U.S. patent application number 17/335799 was filed with the patent office on 2021-11-25 for novel antibodies specifically binding to zika virus epitopes and uses thereof.
The applicant listed for this patent is HUMABS BIOMED SA. Invention is credited to Davide CORTI.
Application Number | 20210363227 17/335799 |
Document ID | / |
Family ID | 1000005756448 |
Filed Date | 2021-11-25 |
United States Patent
Application |
20210363227 |
Kind Code |
A1 |
CORTI; Davide |
November 25, 2021 |
NOVEL ANTIBODIES SPECIFICALLY BINDING TO ZIKA VIRUS EPITOPES AND
USES THEREOF
Abstract
The invention relates to antibodies, and antigen binding
fragments thereof, that potently neutralize infection of ZIKV. The
invention also relates to antigenic sites to which the antibodies
and antigen binding fragments bind, as well as to nucleic acids
that encode and immortalized B cells that produce such antibodies
and antibody fragments. In addition, the invention relates to the
use of the antibodies and antibody fragments of the invention in
screening methods as well as in the diagnosis, prophylaxis and
treatment of ZIKV infection.
Inventors: |
CORTI; Davide; (Bellinzona,
CH) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
HUMABS BIOMED SA |
Bellinzona |
|
CH |
|
|
Family ID: |
1000005756448 |
Appl. No.: |
17/335799 |
Filed: |
June 1, 2021 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
16317533 |
Jan 11, 2019 |
11117954 |
|
|
PCT/EP2017/067581 |
Jul 12, 2017 |
|
|
|
17335799 |
|
|
|
|
Current U.S.
Class: |
1/1 |
Current CPC
Class: |
Y02A 50/30 20180101;
A61P 31/20 20180101; C07K 2317/33 20130101; C07K 2317/21 20130101;
A61K 9/0019 20130101; G01N 33/56983 20130101; A61K 2039/505
20130101; C07K 2317/76 20130101; C07K 16/1081 20130101 |
International
Class: |
C07K 16/10 20060101
C07K016/10; A61P 31/20 20060101 A61P031/20; A61K 9/00 20060101
A61K009/00; G01N 33/569 20060101 G01N033/569 |
Foreign Application Data
Date |
Code |
Application Number |
Jul 13, 2016 |
EP |
PCT/EP2016/066684 |
Claims
1-97. (canceled)
98. A method for in-vitro diagnosis of Zika virus infection, the
method comprising: (i) incubating an isolated sample from a subject
to be diagnosed with ZIKV NS1 protein, (ii) adding an antibody, or
an antigen binding fragment thereof, that specifically binds to NS1
protein of Zika virus, (iii) determining inhibition of binding of
said antibody or antigen binding fragment thereof in the isolated
sample after incubating with ZIKV NS1 protein and the antibody, or
antigen binding fragment thereof.
99. The method of claim 98, wherein the isolated sample from the
subject to be diagnosed is selected from blood, saliva and
urine.
100. The method of claim 98, wherein the antibody, or the antigen
binding fragment thereof is labelled.
101. The method of claim 110, wherein the isolated sample from a
subject to be diagnosed is diluted 1:5-1:50.
102. The method of claim 110, wherein incubating said sample on
said plate is for at least 5 min, and wherein, after adding the
antibody, or the antigen binding fragment thereof, the antibody, or
the antigen binding fragment thereof, is incubated for at least 1
min.
103-109. (canceled)
110. The method of claim 100, wherein the antibody, or the antigen
binding fragment thereof, is biotinylated, conjugated to an enzyme,
conjugated to a fluorescent dye, conjugated to a fluorescent
protein, or conjugated to a second antibody, or an antigen binding
fragment thereof to form an antibody heteroconjugate.
111. The method of claim 98, wherein determining inhibition of
binding of said antibody or antigen binding fragment thereof
comprises determining the optical density of the isolated sample
after incubating with ZIKV NS1 protein and the antibody, or antigen
binding fragment thereof.
112. The method of claim 98, wherein the subject shows symptoms of
Zika virus infection.
113. The method of claim 98, wherein the subject is pregnant.
114. The method of claim 98, wherein the antibody, or antigen
binding fragment thereof, does essentially not bind to Dengue virus
(DENV) NS1 protein, Yellow fever virus (YFV) NS1 protein, West nile
virus (WNV) NS1 protein, Japanese encephalitis virus (JEV) NS1
protein and/or to Tick-borne encephalitis virus (TBEV) NS1
protein.
115. The method of claim 98, wherein the antibody, or antigen
binding fragment thereof, binds to antigenic site S1 of Zika virus
NS1 protein, to antigenic site S2 of Zika virus NS1 protein, or
both.
116. The method of claim 98, wherein the antibody, or antigen
binding fragment thereof, is a monoclonal antibody, or an antigen
binding fragment of a monoclonal antibody.
117. The method of claim 116, wherein the antibody, or antigen
binding fragment thereof, is a recombinant monoclonal antibody, or
an antigen binding fragment of a recombinant monoclonal
antibody.
118. The method of claim 98, wherein the antibody, or antigen
binding fragment thereof, does not comprise an Fc region or is a
single chain antibody, Fab, Fab', F(ab')2, Fv or scFv.
119. The method of claim 98, wherein the antibody, or antigen
binding fragment thereof, comprises CDRH3 amino acid sequence of
SEQ ID NOs: 93, 111, 129, 155, 159, 163, 167, 171, 175, 179, 183,
187, 191, 195, 199, 203, 207, 211, 215, 219, 223, 227, 231.
120. The method of claim 98, wherein in the antibody, or antigen
binding fragment thereof, comprises CDRH1, CDRH2, CDRH3, CDRL1,
CDRL2, and CDRL3 amino acid sequences of: (i) SEQ ID NO: 91, SEQ ID
NO: 92, SEQ ID NO: 93, SEQ ID NO: 94, LGS, and SEQ ID NO: 97,
respectively; or (ii) SEQ ID NO: 91, SEQ ID NO: 92, SEQ ID NO: 93,
SEQ ID NO: 94, SEQ ID NO: 96, and SEQ ID NO: 97, respectively.
121. The method of claim 98, wherein in the antibody, or antigen
binding fragment thereof, comprises CDRH1, CDRH2, CDRH3, CDRL1,
CDRL2, and CDRL3 amino acid sequences of: (i) SEQ ID NO: 109, SEQ
ID NO: 110, SEQ ID NO: 111, SEQ ID NO: 112, QDS, and SEQ ID NO:
115, respectively; or (ii) SEQ ID NO: 109, SEQ ID NO: 110, SEQ ID
NO: 111, SEQ ID NO: 112, SEQ ID NO: 113, and SEQ ID NO: 114,
respectively.
122. The method of claim 98, wherein in the antibody, or antigen
binding fragment thereof, comprises CDRH1, CDRH2, CDRH3, CDRL1,
CDRL2, and CDRL3 amino acid sequences of: (i) SEQ ID NO: 127, SEQ
ID NO: 128, SEQ ID NO: 129, SEQ ID NO: 130, RNN, and SEQ ID NO:
133, respectively; or (ii) SEQ ID NO: 127, SEQ ID NO: 128, SEQ ID
NO: 129, SEQ ID NO: 130, SEQ ID NO: 132, and SEQ ID NO: 133,
respectively.
123. The method of claim 98, wherein the antibody, or the antigen
binding fragment thereof, comprises a heavy chain variable region
(VH) amino acid sequence having at least 70% sequence identity to
the amino acid sequence of SEQ ID NO: 98, and a light chain
variable region (VL) amino acid sequence having at least 70%
sequence identity to the amino acid sequence of SEQ ID NO: 99,
provided that the antibody, or antigen binding fragment thereof,
comprises CDRH1, CDRH2, CDRH3, CDRL1, CDRL2, and CDRL3 amino acid
sequences of: (i) SEQ ID NO: 91, SEQ ID NO: 92, SEQ ID NO: 93, SEQ
ID NO: 94, LGS, and SEQ ID NO: 97, respectively; or (ii) SEQ ID NO:
91, SEQ ID NO: 92, SEQ ID NO: 93, SEQ ID NO: 94, SEQ ID NO: 96, and
SEQ ID NO: 97, respectively.
124. The method of claim 98, wherein the antibody, or the antigen
binding fragment thereof, comprises a heavy chain variable region
(VH) amino acid sequence having at least 70% sequence identity to
the amino acid sequence of SEQ ID NO: 116, and a light chain
variable region (VL) amino acid sequence having at least 70%
sequence identity to the amino acid sequence of SEQ ID NO: 117,
provided that the antibody, or antigen binding fragment thereof,
comprises CDRH1, CDRH2, CDRH3, CDRL1, CDRL2, and CDRL3 amino acid
sequences of: (i) SEQ ID NO: 109, SEQ ID NO: 110, SEQ ID NO: 111,
SEQ ID NO: 112, QDS, and SEQ ID NO: 115, respectively; or (ii) SEQ
ID NO: 109, SEQ ID NO: 110, SEQ ID NO: 111, SEQ ID NO: 112, SEQ ID
NO: 113, and SEQ ID NO: 114, respectively.
125. The method of claim 98, wherein the antibody, or the antigen
binding fragment thereof, comprises a heavy chain variable region
(VH) amino acid sequence having at least 70% sequence identity to
the amino acid sequence of SEQ ID NO: 134, and a light chain
variable region (VL) amino acid sequence having at least 70%
sequence identity to the amino acid sequence of SEQ ID NO: 135,
provided that the antibody, or antigen binding fragment thereof,
comprises CDRH1, CDRH2, CDRH3, CDRL1, CDRL2, and CDRL3 amino acid
sequences of: (i) SEQ ID NO: 127, SEQ ID NO: 128, SEQ ID NO: 129,
SEQ ID NO: 130, RNN, and SEQ ID NO: 133, respectively; or (ii) SEQ
ID NO: 127, SEQ ID NO: 128, SEQ ID NO: 129, SEQ ID NO: 130, SEQ ID
NO: 132, and SEQ ID NO: 133, respectively.
126. The method of claim 98, wherein the method comprises an ELISA,
a sandwich ELISA, a blockade-of-binding assay, an
immunofluorescence assay, an immunohistochemistry assay, or a flow
cytometry assay.
127. A kit comprising: an isolated antibody, or an antigen binding
fragment thereof, that specifically binds to NS1 protein of Zika
virus; one or more solutions usable in a method for in-vitro
diagnosis of Zika virus infection; and a ZIKV NS1 protein.
128. The kit of claim 127, further comprising a label for the
antibody, or antigen binding fragment thereof, wherein the label
comprises biotin, an enzyme, a fluorescent dye, a fluorescent
protein, or a second antibody, or an antigen binding fragment
thereof.
129. The kit of claim 128, wherein the antibody, or antigen binding
fragment thereof is labeled with biotin, an enzyme, a fluorescent
dye, a fluorescent protein, or a second antibody, or an antigen
binding fragment thereof.
130. The kit of claim 128, wherein in the antibody, or antigen
binding fragment thereof, comprises CDRH1, CDRH2, CDRH3, CDRL1,
CDRL2, and CDRL3 amino acid sequences of: (i) SEQ ID NO: 91, SEQ ID
NO: 92, SEQ ID NO: 93, SEQ ID NO: 94, LGS, and SEQ ID NO: 97,
respectively; (ii) SEQ ID NO: 91, SEQ ID NO: 92, SEQ ID NO: 93, SEQ
ID NO: 94, SEQ ID NO: 96, and SEQ ID NO: 97, respectively; (iii)
SEQ ID NO: 109, SEQ ID NO: 110, SEQ ID NO: 111, SEQ ID NO: 112,
QDS, and SEQ ID NO: 115, respectively; (iv) SEQ ID NO: 109, SEQ ID
NO: 110, SEQ ID NO: 111, SEQ ID NO: 112, SEQ ID NO: 113, and SEQ ID
NO: 114, respectively; (v) SEQ ID NO: 127, SEQ ID NO: 128, SEQ ID
NO: 129, SEQ ID NO: 130, RNN, and SEQ ID NO: 133, respectively; or
(vi) SEQ ID NO: 127, SEQ ID NO: 128, SEQ ID NO: 129, SEQ ID NO:
130, SEQ ID NO: 132, and SEQ ID NO: 133, respectively.
131. A complex comprising: an isolated antibody, or an antigen
binding fragment thereof, that specifically binds to NS1 protein of
Zika virus; and a label conjugated to the antibody wherein the
label comprises biotin, an enzyme, a fluorescent dye, or a
fluorescent protein, or a second antibody or an antigen binding
fragment thereof.
132. The complex of claim 131, wherein in the antibody, or antigen
binding fragment thereof, comprises CDRH1, CDRH2, CDRH3, CDRL1,
CDRL2, and CDRL3 amino acid sequences of: (i) SEQ ID NO: 91, SEQ ID
NO: 92, SEQ ID NO: 93, SEQ ID NO: 94, LGS, and SEQ ID NO: 97,
respectively; (ii) SEQ ID NO: 91, SEQ ID NO: 92, SEQ ID NO: 93, SEQ
ID NO: 94, SEQ ID NO: 96, and SEQ ID NO: 97, respectively; (iii)
SEQ ID NO: 109, SEQ ID NO: 110, SEQ ID NO: 111, SEQ ID NO: 112,
QDS, and SEQ ID NO: 115, respectively; (iv) SEQ ID NO: 109, SEQ ID
NO: 110, SEQ ID NO: 111, SEQ ID NO: 112, SEQ ID NO: 113, and SEQ ID
NO: 114, respectively; (v) SEQ ID NO: 127, SEQ ID NO: 128, SEQ ID
NO: 129, SEQ ID NO: 130, RNN, and SEQ ID NO: 133, respectively; or
(vi) SEQ ID NO: 127, SEQ ID NO: 128, SEQ ID NO: 129, SEQ ID NO:
130, SEQ ID NO: 132, and SEQ ID NO: 133, respectively.
Description
STATEMENT REGARDING SEQUENCE LISTING
[0001] The Sequence Listing associated with this application is
provided in text format in lieu of a paper copy, and is hereby
incorporated by reference into the specification. The name of the
text file containing the Sequence Listing is 470082_407
D2_SEQUENCE-LISTING.txt. The text file is 175 kb, was created on
May 26, 2021, and is being submitted electronically via
EFS-Web.
[0002] The present invention relates to antibodies, and antigen
binding fragments thereof, that bind specifically to Zika virus
(ZIKV) epitopes. Such antibodies (i) potently neutralize infection
of Zika virus (ZIKV) or (ii) are directed against NS1 ZIKV and can
be used as diagnostics. The invention also relates to antigenic
sites to which the antibodies and antigen binding fragments bind
to, as well as to nucleic acids that encode the antibodies and
immortalized B cells that produce such antibodies and antibody
fragments. In addition, the invention relates to the use of the
antibodies and antibody fragments of the invention in screening
methods as well as in the diagnosis, prevention and treatment of
ZIKV infection.
[0003] Zika virus (ZIKV), a mosquito-borne flavivirus, is a public
health emergency. ZIKV was first isolated from macaques in 1947 in
the Zika forest in Uganda (G. W. A. Dick, S. F. Kitchen, A. J.
Haddow, Zika virus. I. Isolations and serological specificity.
Trans. R. Soc. Trop. Med. Hyg. 46, 509-520 (1952)) and the first
human infection was reported in Nigeria in 1954 F. N. Macnamara,
Zika virus: a report on three cases of human infection during an
epidemic of jaundice in Nigeria. Trans. R. Soc. Trop. Med. Hyg. 48,
139-145 (1954)). Since then, ZIKV infections were sporadically
reported in Africa and southeast Asia (D. Musso, Van Mai
Cao-Lormeau, D. J. Gubler, Zika virus: following the path of dengue
and chikungunya? The Lancet. 386, 243-244 (2015)), but epidemics
were reported in Micronesia in 2007 (M. R. Duffy et al., Zika virus
outbreak on Yap Island, Federated States of Micronesia. N Engl J
Med. 360, 2536-2543 (2009)) and in French Polynesia in 2013-14,
with the virus subsequently spreading to other countries in the
Oceanian continent (V.-M. Cao-Lormeau, D. Musso, Emerging
arboviruses in the Pacific. Lancet. 384, 1571-1572 (2014); D.
Musso, E. J. Nilles, V.-M. Cao-Lormeau, Rapid spread of emerging
Zika virus in the Pacific area. Clin. Microbiol. Infect. 20, 0595-6
(2014)). After its introduction into Brazil in 2015, ZIKV has
spread rapidly and in February 2016 the World Health Organization
(WHO) declared it a Public Health Emergency of International
Concern (L. R. Baden, L. R. Petersen, D. J. Jamieson, A. M. Powers,
M. A. Honein, Zika Virus. N Engl. J. Med. 374, 1552-1563 (2016); A.
S. Fauci, D. M. Morens, Zika Virus in the Americas--Yet Another
Arbovirus Threat. N Engl J Med, 160113142101009 (2016); D. L.
Heymann et al., Zika virus and microcephaly: why is this situation
a PHEIC? Lancet. 387, 719-721 (2016)). The main route of ZIKV
infection is through bites by Aedes mosquitos, but the virus may
also be sexually (D. Musso et al., Potential sexual transmission of
Zika virus. Emerg Infect Dis. 21, 359-361 (2015)) and vertically
transmitted (J. Mlakar et al., Zika Virus Associated with
Microcephaly. N Engl J Med. 374, 951-958 (2016)). While most of the
ZIKV infections are asymptomatic or cause only mild symptoms, there
is evidence that ZIKV infection can lead to neurological
complications, such as Guillain-Barre Syndrome in adults (V.-M.
Cao-Lormeau et al., Guillain-Barre Syndrome outbreak associated
with Zika virus infection in French Polynesia: a case-control
study. Lancet. O (2016), doi:10.1016/50140-6736(16)00562-6) and
congenital birth defects including microcephaly in the developing
fetus G. Calvet, R. S. Aguiar, A. Melo, S. A. Sampaio, Detection
and sequencing of Zika virus from amniotic fluid of fetuses with
microcephaly in Brazil: a case study. Lancet Infect Dis (2016),
doi:10.1016/s1473-3099(16)00095-5; J. Mlakar et al., Zika Virus
Associated with Microcephaly. N Engl J Med. 374, 951-958 (2016); E.
J. Rubin, M. F. Greene, L. R. Baden, Zika Virus and Microcephaly. N
Engl J Med (2016), doi:10.1056/NEJMe1601862), likely through its
ability to infect human neural progenitor cells H. Tang et al.,
Zika Virus Infects Human Cortical Neural Progenitors and Attenuates
Their Growth. Stem Cell, 1-5 (2016)).
[0004] ZIKV belongs to the genus flavivirus, which also includes
the West Nile virus, dengue virus, tick-borne encephalitis virus,
yellow fever virus, and several other viruses which may cause
encephalitis. Flaviviruses are enveloped, with icosahedral and
spherical geometries. The diameter is around 50 nm. Genomes are
linear positive-sense RNA and non-segmented, around 10-11 kb in
length. The genome of flaviviruses encodes 3 structural proteins
(Capsid, prM, and Envelope) and 8 non-structural proteins (NS1,
NS2A, NS2B, NS3, NS4A, NS4B, NS5 and NS5B).
[0005] While flavivirus envelope (E) proteins mediate fusion and
are the main target of neutralizing antibodies, the non-structural
protein 1 (NS1) is secreted by infected cells and is involved in
immune evasion and pathogenesis (D. A. Muller, P. R. Young, The
flavivirus NS1 protein: molecular and structural biology,
immunology, role in pathogenesis and application as a diagnostic
biomarker. Antiviral Res. 98, 192-208 (2013)). Two recent
structural studies showed a high level of structural similarity
between the E protein of ZIKV and that of other flaviviruses, such
as dengue virus (DENV), yellow fever virus (YFV) and West Nile
virus (WNV) but also revealed unique features that may be related
to the ZIKV neurotropism (L. Dai et al., Structures of the Zika
Virus Envelope Protein and Its Complex with a Flavivirus Broadly
Protective Antibody. Cell Host Microbe (2016),
doi:10.1016/j.chom.2016.04.013; D. Sirohi et al., The 3.8 .ANG.
resolution cryo-EM structure of Zika virus. Science, aaf5316
(2016)). Similarly, the structural analysis of ZIKV NS1 revealed
conserved features with NS1 of other flaviviruses although with
different electrostatic characteristics (J. Kim et al., Zika virus
NS1 structure reveals diversity of electrostatic surfaces among
flaviviruses, 1-6 (2016)).
[0006] A phenomenon that is characteristic of certain flaviviruses
is the disease-enhancing activity of cross-reactive antibodies
elicited by previous infection by heterologous viruses. In the case
of Dengue virus (DENV), for which 4 serotypes are known, there is
epidemiological evidence that a primary infection protects from
reinfection with the same serotype, but represents a risk factor
for the development of severe disease upon reinfection with a
different serotype (S. B. Halstead, Dengue Antibody-Dependent
Enhancement: Knowns and Unknowns. Microbiol Spectr. 2, 249-271
(2014)). The exacerbated disease is triggered by E and prM-specific
antibodies that fail to neutralize the incoming virus but instead
enhance its capture by Fc receptor-expressing (FcR.sup.+) cells,
leading to enhanced viral replication and activation of
cross-reactive memory T cells. The resulting cytokine storm is
thought to be the basis of the most severe form of disease known as
dengue hemorragic fever/dengue shock syndrome (S. B. Halstead,
Neutralization and antibody-dependent enhancement of dengue
viruses. Adv Virus Res. 60, 421-467 (2003); G. Screaton, J.
Mongkolsapaya, S. Yacoub, C. Roberts, New insights into the
immunopathology and control of dengue virus infection. Nat Rev
Immunol. 15, 745-759 (2015). The role of antibodies in severe
dengue is supported by studies showing that waning levels of
maternal antibodies in infants represent a higher risk for
development of severe dengue disease (S. B. Halstead,
Neutralization and antibody-dependent enhancement of dengue
viruses. Adv Virus Res. 60, 421-467 (2003); S. B. Halstead et al.,
Dengue hemorrhagic fever in infants: research opportunities
ignored. Emerging Infect Dis. 8, 1474-1479 (2002); T. H. Nguyen et
al., Dengue hemorrhagic fever in infants: a study of clinical and
cytokine profiles. J Infect Dis. 189, 221-232 (2004); A. L.
Rothman, Dengue: defining protective versus pathologic immunity. J
Clin Invest. 113, 946-951 (2004)).
[0007] Recently, it was shown that most antibodies that reacted to
DENV envelope protein also bound to ZIKV, but those that recognize
the major linear fusion-loop epitope (FLE) did not neutralize ZIKV
and instead promoted antibody-dependent enhancement (ADE) of ZIKV
infection (Dejnirattisai W, Supasa P, Wongwiwat W, Rouvinski A,
Barba-Spaeth G, Duangchinda T, Sakuntabhai A, Cao-Lormeau V M,
Malasit P, Rey F A, Mongkolsapaya J, Screaton G R: Dengue virus
sero-cross-reactivity drives antibody-dependent enhancement of
infection with zika virus. Nat Immunol. 2016 Jun. 23. doi:
10.1038/ni.3515. [Epub ahead of print]).
[0008] Moreover, according to the WHO, the recent increase in cases
of microcephaly and other neurological disorders potentially
associated with Zika virus infection has prompted an increase in
demand for laboratory testing to detect Zika virus infection. In
this context, high specificity of the antibodies is required in
order to distinguish ZIKV infection from infection of other
flaviviruses. However, known anti-Zika antibodies are typically
cross-reactive for other flaviviruses and, thus, not useful to
distinguish ZIKV infection from infection of other
flaviviruses.
[0009] In view of the above, it is an object of the present
invention to provide novel antibodies, which specifically bind to
ZIKV epitopes. It is also an object of the present invention to
provide potently neutralizing anti-ZIKV antibodies. Such antibodies
do preferably not contribute to antibody-dependent enhancement
(ADE) of Zika virus infection. It is also an object of the present
invention to provide highly specific anti-ZIKV antibodies useful in
diagnosis and testing of ZIKV infection and diagnosis methods using
such antibodies.
[0010] The object underlying the present invention is solved by the
claimed subject matter.
[0011] Although the present invention is described in detail below,
it is to be understood that this invention is not limited to the
particular methodologies, protocols and reagents described herein
as these may vary. It is also to be understood that the terminology
used herein is not intended to limit the scope of the present
invention which will be limited only by the appended claims. Unless
defined otherwise, all technical and scientific terms used herein
have the same meanings as commonly understood by one of ordinary
skill in the art.
[0012] In the following, the elements of the present invention will
be described. These elements are listed with specific embodiments,
however, it should be understood that they may be combined in any
manner and in any number to create additional embodiments. The
variously described examples and preferred embodiments should not
be construed to limit the present invention to only the explicitly
described embodiments. This description should be understood to
support and encompass embodiments which combine the explicitly
described embodiments with any number of the disclosed and/or
preferred elements. Furthermore, any permutations and combinations
of all described elements in this application should be considered
disclosed by the description of the present application unless the
context indicates otherwise.
[0013] Throughout this specification and the claims which follow,
unless the context requires otherwise, the term "comprise", and
variations such as "comprises" and "comprising", will be understood
to imply the inclusion of a stated member, integer or step but not
the exclusion of any other non-stated member, integer or step. The
term "consist of" is a particular embodiment of the term
"comprise", wherein any other non-stated member, integer or step is
excluded. In the context of the present invention, the term
"comprise" encompasses the term "consist of". The term "comprising"
thus encompasses "including" as well as "consisting" e.g., a
composition "comprising" X may consist exclusively of X or may
include something additional e.g., X+Y.
[0014] The terms "a" and "an" and "the" and similar reference used
in the context of describing the invention (especially in the
context of the claims) are to be construed to cover both the
singular and the plural, unless otherwise indicated herein or
clearly contradicted by context. Recitation of ranges of values
herein is merely intended to serve as a shorthand method of
referring individually to each separate value falling within the
range. Unless otherwise indicated herein, each individual value is
incorporated into the specification as if it were individually
recited herein. No language in the specification should be
construed as indicating any non-claimed element essential to the
practice of the invention.
[0015] The word "substantially" does not exclude "completely" e.g.,
a composition which is "substantially free" from Y may be
completely free from Y. Where necessary, the word "substantially"
may be omitted from the definition of the invention.
[0016] The term "about" in relation to a numerical value x means
x.+-.10%.
[0017] The term "disease" as used herein is intended to be
generally synonymous, and is used interchangeably with, the terms
"disorder" and "condition" (as in medical condition), in that all
reflect an abnormal condition of the human or animal body or of one
of its parts that impairs normal functioning, is typically
manifested by distinguishing signs and symptoms, and causes the
human or animal to have a reduced duration or quality of life.
[0018] As used herein, reference to "treatment" of a subject or
patient is intended to include prevention, prophylaxis,
attenuation, amelioration and therapy. The terms "subject" or
"patient" are used interchangeably herein to mean all mammals
including humans. Examples of subjects include humans, cows, dogs,
cats, horses, goats, sheep, pigs, and rabbits. In one embodiment,
the patient is a human.
[0019] As used herein, the terms "antigen binding fragment,"
"fragment," and "antibody fragment" are used interchangeably to
refer to any fragment of an antibody of the invention that retains
the antigen-binding activity of the antibody. Examples of antibody
fragments include, but are not limited to, a single chain antibody,
Fab, Fab', F(ab').sub.2, Fv or scFv. Further, the term "antibody"
as used herein includes both antibodies and antigen binding
fragments thereof.
[0020] As used herein, the term "antibody" encompasses various
forms of antibodies including, without being limited to, whole
antibodies, antibody fragments, in particular antigen binding
fragments, human antibodies, chimeric antibodies, humanized
antibodies, recombinant antibodies and genetically engineered
antibodies (variant or mutant antibodies) as long as the
characteristic properties according to the invention are retained.
Human antibodies and monoclonal antibodies are preferred and
especially preferred are human monoclonal antibodies, in particular
as recombinant human monoclonal antibodies.
[0021] Human antibodies are well-known in the state of the art (van
Dijk, M. A., and van de Winkel, J. G., Curr. Opin. Chem. Biol. 5
(2001) 368-374). Human antibodies can also be produced in
transgenic animals (e.g., mice) that are capable, upon
immunization, of producing a full repertoire or a selection of
human antibodies in the absence of endogenous immunoglobulin
production. Transfer of the human germ-line immunoglobulin gene
array in such germ-line mutant mice will result in the production
of human antibodies upon antigen challenge (see, e.g., Jakobovits,
A., et al., Proc. Natl. Acad. Sci. USA 90 (1993) 2551-2555;
Jakobovits, A., et al., Nature 362 (1993) 255-258; Bruggemann, M.,
et al., rear Immunol. 7 (1993) 3340). Human antibodies can also be
produced in phage display libraries (Hoogenboom, H. R., and Winter,
G., J. Mol. Biol. 227 (1992) 381-388; Marks, J. D., et al., J. Mol.
Biol. 222 (1991) 581-597). The techniques of Cole et al. and
Boerner et al. are also available for the preparation of human
monoclonal antibodies (Cole et al., Monoclonal Antibodies and
Cancer Therapy, Alan R. Liss, p. 77 (1985); and Boerner, P., et
al., J. Immunol. 147 (1991) 86-95). Preferably, human monoclonal
antibodies are prepared by using improved EBV-B cell
immortalization as described in Traggiai E, Becker S, Subbarao K,
Kolesnikova L, Uematsu Y, Gismondo M R, Murphy B R, Rappuoli R,
Lanzavecchia A. (2004): An efficient method to make human
monoclonal antibodies from memory B cells: potent neutralization of
SARS coronavirus. Nat Med. 10(8):871-5. The term "human antibody"
as used herein also comprises such antibodies which are modified,
e.g. in the variable region, to generate the properties according
to the invention as described herein. As used herein, the term
"variable region" (variable region of a light chain (V.sub.L),
variable region of a heavy chain (V.sub.H)) denotes each of the
pair of light and heavy chains which is involved directly in
binding the antibody to the antigen.
[0022] Antibodies of the invention can be of any isotype (e.g.,
IgA, IgG, IgM i.e. an .alpha., .gamma. or .mu. heavy chain), but
will preferably be IgG. Within the IgG isotype, antibodies may be
IgG1, IgG2, IgG3 or IgG4 subclass, whereby IgG1 is preferred.
Antibodies of the invention may have a .kappa. or a .lamda. light
chain.
[0023] Preferably, the antibody according to the present invention,
or the antigen binding fragment thereof, is a purified antibody, a
single chain antibody, Fab, Fab', F(ab')2, Fv or scFv.
[0024] The antibodies of the invention may thus preferably be human
antibodies, monoclonal antibodies, human monoclonal antibodies,
recombinant antibodies or purified antibodies. The invention also
provides fragments of the antibodies of the invention, particularly
fragments that retain the antigen-binding activity of the
antibodies. Such fragments include, but are not limited to, single
chain antibodies, Fab, Fab', F(ab')2, Fv or scFv. Although the
specification, including the claims, may, in some places, refer
explicitly to antigen binding fragment(s), antibody fragment(s),
variant(s) and/or derivative(s) of antibodies, it is understood
that the term "antibody" or "antibody of the invention" includes
all categories of antibodies, namely, antigen binding fragment(s),
antibody fragment(s), variant(s) and derivative(s) of
antibodies.
[0025] Fragments of the antibodies of the invention can be obtained
from the antibodies by methods that include digestion with enzymes,
such as pepsin or papain, and/or by cleavage of disulfide bonds by
chemical reduction. Alternatively, fragments of the antibodies can
be obtained by cloning and expression of part of the sequences of
the heavy or light chains. Antibody "fragments" include Fab, Fab',
F(ab')2 and Fv fragments. The invention also encompasses
single-chain Fv fragments (scFv) derived from the heavy and light
chains of an antibody of the invention. For example, the invention
includes a scFv comprising the CDRs from an antibody of the
invention. Also included are heavy or light chain monomers and
dimers, single domain heavy chain antibodies, single domain light
chain antibodies, as well as single chain antibodies, e.g., single
chain Fv in which the heavy and light chain variable domains are
joined by a peptide linker.
[0026] Antibody fragments of the invention may impart monovalent or
multivalent interactions and be contained in a variety of
structures as described above. For instance, scFv molecules may be
synthesized to create a trivalent "triabody" or a tetravalent
"tetrabody." The scFv molecules may include a domain of the Fc
region resulting in bivalent minibodies. In addition, the sequences
of the invention may be a component of multispecific molecules in
which the sequences of the invention target the epitopes of the
invention and other regions of the molecule bind to other targets.
Exemplary molecules include, but are not limited to, bispecific
Fab2, trispecific Fab3, bispecific scFv, and diabodies (Holliger
and Hudson, 2005, Nature Biotechnology 9: 1126-1136).
[0027] Antibodies according to the present invention may be
provided in purified form. Typically, the antibody will be present
in a composition that is substantially free of other polypeptides
e.g., where less than 90% (by weight), usually less than 60% and
more usually less than 50% of the composition is made up of other
polypeptides.
[0028] Antibodies according to the present invention may be
immunogenic in human and/or in non-human (or heterologous) hosts
e.g., in mice. For example, the antibodies may have an idiotope
that is immunogenic in non-human hosts, but not in a human host.
Antibodies of the invention for human use include those that cannot
be easily isolated from hosts such as mice, goats, rabbits, rats,
non-primate mammals, etc. and cannot generally be obtained by
humanization or from xeno-mice.
[0029] As used herein, a "neutralizing antibody" is one that can
neutralize, i.e., prevent, inhibit, reduce, impede or interfere
with, the ability of a pathogen to initiate and/or perpetuate an
infection in a host. The terms "neutralizing antibody" and "an
antibody that neutralizes" or "antibodies that neutralize" are used
interchangeably herein. These antibodies can be used alone, or in
combination, as prophylactic or therapeutic agents upon appropriate
formulation, in association with active vaccination, as a
diagnostic tool, or as a production tool as described herein.
[0030] Doses are often expressed in relation to the bodyweight.
Thus, a dose which is expressed as [g, mg, or other unit]/kg (or g,
mg etc.) usually refers to [g, mg, or other unit] "per kg (or g, mg
etc.) bodyweight", even if the term "bodyweight" is not explicitly
mentioned.
[0031] The term "specifically binding" and similar reference does
not encompass non-specific sticking.
[0032] The term "vaccine" as used herein is typically understood to
be a prophylactic or therapeutic material providing at least one
antigen, preferably an immunogen. The antigen or immunogen may be
derived from any material that is suitable for vaccination. For
example, the antigen or immunogen may be derived from a pathogen,
such as from bacteria or virus particles etc., or from a tumor or
cancerous tissue. The antigen or immunogen stimulates the body's
adaptive immune system to provide an adaptive immune response. In
particular, an "antigen" or an "immunogen" refers typically to a
substance which may be recognized by the immune system, preferably
by the adaptive immune system, and which is capable of triggering
an antigen-specific immune response, e.g. by formation of
antibodies and/or antigen-specific T cells as part of an adaptive
immune response. Typically, an antigen may be or may comprise a
peptide or protein which may be presented by the MHC to
T-cells.
[0033] As used herein, "sequence variant" (also referred to as
"variant") refers to any alteration in a reference sequence,
whereby a reference sequence is any of the sequences listed in the
"Tables of Sequences and SEQ ID Numbers" (sequence listing), i.e.
SEQ ID NO: 1 to SEQ ID NO: 407. Thus, the term "sequence variant"
includes nucleotide sequence variants and amino acid sequence
variants. Of note, the sequence variants referred to herein are in
particular functional sequence variants, i.e. sequence variants
maintaining the biological function of, for example, the antibody.
In the context of the present invention such a maintained
biological function is preferably the neutralization of ZIKV
infection, the binding of the antibody to the ZIKV E protein and/or
the binding of the antibody to the ZIKV NS1 protein. Preferred
sequence variants are thus functional sequence variants having at
least 70%, at least 75%, at least 80%, at least 85%, at least 88%,
at least 90%, at least 92%, at least 95%, at least 96%, at least
97%, at least 98% or at least 99% sequence identity to a reference
sequence. The phrase "functional sequence variant thereof having at
least 70%, at least 75%, at least 80%, at least 85%, at least 88%,
at least 90%, at least 92%, at least 95%, at least 96%, at least
97%, at least 98% or at least 99% sequence identity", as used
herein, means (i) that the sequence variant is functional as
described herein and (ii) the higher the % sequence identity, the
more preferred the sequence variant. In other words, the phrase
"functional sequence variant thereof having at least 70%, at least
75%, at least 80%, at least 85%, at least 88%, at least 90%, at
least 92%, at least 95%, at least 96%, at least 97%, at least 98%
or at least 99% sequence identity", means in particular that the
functional sequence variant has at least 70% sequence identity,
preferably at least 75% sequence identity, preferably at least 80%
sequence identity, more preferably at least 85% sequence identity,
more preferably at least 88% sequence identity, even more
preferably at least 90% sequence identity, even more preferably at
least 92% sequence identity, still more preferably at least 95%
sequence identity, still more preferably at least 96% sequence
identity, particularly preferably at least 97% sequence identity,
particularly preferably at least 98% sequence identity and most
preferably at least 99% sequence identity to the respective
reference sequence.
[0034] The term "sequence variant" includes in particular such
variants that comprise mutations and/or substitutions in comparison
to the reference sequence. Exemplary variants of an Fc moiety
sequence include, but are not limited to, those that have an L to A
substitution at position CH2 4, CH2 5, or both.
[0035] Sequence identity is usually calculated with regard to the
full length of the reference sequence (i.e. the sequence recited in
the application). Percentage identity, as referred to herein, can
be determined, for example, using BLAST using the default
parameters specified by the NCBI (the National Center for
Biotechnology Information; http://www.ncbi.nlm.nih.gov/) [Blosum 62
matrix; gap open penalty=11 and gap extension penalty=1].
[0036] As used herein, a "nucleotide sequence variant" has an
altered sequence in which one or more of the nucleotides in the
reference sequence is deleted, or substituted, or one or more
nucleotides are inserted into the sequence of the reference
nucleotide sequence. Nucleotides are referred to herein by the
standard one-letter designation (A, C, G, or T). Due to the
degeneracy of the genetic code, a "nucleotide sequence variant" can
either result in a change in the respective reference amino acid
sequence, i.e. in an "amino acid sequence variant" or not.
Preferred sequence variants are such nucleotide sequence variants,
which do not result in amino acid sequence variants (silent
mutations), but other non-silent mutations are within the scope as
well, in particular mutant nucleotide sequences, which result in an
amino acid sequence, which is at least 80%, preferably at least
90%, more preferably at least 95% sequence identical to the
reference sequence.
[0037] An "amino acid sequence variant" has an altered sequence in
which one or more of the amino acids in the reference sequence is
deleted or substituted, or one or more amino acids are inserted
into the sequence of the reference amino acid sequence. As a result
of the alterations, the amino acid sequence variant has an amino
acid sequence which is at least 80% identical to the reference
sequence, preferably, at least 90% identical, more preferably at
least 95% identical, most preferably at least 99% identical to the
reference sequence. Variant sequences which are at least 90%
identical have no more than 10 alterations, i.e. any combination of
deletions, insertions or substitutions, per 100 amino acids of the
reference sequence.
[0038] While it is possible to have non-conservative amino acid
substitutions, it is preferred that the substitutions be
conservative amino acid substitutions, in which the substituted
amino acid has similar structural or chemical properties with the
corresponding amino acid in the reference sequence. By way of
example, conservative amino acid substitutions involve substitution
of one aliphatic or hydrophobic amino acids, e.g. alanine, valine,
leucine and isoleucine, with another; substitution of one
hydoxyl-containing amino acid, e.g. serine and threonine, with
another; substitution of one acidic residue, e.g. glutamic acid or
aspartic acid, with another; replacement of one amide-containing
residue, e.g. asparagine and glutamine, with another; replacement
of one aromatic residue, e.g. phenylalanine and tyrosine, with
another; replacement of one basic residue, e.g. lysine, arginine
and histidine, with another; and replacement of one small amino
acid, e.g., alanine, serine, threonine, methionine, and glycine,
with another.
[0039] Amino acid sequence insertions include amino- and/or
carboxyl-terminal fusions ranging in length from one residue to
polypeptides containing a hundred or more residues, as well as
intrasequence insertions of single or multiple amino acid residues.
Examples of terminal insertions include the fusion to the N- or
C-terminus of an amino acid sequence to a reporter molecule or an
enzyme.
[0040] Importantly, the alterations in the sequence variants do not
abolish the functionality of the respective reference sequence, in
the present case, e.g., the functionality of a sequence of an
antibody, or antigen binding fragment thereof, to bind to the same
epitope and/or to sufficiently neutralize infection of ZIKV.
Guidance in determining which nucleotides and amino acid residues,
respectively, may be substituted, inserted or deleted without
abolishing such functionality are found by using computer programs
well known in the art.
[0041] As used herein, a nucleic acid sequence or an amino acid
sequence "derived from" a designated nucleic acid, peptide,
polypeptide or protein refers to the origin of the nucleic acid,
peptide, polypeptide or protein. Preferably, the nucleic acid
sequence or amino acid sequence which is derived from a particular
sequence has an amino acid sequence that is essentially identical
to that sequence or a portion thereof, from which it is derived,
whereby "essentially identical" includes sequence variants as
defined above. Preferably, the nucleic acid sequence or amino acid
sequence which is derived from a particular peptide or protein, is
derived from the corresponding domain in the particular peptide or
protein. Thereby, "corresponding" refers in particular to the same
functionality. For example, an "extracellular domain" corresponds
to another "extracellular domain" (of another protein), or a
"transmembrane domain" corresponds to another "transmembrane
domain" (of another protein). "Corresponding" parts of peptides,
proteins and nucleic acids are thus easily identifiable to one of
ordinary skill in the art. Likewise, sequences "derived from" other
sequence are usually easily identifiable to one of ordinary skill
in the art as having its origin in the sequence.
[0042] Preferably, a nucleic acid sequence or an amino acid
sequence derived from another nucleic acid, peptide, polypeptide or
protein may be identical to the starting nucleic acid, peptide,
polypeptide or protein (from which it is derived). However, a
nucleic acid sequence or an amino acid sequence derived from
another nucleic acid, peptide, polypeptide or protein may also have
one or more mutations relative to the starting nucleic acid,
peptide, polypeptide or protein (from which it is derived), in
particular a nucleic acid sequence or an amino acid sequence
derived from another nucleic acid, peptide, polypeptide or protein
may be a functional sequence variant as described above of the
starting nucleic acid, peptide, polypeptide or protein (from which
it is derived). For example, in a peptide/protein one or more amino
acid residues may be substituted with other amino acid residues or
one or more amino acid residue insertions or deletions may
occur.
[0043] As used herein, the term "mutation" relates to a change in
the nucleic acid sequence and/or in the amino acid sequence in
comparison to a reference sequence, e.g. a corresponding genomic
sequence.
[0044] A mutation, e.g. in comparison to a genomic sequence, may
be, for example, a (naturally occurring) somatic mutation, a
spontaneous mutation, an induced mutation, e.g. induced by enzymes,
chemicals or radiation, or a mutation obtained by site-directed
mutagenesis (molecular biology methods for making specific and
intentional changes in the nucleic acid sequence and/or in the
amino acid sequence). Thus, the terms "mutation" or "mutating"
shall be understood to also include physically making a mutation,
e.g. in a nucleic acid sequence or in an amino acid sequence. A
mutation includes substitution, deletion and insertion of one or
more nucleotides or amino acids as well as inversion of several
successive nucleotides or amino acids. To achieve a mutation in an
amino acid sequence, preferably a mutation may be introduced into
the nucleotide sequence encoding said amino acid sequence in order
to express a (recombinant) mutated polypeptide. A mutation may be
achieved e.g., by altering, e.g., by site-directed mutagenesis, a
codon of a nucleic acid molecule encoding one amino acid to result
in a codon encoding a different amino acid, or by synthesizing a
sequence variant, e.g., by knowing the nucleotide sequence of a
nucleic acid molecule encoding a polypeptide and by designing the
synthesis of a nucleic acid molecule comprising a nucleotide
sequence encoding a variant of the polypeptide without the need for
mutating one or more nucleotides of a nucleic acid molecule.
[0045] Several documents are cited throughout the text of this
specification. Each of the documents cited herein (including all
patents, patent applications, scientific publications,
manufacturer's specifications, instructions, etc.), whether supra
or infra, are hereby incorporated by reference in their entirety.
Nothing herein is to be construed as an admission that the
invention is not entitled to antedate such disclosure by virtue of
prior invention.
[0046] It is to be understood that this invention is not limited to
the particular methodology, protocols and reagents described herein
as these may vary. It is also to be understood that the terminology
used herein is for the purpose of describing particular embodiments
only, and is not intended to limit the scope of the present
invention which will be limited only by the appended claims. Unless
defined otherwise, all technical and scientific terms used herein
have the same meanings as commonly understood by one of ordinary
skill in the art.
[0047] Antibodies Potently Neutralizing Zika Virus Infection
[0048] The present invention is based, amongst other findings, on
the discovery and isolation of antibodies that bind specifically to
Zika virus epitopes. Such antibodies are either (i) highly potent
in neutralizing Zika virus, if directed to an antigenic site of
Zika virus envelope (E) protein or to a ZIKV quaternary epitope or
(ii) useful in diagnosis of Zika virus infection, if directed to
Zika virus NS1 protein. Such antibodies are desirable, as only
small quantities of the antibodies are required in order to
neutralize Zika virus. In particular, there is currently no
prevention/treatment available for Zika virus infection. The
antibodies according to the present invention are highly effective
in preventing as well as treating or attenuating Zika virus
infection. Moreover, due to the specificity of the antibodies for
Zika virus, they do not elicit ADE, but rather block ADE. In
diagnosis, Zika-specific antibodies provide an important tool for
distinguishing Zika virus infection from infection with other
flaviviruses, such as Dengue virus.
[0049] In a first aspect the present invention provides an isolated
antibody, or an antigen binding fragment thereof, that specifically
binds to a Zika virus epitope and neutralizes Zika virus infection.
In other words, the antibody, or the antigen binding fragment
thereof, according to the present invention, reduces viral
infectivity of Zika virus.
[0050] To study and quantitate virus infectivity (or
"neutralization") in the laboratory the person skilled in the art
knows various standard "neutralization assays". For a
neutralization assay animal viruses are typically propagated in
cells and/or cell lines. In the context of the present invention a
neutralization assay is preferred, wherein cultured cells are
incubated with a fixed amount of Zika virus (ZIKV) in the presence
(or absence) of the antibody to be tested. As a readout for example
flow cytometry may be used. Alternatively, also other readouts are
conceivable, such as determining the amount of ZIKV non-structural
proteins (such as ZIKV NS1) secreted into culture supernatant. For
example, a ZIKV nonstructural protein 1 (NS1) antigen capture
enzyme-linked immunosorbent assay (ELISA)-based tissue culture
infectious dose-50 (TCID50) test (TCID50-ELISA) may be used as an
alternative to the standard plaque assay for titrating Zika
virus--in a similar manner as described for dengue virus (DENV) by
Li J, Hu D-M, Ding X-X, Chen Y, Pan Y-X, Qiu L-W, Che X-Y:
Enzyme-linked immunosorbent assay-format tissue culture infectious
dose-50 test for titrating dengue virus. PLoS ONE 2011, 6:e22553.
In such an assay for example the ZIKV NS1-binding antibodies as
described in the present application may be advantageously
used.
[0051] In a preferred embodiment of a ZIKV neutralization assay,
cultured cells, for example Vero cells, are incubated with a fixed
amount of ZIKV in the presence or absence of the antibody to be
tested, for example for about four days. After incubation, cells
may be washed and further cultivated. To measure virus infectivity,
flow cytometry may be used. To this end, cells may be fixed, e.g.
with 2% formaldehyde, permeabilizes, e.g. in PBS (phosphate
buffered saline) 1% FCS (fetal calf serum) 0.5% saponin, and
stained, e.g. with mouse antibody 4G2. Cells may then be incubated
with a goat anti-mouse IgG conjugated to a dye, such as Alexa
Fluor488 and analyzed by flow cytometry.
[0052] Alternatively, viable cells may be detected by flow
cytometry using for example the WST-1 reagent (Roche). A preferred
ZIKV strain to be used in such a neutralization assay is ZIKV
H/PF/2013.
[0053] The antibody and antigen binding fragment of the invention
have high neutralizing potency. The concentration of the antibody
required for 50% neutralization of Zika virus (IC.sub.50) as
compared to no-antibody controls, is, for example, up to about 3
.mu.g/ml or up to about 1 .mu.g/ml. Preferably, the concentration
of the antibody of the invention required for 50% neutralization of
ZIKV (IC.sub.50) is up to about 500 ng/ml, more preferably the
concentration of the antibody of the invention required for 50%
neutralization of ZIKV (IC.sub.50) is up to about 250 ng/ml, even
more preferably the concentration of the antibody of the invention
required for 50% neutralization of ZIKV (IC.sub.50) is up to about
150 ng/ml. Most preferably, the concentration of the antibody of
the invention required for 50% neutralization of ZIKV (IC.sub.50)
is about 100 ng/ml or less, e.g. about 90 ng/ml or less, about 80
ng/ml or less, about 70 ng/ml or less, about 60 ng/ml or less,
about 50 ng/ml or less, about 45 ng/ml or less, about 40 ng/ml or
less, about 35 ng/ml or less, about 30 ng/ml or less, about 25
ng/ml or less, about 20 ng/ml or less or, particularly preferably,
about 15 ng/ml or less. In particular, the concentration of the
antibody of the invention required for 50% neutralization of ZIKV
(IC.sub.50) is preferably about 50 ng/ml or less. This means that
only low concentrations of the antibody are required for 50%
neutralization of ZIKV. The concentration of the antibody of the
invention required for 50% neutralization of ZIKV (IC.sub.50) can
be measured using standard neutralization assays as known to one of
skill in the art or, in particular, as described above.
[0054] In general, binding of an antibody may be assessed by use of
a standard ELISA (enzyme-linked immunosorbent assay), which is
well-known to the skilled person. An exemplary standard ELISA may
be performed as follows: ELISA plates may be coated (e.g.,
overnight at 4.degree. C.) with a sufficient amount (e.g., 1
.mu.g/ml) of the protein/complex/particle to which binding of the
antibody is to be tested (for example, for DENV binding as outlined
below, DENV E proteins and/or DENV VLPs are used), e.g. in PBS.
Plates may then be blocked, e.g. with a 1% w/v solution of Bovine
Serum Albumin (BSA) in PBS, and incubated with the antibody to be
tested (e.g. for about 1.5 hours at room temperature). After
washing, antibody binding can be revealed, e.g. using goat
anti-human IgG coupled to alkaline phosphatase. Plates may then be
washed, the required substrate (e.g., p-NPP) may be added and
plates may be read, e.g. at 405 nm. The relative affinities of
antibody binding may be determined by measuring the concentration
of mAb (EC.sub.50) required to achieve 50% maximal binding at
saturation. The EC.sub.50 values may be calculated by interpolation
of binding curves fitted with a four-parameter nonlinear regression
with a variable slope.
[0055] Preferably, the antibody, or an antigen binding fragment
thereof, according to the present invention does essentially not
bind to Dengue virus-like particles and/or to Dengue envelope
protein. More preferably, the antibody, or an antigen binding
fragment thereof, according to the present invention does
essentially not bind to Dengue virus-like particles and/or to
Dengue envelope protein of any of the four DENV serotypes DENV1,
DENV2, DENV3 and DENV4. Thereby "essentially not binding" means
that for the antibody, or an antigen binding fragment thereof, no
EC.sub.50-value up to 10.sup.2 ng/ml, preferably up to 10.sup.3
ng/ml, more preferably up to 5*10.sup.3 ng/ml, even more preferably
up to 8*10.sup.3 ng/ml, and most preferably up to 10.sup.4 ng/ml
can be determined in a standard ELISA to Dengue virus-like
particles (DENV VLP) and/or to Dengue envelope protein (DENV E
protein). In other words, the concentration of the antibody, or an
antigen binding fragment thereof, required to achieve 50% maximal
binding at saturation (EC.sub.50) to Dengue virus-like particles
(DENV VLP) and/or to Dengue envelope protein (DENV E protein) in a
standard ELISA is typically more than 10.sup.2 ng/ml, preferably
more than 10.sup.3 ng/ml, more preferably more than 5*10.sup.3
ng/ml, even more preferably more than 8*10.sup.3 ng/ml, and most
preferably more than 10.sup.4 ng/ml.
[0056] Preferably, the antibody, or an antigen binding fragment
thereof, according to the present invention does not contribute to
antibody-dependent enhancement (ADE) of Zika virus infection. More
preferably, the antibody, or an antigen binding fragment thereof,
according to the present invention blocks antibody-dependent
enhancement (ADE) of Zika virus infection.
[0057] ADE may be assessed by a flow-cytometry based assay using,
for example cultured cells or cell lines, such as K562 cells. For
example, the antibodies to be tested and ZIKV may be mixed for 1
hour at 37.degree. C. and added to 5000 K562 cells/well. After four
days, cells may be fixed, permeabilized, and stained with m4G2,
e.g. as described above for neutralization assays. The number of
infected cells was determined by flow cytometry, as described above
for neutralization assays.
[0058] Preferably, the antibody, or an antigen binding fragment
thereof, according to the present invention is a human antibody. It
is also preferred that the antibody, or an antigen binding fragment
thereof, according to the present invention is a monoclonal
antibody, preferably a human monoclonal antibody. Furthermore, it
is also preferred that the antibody, or an antigen binding fragment
thereof, according to the present invention is a recombinant
antibody.
[0059] Preferably, the antibody according to the present invention,
or an antigen binding fragment thereof, comprises an Fc moiety.
More preferably, the Fc moiety is derived from human origin, e.g.
from human IgG1, IgG2, IgG3, and/or IgG4, whereby human IgG1 is
particularly preferred.
[0060] As used herein, the term "Fc moiety" refers to a sequence
derived from the portion of an immunoglobulin heavy chain beginning
in the hinge region just upstream of the papain cleavage site
(e.g., residue 216 in native IgG, taking the first residue of heavy
chain constant region to be 114) and ending at the C-terminus of
the immunoglobulin heavy chain. Accordingly, an Fc moiety may be a
complete Fc moiety or a portion (e.g., a domain) thereof. A
complete Fc moiety comprises at least a hinge domain, a CH2 domain,
and a CH3 domain (e.g., EU amino acid positions 216-446). An
additional lysine residue (K) is sometimes present at the extreme
C-terminus of the Fc moiety, but is often cleaved from a mature
antibody. Each of the amino acid positions within an Fc moiety have
been numbered according to the art-recognized EU numbering system
of Kabat, see e.g., by Kabat et al., in "Sequences of Proteins of
Immunological Interest", U.S. Dept. Health and Human Services, 1983
and 1987.
[0061] Preferably, in the context of the present invention an Fc
moiety comprises at least one of: a hinge (e.g., upper, middle,
and/or lower hinge region) domain, a CH2 domain, a CH3 domain, or a
variant, portion, or fragment thereof. In preferred embodiments, an
Fc moiety comprises at least a hinge domain, a CH2 domain or a CH3
domain. More preferably, the Fc moiety is a complete Fc moiety. The
Fc moiety may also comprises one or more amino acid insertions,
deletions, or substitutions relative to a naturally-occurring Fc
moiety. For example, at least one of a hinge domain, CH2 domain or
CH3 domain (or portion thereof) may be deleted. For example, an Fc
moiety may comprise or consist of: (i) hinge domain (or portion
thereof) fused to a CH2 domain (or portion thereof), (ii) a hinge
domain (or portion thereof) fused to a CH3 domain (or portion
thereof), (iii) a CH2 domain (or portion thereof) fused to a CH3
domain (or portion thereof), (iv) a hinge domain (or portion
thereof), (v) a CH2 domain (or portion thereof), or (vi) a CH3
domain or portion thereof.
[0062] It will be understood by one of ordinary skill in the art
that the Fc moiety may be modified such that it varies in amino
acid sequence from the complete Fc moiety of a naturally occurring
immunoglobulin molecule, while retaining at least one desirable
function conferred by the naturally-occurring Fc moiety. Such
functions include Fc receptor (FcR) binding, antibody half-life
modulation, ADCC function, protein A binding, protein G binding,
and complement binding. The portions of naturally occurring Fc
moieties, which are responsible and/or essential for such functions
are well known by those skilled in the art.
[0063] For example, to activate the complement cascade C q binds to
at least two molecules of IgG1 or one molecule of IgM, attached to
the antigenic target (Ward, E. S., and Ghetie, V., Ther. Immunol. 2
(1995) 77-94). Burton, D. R., described (Mol. Immunol. 22 (1985)
161-206) that the heavy chain region comprising amino acid residues
318 to 337 is involved in complement fixation. Duncan, A. R., and
Winter, G. (Nature 332 (1988) 738-740), using site directed
mutagenesis, reported that Glu318, Lys320 and Lys322 form the
binding site to C1q. The role of Glu318, Lys320 and Lys 322
residues in the binding of C1q was confirmed by the ability of a
short synthetic peptide containing these residues to inhibit
complement mediated lysis.
[0064] For example, FcR binding can be mediated by the interaction
of the Fc moiety (of an antibody) with Fc receptors (FcRs), which
are specialized cell surface receptors on hematopoietic cells. Fc
receptors belong to the immunoglobulin superfamily, and were shown
to mediate both the removal of antibody-coated pathogens by
phagocytosis of immune complexes, and the lysis of erythrocytes and
various other cellular targets (e.g. tumor cells) coated with the
corresponding antibody, via antibody dependent cell mediated
cytotoxicity (ADCC; Van de Winkel, J. G., and Anderson, C. L., J.
Leukoc. Biol. 49 (1991) 511-524). FcRs are defined by their
specificity for immunoglobulin classes; Fc receptors for IgG
antibodies are referred to as Fc.gamma.R, for IgE as Fc.epsilon.R,
for IgA as Fc.alpha.R and so on and neonatal Fc receptors are
referred to as FcRn. Fc receptor binding is described for example
in Ravetch, J. V., and Kinet, J. P., Annu. Rev. Immunol. 9 (1991)
457-492; Capel, P. J., et al., Immunomethods 4 (1994) 25-34; de
Haas, M., et al., J Lab. Clin. Med. 126 (1995) 330-341; and
Gessner, J. E., et al., Ann. Hematol. 76 (1998) 231-248.
[0065] Cross-linking of receptors by the Fc domain of native IgG
antibodies (Fc.gamma.R) triggers a wide variety of effector
functions including phagocytosis, antibody-dependent cellular
cytotoxicity, and release of inflammatory mediators, as well as
immune complex clearance and regulation of antibody production.
Therefore, Fc moieties providing cross-linking of receptors
(Fc.gamma.R) are preferred. In humans, three classes of Fc.gamma.R
have been characterized, which are: (i) Fc.gamma.RI (CD64), which
binds monomeric IgG with high affinity and is expressed on
macrophages, monocytes, neutrophils and eosinophils; (ii)
Fc.gamma.RII (CD32), which binds complexed IgG with medium to low
affinity, is widely expressed, in particular on leukocytes, is
known to be a central player in antibody-mediated immunity, and
which can be divided into Fc.gamma.RIIA, Fc.gamma.RIIB and
Fc.gamma.RIIC, which perform different functions in the immune
system, but bind with similar low affinity to the IgG-Fc, and the
ectodomains of these receptors are highly homologuous; and (iii)
Fc.gamma.RIII (CD16), which binds IgG with medium to low affinity
and exists as two types: Fc.gamma.RIIIA found on NK cells,
macrophages, eosinophils and some monocytes and T cells and
mediating ADCC and Fc.gamma.RIIIB, which is highly expressed on
neutrophils. Fc.gamma.RIIA is found on many cells involved in
killing (e.g. macrophages, monocytes, neutrophils) and seems able
to activate the killing process. Fc.gamma.RIIB seems to play a role
in inhibitory processes and is found on B-cells, macrophages and on
mast cells and eosinophils. Importantly, 75% of all Fc.gamma.RIIB
is found in the liver (Ganesan, L. P. et al., 2012: Fc.gamma.RIIb
on liver sinusoidal endothelium clears small immune complexes.
Journal of Immunology 189: 4981-4988). Fc.gamma.RIIB is abundantly
expressed on Liver Sinusoidal Endothelium, called LSEC, and in
Kupffer cells in the liver and LSEC are the major site of small
immune complexes clearance (Ganesan, L. P. et al., 2012:
Fc.gamma.RIIb on liver sinusoidal endothelium clears small immune
complexes. Journal of Immunology 189: 4981-4988).
[0066] Accordingly, in the present invention such antibodies, and
antigen binding fragments thereof, are preferred, which are able to
bind to Fc.gamma.RIIb, for example antibodies comprising an Fc
moiety for binding to Fc.gamma.RIIb, in particular an Fc region,
such as, for example IgG-type antibodies. Moreover, it is possible
to engineer the Fc moiety to enhance Fc.gamma.RIIB binding by
introducing the mutations S267E and L328F as described by Chu, S.
Y. et al., 2008: Inhibition of B cell receptor-mediated activation
of primary human B cells by coengagement of CD19 and FcgammaRIIb
with Fc-engineered antibodies. Molecular Immunology 45, 3926-3933.
Thereby, the clearance of immune complexes can be enhanced (Chu,
S., et al., 2014: Accelerated Clearance of IgE In Chimpanzees Is
Mediated By Xmab7195, An Fc-Engineered Antibody With Enhanced
Affinity For Inhibitory Receptor Fc.gamma.RIIb. Am J Respir Crit,
American Thoracic Society International Conference Abstracts).
Accordingly, in the context of the present invention such
antibodies, or antigen binding fragments thereof, are preferred,
which comprise an engineered Fc moiety with the mutations S267E and
L328F, in particular as described by Chu, S. Y. et al., 2008:
Inhibition of B cell receptor-mediated activation of primary human
B cells by coengagement of CD19 and FcgammaRIIb with Fc-engineered
antibodies. Molecular Immunology 45, 3926-3933.
[0067] On B-cells it seems to function to suppress further
immunoglobulin production and isotype switching to say for example
the IgE class. On macrophages, Fc.gamma.RIIB acts to inhibit
phagocytosis as mediated through Fc.gamma.RIIA. On eosinophils and
mast cells the b form may help to suppress activation of these
cells through IgE binding to its separate receptor.
[0068] Regarding Fc.gamma.RI binding, modification in native IgG of
at least one of E233-G236, P238, D265, N297, A327 and P329 reduces
binding to Fc.gamma.RI. IgG2 residues at positions 233-236,
substituted into IgG1 and IgG4, reduces binding to Fc.gamma.RI by
10.sup.3-fold and eliminated the human monocyte response to
antibody-sensitized red blood cells (Armour, K. L., et al. Eur. J.
Immunol. 29 (1999) 2613-2624). Regarding Fc.gamma.RII binding,
reduced binding for Fc.gamma.RIIA is found e.g. for IgG mutation of
at least one of E233-G236, P238, D265, N297, A327, P329, D270,
Q295, A327, R292 and K414. Regarding Fc.gamma.RIII binding, reduced
binding to Fc.gamma.RIIIA is found e.g. for mutation of at least
one of E233-G236, P238, D265, N297, A327, P329, D270, Q295, A327,
S239, E269, E293, Y296, V303, A327, K338 and D376. Mapping of the
binding sites on human IgG1 for Fc receptors, the above mentioned
mutation sites and methods for measuring binding to Fc.gamma.RI and
Fc.gamma.RIIA are described in Shields, R. L., et al., J. Biol.
Chem. 276 (2001) 6591-6604.
[0069] Regarding binding to the crucial Fc.gamma.RII, two regions
of native IgG Fc appear to be critical for interactions of
Fc.gamma.RIIs and IgGs, namely (i) the lower hinge site of IgG Fc,
in particular amino acid residues L, L, G, G (234-237, EU
numbering), and (ii) the adjacent region of the CH2 domain of IgG
Fc, in particular a loop and strands in the upper CH2 domain
adjacent to the lower hinge region, e.g. in a region of P331
(Wines, B. D., et al., J. Immunol. 2000; 164: 5313-5318). Moreover,
Fc.gamma.RI appears to bind to the same site on IgG Fc, whereas
FcRn and Protein A bind to a different site on IgG Fc, which
appears to be at the CH2-CH3 interface (Wines, B. D., et al., J.
Immunol. 2000; 164: 5313-5318).
[0070] For example, the Fc moiety may comprise or consist of at
least the portion of an Fc moiety that is known in the art to be
required for FcRn binding or extended half-life. Alternatively or
additionally, the Fc moiety of the antibody of the invention
comprises at least the portion of known in the art to be required
for Protein A binding and/or the Fc moiety of the antibody of the
invention comprises at least the portion of an Fc molecule known in
the art to be required for protein G binding. Preferably, the
retained function is the neutralization of Zika virus infection,
which is assumed to be mediated by Fc.gamma.R binding. Accordingly,
a preferred Fc moiety comprises at least the portion known in the
art to be required for Fc.gamma.R binding. As outlined above, a
preferred Fc moiety may thus at least comprise (i) the lower hinge
site of native IgG Fc, in particular amino acid residues L, L, G, G
(234-237, EU numbering), and (ii) the adjacent region of the CH2
domain of native IgG Fc, in particular a loop and strands in the
upper CH2 domain adjacent to the lower hinge region, e.g. in a
region of P331, for example a region of at least 3, 4, 5, 6, 7, 8,
9, or 10 consecutive amino acids in the upper CH2 domain of native
IgG Fc around P331, e.g. between amino acids 320 and 340 (EU
numbering) of native IgG Fc.
[0071] Preferably, the antibody, or antigen binding fragment
thereof, according to the present invention comprises an Fc region.
As used herein, the term "Fc region" refers to the portion of an
immunoglobulin formed by two or more Fc moieties of antibody heavy
chains. For example, the Fc region may be monomeric or
"single-chain" Fc region (i.e., a scFc region). Single chain Fc
regions are comprised of Fc moieties linked within a single
polypeptide chain (e.g., encoded in a single contiguous nucleic
acid sequence). Exemplary scFc regions are disclosed in WO
2008/143954 A2. Preferably, the Fc region is a dimeric Fc region. A
"dimeric Fc region" or "dcFc" refers to the dimer formed by the Fc
moieties of two separate immunoglobulin heavy chains. The dimeric
Fc region may be a homodimer of two identical Fc moieties (e.g., an
Fc region of a naturally occurring immunoglobulin) or a heterodimer
of two non-identical Fc moieties.
[0072] The Fc moieties of the Fc region may be of the same or
different class and/or subclass. For example, the Fc moieties may
be derived from an immunoglobulin (e.g., a human immunoglobulin) of
an IgG1, IgG2, IgG3 or IgG4 subclass. Preferably, the Fc moieties
of Fc region are of the same class and subclass. However, the Fc
region (or one or more Fc moieties of an Fc region) may also be
chimeric, whereby a chimeric Fc region may comprise Fc moieties
derived from different immunoglobulin classes and/or subclasses.
For example, at least two of the Fc moieties of a dimeric or
single-chain Fc region may be from different immunoglobulin classes
and/or subclasses. Additionally or alternatively, the chimeric Fc
regions may comprise one or more chimeric Fc moieties. For example,
the chimeric Fc region or moiety may comprise one or more portions
derived from an immunoglobulin of a first subclass (e.g., an IgG1,
IgG2, or IgG3 subclass) while the remainder of the Fc region or
moiety is of a different subclass. For example, an Fc region or
moiety of an Fc polypeptide may comprise a CH2 and/or CH3 domain
derived from an immunoglobulin of a first subclass (e.g., an IgG1,
IgG2 or IgG4 subclass) and a hinge region from an immunoglobulin of
a second subclass (e.g., an IgG3 subclass). For example, the Fc
region or moiety may comprise a hinge and/or CH2 domain derived
from an immunoglobulin of a first subclass (e.g., an IgG4 subclass)
and a CH3 domain from an immunoglobulin of a second subclass (e.g.,
an IgG1, IgG2, or IgG3 subclass). For example, the chimeric Fc
region may comprise an Fc moiety (e.g., a complete Fc moiety) from
an immunoglobulin for a first subclass (e.g., an IgG4 subclass) and
an Fc moiety from an immunoglobulin of a second subclass (e.g., an
IgG1, IgG2 or IgG3 subclass). For example, the Fc region or moiety
may comprise a CH2 domain from an IgG4 immunoglobulin and a CH3
domain from an IgG1 immunoglobulin. For example, the Fc region or
moiety may comprise a CH1 domain and a CH2 domain from an IgG4
molecule and a CH3 domain from an IgG1 molecule. For example, the
Fc region or moiety may comprise a portion of a CH2 domain from a
particular subclass of antibody, e.g., EU positions 292-340 of a
CH2 domain. For example, an Fc region or moiety may comprise amino
acids a positions 292-340 of CH2 derived from an IgG4 moiety and
the remainder of CH2 derived from an IgG1 moiety (alternatively,
292-340 of CH2 may be derived from an IgG1 moiety and the remainder
of CH2 derived from an IgG4 moiety).
[0073] Moreover, an Fc region or moiety may (additionally or
alternatively) for example comprise a chimeric hinge region. For
example, the chimeric hinge may be derived, e.g. in part, from an
IgG1, IgG2, or IgG4 molecule (e.g., an upper and lower middle hinge
sequence) and, in part, from an IgG3 molecule (e.g., an middle
hinge sequence). In another example, an Fc region or moiety may
comprise a chimeric hinge derived, in part, from an IgG1 molecule
and, in part, from an IgG4 molecule. In another example, the
chimeric hinge may comprise upper and lower hinge domains from an
IgG4 molecule and a middle hinge domain from an IgG1 molecule. Such
a chimeric hinge may be made, for example, by introducing a proline
substitution (Ser228Pro) at EU position 228 in the middle hinge
domain of an IgG4 hinge region. In another embodiment, the chimeric
hinge can comprise amino acids at EU positions 233-236 are from an
IgG2 antibody and/or the Ser228Pro mutation, wherein the remaining
amino acids of the hinge are from an IgG4 antibody (e.g., a
chimeric hinge of the sequence ESKYGPPCPPCPAPPVAGP (SEQ ID NO:
408)). Further chimeric hinges, which may be used in the Fc moiety
of the antibody according to the present invention are described in
US 2005/0163783 A1.
[0074] In the present invention it is preferred that the Fc moiety,
or the Fc region, comprises or consists of an amino acid sequence
derived from a human immunoglobulin sequence (e.g., from an Fc
region or Fc moiety from a human IgG molecule). However,
polypeptides may comprise one or more amino acids from another
mammalian species. For example, a primate Fc moiety or a primate
binding site may be included in the subject polypeptides.
Alternatively, one or more murine amino acids may be present in the
Fc moiety or in the Fc region.
[0075] Preferably, the antibody according to the present invention
comprises, in particular in addition to an Fc moiety as described
above, other parts derived from a constant region, in particular
from a constant region of IgG, preferably from a constant region of
IgG1, more preferably from a constant region of human IgG1. More
preferably, the antibody according to the present invention
comprises, in particular in addition to an Fc moiety as described
above, all other parts of the constant regions, in particular all
other parts of the constant regions of IgG, preferably all other
parts of the constant regions of IgG1, more preferably all other
parts of the constant regions of human IgG1.
[0076] Particularly preferred sequences of constant regions are the
amino acid sequences according to SEQ ID NOs: 145-148 (nucleic acid
sequences according to SEQ ID NOs: 149-152). Preferably, the amino
acid sequence of IgG1 CH1-CH2-CH3 is according to SEQ ID NO: 145 or
a functional sequence variant thereof, as described herein. Even
more preferably, the amino acid sequence of IgG1 CH1-CH2-CH3 is
according to SEQ ID NO: 146 or a functional sequence variant
thereof, as described herein, wherein the "LALA" mutation is
maintained.
[0077] As outlined above, a particularly preferred antibody
according to the present invention comprises a (complete) Fc region
derived from human IgG1. More preferably, the antibody according to
the present invention comprises, in particular in addition to a
(complete) Fc region derived from human IgG1 also all other parts
of the constant regions of IgG, preferably all other parts of the
constant regions of IgG1, more preferably all other parts of the
constant regions of human IgG1.
[0078] Without being bound to any theory, it is believed that
antibody-dependent enhancement (ADE) of Zika virus infection is
brought about by the binding of the Fc moiety of the antibody, in
particular, the Fc moiety of the heavy chain of an IgG molecule, to
an Fc receptor, e.g., an Fc.gamma. receptor on a host cell. It is
thus preferred that the antibody according to the present
invention, or an antigen binding fragment thereof, comprises one or
more mutations in the Fc moiety. The mutation(s) may be any
mutation that reduces binding of the antibody to an Fc receptor
(FcR), in particular reduces binding of the antibody to an
Fc.gamma. receptor (Fc.gamma.R). On the other hand, it is preferred
that the antibody according to the present invention comprises a
(complete) Fc moiety/Fc region, wherein the interaction/binding
with FcRn is not compromised. Accordingly, it is particularly
preferred that the antibody according to the present invention, or
an antigen binding fragment thereof, comprises one or more
mutations in the Fc moiety, which (i) reduce(s) binding of the
antibody to an Fc.gamma. receptor, but do(es) not compromise
interaction with FcRn. One example of such a mutation is the "LALA"
mutation described below.
[0079] In general, binding of the antibody to an Fc receptor may be
assessed by various methods known to the skilled person, such as
ELISA (Hessell A J, Hangartner L, Hunter M, Havenith C E G,
Beurskens F J, Bakker J M, Lanigan C M S, Landucci G, Forthal D N,
Parren PWHI, et al.: Fc receptor but not complement binding is
important in antibody protection against HIV. Nature 2007,
449:101-104; Grevys A, Bern M, Foss S, Bratlie D B, Moen A,
Gunnarsen K S, Aase A, Michaelsen T E, Sandlie I, Andersen J T: Fc
Engineering of Human IgG1 for Altered Binding to the Neonatal Fc
Receptor Affects Fc Effector Functions. 2015, 194:5497-5508) or
flow-cytometry (Perez L G, Costa M R, Todd C A, Haynes B F,
Montefiori D C: Utilization of immunoglobulin G Fc receptors by
human immunodeficiency virus type 1: a specific role for antibodies
against the membrane-proximal external region of gp41. J Virol
2009, 83:7397-7410; Piccoli L, Campo I, Fregni C S, Rodriguez B M
F, Minola A, Sallusto F, Luisetti M, Corti D, Lanzavecchia A:
Neutralization and clearance of GM-CSF by autoantibodies in
pulmonary alveolar proteinosis. Nat Commun 2015, 6:1-9).
[0080] In general, the antibody according to the present invention
may be glycosylated. N-linked glycans attached to the CH2 domain of
a heavy chain, for instance, can influence C1q and FcR binding,
with aglycosylated antibodies having lower affinity for these
receptors. Accordingly, the CH2 domain of the Fc moiety of the
antibody according to the present invention may comprise one or
more mutations, in which a glycosylated residue is substituted by a
non-glycosylated residue. The glycan structure can also affect
activity e.g. differences in complement-mediated cell death may be
seen depending on the number of galactose sugars (0, 1 or 2) at the
terminus of a glycan's biantennary chain. Preferably, the
antibody's glycans do not lead to a human immunogenic response
after administration.
[0081] Furthermore, the antibody according to the present invention
can be modified by introducing random amino acid mutations into
particular region of the CH2 or CH3 domain of the heavy chain in
order to alter their binding affinity for FcR and/or their serum
half-life in comparison to unmodified antibodies. Examples of such
modifications include, but are not limited to, substitutions of at
least one amino acid from the heavy chain constant region selected
from the group consisting of amino acid residues 250, 314, and
428.
[0082] Particularly preferably, the Fc moiety of an antibody of the
invention comprises a substitution at positions CH24, CH25, or
both. In general, the amino acid at positions 4 and 5 of CH2 of the
wild-type IgG1 and IgG3 is a leucine ("L"). Preferably, the
antibody according to the present invention comprises an amino acid
at position CH24, CH25, or both, that is not an L. More preferably,
antibody according to the present invention comprises an alanine
("A") at position CH24, or CH2 5, or both. Most preferably, the
antibody according to the present invention comprises both, a CH2
L4A and a CH2 L5A substitution. Such antibodies are referred to
herein as a "LALA" variant. Interestingly, such a "LALA" mutation
in the Fc moiety does not only result in a lack of contribution of
the respective antibody in antibody-dependent enhancement (ADE) of
Zika virus infection, but also blocks antibody-dependent
enhancement (ADE) of Zika virus infection. An exemplary amino acid
sequence of IgG1 CH1-CH2-CH3 comprising the "LALA" mutation is
according to SEQ ID NO: 146. Accordingly, the amino acid sequence
of IgG1 CH1-CH2-CH3 is preferably according to SEQ ID NO: 146 or a
functional sequence variant thereof, as described herein, wherein
the "LALA" mutation is maintained.
[0083] Preferably, the antibody, or antigen binding fragment
thereof, binds to domain III of Zika virus envelope protein (EDIII,
also referred to as "DIII"). In other words, it is preferred that
the, the antibody, or antigen binding fragment thereof, according
to the present invention binds to an epitope of Zika virus envelope
protein, which includes one or more amino acid residues of domain
III of Zika virus envelope protein (EDIII). ZIKV includes a
nucleocapsid core, which comprising single-stranded RNA wrapped by
core proteins. The nucleocapsid core is encapsulated by a lipid
bilayer membrane with "membrane proteins" and "envelope proteins".
ZIKV envelope protein (E protein) is the dominant antigen. The
ectodomain of the envelope protein comprises three distinct
domains: E protein domain I (EDI), E protein domain II (EDII), and
E protein domain III (EDIII). EDIII is highly conserved among
different ZIKV strains (see FIG. 12 for an alignment of amino acid
sequences of EDIII of different ZIKV strains.
[0084] Accordingly, the antibody, or antigen binding fragment
thereof, more preferably binds to domain III of Zika virus envelope
protein (EDIII) with EDIII having the following amino acid sequence
(SEQ ID NO: 401):
TABLE-US-00001 TAAFTFTKXPAEXXHGTVTVEXQYXGXDGPCKXPXQMAVDXQTLTPVG
RLITANPVITEXTENSKMMLELDPPFGDSYIVIGXGXKKITHHWHRS
[0085] wherein X may be any (naturally occurring) amino acid. In
other words, it is preferred that the, the antibody, or antigen
binding fragment thereof, according to the present invention binds
to an epitope of Zika virus envelope protein, which includes one or
more amino acid residues of SEQ ID NO: 401.
[0086] It is also preferred that the antibody, or antigen binding
fragment thereof, according to the present invention binds to
domain III of Zika virus envelope protein (EDIII) with EDIII having
the following amino acid sequence (SEQ ID NO: 407):
TABLE-US-00002
X.sub.1GX.sub.2X.sub.3YSLCTAAFTFTKX.sub.4PAEX.sub.5X.sub.6HGTVTVEX.sub.7Q-
YX.sub.8GX.sub.9DG
PCKX.sub.10PX.sub.11QMAVDX.sub.12QTLTPVGRLITANPVITEX.sub.13TX.sub.14NSK
MMLELDPPFGDSYIVIGX.sub.15GX.sub.16X.sub.17KITHHWHRSG
[0087] wherein X1 may be any (naturally occurring) amino acid,
preferably K, A, or E; [0088] X2 may be any (naturally occurring)
amino acid, preferably V, F, or L; [0089] X3 may be any (naturally
occurring) amino acid, preferably S or F; [0090] X4 may be any
(naturally occurring) amino acid, preferably I or V; [0091] X5 may
be any (naturally occurring) amino acid, preferably T or V; [0092]
X6 may be any (naturally occurring) amino acid, preferably L or D;
[0093] X7 may be any (naturally occurring) amino acid, preferably V
or G; [0094] X8 may be any (naturally occurring) amino acid,
preferably A or G; [0095] X9 may be any (naturally occurring) amino
acid except R, preferably T or A; [0096] X10 may be any (naturally
occurring) amino acid, preferably V or I; [0097] X11 may be any
(naturally occurring) amino acid, preferably A or V; [0098] X12 may
be any (naturally occurring) amino acid, preferably M or T; [0099]
X13 may be any (naturally occurring) amino acid, preferably S or G;
[0100] X14 may be any (naturally occurring) amino acid, preferably
E or K; [0101] X15 may be any (naturally occurring) amino acid,
preferably V or I; [0102] X16 may be any (naturally occurring)
amino acid, preferably E, A, K, or D; and [0103] X17 may be any
(naturally occurring) amino acid, preferably E, A, or K, more
preferably K or A.
[0104] In other words, it is preferred that the, the antibody, or
antigen binding fragment thereof, according to the present
invention binds to an epitope of Zika virus envelope protein, which
includes one or more amino acid residues of SEQ ID NO: 407.
[0105] For example, EDIII stretches from amino acid 309 to amino
acid 403 of ZIKV E protein of the ZIKV H/PF/2013 strain (Genbank
accession number KJ776791). Accordingly, the antibody, or antigen
binding fragment thereof, most preferably binds to domain III of
Zika virus envelope protein (EDIII) with EDIII having the following
amino acid sequence (SEQ ID NO: 402):
TABLE-US-00003 TAAFTFTKIPAETLHGTVTVEVQYAGTDGPCKVPAQMAVDMQTLTP
VGRLITANPVITESTENSKMMLELDPPFGDSYIVIGVGEKKITHHW HRS.
[0106] In other words, it is preferred that the, the antibody, or
antigen binding fragment thereof, according to the present
invention binds to an epitope of Zika virus envelope protein, which
includes one or more amino acid residues of SEQ ID NO: 402.
[0107] Surprisingly, the present inventors have found that
antibodies binding to domain III of Zika virus envelope protein
(EDIII) show (i) increased neutralization of ZIKV and (ii)
decreased cross-reactivity with DENV (in particular essentially no
cross-reactivity with DENV) as compared to antibodies binding to
domain I/II of Zika virus envelope protein (EDI/II).
[0108] More preferably, the antibody, or antigen binding fragment
thereof, according to the present invention binds to an epitope of
Zika virus envelope protein, which includes one or more amino acid
residues of the lateral ridge (LR) of EDIII and/or one or more
amino acid residues of the EDI-EDIII hinge region. The EDIII
lateral ridge and EDI-EDIII hinge region are known to the skilled
person and described, for example, in Zhao, H., Fernandez, E.,
Dowd, K. A., Speer, S. D., Platt, D. J., Gorman, M. J., Govero, J.,
Nelson, C. A., Pierson, T. C., Diamond, M. S., et al. (2016).
Structural Basis of Zika Virus-Specific Antibody Protection. Cell
166(4):1016-27 and in Kostyuchenko V A, Lim E X, Zhang S,
Fibriansah G, Ng T S, Ooi J S, Shi J, Lok S M. Structure of the
thermally stable Zika virus. Nature. 2016 May 19; 533(7603):425-8.
Without being bound to any theory, it is assumed that (i) binding
to the LR may inhibit fusion by trapping a fusion transitional
state of the virus and (ii) binding to the EDI-EDIII hinge and
EDIII may hinder the movement of EDIII to form the trimeric
post-fusion structure, thereby halting membrane fusion.
[0109] Accordingly, it is preferred that the antibody, or antigen
binding fragment thereof, according to the present invention (is
able to) inhibit(s) a post-attachment step of ZIKV.
"Post-attachment" typically refers to any step of ZIKV infection
after attachment of ZIKV to the cell membrane (of the cell targeted
by ZIKV). For example, the antibody, or antigen binding fragment
thereof, according to the present invention preferably (is able to)
prevent(s) membrane fusion. Furthermore, it is also preferred that
the antibody, or antigen binding fragment thereof, according to the
present invention (is able to) cause(s) aggregation of ZIKV
(particles). Most preferably, the antibody, or antigen binding
fragment thereof, according to the present invention (is able to)
(i) inhibit(s) a post-attachment step of ZIKV and (ii) cause(s)
aggregation of ZIKV (particles). It is also preferred that the
antibody, or antigen binding fragment thereof, binds to a
quaternary epitope displayed on a ZIKV infectious virion. Despite
considerable neutralizing activity, such antibodies show typically
no detectable binding to recombinant ZIKV E protein or to ZIKV
EDIII in a standard ELISA (as described above), i.e. if tested in
vitro, in particular in purified form (i.e. ZIKV E protein
"outside/without" a virion, a virus-like particle or the like).
Thereby, "no detectable binding" typically means that no EC.sub.50
up to 10000 ng/ml was detected in a standard ELISA. In other words,
if the EC.sub.50 detectable in a standard ELISA is above 10000
ng/ml, it is referred to as "no detectable binding".
[0110] Therefore, such antibodies are also referred to herein as
"neutralizing-non-E-binding" (NNB) antibodies. The quaternary
epitope displayed on a ZIKV infectious virion is typically a
conformational epitope. For example, the quaternary epitope
displayed on a ZIKV infectious virion may be formed at the
interface of two envelope protein monomers making up a dimer
("envelope dimer epitope"; EDE) or it may be formed across
neighbouring dimers ("herring-bone epitope").
[0111] In general, the antibody according to the present invention,
or the antigen binding fragment thereof, preferably comprises (at
least) three complementarity determining regions (CDRs) on a heavy
chain and (at least) three CDRs on a light chain. In general,
complementarity determining regions (CDRs) are the hypervariable
regions present in heavy chain variable domains and light chain
variable domains. Typically, the CDRs of a heavy chain and the
connected light chain of an antibody together form the antigen
receptor. Usually, the three CDRs (CDR1, CDR2, and CDR3) are
arranged non-consecutively in the variable domain. Since antigen
receptors are typically composed of two variable domains (on two
different polypeptide chains, i.e. heavy and light chain), there
are six CDRs for each antigen receptor (heavy chain: CDRH1, CDRH2,
and CDRH3; light chain: CDRL1, CDRL2, and CDRL3). A single antibody
molecule usually has two antigen receptors and therefore contains
twelve CDRs. The CDRs on the heavy and/or light chain may be
separated by framework regions, whereby a framework region (FR) is
a region in the variable domain which is less "variable" than the
CDR. For example, a chain (or each chain, respectively) may be
composed of four framework regions, separated by three CDR's.
[0112] The sequences of the heavy chains and light chains of
exemplary antibodies of the invention, comprising three different
CDRs on the heavy chain and three different CDRs on the light chain
were determined. The position of the CDR amino acids are defined
according to the IMGT numbering system (IMGT: http://www.imgt.org/;
cf. Lefranc, M.-P. et al. (2009) Nucleic Acids Res. 37,
D1006-D1012).
[0113] Table 1 shows the SEQ ID NO's of the amino acid sequences of
the heavy chain CDR's (CDRH1, CDRH2, and CDRH3) and of the heavy
chain variable region (referred to as "VH") of exemplary antibodies
according to the present invention:
TABLE-US-00004 Antibody name CDRH1 CDRH2 CDRH3 VH ZKA190 1 2 3 8
ZKA185 19 20 21 26 ZKA230 37 38 39 44 ZKA78 55 56 57 62 ZKA64 73 74
75 80 ZKA3 237 238 239 240 ZKA4 241 242 243 244 ZKA5 245 246 247
248 ZKA6 249 250 251 252 ZKA7 253 254 255 256 ZKA8 257 258 259 260
ZKA76 261 262 263 264 ZKA117 265 266 267 268 ZKB27 269 270 271 272
ZKB29 273 274 275 276 ZKB34 277 278 279 280 ZKB39 281 282 283 284
ZKB46 285 286 287 288 ZKB53 289 290 291 292 ZKC26 293 294 295 296
ZKD5 297 298 299 300 ZKD7 301 302 303 304 ZKD8 305 306 307 308
ZKD15 309 310 311 312 ZKD16 313 314 315 316 ZKD17 317 318 319 320
ZKD20 321 322 323 324 ZKA134 325 326 327 328 ZKA246 329 330 331 332
ZKA256 333 334 335 336 ZKB42 337 338 339 340 ZKB85 341 342 343 344
ZKB47 345 346 347 348 ZKC6 349 350 351 352 ZKA160 353 354 355 356
ZKA172 357 358 359 360 ZKA174 361 362 363 364 ZKA189 365 366 367
368 ZKA195 369 370 371 372 ZKA215 373 374 375 376 ZKA218 377 378
379 380 ZKB75 381 382 383 384 ZKB83 385 386 387 388 ZKC3 389 390
391 392 ZKC18 393 394 395 396 ZKD1 397 398 399 400
[0114] Table 2 below shows the SEQ ID NO's of the amino acid
sequences of the light chain CDR's (CDRL1, CDRL2, and CDRL3) and of
the light chain variable region (referred to as "VL") of exemplary
antibodies according to the present invention:
TABLE-US-00005 Antibody CDRL.sub.2 name CDRL.sub.1 CDRL.sub.2 long
CDRL.sub.3 VL ZKA190 4 5 6 7 9 ZKA185 22 23 24 25 27 ZKA230 40 41
42 43 45 ZKA78 58 59 60 61 63 ZKA64 76 77 78 79 81
[0115] It is thus preferred that the antibody, or the antigen
binding fragment thereof, according to the present invention
comprises amino acid sequences having at least 70%, at least 75%,
at least 80%, at least 85%, at least 88%, at least 90%, at least
92%, at least 95%, at least 96%, at least 97%, at least 98% or at
least 99% identity to at least one of the CDR sequences, the VH
sequence and/or the VL sequence shown in Table 1 and/or in Table
2.
[0116] It is preferred that the antibody, or the antigen binding
fragment thereof, according to the present invention comprises a
heavy chain comprising at least one CDRH1, at least one CDRH2 and
at least one CDRH3 and a light chain comprising at least one CDRL1,
at least one CDRL2 and at least one CDRL3, wherein at least one
CDR, preferably the at least one heavy chain CDRH3, comprises or
consists of an amino acid sequence according to any of SEQ ID NOs:
3, 75, 39, 21, 57, 239, 243, 247, 251, 255, 259, 263, 267, 271,
275, 279, 283, 287, 291, 295, 299, 303, 307, 311, 315, 319, 323,
327, 331, 335, 339, 343, 347, 351, 355, 359, 363, 367, 371, 375,
379, 383, 387, 391, 395, and 399, or a functional sequence variant
thereof having at least 70%, at least 75%, at least 80%, at least
85%, at least 88%, at least 90%, at least 92%, at least 95%, at
least 96%, at least 97%, at least 98% or at least 99% sequence
identity.
[0117] More preferably, the antibody, or the antigen binding
fragment thereof, according to the present invention comprises a
heavy chain comprising at least one CDRH1, at least one CDRH2 and
at least one CDRH3 and a light chain comprising at least one CDRL1,
at least one CDRL2 and at least one CDRL3, wherein at least one
CDR, preferably the at least one heavy chain CDRH3, comprises or
consists of an amino acid sequence according to any of SEQ ID NOs:
3, 21, 39, 57 and 75 or a functional sequence variant thereof
having at least 70%, at least 75%, at least 80%, at least 85%, at
least 88%, at least 90%, at least 92%, at least 95%, at least 96%,
at least 97%, at least 98% or at least 99% sequence identity. More
preferably, the antibody, or the antigen binding fragment thereof,
according to the present invention comprises a heavy chain
comprising at least one CDRH1, at least one CDRH2 and at least one
CDRH3 and a light chain comprising at least one CDRL1, at least one
CDRL2 and at least one CDRL3, wherein at least one CDR, preferably
the at least one heavy chain CDRH3, comprises or consists of an
amino acid sequence according to any of SEQ ID NOs: 3, 21, 39 and
75 or a functional sequence variant thereof having at least 70%, at
least 75%, at least 80%, at least 85%, at least 88%, at least 90%,
at least 92%, at least 95%, at least 96%, at least 97%, at least
98% or at least 99% sequence identity. Even more preferably, the
antibody, or the antigen binding fragment thereof, according to the
present invention comprises a heavy chain comprising at least one
CDRH1, at least one CDRH2 and at least one CDRH3 and a light chain
comprising at least one CDRL1, at least one CDRL2 and at least one
CDRL3, wherein at least one CDR, preferably the at least one heavy
chain CDRH3, comprises or consists of an amino acid sequence
according to SEQ ID NO: 3 or according to SEQ ID NO: 75; or a
functional sequence variant thereof having at least 70%, at least
75%, at least 80%, at least 85%, at least 88%, at least 90%, at
least 92%, at least 95%, at least 96%, at least 97%, at least 98%
or at least 99% sequence identity. It is also preferred that the
antibody, or the antigen binding fragment thereof, according to the
present invention comprises a heavy chain comprising at least one
CDRH1, at least one CDRH2 and at least one CDRH3 and a light chain
comprising at least one CDRL1, at least one CDRL2 and at least one
CDRL3, wherein at least one CDR, preferably the at least one heavy
chain CDRH3, comprises or consists of an amino acid sequence
according to SEQ ID NO: 21 or according to SEQ ID NO: 39; or a
functional sequence variant thereof having at least 70%, at least
75%, at least 80%, at least 85%, at least 88%, at least 90%, at
least 92%, at least 95%, at least 96%, at least 97%, at least 98%
or at least 99% sequence identity. Most preferably, the antibody,
or the antigen binding fragment thereof, according to the present
invention comprises a heavy chain comprising at least one CDRH1, at
least one CDRH2 and at least one CDRH3 and a light chain comprising
at least one CDRL1, at least one CDRL2 and at least one CDRL3,
wherein at least one CDR, preferably the at least one heavy chain
CDRH3, comprises or consists of an amino acid sequence according to
SEQ ID NO: 3 or a functional sequence variant thereof having at
least 70%, at least 75%, at least 80%, at least 85%, at least 88%,
at least 90%, at least 92%, at least 95%, at least 96%, at least
97%, at least 98% or at least 99% sequence identity.
[0118] More preferably, the antibody, or the antigen binding
fragment thereof, according to the present invention comprises a
heavy chain comprising at least one CDRH1, at least one CDRH2 and
at least one CDRH3 and a light chain comprising at least one CDRL1,
at least one CDRL2 and at least one CDRL3, wherein [0119] (i) the
at least one heavy chain CDRH1 comprises an amino acid sequence
according to any of SEQ ID NOs: 1, 19, 37, 55, 73, 237, 241, 245,
249, 253, 257, 261, 265, 269, 273, 277, 281, 285, 289, 293, 297,
301, 305, 309, 313, 317, 321, 325, 329, 333, 337, 341, 345, 349,
353, 357, 361, 365, 369, 373, 377, 381, 385, 389, 393, and 397, or
a functional sequence variant thereof having at least 70%, at least
75%, at least 80%, at least 85%, at least 88%, at least 90%, at
least 92%, at least 95%, at least 96%, at least 97%, at least 98%
or at least 99% sequence identity; [0120] (ii) the at least one
CDRH2 comprises an amino acid sequence according to any of SEQ ID
NOs: 2, 20, 38, 56, 74, 238, 242, 246, 250, 254, 258, 262, 266,
270, 274, 278, 282, 286, 290, 294, 298, 302, 306, 310, 314, 318,
322, 326, 330, 334, 338, 342, 346, 350, 354, 358, 362, 366, 370,
374, 378, 382, 386, 390, 394, and 398, or a functional sequence
variant thereof having at least 70%, at least 75%, at least 80%, at
least 85%, at least 88%, at least 90%, at least 92%, at least 95%,
at least 96%, at least 97%, at least 98% or at least 99% sequence
identity; and/or [0121] (iii) the at least one heavy chain CDRH3
comprises an amino acid sequence according to any of SEQ ID NOs: 3,
21, 39, 57, 75, 239, 243, 247, 251, 255, 259, 263, 267, 271, 275,
279, 283, 287, 291, 295, 299, 303, 307, 311, 315, 319, 323, 327,
331, 335, 339, 343, 347, 351, 355, 359, 363, 367, 371, 375, 379,
383, 387, 391, 395, and 399, or a functional sequence variant
thereof having at least 70%, at least 75%, at least 80%, at least
85%, at least 88%, at least 90%, at least 92%, at least 95%, at
least 96%, at least 97%, at least 98% or at least 99% sequence
identity.
[0122] Still more preferably, the antibody, or the antigen binding
fragment thereof, according to the present invention comprises a
heavy chain comprising at least one CDRH1, at least one CDRH2 and
at least one CDRH3 and a light chain comprising at least one CDRL1,
at least one CDRL2 and at least one CDRL3, wherein [0123] (i) the
at least one heavy chain CDRH1 comprises an amino acid sequence
according to any of SEQ ID NOs: 1, 19, 37, 55 and 73 or a
functional sequence variant thereof having at least 70%, at least
75%, at least 80%, at least 85%, at least 88%, at least 90%, at
least 92%, at least 95%, at least 96%, at least 97%, at least 98%
or at least 99% sequence identity; [0124] (ii) the at least one
CDRH2 comprises an amino acid sequence according to any of SEQ ID
NOs: 2, 20, 38, 56 and 74 or a functional sequence variant thereof
having at least 70%, at least 75%, at least 80%, at least 85%, at
least 88%, at least 90%, at least 92%, at least 95%, at least 96%,
at least 97%, at least 98% or at least 99% sequence identity;
and/or [0125] (iii) the at least one heavy chain CDRH3 comprises an
amino acid sequence according to any of SEQ ID NOs: 3, 21, 39, 57
and 75 or a functional sequence variant thereof having at least
70%, at least 75%, at least 80%, at least 85%, at least 88%, at
least 90%, at least 92%, at least 95%, at least 96%, at least 97%,
at least 98% or at least 99% sequence identity.
[0126] Even more preferably, the antibody, or the antigen binding
fragment thereof, according to the present invention comprises a
heavy chain comprising at least one CDRH1, at least one CDRH2 and
at least one CDRH3 and a light chain comprising at least one CDRL1,
at least one CDRL2 and at least one CDRL3, wherein [0127] (i) the
at least one heavy chain CDRH1 comprises an amino acid sequence
according to any of SEQ ID NOs: 1, 19, 37 and 73 or a functional
sequence variant thereof having at least 70%, at least 75%, at
least 80%, at least 85%, at least 88%, at least 90%, at least 92%,
at least 95%, at least 96%, at least 97%, at least 98% or at least
99% sequence identity; [0128] (ii) the at least one CDRH2 comprises
an amino acid sequence according to any of SEQ ID NOs: 2, 20, 38
and 74 or a functional sequence variant thereof having at least
70%, at least 75%, at least 80%, at least 85%, at least 88%, at
least 90%, at least 92%, at least 95%, at least 96%, at least 97%,
at least 98% or at least 99% sequence identity; and/or [0129] (iii)
the at least one heavy chain CDRH3 comprises an amino acid sequence
according to any of SEQ ID NOs: 3, 21, 39 and 75 or a functional
sequence variant thereof having at least 70%, at least 75%, at
least 80%, at least 85%, at least 88%, at least 90%, at least 92%,
at least 95%, at least 96%, at least 97%, at least 98% or at least
99% sequence identity.
[0130] Most preferably, the antibody, or the antigen binding
fragment thereof, according to the present invention comprises a
heavy chain comprising at least one CDRH1, at least one CDRH2 and
at least one CDRH3 and a light chain comprising at least one CDRL1,
at least one CDRL2 and at least one CDRL3, wherein [0131] (i) the
at least one heavy chain CDRH1 comprises an amino acid sequence
according to SEQ ID NO: 1 or according to SEQ ID NO: 73; or a
functional sequence variant thereof having at least 70%, at least
75%, at least 80%, at least 85%, at least 88%, at least 90%, at
least 92%, at least 95%, at least 96%, at least 97%, at least 98%
or at least 99% sequence identity; [0132] (ii) the at least one
CDRH2 comprises an amino acid sequence according to SEQ ID NO: 2 or
according to SEQ ID NO: 74; or a functional sequence variant
thereof having at least 70%, at least 75%, at least 80%, at least
85%, at least 88%, at least 90%, at least 92%, at least 95%, at
least 96%, at least 97%, at least 98% or at least 99% sequence
identity; and/or [0133] (iii) the at least one heavy chain CDRH3
comprises an amino acid sequence according to SEQ ID NO: 3 or
according to SEQ ID NO: 75; or a functional sequence variant
thereof having at least 70%, at least 75%, at least 80%, at least
85%, at least 88%, at least 90%, at least 92%, at least 95%, at
least 96%, at least 97%, at least 98% or at least 99% sequence
identity.
[0134] It is also preferred that the antibody, or the antigen
binding fragment thereof, according to the present invention
comprises a heavy chain comprising at least one CDRH1, at least one
CDRH2 and at least one CDRH3 and a light chain comprising at least
one CDRL1, at least one CDRL2 and at least one CDRL3, wherein
[0135] (i) the at least one heavy chain CDRH1 comprises an amino
acid sequence according to SEQ ID NO: 19 or according to SEQ ID NO:
37; or a functional sequence variant thereof having at least 70%,
at least 75%, at least 80%, at least 85%, at least 88%, at least
90%, at least 92%, at least 95%, at least 96%, at least 97%, at
least 98% or at least 99% sequence identity; [0136] (ii) the at
least one CDRH2 comprises an amino acid sequence according to SEQ
ID NO: 20 or according to SEQ ID NO: 38; or a functional sequence
variant thereof having at least 70%, at least 75%, at least 80%, at
least 85%, at least 88%, at least 90%, at least 92%, at least 95%,
at least 96%, at least 97%, at least 98% or at least 99% sequence
identity; and/or [0137] (iii) the at least one heavy chain CDRH3
comprises an amino acid sequence according to SEQ ID NO: 21 or
according to SEQ ID NO: 39; or a functional sequence variant
thereof having at least 70%, at least 75%, at least 80%, at least
85%, at least 88%, at least 90%, at least 92%, at least 95%, at
least 96%, at least 97%, at least 98% or at least 99% sequence
identity.
[0138] Particularly preferably, the antibody, or the antigen
binding fragment thereof, according to the present invention
comprises a heavy chain comprising at least one CDRH1, at least one
CDRH2 and at least one CDRH3 and a light chain comprising at least
one CDRL1, at least one CDRL2 and at least one CDRL3, wherein
[0139] (i) the at least one heavy chain CDRH1 comprises an amino
acid sequence according to SEQ ID NO: 1 or a functional sequence
variant thereof having at least 70%, at least 75%, at least 80%, at
least 85%, at least 88%, at least 90%, at least 92%, at least 95%,
at least 96%, at least 97%, at least 98% or at least 99% sequence
identity; [0140] (ii) the at least one CDRH2 comprises an amino
acid sequence according to SEQ ID NO: 2 or a functional sequence
variant thereof having at least 70%, at least 75%, at least 80%, at
least 85%, at least 88%, at least 90%, at least 92%, at least 95%,
at least 96%, at least 97%, at least 98% or at least 99% sequence
identity; and/or [0141] (iii) the at least one heavy chain CDRH3
comprises an amino acid sequence according to SEQ ID NO: 3 or a
functional sequence variant thereof having at least 70%, at least
75%, at least 80%, at least 85%, at least 88%, at least 90%, at
least 92%, at least 95%, at least 96%, at least 97%, at least 98%
or at least 99% sequence identity.
[0142] It is also preferred that the antibody, or the antigen
binding fragment thereof, according to the present invention
comprises a heavy chain comprising at least one CDRH1, at least one
CDRH2 and at least one CDRH3 and a light chain comprising at least
one CDRL1, at least one CDRL2 and at least one CDRL3, wherein
[0143] (i) the at least one CDRL1 comprises an amino acid sequence
according to any of SEQ ID NOs: 4, 22, 40, 58 and 76 or a
functional sequence variant thereof having at least 70%, at least
75%, at least 80%, at least 85%, at least 88%, at least 90%, at
least 92%, at least 95%, at least 96%, at least 97%, at least 98%
or at least 99% sequence identity; [0144] (ii) the at least one
CDRL2 comprises an amino acid sequence according to any of SEQ ID
NOs: 5, 6, 23, 24, 41, 42, 59, 60, 77 and 78 or a functional
sequence variant thereof having at least 70%, at least 75%, at
least 80%, at least 85%, at least 88%, at least 90%, at least 92%,
at least 95%, at least 96%, at least 97%, at least 98% or at least
99% sequence identity; and/or [0145] (iii) the at least one CDRL3
amino comprises an amino acid sequence according to any of SEQ ID
NOs: 7, 25, 43, 61 and 79 or a functional sequence variant thereof
having at least 70%, at least 75%, at least 80%, at least 85%, at
least 88%, at least 90%, at least 92%, at least 95%, at least 96%,
at least 97%, at least 98% or at least 99% sequence identity.
[0146] More preferably, the antibody, or the antigen binding
fragment thereof, according to the present invention comprises a
heavy chain comprising at least one CDRH1, at least one CDRH2 and
at least one CDRH3 and a light chain comprising at least one CDRL1,
at least one CDRL2 and at least one CDRL3, wherein [0147] (i) the
at least one CDRL1 comprises an amino acid sequence according to
any of SEQ ID NOs: 4, 22, 40 and 76 or a functional sequence
variant thereof having at least 70%, at least 75%, at least 80%, at
least 85%, at least 88%, at least 90%, at least 92%, at least 95%,
at least 96%, at least 97%, at least 98% or at least 99% sequence
identity; [0148] (ii) the at least one CDRL2 comprises an amino
acid sequence according to any of SEQ ID NOs: 5, 6, 23, 24, 41, 42,
77 and 78 or a functional sequence variant thereof having at least
70%, at least 75%, at least 80%, at least 85%, at least 88%, at
least 90%, at least 92%, at least 95%, at least 96%, at least 97%,
at least 98% or at least 99% sequence identity; and/or [0149] (iii)
the at least one CDRL3 amino comprises an amino acid sequence
according to any of SEQ ID NOs: 7, 25, 43 and 79 or a functional
sequence variant thereof having at least 70%, at least 75%, at
least 80%, at least 85%, at least 88%, at least 90%, at least 92%,
at least 95%, at least 96%, at least 97%, at least 98% or at least
99% sequence identity.
[0150] Even more preferably, the antibody, or the antigen binding
fragment thereof, according to the present invention comprises a
heavy chain comprising at least one CDRH1, at least one CDRH2 and
at least one CDRH3 and a light chain comprising at least one CDRL1,
at least one CDRL2 and at least one CDRL3, wherein [0151] (i) the
at least one CDRL1 comprises an amino acid sequence according to
SEQ ID NO: 4 or according to SEQ ID NO: 76; or a functional
sequence variant thereof having at least 70%, at least 75%, at
least 80%, at least 85%, at least 88%, at least 90%, at least 92%,
at least 95%, at least 96%, at least 97%, at least 98% or at least
99% sequence identity; [0152] (ii) the at least one CDRL2 comprises
an amino acid sequence according to any of SEQ ID NOs: 5, 6, 77 and
78 or a functional sequence variant thereof having at least 70%, at
least 75%, at least 80%, at least 85%, at least 88%, at least 90%,
at least 92%, at least 95%, at least 96%, at least 97%, at least
98% or at least 99% sequence identity; and/or [0153] (iii) the at
least one CDRL3 amino comprises an amino acid sequence according to
SEQ ID NO: 7 or according to SEQ ID NO: 79; or a functional
sequence variant thereof having at least 70%, at least 75%, at
least 80%, at least 85%, at least 88%, at least 90%, at least 92%,
at least 95%, at least 96%, at least 97%, at least 98% or at least
99% sequence identity.
[0154] It is also preferred that the antibody, or the antigen
binding fragment thereof, according to the present invention
comprises a heavy chain comprising at least one CDRH1, at least one
CDRH2 and at least one CDRH3 and a light chain comprising at least
one CDRL1, at least one CDRL2 and at least one CDRL3, wherein
[0155] (i) the at least one CDRL1 comprises an amino acid sequence
according to SEQ ID NO: 22 or according to SEQ ID NO: 40; or a
functional sequence variant thereof having at least 70%, at least
75%, at least 80%, at least 85%, at least 88%, at least 90%, at
least 92%, at least 95%, at least 96%, at least 97%, at least 98%
or at least 99% sequence identity; [0156] (ii) the at least one
CDRL2 comprises an amino acid sequence according to any of SEQ ID
NOs: 23, 24, 41 and 42 or a functional sequence variant thereof
having at least 70%, at least 75%, at least 80%, at least 85%, at
least 88%, at least 90%, at least 92%, at least 95%, at least 96%,
at least 97%, at least 98% or at least 99% sequence identity;
and/or [0157] (iii) the at least one CDRL3 amino comprises an amino
acid sequence according to SEQ ID NO: 25 or according to SEQ ID NO:
43; or a functional sequence variant thereof having at least 70%,
at least 75%, at least 80%, at least 85%, at least 88%, at least
90%, at least 92%, at least 95%, at least 96%, at least 97%, at
least 98% or at least 99% sequence identity.
[0158] Most preferably, the antibody, or the antigen binding
fragment thereof, according to the present invention comprises a
heavy chain comprising at least one CDRH1, at least one CDRH2 and
at least one CDRH3 and a light chain comprising at least one CDRL1,
at least one CDRL2 and at least one CDRL3, wherein [0159] (i) the
at least one CDRL1 comprises an amino acid sequence according to
SEQ ID NO: 4 or a functional sequence variant thereof having at
least 70%, at least 75%, at least 80%, at least 85%, at least 88%,
at least 90%, at least 92%, at least 95%, at least 96%, at least
97%, at least 98% or at least 99% sequence identity; [0160] (ii)
the at least one CDRL2 comprises an amino acid sequence according
to SEQ ID NO: 5 or 6, or a functional sequence variant thereof
having at least 70%, at least 75%, at least 80%, at least 85%, at
least 88%, at least 90%, at least 92%, at least 95%, at least 96%,
at least 97%, at least 98% or at least 99% sequence identity;
and/or [0161] (iii) the at least one CDRL3 amino comprises an amino
acid sequence according to SEQ ID NO: 7 or a functional sequence
variant thereof having at least 70%, at least 75%, at least 80%, at
least 85%, at least 88%, at least 90%, at least 92%, at least 95%,
at least 96%, at least 97%, at least 98% or at least 99% sequence
identity.
[0162] Preferably, the antibody, or the antigen binding fragment
thereof, according to the present invention comprises CDRH1, CDRH2,
and CDRH3 amino acid sequences (i) according to SEQ ID NOs: 1-3; or
functional sequence variants thereof having at least 70%, at least
75%, at least 80%, at least 85%, at least 88%, at least 90%, at
least 92%, at least 95%, at least 96%, at least 97%, at least 98%
or at least 99% sequence identity; (ii) according to SEQ ID NOs:
19-21; or functional sequence variants thereof having at least 70%,
at least 75%, at least 80%, at least 85%, at least 88%, at least
90%, at least 92%, at least 95%, at least 96%, at least 97%, at
least 98% or at least 99% sequence identity; (iii) according to SEQ
ID NOs: 37-39; or functional sequence variants thereof having at
least 70%, at least 75%, at least 80%, at least 85%, at least 88%,
at least 90%, at least 92%, at least 95%, at least 96%, at least
97%, at least 98% or at least 99% sequence identity; (iv) according
to SEQ ID NOs: 55-57; or functional sequence variants thereof
having at least 70%, at least 75%, at least 80%, at least 85%, at
least 88%, at least 90%, at least 92%, at least 95%, at least 96%,
at least 97%, at least 98% or at least 99% sequence identity; (v)
according to SEQ ID NOs: 73-75; or functional sequence variants
thereof having at least 70%, at least 75%, at least 80%, at least
85%, at least 88%, at least 90%, at least 92%, at least 95%, at
least 96%, at least 97%, at least 98% or at least 99% sequence
identity; (vi) according to SEQ ID NOs: 237-239; or functional
sequence variants thereof having at least 70%, at least 75%, at
least 80%, at least 85%, at least 88%, at least 90%, at least 92%,
at least 95%, at least 96%, at least 97%, at least 98% or at least
99% sequence identity; (vii) according to SEQ ID NOs: 241-243; or
functional sequence variants thereof having at least 70%, at least
75%, at least 80%, at least 85%, at least 88%, at least 90%, at
least 92%, at least 95%, at least 96%, at least 97%, at least 98%
or at least 99% sequence identity; (viii) according to SEQ ID NOs:
245-247; or functional sequence variants thereof having at least
70%, at least 75%, at least 80%, at least 85%, at least 88%, at
least 90%, at least 92%, at least 95%, at least 96%, at least 97%,
at least 98% or at least 99% sequence identity; (ix) according to
SEQ ID NOs: 249-251; or functional sequence variants thereof having
at least 70%, at least 75%, at least 80%, at least 85%, at least
88%, at least 90%, at least 92%, at least 95%, at least 96%, at
least 97%, at least 98% or at least 99% sequence identity; (x)
according to SEQ ID NOs: 253-255; or functional sequence variants
thereof having at least 70%, at least 75%, at least 80%, at least
85%, at least 88%, at least 90%, at least 92%, at least 95%, at
least 96%, at least 97%, at least 98% or at least 99% sequence
identity; (xi) according to SEQ ID NOs: 257-259; or functional
sequence variants thereof having at least 70%, at least 75%, at
least 80%, at least 85%, at least 88%, at least 90%, at least 92%,
at least 95%, at least 96%, at least 97%, at least 98% or at least
99% sequence identity; (xii) according to SEQ ID NOs: 261-263; or
functional sequence variants thereof having at least 70%, at least
75%, at least 80%, at least 85%, at least 88%, at least 90%, at
least 92%, at least 95%, at least 96%, at least 97%, at least 98%
or at least 99% sequence identity; (xiii) according to SEQ ID NOs:
265-267; or functional sequence variants thereof having at least
70%, at least 75%, at least 80%, at least 85%, at least 88%, at
least 90%, at least 92%, at least 95%, at least 96%, at least 97%,
at least 98% or at least 99% sequence identity; (xiv) according to
SEQ ID NOs: 269-271; or functional sequence variants thereof having
at least 70%, at least 75%, at least 80%, at least 85%, at least
88%, at least 90%, at least 92%, at least 95%, at least 96%, at
least 97%, at least 98% or at least 99% sequence identity; (xv)
according to SEQ ID NOs: 273-275; or functional sequence variants
thereof having at least 70%, at least 75%, at least 80%, at least
85%, at least 88%, at least 90%, at least 92%, at least 95%, at
least 96%, at least 97%, at least 98% or at least 99% sequence
identity; (xvi) according to SEQ ID NOs: 277-279; or functional
sequence variants thereof having at least 70%, at least 75%, at
least 80%, at least 85%, at least 88%, at least 90%, at least 92%,
at least 95%, at least 96%, at least 97%, at least 98% or at least
99% sequence identity; (xvii) according to SEQ ID NOs: 281-283; or
functional sequence variants thereof having at least 70%, at least
75%, at least 80%, at least 85%, at least 88%, at least 90%, at
least 92%, at least 95%, at least 96%, at least 97%, at least 98%
or at least 99% sequence identity; (xviii) according to SEQ ID NOs:
285-287; or functional sequence variants thereof having at least
70%, at least 75%, at least 80%, at least 85%, at least 88%, at
least 90%, at least 92%, at least 95%, at least 96%, at least 97%,
at least 98% or at least 99% sequence identity; (xix) according to
SEQ ID NOs: 289-291; or functional sequence variants thereof having
at least 70%, at least 75%, at least 80%, at least 85%, at least
88%, at least 90%, at least 92%, at least 95%, at least 96%, at
least 97%, at least 98% or at least 99% sequence identity; (xx)
according to SEQ ID NOs: 293-295; or functional sequence variants
thereof having at least 70%, at least 75%, at least 80%, at least
85%, at least 88%, at least 90%, at least 92%, at least 95%, at
least 96%, at least 97%, at least 98% or at least 99% sequence
identity; (xxi) according to SEQ ID NOs: 297-299; or functional
sequence variants thereof having at least 70%, at least 75%, at
least 80%, at least 85%, at least 88%, at least 90%, at least 92%,
at least 95%, at least 96%, at least 97%, at least 98% or at least
99% sequence identity; (xxii) according to SEQ ID NOs: 301-303; or
functional sequence variants thereof having at least 70%, at least
75%, at least 80%, at least 85%, at least 88%, at least 90%, at
least 92%, at least 95%, at least 96%, at least 97%, at least 98%
or at least 99% sequence identity; (xxiii) according to SEQ ID NOs:
305-307; or functional sequence variants thereof having at least
70%, at least 75%, at least 80%, at least 85%, at least 88%, at
least 90%, at least 92%, at least 95%, at least 96%, at least 97%,
at least 98% or at least 99% sequence identity; (xxiv) according to
SEQ ID NOs: 309-311; or functional sequence variants thereof having
at least 70%, at least 75%, at least 80%, at least 85%, at least
88%, at least 90%, at least 92%, at least 95%, at least 96%, at
least 97%, at least 98% or at least 99% sequence identity; (xxv)
according to SEQ ID NOs: 313-315; or functional sequence variants
thereof having at least 70%, at least 75%, at least 80%, at least
85%, at least 88%, at least 90%, at least 92%, at least 95%, at
least 96%, at least 97%, at least 98% or at least 99% sequence
identity; (xxvi) according to SEQ ID NOs: 317-319; or functional
sequence variants thereof having at least 70%, at least 75%, at
least 80%, at least 85%, at least 88%, at least 90%, at least 92%,
at least 95%, at least 96%, at least 97%, at least 98% or at least
99% sequence identity; (xxvii) according to SEQ ID NOs: 321-323; or
functional sequence variants thereof having at least 70%, at least
75%, at least 80%, at least 85%, at least 88%, at least 90%, at
least 92%, at least 95%, at least 96%, at least 97%, at least 98%
or at least 99% sequence identity; (xxviii) according to SEQ ID
NOs: 325-327; or functional sequence variants thereof having at
least 70%, at least 75%, at least 80%, at least 85%, at least 88%,
at least 90%, at least 92%, at least 95%, at least 96%, at least
97%, at least 98% or at least 99% sequence identity; (xxix)
according to SEQ ID NOs: 329-331; or functional sequence variants
thereof having at least 70%, at least 75%, at least 80%, at least
85%, at least 88%, at least 90%, at least 92%, at least 95%, at
least 96%, at least 97%, at least 98% or at least 99% sequence
identity; (xxx) according to SEQ ID NOs: 333-335; or functional
sequence variants thereof having at least 70%, at least 75%, at
least 80%, at least 85%, at least 88%, at least 90%, at least 92%,
at least 95%, at least 96%, at least 97%, at least 98% or at least
99% sequence identity; (xxxi) according to SEQ ID NOs: 337-339; or
functional sequence variants thereof having at least 70%, at least
75%, at least 80%, at least 85%, at least 88%, at least 90%, at
least 92%, at least 95%, at least 96%, at least 97%, at least 98%
or at least 99% sequence identity; (xxxii) according to SEQ ID NOs:
341-343; or functional sequence variants thereof having at least
70%, at least 75%, at least 80%, at least 85%, at least 88%, at
least 90%, at least 92%, at least 95%, at least 96%, at least 97%,
at least 98% or at least 99% sequence identity; (xxxiii) according
to SEQ ID NOs: 345-347; or functional sequence variants thereof
having at least 70%, at least 75%, at least 80%, at least 85%, at
least 88%, at least 90%, at least 92%, at least 95%, at least 96%,
at least 97%, at least 98% or at least 99% sequence identity;
(xxxiv) according to SEQ ID NOs: 349-351; or functional sequence
variants thereof having at least 70%, at least 75%, at least 80%,
at least 85%, at least 88%, at least 90%, at least 92%, at least
95%, at least 96%, at least 97%, at least 98% or at least 99%
sequence identity; (xxxv) according to SEQ ID NOs: 353-355; or
functional sequence variants thereof having at least 70%, at least
75%, at least 80%, at least 85%, at least 88%, at least 90%, at
least 92%, at least 95%, at least 96%, at least 97%, at least 98%
or at least 99% sequence identity; (xxxvi) according to SEQ ID NOs:
357-359; or functional sequence variants thereof having at least
70%, at least 75%, at least 80%, at least 85%, at least 88%, at
least 90%, at least 92%, at least 95%, at least 96%, at least 97%,
at least 98% or at least 99% sequence identity; (xxxvii) according
to SEQ ID NOs: 361-363; or functional sequence variants thereof
having at least 70%, at least 75%, at least 80%, at least 85%, at
least 88%, at least 90%, at least 92%, at least 95%, at least 96%,
at least 97%, at least 98% or at least 99% sequence identity;
(xxxviii) according to SEQ ID NOs: 365-367; or functional sequence
variants thereof having at least 70%, at least 75%, at least 80%,
at least 85%, at least 88%, at least 90%, at least 92%, at least
95%, at least 96%, at least 97%, at least 98% or at least 99%
sequence identity; (xxxix) according to SEQ ID NOs: 369-371; or
functional sequence variants thereof having at least 70%, at least
75%, at least 80%, at least 85%, at least 88%, at least 90%, at
least 92%, at least 95%, at least 96%, at least 97%, at least 98%
or at least 99% sequence identity; (xl) according to SEQ ID NOs:
373-375; or functional sequence variants thereof having at least
70%, at least 75%, at least 80%, at least 85%, at least 88%, at
least 90%, at least 92%, at least 95%, at least 96%, at least 97%,
at least 98% or at least 99% sequence identity; (xli) according to
SEQ ID NOs: 377-379; or functional sequence variants thereof having
at least 70%, at least 75%, at least 80%, at least 85%, at least
88%, at least 90%, at least 92%, at least 95%, at least 96%, at
least 97%, at least 98% or at least 99% sequence identity; (xlii)
according to SEQ ID NOs: 381-383; or functional sequence variants
thereof having at least 70%, at least 75%, at least 80%, at least
85%, at least 88%, at least 90%, at least 92%, at least 95%, at
least 96%, at least 97%, at least 98% or at least 99% sequence
identity; (xliii) according to SEQ ID NOs: 385-387; or functional
sequence variants thereof having at least 70%, at least 75%, at
least 80%, at least 85%, at least 88%, at least 90%, at least 92%,
at least 95%, at least 96%, at least 97%, at least 98% or at least
99% sequence identity; (xliv) according to SEQ ID NOs: 389-391; or
functional sequence variants thereof having at least 70%, at least
75%, at least 80%, at least 85%, at least 88%, at least 90%, at
least 92%, at least 95%, at least 96%, at least 97%, at least 98%
or at least 99% sequence identity; (xlv) according to SEQ ID NOs:
393-395; or functional sequence variants thereof having at least
70%, at least 75%, at least 80%, at least 85%, at least 88%, at
least 90%, at least 92%, at least 95%, at least 96%, at least 97%,
at least 98% or at least 99% sequence identity; or (xlvi) according
to SEQ ID NOs: 397-399; or functional sequence variants thereof
having at least 70%, at least 75%, at least 80%, at least 85%, at
least 88%, at least 90%, at least 92%, at least 95%, at least 96%,
at least 97%, at least 98% or at least 99% sequence identity.
[0163] Accordingly, it is also preferred that the antibody, or the
antigen binding fragment thereof, according to the present
invention comprises CDRH1, CDRH2, and CDRH3 amino acid sequences
and CDRL1, CDRL2, and CDRL3 amino acid sequences (i) according to
SEQ ID NOs: 1-5 and 7; or functional sequence variants thereof
having at least 70%, at least 75%, at least 80%, at least 85%, at
least 88%, at least 90%, at least 92%, at least 95%, at least 96%,
at least 97%, at least 98% or at least 99% sequence identity; (ii)
according to SEQ ID NOs: 1-4 and 6-7; or functional sequence
variants thereof having at least 70%, at least 75%, at least 80%,
at least 85%, at least 88%, at least 90%, at least 92%, at least
95%, at least 96%, at least 97%, at least 98% or at least 99%
sequence identity; (iii) according to SEQ ID NOs: 19-23 and 25; or
functional sequence variants thereof having at least 70%, at least
75%, at least 80%, at least 85%, at least 88%, at least 90%, at
least 92%, at least 95%, at least 96%, at least 97%, at least 98%
or at least 99% sequence identity; (iv) according to SEQ ID NOs:
19-22 and 24-25; or functional sequence variants thereof having at
least 70%, at least 75%, at least 80%, at least 85%, at least 88%,
at least 90%, at least 92%, at least 95%, at least 96%, at least
97%, at least 98% or at least 99% sequence identity; (v) according
to SEQ ID NOs: 37-41 and 43; or functional sequence variants
thereof having at least 70%, at least 75%, at least 80%, at least
85%, at least 88%, at least 90%, at least 92%, at least 95%, at
least 96%, at least 97%, at least 98% or at least 99% sequence
identity; (vi) according to SEQ ID NOs: 37-40 and 42-43; or
functional sequence variants thereof having at least 70%, at least
75%, at least 80%, at least 85%, at least 88%, at least 90%, at
least 92%, at least 95%, at least 96%, at least 97%, at least 98%
or at least 99% sequence identity; (vii) according to SEQ ID NOs:
55-59 and 61; or functional sequence variants thereof having at
least 70%, at least 75%, at least 80%, at least 85%, at least 88%,
at least 90%, at least 92%, at least 95%, at least 96%, at least
97%, at least 98% or at least 99% sequence identity; (viii)
according to SEQ ID NOs: 55-58 and 60-61; or functional sequence
variants thereof having at least 70%, at least 75%, at least 80%,
at least 85%, at least 88%, at least 90%, at least 92%, at least
95%, at least 96%, at least 97%, at least 98% or at least 99%
sequence identity; (ix) according to SEQ ID NOs: 73-77 and 79; or
functional sequence variants thereof having at least 70%, at least
75%, at least 80%, at least 85%, at least 88%, at least 90%, at
least 92%, at least 95%, at least 96%, at least 97%, at least 98%
or at least 99% sequence identity; or (x) according to SEQ ID NOs:
73-76 and 78-79; or functional sequence variants thereof having at
least 70%, at least 75%, at least 80%, at least 85%, at least 88%,
at least 90%, at least 92%, at least 95%, at least 96%, at least
97%, at least 98% or at least 99% sequence identity.
[0164] More preferably, the antibody, or the antigen binding
fragment thereof, according to the present invention comprises
CDRH1, CDRH2, and CDRH3 amino acid sequences and CDRL1, CDRL2, and
CDRL3 amino acid sequences (i) according to SEQ ID NOs: 1-5 and 7;
or functional sequence variants thereof having at least 70%, at
least 75%, at least 80%, at least 85%, at least 88%, at least 90%,
at least 92%, at least 95%, at least 96%, at least 97%, at least
98% or at least 99% sequence identity; (ii) according to SEQ ID
NOs: 1-4 and 6-7; or functional sequence variants thereof having at
least 70%, at least 75%, at least 80%, at least 85%, at least 88%,
at least 90%, at least 92%, at least 95%, at least 96%, at least
97%, at least 98% or at least 99% sequence identity; (iii)
according to SEQ ID NOs: 19-23 and 25; or functional sequence
variants thereof having at least 70%, at least 75%, at least 80%,
at least 85%, at least 88%, at least 90%, at least 92%, at least
95%, at least 96%, at least 97%, at least 98% or at least 99%
sequence identity; (iv) according to SEQ ID NOs: 19-22 and 24-25;
or functional sequence variants thereof having at least 70%, at
least 75%, at least 80%, at least 85%, at least 88%, at least 90%,
at least 92%, at least 95%, at least 96%, at least 97%, at least
98% or at least 99% sequence identity; (v) according to SEQ ID NOs:
37-41 and 43; or functional sequence variants thereof having at
least 70%, at least 75%, at least 80%, at least 85%, at least 88%,
at least 90%, at least 92%, at least 95%, at least 96%, at least
97%, at least 98% or at least 99% sequence identity; (vi) according
to SEQ ID NOs: 37-40 and 42-43; or functional sequence variants
thereof having at least 70%, at least 75%, at least 80%, at least
85%, at least 88%, at least 90%, at least 92%, at least 95%, at
least 96%, at least 97%, at least 98% or at least 99% sequence
identity; (vii) according to SEQ ID NOs: 73-77 and 79; or
functional sequence variants thereof having at least 70%, at least
75%, at least 80%, at least 85%, at least 88%, at least 90%, at
least 92%, at least 95%, at least 96%, at least 97%, at least 98%
or at least 99% sequence identity; or (viii) according to SEQ ID
NOs: 73-76 and 78-79; or functional sequence variants thereof
having at least 70%, at least 75%, at least 80%, at least 85%, at
least 88%, at least 90%, at least 92%, at least 95%, at least 96%,
at least 97%, at least 98% or at least 99% sequence identity.
[0165] Even more preferably, the antibody, or the antigen binding
fragment thereof, according to the present invention comprises
CDRH1, CDRH2, and CDRH3 amino acid sequences and CDRL1, CDRL2, and
CDRL3 amino acid sequences (i) according to SEQ ID NOs: 1-5 and 7;
or functional sequence variants thereof having at least 70%, at
least 75%, at least 80%, at least 85%, at least 88%, at least 90%,
at least 92%, at least 95%, at least 96%, at least 97%, at least
98% or at least 99% sequence identity; (ii) according to SEQ ID
NOs: 1-4 and 6-7; or functional sequence variants thereof having at
least 70%, at least 75%, at least 80%, at least 85%, at least 88%,
at least 90%, at least 92%, at least 95%, at least 96%, at least
97%, at least 98% or at least 99% sequence identity; (iii)
according to SEQ ID NOs: 73-77 and 79; or functional sequence
variants thereof having at least 70%, at least 75%, at least 80%,
at least 85%, at least 88%, at least 90%, at least 92%, at least
95%, at least 96%, at least 97%, at least 98% or at least 99%
sequence identity; or (iv) according to SEQ ID NOs: 73-76 and
78-79; or functional sequence variants thereof having at least 70%,
at least 75%, at least 80%, at least 85%, at least 88%, at least
90%, at least 92%, at least 95%, at least 96%, at least 97%, at
least 98% or at least 99% sequence identity.
[0166] It is also preferred that preferably, the antibody, or the
antigen binding fragment thereof, according to the present
invention comprises CDRH1, CDRH2, and CDRH3 amino acid sequences
and CDRL1, CDRL2, and CDRL3 amino acid sequences (i) according to
SEQ ID NOs: 19-23 and 25; or functional sequence variants thereof
having at least 70%, at least 75%, at least 80%, at least 85%, at
least 88%, at least 90%, at least 92%, at least 95%, at least 96%,
at least 97%, at least 98% or at least 99% sequence identity; (ii)
according to SEQ ID NOs: 19-22 and 24-25; or functional sequence
variants thereof having at least 70%, at least 75%, at least 80%,
at least 85%, at least 88%, at least 90%, at least 92%, at least
95%, at least 96%, at least 97%, at least 98% or at least 99%
sequence identity; (iii) according to SEQ ID NOs: 37-41 and 43; or
functional sequence variants thereof having at least 70%, at least
75%, at least 80%, at least 85%, at least 88%, at least 90%, at
least 92%, at least 95%, at least 96%, at least 97%, at least 98%
or at least 99% sequence identity; or (vi) according to SEQ ID NOs:
37-40 and 42-43; or functional sequence variants thereof having at
least 70%, at least 75%, at least 80%, at least 85%, at least 88%,
at least 90%, at least 92%, at least 95%, at least 96%, at least
97%, at least 98% or at least 99% sequence identity.
[0167] Most preferably, the antibody, or the antigen binding
fragment thereof, according to the present invention comprises
CDRH1, CDRH2, and CDRH3 amino acid sequences and CDRL1, CDRL2, and
CDRL3 amino acid sequences (i) according to SEQ ID NOs: 1-5 and 7;
or functional sequence variants thereof having at least 70%, at
least 75%, at least 80%, at least 85%, at least 88%, at least 90%,
at least 92%, at least 95%, at least 96%, at least 97%, at least
98% or at least 99% sequence identity; or (ii) according to SEQ ID
NOs: 1-4 and 6-7; or functional sequence variants thereof having at
least 70%, at least 75%, at least 80%, at least 85%, at least 88%,
at least 90%, at least 92%, at least 95%, at least 96%, at least
97%, at least 98% or at least 99% sequence identity.
[0168] In addition, it is also preferred that the antibody, or the
antigen binding fragment thereof, according to the present
invention comprises a heavy chain variable region (VH) and,
optionally, a light chain variable region (VL), wherein the heavy
chain variable region (VH) comprises or consists of an amino acid
sequence according to any of SEQ ID NOs: 8, 26, 44, 62, 80, 240,
244, 248, 252, 256, 260, 264, 268, 272, 276, 280, 284, 288, 292,
296, 300, 304, 308, 312, 316, 320, 324, 328, 332, 336, 340, 344,
348, 352, 356, 360, 364, 368, 372, 376, 380, 384, 388, 392, 396,
and 400; or a functional sequence variant thereof having at least
70%, at least 75%, at least 80%, at least 85%, at least 88%, at
least 90%, at least 92%, at least 95%, at least 96%, at least 97%,
at least 98% or at least 99% sequence identity.
[0169] Moreover, it is also preferred that the antibody, or the
antigen binding fragment thereof, according to the present
invention comprises (i) a heavy chain variable region (VH) amino
acid sequence according to SEQ ID NO: 8 or a functional sequence
variant thereof having at least 70%, at least 75%, at least 80%, at
least 85%, at least 88%, at least 90%, at least 92%, at least 95%,
at least 96%, at least 97%, at least 98% or at least 99% sequence
identity and/or a light chain variable region (VL) amino acid
sequence according to SEQ ID NO: 9 or a functional sequence variant
thereof having at least 70%, at least 75%, at least 80%, at least
85%, at least 88%, at least 90%, at least 92%, at least 95%, at
least 96%, at least 97%, at least 98% or at least 99% sequence
identity; (ii) a heavy chain variable region (VH) amino acid
sequence according to SEQ ID NO: 26 or a functional sequence
variant thereof having at least 70%, at least 75%, at least 80%, at
least 85%, at least 88%, at least 90%, at least 92%, at least 95%,
at least 96%, at least 97%, at least 98% or at least 99% sequence
identity and/or a light chain variable region (VL) amino acid
sequence according to SEQ ID NO: 27 or a functional sequence
variant thereof having at least 70%, at least 75%, at least 80%, at
least 85%, at least 88%, at least 90%, at least 92%, at least 95%,
at least 96%, at least 97%, at least 98% or at least 99% sequence
identity; (iii) a heavy chain variable region (VH) amino acid
sequence according to SEQ ID NO: 44 or a functional sequence
variant thereof having at least 70%, at least 75%, at least 80%, at
least 85%, at least 88%, at least 90%, at least 92%, at least 95%,
at least 96%, at least 97%, at least 98% or at least 99% sequence
identity and/or a light chain variable region (VL) amino acid
sequence according to SEQ ID NO: 45 or a functional sequence
variant thereof having at least 70%, at least 75%, at least 80%, at
least 85%, at least 88%, at least 90%, at least 92%, at least 95%,
at least 96%, at least 97%, at least 98% or at least 99% sequence
identity; (iv) a heavy chain variable region (VH) amino acid
sequence according to SEQ ID NO: 62 or a functional sequence
variant thereof having at least 70%, at least 75%, at least 80%, at
least 85%, at least 88%, at least 90%, at least 92%, at least 95%,
at least 96%, at least 97%, at least 98% or at least 99% sequence
identity and/or a light chain variable region (VL) amino acid
sequence according to SEQ ID NO: 63 or a functional sequence
variant thereof having at least 70%, at least 75%, at least 80%, at
least 85%, at least 88%, at least 90%, at least 92%, at least 95%,
at least 96%, at least 97%, at least 98% or at least 99% sequence
identity; or (v) a heavy chain variable region (VH) amino acid
sequence according to SEQ ID NO: 80 or a functional sequence
variant thereof having at least 70%, at least 75%, at least 80%, at
least 85%, at least 88%, at least 90%, at least 92%, at least 95%,
at least 96%, at least 97%, at least 98% or at least 99% sequence
identity and/or a light chain variable region (VL) amino acid
sequence according to SEQ ID NO: 81 or a functional sequence
variant thereof having at least 70%, at least 75%, at least 80%, at
least 85%, at least 88%, at least 90%, at least 92%, at least 95%,
at least 96%, at least 97%, at least 98% or at least 99% sequence
identity.
[0170] More preferably, the antibody, or the antigen binding
fragment thereof, according to the present invention comprises (i)
a heavy chain variable region (VH) amino acid sequence according to
SEQ ID NO: 8 or a functional sequence variant thereof having at
least 70%, at least 75%, at least 80%, at least 85%, at least 88%,
at least 90%, at least 92%, at least 95%, at least 96%, at least
97%, at least 98% or at least 99% sequence identity and/or a light
chain variable region (VL) amino acid sequence according to SEQ ID
NO: 9 or a functional sequence variant thereof having at least 70%,
at least 75%, at least 80%, at least 85%, at least 88%, at least
90%, at least 92%, at least 95%, at least 96%, at least 97%, at
least 98% or at least 99% sequence identity; (ii) a heavy chain
variable region (VH) amino acid sequence according to SEQ ID NO: 26
or a functional sequence variant thereof having at least 70%, at
least 75%, at least 80%, at least 85%, at least 88%, at least 90%,
at least 92%, at least 95%, at least 96%, at least 97%, at least
98% or at least 99% sequence identity and/or a light chain variable
region (VL) amino acid sequence according to SEQ ID NO: 27 or a
functional sequence variant thereof having at least 70%, at least
75%, at least 80%, at least 85%, at least 88%, at least 90%, at
least 92%, at least 95%, at least 96%, at least 97%, at least 98%
or at least 99% sequence identity; (iii) a heavy chain variable
region (VH) amino acid sequence according to SEQ ID NO: 44 or a
functional sequence variant thereof having at least 70%, at least
75%, at least 80%, at least 85%, at least 88%, at least 90%, at
least 92%, at least 95%, at least 96%, at least 97%, at least 98%
or at least 99% sequence identity and/or a light chain variable
region (VL) amino acid sequence according to SEQ ID NO: 45 or a
functional sequence variant thereof having at least 70%, at least
75%, at least 80%, at least 85%, at least 88%, at least 90%, at
least 92%, at least 95%, at least 96%, at least 97%, at least 98%
or at least 99% sequence identity; or (iv) a heavy chain variable
region (VH) amino acid sequence according to SEQ ID NO: 80 or a
functional sequence variant thereof having at least 70%, at least
75%, at least 80%, at least 85%, at least 88%, at least 90%, at
least 92%, at least 95%, at least 96%, at least 97%, at least 98%
or at least 99% sequence identity and/or a light chain variable
region (VL) amino acid sequence according to SEQ ID NO: 81 or a
functional sequence variant thereof having at least 70%, at least
75%, at least 80%, at least 85%, at least 88%, at least 90%, at
least 92%, at least 95%, at least 96%, at least 97%, at least 98%
or at least 99% sequence identity.
[0171] Even more preferably, the antibody, or the antigen binding
fragment thereof, according to the present invention comprises (i)
a heavy chain variable region (VH) amino acid sequence according to
SEQ ID NO: 8 or a functional sequence variant thereof having at
least 70%, at least 75%, at least 80%, at least 85%, at least 88%,
at least 90%, at least 92%, at least 95%, at least 96%, at least
97%, at least 98% or at least 99% sequence identity and/or a light
chain variable region (VL) amino acid sequence according to SEQ ID
NO: 9 or a functional sequence variant thereof having at least 70%,
at least 75%, at least 80%, at least 85%, at least 88%, at least
90%, at least 92%, at least 95%, at least 96%, at least 97%, at
least 98% or at least 99% sequence identity; or (ii) a heavy chain
variable region (VH) amino acid sequence according to SEQ ID NO: 80
or a functional sequence variant thereof having at least 70%, at
least 75%, at least 80%, at least 85%, at least 88%, at least 90%,
at least 92%, at least 95%, at least 96%, at least 97%, at least
98% or at least 99% sequence identity and/or a light chain variable
region (VL) amino acid sequence according to SEQ ID NO: 81 or a
functional sequence variant thereof having at least 70%, at least
75%, at least 80%, at least 85%, at least 88%, at least 90%, at
least 92%, at least 95%, at least 96%, at least 97%, at least 98%
or at least 99% sequence identity.
[0172] It is also preferred that the antibody, or the antigen
binding fragment thereof, according to the present invention
comprises (i) a heavy chain variable region (VH) amino acid
sequence according to SEQ ID NO: 26 or a functional sequence
variant thereof having at least 70%, at least 75%, at least 80%, at
least 85%, at least 88%, at least 90%, at least 92%, at least 95%,
at least 96%, at least 97%, at least 98% or at least 99% sequence
identity and/or a light chain variable region (VL) amino acid
sequence according to SEQ ID NO: 27 or a functional sequence
variant thereof having at least 70%, at least 75%, at least 80%, at
least 85%, at least 88%, at least 90%, at least 92%, at least 95%,
at least 96%, at least 97%, at least 98% or at least 99% sequence
identity; or (ii) a heavy chain variable region (VH) amino acid
sequence according to SEQ ID NO: 44 or a functional sequence
variant thereof having at least 70%, at least 75%, at least 80%, at
least 85%, at least 88%, at least 90%, at least 92%, at least 95%,
at least 96%, at least 97%, at least 98% or at least 99% sequence
identity and/or a light chain variable region (VL) amino acid
sequence according to SEQ ID NO: 45 or a functional sequence
variant thereof having at least 70%, at least 75%, at least 80%, at
least 85%, at least 88%, at least 90%, at least 92%, at least 95%,
at least 96%, at least 97%, at least 98% or at least 99% sequence
identity.
[0173] Most preferably, the antibody, or the antigen binding
fragment thereof, according to the present invention comprises a
heavy chain variable region (VH) amino acid sequence according to
SEQ ID NO: 8 or a functional sequence variant thereof having at
least 70%, at least 75%, at least 80%, at least 85%, at least 88%,
at least 90%, at least 92%, at least 95%, at least 96%, at least
97%, at least 98% or at least 99% sequence identity and/or a light
chain variable region (VL) amino acid sequence according to SEQ ID
NO: 9 or a functional sequence variant thereof having at least 70%,
at least 75%, at least 80%, at least 85%, at least 88%, at least
90%, at least 92%, at least 95%, at least 96%, at least 97%, at
least 98% or at least 99% sequence identity.
[0174] Preferably, the antibody, or the antigen binding fragment
thereof, according to the present invention is gZKA190, gZKA64,
gZKA230, gZKA185 or gZKA78, preferably the antibody, or the antigen
binding fragment thereof, is gZKA190, gZKA64, gZKA230 or gZKA185,
more preferably the antibody, or the antigen binding fragment
thereof, is gZKA190 or gZKA64, and most preferably the antibody, or
the antigen binding fragment thereof, is gZKA190.
[0175] The present inventors have isolated monoclonal antibody
(mAb) according to the present invention, which are referred to
herein as ZKA190, ZKA64, ZKA230, ZKA185 and ZKA78 (cf. Tables 1 and
2, Example 1). Based on those antibodies, in particular on the VH
and VL genes of those antibodies, the terms "gZKA190", "gZKA64",
"gZKA230", "gZKA185" and "gZKA78", as used herein, refer to the
respective "generic" antibodies, or antigen binding fragments
thereof.
[0176] Namely, "gZKA190" refers to an antibody, or antigen binding
fragment thereof, having a CDRH1 amino acid sequence according to
SEQ ID NO: 1, a CDRH2 amino acid sequence according to SEQ ID NO:
2, a CDRH3 amino acid sequence according to SEQ ID NO: 3, a CDRL1
amino acid sequence according to SEQ ID NO: 4, a CDRL2 amino acid
sequence according to SEQ ID NO: 5 or 6, and a CDRL3 amino acid
sequence according to SEQ ID NO: 7. The heavy chain variable region
(V.sub.H) has preferably an amino acid sequence according to SEQ ID
NO: 8 and the light chain variable region (V.sub.L) has preferably
an amino acid sequence according to SEQ ID NO: 9.
[0177] "gZKA64" refers to an antibody, or antigen binding fragment
thereof, having a CDRH1 amino acid sequence according to SEQ ID NO:
73, a CDRH2 amino acid sequence according to SEQ ID NO: 74, a CDRH3
amino acid sequence according to SEQ ID NO: 75, a CDRL1 amino acid
sequence according to SEQ ID NO: 76, a CDRL2 amino acid sequence
according to SEQ ID NO: 77 or 78, and a CDRL3 amino acid sequence
according to SEQ ID NO: 79. The heavy chain variable region
(V.sub.H) has preferably an amino acid sequence according to SEQ ID
NO: 80 and the light chain variable region (V.sub.L) has preferably
an amino acid sequence according to SEQ ID NO: 81.
[0178] "gZKA230" refers to an antibody, or antigen binding fragment
thereof, having a CDRH1 amino acid sequence according to SEQ ID NO:
37, a CDRH2 amino acid sequence according to SEQ ID NO: 38, a CDRH3
amino acid sequence according to SEQ ID NO: 39, a CDRL1 amino acid
sequence according to SEQ ID NO: 40, a CDRL2 amino acid sequence
according to SEQ ID NO: 41 or 42, and a CDRL3 amino acid sequence
according to SEQ ID NO: 43. The heavy chain variable region
(V.sub.H) has preferably an amino acid sequence according to SEQ ID
NO: 44 and the light chain variable region (V.sub.L) has preferably
an amino acid sequence according to SEQ ID NO: 45.
[0179] "gZKA185" refers to an antibody, or antigen binding fragment
thereof, having a CDRH1 amino acid sequence according to SEQ ID NO:
19, a CDRH2 amino acid sequence according to SEQ ID NO: 20, a CDRH3
amino acid sequence according to SEQ ID NO: 21, a CDRL1 amino acid
sequence according to SEQ ID NO: 22, a CDRL2 amino acid sequence
according to SEQ ID NO: 23 or 24, and a CDRL3 amino acid sequence
according to SEQ ID NO: 25. The heavy chain variable region
(V.sub.H) has preferably an amino acid sequence according to SEQ ID
NO: 26 and the light chain variable region (V.sub.L) has preferably
an amino acid sequence according to SEQ ID NO: 27.
[0180] "gZKA78" refers to an antibody, or antigen binding fragment
thereof, having a CDRH1 amino acid sequence according to SEQ ID NO:
55, a CDRH2 amino acid sequence according to SEQ ID NO: 56, a CDRH3
amino acid sequence according to SEQ ID NO: 57, a CDRL1 amino acid
sequence according to SEQ ID NO: 58, a CDRL2 amino acid sequence
according to SEQ ID NO: 59 or 60, and a CDRL3 amino acid sequence
according to SEQ ID NO: 61. The heavy chain variable region
(V.sub.H) has preferably an amino acid sequence according to SEQ ID
NO: 62 and the light chain variable region (V.sub.L) has preferably
an amino acid sequence according to SEQ ID NO: 63.
[0181] Preferably, the antibody, or an antigen binding fragment
thereof, according to the present invention is for use as a
medicament. In other words, the antibody, or an antigen binding
fragment thereof, according to the present invention may be used in
the preparation of a medicament. More preferably, the antibody, or
an antigen binding fragment thereof, according to the present
invention is for use in the prevention and/or treatment of Zika
virus infection. In other words, the antibody, or an antigen
binding fragment thereof, according to the present invention may be
used in the preparation of a medicament or use in the prevention
and/or treatment of Zika virus infection. This aspect is described
in more detail below.
[0182] Nucleic Acid Molecule
[0183] In another aspect, the invention also provides a nucleic
acid molecule comprising a polynucleotide encoding the antibody, or
the antigen binding fragment thereof, according to the present
invention as described above. Examples of nucleic acid molecules
and/or polynucleotides include, e.g., a recombinant polynucleotide,
a vector, an oligonucleotide, an RNA molecule such as an rRNA, an
mRNA, an miRNA, an siRNA, or a tRNA, or a DNA molecule such as a
cDNA. Nucleic acid sequences encoding part or all of the light and
heavy chains and CDRs of the antibodies of the present invention
are preferred. Preferably provided herein are thus nucleic acid
sequences encoding part or all of the light and heavy chains, in
particular VH and VL sequences and CDRs of the exemplary antibodies
of the invention. Tables 1 and 2 provide the SEQ ID numbers for the
amino acid sequences of the CDRs and VH and VL of exemplary
antibodies according to the present invention.
[0184] Table 3 below provides the SEQ ID numbers for exemplary
nucleic acid sequences encoding the CDRs and VH and VL of exemplary
antibodies according to the present invention. Due to the
redundancy of the genetic code, the present invention also
comprises sequence variants of these nucleic acid sequences and in
particular such sequence variants, which encode the same amino acid
sequences.
[0185] A nucleic acid molecule is a molecule comprising, preferably
consisting of nucleic acid components. The term nucleic acid
molecule preferably refers to DNA or RNA molecules. In particular,
it is used synonymous with the term "polynucleotide". Preferably, a
nucleic acid molecule is a polymer comprising or consisting of
nucleotide monomers which are covalently linked to each other by
phosphodiester-bonds of a sugar/phosphate-backbone. The term
"nucleic acid molecule" also encompasses modified nucleic acid
molecules, such as base-modified, sugar-modified or
backbone-modified etc. DNA or RNA molecules.
[0186] Table 3 shows exemplary nucleic acid sequences of the CDR's
and the heavy chain variable region (VH) and the light chain
variable region (VL) of five exemplary antibodies according to the
present invention ("ZKA190", "ZKA64", "ZKA230", "ZKA185",
"ZKA78"):
TABLE-US-00006 SEQ ID ZKA190 NO. Nucleic acid sequence CDRH1 10
ggattcaccttcagtaaatatggc CDRH2 11 atatcatatgagggaagtaataaa CDRH3 12
gcgaaatcggggacccaatactatgatactactggttatg
agtataggggtttggaatactttggctac CDRL1 13 cagagtgttagtagcagttac CDRL2
14 gatgcatcc CDRL2 15 ctcatctatgatgcatccagcagggcc long CDRL3 16
cagcagtatggtaggtcaaggtggaca VH 17
caggtgcagctggtggagtctgggggaggcgtggtccagc
ctgggaggtccctgagactctcctgtgcagcctctggatt
caccttcagtaaatatggcatgcactgggtccgccaggct
ccaggcaaggggctggagtgggtggcagttatatcatatg
agggaagtaataaatattatgcagactccgtgaagggccg
attcaccatctccagagacaattccaagaacacgctgtat
ctgcaaatgaacagcctgagagctgaggacacggcagtgt
attactgtgcgaaatcggggacccaatactatgatactac
tggttatgagtataggggtttggaatactttggctactgg
ggccagggaaccctggtcaccgtctcctcag VL 18
gaaattgtgttgacgcagtctccaggcaccctgtctttgt
ctccaggggaaagagccaccctctcctgcagggccagtca
gagtgttagtagcagttacttagcctggtaccagcagaaa
cgtggccaggctcccaggctcctcatctatgatgcatcca
gcagggccactggcatcccagacaggttcagtggcagtgg
gtctgggacagacttcactctcaccatcagcagactggag
cctgaagattttgcagtgtattactgtcagcagtatggta
ggtcaaggtggacattcggccaagggaccaaggtggaaat caaac SEQ ID ZKA185 NO.
Nucleic acid sequence CDRH1 28 ggatatagttttaccagttactgg CDRH2 29
tttgatcctagtgactctcaaacc CDRH3 30
gcgagaagatattgtagtagtagtagttgttatgtggaca at CDRL1 31
gcattgccaaataaattt CDRL2 32 gaggacaac CDRL2 33
gtcatctatgaggacaacaaacgaccc long CDRL3 34
tactcaacagacagcagttctaatcccctgggagta VH 35
gaagtgcagctggtgcagtccggagcagaggtgaaaaagc
ccggggagtctctgaggatctcctgtaagggttctggata
tagttttaccagttactggatcacctgggtgcgccagatg
cccgggaaaggcctggagtggatggcgaagtttgatccta
gtgactctcaaaccaactacagcccgtccttccaaggcca
cgtcaccatctcagttgacaagtccatcagcactgcctac
ttgcagtggagcagcctgaaggcctcggacaccgccatgt
attactgtgcgagaagatattgtagtagtagtagttgtta
tgtggacaattggggccagggaaccctggtcaccatcttc tcag VL 36
tcctatgagctgacacagccaccctcggtgtcagtgtccc
caggacaaacggccaggatcacctgctctggagatgcatt
gccaaataaatttgcttattggtaccggcagaagtcaggc
caggcccctgttctggtcatctatgaggacaacaaacgac
cctccgggatccctgagagattctctggctccagctcagg
gacaatggccaccttgactatcagtggggcccaggtggag
gatgaagctgactaccactgttactcaacagacagcagtt
ctaatcccctgggagtattcggcggagggaccaagctgac cgtcctag SEQ ID ZKA230 NO.
Nucleic acid sequence CDRH1 46 ggtggctccatcagtagtgactac CDRH2 47
atctattacagtgggagcacc CDRH3 48
gcgaggaggaggaagtatgattccctttgggggagttttg cttttgatatc CDRL1 49
agctccaacatcggaggtaattat CDRL2 50 attaatgat CDRL2 51
ctcatctgtattaatgatcaccggccc long CDRL3 52
gcaacatgggatgacagcctgggtggccttgta VH 53
caggtgcagctgcaggagtcgggcccaggcctggtgaagc
cttcggagaccctgtccctcacctgcgcagtctctggtgg
ctccatcagtagtgactactggagctggatccggcagccc
ccagggaagggactggagtggattgggtatatctattaca
gtgggagcaccaactacaacccctccctcaagagtcgagt
caccatatcagtagacacgtccaagaaccacttctccctg
aagctgaactctgtgaccgctgcggacacggccgtgtatt
actgtgcgaggaggaggaagtatgattccctttgggggag
ttttgcttttgatatctggggccaagggacaatggtcacc gtctcttcag VL 54
cagtctgtgctgactcagccaccctcagcgtctgggaccc
ccgggcagagggtcaccatctcttgttctggaagcagctc
caacatcggaggtaattatgtatactggtaccagcagctc
ccaggaacggcccccaaactcctcatctgtattaatgatc
accggccctcaggggtccctgaccgattctctggctccaa
gtctggcacctcagcctccctggccatcagtgggctccag
tccgaggatgaggctgattattactgtgcaacatgggatg
acagcctgggtggccttgtattcggcggagggaccaagct gaccgtcctag SEQ ID ZKA78
NO. Nucleic acid sequence CDRH1 64 ggcttcacttttagtaactatgca CDRH2
65 atcgggcgcaacggggactctatc CDRH3 66
gtgaaagatctggccatccccgagtcctacagaattgaag ctgattat CDRL1 67
cagtccgtgctgtaccgctctaacaacaagaattac CDRL2 68 tgggcttca CDRL2 69
ctgatctattgggcttcaacccgggaa long CDRL3 70
cagcagtactattctagtcctcgaact VH 71
gaggtgcagctggcagaatcaggcgggggactggtccagc
ctggcggcagcctgacactgtcttgcagtggatcaggctt
cacttttagtaactatgcaatggtgtgggcaaggcaggct
cctgggaagggactggagtatgtctctggcatcgggcgca
acggggactctatctactatactgatagtgtgaagggccg
gttcaccatcagcagagacaatagcaaatccatggtgtac
ctgcagatgagctccctgcgaaccgaagacacagcagtgt
actattgcgtgaaagatctggccatccccgagtcctacag
aattgaagctgattattggggacagggcaccctggtcatc gtgagcgccg VL 72
gacatcgtgatgacacagtctccagatagtctggcagtca
gtctgggggagagggccactattaactgcaagagctccca
gtccgtgctgtaccgctctaacaacaagaattacctgtct
tggtatcagcagaagcccggacagccccctaaactgctga
tctattgggcttcaacccgggaaagcggcgtcccagacag
attctcaggcagcgggtccggaacagacttcaccctgaca
attagccccctgcaggcagaggacgtggctgtctactatt
gtcagcagtactattctagtcctcgaactttcggccaggg gaccaaggtggaaatcaaac SEQ
ID ZKA64 NO. Nucleic acid sequence CDRH1 82
ggctacaccttcacagggtatcac CDRH2 83 attaaccctaattctggcgggacc CDRH3 84
gctcggatgagctcctctatttggggcttcgatcat CDRL1 85 cagtctgtgctgattaac
CDRL2 86 ggagcatcc CDRL2 87 ctgatctatggagcatcctccagggct long CDRL3
88 cagcagtacaatgattggccccctatcaca VH 89
caggtgcagctggtccagagcggagcagaggtgaagaaac
ccggcgcctcagtgaaggtcagctgcaaagcttccggcta
caccttcacagggtatcacatcgactgggtgaggcaggca
agaggacagggactggaatggatgggacggattaacccta
attctggcgggaccaactacgcccagaagtttcagggccg
agtgactatgaccagagacaccagcatctccacagcttat
atgcagctgtcccggctgagatctgacgatagtgccgtct
actattgtgctcggatgagctcctctatttggggcttcga
tcattgggggcagggaacactggtgactgtcagttcag VL 90
gagatcgtgatgactcagtctccagccaccctgtcagtca
gcccaggagaacgggcaaccctgtcttgcagagcctccca
gtctgtgctgattaacctggcttggtaccagcagaagcca
ggccaggcaccccgactgctgatctatggagcatcctcca
gggctaccggcattcctgcacgcttcagtggatcaggaag
cggaacagagtttaccctgacaatctctagtctgcagtcc
gaagacttcgctgtctactattgtcagcagtacaatgatt
ggccccctatcacatttggccaggggactagactggagat caagc
[0187] Preferably, the sequence of the nucleic acid molecule
according to the present invention comprises or consists of a
nucleic acid sequence according to any one of SEQ ID NOs: 10-18,
28-36, 46-54, 64-72, and 82-90; or a functional sequence variant
thereof.
[0188] It is also preferred that nucleic acid sequences according
to the invention include nucleic acid sequences having at least
70%, at least 75%, at least 80%, at least 85%, at least 88%, at
least 90%, at least 92%, at least 95%, at least 96%, at least 97%,
at least 98% or at least 99% identity to the nucleic acid encoding
a CDR, a VH sequence and/or a VL sequence used in an (exemplary)
antibody according to the present invention, for example to the
sequences shown in Table 3.
[0189] In general, the nucleic acid molecule may be manipulated to
insert, delete or alter certain nucleic acid sequences. Changes
from such manipulation include, but are not limited to, changes to
introduce restriction sites, to amend codon usage, to add or
optimize transcription and/or translation regulatory sequences,
etc. It is also possible to change the nucleic acid to alter the
encoded amino acids. For example, it may be useful to introduce one
or more (e.g., 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, etc.) amino acid
substitutions, deletions and/or insertions into the antibody's
amino acid sequence. Such point mutations can modify effector
functions, antigen-binding affinity, post-translational
modifications, immunogenicity, etc., can introduce amino acids for
the attachment of covalent groups (e.g., labels) or can introduce
tags (e.g., for purification purposes). Mutations can be introduced
in specific sites or can be introduced at random, followed by
selection (e.g., molecular evolution). For instance, one or more
nucleic acids encoding any of the CDR regions, a VH sequence and/or
a VL sequence of an (exemplary) antibody of the invention can be
randomly or directionally mutated to introduce different properties
in the encoded amino acids. Such changes can be the result of an
iterative process wherein initial changes are retained and new
changes at other nucleotide positions are introduced. Further,
changes achieved in independent steps may be combined. Different
properties introduced into the encoded amino acids may include, but
are not limited to, enhanced affinity.
[0190] Vector
[0191] Further included within the scope of the invention are
vectors, for example, expression vectors, comprising a nucleic acid
molecule according to the present invention. Preferably, a vector
comprises a nucleic acid molecule as described above.
[0192] The term "vector" refers to a nucleic acid molecule,
preferably to a recombinant nucleic acid molecule, i.e. a nucleic
acid molecule which does not occur in nature. A vector in the
context of the present invention is suitable for incorporating or
harboring a desired nucleic acid sequence. Such vectors may be
storage vectors, expression vectors, cloning vectors, transfer
vectors etc. A storage vector is a vector which allows the
convenient storage of a nucleic acid molecule. Thus, the vector may
comprise a sequence corresponding, e.g., to a desired antibody or
antibody fragment thereof according to the present invention. An
expression vector may be used for production of expression products
such as RNA, e.g. mRNA, or peptides, polypeptides or proteins. For
example, an expression vector may comprise sequences needed for
transcription of a sequence stretch of the vector, such as a
promoter sequence. A cloning vector is typically a vector that
contains a cloning site, which may be used to incorporate nucleic
acid sequences into the vector. A cloning vector may be, e.g., a
plasmid vector or a bacteriophage vector. A transfer vector may be
a vector which is suitable for transferring nucleic acid molecules
into cells or organisms, for example, viral vectors. A vector in
the context of the present invention may be, e.g., an RNA vector or
a DNA vector. Preferably, a vector is a DNA molecule. For example,
a vector in the sense of the present application comprises a
cloning site, a selection marker, such as an antibiotic resistance
factor, and a sequence suitable for multiplication of the vector,
such as an origin of replication. Preferably, a vector in the
context of the present application is a plasmid vector.
[0193] Cells
[0194] In a further aspect, the present invention also provides
cell expressing the antibody, or the antigen binding fragment
thereof, according to the present invention; and/or comprising the
vector according the present invention.
[0195] Examples of such cells include but are not limited to,
eukaryotic cells, e.g., yeast cells, animal cells or plant cells.
Preferably, the cells are mammalian cells, more preferably a
mammalian cell line. Preferred examples include human cells, CHO
cells, HEK293T cells, PER.C6 cells, NS0 cells, human liver cells,
myeloma cells or hybridoma cells.
[0196] In particular, the cell may be transfected with a vector
according to the present invention, preferably with an expression
vector. The term "transfection" refers to the introduction of
nucleic acid molecules, such as DNA or RNA (e.g. mRNA) molecules,
into cells, preferably into eukaryotic cells. In the context of the
present invention, the term "transfection" encompasses any method
known to the skilled person for introducing nucleic acid molecules
into cells, preferably into eukaryotic cells, such as into
mammalian cells. Such methods encompass, for example,
electroporation, lipofection, e.g. based on cationic lipids and/or
liposomes, calcium phosphate precipitation, nanoparticle based
transfection, virus based transfection, or transfection based on
cationic polymers, such as DEAE-dextran or polyethylenimine etc.
Preferably, the introduction is non-viral.
[0197] Moreover, the cells of the present invention may be
transfected stably or transiently with the vector according to the
present invention, e.g. for expressing the antibody, or the antigen
binding fragment thereof, according to the present invention.
Preferably, the cells are stably transfected with the vector
according to the present invention encoding the antibody, or the
antigen binding fragment thereof, according to the present
invention. Alternatively, it is also preferred that the cells are
transiently transfected with the vector according to the present
invention encoding the antibody, or the antigen binding fragment
thereof, according to the present invention.
[0198] Optional Additional Features of the Antibodies
[0199] Antibodies of the invention may be coupled, for example, to
a drug for delivery to a treatment site or coupled to a detectable
label to facilitate imaging of a site comprising cells of interest.
Methods for coupling antibodies to drugs and detectable labels are
well known in the art, as are methods for imaging using detectable
labels. Labeled antibodies may be employed in a wide variety of
assays, employing a wide variety of labels. Detection of the
formation of an antibody-antigen complex between an antibody of the
invention and an epitope of interest can be facilitated by
attaching a detectable substance to the antibody. Suitable
detection means include the use of labels such as radionuclides,
enzymes, coenzymes, fluorescers, chemiluminescers, chromogens,
enzyme substrates or co-factors, enzyme inhibitors, prosthetic
group complexes, free radicals, particles, dyes, and the like.
Examples of suitable enzymes include horseradish peroxidase,
alkaline phosphatase, .beta.-galactosidase, or
acetylcholinesterase; examples of suitable prosthetic group
complexes include streptavidin/biotin and avidin/biotin; examples
of suitable fluorescent materials include umbelliferone,
fluorescein, fluorescein isothiocyanate, rhodamine,
dichlorotriazinylamine fluorescein, dansyl chloride or
phycoerythrin; an example of a luminescent material is luminol;
examples of bioluminescent materials include luciferase, luciferin,
and aequorin; and examples of suitable radioactive material include
125I, 131I, 35S, or 3H. Such labeled reagents may be used in a
variety of well-known assays, such as radioimmunoassays, enzyme
immunoassays, e.g., ELISA, fluorescent immunoassays, and the like.
Labeled antibodies according to the present invention may be thus
be used in such assays for example as described in U.S. Pat. Nos.
3,766,162; 3,791,932; 3,817,837; and 4,233,402.
[0200] An antibody according to the invention may be conjugated to
a therapeutic moiety such as a cytotoxin, a therapeutic agent, or a
radioactive metal ion or radioisotope. Examples of radioisotopes
include, but are not limited to, 1-131, 1-123, 1-125, Y-90, Re-188,
Re-186, At-211, Cu-67, Bi-212, Bi-213, Pd-109, Tc-99, In-111, and
the like. Such antibody conjugates can be used for modifying a
given biological response; the drug moiety is not to be construed
as limited to classical chemical therapeutic agents. For example,
the drug moiety may be a protein or polypeptide possessing a
desired biological activity. Such proteins may include, for
example, a toxin such as abrin, ricin A, Pseudomonas exotoxin, or
diphtheria toxin.
[0201] Techniques for conjugating such therapeutic moiety to
antibodies are well known. See, for example, Arnon et al. (1985)
"Monoclonal Antibodies for Immunotargeting of Drugs in Cancer
Therapy," in Monoclonal Antibodies and Cancer Therapy, ed. Reisfeld
et al. (Alan R. Liss, Inc.), pp. 243-256; ed. Hellstrom et al.
(1987) "Antibodies for Drug Delivery," in Controlled Drug Delivery,
ed. Robinson et al. (2d ed; Marcel Dekker, Inc.), pp. 623-653;
Thorpe (1985) "Antibody Carriers of Cytotoxic Agents in Cancer
Therapy: A Review," in Monoclonal Antibodies '84: Biological and
Clinical Applications, ed. Pinchera et al. pp. 475-506 (Editrice
Kurtis, Milano, Italy, 1985); "Analysis, Results, and Future
Prospective of the Therapeutic Use of Radiolabeled Antibody in
Cancer Therapy," in Monoclonal Antibodies for Cancer Detection and
Therapy, ed. Baldwin et al. (Academic Press, New York, 1985), pp.
303-316; and Thorpe et al. (1982) Immunol. Rev. 62:119-158.
[0202] Alternatively, an antibody, or antibody fragment thereof,
can be conjugated to a second antibody, or antibody fragment
thereof, to form an antibody heteroconjugate as described in U.S.
Pat. No. 4,676,980. In addition, linkers may be used between the
labels and the antibodies of the invention, e.g., as described in
U.S. Pat. No. 4,831,175. Antibodies or, antigen-binding fragments
thereof may be directly labeled with radioactive iodine, indium,
yttrium, or other radioactive particle known in the art, e.g., as
described in U.S. Pat. No. 5,595,721. Treatment may consist of a
combination of treatment with conjugated and non-conjugated
antibodies administered simultaneously or subsequently e.g., as
described in WO00/52031; WO00/52473.
[0203] Antibodies of the invention may also be attached to a solid
support. Additionally, antibodies of the invention, or functional
antibody fragments thereof, can be chemically modified by covalent
conjugation to a polymer to, for example, increase their
circulating half-life. Examples of polymers, and methods to attach
them to peptides, are shown in U.S. Pat. Nos. 4,766,106; 4,179,337;
4,495,285 and 4,609,546. In some embodiments the polymers may be
selected from polyoxyethylated polyols and polyethylene glycol
(PEG). PEG is soluble in water at room temperature and has the
general formula: R(O--CH.sub.2--CH.sub.2).sub.nO--R, wherein R can
be hydrogen, or a protective group such as an alkyl or alkanol
group. Preferably, the protective group may have between 1 and 8
carbons. For example, the protective group is methyl. The symbol n
is a positive integer. In one embodiment n is between 1 and 1,000.
In another embodiment n is between 2 and 500. Preferably, the PEG
has an average molecular weight between 1,000 and 40,000, more
preferably the PEG has a molecular weight between 2,000 and 20,000,
even more preferably the PEG has a molecular weight between 3,000
and 12,000. Furthermore, PEG may have at least one hydroxy group,
for example the PEG may have a terminal hydroxy group. For example,
it is the terminal hydroxy group which is activated to react with a
free amino group on the inhibitor. However, it will be understood
that the type and amount of the reactive groups may be varied to
achieve a covalently conjugated PEG/antibody of the present
invention.
[0204] Water-soluble polyoxyethylated polyols are also useful in
the present invention. They include polyoxyethylated sorbitol,
polyoxyethylated glucose, polyoxyethylated glycerol (POG), and the
like. In one embodiment, POG is used. Without being bound by any
theory, because the glycerol backbone of polyoxyethylated glycerol
is the same backbone occurring naturally in, for example, animals
and humans in mono-, di-, triglycerides, this branching would not
necessarily be seen as a foreign agent in the body. POG may have a
molecular weight in the same range as PEG. Another drug delivery
system that can be used for increasing circulatory half-life is the
liposome. Methods of preparing liposome delivery systems are known
to one of skill in the art. Other drug delivery systems are known
in the art and are described in, for example, referenced in
Poznansky et al. (1980) and Poznansky (1984).
[0205] Antibodies of the invention may be provided in purified
form. Typically, the antibody will be present in a composition that
is substantially free of other polypeptides e.g., where less than
90% (by weight), usually less than 60% and more usually less than
50% of the composition is made up of other polypeptides.
[0206] Antibodies of the invention may be immunogenic in non-human
(or heterologous) hosts e.g., in mice. In particular, the
antibodies may have an idiotope that is immunogenic in non-human
hosts, but not in a human host. In particular, antibodies of the
invention for human use include those that cannot be easily
isolated from hosts such as mice, goats, rabbits, rats, non-primate
mammals, etc. and cannot generally be obtained by humanization or
from xeno-mice.
[0207] Production of Antibodies
[0208] Antibodies according to the invention can be made by any
method known in the art. For example, the general methodology for
making monoclonal antibodies using hybridoma technology is well
known (Kohler, G. and Milstein, C., 1975; Kozbar et al. 1983). In
one embodiment, the alternative EBV immortalization method
described in WO2004/076677 is used.
[0209] A preferred method is described in WO 2004/076677. In this
method B cells producing the antibody of the invention are
transformed with EBV and a polyclonal B cell activator. Additional
stimulants of cellular growth and differentiation may optionally be
added during the transformation step to further enhance the
efficiency. These stimulants may be cytokines such as IL-2 and
IL-15. In one aspect, IL-2 is added during the immortalization step
to further improve the efficiency of immortalization, but its use
is not essential. The immortalized B cells produced using these
methods can then be cultured using methods known in the art and
antibodies isolated therefrom.
[0210] Another preferred method is described in WO 2010/046775. In
this method plasma cells are cultured in limited numbers, or as
single plasma cells in microwell culture plates. Antibodies can be
isolated from the plasma cell cultures. Further, from the plasma
cell cultures, RNA can be extracted and PCR can be performed using
methods known in the art. The VH and VL regions of the antibodies
can be amplified by RT-PCR (reverse transcriptase PCR), sequenced
and cloned into an expression vector that is then transfected into
HEK293T cells or other host cells. The cloning of nucleic acid in
expression vectors, the transfection of host cells, the culture of
the transfected host cells and the isolation of the produced
antibody can be done using any methods known to one of skill in the
art.
[0211] The antibodies may be further purified, if desired, using
filtration, centrifugation and various chromatographic methods such
as HPLC or affinity chromatography. Techniques for purification of
antibodies, e.g., monoclonal antibodies, including techniques for
producing pharmaceutical-grade antibodies, are well known in the
art.
[0212] Fragments of the antibodies of the invention can be obtained
from the antibodies by methods that include digestion with enzymes,
such as pepsin or papain, and/or by cleavage of disulfide bonds by
chemical reduction. Alternatively, fragments of the antibodies can
be obtained by cloning and expression of part of the sequences of
the heavy or light chains. Antibody "fragments" include Fab, Fab',
F(ab')2 and Fv fragments. The invention also encompasses
single-chain Fv fragments (scFv) derived from the heavy and light
chains of an antibody of the invention. For example, the invention
includes a scFv comprising the CDRs from an antibody of the
invention. Also included are heavy or light chain monomers and
dimers, single domain heavy chain antibodies, single domain light
chain antibodies, as well as single chain antibodies, e.g., single
chain Fv in which the heavy and light chain variable domains are
joined by a peptide linker.
[0213] Antibody fragments of the invention may impart monovalent or
multivalent interactions and be contained in a variety of
structures as described above. For instance, scFv molecules may be
synthesized to create a trivalent "triabody" or a tetravalent
"tetrabody." The scFv molecules may include a domain of the Fc
region resulting in bivalent minibodies. In addition, the sequences
of the invention may be a component of multispecific molecules in
which the sequences of the invention target the epitopes of the
invention and other regions of the molecule bind to other targets.
Exemplary molecules include, but are not limited to, bispecific
Fab2, trispecific Fab3, bispecific scFv, and diabodies (Holliger
and Hudson, 2005, Nature Biotechnology 9: 1126-1136).
[0214] Standard techniques of molecular biology may be used to
prepare DNA sequences encoding the antibodies or antibody fragments
of the present invention. Desired DNA sequences may be synthesized
completely or in part using oligonucleotide synthesis techniques.
Site-directed mutagenesis and polymerase chain reaction (PCR)
techniques may be used as appropriate.
[0215] Any suitable host cell/vector system may be used for
expression of the DNA sequences encoding the antibody molecules of
the present invention or fragments thereof. Bacterial, for example
E. coli, and other microbial systems may be used, in part, for
expression of antibody fragments such as Fab and F(ab')2 fragments,
and especially Fv fragments and single chain antibody fragments,
for example, single chain Fvs. Eukaryotic, e.g., mammalian, host
cell expression systems may be used for production of larger
antibody molecules, including complete antibody molecules. Suitable
mammalian host cells include, but are not limited to, CHO, HEK293T,
PER.C6, NS0, myeloma or hybridoma cells.
[0216] The present invention also provides a process for the
production of an antibody molecule according to the present
invention comprising culturing a host cell comprising a vector
encoding a nucleic acid of the present invention under conditions
suitable for expression of protein from DNA encoding the antibody
molecule of the present invention, and isolating the antibody
molecule.
[0217] The antibody molecule may comprise only a heavy or light
chain polypeptide, in which case only a heavy chain or light chain
polypeptide coding sequence needs to be used to transfect the host
cells. For production of products comprising both heavy and light
chains, the cell line may be transfected with two vectors, a first
vector encoding a light chain polypeptide and a second vector
encoding a heavy chain polypeptide. Alternatively, a single vector
may be used, the vector including sequences encoding light chain
and heavy chain polypeptides.
[0218] Alternatively, antibodies according to the invention may be
produced by (i) expressing a nucleic acid sequence according to the
invention in a host cell, e.g. by use of a vector according to the
present invention, and (ii) isolating the expressed antibody
product. Additionally, the method may include (iii) purifying the
isolated antibody. Transformed B cells and cultured plasma cells
may be screened for those producing antibodies of the desired
specificity or function.
[0219] The screening step may be carried out by any immunoassay,
e.g., ELISA, by staining of tissues or cells (including transfected
cells), by neutralization assay or by one of a number of other
methods known in the art for identifying desired specificity or
function. The assay may select on the basis of simple recognition
of one or more antigens, or may select on the additional basis of a
desired function e.g., to select neutralizing antibodies rather
than just antigen-binding antibodies, to select antibodies that can
change characteristics of targeted cells, such as their signaling
cascades, their shape, their growth rate, their capability of
influencing other cells, their response to the influence by other
cells or by other reagents or by a change in conditions, their
differentiation status, etc.
[0220] Individual transformed B cell clones may then be produced
from the positive transformed B cell culture. The cloning step for
separating individual clones from the mixture of positive cells may
be carried out using limiting dilution, micromanipulation, single
cell deposition by cell sorting or another method known in the
art.
[0221] Nucleic acid from the cultured plasma cells can be isolated,
cloned and expressed in HEK293T cells or other known host cells
using methods known in the art.
[0222] The immortalized B cell clones or the transfected host-cells
of the invention can be used in various ways e.g., as a source of
monoclonal antibodies, as a source of nucleic acid (DNA or mRNA)
encoding a monoclonal antibody of interest, for research, etc.
[0223] The invention also provides a composition comprising
immortalized B memory cells or transfected host cells that produce
antibodies according to the present invention.
[0224] The immortalized B cell clone or the cultured plasma cells
of the invention may also be used as a source of nucleic acid for
the cloning of antibody genes for subsequent recombinant
expression. Expression from recombinant sources is more common for
pharmaceutical purposes than expression from B cells or hybridomas
e.g., for reasons of stability, reproducibility, culture ease,
etc.
[0225] Thus the invention also provides a method for preparing a
recombinant cell, comprising the steps of: (i) obtaining one or
more nucleic acids (e.g., heavy and/or light chain mRNAs) from the
B cell clone or the cultured plasma cells that encodes the antibody
of interest; (ii) inserting the nucleic acid into an expression
vector and (iii) transfecting the vector into a host cell in order
to permit expression of the antibody of interest in that host
cell.
[0226] Similarly, the invention provides a method for preparing a
recombinant cell, comprising the steps of: (i) sequencing nucleic
acid(s) from the B cell clone or the cultured plasma cells that
encodes the antibody of interest; and (ii) using the sequence
information from step (i) to prepare nucleic acid(s) for insertion
into a host cell in order to permit expression of the antibody of
interest in that host cell. The nucleic acid may, but need not, be
manipulated between steps (i) and (ii) to introduce restriction
sites, to change codon usage, and/or to optimize transcription
and/or translation regulatory sequences.
[0227] Furthermore, the invention also provides a method of
preparing a transfected host cell, comprising the step of
transfecting a host cell with one or more nucleic acids that encode
an antibody of interest, wherein the nucleic acids are nucleic
acids that were derived from an immortalized B cell clone or a
cultured plasma cell of the invention. Thus the procedures for
first preparing the nucleic acid(s) and then using it to transfect
a host cell can be performed at different times by different people
in different places (e.g., in different countries).
[0228] These recombinant cells of the invention can then be used
for expression and culture purposes. They are particularly useful
for expression of antibodies for large-scale pharmaceutical
production. They can also be used as the active ingredient of a
pharmaceutical composition. Any suitable culture technique can be
used, including but not limited to static culture, roller bottle
culture, ascites fluid, hollow-fiber type bioreactor cartridge,
modular minifermenter, stirred tank, microcarrier culture, ceramic
core perfusion, etc.
[0229] Methods for obtaining and sequencing immunoglobulin genes
from B cells or plasma cells are well known in the art (e.g., see
Chapter 4 of Kuby Immunology, 4th edition, 2000).
[0230] The transfected host cell may be a eukaryotic cell,
including yeast and animal cells, particularly mammalian cells
(e.g., CHO cells, NS0 cells, human cells such as PER.C6 or HKB-11
cells, myeloma cells, or a human liver cell), as well as plant
cells, whereby mammalian cells are preferred. Preferred expression
hosts can glycosylate the antibody of the invention, particularly
with carbohydrate structures that are not themselves immunogenic in
humans. In one embodiment the transfected host cell may be able to
grow in serum-free media. In a further embodiment the transfected
host cell may be able to grow in culture without the presence of
animal-derived products. The transfected host cell may also be
cultured to give a cell line.
[0231] The invention also provides a method for preparing one or
more nucleic acid molecules (e.g., heavy and light chain genes)
that encode an antibody of interest, comprising the steps of (i)
preparing an immortalized B cell clone or culturing plasma cells
according to the invention; (ii) obtaining from the B cell clone or
the cultured plasma cells nucleic acid that encodes the antibody of
interest.
[0232] Further, the invention provides a method for obtaining a
nucleic acid sequence that encodes an antibody of interest,
comprising the steps of: (i) preparing an immortalized B cell clone
or culturing plasma cells according to the invention; (ii)
sequencing nucleic acid from the B cell clone or the cultured
plasma cells that encodes the antibody of interest.
[0233] The invention further provides a method of preparing nucleic
acid molecule(s) that encode an antibody of interest, comprising
the step of obtaining the nucleic acid that was obtained from a
transformed B cell clone or cultured plasma cells of the invention.
Thus the procedures for first obtaining the B cell clone or the
cultured plasma cell, and then obtaining nucleic acid(s) from the B
cell clone or the cultured plasma cells can be performed at
different times by different people in different places (e.g., in
different countries).
[0234] The invention also comprises a method for preparing an
antibody (e.g., for pharmaceutical use) according to the present
invention, comprising the steps of: (i) obtaining and/or sequencing
one or more nucleic acids (e.g., heavy and light chain genes) from
the selected B cell clone or the cultured plasma cells expressing
the antibody of interest; (ii) inserting the nucleic acid(s) into
or using the nucleic acid(s) sequence(s) to prepare an expression
vector; (iii) transfecting a host cell that can express the
antibody of interest; (iv) culturing or sub-culturing the
transfected host cells under conditions where the antibody of
interest is expressed; and, optionally, (v) purifying the antibody
of interest.
[0235] The invention also provides a method of preparing an
antibody comprising the steps of: culturing or sub-culturing a
transfected host cell population, e.g. a stably transfected host
cell population, under conditions where the antibody of interest is
expressed and, optionally, purifying the antibody of interest,
wherein said transfected host cell population has been prepared by
(i) providing nucleic acid(s) encoding a selected antibody of
interest that is produced by a B cell clone or cultured plasma
cells prepared as described above, (ii) inserting the nucleic
acid(s) into an expression vector, (iii) transfecting the vector in
a host cell that can express the antibody of interest, and (iv)
culturing or sub-culturing the transfected host cell comprising the
inserted nucleic acids to produce the antibody of interest. Thus
the procedures for first preparing the recombinant host cell and
then culturing it to express antibody can be performed at very
different times by different people in different places (e.g.,
indifferent countries).
[0236] Pharmaceutical Composition
[0237] The present invention also provides a pharmaceutical
composition comprising one or more of: [0238] (i) the antibody, or
the antibody fragment thereof, according to the present invention;
[0239] (ii) the nucleic acid encoding the antibody, or antibody
fragments according to the present invention; [0240] (iii) the
vector comprising the nucleic acid according to the present
invention; and/or [0241] (iv) the cell expressing the antibody
according to the present invention or comprising the vector
according to the present invention.
[0242] In other words, the present invention also provides a
pharmaceutical composition comprising the antibody, or the antigen
binding fragment thereof, according to the present invention, the
nucleic acid according to the present invention, the vector
according to the present invention and/or the cell according to the
present invention.
[0243] The pharmaceutical composition may preferably also contain a
pharmaceutically acceptable carrier, diluent and/or excipient.
Although the carrier or excipient may facilitate administration, it
should not itself induce the production of antibodies harmful to
the individual receiving the composition. Nor should it be toxic.
Suitable carriers may be large, slowly metabolized macromolecules
such as proteins, polypeptides, liposomes, polysaccharides,
polylactic acids, polyglycolic acids, polymeric amino acids, amino
acid copolymers and inactive virus particles. In general,
pharmaceutically acceptable carriers in a pharmaceutical
composition according to the present invention may be active
components or inactive components. Preferably, the pharmaceutically
acceptable carrier in a pharmaceutical composition according to the
present invention is not an active component in respect to Zika
virus infection.
[0244] Pharmaceutically acceptable salts can be used, for example
mineral acid salts, such as hydrochlorides, hydrobromides,
phosphates and sulphates, or salts of organic acids, such as
acetates, propionates, malonates and benzoates.
[0245] Pharmaceutically acceptable carriers in a pharmaceutical
composition may additionally contain liquids such as water, saline,
glycerol and ethanol. Additionally, auxiliary substances, such as
wetting or emulsifying agents or pH buffering substances, may be
present in such compositions. Such carriers enable the
pharmaceutical compositions to be formulated as tablets, pills,
dragees, capsules, liquids, gels, syrups, slurries and suspensions,
for ingestion by the subject.
[0246] Pharmaceutical compositions of the invention may be prepared
in various forms. For example, the compositions may be prepared as
injectables, either as liquid solutions or suspensions. Solid forms
suitable for solution in, or suspension in, liquid vehicles prior
to injection can also be prepared (e.g., a lyophilized composition,
similar to Synagis.TM. and Herceptin.TM., for reconstitution with
sterile water containing a preservative). The composition may be
prepared for topical administration e.g., as an ointment, cream or
powder. The composition may be prepared for oral administration
e.g., as a tablet or capsule, as a spray, or as a syrup (optionally
flavored). The composition may be prepared for pulmonary
administration e.g., as an inhaler, using a fine powder or a spray.
The composition may be prepared as a suppository or pessary. The
composition may be prepared for nasal, aural or ocular
administration e.g., as drops. The composition may be in kit form,
designed such that a combined composition is reconstituted just
prior to administration to a subject. For example, a lyophilized
antibody may be provided in kit form with sterile water or a
sterile buffer.
[0247] It is preferred that the active ingredient in the
composition is an antibody molecule, an antibody fragment or
variants and derivatives thereof, in particular the active
ingredient in the composition is an antibody, an antibody fragment
or variants and derivatives thereof, according to the present
invention. As such, it may be susceptible to degradation in the
gastrointestinal tract. Thus, if the composition is to be
administered by a route using the gastrointestinal tract, the
composition may contain agents which protect the antibody from
degradation but which release the antibody once it has been
absorbed from the gastrointestinal tract.
[0248] A thorough discussion of pharmaceutically acceptable
carriers is available in Gennaro (2000) Remington: The Science and
Practice of Pharmacy, 20th edition, ISBN: 0683306472.
[0249] Pharmaceutical compositions of the invention generally have
a pH between 5.5 and 8.5, in some embodiments this may be between 6
and 8, and in other embodiments about 7. The pH may be maintained
by the use of a buffer. The composition may be sterile and/or
pyrogen free. The composition may be isotonic with respect to
humans. In one embodiment pharmaceutical compositions of the
invention are supplied in hermetically-sealed containers.
[0250] Within the scope of the invention are compositions present
in several forms of administration; the forms include, but are not
limited to, those forms suitable for parenteral administration,
e.g., by injection or infusion, for example by bolus injection or
continuous infusion. Where the product is for injection or
infusion, it may take the form of a suspension, solution or
emulsion in an oily or aqueous vehicle and it may contain
formulatory agents, such as suspending, preservative, stabilizing
and/or dispersing agents. Alternatively, the antibody molecule may
be in dry form, for reconstitution before use with an appropriate
sterile liquid. A vehicle is typically understood to be a material
that is suitable for storing, transporting, and/or administering a
compound, such as a pharmaceutically active compound, in particular
the antibodies according to the present invention. For example, the
vehicle may be a physiologically acceptable liquid, which is
suitable for storing, transporting, and/or administering a
pharmaceutically active compound, in particular the antibodies
according to the present invention. Once formulated, the
compositions of the invention can be administered directly to the
subject. In one embodiment the compositions are adapted for
administration to mammalian, e.g., human subjects.
[0251] The pharmaceutical compositions of this invention may be
administered by any number of routes including, but not limited to,
oral, intravenous, intramuscular, intra-arterial, intramedullary,
intraperitoneal, intrathecal, intraventricular, transdermal,
transcutaneous, topical, subcutaneous, intranasal, enteral,
sublingual, intravaginal or rectal routes. Hyposprays may also be
used to administer the pharmaceutical compositions of the
invention. Preferably, the pharmaceutical composition may be
prepared for oral administration, e.g. as tablets, capsules and the
like, for topical administration, or as injectable, e.g. as liquid
solutions or suspensions, whereby it is particularly preferred that
the pharmaceutical composition is an injectable. Solid forms
suitable for solution in, or suspension in, liquid vehicles prior
to injection are also be preferred, e.g. that the pharmaceutical
composition is in lyophilized form.
[0252] For injection, e.g. intravenous, cutaneous or subcutaneous
injection, or injection at the site of affliction, the active
ingredient will preferably be in the form of a parenterally
acceptable aqueous solution which is pyrogen-free and has suitable
pH, isotonicity and stability. Those of relevant skill in the art
are well able to prepare suitable solutions using, for example,
isotonic vehicles such as Sodium Chloride Injection, Ringer's
Injection, Lactated Ringer's Injection. Preservatives, stabilizers,
buffers, antioxidants and/or other additives may be included, as
required. Whether it is a polypeptide, peptide, or nucleic acid
molecule, other pharmaceutically useful compound according to the
present invention that is to be given to an individual,
administration is preferably in a "prophylactically effective
amount" or a "therapeutically effective amount" (as the case may
be), this being sufficient to show benefit to the individual. The
actual amount administered, and rate and time-course of
administration, will depend on the nature and severity of what is
being treated. For injection, the pharmaceutical composition
according to the present invention may be provided for example in a
pre-filled syringe.
[0253] The inventive pharmaceutical composition as defined above
may also be administered orally in any orally acceptable dosage
form including, but not limited to, capsules, tablets, aqueous
suspensions or solutions. In the case of tablets for oral use,
carriers commonly used include lactose and corn starch. Lubricating
agents, such as magnesium stearate, are also typically added. For
oral administration in a capsule form, useful diluents include
lactose and dried cornstarch. When aqueous suspensions are required
for oral use, the active ingredient, i.e. the inventive transporter
cargo conjugate molecule as defined above, is combined with
emulsifying and suspending agents. If desired, certain sweetening,
flavoring or coloring agents may also be added.
[0254] The inventive pharmaceutical composition may also be
administered topically, especially when the target of treatment
includes areas or organs readily accessible by topical application,
e.g. including diseases of the skin or of any other accessible
epithelial tissue. Suitable topical formulations are readily
prepared for each of these areas or organs. For topical
applications, the inventive pharmaceutical composition may be
formulated in a suitable ointment, containing the inventive
pharmaceutical composition, particularly its components as defined
above, suspended or dissolved in one or more carriers. Carriers for
topical administration include, but are not limited to, mineral
oil, liquid petrolatum, white petrolatum, propylene glycol,
polyoxyethylene, polyoxypropylene compound, emulsifying wax and
water. Alternatively, the inventive pharmaceutical composition can
be formulated in a suitable lotion or cream. In the context of the
present invention, suitable carriers include, but are not limited
to, mineral oil, sorbitan monostearate, polysorbate 60, cetyl
esters wax, cetearyl alcohol, 2-octyldodecanol, benzyl alcohol and
water.
[0255] Dosage treatment may be a single dose schedule or a multiple
dose schedule. In particular, the pharmaceutical composition may be
provided as single-dose product. Preferably, the amount of the
antibody in the pharmaceutical composition--in particular if
provided as single-dose product--does not exceed 200 mg, more
preferably does not exceed 100 mg, and even more preferably does
not exceed 50 mg.
[0256] For example, the pharmaceutical composition according to the
present invention may be administered daily, e.g. once or several
times per day, e.g. once, twice, three times or four times per day,
preferably once or twice per day, more preferable once per day, for
1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19,
20 or 21 or more days, e.g. daily for 1, 2, 3, 4, 5, 6 months.
Preferably, the pharmaceutical composition according to the present
invention may be administered weekly, e.g. once or twice per week,
for 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18,
19, 20 or 21 or more weeks, e.g. weekly for 1, 2, 3, 4, 5, 6, 7, 8,
9, 10, 11, or 12 months or weekly for 2, 3, 4, or 5 years.
Moreover, the pharmaceutical composition according to the present
invention may be preferably administered monthly, e.g. once per
month or, more preferably, every second month for 1, 2, 3, 4, or 5
or more years. It is also preferred that the administration
continues for the lifetime. In addition, also one single
administration only is also envisaged, in particular in respect to
certain indications, e.g. for prevention of Zika virus infection in
case of accidental exposure, e.g. in non-immunised subjects.
However, the most preferred treatment schedule is post-exposure
prophylaxis (PEP), wherein one or more single doses are
administered as soon as possible after Zika infection. A
prophylactic setting, wherein one or more single doses are
administered to prevent Zika infection (i.e. before Zika infection,
in particular in non-Zika-immunised subjects) is also
preferred.
[0257] In particular, it is preferred that for a single dose, e.g.
a daily, weekly or monthly dose, preferably for a weekly dose, the
amount of the antibody, or the antigen binding fragment thereof, in
the pharmaceutical composition according to the present invention,
does not exceed 1 g, preferably does not exceed 500 mg, more
preferably does not exceed 200 mg, even more preferably does not
exceed 100 mg, and particularly preferably does not exceed 50
mg.
[0258] Pharmaceutical compositions typically include an "effective"
amount of one or more antibodies of the invention, i.e. an amount
that is sufficient to treat, ameliorate, attenuate or prevent a
desired disease or condition, or to exhibit a detectable
therapeutic effect. Therapeutic effects also include reduction or
attenuation in pathogenic potency or physical symptoms. The precise
effective amount for any particular subject will depend upon their
size, weight, and health, the nature and extent of the condition,
and the therapeutics or combination of therapeutics selected for
administration. The effective amount for a given situation is
determined by routine experimentation and is within the judgment of
a clinician. For purposes of the present invention, an effective
dose will generally be from about 0.005 to about 100 mg/kg,
preferably from about 0.0075 to about 50 mg/kg, more preferably
from about 0.01 to about 10 mg/kg, and even more preferably from
about 0.02 to about 5 mg/kg, of the antibody of the present
invention (e.g. amount of the antibody in the pharmaceutical
composition) in relation to the bodyweight (e.g., in kg) of the
individual to which it is administered.
[0259] Moreover, the pharmaceutical composition according to the
present invention may also comprise an additional active component,
which may be a further antibody or a component, which is not an
antibody. The additional active component is preferably a
checkpoint inhibitor. It is also preferred that a ZIKV neutralizing
antibody, or an antigen binding fragment thereof, as described
herein is combined with a ZIKV NS1-binding antibody, or an antigen
binding fragment thereof, as described herein as additional active
component (co-agent). Thereby, the pathogenic role of NS1 may be
blocked in addition to neutralization of ZIKV. The pharmaceutical
composition according to the present invention may comprise one or
more of the additional active components, e.g. as described as
co-agents below in the context of a combination therapy.
[0260] The antibody, or the antigen binding fragment, according to
the present invention can be present either in the same
pharmaceutical composition as the additional active component or,
preferably, the antibody, or the antigen binding fragment,
according to the present invention is comprised by a first
pharmaceutical composition and the additional active component is
comprised by a second pharmaceutical composition different from the
first pharmaceutical composition. Accordingly, if more than one
additional active component is envisaged, each additional active
component and the antibody, or the antigen binding fragment,
according to the present invention is preferably comprised by a
different pharmaceutical composition. Such different pharmaceutical
compositions may be administered either combined/simultaneously or
at separate times or at separate locations (e.g. separate parts of
the body).
[0261] Preferably, antibody, or the antigen binding fragment,
according to the present invention and the additional active
component provide an additive therapeutic effect or, preferably, a
synergistic therapeutic effect. The term "synergy" is used to
describe a combined effect of two or more active agents that is
greater than the sum of the individual effects of each respective
active agent. Thus, where the combined effect of two or more agents
results in "synergistic inhibition" of an activity or process, it
is intended that the inhibition of the activity or process is
greater than the sum of the inhibitory effects of each respective
active agent. The term "synergistic therapeutic effect" refers to a
therapeutic effect observed with a combination of two or more
therapies wherein the therapeutic effect (as measured by any of a
number of parameters) is greater than the sum of the individual
therapeutic effects observed with the respective individual
therapies.
[0262] A pharmaceutical composition comprising the antibody
according to gZKA190, gZKA64, gZKA230, gZKA185, gZKA78 or an
antigen binding fragment thereof, and a pharmaceutically acceptable
carrier is preferred.
[0263] In one embodiment, a composition of the invention may
include antibodies of the invention, wherein the antibodies may
make up at least 50% by weight (e.g., 60%, 70%, 75%, 80%, 85%, 90%,
95%, 96%, 97%, 98%, 99% or more) of the total protein in the
composition. In such a composition, the antibodies are preferably
in purified form.
[0264] The present invention also provides a method of preparing a
pharmaceutical composition comprising the steps of: (i) preparing
an antibody of the invention; and (ii) admixing the purified
antibody with one or more pharmaceutically-acceptable carriers.
[0265] In another embodiment, a method of preparing a
pharmaceutical composition comprises the step of: admixing an
antibody with one or more pharmaceutically-acceptable carriers,
wherein the antibody is a monoclonal antibody that was obtained
from a transformed B cell or a cultured plasma cell of the
invention.
[0266] As an alternative to delivering antibodies or B cells for
therapeutic purposes, it is possible to deliver nucleic acid
(typically DNA) that encodes the monoclonal antibody (or active
fragment thereof) of interest derived from the B cell or the
cultured plasma cells to a subject, such that the nucleic acid can
be expressed in the subject in situ to provide a desired
therapeutic effect. Suitable gene therapy and nucleic acid delivery
vectors are known in the art.
[0267] Pharmaceutical compositions may include an antimicrobial
particularly if packaged in a multiple dose format. They may
comprise detergent e.g., a Tween (polysorbate), such as Tween 80.
Detergents are generally present at low levels e.g., less than
0.01%. Compositions may also include sodium salts (e.g., sodium
chloride) to give tonicity. For example, a concentration of
10-.+-.2 mg/ml NaCl is typical.
[0268] Further, pharmaceutical compositions may comprise a sugar
alcohol (e.g., mannitol) or a disaccharide (e.g., sucrose or
trehalose) e.g., at around 15-30 mg/ml (e.g., 25 mg/ml),
particularly if they are to be lyophilized or if they include
material which has been reconstituted from lyophilized material.
The pH of a composition for lyophilization may be adjusted to
between 5 and 8, or between 5.5 and 7, or around 6.1 prior to
lyophilization.
[0269] The compositions of the invention may also comprise one or
more immunoregulatory agents. In one embodiment, one or more of the
immunoregulatory agents include(s) an adjuvant.
[0270] Medical Treatments, Kits and Uses
[0271] Medical Treatments
[0272] In a further aspect, the present invention provides the use
of the antibody, or an antigen binding fragment thereof, according
to the present invention, the nucleic acid according to the present
invention, the vector according to the present invention, the cell
according to the present invention or the pharmaceutical
composition according to the present invention in (i) prevention
and/or treatment of Zika virus infection; or in (ii) diagnosis of
Zika virus infection. Thereby, use of the antibody, or an antigen
binding fragment thereof, according to the present invention (and
in particular its preferred embodiments as described above) is
preferred in (i) prevention and/or treatment of Zika virus
infection as described herein; or in (ii) diagnosis of Zika virus
infection as described herein.
[0273] Methods of diagnosis may include contacting an antibody or
an antibody fragment with a sample. Such samples may be isolated
from a subject, for example an isolated tissue sample taken from,
for example, nasal passages, sinus cavities, salivary glands, lung,
liver, pancreas, kidney, ear, eye, placenta, alimentary tract,
heart, ovaries, pituitary, adrenals, thyroid, brain, skin or blood,
preferably plasma or serum. The methods of diagnosis may also
include the detection of an antigen/antibody complex, in particular
following the contacting of an antibody or an antibody fragment
with a sample. Such a detection step is typically performed at the
bench, i.e. without any contact to the human or animal body.
Examples of detection methods are well-known to the person skilled
in the art and include, e.g., ELISA (enzyme-linked immunosorbent
assay).
[0274] Prevention of Zika virus infection refers in particular to
prophylactic settings, wherein the subject was not diagnosed with
Zika virus infection (either no diagnosis was performed or
diagnosis results were negative) and/or the subject does not show
symptoms of Zika virus infection. Accordingly, prevention of Zika
virus infection includes "post-exposure prophylaxis" (PEP), i.e.
preventive treatment after a possible Zika virus infection, for
example after a mosquito bite in a Zika virus affected area.
Prevention of Zika virus infection is in particular useful in
high-risk subjects, such as in pregnant subjects and/or in subjects
staying in Zika virus affected areas (such as subjects living in
Zika virus affected areas or travelling to Zika virus affected
areas).
[0275] In therapeutic settings, in contrast, the subject is
typically infected by Zika virus, diagnosed with Zika virus
infection and/or showing symptoms of Zika virus infection. Of note,
the terms "treatment" and "therapy"/"therapeutic" of ZIKV infection
include (complete) cure as well as attenuation of ZIKV
infection.
[0276] Preferred methods of diagnosis of Zika virus infection are
the diagnosis methods as described herein, e.g. using the
neutralizing antibody, or antigen binding fragment thereof,
according to the present invention and/or the ZIKV NS1-binding
antibody, or antigen binding fragment thereof, according to the
present invention.
[0277] Accordingly, the antibody, or an antigen binding fragment
thereof, according to the present invention, the nucleic acid
according to the present invention, the vector according to the
present invention, the cell according to the present invention or
the pharmaceutical composition according to the present invention
is preferably used for treatment of Zika virus infection in
subjects diagnosed with Zika virus infection or in subjects showing
symptoms of Zika infection.
[0278] It is also preferred that the antibody, or an antigen
binding fragment thereof, according to the present invention, the
nucleic acid according to the present invention, the vector
according to the present invention, the cell according to the
present invention or the pharmaceutical composition according to
the present invention is used for prevention and/or treatment of
Zika virus infection in asymptomatic subjects. Those subjects may
be diagnosed or not diagnosed with Zika virus infection.
[0279] Preferably, the antibody, or an antigen binding fragment
thereof, according to the present invention, the nucleic acid
according to the present invention, the vector according to the
present invention, the cell according to the present invention or
the pharmaceutical composition according to the present invention
is used for prevention and/or treatment of Zika virus infection in
pregnant subjects, in particular to prevent congenital infection.
For example, this may be performed in a similar manner as for the
prevention of HCMV congential infection as described in Nigro G,
Adler S P, La Torre R, Best A M, Congenital Cytomegalovirus
Collaborating Group: Passive immunization during pregnancy for
congenital cytomegalovirus infection; N Engl J Med 2005,
353:1350-1362.
[0280] Without being bound to any theory, it is assumed that the
antibody, or the antigen-binding fragment thereof, according to the
present invention can pass the placenta through the interaction
with FcRn if administered to a pregnant subject, e.g. by (i.v.)
injection or any other route of administration as described herein.
Importantly, the interaction of "LALA" variants of antibodies as
described herein with FcRn is not compromised. It is believed that
FcRn is already expressed in the first trimester in the
placenta.
[0281] Alternatively, the antibody, or the antigen-binding fragment
thereof, according to the present invention may also be
administered to the extra-amniotic space.
[0282] Preferably, the antibody, or an antigen binding fragment
thereof, according to the present invention, the nucleic acid
according to the present invention, the vector according to the
present invention, the cell according to the present invention or
the pharmaceutical composition according to the present invention
is used for prevention and/or treatment of Zika virus infection,
wherein the antibody, or the antigen binding fragment thereof, the
nucleic acid, the vector, the cell, or the pharmaceutical
composition is administered up to seven days after (a possible)
Zika virus infection, preferably up to five days after (a possible)
Zika virus infection, more preferably up to four days after (a
possible) Zika virus infection, even more preferably up to three
days after (a possible) Zika virus infection, and most preferably
up to one day after (a possible) Zika virus infection. Such a
treatment schedule may be useful in therapeutic settings as well as
in prophylactic settings, in particular in post-exposure
prophylaxis (PEP).
[0283] In PEP typically the first administration of the the
antibody, or an antigen binding fragment thereof, according to the
present invention, the nucleic acid according to the present
invention, the vector according to the present invention, the cell
according to the present invention or the pharmaceutical
composition according to the present invention is as soon as
possible after a possible ZIKV infection, e.g. after a mosquito
bite in a ZIKV affected area. Accordingly, in PEP the first
administration of the the antibody, or an antigen binding fragment
thereof, according to the present invention, the nucleic acid
according to the present invention, the vector according to the
present invention, the cell according to the present invention or
the pharmaceutical composition according to the present invention
is typically up to one or more days after (a possible) ZIKV
infection, as described above.
[0284] It is also preferred that the antibody, or an antigen
binding fragment thereof, according to the present invention, the
nucleic acid according to the present invention, the vector
according to the present invention, the cell according to the
present invention or the pharmaceutical composition according to
the present invention is used for prevention and/or treatment of
Zika virus infection, wherein the antibody, or the antigen binding
fragment thereof, the nucleic acid, the vector, the cell, or the
pharmaceutical composition is administered up to three months
before (a possible) Zika virus infection, preferably up to one
month before (a possible) Zika virus infection, more preferably up
to two weeks before (a possible) Zika virus infection, even more
preferably up to one week before (a possible) Zika virus infection,
and most preferably up to one day before (a possible) Zika virus
infection. Such a treatment schedule refers in particular to a
prophylactic setting.
[0285] In general--and in particular in PEP--after the first
administration of the antibody, or an antigen binding fragment
thereof, according to the present invention, the nucleic acid
according to the present invention, the vector according to the
present invention, the cell according to the present invention or
the pharmaceutical composition according to the present invention,
one or more subsequent administrations may follow, preferably a
single dose per day or per every second day for 1, 2, 3, 4, 5, 6,
7, 8, 9, 10, 11, 12, 13, 1, 15, 16, 17, 18, 19, 20, or 21 days. It
is also preferred that after the first administration of the
antibody, or an antigen binding fragment thereof, according to the
present invention, the nucleic acid according to the present
invention, the vector according to the present invention, the cell
according to the present invention or the pharmaceutical
composition according to the present invention, one or more
subsequent administrations may follow, preferably a single dose
once or twice per week for 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12,
13, 1, 15, 16, 17, 18, 19, 20, or 21 weeks. It is also preferred
that after the first administration of the antibody, or an antigen
binding fragment thereof, according to the present invention, the
nucleic acid according to the present invention, the vector
according to the present invention, the cell according to the
present invention or the pharmaceutical composition according to
the present invention, one or more subsequent administrations may
follow, preferably a single dose every 2 or 4 weeks for 1, 2, 3, 4,
5, 6, 7, 8, 9, 10, 11, 12, 13, 1, 15, 16, 17, 18, 19, 20, or 21
weeks. It is also preferred that after the first administration of
the antibody, or an antigen binding fragment thereof, according to
the present invention, the nucleic acid according to the present
invention, the vector according to the present invention, the cell
according to the present invention or the pharmaceutical
composition according to the present invention, one or more
subsequent administrations may follow, preferably a single dose
every two or four months for 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12,
13, 1, 15, 16, 17, 18, 19, 20, or 21 months. It is also preferred
that after the first administration of the antibody, or an antigen
binding fragment thereof, according to the present invention, the
nucleic acid according to the present invention, the vector
according to the present invention, the cell according to the
present invention or the pharmaceutical composition according to
the present invention, one or more subsequent administrations may
follow, preferably a single dose once or twice per year for 1, 2,
3, 4, 5, 6, 7, 8, 9, or 10 years.
[0286] Preferably, the antibody, or an antigen binding fragment
thereof, according to the present invention, the nucleic acid
according to the present invention, the vector according to the
present invention, the cell according to the present invention or
the pharmaceutical composition according to the present invention
is administered at a (single) dose of 0.005 to 100 mg/kg
bodyweight, preferably at a (single) dose of 0.0075 to 50 mg/kg
bodyweight, more preferably at a (single) dose of 0.01 to 10 mg/kg
bodyweight, even more preferably at a (single) dose of 0.05 to 5
mg/kg bodyweight, and particularly preferably at a (single) dose of
0.1 to 1 mg/kg bodyweight.
[0287] The antibody, or an antigen binding fragment thereof,
according to the present invention, the nucleic acid according to
the present invention, the vector according to the present
invention, the cell according to the present invention or the
pharmaceutical composition according to the present invention may
be administered by any number of routes such as oral, intravenous,
intramuscular, intra-arterial, intramedullary, intraperitoneal,
intrathecal, intraventricular, transdermal, transcutaneous,
topical, subcutaneous, intranasal, enteral, sublingual,
intravaginal or rectal routes. Intravenous administration, or
subcutaneous administration or intramuscular administration are
preferred and intravenous administration or subcutaneous
administration are more preferred.
[0288] In pregnant subjects the antibody, or an antigen binding
fragment thereof, according to the present invention may also be
administered intra- or extra-amniotic, e.g. by injection.
[0289] Accordingly, the present invention also provides a method of
preventing and/or treating Zika virus infection in a subject,
wherein the method comprises administering to a subject in need
thereof the antibody, or an antigen binding fragment thereof,
according to the present invention, the nucleic acid according to
the present invention, the vector according to the present
invention, the cell according to the present invention or the
pharmaceutical composition according to the present invention.
Preferred embodiments of this method correspond to preferred
embodiments of the medical use as described above (and below,
regarding combination therapy). For example, a preferred subject in
this method is a subject diagnosed with Zika virus infection or
showing symptoms of Zika virus infection. Another preferred subject
in this method is a pregnant subject.
[0290] Combination Therapy
[0291] The administration of the antibody, or an antigen binding
fragment thereof, according to the present invention, the nucleic
acid according to the present invention, the vector according to
the present invention, the cell according to the present invention
or the pharmaceutical composition according to the present
invention in the methods and uses according to the invention can be
carried out alone or in combination with a co-agent (also referred
to as "additional active component" herein), which is in particular
useful for preventing and/or treating ZIKV infection.
[0292] The invention encompasses the administration of the
antibody, or an antigen binding fragment thereof, according to the
present invention, the nucleic acid according to the present
invention, the vector according to the present invention, the cell
according to the present invention or the pharmaceutical
composition according to the present invention, wherein it is
administered to a subject prior to, simultaneously or sequentially
with other therapeutic regimens or co-agents useful for treating
and/or preventing ZIKV infection. Said antibody, nucleic acid,
vector, cell or pharmaceutical composition, that is administered
simultaneously with said co-agents can be administered in the same
or different composition(s) and by the same or different route(s)
of administration.
[0293] Said other therapeutic regimens or co-agents may be, for
example, a checkpoint inhibitor.
[0294] Thus, in another aspect of the present invention the
antibody, or an antigen binding fragment thereof, according to the
present invention, the nucleic acid according to the present
invention, the vector according to the present invention, the cell
according to the present invention or the pharmaceutical
composition according to the present invention is administered in
combination with a checkpoint inhibitor for the (medical) uses as
described herein.
[0295] Preferred checkpoint inhibitors are directed to a blockade
of PD-1/PD-L1 and/or of CTLA4 and, thus, include anti-PD-1
antibodies, anti-PD-L1 antibodies and anti-CTLA4 antibodies. Thus,
the pharmaceutical composition according to the present invention
may comprise one or more of the additional active components.
[0296] It is also preferred that a ZIKV neutralizing antibody, or
an antigen binding fragment thereof, as described herein is
combined with a ZIKV NS1-binding antibody, or an antigen binding
fragment thereof, as described herein as additional active
component (co-agent). Thereby, the pathogenic role of NS1 may be
blocked in addition to neutralization of ZIKV. Accordingly, a ZIKV
NS-binding antibody, or an antigen binding fragment thereof, as
described herein is a preferred additional active component
(co-agent).
[0297] The antibody, or the antigen binding fragment, according to
the present invention can be present either in the same
pharmaceutical composition as the additional active component
(co-agent) or, preferably, the antibody, or the antigen binding
fragment, according to the present invention is comprised by a
first pharmaceutical composition and the additional active
component (co-agent) is comprised by a second pharmaceutical
composition different from the first pharmaceutical composition.
Accordingly, if more than one additional active component
(co-agent) is envisaged, each additional active component
(co-agent) and the antibody, or the antigen binding fragment,
according to the present invention is preferably comprised by a
different pharmaceutical composition. Such different pharmaceutical
compositions may be administered either combined/simultaneously or
at separate times or at separate locations (e.g. separate parts of
the body).
[0298] Preferably, the antibody, or the antigen binding fragment,
according to the present invention and the additional active
component (co-agent) provide an additive therapeutic effect or,
preferably, a synergistic therapeutic effect. The term "synergy" is
used to describe a combined effect of two or more active agents
that is greater than the sum of the individual effects of each
respective active agent. Thus, where the combined effect of two or
more agents results in "synergistic inhibition" of an activity or
process, it is intended that the inhibition of the activity or
process is greater than the sum of the inhibitory effects of each
respective active agent. The term "synergistic therapeutic effect"
refers to a therapeutic effect observed with a combination of two
or more therapies wherein the therapeutic effect (as measured by
any of a number of parameters) is greater than the sum of the
individual therapeutic effects observed with the respective
individual therapies.
[0299] Further Use and Kits
[0300] In a further aspect, the present invention also provides the
use of the antibody, or an antigen binding fragment thereof,
according to the present invention, the nucleic acid according to
the present invention, the vector according to the present
invention, the cell according to the present invention or the
pharmaceutical composition according to the present invention for
monitoring the quality of an anti-Zika vaccine by checking that the
antigen of said vaccine contains the specific epitope in the
correct conformation. Preferred antigens comprised by such as
anti-Zika vaccine to be checked include ZIKV envelope protein or
any other molecule/complex comprising or consisting of (i) domain
III of ZIKV E protein (EDIII) as described above or (ii) a
quaternary ZIKV epitope as described above.
[0301] Moreover, the present invention also provides the use of the
antibody, or an antigen binding fragment thereof, according to the
present invention, the nucleic acid according to the present
invention, the vector according to the present invention, the cell
according to the present invention or the pharmaceutical
composition according to the present invention in diagnosis of Zika
virus infection.
[0302] In addition also the use of the antibody, or an antigen
binding fragment thereof, according to the present invention, the
nucleic acid according to the present invention, the vector
according to the present invention, the cell according to the
present invention or the pharmaceutical composition according to
the present invention in determining whether an isolated blood
sample (e.g., whole blood, serum and/or plasma) is infected with
Zika virus is provided.
[0303] As described above, methods of diagnosis may include
contacting an antibody or an antibody fragment with a sample. Such
samples may be isolated from a subject, for example an isolated
tissue sample taken from, for example, nasal passages, sinus
cavities, salivary glands, lung, liver, pancreas, kidney, ear, eye,
placenta, alimentary tract, heart, ovaries, pituitary, adrenals,
thyroid, brain, skin or blood, preferably serum or plasma. The
methods of diagnosis may also include the detection of an
antigen/antibody complex, in particular following the contacting of
an antibody or an antibody fragment with a sample. Such a detection
step is typically performed at the bench, i.e. without any contact
to the human or animal body. Examples of detection methods are
well-known to the person skilled in the art and include, e.g.,
ELISA (enzyme-linked immunosorbent assay).
[0304] In a further aspect, the present invention also provides a
kit of parts comprising at least one antibody, or antigen binding
fragment thereof, according to the present invention, at least one
nucleic acid according to the present invention, at least one
vector according to the present invention, at least one cell
according to the present invention, and/or at least one
pharmaceutical composition according to the present invention. In
addition, the kit may comprise means for administration of the
antibody, or an antigen binding fragment thereof, according to the
present invention, the nucleic acid according to the present
invention, the vector according to the present invention, the cell
according to the present invention or the pharmaceutical
composition according to the present invention, such as a syringe
or a vessel, a leaflet, and/or a co-agent to be administered as
described above.
[0305] Antibodies Specifically Binding to NS1 Protein of Zika
Virus
[0306] In a further aspect, the present invention also provides an
isolated antibody, or an antigen binding fragment thereof, that
specifically binds to NS1 protein of ZIKA virus (ZIKV).
[0307] ZIKV NS1 protein (non-structural protein 1) occurs
intracellular, secreted and cell-surface associated and in
particular secreted ZIKV NS1 protein is typically found in body
fluids, such as serum, saliva, urine etc., of subjects infected
with ZIKV. Secreted and cell-surface-associated NS1 are highly
immunogenic and elicit production of antibodies. NS1 is known to be
an important biomarker for early diagnosis of ZIKV infection.
Accordingly, the antibody, or an antigen binding fragment thereof,
according to the present invention that specifically binds to NS1
protein of ZIKA virus (ZIKV), is for example useful in diagnosis of
ZIKV infection.
[0308] In general, binding may be assessed by standard ELISA as
known to the skilled person and as described above. Thereby, the
relative affinities of antibody binding may be determined by
measuring the concentration of the antibody (EC.sub.50) required to
achieve 50% maximal binding at saturation. Preferably, the
EC.sub.50 of the antibody, or an antigen binding fragment thereof,
according to the present invention to ZIKV NS1 protein is no more
than 50 ng/ml, preferably said EC.sub.50 is no more than 25 ng/ml,
more preferably said EC.sub.50 is no more than 15 ng/ml, even more
preferably said EC.sub.50 is no more than 10 ng/ml, and most
preferably said EC.sub.50 is no more than 5 ng/ml, such as for
example about 2 or 3 ng/ml.
[0309] Preferably, the antibody, or an antigen binding fragment
thereof, according to the present invention, that specifically
binds to NS1 protein of ZIKA virus (ZIKV), does essentially not
bind to Dengue virus (DENV) NS1 protein. Thereby "essentially not
binding" means that for the antibody, or an antigen binding
fragment thereof, no EC.sub.50-value up to 10.sup.2 ng/ml,
preferably up to 10.sup.3 ng/ml, more preferably up to 5*10.sup.3
ng/ml, even more preferably up to 8*10.sup.3 ng/ml, and most
preferably up to 10.sup.4 ng/ml can be determined in a standard
ELISA to Dengue virus (DENV) NS1 protein. In other words, the
concentration of the antibody, or an antigen binding fragment
thereof, required to achieve 50% maximal binding at saturation
(EC.sub.50) to Dengue virus (DENV) NS protein in a standard ELISA
is typically more than 10.sup.2 ng/ml, preferably more than
10.sup.3 ng/ml, more preferably more than 5*10.sup.3 ng/ml, even
more preferably more than 8*10.sup.3 ng/ml, and most preferably
more than 10.sup.4 ng/ml.
[0310] It is also preferred that the antibody, or an antigen
binding fragment thereof, according to the present invention, that
specifically binds to NS1 protein of ZIKA virus (ZIKV), does
essentially not bind to Yellow fever virus (YFV) NS1 protein, West
nile virus (WNV) NS1 protein, Japanese encephalitis virus (JEV) NS1
protein and/or to Tick-borne encephalitis virus (TBEV) NS1 protein.
Thereby "essentially not binding" means that for the antibody, or
an antigen binding fragment thereof, no EC.sub.50-value up to
10.sup.2 ng/ml, preferably up to 10.sup.3 ng/ml, more preferably up
to 5*10.sup.3 ng/ml, even more preferably up to 8*10.sup.3 ng/ml,
and most preferably up to 10.sup.4 ng/ml can be determined in a
standard ELISA to Yellow fever virus (YFV) NS1 protein, West nile
virus (WNV) NS1 protein, Japanese encephalitis virus (JEV) NS1
protein and/or to Tick-borne encephalitis virus (TBEV) NS1 protein.
In other words, the concentration of the antibody, or an antigen
binding fragment thereof, required to achieve 50% maximal binding
at saturation (EC.sub.50) to Yellow fever virus (YFV) NS1 protein,
West nile virus (WNV) NS1 protein, Japanese encephalitis virus
(JEV) NS1 protein and/or to Tick-borne encephalitis virus (TBEV)
NS1 protein in a standard ELISA is typically more than 10.sup.2
ng/ml, preferably more than 10.sup.3 ng/ml, more preferably more
than 5*10.sup.3 ng/ml, even more preferably more than 8*10.sup.3
ng/ml, and most preferably more than 10.sup.4 ng/ml.
[0311] Preferably, the antibody, or an antigen binding fragment
thereof, according to the present invention, that specifically
binds to NS1 protein of ZIKA virus (ZIKV), is a human antibody. It
is also preferred that the antibody, or an antigen binding fragment
thereof, according to the present invention, that specifically
binds to NS1 protein of ZIKA virus (ZIKV), is a monoclonal
antibody, preferably a human monoclonal antibody. Furthermore, it
is also preferred that the antibody, or an antigen binding fragment
thereof, according to the present invention, that specifically
binds to NS1 protein of ZIKA virus (ZIKV), is a recombinant
antibody.
[0312] Preferably, the antibody according to the present invention,
or an antigen binding fragment thereof, that specifically binds to
NS1 protein of ZIKA virus (ZIKV), comprises an Fc moiety. More
preferably, the antibody according to the present invention, or an
antigen binding fragment thereof, that specifically binds to NS1
protein of ZIKA virus (ZIKV), comprises a CH2 L4A mutation, a CH2
L5A mutation, or both. For a detailed description of the antibody
according to the present invention, or an antigen binding fragment
thereof, comprising an Fc moiety and/or comprises a CH2 L4A
mutation, a CH2 L5A mutation, or both, it is referred to the
detailed description of the Fc moiety and of the CH2 L4A mutation,
CH2 L5A mutation, or both in the context of the neutralizing
antibodies according to the present invention as above. The
corresponding detailed description as well as preferred embodiments
apply accordingly also for the antibody according to the present
invention, or an antigen binding fragment thereof, that
specifically binds to NS1 protein of ZIKA virus (ZIKV).
[0313] However, it is also preferred that the antibody according to
the present invention, or an antigen binding fragment thereof, that
specifically binds to NS1 protein of ZIKA virus (ZIKV), does not
comprise an Fc moiety. In particular it is preferred that the
antibody according to the present invention, or an antigen binding
fragment thereof, that specifically binds to NS1 protein of ZIKA
virus (ZIKV), is a purified antibody, a single chain antibody, Fab,
Fab', F(ab')2, Fv or scFv. Even more preferably, the antibody
according to the present invention, or an antigen binding fragment
thereof, that specifically binds to NS1 protein of ZIKA virus
(ZIKV), is labelled as described herein, for example biotinylated,
such as a biotinylated Fab, Fab', or F(ab')2 fragment.
[0314] Preferably, the antibody according to the present invention,
or an antigen binding fragment thereof, that specifically binds to
NS1 protein of ZIKA virus (ZIKV), binds to antigenic site S1 and/or
to antigenic site S2 of Zika virus NS1 protein. The present
inventors have surprisingly found that anti-ZIKV NS1 antibodies
binding to antigenic site S1 and/or to antigenic site S2 of ZIKV
NS1 protein are not cross-reactive with dengue virus NS1 protein
(DENV NS1). Anti-ZIKV NS1 antibodies, which bind neither to
antigenic site S1 nor to antigenic site S2 of ZIKV NS protein, in
contrast, are typically cross-reactive with DENV NS1. This
surprising finding indicates that antigenic sites S1 and S2 on ZIKV
NS1 can be used to distinguish ZIKV NS1-specific antibodies from
antibodies cross-reactive to DENV NS1.
[0315] Most preferably, the antibody according to the present
invention, or an antigen binding fragment thereof, that
specifically binds to NS1 protein of ZIKA virus (ZIKV), binds to
antigenic site S2 of Zika virus NS1 protein. Antigenic site S2 is
highly conserved in multiple ZIKV lineages, but it is not
homologous in sequence and structure with the corresponding site on
NS1 of other flaviviruses, thereby providing a unique specificity
for ZIKV.
[0316] Antigenic sites S1 and S2 of ZIKV NS1 protein were
identified by the present inventors as described in Example 3, FIG.
6. Whether an antibody is binding to antigenic site S1 and/or S2
may be easily identified by the skilled person by using
cross-competition studies, for example as described below or in
Example 3, wherein the S1-specific antibody according to gZKA15
(SEQ ID NOs: 91-99) and/or the S2-specific antibody according to
gZKA35 (SEQ ID NOs: 127-135) may be used as "second antibody". In
such a competition assay, presence of any competition (full or
partial) with gZKA15 and/or gZKA35 indicates that the antibody to
be tested binds to antigenic site S1 and/or to antigenic site S2,
respectively.
[0317] In general, for a competition assay commercially available
systems for characterization of protein-protein binding, such as
for example "Octet.RTM. RED96 System" provided by ForteBio, may be
used, in particular according to the supplier's instructions.
[0318] In an exemplary competition assay, e.g. using "Octet.RTM.
RED96 System" provided by ForteBio, ZIKV-NS1 protein (e.g. diluted
to 2.5 .mu.g/ml in PBS) may be immobilized (e.g. for 7-9 minutes)
on the surface of an APS coated sensor-chip. Coated biosensors may
then be placed into wells containing blocking buffer (e.g., 0.1%
BSA in PBS; e.g. for 6 minutes) to block free Biosensor binding
sites. Coated-Biosensors may then be incubated (e.g., for 8
minutes) with the antibody/antibodies to be tested (e.g. diluted in
blocking buffer for example at 10 .mu.g/ml). After binding of the
antibodies to be tested (step 1), Biosensors were moved to wells
containing the "second antibodies", e.g. gZKA15 and/or gZKA35 (e.g.
for 8 minutes) (step 2). Competition, partial competition or no
competition can thus be determined in step 2, depending on whether
no association (competition), low association (partial competition)
or (strong) association (no competition) is detected.
[0319] As described above, the antibody according to the present
invention, or the antigen binding fragment thereof, preferably
comprises (at least) three complementarity determining regions
(CDRs) on a heavy chain and (at least) three CDRs on a light chain.
In general, complementarity determining regions (CDRs) are the
hypervariable regions present in heavy chain variable domains and
light chain variable domains. Typically, the CDRs of a heavy chain
and the connected light chain of an antibody together form the
antigen receptor. Usually, the three CDRs (CDR1, CDR2, and CDR3)
are arranged non-consecutively in the variable domain. Since
antigen receptors are typically composed of two variable domains
(on two different polypeptide chains, i.e. heavy and light chain),
there are six CDRs for each antigen receptor (heavy chain: CDRH1,
CDRH2, and CDRH3; light chain: CDRL1, CDRL2, and CDRL3). A single
antibody molecule usually has two antigen receptors and therefore
contains twelve CDRs. The CDRs on the heavy and/or light chain may
be separated by framework regions, whereby a framework region (FR)
is a region in the variable domain which is less "variable" than
the CDR. For example, a chain (or each chain, respectively) may be
composed of four framework regions, separated by three CDR's.
[0320] The sequences of the heavy chains and light chains of five
exemplary antibodies of the invention, comprising three different
CDRs on the heavy chain and three different CDRs on the light chain
were determined. The position of the CDR amino acids are defined
according to the IMGT numbering system (IMGT: http://www.imgt.org/;
cf. Lefranc, M.-P. et al. (2009) Nucleic Acids Res. 37,
D1006-D1012).
[0321] Table 4 shows the SEQ ID NO's of the amino acid sequences of
the heavy chain CDR's (CDRH1, CDRH2, and CDRH3) and of the heavy
chain variable region (referred to as "VH") of exemplary antibodies
according to the present invention:
TABLE-US-00007 Antibody name CDRH.sub.1 CDRH.sub.2 CDRH.sub.3 VH
ZKA15 91 92 93 98 ZKA25 109 110 111 116 ZKA35 127 128 129 134 ZKA10
153 154 155 156 ZKA18 157 158 159 160 ZKA28 161 162 163 164 ZKA29
165 166 167 168 ZKA33 169 170 171 172 ZKA39 173 174 175 176 ZKA43
177 178 179 180 ZKA44 181 182 183 184 ZKA46 185 186 187 188 ZKA50
189 190 191 192 ZKA54 193 194 195 196 ZKB18 197 198 199 200 ZKB20
201 202 203 204 ZKB21 205 206 207 208 ZKB23 209 210 211 212 ZKC29
213 214 215 216 ZKC31 217 218 219 220 ZKC32 221 222 223 224 ZKC33
225 226 227 228 ZKC34 229 230 231 232 ZKD25 233 234 235 236
[0322] Table 5 below shows the SEQ ID NO's of the amino acid
sequences of the light chain CDR's (CDRL1, CDRL2, and CDRL3) and of
the light chain variable region (referred to as "VL") of exemplary
antibodies according to the present invention:
TABLE-US-00008 Antibody CDRL.sub.2 name CDRL.sub.1 CDRL.sub.2 long
CDRL.sub.3 VL ZKA15 94 95 96 97 99 ZKA25 112 113 114 115 117 ZKA35
130 131 132 133 135
[0323] It is thus preferred that the antibody, or the antigen
binding fragment thereof, according to the present invention
comprises amino acid sequences having at least 70%, at least 75%,
at least 80%, at least 85%, at least 88%, at least 90%, at least
92%, at least 95%, at least 96%, at least 97%, at least 98% or at
least 99% identity to at least one of the CDR sequences, the VH
sequence and/or the VL sequence shown in Table 4 and/or in Table
5.
[0324] It is preferred that the antibody or antigen binding
fragment thereof, according to the present invention comprises a
heavy chain comprising at least one CDRH1, at least one CDRH2 and
at least one CDRH3 and a light chain comprising at least one CDRL1,
at least one CDRL2 and at least one CDRL3, wherein at least one
CDR, preferably the at least one heavy chain CDRH3, comprises or
consists of an amino acid sequence according to any of SEQ ID NOs:
93, 111, 129, 155, 159, 163, 167, 171, 175, 179, 183, 187, 191,
195, 199, 203, 207, 211, 215, 219, 223, 227, 231, and 235; or a
functional sequence variant thereof having at least 70%, at least
75%, at least 80%, at least 85%, at least 88%, at least 90%, at
least 92%, at least 95%, at least 96%, at least 97%, at least 98%
or at least 99% sequence identity.
[0325] More preferably, the antibody, or the antigen binding
fragment thereof, according to the present invention comprises a
heavy chain comprising at least one CDRH1, at least one CDRH2 and
at least one CDRH3 and a light chain comprising at least one CDRL1,
at least one CDRL2 and at least one CDRL3, wherein at least one
CDR, preferably the at least one heavy chain CDRH3, comprises or
consists of an amino acid sequence according to any of SEQ ID NOs:
93, 111 and 129, or a functional sequence variant thereof having at
least 70%, at least 75%, at least 80%, at least 85%, at least 88%,
at least 90%, at least 92%, at least 95%, at least 96%, at least
97%, at least 98% or at least 99% sequence identity. Thereby, it is
preferred that at least one CDR, preferably the at least one heavy
chain CDRH3, comprises or consists of an amino acid sequence
according to SEQ ID NO: 93 or a functional sequence variant thereof
having at least 70%, at least 75%, at least 80%, at least 85%, at
least 88%, at least 90%, at least 92%, at least 95%, at least 96%,
at least 97%, at least 98% or at least 99% sequence identity. It is
also preferred that at least one CDR, preferably the at least one
heavy chain CDRH3, comprises or consists of an amino acid sequence
according to SEQ ID NO: 111 or a functional sequence variant
thereof having at least 70%, at least 75%, at least 80%, at least
85%, at least 88%, at least 90%, at least 92%, at least 95%, at
least 96%, at least 97%, at least 98% or at least 99% sequence
identity. Moreover, it is also preferred that at least one CDR,
preferably the at least one heavy chain CDRH3, comprises or
consists of an amino acid sequence according to SEQ ID NO: 129 or a
functional sequence variant thereof having at least 70%, at least
75%, at least 80%, at least 85%, at least 88%, at least 90%, at
least 92%, at least 95%, at least 96%, at least 97%, at least 98%
or at least 99% sequence identity.
[0326] Most preferably, the antibody, or the antigen binding
fragment thereof, according to the present invention comprises a
heavy chain comprising at least one CDRH1, at least one CDRH2 and
at least one CDRH3 and a light chain comprising at least one CDRL1,
at least one CDRL2 and at least one CDRL3, wherein at least one
CDR, preferably the at least one heavy chain CDRH3, comprises or
consists of an amino acid sequence according to SEQ ID NO: 129, or
a functional sequence variant thereof having at least 70%, at least
75%, at least 80%, at least 85%, at least 88%, at least 90%, at
least 92%, at least 95%, at least 96%, at least 97%, at least 98%
or at least 99% sequence identity.
[0327] It is also preferred, that the antibody, or the antigen
binding fragment thereof, according to the present invention
comprises a heavy chain comprising at least one CDRH1, at least one
CDRH2 and at least one CDRH3 and a light chain comprising at least
one CDRL1, at least one CDRL2 and at least one CDRL3, wherein
[0328] (i) the at least one heavy chain CDRH1 comprises an amino
acid sequence according to any of SEQ ID NOs: 91, 109, 127, 153,
157, 161, 165, 169, 173, 177, 181, 185, 189, 193, 197, 201, 205,
209, 213, 217, 221, 225, 229, and 233; or a functional sequence
variant thereof having at least 70%, at least 75%, at least 80%, at
least 85%, at least 88%, at least 90%, at least 92%, at least 95%,
at least 96%, at least 97%, at least 98% or at least 99% sequence
identity; [0329] (ii) the at least one CDRH2 comprises an amino
acid sequence according to any of SEQ ID NOs: 92, 110, 128, 154,
158, 162, 166, 170, 174, 178, 182, 186, 190, 194, 198, 202, 206,
210, 214, 218, 222, 226, 230, and 234; or a functional sequence
variant thereof having at least 70%, at least 75%, at least 80%, at
least 85%, at least 88%, at least 90%, at least 92%, at least 95%,
at least 96%, at least 97%, at least 98% or at least 99% sequence
identity; and/or [0330] (iii) the at least one heavy chain CDRH3
comprises an amino acid sequence according to any of SEQ ID NOs:
93, 111, 129, 155, 159, 163, 167, 171, 175, 179, 183, 187, 191,
195, 199, 203, 207, 211, 215, 219, 223, 227, 231, and 235; or a
functional sequence variant thereof having at least 70%, at least
75%, at least 80%, at least 85%, at least 88%, at least 90%, at
least 92%, at least 95%, at least 96%, at least 97%, at least 98%
or at least 99% sequence identity.
[0331] Still more preferably, the antibody, or the antigen binding
fragment thereof, according to the present invention comprises a
heavy chain comprising at least one CDRH1, at least one CDRH2 and
at least one CDRH3 and a light chain comprising at least one CDRL1,
at least one CDRL2 and at least one CDRL3, wherein [0332] (i) the
at least one heavy chain CDRH1 comprises an amino acid sequence
according to any of SEQ ID NOs: 91, 109 and 127, or a functional
sequence variant thereof having at least 70%, at least 75%, at
least 80%, at least 85%, at least 88%, at least 90%, at least 92%,
at least 95%, at least 96%, at least 97%, at least 98% or at least
99% sequence identity; [0333] (ii) the at least one CDRH2 comprises
an amino acid sequence according to any of SEQ ID NOs: 92, 110 and
128, or a functional sequence variant thereof having at least 70%,
at least 75%, at least 80%, at least 85%, at least 88%, at least
90%, at least 92%, at least 95%, at least 96%, at least 97%, at
least 98% or at least 99% sequence identity; and/or [0334] (iii)
the at least one heavy chain CDRH3 comprises an amino acid sequence
according to any of SEQ ID NOs: 93, 111 and 129, or a functional
sequence variant thereof having at least 70%, at least 75%, at
least 80%, at least 85%, at least 88%, at least 90%, at least 92%,
at least 95%, at least 96%, at least 97%, at least 98% or at least
99% sequence identity.
[0335] Even more preferably, the antibody, or the antigen binding
fragment thereof, according to the present invention comprises a
heavy chain comprising at least one CDRH1, at least one CDRH2 and
at least one CDRH3 and a light chain comprising at least one CDRL1,
at least one CDRL2 and at least one CDRL3, wherein [0336] (i) the
at least one heavy chain CDRH1 comprises an amino acid sequence
according to SEQ ID NO: 91 or a functional sequence variant thereof
having at least 70%, at least 75%, at least 80%, at least 85%, at
least 88%, at least 90%, at least 92%, at least 95%, at least 96%,
at least 97%, at least 98% or at least 99% sequence identity;
[0337] (ii) the at least one CDRH2 comprises an amino acid sequence
according to SEQ ID NO: 92 or a functional sequence variant thereof
having at least 70%, at least 75%, at least 80%, at least 85%, at
least 88%, at least 90%, at least 92%, at least 95%, at least 96%,
at least 97%, at least 98% or at least 99% sequence identity;
and/or [0338] (iii) the at least one heavy chain CDRH3 comprises an
amino acid sequence according to SEQ ID NOs: 93 or a functional
sequence variant thereof having at least 70%, at least 75%, at
least 80%, at least 85%, at least 88%, at least 90%, at least 92%,
at least 95%, at least 96%, at least 97%, at least 98% or at least
99% sequence identity.
[0339] It is also even more preferred that the antibody, or the
antigen binding fragment thereof, according to the present
invention comprises a heavy chain comprising at least one CDRH1, at
least one CDRH2 and at least one CDRH3 and a light chain comprising
at least one CDRL1, at least one CDRL2 and at least one CDRL3,
wherein [0340] (i) the at least one heavy chain CDRH1 comprises an
amino acid sequence according to SEQ ID NO: 109 or a functional
sequence variant thereof having at least 70%, at least 75%, at
least 80%, at least 85%, at least 88%, at least 90%, at least 92%,
at least 95%, at least 96%, at least 97%, at least 98% or at least
99% sequence identity; [0341] (ii) the at least one CDRH2 comprises
an amino acid sequence according to SEQ ID NO: 110 or a functional
sequence variant thereof having at least 70%, at least 75%, at
least 80%, at least 85%, at least 88%, at least 90%, at least 92%,
at least 95%, at least 96%, at least 97%, at least 98% or at least
99% sequence identity; and/or [0342] (iii) the at least one heavy
chain CDRH3 comprises an amino acid sequence according to SEQ ID
NOs: 111 or a functional sequence variant thereof having at least
70%, at least 75%, at least 80%, at least 85%, at least 88%, at
least 90%, at least 92%, at least 95%, at least 96%, at least 97%,
at least 98% or at least 99% sequence identity.
[0343] It is particularly preferred, that the antibody, or the
antigen binding fragment thereof, according to the present
invention comprises a heavy chain comprising at least one CDRH1, at
least one CDRH2 and at least one CDRH3 and a light chain comprising
at least one CDRL1, at least one CDRL2 and at least one CDRL3,
wherein [0344] (i) the at least one heavy chain CDRH1 comprises an
amino acid sequence according to SEQ ID NO: 127 or a functional
sequence variant thereof having at least 70%, at least 75%, at
least 80%, at least 85%, at least 88%, at least 90%, at least 92%,
at least 95%, at least 96%, at least 97%, at least 98% or at least
99% sequence identity; [0345] (ii) the at least one CDRH2 comprises
an amino acid sequence according to SEQ ID NO: 128 or a functional
sequence variant thereof having at least 70%, at least 75%, at
least 80%, at least 85%, at least 88%, at least 90%, at least 92%,
at least 95%, at least 96%, at least 97%, at least 98% or at least
99% sequence identity; and/or [0346] (iii) the at least one heavy
chain CDRH3 comprises an amino acid sequence according to SEQ ID
NOs: 129 or a functional sequence variant thereof having at least
70%, at least 75%, at least 80%, at least 85%, at least 88%, at
least 90%, at least 92%, at least 95%, at least 96%, at least 97%,
at least 98% or at least 99% sequence identity.
[0347] More preferably, the antibody, or the antigen binding
fragment thereof, according to the present invention comprises a
heavy chain comprising at least one CDRH1, at least one CDRH2 and
at least one CDRH3 and a light chain comprising at least one CDRL1,
at least one CDRL2 and at least one CDRL3, wherein [0348] (i) the
at least one CDRL1 comprises an amino acid sequence according to
any of SEQ ID NOs: 94, 112 and 130, or a functional sequence
variant thereof having at least 70%, at least 75%, at least 80%, at
least 85%, at least 88%, at least 90%, at least 92%, at least 95%,
at least 96%, at least 97%, at least 98% or at least 99% sequence
identity; [0349] (ii) the at least one CDRL2 comprises an amino
acid sequence according to any of SEQ ID NOs: 95, 96, 113, 114, 131
and 132, or a functional sequence variant thereof having at least
70%, at least 75%, at least 80%, at least 85%, at least 88%, at
least 90%, at least 92%, at least 95%, at least 96%, at least 97%,
at least 98% or at least 99% sequence identity; and/or [0350] (iii)
the at least one CDRL3 amino comprises an amino acid sequence
according to any of SEQ ID NOs: 97, 115 and 133, or a functional
sequence variant thereof having at least 70%, at least 75%, at
least 80%, at least 85%, at least 88%, at least 90%, at least 92%,
at least 95%, at least 96%, at least 97%, at least 98% or at least
99% sequence identity.
[0351] Even more preferably, the antibody, or the antigen binding
fragment thereof, according to the present invention comprises a
heavy chain comprising at least one CDRH1, at least one CDRH2 and
at least one CDRH3 and a light chain comprising at least one CDRL1,
at least one CDRL2 and at least one CDRL3, wherein [0352] (i) the
at least one CDRL1 comprises an amino acid sequence according to
SEQ ID NO: 94 or a functional sequence variant thereof having at
least 70%, at least 75%, at least 80%, at least 85%, at least 88%,
at least 90%, at least 92%, at least 95%, at least 96%, at least
97%, at least 98% or at least 99% sequence identity; [0353] (ii)
the at least one CDRL2 comprises an amino acid sequence according
to SEQ ID NO: 95 or 96, or a functional sequence variant thereof
having at least 70%, at least 75%, at least 80%, at least 85%, at
least 88%, at least 90%, at least 92%, at least 95%, at least 96%,
at least 97%, at least 98% or at least 99% sequence identity;
and/or [0354] (iii) the at least one CDRL3 amino comprises an amino
acid sequence according to SEQ ID NO: 97 or a functional sequence
variant thereof having at least 70%, at least 75%, at least 80%, at
least 85%, at least 88%, at least 90%, at least 92%, at least 95%,
at least 96%, at least 97%, at least 98% or at least 99% sequence
identity.
[0355] It is also even more preferred that the antibody, or the
antigen binding fragment thereof, according to the present
invention comprises a heavy chain comprising at least one CDRH1, at
least one CDRH2 and at least one CDRH3 and a light chain comprising
at least one CDRL1, at least one CDRL2 and at least one CDRL3,
wherein [0356] (i) the at least one CDRL1 comprises an amino acid
sequence according to SEQ ID NO: 112 or a functional sequence
variant thereof having at least 70%, at least 75%, at least 80%, at
least 85%, at least 88%, at least 90%, at least 92%, at least 95%,
at least 96%, at least 97%, at least 98% or at least 99% sequence
identity; [0357] (ii) the at least one CDRL2 comprises an amino
acid sequence according to SEQ ID NO: 113 or 114, or a functional
sequence variant thereof having at least 70%, at least 75%, at
least 80%, at least 85%, at least 88%, at least 90%, at least 92%,
at least 95%, at least 96%, at least 97%, at least 98% or at least
99% sequence identity; and/or [0358] (iii) the at least one CDRL3
amino comprises an amino acid sequence according to SEQ ID NO: 115
or a functional sequence variant thereof having at least 70%, at
least 75%, at least 80%, at least 85%, at least 88%, at least 90%,
at least 92%, at least 95%, at least 96%, at least 97%, at least
98% or at least 99% sequence identity.
[0359] It is particularly preferred that the antibody, or the
antigen binding fragment thereof, according to the present
invention comprises a heavy chain comprising at least one CDRH1, at
least one CDRH2 and at least one CDRH3 and a light chain comprising
at least one CDRL1, at least one CDRL2 and at least one CDRL3,
wherein [0360] (i) the at least one CDRL1 comprises an amino acid
sequence according to SEQ ID NO: 130 or a functional sequence
variant thereof having at least 70%, at least 75%, at least 80%, at
least 85%, at least 88%, at least 90%, at least 92%, at least 95%,
at least 96%, at least 97%, at least 98% or at least 99% sequence
identity; [0361] (ii) the at least one CDRL2 comprises an amino
acid sequence according to SEQ ID NO: 131 or 132, or a functional
sequence variant thereof having at least 70%, at least 75%, at
least 80%, at least 85%, at least 88%, at least 90%, at least 92%,
at least 95%, at least 96%, at least 97%, at least 98% or at least
99% sequence identity; and/or [0362] (iii) the at least one CDRL3
amino comprises an amino acid sequence according to SEQ ID NO: 133
or a functional sequence variant thereof having at least 70%, at
least 75%, at least 80%, at least 85%, at least 88%, at least 90%,
at least 92%, at least 95%, at least 96%, at least 97%, at least
98% or at least 99% sequence identity.
[0363] Most preferably, the antibody, or the antigen binding
fragment thereof, according to the present invention comprises
CDRH1, CDRH2, and CDRH3 amino acid sequences (i) according to SEQ
ID NOs: 91-93; or functional sequence variants thereof having at
least 70%, at least 75%, at least 80%, at least 85%, at least 88%,
at least 90%, at least 92%, at least 95%, at least 96%, at least
97%, at least 98% or at least 99% sequence identity; (ii) according
to SEQ ID NOs: 109-111; or functional sequence variants thereof
having at least 70%, at least 75%, at least 80%, at least 85%, at
least 88%, at least 90%, at least 92%, at least 95%, at least 96%,
at least 97%, at least 98% or at least 99% sequence identity; (iii)
according to SEQ ID NOs: 127-129; or functional sequence variants
thereof having at least 70%, at least 75%, at least 80%, at least
85%, at least 88%, at least 90%, at least 92%, at least 95%, at
least 96%, at least 97%, at least 98% or at least 99% sequence
identity; (iv) according to SEQ ID NOs: 153-155; or functional
sequence variants thereof having at least 70%, at least 75%, at
least 80%, at least 85%, at least 88%, at least 90%, at least 92%,
at least 95%, at least 96%, at least 97%, at least 98% or at least
99% sequence identity; (v) according to SEQ ID NOs: 157-159; or
functional sequence variants thereof having at least 70%, at least
75%, at least 80%, at least 85%, at least 88%, at least 90%, at
least 92%, at least 95%, at least 96%, at least 97%, at least 98%
or at least 99% sequence identity; (vi) according to SEQ ID NOs:
161-163; or functional sequence variants thereof having at least
70%, at least 75%, at least 80%, at least 85%, at least 88%, at
least 90%, at least 92%, at least 95%, at least 96%, at least 97%,
at least 98% or at least 99% sequence identity; (vii) according to
SEQ ID NOs: 165-167; or functional sequence variants thereof having
at least 70%, at least 75%, at least 80%, at least 85%, at least
88%, at least 90%, at least 92%, at least 95%, at least 96%, at
least 97%, at least 98% or at least 99% sequence identity; (viii)
according to SEQ ID NOs: 169-171; or functional sequence variants
thereof having at least 70%, at least 75%, at least 80%, at least
85%, at least 88%, at least 90%, at least 92%, at least 95%, at
least 96%, at least 97%, at least 98% or at least 99% sequence
identity; (ix) according to SEQ ID NOs: 173-175; or functional
sequence variants thereof having at least 70%, at least 75%, at
least 80%, at least 85%, at least 88%, at least 90%, at least 92%,
at least 95%, at least 96%, at least 97%, at least 98% or at least
99% sequence identity; (x) according to SEQ ID NOs: 177-179; or
functional sequence variants thereof having at least 70%, at least
75%, at least 80%, at least 85%, at least 88%, at least 90%, at
least 92%, at least 95%, at least 96%, at least 97%, at least 98%
or at least 99% sequence identity; (xi) according to SEQ ID NOs:
181-183; or functional sequence variants thereof having at least
70%, at least 75%, at least 80%, at least 85%, at least 88%, at
least 90%, at least 92%, at least 95%, at least 96%, at least 97%,
at least 98% or at least 99% sequence identity; (xii) according to
SEQ ID NOs: 185-187; or functional sequence variants thereof having
at least 70%, at least 75%, at least 80%, at least 85%, at least
88%, at least 90%, at least 92%, at least 95%, at least 96%, at
least 97%, at least 98% or at least 99% sequence identity; (xiii)
according to SEQ ID NOs: 189-191; or functional sequence variants
thereof having at least 70%, at least 75%, at least 80%, at least
85%, at least 88%, at least 90%, at least 92%, at least 95%, at
least 96%, at least 97%, at least 98% or at least 99% sequence
identity; (xiv) according to SEQ ID NOs: 193-195; or functional
sequence variants thereof having at least 70%, at least 75%, at
least 80%, at least 85%, at least 88%, at least 90%, at least 92%,
at least 95%, at least 96%, at least 97%, at least 98% or at least
99% sequence identity; (xv) according to SEQ ID NOs: 197-199; or
functional sequence variants thereof having at least 70%, at least
75%, at least 80%, at least 85%, at least 88%, at least 90%, at
least 92%, at least 95%, at least 96%, at least 97%, at least 98%
or at least 99% sequence identity; (xvi) according to SEQ ID NOs:
201-203; or functional sequence variants thereof having at least
70%, at least 75%, at least 80%, at least 85%, at least 88%, at
least 90%, at least 92%, at least 95%, at least 96%, at least 97%,
at least 98% or at least 99% sequence identity; (xvii) according to
SEQ ID NOs: 205-207; or functional sequence variants thereof having
at least 70%, at least 75%, at least 80%, at least 85%, at least
88%, at least 90%, at least 92%, at least 95%, at least 96%, at
least 97%, at least 98% or at least 99% sequence identity; (xviii)
according to SEQ ID NOs: 209-211; or functional sequence variants
thereof having at least 70%, at least 75%, at least 80%, at least
85%, at least 88%, at least 90%, at least 92%, at least 95%, at
least 96%, at least 97%, at least 98% or at least 99% sequence
identity; (xix) according to SEQ ID NOs: 213-215; or functional
sequence variants thereof having at least 70%, at least 75%, at
least 80%, at least 85%, at least 88%, at least 90%, at least 92%,
at least 95%, at least 96%, at least 97%, at least 98% or at least
99% sequence identity; (xx) according to SEQ ID NOs: 217-219; or
functional sequence variants thereof having at least 70%, at least
75%, at least 80%, at least 85%, at least 88%, at least 90%, at
least 92%, at least 95%, at least 96%, at least 97%, at least 98%
or at least 99% sequence identity; (xxi) according to SEQ ID NOs:
221-223; or functional sequence variants thereof having at least
70%, at least 75%, at least 80%, at least 85%, at least 88%, at
least 90%, at least 92%, at least 95%, at least 96%, at least 97%,
at least 98% or at least 99% sequence identity; (xxii) according to
SEQ ID NOs: 225-227; or functional sequence variants thereof having
at least 70%, at least 75%, at least 80%, at least 85%, at least
88%, at least 90%, at least 92%, at least 95%, at least 96%, at
least 97%, at least 98% or at least 99% sequence identity; (xxiii)
according to SEQ ID NOs: 229-231; or functional sequence variants
thereof having at least 70%, at least 75%, at least 80%, at least
85%, at least 88%, at least 90%, at least 92%, at least 95%, at
least 96%, at least 97%, at least 98% or at least 99% sequence
identity; or (xxiv) according to SEQ ID NOs: 233-235; or functional
sequence variants thereof having at least 70%, at least 75%, at
least 80%, at least 85%, at least 88%, at least 90%, at least 92%,
at least 95%, at least 96%, at least 97%, at least 98% or at least
99% sequence identity.
[0364] It is particularly preferred that the antibody, or the
antigen binding fragment thereof, according to the present
invention comprises CDRH1, CDRH2, and CDRH3 amino acid sequences
and CDRL1, CDRL2, and CDRL3 amino acid sequences (i) according to
SEQ ID NOs: 91-95 and 97; or functional sequence variants thereof
having at least 70%, at least 75%, at least 80%, at least 85%, at
least 88%, at least 90%, at least 92%, at least 95%, at least 96%,
at least 97%, at least 98% or at least 99% sequence identity; (ii)
according to SEQ ID NOs: 91-94 and 96-97; or functional sequence
variants thereof having at least 70%, at least 75%, at least 80%,
at least 85%, at least 88%, at least 90%, at least 92%, at least
95%, at least 96%, at least 97%, at least 98% or at least 99%
sequence identity; (iii) according to SEQ ID NOs: 109-113 and 115;
or functional sequence variants thereof having at least 70%, at
least 75%, at least 80%, at least 85%, at least 88%, at least 90%,
at least 92%, at least 95%, at least 96%, at least 97%, at least
98% or at least 99% sequence identity; (iv) according to SEQ ID
NOs: 109-112 and 114-115; or functional sequence variants thereof
having at least 70%, at least 75%, at least 80%, at least 85%, at
least 88%, at least 90%, at least 92%, at least 95%, at least 96%,
at least 97%, at least 98% or at least 99% sequence identity; (v)
according to SEQ ID NOs: 127-131 and 133; or functional sequence
variants thereof having at least 70%, at least 75%, at least 80%,
at least 85%, at least 88%, at least 90%, at least 92%, at least
95%, at least 96%, at least 97%, at least 98% or at least 99%
sequence identity; or (vi) according to SEQ ID NOs: 127-130 and
132-133; or functional sequence variants thereof having at least
70%, at least 75%, at least 80%, at least 85%, at least 88%, at
least 90%, at least 92%, at least 95%, at least 96%, at least 97%,
at least 98% or at least 99% sequence identity.
[0365] Most preferably, the antibody, or the antigen binding
fragment thereof, according to the present invention comprises
CDRH1, CDRH2, and CDRH3 amino acid sequences and CDRL1, CDRL2, and
CDRL3 amino acid sequences (i) according to SEQ ID NOs: 127-131 and
133; or functional sequence variants thereof having at least 70%,
at least 75%, at least 80%, at least 85%, at least 88%, at least
90%, at least 92%, at least 95%, at least 96%, at least 97%, at
least 98% or at least 99% sequence identity; or (ii) according to
SEQ ID NOs: 127-130 and 132-133; or functional sequence variants
thereof having at least 70%, at least 75%, at least 80%, at least
85%, at least 88%, at least 90%, at least 92%, at least 95%, at
least 96%, at least 97%, at least 98% or at least 99% sequence
identity.
[0366] It is also preferred that the antibody, or the antigen
binding fragment thereof, according to the present invention
comprises a heavy chain variable region (VH) and, optionally, a
light chain variable region (VL), wherein the heavy chain variable
region (VH) comprises or consists of an amino acid sequence
according to any of SEQ ID NOs: 98, 116, 134, 156, 160, 164, 168,
172, 176, 180, 184, 188, 192, 196, 200, 204, 208, 212, 216, 220,
224, 228, 232, and 236; or a functional sequence variant thereof
having at least 70%, at least 75%, at least 80%, at least 85%, at
least 88%, at least 90%, at least 92%, at least 95%, at least 96%,
at least 97%, at least 98% or at least 99% sequence identity.
[0367] Most preferably, the antibody, or the antigen binding
fragment thereof, according to the present invention comprises (i)
a heavy chain variable region (VH) amino acid sequence according to
SEQ ID NO: 98 or a functional sequence variant thereof having at
least 70%, at least 75%, at least 80%, at least 85%, at least 88%,
at least 90%, at least 92%, at least 95%, at least 96%, at least
97%, at least 98% or at least 99% sequence identity and/or a light
chain variable region (VL) amino acid sequence according to SEQ ID
NO: 99 or a functional sequence variant thereof having at least
70%, at least 75%, at least 80%, at least 85%, at least 88%, at
least 90%, at least 92%, at least 95%, at least 96%, at least 97%,
at least 98% or at least 99% sequence identity; (ii) a heavy chain
variable region (VH) amino acid sequence according to SEQ ID NO:
116 or a functional sequence variant thereof having at least 70%,
at least 75%, at least 80%, at least 85%, at least 88%, at least
90%, at least 92%, at least 95%, at least 96%, at least 97%, at
least 98% or at least 99% sequence identity and/or a light chain
variable region (VL) amino acid sequence according to SEQ ID NO:
117 or a functional sequence variant thereof having at least 70%,
at least 75%, at least 80%, at least 85%, at least 88%, at least
90%, at least 92%, at least 95%, at least 96%, at least 97%, at
least 98% or at least 99% sequence identity; or (iii) a heavy chain
variable region (VH) amino acid sequence according to SEQ ID NO:
134 or a functional sequence variant thereof having at least 70%,
at least 75%, at least 80%, at least 85%, at least 88%, at least
90%, at least 92%, at least 95%, at least 96%, at least 97%, at
least 98% or at least 99% sequence identity and/or a light chain
variable region (VL) amino acid sequence according to SEQ ID NO:
135 or a functional sequence variant thereof having at least 70%,
at least 75%, at least 80%, at least 85%, at least 88%, at least
90%, at least 92%, at least 95%, at least 96%, at least 97%, at
least 98% or at least 99% sequence identity.
[0368] Most preferably, the antibody, or the antigen binding
fragment thereof, according to the present invention comprises a
heavy chain variable region (VH) amino acid sequence according to
SEQ ID NO: 134 or a functional sequence variant thereof having at
least 70%, at least 75%, at least 80%, at least 85%, at least 88%,
at least 90%, at least 92%, at least 95%, at least 96%, at least
97%, at least 98% or at least 99% sequence identity and/or a light
chain variable region (VL) amino acid sequence according to SEQ ID
NO: 135 or a functional sequence variant thereof having at least
70%, at least 75%, at least 80%, at least 85%, at least 88%, at
least 90%, at least 92%, at least 95%, at least 96%, at least 97%,
at least 98% or at least 99% sequence identity.
[0369] Preferably, the antibody, or the antigen binding fragment
thereof, according to the present invention is gZKA15, gZKA25, or
gZKA35, more preferably the antibody, or the antigen binding
fragment thereof, is gZKA25 or gZKA35, even more preferably the
antibody, or the antigen binding fragment thereof, is gZKA35.
[0370] The present inventors have isolated monoclonal antibody
(mAb) according to the present invention, which are referred to
herein as ZKA15, ZKA25 and ZKA35 (cf Tables 4 and 5, Example 1).
Based on those antibodies, in particular on the VH and VL genes of
those antibodies, the terms "gZKA15", "gZKA25" and "gZKA35", as
used herein, refer to the respective "generic" antibodies, or
antigen binding fragments thereof.
[0371] Namely, "gZKA15" refers to an antibody, or antigen binding
fragment thereof, having a CDRH1 amino acid sequence according to
SEQ ID NO: 91, a CDRH2 amino acid sequence according to SEQ ID NO:
92, a CDRH3 amino acid sequence according to SEQ ID NO: 93, a CDRL1
amino acid sequence according to SEQ ID NO: 94, a CDRL2 amino acid
sequence according to SEQ ID NO: 95 or 96, and a CDRL3 amino acid
sequence according to SEQ ID NO: 97. The heavy chain variable
region (V.sub.H) has preferably an amino acid sequence according to
SEQ ID NO: 98 and the light chain variable region (V.sub.L) has
preferably an amino acid sequence according to SEQ ID NO: 99.
[0372] "gZKA25" refers to an antibody, or antigen binding fragment
thereof, having a CDRH1 amino acid sequence according to SEQ ID NO:
109, a CDRH2 amino acid sequence according to SEQ ID NO: 110, a
CDRH3 amino acid sequence according to SEQ ID NO: 111, a CDRL1
amino acid sequence according to SEQ ID NO: 112, a CDRL2 amino acid
sequence according to SEQ ID NO: 113 or 114, and a CDRL3 amino acid
sequence according to SEQ ID NO: 115. The heavy chain variable
region (V.sub.H) has preferably an amino acid sequence according to
SEQ ID NO: 116 and the light chain variable region (V.sub.L) has
preferably an amino acid sequence according to SEQ ID NO: 117.
[0373] "gZKA35" refers to an antibody, or antigen binding fragment
thereof, having a CDRH1 amino acid sequence according to SEQ ID NO:
127, a CDRH2 amino acid sequence according to SEQ ID NO: 128, a
CDRH3 amino acid sequence according to SEQ ID NO: 129, a CDRL1
amino acid sequence according to SEQ ID NO: 130, a CDRL2 amino acid
sequence according to SEQ ID NO: 131 or 132, and a CDRL3 amino acid
sequence according to SEQ ID NO: 133. The heavy chain variable
region (V.sub.H) has preferably an amino acid sequence according to
SEQ ID NO: 134 and the light chain variable region (V.sub.L) has
preferably an amino acid sequence according to SEQ ID NO: 135.
[0374] The detailed description above referring to "production of
antibodies" (above section "production of antibodies") and
"optional additional features of antibodies" (above section
"optional additional features of antibodies") apply to all
antibodies, and antigen binding fragments thereof, as described in
the present application--i.e. those sections apply not only to the
neutralizing antibodies, and antigen-binding fragments thereof,
according to the present invention, but also to the NS1-protein
binding antibodies, and antigen-binding fragments thereof,
according to the present invention.
[0375] In particular, it is preferred that the antibody, or the
antigen binding fragment thereof, according to the present
invention is labelled, for example biotinylated.
[0376] Preferably, the antibody, or the antigen binding fragment
thereof, according to the present invention is biotinylated.
[0377] It is also preferred that the antibody, or the antigen
binding fragment thereof, according to the present invention is
conjugated to an enzyme, such as horseradish peroxidase (HRP).
Conjugation of antibodies to HRP are, for example, described in
Wisdom G B. Conjugation of antibodies to horseradish peroxidase.
Methods Mol Biol. 2005; 295:127-30 or in Antibodies--a laboratory
manual. Edited by Edward A. Greenfield, Second edition 2012, Cold
Spring Harbor Laboratory Press, ISBN: 9781936113811.
[0378] For example, antibodies of the invention, or the antigen
binding fragments thereof, may be coupled to a detectable label,
for example to provide measurability, e.g. for quantification or to
facilitate imaging. Labeled antibodies may be employed in a wide
variety of assays, in particular in immunoassays, employing a wide
variety of labels. Preferred labels include radionuclides, enzymes,
coenzymes, fluorescers, chemiluminescers, chromogens, enzyme
substrates or co-factors, enzyme inhibitors, prosthetic group
complexes, free radicals, particles, dyes (e.g., fluorescent dyes,
tandem dyes), and the like. Examples of suitable enzymes include
horseradish peroxidase (HRP), alkaline phosphatase,
.beta.-galactosidase, or acetylcholinesterase; examples of suitable
prosthetic group complexes include streptavidin/biotin and
avidin/biotin; examples of suitable fluorescent materials include
umbelliferone, fluorescein, fluorescein isothiocyanate, rhodamine,
dichlorotriazinylamine fluorescein, dansyl chloride or
phycoerythrin; an example of a luminescent material is luminol;
examples of bioluminescent materials include luciferase, luciferin,
and aequorin; and examples of suitable radioactive material include
125I, 131I, 35S, or 3H. Such labeled reagents may be used in a
variety of well-known assays, such as radioimmunoassays, enzyme
immunoassays, e.g., ELISA, fluorescent immunoassays, and the like,
preferably in ELISA. Labeled antibodies according to the present
invention may be thus be used in such assays for example as
described in U.S. Pat. Nos. 3,766,162; 3,791,932; 3,817,837; and
4,233,402.
[0379] Preferred labels include (i) enzymes as described above,
e.g. horseradish peroxidase (HRP) or alkaline phosphatase, in
particular in Blockade-of-binding assay, Western Blotting, ELISA
and immunohistochemistry; (ii) prosthetic group complexes as
described above, e.g. streptavidin/biotin and avidin/biotin, in
particular in ELISA and immunohistochemistry; (iii) fluorescers as
described above, such as fluorescent dyes and fluorescent proteins
(e.g., (enhanced) green fluorescent protein (EGFP); TagBFP,
Turquoise, Venus, K02, Cherry, Apple, Kate2), in particular in
immunofluorescence and flow cytometry; and (iv) tandem dyes in flow
cytometry.
[0380] Preferably, the antibody, or the antigen binding fragment
thereof, is biotinylated. Biotinylation is rapid, specific and is
unlikely to perturb the natural function of the molecule due to the
small size of biotin (MW=244.31 g/mol). Biotin binds to
streptavidin and avidin with an extremely high affinity, fast
on-rate, and high specificity. Biotin-binding to streptavidin and
avidin is resistant to extremes of heat, pH and proteolysis, making
capture of biotinylated molecules possible in a wide variety of
environments. The antibody, or the antigen binding fragment
thereof, according to the present invention may be biotinylated
chemically or enzymatically. Chemical biotinylation utilizes
various conjugation chemistries to yield nonspecific biotinylation
of amines (e.g., NHS-coupling gives biotinylation of any primary
amines in the antibody, see below). Enzymatic biotinylation results
in biotinylation of a specific lysine within a certain sequence by
use of a bacterial biotin ligase.
[0381] Moreover, a second antibody, or antibody fragment thereof,
may also be used as label. In this case, the antibody, or the
antigen binding fragment thereof, according to the present
invention is conjugated to a second antibody, or antibody fragment
thereof, to form an antibody heteroconjugate as described for
example in U.S. Pat. No. 4,676,980. In this case, the second
antibody may optionally be labelled as described herein.
[0382] Methods for coupling antibodies to labels are well known in
the art. For example, in the antibody or in the antigen binding
fragment thereof the side chain of lysine, which terminates in a
primary amine (--NH.sub.2), may be used to link labels covalently
to the antibody or in the antigen binding fragment thereof. Many
variant labeling procedures are described in the literature. For
example, the labelling approach may be selected from the group
consisting of NHS esters, heterobifunctional reagents,
carbodiimides and sodium periodate.
[0383] NHS esters may be used in particular in the case of
fluorescent dye labels. A fluorescent dye label may be purchased in
an activated form of the label with an inbuilt NHS ester (also
called a `succinimidyl ester`). The activated dye can be reacted
under appropriate conditions with the antibody or with the antigen
binding fragment thereof (e.g. via a lysine group). Excess reactive
dye can be removed (e.g. by column chromatography) before the
labeled antibody or antigen binding fragment thereof can be used in
an immunoassay.
[0384] Heterobifunctional reagents may be used in particular if the
label is a protein molecule (e.g. HRP, alkaline phosphatase, or
phycoerythrin). In this case, the antibody or the antigen binding
fragment thereof and the label may have multiple amines. In this
situation some of the lysines on one molecule (e.g. on the antibody
or on the antigen binding fragment thereof) may be modified to
create a new reactive group (X) and lysines on the label to create
another reactive group (Y) (or vice versa). A `heterobifunctional
reagent` is then used to introduce the Y groups, which subsequently
react with X groups when antibody and label are mixed, thus
creating heterodimeric conjugates.
[0385] Carbodiimides, such as EDC, may be used in particular to
create covalent links between amine- and carboxyl-containing
molecules. Carbodiimides activate carboxyl groups, and the
activated intermediate is then attacked by an amine (e.g. provided
by a lysine residue on the antibody or the antigen binding fragment
thereof). Carbodiimides may be used in particular to conjugate
antibodies to carboxylated particles (e.g. latex particles,
magnetic beads), and to other carboxylated surfaces, such as
microwell plates or chip surfaces. Carbodiimides may also be used
to attach dyes or protein labels to antibodies or antigen binding
fragments thereof.
[0386] Sodium periodate may be used in particular for labelling
with horseradish peroxidase (HRP). Periodate activates carbohydrate
chains on the HRP molecule to create aldehyde groups, which are
capable of reacting with lysines on the antibody or antigen binding
fragment thereof. Since HRP itself has very few lysines, it is
relatively easy to create antibody-HRP conjugates without
significant HRP polymerization.
[0387] Optionally, linkers may be used between the labels and the
antibodies of the invention, e.g., as described in U.S. Pat. No.
4,831,175. Antibodies or, antigen-binding fragments thereof may be
directly labeled with radioactive iodine, indium, yttrium, or other
radioactive particle known in the art, e.g., as described in U.S.
Pat. No. 5,595,721.
[0388] Accordingly, the present invention also provides a complex
comprising [0389] (i) the antibody, or the antigen binding fragment
thereof, according to the present invention;
[0390] and [0391] (ii) a label as described above.
[0392] Accordingly, such a complex is preferably a label conjugated
to the antibody, or the antigen binding fragment thereof, according
to the present invention. Preferably, the label and the antibody,
or the antigen binding fragment thereof, according to the present
invention are covalently linked.
[0393] For example, the complex according to the present invention
may be a fusion protein comprising (i) the antibody according to
the present invention and (ii) a label, which is a peptide or a
protein, such as a fluorescent peptide or protein, e.g. EGFP.
[0394] In another aspect, the present invention also provides a
nucleic acid molecule comprising a polynucleotide encoding the
antibody, or the antigen binding fragment thereof, according to the
present invention as described above or the complex according to
the present invention as described above, in particular if the
complex according to the present invention is a fusion protein as
described above.
[0395] Examples of nucleic acid molecules and/or polynucleotides
include, e.g., a recombinant polynucleotide, a vector, an
oligonucleotide, an RNA molecule such as an rRNA, an mRNA, an
miRNA, an siRNA, or a tRNA, or a DNA molecule such as a cDNA.
[0396] Nucleic acid sequences encoding part or all of the light and
heavy chains and CDRs of the antibodies of the present invention
are preferred. Preferably provided herein are thus nucleic acid
sequences encoding part or all of the light and heavy chains, in
particular VH and VL sequences and CDRs of the exemplary antibodies
of the invention. Tables 4 and 5 provide the SEQ ID numbers for the
amino acid sequences of the CDRs and VH and VL of exemplary
antibodies according to the present invention.
[0397] Table 6 below provides the SEQ ID numbers for exemplary
nucleic acid sequences encoding the CDRs and VH and VL of exemplary
antibodies according to the present invention. Due to the
redundancy of the genetic code, the present invention also
comprises sequence variants of these nucleic acid sequences and in
particular such sequence variants, which encode the same amino acid
sequences.
[0398] A nucleic acid molecule is a molecule comprising, preferably
consisting of nucleic acid components. The term nucleic acid
molecule preferably refers to DNA or RNA molecules. In particular,
it is used synonymous with the term "polynucleotide". Preferably, a
nucleic acid molecule is a polymer comprising or consisting of
nucleotide monomers which are covalently linked to each other by
phosphodiester-bonds of a sugar/phosphate-backbone. The term
"nucleic acid molecule" also encompasses modified nucleic acid
molecules, such as base-modified, sugar-modified or
backbone-modified etc. DNA or RNA molecules.
[0399] Table 6 shows exemplary nucleic acid sequences of the CDR's
and the heavy chain variable region (VH) and the light chain
variable region (VL) of three exemplary antibodies according to the
present invention ("ZKA15", "ZKA25", "ZKA35"):
TABLE-US-00009 SEQ ID ZKA15 NO. Nucleic acid sequence CDRH1 100
ggtggcttcatcaatagttactac CDRH2 101 atctataaaagtgggagcacc CDRH3 102
gcgagagatccctacggtgactacgttaaggcttttgatat t CDRL1 103
cagagcctcctgcatagtaatggatacaactat CDRL2 104 ttgggttct CDRL2 105
ctgatctatttgggttctaatcgggcc long CDRL3 106 atgcaagctctacaaactgtcact
VH 107 caggtgcagctgcaggagtcggggccaggactggtgaagcc
ttcggagaccctgtccctcacctgcactgtctccggtggct
tcatcaatagttactactggagctggatccggcagcccgcc
gggaagggactggagtggattgggcgtatctataaaagtgg
gagcaccaactacaacccctccctcaagagtcgagtcacca
tgtcactagacacgtccaagtaccagttctccctgaagctg
aggtctgtgaccgccgctgacacggccgtgtattactgtgc
gagagatccctacggtgactacgttaaggcttttgatattt
ggggccaagggacaatggtcaccgtctcttcag VL 108
gatattgtgatgactcagtctccactctccctgcccgtcac
ccctggagagccggcctccatctcctgcaggtctagtcaga
gcctcctgcatagtaatggatacaactatttgaattggtac
ctgcagaagccagggcagtctccacagctcctgatctattt
gggttctaatcgggcctccggggtccctgacaggttcagtg
gcagtggatcaggcacagattttacactgaaaatcagcaga
gtggaggctgaggatgttggggtttattactgcatgcaagc
tctacaaactgtcactttcggccctgggaccaaagtggata tcaaac SEQ ID ZKA25 NO.
Nucleic acid sequence CDRH1 118 ggattcacctttagaagtcattgg CDRH2 119
ataaaggaagatggatatgagaaa CDRH3 120
gcgagagatttgagggtatatagtgggagaggtttcgaccc c CDRL1 121
aaattgggggataaatat CDRL2 122 caagatagc CDRL2 123
gtcatctatcaagatagcaagcggccc long CDRL3 124
caggcgtgggacagcagcactgtggta VH 125
gaggtgcagttggtggagtctgggggaggcttggtccggcc
tggggggtccctgagactctcctgtgcagcctctggattca
cctttagaagtcattggatgagttgggtccgccaggctcca
gggaaggggctggagtgggtggccaacataaaggaagatgg
atatgagaaatactatgtggactctgtgaagggccgattca
ccatctccagagacaacgccaagaactcactgtatctgcaa
atgaagagcctgagagccgaggacacggccgtgtattactg
tgcgagagatttgagggtatatagtgggagaggtttcgacc
cctggggccagggaaccctggtcaccgtctcctcag VL 126
tcctatgagctgactcagccaccctcactgtccgtgtcccc
aggacagacagccagcatcacctgctctggagataaattgg
gggataaatatgcttgctggtatcagcagaagccaggccag
tcccctgtgttggtcatctatcaagatagcaagcggccctc
agggatccctgcgcgattctctggctccaactctgggaaca
cagccactctgaccatcagcgggacccaggctatggatgag
gctgactattactgtcaggcgtgggacagcagcactgtggt
attcggtggagggaccaagctgaccgtcctag SEQ ID ZKA35 NO. Nucleic acid
sequence CDRH1 136 ggtggctccatcagcactggtggttactac CDRH2 137
atctattacagtgggaacacc CDRH3 138
gcgaaaggaggagggagggagcgaccctttgactac CDRL1 139
agctccaacatcggaagaaattat CDRL2 140 aggaataat CDRL2 141
ctcatctataggaataatcagcggccc long CDRL3 142
gtagcatgggatgacagccggagtggttttgtggta VH 143
caggtgcagctgcaggagtcgggcccaggactggtgaagcc
ttcacagaccctgtccctcacctgcactgtctctggtggct
ccatcagcactggtggttactactggagctggatccgccag
cacccagggaagggcctggagtggattggttacatctatta
cagtgggaacacctactacaacccgtccctcaagagtcgag
ttaccatatcagttgacacctctaagaagcagttctccctg
aagctgagctctgtgactgccgcggacacggccgtgtatta
ctgtgcgaaaggaggagggagggagcgaccctttgactact
ggggccagggaaccctggtcaccgtctcctcag VL 144
cagtctgtgctgactcagccaccctcagcgtctgggacccc
cgggcagagggtcaccatctcttgttctggaagcagctcca
acatcggaagaaattatgtagactggtaccagcaactccca
ggaacggcccccaaactcctcatctataggaataatcagcg
gccctcaggggtccctgagcgattctctggctccaagtctg
gcacctcagcctccctggccatcagtgggctccggtccgag
gatgaggctgattattactgtgtagcatgggatgacagccg
gagtggttttgtggtattcggcggagggaccaaggtgaccg tcctag
[0400] Preferably, the sequence of the nucleic acid molecule
according to the present invention comprises or consists of a
nucleic acid sequence according to any one of SEQ ID NOs: 100-108,
118-126, and 136-144.
[0401] It is also preferred that nucleic acid sequences according
to the invention include nucleic acid sequences having at least
70%, at least 75%, at least 80%, at least 85%, at least 88%, at
least 90%, at least 92%, at least 95%, at least 96%, at least 97%,
at least 98% or at least 99% identity to the nucleic acid encoding
a CDR, a VH sequence and/or a VL sequence used in an (exemplary)
antibody according to the present invention, for example to the
sequences shown in Table 6.
[0402] In general, the nucleic acid molecule may be manipulated to
insert, delete or alter certain nucleic acid sequences. Changes
from such manipulation include, but are not limited to, changes to
introduce restriction sites, to amend codon usage, to add or
optimize transcription and/or translation regulatory sequences,
etc. It is also possible to change the nucleic acid to alter the
encoded amino acids. For example, it may be useful to introduce one
or more (e.g., 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, etc.) amino acid
substitutions, deletions and/or insertions into the antibody's
amino acid sequence. Such point mutations can modify effector
functions, antigen-binding affinity, post-translational
modifications, immunogenicity, etc., can introduce amino acids for
the attachment of covalent groups (e.g., labels) or can introduce
tags (e.g., for purification purposes). Mutations can be introduced
in specific sites or can be introduced at random, followed by
selection (e.g., molecular evolution). For instance, one or more
nucleic acids encoding any of the CDR regions, a VH sequence and/or
a VL sequence of an (exemplary) antibody of the invention can be
randomly or directionally mutated to introduce different properties
in the encoded amino acids. Such changes can be the result of an
iterative process wherein initial changes are retained and new
changes at other nucleotide positions are introduced. Further,
changes achieved in independent steps may be combined. Different
properties introduced into the encoded amino acids may include, but
are not limited to, enhanced affinity.
[0403] Further included within the scope of the invention are
vectors, for example, expression vectors, comprising a nucleic acid
molecule according to the present invention. Preferably, a vector
comprises a nucleic acid molecule as described above.
[0404] The term "vector" refers to a nucleic acid molecule,
preferably to a recombinant nucleic acid molecule, i.e. a nucleic
acid molecule which does not occur in nature. A vector in the
context of the present invention is suitable for incorporating or
harboring a desired nucleic acid sequence. Such vectors may be
storage vectors, expression vectors, cloning vectors, transfer
vectors etc. A storage vector is a vector which allows the
convenient storage of a nucleic acid molecule. Thus, the vector may
comprise a sequence corresponding, e.g., to a desired antibody or
antibody fragment thereof according to the present invention. An
expression vector may be used for production of expression products
such as RNA, e.g. mRNA, or peptides, polypeptides or proteins. For
example, an expression vector may comprise sequences needed for
transcription of a sequence stretch of the vector, such as a
promoter sequence. A cloning vector is typically a vector that
contains a cloning site, which may be used to incorporate nucleic
acid sequences into the vector. A cloning vector may be, e.g., a
plasmid vector or a bacteriophage vector. A transfer vector may be
a vector which is suitable for transferring nucleic acid molecules
into cells or organisms, for example, viral vectors. A vector in
the context of the present invention may be, e.g., an RNA vector or
a DNA vector.
[0405] Preferably, a vector is a DNA molecule. For example, a
vector in the sense of the present application comprises a cloning
site, a selection marker, such as an antibiotic resistance factor,
and a sequence suitable for multiplication of the vector, such as
an origin of replication. Preferably, a vector in the context of
the present application is a plasmid vector.
[0406] In a further aspect, the present invention also provides
cell expressing the antibody, or the antigen binding fragment
thereof, according to the present invention; and/or comprising the
vector according the present invention.
[0407] Examples of such cells include but are not limited to,
eukaryotic cells, e.g., yeast cells, animal cells or plant cells.
Preferably, the cells are mammalian cells, more preferably a
mammalian cell line. Preferred examples include human cells, CHO
cells, HEK293T cells, PER.C6 cells, NS0 cells, human liver cells,
myeloma cells or hybridoma cells.
[0408] In particular, the cell may be transfected with a vector
according to the present invention, preferably with an expression
vector. The term "transfection" refers to the introduction of
nucleic acid molecules, such as DNA or RNA (e.g. mRNA) molecules,
into cells, preferably into eukaryotic cells. In the context of the
present invention, the term "transfection" encompasses any method
known to the skilled person for introducing nucleic acid molecules
into cells, preferably into eukaryotic cells, such as into
mammalian cells. Such methods encompass, for example,
electroporation, lipofection, e.g. based on cationic lipids and/or
liposomes, calcium phosphate precipitation, nanoparticle based
transfection, virus based transfection, or transfection based on
cationic polymers, such as DEAE-dextran or polyethylenimine etc.
Preferably, the introduction is non-viral.
[0409] Moreover, the cells of the present invention may be
transfected stably or transiently with the vector according to the
present invention, e.g. for expressing the antibody, or the antigen
binding fragment thereof, according to the present invention.
Preferably, the cells are stably transfected with the vector
according to the present invention encoding the antibody, or the
antigen binding fragment thereof, according to the present
invention. Alternatively, it is also preferred that the cells are
transiently transfected with the vector according to the present
invention encoding the antibody, or the antigen binding fragment
thereof, according to the present invention.
[0410] In a further aspect, the present invention also provides a
composition comprising the antibody, or the antigen binding
fragment thereof, according to the present invention; the complex
according to the present invention as described above; the nucleic
acid molecule according to the present invention as described
above; the vector according to the present invention as described
above; or the cell according to the present invention as described
above. A composition comprising the antibody, or the antigen
binding fragment thereof, according to the present invention or the
complex according to the present invention as described above is
preferred.
[0411] Such a composition may be a pharmaceutical composition as
described above in the context of the neutralizing antibodies,
whereby the detailed description and preferred embodiments of such
a pharmaceutical composition as described above apply accordingly
to the antibody, or the antigen binding fragment thereof, according
to the present invention, that binds to ZIKV NS1 protein. However,
the composition may also be used for non-pharmaceutical purposes,
such as in diagnosis (of ZIKV infection) or for analytical
purposes.
[0412] Preferably, the composition is in liquid form, e.g. to
provide the antibody or the antigen binding fragment thereof in a
liquid for direct use, e.g. in a diagnosis assay. The liquid
(vehicle) may be chosen according to the purpose, e.g. depending on
the assay. Preferably, the composition according to the present
invention comprises PBS (phosphate-buffered saline) or another
buffer. Such buffers are preferably biological buffers, and the
composition may thus comprise any of MES, BIS-TRIS, ADA, PIPES,
ACES, MOPSO, BIS-TRIS propane, BES, MOPS, TES, HEPES, DIPSO, TAPSO,
Trizma, POPSO, HEPPS, TRICINE, Gly-Gly, BICINE, HEPBS, TAPS, AMPD,
AMPSO, CHES, CAPSO, AMP, CAPS and CABS. It is also preferred that
the composition comprises Ringer's solution. In addition, the
composition may also comprise Tris, e.g., Tris-HCl.
[0413] The composition according to the present invention may also
comprise a detergent e.g., a Tween (polysorbate), such as Tween 20
or Tween 80. Detergents are preferably present at low levels e.g.,
less than 0.01%. Compositions may also include sodium salts (e.g.,
sodium chloride) to give tonicity. For example, a concentration of
10.+-.2 mg/ml NaCl is typical.
[0414] In addition, the composition according to the present
invention may optionally comprise a protein stabilizer, such as BSA
(bovine serum albumin) or HSA (human serum albumin). Further
examples of protein stabilizers, which may optionally be included
in the composition according to the present invention, include
buffers, e.g. as described above; salts, such as sodium chloride;
amino acids, such as histidine, glycine, and arginine;
polyols/disaccharides/polysaccharides, such as trehalose and
sucrose (disaccharides), mannitol and sorbitol (sugar alcohols);
surfactants, such as polysorbate 20, polysorbate 80, and proteins
like HSA or BSA; polymers, such as dextran and polyethylene glycol;
and antioxidants.
[0415] Furthermore, the composition according to the present
invention may optionally comprise a preservative, such as sodium
azide. Preservatives are typically used to prevent microbial
contamination.
[0416] In a further aspect, the present invention also provides a
kit of parts comprising the antibody, or the antigen binding
fragment thereof, according to the present invention, the complex
according to the present invention or the composition according to
the present invention.
[0417] Such a kit of parts may optionally further comprise one or
more of the following: [0418] (i) one or more solutions, e.g. to be
used in a diagnosis assay, e.g. to dilute the antibody or the
antigen binding fragment thereof; [0419] (ii) a leaflet, e.g. with
instructions to use; [0420] (iii) a label as described above and,
optionally, solutions and/or further components required for
labeling; and/or [0421] (iv) vessels or devices, e.g. useful in a
diagnosis assay, for example one or more ELISA plates.
[0422] Preferably, the kit according to the present invention as
described above also comprises a substrate for the development of
the color. Examples of such a substrate include p-NPP, in
particular in case of detection through alkaline phosphatase; or an
enzyme like ABTS, TMB or OPD, in particular in case of use of
horse-radish peroxidase (HRP). Optionally, the substrate may be
diluted in an appropriate buffer, e.g. a buffer as described above
in the context of the composition according to the present
invention. Alternatively, the substrate and the buffer may be
provided as separate entities in the kit.
[0423] With regard to the label, the kit according to the present
invention may also comprise an enzyme conjugated streptavidin, or
another system to detect the binding of the probe antibody. For
example, the probe antibody may be made in murinized form and in
this case the binding may be detected with an anti-mouse secondary
antibody--without the need for biotinylation. The anti-mouse
secondary antibody is typically polyclonal and/or cross-adsorbed
for not reacting with human antibodies.
[0424] Moreover, the kit according to the present invention
preferably comprises one or more ELISA plates. More preferably,
those ELISA plates are pre-coated with ZIKV-NS1 protein.
Optionally, such pre-coated ELISA-plates may be pre-blocked.
[0425] Diagnosis of Zika Virus Infection
[0426] In a further aspect, the present invention also provides the
use of the antibody, or the antigen binding fragment thereof,
according to the present invention, that binds to ZIKV NS1 protein,
the complex according to the present invention, the composition
according to the present invention or the kit of parts according to
the present invention in diagnosis of Zika virus (ZIKV)
infection.
[0427] Diagnosis of Zika virus (ZIKV) infection is typically
performed in vitro, e.g. in an isolated sample of the subject to be
diagnosed. Preferred isolated samples of the subject include
samples of a body fluid and tissue samples. A sample of a body
fluid is more preferred. Preferred body fluids for diagnosis of
ZIKV infection include blood (e.g. whole blood, plasma, serum),
saliva and urine. Blood, in particular plasma or serum, is most
preferred.
[0428] Accordingly, the present invention also provides the use of
the antibody, or the antigen binding fragment thereof, according to
the present invention, that binds to ZIKV NS1 protein, the complex
according to the present invention, the composition according to
the present invention or the kit of parts according to the present
invention in determining whether an isolated sample (of a body
fluid), such as an isolated blood sample, is infected with Zika
virus. As described above, preferred body fluids for diagnosis of
ZIKV infection include blood (e.g. whole blood, plasma, serum),
saliva and urine. Blood, in particular plasma or serum, is more
preferred.
[0429] For diagnosis of Zika virus infection, different diagnosis
assays may be used. Preferred diagnosis assays are immunoassays.
Preferred examples of immunoassays include ELISA,
immunofluorescence, immunohistochemistry and flow cytometry.
Preferably, diagnosis includes ELISA. For example, a standard
ELISA, a sandwich ELISA or a blockade of binding assay may be
used.
[0430] Preferably, the diagnosis assay detects [0431] (i) (the
presence of) ZIKV NS1 protein itself; and/or [0432] (ii) (the
presence of) anti-ZIKV NS1 antibodies
[0433] in an (isolated) sample of a subject to be diagnosed.
[0434] Preferably, a blockade-of-binding assay is used. In this
assay, an isolated sample from a subject to be diagnosed (e.g., a
sample of a body fluid, such as blood (e.g. whole blood, plasma,
serum), saliva and urine) is added to an ELISA plate coated with
ZIKV NS1 protein and incubated (for example, for at least about 30
min or at least about one hour) to allow for binding. Thereafter,
the antibody, or the antigen binding fragment thereof, according to
the present invention is added (as "probe antibody"), wherein the
antibody or the antigen binding fragment thereof according to the
present invention is preferably labelled, e.g. biotinylated or
conjugated to horseradish peroxidase (HRP). After another
incubation time (e.g., at least about 1 min, preferably at least
about 3 min, more preferably at least about 5 min, even more
preferably at least about 10 min, most preferably at least about 15
min), inhibition of binding of the antibody or the antigen binding
fragment thereof according to the present invention can be
determined.
[0435] In general, inhibition of binding shows the presence of
anti-ZIKV NS1 antibodies in the sample of the subject, thus
indicating ZIKV infection of the subject. In samples of
non-infected subjects, in contrast, typically no inhibition of
binding is expected. Importantly, such an assay using the ZIKV
NS1-binding antibodies of the present invention does not score
positive in subjects that were already infected with other
flaviviruses. Flaviviruses typically induce a large number of
antibodies that are cross-reactive with ZIKV. In other words this
assay is highly specific and not affected by cross-reactive
Abs.
[0436] Accordingly, the present invention also provides a
Blockade-of-binding assay for in-vitro diagnosis of Zika virus
infection comprising the following steps: [0437] (i) adding an
isolated sample from a subject to be diagnosed to a plate coated
with ZIKV NS1 protein and incubating said sample on said plate,
[0438] (ii) adding an antibody, or an antigen binding fragment
thereof, that specifically binds to NS1 protein of ZIKA virus or a
complex including the antibody, or an antigen binding fragment
thereof, and a label, [0439] (iii) determining inhibition of
binding of said antibody or antigen-binding fragment thereof.
[0440] The antibody, or antigen binding fragment thereof, in (ii)
or in the complex referenced in (ii) may also have any of the
following characteristics: [0441] the antibody, or antigen binding
fragment thereof, does essentially not bind to Dengue virus (DENV)
NS1 protein; [0442] the antibody, or antigen binding fragment
thereof, does essentially not bind to Yellow fever virus (YFV) NS1
protein, West nile virus (WNV) NS1 protein, Japanese encephalitis
virus (JEV) NS1 protein and/or to Tick-borne encephalitis virus
(TBEV) NS1 protein; [0443] the antibody, or antigen binding
fragment thereof, is a human antibody; [0444] the antibody, or
antigen binding fragment thereof, is a monoclonal antibody,
preferably a human monoclonal antibody, more preferably a
recombinant human monoclonal antibody; [0445] characterized in that
the antibody, or antigen binding fragment thereof, is of the IgG
type, preferably of the IgG1 type; [0446] the antibody, or antigen
binding fragment thereof, comprises an Fc moiety; or additionally
that the antibody, or antigen binding fragment thereof, comprises a
CH2 L4A mutation, a CH2 L5A mutation, or both; [0447] the antibody,
or antigen binding fragment thereof, does not comprise an Fc
region, in particular an Fc moiety; [0448] the antibody, or the
antigen binding fragment thereof, is a purified antibody, a single
chain antibody, Fab, Fab', F(ab')2, Fv or scFv; [0449] the
antibody, or the antigen binding fragment thereof, is biotinylated,
such as a biotinylated Fab, Fab', or F(ab')2 fragment; [0450] the
antibody, or antigen binding fragment thereof, binds to antigenic
site S1 and/or to antigenic site S2 of Zika virus NS1 protein; or
additionally that the antibody, or antigen binding fragment
thereof, binds to antigenic site S2 of Zika virus NS1 protein;
[0451] the antibody or antigen binding fragment thereof, comprises
a heavy chain comprising at least one CDRH1, at least one CDRH2 and
at least one CDRH3 and a light chain comprising at least one CDRL1,
at least one CDRL2 and at least one CDRL3, wherein at least one
CDR, preferably the at least one heavy chain CDRH3, comprises or
consists of an amino acid sequence according to any of SEQ ID NOs:
93, 111, 129, 155, 159, 163, 167, 171, 175, 179, 183, 187, 191,
195, 199, 203, 207, 211, 215, 219, 223, 227, 231, and 235; or a
functional sequence variant thereof having at least 70%, at least
75%, at least 80%, at least 85%, at least 88%, at least 90%, at
least 92%, at least 95%, at least 96%, at least 97%, at least 98%
or at least 99% sequence identity; [0452] the antibody or antigen
binding fragment thereof, comprises a heavy chain comprising at
least one CDRH1, at least one CDRH2 and at least one CDRH3 and a
light chain comprising at least one CDRL1, at least one CDRL2 and
at least one CDRL3, wherein at least one CDR, preferably the at
least one heavy chain CDRH3, comprises or consists of an amino acid
sequence according to any of SEQ ID NOs: 93, 111 and 129, or a
functional sequence variant thereof having at least 70%, at least
75%, at least 80%, at least 85%, at least 88%, at least 90%, at
least 92%, at least 95%, at least 96%, at least 97%, at least 98%
or at least 99% sequence identity; [0453] the antibody or antigen
binding fragment thereof, comprises a heavy chain comprising at
least one CDRH1, at least one CDRH2 and at least one CDRH3 and a
light chain comprising at least one CDRL1, at least one CDRL2 and
at least one CDRL3, wherein at least one CDR, preferably the at
least one heavy chain CDRH3, comprises or consists of an amino acid
sequence according to SEQ ID NO: 129, or a functional sequence
variant thereof having at least 70%, at least 75%, at least 80%, at
least 85%, at least 88%, at least 90%, at least 92%, at least 95%,
at least 96%, at least 97%, at least 98% or at least 99% sequence
identity; [0454] the antibody or antigen binding fragment thereof,
comprises a heavy chain comprising at least one CDRH1, at least one
CDRH2 and at least one CDRH3 and a light chain comprising at least
one CDRL1, at least one CDRL2 and at least one CDRL3, wherein
[0455] (i) the at least one heavy chain CDRH1 comprises an amino
acid sequence according to any of SEQ ID NOs: 91, 109, 127, 153,
157, 161, 165, 169, 173, 177, 181, 185, 189, 193, 197, 201, 205,
209, 213, 217, 221, 225, 229, and 233; or a functional sequence
variant thereof having at least 70%, at least 75%, at least 80%, at
least 85%, at least 88%, at least 90%, at least 92%, at least 95%,
at least 96%, at least 97%, at least 98% or at least 99% sequence
identity; [0456] (ii) the at least one CDRH2 comprises an amino
acid sequence according to any of SEQ ID NOs: 92, 110, 128, 154,
158, 162, 166, 170, 174, 178, 182, 186, 190, 194, 198, 202, 206,
210, 214, 218, 222, 226, 230, and 234; or a functional sequence
variant thereof having at least 70%, at least 75%, at least 80%, at
least 85%, at least 88%, at least 90%, at least 92%, at least 95%,
at least 96%, at least 97%, at least 98% or at least 99% sequence
identity; and/or [0457] (iii) the at least one heavy chain CDRH3
comprises an amino acid sequence according to any of SEQ ID NOs:
93, 111, 129, 155, 159, 163, 167, 171, 175, 179, 183, 187, 191,
195, 199, 203, 207, 211, 215, 219, 223, 227, 231, and 235; or a
functional sequence variant thereof having at least 70%, at least
75%, at least 80%, at least 85%, at least 88%, at least 90%, at
least 92%, at least 95%, at least 96%, at least 97%, at least 98%
or at least 99% sequence identity; or additionally the antibody or
antigen binding fragment thereof, comprises a heavy chain
comprising at least one CDRH1, at least one CDRH2 and at least one
CDRH3 and a light chain comprising at least one CDRL1, at least one
CDRL2 and at least one CDRL3, wherein (i) the at least one heavy
chain CDRH1 comprises an amino acid sequence according to any of
SEQ ID NOs: 91, 109 and 127, or a functional sequence variant
thereof having at least 70%, at least 75%, at least 80%, at least
85%, at least 88%, at least 90%, at least 92%, at least 95%, at
least 96%, at least 97%, at least 98% or at least 99% sequence
identity; [0458] (ii) the at least one CDRH2 comprises an amino
acid sequence according to any of SEQ ID NOs: 92, 110 and 128, or a
functional sequence variant thereof having at least 70%, at least
75%, at least 80%, at least 85%, at least 88%, at least 90%, at
least 92%, at least 95%, at least 96%, at least 97%, at least 98%
or at least 99% sequence identity; and/or [0459] (iii) the at least
one heavy chain CDRH3 comprises an amino acid sequence according to
any of SEQ ID NOs: 93, 111 and 129, or a functional sequence
variant thereof having at least 70%, at least 75%, at least 80%, at
least 85%, at least 88%, at least 90%, at least 92%, at least 95%,
at least 96%, at least 97%, at least 98% or at least 99% sequence
identity; or additionally; the antibody or antigen binding fragment
thereof, comprises a heavy chain comprising at least one CDRH1, at
least one CDRH2 and at least one CDRH3 and a light chain comprising
at least one CDRL1, at least one CDRL2 and at least one CDRL3,
wherein [0460] (i) the at least one heavy chain CDRH1 comprises an
amino acid sequence according to SEQ ID NO: 127, or a functional
sequence variant thereof having at least 70%, at least 75%, at
least 80%, at least 85%, at least 88%, at least 90%, at least 92%,
at least 95%, at least 96%, at least 97%, at least 98% or at least
99% sequence identity; [0461] (ii) the at least one CDRH2 comprises
an amino acid sequence according to SEQ ID NO: 128, or a functional
sequence variant thereof having at least 70%, at least 75%, at
least 80%, at least 85%, at least 88%, at least 90%, at least 92%,
at least 95%, at least 96%, at least 97%, at least 98% or at least
99% sequence identity; and/or [0462] (iii) the at least one heavy
chain CDRH3 comprises an amino acid sequence according to SEQ ID
NO: 129, or a functional sequence variant thereof having at least
70%, at least 75%, at least 80%, at least 85%, at least 88%, at
least 90%, at least 92%, at least 95%, at least 96%, at least 97%,
at least 98% or at least 99% sequence identity; [0463] the
antibody, or the antigen binding fragment thereof, comprises a
heavy chain comprising at least one CDRH1, at least one CDRH2 and
at least one CDRH3 and a light chain comprising at least one CDRL1,
at least one CDRL2 and at least one CDRL3, wherein [0464] (i) the
at least one CDRL1 comprises an amino acid sequence according to
any of SEQ ID NOs: 94, 112 and 130, or a functional sequence
variant thereof having at least 70%, at least 75%, at least 80%, at
least 85%, at least 88%, at least 90%, at least 92%, at least 95%,
at least 96%, at least 97%, at least 98% or at least 99% sequence
identity; [0465] (ii) the at least one CDRL2 comprises an amino
acid sequence according to any of SEQ ID NOs: 95, 96, 113, 114, 131
and 132, or a functional sequence variant thereof having at least
70%, at least 75%, at least 80%, at least 85%, at least 88%, at
least 90%, at least 92%, at least 95%, at least 96%, at least 97%,
at least 98% or at least 99% sequence identity; and/or [0466] (iii)
the at least one CDRL3 comprises an amino acid sequence according
to any of SEQ ID NOs: 97, 115 and 133, or a functional sequence
variant thereof having at least 70%, at least 75%, at least 80%, at
least 85%, at least 88%, at least 90%, at least 92%, at least 95%,
at least 96%, at least 97%, at least 98% or at least 99% sequence
identity; or additionally the antigen binding fragment thereof,
comprises a heavy chain comprising at least one CDRH1, at least one
CDRH2 and at least one CDRH3 and a light chain comprising at least
one CDRL1, at least one CDRL2 and at least one CDRL3, wherein
[0467] (i) the at least one CDRL1 comprises an amino acid sequence
according to SEQ ID NO: 130, or a functional sequence variant
thereof having at least 70%, at least 75%, at least 80%, at least
85%, at least 88%, at least 90%, at least 92%, at least 95%, at
least 96%, at least 97%, at least 98% or at least 99% sequence
identity; [0468] (ii) the at least one CDRL2 comprises an amino
acid sequence according to SEQ ID NO: 131 or 132, or a functional
sequence variant thereof having at least 70%, at least 75%, at
least 80%, at least 85%, at least 88%, at least 90%, at least 92%,
at least 95%, at least 96%, at least 97%, at least 98% or at least
99% sequence identity; and/or [0469] (iii) the at least one CDRL3
comprises an amino acid sequence according to SEQ ID NO: 133, or a
functional sequence variant thereof having at least 70%, at least
75%, at least 80%, at least 85%, at least 88%, at least 90%, at
least 92%, at least 95%, at least 96%, at least 97%, at least 98%
or at least 99% sequence identity; [0470] the antigen binding
fragment thereof, comprises CDRH1, CDRH2, and CDRH3 amino acid
sequences (i) according to SEQ ID NOs: 91-93; or functional
sequence variants thereof having at least 70%, at least 75%, at
least 80%, at least 85%, at least 88%, at least 90%, at least 92%,
at least 95%, at least 96%, at least 97%, at least 98% or at least
99% sequence identity; (ii) according to SEQ ID NOs: 109-111; or
functional sequence variants thereof having at least 70%, at least
75%, at least 80%, at least 85%, at least 88%, at least 90%, at
least 92%, at least 95%, at least 96%, at least 97%, at least 98%
or at least 99% sequence identity; (iii) according to SEQ ID NOs:
127-129; or functional sequence variants thereof having at least
70%, at least 75%, at least 80%, at least 85%, at least 88%, at
least 90%, at least 92%, at least 95%, at least 96%, at least 97%,
at least 98% or at least 99% sequence identity; (iv) according to
SEQ ID NOs: 153-155; or functional sequence variants thereof having
at least 70%, at least 75%, at least 80%, at least 85%, at least
88%, at least 90%, at least 92%, at least 95%, at least 96%, at
least 97%, at least 98% or at least 99% sequence identity; (v)
according to SEQ ID NOs: 157-159; or functional sequence variants
thereof having at least 70%, at least 75%, at least 80%, at least
85%, at least 88%, at least 90%, at least 92%, at least 95%, at
least 96%, at least 97%, at least 98% or at least 99% sequence
identity; (vi) according to SEQ ID NOs: 161-163; or functional
sequence variants thereof having at least 70%, at least 75%, at
least 80%, at least 85%, at least 88%, at least 90%, at least 92%,
at least 95%, at least 96%, at least 97%, at least 98% or at least
99% sequence identity; (vii) according to SEQ ID NOs: 165-167; or
functional sequence variants thereof having at least 70%, at least
75%, at least 80%, at least 85%, at least 88%, at least 90%, at
least 92%, at least 95%, at least 96%, at least 97%, at least 98%
or at least 99% sequence identity; (viii) according to SEQ ID NOs:
169-171; or functional sequence variants thereof having at least
70%, at least 75%, at least 80%, at least 85%, at least 88%, at
least 90%, at least 92%, at least 95%, at least 96%, at least 97%,
at least 98% or at least 99% sequence identity; (ix) according to
SEQ ID NOs: 173-175; or functional sequence variants thereof having
at least 70%, at least 75%, at least 80%, at least 85%, at least
88%, at least 90%, at least 92%, at least 95%, at least 96%, at
least 97%, at least 98% or at least 99% sequence identity; (x)
according to SEQ ID NOs: 177-179; or functional sequence variants
thereof having at least 70%, at least 75%, at least 80%, at least
85%, at least 88%, at least 90%, at least 92%, at least 95%, at
least 96%, at least 97%, at least 98% or at least 99% sequence
identity; (xi) according to SEQ ID NOs: 181-183; or functional
sequence variants thereof having at least 70%, at least 75%, at
least 80%, at least 85%, at least 88%, at least 90%, at least 92%,
at least 95%, at least 96%, at least 97%, at least 98% or at least
99% sequence identity; (xii) according to SEQ ID NOs: 185-187; or
functional sequence variants thereof having at least 70%, at least
75%, at least 80%, at least 85%, at least 88%, at least 90%, at
least 92%, at least 95%, at least 96%, at least 97%, at least 98%
or at least 99% sequence identity; (xiii) according to SEQ ID NOs:
189-191; or functional sequence variants thereof having at least
70%, at least 75%, at least 80%, at least 85%, at least 88%, at
least 90%, at least 92%, at least 95%, at least 96%, at least 97%,
at least 98% or at least 99% sequence identity; (xiv) according to
SEQ ID NOs: 193-195; or functional sequence variants thereof having
at least 70%, at least 75%, at least 80%, at least 85%, at least
88%, at least 90%, at least 92%, at least 95%, at least 96%, at
least 97%, at least 98% or at least 99% sequence identity; (xv)
according to SEQ ID NOs: 197-199; or functional sequence variants
thereof having at least 70%, at least 75%, at least 80%, at least
85%, at least 88%, at least 90%, at least 92%, at least 95%, at
least 96%, at least 97%, at least 98% or at least 99% sequence
identity; (xvi) according to SEQ ID NOs: 201-203; or functional
sequence variants thereof having at least 70%, at least 75%, at
least 80%, at least 85%, at least 88%, at least 90%, at least 92%,
at least 95%, at least 96%, at least 97%, at least 98% or at least
99% sequence identity; (xvii) according to SEQ ID NOs: 205-207; or
functional sequence variants thereof having at least 70%, at least
75%, at least 80%, at least 85%, at least 88%, at least 90%, at
least 92%, at least 95%, at least 96%, at least 97%, at least 98%
or at least 99% sequence identity; (xviii) according to SEQ ID NOs:
209-211; or functional sequence variants thereof having at least
70%, at least 75%, at least 80%, at least 85%, at least 88%, at
least 90%, at least 92%, at least 95%, at least 96%, at least 97%,
at least 98% or at least 99% sequence identity; (xix) according to
SEQ ID NOs: 213-215; or functional sequence variants thereof having
at least 70%, at least 75%, at least 80%, at least 85%, at least
88%, at least 90%, at least 92%, at least 95%, at least 96%, at
least 97%, at least 98% or at least 99% sequence identity; (xx)
according to SEQ ID NOs: 217-219; or functional sequence variants
thereof having at least 70%, at least 75%, at least 80%, at least
85%, at least 88%, at least 90%, at least 92%, at least 95%, at
least 96%, at least 97%, at least 98% or at least 99% sequence
identity; (xxi) according to SEQ ID NOs: 221-223; or functional
sequence variants thereof having at least 70%, at least 75%, at
least 80%, at least 85%, at least 88%, at least 90%, at least 92%,
at least 95%, at least 96%, at least 97%, at least 98% or at least
99% sequence identity; (xxii) according to SEQ ID NOs: 225-227; or
functional sequence variants thereof having at least 70%, at least
75%, at least 80%, at least 85%, at least 88%, at least 90%, at
least 92%, at least 95%, at least 96%, at least 97%, at least 98%
or at least 99% sequence identity; (xxiii) according to SEQ ID NOs:
229-231; or functional sequence variants thereof having at least
70%, at least 75%, at least 80%, at least 85%, at least 88%, at
least 90%, at least 92%, at least 95%, at least 96%, at least 97%,
at least 98% or at least 99% sequence identity; or (xxiv) according
to SEQ ID NOs: 233-235; or functional sequence variants thereof
having at least 70%, at least 75%, at least 80%, at least 85%, at
least 88%, at least 90%, at least 92%, at least 95%, at least 96%,
at least 97%, at least 98% or at least 99% sequence identity; or
additionally the antibody, or the antigen binding fragment thereof,
comprises CDRH1, CDRH2, and CDRH3 amino acid sequences and CDRL1,
CDRL2, and CDRL3 amino acid sequences (i) according to SEQ ID NOs:
91-95 and 97; or functional sequence variants thereof having at
least 70%, at least 75%, at least 80%, at least 85%, at least 88%,
at least 90%, at least 92%, at least 95%, at least 96%, at least
97%, at least 98%
or at least 99% sequence identity; (ii) according to SEQ ID NOs:
91-94 and 96-97; or functional sequence variants thereof having at
least 70%, at least 75%, at least 80%, at least 85%, at least 88%,
at least 90%, at least 92%, at least 95%, at least 96%, at least
97%, at least 98% or at least 99% sequence identity; (iii)
according to SEQ ID NOs: 109-113 and 115; or functional sequence
variants thereof having at least 70%, at least 75%, at least 80%,
at least 85%, at least 88%, at least 90%, at least 92%, at least
95%, at least 96%, at least 97%, at least 98% or at least 99%
sequence identity; (iv) according to SEQ ID NOs: 109-112 and
114-115; or functional sequence variants thereof having at least
70%, at least 75%, at least 80%, at least 85%, at least 88%, at
least 90%, at least 92%, at least 95%, at least 96%, at least 97%,
at least 98% or at least 99% sequence identity; (v) according to
SEQ ID NOs:127-131 and 133; or functional sequence variants thereof
having at least 70%, at least 75%, at least 80%, at least 85%, at
least 88%, at least 90%, at least 92%, at least 95%, at least 96%,
at least 97%, at least 98% or at least 99% sequence identity; or
(vi) according to SEQ ID NOs: 127-130 and 132-133; or functional
sequence variants thereof having at least 70%, at least 75%, at
least 80%, at least 85%, at least 88%, at least 90%, at least 92%,
at least 95%, at least 96%, at least 97%, at least 98% or at least
99% sequence identity; or additionally the antibody, or the antigen
binding fragment thereof, comprises CDRH1, CDRH2, and CDRH3 amino
acid sequences and CDRL1, CDRL2, and CDRL3 amino acid sequences (i)
according to SEQ ID NOs: 127-131 and 133; or functional sequence
variants thereof having at least 70%, at least 75%, at least 80%,
at least 85%, at least 88%, at least 90%, at least 92%, at least
95%, at least 96%, at least 97%, at least 98% or at least 99%
sequence identity; or (ii) according to SEQ ID NOs: 127-130 and
132-133; or functional sequence variants thereof having at least
70%, at least 75%, at least 80%, at least 85%, at least 88%, at
least 90%, at least 92%, at least 95%, at least 96%, at least 97%,
at least 98% or at least 99% sequence identity;
[0471] the antigen binding fragment thereof, comprises a heavy
chain variable region (VH) and, optionally, a light chain variable
region (VL), wherein the heavy chain variable region (VH) comprises
or consists of an amino acid sequence according to any of SEQ ID
NOs: 98, 116, 134, 156, 160, 164, 168, 172, 176, 180, 184, 188,
192, 196, 200, 204, 208, 212, 216, 220, 224, 228, 232, and 236; or
a functional sequence variant thereof having at least 70%, at least
75%, at least 80%, at least 85%, at least 88%, at least 90%, at
least 92%, at least 95%, at least 96%, at least 97%, at least 98%
or at least 99% sequence identity; [0472] the antigen binding
fragment thereof, comprises (i) a heavy chain variable region (VH)
amino acid sequence according to SEQ ID NO: 98 or a functional
sequence variant thereof having at least 70%, at least 75%, at
least 80%, at least 85%, at least 88%, at least 90%, at least 92%,
at least 95%, at least 96%, at least 97%, at least 98% or at least
99% sequence identity and/or a light chain variable region (VL)
amino acid sequence according to SEQ ID NO: 99 or a functional
sequence variant thereof having at least 70%, at least 75%, at
least 80%, at least 85%, at least 88%, at least 90%, at least 92%,
at least 95%, at least 96%, at least 97%, at least 98% or at least
99% sequence identity; (ii) a heavy chain variable region (VH)
amino acid sequence according to SEQ ID NO: 116 or a functional
sequence variant thereof having at least 70%, at least 75%, at
least 80%, at least 85%, at least 88%, at least 90%, at least 92%,
at least 95%, at least 96%, at least 97%, at least 98% or at least
99% sequence identity and/or a light chain variable region (VL)
amino acid sequence according to SEQ ID NO: 117 or a functional
sequence variant thereof having at least 70%, at least 75%, at
least 80%, at least 85%, at least 88%, at least 90%, at least 92%,
at least 95%, at least 96%, at least 97%, at least 98% or at least
99% sequence identity; or (iii) a heavy chain variable region (VH)
amino acid sequence according to SEQ ID NO: 134 or a functional
sequence variant thereof having at least 70%, at least 75%, at
least 80%, at least 85%, at least 88%, at least 90%, at least 92%,
at least 95%, at least 96%, at least 97%, at least 98% or at least
99% sequence identity and/or a light chain variable region (VL)
amino acid sequence according to SEQ ID NO: 135 or a functional
sequence variant thereof having at least 70%, at least 75%, at
least 80%, at least 85%, at least 88%, at least 90%, at least 92%,
at least 95%, at least 96%, at least 97%, at least 98% or at least
99% sequence identity; or additionally, the antigen binding
fragment thereof, comprises a heavy chain variable region (VH)
amino acid sequence according to SEQ ID NO: 134 or a functional
sequence variant thereof having at least 70%, at least 75%, at
least 80%, at least 85%, at least 88%, at least 90%, at least 92%,
at least 95%, at least 96%, at least 97%, at least 98% or at least
99% sequence identity and/or a light chain variable region (VL)
amino acid sequence according to SEQ ID NO: 135 or a functional
sequence variant thereof having at least 70%, at least 75%, at
least 80%, at least 85%, at least 88%, at least 90%, at least 92%,
at least 95%, at least 96%, at least 97%, at least 98% or at least
99% sequence identity, or additionally, the antibody, or the
antigen binding fragment thereof, is gZKA15, gZKA25, or gZKA35,
preferably the antibody, or the antigen binding fragment thereof,
is gZKA35; or [0473] the antigen binding fragment thereof, is
labelled, for example biotinylated or conjugated to horseradish
peroxidase (HRP).
[0474] Preferably, the isolated sample from a subject to be
diagnosed is selected from blood, saliva and urine; preferably the
sample is a blood sample, such as whole blood, plasma or serum.
[0475] It is also preferred that the antibody, or the
antigen-binding fragment thereof, added in step (ii), is labelled,
preferably biotinylated or conjugated to horseradish peroxidase
(HRP).
[0476] Moreover, the isolated sample from a subject to be diagnosed
is preferably diluted, for example 1:5-1:50, preferably 1:5-1:25,
such as 1:10.
[0477] Preferably, the incubation time in step (i) is at least 5
min, preferably at least 15 min, more preferably at least 30 min,
even more preferably at least 45 min and most preferably at least
60 min.
[0478] It is furthermore preferred that in step (ii) after adding
the antibody, or the antigen binding fragment thereof, the
antibody, or the antigen binding fragment thereof, is incubated for
at least 1 min, preferably, at least 3 min, more preferably at
least 5 min, even more preferably at least 10 min and most
preferably at least 15 min.
[0479] Preferably, the antibody, or the antigen binding fragment
thereof, according to the present invention, which is used as probe
antibody in the blockade of binding assay is a preferred antibody,
or antigen binding fragment thereof, according to the present
invention. For example, the antibody, or the antigen binding
fragment thereof, according to the present invention may preferably
be an antibody, or antigen binding fragment thereof, according to
the present invention binding to antigenic site S2 of Zika virus
NS1 protein. Most preferably, the antibody, or the antigen binding
fragment thereof, according to the present invention, which is used
as probe antibody in the blockade of binding assay is an antibody,
or the antigen binding fragment thereof, comprising CDRH1, CDRH2,
and CDRH3 amino acid sequences and CDRL1, CDRL2, and CDRL3 amino
acid sequences (i) according to SEQ ID NOs: 127-131 and 133; or
functional sequence variants thereof having at least 70%, at least
75%, at least 80%, at least 85%, at least 88%, at least 90%, at
least 92%, at least 95%, at least 96%, at least 97%, at least 98%
or at least 99% sequence identity; or (ii) according to SEQ ID NOs:
127-130 and 132-133; or functional sequence variants thereof having
at least 70%, at least 75%, at least 80%, at least 85%, at least
88%, at least 90%, at least 92%, at least 95%, at least 96%, at
least 97%, at least 98% or at least 99% sequence identity.
Particularly preferably, the antibody, or the antigen binding
fragment thereof, according to the present invention, which is used
as probe antibody in the blockade of binding assay is an antibody,
or the antigen binding fragment thereof, comprising a heavy chain
variable region (VH) amino acid sequence according to SEQ ID NO:
134 or a functional sequence variant thereof having at least 70%,
at least 75%, at least 80%, at least 85%, at least 88%, at least
90%, at least 92%, at least 95%, at least 96%, at least 97%, at
least 98% or at least 99% sequence identity and/or a light chain
variable region (VL) amino acid sequence according to SEQ ID NO:
135 or a functional sequence variant thereof having at least 70%,
at least 75%, at least 80%, at least 85%, at least 88%, at least
90%, at least 92%, at least 95%, at least 96%, at least 97%, at
least 98% or at least 99% sequence identity.
[0480] For example, inhibition of binding of exemplary biotinylated
antibodies or antigen binding fragments thereof according to the
present invention may be assessed by determining the optimal
concentration of the antibody or the antigen binding fragment
thereof according to the present invention to achieve 70% maximal
binding to ZIKV NS1 protein. For example, the optimal
concentrations of the exemplary antibodies gZKA15, gZKA25 and
gZKA35 to achieve 70% maximal binding to ZIKV NS1 protein may be
38, 17 and 7 ng/ml, respectively. After performing the above
described blockade-of-binding assay, substrate, such as p-NPP may
be added and the ELISA plate may be read at 405 nm and the
percentage of inhibition of binding may be calculated by the
following equation (I):
% inhib=(1-[(OD sample-OD neg ctr)/(OD pos ctr-OD neg
ctr)]).times.100 (I)
[0481] wherein "% inhib" refers to the percentage of inhibition of
binding of the antibody or the antigen binding fragment thereof
according to the present invention to ZIKV NS1 protein; "OD sample"
refers to the optical density of the sample; "OD neg ctr" refers to
the optical density of a negative control; and "OD pos ctr" refers
to the optical density of a positive control.
[0482] This assay provides several advantages, such as the ability
to detect clinical, sub-clinical and asymptomatic ZIKV infections
at the population level, being able to distinguish them from other
flavivirus infections, such as DENV. In particular, the diagnosis
assay according to the present invention provides higher precision
than direct ELISA binding assays.
[0483] Moreover, the present invention also provides a method for
(in vitro) diagnosis of Zika infection (in an isolated sample),
wherein the antibody, or the antigen binding fragment thereof,
according to the present invention, that binds to ZIKV NS1 protein,
the complex according to the present invention, the composition
according to the present invention or the kit of parts according to
the present invention is used for determining whether an isolated
sample (of a body fluid), such as an isolated blood sample, is
infected with Zika virus.
[0484] Preferred embodiments of the above use in diagnosis apply
also for the diagnosis method. For example, preferred isolated
samples (of the subject) include samples of a body fluid and tissue
samples. A sample of a body fluid is more preferred. Preferred body
fluids for diagnosis of ZIKV infection include blood (e.g. whole
blood, plasma, serum), saliva and urine. Blood, in particular
plasma or serum, is most preferred. Moreover, preferred diagnosis
assays are immunoassays.
[0485] Preferred examples of immunoassays include ELISA,
immunofluorescence, immunohistochemistry and flow cytometry.
Preferably, diagnosis includes ELISA. Most preferably, a
blockade-of-binding assay as described above is used.
[0486] Preferably, the method for (in vitro) diagnosis of Zika
infection (in an isolated sample) comprises a step of [0487] (i)
contacting the isolated sample with the antibody, or the antigen
binding fragment thereof, according to the present invention, that
binds to ZIKV NS1 protein, the complex according to the present
invention, or the composition according to the present
invention.
[0488] More preferably, the method for (in vitro) diagnosis of Zika
infection (in an isolated sample) comprises the following steps:
[0489] (0) adding an isolated sample from a subject to be diagnosed
(e.g., a sample of a body fluid, such as blood (e.g. whole blood,
plasma, serum), saliva and urine) to an ELISA plate coated with
ZIKV NS1 protein; [0490] (i') further adding the antibody or the
antigen binding fragment thereof according to the present invention
to the ELISA plate, wherein the antibody or the antigen binding
fragment thereof according to the present invention is preferably
labelled, e.g. biotinylated; [0491] (ii) optionally, washing the
ELISA plate; and [0492] (iii) determining the inhibition of binding
of the antibody or the antigen binding fragment thereof according
to the present invention.
[0493] In a further aspect, the present invention also provides the
neutralizing antibody, or the antigen binding fragment thereof,
according to the present invention, the nucleic acid according to
the present invention, the vector according to the present
invention, the cell according to the present invention, or the
pharmaceutical composition according to the present invention for
use according in treatment or prevention of ZIKV infection in
subjects diagnosed with Zika virus infection by using the antibody,
or the antigen binding fragment thereof, according to the present
invention, that binds to ZIKV NS1 protein, the complex according to
the present invention, the composition according to the present
invention, the kit of parts according to the present invention or
the method for ZIKV diagnosis according to the present
invention.
[0494] In a further aspect, the present invention also provides a
method of preventing and/or treating Zika virus infection, the
method comprising the following steps: [0495] (i) diagnosing Zika
virus infection in a subject by using the antibody, or the antigen
binding fragment thereof, according to the present invention, that
binds to ZIKV NS1 protein, the complex according to the present
invention, the composition according to the present invention, the
kit of parts according to the present invention or the method for
ZIKV diagnosis according to the present invention; and [0496] (ii)
administering to said subject the neutralizing antibody, or the
antigen binding fragment thereof, according to the present
invention, the nucleic acid according to the present invention, the
vector according to the present invention, the cell according to
the present invention, or the pharmaceutical composition according
to the present invention.
[0497] Preferably, in this method of preventing and/or treating
Zika virus infection, step (i) of diagnosing Zika virus infection
is performed as in-vitro diagnosis on an isolated sample (of a body
fluid), such as an isolated blood sample.
[0498] In a further aspect, the present invention also provides a
kit of parts comprising [0499] (i) the antibody, or the antigen
binding fragment thereof, according to the present invention, that
binds to ZIKV NS1 protein, the complex according to the present
invention, the composition according to the present invention, or
the kit of parts according to the present invention; and [0500]
(ii) the neutralizing antibody, or the antigen binding fragment
thereof, according to the present invention, the nucleic acid
according to the present invention, the vector according to the
present invention, the cell according to the present invention, or
the pharmaceutical composition according to the present
invention.
[0501] Such a kit of parts is particularly useful in a method as
described above. By use of such a method and/or such a kit, ZIKV
infection can be specifically diagnosed as well as prevented and/or
treated.
DESCRIPTION OF FIGURES
[0502] FIG. 1 shows the reactivity (ELISA) and ZIKV and DENV1
neutralizing activity of antibodies derived from four ZIKV immune
donors (ZKA, ZKB, ZKC and ZKD) to E protein of ZIKV and DENV1-4 and
to EDIII-domain of ZIKV E protein; NNB neutralizing, non-E-protein
binding antibodies.
[0503] FIG. 2 shows the reactivity (ELISA) of antibodies derived
from four ZIKV immune donors (ZKA, ZKB, ZKC and ZKD) to NS1 protein
of ZIKV, DENV1-4 and other flaviviruses. YFV--yellow-fever virus;
WVN--West-Nile virus; JEV--Japanese Encephalitis virus; and
TBEV--Tick-borne Encephalitis virus (nd, not determined).
[0504] FIG. 3 shows the binding of ZKA190, ZKA78 and ZKA64
antibodies to ZIKV and DENV1 E and to ZIKV EDIII proteins as
measured by ELISA.
[0505] FIG. 4 shows the binding of ZKA185 and ZKA190 antibodies to
ZIKV E, DENV1 VLP and to ZIKV EDIII proteins as measured by
ELISA.
[0506] FIG. 5 shows the binding of ZKA15, ZKA25 and ZKA35
antibodies to ZIKV and DENV1-4 NS1 proteins as measured by
ELISA.
[0507] FIG. 6 shows for Example 3 ZIKV NS1 protein antigenic site
mapping using cross-competition Octet-binding studies. (A-B)
Cross-competition matrix performed by Octet on 24 mAbs specific for
ZIKV NS1 (A) or cross-reactive to DENV NS1 (B). +, lack of binding
of the secondary Ab; +/-, partial loss of binding of the secondary
mAb; binding of the secondary mAb. Strikethrough cells, not tested.
(C) Map of the antigenic sites targeted by ZIKV NS1-specific mAbs
as defined using BLI (Octet) cross-competition.
[0508] FIG. 7 shows for Example 4 blockade of binding assay using
mAb ZKA35 as a probe to detect ZIKV NS1 in plasma from ZIKV-immune
(n=4), DENV-immune (n=5) and control donors (n=48) (1/10 dilution).
Plasma samples were tested for their capacity to bind NS1 (empty
dots) and to inhibit the binding of the biotinylated mAb ZKA35 to
NS1 (filled dots).
[0509] FIG. 8 shows for Example 5 the neutralizing activity of
ZKA190, ZKA64, ZKA64-LALA, ZKA230 and ZKA78 antibodies against ZIKV
(H/PF/2013 strain) and DENV1 on Vero cells as measured by
flow-cytometry (% of infected cells).
[0510] FIG. 9 shows for Example 5 the neutralizing activity of
ZKA190, ZKA64, ZKA185, ZKA230 and ZKA78 antibodies against ZIKV
(H/PF/2013 strain) on Vero cells as measured with a cell viability
readout (wst-1, Roche).
[0511] FIG. 10 shows for Example 6 the infection enhancing activity
(ADE, antibody-dependent enhancement) of ZKA190, ZKA64, ZKA64-LALA,
ZKA185, ZKA230 and ZKA78 antibodies for ZIKV (H/PF/2013 strain) on
non-permissive K562 cells as measured by flow-cytometry (% of
infected cells).
[0512] FIG. 11 shows for Example 6 that four ZIKV-immune plasma and
one DENV-immune plasma showed similar capacity to enhance ZIKV
infection of K562 cells (upper panel). This ADE effect was
completely blocked in all five immune plasma by the EDIII-specific
ZKA64-LALA antibody (lower panel).
[0513] FIG. 12 shows the amino acid alignment of the EDIII region
of 39 ZIKV strains from the Asian lineage since 2013 (including the
prototypic strain MR766 of the African lineage isolated in 1947).
The EDIII region amino acid sequence of ZIK V H/PF/2013 (SEQ ID NO:
402) is shown at the top of the alignment. Depicted EDIII sequences
differing from ZIKV H/PF/2013 are those of strains MR766 (SEQ ID
NO: 409), KU870645 USA 2 Feb. 2016 (SEQ ID NO: 410), KU729218
Brazil 2015 (SEQ ID NO: 411), KU681081 Thailand 9 Jul. 2014 (SEQ ID
NO: 412), KU926310 Brazil 29 Jan. 2016 (SEQ ID NO: 413), and
KU744693 China 6 Feb. 2016 (SEQ ID NO: 414).
[0514] FIG. 13 shows for Example 5 the neutralizing activity of
ZKA190 and ZKA190-LALA antibody against three strains of ZIKV
(H/PF/2013, MR766 and MRS_OPY_Martinique_PaRi_2015) on Vero cells
as measured by flow-cytometry (% of infected cells).
[0515] FIG. 14 shows for Example 7 NS1 blockade-of-binding analysis
of European residents. Shown are the BOB values for samples
collected in Italy and Switzerland. Plotted are the BOB values in
samples from ZIKV, primary and secondary DENV-, WNV-, and
CHIKV-infected individuals and a panel of samples from healthy
blood donors from Switzerland.
[0516] FIG. 15 shows for Example 8 neutralization of ZKA190 and C8
mAbs tested against a panel of four strains of ZIKV, as determined
by the percentage of infected Vero cells in the presence of
increasing amounts of the mAbs (A). Shown are also the IC50 values
(B) and statistics (C). Data are representative of at least two
independent experiments.
[0517] FIG. 16 shows for Example 9 the neutralization and
enhancement of ZIKV infection by antibody ZKA190. (A)
Neutralization of ZIKV PRVABC59 strain infection of hNPCs by
ZKA190, ZKA190-LALA and a control mAb as determined by plaque assay
on Vero cells (left panel) and indirect immunofluorescence of
infected hNPCs using fluorophore-labelled anti-E antibody (right
panel). (B) ADE of ZIKV infection of non-permissive K562 cells by
ZKA190 and ZKA190-LALA. (C) ADE induced in K562 cells when ZIKV is
pre-incubated with serial dilutions of plasma serum from different
ZIKV-positive patients (left panel). When ZKA190 LALA is added to
the ZIKV-serum complexes, ADE is inhibited (right panel). (D) ADE
induced in K562 cells when ZIKV is pre-incubated with serial
dilutions of a prM cross-reactive mAb (DV62) derived from a
DENV-immune donor. ZKA190-LALA inhibits ADE of ZIKV when complexed
with prM-reactive antibody DV62. (E) Effect on ADE induced by peak
enhancing dilution of a DENV2 plasma (left panel) or anti-prM DV62
mAb (right panel) by serial dilutions of indicated mAbs.
[0518] FIG. 17 shows for Example 10 the identification of ZKA190
epitope and analysis of its conservation in ZIKV strains. (A)
Overlay of [.sup.15N,.sup.1H]-HSQC spectra of .sup.15N-labeled ZIKV
EDIII in absence (black) or presence (red) of unlabelled ZKA190
Fab. Differences identify EDIII residues affected by antibody
binding. (B) NMR epitope mapping of ZKA190 Fab in complex with ZKV
EDIII. The chemical shift perturbation (CSP, y-axis) is plotted
against the EDIII residue number. Residues affected by antibody
binding are in red. (C) Residues in FG loop identified by NMR
epitope mapping is partially hidden in E protein mol A but largely
exposed in mols B and C. EDIII of E protein was coloured in blue.
Residues identified by NMR epitope mapping are coloured in magenta
except those in the FG loop are coloured in green. Adjacent E
proteins are shown as grey surface. (D) Level of amino acid residue
conservation in ZKA190 epitope as calculated by the analysis of
sequences from 217 ZIKV strains found in ZIKV Resources (NCBI)
databases as of Nov. 24 2016. (E) Open-book representation showing
charge complementarity between the epitope and paratope of the
docking result. Boundaries of the epitope and paratope are circled
in green. The borders between heavy and light chains of Fab and its
corresponding footprint on EDIII are shown as yellow dashed
lines.
[0519] FIG. 18 shows for Example 10 the ZKA190 epitope identified
by NMR and Docking. (A) Cartoon representation of the 12 lowest
energy NMR structures of ZIKV EDIII, with residues affected by
ZKA190 binding in red. Flexibility in the N-terminus of the
construct is apparent. (B) Model of the ZKA190:EDIII complex
derived by computational docking and molecular simulation validated
by NMR results. The NMR identified epitope on EDIII (grey) is in
red. The ZKA190 heavy and light chain are colored in dark and light
green, respectively. EDIII residues that affect or not antibody
binding when mutated are shown as orange and blue sticks,
respectively. (C) NMR identified ZKA190 epitope (red) is accessible
on the virus surface (white).
[0520] FIG. 19 shows for Example 10 the binding of wt or mutated
EDIII to ZKA190 IgG. SPR data and binding kinetics are shown. EDIII
mutants that affect (red highlights) or do not affect binding are
shown as indicated in the figure.
[0521] FIG. 20 shows for Example 11 the results of the confocal
microscopy experiments. ZIKV incubated with a concentration
exceeding 10'000-fold the IC50 value of either ZKA190 Fab or full
IgG were added to Vero cells. The ZIKV:antibody complex is detected
inside the cells (green) and co-localizes with endosomes (red,
yellow overlay). Endosomes and acidic organelles are marked by
Lysotracker red; Alexa-488 conjugated ZKA190 is in green. Nuclei
are stained with DAPI (blue).
[0522] FIG. 21 shows for Example 12 prophylactic and therapeutic
efficacy of ZKA190. (A) ZKA190 is strongly protective against ZIKV
infection when administered prophylactically to mice (A129 in (A)
and AG129 in (B)) challenged with a lethal dose of ZIKV strain
MP17451. Experiments used N=4-8 mice per group. Kaplan-Meier
survival curves are shown (A). Significance was determined by using
the Mantel-Cox log-rank test. Panel A, top left: ZKA190 at 5, 1 and
0.2 mg/kg versus Ctr mAb, P=0.0031; ZKA190 at 0.04 mg/kg versus Ctr
mAb, P=0.0116; ZKA190-LALA at 5, 1, 0.2 and 0.04 mg/kg versus Ctr
mAb, P=0.0031. Panel A, top right: Morbidity score of mice
monitored over a 14-15 day period (two different scoring methods
were used; see (Dowall, S. D., Graham, V. A., Rayner, E., Atkinson,
B., Hall, G., Watson, R. J., Bosworth, A., Bonney, L. C., Kitchen,
S., and Hewson, R. (2016). A Susceptible Mouse Model for Zika Virus
Infection. PLoS Negl Trop Dis 10, e0004658-13). Panel A, lower
panels: body weight of mice. Panels B: ZKA190 or ZKA190-LALA were
administered at 15 mg/kg at different time-points after ZIKV
infection. Panel B, top left: A Kaplan-Meier survival curve is
shown. Experiments used N=5 mice per group. Significance was
determined by using the Mantel-Cox log-rank test. ZKA190 and
ZKA190-LALA given either on day 1, 2, 3 or 4 versus Ctr., P=0.0016.
Panel B, top right: Morbidity score of mice monitored over a 14-day
according to (Dowall et al., 2016). Mice were monitored over a 14
day period for body weight loss (Panel B, lower panels). Control
antibody is MPE8 specific for RSV F protein (Corti, D., et al.
Cross-neutralization of four paramyxoviruses by a human monoclonal
antibody. Nature 501, 439-443 (2013)).
[0523] FIG. 22 shows for Example 12 the prophylactic efficacy of
the anti-ZIKV EDIII-specific mAb ZKA190 against ZIKV strains
MP1741. (A) Shown is the viremia measured as PFU/ml on day 5 in
blood of all animals. (B) Viral load was measured as genomic
copies/ml by qPCR on day 5 in blood of all animals and in blood and
indicated tissues when animals were culled at the end of the study
or when the humane end points were met. (C) Mice were monitored
over a 14 day period for body weight loss (D) Human serum IgG
concentration in day 5 blood samples. Significance was determined
compared to control antibody treatment by nonparametric unpaired
Mann-Whitney U test. *p<0.05; **p<0.01; ***p<0.001.
[0524] FIG. 23 shows for Example 12 the therapeutic efficacy of the
anti-ZIKV EDIII-specific mAb ZKA190. (A) Viral loads were measured
as PFUs on day 5 in blood of all animals. (B) Viral loads were
measured as genomic copies by qPCR on day 5 in blood of all animals
and in blood and indicated tissues when animals were culled at the
end of the study or when the human end points were met.
Significance was determined compared to control antibody treatment
by nonparametric unpaired Mann-Whitney U test. *p<0.05;
**p<0.01. (C) Human serum IgG concentration in day 5 blood
samples.
EXAMPLES
[0525] Exemplary embodiments of the present invention are provided
in the following examples. The following examples are presented
only by way of illustration and to assist one of ordinary skill in
using the invention. The examples are not intended in any way to
otherwise limit the scope of the invention.
Example 1: Isolation of ZIKV-Specific Antibodies and Production of
Monoclonal Antibodies
[0526] IgG+ memory B cells were isolated from cryopreserved
peripheral blood mononuclear cells (PBMCs) of four ZIKV-infected
donors (ZKA, ZKB, ZKC and ZKD) using CD22 microbeads (Miltenyi
Biotec), followed by depletion of cells carrying IgM, IgD and IgA
by cell sorting. Memory B cells from the ZIKV-infected donors were
then immortalized with EBV (Epstein Barr Virus) and CpG (CpG
oligodeoxynucleotide 2006) in multiple replicate wells as
previously described (Traggiai, E. et al., Nat. Med. 10, 871-875,
2004) and culture supernatants were then tested in a primary
screening using in parallel a 384-well based micro-neutralization
assay and a binding assay (ELISA) to test their binding to ZIKV NS1
protein or to ZIKV E protein. Results of the binding assay are
shown in FIG. 1 (binding to ZIKV E protein) and FIG. 2 (binding to
ZIKV NS1 protein).
[0527] Neutralization assays were undertaken on Vero cells. In a
384-well plate, ZIKV H/PF/2013 that resulted in an infection rate
(m.o.i, multiplicity of infection) of 0.35 was incubated with
superntanants for 1 h at 37% (5% CO2) before the addition to
pre-seeded 5'000 Vero cells. These were incubated for a further 5
days, after which supernatant was removed and WST-1 reagent (Roche)
was added. Positive cultures were collected and expanded. From
positive cultures the VH and VL sequences were retrieved by RT-PCR.
Antibodies were cloned into human IgG1 and Ig kappa or Ig lambda
expression vectors (kindly provided by Michel Nussenzweig,
Rockefeller University, New York, US) essentially as described
(Tiller T, Meffre E, Yurasov S, Tsuiji M, Nussenzweig M C,
Wardemann H (2008) Efficient generation of monoclonal antibodies
from single human B cells by single cell RT-PCR and expression
vector cloning. J Immunol Methods 329: 112-124). Monoclonal
antibodies were produced from EBV-immortalized B cells or by
transient transfection of 293 Freestyle cells (Invitrogen).
Supernatants from B cells or transfected cells were collected and
IgG were affinity purified by Protein A or Protein G chromatography
(GE Healthcare) and desalted against PBS.
[0528] FIG. 1 provides an overview over selected ZIKV neutralizing
antibodies (cf. Tables 1 and 2 for the amino acid sequences of
their CDRs and heavy/light chain variable regions). The last two
columns of FIG. 1 provide the neutralization activities (IC.sub.50)
of ZIKV and DENV1 (if tested). The other columns provide binding
activities (EC.sub.50) of the antibodies to ZIKV E protein (ZIKV
E), DENV1 E protein (DENV1 E), DENV2 E protein (DENV2 E), DENV3 E
protein (DENV3 E), DENV4 E protein (DENV4 E), DENV1 virus-like
particle (DENV1 VLP), DENV2 virus-like particle (DENV2 VLP), DENV3
virus-like particle (DENV3 VLP), DENV4 virus-like particle (DENV4
VLP), and to EDIII-domain of ZIKV E protein (DIII ZKA).
[0529] Additional antibodies were isolated for their ability to
bind to ZIKV NS1 protein (cf. FIG. 2). Positive cultures were
collected and expanded. From positive cultures the VH and VL
sequences were retrieved by RT-PCR. Antibodies were cloned into
human IgG1 and Ig kappa or Ig lambda expression vectors (kindly
provided by Michel Nussenzweig, Rockefeller University, New York,
US) essentially as described (Tiller T, Meffre E, Yurasov S, Tsuiji
M, Nussenzweig M C, Wardemann H (2008) Efficient generation of
monoclonal antibodies from single human B cells by single cell
RT-PCR and expression vector cloning. J Immunol Methods 329:
112-124). Monoclonal antibodies were produced from EBV-immortalized
B cells or by transient transfection of 293 Freestyle cells
(Invitrogen). Supernatants from B cells or transfected cells were
collected and IgG were affinity purified by Protein A or Protein G
chromatography (GE Healthcare) and desalted against PBS.
[0530] FIG. 2 provides an overview over selected ZIKV NS1-protein
binding antibodies (cf. Tables 4 and 5 for the amino acid sequences
of their CDRs and heavy/light chain variable regions). Namely, FIG.
2 provides binding activities (EC.sub.50) of the antibodies to ZIKV
NS1 protein (ZIKV NS1), DENV1 NS1 protein (DENV1 NS1), DENV2 NS1
protein (DENV2 NS1), DENV3 NS1 protein (DENV3 NS1), DENV4 NS1
protein (DENV4 NS1), yellow-fever virus NS1 protein (YFV NS1),
West-Nile virus NS1 protein (WNV NS1), Japanese-Encephalitis virus
NS1 protein (JEV NS1), and to Tick-borne Encephalitis virus NS1
protein (TBEV NS1).
Example 2: Characterization of Antibodies ZKA190, ZKA185, ZKA230,
ZKA64 and ZKA78
[0531] In Example 1, a large number of ZIKV-neutralizing antibodies
were identified and characterized for their specificity to ZIKV, in
particular ZIKV E protein and ZIKV EDIII as well as for their
cross-reactivity towards DENV. Antibodies ZKA190 (SEQ ID NOs:
1-18), ZKA185 (SEQ ID NOs: 19-36), ZKA230 (SEQ ID NOs: 37-54),
ZKA64 (SEQ ID NOs: 73-90) and ZKA 78 (SEQ ID NOs: 55-72) described
in Example 1 were then selected and further tested against ZIKV E
protein ("ZIKV"), ZIKV EDIII ("DIIIZI") and also tested against the
E protein of dengue virus (DENV, serotype number 1) by ELISA. To
this end, a standard ELISA was used. Briefly, ELISA plates were
coated with ZIKV E protein at 1 or 3 .mu.g/ml, blocked with 10% FCS
in PBS, incubated with sera or human antibodies and washed. Bound
antibodies were detected by incubation with AP-conjugated goat
anti-human IgG (Southern Biotech). Plates were then washed,
substrate (p-NPP, Sigma) was added and plates were read at 405 nm.
The relative affinities of monoclonal antibody binding were
determined by measuring the concentration of antibody (EC50)
required to achieve 50% maximal binding at saturation.
[0532] Results are shown in FIGS. 3 and 4. Of note, ZKA64 and
ZKA190 bound to ZIKV E and ZIKV EDIII ("DIII ZI") with low EC50
values, thereby indicating that ZKA64 and ZKA190 are binding to
domain III of ZIKV E protein (EDIII). ZKA78 bound to ZIKV E, but
not to ZIKV EDIII, indicating that ZKA78 is binding to ZIKV E, but
not targeting the EDIII region. Despite their considerable ZIKV
neutralizing activity (cf. FIG. 1), antibodies ZKA185 and ZKA230
did not show any detectable binding to ZIKV E and ZIKV EDIII (FIG.
4). Accordingly, ZKA185 and ZKA230 were referred to as
"neutralizing-non-E-binding" (NNB) antibodies. Those NNB antibodies
are assumed to recognize quaternary epitopes that are displayed on
the ZIKV infectious virions but not on soluble proteins.
[0533] Moreover, none of ZKA190, ZKA185, ZKA230, and ZKA64 showed
any detectable binding to DENV E proteins (FIG. 1, DENV1-4
serotypes, and FIGS. 3 and 4), indicating that ZKA190, ZKA185,
ZKA230, and ZKA64 are specific for ZIKV and not cross-reactive to
dengue virus. ZKA78, in contrast, which is assumed to bind to ZIKV
EDI/II, but not to ZIKV EDIII (cf. FIG. 3), bound to DENV E
proteins (FIGS. 1 and 3), indicating that ZKA78 is a cross-reactive
antibody binding to both, ZIKV and DENV.
[0534] To further confirm those results, the ZIKV E protein binding
antibodies ZKA190, ZKA64 and ZKA78 were additionally tested against
E protein of dengue virus (DENV, serotypes number 1-4). ZKA64 and
ZKA190 did not bind to DENV1-4 E protein, thereby confirming that
ZKA64 and ZKA190 are specific for ZIKV. ZKA78, in contrast, bound
to DENV1-4 E, confirming that ZKA78 is a cross-reactive antibody
binding to the E protein of both ZIKV and DENV (cf. FIG. 1).
Example 3: Characterization of ZIKV NS1-Specific Antibodies for
Serological Diagnosis
[0535] In Example 1, a large number of NS1-reactive antibodies were
identified and then characterized for their specificity to ZIKV NS1
and cross-reactivity towards other flavivirus NS1 proteins (FIG.
2). Antibodies ZKA15 (SEQ ID NOs: 91-108), ZKA25 (SEQ ID NOs:
109-126) and ZKA35 (SEQ ID NOs: 127-144) were then further
characterized for binding to ZIKV NS1 and DENV1 NS1, DENV2 NS1,
DENV3 NS1 and DENV4 NS1. To this end, a standard ELISA was used.
Briefly, ELISA plates were coated with ZIKV NS1 protein at 1
.mu.g/ml, blocked with 10% FCS in PBS, incubated with sera or human
antibodies and washed. Bound antibodies were detected by incubation
with AP-conjugated goat anti-human IgG (Southern Biotech). Plates
were then washed, substrate (p-NPP, Sigma) was added and plates
were read at 405 nm. The relative affinities of monoclonal antibody
binding were determined by measuring the concentration of antibody
(EC50) required to achieve 50% maximal binding at saturation.
[0536] Results are shown in FIG. 5. All three antibodies (ZKA15,
ZKA25 and ZKA35) bound with high affinity to ZIKV NS1 but not to
the DENV1-4 NS1 antigens (FIG. 5).
[0537] To investigate the binding of the antibodies to ZIKV NS1
further, bio-layer interferometry competition assays were used. A
cross-competition matrix was generated using biolayer
interferometry (BLI; Octet) on 13 antibodies specific for ZIKV NS1
(i.e. not cross-reactive with DENV NS1), namely antibodies ZKA24,
ZKA15, ZKA32, ZKA19, ZKA50, ZKA37, ZKA46, ZKA10, ZKA48, ZKA35,
ZKA25, ZKA44, and ZKA30 (cf. FIG. 6A). As can be retrieved from
FIG. 2 none of those 13 antibodies showed detectable binding to
DENV NS1.
[0538] Competition assays and antigenic sites determination were
determined at 37.degree. C. with a Octet RED96 system, ForteBio.
The ZIKV-NS1 protein diluted to 2.5 .mu.g/ml in PBS was immobilized
for 7-9 minutes on the surface of an APS coated sensor-chip. Coated
biosensors were placed in wells containing blocking buffer (0.1%
BSA in PBS) for 6 minutes to block free Biosensor binding sites.
Coated-Biosensors were then incubated for 8 minutes with a set of
single purified mAbs specific for ZIKV-NS1 diluted in blocking
buffer at 10 .mu.g/ml. After binding of the first set of mAbs (step
1), Biosensors were moved to wells containing different mAbs for 8
minutes (step 2). Association of the second mAb resulted in
recognition of a different antigenic site compared to the first mAb
(e.g. non-competition). Competition or partial competition were
determined in step 2 when no association or low association was
detected, respectively. A cross-competition matrix was created by
multiple runs of competitions in order to predict antigenic site
mapping on ZIKV NS1.
[0539] Results are shown in FIGS. 6A and 6C. Firstly, all of the
ZIKV NS1-specific antibodies tested were binding to antigenic
site(s) S1 and/or S2 (FIG. 6A). However, some of the antibodies did
not compete with others. For example, ZKA15 did not compete for
binding with ZKA25 and ZKA35 and vice versa (FIG. 6A). Accordingly,
antibody ZKA15 was assigned to the antigenic site S1, while
antibodies ZKA25 and ZKA35 were assigned to the antigenic site S2
(FIG. 6C). In summary, based on the antibodies used, antigenic
sites (S1 and S2) on ZIKV NS1 were identified (FIG. 6C).
[0540] Additionally, binding of 10 antibodies cross-reacting to
ZIKV NS1 protein and to DENV NS protein (namely, ZKA18, ZKA29,
ZKA39, ZKA53, ZKA54, ZKB19, ZKB23, ZKC29, ZKC33, and ZKC34; FIG.
6B) to antigenic sites S1 and/or S2 on ZIKV NS1 was investigated.
As can be retrieved from FIG. 2 all of those 10 antibodies showed
binding to DENV NS1. Those 10 cross-reactive antibodies were tested
in a cross-competition assay as described above (for the ZIKV
NS1-specific antibodies) against ZIKV NS1 S1-specific antibody
ZKA15 and against ZIKV NS1 S2-specific antibody ZKA35.
[0541] Results are shown in FIG. 6B. Interestingly, none of the ten
cross-reactive antibodies tested competed with ZKA 15 and/or ZKA35
for binding to antigenic site(s) S1 and/or S2 on ZIKV NS (FIG. 6B).
These results show that ZKA15 and ZKA35 antigenic site is not
targeted by NS cross-reactive antibodies. Thus, NS1 antigenic sites
S1 and S2 were targeted by ZIKV-specific, but not by cross-reactive
antibodies.
Example 4: Use of ZIKV NS1-Specific Antibodies in Diagnosis of ZIKV
Infection
[0542] In the present Example, the usefulness of the ZIKV
NS1-specific antibodies of the present invention in diagnosis of
ZIKV infection was investigated. More specifically, the use of ZIKV
NS1-specific antibodies of the present invention to specifically
detect the presence or absence of antibodies elicited against ZIKV
NS1 in plasma samples of ZIKV- or DENV-infected donors was
determined.
[0543] To this end, a "blockade of binding" assay was used. In
particular, the ability of ZIKV NS1-reactive plasma antibodies to
inhibit the binding of the biotinylated antibody ZKA35 to ZIKV NS1
was measured. To this end, ZIKV NS1-specific antibody ZKA35 was
biotinylated using the EZ-Link NHS-PEO solid phase biotinylation
kit (Pierce). Labelled ZKA35 was tested for binding to ZIKV NS1 to
determine the optimal concentration of ZKA35 to achieve 70% maximal
binding. Plasma samples from ZIKV- (n=4), DENV-immune (n=5) donors
and control (n=48) plasma (1/10 dilution) were added to ELISA
plates coated with ZIKV NS1. After 1 h, biotinylated anti-ZIKV NS1
antibody ZKA35 was added at the concentration achieving 70% maximal
binding and the mixture was incubated at room temperature for 15
minutes. Plates were washed, substrate (p-NPP, Sigma) was added and
plates were read at 405 nm. The percentage of inhibition was
calculated as follow: (1-[(OD sample-OD neg ctr)/(OD pos ctr-OD neg
ctr)]).times.100.
[0544] Results are shown in FIG. 7. Of note, antibody ZKA35 binding
to the antigenic site S2 on NS was inhibited only by plasma samples
from ZIKV-immune donors, but not DENV-immune donors, and its
binding was also not inhibited by 48 control plasma samples (FIG.
7). Accordingly, this assay may be used as to specifically detect
clinical and sub-clinical ZIKV infections at the population
level.
Example 5: The Antibodies According to the Present Invention
Potently Neutralize ZIKV Infection
[0545] The isolated antibodies ZKA190, ZKA185, ZKA230, ZKA64 and
ZKA78 were tested for their ability to neutralize ZIKV and DENV1
infection in vitro.
[0546] Neutralization of DENV and ZIKV infection by antibodies was
measured using a micro-neutralization flow cytometry-based assay.
Different dilutions of antibodies were mixed with ZIKV (MOI of
0.35) or attenuated DENV1 (all at MOI of 0.04) for 1 hour at
37.degree. C. and added to 5000 Vero cells/well in 96-well
flat-bottom plates. After four days for ZIKV and five days for
DENV, the cells were fixed with 2% formaldehyde, permeabilized in
PBS 1% FCS 0.5% saponin, and stained with the mouse mAb 4G2. The
cells were incubated with a goat anti-mouse IgG conjugated to Alexa
Fluor488 (Jackson Immuno-Research, 115485164) and analyzed by flow
cytometry. In other cases the ZIKV neutralization data are also
determined measuring cell viability using the WST-1 reagent
(Roche). The neutralization titer (50% inhibitory concentration
[IC50]) was expressed as the antibody concentration that reduced
the infection by 50% compared to cell-only control wells.
[0547] Results are shown in FIGS. 8, 9 and 13. The EDIII-specific
mAbs ZKA64 and ZKA190 and the NNB mAb ZKA230 were highly potent in
ZIKV neutralization (strain H/PF/2013), with IC50 values of 93, 9
and 10 ng/ml, respectively (FIG. 8, upper panel). In contrast, the
cross-reactive antibody ZKA78 only partially neutralized ZIKV
infectivity and cross-neutralized DENV1 infectivity (FIG. 8, lower
panels). Similar data were obtained by measuring the ZIKV-induced
cytopathic effect as measured with the WST-1 reagent (FIG. 9). In
this second assay, NNB antibody ZKA185 was also included in the
panel of tested antibodies and showed an IC50 similar to the most
potent antibodies ZKA190 (EDIII-specific) and ZKA230 (NNB).
[0548] It is important to note that the ultra-potent ZKA64 and
ZKA190 antibodies in addition to their ability to neutralize the
ZIKV H/PH/2013 strain (present example), also bound to the E
protein and EDIII derived from the ZIKV strains MR766 and SPH2015,
respectively (FIG. 1 and FIG. 3). ZKA190 and ZKA190-LALA was also
confirmed to effectively neutralize two additional ZIKV strains
(MR766 and MRS_OPY_Martinique_PaRi_2015) (FIG. 13). Taken together
the results indicate that the ultra-potent ZKA64 and ZKA190
antibodies cross-react with multiple strains of ZIKV belonging to
different genotypes and origins (East African and Asian from
Uganda, French Polynesia, Martinique and Brazil).
Example 6: The "LALA" Mutation Inhibits Antibody-Dependent
Enhancement of ZIKV Infection by Serum Antibodies
[0549] Neutralizing antibodies were also tested for their ability
to enhance the infection of ZIKV in the non-permissive K562 cells
(antibody-dependent enhancement assay, ADE assay). ADE was measured
by a flow based assay using K562 cells. Antibodies and ZIKV
H/PF/2013 (MOI 0.175) were mixed for 1 hour at 37.degree. C. and
added to 5000 K562 cells/well. After four days, cells were fixed,
permeabilized, and stained with m4G2. The number of infected cells
was determined by flow cytometry.
[0550] Results are shown in FIG. 10. All antibodies enhanced
infection of ZIKV in the non-permissive K562 cells at a broad range
of concentrations, including those that fully neutralized ZIKV
infection on Vero cells (FIG. 10). Of note, while EDIII-specific
antibodies ZKA64 and ZKA190 fully neutralized ZIKV infections of
K562 cells above 1 .mu.g/ml, the NNB antibody ZKA230 failed to do
so, a result that might be due to the different mechanisms of
neutralization of free viruses versus
Fc-gamma-receptor-internalized viruses. In contrast, the
cross-reactive ZKA78 that only partially neutralized ZIKV
infectivity, effectively enhanced ZIKV infection of K562 cells.
These results show that cross-reactive antibodies elicited by
either ZIKV or DENV infection can mediate heterologous ADE.
[0551] In view thereof it was investigated whether ADE could be
also induced by immune sera and whether this could be blocked by
neutralizing antibodies delivered as a "LALA variant". To obtain
the LALA variant, each of the heavy chains was mutated at amino
acids 4 and 5 of CH2 domain by substituting an alanine in place of
the natural leucine using site-directed mutagenesis. As described
above, LALA variants (of human IgG1 antibodies) do not bind to
Fc-gamma-receptors and complement.
[0552] To investigate the effect of ZKA64-LALA antibody in ZIKV
ADE, an inhibition of ADE assay was used. Since ADE of ZIKV is
observed using ZIKV- or DENV-immune plasma, ZIKV (MOI 0.175) was
mixed with plasma from primary ZIKV- or DENV-infected donors for 30
minutes at 37.degree. C. ZKA64-LALA antibody was added at 50
.mu.g/ml, mixed with 5000 K562 cells/well and incubated for three
days. Cells were then stained with 4G2 and analyzed by flow
cytometry.
[0553] Results are shown in FIG. 11. In a homologous setting, four
ZIKV-immune plasma collected from convalescent patients and one
DENV-immune plasma showed similar capacity to enhance ZIKV
infection of K562 cells (FIG. 11, upper panel), and this ADE effect
was completely blocked by the EDIII-specific ZKA64-LALA antibody
(FIG. 11, lower panel).
[0554] Of note, the ADE effect of ZIKV- and DENV-immune plasma was
completely blocked by the EDIII-specific ZKA64-LALA antibody. The
ADE blocking ability of a single EDIII-specific LALA antibody could
be related not only to its capacity to out-compete serum enhancing
antibodies but also to neutralize virus once internalized into
endosomes.
[0555] These results indicate that a potently neutralizing
antibody, such as ZKA190, ZKA230, ZKA185 or ZKA64, developed in the
"LALA" form, have a strong potential to be used in prophylactic or
therapeutic settings to prevent congenital ZIKV infection, e.g. in
pregnant women and/or in people living in high risk areas. The use
of the LALA form avoids the risk of ZIKV ADE and, as shown above,
could also block ADE of pre-existing cross-reactive antibodies,
such as in the case of patients already immune to DENV.
Example 7: Analysis of Samples from European Residents Using ZIKV
NS1-Specific Antibodies for Diagnosis of ZIKV Infection
[0556] The present Example is based on the blockade of binding
assay described in Example 4. To further assess the specificity of
the ZIKV NS1 BOB assay, a large set of samples obtained from
patients infected with DENV, WNV or Chikungunya virus (CHIKV) was
tested.
[0557] To this end, a "blockade of binding" assay was used.
Polystyrene plates were coated overnight with 1 .mu.g/ml of ZIKV
NS1 and blocked for 1 hour with PBS containing 1% BSA. Plasma or
serum (1:10 dilution) were added to NS1-coated ELISA plates.
Thereafter, e.g. after 1 hour, an equal volume of biotinylated
anti-NS1 ZKA35 was added, and the mixture was incubated, e.g. at
room temperature for 15 minutes. Plates were washed and
alkaline-phosphatase-conjugated streptavidin was added, e.g. for 30
minutes. Plates were washed again and the substrate was added. The
percentage of inhibition was calculated as follow: (1-[(OD
sample-OD neg ctr)/(OD pos ctr-OD neg ctr)]).times.100.
[0558] Results are shown in FIG. 14. Thirty-one of 32 samples
(96.9%) from WNV patients collected more than 10 days after symptom
onset scored negative. Of note, the only positive was obtained from
a sample collected in 2016. Two of 27 samples from DENV patients
collected more than 10 days after symptom onset scored positive,
and the two positive samples were derived from secondary DENV
infections. In addition, none of the samples derived from
chikungunya patients or YFV-vaccinees scored positive. A large
number of plasma samples from Swiss blood donors (n=116) collected
between 2010 and 2016 was also tested. None of those samples scored
positive. The results obtained confirmed and strengthened the high
sensitivity and specificity of the NS1 BOB ELISA assay.
Example 8: An Antibody According to the Present Invention
Neutralizes ZIKV More Potently than Prior Art Antibody EDE1 mAb
C8
[0559] To compare the neutralizing antibodies according to the
present invention with highly neutralizing anti-ZIKV antibodies of
the prior art, neutralization performance of ZKA190 was compared to
that of prior art highly neutralizing mAb EDE1 C8 (Barba-Spaeth G,
Dejnirattisai W, Rouvinski A, Vaney M C, Medits I, Sharma A,
Simon-Loriere E, Sakuntabhai A, Cao-Lormeau V M, Haouz A, England
P, Stiasny K, Mongkolsapaya J, Heinz F X, Screaton G R, Rey F A.
Structural basis of potent Zika-dengue virus antibody
cross-neutralization. Nature. 2016 Aug. 4; 536(7614):48-53).
Neutralization of both antibodies was tested against a panel of
four distinct ZIKV strains (H/PF/2013; MR766, MRS-OPY and
PV10552).
[0560] Briefly, neutralization of ZIKV infection by mAbs was
measured using a micro-neutralization flow cytometry-based assay.
Different dilutions of mAbs were mixed with ZIKV (MOI of 0.35) for
1 hour at 37.degree. C. and added to 5000 Vero cells/well in
96-well flat-bottom plates. After four days for ZIKV, the cells
were fixed with 2% formaldehyde, permeabilized in PBS containing 1%
fetal calf serum (Hyclone) and 0.5% saponin, and stained with the
mouse mAb 4G2. The cells were incubated with a goat anti-mouse IgG
conjugated to Alexa Fluor488 (Jackson Immuno-Research, 115485164)
and analyzed by flow cytometry. The neutralization titer (50%
inhibitory concentration [IC50]) is expressed as the antibody
concentration that reduced the infection by 50% compared to
virus-only control wells.
[0561] Results are shown in FIG. 15. ZKA190 mAb potently
neutralized African, Asian and American strains with an IC50
ranging from 0.6 to 8 ng/ml. In comparison, prior art antibody C8
was about 24-fold less potent.
Example 9: Further Characterization of Antibody ZKA190
[0562] The potency of antibody ZKA190 was further investigated in
vitro and in vivo. To this end, the mAb was synthesized in IgG1
wild-type (wt) format and in an IgG1 Fc-LALA format. Briefly, the
VH and VL sequences were cloned into human Ig.gamma.1, Ig.kappa.
and IgA expression vectors (kindly provided by Michel Nussenzweig,
Rockefeller University, New York, N.Y., USA), essentially as
described (Tiller T, Meffre E, Yurasov S, Tsuiji M, Nussenzweig M
C, Wardemann H: Efficient generation of monoclonal antibodies from
single human B cells by single cell RT-PCR and expression vector
cloning. J Immunol Methods 2008, 329:112-124). Recombinant mAbs
were produced by transient transfection of EXPI293 cells
(Invitrogen), purified by Protein A chromatography (GE Healthcare)
and desalted against PBS. To obtain the LALA variant, each of the
heavy chains was mutated at amino acids 4 and 5 of CH2 domain by
substituting an alanine in place of the natural leucine using
site-directed mutagenesis. As described above, LALA variants (of
human IgG1 antibodies) do not bind to Fc-gamma-receptors and
complement.
[0563] As shown in FIG. 15A and described in Example 8, ZKA190 was
tested against a panel of four ZIKV strains. ZKA190 mAb potently
neutralized African, Asian and American strains with an IC50
ranging from 0.004 to 0.05 nM (FIG. 15A; 0.6 to 8 ng/ml).
[0564] Since ZIKV has been shown to infect human neural progenitor
cells (hNPC) leading to heightened cell toxicity, dysregulation of
cell-cycle and reduced cell growth, ZKA190 and ZKA190-LALA were
tested in hNPCs. To this end, adult male fibroblasts obtained from
the Movement Disorders Bio-Bank (Neurogenetics Unit of the
Neurological Institute `Carlo Besta`, Milan) were reprogrammed
using the CytoTune-iPS 2.0 Sendai kit (Life Technologies). hiPSCs
were maintained in feeder-free conditions in mTeSR1 (Stem Cell
Technologies). To generate embryoid bodies (EBs), dissociated
hiPSCs were plated into low adhesion plates in mTeSR1 supplemented
with N2 (0.5.times.) (ThermoFisher Scientific), human Noggin (0.5
mg/ml, R&D System), SB431542 (5 M, Sigma), Y27632 (10 .mu.M,
Miltenyi Biotec) and penicillin/streptomycin (1%, Sigma) (as
described in Marchetto M C N, Carromeu C, Acab A, Yu D, Yeo G W, Mu
Y, Chen G, Gage F H, Muotri A R: A model for neural development and
treatment of Rett syndrome using human induced pluripotent stem
cells. Cell 2010, 143:527-539). To obtain rosettes, EBs were plated
after 10 days onto matrigel-coated plates (1:100, matrigel growth
factor reduced, Corning) in DMEM/F12 (Sigma) with N2 (1:100),
non-essential amino acids (1%, ThermoFisher Scientific) and
penicillin/streptomycin. After 10 days, cells were passaged with
Accutase (Sigma) and seeded onto matrigel coated-flasks in NPC
media containing DMEM/F12, N2 (0.25%), B27 (0.5%, ThermoFisher
Scientific), penicillin/streptomycinand FGF2 (20 ng/ml,
ThermoFisher Scientific). hNPCs (3.times.104) were plated on
coverslips in 24-well plates 3 days prior to infection with
PRVABC59 strain. Virus stock was incubated with the mAbs 1 h prior
to addition to hNPCs to obtain an MOI of 0.5. After 4 h of virus
adsorption, culture supernatant was removed and fresh medium
containing the mAbs was re-added. Supernatant was collected 96 h
post-infection to measure virus titers by plaque assay on Vero
cells. Cells were fixed in 4% paraformaldehyde (PFA, Sigma)
solution in phosphate-buffered saline (PBS, Euroclone) for 30 min
for indirect immunofluorescence. Fixed cells were permeabilized for
30 minutes (min) in blocking solution, containing 0.2% Triton X-100
(Sigma) and 10% donkey serum (Sigma), and incubated overnight at
4.degree. C. with the primary antibodies in blocking solution. The
following antibody was used for detection: anti-envelope (1:200,
Millipore, MAB10216). Then, cells were washed with PBS and
incubated for 1 h with Hoechst and anti-mouse Alexa Fluor-488
secondary antibodies (1:1,000 in blocking solution, ThermoFisher
Scientific). After PBS washes, cells were washed again and mounted.
Results are shown in FIG. 16A. Both, ZKA190 and ZKA190-LALA, fully
abolished infection and replication of ZIKV in hNPCs.
[0565] Next, the ability of ZKA190 and ZKA190-LALA to cause ADE was
tested in the K562 cell line as described in Example 6. Briefly,
ADE was measured by a flow based assay using K562 cells. Briefly,
for ZKA190, ZKA190 and ZIKV H/PF/2013 (MOI 0.175) were mixed for 1
hour at 37.degree. C. and added to 5000 K562 cells/well. After four
days, cells were fixed, permeabilized, and stained with mAb m4G2.
The number of infected cells was determined by flow cytometry. For
ZKA190-LALA, ZIKV (MOI 0.175) was mixed with plasma from primary
ZIKV-infected donors for 30 minutes at 37.degree. C. ZKA190-LALA
was added at 50 .mu.g/ml, mixed with 5000 K562 cells/well and
incubated for three days. Cells were then stained with 4G2 and
analyzed by flow cytometry. Results are shown in FIG. 16B. ZKA190
supports ADE from 0.0001 to 1 nM; as expected, ZKA190-LALA did not
show any ADE activity. The ability of ZKA190-LALA to inhibit ADE
induced by plasma from four ZIKV-immune donors in K562 cells was
also tested. Results are shown in FIG. 16C. It was found that
ZKA190-LALA completely inhibited the ADE induced by plasma
antibodies (FIG. 16C).
[0566] Anti-prM antibodies form part of the predominant antibodies
elicited during the human immune response against flaviviruses and
have been shown to enhance virus infection in vitro (Dejnirattisai,
W., Jumnainsong, A., Onsirisakul, N., Fitton, P., Vasanawathana,
S., Limpitikul, W., Puttikhunt, C., Edwards, C., Duangchinda, T.,
Supasa, S., et al. (2010). Cross-reacting antibodies enhance dengue
virus infection in humans. Science 328, 745-748). K562 cells were
pre-incubated with serial dilutions of prM cross-reactive antibody
DV62 (Beltramello, M., Williams, K. L., Simmons, C. P., Macagno,
A., Simonelli, L., Quyen, N. T. H., Sukupolvi-Petty, S.,
Navarro-Sanchez, E., Young, P. R., de Silva, A. M., et al. (2010).
The human immune response to Dengue virus is dominated by highly
cross-reactive antibodies endowed with neutralizing and enhancing
activity. Cell Host Microbe 8, 271-283) derived from a DENV immune
donor. Results are shown in FIG. 16D. DV62 cross-reacted with ZIKV
prM protein and caused ADE at a broad range of concentrations (FIG.
16D). ZKA190-LALA can fully block anti-prM DV62 mAb-induced ADE of
immature or partially immature ZIKV particles (FIG. 16D).
[0567] Finally, the ability of different concentrations of ZKA190,
ZKA190-LALA and ZKA190 Fab to cause or block ADE of ZIKV in the
presence of enhancing concentrations of human anti-DENV2 plasma or
DV62 was tested. Results are shown in FIG. 16E. ZKA190 at low
concentrations increased the prM DV62-mediated ADE of ZIKV
infection, consistent with its ability to promote the entry of both
immature and mature virions, while at concentrations above 1.3 nM
(i.e., 200 ng/ml) ZKA190 blocked ADE induced by both DENV plasma
and mAb DV62. ZKA190-LALA, as well as its Fab fragment, reduced ADE
at concentrations above 0.06 nM, indicating that both inhibited
virus infection at a post-attachment step, such as fusion.
Example 10: ZKA190 Binds to a Conserved and Highly Accessible
Region of EDIII
[0568] To determine the ZKA190 epitope at the residue level,
solution NMR spectroscopy was used as described in Bardelli, M.,
Livoti, E., Simonelli, L., Pedotti, M., Moraes, A., Valente, A. P.,
and Varani, L. (2015). Epitope mapping by solution NMR
spectroscopy. J. Mol. Recognit. 28, 393-400; Simonelli, L.,
Beltramello, M., Yudina, Z., Macagno, A., Calzolai, L., and Varani,
L. (2010). Rapid structural characterization of human
antibody-antigen complexes through experimentally validated
computational docking. J Mol Biol 396, 1491-1507; and Simonelli,
L., Pedotti, M., Beltramello, M., Livoti, E., Calzolai, L.,
Sallusto, F., Lanzavecchia, A., and Varani, L. (2013). Rational
Engineering of a Human Anti-Dengue Antibody through Experimentally
Validated Computational Docking. PLoS ONE 8, e55561.
[0569] Briefly, spectra were recorded on a Bruker Avance 700 MHz
NMR spectrometer at 300 K. For assignments of backbone resonances
standard triple resonance experiments (HNCO, HN(CA)CO, HN(CO)CACB,
HNCACB were used, while sidechains were annotated using HCCH-TOCSY
and HBHA(CO)NH experiments. All NMR experiments were processed
using Topspin 2.1 (Bruker Biospin) and analysed with CARA. NOESY
cross peaks were automatically assigned using the CYANA "noeassign"
macro based on the manually assigned chemical shifts.
Upper-distance restraints used for the structure calculations in
CYANA using the standard simulated annealing protocol were derived
from 70 ms .sup.15N- and .sup.13C-resolved NOESY spectra. Backbone
dynamics of ZIKV EDIII were derived from .sup.15N relaxation
measurements recorded on 600 and 700 MHz spectrometers.
Proton-detected versions of the CPMG (R2), inversion-recovery (R1)
and 1.sup.5N {1H}-steady-state NOE were utilized. Delay settings
for the T2 series were in the range of 0 to 0.25 sec and for the T1
series between 0.02 to 2 sec. The .sup.15N {.sup.1H}-NOE experiment
used a relaxation delay of 5 s. The R1 and R2 relaxation rates were
derived from least-squares fits of corresponding exponential
functions to the measured data using home-written scripts. The
relaxation data were analyzed in a model-free approach using the
software package DYNAMICS. The program ROTDIF was used to calculate
the overall correlation time from the relaxation data (8.5 ns). NMR
epitope mapping was performed as previously described (Bardelli et
al., 2015; Simonelli et al., 2010; 2013). Briefly, overlay
of.sup.15NHSQC spectra of labelled EDIII free or bound to ZKA190
Fab allowed identification of EDIII residues whose NMR signal
changed upon complex formation, indicating that they were affected
by ZKA190 binding. Changes were identified by manual inspection and
by the Chemical Shift Perturbation (CSP),
CSP=((.DELTA..delta..sub.H).sup.2+(.DELTA..delta..sub.N/10).sup.2).sup.1/-
2. NMR samples were typically 800 .mu.M of [.sup.15N,
.sup.13C]-labeled EDIII in 20 mM sodium phosphate, 50 mM NaCl, pH
6.0. Perdeuterated (nominally 70%) .sup.2H,.sup.15N EDIII samples
were used for NMR epitope mapping with a EDIII:ZKA190 Fab ratio of
1:1.1; EDIII concentration was typically 0.4 mM.
[0570] Since the NMR signal is strongly dependent on the local
chemical environment, changes upon complex formation identify
antigen residues that are affected by antibody binding, either
directly or through allosteric effects. By comparing the NMR
spectra of free and bound EDIII (FIG. 17A), residues affected by
ZKA190 were mapped to the LR of EDIII, in particular to the BC, DE
and FG loops, as well as to part of the EDI-EDIII hinge (FIG. 18A).
These residues are nearly identical among 217 known ZIKV strains,
with the exception of substitutions at V341I and E393D in the
Uganda 1947 isolate (FIG. 17D). These mutations are also present in
the MR766 strain that was efficiently neutralized by ZKA190 (FIG.
15A). Analysis of the ZKA190 epitope on the uncomplexed ZIKV
structure showed that the epitope is highly accessible, except for
the FG loop in the 5-fold vertex (FIGS. 18B and 17C, molecule
A).
[0571] Computational docking followed by molecular dynamics
simulation, guided and validated by NMR-derived epitope information
as well as EDIII mutagenesis, showed that ZKA190 binds through an
interface characterized by shape and charge complementarity (FIGS.
18B and 17E).
[0572] Docking indicates that there are no direct contacts between
ZKA190 and the FG loop on EDIII, suggesting that changes in its NMR
signals upon antibody binding derive from allosteric effects. This
notion is supported by the fact that mutations of FG loop residues
in recombinant EDIII, but not in other epitope regions, did not
affect the binding affinity of ZKA190 for EDIII (FIGS. 18B and
19).
Example 11: Mechanisms of ZKA190 Neutralization
[0573] The ability of ZKA190 to efficiently neutralize the virus
may involve inhibition of either cell attachment or membrane
fusion. A further mechanism might involve virus inactivation
through cross-linking of viral particles.
[0574] ZKA190 Fab can neutralize ZIKV, albeit less efficiently than
the corresponding IgG. By binding to the EDI-EDIII linker, ZKA190
(both Fab and IgG) might inhibit the .about.70 degree rotation of
DIII required for viral fusion to the host cell membrane
(Bressanelli, S., Stiasny, K., Allison, S. L., Stura, E. A.,
Duquerroy, S., Lescar, J., Heinz, F. X., and Rey, F. A. (2004).
Structure of a flavivirus envelope glycoprotein in its
low-pH-induced membrane fusion conformation. Embo J 23, 728-738;
Modis, Y., Ogata, S., Clements, D., and Harrison, S. C. (2004).
Structure of the dengue virus envelope protein after membrane
fusion. Nature 427, 313-319). Alternatively, ZKA190 might prevent
the attachment of ZIKV to target cells.
[0575] The ability of ZKA190 to inhibit membrane fusion is
supported by confocal microscopy analysis. To this end, Vero cells
were plated at 7,500 cells/well on 12 mm-diameter coverslips in
24-well plates and incubated overnight. Cells were infected with
ZIKV H/PF/2013 (MOI of 100) in the presence or absence of
neutralizing concentrations of Alexa-488 conjugated mAbs (0.7
.mu.M) at 37.degree. C. for 3 h, washed with PBS, and fixed with 2%
paraformaldehyde in PBS for 30 min at room temperature. Acidified
endosome were identified with Lysotracker red (Invitrogen) by
adding the dye (50 nM) to the cells for the last 30 min of the
incubation prior to fixation. Fixation was followed by extensive
washes in PBS and 50 mM glycine and finally the coverslips were
prepared for microscopy analysis using Vectashield mounting medium
for fluorescence with DAPI (Vector Laboratories). Samples were
analyzed by confocal microscopy using a Leica TCS SP5 microscope
with a 63.times./1.4 N. A. objective. Image analysis and processing
was performed with FIJI software.
[0576] Results are shown in FIG. 20. Confocal microscopy analysis
shows that ZKA190 (Fab or IgG) can enter Vero cells only when
complexed with ZIKV, at neutralizing concentrations exceeding the
IC50 by 10,000-fold (FIG. 20).
Example 12: In Vivo Characterization of the EDIII-Specific mAb
ZKA190
[0577] To evaluate their prophylactic and therapeutic properties,
ZKA190 and ZKA190-LALA were tested in A129 mice challenged with a
lethal dose of ZIKV strain MP1751 (African lineage). To test their
prophylactic potencies, ZKA190 and ZKA190-LALA were administered
one day before virus challenge.
[0578] Female A129 mice (IFN-alpha/beta receptor -/-) and wild-type
129Sv/Ev mice aged 5-8 weeks were administered mAbs (ZKA190,
ZKA190-LALA and control antibody MPE8 (Corti, D., et al.
Cross-neutralization of four paramyxoviruses by a human monoclonal
antibody. Nature 501, 439-443 (2013)) diluted in PBS at different
doses via the intraperitoneal (i.p.) route in a volume of 500
.mu.l. MAbs were administered either 1 day before or 1, 2, 3 or 4
days after virus challenge. Animals were challenged subcutaneously
with 102 pfu ZIKV (strain MP1751) and followed for 14 days. Weights
and temperatures were monitored daily and clinical observations
were recorded at least twice per day. On day 5 post-challenge, 50
.mu.l of blood was collected from each animal into a RNAprotect
tube (Qiagen, UK) and frozen at -80.degree. C. At the end of the
study (14 days post-challenge) or when animals met human endpoints,
necropsies were undertaken, and blood and sections of brain,
spleen, liver, kidney and ovary were collected for virological
analysis.
[0579] Tissue samples from A129 mice were weighed and homogenized
into PBS using ceramic beads and an automated homogenizer
(Precellys, UK) using six 5 second cycles of 6500 rpm with a 30
second gap. Two hundred .mu.l of tissue homogenate or blood
solution was transferred into 600 .mu.L RLT buffer (Qiagen, UK) for
RNA extraction using the RNeasy Mini extraction kit (Qiagen, UK);
samples were passed through a QlAshredder (Qiagen, UK) as an
initial step. A ZIKV specific realtime RT-PCR assay was utilized
for the detection of viral RNA from subject animals. The primer and
probe sequences were adopted from Quick et al., 2017 (Quick, J,
Grubaugh N D, Pullan S T, Claro I M, Smith A D, Gangavarapu K,
Oliveira G, Robles-Sikisaka R, Rogers T F, Beutler N A, et al.:
Multiplex PCR method for MinION and Illumina sequencing of Zika and
other virus genomes directly from clinical samples. Nat Protoc
2017, 12:1261-1276) with in-house optimization and validation
performed to provide optimal mastermix and cycling conditions.
Real-time RT-PCR was performed using the SuperScript III Platinum
One-step qRT-PCR kit (Life Technologies, UK). The fmal mastermix
(15111) was comprised of 10 .mu.l of 2.times. Reaction Mix, 1.2
.mu.l of PCR-grade water, 0.2 .mu.l of 50 mM MgSO4, 1 .mu.l of each
primer (ZIKV 1086 and ZIKV 1162c both at 18 .mu.M working
concentration), 0.8 .mu.l of probe (ZIKV 1107-FAM at 25 .mu.M
working concentration) and 0.8 .mu.l of SSIII enzyme mix. Five
.mu.l of template RNA was added to the mastermix, yielding a final
reaction volume of 20 .mu.l. The cycling conditions used were
50.degree. C. for 10 minutes, 95.degree. C. for 2 minutes, followed
by 45 cycles of 95.degree. C. for 10 seconds and 60.degree. C. for
40 seconds, plus a final cooling step of 40.degree. C. for 30
seconds. Quantification analysis using fluorescence was performed
at the end of each 60.degree. C. step. Reactions were run and
analyzed on the 7500 Fast platform (Life Technologies, UK) using
7500 software version 2.0.6. Quantification of viral load in
samples was performed using a dilution series of quantified RNA
oligonucleotide (Integrated DNA Technologies). The oligonucleotide
comprised the 77 bases of ZIKV RNA targeted by the assay, based on
GenBank accession AY632535.2 and was synthesized to a scale of 250
nmole with HPLC purification.
[0580] Results are shown in FIGS. 21, 22 and 23. ZKA190 and
ZKA190-LALA were shown to protect mice from mortality and morbidity
at concentrations of 5, 1 or 0.2 mg/kg (FIG. 21A-B). ZKA190-LALA,
and to a lesser extent ZKA190, delayed morbidity and mortality as
compared to the control group at 0.04 mg/kg. Viral titers in blood
and organs were reduced significantly compared to control
antibody-treated animals, even in the presence of serum antibody
levels below 1 .mu.g/ml (FIG. 22A-D).
[0581] To evaluate the therapeutic potential of ZKA190, we
administered ZKA190 and ZKA190-LALA at different time-points
following ZIKV infection. At a dose of 15 mg/kg, survival rates of
80%-100% were achieved, and the morbidity was greatly reduced even
when treatment was given four days post-infection (FIG. 21E-G).
ZKA190 and ZKA190-LALA treatment at all post-infection time-points
resulted in significantly reduced viral titers, compared to animals
treated with control antibody, with a clear trend for greater
reduction with earlier treatment (FIG. 23A-21C). Of note,
ZKA190-LALA showed a significantly reduced antiviral activity in
the blood day 5 sample as compared to ZKA190 when mAbs were given
four days post-infection, a result that might be related to the
impaired ability of the LALA variant to facilitate rapid clearance
of coated virions.
[0582] Tables of Sequences and SEQ ID Numbers
TABLE-US-00010 SEQ ID ZKA190 NO. Amino acid sequence CDRH1 1
GFTFSKYG CDRH2 2 ISYEGSNK CDRH3 3 AKSGTQYYDTTGYEYRGLEYFGY CDRL1 4
QSVSSSY CDRL2 5 DAS CDRL2 6 LIYDASSRA long CDRL3 7 QQYGRSRWT VH 8
QVQLVESGGGVVQPGRSLRLSCAASGFTFSKYGMHWVRQA
PGKGLEWVAVISYEGSNKYYADSVKGRFTISRDNSKNTLY
LQMNSLRAEDTAVYYCAKSGTQYYDTTGYEYRGLEYFGYW GQGTLVTVSS VL 9
EIVLTQSPGTLSLSPGERATLSCRASQSVSSSYLAWYQQK
RGQAPRLLIYDASSRATGIPDRFSGSGSGTDFTLTISRLE
PEDFAVYYCQQYGRSRWTFGQGTKVEVEIK
TABLE-US-00011 SEQ ID ZKA190 NO. Nucleic acid sequence CDRH1 10
ggattcaccttcagtaaatatggc CDRH2 11 atatcatatgagggaagtaataaa CDRH3 12
gcgaaatcggggacccaatactatgatactactggttatg
agtataggggtttggaatactttggctac CDRL1 13 cagagtgttagtagcagttac CDRL2
14 gatgcatcc CDRL2 15 ctcatctatgatgcatccagcagggcc long CDRL3 16
cagcagtatggtaggtcaaggtggaca VH 17
caggtgcagctggtggagtctgggggaggcgtggtccagc
ctgggaggtccctgagactctcctgtgcagcctctggatt
caccttcagtaaatatggcatgcactgggtccgccaggct
ccaggcaaggggctggagtgggtggcagttatatcatatg
agggaagtaataaatattatgcagactccgtgaagggccg
attcaccatctccagagacaattccaagaacacgctgtat
ctgcaaatgaacagcctgagagctgaggacacggcagtgt
attactgtgcgaaatcggggacccaatactatgatactac
tggttatgagtataggggtttggaatactttggctactgg
ggccagggaaccctggtcaccgtctcctcag VL 18
gaaattgtgttgacgcagtctccaggcaccctgtctttgt
ctccaggggaaagagccaccctctcctgcagggccagtca
gagtgttagtagcagttacttagcctggtaccagcagaaa
cgtggccaggctcccaggctcctcatctatgatgcatcca
gcagggccactggcatcccagacaggttcagtggcagtgg
gtctgggacagacttcactctcaccatcagcagactggag
cctgaagattttgcagtgtattactgtcagcagtatggta
ggtcaaggtggacattcggccaagggaccaaggtggaaat caaac
TABLE-US-00012 SEQ ID ZKA185 NO. Amino acid sequence CDRH1 19
GYSFTSYW CDRH2 20 FDPSDSQT CDRH3 21 ARRYCSSSSCYVDN CDRL1 22 ALPNKF
CDRL2 23 EDN CDRL2 24 VIYEDNKRP long CDRL3 25 YSTDSSSNPLGV VH 26
EVQLVQSGAEVKKPGESLRISCKGSGYSFTSYWITWVRQM
PGKGLEWMAKFDPSDSQTNYSPSFQGHVTISVDKSISTAY
LQWSSLKASDTAMYYCARRYCSSSSCYVDNWGQGTLVTIF S VL 27
SYELTQPPSVSVSPGQTARITCSGDALPNKFAYWYRQKSG
QAPVLVIYEDNKRPSGIPERFSGSSSGTMATLTISGAQVE
DEADYHCYSTDSSSNPLGVFGGGTKLTVL
TABLE-US-00013 SEQ ID ZKA185 NO. Nucleic acid sequence CDRH1 28
ggatatagttttaccagttactgg CDRH2 29 tttgatcctagtgactctcaaacc CDRH3 30
gcgagaagatattgtagtagtagtagttgttatgtggaca at CDRL1 31
gcattgccaaataaattt CDRL2 32 gaggacaac CDRL2 33
gtcatctatgaggacaacaaacgaccc long CDRL3 34
tactcaacagacagcagttctaatcccctgggagta VH 35
gaagtgcagctggtgcagtccggagcagaggtgaaaaagc
ccggggagtctctgaggatctcctgtaagggttctggata
tagttttaccagttactggatcacctgggtgcgccagatg
cccgggaaaggcctggagtggatggcgaagtttgatccta
gtgactctcaaaccaactacagcccgtccttccaaggcca
cgtcaccatctcagttgacaagtccatcagcactgcctac
ttgcagtggagcagcctgaaggcctcggacaccgccatgt
attactgtgcgagaagatattgtagtagtagtagttgtta
tgtggacaattggggccagggaaccctggtcaccatcttc tcag VL 36
tcctatgagctgacacagccaccctcggtgtcagtgtccc
caggacaaacggccaggatcacctgctctggagatgcatt
gccaaataaatttgcttattggtaccggcagaagtcaggc
caggcccctgttctggtcatctatgaggacaacaaacgac
cctccgggatccctgagagattctctggctccagctcagg
gacaatggccaccttgactatcagtggggcccaggtggag
gatgaagctgactaccactgttactcaacagacagcagtt
ctaatcccctgggagtattcggcggagggaccaagctgac cgtcctag
TABLE-US-00014 SEQ ID ZKA230 NO. Amino acid sequence CDRH1 37
GGSISSDY CDRH2 38 IYYSGST CDRH3 39 ARRRKYDSLWGSFAFDI CDRL1 40
SSNIGGNY CDRL2 41 IND CDRL2 42 LICINDHRP long CDRL3 43 ATWDDSLGGLV
VH 44 QVQLQESGPGLVKPSETLSLTCAVSGGSISSDYWSWIRQP
PGKGLEWIGYIYYSGSTNYNPSLKSRVTISVDTSKNHFSL
KLNSVTAADTAVYYCARRRKYDSLWGSFAFDIWGQGTMVT VSS VL 45
QSVLTQPPSASGTPGQRVTISCSGSSSNIGGNYVYWYQQL
PGTAPKLLICINDHRPSGVPDRFSGSKSGTSASLAISGLQ
SEDEADYYCATWDDSLGGLVFGGGTKLTVL SEQ ID ZKA230 NO. Nucleic acid
sequence CDRH1 46 ggtggctccatcagtagtgactac CDRH2 47
atctattacagtgggagcacc CDRH3 48
gcgaggaggaggaagtatgattccctttgggggagttttg cttttgatatc CDRL1 49
agctccaacatcggaggtaattat CDRL2 50 attaatgat CDRL2 51
ctcatctgtattaatgatcaccggccc long CDRL3 52
gcaacatgggatgacagcctgggtggccttgta VH 53
caggtgcagctgcaggagtcgggcccaggcctggtgaagc
cttcggagaccctgtccctcacctgcgcagtctctggtgg
ctccatcagtagtgactactggagctggatccggcagccc
ccagggaagggactggagtggattgggtatatctattaca
gtgggagcaccaactacaacccctccctcaagagtcgagt
caccatatcagtagacacgtccaagaaccacttctccctg
aagctgaactctgtgaccgctgcggacacggccgtgtatt
actgtgcgaggaggaggaagtatgattccctttgggggag
ttttgcttttgatatctggggccaagggacaatggtcacc gtctcttcag VL 54
cagtctgtgctgactcagccaccctcagcgtctgggaccc
ccgggcagagggtcaccatctcttgttctggaagcagctc
caacatcggaggtaattatgtatactggtaccagcagctc
ccaggaacggcccccaaactcctcatctgtattaatgatc
accggccctcaggggtccctgaccgattctctggctccaa
gtctggcacctcagcctccctggccatcagtgggctccag
tccgaggatgaggctgattattactgtgcaacatgggatg
acagcctgggtggccttgtattcggcggagggaccaagct gaccgtcctag
TABLE-US-00015 SEQ ID ZKA78 NO. Amino acid sequence CDRH1 55
GFTFSNYA CDRH2 56 IGRNGDSI CDRH3 57 VKDLAIPESYRIEADY CDRL1 58
QSVLYRSNNKNY CDRL2 59 WAS CDRL2 60 LIYWASTRE long CDRL3 61
QQYYSSPRT VH 62 EVQLAESGGGLVQPGGSLTLSCSGSGFTFSNYAMVWARQA
PGKGLEYVSGIGRNGDSIYYTDSVKGRFTISRDNSKSMVY
LQMSSLRTEDTAVYYCVKDLAIPESYRIEADYWGQGTLVI VSA VL 63
DIVMTQSPDSLAVSLGERATINCKSSQSVLYRSNNKNYLS
WYQQKPGQPPKLLIYWASTRESGVPDRFSGSGSGTDFTLT
ISPLQAEDVAVYYCQQYYSSPRTFGQGTKVEIK
TABLE-US-00016 SEQ ID ZKA78 NO. Nucleic acid sequence CDRH1 64
ggcttcacttttagtaactatgca CDRH2 65 atcgggcgcaacggggactctatc CDRH3 66
gtgaaagatctggccatccccgagtcctacagaattgaag ctgattat CDRL1 67
cagtccgtgctgtaccgctctaacaacaagaattac CDRL2 68 tgggcttca CDRL2 69
ctgatctattgggcttcaacccgggaa long CDRL3 70
cagcagtactattctagtcctcgaact VH 71
gaggtgcagctggcagaatcaggcgggggactggtccagc
ctggcggcagcctgacactgtcttgcagtggatcaggctt
cacttttagtaactatgcaatggtgtgggcaaggcaggct
cctgggaagggactggagtatgtctctggcatcgggcgca
acggggactctatctactatactgatagtgtgaagggccg
gttcaccatcagcagagacaatagcaaatccatggtgtac
ctgcagatgagctccctgcgaaccgaagacacagcagtgt
actattgcgtgaaagatctggccatccccgagtcctacag
aattgaagctgattattggggacagggcaccctggtcatc gtgagcgccg VL 72
gacatcgtgatgacacagtctccagatagtctggcagtca
gtctgggggagagggccactattaactgcaagagctccca
gtccgtgctgtaccgctctaacaacaagaattacctgtct
tggtatcagcagaagcccggacagccccctaaactgctga
tctattgggcttcaacccgggaaagcggcgtcccagacag
attctcaggcagcgggtccggaacagacttcaccctgaca
attagccccctgcaggcagaggacgtggctgtctactatt
gtcagcagtactattctagtcctcgaactttcggccaggg gaccaaggtggaaatcaaac
TABLE-US-00017 SEQ ID ZKA64 NO. Amino acid sequence CDRH1 73
GYTFTGYH CDRH2 74 INPNSGGT CDRH3 75 ARMSSSIWGFDH CDRL1 76 QSVLIN
CDRL2 77 GAS CDRL2 78 LIYGASSRA long CDRL3 79 QQYNDWPPIT VH 80
QVQLVQSGAEVKKPGASVKVSCKASGYTFTGYHIDWVRQA
RGQGLEWMGRINPNSGGTNYAQKFQGRVTMTRDTSISTAY
MQLSRLRSDDSAVYYCARMSSSIWGFDHWGQGTLVTVSS VL 81
EIVMTQSPATLSVSPGERATLSCRASQSVLINLAWYQQKP
GQAPRLLIYGASSRATGIPARFSGSGSGTEFTLTISSLQS
EDFAVYYCQQYNDWPPITFGQGTRLEVEIK
TABLE-US-00018 SEQ ID ZKA64 NO. Nucleic acid sequence CDRH1 82
ggctacaccttcacagggtatcac CDRH2 83 attaaccctaattctggcgggacc CDRH3 84
gctcggatgagctcctctatttggggcttcgatcat CDRL1 85 cagtctgtgctgattaac
CDRL2 86 ggagcatcc CDRL2 87 ctgatctatggagcatcctccagggct long CDRL3
88 cagcagtacaatgattggccccctatcaca VH 89
caggtgcagctggtccagagcggagcagaggtgaagaaac
ccggcgcctcagtgaaggtcagctgcaaagcttccggcta
caccttcacagggtatcacatcgactgggtgaggcaggca
agaggacagggactggaatggatgggacggattaacccta
attctggcgggaccaactacgcccagaagtttcagggccg
agtgactatgaccagagacaccagcatctccacagcttat
atgcagctgtcccggctgagatctgacgatagtgccgtct
actattgtgctcggatgagctcctctatttggggcttcga
tcattgggggcagggaacactggtgactgtcagttcag VL 90
gagatcgtgatgactcagtctccagccaccctgtcagtca
gcccaggagaacgggcaaccctgtcttgcagagcctccca
gtctgtgctgattaacctggcttggtaccagcagaagcca
ggccaggcaccccgactgctgatctatggagcatcctcca
gggctaccggcattcctgcacgcttcagtggatcaggaag
cggaacagagtttaccctgacaatctctagtctgcagtcc
gaagacttcgctgtctactattgtcagcagtacaatgatt
ggccccctatcacatttggccaggggactagactggagat caagc
TABLE-US-00019 SEQ ID ZKA15 NO. Amino acid sequence CDRH1 91
GGFINSYY CDRH2 92 IYKSGST CDRH3 93 ARDPYGDYVKAFDI CDRL1 94
QSLLHSNGYNY CDRL2 95 LGS CDRL2 96 LIYLGSNRA long CDRL3 97 MQALQTVT
VH 98 QVQLQESGPGLVKPSETLSLTCTVSGGFINSYYWSWIRQP
AGKGLEWIGRIYKSGSTNYNPSLKSRVTMSLDTSKYQFSL
KLRSVTAADTAVYYCARDPYGDYVKAFDIWGQGTMVTVSS VL 99
DIVMTQSPLSLPVTPGEPASISCRSSQSLLHSNGYNYLNW
YLQKPGQSPQLLIYLGSNRASGVPDRFSGSGSGTDFTLKI
SRVEAEDVGVYYCMQALQTVTFGPGTKVDVEIK
TABLE-US-00020 SEQ ID ZKA15 NO. Nucleic acid sequence CDRH1 100
ggtggcttcatcaatagttactac CDRH2 101 atctataaaagtgggagcacc CDRH3 102
gcgagagatccctacggtgactacgttaaggcttttgata tt CDRL1 103
cagagcctcctgcatagtaatggatacaactat CDRL2 104 ttgggttct CDRL2 105
ctgatctatttgggttctaatcgggcc long CDRL3 106 atgcaagctctacaaactgtcact
VH 107 caggtgcagctgcaggagtcggggccaggactggtgaagc
cttcggagaccctgtccctcacctgcactgtctccggtgg
cttcatcaatagttactactggagctggatccggcagccc
gccgggaagggactggagtggattgggcgtatctataaaa
gtgggagcaccaactacaacccctccctcaagagtcgagt
caccatgtcactagacacgtccaagtaccagttctccctg
aagctgaggtctgtgaccgccgctgacacggccgtgtatt
actgtgcgagagatccctacggtgactacgttaaggcttt
tgatatttggggccaagggacaatggtcaccgtctcttca g VL 108
gatattgtgatgactcagtctccactctccctgcccgtca
cccctggagagccggcctccatctcctgcaggtctagtca
gagcctcctgcatagtaatggatacaactatttgaattgg
tacctgcagaagccagggcagtctccacagctcctgatct
atttgggttctaatcgggcctccggggtccctgacaggtt
cagtggcagtggatcaggcacagattttacactgaaaatc
agcagagtggaggctgaggatgttggggtttattactgca
tgcaagctctacaaactgtcactttcggccctgggaccaa agtggatatcaaac
TABLE-US-00021 SEQ ID Amino acid ZKA25 NO. sequence CDRH1 109
GFTFRSHW CDRH2 110 IKEDGYEK CDRH3 111 ARDLRVYSGRGFDP CDRL1 112
KLGDKY CDRL2 113 QDS CDRL2 114 VIYQDSKRP long CDRL3 115 QAWDSSTVV
VH 116 EVQLVESGGGLVRPGGSLRL SCAASGFTFRSHWMSWVRQA
PGKGLEWVANIKEDGYEKYY VDSVKGRFTISRDNAKNSLY LQMKSLRAEDTAVYYCARDL
RVYSGRGFDPWGQGTLVTVS S VL 117 SYELTQPPSLSVSPGQTASI
TCSGDKLGDKYACWYQQKPG QSPVLVIYQDSKRPSGIPAR FSGSNSGNTATLTISGTQAM
DEADYYCQAWDSSTVVFGGG TKLTVL
TABLE-US-00022 SEQ ID ZKA25 NO. Nucleic acid sequence CDRH1 118
ggattcacctttagaagtcattgg CDRH2 119 ataaaggaagatggatatgagaaa CDRH3
120 gcgagagatttgagggtatatagt gggagaggtttcgacccc CDRL1 121
aaattgggggataaatat CDRL2 122 caagatagc CDRL2 123
gtcatctatcaagatagcaagcggccc long CDRL3 124
caggcgtgggacagcagcactgtggta VH 125
gaggtgcagttggtggagtctgggggaggcttggtccggc
ctggggggtccctgagactctcctgtgcagcctctggatt
cacctttagaagtcattggatgagttgggtccgccaggct
ccagggaaggggctggagtgggtggccaacataaaggaag
atggatatgagaaatactatgtggactctgtgaagggccg
attcaccatctccagagacaacgccaagaactcactgtat
ctgcaaatgaagagcctgagagccgaggacacggccgtgt
attactgtgcgagagatttgagggtatatagtgggagagg
tttcgacccctggggccagggaaccctggtcaccgtctcc tcag VL 126
tcctatgagctgactcagccaccctcactgtccgtgtccc
caggacagacagccagcatcacctgctctggagataaatt
gggggataaatatgcttgctggtatcagcagaagccaggc
cagtcccctgtgttggtcatctatcaagatagcaagcggc
cctcagggatccctgcgcgattctctggctccaactctgg
gaacacagccactctgaccatcagcgggacccaggctatg
gatgaggctgactattactgtcaggcgtgggacagcagca
ctgtggtattcggtggagggaccaagctgaccgtcctag SEQ ID ZKA35 NO. Amino acid
sequence CDRH1 127 GGSISTGGYY CDRH2 128 IYYSGNT CDRH3 129
AKGGGRERPFDY CDRL1 130 SSNIGRNY CDRL2 131 RNN CDRL2 132 LIYRNNQRP
long CDRL3 133 VAWDDSRSGFVV VH 134
QVQLQESGPGLVKPSQTLSLTCTVSGGSISTGGYYWSWIR
QHPGKGLEWIGYIYYSGNTYYNPSLKSRVTISVDTSKKQF
SLKLSSVTAADTAVYYCAKGGGRERPFDYWGQGTLVTVSS VL 135
QSVLTQPPSASGTPGQRVTISCSGSSSNIGRNYVDWYQQL
PGTAPKLLIYRNNQRPSGVPERFSGSKSGTSASLAISGLR
SEDEADYYCVAWDDSRSGFVVFGGGTKVTVL
TABLE-US-00023 SEQ ID ZKA35 NO. Nucleic acid sequence CDRH1 136
ggtggctccatcagcactggtggttactac CDRH2 137 atctattacagtgggaacacc
CDRH3 138 gcgaaaggaggagggagggagcgaccctttgactac CDRL1 139
agctccaacatcggaagaaattat CDRL2 140 aggaataat CDRL2 141
ctcatctataggaataatcagcggccc long CDRL3 142
gtagcatgggatgacagccggagtggttttgtggta VH 143
caggtgcagctgcaggagtcgggcccaggactggtgaagc
cttcacagaccctgtccctcacctgcactgtctctggtgg
ctccatcagcactggtggttactactggagctggatccgc
cagcacccagggaagggcctggagtggattggttacatct
attacagtgggaacacctactacaacccgtccctcaagag
tcgagttaccatatcagttgacacctctaagaagcagttc
tccctgaagctgagctctgtgactgccgcggacacggccg
tgtattactgtgcgaaaggaggagggagggagcgaccctt
tgactactggggccagggaaccctggtcaccgtctcctca g VL 144
cagtctgtgctgactcagccaccctcagcgtctgggaccc
ccgggcagagggtcaccatctcttgttctggaagcagctc
caacatcggaagaaattatgtagactggtaccagcaactc
ccaggaacggcccccaaactcctcatctataggaataatc
agcggccctcaggggtccctgagcgattctctggctccaa
gtctggcacctcagcctccctggccatcagtgggctccgg
tccgaggatgaggctgattattactgtgtagcatgggatg
acagccggagtggttttgtggtattcggcggagggaccaa ggtgaccgtcctag
TABLE-US-00024 SEQ Constant ID regions NO. Sequence IgG1 145
ASTKGPSVFPLAPSSKSTSGGT CH1-CH2- AALGCLVKDYFPEPVTVSWNSG CH3 aa
ALTSGVHTFPAVLQSSGLYSLS SVVTVPSSSLGTQTYICNVNHK
PSNTKVDKRVEPKSCDKTHTCP PCPAPELLGGPSVFLFPPKPKD
TLMISRTPEVTCVVVDVSHEDP EVKFNWYVDGVEVHNAKTKPRE
EQYNSTYRVVSVLTVLHQDWLN GKEYKCKVSNKALPAPIEKTIS
KAKGQPREPQVYTLPPSREEMT KNQVSLTCLVKGFYPSDIAVEW
ESNGQPENNYKTTPPVLDSDGS FFLYSKLTVDKSRWQQGNVFSC
SVMHEALHNHYTQKSLSLSPGK IgG1 146 ASTKGPSVFPLAPSSKSTSGGT CH1-CH2-
AALGCLVKDYFPEPVTVSWNSG CH3 ALTSGVHTFPAVLQSSGLYSLS LALA aa
SVVTVPSSSEGTQTYICNVNHK PSNTKVDKKVEPKSCDKTHTCP
PCPAPEAAGGPSVFLFPPKPKD TLMISRTPEVTCVVVDVSHEDP
EVKFNWYVDGVEVHNAKTKPRE EQYNSTYRVVSVLTVLHQDWLN
GKEYKCKVSNKALPAPIEKTIS KAKGQPREPQVYTLPPSRDELT
KNQVSLTCLVKGFYPSDIAVEW ESNGQPENNYKTTPPVLDSDGS
FFLYSKLTVDKSRWQQGNVFSC SVMHEALHNHYTQKSLSLSPGK IgG 147
RTVAAPSVFIFPPSDEQLKSGT CK aa ASVVCLLNNFYPREAKVQWKVD
NALQSGNSQESVTEQDSKDSTY SLSSTLTLSKADYEKHKVYACE VTHQGLSSPVTKSFNRGEC
IgG 148 GQPKAAPSVTLFPPSSEELQAN CL aa KATLVCLISDFYPGAVTVAWKA
DSSPVKAGVETTTPSKQSNNKY AASSYLSLTPEQWKSHRSYSCQ VTHEGSTVEKTVAPTECS
IgG1 149 gcgtcgaccaagggcccatcgg CH1-CH2- tcttccccctggcaccctcctc CH3
caagagcacctctgggggcaca nucl gcggccctgggctgcctggtca
aggactacttccccgaacctgt gacggtctcgtggaactcaggc
gccctgaccagcggcgtgcaca ccttcccggctgtcctacagtc
ctcaggactctactccctcagc agcgtggtgaccgtgccctcca
gcagcttgggcacccagaccta catctgcaacgtgaatcacaag
cccagcaacaccaaggtggaca agagagttgagcccaaatcttg
tgacaaaactcacacatgccca ccgtgcccagcacctgaactcc
tggggggaccgtcagtcttcct cttccccccaaaacccaaggac
accctcatgatctcccggaccc ctgaggtcacatgcgtggtggt
ggacgtgagccacgaAgaCcct gaggtcaagttcaactggtacg
tggacggcgtggaggtgcataa tgccaagacaaagccgcgggag
gagcagtacaacagcacgtacc gtgtggtcagcgtcctcaccgt
cctgcaccaggactggctgaat ggcaaggagtacaagtgcaagg
tctccaacaaagccctcccagc ccccatcgagaaaaccatctcc
aaagccaaagggcagccccgag aaccacaggtgtacaccctgcc
cccatcccgggaggagatgacc aagaaccaggtcagcctgacct
gcctggtcaaaggcttctatcc cagcgacatcgccgtggagtgg
gagagcaatgggcagccggaga acaactacaagaccacgcctcc
cgtgctggactccgacggctcc ttcttcctctatagcaagctca
ccgtggacaagagcaggtggca gcaggggaacgtcttctcatgc
tccgtgatgcatgaggctctgc acaaccactacacgcagaagag
cctctccctgtccccgggtaaa IgG1 150 gcgtcgaccaagggcccatcgg CH1-CH2-
tcttccccctggcaccctcctc CH3 caagagcacctctgggggcaca LALA
gcggccctgggctgcctggtca nucl aggactacttccccgaacctgt
gacggtctcgtggaactcaggc gccctgaccagcggcgtgcaca
ccttcccggctgtcctacagtc ctcaggactctactccctcagc
agcgtggtgaccgtgccctcca gcagcttgggcacccagaccta
catctgcaacgtgaatcacaag cccagcaacaccaaggtggaca
agagagttgagcccaaatcttg tgacaaaactcacacatgccca
ccgtgcccagcacctgaaGCCG CGgggggaccgtcagtcttcct
cttccccccaaaacccaaggac accctcatgatctcccggaccc
ctgaggtcacatgcgtggtggt ggacgtgagccacgaagaccct
gaggtcaagttcaactggtacg tggacggcgtggaggtgcataa
tgccaagacaaagccgcgggag gagcagtacaacagcacgtacc
gtgtggtcagcgtcctcaccgt cctgcaccaggactggctgaat
ggcaaggagtacaagtgcaagg tctccaacaaagccctcccagc
ccccatcgagaaaaccatctcc aaagccaaagggcagccccgag
aaccacaggtgtacaccctgcc cccatcccgggaggagatgacc
aagaaccaggtcagcctgacct gcctggtcaaaggcttctatcc
cagcgacatcgccgtggagtgg gagagcaatgggcagccggaga
acaactacaagaccacgcctcc cgtgctggactccgacggctcc
ttcttcctctatagcaagctca ccgtggacaagagcaggtggca
gcaggggaacgtcttctcatgc tccgtgatgcatgaggctctgc
acaaccactacacgcagaagag cctctccctgtccccgggtaaa IgG CK 151
cgTacGgtggctgcaccatctg nucl tcttcatcttcccgccatctga
tgagcagttgaaatctggaact gcctctgttgtgtgcctgctga
ataacttctatcccagagaggc caaagtacagtggaaggtggat
aacgccctccaatcgggtaact cccaggagagtgtcacagagca
ggacagcaaggacagcacctac agcctcagcagcaccctgacgc
tgagcaaagcagactacgagaa acacaaagtctacgcctgcgaa
gtcacccatcagggcctgagct cgcccgtcacaaagagcttcaa caggggagagtgt IgG CL
152 ggtcagcccaaggctgccccct nucl cggtcactctgttcccgccctc
ctctgaggagcttcaagccaac aaggccacactggtgtgtctca
taagtgacttctacccgggagc cgtgacagtggcttggaaagca
gatagcagccccgtcaaggcgg gagtggagaccaccacaccctc
caaacaaagcaacaacaagtac gcggccagcagctatctgagcc
tgacgcctgagcagtggaagtc ccacagaagctacagctgccag
gtcacgcatgaagggagcaccg tggagaagacagtggcccctac agaatgttca
TABLE-US-00025 SEQ ID ZKA10 NO. Amino acid sequence CDRH1 153
GFTFSDSY CDRH2 154 ISSSSPFT CDRH3 155 ARGLVRDGYKWLYFFDY CDRH3 156
QVQLVESGGGLVEPRGSLRLSCAASGFTFSDSY MSWIRQAPGKGLEWISYISSSSPFTNYADSVKG
RFTISRDNAKNSLYLQMNSLRAEDTAVYYCARG LVRDGYKWLYFFDYWGQGTLVTVSS SEQ ID
ZKA18 NO. Amino acid sequence CDRH1 157 GFTFSSYG CDRH2 158 IWYDGSNK
CDRH3 159 ARDDSGYSEPFDY CDRH3 160 QVQLVESGGGVVQPGRSLRLSCAASGFTFSSYG
MHWVRQAPGKGLEWVAVIWYDGSNKYYADSVKG RFTITRDNSKNTLYLQMNSLRPEDTAVYYCARD
DSGYSEPFDYWGQGTLVTVSS SEQ ID ZKA28 NO. Amino acid sequence CDRH1
161 GFTVSRNY CDRH2 162 IYSGGST CDRH3 163 ARWINDAFDI CDRH3 164
EVQLVESGGGLIQPGGSLRLSCAASGFTVSRNY MSWVRQAPGKGLEWVSVIYSGGSTYYADSVKGR
FTISRDNSKNTLYLQMNSLRAEDTAVYYCARWI NDAFDIWGQGTMVTVSS SEQ ID ZKA29
NO. Amino acid sequence CDRH1 165 GFTFSRYS CDRH2 166 ISPRSTTI CDRH3
167 AREDCTNGVCYRVDY CDRH3 168 EVQLVESGGGLVQPGGSLRLSCVVSGFTFSRYS
MNWVRQAPGKGLEWVSYISPRSTTIYYADSVEG RFTVSRDNAKNSLYLQLNSLRAEDTAVYYCARE
DCTNGVCYRVDYWGQGTLVTVSS SEQ ID ZKA33 NO. Amino acid sequence CDRH1
169 GFTFSRNW CDRH2 170 IKEDGNEK CDRH3 171 ARPFHQGGYAYGLAY CDRH3 172
EVQLVESGGGLVQPGGSLRLSCAASGFTFSRNW MTWVRQAPGKGLEWVANIKEDGNEKYYVDSVKG
RFTISRDNAKNSLYLQMNSLRAEDTAVYYCARP FHQGGYAYGLAYWGQGTLVTVSS SEQ ID
ZKA39 NO. Amino acid sequence CDRH1 173 GFTFSTYS CDRH2 174 ISPSSSTI
CDRH3 175 AREYCSGGSCYLLDY CDRH3 176
EVQLVESGGGLVQPGGSLRLSCAASGFTFSTYS MNWVRQAPGKGLEWVSYISPSSSTIYYPDSLKG
RFTISRDNAKNSLYLQMDSLRAEDTAQYYCARE YCSGGSCYLLDYWGQGTLVTVSS SEQ ID
ZKA43 NO. Amino acid sequence CDRH1 177 GGSITSYY CDRH2 178 SHYSGST
CDRH3 179 ARGIYSGKNWFDP CDRH3 180 QVQLQESGPGLVKPSETLSLTCTVYGGSITSYY
WTWIRQPPGKGLEWIGYSHYSGSTNYNPSLKSR VTISIDTSKSQFSLNLNSVTAADTAVYYCARGI
YSGKNWFDPWGQGTLVTVSS SEQ ID ZKA44 NO. Amino acid sequence CDRH1 181
GFTVSTSY CDRH2 182 IYSSGST CDRH3 183 ARVSLGGLDP CDRH3 184
EVQLVESGGGLIQPGGSLRLSCVASGFTVSTSY GMNWVRQAPKGLEWVSVIYSSGSTYYADSVKGR
FTISRNTSKNTLYLQMNSLRAEDTAVYYCARVS LGGLDPWGQGTPVTVSS SEQ ID ZKA46
NO. Amino acid sequence CDRH1 185 GFSLSNGRMG CDRH2 186 IFSNDEK
CDRH3 187 ARVEFRAGNYLDS CDRH3 188 QVTLKESGPVLVKPTETLTLTCTVSGFSLSNGR
MGVSWIRQPPGKALEWLAHIFSNDEKYYSTSLK NRLTISKDTSKSQVVLTMTNMDPVDTATYYCAR
VEFRAGNYLDSWGQGTLVTVSS SEQ ID ZKA50 NO. Amino acid sequence CDRH1
189 GYTFTNSW CDRH2 190 IYPGDSDT CDRH3 191 ARQPFFDY CDRH3 192
EVQLVQSGAQVKKPGESLKISCKASGYTFTNSW IGWVRQMPGKGLEWMGIIYPGDSDTRYSPSFQG
QVTISADKSISTAYLQWSSLKASDTAMYYCARQ PFFDYWGQGTLVTVSS SEQ ID ZKA54 NO.
Amino acid sequence CDRH1 193 GYTFTGYY CDRH2 194 INANSGGT CDRH3 195
AHSDIVVVPSDDYYALDV CDRH3 196 QVQLVQSGAEVKKPGASVKVSCKTSGYTFTGY
YMHWVRQAPGQGLEWMGWINANSGGTNFAQRF QGRVTMTWDTSISTAYMELSRLRSDDTAVYYC
AHSDIVVVPSDDYYALDVWGQGTTVTVSS SEQ ID ZKB18 NO. Amino acid sequence
CDRH1 197 GYSFTSYW CDRH2 198 IYPGDSDT CDRH3 199 ARQTPGDY CDRH3 200
EVQLVQSGAEVKKPGESLKISCKTFGYSFTS YWIGWVRQMPGKGLEWMGMIYPGDSDTRYSP
SFQGQVTISADMSISTAYLQWSSLKASDTAM YYCARQTPGDYWGQGTLVTVSS SEQ ID ZKB20
NO. Amino acid sequence CDRH1 201 GYFFTRYV CDRH2 202 INTDNGST CDRH3
203 ARGTGRDGYNSFFAN CDRH3 204 QVQLVQSGAEVKKPGASVRVSCKASGYFFTR
YVILWVRQAPGQRPEWMGWINTDNGSTRYSQ KFQGRVTITKDTSATTAYMDLSSLKSDDTAV
YYCARGTGRDGYNSFFANWGQGTLVTVSP SEQ ID ZKB21 NO. Amino acid sequence
CDRH1 205 GYTFTGYS CDRH2 206 IDTNSGDT CDRH3 207 ARDRERHPFSY CDRH3
208 QVQLVQSGAEVKKPGASVKVSCKASGYTFTGY
SIHWVRQAPGQGLAWMGRIDTNSGDTNYAERF QGRVTMTRDTSISTAYMEVRRLRSDDTAVYYC
ARDRERHPFSYWGQGTLVTVSS SEQ ID ZKB23 NO. Amino acid sequence CDRH1
209 GGSISSGDYS CDRH2 210 ITHSGTT CDRH3 211 ARHFGWFDP CDRH3 212
QLQLQESGSGLVKPSQTLSLTCAVSGGSISSG DYSWSWIRQPPGKGLEWIGYITHSGTTYFNPS
LKSRVTISVDRSRNQFSLKVTSVTAADTAVYY CARHFGWFDPWGQGTLVTVSS SEQ ID ZKC29
NO. Amino acid sequence CDRH1 213 GGSISSGEYF CDRH2 214 IHNRGNT
CDRH3 215 ARGGGDLVVVPDSIWDYYGMDV CDRH3 216
QVQLQESGPGLVRPSQTLSLTCTVSGGSISS GEYFWTWIRQHPKKGLEWIGYIHNRGNTYYN
PSLKSRLSISLDTSKNHLSLRLSSVTAADTA VYYCARGGGDLVVVPDSIWDYYGMDVWGQGT
TVTVSS SEQ ID ZKC31 NO. Amino acid sequence CDRH1 217 GGSISSGGYH
CDRH2 218 IYYSGST
CDRH3 219 ARDRSEPGEYHYYYYAMDV CDRH3 220
QVQLQESGPGLVKPSQTLSLTCTVSGGSISS GGYHWSWIRQHPGKGLEWIGYIYYSGSTYYN
PSLKRRVTISVDTSKNQFSLKLSSVSAADTA VYYCARDRSEPGEYHYYYYAMDVWGQGTTV TVSS
SEQ ID ZKC32 NO. Amino acid sequence CDRH1 221 GFTVSSNY CDRH2 222
IYSSGST CDRH3 223 ARGKKGNAFDI CDRH3 224
EVQLVESGGDLIQPGGSLRLSCAASGFTVSS NYMSWVRQAPGKGLEWVSVIYSSGSTYYADS
VKGRFTISRDNSKNTLYLQMNSLRAGDTA VYYCARGKKGNAFDIWGQGTVVTVSS SEQ ID
ZKC33 NO. Amino acid sequence CDRH1 225 GDSISSRTFS CDRH2 226
IYYSGST CDRH3 227 ARRNAEFFSFWSYYGMDV CDRH3 228
QVQLQESGPGLVKPSQTLSLTCTVSGDSISS RTFSWSWIRQPPGKGLEWVGHIYYSGSTDYN
PSLKSRISISIDTSKNQFSLKLSSVTAADTA VYYCARRNAEFFSFWSYYGMDVWGHGTAVIV SS
SEQ ID ZKC34 NO. Amino acid sequence CDRH1 229 GGSINSGGYY CDRH2 230
ILHSGNT CDRH3 231 ARAGDYYSGYVPPEY CDRH3 232
QVQLQESGPGLVKPSQTLSLTCAVSGGSINS GGYYWSWVRQHPGKGLEWIGYILHSGNTNYN
PSLKSRVNIFVDTSENQFSLKLRSVTAADTA IYFCARAGDYYSGYVPPEYWGPGTLVTVSS SEQ
ID ZKD25 NO. Amino acid sequence CDRH1 233 GFTVSSNY CDRH2 234
IYSGGST CDRH3 235 ARFGGNPSFDY CDRH3 236
EVQLVESGGGLVQPGGSLRLSCAASGFTV SSNYMSWVRQAPGKGLEWVSVIYSGGSTY
YANSVKGRFTISRDKSKNTLYLQMNNLRA EDTAVYFCARFGGNPSFDYWGQGTLVTVSS SEQ ID
ZKA3 NO. Amino acid sequence CDRH1 237 GFIFSNYA CDRH2 238 IGGKGDSI
CDRH3 239 VKDLAVLESDRLEVDQ CDRH3 240 EVQLAESGGGLVQPGGSLRLSCSGSGFIFS
NYAMVWARQAPGKGLEYVSGIGGKGDSIYH IDSVKGRFTISRDNSKRTVYLQMSRLRTED
TAVYYCVKDLAVLESDRLEVDQWGQGTLVI VSA SEQ ID ZKA4 NO. Amino acid
sequence CDRH1 241 GFTFSSYV CDRH2 242 TSYDGSNK CDRH3 243
ARGPVPYWSGESYSGAYFDF CDRH3 244 QVQLVESGGGVVQPGRSLRLSCAASGFTFSS
YVMHWVRQAPGKGLEWVTVTSYDGSNKYYAD SVKGRFTISRDNAKNTLYLQMNSLRGEDTAI
TYYCARGPVPYWSGESYSGAYFDFWGQGILV VSS SEQ ID ZKA5 NO. Amino acid
sequence CDRH1 245 GFTFSNYY CDRH2 246 MSSSETIK CDRH3 247
ARSGIETVAGSIDYYGMDV CDRH3 248 QVQLVESGGGLVKPGGSLRLSCAGSGFTFSNY
YMTWIRQAPGKGLELVSYMSSSETIKYYADSV KGRFTISRDNAKNSLYLQMNSLRADDTARYYC
ARSGIETVAGSIDYYGMDVWGHGTPVTVSS SEQ ID ZKA6 NO. Amino acid sequence
CDRH1 249 DFTVSNYA CDRH2 250 VSYDGSNK CDRH3 251 ATGVTMFQGAQTNAEYLHY
CDRH3 252 QVHLVESGGGVVQPGRSLRLSCEASDFTVSN
YAMHWVRQAPGKGLEWVAVVSYDGSNKYYAD SVKGRFTISRDNSKNTLYLQMNSLRAEDTAL
YYCATGVTMFQGAQTNAEYLHYWGQGSLVTI SS SEQ ID ZKA7 NO. Amino acid
sequence CDRH1 253 GFTFSRYG CDRH2 254 VSGDGSST CDRH3 255
VKDFWSGDQSLESDF CDRH3 256 EVQLVESGGGLVQPGGSLRLSCSASGFTFSR
YGMVWARQAPGKGLEYLSGVSGDGSSTYYAN SVKGRFTISRDNSKNTLYLHMSRLRDEDTAM
YYCVKDFWSGDQSLESDFWGQGALVTVSS SEQ ID ZKA8 NO. Amino acid sequence
CDRH1 257 GFTFSAHA CDRH2 258 ISRNEDYT CDRH3 259 VKDFGTSPQTDF CDRH3
260 DERLVESGGGLVQPGGSLRLVCSASGFTFSA HAMHWVRQPPGKGLEYVSTISRNEDYTYYAD
SVKGRFTISRDNSKNSLYLQMRRLRPEDTAI YYCVKDFGTSPQTDFWGQGTLVAVSS SEQ ID
ZKA76 NO. Amino acid sequence CDRH1 261 GFTFSTYF CDRH2 262 ISSTGSYK
CDRH3 263 ARPFHSEYTYGLDAFDI CDRH3 264
EVQLVESGGGLVKPGGSLRLSCAASGFTFST YFMHWVRQAPGKGLEWVASISSTGSYKFYAD
SVKGRFTISRDNTKNSLFLQMNSLRAEDTAV FYCARPFHSEYTYGLDAFDIWGQGTMLTVSS SEQ
ID ZKA117 NO. Amino acid sequence CDRH1 265 GGSIRRTNSY CDRH2 266
ISYSGST CDRH3 267 ARLNDGSTVTTSSYFDY CDRH3 268
QLQLQESGPGLVKPSETLSLTCTVSGGSIRR TNSYWGWIRQTTGKGLQWIGSISYSGSTFYN
PSLKSRVTISLDTSKDHFSLELSSVTAADTA IYYCARLNDGSTVTTSSYFDYWGQGTLVTVS S
SEQ ID ZKB27 NO. Amino acid sequence CDRH1 269 GYSFTSSW CDRH2 270
IDPSDSYT CDRH3 271 ARHDYSVSENGMDV CDRH3 272
EVQLVQSGAEVKKPGESLRISCKASGYSFTS SWINWVRQMPGKGLEWMGRIDPSDSYTTYNP
SFQGHVTISVDKSIGTAYLQWNSLRASDTAM YYCARHDYSVSENGMDVWGQGTTVTVSS SEQ ID
ZKB29 NO. Amino acid sequence CDRH1 273 GFTFSSYT CDRH2 274 ISYDGSHK
CDRH3 275 ARRSYSISCFDY CDRH3 276 QVQLVESGGGVVQPGRSLRLSCAASGFTFSS
YTMHWVRQAPGKGLEWVAVISYDGSHKFYAD SVKGRFTISRDNSKDTLYLQMNSLRAEDTAL
YYCARRSYSISCFDYWGQGTLVTISS SEQ ID ZKB34 NO. Amino acid sequence
CDRH1 277 GFTFSRSG CDRH2 278 VSYDGSNK CDRH3 279 AKDLTMVRGVHYYYYVMDV
CDRH3 280 QVQLVESGGGVVQPGRSLRLSCAASGFTFSR
SGMHWVRQAPGKGLEWVAVVSYDGSNKYYSD SVKGRFTISRDNSKNTLYLQMNSLRVEDTAV
YYCAKDLTMVRGVHYYYYVMDVWGQGTTVTV SS SEQ ID ZKB39 NO. Amino acid
sequence CDRH1 281 GYTFDDYY CDRH2 282 INPHRGGT CDRH3 283
VRDQYCDGGNCYGIHQPHYGMDV CDRH3 284 QVQLVQSGAEVKKPGASLKVSCKASGYTFD
DYYIHWVRQAPGQGLEWLGRINPHRGGTNY
AQKFQGRVIMTLDMSISTTYMELRRITSDD AAVYYCVRDQYCDGGNCYGIHQPHYGMDVW
GQGTTVTVSS SEQ ID ZKB46 NO. Amino acid sequence CDRH1 285 GYSFTSYW
CDRH2 286 IDPSDSYT CDRH3 287 ARREYSSSSGQEDWFDP CDRH3 288
EVQLVQSGAEVKKPGESLRISCKGSGYSFT SYWISWVRQMPGKGLEWMGRIDPSDSYTNY
SPSFQGHVTISADKSISTAYLQWSSLKASD TAMYYCARREYSSSSGQEDWFDPWGQGTLV TVSS
SEQ ID ZKB53 NO. Amino acid sequence CDRH1 289 GFTFSSYA CDRH2 290
ISYDGSNR CDRH3 291 ARHVEQLPSSGYFQH CDRH3 292
QVQLVESGGGVVQPGRSLRLSCAASGFTFS SYAMHWVRQTPGKGLEWVTVISYDGSNRYY
ADSVKGRFTISRDNSKNTLYLQMNSLRSED TAVYYCARHVEQLPSSGYFQHWGQGTLVTV SS
SEQ ID ZKC26 NO. Amino acid sequence CDRH1 293 GFIFSDFY CDRH2 294
IGHDGSYI CDRH3 295 ARAHGGFRH CDRH3 296
QVQVVESGGGLVKPGGSLRLSCAASGFIFSD FYMSWMRQAPGKGLEWVAYIGHDGSYILYAD
SVKGRFTISRDNAKNSLFLRMNSLRVEDTAV YYCARAHGGFRHWGQGTVVAVSP SEQ ID ZKD5
NO. Amino acid sequence CDRH1 297 GFTFTSYG CDRH2 298 ISYDGSNK CDRH3
299 ARDRDHYDLWNAYTFDY CDRH3 300 QVQLVESGGGVVQPGRSLRLSCAASGFTF
TSYGMHWVRQTPGKGLDWVAVISYDGSNK YYADSVKGRFTISRDNSKDTLYLQMNSLR
AADTALYYCARDRDHYDLWNAYTFDYWGQ GTLVTVSS SEQ ID ZKD7 NO. Amino acid
sequence CDRH1 301 GFTFSNYA CDRH2 302 ISYDVSDK CDRH3 303
AGGPLGVVVIKPSNAEHFHH CDRH3 304 QVQLVESGGGVVQPGRSLRLSCAASGFTF
SNYAMHWVRQAPGKGLEWVAVISYDVSDK YYADSVKGRFTISRDNSKNTLFLQMNSLR
AEDTAAYYCAGGPLGVVVIKPSNAEHFHH WGQGTLVTVSS SEQ ID ZKD8 NO. Amino
acid sequence CDRH1 305 GFTFINYA CDRH2 306 ISYDGSNK CDRH3 307
ATDADAYGDSGANFHY CDRH3 308 QVQLVESGGGVVQPGKSLRLSCAASGFTFI
NYAIHWVRQAPGKGLEWVAVISYDGSNKFY TDSVKGRFTISRDNSKNTLYLQMNSLRADD
TAVYYCATDADAYGDSGANFHYWGQGTLVT VSS SEQ ID ZKD15 NO. Amino acid
sequence CDRH1 309 DASISSGGFS CDRH2 310 IYSSGDT CDRH3 311
ARAHTPTSKFYYYYAMDV CDRH3 312 QLQLQESGSGLVKPSQTLSLTCTVSDASIS
SGGFSWSWIRQPLGKGLEWLGYIYSSGDTF YNPSLQGRVTMSVDIFRSQFSLKLTSVTAA
DTAMYYCARAHTPTSKFYYYYAMDVWGQGT TVTVSS SEQ ID ZKD16 NO. Amino acid
sequence CDRH1 313 GFTFSDHF CDRH2 314 SRNKPNSYTT CDRH3 315
AKVGGCYGGDCHVENDY CDRH3 316 EVQLVESGGDLVQPGGSLRLSCVASGFTFSD
HFMDWVRQAPGKGLEWVGRSRNKPNSYTTEY AASVKGRFSISRDDSKKALYLQMNSLQTEDT
AVYYCAKVGGCYGGDCHVENDYWGQGTLVT VSS SEQ ID ZKD17 NO. Amino acid
sequence CDRH1 317 GFIFSDYA CDRH2 318 ISYDGSSR CDRH3 319
ARGYCSSGTCFSTNAEYFHP CDRH3 320 QVQMVESGGGVVQPGTSLRLSCATSGFIFSD
YAMHWVRQAPGKGLEWVAVISYDGSSRLYAD SVKGRFTVSRDNSKNTLYLQMHSLRAGDTAV
YYCARGYCSSGTCFSTNAEYFHPWGQGTLA TISS SEQ ID ZKD20 NO. Amino acid
sequence CDRH1 321 GFTFSDHF CDRH2 322 SRNKPNSYTT CDRH3 323
ARVGGCNGGDCHVENDY CDRH3 324 EVQLVESGGGLVQPGGSLRLSCVASGFTFSD
HFMDWVRQAPGKGLEWVGRSRNKPNSYTTEY AASVKGRFTISRDDSKNSLYLQMNSLQTEDT
AVYYCARVGGCNGGDCHVENDYWGQGTLVT VSS SEQ ID ZKA134 NO. Amino acid
sequence CDRH1 325 GGTFSAYA CDRH2 326 IIPFFGTA CDRH3 327
ARSDIVSTTRGYHHYGMDV CDRH3 328 QVHLVQSGAEVKKPGSSVNVSCKASGGTFS
AYAISWVRQAPGQGLEWMGGIIPFFGTAYY MAQKFKGRVTVTADKSTSTVYMETSLRSED
TAVYYCARSDIVSTTRGYHHYGMDVWGQGT TVTVSS SEQ ID ZKA246 NO. Amino acid
sequence CDRH1 329 GYTFSDYY CDRH2 330 INPYSGGT CDRH3 331
ARGFTMISDREFDP CDRH3 332 QVQLVQSGAEVKRPGASVKVSCKASGYTFSD
YYMHWVRQAPGQGLEWMGRINPYSGGTNYAQ KFHGRVTVTRDTSISTVYMELRGLRSDDTAV
YYCARGFTMISDREFDPWGQGTLVTVSS SEQ ID ZKA256 NO. Amino acid sequence
CDRH1 333 GFTFSTYW CDRH2 334 IKQDGSEK CDRH3 335
ARDPGYDDFWSGSYSGSFDI CDRH3 336 EVQLVESGGGLVQPGGSLRLSCAASGFTFSTY
WMTWVRQAPGKGLEWVANIKQDGSEKYYVDSV KGRFTISRDNTKNSLYLQVNSLRAEDTAIYYC
ARDPGYDDFWSGSYSGSFDIWGQGTMVTVSS SEQ ID ZKB42 NO. Amino acid
sequence CDRH1 337 GFTFNNYG CDRH2 338 ISYDGNKK CDRH3 339
VKYGERINGYSDPFDH CDRH3 340 QVQVVESGGGVVQPGRSLRLFCAASGFTFNNY
GMHWVRQAPGKGLEWVALISYDGNKKYYADSV KGRFSISRDNSKNTLYLQMNRLRSGDTAVYHC
VKYGERINGYSDPFDHWGQGTLVTVSS SEQ ID ZKB85 NO. Amino acid sequence
CDRH1 341 GYTFTTYA CDRH2 342 INTNTGNP CDRH3 343 ARVIVPYAFDI CDRH3
344 QVQLVQSGSELKKPGASVKVSCKASGYTFTT YAMNWVRQAPGQGPEWVGWINTNTGNPTYAQ
GFTGRFVLSLDTSVSTAFLQISSLKAEDTAV YYCARVIVPYAFDIWGQGTMVTVSS SEQ ID
ZKB47 NO. Amino acid sequence CDRH1 345 GYTFTNYY CDRH2 346 INPSGGPT
CDRH3 347 ARDQYGGYARYGMDV CDRH3 348
QVQLVQSGAEVKKPGASVKVSCQASGYTFTN
YYMHWVRQAPGQGLEWMGIINPSGGPTSYAQ KFQGRVTMTTDTSTSTVYMELSSLRSEDTAV
YYCARDQYGGYARYGMDVWGQGTTVIVSS SEQ ID ZKC6 NO. Amino acid sequence
CDRH1 349 GYTFTGYY CDRH2 350 INPNSGGT CDRH3 351 ARVSDWGFAFDI CDRH3
352 QVQLVQSGTEVKKPGASVKVSCKASGYTF TGYYMHWVRQAPGQGLEWMGRINPNSGGT
NYAQKFQGRVTMTRDTSISTAYMELSGLR SDDTAVYYCARVSDWGFAFDIWGQGTMVT VSQ SEQ
ID ZKA160 NO. Amino acid sequence CDRH1 353 GGSITSYS CDRH2 354
IFYSGST CDRH3 355 ARDQTMPVWVGGMDV CDRH3 356
QVQLQESGPGLVKPSETLSLTCTVSGGSI TSYSWSWIRQPPGKGLEWIGYIFYSGSTD
YNPSLKSRVTISVDTSKDQFSLRLRSVTA ADTAVYYCARDQTMPVWVGGMDVWGQGTT VTVSS
SEQ ID ZKA172 NO. Amino acid sequence CDRH1 357 GYIFTRYW CDRH2 358
IDPSDSYT CDRH3 359 ARQETAREDGMAV CDRH3 360
EVQLVQSGAEVKKPGKSLRISCKGSGYIFT RYWISWVRQMPGKGLEWMGRIDPSDSYTNYS
PSFQGHVTISADKSISTAYLQWSSLKASDTA MYYCARQETAREDGMAVWGQGTTVIVSS SEQ ID
ZKA174 NO. Amino acid sequence CDRH1 361 GGSMSNSYYH CDRH2 362
IYYSGST CDRH3 363 ARNPVFNPLTLTHDAFDI CDRH3 364
QLQLQESGPGLVKPSETLSLTCTVSGGSMS NSYYHWGWIRQPPGKGLEWIGSIYYSGSTY
YNPSLKSRVTISVDTSKNQFSLKLNSVTAA DTAVYYCARNPVFNPLTLTHDAFDIWGQGT
MVTVSS SEQ ID ZKA189 NO. Amino acid sequence CDRH1 365 GFTFSSYA
CDRH2 366 ISGSGDNT CDRH3 367 AKWPYYDFWSGSESYFDP CDRH3 368
GVQLLESGGALVQPGKSLRLSCAASGFTF SSYALTWVRQAPGKGLQWVSAISGSGDNT
YYADSVKGRFTISRDNSKNTLYLQMNSLR AEDTAVYYCAKWPYYDFWSGSESYFDPWG
QGTLVTVSS SEQ ID ZKA195 NO. Amino acid sequence CDRH1 369 GYNFPSYW
CDRH2 370 IDPSDSYT CDRH3 371 ARADCRSTSCYLVFE CDRH3 372
EVQLVQSGAEVKKPGESLRISCKDSGYNFP SYWIHWVRQMPGKGLEWMGTIDPSDSYTNY
SPSFQGHVTISADKSISTAYLQWSSLKASD TAMYYCARADCRSTSCYLVFEGQGTLVTVSS SEQ
ID ZKA215 NO. Amino acid sequence CDRH1 373 GYTFTSYW CDRH2 374
IDPSDSHT CDRH3 375 ARHALPNYFDS CDRH3 376
EVQLVQSGAEVKKPGESLRISCKGSGYTFT SYWISWVRQMPGKGLEWMGRIDPSDSHTD
YSPSFQGHVTISADKSISAAYLQWSSLKAS DTAMYYCARHALPNYFDSWGQGTLVTVSS SEQ ID
ZKA218 NO. Amino acid sequence CDRH1 377 GFPFSSYW CDRH2 378
INSDGRNT CDRH3 379 ARGGYDYDSSGCFDY CDRH3 380
EVQLVESGGGLVQPGGSLRLSCAASGFPFSS YWMHWVRQAPGKGLVWVSRINSDGRNTNYAD
SVKGRFTISRDNAENTVYLQMNSLRAEDTAV YYCARGGYDYDSSGCFDYWGQGTLVTVSS SEQ
ID ZKB75 NO. Amino acid sequence CDRH1 381 GFTFSNYA CDRH2 382
ISGTGGST CDRH3 383 AKDSASRGGYCSGGVCYLNPGHHDY CDRH3 384
EVQVLESGGGLLQPGGSLRLSCAASGFTFSN YAMSWVRQAPGKGLEWVSTISGTGGSTYYAD
SVKGRFTISRDNSKNTLYLQMNSLRAEDTAV YYCAKDSASRGGYCSGGVCYLNPGHHDYWG
QGTLVTVSS SEQ ID ZKB83 NO. Amino acid sequence CDRH1 385 GYSFTNYW
CDRH2 386 IDPSDSYT CDRH3 387 ARLRGSLYCSGGRCYSVPGETPNWFDP CDRH3 388
EVQLVQSGAEVKKPGESLRISCKGSGYSFT NYWITWVRQMPGKGLEWMGSIDPSDSYTNY
SPSFQGHVTISADWSINTAYLQWSSLKASD TAKYYCARLRGSLYCSGGRCYSVPGETPNW
FDPWGQGTLVTVSS SEQ ID ZKC3 NO. Amino acid sequence CDRH1 389
GGSITSYY CDRH2 390 IYYSGST CDRH3 391 ARVGGAPYYYYGMDV CDRH3 392
QVQLQESGPGLVKPSETLSLTCTVSGGSITS YYWSWIRQPPGKGLEWIGYIYYSGSTNYNPS
LKSRVTISVDTSKNQFSLKLSSVTAADTAVY YCARVGGAPYYYYGMDVWGQGTTVTVSS SEQ ID
ZKC18 NO. Amino acid sequence CDRH1 393 GFTFGDYA CDRH2 394
IRSKAYGGTT CDRH3 395 SRDHTGTTYAFDI CDRH3 396
EVQLVESGGGLVQPGRSLRLSCTASGFTFGD YAMSWFRQAPGKGLEWVGFIRSKAYGGTTEY
AASVKGRFTISRDDSKSIAYLQMNSLKTEDT AVYYCSRDHTGTTYAFDIWGQGTMVTVSQ SEQ
ID ZKD1 NO. Amino acid sequence CDRH1 397 GFTFSSYG CDRH2 398
IWYDGSNK CDRH3 399 ARDRRGYGDYVGYYYGMDV CDRH3 400
QVQLVESGGGVVQPGRSLRLSCAASGFTFS SYGMHWVRQAPGKGLEWVAVIWYDGSNKYY
ADSVKGRFTISRDNSKNTLYLQMNSLRAED TAVYYCARDRRGYGDYVGYYYGMDVWGQGT
TVTVSS
TABLE-US-00026 SEQ ID Name NO. Amino acid sequence ZIKV 401
TAAFTFTKXPAEXXHGTVT VEXQYXGXDGPCKXPXQM AVDXQTLTPVGRLITANP
VITEXTENSKMMLELDPP FGDSYIVIGXGXKKITHH WHRS EDIII 402
TAAFTFTKIPAETLHGTV generic TVEVQYAGTDGPCKVPAQ ZIKV
MAVDMQTLTPVGRLITAN H/PF/ PVITESTENSKMMLELDP 2013 PFGDSYIVIGVGEKKITH
EDIII HWHRS ZIKV-NS1 403 TGGAGTTCAACTGACGGTCG forward primer ZIKV-
404 TACCCCGAACCCATGATCCT NS1- reverse primer Gapdh- 405
GGCAAGTTCAAAGGCACAGTC forward primer Gapdh- 406 CACCAGCATCACCCCATTT
reverse primer ZIKV 407 X.sub.1GX.sub.2X.sub.3YSLCTAAFTFTKX.sub.4
EDIII PAEX.sub.5X.sub.6HGTVTVEX.sub.7QYX.sub.8G generic
X.sub.9DGPCKX.sub.10PX.sub.11QMAVDX.sub.12
QTLTPVGRLITANPVITEX.sub.13 TX.sub.14NSKMMLELDPPFGDSYI
VIGX.sub.15GX.sub.16X.sub.17KITHHWHR SG wherein X.sub.1 may be any
(naturally occurring) amino acid, preferably K, A, or E; X.sub.2
may be any (naturally occurring) amino acid, preferably V, F, or L;
X.sub.3 may be any (naturally occurring) amino acid, preferably S
or F; X.sub.4 may be any (naturally occurring) amino acid,
preferably I or V; X.sub.5 may be any (naturally occurring) amino
acid, preferably T or V; X.sub.6 may be any (naturally occurring)
amino acid, preferably L or D; X.sub.7 may be any (naturally
occurring) amino acid, preferably V or G; X.sub.8 may be any
(naturally occurring) amino acid, preferably A or G; X.sub.9 may be
any (naturally occurring) amino acid except R, preferably T or A;
X.sub.10 may be any (naturally occurring) amino acid, preferably V
or I; X.sub.11 may be any (naturally occurring) amino acid,
preferably A or V; X.sub.12 may be any (naturally occurring) amino
acid, preferably M or T; X.sub.13 may be any (naturally occurring)
amino acid, preferably S or G; X.sub.14 may be any (naturally
occurring) amino acid, preferably E or K; X.sub.15 may be any
(naturally occurring) amino acid, preferably V or I; X.sub.16 may
be any (naturally occurring) amino acid, preferably E, A, K, or D;
and X.sub.17 may be any (naturally occurring) amino acid,
preferably E, A, or K, more preferably K or A
* the sequences highlighted in bold are CDR regions (nucleotide or
aa) and the underlined residues are mutated residues as compared to
the "germline" sequence.
Sequence CWU 1 SEQUENCE LISTING <160> NUMBER OF SEQ ID
NOS: 414 <210> SEQ ID NO 1 <211> LENGTH: 8 <212>
TYPE: PRT <213> ORGANISM: Artificial Sequence <220>
FEATURE: <223> OTHER INFORMATION: ZKA190 CDRH1 <400>
SEQUENCE: 1 Gly Phe Thr Phe Ser Lys Tyr Gly 1 5 <210> SEQ ID
NO 2 <211> LENGTH: 8 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKA190 CDRH2 <400> SEQUENCE: 2 Ile Ser Tyr
Glu Gly Ser Asn Lys 1 5 <210> SEQ ID NO 3 <211> LENGTH:
23 <212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA190 CDRH3
<400> SEQUENCE: 3 Ala Lys Ser Gly Thr Gln Tyr Tyr Asp Thr Thr
Gly Tyr Glu Tyr Arg 1 5 10 15 Gly Leu Glu Tyr Phe Gly Tyr 20
<210> SEQ ID NO 4 <211> LENGTH: 7 <212> TYPE: PRT
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA190 CDRL1 <400> SEQUENCE: 4
Gln Ser Val Ser Ser Ser Tyr 1 5 <210> SEQ ID NO 5 <400>
SEQUENCE: 5 000 <210> SEQ ID NO 6 <211> LENGTH: 9
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA190 CDRL2
long <400> SEQUENCE: 6 Leu Ile Tyr Asp Ala Ser Ser Arg Ala 1
5 <210> SEQ ID NO 7 <211> LENGTH: 9 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA190 CDRL3 <400> SEQUENCE: 7
Gln Gln Tyr Gly Arg Ser Arg Trp Thr 1 5 <210> SEQ ID NO 8
<211> LENGTH: 130 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA190 VH <400> SEQUENCE: 8 Gln Val Gln Leu Val
Glu Ser Gly Gly Gly Val Val Gln Pro Gly Arg 1 5 10 15 Ser Leu Arg
Leu Ser Cys Ala Ala Ser Gly Phe Thr Phe Ser Lys Tyr 20 25 30 Gly
Met His Trp Val Arg Gln Ala Pro Gly Lys Gly Leu Glu Trp Val 35 40
45 Ala Val Ile Ser Tyr Glu Gly Ser Asn Lys Tyr Tyr Ala Asp Ser Val
50 55 60 Lys Gly Arg Phe Thr Ile Ser Arg Asp Asn Ser Lys Asn Thr
Leu Tyr 65 70 75 80 Leu Gln Met Asn Ser Leu Arg Ala Glu Asp Thr Ala
Val Tyr Tyr Cys 85 90 95 Ala Lys Ser Gly Thr Gln Tyr Tyr Asp Thr
Thr Gly Tyr Glu Tyr Arg 100 105 110 Gly Leu Glu Tyr Phe Gly Tyr Trp
Gly Gln Gly Thr Leu Val Thr Val 115 120 125 Ser Ser 130 <210>
SEQ ID NO 9 <211> LENGTH: 108 <212> TYPE: PRT
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA190 VL <400> SEQUENCE: 9
Glu Ile Val Leu Thr Gln Ser Pro Gly Thr Leu Ser Leu Ser Pro Gly 1 5
10 15 Glu Arg Ala Thr Leu Ser Cys Arg Ala Ser Gln Ser Val Ser Ser
Ser 20 25 30 Tyr Leu Ala Trp Tyr Gln Gln Lys Arg Gly Gln Ala Pro
Arg Leu Leu 35 40 45 Ile Tyr Asp Ala Ser Ser Arg Ala Thr Gly Ile
Pro Asp Arg Phe Ser 50 55 60 Gly Ser Gly Ser Gly Thr Asp Phe Thr
Leu Thr Ile Ser Arg Leu Glu 65 70 75 80 Pro Glu Asp Phe Ala Val Tyr
Tyr Cys Gln Gln Tyr Gly Arg Ser Arg 85 90 95 Trp Thr Phe Gly Gln
Gly Thr Lys Val Glu Ile Lys 100 105 <210> SEQ ID NO 10
<211> LENGTH: 24 <212> TYPE: DNA <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA190 CDRH1 <400> SEQUENCE: 10 ggattcacct
tcagtaaata tggc 24 <210> SEQ ID NO 11 <211> LENGTH: 24
<212> TYPE: DNA <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA190 CDRH2
<400> SEQUENCE: 11 atatcatatg agggaagtaa taaa 24 <210>
SEQ ID NO 12 <211> LENGTH: 69 <212> TYPE: DNA
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA190 CDRH3 <400> SEQUENCE:
12 gcgaaatcgg ggacccaata ctatgatact actggttatg agtatagggg
tttggaatac 60 tttggctac 69 <210> SEQ ID NO 13 <211>
LENGTH: 21 <212> TYPE: DNA <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA190
CDRL1 <400> SEQUENCE: 13 cagagtgtta gtagcagtta c 21
<210> SEQ ID NO 14 <400> SEQUENCE: 14 000 <210>
SEQ ID NO 15 <211> LENGTH: 27 <212> TYPE: DNA
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA190 CDRL2 long <400>
SEQUENCE: 15 ctcatctatg atgcatccag cagggcc 27 <210> SEQ ID NO
16 <211> LENGTH: 27 <212> TYPE: DNA <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKA190 CDRL3 <400> SEQUENCE: 16 cagcagtatg
gtaggtcaag gtggaca 27 <210> SEQ ID NO 17 <211> LENGTH:
391 <212> TYPE: DNA <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA190 VH
<400> SEQUENCE: 17 caggtgcagc tggtggagtc tgggggaggc
gtggtccagc ctgggaggtc cctgagactc 60 tcctgtgcag cctctggatt
caccttcagt aaatatggca tgcactgggt ccgccaggct 120 ccaggcaagg
ggctggagtg ggtggcagtt atatcatatg agggaagtaa taaatattat 180
gcagactccg tgaagggccg attcaccatc tccagagaca attccaagaa cacgctgtat
240 ctgcaaatga acagcctgag agctgaggac acggcagtgt attactgtgc
gaaatcgggg 300 acccaatact atgatactac tggttatgag tataggggtt
tggaatactt tggctactgg 360 ggccagggaa ccctggtcac cgtctcctca g 391
<210> SEQ ID NO 18 <211> LENGTH: 325 <212> TYPE:
DNA <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA190 VL <400> SEQUENCE: 18
gaaattgtgt tgacgcagtc tccaggcacc ctgtctttgt ctccagggga aagagccacc
60 ctctcctgca gggccagtca gagtgttagt agcagttact tagcctggta
ccagcagaaa 120 cgtggccagg ctcccaggct cctcatctat gatgcatcca
gcagggccac tggcatccca 180 gacaggttca gtggcagtgg gtctgggaca
gacttcactc tcaccatcag cagactggag 240 cctgaagatt ttgcagtgta
ttactgtcag cagtatggta ggtcaaggtg gacattcggc 300 caagggacca
aggtggaaat caaac 325 <210> SEQ ID NO 19 <211> LENGTH: 8
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA185 CDRH1
<400> SEQUENCE: 19 Gly Tyr Ser Phe Thr Ser Tyr Trp 1 5
<210> SEQ ID NO 20 <211> LENGTH: 8 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA185 CDRH2 <400> SEQUENCE:
20 Phe Asp Pro Ser Asp Ser Gln Thr 1 5 <210> SEQ ID NO 21
<211> LENGTH: 14 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA185 CDRH3 <400> SEQUENCE: 21 Ala Arg Arg Tyr
Cys Ser Ser Ser Ser Cys Tyr Val Asp Asn 1 5 10 <210> SEQ ID
NO 22 <211> LENGTH: 6 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKA185 CDRL1 <400> SEQUENCE: 22 Ala Leu
Pro Asn Lys Phe 1 5 <210> SEQ ID NO 23 <400> SEQUENCE:
23 000 <210> SEQ ID NO 24 <211> LENGTH: 9 <212>
TYPE: PRT <213> ORGANISM: Artificial Sequence <220>
FEATURE: <223> OTHER INFORMATION: ZKA185 CDRL2 long
<400> SEQUENCE: 24 Val Ile Tyr Glu Asp Asn Lys Arg Pro 1 5
<210> SEQ ID NO 25 <211> LENGTH: 12 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA185 CDRL3 <400> SEQUENCE:
25 Tyr Ser Thr Asp Ser Ser Ser Asn Pro Leu Gly Val 1 5 10
<210> SEQ ID NO 26 <211> LENGTH: 121 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA185 VH <400> SEQUENCE: 26
Glu Val Gln Leu Val Gln Ser Gly Ala Glu Val Lys Lys Pro Gly Glu 1 5
10 15 Ser Leu Arg Ile Ser Cys Lys Gly Ser Gly Tyr Ser Phe Thr Ser
Tyr 20 25 30 Trp Ile Thr Trp Val Arg Gln Met Pro Gly Lys Gly Leu
Glu Trp Met 35 40 45 Ala Lys Phe Asp Pro Ser Asp Ser Gln Thr Asn
Tyr Ser Pro Ser Phe 50 55 60 Gln Gly His Val Thr Ile Ser Val Asp
Lys Ser Ile Ser Thr Ala Tyr 65 70 75 80 Leu Gln Trp Ser Ser Leu Lys
Ala Ser Asp Thr Ala Met Tyr Tyr Cys 85 90 95 Ala Arg Arg Tyr Cys
Ser Ser Ser Ser Cys Tyr Val Asp Asn Trp Gly 100 105 110 Gln Gly Thr
Leu Val Thr Ile Phe Ser 115 120 <210> SEQ ID NO 27
<211> LENGTH: 109 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA185 VL <400> SEQUENCE: 27 Ser Tyr Glu Leu Thr
Gln Pro Pro Ser Val Ser Val Ser Pro Gly Gln 1 5 10 15 Thr Ala Arg
Ile Thr Cys Ser Gly Asp Ala Leu Pro Asn Lys Phe Ala 20 25 30 Tyr
Trp Tyr Arg Gln Lys Ser Gly Gln Ala Pro Val Leu Val Ile Tyr 35 40
45 Glu Asp Asn Lys Arg Pro Ser Gly Ile Pro Glu Arg Phe Ser Gly Ser
50 55 60 Ser Ser Gly Thr Met Ala Thr Leu Thr Ile Ser Gly Ala Gln
Val Glu 65 70 75 80 Asp Glu Ala Asp Tyr His Cys Tyr Ser Thr Asp Ser
Ser Ser Asn Pro 85 90 95 Leu Gly Val Phe Gly Gly Gly Thr Lys Leu
Thr Val Leu 100 105 <210> SEQ ID NO 28 <211> LENGTH: 24
<212> TYPE: DNA <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA185 CDRH1
<400> SEQUENCE: 28 ggatatagtt ttaccagtta ctgg 24 <210>
SEQ ID NO 29 <211> LENGTH: 24 <212> TYPE: DNA
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA185 CDRH2 <400> SEQUENCE:
29 tttgatccta gtgactctca aacc 24 <210> SEQ ID NO 30
<211> LENGTH: 42 <212> TYPE: DNA <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA185 CDRH3 <400> SEQUENCE: 30 gcgagaagat
attgtagtag tagtagttgt tatgtggaca at 42 <210> SEQ ID NO 31
<211> LENGTH: 18 <212> TYPE: DNA <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA185 CDRL1 <400> SEQUENCE: 31 gcattgccaa
ataaattt 18 <210> SEQ ID NO 32 <400> SEQUENCE: 32 000
<210> SEQ ID NO 33 <211> LENGTH: 27 <212> TYPE:
DNA <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA185 CDRL2 long <400>
SEQUENCE: 33 gtcatctatg aggacaacaa acgaccc 27 <210> SEQ ID NO
34 <211> LENGTH: 36 <212> TYPE: DNA <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKA185 CDRL3 <400> SEQUENCE: 34 tactcaacag
acagcagttc taatcccctg ggagta 36 <210> SEQ ID NO 35
<211> LENGTH: 364 <212> TYPE: DNA <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA185 VH <400> SEQUENCE: 35 gaagtgcagc
tggtgcagtc cggagcagag gtgaaaaagc ccggggagtc tctgaggatc 60
tcctgtaagg gttctggata tagttttacc agttactgga tcacctgggt gcgccagatg
120 cccgggaaag gcctggagtg gatggcgaag tttgatccta gtgactctca
aaccaactac 180 agcccgtcct tccaaggcca cgtcaccatc tcagttgaca
agtccatcag cactgcctac 240 ttgcagtgga gcagcctgaa ggcctcggac
accgccatgt attactgtgc gagaagatat 300 tgtagtagta gtagttgtta
tgtggacaat tggggccagg gaaccctggt caccatcttc 360 tcag 364
<210> SEQ ID NO 36 <211> LENGTH: 328 <212> TYPE:
DNA <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA185 VL <400> SEQUENCE: 36
tcctatgagc tgacacagcc accctcggtg tcagtgtccc caggacaaac ggccaggatc
60 acctgctctg gagatgcatt gccaaataaa tttgcttatt ggtaccggca
gaagtcaggc 120 caggcccctg ttctggtcat ctatgaggac aacaaacgac
cctccgggat ccctgagaga 180 ttctctggct ccagctcagg gacaatggcc
accttgacta tcagtggggc ccaggtggag 240 gatgaagctg actaccactg
ttactcaaca gacagcagtt ctaatcccct gggagtattc 300 ggcggaggga
ccaagctgac cgtcctag 328 <210> SEQ ID NO 37 <211>
LENGTH: 8 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA230
CDRH1 <400> SEQUENCE: 37 Gly Gly Ser Ile Ser Ser Asp Tyr 1 5
<210> SEQ ID NO 38 <211> LENGTH: 7 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA230 CDRH2 <400> SEQUENCE:
38 Ile Tyr Tyr Ser Gly Ser Thr 1 5 <210> SEQ ID NO 39
<211> LENGTH: 17 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA230 CDRH3 <400> SEQUENCE: 39 Ala Arg Arg Arg
Lys Tyr Asp Ser Leu Trp Gly Ser Phe Ala Phe Asp 1 5 10 15 Ile
<210> SEQ ID NO 40 <211> LENGTH: 8 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA230 CDRL1 <400> SEQUENCE:
40 Ser Ser Asn Ile Gly Gly Asn Tyr 1 5 <210> SEQ ID NO 41
<400> SEQUENCE: 41 000 <210> SEQ ID NO 42 <211>
LENGTH: 9 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA230
CDRL2 long <400> SEQUENCE: 42 Leu Ile Cys Ile Asn Asp His Arg
Pro 1 5 <210> SEQ ID NO 43 <211> LENGTH: 11 <212>
TYPE: PRT <213> ORGANISM: Artificial Sequence <220>
FEATURE: <223> OTHER INFORMATION: ZKA230 CDRL3 <400>
SEQUENCE: 43 Ala Thr Trp Asp Asp Ser Leu Gly Gly Leu Val 1 5 10
<210> SEQ ID NO 44 <211> LENGTH: 123 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA230 VH <400> SEQUENCE: 44
Gln Val Gln Leu Gln Glu Ser Gly Pro Gly Leu Val Lys Pro Ser Glu 1 5
10 15 Thr Leu Ser Leu Thr Cys Ala Val Ser Gly Gly Ser Ile Ser Ser
Asp 20 25 30 Tyr Trp Ser Trp Ile Arg Gln Pro Pro Gly Lys Gly Leu
Glu Trp Ile 35 40 45 Gly Tyr Ile Tyr Tyr Ser Gly Ser Thr Asn Tyr
Asn Pro Ser Leu Lys 50 55 60 Ser Arg Val Thr Ile Ser Val Asp Thr
Ser Lys Asn His Phe Ser Leu 65 70 75 80 Lys Leu Asn Ser Val Thr Ala
Ala Asp Thr Ala Val Tyr Tyr Cys Ala 85 90 95 Arg Arg Arg Lys Tyr
Asp Ser Leu Trp Gly Ser Phe Ala Phe Asp Ile 100 105 110 Trp Gly Gln
Gly Thr Met Val Thr Val Ser Ser 115 120 <210> SEQ ID NO 45
<211> LENGTH: 110 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA230 VL <400> SEQUENCE: 45 Gln Ser Val Leu Thr
Gln Pro Pro Ser Ala Ser Gly Thr Pro Gly Gln 1 5 10 15 Arg Val Thr
Ile Ser Cys Ser Gly Ser Ser Ser Asn Ile Gly Gly Asn 20 25 30 Tyr
Val Tyr Trp Tyr Gln Gln Leu Pro Gly Thr Ala Pro Lys Leu Leu 35 40
45 Ile Cys Ile Asn Asp His Arg Pro Ser Gly Val Pro Asp Arg Phe Ser
50 55 60 Gly Ser Lys Ser Gly Thr Ser Ala Ser Leu Ala Ile Ser Gly
Leu Gln 65 70 75 80 Ser Glu Asp Glu Ala Asp Tyr Tyr Cys Ala Thr Trp
Asp Asp Ser Leu 85 90 95 Gly Gly Leu Val Phe Gly Gly Gly Thr Lys
Leu Thr Val Leu 100 105 110 <210> SEQ ID NO 46 <211>
LENGTH: 24 <212> TYPE: DNA <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA230
CDRH1 <400> SEQUENCE: 46 ggtggctcca tcagtagtga ctac 24
<210> SEQ ID NO 47 <211> LENGTH: 21 <212> TYPE:
DNA <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA230 CDRH2 <400> SEQUENCE:
47 atctattaca gtgggagcac c 21 <210> SEQ ID NO 48 <211>
LENGTH: 51 <212> TYPE: DNA <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA230
CDRH3 <400> SEQUENCE: 48 gcgaggagga ggaagtatga ttccctttgg
gggagttttg cttttgatat c 51 <210> SEQ ID NO 49 <211>
LENGTH: 24 <212> TYPE: DNA <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA230
CDRL1 <400> SEQUENCE: 49 agctccaaca tcggaggtaa ttat 24
<210> SEQ ID NO 50 <400> SEQUENCE: 50 000 <210>
SEQ ID NO 51 <211> LENGTH: 27 <212> TYPE: DNA
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA230 CDRL2 long <400>
SEQUENCE: 51 ctcatctgta ttaatgatca ccggccc 27 <210> SEQ ID NO
52 <211> LENGTH: 33 <212> TYPE: DNA <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKA230 CDRL3 <400> SEQUENCE: 52 gcaacatggg
atgacagcct gggtggcctt gta 33 <210> SEQ ID NO 53 <211>
LENGTH: 370 <212> TYPE: DNA <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA230
VH <400> SEQUENCE: 53 caggtgcagc tgcaggagtc gggcccaggc
ctggtgaagc cttcggagac cctgtccctc 60 acctgcgcag tctctggtgg
ctccatcagt agtgactact ggagctggat ccggcagccc 120 ccagggaagg
gactggagtg gattgggtat atctattaca gtgggagcac caactacaac 180
ccctccctca agagtcgagt caccatatca gtagacacgt ccaagaacca cttctccctg
240 aagctgaact ctgtgaccgc tgcggacacg gccgtgtatt actgtgcgag
gaggaggaag 300 tatgattccc tttgggggag ttttgctttt gatatctggg
gccaagggac aatggtcacc 360 gtctcttcag 370 <210> SEQ ID NO 54
<211> LENGTH: 331 <212> TYPE: DNA <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA230 VL <400> SEQUENCE: 54 cagtctgtgc
tgactcagcc accctcagcg tctgggaccc ccgggcagag ggtcaccatc 60
tcttgttctg gaagcagctc caacatcgga ggtaattatg tatactggta ccagcagctc
120 ccaggaacgg cccccaaact cctcatctgt attaatgatc accggccctc
aggggtccct 180 gaccgattct ctggctccaa gtctggcacc tcagcctccc
tggccatcag tgggctccag 240 tccgaggatg aggctgatta ttactgtgca
acatgggatg acagcctggg tggccttgta 300 ttcggcggag ggaccaagct
gaccgtccta g 331 <210> SEQ ID NO 55 <211> LENGTH: 8
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA78 CDRH1
<400> SEQUENCE: 55 Gly Phe Thr Phe Ser Asn Tyr Ala 1 5
<210> SEQ ID NO 56 <211> LENGTH: 8 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA78 CDRH2 <400> SEQUENCE: 56
Ile Gly Arg Asn Gly Asp Ser Ile 1 5 <210> SEQ ID NO 57
<211> LENGTH: 16 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA78 CDRH3 <400> SEQUENCE: 57 Val Lys Asp Leu
Ala Ile Pro Glu Ser Tyr Arg Ile Glu Ala Asp Tyr 1 5 10 15
<210> SEQ ID NO 58 <211> LENGTH: 12 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA78 CDRL1 <400> SEQUENCE: 58
Gln Ser Val Leu Tyr Arg Ser Asn Asn Lys Asn Tyr 1 5 10 <210>
SEQ ID NO 59 <400> SEQUENCE: 59 000 <210> SEQ ID NO 60
<211> LENGTH: 9 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA78 CDRL2 long <400> SEQUENCE: 60 Leu Ile Tyr
Trp Ala Ser Thr Arg Glu 1 5 <210> SEQ ID NO 61 <211>
LENGTH: 9 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA78
CDRL3 <400> SEQUENCE: 61 Gln Gln Tyr Tyr Ser Ser Pro Arg Thr
1 5 <210> SEQ ID NO 62 <211> LENGTH: 123 <212>
TYPE: PRT <213> ORGANISM: Artificial Sequence <220>
FEATURE: <223> OTHER INFORMATION: ZKA78 VH <400>
SEQUENCE: 62 Glu Val Gln Leu Ala Glu Ser Gly Gly Gly Leu Val Gln
Pro Gly Gly 1 5 10 15 Ser Leu Thr Leu Ser Cys Ser Gly Ser Gly Phe
Thr Phe Ser Asn Tyr 20 25 30 Ala Met Val Trp Ala Arg Gln Ala Pro
Gly Lys Gly Leu Glu Tyr Val 35 40 45 Ser Gly Ile Gly Arg Asn Gly
Asp Ser Ile Tyr Tyr Thr Asp Ser Val 50 55 60 Lys Gly Arg Phe Thr
Ile Ser Arg Asp Asn Ser Lys Ser Met Val Tyr 65 70 75 80 Leu Gln Met
Ser Ser Leu Arg Thr Glu Asp Thr Ala Val Tyr Tyr Cys 85 90 95 Val
Lys Asp Leu Ala Ile Pro Glu Ser Tyr Arg Ile Glu Ala Asp Tyr 100 105
110 Trp Gly Gln Gly Thr Leu Val Ile Val Ser Ala 115 120 <210>
SEQ ID NO 63 <211> LENGTH: 113 <212> TYPE: PRT
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA78 VL <400> SEQUENCE: 63
Asp Ile Val Met Thr Gln Ser Pro Asp Ser Leu Ala Val Ser Leu Gly 1 5
10 15 Glu Arg Ala Thr Ile Asn Cys Lys Ser Ser Gln Ser Val Leu Tyr
Arg 20 25 30 Ser Asn Asn Lys Asn Tyr Leu Ser Trp Tyr Gln Gln Lys
Pro Gly Gln 35 40 45 Pro Pro Lys Leu Leu Ile Tyr Trp Ala Ser Thr
Arg Glu Ser Gly Val 50 55 60 Pro Asp Arg Phe Ser Gly Ser Gly Ser
Gly Thr Asp Phe Thr Leu Thr 65 70 75 80 Ile Ser Pro Leu Gln Ala Glu
Asp Val Ala Val Tyr Tyr Cys Gln Gln 85 90 95 Tyr Tyr Ser Ser Pro
Arg Thr Phe Gly Gln Gly Thr Lys Val Glu Ile 100 105 110 Lys
<210> SEQ ID NO 64 <211> LENGTH: 24 <212> TYPE:
DNA <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA78 CDRH1 <400> SEQUENCE: 64
ggcttcactt ttagtaacta tgca 24 <210> SEQ ID NO 65 <211>
LENGTH: 24 <212> TYPE: DNA <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA78
CDRH2 <400> SEQUENCE: 65 atcgggcgca acggggactc tatc 24
<210> SEQ ID NO 66 <211> LENGTH: 48 <212> TYPE:
DNA <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA78 CDRH3 <400> SEQUENCE: 66
gtgaaagatc tggccatccc cgagtcctac agaattgaag ctgattat 48 <210>
SEQ ID NO 67 <211> LENGTH: 36 <212> TYPE: DNA
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA78 CDRL1 <400> SEQUENCE: 67
cagtccgtgc tgtaccgctc taacaacaag aattac 36 <210> SEQ ID NO 68
<400> SEQUENCE: 68 000 <210> SEQ ID NO 69 <211>
LENGTH: 27 <212> TYPE: DNA <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA78
CDRL2 long <400> SEQUENCE: 69 ctgatctatt gggcttcaac ccgggaa
27 <210> SEQ ID NO 70 <211> LENGTH: 27 <212>
TYPE: DNA <213> ORGANISM: Artificial Sequence <220>
FEATURE: <223> OTHER INFORMATION: ZKA78 CDRL3 <400>
SEQUENCE: 70 cagcagtact attctagtcc tcgaact 27 <210> SEQ ID NO
71 <211> LENGTH: 370 <212> TYPE: DNA <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKA78 VH <400> SEQUENCE: 71 gaggtgcagc
tggcagaatc aggcggggga ctggtccagc ctggcggcag cctgacactg 60
tcttgcagtg gatcaggctt cacttttagt aactatgcaa tggtgtgggc aaggcaggct
120 cctgggaagg gactggagta tgtctctggc atcgggcgca acggggactc
tatctactat 180 actgatagtg tgaagggccg gttcaccatc agcagagaca
atagcaaatc catggtgtac 240 ctgcagatga gctccctgcg aaccgaagac
acagcagtgt actattgcgt gaaagatctg 300 gccatccccg agtcctacag
aattgaagct gattattggg gacagggcac cctggtcatc 360 gtgagcgccg 370
<210> SEQ ID NO 72 <211> LENGTH: 340 <212> TYPE:
DNA <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA78 VL <400> SEQUENCE: 72
gacatcgtga tgacacagtc tccagatagt ctggcagtca gtctggggga gagggccact
60 attaactgca agagctccca gtccgtgctg taccgctcta acaacaagaa
ttacctgtct 120 tggtatcagc agaagcccgg acagccccct aaactgctga
tctattgggc ttcaacccgg 180 gaaagcggcg tcccagacag attctcaggc
agcgggtccg gaacagactt caccctgaca 240 attagccccc tgcaggcaga
ggacgtggct gtctactatt gtcagcagta ctattctagt 300 cctcgaactt
tcggccaggg gaccaaggtg gaaatcaaac 340 <210> SEQ ID NO 73
<211> LENGTH: 8 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA64 CDRH1 <400> SEQUENCE: 73 Gly Tyr Thr Phe
Thr Gly Tyr His 1 5 <210> SEQ ID NO 74 <211> LENGTH: 8
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA64 CDRH2
<400> SEQUENCE: 74 Ile Asn Pro Asn Ser Gly Gly Thr 1 5
<210> SEQ ID NO 75 <211> LENGTH: 12 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA64 CDRH3 <400> SEQUENCE: 75
Ala Arg Met Ser Ser Ser Ile Trp Gly Phe Asp His 1 5 10 <210>
SEQ ID NO 76 <211> LENGTH: 6 <212> TYPE: PRT
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA64 CDRL1 <400> SEQUENCE: 76
Gln Ser Val Leu Ile Asn 1 5 <210> SEQ ID NO 77 <400>
SEQUENCE: 77 000 <210> SEQ ID NO 78 <211> LENGTH: 9
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA64 CDRL2
long <400> SEQUENCE: 78 Leu Ile Tyr Gly Ala Ser Ser Arg Ala 1
5 <210> SEQ ID NO 79 <211> LENGTH: 10 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA64 CDRL3 <400> SEQUENCE: 79
Gln Gln Tyr Asn Asp Trp Pro Pro Ile Thr 1 5 10 <210> SEQ ID
NO 80 <211> LENGTH: 119 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKA64 VH <400> SEQUENCE: 80 Gln Val Gln
Leu Val Gln Ser Gly Ala Glu Val Lys Lys Pro Gly Ala 1 5 10 15 Ser
Val Lys Val Ser Cys Lys Ala Ser Gly Tyr Thr Phe Thr Gly Tyr 20 25
30 His Ile Asp Trp Val Arg Gln Ala Arg Gly Gln Gly Leu Glu Trp Met
35 40 45 Gly Arg Ile Asn Pro Asn Ser Gly Gly Thr Asn Tyr Ala Gln
Lys Phe 50 55 60 Gln Gly Arg Val Thr Met Thr Arg Asp Thr Ser Ile
Ser Thr Ala Tyr 65 70 75 80 Met Gln Leu Ser Arg Leu Arg Ser Asp Asp
Ser Ala Val Tyr Tyr Cys 85 90 95 Ala Arg Met Ser Ser Ser Ile Trp
Gly Phe Asp His Trp Gly Gln Gly 100 105 110 Thr Leu Val Thr Val Ser
Ser 115 <210> SEQ ID NO 81 <211> LENGTH: 108
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA64 VL
<400> SEQUENCE: 81 Glu Ile Val Met Thr Gln Ser Pro Ala Thr
Leu Ser Val Ser Pro Gly 1 5 10 15 Glu Arg Ala Thr Leu Ser Cys Arg
Ala Ser Gln Ser Val Leu Ile Asn 20 25 30 Leu Ala Trp Tyr Gln Gln
Lys Pro Gly Gln Ala Pro Arg Leu Leu Ile 35 40 45 Tyr Gly Ala Ser
Ser Arg Ala Thr Gly Ile Pro Ala Arg Phe Ser Gly 50 55 60 Ser Gly
Ser Gly Thr Glu Phe Thr Leu Thr Ile Ser Ser Leu Gln Ser 65 70 75 80
Glu Asp Phe Ala Val Tyr Tyr Cys Gln Gln Tyr Asn Asp Trp Pro Pro 85
90 95 Ile Thr Phe Gly Gln Gly Thr Arg Leu Glu Ile Lys 100 105
<210> SEQ ID NO 82 <211> LENGTH: 24 <212> TYPE:
DNA <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA64 CDRH1 <400> SEQUENCE: 82
ggctacacct tcacagggta tcac 24 <210> SEQ ID NO 83 <211>
LENGTH: 24 <212> TYPE: DNA <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA64
CDRH2 <400> SEQUENCE: 83 attaacccta attctggcgg gacc 24
<210> SEQ ID NO 84 <211> LENGTH: 36 <212> TYPE:
DNA <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA64 CDRH3 <400> SEQUENCE: 84
gctcggatga gctcctctat ttggggcttc gatcat 36 <210> SEQ ID NO 85
<211> LENGTH: 18 <212> TYPE: DNA <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA64 CDRL1 <400> SEQUENCE: 85 cagtctgtgc
tgattaac 18 <210> SEQ ID NO 86 <400> SEQUENCE: 86 000
<210> SEQ ID NO 87 <211> LENGTH: 27 <212> TYPE:
DNA <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA64 CDRL2 long <400>
SEQUENCE: 87 ctgatctatg gagcatcctc cagggct 27 <210> SEQ ID NO
88 <211> LENGTH: 30 <212> TYPE: DNA <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKA64 CDRL3 <400> SEQUENCE: 88 cagcagtaca
atgattggcc ccctatcaca 30 <210> SEQ ID NO 89 <211>
LENGTH: 358 <212> TYPE: DNA <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA64
VH <400> SEQUENCE: 89 caggtgcagc tggtccagag cggagcagag
gtgaagaaac ccggcgcctc agtgaaggtc 60 agctgcaaag cttccggcta
caccttcaca gggtatcaca tcgactgggt gaggcaggca 120 agaggacagg
gactggaatg gatgggacgg attaacccta attctggcgg gaccaactac 180
gcccagaagt ttcagggccg agtgactatg accagagaca ccagcatctc cacagcttat
240 atgcagctgt cccggctgag atctgacgat agtgccgtct actattgtgc
tcggatgagc 300 tcctctattt ggggcttcga tcattggggg cagggaacac
tggtgactgt cagttcag 358 <210> SEQ ID NO 90 <211>
LENGTH: 325 <212> TYPE: DNA <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA64
VL <400> SEQUENCE: 90 gagatcgtga tgactcagtc tccagccacc
ctgtcagtca gcccaggaga acgggcaacc 60 ctgtcttgca gagcctccca
gtctgtgctg attaacctgg cttggtacca gcagaagcca 120 ggccaggcac
cccgactgct gatctatgga gcatcctcca gggctaccgg cattcctgca 180
cgcttcagtg gatcaggaag cggaacagag tttaccctga caatctctag tctgcagtcc
240 gaagacttcg ctgtctacta ttgtcagcag tacaatgatt ggccccctat
cacatttggc 300 caggggacta gactggagat caagc 325 <210> SEQ ID
NO 91 <211> LENGTH: 8 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKA15 CDRH1 <400> SEQUENCE: 91 Gly Gly Phe
Ile Asn Ser Tyr Tyr 1 5 <210> SEQ ID NO 92 <211>
LENGTH: 7 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA15
CDRH2 <400> SEQUENCE: 92 Ile Tyr Lys Ser Gly Ser Thr 1 5
<210> SEQ ID NO 93 <211> LENGTH: 14 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA15 CDRH3 <400> SEQUENCE: 93
Ala Arg Asp Pro Tyr Gly Asp Tyr Val Lys Ala Phe Asp Ile 1 5 10
<210> SEQ ID NO 94 <211> LENGTH: 11 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA15 CDRL1 <400> SEQUENCE: 94
Gln Ser Leu Leu His Ser Asn Gly Tyr Asn Tyr 1 5 10 <210> SEQ
ID NO 95 <400> SEQUENCE: 95 000 <210> SEQ ID NO 96
<211> LENGTH: 9 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA15 CDRL2 long <400> SEQUENCE: 96 Leu Ile Tyr
Leu Gly Ser Asn Arg Ala 1 5 <210> SEQ ID NO 97 <211>
LENGTH: 8 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA15
CDRL3 <400> SEQUENCE: 97 Met Gln Ala Leu Gln Thr Val Thr 1 5
<210> SEQ ID NO 98 <211> LENGTH: 120 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA15 VH <400> SEQUENCE: 98
Gln Val Gln Leu Gln Glu Ser Gly Pro Gly Leu Val Lys Pro Ser Glu 1 5
10 15 Thr Leu Ser Leu Thr Cys Thr Val Ser Gly Gly Phe Ile Asn Ser
Tyr 20 25 30 Tyr Trp Ser Trp Ile Arg Gln Pro Ala Gly Lys Gly Leu
Glu Trp Ile 35 40 45 Gly Arg Ile Tyr Lys Ser Gly Ser Thr Asn Tyr
Asn Pro Ser Leu Lys 50 55 60 Ser Arg Val Thr Met Ser Leu Asp Thr
Ser Lys Tyr Gln Phe Ser Leu 65 70 75 80 Lys Leu Arg Ser Val Thr Ala
Ala Asp Thr Ala Val Tyr Tyr Cys Ala 85 90 95 Arg Asp Pro Tyr Gly
Asp Tyr Val Lys Ala Phe Asp Ile Trp Gly Gln 100 105 110 Gly Thr Met
Val Thr Val Ser Ser 115 120 <210> SEQ ID NO 99 <211>
LENGTH: 111 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA15
VL <400> SEQUENCE: 99 Asp Ile Val Met Thr Gln Ser Pro Leu Ser
Leu Pro Val Thr Pro Gly 1 5 10 15 Glu Pro Ala Ser Ile Ser Cys Arg
Ser Ser Gln Ser Leu Leu His Ser 20 25 30 Asn Gly Tyr Asn Tyr Leu
Asn Trp Tyr Leu Gln Lys Pro Gly Gln Ser 35 40 45 Pro Gln Leu Leu
Ile Tyr Leu Gly Ser Asn Arg Ala Ser Gly Val Pro 50 55 60 Asp Arg
Phe Ser Gly Ser Gly Ser Gly Thr Asp Phe Thr Leu Lys Ile 65 70 75 80
Ser Arg Val Glu Ala Glu Asp Val Gly Val Tyr Tyr Cys Met Gln Ala 85
90 95 Leu Gln Thr Val Thr Phe Gly Pro Gly Thr Lys Val Asp Ile Lys
100 105 110 <210> SEQ ID NO 100 <211> LENGTH: 24
<212> TYPE: DNA <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA15 CDRH1
<400> SEQUENCE: 100 ggtggcttca tcaatagtta ctac 24 <210>
SEQ ID NO 101 <211> LENGTH: 21 <212> TYPE: DNA
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA15 CDRH2 <400> SEQUENCE:
101 atctataaaa gtgggagcac c 21 <210> SEQ ID NO 102
<211> LENGTH: 42 <212> TYPE: DNA <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA15 CDRH3 <400> SEQUENCE: 102 gcgagagatc
cctacggtga ctacgttaag gcttttgata tt 42 <210> SEQ ID NO 103
<211> LENGTH: 33 <212> TYPE: DNA <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA15 CDRL1 <400> SEQUENCE: 103 cagagcctcc
tgcatagtaa tggatacaac tat 33 <210> SEQ ID NO 104 <400>
SEQUENCE: 104 000 <210> SEQ ID NO 105 <211> LENGTH: 27
<212> TYPE: DNA <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA15 CDRL2
long <400> SEQUENCE: 105 ctgatctatt tgggttctaa tcgggcc 27
<210> SEQ ID NO 106 <211> LENGTH: 24 <212> TYPE:
DNA <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA15 CDRL3 <400> SEQUENCE:
106 atgcaagctc tacaaactgt cact 24 <210> SEQ ID NO 107
<211> LENGTH: 361 <212> TYPE: DNA <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA15 VH <400> SEQUENCE: 107 caggtgcagc
tgcaggagtc ggggccagga ctggtgaagc cttcggagac cctgtccctc 60
acctgcactg tctccggtgg cttcatcaat agttactact ggagctggat ccggcagccc
120 gccgggaagg gactggagtg gattgggcgt atctataaaa gtgggagcac
caactacaac 180 ccctccctca agagtcgagt caccatgtca ctagacacgt
ccaagtacca gttctccctg 240 aagctgaggt ctgtgaccgc cgctgacacg
gccgtgtatt actgtgcgag agatccctac 300 ggtgactacg ttaaggcttt
tgatatttgg ggccaaggga caatggtcac cgtctcttca 360 g 361 <210>
SEQ ID NO 108 <211> LENGTH: 334 <212> TYPE: DNA
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA15 VL <400> SEQUENCE: 108
gatattgtga tgactcagtc tccactctcc ctgcccgtca cccctggaga gccggcctcc
60 atctcctgca ggtctagtca gagcctcctg catagtaatg gatacaacta
tttgaattgg 120 tacctgcaga agccagggca gtctccacag ctcctgatct
atttgggttc taatcgggcc 180 tccggggtcc ctgacaggtt cagtggcagt
ggatcaggca cagattttac actgaaaatc 240 agcagagtgg aggctgagga
tgttggggtt tattactgca tgcaagctct acaaactgtc 300 actttcggcc
ctgggaccaa agtggatatc aaac 334 <210> SEQ ID NO 109
<211> LENGTH: 8 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA25 CDRH1 <400> SEQUENCE: 109 Gly Phe Thr Phe
Arg Ser His Trp 1 5 <210> SEQ ID NO 110 <211> LENGTH: 8
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA25 CDRH2
<400> SEQUENCE: 110 Ile Lys Glu Asp Gly Tyr Glu Lys 1 5
<210> SEQ ID NO 111 <211> LENGTH: 14 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA25 CDRH3 <400> SEQUENCE:
111 Ala Arg Asp Leu Arg Val Tyr Ser Gly Arg Gly Phe Asp Pro 1 5 10
<210> SEQ ID NO 112 <211> LENGTH: 6 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA25 CDRL1 <400> SEQUENCE:
112 Lys Leu Gly Asp Lys Tyr 1 5 <210> SEQ ID NO 113
<400> SEQUENCE: 113 000 <210> SEQ ID NO 114 <211>
LENGTH: 9 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA25
CDRL2 long <400> SEQUENCE: 114 Val Ile Tyr Gln Asp Ser Lys
Arg Pro 1 5 <210> SEQ ID NO 115 <211> LENGTH: 9
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA25 CDRL3
<400> SEQUENCE: 115 Gln Ala Trp Asp Ser Ser Thr Val Val 1 5
<210> SEQ ID NO 116 <211> LENGTH: 121 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA25 VH <400> SEQUENCE: 116
Glu Val Gln Leu Val Glu Ser Gly Gly Gly Leu Val Arg Pro Gly Gly 1 5
10 15 Ser Leu Arg Leu Ser Cys Ala Ala Ser Gly Phe Thr Phe Arg Ser
His 20 25 30 Trp Met Ser Trp Val Arg Gln Ala Pro Gly Lys Gly Leu
Glu Trp Val 35 40 45 Ala Asn Ile Lys Glu Asp Gly Tyr Glu Lys Tyr
Tyr Val Asp Ser Val 50 55 60 Lys Gly Arg Phe Thr Ile Ser Arg Asp
Asn Ala Lys Asn Ser Leu Tyr 65 70 75 80 Leu Gln Met Lys Ser Leu Arg
Ala Glu Asp Thr Ala Val Tyr Tyr Cys 85 90 95 Ala Arg Asp Leu Arg
Val Tyr Ser Gly Arg Gly Phe Asp Pro Trp Gly 100 105 110 Gln Gly Thr
Leu Val Thr Val Ser Ser 115 120 <210> SEQ ID NO 117
<211> LENGTH: 106 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA25 VL <400> SEQUENCE: 117 Ser Tyr Glu Leu Thr
Gln Pro Pro Ser Leu Ser Val Ser Pro Gly Gln 1 5 10 15 Thr Ala Ser
Ile Thr Cys Ser Gly Asp Lys Leu Gly Asp Lys Tyr Ala 20 25 30 Cys
Trp Tyr Gln Gln Lys Pro Gly Gln Ser Pro Val Leu Val Ile Tyr 35 40
45 Gln Asp Ser Lys Arg Pro Ser Gly Ile Pro Ala Arg Phe Ser Gly Ser
50 55 60 Asn Ser Gly Asn Thr Ala Thr Leu Thr Ile Ser Gly Thr Gln
Ala Met 65 70 75 80 Asp Glu Ala Asp Tyr Tyr Cys Gln Ala Trp Asp Ser
Ser Thr Val Val 85 90 95 Phe Gly Gly Gly Thr Lys Leu Thr Val Leu
100 105 <210> SEQ ID NO 118 <211> LENGTH: 24
<212> TYPE: DNA <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA25 CDRH1
<400> SEQUENCE: 118 ggattcacct ttagaagtca ttgg 24 <210>
SEQ ID NO 119 <211> LENGTH: 24 <212> TYPE: DNA
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA25 CDRH2 <400> SEQUENCE:
119 ataaaggaag atggatatga gaaa 24 <210> SEQ ID NO 120
<211> LENGTH: 42 <212> TYPE: DNA <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA25 CDRH3 <400> SEQUENCE: 120 gcgagagatt
tgagggtata tagtgggaga ggtttcgacc cc 42 <210> SEQ ID NO 121
<211> LENGTH: 18 <212> TYPE: DNA <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA25 CDRL1 <400> SEQUENCE: 121 aaattggggg
ataaatat 18 <210> SEQ ID NO 122 <400> SEQUENCE: 122 000
<210> SEQ ID NO 123 <211> LENGTH: 27 <212> TYPE:
DNA <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA25 CDRL2 long <400>
SEQUENCE: 123 gtcatctatc aagatagcaa gcggccc 27 <210> SEQ ID
NO 124 <211> LENGTH: 27 <212> TYPE: DNA <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKA25 CDRL3 <400> SEQUENCE: 124 caggcgtggg
acagcagcac tgtggta 27 <210> SEQ ID NO 125 <211> LENGTH:
364 <212> TYPE: DNA <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA25 VH
<400> SEQUENCE: 125 gaggtgcagt tggtggagtc tgggggaggc
ttggtccggc ctggggggtc cctgagactc 60 tcctgtgcag cctctggatt
cacctttaga agtcattgga tgagttgggt ccgccaggct 120 ccagggaagg
ggctggagtg ggtggccaac ataaaggaag atggatatga gaaatactat 180
gtggactctg tgaagggccg attcaccatc tccagagaca acgccaagaa ctcactgtat
240 ctgcaaatga agagcctgag agccgaggac acggccgtgt attactgtgc
gagagatttg 300 agggtatata gtgggagagg tttcgacccc tggggccagg
gaaccctggt caccgtctcc 360 tcag 364 <210> SEQ ID NO 126
<211> LENGTH: 319 <212> TYPE: DNA <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA25 VL <400> SEQUENCE: 126 tcctatgagc
tgactcagcc accctcactg tccgtgtccc caggacagac agccagcatc 60
acctgctctg gagataaatt gggggataaa tatgcttgct ggtatcagca gaagccaggc
120 cagtcccctg tgttggtcat ctatcaagat agcaagcggc cctcagggat
ccctgcgcga 180 ttctctggct ccaactctgg gaacacagcc actctgacca
tcagcgggac ccaggctatg 240 gatgaggctg actattactg tcaggcgtgg
gacagcagca ctgtggtatt cggtggaggg 300 accaagctga ccgtcctag 319
<210> SEQ ID NO 127 <211> LENGTH: 10 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA35 CDRH1 <400> SEQUENCE:
127 Gly Gly Ser Ile Ser Thr Gly Gly Tyr Tyr 1 5 10 <210> SEQ
ID NO 128 <211> LENGTH: 7 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKA35 CDRH2 <400> SEQUENCE: 128 Ile Tyr
Tyr Ser Gly Asn Thr 1 5 <210> SEQ ID NO 129 <211>
LENGTH: 12 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA35
CDRH3 <400> SEQUENCE: 129 Ala Lys Gly Gly Gly Arg Glu Arg Pro
Phe Asp Tyr 1 5 10 <210> SEQ ID NO 130 <211> LENGTH: 8
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA35 CDRL1
<400> SEQUENCE: 130 Ser Ser Asn Ile Gly Arg Asn Tyr 1 5
<210> SEQ ID NO 131 <400> SEQUENCE: 131 000 <210>
SEQ ID NO 132 <211> LENGTH: 9 <212> TYPE: PRT
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA35 CDRL2 long <400>
SEQUENCE: 132 Leu Ile Tyr Arg Asn Asn Gln Arg Pro 1 5 <210>
SEQ ID NO 133 <211> LENGTH: 12 <212> TYPE: PRT
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA35 CDRL3 <400> SEQUENCE:
133 Val Ala Trp Asp Asp Ser Arg Ser Gly Phe Val Val 1 5 10
<210> SEQ ID NO 134 <211> LENGTH: 120 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA35 VH <400> SEQUENCE: 134
Gln Val Gln Leu Gln Glu Ser Gly Pro Gly Leu Val Lys Pro Ser Gln 1 5
10 15 Thr Leu Ser Leu Thr Cys Thr Val Ser Gly Gly Ser Ile Ser Thr
Gly 20 25 30 Gly Tyr Tyr Trp Ser Trp Ile Arg Gln His Pro Gly Lys
Gly Leu Glu 35 40 45 Trp Ile Gly Tyr Ile Tyr Tyr Ser Gly Asn Thr
Tyr Tyr Asn Pro Ser 50 55 60 Leu Lys Ser Arg Val Thr Ile Ser Val
Asp Thr Ser Lys Lys Gln Phe 65 70 75 80 Ser Leu Lys Leu Ser Ser Val
Thr Ala Ala Asp Thr Ala Val Tyr Tyr 85 90 95 Cys Ala Lys Gly Gly
Gly Arg Glu Arg Pro Phe Asp Tyr Trp Gly Gln 100 105 110 Gly Thr Leu
Val Thr Val Ser Ser 115 120 <210> SEQ ID NO 135 <211>
LENGTH: 111 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA35
VL <400> SEQUENCE: 135 Gln Ser Val Leu Thr Gln Pro Pro Ser
Ala Ser Gly Thr Pro Gly Gln 1 5 10 15 Arg Val Thr Ile Ser Cys Ser
Gly Ser Ser Ser Asn Ile Gly Arg Asn 20 25 30 Tyr Val Asp Trp Tyr
Gln Gln Leu Pro Gly Thr Ala Pro Lys Leu Leu 35 40 45 Ile Tyr Arg
Asn Asn Gln Arg Pro Ser Gly Val Pro Glu Arg Phe Ser 50 55 60 Gly
Ser Lys Ser Gly Thr Ser Ala Ser Leu Ala Ile Ser Gly Leu Arg 65 70
75 80 Ser Glu Asp Glu Ala Asp Tyr Tyr Cys Val Ala Trp Asp Asp Ser
Arg 85 90 95 Ser Gly Phe Val Val Phe Gly Gly Gly Thr Lys Val Thr
Val Leu 100 105 110 <210> SEQ ID NO 136 <211> LENGTH:
30 <212> TYPE: DNA <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA35 CDRH1
<400> SEQUENCE: 136 ggtggctcca tcagcactgg tggttactac 30
<210> SEQ ID NO 137 <211> LENGTH: 21 <212> TYPE:
DNA <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA35 CDRH2 <400> SEQUENCE:
137 atctattaca gtgggaacac c 21 <210> SEQ ID NO 138
<211> LENGTH: 36 <212> TYPE: DNA <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA35 CDRH3 <400> SEQUENCE: 138 gcgaaaggag
gagggaggga gcgacccttt gactac 36 <210> SEQ ID NO 139
<211> LENGTH: 24 <212> TYPE: DNA <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA35 CDRL1 <400> SEQUENCE: 139 agctccaaca
tcggaagaaa ttat 24 <210> SEQ ID NO 140 <400> SEQUENCE:
140 000 <210> SEQ ID NO 141 <211> LENGTH: 27
<212> TYPE: DNA <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA35 CDRL2
long <400> SEQUENCE: 141 ctcatctata ggaataatca gcggccc 27
<210> SEQ ID NO 142 <211> LENGTH: 36 <212> TYPE:
DNA <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA35 CDRL3 <400> SEQUENCE:
142 gtagcatggg atgacagccg gagtggtttt gtggta 36 <210> SEQ ID
NO 143 <211> LENGTH: 361 <212> TYPE: DNA <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKA35 VH <400> SEQUENCE: 143 caggtgcagc
tgcaggagtc gggcccagga ctggtgaagc cttcacagac cctgtccctc 60
acctgcactg tctctggtgg ctccatcagc actggtggtt actactggag ctggatccgc
120 cagcacccag ggaagggcct ggagtggatt ggttacatct attacagtgg
gaacacctac 180 tacaacccgt ccctcaagag tcgagttacc atatcagttg
acacctctaa gaagcagttc 240 tccctgaagc tgagctctgt gactgccgcg
gacacggccg tgtattactg tgcgaaagga 300 ggagggaggg agcgaccctt
tgactactgg ggccagggaa ccctggtcac cgtctcctca 360 g 361 <210>
SEQ ID NO 144 <211> LENGTH: 334 <212> TYPE: DNA
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA35 VL <400> SEQUENCE: 144
cagtctgtgc tgactcagcc accctcagcg tctgggaccc ccgggcagag ggtcaccatc
60 tcttgttctg gaagcagctc caacatcgga agaaattatg tagactggta
ccagcaactc 120 ccaggaacgg cccccaaact cctcatctat aggaataatc
agcggccctc aggggtccct 180 gagcgattct ctggctccaa gtctggcacc
tcagcctccc tggccatcag tgggctccgg 240 tccgaggatg aggctgatta
ttactgtgta gcatgggatg acagccggag tggttttgtg 300 gtattcggcg
gagggaccaa ggtgaccgtc ctag 334 <210> SEQ ID NO 145
<211> LENGTH: 330 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: IgG1 CH1-CH2-CH3 aa <400> SEQUENCE: 145 Ala Ser
Thr Lys Gly Pro Ser Val Phe Pro Leu Ala Pro Ser Ser Lys 1 5 10 15
Ser Thr Ser Gly Gly Thr Ala Ala Leu Gly Cys Leu Val Lys Asp Tyr 20
25 30 Phe Pro Glu Pro Val Thr Val Ser Trp Asn Ser Gly Ala Leu Thr
Ser 35 40 45 Gly Val His Thr Phe Pro Ala Val Leu Gln Ser Ser Gly
Leu Tyr Ser 50 55 60 Leu Ser Ser Val Val Thr Val Pro Ser Ser Ser
Leu Gly Thr Gln Thr 65 70 75 80 Tyr Ile Cys Asn Val Asn His Lys Pro
Ser Asn Thr Lys Val Asp Lys 85 90 95 Arg Val Glu Pro Lys Ser Cys
Asp Lys Thr His Thr Cys Pro Pro Cys 100 105 110 Pro Ala Pro Glu Leu
Leu Gly Gly Pro Ser Val Phe Leu Phe Pro Pro 115 120 125 Lys Pro Lys
Asp Thr Leu Met Ile Ser Arg Thr Pro Glu Val Thr Cys 130 135 140 Val
Val Val Asp Val Ser His Glu Asp Pro Glu Val Lys Phe Asn Trp 145 150
155 160 Tyr Val Asp Gly Val Glu Val His Asn Ala Lys Thr Lys Pro Arg
Glu 165 170 175 Glu Gln Tyr Asn Ser Thr Tyr Arg Val Val Ser Val Leu
Thr Val Leu 180 185 190 His Gln Asp Trp Leu Asn Gly Lys Glu Tyr Lys
Cys Lys Val Ser Asn 195 200 205 Lys Ala Leu Pro Ala Pro Ile Glu Lys
Thr Ile Ser Lys Ala Lys Gly 210 215 220 Gln Pro Arg Glu Pro Gln Val
Tyr Thr Leu Pro Pro Ser Arg Glu Glu 225 230 235 240 Met Thr Lys Asn
Gln Val Ser Leu Thr Cys Leu Val Lys Gly Phe Tyr 245 250 255 Pro Ser
Asp Ile Ala Val Glu Trp Glu Ser Asn Gly Gln Pro Glu Asn 260 265 270
Asn Tyr Lys Thr Thr Pro Pro Val Leu Asp Ser Asp Gly Ser Phe Phe 275
280 285 Leu Tyr Ser Lys Leu Thr Val Asp Lys Ser Arg Trp Gln Gln Gly
Asn 290 295 300 Val Phe Ser Cys Ser Val Met His Glu Ala Leu His Asn
His Tyr Thr 305 310 315 320 Gln Lys Ser Leu Ser Leu Ser Pro Gly Lys
325 330 <210> SEQ ID NO 146 <211> LENGTH: 330
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: IgG1
CH1-CH2-CH3 LALA aa <400> SEQUENCE: 146 Ala Ser Thr Lys Gly
Pro Ser Val Phe Pro Leu Ala Pro Ser Ser Lys 1 5 10 15 Ser Thr Ser
Gly Gly Thr Ala Ala Leu Gly Cys Leu Val Lys Asp Tyr 20 25 30 Phe
Pro Glu Pro Val Thr Val Ser Trp Asn Ser Gly Ala Leu Thr Ser 35 40
45 Gly Val His Thr Phe Pro Ala Val Leu Gln Ser Ser Gly Leu Tyr Ser
50 55 60 Leu Ser Ser Val Val Thr Val Pro Ser Ser Ser Leu Gly Thr
Gln Thr 65 70 75 80 Tyr Ile Cys Asn Val Asn His Lys Pro Ser Asn Thr
Lys Val Asp Lys 85 90 95 Lys Val Glu Pro Lys Ser Cys Asp Lys Thr
His Thr Cys Pro Pro Cys 100 105 110 Pro Ala Pro Glu Ala Ala Gly Gly
Pro Ser Val Phe Leu Phe Pro Pro 115 120 125 Lys Pro Lys Asp Thr Leu
Met Ile Ser Arg Thr Pro Glu Val Thr Cys 130 135 140 Val Val Val Asp
Val Ser His Glu Asp Pro Glu Val Lys Phe Asn Trp 145 150 155 160 Tyr
Val Asp Gly Val Glu Val His Asn Ala Lys Thr Lys Pro Arg Glu 165 170
175 Glu Gln Tyr Asn Ser Thr Tyr Arg Val Val Ser Val Leu Thr Val Leu
180 185 190 His Gln Asp Trp Leu Asn Gly Lys Glu Tyr Lys Cys Lys Val
Ser Asn 195 200 205 Lys Ala Leu Pro Ala Pro Ile Glu Lys Thr Ile Ser
Lys Ala Lys Gly 210 215 220 Gln Pro Arg Glu Pro Gln Val Tyr Thr Leu
Pro Pro Ser Arg Asp Glu 225 230 235 240 Leu Thr Lys Asn Gln Val Ser
Leu Thr Cys Leu Val Lys Gly Phe Tyr 245 250 255 Pro Ser Asp Ile Ala
Val Glu Trp Glu Ser Asn Gly Gln Pro Glu Asn 260 265 270 Asn Tyr Lys
Thr Thr Pro Pro Val Leu Asp Ser Asp Gly Ser Phe Phe 275 280 285 Leu
Tyr Ser Lys Leu Thr Val Asp Lys Ser Arg Trp Gln Gln Gly Asn 290 295
300 Val Phe Ser Cys Ser Val Met His Glu Ala Leu His Asn His Tyr Thr
305 310 315 320 Gln Lys Ser Leu Ser Leu Ser Pro Gly Lys 325 330
<210> SEQ ID NO 147 <211> LENGTH: 107 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: IgG CK aa <400> SEQUENCE: 147
Arg Thr Val Ala Ala Pro Ser Val Phe Ile Phe Pro Pro Ser Asp Glu 1 5
10 15 Gln Leu Lys Ser Gly Thr Ala Ser Val Val Cys Leu Leu Asn Asn
Phe 20 25 30 Tyr Pro Arg Glu Ala Lys Val Gln Trp Lys Val Asp Asn
Ala Leu Gln 35 40 45 Ser Gly Asn Ser Gln Glu Ser Val Thr Glu Gln
Asp Ser Lys Asp Ser 50 55 60 Thr Tyr Ser Leu Ser Ser Thr Leu Thr
Leu Ser Lys Ala Asp Tyr Glu 65 70 75 80 Lys His Lys Val Tyr Ala Cys
Glu Val Thr His Gln Gly Leu Ser Ser 85 90 95 Pro Val Thr Lys Ser
Phe Asn Arg Gly Glu Cys 100 105 <210> SEQ ID NO 148
<211> LENGTH: 106 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: IgG CL aa <400> SEQUENCE: 148 Gly Gln Pro Lys
Ala Ala Pro Ser Val Thr Leu Phe Pro Pro Ser Ser 1 5 10 15 Glu Glu
Leu Gln Ala Asn Lys Ala Thr Leu Val Cys Leu Ile Ser Asp 20 25 30
Phe Tyr Pro Gly Ala Val Thr Val Ala Trp Lys Ala Asp Ser Ser Pro 35
40 45 Val Lys Ala Gly Val Glu Thr Thr Thr Pro Ser Lys Gln Ser Asn
Asn 50 55 60 Lys Tyr Ala Ala Ser Ser Tyr Leu Ser Leu Thr Pro Glu
Gln Trp Lys 65 70 75 80 Ser His Arg Ser Tyr Ser Cys Gln Val Thr His
Glu Gly Ser Thr Val 85 90 95 Glu Lys Thr Val Ala Pro Thr Glu Cys
Ser 100 105 <210> SEQ ID NO 149 <211> LENGTH: 990
<212> TYPE: DNA <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: IgG1
CH1-CH2-CH3 nucl <400> SEQUENCE: 149 gcgtcgacca agggcccatc
ggtcttcccc ctggcaccct cctccaagag cacctctggg 60 ggcacagcgg
ccctgggctg cctggtcaag gactacttcc ccgaacctgt gacggtctcg 120
tggaactcag gcgccctgac cagcggcgtg cacaccttcc cggctgtcct acagtcctca
180 ggactctact ccctcagcag cgtggtgacc gtgccctcca gcagcttggg
cacccagacc 240 tacatctgca acgtgaatca caagcccagc aacaccaagg
tggacaagag agttgagccc 300 aaatcttgtg acaaaactca cacatgccca
ccgtgcccag cacctgaact cctgggggga 360 ccgtcagtct tcctcttccc
cccaaaaccc aaggacaccc tcatgatctc ccggacccct 420 gaggtcacat
gcgtggtggt ggacgtgagc cacgaagacc ctgaggtcaa gttcaactgg 480
tacgtggacg gcgtggaggt gcataatgcc aagacaaagc cgcgggagga gcagtacaac
540 agcacgtacc gtgtggtcag cgtcctcacc gtcctgcacc aggactggct
gaatggcaag 600 gagtacaagt gcaaggtctc caacaaagcc ctcccagccc
ccatcgagaa aaccatctcc 660 aaagccaaag ggcagccccg agaaccacag
gtgtacaccc tgcccccatc ccgggaggag 720 atgaccaaga accaggtcag
cctgacctgc ctggtcaaag gcttctatcc cagcgacatc 780 gccgtggagt
gggagagcaa tgggcagccg gagaacaact acaagaccac gcctcccgtg 840
ctggactccg acggctcctt cttcctctat agcaagctca ccgtggacaa gagcaggtgg
900 cagcagggga acgtcttctc atgctccgtg atgcatgagg ctctgcacaa
ccactacacg 960 cagaagagcc tctccctgtc cccgggtaaa 990 <210> SEQ
ID NO 150 <211> LENGTH: 990 <212> TYPE: DNA <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: IgG1 CH1-CH2-CH3 LALA nucl <400> SEQUENCE:
150 gcgtcgacca agggcccatc ggtcttcccc ctggcaccct cctccaagag
cacctctggg 60 ggcacagcgg ccctgggctg cctggtcaag gactacttcc
ccgaacctgt gacggtctcg 120 tggaactcag gcgccctgac cagcggcgtg
cacaccttcc cggctgtcct acagtcctca 180 ggactctact ccctcagcag
cgtggtgacc gtgccctcca gcagcttggg cacccagacc 240 tacatctgca
acgtgaatca caagcccagc aacaccaagg tggacaagag agttgagccc 300
aaatcttgtg acaaaactca cacatgccca ccgtgcccag cacctgaagc cgcgggggga
360 ccgtcagtct tcctcttccc cccaaaaccc aaggacaccc tcatgatctc
ccggacccct 420 gaggtcacat gcgtggtggt ggacgtgagc cacgaagacc
ctgaggtcaa gttcaactgg 480 tacgtggacg gcgtggaggt gcataatgcc
aagacaaagc cgcgggagga gcagtacaac 540 agcacgtacc gtgtggtcag
cgtcctcacc gtcctgcacc aggactggct gaatggcaag 600 gagtacaagt
gcaaggtctc caacaaagcc ctcccagccc ccatcgagaa aaccatctcc 660
aaagccaaag ggcagccccg agaaccacag gtgtacaccc tgcccccatc ccgggaggag
720 atgaccaaga accaggtcag cctgacctgc ctggtcaaag gcttctatcc
cagcgacatc 780 gccgtggagt gggagagcaa tgggcagccg gagaacaact
acaagaccac gcctcccgtg 840 ctggactccg acggctcctt cttcctctat
agcaagctca ccgtggacaa gagcaggtgg 900 cagcagggga acgtcttctc
atgctccgtg atgcatgagg ctctgcacaa ccactacacg 960 cagaagagcc
tctccctgtc cccgggtaaa 990 <210> SEQ ID NO 151 <211>
LENGTH: 321 <212> TYPE: DNA <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: IgG CK
nucl <400> SEQUENCE: 151 cgtacggtgg ctgcaccatc tgtcttcatc
ttcccgccat ctgatgagca gttgaaatct 60 ggaactgcct ctgttgtgtg
cctgctgaat aacttctatc ccagagaggc caaagtacag 120 tggaaggtgg
ataacgccct ccaatcgggt aactcccagg agagtgtcac agagcaggac 180
agcaaggaca gcacctacag cctcagcagc accctgacgc tgagcaaagc agactacgag
240 aaacacaaag tctacgcctg cgaagtcacc catcagggcc tgagctcgcc
cgtcacaaag 300 agcttcaaca ggggagagtg t 321 <210> SEQ ID NO
152 <211> LENGTH: 318 <212> TYPE: DNA <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: IgG CL nucl <400> SEQUENCE: 152 ggtcagccca
aggctgcccc ctcggtcact ctgttcccgc cctcctctga ggagcttcaa 60
gccaacaagg ccacactggt gtgtctcata agtgacttct acccgggagc cgtgacagtg
120 gcttggaaag cagatagcag ccccgtcaag gcgggagtgg agaccaccac
accctccaaa 180 caaagcaaca acaagtacgc ggccagcagc tatctgagcc
tgacgcctga gcagtggaag 240 tcccacagaa gctacagctg ccaggtcacg
catgaaggga gcaccgtgga gaagacagtg 300 gcccctacag aatgttca 318
<210> SEQ ID NO 153 <211> LENGTH: 8 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA10 CDRH1 <400> SEQUENCE:
153 Gly Phe Thr Phe Ser Asp Ser Tyr 1 5 <210> SEQ ID NO 154
<211> LENGTH: 8 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA10 CDRH2 <400> SEQUENCE: 154 Ile Ser Ser Ser
Ser Pro Phe Thr 1 5 <210> SEQ ID NO 155 <211> LENGTH:
17 <212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA10 CDRH3
<400> SEQUENCE: 155 Ala Arg Gly Leu Val Arg Asp Gly Tyr Lys
Trp Leu Tyr Phe Phe Asp 1 5 10 15 Tyr <210> SEQ ID NO 156
<211> LENGTH: 124 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA10 VH <400> SEQUENCE: 156 Gln Val Gln Leu Val
Glu Ser Gly Gly Gly Leu Val Glu Pro Arg Gly 1 5 10 15 Ser Leu Arg
Leu Ser Cys Ala Ala Ser Gly Phe Thr Phe Ser Asp Ser 20 25 30 Tyr
Met Ser Trp Ile Arg Gln Ala Pro Gly Lys Gly Leu Glu Trp Ile 35 40
45 Ser Tyr Ile Ser Ser Ser Ser Pro Phe Thr Asn Tyr Ala Asp Ser Val
50 55 60 Lys Gly Arg Phe Thr Ile Ser Arg Asp Asn Ala Lys Asn Ser
Leu Tyr 65 70 75 80 Leu Gln Met Asn Ser Leu Arg Ala Glu Asp Thr Ala
Val Tyr Tyr Cys 85 90 95 Ala Arg Gly Leu Val Arg Asp Gly Tyr Lys
Trp Leu Tyr Phe Phe Asp 100 105 110 Tyr Trp Gly Gln Gly Thr Leu Val
Thr Val Ser Ser 115 120 <210> SEQ ID NO 157 <211>
LENGTH: 8 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA18
CDRH1 <400> SEQUENCE: 157 Gly Phe Thr Phe Ser Ser Tyr Gly 1 5
<210> SEQ ID NO 158 <211> LENGTH: 8 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA18 CDRH2 <400> SEQUENCE:
158 Ile Trp Tyr Asp Gly Ser Asn Lys 1 5 <210> SEQ ID NO 159
<211> LENGTH: 13 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA18 CDRH3 <400> SEQUENCE: 159 Ala Arg Asp Asp
Ser Gly Tyr Ser Glu Pro Phe Asp Tyr 1 5 10 <210> SEQ ID NO
160 <211> LENGTH: 120 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKA18 VH <400> SEQUENCE: 160 Gln Val Gln
Leu Val Glu Ser Gly Gly Gly Val Val Gln Pro Gly Arg 1 5 10 15 Ser
Leu Arg Leu Ser Cys Ala Ala Ser Gly Phe Thr Phe Ser Ser Tyr 20 25
30 Gly Met His Trp Val Arg Gln Ala Pro Gly Lys Gly Leu Glu Trp Val
35 40 45 Ala Val Ile Trp Tyr Asp Gly Ser Asn Lys Tyr Tyr Ala Asp
Ser Val 50 55 60 Lys Gly Arg Phe Thr Ile Thr Arg Asp Asn Ser Lys
Asn Thr Leu Tyr 65 70 75 80 Leu Gln Met Asn Ser Leu Arg Pro Glu Asp
Thr Ala Val Tyr Tyr Cys 85 90 95 Ala Arg Asp Asp Ser Gly Tyr Ser
Glu Pro Phe Asp Tyr Trp Gly Gln 100 105 110 Gly Thr Leu Val Thr Val
Ser Ser 115 120 <210> SEQ ID NO 161 <211> LENGTH: 8
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA28 CDRH1
<400> SEQUENCE: 161 Gly Phe Thr Val Ser Arg Asn Tyr 1 5
<210> SEQ ID NO 162 <211> LENGTH: 7 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA28 CDRH2 <400> SEQUENCE:
162 Ile Tyr Ser Gly Gly Ser Thr 1 5 <210> SEQ ID NO 163
<211> LENGTH: 10 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA28 CDRH3 <400> SEQUENCE: 163 Ala Arg Trp Ile
Asn Asp Ala Phe Asp Ile 1 5 10 <210> SEQ ID NO 164
<211> LENGTH: 116 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA28 VH <400> SEQUENCE: 164 Glu Val Gln Leu Val
Glu Ser Gly Gly Gly Leu Ile Gln Pro Gly Gly 1 5 10 15 Ser Leu Arg
Leu Ser Cys Ala Ala Ser Gly Phe Thr Val Ser Arg Asn 20 25 30 Tyr
Met Ser Trp Val Arg Gln Ala Pro Gly Lys Gly Leu Glu Trp Val 35 40
45 Ser Val Ile Tyr Ser Gly Gly Ser Thr Tyr Tyr Ala Asp Ser Val Lys
50 55 60 Gly Arg Phe Thr Ile Ser Arg Asp Asn Ser Lys Asn Thr Leu
Tyr Leu 65 70 75 80 Gln Met Asn Ser Leu Arg Ala Glu Asp Thr Ala Val
Tyr Tyr Cys Ala 85 90 95 Arg Trp Ile Asn Asp Ala Phe Asp Ile Trp
Gly Gln Gly Thr Met Val 100 105 110 Thr Val Ser Ser 115 <210>
SEQ ID NO 165 <211> LENGTH: 8 <212> TYPE: PRT
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA29 CDRH1 <400> SEQUENCE:
165 Gly Phe Thr Phe Ser Arg Tyr Ser 1 5 <210> SEQ ID NO 166
<211> LENGTH: 8 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA29 CDRH2 <400> SEQUENCE: 166 Ile Ser Pro Arg
Ser Thr Thr Ile 1 5 <210> SEQ ID NO 167 <211> LENGTH:
15 <212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA29 CDRH3
<400> SEQUENCE: 167 Ala Arg Glu Asp Cys Thr Asn Gly Val Cys
Tyr Arg Val Asp Tyr 1 5 10 15 <210> SEQ ID NO 168 <211>
LENGTH: 122 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA29
VH <400> SEQUENCE: 168 Glu Val Gln Leu Val Glu Ser Gly Gly
Gly Leu Val Gln Pro Gly Gly 1 5 10 15 Ser Leu Arg Leu Ser Cys Val
Val Ser Gly Phe Thr Phe Ser Arg Tyr 20 25 30 Ser Met Asn Trp Val
Arg Gln Ala Pro Gly Lys Gly Leu Glu Trp Val 35 40 45 Ser Tyr Ile
Ser Pro Arg Ser Thr Thr Ile Tyr Tyr Ala Asp Ser Val 50 55 60 Glu
Gly Arg Phe Thr Val Ser Arg Asp Asn Ala Lys Asn Ser Leu Tyr 65 70
75 80 Leu Gln Leu Asn Ser Leu Arg Ala Glu Asp Thr Ala Val Tyr Tyr
Cys 85 90 95 Ala Arg Glu Asp Cys Thr Asn Gly Val Cys Tyr Arg Val
Asp Tyr Trp 100 105 110 Gly Gln Gly Thr Leu Val Thr Val Ser Ser 115
120 <210> SEQ ID NO 169 <211> LENGTH: 8 <212>
TYPE: PRT <213> ORGANISM: Artificial Sequence <220>
FEATURE: <223> OTHER INFORMATION: ZKA33 CDRH1 <400>
SEQUENCE: 169 Gly Phe Thr Phe Ser Arg Asn Trp 1 5 <210> SEQ
ID NO 170 <211> LENGTH: 8 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKA33 CDRH2 <400> SEQUENCE: 170 Ile Lys
Glu Asp Gly Asn Glu Lys 1 5 <210> SEQ ID NO 171 <211>
LENGTH: 15 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA33
CDRH3 <400> SEQUENCE: 171 Ala Arg Pro Phe His Gln Gly Gly Tyr
Ala Tyr Gly Leu Ala Tyr 1 5 10 15 <210> SEQ ID NO 172
<211> LENGTH: 122 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA33 VH <400> SEQUENCE: 172 Glu Val Gln Leu Val
Glu Ser Gly Gly Gly Leu Val Gln Pro Gly Gly 1 5 10 15 Ser Leu Arg
Leu Ser Cys Ala Ala Ser Gly Phe Thr Phe Ser Arg Asn 20 25 30 Trp
Met Thr Trp Val Arg Gln Ala Pro Gly Lys Gly Leu Glu Trp Val 35 40
45 Ala Asn Ile Lys Glu Asp Gly Asn Glu Lys Tyr Tyr Val Asp Ser Val
50 55 60 Lys Gly Arg Phe Thr Ile Ser Arg Asp Asn Ala Lys Asn Ser
Leu Tyr 65 70 75 80 Leu Gln Met Asn Ser Leu Arg Ala Glu Asp Thr Ala
Val Tyr Tyr Cys 85 90 95 Ala Arg Pro Phe His Gln Gly Gly Tyr Ala
Tyr Gly Leu Ala Tyr Trp 100 105 110 Gly Gln Gly Thr Leu Val Thr Val
Ser Ser 115 120 <210> SEQ ID NO 173 <211> LENGTH: 8
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA39 CDRH1
<400> SEQUENCE: 173 Gly Phe Thr Phe Ser Thr Tyr Ser 1 5
<210> SEQ ID NO 174 <211> LENGTH: 8 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA39 CDRH2 <400> SEQUENCE:
174 Ile Ser Pro Ser Ser Ser Thr Ile 1 5 <210> SEQ ID NO 175
<211> LENGTH: 15 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA39 CDRH3 <400> SEQUENCE: 175 Ala Arg Glu Tyr
Cys Ser Gly Gly Ser Cys Tyr Leu Leu Asp Tyr 1 5 10 15 <210>
SEQ ID NO 176 <211> LENGTH: 122 <212> TYPE: PRT
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA39 VH <400> SEQUENCE: 176
Glu Val Gln Leu Val Glu Ser Gly Gly Gly Leu Val Gln Pro Gly Gly 1 5
10 15 Ser Leu Arg Leu Ser Cys Ala Ala Ser Gly Phe Thr Phe Ser Thr
Tyr 20 25 30 Ser Met Asn Trp Val Arg Gln Ala Pro Gly Lys Gly Leu
Glu Trp Val 35 40 45 Ser Tyr Ile Ser Pro Ser Ser Ser Thr Ile Tyr
Tyr Pro Asp Ser Leu 50 55 60 Lys Gly Arg Phe Thr Ile Ser Arg Asp
Asn Ala Lys Asn Ser Leu Tyr 65 70 75 80 Leu Gln Met Asp Ser Leu Arg
Ala Glu Asp Thr Ala Gln Tyr Tyr Cys 85 90 95 Ala Arg Glu Tyr Cys
Ser Gly Gly Ser Cys Tyr Leu Leu Asp Tyr Trp 100 105 110 Gly Gln Gly
Thr Leu Val Thr Val Ser Ser 115 120 <210> SEQ ID NO 177
<211> LENGTH: 8 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA43 CDRH1 <400> SEQUENCE: 177 Gly Gly Ser Ile
Thr Ser Tyr Tyr 1 5 <210> SEQ ID NO 178 <211> LENGTH: 7
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA43 CDRH2
<400> SEQUENCE: 178 Ser His Tyr Ser Gly Ser Thr 1 5
<210> SEQ ID NO 179 <211> LENGTH: 13 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA43 CDRH3 <400> SEQUENCE:
179 Ala Arg Gly Ile Tyr Ser Gly Lys Asn Trp Phe Asp Pro 1 5 10
<210> SEQ ID NO 180 <211> LENGTH: 119 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA43 VH <400> SEQUENCE: 180
Gln Val Gln Leu Gln Glu Ser Gly Pro Gly Leu Val Lys Pro Ser Glu 1 5
10 15 Thr Leu Ser Leu Thr Cys Thr Val Tyr Gly Gly Ser Ile Thr Ser
Tyr 20 25 30 Tyr Trp Thr Trp Ile Arg Gln Pro Pro Gly Lys Gly Leu
Glu Trp Ile 35 40 45 Gly Tyr Ser His Tyr Ser Gly Ser Thr Asn Tyr
Asn Pro Ser Leu Lys 50 55 60 Ser Arg Val Thr Ile Ser Ile Asp Thr
Ser Lys Ser Gln Phe Ser Leu 65 70 75 80 Asn Leu Asn Ser Val Thr Ala
Ala Asp Thr Ala Val Tyr Tyr Cys Ala 85 90 95 Arg Gly Ile Tyr Ser
Gly Lys Asn Trp Phe Asp Pro Trp Gly Gln Gly 100 105 110 Thr Leu Val
Thr Val Ser Ser 115 <210> SEQ ID NO 181 <211> LENGTH: 8
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA44 CDRH1
<400> SEQUENCE: 181 Gly Phe Thr Val Ser Thr Ser Tyr 1 5
<210> SEQ ID NO 182 <211> LENGTH: 7 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA44 CDRH2 <400> SEQUENCE:
182 Ile Tyr Ser Ser Gly Ser Thr 1 5 <210> SEQ ID NO 183
<211> LENGTH: 10 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA44 CDRH3 <400> SEQUENCE: 183 Ala Arg Val Ser
Leu Gly Gly Leu Asp Pro 1 5 10 <210> SEQ ID NO 184
<211> LENGTH: 116 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA44 VH <400> SEQUENCE: 184 Glu Val Gln Leu Val
Glu Ser Gly Gly Gly Leu Ile Gln Pro Gly Gly 1 5 10 15 Ser Leu Arg
Leu Ser Cys Val Ala Ser Gly Phe Thr Val Ser Thr Ser 20 25 30 Tyr
Met Asn Trp Val Arg Gln Ala Pro Gly Lys Gly Leu Glu Trp Val 35 40
45 Ser Val Ile Tyr Ser Ser Gly Ser Thr Tyr Tyr Ala Asp Ser Val Lys
50 55 60 Gly Arg Phe Thr Ile Ser Arg Asn Thr Ser Lys Asn Thr Leu
Tyr Leu 65 70 75 80 Gln Met Asn Ser Leu Arg Ala Glu Asp Thr Ala Val
Tyr Tyr Cys Ala 85 90 95 Arg Val Ser Leu Gly Gly Leu Asp Pro Trp
Gly Gln Gly Thr Pro Val 100 105 110 Thr Val Ser Ser 115 <210>
SEQ ID NO 185 <211> LENGTH: 10 <212> TYPE: PRT
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA46 CDRH1 <400> SEQUENCE:
185 Gly Phe Ser Leu Ser Asn Gly Arg Met Gly 1 5 10 <210> SEQ
ID NO 186 <211> LENGTH: 7 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKA46 CDRH2 <400> SEQUENCE: 186 Ile Phe
Ser Asn Asp Glu Lys 1 5 <210> SEQ ID NO 187 <211>
LENGTH: 13 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA46
CDRH3 <400> SEQUENCE: 187 Ala Arg Val Glu Phe Arg Ala Gly Asn
Tyr Leu Asp Ser 1 5 10 <210> SEQ ID NO 188 <211>
LENGTH: 121 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA46
VH <400> SEQUENCE: 188 Gln Val Thr Leu Lys Glu Ser Gly Pro
Val Leu Val Lys Pro Thr Glu 1 5 10 15 Thr Leu Thr Leu Thr Cys Thr
Val Ser Gly Phe Ser Leu Ser Asn Gly 20 25 30 Arg Met Gly Val Ser
Trp Ile Arg Gln Pro Pro Gly Lys Ala Leu Glu 35 40 45 Trp Leu Ala
His Ile Phe Ser Asn Asp Glu Lys Tyr Tyr Ser Thr Ser 50 55 60 Leu
Lys Asn Arg Leu Thr Ile Ser Lys Asp Thr Ser Lys Ser Gln Val 65 70
75 80 Val Leu Thr Met Thr Asn Met Asp Pro Val Asp Thr Ala Thr Tyr
Tyr 85 90 95 Cys Ala Arg Val Glu Phe Arg Ala Gly Asn Tyr Leu Asp
Ser Trp Gly 100 105 110 Gln Gly Thr Leu Val Thr Val Ser Ser 115 120
<210> SEQ ID NO 189 <211> LENGTH: 8 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA50 CDRH1 <400> SEQUENCE:
189 Gly Tyr Thr Phe Thr Asn Ser Trp 1 5 <210> SEQ ID NO 190
<211> LENGTH: 8 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA50 CDRH2 <400> SEQUENCE: 190 Ile Tyr Pro Gly
Asp Ser Asp Thr 1 5 <210> SEQ ID NO 191 <211> LENGTH: 8
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA50 CDRH3
<400> SEQUENCE: 191 Ala Arg Gln Pro Phe Phe Asp Tyr 1 5
<210> SEQ ID NO 192 <211> LENGTH: 115 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA50 VH <400> SEQUENCE: 192
Glu Val Gln Leu Val Gln Ser Gly Ala Gln Val Lys Lys Pro Gly Glu 1 5
10 15 Ser Leu Lys Ile Ser Cys Lys Ala Ser Gly Tyr Thr Phe Thr Asn
Ser 20 25 30 Trp Ile Gly Trp Val Arg Gln Met Pro Gly Lys Gly Leu
Glu Trp Met 35 40 45 Gly Ile Ile Tyr Pro Gly Asp Ser Asp Thr Arg
Tyr Ser Pro Ser Phe 50 55 60 Gln Gly Gln Val Thr Ile Ser Ala Asp
Lys Ser Ile Ser Thr Ala Tyr 65 70 75 80 Leu Gln Trp Ser Ser Leu Lys
Ala Ser Asp Thr Ala Met Tyr Tyr Cys 85 90 95 Ala Arg Gln Pro Phe
Phe Asp Tyr Trp Gly Gln Gly Thr Leu Val Thr 100 105 110 Val Ser Ser
115 <210> SEQ ID NO 193 <211> LENGTH: 8 <212>
TYPE: PRT <213> ORGANISM: Artificial Sequence <220>
FEATURE: <223> OTHER INFORMATION: ZKA54 CDRH1 <400>
SEQUENCE: 193 Gly Tyr Thr Phe Thr Gly Tyr Tyr 1 5 <210> SEQ
ID NO 194 <211> LENGTH: 8 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKA54 CDRH2 <400> SEQUENCE: 194 Ile Asn
Ala Asn Ser Gly Gly Thr 1 5 <210> SEQ ID NO 195 <211>
LENGTH: 18 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA54
CDRH3 <400> SEQUENCE: 195 Ala His Ser Asp Ile Val Val Val Pro
Ser Asp Asp Tyr Tyr Ala Leu 1 5 10 15 Asp Val <210> SEQ ID NO
196 <211> LENGTH: 125 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKA54 VH <400> SEQUENCE: 196 Gln Val Gln
Leu Val Gln Ser Gly Ala Glu Val Lys Lys Pro Gly Ala 1 5 10 15 Ser
Val Lys Val Ser Cys Lys Thr Ser Gly Tyr Thr Phe Thr Gly Tyr 20 25
30 Tyr Met His Trp Val Arg Gln Ala Pro Gly Gln Gly Leu Glu Trp Met
35 40 45 Gly Trp Ile Asn Ala Asn Ser Gly Gly Thr Asn Phe Ala Gln
Arg Phe 50 55 60 Gln Gly Arg Val Thr Met Thr Trp Asp Thr Ser Ile
Ser Thr Ala Tyr 65 70 75 80 Met Glu Leu Ser Arg Leu Arg Ser Asp Asp
Thr Ala Val Tyr Tyr Cys 85 90 95 Ala His Ser Asp Ile Val Val Val
Pro Ser Asp Asp Tyr Tyr Ala Leu 100 105 110 Asp Val Trp Gly Gln Gly
Thr Thr Val Thr Val Ser Ser 115 120 125 <210> SEQ ID NO 197
<211> LENGTH: 8 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKB18 CDRH1 <400> SEQUENCE: 197 Gly Tyr Ser Phe
Thr Ser Tyr Trp 1 5 <210> SEQ ID NO 198 <211> LENGTH: 8
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKB18 CDRH2
<400> SEQUENCE: 198 Ile Tyr Pro Gly Asp Ser Asp Thr 1 5
<210> SEQ ID NO 199 <211> LENGTH: 8 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKB18 CDRH3 <400> SEQUENCE:
199 Ala Arg Gln Thr Pro Gly Asp Tyr 1 5 <210> SEQ ID NO 200
<211> LENGTH: 115 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKB18 VH <400> SEQUENCE: 200 Glu Val Gln Leu Val
Gln Ser Gly Ala Glu Val Lys Lys Pro Gly Glu 1 5 10 15 Ser Leu Lys
Ile Ser Cys Lys Thr Phe Gly Tyr Ser Phe Thr Ser Tyr 20 25 30 Trp
Ile Gly Trp Val Arg Gln Met Pro Gly Lys Gly Leu Glu Trp Met 35 40
45 Gly Met Ile Tyr Pro Gly Asp Ser Asp Thr Arg Tyr Ser Pro Ser Phe
50 55 60 Gln Gly Gln Val Thr Ile Ser Ala Asp Met Ser Ile Ser Thr
Ala Tyr 65 70 75 80 Leu Gln Trp Ser Ser Leu Lys Ala Ser Asp Thr Ala
Met Tyr Tyr Cys 85 90 95 Ala Arg Gln Thr Pro Gly Asp Tyr Trp Gly
Gln Gly Thr Leu Val Thr 100 105 110 Val Ser Ser 115 <210> SEQ
ID NO 201 <211> LENGTH: 8 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKB20 CDRH1 <400> SEQUENCE: 201 Gly Tyr
Phe Phe Thr Arg Tyr Val 1 5 <210> SEQ ID NO 202 <211>
LENGTH: 8 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKB20
CDRH2 <400> SEQUENCE: 202 Ile Asn Thr Asp Asn Gly Ser Thr 1 5
<210> SEQ ID NO 203 <211> LENGTH: 15 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKB20 CDRH3 <400> SEQUENCE:
203 Ala Arg Gly Thr Gly Arg Asp Gly Tyr Asn Ser Phe Phe Ala Asn 1 5
10 15 <210> SEQ ID NO 204 <211> LENGTH: 122 <212>
TYPE: PRT <213> ORGANISM: Artificial Sequence <220>
FEATURE: <223> OTHER INFORMATION: ZKB20 VH <400>
SEQUENCE: 204 Gln Val Gln Leu Val Gln Ser Gly Ala Glu Val Lys Lys
Pro Gly Ala 1 5 10 15 Ser Val Arg Val Ser Cys Lys Ala Ser Gly Tyr
Phe Phe Thr Arg Tyr 20 25 30 Val Ile Leu Trp Val Arg Gln Ala Pro
Gly Gln Arg Pro Glu Trp Met 35 40 45 Gly Trp Ile Asn Thr Asp Asn
Gly Ser Thr Arg Tyr Ser Gln Lys Phe 50 55 60 Gln Gly Arg Val Thr
Ile Thr Lys Asp Thr Ser Ala Thr Thr Ala Tyr 65 70 75 80 Met Asp Leu
Ser Ser Leu Lys Ser Asp Asp Thr Ala Val Tyr Tyr Cys 85 90 95 Ala
Arg Gly Thr Gly Arg Asp Gly Tyr Asn Ser Phe Phe Ala Asn Trp 100 105
110 Gly Gln Gly Thr Leu Val Thr Val Ser Pro 115 120 <210> SEQ
ID NO 205 <211> LENGTH: 8 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKB21 CDRH1 <400> SEQUENCE: 205 Gly Tyr
Thr Phe Thr Gly Tyr Ser 1 5 <210> SEQ ID NO 206 <211>
LENGTH: 8 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKB21
CDRH2 <400> SEQUENCE: 206 Ile Asp Thr Asn Ser Gly Asp Thr 1 5
<210> SEQ ID NO 207 <211> LENGTH: 11 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKB21 CDRH3 <400> SEQUENCE:
207 Ala Arg Asp Arg Glu Arg His Pro Phe Ser Tyr 1 5 10 <210>
SEQ ID NO 208 <211> LENGTH: 118 <212> TYPE: PRT
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKB21 VH <400> SEQUENCE: 208
Gln Val Gln Leu Val Gln Ser Gly Ala Glu Val Lys Lys Pro Gly Ala 1 5
10 15 Ser Val Lys Val Ser Cys Lys Ala Ser Gly Tyr Thr Phe Thr Gly
Tyr 20 25 30 Ser Ile His Trp Val Arg Gln Ala Pro Gly Gln Gly Leu
Ala Trp Met 35 40 45 Gly Arg Ile Asp Thr Asn Ser Gly Asp Thr Asn
Tyr Ala Glu Arg Phe 50 55 60 Gln Gly Arg Val Thr Met Thr Arg Asp
Thr Ser Ile Ser Thr Ala Tyr 65 70 75 80 Met Glu Val Arg Arg Leu Arg
Ser Asp Asp Thr Ala Val Tyr Tyr Cys 85 90 95 Ala Arg Asp Arg Glu
Arg His Pro Phe Ser Tyr Trp Gly Gln Gly Thr 100 105 110 Leu Val Thr
Val Ser Ser 115 <210> SEQ ID NO 209 <211> LENGTH: 10
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKB23 CDRH1
<400> SEQUENCE: 209 Gly Gly Ser Ile Ser Ser Gly Asp Tyr Ser 1
5 10 <210> SEQ ID NO 210 <211> LENGTH: 7 <212>
TYPE: PRT <213> ORGANISM: Artificial Sequence <220>
FEATURE: <223> OTHER INFORMATION: ZKB23 CDRH2 <400>
SEQUENCE: 210 Ile Thr His Ser Gly Thr Thr 1 5 <210> SEQ ID NO
211 <211> LENGTH: 9 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKB23 CDRH3 <400> SEQUENCE: 211 Ala Arg
His Phe Gly Trp Phe Asp Pro 1 5 <210> SEQ ID NO 212
<211> LENGTH: 117 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKB23 VH <400> SEQUENCE: 212 Gln Leu Gln Leu Gln
Glu Ser Gly Ser Gly Leu Val Lys Pro Ser Gln 1 5 10 15 Thr Leu Ser
Leu Thr Cys Ala Val Ser Gly Gly Ser Ile Ser Ser Gly 20 25 30 Asp
Tyr Ser Trp Ser Trp Ile Arg Gln Pro Pro Gly Lys Gly Leu Glu 35 40
45 Trp Ile Gly Tyr Ile Thr His Ser Gly Thr Thr Tyr Phe Asn Pro Ser
50 55 60 Leu Lys Ser Arg Val Thr Ile Ser Val Asp Arg Ser Arg Asn
Gln Phe 65 70 75 80 Ser Leu Lys Val Thr Ser Val Thr Ala Ala Asp Thr
Ala Val Tyr Tyr 85 90 95 Cys Ala Arg His Phe Gly Trp Phe Asp Pro
Trp Gly Gln Gly Thr Leu 100 105 110 Val Thr Val Ser Ser 115
<210> SEQ ID NO 213 <211> LENGTH: 10 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKC29 CDRH1 <400> SEQUENCE:
213 Gly Gly Ser Ile Ser Ser Gly Glu Tyr Phe 1 5 10 <210> SEQ
ID NO 214 <211> LENGTH: 7 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKC29 CDRH2 <400> SEQUENCE: 214 Ile His
Asn Arg Gly Asn Thr 1 5 <210> SEQ ID NO 215 <211>
LENGTH: 22 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKC29
CDRH3 <400> SEQUENCE: 215 Ala Arg Gly Gly Gly Asp Leu Val Val
Val Pro Asp Ser Ile Trp Asp 1 5 10 15 Tyr Tyr Gly Met Asp Val 20
<210> SEQ ID NO 216 <211> LENGTH: 130 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKC29 VH <400> SEQUENCE: 216
Gln Val Gln Leu Gln Glu Ser Gly Pro Gly Leu Val Arg Pro Ser Gln 1 5
10 15 Thr Leu Ser Leu Thr Cys Thr Val Ser Gly Gly Ser Ile Ser Ser
Gly 20 25 30 Glu Tyr Phe Trp Thr Trp Ile Arg Gln His Pro Lys Lys
Gly Leu Glu 35 40 45 Trp Ile Gly Tyr Ile His Asn Arg Gly Asn Thr
Tyr Tyr Asn Pro Ser 50 55 60 Leu Lys Ser Arg Leu Ser Ile Ser Leu
Asp Thr Ser Lys Asn His Leu 65 70 75 80 Ser Leu Arg Leu Ser Ser Val
Thr Ala Ala Asp Thr Ala Val Tyr Tyr 85 90 95 Cys Ala Arg Gly Gly
Gly Asp Leu Val Val Val Pro Asp Ser Ile Trp 100 105 110 Asp Tyr Tyr
Gly Met Asp Val Trp Gly Gln Gly Thr Thr Val Thr Val 115 120 125 Ser
Ser 130 <210> SEQ ID NO 217 <211> LENGTH: 10
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKC31 CDRH1
<400> SEQUENCE: 217 Gly Gly Ser Ile Ser Ser Gly Gly Tyr His 1
5 10 <210> SEQ ID NO 218 <211> LENGTH: 7 <212>
TYPE: PRT <213> ORGANISM: Artificial Sequence <220>
FEATURE: <223> OTHER INFORMATION: ZKC31 CDRH2 <400>
SEQUENCE: 218 Ile Tyr Tyr Ser Gly Ser Thr 1 5 <210> SEQ ID NO
219 <211> LENGTH: 19 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKC31 CDRH3 <400> SEQUENCE: 219 Ala Arg
Asp Arg Ser Glu Pro Gly Glu Tyr His Tyr Tyr Tyr Tyr Ala 1 5 10 15
Met Asp Val <210> SEQ ID NO 220 <211> LENGTH: 127
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKC31 VH
<400> SEQUENCE: 220 Gln Val Gln Leu Gln Glu Ser Gly Pro Gly
Leu Val Lys Pro Ser Gln 1 5 10 15 Thr Leu Ser Leu Thr Cys Thr Val
Ser Gly Gly Ser Ile Ser Ser Gly 20 25 30 Gly Tyr His Trp Ser Trp
Ile Arg Gln His Pro Gly Lys Gly Leu Glu 35 40 45 Trp Ile Gly Tyr
Ile Tyr Tyr Ser Gly Ser Thr Tyr Tyr Asn Pro Ser 50 55 60 Leu Lys
Arg Arg Val Thr Ile Ser Val Asp Thr Ser Lys Asn Gln Phe 65 70 75 80
Ser Leu Lys Leu Ser Ser Val Ser Ala Ala Asp Thr Ala Val Tyr Tyr 85
90 95 Cys Ala Arg Asp Arg Ser Glu Pro Gly Glu Tyr His Tyr Tyr Tyr
Tyr 100 105 110 Ala Met Asp Val Trp Gly Gln Gly Thr Thr Val Thr Val
Ser Ser 115 120 125 <210> SEQ ID NO 221 <211> LENGTH: 8
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKC32 CDRH1
<400> SEQUENCE: 221 Gly Phe Thr Val Ser Ser Asn Tyr 1 5
<210> SEQ ID NO 222 <211> LENGTH: 7 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKC32 CDRH2 <400> SEQUENCE:
222 Ile Tyr Ser Ser Gly Ser Thr 1 5 <210> SEQ ID NO 223
<211> LENGTH: 11 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKC32 CDRH3 <400> SEQUENCE: 223 Ala Arg Gly Lys
Lys Gly Asn Ala Phe Asp Ile 1 5 10 <210> SEQ ID NO 224
<211> LENGTH: 117 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKC32 VH <400> SEQUENCE: 224 Glu Val Gln Leu Val
Glu Ser Gly Gly Asp Leu Ile Gln Pro Gly Gly 1 5 10 15 Ser Leu Arg
Leu Ser Cys Ala Ala Ser Gly Phe Thr Val Ser Ser Asn 20 25 30 Tyr
Met Ser Trp Val Arg Gln Ala Pro Gly Lys Gly Leu Glu Trp Val 35 40
45 Ser Val Ile Tyr Ser Ser Gly Ser Thr Tyr Tyr Ala Asp Ser Val Lys
50 55 60 Gly Arg Phe Thr Ile Ser Arg Asp Asn Ser Lys Asn Thr Leu
Tyr Leu 65 70 75 80 Gln Met Asn Ser Leu Arg Ala Gly Asp Thr Ala Val
Tyr Tyr Cys Ala 85 90 95 Arg Gly Lys Lys Gly Asn Ala Phe Asp Ile
Trp Gly Gln Gly Thr Val 100 105 110 Val Thr Val Ser Ser 115
<210> SEQ ID NO 225 <211> LENGTH: 10 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKC33, CDRH1 <400> SEQUENCE:
225 Gly Asp Ser Ile Ser Ser Arg Thr Phe Ser 1 5 10 <210> SEQ
ID NO 226 <211> LENGTH: 7 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKC33 CDRH2 <400> SEQUENCE: 226 Ile Tyr
Tyr Ser Gly Ser Thr 1 5 <210> SEQ ID NO 227 <211>
LENGTH: 18 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKC33
CDRH3 <400> SEQUENCE: 227 Ala Arg Arg Asn Ala Glu Phe Phe Ser
Phe Trp Ser Tyr Tyr Gly Met 1 5 10 15 Asp Val <210> SEQ ID NO
228 <211> LENGTH: 126 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKC33 VH <400> SEQUENCE: 228 Gln Val Gln
Leu Gln Glu Ser Gly Pro Gly Leu Val Lys Pro Ser Gln 1 5 10 15 Thr
Leu Ser Leu Thr Cys Thr Val Ser Gly Asp Ser Ile Ser Ser Arg 20 25
30 Thr Phe Ser Trp Ser Trp Ile Arg Gln Pro Pro Gly Lys Gly Leu Glu
35 40 45 Trp Val Gly His Ile Tyr Tyr Ser Gly Ser Thr Asp Tyr Asn
Pro Ser 50 55 60 Leu Lys Ser Arg Ile Ser Ile Ser Ile Asp Thr Ser
Lys Asn Gln Phe 65 70 75 80 Ser Leu Lys Leu Ser Ser Val Thr Ala Ala
Asp Thr Ala Val Tyr Tyr 85 90 95 Cys Ala Arg Arg Asn Ala Glu Phe
Phe Ser Phe Trp Ser Tyr Tyr Gly 100 105 110 Met Asp Val Trp Gly His
Gly Thr Ala Val Ile Val Ser Ser 115 120 125 <210> SEQ ID NO
229 <211> LENGTH: 10 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKC34 CDRH1 <400> SEQUENCE: 229 Gly Gly
Ser Ile Asn Ser Gly Gly Tyr Tyr 1 5 10 <210> SEQ ID NO 230
<211> LENGTH: 7 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKC34 CDRH2 <400> SEQUENCE: 230 Ile Leu His Ser
Gly Asn Thr 1 5 <210> SEQ ID NO 231 <211> LENGTH: 15
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKC34 CDRH3
<400> SEQUENCE: 231 Ala Arg Ala Gly Asp Tyr Tyr Ser Gly Tyr
Val Pro Pro Glu Tyr 1 5 10 15 <210> SEQ ID NO 232 <211>
LENGTH: 123 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKC34
VH <400> SEQUENCE: 232 Gln Val Gln Leu Gln Glu Ser Gly Pro
Gly Leu Val Lys Pro Ser Gln 1 5 10 15 Thr Leu Ser Leu Thr Cys Ala
Val Ser Gly Gly Ser Ile Asn Ser Gly 20 25 30 Gly Tyr Tyr Trp Ser
Trp Val Arg Gln His Pro Gly Lys Gly Leu Glu 35 40 45 Trp Ile Gly
Tyr Ile Leu His Ser Gly Asn Thr Asn Tyr Asn Pro Ser 50 55 60 Leu
Lys Ser Arg Val Asn Ile Phe Val Asp Thr Ser Glu Asn Gln Phe 65 70
75 80 Ser Leu Lys Leu Arg Ser Val Thr Ala Ala Asp Thr Ala Ile Tyr
Phe 85 90 95 Cys Ala Arg Ala Gly Asp Tyr Tyr Ser Gly Tyr Val Pro
Pro Glu Tyr 100 105 110 Trp Gly Pro Gly Thr Leu Val Thr Val Ser Ser
115 120 <210> SEQ ID NO 233 <211> LENGTH: 8 <212>
TYPE: PRT <213> ORGANISM: Artificial Sequence <220>
FEATURE: <223> OTHER INFORMATION: ZKD25 CDRH1 <400>
SEQUENCE: 233 Gly Phe Thr Val Ser Ser Asn Tyr 1 5 <210> SEQ
ID NO 234 <211> LENGTH: 7 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKD25 CDRH2 <400> SEQUENCE: 234 Ile Tyr
Ser Gly Gly Ser Thr 1 5 <210> SEQ ID NO 235 <211>
LENGTH: 11 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKD25
CDRH3 <400> SEQUENCE: 235 Ala Arg Phe Gly Gly Asn Pro Ser Phe
Asp Tyr 1 5 10 <210> SEQ ID NO 236 <211> LENGTH: 117
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKD25 VH
<400> SEQUENCE: 236 Glu Val Gln Leu Val Glu Ser Gly Gly Gly
Leu Val Gln Pro Gly Gly 1 5 10 15 Ser Leu Arg Leu Ser Cys Ala Ala
Ser Gly Phe Thr Val Ser Ser Asn 20 25 30 Tyr Met Ser Trp Val Arg
Gln Ala Pro Gly Lys Gly Leu Glu Trp Val 35 40 45 Ser Val Ile Tyr
Ser Gly Gly Ser Thr Tyr Tyr Ala Asn Ser Val Lys 50 55 60 Gly Arg
Phe Thr Ile Ser Arg Asp Lys Ser Lys Asn Thr Leu Tyr Leu 65 70 75 80
Gln Met Asn Asn Leu Arg Ala Glu Asp Thr Ala Val Tyr Phe Cys Ala 85
90 95 Arg Phe Gly Gly Asn Pro Ser Phe Asp Tyr Trp Gly Gln Gly Thr
Leu 100 105 110 Val Thr Val Ser Ser 115 <210> SEQ ID NO 237
<211> LENGTH: 8 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA3 CDRH1 <400> SEQUENCE: 237 Gly Phe Ile Phe
Ser Asn Tyr Ala 1 5 <210> SEQ ID NO 238 <211> LENGTH: 8
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA3 CDRH2
<400> SEQUENCE: 238 Ile Gly Gly Lys Gly Asp Ser Ile 1 5
<210> SEQ ID NO 239 <211> LENGTH: 16 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA3 CDRH3 <400> SEQUENCE: 239
Val Lys Asp Leu Ala Val Leu Glu Ser Asp Arg Leu Glu Val Asp Gln 1 5
10 15 <210> SEQ ID NO 240 <211> LENGTH: 123 <212>
TYPE: PRT <213> ORGANISM: Artificial Sequence <220>
FEATURE: <223> OTHER INFORMATION: ZKA3 VH <400>
SEQUENCE: 240 Glu Val Gln Leu Ala Glu Ser Gly Gly Gly Leu Val Gln
Pro Gly Gly 1 5 10 15 Ser Leu Arg Leu Ser Cys Ser Gly Ser Gly Phe
Ile Phe Ser Asn Tyr 20 25 30 Ala Met Val Trp Ala Arg Gln Ala Pro
Gly Lys Gly Leu Glu Tyr Val 35 40 45 Ser Gly Ile Gly Gly Lys Gly
Asp Ser Ile Tyr His Ile Asp Ser Val 50 55 60 Lys Gly Arg Phe Thr
Ile Ser Arg Asp Asn Ser Lys Arg Thr Val Tyr 65 70 75 80 Leu Gln Met
Ser Arg Leu Arg Thr Glu Asp Thr Ala Val Tyr Tyr Cys 85 90 95 Val
Lys Asp Leu Ala Val Leu Glu Ser Asp Arg Leu Glu Val Asp Gln 100 105
110 Trp Gly Gln Gly Thr Leu Val Ile Val Ser Ala 115 120 <210>
SEQ ID NO 241 <211> LENGTH: 8 <212> TYPE: PRT
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA4 CDRH1 <400> SEQUENCE: 241
Gly Phe Thr Phe Ser Ser Tyr Val 1 5 <210> SEQ ID NO 242
<211> LENGTH: 8 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA4 CDRH2 <400> SEQUENCE: 242 Thr Ser Tyr Asp
Gly Ser Asn Lys 1 5 <210> SEQ ID NO 243 <211> LENGTH:
20 <212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA4 CDRH3
<400> SEQUENCE: 243 Ala Arg Gly Pro Val Pro Tyr Trp Ser Gly
Glu Ser Tyr Ser Gly Ala 1 5 10 15 Tyr Phe Asp Phe 20 <210>
SEQ ID NO 244 <211> LENGTH: 127 <212> TYPE: PRT
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA4 VH <400> SEQUENCE: 244
Gln Val Gln Leu Val Glu Ser Gly Gly Gly Val Val Gln Pro Gly Arg 1 5
10 15 Ser Leu Arg Leu Ser Cys Ala Ala Ser Gly Phe Thr Phe Ser Ser
Tyr 20 25 30 Val Met His Trp Val Arg Gln Ala Pro Gly Lys Gly Leu
Glu Trp Val 35 40 45 Thr Val Thr Ser Tyr Asp Gly Ser Asn Lys Tyr
Tyr Ala Asp Ser Val 50 55 60 Lys Gly Arg Phe Thr Ile Ser Arg Asp
Asn Ala Lys Asn Thr Leu Tyr 65 70 75 80 Leu Gln Met Asn Ser Leu Arg
Gly Glu Asp Thr Ala Ile Tyr Tyr Cys 85 90 95 Ala Arg Gly Pro Val
Pro Tyr Trp Ser Gly Glu Ser Tyr Ser Gly Ala 100 105 110 Tyr Phe Asp
Phe Trp Gly Gln Gly Ile Leu Val Thr Val Ser Ser 115 120 125
<210> SEQ ID NO 245 <211> LENGTH: 8 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA5 CDRH1 <400> SEQUENCE: 245
Gly Phe Thr Phe Ser Asn Tyr Tyr 1 5 <210> SEQ ID NO 246
<211> LENGTH: 8 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA5 CDRH2 <400> SEQUENCE: 246 Met Ser Ser Ser
Glu Thr Ile Lys 1 5 <210> SEQ ID NO 247 <211> LENGTH:
19 <212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA5 CDRH3
<400> SEQUENCE: 247 Ala Arg Ser Gly Ile Glu Thr Val Ala Gly
Ser Ile Asp Tyr Tyr Gly 1 5 10 15 Met Asp Val <210> SEQ ID NO
248 <211> LENGTH: 126 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKA5 VH <400> SEQUENCE: 248 Gln Val Gln
Leu Val Glu Ser Gly Gly Gly Leu Val Lys Pro Gly Gly 1 5 10 15 Ser
Leu Arg Leu Ser Cys Ala Gly Ser Gly Phe Thr Phe Ser Asn Tyr 20 25
30 Tyr Met Thr Trp Ile Arg Gln Ala Pro Gly Lys Gly Leu Glu Leu Val
35 40 45 Ser Tyr Met Ser Ser Ser Glu Thr Ile Lys Tyr Tyr Ala Asp
Ser Val 50 55 60 Lys Gly Arg Phe Thr Ile Ser Arg Asp Asn Ala Lys
Asn Ser Leu Tyr 65 70 75 80 Leu Gln Met Asn Ser Leu Arg Ala Asp Asp
Thr Ala Arg Tyr Tyr Cys 85 90 95 Ala Arg Ser Gly Ile Glu Thr Val
Ala Gly Ser Ile Asp Tyr Tyr Gly 100 105 110 Met Asp Val Trp Gly His
Gly Thr Pro Val Thr Val Ser Ser 115 120 125 <210> SEQ ID NO
249 <211> LENGTH: 8 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKA6 CDRH1 <400> SEQUENCE: 249 Asp Phe Thr
Val Ser Asn Tyr Ala 1 5 <210> SEQ ID NO 250 <211>
LENGTH: 8 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA6
CDRH2 <400> SEQUENCE: 250 Val Ser Tyr Asp Gly Ser Asn Lys 1 5
<210> SEQ ID NO 251 <211> LENGTH: 19 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA6 CDRH3 <400> SEQUENCE: 251
Ala Thr Gly Val Thr Met Phe Gln Gly Ala Gln Thr Asn Ala Glu Tyr 1 5
10 15 Leu His Tyr <210> SEQ ID NO 252 <211> LENGTH: 126
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA6 VH
<400> SEQUENCE: 252 Gln Val His Leu Val Glu Ser Gly Gly Gly
Val Val Gln Pro Gly Arg 1 5 10 15 Ser Leu Arg Leu Ser Cys Glu Ala
Ser Asp Phe Thr Val Ser Asn Tyr 20 25 30 Ala Met His Trp Val Arg
Gln Ala Pro Gly Lys Gly Leu Glu Trp Val 35 40 45 Ala Val Val Ser
Tyr Asp Gly Ser Asn Lys Tyr Tyr Ala Asp Ser Val 50 55 60 Lys Gly
Arg Phe Thr Ile Ser Arg Asp Asn Ser Lys Asn Thr Leu Tyr 65 70 75 80
Leu Gln Met Asn Ser Leu Arg Ala Glu Asp Thr Ala Leu Tyr Tyr Cys 85
90 95 Ala Thr Gly Val Thr Met Phe Gln Gly Ala Gln Thr Asn Ala Glu
Tyr 100 105 110 Leu His Tyr Trp Gly Gln Gly Ser Leu Val Thr Ile Ser
Ser 115 120 125 <210> SEQ ID NO 253 <211> LENGTH: 8
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA7 CDRH1
<400> SEQUENCE: 253 Gly Phe Thr Phe Ser Arg Tyr Gly 1 5
<210> SEQ ID NO 254 <211> LENGTH: 8 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA7 CDRH2 <400> SEQUENCE: 254
Val Ser Gly Asp Gly Ser Ser Thr 1 5 <210> SEQ ID NO 255
<211> LENGTH: 15 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA7 CDRH3 <400> SEQUENCE: 255 Val Lys Asp Phe
Trp Ser Gly Asp Gln Ser Leu Glu Ser Asp Phe 1 5 10 15 <210>
SEQ ID NO 256 <211> LENGTH: 122 <212> TYPE: PRT
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA7 VH <400> SEQUENCE: 256
Glu Val Gln Leu Val Glu Ser Gly Gly Gly Leu Val Gln Pro Gly Gly 1 5
10 15 Ser Leu Arg Leu Ser Cys Ser Ala Ser Gly Phe Thr Phe Ser Arg
Tyr 20 25 30 Gly Met Val Trp Ala Arg Gln Ala Pro Gly Lys Gly Leu
Glu Tyr Leu 35 40 45 Ser Gly Val Ser Gly Asp Gly Ser Ser Thr Tyr
Tyr Ala Asn Ser Val 50 55 60 Lys Gly Arg Phe Thr Ile Ser Arg Asp
Asn Ser Lys Asn Thr Leu Tyr 65 70 75 80 Leu His Met Ser Arg Leu Arg
Asp Glu Asp Thr Ala Met Tyr Tyr Cys 85 90 95 Val Lys Asp Phe Trp
Ser Gly Asp Gln Ser Leu Glu Ser Asp Phe Trp 100 105 110 Gly Gln Gly
Ala Leu Val Thr Val Ser Ser 115 120 <210> SEQ ID NO 257
<211> LENGTH: 8 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA8 CDRH1 <400> SEQUENCE: 257 Gly Phe Thr Phe
Ser Ala His Ala 1 5 <210> SEQ ID NO 258 <211> LENGTH: 8
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA8 CDRH2
<400> SEQUENCE: 258 Ile Ser Arg Asn Glu Asp Tyr Thr 1 5
<210> SEQ ID NO 259 <211> LENGTH: 12 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA8 CDRH3 <400> SEQUENCE: 259
Val Lys Asp Phe Gly Thr Ser Pro Gln Thr Asp Phe 1 5 10 <210>
SEQ ID NO 260 <211> LENGTH: 119 <212> TYPE: PRT
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA8 VH <400> SEQUENCE: 260
Asp Glu Arg Leu Val Glu Ser Gly Gly Gly Leu Val Gln Pro Gly Gly 1 5
10 15 Ser Leu Arg Leu Val Cys Ser Ala Ser Gly Phe Thr Phe Ser Ala
His 20 25 30 Ala Met His Trp Val Arg Gln Pro Pro Gly Lys Gly Leu
Glu Tyr Val 35 40 45 Ser Thr Ile Ser Arg Asn Glu Asp Tyr Thr Tyr
Tyr Ala Asp Ser Val 50 55 60 Lys Gly Arg Phe Thr Ile Ser Arg Asp
Asn Ser Lys Asn Ser Leu Tyr 65 70 75 80 Leu Gln Met Arg Arg Leu Arg
Pro Glu Asp Thr Ala Ile Tyr Tyr Cys 85 90 95 Val Lys Asp Phe Gly
Thr Ser Pro Gln Thr Asp Phe Trp Gly Gln Gly 100 105 110 Thr Leu Val
Ala Val Ser Ser 115 <210> SEQ ID NO 261 <211> LENGTH: 8
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA76 CDRH1
<400> SEQUENCE: 261 Gly Phe Thr Phe Ser Thr Tyr Phe 1 5
<210> SEQ ID NO 262 <211> LENGTH: 8 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA76 CDRH2 <400> SEQUENCE:
262 Ile Ser Ser Thr Gly Ser Tyr Lys 1 5 <210> SEQ ID NO 263
<211> LENGTH: 17 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA76 CDRH3 <400> SEQUENCE: 263 Ala Arg Pro Phe
His Ser Glu Tyr Thr Tyr Gly Leu Asp Ala Phe Asp 1 5 10 15 Ile
<210> SEQ ID NO 264 <211> LENGTH: 124 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA76 VH <400> SEQUENCE: 264
Glu Val Gln Leu Val Glu Ser Gly Gly Gly Leu Val Lys Pro Gly Gly 1 5
10 15 Ser Leu Arg Leu Ser Cys Ala Ala Ser Gly Phe Thr Phe Ser Thr
Tyr 20 25 30 Phe Met His Trp Val Arg Gln Ala Pro Gly Lys Gly Leu
Glu Trp Val 35 40 45 Ala Ser Ile Ser Ser Thr Gly Ser Tyr Lys Phe
Tyr Ala Asp Ser Val 50 55 60 Lys Gly Arg Phe Thr Ile Ser Arg Asp
Asn Thr Lys Asn Ser Leu Phe 65 70 75 80 Leu Gln Met Asn Ser Leu Arg
Ala Glu Asp Thr Ala Val Phe Tyr Cys 85 90 95 Ala Arg Pro Phe His
Ser Glu Tyr Thr Tyr Gly Leu Asp Ala Phe Asp 100 105 110 Ile Trp Gly
Gln Gly Thr Met Leu Thr Val Ser Ser 115 120 <210> SEQ ID NO
265 <211> LENGTH: 10 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKA117 CDRH1 <400> SEQUENCE: 265 Gly Gly
Ser Ile Arg Arg Thr Asn Ser Tyr 1 5 10 <210> SEQ ID NO 266
<211> LENGTH: 7 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA117 CDRH2 <400> SEQUENCE: 266 Ile Ser Tyr Ser
Gly Ser Thr 1 5 <210> SEQ ID NO 267 <211> LENGTH: 17
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA117 CDRH3
<400> SEQUENCE: 267 Ala Arg Leu Asn Asp Gly Ser Thr Val Thr
Thr Ser Ser Tyr Phe Asp 1 5 10 15 Tyr <210> SEQ ID NO 268
<211> LENGTH: 125 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA117 VH <400> SEQUENCE: 268 Gln Leu Gln Leu
Gln Glu Ser Gly Pro Gly Leu Val Lys Pro Ser Glu 1 5 10 15 Thr Leu
Ser Leu Thr Cys Thr Val Ser Gly Gly Ser Ile Arg Arg Thr 20 25 30
Asn Ser Tyr Trp Gly Trp Ile Arg Gln Thr Thr Gly Lys Gly Leu Gln 35
40 45 Trp Ile Gly Ser Ile Ser Tyr Ser Gly Ser Thr Phe Tyr Asn Pro
Ser 50 55 60 Leu Lys Ser Arg Val Thr Ile Ser Leu Asp Thr Ser Lys
Asp His Phe 65 70 75 80 Ser Leu Glu Leu Ser Ser Val Thr Ala Ala Asp
Thr Ala Ile Tyr Tyr 85 90 95 Cys Ala Arg Leu Asn Asp Gly Ser Thr
Val Thr Thr Ser Ser Tyr Phe 100 105 110 Asp Tyr Trp Gly Gln Gly Thr
Leu Val Thr Val Ser Ser 115 120 125 <210> SEQ ID NO 269
<211> LENGTH: 8 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKB27 CDRH1 <400> SEQUENCE: 269 Gly Tyr Ser Phe
Thr Ser Ser Trp 1 5 <210> SEQ ID NO 270 <211> LENGTH: 8
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKB27 CDRH2
<400> SEQUENCE: 270 Ile Asp Pro Ser Asp Ser Tyr Thr 1 5
<210> SEQ ID NO 271 <211> LENGTH: 14 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKB27 CDRH3 <400> SEQUENCE:
271 Ala Arg His Asp Tyr Ser Val Ser Glu Asn Gly Met Asp Val 1 5 10
<210> SEQ ID NO 272 <211> LENGTH: 121 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKB27 VH <400> SEQUENCE: 272
Glu Val Gln Leu Val Gln Ser Gly Ala Glu Val Lys Lys Pro Gly Glu 1 5
10 15 Ser Leu Arg Ile Ser Cys Lys Ala Ser Gly Tyr Ser Phe Thr Ser
Ser 20 25 30 Trp Ile Asn Trp Val Arg Gln Met Pro Gly Lys Gly Leu
Glu Trp Met 35 40 45 Gly Arg Ile Asp Pro Ser Asp Ser Tyr Thr Thr
Tyr Asn Pro Ser Phe 50 55 60 Gln Gly His Val Thr Ile Ser Val Asp
Lys Ser Ile Gly Thr Ala Tyr 65 70 75 80 Leu Gln Trp Asn Ser Leu Arg
Ala Ser Asp Thr Ala Met Tyr Tyr Cys 85 90 95 Ala Arg His Asp Tyr
Ser Val Ser Glu Asn Gly Met Asp Val Trp Gly 100 105 110 Gln Gly Thr
Thr Val Thr Val Ser Ser 115 120 <210> SEQ ID NO 273
<211> LENGTH: 8 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKB29 CDRH1 <400> SEQUENCE: 273 Gly Phe Thr Phe
Ser Ser Tyr Thr 1 5 <210> SEQ ID NO 274 <211> LENGTH: 8
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKB29 CDRH2
<400> SEQUENCE: 274 Ile Ser Tyr Asp Gly Ser His Lys 1 5
<210> SEQ ID NO 275 <211> LENGTH: 12 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKB29 CDRH3 <400> SEQUENCE:
275 Ala Arg Arg Ser Tyr Ser Ile Ser Cys Phe Asp Tyr 1 5 10
<210> SEQ ID NO 276 <211> LENGTH: 119 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKB29 VH <400> SEQUENCE: 276
Gln Val Gln Leu Val Glu Ser Gly Gly Gly Val Val Gln Pro Gly Arg 1 5
10 15 Ser Leu Arg Leu Ser Cys Ala Ala Ser Gly Phe Thr Phe Ser Ser
Tyr 20 25 30 Thr Met His Trp Val Arg Gln Ala Pro Gly Lys Gly Leu
Glu Trp Val 35 40 45 Ala Val Ile Ser Tyr Asp Gly Ser His Lys Phe
Tyr Ala Asp Ser Val 50 55 60 Lys Gly Arg Phe Thr Ile Ser Arg Asp
Asn Ser Lys Asp Thr Leu Tyr 65 70 75 80 Leu Gln Met Asn Ser Leu Arg
Ala Glu Asp Thr Ala Leu Tyr Tyr Cys 85 90 95 Ala Arg Arg Ser Tyr
Ser Ile Ser Cys Phe Asp Tyr Trp Gly Gln Gly 100 105 110 Thr Leu Val
Thr Ile Ser Ser 115 <210> SEQ ID NO 277 <211> LENGTH: 8
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKB34 CDRH1
<400> SEQUENCE: 277 Gly Phe Thr Phe Ser Arg Ser Gly 1 5
<210> SEQ ID NO 278 <211> LENGTH: 8 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKB34 CDRH2 <400> SEQUENCE:
278 Val Ser Tyr Asp Gly Ser Asn Lys 1 5 <210> SEQ ID NO 279
<211> LENGTH: 19 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKB34 CDRH3 <400> SEQUENCE: 279 Ala Lys Asp Leu
Thr Met Val Arg Gly Val His Tyr Tyr Tyr Tyr Val 1 5 10 15 Met Asp
Val <210> SEQ ID NO 280 <211> LENGTH: 126 <212>
TYPE: PRT <213> ORGANISM: Artificial Sequence <220>
FEATURE: <223> OTHER INFORMATION: ZKB34 VH <400>
SEQUENCE: 280 Gln Val Gln Leu Val Glu Ser Gly Gly Gly Val Val Gln
Pro Gly Arg 1 5 10 15 Ser Leu Arg Leu Ser Cys Ala Ala Ser Gly Phe
Thr Phe Ser Arg Ser 20 25 30 Gly Met His Trp Val Arg Gln Ala Pro
Gly Lys Gly Leu Glu Trp Val 35 40 45 Ala Val Val Ser Tyr Asp Gly
Ser Asn Lys Tyr Tyr Ser Asp Ser Val 50 55 60 Lys Gly Arg Phe Thr
Ile Ser Arg Asp Asn Ser Lys Asn Thr Leu Tyr 65 70 75 80 Leu Gln Met
Asn Ser Leu Arg Val Glu Asp Thr Ala Val Tyr Tyr Cys 85 90 95 Ala
Lys Asp Leu Thr Met Val Arg Gly Val His Tyr Tyr Tyr Tyr Val 100 105
110 Met Asp Val Trp Gly Gln Gly Thr Thr Val Thr Val Ser Ser 115 120
125 <210> SEQ ID NO 281 <211> LENGTH: 8 <212>
TYPE: PRT <213> ORGANISM: Artificial Sequence <220>
FEATURE: <223> OTHER INFORMATION: ZKB39 CDRH1 <400>
SEQUENCE: 281 Gly Tyr Thr Phe Asp Asp Tyr Tyr 1 5 <210> SEQ
ID NO 282 <211> LENGTH: 8 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKB39 CDRH2 <400> SEQUENCE: 282 Ile Asn
Pro His Arg Gly Gly Thr 1 5 <210> SEQ ID NO 283 <211>
LENGTH: 23 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKB39
CDRH3 <400> SEQUENCE: 283 Val Arg Asp Gln Tyr Cys Asp Gly Gly
Asn Cys Tyr Gly Ile His Gln 1 5 10 15 Pro His Tyr Gly Met Asp Val
20 <210> SEQ ID NO 284 <211> LENGTH: 130 <212>
TYPE: PRT <213> ORGANISM: Artificial Sequence <220>
FEATURE: <223> OTHER INFORMATION: ZKB39 VH <400>
SEQUENCE: 284 Gln Val Gln Leu Val Gln Ser Gly Ala Glu Val Lys Lys
Pro Gly Ala 1 5 10 15 Ser Leu Lys Val Ser Cys Lys Ala Ser Gly Tyr
Thr Phe Asp Asp Tyr 20 25 30 Tyr Ile His Trp Val Arg Gln Ala Pro
Gly Gln Gly Leu Glu Trp Leu 35 40 45 Gly Arg Ile Asn Pro His Arg
Gly Gly Thr Asn Tyr Ala Gln Lys Phe 50 55 60 Gln Gly Arg Val Ile
Met Thr Leu Asp Met Ser Ile Ser Thr Thr Tyr 65 70 75 80 Met Glu Leu
Arg Arg Ile Thr Ser Asp Asp Ala Ala Val Tyr Tyr Cys 85 90 95 Val
Arg Asp Gln Tyr Cys Asp Gly Gly Asn Cys Tyr Gly Ile His Gln 100 105
110 Pro His Tyr Gly Met Asp Val Trp Gly Gln Gly Thr Thr Val Thr Val
115 120 125 Ser Ser 130 <210> SEQ ID NO 285 <211>
LENGTH: 8 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKB46
CDRH1 <400> SEQUENCE: 285 Gly Tyr Ser Phe Thr Ser Tyr Trp 1 5
<210> SEQ ID NO 286 <211> LENGTH: 8 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKB46 CDRH2 <400> SEQUENCE:
286 Ile Asp Pro Ser Asp Ser Tyr Thr 1 5 <210> SEQ ID NO 287
<211> LENGTH: 17 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKB46 CDRH3 <400> SEQUENCE: 287 Ala Arg Arg Glu
Tyr Ser Ser Ser Ser Gly Gln Glu Asp Trp Phe Asp 1 5 10 15 Pro
<210> SEQ ID NO 288 <211> LENGTH: 124 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKB46 VH <400> SEQUENCE: 288
Glu Val Gln Leu Val Gln Ser Gly Ala Glu Val Lys Lys Pro Gly Glu 1 5
10 15 Ser Leu Arg Ile Ser Cys Lys Gly Ser Gly Tyr Ser Phe Thr Ser
Tyr 20 25 30 Trp Ile Ser Trp Val Arg Gln Met Pro Gly Lys Gly Leu
Glu Trp Met 35 40 45 Gly Arg Ile Asp Pro Ser Asp Ser Tyr Thr Asn
Tyr Ser Pro Ser Phe 50 55 60 Gln Gly His Val Thr Ile Ser Ala Asp
Lys Ser Ile Ser Thr Ala Tyr 65 70 75 80 Leu Gln Trp Ser Ser Leu Lys
Ala Ser Asp Thr Ala Met Tyr Tyr Cys 85 90 95 Ala Arg Arg Glu Tyr
Ser Ser Ser Ser Gly Gln Glu Asp Trp Phe Asp 100 105 110 Pro Trp Gly
Gln Gly Thr Leu Val Thr Val Ser Ser 115 120 <210> SEQ ID NO
289 <211> LENGTH: 8 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKB53 CDRH1 <400> SEQUENCE: 289 Gly Phe
Thr Phe Ser Ser Tyr Ala 1 5 <210> SEQ ID NO 290 <211>
LENGTH: 8 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKB53
CDRH2 <400> SEQUENCE: 290 Ile Ser Tyr Asp Gly Ser Asn Arg 1 5
<210> SEQ ID NO 291 <211> LENGTH: 15 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKB53 CDRH3 <400> SEQUENCE:
291 Ala Arg His Val Glu Gln Leu Pro Ser Ser Gly Tyr Phe Gln His 1 5
10 15 <210> SEQ ID NO 292 <211> LENGTH: 122 <212>
TYPE: PRT <213> ORGANISM: Artificial Sequence <220>
FEATURE: <223> OTHER INFORMATION: ZKB53 VH <400>
SEQUENCE: 292 Gln Val Gln Leu Val Glu Ser Gly Gly Gly Val Val Gln
Pro Gly Arg 1 5 10 15 Ser Leu Arg Leu Ser Cys Ala Ala Ser Gly Phe
Thr Phe Ser Ser Tyr 20 25 30 Ala Met His Trp Val Arg Gln Thr Pro
Gly Lys Gly Leu Glu Trp Val 35 40 45 Thr Val Ile Ser Tyr Asp Gly
Ser Asn Arg Tyr Tyr Ala Asp Ser Val 50 55 60 Lys Gly Arg Phe Thr
Ile Ser Arg Asp Asn Ser Lys Asn Thr Leu Tyr 65 70 75 80 Leu Gln Met
Asn Ser Leu Arg Ser Glu Asp Thr Ala Val Tyr Tyr Cys 85 90 95 Ala
Arg His Val Glu Gln Leu Pro Ser Ser Gly Tyr Phe Gln His Trp 100 105
110 Gly Gln Gly Thr Leu Val Thr Val Ser Ser 115 120 <210> SEQ
ID NO 293 <211> LENGTH: 8 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKC26 CDRH1 <400> SEQUENCE: 293 Gly Phe
Ile Phe Ser Asp Phe Tyr 1 5 <210> SEQ ID NO 294 <211>
LENGTH: 8 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKC26
CDRH2 <400> SEQUENCE: 294 Ile Gly His Asp Gly Ser Tyr Ile 1 5
<210> SEQ ID NO 295 <211> LENGTH: 9 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKC26 CDRH3 <400> SEQUENCE:
295 Ala Arg Ala His Gly Gly Phe Arg His 1 5 <210> SEQ ID NO
296 <211> LENGTH: 116 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKC26 VH <400> SEQUENCE: 296 Gln Val Gln
Val Val Glu Ser Gly Gly Gly Leu Val Lys Pro Gly Gly 1 5 10 15 Ser
Leu Arg Leu Ser Cys Ala Ala Ser Gly Phe Ile Phe Ser Asp Phe 20 25
30 Tyr Met Ser Trp Met Arg Gln Ala Pro Gly Lys Gly Leu Glu Trp Val
35 40 45 Ala Tyr Ile Gly His Asp Gly Ser Tyr Ile Leu Tyr Ala Asp
Ser Val 50 55 60 Lys Gly Arg Phe Thr Ile Ser Arg Asp Asn Ala Lys
Asn Ser Leu Phe 65 70 75 80 Leu Arg Met Asn Ser Leu Arg Val Glu Asp
Thr Ala Val Tyr Tyr Cys 85 90 95 Ala Arg Ala His Gly Gly Phe Arg
His Trp Gly Gln Gly Thr Val Val 100 105 110 Ala Val Ser Pro 115
<210> SEQ ID NO 297 <211> LENGTH: 8 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKD5 CDRH1 <400> SEQUENCE: 297
Gly Phe Thr Phe Thr Ser Tyr Gly 1 5 <210> SEQ ID NO 298
<211> LENGTH: 8 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKD5 CDRH2 <400> SEQUENCE: 298 Ile Ser Tyr Asp
Gly Ser Asn Lys 1 5 <210> SEQ ID NO 299 <211> LENGTH:
17 <212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKD5 CDRH3
<400> SEQUENCE: 299 Ala Arg Asp Arg Asp His Tyr Asp Leu Trp
Asn Ala Tyr Thr Phe Asp 1 5 10 15 Tyr <210> SEQ ID NO 300
<211> LENGTH: 124 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKD5 VH <400> SEQUENCE: 300 Gln Val Gln Leu Val
Glu Ser Gly Gly Gly Val Val Gln Pro Gly Arg 1 5 10 15 Ser Leu Arg
Leu Ser Cys Ala Ala Ser Gly Phe Thr Phe Thr Ser Tyr 20 25 30 Gly
Met His Trp Val Arg Gln Thr Pro Gly Lys Gly Leu Asp Trp Val 35 40
45 Ala Val Ile Ser Tyr Asp Gly Ser Asn Lys Tyr Tyr Ala Asp Ser Val
50 55 60 Lys Gly Arg Phe Thr Ile Ser Arg Asp Asn Ser Lys Asp Thr
Leu Tyr 65 70 75 80 Leu Gln Met Asn Ser Leu Arg Ala Ala Asp Thr Ala
Leu Tyr Tyr Cys 85 90 95 Ala Arg Asp Arg Asp His Tyr Asp Leu Trp
Asn Ala Tyr Thr Phe Asp 100 105 110 Tyr Trp Gly Gln Gly Thr Leu Val
Thr Val Ser Ser 115 120 <210> SEQ ID NO 301 <211>
LENGTH: 8 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKD7
CDRH1 <400> SEQUENCE: 301 Gly Phe Thr Phe Ser Asn Tyr Ala 1 5
<210> SEQ ID NO 302 <211> LENGTH: 8 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKD7 CDRH2 <400> SEQUENCE: 302
Ile Ser Tyr Asp Val Ser Asp Lys 1 5 <210> SEQ ID NO 303
<211> LENGTH: 20 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKD7 CDRH3 <400> SEQUENCE: 303 Ala Gly Gly Pro
Leu Gly Val Val Val Ile Lys Pro Ser Asn Ala Glu 1 5 10 15 His Phe
His His 20 <210> SEQ ID NO 304 <211> LENGTH: 127
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKD7 VH
<400> SEQUENCE: 304 Gln Val Gln Leu Val Glu Ser Gly Gly Gly
Val Val Gln Pro Gly Arg 1 5 10 15 Ser Leu Arg Leu Ser Cys Ala Ala
Ser Gly Phe Thr Phe Ser Asn Tyr 20 25 30 Ala Met His Trp Val Arg
Gln Ala Pro Gly Lys Gly Leu Glu Trp Val 35 40 45 Ala Val Ile Ser
Tyr Asp Val Ser Asp Lys Tyr Tyr Ala Asp Ser Val 50 55 60 Lys Gly
Arg Phe Thr Ile Ser Arg Asp Asn Ser Lys Asn Thr Leu Phe 65 70 75 80
Leu Gln Met Asn Ser Leu Arg Ala Glu Asp Thr Ala Ala Tyr Tyr Cys 85
90 95 Ala Gly Gly Pro Leu Gly Val Val Val Ile Lys Pro Ser Asn Ala
Glu 100 105 110 His Phe His His Trp Gly Gln Gly Thr Leu Val Thr Val
Ser Ser 115 120 125 <210> SEQ ID NO 305 <211> LENGTH: 8
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKD8 CDRH1
<400> SEQUENCE: 305 Gly Phe Thr Phe Ile Asn Tyr Ala 1 5
<210> SEQ ID NO 306 <211> LENGTH: 8 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKD8 CDRH2 <400> SEQUENCE: 306
Ile Ser Tyr Asp Gly Ser Asn Lys 1 5 <210> SEQ ID NO 307
<211> LENGTH: 16 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKD8 CDRH3 <400> SEQUENCE: 307 Ala Thr Asp Ala
Asp Ala Tyr Gly Asp Ser Gly Ala Asn Phe His Tyr 1 5 10 15
<210> SEQ ID NO 308 <211> LENGTH: 123 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKD8 VH <400> SEQUENCE: 308
Gln Val Gln Leu Val Glu Ser Gly Gly Gly Val Val Gln Pro Gly Lys 1 5
10 15 Ser Leu Arg Leu Ser Cys Ala Ala Ser Gly Phe Thr Phe Ile Asn
Tyr 20 25 30 Ala Ile His Trp Val Arg Gln Ala Pro Gly Lys Gly Leu
Glu Trp Val 35 40 45 Ala Val Ile Ser Tyr Asp Gly Ser Asn Lys Phe
Tyr Thr Asp Ser Val 50 55 60 Lys Gly Arg Phe Thr Ile Ser Arg Asp
Asn Ser Lys Asn Thr Leu Tyr 65 70 75 80 Leu Gln Met Asn Ser Leu Arg
Ala Asp Asp Thr Ala Val Tyr Tyr Cys 85 90 95 Ala Thr Asp Ala Asp
Ala Tyr Gly Asp Ser Gly Ala Asn Phe His Tyr 100 105 110 Trp Gly Gln
Gly Thr Leu Val Thr Val Ser Ser 115 120 <210> SEQ ID NO 309
<211> LENGTH: 10 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKD15 CDRH1 <400> SEQUENCE: 309 Asp Ala Ser Ile
Ser Ser Gly Gly Phe Ser 1 5 10 <210> SEQ ID NO 310
<211> LENGTH: 7 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKD15 CDRH2 <400> SEQUENCE: 310 Ile Tyr Ser Ser
Gly Asp Thr 1 5 <210> SEQ ID NO 311 <211> LENGTH: 18
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKD15 CDRH3
<400> SEQUENCE: 311 Ala Arg Ala His Thr Pro Thr Ser Lys Phe
Tyr Tyr Tyr Tyr Ala Met 1 5 10 15 Asp Val <210> SEQ ID NO 312
<211> LENGTH: 126 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKD15 VH <400> SEQUENCE: 312 Gln Leu Gln Leu Gln
Glu Ser Gly Ser Gly Leu Val Lys Pro Ser Gln 1 5 10 15 Thr Leu Ser
Leu Thr Cys Thr Val Ser Asp Ala Ser Ile Ser Ser Gly 20 25 30 Gly
Phe Ser Trp Ser Trp Ile Arg Gln Pro Leu Gly Lys Gly Leu Glu 35 40
45 Trp Leu Gly Tyr Ile Tyr Ser Ser Gly Asp Thr Phe Tyr Asn Pro Ser
50 55 60 Leu Gln Gly Arg Val Thr Met Ser Val Asp Ile Phe Arg Ser
Gln Phe 65 70 75 80 Ser Leu Lys Leu Thr Ser Val Thr Ala Ala Asp Thr
Ala Met Tyr Tyr 85 90 95 Cys Ala Arg Ala His Thr Pro Thr Ser Lys
Phe Tyr Tyr Tyr Tyr Ala 100 105 110 Met Asp Val Trp Gly Gln Gly Thr
Thr Val Thr Val Ser Ser 115 120 125 <210> SEQ ID NO 313
<211> LENGTH: 8 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKD16 CDRH1 <400> SEQUENCE: 313 Gly Phe Thr Phe
Ser Asp His Phe 1 5 <210> SEQ ID NO 314 <211> LENGTH:
10 <212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKD16 CDRH2
<400> SEQUENCE: 314 Ser Arg Asn Lys Pro Asn Ser Tyr Thr Thr 1
5 10 <210> SEQ ID NO 315 <211> LENGTH: 17 <212>
TYPE: PRT <213> ORGANISM: Artificial Sequence <220>
FEATURE: <223> OTHER INFORMATION: ZKD16 CDRH3 <400>
SEQUENCE: 315 Ala Lys Val Gly Gly Cys Tyr Gly Gly Asp Cys His Val
Glu Asn Asp 1 5 10 15 Tyr <210> SEQ ID NO 316 <211>
LENGTH: 126 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKD16
VH <400> SEQUENCE: 316 Glu Val Gln Leu Val Glu Ser Gly Gly
Asp Leu Val Gln Pro Gly Gly 1 5 10 15 Ser Leu Arg Leu Ser Cys Val
Ala Ser Gly Phe Thr Phe Ser Asp His 20 25 30 Phe Met Asp Trp Val
Arg Gln Ala Pro Gly Lys Gly Leu Glu Trp Val 35 40 45 Gly Arg Ser
Arg Asn Lys Pro Asn Ser Tyr Thr Thr Glu Tyr Ala Ala 50 55 60 Ser
Val Lys Gly Arg Phe Ser Ile Ser Arg Asp Asp Ser Lys Lys Ala 65 70
75 80 Leu Tyr Leu Gln Met Asn Ser Leu Gln Thr Glu Asp Thr Ala Val
Tyr 85 90 95 Tyr Cys Ala Lys Val Gly Gly Cys Tyr Gly Gly Asp Cys
His Val Glu 100 105 110 Asn Asp Tyr Trp Gly Gln Gly Thr Leu Val Thr
Val Ser Ser 115 120 125 <210> SEQ ID NO 317 <211>
LENGTH: 8 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKD17
CDRH1 <400> SEQUENCE: 317 Gly Phe Ile Phe Ser Asp Tyr Ala 1 5
<210> SEQ ID NO 318 <211> LENGTH: 8 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKD17 CDRH2 <400> SEQUENCE:
318 Ile Ser Tyr Asp Gly Ser Ser Arg 1 5 <210> SEQ ID NO 319
<211> LENGTH: 20 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKD17 CDRH3 <400> SEQUENCE: 319 Ala Arg Gly Tyr
Cys Ser Ser Gly Thr Cys Phe Ser Thr Asn Ala Glu 1 5 10 15 Tyr Phe
His Pro 20 <210> SEQ ID NO 320 <211> LENGTH: 127
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKD17 VH
<400> SEQUENCE: 320 Gln Val Gln Met Val Glu Ser Gly Gly Gly
Val Val Gln Pro Gly Thr 1 5 10 15 Ser Leu Arg Leu Ser Cys Ala Thr
Ser Gly Phe Ile Phe Ser Asp Tyr 20 25 30 Ala Met His Trp Val Arg
Gln Ala Pro Gly Lys Gly Leu Glu Trp Val 35 40 45 Ala Val Ile Ser
Tyr Asp Gly Ser Ser Arg Leu Tyr Ala Asp Ser Val 50 55 60 Lys Gly
Arg Phe Thr Val Ser Arg Asp Asn Ser Lys Asn Thr Leu Tyr 65 70 75 80
Leu Gln Met His Ser Leu Arg Ala Gly Asp Thr Ala Val Tyr Tyr Cys 85
90 95 Ala Arg Gly Tyr Cys Ser Ser Gly Thr Cys Phe Ser Thr Asn Ala
Glu 100 105 110 Tyr Phe His Pro Trp Gly Gln Gly Thr Leu Ala Thr Ile
Ser Ser 115 120 125 <210> SEQ ID NO 321 <211> LENGTH: 8
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKD20 CDRH1
<400> SEQUENCE: 321 Gly Phe Thr Phe Ser Asp His Phe 1 5
<210> SEQ ID NO 322 <211> LENGTH: 10 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKD20 CDRH2 <400> SEQUENCE:
322 Ser Arg Asn Lys Pro Asn Ser Tyr Thr Thr 1 5 10 <210> SEQ
ID NO 323 <211> LENGTH: 17 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKD20 CDRH3 <400> SEQUENCE: 323 Ala Arg
Val Gly Gly Cys Asn Gly Gly Asp Cys His Val Glu Asn Asp 1 5 10 15
Tyr <210> SEQ ID NO 324 <211> LENGTH: 126 <212>
TYPE: PRT <213> ORGANISM: Artificial Sequence <220>
FEATURE: <223> OTHER INFORMATION: ZKD20 VH <400>
SEQUENCE: 324 Glu Val Gln Leu Val Glu Ser Gly Gly Gly Leu Val Gln
Pro Gly Gly 1 5 10 15 Ser Leu Arg Leu Ser Cys Val Ala Ser Gly Phe
Thr Phe Ser Asp His 20 25 30 Phe Met Asp Trp Val Arg Gln Ala Pro
Gly Lys Gly Leu Glu Trp Val 35 40 45 Gly Arg Ser Arg Asn Lys Pro
Asn Ser Tyr Thr Thr Glu Tyr Ala Ala 50 55 60 Ser Val Lys Gly Arg
Phe Thr Ile Ser Arg Asp Asp Ser Lys Asn Ser 65 70 75 80 Leu Tyr Leu
Gln Met Asn Ser Leu Gln Thr Glu Asp Thr Ala Val Tyr 85 90 95 Tyr
Cys Ala Arg Val Gly Gly Cys Asn Gly Gly Asp Cys His Val Glu 100 105
110 Asn Asp Tyr Trp Gly Gln Gly Thr Leu Val Thr Val Ser Ser 115 120
125 <210> SEQ ID NO 325 <211> LENGTH: 8 <212>
TYPE: PRT <213> ORGANISM: Artificial Sequence <220>
FEATURE: <223> OTHER INFORMATION: ZKA134 CDRH1 <400>
SEQUENCE: 325 Gly Gly Thr Phe Ser Ala Tyr Ala 1 5 <210> SEQ
ID NO 326 <211> LENGTH: 8 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKA134 CDRH2 <400> SEQUENCE: 326 Ile Ile
Pro Phe Phe Gly Thr Ala 1 5 <210> SEQ ID NO 327 <211>
LENGTH: 19 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA134
CDRH3 <400> SEQUENCE: 327 Ala Arg Ser Asp Ile Val Ser Thr Thr
Arg Gly Tyr His His Tyr Gly 1 5 10 15 Met Asp Val <210> SEQ
ID NO 328 <211> LENGTH: 126 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKA134 VH <400> SEQUENCE: 328 Gln Val His
Leu Val Gln Ser Gly Ala Glu Val Lys Lys Pro Gly Ser 1 5 10 15 Ser
Val Asn Val Ser Cys Lys Ala Ser Gly Gly Thr Phe Ser Ala Tyr 20 25
30 Ala Ile Ser Trp Val Arg Gln Ala Pro Gly Gln Gly Leu Glu Trp Met
35 40 45 Gly Gly Ile Ile Pro Phe Phe Gly Thr Ala Tyr Tyr Ala Gln
Lys Phe 50 55 60 Lys Gly Arg Val Thr Val Thr Ala Asp Lys Ser Thr
Ser Thr Val Tyr 65 70 75 80 Met Glu Met Thr Ser Leu Arg Ser Glu Asp
Thr Ala Val Tyr Tyr Cys 85 90 95 Ala Arg Ser Asp Ile Val Ser Thr
Thr Arg Gly Tyr His His Tyr Gly 100 105 110 Met Asp Val Trp Gly Gln
Gly Thr Thr Val Thr Val Ser Ser 115 120 125 <210> SEQ ID NO
329 <211> LENGTH: 8 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKA246 CDRH1 <400> SEQUENCE: 329 Gly Tyr
Thr Phe Ser Asp Tyr Tyr 1 5 <210> SEQ ID NO 330 <211>
LENGTH: 8 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA246
CDRH2 <400> SEQUENCE: 330 Ile Asn Pro Tyr Ser Gly Gly Thr 1 5
<210> SEQ ID NO 331 <211> LENGTH: 14 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA246 CDRH3 <400> SEQUENCE:
331 Ala Arg Gly Phe Thr Met Ile Ser Asp Arg Glu Phe Asp Pro 1 5 10
<210> SEQ ID NO 332 <211> LENGTH: 121 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA246 VH <400> SEQUENCE: 332
Gln Val Gln Leu Val Gln Ser Gly Ala Glu Val Lys Arg Pro Gly Ala 1 5
10 15 Ser Val Lys Val Ser Cys Lys Ala Ser Gly Tyr Thr Phe Ser Asp
Tyr 20 25 30 Tyr Met His Trp Val Arg Gln Ala Pro Gly Gln Gly Leu
Glu Trp Met 35 40 45 Gly Arg Ile Asn Pro Tyr Ser Gly Gly Thr Asn
Tyr Ala Gln Lys Phe 50 55 60 His Gly Arg Val Thr Val Thr Arg Asp
Thr Ser Ile Ser Thr Val Tyr 65 70 75 80 Met Glu Leu Arg Gly Leu Arg
Ser Asp Asp Thr Ala Val Tyr Tyr Cys 85 90 95 Ala Arg Gly Phe Thr
Met Ile Ser Asp Arg Glu Phe Asp Pro Trp Gly 100 105 110 Gln Gly Thr
Leu Val Thr Val Ser Ser 115 120 <210> SEQ ID NO 333
<211> LENGTH: 8 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA256 CDRH1 <400> SEQUENCE: 333 Gly Phe Thr Phe
Ser Thr Tyr Trp 1 5 <210> SEQ ID NO 334 <211> LENGTH: 8
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA256 CDRH2
<400> SEQUENCE: 334 Ile Lys Gln Asp Gly Ser Glu Lys 1 5
<210> SEQ ID NO 335 <211> LENGTH: 20 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA256 CDRH3 <400> SEQUENCE:
335 Ala Arg Asp Pro Gly Tyr Asp Asp Phe Trp Ser Gly Ser Tyr Ser Gly
1 5 10 15 Ser Phe Asp Ile 20 <210> SEQ ID NO 336 <211>
LENGTH: 127 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA256
VH <400> SEQUENCE: 336 Glu Val Gln Leu Val Glu Ser Gly Gly
Gly Leu Val Gln Pro Gly Gly 1 5 10 15 Ser Leu Arg Leu Ser Cys Ala
Ala Ser Gly Phe Thr Phe Ser Thr Tyr 20 25 30 Trp Met Thr Trp Val
Arg Gln Ala Pro Gly Lys Gly Leu Glu Trp Val 35 40 45 Ala Asn Ile
Lys Gln Asp Gly Ser Glu Lys Tyr Tyr Val Asp Ser Val 50 55 60 Lys
Gly Arg Phe Thr Ile Ser Arg Asp Asn Thr Lys Asn Ser Leu Tyr 65 70
75 80 Leu Gln Val Asn Ser Leu Arg Ala Glu Asp Thr Ala Ile Tyr Tyr
Cys 85 90 95 Ala Arg Asp Pro Gly Tyr Asp Asp Phe Trp Ser Gly Ser
Tyr Ser Gly 100 105 110 Ser Phe Asp Ile Trp Gly Gln Gly Thr Met Val
Thr Val Ser Ser 115 120 125 <210> SEQ ID NO 337 <211>
LENGTH: 8 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKB42
CDRH1 <400> SEQUENCE: 337 Gly Phe Thr Phe Asn Asn Tyr Gly 1 5
<210> SEQ ID NO 338 <211> LENGTH: 8 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKB42 CDRH2 <400> SEQUENCE:
338 Ile Ser Tyr Asp Gly Asn Lys Lys 1 5 <210> SEQ ID NO 339
<211> LENGTH: 16 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKB42 CDRH3 <400> SEQUENCE: 339 Val Lys Tyr Gly
Glu Arg Ile Asn Gly Tyr Ser Asp Pro Phe Asp His 1 5 10 15
<210> SEQ ID NO 340 <211> LENGTH: 123 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKB42 VH <400> SEQUENCE: 340
Gln Val Gln Val Val Glu Ser Gly Gly Gly Val Val Gln Pro Gly Arg 1 5
10 15 Ser Leu Arg Leu Phe Cys Ala Ala Ser Gly Phe Thr Phe Asn Asn
Tyr 20 25 30 Gly Met His Trp Val Arg Gln Ala Pro Gly Lys Gly Leu
Glu Trp Val 35 40 45 Ala Leu Ile Ser Tyr Asp Gly Asn Lys Lys Tyr
Tyr Ala Asp Ser Val 50 55 60 Lys Gly Arg Phe Ser Ile Ser Arg Asp
Asn Ser Lys Asn Thr Leu Tyr 65 70 75 80 Leu Gln Met Asn Arg Leu Arg
Ser Gly Asp Thr Ala Val Tyr His Cys 85 90 95 Val Lys Tyr Gly Glu
Arg Ile Asn Gly Tyr Ser Asp Pro Phe Asp His 100 105 110 Trp Gly Gln
Gly Thr Leu Val Thr Val Ser Ser 115 120 <210> SEQ ID NO 341
<211> LENGTH: 8 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKB85 CDRH1 <400> SEQUENCE: 341 Gly Tyr Thr Phe
Thr Thr Tyr Ala 1 5 <210> SEQ ID NO 342 <211> LENGTH: 8
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKB85 CDRH2
<400> SEQUENCE: 342 Ile Asn Thr Asn Thr Gly Asn Pro 1 5
<210> SEQ ID NO 343 <211> LENGTH: 11 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKB85 CDRH3 <400> SEQUENCE:
343 Ala Arg Val Ile Val Pro Tyr Ala Phe Asp Ile 1 5 10 <210>
SEQ ID NO 344 <211> LENGTH: 118 <212> TYPE: PRT
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKB85 VH <400> SEQUENCE: 344
Gln Val Gln Leu Val Gln Ser Gly Ser Glu Leu Lys Lys Pro Gly Ala 1 5
10 15 Ser Val Lys Val Ser Cys Lys Ala Ser Gly Tyr Thr Phe Thr Thr
Tyr 20 25 30 Ala Met Asn Trp Val Arg Gln Ala Pro Gly Gln Gly Pro
Glu Trp Val 35 40 45 Gly Trp Ile Asn Thr Asn Thr Gly Asn Pro Thr
Tyr Ala Gln Gly Phe 50 55 60 Thr Gly Arg Phe Val Leu Ser Leu Asp
Thr Ser Val Ser Thr Ala Phe 65 70 75 80 Leu Gln Ile Ser Ser Leu Lys
Ala Glu Asp Thr Ala Val Tyr Tyr Cys 85 90 95 Ala Arg Val Ile Val
Pro Tyr Ala Phe Asp Ile Trp Gly Gln Gly Thr 100 105 110 Met Val Thr
Val Ser Ser 115 <210> SEQ ID NO 345 <211> LENGTH: 8
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKB47 CDRH1
<400> SEQUENCE: 345 Gly Tyr Thr Phe Thr Asn Tyr Tyr 1 5
<210> SEQ ID NO 346 <211> LENGTH: 8 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKB47 CDRH2 <400> SEQUENCE:
346 Ile Asn Pro Ser Gly Gly Pro Thr 1 5 <210> SEQ ID NO 347
<211> LENGTH: 15 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKB47 CDRH3 <400> SEQUENCE: 347 Ala Arg Asp Gln
Tyr Gly Gly Tyr Ala Arg Tyr Gly Met Asp Val 1 5 10 15 <210>
SEQ ID NO 348 <211> LENGTH: 122 <212> TYPE: PRT
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKB47 VH <400> SEQUENCE: 348
Gln Val Gln Leu Val Gln Ser Gly Ala Glu Val Lys Lys Pro Gly Ala 1 5
10 15 Ser Val Lys Val Ser Cys Gln Ala Ser Gly Tyr Thr Phe Thr Asn
Tyr 20 25 30 Tyr Met His Trp Val Arg Gln Ala Pro Gly Gln Gly Leu
Glu Trp Met 35 40 45 Gly Ile Ile Asn Pro Ser Gly Gly Pro Thr Ser
Tyr Ala Gln Lys Phe 50 55 60 Gln Gly Arg Val Thr Met Thr Thr Asp
Thr Ser Thr Ser Thr Val Tyr 65 70 75 80 Met Glu Leu Ser Ser Leu Arg
Ser Glu Asp Thr Ala Val Tyr Tyr Cys 85 90 95 Ala Arg Asp Gln Tyr
Gly Gly Tyr Ala Arg Tyr Gly Met Asp Val Trp 100 105 110 Gly Gln Gly
Thr Thr Val Thr Val Ser Ser 115 120 <210> SEQ ID NO 349
<211> LENGTH: 8 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKC6 CDRH1 <400> SEQUENCE: 349 Gly Tyr Thr Phe
Thr Gly Tyr Tyr 1 5 <210> SEQ ID NO 350 <211> LENGTH: 8
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKC6 CDRH2
<400> SEQUENCE: 350 Ile Asn Pro Asn Ser Gly Gly Thr 1 5
<210> SEQ ID NO 351 <211> LENGTH: 12 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKC6 CDRH3 <400> SEQUENCE: 351
Ala Arg Val Ser Asp Trp Gly Phe Ala Phe Asp Ile 1 5 10 <210>
SEQ ID NO 352 <211> LENGTH: 119 <212> TYPE: PRT
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKC6 VH <400> SEQUENCE: 352
Gln Val Gln Leu Val Gln Ser Gly Thr Glu Val Lys Lys Pro Gly Ala 1 5
10 15 Ser Val Lys Val Ser Cys Lys Ala Ser Gly Tyr Thr Phe Thr Gly
Tyr 20 25 30 Tyr Met His Trp Val Arg Gln Ala Pro Gly Gln Gly Leu
Glu Trp Met 35 40 45 Gly Arg Ile Asn Pro Asn Ser Gly Gly Thr Asn
Tyr Ala Gln Lys Phe 50 55 60 Gln Gly Arg Val Thr Met Thr Arg Asp
Thr Ser Ile Ser Thr Ala Tyr 65 70 75 80 Met Glu Leu Ser Gly Leu Arg
Ser Asp Asp Thr Ala Val Tyr Tyr Cys 85 90 95 Ala Arg Val Ser Asp
Trp Gly Phe Ala Phe Asp Ile Trp Gly Gln Gly 100 105 110 Thr Met Val
Thr Val Ser Gln 115 <210> SEQ ID NO 353 <211> LENGTH: 8
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA160 CDRH1
<400> SEQUENCE: 353 Gly Gly Ser Ile Thr Ser Tyr Ser 1 5
<210> SEQ ID NO 354 <211> LENGTH: 7 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA160 CDRH2 <400> SEQUENCE:
354 Ile Phe Tyr Ser Gly Ser Thr 1 5 <210> SEQ ID NO 355
<211> LENGTH: 15 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA160 CDRH3 <400> SEQUENCE: 355 Ala Arg Asp Gln
Thr Met Pro Val Trp Val Gly Gly Met Asp Val 1 5 10 15 <210>
SEQ ID NO 356 <211> LENGTH: 121 <212> TYPE: PRT
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA160 VH <400> SEQUENCE: 356
Gln Val Gln Leu Gln Glu Ser Gly Pro Gly Leu Val Lys Pro Ser Glu 1 5
10 15 Thr Leu Ser Leu Thr Cys Thr Val Ser Gly Gly Ser Ile Thr Ser
Tyr 20 25 30 Ser Trp Ser Trp Ile Arg Gln Pro Pro Gly Lys Gly Leu
Glu Trp Ile 35 40 45 Gly Tyr Ile Phe Tyr Ser Gly Ser Thr Asp Tyr
Asn Pro Ser Leu Lys 50 55 60 Ser Arg Val Thr Ile Ser Val Asp Thr
Ser Lys Asp Gln Phe Ser Leu 65 70 75 80 Arg Leu Arg Ser Val Thr Ala
Ala Asp Thr Ala Val Tyr Tyr Cys Ala 85 90 95 Arg Asp Gln Thr Met
Pro Val Trp Val Gly Gly Met Asp Val Trp Gly 100 105 110 Gln Gly Thr
Thr Val Thr Val Ser Ser 115 120 <210> SEQ ID NO 357
<211> LENGTH: 8 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA172 CDRH1 <400> SEQUENCE: 357 Gly Tyr Ile Phe
Thr Arg Tyr Trp 1 5 <210> SEQ ID NO 358 <211> LENGTH: 8
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA172 CDRH2
<400> SEQUENCE: 358 Ile Asp Pro Ser Asp Ser Tyr Thr 1 5
<210> SEQ ID NO 359 <211> LENGTH: 13 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA172 CDRH3 <400> SEQUENCE:
359 Ala Arg Gln Glu Thr Ala Arg Glu Asp Gly Met Ala Val 1 5 10
<210> SEQ ID NO 360 <211> LENGTH: 120 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA172 VH <400> SEQUENCE: 360
Glu Val Gln Leu Val Gln Ser Gly Ala Glu Val Lys Lys Pro Gly Lys 1 5
10 15 Ser Leu Arg Ile Ser Cys Lys Gly Ser Gly Tyr Ile Phe Thr Arg
Tyr 20 25 30 Trp Ile Ser Trp Val Arg Gln Met Pro Gly Lys Gly Leu
Glu Trp Met 35 40 45 Gly Arg Ile Asp Pro Ser Asp Ser Tyr Thr Asn
Tyr Ser Pro Ser Phe 50 55 60 Gln Gly His Val Thr Ile Ser Ala Asp
Lys Ser Ile Ser Thr Ala Tyr 65 70 75 80 Leu Gln Trp Ser Ser Leu Lys
Ala Ser Asp Thr Ala Met Tyr Tyr Cys 85 90 95 Ala Arg Gln Glu Thr
Ala Arg Glu Asp Gly Met Ala Val Trp Gly Gln 100 105 110 Gly Thr Thr
Val Thr Val Ser Ser 115 120 <210> SEQ ID NO 361 <211>
LENGTH: 10 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA174
CDRH1 <400> SEQUENCE: 361 Gly Gly Ser Met Ser Asn Ser Tyr Tyr
His 1 5 10 <210> SEQ ID NO 362 <211> LENGTH: 7
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA174 CDRH2
<400> SEQUENCE: 362 Ile Tyr Tyr Ser Gly Ser Thr 1 5
<210> SEQ ID NO 363 <211> LENGTH: 18 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA174 CDRH3 <400> SEQUENCE:
363 Ala Arg Asn Pro Val Phe Asn Pro Leu Thr Leu Thr His Asp Ala Phe
1 5 10 15 Asp Ile <210> SEQ ID NO 364 <211> LENGTH: 126
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA174 VH
<400> SEQUENCE: 364 Gln Leu Gln Leu Gln Glu Ser Gly Pro Gly
Leu Val Lys Pro Ser Glu 1 5 10 15 Thr Leu Ser Leu Thr Cys Thr Val
Ser Gly Gly Ser Met Ser Asn Ser 20 25 30 Tyr Tyr His Trp Gly Trp
Ile Arg Gln Pro Pro Gly Lys Gly Leu Glu 35 40 45 Trp Ile Gly Ser
Ile Tyr Tyr Ser Gly Ser Thr Tyr Tyr Asn Pro Ser 50 55 60 Leu Lys
Ser Arg Val Thr Ile Ser Val Asp Thr Ser Lys Asn Gln Phe 65 70 75 80
Ser Leu Lys Leu Asn Ser Val Thr Ala Ala Asp Thr Ala Val Tyr Tyr 85
90 95 Cys Ala Arg Asn Pro Val Phe Asn Pro Leu Thr Leu Thr His Asp
Ala 100 105 110 Phe Asp Ile Trp Gly Gln Gly Thr Met Val Thr Val Ser
Ser 115 120 125 <210> SEQ ID NO 365 <211> LENGTH: 8
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA189 CDRH1
<400> SEQUENCE: 365 Gly Phe Thr Phe Ser Ser Tyr Ala 1 5
<210> SEQ ID NO 366 <211> LENGTH: 8 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA189 CDRH2 <400> SEQUENCE:
366 Ile Ser Gly Ser Gly Asp Asn Thr 1 5 <210> SEQ ID NO 367
<211> LENGTH: 18 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA189 CDRH3 <400> SEQUENCE: 367 Ala Lys Trp Pro
Tyr Tyr Asp Phe Trp Ser Gly Ser Glu Ser Tyr Phe 1 5 10 15 Asp Pro
<210> SEQ ID NO 368 <211> LENGTH: 125 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA189 VH <400> SEQUENCE: 368
Gly Val Gln Leu Leu Glu Ser Gly Gly Ala Leu Val Gln Pro Gly Lys 1 5
10 15 Ser Leu Arg Leu Ser Cys Ala Ala Ser Gly Phe Thr Phe Ser Ser
Tyr 20 25 30 Ala Leu Thr Trp Val Arg Gln Ala Pro Gly Lys Gly Leu
Gln Trp Val 35 40 45 Ser Ala Ile Ser Gly Ser Gly Asp Asn Thr Tyr
Tyr Ala Asp Ser Val 50 55 60 Lys Gly Arg Phe Thr Ile Ser Arg Asp
Asn Ser Lys Asn Thr Leu Tyr 65 70 75 80 Leu Gln Met Asn Ser Leu Arg
Ala Glu Asp Thr Ala Val Tyr Tyr Cys 85 90 95 Ala Lys Trp Pro Tyr
Tyr Asp Phe Trp Ser Gly Ser Glu Ser Tyr Phe 100 105 110 Asp Pro Trp
Gly Gln Gly Thr Leu Val Thr Val Ser Ser 115 120 125 <210> SEQ
ID NO 369 <211> LENGTH: 8 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKA195 CDRH1 <400> SEQUENCE: 369 Gly Tyr
Asn Phe Pro Ser Tyr Trp 1 5 <210> SEQ ID NO 370 <211>
LENGTH: 8 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA195
CDRH2 <400> SEQUENCE: 370 Ile Asp Pro Ser Asp Ser Tyr Thr 1 5
<210> SEQ ID NO 371 <211> LENGTH: 15 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA195 CDRH3 <400> SEQUENCE:
371 Ala Arg Ala Asp Cys Arg Ser Thr Ser Cys Tyr Leu Val Phe Glu 1 5
10 15 <210> SEQ ID NO 372 <211> LENGTH: 121 <212>
TYPE: PRT <213> ORGANISM: Artificial Sequence <220>
FEATURE: <223> OTHER INFORMATION: ZKA195 VH <400>
SEQUENCE: 372 Glu Val Gln Leu Val Gln Ser Gly Ala Glu Val Lys Lys
Pro Gly Glu 1 5 10 15 Ser Leu Arg Ile Ser Cys Lys Asp Ser Gly Tyr
Asn Phe Pro Ser Tyr 20 25 30 Trp Ile His Trp Val Arg Gln Met Pro
Gly Lys Gly Leu Glu Trp Met 35 40 45 Gly Thr Ile Asp Pro Ser Asp
Ser Tyr Thr Asn Tyr Ser Pro Ser Phe 50 55 60 Gln Gly His Val Thr
Ile Ser Ala Asp Lys Ser Ile Ser Thr Ala Tyr 65 70 75 80 Leu Gln Trp
Ser Ser Leu Lys Ala Ser Asp Thr Ala Met Tyr Tyr Cys 85 90 95 Ala
Arg Ala Asp Cys Arg Ser Thr Ser Cys Tyr Leu Val Phe Glu Gly 100 105
110 Gln Gly Thr Leu Val Thr Val Ser Ser 115 120 <210> SEQ ID
NO 373 <211> LENGTH: 8 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKA215 CDRH1 <400> SEQUENCE: 373 Gly Tyr
Thr Phe Thr Ser Tyr Trp 1 5 <210> SEQ ID NO 374 <211>
LENGTH: 8 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA215
CDRH2 <400> SEQUENCE: 374 Ile Asp Pro Ser Asp Ser His Thr 1 5
<210> SEQ ID NO 375 <211> LENGTH: 11 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA215 CDRH3 <400> SEQUENCE:
375 Ala Arg His Ala Leu Pro Asn Tyr Phe Asp Ser 1 5 10 <210>
SEQ ID NO 376 <211> LENGTH: 118 <212> TYPE: PRT
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA215 VH <400> SEQUENCE: 376
Glu Val Gln Leu Val Gln Ser Gly Ala Glu Val Lys Lys Pro Gly Glu 1 5
10 15 Ser Leu Arg Ile Ser Cys Lys Gly Ser Gly Tyr Thr Phe Thr Ser
Tyr 20 25 30 Trp Ile Ser Trp Val Arg Gln Met Pro Gly Lys Gly Leu
Glu Trp Met 35 40 45 Gly Arg Ile Asp Pro Ser Asp Ser His Thr Asp
Tyr Ser Pro Ser Phe 50 55 60 Gln Gly His Val Thr Ile Ser Ala Asp
Lys Ser Ile Ser Ala Ala Tyr 65 70 75 80 Leu Gln Trp Ser Ser Leu Lys
Ala Ser Asp Thr Ala Met Tyr Tyr Cys 85 90 95 Ala Arg His Ala Leu
Pro Asn Tyr Phe Asp Ser Trp Gly Gln Gly Thr 100 105 110 Leu Val Thr
Val Ser Ser 115 <210> SEQ ID NO 377 <211> LENGTH: 8
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA218 CDRH1
<400> SEQUENCE: 377 Gly Phe Pro Phe Ser Ser Tyr Trp 1 5
<210> SEQ ID NO 378 <211> LENGTH: 8 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA218 CDRH2 <400> SEQUENCE:
378 Ile Asn Ser Asp Gly Arg Asn Thr 1 5 <210> SEQ ID NO 379
<211> LENGTH: 15 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA218 CDRH3 <400> SEQUENCE: 379 Ala Arg Gly Gly
Tyr Asp Tyr Asp Ser Ser Gly Cys Phe Asp Tyr 1 5 10 15 <210>
SEQ ID NO 380 <211> LENGTH: 122 <212> TYPE: PRT
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA218 VH <400> SEQUENCE: 380
Glu Val Gln Leu Val Glu Ser Gly Gly Gly Leu Val Gln Pro Gly Gly 1 5
10 15 Ser Leu Arg Leu Ser Cys Ala Ala Ser Gly Phe Pro Phe Ser Ser
Tyr 20 25 30 Trp Met His Trp Val Arg Gln Ala Pro Gly Lys Gly Leu
Val Trp Val 35 40 45 Ser Arg Ile Asn Ser Asp Gly Arg Asn Thr Asn
Tyr Ala Asp Ser Val 50 55 60 Lys Gly Arg Phe Thr Ile Ser Arg Asp
Asn Ala Glu Asn Thr Val Tyr 65 70 75 80 Leu Gln Met Asn Ser Leu Arg
Ala Glu Asp Thr Ala Val Tyr Tyr Cys 85 90 95 Ala Arg Gly Gly Tyr
Asp Tyr Asp Ser Ser Gly Cys Phe Asp Tyr Trp 100 105 110 Gly Gln Gly
Thr Leu Val Thr Val Ser Ser 115 120 <210> SEQ ID NO 381
<211> LENGTH: 8 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKB75 CDRH1 <400> SEQUENCE: 381 Gly Phe Thr Phe
Ser Asn Tyr Ala 1 5 <210> SEQ ID NO 382 <211> LENGTH: 8
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKB75 CDRH2
<400> SEQUENCE: 382 Ile Ser Gly Thr Gly Gly Ser Thr 1 5
<210> SEQ ID NO 383 <211> LENGTH: 25 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKB75 CDRH3 <400> SEQUENCE:
383 Ala Lys Asp Ser Ala Ser Arg Gly Gly Tyr Cys Ser Gly Gly Val Cys
1 5 10 15 Tyr Leu Asn Pro Gly His His Asp Tyr 20 25 <210> SEQ
ID NO 384 <211> LENGTH: 132 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKB75 VH <400> SEQUENCE: 384 Glu Val Gln
Val Leu Glu Ser Gly Gly Gly Leu Leu Gln Pro Gly Gly 1 5 10 15 Ser
Leu Arg Leu Ser Cys Ala Ala Ser Gly Phe Thr Phe Ser Asn Tyr 20 25
30 Ala Met Ser Trp Val Arg Gln Ala Pro Gly Lys Gly Leu Glu Trp Val
35 40 45 Ser Thr Ile Ser Gly Thr Gly Gly Ser Thr Tyr Tyr Ala Asp
Ser Val 50 55 60 Lys Gly Arg Phe Thr Ile Ser Arg Asp Asn Ser Lys
Asn Thr Leu Tyr 65 70 75 80 Leu Gln Met Asn Ser Leu Arg Ala Glu Asp
Thr Ala Val Tyr Tyr Cys 85 90 95 Ala Lys Asp Ser Ala Ser Arg Gly
Gly Tyr Cys Ser Gly Gly Val Cys 100 105 110 Tyr Leu Asn Pro Gly His
His Asp Tyr Trp Gly Gln Gly Thr Leu Val 115 120 125 Thr Val Ser Ser
130 <210> SEQ ID NO 385 <211> LENGTH: 8 <212>
TYPE: PRT <213> ORGANISM: Artificial Sequence <220>
FEATURE: <223> OTHER INFORMATION: ZKB83 CDRH1 <400>
SEQUENCE: 385 Gly Tyr Ser Phe Thr Asn Tyr Trp 1 5 <210> SEQ
ID NO 386 <211> LENGTH: 8 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKB83 CDRH2 <400> SEQUENCE: 386 Ile Asp
Pro Ser Asp Ser Tyr Thr 1 5 <210> SEQ ID NO 387 <211>
LENGTH: 27 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKB83
CDRH3 <400> SEQUENCE: 387 Ala Arg Leu Arg Gly Ser Leu Tyr Cys
Ser Gly Gly Arg Cys Tyr Ser 1 5 10 15 Val Pro Gly Glu Thr Pro Asn
Trp Phe Asp Pro 20 25 <210> SEQ ID NO 388 <211> LENGTH:
134 <212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKB83 VH
<400> SEQUENCE: 388 Glu Val Gln Leu Val Gln Ser Gly Ala Glu
Val Lys Lys Pro Gly Glu 1 5 10 15 Ser Leu Arg Ile Ser Cys Lys Gly
Ser Gly Tyr Ser Phe Thr Asn Tyr 20 25 30 Trp Ile Thr Trp Val Arg
Gln Met Pro Gly Lys Gly Leu Glu Trp Met 35 40 45 Gly Ser Ile Asp
Pro Ser Asp Ser Tyr Thr Asn Tyr Ser Pro Ser Phe 50 55 60 Gln Gly
His Val Thr Ile Ser Ala Asp Trp Ser Ile Asn Thr Ala Tyr 65 70 75 80
Leu Gln Trp Ser Ser Leu Lys Ala Ser Asp Thr Ala Lys Tyr Tyr Cys 85
90 95 Ala Arg Leu Arg Gly Ser Leu Tyr Cys Ser Gly Gly Arg Cys Tyr
Ser 100 105 110 Val Pro Gly Glu Thr Pro Asn Trp Phe Asp Pro Trp Gly
Gln Gly Thr 115 120 125 Leu Val Thr Val Ser Ser 130 <210> SEQ
ID NO 389 <211> LENGTH: 8 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKC3 CDRH1 <400> SEQUENCE: 389 Gly Gly Ser
Ile Thr Ser Tyr Tyr 1 5 <210> SEQ ID NO 390 <211>
LENGTH: 7 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKC3
CDRH2 <400> SEQUENCE: 390 Ile Tyr Tyr Ser Gly Ser Thr 1 5
<210> SEQ ID NO 391 <211> LENGTH: 15 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKC3 CDRH3 <400> SEQUENCE: 391
Ala Arg Val Gly Gly Ala Pro Tyr Tyr Tyr Tyr Gly Met Asp Val 1 5 10
15 <210> SEQ ID NO 392 <211> LENGTH: 121 <212>
TYPE: PRT <213> ORGANISM: Artificial Sequence <220>
FEATURE: <223> OTHER INFORMATION: ZKC3 VH <400>
SEQUENCE: 392 Gln Val Gln Leu Gln Glu Ser Gly Pro Gly Leu Val Lys
Pro Ser Glu 1 5 10 15 Thr Leu Ser Leu Thr Cys Thr Val Ser Gly Gly
Ser Ile Thr Ser Tyr 20 25 30 Tyr Trp Ser Trp Ile Arg Gln Pro Pro
Gly Lys Gly Leu Glu Trp Ile 35 40 45 Gly Tyr Ile Tyr Tyr Ser Gly
Ser Thr Asn Tyr Asn Pro Ser Leu Lys 50 55 60 Ser Arg Val Thr Ile
Ser Val Asp Thr Ser Lys Asn Gln Phe Ser Leu 65 70 75 80 Lys Leu Ser
Ser Val Thr Ala Ala Asp Thr Ala Val Tyr Tyr Cys Ala 85 90 95 Arg
Val Gly Gly Ala Pro Tyr Tyr Tyr Tyr Gly Met Asp Val Trp Gly 100 105
110 Gln Gly Thr Thr Val Thr Val Ser Ser 115 120 <210> SEQ ID
NO 393 <211> LENGTH: 8 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKC18 CDRH1 <400> SEQUENCE: 393 Gly Phe
Thr Phe Gly Asp Tyr Ala 1 5 <210> SEQ ID NO 394 <211>
LENGTH: 10 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKC18
CDRH2 <400> SEQUENCE: 394 Ile Arg Ser Lys Ala Tyr Gly Gly Thr
Thr 1 5 10 <210> SEQ ID NO 395 <211> LENGTH: 13
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKC18 CDRH3
<400> SEQUENCE: 395 Ser Arg Asp His Thr Gly Thr Thr Tyr Ala
Phe Asp Ile 1 5 10 <210> SEQ ID NO 396 <211> LENGTH:
122 <212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKC18 VH
<400> SEQUENCE: 396 Glu Val Gln Leu Val Glu Ser Gly Gly Gly
Leu Val Gln Pro Gly Arg 1 5 10 15 Ser Leu Arg Leu Ser Cys Thr Ala
Ser Gly Phe Thr Phe Gly Asp Tyr 20 25 30 Ala Met Ser Trp Phe Arg
Gln Ala Pro Gly Lys Gly Leu Glu Trp Val 35 40 45 Gly Phe Ile Arg
Ser Lys Ala Tyr Gly Gly Thr Thr Glu Tyr Ala Ala 50 55 60 Ser Val
Lys Gly Arg Phe Thr Ile Ser Arg Asp Asp Ser Lys Ser Ile 65 70 75 80
Ala Tyr Leu Gln Met Asn Ser Leu Lys Thr Glu Asp Thr Ala Val Tyr 85
90 95 Tyr Cys Ser Arg Asp His Thr Gly Thr Thr Tyr Ala Phe Asp Ile
Trp 100 105 110 Gly Gln Gly Thr Met Val Thr Val Ser Gln 115 120
<210> SEQ ID NO 397 <211> LENGTH: 8 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKD1 CDRH1 <400> SEQUENCE: 397
Gly Phe Thr Phe Ser Ser Tyr Gly 1 5 <210> SEQ ID NO 398
<211> LENGTH: 8 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKD1 CDRH2 <400> SEQUENCE: 398 Ile Trp Tyr Asp
Gly Ser Asn Lys 1 5 <210> SEQ ID NO 399 <211> LENGTH:
19 <212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKD1 CDRH3
<400> SEQUENCE: 399 Ala Arg Asp Arg Arg Gly Tyr Gly Asp Tyr
Val Gly Tyr Tyr Tyr Gly 1 5 10 15 Met Asp Val <210> SEQ ID NO
400 <211> LENGTH: 126 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKD1 VH <400> SEQUENCE: 400 Gln Val Gln
Leu Val Glu Ser Gly Gly Gly Val Val Gln Pro Gly Arg 1 5 10 15 Ser
Leu Arg Leu Ser Cys Ala Ala Ser Gly Phe Thr Phe Ser Ser Tyr 20 25
30 Gly Met His Trp Val Arg Gln Ala Pro Gly Lys Gly Leu Glu Trp Val
35 40 45 Ala Val Ile Trp Tyr Asp Gly Ser Asn Lys Tyr Tyr Ala Asp
Ser Val 50 55 60 Lys Gly Arg Phe Thr Ile Ser Arg Asp Asn Ser Lys
Asn Thr Leu Tyr 65 70 75 80 Leu Gln Met Asn Ser Leu Arg Ala Glu Asp
Thr Ala Val Tyr Tyr Cys 85 90 95 Ala Arg Asp Arg Arg Gly Tyr Gly
Asp Tyr Val Gly Tyr Tyr Tyr Gly 100 105 110 Met Asp Val Trp Gly Gln
Gly Thr Thr Val Thr Val Ser Ser 115 120 125 <210> SEQ ID NO
401 <211> LENGTH: 95 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: Zika Virus EDIII <220> FEATURE:
<221> NAME/KEY: misc_feature <222> LOCATION: (9)..(9)
<223> OTHER INFORMATION: Xaa can be any naturally occurring
amino acid <220> FEATURE: <221> NAME/KEY: misc_feature
<222> LOCATION: (13)..(14) <223> OTHER INFORMATION: Xaa
can be any naturally occurring amino acid <220> FEATURE:
<221> NAME/KEY: misc_feature <222> LOCATION: (22)..(22)
<223> OTHER INFORMATION: Xaa can be any naturally occurring
amino acid <220> FEATURE: <221> NAME/KEY: misc_feature
<222> LOCATION: (25)..(25) <223> OTHER INFORMATION: Xaa
can be any naturally occurring amino acid <220> FEATURE:
<221> NAME/KEY: misc_feature <222> LOCATION: (27)..(27)
<223> OTHER INFORMATION: Xaa can be any naturally occurring
amino acid <220> FEATURE: <221> NAME/KEY: misc_feature
<222> LOCATION: (33)..(33) <223> OTHER INFORMATION: Xaa
can be any naturally occurring amino acid <220> FEATURE:
<221> NAME/KEY: misc_feature <222> LOCATION: (35)..(35)
<223> OTHER INFORMATION: Xaa can be any naturally occurring
amino acid <220> FEATURE: <221> NAME/KEY: misc_feature
<222> LOCATION: (41)..(41) <223> OTHER INFORMATION: Xaa
can be any naturally occurring amino acid <220> FEATURE:
<221> NAME/KEY: misc_feature <222> LOCATION: (60)..(60)
<223> OTHER INFORMATION: Xaa can be any naturally occurring
amino acid <220> FEATURE: <221> NAME/KEY: misc_feature
<222> LOCATION: (83)..(83) <223> OTHER INFORMATION: Xaa
can be any naturally occurring amino acid <220> FEATURE:
<221> NAME/KEY: misc_feature <222> LOCATION: (85)..(85)
<223> OTHER INFORMATION: Xaa can be any naturally occurring
amino acid <400> SEQUENCE: 401 Thr Ala Ala Phe Thr Phe Thr
Lys Xaa Pro Ala Glu Xaa Xaa His Gly 1 5 10 15 Thr Val Thr Val Glu
Xaa Gln Tyr Xaa Gly Xaa Asp Gly Pro Cys Lys 20 25 30 Xaa Pro Xaa
Gln Met Ala Val Asp Xaa Gln Thr Leu Thr Pro Val Gly 35 40 45 Arg
Leu Ile Thr Ala Asn Pro Val Ile Thr Glu Xaa Thr Glu Asn Ser 50 55
60 Lys Met Met Leu Glu Leu Asp Pro Pro Phe Gly Asp Ser Tyr Ile Val
65 70 75 80 Ile Gly Xaa Gly Xaa Lys Lys Ile Thr His His Trp His Arg
Ser 85 90 95 <210> SEQ ID NO 402 <211> LENGTH: 95
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZIKAV H PF 2013
EDIII <400> SEQUENCE: 402 Thr Ala Ala Phe Thr Phe Thr Lys Ile
Pro Ala Glu Thr Leu His Gly 1 5 10 15 Thr Val Thr Val Glu Val Gln
Tyr Ala Gly Thr Asp Gly Pro Cys Lys 20 25 30 Val Pro Ala Gln Met
Ala Val Asp Met Gln Thr Leu Thr Pro Val Gly 35 40 45 Arg Leu Ile
Thr Ala Asn Pro Val Ile Thr Glu Ser Thr Glu Asn Ser 50 55 60 Lys
Met Met Leu Glu Leu Asp Pro Pro Phe Gly Asp Ser Tyr Ile Val 65 70
75 80 Ile Gly Val Gly Glu Lys Lys Ile Thr His His Trp His Arg Ser
85 90 95 <210> SEQ ID NO 403 <211> LENGTH: 20
<212> TYPE: DNA <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZIKV-NS1
forward primer <400> SEQUENCE: 403 tggagttcaa ctgacggtcg 20
<210> SEQ ID NO 404 <211> LENGTH: 20 <212> TYPE:
DNA <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZIKV-NS1-reverse primer <400>
SEQUENCE: 404 taccccgaac ccatgatcct 20 <210> SEQ ID NO 405
<211> LENGTH: 21 <212> TYPE: DNA <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: Gapdh- forward primer <400> SEQUENCE: 405
ggcaagttca aaggcacagt c 21 <210> SEQ ID NO 406 <211>
LENGTH: 19 <212> TYPE: DNA <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION:
Gapdh-reverse primer <400> SEQUENCE: 406 caccagcatc accccattt
19 <210> SEQ ID NO 407 <211> LENGTH: 104 <212>
TYPE: PRT <213> ORGANISM: Artificial Sequence <220>
FEATURE: <223> OTHER INFORMATION: ZIKV EDIII generic
<220> FEATURE: <221> NAME/KEY: misc_feature <222>
LOCATION: (1)..(1) <223> OTHER INFORMATION: X may be any
(naturally occurring) amino acid, preferably K, A, or E <220>
FEATURE: <221> NAME/KEY: misc_feature <222> LOCATION:
(3)..(3) <223> OTHER INFORMATION: X may be any (naturally
occurring) amino acid, preferably V, F, or L <220> FEATURE:
<221> NAME/KEY: misc_feature <222> LOCATION: (4)..(4)
<223> OTHER INFORMATION: X may be any (naturally occurring)
amino acid, preferably S or F <220> FEATURE: <221>
NAME/KEY: misc_feature <222> LOCATION: (17)..(17) <223>
OTHER INFORMATION: X may be any (naturally occurring) amino acid,
preferably I or V <220> FEATURE: <221> NAME/KEY:
misc_feature <222> LOCATION: (21)..(21) <223> OTHER
INFORMATION: X may be any (naturally occurring) amino acid,
preferably T or V <220> FEATURE: <221> NAME/KEY:
misc_feature <222> LOCATION: (22)..(22) <223> OTHER
INFORMATION: X may be any (naturally occurring) amino acid,
preferably L or D <220> FEATURE: <221> NAME/KEY:
misc_feature <222> LOCATION: (30)..(30) <223> OTHER
INFORMATION: X may be any (naturally occurring) amino acid,
preferably V or G <220> FEATURE: <221> NAME/KEY:
misc_feature <222> LOCATION: (33)..(33) <223> OTHER
INFORMATION: X may be any (naturally occurring) amino acid,
preferably A or G <220> FEATURE: <221> NAME/KEY:
misc_feature <222> LOCATION: (35)..(35) <223> OTHER
INFORMATION: X may be any (naturally occurring) amino acid except
R, preferably T or A <220> FEATURE: <221> NAME/KEY:
misc_feature <222> LOCATION: (41)..(41) <223> OTHER
INFORMATION: X may be any (naturally occurring) amino acid,
preferably V or I <220> FEATURE: <221> NAME/KEY:
misc_feature <222> LOCATION: (43)..(43) <223> OTHER
INFORMATION: X may be any (naturally occurring) amino acid,
preferably A or V <220> FEATURE: <221> NAME/KEY:
misc_feature <222> LOCATION: (49)..(49) <223> OTHER
INFORMATION: X may be any (naturally occurring) amino acid,
preferably M or T <220> FEATURE: <221> NAME/KEY:
misc_feature <222> LOCATION: (68)..(68) <223> OTHER
INFORMATION: X may be any (naturally occurring) amino acid,
preferably S or G <220> FEATURE: <221> NAME/KEY:
misc_feature <222> LOCATION: (70)..(70) <223> OTHER
INFORMATION: X may be any (naturally occurring) amino acid,
preferably E or K <220> FEATURE: <221> NAME/KEY:
misc_feature <222> LOCATION: (91)..(91) <223> OTHER
INFORMATION: X may be any (naturally occurring) amino acid,
preferably V or I <220> FEATURE: <221> NAME/KEY:
misc_feature <222> LOCATION: (93)..(93) <223> OTHER
INFORMATION: X may be any (naturally occurring) amino acid,
preferably E, A, K, or D <220> FEATURE: <221> NAME/KEY:
misc_feature <222> LOCATION: (94)..(94) <223> OTHER
INFORMATION: X may be any (naturally occurring) amino acid,
preferably E, A, or K, more preferably K or A <400> SEQUENCE:
407 Xaa Gly Xaa Xaa Tyr Ser Leu Cys Thr Ala Ala Phe Thr Phe Thr Lys
1 5 10 15 Xaa Pro Ala Glu Xaa Xaa His Gly Thr Val Thr Val Glu Xaa
Gln Tyr 20 25 30 Xaa Gly Xaa Asp Gly Pro Cys Lys Xaa Pro Xaa Gln
Met Ala Val Asp 35 40 45 Xaa Gln Thr Leu Thr Pro Val Gly Arg Leu
Ile Thr Ala Asn Pro Val 50 55 60 Ile Thr Glu Xaa Thr Xaa Asn Ser
Lys Met Met Leu Glu Leu Asp Pro 65 70 75 80 Pro Phe Gly Asp Ser Tyr
Ile Val Ile Gly Xaa Gly Xaa Xaa Lys Ile 85 90 95 Thr His His Trp
His Arg Ser Gly 100 <210> SEQ ID NO 408 <211> LENGTH:
19 <212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: chimeric hinge
<400> SEQUENCE: 408 Glu Ser Lys Tyr Gly Pro Pro Cys Pro Pro
Cys Pro Ala Pro Pro Val 1 5 10 15 Ala Gly Pro <210> SEQ ID NO
409 <211> LENGTH: 95 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZIKV MR766 (AY632535) EDIII <400>
SEQUENCE: 409 Thr Ala Ala Phe Thr Phe Thr Lys Val Pro Ala Glu Thr
Leu His Gly 1 5 10 15 Thr Val Thr Val Glu Val Gln Tyr Ala Gly Thr
Asp Gly Pro Cys Lys 20 25 30 Leu Pro Val Gln Met Ala Val Asp Met
Gln Thr Leu Thr Pro Val Gly 35 40 45 Arg Leu Ile Thr Ala Asn Pro
Val Ile Thr Glu Ser Thr Glu Asn Ser 50 55 60 Lys Met Met Leu Glu
Leu Asp Pro Pro Phe Gly Asp Ser Tyr Ile Val 65 70 75 80 Ile Gly Val
Gly Asp Lys Lys Ile Thr His His Trp His Arg Ser 85 90 95
<210> SEQ ID NO 410 <211> LENGTH: 95 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZIKV KU870645 (USA) 02-Feb-2016
EDIII <400> SEQUENCE: 410 Thr Ala Ala Phe Thr Phe Thr Lys Ile
Pro Ala Glu Thr Leu His Gly 1 5 10 15 Thr Val Thr Val Glu Val Gln
Tyr Ala Gly Thr Asp Gly Pro Cys Lys 20 25 30 Val Pro Ala Gln Met
Ala Val Asp Met Gln Thr Leu Thr Pro Val Gly 35 40 45 Arg Leu Ile
Thr Ala Asn Pro Val Ile Thr Glu Ser Thr Glu Asn Ser 50 55 60 Lys
Met Met Leu Glu Leu Asp Pro Pro Phe Gly Asp Ser Tyr Ile Val 65 70
75 80 Ile Gly Leu Gly Glu Lys Lys Ile Thr His His Trp His Arg Ser
85 90 95 <210> SEQ ID NO 411 <211> LENGTH: 95
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZIKV KU729218
(Brazil) 2015 EDIII <400> SEQUENCE: 411 Thr Ala Ala Phe Thr
Phe Thr Lys Ile Pro Ala Glu Thr Leu His Gly 1 5 10 15 Thr Val Thr
Val Glu Val Gln Tyr Ala Gly Thr Asp Gly Pro Cys Lys 20 25 30 Val
Pro Ala Gln Met Ala Val Asp Thr Gln Thr Leu Thr Pro Val Gly 35 40
45 Arg Leu Ile Thr Ala Asn Pro Val Ile Thr Glu Ser Thr Glu Asn Ser
50 55 60 Lys Met Met Leu Glu Leu Asp Pro Pro Phe Gly Asp Ser Tyr
Ile Val 65 70 75 80 Ile Gly Val Gly Glu Lys Lys Ile Thr His His Trp
His Arg Ser 85 90 95 <210> SEQ ID NO 412 <211> LENGTH:
95 <212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZIKV KU681081
(Thailand) 9-Jul-14 EDIII <400> SEQUENCE: 412 Thr Ala Ala Phe
Thr Phe Thr Lys Ile Pro Ala Glu Thr Leu His Gly 1 5 10 15 Thr Val
Thr Val Glu Val Gln Tyr Ala Gly Thr Asp Gly Pro Cys Lys 20 25 30
Val Pro Ala Gln Met Ala Val Asp Met Gln Thr Leu Thr Pro Val Gly 35
40 45 Arg Leu Ile Thr Ala Asn Pro Val Ile Thr Glu Gly Thr Glu Asn
Ser 50 55 60 Lys Met Met Leu Glu Leu Asp Pro Pro Phe Gly Asp Ser
Tyr Ile Val 65 70 75 80 Ile Gly Val Gly Glu Lys Lys Ile Thr His His
Trp His Arg Ser 85 90 95 <210> SEQ ID NO 413 <211>
LENGTH: 95 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZIKV
KU926310 (Brazil) 29-Jan-2016 EDIII <400> SEQUENCE: 413 Thr
Ala Ala Phe Thr Phe Thr Lys Ile Pro Ala Glu Thr Leu His Gly 1 5 10
15 Thr Val Thr Val Glu Val Gln Tyr Ala Gly Ala Asp Gly Pro Cys Lys
20 25 30 Val Pro Ala Gln Met Ala Val Asp Met Gln Thr Leu Thr Pro
Val Gly 35 40 45 Arg Leu Ile Thr Ala Asn Pro Val Ile Thr Glu Ser
Thr Glu Asn Ser 50 55 60 Lys Met Met Leu Glu Leu Asp Pro Pro Phe
Gly Asp Ser Tyr Ile Val 65 70 75 80 Ile Gly Val Gly Glu Lys Lys Ile
Thr His His Trp His Arg Ser 85 90 95 <210> SEQ ID NO 414
<211> LENGTH: 95 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZIKV KU744693 (China) 6-Feb-16 EDIII <400>
SEQUENCE: 414 Thr Ala Ala Phe Thr Phe Thr Lys Ile Pro Ala Glu Thr
Val Asp Gly 1 5 10 15 Thr Val Thr Val Glu Gly Gln Tyr Gly Gly Thr
Asp Gly Pro Cys Lys 20 25 30 Val Pro Ala Gln Met Ala Val Asp Met
Gln Thr Leu Thr Pro Val Gly 35 40 45 Arg Leu Ile Thr Ala Asn Pro
Val Ile Thr Glu Ser Thr Glu Asn Ser 50 55 60 Lys Met Met Leu Glu
Leu Asp Pro Pro Phe Gly Asp Ser Tyr Ile Val 65 70 75 80 Ile Gly Val
Gly Glu Lys Lys Ile Thr His His Trp His Arg Ser 85 90 95
1 SEQUENCE LISTING <160> NUMBER OF SEQ ID NOS: 414
<210> SEQ ID NO 1 <211> LENGTH: 8 <212> TYPE: PRT
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA190 CDRH1 <400> SEQUENCE: 1
Gly Phe Thr Phe Ser Lys Tyr Gly 1 5 <210> SEQ ID NO 2
<211> LENGTH: 8 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA190 CDRH2 <400> SEQUENCE: 2 Ile Ser Tyr Glu
Gly Ser Asn Lys 1 5 <210> SEQ ID NO 3 <211> LENGTH: 23
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA190 CDRH3
<400> SEQUENCE: 3 Ala Lys Ser Gly Thr Gln Tyr Tyr Asp Thr Thr
Gly Tyr Glu Tyr Arg 1 5 10 15 Gly Leu Glu Tyr Phe Gly Tyr 20
<210> SEQ ID NO 4 <211> LENGTH: 7 <212> TYPE: PRT
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA190 CDRL1 <400> SEQUENCE: 4
Gln Ser Val Ser Ser Ser Tyr 1 5 <210> SEQ ID NO 5 <400>
SEQUENCE: 5 000 <210> SEQ ID NO 6 <211> LENGTH: 9
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA190 CDRL2
long <400> SEQUENCE: 6 Leu Ile Tyr Asp Ala Ser Ser Arg Ala 1
5 <210> SEQ ID NO 7 <211> LENGTH: 9 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA190 CDRL3 <400> SEQUENCE: 7
Gln Gln Tyr Gly Arg Ser Arg Trp Thr 1 5 <210> SEQ ID NO 8
<211> LENGTH: 130 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA190 VH <400> SEQUENCE: 8 Gln Val Gln Leu Val
Glu Ser Gly Gly Gly Val Val Gln Pro Gly Arg 1 5 10 15 Ser Leu Arg
Leu Ser Cys Ala Ala Ser Gly Phe Thr Phe Ser Lys Tyr 20 25 30 Gly
Met His Trp Val Arg Gln Ala Pro Gly Lys Gly Leu Glu Trp Val 35 40
45 Ala Val Ile Ser Tyr Glu Gly Ser Asn Lys Tyr Tyr Ala Asp Ser Val
50 55 60 Lys Gly Arg Phe Thr Ile Ser Arg Asp Asn Ser Lys Asn Thr
Leu Tyr 65 70 75 80 Leu Gln Met Asn Ser Leu Arg Ala Glu Asp Thr Ala
Val Tyr Tyr Cys 85 90 95 Ala Lys Ser Gly Thr Gln Tyr Tyr Asp Thr
Thr Gly Tyr Glu Tyr Arg 100 105 110 Gly Leu Glu Tyr Phe Gly Tyr Trp
Gly Gln Gly Thr Leu Val Thr Val 115 120 125 Ser Ser 130 <210>
SEQ ID NO 9 <211> LENGTH: 108 <212> TYPE: PRT
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA190 VL <400> SEQUENCE: 9
Glu Ile Val Leu Thr Gln Ser Pro Gly Thr Leu Ser Leu Ser Pro Gly 1 5
10 15 Glu Arg Ala Thr Leu Ser Cys Arg Ala Ser Gln Ser Val Ser Ser
Ser 20 25 30 Tyr Leu Ala Trp Tyr Gln Gln Lys Arg Gly Gln Ala Pro
Arg Leu Leu 35 40 45 Ile Tyr Asp Ala Ser Ser Arg Ala Thr Gly Ile
Pro Asp Arg Phe Ser 50 55 60 Gly Ser Gly Ser Gly Thr Asp Phe Thr
Leu Thr Ile Ser Arg Leu Glu 65 70 75 80 Pro Glu Asp Phe Ala Val Tyr
Tyr Cys Gln Gln Tyr Gly Arg Ser Arg 85 90 95 Trp Thr Phe Gly Gln
Gly Thr Lys Val Glu Ile Lys 100 105 <210> SEQ ID NO 10
<211> LENGTH: 24 <212> TYPE: DNA <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA190 CDRH1 <400> SEQUENCE: 10 ggattcacct
tcagtaaata tggc 24 <210> SEQ ID NO 11 <211> LENGTH: 24
<212> TYPE: DNA <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA190 CDRH2
<400> SEQUENCE: 11 atatcatatg agggaagtaa taaa 24 <210>
SEQ ID NO 12 <211> LENGTH: 69 <212> TYPE: DNA
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA190 CDRH3 <400> SEQUENCE:
12 gcgaaatcgg ggacccaata ctatgatact actggttatg agtatagggg
tttggaatac 60 tttggctac 69 <210> SEQ ID NO 13 <211>
LENGTH: 21 <212> TYPE: DNA <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA190
CDRL1 <400> SEQUENCE: 13 cagagtgtta gtagcagtta c 21
<210> SEQ ID NO 14 <400> SEQUENCE: 14 000 <210>
SEQ ID NO 15 <211> LENGTH: 27 <212> TYPE: DNA
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA190 CDRL2 long <400>
SEQUENCE: 15 ctcatctatg atgcatccag cagggcc 27 <210> SEQ ID NO
16 <211> LENGTH: 27 <212> TYPE: DNA <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKA190 CDRL3 <400> SEQUENCE: 16 cagcagtatg
gtaggtcaag gtggaca 27 <210> SEQ ID NO 17 <211> LENGTH:
391 <212> TYPE: DNA <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA190 VH
<400> SEQUENCE: 17 caggtgcagc tggtggagtc tgggggaggc
gtggtccagc ctgggaggtc cctgagactc 60 tcctgtgcag cctctggatt
caccttcagt aaatatggca tgcactgggt ccgccaggct 120 ccaggcaagg
ggctggagtg ggtggcagtt atatcatatg agggaagtaa taaatattat 180
gcagactccg tgaagggccg attcaccatc tccagagaca attccaagaa cacgctgtat
240 ctgcaaatga acagcctgag agctgaggac acggcagtgt attactgtgc
gaaatcgggg 300 acccaatact atgatactac tggttatgag tataggggtt
tggaatactt tggctactgg 360 ggccagggaa ccctggtcac cgtctcctca g 391
<210> SEQ ID NO 18 <211> LENGTH: 325 <212> TYPE:
DNA <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA190 VL <400> SEQUENCE: 18
gaaattgtgt tgacgcagtc tccaggcacc ctgtctttgt ctccagggga aagagccacc
60 ctctcctgca gggccagtca gagtgttagt agcagttact tagcctggta
ccagcagaaa 120 cgtggccagg ctcccaggct cctcatctat gatgcatcca
gcagggccac tggcatccca 180 gacaggttca gtggcagtgg gtctgggaca
gacttcactc tcaccatcag cagactggag 240 cctgaagatt ttgcagtgta
ttactgtcag cagtatggta ggtcaaggtg gacattcggc 300 caagggacca
aggtggaaat caaac 325 <210> SEQ ID NO 19 <211> LENGTH: 8
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA185 CDRH1
<400> SEQUENCE: 19 Gly Tyr Ser Phe Thr Ser Tyr Trp 1 5
<210> SEQ ID NO 20 <211> LENGTH: 8 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA185 CDRH2 <400> SEQUENCE:
20 Phe Asp Pro Ser Asp Ser Gln Thr 1 5 <210> SEQ ID NO 21
<211> LENGTH: 14 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA185 CDRH3 <400> SEQUENCE: 21 Ala Arg Arg Tyr
Cys Ser Ser Ser Ser Cys Tyr Val Asp Asn 1 5 10 <210> SEQ ID
NO 22 <211> LENGTH: 6 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKA185 CDRL1 <400> SEQUENCE: 22 Ala Leu
Pro Asn Lys Phe 1 5 <210> SEQ ID NO 23 <400> SEQUENCE:
23 000 <210> SEQ ID NO 24 <211> LENGTH: 9 <212>
TYPE: PRT <213> ORGANISM: Artificial Sequence <220>
FEATURE: <223> OTHER INFORMATION: ZKA185 CDRL2 long
<400> SEQUENCE: 24 Val Ile Tyr Glu Asp Asn Lys Arg Pro 1 5
<210> SEQ ID NO 25 <211> LENGTH: 12 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA185 CDRL3 <400> SEQUENCE:
25 Tyr Ser Thr Asp Ser Ser Ser Asn Pro Leu Gly Val 1 5 10
<210> SEQ ID NO 26 <211> LENGTH: 121 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA185 VH <400> SEQUENCE: 26
Glu Val Gln Leu Val Gln Ser Gly Ala Glu Val Lys Lys Pro Gly Glu 1 5
10 15 Ser Leu Arg Ile Ser Cys Lys Gly Ser Gly Tyr Ser Phe Thr Ser
Tyr 20 25 30 Trp Ile Thr Trp Val Arg Gln Met Pro Gly Lys Gly Leu
Glu Trp Met 35 40 45 Ala Lys Phe Asp Pro Ser Asp Ser Gln Thr Asn
Tyr Ser Pro Ser Phe 50 55 60 Gln Gly His Val Thr Ile Ser Val Asp
Lys Ser Ile Ser Thr Ala Tyr 65 70 75 80 Leu Gln Trp Ser Ser Leu Lys
Ala Ser Asp Thr Ala Met Tyr Tyr Cys 85 90 95 Ala Arg Arg Tyr Cys
Ser Ser Ser Ser Cys Tyr Val Asp Asn Trp Gly 100 105 110 Gln Gly Thr
Leu Val Thr Ile Phe Ser 115 120 <210> SEQ ID NO 27
<211> LENGTH: 109 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA185 VL <400> SEQUENCE: 27 Ser Tyr Glu Leu Thr
Gln Pro Pro Ser Val Ser Val Ser Pro Gly Gln 1 5 10 15 Thr Ala Arg
Ile Thr Cys Ser Gly Asp Ala Leu Pro Asn Lys Phe Ala 20 25 30 Tyr
Trp Tyr Arg Gln Lys Ser Gly Gln Ala Pro Val Leu Val Ile Tyr 35 40
45 Glu Asp Asn Lys Arg Pro Ser Gly Ile Pro Glu Arg Phe Ser Gly Ser
50 55 60 Ser Ser Gly Thr Met Ala Thr Leu Thr Ile Ser Gly Ala Gln
Val Glu 65 70 75 80 Asp Glu Ala Asp Tyr His Cys Tyr Ser Thr Asp Ser
Ser Ser Asn Pro 85 90 95 Leu Gly Val Phe Gly Gly Gly Thr Lys Leu
Thr Val Leu 100 105 <210> SEQ ID NO 28 <211> LENGTH: 24
<212> TYPE: DNA <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA185 CDRH1
<400> SEQUENCE: 28 ggatatagtt ttaccagtta ctgg 24 <210>
SEQ ID NO 29 <211> LENGTH: 24 <212> TYPE: DNA
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA185 CDRH2 <400> SEQUENCE:
29 tttgatccta gtgactctca aacc 24 <210> SEQ ID NO 30
<211> LENGTH: 42 <212> TYPE: DNA <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA185 CDRH3 <400> SEQUENCE: 30 gcgagaagat
attgtagtag tagtagttgt tatgtggaca at 42 <210> SEQ ID NO 31
<211> LENGTH: 18 <212> TYPE: DNA <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA185 CDRL1 <400> SEQUENCE: 31 gcattgccaa
ataaattt 18 <210> SEQ ID NO 32 <400> SEQUENCE: 32 000
<210> SEQ ID NO 33 <211> LENGTH: 27 <212> TYPE:
DNA <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA185 CDRL2 long
<400> SEQUENCE: 33 gtcatctatg aggacaacaa acgaccc 27
<210> SEQ ID NO 34 <211> LENGTH: 36 <212> TYPE:
DNA <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA185 CDRL3 <400> SEQUENCE:
34 tactcaacag acagcagttc taatcccctg ggagta 36 <210> SEQ ID NO
35 <211> LENGTH: 364 <212> TYPE: DNA <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKA185 VH <400> SEQUENCE: 35 gaagtgcagc
tggtgcagtc cggagcagag gtgaaaaagc ccggggagtc tctgaggatc 60
tcctgtaagg gttctggata tagttttacc agttactgga tcacctgggt gcgccagatg
120 cccgggaaag gcctggagtg gatggcgaag tttgatccta gtgactctca
aaccaactac 180 agcccgtcct tccaaggcca cgtcaccatc tcagttgaca
agtccatcag cactgcctac 240 ttgcagtgga gcagcctgaa ggcctcggac
accgccatgt attactgtgc gagaagatat 300 tgtagtagta gtagttgtta
tgtggacaat tggggccagg gaaccctggt caccatcttc 360 tcag 364
<210> SEQ ID NO 36 <211> LENGTH: 328 <212> TYPE:
DNA <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA185 VL <400> SEQUENCE: 36
tcctatgagc tgacacagcc accctcggtg tcagtgtccc caggacaaac ggccaggatc
60 acctgctctg gagatgcatt gccaaataaa tttgcttatt ggtaccggca
gaagtcaggc 120 caggcccctg ttctggtcat ctatgaggac aacaaacgac
cctccgggat ccctgagaga 180 ttctctggct ccagctcagg gacaatggcc
accttgacta tcagtggggc ccaggtggag 240 gatgaagctg actaccactg
ttactcaaca gacagcagtt ctaatcccct gggagtattc 300 ggcggaggga
ccaagctgac cgtcctag 328 <210> SEQ ID NO 37 <211>
LENGTH: 8 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA230
CDRH1 <400> SEQUENCE: 37 Gly Gly Ser Ile Ser Ser Asp Tyr 1 5
<210> SEQ ID NO 38 <211> LENGTH: 7 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA230 CDRH2 <400> SEQUENCE:
38 Ile Tyr Tyr Ser Gly Ser Thr 1 5 <210> SEQ ID NO 39
<211> LENGTH: 17 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA230 CDRH3 <400> SEQUENCE: 39 Ala Arg Arg Arg
Lys Tyr Asp Ser Leu Trp Gly Ser Phe Ala Phe Asp 1 5 10 15 Ile
<210> SEQ ID NO 40 <211> LENGTH: 8 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA230 CDRL1 <400> SEQUENCE:
40 Ser Ser Asn Ile Gly Gly Asn Tyr 1 5 <210> SEQ ID NO 41
<400> SEQUENCE: 41 000 <210> SEQ ID NO 42 <211>
LENGTH: 9 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA230
CDRL2 long <400> SEQUENCE: 42 Leu Ile Cys Ile Asn Asp His Arg
Pro 1 5 <210> SEQ ID NO 43 <211> LENGTH: 11 <212>
TYPE: PRT <213> ORGANISM: Artificial Sequence <220>
FEATURE: <223> OTHER INFORMATION: ZKA230 CDRL3 <400>
SEQUENCE: 43 Ala Thr Trp Asp Asp Ser Leu Gly Gly Leu Val 1 5 10
<210> SEQ ID NO 44 <211> LENGTH: 123 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA230 VH <400> SEQUENCE: 44
Gln Val Gln Leu Gln Glu Ser Gly Pro Gly Leu Val Lys Pro Ser Glu 1 5
10 15 Thr Leu Ser Leu Thr Cys Ala Val Ser Gly Gly Ser Ile Ser Ser
Asp 20 25 30 Tyr Trp Ser Trp Ile Arg Gln Pro Pro Gly Lys Gly Leu
Glu Trp Ile 35 40 45 Gly Tyr Ile Tyr Tyr Ser Gly Ser Thr Asn Tyr
Asn Pro Ser Leu Lys 50 55 60 Ser Arg Val Thr Ile Ser Val Asp Thr
Ser Lys Asn His Phe Ser Leu 65 70 75 80 Lys Leu Asn Ser Val Thr Ala
Ala Asp Thr Ala Val Tyr Tyr Cys Ala 85 90 95 Arg Arg Arg Lys Tyr
Asp Ser Leu Trp Gly Ser Phe Ala Phe Asp Ile 100 105 110 Trp Gly Gln
Gly Thr Met Val Thr Val Ser Ser 115 120 <210> SEQ ID NO 45
<211> LENGTH: 110 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA230 VL <400> SEQUENCE: 45 Gln Ser Val Leu Thr
Gln Pro Pro Ser Ala Ser Gly Thr Pro Gly Gln 1 5 10 15 Arg Val Thr
Ile Ser Cys Ser Gly Ser Ser Ser Asn Ile Gly Gly Asn 20 25 30 Tyr
Val Tyr Trp Tyr Gln Gln Leu Pro Gly Thr Ala Pro Lys Leu Leu 35 40
45 Ile Cys Ile Asn Asp His Arg Pro Ser Gly Val Pro Asp Arg Phe Ser
50 55 60 Gly Ser Lys Ser Gly Thr Ser Ala Ser Leu Ala Ile Ser Gly
Leu Gln 65 70 75 80 Ser Glu Asp Glu Ala Asp Tyr Tyr Cys Ala Thr Trp
Asp Asp Ser Leu 85 90 95 Gly Gly Leu Val Phe Gly Gly Gly Thr Lys
Leu Thr Val Leu 100 105 110 <210> SEQ ID NO 46 <211>
LENGTH: 24 <212> TYPE: DNA <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA230
CDRH1 <400> SEQUENCE: 46 ggtggctcca tcagtagtga ctac 24
<210> SEQ ID NO 47 <211> LENGTH: 21 <212> TYPE:
DNA <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA230 CDRH2 <400> SEQUENCE:
47 atctattaca gtgggagcac c 21 <210> SEQ ID NO 48 <211>
LENGTH: 51 <212> TYPE: DNA <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA230
CDRH3 <400> SEQUENCE: 48 gcgaggagga ggaagtatga ttccctttgg
gggagttttg cttttgatat c 51
<210> SEQ ID NO 49 <211> LENGTH: 24 <212> TYPE:
DNA <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA230 CDRL1 <400> SEQUENCE:
49 agctccaaca tcggaggtaa ttat 24 <210> SEQ ID NO 50
<400> SEQUENCE: 50 000 <210> SEQ ID NO 51 <211>
LENGTH: 27 <212> TYPE: DNA <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA230
CDRL2 long <400> SEQUENCE: 51 ctcatctgta ttaatgatca ccggccc
27 <210> SEQ ID NO 52 <211> LENGTH: 33 <212>
TYPE: DNA <213> ORGANISM: Artificial Sequence <220>
FEATURE: <223> OTHER INFORMATION: ZKA230 CDRL3 <400>
SEQUENCE: 52 gcaacatggg atgacagcct gggtggcctt gta 33 <210>
SEQ ID NO 53 <211> LENGTH: 370 <212> TYPE: DNA
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA230 VH <400> SEQUENCE: 53
caggtgcagc tgcaggagtc gggcccaggc ctggtgaagc cttcggagac cctgtccctc
60 acctgcgcag tctctggtgg ctccatcagt agtgactact ggagctggat
ccggcagccc 120 ccagggaagg gactggagtg gattgggtat atctattaca
gtgggagcac caactacaac 180 ccctccctca agagtcgagt caccatatca
gtagacacgt ccaagaacca cttctccctg 240 aagctgaact ctgtgaccgc
tgcggacacg gccgtgtatt actgtgcgag gaggaggaag 300 tatgattccc
tttgggggag ttttgctttt gatatctggg gccaagggac aatggtcacc 360
gtctcttcag 370 <210> SEQ ID NO 54 <211> LENGTH: 331
<212> TYPE: DNA <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA230 VL
<400> SEQUENCE: 54 cagtctgtgc tgactcagcc accctcagcg
tctgggaccc ccgggcagag ggtcaccatc 60 tcttgttctg gaagcagctc
caacatcgga ggtaattatg tatactggta ccagcagctc 120 ccaggaacgg
cccccaaact cctcatctgt attaatgatc accggccctc aggggtccct 180
gaccgattct ctggctccaa gtctggcacc tcagcctccc tggccatcag tgggctccag
240 tccgaggatg aggctgatta ttactgtgca acatgggatg acagcctggg
tggccttgta 300 ttcggcggag ggaccaagct gaccgtccta g 331 <210>
SEQ ID NO 55 <211> LENGTH: 8 <212> TYPE: PRT
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA78 CDRH1 <400> SEQUENCE: 55
Gly Phe Thr Phe Ser Asn Tyr Ala 1 5 <210> SEQ ID NO 56
<211> LENGTH: 8 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA78 CDRH2 <400> SEQUENCE: 56 Ile Gly Arg Asn
Gly Asp Ser Ile 1 5 <210> SEQ ID NO 57 <211> LENGTH: 16
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA78 CDRH3
<400> SEQUENCE: 57 Val Lys Asp Leu Ala Ile Pro Glu Ser Tyr
Arg Ile Glu Ala Asp Tyr 1 5 10 15 <210> SEQ ID NO 58
<211> LENGTH: 12 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA78 CDRL1 <400> SEQUENCE: 58 Gln Ser Val Leu
Tyr Arg Ser Asn Asn Lys Asn Tyr 1 5 10 <210> SEQ ID NO 59
<400> SEQUENCE: 59 000 <210> SEQ ID NO 60 <211>
LENGTH: 9 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA78
CDRL2 long <400> SEQUENCE: 60 Leu Ile Tyr Trp Ala Ser Thr Arg
Glu 1 5 <210> SEQ ID NO 61 <211> LENGTH: 9 <212>
TYPE: PRT <213> ORGANISM: Artificial Sequence <220>
FEATURE: <223> OTHER INFORMATION: ZKA78 CDRL3 <400>
SEQUENCE: 61 Gln Gln Tyr Tyr Ser Ser Pro Arg Thr 1 5 <210>
SEQ ID NO 62 <211> LENGTH: 123 <212> TYPE: PRT
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA78 VH <400> SEQUENCE: 62
Glu Val Gln Leu Ala Glu Ser Gly Gly Gly Leu Val Gln Pro Gly Gly 1 5
10 15 Ser Leu Thr Leu Ser Cys Ser Gly Ser Gly Phe Thr Phe Ser Asn
Tyr 20 25 30 Ala Met Val Trp Ala Arg Gln Ala Pro Gly Lys Gly Leu
Glu Tyr Val 35 40 45 Ser Gly Ile Gly Arg Asn Gly Asp Ser Ile Tyr
Tyr Thr Asp Ser Val 50 55 60 Lys Gly Arg Phe Thr Ile Ser Arg Asp
Asn Ser Lys Ser Met Val Tyr 65 70 75 80 Leu Gln Met Ser Ser Leu Arg
Thr Glu Asp Thr Ala Val Tyr Tyr Cys 85 90 95 Val Lys Asp Leu Ala
Ile Pro Glu Ser Tyr Arg Ile Glu Ala Asp Tyr 100 105 110 Trp Gly Gln
Gly Thr Leu Val Ile Val Ser Ala 115 120 <210> SEQ ID NO 63
<211> LENGTH: 113 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA78 VL <400> SEQUENCE: 63 Asp Ile Val Met Thr
Gln Ser Pro Asp Ser Leu Ala Val Ser Leu Gly 1 5 10 15 Glu Arg Ala
Thr Ile Asn Cys Lys Ser Ser Gln Ser Val Leu Tyr Arg 20 25 30 Ser
Asn Asn Lys Asn Tyr Leu Ser Trp Tyr Gln Gln Lys Pro Gly Gln 35 40
45 Pro Pro Lys Leu Leu Ile Tyr Trp Ala Ser Thr Arg Glu Ser Gly Val
50 55 60 Pro Asp Arg Phe Ser Gly Ser Gly Ser Gly Thr Asp Phe Thr
Leu Thr 65 70 75 80 Ile Ser Pro Leu Gln Ala Glu Asp Val Ala Val Tyr
Tyr Cys Gln Gln 85 90 95 Tyr Tyr Ser Ser Pro Arg Thr Phe Gly Gln
Gly Thr Lys Val Glu Ile 100 105 110 Lys <210> SEQ ID NO 64
<211> LENGTH: 24 <212> TYPE: DNA <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA78 CDRH1 <400> SEQUENCE: 64 ggcttcactt
ttagtaacta tgca 24
<210> SEQ ID NO 65 <211> LENGTH: 24 <212> TYPE:
DNA <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA78 CDRH2 <400> SEQUENCE: 65
atcgggcgca acggggactc tatc 24 <210> SEQ ID NO 66 <211>
LENGTH: 48 <212> TYPE: DNA <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA78
CDRH3 <400> SEQUENCE: 66 gtgaaagatc tggccatccc cgagtcctac
agaattgaag ctgattat 48 <210> SEQ ID NO 67 <211> LENGTH:
36 <212> TYPE: DNA <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA78 CDRL1
<400> SEQUENCE: 67 cagtccgtgc tgtaccgctc taacaacaag aattac 36
<210> SEQ ID NO 68 <400> SEQUENCE: 68 000 <210>
SEQ ID NO 69 <211> LENGTH: 27 <212> TYPE: DNA
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA78 CDRL2 long <400>
SEQUENCE: 69 ctgatctatt gggcttcaac ccgggaa 27 <210> SEQ ID NO
70 <211> LENGTH: 27 <212> TYPE: DNA <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKA78 CDRL3 <400> SEQUENCE: 70 cagcagtact
attctagtcc tcgaact 27 <210> SEQ ID NO 71 <211> LENGTH:
370 <212> TYPE: DNA <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA78 VH
<400> SEQUENCE: 71 gaggtgcagc tggcagaatc aggcggggga
ctggtccagc ctggcggcag cctgacactg 60 tcttgcagtg gatcaggctt
cacttttagt aactatgcaa tggtgtgggc aaggcaggct 120 cctgggaagg
gactggagta tgtctctggc atcgggcgca acggggactc tatctactat 180
actgatagtg tgaagggccg gttcaccatc agcagagaca atagcaaatc catggtgtac
240 ctgcagatga gctccctgcg aaccgaagac acagcagtgt actattgcgt
gaaagatctg 300 gccatccccg agtcctacag aattgaagct gattattggg
gacagggcac cctggtcatc 360 gtgagcgccg 370 <210> SEQ ID NO 72
<211> LENGTH: 340 <212> TYPE: DNA <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA78 VL <400> SEQUENCE: 72 gacatcgtga
tgacacagtc tccagatagt ctggcagtca gtctggggga gagggccact 60
attaactgca agagctccca gtccgtgctg taccgctcta acaacaagaa ttacctgtct
120 tggtatcagc agaagcccgg acagccccct aaactgctga tctattgggc
ttcaacccgg 180 gaaagcggcg tcccagacag attctcaggc agcgggtccg
gaacagactt caccctgaca 240 attagccccc tgcaggcaga ggacgtggct
gtctactatt gtcagcagta ctattctagt 300 cctcgaactt tcggccaggg
gaccaaggtg gaaatcaaac 340 <210> SEQ ID NO 73 <211>
LENGTH: 8 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA64
CDRH1 <400> SEQUENCE: 73 Gly Tyr Thr Phe Thr Gly Tyr His 1 5
<210> SEQ ID NO 74 <211> LENGTH: 8 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA64 CDRH2 <400> SEQUENCE: 74
Ile Asn Pro Asn Ser Gly Gly Thr 1 5 <210> SEQ ID NO 75
<211> LENGTH: 12 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA64 CDRH3 <400> SEQUENCE: 75 Ala Arg Met Ser
Ser Ser Ile Trp Gly Phe Asp His 1 5 10 <210> SEQ ID NO 76
<211> LENGTH: 6 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA64 CDRL1 <400> SEQUENCE: 76 Gln Ser Val Leu
Ile Asn 1 5 <210> SEQ ID NO 77 <400> SEQUENCE: 77 000
<210> SEQ ID NO 78 <211> LENGTH: 9 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA64 CDRL2 long <400>
SEQUENCE: 78 Leu Ile Tyr Gly Ala Ser Ser Arg Ala 1 5 <210>
SEQ ID NO 79 <211> LENGTH: 10 <212> TYPE: PRT
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA64 CDRL3 <400> SEQUENCE: 79
Gln Gln Tyr Asn Asp Trp Pro Pro Ile Thr 1 5 10 <210> SEQ ID
NO 80 <211> LENGTH: 119 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKA64 VH <400> SEQUENCE: 80 Gln Val Gln
Leu Val Gln Ser Gly Ala Glu Val Lys Lys Pro Gly Ala 1 5 10 15 Ser
Val Lys Val Ser Cys Lys Ala Ser Gly Tyr Thr Phe Thr Gly Tyr 20 25
30 His Ile Asp Trp Val Arg Gln Ala Arg Gly Gln Gly Leu Glu Trp Met
35 40 45 Gly Arg Ile Asn Pro Asn Ser Gly Gly Thr Asn Tyr Ala Gln
Lys Phe 50 55 60 Gln Gly Arg Val Thr Met Thr Arg Asp Thr Ser Ile
Ser Thr Ala Tyr 65 70 75 80 Met Gln Leu Ser Arg Leu Arg Ser Asp Asp
Ser Ala Val Tyr Tyr Cys 85 90 95 Ala Arg Met Ser Ser Ser Ile Trp
Gly Phe Asp His Trp Gly Gln Gly 100 105 110 Thr Leu Val Thr Val Ser
Ser 115 <210> SEQ ID NO 81 <211> LENGTH: 108
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA64 VL
<400> SEQUENCE: 81 Glu Ile Val Met Thr Gln Ser Pro Ala Thr
Leu Ser Val Ser Pro Gly 1 5 10 15 Glu Arg Ala Thr Leu Ser Cys Arg
Ala Ser Gln Ser Val Leu Ile Asn 20 25 30 Leu Ala Trp Tyr Gln Gln
Lys Pro Gly Gln Ala Pro Arg Leu Leu Ile 35 40 45
Tyr Gly Ala Ser Ser Arg Ala Thr Gly Ile Pro Ala Arg Phe Ser Gly 50
55 60 Ser Gly Ser Gly Thr Glu Phe Thr Leu Thr Ile Ser Ser Leu Gln
Ser 65 70 75 80 Glu Asp Phe Ala Val Tyr Tyr Cys Gln Gln Tyr Asn Asp
Trp Pro Pro 85 90 95 Ile Thr Phe Gly Gln Gly Thr Arg Leu Glu Ile
Lys 100 105 <210> SEQ ID NO 82 <211> LENGTH: 24
<212> TYPE: DNA <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA64 CDRH1
<400> SEQUENCE: 82 ggctacacct tcacagggta tcac 24 <210>
SEQ ID NO 83 <211> LENGTH: 24 <212> TYPE: DNA
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA64 CDRH2 <400> SEQUENCE: 83
attaacccta attctggcgg gacc 24 <210> SEQ ID NO 84 <211>
LENGTH: 36 <212> TYPE: DNA <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA64
CDRH3 <400> SEQUENCE: 84 gctcggatga gctcctctat ttggggcttc
gatcat 36 <210> SEQ ID NO 85 <211> LENGTH: 18
<212> TYPE: DNA <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA64 CDRL1
<400> SEQUENCE: 85 cagtctgtgc tgattaac 18 <210> SEQ ID
NO 86 <400> SEQUENCE: 86 000 <210> SEQ ID NO 87
<211> LENGTH: 27 <212> TYPE: DNA <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA64 CDRL2 long <400> SEQUENCE: 87 ctgatctatg
gagcatcctc cagggct 27 <210> SEQ ID NO 88 <211> LENGTH:
30 <212> TYPE: DNA <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA64 CDRL3
<400> SEQUENCE: 88 cagcagtaca atgattggcc ccctatcaca 30
<210> SEQ ID NO 89 <211> LENGTH: 358 <212> TYPE:
DNA <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA64 VH <400> SEQUENCE: 89
caggtgcagc tggtccagag cggagcagag gtgaagaaac ccggcgcctc agtgaaggtc
60 agctgcaaag cttccggcta caccttcaca gggtatcaca tcgactgggt
gaggcaggca 120 agaggacagg gactggaatg gatgggacgg attaacccta
attctggcgg gaccaactac 180 gcccagaagt ttcagggccg agtgactatg
accagagaca ccagcatctc cacagcttat 240 atgcagctgt cccggctgag
atctgacgat agtgccgtct actattgtgc tcggatgagc 300 tcctctattt
ggggcttcga tcattggggg cagggaacac tggtgactgt cagttcag 358
<210> SEQ ID NO 90 <211> LENGTH: 325 <212> TYPE:
DNA <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA64 VL <400> SEQUENCE: 90
gagatcgtga tgactcagtc tccagccacc ctgtcagtca gcccaggaga acgggcaacc
60 ctgtcttgca gagcctccca gtctgtgctg attaacctgg cttggtacca
gcagaagcca 120 ggccaggcac cccgactgct gatctatgga gcatcctcca
gggctaccgg cattcctgca 180 cgcttcagtg gatcaggaag cggaacagag
tttaccctga caatctctag tctgcagtcc 240 gaagacttcg ctgtctacta
ttgtcagcag tacaatgatt ggccccctat cacatttggc 300 caggggacta
gactggagat caagc 325 <210> SEQ ID NO 91 <211> LENGTH: 8
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA15 CDRH1
<400> SEQUENCE: 91 Gly Gly Phe Ile Asn Ser Tyr Tyr 1 5
<210> SEQ ID NO 92 <211> LENGTH: 7 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA15 CDRH2 <400> SEQUENCE: 92
Ile Tyr Lys Ser Gly Ser Thr 1 5 <210> SEQ ID NO 93
<211> LENGTH: 14 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA15 CDRH3 <400> SEQUENCE: 93 Ala Arg Asp Pro
Tyr Gly Asp Tyr Val Lys Ala Phe Asp Ile 1 5 10 <210> SEQ ID
NO 94 <211> LENGTH: 11 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKA15 CDRL1 <400> SEQUENCE: 94 Gln Ser Leu
Leu His Ser Asn Gly Tyr Asn Tyr 1 5 10 <210> SEQ ID NO 95
<400> SEQUENCE: 95 000 <210> SEQ ID NO 96 <211>
LENGTH: 9 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA15
CDRL2 long <400> SEQUENCE: 96 Leu Ile Tyr Leu Gly Ser Asn Arg
Ala 1 5 <210> SEQ ID NO 97 <211> LENGTH: 8 <212>
TYPE: PRT <213> ORGANISM: Artificial Sequence <220>
FEATURE: <223> OTHER INFORMATION: ZKA15 CDRL3 <400>
SEQUENCE: 97 Met Gln Ala Leu Gln Thr Val Thr 1 5 <210> SEQ ID
NO 98 <211> LENGTH: 120 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKA15 VH <400> SEQUENCE: 98 Gln Val Gln
Leu Gln Glu Ser Gly Pro Gly Leu Val Lys Pro Ser Glu 1 5 10 15 Thr
Leu Ser Leu Thr Cys Thr Val Ser Gly Gly Phe Ile Asn Ser Tyr 20 25
30 Tyr Trp Ser Trp Ile Arg Gln Pro Ala Gly Lys Gly Leu Glu Trp Ile
35 40 45 Gly Arg Ile Tyr Lys Ser Gly Ser Thr Asn Tyr Asn Pro Ser
Leu Lys 50 55 60 Ser Arg Val Thr Met Ser Leu Asp Thr Ser Lys Tyr
Gln Phe Ser Leu 65 70 75 80 Lys Leu Arg Ser Val Thr Ala Ala Asp Thr
Ala Val Tyr Tyr Cys Ala 85 90 95 Arg Asp Pro Tyr Gly Asp Tyr Val
Lys Ala Phe Asp Ile Trp Gly Gln 100 105 110
Gly Thr Met Val Thr Val Ser Ser 115 120 <210> SEQ ID NO 99
<211> LENGTH: 111 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA15 VL <400> SEQUENCE: 99 Asp Ile Val Met Thr
Gln Ser Pro Leu Ser Leu Pro Val Thr Pro Gly 1 5 10 15 Glu Pro Ala
Ser Ile Ser Cys Arg Ser Ser Gln Ser Leu Leu His Ser 20 25 30 Asn
Gly Tyr Asn Tyr Leu Asn Trp Tyr Leu Gln Lys Pro Gly Gln Ser 35 40
45 Pro Gln Leu Leu Ile Tyr Leu Gly Ser Asn Arg Ala Ser Gly Val Pro
50 55 60 Asp Arg Phe Ser Gly Ser Gly Ser Gly Thr Asp Phe Thr Leu
Lys Ile 65 70 75 80 Ser Arg Val Glu Ala Glu Asp Val Gly Val Tyr Tyr
Cys Met Gln Ala 85 90 95 Leu Gln Thr Val Thr Phe Gly Pro Gly Thr
Lys Val Asp Ile Lys 100 105 110 <210> SEQ ID NO 100
<211> LENGTH: 24 <212> TYPE: DNA <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA15 CDRH1 <400> SEQUENCE: 100 ggtggcttca
tcaatagtta ctac 24 <210> SEQ ID NO 101 <211> LENGTH: 21
<212> TYPE: DNA <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA15 CDRH2
<400> SEQUENCE: 101 atctataaaa gtgggagcac c 21 <210>
SEQ ID NO 102 <211> LENGTH: 42 <212> TYPE: DNA
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA15 CDRH3 <400> SEQUENCE:
102 gcgagagatc cctacggtga ctacgttaag gcttttgata tt 42 <210>
SEQ ID NO 103 <211> LENGTH: 33 <212> TYPE: DNA
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA15 CDRL1 <400> SEQUENCE:
103 cagagcctcc tgcatagtaa tggatacaac tat 33 <210> SEQ ID NO
104 <400> SEQUENCE: 104 000 <210> SEQ ID NO 105
<211> LENGTH: 27 <212> TYPE: DNA <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA15 CDRL2 long <400> SEQUENCE: 105 ctgatctatt
tgggttctaa tcgggcc 27 <210> SEQ ID NO 106 <211> LENGTH:
24 <212> TYPE: DNA <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA15 CDRL3
<400> SEQUENCE: 106 atgcaagctc tacaaactgt cact 24 <210>
SEQ ID NO 107 <211> LENGTH: 361 <212> TYPE: DNA
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA15 VH <400> SEQUENCE: 107
caggtgcagc tgcaggagtc ggggccagga ctggtgaagc cttcggagac cctgtccctc
60 acctgcactg tctccggtgg cttcatcaat agttactact ggagctggat
ccggcagccc 120 gccgggaagg gactggagtg gattgggcgt atctataaaa
gtgggagcac caactacaac 180 ccctccctca agagtcgagt caccatgtca
ctagacacgt ccaagtacca gttctccctg 240 aagctgaggt ctgtgaccgc
cgctgacacg gccgtgtatt actgtgcgag agatccctac 300 ggtgactacg
ttaaggcttt tgatatttgg ggccaaggga caatggtcac cgtctcttca 360 g 361
<210> SEQ ID NO 108 <211> LENGTH: 334 <212> TYPE:
DNA <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA15 VL <400> SEQUENCE: 108
gatattgtga tgactcagtc tccactctcc ctgcccgtca cccctggaga gccggcctcc
60 atctcctgca ggtctagtca gagcctcctg catagtaatg gatacaacta
tttgaattgg 120 tacctgcaga agccagggca gtctccacag ctcctgatct
atttgggttc taatcgggcc 180 tccggggtcc ctgacaggtt cagtggcagt
ggatcaggca cagattttac actgaaaatc 240 agcagagtgg aggctgagga
tgttggggtt tattactgca tgcaagctct acaaactgtc 300 actttcggcc
ctgggaccaa agtggatatc aaac 334 <210> SEQ ID NO 109
<211> LENGTH: 8 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA25 CDRH1 <400> SEQUENCE: 109 Gly Phe Thr Phe
Arg Ser His Trp 1 5 <210> SEQ ID NO 110 <211> LENGTH: 8
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA25 CDRH2
<400> SEQUENCE: 110 Ile Lys Glu Asp Gly Tyr Glu Lys 1 5
<210> SEQ ID NO 111 <211> LENGTH: 14 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA25 CDRH3 <400> SEQUENCE:
111 Ala Arg Asp Leu Arg Val Tyr Ser Gly Arg Gly Phe Asp Pro 1 5 10
<210> SEQ ID NO 112 <211> LENGTH: 6 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA25 CDRL1 <400> SEQUENCE:
112 Lys Leu Gly Asp Lys Tyr 1 5 <210> SEQ ID NO 113
<400> SEQUENCE: 113 000 <210> SEQ ID NO 114 <211>
LENGTH: 9 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA25
CDRL2 long <400> SEQUENCE: 114 Val Ile Tyr Gln Asp Ser Lys
Arg Pro 1 5 <210> SEQ ID NO 115 <211> LENGTH: 9
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA25 CDRL3
<400> SEQUENCE: 115 Gln Ala Trp Asp Ser Ser Thr Val Val 1 5
<210> SEQ ID NO 116 <211> LENGTH: 121 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220>
FEATURE:
<223> OTHER INFORMATION: ZKA25 VH <400> SEQUENCE: 116
Glu Val Gln Leu Val Glu Ser Gly Gly Gly Leu Val Arg Pro Gly Gly 1 5
10 15 Ser Leu Arg Leu Ser Cys Ala Ala Ser Gly Phe Thr Phe Arg Ser
His 20 25 30 Trp Met Ser Trp Val Arg Gln Ala Pro Gly Lys Gly Leu
Glu Trp Val 35 40 45 Ala Asn Ile Lys Glu Asp Gly Tyr Glu Lys Tyr
Tyr Val Asp Ser Val 50 55 60 Lys Gly Arg Phe Thr Ile Ser Arg Asp
Asn Ala Lys Asn Ser Leu Tyr 65 70 75 80 Leu Gln Met Lys Ser Leu Arg
Ala Glu Asp Thr Ala Val Tyr Tyr Cys 85 90 95 Ala Arg Asp Leu Arg
Val Tyr Ser Gly Arg Gly Phe Asp Pro Trp Gly 100 105 110 Gln Gly Thr
Leu Val Thr Val Ser Ser 115 120 <210> SEQ ID NO 117
<211> LENGTH: 106 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA25 VL <400> SEQUENCE: 117 Ser Tyr Glu Leu Thr
Gln Pro Pro Ser Leu Ser Val Ser Pro Gly Gln 1 5 10 15 Thr Ala Ser
Ile Thr Cys Ser Gly Asp Lys Leu Gly Asp Lys Tyr Ala 20 25 30 Cys
Trp Tyr Gln Gln Lys Pro Gly Gln Ser Pro Val Leu Val Ile Tyr 35 40
45 Gln Asp Ser Lys Arg Pro Ser Gly Ile Pro Ala Arg Phe Ser Gly Ser
50 55 60 Asn Ser Gly Asn Thr Ala Thr Leu Thr Ile Ser Gly Thr Gln
Ala Met 65 70 75 80 Asp Glu Ala Asp Tyr Tyr Cys Gln Ala Trp Asp Ser
Ser Thr Val Val 85 90 95 Phe Gly Gly Gly Thr Lys Leu Thr Val Leu
100 105 <210> SEQ ID NO 118 <211> LENGTH: 24
<212> TYPE: DNA <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA25 CDRH1
<400> SEQUENCE: 118 ggattcacct ttagaagtca ttgg 24 <210>
SEQ ID NO 119 <211> LENGTH: 24 <212> TYPE: DNA
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA25 CDRH2 <400> SEQUENCE:
119 ataaaggaag atggatatga gaaa 24 <210> SEQ ID NO 120
<211> LENGTH: 42 <212> TYPE: DNA <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA25 CDRH3 <400> SEQUENCE: 120 gcgagagatt
tgagggtata tagtgggaga ggtttcgacc cc 42 <210> SEQ ID NO 121
<211> LENGTH: 18 <212> TYPE: DNA <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA25 CDRL1 <400> SEQUENCE: 121 aaattggggg
ataaatat 18 <210> SEQ ID NO 122 <400> SEQUENCE: 122 000
<210> SEQ ID NO 123 <211> LENGTH: 27 <212> TYPE:
DNA <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA25 CDRL2 long <400>
SEQUENCE: 123 gtcatctatc aagatagcaa gcggccc 27 <210> SEQ ID
NO 124 <211> LENGTH: 27 <212> TYPE: DNA <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKA25 CDRL3 <400> SEQUENCE: 124 caggcgtggg
acagcagcac tgtggta 27 <210> SEQ ID NO 125 <211> LENGTH:
364 <212> TYPE: DNA <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA25 VH
<400> SEQUENCE: 125 gaggtgcagt tggtggagtc tgggggaggc
ttggtccggc ctggggggtc cctgagactc 60 tcctgtgcag cctctggatt
cacctttaga agtcattgga tgagttgggt ccgccaggct 120 ccagggaagg
ggctggagtg ggtggccaac ataaaggaag atggatatga gaaatactat 180
gtggactctg tgaagggccg attcaccatc tccagagaca acgccaagaa ctcactgtat
240 ctgcaaatga agagcctgag agccgaggac acggccgtgt attactgtgc
gagagatttg 300 agggtatata gtgggagagg tttcgacccc tggggccagg
gaaccctggt caccgtctcc 360 tcag 364 <210> SEQ ID NO 126
<211> LENGTH: 319 <212> TYPE: DNA <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA25 VL <400> SEQUENCE: 126 tcctatgagc
tgactcagcc accctcactg tccgtgtccc caggacagac agccagcatc 60
acctgctctg gagataaatt gggggataaa tatgcttgct ggtatcagca gaagccaggc
120 cagtcccctg tgttggtcat ctatcaagat agcaagcggc cctcagggat
ccctgcgcga 180 ttctctggct ccaactctgg gaacacagcc actctgacca
tcagcgggac ccaggctatg 240 gatgaggctg actattactg tcaggcgtgg
gacagcagca ctgtggtatt cggtggaggg 300 accaagctga ccgtcctag 319
<210> SEQ ID NO 127 <211> LENGTH: 10 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA35 CDRH1 <400> SEQUENCE:
127 Gly Gly Ser Ile Ser Thr Gly Gly Tyr Tyr 1 5 10 <210> SEQ
ID NO 128 <211> LENGTH: 7 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKA35 CDRH2 <400> SEQUENCE: 128 Ile Tyr
Tyr Ser Gly Asn Thr 1 5 <210> SEQ ID NO 129 <211>
LENGTH: 12 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA35
CDRH3 <400> SEQUENCE: 129 Ala Lys Gly Gly Gly Arg Glu Arg Pro
Phe Asp Tyr 1 5 10 <210> SEQ ID NO 130 <211> LENGTH: 8
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA35 CDRL1
<400> SEQUENCE: 130 Ser Ser Asn Ile Gly Arg Asn Tyr 1 5
<210> SEQ ID NO 131 <400> SEQUENCE: 131 000 <210>
SEQ ID NO 132 <211> LENGTH: 9 <212> TYPE: PRT
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA35 CDRL2 long
<400> SEQUENCE: 132 Leu Ile Tyr Arg Asn Asn Gln Arg Pro 1 5
<210> SEQ ID NO 133 <211> LENGTH: 12 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA35 CDRL3 <400> SEQUENCE:
133 Val Ala Trp Asp Asp Ser Arg Ser Gly Phe Val Val 1 5 10
<210> SEQ ID NO 134 <211> LENGTH: 120 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA35 VH <400> SEQUENCE: 134
Gln Val Gln Leu Gln Glu Ser Gly Pro Gly Leu Val Lys Pro Ser Gln 1 5
10 15 Thr Leu Ser Leu Thr Cys Thr Val Ser Gly Gly Ser Ile Ser Thr
Gly 20 25 30 Gly Tyr Tyr Trp Ser Trp Ile Arg Gln His Pro Gly Lys
Gly Leu Glu 35 40 45 Trp Ile Gly Tyr Ile Tyr Tyr Ser Gly Asn Thr
Tyr Tyr Asn Pro Ser 50 55 60 Leu Lys Ser Arg Val Thr Ile Ser Val
Asp Thr Ser Lys Lys Gln Phe 65 70 75 80 Ser Leu Lys Leu Ser Ser Val
Thr Ala Ala Asp Thr Ala Val Tyr Tyr 85 90 95 Cys Ala Lys Gly Gly
Gly Arg Glu Arg Pro Phe Asp Tyr Trp Gly Gln 100 105 110 Gly Thr Leu
Val Thr Val Ser Ser 115 120 <210> SEQ ID NO 135 <211>
LENGTH: 111 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA35
VL <400> SEQUENCE: 135 Gln Ser Val Leu Thr Gln Pro Pro Ser
Ala Ser Gly Thr Pro Gly Gln 1 5 10 15 Arg Val Thr Ile Ser Cys Ser
Gly Ser Ser Ser Asn Ile Gly Arg Asn 20 25 30 Tyr Val Asp Trp Tyr
Gln Gln Leu Pro Gly Thr Ala Pro Lys Leu Leu 35 40 45 Ile Tyr Arg
Asn Asn Gln Arg Pro Ser Gly Val Pro Glu Arg Phe Ser 50 55 60 Gly
Ser Lys Ser Gly Thr Ser Ala Ser Leu Ala Ile Ser Gly Leu Arg 65 70
75 80 Ser Glu Asp Glu Ala Asp Tyr Tyr Cys Val Ala Trp Asp Asp Ser
Arg 85 90 95 Ser Gly Phe Val Val Phe Gly Gly Gly Thr Lys Val Thr
Val Leu 100 105 110 <210> SEQ ID NO 136 <211> LENGTH:
30 <212> TYPE: DNA <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA35 CDRH1
<400> SEQUENCE: 136 ggtggctcca tcagcactgg tggttactac 30
<210> SEQ ID NO 137 <211> LENGTH: 21 <212> TYPE:
DNA <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA35 CDRH2 <400> SEQUENCE:
137 atctattaca gtgggaacac c 21 <210> SEQ ID NO 138
<211> LENGTH: 36 <212> TYPE: DNA <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA35 CDRH3 <400> SEQUENCE: 138 gcgaaaggag
gagggaggga gcgacccttt gactac 36 <210> SEQ ID NO 139
<211> LENGTH: 24 <212> TYPE: DNA <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA35 CDRL1 <400> SEQUENCE: 139 agctccaaca
tcggaagaaa ttat 24 <210> SEQ ID NO 140 <400> SEQUENCE:
140 000 <210> SEQ ID NO 141 <211> LENGTH: 27
<212> TYPE: DNA <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA35 CDRL2
long <400> SEQUENCE: 141 ctcatctata ggaataatca gcggccc 27
<210> SEQ ID NO 142 <211> LENGTH: 36 <212> TYPE:
DNA <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA35 CDRL3 <400> SEQUENCE:
142 gtagcatggg atgacagccg gagtggtttt gtggta 36 <210> SEQ ID
NO 143 <211> LENGTH: 361 <212> TYPE: DNA <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKA35 VH <400> SEQUENCE: 143 caggtgcagc
tgcaggagtc gggcccagga ctggtgaagc cttcacagac cctgtccctc 60
acctgcactg tctctggtgg ctccatcagc actggtggtt actactggag ctggatccgc
120 cagcacccag ggaagggcct ggagtggatt ggttacatct attacagtgg
gaacacctac 180 tacaacccgt ccctcaagag tcgagttacc atatcagttg
acacctctaa gaagcagttc 240 tccctgaagc tgagctctgt gactgccgcg
gacacggccg tgtattactg tgcgaaagga 300 ggagggaggg agcgaccctt
tgactactgg ggccagggaa ccctggtcac cgtctcctca 360 g 361 <210>
SEQ ID NO 144 <211> LENGTH: 334 <212> TYPE: DNA
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA35 VL <400> SEQUENCE: 144
cagtctgtgc tgactcagcc accctcagcg tctgggaccc ccgggcagag ggtcaccatc
60 tcttgttctg gaagcagctc caacatcgga agaaattatg tagactggta
ccagcaactc 120 ccaggaacgg cccccaaact cctcatctat aggaataatc
agcggccctc aggggtccct 180 gagcgattct ctggctccaa gtctggcacc
tcagcctccc tggccatcag tgggctccgg 240 tccgaggatg aggctgatta
ttactgtgta gcatgggatg acagccggag tggttttgtg 300 gtattcggcg
gagggaccaa ggtgaccgtc ctag 334 <210> SEQ ID NO 145
<211> LENGTH: 330 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: IgG1 CH1-CH2-CH3 aa <400> SEQUENCE: 145 Ala Ser
Thr Lys Gly Pro Ser Val Phe Pro Leu Ala Pro Ser Ser Lys 1 5 10 15
Ser Thr Ser Gly Gly Thr Ala Ala Leu Gly Cys Leu Val Lys Asp Tyr 20
25 30 Phe Pro Glu Pro Val Thr Val Ser Trp Asn Ser Gly Ala Leu Thr
Ser 35 40 45 Gly Val His Thr Phe Pro Ala Val Leu Gln Ser Ser Gly
Leu Tyr Ser 50 55 60 Leu Ser Ser Val Val Thr Val Pro Ser Ser Ser
Leu Gly Thr Gln Thr 65 70 75 80 Tyr Ile Cys Asn Val Asn His Lys Pro
Ser Asn Thr Lys Val Asp Lys 85 90 95 Arg Val Glu Pro Lys Ser Cys
Asp Lys Thr His Thr Cys Pro Pro Cys 100 105 110 Pro Ala Pro Glu Leu
Leu Gly Gly Pro Ser Val Phe Leu Phe Pro Pro 115 120 125 Lys Pro Lys
Asp Thr Leu Met Ile Ser Arg Thr Pro Glu Val Thr Cys 130 135 140 Val
Val Val Asp Val Ser His Glu Asp Pro Glu Val Lys Phe Asn Trp 145 150
155 160 Tyr Val Asp Gly Val Glu Val His Asn Ala Lys Thr Lys Pro Arg
Glu 165 170 175
Glu Gln Tyr Asn Ser Thr Tyr Arg Val Val Ser Val Leu Thr Val Leu 180
185 190 His Gln Asp Trp Leu Asn Gly Lys Glu Tyr Lys Cys Lys Val Ser
Asn 195 200 205 Lys Ala Leu Pro Ala Pro Ile Glu Lys Thr Ile Ser Lys
Ala Lys Gly 210 215 220 Gln Pro Arg Glu Pro Gln Val Tyr Thr Leu Pro
Pro Ser Arg Glu Glu 225 230 235 240 Met Thr Lys Asn Gln Val Ser Leu
Thr Cys Leu Val Lys Gly Phe Tyr 245 250 255 Pro Ser Asp Ile Ala Val
Glu Trp Glu Ser Asn Gly Gln Pro Glu Asn 260 265 270 Asn Tyr Lys Thr
Thr Pro Pro Val Leu Asp Ser Asp Gly Ser Phe Phe 275 280 285 Leu Tyr
Ser Lys Leu Thr Val Asp Lys Ser Arg Trp Gln Gln Gly Asn 290 295 300
Val Phe Ser Cys Ser Val Met His Glu Ala Leu His Asn His Tyr Thr 305
310 315 320 Gln Lys Ser Leu Ser Leu Ser Pro Gly Lys 325 330
<210> SEQ ID NO 146 <211> LENGTH: 330 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: IgG1 CH1-CH2-CH3 LALA aa <400>
SEQUENCE: 146 Ala Ser Thr Lys Gly Pro Ser Val Phe Pro Leu Ala Pro
Ser Ser Lys 1 5 10 15 Ser Thr Ser Gly Gly Thr Ala Ala Leu Gly Cys
Leu Val Lys Asp Tyr 20 25 30 Phe Pro Glu Pro Val Thr Val Ser Trp
Asn Ser Gly Ala Leu Thr Ser 35 40 45 Gly Val His Thr Phe Pro Ala
Val Leu Gln Ser Ser Gly Leu Tyr Ser 50 55 60 Leu Ser Ser Val Val
Thr Val Pro Ser Ser Ser Leu Gly Thr Gln Thr 65 70 75 80 Tyr Ile Cys
Asn Val Asn His Lys Pro Ser Asn Thr Lys Val Asp Lys 85 90 95 Lys
Val Glu Pro Lys Ser Cys Asp Lys Thr His Thr Cys Pro Pro Cys 100 105
110 Pro Ala Pro Glu Ala Ala Gly Gly Pro Ser Val Phe Leu Phe Pro Pro
115 120 125 Lys Pro Lys Asp Thr Leu Met Ile Ser Arg Thr Pro Glu Val
Thr Cys 130 135 140 Val Val Val Asp Val Ser His Glu Asp Pro Glu Val
Lys Phe Asn Trp 145 150 155 160 Tyr Val Asp Gly Val Glu Val His Asn
Ala Lys Thr Lys Pro Arg Glu 165 170 175 Glu Gln Tyr Asn Ser Thr Tyr
Arg Val Val Ser Val Leu Thr Val Leu 180 185 190 His Gln Asp Trp Leu
Asn Gly Lys Glu Tyr Lys Cys Lys Val Ser Asn 195 200 205 Lys Ala Leu
Pro Ala Pro Ile Glu Lys Thr Ile Ser Lys Ala Lys Gly 210 215 220 Gln
Pro Arg Glu Pro Gln Val Tyr Thr Leu Pro Pro Ser Arg Asp Glu 225 230
235 240 Leu Thr Lys Asn Gln Val Ser Leu Thr Cys Leu Val Lys Gly Phe
Tyr 245 250 255 Pro Ser Asp Ile Ala Val Glu Trp Glu Ser Asn Gly Gln
Pro Glu Asn 260 265 270 Asn Tyr Lys Thr Thr Pro Pro Val Leu Asp Ser
Asp Gly Ser Phe Phe 275 280 285 Leu Tyr Ser Lys Leu Thr Val Asp Lys
Ser Arg Trp Gln Gln Gly Asn 290 295 300 Val Phe Ser Cys Ser Val Met
His Glu Ala Leu His Asn His Tyr Thr 305 310 315 320 Gln Lys Ser Leu
Ser Leu Ser Pro Gly Lys 325 330 <210> SEQ ID NO 147
<211> LENGTH: 107 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: IgG CK aa <400> SEQUENCE: 147 Arg Thr Val Ala
Ala Pro Ser Val Phe Ile Phe Pro Pro Ser Asp Glu 1 5 10 15 Gln Leu
Lys Ser Gly Thr Ala Ser Val Val Cys Leu Leu Asn Asn Phe 20 25 30
Tyr Pro Arg Glu Ala Lys Val Gln Trp Lys Val Asp Asn Ala Leu Gln 35
40 45 Ser Gly Asn Ser Gln Glu Ser Val Thr Glu Gln Asp Ser Lys Asp
Ser 50 55 60 Thr Tyr Ser Leu Ser Ser Thr Leu Thr Leu Ser Lys Ala
Asp Tyr Glu 65 70 75 80 Lys His Lys Val Tyr Ala Cys Glu Val Thr His
Gln Gly Leu Ser Ser 85 90 95 Pro Val Thr Lys Ser Phe Asn Arg Gly
Glu Cys 100 105 <210> SEQ ID NO 148 <211> LENGTH: 106
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: IgG CL aa
<400> SEQUENCE: 148 Gly Gln Pro Lys Ala Ala Pro Ser Val Thr
Leu Phe Pro Pro Ser Ser 1 5 10 15 Glu Glu Leu Gln Ala Asn Lys Ala
Thr Leu Val Cys Leu Ile Ser Asp 20 25 30 Phe Tyr Pro Gly Ala Val
Thr Val Ala Trp Lys Ala Asp Ser Ser Pro 35 40 45 Val Lys Ala Gly
Val Glu Thr Thr Thr Pro Ser Lys Gln Ser Asn Asn 50 55 60 Lys Tyr
Ala Ala Ser Ser Tyr Leu Ser Leu Thr Pro Glu Gln Trp Lys 65 70 75 80
Ser His Arg Ser Tyr Ser Cys Gln Val Thr His Glu Gly Ser Thr Val 85
90 95 Glu Lys Thr Val Ala Pro Thr Glu Cys Ser 100 105 <210>
SEQ ID NO 149 <211> LENGTH: 990 <212> TYPE: DNA
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: IgG1 CH1-CH2-CH3 nucl <400>
SEQUENCE: 149 gcgtcgacca agggcccatc ggtcttcccc ctggcaccct
cctccaagag cacctctggg 60 ggcacagcgg ccctgggctg cctggtcaag
gactacttcc ccgaacctgt gacggtctcg 120 tggaactcag gcgccctgac
cagcggcgtg cacaccttcc cggctgtcct acagtcctca 180 ggactctact
ccctcagcag cgtggtgacc gtgccctcca gcagcttggg cacccagacc 240
tacatctgca acgtgaatca caagcccagc aacaccaagg tggacaagag agttgagccc
300 aaatcttgtg acaaaactca cacatgccca ccgtgcccag cacctgaact
cctgggggga 360 ccgtcagtct tcctcttccc cccaaaaccc aaggacaccc
tcatgatctc ccggacccct 420 gaggtcacat gcgtggtggt ggacgtgagc
cacgaagacc ctgaggtcaa gttcaactgg 480 tacgtggacg gcgtggaggt
gcataatgcc aagacaaagc cgcgggagga gcagtacaac 540 agcacgtacc
gtgtggtcag cgtcctcacc gtcctgcacc aggactggct gaatggcaag 600
gagtacaagt gcaaggtctc caacaaagcc ctcccagccc ccatcgagaa aaccatctcc
660 aaagccaaag ggcagccccg agaaccacag gtgtacaccc tgcccccatc
ccgggaggag 720 atgaccaaga accaggtcag cctgacctgc ctggtcaaag
gcttctatcc cagcgacatc 780 gccgtggagt gggagagcaa tgggcagccg
gagaacaact acaagaccac gcctcccgtg 840 ctggactccg acggctcctt
cttcctctat agcaagctca ccgtggacaa gagcaggtgg 900 cagcagggga
acgtcttctc atgctccgtg atgcatgagg ctctgcacaa ccactacacg 960
cagaagagcc tctccctgtc cccgggtaaa 990 <210> SEQ ID NO 150
<211> LENGTH: 990 <212> TYPE: DNA <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: IgG1 CH1-CH2-CH3 LALA nucl <400> SEQUENCE: 150
gcgtcgacca agggcccatc ggtcttcccc ctggcaccct cctccaagag cacctctggg
60 ggcacagcgg ccctgggctg cctggtcaag gactacttcc ccgaacctgt
gacggtctcg 120 tggaactcag gcgccctgac cagcggcgtg cacaccttcc
cggctgtcct acagtcctca 180 ggactctact ccctcagcag cgtggtgacc
gtgccctcca gcagcttggg cacccagacc 240 tacatctgca acgtgaatca
caagcccagc aacaccaagg tggacaagag agttgagccc 300 aaatcttgtg
acaaaactca cacatgccca ccgtgcccag cacctgaagc cgcgggggga 360
ccgtcagtct tcctcttccc cccaaaaccc aaggacaccc tcatgatctc ccggacccct
420 gaggtcacat gcgtggtggt ggacgtgagc cacgaagacc ctgaggtcaa
gttcaactgg 480 tacgtggacg gcgtggaggt gcataatgcc aagacaaagc
cgcgggagga gcagtacaac 540 agcacgtacc gtgtggtcag cgtcctcacc
gtcctgcacc aggactggct gaatggcaag 600 gagtacaagt gcaaggtctc
caacaaagcc ctcccagccc ccatcgagaa aaccatctcc 660 aaagccaaag
ggcagccccg agaaccacag gtgtacaccc tgcccccatc ccgggaggag 720
atgaccaaga accaggtcag cctgacctgc ctggtcaaag gcttctatcc cagcgacatc
780 gccgtggagt gggagagcaa tgggcagccg gagaacaact acaagaccac
gcctcccgtg 840 ctggactccg acggctcctt cttcctctat agcaagctca
ccgtggacaa gagcaggtgg 900 cagcagggga acgtcttctc atgctccgtg
atgcatgagg ctctgcacaa ccactacacg 960
cagaagagcc tctccctgtc cccgggtaaa 990 <210> SEQ ID NO 151
<211> LENGTH: 321 <212> TYPE: DNA <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: IgG CK nucl <400> SEQUENCE: 151 cgtacggtgg
ctgcaccatc tgtcttcatc ttcccgccat ctgatgagca gttgaaatct 60
ggaactgcct ctgttgtgtg cctgctgaat aacttctatc ccagagaggc caaagtacag
120 tggaaggtgg ataacgccct ccaatcgggt aactcccagg agagtgtcac
agagcaggac 180 agcaaggaca gcacctacag cctcagcagc accctgacgc
tgagcaaagc agactacgag 240 aaacacaaag tctacgcctg cgaagtcacc
catcagggcc tgagctcgcc cgtcacaaag 300 agcttcaaca ggggagagtg t 321
<210> SEQ ID NO 152 <211> LENGTH: 318 <212> TYPE:
DNA <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: IgG CL nucl <400> SEQUENCE:
152 ggtcagccca aggctgcccc ctcggtcact ctgttcccgc cctcctctga
ggagcttcaa 60 gccaacaagg ccacactggt gtgtctcata agtgacttct
acccgggagc cgtgacagtg 120 gcttggaaag cagatagcag ccccgtcaag
gcgggagtgg agaccaccac accctccaaa 180 caaagcaaca acaagtacgc
ggccagcagc tatctgagcc tgacgcctga gcagtggaag 240 tcccacagaa
gctacagctg ccaggtcacg catgaaggga gcaccgtgga gaagacagtg 300
gcccctacag aatgttca 318 <210> SEQ ID NO 153 <211>
LENGTH: 8 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA10
CDRH1 <400> SEQUENCE: 153 Gly Phe Thr Phe Ser Asp Ser Tyr 1 5
<210> SEQ ID NO 154 <211> LENGTH: 8 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA10 CDRH2 <400> SEQUENCE:
154 Ile Ser Ser Ser Ser Pro Phe Thr 1 5 <210> SEQ ID NO 155
<211> LENGTH: 17 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA10 CDRH3 <400> SEQUENCE: 155 Ala Arg Gly Leu
Val Arg Asp Gly Tyr Lys Trp Leu Tyr Phe Phe Asp 1 5 10 15 Tyr
<210> SEQ ID NO 156 <211> LENGTH: 124 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA10 VH <400> SEQUENCE: 156
Gln Val Gln Leu Val Glu Ser Gly Gly Gly Leu Val Glu Pro Arg Gly 1 5
10 15 Ser Leu Arg Leu Ser Cys Ala Ala Ser Gly Phe Thr Phe Ser Asp
Ser 20 25 30 Tyr Met Ser Trp Ile Arg Gln Ala Pro Gly Lys Gly Leu
Glu Trp Ile 35 40 45 Ser Tyr Ile Ser Ser Ser Ser Pro Phe Thr Asn
Tyr Ala Asp Ser Val 50 55 60 Lys Gly Arg Phe Thr Ile Ser Arg Asp
Asn Ala Lys Asn Ser Leu Tyr 65 70 75 80 Leu Gln Met Asn Ser Leu Arg
Ala Glu Asp Thr Ala Val Tyr Tyr Cys 85 90 95 Ala Arg Gly Leu Val
Arg Asp Gly Tyr Lys Trp Leu Tyr Phe Phe Asp 100 105 110 Tyr Trp Gly
Gln Gly Thr Leu Val Thr Val Ser Ser 115 120 <210> SEQ ID NO
157 <211> LENGTH: 8 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKA18 CDRH1 <400> SEQUENCE: 157 Gly Phe
Thr Phe Ser Ser Tyr Gly 1 5 <210> SEQ ID NO 158 <211>
LENGTH: 8 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA18
CDRH2 <400> SEQUENCE: 158 Ile Trp Tyr Asp Gly Ser Asn Lys 1 5
<210> SEQ ID NO 159 <211> LENGTH: 13 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA18 CDRH3 <400> SEQUENCE:
159 Ala Arg Asp Asp Ser Gly Tyr Ser Glu Pro Phe Asp Tyr 1 5 10
<210> SEQ ID NO 160 <211> LENGTH: 120 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA18 VH <400> SEQUENCE: 160
Gln Val Gln Leu Val Glu Ser Gly Gly Gly Val Val Gln Pro Gly Arg 1 5
10 15 Ser Leu Arg Leu Ser Cys Ala Ala Ser Gly Phe Thr Phe Ser Ser
Tyr 20 25 30 Gly Met His Trp Val Arg Gln Ala Pro Gly Lys Gly Leu
Glu Trp Val 35 40 45 Ala Val Ile Trp Tyr Asp Gly Ser Asn Lys Tyr
Tyr Ala Asp Ser Val 50 55 60 Lys Gly Arg Phe Thr Ile Thr Arg Asp
Asn Ser Lys Asn Thr Leu Tyr 65 70 75 80 Leu Gln Met Asn Ser Leu Arg
Pro Glu Asp Thr Ala Val Tyr Tyr Cys 85 90 95 Ala Arg Asp Asp Ser
Gly Tyr Ser Glu Pro Phe Asp Tyr Trp Gly Gln 100 105 110 Gly Thr Leu
Val Thr Val Ser Ser 115 120 <210> SEQ ID NO 161 <211>
LENGTH: 8 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA28
CDRH1 <400> SEQUENCE: 161 Gly Phe Thr Val Ser Arg Asn Tyr 1 5
<210> SEQ ID NO 162 <211> LENGTH: 7 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA28 CDRH2 <400> SEQUENCE:
162 Ile Tyr Ser Gly Gly Ser Thr 1 5 <210> SEQ ID NO 163
<211> LENGTH: 10 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA28 CDRH3 <400> SEQUENCE: 163 Ala Arg Trp Ile
Asn Asp Ala Phe Asp Ile 1 5 10 <210> SEQ ID NO 164
<211> LENGTH: 116 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA28 VH <400> SEQUENCE: 164 Glu Val Gln Leu Val
Glu Ser Gly Gly Gly Leu Ile Gln Pro Gly Gly 1 5 10 15 Ser Leu Arg
Leu Ser Cys Ala Ala Ser Gly Phe Thr Val Ser Arg Asn 20 25 30 Tyr
Met Ser Trp Val Arg Gln Ala Pro Gly Lys Gly Leu Glu Trp Val 35 40
45
Ser Val Ile Tyr Ser Gly Gly Ser Thr Tyr Tyr Ala Asp Ser Val Lys 50
55 60 Gly Arg Phe Thr Ile Ser Arg Asp Asn Ser Lys Asn Thr Leu Tyr
Leu 65 70 75 80 Gln Met Asn Ser Leu Arg Ala Glu Asp Thr Ala Val Tyr
Tyr Cys Ala 85 90 95 Arg Trp Ile Asn Asp Ala Phe Asp Ile Trp Gly
Gln Gly Thr Met Val 100 105 110 Thr Val Ser Ser 115 <210> SEQ
ID NO 165 <211> LENGTH: 8 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKA29 CDRH1 <400> SEQUENCE: 165 Gly Phe
Thr Phe Ser Arg Tyr Ser 1 5 <210> SEQ ID NO 166 <211>
LENGTH: 8 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA29
CDRH2 <400> SEQUENCE: 166 Ile Ser Pro Arg Ser Thr Thr Ile 1 5
<210> SEQ ID NO 167 <211> LENGTH: 15 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA29 CDRH3 <400> SEQUENCE:
167 Ala Arg Glu Asp Cys Thr Asn Gly Val Cys Tyr Arg Val Asp Tyr 1 5
10 15 <210> SEQ ID NO 168 <211> LENGTH: 122 <212>
TYPE: PRT <213> ORGANISM: Artificial Sequence <220>
FEATURE: <223> OTHER INFORMATION: ZKA29 VH <400>
SEQUENCE: 168 Glu Val Gln Leu Val Glu Ser Gly Gly Gly Leu Val Gln
Pro Gly Gly 1 5 10 15 Ser Leu Arg Leu Ser Cys Val Val Ser Gly Phe
Thr Phe Ser Arg Tyr 20 25 30 Ser Met Asn Trp Val Arg Gln Ala Pro
Gly Lys Gly Leu Glu Trp Val 35 40 45 Ser Tyr Ile Ser Pro Arg Ser
Thr Thr Ile Tyr Tyr Ala Asp Ser Val 50 55 60 Glu Gly Arg Phe Thr
Val Ser Arg Asp Asn Ala Lys Asn Ser Leu Tyr 65 70 75 80 Leu Gln Leu
Asn Ser Leu Arg Ala Glu Asp Thr Ala Val Tyr Tyr Cys 85 90 95 Ala
Arg Glu Asp Cys Thr Asn Gly Val Cys Tyr Arg Val Asp Tyr Trp 100 105
110 Gly Gln Gly Thr Leu Val Thr Val Ser Ser 115 120 <210> SEQ
ID NO 169 <211> LENGTH: 8 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKA33 CDRH1 <400> SEQUENCE: 169 Gly Phe
Thr Phe Ser Arg Asn Trp 1 5 <210> SEQ ID NO 170 <211>
LENGTH: 8 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA33
CDRH2 <400> SEQUENCE: 170 Ile Lys Glu Asp Gly Asn Glu Lys 1 5
<210> SEQ ID NO 171 <211> LENGTH: 15 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA33 CDRH3 <400> SEQUENCE:
171 Ala Arg Pro Phe His Gln Gly Gly Tyr Ala Tyr Gly Leu Ala Tyr 1 5
10 15 <210> SEQ ID NO 172 <211> LENGTH: 122 <212>
TYPE: PRT <213> ORGANISM: Artificial Sequence <220>
FEATURE: <223> OTHER INFORMATION: ZKA33 VH <400>
SEQUENCE: 172 Glu Val Gln Leu Val Glu Ser Gly Gly Gly Leu Val Gln
Pro Gly Gly 1 5 10 15 Ser Leu Arg Leu Ser Cys Ala Ala Ser Gly Phe
Thr Phe Ser Arg Asn 20 25 30 Trp Met Thr Trp Val Arg Gln Ala Pro
Gly Lys Gly Leu Glu Trp Val 35 40 45 Ala Asn Ile Lys Glu Asp Gly
Asn Glu Lys Tyr Tyr Val Asp Ser Val 50 55 60 Lys Gly Arg Phe Thr
Ile Ser Arg Asp Asn Ala Lys Asn Ser Leu Tyr 65 70 75 80 Leu Gln Met
Asn Ser Leu Arg Ala Glu Asp Thr Ala Val Tyr Tyr Cys 85 90 95 Ala
Arg Pro Phe His Gln Gly Gly Tyr Ala Tyr Gly Leu Ala Tyr Trp 100 105
110 Gly Gln Gly Thr Leu Val Thr Val Ser Ser 115 120 <210> SEQ
ID NO 173 <211> LENGTH: 8 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKA39 CDRH1 <400> SEQUENCE: 173 Gly Phe
Thr Phe Ser Thr Tyr Ser 1 5 <210> SEQ ID NO 174 <211>
LENGTH: 8 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA39
CDRH2 <400> SEQUENCE: 174 Ile Ser Pro Ser Ser Ser Thr Ile 1 5
<210> SEQ ID NO 175 <211> LENGTH: 15 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA39 CDRH3 <400> SEQUENCE:
175 Ala Arg Glu Tyr Cys Ser Gly Gly Ser Cys Tyr Leu Leu Asp Tyr 1 5
10 15 <210> SEQ ID NO 176 <211> LENGTH: 122 <212>
TYPE: PRT <213> ORGANISM: Artificial Sequence <220>
FEATURE: <223> OTHER INFORMATION: ZKA39 VH <400>
SEQUENCE: 176 Glu Val Gln Leu Val Glu Ser Gly Gly Gly Leu Val Gln
Pro Gly Gly 1 5 10 15 Ser Leu Arg Leu Ser Cys Ala Ala Ser Gly Phe
Thr Phe Ser Thr Tyr 20 25 30 Ser Met Asn Trp Val Arg Gln Ala Pro
Gly Lys Gly Leu Glu Trp Val 35 40 45 Ser Tyr Ile Ser Pro Ser Ser
Ser Thr Ile Tyr Tyr Pro Asp Ser Leu 50 55 60 Lys Gly Arg Phe Thr
Ile Ser Arg Asp Asn Ala Lys Asn Ser Leu Tyr 65 70 75 80 Leu Gln Met
Asp Ser Leu Arg Ala Glu Asp Thr Ala Gln Tyr Tyr Cys 85 90 95 Ala
Arg Glu Tyr Cys Ser Gly Gly Ser Cys Tyr Leu Leu Asp Tyr Trp 100 105
110 Gly Gln Gly Thr Leu Val Thr Val Ser Ser 115 120 <210> SEQ
ID NO 177 <211> LENGTH: 8 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKA43 CDRH1 <400> SEQUENCE: 177 Gly Gly
Ser Ile Thr Ser Tyr Tyr 1 5 <210> SEQ ID NO 178 <211>
LENGTH: 7
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA43 CDRH2
<400> SEQUENCE: 178 Ser His Tyr Ser Gly Ser Thr 1 5
<210> SEQ ID NO 179 <211> LENGTH: 13 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA43 CDRH3 <400> SEQUENCE:
179 Ala Arg Gly Ile Tyr Ser Gly Lys Asn Trp Phe Asp Pro 1 5 10
<210> SEQ ID NO 180 <211> LENGTH: 119 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA43 VH <400> SEQUENCE: 180
Gln Val Gln Leu Gln Glu Ser Gly Pro Gly Leu Val Lys Pro Ser Glu 1 5
10 15 Thr Leu Ser Leu Thr Cys Thr Val Tyr Gly Gly Ser Ile Thr Ser
Tyr 20 25 30 Tyr Trp Thr Trp Ile Arg Gln Pro Pro Gly Lys Gly Leu
Glu Trp Ile 35 40 45 Gly Tyr Ser His Tyr Ser Gly Ser Thr Asn Tyr
Asn Pro Ser Leu Lys 50 55 60 Ser Arg Val Thr Ile Ser Ile Asp Thr
Ser Lys Ser Gln Phe Ser Leu 65 70 75 80 Asn Leu Asn Ser Val Thr Ala
Ala Asp Thr Ala Val Tyr Tyr Cys Ala 85 90 95 Arg Gly Ile Tyr Ser
Gly Lys Asn Trp Phe Asp Pro Trp Gly Gln Gly 100 105 110 Thr Leu Val
Thr Val Ser Ser 115 <210> SEQ ID NO 181 <211> LENGTH: 8
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA44 CDRH1
<400> SEQUENCE: 181 Gly Phe Thr Val Ser Thr Ser Tyr 1 5
<210> SEQ ID NO 182 <211> LENGTH: 7 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA44 CDRH2 <400> SEQUENCE:
182 Ile Tyr Ser Ser Gly Ser Thr 1 5 <210> SEQ ID NO 183
<211> LENGTH: 10 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA44 CDRH3 <400> SEQUENCE: 183 Ala Arg Val Ser
Leu Gly Gly Leu Asp Pro 1 5 10 <210> SEQ ID NO 184
<211> LENGTH: 116 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA44 VH <400> SEQUENCE: 184 Glu Val Gln Leu Val
Glu Ser Gly Gly Gly Leu Ile Gln Pro Gly Gly 1 5 10 15 Ser Leu Arg
Leu Ser Cys Val Ala Ser Gly Phe Thr Val Ser Thr Ser 20 25 30 Tyr
Met Asn Trp Val Arg Gln Ala Pro Gly Lys Gly Leu Glu Trp Val 35 40
45 Ser Val Ile Tyr Ser Ser Gly Ser Thr Tyr Tyr Ala Asp Ser Val Lys
50 55 60 Gly Arg Phe Thr Ile Ser Arg Asn Thr Ser Lys Asn Thr Leu
Tyr Leu 65 70 75 80 Gln Met Asn Ser Leu Arg Ala Glu Asp Thr Ala Val
Tyr Tyr Cys Ala 85 90 95 Arg Val Ser Leu Gly Gly Leu Asp Pro Trp
Gly Gln Gly Thr Pro Val 100 105 110 Thr Val Ser Ser 115 <210>
SEQ ID NO 185 <211> LENGTH: 10 <212> TYPE: PRT
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA46 CDRH1 <400> SEQUENCE:
185 Gly Phe Ser Leu Ser Asn Gly Arg Met Gly 1 5 10 <210> SEQ
ID NO 186 <211> LENGTH: 7 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKA46 CDRH2 <400> SEQUENCE: 186 Ile Phe
Ser Asn Asp Glu Lys 1 5 <210> SEQ ID NO 187 <211>
LENGTH: 13 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA46
CDRH3 <400> SEQUENCE: 187 Ala Arg Val Glu Phe Arg Ala Gly Asn
Tyr Leu Asp Ser 1 5 10 <210> SEQ ID NO 188 <211>
LENGTH: 121 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA46
VH <400> SEQUENCE: 188 Gln Val Thr Leu Lys Glu Ser Gly Pro
Val Leu Val Lys Pro Thr Glu 1 5 10 15 Thr Leu Thr Leu Thr Cys Thr
Val Ser Gly Phe Ser Leu Ser Asn Gly 20 25 30 Arg Met Gly Val Ser
Trp Ile Arg Gln Pro Pro Gly Lys Ala Leu Glu 35 40 45 Trp Leu Ala
His Ile Phe Ser Asn Asp Glu Lys Tyr Tyr Ser Thr Ser 50 55 60 Leu
Lys Asn Arg Leu Thr Ile Ser Lys Asp Thr Ser Lys Ser Gln Val 65 70
75 80 Val Leu Thr Met Thr Asn Met Asp Pro Val Asp Thr Ala Thr Tyr
Tyr 85 90 95 Cys Ala Arg Val Glu Phe Arg Ala Gly Asn Tyr Leu Asp
Ser Trp Gly 100 105 110 Gln Gly Thr Leu Val Thr Val Ser Ser 115 120
<210> SEQ ID NO 189 <211> LENGTH: 8 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA50 CDRH1 <400> SEQUENCE:
189 Gly Tyr Thr Phe Thr Asn Ser Trp 1 5 <210> SEQ ID NO 190
<211> LENGTH: 8 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA50 CDRH2 <400> SEQUENCE: 190 Ile Tyr Pro Gly
Asp Ser Asp Thr 1 5 <210> SEQ ID NO 191 <211> LENGTH: 8
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA50 CDRH3
<400> SEQUENCE: 191 Ala Arg Gln Pro Phe Phe Asp Tyr 1 5
<210> SEQ ID NO 192 <211> LENGTH: 115 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA50 VH <400> SEQUENCE:
192
Glu Val Gln Leu Val Gln Ser Gly Ala Gln Val Lys Lys Pro Gly Glu 1 5
10 15 Ser Leu Lys Ile Ser Cys Lys Ala Ser Gly Tyr Thr Phe Thr Asn
Ser 20 25 30 Trp Ile Gly Trp Val Arg Gln Met Pro Gly Lys Gly Leu
Glu Trp Met 35 40 45 Gly Ile Ile Tyr Pro Gly Asp Ser Asp Thr Arg
Tyr Ser Pro Ser Phe 50 55 60 Gln Gly Gln Val Thr Ile Ser Ala Asp
Lys Ser Ile Ser Thr Ala Tyr 65 70 75 80 Leu Gln Trp Ser Ser Leu Lys
Ala Ser Asp Thr Ala Met Tyr Tyr Cys 85 90 95 Ala Arg Gln Pro Phe
Phe Asp Tyr Trp Gly Gln Gly Thr Leu Val Thr 100 105 110 Val Ser Ser
115 <210> SEQ ID NO 193 <211> LENGTH: 8 <212>
TYPE: PRT <213> ORGANISM: Artificial Sequence <220>
FEATURE: <223> OTHER INFORMATION: ZKA54 CDRH1 <400>
SEQUENCE: 193 Gly Tyr Thr Phe Thr Gly Tyr Tyr 1 5 <210> SEQ
ID NO 194 <211> LENGTH: 8 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKA54 CDRH2 <400> SEQUENCE: 194 Ile Asn
Ala Asn Ser Gly Gly Thr 1 5 <210> SEQ ID NO 195 <211>
LENGTH: 18 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA54
CDRH3 <400> SEQUENCE: 195 Ala His Ser Asp Ile Val Val Val Pro
Ser Asp Asp Tyr Tyr Ala Leu 1 5 10 15 Asp Val <210> SEQ ID NO
196 <211> LENGTH: 125 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKA54 VH <400> SEQUENCE: 196 Gln Val Gln
Leu Val Gln Ser Gly Ala Glu Val Lys Lys Pro Gly Ala 1 5 10 15 Ser
Val Lys Val Ser Cys Lys Thr Ser Gly Tyr Thr Phe Thr Gly Tyr 20 25
30 Tyr Met His Trp Val Arg Gln Ala Pro Gly Gln Gly Leu Glu Trp Met
35 40 45 Gly Trp Ile Asn Ala Asn Ser Gly Gly Thr Asn Phe Ala Gln
Arg Phe 50 55 60 Gln Gly Arg Val Thr Met Thr Trp Asp Thr Ser Ile
Ser Thr Ala Tyr 65 70 75 80 Met Glu Leu Ser Arg Leu Arg Ser Asp Asp
Thr Ala Val Tyr Tyr Cys 85 90 95 Ala His Ser Asp Ile Val Val Val
Pro Ser Asp Asp Tyr Tyr Ala Leu 100 105 110 Asp Val Trp Gly Gln Gly
Thr Thr Val Thr Val Ser Ser 115 120 125 <210> SEQ ID NO 197
<211> LENGTH: 8 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKB18 CDRH1 <400> SEQUENCE: 197 Gly Tyr Ser Phe
Thr Ser Tyr Trp 1 5 <210> SEQ ID NO 198 <211> LENGTH: 8
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKB18 CDRH2
<400> SEQUENCE: 198 Ile Tyr Pro Gly Asp Ser Asp Thr 1 5
<210> SEQ ID NO 199 <211> LENGTH: 8 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKB18 CDRH3 <400> SEQUENCE:
199 Ala Arg Gln Thr Pro Gly Asp Tyr 1 5 <210> SEQ ID NO 200
<211> LENGTH: 115 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKB18 VH <400> SEQUENCE: 200 Glu Val Gln Leu Val
Gln Ser Gly Ala Glu Val Lys Lys Pro Gly Glu 1 5 10 15 Ser Leu Lys
Ile Ser Cys Lys Thr Phe Gly Tyr Ser Phe Thr Ser Tyr 20 25 30 Trp
Ile Gly Trp Val Arg Gln Met Pro Gly Lys Gly Leu Glu Trp Met 35 40
45 Gly Met Ile Tyr Pro Gly Asp Ser Asp Thr Arg Tyr Ser Pro Ser Phe
50 55 60 Gln Gly Gln Val Thr Ile Ser Ala Asp Met Ser Ile Ser Thr
Ala Tyr 65 70 75 80 Leu Gln Trp Ser Ser Leu Lys Ala Ser Asp Thr Ala
Met Tyr Tyr Cys 85 90 95 Ala Arg Gln Thr Pro Gly Asp Tyr Trp Gly
Gln Gly Thr Leu Val Thr 100 105 110 Val Ser Ser 115 <210> SEQ
ID NO 201 <211> LENGTH: 8 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKB20 CDRH1 <400> SEQUENCE: 201 Gly Tyr
Phe Phe Thr Arg Tyr Val 1 5 <210> SEQ ID NO 202 <211>
LENGTH: 8 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKB20
CDRH2 <400> SEQUENCE: 202 Ile Asn Thr Asp Asn Gly Ser Thr 1 5
<210> SEQ ID NO 203 <211> LENGTH: 15 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKB20 CDRH3 <400> SEQUENCE:
203 Ala Arg Gly Thr Gly Arg Asp Gly Tyr Asn Ser Phe Phe Ala Asn 1 5
10 15 <210> SEQ ID NO 204 <211> LENGTH: 122 <212>
TYPE: PRT <213> ORGANISM: Artificial Sequence <220>
FEATURE: <223> OTHER INFORMATION: ZKB20 VH <400>
SEQUENCE: 204 Gln Val Gln Leu Val Gln Ser Gly Ala Glu Val Lys Lys
Pro Gly Ala 1 5 10 15 Ser Val Arg Val Ser Cys Lys Ala Ser Gly Tyr
Phe Phe Thr Arg Tyr 20 25 30 Val Ile Leu Trp Val Arg Gln Ala Pro
Gly Gln Arg Pro Glu Trp Met 35 40 45 Gly Trp Ile Asn Thr Asp Asn
Gly Ser Thr Arg Tyr Ser Gln Lys Phe 50 55 60 Gln Gly Arg Val Thr
Ile Thr Lys Asp Thr Ser Ala Thr Thr Ala Tyr 65 70 75 80 Met Asp Leu
Ser Ser Leu Lys Ser Asp Asp Thr Ala Val Tyr Tyr Cys 85 90 95 Ala
Arg Gly Thr Gly Arg Asp Gly Tyr Asn Ser Phe Phe Ala Asn Trp 100 105
110 Gly Gln Gly Thr Leu Val Thr Val Ser Pro 115 120 <210> SEQ
ID NO 205 <211> LENGTH: 8 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKB21 CDRH1
<400> SEQUENCE: 205 Gly Tyr Thr Phe Thr Gly Tyr Ser 1 5
<210> SEQ ID NO 206 <211> LENGTH: 8 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKB21 CDRH2 <400> SEQUENCE:
206 Ile Asp Thr Asn Ser Gly Asp Thr 1 5 <210> SEQ ID NO 207
<211> LENGTH: 11 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKB21 CDRH3 <400> SEQUENCE: 207 Ala Arg Asp Arg
Glu Arg His Pro Phe Ser Tyr 1 5 10 <210> SEQ ID NO 208
<211> LENGTH: 118 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKB21 VH <400> SEQUENCE: 208 Gln Val Gln Leu Val
Gln Ser Gly Ala Glu Val Lys Lys Pro Gly Ala 1 5 10 15 Ser Val Lys
Val Ser Cys Lys Ala Ser Gly Tyr Thr Phe Thr Gly Tyr 20 25 30 Ser
Ile His Trp Val Arg Gln Ala Pro Gly Gln Gly Leu Ala Trp Met 35 40
45 Gly Arg Ile Asp Thr Asn Ser Gly Asp Thr Asn Tyr Ala Glu Arg Phe
50 55 60 Gln Gly Arg Val Thr Met Thr Arg Asp Thr Ser Ile Ser Thr
Ala Tyr 65 70 75 80 Met Glu Val Arg Arg Leu Arg Ser Asp Asp Thr Ala
Val Tyr Tyr Cys 85 90 95 Ala Arg Asp Arg Glu Arg His Pro Phe Ser
Tyr Trp Gly Gln Gly Thr 100 105 110 Leu Val Thr Val Ser Ser 115
<210> SEQ ID NO 209 <211> LENGTH: 10 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKB23 CDRH1 <400> SEQUENCE:
209 Gly Gly Ser Ile Ser Ser Gly Asp Tyr Ser 1 5 10 <210> SEQ
ID NO 210 <211> LENGTH: 7 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKB23 CDRH2 <400> SEQUENCE: 210 Ile Thr
His Ser Gly Thr Thr 1 5 <210> SEQ ID NO 211 <211>
LENGTH: 9 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKB23
CDRH3 <400> SEQUENCE: 211 Ala Arg His Phe Gly Trp Phe Asp Pro
1 5 <210> SEQ ID NO 212 <211> LENGTH: 117 <212>
TYPE: PRT <213> ORGANISM: Artificial Sequence <220>
FEATURE: <223> OTHER INFORMATION: ZKB23 VH <400>
SEQUENCE: 212 Gln Leu Gln Leu Gln Glu Ser Gly Ser Gly Leu Val Lys
Pro Ser Gln 1 5 10 15 Thr Leu Ser Leu Thr Cys Ala Val Ser Gly Gly
Ser Ile Ser Ser Gly 20 25 30 Asp Tyr Ser Trp Ser Trp Ile Arg Gln
Pro Pro Gly Lys Gly Leu Glu 35 40 45 Trp Ile Gly Tyr Ile Thr His
Ser Gly Thr Thr Tyr Phe Asn Pro Ser 50 55 60 Leu Lys Ser Arg Val
Thr Ile Ser Val Asp Arg Ser Arg Asn Gln Phe 65 70 75 80 Ser Leu Lys
Val Thr Ser Val Thr Ala Ala Asp Thr Ala Val Tyr Tyr 85 90 95 Cys
Ala Arg His Phe Gly Trp Phe Asp Pro Trp Gly Gln Gly Thr Leu 100 105
110 Val Thr Val Ser Ser 115 <210> SEQ ID NO 213 <211>
LENGTH: 10 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKC29
CDRH1 <400> SEQUENCE: 213 Gly Gly Ser Ile Ser Ser Gly Glu Tyr
Phe 1 5 10 <210> SEQ ID NO 214 <211> LENGTH: 7
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKC29 CDRH2
<400> SEQUENCE: 214 Ile His Asn Arg Gly Asn Thr 1 5
<210> SEQ ID NO 215 <211> LENGTH: 22 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKC29 CDRH3 <400> SEQUENCE:
215 Ala Arg Gly Gly Gly Asp Leu Val Val Val Pro Asp Ser Ile Trp Asp
1 5 10 15 Tyr Tyr Gly Met Asp Val 20 <210> SEQ ID NO 216
<211> LENGTH: 130 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKC29 VH <400> SEQUENCE: 216 Gln Val Gln Leu Gln
Glu Ser Gly Pro Gly Leu Val Arg Pro Ser Gln 1 5 10 15 Thr Leu Ser
Leu Thr Cys Thr Val Ser Gly Gly Ser Ile Ser Ser Gly 20 25 30 Glu
Tyr Phe Trp Thr Trp Ile Arg Gln His Pro Lys Lys Gly Leu Glu 35 40
45 Trp Ile Gly Tyr Ile His Asn Arg Gly Asn Thr Tyr Tyr Asn Pro Ser
50 55 60 Leu Lys Ser Arg Leu Ser Ile Ser Leu Asp Thr Ser Lys Asn
His Leu 65 70 75 80 Ser Leu Arg Leu Ser Ser Val Thr Ala Ala Asp Thr
Ala Val Tyr Tyr 85 90 95 Cys Ala Arg Gly Gly Gly Asp Leu Val Val
Val Pro Asp Ser Ile Trp 100 105 110 Asp Tyr Tyr Gly Met Asp Val Trp
Gly Gln Gly Thr Thr Val Thr Val 115 120 125 Ser Ser 130 <210>
SEQ ID NO 217 <211> LENGTH: 10 <212> TYPE: PRT
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKC31 CDRH1 <400> SEQUENCE:
217 Gly Gly Ser Ile Ser Ser Gly Gly Tyr His 1 5 10 <210> SEQ
ID NO 218 <211> LENGTH: 7 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKC31 CDRH2 <400> SEQUENCE: 218 Ile Tyr
Tyr Ser Gly Ser Thr 1 5 <210> SEQ ID NO 219 <211>
LENGTH: 19 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKC31 CDRH3
<400> SEQUENCE: 219 Ala Arg Asp Arg Ser Glu Pro Gly Glu Tyr
His Tyr Tyr Tyr Tyr Ala 1 5 10 15 Met Asp Val <210> SEQ ID NO
220 <211> LENGTH: 127 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKC31 VH <400> SEQUENCE: 220 Gln Val Gln
Leu Gln Glu Ser Gly Pro Gly Leu Val Lys Pro Ser Gln 1 5 10 15 Thr
Leu Ser Leu Thr Cys Thr Val Ser Gly Gly Ser Ile Ser Ser Gly 20 25
30 Gly Tyr His Trp Ser Trp Ile Arg Gln His Pro Gly Lys Gly Leu Glu
35 40 45 Trp Ile Gly Tyr Ile Tyr Tyr Ser Gly Ser Thr Tyr Tyr Asn
Pro Ser 50 55 60 Leu Lys Arg Arg Val Thr Ile Ser Val Asp Thr Ser
Lys Asn Gln Phe 65 70 75 80 Ser Leu Lys Leu Ser Ser Val Ser Ala Ala
Asp Thr Ala Val Tyr Tyr 85 90 95 Cys Ala Arg Asp Arg Ser Glu Pro
Gly Glu Tyr His Tyr Tyr Tyr Tyr 100 105 110 Ala Met Asp Val Trp Gly
Gln Gly Thr Thr Val Thr Val Ser Ser 115 120 125 <210> SEQ ID
NO 221 <211> LENGTH: 8 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKC32 CDRH1 <400> SEQUENCE: 221 Gly Phe
Thr Val Ser Ser Asn Tyr 1 5 <210> SEQ ID NO 222 <211>
LENGTH: 7 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKC32
CDRH2 <400> SEQUENCE: 222 Ile Tyr Ser Ser Gly Ser Thr 1 5
<210> SEQ ID NO 223 <211> LENGTH: 11 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKC32 CDRH3 <400> SEQUENCE:
223 Ala Arg Gly Lys Lys Gly Asn Ala Phe Asp Ile 1 5 10 <210>
SEQ ID NO 224 <211> LENGTH: 117 <212> TYPE: PRT
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKC32 VH <400> SEQUENCE: 224
Glu Val Gln Leu Val Glu Ser Gly Gly Asp Leu Ile Gln Pro Gly Gly 1 5
10 15 Ser Leu Arg Leu Ser Cys Ala Ala Ser Gly Phe Thr Val Ser Ser
Asn 20 25 30 Tyr Met Ser Trp Val Arg Gln Ala Pro Gly Lys Gly Leu
Glu Trp Val 35 40 45 Ser Val Ile Tyr Ser Ser Gly Ser Thr Tyr Tyr
Ala Asp Ser Val Lys 50 55 60 Gly Arg Phe Thr Ile Ser Arg Asp Asn
Ser Lys Asn Thr Leu Tyr Leu 65 70 75 80 Gln Met Asn Ser Leu Arg Ala
Gly Asp Thr Ala Val Tyr Tyr Cys Ala 85 90 95 Arg Gly Lys Lys Gly
Asn Ala Phe Asp Ile Trp Gly Gln Gly Thr Val 100 105 110 Val Thr Val
Ser Ser 115 <210> SEQ ID NO 225 <211> LENGTH: 10
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKC33, CDRH1
<400> SEQUENCE: 225 Gly Asp Ser Ile Ser Ser Arg Thr Phe Ser 1
5 10 <210> SEQ ID NO 226 <211> LENGTH: 7 <212>
TYPE: PRT <213> ORGANISM: Artificial Sequence <220>
FEATURE: <223> OTHER INFORMATION: ZKC33 CDRH2 <400>
SEQUENCE: 226 Ile Tyr Tyr Ser Gly Ser Thr 1 5 <210> SEQ ID NO
227 <211> LENGTH: 18 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKC33 CDRH3 <400> SEQUENCE: 227 Ala Arg
Arg Asn Ala Glu Phe Phe Ser Phe Trp Ser Tyr Tyr Gly Met 1 5 10 15
Asp Val <210> SEQ ID NO 228 <211> LENGTH: 126
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKC33 VH
<400> SEQUENCE: 228 Gln Val Gln Leu Gln Glu Ser Gly Pro Gly
Leu Val Lys Pro Ser Gln 1 5 10 15 Thr Leu Ser Leu Thr Cys Thr Val
Ser Gly Asp Ser Ile Ser Ser Arg 20 25 30 Thr Phe Ser Trp Ser Trp
Ile Arg Gln Pro Pro Gly Lys Gly Leu Glu 35 40 45 Trp Val Gly His
Ile Tyr Tyr Ser Gly Ser Thr Asp Tyr Asn Pro Ser 50 55 60 Leu Lys
Ser Arg Ile Ser Ile Ser Ile Asp Thr Ser Lys Asn Gln Phe 65 70 75 80
Ser Leu Lys Leu Ser Ser Val Thr Ala Ala Asp Thr Ala Val Tyr Tyr 85
90 95 Cys Ala Arg Arg Asn Ala Glu Phe Phe Ser Phe Trp Ser Tyr Tyr
Gly 100 105 110 Met Asp Val Trp Gly His Gly Thr Ala Val Ile Val Ser
Ser 115 120 125 <210> SEQ ID NO 229 <211> LENGTH: 10
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKC34 CDRH1
<400> SEQUENCE: 229 Gly Gly Ser Ile Asn Ser Gly Gly Tyr Tyr 1
5 10 <210> SEQ ID NO 230 <211> LENGTH: 7 <212>
TYPE: PRT <213> ORGANISM: Artificial Sequence <220>
FEATURE: <223> OTHER INFORMATION: ZKC34 CDRH2 <400>
SEQUENCE: 230 Ile Leu His Ser Gly Asn Thr 1 5 <210> SEQ ID NO
231 <211> LENGTH: 15 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKC34 CDRH3 <400> SEQUENCE: 231 Ala Arg
Ala Gly Asp Tyr Tyr Ser Gly Tyr Val Pro Pro Glu Tyr 1 5 10 15
<210> SEQ ID NO 232 <211> LENGTH: 123 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKC34 VH <400> SEQUENCE: 232
Gln Val Gln Leu Gln Glu Ser Gly Pro Gly Leu Val Lys Pro Ser Gln 1 5
10 15 Thr Leu Ser Leu Thr Cys Ala Val Ser Gly Gly Ser Ile Asn Ser
Gly 20 25 30 Gly Tyr Tyr Trp Ser Trp Val Arg Gln His Pro Gly Lys
Gly Leu Glu 35 40 45 Trp Ile Gly Tyr Ile Leu His Ser Gly Asn Thr
Asn Tyr Asn Pro Ser
50 55 60 Leu Lys Ser Arg Val Asn Ile Phe Val Asp Thr Ser Glu Asn
Gln Phe 65 70 75 80 Ser Leu Lys Leu Arg Ser Val Thr Ala Ala Asp Thr
Ala Ile Tyr Phe 85 90 95 Cys Ala Arg Ala Gly Asp Tyr Tyr Ser Gly
Tyr Val Pro Pro Glu Tyr 100 105 110 Trp Gly Pro Gly Thr Leu Val Thr
Val Ser Ser 115 120 <210> SEQ ID NO 233 <211> LENGTH: 8
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKD25 CDRH1
<400> SEQUENCE: 233 Gly Phe Thr Val Ser Ser Asn Tyr 1 5
<210> SEQ ID NO 234 <211> LENGTH: 7 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKD25 CDRH2 <400> SEQUENCE:
234 Ile Tyr Ser Gly Gly Ser Thr 1 5 <210> SEQ ID NO 235
<211> LENGTH: 11 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKD25 CDRH3 <400> SEQUENCE: 235 Ala Arg Phe Gly
Gly Asn Pro Ser Phe Asp Tyr 1 5 10 <210> SEQ ID NO 236
<211> LENGTH: 117 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKD25 VH <400> SEQUENCE: 236 Glu Val Gln Leu Val
Glu Ser Gly Gly Gly Leu Val Gln Pro Gly Gly 1 5 10 15 Ser Leu Arg
Leu Ser Cys Ala Ala Ser Gly Phe Thr Val Ser Ser Asn 20 25 30 Tyr
Met Ser Trp Val Arg Gln Ala Pro Gly Lys Gly Leu Glu Trp Val 35 40
45 Ser Val Ile Tyr Ser Gly Gly Ser Thr Tyr Tyr Ala Asn Ser Val Lys
50 55 60 Gly Arg Phe Thr Ile Ser Arg Asp Lys Ser Lys Asn Thr Leu
Tyr Leu 65 70 75 80 Gln Met Asn Asn Leu Arg Ala Glu Asp Thr Ala Val
Tyr Phe Cys Ala 85 90 95 Arg Phe Gly Gly Asn Pro Ser Phe Asp Tyr
Trp Gly Gln Gly Thr Leu 100 105 110 Val Thr Val Ser Ser 115
<210> SEQ ID NO 237 <211> LENGTH: 8 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA3 CDRH1 <400> SEQUENCE: 237
Gly Phe Ile Phe Ser Asn Tyr Ala 1 5 <210> SEQ ID NO 238
<211> LENGTH: 8 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA3 CDRH2 <400> SEQUENCE: 238 Ile Gly Gly Lys
Gly Asp Ser Ile 1 5 <210> SEQ ID NO 239 <211> LENGTH:
16 <212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA3 CDRH3
<400> SEQUENCE: 239 Val Lys Asp Leu Ala Val Leu Glu Ser Asp
Arg Leu Glu Val Asp Gln 1 5 10 15 <210> SEQ ID NO 240
<211> LENGTH: 123 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA3 VH <400> SEQUENCE: 240 Glu Val Gln Leu Ala
Glu Ser Gly Gly Gly Leu Val Gln Pro Gly Gly 1 5 10 15 Ser Leu Arg
Leu Ser Cys Ser Gly Ser Gly Phe Ile Phe Ser Asn Tyr 20 25 30 Ala
Met Val Trp Ala Arg Gln Ala Pro Gly Lys Gly Leu Glu Tyr Val 35 40
45 Ser Gly Ile Gly Gly Lys Gly Asp Ser Ile Tyr His Ile Asp Ser Val
50 55 60 Lys Gly Arg Phe Thr Ile Ser Arg Asp Asn Ser Lys Arg Thr
Val Tyr 65 70 75 80 Leu Gln Met Ser Arg Leu Arg Thr Glu Asp Thr Ala
Val Tyr Tyr Cys 85 90 95 Val Lys Asp Leu Ala Val Leu Glu Ser Asp
Arg Leu Glu Val Asp Gln 100 105 110 Trp Gly Gln Gly Thr Leu Val Ile
Val Ser Ala 115 120 <210> SEQ ID NO 241 <211> LENGTH: 8
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA4 CDRH1
<400> SEQUENCE: 241 Gly Phe Thr Phe Ser Ser Tyr Val 1 5
<210> SEQ ID NO 242 <211> LENGTH: 8 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA4 CDRH2 <400> SEQUENCE: 242
Thr Ser Tyr Asp Gly Ser Asn Lys 1 5 <210> SEQ ID NO 243
<211> LENGTH: 20 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA4 CDRH3 <400> SEQUENCE: 243 Ala Arg Gly Pro
Val Pro Tyr Trp Ser Gly Glu Ser Tyr Ser Gly Ala 1 5 10 15 Tyr Phe
Asp Phe 20 <210> SEQ ID NO 244 <211> LENGTH: 127
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA4 VH
<400> SEQUENCE: 244 Gln Val Gln Leu Val Glu Ser Gly Gly Gly
Val Val Gln Pro Gly Arg 1 5 10 15 Ser Leu Arg Leu Ser Cys Ala Ala
Ser Gly Phe Thr Phe Ser Ser Tyr 20 25 30 Val Met His Trp Val Arg
Gln Ala Pro Gly Lys Gly Leu Glu Trp Val 35 40 45 Thr Val Thr Ser
Tyr Asp Gly Ser Asn Lys Tyr Tyr Ala Asp Ser Val 50 55 60 Lys Gly
Arg Phe Thr Ile Ser Arg Asp Asn Ala Lys Asn Thr Leu Tyr 65 70 75 80
Leu Gln Met Asn Ser Leu Arg Gly Glu Asp Thr Ala Ile Tyr Tyr Cys 85
90 95 Ala Arg Gly Pro Val Pro Tyr Trp Ser Gly Glu Ser Tyr Ser Gly
Ala 100 105 110 Tyr Phe Asp Phe Trp Gly Gln Gly Ile Leu Val Thr Val
Ser Ser 115 120 125 <210> SEQ ID NO 245 <211> LENGTH: 8
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA5 CDRH1
<400> SEQUENCE: 245 Gly Phe Thr Phe Ser Asn Tyr Tyr 1 5
<210> SEQ ID NO 246
<211> LENGTH: 8 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA5 CDRH2 <400> SEQUENCE: 246 Met Ser Ser Ser
Glu Thr Ile Lys 1 5 <210> SEQ ID NO 247 <211> LENGTH:
19 <212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA5 CDRH3
<400> SEQUENCE: 247 Ala Arg Ser Gly Ile Glu Thr Val Ala Gly
Ser Ile Asp Tyr Tyr Gly 1 5 10 15 Met Asp Val <210> SEQ ID NO
248 <211> LENGTH: 126 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKA5 VH <400> SEQUENCE: 248 Gln Val Gln
Leu Val Glu Ser Gly Gly Gly Leu Val Lys Pro Gly Gly 1 5 10 15 Ser
Leu Arg Leu Ser Cys Ala Gly Ser Gly Phe Thr Phe Ser Asn Tyr 20 25
30 Tyr Met Thr Trp Ile Arg Gln Ala Pro Gly Lys Gly Leu Glu Leu Val
35 40 45 Ser Tyr Met Ser Ser Ser Glu Thr Ile Lys Tyr Tyr Ala Asp
Ser Val 50 55 60 Lys Gly Arg Phe Thr Ile Ser Arg Asp Asn Ala Lys
Asn Ser Leu Tyr 65 70 75 80 Leu Gln Met Asn Ser Leu Arg Ala Asp Asp
Thr Ala Arg Tyr Tyr Cys 85 90 95 Ala Arg Ser Gly Ile Glu Thr Val
Ala Gly Ser Ile Asp Tyr Tyr Gly 100 105 110 Met Asp Val Trp Gly His
Gly Thr Pro Val Thr Val Ser Ser 115 120 125 <210> SEQ ID NO
249 <211> LENGTH: 8 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKA6 CDRH1 <400> SEQUENCE: 249 Asp Phe Thr
Val Ser Asn Tyr Ala 1 5 <210> SEQ ID NO 250 <211>
LENGTH: 8 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA6
CDRH2 <400> SEQUENCE: 250 Val Ser Tyr Asp Gly Ser Asn Lys 1 5
<210> SEQ ID NO 251 <211> LENGTH: 19 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA6 CDRH3 <400> SEQUENCE: 251
Ala Thr Gly Val Thr Met Phe Gln Gly Ala Gln Thr Asn Ala Glu Tyr 1 5
10 15 Leu His Tyr <210> SEQ ID NO 252 <211> LENGTH: 126
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA6 VH
<400> SEQUENCE: 252 Gln Val His Leu Val Glu Ser Gly Gly Gly
Val Val Gln Pro Gly Arg 1 5 10 15 Ser Leu Arg Leu Ser Cys Glu Ala
Ser Asp Phe Thr Val Ser Asn Tyr 20 25 30 Ala Met His Trp Val Arg
Gln Ala Pro Gly Lys Gly Leu Glu Trp Val 35 40 45 Ala Val Val Ser
Tyr Asp Gly Ser Asn Lys Tyr Tyr Ala Asp Ser Val 50 55 60 Lys Gly
Arg Phe Thr Ile Ser Arg Asp Asn Ser Lys Asn Thr Leu Tyr 65 70 75 80
Leu Gln Met Asn Ser Leu Arg Ala Glu Asp Thr Ala Leu Tyr Tyr Cys 85
90 95 Ala Thr Gly Val Thr Met Phe Gln Gly Ala Gln Thr Asn Ala Glu
Tyr 100 105 110 Leu His Tyr Trp Gly Gln Gly Ser Leu Val Thr Ile Ser
Ser 115 120 125 <210> SEQ ID NO 253 <211> LENGTH: 8
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA7 CDRH1
<400> SEQUENCE: 253 Gly Phe Thr Phe Ser Arg Tyr Gly 1 5
<210> SEQ ID NO 254 <211> LENGTH: 8 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA7 CDRH2 <400> SEQUENCE: 254
Val Ser Gly Asp Gly Ser Ser Thr 1 5 <210> SEQ ID NO 255
<211> LENGTH: 15 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA7 CDRH3 <400> SEQUENCE: 255 Val Lys Asp Phe
Trp Ser Gly Asp Gln Ser Leu Glu Ser Asp Phe 1 5 10 15 <210>
SEQ ID NO 256 <211> LENGTH: 122 <212> TYPE: PRT
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA7 VH <400> SEQUENCE: 256
Glu Val Gln Leu Val Glu Ser Gly Gly Gly Leu Val Gln Pro Gly Gly 1 5
10 15 Ser Leu Arg Leu Ser Cys Ser Ala Ser Gly Phe Thr Phe Ser Arg
Tyr 20 25 30 Gly Met Val Trp Ala Arg Gln Ala Pro Gly Lys Gly Leu
Glu Tyr Leu 35 40 45 Ser Gly Val Ser Gly Asp Gly Ser Ser Thr Tyr
Tyr Ala Asn Ser Val 50 55 60 Lys Gly Arg Phe Thr Ile Ser Arg Asp
Asn Ser Lys Asn Thr Leu Tyr 65 70 75 80 Leu His Met Ser Arg Leu Arg
Asp Glu Asp Thr Ala Met Tyr Tyr Cys 85 90 95 Val Lys Asp Phe Trp
Ser Gly Asp Gln Ser Leu Glu Ser Asp Phe Trp 100 105 110 Gly Gln Gly
Ala Leu Val Thr Val Ser Ser 115 120 <210> SEQ ID NO 257
<211> LENGTH: 8 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA8 CDRH1 <400> SEQUENCE: 257 Gly Phe Thr Phe
Ser Ala His Ala 1 5 <210> SEQ ID NO 258 <211> LENGTH: 8
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA8 CDRH2
<400> SEQUENCE: 258 Ile Ser Arg Asn Glu Asp Tyr Thr 1 5
<210> SEQ ID NO 259 <211> LENGTH: 12 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA8 CDRH3 <400> SEQUENCE: 259
Val Lys Asp Phe Gly Thr Ser Pro Gln Thr Asp Phe 1 5 10 <210>
SEQ ID NO 260 <211> LENGTH: 119 <212> TYPE: PRT
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA8 VH <400> SEQUENCE: 260
Asp Glu Arg Leu Val Glu Ser Gly Gly Gly Leu Val Gln Pro Gly Gly 1 5
10 15 Ser Leu Arg Leu Val Cys Ser Ala Ser Gly Phe Thr Phe Ser Ala
His 20 25 30 Ala Met His Trp Val Arg Gln Pro Pro Gly Lys Gly Leu
Glu Tyr Val 35 40 45 Ser Thr Ile Ser Arg Asn Glu Asp Tyr Thr Tyr
Tyr Ala Asp Ser Val 50 55 60 Lys Gly Arg Phe Thr Ile Ser Arg Asp
Asn Ser Lys Asn Ser Leu Tyr 65 70 75 80 Leu Gln Met Arg Arg Leu Arg
Pro Glu Asp Thr Ala Ile Tyr Tyr Cys 85 90 95 Val Lys Asp Phe Gly
Thr Ser Pro Gln Thr Asp Phe Trp Gly Gln Gly 100 105 110 Thr Leu Val
Ala Val Ser Ser 115 <210> SEQ ID NO 261 <211> LENGTH: 8
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA76 CDRH1
<400> SEQUENCE: 261 Gly Phe Thr Phe Ser Thr Tyr Phe 1 5
<210> SEQ ID NO 262 <211> LENGTH: 8 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA76 CDRH2 <400> SEQUENCE:
262 Ile Ser Ser Thr Gly Ser Tyr Lys 1 5 <210> SEQ ID NO 263
<211> LENGTH: 17 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA76 CDRH3 <400> SEQUENCE: 263 Ala Arg Pro Phe
His Ser Glu Tyr Thr Tyr Gly Leu Asp Ala Phe Asp 1 5 10 15 Ile
<210> SEQ ID NO 264 <211> LENGTH: 124 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA76 VH <400> SEQUENCE: 264
Glu Val Gln Leu Val Glu Ser Gly Gly Gly Leu Val Lys Pro Gly Gly 1 5
10 15 Ser Leu Arg Leu Ser Cys Ala Ala Ser Gly Phe Thr Phe Ser Thr
Tyr 20 25 30 Phe Met His Trp Val Arg Gln Ala Pro Gly Lys Gly Leu
Glu Trp Val 35 40 45 Ala Ser Ile Ser Ser Thr Gly Ser Tyr Lys Phe
Tyr Ala Asp Ser Val 50 55 60 Lys Gly Arg Phe Thr Ile Ser Arg Asp
Asn Thr Lys Asn Ser Leu Phe 65 70 75 80 Leu Gln Met Asn Ser Leu Arg
Ala Glu Asp Thr Ala Val Phe Tyr Cys 85 90 95 Ala Arg Pro Phe His
Ser Glu Tyr Thr Tyr Gly Leu Asp Ala Phe Asp 100 105 110 Ile Trp Gly
Gln Gly Thr Met Leu Thr Val Ser Ser 115 120 <210> SEQ ID NO
265 <211> LENGTH: 10 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKA117 CDRH1 <400> SEQUENCE: 265 Gly Gly
Ser Ile Arg Arg Thr Asn Ser Tyr 1 5 10 <210> SEQ ID NO 266
<211> LENGTH: 7 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA117 CDRH2 <400> SEQUENCE: 266 Ile Ser Tyr Ser
Gly Ser Thr 1 5 <210> SEQ ID NO 267 <211> LENGTH: 17
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA117 CDRH3
<400> SEQUENCE: 267 Ala Arg Leu Asn Asp Gly Ser Thr Val Thr
Thr Ser Ser Tyr Phe Asp 1 5 10 15 Tyr <210> SEQ ID NO 268
<211> LENGTH: 125 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA117 VH <400> SEQUENCE: 268 Gln Leu Gln Leu
Gln Glu Ser Gly Pro Gly Leu Val Lys Pro Ser Glu 1 5 10 15 Thr Leu
Ser Leu Thr Cys Thr Val Ser Gly Gly Ser Ile Arg Arg Thr 20 25 30
Asn Ser Tyr Trp Gly Trp Ile Arg Gln Thr Thr Gly Lys Gly Leu Gln 35
40 45 Trp Ile Gly Ser Ile Ser Tyr Ser Gly Ser Thr Phe Tyr Asn Pro
Ser 50 55 60 Leu Lys Ser Arg Val Thr Ile Ser Leu Asp Thr Ser Lys
Asp His Phe 65 70 75 80 Ser Leu Glu Leu Ser Ser Val Thr Ala Ala Asp
Thr Ala Ile Tyr Tyr 85 90 95 Cys Ala Arg Leu Asn Asp Gly Ser Thr
Val Thr Thr Ser Ser Tyr Phe 100 105 110 Asp Tyr Trp Gly Gln Gly Thr
Leu Val Thr Val Ser Ser 115 120 125 <210> SEQ ID NO 269
<211> LENGTH: 8 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKB27 CDRH1 <400> SEQUENCE: 269 Gly Tyr Ser Phe
Thr Ser Ser Trp 1 5 <210> SEQ ID NO 270 <211> LENGTH: 8
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKB27 CDRH2
<400> SEQUENCE: 270 Ile Asp Pro Ser Asp Ser Tyr Thr 1 5
<210> SEQ ID NO 271 <211> LENGTH: 14 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKB27 CDRH3 <400> SEQUENCE:
271 Ala Arg His Asp Tyr Ser Val Ser Glu Asn Gly Met Asp Val 1 5 10
<210> SEQ ID NO 272 <211> LENGTH: 121 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKB27 VH <400> SEQUENCE: 272
Glu Val Gln Leu Val Gln Ser Gly Ala Glu Val Lys Lys Pro Gly Glu 1 5
10 15 Ser Leu Arg Ile Ser Cys Lys Ala Ser Gly Tyr Ser Phe Thr Ser
Ser 20 25 30 Trp Ile Asn Trp Val Arg Gln Met Pro Gly Lys Gly Leu
Glu Trp Met 35 40 45 Gly Arg Ile Asp Pro Ser Asp Ser Tyr Thr Thr
Tyr Asn Pro Ser Phe 50 55 60 Gln Gly His Val Thr Ile Ser Val Asp
Lys Ser Ile Gly Thr Ala Tyr 65 70 75 80 Leu Gln Trp Asn Ser Leu Arg
Ala Ser Asp Thr Ala Met Tyr Tyr Cys 85 90 95 Ala Arg His Asp Tyr
Ser Val Ser Glu Asn Gly Met Asp Val Trp Gly 100 105 110 Gln Gly Thr
Thr Val Thr Val Ser Ser
115 120 <210> SEQ ID NO 273 <211> LENGTH: 8 <212>
TYPE: PRT <213> ORGANISM: Artificial Sequence <220>
FEATURE: <223> OTHER INFORMATION: ZKB29 CDRH1 <400>
SEQUENCE: 273 Gly Phe Thr Phe Ser Ser Tyr Thr 1 5 <210> SEQ
ID NO 274 <211> LENGTH: 8 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKB29 CDRH2 <400> SEQUENCE: 274 Ile Ser
Tyr Asp Gly Ser His Lys 1 5 <210> SEQ ID NO 275 <211>
LENGTH: 12 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKB29
CDRH3 <400> SEQUENCE: 275 Ala Arg Arg Ser Tyr Ser Ile Ser Cys
Phe Asp Tyr 1 5 10 <210> SEQ ID NO 276 <211> LENGTH:
119 <212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKB29 VH
<400> SEQUENCE: 276 Gln Val Gln Leu Val Glu Ser Gly Gly Gly
Val Val Gln Pro Gly Arg 1 5 10 15 Ser Leu Arg Leu Ser Cys Ala Ala
Ser Gly Phe Thr Phe Ser Ser Tyr 20 25 30 Thr Met His Trp Val Arg
Gln Ala Pro Gly Lys Gly Leu Glu Trp Val 35 40 45 Ala Val Ile Ser
Tyr Asp Gly Ser His Lys Phe Tyr Ala Asp Ser Val 50 55 60 Lys Gly
Arg Phe Thr Ile Ser Arg Asp Asn Ser Lys Asp Thr Leu Tyr 65 70 75 80
Leu Gln Met Asn Ser Leu Arg Ala Glu Asp Thr Ala Leu Tyr Tyr Cys 85
90 95 Ala Arg Arg Ser Tyr Ser Ile Ser Cys Phe Asp Tyr Trp Gly Gln
Gly 100 105 110 Thr Leu Val Thr Ile Ser Ser 115 <210> SEQ ID
NO 277 <211> LENGTH: 8 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKB34 CDRH1 <400> SEQUENCE: 277 Gly Phe
Thr Phe Ser Arg Ser Gly 1 5 <210> SEQ ID NO 278 <211>
LENGTH: 8 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKB34
CDRH2 <400> SEQUENCE: 278 Val Ser Tyr Asp Gly Ser Asn Lys 1 5
<210> SEQ ID NO 279 <211> LENGTH: 19 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKB34 CDRH3 <400> SEQUENCE:
279 Ala Lys Asp Leu Thr Met Val Arg Gly Val His Tyr Tyr Tyr Tyr Val
1 5 10 15 Met Asp Val <210> SEQ ID NO 280 <211> LENGTH:
126 <212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKB34 VH
<400> SEQUENCE: 280 Gln Val Gln Leu Val Glu Ser Gly Gly Gly
Val Val Gln Pro Gly Arg 1 5 10 15 Ser Leu Arg Leu Ser Cys Ala Ala
Ser Gly Phe Thr Phe Ser Arg Ser 20 25 30 Gly Met His Trp Val Arg
Gln Ala Pro Gly Lys Gly Leu Glu Trp Val 35 40 45 Ala Val Val Ser
Tyr Asp Gly Ser Asn Lys Tyr Tyr Ser Asp Ser Val 50 55 60 Lys Gly
Arg Phe Thr Ile Ser Arg Asp Asn Ser Lys Asn Thr Leu Tyr 65 70 75 80
Leu Gln Met Asn Ser Leu Arg Val Glu Asp Thr Ala Val Tyr Tyr Cys 85
90 95 Ala Lys Asp Leu Thr Met Val Arg Gly Val His Tyr Tyr Tyr Tyr
Val 100 105 110 Met Asp Val Trp Gly Gln Gly Thr Thr Val Thr Val Ser
Ser 115 120 125 <210> SEQ ID NO 281 <211> LENGTH: 8
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKB39 CDRH1
<400> SEQUENCE: 281 Gly Tyr Thr Phe Asp Asp Tyr Tyr 1 5
<210> SEQ ID NO 282 <211> LENGTH: 8 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKB39 CDRH2 <400> SEQUENCE:
282 Ile Asn Pro His Arg Gly Gly Thr 1 5 <210> SEQ ID NO 283
<211> LENGTH: 23 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKB39 CDRH3 <400> SEQUENCE: 283 Val Arg Asp Gln
Tyr Cys Asp Gly Gly Asn Cys Tyr Gly Ile His Gln 1 5 10 15 Pro His
Tyr Gly Met Asp Val 20 <210> SEQ ID NO 284 <211>
LENGTH: 130 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKB39
VH <400> SEQUENCE: 284 Gln Val Gln Leu Val Gln Ser Gly Ala
Glu Val Lys Lys Pro Gly Ala 1 5 10 15 Ser Leu Lys Val Ser Cys Lys
Ala Ser Gly Tyr Thr Phe Asp Asp Tyr 20 25 30 Tyr Ile His Trp Val
Arg Gln Ala Pro Gly Gln Gly Leu Glu Trp Leu 35 40 45 Gly Arg Ile
Asn Pro His Arg Gly Gly Thr Asn Tyr Ala Gln Lys Phe 50 55 60 Gln
Gly Arg Val Ile Met Thr Leu Asp Met Ser Ile Ser Thr Thr Tyr 65 70
75 80 Met Glu Leu Arg Arg Ile Thr Ser Asp Asp Ala Ala Val Tyr Tyr
Cys 85 90 95 Val Arg Asp Gln Tyr Cys Asp Gly Gly Asn Cys Tyr Gly
Ile His Gln 100 105 110 Pro His Tyr Gly Met Asp Val Trp Gly Gln Gly
Thr Thr Val Thr Val 115 120 125 Ser Ser 130 <210> SEQ ID NO
285 <211> LENGTH: 8 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKB46 CDRH1 <400> SEQUENCE: 285 Gly Tyr
Ser Phe Thr Ser Tyr Trp 1 5 <210> SEQ ID NO 286 <211>
LENGTH: 8 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKB46
CDRH2 <400> SEQUENCE: 286
Ile Asp Pro Ser Asp Ser Tyr Thr 1 5 <210> SEQ ID NO 287
<211> LENGTH: 17 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKB46 CDRH3 <400> SEQUENCE: 287 Ala Arg Arg Glu
Tyr Ser Ser Ser Ser Gly Gln Glu Asp Trp Phe Asp 1 5 10 15 Pro
<210> SEQ ID NO 288 <211> LENGTH: 124 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKB46 VH <400> SEQUENCE: 288
Glu Val Gln Leu Val Gln Ser Gly Ala Glu Val Lys Lys Pro Gly Glu 1 5
10 15 Ser Leu Arg Ile Ser Cys Lys Gly Ser Gly Tyr Ser Phe Thr Ser
Tyr 20 25 30 Trp Ile Ser Trp Val Arg Gln Met Pro Gly Lys Gly Leu
Glu Trp Met 35 40 45 Gly Arg Ile Asp Pro Ser Asp Ser Tyr Thr Asn
Tyr Ser Pro Ser Phe 50 55 60 Gln Gly His Val Thr Ile Ser Ala Asp
Lys Ser Ile Ser Thr Ala Tyr 65 70 75 80 Leu Gln Trp Ser Ser Leu Lys
Ala Ser Asp Thr Ala Met Tyr Tyr Cys 85 90 95 Ala Arg Arg Glu Tyr
Ser Ser Ser Ser Gly Gln Glu Asp Trp Phe Asp 100 105 110 Pro Trp Gly
Gln Gly Thr Leu Val Thr Val Ser Ser 115 120 <210> SEQ ID NO
289 <211> LENGTH: 8 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKB53 CDRH1 <400> SEQUENCE: 289 Gly Phe
Thr Phe Ser Ser Tyr Ala 1 5 <210> SEQ ID NO 290 <211>
LENGTH: 8 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKB53
CDRH2 <400> SEQUENCE: 290 Ile Ser Tyr Asp Gly Ser Asn Arg 1 5
<210> SEQ ID NO 291 <211> LENGTH: 15 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKB53 CDRH3 <400> SEQUENCE:
291 Ala Arg His Val Glu Gln Leu Pro Ser Ser Gly Tyr Phe Gln His 1 5
10 15 <210> SEQ ID NO 292 <211> LENGTH: 122 <212>
TYPE: PRT <213> ORGANISM: Artificial Sequence <220>
FEATURE: <223> OTHER INFORMATION: ZKB53 VH <400>
SEQUENCE: 292 Gln Val Gln Leu Val Glu Ser Gly Gly Gly Val Val Gln
Pro Gly Arg 1 5 10 15 Ser Leu Arg Leu Ser Cys Ala Ala Ser Gly Phe
Thr Phe Ser Ser Tyr 20 25 30 Ala Met His Trp Val Arg Gln Thr Pro
Gly Lys Gly Leu Glu Trp Val 35 40 45 Thr Val Ile Ser Tyr Asp Gly
Ser Asn Arg Tyr Tyr Ala Asp Ser Val 50 55 60 Lys Gly Arg Phe Thr
Ile Ser Arg Asp Asn Ser Lys Asn Thr Leu Tyr 65 70 75 80 Leu Gln Met
Asn Ser Leu Arg Ser Glu Asp Thr Ala Val Tyr Tyr Cys 85 90 95 Ala
Arg His Val Glu Gln Leu Pro Ser Ser Gly Tyr Phe Gln His Trp 100 105
110 Gly Gln Gly Thr Leu Val Thr Val Ser Ser 115 120 <210> SEQ
ID NO 293 <211> LENGTH: 8 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKC26 CDRH1 <400> SEQUENCE: 293 Gly Phe
Ile Phe Ser Asp Phe Tyr 1 5 <210> SEQ ID NO 294 <211>
LENGTH: 8 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKC26
CDRH2 <400> SEQUENCE: 294 Ile Gly His Asp Gly Ser Tyr Ile 1 5
<210> SEQ ID NO 295 <211> LENGTH: 9 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKC26 CDRH3 <400> SEQUENCE:
295 Ala Arg Ala His Gly Gly Phe Arg His 1 5 <210> SEQ ID NO
296 <211> LENGTH: 116 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKC26 VH <400> SEQUENCE: 296 Gln Val Gln
Val Val Glu Ser Gly Gly Gly Leu Val Lys Pro Gly Gly 1 5 10 15 Ser
Leu Arg Leu Ser Cys Ala Ala Ser Gly Phe Ile Phe Ser Asp Phe 20 25
30 Tyr Met Ser Trp Met Arg Gln Ala Pro Gly Lys Gly Leu Glu Trp Val
35 40 45 Ala Tyr Ile Gly His Asp Gly Ser Tyr Ile Leu Tyr Ala Asp
Ser Val 50 55 60 Lys Gly Arg Phe Thr Ile Ser Arg Asp Asn Ala Lys
Asn Ser Leu Phe 65 70 75 80 Leu Arg Met Asn Ser Leu Arg Val Glu Asp
Thr Ala Val Tyr Tyr Cys 85 90 95 Ala Arg Ala His Gly Gly Phe Arg
His Trp Gly Gln Gly Thr Val Val 100 105 110 Ala Val Ser Pro 115
<210> SEQ ID NO 297 <211> LENGTH: 8 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKD5 CDRH1 <400> SEQUENCE: 297
Gly Phe Thr Phe Thr Ser Tyr Gly 1 5 <210> SEQ ID NO 298
<211> LENGTH: 8 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKD5 CDRH2 <400> SEQUENCE: 298 Ile Ser Tyr Asp
Gly Ser Asn Lys 1 5 <210> SEQ ID NO 299 <211> LENGTH:
17 <212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKD5 CDRH3
<400> SEQUENCE: 299 Ala Arg Asp Arg Asp His Tyr Asp Leu Trp
Asn Ala Tyr Thr Phe Asp 1 5 10 15 Tyr <210> SEQ ID NO 300
<211> LENGTH: 124 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKD5 VH <400> SEQUENCE: 300 Gln Val Gln Leu Val
Glu Ser Gly Gly Gly Val Val Gln Pro Gly Arg
1 5 10 15 Ser Leu Arg Leu Ser Cys Ala Ala Ser Gly Phe Thr Phe Thr
Ser Tyr 20 25 30 Gly Met His Trp Val Arg Gln Thr Pro Gly Lys Gly
Leu Asp Trp Val 35 40 45 Ala Val Ile Ser Tyr Asp Gly Ser Asn Lys
Tyr Tyr Ala Asp Ser Val 50 55 60 Lys Gly Arg Phe Thr Ile Ser Arg
Asp Asn Ser Lys Asp Thr Leu Tyr 65 70 75 80 Leu Gln Met Asn Ser Leu
Arg Ala Ala Asp Thr Ala Leu Tyr Tyr Cys 85 90 95 Ala Arg Asp Arg
Asp His Tyr Asp Leu Trp Asn Ala Tyr Thr Phe Asp 100 105 110 Tyr Trp
Gly Gln Gly Thr Leu Val Thr Val Ser Ser 115 120 <210> SEQ ID
NO 301 <211> LENGTH: 8 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKD7 CDRH1 <400> SEQUENCE: 301 Gly Phe Thr
Phe Ser Asn Tyr Ala 1 5 <210> SEQ ID NO 302 <211>
LENGTH: 8 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKD7
CDRH2 <400> SEQUENCE: 302 Ile Ser Tyr Asp Val Ser Asp Lys 1 5
<210> SEQ ID NO 303 <211> LENGTH: 20 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKD7 CDRH3 <400> SEQUENCE: 303
Ala Gly Gly Pro Leu Gly Val Val Val Ile Lys Pro Ser Asn Ala Glu 1 5
10 15 His Phe His His 20 <210> SEQ ID NO 304 <211>
LENGTH: 127 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKD7
VH <400> SEQUENCE: 304 Gln Val Gln Leu Val Glu Ser Gly Gly
Gly Val Val Gln Pro Gly Arg 1 5 10 15 Ser Leu Arg Leu Ser Cys Ala
Ala Ser Gly Phe Thr Phe Ser Asn Tyr 20 25 30 Ala Met His Trp Val
Arg Gln Ala Pro Gly Lys Gly Leu Glu Trp Val 35 40 45 Ala Val Ile
Ser Tyr Asp Val Ser Asp Lys Tyr Tyr Ala Asp Ser Val 50 55 60 Lys
Gly Arg Phe Thr Ile Ser Arg Asp Asn Ser Lys Asn Thr Leu Phe 65 70
75 80 Leu Gln Met Asn Ser Leu Arg Ala Glu Asp Thr Ala Ala Tyr Tyr
Cys 85 90 95 Ala Gly Gly Pro Leu Gly Val Val Val Ile Lys Pro Ser
Asn Ala Glu 100 105 110 His Phe His His Trp Gly Gln Gly Thr Leu Val
Thr Val Ser Ser 115 120 125 <210> SEQ ID NO 305 <211>
LENGTH: 8 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKD8
CDRH1 <400> SEQUENCE: 305 Gly Phe Thr Phe Ile Asn Tyr Ala 1 5
<210> SEQ ID NO 306 <211> LENGTH: 8 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKD8 CDRH2 <400> SEQUENCE: 306
Ile Ser Tyr Asp Gly Ser Asn Lys 1 5 <210> SEQ ID NO 307
<211> LENGTH: 16 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKD8 CDRH3 <400> SEQUENCE: 307 Ala Thr Asp Ala
Asp Ala Tyr Gly Asp Ser Gly Ala Asn Phe His Tyr 1 5 10 15
<210> SEQ ID NO 308 <211> LENGTH: 123 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKD8 VH <400> SEQUENCE: 308
Gln Val Gln Leu Val Glu Ser Gly Gly Gly Val Val Gln Pro Gly Lys 1 5
10 15 Ser Leu Arg Leu Ser Cys Ala Ala Ser Gly Phe Thr Phe Ile Asn
Tyr 20 25 30 Ala Ile His Trp Val Arg Gln Ala Pro Gly Lys Gly Leu
Glu Trp Val 35 40 45 Ala Val Ile Ser Tyr Asp Gly Ser Asn Lys Phe
Tyr Thr Asp Ser Val 50 55 60 Lys Gly Arg Phe Thr Ile Ser Arg Asp
Asn Ser Lys Asn Thr Leu Tyr 65 70 75 80 Leu Gln Met Asn Ser Leu Arg
Ala Asp Asp Thr Ala Val Tyr Tyr Cys 85 90 95 Ala Thr Asp Ala Asp
Ala Tyr Gly Asp Ser Gly Ala Asn Phe His Tyr 100 105 110 Trp Gly Gln
Gly Thr Leu Val Thr Val Ser Ser 115 120 <210> SEQ ID NO 309
<211> LENGTH: 10 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKD15 CDRH1 <400> SEQUENCE: 309 Asp Ala Ser Ile
Ser Ser Gly Gly Phe Ser 1 5 10 <210> SEQ ID NO 310
<211> LENGTH: 7 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKD15 CDRH2 <400> SEQUENCE: 310 Ile Tyr Ser Ser
Gly Asp Thr 1 5 <210> SEQ ID NO 311 <211> LENGTH: 18
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKD15 CDRH3
<400> SEQUENCE: 311 Ala Arg Ala His Thr Pro Thr Ser Lys Phe
Tyr Tyr Tyr Tyr Ala Met 1 5 10 15 Asp Val <210> SEQ ID NO 312
<211> LENGTH: 126 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKD15 VH <400> SEQUENCE: 312 Gln Leu Gln Leu Gln
Glu Ser Gly Ser Gly Leu Val Lys Pro Ser Gln 1 5 10 15 Thr Leu Ser
Leu Thr Cys Thr Val Ser Asp Ala Ser Ile Ser Ser Gly 20 25 30 Gly
Phe Ser Trp Ser Trp Ile Arg Gln Pro Leu Gly Lys Gly Leu Glu 35 40
45 Trp Leu Gly Tyr Ile Tyr Ser Ser Gly Asp Thr Phe Tyr Asn Pro Ser
50 55 60 Leu Gln Gly Arg Val Thr Met Ser Val Asp Ile Phe Arg Ser
Gln Phe 65 70 75 80 Ser Leu Lys Leu Thr Ser Val Thr Ala Ala Asp Thr
Ala Met Tyr Tyr 85 90 95 Cys Ala Arg Ala His Thr Pro Thr Ser Lys
Phe Tyr Tyr Tyr Tyr Ala 100 105 110 Met Asp Val Trp Gly Gln Gly Thr
Thr Val Thr Val Ser Ser 115 120 125 <210> SEQ ID NO 313
<211> LENGTH: 8 <212> TYPE: PRT
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKD16 CDRH1 <400> SEQUENCE:
313 Gly Phe Thr Phe Ser Asp His Phe 1 5 <210> SEQ ID NO 314
<211> LENGTH: 10 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKD16 CDRH2 <400> SEQUENCE: 314 Ser Arg Asn Lys
Pro Asn Ser Tyr Thr Thr 1 5 10 <210> SEQ ID NO 315
<211> LENGTH: 17 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKD16 CDRH3 <400> SEQUENCE: 315 Ala Lys Val Gly
Gly Cys Tyr Gly Gly Asp Cys His Val Glu Asn Asp 1 5 10 15 Tyr
<210> SEQ ID NO 316 <211> LENGTH: 126 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKD16 VH <400> SEQUENCE: 316
Glu Val Gln Leu Val Glu Ser Gly Gly Asp Leu Val Gln Pro Gly Gly 1 5
10 15 Ser Leu Arg Leu Ser Cys Val Ala Ser Gly Phe Thr Phe Ser Asp
His 20 25 30 Phe Met Asp Trp Val Arg Gln Ala Pro Gly Lys Gly Leu
Glu Trp Val 35 40 45 Gly Arg Ser Arg Asn Lys Pro Asn Ser Tyr Thr
Thr Glu Tyr Ala Ala 50 55 60 Ser Val Lys Gly Arg Phe Ser Ile Ser
Arg Asp Asp Ser Lys Lys Ala 65 70 75 80 Leu Tyr Leu Gln Met Asn Ser
Leu Gln Thr Glu Asp Thr Ala Val Tyr 85 90 95 Tyr Cys Ala Lys Val
Gly Gly Cys Tyr Gly Gly Asp Cys His Val Glu 100 105 110 Asn Asp Tyr
Trp Gly Gln Gly Thr Leu Val Thr Val Ser Ser 115 120 125 <210>
SEQ ID NO 317 <211> LENGTH: 8 <212> TYPE: PRT
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKD17 CDRH1 <400> SEQUENCE:
317 Gly Phe Ile Phe Ser Asp Tyr Ala 1 5 <210> SEQ ID NO 318
<211> LENGTH: 8 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKD17 CDRH2 <400> SEQUENCE: 318 Ile Ser Tyr Asp
Gly Ser Ser Arg 1 5 <210> SEQ ID NO 319 <211> LENGTH:
20 <212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKD17 CDRH3
<400> SEQUENCE: 319 Ala Arg Gly Tyr Cys Ser Ser Gly Thr Cys
Phe Ser Thr Asn Ala Glu 1 5 10 15 Tyr Phe His Pro 20 <210>
SEQ ID NO 320 <211> LENGTH: 127 <212> TYPE: PRT
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKD17 VH <400> SEQUENCE: 320
Gln Val Gln Met Val Glu Ser Gly Gly Gly Val Val Gln Pro Gly Thr 1 5
10 15 Ser Leu Arg Leu Ser Cys Ala Thr Ser Gly Phe Ile Phe Ser Asp
Tyr 20 25 30 Ala Met His Trp Val Arg Gln Ala Pro Gly Lys Gly Leu
Glu Trp Val 35 40 45 Ala Val Ile Ser Tyr Asp Gly Ser Ser Arg Leu
Tyr Ala Asp Ser Val 50 55 60 Lys Gly Arg Phe Thr Val Ser Arg Asp
Asn Ser Lys Asn Thr Leu Tyr 65 70 75 80 Leu Gln Met His Ser Leu Arg
Ala Gly Asp Thr Ala Val Tyr Tyr Cys 85 90 95 Ala Arg Gly Tyr Cys
Ser Ser Gly Thr Cys Phe Ser Thr Asn Ala Glu 100 105 110 Tyr Phe His
Pro Trp Gly Gln Gly Thr Leu Ala Thr Ile Ser Ser 115 120 125
<210> SEQ ID NO 321 <211> LENGTH: 8 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKD20 CDRH1 <400> SEQUENCE:
321 Gly Phe Thr Phe Ser Asp His Phe 1 5 <210> SEQ ID NO 322
<211> LENGTH: 10 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKD20 CDRH2 <400> SEQUENCE: 322 Ser Arg Asn Lys
Pro Asn Ser Tyr Thr Thr 1 5 10 <210> SEQ ID NO 323
<211> LENGTH: 17 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKD20 CDRH3 <400> SEQUENCE: 323 Ala Arg Val Gly
Gly Cys Asn Gly Gly Asp Cys His Val Glu Asn Asp 1 5 10 15 Tyr
<210> SEQ ID NO 324 <211> LENGTH: 126 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKD20 VH <400> SEQUENCE: 324
Glu Val Gln Leu Val Glu Ser Gly Gly Gly Leu Val Gln Pro Gly Gly 1 5
10 15 Ser Leu Arg Leu Ser Cys Val Ala Ser Gly Phe Thr Phe Ser Asp
His 20 25 30 Phe Met Asp Trp Val Arg Gln Ala Pro Gly Lys Gly Leu
Glu Trp Val 35 40 45 Gly Arg Ser Arg Asn Lys Pro Asn Ser Tyr Thr
Thr Glu Tyr Ala Ala 50 55 60 Ser Val Lys Gly Arg Phe Thr Ile Ser
Arg Asp Asp Ser Lys Asn Ser 65 70 75 80 Leu Tyr Leu Gln Met Asn Ser
Leu Gln Thr Glu Asp Thr Ala Val Tyr 85 90 95 Tyr Cys Ala Arg Val
Gly Gly Cys Asn Gly Gly Asp Cys His Val Glu 100 105 110 Asn Asp Tyr
Trp Gly Gln Gly Thr Leu Val Thr Val Ser Ser 115 120 125 <210>
SEQ ID NO 325 <211> LENGTH: 8 <212> TYPE: PRT
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA134 CDRH1 <400> SEQUENCE:
325 Gly Gly Thr Phe Ser Ala Tyr Ala 1 5 <210> SEQ ID NO 326
<211> LENGTH: 8 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA134 CDRH2 <400> SEQUENCE: 326 Ile Ile Pro Phe
Phe Gly Thr Ala 1 5 <210> SEQ ID NO 327 <211> LENGTH:
19
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA134 CDRH3
<400> SEQUENCE: 327 Ala Arg Ser Asp Ile Val Ser Thr Thr Arg
Gly Tyr His His Tyr Gly 1 5 10 15 Met Asp Val <210> SEQ ID NO
328 <211> LENGTH: 126 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKA134 VH <400> SEQUENCE: 328 Gln Val His
Leu Val Gln Ser Gly Ala Glu Val Lys Lys Pro Gly Ser 1 5 10 15 Ser
Val Asn Val Ser Cys Lys Ala Ser Gly Gly Thr Phe Ser Ala Tyr 20 25
30 Ala Ile Ser Trp Val Arg Gln Ala Pro Gly Gln Gly Leu Glu Trp Met
35 40 45 Gly Gly Ile Ile Pro Phe Phe Gly Thr Ala Tyr Tyr Ala Gln
Lys Phe 50 55 60 Lys Gly Arg Val Thr Val Thr Ala Asp Lys Ser Thr
Ser Thr Val Tyr 65 70 75 80 Met Glu Met Thr Ser Leu Arg Ser Glu Asp
Thr Ala Val Tyr Tyr Cys 85 90 95 Ala Arg Ser Asp Ile Val Ser Thr
Thr Arg Gly Tyr His His Tyr Gly 100 105 110 Met Asp Val Trp Gly Gln
Gly Thr Thr Val Thr Val Ser Ser 115 120 125 <210> SEQ ID NO
329 <211> LENGTH: 8 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKA246 CDRH1 <400> SEQUENCE: 329 Gly Tyr
Thr Phe Ser Asp Tyr Tyr 1 5 <210> SEQ ID NO 330 <211>
LENGTH: 8 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA246
CDRH2 <400> SEQUENCE: 330 Ile Asn Pro Tyr Ser Gly Gly Thr 1 5
<210> SEQ ID NO 331 <211> LENGTH: 14 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA246 CDRH3 <400> SEQUENCE:
331 Ala Arg Gly Phe Thr Met Ile Ser Asp Arg Glu Phe Asp Pro 1 5 10
<210> SEQ ID NO 332 <211> LENGTH: 121 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA246 VH <400> SEQUENCE: 332
Gln Val Gln Leu Val Gln Ser Gly Ala Glu Val Lys Arg Pro Gly Ala 1 5
10 15 Ser Val Lys Val Ser Cys Lys Ala Ser Gly Tyr Thr Phe Ser Asp
Tyr 20 25 30 Tyr Met His Trp Val Arg Gln Ala Pro Gly Gln Gly Leu
Glu Trp Met 35 40 45 Gly Arg Ile Asn Pro Tyr Ser Gly Gly Thr Asn
Tyr Ala Gln Lys Phe 50 55 60 His Gly Arg Val Thr Val Thr Arg Asp
Thr Ser Ile Ser Thr Val Tyr 65 70 75 80 Met Glu Leu Arg Gly Leu Arg
Ser Asp Asp Thr Ala Val Tyr Tyr Cys 85 90 95 Ala Arg Gly Phe Thr
Met Ile Ser Asp Arg Glu Phe Asp Pro Trp Gly 100 105 110 Gln Gly Thr
Leu Val Thr Val Ser Ser 115 120 <210> SEQ ID NO 333
<211> LENGTH: 8 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA256 CDRH1 <400> SEQUENCE: 333 Gly Phe Thr Phe
Ser Thr Tyr Trp 1 5 <210> SEQ ID NO 334 <211> LENGTH: 8
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA256 CDRH2
<400> SEQUENCE: 334 Ile Lys Gln Asp Gly Ser Glu Lys 1 5
<210> SEQ ID NO 335 <211> LENGTH: 20 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA256 CDRH3 <400> SEQUENCE:
335 Ala Arg Asp Pro Gly Tyr Asp Asp Phe Trp Ser Gly Ser Tyr Ser Gly
1 5 10 15 Ser Phe Asp Ile 20 <210> SEQ ID NO 336 <211>
LENGTH: 127 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA256
VH <400> SEQUENCE: 336 Glu Val Gln Leu Val Glu Ser Gly Gly
Gly Leu Val Gln Pro Gly Gly 1 5 10 15 Ser Leu Arg Leu Ser Cys Ala
Ala Ser Gly Phe Thr Phe Ser Thr Tyr 20 25 30 Trp Met Thr Trp Val
Arg Gln Ala Pro Gly Lys Gly Leu Glu Trp Val 35 40 45 Ala Asn Ile
Lys Gln Asp Gly Ser Glu Lys Tyr Tyr Val Asp Ser Val 50 55 60 Lys
Gly Arg Phe Thr Ile Ser Arg Asp Asn Thr Lys Asn Ser Leu Tyr 65 70
75 80 Leu Gln Val Asn Ser Leu Arg Ala Glu Asp Thr Ala Ile Tyr Tyr
Cys 85 90 95 Ala Arg Asp Pro Gly Tyr Asp Asp Phe Trp Ser Gly Ser
Tyr Ser Gly 100 105 110 Ser Phe Asp Ile Trp Gly Gln Gly Thr Met Val
Thr Val Ser Ser 115 120 125 <210> SEQ ID NO 337 <211>
LENGTH: 8 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKB42
CDRH1 <400> SEQUENCE: 337 Gly Phe Thr Phe Asn Asn Tyr Gly 1 5
<210> SEQ ID NO 338 <211> LENGTH: 8 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKB42 CDRH2 <400> SEQUENCE:
338 Ile Ser Tyr Asp Gly Asn Lys Lys 1 5 <210> SEQ ID NO 339
<211> LENGTH: 16 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKB42 CDRH3 <400> SEQUENCE: 339 Val Lys Tyr Gly
Glu Arg Ile Asn Gly Tyr Ser Asp Pro Phe Asp His 1 5 10 15
<210> SEQ ID NO 340 <211> LENGTH: 123 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKB42 VH <400> SEQUENCE: 340
Gln Val Gln Val Val Glu Ser Gly Gly Gly Val Val Gln Pro Gly Arg 1 5
10 15 Ser Leu Arg Leu Phe Cys Ala Ala Ser Gly Phe Thr Phe Asn Asn
Tyr 20 25 30 Gly Met His Trp Val Arg Gln Ala Pro Gly Lys Gly Leu
Glu Trp Val
35 40 45 Ala Leu Ile Ser Tyr Asp Gly Asn Lys Lys Tyr Tyr Ala Asp
Ser Val 50 55 60 Lys Gly Arg Phe Ser Ile Ser Arg Asp Asn Ser Lys
Asn Thr Leu Tyr 65 70 75 80 Leu Gln Met Asn Arg Leu Arg Ser Gly Asp
Thr Ala Val Tyr His Cys 85 90 95 Val Lys Tyr Gly Glu Arg Ile Asn
Gly Tyr Ser Asp Pro Phe Asp His 100 105 110 Trp Gly Gln Gly Thr Leu
Val Thr Val Ser Ser 115 120 <210> SEQ ID NO 341 <211>
LENGTH: 8 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKB85
CDRH1 <400> SEQUENCE: 341 Gly Tyr Thr Phe Thr Thr Tyr Ala 1 5
<210> SEQ ID NO 342 <211> LENGTH: 8 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKB85 CDRH2 <400> SEQUENCE:
342 Ile Asn Thr Asn Thr Gly Asn Pro 1 5 <210> SEQ ID NO 343
<211> LENGTH: 11 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKB85 CDRH3 <400> SEQUENCE: 343 Ala Arg Val Ile
Val Pro Tyr Ala Phe Asp Ile 1 5 10 <210> SEQ ID NO 344
<211> LENGTH: 118 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKB85 VH <400> SEQUENCE: 344 Gln Val Gln Leu Val
Gln Ser Gly Ser Glu Leu Lys Lys Pro Gly Ala 1 5 10 15 Ser Val Lys
Val Ser Cys Lys Ala Ser Gly Tyr Thr Phe Thr Thr Tyr 20 25 30 Ala
Met Asn Trp Val Arg Gln Ala Pro Gly Gln Gly Pro Glu Trp Val 35 40
45 Gly Trp Ile Asn Thr Asn Thr Gly Asn Pro Thr Tyr Ala Gln Gly Phe
50 55 60 Thr Gly Arg Phe Val Leu Ser Leu Asp Thr Ser Val Ser Thr
Ala Phe 65 70 75 80 Leu Gln Ile Ser Ser Leu Lys Ala Glu Asp Thr Ala
Val Tyr Tyr Cys 85 90 95 Ala Arg Val Ile Val Pro Tyr Ala Phe Asp
Ile Trp Gly Gln Gly Thr 100 105 110 Met Val Thr Val Ser Ser 115
<210> SEQ ID NO 345 <211> LENGTH: 8 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKB47 CDRH1 <400> SEQUENCE:
345 Gly Tyr Thr Phe Thr Asn Tyr Tyr 1 5 <210> SEQ ID NO 346
<211> LENGTH: 8 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKB47 CDRH2 <400> SEQUENCE: 346 Ile Asn Pro Ser
Gly Gly Pro Thr 1 5 <210> SEQ ID NO 347 <211> LENGTH:
15 <212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKB47 CDRH3
<400> SEQUENCE: 347 Ala Arg Asp Gln Tyr Gly Gly Tyr Ala Arg
Tyr Gly Met Asp Val 1 5 10 15 <210> SEQ ID NO 348 <211>
LENGTH: 122 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKB47
VH <400> SEQUENCE: 348 Gln Val Gln Leu Val Gln Ser Gly Ala
Glu Val Lys Lys Pro Gly Ala 1 5 10 15 Ser Val Lys Val Ser Cys Gln
Ala Ser Gly Tyr Thr Phe Thr Asn Tyr 20 25 30 Tyr Met His Trp Val
Arg Gln Ala Pro Gly Gln Gly Leu Glu Trp Met 35 40 45 Gly Ile Ile
Asn Pro Ser Gly Gly Pro Thr Ser Tyr Ala Gln Lys Phe 50 55 60 Gln
Gly Arg Val Thr Met Thr Thr Asp Thr Ser Thr Ser Thr Val Tyr 65 70
75 80 Met Glu Leu Ser Ser Leu Arg Ser Glu Asp Thr Ala Val Tyr Tyr
Cys 85 90 95 Ala Arg Asp Gln Tyr Gly Gly Tyr Ala Arg Tyr Gly Met
Asp Val Trp 100 105 110 Gly Gln Gly Thr Thr Val Thr Val Ser Ser 115
120 <210> SEQ ID NO 349 <211> LENGTH: 8 <212>
TYPE: PRT <213> ORGANISM: Artificial Sequence <220>
FEATURE: <223> OTHER INFORMATION: ZKC6 CDRH1 <400>
SEQUENCE: 349 Gly Tyr Thr Phe Thr Gly Tyr Tyr 1 5 <210> SEQ
ID NO 350 <211> LENGTH: 8 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKC6 CDRH2 <400> SEQUENCE: 350 Ile Asn Pro
Asn Ser Gly Gly Thr 1 5 <210> SEQ ID NO 351 <211>
LENGTH: 12 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKC6
CDRH3 <400> SEQUENCE: 351 Ala Arg Val Ser Asp Trp Gly Phe Ala
Phe Asp Ile 1 5 10 <210> SEQ ID NO 352 <211> LENGTH:
119 <212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKC6 VH
<400> SEQUENCE: 352 Gln Val Gln Leu Val Gln Ser Gly Thr Glu
Val Lys Lys Pro Gly Ala 1 5 10 15 Ser Val Lys Val Ser Cys Lys Ala
Ser Gly Tyr Thr Phe Thr Gly Tyr 20 25 30 Tyr Met His Trp Val Arg
Gln Ala Pro Gly Gln Gly Leu Glu Trp Met 35 40 45 Gly Arg Ile Asn
Pro Asn Ser Gly Gly Thr Asn Tyr Ala Gln Lys Phe 50 55 60 Gln Gly
Arg Val Thr Met Thr Arg Asp Thr Ser Ile Ser Thr Ala Tyr 65 70 75 80
Met Glu Leu Ser Gly Leu Arg Ser Asp Asp Thr Ala Val Tyr Tyr Cys 85
90 95 Ala Arg Val Ser Asp Trp Gly Phe Ala Phe Asp Ile Trp Gly Gln
Gly 100 105 110 Thr Met Val Thr Val Ser Gln 115 <210> SEQ ID
NO 353 <211> LENGTH: 8 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKA160 CDRH1 <400> SEQUENCE: 353 Gly Gly
Ser Ile Thr Ser Tyr Ser 1 5 <210> SEQ ID NO 354
<211> LENGTH: 7 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA160 CDRH2 <400> SEQUENCE: 354 Ile Phe Tyr Ser
Gly Ser Thr 1 5 <210> SEQ ID NO 355 <211> LENGTH: 15
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA160 CDRH3
<400> SEQUENCE: 355 Ala Arg Asp Gln Thr Met Pro Val Trp Val
Gly Gly Met Asp Val 1 5 10 15 <210> SEQ ID NO 356 <211>
LENGTH: 121 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA160
VH <400> SEQUENCE: 356 Gln Val Gln Leu Gln Glu Ser Gly Pro
Gly Leu Val Lys Pro Ser Glu 1 5 10 15 Thr Leu Ser Leu Thr Cys Thr
Val Ser Gly Gly Ser Ile Thr Ser Tyr 20 25 30 Ser Trp Ser Trp Ile
Arg Gln Pro Pro Gly Lys Gly Leu Glu Trp Ile 35 40 45 Gly Tyr Ile
Phe Tyr Ser Gly Ser Thr Asp Tyr Asn Pro Ser Leu Lys 50 55 60 Ser
Arg Val Thr Ile Ser Val Asp Thr Ser Lys Asp Gln Phe Ser Leu 65 70
75 80 Arg Leu Arg Ser Val Thr Ala Ala Asp Thr Ala Val Tyr Tyr Cys
Ala 85 90 95 Arg Asp Gln Thr Met Pro Val Trp Val Gly Gly Met Asp
Val Trp Gly 100 105 110 Gln Gly Thr Thr Val Thr Val Ser Ser 115 120
<210> SEQ ID NO 357 <211> LENGTH: 8 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA172 CDRH1 <400> SEQUENCE:
357 Gly Tyr Ile Phe Thr Arg Tyr Trp 1 5 <210> SEQ ID NO 358
<211> LENGTH: 8 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA172 CDRH2 <400> SEQUENCE: 358 Ile Asp Pro Ser
Asp Ser Tyr Thr 1 5 <210> SEQ ID NO 359 <211> LENGTH:
13 <212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA172 CDRH3
<400> SEQUENCE: 359 Ala Arg Gln Glu Thr Ala Arg Glu Asp Gly
Met Ala Val 1 5 10 <210> SEQ ID NO 360 <211> LENGTH:
120 <212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA172 VH
<400> SEQUENCE: 360 Glu Val Gln Leu Val Gln Ser Gly Ala Glu
Val Lys Lys Pro Gly Lys 1 5 10 15 Ser Leu Arg Ile Ser Cys Lys Gly
Ser Gly Tyr Ile Phe Thr Arg Tyr 20 25 30 Trp Ile Ser Trp Val Arg
Gln Met Pro Gly Lys Gly Leu Glu Trp Met 35 40 45 Gly Arg Ile Asp
Pro Ser Asp Ser Tyr Thr Asn Tyr Ser Pro Ser Phe 50 55 60 Gln Gly
His Val Thr Ile Ser Ala Asp Lys Ser Ile Ser Thr Ala Tyr 65 70 75 80
Leu Gln Trp Ser Ser Leu Lys Ala Ser Asp Thr Ala Met Tyr Tyr Cys 85
90 95 Ala Arg Gln Glu Thr Ala Arg Glu Asp Gly Met Ala Val Trp Gly
Gln 100 105 110 Gly Thr Thr Val Thr Val Ser Ser 115 120 <210>
SEQ ID NO 361 <211> LENGTH: 10 <212> TYPE: PRT
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA174 CDRH1 <400> SEQUENCE:
361 Gly Gly Ser Met Ser Asn Ser Tyr Tyr His 1 5 10 <210> SEQ
ID NO 362 <211> LENGTH: 7 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKA174 CDRH2 <400> SEQUENCE: 362 Ile Tyr
Tyr Ser Gly Ser Thr 1 5 <210> SEQ ID NO 363 <211>
LENGTH: 18 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA174
CDRH3 <400> SEQUENCE: 363 Ala Arg Asn Pro Val Phe Asn Pro Leu
Thr Leu Thr His Asp Ala Phe 1 5 10 15 Asp Ile <210> SEQ ID NO
364 <211> LENGTH: 126 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKA174 VH <400> SEQUENCE: 364 Gln Leu Gln
Leu Gln Glu Ser Gly Pro Gly Leu Val Lys Pro Ser Glu 1 5 10 15 Thr
Leu Ser Leu Thr Cys Thr Val Ser Gly Gly Ser Met Ser Asn Ser 20 25
30 Tyr Tyr His Trp Gly Trp Ile Arg Gln Pro Pro Gly Lys Gly Leu Glu
35 40 45 Trp Ile Gly Ser Ile Tyr Tyr Ser Gly Ser Thr Tyr Tyr Asn
Pro Ser 50 55 60 Leu Lys Ser Arg Val Thr Ile Ser Val Asp Thr Ser
Lys Asn Gln Phe 65 70 75 80 Ser Leu Lys Leu Asn Ser Val Thr Ala Ala
Asp Thr Ala Val Tyr Tyr 85 90 95 Cys Ala Arg Asn Pro Val Phe Asn
Pro Leu Thr Leu Thr His Asp Ala 100 105 110 Phe Asp Ile Trp Gly Gln
Gly Thr Met Val Thr Val Ser Ser 115 120 125 <210> SEQ ID NO
365 <211> LENGTH: 8 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKA189 CDRH1 <400> SEQUENCE: 365 Gly Phe
Thr Phe Ser Ser Tyr Ala 1 5 <210> SEQ ID NO 366 <211>
LENGTH: 8 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA189
CDRH2 <400> SEQUENCE: 366 Ile Ser Gly Ser Gly Asp Asn Thr 1 5
<210> SEQ ID NO 367 <211> LENGTH: 18 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA189 CDRH3 <400> SEQUENCE:
367 Ala Lys Trp Pro Tyr Tyr Asp Phe Trp Ser Gly Ser Glu Ser Tyr Phe
1 5 10 15 Asp Pro <210> SEQ ID NO 368 <211> LENGTH: 125
<212> TYPE: PRT
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA189 VH <400> SEQUENCE: 368
Gly Val Gln Leu Leu Glu Ser Gly Gly Ala Leu Val Gln Pro Gly Lys 1 5
10 15 Ser Leu Arg Leu Ser Cys Ala Ala Ser Gly Phe Thr Phe Ser Ser
Tyr 20 25 30 Ala Leu Thr Trp Val Arg Gln Ala Pro Gly Lys Gly Leu
Gln Trp Val 35 40 45 Ser Ala Ile Ser Gly Ser Gly Asp Asn Thr Tyr
Tyr Ala Asp Ser Val 50 55 60 Lys Gly Arg Phe Thr Ile Ser Arg Asp
Asn Ser Lys Asn Thr Leu Tyr 65 70 75 80 Leu Gln Met Asn Ser Leu Arg
Ala Glu Asp Thr Ala Val Tyr Tyr Cys 85 90 95 Ala Lys Trp Pro Tyr
Tyr Asp Phe Trp Ser Gly Ser Glu Ser Tyr Phe 100 105 110 Asp Pro Trp
Gly Gln Gly Thr Leu Val Thr Val Ser Ser 115 120 125 <210> SEQ
ID NO 369 <211> LENGTH: 8 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKA195 CDRH1 <400> SEQUENCE: 369 Gly Tyr
Asn Phe Pro Ser Tyr Trp 1 5 <210> SEQ ID NO 370 <211>
LENGTH: 8 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA195
CDRH2 <400> SEQUENCE: 370 Ile Asp Pro Ser Asp Ser Tyr Thr 1 5
<210> SEQ ID NO 371 <211> LENGTH: 15 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA195 CDRH3 <400> SEQUENCE:
371 Ala Arg Ala Asp Cys Arg Ser Thr Ser Cys Tyr Leu Val Phe Glu 1 5
10 15 <210> SEQ ID NO 372 <211> LENGTH: 121 <212>
TYPE: PRT <213> ORGANISM: Artificial Sequence <220>
FEATURE: <223> OTHER INFORMATION: ZKA195 VH <400>
SEQUENCE: 372 Glu Val Gln Leu Val Gln Ser Gly Ala Glu Val Lys Lys
Pro Gly Glu 1 5 10 15 Ser Leu Arg Ile Ser Cys Lys Asp Ser Gly Tyr
Asn Phe Pro Ser Tyr 20 25 30 Trp Ile His Trp Val Arg Gln Met Pro
Gly Lys Gly Leu Glu Trp Met 35 40 45 Gly Thr Ile Asp Pro Ser Asp
Ser Tyr Thr Asn Tyr Ser Pro Ser Phe 50 55 60 Gln Gly His Val Thr
Ile Ser Ala Asp Lys Ser Ile Ser Thr Ala Tyr 65 70 75 80 Leu Gln Trp
Ser Ser Leu Lys Ala Ser Asp Thr Ala Met Tyr Tyr Cys 85 90 95 Ala
Arg Ala Asp Cys Arg Ser Thr Ser Cys Tyr Leu Val Phe Glu Gly 100 105
110 Gln Gly Thr Leu Val Thr Val Ser Ser 115 120 <210> SEQ ID
NO 373 <211> LENGTH: 8 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKA215 CDRH1 <400> SEQUENCE: 373 Gly Tyr
Thr Phe Thr Ser Tyr Trp 1 5 <210> SEQ ID NO 374 <211>
LENGTH: 8 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKA215
CDRH2 <400> SEQUENCE: 374 Ile Asp Pro Ser Asp Ser His Thr 1 5
<210> SEQ ID NO 375 <211> LENGTH: 11 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA215 CDRH3 <400> SEQUENCE:
375 Ala Arg His Ala Leu Pro Asn Tyr Phe Asp Ser 1 5 10 <210>
SEQ ID NO 376 <211> LENGTH: 118 <212> TYPE: PRT
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA215 VH <400> SEQUENCE: 376
Glu Val Gln Leu Val Gln Ser Gly Ala Glu Val Lys Lys Pro Gly Glu 1 5
10 15 Ser Leu Arg Ile Ser Cys Lys Gly Ser Gly Tyr Thr Phe Thr Ser
Tyr 20 25 30 Trp Ile Ser Trp Val Arg Gln Met Pro Gly Lys Gly Leu
Glu Trp Met 35 40 45 Gly Arg Ile Asp Pro Ser Asp Ser His Thr Asp
Tyr Ser Pro Ser Phe 50 55 60 Gln Gly His Val Thr Ile Ser Ala Asp
Lys Ser Ile Ser Ala Ala Tyr 65 70 75 80 Leu Gln Trp Ser Ser Leu Lys
Ala Ser Asp Thr Ala Met Tyr Tyr Cys 85 90 95 Ala Arg His Ala Leu
Pro Asn Tyr Phe Asp Ser Trp Gly Gln Gly Thr 100 105 110 Leu Val Thr
Val Ser Ser 115 <210> SEQ ID NO 377 <211> LENGTH: 8
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKA218 CDRH1
<400> SEQUENCE: 377 Gly Phe Pro Phe Ser Ser Tyr Trp 1 5
<210> SEQ ID NO 378 <211> LENGTH: 8 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA218 CDRH2 <400> SEQUENCE:
378 Ile Asn Ser Asp Gly Arg Asn Thr 1 5 <210> SEQ ID NO 379
<211> LENGTH: 15 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKA218 CDRH3 <400> SEQUENCE: 379 Ala Arg Gly Gly
Tyr Asp Tyr Asp Ser Ser Gly Cys Phe Asp Tyr 1 5 10 15 <210>
SEQ ID NO 380 <211> LENGTH: 122 <212> TYPE: PRT
<213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKA218 VH <400> SEQUENCE: 380
Glu Val Gln Leu Val Glu Ser Gly Gly Gly Leu Val Gln Pro Gly Gly 1 5
10 15 Ser Leu Arg Leu Ser Cys Ala Ala Ser Gly Phe Pro Phe Ser Ser
Tyr 20 25 30 Trp Met His Trp Val Arg Gln Ala Pro Gly Lys Gly Leu
Val Trp Val 35 40 45 Ser Arg Ile Asn Ser Asp Gly Arg Asn Thr Asn
Tyr Ala Asp Ser Val 50 55 60 Lys Gly Arg Phe Thr Ile Ser Arg Asp
Asn Ala Glu Asn Thr Val Tyr 65 70 75 80 Leu Gln Met Asn Ser Leu Arg
Ala Glu Asp Thr Ala Val Tyr Tyr Cys 85 90 95 Ala Arg Gly Gly Tyr
Asp Tyr Asp Ser Ser Gly Cys Phe Asp Tyr Trp 100 105 110 Gly Gln Gly
Thr Leu Val Thr Val Ser Ser 115 120 <210> SEQ ID NO 381
<211> LENGTH: 8 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKB75 CDRH1 <400> SEQUENCE: 381 Gly Phe Thr Phe
Ser Asn Tyr Ala 1 5 <210> SEQ ID NO 382 <211> LENGTH: 8
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKB75 CDRH2
<400> SEQUENCE: 382 Ile Ser Gly Thr Gly Gly Ser Thr 1 5
<210> SEQ ID NO 383 <211> LENGTH: 25 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKB75 CDRH3 <400> SEQUENCE:
383 Ala Lys Asp Ser Ala Ser Arg Gly Gly Tyr Cys Ser Gly Gly Val Cys
1 5 10 15 Tyr Leu Asn Pro Gly His His Asp Tyr 20 25 <210> SEQ
ID NO 384 <211> LENGTH: 132 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKB75 VH <400> SEQUENCE: 384 Glu Val Gln
Val Leu Glu Ser Gly Gly Gly Leu Leu Gln Pro Gly Gly 1 5 10 15 Ser
Leu Arg Leu Ser Cys Ala Ala Ser Gly Phe Thr Phe Ser Asn Tyr 20 25
30 Ala Met Ser Trp Val Arg Gln Ala Pro Gly Lys Gly Leu Glu Trp Val
35 40 45 Ser Thr Ile Ser Gly Thr Gly Gly Ser Thr Tyr Tyr Ala Asp
Ser Val 50 55 60 Lys Gly Arg Phe Thr Ile Ser Arg Asp Asn Ser Lys
Asn Thr Leu Tyr 65 70 75 80 Leu Gln Met Asn Ser Leu Arg Ala Glu Asp
Thr Ala Val Tyr Tyr Cys 85 90 95 Ala Lys Asp Ser Ala Ser Arg Gly
Gly Tyr Cys Ser Gly Gly Val Cys 100 105 110 Tyr Leu Asn Pro Gly His
His Asp Tyr Trp Gly Gln Gly Thr Leu Val 115 120 125 Thr Val Ser Ser
130 <210> SEQ ID NO 385 <211> LENGTH: 8 <212>
TYPE: PRT <213> ORGANISM: Artificial Sequence <220>
FEATURE: <223> OTHER INFORMATION: ZKB83 CDRH1 <400>
SEQUENCE: 385 Gly Tyr Ser Phe Thr Asn Tyr Trp 1 5 <210> SEQ
ID NO 386 <211> LENGTH: 8 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKB83 CDRH2 <400> SEQUENCE: 386 Ile Asp
Pro Ser Asp Ser Tyr Thr 1 5 <210> SEQ ID NO 387 <211>
LENGTH: 27 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKB83
CDRH3 <400> SEQUENCE: 387 Ala Arg Leu Arg Gly Ser Leu Tyr Cys
Ser Gly Gly Arg Cys Tyr Ser 1 5 10 15 Val Pro Gly Glu Thr Pro Asn
Trp Phe Asp Pro 20 25 <210> SEQ ID NO 388 <211> LENGTH:
134 <212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZKB83 VH
<400> SEQUENCE: 388 Glu Val Gln Leu Val Gln Ser Gly Ala Glu
Val Lys Lys Pro Gly Glu 1 5 10 15 Ser Leu Arg Ile Ser Cys Lys Gly
Ser Gly Tyr Ser Phe Thr Asn Tyr 20 25 30 Trp Ile Thr Trp Val Arg
Gln Met Pro Gly Lys Gly Leu Glu Trp Met 35 40 45 Gly Ser Ile Asp
Pro Ser Asp Ser Tyr Thr Asn Tyr Ser Pro Ser Phe 50 55 60 Gln Gly
His Val Thr Ile Ser Ala Asp Trp Ser Ile Asn Thr Ala Tyr 65 70 75 80
Leu Gln Trp Ser Ser Leu Lys Ala Ser Asp Thr Ala Lys Tyr Tyr Cys 85
90 95 Ala Arg Leu Arg Gly Ser Leu Tyr Cys Ser Gly Gly Arg Cys Tyr
Ser 100 105 110 Val Pro Gly Glu Thr Pro Asn Trp Phe Asp Pro Trp Gly
Gln Gly Thr 115 120 125 Leu Val Thr Val Ser Ser 130 <210> SEQ
ID NO 389 <211> LENGTH: 8 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKC3 CDRH1 <400> SEQUENCE: 389 Gly Gly Ser
Ile Thr Ser Tyr Tyr 1 5 <210> SEQ ID NO 390 <211>
LENGTH: 7 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKC3
CDRH2 <400> SEQUENCE: 390 Ile Tyr Tyr Ser Gly Ser Thr 1 5
<210> SEQ ID NO 391 <211> LENGTH: 15 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKC3 CDRH3 <400> SEQUENCE: 391
Ala Arg Val Gly Gly Ala Pro Tyr Tyr Tyr Tyr Gly Met Asp Val 1 5 10
15 <210> SEQ ID NO 392 <211> LENGTH: 121 <212>
TYPE: PRT <213> ORGANISM: Artificial Sequence <220>
FEATURE: <223> OTHER INFORMATION: ZKC3 VH <400>
SEQUENCE: 392 Gln Val Gln Leu Gln Glu Ser Gly Pro Gly Leu Val Lys
Pro Ser Glu 1 5 10 15 Thr Leu Ser Leu Thr Cys Thr Val Ser Gly Gly
Ser Ile Thr Ser Tyr 20 25 30 Tyr Trp Ser Trp Ile Arg Gln Pro Pro
Gly Lys Gly Leu Glu Trp Ile 35 40 45 Gly Tyr Ile Tyr Tyr Ser Gly
Ser Thr Asn Tyr Asn Pro Ser Leu Lys 50 55 60 Ser Arg Val Thr Ile
Ser Val Asp Thr Ser Lys Asn Gln Phe Ser Leu 65 70 75 80 Lys Leu Ser
Ser Val Thr Ala Ala Asp Thr Ala Val Tyr Tyr Cys Ala 85 90 95 Arg
Val Gly Gly Ala Pro Tyr Tyr Tyr Tyr Gly Met Asp Val Trp Gly 100 105
110 Gln Gly Thr Thr Val Thr Val Ser Ser 115 120 <210> SEQ ID
NO 393 <211> LENGTH: 8 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKC18 CDRH1 <400> SEQUENCE: 393 Gly Phe
Thr Phe Gly Asp Tyr Ala 1 5 <210> SEQ ID NO 394 <211>
LENGTH: 10 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKC18
CDRH2 <400> SEQUENCE: 394
Ile Arg Ser Lys Ala Tyr Gly Gly Thr Thr 1 5 10 <210> SEQ ID
NO 395 <211> LENGTH: 13 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKC18 CDRH3 <400> SEQUENCE: 395 Ser Arg
Asp His Thr Gly Thr Thr Tyr Ala Phe Asp Ile 1 5 10 <210> SEQ
ID NO 396 <211> LENGTH: 122 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZKC18 VH <400> SEQUENCE: 396 Glu Val Gln
Leu Val Glu Ser Gly Gly Gly Leu Val Gln Pro Gly Arg 1 5 10 15 Ser
Leu Arg Leu Ser Cys Thr Ala Ser Gly Phe Thr Phe Gly Asp Tyr 20 25
30 Ala Met Ser Trp Phe Arg Gln Ala Pro Gly Lys Gly Leu Glu Trp Val
35 40 45 Gly Phe Ile Arg Ser Lys Ala Tyr Gly Gly Thr Thr Glu Tyr
Ala Ala 50 55 60 Ser Val Lys Gly Arg Phe Thr Ile Ser Arg Asp Asp
Ser Lys Ser Ile 65 70 75 80 Ala Tyr Leu Gln Met Asn Ser Leu Lys Thr
Glu Asp Thr Ala Val Tyr 85 90 95 Tyr Cys Ser Arg Asp His Thr Gly
Thr Thr Tyr Ala Phe Asp Ile Trp 100 105 110 Gly Gln Gly Thr Met Val
Thr Val Ser Gln 115 120 <210> SEQ ID NO 397 <211>
LENGTH: 8 <212> TYPE: PRT <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: ZKD1
CDRH1 <400> SEQUENCE: 397 Gly Phe Thr Phe Ser Ser Tyr Gly 1 5
<210> SEQ ID NO 398 <211> LENGTH: 8 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZKD1 CDRH2 <400> SEQUENCE: 398
Ile Trp Tyr Asp Gly Ser Asn Lys 1 5 <210> SEQ ID NO 399
<211> LENGTH: 19 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZKD1 CDRH3 <400> SEQUENCE: 399 Ala Arg Asp Arg
Arg Gly Tyr Gly Asp Tyr Val Gly Tyr Tyr Tyr Gly 1 5 10 15 Met Asp
Val <210> SEQ ID NO 400 <211> LENGTH: 126 <212>
TYPE: PRT <213> ORGANISM: Artificial Sequence <220>
FEATURE: <223> OTHER INFORMATION: ZKD1 VH <400>
SEQUENCE: 400 Gln Val Gln Leu Val Glu Ser Gly Gly Gly Val Val Gln
Pro Gly Arg 1 5 10 15 Ser Leu Arg Leu Ser Cys Ala Ala Ser Gly Phe
Thr Phe Ser Ser Tyr 20 25 30 Gly Met His Trp Val Arg Gln Ala Pro
Gly Lys Gly Leu Glu Trp Val 35 40 45 Ala Val Ile Trp Tyr Asp Gly
Ser Asn Lys Tyr Tyr Ala Asp Ser Val 50 55 60 Lys Gly Arg Phe Thr
Ile Ser Arg Asp Asn Ser Lys Asn Thr Leu Tyr 65 70 75 80 Leu Gln Met
Asn Ser Leu Arg Ala Glu Asp Thr Ala Val Tyr Tyr Cys 85 90 95 Ala
Arg Asp Arg Arg Gly Tyr Gly Asp Tyr Val Gly Tyr Tyr Tyr Gly 100 105
110 Met Asp Val Trp Gly Gln Gly Thr Thr Val Thr Val Ser Ser 115 120
125 <210> SEQ ID NO 401 <211> LENGTH: 95 <212>
TYPE: PRT <213> ORGANISM: Artificial Sequence <220>
FEATURE: <223> OTHER INFORMATION: Zika Virus EDIII
<220> FEATURE: <221> NAME/KEY: misc_feature <222>
LOCATION: (9)..(9) <223> OTHER INFORMATION: Xaa can be any
naturally occurring amino acid <220> FEATURE: <221>
NAME/KEY: misc_feature <222> LOCATION: (13)..(14) <223>
OTHER INFORMATION: Xaa can be any naturally occurring amino acid
<220> FEATURE: <221> NAME/KEY: misc_feature <222>
LOCATION: (22)..(22) <223> OTHER INFORMATION: Xaa can be any
naturally occurring amino acid <220> FEATURE: <221>
NAME/KEY: misc_feature <222> LOCATION: (25)..(25) <223>
OTHER INFORMATION: Xaa can be any naturally occurring amino acid
<220> FEATURE: <221> NAME/KEY: misc_feature <222>
LOCATION: (27)..(27) <223> OTHER INFORMATION: Xaa can be any
naturally occurring amino acid <220> FEATURE: <221>
NAME/KEY: misc_feature <222> LOCATION: (33)..(33) <223>
OTHER INFORMATION: Xaa can be any naturally occurring amino acid
<220> FEATURE: <221> NAME/KEY: misc_feature <222>
LOCATION: (35)..(35) <223> OTHER INFORMATION: Xaa can be any
naturally occurring amino acid <220> FEATURE: <221>
NAME/KEY: misc_feature <222> LOCATION: (41)..(41) <223>
OTHER INFORMATION: Xaa can be any naturally occurring amino acid
<220> FEATURE: <221> NAME/KEY: misc_feature <222>
LOCATION: (60)..(60) <223> OTHER INFORMATION: Xaa can be any
naturally occurring amino acid <220> FEATURE: <221>
NAME/KEY: misc_feature <222> LOCATION: (83)..(83) <223>
OTHER INFORMATION: Xaa can be any naturally occurring amino acid
<220> FEATURE: <221> NAME/KEY: misc_feature <222>
LOCATION: (85)..(85) <223> OTHER INFORMATION: Xaa can be any
naturally occurring amino acid <400> SEQUENCE: 401 Thr Ala
Ala Phe Thr Phe Thr Lys Xaa Pro Ala Glu Xaa Xaa His Gly 1 5 10 15
Thr Val Thr Val Glu Xaa Gln Tyr Xaa Gly Xaa Asp Gly Pro Cys Lys 20
25 30 Xaa Pro Xaa Gln Met Ala Val Asp Xaa Gln Thr Leu Thr Pro Val
Gly 35 40 45 Arg Leu Ile Thr Ala Asn Pro Val Ile Thr Glu Xaa Thr
Glu Asn Ser 50 55 60 Lys Met Met Leu Glu Leu Asp Pro Pro Phe Gly
Asp Ser Tyr Ile Val 65 70 75 80 Ile Gly Xaa Gly Xaa Lys Lys Ile Thr
His His Trp His Arg Ser 85 90 95 <210> SEQ ID NO 402
<211> LENGTH: 95 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZIKAV H PF 2013 EDIII <400> SEQUENCE: 402 Thr
Ala Ala Phe Thr Phe Thr Lys Ile Pro Ala Glu Thr Leu His Gly 1 5 10
15 Thr Val Thr Val Glu Val Gln Tyr Ala Gly Thr Asp Gly Pro Cys Lys
20 25 30 Val Pro Ala Gln Met Ala Val Asp Met Gln Thr Leu Thr Pro
Val Gly 35 40 45 Arg Leu Ile Thr Ala Asn Pro Val Ile Thr Glu Ser
Thr Glu Asn Ser 50 55 60 Lys Met Met Leu Glu Leu Asp Pro Pro Phe
Gly Asp Ser Tyr Ile Val 65 70 75 80 Ile Gly Val Gly Glu Lys Lys Ile
Thr His His Trp His Arg Ser 85 90 95 <210> SEQ ID NO 403
<211> LENGTH: 20 <212> TYPE: DNA <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZIKV-NS1 forward primer <400> SEQUENCE: 403
tggagttcaa ctgacggtcg 20 <210> SEQ ID NO 404
<211> LENGTH: 20 <212> TYPE: DNA <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZIKV-NS1-reverse primer <400> SEQUENCE: 404
taccccgaac ccatgatcct 20 <210> SEQ ID NO 405 <211>
LENGTH: 21 <212> TYPE: DNA <213> ORGANISM: Artificial
Sequence <220> FEATURE: <223> OTHER INFORMATION: Gapdh-
forward primer <400> SEQUENCE: 405 ggcaagttca aaggcacagt c 21
<210> SEQ ID NO 406 <211> LENGTH: 19 <212> TYPE:
DNA <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: Gapdh-reverse primer <400>
SEQUENCE: 406 caccagcatc accccattt 19 <210> SEQ ID NO 407
<211> LENGTH: 104 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZIKV EDIII generic <220> FEATURE: <221>
NAME/KEY: misc_feature <222> LOCATION: (1)..(1) <223>
OTHER INFORMATION: X may be any (naturally occurring) amino acid,
preferably K, A, or E <220> FEATURE: <221> NAME/KEY:
misc_feature <222> LOCATION: (3)..(3) <223> OTHER
INFORMATION: X may be any (naturally occurring) amino acid,
preferably V, F, or L <220> FEATURE: <221> NAME/KEY:
misc_feature <222> LOCATION: (4)..(4) <223> OTHER
INFORMATION: X may be any (naturally occurring) amino acid,
preferably S or F <220> FEATURE: <221> NAME/KEY:
misc_feature <222> LOCATION: (17)..(17) <223> OTHER
INFORMATION: X may be any (naturally occurring) amino acid,
preferably I or V <220> FEATURE: <221> NAME/KEY:
misc_feature <222> LOCATION: (21)..(21) <223> OTHER
INFORMATION: X may be any (naturally occurring) amino acid,
preferably T or V <220> FEATURE: <221> NAME/KEY:
misc_feature <222> LOCATION: (22)..(22) <223> OTHER
INFORMATION: X may be any (naturally occurring) amino acid,
preferably L or D <220> FEATURE: <221> NAME/KEY:
misc_feature <222> LOCATION: (30)..(30) <223> OTHER
INFORMATION: X may be any (naturally occurring) amino acid,
preferably V or G <220> FEATURE: <221> NAME/KEY:
misc_feature <222> LOCATION: (33)..(33) <223> OTHER
INFORMATION: X may be any (naturally occurring) amino acid,
preferably A or G <220> FEATURE: <221> NAME/KEY:
misc_feature <222> LOCATION: (35)..(35) <223> OTHER
INFORMATION: X may be any (naturally occurring) amino acid except
R, preferably T or A <220> FEATURE: <221> NAME/KEY:
misc_feature <222> LOCATION: (41)..(41) <223> OTHER
INFORMATION: X may be any (naturally occurring) amino acid,
preferably V or I <220> FEATURE: <221> NAME/KEY:
misc_feature <222> LOCATION: (43)..(43) <223> OTHER
INFORMATION: X may be any (naturally occurring) amino acid,
preferably A or V <220> FEATURE: <221> NAME/KEY:
misc_feature <222> LOCATION: (49)..(49) <223> OTHER
INFORMATION: X may be any (naturally occurring) amino acid,
preferably M or T <220> FEATURE: <221> NAME/KEY:
misc_feature <222> LOCATION: (68)..(68) <223> OTHER
INFORMATION: X may be any (naturally occurring) amino acid,
preferably S or G <220> FEATURE: <221> NAME/KEY:
misc_feature <222> LOCATION: (70)..(70) <223> OTHER
INFORMATION: X may be any (naturally occurring) amino acid,
preferably E or K <220> FEATURE: <221> NAME/KEY:
misc_feature <222> LOCATION: (91)..(91) <223> OTHER
INFORMATION: X may be any (naturally occurring) amino acid,
preferably V or I <220> FEATURE: <221> NAME/KEY:
misc_feature <222> LOCATION: (93)..(93) <223> OTHER
INFORMATION: X may be any (naturally occurring) amino acid,
preferably E, A, K, or D <220> FEATURE: <221> NAME/KEY:
misc_feature <222> LOCATION: (94)..(94) <223> OTHER
INFORMATION: X may be any (naturally occurring) amino acid,
preferably E, A, or K, more preferably K or A <400> SEQUENCE:
407 Xaa Gly Xaa Xaa Tyr Ser Leu Cys Thr Ala Ala Phe Thr Phe Thr Lys
1 5 10 15 Xaa Pro Ala Glu Xaa Xaa His Gly Thr Val Thr Val Glu Xaa
Gln Tyr 20 25 30 Xaa Gly Xaa Asp Gly Pro Cys Lys Xaa Pro Xaa Gln
Met Ala Val Asp 35 40 45 Xaa Gln Thr Leu Thr Pro Val Gly Arg Leu
Ile Thr Ala Asn Pro Val 50 55 60 Ile Thr Glu Xaa Thr Xaa Asn Ser
Lys Met Met Leu Glu Leu Asp Pro 65 70 75 80 Pro Phe Gly Asp Ser Tyr
Ile Val Ile Gly Xaa Gly Xaa Xaa Lys Ile 85 90 95 Thr His His Trp
His Arg Ser Gly 100 <210> SEQ ID NO 408 <211> LENGTH:
19 <212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: chimeric hinge
<400> SEQUENCE: 408 Glu Ser Lys Tyr Gly Pro Pro Cys Pro Pro
Cys Pro Ala Pro Pro Val 1 5 10 15 Ala Gly Pro <210> SEQ ID NO
409 <211> LENGTH: 95 <212> TYPE: PRT <213>
ORGANISM: Artificial Sequence <220> FEATURE: <223>
OTHER INFORMATION: ZIKV MR766 (AY632535) EDIII <400>
SEQUENCE: 409 Thr Ala Ala Phe Thr Phe Thr Lys Val Pro Ala Glu Thr
Leu His Gly 1 5 10 15 Thr Val Thr Val Glu Val Gln Tyr Ala Gly Thr
Asp Gly Pro Cys Lys 20 25 30 Leu Pro Val Gln Met Ala Val Asp Met
Gln Thr Leu Thr Pro Val Gly 35 40 45 Arg Leu Ile Thr Ala Asn Pro
Val Ile Thr Glu Ser Thr Glu Asn Ser 50 55 60 Lys Met Met Leu Glu
Leu Asp Pro Pro Phe Gly Asp Ser Tyr Ile Val 65 70 75 80 Ile Gly Val
Gly Asp Lys Lys Ile Thr His His Trp His Arg Ser 85 90 95
<210> SEQ ID NO 410 <211> LENGTH: 95 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZIKV KU870645 (USA) 02-Feb-2016
EDIII <400> SEQUENCE: 410 Thr Ala Ala Phe Thr Phe Thr Lys Ile
Pro Ala Glu Thr Leu His Gly 1 5 10 15 Thr Val Thr Val Glu Val Gln
Tyr Ala Gly Thr Asp Gly Pro Cys Lys 20 25 30 Val Pro Ala Gln Met
Ala Val Asp Met Gln Thr Leu Thr Pro Val Gly 35 40 45 Arg Leu Ile
Thr Ala Asn Pro Val Ile Thr Glu Ser Thr Glu Asn Ser 50 55 60 Lys
Met Met Leu Glu Leu Asp Pro Pro Phe Gly Asp Ser Tyr Ile Val 65 70
75 80 Ile Gly Leu Gly Glu Lys Lys Ile Thr His His Trp His Arg Ser
85 90 95 <210> SEQ ID NO 411 <211> LENGTH: 95
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZIKV KU729218
(Brazil) 2015 EDIII <400> SEQUENCE: 411 Thr Ala Ala Phe Thr
Phe Thr Lys Ile Pro Ala Glu Thr Leu His Gly 1 5 10 15
Thr Val Thr Val Glu Val Gln Tyr Ala Gly Thr Asp Gly Pro Cys Lys 20
25 30 Val Pro Ala Gln Met Ala Val Asp Thr Gln Thr Leu Thr Pro Val
Gly 35 40 45 Arg Leu Ile Thr Ala Asn Pro Val Ile Thr Glu Ser Thr
Glu Asn Ser 50 55 60 Lys Met Met Leu Glu Leu Asp Pro Pro Phe Gly
Asp Ser Tyr Ile Val 65 70 75 80 Ile Gly Val Gly Glu Lys Lys Ile Thr
His His Trp His Arg Ser 85 90 95 <210> SEQ ID NO 412
<211> LENGTH: 95 <212> TYPE: PRT <213> ORGANISM:
Artificial Sequence <220> FEATURE: <223> OTHER
INFORMATION: ZIKV KU681081 (Thailand) 9-Jul-14 EDIII <400>
SEQUENCE: 412 Thr Ala Ala Phe Thr Phe Thr Lys Ile Pro Ala Glu Thr
Leu His Gly 1 5 10 15 Thr Val Thr Val Glu Val Gln Tyr Ala Gly Thr
Asp Gly Pro Cys Lys 20 25 30 Val Pro Ala Gln Met Ala Val Asp Met
Gln Thr Leu Thr Pro Val Gly 35 40 45 Arg Leu Ile Thr Ala Asn Pro
Val Ile Thr Glu Gly Thr Glu Asn Ser 50 55 60 Lys Met Met Leu Glu
Leu Asp Pro Pro Phe Gly Asp Ser Tyr Ile Val 65 70 75 80 Ile Gly Val
Gly Glu Lys Lys Ile Thr His His Trp His Arg Ser 85 90 95
<210> SEQ ID NO 413 <211> LENGTH: 95 <212> TYPE:
PRT <213> ORGANISM: Artificial Sequence <220> FEATURE:
<223> OTHER INFORMATION: ZIKV KU926310 (Brazil) 29-Jan-2016
EDIII <400> SEQUENCE: 413 Thr Ala Ala Phe Thr Phe Thr Lys Ile
Pro Ala Glu Thr Leu His Gly 1 5 10 15 Thr Val Thr Val Glu Val Gln
Tyr Ala Gly Ala Asp Gly Pro Cys Lys 20 25 30 Val Pro Ala Gln Met
Ala Val Asp Met Gln Thr Leu Thr Pro Val Gly 35 40 45 Arg Leu Ile
Thr Ala Asn Pro Val Ile Thr Glu Ser Thr Glu Asn Ser 50 55 60 Lys
Met Met Leu Glu Leu Asp Pro Pro Phe Gly Asp Ser Tyr Ile Val 65 70
75 80 Ile Gly Val Gly Glu Lys Lys Ile Thr His His Trp His Arg Ser
85 90 95 <210> SEQ ID NO 414 <211> LENGTH: 95
<212> TYPE: PRT <213> ORGANISM: Artificial Sequence
<220> FEATURE: <223> OTHER INFORMATION: ZIKV KU744693
(China) 6-Feb-16 EDIII <400> SEQUENCE: 414 Thr Ala Ala Phe
Thr Phe Thr Lys Ile Pro Ala Glu Thr Val Asp Gly 1 5 10 15 Thr Val
Thr Val Glu Gly Gln Tyr Gly Gly Thr Asp Gly Pro Cys Lys 20 25 30
Val Pro Ala Gln Met Ala Val Asp Met Gln Thr Leu Thr Pro Val Gly 35
40 45 Arg Leu Ile Thr Ala Asn Pro Val Ile Thr Glu Ser Thr Glu Asn
Ser 50 55 60 Lys Met Met Leu Glu Leu Asp Pro Pro Phe Gly Asp Ser
Tyr Ile Val 65 70 75 80 Ile Gly Val Gly Glu Lys Lys Ile Thr His His
Trp His Arg Ser 85 90 95
* * * * *
References