U.S. patent application number 11/364978 was filed with the patent office on 2009-05-07 for surface-enhanced raman nanobiosensor.
This patent application is currently assigned to Northwestern University. Invention is credited to Matthew R. Glucksberg, Christy L. Haynes, Olga Lyandres, Karen E. Pettier, Nilam C. Shah, Douglas A. Stuart, Richard P. Van Duyne, Joseph T. Walsh, Chanda Ranjit Yonzon, Jonathan M. Yuen.
Application Number | 20090118605 11/364978 |
Document ID | / |
Family ID | 40588839 |
Filed Date | 2009-05-07 |
United States Patent
Application |
20090118605 |
Kind Code |
A1 |
Van Duyne; Richard P. ; et
al. |
May 7, 2009 |
Surface-enhanced raman nanobiosensor
Abstract
The present invention relates to biosensors, in particular to
surface-enhanced Raman biosensors for detection of in vivo and ex
vivo analytes. In particular, the present invention provides
compositions and methods for the in vivo detection of analytes such
as glucose.
Inventors: |
Van Duyne; Richard P.;
(Wilmette, IL) ; Glucksberg; Matthew R.;
(Evanston, IL) ; Pettier; Karen E.; (Lawrence,
KS) ; Haynes; Christy L.; (Minneapolis, MN) ;
Walsh; Joseph T.; (Evanston, IL) ; Yonzon; Chanda
Ranjit; (Springfield, NJ) ; Shah; Nilam C.;
(Chicago, IL) ; Lyandres; Olga; (Chicago, IL)
; Stuart; Douglas A.; (Downers Grove, IL) ; Yuen;
Jonathan M.; (Chicago, IL) |
Correspondence
Address: |
Casimir Jones, S.C.
440 Science Drive, Suite 203
Madison
WI
53711
US
|
Assignee: |
Northwestern University
Evanston
IL
|
Family ID: |
40588839 |
Appl. No.: |
11/364978 |
Filed: |
March 1, 2006 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
10652280 |
Aug 29, 2003 |
|
|
|
11364978 |
|
|
|
|
60407061 |
Aug 30, 2002 |
|
|
|
Current U.S.
Class: |
600/365 ;
977/920 |
Current CPC
Class: |
G01N 33/54346 20130101;
A61B 5/14735 20130101; G01N 33/54373 20130101; A61B 5/14532
20130101; B82Y 30/00 20130101; A61B 5/14546 20130101; G01N 21/658
20130101; A61B 2562/0285 20130101; B82Y 15/00 20130101 |
Class at
Publication: |
600/365 ;
977/920 |
International
Class: |
A61B 5/145 20060101
A61B005/145 |
Goverment Interests
[0002] This invention was made with government support under N.I.H.
grants EY13002, 13015, and DK6990-01A1, National Science Foundation
grants CHE0414554, EEC-0118025 and DMR-0076097, the Air Force
Office of Scientific Research MURI program grant F49620-02-1-0381,
and the U.S. Army Medical Research and Materiel Command grant
W81XWH-04-1-0630.
[0003] The government may have certain rights in the invention.
Claims
1. A biosensor, comprising: a) a substrate; and b) a plurality of
nanobiosensors adherent to said substrate, comprising: a plurality
of nanospheres; a metal film over said nanospheres; and a
self-assembled partition layer formed on the surface of said metal
film over said nanospheres comprising at least two compounds,
wherein said nanobiosensors are configured for the quantitative
detection of an analyte such that the spectrum of a
surface-enhanced Raman scattering signal detected from said
plurality of nanobiosensors in the presence of said analyte is
correlated with the concentration of said analyte in a medium.
2. The biosensor of claim 1, wherein said substrate is copper or
silicon dioxide.
3. The biosensor of claim 1, wherein said substrate is configured
to provide a plurality of nanowells containing said
nanospheres.
4. The nanobiosensor of claim 1, wherein said nanospheres comprise
polystyrene or silica nanospheres.
5. The nanobiosensor of claim 1, wherein said self-assembled
partition layer comprises a hydrophilic compound and a hydrophobic
compound.
6. The self-assembled partition layer of claim 5, wherein said
hydrophilic compound and said hydrophobic compound comprise
modified alkanes.
7. The modified alkanes of claim 6, wherein said alkanes comprise a
chain length of at least 5 carbon atoms.
8. The self-assembled partition layer of claim 5, wherein said
hydrophobic compound comprises decanethiol and said hydrophilic
compound comprises mercaptohexanol.
9. The biosensor of claim 1, further comprising a receptor specific
for said analyte, wherein said receptor is configured to bind
reversibly to said analyte.
10. The biosensor of claim 1, wherein said analyte is glucose.
11-20. (canceled)
21. The biosensor of claim 1, wherein said analyte is lactic acid
or lactate.
22. The biosensor of claim 1, wherein said analyte is a chemical
warfare agent.
23. A biosensor, comprising: a) a substrate; and b) a plurality of
nanobiosensors adherent to said substrate, comprising: a plurality
of nanospheres; a metal film over said nanospheres; and a
self-assembled partition layer formed on the surface of said metal
film over said nanospheres comprising at least two compounds,
wherein said nanobiosensors are configured for the quantitative
detection of lactic acid or lactate such that the spectrum of a
surface-enhanced Raman scattering signal detected from said
plurality of nanobiosensors in the presence of said lactic acid or
lactate is correlated with the concentration of said lactic acid or
lactate in a medium.
24. The biosensor of claim 23, wherein said self-assembled
partition layer comprises decanethiol and mercaptohexanol.
25. The biosensor of claim 23, wherein said biosensor is configured
for use in vivo.
26-32. (canceled)
Description
[0001] This application is a continuation in part of co-pending
U.S. patent application Ser. No. 10/652,280, filed Aug. 29, 2003,
which in turn claims priority to U.S. Provisional Patent
Application Ser. No. 60/407,061, filed Aug. 30, 2002, both of which
are herein incorporated by reference in their entireties.
FIELD OF THE INVENTION
[0004] The present invention relates to biosensors, in particular
to surface-enhanced Raman biosensors for detection of analytes.
BACKGROUND
[0005] In diabetes mellitus, the body either fails to produce or to
respond to insulin, which regulates glucose metabolism, resulting
in large fluctuations in glucose levels. These fluctuations can
cause a range of secondary complications, including kidney disease,
heart disease, blindness, nerve damage, and gangrene. Current
treatment of diabetes consists of self-regulation of blood glucose
levels through frequent monitoring and a combination of diet,
medication, and insulin injection, depending on the type of
diabetes. Most patients measure their glucose levels by withdrawing
small samples of blood using a "finger-stick" apparatus followed by
electrochemical detection of an oxidation product of glucose. This
type of measurement is both painful and inconvenient. As a result,
many patients fail to adequately monitor their glucose levels,
risking secondary complications. A faster, easier, and less painful
method for frequently measuring glucose levels would be of great
individual, clinical, and societal benefit. Continuous monitoring
of blood glucose would open the door to feedback control of
implanted insulin pumps. In fact, reliable and robust sensor
technology is the single stumbling block in an artificial
pancreas.
SUMMARY OF THE INVENTION
[0006] The present invention relates to biosensors, in particular
to surface-enhanced Raman biosensors for detection of analytes.
[0007] Accordingly, in some embodiments, the present invention
provides a composition comprising a plurality of nanobiosensors,
the nanobiosensors configured for surface enhanced Raman
spectroscopy detection of an analyte. In some embodiments, the
nanobiosensors are coated with a noble metal (e.g., silver, gold,
platinum, etc. and combinations thereof). In some embodiments, the
nanobiosensors are configured for quantitative detection of the
analyte. In some embodiments, the nanobiosensors are configured for
use in vivo (e.g., including, but not limited to, implantation of
the nanobiosensor under the skin or in the eye). In some
embodiments, the nanobiosensors comprise a biocompatible coating.
In some embodiments, the nanobiosensors are configured for
detection of an analyte in a bodily fluid. In some embodiments, the
analyte is glucose. In some embodiments, the analyte is selected
from the group consisting of ascorbate, lactic acid, urea,
pesticides, chemical warfare agents, pollutants, and explosives,
although the systems may be used for the detection of any type of
analyte. In some embodiments, the nanobiosensors further comprise a
surface bound reversibly-binding analyte receptor, the receptor
specific for the analyte of interest. In some embodiments, the
analyte is glucose and the reversibly-binding receptor is
concanavalin A.
[0008] In other embodiments, the nanobiosensors further comprise a
self-assembled monolayer formed on the surface of the
nanobiosensors. In some embodiments, the self-assembled monolayer
is selected from the group consisting of 4-aminothiophenol,
L-cystein, 3-mercaptopropionic acid, 11-mercaptoundecanoic acid,
1-hexanethiol, 1-octanethiol, 1-decanethiol (1-DT),
1-hexadecanethiol, mercaptoethanol, poly-DL-lysine,
3-mercapto-1-propanesulfonic acid, benzenethiol, and
cyclohexylmercaptan. In other embodiments, the self-assembled
monolayer is a combination of two or more components. In some
preferred embodiments, the self-assembled monolayer is 1-DT. In
other embodiment, the self-assembled monolayer is
(1-mercaptoundeca-11-yl) tri(ethylene glycol)
(HS(CH.sub.2).sub.11(OCH.sub.2CH.sub.2).sub.3OH. In some
embodiments, the nanobiosensors are embedded in nanowells. In some
embodiments, the nanowells are fabricated out of silica.
[0009] In some embodiments, the nanobiosensors are configured for
quantitative detection of glucose or other analytes in a
physiological concentration range (e.g., 0-450 mg/dL). In some
particularly preferred embodiments, the nanobiosensors are
configured for detection of the analyte for at least 3 days. In
some embodiments, the nanobiosensors are configured for reversible
detection of the analyte. In certain embodiments, the
nanobiosensors are configured for detection of the analyte in the
presence of interfering molecules, for example, proteins.
[0010] The present invention further provides a kit comprising a
plurality of nanobiosensors, the nanobiosensors configured for
surface-enhanced Raman spectroscopy detection of an analyte.
[0011] The present invention also provides a system, comprising a
plurality of nanobiosensors, the nanobiosensors configured for
surface enhanced Raman spectroscopy detection of an analyte; and a
device configured for carrying out the surface-enhanced Raman
spectroscopy detection of the analyte. In some embodiments, the
device comprises delivery and collection optics, a laser source, a
notch filter, and a detector. In some embodiments, the delivery and
collection optics and the notch filter are incorporated into a
fiber optic probe. In some embodiments, the fiber optic probe is in
communication with the laser source and the detector. In some
embodiments, the system further comprises a second device
configured for the delivery of insulin or other agents to a
subject.
[0012] The present invention additionally provides a method for
detection of an analyte, comprising providing a plurality of
nanobiosensors, the nanobiosensors configured for surface-enhanced
Raman spectroscopy detection of an analyte; and a device configured
for the surface-enhanced Raman spectroscopy detection of the
analyte; and contacting the plurality of nanobiosensors with a
bodily fluid comprising the analyte; and detecting a
surface-enhanced Raman signal from the nanobiosensor using the
device. In some embodiments, the level of the surface-enhanced
Raman signal is correlated with the concentration of the analyte in
the bodily fluid. In some embodiments, the detecting is in vivo. In
some embodiments, the nanobiosensors are implanted under the skin.
In other embodiments, the nanobiosensors are implanted in an
eye.
[0013] The present invention further provides a composition
comprising a fiber optic tip coated with a plurality of
nanobiosensors configured for surface-enhanced Raman spectroscopic
detection of glucose. In some embodiments, the nanobiosensors are
configured for use in vivo, for example under the skin.
[0014] In some embodiments, the present invention provides a
composition comprising a biosensor comprising a substrate, and a
plurality of nanobiosensors adherent to the substrate, comprising:
a plurality of nanospheres; a metal film over nanospheres (MFON);
and a self-assembled partition layer formed on the surface of the
metal film over nanospheres comprising at least two compounds,
wherein the nanobiosensors are configured for the quantitative
detection of an analyte such that the spectrum of a
surface-enhanced Raman signal detected from the plurality of
nanobiosensors in the presence of the analyte is correlated with
the concentration of the analyte in a medium. In further
embodiments, the substrate is copper or silicon dioxide. In some
embodiments, the substrate is configured to provide a plurality of
nanowells containing the nanospheres. In other embodiments, the
nanospheres comprise polystyrene or silica nanospheres. In further
embodiments, the self-assembled partition layer comprises a
hydrophilic compound and a hydrophobic compound. In still further
embodiments, the hydrophilic compound and the hydrophobic compound
comprise modified alkanes. In preferred embodiments, the alkanes
comprise a chain length of at least 5 carbon atoms. In particularly
preferred embodiments, the hydrophobic compound comprises
decanethiol and the hydrophilic compound comprises mercaptohexanol.
Other embodiments further comprise a receptor specific for the
analyte, wherein the receptor is configured to bind to the analyte
reversibly. In a preferred embodiment the analyte is glucose.
[0015] In some embodiments, the present invention provides a method
for detection of an analyte, comprising providing a biosensor,
comprising a substrate, and a plurality of nanobiosensors adherent
to the substrate, comprising: a plurality of nanospheres; a metal
film over the nanospheres (MFON); and a self-assembled partition
layer formed on the surface of the metal film over the nanospheres
comprising at least two compounds, wherein the nanobiosensors are
configured for the quantitative detection of an analyte such that
the spectrum of a surface-enhanced Raman signal obtained from the
plurality of the nanobiosensors in the presence of the analyte is
correlated with the concentration of the analyte in a medium; and a
device configured for the surface-enhanced Raman spectroscopy
detection of the analyte; a medium comprising the analyte; and
contacting the biosensor with the medium; and detecting a
surface-enhanced Raman signal from the biosensor using the device.
In some embodiments, the nanospheres comprise polystyrene or silica
nanospheres. In other embodiments, the self-assembled partition
layer comprises a hydrophilic compound and a hydrophobic compound.
In further embodiments, the hydrophilic compound and the
hydrophobic compound comprise modified alkanes. In still further
embodiments, the alkanes comprise a chain length of at least 5
carbon atoms. In preferred embodiments the hydrophobic compound
comprises decanethiol and the hydrophilic compound comprises
mercaptohexanol. In another embodiment, the biosensor comprises a
receptor specific for the analyte, wherein the receptor is
configured to bind to the analyte reversibly. In a preferred
embodiment, the analyte is glucose. In a particularly preferred
embodiment the detecting is in vivo detecting, and the biosensor is
implanted in a subject.
DESCRIPTION OF THE FIGURES
[0016] FIG. 1 shows the size and shape of tunable LSPR spectra of
Ag nanoparticles fabricated by NSL in some embodiments of the
present invention.
[0017] FIG. 2 depicts exemplary steps in nanosphere lithography.
FIG. 2A shows a depiction of a nanosphere monolayer and FIG. 2B
shows an atomic force micrograph of the resulting nanoparticle
array.
[0018] FIG. 3 shows FT-Raman spectra of the major components of the
aqueous humor.
[0019] FIG. 4 shows a schematic depicting SERS sensing modality
with embedded nanoparticle substrate.
[0020] FIG. 5 shows spectrum of 1-DT (FIG. 5A) subtracted from
combined 1-DT and glucose spectrum (FIG. 5B) to produce spectrum of
glucose (FIG. 5C). *indicates glucose peaks
[0021] FIG. 6 depicts exemplary steps in nanowell formation. FIG.
6A shows a schematic of nanowell fabrication and FIG. 6B shows an
atomic force micrograph of a nanowell structure.
[0022] FIG. 7 shows a LSPR spectra of Ag nanoparticles embedded in
SiO.sub.2 nanowells of varying depths (30 to 300 nm).
[0023] FIG. 8 shows a partial least-squares leave-one-out
prediction of glucose concentration versus actual concentration
based on measurements made from silver SERS substrate coated with a
single monolayer of 1-octanethiol. Primary peak used for prediction
is 1121 cm.sup.-1. The root-mean-squared error of prediction is 2.5
mM.
[0024] FIG. 9 shows a schematic of nanoparticles embedded in wells
and coated with capture layer to increase analyte interaction with
the nanoparticles.
[0025] FIG. 10 shows a schematic showing placement of eye and skin
implants.
[0026] FIG. 11 shows hypothetical glucose concentration gradient
created by 1-DT capture layer.
[0027] FIG. 12 shows spectra used in quantitative analysis. FIG.
12A shows a 1-DT monolayer on AgFON substrate, .lamda..sub.ex=532
nm, P=1.25 mW, acquisition time (t)=30 seconds. FIG. 12B shows a
mixture of 1-DT monolayer and glucose partitioned from a 100 mM
solution, .lamda..sub.ex=532 nm, P=1.25 mW, acquisition time (t)=30
seconds. FIG. 12C shows residual glucose spectrum produced by
subtracting FIG. 12A from FIG. 12B. FIG. 12D shows normal Raman
spectrum of crystalline glucose for comparison,
.lamda..sub.ex=632.8 nm, P=5 mW, acquisition time (t)=30
seconds.
[0028] FIG. 13 shows a plot of partial least-squares predicted
glucose concentrations versus actual glucose concentrations using
leave-on-out cross-validation (21 loading vectors). Each micro-SERS
measurement was made under ambient conditions, using
.lamda..sub.ex=632.8 nm (P=4.7 mW, acquisition time (t)=90 sec).
The dashed line represents perfect predictions. The inset shows the
root-mean-squared error of calibration as a function of number of
loading vectors used in the PLS algorithm.
[0029] FIG. 14 shows a plot of partial least-squares predicted
physiologically-relevant glucose concentrations versus actual
glucose concentrations using leave one-out cross-validation (10
loading vectors). Each micro-SERS measurement was made while
samples were in an environmental control cell filled with glucose
solution, using .lamda..sub.ex=632.8 nm (P=3.25 mW, acquisition
time (t)=30 sec). The dashed line represents perfect predictions.
The inset shows the root-mea-squared error of calibration as a
function of number of loading vectors used in the PLS
algorithm.
[0030] FIG. 15 shows calibration vectors used to produce
predictions shown in FIGS. 13, and 14, respectively.
[0031] FIG. 16 shows a Clarke error grid of glucose detection by
nanobiosensors of some embodiments of the present invention.
[0032] FIG. 17 shows SER spectra from nanobiosensors of the present
invention captured every 24 hours from the same sample location for
72 hours.
[0033] FIG. 18 shows SER spectra demonstrating the
partition/departition capability of the EG3-modified AgFON
substrate used in some embodiments of the present invention.
[0034] FIG. 19 shows SER spectra of the detection of glucose in
presence of serum albumin.
[0035] FIG. 20 shows a schematic diagram for fabricating
DT/MH-functionalized AgFON.
[0036] FIG. 20A shows nanospheres (diameter=390 nm) that have
self-assembled to form a hexagonal close packed 2D crystal. Metal
(Ag) was then deposited via electron beam deposition. FIG. 20B
shows that the FON surface was then incubated in a solution of 1 mM
DT in ethanol for 45 minutes and then transferred to 1 mM MH in
ethanol for at least 12 hours.
[0037] FIG. 21 shows the temporal stability of the
DT/MH-functionalized FON. FIG. 21A shows the SERS spectrum of
DT/MH-functionalized FON (Day 2). FIG. 21B shows the time-course of
intensity of the 1119 cm.sup.-1 peak. Signal intensities remained
stable over a 10 day period with only a 2.08% change in intensity,
.lamda..sub.ex=785 nm, P.sub.laser=55 mW, acquisition time (t)=2
min.
[0038] FIG. 22 shows a glucose pulsing sequence on the SAM modified
AgFON surface (inset). SERS spectra of the sample cycled between 0
and 100 mM aqueous glucose solutions (spectra A, B, C, D),
.lamda..sub.ex=532 nm, P.sub.laser=10 mW, acquisition time (t)=20
min, pH.about.7. Spectrum E shows the normal Raman spectrum of
aqueous saturated glucose solution. Difference spectra F, G, H, and
I show the partitioning/departitioning of glucose. Note that
imperfect subtraction of the narrow band at 1053 cm.sup.-1 due to
nitrate results in a very sharp peak in the difference spectra
(diamond).
[0039] FIG. 23 shows calibration (.diamond-solid.) and validation (
) plots using two substrates and multiple spots. A PLS calibration
plot was constructed using 46 data points. The validation plot was
constructed using 23 data points taken over a range of glucose
concentrations (10-450 mg/dL) in 1 mM lactate and 2.5 mM urea at
pH.about.7. RMSEC=9.89 mg/dL (0.55 mM) and RMSEP=92.17 mg/dL (5.12
mM). .lamda..sub.ex=785 nm, P.sub.laser=8.4 mW, acquisition time
(t)=2 min.
[0040] FIG. 24 shows calibration (.diamond-solid.) and validation (
) plots using three substrates and multiple spots acquired in two
days. A PLS calibration plot was constructed using 92 data points.
The validation plot was constructed using 46 data points taken over
a range of glucose concentrations (10-450 mg/dL) in bovine plasma.
RMSEC=34.3 mg/dL (1.9 mM) and RMSEP=83.16 mg/dL (4.62 mM).
.lamda..sub.ex=785 nm, P.sub.laser=10-30 mW, acquisition time (t)=2
min.
[0041] FIG. 25 shows real-time SERS responses to step changes in
glucose concentrations in bovine plasma. FIG. 25A SERS spectra of
the SAM and glucose at various times. Peaks at 1451 and 1428
cm.sup.-1 are features of SAM, and 1462 cm.sup.-1 indicates
glucose. Glucose was injected at t=0 sec, and the cell was flushed
with bovine plasma at t=225 sec. FIG. 25B shows an expanded scale
version of FIG. 25A from 1480 to 1440 cm.sup.-1. FIG. 25C shows
partitioning and departitioning of glucose. .lamda..sub.ex=785 nm,
P.sub.laser=100 mW, acquisition time (t)=15 sec. The 1/e time
constants were calculated to be 28 sec for partitioning, and 25 sec
for departitioning.
[0042] FIG. 26 shows a schematic diagram of the instrumental
apparatus (FIG. 26A) and substrate preparation (FIG. 26B) used for
in vivo experiments. FIG. 26A shows a rat, with a surgically
implanted substrate and optical window integrated into a Raman
spectroscopy system consisting of a Ti:Saphire laser
(.lamda..sub.ex=785 nm) band pass filter (BP), steering and
collection optics, and a long pass filter (LP) that rejected
Raleigh scattered light. FIG. 26B depicts AgFONs used in the in
vivo experiments prepared by depositing metal through a mask of
self-assembled nanospheres. The resultant structure is shown in the
atomic force micrograph in the inset The AgFON was then
functionalized by successive emersions in ethanolic solutions of
decanethiol and mecaptohexanol. While the present invention is not
limited to a particular mechanism of action, and an understanding
of the mechanism is not necessary to practice the present
invention, it is nonetheless contemplated that glucose is able to
partition into and out of the DT/MH layer shown in the left of FIG.
26B.
[0043] FIG. 27 shows the time course of an in vivo glucose
measurement experiment. Paired data from FIG. 27 are displayed
versus time of acquisition, plotting the glucose concentration
change over time. The glucose bolus was started at t=60 min.
Triangles (.tangle-solidup.) represent measurements made using a
One Touch II blood glucose meter, and squares (.box-solid.)
represent measurements made using a SERS sensor. The glucose
infusion was started at 1 hour, as demarcated by the arrow. The
inset shows an exemplary in vivo SERS spectrum after baseline
correction and smoothing. (.lamda..sub.ex=785 nm, P=50 mW, t=2
min).
[0044] FIG. 28 depicts an exemplary Clarke glucose error grid
analysis showing in vivo data from calibration (.diamond-solid.)
and validation ( ) plots using a single substrate and a single spot
on a DT-MH functionalized AgFON surface. The calibration set was
constructed using 21 data points correlated with a commercial
glucometer. The validation set was constructed using 5 data points.
RMSEC=7.46 mg/dL (0.41 mM) and RMSEP=53.42 mg/dL (2.97 mM).
(.lamda..sub.ex=785 nm, P=50 mW, acquisition time (t)=2 min).
[0045] FIG. 29 shows an ex vivo analysis of nanobiosensor response
time. Real-time SERS were recorded in response to step changes in
glucose concentration after being implanted in a rat for 5 hours
and then removed. Glucose was injected into the flow cell at t=0
sec, and the cell was then flushed with bovine plasma at t=225 sec.
FIG. 29 depicts the partitioning and departitioning of glucose from
the nanobiosensor over time in the presence of varying levels of
glucose in the media, with 1/e time constants calculated to be 9
sec for partitioning, and 27 sec for departitioning.
(.lamda..sub.ex=785 nm, P=100 mW, acquisition time (t)=15 s).
GENERAL DESCRIPTION
[0046] The present invention relates to biosensors, in particular
to surface-enhanced Raman (SERS) biosensors for detection of in
vivo and ex vivo analytes. Because of the clinical importance of
the detection of blood glucose, many groups are researching methods
for minimally invasive, biologically compatible, quantitative
glucose detection (McNichols et al., J. Biomed. Opt. 5:5-16 [2000];
Steffes, Diabetes Tech. Ther. 1:129 [1999]). Mid-infrared
absorption, one of the more promising techniques, is sensitive to
temperature, pH, and competing absorption by water. Current
mid-infrared absorption studies utilize an indwelling probe to
minimize complicating factors (Klonoff et al., IEEE LEOS Newsletter
12:13 [1998]). In laser polarimetry, another approach being
developed, polarized light is rotated by chiral molecules, such as
glucose, while passing through the aqueous humor of the eye. This
technique is capable of detecting glucose concentrations as low as
20 mg/dL (.about.2.0 mM) in vitro, however the optical activity of
the other constituents of the aqueous humor, such as ascorbate and
albumin, as well as the birefringence of the cornea make this
approach extremely difficult (Cameron et al., Diabetes Tech. Ther.
1:125 [1999]). Indirect detection of glucose is also done using
fluorescence or other optical techniques (Russell et al., Anal.
Chem. 71:3126 [1999]; Jin et al., Anal. Chem. 69:1326 [1997]).
These techniques rely on the enzymatic reaction of glucose to
produce the detected by-product. Biomolecules similar to the
analyte can interfere with this multi-step process, giving false
positives.
[0047] One technique capable of addressing the major weaknesses of
the aforementioned methods (interfering water absorption,
overlapping signals from competing analytes, and indirect
measurement complications) is by using vibrational Raman
spectroscopy. It has been shown that normal Raman spectroscopy
(NRS) can readily detect physiological concentrations of glucose in
vitro from a simulated aqueous humor solution (Lambert et al., IEEE
LEOS Newsletter 12:19 [1998]). Using partial-least squares (PLS),
Lambert et al. were able to predict glucose levels ranging from 50
mg/dL (2.8 mM, hypoglycemic) to 1300 mg/dL (72.2 mM, severe
diabetic) with a standard error of 24.7 mg/dL (1.5 mM). Berger et
al. were able to detect glucose concentrations with an accuracy of
26 mg/dL (1.4 mM) in serum and 79 mg/dL (4.4 mM) in whole blood
using PLS (Berger et al., Appl. Opt. 38:2916 [1999]). However, the
laser exposure in both experiments is significantly higher than is
biologically permissible (American National Standards Institute,
Laser institute of America: Orlando, Fl 1993). The high laser
powers and long acquisition times are required due to the
inherently small normal Raman scattering cross section of glucose,
5.6.times.10.sup.-30 cm.sup.2 molecule.sup.-7 sr.sup.-1 according
to McCreery and coworkers (McCreery, R. L. Raman Spectroscopy for
Chemical Analysis; John Wiley & Sons, Inc.: New York, 2000;
Vol. 157). The reported Raman cross section for glucose is five
times smaller than that of benzene and 50 times larger than that of
water.
[0048] Raman optical activity spectroscopy and Raman difference
spectroscopy are both examples of highly sensitive Raman techniques
capable of detecting small differences in the Raman cross section.
In both of these techniques, however, the resultant difference
signals are very small and long data acquisition times are required
(Bell et al., Carbohydr. Res. 257:11 [1994]; Chaiken et al., Proc.
SPIE 4254:216 [2001]). Such an approach is not desirable in a
rapid, robust, clinical analysis method. One way to increase the
Raman cross section is to exploit resonance Raman spectroscopy
(Asher, Anal chem. 65:201 A [1993]). In the case of glucose, this
would require excitation in the deep ultraviolet region
(.lamda..about.200 nm) of the spectrum. However, ultraviolet
excitation is unlikely to be appropriate for in vivo sensing due to
photodamage of DNA.
[0049] The compositions and methods of the present invention
overcome these limitations by employing surface enhanced Raman
spectroscopy (SERS). SERS retains all of the advantages of normal
Raman spectroscopy while achieving significantly stronger signal
intensity. SERS is a process whereby the Raman scattering signal is
increased when a Raman-active molecule is spatially confined within
range of the electromagnetic fields generated upon excitation of
the localized surface plasmon resonance of nanostructured noble
metal surfaces. The ensemble averaged Raman signal increases by up
to eight orders of magnitude while the non-ensemble-averaged Raman
signal can increase by 14 or 15 orders of magnitude in special
cases (Emory and Nie, Science 275:1102 [1997]; Kneipp et al., Phys.
Rev. Lett. 78:1667 [1997]). Both chemical and conformational
information can be elucidated from SERS. Theoretical analysis
suggests that molecules confined within the decay length of the
electromagnetic fields, viz. 0-4 nm, will exhibit SER spectra even
if they are not chemisorbed (Schatz et al., In Handbook of
Vibrational Spectroscopy; Chalmers, J. M., Griffiths, P. R. Eds.;
John Wiley & Sons: Chichester, UK, 2002; Vol. 1 pp 759-774).
SERS possesses many desirable characteristics as a tool for the
chemical analysis of in vivo molecular species including high
specificity, attomole to high zeptomole mass sensitivity,
micromolar to picomolar concentration sensitivity, and interfacial
generality (Smith and Rodger, In Handbook of Vibrational
Spectroscopy; Chalmers, J. M., Griffiths, P. R. Eds.; John Wiley
& Sons: Chichester, UK, 2002; Vol. 1 pp 775-784).
[0050] Experiments conducted during the course of development of
the present invention that sought to observe glucose on silver film
over nanosphere (AgFON) surfaces using SERS without a partition
layer were unsuccessful. This result is in agreement with all
previous attempts to measure glucose using SERS that are known.
Published SERS spectra of glucose use a multi-step surface
preparation technique that is likely to be rather unwieldy for
field or clinical applications (Mrozek et al., Anal. Chem. 74:4069
[2002]). The present invention is not limited to a particular
mechanism of action. Indeed, an understanding of the mechanism is
not necessary to practice the present invention. Nonetheless, it is
contemplated that, based on the described substrate preparation and
the resultant SER spectra in Mozek et al., it is possible that
recrystallized rather than adsorbed glucose was observed. Historic
difficulty of SERS detection of glucose is likely to be
attributable to its weak or non-existent binding to bare silver
surfaces since its normal Raman cross section should provide
sufficient signal.
[0051] The present invention provides novel methods for increasing
glucose interaction with the AgFON surface, such as the formation
of a self-assembled monolayer (SAM) on the surface of biosensors to
pre-concentrate the analyte of interest (See e.g., FIG. 11), in a
manner analogous to that used to create the stationary phase in
high performance liquid chromatography (HPLC) (Freunshct et al.
Chem. Phys. Lett. 281:372 [1997]; Blanco et al., J. Anal. Chim.
Acta 436:173 [2001]; Yang et al., Anal. Chem. 34:1326 [1995];
Carron et al., J. Anal. Chem. 67:3353 [1995]; Deschaines et al.,
Appl. Spectrosc. 51:1355 [1997]). Experiments conducted during the
course of development of the present invention demonstrated that
SERS can be utilized for the detection of analytes such as glucose.
The present invention thus provides improved methods of detecting
physiologically relevant analytes.
[0052] Further experiments conducted during the course of
development of the present invention (See Example 2) demonstrated
quantitative detection of glucose in the physiological range (0-450
mg/dL, 0-25 mM) under physiological conditions, three-day sensor
stability, partition/departition efficacy of the sensor, and
glucose detection in the presence of an interfering protein.
[0053] The accuracy of the SERS glucose sensor was evaluated using
the Clarke error grid, the accepted metric for judging the
prediction capability of glucose sensors in the clinical
concentration range (Clarke et al., Diabetes Care 10:622 [1987]).
94% of the predictions fell in zones A and B, signifying that
correct treatment choices can be made using this sensor.
Additionally, the EG3-modified AgFON sensor quantitatively detects
glucose in the physiological range with a corresponding prediction
error of 82 mg/dL (4.5 mM). The stability of the EG3-modified AgFON
SERS substrate is evident as the SERS bands and intensities do not
change significantly during a three-day period in saline with
pH=7.4 at room temperature. The molecular order of the EG3 SAM
increases with incubation time (Biebuyck et al., Langmuir 10:1825
[1994]), and this rearrangement gives rise to slightly larger SERS
intensities. The glucose partition/departition capability of the
EG3-modified AgFON sensor was demonstrated by exposing the sensor
to cycles of 250 mM and 0 mM glucose solutions. The relatively high
glucose concentration used in this experiment caused incomplete
departitioning after each cycle, and accordingly, the glucose
accumulated in each step. The present invention is not limited to a
particular mechanism. Indeed, an understanding of the mechanism is
not necessary to practice the present invention. Nonetheless, it is
contemplated that physiological concentrations of glucose will not
likely cause such accumulation in the partition layer, and the
natural flow of aqueous humor (Vanlandingham et al., Am. J. Opthal.
126:191 [1998]) or interstitial fluid will assist glucose
departitioning. This work further demonstrates that an EG3
partition layer can capture glucose near the surface, while showing
resistance to serum albumin, the most abundant protein in plasma
(Baker et al., FEBS Lett. 439:9 [1998]).
[0054] The present invention further provides methods for the
simultaneous detection of multiple (e.g., two or more) analytes. In
some embodiments, the nanobiosensors contain arrays of regions,
where each region is specific for the detection of a different
analyte. The nanobiosensors can then be scanned with a detection
device to obtain information about the concentration of multiple
analytes.
DEFINITIONS
[0055] As used herein, the term "nanobiosensors," as in
"nanobiosensors configured for surface enhanced Raman spectroscopy
detection of an analyte" refers to any sensor that is small enough
to be implanted internally (e.g., under the skin or in the eye), is
specific for detection of one or more analytes, and is capable of
having an altered surface enhanced Raman signal in the presence of
the specific analyte(s). In preferred embodiments, the
nanobiosensors comprise components for specifically, but
reversibly, interacting with the specific analyte.
[0056] As used herein, the term "surface bound reversibly-binding
receptor" refers to a receptor bound to the surface of a
nanobiosensor of the present invention that binds reversibly to a
specific analyte. In preferred embodiments, the interaction of the
receptor and the analyte lasts long enough for detection of the
analyte by the sensor.
[0057] As used herein, the term "self-assembled monolayer" refers
to a material that forms single layer or multilayers of molecules
on the surface of a nanobiosensor. As used herein, the term
"self-assembled partition layer" refers to material that forms a
layer or multilayers on the surface of a nanobiosensor.
[0058] As used herein, the term "nanowell" refers to a solid
surface comprising wells for immobilizing the nanobiosensors of the
present invention. In preferred embodiments, the nanowells are made
of an inert material and are large enough to hold a plurality of
nanobiosensors.
[0059] As used herein, the term "bodily fluid" refers to any fluid
normally found in the body of a mammal (e.g., a human). Exemplary
bodily fluids include, but are not limited to, blood, serum, lymph,
aqueous humor, interstitial fluid, and urine. The term "bodily
fluid" encompasses both bodily fluid found in its natural state
(e.g., in the body) and bodily fluid removed from the body.
[0060] As used herein, the term "analyte" refers to any molecule or
atom or molecular complex suitable for detection by the
nanobiosensors of the present invention. Exemplary analytes
include, but are not limited to, various biomolecules (e.g.,
proteins, nucleic acids, lipids, etc.), glucose, ascorbate, lactic
acid, urea, pesticides, chemical warfare agents, pollutants, and
explosives.
[0061] As used herein, the term "a device configured for the
detection of surface enhanced Raman scattering signal from said
nanobisoensors" refers to any device suitable for detection of a
signal from the nanobiosensors of the present invention. In some
embodiments, the device includes delivery and collection optics, a
laser source, a notch filter, and detector.
[0062] As used herein, the term "instructions for using said kit
for detection of said analyte" includes instructions for using the
nanobiosensors and devices of present invention for the detection
of any suitable "analyte." In preferred embodiments, the
instructions include instructions for the quantitative detection of
the analyte. In some embodiments, the instructions further comprise
the statement of intended use required by the U.S. Food and Drug
Administration (FDA) in labeling medical devices. The FDA requires
that medical devices be approved through the 510(k) procedure.
Information required in an application under 510(k) includes: 1)
The product name, including the trade or proprietary name, the
common or usual name, and the classification name of the device; 2)
The intended use of the product; 3) The establishment registration
number, if applicable, of the owner or operator submitting the
510(k) submission; the class in which the product was placed under
section 513 of the FD&C Act, if known, its appropriate panel,
or, if the owner or operator determines that the device has not
been classified under such section, a statement of that
determination and the basis for the determination that the product
is not so classified; 4) Proposed labels, labeling and
advertisements sufficient to describe the diagnostic product, its
intended use, and directions for use, including photographs or
engineering drawings, where applicable; 5) A statement indicating
that the device is similar to and/or different from other products
of comparable type in commercial distribution in the U.S.,
accompanied by data to support the statement; 6) A 510(k) summary
of the safety and effectiveness data upon which the substantial
equivalence determination is based; or a statement that the 510(k)
safety and effectiveness information supporting the FDA finding of
substantial equivalence will be made available to any person within
30 days of a written request; 7) A statement that the submitter
believes, to the best of their knowledge, that all data and
information submitted in the premarket notification are truthful
and accurate and that no material fact has been omitted; and 8) Any
additional information regarding the in vitro diagnostic product
requested that is necessary for the FDA to make a substantial
equivalency determination. Additional information is available at
the Internet web page of the U.S. FDA.
[0063] As used herein, the term "physiological concentration range"
refers to the concentration range of an analyte that is typically
found in an animal (e.g., a human). The physiological concentration
range covers both the physiological concentration in a healthy
animal and in an animal with a disease (e.g., diabetes).
[0064] As used herein, the term "detection of said analyte for at
least 3 days" refers to nanobiosensors of the present invention
that are capable of detecting an analyte for at least 3 days in
vitro or in vivo. Detection of said analyte for at least 3 days
does not require that the nanobiosensor take continuous
measurements for 3 days, but that the sensor functions (e.g., by
taking periodic measurements) for at least 3 days. In preferred
embodiments, the measurements are quantitative and maintain
precision and accuracy for at least 3 days.
[0065] As used herein, the term "reversible detection of said
analyte" refers to nanobiosensors of the present invention that are
capable of repeated detection of an analyte. For example, in some
embodiments, nanobiosensors measure the concentration of glucose in
a biological fluid multiple times (e.g., from one time per second
to one time per hour) over the course of the usable life span of
the sensor (e.g., at least 3 days).
[0066] As used herein, the term "detection of said analyte in the
presence of interfering proteins" refers to nanobiosensors of the
present invention that are able to function in the presence of
proteins other than the analyte (e.g., biological proteins).
[0067] As used herein, the term "biological macromolecule" refers
to large molecules (e.g., polymers) typically found in living
organisms. Examples include, but are not limited to, proteins,
nucleic acids, lipids, and carbohydrates.
[0068] A "solvent" is a liquid substance capable of dissolving or
dispersing one or more other substances. It is not intended that
the present invention be limited by the nature of the solvent
used.
[0069] As used herein, the term "polymer" refers to material
comprised of repeating subunits. Examples of polymers include, but
are not limited to polyacrylamide and poly(vinyl chloride),
poly(vinyl chloride) carboxylated, and poly(vinyl chloride-co-vinyl
acetate co-vinyl) alcohols.
[0070] As used herein, the term "polymerization" encompasses any
process that results in the conversion of small molecular monomers
into larger molecules consisting of repeated units. Typically,
polymerization involves chemical crosslinking of monomers to one
another. As used herein, the term "spectrum" refers to the
distribution of electromagnetic energies arranged in order of
wavelength.
[0071] As used the term "visible spectrum" refers to light
radiation that contains wavelengths from approximately 360 nm to
approximately 800 nm.
[0072] As used herein, the term "ultraviolet spectrum" refers to
radiation with wavelengths less than that of visible light (i.e.,
less than approximately 360 nm) but greater than that of X-rays
(i.e., greater than approximately 0.1 nm).
[0073] As used herein, the term "infrared spectrum" refers to
radiation with wavelengths of greater than 800 nm.
[0074] As used herein, the term "sample" is used in its broadest
sense. In one sense, it is meant to include a specimen or culture
obtained from any source, as well as biological and environmental
samples. Biological samples may be obtained from animals (including
humans) and encompass fluids, solids, tissues, and gases.
Biological samples include blood products, such as plasma, serum
and the like. Environmental samples include environmental material
such as surface matter, soil, water, crystals and industrial
samples. Such examples are not however to be construed as limiting
the sample types applicable to the present invention.
[0075] As used herein, the term "medium" refers to the fluid
environment of an analyte of interest. In some embodiments, the
medium refers to a bodily fluid. The bodily fluid may be, for
example, blood, plasma, serum, cerebrospinal fluid, vitreous or
aqueous humor, urine, extracellular fluid, or interstitial fluid.
In some embodiments, the medium is an in vivo medium. In other
embodiments, the medium is an ex vivo or in vitro medium, for
example, a fluid sample taken from a subject.
DETAILED DESCRIPTION OF THE INVENTION
[0076] The present invention relates to biosensors, in particular
to surface-enhanced Raman scattering (SERS) biosensors for
detection of intracellular analytes. The compositions and methods
of the present invention provide sensitive, real time measurement
of physiologically relevant analytes such as glucose.
I. Surface-Enhanced Raman Spectroscopy
[0077] In some embodiments, the present invention provides
nanobiosensors that utilize surface-enhanced Raman spectroscopy to
detect intracellular analytes.
A. Localized Surface Plasmon Resonance
[0078] The signature optical property of a noble metal nanoparticle
is the localized surface plasmon resonance (LSPR). This resonance
occurs when the correct wavelength of light strikes a noble metal
nanoparticle, causing the plasma of conduction electrons to
oscillate collectively. The term LSPR is used to emphasize that
this collective oscillation is localized within the near surface
region of the nanoparticle and to differentiate it from propagating
surface plasmons which are often referred to simply as surface
plasmons. The two consequences of LSPR excitation are: 1) selective
photon absorption and 2) generation of locally enhanced or
amplified electromagnetic fields at the nanoparticle surface. The
LSPR for noble metal nanoparticles in the 20-few hundred nanometer
size regime occurs in the visible and IR regions of the spectrum
and can be measured by UV-visible-IR extinction spectroscopy (FIG.
1) (Haynes et al., J. Phys. Chem. B 105:5599 [2001]). The spectral
location of the LSPR is intricately related to the resulting SERS
spectrum.
B. Nanosphere Lithography
[0079] Nanosphere lithography (NSL) is a fabrication technique to
inexpensively produce nanoparticle arrays with precisely controlled
shape, size, and interparticle spacing, and accordingly precisely
controlled LSPRs (Hulteen et al., J. Vac. Sci. Technol. A 13:1553
[1995]). The need for monodisperse, reproducible, and materials
general nanoparticles has driven the development and refinement of
the most basic NSL architecture as well as many new nanostructure
derivatives. Every NSL structure begins with the self-assembly of
size-monodispersed nanospheres to form a two-dimensional colloidal
crystal deposition mask (FIG. 2A). As in all naturally occurring
crystals, nanosphere masks include a variety of defects that arise
as a result of nanosphere polydispersity, site randomness, point
(vacancy) defects, line defects (slip dislocations) and
polycrystalline domains. Typical defect-free domain sizes are in
the 10-100 micron range. Following self-assembly of the nanosphere
mask, a noble metal or other material is then deposited by thermal
evaporation, electron beam deposition, or pulsed laser deposition
from a source normal to the substrate through the nanosphere mask
to a controlled mass thickness, dm. After noble metal deposition,
the nanosphere mask is removed by sonicating the entire sample in a
solvent, leaving behind the material deposited through the
nanosphere mask to the substrate (FIG. 2B). The LSPR of NSL-derived
nanoparticles depends on nanoparticle material, size, shape,
interparticle spacing, substrate, solvent, dielectric thin film
overlayers, and molecular adsorbates (Haynes et al., supra).
C. Surface-Enhanced Raman Scattering
[0080] Normal Raman scattering is an inelastic scattering process
in which photons incident on a sample transfer energy to or from
the sample's vibrational or rotational modes. Individual bands in a
Raman spectrum are characteristic of specific molecular motions. As
a result, each chemical analyte has its own unique Raman signature.
For example, the four biochemicals commonly found in aqueous humor
each have very different Raman spectra (FIG. 3). When a
Raman-active molecule is positioned within the electromagnetic
fields generated upon excitation of the LSPR of NSL-derived
nanoparticles, the Raman signal increases by up to eight orders of
magnitude. Both chemical and conformational information can be
elucidated from SERS data. Current estimates suggest that the
electromagnetic fields reach further than 65 nanometers from the
noble metal surface, allowing one to probe molecular species using
the surface of embedded nanoparticles (Malinsky et al., J. Am.
Chem. Soc. 123:1471 [2001]). SERS possesses many desirable
characteristics as a tool for the chemical analysis of in vivo
molecular species including high specificity, attomole to high
zeptomole mass sensitivity, micromolar to picomolar concentration
sensitivity, and interfacial generality (Handbook of Vibrational
Spectroscopy; Chalmers, J. M., Griffiths, P. R. Eds.; John Wiley
& Sons: Chichester, UK, 2002; Vol. 1 pp 392).
[0081] In order to evaluate the potential of embedded nanoparticle
multianalyte SERS sensors, it is preferred to consider the
theoretical SERS signal from physiologically relevant analyte
concentrations. The ocular in vivo concentrations of glucose,
lactate, urea, ascorbate, and protein have not been evaluated in
humans. The sensing mechanism of the present invention allows
determination of these concentrations. In some embodiments, the
intensity of the SERS signal is calculated using the following
equation (Van Duyne, R. P. In Chemical and Biochemical Applications
of Lasers; Moore, C. B. Ed.; Academic Press: New York, 1979; Vol.
4, pp 101-184).
I if ( .omega. s ) = .OMEGA. .sigma. ( .omega. s ) .OMEGA. N surf P
L ( .omega. L ) ( .omega. L ) - 1 QT m T o EF ##EQU00001##
[0082] In this equation, I.sub.if(.omega..sub.s) is the intensity
of the SERS peak in photoelectron counts per second, N.sub.surf is
the number of molecules in the probed area of the surface,
.OMEGA.(d.sigma.(.omega..sub.s)/d.OMEGA.) is the scattering
cross-section in molecules.sup.-1 (accounting for the solid
collection angle in steradians and illumination area in cm.sup.2),
P.sub.L(.omega..sub.L).epsilon.(.omega..sub.L).sup.-1 describes the
photon flux in photons per second, QT.sub.mT.sub.o describes the
efficiency of the detection system (unitless), and EF is the
enhancement factor (unitless). Using a Raman cross-section of
10.sup.-30 cm.sup.2sr.sup.-1molecule.sup.-1, an enhancement factor
of 10.sup.8, and the expected collection parameters, a conservative
estimate of the glucose detection limit is 1.51.times.10.sup.-2
mg/dL. This value is almost three orders of magnitude lower than
the expected physiological concentration of 97 mg/dL (in rabbits)
(Lambert et al., IEEE LEOS Newsletter 12:19 [1998]). It is
contemplated that lactate, urea, and ascorbate have similar
detection limits. The present invention thus provides methods for
simultaneously detecting and quantitating a variety of analytes for
both fundamental and applied circumstances.
D. Optimum Parameters for Biocompatible SERS Nanosensors
[0083] Many current attempts at in vivo sensing detect the molecule
of interest indirectly, based on binding events or pH change. The
SERS sensors have the advantage of directly detecting the analytes
of interest, allowing facile quantification. A nanowell structure
(discussed in more detail below) is used in SERS sensors for both
the eye and the skin. Embedded nanoparticle properties (material,
size, and spacing) are chosen to optimize the SERS signal resulting
from Brownian approach of analyte molecules to the SERS-active
substrate (FIG. 4).
[0084] In preferred embodiments, the SERS biosensors of the present
invention are coated with a noble metal. In some embodiments, the
metal is silver. The present invention is not limited to the use of
silver. Any noble metal may be utilized, including, but not limited
to, gold and platinum. In certain embodiments, a 1 nm layer of
titanium or chromium is added to the surface of the particles prior
to the silver in order to improve the adhesion of the silver to the
surface.
[0085] To prolong analyte interaction with the noble metal
nanoparticle surface, in some embodiments, a reversibly-binding
receptor is used to temporarily bind the analyte to the surface. In
the case of glucose, in some embodiments a receptor such as
concanavalin A is used as a reversible-binding agent (See e.g.,
Russell et al., Ana. Chem. 71:3126 [1999]) and/or an alkanethiol,
such as 1-decanethiol, is used to form the self-assembled capture
layer (Blanco Gomis et al., J. Anal. Chim. Acta 436:173 [2001];
Yang et al., Anal. Chem. 34:1326 [1995]). Other exemplary capture
molecules include longer-chained alkanethiols, cyclohexyl
mercaptan, glucosamine, boronic acid and mercapto carboxylic acids
(e.g., 11-mercaptoundecanoic acid). In other embodiments,
apo-glucose oxide is used as the capture molecule.
[0086] Alternatively, a self-assembled monolayer (SAM) is formed on
the nanoparticle surface to concentrate the analyte of interest
near the nanoparticle surface, an adaptation of common high
performance liquid chromatography technology. Exemplary SAMs
include, but are not limited to, 4-aminothiophenol, L-cystein,
3-mercaptopropionicacid, 11-mercaptoundecanoic acid, 1-hexanethiol,
1-octanethiol, 1-DT, 1-hexadecanethiol, poly-DL-lysine,
3-mercapto-1-propanesulfonic acid, benzenethiol, and
cyclohexylmercaptan. In preferred embodiments, the SAM is comprised
of straight chain alkanethiols. In some particularly preferred
embodiments, the SAM is 1-decanethiol. In other particularly
preferred embodiments, the SAM is EG3 (See Example 2). In still
further embodiments, the SAM is a thiolated boronic acid. In yet
other embodiments, the SAM is polyethylene glycol (PEG) or a
thiolated PEG derivative. Preferred SAMs are those that efficiently
and reversibly bind analytes but have capture and release kinetic
rapid enough to follow fast changes in analyte levels (e.g.,
physiological glucose levels). In particularly preferred
embodiments, the SAM comprises mixed components, for example, DT/MH
(decanethiol/mercaptohexanol). In other embodiments, the SAM is
modified to substitute a halogen, for example fluorine, for
hydrogen.
[0087] In some embodiments, a dialysis membrane is utilized to
exclude molecules significantly larger than the analyte (e.g.,
glucose) from contacting the nanoparticle surface. The present
invention is not limited to a particular mechanism. Indeed, an
understanding of the mechanism is not necessary to understand the
present invention. Nonetheless, it is contemplated that the
exclusion of large molecules will increase the accuracy and
precision of measurement of small molecule analytes such as
glucose.
[0088] In other embodiments, nanoparticles are coated to prevent
the accumulation of interfering proteins on the particle surface.
In some embodiments, PEG is immobilized on nanoparticle surfaces to
prevent protein fouling. In some embodiments, silica sensor
surfaces not coated with silver are PEGylated with silane
terminated monomethoxy-PEG and silver coated nanoparticle surfaces
are coated with oligoethyleneglycol terminated alkanethiols. In
some embodiments, the PEGylated surfaces are analyzed using X-ray
photoelectron spectroscopy and secondary ion mass spectra to
determine the presence and homogeneity of PEG on surfaces. In some
embodiments, protein adhesion to modified surfaces is measured by
placing sensors in a culture of fibroblasts for several weeks,
removing unattached cells, and counting the number of adhered
cells. The effect of suitable anti-fouling coatings on sensor
performance can be tested using any suitable method, including, but
not limited to, those disclosed in Example 2 below.
[0089] While the skin sensor is based on a simple chip implant that
can include a SiO.sub.2 substrate, the eye sensor is adapted for
incorporation into an intraocular by etching nanowells directly
into the intraocular lens surface. The choice of excitation
wavelength is optimized for data collection in the eye and the
skin.
E. Durability of Nanoparticle Arrays
[0090] In preferred embodiments, embedded nanoparticles for use in
in vivo systems exhibit both optical and physical durability. In
experiments conducted during the course of development of the
present invention, degradation of the optical signals as the
nanoparticles were exposed to many cycles of buffer and solvent
rinsing was observed. AFM data indicate that the sharp tips of the
triangular nanoparticles are annealed when exposed to these rinse
cycles. This change in particle shape causes an uncontrolled shift
in the LSPR. In some experiments, the nanoparticles were found to
be unintentionally released from the surface into solution. Such
release is undesirable for in vivo applications.
[0091] In some embodiments, a new nanostructure is used to combat
both the uncontrolled shape change and release of nanoparticles. In
this nanostructure, the triangular nanoparticles are embedded in
SiO.sub.2 or polymethylmethacrylate nanowells, effectively
immobilizing the nanoparticle and preventing geometric changes
while maintaining the advantages of ordered arrays of
nanoparticles. This design uses polystyrene or silica nanospheres
as a reactive ion etching (RIE) mask. Polystyrene nanospheres are
used to create nanowells in the silica substrate for the
subcutaneous implant. When CF.sub.4 plasma strikes the polystyrene
nanospheres, the hydrocarbons are fluorinated. This non-volatile
product is not etched away, so the spheres act as an etch stop.
Meanwhile, as the CF.sub.4 plasma penetrates the pores in the
nanosphere mask, volatile SiF.sub.2 radicals and SiF.sub.x products
are etched away.
[0092] In some embodiments, silica nanospheres are used to create
nanowells in the polymer substrate for the intraocular implant. In
this situation, when the O.sub.2 plasma strikes the silica
nanospheres, only oxygen exchange will occur. Reaction between the
O.sub.2 plasma and the hydrocarbon intraocular lens produces
volatile COX products. The resulting structures in both cases are
nanowells with a triangular cross-section. Deposition of material
through the nanosphere mask after etching embeds nanoparticles
within the substrate (FIG. 6). Etched SiO.sub.2 samples have been
characterized by AFM line scans to show an average etch of 15 nm
per minute with 60 mTorr CF.sub.4 plasma pressure.
[0093] Experiments conducted during the course of the present
invention (See e.g., Experimental Section below) demonstrated that
LSPRs can be measured from embedded nanoparticles and are both
measurable and tunable. Seven 400 nm polystyrene diameter
nanosphere masks were etched for varied times in a constant 60
mTorr CF.sub.4 plasma. The depths of these nanowell structures
ranged from 30 nm to 300 nm. Before removing the nanosphere masks,
50 nm of Ag was evaporated onto each sample. The extinction spectra
of these embedded nanoparticle structures were then measured (FIG.
7). In order to predict the extinction response of the embedded
nanoparticles after being exposed to a physiologically relevant
environment, the buried nanoparticles were thermally annealed under
vacuum at 300.degree. C. for 1 hour. The general trend for silver
nanoparticles was that they become more spherical and increase in
height when annealed, yielding a blue shift in the LSPR (FIG.
7).
[0094] In other embodiments, nanowells are fabricated on the tip of
an optical fiber. In some embodiments, the fiber tip is cleaved and
polished prior to use. In some embodiment, a broad reflective
dielectric coating is deposited on the tip. In some embodiments,
the surface of a fiber optic probe is treated to make the surface
clean and hydrophilic (e.g., using 3:1 H.sub.2SO.sub.4; 30%
H.sub.2O.sub.2 at 80.degree. C. for one hour followed by 5:1:1
H.sub.20:NH.sub.4OH:30% H.sub.2O.sub.2 with sonication for one
hour). In some embodiments, a polystryrene nanosphere solution is
then drop-coated onto each substrate and allowed to dry. In certain
embodiments, the nanosphere coated tip is CF4 plasma reactive ion
etched to create wells from 0-300 nm in depth. In some embodiments,
silver is vacuum deposited, followed by sonication in ethanol to
remove the nanopsheres and leave a tip filled with Ag filled
nanowells.
F. Detection and Quantitative Analysis of SERS Signals
[0095] In some embodiments, SERS signals are obtained and detected
using a laser for excitation. In some embodiments, excitation is at
632.8 nm or 532.0 mm. In preferred embodiments, near infra-red
excitation within the "therapeutic window", between 700 and 1200
nm, where absorption by skin is at its minimum is utilized. In some
preferred embodiments, the laser power density is below the
American National Standards Institute guidelines for human exposure
(<2.5 mW cm-2 for 0.25 s, .lamda.=633 nm, directed at the
eye).
[0096] In preferred embodiments, both ocular and skin sensors are
adapted for quantitative analysis. Manoharan et al. have shown that
the normal Raman spectrum of a mixture is a linear combination of
the mixture's component spectra, and that there is a linear
relationship between signal intensity and chemical concentration
(Manoharan et al., J. Photochem. Photobiol. B: Biol. 16:211
[1992]). Experiments conducted during the course of development of
the present invention used partial least-squares leave-one-out
analysis to show quantitative prediction capability for glucose
concentrated by a 1-octanethiol monolayer (FIGS. 8, 13, and 14).
Exemplary calibration techniques include, but are not limited to,
linear multivariate calibration techniques such as partial-least
squares (Geladi et al., Anal. Chim. Acta 185:1 [1986]) and hybrid
linear analysis (Berger et al., Anal. Chem. 70:623 [1998]), as well
as non-linear techniques such as non-linear partial least-squares
and neural networks (Robb et al., Mikrochim. Acta 1:131 [1990]). In
some embodiments, an internal standard is incorporated into the
sensor device to monitor sensor degradation. Calibration algorithms
are optimized for each system and then validated. In preferred
embodiments, tissue scattering and absorption are accounted for in
subcutaneous measurements.
[0097] In some embodiments, the detection system is miniaturized.
Miniaturization is preferable for a clinical application in which a
subject may wear a detection unit and sensor for continuous
monitoring of an analyte. In some embodiments, the
spectrophotometer component of the detection system is limited to a
narrow, relevant wavelength range in order to decrease the size of
the spectrophotometer.
II. Surface-Enhanced Raman Nanobiosensor for Analyte Detection
[0098] The following section describes certain preferred
embodiments of the invention, but the invention is not limited to
these embodiments. In some embodiments, the present invention
provides a nanobiosensor for use in the detection of analytes. In
some preferred embodiments, the sensor is a surface-enhanced Raman
(SERS) nanobiosensor. The in vivo biochemical sensor of the present
invention is designed to take advantage of the surface-enhancing
properties of noble metallic nanoparticles to acquire Raman spectra
from eye (e.g., aqueous humor) or skin (e.g., interstitial fluid,
blood), or other organs. Preferred organs for implantation of the
sensor are accessible without invasive procedures (e.g., are
external) and contain a bodily fluid that is in contact with or
exchanges analytes with the entire body.
[0099] The surface-enhanced Raman nanobiosensor enables real-time,
continuous measurement of multiple analytes (such as glucose, urea,
and ascorbate) simultaneously. Another advantage of this technique
is that it directly detects the presence of the analytes, rather
than relying on an indirect measurement. In some embodiments, the
initial placement of the sensor requires surgery, but once in place
subsequent measurements are non-invasive.
[0100] A. Sensor Fabrication
[0101] In some embodiments, the sensor is fabricated from a
substrate including, but not limited to, polymethacrylate, acrylic,
or silicone for the eye and SiO.sub.2 for under the skin. In some
embodiments, noble metal nanoparticles are deposited into shallow
wells in the substrate. In preferred embodiments, the sensor region
is only a few millimeters in its longest dimension. In some
embodiments, the particles are then coated with a self-assembled
monolayer (SAM) to protect them from fouling and to prolong
interaction between the analytes of interest and the surface (FIG.
9). In some embodiments, reversibly-binding receptors are
incorporated into this SAM. The sensor is implanted either under
the skin or used to replace the intraocular lens (FIG. 10). To
detect surface-enhanced Raman signals from the sensor, delivery and
collection optics as well as a laser source, an optical filter, and
a detector are used. In some embodiments, the delivery and
collection optics (as well as filters) are incorporated into a
fiber optic probe, which is connected to the laser and
detector.
[0102] B. Mixed Partition Layers Enables Real-Time Glucose Sensing
by Surface Enhanced Raman Spectroscopy in Plasma
[0103] In some embodiments, the present invention provides a
composition comprising a biosensor comprising a substrate, and a
plurality of nanobiosensors adherent to the substrate, comprising:
a plurality of nanospheres; a metal film over the nanospheres
(MFON); and a self-assembled partition layer formed on the surface
of the metal film over the nanospheres comprising at least two
compounds, wherein the nanobiosensors are configured for the
quantitative detection of an analyte such that the spectrum of a
surface-enhanced Raman signal detected from the plurality of
nanobiosensors in the presence of the analyte is correlated with
the concentration of the analyte in a medium. Experiments conducted
during the course of development of the present invention led to
the discovery of a mixed decanethiol (DT) partition layer with
improved properties that has been developed for glucose sensing by
surface-enhanced Raman spectroscopy (See Experimental Example 4,
below). DT is hydrophobic and not compatible with an aqueous
environment. The mixed partition layer based on two commercially
available components, decanethiol (DT) and mercaptohexanol (MH)
exhibits (1) temporal stability on an AgFON surface of the
nanobiosensor, (2) rapid (less than one min), reversible
partitioning and departitioning of glucose on a
DT/MH-functionalized AgFON surface, (3) quantitative detection of
glucose in aqueous solution with interfering analytes and in bovine
plasma over the physiological and pathological concentration
ranges, and (4) real-time kinetics of glucose partitioning and
departitioning. Moreover, the DT/MH-functionalized surface is
simple to assemble and to control.
[0104] While the present invention is not limited to a particular
mechanism of action, and an understanding of the mechanism is not
necessary to practice the present invention, it is nonetheless
contemplated that on space filling models the hydroxyl-terminated
chains form hydrophilic pockets, thus partitioning glucose closer
to the SERS-active surface. In turn, the DT/MH SAM has dual
hydrophobic/hydrophilic functionality thereby excluding non-target
molecules such as proteins that could give rise to spectral
congestion. This property facilitates detection by simplifying the
composition of the solution at the surface, while the SERS spectra
provide a vibrational "fingerprint" that is unique to each
molecule.
[0105] C. Uses of Sensors
[0106] In some embodiments, the eye implant is a modified
intraocular lenses commonly used in lens replacements when
cataracts occur. The noble metal nanoparticles are incorporated
into a small portion of these lenses to form the sensor.
[0107] In some embodiments, for skin implant sensor fabrication,
the noble metal nanoparticles are deposited in shallow wells in a
chip (e.g., only a few millimeters in its longest dimension)
composed of SiO.sub.2.
[0108] In some embodiments, the surface-enhanced Raman
nanobiosensors of the present invention enable faster, easier, and
continuous measurement glucose levels for diabetics. In other
embodiments, the nanobiosensors are used in the measurement of
previously unmonitored analytes critical in other diseases.
Continuous measurements of blood glucose levels open the door to
implanted insulin pumps. In some embodiments, a SERS nanobiosensor
is used for monitoring drug-delivery in many situations, enabling
tighter control over drug administration.
[0109] The methods of the present invention are not limited to the
detection of glucose. Previously, SERS has been used to detect a
wide variety of analytes present at low concentrations, including,
but not limited to, pollutants (Weissenbacher et al., J. Mol.
Struct. 410-411:539 [1997]), explosives (McHugh et al., Chem.
Commun. 580:-581 [2002]; Sylvia et al., Anal. Chem. 72:5834
[2000]), chemical warfare agents (Taranenko et al., J. Raman Spec.
27:379 [1996]), and DNA (Vo Dinh et al., J. Raman Spec. 30:785
[1999]). The methods of the present invention are thus applicable
to the in vivo detection of exposure (e.g., monitoring) of
individuals exposed to such agents.
[0110] D. In Vivo Measurement of Glucose Concentration by
Surface-Enhanced Raman Spectroscopy on Chemically Modified Metal
Film Over Nanosphere Substrates
[0111] In some embodiments, the present invention provides a method
for detection of an analyte in vivo with the biosensor of the
present invention implanted in a subject. In experiments conducted
during the course of development of the present invention SERS was
used to obtain in vivo quantitative glucose measurements from an
animal model (See Experimental Example 5, below). Silver film over
nanosphere (AgFON) substrates were functionalized with a two
component self-assembled monolayer (SAM) (See Experimental Example
4, below), and the biosensor of the present invention was
subcutaneously implanted in a Sprague-Dawley rat such that the
glucose concentration of the interstitial fluid could be
spectroscopically addressed through an optical window.
[0112] In vivo applications of SERS confront a number of
challenges. Analytes of interest must be in close physical
proximity to (.about.1-2 nm), or adsorbed on, a roughened metal
surface. In turn, the complexity and structural similarity of many
molecules (e.g., proteins) may yield SERS spectra that are
difficult to interpret. Moreover, placement of the SERS active
surface in living systems must avoid damage to either the host or
the surface. While colloid-based substrates may be difficult to
introduce into cells and can coalesce in the extracellular space,
solid substrates are more robust, but require surgical
implantation. Additional problems may become apparent after the
substrate is surgically implanted and surrounded by a biological
medium. For example, there may be little of no control over which
species adsorb, perhaps irreversibly, to the SERS active surface.
This condition potentially creates undesired spectral noise from
non-target molecules, while simultaneously blocking the access of
the desired species, lowering the possible signal. Cellular and in
vivo environments may be awash with a multitude of interfering
molecules whose presence and concentration are in a constant state
of flux. Similarly, the concentration of the target analyte itself
may be invariably changing. These challenges are compounded in vivo
by host immune responses, clotting factors, and the concentration
of target species in the extracellular matrix. In aggregate, these
factors may contribute to surface contamination and cause unwanted
effects.
[0113] As shown in Experimental Example 5, the present invention
addresses and surmounts these challenges. In experiments conducted
during the course of development of the present invention, a
technique that addresses the critical problems previously limiting
the use of SERS to glucose measurements, particularly in vivo, has
been developed. Advances include the development of stable and
strongly enhancing SERS active surfaces, and the chemical
functionalization of those surfaces with self-assembled monolayers
(SAMs). (See Experimental Examples 1, 2 and 4, below).
[0114] While the present invention is not limited to a particular
mechanism of action, and an understanding of the mechanism is not
necessary to practice the present invention, it is nonetheless
contemplated that the multiple roles performed by SAMs in the in
vivo detection system include limiting fouling, providing an
internal standard, segregating classes of interferants from the
detection surface, and amplifying the analyte signal. Stability of
the SERS signal, and substrate stability itself is primarily
determined by the material properties of the enhancing substrate.
(Stuart, et al., The Analyst [2005] (in press); McFarland et al.,
Phys. Chem. B.109:11279 [2005]). Experiments were conducted during
the course of development of the present invention that verified
the parameters required to optimize the plasmonic properties of the
FON type substrates. Although the FON variant of NSL is
intrinsically less enhancing (EF=10.sup.6) than other NSL varieties
(10.sup.8), FONs provide higher overall SERS signals. While the
present invention is not limited to a particular mechanism of
action, and an understanding of the mechanism is not necessary to
practice the present invention, it is nonetheless contemplated that
this is because the total signal is related to both the SERS EF and
the number of analyte molecules probed, which is quite high for
FONs because of their relatively large viable surface area. The
high radius of curvature imparted by the underlying nanospheres
prevents annealing or loss of the nanoscale roughness features that
give rise to SERS. Hence, the use of SAM functionalized substrates
allows the present invention to overcome many of the in vivo the
hurdles cited above (Lyandres et al., Analytical Chemistry
77:6134[1005]; Sulk et al., Journal of Raman Spectroscopy 30:853
[1999]), thereby providing a minimally invasive, real-time
continuous, reusable, quantitative SERS glucose sensor as a
candidate for implantable sensing.
III. Kits
[0115] In some embodiments, the present invention provides kits and
systems for use in monitoring the level of an analyte in an
individual. In some embodiments, the kits are kits for home use by
a subject (e.g., a subject with diabetes). For example, in some
embodiments, a sensor is implanted in the skin or the eye of a
subject (e.g., by a medical professional) and the subject is
provided with a device for monitoring levels of analyte (e.g., the
subject places the device near the sensor and the device reads-out
glucose levels). The subject can then use this information to
maintain better control of blood glucose levels and avoid
complications of the disease. In some embodiments, the sensor is
used extra-corporeally by introducing a biological sample (e.g.,
blood) to the device.
[0116] In other embodiments, the present invention provides kits
for use by medical professionals. For example, in some embodiments,
the present invention provides kits for monitoring military
personnel in a war situation where they may be exposed to toxins.
The sensors are implanted prior to potential exposure (e.g., prior
to departing for active duty). Personnel are then monitored by
medical professionals using a detection device.
[0117] In still further embodiments, the present device is used at
home or by a medical professional to monitor exposure to pesticides
(e.g., in agricultural workers). The workers receive a sensor and
are then monitored using a detection device.
[0118] In yet other embodiments, the present invention provides
systems comprising nanobiosensors and detection devices. For
example, in some embodiments, the systems are combined with an
insulin delivery device (e.g., an insulin pump) for use as an
artificial pancreas. Such a device finds use in the treatment of
individuals with diabetes who require regular insulin doses. In
some embodiments, the detection device and pump are external (e.g.,
combined into one unit). The device takes readings from a sensor
(e.g., implanted in the skin near the device), calculates blood
glucose concentration, and administers an appropriate level of
insulin. In other embodiments, the entire system is internal (e.g.,
implanted underneath the skin or located in the abdominal cavity).
In some embodiments, the entire system is a single unit comprising
a sensor, a detection device, and an insulin delivery device.
EXPERIMENTAL
[0119] The following examples serve to illustrate certain preferred
embodiments and aspects of the present invention and are not to be
construed as limiting the scope thereof.
Example 1
Optimization of SAMs for Biosensors
[0120] This Example describes the characterization of glucose
sensing biosensors comprising a variety of SAMs.
A. Materials
[0121] Ag (99.99%, 0.04'' diameter) was purchased from D. F.
Goldsmith (Evanston, Ill.). Glass substrates were 18 mm diameter,
No. 2 coverslips from Fisher Scientific (Fairlawn, Va.).
Pretreatment of substrates required H2SO4, H2O2, and NH4OH, all
purchased from Fisher Scientific (Fairlawn, Va.). Surfactant-free
white carboxyl-substituted polystyrene latex nanospheres with
diameters of 390.+-.19.5 nm were obtained from Duke Scientific
Corporation (Palo Alto, Calif.). Tungsten vapor deposition boats
were purchased from R. D. Mathis (Long Beach, Calif.).
4-aminothiophenol (90%), L-cysteine (97%), 3-mercaptoproprionic
acid (99+%), 11-mercaptoundecanoic acid (95%), 1-hexanethiol (95%),
1-octanethiol (98%), 1-DT (96%), 1-hexadecanethiol (92%),
3-mercapto-1-propanesulfonic acid (Na+ salt, 90%), benzenethiol
(99+%), cyclohexylmercaptan (97%), .alpha.-D-Glucose (ACS Reagent
Grade) were purchased from Aldrich (Milwaukee, Wis.) and used as
received. Poly-DL-lysine hydrobromide was purchased from Sigma (St.
Louis, Mo.). Ethanol was purchased from Pharmco (Brookfield,
Conn.). For all steps of substrate and solution preparation,
ultrapure water (18.2 M.OMEGA. cm-1) from a Millipore academic
system (Marlborough, Mass.) was used.
AgFON Fabrication and Incubation Procedure.
[0122] Borosilicate glass substrates were pretreated in two steps
(1) pianha etch, 3:1 H.sub.2SO.sub.4:30% H.sub.2O.sub.2 at
80.degree. C. for 1 hr, was used to clean the substrate, and (2)
base treatment, 5:1:1 H.sub.2O:NH.sub.4OH:30% H.sub.2O.sub.2 with
sonication for 1 hour, was used to render the surface hydrophilic.
Approximately 2 .mu.L of undiluted nanosphere solution (4% solids)
were drop coated onto each substrate and allowed to dry in ambient
conditions. The metal films were deposited in a modified
Consolidated Vacuum Corporation vapor deposition system (Hulteen et
al., J. Vac. Sci. Technol. A 13:1553 [1995]) with a base pressure
of 10.sup.-7 torr. The mass thickness of Ag in all cases was 200 nm
and deposition rates for each film (1 nm/sec) were measured using a
Leybold Inficon XTM/2 quartz-crystal microbalance (QCM) (East
Syracuse, N.Y.). Fresh AgFON samples were incubated in 1 mM
solutions of the partition layer self-assembled monolayers (SAMs)
in ethanol for >12 hours before being exposed to glucose
solutions of the desired concentration. Each sample was dosed in a
separate vial. Glucose solutions ranged in concentration from 0-250
mM in 80% ethanol:20% water.
Micro-SERS Apparatus
[0123] Spatially-resolved SER spectra were measured using a
modified Nikon Optiphot (Frier Company, Huntley, Ill.) confocal
microscope with a 20.times. objective in backscattering geometry.
The laser light from a Coherent (Santa Clara, Calif.) model 590 dye
laser operating at .lamda..sub.ex=632.8 nm or a Spectra-Physics
(Moutainview, Calif.) model Millenia Vs laser operating at
.lamda..sub.ex=532.0 nm was coupled into a 200 .mu.m core diameter
fiber using a Thorlabs (Newton, N.J.) fiber launch. Appropriate
Edmund Scientific (Barrington, N.J.) interference filters and
Kaiser (Ann Arbor, Mich.) holographic notch filters were placed in
the beam path. The back-scattered light was collected by an output
fiber optic coupled to an Acton (Acton, Mass.) VM-505 monochromator
(entrance slit set at 250 .mu.m) with a Roper Scientific (Trenton,
N.J.) Spec-10:400B liquid N2-cooled CCD detector.
Chemometrics Method
[0124] All data processing was performed using MATLAB (MathWorks,
Inc., Natick, Mass.) and PLS_Toolbox (Eigenvector Research, Inc.,
Manson, Wash.). Prior to analysis, cosmic rays were removed from
the spectra using a derivative filter and the slowly-varying
background, commonly seen in SERS experiments, was removed by
subtracting a fourth-order polynomial. The data was then
mean-centered. Data analysis was performed using partial least
squares (PLS) leave-one-out (LOO) analysis. PLS was chosen from
among the many chemometric techniques available because it only
requires knowledge of the concentrations of the analyte of interest
during calibration (Geladi et al., Anal. Chim. Acta. 185:1 [1986];
Haaland et al., Anal. Chem. 60:1193 [1988]). Other techniques, such
as classical least-squares require knowledge of all of the
chemicals present in the sample. Although the precise amount of
glucose added to each sample is known in the presented experiments,
the knowledge of the other chemicals in the background (e.g.
polystyrene from substrate preparation or impurities in the
partition layers) was not known.
[0125] Whenever a chemometric technique is used, proper validation
is preferred to aid in obtaining meaningful results. Usually two
separate data sets are used, one for calibration and one for
validation. Because of the limited number of samples in the data
set, LOO was chosen as the cross-validation technique (Martens et
al., Multivariate Calibration; Wiley:Chichester, 1989). In LOO
analysis, one sample at a time is left out of the calibration set.
The PLS model is developed using the remaining data and then
applied to the lone sample. The predicted concentration of this
sample is then compared to the actual concentration and used to
evaluate the quality of the model. The process is then repeated,
leaving each sample out, one at a time, to build up a set of
validation results. LOO cross-validation enables evaluation of a
new technique despite a relatively small data set. Prediction error
in the calibration and validation sets was determined by
calculating the root-mean-squared error of prediction (RMSEP),
R M S E P = ( conc 1 - pred 1 ) 2 + ( conc 2 - pred 2 ) 2 + + (
conc n - pred n ) 2 n ( 1 ) ##EQU00002##
In this equation, conc represents the actual concentration of a
sample, pred represents the predicted concentration for that
sample, and n is the total number of samples. The choice of the
number of loading vectors to use in the PLS results discussed here
was determined by the number of loading vectors needed for the
root-mean-squared error of calibration (RMSEC) to stabilize at a
minimum value.
B. Results
[0126] Several SAMs were tested to determine their effectiveness as
a partition layer. The twelve SAMs tested were 4-aminothiophenol,
L-cystein, 3-mercaptopropionicacid, 11-mercaptoundecanoic acid,
1-hexanethiol, 1-octanethiol, 1-DT, 1-hexadecanethiol,
poly-DL-lysine, 3-mercapto-1-propanesulfonic acid, benzenethiol,
and cyclohexylmercaptan. Of these, only the straight chain
alkanethiols were found to be effective partition layers,
especially 1-DT (which forms a monolayer on silver .about.1.9 nm
thick) (Walczak et al., J. Am. Chem. Soc. 113:2370 [1991]). 1-DT
almost completely fills the theoretical first decay length of the
electromagnetic fields from the SERS substrate (Schatz et al., In
Handbook of Vibrational Spectroscopy; Chalmers, J. M., Griffiths,
P. R. Eds.; John Wiley & Sons: Chichester, UK, 2002; Vol. 1 pp
775-784. FIG. 12 shows example spectra from the different stages of
assembly of the glucose/1-DT/AgFON surface. FIG. 12A shows the SER
spectrum of 1-DT on a AgFON surface. After 10 minutes incubation in
100 mM glucose solution, the SER spectrum in FIG. 12B was observed.
This spectrum is the superposition of the SER spectra for the
partition layer and glucose. FIG. 12B shows vibrational features
from both the analyte glucose (1123 and 1064 cm-1) and 1-DT (1099,
864, and 681 cm-1) constituents. The SERS difference spectrum
resulting from subtraction of spectrum 12A from spectrum 12B is
shown in FIG. 12C. The difference spectrum can be compared directly
to the normal Raman spectrum of crystalline glucose shown in FIG.
12D. The vibrational bands seen at 914 cm-1 and 840 cm-1 in the
crystalline glucose spectrum (FIG. 12D) are not observed in the
spectra shown in FIGS. 12B and 12C because these bands are
strongest in crystalline glucose; this phenomenon has been
previously observed (Mrozek et al., J. Anal. Chem. 74:4069
[2002]).
[0127] In the initial quantitative experiment, AgFON surfaces with
a monolayer of 1-DT were incubated for ten minutes in a solution
containing glucose concentrations ranging from 0-250 mM. SER
spectra were then measured from each sample using
.lamda..sub.ex=632.8 nm (P.sub.laser=4.7 mW, acquisition time
(t)=90s). In all 36 cases, the measurements were made on samples in
dry, ambient conditions. Upon performing LOO-PLS analysis, 21
loading vectors were found to minimize the root-mean-squared error
of calibration (RMSEC), see inset of FIG. 13. The resulting
cross-validated glucose concentration predictions, using 21 loading
vectors, can be seen in FIG. 13. The corresponding error of
prediction is 3.3 mM. This result was repeated with multiple,
similar data sets. While quantitative SERS detection is
demonstrated in the aforementioned data set, a clinically-relevant
concentration range is preferred. Accordingly, AgFONs with a
monolayer of 1-DT were incubated for an hour in glucose solutions
diluted by a factor of 10 (0-25 mM, 0-450 mg/dL). SER spectra were
then measured from each sample using .lamda..sub.ex=632.8 nm
(P.sub.laser=3.25 mW, acquisition time (t)=30 s). In all 13 cases,
the measurements were made on samples in a simple environmentally
controlled cell, bathed in the corresponding glucose solution. Upon
performing LOO-PLS analysis, 10 loading vectors were found to
minimize the root-mean-squared error of calibration (RMSEC), see
inset of FIG. 14. The resulting cross-validated glucose
concentration predictions, using 10 loading vectors, can be seen in
FIG. 14. The corresponding error of prediction is 1.8 mM. Fewer
loading vectors and a lower RMSEP in the smaller concentration
range experiment may be attributable to the onset of a non-linear
signal versus glucose concentration relationship (i.e. the
non-linear portion of the partition isotherm) as higher
concentrations are partitioned.
[0128] In the calibration vectors (FIG. 15A and 15B) used to
generate the prediction plots seen in FIGS. 13 and 14, the
characteristic vibrational bands of glucose are clearly visible at
1121 cm.sup.-1 and 1071 cm.sup.-1. These calibration vectors
represent the portions of glucose that do not overlap with bands of
the partition layer or analytes present in the background.
Accordingly, some glucose features are absent, while others
represent the portion of the glucose band not overlapping with
those bands of 1-DT.
[0129] In conclusion, the first systematic detection of glucose
using SERS is described. The SERS bands observed clearly at 1123
cm.sup.-1 and 1064 cm.sup.-1 demonstrate the vibrational features
of glucose in solution. The adsorption problem has been
circumvented by partitioning glucose into an alkanethiol monolayer
adsorbed on the silver surface thereby pre-concentrating it within
the zone of electromagnetic field enhancement. Of the 12 partition
layers studied, only straight chain alkanethiols were found to be
effective. Consequently, 1-DT was chosen as the partition layer for
all the studies.
[0130] Two data sets are presented to support the quantitative
detection of glucose using SERS. The first probes the quantitative
prediction of glucose over a large concentration range (0-250 mM),
demonstrating a root-mean-squared error of prediction (RMSEP) of
3.3 mM. The second covers the clinically-relevant concentration
range (0-25 mM/0-450 mg/dL), performed in a liquid environment with
short (viz. 30 second) data acquisition times. This data set is
effectively treated using LOO-PLS and displays a RMSEP of 1.8 mM
(33.1 mg/dL), near that desired for medical applications. The
calibration vectors derived in both experiments using the PLS
algorithm show the characteristic vibrational features of
glucose.
Example 2
Biosensors Utilizing EG3 Monolayers
[0131] This Example describes the characterization of
glucose-sensing biosensors comprising EG3 self assembled
monolayers.
A. Methods
Materials
[0132] All the chemicals were of reagent grade or better, and used
as purchased. Ag wire (99.99%, 0.04 inch diameter) was purchased
from D. F. Goldsmith (Evanston, Ill.). Oxygen-free high
conductivity copper was obtained from McMaster-Carr (Chicago, Ill.)
and cut into 18-mm-diameter discs. CH.sub.3CH.sub.2OH,
H.sub.2O.sub.2, and NH.sub.4OH were purchased from Fisher
Scientific (Fairlawn, Va.). Surfactant-free white
carboxyl-substituted latex polystyrene nanosphere suspensions
(390.+-.19.5 nm diameter, 4% solid) were acquired from Duke
Scientific Corporation (Palo Alto, Calif.). Tungsten vapor
deposition boats were purchased from R. D. Mathis (Long Beach,
Calif.). For substrate and solution preparations, ultrapure water
(18.2 M.OMEGA. cm-1) from a Millipore academic system (Marlborough,
Mass.) was used. Bovine serum albumin and saline were obtained from
Sigma (St. Louis, Mo.). The disposable filters with 0.45-.mu.m-pore
size were acquired from Gelman Sciences (Ann Arbor, Mich.).
(1-Mercaptoundeca-11-yl) tri(ethylene glycol)
(HS(CH.sub.2).sub.11(OCH.sub.2CH.sub.2).sub.3OH, EG3) was
synthesized (Palegrosdemange et al., J. Am. Chem. Soc. 113:12-20
[1991]) and donated by the Mrksich group at the University of
Chicago (Hodneland et al., Proc. Natl. Acad. Sci. 99:5048
[2002]).
AgFON Fabrication and Incubation Procedure
[0133] AgFON substrates were used because of their stable SERS
activity in electrochemical ultrahigh vacuum (Dick et al., J. Phys.
Chem. B 106:853 [2002]; Dick et al., J. Phys. Chem. B 104:11752
[2000]; Litorja et al., J. Phys. Chem. B 105: 6907 [2001]) and
ambient experiments (Shafer-Peltier et al., J. Am. Chem. Soc. 125:
588 [2003]). In this work, AgFONs were fabricated on copper
substrates. The copper substrates were cleaned by sonicating in
10:1:1 H.sub.2O:30% H.sub.2O.sub.2:NH.sub.4OH. Approximately 12
.mu.L of nanosphere solution was drop-coated onto a clean copper
substrate and allowed to dry at room temperature. Then,
200-nm-thick Ag films were deposited onto and through the
nanosphere mask using a modified Consolidated Vacuum Corporation
vapor deposition system (base pressure 10-7 Torr) (Hulteen et al.,
J. Vac. Sci. Technol. A 13:1553 [1995]). The mass thickness and
deposition rate (.about.1 nm/sec) of the Ag metal were measured by
a Leybold Inficon XTM/2 quartz-crystal microbalance (East Syracuse,
N.Y.). AgFON substrates were first incubated in 1 mM EG3 in ethanol
for more than 12 hours. Then, the EG3-modified substrates were
mounted into a small volume flow cell and exposed to glucose
solutions for 10 minutes to ensure complete partitioning of the
glucose into the EG3 monolayer.
Surface-Enhanced Raman Scattering Spectroscopy
[0134] A Spectra-Physics model 120 HeNe laser was used to produce
the 632.8 nm excitation wavelength (.lamda.ex); the laser spot size
was less than 2 mm in diameter. The SERS measurement system
includes an interference filter (Edmund Scientific, Barrington,
N.J.), a holographic notch filter (Kaiser Optical Systems, Ann
Arbor, Mich.), a model VM-505 single-grating monochromator with the
entrance slit set at 100 .mu.m (Acton Research Corp., Acton,
Mass.), and a liquid N.sub.2-cooled CCD detector (Roper Scientific,
Trenton, N.J.). The small volume flow cell (Malinsky et al., J. Am.
Chem. Soc. 123:1471 [2001]) was used to control the external
environment of AgFON surfaces throughout the SERS experiment.
Chemometrics Method
[0135] All data processing was performed using MATLAB (MathWorks,
Inc., Natick, Mass.) and PLS_Toolbox (Eigenvector Research, Inc.,
Manson, Wash.). Prior to analysis, cosmic rays were removed from
the spectra using a derivative filter. The slowly-varying
background, commonly seen in SERS experiments, was also removed by
subtracting a fourth-order polynomial. Data analysis was performed
using partial least-squares (PLS) leave-one-out (LOO) analysis.
B. Results
[0136] Significant progress has been made toward achieving a
real-time, non-invasive, biocompatible SERS glucose sensor. In
previous work (See Example 1), decanethiol was used as a partition
layer for glucose, but the required sensor characteristics of
temporal stability, reversibility, and biocompatibility were not
studied in detail. Herein, EG3 was chosen as a partition layer
because of its biocompatibility and hydrophilic properties,
progressing toward the long-term goal of fabricating an implantable
glucose sensor. The EG3-modified AgFON substrate was exposed to
various concentrations of glucose under physiological conditions,
promoting preconcentration of glucose near the AgFON surface. After
data analysis using LOO-PLS, the results are presented in a Clarke
error grid (FIG. 16). Clarke and coworkers established the Clarke
error grid as a metric for evaluating glucose sensor efficacy in
the clinical concentration range (Clarke et al., Diabetes Care
10:622 [1987]). The Clarke error grid is divided into five major
zones: zone A predictions lead to clinically correct treatment
decisions; zone B predictions lead to benign errors or no
treatment; zone C predictions lead to overcorrecting acceptable
blood glucose concentrations; zone D predictions lead to dangerous
failure to detect and treat; and zone E predictions lead to further
aggravating abnormal glucose levels.
Quantitative Study of Glucose Using an EG3 Partition Layer
[0137] A viable glucose biosensor should be capable of detecting
0-450 mg/dL (0-25 mM) glucose under physiological conditions. Each
EG3-modified AgFON sample was incubated for 10 minutes in a pH=7.4
saline solution containing glucose concentrations from 0-450 mg/dL
(0-25 mM). The samples were placed in an environmental control flow
cell under saline, and SERS spectra were then measured
(.lamda.ex=632.8 nm, Plaser=2.5 mW, acquisition time (t)=30 s).
After spectral normalization using EG3 peak intensities, the SERS
spectra were analyzed with the LOO-PLS method.
[0138] In the data presented in FIG. 16, five loading vectors were
found to minimize the root-mean-squared error of cross validation
(RMSECV). The resulting cross-validated glucose concentration
predictions are presented in the Clarke error grid (FIG. 16). The
EG3-modified AgFON sensor quantitatively detects glucose in the
physiological range with a corresponding RMSECV of 82 mg/dL (4.5
mM). In FIG. 16, 94% of the predictions fall in zones A and B,
while a few data points overlap in zone D within the hypoglycemic
area (<70 mg/dL, <3.9 mM). The prediction error of 82 mg/dL
(4.5 mM) can be partially attributed to variation of the SERS
enhancement factor on different AgFON samples. The nanostructure on
a AgFON substrate varies from point to point, affecting the
localized surface plasmon resonance, and accordingly, the SERS
enhancement factor (Haynes et al., J. Phys. Chem. 107:7426
[2003]).
Temporal Stability of the EG3-Modified Substrate
[0139] It is preferred that implantable glucose sensors are stable
for at least a three-day period (Kaufman et al., Diabetes Care
24:2030 [2001]). Previous work has demonstrated that bare AgFON
surfaces display extremely stable SERS activity when challenged
with high potentials (Dick et al., [2002]; supra) and high
temperatures in ultrahigh Vacuum (Litoija et al., J. Phys. Chem. B
105:6907 [2001]). Here, the stability of the EG3-modified AgFON
SERS substrate is studied over a period of three days in saline
with pH=7.4 at room temperature. SER spectra were captured every 24
hours from the same sample location (.lamda.ex=632.8, acquisition
time (t)=60 s) (FIG. 17). The EG3 spectral band positions do not
vary significantly over the course of 72 hours. Peaks at 1107 and
1064 cm-1 increase in intensity by 7.5% and 13% over 48 hours,
respectively (inset in FIG. 17). The molecular order of self
assembled monolayers (SAMs) increases with incubation time
(Biebuyck et al., Langmuir 10:1825 [1994]); the rearrangement of
the SAM gives rise to peaks with increasing intensity. The SERS
peaks at 1341 and 834 cm-1 have been identified as a signature of
highly ordered SAMs (Clarke et al., J. Phys. Chem. B 103: 8201
[1999]; Gregory et al., J. Phys. Chem. B 105:4684 [2001]) and are
the subject of further investigation.
Reversible Glucose Sensing
[0140] While the quantitative detection of glucose using the
EG3-modified AgFON sensor and the stability of the sensor have been
demonstrated, an implantable sensor is preferably reusable. In
order to examine the partition/departition capability of the
EG3-modified AgFON sensor, it was exposed to cycles of 250 mM and 0
mM glucose solutions (FIG. 18 inset). SER spectra were captured
after each concentration variation (.lamda.ex=632.8, Plaser=1.5 mW,
t=30.times.20 s) (FIG. 18A). F and G are the difference spectra
representing glucose partitioned into the EG3 SAM. I is the Raman
spectrum of crystalline glucose for comparison. Vibrational modes
at 1342 cm.sup.-1 (C--C--H bend), 1270 cm -1, 1164 cm-1, 1116
cm.sup.-1 (C--C+C--O stretch), 1070 cm.sup.-1 (C1-H stretch), 914
cm-1 (O-C1-H1 bend), and 840 cm.sup.-1 (C--C stretch) are known to
be signatures of crystalline glucose (Soderholm et al., J. Raman
Sprectrosc. 30:1009 [1999]). The literature has shown that SER
spectral bands shift up to 25 cm-1 when compared to the normal
Raman scattering bands of the same analyte (Stacy et al., Chem.
Phys. Lett. 102:365 [1983]). Peaks in the SERS difference spectrum
(FIG. 18; spectra F) at 1320, 1260, 1168, 1124, and 1076 cm-1
correspond with the Raman spectrum of crystalline glucose. In order
to evaluate the glucose departitioning, spectral subtraction of two
glucose-containing cycles was performed (FIG. 18; spectra H).
Spectra H shows spectral features that match with the glucose peaks
at 1320 and 1076 cm-1, but with lower intensities. Based on the
1076 cm-1 peak area, up to 33% of the glucose may remain in the EG3
layer after the 0 mM glucose cycle. The high glucose concentration
used in this experiment caused incomplete departitioning after each
cycle, and accordingly, the glucose accumulated in each step.
However, it is contemplated that physiological concentrations
(0-450 mg/dL, 0-25 mM) of glucose will not likely cause such
accumulation in the partition layer, and the natural flow of
aqueous humor (Vanlandingham et al., Am. J. Ophthal. 126:191
[1998]) and interstitial fluid will assist glucose
departitioning.
Selectivity of the Sensor for Glucose in the Presence of Blood
Serum Protein Mimic
[0141] Quantitative detection, temporal stability, and reusability
are preferred characteristics of a viable biosensor. It is also
preferred that the glucose sensor be effective in the presence of
interfering proteins. Serum albumin is a blood serum protein mimic
for challenging the glucose sensor. In this work, 1 mg/mL serum
albumin in saline was used after it was centrifuged, and the
supernatant was filtered to remove any undissolved particulate.
FIG. 18; spectra A shows the SER spectrum of the EG3-modified AgFON
substrate in saline (.lamda.ex=632.8, Plaser=0.8 mW, acquisition
time (t)=240 s). When the serum albumin solution was injected into
the flow cell, the SERS spectrum was collected throughout the
240-second incubation (FIG. 19; spectra B). Finally, the sample was
exposed to 100 mM glucose, and the SER spectrum was collected (FIG.
19; spectra C). Spectra D is the difference spectrum demonstrating
that serum albumin does not have a measurable SER spectrum. The
present invention is not limited to a particular mechanism. Indeed,
an understanding of the mechanism is not necessary to practice the
present invention. Nonetheless, it is contemplated that the lack of
SERS serum albumin bands is either due to the small Raman
scattering cross section of serum albumin or inefficient adsorption
of serum albumin to the EG3 partition layer. Spectra E demonstrates
that the SERS glucose sensor is still effective after substrate
exposure to an interfering protein. The peaks at 1449, 1433, 1339,
1291, 1108, 1077, 1059, and 855 cm.sup.-1 (FIG. 19; spectra E)
correspond with the crystalline glucose peaks shown in spectra F.
This experiment shows that glucose partitioning into EG3 is not
affected by the presence of large molecules such as serum albumin.
The peak at 695 cm.sup.-1 (FIG. 19; spectra A) shifts to 710
cm.sup.-1 (spectra C) in the presence of glucose. This shift may be
due to the rearrangement of the SAM when the glucose molecules
partition into EG3. The observed shift further supports the
hypothesis of glucose penetrating deeply into the EG3 monolayer,
affecting even the character of the C--S bond.
Example 3
In Vivo Glucose Analysis
[0142] In some embodiments, an in vivo animal system is used to
test the nanobiosensors of the present invention. A SERS biosensor
(e.g., a fiber optic glucose sensor) is quantified in 10
Sprague-Dawley rats. Diabetes is induced with a single injection of
streptozotocin (35 mg/kg) given IP. The blood glucose levels of
each rat are measured daily using blood drawn from the dorsal tail
vein until a diabetic state is confirmed by glucose measurements
over 200 mg/dL.
[0143] In order to implant the sensors, the rats are anesthetized
with 50 mg/kg sodium pentobarbital given IP. Every hour, or earlier
if the rat responds to external stimuli, an additional dose (1/5 of
the original dose) is given. The hair on the abdomen of the animal
is removed with an electric razor following assurance that the
animal does not feel pain. The skin is then scrubbed with a tamed
iodine soap. Subsequent to sterilization of the skin surface, an
approximately 10 mm long incision is made in the abdominal skin. A
separation in the fascial plane between the skin and underlying
abdominal muscles is created using blunt dissection with sterile
surgical scissors and forceps. Prior to delivery, the optical fiber
tip is placed 2/3 of the way down the barrel of a standard 25 gauge
hypodermic needle. The barrel of the needle and fiber are passed
through a silicon membrane such that when the membrane sits flush
on the rat skin surface the fiber tip is just subcutaneous. The
membrane adheres to the surface of the rat skin. The needle is then
withdrawn from the skin leaving the optical fiber tip in place. The
membrane closes around the optical fiber, holding it firmly. The
fiber is held in place with a suture or adhesive and the skin is
closed with 5-0 nylon suture.
[0144] The proximal end of the optical fiber is connected to
instrumentation for collection of Raman spectra. The glucose levels
of the rats are varied by IV injection of glucose and insulin thru
an indwelling IV catheter. The actual blood glucose is monitored
using a standard laboratory system (e.g., those commercially
available from YSI or Beckman). The glucose level is varied from
approximately 40-500 mg/dL. The glucose concentration in
interstitial fluid is allowed to stabilize for approximately 10-15
minutes and then Raman spectra are collected continuously. The
sensor is left in place for at least 5 days to monitor the accuracy
and durability of the sensor. The rats are euthanized with an
overdose of sodium pentobarbital (150 mg/kg) given IP following
surgery.
[0145] The collected Raman spectra are analyzed chemometrically and
interstitial glucose levels are determined and correlated with
blood glucose measurements. The results are plotted on a Clarke
error grid.
[0146] The rat model is also used to test the control of glucose
levels using a feedback loop and insulin delivery system. The
diabetic rat is constrained and catherized for infusion of glucose
and insulin and withdrawal of blood for glucose measurements.
Insulin is delivered subcutaneously using a standard catheter set
from MiniMed via a catheter connected to a motor-driven syringe
pump. The speed of insulin delivery (i.e., the motor speed) is
modulated based on the glucose level. A
proportional-integral-differential (PID) controller with upper and
lower limit constraints in the feedback loop is used to determine
the amount of insulin to be injected. Initial active variation in
the PID parameters is used to achieve reasonable control with
limited oscillations in the blood glucose. The control system is
challenged with periodic injections of glucose and insulin.
Example 4
Biosensors Utilizing Decanethiol/Mercaptohexanol Partition
Layers
[0147] This Example describes characterization of glucose-sensing
biosensors comprising decanethiol/mercaptohexanol mixed partition
layers as an example of multiple component SAMs.
A. Methods
Materials
[0148] All the chemicals were reagent grade or better, and used as
purchased. Silver pellets (99.99%) were purchased from Kurt J.
Lesker Company (Clairton, Pa.). Oxygen free high conductivity
copper was obtained from McMaster-Carr (Chicago, Ill.) and cut into
18 mm diameter disks. To clean substrates, NH.sub.4OH,
H.sub.2O.sub.2, and CH.sub.3CH.sub.2OH were used from Fisher
Scientific (Fairlawn, Va.). Surfactant-free, white
carboxyl-substituted latex polystyrene nanosphere suspensions
(390+19.5 nm diameter, 4% solid) were purchased from Duke
Scientific Corporation (Palo Alto, Calif.). Ultrapure water (18.2
M.OMEGA.cm.sup.-1) from a Millipore system (Marlborough, Mass.) was
used for substrate and solution preparation. Bovine plasma was
obtained from Hemostat Laboratories (Dixon, Calif.). Glucose,
lactate, and urea were purchased from Sigma (St. Louis, Mo.).
Decanethiol (CH.sub.3(CH.sub.2).sub.9SH), and 6-mercapto-1-hexanol
(HS(CH.sub.2).sub.6OH) were purchased from Aldrich (Milwaukee,
Wis.). Disposable filters, pore size 0.45 .mu.m, were acquired from
Gelman Sciences (Ann Arbor, Mich.).
AgFON Fabrication and Incubation Procedure
[0149] The copper substrates were cleaned by sonicating in 5:1:1
H.sub.20/30% H.sub.2O.sub.2/NH.sub.4OH. Approximately 10 .mu.l of
nanosphere solution was drop-coated onto a clean copper substrate
and allowed to dry at room temperature. Then, 200 nm thick Ag films
were deposited onto and through the nanosphere mask using the Kurt
J. Lesker electron beam deposition system (Clairton, Pa.) to form
AgFON substrates. The mass thickness and deposition rate (2
.ANG./s) of the Ag metal were measured by 6 MHz gold plated
quartz-crystal microbalance purchased from Sigma Instruments (Fort
Collins, Colo.). AgFON substrates were first incubated in 1 mM DT
in ethanol for 45 minutes and then transferred to 1 mM MH in
ethanol for at least 12 hours (FIG. 20). Then the
SAM-functionalized surfaces were mounted into a small volume flow
cell for SER spectra collection.
Surface-Enhanced Raman Spectroscopy
[0150] A Spectra-Physics model Millennia Vs laser
(.lamda..sub.ex=532 nm) was used to excite a Spectra-Physics model
3900 Ti-sapphire laser to produce the 785 nm excitation wavelength
(.lamda..sub.ex); the laser spot size on the sample was less than
0.5 mm in diameter. This excitation wavelength was chosen to
minimize autofluorescence of proteins (Anderson and Parrish, The
Science of Photomedicine; Plenum Press: New York, p 147 [1982];
Weissleder, Nat. Biotechnol. 19:316 [2001]). The SERS measurement
system includes an interference filter, an edge filter (Semrock,
Rochester, N.Y.), a model VM-505 single-grating monochromator with
the entrance slit set at 100 .mu.m (Acton Research Corp., Acton,
Mass.), and a LN.sub.2-cooled CCD detector (Roper Scientific,
Trenton, N.J.). A collection lens with magnification 5 was used to
collect the scattered light. The small volume flow cell was used to
control the external environment of the AgFON surfaces throughout
the SERS experiments.
Quantitative Multivariate Analysis
[0151] All data processing was performed using MATLAB (MathWorks,
INC., Natick, Mass.) and PLS_Toolbox (Eigenvector Research, Inc.,
Manson, Wash.). Prior to analysis, the spectra were smoothed using
the Savitsky-Golay method with a second order polynomial and window
size of 9. Cosmic rays were removed from the spectra using a
derivative filter. The slowly varying background, commonly seen in
SERS experiments was removed by subtracting a fourth-order
polynomial fit. This method greatly reduced varying background
levels with minimum effect on the SERS peaks. The chemometric
analysis was performed using the partial least-squares (PLS) method
and leave-one-out (LOO) cross validation algorithm.
Time Constant Analysis
[0152] The data was processed using PeakFit 4.12 software (Systat
Software Inc, Richmond, Calif.). To remove the varying background
in SER spectra, a fourth order polynomial was subtracted from the
baseline using MATLAB software. The spectra were further
preprocessed in PeakFit with linear best fit baseline correction
and Savitsky-Golay smoothing. The amplitude of the Raman bands was
obtained by fitting the data to the superposition of the Lorentzian
amplitude lineshapes.
B. Results
[0153] The results presented below show significant advancement
towards an implantable, real-time continuous SERS based glucose
sensor. Our previous work (See Example 1) demonstrated the ability
to detect glucose with SERS using decanethiol as the partition
layer. Subsequently, EG3 was used as the partition layer because it
is biocompatible and has the ability to resist nonspecific binding
of proteins. (See Example 2). Moreover, stability, reversibility,
and resistance to serum protein interference of the
EG3-functionalized glucose sensor were demonstrated. Finally, the
SERS based glucose sensor was optimized for NIR laser excitation
wavelength with the EG3 partition layer to reduce photodamage of
tissue and optimize signal on Au surfaces (Stuart et al., Anal.
Chem. 77: 4013 [2005]). This development showed enhanced spectral
stability and gave more accurate measurements. However, due to the
intricate synthesis, availability of EG3 is scarce. In the present
Example, a new mixed SAM layer, consisting of decanethiol (DT) and
mercaptohexanol (MH) has been developed. We also demonstrate (1)
long term stability of the DT/MH-functionalized AgFON surface (2)
reversibility of the sensor (3) quantitative measurement of
glucose, and (4) real-time partitioning and departitioning of the
glucose sensor.
Temporal Stability of DT/MH-Modified Substrate.
[0154] An implantable glucose sensor must be stable for at least 3
days (Heller, Annu. Rev. Biomed. Eng. 1:153 [1999]). In previous
work, we have demonstrated that SAM-functionalized AgFON substrates
were stable for at least 3 days in phosphate buffered saline by
electrochemical and SERS measurements (Stuart et al., Anal. Chem.
77:4013 [2005]). Here, we demonstrate the stability of the
DT/MH-functionalized AgFON surface for 10 days in bovine plasma
(FIG. 21). SER spectra were captured every 24 hours from three
different samples and three spots on each sample
(.lamda..sub.ex=785 nm, acquisition time (t)=2 min). FIG. 21A
represents the DT/MH spectrum acquired on day 2. FIG. 21B shows the
average intensity of the 1119 cm.sup.-1 peak for DT/MH on the AgFON
for each day as a function of time. The 1119 cm.sup.-1 band
corresponds to a symmetric stretching vibration of a C--C bond
(Bryant et al., J. Am. Chem. Soc. 113, 8284, [1991]). The change in
intensity of the 1119 cm.sup.-1 peak from the first day to the last
day is 2.08%, indicating that it did not vary significantly during
the ten day period. The 2% change in the intensity can be
attributed to the rearrangement of the SAM during the incubation in
bovine plasma (Biebuyck et al., Langmuir 10:1825 [1994]). The
temporal stability of the 1119 cm.sup.-1 peak intensity indicates
that the DT/MH SAM was intact and well ordered, making this
SAM-functionalized surface a potential candidate for an implantable
sensor.
Reversible Glucose Sensing
[0155] An implantable glucose sensor must also be reversible in
order to successfully monitor fluctuation in glucose concentration
throughout the day. To demonstrate the reversibility of the sensor,
the DT/MH-modified AgFON sensor was exposed to cycles of 0 and 100
mM aqueous glucose solutions (pH.about.7) without flushing the
sensor in between measurements to simulate real-time sensing (FIG.
22 inset). Nitrate was used as an internal standard in all the
experiments (1053 cm.sup.-1 peak) to minimize effective laser power
fluctuations. The 1053 cm.sup.-1 band corresponds to a symmetric
stretching vibration of NO.sub.3.sup.- and was used to normalize
the spectra (Soderholm et al., J. Raman Spectrosc. 30:1009 [1999]).
SERS spectra were collected for each step (.lamda..sub.ex=532 nm,
P=10 mW, acquisition time (t)=20 min) (FIG. 22A, 22B, 22C, 22D).
FIG. 22E shows the normal Raman spectrum of a saturated aqueous
glucose solution for comparison. In the normal Raman spectrum of a
saturated aqueous glucose solution, peaks at 1462, 1365, 1268,
1126, 915, and 850 cm.sup.-1 correspond to crystalline glucose
peaks. The difference spectra (FIG. 22F, 22G) represent
partitioning of glucose in DT/MH SAM, which clearly show the
glucose features at 1461, 1371, 1269, 1131, 916, and 864 cm.sup.-1.
This corresponds to the peaks in the normal Raman spectrum of
glucose in aqueous solution (FIG. 22E). SERS bands can shift up to
25 cm.sup.-1 when compared to normal Raman bands of the same
analyte (Stacy et al., Chem. Phys. Lett. 102:365 [1983]). The sharp
peak seen in all of the difference spectra at 1053 cm.sup.-1
represent imperfect subtraction of the nitrate internal standard.
The absence of glucose spectral features in the difference spectra
(FIG. 22H, 22I) represents complete departitioning of glucose. The
DT/MH mixed SAM presents a completely reversible sensing surface
for optimal partitioning and departitioning of glucose.
Quantitative Detection of Glucose Using DT/MH Partition Layer
[0156] In order for a glucose sensor to be viable, it should be
able to detect glucose in the clinically relevant range 10-450
mg/dL (0.56-25 mM), under physiological pH, and in the presence of
interfering analytes (FIG. 23). The data is presented in the Clarke
error grid, a standard for evaluating the reliability of glucose
sensors in the clinically relevant concentration range (10-450
mg/dL) (Clarke et al., Diabetes Care 10:622 [1987]). Data points
that fall in the A and B range are acceptable values. Values
outside the A and B range result in potential failure to detect
blood glucose levels outside of the target range and erroneous
diagnosis. DT/MH-functionalized AgFON samples were placed in a flow
cell containing water (pH.about.7) with lactate (1 mM) and urea
(2.5 mM) in physiological concentrations, which are potential
interferents for glucose detection. Glucose solutions ranging from
10 to 450 mg/dL with lactate and urea were then randomly introduced
in the cell and incubated for 2 min to ensure complete
partitioning. SER spectra were collected using two substrates and
multiple spots with a near-infrared laser source
(.lamda..sub.ex=785 nm, P=8.4 mW, acquisition time (t)=2 min). A
calibration model was constructed using partial least squares
leave-one-out (PLS-LOO) analysis with 46 randomly chosen
independent spectral measurements of known glucose concentrations.
The calibration model was based upon 7 latent variables that take
into account variation in laser power, the environment in the lab,
and SERS enhancement at different locations. The PLS analysis
results in a root mean square error of calibration (RMSEC) of 9.89
mg/dL (0.549 mM). This RMSEC value is lower than that reported in
our previous work using the EG3-modified AgFON.
[0157] In addition to having a low RMSEC, it is important to use an
independent validation set to test the calibration model (Beebe et
al., Chemometrics: A Practical Guide; Wiley Interscience: New York,
[1998]). For this model, a set of 23 data points was used to
validate the model. The root mean square error of prediction
(RMSEP) was calculated to be 92.17 mg/dL (5.12 mM). FIG. 23 depicts
that 98% of the calibration points and 87% of the validation points
fall in the A and B range of the Clarke error grid. The RMSEP can
be improved by increasing the number of data points in the
calibration set.
[0158] To transition from the in vitro sensor to an in vivo sensor,
the sensor should also demonstrate quantitative detection in a more
complex medium. Bovine plasma was used to simulate the in vivo
environment of an implantable glucose sensor. Prior to use, bovine
plasma was filtered using 0.45 .mu.m diameter pore size. The
filtered plasma was then spiked with glucose concentrations ranging
from 10-450 mg/dL. DT/MH-functionalized AgFON substrates were
placed in the flow cell and exposed to glucose spiked bovine
plasma. SERS spectra were collected at each concentration using
multiple samples and multiple spots in random order to construct a
robust calibration model (.lamda..sub.ex=785 nm, P=10-30 mW,
acquisition time (t)=2 min). Calibration was constructed using
PLS-LOO analysis described above using 7 latent variables and
presented in a Clarke error grid (FIG. 24). To construct the
calibration, 92 randomly chosen data points were used, resulting in
an RMSEC of 34.3 mg/dL (1.90 mM). For the validation, 46 data
points were used with an RMSEP of 83.16 mg/dL (4.62 mM). In the
Clarke error grid, 98% for calibration and 85% for validation fall
in the A and B range. Errors in both experiments can be reduced by
the use of additional data points for the calibration. In addition,
error can also be attributed to variation in SERS enhancement at
different spots and different substrates (Haynes et al., J. Phys.
Chem. 107:7426 [2003]). The results show that the DT/MH-modified
AgFON glucose sensor is capable of making accurate glucose
measurements in the presence of many interfering analytes.
Real-Time Study of Partitioning and Departitioning of Glucose
[0159] In addition to reversibility, which is an important
characteristic for a viable sensor, the sensor should be able to
partition and departition glucose on a reasonable time scale. The
real-time response was examined in a system with bovine plasma
simulating the in vivo environment. In order to evaluate the
real-time response of the sensor, the 1/e time constant for
partitioning and departitioning was calculated.
[0160] A DT/MH-functionalized AgFON was placed in bovine plasma for
.about.5 hours. The AgFON surface was then placed in a flow cell.
SER spectra were collected continuously (.lamda..sub.ex=785 nm)
with a 15 second integration time. To observe partitioning, 50 mM
glucose solution in bovine plasma was injected at t=0. At t=225
sec, 0 mM glucose solution in bovine plasma was injected into the
flow cell to evaluate the departitioning of glucose. An excitation
wavelength of 785 nm was used to reduce autofluorescence caused by
proteins (Anderson and Parrish, The Science of Photomedicine;
Plenum Press: New York, p 147 [1982], Weissleder, Nat. Biotechnol.
19:316 [2001]). The amplitude was then plotted versus time as shown
in FIG. 25C. The 1/e time constant was calculated from the
exponential curve fitted to the data points.
[0161] The spectra shown in FIGS. 25A and 25B demonstrate real-time
amplitude changes in the 1462 cm.sup.-1 peak as glucose
concentrations vary. The amplitude of the 1462 cm.sup.-1 peak was
obtained by fitting the data to the superposition of three
Lorentzian lineshapes using PeakFit. The 1/e time constant is 28
seconds for partitioning, and 25 seconds for departitioning,
calculated from the exponential fit (FIG. 25C). These experiments
demonstrate that partitioning and departitioning occur rapidly
making the DT/MH SERS-based glucose sensor suitable for
implantable, continuous sensing.
Example 5
In Vivo Glucose Analysis
[0162] This Example describes the characterization of
glucose-sensing biosensors comprising DT/MH partition layers in
vivo.
A. Methods
Materials
[0163] The SAM functionalized SERS active surfaces were prepared in
two steps, as shown in FIG. 26. First, AgFON surfaces were
fabricated by drop-coating 10 .mu.l of 390 nm diameter nanosphere
solution onto clean copper substrates, and then depositing a
200-nm-thick Ag film onto the nanosphere mask. The AgFON substrates
were then incubated in 1 mM decanethiol (DT) for 45 minutes and
subsequently transferred to 1 mM mercaptohexanol (MH) solution for
at least 12 hours to form a mixed DT/MH SAM. The substrates were
kept in 1 mM MH in ethanol prior to surgically implantation.
Animal Care and Preparation
[0164] The surgical procedure followed the protocol filed with
Northwestern University and in accord with IACUC provisions.
Sprague-Dawley rats (300-500 g, N=4 were anesthetized with
pentobarbital (Office of Research Safety, Northwestern University)
with an initial dose of 50 mg/kg IP. The animals were checked for
pain reactions by toe-tug and blink tests. The rats were kept
anesthetized by hourly administration of pentobarbital at 25 mg/kg.
After the anesthetic had taken effect, the surgical areas were
prepared by removal of hair (shaving and chemical depilatory) and
cleaning. Then, the femoral vein was cannulated using PE 50 tubing
(Clay Adams) for glucose injections. The carotid artery was
cannulated with PE 90 tubing for blood glucose measurements with
FDA-qualified home medical equipment (One Touch II Meter). A
tracheotomy and intubation was performed to enable the attachment
of a ventilator to aid breathing. The incisions were shut with
surgical clips. The rat was thermally stabilized by an electric
heating pad throughout the course of the surgery and experiment. A
metal frame containing a glass window was placed along the midline
of the rat's back. A circular incision was made to allow the
positioning of a DT/MH functionalized AgFON substrate
subcutaneously such that the substrate was in contact with the
interstitial fluid, and optically addressed through the window.
Following the experiment, the animals were sacrificed with an
overdose of anesthetic and bilateral thorachotomy.
Experimental Protocols
[0165] Glucose was varied in the rat through intermittent
intravenous infusion for three hours. A bolus of glucose was
delivered over 5 to 10 min, at a concentration of 1 g/mL in sterile
phosphate buffered saline. A droplet of blood was drawn from the
rat, the glucose level was measured with the One Touch II
glucometer, and corresponding SERS measurements were taken. The
SERS spectra were acquired through the optical window using a
Ti:Saph laser (.lamda..sub.ex=785 nm, P=50 mW, acquisition time
(t)=2 min).
Data Analysis
[0166] The data were collected and analyzed by the partial least
squares method (Yonzon et al., Analytical Chemistry 76:78 [2004];
Stuart et al., Analytical Chemistry 77:4019 [2005]; Lyandres et
al., Analytical Chemistry 77:6134 [2005]; Shafer-Peltier et al.,
Journal of the American Chemical Society 125:588 [2003]). FIG. 27
shows corresponding glucose concentration variations in an
exemplary rat using SERS and the One Touch II blood glucose meter.
Both the standard glucometer and the SERS-based measurements
tracked the change in glucose concentration after infusion i.e., a
sharp rise in glucose concentration is detected by both techniques
after the start of the glucose infusion (t=60 min). FIG. 28 plots
the time-independent data on the Clarke error grid to more
precisely gauge the performance of the SERS measurement system. The
Clarke error grid was developed as a convenient and
modality-independent means to compare the accuracy and performance
of glucose sensors in the clinically relevant range Clarke et al.,
Diabetes Care 10:622 [10987]). The grid is divided into five zones.
Predictions within these zones lead to: A) clinically correct
measurement and treatment, B) benign errors or no treatment, C)
incorrect measurements leading to overcorrection of acceptable
glucose levels, D) dangerous failure to detect and treat, and E)
treatments that further aggravate abnormal glucose levels.
B. Results
[0167] The majority of measurements from all samples fell within
the acceptable range. FIG. 28 shows a representative Clarke error
grid analysis of a single rodent. The 26 measurements are from a
single spot on the implanted DT/MH functionalized AgFON surface.
The calibration set was constructed using 21 data points that were
correlated with the commercial glucometer. The validation set was
constructed using 5 independent data points. The sensor had
relatively low error (RMSEC=7.46 mg/dL (0.41 mM) and RMSEP=53.42
mg/dL (2.97 mM). These data compare favorably with our previous in
vitro results (See Examples 1, 2 and 4) as well as those of other
optically based glucose measurements Solnica et al., J. Clinica
Chimica Acta 331:29 [2003]).
[0168] Important parameters governing the overall efficacy of a
given sensor are its response time, reversibility, and long-term
stability. Previous research has demonstrated that FON based
sensors are stable with good reversibility under a variety of
conditions. The present work shows that the SERS sensors of the
present invention show rapid response to detect the glucose
injection, keeping pace with a conventional glucometer. Because it
is impossible to rapidly and accurately vary glucose levels in
vivo, additional in vitro experiments were conducted to determine
whether the sensor is capable of exhibiting response times rapid
enough for continuous or semi-continuous monitoring (i.e.,
.ltoreq.the 2 min collection time). After being implanted in the
rat for 5 hr, a DT/MH functionalized AgFON surface was removed and
placed in a flow cell containing bovine plasma to simulate the in
vivo environment. Then, the 1/e time constant response to a step
change in glucose concentration was determined (FIG. 29). The AgFON
surface was exposed to 50 mM glucose in plasma at t=0 seconds and
then flushed with plasma at t 225 seconds. SERS spectra were
collected every 15 seconds (.lamda..sub.ex=785 nm, P=100 W). Based
upon amplitude calculations for the 1462 cm.sup.-1 peak, the 1/e
time constant was 9 seconds for partitioning, and 27 seconds for
departitioning. These values indicate that glucose binds reversibly
to the SERS active surface, and that changes in concentration as
rapid as .about.30 seconds can be detected spectroscopically.
[0169] All publications and patents mentioned in the above
specification are herein incorporated by reference. Various
modifications and variations of the described method and system of
the invention will be apparent to those skilled in the art without
departing from the scope and spirit of the invention. Although the
invention has been described in connection with specific preferred
embodiments, it should be understood that the invention as claimed
should not be unduly limited to such specific embodiments. Indeed,
various modifications of the described modes for carrying out the
invention which are obvious to those skilled in the relevant fields
are intended to be within the scope of the present invention.
* * * * *