Detection of a Biomarker for Alzheimer’s Disease from
Synthetic and Clinical Samples Using a Nanoscale Optical
Amanda J. Haes,†,§Lei Chang,‡William L. Klein,*,‡and Richard P. Van Duyne*,†
Contribution from the Department of Chemistry, Northwestern UniVersity, 2145 Sheridan Road,
EVanston, Illinois 60208-3113, and Department of Neurobiology and Physiology,
Northwestern UniVersity, 2205 Tech DriVe, EVanston, Illinois 60208
Received September 28, 2004; E-mail: email@example.com; firstname.lastname@example.org
Abstract: A nanoscale optical biosensor based on localized surface plasmon resonance (LSPR)
spectroscopy has been developed to monitor the interaction between the antigen, amyloid-? derived diffusible
ligands (ADDLs), and specific anti-ADDL antibodies. Using the sandwich assay format, this nanosensor
provides quantitative binding information for both antigen and second antibody detection that permits the
determination of ADDL concentration and offers the unique analysis of the aggregation mechanisms of
this putative Alzheimer’s disease pathogen at physiologically relevant monomer concentrations. Monitoring
the LSPR-induced shifts from both ADDLs and a second polyclonal anti-ADDL antibody as a function of
ADDL concentration reveals two ADDL epitopes that have binding constants to the specific anti-ADDL
antibodies of 7.3 × 1012M-1and 9.5 × 108M-1. The analysis of human brain extract and cerebrospinal
fluid samples from control and Alzheimer’s disease patients reveals that the LSPR nanosensor provides
new information relevant to the understanding and possible diagnosis of Alzheimer’s disease.
The development of highly sensitive and selective biological
sensors for the monitoring of disease biomarkers is an important
motivation for nanoscience research. The localized surface
plasmon resonance (LSPR) nanosensor has been demonstrated
to be an effective platform for the quantitative detection of
biological and chemical species.1-8The signal transduction
mechanism of the LSPR nanosensor is based on its sensitivity
to local refractive index changes near the surfaces of substrate-
confined, size- and shape-controlled, silver and gold nano-
particles.9-11The LSPR of noble metal nanoparticles arises
when electromagnetic radiation induces a collective oscillation
of the conduction electrons of the individual nanoparticles9,10,12-18
and has two primary consequences: (1) selective photon
absorption which allows the optical properties of these nano-
particles to be monitored with UV-vis spectroscopy and (2)
the enhancement of the electromagnetic fields surrounding the
nanoparticles which is responsible for all surface-enhanced
spectroscopies. The reader is referred to recent reviews14,16,19-21
for a detailed description of the physics behind LSPR spectros-
copy. The LSPR spectrum is measured by either transmission
extinction spectroscopy11,22or dark-field light scattering spec-
troscopy.23,24The extinction spectra of the nanoparticles exhibit
easily measured wavelength shifts that correspond to small
changes in the refractive index within the electromagnetic fields
surrounding the nanoparticles. It is well established that the
extinction spectra of Ag and Au nanoparticles synthesized using
nanosphere lithography (NSL) have refractive index sensitivities
†Department of Chemistry.
‡Department of Neurobiology and Physiology.
§Current address: Naval Research Laboratory, Washington D.C.
(1) Haes, A. J.; Hall, W. P.; Chang, L.; Klein, W. L.; Van Duyne, R. P. Nano
Lett. 2004, 4, 1029.
(2) Haes, A. J.; Van Duyne, R. P. J. Am. Chem. Soc. 2002, 124, 10596.
(3) Haes, A. J.; Zou, S.; Schatz, G. C.; Van Duyne, R. P. J. Phys. Chem. B
2004, 108, 6961.
(4) Haes, A. J.; Zou, S.; Schatz, G. C.; Van Duyne, R. P. J. Phys. Chem. B
2004, 108, 109.
(5) Malinsky, M. D.; Kelly, K. L.; Schatz, G. C.; Van Duyne, R. P. J. Am.
Chem. Soc. 2001, 123, 1471.
(6) Riboh, J. C.; Haes, A. J.; McFarland, A. D.; Yonzon, C. R.; Van Duyne,
R. P. J. Phys. Chem. B 2003, 107, 1772.
(7) Englebienne, P. Analyst 1998, 123, 1599.
(8) Hirsch, L. R.; Jackson, J. B.; Lee, A.; Halas, N. J.; West, J. L. Anal. Chem.
2003, 75, 2377.
(9) El-Sayed, M. A. Acc. Chem. Res. 2001, 34, 257.
(10) Link, S.; El-Sayed, M. A. J. Phys. Chem. B 1999, 103, 8410.
(11) Haynes, C. L.; Haes, A. J.; Van Duyne, R. P. Mater. Res. Soc. Symp. Proc.
2001, 635, C6.3/1.
(12) Haynes, C. L.; Van Duyne, R. P. J. Phys. Chem. B 2001, 105, 5599.
(13) Mulvaney, P. MRS Bull. 2001, 26, 1009.
(14) Kreibig, U.; Gartz, M.; Hilger, A.; Hovel, H. Optical investigations of
surfaces and interfaces of metal clusters. In AdVances in Metal and
Semiconductor Clusters; Duncan, M. A., Ed.; JAI Press Inc.: Stamford,
CT, 1998; Vol. 4, p 345.
(15) Mulvaney, P. Langmuir 1996, 12, 788.
(16) Kreibig, U. Optics of nanosized metals. In Handbook of Optical Properties;
Hummel, R. E., Wissmann, P., Eds.; CRC Press: Boca Raton, FL, 1997;
Vol. II, p 145.
(17) Kreibig, U.; Gartz, M.; Hilger, A. Ber. Bunsen-Ges. 1997, 101, 1593.
(18) Jensen, T. R.; Kelly, K. L.; Lazarides, A.; Schatz, G. C. J. Cluster Sci.
1999, 10, 295.
(19) Schatz, G. C.; Van Duyne, R. P. Electromagnetic Mechanism of Surface-
Enhanced Spectroscopy; Wiley: New York, 2002; Vol. 1.
(20) Haes, A. J.; Van Duyne, R. P. Expert ReView of Molecular Diagnostics
2004, 4, 527.
(21) Haes, A. J.; Stuart, D. A.; Nie, S.; Van Duyne, R. P. J. Fluoresc. 2004, 14,
(22) Jensen, T. R.; Malinsky, M. D.; Haynes, C. L.; Van Duyne, R. P. J. Phys.
Chem. B 2000, 104, 10549.
(23) Van Duyne, R. P.; Haes, A. J.; McFarland, A. D. SPIE 2003, 5223, 197.
(24) McFarland, A. D.; Van Duyne, R. P. Nano Lett. 2003, 3, 1057.
Published on Web 00/00/0000
10.1021/ja044087q CCC: $30.25 © xxxx American Chemical Society
J. AM. CHEM. SOC. XXXX, XXX,
PAGE EST: 7.8
of ∼200 nm/RIU3,5and a sensing volume that can be tuned by
controlling nanoparticle size, shape, and composition.3,4
The need for ultrasensitive detection methods for biological
and chemical sample screening is an important issue in disease
diagnosis and mechanistic understanding. For example, Alz-
heimer’s disease (AD), a progressive neurodegenerative disease
for which there is neither a cure nor a good clinical diagnostic,
is the leading cause of dementia in people over age 65 and
affects over four million Americans. Genetic, biochemical, and
animal model studies strongly suggest a central role for
amyloid-? (A?) in the pathogenesis of AD.25A? is a 42-amino
acid peptide that was first discovered as the monomeric subunit
of the large insoluble amyloid fibrils of AD plaques. In the past
several years, it has been found that, in addition to plaques, A?
will also self-assemble into small soluble oligomers, termed
ADDLs, (amyloid-derived diffusible ligands), and that ADDLs
will cause neurological dysfunctions relevant to memory.26-28
Two comprehensive reviews of current data have concluded that
early stages of the disease are likely to be caused by the synaptic
impact of such soluble oligomeric assemblies of A?.29,30Despite
the fact that several molecules including A?1-40, A?1-42, tau,
phosphorylated tau, and seeded A? forms or their combination
have been suggested as potential AD biomarkers,31-33at the
present time, there is no definite clinical diagnosis for AD other
than autopsy. Recently, it has been demonstrated that ADDLs
are present at significantly elevated levels in autopsied brain
samples from humans with AD.34The association of ADDLs
with disease mechanisms and brain pathology suggests that a
sensitive means to detect ADDLs in body fluid could provide
a definitive molecular basis for the laboratory diagnosis of AD.
An ultrasensitive method for ADDLs detection potentially could
emerge from LSPR nanosensor technology, providing an
opportunity to develop the first clinical lab diagnostic for AD.
In this study, we combine LSPR nanosensor technology with
A? oligomer specific antibodies to determine if ADDLs are a
potential biomarker for AD. In comparison with other biosensing
approaches, the LSPR nanosensor allows for the analysis of
biological species in a surfactant-free environment. This is
extremely important because the absence of a surfactant allows
for biological species to be analyzed in their native state (not
one influenced by the presence of physiologically unnatural
molecules). In addition, the LSPR nanosensor is demonstrated
to be sufficiently sensitive for the detection of ultralow
concentrations of ADDLs in biological samples. By optimizing
the size of the nanoparticles, the sensing distance of the
nanoparticles extends ∼35 nm from the surface. This sensing
distance is required for the implementation of a sandwich assay
for the target ADDL molecules. Quantitative detection models
for both the ADDL and the second anti-ADDL antibody
response are developed to extract information regarding the
affinity constants between the antigen and antibody species and
to aid in the analysis of human samples. Human brain extracts
and cerebrospinal fluid (CSF) samples from control and diseased
patients exhibit drastically different responses, which indicate
the presence of elevated ADDL concentrations in the samples
from diseased patients in comparison to the control patients.
These preliminary results indicate that LSPR nanosensor
technology is useful as a screening method of human samples
for disease diagnosis as well as a more general approach to a
mechanistic understanding of diseases and drug target interac-
Methods and Materials
Materials. 11-Mercaptoundecanoic acid (11-MUA) and 1-octanethiol
(1-OT) were purchased from Aldrich (Milwaukee, WI). Phosphate-
buffered saline (PBS) solutions of 10 and 20 mM, pH 7.4, was
purchased from Sigma (St. Louis, MO). 1-Ethyl-3-[3-dimethy-lami-
nopropyl]carbodiimide hydrochloride (EDC) was obtained from Pierce
(Rockford, IL). Absolute ethanol was obtained from Pharmco (Brook-
field, CT). Methanol was purchased from Fisher Scientific (Pittsburgh,
PA). Silver wire (99.99%, 0.5-mm diameter) was purchased from D.
F. Goldsmith (Evanston, IL). Tungsten vapor deposition boats and
chromium rods were acquired from R. D. Mathis (Long Beach, CA).
Polystyrene spheres of 390 nm ( 19.5 nm were received as a suspension
in water from Duke Scientific (Palo Alto, CA). Millipore cartridges
(Marlborough, MA) were used to purify water to a resistivity of 18.2
MΩ cm-1. Triton X-100 was purchased from Aldrich (Milwaukee, WI).
Ruby red muscovite mica substrates were purchased from Asheville-
Schoonmaker Mica Co. (Newport News, VA).
Synthetic ADDL and Anti-ADDL Antibody Preparation. A?1-42
peptide (California Peptide Research, Napa, CA) was used to prepare
synthetic ADDLs according to published protocols.26An aliquot of
A?1-42was dissolved in anhydrous DMSO to a concentration of 22.5
mg/mL (5 mM), pipet mixed, and further diluted into ice-cold F12
medium (phenol red free) (BioSource, CA) at a ratio of 1:50. The
mixture was quickly vortexed, incubated at 6-8 °C for 24 h, and
centrifuged at 14 000g for 10 min. The oligomers were then collected
from the supernatant. The concentration of synthetic ADDLs was
determined by a micro-BCA assay (Pierce, Rockford, IL). A similar
protocol was followed for vehicle preparation except no A? peptide
was added. Antibodies targeting ADDLs in the LSPR immunoassay
(polyclonal M71 and monoclonal 20C2) were generated and character-
ized as previously described.35
CSF and Brain Extract Preparation. CSF samples were provided
by Dr. John Lee, Loyola University Medical School, Maywood, IL.
Premortem samples were obtained via lumbar puncture and kept frozen
until used. Tissues of cortex of AD (pathology diagnosis based on Braak
& Braak, CERAD, and NIA/Reagan Institute Criteria) and age-matched
control brains were obtained from Northwestern University Alzheimer’s
Brain Bank and stored at -80 °C before using. Brain extracts were
prepared following the literature.34One hundred milligrams of tissue
was homogenized in 1 mL Ham’s F12 medium (phenol-free BioSource,
Camarillo, CA) containing protease inhibitors (complete mini-EDTA
free tablet, Roche, Indianapolis, IN) on ice using A Tissue Tearor
(Biospec Products, Bartlesville, OK). After centrifugation at 20 000g
for 10 min, the supernatant was centrifuged at 100 000g for 60 min.
(25) Selkoe, D. J.; Hardy, J. Science 2002, 298, 963.
(26) Lambert, M. P.; Barlow, A. K.; Chromy, B. A.; Edwards, C.; Freed, R.;
Liosatos, M.; Morgan, T. E.; Rozovsky, I.; Trommer, B.; Viola, K. L.;
Wals, P.; Zhang, C.; Finch, C. E.; Krafft, G. A.; Klein, W. L. Proc. Natl.
Acad. Sci. U.S.A. 1998, 95, 6448.
(27) Walsh, D. M.; Klyubin, I.; Fadeeva, J. V.; Cullen, W. K.; Anwyl, R.; Wolfe,
M. S.; Rowan, M. J.; Selkoe, D. J. Nature 2002, 416, 535.
(28) Chromy, B. A.; Nowak, R. J.; Lambert, M. P.; Viola, K. L.; Chang, L.;
Velasco, P. T.; Jones, B. W.; Fernandez, S. J.; Lacor, P. N.; Horowitz, P.;
Finch, C. E.; Krafft, G. A.; Klein, W. L. Biochemistry 2003, 42, 12749.
(29) Mattson, M. P. Nature 2004, 430, 631.
(30) Hardy, J.; Selkoe, D. J. Science 2002, 297, 353.
(31) Hampel, H.; Mitchell, A.; Blennow, K.; Frank, R. A.; Brettschneider, S.;
Weller, L.; Moller, H. J. J. Neural Transmission 2004, 111, 247.
(32) Pitschke, M.; Prior, R.; Haupt, M.; Riesner, D. Nature Medicine 1998, 4,
(33) Seubert, P.; Vigopelfrey, C.; Esch, F.; Lee, M.; Dovey, H.; Davis, D.; Sinha,
S.; Schlossmacher, M.; Whaley, J.; Swindlehurst, C.; McCormack, R.;
Wolfert, R.; Selkoe, D.; Lieberburg, I.; Schenk, D. Nature 1992, 359, 325.
(34) Gong, Y.; Chang, L.; Viola, K. L.; Lacor, P. N.; Lambert, M. P.; Finch, C.
E.; Krafft, G. A.; Klein, W. L. Proc. Natl. Acad. Sci. U.S.A. 2003, 100,
(35) Lambert, M. P.; Viola, K. L.; Chromy, B. A.; Chang, L.; Morgan, T. E.;
Yu, J. X.; Venton, D. L.; Krafft, G. A.; Finch, C. E.; Klein, W. L. J.
Neurochem. 2001, 79, 595.
A R T I C L E S
Haes et al.
B J. AM. CHEM. SOC.
Protein concentration of 100 000g supernatant was determined by
standard BCA assay. Fractions were stored at -80 °C. Gloves were
worn at all times when handling all biological materials.
Nanoparticle Fabrication. Nanosphere lithography (NSL) was used
to create monodisperse, surface-confined Ag nanotriangles.2,4Polysty-
rene nanospheres (diameter ) 390 nm + 19.5 nm, Duke Scientific)
were diluted as a 1:1 solution with Triton X-100 and methanol (1:400
by volume). Approximately 4 µL of this solution was drop-coated onto
the freshly cleaved mica substrates2and allowed to dry, forming a
monolayer in a close-packed hexagonal formation, which served as a
deposition mask. The samples were mounted into a Consolidated
Vacuum Corporation vapor-deposition chamber. A Leybold Inficon
XTM/2 quartz crystal microbalance was used to monitor the thickness
of the metal being deposited. For all experiments, 25 nm of Ag (D. F.
Goldsmith) was evaporated onto the samples. Following metal deposi-
tion, the samples were sonicated for 3-5 min in ethanol (Pharmco) to
remove the polystyrene nanosphere mask, creating Ag triangular
nanoparticles on the mica substrate.
Ultraviolet-Visible Extinction Spectroscopy. Macroscale UV-
vis extinction measurements were collected using an Ocean Optics
(Dunedin, FL) S2000 fiber optically coupled spectrometer with a CCD
detector. All spectra in this study are macroscopic measurements
performed in standard transmission geometry with unpolarized light.
The probe diameter was approximately 1 mm. A home-built flow cell5
was used to control the external environment of the Ag nanoparticle
Nanoparticle Functionalization. Immediately following nanosphere
removal, the samples were incubated for 24-48 h in 3:1 1 mM 1-OT/1
mM 11-MUA solution (in ethanol). Next, the samples were rinsed
thoroughly with ethanol and dried with N2. Following SAM function-
alization, the samples were incubated in a 100 nM anti-ADDL antibody
solution in 10 mM PBS. Covalent linking was ensured via the presence
of 100 mM EDC during this 1-h incubation period. Exposing the sample
to varying concentrations of ADDL in 10 mM PBS for 30 min was
the first detection step of the immunoassay. The assay was completed
via the incubation of the sample in 100 nM anti-ADDL antibody in 10
mM PBS for 30 min. Experiments were performed using a monoclonal
antibody for the second antibody; however, no notable differences were
observed in the LSPR shift response data. For this reason, only
polyclonal antibodies were used for these sandwich assays.
Nonspecific interactions were monitored by exposing the SAM
functionalized nanoparticles to a 250 nM anti-ADDL antibody solution
(no EDC coupling agent) for 1 h. For the sandwich assay experiments
involving the human samples, following SAM modification, the samples
were incubated in a 100 nM anti-ADDL/100 mM EDC solution in 10
mM PBS for 1 h. Next, the sample was exposed to varying concentra-
tions of human brain extract or CSF for 30 min. Finally, the sample
was exposed to 100 nM ADDL solution for 30 min. Samples were
rinsed with cycles of 10 mM PBS and 20 mM PBS (with 0.1% Tween
20) following each functionalization step to ensure the removal of
nonspecifically bound species. For the binding curves in Figure 4A
and 4B, each data point represents averaged experimental results from
a minimum of two fresh samples.
Atomic Force Microscopy. Atomic force microscope (AFM) images
were collected using a Digital Instruments Nanoscope IV microscope
and Nanoscope IIIa controller operating in tapping mode. The resulting
AFM linescan analysis reveal that the bare triangular nanoparticles have
∼90-nm perpendicular bisectors and ∼25-nm heights.
Detection of ADDLs Using an Antibody Sandwich Assay.
For the first time, a sandwich assay has been accomplished using
the LSPR nanosensor technology. Figure 1A shows a schematic
representation of the nanoparticle structure and chemical func-
tionalization for the sandwich assay. Briefly, triangular Ag
nanoparticles (perpendicular bisectors ) 90 nm, heights ) 25
nm) are synthesized on mica substrates and functionalized with
antibodies specific for ADDLs. Next, the nanoparticle surface
is exposed to varying concentrations of synthetic ADDLs.
Finally, the nanosensor is exposed to a fixed concentration of
anti-ADDL antibody for 30 min to complete the assay. A
polyclonal anti-ADDL antibody was used in both instances of
By functionalizing the Ag nanoparticles with a self-assembled
monolayer (SAM), the stability of the samples is greatly
increased and consistent red shifts2are produced upon incubation
in given concentrations of ADDLs and the second anti-ADDL
antibody. Because it is hypothesized that the formation ADDLs
may be implicated in the early symptoms associated with AD,
an assay that targets these molecules at sub-nanomolar concen-
trations is desired. A sandwich assay that relies on the specific
interactions between ADDLs and their antibodies is the most
Figure 1. Design and experimental setup for the LSPR biosensor for the detection of ADDLs using a sandwich assay. Transmission UV-vis spectroscopy
is used to monitor the optical properties (LSPR) of Ag nanoparticles. The schematic illustration displays the sandwich assay and surface chemistry of the
LSPR nanosensor. First, surface-confined Ag nanoparticles (see AFM inset, nanoparticle width ) 90 nm, nanoparticle height ) 25.0 nm) are synthesized
using NSL on mica substrates. Next, a SAM layer consisting of a mixed monolayer of OT and MUA passivates the nanoparticles for nonspecific binding
and activates the nanoparticles for the attachment of the first anti-ADDL antibody, respectively. The first anti-ADDL antibody is covalently attached to the
nanoparticles via incubation in 100 mM EDC/100 nM anti-ADDL antibody solution for 1 h. Samples are then incubated in varying concentrations of ADDLs
for 30 min. Finally, to enhance the LSPR shift response of the ADDLs, the samples are incubated in a 100 nM anti-ADDL solution for an additional 30 min.
Detection of a Biomarker Using an Optical Biosensor
A R T I C L E S
J. AM. CHEM. SOC. C
straightforward method to analyze and detect these molecules.
To perform this sandwich assay, implementation of the opti-
mization of nanoparticle size and sensing distances were
employed.4That is, the nanoparticles were fabricated to have
electromagnetic fields or an effective sensing distance that
extends ∼35 nm away from their surface.4
Representative sandwich assays for the direct detection of
high and low concentrations of ADDLs are displayed in Figure
2. In this study, anti-ADDL antibody was specifically linked to
the SAM functionalized nanoparticles over a 1-h period. Next,
the nanosensor is exposed to a 10 pM ADDL (Figure 2A)
solution for 30 min. During the ADDL incubation, the extinction
maximum of the sample shifts from 718.5 nm (Figure 2A-1) to
722.7 nm (Figure 2A-2), a 4.2-nm shift. Next, this shift is
amplified by exposing the sample to an additional anti-ADDL
antibody for 30 min resulting in an additional 7.6-nm wavelength
shift or a total LSPR shift of 11.8 nm. When this sandwich
assay is repeated with 100 fM ADDL (Figure 2B), the LSPR
extinction maximum shifts 1.9 nm (from 741.7 to 743.6 nm)
upon exposure to ADDLs. An additional 3.4-nm shift is
observed upon incubation in the second antibody for a total
LSPR shift of 5.3 nm. The samples were checked for nonspecific
binding after step 1, and no nonspecific binding was observed.
Elimination of Nonspecific Binding Using Mica Substrates.
The development of a successful sensor protocol requires the
demonstration of minimal to no nonspecific binding. In our
previous studies for implementing the LSPR nanosensor tech-
nology with an immunoassay specific for targeting antibodies
of AD markers,1the LSPR nanosensor response was dominated
by nonspecific interactions at low target molecule concentra-
tions. The source of this undesirable response was attributed to
a molecular interaction with the Cr adhesion layer. While silane
passivation decreased this undesirable result, it did not totally
eliminate the problem. Presently, we have made great strides
to overcome the nonspecific binding observed with the Cr
containing nanoparticles. First, mica substrates have been
substituted in place of the glass substrates and the use of Cr as
an adhesion layer has been eliminated. The adhesion of silver
to mica is sufficiently great that the nanoparticles remain
surface-bound throughout an experiment.6Additionally, as seen
in Figure 3A, when a SAM functionalized nanoparticle surface
is exposed to 250 nM anti-ADDL solution for an hour without
the use of a coupling agent (EDC), no shift in the LSPR
wavelength is observed. This indicates that no antibody is
randomly attaching to the nanoparticle sample and that the
nonspecific interactions that had previously plagued the LSPR
nanosensor shift responses have been eliminated.
A second nonspecific binding experiment is performed to
analyze the influence of the biological vehicle (see Materials
and Methods for description) on the LSPR response of an
antibody functionalized nanoparticle sample (Figure 3B). To
test for this nonspecific interaction, the LSPR nanosensor is first
specifically functionalized with anti-ADDL antibodies using an
EDC coupling agent (λmax ) 798.4 nm, Figure 3B-1). Next,
the sensor surface was exposed to the biological vehicle for 30
min revealing a slight blue-shift to 797.4 nm, Figure 3B-2.
Finally, the sensor surface was exposed to 100 nM anti-ADDL
Figure 2. Demonstration of the LSPR sandwich assay at high and low
concentrations. (A) LSPR spectra for each step of the sandwich assay for
10 pM ADDL. Ag nanoparticles after modification with (A-1) 100 nM anti-
ADDL antibody, λmax) 718.5 nm, (A-2) 10 pM ADDL, λmax) 722.7 nm,
and (A-3) 100 nM anti-ADDL, λmax ) 730.3 nm. (B) LSPR spectra for
each step of the sandwich assay for 100 fM ADDL. Ag nanoparticles after
modification with (B-1) 100 nM anti-ADDL antibody, λmax) 741.7 nm,
(B-2) 100 fM ADDL, λmax) 743.6 nm, and (B-3) 100 nM anti-ADDL,
λmax) 747.0 nm. All spectra were collected in N2.
Figure 3. LSPR spectra illustrating the lack of nonspecific binding for the
LSPR nanosensor. (A) Demonstration of nonspecific binding elimination
from anti-ADDL antibody to SAM-functionalized nanoparticles on mica.
(A-1) Ag nanoparticles after modification with 1:3 1 mM 11-MUA:1-OT,
λmax) 596.6 nm, and (A-2) Ag nanoparticles after exposure to 250 nM
anti-ADDLs (in the absence of EDC), λmax) 596.6 nm. (B) Demonstration
of the absence of nonspecific binding from a vehicle solution to anti-ADDL
antibody functionalized nanoparticles on mica. Ag nanoparticles after
modification with (B-1) 100 nM anti-ADDL antibody, λmax) 798.4 nm,
(B-2) vehicle, λmax ) 797.4 nm, and (B-3) 100 nM anti-ADDL, λmax )
797.1 nm. All spectra were collected in a N2environment.
A R T I C L E S
Haes et al.
D J. AM. CHEM. SOC.
antibody for an additional 30 min. Once again, a slight blue-
shift in the LSPR extinction maximum to 797.1 nm (Figure 3B-
3) is observed. These small blue-shifts, which indicate that no
additional mass is being detected at the nanoparticle sensor chip,
can be attributed either to the removal of a small amount of
covalently bound anti-ADDL antibody from the substrate or to
slight variations in spectrometer noise. Considering nonspecific
binding and the peak-to-peak wavelength shift noise of the
baseline in repetitive experiments (∼0.3 nm), one can conser-
vatively estimate the smallest, reliable LSPR wavelength shift
as 3 times this value (∼0.9 nm).
Quantification of ADDL and Second Anti-ADDL Anti-
body Response. To thoroughly investigate the influence of
varying concentrations of ADDLs on the LSPR nanosensor
response, the shift induced by the direct binding of ADDLs onto
the anti-ADDL antibody functionalized nanoparticle surface has
been analyzed. To develop a more quantitative understanding
of the concentration-dependent response, only the shift from
the ADDLs and not the response from the second antibody is
included. The LSPR λmaxshift, ∆λmax) ∆R, versus [ADDL]
response curve was measured over the concentration range 1
fM < [ADDL] < 10 µM (Figure 4A). While the thermodynamic
affinity constant between antibodies and their specific antigens
is by no means large enough to be considered irreversible,
sample incubation and rinsing conditions have been optimized
to minimize nonspecific interactions while retaining the largest
possible LSPR shift response.
Figure 4A shows the experimental data for the ADDL shifts
plotted as the LSPR λmax shift, ∆R versus [ADDL]. The
experimental response curve for ADDLs has been quantitatively
interpreted in terms of a model that makes the following
assumptions: (1) 1:1 binding of a solution-phase multivalent
analyte (ADDL) with different sites but invariant affinities to
the surface-bound capture ligand (anti-ADDL antibody); (2) the
only operative nanoparticle sensing mechanism is the change
in the local refractive index caused by the adsorbed analyte
(ADDL); and (3) the measured LSPR λmaxshift response, ∆R,
is proportional to the thickness (and mass) of the adsorbed
analyte layer and its refractive index. The response curve can
be described by the following equation:6
∆R ) ∆Rmax[
where ∆R ) ∆λmax, LSPR λmaxshift for a given concentration,
∆Rmaxis the maximum LSPR response at high concentrations,
Figure 4. Quantitative response curves. (A) LSPR shift, ∆R or ∆λmax, versus [ADDL] response curve for the binding of ADDLs to an anti-ADDL antibody
functionalized Ag nanobiosensor. All measurements were collected in a N2environment. The solid line is the calculated value of ∆R using eq 1. The shaded
box indicates the limit of detection as determined from the spectrometer noise (3 times the smallest reliable wavelength shift measured). Error bars represent
the spread in the data. (B) Enhanced LSPR shift, ∆R or ∆λmax, versus [anti-ADDL antibody] response curve for the binding of second anti-ADDL antibody
to the ADDL functionalized Ag nanobiosensor. All measurements were collected in a N2environment. The solid line is the calculated value of ∆R using eq
2. The shaded box indicates the limit of detection as determined from the spectrometer (3 times the smallest reliable wavelength shift measured). Error bars
represent the spread in the data. The data are divided into three detection regions: (region 1) response for concentrations less than 10 pM, (region 2)
response for concentrations between 10 pM through 10 nM, and (region 3) response for concentrations greater than 10 nM.
1 + Ka,surf[ADDL]]
Detection of a Biomarker Using an Optical Biosensor
A R T I C L E S
J. AM. CHEM. SOC. E
Ka,surfis the surface-confined thermodynamic affinity constant,
and [ADDL] is the concentration of the ADDL solution.
Comparing the experimentally measured ∆R versus [ADDL]
response (points) to eq 1 (solid line, Figure 4A) yields
approximate values for ∆Rmax) 11.7 nm and a surface-confined
thermodynamic affinity constant Ka,surf) 4.5 × 1010M-1. In
addition, the data in Figure 4A allows one to estimate the limit
of detection (LOD) for the LSPR nanobiosensor for ADDL
detection separate from the enhanced shift from the second anti-
ADDL antibody. The peak-to-peak wavelength shift noise of
the baseline in repetitive experiments is ∼0.3 nm. Taking the
limit of detection as 3 times this value, one conservatively
estimates a LOD of ∼10 pM.
Figure 4B displays the experimental data for the second anti-
ADDL antibody shift response plotted as the LSPR λmax, ∆R
versus [ADDL]. As a reminder, a fixed (100 nM) second anti-
ADDL antibody concentration was exposed to the LSPR
nanosensor substrates that had been exposed to varying ADDL
concentrations. These data are not accurately described by the
previous binding model. Instead, the experimental response
curve has been quantitatively predicted in terms of a model that
makes the following assumptions: (1) multiple (two) receptors
of solution phase multivalent analytes (anti-ADDL antibody)
with different sites but invariant affinities to the surface-bound
capture ligands (ADDL); (2) the only operative nanoparticle
sensing mechanism is the change in the local refractive index
caused by the adsorbed analyte (anti-ADDL antibody); and (3)
the measured LSPR λmaxshift response, ∆R, is determined only
by the thickness of the adsorbed analyte layer and its refractive
index. The response curve can be described by the following
∆R ) ∆Rmax,1[
where ∆R ) ∆λmax, LSPR λmax shift induced by the second
anti-ADDL antibody for a given ADDL concentration, ∆Rmax,1
is the maximum LSPR response at concentrations in region 1,
∆Rmax,2 is the added LSPR shift response above ∆Rmax,1 at
concentrations in region 2, Ka,surf,1 is the surface-confined
thermodynamic affinity constant in region 1, Ka,surf,2 is the
surface-confined thermodynamic affinity constant in region 2,
and [ADDL] is the concentration of ADDLs in solution.
This model accurately predicts the second anti-ADDL
antibody response for ADDL concentrations ranging from 1 fM
to 100 nM. As displayed in Figure 4B, three binding regimes
are revealed. Region 1, which covers ADDL concentrations
below 10 pM, reveals an extremely strong antigen/antibody
interaction with a binding constant (Ka,surf,1) of 7.3 × 1012M-1
and a limit of detection (LOD) <100 fM. Region 2, which
covers ADDL concentrations between 10 pM to 100 nM,
describes an average antigen/antibody interaction with a binding
constant (Ka,surf,2) of 9.5 × 108M-1. At concentrations greater
than 100 nM (region 3), the second anti-ADDL antibody
response begins to deviate from the binding model. While the
quantitation of ADDLs at these concentrations is biologically
irrelevant, we have included these data for completeness. The
origin of the observed decrease in response at high ADDL
concentrations is under investigation.
Alzheimer’s Disease Assay Using Human Samples. To
date, the only systems probed with the LSPR nanosensor have
been model systems such as biotin/streptavidin2and biotin/
antibiotin.6Because of our encouraging nonspecific binding
results and the quantitative models revealed in the standard
sandwich assay described above, it was decided that two types
of human samples would be attempted: human brain extract
and cerebrospinal fluid. In both cases, duplicate samples from
a control and an AD patient were analyzed. These initial data
suggest promising results for clinical sample analysis using
LSPR nanosensor technology.
In the first comparison, human brain extract was prepared
and exposed to an anti-ADDL antibody functionalized LSPR
nanosensor chip for 30 min. After thorough rinsing, the sample’s
response was enhanced via exposure to a second anti-ADDL
antibody (Figure 5). The assays were performed in duplicate.
Figure 5B reveals a representative assay from an age-matched
brain extract sample. During the brain extract incubation, the
extinction maximum of the sample shifted from 782.3 nm
(Figure 5B-1) to 782.4 nm (Figure 5B-2), a 0.1-nm shift. Next,
this shift was amplified by exposing the sample to an additional
Figure 5. Analysis of human brain extract samples using a sandwich assay
and the LSPR nanosensor. (A) Surface chemistry for the possible ADDL
detection in human brain extract samples using the antibody sandwich assay.
(B) Control patient: LSPR spectra for each step of the assay. Ag
nanoparticles after functionalization with (B-1) 100 nM anti-ADDL (100
mM EDC) (λmax) 782.3 nm), (B-2) control brain extract (λmax) 782.4
nm), and (B-3) 100 nM anti-ADDL (λmax) 782.9 nm). (C) AD patient:
LSPR spectra for each step of the assay. Ag nanoparticles after function-
alization with (C-1) 100 nM anti-ADDL (100 mM EDC) (λmax ) 751.6
nm), (C-2) diseased brain extract (λmax)758.3 nm), and (C-3) 100 nM anti-
ADDL (λmax) 762.3 nm). All measurements were collected in N2.
1 + Ka,surf,1[ADDL]]+
1 + Ka,surf,2[ADDL]](2)
A R T I C L E S
Haes et al.
F J. AM. CHEM. SOC.
anti-ADDL antibody for 30 min resulting in an additional 0.5-
nm wavelength shift or a total LSPR shift of 0.6 nm. This
response is below the reliable detection capabilities of the LSPR
nanosensor sandwich assay as predicted from the quantitative
analysis in Figure 4A and 4B. Additionally, this small response
indicates that the LSPR nanosensor can be used for human
sample analysis without undue concern for nonspecific binding.
Analysis of the AD sample reveals completely different
behavior. A representative assay is shown in Figure 5C. Upon
incubation of the antibody functionalized sensor substrate in
the diseased brain extract, the LSPR extinction maximum shifts
from 751.6 nm (Figure 5C-1) to 758.3 nm (Figure 5C-2) giving
an LSPR shift of 6.7 nm. This response is amplified by an
additional 4.0 nm by the second anti-ADDL antibody revealing
an easily detected total shift of 10.7 nm.
While detection of ADDL in brain extract is encouraging,
this assay can only be completed postmortem. Ideally, one would
perform this assay in a body fluid that can be extracted while
a person is still alive. For that reason, the LSPR sandwich assay
was used to analyze cerebrospinal fluid (CSF) from a control
and AD patient. Preliminary results indicate a large difference
between aging and AD patient samples. As seen in Figure 6A,
a small 2.9-nm LSPR shift was observed after exposing the CSF
sample from an aging patient onto the antibody functionalized
substrate. This shift was amplified by 4.3 nm with the second
anti-ADDL antibody. The results in Figure 6B are dramatically
different. When this antibody functionalized sample is exposed
to a CSF sample from a patient with AD, an 18.5-nm shift is
observed. This shift is magnified by an additional 15.4 nm after
exposure to the second antibody resulting in a total shift of 33.9
To date, much of nanoscience research has been focused on
understanding the fundamental science involved rather than on
the implementation of successful applications for biological or
chemical analysis. As demonstrated in this work, the potential
of nanoscale optical biosensors when combined with specific
and novel biological materials are far reaching and will aid in
the development of other successful nanotechnology-based
devices. The development of nanodevices, including nanosensors
that are highly sensitive and selective (give low false positives,
low false negatives), will provide a major improvement over
current technologies for disease understanding, treatment, and
monitoring. Instruments that provide high throughput screening
for drug discovery and disease diagnosis will uncover informa-
tion vital to the understanding and monitoring of disease that
may lead to the design of better drug candidates for its treatment
Specifically, this work demonstrates that nanoscale devices
do, in fact, provide information that is not attainable with
traditional macroscale devices. In terms of the LSPR nanosensor
technology, several major accomplishments have been shown
here. First, a successful sandwich assay for an antigen between
two antibodies has been successfully demonstrated for the first
time. This was accomplished, in part, because of our funda-
mental understanding and optimization of the spatial distribution
of the electromagnetic fields surrounding the nanoparticles by
controlling nanoparticle size and shape.3,4Next, without sacrific-
ing the strength of nanoparticle adhesion to the substrate, the
sensor’s response exhibited no nonspecific binding in a pair of
important assays. Quantitative binding models have been
presented for both the detection of the target antigen and the
second antibody in the antibody sandwich.
These models reveal important information regarding the
nature of the ADDL targets as well as revealing two classes of
binding affinity constants between the target antigens and
antibodies. In contrast to our previous studies,2,6this binding
model for direct ADDL detection does not perfectly fit the data
(Figure 4A). At high concentrations, the response model
accurately predicts the LSPR shift response; however, at low
concentrations (that is, <10 pM ADDL), the response model
underestimates the observed response. Because the LSPR shift
response eventually goes to zero, the underestimation of the
response cannot arise for nonspecific binding interactions.
Instead, we hypothesize that this behavior reveals information
regarding the nature of the endogenous ADDL molecules.
Surprisingly, this hypothesis is fully supported upon the
analysis of the second anti-ADDL antibody shift response. This
response can be divided into three ADDL concentration regions.
In region 1 (1 fM to 10 pM), the ADDL molecules have an
extremely high affinity (Ka,surf,1 ) 7.3 × 1012M-1) for the
Figure 6. Analysis of human CSF samples using a sandwich assay and
the LSPR nanosensor. (A) Surface chemistry for the possible ADDL
detection in human CSF samples using the antibody “sandwich” assay. (B)
Aging patient: LSPR spectra for each step of the assay. Ag nanoparticles
after functionalization with (B-1) 100 nM anti-ADDL (100 mM EDC) (λmax
) 759.7 nm), (B-2) CSF (λmax) 762.6 nm), and (B-3) 100 nM anti-ADDL
(λmax ) 766.9 nm). (C) AD patient: LSPR spectra for each step of the
assay. Ag nanoparticles after functionalization with (C-1) 100 nM anti-
ADDL (100 mM EDC) (λmax) 780.6 nm), (C-2) CSF (λmax) 809.1 nm),
and (C-3) 100 nM anti-ADDL (λmax) 824.5 nm). All measurements were
collected in a N2environment.
Detection of a Biomarker Using an Optical Biosensor
A R T I C L E S
J. AM. CHEM. SOC. G
antibody. From the quantitative model, which underestimates
the LSPR response for ADDL detection in the sub-10 pM
region, the actual response observed here suggests that the
molecules have a higher molecular weight than the second class
of ADDLs detected. This is apparent from the larger than
predicted LSPR shifts observed in the direct ADDL response
analysis. The results in region 2 (10 pM to 10 nM) verify this
claim. The second class of molecules was revealed to have a
binding constant, Ka,surf,2, about 9.5 × 108M-1. A comparison
study using size exclusion chromatography (SEC) and ELISA
revealed that ADDLs are comprised of two major types of
molecules with different molecular weights (unpublished data).
This is consistent with studies by Bitan and Teplow who
detected different subpopulations in experiments with cross-
linking reagents.36The data in region 3 (10 nM to 10 µM)
reveals unusual behavior. The drop-off in the response was
initially thought to arise from the so-called “Hook effect”.37,38
However, this type of behavior should be eliminated when a
monoclonal antibody is used in place of the second polyclonal
antibody. When this assay is repeated with a monoclonal second
antibody, the hook effect is still observed (data not shown).
Consequently, this response cannot be attributed to the Hook
effect. Further work is in progress that seeks to understand this
phenomenon. Our working hypothesis regarding this deviation
is that at extremely high ADDL concentrations, small surface
ADDLs (with a weak affinity to the first antibody) begin to
interact and begin to behave as large ADDLs with higher
binding affinities to the second antibody thereby being pulled
off of the nanoparticles and exhibiting a reduction in second
Finally, this work presents the first analysis of endogenous
biological samples using LSPR detection. These extremely
promising results indicate that the surface chemistry of the LSPR
nanosensor has been designed for optimal analysis of complex
biological species. The first set of samples discussed here
confirms the previous observation that ADDLs are present in
elevated concentrations in AD brain in comparison to a control.34
Using the quantitative models, the magnitude of the shifts
indicate that the ADDL concentration is ∼1 pM in the diseased
brain while the signal from the control sample is undetectable
with the noise level (i.e., ∼0). Again, the magnitude of the CSF
shift indicates that a higher concentration of ADDLs or ADDL-
related molecules is present in the CSF AD sample in
comparison to the control. While the magnitude of the shift from
the CSF sample is much larger than predicted by the ADDL
response binding curve (Figure 4A), the second anti-ADDL
antibody response indicates that the CSF sample contains
ADDLs (possibly complexed to other molecules). This response
is much larger than that observed in the analysis of the AD
brain extract. At this time, we hypothesize that the oligomer
size in CSF versus brain extract is quite different, perhaps
influenced by the differing molecular milieus of CSF and brain
extracts. Thus, it is most relevant to assess the two types of
samples separately. Even though these studies must be regarded
as preliminary given the small number of samples analyzed,
they are, nonetheless, very exciting. They indicate that the LSPR
nanobiosensor can be used to study human samples and may
aid in the understanding of the mechanism and diagnosis of
The success of this LSPR nanosensor was directly related to
the synthesis and isolation of target biomolecules. By integrating
this technology with the newly modified amyloid hypothesis,
previously unavailable information has been revealed regarding
the nature of ADDLs. All previous biophysical characterization
of the oligomerization and fibrillogensis of these molecules
typically required concentrations so high as to be irrelevant to
in vivo conditions. The LSPR nanosensor has been demonstrated
to be a powerful tool for studying the oligomerization of low
concentrations of amyloid precursors. Given this success, the
use of LSPR technology also holds promise as one of the best
detection techniques for the screening of oligomerization-
blocking drugs. Because the molecular causes and mechanisms
of AD are not fully understood, devices that provide insight
into the aggregated states of biological species and their
interactions at native concentrations will help in screening
patients for disease and possibly for studying drug interactions
with target species. This work represents the first steps toward
making this possible.
Acknowledgment. The authors gratefully acknowledge fi-
nancial support from the Nanoscale Science and Engineering
Initiative of the National Science Foundation under NSF Award
EEC-0118025. Any opinions, findings, and conclusions or
recommendations expressed in this material are those of the
authors and do not necessarily reflect those of the National
Science Foundation. W.L.K. gratefully acknowledges support
from NIH. W.L.K. is co-founder of Acumen Pharmaceuticals,
which has the sole license to patent rights owned by North-
western University and the University of Southern California
for use of ADDLs in the development of Alzheimer’s-related
therapeutics and diagnostics.
(36) Bitan, G.; Kirkitadze, M. D.; Lomakin, A.; Vollers, S. S.; Benedek, G. B.;
Teplow, D. B. Proc. Natl. Acad. Sci. U.S.A. 2003, 100, 330.
(37) Fernando, S. A.; Wilson, G. S. J. Immunol. Methods 1992, 151, 67.
(38) Fernando, S. A.; Wilson, G. S. J. Immunol. Methods 1992, 151, 47.
A R T I C L E S
Haes et al.
H J. AM. CHEM. SOC.PAGE EST: 7.8