A Colorimetric Gold Nanoparticle Sensor To Interrogate Biomolecular

Interrogate Biomolecular Interactions in Real Time on a Surface. Nidhi Nath and Ashutosh Chilkoti*. Department of Biomedical Engineering, Duke Univers...
7 downloads 0 Views 224KB Size
Download PDF
Anal. Chem. 2002, 74, 504-509

A Colorimetric Gold Nanoparticle Sensor To Interrogate Biomolecular Interactions in Real Time on a Surface Nidhi Nath and Ashutosh Chilkoti*

Department of Biomedical Engineering, Duke University, Box 90281, Durham, North Carolina 27708-0281

This paper presents a new label-free optical method to study biomolecular interactions in real time at the surface of an optically transparent substrate. The method relies on the change in the absorbance spectrum of a selfassembled monolayer of colloidal gold on glass, as a function of biomolecular binding to the surface of the immobilized colloids. Using this approach, we demonstrate proof of principle of a label-free optical biosensor to quantify biomolecular interactions in real time on a surface in a commercially available UV-visible spectrophotometer and of a colorimetric end-point assay using an optical scanner. The spectrophotometric sensor shows concentration-dependent binding and a detection limit of 16 nM for streptavidin. The sensor is easy to fabricate, is reproducible in its performance, has minimal technological requirements, namely, the availability of an UVvisible spectrophotometer or an optical scanner, and will enable high-throughput screening of biomolecular interactions in real time in an array-based format. In the past decade, surface plasmon resonance (SPR)1 methods have made an important contribution to the quantification of biomolecular interactions. Unfortunately, conventional SPR reflectometry (e.g., the Biacore sensor) is difficult to realize in a large-scale array format, because of the optics associated with the detection system. This latter limitation is significant, because highthroughput biochemical assays based on protein arrays are urgently needed to measure the protein-protein and the proteinligand interactions for the many thousands of proteins identified by the human genome project.2-4 This paper presents a new label-free optical sensor, which retains many of the desirable features of conventional SPR reflectometry, namely, the ability to monitor the kinetics of biomolecular interactions in real time without a label, but which has several important advantages: the sensor is easy to fabricate, and simple to implement, requiring only an UV-visible spectrophotometer or flatbed scanner. Importantly, the sensor can be easily multiplexed to enable high-throughput screening of biomolecular interactions in an array-based format. * To whom correspondence should be addressed: (e-mail) [email protected]. (1) Homola, J.; Yee, S. S.; Gauglitz, G. Sens. Actuators, B 1999, 54, 3-15. (2) Emili, A. Q.; Cagney, G. Nat. Biotechnol. 2000, 18, 393-397. (3) Pandey, A.; Mann, M. Nature 2000, 405, 837-846. (4) Fields, S. Science 2000, 291, 1221-+ 2001.

504 Analytical Chemistry, Vol. 74, No. 3, February 1, 2002

Colloidal SPR is responsible for the intense colors exhibited by colloidal solutions of noble metals and is attributed to the collective oscillations of surface electrons induced by visible light.5 Colloidal SPR is an interfacial phenomenon and can be used in two complementary modes to transduce binding events at the colloid surface. In the first mode, changes in the proximity of colloids due to their aggregation in suspension cause a large change in the absorbance spectrum of the colloidal suspension due to long-range coupling of surface plasmons. The interparticle distance-dependent color change of colloidal gold due to aggregation of gold colloids has been used in solution-based immunoassays6 and has recently been used by Mirkin and colleagues to design a sensor capable of determining single-base mismatches in DNA hybridization,7-8 In the second mode, the optical signal arises from the dependence of the peak intensity and position of the surface plasmon absorbance of gold nanoparticles upon the local refractive index of the surrounding medium, which is altered due to binding at the colloid-solution interface.9 This mode, which is analogous to conventional SPR, has been previously utilized to determine biomolecular binding on the surface of a colloid in suspension10-11 We sought to develop a SPR biosensor in a planar, chip-based format using immobilized gold colloids on an optically transparent substrate for the following reasons: first, because gold colloids permit the transmission of light in the visible region of the electromagnetic spectrum, in principle, a chip-based colloidal SPR sensor should enable SPR to be performed in transmission mode in a UV-visible spectrophotometer. In essence, we sought to make SPR sensing accessible to any researcher with an UVvisible spectrophotometer. Second, our choice of a surface, rather than a solution-based sensor was based on the obvious realization that chip-based sensors can be designed in an array format for rapid, high-throughput screening of biomolecular interactions. Compared to other direct optical sensors utilizing SPR of gold or (5) Link, S.; El-Sayed, M. A. Int. Rev. Phys. Chem. 2000, 19, 409-453. (6) Hayat, M. A. Colloidal Gold: Principles, Methods, and Applications; Academic Press: San Diego, CA, 1989. (7) Elghanian, R.; Storhoff, J. J.; Mucic, R. C.; Letsinger, R. L.; Mirkin, C. A. Science 1997, 277, 1078-1081. (8) Storhoff, J. J.; Elghanian, R.; Mucic, R. C.; Mirkin, C. A.; Letsinger, R. L. J. Am. Chem. Soc. 1998, 120, 1959-1964. (9) Kreibig, U.; Vollmer, M. Optical Properties of Metal Clusters; SpringerVerlag: Heidelberg, Germany, 1995. (10) Englebienne, P. Analyst 1998, 123, 1599-1603. (11) Eck, D.; Helm, C. A.; Wagner, N. J.; Vaynberg, K. A. Langmuir 2001, 17, 957-960. 10.1021/ac015657x CCC: $22.00

© 2002 American Chemical Society Published on Web 12/29/2001

silver nanoparticles12-14 that have recently been suggested, the method reported here uses solution self-assembly to fabricate the sensor, which is simple enough to be implemented in most biochemical laboratories, and yields a sensor with good sensitivity and reproducibility. Furthermore, as we demonstrate, this methodology can easily be extended to the fabrication of an array-based sensor. Previous studies by Natan and colleagues15-17 have shown that gold colloids can be self-assembled from solution onto a functionalized glass surface to yield an optically transparent monolayer, where the assembly is stabilized by strong attractive colloidsurface interactions and laterally by repulsive colloid-colloid electrostatic interactions. More recently, Okamoto et al.18 showed that the absorbance of an immobilized monolayer of gold colloids is sensitive to the refractive index of the surrounding solvent, but they made no attempt to detect biomolecular interactions with their sensor. Using this approach, we demonstrate proof of principle of a simple, label-free optical biosensor; we show that a monolayer of immobilized gold nanoparticles can be prepared on glass by self-assembly from solution, that the immobilized gold nanoparticles can be subsequently functionalized with a biological ligand, and that the biofunctionalized surface exhibits colloidal SPR with a sensitivity and dynamic range that is suitable to quantify biomolecular interactions in real time on a surface in a commercially available UV-visible spectrophotometer. We also report preliminary results of a colorimetric end-point assay in an array format using a low-cost, flatbed optical scanner to image the absorbance change associated with biomolecular binding at the surface of the immobilized colloids. EXPERIMENTAL SECTION Synthesis and Characterization of Gold Colloids. All glassware used for preparation of colloids were thoroughly washed with aqua regia (3:1 HNO3-HCl), rinsed extensively with distilled water, and then dried in an oven at 100 °C for 2 h. Gold colloids were prepared by sodium citrate reduction of HAuCl4‚3H2O as reported earlier.15 A 250-mL sample of 1 mM HAuCl4 (Aldrich) was brought to a vigorous boil with stirring in a round-bottom flask fitted with a reflux condenser, and 25 mL of 38.8 mM sodium citrate was rapidly added to the solution. The solution was boiled for another 15 min, during which time the solution changed color from pale yellow to deep red. The solution was allowed to cool to room temperature with continued stirring. The suspension was filtered using a 0.22-µm filter (Corning, NY) and stored at 4 °C until further use. The diameter of the colloids was determined by transmission electron microscopy (TEM). A small drop of the colloidal gold suspension was placed on a lysine-coated Formvar grid, and (12) Kalyuzhny, G.; Schneeweiss, M. A.; Shanzer, A.; Vaskevich, A.; Rubinstein, I. J. Am. Chem. Soc. 2001, 123, 3177-3178. (13) Malinsky, M. D.; Kelly, K. L.; Schatz, G. C.; Van Duyne, R. P. J. Am. Chem. Soc. 2001, 123, 1471-1482. (14) Himmelhaus, M.; Takei, H. Sens, Actuators, B 2000, 63, 24-30. (15) Grabar, K. C.; Freeman, R. G.; Hommer, M. B.; Natan, M. J. Anal. Chem. 1995, 67, 735-743. (16) Freeman, R. G.; Grabar, K. C.; Allison, K. J.; Bright, R. M.; Davis, J. A.; Guthrie, A. P.; Hommer, M. B.; Jackson, M. A.; Smith, P. C.; Walter, D. G.; Natan, M. J. Science 1995, 267, 1629-1632. (17) Grabar, K. C.; Brown, K. R.; Keating, C. D.; Stranick, S. J.; Tang, S. L.; Natan, M. J. Anal. Chem. 1997, 69, 471-477. (18) Okamoto, T.; Yamaguchi, I.; Kobayashi, T. Opt. Lett. 2000, 25, 372-374.

excess solution was wicked away by a filter paper. The grid was subsequently dried in air and imaged on a Philips 400S transmission electron microscope. The accelerating voltage was 80 kV. The size of the gold colloids was determined by image analysis of TEM images of gold colloids (n ) 150). Fabrication and Characterization of Colloidal Gold Monolayer on Glass (AuCM). Glass coverslips (VWR Scientific) cut into 10 × 50 mm pieces, were used as the substrate for assembly of the colloidal gold monolayers. The glass substrates were cleaned by sonication for 5 min in hot RBS 35 detergent (Pierce) and washed extensively with distilled water. The substrate were further cleaned in a 1:1 solution of methanol and HCl for 30 min, washed extensively with distilled water, and dried overnight at 60 °C. The cleaned glass substrates were immersed in a 10% (v/v) solution of γ-(aminopropyl)triethoxysilane (APTES, Sigma) in anhydrous ethanol for 15 min, rinsed five times in ethanol with sonication, and dried at 120 °C for 3 h. The silanized glass coverslips were subsequently immersed overnight in a colloidal gold solution (11.6 nM) to form a self-assembled monolayer of the gold colloids on both sides of the glass coverslip (AuCM). The immobilized colloids were imaged by atomic force microscopy (AFM) in tapping mode in air using standard Si3N4 cantilevers on a Multimode NanoscopeIIIa (Digital Instruments Inc.). Functionalization of AuCM. AuCM was modified by the formation of a self-assembled monolayer (SAM) of mercaptopropionic acid (MPA) by incubation of the colloid monolayer on glass in a 1 mM solution of MPA in absolute ethanol for 10 min at room temperature (termed AuCM-MPA). These samples were used for fibrinogen adsorption studies. AuCM-MPA samples were functionalized with biotin as follows: AuCM-MPA samples were immersed in an ethanol solution of 0.1 M 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide (EDAC, Sigma) and 0.2 M pentafluorophenol (PFP, Sigma) for 20 min at room temperature, rinsed three times with ethanol, immersed in a 100 µg/mL solution of (+)-biotinyl-3,6,9-trioxaundecanediamine (trade name, EZ-Linkbiotin-PEO-LC-amine; Pierce) in ethanol for 2 h, washed with ethanol, and stored in PBS at 4 °C until further use. Absorbance Measurements of Immobilized Gold Colloids on Glass. A temperature-controlled spectrophotometer (Cary 300Bio, Varian Instruments) was used to measure the absorbance of the immobilized gold colloids on glass coverslips. A rectangular glass flow cell of 4 mm width was designed in-house to hold the samples. Samples were positioned in the center of the glass cell using two Teflon guides on top of the cell. Spectra were collected in transmission mode over a range of 350-850 nm. The glass coverslips were scanned using a commercial UMAX Super Vista S-12 flatbed scanner (UMAX Technologies, Inc.). The color image was converted to gray scale in NIH Image and analyzed for mean and standard deviation. The data were analyzed by one-way ANOVA and Bonferroni multiple comparison test using InStat for MacIntosh version 2.0 (GraphPad Software, Inc., San Diego, CA). RESULTS AND DISCUSSION The physical principle underlying the sensor and its fabrication is illustrated in Figure 1. Exploiting the high-affinity of gold colloids for thiol and amine functional groups, a glass surface was transformed into a sensor chip by self-assembly of gold nanoparticles to form a reactive monolayer on an amine-terminated glass substrate16 (Figure 1A). Molecular binding on the sensor surface Analytical Chemistry, Vol. 74, No. 3, February 1, 2002

505

Figure 1. (A) Schematic of the steps involved in the fabrication of the immobilized colloidal gold sensor chip on glass. Glass substrate was functionalized with APTES (1) to provide an amine-terminated surface for formation of a monolayer of gold nanoparticles (AuCM). SAM of MPA (2) on gold nanoparticles provides a reactive carboxyl group that can be further modified by biotin (3) to study specific binding of streptavidin. (B) Biomolecular binding at the surface of the functionalized gold monolayer results in a shift in peak wavelength as well as an increase in intensity.

is transduced to a colorimetric signal due to the changes in surface plasmon absorbance of the immobilized gold nanoparticles (Figure 1B). Our strategy for the fabrication of the sensor chip involves solution self-assembly at each critical step, as follows. First, a glass substrate was functionalized by the formation of a SAM of APTES to present an amine-terminated SAM on the glass surface. Next, the surface was immersed in a solution of colloidal gold, which resulted in the spontaneous self-assembly of a monolayer (CM) of gold colloids on glass (AuCM). The attachment of gold nanoparticles to an amine-terminated surface is strong enough to withstand subsequent chemical modification of the gold nanoparticles without causing their detachment from the surface. The gold colloids were synthesized by trisodium citrate reduction of gold tetrachloroaurate.15 This is a simple, one-step reduction, and the reaction conditions can be controlled to yield monodisperse gold nanoparticles of any desired size in the 5-100nm range. Image analysis of the TEM micrographs of colloidal gold showed that the diameter of gold nanoparticles was 13.4 ( 0.9 nm (n ) 150). The self-assembled gold nanoparticles on glass were characterized by UV-visible spectrophotometry as well as by AFM to check the quality of the monolayer. The aggregation of colloidal gold on glass or the formation of multilayers would result in the coupling of plasmons of individual particles and would be reflected in the UV-visible spectrum as an increased absorbance at wavelengths greater than 600 nm.15 The absence of such a feature in the UV-visible spectrum of AuCM (Figure 2A) when compared to the spectrum of an aqueous solution of colloidal gold indicates 506 Analytical Chemistry, Vol. 74, No. 3, February 1, 2002

that the gold nanoparticles are immobilized in a monolayer on the glass substrate and are isolated from each other. The observed diameter of the gold nanoparticles determined by AFM was ∼20 nm (Figure 2B), which is significantly greater than that determined by TEM. This discrepancy is caused by the finite (∼20nm) radius of the AFM tip, which results in an overestimation of the diameter of the colloids due to convolution of the tip geometry into the AFM image.17 A surface density of 1.46 × 1011 gold particles/cm2 was determined from AFM. Using the colloid diameter of 13.4 nm determined by TEM, and the surface density determined by AFM, an average center-to-center interparticle spacing of 28.1 nm was calculated by assuming a hexagonal arrangement of the colloids. These results suggest that on average the immobilized gold colloids on glass are separated by ∼15 nm. We first examined the ability of the immobilized monolayer of gold nanoparticles to transduce changes in the surrounding refractive index into the absorbance spectrum. We found that the absorbance spectrum of AuCM exhibited a red shift in the peak wavelength along with an increase in the absorbance at the peak wavelength as a function of the refractive index of the solvent in the range of 1.33-1.49 (Figure 3). Mie theory predicts a similar red shift in the position of the absorbance peak (λmax) and an increase in the absorbance maximum as shown by Mulvaney et al.,19 consistent with these experimental observations. The inset in Figure 3 shows that a linear fit to the plot of the λmax as a function of refractive index gives a sensitivity of 76.4 nm/refractive index unit (RIU). A nonlinear response is expected for the (19) Templeton, A. C.; Pietron, J. J.; Murray, R. W.; Mulvaney, P. J. Phys. Chem. B 2000, 104, 564-570.

Figure 4. (A) Absorbance spectrum of an MPA-functionalized colloidal gold monolayer on glass AuCM-MPA (a). Change in spectrum after 2-h incubation of AuCM-MPA in (b) 10 and (c) 100 µg/mL solution of fibrinogen. (B) Adsorption of fibrinogen onto AuCMMPA as a function of time, monitored at 550 nm, for three different solution concentrations of the protein.

Figure 2. (A) Absorbance spectrum of colloidal gold in aqueous medium (s) and that of a monolayer of gold colloids on APTESfunctionalized glass (- - -). Spectra were normalized to their absorption maximums. (B) Tapping mode AFM images in air of a SAM of gold colloids on glass. Scan area, 0.5 µm × 0.5 µm.

Figure 3. Absorbance spectra of a monolayer of immobilized gold colloids on glass in the following: (a) water (n ) 1.33); (b) ethanol (n ) 1.36); (c) 3:1 (v/v) ethanol-toluene (n ) 1.39); (d) 1:1 (v/v) ethanol-toluene (n ) 1.429); (e) 1:3 (v/v) ethanol-toluene (n ) 1.462); (f) toluene (n ) 1.495). The inset shows the dependence of the absorbance maximum as well as the absorbance at 500 nm on the refractive index of the surrounding medium.

absorbance change at a fixed wavelength (e.g., 550 nm) as a function of refractive index due to change in the peak wavelength as well as the absorbance. Over the range of refractive index studied here, however, the response was linear (R ) 0.992 ( 0.004) with a slope of 0.46 absorbance unit (AU)/RIU. We suggest that, within this limited refractive index range, it is reasonable to assume a linear response of the sensor, which results in a sensitivity of 0.46 AU/RIU. However, for a wider range of refractive index, a linear fit may not apply.

The inset shows the dependence of the absorbance maximum as well as the absorbance at 550 nm on the refractive index of the surrounding medium. Both the wavelength shift and the increase in absorbance can be used as an optical signature for the change in refractive index. We chose to monitor the absorbance shift at 550 nm for subsequent measurements because of the somewhat higher sensitivity that can be achieved as well as the ease of continuous monitoring at a fixed wavelength. These results clearly show that the optical absorbance of an immobilized monolayer of gold colloids is sensitive to the refractive index of the surrounding bulk medium. Encouraged by these results, we investigated whether the refractive index change at the surface of individual gold nanoparticles due to biomolecular binding events could also be transduced into an experimentally detectable change in the absorbance spectrum. A SAM of MPA was formed on AuCM to present terminal COOH groups (AuCM-MPA). We chose this surface for the adsorption and binding experiments because it provides a welldefined surface as compared to bare gold (which is prone to adventitious contamination) and because it enables ligands to be covalently tethered at the colloid-solution interface. We investigated two different classes of binding interactions at this interface in order to determine whether the change in local refractive index due to binding of biomolecules could be transduced to an optical signal. These are as follows: (1) the adsorption of fibrinogen onto AuCM-MPA and (2) molecular recognition of an anti-biotin monoclonal antibody (mAb) and streptavidin by AuCM-MPA that had been covalently functionalized with a biotin derivative. Figure 4A is the spectrum of AuCM-MPA before and after incubation with fibrinogen at two different solution concentrations for 2 h at room temperature. A significant increase in the absorbance at 550 nm was observed, due to the increase in the local refractive index at the colloid-solution interface caused by the adsorption of fibrinogen. We further investigated the kinetics of fibrinogen adsorption by monitoring the absorbance change at 550 nm in real time using a home-built, rectangular glass flow chamber. Both the concentration and time-dependent adsorption of fibrinogen adsorption can be successfully transduced using this Analytical Chemistry, Vol. 74, No. 3, February 1, 2002

507

Figure 5. (A) Scanned images of AuCM on glass functionalized with APTES (1), after formation of MPA SAM (2), and after 30-min incubation with 10 (3) and 1000 µg/mL (4) fibrinogen. (B) Histogram of the average intensity of each chip. Error bar represents standard deviation of the intensity distribution of each chip.

sensor, as shown in Figure 4B. These results show that the kinetics of adsorption are directly related to the solution concentration of the protein and that the maximum amount of fibrinogen bound to the surface at steady state is directly related to the solution concentration. These results are consistent with previous observations and can be attributed to different packing configurations of the protein on the surface.20 The absorbance changes observed in the spectrophotometric assay raised the intriguing possibility that optical scanning might directly capture these colorimetric changes, which would enable high-throughput assays to be developed in an array format using widely available optical scanners. To examine this possibility, we scanned four coverslips on a commercially available flatbed optical scanner (Figure 5A), which were functionalized as follows: AuCM (1), AuCM-MPA (2), and AuCM-MPA incubated with 10 (3) and 1000 µg/mL (4) solution of fibrinogen for 30 min. The color difference between the four sensor chips could be discerned visually and was quantitatively confirmed by conversion of the images to gray scale and calculation of the average intensity, as shown in Figure 5B. The difference between all samples were significant (P < 0.001) as determined by ANOVA and by Bonferroni’s multiple comparisons test. We suggest that the large increase in the absorbance observed for AuCM-MPA compared to AuCM is due to the contribution from the following two phenomena: first the change in refractive index of the local environment and second the chemisorption of thiols to the gold that results in damping and broadening of the plasmon band.21,22 In contrast, the smaller change in the gray scale intensity upon adsorption of fibrinogen at the surface of the immobilized colloids is largely due to the change in the local refractive index. These results suggest that, apart from its utility as a colorimetric SPR sensor, this sensor can be used in an array format for endpoint-assays. We note that this proof-of-principle experiment was performed on a low-cost optical scanner that was available in our laboratory with a small dynamic range. Use of state-of-the-art colorimetric scanners23 with high-resolution, cooled CCD cameras (20) Andrade, J. D.; Hlady, V. Adv. Polym. Sci. 1986, 79, 1-63. (21) Link, S.; El-Sayed, M. A. J. Phys. Chem B 1999, 103, 8410-8426. (22) Mulvaney, P. Langmuir 1996, 12, 788-800.

508 Analytical Chemistry, Vol. 74, No. 3, February 1, 2002

Figure 6. Binding of streptavidin and antibiotin mAb to a biotinfunctionalized gold colloid monolayer (AuCM-biotin) studied by surface plasmon absorbance of colloidal gold at 550 nm. (A) Baseline absorbance in PBS + 0.05% (v/v) Tween as a function of time. (B) Incubation of the biotin-functionalized surface with streptavidin (10 µg/mL) or antibiotin mAb (50 µg/mL) results in an increase in the absorbance due to protein-ligand binding. No increase in absorbance was observed on incubation of biotin-functionalized surface with BSA (10 µg/mL) (a), human IgG (50 µg/mL) (b), or streptavidin (30 µg/ mL) preincubated with 1.0 mM biotin (c). (C) Incubation of the proteinligand complex on the surface with 1 mM biotin in solution causes decrease in signal due to dissociation of biotin-mAb complex. No dissociation was observed for biotin-streptavidin complex due to its slow off-rate constant.

should yield a sensitivity that is comparable to that obtained with the UV-visible spectrophotometer used in the kinetic assays. We also tested the sensor for use in ligand-receptor binding studies using the model biotin-streptavidin and biotin-anti-biotin mAb receptor-ligand pairs. A monolayer of gold nanoparticles was covalently functionalized with biotin by activating the terminal COOH group of AuCM-MPA with 0.1 M EDAC and 0.2 M PFP in ethanol. The activated surface, which presented pentafluorophenyl esters, was reacted with an amine-terminated biotin derivative (EZLink-biotin-PEO-LC-amine) (Pierce) to covalently tether biotin to the immobilized colloid monolayer (AuCM-biotin). The binding experiments were performed in batch mode where biotin-functionalized surfaces were sequentially immersed in PBSTween 20 buffer (to characterize baseline stability), a solution of streptavidin or anti-biotin mAb (to initiate binding), followed by a 1 mM solution of biotin (to initiate dissociation of the ligandreceptor complex). A small negative baseline drift of 3 × 10-4 absorbance unit/min was observed for all AuCM-biotin sensor chips in buffer, but this drift was extremely reproducible, as seen by the overlaid plots for two different AuCM-biotin chips (Figure 6A). Incubation of the AuCM-biotin surface with a 10 µg/mL solution of streptavidin or 50 µg/mL anti-biotin mAb resulted in a dramatic, time-dependent increase in absorbance (Figure 6B). In control experiments, incubation of the biotin-functionalized surface with BSA or with anti-human IgG or with streptavidin whose biotin binding sites were blocked by preincubation with 1.0 mM biotin did not result in an absorbance change (Figure 6B), which confirmed that the increase in absorbance observed upon incubation with streptavidin and anti-biotin mAb is due to (23) Brush, M. D. Scientist 2000, 19, 26-29.

Figure 7. (A) Time-dependent change in surface plasmon absorbance at 550 nm as a result of specific binding of streptavidin to biotinylated gold nanoparticles surface: (a) 100, (b) 30, (c-e) 10, and (f) 5 µg/mL, (g) BSA, (h) biotin-saturated streptavidin, and (i) human IgG. (B) Absorbance change at 550 nm as a function of streptavidin concentration.

molecular recognition between the protein and immobilized ligand on the surface. Incubating the biotin-mAb complex in an aqueous solution of 1 mM biotin resulted in a decrease in absorbance as a function of time, due to dissociation of the mAb from the surface (Figure 6C). The shallow negative slope of the absorbance of the streptavidin-biotin complex as a function of time, post-biotin injection in Figure 6C is similar to the baseline drift (Figure 6A) and, hence, cannot be attributed to the dissociation of the streptavidin-biotin complex. The lack of dissociation of the streptavidin-biotin complex upon incubation in a biotin solution is consistent with the extremely slow off-rate constant of the complex (∼10-6 s-1).24 We further investigated the concentration-dependent absorbance change to determine the dynamic range and sensitivity that can be achieved for streptavidin-biotin binding using this sensor. An AuCM-biotin chip was incubated in streptavidin as a function of solution concentrations ranging from 0.03 to 100 µg/mL, and the absorbance change at 550 nm was measured as a function of time. Figure 7A shows representative plots of the response of the sensor as a function of time for different streptavidin concentrations. Both the kinetic and steady-state responses of the sensor are highly reproducible, as shown for three different replicates at the same solution concentration of streptavidin (10 µg/mL) in Figure 7A (plots c-e). A calibration plot of the absorbance change at 550 nm, after 30-min incubation, versus streptavidin concentration yielded a detection limit of ∼1 µg/mL streptavidin (16.6 nM streptavidin tetramer) (Figure 7B). The dynamic range of the sensor is ∼1.0-30 µg/mL. These results indicate that a selfassembled monolayer of gold nanoparticles on glass can be used to transduce ligand-receptor binding at a surface into an absorbance change with a sensitivity that is useful for biosensor applications. We believe that a much lower detection limit can be achieved by optimization of both the surface chemistry (e.g., size of colloid and surface density) and the detector, though we note that the current detection limit is low enough for many practical applications. (24) Jung, L. S.; Nelson, K. E.; Stayton, P. S.; Campbell, C. T. Langmuir 2000, 16, 9421-9432.

Although these proof-of-principle results reported here are encouraging, much remains to be done in order to develop this sensor for practical applications. These include the following: (1) further characterization of the sensitivity and dynamic range of the sensor as a function of the size of the analyte and its binding affinity in absolute terms, namely, molecules per unit area, (2) a direct comparison of the kinetics, steady-state response, and sensitivity of binding obtained with this method with conventional SPR, and (3) characterization of the effect of colloid size and surface density on the response of the sensor. Finally, the experimental results need to be understood within an appropriate model that can account for quantitative estimation of protein binding using a modified Mie theory for colloidal gold encapsulated in a shell of well-known thickness and the refractive index taking into account both chemical interface damping due to chemisorption of a thiol to gold and subsequent biomolecular interactions at that interface as a function of the distancedependent electromagnetic field at the solid-liquid interface. In conclusion, we have demonstrated proof of principle of a label-free, optical biosensor that exploits colloidal SPR of immobilized, self-assembled gold nanoparticles on an optically transparent substrate. Our implementation of colloidal SPR, reported here, provides an experimentally simple and convenient biosensor that can be easily implemented in most laboratories. A key finding of this paper, which enabled realization of this biosensor, is our observation that the change in the absorbance spectrum associated with biomolecular binding to the immobilized colloids can be easily measured in a UV-visible spectrophotometer with an analytical sensitivity and temporal resolution that is sufficient to quantify these interactions. The primary advantage of this sensor is its simplicity and flexibility at several different levels in our implementation. First, gold nanoparticles are easily prepared, and can be easily and reproducibly deposited on glass (or other optically transparent substrate) by solution self-assembly. Second, the spontaneous selfassembly of alkanethiols on gold allows convenient fabrication of surfaces with well-defined interfacial properties and reactive groups, which allows the chemistry at the interface to be easily tailored for a specific application of interest, an advantage this sensor shares with conventional SPR on gold or silver films. Third, this sensor enables label-free detection of biomolecular interactions. The biosensor can also be easily multiplexed to enable highthroughput screening in an array-based format for applications in genomics, proteomics, and drug discovery. ACKNOWLEDGMENT This work was supported in part by grants to A.C. from the National Science Foundation (NSF-BES-97-33009-CAREER and NSF-BES-99-86477-NANOSCALE) and the National Institutes of Health (R01-GM-61232-01). We thank Dr. Thomas LaBean and Dr. Dage Liu at Duke University for the AFM measurements. The AFM instrumentation used in these studies was acquired through multi-investigator research instrumentation awards from the National Science Foundation (NSF-DBI-96-04785) and the North Carolina Biotechnology Center (9703-IDG-1002) to Duke University. Received for review November 20, 2001.

October

24,

2001.

Accepted

AC015657X Analytical Chemistry, Vol. 74, No. 3, February 1, 2002

509