High-Throughput Nanohole Array Based System ... - ACS Publications

The core of the system is a sensing chip containing multiple nanohole arrays ... a temporal resolution on milliseconds scale that is decided only by t...
0 downloads 0 Views 327KB Size
Download PDF
Anal. Chem. 2008, 80, 2491-2498

High-Throughput Nanohole Array Based System To Monitor Multiple Binding Events in Real Time Jin Ji,*,† J. Garland O’Connell,† David J. D. Carter,‡ and Dale N. Larson†

Technology and Engineering Center, Department of Biochemistry and Molecular Pharmacology, Harvard Medical School, 240 Longwood Avenue, Boston, Massachusetts 02115, and The Charles Stark Draper Laboratory, 555 Technology Square, Cambridge, Massachusetts 02139

We have developed an integrated label-free, real-time sensing system that is able to monitor multiple biomolecular binding events based on the changes in the intensity of extraordinary optical transmission (EOT) through nanohole arrays. The core of the system is a sensing chip containing multiple nanohole arrays embedded within an optically thick gold film, where each array functions as an independent sensor. Each array is a square array containing 10 × 10 nanoholes (150 nm in diameter), occupying a total area of 3.3 µm × 3.3 µm. The integrated system includes a laser light source, a temperature-regulated flow cell encasing the sensing chip, motorized optics, and a charge-coupled detector (CCD) camera. For demonstration purposes, sensing chips containing 25 nanohole arrays were studied for their use in multiplexed detection, although the sensing chip could be easily populated to contain up to 20 164 nanohole arrays within its 0.64 cm2 sensing area. Using this system, we successfully recorded 25 separate binding curves between glutathione S-transferase (GST) and antiGST simultaneously in real time with good sensitivity. The system responds to binding events in a concentrationdependent manner, showing a sharp linear response to anti-GST at concentrations ranging from 13 to 290 nM. The EOT intensity-based approach also enables the system to monitor multiple bindings simultaneously and continuously, offering a temporal resolution on milliseconds scale that is decided only by the camera speed and exposure time. The small footprint of the sensing arrays combined with the EOT intensity-based approach enables the system to resolve binding events that occurred on nanohole sensing arrays spaced 96 µm apart, with a reasonable prediction of resolving binding events spaced 56 µm apart. This work represents a new direction that implements nanohole arrays and EOT intensity to meet high-throughput, spatial and temporal resolution, and sensitivity requirements in drug discovery and proteomics studies.

* To whom correspondence should be addressed. E-mail: [email protected]. edu. Fax: 617-432-1889. † Harvard Medical School. ‡ The Charles Stark Draper Laboratory. 10.1021/ac7023206 CCC: $40.75 Published on Web 02/29/2008

© 2008 American Chemical Society

Protein interactions in cellular environment regulate many processes that are vital to gene expression, virology, and cell signaling.1-4 These interactions include protein-protein interactions and interactions between proteins and other molecules such as metabolites, lipids, nucleic acids, carbohydrates, and drug molecules.5 Monitoring these interactions in real time can reveal the dynamic binding characteristics of these binding events, helping researchers to decipher the functional responsibilities of proteins and the cellular network where they operate. Due to the interdependence and the fast kinetics rate of many binding events, techniques that can monitor multiple binding events simultaneously with high temporal and spatial resolution are particularly desirable. A system offering these capabilities in a high-throughput manner would tremendously advance the researches in proteomics and drug discovery where thousands of proteins interactions need to be identified and characterized. Traditional approaches for studying protein interactions include the yeast two-hybrid (YTH) system, fluorescence techniques, and solution biochemistry based methods such as tandem affinity purification (TAP) and immunoprecipitation (IP). All these methods require a label or use mass spectrometry (MS) to report the binding partners. Several large-scale efforts have used IP coupled with a mass spectrometer (IP/MS) to identify protein interactions in the yeast proteome.6,7 These studies generated a large amount of data that revealed a complex proteomic network. The data also revealed that comparable efforts from multiple laboratories (1) Lin, L.; Harris, J. W.; Thompson, H. G. R.; Brody, J. P. Anal. Chem. 2004, 76, 6555-6559. (2) Rich, R. L.; Myszka, D. G. Drug Discovery Today 2004, 1, 301-308. (3) Shumaker-Parry, J. S.; Aebersold, R.; Campbell, C. T. Anal. Chem. 2004, 76, 2071-2082. (4) Rao, J.; Yan, L.; Xu, B.; Whitesides, G. M. J. Am. Chem. Soc. 1999, 121, 2629-2630. (5) Ramachandran, N.; Larson, D. N.; Stark, P. R. H.; Hainsworth, E.; LaBaer, J. FEBS J. 2005, 272, 5412-5425. (6) Li, S.; Armstrong, C. M.; Bertin, N.; Ge, H.; Milstein, S.; Boxem, M.; Vidalain, P.-O.; Han, J.-D. J.; Chesneau, A.; Hao, T.; Goldberg, D. S.; Li, N.; Martinez, M.; Rual, J.-F.; Lamesch, P.; Xu, L.; Tewari, M.; Wong, S. L.; Zhang, L. V.; Berriz, G. F.; Jacotot, L.; Vaglio, P.; Reboul, J.; Hirozane-Kishikawa, T.; Li, Q.; Gabel, H. W.; Elewa, A.; Baumgartner, B.; Rose, D. J.; Yu, H.; Bosak, S.; Sequerra, R.; Fraser, A.; Mango, S. E.; Saxton, W. M.; Strome, S.; van den Heuvel, S.; Piano, F.; Vandenhaute, J.; Sardet, C.; Gerstein, M.; DoucetteStamm, L.; Gunsalus, K. C.; Harper, J. W.; Cusick, M. E.; Roth, F. P.; Hill, D. E.; Vidal, M. Science 2004, 303, 540-544. (7) Uetz, P.; Giot, L.; Cagney, G.; Mansfield, T. A.; Judson, R. S.; Knight, J. R.; Lockshon, D.; Narayan, Y. a.; Srinivasan, M.; Pochart, P.; Qureshi-Emili, A.; Li, Y.; Godwin, B.; Conover, D.; Kalbfleisch, T.; Vijayadamodar, G.; Yang, M.; Johnston, M.; Fields, S.; Rothberg, J. M. Nature 2000, 403, 623-627.

Analytical Chemistry, Vol. 80, No. 7, April 1, 2008 2491

resulted in only a 10% overlap in the number of interactions identified in yeast,8 suggesting the need for a more direct and accurate method for protein interaction studies. Furthermore, these methods are limited to detecting protein-protein interactions.5 Methods allowing studies on bindings between proteins and other biomolecules such as lipids, DNA, and small molecules in high-throughput manner are demanded. A number of label-free techniques, including surface plasmon resonance (SPR), microcantilevers, and semiconductor nanowires, have been developed to monitor these biological binding events. Such techniques first immobilize one component of a binding pair on a signal transducer. Then, upon exposing the transducer to the complementary component of the binding pair, these techniques detect the binding event based on changes of refractive index, bending or vibration, and conductivity of the transducer, respectively. Among all the label-free techniques, SPR techniques are the most mature. Two common configurations used in this approach are prism-based SPR, also called conventional SPR, and grating-coupled SPR (GC-SPR). Prism-based SPR systems have widely been considered more sensitive than GC-SPR, though commonly they have low throughput.9-11 A few commercial SPR instruments and prototypes have been recently developed to address the high-throughput issue, the most remarkable ones of which are Biacore and Lumera.12,13 Microcantilever techniques that monitor the bending or the mechanical resonant frequency of the cantilever to monitor binding events have been traditionally viewed as off-line methods with low sensitivity.14,15 Recent advance in piezoelectric materials has enabled microcantilevers in monitoring binding in real time with much improved sensitivity,16 but the techniques are still far from achieving high-throughput capability. Nanowires17-19 monitor binding events by measuring charge at the nanowire surface. This technology takes advantage of wellestablished electrical detection methods, positioning itself as a sensitive, real-time method with high temporal resolution. Recent work in controlled nanowire assembly also demonstrated high multiplexing capability with this technology.19 However, the technology faces a major limitation that its detection sensitivity depends on solution ionic strength, making it unsuitable to use in real biological samples.18 There is also a practical constraint that the fabrication of nanowires requires extremely specific training that is unavailable in traditional laboratories. Spinning-disk microinterferometry is another label-free technique that integrates a large number of interferometers onto a spinning disk, analogous to an optical CD, to achieve high(8) Ito, T.; Ota, K.; Kubota, H.; Yamaguchi, Y.; Chiba, T.; Sakuraba, K.; Yoshida, M. Mol. Cell. Proteomics 2002, 1, 561-566. (9) Homola, J. Sens. Actuators, B 1997, B41, 207-211. (10) Homola, J.; Koudela, I.; Yee, S. S. Sens. Actuators, B 1999, B54, 16-24. (11) Homola, J.; Yee, S. S.; Gauglitz, G. Sens. Actuators, B 1999, B54, 3-15. (12) Saefsten, P.; Klakamp, S. L.; Drake, A. W.; Karlsson, R.; Myszka, D. G. Anal. Biochem. 2006, 353, 181-190. (13) Campbell, C. T.; Kim, G. Biomaterials 2007, 28, 2380-2392. (14) Verbridge, S. S.; Bellan, L. M.; Parpia, J. M.; Craighead, H. G. Nano Lett. 2006, 6, 2109-2114. (15) Cooper, M. A.; Singleton, V. T. J. Mol. Recognit. 2007, 20, 154-184. (16) Kwon, T. Y.; Eom, K.; Park, J. H.; Yoon, D. S.; Kim, T. S.; Lee, H. L. Appl. Phys. Lett. 2007, 90, 223903/223901-223903/223903. (17) Cui, Y.; Wei, Q.; Park, H.; Lieber, C. M. Science 2001, 293, 1289-1292. (18) Patolsky, F.; Zheng, G.; Lieber, C. M. Anal. Chem. 2006, 78, 4260-4269. (19) Jin, S.; Whang, D.; McAlpine, M. C.; Friedman, R. S.; Wu, Y.; Lieber, C. M. Nano Lett. 2004, 4, 915-919.

2492

Analytical Chemistry, Vol. 80, No. 7, April 1, 2008

throughput, label-free detection of binding events.20,21 The approach is novel but does not offer real-time detection, presumably due to the difficulty of incorporating a spinning component within a flow path to allow the binding events to occur in real time. There are also a few other label-free techniques including the ellipsometry-based microscope,22,23 Kelvin nanoprobes,24 micro-surfaceenhanced Raman scattering (µSERS),25 nanogap dielectric biosensor,26 and localized surface plasmon resonance (LSPR).27 These techniques are in various stages of development and are not discussed here. Recent work with the extraordinary optical transmission (EOT) presented a new approach in probing surface properties. The discovery, reported by Ebbesen and co-workers in 1998, indicates that the light transmission through a periodic array of subwavelength holes in an optically thick metallic film can be several orders of magnitude higher than that described in the classical theorysBethe’s theory.28-30 Since its discovery, EOT through nanohole arrays has been studied extensively in all aspects, including the theory and evidence of its origin,31,32 parameters affecting its intensity,33-37 and other optical properties such as its spectra and divergence.38-40 Although still not fully understood, this extraordinary transmission is commonly explained as a result of resonant coupling between incident light and surface plasmon waves at the metal-dielectric interface through nanohole arrays. The extent of the EOT depends on several parameters including the refractive index of the medium on the metal film surface, the wavelength of the incident light, the hole geometry and periodicity. In 2002, our group first presented the concept of using EOT and (20) Peng, L.; Varma, M. M.; Cho, W.; Regnier, F. E.; Nolte, D. D. Appl. Opt. 2007, 46, 5384-5395. (21) Varma, M. M.; Inerowicz, H. D.; Regnier, F. E.; Nolte, D. D. Biosens. Bioelectron. 2004, 19, 1371-1376. (22) Landry, J. P.; Zhu, X. D.; Gregg, J. P. Opt. Lett. 2004, 29, 581-583. (23) Zhu, X.; Landry, J. P.; Sun, Y. S.; Gregg, J. P.; Lam, K. S.; Guo, X. Appl. Opt. 2007, 46, 1890-1895. (24) Thompson, M.; Cheran, L.-E.; Zhang, M.; Chacko, M.; Huo, H.; Sadeghi, S. Biosens. Bioelectron. 2005, 20, 1471-1481. (25) Grow, A. E.; Wood, L. L.; Claycomb, J. L.; Thompson, P. A. J. Microbiol. Methods 2003, 53, 221-233. (26) Yi, M.; Jeong, K.-H.; Lee, L. P. Biosens. Bioelectron. 2005, 20, 1320-1326. (27) Yonzon, C. R.; Jeoung, E.; Zou, S.; Schatz, G. C.; Mrksich, M.; Van Duyne, R. P. J. Am. Chem. Soc. 2004, 126, 12669-12676. (28) Ebbesen, T. W.; Lezec, H. J.; Ghaemi, H. F.; Thio, T.; Wolff, P. A. Nature 1998, 391, 667-669. (29) Ghaemi, H. F.; Thio, T.; Grupp, D. E.; Ebbesen, T. W.; Lezec, H. J. Phys. Rev. B 1998, 58, 6779-6782. (30) Bethe, H. A.; Von der Lage, F. C. Phys. Rev. 1944, 65, 255. (31) Gao, H.; Henzie, J.; Odom, T. W. Nano Lett. 2006, 6, 2104-2108. (32) Martin-Moreno, L.; Garcia-Vidal, F. J.; Lezec, H. J.; Pellerin, K. M.; Thio, T.; Pendry, J. B.; Ebbesen, T. W. Phys. Rev. Lett. 2001, 86, 1114-1117. (33) Gordon, R.; Hughes, M.; Leathem, B.; Kavanagh, K. L.; Brolo, A. G. Nano Lett. 2005, 5, 1243-1246. (34) Degiron, A.; Lezec, H. J.; Barnes, W. L.; Ebbesen, T. W. Appl. Phys. Lett. 2002, 81, 4327-4329. (35) Koerkamp, K. J. K.; Enoch, S.; Segerink, F. B.; Van Hulst, N. F.; Kuipers, L. Phys. Rev. Lett. 2004, 92, 183901/183901-183901/183904. (36) Amarie, D.; Rawlinson, N. D.; Schaich, W. L.; Dragnea, B.; Jacobson, S. C. Nano Lett. 2005, 5, 1227-1230. (37) Williams, S. M.; Stafford, A. D.; Rogers, T. M.; Bishop, S. R.; Coe, J. V. Appl. Phys. Lett. 2004, 85, 1472-1474. (38) Tetz, K. A.; Pang, L.; Fainman, Y. Opt. Lett. 2006, 31, 1528-1530. (39) Martin-Moreno, L.; Garcia-Vidal, F. J.; Lezec, H. J.; Degiron, A.; Ebbesen, T. W. Phys. Rev. Lett. 2003, 90, 167401. (40) Lezec, H. J.; Degiron, A.; Devaux, E.; Linke, R. A.; Martin-Moreno, L.; GarciaVidal, F. J.; Ebbesen, T. W. Science 2002, 297, 820-822.

nanohole arrays as biological sensors.41 Since EOT is in part determined by the refractive index of the medium on the metal film surface, one could use EOT to monitor the surface refractive index, which is directly related to the physical and chemical properties of the material on the surface. This concept prompted much research that studied surface chemical or physical reactions on a metal film based on either the intensity42 or spectra38,43-46 of EOT through nanohole arrays sensors. Functioning as biosensors, nanohole arrays have several unique advantages: (1) The extremely small footprint of a nanohole array enables a high density of the sensors, facilitating their integration with microfluidics and other miniaturized electronics to achieve real-time detection as well as high throughput. (2) In contrast to detecting reflected light as in conventional label-free sensors such as prism- or gratingbased SPR sensors, nanohole array sensors detect EOT, i.e., transmitted light through nanoholes. This dramatic departure simplifies the instrumentation integration and miniaturization. (3) EOT through nanohole arrays is affected by many parameters including the shape and periodicity of the holes. These parameters can be used to tailor sensor signal according to various applications. (4) When fabricated by electron-beam or stepper-based lithography, the nanohole arrays can be mass-produced at a fast rate with high reproducibility, making this sensing approach truly high throughput. This overcomes the shortcoming of low productivity experienced by current fabrication techniquesfocused ion beam milling. Work to date in this area has focused on studying the spectra of isolated nanohole arrays to detect bulk solution and absorbed biomaterial on a metal film through nanohole arrays. In the direction of multiplexing, it is only recently that Brolo and Gordon’s group reported a 12-channel, EOT spectra-shift-based nanohole sensing array that detects absorption of biological material sequentially over 12 nanohole sensing arrays.43 Although promising, the spectra-based approach limits the sensor to detecting multiple events sequentially instead of simultaneously, which would be unsuitable when high temporal resolution is required. Here we demonstrate simultaneous detection of multiple biological binding events in real time by monitoring the intensity of EOT through nanohole sensing arrays. Besides the small sensor footprint, easy system integration, and abundant optimization parameters that all nanohole array based sensing techniques enjoy, we offer extremely high potential of multiplexing by using intensity of EOT to probe binding events, a dramatic departure from previously reported spectra interrogation of EOT. Our previous studies also showed that EOT intensity-based sensing approach offers superb sensitivity of 9.4 × 10-8 RIU, exceeding the theoretical limit of what prism-based SPRs could offer.11,42 Work reported below indicates that we can monitor binding events (41) Stark, P.; Larson, D. Surface Plasmon Enhanced Illumination and Its Applications. In Progress in Electromagnetics Research Symposium (PIERS); Cambridge, MA, 2002. (42) Stark, P. R. H.; Halleck, A. E.; Larson, D. N. Methods 2005, 37, 37-47. (43) De Leebeeck, A.; Kumar, L. K. S.; De Lange, V.; Sinton, D.; Gordon, R.; Brolo, A. G. Anal. Chem. 2007, 79, 4094-4100. (44) Rindzevicius, T.; Alaverdyan, Y.; Dahlin, A.; Hook, F.; Sutherland, D. S.; Kall, M. Nano Lett. 2005, 5, 2335-2339. (45) Williams, S. M.; Rodriguez, K. R.; Teeters-Kennedy, S.; Stafford, A. D.; Bishop, S. R.; Lincoln, U. K.; Coe, J. V. J. Phys. Chem. B 2004, 108, 1183311837. (46) Brolo, A. G.; Gordon, R.; Leathem, B.; Kavanagh, K. L. Langmuir 2004, 20, 4813-4815.

at a sampling rate of a millisecond scale and resolve bindings that occurred on locations only 96 µm apart with a reasonable prediction of spatial resolution of 56 µm. Such combination of high throughput, sensitivity, and temporal and spatial resolution has not been achieved by any label-free or labeled technologies. To implement the studies, we designed and built an integrated system that includes a miniaturized flow cell to encase the nanohole sensing arrays and maintain a stable temperature, a motorized stage to align the sensors with the laser light source, optics and a high-performance charge-coupled detector (CCD) for real-time detection, and a syringe pump and injector for sample injection. For fast chip fabrication, we developed an electron-beam directwrite lithography process to create multiple nanohole sensing arrays on a thin gold film, each nanohole sensing array containing 10 × 10 nanoholes with diameters of 150 nm. The combined signal from these 100 nanoholes in one sensing array functions as one independent probe. The size of each nanohole array is 3.3 µm by 3.3 µm. We successfully functionalized the sensor surface with glutathione S-transferase (GST), a protein used as a model antigen, and monitored in real time the specific binding of its counterpart, anti-GST antibody. For demonstration purpose, we present here the studies done on 25 nanohole sensing arrays that are located 96 µm apart, although we reasonably predict that we could resolve up to 20 164 independent binding events spacing 56 µm apart. EXPERIMENTAL SETUP Nanohole Array Fabrication. The nanohole arrays used in this study were fabricated by two methods: focused ion beam (FIB) milling and direct-write electron-beam lithography. The FIB method was developed in previous studies30 and was used in this work to fabricate small numbers of sensing chips of various designs for exploratory studies. To achieve fast and reproducible nanohole array fabrication, we also developed an electron-beam lithography procedure to produce a large number of sensing chips. To the best of our knowledge, such a method (schematics shown in Figure 1) has not been reported. In this method, nanohole arrays were fabricated by electron-beam lithography and lift-off of metals on 762 µm thick glass wafers. We first coated wafers with 210 nm of spun-on LOR resist (Microchem), 80 nm of electron-beam evaporated SiO2, 200 nm of spun-on maN-2403 negative electron-beam resist (MicroResist Technology), and a few nanometers of spun-on ESPACER organic conductor (Showa Denko). We then performed electronbeam lithography using a Raith 150 system at 30 keV acceleration voltage. Each independent sensor on the chip consisted of a 10 × 10 array of nanoholes pitched at 350 nm. Each sensing chip contained 25 sensors spaced 96 µm apart. After electron-beam patterning, the wafer was rinsed with deionized water to strip the ESPACER, then the maN-2403 resist was developed in MF-319 developer (Shipley). The remaining layers were then anisotropically etched with reactive ion etching (RIE) in a Surface Technology Systems (STS) Multiplex RIE. SiO2 was etched with CHF3 and CF4, and the LOR resist was etched with O2. The LOR etch was endpoint monitored with a Plasmascope optical emission spectrometer (Horiba-Yvon) to minimize overetch and undercutting of the LOR. After etching, 25 nm of Cr and 105 nm of Au were deposited by electron-beam evaporation. This process resulted in a resist stack profile that gave a clean break between the metal on the substrate and the Analytical Chemistry, Vol. 80, No. 7, April 1, 2008

2493

Figure 1. Schematic of the electron-beam lithography procedure. The drawing is not to scale.

Figure 2. (A) Schematic of the optical and mechanical setup of the biosensing system. The drawing is not to scale. (B) Photograph of the flow chamber.

small dots deposited on the resist nanoposts. The nanoposts were then lifted off, resulting nanohole pattern in the thin gold film. System Instrumentation. Figure 2A depicts the optical and mechanical setup of the prototype biosensing instrumentation. A 635 nm diode laser (Coherent ULN series) with a 1 mm beam diameter is used as the light source. The light is expanded through 2494

Analytical Chemistry, Vol. 80, No. 7, April 1, 2008

a 10× beam expander and directed to the nanohole array based sensing chip through two front surface mirrors. An optical aperture is placed in between these two mirrors to block out unwanted stray light. The position of the second mirror is controlled by a motorized stage (Thorlab) to reflect the laser onto the sensing chip at incident angles ranging from normal to 30°. A flow cell assembly was designed to encase the nanohole array based sensing chip and control its temperature. The flow cell assembly includes a flow chamber that is made by punching out a small chamber from a 76 µm thick silicone membrane (Bentec Medical) and sandwiching the small chamber between the sensing chip and a cover glass. The cover glass includes two drilled holes and inlet and outlet ports to allow sample flow. The volume of the flow chamber is ∼3 µL. A photograph of the flow chamber is shown in Figure 2B. A syringe pump located upstream of the flow chamber continuously drives buffer through the flow chamber, while a sample injector (Rheodyne) placed immediately before the flow chamber introduces samples of interest. The flow cell assembly also contains temperature control components including a small resistive heater, a thermoelectric cooler, and a custommade aluminum heat sink. Two thermistors (Wavelength Electronics) are also embedded in the flow cell. One thermistor monitors the temperature in the flow chamber, while the other feeds back to a temperature controller (Wavelength Electronics). The flow cell assembly is also controlled by a motorized stage that aligns the sensing chip with the optical path of the laser and the CCD camera. Labview programs developed with Labview 8.0 (National Instruments) were used to control all the motorized motions. The precision of the motion is 0.05 µm. A 1× telecentric lens with numerical aperture (NA) of 0.14 (Schneider) is then used to project the transmitted light from nanohole arrays on to a CCD Camera (Hamamatsu) that has a frame-grabbing rate of 30 frames/s. Transmission Spectra. Transmission spectra of the nanohole arrays were taken using an inverted microscope (Nikon eclipse TE300) equipped with an uncooled CCD (Qimaging). White light from a Xe arc lamp (PTI 1010B 75W) is passed through a double monochromator (Jobin-Yvon SPEX 1680B) and optics to produce collimated light with various wavelengths. Spectral transmission data from the arrays were collected by scanning the wavelength of the light through the monochromator while monitoring the transmission through the CCD. Functionalize Nanohole Array Sensing Chip with GST. We modified the nanohole array sensing chip with GST by covalently attaching the protein onto a self-assembled monolayer (SAM) of amino-functionalized thiol (HSC11EG6NH2, Prochimia) on the gold surface. We first functionalized the gold surface with amine functional groups by forming an amino-functionalized SAM. We then attached the protein GST onto the gold surface via a bifunctional cross-linker that covalently attaches to amine groups on both GST and the gold surface. Before exposing to thiol solution to form SAM on its surface, the gold sensing chip was cleaned by 15 min of sonication in acetone and ethanol, respectively. The chip surface was further treated with freshly prepared piranha solution for 1 min to be free of any organic residues. After a thorough rinse in distilled water and air-can blow dry, the chip was immersed in freshly prepared 2 mM amino-functionalized thiol in anhydrous ethanol for 16 h to

form an ordered amino-functionalized SAM. The chip was then rinsed with ethanol to be free of loosely bound SAM and dried with an air-can. After obtaining SAM on the sensing chip, 50 µL of 500 µg/mL GST (Sigma) containing 2 mM bis(sulfosuccinimidyl)suberate (BS3, obtained from Pierce) in PBS 7.4 buffer was deposited onto the center of the chip where the nanohole arrays were. BS3 is a homobifunctional cross-linker containing an amine-reactive Nhydroxysulfosuccinimide (NHS) ester at each end of an eightcarbon spacer arm. Here, BS3 functions as a cross-linker that attaches GST molecules onto the gold film through the amine functional groups on both GST and SAM. One hour later, the GST/BS3 solution was removed and the gold chip was treated with 1 M glycine (Lancaster Synthesis, Inc.), pH 10 solution, to quench any ongoing reaction. The gold chip was then rinsed with water and blown dry. This resulted in a sensing chip with GST covalently attached to its surface. The chip was stored in air at 4 °C before use. Antigen-Antibody Binding Events Detection and Data Analysis. Different concentrations of anti-GST antibodies (GE Health) prepared in PBS 7.4 buffer (tablets obtained from Sigma) were introduced to the flow cell through a sample injector described above. Images of nanohole sensing arrays were monitored through a CCD camera to interpret the binding events between the GST immobilized on the sensing chip and the antibodies injected into the flow cell. Custom-developed Matlab programs were used to process the CCD images and output an EOT intensity versus time graph. To eliminate the effect of laser fluctuation and other systematic effect on sensor signal, the intensity of the film where no nanohole sensing array was present was used as the background and was subtracted from the intensity of nanohole arrays. After a binding event occurred, 10 mM glycine, pH 2.2 solution, was injected into the flow cell to regenerate the sensing chip. Other Reagents and Materials. Antihuman IgG was purchased from Sigma. Unless otherwise stated, all aqueous solutions were prepared with 18.2 MΩ‚cm at 25 °C deionized-distilled water produced from a Millipore Simplicity water purification system. RESULTS AND DISCUSSION Design of Nanohole Arrays. As described above EOT results from incoming light resonating with the surface plasmon (SP) waves on the metal surface via nanohole arrays. Similar to other periodic systems, the degree of EOT depends on the shape of the basis and the lattice of the arrays. Our group has been investigating nanohole arrays with different lattice constants, hole diameters, hole patterns, and number of holes. For this study we standardized the array design to be a 10 by 10 nanohole array with hole diameter (d) ) 150 nm, lattice constant (a) ) 350 nm, and thickness of the gold film (t) of 105 nm. The choice of hole diameter of 150 nm is decided by the easy fabrication of relatively large holes using FIB and electron-beam lithography. The lattice constant was determined by the wavelength of the monochromatic laser light used (635 ( 5 nm) based on the equation below:42,47 (47) Raether, H. Surface Plasmons on Smooth and Rough Surfaces and on Gratings; Springer-Verlag: New York, 1988.

( ) 12 1 + 2

1/2

- sin θ )

νλ a

Here ν is an integer; 1 and 2 are dielectric constants of the interfacing media (gold and buffer solution in our case). θ is the incident angle of the light with the sensor surface. The transmission spectra of different sensing arrays showed a consistent peak around 630 nm, close to the theoretical prediction. We chose a red light source because light in this region does not trigger significant intrinsic fluorescence of biological samples. The long wavelength is also compatible with optics made of a broad range of materials, making it easy for system design and building. We selected the thickness of 105 nm based on the combination of several considerations: (1) Ebbesen’s group reported that when the ratio of hole depth and hole diameter t/d, i.e., the ratio of the thickness of metal film and hole diameter, is close to 1, maximum transmission peak sharpness is obtained.28,48 This indicates that t/d ∼ 1 is a possible optimum condition when using EOT through nanohole arrays as a monitoring tool. (2) Thickness of 105 nm is more than twice the SP penetration depth for 635 nm light in gold film, which means SPs on one side of the film have minimum direct communication with the SPs on the other side of such optically thick film.42 (3) The electron-beam lithography and lift-off fabrication process favors thin films in order to achieve good yield of nanohole sensing arrays. Studies have shown that the type of lattice, for instance square or triangle, has significant impact on the spectra of the EOT,28 therefore potentially the intensity of EOT. For this study, we randomly chose 10 × 10 square nanohole arrays as shown in Figure 3B. Presently, we are actively studying the effect of different array designs on their biosensing capability to optimize the array design. Electron-Beam Lithography Approach. Patterning nanoholes in gold with electron-beam lithography is difficult; lack of a controllable dry etch for gold requires a lift-off process where small nanoposts are patterned in resist. Electron proximity effects dictate that a negative electron-beam resist must be used. Because of backscattering of electrons from the substrate, a single-layer negative resist process yields a sloped profile to the resist that is not amenable to lift-off. In this work, we adapted trilevel resist processes previously reported for positive electron-beam resist49 and for optical interference lithography50 to provide an optimal profile for lift-off. Our adaptation from previous work resulted in an oxide interlayer, a bottom LOR layer for easier removal, and a negative resist layer on top. Nanoholes fabricated using this approach showed well-defined shapes and high batch-to-batch consistency. Figure 3A shows an optical micrograph of nanohole array chip with 25 independent sensors. The footprint of the entire 25 independent sensors is less than 0.2 mm2. Figure 3B is a scanning electron micrograph of an individual sensor, i.e., a 10 × 10 (48) Degiron, A.; Lezec, H. J.; Yamamoto, N.; Ebbesen, T. W. Opt. Commun. 2004, 239, 61-66. (49) Tennant, D. M.; Jackel, L. D.; Howard, R. E.; Hu, E. L.; Grabbe, P.; Capik, R. J.; Schneider, B. S. J. Vac. Sci. Technol. 1981, 19, 1304-1307. (50) Schattenburg, M. L.; Aucoin, R. J.; Fleming, R. C. J. Vac. Sci. Technol., B 1995, 13, 3007-3011.

Analytical Chemistry, Vol. 80, No. 7, April 1, 2008

2495

Figure 3. (A) Optical micrograph of 25 independent nanohole sensing arrays on a gold chip. The array-to-array distance is 96 µm. (B) Scanning electron microscope (SEM) image of an individual sensing array that contains a 10 × 10 nanohole array. The array is fabricated by electron-beam direct-write lithography method. Magnification, 14.95k×; hole diameter, 150 nm; hole-to-hole distance, 350 nm.

Figure 5. Response of 25 nanohole sensing arrays to 290 nM antiGST followed by surface regeneration reagent.

Figure 4. CCD image of 25 nanohole sensing arrays. Each bright spot is the image of one sensing array containing 100 nanoholes. Laser incident angle: 13°.

nanohole array fabricated by electron-beam lithography. The area of one sensor, i.e., the total area of 100 nanoholes, is 3.3 µm × 3.3 µm. Using the developed electron-beam method, we can batch fabricate sensing chips at a rate of 20 min per chip. The active personnel time required for batch fabrication is 7 min per chip High-Throughput Capability of the Nanohole Array Based Biosensing System. We fabricated 25 nanohole arrays spacing at 96 µm apart (shown in Figure 3B) to demonstrate the use of these arrays as multiplexed sensors to monitor binding events. Under our optical setting that applies a 1× telecentric lens, the images of individual nanohole array based sensors should also be 96 µm apart. A CCD image of the sensors as shown in Figure 4 indicates that the images of individual sensors are 12 pixels apart. The physical size of the pixel in our CCD is 8 µm by 8 µm, indicating that the images of sensors are 96 µm apart as predicted. Figure 4 also suggests that the image of an individual sensor occupies an area of less than 4 pixels by 4 pixels. This means the images can be sufficiently resolved if sensors are spaced 56 µm apart, i.e., images of sensors are 7 pixels apart, leaving 3 pixels 2496 Analytical Chemistry, Vol. 80, No. 7, April 1, 2008

between sensors to ensure signal resolution. The CCD camera used in our system has 1000 by 1000 active pixels, with the pixel size of 8 µm by 8 µm. Through a 1× telecentric lens, the CCD chip can monitor a sensing area of 8 mm by 8 mm. When spacing sensors 56 µm apart, we can fabricate 142 by 142 sensors, i.e., 20 164 sensors on the area of 8 mm by 8 mm, while the resulted CCD images are still 7 pixels apart, sufficiently resolved. Such potential of high throughput and spatial resolution by a real-time, label-free biosensing technique has not been reported. Thermal Control of the Nanohole Array Based Biosensing System. Nanohole array based biosensors monitor the surface refractive index of gold film through EOT to interpret binding events. Since temperature variation could also affect the refractive index of surface bound proteins as well as the bulk buffer solutions, we, as well as all other SPR-based techniques, must account for the temperature effect. Temperature control is particularly necessary when pursuing sensitive detection. In our system, we minimized potential temperature interference by implementing miniature temperature control components in the flow cell. Controlling these components through a temperature controller (Wavelength Electronics), we were able to maintain the sample temperature to within (0.005 °C. Our results indicate that this level of temperature consistency eliminated temperature effect on sensor signal to a negligible degree, offering a stable baseline. Monitoring Antigen-Antibody Binding Events through Transmission Intensity of Nanohole Arrays. Our group has previously demonstrated the use of a single nanohole array sensor to resolve very small changes in bulk indexes of refraction.42 Using specific binding events between GST and anti-GST as a model system, we demonstrate here that by monitoring EOT through multiple nanohole arrays, we were able to monitor multiple binding events between GST that was immobilized on nanohole array embedded gold film and anti-GST antibodies in a flow. Twenty-Five Binding Events Monitored Simultaneously. Figure 5 shows 25 simultaneous, separate binding events monitored through our sensing system based on EOT intensity of nanohole

Figure 6. Specific binding of the sensing arrays to anti-GST. The nanohole sensing arrays are modified with GST: (A) no binding observed between GST-modified sensing arrays and antihuman IgG; (B) binding events occurred between the sensing arrays and anti-GST.

sensing arrays. At t ) 16.75 min, we triggered the binding events by introducing 200 µL of 290 nM GST into the flow cell. Within 1 min, the EOT of the 25 nanohole sensing arrays displayed a significant increase, signaling the attachment of anti-GST onto the gold film through binding with the immobilized GST, which then triggered a change of surface refractive index. These binding events were further confirmed by a follow-up regeneration event. At t ) 37.47 min, we introduced 100 µL of the regeneration solution, 10 mM glycine, pH 2.2, to dissociate the binding between GST and anti-GST. Within 1 min, the intensity of EOT dropped down, returning to its baseline level. This set of data was taken when the laser incident angle was set at 13° off normal and the flow rate was 30 µL/mL. The choice of the laser incident angle will be discussed below. Data showed that under this setting, the baseline is fairly stable, with the noise, i.e., the standard deviation of the signal, remaining within 0.1% of the signal read-out. The average signal increase to noise ratio (∆S/N) of the data is 21.6. Angle of Incident Light. Previous studies have reported that the intensity and spectra of EOT could be affected by the incident angle of the light.28,51 The incident light angle therefore could affect the nanohole sensing array’s biosensing capability due to its impact on the intensity and spectra of EOT. We found in our studies that the incident light angle could also affect sensor biosensing capability due to its impact on background transmission noise. The skin depth for gold at 635 nm illumination is 86 µm, which means the optical power of the laser beam, after traveling through 86 µm of gold, would be attenuated to 1/e times.42 Although significantly attenuated, we observed that the amount of the light directly transmitted through 105 nm of background film is still a significant noise, presumably due to the relatively large background film area compared to the small footprint of the nanohole arrays. The direction of EOT is normal (51) Laluet, J. Y.; Devaux, E.; Genet, C.; Ebbesen, T. W. Opt. Express 2007, 15, 3488-3495.

to the sensing chip, whereas the directly transmitted light follows the incident angle of the laser light. Since we used a 1× telecentric lens with a low NA of 0.14 that favors light that is normal to the sensing chip, increasing the incident angle of the laser light, therefore the angle of directly transmitted light, would decrease the amount of background light collected by the lens. We conducted a series of tests to optimize the incident angle to achieve the best biosensing capability. Here, we scanned the incident angle from normal to 30° (the maximum angle achieved by our system). At each angle, we challenged the biosensors with 200 µL of 290 nM anti-GST followed by 10 mM glycine, pH 2.2, to regenerate the sensor. We found that the incident angle of the light had dramatic influence on sensor response as predicted (data not shown). When the angle is between 0° and 12°, ∆S/N is low with relatively large sensor-to-sensor variation. The average ∆S/N during this angle range is 5. When the laser incident angle is 13°, however, we observed a dramatic increase in ∆S/N, reaching 21.6 with a small sensor-to-sensor variation. Increasing the laser incident angle further did not significantly enhance ∆S/N. When the angle is in the range from 13° to 30°, the average ∆S/N observed is 21.9. The optimum angle of incident light is 13° where ∆S/N as well as sensor-to-sensor consistency are high. Therefore, we chose 13° as the incident laser angle for the rest of our studies. The significant angular dependence that we observed could be caused by both effects we speculated above. Our study indicated that incident angle is significant and should be optimized when using EOT-based technology in sensing application. Specificity, Sensitivity, and Dynamic Range of Nanohole Array Based Sensors. We then conducted a series of characterization studies on the nanohole array sensors with the laser incident angle at 13°. Figure 6 indicates that the GST-modified sensors responded specifically to anti-GST, but not to antihuman IgG which was used as a negative control. Figure 6A shows that after injection of 100 µL of 333 nM antihuman IgGsan antibody that was used as a Analytical Chemistry, Vol. 80, No. 7, April 1, 2008

2497

Figure 7. Concentration-dependent response of GST-functionalized nanohole sensing arrays. Each data point represents the average signal of 25 sensors. Error bars represent one standard deviation. A linear dynamic range with sensitive response was found to be from 13 to 290 nM.

negative controlsall five sensors showed no response, suggesting that no binding events occurred even though a high concentration of antibody was used. When challenged with the same concentration of anti-GST, all five sensors that were used for this study showed a significant increase in EOT (Figure 6B), signaling the binding events occurred between the anti-GST and GST immobilized on the sensors. To study the sensitivity of the nanohole array based sensors, we tested the response of 25 sensors to a series of anti-GST solutions of different concentrations. The average signal response (∆S) of 25 sensors was plotted against concentration in Figure 7. Data showed that the sensors could detect as low as 13 nM antiGST in a flow stream, where the average ∆S/N reaches 3. The signal response is concentration-dependent, showing a sharp linear response when anti-GST is within the range of 13 to 290 nM. A linear curve fitting for this range showed an R-square value of 0.9637. Further increasing the concentration of anti-GST beyond 290 nM also induced a linear response but with a much smaller slope, indicating less sensitivity in this region. The error bars in Figure 7 represent one standard deviation of the signal response generated from all 25 sensors. The average coefficient variance (CV) of the sensor response from 25 sensors is ∼20% under tested concentrations, representing a good sensor-to-sensor consistency. Similar studies on different sensor chips indicate that the automated electron-beam lithography procedure produced consistent nanohole arrays from chip-to-chip. Sensitivity and dynamic range of different sensor chips are comparable. SUMMARY AND CONCLUSION This paper demonstrated the use of nanohole sensing arrays and the intensity of EOT as a label-free, real-time approach to monitor biological binding events with extremely high throughput and temporal and spatial resolution. To achieve real-time detection, we designed and built a biosensing system that integrated a

2498

Analytical Chemistry, Vol. 80, No. 7, April 1, 2008

temperature-regulated flow cell encasing the nanohole sensing arrays, a laser light source, motorized optics, and a highperformance CCD camera. We have also developed an electronbeam direct-write lithography method to produce nanohole arrays in a fast and reproducible manner, more than 10 times faster than what the previously used FIB method could achieve. For demonstration purpose, we used the developed system to record the binding curves of 25 GST and anti-GST binding events simultaneously in real time; each binding event occurred on locations that are only 96 µm apart. The system responded to the anti-GST in a concentration-dependent manner, showing a sharp linear response to anti-GST ranging from 13 to 290 nM. Using EOT intensity as a probe, a dramatic shift from commonly studied EOT spectra, we were able to monitor multiple binding events simultaneously and continuously, in real time, without the need to scan and evaluate the binding locations one-by-one. This offers our system an extremely high temporal resolution, limited only by the camera speed and exposure time. Currently, we have been successfully capturing binding events at a sampling rate of 50 ms. Each of our sensing arrays contains 100 nanoholes, occupying a total area of 3.3 µm × 3.3 µm. The large number of nanoholes in one sensing array offered sufficient amount of the combined signal to achieve good sensitivity, while at the same time, the total footprint of 100 nanoholes is still within the micrometer scale. Such small footprint of the sensor combined with the system ability of resolving tightly spaced sensors offered a potential for extreme high throughput. Our data indicate that the system could be easily scaled up to monitor 20 164 binding events simultaneously, with a spatial resolution of 56 µm, without modification to the basic chip design or prototype instrumentation. This combination of remarkable high throughput, sensitivity, and temporal and spatial resolution would tremendously advance proteomics studies and drug discovery, where thousands of protein interactions and their interdependence need to be identified and characterized. Currently, we are actively studying the effect of a number of parameters on EOT intensity to optimize system performance. The parameters we are investigating include laser fluctuation and dispersion characterization, polarization of the laser, design of nanohole arrays such as the array period, shape, and material, and the sensing chip surface modification scheme. The results of these studies are expected to enhance the system sensitivity further to study more subtle binding systems such as the bindings between proteins and small molecules. ACKNOWLEDGMENT The authors are grateful for the financial support of the National Institutes of Health (Grant Nos. 5R01 HG003828-03 and 1R21 EB004333-01A2). We also thank Dr. Zeev Rosenzweig of National Science Foundation for his helpful review and discussion.

Received for review November 9, 2007. Accepted January 25, 2008. AC7023206