Anal. Chem. 2006, 78, 5578-5583
Theoretical and Experimental Analysis of Arrayed Imaging Reflectometry as a Sensitive Proteomics Technique Charles R. Mace,† Christopher C. Striemer,‡ and Benjamin L. Miller*,†,§
Department of Dermatology, Department of Biochemistry and Biophysics, Pathologics, LLC, and The Center for Future Health, University of Rochester, Rochester, New York 14642
Arrayed imaging reflectometry (AIR) is a newly developed label-free optical biosensing technique based on the creation and perturbation of a condition of zero reflectance on a silicon substrate. The antireflective coating is formed by covalently immobilizing arrayed probes on a silicon dioxide film. Probe-target complex formation causes a localized increase in optical thickness and a measurable reflectance change. To evaluate the performance of AIR, we have employed two proteins, intimin and tir, from enteropathogenic E. coli that are critical to the bacterium’s mechanism of host infection. Using substrates functionalized with the intimin-binding domain of tir, we demonstrate detection of the extracellular domain of intimin at concentrations as low as 10 pM. Through the use of a diffusion-limited model for the intimin-tir binding interaction at this concentration, we estimate the detected intimin surface concentration to be 0.33 pg/mm2. Scientific advancement in the fields of genomics and proteomics continues to rely heavily on high-throughput arrays for the selective detection and quantitation of specified targets. Highdensity arrays for the detection of oligonucleotide hybridization have been developed1,2 based on fluorescence tagging, and analogous techniques for protein arrays based on enzyme-linked immunosorbant assays have also been described.3,4 However, a major focus of current research efforts is the design of biosensing platforms that are both arrayable and label-free. Label-free detection reduces the complexity and cost of a technique, while simultaneously eliminating any potential errors stemming from the labeling process itself. The need for label-free sensing strategies has driven the development of optical transducers, techniques that are responsive to changes in optical thickness * To whom correspondence should be addressed. E-mail: benjamin_miller@ futurehealth.rochester.edu. † Department of Biochemistry and Biophysics. ‡ Pathologics, LLC. § Department of Dermatology. (1) Lipshutz, R. J.; Fodor, S. P. A.; Gingeras, T. R.; Lockhart, D. J. Nat. Genet. (Suppl) 1999, 21, 20-24. (2) McCaffrey, R. L.; Fawcett, P.; O’Riordan, M.; Lee, K.-D.; Havell, E. A.; Brown, P. O.; Portnoy, D. A. Proc. Natl. Acad. Sci. U.S.A. 2004, 101, 11386-11391. (3) Wiese, R.; Belosludstev, Y.; Powdrill, T.; Thompson, O.; Hogan, M. Clin. Chem. 2001, 47, 1451-1457. (4) Woodbury, R. L.; Varnum, S. M.; Zangar, R. C. J. Prot. Res. 2002, 1, 233237.
5578 Analytical Chemistry, Vol. 78, No. 15, August 1, 2006
(∆nd) of a surface film. Numerous biosensing methods rely on spectroscopic, angular, or imaging optical transducers for the detection of molecular interactions. Surface plasmon resonance5 (SPR) and reflectometric interference spectrometry6 examine the shift of a characteristic minimum or other spectral feature as a means of detecting binding or surface adsorption. Each experiment is performed in a single, isolated channel, so array data must be gathered separately. Alternatively, SPR imaging7 and imaging ellipsometry8 are two-dimensional analogues of known techniques, but sensitivity is compromised for this multiplexed capability. We have developed arrayed imaging reflectometry (AIR) as an optical, label-free technique for the quantitative detection of specific target analytes over a wide range of concentrations. AIR is capable of detecting small localized changes in the thickness of a thin film and can be used to simultaneously measure target binding on a substrate with arrayed probes. This method was initially introduced into the literature as “reflective interferometry”9,10 but has been renamed in order to avoid confusion with other interferometric techniques.11,12 AIR relies on the destructive interference of polarized light reflected off a silicon dioxide/silicon wafer substrate. For s-polarized light, there exist combinations of the silicon dioxide layer thickness, incident angle, and wavelength that will result in the total destructive interference of reflected light (Figure 1). For a fixed wavelength, a well-defined reflectivity zero exists, and deviations in angle or in thickness will increase the observed reflectivity. The silicon dioxide interference layer is readily functionalized with biomolecular probes; therefore, by tailoring the silicon dioxide layer thickness to achieve the zero reflectance condition postprobe immobilization, a highly sensitive system for the detection of probe-target complex formation is created. (5) Homola, J.; Yee, S. S.; Gauglitz, G. Sens. Actuators, B 1999, 54, 3-15. (6) Schmitt, H.-M.; Brecht, A.; Piehler, J.; Gauglitz, G. Biosens. Bioelectron. 1997, 12, 809-816. (7) Nelson, B. P.; Grimsrud, T. E.; Liles, M. R.; Goodman, R. M.; Corn, R. M. Anal. Chem. 2001, 73, 1-7. (8) Jin, G.; Tengvall, P.; Lundstrom, I.; Arwin, H. Anal. Biochem. 1995, 232, 69-72. (9) Lu, J.; Strohsahl, C. M.; Miller, B. L.; Rothberg, L. J. Anal. Chem. 2004, 76, 4416-4420. (10) Horner, S. R.; Mace, C. M.; Rothberg, L. J.; Miller, B. L. Biosens. Bioelectron. 2006, 21, 1659-1663. (11) Sauer, M.; Brecht, A.; Charisse, K.; Gerster, M.; Stemmler, I.; Gauglitz, G.; Bayer, E. Anal. Chem. 1999, 71, 2850-2857. (12) Ymeti, A.; Kanger, J. S.; Greve, J.; Besselink, G. A. J.; Lambeck, P. V.; Wign, R.; Heideman, R. G. Biosens. Bioelectron. 2005, 20, 1417, 1421. 10.1021/ac060473+ CCC: $33.50
© 2006 American Chemical Society Published on Web 07/04/2006
Figure 1. Simulated s-polarization reflectance spectra for AIR. The graph depicts broad ranges for both incident angle and silicon dioxide layer thickness at a fixed wavelength (632.8 nm). Total reflectance is shown as a logarithmic color scale ranging from 1 × 10-4 (blue) to 1 (red). The low dynamic range shown is due to the limited resolution of the simulation for this broad range of angle and thickness.
We have previously reported the utilization of cloned extracellular protein domains in concert with the AIR technique in order to create a biosensor capable of detecting the presence of enteropathogenic Escherichia coli (EPEC), a major cause of infantile diarrhea,13 directly in culture.10 Intimin and the translocated intimin receptor (tir) are two proteins involved in the attachment and effacement of numerous pathogenic E. coli strains, including EPEC, to host intestinal epithelium. Intimin is naturally expressed as an outer membrane protein, while tir is secreted via EPEC’s type III secretion system. Upon secretion, tir inserts itself into the plasma membrane of a host and acts as a cell anchoring point by binding to intimin.14 Because of the stability, level of structural characterization, and importance to human health of the intimin-tir system, it represented an outstanding test system for the further evaluation of AIR as a sensitive and quantitative technique. EXPERIMENTAL SECTION Surface Preparation. Polished crystalline silicon wafers coated with a highly uniform 1400-Å silicon dioxide layer (Rochester Institute of Technology) were diced into chips with dimensions of 12 mm × 20 mm. To finely tune the thicknesses of the starting material, individual chips were etched in a solution of dilute hydrofluoric acid (300:1 in Milli-Q water) to 1380 Å as measured by spectroscopic ellipsometry (J. A. Woollam, M-2000). The ellipsometer used in this study has a reproducible accuracy of (0.1 Å at this thickness. Protein Preparation. Vectors for both the intimin extracellular domain and the intimin-binding domain of tir (referred to as intimin-ECD and tir-IBD, respectively) were graciously donated by Professors Natalie Strynadka and Brett Finlay of the University of British Columbia. Proteins were expressed, purified, and quantified as described previously.10 (13) Clarke, S. C. Diag. Microbiol. Infect. Dis. 2001, 41, 93-98. (14) Luo, Y.; Frey, E. A.; Pfeutzner, R.; Craigh, L.; Knoechel, D.; Haynes, C.; Finlay, B. B.; Strynadka, N. C. Nature 2000, 405, 1073-1077.
Figure 2. Schematic of the imaging apparatus.
Probe Immobilization. Freshly HF etched chips were immersed in a solution of (γ-aminopropyl)triethoxysilane (APTES, Sigma Aldrich) and acetone (1.5 mL of a solution of 5% APTES in distilled water was added to 28.5 mL of acetone) for 15 min. Chips were then washed with distilled water, dried under a stream of N2, and cured at 100 °C for 15 min. After curing, chips were immersed in a solution of buffered glutaraldehyde (Sigma Aldrich, 1.25 mL of 50% aqueous glutaraldehyde was added to 48.75 mL of a solution containing 100 mM Na2HPO4 and 150 mM NaCl at pH 7.2) for 60 min. Note that glutaraldehyde must be handled in a fume hood and disposed of in a hazardous waste container. After disposal of the glutaraldehyde solution, chips were again washed with acetone and distilled water and then dried under a stream of N2. Chip surfaces were functionalized with a terminal aldehyde, thus making them susceptible to nucleophilic attack by the free amine of a probe molecule. Each chip was composed of four spots of each probe, tir-IBD (experimental) and ubiquitin (control). Probe spots were manually arrayed at concentrations of 500 µM and volumes of 1.5 µL and allowed to immobilize for 60 min in a humidity chamber. The remaining aldehydes were deactivated by immersing the chips in a solution of bovine serum albumin (BSA, BSA solution contains 100 µg/mL BSA, 20 mM HEPES, and 150 mM NaCl at pH 7.2) for 90 min. Apparatus. The detection apparatus (Figure 2) used for this study utilizes a linearly polarized 10-mW HeNe laser (Melles Griot) operating at a wavelength of 632.8 nm. The beam is then spatially filtered, expanded, and collimated to a diameter of 25 mm. The Analytical Chemistry, Vol. 78, No. 15, August 1, 2006
5579
Figure 3. Composition of the multilayer substrate used for simulated AIR data.
sample stage is a modified vacuum chuck to allow for vertical placement of the chip. Laser light reflected off of the chip is then captured by a Sony XCD-SX910 CCD camera. Experimental Procedure. Individual chips, previously derivitized and blocked as described above, were washed with distilled water, dried under N2, and imaged at various integration times. These images were catalogued as references for background subtraction, so that the intensity of each spot could be compared preand postapplication of protein solution. Next, 100 µL of intiminECD solutions, as well as a buffer control, were pipetted onto the surface and allowed to incubate for 60 min. The intimin-ECD solutions, ranging in concentration from 10 pM to 1 µM, were prepared by successive 1:10 dilutions from a common stock. Chips were next washed with distilled water, dried under a stream of N2, and imaged. Data Analysis. Acquired images were imported into the ImageJ software package.15 Each spot in the array was analyzed by determining the mean intensity from the curve fit to a normal distribution of the pixel intensity profile. A raw reflectivity difference for each spot was computed with respect to the background image by subtracting the mean intensities (∆R). ∆R intensities were normalized with respect to the buffer control. RESULTS AND DISCUSSION Theoretical Analysis. The reflectance of our structure (Figure 3) was simulated using a standard multilayer thin-film technique.16 Each layer was represented by a 2 × 2 element characteristic matrix that models the propagation of the electromagnetic field though the single film. The characteristic matrix includes information regarding the optical thickness and propagation angle for each layer. These matrixes are then multiplied in sequence to yield a single 2 × 2 matrix for the complete structure. The total reflectance, a function of the properties of every layer, can be directly calculated from the terms of this resultant matrix. Using this matrix analysis, we have the ability to purvey a theoretical representation of AIR by accounting for the distinct properties of the substrate and each bioorganic layer. By ellipsometry, the tirIBD, glutaraldehyde, and APTES layers were found to behave as a single layer with a refractive index of 1.50 (data not shown) and were modeled as such. This is expected for a layer that predominantly consists of protein, and it agrees with the reported literature value for a typical protein.17,18 (15) Abramoff, M. D.; Magelhaes, P. J.; Ram, S. J. Biophotonics Int. 2004, 11, 36-42. (16) Hecht, E.; Zajac, A. Optics, 1st ed.; Addison-Wesley: London, 1974. (17) Fontana, E.; Pantell, R. H.; Strober, S. Appl. Opt. 1990, 29, 4694-4704. (18) Tronin, A.; Edwards, A. M.; Wright, W. W.; Vanderkooi, J. M.; Blasie, J. K. Biophys. J. 2002, 82, 996-1003.
5580 Analytical Chemistry, Vol. 78, No. 15, August 1, 2006
Figure 4. Compilation of simulated reflectance spectra over the course of an experiment. From top to bottom: bare silicon dioxide on silicon, functionalized up to the tir-IBD probe layer, a 1-Å increase due to intimin-ECD binding, and a 10-Å increase due to intimin-ECD binding.
Fixing both the wavelength of light and the incident angle requires precise engineering of the original oxide layer thickness, as well as control over the deposition of our covalent linking chemistries and immobilization of the probe set. Using a slow HF etch, the starting silicon dioxide thickness can be tailored to 0.1 Å. Figure 4 depicts the simulated reflectivity of our substrate along subsequent steps in the experimental process. The black curve represents our starting oxide at a thickness of 1369.6 Å. As indicated by the red line, successive functionalization of the substrate up to the probe layer allows us to approach the reflectivity minimum of our system. In addition to linker molecules, this tir-IBD layer, which alone accounts for 23 Å, corresponds to an overall thickness increase of 38 Å. Finally, the green and blue traces model the spectrum shifts that result from binding 1 and 10 Å of our target intimin-ECD, respectively. Reflectance Measurements. While the zero reflectance condition inherently imparts a large dynamic range to AIR, it is important to determine the experimental operating range of the technique. To do so, a set of chips, comprising known thicknesses of SiO2 near the minimum reflectivity thickness, was created as a simple test system. The reflected intensity was compared to the reference beam intensity and calculated as the characteristic reflectance for the known thickness. In addition, the laser beam was analyzed to determine the purity of the polarization state. It was determined that, in a nonexpanded state, our system performs at ∼30 000:1 s- to p-polarization. Figure 5 shows the experimental data overlaid onto simulated spectra for both a pure s-polarization state and our experimentally determined polarization state. It can be seen that the data closely align with theory; however, the measured minimum is somewhat broader than desired. This may be caused by localized nonuniformities in the SiO2 that skew our ability to definitively assign a reflectance to a thickness. In the short term, the broad minimum will dictate creating chips slightly thicker than the minimum in order to avoid the range of nearequivalent minimal reflectance; in the long term, it should be possible to further improve sensitivity with additional refinement of the AIR substrate and measurement apparatus.
Figure 5. Measured vs simulated reflectance. Measured reflectance values from known chip thicknesses (open circles) are plotted against simulated reflectance spectra corresponding to conditions of 30 000: 1 s/p-polarization (line) and pure s-polarization (dash).
Experimental Analysis. Figure 6 shows a selection from a typical experimental image that focuses on consecutive spots of ubiquitin and tir-IBD. The reflectance change due to the addition of intimin-ECD is clearly seen between a background image (preintimin-ECD) and an exposed image (post-intimin-ECD), and that
change is further clarified in the associated profiles. Figure 7 depicts the change in reflectance for concentrations of intiminECD ranging from 10 pM to 1 µM for images acquired at 100 ms. All reflectance changes were normalized with respect to the difference observed in a set of three control chips. The control chips were treated identically to chips exposed with solutions containing intimin-ECD but received buffer only; they served to account for any bulk decrease in reflectance that occurs due to the displacement of material from the sensor surface by the small amount of surfactant contained in the buffer. Close examination of the lower end of the concentration curve shows that the error in measurement for 10 and 100 pM intimin-ECD overlap. While this precludes using the current concentration curve for intiminECD quantitation at low protein concentrations, or for calculating a precise limit of detection, the increased spot intensity produced by the 10 pM intimin-ECD solution relative to buffer control is highly reproducible. As future experiments are carried out on larger (machine-produced) arrays, the increased spot redundancy will allow for more effective statistical analysis. Moreover, enhancement of the measured reflectance is also possible by increasing the image acquisition time. This is another potential method of increasing the statistical significance of data close to the detection limit, at the expense of overall dynamic range and will be the subject of further study. Finally, the normalized reflectance changes for the ubiquitin negative control spots returned a baseline for all screened concentrations.
Figure 6. Background (A) and exposure (B) images depicting the detection of 100 nM intimin-ECD. In both images, the ubiquitin (control) probe is the left spot and the tir-IBD (experimental) probe is the right spot. Smoothed intensity profiles derived from these images (C) show the signal changes that occur following target binding.
Figure 7. Measured change in normalized reflectance as a function of intimin-ECD concentration. The left graph shows the results from the full range of concentrations. The right graph focuses on the lower limit of the experiment.
Analytical Chemistry, Vol. 78, No. 15, August 1, 2006
5581
Quantification of the Lower Limit. While the concentrationdependent sensor response is an important descriptor of performance, the mass/area (pg/mm2) sensitivity is a more suitable value to allow for comparison across sensor platforms.19 For the sensing system employed here to yield quantitative information, the analytical procedure must be approximated to follow solution kinetics. The correlation of planar to solution geometries has been studied previously as applied to SPR,20,21 and we have applied a similar approach to analyze our AIR data. The reported literature value of the dissociation constant for the intimin-ECD/tir-IBD binding interaction is 312.5 nM.14 Therefore, by detecting 10 pM for this system, we are measuring a thickness increase that corresponds to much less than a monolayer. To fully quantify the amount of bound intimin we need to know the concentration of immobilized tir-IBD on the surface of our substrate, which will be dependent upon the size of the array, the size of tir-IBD, and how densely the monolayer assembles. The crystal structure solved for tir-IBD (pdb: 1F02T) by Luo et al.14 contained the β-hairpin binding site and two flanking, ordered R helices. Our tir-IBD probe, on the other hand, is significantly larger. The N-terminus is extended with a six-histidine tag and a thrombin cleavage recognition site, products of the expression vector, which result in 21 added amino acids. Additionally, the C-terminus is extended with 26 amino acids contained in full-length tir. It is necessary to our understanding of how our probe immobilizes to dimensionally describe tir-IBD as a larger protein than what is presented in the crystal structure. Preliminary modeling studies of the larger tir-IBD molecule suggests, depending on the folding, flexibility, and orientation of the extra sequences, that the height could increase by more than 30 Å. While the reported sensitivity limit is of course dependent on any overestimation in probe size, the actual calculated sensitivity increase obtained by deviating from the reported crystal structure dimensions is not large. To model the surface of the substrate, we assume a tightly packed monolayer with each tir-IBD molecule represented by a rigid cylinder with a height of 70 Å and a radius of 20 Å. Taking the area each probe occupies and the area of the array based on spots with a 3-mm diameter, the maximum amount of immobilized tir-IBD is 847 fmol. The thickness of this modeled tir-IBD layer is 70 Å, but since we immobilize 23 Å, only 278 fmol is present. While the number of moles of immobilized tir-IBD is constant, a volume must be chosen to determine an equivalent solution concentration for kinetic evaluation. If it assumed that the experiment reaches equilibrium within the 60-min time frame of the experiment, then the volume containing the immobilized tir-IBD is 100 µL and the surface concentration is 2.78 nM. The amount of bound intimin-ECD was determined using the equation for the dissociation constant (Kd, eq 1).
Kd )
Kd )
[TirIBD,free][IntECD,free]
[bound] [TirIBD,S - bound][IntECD,S - bound] [bound]
Analytical Chemistry, Vol. 78, No. 15, August 1, 2006
When eq 1 is rearranged in terms of known quantities, such as the concentration of immobilized tir-IBD (TirIBD,S) and the concentration of intimin-ECD in solution (IntECD,S), the dissociation constant becomes a function of the concentration of bound intiminECD (eq 2). The amount bound at any screened concentration can then be solved by using the quadratic equation following a rearrangement of eq 2 to eq 3. This can be presented as a mass/ area by accounting for the experimental volume, the molecular mass of intimin-ECD (32.3 kDa), and the size of the tir-IBD spot. At the lower detection limit of 10 pM, eq 3 yields 0.04 pg/mm2 bound intimin-ECD. The equilibrium analysis returns a value that is remarkable but not plausible for the current working limitations of the technique. Due to the static nature of the experimental procedure, i.e., no introduction of flow or stirring to the analyte solution, it is probable that the system is severely diffusion limited. Therefore, in determining the immobilized tir-IBD concentration, one must not use the experimental analyte solution volume because this will both decrease the probe concentration and increase sensitivity; rather, the effective sensing volume (ESV) and the local sensor volume (LSV) must be used (Figure 8). The ESV is the limited analyte volume that will be accessible by diffusion to the sensor surface, regardless of a larger screened volume. It is determined by the dimensions of the tir-IBD probe spot as well as the diffusion length of intimin-ECD,
L ) x2Dt
(4)
where L is the diffusion length, D is the diffusion coefficient, and t is the time interval for the experiment. Using the diffusion coefficient for a typical protein22 of 5 × 10-7 cm2/s over a 60-min experiment, the diffusion length of intimin-ECD equals 0.6 mm. The concentration of immobilized tir-IBD in the ESV is 37.8 nM as determined using the volume of the ESV (eq 5), where R is the radius of the tir-IBD spot (1.5 mm) and L is the aforementioned diffusion length.
V ) πL
[
]
2L2 πRL + + R2 3 2
(5)
(1) The LSV, on the other hand, is defined as the volume located at
(2)
1 [bound] ) (Kd + TirIBD,S + IntECD,S) 2 1 (K + TirIBD,S + IntECD,S)2 - 4TirIBD,SIntECD,S (3) 2x d 5582
Figure 8. Schematic representing the volumes involved in the calculation for the quantification at the lower detection limit.
(19) Delouise, L. A.; Kuo, P. M.; Miller, B. L. Anal. Chem. 2005, 77, 32223230. (20) Day, Y. S. N.; Baird, C. L.; Rich, R. L.; Myszka, D. G. Protein Sci. 2002, 11, 1017-1025. (21) Karlsson, R.; Falt, A. J. Immunol. Methods 1997, 200, 121-133. (22) Beverung, C. J.; Radke, C. J.; Blanch, H. W. Biophys. Chem. 1999, 81, 59-80.
the sensor surface where at least 99% of the screened analyte will be bound, as dictated by the Kd of the binding interaction. It can be implicitly determined that this condition occurs 1.2 µm from the sensor surface, and thus, the associated immobilized tir-IBD concentration becomes 32.7 mM. Since there is a vast excess of probe within the LSV, dissociated analyte will be immediately rebound by a neighbor probe; therefore, we can consider interactions inside this volume to be irreversible. While the LSV is inherently small, the ESV acts as a reservoir for the sensor surface. Due to diffusion limits, molecules in the ESV can be considered to reach the surface while those outside cannot. Using this diffusion-limited model, the bound intimin-ECD that we detect at 10 pM is equivalent to 0.33 pg/mm2. While this is 1 order of magnitude above the equilibrium-only model, it is still a considerable level of sensitivity. The immobilized concentration assumes, however, that each molecule of tir-IBD bound to the surface is both viable and in an orientation capable of binding intimin-ECD. One must consider that neither may be true because the probe spot is washed and dried before analyte introduction, and there are five solventexposed lysines in tir that may react with the surface during immobilization (potentially resulting in an inactive molecule). Nevertheless, these assumptions allow us to conservatively estimate the sensitivity of the AIR technique at its current level of refinement. Once the transition to a solution system is made for the analysis, four critical estimates are used: the tir-IBD construct is structured and oriented as proposed, all tir-IBD is active, the sensing volume is accurately modeled, and 10 pM is a lower limit for our current detection capabilities. All estimates are made such that they handicap our sensitivity calculation for this particular system, so it is probable that AIR is ultimately more sensitive than presented. Even if we neglect the excess size of tir-IBD and rely solely on the crystal structure’s dimensions, the sensitivity limit does not appreciably change. It also remains to be seen how diffusion-limited this method is. If we are, indeed, inhibited by diffusion to the surface of our sensor, improvements in analyte
introduction coupled with continued refinement of the apparatus may provide further increases in sensitivity. CONCLUSIONS Techniques for the rapid, parallel, and label-free detection of proteins are vital to the advancement of proteomics, basic biology, and biomedical science. Through the use of a model system based on the bacterial proteins intimin-ECD and tir-IBD, we have demonstrated that AIR is an exceptionally sensitive, quantitative analytical technique for the submonolayer detection of biomolecules. The arrayable aspect of AIR inherently tracks it toward development as a high-throughput sensing platform. The 1-h analyte equilibration time used in the experiments described above is consistent with standard protocols for comparable methods of protein detection and only slightly hinders the number of parallel experiments that can be performed. Since the duration of the experiment is largely determined by the diffusion of the analyte and the degree of sensitivity desired, one could of course increase throughput by sacrificing some measure of sensitivity. Further experiments designed to test the applicability of AIR to several other problems in protein detection are currently underway in our laboratories. ACKNOWLEDGMENT We thank Professors Natalie Strynadka and Brett Finlay (University of British Columbia) for donation of materials. We thank Professor Lewis J. Rothberg (University of Rochester) and Dr. Scott R. Horner (Pathologics, LLC) for many helpful discussions. Portions of this work were supported by the Infotonics Technology Center, Inc., via a grant from the Department of Energy (DE-FG02-02ER63410.000), and by the NIH (R24-AL054953, via the Human Immunology Center at the University of Rochester). Received for review March 15, 2006. Accepted May 28, 2006. AC060473+
Analytical Chemistry, Vol. 78, No. 15, August 1, 2006
5583