Extracting kinetic rate constants from surface plasmon resonance array systems

Extracting kinetic rate constants from surface plasmon resonance array systems

Available online at www.sciencedirect.com ANALYTICAL BIOCHEMISTRY Analytical Biochemistry 373 (2008) 112–120 www.elsevier.com/locate/yabio Extractin...

2MB Sizes 0 Downloads 53 Views

Available online at www.sciencedirect.com

ANALYTICAL BIOCHEMISTRY Analytical Biochemistry 373 (2008) 112–120 www.elsevier.com/locate/yabio

Extracting kinetic rate constants from surface plasmon resonance array systems Rebecca L. Rich a, Michelle J. Cannon a, Jerry Jenkins b, Prabhakar Pandian b, Shankar Sundaram b, Rachelle Magyar c, Jennifer Brockman c, Jeremy Lambert c, David G. Myszka a,* a

Center for Biomolecular Interaction Analysis, School of Medicine, University of Utah, Salt Lake City, UT 84132, U.S.A. b CFD Research Corp., Huntsville, AL 35805, U.S.A. c HTS Biosystems, Hopkinton, MA 01748, U.S.A. Received 17 July 2007 Available online 19 August 2007

Abstract Surface plasmon resonance imaging systems, such as Flexchip from Biacore, are capable of monitoring hundreds of reaction spots simultaneously within a single flow cell. Interpreting the binding kinetics in a large-format flow cell presents a number of potential challenges, including accounting for mass transport effects and spot-to-spot sample depletion. We employed a combination of computer simulations and experimentation to characterize these effects across the spotted array and established that a simple two-compartment model may be used to accurately extract intrinsic rate constants from the array under mass transport-limited conditions. Using antibody systems, we demonstrate that the spot-to-spot variability in the binding kinetics was <9%. We also illustrate the advantage of globally fitting binding data from multiple spots within an array for a system that is mass transport limited.  2007 Elsevier Inc. All rights reserved. Keywords: Biacore; Protein–protein interaction; Affinity; Flexchip

Surface plasmon resonance (SPR)1 optical biosensors monitor binding events in real time and without reactant labeling. This technology has revolutionized the way in which we characterize biomolecular interactions. Commonly employed SPR instruments (e.g., Biacore 3000 and T100 platforms) monitor at most four interactions simultaneously. While these instruments are widely used [1], next– generation SPR array-based technology such as Biacore’s Flexchip, Genoptics’ SPRi-Plex array system, Plexera’s ProteomicProcessor, GWC’s SPRimagerII, and Graffinity’s SPR Imaging Technology offer the ability to measure reactions in a multiplexed format. *

Corresponding author. Fax: +1 (801) 585 3015. E-mail address: [email protected] (D.G. Myszka). 1 Abbreviations used: BSA, bovine serum albumin; PBS, phosphatebuffered saline; ROI, region of interest; SPR, surface plasmon resonance; 3D, three-dimensional. 0003-2697/$ - see front matter  2007 Elsevier Inc. All rights reserved. doi:10.1016/j.ab.2007.08.017

Here we focus on the Flexchip system but our conclusions may be applied to any large-format flow cell SPR array-based technology. The Flexchip incorporates advances in hardware components, making it possible to monitor analyte (P5000Da) binding up to 400 ligands at one time. Flexchip uses grating-coupled SPR to monitor refractive index changes across the target area (1 cm2) as analyte is flowed across a matrix of ligand spots within the single large flow cell (see Fig. 1). Additional novel Flexchip features include using unmodified areas between the ligand spots for referencing and recirculating analyte through the flow cell to collect association-phase data for extended periods of time (1 to 2 h). Since its commercial release, a few researchers have described Flexchip’s utility in higher-throughput biosensor-based screening applications in epitope mapping and protein expression profiling [2–6]. These groups primarily used the Flexchip in a qualitative format, obtaining

Extracting kinetics from SPR array systems / R.L. Rich et al. / Anal. Biochem. 373 (2008) 112–120

113

Fig. 1. Ligand spotting and flow cell assembly. (A) Flexchip flow cell. A gasket with a hexagonal cutout, sandwiched between the gold surface and the glass window, forms the flow cell walls. Analyte enters through the portal on the left, flows across the ligand spots, and exits through the portal on the right. In this example, a 20 · 20 matrix of protein spots was prepared by solid-pin deposition within 1 cm2 of an underivatized target slide. (B) Alternate view of a slide in which the prism effect from the grating molded into the sensor surface is apparent.

single-point measurements for analytes binding to large collections of antibodies, peptides, or protein-binding partners. As the first approach to fully recognizing Flexchip’s potential for monitoring interactions in real time, Sexton and co-workers [7] recently obtained kinetic and affinity parameters for Fabs screened against human tissue kallikrein. Here we show how Flexchip can be used in a highthroughput, high-resolution format to determine robust kinetic rate constants from each spot within a ligand array. We outline the particularly novel features of this next-generation SPR biosensor and describe the approach we have taken to establish the reliability of this instrument. We used full 3D computer models to characterize mass transport properties of the large-format flow cell and developed the data analysis tools that make it possible to extract reliable kinetic constants from arrays. We also established the Flexchip’s selectivity, sensitivity, and reproducibility (both spot-to-spot and run-to-run) with two antibody/antigen model systems.

The modeling software package CFD-ACE+ allowed us to simulate reactions under different conditions to elucidate the effects of flow cell height, flow rate, binding constants, and grid position on the binding responses associated with the array biosensor. During the simulation the amount of bound analyte was output at each time step, with each ligand patch having a separate output file. The simulation protocol was specified as follows: The analytes were flowed in for 10 min (the association phase). The wash step (dissociation phase) began immediately after the association phase, and buffer was flowed in for an additional 10 min. The conditions under which simulations were performed include four analyte inlet concentrations (80.0, 26.7, 9.88, and 2.96 nM), two flow rates (1000 and 100 lL/min), two kas (1 · 105 and 1 · 107 M 1 s 1), two surface capacities (667 and 140,000 moles/lm2), a dissociation rates constant of 0.001 s 1, and an analyte diffusion coefficient of 693.8 lm2/s.

Materials and methods

Protein A/G spotting

Instruments and reagents

For the spot array of uniform densities, Protein A/G (50 lg/mL in 0.1· PBS (1· = 1.9 mM NaH2PO4, 8.1 mM Na2HPO4, 150 mM NaCl, pH 7.4)) was spotted directly using solid pins onto a plain gold-coated target slide and allowed to air dry. For the spot array of different Protein A/G densities, stock solutions of 0, 3.3, 6.7, 12.5, 25, and 50 lg/mL protein in 0.1· PBS were spotted onto a plain gold-coated target slide and allowed to air dry. After the flow cell was assembled and installed in the Flexchip instrument, the slide surface was blocked for 5 min with 1· PBS supplemented with 5 mg/mL BSA to minimize spot spreading and then filled with running buffer (1· PBS, pH 7.4, 0.2 mg/mL BSA).

The Flexchip instrument, sensor slides, gaskets, and windows were supplied by Biacore AB (Uppsala, Sweden) and the QArrayMini microarrayer was from Genetix (Boston, MA). CFD-ACE+ advanced multiphysics software was from ESI Group (Huntsville, AL). Purified rat myeloma IgG2b (Zymed Catalog No. 02-9288), purified recombinant Protein A/G (Cat. No. 21186), and human IgG1j (Cat. No. I5154) were purchased from Invitrogen, Pierce Biotechnology, Inc., and Sigma Chemical Co., respectively. General laboratory reagents and supplies were purchased from Fisher Scientific.

Modeling

114

Extracting kinetics from SPR array systems / R.L. Rich et al. / Anal. Biochem. 373 (2008) 112–120

IgG binding studies All binding studies were performed at 25 C using 1· PBS, pH 7.4, 0.2 mg/mL BSA as the running buffer. Before each analysis, the flow cell was filled with running buffer and ‘‘bubble-blasted’’ at least twice to ensure that it was free of air bubbles. Mock binding analyses using running buffer as the analyte were performed until the baseline signal stabilized and was reproducible (the final blank run was used for referencing during data processing). Solutions of rat IgG (5–200 nM) were prepared in running buffer and tested in duplicate for binding to the 12 · 13 matrix of uniform Protein A/G spots. Solutions of human IgG (3–100 nM) were prepared in running buffer and tested in triplicate for binding to the 156 matrix of different-density Protein A/G spots. Each antibody solution was flowed across the Protein A/G spots for 10 min at a flow rate of 200 lL/min and antibody dissociation from the surface was monitored for another 10 min at the same flow rate. At the end of each binding cycle, the Protein A/G surface was regenerated with a short injection of 20 mM H3PO4. Data processing and fitting All data processing and fitting were carried out using Scrubber2 (BioLogic Software, Pty., Australia). The binding responses were initially referenced by subtracting the signal over reference ROIs located adjacent to each protein A/G spot. The signal of a buffer injection over the Protein A/G spots was then subtracted from the single-referenced data of IgG binding to each spot. The resulting responses were then zeroed using a 30-s window prior to the IgG injection. The data sets of increasing concentrations of IgG over the same spot were then assembled together. Data were fit to a 1:1 interaction model (either A + B = AB or Ao = A + B = AB). Results and discussion Ligand immobilization We adapted a commercially available solid pin spotting technology used to produce DNA arrays to make protein chips for the Flexchip. An example of protein spotted in a 20 · 20 matrix on the Flexchip sensor surface is shown in Fig. 1A. The sensor surface is a plastic slide with an embedded optical grating and is coated with a thin (80 nm) gold layer for generating plasmon waves (Fig. 1B). In this study, we adsorbed Protein A/G onto this gold surface using solid-pin spotting; the ability to deposit protein in discrete locations is apparent in the separation of spots visible in Fig. 1. Flow cell design and sample delivery The Flexchip flow cell is composed of three parts: the sensor surface, an adhesive gasket (whose thickness deter-

mines the flow cell height), and a transparent flow cell window. The gasket and window are manually affixed to the surface to make a single large-format flow cell having one inlet and outlet (Fig. 1A). Analyte and buffer solutions can be flowed across the ligand surface at rates of 100 to 1500 lL/min with a minimum sample requirement of 1.6 mL. While fast flow rates decrease mass transport effects, they of course increase sample consumption. To counter this, the Flexchip’s fluidic arrangement allows analyte to be circulated repeatedly through the flow cell. This allows one to conserve material and, if necessary, to collect association phase data for hours (rather than minutes, which is a limitation in other biosensor platforms). This feature is particularly advantageous for examining high-affinity interaction that take a long time to reach equilibrium and/or collecting measurable binding signals from extremely dilute analyte solutions. Signal referencing and detection Fig. 2A shows a 12 · 13 array of protein spots and reference areas within the Flexchip flow cell. The white circles represent reaction regions of interest (ROIs). Reaction ROIs are areas where a specific ligand has been immobilized. In between every reaction ROI is also a blank area on the chip (reference ROI) that can be monitored to correct for bulk refractive index changes, nonspecific binding, and instrument drift (similar to using a reference flow cell in Biacore 3000 or T100). The number, position, and size of the reaction and reference ROIs are determined by the user, thereby permitting flexibility in spot patterning on the sensor slide. The array’s detection system is sufficiently selective and sensitive to differentiate between the binding response of the reaction ROI and the background response of neighboring reference ROIs. The Flexchip uses grating-coupled SPR to monitor refractive index changes across the target area. Compared to the optical system used in the traditional Biacore platforms (in which a wedge of light at different angles is refracted through a prism and onto a metal surface to generate a signal [8]), the Flexchip uses a graded metal surface to generate the surface plasmon wave (Fig. 2B). One advantage of this approach is that there is no need to match the refractive index of the prism with the gold substrate, making it easier to dock sensor chips. The chargecoupled device camera of this two-dimensional SPR detector measures the intensity of light refracted from the entire chip surface at different angles. Modeling flexchip flow cell The CFD-ACE+ modeling software allows us to simulate reactions with any set of rate constants and array pattern in the flow cell. For example, Fig. 3A illustrates the geometry and setup of the flow cell created using an automated scripted program that we developed for the Flexchip

Extracting kinetics from SPR array systems / R.L. Rich et al. / Anal. Biochem. 373 (2008) 112–120

115

Fig. 2. Sample referencing and detection. (A) Charge-coupled device camera shot of the region of interest (ROI) map for a 12 · 13 matrix of protein A/G spots. Protein spot sizes are 200 lm in diameter with a 400 lm center-to-center separation. Reaction ROIs are marked by white circles that surround the protein spots. The smaller (120 lm) darker circles represent reference spots which have been centered between the reaction spots. Although the reference ROIs are not actually marked on the slide surface, in this image they are shown as dark circles between reaction spots. One row each of reaction and reference ROIs are highlighted in red and blue, respectively. (B) Schematic of the detection system in Flexchip. The detector monitors the change in refracted light intensity across the flow cell surface as the analyte binds to, and dissociates from, the array of immobilized ligand spots.

flow cell. The flow cell is approximately 13 mm wide, 0.5 mm high, and 20 mm long. Inlet and outlet sections have a diameter of 1 mm each. A 12 · 12 array of receptor spots (144 total) was specified at the flow cell bottom. A numbering scheme consistent with microarray formats is adopted and is as follows: rows are indicated by alphabet (top to bottom) and columns are indicated by a number (left to right). For example, C05 would be a spot located in row 3 and column 5. During the simulation the amount of bound analyte was output at each time step, with each ligand patch having a separate output file. For simulation, a 3D mesh with approximately 130,000 cells was created for different flow cell designs (Fig. 3B shows an example of a flow cell with a notch in the window). Figs. 3C–3E show analyte migrating through the a flow cell, mapping the flow velocities and analyte concentration gradients in solution and at the sensor surface. Reaction- and transport-limited binding events We applied our 3D modeling capabilities to explore the limits of association rates obtainable with the Flexchip flow cell configuration. Using CFD-ACE+ and the 3D model of the array-based flow cell shown in Fig. 3A, we simulated binding reactions under two conditions: reaction limited and mass transport limited. Fig. 4 depicts response data sets simulated under the conditions where the binding responses are limited by the rate of the reaction (having a slow association rate constant of 1 · 105 M 1 s 1) and the other conditions where the responses are limited by the transport rate of analyte to the surface (here the association rate is modeled at 1 · 107 M 1 s 1). Note that the

binding responses appear curved (they are in fact exponentials) when complex formation is not mass transport limited (Fig. 4A) and linear under mass transport-limited conditions (Fig. 4B). The simulated array data were fit to test the ability of our data extraction program to accurately recover the values used in the simulation. Fig. 4A illustrates a fit to 12 data sets collected along column 5 within the sample grid. All of these data sets fit very well to a simple 1:1 interaction model. The association and dissociation rate constants determined from analysis of the data match the values used in the simulation. These results confirm that the scripting program is functioning properly and show that our data fitting program (Scrubber2) is working correctly. We extended our analysis to test how well our simple two-compartment model for mass transport [9] would interpret array data that were highly influenced by mass transport. Fig. 4B shows that this model is capable of fitting data simulated with very fast association rates. We can tell by visually inspecting the data that these reactions are mass transport limited because the binding profiles appear essentially linear. If we apply a simple 1:1 (A + B = AB) model to these data sets, the model fails to fit the data and the rate constants do not match the values used in the simulation. If, however, we include one more parameter to the model, which describes transport of the analyte from bulk flow down to the sensor surface (Ao = A + B = AB), then the model fits the data (Fig. 4B) and the returned rate constants match the simulated values. These results confirm that it is possible to extract accurate values for the intrinsic reaction rate constants even though the binding responses are heavily influenced by mass transport.

116

Extracting kinetics from SPR array systems / R.L. Rich et al. / Anal. Biochem. 373 (2008) 112–120

Fig. 3. Design and analysis of Flexchip flow cell created in CFD-ACE+ software. (A) Model flow cell created by CFD-ACE+ scripting program. The text in the figure illustrates the parameters that can be input by the user and the convention of naming the individual ligand spots. (B) Wire frame image. (C) Analyte velocity profiles. (D) Analyte solution gradients. (E) Analyte surface gradients within detection region.

Fig. 4. Global analysis of reaction- and transport-limited data. (A) A simple A + B = AB model (red lines) fit the array data (black lines) simulated under non-mass-transport-limited conditions (105 M 1 s 1) (black lines). (B) A two-compartment model for mass transport Ao = A + B = AB (red lines) fit the array data (black lines) simulated under mass-transport-limited conditions (107 M 1 s 1). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Extracting kinetics from SPR array systems / R.L. Rich et al. / Anal. Biochem. 373 (2008) 112–120

Fig. 5B shows the apparent transport coefficient in relationship to spot position within column 5 of the array. Again we see an unusual effect: the transport rate is slowest in the center of the flow cell, increases at the midpoint toward the wall, and then decreases near the walls. Again, this behavior is attributed to the effects of radial dispersion of the concentration profile. Our findings demonstrate that the difference in transport rate can be as great at 10% depending on the spot positions. These results illustrate that a separate transport rate should be used to describe the reaction at each position that is along the path of the flow and in the cross direction. However, given the twofold symmetry with regard to the cross flow, it could be possible to globally fit data from spots that have mirror equivalence.

Variability of transport coefficients across the sensor surface To evaluate the degree to which the transport rate is dependent on position within the spot matrix, we floated a different transport coefficient for each reaction spot within the array grid. The plot in Fig. 5A illustrates the value of the transport rate verses its position within the array. If we focus on the data for row A, which is nearest to the wall, we see that, as expected, the mass transport rate (km) decreases as we move down the field, until around position A10 when the transport rate bottoms out and then actually increases at the back of the field. A similar trend in transport rate is observed for row D, but as we would expect, the overall transport rates are higher for row D than for row A because it is nearer to the entrance of the flow cell. What was unexpected was the observation that the transport rates measured at many of the positions for row F (in the center of the flow cell) were actually lower than the transport rates observed for row A near the far wall. The radial velocity diagram in Fig. 5A shows effective concentration gradients caused by radial dispersion introduced by the shape of the flow cell. Analyte is essentially depleted from the center of the flow chamber as it diffuses toward the outer walls. At the back of the flow cell, analyte is effectively concentrated as it exits the flow cell.

A

117

Reproducibility of kinetic rate constants across the flow cell surface To establish the overall reliability of the responses obtained across the entire sensor slide surface with a real system, we tested the binding of a concentration series of rat IgG to a 12 · 13 matrix of Protein A/G spots. Fig. 6 shows the responses obtained for this interaction. An expanded view of data collected from one spot (shown in

Apparent Transport rate km

2.25 x106 2.20 x106 Inlet

Outlet

2.15 x106 2.10 x106

D

2.05 x106

A 2.00 x106

F

1.95 x106 0

2

4

6

8

10

12

14

Radial velocity

Column Position

Row Position

B

Column 5

A B C D E F G H I J K L 2.13 x106

Inlet

2.12 x106

2.11 x106

2.10 x106

2.09 x106

2.08 x106

Outlet

Radial velocity

Apparent Transport rate km Fig. 5. (A) Left: calculated transport coefficients plotted against target ROI position along the direction of flow. Right: 3D rendering of the Flexchip flow cell with positions of rows A, D, F, highlighted (top) and a radial velocity plot (bottom) illustrating analyte concentration gradients within the flow cell due to convection corresponding to change in chamber cross section. (bottom). (B) Left: calculated transport coefficients plotted against the position in column 5 of the spotted array. Right: row positions highlighted in column 5 (top) and the corresponding radial velocity plot (bottom).

118

Extracting kinetics from SPR array systems / R.L. Rich et al. / Anal. Biochem. 373 (2008) 112–120

Fig. 6A) demonstrates that the responses were concentration dependent, reproducible, and well described by a simple interaction model. No transport step was required because the association rate of rat IgG was too slow to be mass transport limited under these conditions. Including a step for mass transport does not improve the quality of the fit nor does it change the values of the binding constants or their standard errors. Fig. 6B shows that similar binding responses were observed at all the reaction spots within the array. In fact, the average association and dissociation rates for the 156 spots were (1.68 ± 0.2) · 105 M 1 s 1 and (1.57 ± 0.1) · 10 3 s 1, respectively. This represents a standard error in the kinetics between spots of <9% and confirms that the Flexchip can provide consistent kinetics for binding interactions across the flow cell surface. It is important to note that there was no trend in the variability of the kinetic rate constants. In other words, we did not for instance observe a higher intrinsic association rate for the spots closer to the flow cell entrance. The var-

iability in the rate constants appeared random throughout the array. Surface-density dependence of analyte binding responses To test Flexchip’s limits in sensitivity and reproducibility, a human IgG sample was injected over a concentration range of 3–100 nM over an array of Protein A/G spots that was generated with a gradient density of spots prepared from 3.3 to 50 lg/mL protein solutions. Fig. 7 shows the binding profiles from six of the different density protein spots. As expected, the binding responses increase in intensity as the capacity of the spot is increased. Each IgG concentration was injected three times and the overlay of these replicates is shown in Fig. 7. It is in fact difficult to resolve the independent injections because the binding responses within a given spot are very reproducible. Importantly, no binding of IgG was observed at a reference spot (Fig. 7, lower-right panel) that was located

Fig. 6. Kinetic analysis of rat IgG binding to Protein A/G. (A) Duplicate responses (black lines) for 5–200 nM IgG obtained from one spot (position D8). Red lines represent a global fit of the data set to a simple 1:1 interaction model. (B) Responses and fits for IgG binding to the array of 156 spots. Panel B reproduced from reference 10 with permission from RJ Communications & Media Ltd. 2003.

Fig. 7. Triplicate responses for human IgG1 (100, 50, 25, 12.5, 6, and 3 nM) binding to spots of different Protein A/G surface density (prepared from 3.3, 6.7, 12.5, 25, 50, and 0 lg/mL stock solutions).

Extracting kinetics from SPR array systems / R.L. Rich et al. / Anal. Biochem. 373 (2008) 112–120

119

Fig. 8. Global fits (red lines) of human IgG binding responses (black lines) from spots of different Protein A/G surface densities (low, medium, and high) to a 1:1 interaction model that includes a step for mass transport. Each IgG concentration was tested three times. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

immediately downstream from the highest-density spot of Protein A/G. This illustrates that no spot spreading occurred. We found that the lower limit on spotting density is 2 lg/mL to obtain interpretable analyte binding responses, although this limit of course depends partially on the spotting method used and size of the spotted area. In addition, the lower limit of analyte detection in this example is 1 nM; however, this limit will also depend on surface density and the mass of the analyte. To determine binding kinetics for the IgG/Protein A/G interaction, response data were fit to a 1:1 interaction model that included a step for mass transport. In this case, the kinetic values were globally fit parameters with an independent parameter for the transport rate for each spot. Fig. 8 shows an overlay of the best-fit model to data from three surfaces densities. The excellent fit of this model to the data again helps validate the kinetics obtained from Flexchip. The association and dissociation rate constants were calculated to be ka = 4.26 · 105 M 1 s 1 (± 0.3%) and kd = 4.85 · 10 5 s 1 (±2.5%), respectively. The very low standard errors for the parameter values stem from the fact that global analysis incorporates a large amount of information during the fitting process. The rate constants yield an equilibrium dissociation constant KD = 113 pM (±2.5%). It is perhaps surprising that a 1:1 interaction model could describe the response data for a mixture of IgG interacting with a Protein A/G surface. It is likely that if multisite complexes are formed within the IgG (either through the two available sites within the constant region or through interactions with the Fab domain or with Protein A and Protein G) these events occur on a time scale that we cannot resolve with SPR. We essentially see the overall interaction of the IgG with the sensor surface. It is the sum of these interactions that yields the high affinity. It may be possible

in the future to study individual domains of the IgG interacting with pure Protein A and Protein G surfaces to resolve how much each component contributes to the interaction. Given the minimum sample requirements of 1.6 mL for the Flexchip instrument and the binding constants for the IgGProtein A/G interaction, a complete binding profile for this system (assuming an IgG concentration of 1 nM) could be collected with as little as 240 ng of the IgG.

Conclusion The interest in high-throughput screening systems and the proteomics revolution is driving the development of higher-capacity SPR biosensor technology [10]. Our modeling and experimental analyses demonstrate that SPR array-based systems can yield reliable rate constants. A two-compartment model can accurately describe the intrinsic binding kinetics for mass transport-limited reactions within an array flow cell. Our simulations revealed that the measured transport rate (km) varies based on the spot position within the flow cell because of analyte dispersion. Therefore, when analyzing kinetic data obtained from array-based flow cells, it is necessary to fit km independently at each spot position. The spot-tospot variation in the binding kinetics observed within the Flexchip was <9%. Data from spots with varying surface density can be globally fit to provide more information about the binding constants, especially for mass transport-limited reactions. In summary, the ability to extract kinetic information from hundreds of interactions at one time dramatically expands the potential impact of optical biosensor technology in both basic and applied research fields.

120

Extracting kinetics from SPR array systems / R.L. Rich et al. / Anal. Biochem. 373 (2008) 112–120

Acknowledgment This work was funded in part by NSF Grant EF0427665 (awarded to D.G.M).

[5]

References [6] [1] R.L. Rich, D.G. Myszka, Survey of the year 2005 commercial optical biosensor literature, J. Mol. Recognit. 19 (2006) 478–534. [2] G.J. Carven, S. Chitta, I. Hilgert, M.M. Rushe, R.F. Baggio, M. Palmer, J.E. Arenas, J.L. Strominger, V. Horejsi, L. Santambrogio, L.J. Stern, Monoclonal antibodies specific for the empty conformation of HLA-DR1 reveal aspects of the conformational change associated with peptide binding, J. Biol. Chem. 279 (2004) 16561– 16570. [3] R. Baggio, G.J. Carven, A. Chiulli, M. Palmer, L.J. Stern, J.E. Arenas, Induced fit of an epitope peptide to a monoclonal antibody probed with a novel parallel surface plasmon resonance assay, J. Biol. Chem. 280 (2005) 4188–4194. [4] R.M. Hoet, E.H. Cohen, R.B. Kent, K. Rookey, S. Schoonbroodt, S. Hogan, L. Rem, N. Frans, M. Daukandt, H. Pieters, R. van Hegelsom, N.C. Neer, H.G. Nastri, I.J. Rondon, J.A. Leeds, S.E. Hufton, L. Huang, I. Kashin, M. Devlin, G. Kuang, M. Steukers, M.

[7]

[8] [9]

[10]

Viswanathan, A.E. Nixon, D.J. Sexton, H.R. Hoogenboom, R.C. Ladner, Generation of high-affinity human antibodies by combining donor-derived and synthetic complementarity-determining-region diversity, Nat. Biotechnol. 23 (2005) 344–348. D.W. Unfricht, S.L. Colpitts, S.M. Fernandez, M.A. Lynes, Gratingcoupled surface plasmon resonance: a cell and protein microarray platform, Proteomics 5 (2005) 4432–4442. K. Usui-Aoki, K. Shimada, M. Nagano, M. Kawai, H. Koga, A novel approach to protein expression profiling using antibody microarrays combined with surface plasmon resonance technology, Proteomics 5 (2005) 2396–2401. D. Wassaf, G. Kuang, K. Kopacz, Q.-L. Wu, Q. Nguyen, M. Toews, J. Cosic, J. Jacques, S. Wiltshire, J. Lambert, C.C. Pazmany, S. Hogan, R.C. Ladner, A.E. Nixon, D.J. Sexton, High-thoughput affinity ranking of antibodies using surface plasmon resonance microarrays, Anal. Biochem. 351 (2006) 241–253. R.L. Rich, D.G. Myszka, Why you should be using more SPR biosensor technology, Drug Discov. Today Technol. (2004) 301–308. D.G. Myszka, X. He, M. Dembo, T.A. Morton, B. Goldstein, Extending the range of rate constants available from BIACORE: interpreting mass transport-influenced binding data, Biophys. J. 75 (1998) 583–594. R.L. Rich, D.G. Myszka, Higher-throughput, label-free, real-time molecular interaction analysis, Anal. Biochem. 361 (2007) 1–6.