Near-Simultaneous and Real-Time Detection of ... - ACS Publications

Sep 28, 2004 - Cancer Center and Department of Pathology, University of New Mexico School of Medicine, Department of Chemical and. Nuclear Engineering...
0 downloads 0 Views 170KB Size
Download PDF
Anal. Chem. 2004, 76, 6266-6273

Near-Simultaneous and Real-Time Detection of Multiple Analytes in Affinity Microcolumns Menake E. Piyasena,† Tione Buranda,*,‡ Yang Wu,§ Jinman Huang,§ Larry A. Sklar,*,‡,§ and Gabriel P. Lopez*,†,§

Cancer Center and Department of Pathology, University of New Mexico School of Medicine, Department of Chemical and Nuclear Engineering, and Department of Chemistry, University of New Mexico, Albuquerque, New Mexico 87131

A miniaturized immunoassay system based on beads in poly(dimethylsiloxane) microchannels for analyzing multiple analytes has been developed. The method involves real-time detection of soluble molecules binding to receptor-bearing microspheres, sequestered in affinity column format inside a microfluidic channel. Identification and quantitation of analytes occurs via direct fluorescence measurements or fluorescence resonance energy transfer. A preliminary account of this work based on single-analyte format has been published in this journal (Buranda, T.; Huang, J.; Perez-Luna, V. H.; Schreyer, B.; Sklar, L. A.; Lopez, G. P. Anal. Chem. 2002, 74, 1149-1156). We have extended the work to a multianalyte model system composed of discrete segments of beads that bear distinct receptors. Near-simultaneous and real-time detection of diverse analytes is demonstrated. The importance of this work is established in the exploration of important factors related to the design, assessment, and utility of affinity microcolumn sensors. First, beads derivatized with surface chemistry suitable for the attachment of fluorescently labeled biomolecules of interest are prepared and characterized in terms of functionality and receptor site densities by flow cytometry. Second, calibrated beads are incorporated in microfluidic channels. The analytical device that emerges replicates the basic elements of affinity chromatography with the advantages of microscale and real-time direct measurement of bound analyte on beads rather than the indirect determination from eluted sample typical of affinity chromatography. In addition, the two-compartment analysis of the assay data as demonstrated in single-analyte columns provides a template upon which the dynamics of multiple-analyte assays can be characterized using existing theoretical models and be tested experimentally. The assay can potentially detect subfemtomole quantities of protein with high signal-tonoise ratio and a large dynamic range spanning nearly 4 orders of magnitude in analyte concentration in microliter to submicroliter volumes of analyte fluid. The approach * To whom correspondence should be addressed. E-mail: buranda@ unm.edu; [email protected]; [email protected]. † Department of Chemistry. ‡ Cancer Center and Department of Pathology, University of New Mexico School of Medicine. § Department of Chemical and Nuclear Engineering.

6266 Analytical Chemistry, Vol. 76, No. 21, November 1, 2004

has the potential to be generalized to a host of bioaffinity assay methods including analysis of protein complexes (e.g., biomolecular indicators of deseases). Proof-ofprinciple analytes include FLAG peptide and carcinoembryonic antigen detected at physiologically relevant concentration levels. Assays derived from the molecular recognition interactions of biological molecules have been the mainstay of many clinical, biochemical, and environmental research efforts.1-13 A variety of approaches have been developed for performing immunoassays in single as well as multianalyte format.8,14-22 Besides the conventional sandwich, direct binding, and competition-based assays, a number of approaches such as flow injection analysis21,23-32 and those based on small-volume microfluidic devices have emerged33-41 Common themes in the design and optimal use of these microanalytical devices have emerged:42 (a) The presence of beads (1) Bangs, L. B. Pure Appl. Chem. 1996, 68, 1873-1879. (2) Hage, D. S. J. Clin. Ligand Assay 1997, 20, 293-301. (3) Hage, D. S. Anal. Chem. 1999, 71, R294-R304. (4) Hage, D. S.; Nelson, M. A. Anal. Chem. 2001, 73, 198A-205A. (5) Holt, D.; Rabbany, S. Y.; Kusterbeck, A. W.; Ligler, F. S. Rev. Anal. Chem. 1999, 18, 107-132. (6) Jones, G.; Wortberg, M.; Rocke, D. M.; Hammock, B. D. ACS Symp. Ser. 1997, 657, 331-342. (7) Ohmura, N.; Lackie, S. J.; Saiki, H. Anal. Chem. 2001, 73, 3392-3399. (8) Phillips, T. M. J. Biochem. Biophys. Methods 2001, 49, 253-262. (9) Rabbany, S. Y.; Lane, W. J.; Marganski, W. A.; Kusterbeck, A. W.; Ligler, F. S. J. Immunol. Methods 2000, 1, 1-2. (10) Sapsford, K. E.; Charles, P. T.; Patterson, C. H.; Ligler, F. S. Anal. Chem. 2002, 74, 1061-1068. (11) Sato, K.; Tokeshi, M.; Kimura, H.; Kitamori, T. Anal. Chem. 2001, 73, 12131218. (12) Solassol, I.; Granier, C.; Pelegrin, A. Tumor Biol. 2001, 22, 184-190. (13) Swartzman, E. E.; Miraglia, S. J.; MellentinMichelotti, J.; Evangelista, L.; Yuan, P. M. Anal. Biochem. 1999, 271, 143-151. (14) Albert, K. J.; Lewis, N. S.; Schauer, C. L.; Sotzing, G. A.; Stitzel, S. E.; Vaid, T. P.; Walt, D. R. Chem. Rev. 2000, 100, 2595-2626. (15) Albert, K. J.; Walt, D. R.; Gill, D. S.; Pearce, T. C. Anal. Chem. 2001, 73, 2501-2508. (16) Crabtree, H. J.; Cheong, E. C. S.; Tilroe, D. A.; Backhouse, C. J. Anal. Chem. 2001, 73, 4079-4086. (17) Dickinson, T. A.; Michael, K. L.; Kauer, J. S.; Walt, D. R. Anal. Chem. 1999, 71, 2192-2198. (18) Dolan, P. L.; Wu, Y.; Ista, L. K.; Metzenberg, R. L.; Nelson, M. A.; Lopez, G. P. Nucleic Acids Res. 2001, 29, U37-U44. (19) Fang, Y.; Frutos, A. G.; Lahiri, J. J. Am. Chem. Soc. 2002, 124, 2394-2395. (20) Figeys, D.; Pinto, D. Electrophoresis 2001, 22, 208-216. (21) Li, J. J.; Tremblay, T. L.; Thibault, P.; Wang, C.; Attiya, S.; Harrison, D. J. Eur. J. Mass Spectrom. 2001, 7, 143-155. (22) Walt, D. R. Science 2000, 287, 451-452. 10.1021/ac049260f CCC: $27.50

© 2004 American Chemical Society Published on Web 09/28/2004

in microchannels provides a high surface area-to-volume ratio. As a result, the analyte diffusion distances are reduced in magnitude to those equivalent to the interstitial space between the beads. (b) The site coverage of the beads can be optimized for an assay outside the channel, with the potential for prior quantitation by flow cytometry or spectrofluorometry.41,43,44 (c) The total number of binding sites within the detection scheme should be optimized to match the needs of detection limits and sensitivity.7 (d) The volumetric flow rates must be limited by the kinetic restrictions of the immunoreaction. (e) For closely packed beds, the flow rate may typically be limiting due to pressure loss, depending on the size of beads and length of bed. The flow-through immunoaffinity microsystems, such as the one described here, are governed by the same basic principles.42,45 However, important differences in design emphasis have been presented, where such variations are manifested in the performance characteristics of the devices. Because of the nascent nature of these approaches, the necessary scaling laws have not yet been fully developed to optimize their universal working characteristics. The crucial factor from which the microsystems derive their significant operational advantage over the conventional titer plate assay is derived from the “specific interface”.46 Specific interface is defined as the ratio between substrate surface area and the fluid volume in the microchannel space occupied by the beads. The density of available binding sites in the detection volume is referred to as the “reaction field”.46 Taken together, the magnitudes of the specific interface and reaction field in microfluidic devices can limit the reaction time to minutes or (23) AbdelHamid, I.; Ivnitski, D.; Atanasov, P.; Wilkins, E. Anal. Chim. Acta 1999, 399, 99-108. (24) Bruin, G. J. M. Electrophoresis 2000, 21, 3931-3951. (25) Hage, D. S.; Thomas, D. H.; Chowdhuri, A. R.; Clarke, W. Anal. Chem. 1999, 71, 2965-2975. (26) Hodder, P. S.; Ruzicka, J. Anal. Chem. 1999, 71, 1160-1166. (27) Howard, M. E.; Holcombe, J. A. Anal. Chem. 2000, 72, 3927-3933. (28) Lahdesmaki, I.; Ruzicka, J.; Ivaska, A. Analyst 2000, 125, 1889-1895. (29) Ruzicka, J. Analyst 1998, 123, 1617-1623. (30) Ruzicka, J.; Scampavia, L. Anal. Chem. 1999, 71, A257-A263. (31) Ruzicka, J. Analyst 2000, 125, 1053-1060. (32) Scampavia, L. D.; Hodder, P. S.; Lahdesmaki, I.; Ruzicka, J. Trends Biotechnol. 1999, 17, 443-447. (33) Andersson, H.; van der Wijngaart, W.; Enoksson, P.; Stemme, G. Sens. Actuators, B 2000, 10, 1-2. (34) Choi, J. W.; Oh, K. W.; Thomas, J. H.; Heineman, W. R.; Halsall, H. B.; Nevin, J. H.; Helmicki, A. J.; Henderson, H. T.; Ahn, C. H. Lab Chip 2002, 2, 27-30. (35) Lettieri, G. L.; Dodge, A.; Boer, G.; de Rooij, N. F.; Verpoorte, E. Lab Chip 2003, 3, 34-39. (36) Oleschuk, R. D.; Shultz-Lockyear, L. L.; Ning, Y. B.; Harrison, D. J. Anal. Chem. 2000, 1, 585-590. (37) Li, J. J.; LeRiche, T.; Tremblay, T. L.; Wang, C.; Bonneil, E.; Harrison, D. J.; Thibault, P. Mol. Cell. Proteomics 2002, 1, 157-168. (38) Eyal, S.; Quake, S. R. Electrophoresis 2002, 23, 2653-2657. (39) Kohara, Y. Anal. Chem. 2003, 1, 3079-3085. (40) Malmstadt, N.; Yager, P.; Hoffman, A. S.; Stayton, P. S. Anal. Chem. 2003, 75, 2943-2949. (41) Buranda, T.; Huang, J.; Perez-Luna, V. H.; Schreyer, B.; Sklar, L. A.; Lopez, G. P. Anal. Chem. 2002, 74, 1149-1156. (42) Hayes, M. A.; Polson, N. A.; Phayre, A. N.; Garcia, A. A. Anal. Chem. 2001, 73, 5896-5902. (43) Buranda, T.; Jones, G.; Nolan, J.; Keij, J.; Lopez, G. P.; Sklar, L. A. J. Phys. Chem. B 1999, 103, 3399-3410. (44) Buranda, T.; Lopez, G. P.; Simons, P.; Pastuszyn, A.; Sklar, L. A. Anal. Biochem. 2001, 298, 151-162. (45) Buranda, T.; Sklar, L. A.; Lopez, G. P. In Biotechnology in Flow Cytometry; Sklar, L., Ed.; Oxford University Press: Oxford, in press. (46) Sato, K.; Tokeshi, M.; Odake, T.; Kimura, H.; Ooi, T.; Nakao, M.; Kitamori, T. Anal. Chem. 2000, 72, 1144-1147.

seconds in the flow-through microimmunoassays, compared to the standard g24 h associated with conventional titer plate assays. In a previous publication,41 we described a flow-through immunoassay approach based on the detection of defined molecular assemblies on beads trapped in a microchannel. Accurate quantitation was achieved by the use of known quantities of beads previously characterized by flow cytometry and spectrofluorimetry to account for surface coverage of receptor proteins. The detection relies on fluorescence resonance energy transfer (FRET); thus, the fluorescence of the beads corresponds to a known concentration of surface receptors, where the subsequent changes define the amount of specifically bound analytes, without signal interference from unbound or nonspecifically adsorbed analytes.41 In this work, we have expanded this approach to a multianalyte model system composed of discrete segments of beads that bear distinct receptors for the simultaneous detection of diverse analytes. In a demonstration assay, we show the FRET-based detection of the molecular recognition of FLAG peptide/antiFLAG and carcinoembryonic antigen (CEA)/anti-CEA. An additional segment of beads bearing fluorescein biotin was used as a control “channel” for recording photobleaching or nonspecific analyte interaction with receptor bead segments. The FLAG system is a commonly used epitope tag that relies on the octapeptide, DYKDDDDK, with readily available monoclonal antibodies.47-49 This system is an ideal model for the development of generalizable assays for proteins with known epitopes. CEA is a 180-kDa protein (CD66 family of antigens)50 that is expressed at elevated levels in patients afflicted by several forms of cancer including the following: colon, breast, lung, and pancreatic cancers as well as liver deseases.51-54 The data are analyzed in terms of mass transport limited kinetics.55,56 Furthermore, because the detection system allows for the quantitative characterization of the spatial distribution of the analyte during passage through the column, a framework for robust modeling of real-time analysis of unsteady-state phenomena is presented.57 EXPERIMENTAL SECTION Materials. L-R-Phosphotidylcholine (egg-PC), 1,2-dioleoyl-snglycero-3-phosphoethanolamine-N-(carboxyfluorescein) (fluoresceinPE), and 1,2-dipalmitoyl-sn-glycero-3-phosphoethanolamine-N(biotinyl) (biotin-PE) were purchased from Avanti Polar Lipids, Inc. (Alabaster, AL). The 5-, 10-, and 20-µm-diameter glass beads were obtained in dry form from Duke Scientific Corp. (Palo Alto, CA). Biotin and fluorescein biotin were purchased from Molecular (47) Brizzard, B. L.; Chubet, R. G.; Vizard, D. L. Biotechniques 1994, 16, 730735. (48) Knappik, A.; Pluckthun, A. Biotechniques 1994, 17, 754-761. (49) Slootstra, J. W.; Kuperus, D.; Pluckthun, A.; Meloen, R. H. Mol. Diversity 1997, 2, 156-164. (50) GrayOwen, S. D.; Dehio, C.; Haude, A.; Grunert, F.; Meyer, T. F. Embo J. 1997, 16, 3435-3445. (51) Berinstein, N. L. J. Clin. Oncol. 2002, 20, 2197-2207. (52) Gold, P.; Goldenberg, N. A. Perspect. Colon Rectal Surg. 1996, 9. (53) Maxwell, P. Br. J. Biomed. Sci. 1999, 56, 209-214. (54) Rymer, J. C.; Sabatier, R.; Daver, A.; Bourleaud, J.; Assicot, M.; Bremond, J.; Rapin, J.; Salhi, S. L.; Thinion, B.; Vassault, A.; Ingrand, J.; Pau, B. Clin. Chem. 1999, 45, 869-881. (55) Myszka, D. G.; He, X.; Dembo, M.; Morton, T. A.; Goldstein, B. Biophys. J. 1998, 75, 583-594. (56) Schuck, P.; Minton, A. P. Anal. Biochem. 1996, 5, 262-272. (57) R. B., B.; Stewart, W. E.; Lightfoot, E. N. Transport Phenomena; John Wiley & Sons: NewYork, 1960.

Analytical Chemistry, Vol. 76, No. 21, November 1, 2004

6267

Probes (Eugene, OR) and used without further purification. Biotinylated, fluorescein-labeled FLAG peptide (DYKDDDDK) was synthesized in-house as described previously.44 Anti-FLAG antibodies were purchased from Sigma Chemical Co. (St. Louis, MO). CEA and biotinylated human CD66e anti-CEA monoclonal antibodies (anti-CEA) were purchased from Chemicon Int. (Temecula, CA). p-Aminophenyltrimethoxysilane (ATMS 90%) were from Gelset Inc.. Sodium nitrite and sodium acetate were from Sigma. Ethanol (absolute, 200 proof) was purchased from Aaper Alcohols and Chemical Co., and glycine was purchased from Aldrich. Fluorescent Labeling of Monoclonal Antibodies and CEA. Proteins were labeled with the desired probe as described previously.44 The samples were purified and concentrated by ultrafiltration from phosphate-buffered saline using YM-30 Microcon centrifugal filter devices (Millipore Corp., Bedford, MA). The fluorophore to protein (f/p) ratios were determined following standard procedures from the manufacturers. The f/p ratios were generally on the order of 6:1 for Texas Red-labeled proteins and 2< for the fluorescein-labeled proteins. Preparation of Receptor Beads. In previous work, we have used standard commercial 6.2-µm streptavidin-coated beads, as substrates for molecular assembly in microchannels.41 During the course of that study it became clear that because of the substantial pressure losses suffered during fluid transport, larger biofunctionalized beads were better suited for the application. Therefore, we turned to 20-µm glass beads whose biofunctionalization is described below. Lipid-Coated Glass Beads. Lipid-coated glass beads were prepared as previously described.58 Briefly, glass beads were cleaned and surface modified by treating with a mixture of 4% H2O2 and 4% NH4OH (by weight) at 80 °C followed by a mixture of 4% H2O2 and 0.4 M HCl at 80 °C. After washing with distilled water and TRIS buffer (50 mM TRIS, 50 mM NaCl, pH 7.4) several times, beads were suspended in TRIS buffer. Hemocytometer analysis gave a bead concentration of 2.78 × 107 beads/mL. Known amounts of beads were incubated in a unilamellar lipid vesicle suspension consisting of biotin-PE and egg-PC in 1:400 molar ratio. The bead suspension was washed several times with TRIS buffer to remove unbound phospholipids. Streptavidin was then added to the beads as previously described.58 Diazotized Glass Beads. The 20-µm glass beads were sonicated for 5 min in deionized water and were subsequently cleaned by vortexing in piranha solution (3/7 by volume of 30% H2O2 and H2SO4; CAUTION: piranha solution reacts violently with most organic materials and must be handled with extreme care). After 30 min of vortexing, the beads were thoroughly rinsed in deionized water followed by several rinses in 100% ethanol. The beads were subsequently coated with ATMS by immersion in a 1 mM ethanolic solution of ATMS and vortexing for up to 4 h. The glass beads were then rinsed in ethanol and deionized water. This procedure resulted in the formation of amine-terminated monolayer on the glass substrate. ATMS-coated beads were converted to the diazobenzyl form by treatment with a solution containing 40 mL of water, 80 mL of 1.8 M HCl, and 3.2 mL of freshly prepared solution of NaNO2 (10 mg/mL, pH 0.9) for 30 min at 4

°C.18 After 30 min, the ATMS-treated surfaces were washed 3 times, each for 3 min, with ice-cold sodium acetate buffer (50 mM, pH 4.7) followed by washing with ice-cold deionized water and ethanol 2 times each (5-min washes). The diazotized beads were reacted with enough streptavidin to ensure complete surface coverage of the beads. The suspension was incubated with vigorous vortexing for 3 h at room temperature. The sample was then rinsed with doubly deionized water and resususpended in 100 mM Tris buffer (pH 7.5). The surface density of streptavidin sites was determined by centrifugation assays and flow cytometry.43,44 Flow Cytometry. The flow cytometric analysis used a BectonDickinson FACScan flow cytometer (Sunnyvale, CA) interfaced to a Power PC Macintosh using the CellQuest software package. The FACScan is equipped with a 15-mW air-cooled argon laser. It has been shown elswhere43,44 that the mean of the histogram is the quantity relevant to binding capacity. The average fluorescence of a single bead is converted to the number of fluorophores per bead on the basis of flow cytometric calibration beads. Characterization of Molecular Assembly Components: Biotinylated and Fluorescein-Tagged FLAG Peptide and Fluorescently Tagged Antibodies Equilibrium Binding to Beads. The equilibrium binding characteristics of fluorescein biotin and the FLAG peptide have already been described elsewhere.43,44,58 Because of their known binding constants to beads, fluorescein biotin or FLAG peptide were assembled on beads such that the site densities of the fluorophores were limited to 1 million/bead in each case. The binding of biotinylated antiCEA antibody was performed following procedures described previously for anti-FLAG antibodies.44,59 Binding of Texas Red-Labeled CEA (TR-CEA) to Fluorescein-Labeled Anti-CEA Antibody Bearing Beads. The experimental protocol was similar to that previously described for FLAG peptides41,43,44,58 Briefly, 5-µL suspensions of anti-CEA antibody bearing beads were added to microcentrifuge tubes. TRCEA was then added in 2- and 6-µL volumes, and then buffer was added such that the final volume for each sample was 20 µL with final concentrations ranging from 1 to 100 nM in the respective tubes. The samples were incubated for 10 min while shaking. The samples were then transferred to FACScan tubes, and buffer was added to a final volume of 200 µL for analysis by flow cytometry. The data were analyzed in terms of energy transfer between fluorescein labeled anti-CEA antibodies and TR-CEA. Kinetic Cytometric Analysis. The time course of association was typically measured by acquiring a 10-s baseline of beads bearing fluorescent FLAG peptide and then adding an aliquot of Texas Red-labeled antibody to the sample, briefly vortexing (thus creating a 5-10-s gap in the time course), and resuming data collection continuously to completion at some desired time (e.g., 60-1000 s). The raw data were converted to ASCII format, suitable for kinetic fitting, using FCSQuery software developed by Dr, Bruce Edwards at UNM. The time-resolved FRET quenching data were fit by nonlinear regression methods using the software package Scientist from Micromath Scientific Software (Salt LakeCity UT). The binding data were analyzed using a model based on the law of mass action shown in eq 1,

(58) Buranda, T.; Huang, J.; Rama Rao, G. V.; Ista, L. K.; Larson, R. S.; Ward, T. L.; Sklar, L. A.; Lopez, G. P. Langmuir 2003, 19, 1654-1663.

L + R y\ z LR k

6268

Analytical Chemistry, Vol. 76, No. 21, November 1, 2004

k1

-1

(1)

where L refers to the Texas Red-labeled anti-FLAG antibody and R to the fluorescent FLAG peptides on beads. k1 is the binding rate constant, and k-1 is the dissociation rate constant. Assessing the Extent of Surface Coverage of FLAG Peptide-Anti-FLAG Complex on Beads. The surface occupancy of streptavidin-coated beads by FLAG peptides was determined on the basis of flow cytometry intensity histograms as previously described.43 The average quantity of anti-FLAG antibodies bound to each bead was proportional to the intensity (Ii) (relative to the intensity of antibody-free beads normalized to unity) of the beads that were exposed to a given concentration of the anti-FLAG antibody. The quantity of bound antibodies can be expressed in terms of fractional coverage per bead (eq 2).

site coverage ) θi ) (1 - Ii)/(1 - If) (1 - Ii) )

(1 - Ii)[IgG]i Kd + [IgG]i

(2)

(3)

Here, If is the intensity of FLAG peptide bearing beads saturated with anti-FLAG antibodies. For a series of concentrations of antiFLAG antibodies, [IgG]i, the law of mass action-derived expression (eq 3) can be used to determine the dissociation constant (Kd) of the FLAG peptide/antibody complex. Fabrication of Microchannels. PDMS microchannels were constructed using soft lithographic techniques adapted from the literature.60 The microfluidic channels were fabricated with weirs to hold the beads in place. Test grade N-type silicon wafers (WaferNet Inc.) were cleaned in piranha etch for 2 h and rinsed with ultrapure water and ethanol, respectively. Cleaned Si wafers were dried in a stream of N2. A thin layer of negative photoresist (SU8-50, Microchem Corp.) was spin-coated on to the cleaned wafer at 500 rpm for 12 s followed by 3000 rpm for 1 min and then baked for 20 min at 75 °C. A region on the baked sample was then covered with 2 mm by 3 cm Scotch brand clear tape. Another layer of negative photoresist (SU8-100, Microchem Corp.) was then spin-coated at 500 rpm for 12 s and 2000 rpm for 1 min. The sample was baked for 45 min at 75 °C, after which the tape was removed. A photoresist layer with two regions of different thickness was thus fabricated. The pattern of the microchannel was drawn using CorelDraw10 software and printed as 2400 dpi positive film using an Agfa Selectset 5000 Imagesetter (SUBIA, Albuquerque, NM). The mask was aligned with the photoresistcoated silicon wafer such that the thinner region of the photoresist was under the region near to the end corresponding to the outlet of the microchannel. This sample was exposed to UV radiation for 15 min, baked for 10 min, and then rinsed with SU-8 developer to remove unpolymerized photoresist from the wafer. This served as the master for the microchannel. The sample was dried at room temperature for 0.5 h. Two-pieces of 0.3-mm-long Cu wire (0.5mm diameter) were then glued to the two ends of the channel to support two 1-cm-long silicone tubings (0.64-mm i.d., Cole Parmer, Vernon Hills, IL) that were attached to the two ends. These (59) Simons, P.; Shi, M.; Foutz, T.; Lewis, J.; Buranda, T.; Lim, W. K.; Neubig, R.; Garrison, J.; Prossnitz, E. R.; Sklar, L. A. Mol. Pharmacol. 2003, 64, 1227-1238. (60) Duffy, D. C.; McDonald, J. C.; Schueller, O. J. A.; Whitesides, G. M. Anal. Chem. 1998, 70, 4974-4984.

Figure 1. (A) Elastomeric silicone microchannel, mounted on glass slide with two openings for sample delivery and egress. The microchannel is 300 µm wide, 70 µm deep. and 2 cm long. A ∼60-µm weir is used to trap 30-µm borosilicate naked beads. The 20-µm-size biofunctionalized beads coated with receptors (anti-CEA, FLAG peptide, and fluorescein biotin) packed in tandem and separated with naked beads to form a ∼5.4-mm-long affinity microcolumn, composed of 600-µm receptor bead segments and two 1.5- and 0.6-mm spacer beads segments. (B) Schematic of a microfluidic apparatus showing the configuration in which sample was delivered and fluorescence measurements taken with a spectrofluorometer. (C) Plot of emission intensity of biofunctionalized (peaks) and naked (valleys) bead segments versus translation along the column length. The scan rate was kept at ∼0.05-0.07 mm/s.

tubings served as fluidic connectors to the microchannel. The PDMS prepolymer (Sylgard 184, Dow Corning, Midland, MI) was poured over the master and was allowed to release trapped air in the prepolymer solution for ∼30 min. The sample was cured at 90 °C for 1 h, and the PDMS layer was peeled off from the master. The PDMS channel was then irreversibly sealed onto a glass slide using an Ar plasma (Harrick Plasma Cleaner, 100 W, 50 mTorr). A schematic drawing of the finished microchannel is shown in Figure 1A. Channel dimensions were typically 2 cm, 300 µm, 6070 µm in length, breadth, and depth, respectively. To trap beads Analytical Chemistry, Vol. 76, No. 21, November 1, 2004

6269

near the outlet, the depth of the channel was limited to 12-15 µm. Packing of Microchannels with Beads. Single-analyte (FLAG) microcolumns were packed as previously described.41 For multianalyte columns, 5-µL aliquots of receptor bearing bead solutions were injected into the column by applying a vacuum at the outlet. Each bead segment was packed by flowing several microliters of TRIS buffer. Each 600-µm receptor bead segment was separated by a 1.5-mm segment of naked beads. The 600-µm receptor bead segments were designed to optimize the signal. The length of the spacer beads was optimized to circumvent the incidence of signal cross talk between the separated receptor bead segments. The total length (∼5.4 mm) of the column comprised three (600 µm) segments of receptor beads with two (1.5 mm) spacer bead segments in addition to a 600-µm foundation segment of spacer beads at the base of the column. The order in which the receptor beads were packed in the columns was permuted to check the spatial sensitivity to photobleaching and nonspecific binding artifacts. Bead-packed channels were kept wet with TRIS buffer that was allowed to continuously percolate through the column under gravity until ready for use. Analyte Detection in Affinity Microcolumns. The multianalyte columns were mounted onto a motorized vertical translational stage (Figure 1B) located in the sample holder space of a Model Fluorolog-3 SPEX fluorometer (Instruments S.A.). Each segment of surface-functionalized beads was irradiated with 8-10 mW of 488-nm laser excitation in sequence with the periodic up and down movement (at 0.05-0.07 mm/s) of the translation stage. The laser power and scan rate were empirically optimized to maximize the signal, while minimizing photobleaching. The inlet of the column was connected to a buffer reservoir, while the outlet was connected to a vacuum source. Several microliters of TRIS buffer were passed through the microchannel before the injection of the sample. While applying the vacuum at the outlet, 4-µL volume samples of analytes were injected directly into the column through the inlet silicone tubing using a 10-µL Hamilton syringe. The flow rate was maintained at 0.26 µL/min by controlling the vacuum at the outlet. The interaction of the flowing analyte fluid was monitored as the change in the original intensity of fluorescence signal of FITCanti-CEA antibody, FITC-FLAG peptide and fluorescein biotin bearing beads at 520 nm. The FRET data were analyzed in terms of a Langmuirian twocompartment transport-kinetic model of the type shown in eq 4

dΓAB D ) kfC0ΓA - kbΓAB ) (Cb - C0) dt δ

(4)

as previously described.41,61 The model relates the association and dissociation rate constants for bead-analyte as well as the analyte’s mass transport properties to the time evolution of analyte binding. Cb and C0 represent the concentrations of antibody in the bulk and at the liquid-solid interface, respectively; ΓAB is the surface concentration of FLAG peptides bound to antibodies; ΓA is the surface concentration of unbound peptides; and kf and kb are the forward and reverse kinetic rate constants. D is the diffusion (61) Bourdillon, C.; Demaille, C.; Moiroux, J.; Saveant, J. M. J. Am. Chem. Soc. 1999, 121, 2401-2408.

6270 Analytical Chemistry, Vol. 76, No. 21, November 1, 2004

Figure 2. Equilibrium binding curve of CEA/anti-CEA antibody complexes on beads. The derived affinity constant is 26.0 nM.

coefficient of the antibody and δ is the thickness of the steadystate diffusion-convection boundary layer established by fluid transport in which we assumed a linear gradient in concentrations (between Cb and C0). The parameter D/δ represents the effects of diffusive/convective transport of analytes to the surface receptors. RESULTS AND DISCUSSION Flow Cytometry Measurement of FRET and Binding of CEA to Anti-CEA Antibodies on Beads. The chemical labeling of the CEA with Texas Red fluorophores resulted in an average of ∼6.0 fluorophores/protein while retaining binding activity. The high density of Texas Red fluorophores per protein favors the likelihood of having at least one Texas Red moiety in proximity to the fluorescein tag(s) on the antibody (f/p ≈ 2e), thus enhancing the effectiveness of FRET as a transducer of the binding of antibodies to the 180-kDa protein. The equilibrium binding data of Texas Red-labeled CEA to fluorescein-labeled anti-CEA antibody beads is shown in Figure 2. Binding of the Texas Red-tagged CEA is characterized by the FRET quenching of antibody fluorescence. The TR-CEA was determined to bind to the anti-CEA bearing beads with an affinity constant of Kd ≈ 26.0 nM using Prism (GraphPad Software, Inc.) It is worth mentioning that numerous epitope regions on the CEA protein have been mapped and monoclonal antibodies for these regions are commercially available and that the Kd measured here refers to the CD66e antibody. Kinetic Analysis of Texas Red-Labeled Anti-FLAG Antibodies to FLAG Peptide-Bearing Beads. The binding of Texas Red-labeled anti-FLAG antibodies (TR-anti-FLAG) to a suspension of FLAG peptide-bearing beads was monitored in real time by flow cytometry. The data are summarized in Figure 3 for three different concentrations of TR-anti-FLAG. The intensity data were converted to surface coverage units using eq 2. At saturation, it was assumed that the site density of the antibody was equivalent to the number of peptides on the beads, ∼0.75 nM. The data were fit to the model given in eq 1 using the software package Scientist (Micromath, San Diego, CA). The fits to the experimental data yield the following parameter values: Kd ) 7 ( 3.0 nM, k1 ) (6.0 ( 2.6) × 105 M-1 s-1, and k-1 ) (3.3 ( 1.0) × 10-3 s-1. The affinity constant derived from the kinetic data was similar to that determined from equilibrium binding measurements.44,59 Two-Compartment Transport-Kinetic55,62-65 Model Analysis of the Passage of TR-Anti-FLAG through Affinity Microcolumn. Several concentrations of TR-anti-FLAG were analyzed with affinity microcolumns. The kinetics of binding were analyzed

Figure 3. Analysis of the kinetics of binding of Texas Red-labeled anti-FLAG antibodies to fluorescein-labeled FLAG peptides on beads. The raw data were measured as FRET quenching on beads using a flow cytometer. The data were converted to surface coverage units using eq 2. The fit of the data to a simple bimolecular binding model gave affinity, binding, and dissociation rate constants of Kd ) 7 ( 3.0 nM, k1 ) (6.0 ( 2.6) × 105 M-1 s-1, and k-1 ) (3.3 ( 1.0) × 10-3 s-1 respectively.

using eq 4. Equation 4 describes the binding and dissociation kinetics with a simple first-order reaction rate law. Equation 4 was solved numerically using the software packages Scientist and Berkeley Madonna (Robert I. Macey and George F. Oster, UC Berkeley, Berkeley, CA). The analyzed data are shown in Figure 4. The binding and dissociation rate constants are given separately for each curve and they average kf ) (9.0 ( 6.0) × 104 M-1s-1 and kb ) (1.2 ( 0.8) × 10-3 s-1. The averaged magnitude of the binding and dissociation rate constants derived for the different concentrations of analyte (Figure 4) are of the same order as those derived for the well-mixed flow cytometry experiments. However, the microcolumn-derived parameters are notably smaller than those obtained from flow cytometry measurements. There is a notable parallel decrease in the values of kf and kb with increasing concentration of the analyte (anti-FLAG antibody). This trend is a probable indicator of the increasing dominance of mass transport over kinetic considerations as the analyte concentration is increased. This assertion can be clarified by using a couple of simple semiquantitative approaches. First, a cursory examination of the 96 nM data set (in Figure 4A) suggests that the half-time to saturation (t1/2) of receptor sites by the analyte is on the order of 200 s. In contrast, the t1/2 () ln2/r) calculated from the quasiunimolecular rate constant r ) kf[IgG] + kb obtains a t1/2 value of 70 s. Second, the Damko¨hler number (Da) is a dimensionless quantity that is often used to represent the ratio between the rate of a reaction and the rate of delivery of reactants to the reaction interface.66 If Da >1, then the process is mass transport limited and when Da < 1 then the processes is reaction rate limited. Da can be evaluated from eq 5. where the terms as used in our

Da ) kf[IgG]d/(D/δ)

(5)

formalism are d is the interstitial space between the packed beads and D/δ is the rate of diffusive/convective transport of [IgG] to (62) Vijayendran, R. A.; Ligler, F. S.; Leckband, D. E. Anal. Chem. 1999, 71, 5405-5412. (63) Goldstein, B.; Coombs, D.; He, X. Y.; Pineda, A. R.; Wofsy, C. J. Mol. Recognit. 1999, 12, 293-299. (64) Mason, T.; Pineda, A. R.; Wofsy, C.; Goldstein, B. Math. Biosci. 1999, 159, 123-144. (65) Wofsy, C.; Goldstein, B. Biophys. J. 2002, 82, 1743-1755. (66) Fogler, H. S. Elements of Chemical Reaction Engineering, 3rd ed.; Prentice Hall: Upper Saddle River, NJ, 1999.

Figure 4. (A) Binding curves of 2-µL plugs of Texas Red-labeled monoclonal anti-FLAG antibodies through affinity microcolumns of fluorescein-labeled FLAG peptide-bearing beads. The points refer to the normalized intensity readings taken during each run. The lines represent simulations using parameters from kinetic analysis shown in panel B. Panel A was reproduced with permission from ref 4. (B) Kinetic analysis of the binding of TR-anti-FLAG to beads. The derived rate constants: (96.0 nM) kf ) 2.9 × 104 M-1 s-1; kb)3.5 × 10-3 s-1, (48.0 nM) kf ) 5.2 × 104 M-1 s-1; kb ) 9.9 × 10-4 s-1. (24.0 nM) kf ) 8.6 × 104 M-1 s-1; kb ) 1.0 × 10-3 s-1, (9.6 nM) kf ) 1.35 × 105 M-1 s-1, kb ) 2.0 × 10-3 s-1, (4.8 nM) kf ) 2.5 × 105 M-1 s-1, kb ) 2.7 × 10-3 s-1, (0.48 nM) kf ) 6 × 105 M-1 s-1; kb ) 5 × 10-3 s-1. The variations in the derived results are likely due to uncertainties in the initial concentrations of reagents (surface occupancy and number of beads) as well as mass transport limitations; see text for details. Some data sets were omitted for figure clarity.

the surface-immobilized peptides. The numerical values of the parameters in eq 5refer to kf(k1) derived from the well-mixed flow cytometry data (Figure 3), an estimate of d on the order of 3.1 × 10-6 m,67 and the phenomenological value of D/δ ≈ 1.2 × 10-10 ms-1 derived from the fit of eq 3. This analysis shows that only the 0.48 nM sample is in the regime of reaction-limited kinetics, whereas the rest of the data are dominated by mass transport considerations. This analysis also suggests that increasing the flow rate would reduce the reaction time of the transport-limited reactions to the limit defined by kinetics. Thus, it is worth noting that, in the limit of low analyte concentration, such as the 0.48 nM assay (Da ) 0.48 ≈1), where the quasi-unimolecular rate (r) is dominated by the dissociation rate constant kb (t1/2 ≈ 552 s), a higher flow rate may do little to reduce the reaction time and may likely be counterproductive, because of the reduced analyte residence time. Detection of TR-Anti-FLAG and TR-CEA via FRET in Multianalyte Affinity Columns. Figure 1 shows a prototypical (67) Alargova, R. G.; Petkov, J. T.; Denkov, N. D.; Petsev, D. N.; Ivanov, I. B. Colloids Surf. A 1998, 134, 331-342.

Analytical Chemistry, Vol. 76, No. 21, November 1, 2004

6271

Figure 5. (A) Normalized intensity of biofunctionalized beads versus time. The data show the extent of photobleaching of beads over time of periodic laser scanning. The data represent an average of three separate measurements. Each analyte was in 2-µL, 1 µM aliquots (B) Emission intensity of biofunctionalized beads over 30 min during passage of soluble CEA and anti-FLAG antibody. The binding of analytes to the FLAG (FL) and CEA beads is indicated by FRET quenching of those bead segments. The intensity of the fluorescein biotin beads (B) remained similar to panel A. The data shown here has been corrected for photobleaching using the fluorescein biotin data set as a reference. The flow rate was maintained at 0.26 µL/ min.

setup of the affinity microcolumn multianalyte detection system described herein. Figure 1A is a schematic depiction of the microchannel where receptor functionalized beads are packed in tandem, with intervening segments of inert beads to achieve spatial resolution. Figure 1B displays a schematic illustration of the affinity microcolumn mounted on a vertical translation stage inside the sample compartment of a spectrofluorometer. Figure 1C shows the emission intensity measurement of receptor-bearing (the peaks) and naked beads (the valleys) after one full z-axis translation of the entire length of the packed column past the laser excitation. The motorized back and forth translation of the column was kept at the same velocity (∼0.05-0.07 mm/s) during experimental runs. The intensity data for each receptor bead segment were integrated over the entire peak corresponding to the target bead segment and normalized to the initial intensity at time zero. The peak profile is the result of the transient simultaneous irradiation of spacer and sensor beads and the segment boundaries during the scans. In addition, inhomogeneities of packing (slight mixing of sensor beads with spacer beads near two boundaries) could potentially affect the line shape of the peak. Figure 5A shows the normalized intensities of beads bearing anti-CEA antibody, FLAG peptide, and fluorescein biotin. The affinity microcolumn was scanned at 5-min intervals for 30 min to minimize photobleaching. This experiment demonstrates that the receptor beads experienced 5-10% reduction in intensity from photobleaching alone. Because the extent of photobleaching was 6272 Analytical Chemistry, Vol. 76, No. 21, November 1, 2004

Figure 6. (A) Schematic representation of a competitive binding assay. For example, TR-anti FLAG antibodies are premixed with soluble FLAG peptide after which, the sample is pumped through the affinity microcolumn. The binding of the soluble peptide to antibody is measured in terms of reduced FRET relative to neat TR-antiFLAG. (B) Normalized and photobleaching-corrected receptor segment beads. Competition for soluble TR-antibody between receptor beads and 100-fold soluble FLAG peptide results in the complete inhibition of FRET. (C) Competitive binding between excess soluble CEA and TR-CEA for CEA receptor beads.

the same for all three segments, the fluorescein biotin segment was used as an indicator of baseline drift due to photobleaching as well a marker of potential nonspecific interactions between the beads and the soluble TR-CEA or TR-anti-FLAG analytes. Figure 5B shows intensity data of receptor beads after passage through the column of a mixed sample of TR-CEA and TR-anti-FLAG. It is worth mentioning that the order in which the fluorescein biotin, anti-CEA-CEA, or FLAG beads appeared in the column had no effect on their respective signal. The data are plotted after correction for “baseline drift” that is based on the fluorescein biotin segment. Figure 6A shows a plot of baseline drift-corrected data after passage of TR-CEA, TR-anti-FLAG, and excess nonfluorescent FLAG peptide. Because the large excess of soluble FLAG peptide competitively binds to TR-anti-FLAG at the expense of the beadborne fluorescent FLAG peptide, TR-anti-FLAG does not bind to the beads. Thus, FRET is inhibited in this case. Figure 6B shows a plot of baseline drift-corrected data after passage of TRCEA, excess unlabeled anti-CEA antibody and TR-anti-FLAG. As in Figure 6A, the soluble anti-CEA is shown to effectively block

bead characterization in flow cytometry (cf. Figures 2 and 3), the two-compartment analysis of the assays in single-analyte columns (cf Figure 4) provides a template upon which the dynamics of multiple-analyte assays can be characterized using existing theoretical models and be tested experimentally (cf. Figures 6 and 7), once more robust materials are used to fabricate microcolumns.

Figure 7. (a) Intensity profile of fluorescently labeled analyte (antiFLAG) during passage through the affinity microcolumn driven by transport buffer (0.26 µL/min). (b) Intensity profile of soluble analyte during passage in affinity column in the absence of dilution and dispersion in transport buffer.

the binding interaction between the TR-CEA and the fluoresceinlabeled anti-CEA antibodies on beads. To visualize the effect of analyte dispersion in the transport buffer fluid during passage through the microcolumn, we measured the fluorescence intensity of soluble anti-FLAG as it passed a static point of laser excitation, through a column identical to those described above but without receptors. Figure 7 shows the emission intensity profile of a 4 µL of 10-6 M sample of fluoresceinlabeled anti-FLAG antibody using 488-nm laser excitation. The data show that the 4-µL plug of analyte fluid is dispersed in the transport buffer ultimately resulting in the dilution of the analyte on average, by a factor of ∼3. At a flow rate of 0.26 µL/min, the data indicate that it takes well over 1 h for the originally 4-µL volume analyte to pass through the column. The extent of analyte dispersion and the flow rate of analytes suggest that the reaction of analytes with the column beads is strongly limited by mass transport considerations. The concentration profile of the analyte flowing through the channel is reminiscent of matters associated with unsteady-state phenomena in chromatographic separations and chemical reactor operations.27,57,68 The clear dominance of mass transport over kinetics suggests that the data collected under current experimental conditions can only be analyzed as end point assays. Because the microchannels fabricated from PDMS cannot withstand the greater than 2-atm pressure loss, the extent of analyte dilution and dispersion cannot be mitigated by increasing the flow rate without causing the failure of the microchannel.41 It is however important to note that the results presented here represent a significant step in the development of this type of assay. We have demonstrated a methodology by which molecular recognition-based FRET assays can be achieved for multiple analytes in affinity columns. Beginning with (68) Hage, D. S. J. Chromatogra., B 1998, 715, 3-28. (69) Selvaganapathy, P. R.; Carlen, E. T.; Mastrangelo, C. H. Proc. IEEE 2003, 91, 954-975. (70) Reyes, D. R.; Iossifidis, D.; Auroux, P. A.; Manz, A. Anal. Chem. 2002, 15, 2623-2636. (71) Vilkner, T.; Janasek, D.; Manz, A. Anal. Chem. 2004, 15, 3373-3385. (72) Svitel, J.; Balbo, A.; Mariuzza, R. A.; Gonzales, N. R.; Schuck, P. Biophys. J. 2003, 84, 4062-4077. (73) Balgi, G.; Leckband, D. E.; Nitsche, J. M. Biophys. J. 1995, 68, 2251-2260.

SUMMARY AND CONCLUSIONS This study has demonstrated some important factors related to the design, assessment, and utility of affinity microcolumn sensors. First, beads derivatized with surface chemistry suitable for the attachment of fluorescently labeled biomolecules of interest are prepared and characterized in terms of functionality and receptor site densities by flow cytometry. Second, calibrated beads are incorporated in microfluidic channels. The analytical device that emerges replicates the basic elements of affinity chromatography with the advantages of (1) scale, (2) direct measurement of bound analyte on beads rather than the indirect determination from eluted sample typical of affinity chromatography, and (3) simultaneous detection of multiple analytes from columns composed of discrete segments of diverse populations of receptor beads. An optimally efficient device is likely to emerge from design criteria inherent in the examples described in the literature.69-71 Our approach so far has addressed analyte detection and quantitation. The next step involves selection of a more robust fabrication material other than PDMS that would allow the integration of more segments to the assay. Current effort is underway to fabricate microchannels out of (400-µm i.d.) square capillaries in order to fabricate affinity microcolumns capable assaying a larger diversity of analytes, as well as achieving flow rates of sufficient magnitude to enable a limiting analysis of the temporal dynamics of the passage of analytes through the columns.55,56,62-65,72,73 We have successfully packed test columns that exceed 7 cm in length (equal to discrete 38 analyte segments in the current configuration). Flow rates on the order of 0.67 µL/ min (cf. 0.26 µL/min for the current 5.4-mm PDMS column) at applied pressures of ∼60 PSI have been achieved. To circumvent the ungainly nature of a 7-cm-long column, we will design a system based on tandem serpentine-linked arrays of microcapillary columns similar to an approach described by Phillips.8 ACKNOWLEDGMENT This work was supported by the National Science Foundation MCB-9907611 (T.B.) NSF CTS0332315 (G.P.L.), NIH-BECON GM60799-02 (L.A.S.), and NM State Cigarette tax to the UNM Cancer Center.

Received for review May 20, 2004. Accepted August 4, 2004. AC049260F

Analytical Chemistry, Vol. 76, No. 21, November 1, 2004

6273