Anal. Chem. 2004, 76, 7323-7328
Transfected Cell Microarrays for the Expression of Membrane-Displayed Single-Chain Antibodies James B. Delehanty,* Kara M. Shaffer, and Baochuan Lin
Center for Bio/Molecular Science and Engineering, Naval Research Laboratory, Washington, D.C. 20375-5348
The expression of recombinant antibody fragments on the surface of mammalian cells has recently emerged as a therapeutic strategy, particularly in the treatment of a number of cancers. Screening technologies that allow for the facile characterization of fragments expressed on the cell surface would hasten the identification and isolation of reagents to be used as therapeutics. In this report, we describe a cellular microarray-based platform for the comparative functional analysis of single-chain antibodies (scFvs) expressed on the plasma membrane of mammalian cells. Using the anti-fluorescein monoclonal antibody 4-4-20 as a model system, the native binding site and three mutants were expressed as scFvs on the membrane of HEK 293T/17 cells in a microarray format. Collectively, the equilibrium dissociation constants of the soluble forms of the wild-type scFv and the three mutants spanned nearly 3 orders of magnitude. Expression of the scFvs on the surface of mammalian cells was achieved by the deposition of plasmid DNAs in micrometer-sized spots onto the surface of a glass microscope slide. The addition of cells to the printed array resulted in the expression of the scFvs in clusters of cells in spatially discrete locations. Ligand binding assays performed with a fluorescein-bovine serum albumin conjugate demonstrated the ability of the transfected cell microarray to differentiate the relative binding affinities of the expressed scFvs. Further, the apparent affinities of the membranedisplayed scFvs were within 10-fold of those reported for the soluble forms of the scFvs. The assays described herein demonstrate the potential for cellular microarrays to be used for the high-throughput screening of potential therapeutic reagents. More generally, our work details the utility of transfected cell microarrays in mediating the functional characterization of expressed membrane receptor proteins. Single-chain antibodies (scFvs) and other recombinant antibody fragments have enormous potential as diagnostic and therapeutic reagents. Their small size coupled with their ability to retain the binding characteristics of the native immunoglobulin have led to their use as reagents for the detection of tumorassociated antigens (TAAs), such as carcinoembryonic antigen * To whom correspondence should be addressed. Phone 202-767-0291. Fax 202-767-9594. E-mail:
[email protected]. 10.1021/ac049259g CCC: $27.50 Published on Web 11/10/2004
© 2004 American Chemical Society
(CEA),1 the in vivo imaging of tumor sites bearing markers of angiogenesis2 and CEA,3,4 and the targeted delivery of anticancer therapeutics.5-8 More recently, the expression of anti-TAA scFvs both within cells9 and on the mammalian cell surface10-12 has emerged as another therapeutic avenue for the treatment of certain cancers. For example, Kuroki et al.12 were able to achieve the localization of cytotoxic T cells to the site of CEA-expressing tumor cells through the expression of an anti-CEA scFv on the T cell plasma membrane. Concomitant with the development of this new therapeutic avenue has emerged the need for a highthroughput means to assess the binding activity of scFvs and other recombinant fragments as they are displayed on the surface of mammalian cells. Cellular microarrays, arrays of cells expressing defined cDNAs, hold promise as a potential platform for the parallel analysis of the binding characteristics of cell-surface expressed receptors. In the cellular microarray technique, cDNAs in the form of plasmids are deposited at high density onto a microscope slide, and mammalian cells are subsequently added to the printed array. Upon attachment to the substrate, the cells take up the plasmid DNA and express the particular protein encoded at each location. The result is an array whose features are micrometer-sized clusters of cells expressing defined genes at a high spatial density.13-15 (1) Paus, E.; Almasbak, H.; Bormer, O. P.; Warren, D. J. J. Immunol. Methods 2003, 283, 125-139. (2) Santimaria, M.; Moscatelli, G.; Viale, G. L.; Giovannoni, L.; Neri, G.; Viti, F.; Leprini, A.; Borsi, L.; Castellani, P.; Zardi, L.; Neri, D.; Riva, P. Clin. Cancer Res. 2003, 9, 571-579. (3) Olafsen, T.; Cheung, C.-W.; Yazaki, P. J.; Li, L.; Sundaresan, G.; Gambhir, S. S.; Sherman, M. A.; Williams, L. E.; Shively, J. E.; Raubitschek, A. A.; Wu, A. M. Protein Eng., Des. Sel. 2004, 17, 21-27. (4) Sundaresan, G.; Yazaki, P. J.; Shively, J. E.; Finn, R. D.; Larson, S. M.; Raubitschek, A. A.; Williams, L. E.; Chatziioannou, A. F.; Gambhir, S. S.; Wu, A. M. J. Nucl. Med. 2003, 44, 1962-1969. (5) Borsi, L.; Balza, E.; Carnemolla, B.; Sassi, F.; Castellani, P.; Berndt, A.; Kosmehl, H.; Biro, A.; Siri, A.; Orecchia, P.; Grassi, J.; Neri, D.; Zardi, L. Blood 2003, 102, 4384-4392. (6) Weisbart, R. H.; Miller, C. W.; Chan, G.; Wakelin, R.; Ferreri, K.; Koeffler, H. P. Cancer Lett. 2003, 195, 211-219. (7) Onda, M.; Wang, Q.-C.; Guo, H.-f.; Cheung, N.-K. V.; Pastan, I. Cancer Res. 2004, 64, 1419-1424. (8) Winthrop, M. D.; DeNardo, S. J.; Albrecht, H.; Mirick, G. R.; Kroger, L. A.; Lamborn, K. R.; Venclovas, C.; Colvin, M. E.; Burke, P. A.; DeNardo, G. L. Clin. Cancer Res. 2003, 9, 3845S-3853S. (9) Chen L, L. G., Tang L, Wang J, Ge XR. Cell Res. 2002, 12, 47-54. (10) Kuroki, M.; Shibaguchi, H.; Imakiire, T.; Uno, K.; Shirota, K.; Higuchi, T.; Shitama, T.; Yamada, H.; Hirose, Y.; Nagata, A. Anticancer Res. 2003, 23, 4377-4381. (11) Arakawa, F.; Shibaguchi, H.; Xu, Z.; Kuroki, M. Anticancer Res. 2002, 22, 4285-4289. (12) Kuroki, M.; Arakawa, F.; Khare, P. D.; Liao, S.; Matsumoto, H.; Abe, H.; Imakiire, T. Anticancer Res. 2000, 20, 4067-4071.
Analytical Chemistry, Vol. 76, No. 24, December 15, 2004 7323
To date, the utility of the cellular microarray platform has been demonstrated for the identification of cellular gene products that induce particular cellular phenotypes as well as for the detection of the interaction of drugs with their cognate receptors.14,16 However, the use of the cellular microarray approach to achieve the quantitative characterization of the binding activity of cellsurface expressed receptors has been limited. In this report, we demonstrate the use of cellular microarrays to characterize the relative binding affinities of a model single-chain antibody expressed on the surface of mammalian cells. Using a membranetargeting expression vector, an scFv version of the anti-fluorescein antibody, 4-4-20, was expressed on the mammalian cell surface as a fusion with the transmembrane domain of the platelet-derived growth factor receptor. We demonstrate the ability of the microarray to differentiate the relative binding affinities of the native scFv and three binding site mutants whose affinities span almost 3 orders of magnitude. To our knowledge, this represents the first instance in which the cellular microarray format has been used to achieve the quantitative characterization of cell-surface expressed scFvs. The work described herein demonstrates the utility of transfected cell microarrays for the screening and characterization of cell-surface expressed scFvs and has applications for the high-throughput screening of potential therapeutic agents. EXPERIMENTAL SECTION Materials. Polystyrene plates (medium-binding) and GAPS II microscrope slides (3-aminopropyl triethoxysilane) were obtained from Corning (Acton, MA). The HEK 293T/17 cell line and fetal bovine serum were products of the American Type Culture Collection (Manassas, VA). Press-to-Seal silicone well forms were purchased from Schleicher and Schuell (Keene, NH). Dulbecco’s modified Eagle medium (high glucose) was purchased from Invitrogen (Carlsbad, CA). The FluoReporter FITC protein labeling kit was purchased from Molecular Probes (Eugene, OR). FITC-conjugated anti-myc epitope polyclonal antibody was a product of Novus Biologicals (Littleton, CO). The antibiotic/ antimycotic, bovine serum albumin, gelatin (type B, 225 Bloom), sodium bicarbonate, sodium pyruvate, and sucrose were purchased from Sigma (St. Louis, MO). The pDsRed2-Nuc expression vector was obtained from BD Biosciences (Palo Alto, CA). VectaShield mounting medium containing DAPI (4′,6-diamidino2-phenylindole) was purchased from Vector Laboratories (Burlingame, CA). The Effectene transfection kit and plasmid preparation kits were obtained from Qiagen (Valencia, CA). Pfu Turbo DNA polymerase and the QuickChange Site-Directed Mutagenesis kit were purchased from Stratagene (La Jolla, CA). All other reagents were obtained as noted in the text. Preparation of Fluorescein-BSA Conjugate. FITC (fluorescein-5-isothiocyanate, isomer I) was covalently coupled to bovine serum albumin (BSA) using the FluoReporter FITC proteinlabeling kit according to the manufacturer’s instructions. Briefly, FITC was reacted with BSA at a molar ratio of 10:1 in 0.1 M sodium bicarbonate buffer (pH 8.0) for 2 h at room temperature. (13) Ziauddin, J.; Sabatini, D. M. Nature 2001, 411, 107-110. (14) Wu, R. Z.; Bailey, S. N.; Sabatini, D. M. Trends Cell Biol. 2002, 12, 485488. (15) Delehanty, J. B.; Shaffer, K.; Lin B. Biosens. Bioelectron. 2004 (in press). (16) Bailey, S. N.; Wu, R. Z.; Sabatini, D. M. Drug Discovery Today 2002, 7, S113-S118.
7324 Analytical Chemistry, Vol. 76, No. 24, December 15, 2004
Figure 1. 4-4-20 scFv domain orientation and transfected cell microarray format. (A) In the 4-4-20 scFv, the light chain variable (VL) and heavy chain variable (VH) domains are joined by a 25-amino acid linker. The scFv is expressed as a fusion with a C-terminal mycepitope tag and a PDGF receptor transmembrane domain (PDGFRTM). (B) In the transfected cell microarray format, scFv plasmid DNA is mixed with transfection reagent and deposited in a gelatin/sucrose printing buffer onto the surface of a glass slide. A multiple-well silicone well form is then aligned on the printed plasmid DNA array. The addition of cells to the printed array results in the transfection of cells in spatially defined locations.
Unincorporated FITC was removed by buffer exchange using a Microcon-30 membrane filter device (Millipore, Bedford, MA). The dye/protein ratio was determined by spectrophotometry to be 3:1. Molecular Cloning of Anti-Fluorescein Single-Chain Antibodies and Site-Directed Mutagenesis. The anti-fluorescein single-chain antibody 4-4-20 (scFv 4-4-20) construct employed in these studies was the 4-4-20/205c clone.17 In this construct, a 25amino acid linker joins the C terminus of the light-chain variable region to the N terminus of the heavy-chain variable region. The coding sequence of 4-4-20/205c was amplified by polymerase chain reaction (PCR) with Pfu DNA polymerase from the vector pRS316 prepro/MBP/ 4-4-20/Aga2 (a gift from Dr. Robert Siegel) using the primers 4-4-20 BACK-SfiI and 4-4-20 FOR-SalI. (The sequences of all primers used in this study are given in the Supporting Information). This amplification appended the appropriate restriction sites for cloning into the SfiI-SalI site of the vector pDisplay (Invitrogen Life Technologies, Carlsbad, CA). Upon ligation into the pDisplay plasmid, the 3′-end of the scFv coding sequence was placed in frame with both a c-myc epitope tag for detection and the platelet-derived growth factor receptor transmembrane domain for display on the cell surface. The domain orientation of the expressed scFv 4-4-20 protein is shown schematically in Figure 1A. Mutant clones were generated using the Quick Change SiteDirected Mutagenesis Kit (Stratagene, La Jolla, CA). All sequences were confirmed by automated DNA sequencing in an institutional facility. Sample Preparation and Microarray Fabrication. All microarrays were printed using a BioChip Arrayer I noncontact microarray instrument (Perkin-Elmer Life Sciences, Boston, MA) housed in an environmentally isolated chamber with a relative (17) Denzin, L.; Whitlow, M.; Voss, E., Jr. J. Biol. Chem. 1991, 266, 1409514103.
humidity of 35-45% at room temperature. To prepare arrays for transfection, samples (at a final volume of 50 µL) were prepared in the wells of a 96-well medium-binding plate by diluting 0.162.0 µg of plasmid DNA in EC buffer. Enhancer reagent was added to achieve a final ratio of 6:1 (µL of reagent/µg plasmid), and the mixture was incubated for 5 min at room temperature. Effectene reagent was then added to obtain a final ratio of 6:1 (µL of reagent: µg plasmid) and incubated for 20 min at room temperature. To each sample, an equal volume of 2× transfection printing buffer (0.2% gelatin and 0.2 M sucrose in water) was added to achieve final concentrations of 0.1% gelatin and 0.1 M sucrose in a final volume of 100 µL. The arrayer was programmed to deliver solutions at 1.5 nL per spot at a spacing of 650 µm between spots to the surface of GAPS II microscope slides. As an internal control, a row of spots encoding the red fluorescent protein, DsRed2, was deposited to monitor the transfection efficiency. As shown in Figure 1B, the sample spots were deposited in horizontal rows across the surface of the slide for later alignment with the wells of a silicone well form. For all printing operations, a rinse step was performed between each sample aspiration/dispense cycle in which the tips were rinsed liberally in a sonicating bath containing 0.2% Tween-20 in distilled water. This was followed by a final rinse with distilled water. To control for intercapillary variability, all printing operations were performed using the same glass capillary tip. After printing, the slides were stored at 4 °C in a desiccator, and transfections were typically performed within 1-2 days. Cell Culture, Microarray Transfection and Assays. At the time of transfection, the printed slides were equilibrated to room temperature, and a Press-to-Seal silicone well form was affixed to the slide such that the arrays were centered in the wells (Figure 1B). The assembly was then placed between two pieces of plexiglass that were screwed together to provide constant pressure on the assembly to prevent the leakage of media during incubation. HEK 293T/17 cells were cultured at 37 °C in a 5% CO2 atmosphere in DMEM supplemented with L-glutamine, sodium bicarbonate, sodium pyruvate, 1% antibiotic/antimycotic, and 10% FBS. Cells were trypsinized from tissue culture flasks and pooled, and an aliquot of cells (150 µL at a concentration of 2.6 × 105 cells/mL) was added to replicate wells, and the slides were incubated for 24 h. After 24 h, the transfected slides were removed from the incubator and assayed for both the level of scFv expression and fluorescein binding activity. For the detection of expressed scFv, the arrays were incubated for 1 h with an anti-myc polyclonal antibody diluted to a final concentration of 5 ug/mL in complete growth medium. For the detection of fluorescein binding activity, a fluorescein-BSA conjugate was diluted to the effective working fluorescein concentration in complete growth medium and incubated with the array for 1 h. For all assays, at the end of the incubation period, both the plexiglass assembly and the silicone form were removed prior to the removal of culture media. The slides were then rinsed in PBS (warmed to 37 °C) for 10 s and fixed in 4% paraformaldehyde in PBS for 30 min. This was followed by two rinses in PBS (room temperature) for 10 min each. Finally, 2 drops of Vectashield mounting medium containing DAPI were added to the slide, which was then covered with a coverslip and sealed with nail polish. All tissue culture and handling of cellular
microarrays and all contaminated disposables (pipet tips, used microarray slides, etc.) were performed in compliance with Biosafety Level 2 safety regulations. Image Collection and Determination of Apparent Dissociation Constants. Individual images of transfected cell spots were collected using a Nikon E800 Eclipse fluorescence microscope equipped with fluorescein and rhodamine filter sets. Within each isolated array area, control spots were included to monitor the transfection efficiency within each array. The quantification of the transfection efficiency was performed as described previously.15 Briefly, within each control spot, the number of transfected cells was determined by counting those cells expressing the red fluorescent protein, DsRed2, localized to the nucleus. The total number of cells present on each spot was determined by counting the number of DAPI-stained nuclei. Thus, the transfection efficiency was determined as the number of transfected (red) nuclei divided by the total (blue) nuclei. For the quantification of the relative expression levels of cell-surface displayed scFvs and for the determination of apparent dissociation constants, images of cell clusters consisting of 80-100 cells were collected, and the fluorescence intensities within individual images were determined using MCID Analysis image analysis software (version 7, Imaging Research, Inc., Ontario, Canada). The total fluorescence signal within each image was determined as the product of the average pixel intensity and the total number of pixels within the image minus the fluorescence intensity of the background. For the determination of apparent dissociation constants, the fluorescence intensity values were determined at each fluorescein concentration, and each value was transformed into a corresponding value of percent maximum fluorescence relative to the fluorescence value corresponding to a saturating concentration of fluorescein. The percent maximum fluorescence was plotted versus the fluorescein concentration, and the apparent dissociation constant was determined from the following equation,
% maximum fluorescence ) X/(X + Kd)
(1)
where X represents the fluorescein concentration and Kd represents the apparent dissociation constant. Representative images of complete arrays were assembled from individual images using Adobe PhotoShop. RESULTS AND DISCUSSION Wild-Type scFv 4-4-20 and Binding Site Mutants. To investigate the ability of the cellular microarray format to characterize the binding activity of scFvs expressed on the surface of mammalian cells, we used an scFv version of the anti-fluorescein monoclonal antibody, 4-4-20, as a model system. The native antibody binds fluorescein with an equilibrium dissociation constant (Kd) of 2.0 × 10-10 M17 while the scFv form of the antibody exhibits an affinity for the fluorescyl ligand that is within 2- to 3-fold of that of the native binding site. Further, crystallographic analyses and site-directed mutagenesis studies have identified specific residues responsible for maintaining the affinity of the binding site.18 The mutation of (L) W96 to phenylalanine ((L) W96F) results in a 170-fold decrease in the affinity of the (18) Herron, J. N.; He, X. M.; Mason, M. L.; Voss, E. W.; Edmundson, A. B. Proteins 1989, 5, 271-280.
Analytical Chemistry, Vol. 76, No. 24, December 15, 2004
7325
Figure 2. Representative transfected cell microarray for the expression of a single scFv. (A) Expression of wild-type scFv 4-4-20. Plasmid DNA encoding wild-type 4-4-20 scFv was deposited at amounts of 0, 2.5, 5, 10, 20, and 30 pg, as indicated. Upon the addition of cells, expression proceeded for 24 h. After exposure of the array to a fluorescein-labeled goat anti-myc antibody, the slides were washed and fixed. Individual images were collected using a fluorescence microscope. The row labeled “control” depicts replicate spots of cells expressing the red fluorescent protein, DsRed2, localized to the nucleus to monitor the transfection efficiency. (B) Dose-response curve for the expression of wild-type scFv 4-4-20. Fluorescence values were plotted for the replicate spots at each plasmid DNA amount. Each data point represents the mean ( SEM for 10 replicate spots on replicate slides. For each data set, a 3-parameter curve fit function (SlideWrite version 5.01) was used to determine the best line through the data points.
scFv for fluorescein, whereas the mutation of (L) R34 to either lysine ((L) R34K) or histidine ((L) R34H) decreases the affinity of the binding site by 415- and 750-fold, respectively.17-19 Characterization of Wild-Type scFv 4-4-20 Expression. To characterize the cellular microarray format, it was of interest to first determine the array response for the expression of a single scFv. Figure 2A shows an image of a representative array obtained when HEK 293T/17 cells were transfected with plasmid DNA encoding the wild-type scFv 4-4-20. Twenty-four hours after the addition of cells to the printed array, the expressed protein on (19) Denzin L. K.; Gulliver, G. A.; Voss, E. W., Jr. Mol. Immunol. 1993, 30, 1331-1345.
7326
Analytical Chemistry, Vol. 76, No. 24, December 15, 2004
the cell surface was detected using a fluoresein-labeled antibody directed against the myc epitope on the scFv protein. It was apparent that an increase in the amount of deposited DNA resulted in a corresponding increase in the amount of scFv expressed on the cell surface. The fluorescence signals of the replicate spots steadily increased as the amount of deposited plasmid DNA was increased from 2.5 to 10 pg of DNA. The fluorescence resulting from the no-DNA control row (0 pg) was very low (
