Multiphenotypic Whole-Cell Sensor for Viability Screening - Analytical

Here, we describe the fabrication of whole mammalian cell biosensors for the optical monitoring of cell viability. Three phenotypes were examined for ...
0 downloads 0 Views 487KB Size
Download PDF
Anal. Chem. 2005, 77, 7887-7893

Multiphenotypic Whole-Cell Sensor for Viability Screening Laura J. Itle† and Michael V. Pishko*,‡

Department of Chemical Engineering and The Huck Institute for the Life Sciences and Departments of Chemical Engineering, Materials Science & Engineering, and Chemistry, The Pennsylvania State University, University Park, Pennsylvania 16802-4420

Here, we describe the fabrication of whole mammalian cell biosensors for the optical monitoring of cell viability. Three phenotypes were examined for their response to the addition of two model chemotoxins: sodium hypochlorite and sodium azide, and one model biotoxin, concanavalin A. Two sensing platforms containing cells, hydrogel microspheres, or hydrogel arrays, were also explored. Step changes in viability in response to small doses of sodium hypochlorite were seen nearly instantaneously in all cell lines, in solution, microspheres, and microarrays. Linear detection of sodium azide by entrapped hepatocytes was 0-10 µM, whereas the linear detection range for macrophages and endothelial cells was 0-75 µM. Macrophages and hepatocytes have a greater sensitivity, as indicated by a 40% change in fluorescence over the linear range, whereas endothelial cells show only a 15% change in fluorescence over the linear range. Using photoreaction injection molding, we were also able to generate a multiphenotype sensor that enables the measurement of the toxic effect of 100 µg/mL concanavalin A on macrophages and hepatocytes, but not on endothelial cells. Whole-cell biosensors for the detection of toxins and the efficacy for drug candidates during high-throughput drug screening has traditionally been achieved using reporter genes in prokaryotic cells (i.e., bacteria) and simple eukaryotic cells (i.e., yeasts).1-3 Although much success has been achieved using luminescence genes (lux) to organic contaminants such as benzene,4 toluene,4 xylene,4 naphthalene,5,6 3,5-dicholorphenol,7 * To whom correspondence should be addressed. Department of Chemical Engineering, The Pennsylvania State University, 204 Fenske Laboratory, University Park, PA 16802-4420. Phone: (814) 863-4810. Fax: (814) 865-7846. E-mail: [email protected]. † Department of Chemical Engineering and The Huck Institute for the Life Sciences. ‡ Departments of Chemical Engineering, Materials Science & Engineering, and Chemistry. (1) Bousse, L. Sens. Actuators, B 1996, 34, 270-275. (2) D’Souza, S. F. Biosens. Bioelectron. 2001, 16, 337-353. (3) Durick, K.; Negulescu, P. Biosens. Bioelectron. 2001, 16, 587-592. (4) Applegate, B. M.; Kehrmeyer, S. R.; Sayler, G. S. Appl. Environ. Microbiol. 1998, 64, 2730-2735. (5) Burlage, R. S.; Sayler, G. S.; Larimer, F. J. Bacteriol. 1990, 172, 47494757. (6) Heitzer, A.; Malachowsky, K.; Thonnard, J. E.; Bienkowski, P. R.; White, D. C.; Sayler, G. S. Appl. Environ. Microbiol. 1994, 60, 1487-1494. (7) Lagido, C.; Pettitt, J.; Porter, A. J.; Paton, G. I.; Glover, L. A. FEBS Lett. 2001, 493, 36-39. 10.1021/ac051012b CCC: $30.25 Published on Web 11/12/2005

© 2005 American Chemical Society

metal ions such as copper and lead,7 and nutrients such as iron8 and phosphates,9,10 bioluminescent protein expression generally requires the addition of a cofactor or substrate for luminescence.11 Although luminescent reporter genes have been introduced into mammalian cell lines derived from monkeys12,13 and rabbits,14 the necessity of a cofactor or substrate for maximal activity is a distinct disadvantage for a self-contained biosensor. As an alternative to bioluminescent gene systems, green fluorescent proteins (GFPs) have been exploited for real-time detection within cells without the addition of cofactors or substrates.15 GFPs can be used in conjunction with multiphenotypic biosensors to measure the up-regulation of gene expression15,16 and intracellular protein trafficking17 and to selectively target specific organelles.18 However, as with the lux system, the use of a GFP reporting system relies on the ability of researchers to know which specific gene or protein to monitor and then to generate the relevant transfected cell line. Although the specificity generated by this method could be advantageous for screening drugs whose ultimate goal is the up-regulation of a specific protein, it is not necessarily applicable for the determination of cell viability or changes in intracellular environment. The use of small, fluorescent molecules in the development of optically based whole-cell biosensors overcomes problems associated with protein-based systems by labeling ions, organelles, or cytoplasm in a way that does not interfere with normal cell operations.19 Loading of cells with the fluorescent dyes also occurs in under an hour, as opposed to the time and costprohibitive development of gene-based reporter systems, eliminat(8) Durham, K. A.; Porta, D.; Twiss, M. R.; McKay, R. M.; Bullerjahn, G. S. FEMS Microbiol. Lett. 2002, 209, 215-221. (9) Schreiter, P. P.; Gillor, O.; Post, A.; Belkin, S.; Schmid, R. D.; Bachmann, T. T. Biosens. Bioelectron. 2001, 16, 811-818. (10) Dollard, M. A.; Billard, P. J. Microbiol. Methods 2003, 55, 221-229. (11) Rosochacki, S. J.; Matejczyk, M. Acta Microbiol. Pol. 2002, 51, 205-216. (12) de Wet, J. R.; Wood, K. V.; DeLuca, M.; Helinski, D. R.; Subramani, S. Mol. Cell Biol. 1987, 7, 725-737. (13) Stanford, C.; Keller, J.; Solursh, M. J. Dent. Res. 1993, 73, 1061-1071. (14) Deuschle, U.; Pepperkok, R.; Wang, F. B.; Giordano, T. J.; McAllister, W. T.; Ansorge, W.; Bujard, H. Proc. Natl. Acad. Sci. U.S.A. 1989, 86, 54005404. (15) Chalfie, M.; Tu, Y.; Euskirchen, G.; Ward, W. W.; Prasher, D. C. Science 1994, 263, 802-805. (16) Yeh, E.; Gustafson, K.; Boulianne, G. L. Proc. Natl. Acad. Sci. U.S.A. 1995, 92, 7036-7040. (17) Girotti, M.; Banting, G. J. Cell Sci. 1996, 109 (part 12), 2915-2926. (18) Rizzuto, R.; Brini, M.; Pizzo, P.; Murgia, M.; Pozzan, T. Curr. Biol. 1995, 5, 635-642. (19) Johnson, I. Histochem. J. 1998, 30, 123-140.

Analytical Chemistry, Vol. 77, No. 24, December 15, 2005 7887

ing another impediment to the generation of whole-cell biosensors.19 Additionally, the fabrication of whole-cell viability biosensors utilizing calcein AM is further simplified with the encapsulation and patterning of fluorescently labeled cells using photolithography20 and photoreaction injection molding.21 The optically transparent nature of poly(ethylene) glycol, which is used to encapsulate the cells, allows for good transmission of the fluorescent signal and the determination of cell viability as a function of chemical toxin exposure. Here, we demonstrate the use of a commercially available, small, fluorescent molecule, calcein AM, for the real-time monitoring of cellular viability in response to chemical and biological toxins. Instead of relying on the sensing of secondary products,22,23 this sensor is used to directly monitor changes in cell viability in multiple cell lines utilizing two model chemical toxins, sodium azide and sodium hypochlorite. The feasibility of using a multiphenotypic sensor to determine selective viability is explored through the use of a biological toxin, concanavalin A. EXPERIMENTAL SECTION Materials. Poly(ethylene glycol) diacrylate (PEG-DA, MW 575), anhydrous carbon tetrachloride, and n-heptane were obtained from Aldrich Chemical Co. (Milwaukee, WI), and 2-hydroxy-1[4-(hydroxyethoxy)phenyl]-2-methyl-1-propanone (Ciba, Tarrytown, NY) was used as a photoinitiator. Sulfuric acid was purchased from Fisher Scientific (Fair Lawn, NJ). Sodium hypochlorite, sodium azide, and concanavalin A were purchased from Sigma (St. Louis, MO). SV-40-transformed murine hepatocytes, murine peritoneal macrophages, and murine endothelial cells were obtained from American Type Culture Collection (Manassas, VA). For cell culture, Dulbecco’s modified Eagle’s medium (DMEM), fetal bovine serum (FBS), antibiotic/antimycotic solution, dexamethasone, trypsin, and ethylenediaminetetraacetate (EDTA) were purchased from Sigma Chemical Co. (St. Louis, MO). For viability measurements, Trypan blue was obtained from Sigma (St. Louis, MO), and calcein AM was obtained from Invitrogen (Eugene, OR). Calcein AM is a cell permeant dye that is fluorescently active in the presence of esterases in viable cells. Upon cell death, calcein ceases to fluoresce. Cell Culture. SV-40-transformed murine hepatocytes were cultured in DMEM containing 1.0 g/L glucose, 200 nM dexamethasone, 4% FBS, and 1% antibiotic/antimycotic solution, while macrophages and endothelial cells were cultured in DMEM containing 1.0 g/L glucose, 10% FBS, and 1% antibiotic/ antimycotic solution. Cultures were incubated at 37 °C in 5% CO2 and 95% humidified air. Cells were grown to confluence in 75 cm2 polystyrene tissue culture flasks. Confluent hepatocytes and endothelial cells were subcultured every 2-3 days by trypsinization with 0.25% (w/v) trypsin and 0.13% (w/v) EDTA. Confluent macrophages were subcultured every (20) Koh, W.; Revzin, A.; Pishko, M. Langmuir 2002, 18, 2459. (21) Koh, W. G.; Itle, L. J.; Pishko, M. V. Anal. Chem. 2003, 75, 57835789. (22) Lorenzelli, L.; Margesin, B.; Martinoia, S.; Tedesco, M. T.; Valle, M. Biosens. Bioelectron. 2003, 18, 621-626. (23) O’Riordan, T. C.; Buckley, D.; Ogurtsov, V.; O’Connor, R.; Papkovsky, D. B. Anal. Biochem. 2000, 278, 221-227.

7888

Analytical Chemistry, Vol. 77, No. 24, December 15, 2005

2-3 days via cell scraping. Viability of monolayer cultures was monitored using the Trypan blue exclusion method.24 Real-Time Viability Sensor Fabrication. For the detection of cell viability, 5 µM calcein AM (Invitrogen, Eugene, OR) was incubated with monolayer cultures for 1 h. After incubation, cells were detached from tissue culture flasks via trypsinization or scraping, as described above. For step change experiments, cells were allowed to equilibrate in cuvettes until a steady state was reached, and 100 µL of 5.25 wt % sodium hypochlorite solution was added. The relative fluorescence of each sample was determined using a fluorescence spectrometer (Photon Technology International; 75-W mercury lamp; excitation, 470 nm; emission, 505-650 nm). To determine the working range of the dye in response to a model chemical toxin, sodium azide, cells were resuspended in sodium azide solutions of 0.1-1000 µM at concentrations of 0.20.4 × 106 cells/mL. Initial fluorescence emission of spectra of cells was determined using a fluorescence spectrometer (Photon Technology International; 75-W mercury lamp; excitation, 470 nm; emission, 505-650 nm) and monitored for 2 h, effectively monitoring cell death. After the determination of the steady-state time value, cells were incubated at 37 °C for 2 h, and final emission spectra were obtained to determine the working range of the sensor. The percent change in relative fluorescence units was determined utilizing the initial and final emission spectra at 515 nm. Percent change is defined as the difference between the final and initial emission at 515 nm and divided by the initial emission at 515 nm. After initial ranges were determined, cell-based sensors were generated using poly(ethylene) glycol hydrogels. Preparation of Cell-Containing, PEG Hydrogel Spheres for Viability Sensing. Hydrogel spheres were prepared by following a previously described procedure.25 A poly(ethylene) glycol diacrylate (PEG-DA, Sigma, St. Louis, MO) cell-containing precursor solution was extruded through a 16-gauge needle into a graduated cylinder of mineral oil. The precursor droplets were photopolymerized with a 365-nm, 300 mW/cm2 light source (EFOS Ultracure 100ss Plus, UV spot lamp, Mississauga, Ontario). The spheres were collected and washed repeatedly with water. The precursor solution was composed of (1.0-1.5) × 106 cells/ mL resuspended in 90% phosphate buffered saline, 10% PEG-DA (MW 575), and 1% 2-hydroxy-2-methyl-1-phenyl-1-propanone (Darocur 1173, Ciba, Tarrytown, NY) photinitiator. For step change experiments, cells were allowed to equilibrate in cuvettes before 1, 10, or 100 µL of a 5.25% sodium hypochlorite was added. For dose-response curves, spheres containing a single phenotype of cells were resuspended in sodium azide solution for viability monitoring as well as to confirm that entrapment of cells in PEG hydrogel spheres did not significantly alter the detectable range of sodium azide. Encapsulation of Cells in Hydrogel Arrays for Viability Sensing. Cell-containing microarrays were generated as described previously.20,21 In brief, mammalian cells ((1.0-1.5) × 106 cells/ mL) were suspended in a sterile polymer precursor solution containing 90% serum-free DMEM or phosphate buffered saline, 10% poly(ethylene) glycol diacrylate (MW 575, Aldrich Chemical (24) Fong, R.; Kissmeyer-Nielsen, F. Tissue Antigens 1972, 2, 57-63. (25) Russell, R.; Pishko, M.; Gefrides, C.; McShane, M.; Cote, G. Anal. Chem. 1999, 71, 3126-3132.

Co. Milwaukee, WI), and 1% 2-hydroxy-1[4-(hydroxyethoxy) phenyl]-2-methyl-1-propanone (Irgacure 2959, Ciba, Tarrytown, NY) as a photoinitiator. The cell-containing precursor solution was placed on 1 cm × 1 cm glass substrates modified with 3-(trichlorosilyl)propyl methacrylate (TPM) (Fluka Chemicals, Milwaukee, WI) to prevent delamination of the hydrogel microstructures. UV light (365 nm, 300 mW/cm2, 5 s exposure time, EFOS Ultracure 100ss Plus, UV spot lamp, Mississauga, Ontario) was shown through a 500-µm circular photomask (Advanced Reproductions, Andover, MA) to generate an array of cell-containing hydrogels on the silicon substrate. For step change experiments, cells were exposed to minute quantities of a 5.25% sodium hypochlorite solution for 10 minutes. For dose-response curves, arrays containing a single phenotype of cells were resuspended in sodium azide solution of the predetermined range. An array element was imaged every 10 min for 2 h at a constant exposure using an Axiovert Zeiss 200M fluorescence microscope with a mercury light source and a FITC (exciter, 480 ( 20 nm; emitter, 535 ( 25 nm; beam splitter, 505 long band-pass) filter cube (Zeiss, Thornwood, NY). Images were broken down into their color components. The percent change in the mean green intensity was determined. Multiphenotypic arrays of cells for viability sensing were generated as previously described.21 Briefly, poly(dimethyl) siloxane microfluidic channels were generated using SU-8 photoresist on silicon masters. A 10:1 ratio of prepolymer to curing agent was poured over the master and allowed to cure at room temperature for 2 h. PDMS microchannels were removed from the masters and reversibly sealed on TPM-treated glass substrate. The microchannels were filled with polymer precursor solutions containing one of three cell lines, and a photomask was positioned over the channels to generate multiphenotype arrays. Cells preincubated with calcein AM and resuspended in a polymer precursor solution were exposed to UV light (365 nm, 300 mW/ cm2, EFOS Ultracure 100ss Plus, UV spot lamp, Mississauga, Ontario) for 10 s. To determine if we could selectively induce the death of one cell line in the array, multiphenotypic arrays were exposed to 100 µg/mL of concanavalin A for 96 h. Concanavalin A is known to be cytotoxic to hepatocytic and macrophagic cell lines, while not being cytotoxic to endothelial cell lines.26 RESULTS AND DISCUSSION Viability Measurements in Cells in Free Solution. To ensure that calcein AM could be used to monitor dramatic changes in cell viability, macrophages were preincubated in calcein AM for 1 h and then resuspended in phosphate buffered saline. A 1.5-mL portion of a 0.5 × 106 cells/mL cell suspension was placed in a cuvette and allowed to equilibrate. Upon reaching steady state, cells were dosed with 100 µL of a 5.25% sodium hypochlorite solution. Sodium hypochlorite kills cells by releasing free oxygen into the cell, combining with cellular proteins and causing them to denature. Secondarily, the hypochlorite ion also disrupts the lipoprotein structure of the cell membrane, causing cytoplasm leakage and killing the cell. A step change in fluorescence that corresponded to complete cell death was observed within 2 min of dosing with sodium hypochlorite (see Figure 1). After this ability to sense dramatic changes in cell viability was determined, cells were exposed to a second model chemical toxin, (26) Leist, M.; Wendel, A. J. Hepatol. 1996, 25, 948-959.

Figure 1. Macrophages preincubated with calcein AM can be used to detect step changes in cell viability, caused by exposure of cells to 100 µL of 5.25 wt % sodium hypochlorite.

sodium azide, that acts more gradually to diminish cell viability. In general, azide damages cellular mitochondria, effectively blocking the flow of electrons to oxygen and halting the production of ATP in cells. Without ATP production, cell death occurs. To determine if cells containing calcein AM could be used to distinguish between different doses of sodium azide, cells were resuspended in phosphate buffered saline containing differing amounts of sodium azide. The linear range for sodium azide detection was determined for each of three cell lines. Figure 2A shows the linear range of sodium azide detectable by murine macrophages (closed circles) and endothelial cells (open circles). Although macrophages and endothelial cells have similar detection ranges, from 0 to 50 µM, macrophages showed greater sensitivity to azide. The endothelial cells had an overall change in fluorescence of 15%, but macrophages over the same range demonstrated a 40% change in fluorescence. Murine hepatocytes were used to detect much lower sodium azide concentrations than either macrophages or endothelial cells. As shown in Figure 2B, hepatocytes were capable of discriminating between much lower levels of sodium azide, from 0 to 1.0 µM. Moreover, like macrophages, hepatocytes showed a much greater sensitivity, a 40% change in fluorescence intensity over the linear range. The decrease in viability of cells incubated in phosphate buffered saline containing sodium azide was confirmed using the Trypan blue exclusion method (data not shown) to ensure that sodium azide was, in fact, resulting in cell death and not simply disrupting the fluorescent dye. Viability Sensing Using Cells Entrapped in PEG Hydrogel Spheres. Upon determining the range of the model chemical toxin, sodium azide, that was detectable using mammalian cells, cells were entrapped in poly(ethylene) glycol (PEG) spheres. Poly(ethylene) glycol can be used to entrap viable cells to provide a more in vivo-like environment for sensing. Microspheres entrapping cells were used to determine if entrapment of PEG alters the range of sodium azide that is detectable. Before ranges of sodium azide were confirmed, macrophages entrapped in PEG spheres were subjected to dosing with 100 µL of a 5.25% sodium hypochlorite solution after a steady state had been reached. A step change in fluorescence identical to that in cells in solution was achieved (Figure 3). Analytical Chemistry, Vol. 77, No. 24, December 15, 2005

7889

Figure 2. Detectable ranges of sodium azide by calcein-containing cells suspended in phosphate buffered saline. (A) Linear ranges for sodium azide using macrophages (0-50 µM, b) and endothelial cells (0-40 µM, O). (B) Linear range for sodium azide using hepatocytes (0-1 µM, b).

Figure 3. Macrophages preincubated with calcein AM and entrapped in PEG hydrogel spheres can be used to detect step changes in cell viability, caused by exposure of cells to 100 µL of 5.25 wt % sodium hypochlorite.

As with cells in solution, the range of sodium azide detectable for all three cell lines can be seen in Figure 4. Entrapment in PEG hydrogel microspheres increased the detectable range of 7890 Analytical Chemistry, Vol. 77, No. 24, December 15, 2005

Figure 4. Detectable ranges of sodium azide by calcein containing cells in phosphate buffered saline. (A) Linear ranges for sodium azide using macrophages (0-75 µM, b) and endothelial cells (0-75 µM, O). (B) Linear range for sodium azide using hepatocytes (0-10 µM, b).

sodium azide from 0-50 to 0-75 µM in endothelial cells and macrophages (Figure 4A), and entrapment in PEG hydrogel microspheres increased the range of sodium azide detectable from 0-1 to 0-10 µM (Figure 4B) in hepatocytes. Although a slight change in fluorescence might not be detectable when cells are suspended in a saline solution, cells entrapped in hydrogels might experience total internal reflection. It is postulated that the increase in the linear detection range was likely caused by amplification of the fluorescence due to internal reflection within the hydrogel microspheres. As an added advantage, an order of magnitude fewer cells used in the fabrication of microspheres achieved detection limits similar to that of cells in solution. Finally, to determine if online detection of changes in viability could be determined in less than 2 h, PEG microsphere-cell sensors utilizing macrophages were monitored continually for 2 h. A representative graph of real-time monitoring of viability is shown in Figure 5. As shown in Figure 5, differing doses of sodium azide can be differentiated using a PEG-entrapped whole-cell sensor as early as 30 min. This faster detection demonstrated the feasibility of using whole-cell sensors for situations that require more rapid responses. Single Phenotype Arrays for Viability Sensing. To determine if the results obtained using PEG microspheres were

Figure 5. PEG microspheres with cells containing calcein AM were incubated with 0, 50, 75, and 100 µM azide. The fluorescence intensity of the spheres was monitored as a function of time to determine the earliest point at which differing doses of azide could be distinguished using whole-cell sensors. Doses are distinguishable after 30 min.

transferable to a microdevice format, arrays of PEG microstructures containing macrophages were generated. The arrays were subjected to a small dose (100 µL) of 5.25% sodium hypochlorite and monitored to see if arrays could be used to determine step changes in cell viability. Figure 6 shows an array of PEG cylinders containing macrophages before and after exposure to sodium hypochlorite. The results clearly indicate that step changes in viability are visually detectable. Second, an array of PEG microstructures containing macrophages were subjected to varying doses of sodium azide and monitored over time to determine whether it was possible to distinguish between different doses of sodium azide. In Figure 7, the mean green pixel intensity of fluorescence images was monitored over the time course of the experiment and converted to a percent change of pixel intensity. Figure 7 demonstrates that changes in cell viability, as they relate to decreases in fluorescence output, can be measured in a dose-dependent manner. Figure 8 shows single array elements on the whole-cell viability sensor with and without exposure to sodium azide. The diminished calcein AM fluorescence due to cell death caused by sodium azide is clearly visible in these images (the difference in intensity between images C and D as compared to the control). Although distinguishing between arrays that are exposed to no sodium azide and those which are was easily achieved, distinguishing between doses of sodium azide became more difficult; however, this was more likely due to the acquisition of fluorescence images, rather than a limitation of the physical sensor. Fluorescence spectroscopy, in this case as demonstrated by the use of PEG microspheres, provides more sensitive detections of variations than does fluorescence microscopy. However, the use of a more sensitive optical detector would allow for a greater ability to distinguish between different doses. Because, unlike the detection of step changes in viability, differing doses of sodium azide cannot be distinguished by the human, there is a great need for sensitive optics to be coupled with this type of sensor. Multiphenotype Array. Finally, to demonstrate that selectivity is possible using a whole-cell format, a multiphenotype sensor was

Figure 6. Single phenotype array containing macrophages before (A) and after (B) exposure to 100 µL of a 5.25 wt % of a sodium hypochlorite solution at 2.5× magnification.

Figure 7. Single phenotype array containing macrophages exposed to 0, 25, 50, or 75 µM sodium azide. Cells exposed to no sodium azide were readily distinguishable from cells that were exposed to sodium azide. However, differentiating between doses was difficult unless the scale of the graph was enlarged.

created. A picture of the multiphenotype sensor can be seen in Figure 9. From left to right, the cell lines contained with in the sensor are hepatocytes, macrophages, and endothelial cells. The multiphenotypic array was subjected to concanavalin A for 96 h. Each array element was decomposed into its RGB components. The mean green intensity of each cell line was determined. All Analytical Chemistry, Vol. 77, No. 24, December 15, 2005

7891

Figure 8. Fluorescence images showing an initial (A and C) and 2-h time lapse (B and D) images of a single array element upon exposure to 0 µM (A and B) and 100 µM (C and D) sodium azide.

Figure 9. A multiphenotypic microdevice containing (left to right) hepatocytes, macrophages, and endothelial cells stained with calcein AM.

cell lines were compared to determine if concanavalin A was able to selectively damage two of the three phenotypes. These results are presented in Figure 10. As demonstrated in Figure 10, after 6 h, a sharp decrease in cellular viability (as shown by a decrease in percent fluorescence) was achieved in both hepatocytic and macrophagic cell lines, while endothelial cells showed increased fluorescence, an indicator of good cell health. Hepatocytes and macrophages appeared to 7892

Analytical Chemistry, Vol. 77, No. 24, December 15, 2005

recover somewhat in the time between 6 and 96 h; however, endothelial cells remained relatively constant. In the case of hepatocytes, these results may be aniticipated, since hepatocyte injury was shown to stimulate the proliferation of nearby cells,27 and we have previously demonstrated hepatocyte proliferation within these hydrogel matrixes.28 After 96 h, macrophages and (27) Trautwein, C.; Rakemann, T.; Malek, N. P.; Plumpe, J.; Tiegs, G.; Manns, M. P. J. Clin. Invest. 1998, 101, 1960-1969.

Figure 10. Percent change in mean green intensity over time caused by exposure to 100 µg/mL concanavalin A for 96 h. Each cell line in the multiphenotypic microdevice was monitored for decreases in mean green intensity. Hepatocytes and macrophages showed decreases in viability, while endothelial cells showed a relatively constant intensity over time.

hepatocytes showed drops and viability, with endothelial cells remaining at a constant value. These data demonstrate that multiphenotypic arrays have the potential to determine whether a chemical selectively damages a specific cell line. This would be beneficial in determining the efficacy of a drug targeted at destroying a specific cell type in vivo, that is, cancer cells. CONCLUSION The generation of whole-cell biosensors with rapid response times using mammalian cells has been a subject of much interest in recent years. Of particular importance is the application of cellbased microdevices to drug discovery, because cells represent the ultimate target for pharmaceuticals. In this instance, a (28) Itle, L. J.; Koh, W. G.; Pishko, M. V. Biotechnol. Prog. 2005, 21, 926-932.

multiphenotypic biosensor would allow systemic information to be obtained for a drug candidate in a one-pass system. When coupled with a microfluidic delivery device, small quantities of a drug candidate could be tested on multiple cell lines simultaneously to gauge a drug’s ADMET (adsorption, distribution, metabolism, elimination, and toxicology) properties using the platform of macrophages, hepatocytes, and endothelial cells described in this paper, liver function, immune response, and inflammation response in a single assay step, all parts of the ADMET response. Overall toxicity on multiple cell lines can also be gauged using this type of sensor. This would provide a distinct advantage in screening chemotherapeutic effectiveness in the treatment of cancer. A multiphenotypic sensor could be used to test drug toxicity on a cancer cell line as well as cell lines representing the surrounding tissue. Here, we have presented a whole-cell mammalian sensor using whole-cell staining, coupled with entrapment of the cells in an easy to fabricate and use optical platform. The whole-cell sensors have the ability to exist in two physical formats, as hydrogel microspheres or in array formats. Using hydrogel encapsulation, we can detect sodium azide, a model chemotoxin, in micromolar quantities in under 30 min. By using the microdevice format, we have been able to successfully generate a multiphenotypic chip that enables one to measure the toxic effect of concanavalin A on two cell types, but not on the third. In the future, this work can be expanded to included more cell phenotypes and dyes sensitive for a variety of other internal conditions, such as nitric oxide production, intracellular pH, or caspase production. ACKNOWLEDGMENT The authors thank NASA Biotech-01-0023-0131 for funding this research.

Received for review June 8, 2005. Accepted October 19, 2005. AC051012B

Analytical Chemistry, Vol. 77, No. 24, December 15, 2005

7893