Vacuum-Assisted Cell Seeding in a Microwell Cell Culture System

Anal. Chem. 2010, 82, 2380–2386

Vacuum-Assisted Cell Seeding in a Microwell Cell Culture System Nicholas Ferrell, Daniel Gallego-Perez, Natalia Higuita-Castro, Randall T. Butler, Rashmeet K. Reen, Keith J. Gooch, and Derek J. Hansford* Department of Biomedical Engineering and Department of Materials Science and Engineering, Ohio State University, Columbus, Ohio 43210 We present a simple method to actively pattern individual cells and groups of cells in a polymer-based microdevice using vacuum-assisted cell seeding. Soft lithography is used to mold polymer microwells with various geometries on top of commercially available porous membranes. Cell suspensions are placed in a vacuum filtration setup to pull culture medium through the microdevice, trapping the cells in the microwells. The process is evaluated by determining the number of cells per microwell for a given cell seeding density and microwell geometry. This method is tested with adherent and nonadherent cells (NIH 3T3 fibroblasts, PANC-1 pancreatic ductal epithelial-like cells, and THP-1 monocytic leukemia cells). These devices could find applications in high-throughput cell screening, cell transport studies, guided formation of cell clusters, and tissue engineering. The ability to spatially manipulate cell location and morphology with a high degree of control is important for many cell biology applications. Isolation of individual cells is critical in single-cell electroporation,1,2 planar patch clamping,3,4 and cell-based biosensors.5-7 It is also well-established that cell morphology plays a critical role in a variety of cellular processes including proliferation, differentiation, apoptosis, and motility.8-11 A device with the ability to simultaneously manipulate the spatial arrangement and morphology of single cells or small groups of cells could therefore be used in a wide range of biological applications. In addition, guided assembly of three-dimensional (3D) microtissue subunits has been useful in cell therapy, tissue engineer* To whom correspondence should be addressed. Phone: 614-688-3449. Fax: 614-292-7301. E-mail: [email protected]. (1) Huang, Y.; Rubinsky, B. Sens. Actuators, A 2001, 89, 242–249. (2) Khine, M.; Lau, A.; Ionescu-Zanetti, C.; Seo, J.; Lee, L. P. Lab Chip 2005, 5, 38–43. (3) Klemic, K. G.; Klemic, J. F.; Reed, M. A.; Sigworth, F. J. Biosens. Bioelectron. 2002, 17, 597–604. (4) Li, X.; Klemic, K. G.; Reed, M. A.; Sigworth, F. J. Nano Lett. 2006, 6, 815– 819. (5) Wang, P.; Xu, G.; Qui, L.; Xu, Y.; Li, Y.; Li, R. Sens. Actuators, B 2005, 108, 567–584. (6) Nishizawa, M.; Matsue, T. Langmuir 2002, 18, 3645–3649. (7) Chen, C. S.; Mrksich, M.; Huang, S.; Whitesides, G. M. Biotechnol. Prog. 1998, 14, 356–363. (8) Chen, C. S.; Mrksich, M.; Huang, S.; Whitesides, G. M.; Ingber, D. E. Science 1997, 276, 1425–1428. (9) Ito, Y. Biomaterials 1999, 20, 2333–2342. (10) Brock, A.; Chang, D.; Ho, C.; LeDuc, P.; Jiang, X.; Whitesides, G. M.; Ingber, D. E. Langmuir 2003, 19, 1611–1617. (11) Bhatia, S. N.; Chen, C. S. Biomed. Microdevices 1999, 2, 131–144.

2380

Analytical Chemistry, Vol. 82, No. 6, March 15, 2010

ing, cellular and developmental biology, drug discovery, and other applications. Many cell types (e.g., hepatocytes, pancreatic islet cells, embryonic stem cells, and cardiomyocytes) exhibit more in vivo-like behavior when cultured in 3D aggregates as opposed to traditional 2D culture conditions.12-16 Geometrical factors such as aggregate size and morphology have been known to influence cell responses.17-19 The ability to control aggregate formation in a high-throughput manner could have significant implications in a number of biotechnology applications. Microfabrication is an attractive approach for interacting with individual and small groups of cells because of the ability to fabricate structures and devices on the same size scale as cells. Microcontact printing (µCP)20 has been widely used to locally alter surface chemistry for cell patterning. µCP is a soft lithography process that uses a poly(dimethylsiloxane) (PDMS) stamp “inked” with various chemistries including self-assembled monolayers (SAM).21,22 The stamp is then brought into contact with the surface of interest to transfer the chemical pattern. The spatial difference in the surface chemistry alters subsequent protein adsorption, which leads to preferential cell attachment. Cell adhesion proteins such as fibronectin have also been directly patterned on surfaces to alter cell attachment.6,23 In addition to alteration of surface chemistry, a combination of surface topography and chemical modifications have been used to pattern adherent cells.24,25 These techniques rely on adhesive versus (12) Kelm, J. M.; Fussenegger, M. Trends Biotechnol. 2004, 22, 195–202. (13) Khademhosseini, A.; Langer, R.; Borenstein, J.; Vacanti, J. P. Proc. Natl. Acad. Sci. U.S.A. 2006, 103, 2480–2487. (14) Khalil, M.; Shariat-Panahi, A.; Tootle, R.; Ryder, T.; McCloskey, P.; Roberts, E.; Hodgson, H.; Selden, C. J. Hepatol. 2001, 297, 68–77. (15) Boretti, M. I.; Gooch, K. J. Tissue Eng., Part A 2008, 14, 1927–1937. (16) Boretti, M. I.; Gooch, K. J. Tissue Eng. 2006, 12, 939–948. (17) Glicklis, R.; Merchuk, J. C.; Cohen, S. Biotechnol. Bioeng. 2004, 86, 672– 680. (18) Karp, J. M.; Yeh, J.; Eng, G.; Fukuda, J.; Blumling, J.; Suh, K. Y.; Cheng, J.; Mahdavi, A.; Borenstein, J.; Langer, R.; Khademhosseini, A. Lab Chip 2007, 7, 786–794. (19) Dang, S. M.; Kyba, M.; Perlingeiro, R.; Daley, G. Q.; Zandstra, P. W. Biotechnol. Bioeng. 2002, 78, 442–453. (20) Xia, Y.; Whitesides, G. M. Annu. Rev. Mater. Sci. 1998, 28, 153–184. (21) Mrksich, M.; Dike, L. E.; Tien, J.; Ingber, D. E.; Whitesides, G. M. Exp. Cell Res. 1997, 235, 305–313. (22) Kane, R. S.; Takayama, S.; Ostuni, E.; Ingber, D. E.; Whitesides, G. M. Biomaterials 1999, 20, 2363–2376. (23) Kaji, H.; Takoh, K.; Nishizawa, M.; Matsue, T. Biotechnol. Bioeng. 2003, 81, 748–751. (24) Suh, K. Y.; Seong, J.; Khademhosseini, A.; Laibinis, P. E.; Langer, R. Biomaterials 2004, 25, 557–563. (25) Dusseiller, M. R.; Schlaepfer, D.; Koch, M.; Kroschewski, R.; Textor, M. Biomaterials 2005, 26, 5917–5925. 10.1021/ac902596b  2010 American Chemical Society Published on Web 02/24/2010

nonadhesive areas to achieve cell patterning. Although this approach is good for certain applications, it is not amenable to isolation of nonadherent cells or formation of 3D cell aggregates. In the case of aggregate formation, it is not desirable to have cells on strongly adhesive surfaces since this alters cell differentiation and affects cluster formation.16 Several other devices have also been fabricated for patterning individual cells or groups of cells.26-31 Elastomeric stencils have been used to pattern groups of adherent cells.27 Cells were patterned by culturing them on a PDMS membrane with throughholes and removing the membrane after cell attachment. Revzin et al.28,29 patterned cell-repelling poly(ethylene glycol) (PEG) microwells with surface modified glass bottoms to selectively adhere 3T3 fibroblasts30 and leukocytes.31 Rettig and Folch30 developed a system for trapping cells in PDMS microwells and studied the effects of microwell geometry and settling time on the efficiency of trapping. Rosenbluth et al.31 used microfabricated SU-8 wells for isolating individual nonadherent cells for atomic force microscopy (AFM) measurements, for which the microwells had to be relatively shallow (11 µm) to allow easy access to the cells by the AFM tip. Fukuda and Nakazawa32 used micromilled polystyrene wells of various diameters to create hepatocyte spheroids of different sizes. All these systems rely on passive and often lengthy loading of cells via sedimentation into the wells. In addition, for achieving high trapping efficiency, cell seeding densities in large excess of the number of microwells are required. We have developed a simple method of vacuum-assisted cell seeding for active placement of adherent and nonadherent cells in microwells with defined geometries. The method was evaluated with adherent cells (NIH 3T3), nonadherent cells (THP-1), and aggregate-forming cells (PANC-1). This method provides several advantages over available cell trapping techniques, including fast seeding times, minimal cell loss, and improved nutrient/waste exchange. In addition, the device can easily be integrated into two-chamber membrane-based cell culture systems to facilitate the study of important cellular parameters such as transcellular drug and solute transport, chemotaxis, secretion, and absorption. This device could find applications in high-throughput cell screening, cell-based biosensor arrays, guided assembly of microtissues, and tissue engineering. EXPERIMENTAL SECTION Device Fabrication. Microwells were fabricated from either poly(propyl methacrylate) (PPMA, Scientific Polymer Products, 150 000 MW), poly(methyl methacrylate) (PMMA, Scientific Polymer Products, 75 000 MW), or polystyrene (PS, Aldrich, melt flow index 4.0). Polymer solutions were made by dissolving the polymers in anisole (Sigma-Aldrich) at various concentrations depending on the desired microwell depth (thickness of the microwell layer). Commercially available micro- and nanoporous (26) Jager, E. W. H.; Immerstrand, C.; Peterson, K. H.; Magnusson, K.; Lundstro ¨m, I.; Ingana¨s, O. Biomed. Microdevices 2002, 4, 177–187. (27) Folch, A.; Jo, B.; Hurtado, O.; Beebe, D. J.; Toner, M. J. Biomed. Mater. Res. 2000, 52, 346–353. (28) Revzin, A.; Tompkins, R. G.; Toner, N. Langmuir 2003, 19, 9855–9862. (29) Revzin, A.; Sekine, K.; Sin, A.; Tompkins, R. G.; Toner, M. Lab Chip 2005, 5, 30–37. (30) Rettig, J. R.; Folch, A. Anal. Chem. 2005, 77, 5628–5634. (31) Rosenbluth, M. J.; Lam, W. A.; Fletcher, D. A. Biophys. J. 2006, 90, 2994– 3003.

membranes were used as underlying supports. Track-etched polycarbonate membranes (Isopore, Millipore) with 200 and 400 nm diameter pores, anodized alumina membranes (Anodisc, Whatman) with 20 and 200 nm pores, and polyester membranes (Transwell, Corning) with 400 nm pores were all used for different isolation experiments. Transwell membranes were used for the majority of the experiments in this study because they offered the advantage of being optically transparent. This allowed cell observation using an inverted phase contrast microscope. Microwells were fabricated using a two-step soft lithographic micromolding process combined with spin-dewetting.34 The geometry of the structure was first defined in photoresist using standard photolithography. Negative tone SU-8 25 and SU-8 2075 (Microchem Corporation) and positive tone SPR 220-7 (Shipley) photoresists were used in the fabrication. Photoresists were spincoated on clean silicon at 3000 rpm for SU-8 25 and 2000 rpm for SU-8 2075 and SPR 220-7. The resulting film thickness was approximately 9 µm for SU-8 25, 100 µm for SU-8 2075, and 8 µm for SPR 220-7. The photoresist films were exposed to ultraviolet light through chrome/glass photomasks with various geometries, and films were processed according to the parameters suggested by the manufacturer. The photolithographically patterned photoresist features on silicon wafers were used as masters for the replication of a PDMS mold. The mold was made by mixing a 10:1 (w/w) ratio of PDMS base with curing agent (Silastic T-2, Dow Corning). The base and curing agent were mixed thoroughly and poured over the patterned wafer. The PDMS was degassed in a vacuum desiccator until all bubbles were removed. The PDMS was allowed to cure at room temperature for 48 h before removing the mold. Figure 1a-g shows the membrane fabrication process after making the PDMS mold. The mold was spin-coated with a solution of PPMA, PMMA, or PS in anisole. Solutions of 15% (w/w) were used on the 8 µm deep molds, 20% (w/w) solutions were used on 9 µm deep molds, and 25% (w/w) were used on the 100 µm deep molds. Solutions were spin-coated at 3000 rpm for 1 min. After spin-coating, dewetting of the polymer was observed, as the polymer separated to form independent particles on top of the pillar structures of the PDMS molds. The particles were removed by contacting the polymer with a glass slide at 180 °C (PPMA and PMMA) or 200 °C (PS) using low pressure (∼17 KPa). This is shown in Figure 1, parts c and d. After removal of the surface material, the mold was brought into conformal contact with the membrane. The membrane was then placed on a hot plate at 95 °C (PPMA) or 125 °C (PMMA and PS), and a pressure of ∼415 KPa was used to transfer the remaining polymer onto the membrane. The resulting structure consisted of geometrically uniform through-holes with unobstructed access to the underlying membrane. Cell Culture. NIH 3T3 mouse fibroblasts and human PANC-1 cells were maintained in Dulbecco’s modified Eagle’s medium (DMEM). NIH 3T3 cells were supplemented with 10% fetal calf serum, and PANC-1 cells were supplemented with 10% fetal bovine serum (FBS). Human THP-1 monocytes were cultured in RPMI 1640 medium supplemented with 10% FBS. Penicillin (100 IU/ mL) and streptomycin (100 µg/mL) were added to all culture media. All cell lines and reagents were obtained from the American (32) Fukuda, J.; Nakazawa, K. Tissue Eng. 2005, 11, 1254–1262.


2381

Figure 1. (a-g) Schematic diagram of the membrane fabrication process: (a) uncoated PDMS mold; (b) mold spin-coated with polymer (particles on the top of the pillars form by spin-dewetting); (c and d) the coated mold is brought into contact with heated glass to remove the particles; (e) selectively coated mold; (f) the mold is inverted and brought into contact with the heated filter by applying pressure; (g) the final structure after removing the mold. (h and i) Experimental setup for cell isolation: (h) the membrane is placed in vacuum filtration setup; (i) a cell suspension is loaded in the upper chamber, and gentle vacuum (17-85 kPa) is used to pull the cells into the wells.

Type Culture Collection (ATCC) unless otherwise noted. Cells were incubated at 37 °C in a humidified atmosphere with 5% CO2. Cell Trapping. Cells were trapped in the microwells using a glass microanalysis vacuum filtration setup (Fisher Scientific). A schematic of the experimental setup is shown in the Figure 1, parts h and i. The microwell membrane was clamped between a glass funnel and the underlying vacuum filter flask as shown in Figure 1h. A cell suspension in culture medium, with the desired cell concentration, was placed in the funnel on top of the device, and vacuum was applied for 0.5-5 min to pull the cell culture medium through the membrane, trapping the cells in the microfabricated wells. Cell seeding time depended upon several factors, including cell density, suspension volume, vacuum pressure, and microdevice porosity. This process is shown in Figure 1i. After filtration, the membranes were removed from the setup and cultured for different time periods with the exception of the THP-1 cells, which were immediately prepared for fluorescence microscopy. Cell seeding density was chosen based on cell type and microwell size and density. For 3T3 cells, seeding densities ranged from 50 to 950 cells/mm2. For THP-1, a seeding density of 340 cells/mm2 was used. PANC-1 cells were vacuum-seeded on PMMA and PS microwell arrays (300 and 500 µm diameter) stamped on anodized alumina filters at a density of 1600 cells/ mm2. 3T3 cells were cultured for 1-2 h (to allow adhesion) under normal conditions after filtration. To induce cluster formation, PANC-1 were cultured for 4 days using serum-free medium, which is a combination of DMEM/F12 supplemented with 17.5 mM glucose, 1% albumin bovine fraction V fatty acid free (no. 152401, ICN), and 1× insulin-transferrin-selenium (GIBCO).35,36 (33) Park, J.; Cho, C. H.; Parashurama, N.; Li, Y.; Berthiaume, F.; Toner, M.; Tilles, A. W.; Yarmush, M. L. Lab Chip 2007, 7, 1018–1028. (34) Ferrell, N.; Hansford, D. Macromol. Rapid Commun. 2007, 28, 964–967.

2382


Scanning Electron Microscopy. Microwell devices (before and after cell seeding) were characterized using scanning electron microscopy (SEM, Hitachi S3000H). Devices with no cells were sputter-coated with a thin layer of gold-palladium prior to imaging. Devices with cells were prepared for SEM by dehydration in graded ethanol solutions (70%, 80%, 90%, and 100%) and hexamethyldisilazane according to the procedure described by Braet et al.37 Fluorescence Microscopy. For fluorescence microscopy, cells were fixed in 70% ethanol for 30 min at -20 °C. Cells were then stained with propidium iodide RNase (PI RNase, BD Biosciences) for 10 min at 4 °C. Samples were washed three times in phosphate-buffered saline (PBS) and placed on glass slides for observation. Samples were imaged on an inverted fluorescence microscope (Nikon TS 100). Image Analysis. The number of cells in each microwell was determined by analyzing the fluorescence images. NIH 3T3 cells were seeded in 20 and 50 µm diameter wells and cultured for 2 h. Three seeding densities were used for each well diameter with the density expressed as the number of cells per unit of sample surface area. For 50 µm wells, cells were seeded at 100, 200, and 300 cells/mm2. For 20 µm wells, cells were seeded at 300, 450, and 600 cells/mm2. The densities tested were determined by the microwell density on the device, which is defined by the geometrical parameters of the patterned features (e.g., shape, size, and spacing). The number of cells in each well was counted manually. For the 50 µm wells, five images were analyzed for each of three samples with counts of 48 (35) Hardikar, A. A.; Marcus-Samuels, B.; Geras-Raaka, E.; Raaka, B. M.; Gershengorn, M. C. Proc. Natl. Acad. Sci. U.S.A. 2003, 100, 7117–7122. (36) Gershengorn, M. C.; Hardikar, A. A.; Wei, C.; Geras-Raaka, E.; MarcusSamuels, B.; Raaka, B. M. Science 2004, 306, 2261–2264. (37) Braet, F.; De Zanger, R.; Wisse, E. J. Microsc. 1997, 186, 84–87.

Figure 2. (a-c) SEM micrographs of cell isolation membranes. Feature geometry and size are (a) 20 µm diameter circles, (b) 30 µm squares, and (c) 20 µm (side to side) hexagons. (d-i) NIH 3T3 cells trapped in the microwells at lower (d-f) and higher (g-i) magnifications.

wells/image (240 wells total per sample). For the 20 µm wells, 80 wells/image were counted (400 wells total per sample). RESULTS AND DISCUSSION Scanning electron micrographs of several microwell geometries used for patterning 3T3 cells are shown in Figure 2a-c. The features shown are (Figure 2a) 20 µm circles, (Figure 2b) 30 µm squares, and (Figure 2c) 20 µm hexagons (side to side). 3T3 cells were also patterned in 50 µm diameter circular microwells. Additional geometries (not shown) were used for THP-1 cells (15 µm diameter circular features) and PANC-1 cells (300 and 500 µm diameter circular features). Parts a and b of Figure 2 are microwells on Isopore membranes, and (Figure 2c) shows microwells on an Anodisc membrane. It is critical that unobstructed access to the pores is maintained throughout the fabrication process to provide an open pathway through the device for vacuum filtration of the cells. Figure 2d-i presents scanning electron micrographs of NIH 3T3 cells patterned in the devices. Figure 2d shows one NIH 3T3 cell in each of four 20 µm circular wells. Figure 2e shows cells patterned in 30 µm squares, with multiple cells observed in the upper left and lower right features, whereas single cells are observed in the other two features. Figure 2f shows cells in two hexagonal wells with two wells empty. Figure 2g-i shows highmagnification images of individual cells patterned in three different microwell configurations. The morphology of the cells conformed

to the geometry of the microwell. Control of cell morphology was further confirmed by SEM of the sample after removal of the microwell layer (see Supporting Information Figure S-1) which shows that the cells conformed to the geometry of the microwells. Figure 3 shows fluorescent micrographs of the cells with overlaid phase contrast images to show the microwells. Parts a and b of Figure 3 show NIH 3T3 cells in 20 and 50 µm diameter circular wells, respectively. Nonadherent THP-1 cells were isolated in 15 µm circular wells as shown in Figure 3c. The ability to isolate nonadherent cells is significant, given the difficulties presented in manipulating these cells. For nonadherent cells, the surface chemistry approaches cannot be easily employed for isolation and positioning. Some passive techniques have been used for isolating nonadherent cells,31 but to our knowledge this is the first active mechanism that has been employed for isolation of single and multiple adherent and nonadherent cell types. Figure 4 shows fluorescence microscopy images of the PANC-1 cells after 4 days of culture. Vacuum-assisted seeding successfully placed the cells inside the wells. After 4 days the cells remained confined in the wells and started to show evidence of cluster formation as indicated by confocal fluorescence microscopy (see Supporting Information Figure S-2). Previous reports have shown that these cells transdifferentiate when cultured in 3D clusters, with dedifferentiation from their original lineage accompanied by a differentiation toward an insulin-producing β cell phenotype.32,33 Analytical Chemistry, Vol. 82, No. 6, March 15, 2010

2383

Figure 4. Fluorescence images of PANC-1 cell aggregates in (a) 300 µm diameter wells and (b) 500 µm diameter wells.

Figure 3. Fluorescence/phase contrast merged images of (a) 20 µm diameter wells with NIH 3T3 cells, (b) 50 µm diameter wells with NIH 3T3 cells, and (c) 15 µm diameter wells with THP-1 cells.

Many factors, including the number of seeded cells, well size, and chemical and physical properties of the underlying porous substrate, can be easily altered to control cell clustering. Functional studies of the cell clusters are outside the scope of this article and will be addressed in future work. Cell viability was evaluated with PANC-1 cells to determine if the vacuum pressure used in the seeding process was detrimental to the cells. Cells were seeded at 17, 51, and 85 kPa. No significant loss of viability was observed at any of the vacuum pressures used (Supporting Information Figure S-3). Even at the lowest pressure used, cells were confined to the microwells. Although no major loss of viability was observed for any of the vacuum pressures used, it may be advisable to use low vacuum pressure if the 2384


process were used with more sensitive cell lines such as primary cultures or stem cells. Analysis of the fluorescent images using two different microwell sizes and three different cell seeding densities was used to evaluate the cell trapping with NIH 3T3 cells. In Figure 5, the number of cells in each well for a specific seeding density and microwell diameter is given. The data is presented as a percentage of the total number of available microwells. The average percentage for the three samples is given with the error bar representing the negative intersample standard deviation. For the 20 µm diameter wells seeded at the lowest density tested (300 cells/ mm2), an average 41% of the wells were empty and about 45% of the wells had one cell each. By doubling the seeding density to 600 cells/mm2, 50% of the wells had one cell, and nearly 85% had one or more cells per well. No wells had more than three cells for any of the seeding densities tested. Figure 5b shows the data for the 50 µm diameter wells. Again, at the lowest seeding density tested (100 cells/mm2), a relatively large percentage of wells were empty (46%). At 200 cells/mm2, the number of empty wells decreased to 24%, with 39% having one cell per well and the remaining wells having two to four cells per well. With a seeding density of 300 cells/mm2, almost 95% of the wells were occupied by at least one cell. A large proportion of the wells (74%) had one to three cells per well with 18% having four or five cells per well. About 2% of the wells had more than five cells per well. Although additional optimization of the seeding conditions could have been performed based on a wide range of experimental parameters, the

Figure 5. Number of NIH 3T3 cells per microwell as a percentage of the total number of microwells: (a) cells seeded at 300, 450, and 600 cells/mm2 in 20 µm diameter circular microwells; (b) cell seeded at 100, 200, and 300 cells/mm2 in 50 µm diameter circular microwells.

data shows that the number of cells in each microwell can be reasonably well-controlled by adjusting the diameter of the microwell and the cell density in the initial cell suspension. Initial theoretical values of “cell trapping” flow rates and wall shear stresses in each microwell were calculated for 20 and 50 µm diameter microwells with Transwell membranes at pressures from 17-85 kPa. The equations used to calculate these factors can be found in the Supporting Information (S-4). Assuming a water-like fluid (condition that holds true right before the cells start to get trapped in the microwells), and based on the porous membrane manufacturer’s specifications (pore size of 400 nm, porosity of 4 × 106 pores/cm2, and membrane thickness of 10 µm), we found that “cell trapping” flow rates in each well can range from 0.8 to 4 and from 5 to 25 nL/min/well, for 20 and 50 µm diameter microwells at 17 and 85 kPa, respectively.

Shear stresses on the microwell wall were less than 1 dyn/ cm2 in all cases. It should be noted that the flow rates and shear stresses can vary considerably based on the microwell geometry and membrane properties and could therefore be tailored for a given application. Volume flow rates and stresses for the overall device will tend to decrease over time as cells are loaded into the microwells. This technique possesses several advantages over previously developed microwell arrays for single and multiple cell trapping based on passive cell seeding. These include the ability to actively place the cells inside the wells through a fast and simple vacuumseeding step, significantly reducing the cell seeding time. This is crucial for maximizing cell viability/functionality and allows more rapid screening. Moreover, with the systems that rely on passive cell settling, cells need to be seeded at a relatively high density Analytical Chemistry, Vol. 82, No. 6, March 15, 2010

2385

(compared to the number of microwells) to guarantee proper microwell filling before washing the undocked cells. With our system, virtually every cell in the suspension to be vacuumed can be seeded within a well. This is particularly important when working with a scarce cell source. In addition, since cell trapping is not achieved by patterning adhesive islands as used in some of the other processes, cells can be selectively seeded onto relatively nonadhesive surfaces, which facilitates the guided assembly of cell clusters.16 Furthermore, the underlying nanoporous substrate could facilitate more suitable nutrient/waste exchange conditions to the cell clusters which can also be a critical factor for avoiding necrotic cell cores and achieving adequate cell functionality. Preliminary modeling studies (see the supplementary model description S-5 in the Supporting Information) of nutrient transport through a cell aggregate that conforms to the geometry of the microwell predict that the concentration gradient for a given solute is smaller in a microwell with a porous bottom (with or without convection flow) than in one with a solid bottom. This gradient is further reduced when perfusion of medium through the cluster and across the porous membrane is considered. CONCLUSION Several adherent and nonadherent cells were actively placed in polymer microwells using vacuum-assisted cell seeding. The process provides a simple method for spatial control of cell location as well as control of cell morphology in the case of single or small numbers of adherent cells. The number of cells in each well was controlled by changing the geometry of the microwell and the

2386


seeding density. Adherent cells were able to attach and spread on the membrane material. These devices are well-suited for experiments where spatial isolation of single and small group of cells is needed. In addition, nonadherent cells were successfully isolated. Finally, clustering of larger groups of 3D cell subunits was demonstrated. These devices could be used in a number of applications including biological studies of single and small groups of cells, high-throughput cell screening, tissue engineering, and guided assembly of 3D cell aggregates. ACKNOWLEDGMENT Support for this work was provided by the Air Force Office of Scientific Research Multidisciplinary Research Initiative (Grant No. F49620-03-1-0421) and the Juvenile Diabetes Research Foundation (Grant No. 5-2009-511). N.F. is supported by an NSF IGERT fellowship (Grant No. DGE0221678). SUPPORTING INFORMATION AVAILABLE SEM of 3T3 cells after removal of microwells, confocal fluorescence micrographs of PANC-1 cells, fluorescence micrographs of PANC-1 cell viability assay, description of flow rate and shear stress calculations, and model predicting diffusion behavior through a cell cluster. This material is available free of charge via the Internet at http://pubs.acs.org. Received for review November 12, 2009. Accepted January 28, 2010. AC902596B

Vacuum-Assisted Cell Seeding in a Microwell Cell Culture System

Recommend Documents