Peptide-Mediated Selective Adhesion of Smooth Muscle and

Mar 21, 2007 - CMTMR) cell tracker dyes were also purchased from Molecular. Probes. .... each location, three cell counts were taken, each representin...
0 downloads 0 Views 80KB Size
Download PDF
5050

Langmuir 2007, 23, 5050-5055

Peptide-Mediated Selective Adhesion of Smooth Muscle and Endothelial Cells in Microfluidic Shear Flow Brian D. Plouffe,† Danson N. Njoka,† Joscelyn Harris,‡ Jiahui Liao,‡ Nora K. Horick,§ Milica Radisic,| and Shashi K. Murthy*,† Department of Chemical Engineering, Northeastern UniVersity, Boston, Massachusetts 02115, Health Careers Academy, Boston, Massachusetts 02115, Center for Biostatistics, Massachusetts General Hospital, Boston, Massachusetts 02114, Institute of Biomaterials and Biomedical Engineering and Department of Chemical Engineering and Applied Chemistry, UniVersity of Toronto, Toronto, Ontario M5S 3G9, Canada ReceiVed January 4, 2007. In Final Form: February 7, 2007 Microfluidic devices have recently emerged as effective tools for cell separation compared to traditional techniques. These devices offer the advantages of small sample volumes, low cost, and high purity. Adhesion-based separation of cells from heterogeneous suspensions can be achieved by taking advantage of specific ligand-receptor interactions. The peptide sequences Arg-Glu-Asp-Val (REDV) and Val-Ala-Pro-Gly (VAPG) are known to bind preferentially to endothelial cells (ECs) and smooth muscle cells (SMCs), respectively. This article examines the roles of REDV and VAPG and fluid shear stress in achieving selective capture of ECs and SMCs in microfluidic devices. The adhesion of ECs in REDV-coated devices and SMCs in VAPG-coated devices increases significantly compared to that of the nontargeted cells with decreasing shear stress. Furthermore, the adhesion of these cells is shown to be independent of whether these cells flow through the devices as suspensions of only one cell type or as a heterogeneous suspension containing ECs, SMCs, and fibroblasts. Whereas the overall adhesion of cells in the devices is determined mainly by shear stress, the selectivity of adhesion depends on the type of peptide and on the device surface as well as on the shear stress.

Introduction The manipulation of cells within microscale devices has been an active area of research in recent years.1-7 The relatively small scale of these devices offers certain advantages over conventional macroscale systems. For example, in the area of medical diagnostics, microfluidic devices have the ability to process small sample volumes rapidly and inexpensively and provide information such as the presence of cells associated with diseases8,9 or the activation of genes as part of the immune response to injury or disease.10 The separation or isolation of key cell types is often an important aspect in the design of these devices.4 Several strategies exist for the separation of different cell types in microfluidic devices. These include separation on the basis of size, adhesion, electrophoretic mobility, and affinity to fluorescent or magnetic tags.4 The principle of affinity chromatography, which applies to separation based on adhesion, is a particularly facile mode of cell separation when the various * To whom correspondence should be addressed. E-mail: smurthy@ coe.neu.edu. † Northeastern University. ‡ Health Careers Academy. § Massachusetts General Hospital. | University of Toronto. (1) Beebe, D. J.; Mensing, G. A.; Walker, G. M. Annu. ReV. Biomed. Eng. 2002, 4, 261-286. (2) Fiorini, G. S.; Chiu, D. T. Biotechniques 2005, 38, 429-446. (3) Pihl, J.; Karlsson, M.; Chiu, D. T. Drug DiscoVery Today 2005, 10, 13771383. (4) Radisic, M.; Iyer, R. K.; Murthy, S. K. Int. J. Nanomed. 2006, 1, 3-14. (5) Toner, M.; Irimia, D. Annu. ReV. Biomed. Eng. 2005, 7, 77-103. (6) Voldman, J. Curr. Opin. Biotechnol. 2006, 17, 532-537. (7) El-Ali, J.; Sorger, P. K.; Jensen, K. F. Nature 2006, 442, 403-411. (8) Yang, J.; Huang, Y.; Wang, X. B.; Becker, F. F.; Gascoyne, P. R. C. Anal. Chem. 1999, 71, 911-918. (9) Du, Z.; Colls, N.; Cheng, K. H.; Vaughn, M. W.; Gollahon, L. Biosens. Bioelectron. 2006, 21, 1991-1995. (10) Sethu, P.; Moldawer, L. L.; Mindrinos, M. N.; Scumpia, P. O.; Tannahill, C. L.; Wilhelmy, J.; Efron, P. A.; Brownstein, B. H.; Tompkins, R. G.; Toner, M. Anal. Chem. 2006, 78, 5453-5461.

cell populations in a sample possess unique markers.9,11,12 For example, leukocyte subpopulations in blood express different surface antigens depending on the cell type (e.g., CD19 expressed by B lymphocytes and CD5 expressed by certain T lymphocytes). By flowing a mixed suspension containing B and T lymphocytes through channels coated with antibodies against one particular antigen, cells expressing that antigen can be separated with high purity.11 This mode of separation does not require any preprocessing incubation steps, such as attaching fluorescent or magnetic antibody tags, which are required for separation in conventional systems (namely, fluorescence-activated cell sorting, FACS, and magnetic cell sorting, MACS).13,14 Adhesion-based microfluidic cell separation would be advantageous in the broad areas of medical diagnostics (where, for example, specific cell surface markers are indicative of disease states9) and tissue engineering (where selected cell types from a heterogeneous suspension obtained from a tissue sample must be enriched prior to culturing on scaffolds15,16). This article examines the role of surface-immobilized peptides and fluid shear stress in the selective capture of smooth muscle cells (SMCs) and endothelial cells (ECs) from mixed suspensions containing SMCs, ECs, and fibroblasts. These cell types are collectively present in different types of tissue (such as cardiac muscle, smooth muscle, heart valves, skin, etc.), and the separation of these cell types is significant from the standpoint of enriching (11) Murthy, S. K.; Sin, A.; Tompkins, R. G.; Toner, M. Langmuir 2004, 20, 11649-11655. (12) Sin, A.; Murthy, S. K.; Revzin, A.; Tompkins, R. G.; Toner, M. Biotechnol. Bioeng. 2005, 91, 816-826. (13) Thiel, A.; Scheffold, A.; Radbruch, A. Immunotechnology 1998, 4, 8996. (14) Putnam, D. D.; Namasivayam, V.; Burns, M. A. Biotechnol. Bioeng. 2003, 81, 650-665. (15) Levenberg, S.; Rouwkema, J.; Macdonald, M.; Garfein, E. S.; Kohane, D. S.; Darland, D. C.; Marini, R.; van Blitterswijk, C. A.; Mulligan, R. C.; D’Amore, P. A.; Langer, R. Nat. Biotechnol. 2005, 23, 879-884. (16) Murthy, S. K.; Sethu, P.; Vunjak-Novakovic, G.; Toner, M.; Radisic, M. Biomed. MicrodeVices 2006, 8, 231-237.

10.1021/la0700220 CCC: $37.00 © 2007 American Chemical Society Published on Web 03/21/2007

Adhesion of Smooth Muscle and Endothelial Cells

key cell types for tissue engineering. In addition, selective capture of a single cell type from a mixed cell population in a microfluidic device can subsequently enable cultivation in the same microscale system rather than using separate systems for separation and cultivation. This approach may prove especially useful in studies aimed at determining the effects of different levels of shear and/ or concentrations of biochemical factors on selected cell populations present in small sample volumes. The basis for cell capture in this work is the selective binding of SMCs and ECs to surfaces coated with the peptide sequences Val-Ala-Pro-Gly (VAPG, derived from elastin) and Arg-GluAsp-Val (REDV, derived from fibronectin), respectively. As demonstrated by Gobin and West,17 VAPG-modified surfaces are adhesive to SMCs but not to fibroblasts or endothelial cells. REDV-coated surfaces, on the other hand, will selectively bind ECs but not allow the adhesion of fibroblasts or SMCs, as shown by Hubbell et al.18 In addition to investigating peptide-mediated adhesion, this article also examines the role of shear stress by utilizing a flow-chamber geometry that generates a linear gradient of shear stress along its axis.19 Such a design allows a range of shear stresses to be examined in a single experiment. The results indicate that the selective adhesion observed in static systems by Gobin et al. and Hubbell et al. can be extended to flow systems and used as a basis for cell separation.

Langmuir, Vol. 23, No. 9, 2007 5051

Figure 1. Flow chamber geometry and shear stress profile based on the design developed by Murthy et al.11 The relationship between shear stress (τw) and axial position (z) is given by the equation derived by Usami et al.19

Materials and Methods Materials. Ethanol (200 proof), glass slides, and fetal bovine serum (FBS) were purchased from Fisher Scientific (Fair Lawn, NJ). 3-Mercaptopropyl trimethoxysilane was obtained from Gelest Inc. (Morrisville, PA), and the coupling agent GMBS (N-ymaleimidobutyryloxy succinimide ester) was obtained from Pierce Biotechnology (Rockford, IL). For chamber fabrication, the SU-8 photoresist and developer were obtained from MicroChem (Newton, MA); the silicone elastomer and curing agent were obtained from Dow Corning (Midland, MI). Phosphate-buffered saline (PBS; 1×, without calcium or magnesium) and Dulbecco’s modified Eagle’s medium (DMEM) were purchased from Mediatech (Herndon, VA). Penicillin-streptomycin (PS) and 0.25% trypsin/ethylene diaminetetraacetic acid (EDTA) solution in Hank’s buffered salt solution (HBSS) (0.25% trypsin-EDTA) were obtained from Hyclone (Logan, UT). The adhesion peptides Val-Ala-Pro-Gly (VAPG) and Arg-Glu-Asp-Val (REDV) were obtained from Sigma-Aldrich (St. Louis, MO) and American Peptide Company (Sunnyvale, CA), respectively. A live/dead cell viability/cytotoxicity kit for mammalian cells containing calcein (494/517 nm) and ethidium homodimer-1 (528/617 nm; EthD-1) was purchased from Molecular Probes (Eugene, OR). Blue [7-amino-4-chloro-methylcourmain, CMAC], green [5-chloromethylfluorescien diacetate, CMFDA], and orange [5-(and 6)-(((4-chloromethyl)-benzoyl)-amino)-tetramethyl-rhodamine, CMTMR) cell tracker dyes were also purchased from Molecular Probes. The A7r5 rat aortic smooth muscle cell line was purchased from American Type Culture Collections (Manassas, VA). 3T3-J2 mouse embryonic fibroblasts and H5V mouse cardiac ECs were kindly provided by Dr. Yaakov Nahmias at the Massachusetts General Hospital and Dr. George Coukos at the University of Pennsylvania, respectively. Device Fabrication. The design of the microfluidic device utilized in the present work is such that the shear stress along the longitudinal axis of the device decreases linearly with device length, as shown in Figure 1. This design is based on the theory of Hele-Shaw flow and design equations derived by Usami et al.19 Microfluidic devices were fabricated and assembled at the BioMEMS Resource Center (17) Gobin, A. S.; West, J. L. J. Biomed. Mater. Res., Part A 2003, 67, 255259. (18) Hubbell, J. A.; Massia, S. P.; Desai, N. P.; Drumheller, P. D. Bio/Technology 1991, 9, 568-572. (19) Usami, S.; Chen, H. H.; Zhao, Y. H.; Chien, S.; Skalak, R. Ann. Biomed. Eng. 1993, 21, 77-83.

at the Center of Engineering in Medicine, Massachusetts General Hospital. A 2D projection of the device was drawn using AutoCAD in house, and the image was printed at high resolution on a transparency (FineLine Imaging, Colorado Springs, CO). This photomask was utilized to generate a negative master. Briefly, a silicon wafer was coated with SU 8-50 photoresist to a thickness of approximately 43 µm. With the transparency overlaid, the wafer was exposed to 365 nm, 11 mW/cm3 UV light from a Q2001 mask aligner (Quintel Co, San Jose, CA). Unexposed photoresist was then removed using SU 8 developer. The feature height was verified using a Dektak surface profiler (Veeco Instruments, Santa Barbara, CA). For device fabrication, the silicone elastomer and curing agent were mixed in a 10:1 (w/w) ratio and poured on top of the negative master wafers, degassed, and allowed to cure overnight at 65 °C. PDMS replicas were then pulled off the wafers prior to punching inlet and outlet holes with a 19-gauge blunt-nose needle. The replicas and glass slides were exposed to an oxygen plasma (100 mW with 8% oxygen for 30 s) in a PX-250 plasma chamber (March Instruments, Concord, MA) and then immediately placed in contact with each other. The irreversible bonding between PDMS and glass was completed by baking for 5 min at 65 °C. Surface modification of the devices was carried out immediately after the baking step. Surface Modification. Stock solutions of GMBS were prepared by dissolving each 50 mg batch of GMBS in 0.5 mL of DMSO. Prior to surface functionalization, a 4% (v/v) solution of 3-mercaptopropyl trimethoxysilane in ethanol was prepared under a nitrogen atmosphere, and 0.1 mg/mL solutions of each peptide were prepared in PBS. Surface functionalization of the microfluidic devices was carried out in three steps. In the first step, the devices were flushed with silane solution and allowed to react at room temperature for 30 min. Unreacted silane was removed by flushing with ethanol. Prior to the next step, an appropriate volume of GMBS stock solution was dissolved in ethanol to create a 0.28% (v/v GMBS stock/ethanol) solution. In the second step, GMBS solution was flowed through the devices and allowed to react for 15 min. After flushing with ethanol to remove unreacted GMBS, the devices were flushed with PBS, followed by the appropriate peptide solution in PBS (REDV or VAPG; 0.1 mg/mL concentration). Following a 30 min period, the devices were flushed with PBS and either used directly in experiments or stored at 4 °C.

5052 Langmuir, Vol. 23, No. 9, 2007 Cell Culture. A7r5 SMCs, H5V ECs, and mouse 3T3-J2 fibroblasts were cultured in 75 cm2 tissue culture flasks at 37 °C in a humidified atmosphere with 5% CO2 and 95% air. The cells were incubated in DMEM supplemented with 4.5 g/L glucose and L-glutamine, 10% fetal bovine serum, 100 U/mL penicillin, and 100 µg/mL streptomycin. Cells were grown to preconfluence and isolated for experiments by trypsinization using a 0.25% trypsin-EDTA solution. For all experiments, cell suspensions were centrifuged at 190g and then resuspended in PBS to a concentration of 1 × 105 cells/mL (measured using a hemacytometer). For experiments with mixed suspensions, 1 × 105 cells/mL solutions of each cell type were prepared separately and incubated with cell tracker dyes. SMCs, ECs, and fibroblasts were stained red, green, and blue, respectively. For this step, the cell suspension drawn from the culture flask was centrifuged at 190g for 5 min, resuspended in 15 mL of serum-free DMEM containing 3 µM of cell-tracker dye, and incubated for 1 h at 37 °C. Centrifugation and resuspension in PBS were subsequently performed as described above. The three suspensions were then combined such that the concentration of each of the three cell types was 0.33 × 105 cells/mL, giving a total cell concentration of 1 × 105 cells/mL. Having the same total cell concentration for the unmixed and mixed suspension experiments is a way to ensure that overall adhesion levels in both types of experiments are of the same order of magnitude, particularly at the lowest shear stress positions. Flow Experiments. Cell suspensions were flowed through microfluidic devices at a flow rate of 40 µL/min for a period of 15 min using a Harvard Apparatus PHD 2000 syringe pump (Holliston, MA). Experiments were performed with both single-cell-type (unmixed) suspensions and mixed suspensions of all three cell types. Cell adhesion within the devices was measured using a field finder (with 1 mm × 1 mm grids) placed under the microfluidic chamber. Adhered cells were manually counted at selected points along the device axis under a Nikon Eclipse TE2000 inverted microscope. At each location, three cell counts were taken, each representing a 1 mm2 square, and then averaged. Cell counts were taken between 3 and 35 mm from the device inlet, along the device axis. All flow experiments were performed at room temperature. For mixed-suspension experiments, cells were imaged by fluorescence microscopy using an Olympus IX51 microscope at 10× magnification using fluorescein (480 ( 30 nm/535 ( 40 nm), rhodamine (540 ( 25 nm/605 ( 50 nm), and DAPI (360 ( 40 nm/460 ( 50 nm) excitation/emission filters. Fluorescence images were acquired using each filter at locations on the devices corresponding to three shear stresses: 3.9, 2.9, and 1.9 dyn/cm2. (The locations were determined using a field finder and bright-field visualization prior to fluorescence image acquisistion.) Fluorescence images acquired using each of the three filters were merged using Nikon NIS-Elements Advanced Research software. Note that these images cover an area of 1.3 mm × 1.7 mm. Cell counts at each shear stress location were obtained by manually counting each cell type (distinguished by color). For comparison with the unmixed cell suspension experiments, the cell counts were normalized by dividing by the image area (2.21 mm2) and multiplying by a factor of 3. The factor of 3 accounts for the fact that in the mixed suspension experiments the concentration of each cell type is 1/3 the concentration in the single-cell-type experiments. Multiplying the mixed-suspension cell adhesion measurements by 3 then allows for direct comparison with the measurements made with the unmixed suspensions. Viability Assay. Viability assays were performed on the cells captured within the microfluidic devices following experiments with the unmixed cell suspensions. Following the flow experiments, a solution containing 2 µM calcein (live cell indicator) and 4 µM EthD-1 (dead cell indicator) in PBS was pumped through the device at 20 µL/min for 12.5 min and allowed incubation for 30 min. Live and dead cells were visualized and counted at 10× magnification using fluorescein and rhodamine filters, respectively. Statistics and Data Analysis. For experiments with unmixed suspensions, cell adhesion measurements were obtained for 9 shear stresses per experiment, and each experiment was repeated 10 times. The cell adhesion measurements that are reported represent average values over 10 repetitions, and the error bars represent the standard

Plouffe et al. Table 1. Cell Adhesion from Mixed and Unmixed Suspensions in REDV-Coated Devices cell adhesion [number of cells/mm2]a,b smooth muscle cells

endothelial cellsc

shear stress (dyn/cm2) unmixed mixed unmixed 3.9 2.9 1.9

mixed

fibroblasts unmixed mixed

3(1 4(1 5(1 5(1 2(0 2(0 9 ( 3 9 ( 1 23 ( 6 23 ( 2 3(1 3(1 21 ( 7 19 ( 2 175 ( 24 187 ( 10 10 ( 1 12 ( 2

a Average number of cells and standard error are rounded to the nearest integer. b Inlet concentration of each cell type is normalized to 105 cells/ mL. c Denotes target cell type.

Table 2. Cell Adhesion from Mixed and Unmixed Suspensions in VAPG-Coated Devices cell adhesion [number of cells/mm2]a,b smooth muscle cellsc shear stress (dyn/cm2) unmixed 3.9 2.9 1.9

5(1 14 ( 3 38 ( 3

endothelial cells

mixed 6(1 15 ( 1 39 ( 2

fibroblasts

unmixed mixed unmixed mixed 1(0 1(0 6(1

1(0 2(0 6(1

1(0 1(0 5(1

1(0 1(0 7(1

a Average number of cells and standard error are rounded to the nearest integer. b Inlet concentration of each cell type is normalized to 105 cells/ mL. c Denotes target cell type.

errors of the mean (standard deviation/xn, where n ) 10). Statistical analysis was performed using two-way analysis of variance (ANOVA) models to examine the relationship among cell adhesion, shear stress, and cell type for each peptide. This analysis was carried out using R version 2.3.1 statistical software. For experiments with mixed suspensions containing all three cell types, 10 experiments were carried out, and the cell adhesion data in Tables 1 and 2 represent averages over these experiments along with the associated standard error. Cell adhesion values and standard errors were rounded to the nearest integer in these Tables to ensure that the data make physical sense. From this data, enrichment values were calculated as the number of adherent cells of a given type relative to the total number of cells that adhered at each level of shear stress. Two-way ANOVA models were then used to investigate how enrichment varied by shear stress level and cell type.

Results Figure 2 shows the results of flow experiments performed with one cell type at a time (no mixed suspensions) using devices coated with REDV and VAPG. The affinity of ECs for REDV is significantly greater than that of SMCs and fibroblasts (Figure 2a), suggesting that REDV-coated microfluidic devices can be used to separate ECs from a mixed suspension of the three cell types. In VAPG-coated devices, SMCs show significant preferential adhesion over ECs and fibroblasts (Figure 2b), but the magnitude of adhesion is not as large as with ECs in REDVcoated devices. For both REDV- and VAPG-coated devices, the variation in cell adhesion between experiments is relatively small, as reflected by the small error bars. Because there is no change in cell culture or flow conditions between independent repetitions of each experiment, the small error bars suggest that the surface coverage of REDV and VAPG within each device is highly reproducible between successive fabrication batches. The results from flow experiments with mixed suspensions are shown in Figure 3. Staining each cell type with a cell-tracker dye prior to creating the mixed suspension allows for easy visualization under a fluorescence microscope. The data shown in Figure 3 were obtained at three positions along the device axis

Adhesion of Smooth Muscle and Endothelial Cells

Langmuir, Vol. 23, No. 9, 2007 5053

Figure 2. Cell adhesion as a function of shear stress observed with smooth muscle cells, endothelial cells, and fibroblasts on microfluidic devices coated with (a) REDV and (b) VAPG, showing preferential adhesion of endothelial cells and smooth muscle cells, respectively. Each data set represents a distinct experiment performed with only one cell type. Error bars denote standard errors for each point based on 10 repetitions.

Figure 3. Fluorescence micrographs showing cells captured in (a-c) a microfluidic device coated with REDV and (d-f) a device coated with VAPG. Images were acquired at shear stress positions of (a, d) 3.9, (b, e) 2.9, and (c, f) 1.9 dyn/cm2. Endothelial cells are stained green, smooth muscle cells are red, and fibroblasts are blue. The input to the device in both instances was a suspension containing equal concentrations of all three cell types.

corresponding to high (3.9 dyn/cm2), medium (2.9 dyn/cm2), and low (1.9 dyn/cm2) shear stress. The area covered by each image in Figure 3 is 1.3 mm × 1.7 mm. Qualitatively, these images are consistent with the variation of cell adhesion with shear stress for each cell type shown in Figure 2. In REDVcoated devices, the proportion of ECs (the target cell type) to SMCs and fibroblasts is greatest at the highest shear stress (Figure 3a-c). A similar trend is shown by SMCs in VAPG-coated devices relative to ECs and fibroblasts. The adhesion of ECs to REDV-coated devices is, in general, of a greater magnitude than that of SMCs to VAPG-coated devices. A two-way ANOVA model revealed that for REDV-coated devices adhesion was significantly affected by the shear stress level (p < 0.01) and cell type (p < 0.01). Furthermore, there was a significant interaction between the shear stress and cell type (p < 0.01), indicating that the difference in adhesion between cell types varied across the levels of shear stress. Similarly, for

VAPG-coated devices, adhesion was significantly affected by both the shear stress level and cell type (p < 0.01 for both factors), and the interaction of the two factors was statistically significant (p < 0.01). A quantitative comparison between the experiments carried out with unmixed and mixed suspensions in REDV- and VAPGcoated devices is given in Tables 1 and 2, respectively. For unmixed cell suspensions, the number of cells adhered at shear stresses of 3.9, 2.9, and 1.9 dyn/cm2 is drawn directly from the data in Figure 2. For the experiments with mixed suspensions containing equal concentrations of all three cell types, cell adhesion data was obtained from fluorescence images such as those shown in Figure 3 as described in the Materials and Methods section. The data in Tables 1 and 2 show that the SMCs, ECs, and fibroblasts adhere in exactly the same way to REDV- and VAPG-coated devices irrespective of whether they flow through

5054 Langmuir, Vol. 23, No. 9, 2007

Plouffe et al.

Table 3. Relative Compositions of Cells Captured from Mixed Suspensionsa shear stress (dyn/cm2)

smooth muscle cells (%)

endothelial cells (%)

fibroblasts (%)

3.9 2.9 1.9

REDV-Coated Devices 33 ( 5 44 ( 9 27 ( 3 64 ( 3 9(1 86 ( 1

23 ( 6 9(2 5(1

3.9 2.9 1.9

VAPG-Coated Devices 66 ( 6 14 ( 4 83 ( 2 9(2 75 ( 2 11 ( 1

20 ( 7 8(3 14 ( 1

a Note that the input composition of suspensions that flowed into both types of devices was 33 ( 5%.

the devices as suspensions of only one cell type or as mixed suspensions of all three cell types. Table 3 shows percentage compositions of the cells captured on the device surface for both peptide types. These values were obtained from the cell adhesion data for the mixed suspensions. In REDV-coated devices, at the low shear stress value of 1.9 dyn/cm2, ECs are captured with a purity of 86%, which is significant compared to the inlet composition of 33%. In VAPGcoated devices, the purity of the captured SMCs is highest (83%) at the intermediate shear stress value of 2.9 dyn/cm2. A two-way ANOVA model for the mixed suspension data for the REDV-coated devices yielded similar conclusions to those obtained for the unmixed data. Enrichment values were significantly associated with shear stress (p < 0.01) and cell type (p < 0.01), and there was a significant interaction between shear stress and cell type (p < 0.01). Similarly, for the VAPG-coated devices, a two-way ANOVA model indicated that enrichment varied significantly by shear stress (p < 0.01) and cell type (p < 0.01) and that there was significant interaction between the two factors (p < 0.01) (i.e. the difference in enrichment between cell types was not constant across the shear stress).

Discussion Adhesion peptides REDV and VAPG were chosen because of their ability to selectively bind ECs and SMCs, respectively, in static systems.17,18 Hubbell et al.18 demonstrated that REDVcoated surfaces selectively bind ECs but not fibroblasts or SMCs. REDV is derived from the type III connecting segment region of fibronectin.20 Through studies of cell adhesion and by using selected anti-integrin antibodies, Hubbell et al. showed that integrin R4β1 is the receptor for REDV on human umbilical vein ECs (HUVECs). VAPG is one of several repeating hydrophobic sequences that are present in tropoelastin, the soluble precursor of elastin.21 Gobin and West17 have shown that VAPG-modified surfaces are adhesive to SMCs but not to fibroblasts or endothelial cells. Peptide sequences related to VAPG, such as VGVAPG, VGAPG, and VGAPG, bind to elastin binding protein, a peripheral membrane protein found on SMCs as well as fibroblasts and several other cell types.22-24 The peptide VAPG, however, is thought to interact with a receptor other than elastin binding protein, as reported by Castiglione Morelli et al.22 This difference (20) Humphries, M. J.; Akiyama, S. K.; Komoriya, A.; Olden, K.; Yamada, K. M. J. Cell Biol. 1986, 103, 2637-2647. (21) Senior, R. M.; Griffin, G. L.; Mecham, R. P.; Wrenn, D. S.; Prasad, K. U.; Urry, D. W. J. Cell Biol. 1984, 99, 870-874. (22) Morelli, M. A. C.; Bisaccia, F.; Spisani, S.; DeBiasi, M.; Traniello, S.; Tamburro, A. M. J. Pept. Res. 1997, 49, 492-499. (23) Hinek, A.; Wrenn, D. S.; Mecham, R. P.; Barondes, S. H. Science 1988, 239, 1539-1541. (24) Mecham, R. P.; Hinek, A.; Entwistle, R.; Wrenn, D. S.; Griffin, G. L.; Senior, R. M. Biochemistry 1989, 28, 3716-3722.

in interaction is most likely why VAPG-coated surfaces can selectively bind SMCs but not fibroblasts or endothelial cells. The results shown in Figures 2 and 3 indicate that REDV and VAPG can be used as surface coatings in adhesion-based cell separation processes to selectively remove ECs and SMCs, respectively, from mixed suspensions. The linear shear stress gradient design was chosen for this work in order to examine the role of shear stress in combination with that of surface functionalization. All three cell types show the expected trend of increasing number of adhered cells with decreasing shear stress in both types of devices (Figure 2a,b). However, the adhesion profiles of the targeted cell types are markedly different. At the lowest shear stress, endothelial cell adhesion in REDVcoated devices is significantly higher than smooth muscle cell adhesion in VAPG-coated devices. Considering the scale on the y axis of Figure 2a, however, it is readily apparent that the nontarget cell (SMC and fibroblasts) adhesion in REDV-coated devices is almost at the same magnitude as that of the target-cell (SMC) adhesion in VAPG-coated devices. These observations suggest that fibroblasts and SMCs are able to bind to REDVcoated surfaces, albeit to a much smaller extent than ECs. By contrast, nontarget cell adhesion on VAPG-coated devices is much smaller, comparable to the adhesion of these cells in devices with no surface coating (data not shown). The adhesion of the targeted SMCs in VAPG-coated devices, however, is certainly preferential and statistically significant. Another characteristic of the adhesion profiles in Figure 2 is the relative magnitude of target cell adhesion compared to nontarget cell adhesion. In REDV-coated devices, the difference between target (ECs) and nontarget (SMCs and fibroblasts) cell adhesion begins to increase in magnitude only below a shear stress value of approximately 2.6 dyn/cm2. By contrast, the gap between target and nontarget cell adhesion in VAPG-coated devices is relatively large up to a shear stress of around 3.6 dyn/cm2. If the critical shear stress is defined as the shear stress below which the difference in cell adhesion between two cell types is greater than 10 cells/mm2, then these two shear stresses can be used as bases for comparison. The greater critical shear stress for SMCs on VAPG suggests a smaller dissociation constant for VAPG-SMC receptor binding compared to that for REDVEC receptor binding, a conclusion that is supported by values from the literature. The former can be estimated to be in the 10-9 M range on the basis of reported values for the dissocation constant of VGVAPG binding to Lewis lung carcinoma cells25 and chemotaxis data for monocytes exposed to GVAPG.22 The latter has been measured by Hubbel et al. to be 2.2 × 10-6 M using HUVECs.18 The higher magnitude of cell adhesion of ECs on REDV compared to that of SMCs on VAPG could then be attributed to a greater density of REDV-binding sites on the EC surface compared to the density of VAPG-binding sites on SMCs. The effect of fluid shear stress on the number and density of receptors on ECs and SMCs was not investigated in the present study. It is known that shear stress can influence the expression and activity of integrins in endothelial cells. Urbich et al.26 have shown that when HUVECs are exposed to shear stresses near physiological levels (12-15 dyn/cm2) for short periods of time (0-5 min) changes in integrin conformation occur, resulting in an increased strength of integrin-ligand bonds. Furthermore, there are changes in integrin mobility that result in integrin clustering, which can lead to an increase in the number of binding sites on the cell surface. Changes in mRNA transcription and the (25) Blood, C. H.; Zetter, B. R. J. Biol. Chem. 1989, 264, 10614-10620. (26) Urbich, C.; Walter, D. H.; Zeiher, A. M.; Dimmeler, S. Circ. Res. 2000, 87, 683-689.

Adhesion of Smooth Muscle and Endothelial Cells

protein level generally require shear stress exposure over several hours.27 An analysis of the effect of fluid shear stress on receptors requires the ability to stain the adhered cells with radio- or fluorescently labeled antibodies without exposing the cells to additional shear. A shortcoming of the linear shear gradient device used in the present study is the inability to carry out this type of analysis. The adhesion profiles of the target and nontarget cells in both types of devices show very little variability between experiments. This is in contrast to prior work with similar devices used to separate T and B lymphocytes on the basis of antibody-antigen binding.11 In that work, avidin was immobilized on the flow channel surfaces by coupling with GMBS and 3-mercaptopropyl trimethoxysilane using protocols similar to those employed in the present study. Antibodies were then attached to the avidin layer by avidin-biotin coupling. The adhesion profiles of T and B lymphocytes showed a high degree of variability that was mitigated by the use of biotinylated poly(ethylene glycol) (PEG) in conjunction with the biotinylated antibodies. The high degree of variability in cell adhesion was attributed to inadequate coverage of the flow channel surface by antibodies. In the present work, the adhesion peptide was tethered directly to GMBS without an intermediate step (such as binding avidin), and the concentration of adhesion peptide that flowed into the device for binding to GMBS was much higher (0.1 mg/mL compared to 0.01 mg/mL in previous work). It is postulated that these two factors lead to uniform monolayer coverage of the flow channel surface with the peptide, ensuring minimal nonspecific binding. The fluorescence micrographs shown in Figure 3 and the data in Tables 1 and 2 show that the adhesion behavior of all three cell types in both types of devices is independent of whether the cells are in mixed suspensions or in suspensions of only one cell type. As shown in Tables 1 and 2, the shear stress data from Figure 2 is therefore directly applicable to mixed suspensions and can be used as a basis for the design of an adhesion-based cell-separation device. The results obtained from mixed suspensions also indicate that the total number of adherent cells depends strongly on shear stress. However, at each shear stress level, the relative enrichment for the target cell type is mainly determined by the type of peptide used for surface functionalization, consistent with the static studies performed by Hubbell18 et al. and Gobin and West.17 Specifically, at each shear stress level the relative fraction of ECs that adhered to the REDV-coated devices is significantly higher than the (27) Tzima, E.; del Pozo, M. A.; Shattil, S. J.; Chien, S.; Schwartz, M. A. EMBO J. 2001, 20, 4639-4647.

Langmuir, Vol. 23, No. 9, 2007 5055

relative fraction of nontargeted cell types. At a shear stress value of 1.9 dyn/cm2, 86% of the cells that adhered to the surface are EC, 9% are SMCs, and 5% are fibroblasts. Thus, significant enrichment is achieved in the REDV-coated devices compared to the percentage of ECs in the inlet suspension (33%). Similarly, in VAPG-coated devices, the relative fraction of the targeted cell type, SMCs, is significantly higher than the relative fractions of nontargeted cell types (ECs and fibroblasts). At a shear stress value of 2.9 dyn/cm2, 83% of the adhered cells were SMCs, 9% were ECs, and 8% were fibroblasts. Overall our findings provide a set of design criteria for microfluidic devices that can separate ECs and SMCs on the basis of adhesion. Specifically, the yield of the separation device (i.e., the number of adhered cells) is mainly determined by the shear stress level, with low values of shear stress (below 2.9 dyn/cm2) providing the highest number of adhered cells. The selectivity of the device is primarily determined by the type of peptide immobilized on the surface. However, selectivity is also influenced by shear stress, with lower shear stresses providing greater selectivity. Whereas high values of target cell enrichment were obtained with both REDV- and VAPG-coated devices, our results suggest that ligands consisting of only four amino acids may not be selective enough to obtain purity levels closer to 100%.

Conclusions Microfluidic devices coated with REDV and VAPG can be used for the adhesion-based separation of ECs and SMCs from heterogeneous suspensions containing both cell types in addition to fibroblasts. The adhesion of ECs in REDV-coated devices and SMCs in VAPG-coated devices becomes significantly greater than that of the nontargeted cells with decreasing shear stress. Furthermore, the adhesion of these cells is shown to be independent of whether these cells flow through the devices as suspensions of only one cell type or as a heterogeneous suspension containing ECs, SMCs, and fibroblasts. Whereas both REDVand VAPG-coated devices can capture the target cell types (ECs and SMCs, respectively) with high purity (83% or greater), more ECs are captured in REDV-coated devices than SMCs in VAPGcoated devices. Acknowledgment. We gratefully acknowledge support from the Northeastern University Provost’s Office through grants to S.K.M. and D.N. We also thank Professor Rebecca Carrier for access to the Olympus fluorescence microscope and Professor Mehmet Toner and Mr. Octavio Hurtado for access to facilities at the BioMEMS Resource Center at the Massachusetts General Hospital. LA0700220