11552
Langmuir 2004, 20, 11552-11556
Integrated Lithographic Membranes and Surface Adhesion Chemistry for Three-Dimensional Cellular Stimulation James D. Kubicek, Stephanie Brelsford, Punit Ahluwalia, and Philip R. LeDuc* Departments of Mechanical and Biomedical Engineering, Carnegie Mellon University, 5000 Forbes Avenue, Pittsburgh, Pennsylvania 15213 Received May 18, 2004. In Final Form: September 22, 2004 The complex spatiotemporal organization of cellular and molecular interactions dictates the physiological function of cells. These behaviors are indications of an integrated response to a three-dimensional cellular environment and anchored in cell adhesion on scaffolds. Here, we are able to control interconnected structural, mechanical, and chemical stimuli by dictating the cellular environment through chemical surface modifications, soft lithography, and mechanical deformation. Control of these variables is obtained through the use of an elastomeric membrane chemically modified for cell adhesion with a pressure-driven cellstretching device which creates a pattern of forces similar to those encountered in physiological environments. Further, the integration of lithographic methods and chemical patterning allows the introduction of spaceand time-dependent parameters by combining mechanical stimulation, biochemical regulation, and scaffolding design. The method is applied to stimulate single cells and cell populations to examine cellular response with spatiotemporal control. This research provides the capacity to probe biological patterns and tissue formation under the influence of mechanical stress.
Introduction Intensive studies of the link between the mechanics and biochemistry of cells have provided insights into physiological mechanotransduction, including gravity sensation, audio-sensory channels, and baro-reception and has stimulated research in related areas, including polymer physics and bioengineering.1-3 Mechanical stimulation is transmitted through cell adhesion by a family of transmembrane proteins, the integrins, which are linked intracellularly to a series of proteins including talin, vinculin, and paxillin.4-6 These proteins form focal adhesion complexes around the integrins upon their engagement by ligands in the extracellular matrix (ECM) and connect to actin filaments. Focal adhesion complexes, therefore, represent a junction for force transmission from the ECM to the intracellular cytoskeleton. This mechanism for force transfer from the extracellular arena to the intracellular molecular interactions allows signal communication to have multi-variable constraints. Through the application of force to cells, an understanding of the regulation of canonical intracellular signal-transduction pathways, such as the mitogen-activated protein kinase pathways7-9 and the regulation of the second messenger, * Author to whom correspondence should be addressed. Phone: 412-268-2504. Fax: 412-268-3348. E-mail:
[email protected]. (1) Wang, N.; Butler, J. P.; Ingber, D. E. Science 1993, 260, 11241127. (2) LeDuc, P.; Haber, C.; Bao, G.; Wirtz, D. Nature 1999, 399, 564566. (3) Perkins, T. T.; Smith, D. E.; Larson, R. G.; Chu, S. Science 1995, 268, 83-87. (4) Moulder, G. L.; Huang, M. M.; Waterston, R. H.; Barstead, R. J. Mol. Cell. Biol. 1996, 7, 1181-1193. (5) Wood, C. K.; Turner, C. E.; Jackson, P.; Critchley, D. R. J. Cell Sci. 1994, 107, 709-717. (6) Rose, D. M.; Liu, S.; Woodside, D. G.; Han, J.; Schlaepfer, D. D.; Ginsberg, M. H. J. Immunol. 2003, 170, 5912-5918. (7) Ferrer, I.; Blanco, R.; Carmona, M.; Puig, B.; Barrachina, M.; Gomez, C.; Ambrosio, S. J. Neural. Transm. 2001, 108, 1383-1396. (8) Li, C.; Hu, Y.; Mayr, M.; Xu, Q. J. Biol. Chem. 1999, 274, 2527325280. (9) Shrode, L. D.; Rubie, E. A.; Woodgett, J. R.; Grinstein, S. J. Biol. Chem. 1997, 272, 13653-13659.
cyclic adenosine monophosphate, has been advanced.10-13 Further, ultimate mammalian cell behavior is linked to mechanical stimuli since the application of stress influences proliferation, differentiation, and apoptosis.14-19 Prior observations have largely been made in in vitro experiments using limited environmental constraints such as prescribed material scaffolding or two-dimensional mechanical-stimulation materials.20-23 These methods, however, were unable to simulate the complex multidimensional array of environmental stimulation simultaneously, including chemical, mechanical, and scaffolding, which influence cellular responses in physiological systems. This lack of effective methods to explore cell behavior in an accurate three-dimensional environment has hampered progress in the understanding of the normal cellular response in living tissue. The integration of this mechanical connection with the scaffolding system of the cell (10) Meyer, C. J.; Alenghat, F. J.; Rim, P.; Fong, J. H.; Fabry, B.; Ingber, D. E. Nat. Cell Biol. 2000, 2, 666-668. (11) Garcia-Cardena, G.; Comander, J. I.; Blackman, B. R.; Anderson, K. R.; Gimbrone, M. A. Ann. NY Acad. Sci. 2001, 947, 1-6. (12) Topper, J. N.; Gimbrone, M. A., Jr. Mol. Med. Today 1999, 5, 40-46. (13) Resnick, N.; Yahav, H.; Khachigian, L. M.; Collins, T.; Anderson, K. R.; Dewey, F. C.; Gimbrone, M. A., Jr. Adv. Exp. Med. Biol. 1997, 430, 155-164. (14) Matsuda, N.; Morita, N.; Matsuda, K.; Watanabe, M. Biochem. Biophys. Res. Commun. 1998, 249, 350-354. (15) Liu, S. Q.; Ruan, Y. Y.; Tang, D.; Li, Y. C.; Goldman, J.; Zhong, L. Biomech. Model Mechanobiol. 2002, 1, 17-27. (16) Weyts, F. A.; Bosmans, B.; Niesing, R.; Leeuwen, J. P.; Weinans, H. Calcif. Tissue Int. 2003, 72, 505-512. (17) Hammerschmidt, S.; Kuhn, H.; Grasenack, T.; Gessner, C.; Wirtz, H. Am. J. Respir. Cell Mol. Biol. 2003. (18) Husse, B.; Sopart, A.; Isenberg, G. Am. J. Physiol. Heart Circ. Physiol. 2003, 285, H1521-1527. (19) Chess, P. R.; Toia, L.; Finkelstein, J. N. Am. J. Physiol. Lung Cell Mol. Physiol. 2000, 279, L43-51. (20) Levenberg, S.; Huang, N. F.; Lavik, E.; Rogers, A. B.; ItskovitzEldor, J.; Langer, R. Proc. Natl. Acad. Sci. USA 2003, 100, 1274112746. (21) Sumpio, B. E.; Banes, A. J.; Levin, L. G.; Johnson, G., Jr. J. Vasc. Surg. 1987, 6, 252-256. (22) Camargo, M. J.; Sumpio, B. E.; Maack, T. Am. J. Physiol. 1984, 247, F656-664. (23) Boitano, S.; Sanderson, M. J.; Dirksen, E. R. J. Cell Sci. 1994, 107, 3037-3044.
10.1021/la0487646 CCC: $27.50 © 2004 American Chemical Society Published on Web 11/24/2004
Lithographic Membranes and Surface Adhesion Chemistry
Langmuir, Vol. 20, No. 26, 2004 11553
fabricated networks and composite organization within the membrane of varied elasticities produce a threedimensional strain on the cells in the PreCS device. For example, cells that are cultured in fabricated structures inside the membrane can experience stress at a minimum on their basal and apical surfaces simultaneously. Experimental Section
Figure 1. Diagram and image of the combined soft lithography for chemical surface modification with the Pressure-driven Cell Stretching (PreCS) device. Cells were cultured in lithographically fabricated channels within deformable membranes providing a strain gradient. Flexible membranes were prepared using Sylgard 184. The poly(dimethylsiloxane) (PDMS) membrane was poured, or spun at speeds up to 5000 rpm, into a precast mold then thermally cured at 80 °C for 20 min to obtain variable thicknesses between 50 µm and 2 cm. The thickness must be tightly regulated because it is closely correlated with the applied strain during pressurized deformation. The 60mm-diameter membrane of the PDMS for stretching was clamped in the pressure-driven device. The substrate was sterilized by the addition of 100% ethanol and rinsed with phosphate-buffered saline (PBS). Fibronectin was added to the surface of the membrane before seeding NIH 3T3 fibroblast cells with Dulbecco’s Modified Eagle’s Medium supplemented by 10% calf serum, glutamine, 0.3 mg/mL, penicillin, 100 U/mL, streptomycin, 100 µg/mL, and 20 mM N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid at pH of 7.4, under 5% carbon dioxide. The scale bar is 5 cm.
attachment along with the biochemical environment, which is inherent in mammalian cell systems, requires the simultaneous control of all components for realistic results to be extracted. To address this multi-variable control dilemma, chemical patterning and soft lithography have been integrated into the Pressure-driven Cell-Stretching device (PreCS). This system constrains an elastomeric membrane to modulate the mechanical stimulation of cells through their ECM connections while also incorporating scaffolding and chemical control to simulate a physiologically functional environment. A membrane such as poly(dimethylsiloxane) (PDMS) is utilized to mechanically stimulate cells by virtue of its quasi-elastic stress-strain behavior; a regulator applies pressure ranging from -15 to +15 lb/in2 to the bottom surface of the membrane to modulate the deformation. Cells are initially cultured on this membrane to form an adherent monolayer and are then placed in the PreCS device (Figure 1). A controlled increase in the pressure on the lower surface of a radially symmetric membrane in a PreCS device creates an associated equibiaxial stretch due to the fixed-displacement boundary conditions. The strain is transferred to cells through their basal side attachments where the ECM-coated PDMS surface allows attachment and spreading of single cells. Since the differential pressure is physically separated from the surface of the membrane-cell attachment, an associated deformation of the flexible membrane strains the cells mechanically. This approach can be used with various porous and elastic materials, including biodegradable membranes, for engineered tissues and three-dimensional scaffolding control.24-26 Although the substrate that is being stretched by pressurization is planar, imbedded (24) Jansson, K.; Haegerstrand, A.; Kratz, G. Scand. J. Plast. Reconstr. Surg. Hand Surg. 2001, 35, 369-375. (25) Wang, J. H.; Yao, C. H.; Chuang, W. Y.; Young, T. H. J. Biomed. Mater. Res. 2000, 51, 761-770. (26) Wang, Y.; Ameer, G. A.; Sheppard, B. J.; Langer, R. Nat. Biotechnol. 2002, 20, 602-606.
Cell Culture. Before being seeded on the flexible membranes, the cells were washed once with phosphate-buffered saline solution (PBS) and then exposed to trypsin-ethylenediaminetetraacetate for 3 min. After dissociation from the tissue culture plates, the cells were counted and cultured on the flexible membranes of PDMS at 4300 cells/cm2. The cells were subsequently incubated for at least 6 h to allow for attachment and spreading prior to mechanical stimulation. The medium was replaced, while a thermostatically regulated heat source maintained the sample temperature at 37 °C for the duration of the experiment; this allowed for a controlled aqueous cell-culture environment prior to mechanical stimulation. Membrane Coating for Cell Attachment. The membrane was coated with 150 µL of human fibronectin (FN) dissolved in PBS, and after 60 min of incubation at 23 °C, the substrate was seeded with NIH 3T3 fibroblasts.27 Coatings of varying concentrations of FN were applied on the membrane to optimize the conditions for attachment and spreading, as at low FN concentrations, the cells remained not fully spread or in suspension. After cell attachment and spreading on the elastomeric substrate, the membrane was rinsed with PBS and manually clamped in the equibiaxial cell stretcher for deformation by externally applied pressure. Microscope Visualization. The morphology of single cells was observed with an Axiovert inverted Zeiss microscope. For examination of cell morphology, the fibroblasts were fixed in 4% paraformaldehyde diluted in PBS and then permeabilized in 0.1% TritonX-100 diluted in PBS. The cells were immunofluorescently labeled with fluorescein isothiocyanate phalloidin and Dapi to observe actin filaments and the nucleus. The sample was mounted on a #1 borosilicate coverslide with Fluoromount-G. Differential interference contrast and epi-fluorescence microscopy were performed with an Insight digital camera and NIH Image analysis software with dapi and fluorescein isothiocyanate filter sets under 10×, 20×, and 63× (1.4 numerical aperture) objectives.
Results and Discussion The PreCS integrates fabrication and molecular technologies to generate cellular patterns which mimic ECMcell interactions in vivo and enables the application of three-dimensional mechanical stimuli. First, we modulated the adherent molecular regions through the application of spatially constrained areas of FN. To analyze the structural form of cell regulation without the limitations of overlaying conventional tissue-culture dishes with varying amounts or distributions of nonadhesive blocking polymer, we built patterns of adhesive regions with size, shape, and position defined on the micrometer scale, surrounded by nonadhesive boundary regions established with pluronics or bovine serum albumin. The substrates were coated with FN, and the cells were placed on these surfaces in a medium devoid of serum to prevent the deposition of ECM molecules on the nonadhesive surfaces. We maintained control over the size and shape of the adhesive surface, as well as controlled the extent to which the cells could distend (Figure 2A). When cells adhered to the ECM-coated region, they spread over fixed ECM anchors to cover the adhesive region, stopping when their periphery had reached the nonadhesive boundary on the PDMS. As a result, they changed their morphology to assume the geometry of their molecularly patterned (27) Cunningham, C. C.; Stossel, T. P.; Kwiatkowski, D. J. Science 1991, 251, 1233-1236.
11554
Langmuir, Vol. 20, No. 26, 2004
Figure 2. Spatial cellular control using nonadhesive molecules and soft lithography. (A) A differential-contrast (DIC) image of fibroblasts cultured with 0% calf serum on PDMS coated with fibronectin (FN) is shown in the lower region of the image. After adhesive molecules were coated on defined regions of the surface (FN+), solutions with nonadhesive molecules including co-block polymers were added to the entire surface, coating the upper region in the image (FN-). (B) DIC image of cells cultured on a flexible PreCS membrane with cellular attachment controlled through fluidic channels. Scale bars ) 30 µm.
Figure 3. Multilayered PDMS membrane for cellular stimulation with PreCS (A) drawing and (B) photomicrographic image of fabricated fluidic channels in the membrane, which are used for molecular patterning, differential strains, and the introduction of complex three-dimensional geometries. Scale bar ) 500 µm.
substrate. This allowed us to analyze the effects of varying cell populations and single-cell sizes and shapes on cell functions (e.g., proliferation, apoptosis, differentiation) in combination with three-dimensional mechanical stimulation with the PreCS apparatus. This was a critical step since local changes in cell-ECM interactions heavily influence growth differentials that drive pattern formation and the structural basis of morphogenic regulation in vivo. By controlling the membrane architecture with fabrication techniques, three-dimensional networks were integrated with the PreCS device to create mechanical stimulation environments mimicking in vivo conditions. This is useful in a multitude of physiological environments where heterogeneous scaffolding and mechanical stimulants are inherent, such as in the simulation of vasculature for tissue-engineering applications. Enclosed geometries inside the PDMS were fabricated through multilevel lithographic techniques to spatially control cell attachment and spreading (Figure 2B). Rectangular channels were created using negative photoresist layers on silicon wafers and then multilayered systems could be combined for three-dimensional organization of the channels (Figure 3A). Patterns were printed onto transparent masks, after which the wafers were exposed to ultraviolet light with the mask adherent to the resist layer. After the photoresist was exposed and developed, the remaining resist was used as a positive mold and then the PDMS was placed on the surface of the wafer and thermally cured (Figure 3B). The thickness was defined by the rotational speed of the wafer during introduction of the PDMS. The exposed channel was inverted onto a second membrane and thermally cured. The size of the channels could be controlled in three dimensions down to single micrometers. Cells were cultured in these imbedded membrane networks, as well as in exposed channels lacking an upper membrane. The cells attached themselves to one or more supporting
Kubicek et al.
Figure 4. Membrane constraining system and cellular staining with the PreCS device. (A) Clamping device for constraining the membrane periphery. The clamping configuration can be circular or elliptical with major-to-minor-axis ratios of 2, 4, and 8 with (B) the strain vector profiles indicated for the increasing ratios. Scale bar ) 2 cm. (C) Epi-fluorescent image of immunolabeled NIH 3T3 fibroblasts on PreCS membranes for the nucleus (blue) and the actin cytoskeleton (green) with Dapi and fluorescein isothiocyanate phalloidin, respectively. Scale bar ) 15 µm.
surfaces of the channels, exposing them to threedimensional mechanical stimulation and scaffolding support. To study more-complex, anisotropic stimulation, we controlled the boundary constraints of the PreCS device and created membranes with materials of varying elastic moduli or topography. We modified the circular clamping constraints to obtain an elliptical configuration and impose gradients of strain (Figure 4A). The major axis of the elliptical membrane remained constant while the minor axis was decreased in ratios of 2, 4, and 8. This imposed a curvature that was greater over the length of the minor than of the major axis. The minor-axis length was decreased; therefore, the strain, , of the membrane and associated cells approached uniaxial conditions (minor axis . major axis) in this direction (Figure 4B). This made it possible to impose varying mechanical deformation profiles on cells cultured on these membranes, over the surface of the membrane. The methodology was expanded to impose heterogeneous three-dimensional deformation by varying the elastomeric properties of the membrane in defined domains throughout the substrate. The PreCS device also allowed cellular immunostaining, which we used to observe the cytoskeleton (Figure 4C) as stress affects morphology and actin cytoskeleton alignment when individual cells sense and respond to the application of external forces.28 This is essential to control mechanics, chemistry, and scaffolding as this underlying cytoskeletal structure will change under the combination of all three variables. PreCS is able to modulate these experimental conditions to examine the cytoskeleton and separate these interconnected factors. Thus, the extracellular and intracellular stimulation, which affect cell responses from cell chemotaxis to proliferation and are linked to the cytoskeleton through the ECM and focal adhesions, can be elucidated. Variations of the elastic moduli also allow us to control over stimulation profiles. An infusion of materials with higher elastic moduli than the PDMS, including a woven nylon matrix (EPDMS ) 2 × 105 Pa, Enylon ) 2 × 109 Pa) were introduced into the membrane. Defined stress-strain profiles were developed through imbedding materials and varying surface topologies over the diameter of the film. Through repeating patterns of materials at the micrometer scale imbedded in the polymer, we were able to produce topologies with strain gradients across the width of the membrane (Figure 5A). Alternately, the thickness profile (28) Costa, K. D.; Hucker, W. J.; Yin, F. C. Cell Motil. Cytoskeleton 2002, 52, 266-274.
Lithographic Membranes and Surface Adhesion Chemistry
Langmuir, Vol. 20, No. 26, 2004 11555
Figure 5. Imbedded polymer composite and membrane deformation. (A) Metallic grid imbedded in PDMS. This micrometer-scale grid creates gradients of deformation of membranes (inset) for cell stretching. Scale bar ) 50 µm. (B) Interferometric measurements of vertical deflection for pressures from 0 to 0.4 lb/in2 with a PDMS membrane 1 mm in thickness. The deflection is measured at the center of the circular membrane (inset). The error bars represent the standard deviation.
of the membrane and the resultant strain profile were controlled by membrane topologies through a time-delay thermal curing cycle. The variations in the PDMS surface produced complex deformation of the cells. To quantify the strain applied by the PreCS device, we first measured the vertical displacement of the PDMS membrane optically by noncontact laser interferometry at the center point of the circular PDMS membrane. A retroreflective coating was applied to the surface of the PDMS membrane to increase the signal-to-noise ratio from the optical head. The displacement was measured with a Michelson interferometer (Polytec; GmbH, Waldbronn, Germany), which employs fiber-optic bundles to control the paths of the reference and target laser beams. Due to the edge constraints and circular nature of the equidistant constrained system, the midpoint of the membrane reflected the region with the maximum vertical displacement (from 100 µm to 2 cm). Mathematical techniques are needed to precisely describe the strain on the system under the influence of a pressure-driven membrane.29,30 We characterized the relationship between pressure and strain in the substrate, which revealed that circularly clamped elastic membranes behave like spherical membranes when exposed to a pressure exceeding a stress threshold. The shape of the deformation is approximated by the surface of a sphere where the bulgetest equation is valid. An approximately linear relationship between applied pressure and vertical displacement is revealed over a range of strains up to 25% (Figure 5B). The threshold of this thin-film approximation is correlated with the ratio of the change in vertical displacement of the film to the pre-strain thickness of the film (data not shown). The membrane was exposed to a constant-pressure environment with elastic-modulus mismatches, which induced strains that varied across the section. This displacement across the membrane was measured with a MicroVal Coordinate Measuring Machine (Brown and Sharpe; North Kingstown, Rhode Island), which is accurate to 10 µm for the x, y, and z coordinates of an orthogonal Cartesian-coordinate system. As we traversed the surface of the membrane, the vertical deflection followed an essentially spherical configuration with the slope of the deflection increasing as the radial distance from the center of the membrane increased (Figure 6A). The deformation of the membrane was also modeled using finite-element software (ANSYS; Canonsburg, Pennsyl(29) Gilbert, J. A.; Weinhold, P. S.; Banes, A. J.; Link, G. W.; Jones, G. L. J. Biomech. 1994, 27, 1169-1177. (30) Williams, J. L.; Chen, J. H.; Belloli, D. M. J. Biomech. Eng. 1992, 114, 377-384.
Figure 6. Experimental and simulation displacement profiles under uniform pressure. (A) MicroVal coordinate measurements of vertical deflection with respect to the horizontal position across the membrane surface under pressure with a PDMS membrane 1 mm in thickness. (B) Displacements of the circular membrane are modeled using finite-element methods with shell elements in ANSYS. A top view of the deflection of homogeneous PDMS is shown under uniform pressure constrained at the periphery of the membrane. The deformation results are coded with the greatest displacements at the center of the membrane. (C) Local vertical displacements due to the gradient of elastic moduli in the membrane of the PreCS device. MicroVal coordinate measurements of vertical deflection with respect to horizontal position across the surface of a membrane with a rectangular nylon section imbedded in the membrane at 6.8 mm on the graph. (D) Side view of a membrane showing the displacements modeled as in (B) with an imbedded rectangular section. The vertical displacement profile reveals a greater deformation at either side of the x axes than the position centered along the x axis where the embedded material is located.
vania) using clamped boundary conditions at the periphery of an elastic membrane modeled with shell elements. The elastic modulus of PDMS was used as a homogeneous material property with a constant pressure from below the basal side of a multilayered element system. The resulting displacements in a circularly constrained membrane deformation system revealed the deformation due to the pinned periphery constraint on the membrane, consistent with our experimental observations (Figure 6B). Experimental and simulation results when imbedding a thin rectangular section of nylon matrix at the interior of the membrane reveals a divergence in deformation from the previously described spherical deformation of the homogeneous membrane (Figure 6C, D). Thus, the PreCS device is able to apply a strain gradient distribution due to composite formed from the imbedded section within PDMS. Conclusion The integration of soft lithography and chemical surface modification with the PreCS device represents a significant resource for simultaneously modulating multiple environmental parameters (scaffolding, mechanics, and chemistry) with concurrent measurements of cell responses using biological staining and imaging. With this device, we can explore the local activation of biochemical responses and spatial distribution of the focal adhesion complexes and examine transmembrane proteins involved in mechanotransduction. We are also developing an enhanced device to analyze these three physical dimensions in real time. The ability to control the scaffolding and chemistry
11556
Langmuir, Vol. 20, No. 26, 2004
while simultaneously applying mechanical stimulation with the PreCS device in a three-dimensional configuration will greatly enhance our understanding of cellular responses. This research has led to significant advances in establishing a multifaceted in vitro technique, which mimics in vivo environments over a wide range of applications. These observations, in turn, will improve our knowledge of cellular and molecular behavior and will provide useful insights in related fields, including in vitro diagnostics, biomechanics, tissue engineering, and drug discovery.
Kubicek et al.
Acknowledgment. We would like to sincerely thank Matthew Brake and Jonathan Wickert for their help with the interferometry measurements, Julie Jadlowiec and Ayse Celil in the Carnegie Mellon Bone Tissue Engineering Center for the NIH 3T3 fibroblasts, and William Messner, Jerry Griffin, and Sui Huang for their helpful suggestions. This work was supported by the National Science Foundation CAREER, Department of Energy Genome to Life, Pennsylvania Infrastructure Technology Alliance, and Berkman Faculty Development Fund. LA0487646
