Endothelial Cell Migration on Surface-Density ... - ACS Publications

Sep 25, 2007 - Joseph Walpole , Feilim Mac Gabhann , Shayn M Peirce , John C ... Felix Jakob Segerer , Peter Johan Friedrich Röttgermann , Simon Schus...
0 downloads 0 Views 234KB Size
11168

Langmuir 2007, 23, 11168-11173

Endothelial Cell Migration on Surface-Density Gradients of Fibronectin, VEGF, or Both Proteins Lingyun Liu,†,| Buddy D. Ratner,*,†,‡ E. Helene Sage,*,§ and Shaoyi Jiang*,†,‡ Department of Bioengineering, UniVersity of Washington, Seattle, Washington 98195, Department of Chemical Engineering, UniVersity of Washington, Seattle, Washington 98195, and Hope Heart Program, Benaroya Research Institute at Virginia Mason, Seattle, Washington 98101 ReceiVed May 16, 2007. In Final Form: July 6, 2007 Cell migration is essential to many physiological processes, including angiogenesis, which is critical to the success of implanted biomaterials and tissue-engineered constructs. Gradients play an important role in cell migration. Previous work on cell migration has been mostly executed either in the concentration gradients of stimuli (e.g., VEGF) in bulk or hydrogels or on the surface-density gradients of ECM proteins (e.g., fibronectin) or small ligands (e.g., RGD). Little work has been done to investigate how cell migration responds to the surface-density gradients of growth factors. No work has been done to study how the surface gradients of both adhesive proteins and growth factors influence cell migration. In this work, we studied the effect of the surface-density gradients of fibronectin (FN), VEGF, or both proteins on endothelial cell migration. Gradients with different slopes were prepared to study how the gradient slope affects cell migration. The gradients were generated by first forming a counter-propagating C15COOH/C11OH selfassembled monolayer (SAM) gradient using a surface electrochemistry approach, followed by activating the -COOH moieties and covalently immobilizing proteins onto the surface. Fourier transform infrared spectra and X-ray photoelectron spectroscopy were used to characterize the SAM and protein gradients, respectively. A free cell migration assay using bovine aortic endothelial cells was performed on various gradient surfaces or on surfaces with uniform protein density. Results showed that cells on the surface-density gradients of FN, VEGF, or both proteins moved faster along the gradient direction than on the respective uniform control surface after 24-h cell culture. It is also shown that for each protein or protein combination, the directional cell displacement was not statistically different between two gradients with different slopes. Results show that the directional cell migration was increased by about 2-fold on the VEGF gradient as compared to the FN gradient and was further increased by another 2-fold on the combined gradients of both proteins as compared to the VEGF gradient alone. This is the first work to create surface-density gradients of VEGF and the first study to generate a combined surface gradient of growth factor and ECM protein to investigate their effect on cell migration on surfaces. This work broadens our understanding of the directional movement of endothelial cells. Our findings provide useful information for directing cell migration into tissue-engineered constructs and can be potentially used for those applications where cell migration is critical, such as angiogenesis.

Introduction Cell migration is essential in many physiological processes such as angiogenesis, wound healing, the immune response, and tumor metastasis. Angiogenesis is critical to wound healing (since the newly formed blood vessels provide nutrition and oxygen to growing tissues) and thus is critical to the success of implanted biomaterials and tissue-engineered constructs. Endothelial cells secrete proteases to degrade the extracellular matrix, then migrate into the perivascular space, proliferate, and align themselves to form new vessels directed by a gradient of nutrients and cytokines.1 Therefore, a gradient of proteins plays a very important role in directed cell migration and angiogenesis. Many interesting studies have been performed to investigate how cells respond to the gradients of different stimuli. Gradients of soluble growth factors such as VEGF2 and aFGF3 were shown to promote directed cell migration through chemotaxis. Recent * To whom correspondence should be addressed. E-mail: sjiang@u. washington.edu; [email protected]; hsage@benaroyaresearch. org. † Department of Bioengineering, University of Washington. ‡ Department of Chemical Engineering, University of Washington. § Benaroya Research Institute at Virginia Mason. | Present address: Department of Chemical and Biomolecular Engineering, University of Akron, Akron, OH 44325-3906. (1) Bischoff, J. Trends Cell Biol. 1995, 5, 69-74. (2) Soga, N.; Namba, N.; McAllister, S.; Cornelius, L.; Teitebaum, S. L.; Dowday, S. F.; Kawamura, J.; Hruska, K. A. Exp. Cell Res. 2001, 269, 73-87. (3) Stokes, C. L.; Rupnick, M. A.; Williams, S. K.; Lauffenburger, D. A. Lab. InVest. 1990, 63, 657-668.

studies have been reported to characterize cell behavior, such as attachment, alignment, and migration, in response to the gradients of surface-bound proteins or ligands. Plummer et al. showed that fibroblast attachment was affected by a fibronectin gradient on a surface.4 Human umbilical vein endothelial cell attachment was shown to vary spatially across PEG hydrogels when RGDS was tethered in a gradient.5 Protein expression was found to vary spatially from cells cultured on covalently immobilized countergradients of laminin and collagen I.6 Fibroblasts cultured on the surface density gradients of RGD aligned themselves parallel to the gradient and elongated compared to their behavior on uniform RGD surfaces.7 Neuron axon was shown to orient in the direction of increasing surface density of laminin.8 Smooth muscle cells were observed to align on hydrogels modified with a bFGF gradient in the direction of increasing tethered bFGF concentration.9 Recently, Smith et al. showed that the directed migration of bovine aortic endothelial cells (BAECs) was increased on a surface-bound fibronectin gradient compared to a uniform control (4) Plummer, S. T.; Wang, Q.; Bohn, P. W. Langmuir 2003, 19, 7528-7536. (5) Burdick, J. A.; Khademhosseini, A.; Langer, R. Langmuir 2004, 20, 51535156. (6) Gunawan, R. C.; Choban, E. R.; Conour, J. E.; Silvestre, J.; Schook, L. B.; Gaskins, H. R.; Leckband, D. E.; Kenis, P. J. A. Langmuir 2005, 21, 30613068. (7) Kang, C. E.; Gemeinhart, E. J.; Gemeinhart, R. A. J. Biomed. Mater. Res. A 2004, 71A, 403-411. (8) Dertinger, S. K. W.; Jiang, X.; Li, Z.; Murthy, V. N.; Whitesides, G. M. Proc. Natl. Acad. Sci. U.S.A. 2002, 99, 12542-12547. (9) DeLong, S. A.; Moon, J. J.; West, J. L. Biomaterials 2005, 26, 3227-3234.

10.1021/la701435x CCC: $37.00 © 2007 American Chemical Society Published on Web 09/25/2007

Cell Migration with Surface Gradient of Growth Factors

Langmuir, Vol. 23, No. 22, 2007 11169

substrate.10 Furthermore, cell migration is regulated simultaneously by growth factors and extracellular molecules.11 So far little work has been done to investigate how cell migration responds to the surface-density gradients of growth factors, and no work has been done to study how the surface gradients of both adhesive protein and growth factor influence cell migration. In this contribution, we demonstrate the effect of the surfacedensity gradients of fibronectin (FN), vascular endothelial cell growth factor (VEGF), and both proteins on the migration of endothelial cells. Fibronectin is one of the most common proteins in the extracellular matrix and has been shown to promote angiogenesis.12 VEGF is one of the most potent angiogenic factors in the angiogenesis process. When VEGF binds to its cell receptor, it stimulates proliferation, migration, and survival of endothelial cells and promotes angiogenesis.13 It has been reported recently that endothelial cells guide outgrowing capillaries in response to the gradients of extracellular VEGF in retinal angiogenesis.14 By combining both FN and VEGF gradients on a surface, it represents the conditions closer to the environment of cells in the body. The protein gradients in this work were generated by an electrochemical method. Gradients with two different slopes were prepared to study how the gradient slopes influence the directed cell migration. The derived gradients were characterized by Fourier transform infrared (FTIR) spectra and X-ray photoelectron spectroscopy (XPS). A free cell migration assay using bovine aortic endothelial cell was performed to study endothelial cell migration on various protein gradients. These experiments broaden the understanding of the directional movement of endothelial cells and provide useful information for designing new biomaterials or tissue-engineered scaffolds with enhanced angiogenesis.

Experimental Methods Materials. Rectangular glass coverslips (9 mm × 22 mm, Electron Microscopy Sciences, Hatfield, PA) were deposited with a 2-nm chromium adhesion layer followed by a 20-nm gold layer using a CHA SEC-600 E-beam evaporator (CHA Industries, Fremont, CA) at a base pressure of 10-6 Torr. The deposition rates for chromium and gold were 0.5 Å/s and 5 Å/s, respectively. 11-Mercapto-1-undecanol (HS(CH2)11OH) and 16-mercaptohexadecanoic acid (HS(CH2)15COOH) were purchased from Aldrich (St. Louis, MO) and used as received. N-hydroxysuccinimide (NHS) and 1-ethyl-3-(3-(dimethylamino)propyl)carbodiimide (EDC) were obtained from Sigma. Human plasma fibronectin was purchased from Chemicon (Temecula, CA). Recombinant human VEGF165 was obtained from PeproTech (Rocky Hill, NJ). Bovine aortic endothelial cells were a gift from Prof. Cecilia Giachelli (University of Washington). The CO2-independent medium, L-glutamine, and fetal bovine serum were purchased from Gibco Invitrogen (Carlsbad, CA). Preparation of Alkanethiol Gradients. The alkanethiol gradients were prepared using an electrochemical method.4,15 The gold-coated substrates were first cleaned by irradiation under UV/ozone in a Jelight UVO-cleaner (model 42, Irvine, CA) for (10) Smith, J. T.; Tomfohr, J. K.; Wells, M. C.; Beebe, T. P.; Kepler, T. B.; Reichert, W. M. Langmuir 2004, 20, 8279-8286. (11) Maheshwari, G.; Wells, A.; Griffith, L. G.; Lauffenburger, D. A. Biophys. J. 1999, 76, 2814-2823. (12) Nicosia, R. F.; Bonanno, E.; Smith, M. J. Cell. Physiol. 1993, 154, 654661. (13) Sahni, A.; Francis, C. W. Blood 2000, 96, 3772-3778. (14) Gerhardt, H.; Golding, M.; Fruttiger, M.; Rubrberg, C.; Lundkvist, A.; Abramsson, A.; Jeltsch, M.; Mitchell, C.; Alitalo, K.; Shima, D.; Betsholtz, C. J. Cell Biol. 2003, 16, 1163-1177. (15) Wang, Q.; Bohn, P. W. J. Phys. Chem. B 2003, 107, 12578-12584.

Figure 1. Schematic of the process to produce the surface density gradient of a protein using an electrochemical approach.

20 min, extensively rinsed with DI water and ethanol, and dried with nitrogen. To form OH-terminated self-assembled monolayers (SAMs) on gold, clean gold-coated substrates were soaked in a 1 mM HS(CH2)11OH solution in ethanol overnight, rinsed with ethanol, and dried with N2. Using a pair of Teflon clips, platinum wires were pressed into intimate contact with opposite ends of the Au substrate precoated with C11OH SAM to establish working electrodes E1 and E2. The sample was then placed in a Teflon electrochemical reaction cell containing an Ag/AgCl reference electrode and a Pt counter electrode. A 0.5 M methanolic KOH solution was used as the electrolyte and was purged with Ar for 15 min prior to use. A potential window of -1400 to -400 mV or -900 to -800 mV was applied to both ends of the C11OHcoated Au substrate by a bipotentiostat for 1 min. Samples were then quickly removed from the electrochemical cell, rinsed thoroughly with methanol and ethanol, and reimmersed in a 1 mM ethanolic solution of HS(CH2)15COOH. After a 5-min assembly time, the sample was rinsed with ethanol, dried with N2, and immediately used for protein immobilization. Protein Immobilization. To create protein gradients on surfaces, FN, VEGF, or both proteins were covalently attached to the carboxyl terminus of HS(CH2)15COOH having a gradient across the surface (Figure 1). NHS (2 mg/mL) in dioxane and EDC (20 mg/mL) in H2O were freshly prepared and mixed

11170 Langmuir, Vol. 23, No. 22, 2007

immediately before use in a 9:1 ratio. The alkanethiol gradient of C15COOH/C11OH was activated by treatment with the mixture of EDC and NHS for 1 h under stirred conditions. After being washed briefly with water, the activated substrate was incubated with 20 µg/mL FN, 10 µg/mL VEGF, or a mixture of 20 µg/mL FN and 10 µg/mL VEGF in PBS (PH 7.4, 10 mM, 138 mM NaCl, 2.7 mM KCl, Sigma) overnight at 4 °C. The samples were then rinsed with water, soaked in 100 mM NaOH on a shaker for 15 min, and rinsed with PBS. FTIR Characterization. FTIR spectra were measured to characterize the EDC-/NHS-activated C15COOH/C11OH gradients. The spectra were measured using a Tensor 27 spectrometer (Bruker, Billerica, MA) with 80 scans at a resolution of 2 cm-1. All spectra were baseline corrected for accurate interspectrum comparison. Data points were acquired at 2-3 mm intervals, starting 2 mm from the substrate edge, with a beam size of approximately 2 mm. X-ray Photoelectron Spectroscopy Measurements. To characterize protein gradients, a series of XPS atomic intensity signals were measured iteratively along a surface. The elemental compositions of the protein layer at various positions were quantitatively determined, and the intensity ratio of protein relating the nitrogen peak (relating to protein) to the Au peak (relating to substrate) was calculated. For XPS characterization, the fibronectin gradients prepared as described above were washed with water, dried with N2, and left in a dryer overnight. The XPS spectra were recorded at a step size of 2-3 mm along surfaces, using Surface Science Instruments X-probe and M-probe spectrometers with monochromatic Al KR X-ray sources (hν )1486.6 eV). The binding energy of the Au4f7/2 was calibrated to 84.0 eV. Surface Plasmon Resonance Measurements. A surface plasmon resonance (SPR) sensor was used to determine the mass of protein bound to the pure C15COOH or C11OH surface. The custom-built SPR sensor for measuring SPR curves has been described previously.16 The sensor chip was deposited with C15COOH or C11OH SAMs. After the chip was mounted, the sensor was stabilized with PBS buffer. Then, a mixture of EDC and NHS prepared freshly as described above was flowed into the SPR cell for 15 min, followed by flushing with PBS buffer for 5 min. A FN solution at 20 µg/mL was then flowed for 30 min, followed by washing with PBS for 20 min. To remove nonspecific protein adsorption, a solution of 100 mM NaOH was used to wash the surface for 30 min, followed by PBS for 20 min. Cell Culture and Migration Assay. BAECs were maintained in continuous growth in Dulbecco’s Modified Eagle Medium (DMEM), supplemented with 10% fetal bovine serum, 1% sodium pyruvate, 1% non-esssential amino acids, and 2% penicillinstreptomycin solution at 37 °C in a humidified atmosphere containing 5% CO2 on TCPS flasks. Cells were used before passage 15. The cell culture medium and all reagents were obtained from Gibco (Gaithersburg, MD). The free cell migration assay10 was used to determine the effect of protein gradients on BAEC migration. The coverslips immobilized with gradients of FN, VEGF, or both proteins were transferred to a six-well culture plate and washed with sterile PBS three times. All samples were then blocked with 1 mg/mL sterile BSA for 30 min. In the interim, BAECs were trypsinized, washed, and resuspended in the CO2-independent medium supplemented with 4 mM glutamine and 0.2% fetal bovine serum. After the BSA-blocking solutions were removed and substrates were washed with sterile PBS three times, cell suspension at (16) Liu, L. Y.; Chen, S. F.; Giachelli, C. M.; Ratner, B. D.; Jiang, S. Y. J. Biomed. Mater. Res. A 2005, 74A, 23-31.

Liu et al.

6000 cells/cm2 was added to each well and incubated with the samples at 37 °C under a phase contrast microscope equipped with a temperature control device. Time-lapse phase contrast images (10×) were captured every 30 min using a Nikon TE200 inverted microscope combined with a Metamorph program. Cells were continuously observed for 24 h. Images were placed sequentially into stacks for further analysis. The movement of individual cells was manually tracked and quantified. Uniform surfaces with FN, VEGF, or both proteins covalently bound to pure C15COOH surfaces were used as control in these experiments. At least 100 cells were tracked for each case. Numerical data were analyzed using one-way analysis of variance (ANOVA). If probability values for a set of data were found to be less than 0.05 (p < 0.05), the differences between the data were considered statistically significant. Statistical analyses were performed using Origin 6.1 (OriginLab Corporation, MA).

Results and Discussion In this paper the surface gradients of FN, VEGF, or both proteins were produced by first forming a counter-propagating two-component C15COOH/C11OH SAM gradient using an electrochemical approach, then activating reactive -COOH moieties via EDC/NHS coupling chemistry, and finally covalently immobilizing the protein onto this surface through its amine groups (Figure 1). A variety of methods for producing surface-bound gradients of proteins or small molecules have been reported, including cross-diffusion,10,17-19 electrochemistry,4,15 photopolymerization,20,21 microfluidics,8 and ink-jet printing.22 From these methods, the electrochemical method was chosen to form protein gradients in this study due to a greater ease than other methods to tune gradient properties such as slope. In principle, a spatially varying in-plane electrochemical potential is mapped onto thiol molecules on a surface to enable the control of the composition profile of SAMs laterally, using their characteristic reductive desorption/oxidative adsorption reactions. Thiol molecules desorb in the region where the potential is lower than the reductive desorption potential and adsorb in the region where the potential is higher than the oxidative adsorption potential. A transition region with spatially varying composition forms between the covered and bare regions. Once a gradient of one thiol component is established, a second thiol component can be backfilled into the bare Au region, resulting in a counter-propagating twocomponent SAM gradient. The rate of the compositional change in the transition region can be easily controlled by adjusting the width of the applied electrochemical potential window. In this work, a C11OH SAM, which is low-fouling to protein adsorption,16 was first formed on the gold substrate. After the gradient of C11OH was formed,4 C15COOH molecules were backfilled to provide reactive sites for covalently tethering protein via EDC/ NHS activation. FTIR was used to characterize the C15COOH/C11OH gradients formed from different potential windows. Figure 2 shows the integrated area of the carbonyl band at 1735 cm-1 as a function (17) Liedberg, B.; Tengvall, P. Langmuir 1995, 11, 3821-3827. (18) Liedberg, B.; Wirde, M.; Tao, Y. T.; Tengvall, P.; Gelius, U. Langmuir 1997, 13, 5329-5334. (19) Riepl, M.; Ostblom, M.; Lundsrom, I.; Svensson, S. C. T.; Denier van der Gon, A. W.; Schaferling, M.; Liedberg, B. Langmuir 2005, 21, 1042-1050. (20) Hypolite, C. L.; McLernon, T. L.; Adams, D. N.; Chapman, K. E.; Herbert, C. B.; Huang, C. C.; Distefano, M. D.; Hu, W. S. Bioconjugate Chem. 1997, 8, 658-663. (21) Harris, B. P.; Kutty, J. K.; Fritz, E. W.; Webb, C. K.; Burg, K. J. L.; Metters, A. T. Langmuir 2006, 22, 4467-4471. (22) Roth, E. A.; Xu, T.; Das, M.; Gregory, C.; Hickman, J. J.; Boland, T. Biomaterials 2004, 25, 3707-3715.

Cell Migration with Surface Gradient of Growth Factors

Figure 2. Integrated area of carbonyl band at 1735 cm-1 measured from FTIR as a function of position for two C15COOH/C11OH gradients after EDC/NHS activation. The gradients were formed from different potential windows: -1400 to -400 mV (empty square and solid line) and -900 to -800 mV (solid square and dotted line).

of position after C15COOH was activated with EDC/NHS. The C15COOH/C11OH gradients before EDC/NHS activation were also measured by FTIR and showed a much weaker carbonyl signal than after activation. Control experiments showed a weak CdO signal from the uniform C15COOH SAM on rough gold surfaces deposited by e-beam evaporation and a strong CdO signal on super flat single-crystal Au (111) surfaces deposited by a modified thermal evaporator. It is likely that surface roughness reduces the CdO signal. The carbonyl signal was amplified after activation because more carbonyl groups were introduced onto the surfaces from the activated NHS-ester. It was obvious that carbonyl bands decreased in intensity as the local potential shifted positively, i.e., toward the -OH end of the gradient. The transition area showed a characteristic sigmoidal profile. With a potential window of -1400 to -400 mV, the potential width is 1000 mV. The gradient transition area was noticeably sharp with an approximately 1-mm width, and the gradient center was located slightly to the right of the film center. With a potential window of -900 to -800 mV, the potential width was decreased to 100 mV. The SAM gradient became much wider in physical space, covering almost the whole length of the substrate. As the potential width decreases and the physical length of the substrate is kept constant, the potential drop per unit length decreases, resulting in a larger transition region with a smoother gradient. Plummer et al. reported that the reductive desorption potential of C11OH from Au measured by cyclic voltammetry was -975 mV.4 It has been previously observed that the desorption of organothiols from Au in static systems occurs at potentials 100-300 mV positive of the sweeping desorption potential, due to the difference between the temporally varying potential in cyclic voltammetry and the static potential window applied in the gradient formation experiment.4,23,24 Our results are consistent with these previous reports. Once C15COOH/C11OH SAM gradients with different slopes were obtained, protein gradients were produced on these gradients by activating reactive -COOH moieties using carbodiimidecoupling chemistry and covalently immobilizing the proteins onto surfaces through their amine groups. SPR experiments were performed to determine the mass of the proteins bound to the C15COOH or C11OH surface. Figure 3 shows the SPR curves (23) Terrill, R. H.; Balss, K. M.; Zhang, Y.; Bohn, P. W. J. Am. Chem. Soc. 2000, 122, 988-989. (24) Plummer, S. T.; Bohn, P. W. Langmuir 2002, 18, 4142-4149.

Langmuir, Vol. 23, No. 22, 2007 11171

Figure 3. SPR measurements of fibronectin adsorption on C15COOH and C11OH surfaces.

when various reagents were flowed over the C15COOH or C11OH surface. The results show that the amount of adsorbed proteins on the C15COOH surface was much higher than that on the C11OH surface. Introduction of FN to the activated -COOH surface caused a SPR shift of 9.3 nm, which decreased to 8.9 nm after it was treated with NaOH, indicating that most FN molecules were covalently immobilized onto the activated -COOH surface. Introduction of FN to the -OH surface caused a SPR shift of 3.7 nm, which decreased to 1.6 nm after the surface was washed with NaOH. The final SPR wavelength shift caused by protein adsorption at 20 µg/mL on C15COOH and C11OH surfaces corresponds to the surface coverage of 1.76 and 0.32 ng/mm2, respectively. Brief exposure to NaOH at the final step facilitated the removal of loosely bound proteins. This step, which did little to alter protein bioactivity as shown in the cell migration assay, was included for preparing all the protein gradients used in this work. Because proteins react only with the activated esters derived from -COOH terminal groups, the gradient structure of immobilized proteins should follow the spatial distribution of C15COOH. The presence of protein gradients was confirmed with XPS. A series of XPS atomic intensity signals were measured along the substrates to characterize the atomic profiles. Atomic intensity information was used to determine the ratio of proteinrelated peaks (N) to substrate-related peaks (Au), defined as the “protein ratio”. The change in protein ratio over the distance to the -COOH end is shown in Figure 4. There was a maximal amount of fibronectin at the -COOH end and a minimal amount of fibronectin at the -OH end. When the potential window was -1400 < E < -400 mV, a steep FN gradient was obtained. Based on the surface coverage of FN on the pure -COOH or -OH surface from SPR, the gradient was from 1.76 to 0.32 ng/mm2 with an approximately 1-mm wide transition region and a slope of 1.44 ng/mm3. With the potential window of -900 to -700 mV, a much smoother FN gradient was generated. Calibrated with the SPR results, the gradient was from 1.76 to 0.32 ng/mm2 with an approximately 10-mm wide transition region and a slope of 0.14 ng/mm3. XPS results are consistent with FTIR results, suggesting that the protein gradients with two different slopes were successfully generated on surfaces using the electrochemical approach. BAECs were seeded on uniform or gradient surfaces of FN, VEGF, or both proteins and cultured for 24 h for cell migration experiments. Since cell culture was performed outside the incubator for continuous microscope observation, a CO2independent medium was used for cell culture. Fetal bovine serum

11172 Langmuir, Vol. 23, No. 22, 2007

Liu et al.

at a concentration as low as 0.2%, which does not significantly alter gradient surfaces, was added to the culture medium to reduce cell death during the time-lapse observation period. Several methods have been developed to study cell migration on surfaces, such as the wound assay,25 the Teflon fence assay,26 and free cell migration assay.10 In the wound assay, cells are scraped from the surface which will probably remove or destroy the protein layer on the surface. The fence model is widely used, but the population-related effects cannot be avoided with this method. To investigate how surface-density gradients of FN, VEGF, or both proteins affect cell migration, the free cell migration model was used in our work. A low cell-seeding density was used to reduce cell-cell interactions. Typical cell images on surfaces with protein gradients at different time points after cell seeding are given in Figure 5. Cell displacements along protein gradients toward higher surface protein densities after 24-h culture are shown in Figure 6. It was shown that cells on the surface-density gradients of FN moved further than on the uniform FN surface, which is consistent with the previous report that the drift speed of BAECs increased along a FN gradient as compared to the uniform control substrate.10 In our work, cells on the FN gradient surface with a slope of 0.14 ng/mm3 shifted 7 µm after 24-h culturing. In the work by Smith et al., cells on the FN gradient surface with a similar slope shifted 30 µm after 24 h.10 This difference might be caused by different conditions used such as the serum concentration in culture medium. A serum concentration as low as 0.2% was used in our study, whereas 10% serum was used in Smith’s work. It was also observed in our work that cells on the gradients of VEGF or two proteins (FN and VEGF) moved further along the gradient direction than on the uniform control surface. The directed migration of endothelial cells along the gradients of FN, VEGF, or both proteins suggests that cells respond haptotactically. Gradients with different slopes were electrochemically derived for determining how the gradient steepness affects cell migration. Our results show that for FN, VEGF, or both proteins, the directional cell migrations on sharp and smooth gradients of each case were not statistically different (Figure 6). Gunawan et al. also found that the migration rate of rat small intestine

epithelial cells was independent of the slopes of the immobilized laminin gradients prepared in microfluidic devices,6 which is consistent with our results. Smith et al., however, found that the drift speed of human microvascular endothelial cells increased with increasing gradient slope.27 The discrepancy might be caused by large standard deviations observed in both studies, different cell types used, or different culturing conditions. It is also possible that on the chip with a sharp gradient some cells beyond the gradient area were included for analysis due to the sharpness of the slope, which could decrease the average cell shift rate. A more delicate approach is desired that can display the protein gradient range and allow cell observation simultaneously. Cells are able to sense small differences in extracellular stimuli concentration between its two ends and amplify this small difference into a much greater intracellular biochemical gradient, which is then translated into the mechanical force to drive cell migration.28 It is possible that both gradients used in this study had reached a saturated level so that after cell amplification and translation the final driving forces for cell migration were similar, and thus no differences in cell migration were detected. Gradients with smaller slopes might be needed to detect statistically significant differences in directed cell migration. Interestingly, it was found that the directional cell migration on the VEGF gradient surfaces was increased as compared to that on the FN gradient surfaces for both gradient slopes (Figure 6). Cells displaced 10 or 13 µm, respectively, along the sharp or smooth VEGF gradient after 24-h culture, as compared to 4 or 7 µm, respectively, on the FN gradient surfaces. VEGF has been known to be important to cell migration. A recent study shows that endothelial cells guide capillary outgrowth toward the gradients of extracellular VEGF in retinal angiogenesis.14 Our results here, in the first work to create the surface-density gradients of VEGF to study its influence on cell migration, show that the VEGF surface gradient promotes cell migration more than the FN surface gradient. It was also found that when cells were cultured on the gradients of two proteins (VEGF and FN), the directional endothelial cell migration was further increased, as compared to that with VEGF alone or FN alone (Figure 6). Cells displaced 23 or 26 µm along the sharp or smooth gradient surface of both VEGF and FN after 24-h culture, as compared to 10 or 13 µm, respectively, on the VEGF gradient surfaces, and 4 or 7 µm, respectively, on the FN gradient surfaces. The immobilized amount of FN or VEGF on a surface from a mixture of VEGF and FN should be less than that when the immobilization was done separately from pure FN or VEGF with a similar concentration to that in the mixture. Furthermore, VEGF and FN have different mechanisms to promote cell adhesion and migration. FN binds to integrins, whereas VEGF binds to VEGF receptors. FN and VEGF stimulate cell migration through binding to different cell receptors and taking different signaling pathways. With the combined gradients of FN and VEGF, both integrins and VEGF receptors were engaged, leading to an increased cell migration compared with the effect of the onecomponent gradient. The combined gradient of growth factor and ECM protein might better mimic the real extracellular environment in the body. This is the first study to generate the combined surface gradients of growth factor and ECM protein and to investigate their effects on cell directional migration on surfaces. Kumar et al. used micropattern arrays to guide the long-range (>600 µm) directional migration of attached mammalian cells

(25) Majack, R. A.; Clowes, A. W. J. Cell. Physiol. 1984, 118, 253-256. (26) Pratt, B. M.; Harris, A. S.; Morrow, J. S.; Madri, J. A. Am. J. Pathol. 1984, 117, 349-354.

(27) Smith, J. T.; Elkin, J. T.; Reichert, W. M. Exp. Cell Res. 2006, 312, 2424-2432. (28) Wu, D. Cell Res. 2005, 15, 52-56.

Figure 4. XPS characterization of fibronectin gradients on C15COOH/C11OH SAM gradients formed from different potential windows: -1400 to -400 mV (empty squares and solid line) and -900 to -700 mV (solid squares and dotted line). The curves represent the ratio of N/Au versus position, with the left end of each curve representing the C15COOH end.

Cell Migration with Surface Gradient of Growth Factors

Langmuir, Vol. 23, No. 22, 2007 11173

Figure 5. Typical phase contrast images of cells on surfaces with protein gradients at different time points after cell seeding. Cells marked with red underlines represent the same cell after culturing for 0 h (left), 12 h (middle), and 24 h (right).

Conclusions

Figure 6. Bovine aortic endothelial cell displacements along gradients toward higher protein surface densities after 24-h cell culture. Data shown are for uniform, steep gradient, and smooth gradient surfaces of fibronectin, VEGF or both. * means significant difference between two compared results (p < 0.05).

in the absence of stimulus gradients.29 The difference in maximum migration distances between two studies is most likely attributed to different approaches used (i.e., patterns vs stimulus gradients) and different cell culture conditions (e.g., 10% vs 0.2% serum). Low serum concentration (e.g., 0.2%) was used in our study to minimize the effect of serum on protein gradients on the surfaces. It was shown in other studies that 10% serum increased cell migration by >4 fold.4 Thus, different cell migration rates can be obtained using various approaches (e.g., gradients vs patterns) and under various conditions (e.g., serum concentrations) and can be applied for applications where controllable extents of migration are desired, e.g., to match polymer scaffolds with different biodegradation abilities. Our results provide useful information for the design of better biocompatible biomaterials or tissue-engineered scaffolds. For example, incorporation of growth factors or both growth factor and ECM protein can improve cell migration, possibly leading to enhanced angiogenesis in engineered tissues. (29) Kumar, G.; Ho, C. C.; Co, C. C. AdV. Mater. 2007, 19, 1084-1090.

In this work, we demonstrate that the surface gradients of FN, VEGF, and both proteins with different slopes can be generated on surfaces by forming C15COOH/C11OH SAM gradients using the electrochemical method, followed by EDC/NHS activation and the chemical immobilization of proteins. Different gradient slopes were easily derived by changing the width of the electrochemical potential window, confirmed by FTIR and XPS. Cell migration experiments show that bovine aortic endothelial cells on the surface-density gradients of FN, VEGF, or both proteins moved further along protein gradients than on the respective uniform surface control after 24-h culture. For cells on the gradients of FN, VEGF, or both proteins, the directional cell displacements were not statistically different between two gradients with different slopes. It was also found that the directional cell migration on the VEGF gradients was increased as compared to the FN gradients. On the combined gradients of both proteins, the directional endothelial cell migration was further increased as compared to VEGF alone or FN alone. Cells displaced 23 or 26 µm along the sharp or smooth gradients, respectively, of both VEGF and FN after 24-h culture, as compared to 10 or 13 µm, respectively, on the VEGF gradients, and 4 or 7 µm, respectively, on the FN gradients. This work can be readily extended to study the gradient effects of other proteins on cell behavior. It provides useful information for directing cell migration into tissue-engineered scaffolds and for developing better biocompatible materials, e.g., those with controlled angiogenesis, for biomedical applications. Acknowledgment. We thank Allen D. Taylor for his assistance to fabricate the Teflon electrochemical reaction setup and Dr. Shengfu Chen for helpful discussions. We thank Prof. Daniel Schwartz for the use of electrochemical reaction devices. We also thank Dr. Kip Hauch for the use of optical microscopes at the University of Washington Engineered Biomaterials (UWEB) Optical Microscopy and Image Analysis Shared Resource. This work is supported by the National Science Foundation under Grants NSF EEC-9529161 and CTS-0092699. Supporting Information Available: Figure S1, experimental setup used to create a C15COOH/C11OH alkanethiol gradient on a gold substrate via electrochemical reaction. This material is available free of charge via the Internet at http://pubs.acs.org. LA701435X