A Novel Technique for Micro-patterning Proteins and Cells on Polyacrylamide Gels

2.1.Microfabrication and soft lithography

A Si master with geometric patterns was fabricated using standard photolithographic technique as shown in Fig. 1a1–1a4. The patterns were designed using AutoCAD (Autodesk, Inc., CA). Chrome masks were manufactured by Fine line Imaging (Colorado Springs, CO). The Si wafer was cleaned, degreased, blown dry with N₂ gas and baked at 110°C (4 min). Descumming of the Si wafer was performed using O₂/Ar plasma (flow rate ratio of 2:1) at Reactive Ion Etcher (March Instruments, Concord, CA) for 1 min at 100 W. The Si wafer was then spin-coated (maximum spin speed of 4000 rpm) with a thin layer of SU-8 negative photoresist (SU-8 2002, MicroChem, Newton, MA). After soft baking at 110°C for 4 min on a flat hot plate, the wafer was brought in hard contact with the chrome mask and was exposed to UV light for 7s using a flood exposure system (ABM, Inc., Fig. 1a1). After the post exposure bake at 95°C for 2 min, the wafer was developed using SU-8 developer (MicroChem, Newton, MA) for 10s (Fig. 1a2) and hard baked at 150 °C for 18 min. Finally the wafer was treated with molecular vapor deposition (MVD) of FDTS (heptadecafluro-1H, 1H, 2H, 2H-hydradecyltrichlorosilane) using MVD System (Applied MST, San Jose, CA) to ensure easy removal of the cured PDMS stamp in the subsequent steps (Fig. 1a3–1a4) ^49–52. PDMS prepolymer was obtained via mixing Silicone elastomer and curing agent at 10:1 ratio (Dow Corning, Midland, MI) and degassing the mixture in a vacuum jar for 45 min ^52–54. The prepolymer solution was poured over the Si master mold (Fig. 1b1) and cured in the oven at 70°C for 12 h. The elastomeric stamp was then peeled off carefully (Fig. 1b2) and cut in 2 × 2 cm² pieces for μCP.

2.2.Micro-contact printing (μCP) and protein pattern transfer

Fibronectin (FN), laminin (LN) and collagen type I (CN) (from BD Biosciences, NJ) were respectively diluted in PBS solution to a concentration of 50 μg/ml. For FN, the diluted solution was oxidized to contain aldehyde groups by using sodium periodate (NaIO₄) to facilitate the FN conjugation to PA gel solution ^{44, 55}. The PDMS stamp obtained via soft lithography was then inked with the oxidized FN solution for 20 min at room temperature. Then, the excess solution was blown off (Fig. 1b3–1b4). The stamp was brought into complete conformal contact with a glass substrate for 45 min at 37°C (Fig. 1b5). Small weights (25 g) were placed over the PDMS stamp to aid complete protein pattern transfer from PDMS to intermediate glass. Two different types of glass cover slips were used: (1) cover slips as received from the manufacturer (No. 1, 2.5 × 2.5 cm², Corning life Sciences, Netherlands) without any surface treatment; (2) cover slips after immersion in piranha bath (96% H₂SO₄+30% H₂O₂ in 3:1 molar ratio) for over 1 day, which makes the glass hydrohiphilic. Wet treatment of glass in piranha creates both hydroxyl (−OH) and sulfhydryl (−SH) groups on glass surface responsible for the increased surface hydrophilicity ^{49, 56–58}. The surface hydrophilicity characteristics for both the as-received glass slide and the ones after piranha treatment were quantified by performing the static contact angle measurements. A CAM200 Goniometer from KSV NIMA (Finland) was used and 7 μl droplets of DI water were dispensed on the surfaces to measure the static contact angles. The contact angles of DI water on slides before and after piranha treatment were 42°±3° and 0°, respectively (Fig. 6 insets). A drop of PA gel solution (20 μL) was placed on the cover slip with the protein patterns. An activated glass cover slip which can covalently bind to gel (see Section 2.3) was placed (floated) on the drop. The drop spread between the cover slips due to capillarity and was sandwiched with uniform thickness ^59–61(Fig. 1b6). Curing of the PA gel was performed for 45 mins at two different temperatures (room temperature, 24°C, and elevated temperature, 37°C) to explore the effect of curing temperature on the efficacy of protein pattern transfer (Fig. 1b7). The cover slips were manually peeled from each other. During peeling, detachment proceeded from one edge of the sandwich. The gel remained adherent to the activated glass slide. The protein patterns were fully peeled off from the piranha-treated glass, and partially from the as-received glass. The patterns were monitored by both phase contrast and immunofluorescence microscopy. The patterning of laminin and collagen type I was done similarly, except the proteins were not oxidized.

2.3.PA gel and functionalized glass preparation

PA gel preparation and slide activation for the covalent gel attachment were performed following the protocols described elsewhere ⁴². In brief, gel solutions were obtained by mixing 8% w/v acrylamide solution (Sigma, St. Louis, MO) and 0.27% N, N’- methylenebisacrylamide (bis) solution (Midwest Scientific, MO) in 10 mM HEPES-buffered saline (Lonza, Walkersville, MD). 1:200 ammonium persulfate (10% w/v) and 1:2000 N, N, N’, N’-tetramethylethylenediamine (TEMED) (both from Bio-Rad Laboratories, Hercules, CA) were used as the initiator and catalyst for polymerization process, respectively. AFM (Asylum Research, Santa Barbara, CA) with a pre-calibrated silicon nitride tip was used to characterize the stiffness of PA gels in phosphate-buffered saline (PBS; Lonza, Walkersville, MD) solution ^{62, 63}. The nominal stiffness of the PA gels was 20 kPa ^{33, 36}. Glass slides were chemically activated to ensure covalent binding of the hydrogel as described earlier ⁴². Briefly, slides were treated with 3- Aminopropyltrymethoxysilane (ATS) from Sigma, MO for 7 min at room temperature. Followed by the removal of the ATS completely with DI water rinse, slides were treated with 0.5% Glutaraldehyde (70% Glutaraldehyde from Polysciences, Inc, diluted in PBS) solution for 30 min.

2.4. Cell Culture and Trypan Blue Live-dead Assay

Normal fibroblast cells (ATCC, Manassas, VA) were cultured in a medium with 90% Dulbecco’s Modified Eagle’s Medium (DMEM; ATCC) and 10% fetal bovine serum (FBS; ATCC) ^{33, 36, 64}. The 0.4% trypan blue solution (Sigma–Aldrich, St. Louis, MO) was applied in the cell medium to test the biocompatibility of the micro-patterned gel substrate with cells.

2.5. Immuno-fluorescent-staining of Fibronectin (FN)

PDMS, intermediate glass and PA gel with FN patterns were immunolabeled right after the protein transfer in each step. After rinsing with PBS 3 times, the substrates with FN were blocked with Image-iT FX signal enhancer (Invitrogen, CA) for 30 min to eliminate non-specific proteins binding. Samples were incubated in monoclonal anti-human fibronectin (Sigma-Aldrich, MO) at a 1:100 dilution in PBS at room temperature for 45 min. The substrates were then incubated for 30 min alexa fluor™ 488 goat anti-mouse IgG (Invitrogen, CA) at a 1:200 dilution in PBS at room temperature. The immunolabeled samples were mounted in ProLong gold antifade reagent (Invitrogen, CA) to prevent photobleaching. All the samples were imaged either using the Zeiss LSM 710 confocal scanning laser microscope (Carl Zeiss, Inc.,) at 63×, NA 1.4 oil immersion objective or Olympus IX81 microscope.

2.6.Functionalization of AFM particle tip for protein (FN) attachment

In order to determine the adhesion strength between fibronectin and glass substrates, a particle tip AFM probe was used. The Silicon Nitride (Si₃N₄) AFM cantilevers with a well-defined spherical SiO₂ particle tip (1 μm diameter coated with a 30 nm gold layer) were chosen (Novascan Technologies, Inc., Ames, IA). The gold-coated microsphere was functionalized to ensure its attachment with FN through FN-MUA-EDC+NHS-Au covalent bonding. The functionalization process consists of: (a) self-assembly of a layer of MUA (11-mercaptoundecanoic acid) on the gold coated probe by soaking it in a 1 mM MUA (in PBS) solution for 30 min, (b) immersion of the probe in a mixer of 75 mM EDC (1-ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride) and 15 mM NHS (N-hydroxysuccinimide) in PBS for 15 min, and (c) exposure of the probe to the 50 μg/ml fibronectin solution in PBS for an hour ⁶⁵. MUA, EDC and NHS were obtained from Sigma-Aldrich, MO.

2.7.Adhesion force measurement experiments using particle tip AFM

All adhesion force measurement experiments were performed using AFM (MFP 3D, Asylum Research, Santa Barbara, CA) and in PBS solution. The particle AFM probes were properly functionalized as described in previous section. Prior to the experiments, functionalized probes were stored at 4°C. Multiple functionalized probes were prepared at the same time to check the consistency between experiments. The spring constant (k) of the AFM cantilevers was specified by the manufacturer as 0.12 N/m^66–68. Calibration using the thermal tuning method ^{69, 70} yielded the nominal value of k = 0.123 N/m. All experiments were performed with the parameters tabulated in Table S1 and at room temperature. The probes were inspected with an Olympus IX81 microscope to confirm their integrity before experiments. The force measurements were carried out at randomly selected locations for every sample and a 5 × 2 array of force curves were obtained for each location. In order to measure the adhesion strength between FN and glass (piranha treated hydrophilic glass and as received glass), a drop of PBS solution was deposited on the glass surface. The AFM tip covalently bonded with FN was immersed in the drop and brought in contact with the glass surface. To ensure the complete contact between FN and glass, the AFM tip was bent on the glass surface until reaching a compressive force of 12 nN. The tip was then retracted while the AFM cantilever deflection was monitored to calculate the force of interaction between the tip and the glass. The maximum force prior to detachment was considered as a measure of adhesion between the glass and FN. All of the AFM parameters were kept constant during the adhesion experiments.

3.1. Micro-patterning Fibronectin (FN), Laminin (LN) and Collagen I (CN) on PA Gels

Figs. 2a and 2b show the patterns and surface topography of the Si master measured by profilometer (Alpha-step IQ, KLA Tencor). The nominal thickness of the SU-8 photoresist layer is 1.98 ± 0.05 μm. The structures are well defined as is evidenced from their sharp edges and flat surfaces (Fig. 2b). Micrographs of the well-defined geometric patterns on silicone elastomers (PDMS) are shown in Fig. 2c. As a demonstration, a typical immune-florescent micrograph of the FN patterns transferred onto PA gel are displayed in Fig. 2d3, where one of the glass slides was functionalized for bonding with PA gel, the other was as-received (not piranha treated). Patterning of laminin and collagen type I on gel was carried out similarly. The pattern consists of parallel FN lines, 25 μm wide with 25μm spacing between them. However, there are discontinuities in FN line patterns along the center region on the gel surface, i.e., not all the proteins were transferred successfully. To investigate at which step the protein patterns lost their integrity, immuno-fluorescence staining of FN patterns were performed on PDMS, glass and PA gels. Images from PDMS and glass substrates revealed uniform and 100% deposition of FN patterns (no defects) (Fig. 2d1 and 2d2). The discontinuous regions appeared only on the PA gel (Fig. 2d3). Based on these step-wise examinations, we believe that the loss of patterns integrity occurs when the as received glass slide is peeled off from the gel. The effectiveness of the pattern transfer to the gel during the peeling process is determined by the relative adhesion between FN and glass vs. adhesion between FN and gel. If FN-glass adhesion is weaker than that between FN and gel, FN will stay with the gel. Otherwise, only part of the FN pattern can be transferred to the gel. In the following sections, we explain this partial pattern transfer using the concepts of fracture and interface mechanics.

3.2. Gel/glass Separation Experiments Reveal 3 types of Defects

To explore the mechanism of defects, we designed and performed a series of gel-glass separation experiments with the peeling force applied at directions as shown in Figs. 3a and 3b. In all cases, the peeling force is orthogonal to the glass surface. But the location of the force is such that the peeling proceeds either along the pattern lines (Fig. 3a) or perpendicular to the lines (Fig. 3b). We found, for wide (100 μm) patterns, FN fails to transfer to the gel in the mid region of the patterns. The FN-free region is at the center of the pattern when the peeling proceeds along the pattern lines (Fig. 3a and 4a1), but is shifted from the center towards the edge (along the direction of peeling) when peeling is orthogonal to the lines (Fig. 3b and 4b2). Narrow line patterns (5 μm wide) have regions where patterns are either completely transferred to gel or they completely remained on glass (Fig. 3c).

3.3. Interfacial Fracture Mechanics Interpretation of FN Defects

The peeling of glass from gel can be considered as an interface fracture problem, where an advancing crack separates the two. Pattern transfer involves detachment or debonding of FN from intermediate glass. Such detachment processes have been studied in the literature as an interface fracture between two solids forming the interface ^71–74. In the case of the gel-glass system, we note the following: (1) gel is significantly more compliant than glass, i.e., there is a large elastic modulus mismatch between the two materials; (2) adhesion between FN and gel is likely to be stronger compared to that between FN and glass. Recall that FN is oxidized to contain the aldehydes group which allows covalent conjugation of FN with the gels; (3) the thickness of FN layer is on the order of few nanometers (Fig. 5c and 5d), orders smaller than the gel layer thickness which is on the order of 100 μm. Thus FN and gel can be considered as one combined layer. Failure of pattern transfer implies that part of the gel remains with the glass together with FN during the peeling process (Fig. S2); (4) the spaces between the FN patterns do not adhere to the gel, i.e., these regions detach from the glass surface during peeling without any resistance. In the current pattern, they are the linear spaces between the FN lines and have the same widths as the FN pattern widths (e.g., 10 μm wide linear space when the FN line is 10μm wide). These regions act as entrapped cracks and generate crack tip stresses on the FN-coated bonded regions during peeling.

During peeling, a force is applied orthogonal to the glass. As a result, the gel stretches orthogonal to the glass and tends to shrink laterally, in-plane to the glass, due to Poisson’s effect. However, the glass substrate, being much harder than gel, restrains the lateral shrinking resulting in shear stresses along the glass-gel interface. The stress gets intensified at the tip of the cracks, i.e., the cracks are loaded in mode II (shear mode) in addition to mode I (opening mode) ^{75, 76}. Thus, the crack tips are loaded in mixed mode ^{77, 78}. If the compliance of the gel and the glass were similar, then the crack would be loaded in pure mode I only, and thus the debonding or the crack growth would proceed along the interface resulting in complete pattern transfer. Due to mode mixity, the crack tip tends to kink into the gel to propagate along the maximum resultant mode I (opening) direction. Two competing factors determine this kinking, namely, the interface toughness (depends on the strength of adhesion between FN and glass) and the gel toughness. Interface toughness also depends on the degree of mode mixity. In general, the higher the ratio of mode II to mode I components, the higher is the interface toughness. If the interface strength is weak, then the crack is expected to grow along the interface. If the strength is higher, the crack will kink into the gel and run parallel to the interface. For an intermediate strength, the crack may grow along the interface and then kink due to local interfacial imperfections (Fig. 5c). We believe, when the FN is patterned on as-received glass, the adhesion is in the last category resulting in partial pattern transfer. For a given compliance mismatch between glass and gel, one potential parameter to adjust is the adhesion between FN and intermediate glass. The adhesion needs to be weak enough for interfacial crack growth, but strong enough for the pattern to be defined. We achieve this goal by using piranha-treated hydrophilic glass. A quantitative measure of adhesion between glass and FN using AFM, discussed below, shows a seven-fold reduction in adhesion strength when piranha-treated glass is used. Consequently, with appropriate adhesion strength between the FN and glass, the pattern transfer was free from defects. Earlier, AFM was used to understand the effect of hydrophilicity of different substrates on the adhesion of different proteins (except FN) ^{79, 80}. The findings demonstrated, in general, an inverse relationship between the adhesion force and the surface hydrophilicity.

3.4. AFM measurement of FN-glass adhesion–increased surface hydrophilicity of glass substrate greatly reduces the adhesion

Fig. 6a and 6b show the typical AFM force-distance curves obtained for the slides without and with piranha treatment, respectively. The maximum pulling-off force of the AFM cantilevers from the substrates was used as a measure of adhesion strength between glass and FN (Table S2, Fig. 6c). The maximum adhesive force for regular glass slide was 48.05 ± 0.94 nN, whereas that for a treated one was 7.57 ± 1.58 nN (Fig. 6d). The hydrophilic glass shows approximately seven folds lower adhesion compared to the as-received glass. These observations of decrease in adhesion force corresponding to the increase in hydrophilicity are consistent with earlier data available for other kinds of protein-substrate interactions ^79–82.

3.5. Patterning with hydrophilic glass (gel polymerization at 37 °C)

Micro-patterning of PA gels with FN were repeated using piranha-treated glass to reduce FN-glass adhesion. Polymerization of the PA gel was carried out at 37 °C. Fig. 7a-f shows both the phase contrast and immunofluorescent staining images of the FN patterns of different widths transferred onto PA gels. The patterns show no defect irrespective of the pattern widths. Similarly, the peeled-off intermediate glass did not reveal any notable FN residues. When polymerization was carried out at 24 °C, and hydrophilic glass was used, the FN patterns on PA gels were still not completely transferred (Fig. S3). Alternatively, the FN patterns on PA gels were incomplete with as-received glass and at 37 °C curing (Fig. S2). Both hydrophilic glass and polymerization at 37 °C are needed to achieve the complete FN pattern transfer. We used same procedure for patterning LN and CN, following the procedure outlined above for FN.

3.6 Culturing living cells on different ECM patterns

To demonstrate that the newly micro-patterned ECM zones (FN, LN and CN) were able to produce high quality spatial organization of cells, normal fibroblast cells (MKF) were plated and cultured on PA gel surface with patterns of varied widths. The cells were imaged on day 3 as illustrated in Fig. 8. Clearly, the cells adhered to ECM patterns only, and they form well-defined linear clusters. To test the biocompatibility of the micro-patterned gel substrate with cells, we applied 0.4% trypan blue solution in the cell medium. We found that the cells within the patterns are viable (Fig. 8c, f, i).