Directed Endothelial Cell Morphogenesis in Micropatterned Gelatin Methacrylate Hydrogels

Gelatin methacrylate synthesis

Gelatin methacrylation was synthesized as described previously [33]. Briefly, type A porcine skin gelatin was mixed at 10% (w/v) into Dulbecco’s phosphate buffered saline (DPBS) (Sigma) at 60 °C and stirred until its components were fully dissolved. Subsequently, 8% (v/v) of methacrylic anhydride was added drop by drop to the gelatin solution and reacted at 50 °C for 3 hours to form GelMA solutions. The solutions were then diluted and dialyzed against distilled water by using 12–14 kDa cutoff dialysis tubing water at 40 °C for one week. The solution was finally lyophilized for 4 days to produce white porous foam. The produced foam was stored at −80 °C for the experimental use.

Cell culture preparation

Green fluorescent protein (GFP)-expressing human umbilical vein endothelial cells (HUVECs) were used in this study. The cells were maintained in physiological condition in 37 °C in a humidified, 5% CO₂–95% air atmosphere and cultured in T-75 flasks in endothelial cells basal medium (EBM-2; Lonza) supplemented with endothelial growth BulletKit (EGM-2; Lonza), 2% fetal bovine serum, vascular endothelial growth factor (VEGF), human fibroblast growth factor (hFGF), insulin growth factor (R3-IGF-I), human epidermal growth factor (hEGF), hydrocortisone, ascorbic acid, heparin, and 1 % GA-1000 (gentamicin and amphotericin B). The media was changed every two week and the cells were passaged once per week.

Microfabrication of patterned cell-laden GelMA hydrogel

Prior to cell encapsulation experiments, glass slides were treated with 3-(trimethoxysilyl)propyl methacrylate (TMSPMA) (Sigma) and then coated with a layer of 4.5% (w/v) poly(2-hydroxyethyl methacrylate) (PolyHEMA) (Sigma) solution in pure ethanol. The slides were then air dried for 3 days and sterilized by using UV light prior to encapsulation (Figure 1(A)). GelMA prepolymer solution was prepared by dissolving 5% (w/v) GelMA macromers in PBS containing 0.5% (w/v) photoinitiator ((2-hydroxy-1-(4-(hydroxyethoxy) phenyl)-2-methyl-1-propanone, Irgacure 2959, CIBA Chemicals). The solution was then placed inside an 80 °C oven until it fully dissolved.

To prepare cell-laden micropatterned hydrogel constructs, HUVECs were trypsinized, counted and gently mixed in GelMA prepolymer solution with a final concentration of 2.5×10⁷ cells/ml of GelMA solution. A 15 µl of prepolymer solution containing cells was then dropped on a petri dish in between the spacers with desirable height and a PolyHEMA-coated glass slide was inverted on the petri dish (Figure 1(B)). Subsequently, the photomask layout (designed by AutoCAD software) was placed on top of the glass slide and the cell-encapsulated prepolymer solution was exposed to ultraviolet (UV) light (360–480 nm, 6.9 mW/cm² intensity) for 30 s to develop cell-laden micropatterned hydrogel constructs. Upon UV exposure, the polyHEMA-coated glass slides containing patterned, crosslinked hydrogel constructs were submerged into warm DPBS to develop the final hydrogel microconstructs (Figure 1(C)). We used a 1 cm × 1 cm single photomaks layout to develop multiple parallels 8 mm-long by 50 µm-wide rectangular micropatterned constructs surrounded by a 1 mm-wide unpatterned region. Micropatterned GelMA constructs with different geometries were developed by changing the spacer height (H, 50, 100, 150) (Figure 1B). Following encapsulation, the microfabricated cell-laden hydrogel constructs were cultured in 24-well cell culture plates and supplemented with endothelial media for 1, 3 and 5 days prior to biological analyses.

Cell viability assay

Live/Dead Assay Kit (Invitrogen) that includes calcein AM (Cl) and ethidium homodimer (ETD) was used to quantify the percentage of viable cells after 1, 3, and 5 days of culture. A solution of 0.5 µl/ml Cl and 2 µl/ml ETD in DPBS were first prepared. The cell-laden GelMA gels were then rinsed with warm DPBS after which 200 µl of live/dead solution was added to each well. The well-plate was placed inside the incubator for 30 minutes and then imaged for live (green stain) and dead (red stain) cells using an inverted fluorescence microscope (Nikon TE 2000-U, Nikon instruments Inc., USA) with 10X magnifications. As HUVECs were expressing GFP, the total number of the cells (green stain) and the number of the dead cells stained with ETD, were counted within each image using ImageJ v. 1.4 software (NIH). Then the number of the live cells was calculated by subtracting the number of the dead cells from the total number of the cells within each image. Finally, the cell viability was calculated based on the number of live cells divided by total cell number. At least three samples were prepared for each condition and three images were taken from each sample. The quantified values reported based on the average ± standard deviation (SD).

Quantification of cell proliferation

Cell proliferation within the micropatterned and unpatterned regions was quantified by counting DAPI-stained nuclei over different culture time. To stain the cells nuclei, encapsulated within the hydrogel, the samples were rinsed in DPBS and fixed in 4% paraformaldehyde (PF) (Sigma) solution in DPBS. A 0.1 % (v/v) DAPI (4',6-diamidino-2-phenylindole) (Sigma) solution in DPBS was prepared and added to each sample. The samples were then placed inside the incubator for 10 minutes to stain the cells nuclei, after which the samples were washed three times in DPBS. The fluorescence images of the cells’ nuclei were captured in 3–5 individual fields of micropatterned and unpatterned regions within each sample. The number of cell nuclei within each field was counted using NIH ImageJ software to assess the proliferation of HUVECs at different culture times (days 1, 3, and 5). At least three samples were stained for each condition.

Quantification of cell alignment

After 5 days of culture, the samples were fixed in 4% PF solution in DPBS and the cells’ nuclei were stained with DAPI as described previously. Alignment analysis was performed by using fluorescence images of DAPI-stained nuclei (3–5 images for each sample). The shape of individual nucleus within each image was fitted with an elliptical shape using built-in functions of ImageJ software. Then, the angles of the nuclei major elliptical axis with respect to the horizontal axis were defined as the cells nuclei alignment angles. Following measurement of individual nucleus alignment, all the nuclei alignment angles were normalized with respect to the average of all nuclei alignment within each image [32]. The normalized alignment angles were grouped in 10 degrees increments to compare the nuclei alignment within patterned or unpatterned regions of each sample. Three samples were analyzed for each condition.

Immunofluorescence for cytoskeletal organization and angiogenic markers

Confocal microscopy was used to assess actin cytoskeleton organization, CD31 (PECAM-1), and VE-cadherein expression of cells encapsulated within the micropatterned regions of GelMA hydrogel constructs. The cell-laden GelMA gels were washed three times in DPBS and fixed in 4% PF solution in DPBS for 30 minutes. The samples were then soaked in 0.1% Triton X-100 in DPBS for 30 minutes to permeabilize the cells membrane. For actin cytoskeleton staining, the gels were blocked in 1% bovine serum albumin (BSA) in DPBS for 1 h. Then, a 1/40 dilution of Alexa Fluor-594 phalloidin (Invitrogen) in 0.1% BSA was added to the samples to stain the actin cytoskeleton following 45 minutes incubation at room temperature. For CD31 and VE-cadherin staining, upon fixation, the samples were blocked in 10% horse serum in DPBS for 1 h after cell fixation. To stain for CD31, a 1/100 dilution of monoclonal mouse anti-CD31 antibody (Abcam) in 10% horse serum was added to the cell-laden gels and the samples were kept at 4 °C for 24 h. The gels were then washed in DPBS three times with 1 h intervals in between the washing steps. After primary antibody staining, the samples were incubated in 1/200 dilution of Alexa Fluor-594 conjugated goat anti-mouse secondary antibody (Abcam) in 10% horse serum in DPBS for 6 h at room temperature (25 °C). Upon completion of staining, the samples were washed three times in DPBS. Similar procedure was used to stain VE-cadherin with the exception of using monoclonal rabbit anti-VE-cadherin antibody as the primary antibody and Alexa Fluor-594 conjugated goat anti-rabbit antibody as the secondary antibody. Following actin, CD31 and VE-cadherein staining, DAPI staining was also performed as described previously and the samples were mounted on Fluoromount-G solution (Southern Biotech) for confocal imaging.

Inverted laser scanning confocal microscope (SP5 X MP, Leica) and Leica application suite (LAS) software were used to acquire 3D serial section images of cord structures within GelMA hydrogels at different depths of foci. Diode laser (405nm) and white light laser (470–670nm) were used to image cell nuclei and actin, CD31 and VE-cadherin respectively. Acquired confocal images were stacked with Imaris software (Bitplane, Zurich, Swiss) to build the final 3D projected images.

Statistical Analysis

Statistical significance was performed using one way analysis of variance (ANOVA) followed by Bonferroni’s post-hoc test for multiple comparisons of different conditions using Graph Pad Prism 5.0 (San Diego, USA) statistical software. Data are presented in mean ± SD.

Microfabrication of cell-laden GelMA hydrogel constructs with variable geometries

The micropatterned constructs were developed with 5 % GelMA with high methacrylation degree (~ 70 %) due to its mechanical robustness and suitability for micropatterning [33]. For the first experiment, we used 0.02 % fluorescein dye to stain and visualize the GelMA hydrogel containing HUVECs on day 0 of culture. A cell-encapsulated micropatterned construct network surrounded by an unpatterned region was successfully formed upon UV exposure and hydrogel crosslinking process (Figure 2A). Micropatterned constructs with different geometries were formed by changing the height of the spacers (i.e. 50, 100, and 150 µm). The cross-sectional views of the fabricated microconstructs (Figure 2B) demonstrated the formation of structures with rounded cross-sectional features due to the UV light diffraction during the microfabrication process. Microconstruct widths and heights were measured using the phase contrast images. The height of cell-laden microconstructs were varied from 56.8±6.6 µm (n=10) to 97.3±7.2 µm (n=17) and 156.4±12.4 µm (n=18) when the spacer heights were changed from 50 µm to 100 µm and 150 µm, respectively. Despite using the same mask layout, the microconstruct width increased as a function of its height (Figure 2C). Fluorescence and phase contrast images demonstrated successful cell encapsulation within microfabricated constructs of all the three different heights at day 0 of cell encapsulation. HUVECs were mostly round upon encapsulation and were homogeneously distributed within the 3D microenvironment of the hydrogel constructs. Overall, these data clearly indicated the formation of 3D cell-laden micropatterned hydrogel constructs with high fidelity and reproducibility.

Cell viability and proliferation within Microfabricated GelMA constructs

We assessed viability and proliferation of HUVECs within micropatterned constructs as well as the unpatterned regions of GelMA hydrogels after 1, 3 and 5 days of culture. Cell viability was above 90 % within the microconstructs for at least 5 days of culture. Interestingly, we detected a slight loss of cell viability as a function of hydrogel height and culture time within the unpatterned regions (Supplementary Figure S1). Overall, high cellular viability within the microconstructs indicated that the presence of the photoinitiator in GelMA hydrogel along with photolithography process (i.e. UV exposure) had minimal effect on endothelial cell survival.

Phase contrast and fluorescence images of GFP-expressing HUVECs demonstrated that the cells exhibited a rounded morphology on day 1 after cell encapsulation. However, after longer culture periods (days 3 and 5), interconnected networks of neighboring cells were formed within the hydrogel microconstructs (Figure 3A). The cells remained within the micropatterned regions and did not migrate to the glass slide surface due to the cell-repellent properties of the polyHEMA coating. Within the micropatterned constructs, cells exhibited an elongated and spindle-like morphology and aligned along the direction of the microconstructs and formed organized networks with neighboring cells on day 5 of culture. Conversely, within the unpatterned regions, cells were randomly distributed within the 3D microenvironment of the hydrogel construct (Figure 3A). Quantification of cell number within GelMA constructs demonstrated that the average cell number was significantly increased with culture time when the microconstruct height was less than 100 um, demonstrating cellular proliferation within microconstructs with lower heights (Figure 3B). However, there were no significant changes in cell number within the unpatterned regions of hydrogels (Figure 3C).

Cellular alignment with micropatterned hydrogels

To assess the role of microconstructs geometries on cellular alignment, we quantified HUVECs nuclei alignment within patterned and unpatterned regions of the hydrogels as a function of construct height after 5 days of culture. Cells nuclei alignment was grouped in 10° increments to compare the alignment distribution in different conditions. As shown in Figure 3(A) no alignment was observed within the micropatterned constructs on day 1 of encapsulation and nuclei alignment began after 3 days of encapsulation. On day 5 of culture, the cells filled the microconstructs and mostly aligned along the major axis of the microconstructs. In addition, we found dependence of cellular alignment on geometrical features of the patterns layout. As indicated in Figure 4A–C, decreasing the height increased the percentage of nuclear alignment within the micropatterned constructs. For instance, in 50 µm-and 100 µm-high microconstructs, approximately 55 % of the cells (n=389, n=348) aligned within the 0–10° increment angle, which was significantly higher than 39.9±13.7 % (n=453) for 150 µm-high microconstructs (p<0.001) . In contrast to micropatterned regions, the cells nuclei were of randomly oriented within the unpatterned regions (Figure 4D–F).

3D cord formation of endothelial cells within micropatterned GelMA hydrogels

As shown previously (Figure 3), HUVECs cell spread and aligned along the length of the microconstructs after 5 days of culture. The expressions of the endothelial cell specific markers including CD31 and VE-cadherin, as well as F-actin within the patterned regions were evaluated. We chose 100 µm-high constructs as a representative feature for the primary studies. Both CD31 and VE-cadherin are endothelial intercellular junctional proteins; CD31, also known as platelet endothelial cell adhesion molecule 1 (PECAM-1), makes up a large portion of endothelial cell intercellular junctions and is involved in angiogenesis [34, 35]. Similarly, VE-cadherin is well known to have a substantial role in endothelial cell morphogenesis, angiogenesis, and vascular stability [36, 37]. Confocal images of the cell-laden GelMA hydrogels revealed expression of CD31 and VE-cadherin within the micropatterned hydrogel constructs—an essential requirement to achieve a functional vasculature—only after 5 days of culture (Figure 5A–D). The results of actin staining demonstrated that actin fibers aligned along the direction of the microconstructs. As shown in Figure 5E–F, the cells reorganized within the hydrogel microconstructs to form cords with circular/elliptical cross sections (Arrows). To study the role micropatterned hydrogel construct on endothelial cells cord formation, we repeated the experiments using FITC conjugated GelMA hydrogel. As shown in Figure 6, confocal images revealed the presence of the hydrogel within the inner layer of cord structures after 5 days of culture (Arrows). The cells reorganized toward the outer surfaces and the periphery of the hydrogel construct to form cord structures. These observations indicate the significant role of patterned hydrogel microconstructs in guiding the assembly of endothelial cells toward cord formation.

To investigate the process of endothelial morphogenesis over the course of time within hydrogel constructs, we performed experiments with longer period of culture time (15 days) using samples of three different heights (50, 100 and 150 µm). Initially, the cells were immediately fixed upon encapsulation and imaged using confocal microscopy for GFP-expressing HUVECs. 3D projected images illustrated that GFP-expressing HUVECs were comprised of round morphology and randomly distributed throughout the hydrogel layer for all the three different heights (Figure 7A–F) consistent with our previous observation shown in Figure 3. However, regardless of the height, confocal images demonstrated that the cells reorganized toward the periphery of the microconstructs, and formed stable cords after 15 days of culture (Figure 7G–L). Interestingly, the cord structures were stable for two weeks of culture time. Specifically, for the 50 µm-high microconstructs, the cross-section of the cord structure resembled a semi-elliptical shape (Figure 7G and 7H), while cords of the 100 µm-high microconstructs comprised a more circular cross-section (Figure 7I and 7J). Although the cord structures were retained inside 150 µm-high microconstructs, the cells were poorly organized compared to 50 µm and 100 µm-high micropatterned constructs (Figure 7K and 7L). Notably, confocal images clearly demonstrated that the actin fibers were distributed in a more organized fashion within the 100 µm-high micropatterned hydrogel constructs compared to the other geometrical features (i.e. 50 µm and 150 µm). These data suggest that 100 µm-high constructs provide the optimal microenvironment in terms of geometrical feature for the formation of circular and stable cords.