TGF-β1 promotes cell barrier function upon maturation of corneal endothelial cells

Human corneal endothelial cells (HCECs) easily become fibroblastic-like when cultured, rendering them unsuitable for tissue engineering of the cornea. Transforming growth factor β (TGF-β) could be a key factor in this phenomenon; however, TGF-β is also known to maintain the endothelium in a quiescent state in vivo. This work aimed to compare the effects of TGF-β1 on the phenotype of HCECs during the proliferation and maturation phases. Our results show that addition of TGF-β1 during the active proliferation phase produced fibroblastic HCECs and loss of the cell junction markers ZO-1 and n-cadherin, independent from the presence of epidermal growth factor (EGF). By contrast, addition of TGF-β1 in maturation media containing few mitogens led to an endothelial phenotype and functional cell junctions as HCECs developed a high trans-endothelial resistance. Furthermore, addition of AG-1478, an epithelial growth factor receptor inhibitor, enhanced the gain of the endothelial phenotype and cell barrier function. Overall, these results show that TGF-β1 can be used to promote the formation of a typical leaky endothelial barrier during the maturation phase of cultured HCECs. A two-phase culture of HCECs using distinct proliferation and maturation media could also be key for developing ideal HCEC culture conditions.

factor receptor. Cells were cultured until confluency and immunofluorescence was then performed. c) Dose-response assessment for TGF-β1. Analysis of the ratio of the area covered by α-SMA and total actin in two different populations of HCECs (grey bars: Population 1; black bars: population 2) depending of the TGF-β1 concentration in the culture medium. One-way ANOVA followed by Tukey's multiple comparison tests were performed. Results are presented as mean ± standard deviation for each condition and population. d) Dose-response assessment for SB431542. Analysis of the ration of the area covered by α-SMA and total actin in one population of HCECs depending of the SB431542 concentration in a culture medium containing 2ng/ml TGF-β1 (P+TGF). One-way ANOVA followed by Tukey's multiple comparison tests were performed. Results are presented as mean ± standard deviation for each condition. e) Dose-response assessment for AG-1478. Analysis of absorbance following a MTS proliferation assay in cells cultured in a basal proliferation medium (P+EGF) and exposed to growing AG-1478 concentrations for 72h. One-way ANOVA followed by Tukey's multiple comparison tests were performed. Results are presented as mean ± standard deviation for each condition. **p<0.005; ***p<0.0005; ****p<0.0001 α-SMA: α-smooth muscle actin; TGF-β1: transforming growth factor β-1; EGF: epidermal growth factor.
Determination of the ratio of the area covered by α-SMA and total actin was made using Image J Supplementary data SREP-17-46670_R1 -Beaulieu-Leclerc et al.

4
software. Mean ratio and standard deviation were calculated for each concentration. The ideal dose was determined by the concentration that induced higher area of α-SMA without inducing cell mortality.
A similar protocol was used to determine the ideal concentration of SB431542, an inhibitor of type I transforming growth factor receptor (TGF-beta RI Kinase Inhibitor, SB431542 hydrate, Sigma-Aldrich). After determining the ideal concentration of TGF-β1, HCECs (n=1 population, 2 coverslips/condition, 3 micrographs/coverslip) were cultured in basal proliferation medium (P) containing 2 ng/ml of TGF-β1 and supplemented with SB421542 (0, 1, 10 and 100 µM) until confluency. Immunofluorescence against α-SMA and total actin was analyzed using image J software as described in the preceding paragraph. The ideal dose was determined by the concentration that blocked the most the expression α-SMA induced by TGF-β1 without inducing cell mortality.
Dose-response to AG-1478 hydrochlodride (Tocris Bioscience, Minneapolis, MN), an inhibitor of EGF receptor, was made using a MTS proliferation assay (CellTiter 96® Non-radioactive Cell Proliferation Assay, Promega, Madison, WI). Briefly, one primary population of HCECs (4 wells per condition) was plated onto 96 wells-plates at a seeding density of 15 000 cells/cm 2 in the basal maturation medium (M). After 24h of adhesion, medium was changed to a basal proliferation medium (P) supplemented with EGF 5 ng/ml and increasing concentrations of AG-1478 (0.01, 0.1, 1, 10 and 100 µM). After 72h of culture, 20 µl of MTS reagent was added to the medium. After 2h of incubation in the dark, absorbance at 490 nm was acquired using a microplate reader (Bio-Rad Model 550 microplate reader, Mississauga, Ontario, Canada) and Microplate Manager 5.0 software. Mean absorbance and standard deviation were calculated for each concentration of AG-1478 (4 wells per condition). The ideal concentration was determined by the concentration that blocked the most proliferation induced by EGF without inducing cell mortality.

Statistical analysis:
Results are presented as mean of all measurements and standard deviation (SD). Statistical significance was calculated with one-way ANOVAs, followed by Tukey's multiple comparisons test using GraphPadPrism©. A p<0,05 was considered significant.
5 assessed with dose-response assays. Response to various concentrations of TGF-β1 and SB431542 was evaluated by the ratio of the area covered by α-SMA on immunofluorescence micrographs to the area covered by total actin.
The TGF-β1 ideal concentration was determined to be 2 ng/ml ( fig.S1a and c) because it induced the most important expression of α-SMA in both cell population that were tested. It was higher than with 0.2 ng/ml, although not statistically different for population 2. Cells proliferated less in 20 ng/ml of TGF-β1 (experimenter's observation). 10 µM of SB431542 was sufficient to totally block expression of α-SMA induced by 2 ng/ml TGF-β1 ( fig.S1b and d). 7 a) Circularity of HCECS after the proliferation phase. Cells were grown until confluency in a proliferation medium containing 5 ng/ml of EGF plus 2 ng/ml TGF-β1, 10 µM SB431542 or 1 µM Ag-1478. One-way ANOVA followed by Tukey's multiple comparison tests were performed.
Results are presented as mean ± standard deviation for each condition. b) Circularity of HCECS after the maturation phase. Cells were grown until confluency in a proliferation medium containing 5 ng/ml of EGF (P). They were then matured for 7 days in a maturation medium (M) and various additives. Cells in the M+TGF/M condition were matured for 7 days in M+TGF-β1 then for another 7 days in the M medium only (a total of 14 days of maturation). One-way ANOVA followed by Tukey's multiple comparison tests were performed.
Results are presented as mean ± standard deviation for each condition. * p<0.05

Supplementary methods S2
HCECs (n=4 populations, 2 or 3 coverslips per condition, 3 micrographs per coverslip) were cultured as described before in different proliferation and maturation media (see table 1 in the main article). Phase contrast images (0 and 7 or 14 days post-confluency) were used to define cell circularity. At least 100 randomly selected cells per condition were measured. Their area and perimeter were acquired with ImageJ software and then circularity was determined by the formula: Hexagonal cells, an hallmark of endothelial morphology, have a circularity close to 1.0 and cells with fibroblastic morphology get a score closer to zero. Statistical difference between the results was calculated with Tukey's multiple comparison test, where p<0.05 was considered statistically significant.

Supplementary results S2
Measurement of cell circularity after the proliferation ( fig. S2a) (0 days post-confluency) show that addition of TGF-β1 to the medium tended to lower the circularity, compared to the control P medium. The negative effect of TGF-β1 was also neutralised by the addition of SB431542, were the circularity is similar to the P control. However, these results are not statistically significant.
After 7 days of maturation in different additives, there were statistically significant differences in the cell circularity. First, addition of TGF-β1 to the maturation medium improved circularity when Supplementary data SREP-17-46670_R1 -Beaulieu-Leclerc et al.

8
compared to the maturation control (M) (0.7338 vs 0.5425, p=0.02). Circularity was also higher than maturation control in TGF-β1 + EGF + AG-1478 and the 14 days post-confluency conditions. Finally, addition of SB431542 to the medium counteracted the beneficial effect of TGF-β1 to the circularity. Circularity was similar to the control maturation condition and lower than all of the TGF-β1 containing conditions. This confirms the positive effect of TGF-β1 on cell circularity of maturing HCECs and shows that it can improve endothelial morphology.  Fig. 2b (a) and Fig. 2e (b)