Hypoxia reduces TGFβ1-induced corneal keratocyte myofibroblast transformation

Materials

The medium, additives, Alexa Fluor 488 goat anti-mouse IgG, and collagenase were purchased from Invitrogen (Cat: A-11001; Carlsbad, CA). TGFβ1 was obtained from Biosource (Camarillo, CA). Monoclonal anti-α-SM actin antibody was obtained from Sigma Chemical Company (Cat: 2547; St. Louis, MO). Phospho-MAPK family antibodies were purchased from Cell Signaling Technology (Danvers, MA).

Cell culture and cell treatment

New Zealand White rabbit eyes were delivered from Pel-Frez (Rogers, AR). Rabbit corneal keratocytes were cultured as previously described [13]. Briefly, the epithelium was scraped off, the cornea was dissected, and the endothelium was wiped off. The entire stroma was put into DMEM (Dulbecco’s Modified Eagle’s Medium) with 2 mg/ml collagenase, 0.5 mg/ml hyaluronidase, and antibiotics at 37 °C overnight. The resultant isolated cells were washed once in DMEM and seeded in serum-free DMEM with non-essential amino acids, MEM vitamins, and sodium pyruvate onto 12 mm glass coverslips double coated with poly-lysine and collagen for immunocytochemistry and onto 60 mm or 100 mm Petri dishes for immunochemistry, pull down assays, and flow cytometry at 5×10⁴ cells/cm². For immunochemistry, 5 ng/ml TGFβ1 was used to treat the cells unless otherwise stated. For 4 h and 24 h treatments, drugs were added once. For 72 h treatments, drugs were added every day with medium change every other day.

Hypoxia treatment

For hypoxia treatment, cells were placed in a hypoxia chamber (Coy Lab Products Inc., Grass Lake, MI) equilibrated with 1% oxygen/5% CO₂-balance nitrogen and incubated for 4 h. For phenotypic experiments, cells were treated with hypoxia (1% oxygen for 4 h every day) for three days concurrent with or without TGFβ1. For evaluation of pSmad3 and MAPK signaling, primary isolated rabbit corneal keratocytes were treated with hypoxia (1% O₂) for 4 h with or without TGFβ1. Cells were immediately prepared for analysis after treatment.

Immunocytochemistry

Cells were washed with fresh medium and fixed in PBS containing 3% paraformaldehyde for 10 min at room temperature. Fixed cells were then permeabilized with acetone (−20 °C for 5 min). Coverslips were dried at room temperature (RT), immediately rehydrated with PBS for 5 min, blocked in PBS:goat serum (1:1) for 1 h, incubated with anti-α-SM actin (Sigma) at 1:100 in blocking medium for 1 h at RT, and washed twice with PBS and then blocking medium for 10 min. Cells were then incubated in goat anti-mouse IgG Alexa Fluor 488 at 1:20 for 40 min (Molecular Probes, Eugene, OR), washed twice with PBS for 10 min, quickly washed with dH₂O, and mounted in Prolong mounting medium (Molecular Probes) with DAPI (4'-6-Diamidino-2-phenylindole).

For western blots, cells were incubated with lysis buffer (Pierce, Rockford, ILf) for 5 min, scraped, and centrifuged at 14,000× g for 10 min. The supernatants were collected and stored at −80 °C. Protein concentrations were determined by the BCA method. Samples were separated on 10% polyacrylamide-SDS gels and electroblotted onto PVDF (Polyvinylidene Fluoride) membranes. After blocking with PBST/5% skim milk, the membrane was incubated overnight at 4 °C with primary antibodies against α-SM actin at 1:1,000 followed by peroxidase conjugated anti-mouse IgG at 1:1,000 for 1 h at room temperature. Signals were detected by ECL (enhanced luminol-based chemiluminescent). Each experiment was repeated at least three times. For statistical analysis, band density was analyzed using Un-scan-it gel analysis software (Silk Scientific, Orem, UT). The intensity of the band of interest is divided by the intensity of the internal control for all statistical analysis.

Flow cytometry

Rabbit corneal keratocytes (1.5×10⁶) in 100 mm Petri dishes were detached with 0.25 mg/ml trypsin for 3 minutes, collected, and fixed in 2% paraformaldehyde for 5 min at RT, and permeabilized in 0.02% Triton with buffered PBS. Cells were then blocked with 1:1 PBS:filtered goat serum for 1 h, stained with mouse anti-αSMA antibody (1:100) in blocking medium for 40 min at RT, washed twice with PBS, and stained with goat anti-mouse IgG Alexa Fluor 488 (Invitrogen-Molecular Probes) at 1:20 for 40 min. Cells were washed twice, and at least 5,000 cells per sample were loaded in a FACSCalibur flow cytometer (Becton Dickinson, San Jose, CA) and analyzed by CellQuest software (BD biosciences PharMingen, San Diego, CA) using manual gating according to intensity of fluorescence and cell size.

Co-immunoprecipitation

Co-immunoprecipitation (Co-IP) was performed on nuclear extracts using the ExactaCruz C: sc-45040 (Santa Cruz Biotechnology, Santa Cruz, CA) and One-Step IP-Western Kit (GenScript Corporation, Piscataway, NJ). Keratocytes (1.2×10⁶) were plated onto 100 mm plates in serum free DMEM as stated above. Monoclonal mouse anti-CBP from two companies were pooled together to increase the binding efficiency. One hundred microliters of the IP matrix with 15 μl of anti-CBP from Chemicon (Cat: MAB1133; Temecula, CA) and 15 μl of anti-CBP from R&D Systems (Cat: MAB2627; Minneapolis, MN) were incubated by end-over-end rotation overnight at 4 °C to make the IP antibody-IP matrix solution. Cells were then treated with forskolin alone or together with TGFβ1 for 4 h. Nuclear extracts were collected using the specified lysis buffer together with proteinase and phosphatase inhibitors according to the manufacturer’s protocol (Pierce). The lysate was incubated with the IP antibody-matrix by end-over-end rotation for 2 h at room temperature to pull-down CBP. Elutes from the pull-down were run on 10% polyacrylamide-SDS gels and electroblotted onto nitrocellulose membranes (Bio-Rad Laboratories, Hercules, CA). The membrane was cut in half along the 100 kDa marker. The higher molecular weight half was blotted with pooled anti-CBP antibody to check the loading. The lower molecular weight half was blotted with anti-pSmad3 (Cat: AB3226; R&D system) using GenScript One-Step Western Blot detection.

RhoA activity assay

RhoA activity was measured semi-quantitatively by pull-down assay (Cat. BK036; Cytoskeleton, Denver, CO) according to the manufacturer’s protocols. For pull-down assays, cells in 60 mm Petri dishes were lysed in 550 μl of lysis buffer and centrifuged at 10,000× g for 2 min, and 450 μl of the supernatant was added to 25 μg Rhotekin-RBD beads and incubated for 1 h at 4 °C. Samples were washed once with washing buffer, boiled with 5× Laemmli sample buffer, run on 12% polyacrylamide-SDS gels, electroblotted onto PVDF membranes, and detected with 1:2,000 mouse anti-RhoA antibody. Thirty-five microliters of supernatant was used to detect total RhoA.

Myofibroblast cell counting

For keratocytes cultured on 12 mm coverslips, five random distinct 200× microscopic fields were photographed on each coverslip. DAPI (+) cells were counted to obtain the total cell count. DAPI (+) and α-SM actin (+) cells were counted as myofibroblasts. Each experimental condition had duplicate or triplicate coverslips. Data was collected from about 750 cells for each condition in each experiment. Experiments were repeated at least three times giving a total of at least 2,000 cells counted per condition. Data are represented as mean±standard error.

Statistical analysis

Data are presented as the mean±SEM for at least three separate experiments. One way analysis of variance (ANOVA) was employed for statistical analysis with significant differences determined as p<0.05.

Hypoxia suppresses TGFβ1-induced myofibroblast differentiation

We used periodic hypoxia (only 4 h/day), which we know induces HIF1α [3], to test the effect of hypoxia on rabbit keratocyte myofibroblast transformation. Figure 1A shows that hypoxia treatment significantly reduced the TGFβ1-induced myofibroblast phenotype from 4.8%±1% to 1.2%±0.3% of the cell populations as determined from immunocytochemistry. Hypoxia alone (no TGFβ) over three days did not increase the proportion of myofibroblasts compared to the control (Figure 1A). To confirm that hypoxia significantly reduced the proportion of TGFβ1-induced myofibroblasts, we repeated the experiment using quantitative flow cytometry analysis for α-SM actin. Figure 1B shows that hypoxia treatment significantly reduced TGFβ1-induced myofibroblast differentiation from 4.0%±1.1% to 1.8%±0.4% 72 h after treatment. Overall, Figure 1 shows that hypoxia reduces TGFβ1-induced myofibroblast transformation of corneal keratocytes and that hypoxia alone does not induce the myofibroblast phenotype. These results were confirmed at the protein level by western blotting. Figure 1C shows that TGFβ1 induces a significant amount of α-SM actin expression while hypoxia suppresses TGFβ1-dependent induction of α-SM actin expression.

Hypoxia does not reduce TGFβ1-induced pSmad3 level

To explore the mechanism by which hypoxia suppresses myofibroblast transformation, we examined the effect of hypoxia on early TGFβ-induced signaling events. Smad signaling is the major TGFβ-induced signaling pathway. pSmad3 is needed for TGFβ-induced α-SM actin expression in fibroblasts [14]. pSmad3 is a transcriptional factor that is upregulated within 4 h of TGFβ1 treatment in corneal keratocytes [15]. Therefore, we examined the effect of hypoxia on TGFβ1-induced α-SM actin expression and Smad3 phosphorylation at 4 h.

Figure 2A shows that TGFβ1 increases α-SM actin expression by 1.7±0.1 fold after only 4 h. Interestingly, hypoxia alone revealed a slight but not significant increase (1.3±0.2 fold increase) in α-SM actin expression over the control. Hypoxia significantly reduced the α-SM actin expression induced by TGFβ1 at 4 h, which is consistent with the reduction in the myofibroblast phenotype (Figure 1). Figure 2B shows that TGFβ1 induced a significant increase (3.75±1.25 fold increase) in pSmad3 4 h after treatment in rabbit keratocytes compared to the control. Hypoxia by itself slightly increased pSmad3, but this increase was not significant. The combination of hypoxia with TGFβ1 had no effect on induced pSmad3 levels, indicating that inhibition of TGFβ-induced α-SM actin expression by hypoxia is not through interference with pSmad3.

Hypoxia does not alter the level of interaction of CBP and pSmad3 induced by TGFβ1

TGFβ1-induced Smad signaling needs the binding of the coactivator, CBP/p300, to exert its function. The transcription factor, HIF1α, which is induced by hypoxia, also needs CBP as a coactivator. Concurrent activation of Smads and Hif1α might lead to a competitive inhibition of pSmad3-CBP interaction. We tested this possibility by co-immunoprecipitation of CBP from nuclear extracts 4 h after treatment and then probed for pSmad3.

Figure 3 shows that hypoxia or TGFβ1 alone induced a significant increase in interaction between pSmad3 and CBP (1.8±0.4 and 2.4±0.1 fold increase, respectively). However, hypoxia+TGFβ1 (2.6±0.3 fold increase) did not significantly change the TGFβ-induced pSmad3-CBP interaction. These results indicate that the reduction of TGFβ-induced myofibroblast formation and α-SM actin expression by hypoxia is not through the reduction of pSmad3 or pSmad3-CBP interaction.

TGFβ induced pERK is not altered by hypoxia

In addition to increasing pSmad3 activation, TGFβ significantly increases phosphorylation of ERK at 4 h. Both TGFβ-induced αSM actin and pERK are significantly reduced by the MEK (mitogen-activated protein kinase kinase) inhibitor, U0126, indicating an important role for pERK in inducing myofibroblasts [15]. Figure 4 shows that hypoxia alone significantly reduced pERK below the control level in rabbit keratocytes. Interestingly, hypoxia did not reduce TGFβ-induced ERK phosphorylation. Instead, pERK was significantly increased relative to TGFβ alone. Figure 4 also shows that there was no change in total ERK, pJNK (phosphorylated jun kinase), or phospho-p38 levels in all experimental groups at 4 h after treatment. These results indicate that hypoxia inhibition of TGFβ1-induced α-SM actin expression in rabbit keratocytes is not through interference with ERK activation.

Hypoxia reduces TGFβ-induced active RhoA

TGFβ can also activate the small GTP-binding protein, RhoA. The RhoA pathway has a role in α-SM actin expression in rabbit corneal myofibroblasts [16]. Recently, we found that in rabbit keratocytes, TGFβ1 transiently induces GTP-RhoA between 5 and 15 min, peaking at approximately 7 min, and that interfering with RhoA activation by ROCK (Rho-associated coil-containing protein kinase) inhibition reduced αSM actin expression after 4 h of TGFβ exposure [15]. Here, we tested the effect of hypoxia on GTP-RhoA following 7 min of TGFβ1 treatment. Cells were pre-equilibrated in hypoxic media for 4 h. TGFβ was added to the cells within the hypoxia chamber, and cells harvested for western blot analysis 7 min later. Figure 5 shows that TGFβ1 significantly increased the GTP-RhoA level (1.38±0.15 fold increase) after 7 min compared to the control. Hypoxia alone did not significantly change GTP-RhoA. However, TGFβ+hypoxia significantly reduced the GTP-RhoA level to less than the control level.