Activation of Prostaglandin FP and EP2 Receptors Differently Modulates Myofibroblast Transition in a Model of Adult Primary Human Trabecular Meshwork Cells

Abstract

Purpose: Prostaglandin F2α analogues are the first-line medication for the treatment of ocular hypertension (OHT), and prostanoid EP2 receptor agonists are under clinical development for this indication. The goal of this study was to investigate the effects of F prostanoid (FP) and EP2 receptor activation on the myofibroblast transition of primary trabecular meshwork (TM) cells, which could be a causal mechanism of TM dysfunction in glaucoma.

Methods: Human primary TM cells were treated with either latanoprost or butaprost and TGF-β2. Trabecular meshwork contraction was measured in a three-dimensional (3D) TM cell–populated collagen gel (CPCG) model. Expression of α-smooth muscle actin (α-SMA) and phosphorylation of myosin light chain (MLC) were determined by Western blot. Assembly of actin stress fibers and collagen deposition were evaluated by immunocytochemistry. Involvement of p38, extracellular signal-regulated kinase (ERK), and Rho-associated kinase (ROCK) pathways as well as matrix metalloproteinase activation was tested with specific inhibitors.

Results: In one source of validated adult TM cells, latanoprost induced cell contraction as observed by CPCG surface reduction and increased actin polymerization, α-SMA expression, and MLC phosphorylation, whereas butaprost inhibited TGF-β2–induced CPCG contraction, actin polymerization, and MLC phosphorylation. Both agonists inhibited TGF-β2–dependent collagen deposition. The latanoprost effects were mediated by p38 pathway.

Conclusions: Latanoprost decreased TM collagen accumulation but promoted a contractile phenotype in a source of adult TM cells that could modulate the conventional outflow pathway. In contrast, butaprost attenuated both TM contraction and collagen deposition induced by TGF-β2, thereby inhibiting myofibroblast transition of TM cells. These results open new perspectives for the management of OHT.

Primary open-angle glaucoma (POAG) is the leading cause of irreversible blindness worldwide.¹ Elevation of intraocular pressure (IOP) is the most critical risk factor for glaucoma and is associated with a dysfunction of the trabecular meshwork (TM) responsible for an increased resistance to the aqueous humor (AH) outflow. The TM of POAG patients is characterized by a decreased TM cell number,^2,3 an accumulation of extracellular matrix (ECM),^4–6 and an increased rigidity⁷ that may account for IOP elevation. Alterations of the TM biophysical properties have also been extensively linked to the increased concentration of the profibrotic transforming growth factor (TGF)-β2 in the AH of glaucomatous patients.^8–10 

The TM tissue is mainly composed of ECM proteins¹¹ and possesses contractile features as highlighted by the presence of a subpopulation of resident myofibroblastic cells expressing α-smooth muscle actin (α-SMA).^12,13 Moreover, the IOP-lowering effect of agents disrupting the TM actomyosin system demonstrates that the TM contraction state regulates conventional AH outflow.¹⁴ It could therefore be hypothesized that fibrotic mechanisms may be responsible for the TM alterations observed in POAG, in which TM cells undergo a deregulated myofibroblast transition promoted by fibrotic factors like TGF-β2. Indeed, the acquisition of a myofibroblastic phenotype by resident cells is a common feature of all fibrotic pathologies and is defined by collagen accumulation and increased contractile properties.^15,16 

Latanoprost, a prostaglandin (PG) F2α analogue, is the first F prostanoid (FP) receptor agonist approved for POAG treatment and is currently used as first-line therapy.¹⁷ However, agonists of another prostanoid receptor, EP2, have also been shown to promote potent hypotensive effects in several animal models^18,19 and are currently in clinical development for the treatment of ocular hypertension (OHT).^20–22 The action of both FP and EP2 agonists seems to be mediated by an increased uveoscleral outflow pathway through activation of matrix metalloproteinases (MMP) and ECM remodeling.^18,23–26 However, their effects on the TM outflow facility, the main route for AH drainage, is still controversial.^27,28 Nonetheless, the presence of both FP and EP2 receptors on TM tissues,^29,30 as well as their potential effects on TM cell survival (Kalouche et al., unpublished observations, 2015), suggests that PG analogues may have direct physiological functions on TM cells. 

F prostanoid and EP2 are respectively qualified as contractile and relaxant receptors due to their effect on smooth muscle cell contraction.^31–35 Furthermore, EP2 receptor mediates Schlemm's canal endothelial cell relaxation³⁶ and has been shown to regulate the mechanisms of contractility and relaxation in the TM.^35,36 F prostanoid receptor has also been implicated in the development of pulmonary³⁷ and myocardial fibrosis,³⁸ while loss of EP2 receptor exacerbates lung fibrotic lesions.³⁹ This dichotomy of action between FP and EP2 receptors may thus potentially lead to different outcomes on the TM alterations observed in POAG. 

The aim of the present work was to compare the effects of FP and EP2 agonists on the myofibroblast transition of TM cells. For that purpose, TM contraction, myofibroblast markers, and collagen deposition were evaluated in the presence of TGF-β2 and PG analogues. We show that the FP agonist latanoprost promotes TM cell contraction, increases myofibroblastic markers, and decreases TGF-β2–mediated collagen deposition. On the other hand, the EP2 agonist butaprost inhibits TGF-β2–mediated myofibroblast transition of TM cells by inhibiting the TGF-β2–dependent contraction and collagen deposition. 

Methods

Reagents

Butaprost and latanoprost free acid forms and AL-8810 were purchased from Cayman (Ann Harbor, MI, USA). The inhibitors FR180204, SB203580, Y-27632 dihydrochloride, and marimastat as well as dexamethasone were from Sigma-Aldrich Corp. (St. Louis, MO, USA). Forskolin was obtained from EMD Millipore (Billerica, MA, USA). Recombinant human TGF-β2 was purchased from PeproTech (Rocky Hill, NJ, USA). The anti-human glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (1/20000) and phospho(Thr18/Ser19)-myosin light chain (MLC) (1/1000) antibodies were from Cell Signaling Technologies (Danvers, MA, USA), while anti-human α-SMA, collagen type I (1/400), vinculin, fibronectin, and myocilin (1/1000) antibodies were from Sigma-Aldrich Corp. 

Cell Culture and Treatments

Three commercially available primary human TM cells were used (ScienCell Research Laboratories, Carlsbad, CA, USA). Two cell batches (nos. 5987 and 7278) were obtained from fetal eyes (22 and 20 weeks of age, respectively), whereas a third cell batch (no. 4973) was isolated from the juxtacanalicular region of a 25-year-old Caucasian male. The cryopreserved primary human TM cells were thawed and plated on poly-L-lysine–coated surfaces and cultured in Dulbecco's modified Eagle's medium (DMEM) (Life Technologies, Carlsbad, CA, USA) supplemented with 10% fetal bovine serum (FBS), sodium pyruvate, and antibiotics (100 U/mL penicillin and 100 μg/mL streptomycin). Cells were maintained in humidified 5% CO₂ and 95% air atmosphere at 37°C. The first passage allowed cryopreservation of approximately 30 vials of TM cells in culture medium supplemented with 10% dimethyl sulfoxide (DMSO). The cells were used for experiments from passages 2 to 5. Trabecular meshwork cells were serum starved for 24 hours before pretreatment with inhibitor compounds or 0.1% DMSO as a control for 30 minutes unless otherwise stated. Dimethyl sulfoxide (0.1%), 1 μM butaprost, or 1 μM latanoprost was added to the cell medium with or without TGF-β2. 

Immunocytochemistry

The three batches of TM cells were grown on poly-L-lysine–coated 96-well μ-plates (Ibidi, Planegg / Martinsried, Germany). For the detection of collagen deposition, L-ascorbic acid 2-phosphate (Sigma-Aldrich Corp.) was added to the TM cell medium to allow collagen maturation.⁴⁰ Ninety-six hours after cell treatment with the various compounds tested, TM cells were fixed either with ice-cold methanol for the detection of collagen deposition or fibronectin, or with 4% paraformaldehyde for the detection of actin stress fibers, α-SMA, and vinculin. The cells were treated with 0.1% Triton X-100 and 3% normal goat serum (NGS) for 1 hour and then incubated overnight at +4°C with relevant antibodies in a 1% NGS solution. Wells were washed three times and incubated with appropriate Alexa Fluor–conjugated secondary antibodies (Life Technologies) and 4′,6-diamidino-2-phenylindole (DAPI) or TO-PRO-3 for 2 hours. For the detection of actin stress fibers, Alexa Fluor 546–conjugated phalloidin (Life Technologies) was incubated with DAPI for 2 hours. Plates were then rinsed in DPBS four times and wells kept in Dulbecco's phosphate buffered saline (DPBS) four times and wells kept in DPBS for fluorescence detection. Total fluorescence was measured by a Tecan M1000 plate reader (Männedorf, Switzerland) at the appropriate wavelengths. Pictures were obtained by an inverted Olympus FV1000 laser scanning confocal microscope (Tokyo, Japan). Z-sections were processed by ImageJ software (http://imagej.nih.gov/ij/; provided in the public domain by the National Institutes of Health, Bethesda, MD, USA) by maximum-intensity projection. 

Immunoblotting

For Western blot analysis, cell supernatants and cell lysis in radioimmunoprecipitation assay (RIPA) buffer supplemented with proteases and phosphatase cocktail inhibitors 2 and 3 (all from Sigma-Aldrich Corp.) were collected. After centrifugation for 10 minutes, equal amounts of proteins were separated by SDS-polyacrylamide gel electrophoresis and transferred to nitrocellulose membranes (all from Life Technologies). Membranes were blocked with Tris-buffered saline containing 0.1% Tween-20 (TBS-T) and 5% nonfat dry milk for 1 hour. Primary antibodies were incubated overnight at +4°C in a 5% bovine serum albumin TBS-T solution. Membranes were washed three times in TBS-T and incubated with relevant horseradish peroxidase–conjugated secondary antibodies (Vector Laboratories, Burlingame, CA, USA). After three additional washes, membranes were developed with enhanced chemiluminescence reagents (Pierce, Bonn, Germany) and imaged by the Fusion FX7 acquisition system (Vilber Lourmat, Torcy, France). 

Cell Viability

Total amount of adenosine triphosphate (ATP) of metabolically active cells was assessed by CellTiter-Glo (Promega, Madison, WI, USA) following manufacturer's instructions. Luminescence was measured using a Tecan M1000 plate reader. 

CPCG Contraction

Rat tail collagen I solution (Corning, Inc., Corning, NY, USA) was used at a sufficient final concentration of 3 mg/mL to avoid spontaneous TM cell–populated cell gel (CPCG) contraction. Cold stock collagen was mixed with DPBS (1× final concentration), 0.1 N NaOH (2.5% final concentration), and a solution containing TM cells. Trabecular meshwork cell concentration was adjusted to 200,000 cells/mL in DMEM and quickly added to the collagen/DBPS/NaOH mixture kept on ice. After gentle mixing, 500 μL of the solution was poured into each well of a 24-well plate. The plate was incubated in humidified 5% CO₂ and 95% air atmosphere at 37°C for 15 minutes to allow gelling. Then 1 mL DMEM was added on the solidified CPCG, which were freed from the well borders using thin tweezers. Twenty-four hours after seeding in collagen gels, the cells were serum starved for another 24 hours and then treated in the same way as the cells grown on poly-L-lysine–coated surfaces. 

Cell-populated collagen gels were imaged using a Leica Z6 APO macroscope (Wetzlar, Germany). Cell-populated collagen gel pictures were calibrated using the well diameter, and CPCG surfaces were calculated by ImageJ software and expressed as a percentage of the initial surface measured prior to treatments. 

Quantitative Real-Time PCR

Total RNA was automatically isolated by a QIACube system using a RNeasy kit with a DNase digestion step (all from Qiagen, Valencia, CA, USA). Complementary DNA was synthesized with a reverse transcription kit using random primers for amplification (Applied Biosystems, Foster City, CA, USA) in a T100 thermal cycler (Bio-Rad Laboratories, Hercules, CA, USA). For real-time quantitative PCR, cDNA, TaqMan Universal PCR Master Mix, and TaqMan gene expression assays (PTGER2: Hs04183523_m1, PTGFR: Hs00168763-m1, COL1A1: Hs00164004_m1, GAPDH: Hs99999905_m1) were mixed in a 96-well optical reaction plate (all from Applied Biosystems). Cycling conditions included an initial denaturation step at 95°C for 10 minutes, followed by 40 cycles consisting of a 15-second denaturation at 95°C and a 1-minute annealing and primer extension at 60°C. GAPDH was used as housekeeping gene, and amplification of its cDNA was done in parallel with the gene of interest. Threshold cycle (Ct) values were determined using the Applied Biosystems software. 

Statistical Analysis

All values were expressed as mean ± SEM of at least three independent experiments. Data were analyzed using GraphPad Prism 6 (GraphPad Software Inc., San Diego, CA, USA). Significance was tested by a 2-way ANOVA with Sidak's or Tukey's correction for multiple comparisons to compare the effects of TGF-β2 and PG analogues, respectively. Significance was denoted as follows: P < 0.05 (*), P < 0.01 (**), or P < 0.001 (***). 

Results

Glucocorticoid-dependent secretion of myocilin being the most specific phenotypic TM cell marker described,^41–43 the three different batches of commercially available human primary TM cells were challenged with dexamethasone for 96 hours. The batch from an adult donor (no. 4973) showed an increased secretion of myocilin in TM supernatants in response to dexamethasone, while myocilin was not detected in the supernatants of the two other strains from fetal origin (nos. 5987 and 7278) (Fig. 1a). The absence of myocilin secretion was not due to a potential toxic effect of dexamethasone, as the cell viability seemed even slightly increased with the glucocorticoid treatment (Fig. 1b). Before comparing the effects of FP and EP2 receptor agonists, we evaluated the expression level of both receptors. The TM cells from fetal origin showed a significant decrease in mRNA expression of the EP2 receptor (batch no. 5987) and of the FP receptor (batch nos. 5987 and 7278) in comparison with the adult human TM cells (batch no. 4973) (Fig. 1c). As fetal TM cells did not express the most specific marker of TM cells and, importantly, did not allow comparison of the effects of EP2 and FP receptor agonists, only the adult TM cells were used for subsequent experiments and are referred as hTM cells. Furthermore, phase-contrast microscopy revealed an elongated fibroblast-like phenotype of the hTM cells in culture (Fig. 1d) as previously observed in the literature for adult hTM cells.^44,45 Finally, hTM cells exhibited significant increases in α-SMA, vinculin, fibronectin, and collagen type I levels, as well as increased actin stress fibers in response to TGF-β2 (Fig. 1e) in accordance with the TGF-β2–dependent myofibroblastic transition of TM cells.^46,47 

Figure 1

 

Characterization of the human trabecular meshwork cells. Three batches of TM cells from ScienCell were grown on poly-L-lysine–coated surfaces and serum starved for 24 hours. (a) Cells were incubated in the presence of 100 nM dexamethasone (dex) for 96 hours. Cell supernatants were subjected to Western blot to detect myocilin secretion. (b) Cell viability was measured by CellTiter-Glo 96 hours after treatment with dexamethasone (100 nM) and expressed as a percentage of the control DMSO condition (mean ± SEM, n = 3). **P < 0.01, ***P < 0.001 (2-way ANOVA with Sidak's multiple comparison test). (c) mRNA expression of the EP2 and the FP receptors was analyzed after the 24-hour serum starvation period. mRNA level was expressed as fold change over the control condition (mean ± SEM, n = 3). *P < 0.01, **P < 0.01, ***P < 0.001 (1-way ANOVA with Tukey's multiple comparison test). (d) The adult hTM cells (batch no. 4973) were imaged using a contrast-phase microscope. (e) Representative confocal pictures of actin stress fibers (red), α-SMA, vinculin, fibronectin, and collagen type I (green), counterstained with TO-PRO-3 (gray), were obtained after treatment of adult hTM cells (no. 4973) with TGF-β2 for 96 hours.

In order to assess the acquisition of a contractile phenotype by hTM cells and the effects of PG analogues, the cells were embedded in a collagen type I solution. After gelling, the generated three-dimensional (3D) model of hTM cells, called hTM CPCG, was used as a force-sensing device as previously described.⁴⁸ Transforming growth factor-β2 induced a time- and concentration-dependent decrease of the hTM CPCG surface (Fig. 2a), suggesting a contraction of hTM cells. Transforming growth factor-β2 was significantly effective in inducing hTM CPCG contraction from 0.02 ng/mL and reduced CPCG surface to approximately 20% of its initial size after 96 hours at the 2 ng/mL concentration. Furthermore, the EP2 agonist butaprost and the FP agonist latanoprost were incubated either alone or in the presence of 0.2 ng/mL TGF-β2, a concentration producing approximately 50% contraction at 96 hours. Latanoprost significantly induced hTM CPCG contraction from 48 hours and reduced CPCG surface by 22 ± 3% (mean ± SEM) at 96 hours (Fig. 2b). Moreover, latanoprost significantly potentiated TGF-β2–mediated hTM CPCG contraction at 96 hours. Butaprost, on the other hand, did not modify hTM CPCG by itself but significantly inhibited TGF-β2–mediated contraction (Fig. 2c). The relaxing effect of butaprost was observed throughout the 96 hours and was most pronounced at 24 hours as observed by a 20 ± 4% (mean ± SEM) increase of CPCG surface in comparison to TGF-β2 alone. As the EP2 receptor is mainly coupled to Gα_s protein,³¹ we used forskolin, an adenylate cyclase activator, which increases cAMP levels. Forskolin (10 μM) completely inhibited TGF-β2–dependent contraction over the 96-hour experiment (data not shown), suggesting that the EP2 receptor–dependent inhibition of contraction is mediated by cAMP. 

Figure 2

 

Latanoprost induces contraction of hTM cells while butaprost relaxes TGF-β2–mediated contraction. hTM cells were embedded into a 3D collagen matrix to evaluate CPCG contraction. hTM CPCG were imaged before addition of compounds to define the initial area and then at 24, 48, and 96 hours using a macroscope. hTM CPCG surfaces were expressed as a percentage of the initial area. (a) hTM CPCG were treated with various concentrations of TGF-β2 (left, mean ± SEM, n = 3). Illustrations of the hTM CPCG contraction were obtained at 96 hours (right). (b) hTM CPCG were treated with either 0.1% DMSO, as a control, or 1 μM latanoprost (latano) with or without 0.2 ng/mL TGF-β2 (mean ± SEM, n = 4). (c) hTM CPCG were treated with either 0.1% DMSO, as a control, or 1 μM butaprost (buta) with or without 0.2 ng/mL TGF-β2 (mean ± SEM, n = 3). *P < 0.05, **P < 0.01, ***P < 0.001 when comparing DMSO with PG analogue for each time point with or without TGF-β2 (2-way ANOVA with Tukey's multiple comparison test).

In correlation with the results from the hTM CPCG contraction assay, latanoprost increased the formation of actin stress fibers and potentiated TGF-β2–mediated actin bundle polymerization at 96 hours (Fig. 3a). Measurement of hTM cell viability showed that latanoprost, as well as TGF-β2, increased viable hTM cells at 96 hours (Fig. 3b), and these results were correlated with hTM cell nuclei counting (data not shown). Furthermore, butaprost inhibited TGF-β2–dependent actin stress fiber formation without impacting hTM cell viability. These data were not observed with the two fetal TM strains (data not shown). We also studied the expression of the widely used myofibroblast marker, α-SMA,⁴⁹ and the phosphorylation of the regulatory light chain of myosin II (MLC) leading to an increase in actomyosin contractility.⁵⁰ Latanoprost alone slightly increased the expression of α-SMA at 24 and 96 hours although not significantly (Fig. 3c). In comparison, α-SMA expression was strongly increased at 2 ng/mL TGF-β2. Furthermore, MLC phosphorylation was enhanced by latanoprost and TGF-β2 but following different kinetics. Latanoprost induced a weak but significant phosphorylation at 24 hours, persisting at 96 hours, while TGF-β2–induced phosphorylation decreased by 96 hours. In contrast, butaprost inhibited MLC phosphorylation at 24 hours (Fig. 3c) in correlation with the significant relaxing effect of butaprost on TGF-β2–mediated hTM CPCG contraction (Fig. 2c). 

Figure 3

 

Latanoprost, but not butaprost, increases myofibroblast markers in hTM cells. hTM cells were grown on a poly-L-lysine–coated surface to assess actin fiber formation, α-SMA expression, and phosphorylation of MLC. (a) hTM cells were treated with either 0.1% DMSO, as a control, or 1 μM butaprost or latanoprost with or without 2 ng/mL TGF-β2. Representative confocal pictures of actin stress fibers (red) counterstained with DAPI (blue) were obtained at 96 hours (left). Total fluorescence was measured at 570 nm (right, mean ± SEM, n = 3). (b) hTM cells were treated with either 0.1% DMSO, as a control, or 1 μM butaprost or latanoprost with or without 2 ng/mL TGF-β2. Cell viability was measured at 96 hours by CellTiter-Glo and expressed as a percentage of the control DMSO condition (mean ± SEM, n = 3). *P < 0.05, **P < 0.01, ***P < 0.001 when comparing DMSO, butaprost, and latanoprost with or without TGF-β2 (2-way ANOVA with either Tukey's or Sidak's multiple comparison tests for the effects of PG analogues or TGF-β2, respectively). (c) hTM cells were treated with either 0.1% DMSO, as a control, or 1 μM butaprost (buta) or latanoprost (latano) with or without TGF-β2 at the specified concentrations. Cell lysates obtained at the indicated times were analyzed by Western blot. Values were obtained from the densitometric analyses of three independent experiments and expressed as fold change over the control condition (mean ± SEM). *P < 0.05, **P < 0.01, ***P < 0.001 when comparing the tested conditions to the DMSO control (2-way ANOVA for each TGF-β2 concentration with Sidak's multiple comparison test).

The contracting effect of latanoprost was further studied. The latanoprost-dependent hTM CPCG contraction was similar at 1 μM and 100 nM (Fig. 4a) and was blocked by the FP receptor antagonist (Fig. 4b), AL-8810 (30 μM),⁵¹ without affecting hTM cell viability at 96 hours (data not shown). These results demonstrated that the contracting effect mediated by latanoprost was specific to the FP receptor. To gain insight into the mechanism of latanoprost-induced contraction, several compounds were used to assess the contribution of signaling pathways. Firstly, Y-27632 (10 μM), an inhibitor of the Rho-associated kinase (ROCK), an upstream effector of MLC phosphorylation, suppressed the latanoprost-dependent hTM CPCG contraction (Fig. 4c) without inducing cytotoxic effect at 96 hours (data not shown), confirming the involvement of MLC phosphorylation. Moreover, as extracellular signal-regulated kinase (ERK) and p38 signaling pathways are known to play a role in fibroblast contraction,⁵² their implication in the latanoprost-mediated contraction was tested by using FR180204 (10 μM), an ERK inhibitor, and SB203580 (10 μM), a p38 inhibitor (Figs. 4d, 4e). No cytotoxic effect was observed at 96 hours (data not shown), but the p38 inhibitor abolished latanoprost-mediated contraction while the ERK inhibitor induced an additive contraction, suggesting that the FP-dependent contraction involved p38 but not ERK activation. 

Figure 4

 

Latanoprost contraction is dependent on FP receptor and is mediated by p38 but not by ERK activation. hTM cells were embedded into a 3D collagen matrix to evaluate CPCG contraction induced by latanoprost. hTM CPCG were imaged before addition of compounds to define the initial area and then at 24, 48, and 96 hours using a macroscope. hTM CPCG surfaces were expressed as the percentage of the initial area. (a) hTM CPCG were treated with 0.1% DMSO, as a control, or various concentrations of latanoprost (mean ± SEM, n = 3). hTM CPCG were preincubated for 30 minutes with either (b) the FP antagonist AL-8810 (30 μM), (c) the ROCK inhibitor Y-27632 (10 μM), (d) the ERK inhibitor FR180204 (10 μM), or (e) the p38 inhibitor SB203580 (10 μM). Control (0.1% DSMO) or 1 μM latanoprost (latano) was then added to the cell medium (mean ± SEM, n = 3–4). *P < 0.05, **P < 0.01, ***P < 0.001 when comparing DMSO with latanoprost for each time point (2-way ANOVA with Tukey's multiple comparison test).

A rise in collagen synthesis is another characteristic of myofibroblast transition.¹⁶ Consequently, we evaluated by immunocytochemistry the end product of the complex collagen production process, the deposition of the insoluble collagen matrix. After 96 hours of culture, hTM cells did not produce significant amounts of extracellular mature collagen (Fig. 5a). On the other hand, TGF-β2 induced a significant increase in the pericellular collagen deposited in a reticular pattern. Interestingly, both PG analogues decreased the extracellular accumulation of collagen induced by TGF-β2 in hTM. However, such modulation of TGF-β2–dependent collagen deposition by PG analogues was not observed in fetal TM cells (data not shown). Furthermore, the mRNA expression level of the pro-α1 chain encoded by the COL1A1 gene, a major component of collagen type I, increased 6 hours after TGF-β2 treatment, culminating at 24 hours and persistent to 48 hours (Fig. 5b). Butaprost significantly decreased TGF-β2–induced COL1A1 upregulation at 6 and 24 hours while latanoprost did not, suggesting that only butaprost inhibited the transcriptional activity of the TGF-β2 signaling pathway. 

Figure 5

 

Latanoprost and butaprost inhibit TGF-β2–mediated collagen deposition in hTM cells but differently affect TGF-β2–dependent collagen I transcription. hTM cells were grown on a poly-L-lysine–coated surface to assess collagen type I production. For that purpose, hTM cells were treated with either 0.1% DMSO, as a control, or 1 μM butaprost or latanoprost with or without 2 ng/mL TGF-β2. (a) Representative confocal pictures (left) were obtained at 96 hours by staining collagen deposition (green) and counterstaining with DAPI (blue). Total fluorescence was measured at 520 nm (right, mean ± SEM, n = 12). (b) mRNA expression of collagen type I gene COL1A1 was analyzed at the indicated times. mRNA level was expressed as fold change over the control condition (mean ± SEM, n = 3). **P < 0.01, ***P < 0.001 when comparing DMSO, butaprost, and latanoprost with or without TGF-β2 (2-way ANOVA with either Tukey's or Sidak's multiple comparison tests for the effects of PG analogues or TGF-β2, respectively).

The remodeling effect of PG analogues on the uveoscleral outflow pathway has been extensively attributed to MMP activation.^24–26 Consequently, the broad-spectrum inhibitor of MMP, marimastat (10 μM), was evaluated (Fig. 6a). Marimastat did not modify collagen deposition in the control conditions (data not shown) but significantly increased collagen deposition induced by TGF-β2, suggesting that MMP did regulate collagen turnover in TM cell cultures. However, MMP activation did not seem to be involved in the inhibitory effect of PG analogues on TGF-β2–induced collagen deposition, as marimastat did not block their effect. In contrast, latanoprost-dependent inhibition of collagen deposition was not modified by the ERK inhibitor FR180204 (10 μM) but abolished by the p38 inhibitor SB203580 (10 μM) (Figs. 6b, 6c), while these inhibitors did not affect collagen deposition in the control conditions (data not shown). Neither of those inhibitors modulated butaprost effects. 

Figure 6

 

PG analogue–mediated inhibition of TGF-β2–dependent collagen deposition is independent of MMP activation, but the latanoprost effect depends on p38 activation. hTM cells were grown on a poly-L-lysine–coated surface and preincubated for 30 minutes with (a) the broad-spectrum MMP inhibitor marimastat (10 μM), (b) the ERK inhibitor FR180204 (10 μM), or (c) the p38 inhibitor SB203580 (10 μM). hTM cells were then treated with either 0.1% DMSO, as a control, or 1 μM butaprost or latanoprost with or without 2 ng/mL TGF-β2 for 96 hours. Total fluorescence from collagen deposition stained with an anti-human collagen type I antibody and Alexa Fluor 488–conjugated secondary antibody was measured at 520 nm (mean ± SEM, n = 4–5). *P < 0.05, **P < 0.01, ***P < 0.001 when comparing DMSO, butaprost, and latanoprost with or without TGF-β2 (2-way ANOVA with Tukey's multiple comparison test).

Discussion

The myofibroblastic transition of TM cells, defined by the acquisition of a contractile phenotype and an increased collagen synthesis, could be responsible for the TM dysfunction observed in glaucomatous patients. If PG analogues mediate potent hypotensive effects and undoubtedly increase the uveoscleral outflow pathway, their effects on the pathophysiology of the TM remain largely unknown. We demonstrate, in one validated adult hTM cell model, that latanoprost, the most commonly used glaucoma medication, induces cell contraction and decreases TGF-β2–mediated collagen deposition. Alternatively, the EP2 agonist butaprost inhibits both TGF-β2–dependent contraction and collagen deposition and, consequently, the myofibroblast transition of hTM cells. 

To avoid the inherent bias of immortalized cell lines, primary hTM cells were evaluated based on the dexamethasone-dependent secretion of myocilin. One source of adult TM cells was validated while two batches of fetal TM cells failed to show an increased secretion of myocilin in response to dexamethasone. Furthermore, PG receptor transcripts were significantly more strongly expressed in the adult TM strain. While other markers of differentiation in TM cell lineage would be necessary to validate these findings, as well as extension to various primary hTM cell strains, the observed differences may be due to the early stage of development of the fetal hTM cells. 

The effect of PG analogues on adult hTM cell contraction was thus evaluated by embedding the cells in a 3D collagen matrix, which allows an indirect measurement of the cellular traction forces applied to the ECM.⁴⁸ The assay parameters were defined to avoid hTM CPCG contraction in control conditions. For that purpose, the adult hTM cells were serum starved for 24 hours and the collagen gel was used at a sufficiently high concentration. In these settings, latanoprost, just like the profibrotic cytokine TGF-β2, induced a time- and concentration-dependent contraction of the hTM CPCG. Furthermore, the latanoprost-mediated contraction correlated with increased actin stress fiber formation, MLC phosphorylation, and α-SMA expression, known hallmarks of myofibroblast transition. In contrast, if butaprost had no effect per se, it significantly inhibited TGF-β2–dependent contraction, which was correlated with MLC phosphorylation inhibition and a decrease in actin stress fiber formation. Interestingly, TM from cynomolgus monkeys treated for a year with latanoprost exhibited myofibroblastic-like cells in the cribriform region,²³ suggesting that our in vitro observations have physiological relevancy in vivo. Other studies using CPCG as a force-sensing device showed that latanoprost induced contraction of human Tenon fibroblasts (HTF) from the conjunctiva and correlated with the formation of actin stress fibers, consistent with our observations.⁵³ In contrast, endothelin-dependent short-time contraction of bovine TM strips has been reported to be inhibited by FP receptor agonists, suggesting that the FP receptor activation could potentially lead to opposite effects at shorter time points.⁵⁴ Finally, EP2 receptor activation decreased CPCG contraction by dermal fibroblasts.^55,56 These authors also demonstrated that the butaprost relaxing effect was mediated by activation of adenylate cyclase and generation of cAMP as we also observed in our hTM CPCG model by using forskolin (data not shown). 

To further explore the mechanism of action of latanoprost-mediated CPCG contraction, we investigated the signaling pathways involved on only the adult hTM cell batch. Firstly, we showed that latanoprost-mediated contraction was dependent on ROCK activation. Because ROCK is an upstream activator of MLC phosphorylation, this result is in accordance with our finding of increased phosphorylation of MLC induced by latanoprost treatment. Interestingly, ROCK inhibitors have been shown to increase the outflow pathway and to reduce IOP,^57–59 suggesting that latanoprost-mediated contraction through ROCK signaling could decrease the outflow through the TM. However, these results should be extended to other TM cell strains and confirmed by in vivo experiments. Furthermore, because p38 and ERK signaling have been involved in contractile and actomyosin relaxation processes,^46,52 their contribution to latanoprost contraction was tested. A p38 inhibitor suppressed latanoprost-induced contraction while an ERK inhibitor resulted in hTM CPCG contraction by itself. Interestingly, these results are in accordance with a previous report on HTF cells showing that latanoprost-induced HTF CPCG contraction was dependent on p38 activation but also seemed mediated by ERK and c-Jun N-terminal kinase signaling.⁵³ In another study, p38 inhibition blocked TGF-β–mediated HTF CPCG contraction and myofibroblast transition while ERK signaling inhibition, by U0126, induced a spontaneous contraction of HTF CPCG.⁵² Hence, we extended these observations made on HTF to TM cells and showed that latanoprost induced the generation of contractile forces in TM cells, a characteristic of myofibroblast phenotype, through ROCK and p38 activation. 

Collagen production is another feature of myofibroblast transition. Both latanoprost and butaprost inhibited TGF-β2–mediated collagen deposition. Interestingly, studies on cynomolgus monkeys treated for a year with either latanoprost²³ or butaprost¹⁸ reported a loss of collagen materials within the TM in accordance with our present in vitro results. Furthermore, the FP agonist fluprostenol has been shown to decrease expression of collagen types IV and VI induced by connective tissue growth factor (CTGF) in cultured TM cells.⁶⁰ To gain insight into the mechanism of PG analogue–mediated inhibition of collagen deposition, several hypotheses were tested on the only adult source of hTM cells used. Firstly, at the transcriptional level, the activation of EP2 receptor led to the attenuation of mRNA expression of the major component of collagen type I, the pro-α1 chain, while latanoprost did not. These results are in agreement with the suppressed mRNA level of α1-collagen induced by an EP2 agonist in lung fibroblasts.⁶¹ Moreover, we showed that p38 activation was involved in latanoprost-mediated inhibition of collagen deposition, but not ERK, while butaprost effects were not affected by inhibition of either pathway. These results confirm the pivotal role of p38 pathway in latanoprost effects and provide further evidence of the activation of distinct signaling pathways in the regulation of collagen deposition by FP and EP2 agonists. Finally, as the mechanism of action of FP agonists is known to induce MMP activation in the ciliary body and the sclera,^24–26 we tested the involvement of MMP activation in our model. The broad-spectrum MMP inhibitor, marimastat, did not block the effects of either PG analogue, suggesting that the inhibition of collagen accumulation is independent of MMP activation in hTM cells. These results are in accordance with the absence of MMP modulation by latanoprost in the TM, as opposed to the profound upregulation of MMP in the ciliary body.²⁴ Thus, mechanisms other than MMP regulation can be involved in PG analogue–mediated inhibition of ECM accumulation to potentially lead to increased AH drainage through the TM. 

Myofibroblasts are able to develop strong contractile forces and synthesize ECM components crucial for normal wound healing processes and for endowing resistance to tissues subjected to high pressures. However, the increased myofibroblast cell population in the TM, due to fibrotic agents such as TGF-β2 or other pathologic events, could lead to the known alterations observed in the TM of POAG patients. Here, we provide evidence, in our model of adult hTM cells, that the PG analogue latanoprost, used to treat OHT, promotes a contractile phenotype and decreases collagen deposition induced by TGF-β2 through activation of p38. The two mechanisms differently regulate the AH outflow pathway, as a decrease of ECM accumulation would reduce the outflow resistance whereas TM contraction would increase it. Nonetheless, the potent hypotensive effect of FP agonists should overwhelm the potential adverse effect of TM cell contraction. In contrast, the ability of the EP2 receptor to both relax TM cell contractility and decrease collagen deposition may be potentially beneficial for long-term treatment of glaucoma, as both mechanisms decrease AH outflow resistance. Such novel data warrant extension to various adult primary hTM cell strains and validation in in vivo models. 

Acknowledgments

The authors thank Charlotte Vernhes (Sanofi Pasteur, Florida, USA) for critical reading and editing of the manuscript. They thank Stéphane Fouquet and David Godefroy (Imaging Core Facility, Vision Institute, France) and Sup'Biotech school (Villejuif, France) for their collaboration. They also thank Patrick Avenet (Sanofi, France), who made this work possible. 

Supported by Sanofi Research & Development and by the Vision Institute (Paris). 

Disclosure: G. Kalouche, Sanofi (E, F); F. Beguier, None; M. Bakria, None; S. Melik-Parsadaniantz, None; C. Leriche, Sanofi (E); T. Debeir, Sanofi (E); W. Rostène, None; C. Baudouin, None; X. Vigé, Sanofi (E)