GSK3β Inhibitors Inhibit TGFβ Signaling in the Human Trabecular Meshwork

Abstract

Purpose: Primary open-angle glaucoma (POAG) is a leading cause of blindness, and its primary risk factor is elevated intraocular pressure (IOP) due to pathologic changes in the trabecular meshwork (TM). We previously showed that there is a cross-inhibition between TGFβ and Wnt signaling pathways in the TM. In this study, we determined if activation of the Wnt signaling pathway using small-molecule Wnt activators can inhibit TGFβ2-induced TM changes and ocular hypertension (OHT).

Methods: Primary human TM (pHTM) cells and transduced SBE-GTM3 cells were treated with or without Wnt and/or TGFβ signaling activators and used for luciferase assays; for the extraction of whole-cell lysate, conditioned medium, cytosolic proteins, and nuclear proteins for Western immunoblotting (WB); or for immunofluorescent staining. Human donor eyes were perfusion cultured to study the effect of Wnt activators on IOP.

Results: We found that the small-molecule Wnt activators (GSK3β inhibitors) (BIO, SB216763, and CHIR99021) activated canonical Wnt signaling in pHTM cells without toxicity at tested concentrations. This activation inhibited TGFβ signaling as well as TGFβ2-induced extracellular matrix deposition and formation of cross-linked actin networks in pHTM cells or SBE-GTM3 cells. We also observed nuclear translocation of both Smad4 and β-catenin in pHTM cells, which suggested that the cross-inhibition between the TGFβ and Wnt signaling pathways may occur in the nucleus. Using our ex vivo model, we found that CHIR99021 inhibited TGFβ2-induced OHT in perfusion-cultured human eyes.

Conclusions: Our results showed that small-molecule Wnt activators have the potential for treating TGFβ signaling-induced OHT in patients with POAG.

Glaucoma is a leading cause of irreversible blindness in the United States and the world.¹ Patients with glaucoma have gradual loss of retinal ganglion cells and their axons, which leads to concentric and painless loss of visual function. The most prevalent type of glaucoma is primary open-angle glaucoma (POAG). The most important risk factor for the development and progression of POAG and other glaucoma types such as pigmentary glaucoma, glucocorticoid-induced glaucoma, and pseudoexfoliation glaucoma is elevated intraocular pressure (IOP).² Even for normal-tension glaucoma, lowering IOP has been clinically proven to protect the optic nerve and visual field.³ Currently, lowering IOP is the only clinically available treatment for glaucoma. Intraocular pressure is determined by aqueous humor production, aqueous humor outflow, and episcleral vein pressure. Many studies showed that in patients with POAG, elevated IOP is due to decreased outflow facility at the trabecular meshwork (TM), especially the juxtacanalicular region and the inner wall of Schlemm's canal. Glaucomatous changes in the TM include loss of TM cells^4,5; compromised TM cell functions^6,7; excessive formation of cross-linked actin networks (CLANs)^8,9; abnormal expression, deposition, and crosslinking of extracellular matrix (ECM) proteins^10–13; and high tissue stiffness.^14–16 

Many factors are associated with POAG. Among those factors, TGFβ2 is one of the most well recognized. Studies have shown increased TGFβ2 levels in the aqueous humor and TM of patients with glaucoma.^17–19 TGFβ2 activates the TGFβ signaling through both the SMAD (canonical) and non-SMAD (noncanonical) signaling pathways. In the SMAD signaling pathway, TGFβ2 binds to TGFβ receptor II, phosphorylates TGFβ receptor I, and then phosphorylates SMAD2/3. Phosphorylated SMAD2/3 associates with the co-SMAD protein SMAD4, translocates into the nucleus, binds to the SMAD binding elements (SBEs), and alters gene expression. Excessive TGFβ2 induces pathologic changes in the TM^20–23 as well as elevates IOP in perfusion cultured human^20,24 and in vivo rodent eyes.²⁵ 

Our group previously showed that there is a functional canonical Wnt signaling pathway in the TM.²⁶ Studies show that canonical Wnt signaling plays an important role in the regulation of IOP in the mouse eye.^26–30 In the absence of Wnt ligands, a complex formed by GSK3β, Axin2, APC, and CKI phosphorylates β-catenin and induces β-catenin degradation. When Wnt ligands bind to their transmembrane receptor Frizzled and coreceptor lipoprotein receptor-related protein 5/6 (LRP 5/6), the complex disassembles, which allows cytosolic β-catenin to accumulate and translocate into the nucleus. Nuclear β-catenin associates with T-cell factors 1/3/4 (TCF1/3/4) or lymphoid enhancer binding factor 1 (LEF-1) and binds to the TCF/LEF binding element to regulate gene expression. We and other groups previously reported increased levels of sFRP1 and DKK1 (both are Wnt pathway inhibitors) in the glaucomatous TM, and the overexpression of either Wnt inhibitors induces ocular hypertension in human and/or mouse eyes.^26,31–33 

TGFβ and Wnt signaling pathways are both known to play key roles in cell differentiation, proliferation, and cell survival. The interaction between TGFβ and Wnt signaling pathways has been found in many pathologic conditions.^34–36 In non-TM cells/tissues, the two pathways seem to work synergistically or there is one-way inhibition. In the kidney, TGFβ and canonical Wnt signaling pathways work together to promote renal fibrosis.³⁷ However, in diabetic renal nephropathy, canonical Wnt signaling inhibits TGFβ signaling-induced renal injury.³⁴ Impaired cross-talk facilitates colorectal and lung cancer development.^35,36 In bone marrow–derived mesenchymal stem cells, the formation of Smad3 and the β-catenin complex promotes nuclear translocation.³⁸ In chondrocytes, β-catenin is protected by complexing with Smad3.³⁹ In pancreatic carcinoma cells, loss of Smad4 facilitates the degradation of β-catenin, suggesting Smad4 and β-catenin promote tumorigenesis in a synergistic manner.⁴⁰ In contrast, we reported a unique cross-inhibition between TGFβ2 and Wnt signaling pathways in the TM.⁴¹ 

Since the Wnt signaling pathway has the potential to inhibit the excessively activated TGFβ signaling pathway, in this study, we determined if small-molecule Wnt activators can be used to inhibit TGFβ2-induced pathologic TM changes and ocular hypertension. The Wnt activators used in this study were GSK3β inhibitors.⁴² Since they target the β-catenin degradation complex, the effect is relatively specific for the canonical Wnt signaling pathway. Our published studies have shown that they are effective in human TM (HTM) cells.³¹ 

Methods

Human TM Cell Cultures

Transformed human glaucomatous TM (GTM3) cells were a kind gift from Alcon Research Ltd. (Fort Worth, TX, USA). Primary human trabecular meshwork (pHTM) cells were established using corneal tissues procured from eye banks and were characterized based on dexamethasone-induced myocilin expression according to published consensus.⁴³ All HTM cells were cultured in Opti-Modified Eagle's Minimum Essential Medium (Opti-MEM; Thermofisher Scientific, Waltham, MA, USA) containing 10% fetal bovine serum (FBS; Atlas Biologicals, Fort Collins, CO, USA) as well as 1% penicillin and streptomycin and 2 mM glutamine (Thermofisher Scientific) in an incubator with 5% CO₂ at 37°C (Thermofisher Scientific). 

SBE-GTM3 Cell Line Establishment

Cignal lenti Smad luciferase reporter and Cignal lenti CMV-Renilla control virus were purchased from Qiagen (Germantown, MD, USA). GTM3 cells (a kind gift from Alcon Research) were transduced first with the Cignal lenti SBE firefly luciferase reporter lentivirus (Qiagen; used to study Smad-dependent TGFβ signaling), and the cells were selected using puromycin (1–16 µg/mL). The transduced GTM3 cells resistant to puromycin were transduced a second time with the Cignal lenti CMV-Renilla luciferase lentivirus (Qiagen; used as an internal control for normalization) and selected using hygromycin (0.1–0.4 mg/mL; Thermofisher Scientific). This cell line was named SBE-GTM3 cells. 

GSK3β Inhibitors

GSK3β inhibitors, 6-bromoindirubin-3-oxime (BIO) (cat. B1686), SB216763 (cat. S3442), and CHIR99021 (cat. SML1046), were purchased from Sigma Aldrich (St. Louis, MO, USA) and dissolved in DMSO. Stock concentrations were 14 mM BIO, 20 mM SB216763, and 4 mM CHIR99021. CHIR99021 purchased from Toris (cat. 4953; Minneapolis, MN, USA) was used for perfusion culture experiments. 

Viability Assay

Two pHTM cells strains were used for viability assay. About 2 × 10⁴ cells were seeded into individual wells in the 96-well plate. The cells were cultured in Opti-MEM with FBS until they became confluent. Then, medium was changed to Opti-MEM without FBS, and the cells were treated with GSK3β inhibitors at indicated concentrations or with DMSO as a vehicle control. After 72 hours of treatment, the cells were used for a viability assay using the CyQuant XTT assay kit (Thermo Fisher Scientific, cat. X12223) following the manufacturer's instructions. Signal was detected using a plate reader at 450 nm. 

Luciferase Assays

For luciferase assays using pHTM cells, 2 × 10⁴ cells/well of pHTM cells were cotransduced with the lentiviral SBE-firefly reporter vector (cat. 336851, Qiagen) and the lentiviral CMV-Renilla luciferase reporter vector (for normalization) (1:50 multiplication of infection (MOI); Qiagen) (cat. 336891, Qiagen) using the Sure Entry transfection reagent (1:4000; Qiagen, cat. 336921). The next day, culture medium was changed, and the cells were treated with or without recombinant human TGFβ2 (5 ng/mL), DMSO (0.1%), and/or GSK3β inhibitors (BIO, SB216763 or CHIR99021) in assay medium (Opti-MEM with 5% FBS) for an additional 24 or 72 hours. At the end of treatment, the Dual-Glow Luciferase Reporter Assay System (cat. E2940; Promega, Madison, WI, USA) was used to develop luciferase signals, which were measured using the Synergy H1 Hybrid Multi-Mode reader (BioTek Instruments, Winooski, VT, USA). Firefly luciferase levels were normalized to Renilla luciferase levels. All luciferase experiments were performed in replicates. 

Cytosolic and Nuclear Protein Extraction

In total, 1 × 10⁵ primary HTM cells (HTM 2019-009, HTM 71FOS, and HTM 9632) were seeded in the 12-well plate. After cells reached 100% confluency, they were treated with or without 0.1% DMSO, GSK3β inhibitors (BIO 1 µM, SB216763 10 µM, CHIR-99021 5 µM), recombinant Wnt3a (100 ng/mL, cat. 1324-WN; R&D Systems Minneapolis, MN, USA), and recombinant TGFβ2 (5 ng/mL, cat. 302-B2). After 24 hours of treatment, the cells were harvested, and cytosolic and nuclear proteins were extracted using the NE-PER Nuclear and Cytoplasmic Extraction kit (cat. 78835; Thermofisher Scientific) according to the manufacturer's instructions. Protein fractions were quantified using the DC protein assay kit (cat. 5000111; Bio-Rad, Hercules, CA, USA) and used for Western immunoblotting (WB). The primary antibodies were rabbit anti–β-catenin (1:1000; cat. 8480S; Cell Signaling Technologies, Danvers, MA, USA), mouse anti-Smad4 (1:1000, cat. SC7966X; Santacruz, Dallas, Texas, USA), rabbit anti-GAPDH (1:1000, cat. 5174S; Cell Signaling Technologies), mouse anti-Lamin A/C (1:1000, cat. 2032; Cell Signaling Technologies), and rabbit anti-Histone H3 (1:1000, cat. 9717; Cell Signaling Technologies). Protein–antibody complexes were detected as mentioned below. 

Western Immunoblotting

Human TM cells were treated as described and then harvested using the M-PER buffer (cat. 78501; Thermofisher Scientific) containing the protease and phosphatase cocktail (Thermofisher Scientific) for collecting whole-cell lysates (WCLs). Conditioned medium (CM) was also collected. Whole-cell lysates were quantified using the DC protein assay kit (cat. 5000111; Bio-Rad). Some pHTM cells were used for cytosolic and nuclear protein extraction as described previously. 

Equal amounts of protein of WCL/cytosolic/nuclear proteins or equal volumes of CM were boiled in Laemmli SDS buffer (Alfa Aesar, MA, USA) under reduced conditions, separated on SDS-PAGE gels, transferred to PVDF membranes, blocked with 5% dry milk in TBST buffer, and probed with primary antibodies and corresponding secondary antibodies conjugated with horseradish peroxidase. 

The primary antibodies used in our study included rabbit anti–β-catenin (1:1000; Cell Signaling Technologies), rabbit anti-fibronectin (FN) (1:1000; cat. AB1945; Millipore, Burlington, MA, USA), rabbit anti–collagen-I (1:1000; cat. ab215969; Abcam, Cambridge, MA, USA), rabbit anti-GAPDH (1:1000; Cell Signaling Technology). pSmad3 (1:1000, cat. 8769S; Cell Signaling Technologies), Smad2/3 (1:1000, cat. 8685S; Cell Signaling Technologies), mouse anti-Smad4 (1:1000; Santacruz), mouse anti-Lamin A/C (1:1000; Cell Signaling Technologies), and rabbit anti-Histone H3 (1:1000; Cell Signaling Technology). Protein–antibody complexes were detected using SuperSignal West Femto Maximum sensitivity Substrate (Thermofisher Scientific) and visualized using the ChemiDoc Imaging Systems (Bio-Rad). Densitometry was done using ImageJ (National Institutes of Health, Bethesda, MD, USA). 

Immunocytochemistry

Primary HTM cells were plated onto the gelatin-coated #1.5H glass-bottom 24-well plate (Mattek, Ashland, MA, USA or CellVis, Mountain View, CA, USA, respectively). Cells were incubated with fresh medium with or without TGFβ2 (5 ng/mL) or TGFβ2 + GSK3β inhibitors (BIO 1 µM, SB216763 10 µM, CHIR99021 5 µM) for 2 weeks, and the medium was replaced every 3 days throughout the course of the experiment. Two weeks after treatment, the culture medium was removed, and the cells were washed with room temperature PBS. The cells were fixed using prechilled 100% methanol for 20 minutes in a –20°C freezer and then washed three times with PBS. The cells were blocked using Superblock blocking buffer in PBS (cat. 37515; Thermofisher Scientific). After blocking, the cells were incubated with the primary antibody (mouse-Anti-FN-IST9 [cat. ab6328] or mouse anti–FN-BC1 [ab154210] at 1:250; Abcam) and rabbit anti-FN (1:250, cat. ab2413; Abcam) or rabbit anti–collagen-I (1:250; cat. ab215969; Abcam) at 4°C overnight. After three times washing with PBS, the cells were incubated for 2 hours with the secondary goat anti-mouse/rabbit–Alexa 488 or 594 antibodies (1:500; Life Technologies). The cells were washed with PBS three times and cell nuclei were stained using DAPI (1:1000; Thermofisher Scientific). Then cells were imaged using the Zeiss confocal microscope (LSM 700; Zeiss, White Plains, NY, USA). 

Cytoskeletal Reorganization and Formation of CLANs

A similar experimental strategy was followed for the treatment of the cells, as explained in the immunostaining section. Two weeks after treatment, the cells were washed with PBS and fixed using 4% formaldehyde at room temperature for 10 minutes. After being washed with PBS three times, the cells were permeabilized with 0.5% Triton-X-100 for 2 minutes followed by a PBS wash. The cells were blocked using Superblock blocking buffer in PBS (Thermofisher Scientific). After blocking, the cells were incubated with phalloidin–Alexa 488 (Life Technologies) at 4°C overnight. After PBS wash, cell nuclei were stained using DAPI (1:1000; Thermofisher Scientific). We used the definition of CLANs published by Wade et al.⁴⁴: “A combination of no less than 5 identifiable hubs and at least 3 triangulated arrangements of actin spokes were considered the minimum necessary to be a CLAN.” CLAN structures were counted using the Nikon Eclipse Ti2 Fluorescence microscope (Nikon Instruments, Inc., Melville, NY, USA). The percentage of CLAN⁺ cells to DAPI⁺ cells was used for one-way ANOVA analysis. 

Human Anterior Segment Perfusion Culture

Human donor eyes were procured from the eye bank (Saving Sight, Kansas City, MO, USA). The setup of perfusion culture was described previously.⁴⁵ Briefly, the extraocular tissue of the eye was removed. The eyes were sterilized using povidone-iodine for about 2 minutes, followed by PBS wash. The eyes were then bisected at the equator to remove a posterior portion of the eye, vitreous humor, lens, and so on without disturbing TM. The anterior segment was mounted on the perfusion culture dish, and a size-matched O-ring was used to clamp the anterior segment against the dish using four screws. Besides using whole eye globes, human donor corneas containing the rim tissue (which has the TM) were also used for perfusion according to our recently published methods with modification (three-dimensional [3D] print files were published as supplemental data in that study and can be downloaded).⁴⁶ To avoid tissue damage, we used the BioClear V1 resin and the Formlab 3B 3D printer (Formlabs, Boston, MA, USA) to print the perfusion plate. Briefly, the remaining uveal tract tissue was removed. The rim of the cornea was first glued to a 3D-printed perfusion plate using Gluture (Zoetis, Inc., Kalamazoo, MI, USA). After curing Gluture in the cell culture incubator for about 10 to 20 minutes, the edge of the sclera was sealed using the Bondic glue kit (Bondic, Aurora, ON, Canada), purchased from Amazon.com. Bondic is a resin that cures within a few seconds upon UV exposure. Cured Bondic helps form a watertight anterior chamber on the perfusion plate. The perfusion plate, after connecting to plastic tubing, was placed in a plate holder (see our published study for plate holder preparation).⁴⁶ 

All the dishes/plates were equipped with two cannulas: one was connected to a syringe mounted on a syringe pump for medium infusion while the other was connected to a pressure transducer for IOP measurement. Perfusion culture medium (OptiMEM + 1% glutamine + 1% penicillin and streptomycin +1% amphotericin B) (Thermofisher Scientific) was used to infuse at a constant rate at 2.5 µL/min using a syringe pump (PHD2000; Harvard Apparatus, Holliston, MA, USA). Pressure transducers (ADInstruments, Colorado Springs, CO, USA) were connected to a data acquisition system (PowerLab; ADInstruments) consisting of a signal amplifier, a bridge amplifier, and a computer with LabChart software (ADInstruments). After stable baseline IOP was established, the eyes were treated with the indicated treatment. For data analysis, baseline IOP was defined as the average of IOP measured 12 hours prior to treatment. The IOP after treatment started was also averaged every 12 hours. ΔIOP was defined as IOP minus baseline IOP. 

Histology and Immunohistology Studies

Perfusion cultured ocular tissues were fixed using 4% paraformaldehyde, embedded in paraffin and sectioned at 10 µm in thickness, and transferred onto SuperFrost Plus slides (Thermofisher Scientific). The slides were heated for 1 hour on a slide warmer. Deparaffinization was performed by washing twice in xylene and twice in 100% ethanol, 95% ethanol, and 50% ethanol for 3 minutes. The slides were soaked in running distilled water. 

For histology studies, the sections were stained with hematoxylin and eosin. After dehydration with 50%, 95%, and 100% ethanol, the sections were washed with xylene and persevered in Permount (Thermofisher Scientific). Images were taken using the Nikon Ti2 inverted microscope (Nikon Instruments, Inc.). 

For immunohistology studies, the sections were used for antigen retrieval using the Tris-EDTA buffer (10 mmol/L Tris base, 1 mmol/L EDTA solution, and 0.05% Tween 20, pH 9.0) and the 2100 Antigen retriever (Electron Microscopy Sciences, Hatfield, PA, USA). The sections were cooled to room temperature and incubated with Superblock TW Blocking Buffer in TBS (Thermofisher Scientific) for 60 minutes. The tissue sections were incubated with primary rabbit anti–β-catenin (1:250, cat. 8480S; Cell Signaling Technologies) and rabbit anti-pSmad3 (1:250, cat. 9520; Cell Signaling Technology) and then followed by the secondary antibody Alexa Fluor–labeled donkey anti-rabbit IgG (cat. A-21206; Invitrogen, Carlsbad, CA, USA). A no primary antibody control was also included to determine the nonspecific binding of the secondary antibody. Slides were mounted with the Prolong Gold mounting medium containing DAPI (cat. P36931; Invitrogen). Images were taken using the Zeiss LSM700 Confocal Microscope. 

Statistical Analysis

Prism v8 (GraphPad Software, San Diego, CA, USA) was used for statistical analysis. One-way ANOVA and Sidak post hoc tests were used, and P values < 0.05 were considered statistically significant. 

Results

GSK3β Inhibitors Activated the Canonical Wnt Signaling Pathway in pHTM Cells

In our previous studies, we found that GSK3β inhibitors (BIO, SB216763, and CHIR99021) activated the canonical Wnt signaling pathway with the optimum concentration at 1 µM, 10 µM, and 5 µM in transformed GTM3 cells, respectively.³¹ In this study, we first determined if these compounds are toxic to pHTM cells using a viability assay (XTT assay). Two pHTM cell strains were seeded in 96-well plates and treated with indicated compounds and concentrations for 72 hours before the XTT assay. We found that the concentrations that we used in this study did not show significant toxicity to pHTM cells (Fig. 1). Instead, SB showed a slight but significant increase in cell viability in one of the strains (Fig. 1B). 

Figure 1.

 

GSK3β inhibitors were not toxic to pHTM cells. Two pHTM cell strains (A and B, respectively) were used for cell viability assays. The cells were grown into confluency and treated with indicated compounds for 72 hours before assay. DMSO was used as a vehicle control, and the readings of the control was set at 1.0 for comparison. One-way ANOVA and Sidak post hoc tests were used for statistical analysis. n = 8. NS, not significant. **P < 0.01.

Then, to determine if these compounds can also activate the canonical Wnt signaling in pHTM cells, three different pHTM cell strains (HTM 2180, HTM 2019-009, and HTM 2019-022) were first transduced with the lentiviral canonical Wnt signaling reporter vector and then treated with or without 0.1% DMSO (vehicle control), BIO (1 µM), SB216763 (10 µM), CHIR99021 (5 µM), or recombinant Wnt3a (100 ng/mL; as a positive control) for 24 hours. Luciferase assays showed that BIO and CHIR99021, but not SB216763, significantly increased canonical Wnt signaling activity in the three pHTM cell strains (one-way ANOVA, n = 6, P < 0.05) (Figs. 2A–C). 

Figure 2.

 

GSK3β inhibitors activated Wnt signaling and inhibited TGFβ signaling in HTM cells. HTM 2180 (A), HTM 2019-009 (B), and HTM 2019-022 (C) cells (all pHTM cells) were transduced with lentiviral Wnt signaling reporter vector and treated with or without 0.1% DMSO (vehicle control), 100 ng/mL Wnt3a, or one of the GSK3β inhibitors (1 µM BIO, 10 µM SB216763, or 5 µM CHIR99021) for 24 hours before luciferase assays. Columns and error bars: means and SDs of Wnt signaling activity. The data were analyzed using one-way ANOVA and Sidak post hoc tests. n = 6. *P < 0.1, **P < 0.01, ****P < 0.0001. (D) SBE-GTM3 cells were treated with or without 0.1% DMSO, 5 ng/mL TGFβ2 ± one of the GSK3β inhibitors (1 µM BIO, 10 µM SB216763, or 5 µM CHIR99021), or GSK3β inhibitor alone before luciferases assays. (E) HTM 9632 cells were transduced with lentiviral TGFβ signaling reporter vectors and treated similarly as shown in (A) before luciferase assay. Columns and error bars: means and SDs of TGFβ signaling activity. The data were analyzed using one-way ANOVA and Sidak post hoc tests. n = 6 (A) or 8 (B). *P < 0.05, **P < 0.01, ****P < 0.0001.

GSK3β Inhibitors Inhibited the TGFβ Signaling Pathway in HTM cells

We have previously shown that Wnt3a inhibits TGFβ signaling in HTM cells.⁴¹ To determine whether GSK3β inhibitors also inhibit TGFβ2 signaling, SBE-GTM3 cells were used for luciferase assays. This transgenic HTM cell line was established by stable transduction of GTM3 cells with two lentiviral vectors and antibiotic selection. This cell line contains the SBE-firefly luciferase expression cassette (for the determination of TGFβ signaling) and the CMV-Renilla luciferase cassette (for normalization). 

SBE-GTM3 cells were treated with or without 5 ng/mL recombinant TGFβ2 and/or GSK3β inhibitors (1 µM BIO, 10 µM SB216763, or 5 µM CHIR99021) for 72 hours. Luciferase assays showed that BIO and CHIR99021 significantly inhibited TGFβ signaling with or without the presence of TGFβ2 (one-way ANOVA, n = 6, P < 0.05) (Fig. 2D). In contrast, SB216763 inhibited TGFβ2 signaling at the baseline level but not at the induced level (Fig. 2D). 

Since the SBE-GTM3 cell is a cell line, its biology may be different from that of pHTM cells. Therefore, we performed similar experiments using pHTM cells. Primary HTM cells were cotransduced with the same lentiviral vectors (firefly and Renilla luciferase vectors) used to establish SBE-GTM3 cells. Two days after transduction, the cells were treated with or without recombinant TGFβ2 and/or GSK3β inhibitors for 24 hours. We found that the three GSK3β inhibitors significantly inhibited TGFβ2 signaling in the presence of TGFβ2 (one-way ANOVA, n = 6, P < 0.05) but not at the baseline level (Fig. 2E). 

GSK3β Inhibitors Inhibited TGFβ2-Induced ECM Proteins in pHTM Cells

TGFβ2-induced pathologic changes in the TM, including excessive ECM protein accumulation, have been well documented. Since we previously showed that GSK3β inhibitors inhibited TGFβ2 signaling, we determined if they also inhibit TGFβ2-induced ECM proteins in pHTM cells. 

Three pHTM cell strains were treated with or without 5 ng/mL TGFβ2 with or without GSK3β inhibitors (1 µM BIO, 10 µM SB216763, or 5 µM CHIR99021) for 72 hours. Western immunoblotting and/or densitometry studies showed that GSK3β inhibitors inhibited TGFβ2-induced FN and collagen 1 levels in conditioned medium as well as in WCLs (Figs. 3A–C, a representative strain). 

Figure 3.

 

Western immunoblotting showed that GSK3β inhibitors inhibited TGFβ2-induced ECM proteins in pHTM cells. HTM 71FOS (A) and other pHTM cell strains (not shown) were treated with 0.1% DMSO (vehicle control), 5 ng/mL TGFβ2 ± one of the GSK3β inhibitors (1 µM BIO, 10 µM SB216763, or 5 µM CHIR99021), or GSK3β inhibitor alone for 72 hours. CM and WCLs were used for WB. GAPDH was used as a loading control for WCLs. Densitometric analysis of FN (B), collagen 1 (C), β-catenin (D), and pSmad3 (E) was conducted, and the data were analyzed using one-way ANOVA and Sidak post hoc tests. N = 3 or 4 different pHTM strains. *P < 0.05, **P < 0.01, ****P < 0.0001.

As a quality control, we also probed the level of total β-catenin and phosphorylated Smad3 (pSmad3)/total Smad3 in WCLs (an anti-Smad2/3 antibody was used for total Smad3) (Fig. 3A). We selected pSmad3 because studies showed that Smad3, not Smad2, plays a major role in the TM.⁴⁷ We found that GSK3β inhibitor treatment alone increased total β-catenin but not pSmad3 (Figs. 3A, 3D, 3E). GSK3β inhibitors also decreased TGFβ2-induced pSmad3. Surprisingly, we consistently found that TGFβ2 treatment alone as well as TGFβ2 + GSK3β treatment increased total β-catenin. 

Since FN exists in different isoforms, including EDA-FN and EDB-FN, which play important roles in TM homeostasis,^12,48,49 we determined the effect of GSK3β inhibitors on their expression and localization in multiple pHTM cell strains using immunostaining (Fig. 4, a representative strain). The cells were treated with or without TGFβ2 ± GSK3β inhibitors for 14 days and the proteins were immunostained. We found that GSK3β inhibitors inhibited TGFβ2-induced expression of EDA-FN and total FN (FN) (Fig. 4A) as well as EDB-FN and collagen I (Fig. 4B) in pHTM cells, which is consistent with our findings using WB. 

Figure 4.

 

Immunostaining showed that GSK3β inhibitors inhibited TGFβ2-induced ECM proteins in pHTM cells. HTM 9632 (A, B) cells were treated with 0.1% DMSO (vehicle control) or TGFβ2 ± one of the GSK3β inhibitors (1 µM BIO, 10 µM SB216763, or 5 µM CHIR99021) for 14 days and then immunostained with anti–EDA-FN (green, A) and total FN (red, A) or EDB-FN (green, B) and collagen 1 antibodies (red, B). Nuclei were stained using DAPI (blue). TGFβ2-treated cells showed elevated ECM protein expression while BIO-, SB-, or CHIR-treated cells showed ECM protein expression levels similar to control. Scale bars: 20 µm.

Wnt or TGFβ Signaling Activation Induced Nuclear Translocation of β-catenin and Smad4 Proteins Simultaneously

As described previously, we found that TGFβ2 induced total β-catenin in the pHTM. The accumulation of β-catenin is usually considered a marker of canonical Wnt signaling activation. However, our published and current luciferase assay data consistently showed that TGFβ signaling inhibits canonical Wnt signaling at the transcriptional level, and this inhibition requires both β-catenin and Smad4.⁴¹ 

To determine the effect of Wnt signaling and TGFβ signaling activation on both β-catenin and Smad4 nuclear localization, we treated three pHTM cell strains (HTM 2019-009, HTM 71, and HTM 9632) with 0.1% DMSO, BIO (1 µM), SB216763 (10 µM), CHIR99021 (5 µM), Wnt3a, and/or TGFβ2. The cytosolic and nuclear proteins were extracted for WB, and LaminA/C or Histone H3 was used as a nuclear protein loading control. 

We found that these treatments, either Wnt activators or TGFβ signaling activators, induced nuclear translocation of β-catenin and Smad4 in all three pHTM cell strains (Fig. 5). Our data suggested that both Wnt and TGFβ signaling activation induce nuclear translocation of β-catenin and Smad4 (see Discussion). 

Figure 5.

 

Wnt or TGFβ signaling activators induced nuclear translocation of β-catenin and Smad4 in pHTM cells. HTM 2019-009 (A), HTM 71FOS (B), and HTM 9632 (C) cells were treated with 0.1% DMSO (vehicle control), 1 µM BIO, 10 µM SB216763, 5 µM CHIR99021, 100 ng/mL Wnt3a, or 5 ng/mL TGFβ2 for 24 hours. Cytosolic and nuclear fractions were used for WB. GAPDH was used as a cytosolic protein loading control, and Lamin A/C or histone H3 was used as a nuclear protein loading control. Densitometric analysis of nuclear β-catenin (D) and Smad4 (E) was conducted, and the data were analyzed using one-way ANOVA and Sidak post hoc tests. n = 3. *P < 0.05.

GSK3β Inhibitors Decreased TGFβ2-Induced CLAN Formation in pHTM Cells

TGFβ2-induced cytoskeletal reorganization and formation of CLANs are associated with glaucomatous TM and glaucomatous insults.^8,9,21 More important, our recent studies, for the first time, show that CLANs increase TM cell stiffness (an important factor that contributes to higher aqueous humor outflow resistance and ocular hypertension [OHT]).⁵⁰ Therefore, we determined if GSK3β inhibitors affect CLAN formation. 

Four different pHTM cell strains were treated with or without TGFβ2 ± one of the GSK3β inhibitors (1 µM BIO, 10 µM SB216763, or 5 µM CHIR99021) for 14 days (Supplementary Fig. S1). We found that TGFβ2 significantly increased CLAN formation (Fig. 6). Also, in three of the four pHTM cell strains, this induction of CLANs was inhibited by all three GSK3β inhibitors (Fig. 6). In one strain, only SB216763 showed significant inhibition (Fig. 6D). However, although not statistically significant, BIO and CHIR99021 showed a consistent trend of inhibition similar to the other cell strains (Fig. 6D), suggesting variations in cell responses to GSK3β inhibitor treatment. 

Figure 6.

 

GSK3β inhibitors inhibited TGFβ2-induced CLAN formation in four different pHTM cell strains. HTM 21-1017L (A), HTM 21-0658 (B), HTM 21-0777 (C), and HTM 1646 (D) cells were cultured in 96-well glass-bottom plates and treated with 0.1% DMSO (vehicle), 5 ng/mL TGFβ2, or TGFβ2 + one of the GSK3β inhibitors (1 µM BIO, 10 µM SB216763, 5 µM CHIR99021) for 14 days. The cells were fixed and stained with phalloidin–Alexa 488 and DAPI, and CLAN⁺ and DAPI⁺ cells were counted using an epifluorescent microscope. The percentage of CLAN-positive cells to total cells (DAPI⁺ cells) was calculated. Columns and error bars: means and SDs. Data were analyzed using one-way ANOVA and Sidak post hoc tests. NS, not significant. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Activation of Canonical Wnt Signaling Decreased TGFβ2-Induced Ocular Hypertension in Perfusion Cultured Human Anterior Segments

Since our data showed that the activation of canonical Wnt signaling inhibited TGFβ2-induced pathologic changes in pHTM cells, we determined if the activation of canonical Wnt signaling inhibits TGFβ2-induced OHT. 

The human anterior segments or corneas were perfused using our published methods in constant flow mode (2.5 µL/min, which mimics the aqueous humor production rate in human eyes).^45,46 The perfusion culture dish and plate were 3D printed. After a stable IOP baseline was reached, the anterior segments/cornea were perfused with 5 ng/mL TGFβ2 to induce OHT. The eyes with more than 3 mm Hg IOP elevation were perfused with TGFβ2 and 150 µM CHIR99021 (mixed with TGFβ2 in perfusion medium) since CHIR99024 at this concentration did not show toxicity to pHTM cells (Fig. 1). We found that cotreatment with CHIR99021 lowered IOP compared to the fellow eyes that received TGFβ2 only (Figs. 7A, 7B). A similar IOP-lowering effect was also observed in other individual eyes (Figs. 7C, 7D). 

Figure 7.

 

CHIR99021 decreased TGFβ2-induced ocular hypertension in perfusion culture human eyes. (A, B) Two pairs of human donor eyes were perfusion cultured. One eye was treated with TGFβ2 (blue), and the fellow eye was treated with TGFβ2 + CHIR99021 (red). Two unpaired eyes (C, D) were initially treated with TGFβ2, followed by treatment with TGFβ2 + CHIR99021. Another two unpaired eyes (E, F) were initially treated with TGFβ2, followed by treatment with TGFβ2 + CHIR99021, a washout period (plain medium), and a second treatment with TGFβ2. ΔIOP = IOP – baseline IOP.

Although we found that CHIR99021 inhibited TGFβ2-induced OHT and CHIR99021 was safe for cultured pHTM cells (Fig. 1), it was still possible that CHIR99021 caused excessive TM cell loss under perfusion culture conditions and therefore lowered IOP. To rule out this possibility, two individual (nonpaired) human donor eyes were perfused with 5 ng/mL TGFβ2 to induce ocular hypertension, and then they were cotreated with CHIR99021 to lower IOP (Figs. 7E, 7F). After IOP lowering, the eyes were perfused with plain medium (no TGFβ2 or CHIR99021) to wash out any treatment. The eyes were then treated again with TGFβ2. The second TGFβ2 treatment re-elevated IOP, suggesting that there were viable TM cells after previous CHIR99021 treatment (Figs. 7E, 7F). 

To further validate the condition of perfusion cultured eyes, we fixed some perfusion cultured eyes for morphologic studies (Fig. 8) and immunostaining (Fig. 9). Hematoxylin and eosin staining showed that the morphology of the TM and Schlemm's canal (SC) was similar after TGFβ2 and TGFβ2 + CHIR99021 treatment (Fig. 8). After both treatments, no obvious damage was observed in the TM or SC. Also, we found that TGFβ2 + CHIR99021 induced more β-catenin in the eye, especially in the TM, compared to TGFβ2 alone (Fig. 9A). Since all the eyes used in this study were treated with TGFβ2, we did not have non-TGFβ2 treatment controls. In contrast, TGFβ2 + CHIR99021 decreased p-Smad3, a key component for TGFβ signaling, compared to TGFβ2 alone (Fig. 9B), which matched what we observed in cultured pHTM cells (Fig. 3). Besides, the staining of DAPI (blue) was similar between TGFβ2 + CHIR99021 and TGFβ2-treated eyes, which further indicated that there was no obvious cell loss caused by CHIR99021. 

Figure 8.

 

Morphology of perfusion cultured human eyes. A pair of human donor eyes were fixed for hematoxylin and eosin staining after being perfusion cultured for 34 days. One eye was treated with TGFβ2 (A) and the fellow eye was treated with TGFβ2 + CHIR99021 (B). Scale bar: 100 µm.

Figure 9.

 

TGFβ2 + CHIR99021 increased the expression of β-catenin but decreased the expression of p-Smad3 in the TM compared to TGFβ2 treatment. Two pairs of human eyes were perfusion cultured, and one eye was treated with TGFβ2 and the fellow eye was treated with TGFβ2 + CHIR99021. After perfusion, the eyes were fixed, embedded in paraffin, sectioned, and immunostained with the anti–β-catenin antibody (green, A) or anti-pSmad3 antibody (green, B) and DAPI (blue). We found that TGFβ2 + CHIR99021 induced more expression of β-catenin in the eye, but decreased p-Smad3, compared to TGFβ2 alone. Scale bar: 100 µm.

Discussion

We previously showed that Wnt and TGFβ signaling pathways cross-inhibit each other in the TM,⁴¹ which provides a potential therapeutic strategy for treating glaucomatous ocular hypertension. As described in the Introduction, in many non-TM cells, such cross-inhibition has not been reported based on our best knowledge. It could be due to the unique features of the TM. For example, confluent TM cells form CLANs, TM cells express more myocilin upon glucocorticoid treatment, there is no single unique protein marker for the TM, TM cells are not “authentic” macrophages but have the ability of phagocytosis, and primary TM cells are very difficult to be transfected with expression plasmids.⁴³ 

The exact mechanism of TGFβ2-Wnt signaling cross-inhibition is not clear. Nishita and colleagues⁵¹ showed that Smad4 and β-catenin bind to each other in Xenopus cells. Based on our observation that the activation of either pathway induced nuclear translocation of both Smad4 and β-catenin, it is likely that the two proteins form a complex in the nucleus that turns each other from a transactivation to a transrepressor (for example, this complex can block protein–DNA interaction by masking the protein–DNA binding domain, prevent the recruitment of other transcriptional factors due to conformational changes, etc.), which mediates this cross-inhibition. Also, it is possible that the formation of this complex inhibits the nuclear translocation of p-Smad3, which inhibits TGFβ signaling. However, further research is needed to elucidate the mechanism. Our hypothesis (summarized in Fig. 10), if proven to be correct, will suggest that Wnt signaling activation is a promising way to lower IOP as well as treat the pathology in the TM since this Smad–β-catenin complex will selectively work on the downstream, POAG-related genes. 

Figure 10.

 

Illustration of the hypothesized mechanism of the cross-inhibition between TGFβ2 and Wnt signaling pathways in the TM.

In this study, we found that not all human donor eyes developed TGFβ2-induced OHT. It could be due to several reasons: 

• 1. Individual's genetic background. For example, the first study reporting elevated TGFβ2 in glaucomatous aqueous humor, which was conducted by Tripathi, showed that glaucomatous patients had elevated TGFβ2 (2.7 ng/mL [total] and 0.45 ng/mL [active form]) compared to nonglaucomatous eyes (1.48 ng/mL [total] and 0.2 ng/mL [active form]).¹⁸ However, in 1 of 10 nonglaucomatous eyes, the level of TGFβ2 was 0.83 ng/mL (active form).¹⁸ Also, 2 of 15 glaucomatous eyes had TGFβ2 levels lower than the average of nonglaucomatous eyes.¹⁸ Therefore, the human eyes respond differently to TGFβ2.
• 2. According to our hypothesis, if an individual has elevated TGFβ2 and a “healthy” canonical Wnt signaling pathway, that person is unlikely to develop OHT. In contrast, if that individual has elevated TGFβ2 and Wnt inhibitors such as sFRP1 and/or Dkk1,^31–33 they are likely to develop OHT since the canonical Wnt signaling is inhibited and cannot antagonize the TGFβ2 signaling.
• 3. In mouse eyes, similar effects were also reported. We and Zode's group showed that not all mice injected with lentiviral vectors expressing an active form of TGFβ2 developed OHT.^52,53

Also, we found that small-molecule Wnt signaling activators (GSK3β inhibitors) inhibited TGFβ signaling and associated changes (ECM and CLANs) in the TM as well as IOP, suggesting the feasibility of this strategy. To apply these compounds in clinical use, there are several issues that need to be considered: 

• 1. Efficacy
• The effective concentration of the three GSK3β inhibitors tested was in the micromolar range. However, this concentration was tested in cell cultures and tissue perfusion cultures. In in vivo application, due to drug clearance, binding to other proteins, and distribution, higher concentrations may be needed. For small-molecule compounds, higher concentrations usually result in lower specificity. For example, we observed that SB216763 did not induce Wnt signaling from baseline in pHTM cells but showed an inhibition of TGFβ2-induced TGFβ signaling and TGFβ2-induced ECM in pHTM cells. This could be due to compound specificity (e.g., SB216763 may directly inhibit TGFβ signaling pathway molecules). Therefore, a compound screening may be needed to identify more potent candidates.
• 2.Interpersonal variations
• In this study, we used several pHTM cell strains and observed cell strain–dependent responses to these compounds. We believe that the differential responses could be due to receptor levels, basal signaling activity levels, epigenetics, age, gender, and so on. Therefore, developing multiple compounds and using a combination of treatment strategies are important for treating glaucoma since it is a multifactorial disease.
• 3.Toxicity
• We did not observe obvious cell death in cell cultures at tested concentrations of these compounds. However, prolonged treatment for months or years may have an impact on the ocular surface tissue. In the future, more in vivo studies are needed to determine their long-term toxicity.
• 4.Disease stage
• We believe that if the compound could be used clinically, it should be used at the early stage of the disease. At the late stage, since the majority of the TM cells are lost, rescuing the remaining cells may not be sufficient to reverse the pathologic changes in the TM.

In summary, we found that small-molecule Wnt pathway activators are a useful tool for studying Wnt and TGFβ signaling crosstalk in the TM, and they have the potential to serve as antiglaucoma agents. 

Acknowledgments

Supported by the National Institute of Health/National Eye Institute Award Numbers R01EY026962 (WM), R01EY031700 (WM), and R21EY033929 (WM); BrightFocus Foundation G2023009S (WM); Indiana University School of Medicine Showalter Scholarship (WM); the Indiana Clinical and Translational Sciences Institute funded, in part by Award Number UL1TR002529 from the National Institutes of Health, National Center for Advancing Translational Sciences, Clinical and Translational Sciences Award (WM); and a challenge grant from Research to Prevent Blindness (Department of Ophthalmology, Indiana University School of Medicine). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. 

Disclosure: C.K. Sugali, None; N.P. Rayana, None; J. Dai, None; D.H. Harvey, None; K. Dhamodaran, None; W. Mao, None