TGFβ2 Induces the Formation of Cross-Linked Actin Networks (CLANs) in Human Trabecular Meshwork Cells Through the Smad and Non-Smad Dependent Pathways

Purpose Increased intraocular pressure results from increased aqueous humor (AH) outflow resistance at the trabecular meshwork (TM) due to pathologic changes including the formation of cross-linked actin networks (CLANs). Transforming growth factor β2 (TGFβ2) is elevated in the AH and TM of primary open angle glaucoma (POAG) patients and induces POAG-associated TM changes, including CLANs. We determined the role of individual TGFβ2 signaling pathways in CLAN formation. Methods Cultured nonglaucomatous human TM (NTM) cells were treated with control or TGFβ2, with or without the inhibitors of TGFβ receptor, Smad3, c-Jun N-terminal kinases (JNK), extracellular signal regulated kinase (ERK), P38, or Rho-associated protein kinase (ROCK). NTM cells were cotreated with TGFβ2 plus inhibitors for 10 days or pretreated with TGFβ2 for 10 days followed by 1-hour inhibitor treatment. NTM cells were immunostained with phalloidin-Alexa-488 and 4′,6-diamidino-2-phenylindole (DAPI). Data were analyzed using 1-way ANOVA and Dunnett's post hoc test. Results TGFβ2 significantly induced CLAN formation (n = 6 to 12, P < 0.05), which was completely inhibited by TGFβ receptor, Smad3, and ERK inhibitors, as well as completely or partially inhibited by JNK, P38, and ROCK inhibitors, depending on cell strains. One-hour exposure to ROCK inhibitor completely resolved formed CLANs (P < 0.05), whereas TGFβ receptor, Smad3 inhibitor, and ERK inhibitors resulted in partial or complete resolution. The JNK and P38 inhibitors showed partial or no resolution. Among these inhibitors, the ROCK inhibitor was the most disruptive to the actin stress fibers, whereas ERK inhibition showed the least disruption. Conclusions TGFβ2-induced CLANs in NTM cells were prevented and resolved using various pathway inhibitors. Apart from CLAN inhibition, some of these inhibitors also had different effects on actin stress fibers.

G laucoma is a progressive optic neuropathy affecting both the anterior and posterior segments of the eye and is a leading cause of irreversible vision loss and blindness worldwide. It is estimated that glaucoma will affect more than 80 million people by the year 2020. [1][2][3] The most common form of glaucoma is primary open-angle glaucoma (POAG), which is characterized by painless, progressive, and irreversible vision loss. Although the exact disease mechanism(s) of POAG is not fully understood, elevated intraocular pressure (IOP) is the primary risk factor for the development and progression of POAG. In these patients, IOP elevation is caused by increased aqueous humor (AH) outflow resistance at the trabecular meshwork (TM). This increased resistance has been associated with loss of TM cells, excessive extracellular matrix (ECM) accumulation, and cytoskeletal reorganization. [4][5][6][7][8][9][10][11] We are particularly interested in the actin cytoskeletal reorganization in TM cells. We first reported the formation of cross-linked actin networks (CLANs) in both human trabecular meshwork (HTM) cells and tissues. 4,8,12 CLANs are threedimensional, geodesic dome-like structures formed primarily around the nucleus, although they can also be found throughout the cell. In two-dimensional microscopic views, CLANs appear to be web-like structures and composed of numerous ''hubs and spokes.'' Although cultured cell types other than TM cells form transient CLAN-like structures during the process of cell attachment and spreading, [13][14][15][16][17] only TM cells form and retain CLANs when they are confluent. 12 In cultured TM cells, CLAN formation can be induced by glucocorticoids and by transforming growth factor b2 (TGFb2). TGFb2 is elevated in the AH and TM of POAG patients and induces ocular hypertension in perfusion cultured human and porcine anterior segments, as well as in mouse eyes. 9,[18][19][20][21][22][23] In cultured TM cells, TGFb2 induces the expression of ECM proteins such as fibronectin (FN) and factors that suppress proteolytic degradation of the ECM. [24][25][26] Additionally, TGFb2 increases the expression of ECM cross-linking enzymes such as lysyl oxidase (LOX) 27 and transglutaminase-2. 28, 29 We previously identified an association between CLANs and POAG by comparing glaucomatous human TM (GTM) cells to nonglaucomatous TM (NTM) cells. 5,30 In those studies, we observed that approximately 40% of the GTM cells contain CLANs, whereas only 4% of the NTM cells have CLANs. Specific mathematical models predict that CLANs increase the stiffness of actin filaments by two orders of magnitude. 31 Other studies showed that GTM tissues are stiffer compared with NTM tissues. 32,33 Although those studies did not evaluate the direct correlation between CLAN formation and TM stiffness, it is believed that CLANs increase AH outflow resistance and IOP by increasing TM stiffness, as well as disturbing TM homeostasis. 12 TGFb2 signaling has two primary pathways for signal transduction: the Smad-dependent (Smad pathway) and -independent (non-Smad pathway) pathways. Both pathways are activated by TGFb2 binding to the TGFb receptor complex, which is then autophosphorylated. In the Smad pathway, the activated receptor complex phosphorylates Smad2 and Smad3, which then binds with Smad4 prior to translocation to the nucleus. The non-Smad pathway includes various branches of mitogen-activated protein kinase (MAPK) pathways such as extracellular signal regulated kinase (ERK), c-Jun N-terminal kinases (JNK), and P38 kinases, as well as Rho-like guanosine triphosphatase (GTPase) signaling pathways. 34 Rho-associated protein kinase (ROCK) inhibitors have the ability to reduce trabecular outflow resistance and IOP. [35][36][37][38][39][40] The Rho/ROCK signaling pathway acts as a molecular switch in the regulation of focal adhesions, cellular contraction, cellular motility, cytokinesis, and the formation of actin stress fibers. 41 It is believed that the ocular hypotensive effect of ROCK inhibitors is due to ''relaxation'' of the TM cytoskeleton. 37,[41][42][43] Inhibition of ROCK decreases actin polymerization, relaxes the TM cells, and decreases outflow resistance. However, whether ROCK inhibition affects CLAN formation is not clear.
Because of the complexity of the TGFb signaling, the delineation of which exact pathway is responsible for a particular biological response is difficult to predict. In the present study, we used NTM cells and various pathway inhibitors to determine the role of individual TGFb signaling pathways in the formation and stabilization of CLANs.

Cell Treatments
Treatments were administered in high glucose DMEM with 1% FBS. High glucose medium was used to improve cell survival, and a low concentration of FBS minimized confounding effects of endogenous growth factors that may be present in the serum. Confluent NTM cells were treated with TGFb2 (5 ng/mL) (RD Systems, Minneapolis, MN, USA) to induce CLAN formation or DMEM with or without dimethyl sulfoxide (DMSO; vehicle control for MAPK inhibitors). To inhibit the Smad and non-Smad pathways, NTM cells were cotreated with TGFb2 and TGFb receptor type I inhibitor SB431542 (5 lM; Sigma, Saint Louis, MO, USA). 27 To inhibit the Smad pathway, NTM cells were cotreated with TGFb2 and the Smad3 phosphorylation inhibitor SIS3 (10 lM; Sigma). 27 To inhibit the non-Smad pathways, NTM cells were cotreated with TGFb2 and the JNK inhibitor SP600125 (10 lM; CalBioChem, San Diego, CA, USA), 27 MEK/Erk inhibitor U0126 (25 lM; Promega, Maddison, WI, USA), 44 P38 inhibitor SB203580 (5lM; Tocris BioSci, Ellisville, MO, USA), 27 or ROCK inhibitor Y27632 (10 lM; Sigma). 35,37,44,45 All these inhibitors were used in previous TM studies and have demonstrated successful pathway inhibition at the concentrations described previously. Cells were cotreated with TGFb2 together with pathway inhibitors for 10 days for studying the prevention of CLAN formation or pretreated with TGFb2 for 10 days followed by treatment with individual inhibitors for 1 hour to study CLAN resolution. Each treatment group consisted of 6 to 12 coverslips (n ¼ 6 to 12). Medium was changed every 2 to 3 days.

Evaluation of CLANs
CLANs were visualized using the Nikon Eclipse Ti inverted fluorescence microscope (Nikon, Inc., Melville, NY, USA) with 6003 magnification. Cytoskeletal images were taken using the Nikon Eclipse Ti inverted fluorescence microscope equipped with the Cri Nuance FX Camera System (Perkin-Elmer, Inc., Waltham, MA, USA).
CLANs were defined as F-actin-containing cytoskeletal structures with at least one triangulated actin arrangement consisting of actin spokes and at least three identifiable hubs. 46 Representative images of CLANs are shown in Figures  1A-1C. Each coverslip was assessed at 10 locations ( Fig. 1D) with approximately 100 to 150 cells per coverslip. Six to 12 coverslips were evaluated per treatment group.
All CLAN counting was done in a masked manner. CLANpositive cells (CPCs) were defined as any cell containing at least one CLAN or multiple CLANs. The formation of CLANs was compared using the percentage of CPCs, which is calculated by dividing the number of CPCs by the number of DAPI-positive cells.

Western Immunoblotting
Primary NTM cells were cultured in six-well plates as previously described. When cells were confluent, they were treated with 5 ng/mL TGFb2 with or without the TGFb pathway inhibitors described previously for 24 hours. After treatment, whole cell lysates were collected using M-PER buffer (Thermo Scientific) or two-dimensional electrophoresis sample rehydration buffer. After protein estimation using the DC protein assay kit (Bio-Rad, Hercules, CA, USA) or the EZQ kit (Thermo Scientific), 38 lg total protein was used for SDS-PAGE. After electrophoresis, proteins were transferred onto a polyvinylidene fluoride membrane. The membrane was blocked with Odyssey blocking buffer (Li-Cor, Licoln, NE, USA), incubated with one of the following antibodies: rabbit polyclonal anti-a-smooth muscle actin antibody (1:1000  After incubation, the membrane was washed, incubated with a secondary antibody (goat anti-rabbit or goat anti-mouse) conjugated with horseradish peroxidase (Cell Signaling Technology). Signals were developed using the SuperSignal West Femto Maximum Sensitivity Substrate (Thermo Scientific). Images were taken using Odyssey Fc Dual-Mode imaging system (Li-Cor).

Statistical Analysis
The percentage of CPCs was compared using 1-way ANOVA followed by Dunnett's multiple comparisons post hoc test (GraphPad Prism 6.02; GraphPad Software, Inc., La Jolla, CA, USA). Data are presented as mean 6 SEM, with the significance level set at P < 0.05.

Smad and Non-Smad Pathway Inhibitors Prevented CLAN Formation
We first studied whether inhibition of Smad and/or non-Smad pathways would inhibit CLAN formation. We treated human NTM cells with TGFb2 together with inhibitors against the TGFb pathways (SB431542), the Smad pathway (SIS3), the ERK pathway (U0126), the JNK pathway (SP600125), the P38 pathway (SB203580), or the ROCK pathway (Y27632). Because CLAN formation has been shown to peak after 10 to 14 days of TGFb2 exposure, 47 we treated NTM cells for 10 days to ensure CLAN induction. Data are presented as the percentage of CPCs.
The formation of CLANs was mostly consistent in the two NTM cell strains studied. The exceptions in NTM1022-02 cells are (1) the ROCK pathway inhibitor showed partial inhibition of CLAN formation and (2) the P38 pathway inhibitor alone slightly elevated CLAN formation (P < 0.05 versus control; Fig.  2B). These data suggest that the response of TM cells to individual pathway inhibitors may be strain dependent.
Because there is a difference in CLAN prevention between JNK and ERK inhibitors, we studied their efficacy in inhibition of JNK and ERK phosphorylation using Western blotting (WB) as a validation. We found that both JNK and ERK inhibitors blocked the phosphorylation of JNK and ERK, respectively (Fig. 2C).

TGFb Pathway Inhibitors Resolved Already Formed CLANs
Although we found that TGFb pathway inhibitors prevented TGFb2-induced CLAN formation, it is still unclear whether they can also resolve already formed CLANs. Therefore, we pretreated NTM cells with TGFb2 for 10 days to induce CLANs and then treated them with individual inhibitors for 1 hour. Cotreatment of TGFb2 and those inhibitors for 10 days was also incorporated as a positive control.

Differential Effects of Smad and Non-Smad Pathway Inhibition on Actin Stress Fiber Formation
It is unclear whether the TGFb pathway regulates CLANs and actin stress fibers in the same manner. Therefore, we evaluated the effects of individual TGFb pathway inhibitors on actin stress fibers in TM cells. It is our belief that a compound that is able to prevent/remove CLANs but has minimal impact on normal actin cytoskeleton structures such as stress fibers will treat one of the glaucomatous pathologies (CLANs), while potentially preserving/restoring the normal actin cytoskeleton.
As previously reported, 44 TGFb2 increased and reorganized actin stress fiber formation compared to control (Figs. 4A, 4B). Transforming growth factor b receptor and Smad3 inhibitors restored actin stress fibers to baseline levels (versus control). The JNK or P38 inhibitors had little effect on TGFb2-induced stress fibers. The ERK inhibitor, although it did not block TGFb2-induced stress fibers, seemed to preserve the normal arrangement of those fibers. Different from these other inhibitors, 1-hour treatment with the ROCK inhibitor almost completely eliminated stress fibers. However, some stress fibers reformed after 10-day cotreatment with TGFb2 plus ROCK inhibitor.
Because a-smooth muscle actin (a-SMA) is a key component of CLANs 12 and is closely associated with actin, we studied whether kinase inhibitors affect TGFb2-induced a-SMA expression in HTM cells using WB. We did not observe obvious inhibition of a-SMA expression (Fig. 4C).

DISCUSSION
In this study, we found that TGFb2-induced CLANs in NTM cells could be both prevented and resolved using different TGFb pathway inhibitors. These inhibitors had different effects on actin stress fibers, which did not completely match their effects on CLANs.
Our results suggest that TGFb2 induced CLAN formation occurs via both the Smad and non-Smad TGFb pathways (summarized in Fig. 5). However, the contribution of individual TGFb2 signaling pathways to CLAN formation seems to be different. In the two NTM cell strains studied, both the Smad pathway and non-Smad ERK pathways seemed to play major roles in TGFb2-induced CLAN formation. The JNK and P38 pathway only partially participated in CLAN induction, whereas the contribution of the ROCK pathway was cell strain dependent. However, the overall trend was consistent.
O'Reilly et al. previously showed that, in bovine TM cells, the TGFb receptor inhibitors LY-364997 and SB431542 partially inhibited TGFb2-induced CLAN formation, 47 which is different from our findings where we found total inhibition. However, the Smad3 inhibitor SIS3 completely inhibited CLAN formation, which is consistent with our observations in NTM cells. The difference between the two studies may be due to the difference in species, suggesting that for results generated in nonhuman TM cells/tissues, a confirmation in human TM is necessary.
We also studied whether different TGFb pathway inhibitors could remove already formed CLANs, which to our best knowledge, has not been previously determined. It surprised us that only 1-hour treatment with TGFb receptor inhibitor, Smad inhibitor, ERK inhibitor, and ROCK inhibitor completely or partially removed already formed CLANs. Because 1 hour is generally believed to be too short to affect both transcription and translation, these changes are very likely to be due to alterations in protein modification, especially protein phosphorylation. Alternatively, CLAN formation may be a more dynamic process than initially thought, and TGFb signaling is required to maintain CLANs. Therefore, even a short-term disruption of the TGFb pathway is able to disrupt CLANs.
Whether the formation of CLANs and actin stress fibers share a similar mechanism is not clear. Our findings suggest that each TGFb signaling pathway had different impacts on actin stress fiber formation, which was different from their impact on CLANs. Although it is known that relaxation of the actin cytoskeleton (disassembly of actin stress fibers) is a valid approach to lower IOP (as seen with ROCK inhibitors), it still remains unknown if long-term disruption of normal actin stress fibers will impair TM function. Also, the most common side effect with the use of ROCK inhibitors is ocular hyperemia because they also ''relax'' conjunctival blood vessels similar to their effect on the TM. Therefore, if a compound is able to remove CLANs with minimal or no changes in actin stress fibers, it may have ocular hypotensive effects with less side effects such as hyperemia.
This is our first attempt to dissect the role of individual TGFb signaling pathways in CLAN formation. There are still many unanswered questions.
First, it is still unclear whether these individual pathways work independently or in a synergistic manner. We found that inhibition of either the Smad or Non-Smad pathway was able to completely inhibit CLAN formation. Therefore, we believe that these pathways are more likely to work synergistically. If they FIGURE 5. Hypothesized roles of the TGFb2 pathway and inhibitors in CLAN formation. Solid green lines, pathways that may play major roles in CLAN formation or maintenance; dotted lines, pathways that may play minor roles in CLAN formation or maintenance. do work together, it will be important to determine at which molecules this cross-regulation occurs.
Second, the downstream molecules of the TGFb pathway that mediate CLAN formation are not entirely clear. Peters et al. performed a proteomic study using spreading TM cells. 48 The authors first treated confluent TM cells with dexamethasone (DEX) to induce CLAN formation or with ethanol as a vehicle control. Then they dissociated and seeded TM cells into fibronectin coated dishes. The TM proteins were fractionated and compared using mass spectrometry. They observed a change in 318 cytoskeletal proteins, some of which contained phosphorylated residues suggesting DEX affects the expression of these proteins at the transcriptional and translational levels. That study provided insightful information of CLAN-related proteins. However, many adherent cell types form transient CLANs when they are first attaching and spreading, whereas only TM cells form and retain CLANs when they are confluent. Therefore, it is still unclear whether the CLANs formed during cell spreading share the same signaling pathway and biological components as the CLANs formed in confluent cells. Also, it is known that TGFb2 induces mesenchymal-like features in the TM including a-SMA, 44 which is an important component of CLANs. 12 We found that TGFb2-induced a-SMA was not obviously affected by the inhibitors after a 24-hour cotreatment, which is different from a study reported by Padmanabhan et al. in which the authors showed the ROCK inhibitor Y27632 decreased a-SMA after a 2-hour treatment without TGFb2. 44 We believe this could result from different treatment times and regimens. A prolonged treatment may lead to desensitization. Also, the presence of TGFb2 may counteract the effect of Y27632, especially after prolonged treatment because Y27632 is a rapid acting compound. However, our findings do not exclude the role of a-SMA in CLAN formation because our WB data did not provide information of posttranslational modification of a-SMA. Besides a-SMA, myosin light chain kinase (MLCK), a key component of the Rho-Rock pathway and stress fiber formation, may also be an important player in CLAN formation. Although there has been no direct evidence, our recent publication shows that myosin light chain, a downstream molecule and substrate of MLCK, is in close proximity to CLANs. 49 Further research is required to determine whether MLCK is involved in the initial assembly of CLANs and/or the maintenance of CLANs.
Third, although both glucocorticoids and TGFb2 are able to induce morphologically similar CLANs via different pathways, it is unknown if TGFb2-induced CLANs and DEX-induced CLANs are biochemically and/or mechanically identical. If the mechanisms are the same, there must be a set of common mediator molecules shared by the two pathways.
Finally, the role of connective tissue growth factor (CTGF) in CLAN formation has not yet been determined. CTGF is a profibrotic cytokine that mediates TGFb2 signaling in the TM. 50 CTGF is among the most highly expressed genes in the TM. 51 It is elevated in the AH of exfoliation glaucoma patients 50,52 and plays a critical role in cell migration, adhesion, proliferation, matrix production, and mediates several of the downstream actions of TGFb. 53 Overexpression of CTGF in a mouse model induces ocular hypertension and optic nerve damage. 54 ROCK inhibitors also affect both TGFb2 and CTGF-induced cellular and cytoskeletal changes in a similar manner. 54,55 Therefore, more research is needed to determine the effect of CTGF and its inhibitors on CLAN formation.
In summary, we found that different TGFb signaling pathways play different roles in the inhibition and disassembly of CLANs, as well as in actin stress fiber formation. This information may help in the development of novel ocular hypotensive agents.