Human TM cells were isolated from donor eyes using a blunt dissection technique followed by an extracellular matrix digestion protocol, exactly as previously described (Stamer et al., 1995). TM cells were characterized following established standards, including confirmation of proper TM cell morphology, and dexamethasone (100 nM for 5 days) induction of myocilin protein by immunofluorescence microscopy and western blot analysis according to consensus recommendations (Keller et al., 2018). Cells were cultured in low glucose Dulbecco’s modified Eagles’s medium (DMEM), containing 10% fetal bovine serum (FBS), 100 U/ml penicillin, 0.1 mg/ml streptomycin and maintained in humidified air containing 5% CO2 at 37°C. See Table 1 for a list of the TM cell strains used in this study, with available donor information.

Cav1 expression in hTM cells was silenced using replication-deficient adenoviral constructs containing CAV1-silencing shRNAs and a GFP reporter (human adenovirus Type 5, dE1/E3 backbone, SKU# shADV-204148; Vector Biolabs, Malvern, PA). Control cells were transduced with particles containing the same vector encoding GFP and scrambled control shRNAs (Cat. #1122, Ad-GFP-U6-shRNA; Vector Biolabs). The optimal multiplicity of infection (MOI) for maximal Cav1 knockdown and minimal viral side effects was determined to be 10–15. We tested 1, 5, 10, 15, 20, and 50 MOI in the hTM120 cell strain to determine MOI that produced the greatest level of Cav1 knockdown with the least amount of observable undesirable effects, including change in cell morphology, and elevated expression of αSMA. As per manufacturer guidelines, cells were incubated in adenovirus solution (1 ml per well in 6-well plate; 2 ml per T25 flask; or 5 ml per T75 flask) for 6 h (5% CO₂, 37°C), then topped up with fresh 10% FBS DMEM. Proteins were extracted at 72–96 h post viral transduction.

Cells were seeded at confluence on 6-well Flexcell plates (Flexcell International Corp., Burlington, NC) that were coated with type IV collagen-coated as previously described (Hauser et al., 2015; Elliott et al., 2016; Youngblood et al., 2020). Cells were incubated in 1% FBS DMEM 2 h prior to starting cyclic stretch. In vivo, TM cells are bathed and nourished in aqueous humor, which contains about 1% serum proteins (Freddo 2013; Keller et al., 2018). Thus, in an attempt to simulate an in vivo environment, we chose to use 1% FBS instead of 0% FBS for use in experiments, which can unduly stress the cells. Cell stretch (20% stretch) was performed at frequency of 1 Hz, mimicking the ocular pulse, for 24 h in 1% FBS DMEM using the Flexcell FX-5000 (Flexcell International Corp., Burlington, NC). Evaluation of cells at 24-h was selected based on previous studies (Bradley et al., 2001; Bradley et al., 2003; Vittal et al., 2005).

After 24 h of cyclic stretching, Flexcell plates were immediately placed on ice, cells were rinsed three times with ice-cold phosphate-buffered saline (PBS) and harvested by scraping into 2X Laemmli sample buffer with 1/10 dilution of beta mercaptoethanol (200μl/well of a 6-well plate), as performed previously (Dismuke et al., 2014). Protein lysates were boiled for 5 min and stored at −80°C. Equal volumes of protein lysate were loaded into 12% polyacrylamide gels, and proteins were separated by SDS-PAGE and transferred electrophoretically to nitrocellulose membranes using transfer buffer with 20% methanol. Blots were blocked at room temperature, on a rocker, for 60 min with 5% bovine serum albumin (BSA) in tris-buffered saline with 0.2% tween-20 (TBS-T). Primary antibodies were diluted in 5% BSA in TBS-T, and blots were incubated in primary antibody solution at 4°C overnight, on a rocker. The phosphorylation status of MLC was used as a surrogate indicator for Rho/ROCK activity and contractile tone as done previously (Kimura et al., 1996; Kureishi et al., 1997; Kaneko-Kawano et al., 2012). For analysis of MLC phosphorylation following Cav1 knockdown in TM cells plated on soft or hard surface, we combined data from multiple experiments that were completed for several different datasets, where cells were either plated on hard or soft surfaces. Therefore, some cells were treated with DMSO (0.05%) for 24 h prior to cell extraction, and others were not. Primary antibodies used are as follows: rabbit polyclonal anti-MLC (catalog number 3672; 1/1,000; Cell Signaling Technology), rabbit polyclonal anti-phospho-MLC (catalog number 3674; 1/1,000; Cell Signaling Technology), mouse monoclonal anti-alpha-tubulin HRP conjugate (catalog number 12351; 1/1,000; Cell Signaling Technology), rabbit monoclonal anti-caveolin-1 HRP conjugate (catalog number 12506; 1/1,000; Cell Signaling Technology), mouse monoclonal anti-phospho-caveolin-1 (catalog number 611338; 1/1,000; BD Biosciences), rabbit polyclonal anti-PTRF/cavin-1 (catalog number ABT131; 1/1,000; Millipore Sigma), mouse monoclonal anti-RhoA (catalog number sc-418; 1/200; Santa Cruz), and mouse monoclonal anti-alpha-smooth muscle actin (catalog number A2547; 1/5,000; Millipore Sigma). The blots were then washed three times (10 min in TBS-T). Membranes probed with non-HRP-conjugate antibodies were then incubated with goat anti-rabbit-HRP or anti-mouse-HRP secondary antibodies (1:5,000) for 1 h at room temperature, then washed three times (5 min in TBS-T). Blots were exposed using the ChemiDoc Imaging System (BioRad) and densitometry of protein bands was measured by Image Lab software (BioRad). It was not possible to run all samples on the same blot as there were not enough wells in the polyacrylamide gel to allow for all conditions for each sample. We addressed this issue in two ways. Firstly, all conditions for each cell strain or independent experiment were always probed on the same blot. Secondly, target protein band signal intensity was normalized to corresponding alpha-tubulin signal intensity on the same membrane, to correct for loading and variations in protein concentration for each respective sample. Additionally, pMLC and MLC were probed on separate blots due to inability to strip and reprobe for MLC proteins. For this reason, pMLC and MLC proteins were normalized to α-tubulin on respective gels to correct for loading, prior to pMLC normalization to MLC.

We tested two Cav-1 scaffolding domain peptides and a vehicle control. Cavtratin (catalog number 219482; Millipore Sigma) is a Cav-1 scaffolding domain peptide (amino acids 82–101) fused at the N-terminus to the cell-permeable Antennapedia internalization sequence (43–58). Cavnoxin was synthesized by Elim Biosciences (Hayward, CA) is identical to cavtratin, although it contains the T90,91 and F92 substitutions to alanine, disrupting the inhibitory action of the scaffolding domain on eNOS (Bernatchez et al., 2005), along with N-terminus acetylation and C-terminus amidation for peptide stability. The Antennapedia internalization sequence alone was used as a peptide control. Cells were washed once with serum-free media prior to treatment. Peptides were diluted in serum-free media and cells were treated for 24 h prior to cell lysis. During the dose optimization study, we found that 1 μM of peptides was tolerated well by the cells, but 10 μM of peptides [as used previously (Bernatchez et al., 2011)] had adverse effects on readouts and cell morphology. For example, high doses of the peptides caused the cells to shrink and detach from the substratum.

Human TM cells were grown to confluence in T-75 cell culture flasks in 10% FBS DMEM. Adenoviral transduction to silence Cav1 expression was performed 72–96 h prior to cell lysis. GFP signal was confirmed using fluorescence microscopy 48 h post-transduction. Cells were incubated in 1% FBS DMEM 2 h prior to cell lysis. The Active Rho Detection Kit (catalog number 8820; Cell Signaling) was used to perform the Rho activity assay. Cells were washed once with ice cold PBS and lysed with Lysis/Binding/Wash Buffer supplemented with 1 mM of phenylmethylsulfonyl fluoride (LBP). Cells were scraped and transferred to appropriate centrifuge tubes. Tubes were centrifuged at 16,000 XG for 15 min at 4°C, and supernatant was transferred to a new tube, while cell pellet was discard. Protein concentration was determined using the Pierce BCA Protein Assay Kit (catalog number 23225; Thermofisher Scientific). Lysate from each of the treatments (500 μg) was incubated with GST-Rhotekin-Rho binding domain-agarose slurry by gently rocking at 4°C for 1 h (as per the manufacturer’s instructions). Equal amounts of cell lysate protein from untreated cells were incubated either with GTPγS or with GDP, which served as positive and negative controls, respectively. The agarose beads were then washed three times with LBP and suspended in reducing sample buffer (2X Laemmli sample buffer with 200 mM dithiothreitol). The samples were vortexed and incubated at room temperature for 2 min. The samples were then centrifuged at 6000xG for 2 min, and the eluted samples were boiled for 5 min and stored at −80°C for future use. The GTP-bound and total Rho was detected by western blot analysis using Rho rabbit antibody (specific for RhoA, RhoB, and RhoC) supplied in Active Rho Detection Kit. Blots were exposed using the ChemiDoc Imaging System (BioRad) and densitometry of protein bands was measured by Image Lab software (BioRad). Signal obtained from refined protein lysate containing GTP-bound Rho was normalized to total Rho from crude protein lysate.

Circular glass coverslips (12 mm diameter) were sterilized with 70% ethanol and inserted into wells of Greiner 24-well tissue culture treated plates (catalog number M8812; Millipore Sigma). Two days before fixation, cells were seeded at a density of 0.3 × 10⁵ cells per well, and incubated in 10% FBS DMEM overnight to allow the cells to adhere to the glass coverslips. Cell media was then replaced with 1% FBS DMEM and incubated overnight. Cells were then rinsed twice with ice cold PBS and fixed with 4% PFA for 30 min on ice. Cells were washed 3 times with TX solution (1/1,000 dilution of Triton X-100 in PBS) and blocked at room temperature for 60 min in blocking buffer (1/10 dilution of goat serum in TX-solution). Cells were then probed at 4°C overnight on a rocker with rabbit monoclonal anti-caveolin-1 primary antibody (catalog number 3267; 1/400; Cell Signaling Technology), diluted in blocking buffer. Cells were washed 3 times for 10 min with TX solution, and then probed with species-specific secondary antibody (Alexa Fluor 488; goat anti-rabbit; final concentration 3.75 μg/ml; catalog number 111–545–144; Jackson Immuno Research Labs), diluted in TX-solution for 2 h at room temperature on a rocker. Cells were then washed 3 times on a rocker for 10 min each, and the protocol was repeated with mouse monoclonal anti-RhoA (catalog number sc-418; 1/50; Santa Cruz) and respective species-specific secondary antibody (Alexa Fluor 647; goat anti-mouse; final concentration 3.75 μg/ml; catalog number 115-605-146; Jackson Immuno Research Labs). Cells were then incubated with DAPI (1 μg/ml diluted in PBS) for 30 min at room temperature. Cells were briefly rinsed once with PBS, and cover slips were mounted onto microscope slides using Immu-Mount (catalog number 9990402; Fisher Scientific). Images were captured using a Nikon A1R confocal laser scanning microscope with 4X air, 20X air objective lenses. For the signal emitted by DAPI, the excitation wavelength was 405 nm with emission filters set to detect wavelengths between 425 and 475 nm. For the signal emitted by Alexa Fluor 488, the excitation wavelength was 488 nm with emission filters set to detect wavelengths between 500 and 550 nm. For the signal emitted by Alexa Fluor 647, the excitation wavelength was 640 nm with emission filters set to detect wavelengths between 663 and 738 nm. Fiji (ImageJ) software (United States National Institutes of Health) was used to identify subcellular colocalization of Cav1 and RhoA, by measuring signal intensities as a function of cross-sectional distance per image, as previously described (De Ieso et al., 2019; Pei et al., 2019). Unidimensional coincident peaks from Cav1 and RhoA signal were interpreted as colocalization. The inclusion criteria for selecting cross sections for analysis were as follows: 1) to be considered a peak, the maximal Y-value of the peak must have been greater than 10 units. 2) For each single image depicting a specific cell line, three cross-sectional lines were drawn. 3) Each line needed to pass through at least three cells.

At 96 h post viral transduction (as described above) cells were washed 3 times with ice-cold PBS and harvested by scraping into detergent-free buffer (PBS with 0.5 mM ethylenediaminetetraacetic acid, catalog number 03690; Millipore Sigma, Halt Protease Inhibitor Cocktail, catalog number 78425; ThermoFisher Scientific, and Phosphatase Inhibitor Cocktail 2, catalog number P5726; Millipore Sigma). Cell lysate was homogenized by drawing lysate up and down 10 times with a 27-gauge needle. Lysates were then gently centrifuged (800 XG) for 5 min to pellet nuclei and large debris. The supernatant (cytosol + membrane fractions) was removed and centrifuged at 16,000 XG for 15 min to separate the cytosol and membrane fractions. Detergent-free buffer (50 μl) was added to the nuclear fraction, which was then stored at −80°C. After the cytosolic and membrane fractions were separated, the supernatant (cytosol-enriched fraction) was removed and stored at −80°C. The membrane-enriched pellet was washed twice with detergent-free buffer and centrifuged again at 16,000 XG for 15 min. The supernatant was removed and discarded, and the pellet was solubilized in a buffer containing 1% Triton X-100 (40 μl) to lyse the membrane pellet. The membrane fraction was also stored at −80°C until protein quantification. Protein concentration was determined using the Pierce BCA Protein Assay Kit (catalog number 23225; Thermofisher Scientific). Protein lysates were mixed with reducing sample buffer (4X Laemmli sample buffer with 1/10 dilution of beta mercaptoethanol) and boiled for 5 min. Equal volumes of protein lysate (10 μg per well) were loaded into TGX Stain-Free FastCast 12% polyacrylamide gels (catalog number 1610184; Biorad). Proteins were separated by SDS-PAGE and gels were activated prior to being transferred electrophoretically to nitrocellulose membranes using transfer buffer with 20% methanol. Protein expression was corrected for loading via normalization to total protein. Alpha-tubulin was used as a cytosolic marker protein and Cav1 was used as a membrane marker protein to confirm purity of membrane and cytosolic fractions.

Graphing and statistical analyses were performed using GraphPad Prism v.9.2.0 (GraphPad Software, La Jolla, CA). A p value of 0.05 or less was considered significant and data are presented as mean ± SEM, unless otherwise stated. Significance was determined using paired and unpaired t-tests. We used unpaired t-tests when comparing the means of two independent or unrelated groups, for example, basal pMLC levels across different cell lines. Alternatively, we used a paired t-test when comparing means of the same group or item (the experiment/cell line) under two separate scenarios (e.g., stretch verses no stretch). This is to account for variability between cell strains and experiments, and a paired t-test is applicable here because within any given experiment, the cell strains were under identical conditions until the treatments (stretch or virus) were applied. Outliers were defined as values that lie two standard deviations outside of the mean.

Actin stress fiber formation (an effector of Rho/ROCK activity) is elevated in human TM cells depleted of Cav1 (Aga et al., 2014), and smooth muscle contractility is elevated in the absence of Cav1 in several other cell types (Shakirova et al., 2006). Therefore, we hypothesized that Cav1 depletion in the TM results in an elevation of Rho/ROCK activity and subsequently contractile tone. To test our hypothesis, we silenced Cav1 in human TM cells and monitored the phosphorylation status of myosin light chain (MLC), a surrogate indicator for Rho/ROCK activity and contractile tone (Kimura et al., 1996; Kureishi et al., 1997; Kaneko-Kawano et al., 2012). Results show a reproducible reduction of Cav1 in TM cells by over 50% using CAV1-silencing shRNAs compared to scrambled shRNAs (Figure 1A, 0.49 ± 0.032, n = 9, p < 0.0001). We also monitored a second indicator of Cav1 knockdown, CAVIN1, a Cav1 interacting transcription factor (Hill et al., 2008; Liu and Pilch 2008; Hansen et al., 2013). Accordingly, we observed that CAVIN1 was significantly reduced to similar levels as Cav1 in Cav1-deficient TM cells compared to control (Figure 1B, 0.52 ± 0.041, n = 7, p < 0.0001). In contrast, Cav1 knockdown had no effect on expression levels of αSMA, RhoA, or MLC (Supplementary Figure S1). In terms of MLC activation, we found that phosphorylated MLC was significantly elevated in Cav1-deficient TM cells compared to Cav1-competent TM cells (Figure 1C, 1.31 ± 0.1, n = 10, p = 0.016), suggesting Cav1 negatively regulates Rho/ROCK activity, and subsequently contractile tone in human TM cells.

TM cells cultured on rigid substrates results in transcriptomic modifications for genes involved in ECM regulation, mechanotransduction, and glaucoma pathogenesis (Thomasy et al., 2012; Tie et al., 2020). Additionally, downstream effectors of Rho/ROCK signaling, such as stress fiber formation, are elevated in TM cells cultured on rigid substrates (Schlunck et al., 2008). Therefore, we hypothesized that the elevation of pMLC in response to Cav1 knockdown would be less pronounced when cells were plated on a hard surface rather than a soft surface. To test this hypothesis, we compared the effect of Cav1 knockdown between TM cells plated on flexible silicone bottom plates (soft) verses plastic (hard). As such, phosphorylated MLC was significantly elevated in Cav1-deficient TM cells compared to Cav1-competent TM cells when plated on a soft surface (Figure 2B, 1.37 ± 0.07, n = 24, p < 0.0001). In contrast, phosphorylated MLC was slightly elevated in Cav1-deficient TM cells compared to Cav1-competent TM cells, although not significantly for TM cells plated on hard surfaces, (Figure 2A, 1.22 ± 0.17, n = 12, p = 0.224). These data suggest that substrate stiffness influences Cav1-facilitated regulation of Rho/ROCK activity in human TM cells.

We observed that the dampened response by TM cells plated on stiff surfaces could be segregation into two groups; “responders” having elevated pMLC and “non-responders” having low levels of pMLC with Cav1 knockdown. We hypothesized that basal (shSCR-treated) pMLC levels are high in non-responders, which would result in a “ceiling effect”; meaning further elevation of pMLC in response to Cav1 knockdown would be less pronounced. As depicted in Figure 3A, we discovered that in shSCR-treated TM cells, basal pMLC levels were more than 80% lower in responders (0.31 ± 0.12, n = 5) compared to non-responders (1.63 ± 0.21, n = 4, p = 0.0007). This suggested that elevation of pMLC in response to Cav1-deficiency primarily occurs in hTM cells where basal pMLC levels are not already elevated. As before, we reproducibly and equivalently reduced Cav1 in TM cells with CAV1-silencing shRNAs compared to scrambled shRNAs in non-responders (Figure 3B, 0.38 ± 0.15, n = 4, p = 0.027) and responders (Figure 3C, 0.38 ± 0.06, n = 6, p = 0.0002) and assayed for Rho activity. We observed that within any given experiment, changes in pMLC levels were directly proportional to changes in Rho activity following Cav1 knockdown. As such, in non-responders, Rho activity did not significantly change in Cav1-deficient compared to Cav1-competent TM cells (Figure 3D, 0.927 ± 0.03, n = 4, p = 0.097). In contrast, Rho activity in responders was significantly elevated in Cav1-deficient compared to Cav1-competent TM cells (Figure 3E, 1.23 ± 0.07, n = 6, p = 0.017), suggesting that Cav1-knockdown elevates MLC phosphorylation via Rho/ROCK pathway. As positive and negative controls for the Rho activity assays, we treated cell lysates with GTPγS or GDP, respectively (Supplementary Figure S2).

The Cav1 scaffolding domain has been shown to inhibit signal transduction via sequestration of a number of key signaling molecules, including RhoA (Oka et al., 1997; Engelman et al., 1998; Razani et al., 1999; Bucci et al., 2000; Razani and Lisanti 2001; Gratton et al., 2003; Labrecque et al., 2003; Lu et al., 2018). For example, Cav1 physically interacts with, and inhibits RhoA in several cell types (Gingras et al., 1998), likely via the scaffolding domain (Okamoto et al., 1998; Taggart et al., 2000). Thus, as MLC phosphorylation is elevated following Cav1 knockdown, we hypothesized that treatment of TM cells with the Cav1 scaffolding domain peptides would mimic upregulation of Cav1 activation and cause MLC dephosphorylation. Interestingly, we found that treatment of TM cells with cavtratin for 24 h resulted in a 22% reduction in relative Cav1 protein as compared to control (Figure 4A, 0.775 ± 0.07, n = 7, p = 0.021). In contrast, treatment with cavnoxin had no significant effect on relative Cav1 protein levels as compared to control (Figure 4A, 0.933 ± 0.06, n = 7, p = 0.332). These data suggest that treatment with cavtratin initiated a negative feedback loop, causing a downregulation of Cav1 protein expression. In addition to decreased Cav1 expression, TM cells treated with cavtratin exhibited a 30% reduction in MLC phosphorylation as compared to control (Figure 4B, 0.699 ± 0.05, n = 7, p = 0.001). Cavnoxin was less efficacious than cavtratin, exhibiting an 18% reduction in MLC phosphorylation in TM cells as compared to AP control (Figure 4B, 0.82 ± 0.07, n = 7, p = 0.043). These data support the hypothesis that Cav1 negatively regulates Rho/ROCK signaling in human TM cells, via the Cav1 scaffolding domain.

Caveolae are mechanotransducers in some tissues (Yu et al., 2006). For example, mechanical stimulation induces phosphorylation of Cav1 in breast cancer cells (Joshi et al., 2012). Since TM cells are subject to continual cyclic stretch (Xin et al., 2018), we aimed to determine whether cyclic mechanical stretch induces Cav1 phosphorylation in human TM cells. In Cav1-competent TM cells, 24 h of cyclic stretch caused a significant increase in Cav1 phosphorylation as compared to no stretch (Figure 5A, 1.25 ± 0.03, n = 9, p < 0.0001). In Cav1-deficient TM cells, 24 h of cyclic stretch elevated phosphorylated Cav1 levels to similar levels, but not significantly (Figure 5A, 1.27 ± 0.2, n = 10, p = 0.213). These results show that cyclic mechanical stretch stimulates phosphorylation of Cav1 in Cav1-competent TM cells, and that this effect is disrupted in the TM cells with lower levels of Cav1.

The ability of the conventional outflow tissues to adapt to fluctuating biomechanical loads is crucial to IOP homeostasis (Overby et al., 2014; Stamer et al., 2015). As MLC dephosphorylation and subsequent TM relaxation elevates outflow facility and reduces IOP (Rao et al., 2001; Rao et al., 2005; Ren et al., 2016), we hypothesized that TM cells respond to cyclic mechanical stretch by dephosphorylating MLC, and subsequently relaxing. Indeed, 24 h of cyclic stretch caused a significant decrease in MLC phosphorylation as compared to no stretch in Cav1-competent TM cells (Figure 5B, 0.693 ± 0.07, n = 9, p = 0.002). One hour of cyclic stretch was not sufficient to cause MLC dephosphorylation in Cav1-competent TM cells (Supplementary Figure S4). We then tried to determine whether the stretch-induced relaxation was Cav1 dependent. We observed that the response was more variable and did not reach significance in Cav1-deficient TM cells, whereby 24 h of cyclic stretch caused a 23% decrease in MLC phosphorylation as compared to no stretch (Figure 5B, 0.772 ± 0.11, n = 10, p = 0.059). These data suggest that cyclic mechanical stretch stimulates dephosphorylation of MLC in Cav1-competent TM cells, and that this effect is disrupted in the TM cells with lower levels of Cav1. We then only looked at TM cells that had been stretched for 24 h, and within this group we found that Cav1 was reduced by 53% with CAV1-silencing shRNAs compared to scrambled shRNAs (Figure 5C, 0.465 ± 0.02, n = 9, p < 0.0001). Moreover, phosphorylated MLC was significantly elevated by 31% in Cav1-deficient compared to Cav1 competent TM cells (Figure 5C, 1.31 ± 0.14, n = 10, p = 0.049). Importantly, we observed that 24 h of cyclic stretch had no effect on protein expression of Cav1, CAVIN1, αSMA, MLC, or RhoA (Supplementary Figure S3). These data suggest that in the presence or absence of cyclic mechanical stretch, MLC phosphorylation is elevated in Cav1-deficient TM cells.

Cav1 physically interacts with Rho GTPases in non-TM cell types, likely via its scaffolding domain (Gingras et al., 1998; Okamoto et al., 1998; Taggart et al., 2000; Peng et al., 2007; Peng et al., 2008; Arpaia et al., 2012). These data plus our results showing that Cav1 regulates RhoA activity in TM cells, led us to hypothesize that Cav1 colocalizes with RhoA in human TM cells. Using immunofluorescent microscopy, we observed that the majority of Cav1 is expressed on the plasma membrane of TM cells. Similarly, we found that RhoA expression is primarily concentrated at cell borders, with some distributed throughout the cytoplasm (Figure 6A and Supplementary Figure S5A). Colocalization was confirmed in confocal images analyzed in cross sections (Figure 6B and Supplementary Figure S5B).

Our data is consistent with previous studies, showing RhoA localizes to both the cytoplasm and plasma membrane (Adamson et al., 1992); translocating from the cytosol to the plasma membrane upon activation (Philips et al., 1993; Boivin and Beliveau 1995; Fleming et al., 1996; Kranenburg et al., 1997; Michaelson et al., 2001). Although, RhoA is negatively regulated when compartmentalized within caveolar lipid rafts (Nuno et al., 2012). Thus, we interrogated the effect of Cav1 expression levels on RhoA localization in TM cells. Under baseline conditions, Cav1 near exclusively localized to the membrane, with close to undetectable levels in the cytosolic fraction of TM cells (Figure 7A, 0.004 ± 0.002, n = 6, p < 0.0001). Upon knockdown, Cav1 was reduced by 53% (Figure 7B, 0.47 ± 0.08, n = 6, p = 0.001). As expected, α-tubulin preferentially localized to the cytosolic as compared to membrane cellular compartment (Figure 7C, 5.5 ± 1.2, n = 6, p = 0.012). Similar to immunofluorescence studies under basal conditions (Figure 6), RhoA localized predominantly in the membrane verses cytosolic fractions (Figure 7D, 0.19 ± 0.04, n = 6, p < 0.0001). Interestingly, we found that RhoA protein levels in the membrane fraction were not significantly different between Cav1-competent and Cav1-deficient TM cells (Figure 7E, 0.72 ± 0.16, n = 6, p = 0.13). However, RhoA was significantly reduced in cytosolic fractions by 38% in Cav1-deficient verses Cav1-competent TM cells (Figure 7F, 0.62 ± 0.05, n = 6, p = 0.0004). These data suggest that while membrane-bound RhoA expression is unchanged, cytoplasmic RhoA expression is downregulated in Cav1-deficient TM cells.