TRPC3-GEF-H1 axis mediates pressure overload-induced cardiac fibrosis

Structural cardiac remodeling, accompanying cytoskeletal reorganization of cardiac cells, is a major clinical outcome of diastolic heart failure. A highly local Ca2+ influx across the plasma membrane has been suggested to code signals to induce Rho GTPase-mediated fibrosis, but it is obscure how the heart specifically decodes the local Ca2+ influx as a cytoskeletal reorganizing signal under the conditions of the rhythmic Ca2+ handling required for pump function. We found that an inhibition of transient receptor potential canonical 3 (TRPC3) channel activity exhibited resistance to Rho-mediated maladaptive fibrosis in pressure-overloaded mouse hearts. Proteomic analysis revealed that microtubule-associated Rho guanine nucleotide exchange factor, GEF-H1, participates in TRPC3-mediated RhoA activation induced by mechanical stress in cardiomyocytes and transforming growth factor (TGF) β stimulation in cardiac fibroblasts. We previously revealed that TRPC3 functionally interacts with microtubule-associated NADPH oxidase (Nox) 2, and inhibition of Nox2 attenuated mechanical stretch-induced GEF-H1 activation in cardiomyocytes. Finally, pharmacological TRPC3 inhibition significantly suppressed fibrotic responses in human cardiomyocytes and cardiac fibroblasts. These results strongly suggest that microtubule-localized TRPC3-GEF-H1 axis mediates fibrotic responses commonly in cardiac myocytes and fibroblasts induced by physico-chemical stimulation.

calcineurin/nuclear factor-activated T cells 8,9 , mitogen-activated protein kinases, phosphatidylinositol-3-kinase/Akt, and small GTPases, Ras, Rho, and Rac 10 . Among them, Rho-mediated signaling has been revealed as a critical mediator of fibrosis through actin cytoskeletal reorganization-dependent fibrotic gene transcription 11,12 . However, it is still obscure whether responsive Rho-mediated fibrosis can be clearly distinguished from reparative fibrosis during the development of heart failure.
The Rho GTPase activity is fundamentally regulated by its guanine nucleotide exchange factor (GEF), GTPase-activating proteins, and guanine nucleotide dissociation inhibitors, while physical and chemical stimuli primarily stimulate GTP binding to Rho through activating specific RhoGEFs 13 . Among 69 distinct RhoGEF homologues, RhoGEF12 reportedly controls both hypertrophy and fibrosis induced by pressure overload 14 and A-kinase anchoring protein-Lbc reportedly participates in myofibroblast formation of cardiac fibroblasts induced by angiotensin II or transforming growth factor (TGF)-β 15 . Although several RhoGEFs may participate in the development of cardiac remodeling, the RhoGEF that specifically encodes a signal to induce responsive fibrosis has not been identified.
NADPH oxidase isoform 2 (Nox2) is a microtubule-associated reactive oxygen species (ROS)-producing enzyme that acts as a key mediator of mechanotransductive signaling in normal hearts 16 . Nox2-deficient mice show specific suppression of pressure overload-induced cardiac fibrosis but not hypertrophy 17 . The intracellular Ca 2+ concentration plays a key role in receptor-stimulated sustained Nox2 activation, and we previously reported that mechanical stress-induced local Ca 2+ influx through transient receptor potential canonical (TRPC) 3 channel increases Nox2-mediated ROS production in neonatal rat cardiomyocytes (NRCMs) 18,19 . TRPC3 forms stable protein complex with Nox2 in myocardial T-tubule, which leads to amplification of ROS signaling in heart 19 . In addition, pharmacological inhibition of TRPC3 actually attenuates LV diastolic dysfunction as well as responsive fibrosis in mouse hearts with dilated cardiomyopathy. However, how the TRPC3-Nox2 axis regulates Rho-mediated responsive fibrosis is unclear. We here demonstrate that TRPC3 deletion specifically inhibits RhoA-mediated maladaptive fibrosis in pressure-overloaded mouse hearts. We also show that a microtubule-associated RhoGEF, GEF-H1, plays a key role in maladaptive fibrosis induced by mechanical stress and TGF-β stimulation.
GEF-H1 participates in TRPC3-mediated cardiac fibrosis. The small GTP-binding protein RhoA plays a central role in cardiac fibrotic signaling in both cardiomyocytes and cardiac fibroblasts 20 . TAC significantly increased myocardial RhoA activity (Fig. 2a), and the TAC-induced RhoA activation was significantly lower in TRPC3 (−/−) than TRPC3 (+/+) hearts. Mechanical stretch of cardiomyocytes is well accepted as an appropriate stimulation to initiate profibrotic factors release from cardiomyocytes during pressure overload 21,22 , while collagen production from myofibroblasts predominantly determines the severity of fibrosis in heart 22 . As TGF-β s are most prominent profibrotic factors that promote differentiation of cardiac fibroblasts into myofibroblasts and collagen production from myofibroblasts 22 , we examined which RhoGEF(s) can be activated upon TGF-β stimulation in rat cardiac fibroblasts. Pull-down assay using agarose-conjugated nucleotide-free RhoA mutant (RhoA G17A ) showed that intensities of three bands were significantly increased by TGF-β 2 stimulation (Fig. 2b). Among them, only one RhoGEF, GEF-H1, was detected from 110 kDa band by proteomic analysis. RhoGEFs are upstream positive regulators of Rho, and two GEFs, LARG and GEF-H1, are reportedly responsive to mechanical stress 23 . This implies that GEF-H1 acts as a common mediator of pressure overload-induced TRPC3-dependent fibrosis both in cardiomyocytes and cardiac fibroblasts. The level of GEF-H1 protein expression in TAC-operated TRPC3 (+/+) heart significantly increased, that was unchanged in TRPC3 (−/−) hearts (Fig. 2c). Furthermore, RhoA G17A agarose-dependent pull-down assays revealed that TAC clearly increased binding activity of GEF-H1 in TRPC3 (+/+) hearts, and the effect was absent in TRPC3 (−/−) hearts (Fig. 2c). In contrast, TAC did not change the level of LARG protein expression (Fig. 2d), suggesting that TAC-induced RhoA activation is mediated predominantly by GEF-H1 in 129 Sv mouse hearts. The TRPC3-GEF-H1 axis underlies mechanical stretch-induced RhoA activation and fibrotic gene expressions in NRCMs. Excess LV diastolic filling due to chronic pressure overload is believed as a major cause of cardiac fibrosis in vivo, and mechanical stretch of cardiomyocytes is accepted as an appropriate in vitro model to mimic chronic pressure overload in heart 22 . We have also reported that mechanical stretch-induced ATP/UDP release from cardiomyocytes through pannexin-1 channels triggers pressure overload-induced cardiac fibrosis in in vivo mouse hearts 21 . We next examined whether a TRPC3-GEF-H1 axis underlies mechanical stretch-induced RhoA activation and fibrotic gene expressions in NRCMs. Mechanical stretch of NRCMs increased RhoA activity and expression of two RhoA-dependent mRNAs, connective tissue growth factor (CTGF) and TGF-β 2, and these RhoA-dependent responses were suppressed by TRPC3 inhibition (Fig. 3a,b). Consistent with the results of TAC-operated heart, mechanical stretch-activated GEF-H1 was significantly suppressed in TRPC3-silencing NRCMs 19 (Fig. 3c). In addition, GEF-H1-silencing NRCMs also suppressed mechanical stretch-induced increases in CTGF mRNA (Fig. 3d,e). These results suggest that the TRPC3-GEF-H1 axis underlies Rho-dependent fibrotic responses of cardiomyocytes induced by mechanical stretch.
GEF-H1 is predominantly localized in microtubules, and both microtubule-dependent and -independent activation mechanism are involved in ligand-stimulated GEF-H1 activation 26 . Our previous study suggested a functional coupling between TRPC3 and microtubule-associated Nox2, and that TRPC3/Nox2-mediated production of reactive oxygen species (ROS) is highly correlated with the severity of heart failure 19 . Microtubule-associated GEF-H1 is reportedly activated by ROS through dissociation from tubulin 27 , and we previously reported that TRPC3 mediates mechanical stretch-induced Nox2 activation in NRCMs 19 . Therefore, we tested whether the TRPC3-Nox2 signaling pathway contributed to GEF-H1 activation. A pretty polymerized tubulin was observed in normal rat cardiac fibroblasts (Fig. 5a). TGF-β 2 stimulation significantly reduced the density of microtubule, which was canceled by TRPC3 inhibition. This result was correlated well with that of GEF-H1 activity (Fig. 4e). Inhibition of Nox2 by the treatment with Nox2 siRNA or diphenyleneiodonium (DPI) significantly suppressed the mechanical stretch-induced GEF-H1 activation in NRCMs (Fig. 5b,c). Despite almost complete knockdown of Nox2 protein by Nox2 siRNA, partial GEF-H1 activation was still observed in Nox2-silencing NRCMs. In contrast, mechanical stretch-induced GEF-H1 activation was abolished completely in NRCMs pretreated with taxol, a microtubule stabilizing agent (Fig. 5d). This indicates that mechanical-stretch induced GEF-H1 activation mainly depends on microtubule depolymerization in cardiomyocytes. As the residual GEF-H1 activation in Nox2-silencing NRCMs was completely abolished by okadaic acid, an inhibitor of PP2A phosphatase, which is reported to positively regulates GEH-H1 activity in a microtubule-independent manner 26 (Fig. 5e). These results suggest that TRPC3 participates in mechanical stress-induced GEH-H1 activation, in part through Nox2/ ROS-mediated microtubule depolymerization.
Inhibition of TRPC3 attenuates fibrotic responses of human cardiomyocytes and cardiac fibroblasts. Finally, we used human induced pluripotent stem cell (iPSC)-derived cardiomyocytes and human cardiac fibroblasts to investigate whether TRPC3-induced fibrogenic signaling is conserved across species. Mechanical stretch of human iPSC cardiomyocytes increased expression of CTGF, TGF-β 1 and TGF-β 2 mRNAs, which were significantly suppressed by TRPC3 inhibition (Fig. 6a-c). Stimulation of human cardiac fibroblasts with TGF-β 2 increased the levels of CTGF mRNA expression and α -SMA protein expression; both were suppressed by TRPC3 inhibition (Fig. 6d-g). These results suggest that TRPC3/Nox2 communication may be a common mechanism underlying activation of RhoA-dependent fibrogenic signaling induced by mechanical stretch and TGF-β stimulation in human cardiac cells.

Discussion
The roles of cytoskeletal alterations especially of microtubules, formed by polymerization of α -and β -tubulin, and desmin have been implicated in the development of cardiac hypertrophy and heart failure in   numerous experimental studies. The contractile myofilaments are reportedly reduced during the development of cardiac remodeling, while cytoskeletal proteins including microtubules are compensatively disorganized, depolymerized, and increased in amount thereby posing an increased load on myocytes which impedes sarcomere motion and promotes cardiac dysfunction. Although the molecular mechanism underlying microtubule destabilization-mediated cardiac fibrosis has been precisely unclear, the present study demonstrated that GEF-H1 mediates Rho-dependent fibrotic responses of cardiomyocytes induced by mechanical stress and cardiac fibroblasts induced by TGF-β stimulation. Although TGF-β -induced epithelial-to-mesenchymal transition of normal murine mammary gland (NMuMG) epithelial cells results in decreased stiffness and loss of normal stiffening response to force applied on integrins through TGF-β /ALK5-enhanced proteasomal degradation of LARG and GEF-H1 28 , GEF-H1 protein expression levels were not reduced in pressure-overloaded mouse hearts and TGF-β -stimulated rat cardiac fibroblasts. We cannot explain the difference of GEF-H1 stability after TGF-β stimulation between NMuMG cells and cardiac cells, but the maintenance of GEF-H1 stability can explain the mechanism why TGF-β preferentially induces cardiac fibrosis as well as cardiac stiffness.
We have identified how TRPC3-mediated local Ca 2+ influx specifically encodes signals to induce maladaptive fibrosis. TRPC3 contributes to the transition from adaptive hypertrophy to maladaptive hypertrophy, including fibrosis, induced by pressure overload in mice (Fig. 1). The inhibition of TRPC3 suppressed GEF-H1 activation induced by mechanical stretch in cardiomyocytes (Fig. 3) and TGF-β stimulation in rat cardiac fibroblasts (Fig. 4). Specifically, the TRPC3-Nox2 communication mediates mechanical stress-induced fibrosis through activation of a GEF-H1-RhoA signaling axis (Fig. 5). GEF-H1 is reportedly activated by ROS-mediated microtubule depolymerization 27 , which establishes the specificity of TRPC3/Nox2-derived ROS signaling in which microtubules are a critical component 16 . Mechanical stretch-induced GEF-H1 activation was abolished by microtubule stabilization, suggesting that microtubule-dependent GEF-H1 regulation is predominant. However, Nox2 downregulation significantly suppressed GEF-H1 activity but not completely. Recently, G α and G β subunits of trimeric G protein and PP2A phosphatase were identified as other positive regulators for GEF-H1 activation 26 . The residual activity of GEF-H1 was indeed sensitive to PP2A inhibitor, okadaic acid (Fig. 5). These results suggest that GEF-H1 activation by mechanical stretch is not solely regulated by microtubule, but other pathways such as dephosphorylation by PP2A are also important. It is unlikely that mechanical stretch activates all the GEF-H1 expressed in NRCMs. The small pool of GEF-H1 activated by mechanical stretch would be localized presumably near TRPC3-Nox2 complex. This is consistent well with our concept which TRPC3-mediated local Ca 2+ influx specifically encodes signals to induce maladaptive fibrosis via close coupling with Nox2.
Another novel finding is to demonstrate that TRPC3 can be downstream mediator of TGF-β stimulation in cardiac fibroblasts. Our aim is to investigate whether TRPC3-GEF-H1 axis can act as a common mediator of fibrotic signaling both in cardiomyocytes and cardiac fibroblasts, while there were no evidence supporting a functional relationship between TRPC3 and TGF-β s. TRPC3 was originally identified as a molecular candidate of receptor-activated cation channels that can be activated by inositol 1,4,5-trisphosphate (IP 3 ) and 1,2-diacylglycerol (DAG) produced by phophatidylinositol-dependent phospholipase C (PI-PLC) 29 , and we have previously reported that TRPC3 can be also activated by mechanical stretch in NRCMs 18,19 . However, TGF-β stimulation seems not to activate PI-PLC in cardiac fibroblasts, because we have never observed a transient increase in intracellular Ca 2+ concentration caused by IP 3 -mediated Ca 2+ release from endoplasmic reticulum using whole cell Ca 2+ imaging. In contrast, TGF-β stimulation actually increased the number of Ca 2+ spike frequency (i.e., spontaneous activity) in NRCMs, which was completely suppressed by pyrazole-3 ( Supplementary Fig. 2). This strongly suggests that TGF-β stimulation actually induces cation influx through TRPC3 channel, leading to increase in plasma membrane potential (i.e., depolarization) 9 . Although our results indirectly suggested a requirement of TRPC3-mediated Ca 2+ influx in TGF-β -stimulated ROS production and GEF-H1 activation through Nox2, future work focusing on microtubule-localized Ca 2+ signaling will be required to elucidate the molecular mechanism underlying activation of TRPC3 by TGF-β stimulation.
Fibrotic cardiac disease, emerging as loss of LV function following maladaptive tissue remodeling, is a leading cause of death worldwide. The pharmacological inhibition of TRPC3 attenuates mechanical stretch-induced fibrotic responses in human iPSC-derived cardiomyocytes and cardiac fibroblasts (Fig. 6). Since TRPC3 is ubiquitously expressed not only in the heart but also in blood vessels, kidney, liver and lung, TRPC3 could be a novel therapeutic target for the treatment of human fibrotic diseases.

Methods
Animals. All protocols using mice and rats were reviewed and approved by the ethics committees at the National Institutes of Natural Sciences or the Animal Care and Use Committee, Kyushu University, and were performed according to the institutional guidelines concerning the care and handling of experimental animals. 129 Sv mice with homozygous deletion of the gene encoding TRPC3 were provided by the Comparative Medicine Branch, National Institute of Environmental Health Sciences, Research Triangle Park, North Carolina 27709. Genotyping was performed using PCR primers; TRPC3-A 5′ -GAATCCACCTGCTTACAACCATGTG-3′ and TRPC3-B 5′ -GGTGGAGGTAACACACAGCTAAGCC-3′ . The PCR was performed using Phusion High-Fidelity DNA polymerase (Thermo Scientific). Mice were maintained in specific-pathogen-free area under a 12 h/12 h light/dark cycle. C57BL/6J mice were purchased from SLC. Sprague-Dawley rats were purchased from Kyudo or SLC.
Pressure overload study in mice. Pressure overload by TAC was performed as described previously 19,30 .
Briefly, male mice (6-8 weeks old) were anaesthetized using a mixture of domitor (Zenoaq), midazolam (Sando) and butorphanol (Meiji Seika Pharma). After orotracheal intubation and ventilation, an intercostal space was opened. The transverse aorta was then exposed and constricted between the brachiocephalic artery and left carotid artery to the width of a 27-G needle using a 5-0 silk braid. Sham treatment was performed similarly but without constriction of the silk braid. An osmotic minipump (model 2004 (Alzet)) filled with vehicle (polyethylene glycol) or pyrazole-3 (10, 30 or 100 μ g kg −1 day −1 ) was implanted intraperitoneally 19,30 . Morphological analysis. The LV functions of mice 6-week after TAC were assessed using a micronanometer catheter, and results of hemodynamic parameters were reported elsewhere 19,30 . Hearts were removed after LV pressure measurement, washed in PBS and fixed in 10% neutral buffered formalin. For quantitative assessment of collagen type I and III deposition, the hearts were embedded in paraffin, sectioned at a thickness of 3 μ m, and stained with picrosirius red using 0.1% Direct Red 80. To assess CSA of cardiomyocytes, the sections were stained with Alexa Fluor 488-conjugated wheat germ agglutinin (WGA) (Life technologies). Three regions were selected at random for each left ventricle, and the average values were calculated using a BZ-II Analyzer (Keyence).

Isolation of cardiomyocytes and cardiac fibroblasts from neonatal rats.
Rat pups were sacrificed on postnatal day 1-3, after which the left ventricles were removed and minced. The minced tissue was pre-digested in 0.05% trypsin-EDTA (Gibco) over night at 4 °C and then digested in 1 mg ml −1 collagenase type 2 (Worthington) in PBS for 30 min at 37 °C. The dissociated cells were plated in a 10-cm culture dish and incubated at 37 °C in a humidified atmosphere (5% CO 2 , 95% air) for 1 hour in DMEM containing 10% FBS and 1% penicillin and streptomycin. Attached cells were cardiac fibroblast and cultured in same medium. Floating cells were collected and plated into gelatin-coated culture dishes or laminin-coated stretch chamber dishes at a density of around 1.5 × 10 5 cells/cm 2 . After 24 h, the culture medium was changed to serum-free DMEM. For protein knockdown, cells were transfected with siRNAs (100 nM) using Lipofectamine 2000 for 72 h. Primary human cardiac fibroblasts were purchased from Lonza and cultured according to manufacturer's instruction.
iPSC-derived cardiomyocytes. The 253G1 human iPSC line was provided by the RIKEN BRC through the Project for Realization of Regenerative Medicine and the National Bio-Resource Project of the MEXT, Japan 31 . Differentiation of human iPSCs into cardiomyocytes was performed using the previously described protocol 32 . The iPSCs were cultured in mTeSR1 medium (STEMCELL Technologies) on Matrigel (BD Biosciences)-coated dishes, and a ROCK inhibitor (Y-27632; Wako, 10 μ M) was added to the cultured medium for 1 h. The iPSCs were then dissociated into single cells using Accutase (Life technologies) for 10 min at 37 °C, seeded onto Matrigel-coated dishes at density of 10 5 cells/cm 2 , and incubated for 24 h in mTeSR1 medium supplemented with 10 μ M Y-27632. The medium was then changed to mTeSR1 without Y-27632 and refreshed daily for 4 days. Thereafter, on day 0 the cells were treated with 12 μ M GSK3 inhibitor (CHIR99021; STEMGENT) in RPMI/B27-insulin medium for 24 h. The medium was then changed, and the cells were incubated in RPMI/B27-insulin for an additional 48 h (days 1-2). The cells were then treated with 5 μ M inhibitor of Wnt production-4 (IWP4; STEMGENT) in RPMI/ B27-insulin for 48 h (days 3-4), after which the medium was changed to RPMI/B27-insulin for 48 h (day 5-6). Beginning on day 7, the cells were maintained in RPMI/B27, and the medium was changed every 3 days. On day 20, the cells were dissociated into single cells using Accutase for 30 min at 37 °C and seeded onto human Laminin-211 (BioLamina)-coated stretch chamber dishes (Menicon) at a density of 10 6 cells/dish in RPMI/B27 supplemented with 10 μ M Y-27632. After 24 h, the medium was replaced with fresh RPMI for 24 h, and the cells were used for experimentation.
Measuring mRNA expression in cells and tissues. Total RNA was isolated from frozen mouse heart samples using an RNeasy Fibrous Tissue Mini Kit (Qiagen) or from cardiac cells using an RNeasy Mini Kit (Qiagen) according to the manufacturer's instructions. Quantitative real-time PCR was performed using an ABI PRISM 7500 Real-Time PCR System (Applied Biosystems) and a OneStep RT-PCR Kit (Qiagen) according to the manufacturer's instructions. All Taqman probes used were purchased from Applied Biosystems.