TGF-β1 regulates the expression and transcriptional activity of TAZ protein via a Smad3-independent, myocardin-related transcription factor-mediated mechanism

Hippo pathway transcriptional coactivators TAZ and YAP and the TGF-β1 (TGFβ) effector Smad3 regulate a common set of genes, can physically interact, and exhibit multilevel cross-talk regulating cell fate-determining and fibrogenic pathways. However, a key aspect of this cross-talk, TGFβ-mediated regulation of TAZ or YAP expression, remains uncharacterized. Here, we show that TGFβ induces robust TAZ but not YAP protein expression in both mesenchymal and epithelial cells. TAZ levels, and to a lesser extent YAP levels, also increased during experimental kidney fibrosis. Pharmacological or genetic inhibition of Smad3 did not prevent the TGFβ-induced TAZ up-regulation, indicating that this canonical pathway is dispensable. In contrast, inhibition of p38 MAPK, its downstream effector MK2 (e.g. by the clinically approved antifibrotic pirferidone), or Akt suppressed the TGFβ-induced TAZ expression. Moreover, TGFβ elevated TAZ mRNA in a p38-dependent manner. Myocardin-related transcription factor (MRTF) was a central mediator of this effect, as MRTF silencing/inhibition abolished the TGFβ-induced TAZ expression. MRTF overexpression drove the TAZ promoter in a CC(A/T-rich)6GG (CArG) box-dependent manner and induced TAZ protein expression. TGFβ did not act by promoting nuclear MRTF translocation; instead, it triggered p38- and MK2-mediated, Nox4-promoted MRTF phosphorylation and activation. Functionally, higher TAZ levels increased TAZ/TEAD-dependent transcription and primed cells for enhanced TAZ activity upon a second stimulus (i.e. sphingosine 1-phosphate) that induced nuclear TAZ translocation. In conclusion, our results uncover an important aspect of the cross-talk between TGFβ and Hippo signaling, showing that TGFβ induces TAZ via a Smad3-independent, p38- and MRTF-mediated and yet MRTF translocation-independent mechanism.

Yes-associated protein (YAP) 2 and its paralog, transcriptional coactivator with a PDZ-binding domain (TAZ), are central effectors of the Hippo pathway and play essential roles in the control of organ size, proliferation (contact inhibition), stemness, differentiation, cellular plasticity (e.g. epithelial-mesenchymal transition (EMT)), regeneration, and the mechanosensitive regulation of gene expression (1)(2)(3)(4). Congruent with these cell fate-determining functions, YAP and TAZ have emerged as key mediators of major disease entities, particularly cancer (5,6) and, as recent studies by us (7)(8)(9)(10) and others (11)(12)(13)(14) reveal, organ fibrosis. YAP and TAZ are primarily regulated at the level of their nucleocytoplasmic traffic. Under resting conditions (e.g. in contact-inhibited cells) the constitutive activity of Hippo kinases, Mst1/2, and their downstream targets, Lats1/2, keep YAP and TAZ in a phosphorylated state thereby ensuring their cytosolic retention via binding to sequestering proteins (e.g. 14-3-3) (15). Upon Hippo kinase inhibition, YAP and TAZ get dephosphorylated and translocate to the nucleus, where they bind to cognate transcription factors (TFs), predominantly to members of the TEAD family, and drive a large set of genes involved in the above-mentioned functions (16,17). Another major input regulating YAP/TAZ nuclear accumulation is the state of the cytoskeleton; actin polymerization accompanied by myosin phosphorylation (e.g. as a result of Rho activation) leads to nuclear YAP/TAZ translocation through partially Hippo-independent, incompletely understood mechanisms (18 -20). Through this cytoskeletal pathway mechanical cues, such as cell contractility, stretch, or extracellular matrix stiffness, impact YAP/TAZ distribution, thereby initiating mechanoresponsive gene transcription (4,21).
Although nucleocytoplasmic shuttling of YAP and TAZ is a central aspect of their regulation, mounting evidence indicates that the expression of YAP and/or TAZ also exhibits significant changes under various conditions. In fact, increased YAP or TAZ levels are not only characteristic of a wide range of cancers, but they are often negative prognostic factors, likely due to their contribution to proliferation and metastasis (5,6). Recently, diabetic nephropathy, a fibrogenic state, has also been associated with increased YAP expression (22). Despite the potentially crucial importance of changes in net YAP and/or TAZ levels, and the demonstration of the involvement of some TFs in this process (see under "Discussion"), the regulation of YAP and TAZ expression (transcription), the relevant stimuli, and the underlying mechanisms remain largely unexplored.
YAP/TAZ signaling exhibits extensive cross-talk with other pathways; a chief example is transforming growth factor ␤1 (TGF␤) signaling (23)(24)(25). This pleiotropic cytokine, which upon binding to its receptors triggers both "canonical," i.e. Smad2/3-dependent, and non-canonical signaling, is the main inducer of EMT and fibrogenesis (26 -28). Because TGF␤ also regulates cancer cell proliferation (29,30), its cardinal effects show a strong functional overlap with those of YAP/TAZ. The molecular underpinning of the cross-talk between TGF␤ and YAP/TAZ signaling is at least 2-fold. First, YAP and TAZ were shown to bind Smad2 and -3, and nuclear YAP/TAZ were proposed to act as retention factors for Smads (31,32). Second, a multitude of genes harbors both Smad-binding elements (SBEs) and TEAD-binding elements in their promoters (33). Binding of the Smad3-TAZ or YAP-TEAD complexes to one or both of these cis-elements has been shown to exert synergistic (or in certain cases antagonistic) transcriptional effects in a promoter-dependent fashion (33,34). Considering fibrogenic gene expression, we have recently shown that TAZ confers Smad3 sensitivity to the promoter of ␣-smooth muscle actin (SMA), the hallmark of the myofibroblast (9). Interestingly, an impact of TGF␤ on TAZ expression has also been noted in a few studies, including our own (7,31,35). However, despite the potential key importance of TGF␤-induced changes in TAZ expression in the TGF␤/TAZ cross-talk, the underlying signaling mechanisms (canonical versus non-canonical), the relevant transcription factors, their mode of regulation, and the functional significance of this phenomenon have not been elucidated. Therefore, we set out to characterize if and, if so, how TGF␤ impacts TAZ (or YAP) expression, and whether this manifests in altered TAZ activity.
To address this issue, we built on our recent study wherein we discovered that myocardin-related transcription factor (MRTF) regulates TAZ at multiple levels (9). MRTF is a transcriptional coactivator of serum-response factor (SRF), which stimulates gene expression via CC(A/T) rich GG elements, the so-called CArG box (36). MRTF is primarily regulated by its nuclear uptake; binding of monomeric (G) actin to MRTF prevents its nuclear accumulation (37). Upon actin polymerization, G-actin dissociates from MRTF allowing its nuclear entry. We previously identified a CArG box in the TAZ promoter and showed that MRTF silencing reduces basal TAZ expression (9). We also demonstrated that MRTF can bind TAZ. This direct interaction inhibits both MRTF and TAZ activity on the SMA promoter (but is synergistic on others), and it is disrupted by TGF␤, which redistributes TAZ from MRTF to Smad3 (9). Cognizant of this scenario, we asked if MRTF might also be mediating the effect of TGF␤ on TAZ expression and, if so, whether this occurs via Smad3-dependent or non-canonical pathways. We also asked whether MRTF acts in a "classic" translocation-dependent fashion or via alternative mechanisms. Here, we show that TGF␤ induces robust TAZ expression by activating MRTF in a p38 MAPK (p38)-and redox-dependent, translocation-independent manner and that this phenomenon primes cells for exaggerated TAZ activation in response to stimuli inducing TAZ translocation.

TAZ protein expression increases in various cell types upon TGF␤ treatment and in an experimental model of fibrosis
To test whether TGF␤ impacts TAZ protein expression, we exposed C3H/10T1/2 cells (a pericyte-like fibroblast line) to this cytokine for varying times. This line was chosen because both fibroblasts and pericytes have been implicated as major drivers of fibrogenesis (38,39). TGF␤ caused a significant increase in TAZ protein, which was detectable as soon as 3 h and rose to Ϸ6-fold above the resting level at 6 h. TAZ levels then slowly declined but remained severalfold higher than the basal level even after 48 h of stimulation (Fig. 1A). Remarkably, this response was specific for TAZ, as YAP showed no change or only marginal change throughout the entire time course (Fig.  1A). The effect of TGF␤ on TAZ expression was prevented by SB-431542, indicating that TGF␤ receptor 1 kinase is required for this effect (Fig. 1B). TGF␤ induced a robust increase in the TAZ protein level in normal rat kidney fibroblasts (NRKFs) as well (Fig. 1C). To test whether a similar response was present also in epithelial cells, LLC-PK1 cells were exposed to TGF␤ either under normal conditions or under low calcium condition, wherein intercellular contacts are disassembled. This dual stimulation regimen (two-hit scheme) was used as our earlier studies have indicated that cell contact injury (uncoupling) synergizes with TGF␤ to cause epithelial-myofibroblast transition (hallmarked by SMA expression) (40 -42). Although TGF␤ alone is insufficient to induce mesenchymal transformation in intact epithelial cells (40 -42), it was capable of enhancing TAZ protein expression, and this effect was further potentiated by low calcium (Fig. 1D). Next, we asked whether TAZ levels would also change under pathological conditions in which TGF␤ is known to play a key role. To this end, we probed whole-kidney lysates obtained from mice, which were either sham-operated or underwent unilateral ureteral obstruction (UUO). This procedure provokes a robust renal fibrosis which, as our recent study shows, is mitigated by YAP/TAZ inhibition (8). UUO provoked a substantial increase in TAZ protein expression, which was seen as early as 3 days after UUO and increased further by day 7 (Fig. 1E). Albeit less pronounced than TAZ expression, YAP expression also increased in most samples. Taken together, these results indicate that TGF␤ induces robust TAZ protein expression in fibroblast-type and epithelial cells and in an in vivo model of fibrosis.

TGF␤ induces p38-dependent, MRTF-mediated TAZ expression TGF␤-induced TAZ expression required non-canonical (Smad3-independent) pathways
TGF␤ activates both Smad2/3-dependent (canonical) and a variety of Smad2/3-independent (non-canonical) signaling pathways (28). To assess the role of the canonical signaling in TGF␤-induced TAZ expression, we applied two approaches. First, we treated C3H/10T1/2 cells with the Smad3 inhibitor SIS3. This failed to prevent the effect of TGF␤ on TAZ protein expression ( Fig. 2A). To verify that SIS3 was effective under these conditions, cells were transfected with a Smad3 reporter luciferase construct (harboring tandem SBE) and exposed to TGF␤ in the absence or presence of the inhibitor (Fig. 2C). SIS3 completely abolished the TGF␤-induced rise in luciferase activity, verifying effective prevention of Smad3-dependent transcription. Accordingly, SIS3 also prevented the TGF␤-induced PAI-1 up-regulation ( Fig. 2A). To substantiate these pharmacological observations, cells were transfected with control (non-related, NR) or Smad3-specific siRNAs for 24 h prior to stimulation (Fig. 2B). Although the Smad3 siRNA reduced Smad3 expression to 21 Ϯ 4% (n ϭ 3, data not shown) and inhibited SBE-luciferase activity to a similar extent (Fig. 2D), it failed to reduce the ensuing Figure 1. TGF␤ induces a rapid increase in TAZ protein expression both in mesenchymal and epithelial cells. TAZ levels also increase in experimental kidney fibrosis. A, C3H/10T1/2 cells were treated with 5 ng/ml TGF␤ for the indicated times, and then TAZ and YAP protein levels were determined by Western blotting and densitometry (n ϭ 5). Data are normalized to ␤-actin as loading control. Similar results were obtained using tubulin or GAPDH as loading controls. B, C3H/10T1/2 cells were treated with vehicle (DMSO) or ALK5 inhibitor SB431542 (10 M, 30-min preincubation) and then exposed to TGF␤ for 6 h. Changes in TAZ and YAP expression were detected as in A. C and D, NRKF (C) and LLC-PK1 kidney tubular cells (D) were incubated without or with TGF␤ for 48 h and then processed for Western blotting for the indicated proteins. E, whole-kidney lysates were prepared from mice 1, 3, and 7 days after sham operation or UUO and probed for the indicated proteins. Representative blots are shown from four similar experiments. Densitometric quantitation of TAZ expression normalized to GAPDH is shown below the blot. Errors bars represent standard deviation.
Smad2 is also activated by TGF␤, and in a small subset of effects Smad2 and Smad3 play selective roles (43,44). Therefore, we also tested the impact of Smad2 by using Smad2 siRNA. Interestingly, Smad2 silencing led to a dramatic decrease in the basal (TGF␤-independent) expression of YAP and also reduced basal TAZ expression (Fig. 2F). TGF␤ had no effect on YAP levels in the Smad2 silenced cells (similar to the controls) but was still able to increase TAZ expression. Nevertheless, at least partly due to the lower basal level, the attained TAZ expression was reduced (Fig. 2F). The fact that Smad2 (as opposed to Smad3) silencing affected YAP levels, and that this effect was independent of TGF␤, pointed to a fundamentally different role and mechanism of Smad2 and -3, wherein the former is indispensable for basal YAP expression and may play a permissive role in maintaining TAZ levels and regulation as well (see also "Discussion").
To investigate the involvement of major non-canonical pathways, we first used their respective pharmacological inhibitors. and then exposed to TGF␤ for 6 h. TAZ expression was determined by Western blotting and quantified by densitometry (graph below the blot). TAZ expression in the TGF␤-treated samples was taken as unity (n ϭ 3). Membranes were also developed for plasminogen activator inhibitor (PAI-1), a direct transcriptional target of Smad3. B, C3H/10T1/2 cells were transfected with 50 nM non-related (siNR) or Smad3-specific siRNA (siSmad3) for 24 h and then, where indicated, exposed to TGF␤ for 6 h and processed as in A. TAZ expression (graph below the blot) was quantified by densitometry (n ϭ 3). C and D, to verify the efficiency of SIS3 (C) and the siRNA (D) on Smad3 activity, luciferase assays were performed using an SBE firefly luciferase reporter construct (SBE-Luc). C3H/10T1/2 cells were cotransfected with SBE-Luc and Renilla luciferase (for normalization), 24 h later serum-starved (2 h), and then incubated in the presence or absence of TGF␤ for 6 h. E, LLC-PK1 cells were transfected with siNR or siSmad3 (50 nM, 24 h), and then exposed to TGF␤ for 24 h. TAZ expression was quantified as in A. F, C3H/10T1/2 cells were transfected with siNR or siSmad2 (50 nM, 24 h), and then exposed to TGF␤ for 6 h. Changes in TAZ expression were determined through Western blotting. n.s., non-significant.

TGF␤ induces p38-dependent, MRTF-mediated TAZ expression
Thus, we determined the TGF␤-induced change in TAZ expression in the absence (vehicle controls, DMSO) or presence of the selective p38 MAPK inhibitor, doramapimod (DORA), the Akt inhibitor AKTi-1/2 (AKTi), the MEK/ERK1/2 inhibitor U0126, and the Rho kinase inhibitor Y-27632. The corresponding Western blottings and their quantifications show that DORA (Fig. 3A) and AKTi (Fig. 3B) significantly suppressed the TGF␤-induced TAZ response, whereas U0126 (Fig. 3C) and Y-27632 (Fig. 3D) did not exert significant effects. Efficiency of the latter two drugs (on ERK phosphorylation and the actin skeleton) was verified in separate experiments (data not shown). These results suggest that p38 and Akt pathways significantly contribute to the mediation of the TGF␤-induced Smad3-independent induction of TAZ expression.

Inhibition of TAZ degradation cannot account for the TGF␤-induced rise in TAZ expression
Several kinase pathways have been reported to induce the proteolysis of TAZ through phosphorylation of a phosphodegron site (45,46). Because some of these (e.g. GSK-3␤) are inhibited by TGF␤ (47), it was conceivable that TGF␤ elevates TAZ levels by suppressing TAZ degradation. To assess this possibility, we first tested the impact of proteasome inhibition on TAZ expression in the relevant time period (Fig. 4A). The proteasome inhibitor MG132 caused a strong increase in steadystate TAZ protein expression within 3-6 h, an effect whose time dependence and magnitude were comparable with that of TGF␤ (Fig. 4B) and therefore compatible with a potential role of TGF␤-inhibited TAZ proteolysis. TGF␤ and MG132 added together appeared to exert a slightly stronger effect than MG132 alone (Fig. 4B). Thus, we tested the half-life of TAZ upon blocking protein synthesis by cycloheximide (CHX) in the absence or presence of TGF␤ (Fig. 4C). CHX induced a rapid reduction in TAZ levels without causing significant changes in YAP or actin levels, verifying that TAZ is a fast turnover protein (45) with a half-life of ϳ2 h in C3H/10T1/2 cells. However, the drop in TAZ levels upon CHX treatment was similar in the absence and presence of TGF␤ (Fig. 4C). We considered that CHX might have reduced the expression of a protein necessary to inhibit TAZ proteolysis. To address this possibility, we expressed HA-tagged TAZ in C3H/10T1/2 cells for various times in the absence and presence of TGF␤ (Fig. 4D). This approach has the following two advantages: it allowed us to test the potential TAZ-stabilizing effect of TGF␤ in the absence of CHX, and it eliminated any transcriptional effects via the TAZ promoter, because HA-TAZ expression was driven by an artificial CMV-based promoter. Under these conditions, we did not observe a significant increase in TAZ protein expression in TGF␤-treated cells at any of the investigated time points. Although these experiments do not fully exclude some contribution of altered degradation, they strongly suggest that other (e.g. transcriptional) mechanisms play a central role.

MRTF plays a key role in TGF␤-induced TAZ mRNA and protein expression
We next tested TAZ mRNA expression and found that TGF␤ induced a significant rise in the TAZ message in C3H/10T1/2 mesenchymal cells (Fig. 5A). Interestingly, the increase in TAZ mRNA was prevented by DORA but not by AKTi, suggesting that p38 but not Akt promotes TAZ expression by mediating TGF␤-induced TAZ transcription (Fig. 5B). Because Akt is an important regulator of protein translation (e.g. via the mTOR/S6 kinase pathway (48)), we surmised that Akt might contribute to TAZ expression through this mechanism. Given that TAZ is a very fast turnover protein, suppression of translation is expected to impact TAZ levels to a greater extent (i.e. more rapidly) than more stable proteins. Indeed, AKTi significantly reduced TGF␤-induced S6 protein phosphorylation (Fig. 5C), whereas the mTOR inhibitor rapamycin suppressed the TGF␤-induced TAZ expression (Fig.  5D). These findings are consistent with the notion that inhibition of TAZ translation, at least in part, accounts for the effect of AKTi.
We next focused on the mechanism underlying TGF␤-induced changes in transcription. Because our recent studies implicated the transcriptional coactivator MRTF in basal TAZ expression, and the TAZ promoter harbors a CArG box, we asked whether MRTF might be a mediator of TGF␤-induced for 30 min and, where indicated, exposed to TGF␤ for 6 h. TAZ expression was quantified by densitometry using the TGF␤-treated sample as unity. The inhibitors used were as follows: A, DORA for p38 (n ϭ 4); B, AKTi for Akt1/2 (n ϭ 3); C, U0126 for MEK/ERK1/2 (n ϭ 7); and D, Y27632 for ROK (n ϭ 10). ns, non-significant.

TGF␤ induces p38-dependent, MRTF-mediated TAZ expression
TAZ transcription. To this end we treated the cells with CCG-1423, an inhibitor of MRTF/SRF-dependent transcription (49). The drug nearly abolished the stimulatory effect of TGF␤ on TAZ mRNA expression (Fig. 5B). Similar observations were made in LLC-PK1 tubular cells, in which TGF␤ induced a significant rise in TAZ mRNA, which was nearly completely prevented by CCG-1423 (Fig. 5E).
To substantiate these findings, we next determined the TGF␤-induced TAZ protein expression upon pharmacological or genetic inhibition of MRTF. CCG-1423 dramatically reduced the TGF␤-provoked rise in TAZ protein level both in C3H/10T1/2 and LLC-PK1 cells (Fig. 6, A and C). Moreover, siRNA-induced silencing of MRTF-A and -B fully abolished TAZ induction (Fig. 6B). It is noteworthy that MRTF siRNA not only prevented the TGF␤-induced rise but also suppressed basal TAZ expression (Fig. 6B).

TGF␤ drives TAZ expression by inducing post-translational modification rather than inducing a net increase in nuclear translocation of MRTF
Activation of MRTF-dependent transcription is, in most cases, due to the nuclear translocation of MRTF. However, previous reports have indicated that TGF␤ is a weak/marginal inducer of MRTF translocation in C3H/10T1/2 cells (50) and our previous studies led to a similar conclusion in LLC-PK1 cells as well (41,42). Nonetheless, we tested the impact of TGF␤ on the localization of both endogenous and GFP-tagged MRTF C3H/10T1/2 cells. Under resting conditions, endogenous . Altered TAZ degradation cannot account for the TGF␤-induced increase in TAZ expression. A, to assess the role of the proteasome in regulating steady-state TAZ levels, C3H/10T1/2 cells were treated with the proteasome inhibitor MG132 (40 M) for the indicated times, and changes in TAZ expression were determined by Western blotting. B and C, to assess whether TGF␤ prevents the constitutive degradation of TAZ, cells were treated with MG132 (B) or the protein synthesis inhibitor CHX (25 g/ml) (C) in the absence or presence of TGF␤. B, cells were treated with vehicle (DMSO) or MG132 (40 M) and, where indicated, exposed to TGF␤ for 6 h. C, cell lysates were collected at the allotted times and change in TAZ protein was quantified relative to the 0-h conditions. The graph shows the respective degradation profiles for both conditions (n ϭ 3). D, cells were transfected with a HA-TAZ construct (0.5 g/ml) and 24 h later treated with or without TGF␤ for the indicated times. HA-TAZ expression was monitored using an anti-HA antibody.

TGF␤ induces p38-dependent, MRTF-mediated TAZ expression
MRTF resided both in the cytosol and the nucleus (Fig. 7A). TGF␤ caused no change or only marginal changes in nuclear localization both after 0.5 and 3 h of treatment (i.e. when the TAZ transcription was already markedly increased), and only a slight increase after 6 h, as verified by immunofluorescence microscopy and the quantitation of the cytoplasmic to nuclear ratio (Fig. 7A). This finding was not due to a general unresponsiveness of MRTF localization in C3H/10T1/2 cells, as serum provoked rapid and sizable MRTF accumulation in the nucleus. Because non-specific (background) antibody binding might mask small but observable changes in MRTF distribution, we monitored MRTF localization in an antibody-independent manner as well. We transfected cells with GFP-MRTF and determined the percentages of cells showing cytosolic or nuclear GFP-MRTF distribution. As shown by Fig. 7B, 90% of resting cells exhibited cytosolic or even GFP-MRTF localization, and this distribution was unaltered upon TGF␤ treatment. In contrast, serum caused a 5-fold increase in the percentage of cells showing nuclear GFP-MRTF accumulation. Finally, these results were confirmed by determining MRTF protein levels in cytoplasmic and nuclear preparations (Fig. 7C). Some MRTF resided in the nucleus in control cells, and there was no measurable increase in the nuclear MRTF level upon TGF␤ treatment (Fig. 7C). However, both the cytosolic and the nuclear MRTF fraction exhibited an upward shift in the apparent molecular mass of MRTF (Fig. 7C), a characteristic sign of phosphorylation (51). MRTF has recently been shown to contain 24 serine/threonine phosphorylation sites, the collective mutation of which (to alanine) inhibited MRTF activity, even when MRTF was rendered constitutively nuclear (see under "Discussion") (52). We therefore tested whether such post-translational modifications, even without a net increase in the nuclear MRTF content, could account for increased MRTF activity, leading to increased TAZ expression. To address this issue, first we checked whether the transcriptional activity of MRTF was indeed increased by TGF␤ (despite the lack of translocation) under our conditions. Cells were transfected with 3DA-Luc, a reporter construct in which three CArG boxes precede the coding region of luciferase. TGF␤ readily activated 3DA-Luc, and importantly, both the basal activity and the TGF␤-induced acti- Figure 5. TGF␤ increases TAZ mRNA via a p38-and SRF/MRTF inhibitorsensitive mechanism. A, C3H/10T1/2 cells were left untreated or stimulated with TGF␤ for 6 h. qPCR was performed, and TAZ mRNA was normalized to GAPDH mRNA. Fold change was calculated using the Pfaffl method. Cntrl, control. B, C3H/10T1/2 cells were pretreated with vehicle (DMSO) or pharmacological inhibitors DORA (10 M), AKTi (10 M), or CCG (3 M) for 30 min and then exposed to TGF␤ for 3 h. qPCR was performed, and TAZ mRNA was normalized to an alternative housekeeping gene, RLP13a mRNA. C, immunofluorescence quantitation of S6 protein phosphorylation. Cells were pretreated with vehicle (DMSO) or AKTi (10 M, AKTi1/2) for 30 min and, where indicated, exposed to TGF␤ for 6 h. S6 phosphorylation was determined by immunofluorescence staining and automated image analysis as described under "Experimental procedures" (n ϭ 3 experiments in each of which 1000 -3000 cells/condition were quantified). D, cells were pretreated with vehicle (DMSO) or rapamycin (100 nM, RAPA) for 30 min and, where indicated, exposed to TGF␤ for 6 h. Changes in TAZ expression was determined by Western blotting. E, LLC-PK1 cells were pretreated with DMSO or CCG, then exposed to TGF␤ for 24 h, and processed for qPCR to determine TAZ mRNA as in A (n ϭ 3). ns, non-significant.

TGF␤ induces p38-dependent, MRTF-mediated TAZ expression
vation were completely abolished by siRNA-mediated downregulation of MRTF (Fig. 7D). These findings indicated that TGF␤ activates MRTF-dependent transcription without inducing robust MRTF translocation. Finally, to check that MRTF/SRF signaling can indeed regulate the endogenous TAZ promoter, we performed ChIP experiments (Fig. 7E). Indirect immunoprecipitation using anti-MRTF antibody resulted in a 3.5-fold increase in the associated TAZ promoter region over the isotype control, and this was further increased by TGF␤ (despite the fact that the MRTF antibody is suboptimal for ChIP). Direct immunoprecipitation with anti-SRF indicated strong basal binding, which was substantially elevated by TGF␤. These experiments verified that MRTF/SRF interacts with the endogenous TAZ promoter. The graph shows the quantification of subcellular distribution (nuclear, cytoplasmic, or diffuse) of GFP-MRTF, using Ͼ100 cells in each category at each time point. Scale bar, 20 m. C, MRTF localization was determined using nuclear preparations. C3H/10T1/2 cells were incubated without or with TGF␤ for 3 h. Cytoplasmic and nuclear fractions were then separated (see under "Experimental procedures") and probed for MRTF, tubulin as a cytosolic marker, and histone H3 as a nuclear marker. D, to determine the effect of TGF␤ on MRTF activity, luciferase assays were performed using a CArG box containing firefly luciferase reporter construct (3DA). Cells were cotransfected with 3DA and Renilla luciferase along with NR or MRTF-A/B siRNA (100/200 nM, respectively, 48 h). Cells were then incubated in the presence or absence of TGF␤ for 6 h, and 3DA-Luc activity was measured and normalized to that of Renilla. E, for chromatin immunoprecipitation, C3H/10T1/2 cells were treated with vehicle (DMSO) or TGF␤ for 3 h. Chromatin-protein complexes were then cross-linked, sheared, and immunoprecipitated (IP) using either anti-MRTF or anti-SRF antibodies as described under "Experimental procedures." qPCR was performed to quantify the coprecipitating TAZ promoter, using specific primers flanking the CArG box. Data are normalized to the signal obtained by control (unrelated) IgG.

TGF␤ induces p38-dependent, MRTF-mediated TAZ expression
Next, we investigated whether the kinase inhibitors that mitigated TAZ expression impacted the post-translational modification of MRTF. The TGF␤-induced shift was reduced by 50% by the p38 inhibitor DORA (Fig. 8A), whereas AKTi did not significantly alter the shift (Fig. 8B). To substantiate the role of p38, we used siRNA as well. Combined targeting of p38 isoforms significantly reduced both the TGF␤-induced TAZ expression and the molecular weight shift in MRTF (Fig. 8, C and D). Accordingly, DORA but not AKTi significantly suppressed the TGF␤-dependent activation of 3DA-Luc (Fig. 8E,  panels i and ii). Of note, Smad3 silencing not only failed to prevent but in fact potentiated the basal activity and TGF␤induced activation of 3DA-Luc (Fig. 8E, panel iii). The latter data are in agreement with our earlier report showing that Smad3 is an inhibitor of MRTF (42). Together, these findings indicate that both TAZ mRNA and reporter 3DA-Luc activity and, where indicated, exposed to TGF␤ for 6 h. Whole-cell extracts were run in 6% SDS gels, and the relative shift is the molecular mass of MRTF (see dotted lines) was determined and normalized to the change detected in the vehicle-treated controls as described under "Experimental procedures." C and D, cells were transfected with 300 nM non-related (siNR) or ␣ (100 nM)-, ␤ (100 nM)-, and ␥ (100 nM)-p38 siRNA (sip38) for 48 h and then, where indicated, exposed to TGF␤ for 6 h. Cortactin was used as loading control. TAZ expression (C) and the relative molecular shift of MRTF (D) were assessed by Western blotting (n ϭ 3) as in Figs. 1A and 8B, respectively. ␣-Actinin (␣-act) was used as loading control. E, cells were cotransfected with 3DA firefly reporter and Renilla luciferase and, where indicated (panel iii) with Smad3 siRNA or NR siRNA for 24 h. Subsequently cells were preincubated with vehicle (DMSO) or DORA (panel i) or AKTi (panel ii) for 30 min. Cells were then exposed to TGF␤ for 6 h, as indicated. Normalized luciferase activities were determined (n ϭ 3 for each condition). F and G, C3H/10T1/2 cells were pretreated with vehicle or the MK2 inhibitor PF-3644022 (PF, 10 M) followed by 6 h of treatment with TGF␤ as indicated and then processed for Western blotting to assess TAZ expression (F) or the shift of MRTF (G) (n ϭ 3). H and I, conditions were as in F and G, except the cells were pretreated where indicated with the antifibrotic drug, pirfenidone (Pirf, 1 mg/ml). n.s., non-significant.
A recent elegant study has shown that MRTF can be directly phosphorylated both by p38 and its major downstream substrate MK2 (53). We therefore asked whether MK2 might be involved in TGF␤-induced TAZ expression. To assess this, cells were pretreated with PF 3644022 (PF), a potent MK2 inhibitor (54). PF significantly reduced the TGF␤-induced TAZ expression (Fig. 8F) and also mitigated the TGF␤-induced shift in the apparent molecular weight of MRTF (Fig. 8G). Finally, we tested the impact of pirfenidone, the first clinically approved drug against pulmonary fibrosis (55). Although its mechanism of action remains incompletely defined, pirfenidone was proposed to interfere with the p38/MK2 axis (56). Pirfenidone markedly suppressed both the TGF␤-induced TAZ expression (Fig. 8H) and the concomitant MRTF phosphorylation (Fig. 8I). Taken together, these experiments suggest that TGF␤ induces MRTF phosphorylation partially via p38 and possibly MK2, which is in turn critical for TGF␤-induced MRTF translocation-independent activation of MRTF.

MRTF is sufficient to drive the TAZ promoter and increase TAZ protein expression
Although the data so far suggested that MRTF is indispensable for TAZ expression, we sought to establish whether TGF␤ can indeed activate the TAZ promoter and whether MRTF and/or SRF is sufficient to induce TAZ protein expression. For this purpose, we cloned a 1200-bp segment of the human TAZ promoter and inserted it into pGL3 luciferase vector (TAZ-Luc). Cells were then transfected with TAZ-Luc and exposed to vehicle or TGF␤. As shown in Fig. 9A, TGF␤ caused a 2.5-fold increase in TAZ-Luc activity. Moreover, DORA prevented the TGF␤-triggered activation of the TAZ promoter (Fig. 9A). We then overexpressed either MRTF or SRF along with the TAZ-Luc reporter and found that MRTF but not SRF drove the TAZ promoter (Fig. 9B). In accordance with this finding, overexpression of MRTF but not SRF was capable of inducing TAZ protein expression; cells transfected with HA-MRTF but not HA-SRF showed enhanced staining for endogenous TAZ protein, compared with their non-transfected neighbors (Fig. 9C). Of note, endogenous SRF is abundantly expressed and largely nuclear, likely accounting for the fact that further SRF expression, without concomitant activation, failed to evoke TAZ expression. Finally, to test whether the observed transcriptional effect of MRTF was indeed mediated directly via the CArG box within the TAZ promoter, we mutated and thereby inactivated this sequence (TAZ mut -Luc). In contrast to TAZ-Luc (Fig. 9A), TAZ mut -Luc was not activated by TGF␤ (in fact it was significantly inhibited by the cytokine) (Fig. 9D). MRTF overexpression also failed to stimulate this promoter construct (Fig. 9E). Taken together, MRTF is sufficient to drive the TAZ promoter activation and protein expression in a CArG-box-dependent manner.

NADPH oxidase Nox4 is a mediator of TGF␤-induced, MRTF-mediated expression and transcriptional activity of TAZ
TGF␤ is a prime inducer of Nox4 (57-59), and this enzyme has recently been implicated as a potential regulator of MRTF phosphorylation via a redox-dependent mechanism (51). Furthermore, Nox4 has been shown to activate p38 (59,60). Therefore, we asked whether Nox4 might be involved in the regulation of TGF␤-induced TAZ expression and TAZ promoter activation. To test this, we preincubated the cells with VAS2870 (VAS), a potent inhibitor of Nox4. Remarkably, VAS reduced the TGF␤-triggered TAZ expression by 60% (Fig 10A) and, concomitantly, mitigated the TGF␤-induced shift in the molecular mass of MRTF by 50% (Fig 10B). To assess whether the inhibition of TAZ protein expression by VAS could be attributed to an effect on the TAZ promoter, we transfected the cells with TAZ-Luc and stimulated them with TGF␤ in the absence or presence of VAS. The inhibitor caused a 50% reduction in the TGF␤-evoked TAZ-Luc activity, reflecting a strong suppression of the relative (fold) change compared with the TGF␤ effect measured in the vehicle-treated controls ( Fig 10C). Together, these data imply that Nox4 plays an important role in the TGF␤-induced TAZ promoter activation and protein expression, predominantly as an upstream activator of MRTF.

Functional significance of increased TAZ expression: Priming and synergy with TAZ-translocating inputs
Although our experiments so far document that TGF␤ induces TAZ expression, and allows insight into the underlying mechanisms, they do not inform us about the potential functional significance of this phenomenon. Given that TAZ, as a transcription factor, acts in the nucleus, we surmised that increased overall TAZ expression might influence TAZ-dependent transcription by two ways. First, having more TAZ could proportionally increase the nuclear presence and thus the transcriptional activity of TAZ, even without a change in the nuclear/cytoplasmic ratio. Second, a larger TAZ pool might prime the cells for enhanced TAZ-mediated transcription once a second stimulus, promoting TAZ nuclear translocation, challenges the cell. One difficulty in conclusively addressing this problem using natural TAZ target genes is that a large number of these harbor both TEAD-binding elements and CArG boxes (61), as well as Smad-binding elements (33) in their promoter. Moreover, studies from us and others have shown that, in addition to the individual effect of these factors, they may negatively or positively influence each other's action at the promoter, depending on the distance of their respective binding sites (9,62). To overcome these complexities, we used a TEAD reporter construct (TEAD-Luc), which contains exclusively TEADbinding elements. As shown in Fig 11A, TGF␤ stimulated TEAD-Luc. Importantly, siRNA-mediated down-regulation of TAZ strongly (80%) inhibited the activation of TEAD-Luc. These findings imply that TGF␤, without any additional stimulus, is capable of activating TEAD-mediated transcription in a TAZ-dependent manner. We next assessed whether TGF␤-induced TAZ up-regulation can prime the cell for a TAZ-mobilizing stimulus. Sphingosine 1-phosphate (S1P) has been shown to promote nuclear localization of TAZ predominantly by inhibiting Hippo kinases (63). Indeed, S1P promoted the nuclear translocation of TAZ (Fig 11B) without affecting net TAZ expression (Fig 11C). In contrast, TGF␤ increased TAZ protein expression (Fig 11C), which occurred roughly propor-

TGF␤ induces p38-dependent, MRTF-mediated TAZ expression
tionally in the cytosol and nucleus (Fig 11B). Importantly, TGF␤ treatment (to induce enhanced expression) followed by S1P (to promote nuclear translocation) resulted in much stronger nuclear accumulation of TAZ than either stimulus alone. Accordingly, TGF␤ and S1P showed strong synergy on the TEAD reporter (Fig 11D). Thus, although TGF␤ alone had a significant but modest effect (2.5-fold), and S1P alone only marginally elevated TEAD-Luc activity at 24 h, the two stimuli together exerted a much more than additive effect ( Fig  11D). Of note, TGF␤-induced TAZ expression is not modulated by S1P (Fig 11C). Collectively, these results imply that TGF␤ efficiently primes for enhanced TAZ transcriptional activity provoked by a TAZ-translocating stimulus, and conversely, the TAZ-mediated transcriptional effect of TGF␤ is strongly potentiated by stimuli causing concomitant TAZ translocation.

Discussion
Cross-talk between TGF␤-induced and YAP/TAZ-mediated signaling has been shown to play a key role in cell plasticity and the pathogenesis of cancer and organ fibrosis (8,23,31,33). Our studies uncover a new aspect of this important interplay; we show that TGF␤ is a strong and fast inducer of TAZ (but not Figure 9. TGF␤ or overexpression of MRTF but not SRF can induce TAZ protein expression and TAZ promoter activation. A, C3H/10T1/2 cells were cotransfected with a Ϸ 1.2 kb segment of the TAZ promoter coupled to firefly luciferase (TAZ-Luc) along with Renilla luciferase for 24 h. Subsequently cells were pretreated with vehicle (DMSO) or DORA (10 M) for 30 min, then stimulated with TGF␤ for 24 h. Normalized TAZ promoter activity was determined by the dual-luciferase assay. B, to assess whether MRTF or SRF was sufficient to drive the TAZ promoter, C3H/10T1/2 cells were cotransfected with TAZ-Luc and Renilla along with either FLAG-MRTF or HA-SRF. Twenty four h later TAZ-Luc activity was measured. C, verification of expression of FLAG-MRTF and HA-SRF and assessment of their impact on TAZ expression was achieved by double immunofluorescence staining for the corresponding epitope tag and endogenous TAZ. Note that HA-SRF is strongly expressed and exhibits robust nuclear localization, yet it does not cause a detectable increase in TAZ expression, as opposed to FLAG-MRTF. Scale bar, 20 m. D, C3H/10T1/2 cells were transfected with a mutant version (TAZ mut -Luc) of TAZ-Luc, in which the CArG box has been inactivated, and with Renilla. Twenty four h later cells were left untreated or treated with TGF␤ for an additional 24 h, and luciferase activities were determined (n ϭ 3). E, conditions were as in D, except cells were transfected with FLAG-MRTF-B along with the TAZ mut -Luc/Renilla system for 24 h, and luciferase activities were measured without further treatment (n ϭ 3). Cntrl, control. n.s., non-significant.

TGF␤ induces p38-dependent, MRTF-mediated TAZ expression
YAP) expression in both fibroblasts and epithelial cells, and we provide new insight into the underlying mechanisms. Specifically, we demonstrate that TGF␤ stimulates TAZ expression by a Smad3-independent, p38-mediated and redox-sensitive pathway, leading to non-conventional (net translocation-independent) activation of MRTF, which in turn stimulates the TAZ promoter. TGF␤-induced TAZ expression increases TAZ-dependent transcription and, importantly, strongly sensitizes the cells for a subsequent TAZ-activating stimulus (TGF␤-induced priming).
Enhanced TAZ expression upon TGF␤ exposure was observed in a few previous studies including our own (7,31,35), but the mechanism, the critical TFs, and the significance of this process remained undefined. The rationale to test the potential role of MRTF originated from our observations that (a) the TAZ promoter contains a CArG box and (b) that MRTF downregulation reduced basal TAZ expression in tubular cells (9). The conclusions that MRTF is both an indispensable and a sufficient mediator of TGF␤-induced TAZ expression are based on our current findings that (a) pharmacological inhibition (CCG-1423) or genetic down-regulation (siRNA) of MRTF abolishes the TGF␤triggered TAZ expression and (b) that overexpression of MRTF (but not SRF) drives the TAZ promoter in a CArG box-dependent manner and increases TAZ protein expression.
Considering the signaling that links TGF␤ to TAZ expression (and thus MRTF), our finding that neither Smad3 down-regulation nor its pharmacological inhibition interfered with this process excluded the involvement of this canonical pathway. In fact, the absence of Smad3 tended to increase the TGF␤-triggered rise in TAZ protein expression, which may well be explained by our earlier finding that Smad3 can bind to MRTF and inhibit its transcriptional activity on certain promoters (e.g. that of SMA) (9,42). Silencing Smad2 induced a dramatic drop in basal YAP expression and reduced TAZ levels as well. This effect was observed in the absence of TGF␤, suggesting a constitutive role for Smad2 in YAP and likely TAZ expression. This effect is distinct from the TAZ-specific and TGF␤-induced regulation. Although TAZ remains responsive to TGF␤ stimulation in Smad2-silenced cells, the attained levels are lower. Thus, at this point we cannot exclude that Smad2 might play a role in the TGF␤-regulated TAZ response as well. It is noteworthy in this regard that Smad2 was reported to interact with MRTF-B, and the complex may synergize on the SMA promoter in smooth muscle cells (64). Further studies are warranted to test whether a similar mechanism might be relevant for TAZ as well. If so, it would point to an intriguing, distinct, and opposite regulation by Smad2, since our earlier studies have shown that Smad3 inhibits MRTF signaling. Thus, it remains to be tested whether one arm of the canonical pathway (Smad2) can selectively synergize with non-canonical pathways regulating TAZ expression.
Of the non-canonical TGF␤ pathways, activation of the MEK/ERK signaling does not seem to play a role in this process either. This finding may be surprising in light of a recent study wherein insulin-like growth factor was found to increase TAZ expression in a MEK/ERK-dependent manner during osteogenic differentiation of mesenchymal stem cells (65). However, the insensitivity to ERK inhibition in our setting is in full agreement with the central role of MRTF as the key mediator of the TGF␤ effect, inasmuch as ERK-mediated phosphorylation (at Ser-454) was shown to inhibit MRTF activity by increasing actin binding and promoting MRTF efflux from the nucleus (66,67). As a recent detailed analysis shows, the picture is more complex because MRTF can be phosphorylated at least on 24 Ser/Thr sites, many of which are targeted by ERK, and a subset of which plays either negative or positive roles, affecting nuclear influx, efflux, or transcriptional activity (52). Although such modifications may fine-tune MRTF responses to various stimuli, our data reveal that ERK-dependent phosphorylation cannot account for the TGF␤-induced elevated TAZ expression. Similarly, ROK also fails to convey this effect, although it can promote MRTF phosphorylation (37), and its inhibition can mitigate Rho-mediated MRTF translocation in response to certain stimuli (41,68). In contrast, the lack of Rho/ROK involvement is consistent with the facts that (a) in many cells TGF␤ is a weak and late-onset activator of Rho (42) and (b) it induces none or only marginal MRTF translocation (42,50). Indeed, we observed only a late onset (Ն6h) and mild nuclear accumulation of MRTF upon TGF␤ exposure, an effect likely due to TGF␤-induced expression of Rho-GEFs (69). Because TAZ expression preceded these changes, their causal contribution (at the early stage investigated) can be excluded. However, this does not mean that the effect of TGF␤ on TAZ was independent of the F-actin status. Actin depolymerization by latruncu-

TGF␤ induces p38-dependent, MRTF-mediated TAZ expression
lin B prevented the effect of TGF␤ on TAZ (data not shown). Because actin depolymerization drives MRTF out of the nucleus by inhibiting influx and activating efflux (37,70), this finding points to the need for the constitutive nuclear presence of some MRTF, which then can be post-translationally modified. It is noteworthy in this regard that during the late stages of our studies, a paper by Liu et al. (71) was published, which implicated MRTF as a mediator of heregulin ␤1 (HRG1)-induced TAZ expression in breast cancer cells. However, HRG1, as opposed to TGF␤, caused robust nuclear translocation of MRTF. Together these observations imply that, depending on the stimulus, MRTF may drive TAZ expression by two distinct mechanisms, namely in a nuclear translocation-dependent and -independent manner. Perhaps such differential regulation might confer target selectivity (e.g. turning on only a subset of MRTF-dependent genes, whose promoter might show selective affinity for such post-translationally modified MRTF). Clearly, these modes do not have to be mutually exclusive. Indeed, we found that serum induced robust MRTF translocation and also induced TAZ expression, albeit its potency was less than that of TGF␤, which failed to induce translocation. Such dual regulation (which is supported by the fact that the transcriptional activity of nuclear resident MRTF mutants is still modified by mutations in the phospho-target sites (52)) offers distinct cytoskeleton remodeling-dependent (e.g. mechanosensitive) and -independent (e.g. TGF␤) control of TAZ expression, integrated via MRTF.
What could this translocation-independent, TGF␤-provoked signaling be, and how could it impact MRTF activity? Our findings suggest a key role for p38, a well-known non- Figure 11. TGF␤-induced TAZ expression increases TAZ-dependent transcriptional activity and primes cells for augmented TAZ activity in response to a stimulus inducing TAZ translocation. A, to demonstrate TAZ-dependent transcriptional activity, luciferase assays were performed using a reporter construct harboring tandem TEAD-binding elements in front of firefly luciferase (TEAD-Luc). Cells were transfected with the TEAD-Luc/Renilla system along with siNR or TAZ-specific siRNA (siTAZ, 25 nM) for 24 h and then left untreated or exposed to TGF␤ for an additional 24 h, followed by the determination of TEAD activity (n ϭ 3). TAZ knockdown was confirmed by Western blotting performed in parallel. B, cells were incubated in the absence or presence of TGF␤ for 5 h and subsequently challenged, where indicated, with S1P (1 M) for 1 h. Cells were then stained for endogenous TAZ, and nuclei were visualized by DAPI. Scale bar, 20 m. C and D, to demonstrate the priming effect of TGF␤ on TAZ activity, cells were pretreated with TGF␤ for 6 h and then further stimulated with S1P (1 M) to induce translocation of TAZ to the nucleus. The effect on TAZ protein expression (C) and TEAD-Luc activity (D) was then measured (n ϭ 3). Cntrl, control. n.s., non-significant.

TGF␤ induces p38-dependent, MRTF-mediated TAZ expression
canonical transducer in the TGF␤ pathway (72,73). Inhibition or down-regulation of p38 strongly reduced the TGF␤-induced TAZ mRNA and protein expression and the concomitant shift in the apparent molecular weight of MRTF (phosphorylation). Importantly, a very recent and elegant study has shown that p38 can promote MRTF phosphorylation by two mechanisms: it can directly phosphorylate MRTF at unidentified site(s), and it activates MK2, which phosphorylates MRTF at Ser-312 and Ser-333 (53). However, the authors found no functional consequence (e.g. altered nuclear transport, SRF binding, and transcriptional effect) of the MK2-mediated MRTF phosphorylation. We found that pharmacological inhibition of MK2 reduced TAZ expression and MRTF phosphorylation. Although the exact mechanism warrants further studies, these results suggest that p38-mediated MRTF phosphorylation alone or in combination with MK2-mediated phosphorylation promotes TAZ transcription. Of note, TGF␤-induced TAZ and SMA expression show remarkable similarities in that both are MRTF-dependent, partly p38, and MK2-mediated processes (59,74,75). In fact, inhibition of p38 mitigates the MRTF-dependent activation of SMA expression by other stimuli (contact uncoupling, osmotic stress) (68, 76) as well.
As opposed to MRTF, the overexpression of its partner SRF is insufficient to increase TAZ expression. Nonetheless, in agreement with previous reports (77), we found that TGF␤ also increases SRF expression in a p38-, Akt-, and MRTF-dependent manner (data not shown). This effect might also facilitate TAZ expression when combined with MRTF activation. It is also conceivable that p38 and/or MK2 may modify SRF, thereby increasing the activity of the MRTF-SRF complex. Furthermore, Akt likely acts via a post-transcriptional mechanism because its inhibition did not or only marginally affected TAZ mRNA. Our data suggest that Akt might support TAZ expression via the Akt/mTOR/S6 kinase pathway, which increases protein translation. Interestingly, SMA expression can be regulated by this pathway (77).
Besides their direct actions, MRTF and p38 may also promote TAZ expression via indirect mechanisms, some of which involve strong positive feedback loops. For example, we have recently described that the Nox4 promoter harbors a CArG box, and MRTF potentiates Nox4 expression (7). In parallel, Lee et al. (51) reported that Nox4 promotes MRTF phosphorylation in a redox-sensitive manner. We confirmed this finding and found that inhibition of Nox4 reduced TAZ expression and TAZ-mediated transcription. Because Nox4-derived H 2 O 2 can activate p38 (59), this could be a mechanism whereby Nox4 promotes MRTF phosphorylation. Together, these findings not only suggest that TAZ expression is a redox-regulated process, but they also imply that the interplay among TGF␤, p38, MRTF, and Nox4 contains at least two self-augmenting circuits. Interestingly, TAZ itself may participate in these feed-forward processes. We recently found that YAP and TAZ are also required (at least in epithelial cells) for the TGF␤-induced Nox4 expression (7). Finally, our ongoing studies suggest that mechanical stimuli induce TGF␤ production via MRTF-and TAZ-dependent processes, and a recent paper reported that enhanced YAP signaling results in the overexpression of the TGF␤ and the TGF␤ receptor (78). In summary, the TGF␤/p38/MRTF path-way and its interplay with YAP/TAZ signaling contains an abundance of self-perpetuating interactions. Such positive feedbacks may underlie the focal and rapidly progressive character of fibrogenic processes.
So far, the majority of studies addressing the TF control of Hippo transcriptional coactivators concentrated on YAP. Given that the effect of MRTF is specific to TAZ, it remains to be tested whether previously identified YAP-regulating factors (e.g. cAMP-response element-binding protein (79) and GAbinding protein (80)) modulate or not TAZ expression as well. HIF-1 (81) and NFB (82) were shown to transcriptionally regulate TAZ. Future studies should clarify whether these TFs act in parallel or (at least in part) in series with SRF/MRTF.
Finally, we briefly consider the physiological and pathological relevance of our findings. We have shown that TGF␤-promoted TAZ expression results in enhanced TAZ transcriptional activity and strongly sensitizes cells to the effect of other TAZ activators. Physiologically increased TAZ expression was observed in the context of TGF␤-induced osteogenic differentiation of mesenchymal stem cells (35) and during embryonic stem cell renewal or differentiation (31). TAZ is a key mediator of osteogenesis and stem cell differentiation, and its overexpression is likely an essential contributor to these processes. Interestingly, MRTF was also reported to be a key regulator of osteogenesis and mesenchymal cell fate (34,83,84), effects that may well be related to its TAZ-inducing capacity. Pathologically, TAZ overexpression is characteristic in several tumors. Moreover, this study shows that TAZ is strongly up-regulated in UUO, a robust fibrosis model, and a recent work demonstrates YAP overexpression in diabetic nephropathy (22). Thus, the picture is emerging that TAZ (and YAP) overexpression is a major feature of cancer and fibrosis. It is noteworthy in this regard that pirfenidone, a clinically used antifibrotic drug potently inhibited TGF␤-induced TAZ expression. Heightened TAZ expression may have two important functional consequences, particularly in the context of increased/altered TGF␤ signaling. First, the higher TAZ levels, especially when combined with TAZ-translocating mechanical stimuli (such as a stiff extracellular matrix) will result in vastly enhanced TAZmediated transcriptional responses. These will augment the production of extracellular matrix components and fibrogenic mediators, creating a feed-forward pathomechanism. Second, prolonged TGF␤ exposure, both in cancer and fibrosis, often results in a drop of Smad3 expression (85,86). This phenomenon contributes to the switch between the tumor suppressor to the tumor promoter role of TGF␤ in cancer and may facilitate MRTF-mediated fibrogenic processes in the late phases of fibrosis (42,87). Because Smad3 facilitates TAZ nuclear localization and transcription, the loss of Smad3 could result in diminished TAZ signaling. However, Smad3-independent, MRTF-mediated overexpression of TAZ might compensate for the loss of Smad3, thereby maintaining the fibrogenic process.
Taken together, we propose that the TGF␤-induced, MRTFmediated TAZ up-regulation described herein is an important pathomechanism in organ fibrosis. Accordingly, strategies targeting MRTF and TAZ may be promising therapeutic approaches in this disease entity.

Chromatin immunoprecipitation (ChIP)
ChIP was performed essentially as in our previous studies (9). Briefly, C3H/10T1/2 cells were grown in 10-cm dishes until reaching confluence and then treated in the presence or absence of TGF␤ (5 ng/ml) for 3 h. The cells were then treated with formaldehyde to achieve cross-linking. The lysis and sonication of isolated chromatin to shear the DNA were performed using the recommended guidelines and reagents supplied in the Magna ChIP A/G kit (Millipore). Immunoprecipitation was carried out using either 2 g of anti-SRF antibody (Cell Signaling; catalog no. 5147), 5 g of anti-MRTF A antibody (Santa Cruz Biotechnology; catalog no. sc-32909), or normal rabbit IgG (Santa Cruz Biotechnology; catalog no. sc-2027) as a negative control. After reverse cross-linking and DNA purification, qPCR (IQ cycler; Bio-Rad) was performed using the following primer sequences found within the mouse TAZ promoter, 5Ј-AAACCGTCTCGCAGACAACT-3Ј and 5Ј-GGATCTGC-CAGAGGTCGGA-3Ј. Immunoprecipitation was carried out, and qPCR was performed in triplicate.

Western blotting
Following the indicated treatments, cells were lysed on ice with cold Triton X-100 lysis buffer (30 mM HEPES, pH 7.4, 100 mM NaCl, 1 mM EGTA, 20 mM sodium fluoride, and 1% Triton X-100) supplemented with 1 mM Na 3 VO 4 , 1 mM phenylmethylsulfonyl fluoride, and Complete Mini Protease Inhibitor (Roche Applied Science, Mississauga, Ontario, Canada). Lysates were boiled in 2ϫ Laemmli buffer, and protein concentration was determined using BCA assay (Thermo Fisher Scientific) to ensure equal loading for SDS-PAGE. Subsequent steps were performed, as described previously (42), using Bio-Rad Transfer System. Nitrocellulose membranes were blocked with 3% BSA or skimmed milk in Tris-buffered saline containing 0.1% Tween (TBST), incubated with primary antibody overnight, washed with TBST, and incubated with HRP-conjugated secondary antibody for 1 h. Blots were developed using enhanced chemiluminescent assay (GE Healthcare, Mississauga, Ontario, Canada). The signal was captured using X-ray film or a ChemiDocTM Touch Imaging System version 1.2.0.12. Densitometry was performed using a GS800 densitometer and Quantity One software or the Image Lab software (Bio-Rad). Fold changes in protein expression were similar with the two methods. To increase accuracy, if the basal signal was near-threshold, data were normalized to signal levels obtained upon TGF␤ stimulation, which was taken as 100%.

Quantification of S6 protein phosphorylation
C3H/10T1/2 cells were plated at 4000 cells/well in 96-well plates (Corning 3882) 1 day before drug treatment. Cells were the serum-starved for 2 h and then exposed to DMSO, AKTi (10 M), TGF␤ (5 ng/ml), or AKTi ϩ TGF␤ for 6 h. Cells were fixed with 4% PFA for 30 min and permeabilized with 0.3% Triton X-100 and blocked with 5% goat serum. Cells were incubated with the primary antibodies (S6 and pS6) at 1:300 dilution for 2 days at 4°C, washed, and exposed to secondary antibodies (goat anti-mouse catalog no. A28175 and goat anti-rabbit catalog no. A27039 from Molecular Probes) (1:3000) overnight at 4°C. The secondary antibodies were removed; the cells were stained with Hoechst 33342 (1 M for 15 min) and washed three times with PBS. The 96-well plate was loaded onto INCELL6000 (GE Healthcare) high-content fluorescence microscope and imaged using the appropriate channels. Twelve random fields were imaged within each well. TIFF images were then loaded into the Acapella automated image analysis software (PerkinElmer Life Sciences) for nuclei identification and quantification of fluorescence intensities. Routinely 1000 -3000 cells were quantified per condition. Data were imported into FlowJo (version 10.2) cytometry analysis software to determine the mean fluorescence intensities for the different channels. The pS6 kinase signal was normalized to the corresponding total S6 kinase signal. None of the treatments caused significant changes in the total S6 protein level.

Nuclear fractionation
After the indicated treatments, lysates were prepared and fractioned into cytoplasmic and nuclear components using the Pierce Ne-Per kit (Thermo Fisher Scientific) as per the manufacturer's instructions. To reduce viscosity, nuclear extracts were incubated with 1 g/ml DNase on ice for 20 min.

Detection of molecular weight shift
To monitor changes in the apparent molecular weight of MRTF, proteins from cell lysates were separated using 6% mini gels. Samples were run until the Ϸ75-kDa marker reached the edge of the gel. MetaMorph Premiere software (Molecular Devices) line tool was used to determine the distance between the lowest ends of the MRTF bands obtained from control and treated samples. The shift (S) between vehicle-and TGF␤treated samples was taken as unity (100%). Changes detected TGF␤ induces p38-dependent, MRTF-mediated TAZ expression upon inhibitor treatments (I) were compared with this control and expressed as I/S ratio.

Immunofluorescence microscopy
Cells were plated on glass coverslips. Subconfluent cells were treated and fixed after indicated times with 4% PFA (Canemco and Marivac, Lakefield, Quebec, Canada) and incubated in 100 mM glycine and 0.1% Triton X-100 in PBS. Samples were then blocked with 3% BSA/PBS, incubated with primary antibody overnight, washed with PBS, and incubated with 4,6diamidino-2-phenylindole, dihydrochloride (DAPI) (Lonza, Basel, Switzerland) and fluorescent-conjugated secondary antibody (Jackson ImmunoResearch, West Grove, PA) for 1 h in the dark. Coverslips were mounted on slides with DAKO mounting fluid (Agilent, Burlington, Ontario, Canada). Images were taken using a WaveFX spinning-disk confocal microscope system (Quorum Technologies, Guelph, Canada) equipped with an ORCA-Flash4.0 digital camera or an Olympus IX81 microscope (Olympus, Tokyo, Japan) coupled to an Evolution QEi monochrome camera (Media Cybernetics, Silver Spring, MD). Images were processed and analyzed with MetaMorph Premiere software (Molecular Devices). Cell midsections were taken to analyze the nuclear/cytoplasmic distribution of the indicated proteins. Signal intensity in the nucleus was divided by intensity in the cytoplasm for all cells. The nuclear/cytosolic ratio was determined under each condition and compared with the untreated control.

Unilateral ureteral obstruction (UUO)
The UUO model was performed as described previously (7). Briefly, 6 -8-week-old male C57BL/6 mice (The Jackson Laboratory, Bar Harbor, ME) underwent either left-sided UUO or sham surgery. Mice were anesthetized, and a flank incision was made on the left side to identify the left kidney and ureter. Subsequently, two 4-0 silk suture knots were tied on the left ureter. One, 3 or 7 days post-surgery, the mice were sacrificed, and the left kidneys were harvested. Half of the kidney was snap-frozen in liquid nitrogen, and the samples were stored at Ϫ80°C until processed. The Animal Care Committee of the St. Michael's Hospital approved this study.

Whole-kidney lysates
Kidney samples were processed as described previously (7). Briefly, kidney samples from sham and UUO mice were cut and added to ice-cold RIPA buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1% Triton X-100, 1% sodium deoxycholate, 0.1% SDS, and 1 mM EDTA) supplemented with 1 mM Na 3 VO 4 , 1 mM phenylmethylsulfonyl fluoride, and Complete Mini Protease Inhibitor (Roche Applied Science, Mississauga, Ontario, Canada). Samples were homogenized with a rotator-stator homogenizer and then centrifuged at 12,000 rpm for 20 min at 4°C. Supernatant was collected and used to determine protein concentration, and samples were analyzed using SDS-PAGE and Western blotting as above.

Statistical analysis
Data are presented as representative blots or images from at least three similar experiments or as means Ϯ S.E. of the mean (S.E.) for the number of experiments indicated. Statistical significance was determined by two-tailed Student's t test or ANOVA (Tukey or Dunn post hoc testing for parametric and nonparametric ANOVA, as appropriate) using Excel (Microsoft) or Prism software. p Ͻ 0.05 was accepted as significant. *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001; ****, p Ͻ 0.0001.
Author contributions-M. Z. M. and J. F. B. performed all the experiments on fibroblasts and epithelial cells, respectively, analyzed the data, prepared the corresponding figures, and contributed to the writing of the manuscript. P. S. generated constructs by mutagenesis, performed the ChIP assay, and provided experimental advice; Q. D. performed the in vivo experiments and data analysis; T. Y. helped with experiments for the revision; K. S. and S. F. P. contributed to the design of studies, to the critical interpretation of the data, and to the final assembly of the manuscript; and A. K. conceived the study and wrote the majority of the manuscript. All authors reviewed the results and approved the final version of the manuscript.