Hyperosmotic stress initiates several adaptive responses, including the remodeling of the cytoskeleton. Besides maintaining structural integrity, the cytoskeleton has emerged as an important regulator of gene transcription. Myocardin-related transcription factor (MRTF), an actin-regulated coactivator of serum response factor, is a major link between the actin skeleton and transcriptional control. We therefore investigated whether MRTF is regulated by hyperosmotic stress. Here we show that hypertonicity induces robust, rapid, and transient translocation of MRTF from the cytosol to the nucleus in kidney tubular cells. We found that the hyperosmolarity-triggered MRTF translocation is mediated by the RhoA/Rho kinase (ROK) pathway. Moreover, the Rho guanine nucleotide exchange factor GEF-H1 is activated by hyperosmotic stress, and it is a key contributor to the ensuing RhoA activation and MRTF translocation, since siRNA-mediated GEF-H1 downregulation suppresses these responses. While the osmotically induced RhoA activation promotes nuclear MRTF accumulation, the concomitant activation of p38 MAP kinase mitigates this effect. Moderate hyperosmotic stress (600 mosM) drives MRTF-dependent transcription through the cis-element CArG box. Silencing or pharmacological inhibition of MRTF prevents the osmotic stimulation of CArG-dependent transcription and renders the cells susceptible to osmotic shock-induced structural damage. Interestingly, strong hyperosmolarity promotes proteasomal degradation of MRTF, concomitant with apoptosis. Thus, MRTF is an osmosensitive and osmoprotective transcription factor, whose intracellular distribution is regulated by the GEF-H1/RhoA/ROK and p38 pathways. However, strong osmotic stress destabilizes MRTF, concomitant with apoptosis, implying that hyperosmotically induced cell death takes precedence over epithelial-myofibroblast transition, a potential consequence of MRTF-mediated phenotypic reprogramming.
- cell volume
deviation from the physiologic hydration state (i.e., “osmotic stress”) represents a major threat for normal cell functions and survival. Accordingly, osmotically challenged cells mobilize a set of adaptive responses, which include the activation of volume-correcting transport systems (29), the expression of osmoprotective genes (8), and the remodeling of the cytoskeleton (17). Hyperosmotic stress-induced cytoskeletal restructuring manifests in enhanced peripheral actin polymerization (18, 26, 52, 57), the formation of a polygonal actin-myosin lattice (38), and enhanced cell contractility (15, 16, 51). These responses are thought to reinforce the cell enabling it to withstand the shrinkage-provoked mechanical trauma. Regarding the underlying mechanisms, we and others have shown that hyperosmolarity activates several Rho family GTPases, which in turn are central mediators of the ensuing cytoskeletal effects (16, 37, 56, 68, 75). For example, RhoA/Rho kinase-mediated cofilin phosphorylation is a key contributor to the shrinkage-induced increase in F-actin (66).
While crucial for structural reinforcement, the hyperosmotic activation of Rho family GTPases and the consequent cytoskeleton remodeling itself may fulfill other roles too. Namely, the Rho family and the cytoskeleton have emerged as important regulators of gene expression (49). One key link between the state of the actin skeleton and transcriptional control is myocardin-related transcription factor (MRTF), the two isoforms of which, MRTF-A (also known a MAL or MKL1) and MRTF-B (MKL2), are ubiquitously expressed. MRTF is an actin-regulated transcriptional coactivator, which, when stimulated, forms a complex with serum response factor (SRF) thereby driving the expression of a variety of cytoskeletal genes through the CC(AT)6GG cis-element, called the CArG box (50, 53, 72). Under resting conditions, MRTF is bound to monomeric (G) actin and resides in the cytosol. According to the current view, upon enhanced actin polymerization, G-actin may be “stolen away” from MRTF, which consequently unmasks its nuclear localization sequence thereby promoting its nuclear accumulation (43, 45, 69). This scenario raises the intriguing possibility that MRTF might be an osmosensitive molecule whose nuclear transport is affected by cell shrinkage. This is of particular interest, since so far the only well-characterized mechanism linking osmotic alterations to transcriptional control is the one mediated by the tonicity-responsive enhancer-binding protein (TonEBP, also termed as OREBP or NFAT5), which is regulated by osmolarity at multiple levels (transport, activity, expression) and drives the expression of osmosensitive solute transporters and osmolyte-generating enzymes through the osmotic response element (8, 30). Thus, while TonEBP is the master regulator of the synthesis and transport of nonperturbing osmolytes (and thus cell survival), it is conceivable that MRTF, a cytoskeleton-regulated and cytoskeleton-regulating transcription factor, might be involved in the structural adaptation to hyperosmolarity. However, regulation by MRTF is a potentially dangerous mechanism, especially in epithelial cells, since it can induce major transcriptional reprogramming of the cytoskeleton, facilitating the expression of mesenchymal or smooth muscle-specific proteins. Indeed, recent studies from our laboratory (14, 21, 40) and from others (31, 42, 44) have shown that MRTF is a key mediator of epithelial-mesenchymal transition (EMT), a phenotypic change that is thought to play an important role in the pathogenesis of organ fibrosis (13, 33, 48, 55). Accordingly, sustained translocation of MRTF induces epithelial-myofibroblast transition (EMyT), the most robust form of EMT characterized by the expression of α-smooth actin (SMA) (21, 40). Importantly, the tissue accumulation of SMA shows close correlation with the severity of organ (e.g., kidney) fibrosis (3, 54). Indeed, MRTF has recently been shown to be a direct mediator of myofibroblast generation during experimental fibrosis (63).
With this scenario in mind, we sought to determine the effect of osmotic stress on the intracellular trafficking of MRTF and to discern the underlying signaling and the potential consequences. We also investigated the hitherto largely unexplored upstream mechanism for RhoA activation under hyperosmotic stress. Our results suggest that osmolarity regulates MRTF at multiple levels in tubular cells. Modest hyperosmolarity induces nuclear MRTF accumulation through the GEF-H1/RhoA/Rho kinase pathway and facilitates nuclear efflux of MRTF through p38 MAP kinase. MRTF is an osmoprotective factor under these conditions. Strong hyperosmotic stress leads to rapid degradation of MRTF, ensuring that the “preferred” cell fate is death rather than epithelial-myofibroblast transition.1
MATERIALS AND METHODS
Chemicals and antibodies.
SB203580, Y-27632, CCG-1432, and zVAD-fmk were from EMD Biosciences (Mississauga, ON, Canada). MG132 was obtained from Sigma (Mississauga, ON, Canada). Bovine serum albumin (BSA) was from BioShop Canada (Burlington, ON, Canada). The Complete Mini Protease inhibitor tablets were from Roche Diagnostics (Laval, QC, Canada). Antibodies against the following proteins were used: RhoA, GEF-H1 (55B6), and SRF from Cell Signaling Technology (Danvers, MA); histones from Millipore (Temecula, CA); GAPDH, lamin A/C (N-18), p38 MAP kinase (designated as p38) and phospho-Thr180 p38 (pp38) from Santa Cruz Biotechnology (Santa Cruz, CA). The polyclonal MRTF antibody (BSAC) has been described previously (59). Peroxidase and Cy3-labeled secondary antibodies were from Jackson ImmunoResearch (West Grove, PA). DAPI nucleic acid stain was from Invitrogen.
Cells and cell treatment.
LLC-PK1, a kidney proximal tubule epithelial cell line (AT1), was used as in our earlier studies (16, 66). Cells were maintained in DMEM medium (Gibco Life Technologies, Grand Island, NY) supplemented with 10% fetal bovine serum and 1% antibiotic suspension (penicillin and streptomycin, Invitrogen) in an atmosphere containing 5% CO2. The isotonic Na+ medium (290 ± 5 mosM) contained 130 mM NaCl, 3 mM KCl, 1 mM MgCl2, 1 mM CaCl2, 5 mM glucose, and 20 mM Na-HEPES pH 7.4. Confluent cells were serum depleted for 3 h in DMEM before the experiments, followed by 15 min of incubation in Na+-medium before the indicated treatments. If not otherwise stated, hyperosmolarity was induced by adding 150 mM NaCl to the isotonic Na+ medium. When osmolarity was varied, the desired osmotic concentration was obtained by adding the appropriate concentration of NaCl. In a subset of experiments, instead of NaCl, equiosmotic sucrose was added with identical results. In long-term experiments (6 h or more), NaCl was added to the serum-free DMEM medium. To induce cell contact disassembly, cells were thoroughly washed with phosphate-buffered saline (PBS, Invitrogen) and cultured in nominally calcium chloride-free DMEM (low-calcium medium: LCM) as in our earlier studies (39, 40).
Gene silencing using siRNA.
The siRNAs targeting the sequence in porcine MRTF-A (CCAAGGAGCTGAAGCCAAA), MRTF-B (CGACAAACACCGTAGCAAA), GEF-H1 [AACAAGAGCATCACAGCCAAG (no. 1) and AACGGGCATCTCTTCACCACC (no. 2)] (32, 71), and RhoA (AAAGCAGGTAGAGTTGGCTTT) were obtained from Applied Biosystems/Ambion (Austin, TX). Cells were transfected with either 100 nM siRNA (GEF-H1 or RhoA siRNA) or (50 nM MRTF-A+50 nM MRTF-B siRNAs) oligonucleotide using the Lipofectamine RNAiMAX Transfection Reagent (Invitrogen) according to the manufacturer's instructions. Control cells were transfected with 100 nM Silencer siRNA negative control no. 2 (nonrelated siRNA) (Applied Biosytems/Ambion). Experiments were performed 24–48 h after transfection. The levels of the silenced proteins were routinely checked by Western blotting.
Constructs and transient transfection.
GST-RBD (RhoA-binding domain of Rhotekin, amino acids 7–89) and GST-RhoA(G17A) (2, 23) were kind gifts from Dr. K. Burridge (Univ. of North Carolina, Chapel Hill). The hemagglutinin (HA)-tagged MRTF construct was generated as described previously (14). LLC-PK1 cells were transfected using the transfection reagent FuGENE 6 (Roche Molecular Biochemicals) according to the manufacturer's instructions using 1 μg DNA/well for six-well plates or as specified in the figures.
Western blot analysis.
Following treatment, cells were lysed on ice with cold lysis buffer [100 mM NaCl, 30 mM HEPES (pH 7.5), 20 mM NaF, 1 mM EGTA, 1% Triton X-100, supplemented with 1 mM Na3VO4, 1 mM PMSF, and Complete Mini Protease Inhibitor Cocktail Tablet (Roche)]. Protein concentration was determined by the bicinchoninic acid assay (BCA) (Pierce Biotechnology) with BSA used as standard. SDS-PAGE and Western blotting were performed as described previously (40). Blots were blocked in Tris-buffered saline containing 3% BSA and incubated with the primary antibody overnight. Antibody binding was visualized with the corresponding peroxidase-conjugated secondary antibodies and the enhanced chemiluminescence method (kit from GE Healthcare Lifesciences). Where indicated, blots were stripped and reprobed to demonstrate equal loading.
Preparation of nuclear extracts.
Nuclear extracts were prepared from confluent layers of LLC-PK1 cells grown on 6-cm dishes, as described previously (40), using the NE-PER Nuclear Extraction Kit from Pierce Biotechnology (Rockford, IL) according to the manufacturer's recommendation. The nuclear extracts were collected, and their protein concentration was determined. Samples containing 10 μg protein were analyzed by Western blotting. Antibodies against histones or lamin were used as markers of nuclear fraction.
GEF-H1 and RhoA activation assays.
GST-RBD and GST-RhoA(G17A) were used to capture active RhoA and GEFs, respectively, as described (32, 70, 71). Briefly, confluent LLC-PK1 cells were grown on 6- or 10-cm dishes and treated as indicated in the respective figures. Cells were lysed with ice-cold buffer containing 100 mM NaCl, 50 mM Tris base (pH 7.6), 20 mM NaF, 10 mM MgCl2, 1% Triton X-100, 0.5% deoxycholic acid, 0.1% SDS, 1 mM PMSF, 1 mM Na3VO4 and protease inhibitors (RhoA activation assay) or 20 mM HEPES (pH 7.5), 150 mM NaCl, 5 mM MgCl2, 1% Triton X-100, 1 mM DTT, 1 mM PMSF, and protease inhibitors (GEF activation assay). After centrifugation, aliquots for determination of total RhoA or GEF-H1 were removed. The remaining supernatants were incubated at 4°C for 45 min with 20–25 μg of GST-RBD or GST-RhoA(G17A)-covered beads, followed by extensive washing. Total cell lysates and the captured proteins were analyzed by Western blotting using RhoA or GEF-H1 antibody. Results were quantified by densitometry (see below).
To quantify results, films with nonsaturated exposures were scanned and densitometry analysis was performed using a GS-800 calibrated densitometer and the “band analysis” option of the Quantity One software (Bio-Rad). In each assay the amount of the investigated protein species was normalized to the appropriate control (e.g., active RhoA to total RhoA, active GEF-H1 to total GEF-H1, nuclear MRTF to nuclear marker protein). Data are expressed as fold changes. Since the basal levels of active RhoA and nuclear MRTF were often either undetectable or just slightly above the background, to achieve accurate and stringent comparison, signals were expressed relative to the response detected in stimulated cells at a selected time point (taken as unity) as described in the figures.
Confluent cells grown on coverslips were treated as indicated in the corresponding figures and fixed with 4% paraformaldehyde. Immunofluorescence staining was carried out as described (39). Briefly, following permeabilization with 0.1% Triton X-100, the coverslips were blocked with 3% BSA in PBS. Next, cells were incubated with anti-MRTF (1:100). Bound antibody was detected using the corresponding fluorescent secondary antibody (1:1,000), which also contained DAPI to counterstain nuclei. All samples were viewed using an Olympus IX81 microscope (Melville, NY) coupled to an Evolution QEi Monochrome camera (Media Cybernetics, Bethesda, MD). Nuclear translocation was quantified as described in our earlier studies (21, 40). Briefly, MRTF localization was denoted as cytosolic, if clear nuclear exclusion was seen, and nuclear if MRTF was accumulated in the nucleus to the extent that the demarcation of nucleus was obvious. These conditions correspond to a nuclear/cytoplasmic ratio of <0.6 and >1.5, respectively, as determined by quantitative image analysis (40). At least 10 randomly selected fields (>200 cells) were quantified for each condition in three experiments.
Phase contrast images were obtained by a Nikon eclipse TS100 microscope (×40 objective), using the Nikon digital sight system and SPOT software.
Caspase-3 activity assay.
For total caspase-3 activity assay, cells were grown in 12-well plates and treated as indicated in the corresponding figures. Following treatment, cells were permeabilized with caspase-3 lysis buffer (10 mM Tris·HCl, 10 mM NaH2PO4/NaHPO4 pH 7.5, 130 mM NaCl, 10 mM sodium pyrophosphate, 1% Triton X-100). The collected lysates were then added to the protease assay buffer (20 mM HEPES pH 7.5, 10% glycerol, 2 mM DTT) in conjunction with 20 μM fluorogenic caspase-3 substrate (Ac-DEVD-AMC) (BD Biosciences). The samples were incubated at 37°C for 1 h. The generated fluorescence of the cleaved reporter subunit was measured in a cuvette fluorimeter (Photon Technologies International) using excitation and emission wavelengths of 380 nm and 440 nm, respectively. The obtained readings were normalized to total protein levels as determined by BCA assay.
Luciferase reporter assays.
These were performed as described previously (40). Briefly, cells were plated onto 12-well plates and at ∼60% confluence were transfected with 0.5 μg/well reporter plasmid pGL3–3DA-Luc, which contains the firefly luciferase gene under the control of a CArG box triplet (64), or pSMA-Luc, which harbors the proximal 765-base pair segment of the rat SMA promoter in the pGL3 vector (40). Normalization was carried out by cotransfection of the enhancer-less Renilla luciferase control plasmid pHRG-B (Promega) (0.125 μg/well), which is optimal for this purpose, as it is not responsive to hypertonicity, as verified in our previous studies (24). At 16–20 h posttransfection, cells were serum deprived for 3 h, and treatments were initiated by adding fresh, serum-free medium supplemented with NaCl, in the presence or absence of inhibitors, as indicated. Finally, cells were lysed and luciferase activity was determined using the Dual-Luciferase Reporter Assay System kit (Promega) and a Berthold Lumat 9507 luminometer according to the manufacturers' instructions. For each condition, treatments were done in duplicate and experiments were repeated at least three times. From each sample the firefly luciferase activity corresponding to a specific promoter construct was normalized to the Renilla luciferase activity of the same sample. Results are expressed as fold changes compared with the mean firefly/Renilla ratio of the untreated controls taken as a unity.
Data are presented as representative blots or images from at least three similar experiments or as the means ± SE for the number of experiments indicated. Statistical significance was determined by one-way analysis of variance (Tukey or Dunn post hoc testing for parametric and nonparametric ANOVA, as appropriate), or with Student's t-test. P < 0.05 was accepted as significant.
Hyperosmotic stress induces nuclear translocation of MRTF.
To test whether hyperosmolarity impacts on MRTF traffic, confluent monolayers of LLC-PK1 tubular cells were left untreated or exposed to hyperosmolarity (600 mosM total, 20 min), and MRTF distribution was visualized by immunofluorescence staining. Under isotonic conditions, MRTF was cytosolic, as verified by the nuclear exclusion of the label, visible in nearly every cell (Fig. 1A left, see DAPI-positive, MRTF-negative nuclear areas). Upon hypertonic exposure, some cells showed strong nuclear MRTF accumulation (over the cytosolic level), while in many others the nuclear exclusion disappeared, indicating that the nuclear MRTF concentration rose to the cytosolic level (Fig. 1A, right). We quantified MRTF localization by determining the percentages of cells that showed cytosolic, even, and nuclear distribution (Fig. 1B), as defined in our previous studies (Refs. 21 and 40 and see materials and methods). While under isotonic conditions no cells exhibited nuclear MRTF accumulation, and <10% had even distribution, hyperosmolarity provoked MRTF redistribution in ∼80% of the cells, with 20% and 60% showing highly nuclear and even localization, respectively. Essentially similar results were obtained irrespective of whether hyperosmolarity was induced by NaCl (Fig. 1, A and B) or osmotically equivalent sucrose (not shown). To quantify MRTF protein accumulation and to investigate the time dependence of the effect, we prepared nuclear extracts from cells that had been exposed to hyperosmolarity for 0–120 min. An increase in nuclear MRTF was clearly detectable after 5 min, peaked at 10–20 min, and gradually decayed thereafter (Fig. 1, C and D). At the peak, the nuclear MRTF content of the monolayer was approximately 20-fold compared with the resting (isotonic) level. To assess the relative magnitude of this effect, we also used low-calcium medium as a stimulus (LCM, 30 min), which, as our previous studies have shown (39, 40), uncouples intercellular contacts and thereby induces robust nuclear accumulation of MRTF (Fig. 1C). The maximal translocation elicited by LCM was approximately sixfold higher than the peak response induced by hyperosmolarity (normalized to 1 in each experiment) (Fig. 1E). Next we tested the osmotic dependence of the effect: increasing the NaCl concentration added over the isotonic level from 75 to 225 mM was accompanied by a gradual rise in the nuclear MRTF content, while stronger osmotic stress (300 mM extra) consistently resulted in less translocation, indicating that at the selected time point (20 min), the maximal effect was obtained at an overall osmolarity of 600–750 mosM (Fig. 1F). Taken together, these data show that hyperosmolarity induces a rapid, robust (20-fold) but submaximal (compared with LCM), transient and osmolarity-dependent nuclear accumulation of MRTF.
The GEF-H1/RhoA/Rho kinase pathway mediates hyperosmolarity-induced nuclear translocation of MRTF.
Next we wished to determine which signaling pathway is responsible for MRTF translocation. Hypertonicity has been shown to activate various Rho family GTPases (RhoA, Rac, Cdc42) with varying magnitude and kinetics in different cell types (16, 37, 56, 66, 68). Both RhoA and Rac have been shown to impact MRTF nuclear translocation (9, 43, 61) and both are osmotically sensitive. Because in LLC-PK1 cells hypertonicity elicits only a small and transient (<1 min) Rac activation followed by a suppression (66), while RhoA shows sizable and sustained activation (Ref. 16 and see Fig. 4A), we hypothesized that the latter might be involved in the osmotically induced MRTF translocation. To test this possibility, we used nonrelated and RhoA-specific siRNA, the latter of which essentially eliminated RhoA expression in the monolayer as verified by Western blotting (Fig. 2A). RhoA siRNA did not change MRTF distribution under isotonic conditions, but it completely abolished the osmotically induced nuclear uptake of MRTF (Fig. 2B). Since the Rho effector Rho kinase (ROK) was implicated in osmotically induced actin polymerization (66) and MRTF can undergo Rho-dependent phosphorylation (43), we tested the potential involvement of ROK in MRTF redistribution. The potent ROK inhibitor Y-27632 fully prevented the hypertonicity-provoked nuclear accumulation of MRTF, as demonstrated both by immunostaining (Fig. 2C) and Western blotting of nuclear extracts (Fig. 2D). Together, these data imply that the RhoA/Rho kinase pathway is indispensable for the hypertonicity-induced nuclear translocation of MRTF.
Little is known about the direct upstream mediators of osmotic RhoA activation. Our previous studies have proposed that ezrin-mediated inhibition of Rho guanine nucleotide dissociation inhibitor (Rho-GDI) might be involved in this process in Ehrlich-Lettre ascites cells (56), but no osmotically regulated guanine nucleotide exchange factor (GEF) (i.e., direct RhoA activator) has been so far identified. We considered GEF-H1 as a good candidate, because this key RhoA-activating molecule has been shown to be regulated by mechanical (shear) stress and other physical factors (6, 25, 71), and it can be associated with intercellular contacts (4), which have been proposed to transmit volume-dependent signals (29). To address the potential role of GEF-H1, first we asked whether it is activated by hyperosmolarity. To this end we performed affinity pull-down assays, using a RhoA(G17A) mutant-GST fusion protein, which mimics nucleotide-free RhoA, and therefore specifically binds to activated GEFs (23, 32, 70). GEFs in lysates obtained from iso- and hypertonically treated cells were precipitated with RhoA(G17A)-GST-covered beads, and the precipitates were immunoblotted with an anti-GEF-H1 antibody. Hyperosmolarity induced a rapid increase in the amount of captured GEF-H1, which peaked after 2 min at 2.5-fold over the isotonic level (Fig. 3A). To test the functional significance of the observed GEF-H1 activation, we treated the cells with nonrelated or GEF-H1-specific siRNA and performed RhoA activation assays after iso- and hypertonic exposure (Fig. 3B). GEF-H1 siRNA induced a strong reduction in GEF-H1 expression and concomitantly suppressed basal RhoA activity and prevented its rise over the isotonic (basal) level. The residual RhoA activation in the presence of GEF-H1 siRNA is likely due to the remaining GEF-H1 expression in a small portion of the cell population and might also reflect the contribution of other RhoA-activating pathways. Having seen that GEF-H1 is a key component of the osmotically induced RhoA activation, we tested its impact on MRTF translocation. As expected, GEF-H1 knockdown strongly mitigated the hypertonicity-triggered MRTF translocation, as detected both by immunofluorescence microscopy (Fig. 3C) and Western blotting of nuclear extracts (Fig. 3D).
Taken together, these results imply that the RhoA-ROK pathway is indispensable for the hypertonicity-induced MRTF translocation and that GEF-H1 is activated by hyperosmolarity and significantly contributes to the ensuing RhoA activation and consequent MRTF redistribution.
RhoA and p38 MAPK oppositely regulate nuclear accumulation of MRTF after hypertonic stimulation.
The fact that hyperosmolarity-induced MRTF translocation, while substantial, was significantly less than that induced by low-calcium medium (Fig. 1E) prompted us to consider whether this difference is due to a proportional difference in RhoA activation or may reflect additional regulatory inputs. To address this question, first we compared RhoA activation elicited by the two stimuli. Hypertonicity provoked robust RhoA activation that at every time point between 0 and 60 min was either larger than (0–10 min) or equal to (later times) the response seen upon LCM stimulation (Fig. 4A). In agreement with our previous studies, these data show that hyperosmotic stress is a strong and sustained activator of RhoA in LLC-PK1 cells, while LCM is a weaker stimulus. These findings suggested that other factors than RhoA must be responsible for the observed differences. Since RhoA activation is an absolute requirement for nuclear MRTF uptake with regard to both stimuli, we surmised that differences in pathways regulating nuclear retention/efflux of MRTF might underlie the observed difference in MRTF accumulation. Relevantly, members of the MAPK family were reported to phosphorylate MRTF (43, 46), which in turn facilitates its nuclear efflux (46). Since hyperosmolarity activates predominantly p38 MAPK (p38) in tubular cells, we compared the effect of hypertonicity and LCM on p38 phosphorylation (an accurate marker of p38 activation). Hyperosmolarity induced much stronger and longer lasting p38 activation than LCM (Fig. 4B). To discern whether p38 activation could indeed influence MRTF trafficking, we pretreated cells with SB203580, a potent p38 MAPK inhibitor, and followed MRTF distribution by immunofluorescence microscopy (Fig. 4C) and Western blotting of nuclear extracts (Fig. 4D). While SB203580 had no significant effect on MRTF distribution under isotonic conditions, osmotic stress induced markedly stronger nuclear MRTF accumulation in the presence of the inhibitor (Fig. 4C). Time-dependent monitoring of nuclear MRTF content revealed that SB203580-treated cells had more nuclear MRTF at every time point after stimulation, but the most pronounced (more than twofold) and highly significant difference was observed after 30 min of exposure to hypertonicity. Thus inhibition of p38 promoted nuclear accumulation of MRTF, presumably by mitigating or delaying MRTF efflux from the nucleus. Nonetheless, MRTF eventually left the nucleus in SB203580-treated cells as well, indicating the additional involvement of other mechanisms. In summary, the data indicate that p38 contributes to the removal of MRTF from the nucleus under hypertonic conditions, suggesting that hyperosmotic stress has a dual effect on MRTF traffic, by stimulating both RhoA-dependent nuclear uptake and p38-promoted release of MRTF.
The fate and role of MRTF under hyperosmotic conditions.
In the following experiments we wished to investigate whether hyperosmolarity could trigger MRTF-dependent transcription and whether MRTF might mediate osmoprotective effects. As an introductory study, we tested whether long-term exposure to various osmolarities impacts MRTF protein expression itself. Supplementing the isotonic medium with 100 or 150 mM NaCl for 24 h resulted in a slight increase in MRTF protein levels, whereas further elevation of the added concentration of NaCl caused a dramatic reduction in the MRTF-immunoreactive band (Fig. 5A). To better characterize this surprising effect, we applied mild or strong hyperosmolarity for a much shorter time (2 h). Again, after the addition of ≥200 mM NaCl we observed a near-complete elimination of the MRTF band (Fig. 5B), indicating that strong osmotic stress may cause acute MRTF degradation, or alternatively elicit posttranslational modifications that mask the immunoreactive epitopes (see below). Addition of 200 mM NaCl (700 mosM total) or more also induced robust activation of caspase-3 (Fig. 5C), indicating that acute exposure of LLC-PK1 to an osmotic concentration of 700 mosM or higher initiates apoptosis concomitant with the loss of the MRTF protein. Thus MRTF expression is sensitive to osmolarity, and the destabilization of the molecule occurs at a rather sharp threshold, which is similar to the threshold of apoptosis induction. Furthermore, these experiments instructed us that any potential transcriptional or protective effect should be tested at relatively mild hyperosmolarity (e.g., 600 mosM in total) and for shorter exposures. Therefore we checked the impact of 150 mM NaCl (600 mosM total, our standard conditions in the translocation experiments) on MRTF and SRF at 6 and 48 h (Fig. 5D). At this level of hyperosmolarity MRTF was stable (albeit after 48 h we occasionally detected a decrease). SRF manifested as two immunoreactive bands. A 6-h exposure to extra 150 mM NaCl reduced the intensity of the lower band, whereas after 48 h usually both bands were suppressed (see discussion).
In light of these data we checked the impact of hyperosmolarity (150 mM NaCl, 6-h exposure) on MRTF/SRF-dependent transcription using a CArG box-containing luciferase reporter construct (3DA) (Fig. 6, A and B). Hypertonic exposure caused a ≈1.8-fold increase in CArG-mediated transcription. Importantly, CCG-1423, a compound which inhibits MRTF/SRF interaction (20), completely prevented the hypertonicity-provoked activation of the reporter. As an alternative (nonpharmacologic) approach we used nonrelated siRNA or a mixture of siRNAs against MRTF-A and MRTF-B, the two isoforms of the protein. Transfection of the cells with the MRTF-specific siRNAs for 48 h strongly reduced MRTF expression (Fig. 6B, top), and it concomitantly decreased the activity of the CArG-dependent reporter under isotonic conditions and efficiently suppressed its activation by hypertonicity (Fig. 6B). The difference between the effect of CCG-1423 and MRTF siRNAs on the basal activity of the reporter is likely due to the fact that the pharmacological inhibitor was applied only 30 min before hypertonic stimulation, whereas the transfection with the siRNA was performed 48 h before hypertonic exposure, during which the cumulative expression of the reporter could also be suppressed. Taken together, these data clearly imply that moderate, intermediate duration hyperosmolarity activates MRTF/SRF-dependent transcription in tubular cells. Since the accumulation of MRTF in the nucleus is transient (see Fig. 1, C and D), these results also suggest that an early rise in nuclear MRTF concentration is sufficient to initiate or prime for a subsequent transcriptional response that manifests at a time when the bulk MRTF concentration is no longer detectably elevated. This protracted mode of action is similar to our previous results obtained with LCM as a stimulus (40).
We next tested whether MRTF might be involved in osmoprotection under these conditions. Inhibition (by CCG-1423) or downregulation (by siRNAs) of MRTF did not seem to cause gross morphological changes under isotonic conditions (Fig. 6, C and D). Dramatically different observations were made after a hypertonic challenge. In the control monolayers, 6 h of exposure to hyperosmolarity provoked only occasional rounding or detachment, seen only in a small fraction of cells; in contrast, hypertonicity caused major morphological alterations in MRTF-inhibited or -depleted cells, characterized by the acquisition of elongated and irregular shapes, formation of dendrite-like processes or peripheral blebs and shriveled appearance, i.e., often the complete loss of the cobblestone-like epithelial character. These findings suggest that MRTF expression and activity are critical for withstanding osmotic stress.
We also investigated the effect of longer-lasting osmotic exposure on transcription. Interestingly, after a 24-h hypertonic treatment, there was a moderate but highly significant decrease in the CArG-dependent luciferase activity (Fig. 6E). Similar observations were made using another reporter, a 765-bp segment of the SMA promoter, which contains two CArG boxes among other cis-elements (40). These findings suggest that the MRTF-dependent transcriptional effect is relatively short-term in hypertonically challenged tubular cells, and it is followed by a reversal, a phenomenon that may be related to the destabilization of SRF under these conditions.
Finally, we sought to obtain insight into the mechanism underlying the downregulation of MRTF under strongly hyperosmotic conditions. Exposure to 200 mM extra NaCl for 2 h caused a substantial drop both in endogenous MRTF and in the heterologously expressed, HA-tagged MRTF, as verified by Western blotting using anti-MRTF and anti-HA antibodies (Fig. 7, A–C). The latter observation is informative in that it excludes the possibility that epitope masking (due to some posttranslational modification) is the main (or exclusive) reason for the observed drop in the MRTF signal. Furthermore, these results also suggest that an inhibition of MRTF transcription is unlikely to be the major cause of the decreased expression either, since the expression of HA-MRTF is driven by the artificial pCMV promoter. Instead, these findings together with the fast decrease in MRTF protein are consistent with enhanced MRTF degradation under highly hyperosmotic conditions. Since MRTF degradation was concomitant with caspase-3 activation, we considered that apoptosis-inducing caspases might be involved in this process. The pan-caspase inhibitor zVAD-fmk caused a strong (albeit not complete) reduction in caspase-3 activation but failed to prevent MRTF degradation (Fig. 7, A–C). zVAD-fmk slightly elevated the steady-state level of MRTF under isotonic conditions (Fig. 7B, top); however, the hypertonicity-induced relative (%) decrease in the endogenous MRTF level was unchanged (Fig. 7B, bottom). High concentration of zVAD-fmk seemed to exert a marginal protective effect with respect to the degradation of the heterologously expressed HA-MRTF. We then asked whether hyperosmolarity might trigger proteasomal MRTF degradation. To assess this possibility, cells were pretreated with the proteasome inhibitor MG132. This compound prevented the osmotically induced decrease in both the endogenous and the heterologously expressed MRTF protein (Fig. 7, A and C). Taken together, the data indicate that strong osmolarity destabilizes MRTF by facilitating its proteasomal degradation.
Cytoskeletal remodeling is an immediate response to osmotic stress, which helps the cell to withstand the ensuing mechanical trauma (17, 36). Our current studies show that hyperosmotically induced, cytoskeleton-regulating signaling pathways and the consequent cytoskeletal changes themselves are not only responsible for the acute structural adaptation, but they also mobilize transcription factors that can impact the expression of cytoskeletal genes. Our recent studies have shown that SRF is phosphorylated and activated upon hyperosmotic stimulation in ELA cells (24). However, SRF is a dual-function transcription factor that can drive both proliferation/survival-promoting early genes and cytoskeleton/muscle differentiation-specific genes (41). Moreover, these two functional modalities were found to be competitive (toggle-switch mechanism), and the selection or ratio between them is governed by the interaction of SRF with distinct transcriptional coactivators, namely with components of the ternary complex (for proliferation) and MRTF (for cytoskeletal control) (7, 41). Therefore we sought to determine whether hyperosmolarity can directly impact the cytoskeletal arm, i.e., MRTF, the activity of which is predominantly regulated through its localization. Our results show that MRTF is an osmosensitive molecule that is rapidly translocated into the nucleus upon hypertonic treatment in a RhoA- and ROK-dependent manner. Consistent with the possibility that the RhoA/ROK pathway acts primarily by inducing net F-actin polymerization, which is a key regulator of MRTF localization, we have shown earlier that the activation of ROK is an important contributor to the osmotically induced rise in F-actin, presumably because ROK mediates cofilin phosphorylation, which in turn reduces the F-actin severing activity of this protein (66). In addition, ROK (directly or indirectly) might induce MRTF phosphorylation as well. This possibility stems from the observations that MRTF was shown to undergo RhoA-dependent phosphorylation (43) and Y-27632 prevents both the stimulus-induced shift in the molecular mass of MRTF (61) and its concomitant translocation. Nonetheless, the role of this phosphorylation in the transport or activity of MRTF remains to be clarified. Finally, cell contractility [increased myosin light chain (MLC) phosphorylation] has also been shown to potentiate MRTF accumulation (21). Since hyperosmolarity provokes RhoA/ROK-dependent MLC phosphorylation in tubular cells (18), this mechanism may also facilitate nuclear MRTF accumulation under hyperosmolar conditions.
Our studies also provide insight into the hitherto unknown upstream mechanisms responsible for hypertonicity-induced RhoA activation. Based on our findings that hyperosmotic stress activates the RhoA exchange factor GEF-H1 and that GEF-H1 downregulation reduces the osmotically provoked RhoA activation and MRTF translocation, we conclude that GEF-H1 is an osmotically sensitive signal transducer and the GEF-H1/RhoA/ROK pathway is a major mediator of the hypertonicity-triggered MRTF translocation. This conclusion is in accord with a recent report showing that GEF-H1 can regulate SRF-dependent transcription (SMA expression) in TGF-β-stimulated retinal pigment cells (67). Furthermore, GEF-H1 has been recently shown to be activated by stretch (6), extracellular matrix stiffening (27), and depolarization (71), which, together with its osmosensitivity documented herein, implies that this molecule is a key mechanotransducer coupling physical changes to RhoA activation. However, GEF-H1 silencing did not completely eliminate RhoA activation and MRTF translocation. While this could be due to some residual GEF-H1 activity, it is likely that GEF-H1 is not the only link between osmotic stress and RhoA. In this regard previous studies have revealed that ezrin is activated by hyperosmotic stress and its downregulation mitigates the shrinkage-induced RhoA activation in ELA cells (56). Since active ezrin counteracts the RhoA-sequestering capacity of Rho-GDI (65), this mechanism may represent a significant permissive input. Another intriguing possibility comes from the elegant studies of Guilley et al. (25), who showed that integrin-mediated force transduction activates two GEFs, GEF-H1 and leukemia-associated Rho GEF (LARG), in parallel and each of these is responsible for approximately half of the ensuing RhoA activation. LARG was activated via integrin-mediated stimulation of the Src-family kinase Fyn. These findings point to LARG as a candidate in the regulation of osmotic RhoA activation as well, because integrins were proposed to transmit cell volume-dependent signals (60) and previous studies by us (34, 35) and others (11) have shown that hyperosmotic stress selectively activates Fyn (and in certain cell types Yes) but not p60 Src, the third ubiquitous member of the family.
Although the investigation of the upstream mechanisms that connect the osmotic insult (or other mechanical stimuli) to GEF-H1 activation is beyond the scope of the current work, considering some potential mechanisms may facilitate future studies in this direction. GEF-H1 can be stimulated by its release from microtubules or the tight junctions (1, 12) [either due to microtubule disassembly or possibly due to GEF-H1 phosphorylation (10, 73)] and by enhancing its intrinsic activity (again via phosphorylation). A variety of kinases (FAK, ERK, PAK family members, etc.) have been implicated in GEF-H1 regulation (22, 25, 32, 73), and several of these are also affected by osmotic shock (see Ref. 29). Future studies are warranted to test their involvement.
The discrepancies between the magnitude of RhoA activation and MRTF translocation by various stimuli prompted us to consider the role of additional regulatory inputs. We found that inhibition of p38 facilitated the nuclear accumulation of MRTF, an effect that may be due to p38-mediated phosphorylation of MRTF. This possibility is supported by the facts that MRTF contains potential p38 target sites (some of which are ERK targets as well), and ERK-mediated MRTF phosphorylation was shown to increase MRTF's affinity for actin, which in turn enhances nuclear export and may inhibit import (46). Irrespective of the exact mechanism, the observed effect was selective for p38 inhibition since pharmacological suppression of the ERK pathway did not detectably influence the hypertonicity-triggered MRTF accumulation (data not shown). This observation is consistent with the fact that hyperosmolarity is a weak and transient stimulus for ERK in many cells, while in others it actually inhibits this kinase (Ref. 24; see also Ref. 29). Furthermore, this finding also implies that ERK, which has been shown to mediate GEF-H1 activation in response to various stimuli (25, 32, 71), is unlikely to fulfill such a role with regard to hyperosmolarity (present study) or membrane stiffness (27). While p38 activity mitigates MRTF's nuclear accumulation during osmotic stress, its effect on MRTF transport is stimulus dependent and its overall impact on SRF-driven transcription is complex. Specifically, LCM-triggered MRTF translocation is suppressed by p38 inhibition, indicating that p38 is needed for upstream signaling steps induced by contact disruption (61). More importantly, SRF activity itself can be enhanced by p38, as indicated by the observation that p38 inhibition significantly reduced the hypertonicity-provoked activation of the SRF reporter in ELA cells (24). The likely mechanism underlying this finding is that p38 activates MAPKAP kinase-2, which directly phosphorylates and activates SRF (28, 58). This scenario raises the interesting possibility that p38 might modulate SRF activity during osmotic stress by enhancing it towards early genes (ternary factors) but mitigating it towards the cytoskeletal (MRTF-dependent) genes. A further level of complexity is that p38 has four isoforms, of which hyperosmolarity activates p38α and p38δ, both of which are inhibited by SB203580 but exert opposite effects on the activity of TonEBP (74). Clearly, the potential isoform-specificity as well as the exact mechanism(s) whereby p38 fine-tunes the activity of MRTF and SRF await elucidation.
Our studies also revealed the functional relevance of the osmotically induced MRTF translocation. We showed that pharmacological or genetic interference with MRTF suppresses the hyperosmolarity-induced increase in CArG-dependent transcription. Moreover, elimination or inhibition of MRTF increased the susceptibility of the cells to hyperosmotic damage, which manifested in gross morphological alterations at osmolarities that did not bring about major structural changes in control cells. Shortly before the submission of this work, Nie et al. (47) reported that downregulation of GEF-H1 augmented hyperosmolarity-induced lactate dehydrogenase release (necrotic damage) in MDCK cells, a finding consistent with the proposed osmoprotective role of the RhoA/GEF-H1/MRTF pathway. Interestingly, in various systems, MRTF has been reported to act either as an inhibitor or as a potential inducer of apoptosis. Specifically, MRTF overexpression protected cells against TNF-α- or hypoxia-induced apoptosis (27, 59), and the SRF/MRTF inhibitor CCG-1423 promoted apoptosis in melanoma cells (20). In contrast, activated MRTF was reported to drive the expression of proapoptotic Bcl family members (62). In light of these findings we sought to investigate whether MRTF could alter the level of osmotically provoked apoptosis. However, an unexpected phenomenon precluded the assessment of this question: we found that at apoptosis-inducing osmolarities, MRTF itself is rapidly degraded. Since osmotic stress activates caspases and SRF has been reported to be a target of caspase-3 (5, 19), it was conceivable that the decrease in MRTF protein might be a caspase-mediated process. However, the inability of the pan-caspase inhibitor zVAD-fmk to prevent MRTF degradation strongly argues against this possibility. Nonetheless, contribution of caspase activation cannot be ruled out since caspase-3 cleavage was not completely abolished by the inhibitor. Importantly, the efficient reduction in MRTF degradation by MG132 indicates that the proteasomal pathway is a main mechanism in the osmotically induced MRTF degradation. As mentioned, high osmolarities destabilize SRF as well. Since osmotic stress has been shown to induce SRF phosphorylation (24), the initial up-shift in the molecular mass of SRF observed after hypertonic treatment may reflect such modification. The osmotically induced phosphorylation may prime SRF for subsequent degradation, a concept that remains to be assessed in future studies. Overall, these observations suggest that while MRTF is necessary for adaptation to moderate osmotic stress, its activity and expression are very tightly controlled under hypertonic conditions. Accordingly, its translocation is transient and relatively modest, its RhoA-dependent nuclear accumulation is mitigated by p38-dependent and -independent mechanisms, and at higher osmolarities its stability (and that of SRF) is compromised. In keeping with this, the ensuing transcriptional effects are moderate and short term, present at 6 but not 24 h after stimulation. These observations indicate that under strongly hyperosmotic conditions, the “preferable” cell fate is death over epithelial-mesenchymal transition (EMT). Indeed, we never detected SMA expression, the hallmark of the myofibroblast phenotype, after stimulation with hyperosmolarity either alone or in combination with TGF-β (not shown). This is in sharp contrast with the combined effect of LCM and TGF-β, which induce robust, MRTF-dependent SMA expression and epithelial-myofibroblast transition although none of these stimuli can induce SMA expression in itself in our cells (39, 40). The above-mentioned MRTF-inhibiting mechanisms can ensure that under hyperosmotic conditions the activation of RhoA, which is necessary for cytoskeletal rearrangement and the maintenance of functional integrity, can be uncoupled from EMT induction.
Our experiments were performed on LLC-PK1 proximal tubule cells, because our earlier studies have thoroughly characterized both the hypertonicity-induced changes in the activity of Rho GTPases (16, 66) and MRTF/SRF-dependent EMyT (14, 21, 39, 40) in this cell type. However, proximal tubule cells do not reside in a hyperosmotic milieu in vivo. Thus while these cells are appropriate for studying the effect of acute osmostress on the epithelium, the role and behavior of MRTF should also be investigated in cells originating from the distal nephron, or in cells that have been adapted to chronic hyperosmolarity. Such future studies will show whether MRTF can be preserved in adapted cells and whether MRTF signaling plays a role in the regulation of the cytoskeleton and integrity in cells functioning in a chronically hyperosmotic environment.
In summary, we have identified the GEF-H1/RhoA/ROK/MRTF pathway as a tightly controlled osmosensitive and osmoprotective mechanism that provides a link between the osmotic environment and the transcriptional control of the cytoskeleton.
This work was supported by grants from the National Science and Engineering Research Council (NSERC), the Canadian Institutes of Health Research (CIHR, MOP-86535 and MOP-106625), and the Kidney Foundation of Canada to A. Kapus and CIHR grant to K. Szászi (MOP-97774). F. Waheed is supported by a Li Ka Shing fellowship and K. Szászi is the recipient of an Early Researcher Award from the Ontario Ministry of Innovation and Research.
No conflicts of interest, financial or otherwise, are declared by the author(s).
D.L.L., F.W., M.L., P.S., M.H., and A.K. performed the experiments; D.L.L., F.W., M.L., and A.K. analyzed the data; F.W., M.L., P.S., A.M., H.N., S.F.P., K.S., and A.K. approved the final version of the manuscript; M.L., K.S., and A.K. prepared the figures; M.L., S.F.P., K.S., and A.K. edited and revised the manuscript; A.M., S.F.P., K.S., and A.K. conception and design of the research; H.N. and A.K. interpreted the results of the experiments; A.K. drafted the manuscript.
- Copyright © 2013 the American Physiological Society