In addition to soluble factors, mechanical constraints and extracellular matrix stiffness are important regulators of cell fate that are mediated by cytoskeletal modifications. The EMT (epithelial–mesenchymal transition) that occurs during normal development and malignant progression is a typical example of the phenotypic switch associated with profound actin remodelling and changes in gene expression. For instance, actin dynamics control motile cell functions in EMT, in part, through regulating the subcellular localization of the myocardin-related transcription factor MKL1 (megakaryoblastic leukaemia translocation 1), a co-activator of SRF (serum-responsive factor). In the present paper, we show that MKL1 participates also to the control of the cellular switch between growth and quiescence. Experimental disconnection between MKL1 and G-actin (globular actin), by using an MKL1 mutant or enhancing the F (filamentous)-/G-actin ratio, generates a widely open chromatin state and a global increase in biosynthetic activity, classically associated with cell growth. Conversely, G-actin accumulation favours nuclear condensation and cell quiescence. These large-scale chromatin changes rely upon extensive histone modifications, exemplified by that of H3K9 (H3 Lys9) shifting from trimethylation, a heterochromatin mark, to acetylation, a mark of euchromatin. The present study provides the first evidence for a global reversible hetero/euchromatinization phenomenon triggered by the actin/MKL1 signalling pathway.
The most dramatic step in cancer progression is the individualization, from primary tumours, of some cells with dedifferentiated mesenchymal phenotypes and which are capable of disseminating throughout the organism. Once individualized after loss of cell–cell contacts and disruption of E-cadherin, the cells committed to stemness and invasiveness acquire typical architectures supposed to be governed by specific gene regulations . However, reciprocal relationships have also been shown between mechanical constraints exogenously imposed to the cell and genetic reprogramming [2,3]. The genetic effects of mechanical tensions generated by extracellular matrix or cell–cell contacts are transmitted through the cytoskeleton. Experimental cell shape alterations are sufficient to modify cell fate and several links have been proposed to explain how gene expression can sense cellular structure. For example, disruption of E-cadherin contacts leads to the release of components of the β-catenin and NF-κB (nuclear factor κB) pathways . Cellular constraints can also activate the co-activator TAZ (tafazzin), which is overexpressed in breast cancer [5,6]. Furthermore, transcriptional regulators, such as MKL1 (megakaryoblastic leukaemia translocation 1), can directly sense the degree of actin polymerization [7–9].
MKL1, also known as MRTF-A (myocardin-related transcription factor-A), MAL (megakaryocytic acute leukaemia) and BSAC (basic, SAP and coiled–coil), belongs to the family of the myocardin-related transcription factors, whose members control motile or contractile cell functions during vascular smooth muscle cell and cardiac myocyte differentiation, neurite outgrowth, neuronal migration or cancer progression, and metastasis [10,11]. MKL1 has a strong transcription activation domain, but no clear DNA-binding domain [12,13], so that it is supposed to mainly work as a transcriptional co-activator through binding to DNA-bound factors. The currently admitted mode of action of MKL1 in relation with actin organization can be summarized as follows. MKL1 can be sequestered by free actin monomers [G-actin (globular actin)] in the cytoplasm, preventing it from translocating into the nucleus. Upon actin polymerization in the form of F-actin (filamentous actin), MKL1 enters the nucleus where it associates with DNA-bound SRF (serum-responsive factors) and strongly activates transcription at these loci [12,14]. MKL1 can also bind to G-actin in the nucleus, leading to its nuclear export and inactivation of SRF target genes. In this scenario, the ratio of polymeric over monomeric actin in the cell is determinant for the action of MKL1 [12,15]. MKL1 is a master link in the circular relationship between gene regulatory networks and cellular shape since, on the one hand, as described above, it can sense the degree of actin polymerization and, on the other hand, it governs the expression of a collection of gene products regulating F-actin formation .
Beside these established functions, we have shown previously that the Rho/actin/MKL1 signalling pathway is also a main actor in oestrogenic signalling, by controlling the cell-specific activity of both transactivation functions of the ERα (oestrogen receptor α) . The transcriptional activity of ERα is mediated through two transactivation functions, called AF-1 (activation function-1) and AF-2, whose respective involvement varies in a cell differentiation stage-dependent manner . We demonstrated that when MKL1 is sequestered in an inactive form by G-actin, the transcriptional activity of ERα is high and mainly relies on AF-1. Conversely, activation of MKL1 silences AF-1 activity, thus dramatically reducing the transactivation efficiency of ERα, then acting exclusively through AF-2 . During that study, we noticed that the basal activity of ERE (oestrogen-responsive element)-less and CarG-less reporter genes were also affected by MKL1 activity .
In the present study, complementary approaches allowed us to decipher the multifaceted influences of MKL1 and to show that it has the dual capacity to either prevent or favour global euchromatization mediated by major epigenomic changes, and thereby controls cellular growth. This study expands the implication of MKL1 far beyond its previously identified roles.
MATERIALS AND METHODS
Plasmids, siRNAs and antibodies
pcDNA6/TR and pcDNA4/TO plasmids (T-Rex™ system) were purchased from Invitrogen. MKL1 pcDNA4/TO expression vectors were generated from the p3Xflag-MKL1, p3Xflag-MKL1ΔN200 and p3Xflag-MKL1ΔC301 expression vectors kindly provided by Professor R. Prywes (Colombia University, New York, NY, U.S.A.) . The ERE-tk-LUC, C3-LUC, c-fos-LUC, pTAL-LUC, CMV-βgal reporter genes and the pCR-ERα, pCR MKL1, pCR MKL1ΔN200 and pCR MKL1ΔC301 expression vectors have been described previously . The MMTV-LUC, GRE-tk-LUC, pSG GR and pSG AR were a gift from Professor Bernadette Ducouret (University of Rennes I, Rennes, France). Actin WT (wild-type), R62D, V159N and S14C expression vectors were kindly provided by Professor M.K. Vartiainen (University of Helsinki, Helsinki, Finland). Human MKL1 (ON-TARGETplus® SMARTpool®) was purchased from Thermo Scientific. Human SRF siRNAs (Mission esiRNA) were obtained from Sigma–Aldrich. The negative universal control siRNA was purchased from Invitrogen. The antibodies used were: anti-E-cadherin (ab15148; Abcam), anti-vimentin (V9; Sigma–Aldrich), anti-(α smooth muscle actin) (ab5694; Abcam), anti-ERα (HC20; Santa Cruz Biotechnology), anti-MKL1 (ab113264; Abcam), anti-FLAG (M2; Sigma–Aldrich), anti-PCNA (proliferating-cell nuclear antigen; M0879; Dako), anti-c-fos (4; Santa Cruz Biotechnology), anti-c-Myc (C-33; Santa Cruz Biotechnology), anti-Rb (retinoblastoma; ab6075, Abcam), anti-ERK1 (extracellular-signal-regulated kinase 1; K-23, Santa Cruz Biotechnology), anti-SRF (G-20; Santa Cruz Biotechnology), anti-β-actin (AC-15; Santa Cruz Biotechnology), anti-H3K9ac (histone H3 acetylated at Lys9; ab10812; Abcam), anti-H3K9me1 (histone H3 methylated at Lys9; ABE101, Millipore), anti-H3K9me3 (histone H3 trimethylated at Lys9; ab8898, Abcam), anti-H3K4me3 (histone H3 trimethylated at Lys4; MC315; Millipore), anti-H3K27me3 (histone H3 trimethylated at Lys27; 07-449; Millipore), anti-(histone H3) (E173-58; Epitomics), anti-HP1 (heterochromatin protein 1; FL-191; Sigma–Aldrich) and dye-conjugated secondary antibodies (Alexa Fluor™; Invitrogen).
Establishment of MCF7 T-REx™ subclones
Stably transfected MCF7 subclones, MCF7-control, MCF7-MKL1 WT, MCF7-MKL1ΔN200 and MCF7-MKL1ΔC301, were obtained by transfecting MCF7 cells with pcDNA6/TR plasmid and corresponding pcDNA4/TO expression vectors (T-REx™ system) with jetPEI reagent (Polyplus Transfection), and selection with 5 μg/ml blasticidin and 100 μg/ml zeocin (Invitrogen). Expression of the proteins of interest was induced by a 48 h treatment of the MCF7 subclones with tetracyclin.
Cell culture and transfection
HepG2, HeLa, MCF7, MDA-MB 231 and MCF7 T REx™ subclones were maintained in DMEM (Dulbecco's modified Eagle's medium; Invitrogen) supplemented with 10% FBS (Biowest) and antibiotics (Invitrogen) at 37°C in 5% CO2. At 1 h before transfection the medium was replaced with Phenol Red-free DMEM (Invitrogen) containing 2.5% charcoal-stripped FBS (Biowest). Plasmid and siRNA transfections were carried out using respectively jetPEI and Lipofectamine™ RNAiMAX (Invitrogen) reagents according to the manufacturer's instructions. At 1 day before transfection, cells were plated in 24-well plates at 30% confluence. At 1 h before transfection the medium was replaced with Phenol Red-free DMEM (Invitrogen) containing 2.5% charcoal-stripped FBS. For the luciferase assays, transfection was carried out with 100 ng of reporter genes, 100 ng of CMV-βGal internal control and appropriate combinations of expression vectors. Plasmid mix was made up to 500 ng of total DNA per well with empty vector. Following an overnight incubation, cells were treated for 24 h when required with ligands (oestradiol, dexamethasone or testosterone) or ethanol (vehicle control). Cells were then harvested and luciferase and β-galactosidase assays were performed as described previously . Luciferase reporter gene activity was normalized to the β-galactosidase. For siRNA transfection, 20 pmol of siRNA per well was used. As determined by Western blotting, the down-regulation efficiency was at least 70% for each protein of interest. For actin studies, transfection was carried out with 100 ng of actin WT, R62D, V159N and S14C expression vectors. The transfection of actin expression vectors was performed 24 h after the siRNA transfection.
Flow cytometry analysis
Cells were grown in 10-cm-diameter dishes in the presence of DMEM containing 2.5% charcoal-stripped FBS. Following trypsinization, cells were collected in PBS containing 30% IFA buffer [10 mM Hepes (pH 7.4), 150 mM NaCl, 4% FBS and 0.1% NaN3], pelleted at 1000 rev./min for 10 min and fixed in 70% ethanol as described previously . Fixed cells were incubated in IFA buffer containing 100 μg/ml RNase A for 15 min at 37°C and 25 μg/ml propidium iodide was added before analysis with FACScan equipment (Beckton Dickinson).
Cells were grown on 10-mm-diameter coverslips in 24-well plates in the presence of DMEM containing 2.5% charcoal-stripped FBS. Cells were fixed with PBS/4% PFA (paraformaldehyde) for 10 min and then permeabilized in PBS/0.3% Triton X-100 for 10 min. Incubation with the primary antibody (1:1000 dilution) was performed overnight at 4°C. Dye-conjugated secondary antibodies were incubated for 1 h at room temperature. After mounting in Vestashield® mounting medium with DAPI (Vector), images were obtained with an Imager.Z1 ApoTome AxioCam (Zeiss) epifluorescent microscope and processed with Axio Vision Software. Fluorescence was quantified with ImageJ software (http://imagej.nih.gov/ij/) from images obtained with identical time exposures. Immunofluorescence was scored for at least 20 cells by image on n images obtained from different experiments. The means obtained for every image were then averaged.
Cells were lysed in RIPA buffer [50 mM Tris/HCl (pH 7.5), 150 mM NaCl, 1% Nonidet P40, 0.5% sodium deoxycholate and 0.1% SDS] containing a cocktail of protease inhibitors (Roche). Western blots were performed as described previously .
Protein, RNA and DNA dosage
Dosages were based on an acid base hydrolysis assay. Briefly, cells were grown in 10-cm-diameter dishes, trypsinized and washed twice in PBS. Macromolecules were precipitated in 0.3 M HClO4 and then solubilized in 0.6 M KOH. RNA hydrolysis was performed at 37°C for 1 h. DNA and proteins were then precipitated by the addition of 11.6 M HClO4 and hydrolysed RNA was quantified in the supernatant. The pellet was resuspended in 0.6 M HClO4, warmed at 80°C for 10 min and then cooled down to 0°C for 1 h. After centrifugation, the supernatant and pellet contained hydrolysed DNA and proteins respectively.
In vivo run-on
In vivo run-on transcription was performed as described previously . Briefly, cells were grown in 24-well plates in the presence of DMEM containing 2.5% charcoal-stripped FBS. For BrUTP (5-bromouridine 5′-triphosphate) administration, cells were incubated over 15 min at 4°C with a mixture composed of BrUTP and FuGENE® 6 (Roche). Cells were then washed twice with PBS and maintained for 30 min at 37°C in culture medium before PFA fixation. Incorporated BrUTP was detected by immunofluorescence with an anti-BrUTP antibody (Roche). Fluorescence was quantified with ImageJ software from images obtained with identical time exposures.
MNase accessibility assay
Adherent cells were treated with lysolecithin solution [0.025% lysolecithin, 150 mM sucrose, 80 mM KCl, 35 mM Hepes (pH 7.4), 5 mM K2HPO4 and 5 mM MgCl2] for 2 min at room temperature. MNase digestion of permeabilized cells was performed at room temperature for 5 min using 0–50 units of MNase (micrococcal nuclease) diluted in buffer [150 mM sucrose, 50 mM Tris/HCl (pH 7.5), 50 mM NaCl and 2 mM CaCl2]. The reaction was stopped with stop solution [20 mM Tris/HCl (pH 8), 20 mM NaCl, 20 mM EDTA and 1% SDS]. After extraction, genomic DNA was analysed by electrophoresis on an agarose gel stained with ethidium bromide.
Significance of the treatments was determined using ANOVA followed by a Fisher's post-hoc test using the Statview 5.0 software (SAS Institute). In all these experiments, data resulted from n separate experiments and are expressed as means±S.E.M.
MKL1 regulates the basal transcriptional activity of CarG-less reporter genes and the transactivation efficiency of nuclear receptors
As a co-activator of SRF, MKL1 stimulates the expression of several genes carrying a serum-responsive element (CarG/SRE), including immediate early, cytoskeletal and muscle-specific genes . We verified this stimulatory effect of MKL1 in our transient expression assays using the c-fos-LUC reporter construct. In HepG2 cells, expression of a constitutively active mutant form of MKL1 (MKL1 ΔN200) deleted from the N-terminal RPEL motifs (Figure 1A) and, to a lesser extent WT MKL1, enhanced the basal activity of the reporter gene (Figure 1B). Conversely, a dominant-negative MKL1 ΔC301, with its C-terminal transactivation domain deleted, had no effect as expected. But, very surprisingly, similar results were also observed when using CarG-less reporter genes with completely different plasmid backbones, including MMTV-LUC, pTAL-LUC, ERE-tk-LUC and others (Figures 1B and 1C). Interestingly, MKL1-induced changes in the basal activity of reporter genes interfere with the transcriptional activity of ligand-inducible transcription factors, such as the GR (glucocorticoid receptor) and the AR (androgen receptor), and, as observed previously , the ERα. Fold induction or the transactivation efficiency of these nuclear receptors was dramatically reduced by transfection of MKL1 ΔN200 (Figure 1C), whereas, inversely, MKL1 ΔC301 enhanced the transactivation efficiency of these receptors, suggesting that the transcriptional effects of MKL1 are more general than defined previously.
MKL1 regulates the basal transcriptional activity of CarG-less reporter genes and the transactivation efficiency of steroid receptors
MKL1 controls global transcriptional activity and cell growth
To investigate the mechanisms involved in the general transcriptional effects of MKL1, MCF7 subclones expressing WT or mutant forms of MKL1 were constructed using the tetracyclin-inducible vector system (T-Rex™). The functionality of these constructs was first tested by Western blot analysis and immunocytology (Figures 2A and 2B). The proportion of cells with a nuclear localization of the flagged MKL1 WT and MKL1 ΔC301 was similar to that found for endogenous MKL1 (Figures 2B and 2C), with approximately 8% of the cells exhibiting a nuclear localization of MKL1 proteins. In contrast, MKL1 ΔN200 was detected in the nucleus of at least 80% of the cells, but this location was expected because the actin-interacting domain, deleted from this construct, is also necessary for the nuclear export of MKL1 [12,14]. In order to control the functionality of MKL1 mutants produced by the tetracyclin-inducible vector system, the activity of MKL1 proteins was measured on the CarG-driven c-fos-LUC reporter gene (Figure 2D) and their capacity to stimulate transcription of a CarG-less reporter gene was also monitored from transfected ERE-driven C3-LUC reporter plasmid (Figure 2D).
MKL1 impacts global transcriptional activity
The potent transcriptional effects of MKL1 were further analysed by measuring the global transcriptional activity of the different MCF7 T-Rex™ subclones, through in vivo run-on experiments. As shown in Figure 2(E), MKL1 ΔN200 induced a more than 2-fold increase in global steady state transcriptional activity, whereas MKL1 ΔC301 slightly decreased it. In parallel, quantification of the total RNA/DNA ratio in the different MCF7 T-Rex™ subclones revealed a 2-fold increase in RNA (Figure 2F) and protein (Figure 2G) synthesis in MKL1 ΔN200-expressing MCF7 cells, which reflects a higher global transcriptional activity of these cells in line with the results of transient expression assays described above. Since such an intense RNA and protein synthesis is characteristic of cell growth, MCF7 T-Rex™ subclones were counted, but no obvious difference was observed between MCF7-control and MCF7-MKL1 WT and MCF7-MKL1 ΔN200 clones. In contrast, MKL1 ΔC301-expressing MCF7 cells proliferate more slowly (Figure 3A). Flow cytometry analysis was then performed to evaluate the relative proportion of cells in the different cell-cycle phases (Figure 3B, and Supplementary Figure S1 at http://www.biochemj.org/bj/461/bj4610257add.htm). Although similar profiles were obtained for MCF7-control and MCF7-MKL1 WT clones, MKL1 ΔN200 expression enhanced both the percentage of cells in S- (×1.4) and subG1- (apoptosis; ×3) phases. Conversely, a clear decrease in the percentage of cells in S-phase was observed in MCF7 cells expressing MKL1 ΔC301 (Figure 3B). To further assess the influence of MKL1 mutants on cell growth, we then analysed by immunofluorescence the expression of major actors of this cellular activity: c-Myc and Rb. As shown in Figure 3(C), expression of c-Myc was higher in MCF7-MKL1 ΔN200 and lower in MCF7-MKL1 ΔC301 compared with unmodified MCF7. Nuclear accumulation of Rb was noticeably enhanced in MCF7-MKL1 ΔC301 (Figure 3C), corroborating the low percentage of cells in S-phase of this clone.
MKL1 is involved in the control of the balance between cell growth and differentiation
Finally, the phenotype and differentiation status of MCF7 T-Rex™ subclones were examined through Western blotting and morphological immunocytology analyses (Figures 3E and 3F). As shown in Figure 3(F), induction of the MCF7-MKL1 ΔN200 transgene led to a striking increase in the size of cells with formation of stress fibres and lamellipodia. In addition, MKL1 ΔN200 dramatically disrupted the E-cadherin network without affecting its expression level (Figures 3D and 3F). It stimulated the expression α-smooth muscle actin and vimentin, two hallmarks of EMT (epithelial–mesenchymal transition). Interestingly, the MKL1 ΔN200 down-regulated ERα expression, whose abrogation is a well-established symptom of the commitment of hormone-dependent breast cancer cells towards EMT. In contrast, MKL1 ΔC301 accentuated the differentiated epithelial phenotype of MCF7 cells, illustrated by the E-cadherin network and ERα expression. Phenotype changes of MCF7 T-Rex™ subclones were accompanied by the corresponding changes in balance between G- and F-actin (Figure 3E).
MKL1 influences chromatin structure on a global scale
The profound transcriptional influences of MKL1 shown above prompted us to determine whether they could be mediated by generalized chromatin organization. DNA accessibility was compared between the different MCF7 T-Rex™ subclones using the MNase assay. As shown in Figure 4(A), although similar internucleosomal fragmentation profiles where obtained for MCF7-control and MCF7-MKL1 WT clones, clearly different patterns were observed for the two mutant forms of MKL1. An enrichment and inversely a depletion in low-molecular-mass fragments were obtained in cells expressing MKL1 ΔN200 and MKL1 ΔC301 respectively. Densitometric profiles of DNA fragments clearly show that the ratio between mono- over di-nucleosomal fragments is lower for MCF7-MKL1 ΔC301, intermediate for MCF7-MKL1 WT and higher for MCF7-MKL1 ΔN200 (Supplementary Figure S2 at http://www.biochemj.org/bj/461/bj4610257add.htm). Given that the relative representation of these DNA fragments is an index of DNA accessibility for MNase, which itself reflects the degree of chromatin condensation , these results strongly suggest that the stimulating and inhibitory action on transcription of MKL1 ΔN200 and MKL1 ΔC301 respectively are mediated, at least in part, through regulating chromatin structure.
MKL1 influences global chromatin structure
Since chromatin packaging is known to correlate with specific chromatin modifications, and particularly PTMs (post-translational modifications) of histones , we tested a series of H3 histone modifications classically associated to either euchromatin (H3K9ac, H3K9me1 and H3K4me3) or heterochromatin (H3K9me3 and H3K27me3). The most dramatic effect of MKL1 was observed at the level of lysine 9 (H3K9) (Figure 4B). A 3-fold increase in H3K9ac and H3K9me1 was measured in the presence of MKL1 ΔN200. Quite the opposite, the lowest level of H3K9ac and the highest level of H3K9me3 were obtained in MCF7 cells expressing MKL1 ΔC301. The increase in H3K9me3 in the MKL1 ΔC301 clone was concomitant with an increase in the amount of HP1 (Figure 4B), known to be recruited by this modification. It is interesting to remember that the ratio between mono- over di-nucleosomal fragments was the lowest in cells expressing MKL1 ΔC301, in agreement with a model of dinucleosomal organization of HP1-H3K9me3 heterochromatin through bridging adjacent nucleosomes by HP1 dimers (Supplementary Figure S2) . Singularly, H3K4me3 and H3K27me3 levels varied co-ordinately in the different cellular clones. Despite the fact that these marks are generally associated to opposite chromatin configurations, their coexistence has been associated to cellular stemness . Finally, total H3 remained equivalent in the different MCF7 T-Rex™ subclones, validating the observed changes in histone PTMs.
MKL1 is involved in the control by actin dynamics, of large-scale chromatin changes and cell proliferation
Besides using mutant forms of MKL1, we wanted to verify whether endogenous MKL1 could be involved in specific histone modifications. To this end, we manipulated its activity through actin organization by transfecting either WT or mutant forms of actin that can alter endogenous actin polymerization. Actin WT that increases the level of G-actin and the non-polymerizable actin R62D, have both been shown to repress MKL1 activity, whereas the polymerization-priming actin versions V159N and S14C strongly activate SRF target genes in the absence of external signals [15,25]. At 48 h after transfection with these different actin constructs, the expected changes in actin polymerization were first verified using fluorescent phalloidin and DNase I staining. As shown in Figure 5(A), WT and R62D actin decreased the global F-/G-actin ratio, whereas the mutant actins V159N and S14C enhanced it. Interestingly, most of the cells in the dishes were affected, suggesting that low amounts of transfected plasmids are sufficient to induce these changes and that actin dynamics could be, as other polymerization phenomena, very sensitive to disrupting molecules. The subcellular localization of MKL1 has been reported to depend on the degree of actin polymerization. Significant increase in the nuclear translocation of MKL1 was indeed observed following mutant V159N and S14C actin expression (Figure 5B). The impact of mutant actins on gene expression was also confirmed by the alterations of c-fos expression (Figure 5C). To verify whether the actin/MKL1 signalling pathway is involved in the control of cell proliferation, the percentage of cells in S-phase was determined through nuclear PCNA staining after transfection of actin variants and of MKL1 siRNA (Figure 5D). Immunocytology results show that actins favouring G-actin (WT and R62D) were unable to modify PCNA, whereas mutant actins favouring F-actin formation (V159N and S14C) led to a 1.5-fold increase in the percentage of cells in S-phase. Very interestingly, MKL1 siRNA, but not control siRNA, cancelled the proliferative effect of polymerization-promoting actin mutants, highlighting the involvement of MKL1 in the effect of actin dynamics on cell growth. The H3K9 PTM changes were then monitored by quantitative immunofluorence. As shown in Figure 5(E), actin mutants favouring F-actin formation, increased H3K9 acetylation, concomitant with a decrease in H3K9 trimethylation, two correlated hallmarks of open chromatin. An opposite profile, typical of heterochromatin, was induced by the actin forms decreasing the F-/G-actin ratio. Strikingly, these changes were totally abolished when cells were transfected with MKL1 siRNA (Figure 5E).
Actin dynamics control large-scale chromatin changes and cell proliferation through MKL1
Differential requirement of SRF in MKL1-mediated chromatin reprogramming
As MKL1 has been characterized as a co-activator of SRF , we examined the role of SRF in the control of cell proliferation and chromatin structure by the actin/MKL1 signalling pathway. Actin mutants R62D and V159N, known to influence the F-/G- actin ratio in opposite manners, were tested for their effect on two parameters: cell proliferation and H3K9 modification, in MCF7 cells in which SRF was knocked down. Nuclear PCNA staining revealed opposite effects of the different actin mutants on cell proliferation as expected. Knockdown of SRF by siRNA abolished the mitotic effect of actin V159N, but not the inhibitory effect of actin R62D on cell proliferation (Figure 6A). SRF knockdown enhanced both H3K9 acetylation and trimethylation (Figure 6B). Actin R62D reduced the stimulatory effect of SRF knockdown on H3K9 acetylation, whereas actin V159N was incapable of inhibiting the stimulation of H3K9 trimethylation. Together, these results demonstrate that SRF is necessary for mediating the effects of actin polymerization on both cell proliferation and the H3K9me3 demethylation, whereas SRF seems to be dispensable for the G-actin-dependent processes of inhibition of cell proliferation and H3K9 acetylation.
Role of SRF in the control of cell proliferation and chromatin structure by the actin/MKL1 signalling pathway
MKL1-mediated cell proliferation and bulk H3K9ac correlate with cytoskeletal reorganization during epithelial-like cell dedifferentiation
To verify whether the importance of the actin/MKL1 system in the control of cell growth shown above is generalizable to other cellular contexts we examined four cell lines, HepG2, MCF7, HeLa and MDA-MB 231, selected for their well-established differential status with respect to EMT and proliferation. As probed by Western blot analysis and immunocytology using E-cadherin and vimentin markers, HepG2 and MCF7 cells can be classified as epithelial-like, whereas HeLa and MDA-MB 231 cells have a less differentiated phenotype (Figures 7A and 7B). Consistent with these structural differences, kinetic cell counting showed that HeLa and MDA-MB 231 cells proliferate more vigorously than HepG2 and MCF7 cells. Furthermore, the difference in transactivation efficiency of the ERα, GR and AR observed in the four cell lines in transient transfection experiments (Figure 7C) was indicative of different differentiation stages of the cells  and distinct functional states of MKL1 in these cells according to our previous observations (Figure 1) . We wanted first to determine whether the four cell lines are associated with different actin dynamics. As shown in Figures 7(A) and 7(D), the more differentiated cells, HepG2 and MCF7, indeed contained higher amounts of G-actin, especially in the nucleus (40–100%). F-actin staining was equivalent in the four cell lines, but with more prominent stress fibres in HeLa and MDA-MB 231 cells, a phenomenon associated with the Rho/ROCK (Rho-associated kinase)-dependent invasion programme . Accordingly, MKL1 subcellular localization was correlated with the F-/G-actin ratio in these cells (Figures 7A and 7E). MKL1 was cytoplasmic in most HepG2 and MCF7 cells, whereas a nuclear translocation of MKL1 was observed in more than 80% of HeLa and MDA-MB-231 cells. Then, the cell lines were subjected or not to siRNA MKL1 knockdown and the percentage of cells in S-phase was determined through nuclear PCNA staining. HeLa and MDA-MB 231 cells displayed higher accumulation of nuclear PCNA (50–60%) than HepG2 and MCF7 cells (20–30%). MKL1 depletion (Figure 7B) induced a 40% reduction in the S-phase in HeLa and MDA-MB 231 cells (Figure 7F). These results show that the actin/MKL1 signalling pathway is specifically involved in the higher proliferation of the cells that underwent EMT. MKL1 depletion was also shown to impact vimentin expression in EMT-established cell lines (Figures 7B and 7G). E-cadherin expression and F-actin level were not really affected (results not shown). Finally, bulk H3K9ac and H3K9me3 were measured in HepG2, HeLa, MCF7 and MDA-MB 231 cell lines, previously shown to exhibit different MKL1 activities. The highest degree of H3K9 acetylation was found in the cells with an activated actin/MKL1 pathway (HeLa and MDA-MB 231). Consistently, this acetylation was reduced by MKL1 siRNA treatment only in these two cell lines (Figure 7H). The correlation between bulk H3K9me3 and MKL1 activity was less obvious. However, a slight increase in H3K9me3 level was observed in HeLa and MDA-MB 231 cells after MKL1 siRNA treatment.
Actin/MKL1 signalling pathway in cell lines with different differentiation status
Members of the MRTF family are critical co-activators involved in the regulation of vascular smooth muscle cell and cardiac myocyte differentiation, neurite outgrowth and neuronal migration through the transcriptional regulation of sets of SRF-target genes encoding contractile and cytoskeletal proteins [9,27,28]. In the context of epithelial cells, MKL1 was shown to be activated following the loss of E-cadherin-mediated cell contacts during EMT and is thus associated with tumour dedifferentiation and metastasis [27,28]. The pro-metastatic role of MKL1 is generally explained by the expression of genes involved in motile functions . However, we show that the influence of MKL1 on gene expression is not limited to SRF-target genes. In previous studies, we observed that MKL1 also influences oestrogen signalling in breast cancer cell lines by controlling ERα transcriptional activity . This observation is extended here to other steroid signalling as shown by MKL1-induced changes in the transactivation efficiency of steroid receptors on their respective target reporter genes. Interestingly, reduced transactivation by steroid receptors was correlated with an increased basal activity of the reporter genes, and vice versa. Therefore changes in steroid receptor transactivation activity result in fact from changes in the basal activity of reporter genes, irrespective of whether the reporter plasmids contain the CarG element or not. These results suggest that the transcriptional effects of MKL1 are more general than previously defined. This assumption is strikingly confirmed by in vivo run-on experiments and analysis of total RNA/DNA ratio in MCF7 subclones expressing WT or mutant forms of MKL1.
The broad effect of MKL1 on the biosynthetic cellular activity was associated with extensive chromatin modifications, leading to either its opening or condensation, depending on MKL1 activity. These opposite effects of MKL1 were also correlated with different cell phenotypes, from quiescent-differentiated to growing-dedifferentiated epithelial-like cells. The effect of MKL1 on euchromatin during dedifferentiation leads to a striking increase of biosynthetic cellular activity. Interestingly, this global action on gene expression is highly reminiscent of that long reported for Myc proteins, which are pleiotropic effectors of a wide range of signalling pathways involved in proliferation, differentiation and metabolic growth . Accordingly, numerous cellular activities are shared by both factors, including (i) the capacity to induce genes transcribed by all types of RNA polymerases I, II and III . (ii) the global reshaping of chromatin structure . and (iii) the induction of stem-like properties . Several additional stem-like phenomena have been evidenced in the present study. Both H3K27me3 and H3K4me3 marks were clearly increased in MCF7-MKL1 ΔN200 clone. The biological meaning of H3K27me3, generally associated to gene repression, is in fact ambiguous. Although H3K9 and H3K4 trimethylation are mutually exclusive, H3K27 and H3K4 trimethylation sometimes coexist, defining a so-called ‘bivalent’ chromatin state, found for example in the context of undifferentiated embryonic pluripotent stem cells . It is likely that their co-enrichment in MCF7-MKL1 ΔN200 cells reflects a de-differentiated status. The permissive modifications induced by MKL1 coincide with the observations of McDonald et al.  of a global reduction in the H3K9 heterochromatin mark during TGF-β (transforming growth factor-β)-induced EMT revealed by genome-scale mapping and are correlated with a striking accumulation of RNA and proteins in the cells. However, comparative transcriptomic and proteomic analyses of normal and oncogene-transformed cells have revealed that certain genes are repressed during EMT [35,36], as observed in the present study for E-cadherin and ERα in MCF7 cells. These transcriptional repressions appear somewhat contradictory with a generalized euchromatinization. This apparent discrepancy could be explained by the fact that the repression of certain genes in early dedifferentiation could be regulated in a very specific manner by specialized repressors in the context of open chromatin. This possibility would be particularly consistent with the intriguing number of transcriptional repressors reported during early EMT, such as Zeb1 (zinc finger E-box-binding homeobox 1), Snail1 (snail family zinc finger 1) or Twist (twist family bHLH transcription factor), shown for example to repress E-cadherin and ERα [37,38].
In agreement with the literature, the antagonistic roles of MKL1 were frequently associated with different nucleocytoplasmic distributions of MKL1. MKL1-mediated cell proliferation and bulk H3K9ac were observed in cell lines exhibiting a nuclear accumulation of MKL1, whereas MKL1-mediated cell quiescence and bulk H3K9me3 were correlated with a cytoplasmic localization of MKL1. However, these correlations were not systematic, despite the dependency of the observed phenomena with respect to MKL1. Cell support, serum or antibody used could make amounts or changes in nuclear MKL1 just below their ability to be detected. In addition, one cannot exclude that MKL1 could be more than a transcription factor, capable of controlling specific signalling pathways outside the nucleus. It can also ensure specific nuclear roles in the absence of massive nuclear accumulation. It is interesting to note that results obtained from ChIP experiments showed that MKL1 is associated with the promoter region of oestrogen-responsive genes in MCF-7 cells , whereas less than 10% of these cells exhibit a massive nuclear localization of MKL1 under conventional culture conditions. MKL1 continuously shuttles between the nucleus and the cytoplasm through a process controlled by the cellular pool of G-actin. Nuclear import of MKL1 is regulated by RhoA activity and actin dynamic, but nuclear G-actin and has also been demonstrated to facilitate nuclear export of MKL1 [14,39]. Although the localization and activity of MKL1 are both controlled by G-actin, the two events remain independent from each other.
The classical interaction partner of MKL1, SRF, is shown in the present study to participate in the control of large-scale H3K9 PTM. Knockdown of SRF enhances bulk acetylation and trimethylation of H3K9 at once. However, SRF appears to be partly implicated in the control of chromatin structure by actin dynamics. Indeed, it is involved in H3K9me3 demethylation induced by actin polymerization, but is dispensable in the G-actin-dependent processes of H3K9 deactylation. Concordantly, the role of SRF in the control of cell growth by actins dynamics was limited to the increase of the percentage of cells in S-phase induced by a high F-/G-actin ratio. Considering the classification of MKL1 as a co-activator of SRF, the global transcriptional effects of MKL1, shown in the present study to be impacting on total RNA synthesis, were unexpected as they seem discordant with the limited number of genes containing putative CarG elements in their proximal regulatory regions . In fact, CarG elements were shown diffusely distributed in the genome, sometimes far away from transcription start sites  and enhancer elements might regulate the association of distant genes through transcription factories . Furthermore, half of the approximately 200 identified SRF target genes encode proteins whose function is restricted to the field of actin dynamics and contractile mechanisms . Likewise, the expression plasmids used in our transient expression assays were also devoid of a clear CarG box, although they were clearly regulated by MKL1. It is possible that widespread permissive chromatin could increase in trans the expression level of episomic vectors, for example through changes in the concentration of diffusing factors. MKL1 might also directly or indirectly affect the folding of the chromatin-like structure adopted by plasmids in the nucleus. So, it is possible that MKL1 acts on chromatin structure through a mechanism which is independent of its direct interaction with SRF. As the other members of the MRTF family, MKL1 contains a SAP [SAF-A/B (scaffold attachment factor A/B)/acinus/PIAS (protein inhibitor of activated STAT)] domain, which is a putative DNA-binding motif involved in chromosomal organization . This module was first identified in the SAF-A/B that bond scaffold- or matrix-attachment regions, acinus involved in apoptotic chromatin condensation and the STAT (signal transducer and activator of transcription) inhibitor PIAS. A role of the SAP domain of MKL1 was recently described in the SRF-independent activation of tenascin-C transcription by mechanical stress . Therefore the potential implication of this domain in the control of chromatin structure by MKL1 should be investigated in the future.
In conclusion, MKL1 appears to integrate information from the cellular environment and actin treadmilling and, in turn, can modulate chromatin structure leading to stimulatory or repressive influences on the global transcriptional activity of the cell. Further studies are now awaited to obtain new insights into molecular mechanisms engaged by MKL1 and its partners to influence chromatin structure. MKL1 clearly exerts various and sometimes opposite biological effects as a tumour-promoting or tumour-suppressing protein, depending on the cellular context. MKL1 overexpression has an antiproliferative effect in various fibroblast and epithelial cell lines , whereas in other cell types depletion of MKL1 suppresses cell migration, cell proliferation and anchorage-independent cell growth . These concurrent conclusions might be reconciled by the present study showing that distinct functional states of MKL1 could force cells to either dedifferentiate or differentiate by controlling global chromatin structure.
Dulbecco’s modified Eagle’s medium
oestrogen receptor α
histone H3 acetylated at Lys9
histone H3 methylated at Lys9
histone H3 trimethylated at Lys4
histone H3 trimethylated at Lys9
histone H3 trimethylated at Lys27
heterochromatin protein 1
megakaryoblastic leukaemia translocation 1
myocardin-related transcription factor
proliferating-cell nuclear antigen
protein inhibitor of activated STAT
scaffold attachment factor A/B
signal transducer and activator of transcription
Gilles Flouriot contributed to the design and execution of the experiments, data analysis and writing the paper. Guillaume Huet and Florence Demay performed and analysed experiments. Denis Michel contributed to the data analysis and the writing of the paper with ideas and comments from Farzad Pakdel and Noureddine Boujrad.
We thank B. Ducouret, M.K. Vartiainen and R. Prywes for providing plasmids. We are grateful to Dr L. Kular and A. Le Bescont for helpful discussion.
This work was supported by the Ligue contre le cancer, Rennes Métropole, the CNRS (Centre National de la Recherche Scientifique), the INSERM (Institut National de la Santé et de la Recherche Médicale) and the University of RENNES 1.