Open Access

Nuclear transport of the serum response factor coactivator MRTF‐A is downregulated at tensional homeostasis

Karen M McGee, Maria K Vartiainen, Peng T Khaw, Richard Treisman, Maryse Bailly

Author Affiliations

  1. Karen M McGee1,
  2. Maria K Vartiainen2,
  3. Peng T Khaw3,
  4. Richard Treisman2 and
  5. Maryse Bailly*,1
  1. 1 Department of Cell Biology, UCL Institute of Ophthalmology, 11–43 Bath Street, London, EC1V 9EL, UK
  2. 2 Transcription Laboratory, Cancer Research UK, London Research Institute, Lincolns Inn Fields Laboratories, 44 Lincolns Inn Fields, London, WC2A 3PX, UK
  3. 3 Department of Ocular Biology and Therapeutics, NIHR Biomedical Research Centre for Ophthalmology at Moorfields Eye Hospital, UCL Institute of Ophthalmology, 11–43 Bath Street, London, EC1V 9EL, UK
  1. *Corresponding author. Tel: +44 (0) 20 7608 6825; Fax: +44 (0) 20 7608 4034; E-mail: m.bailly{at}
  • Present address: Institute of Biotechnology, University of Helsinki, Viikinaari 9, Helsinki 00014, Finland


The serum response factor (SRF) coactivator myocardin‐related transcription factor A (MAL/MKL1/MRTF‐A), the nuclear transport and activity of which is regulated by monomeric actin, has been implicated in tension‐based regulation of SRF‐mediated transcriptional activity. However, the mechanisms involved remain unclear. We used fibroblasts grown within collagen matrices to explore whether MRTF‐A transport is regulated by tissue tension. We show that MRTF‐A nuclear accumulation following stimulation with serum, actin drugs or acute mechanical stress is prevented within mechanically loaded, anchored matrices at tensional homeostasis. This is accompanied by a higher G/F actin ratio, defective nuclear import and increased cofilin expression. We propose that tension regulates MRTF‐A/SRF activity through cofilin‐mediated modulation of actin dynamics.


Mechanical sensing through attachment to the matrix regulates many cell features such as shape, motility, proliferation and differentiation (Discher et al, 2005; Ingber, 2006). In tissues, cells constantly readjust the balance between external (matrix) and internal (cytoskeleton) forces to maintain a preferred tension, a process known as tensional homeostasis (Brown et al, 1998; Paszek et al, 2005; Ingber, 2006). When this balance is perturbed, cell behaviour is modified, resulting in pathological outcomes. Tensional homeostasis and the response to mechanical stress are believed to be controlled by the actin cytoskeleton. However, the mechanisms by which actin dynamics are used to sense and respond to external tension, and the way in which this affects transcriptional regulation to induce changes in cell fate, are not known.

Serum response factor (SRF) is a ubiquitous transcription factor that controls growth‐factor‐regulated immediate early genes such as c‐fos and actin, as well as muscle‐specific genes, most of which are also activated by mechanical stress (Zhao et al, 2007; Olson & Nordheim, 2010). SRF was shown to be activated following mechanical stress, originally through its coactivator myocardin in cardiomyocytes (Liu & Olson, 2006), and more recently through the more‐ubiquitous megakaryocytic acute leukaemia protein (MAL/MRTF‐A; Somogyi & Rorth, 2004; Zhao et al, 2007; Chan et al, 2010; Gomez et al, 2010). MRTF‐A is a member of the myocardin‐related transcription factor family of SRF coactivators (MRTFs)—including MRTF‐A/MKL1/MAL and MRTF‐B/MKL2—the activity of which is regulated by the cellular concentration of monomeric actin (Miralles et al, 2003; Vartiainen et al, 2007). We show here that actin dynamics and MRTF‐A/SRF activation are central to the mechanism by which cells sense external tension and regulate tensional homeostasis.

Results And Discussion

Nuclear transport is prevented at tensional homeostasis

To investigate the way in which tissue tension regulates MRTF‐A/SRF activation, we analysed MRTF‐A transport in NIH3T3 cells stably expressing MRTF‐A–green fluorescent protein (GFP), plated on either a non‐compliant substrate—glass coverslips—or within high‐compliance free‐floating or anchored collagen matrices (Fig 1A). Free‐floating matrices are a tension‐free environment in which the cell pulling force is greater than matrix resistance, leading to gel contraction (Grinnell, 2003; Dahlmann‐Noor et al, 2007). In anchored gels, the matrix becomes mechanically loaded—that is, under tension—as the cells pull on the collagen until tensional homeostasis is reached (Grinnell, 2003; Marenzana et al, 2006). In resting fibroblasts, MRTF‐A transport is dominated by rapid nuclear export, maintaining a mostly cytoplasmic pool (Vartiainen et al, 2007). Accordingly, MRTF‐A–GFP was essentially cytoplasmic in all conditions in low serum (Fig 1A,B). Following stimulation with 15% serum, MRTF‐A rapidly accumulated in the nucleus in cells on coverslips and within free‐floating gels, but not in anchored gels (Fig 1A,B).

Figure 1.

Serum and drug‐induced nuclear accumulation of MRTF‐A is inhibited at tensional homeostasis. (A,B) Starved MRTF‐A–GFP‐expressing cells on glass coverslips or within gels were incubated with medium containing 15% FBS, 0.5 μM jasplakinolide (Jas) or 2.5 μM cytochalasin D (cytoD), and fixed and stained with rhodamine–phalloidin. (A) MRTF‐A–GFP cells before (0 min) and after (15 min) addition of serum. Scale bar, 20 μm. (B) Subcellular distribution of MRTF‐A–GFP. Shown is the mean±s.e.m. for MRTF‐A nuclear/cytoplasmic ratio from a pool of cells from at least two experiments (except for CytoD on coverslips, for which one representative experiment is shown). FBS, 30–50 cells per time point from three experiments; Jas, 31–51 cells per time point for coverslips and 13–24 for gels, each from two experiments; CytoD, 15–23 cells per time point (coverslips) and 18–38 cells from three experiments for the gels. Asterisk indicates difference from the value at time 0, P<0.05. (C,D) Starved MRTF‐A–GFP‐expressing cells within free‐floating gels of increasing collagen concentration (1.4–6 mg/ml) were left unstimulated (t0) or were stimulated with 15% FBS for 15 min (t15) and fixed and stained with rhodamine–phalloidin. (C) Percentage of cells with MRTF‐A predominantly in the nucleus (nuclear), evenly distributed between the nucleus and the cytoplasm (pancellular) or cytoplasmic. Scale bars show a total of 41–286 cells per condition pooled from three experiments. (D) Representative cells in gels of different collagen concentration following serum stimulation. Red, F‐actin; green, MRTF‐A–GFP. FBS, fetal bovine serum; GFP, green fluorescent protein; MRTF‐A, myocardin‐related transcription factor A.

MRTF‐A transport is regulated through a direct interaction with monomeric G‐actin (Vartiainen et al, 2007), and alterations in the G/F actin ratio are both necessary and sufficient for MRTF‐A nuclear accumulation and SRF activation in cells on coverslips (Miralles et al, 2003). Hence, the actin‐binding drugs cytochalasin D and jasplakinolide, which dissociate the MRTF‐A–actin complex, induce strong MRTF‐A nuclear accumulation and SRF activation (Sotiropoulos et al, 1999; Miralles et al, 2003). Both drugs induced MRTF‐A nuclear accumulation in free‐floating gels, albeit to a lesser extent compared with coverslips, suggesting that cells in three‐dimension buffer the effect of actin drugs on MRTF‐A transport (Fig 1B). The drugs induced only minimal and delayed MRTF‐A nuclear accumulation in anchored gels (Fig 1B). Significant nuclear accumulation in these gels was only achieved by using sub‐lethal concentrations of cytochalasin D or mycalolide B (also disrupting the MRTF‐A/G‐actin complex; supplementary Fig S1A,B online), suggesting that drug‐induced MRTF‐A nuclear accumulation is prevented in anchored gels.

Anchored gels show greater tension and stiffness compared with free‐floating gels, as cells locally displace and compact collagen fibres until they reach tensional homeostasis. To determine whether the responses of the cells in three dimensions could be linked to such differences, we analysed MRTF‐A transport in free‐floating matrices of increasing collagen concentration (Fig 1C,D). Apart from a slightly less‐spread phenotype in high collagen, there was little difference in cell morphology in the gels (Fig 1D), and MRTF‐A was consistently cytoplasmic before stimulation (Fig 1C). However, on serum stimulation, there was a peak in the concentration range within which cells downregulated MRTF‐A nuclear accumulation, presumably at the point at which matrix stiffness was close to tensional homeostasis levels. This suggested that MRTF‐A transport is hypersensitive to stimulation in cells remote from tensional homeostasis.

MRTF‐A accumulation in the nucleus has been linked to mechanical stress both in vivo in Drosophila (Somogyi & Rorth, 2004) and in vitro in mammalian cells (Zhao et al, 2007; Chan et al, 2010; Gomez et al, 2010). By using a microneedle to stretch individual cells on coverslips, we showed that acute stress caused rapid nuclear accumulation of MRTF‐A–GFP (Fig 2A,B). When a similar stress was applied to cells within anchored gels—by pulling on the adjacent matrix—no nuclear accumulation was recorded (Fig 2A,B), even when matrix distortion was strong enough to trigger cell blebbing (not shown). Control cells expressing GFP–pyruvate kinase showed no nuclear accumulation following stress on coverslips or in gels. To stress cells in free‐floating matrices without disrupting their ‘tension‐free’ state, we rapidly sheared the whole gel using a pipette. This triggered MRTF‐A accumulation in the nucleus, independently of how long the cells were in the matrix (Fig 2C; supplementary Fig S1C online). Cells in the sheared gels were not damaged, but mostly showed pancellular MRTF‐A (Fig 2C, inset).

Figure 2.

Stress‐induced MRTF‐A nuclear accumulation is prevented in anchored gels. (A) Starved MRTF‐A–GFP‐expressing cells on glass coverslips or within anchored gels were placed in L‐15 medium with 0.3% FBS and subjected to live imaging. By using a microneedle, an individual cell (green arrow), or the matrix adjacent to a cell (blue arrow), was pulled and the tension was maintained (white arrow). Stills from representative movies are shown. (B) MRTF‐A–GFP localization following stress; the mean±s.e.m is shown. MRTF‐A nuclear/cytoplasmic ratio for n cells pooled from at least two independent experiments. Stressed cells on coverslips (green, n=7), neighbouring control cell (red, n=6) and cells stressed in anchored gels (blue, n=4). GFP–pyruvate kinase localization was monitored in transiently transfected cells on coverslips (turquoise, n=11) or in anchored gels (purple, n=6). (C) MRTF‐A–GFP cells were seeded in free‐floating (FF) and anchored (A) collagen matrices for 24 h (24 h), followed by 24 h in low serum (48 h). The gels were rapidly sheared, fixed with formaldehyde and stained for F‐actin. MRTF‐A–GFP localization was monitored as the proportion of cells showing cytoplasmic, nuclear or pancellular distribution (one of three independent experiments, 30–60 cells/condition). Inset shows MRTF‐A–GFP cytoplasmic localization in a cell within control anchored (A) and free‐floating (FF) gels, and pancellular localization in a cell within a sheared free‐floating (FFS) gel. FBS, fetal bovine serum; GFP, green fluorescent protein; MRTF‐A, myocardin‐related transcription factor A; Pyrk, pyruvate kinase; 2D, two dimensions; 3D, three dimensions.

Overall, this suggested that MRTF‐A nuclear accumulation upon acute stimuli—biochemical (serum), chemical (actin drugs) or mechanical (direct stress)—is markedly downregulated in cells at tensional homeostasis. By contrast, cells on a non‐compliant two‐dimensional or three‐dimensional or on a highly compliant three‐dimensional substrate show greatly enhanced sensitivity to stimuli, consistent with the recently described connection between optimal substrate stiffness and cell behaviour (Lam et al, 2010; Raab et al, 2010).

Nuclear import is defective at tensional homeostasis

G‐actin levels regulate MRTF‐A transport and SRF activation on two‐dimensional substrate (Miralles et al, 2003; Vartiainen et al, 2007). To determine whether transport defects in anchored gels were linked to changes in G‐actin levels, we analysed the levels of monomeric (G‐) and filamentous (F‐) actin in cells in gels. Cells in anchored gels showed a significantly higher G/F actin ratio, and an altered G/F pattern following stimulation (Fig 3A). To test whether this increase in G‐actin was sufficient to perturb MRTF‐A transport, we used leptomycin B (LMB) to block CRM1‐dependent nuclear export. Under resting conditions, MRTF‐A transport is dominated by rapid nuclear export (Vartiainen et al, 2007). Accordingly, treatment with LMB led to rapid nuclear accumulation in free‐floating gels, but not in anchored gels, suggesting a defect in nuclear import (Fig 3B). Increasing G‐actin levels in free‐floating gels using latrunculin B (LatB), a G‐actin sequestering drug, prevented both LMB‐induced basal MRTF‐A nuclear accumulation and accumulation in the presence of serum (Fig 3B). However, LMB treatment only partly restored MRTF‐A nuclear accumulation following serum stimulation in anchored gels (Fig 3B) and did not improve accumulation following a saturating concentration of cytochalasin D (Fig 3C), suggesting that the main barrier to nuclear MRTF‐A accumulation in anchored gels is import. Finally, a MRTF‐A mutant defective in actin binding—constitutively nuclear in cells on coverslips—was also nuclear in both free‐floating and anchored gels (supplementary Fig S1D online). Altogether, this suggested that increased G‐actin levels in anchored gels could account for defective MRTF‐A nuclear accumulation through a reduction in import.

Figure 3.

Cells in anchored gels have increased G‐actin levels and defective MRTF‐A nuclear import. (A) Starved NIH3T3 cells within gels were stimulated with 15% FBS for the indicated times, fixed and stained with Alexa Fluor 488‐DNaseI (G‐actin; white bars) and rhodamine–phalloidin (F‐actin; black bars). Graph shows the mean fluorescence intensity (grey level/cell) normalized to starved cells (0 min) in free‐floating gels (29–45 cells per time point, n=3). (B) Serum‐starved MRTF‐A–GFP‐expressing cells in gels were incubated with the following: leptomycin B (LMB, 50 nM) plus either 0.3% FBS (n=4) or 15% FBS (n=2); 5 min pretreatment with latrunculin B (LatB 0.5 μM) in 0.3% FBS, followed by addition of LMB in either 0.3% FBS (n=3) or 15% FBS (n=1); or 15 min treatment with LatB in 0.3% FBS which was washed and replaced with 15% FBS and LMB (n=1). Graph shows the mean±s.e.m. for MRTF‐A nuclear/cytoplasmic ratio from a pool of cells (24–48 and 8–14 cells for free‐floating and anchored gels, respectively) from individual experiments (n). Asterisk indicates P<0.05 significant difference compared with time 0. (C) Serum‐starved cells in gels were incubated with 4 μM cytochalasin D in the presence or absence of LMB for the indicated times. Cells were subsequently fixed and MRTF‐A localization was quantified as the proportion of cells with nuclear, pancellular or cytoplasmic MRTF‐A. One of three experiments with 100 cells/condition is shown. FBS, fetal bovine serum; GFP, green fluorescent protein; MRTF‐A, myocardin‐related transcription factor A.

Defective import is linked to increased cofilin levels

Cofilin is an actin‐depolymerizing protein that functions as a master regulator of actin dynamics. Cofilin expression has been linked to both mechanical stress (Campbell et al, 2007; Zhao et al, 2007; Pho et al, 2008) and SRF regulation (Sotiropoulos et al, 1999; Verdoni et al, 2008). We found that cofilin was overexpressed by approximately 50% in cells in anchored gels, compared with free‐floating gels (Fig 4A,B). The amount of phosphorylated—that is, inactive—cofilin was unchanged, suggesting that cells in anchored gels have an increased pool of potentially active cofilin, which could account for increased G‐actin levels and defective MRTF‐A transport. We used short‐hairpin RNA interference constructs to downregulate both cofilin 1 and actin depolymerizing factor (ADF; Garvalov et al, 2007)—the main cofilin isoforms in NIH3T3 cells—in MRTF‐A–GFP cells and examined their behaviour in gels following serum stimulation. As expected, mCherry‐transfected control cells responded to serum stimulation in free‐floating gels, but not in anchored ones (Fig 4C). Cofilin downregulation by approximately 50% (supplementary Fig S2A online) did not affect MRTF‐A transport in free‐floating gels, but fully restored serum‐stimulated MRTF‐A nuclear accumulation in anchored gels (Fig 4C). To explore further the role of cofilin in MRTF‐A transport, we expressed wild‐type or constitutively active LIM kinase (LIMK) to block cofilin activity, or either wild‐type cofilin or constitutively active S3A or inactive S3E mutants to perturb cofilin levels. Previous work has shown that although LIMK activates SRF in the absence of serum, it is not essential for SRF activation by serum in NIH3T3 cells (Sotiropoulos et al, 1999; Geneste et al, 2002). Similarly, overexpression of a constitutively active cofilin mutant did not prevent serum‐mediated SRF activation (Geneste et al, 2002). Surprisingly then, overexpressing wild‐type or constitutively inactive LIMK did not increase nuclear MRTF‐A accumulation in resting cells in free‐floating gels (supplementary Fig S2B online), but all the constructs reduced nuclear MRTF‐A following serum stimulation (Fig 4D; supplementary Fig S2B online). Despite a tendency towards an increase in nuclear accumulation following serum stimulation for cells expressing LIMK, overall there was no significant effect for any of the constructs in anchored gels. This suggests that the link between the increased cofilin levels and defective MRTF‐A transport in anchored gels is not only due to a change in cofilin activity. Instead, it is possible that an as‐yet‐unknown mechanism uses the net amount of cofilin to control MRTF‐A transport and SRF activity.

Figure 4.

MRTF‐A transport defect in anchored gels is linked to increased cofilin levels. (A) Cofilin is overexpressed in cells in anchored gels. Serum‐starved NIH3T3 fibroblasts in gels were processed for western blotting, with cell number equalized before loading. (B) Band intensity was quantified using the Gel Analysis tool in Image J, and proteins levels were normalized to GAPDH and to the corresponding values in free‐floating (FF) gels. The mean±s.e.m. of n gels pooled from at least three experiments is shown (actin, n=4; cofilin n=19, phospho‐cofilin n=10), asterisk indicates P=0.0002. (C) Downregulating cofilin restores MRTF‐A transport in anchored gels. MRTF‐A–GFP‐expressing fibroblasts were transfected with a control plasmid expressing mCherry with/without both cofilin 1 and ADF short‐hairpin RNA constructs. After 24 h, the cells were seeded in gels, grown for 24 h and starved for 24 h in low serum. The cells were then left untreated (t0) or stimulated with 15% serum for 15 min (t15) and MRTF‐A–GFP localization and cofilin levels were monitored in 30–50 mCherry‐expressing cells using confocal microscopy (n=3; difference between siRNA t15 and mCherry t15 in anchored gels: P=0.006). (D) MRTF‐A–GFP‐expressing fibroblasts were transfected with a control mCherry construct with or without LIMK or LIMK‐CA‐expressing constructs, or with RFP‐tagged cofilin WT, S3A and S3E constructs. After 24 h, the cells were seeded in gels, grown for 24 h and starved for 24 h in low serum. The cells were then stimulated with 15% serum for 15 min (t15) and MRTF‐A–GFP localization was monitored on confocal images (n=2, 30–40 cells/condition in each experiment). FBS, fetal bovine serum; GAPDH, glyceraldehyde‐3‐phosphate dehydrogenase; GFP, green fluorescent protein; LIMK, LIM kinase; LIMK‐CA, constitutively active LIM kinase; MRTF‐A, myocardin‐related transcription factor A; siRNA, small interfering RNA; RFP, red fluorescent protein; WT, wild type.

In summary, our work shows for the first time that activation of the MRTF‐A/SRF pathway is controlled by the matrix tensional state, rather than being a response to local stress or matrix compliance. Cells markedly downregulate MRTF‐A transport when they have attained their preferred level of tension, independently of the means by which the pathway is activated, such as chemical or mechanical signals. Although recent studies have underlined the importance of mechanical sensing in the regulation of cell behaviour (Paszek et al, 2005; Ingber, 2006; Engler et al, 2008; Gomez et al, 2010), this is the first report linking tensional homeostasis to the downregulation of a main transcriptional‐activation pathway. The tensional‐homeostasis level is an intrinsic property of cells (Brown et al, 1998); although permanent changes in mechanosensitivity occur in disease (Paszek et al, 2005; Arnoczky et al, 2008; Townley et al, 2009; Ezra et al, 2010), the mechanisms by which cells define and regulate their preferred tensional level are unknown. Our results indicate that cofilin/ADF is implicated in mechanotransduction and tensional homeostasis through the modulation of G‐actin levels, and subsequently through the MRTF‐A/SRF pathway.


See the supplementary information online for details of cells, reagents, immunofluorescence and western blotting.

Cells and reagents. Stable tetracycline‐inducible MRTF‐A–GFP cells were described by Vartiainen et al, 2007. The GFP–pyruvate kinase construct was prepared by subcloning a partial sequence of chicken pyruvate kinase into pEGFP‐C1 (Clontech). Cofilin1 and actin depolymerizing factor (ADF) short‐hairpin RNA constructs (Garvalov et al, 2007), and wild‐type cofilin, S3A and S3E in pmRFP‐N1, as well as LIMK1 in pCMV‐Tag1 and constitutively active LIMK (T508EE) in pcDNA3.1, were a gift from James Bamburg (Colorado State University). Cells were transfected using Lipofectamine 2000 (Invitrogen). All drugs were obtained from Calbiochem, except for cytochalasin D (Sigma‐Aldrich).

Collagen gels. A volume of 1.5 mg/ml collagen gels with 2 × 105 cells/ml were prepared in Mattek dishes (Dahlmann‐Noor et al, 2007; Martin‐Martin et al, 2011) and, following polymerization, were either manually detached from the well (free‐floating gels) or remained attached (anchored gels). Gels were maintained in DMEM with 10% fetal bovine serum (FBS) for 24 h before serum starvation.

Mechanical stress assays. MRTF‐A–GFP‐expressing cells were seeded on collagen‐coated coverslips or within anchored collagen gels, grown in DMEM with 10% FBS for 24 h and starved in low‐serum (0.3% FBS) medium for a further 24 h. The medium was replaced with L15 medium (0.3% FBS) for live imaging using a × 63 objective on an inverted microscope enclosed in an environmental chamber. Images were captured every 60 s with an ORCA‐ER CCD digital camera (Hammamatsu Photonics, UK), using Openlab software (Perkin Elmer). By using a microneedle controlled by a micromanipulator (Eppendorf), an individual cell (coverslip) or the matrix next to a cell (gel) was placed under stress and the tension was maintained throughout the time lapse. For shearing assays, cells were seeded in the gels for 24 h (with or without a further 24 h in low serum) and stress was applied to the whole gel by pipetting it up and down twice with a transfer pipette, followed by fixation and F‐actin staining.

Immunofluorescence and western blotting. F‐ and G‐actin staining were done as described previously (Cramer et al, 2002; Shao et al, 2006). Fluorescence levels were measured in Image J on projected confocal stacks acquired with a BioRad Radiance 2000 confocal microscope. MRTF‐A nuclear/cytoplasmic ratio was calculated using Volocity (Perkin Elmer) or Image J on single confocal slices taken within the central region of the cells.

For western blotting, cells were extracted from the gels with 0.05% Collagenase D and subjected to SDS–PAGE, and transferred to nitrocellulose membranes. Proteins were visualized using chemiluminescence reagents (Pierce) and quantified using Image J. Details of the antibodies used are provided in the supplementary information online.

Statistical analysis. All graphs show mean and standard error. Statistical analysis was performed using Student's t‐test.

Supplementary information is available at EMBO reports online (

Conflict of Interest

The authors declare that they have no conflict of interest.

Supplementary Information

Supplementary Information [embor2011141-sup-0001.pdf]


We thank J. Bamburg for the LIMK and cofilin constructs and S. Gokool for technical help. This work was supported by a Medical Research Council grant GO500927 (M.B./R.T./P.T.K.), the Royal Society (M.B.), Cancer Research UK (R.T.) and the National Institute for Health Research Biomedical Research Centre for Ophthalmology at Moorfields Eye Hospital and University College London Institute of Ophthalmology (P.T.K.). Laboratories and imaging facilities were supported by the Wellcome Trust and the Medical Research Council.