Substrate stiffness-dependent regulation of the SRF−Mkl1 co-activator complex requires the inner nuclear membrane protein Emerin

Wound healing requires the differentiation of fibroblasts and other cell types into myofibroblasts, which synthesize and remodel the extracellular matrix (ECM) to promote repair (Desmouliere et al., 1993; Hinz, 2010). Persistent myofibroblast activity, however, leads to tissue stiffening and fibrosis, and has also been linked to the cancer progression (Butcher et al., 2009; Hinz, 2009; Huang et al., 2012). The co-activator complex comprising serum response factor (SRF) and megakaryoblastic leukemia 1 protein (Mkl1; also known as MRTF-A, MAL) (SRF−Mkl1) is critical for myofibroblast differentiation (Crider et al., 2011; Small et al., 2010). Though chemical signals including TGFβ1 induce SRF−Mkl1-dependent gene activation during wound healing, tension in the actin cytoskeleton developed through cell-ECM interactions is required for this response (Gomez et al., 2010; Huang et al., 2012; McGee et al., 2011; O'Connor et al., 2015; Varney et al., 2016). Moreover, a feed-forward loop that couples tissue stiffness and SRF−Mkl1-dependent transcription is thought to underlie persistent myofibroblast activation (Zhou et al., 2013). Thus, determining how mechanical cues modulate SRF−Mkl1-dependent gene expression is critical to understand wound healing and pathologies caused by altered ECM mechanics.

The nuclear lamina is a network of nuclear envelope-associated intermediate filaments and their associated proteins. Recently, the nuclear lamina proteins Lamins A/C and Emerin were shown to promote SRF−Mkl1-dependent transcription (Ho et al., 2014). Actin-generated forces regulate Lamins A/C and Emerin (Buxboim et al., 2014; Guilluy et al., 2014; Swift et al., 2013), but whether the nuclear lamina contributes to the mechanical control of SRF−Mkl1-dependent gene expression has not been tested. Here, we show that Emerin specifically enhances SRF−Mkl1-dependent transcription in cells grown on stiff substrates, thereby linking ECM mechanics to gene expression.

Mkl1 is sequestered in the cytoplasm of resting cells by actin monomers, which both inhibit the nuclear import of Mkl1 and promote its nuclear export (Guettler et al., 2008; Miralles et al., 2003; Mouilleron et al., 2011; Pawłowski et al., 2010; Vartiainen et al., 2007). In response to stimuli that initiate actin polymerization, Mkl1 accumulates in the nucleus and, together with SRF, induces the expression of genes important for myofibroblast differentiation (Cen et al., 2003; Crider et al., 2011; Esnault et al., 2014; Miralles et al., 2003). Mkl1 nuclear localization after stimulation is significantly increased in cells on stiff substrates when compared to cells plated on more compliant substrates (McGee et al., 2011; O'Connor et al., 2015; Varney et al., 2016). We tested whether the nuclear lamina is important for mechanosensing in wild-type (Emd^WT) vs Emerin-knockout (Emd^KO) mouse embryonic fibroblasts (MEFs) cultured on glass coverslips and collagen-coated polyacrylamide hydrogels with elastic moduli between 0.2−50 kPa, which spans the range observed in vertebrate tissues in vivo (Butcher et al., 2009).

After serum-starvation, Mkl1 localization − which we quantify by measuring the nuclear-to-cytoplasmic (N:C) ratio of Mkl1 − was similar in Emd^WT and Emd^KO MEFs, regardless of the underlying substrate (N:C ratio of ∼1−1.2) (Fig. 1). Mkl1 accumulated in the nucleus within 5 min of serum stimulation in Emd^WT MEFs grown on glass coverslips, reaching a maximum N:C ratio of 3.5±0.2 (Fig. 1A,B). The level of Mkl1 in the nucleus then steadily declined, leading to an intermediate N:C ratio of 1.9±0.1 by 30 min that was stable for the remainder of the experiment (up to 90 min) (Fig. 1B). The kinetics of Mkl1 nuclear accumulation were not affected by substrate stiffness (Fig. S1A), but the maximum level of Mkl1 nuclear accumulation 5 min after serum stimulation was reduced in Emd^WT MEFs on soft substrates (≤1 kPa) relative to the stiffest substrate (50 kPa) (Fig. 1C,D), consistent with previous reports demonstrating that Mkl1 nuclear accumulation is substrate stiffness-dependent (McGee et al., 2011; O'Connor et al., 2015; Varney et al., 2016). Emerin was required for maximal Mkl1 nuclear accumulation on glass coverslips and on the stiffest hydrogel substrate (maximum N:C ratios in Emd^KO MEFs of 2.5±0.1 and 1.8±0.1, respectively) (Fig. 1A-D). A similar result was obtained when Emd^KO MEFs were plated on substrates with an elastic modulus of 25 kPa (Fig. 1D), though the data did not rise to the level of statistical significance (P<0.07). On substrates with elastic moduli of ≤12 kPa, however, the N:C ratio of Mkl1 after serum stimulation was comparable between Emd^WT and Emd^KO cells (Fig. 1C,D). Mkl1 protein levels were identical in Emd^WT and Emd^KO MEFs (Fig. S1B). Further, nuclear accumulation of the Smad 2/3 transcription factors after stimulation with TGFβ1 was similar in Emd^WT and Emd^KO MEFs cultured on glass coverslips or hydrogel substrates with elastic moduli of 50 or 1 kPa (Fig. S1C-F). Thus, Emerin specifically enhances Mkl1 nuclear accumulation in cells grown on stiff substrates.

A similar dependence on Lamins A/C for Mkl1 nuclear accumulation has previously been reported in MEFs grown on glass coverslips (Ho et al., 2014). Consistent with this study, we found that depletion of Lamins A/C by using specific small interfering RNAs (siRNAs) impaired Mkl1 nuclear accumulation 5 min after serum stimulation in MEFs grown on a stiff hydrogel substrate [50 kPa; N:C ratio: 2.6±0.2 (siControl) vs 1.8±0.2 (siLamins A/C)] (Fig. 1E, Fig. S2A). On a more compliant substrate, however, the maximum Mkl1 nuclear localization was not significantly different between control and Lamin A/C-depleted cells [1 kPa; N:C ratio: 1.7±0.3 (siControl) vs 1.8±0.3 (siLamins A/C)] (Fig. 1E, Fig. S2A). The nuclear lamina is, therefore, specifically required to couple ECM mechanics and Mkl1 localization following serum stimulation of cells grown on stiff substrates. Interestingly, depletion of Lamins A/C, which anchor Emerin at the inner nuclear membrane (INM), further reduced Mkl1 nuclear accumulation in Emd^KO MEFs on glass coverslips from a maximum N:C ratio of 2.3±0.1 (siControl) to a N:C ratio of 1.7±0.1 (siLA/C) (Fig. 1F, see also Fig. S2B). Thus, Lamins A/C are likely to have both Emerin-dependent and Emerin-independent roles in mediating Mkl1 nuclear accumulation after serum stimulation.

We next asked how reduced Mkl1 nuclear localization in Emd^KO cells affected SRF−Mkl1-dependent gene expression. Emerin was required for the full induction of an SRF-specific luciferase reporter in cells grown on plastic or 50 kPa hydrogel substrates (Fig. 2A). In contrast, there was no significant difference in reporter activity between Emd^WT and Emd^KO cells grown on substrates with elastic moduli of 1 kPa (Fig. 2A). Furthermore, wild-type MEFs showed similar levels of Mkl1 accumulation in the nucleus, relative to Emd^KO MEFs, on plastic and stiff hydrogels but showed no significant difference on a more compliant substrate [Fig. 2A, ratio (Emd^WT:Emd^KO)], suggesting that Emerin is required to link SRF−Mkl1-dependent gene expression to substrate stiffness.

Surprisingly, the mRNA levels of endogenous SRF−Mkl1 target genes, including VCL (which encodes vinculin), ACTB (β-actin), and ACTA2 (smooth-muscle actin, SMA), were not significantly different between Emd^WT and Emd^KO MEFs grown on plastic before or 90 min after serum stimulation (Fig. 2B). Moreover, while TAGLN (smooth muscle protein 22, SM22) mRNA levels were lower before and after serum stimulation in Emd^KO cells than in Emd^WT cells, the fold-increase in mRNA levels was similar in both cell lines (Fig. 2B). Emerin was, therefore, not required for endogenous SRF−Mkl1-dependent gene activation after serum stimulation. Emerin was, however, required for VCL, ACTA2 and TAGLN expression in cells exposed to serum on plastic dishes for 24 h, which we refer to as steady state (Fig. 2C). Together, these data demonstrate that Emerin promotes Mkl1 localization specifically in cells grown on stiff substrates and is required to maintain SRF−Mkl1-dependent gene activation over a 24 h time period. These data agree well with the recent analysis of VCL expression in Lamin A/C null fibroblasts (Ho et al., 2014). In that study, VCL mRNA levels were comparable to control cells 1 h after serum stimulation but were significantly reduced 5 h later.

A critical mechanism for sensing ECM mechanics occurs through focal adhesions (Schwartz, 2010). We, therefore, investigated whether Emerin is required for focal adhesion assembly. Complicating these studies was the fact that Emd^KO MEFs have a reduced area and volume compared to control cells (Fig. 3A,B). Hence, we quantified the total focal adhesion area in individual cells and the percentage of cell area associated with focal adhesions, which provides an area-normalized measure of focal adhesion size. Emd^KO MEFs on glass coverslips had reduced focal adhesion size by both metrics at steady state (Fig. 3A,B). Emerin, therefore, plays a critical role in the ability of cells to sense ECM mechanics. Further, the focal adhesion defects persisted in Emd^KO cells after 24 h of serum starvation (Fig. 3A,B), which is likely to underlie the deficiency in Mkl1 localization and SRF−Mkl1-dependent gene activation in response to serum stimulation of cells grown on stiff substrates (Figs 1D, 2A).

The reduced focal adhesion size in Emd^KO cells correlated with the reduced transcription of SRF−Mkl1 target genes at steady state (Figs 2C, 3A,B). In contrast, these parameters were partially uncoupled after serum starvation; while differences in focal adhesions remained (Fig. 3A,B), the mRNA levels of VCL, ACTB, and ACTA2 were not significantly different between control and Emd^KO MEFs (Fig. 2B). To explore this paradox further, we quantified the levels of vinculin, β-actin, SMA and SM22 protein in both control and Emd^KO cells grown on plastic. Interestingly, the levels of each protein were significantly reduced relative to Emd^WT MEFs in Emd^KO MEFs at both steady state and after serum starvation (Fig. 3C,D). When taken together, these data indicate that, after serum deprivation, transcript levels of some genes equalize more rapidly between Emd^WT and Emd^KO than their protein products do, suggesting that the rate of protein turnover in the absence of serum influences how cells respond to subsequent rounds of serum stimulation.

If the primary means by which Emerin contributes to sensing ECM mechanics occur through SRF−Mkl1-dependent gene expression, we would predict that a constitutively active form of Mkl1 (CA-Mkl1) can rescue focal adhesion area in Emd^KO cells. We, therefore, generated a stable cell line that expressed CA-Mkl1 from a doxycycline-inducible promoter (see Materials and Methods; Cen et al., 2003; Miralles et al., 2003). Although localized throughout the cell (Fig. 4A), CA-Mkl1 expression increased the mRNA levels of all tested SRF−Mkl1 target genes and eliminated the difference in TAGLN and ACTA2 expression between Emd^WT and Emd^KO cells in serum-starved cells grown on plastic (Fig. 4B). Similarly, CA-Mkl1 induction equalized VCL and TAGLN expression, and substantially reduced the difference in ACTA2 expression between control and Emd^KO MEFs at steady state (Fig. S3A). The lower levels of vinculin, actin, SM22 and SMA protein in Emd^KO MEFs grown on plastic was also reversed by CA-Mkl1 expression (Fig. 4C,D, Fig. S3B,C). CA-Mkl1, therefore, effectively bypassed the requirement for Emerin in SRF−Mkl1-dependent gene activation.

CA-Mkl1 did not rescue the spread area or volume deficiencies evident in Emd^KO MEFs (Fig. 4E,F, Fig. S3D,E). The total focal adhesion area after CA-Mkl1 expression was similar in serum-starved Emd^WT and Emd^KO cells grown on glass, which resulted in a greater percentage of the cell area in Emd^KO MEFs being associated with focal adhesions than in control cells (Fig. 4E,F). A similar, although somewhat muted, effect was observed in cells at steady state (Fig. S3D,E). Thus, CA-Mkl1 expression largely suppressed the focal adhesion defect in Emd^KO MEFs, suggesting that Emerin contributes to the assembly of an actin-based mechanosensory apparatus by promoting SRF−Mkl1-dependent transcription.

A crucial question is how Emerin specifically promotes SRF−Mkl1-dependent gene activation in cells grown on stiff substrates. Emerin binds to actin and promotes actin polymerization in vitro (Holaska and Wilson, 2007; Holaska et al., 2004), and mutations that disrupt actin binding do not support SRF−Mkl1-dependent transcription (Ho et al., 2014). The ability of Emerin to promote actin polymerization can, therefore, be linked to ECM mechanics. Indeed, Emerin is tyrosine phosphorylated in cells grown on stiff substrates and was recently shown to move from the INM to the outer nuclear membrane (ONM) after exposure to biaxial strain (Guilluy et al., 2014; Le et al., 2016); although we did not observe a change in Emerin localization at the INM or ONM on hydrogel substrates of differing stiffness in this study (Fig. S4A,B). Further studies exploring the relationship between Emerin, substrate stiffness and actin dynamics will be crucial to address these issues. Equally important is to discover how the Emerin-independent functions of Lamins A/C promote Mkl1 nuclear localization in cells grown on stiff substrates. We have demonstrated recently that nuclear envelope-spanning linker of nucleoskeleton and cytoskeleton (LINC) complexes, which couple F-actin, microtubules and intermediate filaments to the nuclear lamina, promote focal adhesion assembly through a transcription-independent mechanism (Thakar et al., 2017). Thus, Lamins A/C could enhance Mkl1 nuclear accumulation through LINC complexes in Emd^KO cells.

These studies have important implications for our understanding of fibrosis and cancer, both of which are linked to persistent myofibroblast activity. Specifically, our results suggest that Emerin is a crucial component of the mechanical feed-forward loop that sustains SRF−Mkl1-dependent gene expression in response to tissue stiffening (Duffield et al., 2013; Small et al., 2010; Zhou et al., 2013). The nuclear lamina may, therefore, drive pathologies that involve altered ECM mechanics.

Emd^WT and Emd^KO (Emd^+/Y and Emd^−/Y, respectively) MEFs were maintained in Dulbecco's modified Eagle's medium (DMEM) and 10% fetal bovine serum (FBS; Sigma, St Louis, MO) and starved in serum-free medium containing 0.2% FBS or in 0.5% fatty acid-free BSA (Sigma) for 16-24 h. Following starvation, MEFs were stimulated with 10% FBS. Cells at steady state were kept in a constant serum concentration for 24 h. Cell lines expressing doxycycline-inducible constitutively active Mkl1 tagged to GFP (CA-Mkl1−GFP) (residues 81−929 of mouse Mkl1) that lacks the N-terminal actin-binding domain, were generated as described (Thoreen et al., 2012). CA-Mkl1 expression was induced with 1 μg/ml doxycycline for 24 h before cells were used.

Coverslips were either coated with 50 μg/ml fibronectin or with collagen pre-coated polyacrylamide hydrogels (Matrigen, Brea, CA). Immunofluorescence was performed as described (Carroll et al., 2009). To assess Emerin localization, fixed cells were permeabilized with 0.25% Triton-X 100 or 0.003% digitonin. Images were captured on a Nikon TE2000-S by using a 40× oil immersion objective or a Leica SP5 laser scanning confocal microscope by using a 100× oil immersion objective. Image analysis and quantification was performed in ImageJ. N:C ratios were calculated by dividing the total fluorescence signal in the nucleus by the total fluorescence signal in a ring of cytoplasm surrounding the nucleus after accounting for differences in the measurement areas. 100 cells were quantified for each condition in each experiment. Emerin localization to the INM or ONM was quantified by measuring the intensity levels of Emerin fluorescence at the nuclear envelope in Triton X-100 or digitonin-permeabilized cells, respectively (Le et al., 2016). Focal adhesions were quantified as described (Thakar et al., 2017). The area of all FAs from each cell was summed and then divided by the cell area to determine total FA area per cell and the percent of vinculin-positive area per cell, respectively. The radius of spherical cells after trypsinization, measured with a Cellometer Auto T4 cell counter (Nexcelom Bioscience, Lawrence, MA), was used to calculate cell volume. Actin filaments were visualized with phalloidin (Life Technologies, Eugene, OR). Antibodies are listed in Table S1.

MEFs were lysed directly in 1× Laemmli buffer for western blots. mRNA for quantitative PCR (qPCR) was isolated by using the RNeasy Mini Kit (Qiagen, Germany). cDNA was synthesized from 100 ng mRNA with SuperScript III First-Strand synthesis system (Invitrogen, Carlsbad, CA) using an oligo dT primer. qPCR was performed using SYBR green detection in a CFX96 Real-Time System (Bio-Rad, Hercules, CA). mRNA levels were calculated from qPCR data by using the dCT method after normalization to TATA-box binding protein (TBP) mRNA. Oligonucleotids for qPCR are listed in Table S2.

Cells were transfected with a 50:1 ratio of 5xSRF-RE (Promega, Madison, WI) and pRenilla plasmids (gift from D. Krause). Four hours later, the medium was replaced with serum-free medium for 24 h. Cells were then harvested or stimulated for 3 h with FBS. Luciferase activity was measured using the Dual Luciferase Reporter Kit (Promega, Madison, WI). Values for each experimental condition are the average ratio of firefly-to-Renilla luciferase activity in three technical replicates. The specificity of the luciferase reporter was verified using cytochalasin D and latrunculin B as described (Vartiainen et al., 2007).

The authors thank Christopher May (Yale University, New Haven, CT) for comments on the manuscript. We also thank the King, Lusk, Yao, and Bahnmanyar labs for helpful comments on the project, Jan Lammerding (Cornell University, Ithaca, NY) for the Emd^WT and Emd^KO MEFs, Diane Krause (Yale University, New Haven, CT) for Mkl1 and pRenilla plasmids, and Carson Thoreen (Yale University, New Haven, CT) for lentiviral vectors.