Research ArticlesMEDICINE

Filamin A interacts with the coactivator MKL1 to promote the activity of the transcription factor SRF and cell migration

See allHide authors and affiliations

Science Signaling  10 Nov 2015:
Vol. 8, Issue 402, pp. ra112
DOI: 10.1126/scisignal.aad2959

F-actin delivers MKL1 to the nucleus

The highly dynamic actin cytoskeleton maintains cell shape, enables cell movement, and contributes to cell division. Perhaps surprisingly, actin cytoskeletal dynamics also regulate gene expression. Persistent cell migration involves both cytoskeletal reorganization and expression of specific genes. The transcription factor serum response factor (SRF) and the coactivator megakaryoblastic leukemia 1 (MKL1) mediate transcriptional changes in response to external signals that affect actin dynamics. Globular actin (G-actin) keeps MKL1 in the cytosol, and signals that promote the formation of filamentous actin (F-actin) trigger the nuclear translocation of MKL1 and transcriptional activation of SRF. MKL1 not only had to dissociate from G-actin but also had to associate with the F-actin binding protein filamin A (FLNA) to promote SRF activity. Indeed, FLNA bound to the promoters of the same genes that are regulated by MKL1-SRF. Blocking the interaction between FLNA and MKL1—by inhibiting actin polymerization, mutating the FLNA interaction site on MKL1, or reducing the amount of FLNA—impaired the expression of MKL1-SRF target genes. Thus, actin dynamics not only actively move cells but also actively mediate the transcriptional activity of regulators needed for cellular movement.

Abstract

Megakaryoblastic leukemia 1 (MKL1) is a coactivator of serum response factor (SRF) that promotes the expression of genes associated with cell proliferation, motility, adhesion, and differentiation—processes that also involve dynamic cytoskeletal changes in the cell. MKL1 is inactive when bound to monomeric globular actin (G-actin), but signals that activate the small guanosine triphosphatase RhoA cause actin polymerization and MKL1 dissociation from G-actin. We found a new mechanism of MKL1 activation that is mediated through its binding to filamin A (FLNA), a protein that binds filamentous actin (F-actin). The interaction of FLNA and MKL1 was required for the expression of MKL1 target genes in primary fibroblasts, melanoma, mammary and hepatocellular carcinoma cells. We identified the regions of interaction between MKL1 and FLNA, and cells expressing an MKL1 mutant that was unable to bind FLNA exhibited impaired cell migration and reduced expression of MKL1-SRF target genes. Induction and repression of MKL1-SRF target genes correlated with increased or decreased MKL1-FLNA interaction, respectively. Lysophosphatidic acid–induced RhoA activation in primary human fibroblasts promoted the association of endogenous MKL1 with FLNA, whereas exposure to an actin polymerization inhibitor dissociated MKL1 from FLNA and decreased MKL1-SRF target gene expression in melanoma cells. Thus, FLNA functions as a positive cellular transducer linking actin polymerization to MKL1-SRF activity, counteracting the known repressive complex of MKL1 and monomeric G-actin.

INTRODUCTION

Megakaryoblastic leukemia 1 (MKL1; also known as MRTF-A, MAL, or BSAC) is a transcriptional coactivator of serum response factor (SRF). SRF regulates gene expression by binding to a highly conserved sequence of nucleotides [CC(A/T)6GG] referred to as CArG box or serum response element (SRE) in the promoters of various target genes associated with the immediate early response, cell growth, differentiation, cell migration, and organization of the cytoskeleton (14). SRF translates rearrangements of the actin cytoskeleton into gene expression through association with its coactivator, MKL1. According to the current paradigm, the N-terminal RPEL domains of MKL1 bind to monomeric globular actin (G-actin), and the dissociation of this protein complex after cytoplasmic actin polymerization enables MKL1 to translocate to the nucleus and to activate gene transcription through SRF (5). However, MKL1-SRF regulation appears to be more complex because several lines of evidence suggest that, in addition to that of cytoplasmic actin, the polymerization of nuclear actin is also involved in MKL1-SRF activation—an unexpected phenomenon whose requirement is not intuitively clear. First, signal-operated nuclear actin polymerization, driven by the mDia family of formins, leads to activation of MKL1 and SRF (6, 7). Second, nucleus-resident actin mutants, such as S14C-actin and G15S-actin, that stabilize polymerized actin promote SRF activity (8, 9). Third, the nuclear protein MICAL-2 decreases nuclear G-actin in a redox-dependent fashion and subsequently increases MKL1-SRF activity (10). Thus, hitherto unappreciated regulatory mechanisms appear to determine MKL1 activity in the nucleus.

Filamin A (FLNA) is an actin binding protein with cross-linking activity that regulates cell shape and migration (1114). FLNA is composed of an N-terminal actin binding domain followed by 24 repeat β-pleated sheet units. Two calpain-sensitive “hinges” separate the repeats into Rod1 (repeats 1 to 15), Rod2 (repeats 16 to 23), and the self-association domain (repeat 24) (11). In addition to its well-characterized function as a cytoplasmic architectural molecule, FLNA also has important roles in the nucleus. Both full-length FLNA and proteolytic fragments of FLNA interact with nuclear proteins, including transcription factors such as steroid hormone receptors, the Smad family, FOXC1, hypoxia-inducible factor-1α, and proteins involved in DNA repair (for example, BRCA2) (1520). FLNA interacts with refilin B to organize perinuclear actin networks and to regulate nuclear shape (21). Furthermore, FLNA is an abundant nucleolar protein that associates with RNA polymerase I to suppress rRNA transcription (22).

A growing body of evidence indicates an important role for nuclear MKL1 in tumor development and progression. Recently, SCAI (suppressor of cancer cell invasion) has been reported to bind and inhibit MKL1 in the nucleus (23). Nuclear MKL1, as found in mammary and hepatocellular carcinoma cells lacking the tumor suppressor DLC1 (deleted in liver cancer 1), is critical for tumor growth because the therapeutic knockdown of MKL1 abolishes hepatocellular carcinoma xenograft growth (24, 25). However, specific nuclear factors involved in MKL1-SRF transcriptional regulation, particularly in tumor cells, remain to be identified.

Tissue-specific deletion of SRF revealed essential biological functions of SRF target genes in the development of the cardiovascular system, muscle, liver, and brain (2634). Mouse embryos homozygous for an SRF-null mutation die at gastrulation as a result of defective migration of embryonic tissue layers (35). Similarly, deletion of MKL1 and MKL2 has been shown to account for a deficit in cell migration (36). Recent studies have demonstrated that knockdown of MKL1 and MKL2 reduces the migration of hepatocellular and mammary carcinoma cells (25, 37). Furthermore, MKL1 and MKL2 are required for experimental metastasis (37). Some MKL-SRF target genes, including integrin α5 (ITGA5), FHL1, and Pkp2, have been shown to mediate at least in part the migratory MKL effects (38). Despite the importance of MKL1 and MKL2 in cell migration and metastasis, it remains unclear how MKL1 executes motile cell functions.

Here, we found a new mode of MKL1 regulation through its binding of FLNA, suggesting that FLNA functions as a cellular transducer linking actin polymerization to MKL1-SRF activity. In contrast to the known repressive complex consisting of MKL1 and monomeric actin, the interaction between FLNA and MKL1 was indispensable for MKL1-SRF transcriptional activity. These findings advance our mechanistic understanding of MKL1-SRF activation, which regulates cell migration, differentiation, and cell growth.

RESULTS

Identification of FLNA as a novel MKL1 interacting protein

To search for MKL1 interacting proteins, we performed a yeast two-hybrid screen using a human keratinocyte library, with amino acids 100 to 723 of MKL1 as bait. We identified the actin filament cross-linking protein FLNA as a novel MKL1 interaction partner. To investigate functional consequences of the MKL1-FLNA interaction, we took advantage of the FLNA-deficient melanoma cell line M2 and the corresponding FLNA-expressing cell line A7 (Fig. 1A) (39). FLNA appeared as a prominent 280-kD band representing unprocessed full-length FLNA and a very weak (190-kD N-terminal and 90-kD C-terminal) calpain cleavage fragment of FLNA, as determined by treatment with calpain inhibitor III (fig. S1A).

Fig. 1 Identification of FLNA as a novel MKL1 interacting protein.

(A) Immunoblotting for FLNA, MKL1, and heat shock protein 90 (HSP90) as a loading control in lysates of A7 cells endogenously expressing FLNA and FLNA-deficient M2 cells. (B to D) Immunoprecipitation (IP) for FLNA and Western blot (IB) for MKL1 and FLNA in lysates from (B) A7 cells transfected with FLAG-tagged MKL1, (C) MEFs, or (D) HuH7 and MDA-MB-468 cells. BO, Sepharose beads–only control, without antibody. (E) Immunofluorescence analysis of MKL1 and FLNA in A7 cells. Scale bar, 20 μm. Representative images (E) and blots (A to D) are shown for n ≥3 biological replicates.

First, we confirmed the interaction of MKL1 and FLNA by immunoprecipitation of A7 melanoma cells expressing FLAG-tagged MKL1. MKL1 was readily detectable after immunoprecipitation of FLNA (Fig. 1B). As a control experiment, we performed immunoprecipitation in the FLNA-negative cell line M2. FLNA and MKL1 were only recovered from FLNA immunoprecipitates when FLNA was reexpressed in M2 cells (fig. S1B). Likewise, neither FLNA nor MKL1 was observed in immunoprecipitates with an unspecific primary antibody (fig. S1C), further confirming the specificity of the MKL1-FLNA interaction. Binding between endogenous FLNA and MKL1 was also observed in mouse embryonic fibroblasts (MEFs) (Fig. 1C), as well as in HuH7 hepatocellular carcinoma cells and MDA-MB-468 mammary carcinoma cells (Fig. 1D).

Next, we investigated where in the cell the interaction of MKL1 and FLNA takes place. Double immunofluorescence with differentially labeled antibodies specific for MKL1 and FLNA revealed that MKL1 and FLNA colocalized predominantly in the nucleus (Fig. 1E). Increased abundance of nuclear FLNA in A7 melanoma cells has been described previously (17), consistent with our previous studies showing nuclear localization of MKL1 in tumor cells lacking the tumor suppressor DLC1 (24, 25). MKL1 accumulated in the nucleus of DLC1-deficient A7 melanoma cells (fig. S1D). Together, these results indicate that FLNA is a nuclear MKL1 binding partner.

Mapping of MKL1-FLNA binding sites

To map MKL1 regions responsible for the interaction with FLNA, we genetically engineered FLAG-tagged MKL1 by in vitro mutagenesis and expressed different constructs in A7 melanoma cells. As a starting point, we used different MKL1 deletion variants to identify protein segments essential for FLNA interaction (Fig. 2, A and B). Immunoprecipitation of cell extracts with an antibody against the FLAG-tag and detection with FLNA antibodies showed that an N-terminal deletion construct of MKL1 lacking the first 300 amino acid residues (N300) was still able to bind FLNA (Fig. 2C). C-terminal deletion constructs of MKL1 (C500, C630, and C830) also interacted with FLNA (Fig. 2D). We then engineered internal deletions of amino acids 301 to 380 (Δ301–380) and amino acids 381 to 506 (Δ381–506). Whereas MKL1 Δ381–506 still associated with FLNA, an interaction between MKL1 Δ301–380 and FLNA was hardly detectable (Fig. 2E and fig. S1E). Because the SAP domain lies between amino acids 343 and 378 in MKL1, we wondered whether this specific MKL1 domain contributed to the interaction; however, FLNA still interacted with an MKL1 mutant lacking the SAP domain (Fig. 2F). Therefore, we predicted that the FLNA binding site was located in a region spanning residues 301 to 342 in MKL1. Indeed, an MKL1 Δ301–342 construct displayed diminished FLNA binding in immunoprecipitation assays (Fig. 2G), and binding further decreased with a mutant lacking amino acids 301 to 310, supporting our conclusion that these amino acids in MKL1 are required for FLNA binding (Fig. 2G).

Fig. 2 Mapping of MKL1-FLNA binding sites.

(A) Schematics of the MKL1 derivatives used for mapping MKL1-FLNA binding sites. RPEL, conserved N-terminal domain; B, basic domain; Q, glutamine-rich domain; SAP, SAF-A/B–Acinus–PIAS domain; LZ, leucine zipper–like domain; TAD, transactivation domain. (B) Table of MKL1 derivatives indicating binding (+) or no binding (−) to FLNA. (C to G) Immunoprecipitation and Western blot as indicated in A7 cells transfected with FLAG-tagged wild-type (WT) MKL1 or the indicated MKL1 mutants. BO, Sepharose beads–only control. A portion (1/30) of the cell lysate was also directly immunoblotted. (H) Immunoprecipitation and immunoblotting as indicated in A7 cells cotransfected with a HA-tagged FLNA construct (1: FLNA amino acids 276 to 570; 2: FLNA amino acids 571 to 866; 3: FLNA amino acids 1155 to 1442; 4: FLNA amino acids 1779 to 2284; 5: FLNA amino acids 2285 to 2729) and FLAG-MKL1. Blots are representative of n ≥3 biological replicates.

Subsequently, we set out to define protein segments of FLNA that are essential for MKL1-FLNA interaction. FLNA contains a filamentous actin (F-actin) binding domain at the N terminus and a rod segment consisting of up to 24 homologous repeats, separated into Rod1 (repeats 1 to 15) and Rod2 (repeats 16 to 23) by two hinge domains (40). A series of vectors expressing hemagglutinin (HA)–tagged FLNA fragments spanning the FLNA protein were transfected into A7 melanoma cells together with FLAG-MKL1 cDNA. Immunoprecipitation experiments revealed that FLNA amino acids 571 to 866 and amino acids 1779 to 2284 (corresponding to repeats 4 to 7 in the Rod1 domain and repeats 16 to 18 in the Rod2 domain, respectively) were sufficient for the interaction with MKL1, whereas other FLNA regions did not contribute to the interaction (Fig. 2H and fig. S1F). Therefore, potential interaction sites appear to be located in both Rod1 and Rod2 domains of FLNA.

Correlation of MKL1-FLNA interaction with the induction and repression of MKL1-SRF target genes

We next studied whether activation or inhibition of Rho-actin signaling alters the MKL1-FLNA association. To achieve activation of Rho-actin signaling, we treated primary human and 3T3 fibroblasts with lysophosphatidic acid (LPA). LPA stimulation increased the amount of endogenous MKL1 that coimmunoprecipitated with endogenous FLNA (Fig. 3A). As expected, MKL1 largely translocated to the nucleus of primary fibroblasts that were incubated with LPA (Fig. 3B). MKL1 also accumulated in the nucleus of LPA-treated FLNA-depleted fibroblasts (fig. S1G), suggesting that MKL1 nuclear translocation upon LPA addition is not dependent on FLNA. The association of endogenous MKL1 and FLNA upon LPA treatment was accompanied by the induction of the MKL target genes SM22, SRF, CTGF, ITGA5, and CNN1 in primary and 3T3 fibroblasts (Fig. 3C and fig. S1H).

Fig. 3 Correlation of MKL1-FLNA interaction with the induction and repression of MKL1-SRF target genes.

(A) Immunoprecipitation using FLNA antibody and Western blot for the indicated endogenous proteins in 3T3 fibroblasts (left) and primary human fibroblasts (right) incubated with or without 10 μM LPA for 2 hours. A portion (1/30) of the cell lysate was also directly immunoblotted. (B) Immunofluorescence analysis of MKL1 in primary fibroblasts treated with or without LPA. 4′,6-Diamidino-2-phenylindole (DAPI) (blue), nuclear counterstain. Scale bar, 200 μm. (C) Quantitative real-time polymerase chain reaction (qRT-PCR) analysis of the mRNA expression of SM22, CTGF, ITGA5, and CNN1 in primary human fibroblasts after 2 hours of LPA stimulation compared with controls. Data are means ± SD (n = 3 experiments). *P < 0.05, **P < 0.01, unpaired Student’s t test. (D) Immunoprecipitation and Western blot as in (A) in A7 cells expressing FLAG-tagged MKL1 treated with or without 0.3 μM LatB for 45 min. (Bottom) Immunofluorescence analysis of MKL1. Scale bar, 200 μm. (E) qRT-PCR analysis of the mRNA expression of SRF, CTGF, SM22, and CNN1 in A7 cells after 45min of 0.3 μM LatB treatment compared with controls. Data are means ± SD (n = 3 experiments). *P < 0.05, **P < 0.01, unpaired Student’s t test. (F)Immunoblotting for the indicated proteins in A7 cells, M2 cells, and A7 cells expressing 50 nM FLNA siRNA (siFLNA) cells. P, phosphorylated. (G) Immunoblotting as indicated in M2 cells transfected with or without myc-FLNA. Representative images and blots (A to G) are shown for n ≥ 3 biological replicates.

We next tested whether the suppression of MKL1-SRF target genes would also be reflected in the MKL1-FLNA interaction. Treatment with the actin polymerization inhibitor latrunculin B (LatB) effectively blocked complex formation between MKL1 and FLNA and led to redistribution of MKL1 to the cytoplasm (Fig. 3D). Dissociation of the MKL1-FLNA interaction upon LatB treatment was accompanied by suppression of MKL1 target genes (Fig. 3E).

Because administration of LatB results in the formation of a repressive MKL1–G-actin complex in which MKL1 is phosphorylated (5, 41, 42), we wondered whether FLNA may switch MKL1 binding from G-actin to F-actin by altering the phosphorylation status of MKL1. Using a mobility shift in SDS gels and an antibody that specifically recognizes phosphorylated MKL1 as experimental approaches, we observed a change in mobility that was due to MKL1 phosphorylation, as verified by the phospho-specific MKL1 antibody detected in FLNA-deficient M2 and FLNA-depleted A7 cells, but not in A7 cells (Fig. 3F). In line with this observation, recombinant expression of myc-FLNA resulted in the appearance of a faster-migrating unphosphorylated form of MKL1 (Fig. 3G). These data show that FLNA suppresses MKL1 phosphorylation and may therefore counteract the repressive MKL1–G-actin complex. Furthermore, it indicates that the extent of the MKL1-FLNA interaction tightly correlated with the induction of MKL1-SRF target genes.

Identification of FLNA as a transducer of actin polymerization to SRF activity

The sensitivity of the formation of the MKL1-FLNA complex to inducers or inhibitors of actin polymerization prompted us to test whether there is a mechanistic link between actin polymerization, FLNA, and MKL1. We first tested the actin binding properties of FLNA by using actin mutants that do not polymerize (R62D-actin) or that favor actin polymerization (S14C-actin). Consistent with the crucial role of FLNA as an F-actin binding protein, FLNA bound the actin mutant enhancing actin polymerization (S14C-actin), but not the actin polymerization defective mutant (R62D-actin) (Fig. 4A).

Fig. 4 Identification of FLNA as a transducer of actin polymerization to SRF activity.

(A) Immunoprecipitation using FLNA antibody and Western blot for the indicated proteins in A7 cells transfected with FLAG-tagged R62D-actin or S14C-actin. A portion of the cell lysate (1/30)(bottom) was immunoblotted. BO, beads only. (B) Immunoprecipitation and Western blot as in (A) in A7 and M2 cells transfected with GFP-MKL1 and FLAG–R62D-actin or FLAG–S14C-actin. Separate cultures of M2 cells transfected as described were reconstituted with myc-tagged FLNA or an empty vector (EV). (C) Immunofluorescence analysis of phalloidin or FLAG in A7 cells transfected with FLAG–S14C-actin. Scale bar, 200 μm. (D) Luciferase assays for 5×SRE reporter activity in A7 or M2 cells transfected with FLAG–S14C-actin or EV along with a 5×SRE reporter gene and a Renilla luciferase internal control vector (pRL-SV40P). Data are means ± SD (n = 3). *P < 0.05, unpaired Student’s t test. (E) Luciferase assays performed as in (D) in A7 and M2 cells treated with or without 0.5 μM Jasplakinolide (Jasp.) for 7 hours. **P < 0.01, unpaired Student’s t test. (F) Luciferase assays as in (D) in M2 cells expressing increasing amounts of FLNA expression vector and a 5×SRE reporter gene. (G) Immunofluorescence analysis of phalloidin and lamin A/C in A7 and M2 cells cultured in 0.5% fetal calf serum for 16 hours before serum stimulation for 2 min. Scale bar, 10 μm. (Right) Filament length was measured from 20 randomly chosen nuclei of either A7 or M2 cells. Representative images and blots (A to G) are depicted for n ≥ 3 biological replicates.

Surprisingly, we found that FLNA was required for an association between the actin mutant favoring actin polymerization (S14C-actin) and MKL1. S14C-actin could be coimmunoprecipitated with MKL1 in FLNA-expressing cells, but not in FLNA-deficient cells (Fig. 4B). Binding of the nonpolymerizable actin mutant (R62D-actin) was detectable in both FLNA-expressing and FLNA-deficient cells (Fig. 4B). To underline the specificity of the obtained results, we reintroduced FLNA into M2 cells, which rescued S14C-actin–MKL1 binding (Fig. 4B).

These data suggest that FLNA may mediate an association between polymerized actin and MKL1 and thereby transduce the signal of polymerized actin to SRF activation. To further investigate this concept, we performed reporter gene assays with a 5×SRE reporter gene and S14C-actin, whose expression favors F-actin formation. FLAG-tagged S14C-actin enabled the identification of transfected cells by immunocytochemistry with FLAG antibody and phalloidin, which specifically binds F-actin. Compared to surrounding untransfected cells, cells transfected with S14C-actin displayed enhanced actin polymerization (Fig. 4C). We obtained a 13-fold induction of luciferase activity in S14C-actin and FLNA-expressing cells. In contrast, there was only a negligible increase in luciferase expression in S14C-actin–expressing FLNA-deficient cells (Fig. 4D). These data suggest that FLNA plays an important role in actin-dependent SRF activation. The same result was obtained after actin polymerization by the administration of Jasplakinolide (Fig. 4E) or by the expression of a constitutively active version of the formin mDia (fig. S1I). SRF activity was restored, however, when M2 cells were reconstituted with FLNA (Fig. 4F).

MKL1-SRF activation has been shown to require nuclear actin polymerization, driven by the formin mDia (6). We therefore sought to clarify whether nuclear actin polymerization requires FLNA. Only FLNA-expressing A7 cells were able to efficiently form nuclear actin filaments after serum stimulation and overexpression of S14C-actin (Fig. 4G and fig. S1J), as determined by immunostaining with phalloidin and the nuclear envelope marker lamin A/C and by quantitation of the number of actin filaments, suggesting that FLNA is necessary for nuclear actin polymerization to mediate MKL1 activation. Together, these results are consistent with the concept that FLNA couples actin polymerization to MKL1-SRF transcriptional activation.

Interaction of FLNA and MKL1 in cell migration and invasion

Knockdown of MKL1 and MKL2 in hepatocellular and mammary carcinoma cells has been recently reported to impair cell migration (25, 37). Thus, we next sought to clarify whether FLNA plays an important role in MKL1-mediated cellular functions such as migration.

We first assessed whether RNA interference (RNAi)–mediated depletion of MKL1 and FLNA impairs cell migration. Expression of MKL1 and FLNA at the protein level was reduced by 90% (Fig. 5A, bottom). Wound healing assays revealed a strong decrease in cell motility in the absence of MKL1, resembling the effect of small interfering RNA (siRNA)–mediated FLNA suppression on cell motility (Fig. 5A) (39). Reintroduction of siRNA-resistant wild-type MKL1, but not siRNA-resistant mutant MKL1 Δ301–380 or Δ301–342 that was unable to interact with FLNA, partially restored the migratory capacity of A7 melanoma cells depleted of endogenous MKL1 (Fig. 5B), suggesting that MKL1 orchestrates cell migration in concert with FLNA. The same result was obtained from HuH7 hepatocellular carcinoma cells (fig. S1K).

Fig. 5 Interaction of FLNA and MKL1 in cell migration and invasion.

(A) Cell migration (assessed by a culture scratch-wound assay) by A7 cells transfected with 50 nM negative control siRNA (ctrl), 50 nM FLNA siRNA (siFLNA), or 50 nM MKL1 siRNA (siMKL1). Data are means ± SD (n = 3 experiments). **P < 0.01, ***P < 0.001. (Right) Representative images. (Bottom) Knockdown efficiencies. (B) Cell migration assessed as in (A) by A7 cells transfected with 50 nM negative control siRNA (ctrl) or 50 nM MKL1 siRNA (siMKL1) and reconstituted with FLAG-tagged WT MKL1, MKL1 Δ301–380, or MKL1 Δ301–342, confirmed by Western blot. Data are means ± SD (n = 3). *P < 0.05. (C) Cell invasion, assessed by a three-dimensional Matrigel invasion assay, by A7 cells transfected with 50 nM negative control siRNA (ctrl) or 50 nM MKL1 siRNA (siMKL1) and reconstituted with FLAG-tagged WT or mutant MKL1. The statistical significance of (A) and (B) was calculated using unpaired Student’s t test and closed testing.

To study the effects of FLNA-MKL1 interaction on invasive cell migration, we performed Transwell invasion assays. A7 cells were allowed to penetrate a three-dimensional Matrigel in response to a gradient of conditioned medium. As a result, invasive migration of A7 cells was strongly reduced by the absence of MKL1 (Fig. 5C). Reconstitution with siRNA-resistant wild-type MKL1 enhanced the invasive behavior of MKL1 siRNA-expressing A7 cells. In contrast, invasive cell migration remained nearly unchanged upon reconstitution with siRNA-resistant MKL1 Δ301–380 (Fig. 5C). Together, these results indicate that FLNA promotes MKL1-dependent cell motility.

Interaction of FLNA and MKL1 in the expression of MKL1 target genes

To delineate the impact of FLNA-MKL1 interaction on the transcriptional activity of MKL1, we first investigated the effect of FLNA on the expression of established MKL1 target genes, such as SRF, CTGF, SM22, and ITGA5, in FLNA-expressing A7 cells and FLNA-deficient M2 cells. Expression of SRF, SM22, CTGF, and ITGA5 was strongly inhibited in M2 cells lacking FLNA (Fig. 6A).

Fig. 6 Interaction of FLNA and MKL1 in the expression of MKL1 target genes.

(A) A7 and M2 cells were subjected to qRT-PCR using SRF, CTGF, SM22, and ITGA5 primers. Data are means ± SD (n = 3 experiments). *P < 0.05, **P < 0.01, ***P < 0.001. rel, relative to 18S rRNA. (B and C) qRT-PCR as in (A) for SRF, SM22, CTGF, and ITGA5 (B) or GLIPR1, CNN1, MYH9, and FHL2 mRNA expression in primary MEFs (B) and A7 cells (C) transfected with negative control siRNA (ctrl) or 50 nM FLNA siRNA (siFLNA). Data are means ± SD (n = 3). **P < 0.01, ***P < 0.001. (D) CTGF mRNA expression, determined by qRT-PCR as in (A), in M2 cells expressing negative control siRNA (ctrl), 50 nM MKL1 siRNA (siMKL1), or 50 nM SRF siRNA (siSRF) and reconstituted with myc-FLNA (+FLNA) or empty vector (EV). Data are means ± SD (n = 3). ***P < 0.001. (E) SM22 mRNA expression by qRT-PCR either in A7 cells expressing negative control siRNA (ctrl) or 50 nM FLNA siRNA (siFLNA) and reconstituted with N100-MKL1, or in M2 cells transfected with N100-MKL1 (+N100) or EV. Data are means ± SD (n = 3). *P < 0.05. (F) The abundance of CTGF and SM22 mRNA in A7 cells transfected with FLAG-MKL1, MKL1 Δ301–380, MKL1 Δ301–342, MKL1 Δ301–310, or EV for 48 hours, as analyzed by qRT-PCR. Data are means ± SD (n = 3). **P < 0.01. (G) FLNA recruitment to the CTGF promoter. ChIP was performed using A7 and M2 cells and a specific antibody against FLNA. Enrichment of the CTGF and GAPDH promoters was quantified by qRT-PCR. Data are means ± SD (n= 3 independent chromatin preparations). **P < 0.01. The statistical significance of (A) to (G) was calculated by unpaired Student’s t test and Bonferroni adjustment.

To further corroborate that FLNA deficiency accounts for impaired MKL-dependent target gene expression, we silenced FLNA in MEFs using RNAi (Fig. 6B). Consistent with the data obtained from A7 and M2 cells, silencing of FLNA expression in MEFs resulted in a strong reduction of SRF, SM22, CTGF, and ITGA5 expression (Fig. 6B). Further assessing the impact of FLNA on MKL1-dependent target gene expression, we found that GLIPR1, CNN1, MYH9, and FHL2 are additional MKL1-dependent target genes because of their markedly reduced expression upon MKL1 depletion (fig. S2A). Expression of all of these genes was also strongly decreased after FLNA depletion (Fig. 6C), further underscoring the importance of FLNA for MKL1-dependent target gene expression. Similar observations were made for primary human and 3T3 fibroblasts, HuH7 and HepG2 hepatocellular carcinoma cells, and MDA-MB-468 mammary carcinoma cells depleted of FLNA (fig. S2, B to E), underscoring the general conclusion that FLNA is required for MKL1 activity.

Given the strong inhibitory effect of FLNA silencing, we investigated whether FLNA in turn induces MKL-SRF–dependent target gene expression. FLNA overexpression strongly induced the expression of MKL-SRF target genes, which was prevented by knocking down SRF or MKL1 (Fig. 6D and fig. S3A). These data, here exemplified for CTGF, demonstrate the SRF and MKL1 dependency of FLNA-induced target gene expression (Fig. 6D and fig. S3A). Likewise, gene targets such as SM22 induced by constitutively active MKL1 (MKL1 N100) proved to be FLNA-dependent because they were strongly reduced in FLNA-deficient cells compared to FLNA-expressing cells (Fig. 6E and fig. S3B).

To delineate the importance of the interaction between FLNA and MKL1 for the expression of MKL1 target genes, we tested whether introduction of the MKL1 mutants unable to bind FLNA would also impinge on the transcription of MKL1 target genes. Indeed, expression of SM22 and CTGF mRNA was markedly reduced in the presence of the MKL1 mutants Δ301–380, Δ301–342, and Δ301–310 (Fig. 6F).

To elucidate the underlying mechanism, we directly analyzed FLNA recruitment to the CTGF promoter in FLNA+ (A7) and FLNA (M2) cells using chromatin immunoprecipitation (ChIP) by an FLNA antibody. The CTGF promoter, but not the GAPDH promoter, was increased in FLNA immunoprecipitates (Fig. 6G), suggesting that CTGF expression is directly regulated by FLNA positioned in the immediate vicinity of its promoter.

DISCUSSION

Here, we have identified FLNA as a novel MKL1 interacting protein in a yeast two-hybrid screen. Mutational analysis revealed that amino acids 301 to 310 in MKL1 are required for the interaction with FLNA. Thus, we have identified a previously unrecognized region between MKL1 amino acids 301 and 310 that modulates the interaction with FLNA.

Mapping of the FLNA protein revealed that amino acids 571 to 866 and amino acids 1779 to 2284 are essential for the interaction with MKL1. The transcription factor FOXC1 associates with FLNA by means of the same binding sites (17) that span repeats 4 to 7 in the Rod1 domain and repeats 16 to 18 in the Rod2 domain of FLNA. The fourth repeat unfolds very easily and thereby differs from all other 24 FLNA repeats. Mechanical stress might cause unfolding of this repeat together with dissociation of binding partners (12), leading to activation of downstream signaling pathways. It is tempting to speculate that MKL1’s ability to transduce mechanical stress into gene expression (43, 44) may rely on this mechanism.

We found that the interaction of FLNA and MKL1 regulates MKL-SRF gene expression. This new concept is substantiated by the strong reduction of MKL-SRF target genes (such as SM22, CTGF, ITGA5, and MYH9) after FLNA depletion and the induction of MKL- and SRF-dependent target genes upon FLNA overexpression. Furthermore, we demonstrate that MKL1 deletion mutants that are unable to bind FLNA strongly decreased the expression of MKL target genes.

The interaction of FLNA and MKL1 proved to be essential for regulating MKL-SRF gene expression in primary murine and human fibroblasts, 3T3 fibroblasts, A7 melanoma cells, HepG2 and HuH7 hepatocellular carcinoma cells, and MDA-MB-468 mammary carcinoma cells, thereby confirming the broad significance of the results obtained. We observed that the extent of FLNA-MKL1 interaction tightly correlated with the induction and repression of MKL target genes. In primary and 3T3 fibroblasts with inactive MKL1 residing in the cytoplasm, stimulation with LPA strongly enhanced the association between MKL1 and FLNA and concomitantly induced MKL1-SRF target gene expression. MKL1 accumulated in the nucleus of LPA-treated and FLNA-depleted fibroblasts. Whether nuclear translocation of MKL1 upon LPA treatment is not dependent on FLNA or is due to residual FLNA abundance will be an important question to resolve. In contrast, treatment with LatB dissociated the MKL1-FLNA complex and reduced the expression of MKL target genes. These data are consistent with a model in which FLNA binds MKL1 to activate MKL1 target genes by two mechanisms: first, direct recruitment to their promoters and, second, interference with MKL1 phosphorylation. According to a previous study, the latter phosphorylation step serves as a prerequisite for G-actin binding and may therefore counteract the formation of the known repressive complex of MKL1 and monomeric G-actin, as illustrated in Fig. 7 (42).

Fig. 7 Model for the novel activating MKL1-FLNA complex.

MKL1 exists in (left) a repressive G-actin complex or (right) an activating FLNA complex. FLNA impairs MKL1 phosphorylation, which is a prerequisite for G-actin binding, thereby switching the repressive MKL1–G-actin complex to an MKL1-FLNA complex that transduces actin polymerization into SRF activity.

MKL target genes such as ITGA5 have been shown to mediate cell migration (38). This notion and the resemblance of the antimigratory phenotypes caused by FLNA and MKL1 depletion prompted us to investigate whether the interaction between MKL1 and FLNA might control MKL1-dependent cell motility. Compared to wild-type MKL1, the MKL1 deletion mutant that was unable to bind FLNA exhibited strongly reduced motile and invasive properties. These data corroborate that the interaction with FLNA is needed for MKL1 to execute its motile functions and invasive behavior and indicates an important role of the MKL1-FLNA interaction in tumor progression.

The interaction between MKL1 and FLNA as an F-actin binding protein in the nucleus is of particular interest because the relevance of nuclear actin dynamics to MKL1 regulation is far from being fully understood. Work by Baarlink et al. (6) showed that MKL1 activation requires nuclear actin polymerization. This finding, however, raises the pivotal question of how the signal for F-actin polymerization is transduced to initiate MKL1-SRF transcriptional activation. We found that FLNA acts as a cellular transducer linking nuclear actin polymerization with MKL1-dependent transcriptional activity. In the absence of FLNA, nuclear actin polymerization and SRF activation induced by Jasplakinolide or an actin mutant (S14C-actin) that augments F-actin do not take place.

In conclusion, we identified FLNA as an MKL1 interacting protein and an important participant in MKL1-SRF–mediated transcription. We provide evidence that the interaction between FLNA and MKL1 is required for actin-MKL1–dependent gene expression and cell migration. Therefore, binding to FLNA represents a newly identified mechanism that positively regulates MKL1 activity, thus opposing the known repressive complex of MKL1 and monomeric G-actin. In the future, disruption of the functionally relevant FLNA-MKL1 interaction might be therapeutically exploited to contain MKL1-SRF–dependent gene transcription and invasion by cancer cells.

MATERIALS AND METHODS

Cell culture, transfections, and reagents

A human FLNA-deficient melanoma cell line (M2) and the same cells restored with FLNA (A7) were gifts from W. Ziegler and were maintained as previously described (39). MEFs, HuH7 cells, 3T3 fibroblasts, and MDA-MB-468 cells were grown in Dulbecco’s modified Eagle’s medium (Sigma-Aldrich) supplemented with 10% fetal bovine serum (FBS; Invitrogen). HepG2 cells were cultivated in RPMI 1640 medium with 10% FBS. Human primary fibroblasts (HDFA) were maintained in Medium 106 including 10% low serum growth supplement (Life Technologies). For transient transfections of plasmids and siRNA, Lipofectamine 2000 (Invitrogen) was used according to the manufacturer’s instructions or the standard calcium phosphate DNA precipitation method (45). FLNA siRNA was purchased from Dharmacon, MKL1 siRNA (5′-GAAUGUGCUACAGUUGAAA[dt][dt]-3′) was purchased from Sigma-Aldrich, LatB was purchased from Merck, LPA was purchased from Sigma-Aldrich, calpain inhibitor III was purchased from AppliChem, and Jasplakinolide was purchased from Calbiochem.

Cloning and plasmids

The MKL1 coding region was expressed in p3x-FLAG-CMV-7.1. The MKL1 deletion mutants lack the region up to and including the amino acids in their names; for example, the N-terminal deletion of N300 lacks amino acids 1 to 300, the C-terminal deletion of C500 lacks amino acids 500 to 931, and the internal deletion MKL1 Δ301–310 lacks amino acids 301 to 310. Myc-FLNA expressed in pcDNA3 was provided by J. Blenis, HA-FLNA fragments in pCI-HA vector were provided by F. Berry, and the R62D-actin and S14C-actin constructs were provided by G. Posern.

Immunoprecipitation and immunoblots

For immunoprecipitation assays, cells from 6-cm dishes were rinsed once in ice-cold phosphate-buffered saline (PBS) and then lysed in 500 μl of immunoprecipitation buffer containing 10% glycerol, 150 mM NaCl, 50 mM tris-HCl, 1% Triton X-100, 0.2% phenylmethylsulfonyl fluoride, and a protease inhibitor cocktail (Calbiochem). After 45 min of incubation on ice, lysates were cleared by centrifugation at 13,000g for 10 min at 4°C. FLNA, FLAG-MKL1, or endogenous MKL1 was immunoprecipitated by adding 1 μl of antibody against FLNA (Merck Millipore and Abcam) or FLAG (Sigma-Aldrich) overnight. Fifty microliters of recombinant protein G–Sepharose (Zymed) in a 50% slurry in immunoprecipitation buffer was added, and the lysates were rotated for 3 hours at 4°C. Immunoprecipitates were collected by centrifugation at 13,000g, washed four times with immunoprecipitation buffer, and then resolved with SDS–polyacrylamide gel electrophoresis. The proteins were then transferred to polyvinylidene difluoride membranes and immunoblotted with the respective antibody. The primary antibodies used were directed against MKL1 (C-19), SRF (A-11), CTGF (L-20) (Santa Cruz Biotechnology), and DLC1 (catalog no. 612020; BD Biosciences). The phospho-specific MKL1 antibody was described previously (24). An antibody recognizing heat shock protein 90 (Santa Cruz Biotechnology) and GAPDH (Sigma-Aldrich) served as a loading control. Incubation with a horseradish peroxidase–coupled secondary antibody enabled visualization of immunoreactive proteins by enhanced chemiluminescence (SuperSignal West Femto; Thermo Scientific) on a luminescent imager (Peqlab).

Immunofluorescence

Cells were fixed with 4% paraformaldehyde in PBS for 10 min at room temperature and extracted with 0.2% Triton X-100 in PBS for 7 min. Incubation with antibodies specific for MRTF-A (C-19) (Santa Cruz Biotechnology) or FLAG (Sigma-Aldrich) was followed by incubation with secondary antibodies labeled with Alexa Fluor 488/555 dye (Invitrogen). F-actin filaments were stained with phalloidin coupled to Alexa Fluor 488 (Invitrogen), and nuclei were stained with 4′,6-diamidino-2-phenylindole (Sigma-Aldrich). Images were obtained using a confocal microscope (Zeiss).

Scratch-wound assay

Cells were allowed to grow to a confluent monolayer. After a wound was scratched with the tip of a pipette, images were acquired on a Zeiss microscope, and the mobilization of cells behind the wound edge was determined.

Luciferase reporter assay

A7 cells were transfected with the 5×SRE reporter plasmid and the Renilla luciferase simian virus 40 (SV40) reference reporter plasmid as internal control. After transfection, luciferase assays were performed using the dual luciferase system (Promega). Firefly luciferase activities were normalized to Renilla luciferase activities.

RNA extraction, cDNA synthesis, and qRT-PCR analysis

Total RNA was isolated using TRIzol reagent according to the manufacturer’s instructions (Invitrogen). cDNA was synthesized from the RNA (1 μg) with SuperScript II Reverse Transcriptase (Invitrogen). SYBR Green–based RT-PCRs were performed using the LightCycler 480 Real-Time PCR System (Roche). mRNA expression was normalized to the expression of the endogenous housekeeping control gene 18S rRNA.

Invasion assay

Cell invasion was tested using a BD BioCoat Matrigel Invasion Chamber (BD Biosciences). A7 cells were resuspended in serum-free minimum essential medium and then added to the upper chamber at a density of 40,000 cells per well. Cell invasion into the Matrigel was determined after 24 hours at 37°C. The membrane containing invading cells was fixed using methanol and stained with toluidine blue. After the removal of noninvading cells from the upper side of the membrane using sterile cotton swabs, cells were quantified by counting the cell number.

Chromatin immunoprecipitation

ChIP was performed in A7 and M2 cells. Cells were cross-linked with 1% formaldehyde and quenched with 0.125 M glycine. Nuclei were pelleted and lysed for 10 min on ice. Lysates were sonicated four times for 30 s into DNA fragments of 200 to 2000 bp. Sheared chromatin was immunoprecipitated with an antibody specific for FLNA (Merck Millipore) coupled to Dynabeads Protein G (Life Technologies) overnight at 4°C. A fraction (1%) of sonicated chromatin was separated as input without antibody. After the washing of immune complexes and the elution of the DNA of both input and ChIP samples, qRT-PCR was performed using primers specific for GAPDH and CTGF promoters.

Statistical analysis

Statistical analysis was performed using Student’s t test, Bonferroni correction, or closed testing. Unless otherwise indicated, the values are presented as means ± SD.

SUPPLEMENTARY MATERIALS

www.sciencesignaling.org/cgi/content/full/8/402/ra112/DC1

Fig. S1. Control experiments for FLNA-MKL1 interaction and its functional effects.

Fig. S2. Requirement of FLNA for MKL-SRF target gene expression in various cells.

Fig. S3. More functional data for the FLNA dependency of MKL target gene expression.

REFERENCES AND NOTES

Acknowledgments: We thank W. Ziegler for the A7 and M2 cells, F. Berry for the FLNA deletion constructs, J. Blenis for the myc-FLNA plasmid, G. Posern for the R62D/S14C-actin constructs, and P. Chinchilla for assistance with nuclear actin visualization. Funding: The work was funded by grants MU 2737/2-1 and 2737/2-2 and by Transregional Collaborative Research Center 152, Project P15, of the Deutsche Forschungsgemeinschaft. Author contributions: S.M. and T.G. conceived the study and wrote the manuscript. S.M. designed and performed the experiments together with P.K., C.H., M.N., and M.K.D., whereas R.G. assessed nuclear actin polymerization. M.F., A.S., and J.P. generated an MKL1 deletion construct. T.L. and R.P. performed the yeast two-hybrid screen. R.P. and R.G. discussed the data and corrected the manuscript. All authors read and approved the manuscript. Competing interests: The authors declare that they have no competing interests.
View Abstract

Navigate This Article