Myocardin-Related Transcription Factor A Activation by Competition with WH2 Domain Proteins for Actin Binding

Abstract

The myocardin-related transcription factors (MRTFs) are coactivators of serum response factor (SRF)-mediated gene expression. Activation of MRTF-A occurs in response to alterations in actin dynamics and critically requires the dissociation of repressive G-actin–MRTF-A complexes. However, the mechanism leading to the release of MRTF-A remains unclear. Here we show that WH2 domains compete directly with MRTF-A for actin binding. Actin nucleation-promoting factors, such as N-WASP and WAVE2, as well as isolated WH2 domains, including those of Spire2 and Cobl, activate MRTF-A independently of changes in actin dynamics. Simultaneous inhibition of Arp2-Arp3 or mutation of the CA region only partially reduces MRTF-A activation by N-WASP and WAVE2. Recombinant WH2 domains and the RPEL domain of MRTF-A bind mutually exclusively to cellular and purified G-actin in vitro. The competition by different WH2 domains correlates with MRTF-SRF activation. Following serum stimulation, nonpolymerizable actin dissociates from MRTF-A, and de novo formation of the G-actin–RPEL complex is impaired by a transferable factor. Our work demonstrates that WH2 domains activate MRTF-A and contribute to target gene regulation by a competitive mechanism, independently of their role in actin filament formation.

INTRODUCTION

Myocardin-related transcription factor A (MRTF-A) and its close relative MRTF-B translate changes of the actin cytoskeleton to gene transcription controlled by serum response factor (SRF) (1, 2). In turn, MRTFs regulate various cytoskeletal and cell adhesion components, thereby acting as a unique signaling node ensuring actin homeostasis and appropriate contraction, adhesion, and migration properties (3–5). Transcriptome analysis by microarrays and deep sequencing of chromatin immunoprecipitations (ChIP-Seq) identified a minimum of 683 direct actin-regulated MRTF-SRF target genes and highlighted the pathway as the major transcriptional response to serum in fibroblasts (6, 7). However, MRTF-A and -B are widely expressed and play essential, though partially redundant, roles in many tissues, including smooth muscle, myoepithelium, neuronal tissue, endothelium, and the hematopoietic system, as shown in knockout studies (8–14).

The activation of MRTF-A correlates with altered actin dynamics. Rho family GTPases and their effectors are required and sufficient for inducing MRTF-SRF-mediated transcription (1, 15). In the repressed state, monomeric G-actin binds in a 5:1 complex to the N-terminal region of MRTF-A, consisting of three RPEL motifs and the intervening linkers, which occludes a bipartite nuclear localization signal (1, 16). Structural analysis revealed that the RPEL domain binds each of the five monomers at the common interaction surface of actin between its subdomains 1 and 3, resulting in an assembly distinctively different from that of F-actin (16, 17). Transcriptionally inactive actin–MRTF-A complexes are found both outside and inside the nucleus and either inhibit the nuclear import or foster the nuclear export of MRTF-A (18–20). In parallel to activation and actin remodeling, MRTF-A dissociates from monomeric actin, at least partially (1).

The assembly of cellular F-actin is catalyzed by three groups of actin nucleators: the formin family, multidomain proteins such as Spire and Cobl, and the Arp2-Arp3 (Arp2/3) complex (21–23). Arp2/3-induced formation of branched filaments requires nucleation-promoting factors (NPF) of the WASP/WAVE family. Upon signaling, NPF recruit and activate the Arp2/3 complex via their C-terminal CA (central/connecting, acidic) regions (24–26). Adjacent to their CA regions, the widely expressed WAVE2 and N-WASP proteins have one or two WASP homology 2/verprolin homology (WH2/V) domains, respectively. The WH2 domains bind and deliver G-actin to the nascent daughter filament. They are found in ∼80 actin binding proteins and adopt the characteristic β-thymosin fold upon binding to the hydrophobic cleft between actin subdomains 1 and 3 (27, 28). Arrays of WH2 domains, such as those present in the multidomain nucleators Spire and Cobl, are thought to facilitate the formation of the energetically unfavorable actin trimers (21, 23). However, WH2 proteins also sequester actin, sever filaments, and cap barbed filament ends, thereby controlling various aspects of actin dynamics (29, 30).

Actin is the critical convergence point for activating MRTF-A and requires the dissociation of the actin-MRTF protein complex, while nuclear accumulation per se is not sufficient (18, 20, 31). It is thought that MRTF-A release is caused by G-actin depletion following polymerization, which accompanies most inducing stimuli. However, our previous work showed robust activation of MRTF-A by thymosin β4 and other barbed-end binding factors that is independent of decreases in G-actin levels (9, 32). Thus, the precise mechanism leading to the dissociation of MRTF-A from G-actin is unclear. Here we show that N-WASP and WAVE2, as well as WH2 domains isolated from N-WASP, WAVE2, Spire2, and Cobl, activate MRTF-SRF-mediated transcription by competing for G-actin binding. MRTF nuclear translocation and transcriptional activation were independent of observable F-actin polymerization and the CA region of N-WASP or WAVE2. Cell-free in vitro competition studies using purified components showed mutually exclusive binding of MRTF-A and WH2 domain-containing proteins to actin. Considerable MRTF-A activation by WH2 domains occurred in fibroblasts despite Arp2/3 inhibition and latrunculin treatment. Furthermore, MRTF-A was released from nonpolymerizable actin upon serum stimulation or was outcompeted by WH2 domains in vitro, results indicative of a MRTF-A-activating mechanism that is separable from actin dynamics.

MATERIALS AND METHODS

Plasmids and reagents.

The SRF reporter plasmids p3D.A-Luc and pRL-TK, actin expression plasmids (pEF-Flag-actin-WT [with wild-type actin] and pEF-Flag-actin-R62D), and MRTF-A plasmids have been described recently (1, 15, 31, 32). Murine thymosin β4 was cloned into pEF. Murine pEGFP-C1-N-WASP and pEGFP-C1-WAVE2 were kindly provided by Theresia Stradal and were subcloned into pEF-myc. pEF-myc-N-WASP-R474E and pEF-myc-WAVE2-R474E were generated by the single oligonucleotide mutagenesis and cloning approach (33). The pEF-myc-N-WASP-ΔA(1-484) and pEF-myc-WAVE2-ΔA(1-484) truncation constructs were generated by inserting a premature stop codon via PCR. Murine glutathione S-transferase (GST)- and green fluorescent protein (GFP)-tagged Cobl-V2(1209-1276) constructs have been described recently (21). Murine GST- and Flag-GFP-Spire2-V2(266-311) plasmids were cloned by inserting an EcoRI-XhoI fragment into pGEX-5X1 and pCMV-Tag2B-Flag-GFP, respectively. The isolated V or VCA constructs N-WASP-VV(388-458), N-WASP-VVCA(388-501), WAVE2-V(417-463), and WAVE2-VCA(417-497) were obtained by PCR and were cloned into pGEX-6P-1 (GE Healthcare, Little Chalfont, Buckinghamshire, United Kingdom) and pEGFP-C2 (Clontech Laboratories, Mountain View, CA) for bacterial and mammalian expression, respectively.

The antibodies used were anti-Flag, antihemagglutinin (anti-HA), anti-GFP, anti-GST, anti-β-actin, antitubulin (all from Sigma-Aldrich, Steinheim, Germany), anti-Myc (Cell Signaling, Cambridge, United Kingdom, and Invitrogen, Karlsruhe, Germany), anti-Arp3 (Proteintech Europe, Manchester, United Kingdom), and anti-MRTF-A (homemade rabbit antiserum [6] or monoclonal antibody 1A11 [34]). The Alexa Fluor-conjugated secondary antibodies used were from Thermo Fisher Scientific (Schwerte, Germany). F-actin was visualized with Atto 488 phalloidin (1:100, Sigma).

Cell culture and transfection.

The mouse fibroblast cell line NIH 3T3 was cultured at 37°C under 5% CO₂ in Dulbecco's modified Eagle medium (DMEM) supplemented with 10% (vol/vol) fetal calf serum (FCS), 2 mM l-glutamine, and antibiotic-antimycotic (Thermo Fisher Scientific). For serum-starved conditions, the culture medium was changed to DMEM supplemented with 0.5% (vol/vol) FCS 16 to 24 h prior to the experiment. Cells either were serum stimulated with 15% (vol/vol) FCS for 1 to 7 h or were treated with 100 μM CK-666 (Sigma) or 0.5 to 1 μM latrunculin B (Calbiochem, Nottingham, United Kingdom) for 1 to 7.5 h where indicated.

Transient transfections were carried out using X-tremeGENE 9 DNA transfection reagent (Roche, Mannheim, Germany). A total of 3 × 10⁵ cells (6-well plates) were grown overnight following a medium change and the transfection of 1 μg cDNA. Cells were lysed 24 h posttransfection in 200 μl buffer (50 mM Tris-HCl [pH 7.4], 150 mM NaCl, 2 mM EDTA, 1% [vol/vol] Triton X-100, 0.1% [vol/vol] SDS, Complete EDTA-free protease inhibitor cocktail [Roche]). The protein concentration was measured using the Micro BCA (bicinchoninic acid) protein assay kit (Thermo Fisher Scientific). Thirty picomoles of small interfering RNA (siRNA) (ON-TARGETplus mouse Actr3; GE Healthcare) was transfected using the Lipofectamine RNAiMAX reagent (Invitrogen) according to the manufacturer's instructions.

Protein purification.

GST fusion proteins were expressed in Escherichia coli BL21(DE3) Rosetta cells following induction with isopropyl-β-d-thiogalactopyranoside (IPTG). Cell pellets containing GST-V or GST-VCA constructs were resuspended in lysis buffer (1 mM dithiothreitol [DTT], Complete EDTA-free protease inhibitor cocktail in phosphate-buffered saline [PBS]), sonicated, and incubated with 1% (vol/vol) Triton X-100 for 30 min. After centrifugation at 10,800 × g for 30 min, GST-V/VCA constructs were purified from cell extracts with glutathione-Sepharose 4B beads (GE Healthcare). Cells expressing GST–MRTF-A(2-261) were lysed in TPE buffer (1% [vol/vol] Triton X-100, 100 mM EDTA, 5% [vol/vol] glycerol, 1 mM DTT, Complete EDTA-free protease inhibitor cocktail in PBS), sonicated, and cleared by centrifugation. GST–MRTF-A(2-261) was affinity precipitated from lysates with glutathione-Sepharose 4B beads (GE Healthcare). After several washing steps, the GST tag was cleaved off in 3C buffer (50 mM Tris-HCl [pH 8], 100 mM NaCl, 1 mM β-mercaptoethanol) using a 3C protease. The yield of purified proteins was quantified using the Bradford assay.

Reporter assays and reverse transcription-quantitative PCR (qRT-PCR).

A total of 7 × 10⁴ cells (12-well plates) were cotransfected with 50 ng of the p3D.A-Luc firefly luciferase reporter plasmid, 5 ng of the pRL-TK Renilla luciferase control plasmid, and 250 to 300 ng of WH2-containing constructs in a total of 500 ng of DNA. A luciferase reporter assay was performed using the Dual-Glo luciferase assay kit (Promega, Madison, WI). Firefly luciferase results were normalized to Renilla luciferase readings and are expressed as fold induction.

Endogenous target gene expression was analyzed following the isolation of total RNA using the RNeasy minikit (Qiagen, Hilden, Germany). First-strand cDNA was synthesized from 500 ng RNA using random hexamers and a Verso cDNA synthesis kit (Thermo Fisher Scientific) according to the manufacturer's instructions. qRT-PCR was performed using 1.5 μl of 1:5-diluted cDNA, 0.5 mM primers, and a DyNAmo ColorFlash SYBR green qPCR kit (Thermo Fisher Scientific) in a 10-μl reaction volume using a LightCycler 480 II instrument (Roche). Relative gene expression levels were calculated according to the 2^−ΔΔCT method. The results shown in the figures are averaged from at least three independent biological replicates.

Immunofluorescence and microscopy.

A total of 7 × 10⁴ cells (12-well plates) were grown overnight on coverslips and were fixed with 3.7% formaldehyde in PBS for 15 min, followed by extraction with 0.2% (vol/vol) Triton X-100 in PBS for 10 min and blocking in 10% FCS–1% bovine serum albumin (BSA)–0.05% Triton X-100 (vol/vol) in PBS for 30 min. Primary antibodies were diluted 1:100 to 1:1,000 and were incubated for 1 h at room temperature. Alexa Fluor-conjugated secondary antibodies were diluted 1:200 and were applied for 1 h at room temperature. Samples were covered with the ProLong Gold antifade reagent (Life Technologies) and were imaged using a TCS SP2 AOBS confocal microscope (Leica, Wetzlar, Germany) or an Axio Imager M microscope (Zeiss, Jena, Germany) equipped with a 63× oil objective and a monochrome Axiocam MRm camera.

Protein precipitation and immunoblotting.

For coimmunoprecipitation, 1 × 10⁶ cells (10-cm dish) were transfected with 5 μg cDNA in total. For isolated V/VCA domain analysis, cells were cotransfected with 2.5 μg of pEF-MRTF-f.l.-HA (with a full-length MRTF) and 2.5 μg of either pEF-Flag-actin-WT or pEF-Flag-actin-R62D. For NPF analysis, cells were cotransfected with 1.5 μg of pEF-Flag-actin-WT and 3.5 μg of pEF-myc-NPF according to the experiment. Twenty-four hours posttransfection, the medium was changed to starvation medium for 24 h. After harvesting of cells in lysis buffer (50 mM Tris-HCl [pH 7.4], 150 mM NaCl, 1 mM EDTA, 1% [vol/vol] Triton X-100, Complete EDTA-free protease inhibitor cocktail), extracts were incubated with anti-Flag M2 magnetic beads (Sigma) together with 4.8 μg purified GST tag, GST-V/VCA constructs, or MRTF-A(2-261) as indicated in the figures. Binding was carried out for 2 h at 4°C under constant rotation. Cells were lysed in 20 mM HEPES (pH 7.7), 50 mM NaCl, 1 mM EDTA, 0.5% (vol/vol) Triton X-100, and Complete EDTA-free protease inhibitor cocktail; the lysates were centrifuged at high speed (100,000 × g); and supernatants were prepared.

For GST pulldown assays, NIH 3T3 cells (10-cm dish) were lysed in 500 μl pulldown buffer (10 mM Tris-HCl [pH 8], 50 mM NaCl, 0.2 mM CaCl₂, 0.2 mM DTT, Complete EDTA-free protease inhibitor cocktail) as described previously (32). After centrifugation at 400,000 × g, G-actin extracts (200 μl) were preincubated with a truncated MRTF-A variant, MRTF-A(2-261), for 1 h at 4°C under constant rotation in pulldown buffer prior to addition to GST-V/VCA domain-coupled glutathione-Sepharose 4B beads (GE Healthcare). Molar ratios of 1:1 [MRTF-A(2-261)/GST-V/VCA constructs] and 0.4:1 (for N-WASP constructs) or 0.2:1 were used. Actin was precipitated for 1 h at 4°C under constant rotation in pulldown buffer. Five hundred nanograms of biotinylated rabbit skeletal muscle actin (Cytoskeleton Inc., Denver, CO) was incubated with purified GST tag, GST–WAVE2-V, or GST–Spire2-V2 on glutathione-Sepharose 4B beads in 10 mM Tris-HCl (pH 8), 0.2 mM CaCl₂, 0.2 mM DTT, 100 mM NaCl, 5% (vol/vol) glycerol, 0.5% (vol/vol) Triton X-100, 0.2 mM ATP, and 0.1% (wt/vol) BSA, supplemented with protease inhibitors. After increasing amounts (up to 10 μg) of MRTF-A(2-261) were added, biotinylated actin was precipitated for 2 h at 4°C under constant rotation.

Precipitates were washed at least three times, resuspended in Laemmli buffer, and analyzed by immunoblotting according to standard protocols. Primary antibodies were used at a 1:500 to 1:2,000 dilution and were incubated overnight at 4°C. Fluorophore-labeled secondary antibodies (Li-Cor, Lincoln, NE) and streptavidin-conjugated Alexa Fluor 680 (Thermo Fischer Scientific) were incubated for 1 h at room temperature. The fluorescence signals were detected with an Odyssey CLx system (Li-Cor) and were quantified by the associated software. Quantitative results were calculated from at least three independent biological experiments.

RESULTS

The critical regulatory step for MRTF-A activation is its dissociation from G-actin. It is believed that this dissociation follows changes in G-actin availability. However, the mechanism is not fully understood, given the vast excess of G-actin over MRTF-A. We therefore analyzed whether WH2/verprolin homology (V) domains, present in various nucleation-promoting factors, play a role in MRTF activation. Fragments of N-WASP, WAVE2, Cobl, and Spire2 fused to GFP were expressed in NIH 3T3 fibroblasts, and correct protein sizes and expression levels were validated (Fig. 1A and D). Upon ectopic expression, we observed that the MRTF-SRF reporter plasmid was significantly activated by all constructs over levels with the negative control, GFP (Fig. 1B). This activation occurred following transfection of the isolated WH2 domains alone, of which N-WASP-VV had the weakest effect and the second WH2 domain of Spire2 (referred to below as Spire2-V2) had the strongest. For the N-WASP and WAVE2 WH2 domains, the activation was further enhanced by the presence of the Arp2/3-activating central and acidic (CA) regions. However, thymosin β4, which resembles an individual WH2 domain without any nucleation activity, also activated MRTF-mediated transcription strongly, in line with our previous findings (9).

FIG 1.

WH2/verprolin homology (V) domains of NPF activate MRTF-SRF-dependent gene expression. NIH 3T3 cells expressing the indicated fusion proteins were serum starved for 24 h and were then analyzed for SRF target gene expression. (A) Schematics of NPF constructs and domain topology. (B) Relative MRTF-SRF luciferase reporter activity upon cotransfection with isolated V or VCA regions from several NPF fused to GFP and thymosin β4. As a control, cells transfected with a GFP vector control (GFP) were stimulated for 7 h with 15% serum (GFP + FCS). All data were normalized to the value for the starvation control, which was set to 1. Error bars, standard errors of the means (n ≥ 3). Asterisks indicate significant differences (*, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001) according to an unpaired one-sample Student t test. (C) Endogenous expression of smooth muscle α-actin (Acta2) mRNA upon transient transfection of isolated V or VCA regions relative to expression with the serum-starved control (GFP). (D) Cell lysates were immunoblotted with a GFP-specific antibody and with tubulin as a control. (E) Ectopic expression of Myc-tagged full-length N-WASP or WAVE2 increases SRF reporter activity. ctrl., control. (F) Endogenous Acta2 expression upon N-WASP or WAVE2 overexpression. (G) Myc-specific immunoblot to analyze levels of Myc-tagged full-length N-WASP or WAVE2. The apparent molecular masses in kilodaltons are indicated on the left of each immunoblot.

To analyze the induction of endogenous transcription by WH2 domains, the known MRTF-SRF target gene Acta2 was investigated. The results showed increased mRNA levels, with a pattern broadly resembling that observed for the reporter assays (Fig. 1C). Moreover, ectopic expression of Myc-tagged full-length N-WASP or WAVE2 strongly activated the luciferase reporter (Fig. 1E and G). In line with this, Acta2 mRNA expression was found to be induced by N-WASP and WAVE2 (Fig. 1F). However, the induction of endogenous Acta2 mRNA levels was generally lower than the induction of the SRF reporter, due to the transient transfection.

Transcriptional activation of target gene expression correlates with MRTF accumulation in the nuclei of fibroblasts. We therefore stained endogenous MRTF-A following transient overexpression of WH2 domain-containing constructs. Cells expressing the single WH2 domains from Cobl, Spire2, or WAVE2, the tandem WH2 domains from N-WASP, or the WAVE2 or N-WASP fragments, including the CA region, all showed increased nuclear MRTF-A staining in the absence of serum (Fig. 2A and B). Interestingly, the proportion of cells with nuclear MRTF-A correlated with the extent of reporter activity and target gene transcription, as shown in Fig. 1. This suggests that the effectiveness of the constructs for MRTF-A activation is determined by their amino acid sequences and/or the cloning approach, since the intrinsically disordered WH2 domains are not highly conserved.

FIG 2.

Nuclear accumulation of MRTF-A upon expression of NPF or their V/VCA domains. Serum-starved NIH 3T3 cells were transiently transfected with the indicated proteins and were immunostained for endogenous MRTF-A. Cells that were starved (GFP; control [ctrl.]) or stimulated with serum for 1 h (GFP + FCS; FCS) served as controls. (A) Cells expressing GFP-V or GFP-VCA. Arrows indicate GFP-expressing cells. Bars, 20 μm. (B) Quantification of MRTF-A localization by counting of 50 GFP-positive cells under each condition. Error bars, standard errors of the means (n ≥ 3). Asterisks indicate significant differences (*, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001) according to an unpaired two-sample Student t test. (C) Overexpression of Myc-tagged full-length N-WASP or WAVE2. Arrows indicate Myc-expressing cells. (D) Quantification of MRTF-A localization by counting of 50 Myc-positive cells under each condition.

We also tested the Myc-tagged full-length NPF and assessed the resulting MRTF-A localization. Full-length WAVE2 localized predominantly to the cytoplasm, while considerable amounts of N-WASP were found in the nucleus. Both nucleation-promoting factors, however, facilitated nuclear MRTF-A accumulation (Fig. 2C and D). Together, these results suggest that ectopically expressed NPF exhibit considerable activity toward MRTF-A in the cytoplasm.

WH2 domains and RPEL motifs bind actin via the same hydrophobic surface between subdomains 1 and 3. We therefore speculated that WH2-mediated MRTF-A activation occurs by direct competition between MRTF-A and WH2 domains for the actin monomer. First, to directly test this, we purified recombinant WH2 domains as GST fusion constructs and added them to actin–MRTF complexes coimmunoprecipitated from cell lysates (Fig. 1A and 3A, top). The WH2 domain of WAVE2 and its VCA fragment, as well as Spire2-V2, impaired MRTF-A binding to actin, in contrast to the negative control, GST (Fig. 3B and C). Quantification of results from at least three independent experiments showed that the amounts of MRTF-A were decreased between 20% (for WAVE2-V) and 90% (for Spire2-V2) (Fig. 3D). This shows that MRTF-A can be released from a preformed complex with actin by isolated WH2 domains and suggests that the competing effect varies depending on the identity of the WH2 domain.

FIG 3.

Isolated V and VCA regions of NPF compete with MRTF-A for actin binding. (A) Overview of the immunoprecipitation (IP) or pulldown assays. The binding of MRTF-A to Flag-tagged actin was analyzed with the addition of GST-V or -VCA domains in Flag-IP experiments. In GST pulldown assays, the binding of G-actin to purified GST-V or GST-VCA domains was challenged by adding purified MRTF-A(2-261). (B and C) Flag-actin-WT and MRTF-A-f.l.-HA were expressed in NIH 3T3 cells and were immunoprecipitated with anti-Flag magnetic beads in the presence of the indicated GST fusion proteins. Precipitated (IP) (top) and input (bottom) proteins were detected with tag-specific antibodies. (D) Quantification of MRTF-A bound to Flag-actin. Data are normalized to the value for the GST tag control (ctrl.). Error bars, standard errors of the means (n ≥ 3). Asterisks indicate significant differences (*, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001) according to an unpaired one-sample Student t test. (E) Representative GST pulldown. The G-actin lysate from NIH 3T3 cells was incubated with the purified MRTF-A(2-261) RPEL domain, followed by affinity precipitation with GST–N-WASP-VV. MRTF-A(2-261) and GST–N-WASP-VV were added at the indicated molar ratios. All lanes for GST–N-WASP-VV or β-actin are from the same immunoblot, respectively. (F) Quantification of the pulldown of endogenous actin with GST-V or GST-VCA fusion proteins. All data were normalized to the value for the control without MRTF-A(2-261) (black bars).

Second, in an inverse approach, we asked whether MRTF-A can compete with actin for binding to WH2 domains. For this purpose, we used the recombinant RPEL domain of MRTF-A, MRTF-A(2-261), and actin, exposing them to immobilized GST-WH2 domains in a pulldown approach in vitro (Fig. 3A, bottom). Upon the addition of equimolar amounts of MRTF-A(2-261) and GST–N-WASP-VV (1:1), the amount of bound actin was visibly reduced from the amount of actin precipitated by GST–N-WASP-VV alone (Fig. 3E). Considering that the RPEL domain of MRTF-A harbors as many as five G-actin binding sites, we also performed the experiment with less MRTF-A(2-261) protein, using molar ratios of 0.4:1 (MRTF-A to N-WASP-VV). Quantification of the actin precipitated by GST–N-WASP-VV and the other GST-WH2 fusion constructs revealed dose-dependent competition by MRTF-A(2-261) when N-WASP and WAVE2 constructs were used (Fig. 3F). However, no competition of MRTF-A(2-261) with the isolated Cobl and Spire2 WH2 domains was observable. In line with the results of coimmunoprecipitation experiment, this again suggests that Spire2-V2, especially, has a higher actin binding affinity than the other WH2 constructs and the MRTF-A RPEL domain.

Next, we validated a direct competition of MRTF-A and WH2 constructs by using purified recombinant proteins. We performed cell-free GST pulldown assays with GST–WAVE2-V or GST–Spire2-V2 and biotinylated actin (Fig. 4). Addition of increasing amounts of MRTF-A(2-261) significantly reduced the binding of biotinylated actin to GST–WAVE2-V (Fig. 4A and C). Furthermore, MRTF-A(2-261) also competed with the WH2 domain of Spire2, an effect visible by the increased level of unbound actin (Fig. 4B and D). Interestingly, Spire2-V2 precipitated much of the biotinylated input actin under all conditions, thereby masking the minor decrease in the level of bound actin (Fig. 4B, top). In contrast, only a fraction of the input actin was bound by the WH2 domain of WAVE2, even in the absence of MRTF-A (Fig. 4A). This indicates a significantly higher K_d (dissociation constant) for the actin–WAVE2-V complex than for actin–Spire2-V2. Furthermore, we could indeed confirm that MRTF-A(2-261) competes directly with WH2 domains for G-actin binding in the absence of additional cellular factors.

FIG 4.

Competition of the recombinant RPEL domain and the V region of WAVE2 or Spire2 for binding to purified actin in vitro. (A and B) GST pulldown of biotinylated rabbit skeletal muscle actin with GST–WAVE2-V (A) or GST–Spire2-V2 (B) in the presence of increasing amounts (0.1 to 10 μg) of MRTF-A(2-261). Precipitated (GST pulldown) and unbound (flowthrough) proteins were immunoblotted with the indicated antibodies. (C and D) Quantification of bound (C) or unbound (D) biotinylated rabbit skeletal muscle actin with GST–WAVE2-V or GST–Spire2-V2. All data were normalized to the value for the control without MRTF-A(2-261) (black bars). Error bars, standard errors of the means (n = 3). Asterisks indicate significant differences (*, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001) according to an unpaired one-sample Student t test.

To determine whether such a competitive mechanism regulates MRTF-A target gene expression in the absence of nucleation and treadmilling, we created N-WASP and WAVE2 constructs with impaired Arp2/3 recruitment and/or activation. The central and acidic (CA) regions of N-WASP and WAVE2 are involved in Arp2/3-mediated actin nucleation, and a critical role of the conserved residue R474 has been reported for N-WASP (25). We therefore compared the effects of wild-type N-WASP and WAVE2 with those of constructs with deletions of the acidic region (ΔA) or with point mutations of R474E in the C region (Fig. 1A). We found that cotransfected N-WASP-ΔA or WAVE2-ΔA efficiently reduced MRTF-A binding to Flag-actin (Fig. 5A to C). However, wild-type N-WASP had little effect on the binding of actin to MRTF-A under these circumstances, suggesting that the major transcriptional effects of the full-length protein are mediated by the CA region. Yet deletion of the acidic region, which places N-WASP in an unfolded active state, exposing its WH2 domain, was sufficient to compete away MRTF-A (Fig. 5B). This result correlated with efficient MRTF-A nuclear accumulation upon the expression of N-WASP-ΔA or WAVE2-ΔA, despite the absence of any obvious alteration in the phalloidin-positive actin cytoskeleton (Fig. 5D to G). In line with this, SRF reporter activity was considerably induced by N-WASP-ΔA and WAVE2-ΔA (Fig. 5H and I). This demonstrates that N-WASP and WAVE2 are able to activate SRF activity by releasing MRTF-A from its inhibitory actin complex, even though they lack the Arp2/3-activating acidic region.

FIG 5.

NPF compete with MRTF-A for actin binding independently of the CA region and actin remodeling. (A) Flag-actin-WT was coexpressed with N-WASP, WAVE2, or their mutants as indicated. Lysates were mixed with purified MRTF-A(2-261) and were subjected to anti-Flag immunoprecipitation. Precipitated proteins were analyzed by MRTF-A- and Flag-specific antibodies. (B and C) Quantification of MRTF-A(2-261) bound to actin in the presence of N-WASP (B) or WAVE2 (C) constructs. Data are normalized to the value for the control without NPF (ctrl.). Error bars, standard errors of the means (n = 3). Asterisks indicate significant differences (*, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001) according to an unpaired one-sample Student t test. (D to F) Analysis of endogenous MRTF-A localization and phalloidin staining upon expression of NPF mutants. (D) Control cells under starved (ctrl.) or serum-stimulated (FCS) conditions. (E and F) Overexpression of Myc-tagged N-WASP (E) or WAVE2 (F) mutant constructs. Arrows indicate Myc-expressing cells. Bars, 20 μm. (G) Quantification of MRTF-A localization (as shown in panels D to F) by counting of 50 Myc-positive cells under each condition. (H and I) Relative SRF luciferase activities of serum-starved cells expressing N-WASP (H) or WAVE2 (I). All data are normalized to the value for the starvation control, which is set to 1.

A similar effect was observed when the R474 residue of the C region was mutated to glutamate: WAVE2-R474E induced SRF activity comparable to that induced by wild-type WAVE2 (Fig. 5I), a finding that correlated with considerable MRTF competition and nuclear accumulation (Fig. 5C, F, and G). However, N-WASP-R474E induction of SRF activity was not significant (Fig. 5H), and MRTF competition was weak (Fig. 5B). Taking these results together, whereas they point to a role for the WH2 domains of N-WASP and WAVE2, the deletions/mutations made are unlikely to fully distinguish WH2-specific effects from Arp2/3-mediated contributions.

We therefore utilized a chemical Arp2/3 inhibitor, siRNA-mediated knockdown, and cytoskeletal drugs to further address the mechanism of MRTF-SRF activation. Despite inhibition of Arp2/3 by CK-666, reporter activity was increased 10- to 20-fold over that with the untreated control following serum stimulation or transfection of N-WASP or WAVE2 (Fig. 6A). In general, however, treatment with CK-666 reduced reporter activation by ∼2-fold (data not shown).

FIG 6.

Induction of MRTF-SRF activity by NPF and WH2 domains despite Arp2/3 inhibition or latrunculin treatment. (A and B) Serum-starved NPF-expressing cells were analyzed for MRTF-SRF luciferase reporter activation upon Arp2/3 inhibition by CK-666 or Arp3 siRNA. Shown is SRF activation by N-WASP and WAVE2 in comparison to that in control transfected cells without (ctrl.) or with (FCS) serum stimulation, following CK-666 treatment (100 μM, 7 h) (A) or Arp3-specific siRNA knockdown (B). (C and D) Validation of Arp3 inhibition by immunofluorescence staining for endogenous Arp3 (red) and the F-actin cytoskeleton (Atto 488 phalloidin) (green) (C) and by immunoblotting with an Arp3-specific antibody for the endogenous Arp3 protein level (D). Bars, 20 μm. (E and F) Effect of the actin-depolymerizing drug latrunculin B (LatB) (7.5 h) on SRF activity in control-transfected cells (ctrl.; GFP) or cells overexpressing NPF (E) or isolated WH2-containing constructs as GFP-V or GFP-VCA regions (F). Error bars, standard errors of the means (n = 3). Asterisks indicate significant differences (*, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001) according to an unpaired two-sample Student t test (A and B) or an unpaired one-sample Student t test (E).

Furthermore, SRF activity was still increased by serum, N-WASP, or WAVE2 despite siRNA-mediated Arp3 inhibition (Fig. 6B). Moreover, transfection with Arp3-specific siRNA did not substantially repress serum- or WAVE2-induced SRF activity, while N-WASP-mediated SRF activation was reduced by only 60% (data not shown). The efficiency of Arp2/3 inhibition was quantitatively analyzed by immunoblotting and functionally assessed by immunofluorescence staining (Fig. 6C and D). Either Arp3 siRNA or CK-666 treatment was sufficient to erase Arp3 localization at the phalloidin-positive protrusive cell edge without destroying the inner actin cytoskeleton. These results demonstrate that a considerable amount of serum-, N-WASP-, or WAVE2-induced SRF activation is independent of Arp2/3-mediated actin nucleation.

Next, we tested the effect of latrunculin B, which depolymerizes F-actin and substantially increases the G-actin pool capable of binding MRTF-A. As shown before, latrunculin B completely abolished serum-induced SRF activity (15) (Fig. 6E). Strikingly, however, latrunculin B failed to significantly impair the SRF activation mediated by N-WASP or WAVE2 (Fig. 6E). We thus tested the possibility that N-WASP and WAVE2 elicit latrunculin-insensitive SRF activation via WH2 domains. Cells were transfected with the WH2 domain constructs of N-WASP, WAVE2, Cobl, and Spire2, together with thymosin β4, and were subsequently treated with increasing amounts of latrunculin B. None of the constructs showed dose-dependent reduction of their SRF reporter activation, except for the serum stimulation control, which was efficiently blocked (Fig. 6F). These results suggest that WH2 domain-containing nucleation-promoting factors, such as N-WASP and WAVE2, induce MRTF-SRF activity independently (at least in part) of actin treadmilling.

We previously used a polymerization-deficient actin mutant, actin R62D, to show that G-actin binds to and thereby inhibits MRTF-A (1, 31). Using this mutant actin, we challenged the current model that MRTF-A release from the inhibitory G-actin complex strictly requires changes in actin treadmilling, nucleation, or filament stability. Flag-tagged actin R62D readily coimmunoprecipitated endogenous MRTF-A. Intriguingly, however, serum stimulation for 10 or 30 min led to a pronounced release of MRTF-A from actin R62D, which cannot polymerize into F-actin (Fig. 7A). This indicates that complex dissociation upon serum stimulation is not dependent exclusively on changes in actin polymerization.

FIG 7.

Impaired association of MRTF-A with monomeric actin in response to serum stimulation is mediated by a trans-acting factor. (A) Serum-starved NIH 3T3 cells expressing the nonpolymerizable mutant actin R62D were treated with FCS or 5 μM latrunculin B for the indicated times (10 min or 30 min). Lysates were analyzed for binding of endogenous MRTF-A to Flag-actin-R62D immunoprecipitates, which was detected by immunoblotting with anti-Flag and anti-MRTF-A antibodies. (B) Serum-starved NIH 3T3 cells expressing Flag-actin-R62D with or without MRTF-A-f.l.-HA were analyzed for binding of MRTF-A-f.l.-HA to actin-R62D (see also Fig. 3). (C) Recombinant MRTF-A(2-261) shows reduced de novo association with Flag-actin-WT from serum-stimulated cells. (D) Ectopic addition of serum-stimulated lysates from untransfected cells to Flag-actin-transfected cells (serum starved) decreases the association of MRTF-A(2-261) with actin. Cell extracts (supernatants) were prepared by high-speed (100,000 × g) centrifugation, and de novo formation of G-actin–MRTF-A complexes was analyzed in the presence of lysates from either serum-starved cells or cells treated with 15% FCS for 30 min. Error bars, standard errors of the means (n = 3). Asterisks indicate significant differences (*, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001) according to an unpaired one-sample Student t test.

We tested whether the complex of MRTF-A with nonpolymerizable actin R62D is dissociated by WH2-containing proteins. Isolated WH2 domains from Spire2 and Cobl indeed impaired the coimmunoprecipitation of MRTF-A with actin R62D, in contrast to the control, GST (Fig. 7B, left). Similarly, N-WASP constructs (Fig. 7B, right) and WAVE2 constructs (data not shown) with or without the Arp2/3-binding CA region strongly reduced the binding of MRTF-A to actin R62D (results are comparable to those shown in Fig. 3B and D). These results demonstrate that the release of MRTF-A from the inhibitory actin complex by WH2 domain-containing proteins can occur independently of treadmilling or nucleation.

Finally, we asked whether serum stimulation affects the ability of G-actin to associate with the recombinant RPEL domain of MRTF-A in vitro. MRTF-A(2-261) shows a much lower level of binding to Flag-actin from serum-stimulated cells than to Flag-actin from starved cells (Fig. 7C). This suggests a central role for actin, while the MRTF-A residues beyond amino acid (aa) 261 appear to be dispensable for regulated binding under these circumstances. In an attempt to analyze how actin has lost its ability to bind to MRTF-A, we asked whether a transferable factor can affect G-actin–MRTF-A association. To this end, we prepared supernatants of Flag-actin from serum-starved cells by high-speed centrifugation and measured their ability to coprecipitate recombinant MRTF-A(2-261) in the presence of ectopically added extracts from either serum-starved or serum-stimulated untransfected cells. The addition of serum-stimulated cell extracts considerably reduced the association of Flag-actin with the MRTF-A RPEL domain (as much as 40%) (Fig. 7D). This suggests the existence of an inducible trans-acting factor that mediates the dissociation of actin–MRTF-A complexes upon serum stimulation.

DISCUSSION

WH2 domains are frequently found in various actin binding proteins, where they are often involved in regulating actin dynamics. Here we provide evidence that they compete with MRTF-A for G-actin binding. Isolated WH2 domains without any nucleating activity release MRTF-A from its repressive G-actin complex, resulting in nuclear accumulation and transcriptional activation of MRTF-SRF-dependent reporters and target genes. The coimmunoprecipitation and pulldown experiments in vitro showed that this competition for G-actin works in both directions: purified WH2 domains compete with MRTF-A, and the purified RPEL domain of MRTF-A competes with WH2 domains. Structurally, this is readily explainable, since the RPEL motifs as well as the intervening linkers cover the same hydrophobic cleft on G-actin as various WH2 domains, including thymosin β4 (17, 27). Despite their opposite orientations, RPEL helix α1 can be superimposed onto WH2 helix 1 bound on actin, in accordance with mutually exclusive binding.

WH2 domain-containing NPF, such as the widely expressed proteins WAVE2 and N-WASP, showed behavior similar to that of isolated WH2 domains. WAVE2 and N-WASP are activated by the Rho family GTPases Rac1 and Cdc42, respectively, and activation of SRF-mediated transcription by N-WASP and its VCA region has been demonstrated previously (15). However, we now show, by using CA mutant constructs, that this activation is, at least in part, independent of Arp2/3-mediated F-actin nucleation. Arginine 474 of N-WASP, which we chose to mutate, facilitates tight contacts with Arp2/3 and is thought to induce structural rearrangement and mother filament binding by the Arp2/3 complex (25, 35). In addition, this residue is critical for intramolecular binding to the GTPase binding domain (GBD) and thus for autoinhibition in the absence of GTPase signaling. Therefore, the R474E mutant would be expected to exhibit an unfolded conformation, which nevertheless is unable to activate Arp2/3. However, the remaining activity of competing with and activating MRTF-A was only moderate for N-WASP-R474E and -ΔA but was more pronounced in the context of WAVE2 (Fig. 5). It remains to be determined whether this reflects a lesser contribution of the WH2 domains in the context of N-WASP, differences in WH2 affinities, or other structural reasons.

Both N-WASP and WAVE2 maintained considerable activity toward MRTF-A when Arp2/3 was inhibited by knockdown or CK-666 or upon latrunculin treatment. While latrunculin effectively blocked serum stimulation, it did not block activation by isolated WH2 domains. Since latrunculin effectively destroys filamentous actin and MRTF-A readily forms inhibitory complexes with latrunculin-bound G-actin, the remaining MRTF-A activation is probably caused by competitive binding of the WH2 domains of N-WASP or WAVE2.

As yet, it is thought that MRTF release is caused by the polymerization of G-actin into F-actin, depleting the G-actin pool available for repressive MRTF binding. This view has been consistent with the strict correlation between MRTF activation and actin remodeling by many inducing stimuli and factors, such as Rho GTPases, their effectors, and serum stimulation (1, 2, 15). However, some have suggested that this mechanism does not fully explain MRTF release. Although readily detectable, the shift of cellular G-actin to F-actin upon stimulation is minute, considering the vast molar excess of actin over MRTF-A (32). Moreover, our previous work suggested that MRTF-A activation by profilin and thymosin β4 is apparently independent of decreases in G-actin levels (9, 32).

The experiment with nonpolymerizable actin R62D now demonstrates that polymerization followed by a reduction in G-actin levels does not sufficiently explain MRTF release. Since the purified RPEL domain, MRTF-A(2-261), shows reduced binding to G-actin following stimulation, essential roles for cellular modifications of MRTF-A, such as changes in S454 phosphorylation, can be excluded in these experimental settings (36). Rather, as we propose, the dissociation of the complex may involve molecular competition with WH2-containing proteins. In line with this, we showed here that a transferable factor present in lysates of serum-stimulated cells inhibits de novo association of the RPEL domain with G-actin (Fig. 7C and D). This finding is inconsistent with a simple model of mass action via the cellular G-actin amount but explainable by a competitor acting in trans. However, other mechanisms cannot be excluded. For instance, inducible oxidation of actin, as demonstrated recently for MICAL-2, potentially also alters the binding properties of G-actin (37).

Nuclear localization of MRTF-A is not sufficient for transcriptional activity per se, since in some cell types or upon experimental treatment, the amount of MRTF-A in the nucleus is enriched, but MRTF-A remains repressed unless it dissociates from G-actin (18, 20, 31). Moreover, recent evidence showed that both monomeric and filamentous actin are present in the interphase nucleus. Extracellular stimuli, such as serum or integrin engagement, as well as Diaphanous-related formins, trigger actin filament formation in the nucleus, resulting in activated nuclear MRTF-A (38, 39). Again, actin polymerization and G-actin depletion are thought to underlie MRTF-A release during these nuclear events. Indeed, the MRTF/actin ratio in the nucleus is much higher than that in the cytoplasm, increasing the possibility of a simple depletion model. Whether molecular competition by, e.g., WH2 domains plays a role in nuclear MRTF release remains to be addressed.

Several studies have shown that WH2 domains differ significantly in their affinities for G-actin (reviewed in reference 26). Due to the fact that WH2 domains are intrinsically disordered and exhibit low sequence homology, however, it remains difficult to determine the correct domain borders (27). Sequences outside the “homology” region may well contribute to actin binding and protein stability, especially when purified proteins are used. We chose the WH2 domains we used by the following criteria: (i) they are derived from inducible NPF, (ii) they can be recombinantly expressed, or (iii) they show profound actin binding. These criteria resulted in strong actin binders, such as the second WH2 domains from Spire2 and Cobl, and weaker actin binders, such as the V regions of WAVE2 and N-WASP. Functionally, the WH2 domain of Spire2 had the strongest effect on competition and MRTF-A activation in our experiments, followed by Cobl, WAVE2, and N-WASP. For instance, Spire2-V2 readily outcompeted MRTF-A, while in turn, the RPEL domain failed to outcompete Spire2-V2 (Fig. 3D and F). In contrast, N-WASP-VV showed competition in both directions, suggesting an affinity for actin lower than that of Spire2-V2 but similar to that of MRTF-A. Nevertheless, the limitations mentioned above do not permit reliable determination of dissociation constants from our data. Previous studies, however, have measured various K_d and have shown that MRTF-A and some WH2 domains bind to actin with comparable affinity in the low micromolar range (17, 32, 35).

We have demonstrated that representative WH2 domains can activate MRTF-A by competing for G-actin binding. Previously it has been difficult to clearly distinguish polymerization-dependent from polymerization-independent competitive mechanisms of MRTF-A activation, because they are inextricable functions of the native WH2 protein. More than 80 mammalian WH2 domain proteins are known, and the number is increasing further due to bioinformatics-based predictions (40). If accessible, all of these WH2 domains bind to G-actin in a comparable fashion, despite little sequence homology. Given this, identification of critical WH2-containing proteins for a particular MRTF-stimulating signal (such as serum) might be challenging if not impossible. However, the prototypical protein with a WH2 fold used here, thymosin β4, and another barbed-end binder, profilin, are unlikely candidates for physiological MRTF-A regulators, since they are extremely abundant and not obviously controllable by extracellular stimuli. WH2-containing proteins with tissue-restricted expression, such as Spire2 and Cobl, cannot generally be responsible for MRTF-A activation, and their regulation by external stimuli is unknown (21, 23). On the other hand, widely expressed WH2 domain proteins, such as inducible NPF, could in theory serve as signal transducers. In resting cells, endogenous N-WASP and WAVE2 are autoinhibited by intramolecular folding or the pentameric WAVE regulatory complex. Upon signaling, GTPase-mediated unfolding exposes the WH2 domains, which rapidly become accessible for actin nucleation and, potentially, MRTF-A competition. Thus, our work proposes that WH2 domain proteins, especially those with inducible conformational changes, are candidates for signal transducers that contribute to the physiological activation of MRTF-A.

Funding Statement

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.