The SRF Target Gene Fhl2 Antagonizes RhoA/MALDependent Activation of SRF The MIT Faculty has made this article openly available. Please share how this access benefits you. Your story matters. Citation Philippar, Ulrike, Gerhard Schratt, Christoph Dieterich, Judith M. Müller, Petra Galgóczy, Felix B. Engel, Mark T. Keating, et al. “The SRF Target Gene Fhl2 Antagonizes RhoA/MAL-Dependent Activation of SRF.” Molecular Cell 16, no. 6 (December 2004): 867-880. Copyright © 2004 Cell Press As Published http://dx.doi.org/10.1016/j.molcel.2004.11.039 Publisher Elsevier Version Final published version Accessed Wed May 25 19:02:24 EDT 2016 Citable Link http://hdl.handle.net/1721.1/83480 Terms of Use Article is made available in accordance with the publisher's policy and may be subject to US copyright law. Please refer to the publisher's site for terms of use. Detailed Terms Molecular Cell, Vol. 16, 867–880, December 22, 2004, Copyright ©2004 by Cell Press The SRF Target Gene Fhl2 Antagonizes RhoA/MAL-Dependent Activation of SRF Ulrike Philippar,1,5,6 Gerhard Schratt,1,6,7 Christoph Dieterich,2 Judith M. Müller,3 Petra Galgóczy,3 Felix B. Engel,4 Mark T. Keating,4 Frank Gertler,5 Roland Schüle,3 Martin Vingron,2 and Alfred Nordheim1,* 1 Abt. Molekularbiologie Institut für Zellbiologie Eberhard-Karls-Universität Tübingen 72076 Tübingen Germany 2 Max-Planck-Institut für Molekulare Genetik Berlin Germany 3 Universitäts-Frauenklinik Zentrale Klinische Forschung Klinikum der Universität Freiburg 79106 Freiburg Germany 4 Howard Hughes Medical Institute Department of Cell Biology Harvard Medical School Children’s Hospital Boston, Massachusetts 02115 5 Department of Biology Massachusetts Institute of Technology Cambridge, Massachusetts 02139 Summary RhoA signaling regulates the activity of the transcription factor SRF (serum response factor) during muscle differentiation. How RhoA signaling is integrated at SRF target promoters to achieve muscle-lineage-specific expression is largely unknown. Using large-scale expression profiling combined with bioinformatic and biochemical approaches, we identified several SRF target genes, including Fhl2, encoding a transcriptional cofactor that is highly expressed in the heart. SRF binds the Fhl2 promoter in vivo and regulates Fhl2 expression in response to RhoA activation. FHL2 protein and SRF interact physically, and FHL2 binds the promoters of SRF-responsive smooth muscle (SM) genes, but not the promoters of immediate-early genes (IEGs), in response to RhoA. FHL2 antagonizes induction of SM genes, but not IEGs or cardiac genes, by competing with the coactivator MAL/MRTF-A for SRF binding. Our findings identify an autoregulatory mechanism to selectively regulate subsets of RhoAactivated SRF target genes. Introduction Serum response factor (SRF), a MADS-box transcription factor, regulates the expression of immediate-early *Correspondence: alfred.nordheim@uni-tuebingen.de 6 These authors contributed equally to this work. 7 Present address: Division of Neuroscience, Children’s Hospital and Department of Neurobiology, Harvard Medical School, Boston, Massachusetts 02115. genes (IEGs), cytoskeletal, and muscle-specific genes (Miano, 2003; Treisman, 1995). Mouse embryos lacking SRF die before gastrulation and do not form any detectable mesoderm (Arsenian et al., 1998). SRF mediates transcriptional activation by binding to CArG box sequences in target gene promoters and by recruiting various cofactors. At IEG promoters, SRF associates with ternary complex factors (TCFs), which bind to Ets binding sites adjacent to the CArG box, to regulate transcription downstream of MAPK signaling (Buchwalter et al., 2004). Recently, SRF cofactors have been described that regulate different subsets of muscle-specific SRF target genes. Myocardin is predominantly expressed in developing cardiac and smooth muscle and activates the promoters of cardiac as well as smooth muscle (SM)specific SRF target genes (Wang et al., 2001). In addition, other members of the myocardin family, MRTF-A and -B, are expressed in muscle (Wang et al., 2002). MRTF-A/ MAL represents a potent coactivator of muscle-specific SRF target genes. MAL translocates to the nucleus upon RhoA-induced actin polymerization, thereby linking Rho signaling to SRF activation (Miralles et al., 2003). The cysteine-rich LIM-only proteins CRP1 and CRP2 serve as bridging molecules between SRF and GATA proteins and selectively induce the activation of SM promoters (Chang et al., 2003). In addition to coactivators, corepressors of SRF-dependent transcription have been identified. GATA transcription factors suppress the transcriptional activity of myocardin on a subset of SRF target promoters (Oh et al., 2004). The cardiac-specific homeodomain-only protein (HOP) associates with SRF and interferes with SRF DNA binding (Chen et al., 2002). A recent study identified the TCF Elk-1 as an antagonist of myocardin and a repressor of SRF-mediated SM gene transcription in response to MAPK signaling (Wang et al., 2004). RhoA signaling activates SRF independently of TCFs, thereby mediating activation of muscle-specific SRF target genes (Hill et al., 1995) and promoting myogenesis (Sordella et al., 2003). The LIM-only protein FHL2/DRAL/Slim3 (four-and-ahalf LIM-domain protein 2) is a member of the FHL protein family and functions as a transcriptional modulator. FHL1–3 are specifically expressed in different muscle types, whereas FHL4 and ACT (activator of CREM in testis) are exclusively found in testis (Fimia et al., 1999). FHL2 is highly expressed in early cardiac precursor cells and in the heart of adult mice (Chan et al., 1998). FHL1 is expressed in cardiac and skeletal muscle (Fimia et al., 2000). Although FHL2-deficient mice maintain normal cardiac function, they exhibit cardiac hypertrophy in response to -adrenergic stimulation (Kong et al., 2001). FHL2 has been shown to act as a transcriptional coactivator of several transcription factors including androgen receptor (Müller et al., 2000), CREB (Fimia et al., 2000), and AP-1 (Morlon and Sassone-Corsi, 2003). In addition, FHL2 negatively regulates MAPK signaling in cardiomyocytes (Purcell et al., 2004) and is an antagonist of the promyelocytic leukemia zinc finger protein (PLZF) Molecular Cell 868 (McLoughlin et al., 2002). FHL2 localizes to focal adhesions and translocates to the nucleus upon RhoA stimulation, thereby linking extracellular signals to gene control (Müller et al., 2002; Wixler et al., 2000). Here we present a microarray approach to identify new SRF target genes using overexpression of SRFVP16 in SRF-deficient embryonic stem (ES) cells, which led to the identification of SRF target genes implicated in cytoskeletal organization, apoptosis, wound healing, and muscle differentiation. We identified Fhl2 as an SRF target gene. FHL2 is upregulated in an SRF-dependent manner during ES cell differentiation and in response to RhoA activation. SRF and FHL2 interact in vitro and in vivo and bind to the promoters of a subset of SRF target genes following RhoA activation. FHL2 inhibits SRF-mediated transcription and antagonizes the RhoA/ MAL-induced activation of SM promoters by competing with MAL for SRF binding. We propose that RhoA, MAL, SRF, and FHL2 constitute an autoregulatory feedback mechanism regulating the expression of subsets of SRF target genes during myogenesis. Results Expression Profiling to Identify SRF-Regulated Genes in ES Cells Despite increasing evidence for important functions of SRF in various biological processes, further insight is hampered by incomplete knowledge of SRF target genes. We used a microarray approach to monitor SRFdependent gene expression at the whole-genome level. We took advantage of the Srf⫺/⫺ ES cell system, which allows robust induction of SRF target genes independent of signal transduction upon overexpression of the constitutively active SRF fusion protein SRF-VP16. SRF⌬M-VP16, a mutant defective in DNA binding, served as control (Schratt et al., 2002). To monitor gene expression profiles of cells transfected with SRF-VP16 or SRF-⌬MVP16, mRNA from two independent transfections was hybridized to Affymetrix microarrays (see Supplemental Table S1 at http://www.molecule.org/cgi/content/full/ 16/6/867/DC1/). Two independent Srf⫺/⫺ ES cell lines, 81 Srf⫺/⫺ and 100 Srf⫺/⫺, were used to control for cellbased variations (Weinhold et al., 2000). We considered all genes regulated by SRF whose expression was at least 3-fold higher in each of the samples derived from SRF-VP16-transfected cells, compared to SRF-⌬MVP16-transfected cells. Expression of a set of 86 genes was reproducibly induced by SRF-VP16, and Table 1 provides their functional clustering. Several genes shown in Table 1 were reported to function in different muscle lineages, others play roles in cell cycle/apoptosis, cytoskeletal organization, wound healing, or cellular metabolism. Our analysis recovered 17 previously known SRF target genes, including Egr-1, SMA (smooth muscle actin), and vinculin. However, regulation by SRF had not been demonstrated previously for several other genes, e.g., Fhl2, tuftelin, CTGF (chondrocyte tissue growth factor), keratin 17 (KRT-17), and endothelin-1 (ET-1). Taken together, we confirmed SRFregulation for 17 known SRF targets representing about 50% of all known SRF target genes, and identified a putative role for SRF in the transcriptional regulation of 69 additional genes. Since our arrays covered approximately 25% of all murine genes, we anticipate additional SRF target genes to exist. In addition, our approach misses SRF target genes, which are activated less than 3-fold or not activated by SRF-VP16, e.g., Mcl-1 (Schratt et al., 2004). SRF-VP16-Inducible Genes Display Distinct Responses to Serum Stimulation Expression of SRF-VP16 in Srf⫺/⫺ ES cells induces morphological changes (Schratt et al., 2002) which might affect gene expression indirectly. We therefore used serum stimulation of ES cells as an independent way to induce SRF target genes. Srf⫹/⫹, Srf⫺/⫺ or Srf⫺/⫺rescue ES cells were stimulated up to 180 min by serum addition in the presence of the protein synthesis inhibitor cycloheximide (Supplemental Table S2). Gene expression profiles were monitored using Affymetrix microarrays. For further analysis, we focused on the subset of 86 genes, which were also activated by SRF-VP16 (see above). Self-organizing map (SOM) clustering of the selected 86 genes yielded four different clusters, each representing a group of genes sharing a common serum induction profile (Figure 1A). Only genes found in clusters b (51 genes) and d (8 genes) displayed a robust serum induction. The remaining 27 genes in clusters a and c, although activated by SRF-VP16, were not induced strongly by serum. Genes in cluster d displayed the classical induction profile of IEGs and included the known SRF-responsive IEGs Egr-1, Egr-2, JunB, Cyr61, and ␥-actin (Figure 1B). SRF-regulated cytoskeletal genes, such as SM22␣, muscle actins and vinculin, displayed a more transient serum induction profile and were primarily found in cluster b. In total, 14 out of 17 previously known SRF target genes were found in clusters b and d, confirming our initial hypothesis that the expression of most SRF target genes would be induced by both SRF-VP16 and serum stimulation. We infer that clusters b and d are enriched for novel SRF target genes. Therefore, scoring for independent activation by SRF-VP16 and serum represents a stringent filter to identify novel bona fide SRF target genes. Identification of Conserved SRF Binding Sites by a Comparative Genomic Approach To address whether SRF directly binds to the promoters of genes activated by SRF-VP16, we screened the regulatory regions of the 86 identified genes for putative SRF consensus binding sites. A 15 kb DNA sequence upstream of the first at least partially translated exon in the genomic sequences of mouse and human was analyzed. To narrow down the search space, we only considered binding sites that localized within conserved noncoding blocks (CNBs) of mouse and human genomic sequences (Dieterich et al., 2003). By restricting our search space, we necessarily miss SRF binding sites within intronic regions and upstream regions with insufficient homology for the identification of CNBs. Within CNBs, we screened for putative SRF binding sites, allowing one base pair deviation (CArG-like) from the canonical SRF consensus sequence (CC(A/T)6GG) (CArG). CArG-like sequences often constitute functional SRF binding sites (Miano, 2003). Using these criteria, we FHL2, a Rho-Dependent Antagonist of SRF 869 Table 1. Functional Clustering of Genes Upregulated by SRF-VP16 in Two Independent Srf⫺/⫺ ES Cell Lines, 81 Srf ⫺/⫺ and 100 Srf⫺/⫺ Fold Activation Accession Number Gene 81 Srf⫺/⫺ (1) 81 Srf⫺/⫺ (2) 100 Srf⫺/⫺ (1) 100 Srf⫺/⫺ (2) SM22␣ ␣-actin (smooth) ACTA1 (skeletal) NPPB CRP-1 ACTC (cardiac) cTnC H-FABP Fhl2 CNN-1 MLC-C LPP 235.1 98.6 80.1 31.8 75.6 9.0 3.4 24.6 14.7 5.4 3.6 3.1 151.2 11.8 73.6 63.0 32.3 26.3 20.1 4.5 10.6 4.8 7.5 12.4 251.4 104.3 201.9 71.5 31.2 8.5 14.6 7.6 3.9 12.8 20.5 6.1 151.4 137.8 88.3 54.0 28.4 14.5 9.8 3.7 10.6 7.4 5.4 4.5 144.1 90.8 23.9 17.0 4.0 15.6 6.8 16.9 3.2 15.7 169.9 151.5 56.6 36.4 26.6 20.4 16.6 22.1 18.8 4.1 271.8 112.7 173.4 7.6 11.5 14.5 10.5 1.9 4.4 6.1 185.4 73.1 56.6 12.5 28.9 11.7 24.2 2.3 5.2 3.0 tissue factor CTGF endothelin-1 inhibin-betaA Cyr61 WISP-1 MuPAR1 28.5 38.7 15.8 20.5 17.0 7.3 7.5 104.4 13.5 22.6 15.8 16.6 6.2 5.4 26.5 58.6 36.9 11.3 19.6 11.0 3.3 33.9 5.2 18.6 19.8 10.3 3.9 3.1 lipocortin 1 (2x) keratin 17 (2x) annexin A2 keratin 19 keratin 18 NICE-1 annexin III vinculin perlecan gelsolin APP keratin 7 desmoplakin keratin 8 claudin-6 tropomyosin 4 TC10 procollagen, type IV CPE receptor tuftelin Fyn ␥-actin (smooth) beta-galactoside specific lectin Aim keratin 14 Arhu 101.0 17.8 179.0 13.5 22.5 11.3 26.0 70.4 3.6 15.4 25.4 4.9 23.9 14.1 12.0 23.6 5.8 16.3 7.7 4.4 4.8 4.8 7.2 6.9 6.5 3.9 66.8 77 34.8 25.4 36.0 46.1 69.3 8.5 47.3 12.9 10.9 9.7 7.6 16.2 20.0 5.3 25.6 7.9 6.5 5.6 4.7 3.1 7.2 9.2 6.6 5.4 55.8 104.2 19.9 121.0 54.2 13.0 13.2 10.9 7.8 24.0 14.3 25.8 15.2 12.0 10.6 11.1 10.9 3.5 6.1 6.4 4.6 3.1 5.8 16.5 17.0 4.3 81.3 68.4 15.6 20.3 18.6 58.8 17.0 7.3 4.6 7.5 8.1 14.1 4.4 8.1 7.0 5.7 3.1 3.7 3.8 4.4 4.5 4.4 4.6 19.0 11.6 5.7 cytochrome b-558 (2x) 5.6 11.2 7.3 4.8 Muscle Specific (12) Z68618 X13297 M12347 D16497 D88793 M15501 M29793 X14961 AF055889 U28932 AI842649 AI850370 Cell Cycle/Apoptosis (10) X81584 X71922 M28845 AF058798 U09268 M35523 M24377 U20735 X67644 Z38110 IGFBP-6 IGF-2 Egr-1 14-3-3 PAC-1 CRABP-2 Egr-2 Jun B IER3 PMP-22 Wound Healing/Angiogenesis (7) M26071 M70642 U35233 X69619 M32490 AF100777 X62700 Cytoskeleton (28) M69260 M13805 M14044 M36120 M22832 AI604345 AJ001633 AI462105 M77174 J04953 U82624 AA755126 AA600542 X15662 AF087824 AI835858 AW060401 M15832 AB000713 AF047704 AW046449 M21495 X15986 AA711704 AA606367 AW121294 Metabolism (9) M31775 (continued) Molecular Cell 870 Table 1. Continued Fold Activation Accession Number Gene 81 Srf⫺/⫺ (1) 81 Srf⫺/⫺ (2) 100 Srf⫺/⫺ (1) 100 Srf⫺/⫺ (2) malic enzyme YSPL-1 secretin Sid478p lipoprotein lipase calcyclin cathepsinB 6.8 3.5 4.0 7.3 6.2 8.4 5.8 6.4 7.0 7.4 4.3 16.4 14.0 4.2 7.0 9.4 8.0 5.2 7.6 3.8 3.7 4.5 10.4 5.8 5.0 4.0 6.7 3.3 interferon-induced 15 kDa protein PRNP (prion) Stra13/Clast5 beta2 microglobulin IGFBP-5 protease C/EBP beta nestin PSD-95 calcium channel 8 TIP-1 PolI transcription related factor cystatin MSSP IFI-30 protective protein for -galactoside serine protease, 43 kD TAP-binding protein Hes-related unknown EST KDT-1 25.3 27.6 27.7 32.5 23.4 12.5 11.1 5.6 6.9 3.1 7.3 6.4 5.0 3.4 4.2 7.4 7.9 9.8 8.7 3.9 11.1 23.3 27.9 7.5 17.9 8.1 7.2 9.4 7.5 5.7 6.0 5.1 5.0 6.5 5.3 6.5 11.6 4.9 6.3 5.4 93.6 10.2 6.0 15.0 4.0 16.9 10.0 7.9 6.3 7.9 3.6 3.9 3.8 3.5 3.0 14.0 8.3 7.4 9.6 4.3 21.2 9.6 4.3 3.8 4.7 2.4 4.3 6.9 3.1 5.6 3.8 3.3 4.9 3.7 3.1 8.4 6.9 2.5 4.2 5.7 Metabolism (9) J02652 U25739 X7380 AB025408 AA726364 X66449 M65270 Others (20) X56602 M18070 Y07836 X01838 AW125478 M61007 AW061260 AI840413 AI849587 AI842665 AI845915 U59807 AB026569 AI844520 J05261 AW228316 AI836367 AW214298 AW122893 U13371 Values represent individual activation of all genes upregulated at least 3-fold in two independent experiments, (1) and (2). Known SRF target genes are displayed in bold. identified a total of 28 putative SRF binding sites within the upstream regions of the 86 genes induced by SRFVP16 (Supplemental Table S3). This frequency (32.6%) is significantly higher (p ⫽ 1.023 ⫻ 10⫺5) than the frequency obtained if one were to pick 86 genes randomly from the mouse genome (1,864 in a total of 13,540 mouse genes, resulting in a frequency of 13.8%). If the analysis is restricted to perfect CArG consensus sequences, the difference is more pronounced (11.6% [10 out of 86 genes] versus 1.1% [151 out of 13,540 genes]; p ⫽ 4.289 ⫻ 10⫺8). This illustrates that our gene expression profiling significantly enriched for putative SRF targets. The majority of genes containing putative SRF binding sites (23 out of 28) display a serum induction profile of the clusters b and d. We found SRF binding sites in CNBs of 10 out of 17 previously known SRF targets, illustrating that the majority of functional SRF binding sites can be identified with our comparative genomics approach. Eighteen previously unrecognized SRF target genes were identified (Supplemental Table S3). Three of these genes contain a conserved consensus CArG sequence (tuftelin, Fhl2, and KRT-17). The remaining 15 genes carry CArG-like sequences (e.g., ET-1), some of which are only partially conserved between mouse and human genomes (e.g., CTGF). In summary, we identified SRF binding sites in 10 previously known and in 18 new SRF target genes using comparative genomic sequence analysis. Validation of SRF-Regulated Genes by RT-PCR and Chromatin Immunoprecipitation We next verified SRF-dependent expression for several genes identified in our screen using quantitative RTPCR. We focused on three genes containing a conserved consensus CArG box (tuftelin, Fhl2, and KRT17) and two genes containing CArG-like sequences (CTGF and ET-1). In Srf⫺/⫺ ES cells, SRF-VP16 significantly upregulated the mRNA levels of tuftelin, CTGF, KRT-17, ET-1, and Fhl2 compared to control cells (Figures 2E–2I). A similar induction was observed for the known SRF target genes Egr-1, SM22␣, and SMA (Figures 2B–2D). Expression of the housekeeping gene GAPDH did not change upon expression of SRF-VP16 (Figure 2A). Similar results were obtained using an independent Srf⫺/⫺ ES cell line (data not shown). Using quantitative RT-PCR, we validated that SRF-VP16 induced the expression of at least five of the SRF-responsive genes identified in our screen. To determine whether SRF binds directly to the identified CArG box sequences in the promoter regions of tuftelin, CTGF, KRT-17, ET-1, and Fhl2 in vivo, we performed ChIP assays (Figure 2J). Differentiated ES cells were used for the ChIP assays since some SRF target promoters, e.g., muscle-specific promoters, might not be occupied by SRF in undifferentiated ES cells (Manabe and Owens, 2001). SRF bound specifically to CArG box sequences of the known SRF target genes Egr-1, Srf, and SMA in E14 Srf⫹/⫹ ES cells (lane 6). No signal was FHL2, a Rho-Dependent Antagonist of SRF 871 Figure 1. Genes Inducible by SRF-VP16 Display Different Serum Induction Profiles (A) Serum induction profiles of genes in E14 Srf⫹/⫹ ES cells were clustered based on the self-organizing map (SOM) algorithm. Four different clusters were obtained: a, weak; b, moderate and transient; c, no; and d, strong and sustained serum induction. (B) Individual serum induction profiles of genes grouped into clusters a–d according to Figure 1A. In addition to E14 Srf⫹/⫹ ES cells, the induction kinetics for 100 Srf⫺/⫺ and 100-2 Srf⫺/⫺rescue ES cells are shown. Degree of induction is reflected by relative color coding: blue, low expression; red, high expression. The numbers to the right of the induction profiles represent the fold activation for serum-treated E14 Srf⫹/⫹ ES cells at the 60 min time point. Molecular Cell 872 Figure 2. Validation of SRF-Dependent Activation of Selected Genes Identified by Microarray Profiling (A–I) Quantitative RT-PCR was performed with mRNA from 100 Srf⫺/⫺ ES cells, transiently transfected with SRF-VP16 or SRF-⌬M-VP16, using specific primers for GAPDH (A), Egr-1 (B), SM22␣ (C), SMA (D), tuftelin (E), CTGF (F), KRT-17 (G), ET-1 (H), and Fhl2 (I). Values represent transcript levels relative to the endogenous housekeeping gene HPRT and are the mean of three independent experiments. Asterisk denotes statistically significant induction (p ⬍ 0.05, Student’s t test). (J) ChIP analysis of SRF binding to promoters of SRF target genes in vivo. Chromatin of day 8 differentiated 100 Srf⫺/⫺ and E14 Srf⫹/⫹ ES cells was immunoprecipitated using a polyclonal anti-SRF antibody (lanes 3 and 6). Bound DNA fragments were amplified by PCR using primers specific for the CArG-containing promoter regions of the indicated genes. Incubation without antibody was used as control (lanes 2 and 5). 1% of the isolated genomic DNA served as input (lanes 1 and 4). observed in 100 Srf⫺/⫺ ES cells (lane 3). SRF also bound specifically to the CArG box sequences in the tuftelin, CTGF, KRT-17, ET-1, and Fhl2 promoters (lane 6), but not to the CArG box-deficient -globin promoter. Thus, our results demonstrate that the promoter regions of tuftelin, CTGF, KRT-17, ET-1, and Fhl2 are bound by SRF in native chromatin of differentiating ES cells, identifying these genes as direct SRF target genes. FHL2, a Rho-Dependent Antagonist of SRF 873 Figure 3. Expression of FHL2 Is Regulated by SRF in a RhoA- and Differentiation-Dependent Manner (A) 81 Srf⫺/⫺ and 100 Srf⫺/⫺ ES cells were transiently transfected with SRF-VP16 or SRF-⌬M-VP16, and protein extracts were prepared 72 hr after transfection. Western blotting was performed using polyclonal anti-FHL2 or anti-␣-actinin antibody. (B) NIH3T3 cells were transfected with luciferase reporter constructs driven by the human FHL2 promoter with an intact or mutated CArG sequence (500 ng), in the presence or absence of RhoAV14 (100 ng). Values are given as fold activation over vector-transfected samples and represent the mean of four independent experiments. (C) 100 Srf⫺/⫺ and E14 Srf⫹/⫹ ES cells were differentiated by LIF removal under monolayer conditions for up to 8 days. Protein lysates were prepared on the indicated days and subjected to Western Blotting using polyclonal anti-FHL2 or anti-␣-actinin antibody. FHL2 protein is indicated by the arrow. (D) 99 Srf⫺/⫹ ES cells were differentiated in the presence of retinoic acid for up to 8 days and mRNA expression was analyzed by quantitative RT-PCR. Values represent the mean of two independent experiments ⫾ SD. SRF Regulates the Expression of FHL2 in a RhoAand Differentiation-Dependent Manner The Fhl2 gene was chosen for further study since FHL2 can function as both transcriptional coactivator and corepressor of several transcription factors and since FHL2 activity is regulated by the RhoA signaling pathway. Transient transfection of SRF-VP16 specifically induced Fhl2 protein expression in Srf⫺/⫺ ES cells (Figure 3A). Therefore, SRF-VP16 is sufficient to induce expression of endogenous Fhl2 protein in Srf⫺/⫺ ES cells. SRFdependent regulation of Fhl2 was further analyzed using luciferase reporter genes driven by the CArG box-con- Molecular Cell 874 Figure 4. SRF and FHL2 Interact In Vitro and In Vivo (A) GST pull-down assays were performed with the indicated GST fusion proteins and in vitro-translated 35S-labeled SRF. 1% of 35Slabeled SRF was loaded as input. Coomassie staining shows comparable expression of the GST fusion proteins used. (B) Mapping of the SRF interaction domain of FHL2. GST pull-down assays were performed as in (A). (C) SRF Core is sufficient to interact with FHL2. GST pull-down assay was performed with GST-FHL2 and in vitro translated 35Slabeled SRF Core (aa 133–265). 10% of 35Slabeled SRF Core was loaded as input. (D) Immunoprecipitation of 293T cells transfected with HA-SRF and Flag-FHL2 in the absence or presence of RhoAV14 was performed using M2 anti-Flag or an IgG control antibody. Immunoprecipitated SRF or FHL2 protein was detected by Western blotting with anti-SRF or anti-Flag M2 antibody, respectively. Input represents 4% of the starting protein lysate. (E) Immunoprecipitation of rat embryonic heart extracts was performed using anti-SRF or an IgG control antibody. Immunoprecipitated SRF or FHL2 protein was detected by Western blotting with anti-SRF or anti-FHL2 antibody. (F) SRF and FHL2 bind to the promoters of SM genes. Differentiated E14 Srf⫹/⫹ ES cells were transfected with HA-SRF, Flag-FHL2, and RhoAV14 as indicated. ChIP assay was performed using an anti-SRF (lanes 3 and 7) or anti-Flag antibody (lanes 4 and 8). No antibody incubation served as control (lanes 2 and 6). 1% of the isolated genomic DNA served as input (lanes 1 and 5). taining human FHL2 promoter. As a control, a CArG box mutation abolishing SRF binding was generated. In NIH3T3 cells, the FHL2 promoter construct was activated 12-fold by coexpression of constitutively active RhoAV14. CArG box mutation completely prevented RhoA-mediated activation (Figure 3B). Therefore, SRF binding to the identified CArG box in the FHL2 promoter is essential for its activation by RhoA. Upon monolayer ES cell differentiation, Fhl2 protein expression increased strongly, reaching maximum levels at day 4 (Figure 3C). This effect is SRF dependent, since Fhl2 protein expression was not induced in differentiated Srf⫺/⫺ ES cells. Similarly, ES cell differentiation in the presence of retinoic acid resulted in a gradual increase of Fhl2 mRNA expression from day 2 until day 8 (Figure 3D). Fhl1 mRNA levels increased in a comparable manner. SM22␣, SMA, and cardiac actin mRNA peaked at day 4 and decreased afterwards (Figure 3D). Therefore, expression of FHL family members precedes expression of smooth and cardiac muscle markers in differentiating ES cells, suggesting a role for FHL proteins in the regulation of SRF-dependent muscle genes. SRF and FHL2 Interact In Vitro and In Vivo Since SRF is able to interact with the LIM-only proteins CRP1 and 2, we investigated a potential direct interac- tion between SRF and FHL family members. We performed GST pull-down assays using in vitro-translated 35 S-labeled SRF and GST-FHL fusion proteins. FHL2 efficiently bound SRF, demonstrating that SRF and FHL2 interact directly in vitro (Figure 4A). No interaction of SRF was observed with GST alone. SRF interacted with FHL1 to a similar extent as with FHL2. In contrast, interaction of SRF with FHL3, FHL4, and ACT was either weak or not detectable (Figure 4A). Therefore, SRF is able to interact in vitro with at least two members of the FHL protein family, FHL1 and FHL2. To map the SRF binding domain of FHL2, we tested different FHL2 mutants in the GST pull-down assay. LIM3-4 bound SRF to a similar extent as full-length FHL2 (Figure 4B). In contrast, neither LIM0-2 nor LIM3/4 alone bound SRF. This indicates that LIM domains 3 and 4 of FHL2 are necessary and sufficient for the interaction with SRF. An SRF mutant consisting only of the DNA binding/dimerization domain (SRF Core) still interacts with FHL2 (Figure 4C). To investigate whether SRF and FHL2 interact in vivo, we performed coimmunoprecipitation assays using 293T cells transfected with HA-SRF and Flag-FHL2 in the absence or presence of RhoAV14. HA-SRF coimmunoprecipitated with Flag-FHL2 only in the presence of active RhoA (Figure 4D). To explore a potential interac- FHL2, a Rho-Dependent Antagonist of SRF 875 tion between endogenous SRF and FHL2, we performed coimmunoprecipitation assays using rat embryonic heart extracts. We detected FHL2 in anti-SRF immunoprecipitates, indicating that SRF and FHL2 interact in cells of the developing heart (Figure 4E). FHL2 and SRF colocalize in the nucleus of rat cardiomyocytes, suggesting that the FHL2/SRF interaction occurs in the nucleus (Supplemental Figure S4). Together, these experiments show that SRF and FHL2 directly interact in vitro and form a RhoAV14-dependent complex in vivo. FHL2 and SRF Bind to the Promoters of SM-Specific SRF Target Genes We used ChIP to determine whether FHL2 and SRF bind to the promoters of SRF target genes in differentiated E14 Srf⫹/⫹ ES cells (d8). As expected, SRF was bound to the promoters of the SRF target genes Egr-1, Srf, SMA, and SM22␣ (Figure 4F, lanes 3 and 7). FHL2 specifically bound to the same SM22␣ and SMA promoter regions recognized by SRF, and FHL2 binding was strongly increased upon RhoA activation (lanes 4 and 8). In contrast, FHL2 binding was not detectable at the Egr-1 and Srf promoters irrespective of RhoA activity. Neither SRF nor FHL2 were able to bind to the CArG box-deficient -globin promoter. Our results indicate that FHL2 is selectively recruited to the promoters of the SM-specific SRF target genes SMA and SM22␣ in response to RhoA signaling. FHL2 Inhibits SRF-Dependent Transcription of SM Genes We next investigated the functional consequences of FHL2 binding to SRF target promoters. FHL2 and RhoAV14 were transiently expressed in 293T cells, and the activity of SRF-dependent reporter genes was monitored. RhoAV14 expression induced activation of the SM22␣ and SMA promoters, which was significantly inhibited by FHL2 in a dose-dependent manner (Figures 5A and 5B). Mutation of CArG box 1 of the SMA promoter diminished activation by RhoAV14, whereas mutation of CArG box 2 almost completely abolished this activation. FHL2 was not able to further repress the RhoA nonresponsive SMA promoter constructs (Figure 5B). Although the cardiac-specific ANF promoter was efficiently activated by RhoAV14, no significant inhibition by FHL2 was observed, suggesting promoter selectivity by FHL2 (Figure 5C). Similarly, FHL2 failed to diminish the activity of the c-fos or tk80 promoters (Figures 5D and 5E). The FHL2 point mutant C7A fails to accumulate in the nucleus upon RhoA activation (Müller et al., 2002). In contrast to wt FHL2, expression of FHL2 C7A did not antagonize RhoA-mediated SM22␣ promoter activation (Figure 5F), indicating that inhibition of SRF-dependent transcription by FHL2 requires nuclear FHL2. To address whether SRF binding is critical for the FHL2-mediated antagonism, we used the FHL2 mutant LIM0-2, which is unable to bind SRF (Figure 4B), but efficiently translocates to nuclei upon RhoA activation (Müller et al., 2002). FHL2 LIM0-2 was no longer able to antagonize RhoA-mediated SM22␣ promoter activation in 293T cells, demonstrating that interaction with SRF is required for FHL2-mediated inhibition (Figure 5G). Equal expression of full-length and mutant FHL2 proteins was confirmed by Western blotting (data not shown). We conclude that FHL2 antagonizes RhoA-dependent activation of specific SRF target genes. FHL2 Antagonizes MAL MAL is a potent SRF-dependent activator of both smooth and cardiac muscle genes. Similar to FHL2, MAL has been shown to accumulate in the nucleus after RhoA activation. We investigated whether FHL2 could interfere with MAL activation of SRF target genes in response to RhoA in 293T cells. MAL and RhoAV14 synergistically activated the SMA, SM22␣, and ANF promoters. Coexpression of FHL2 led to a significant reduction in SM22␣ and SMA activation, but did not interfere with activation of the ANF promoter (Figures 6A–6C). Similarly, expression of FHL2 did not affect c-fos or tk80 promoter activities (Figures 6D and 6E). This suggests that FHL2 can interfere with MAL-mediated activation of a specific subset of RhoA-responsive SRF target genes, including the SM-specific genes SM22␣ and SMA. In undifferentiated Srf⫺/⫺ ES cells, expression of SRF and MAL led to an efficient induction of SMA and SM22␣ mRNA levels, which could be further increased by coexpression of RhoAV14. Simultaneous coexpression of FHL2 significantly reduced RhoA/MAL-mediated activation (Figures 6F and 6G), demonstrating that FHL2 is able to interfere with RhoA/MAL-stimulated transcription of the endogenous SM-specific SRF target genes SMA and SM22␣. In contrast, RNAi-mediated knockdown of Fhl2 had no effect on the expression of SM22␣, SMA, and cardiac actin during ES cell differentiation (Supplemental Figure S5), possibly due to functional redundancy with other FHL proteins that are expressed in ES cells and interact with SRF (Figures 3D and 4A). Competition between FHL2 and MAL for Interaction with SRF FHL2 antagonizes MAL-mediated activation of SM genes and binds to the same region of SRF as MAL (Figures 4C and 6A–6G). We used EMSA to test whether FHL2 could compete with MAL for SRF binding. Consistent with a previous report, we observed an SRF-containing ternary complex when extracts expressing an N-terminal deletion mutant, MAL⌬N, were incubated with a CArG box oligonucleotide (Miralles et al., 2003) (Figure 6H). The identity of the complexes was confirmed using in vitro-translated MAL⌬N and recombinant SRF protein (data not shown). Addition of increasing amounts of in vitro-translated FHL2 diminished the formation of the MAL⌬N/SRF/DNA complex in a dose-dependent manner (Figure 6H). We were unable, however, to detect an SRF/FHL2 complex with these EMSA conditions. This result suggests that FHL2 and MAL compete for the same binding site on SRF, thereby providing a possible mechanism for the observed inhibition of MAL-mediated activation of SRF target genes by FHL2. Discussion We describe here a microarray expression profiling study that led to the identification of several SRF-regulated genes. In particular, we show that the gene encod- Molecular Cell 876 Figure 5. FHL2 Represses SRF-Dependent SM Promoters (A–E) 293T cells were transiently transfected with the indicated luciferase promoter constructs, increasing amounts of FHL2 expression plasmid (100 ng, 500 ng, 1 g, 1.4 g), and 100 ng RhoAV14 as indicated. Values represent fold activation compared to vector control-transfected cells and are the mean of three independent experiments. Asterisk denotes statistically significant repression (p ⬍ 0.05, Student’s t test). (F) FHL2-mediated repression is dependent on nuclear localization. 293T cells were transfected with SM22␣-Luc and the FHL2 or C7A mutant expression plasmid in the presence or absence of RhoAV14. Values represent percent activation relative to RhoAV14-transfected cells and are the mean of three independent experiments. (G) FHL2-mediated repression is dependent on interaction with SRF. 293T cells were transfected with SM22␣-Luc and the indicated FlagFHL2 expression plasmid in the presence or absence of RhoAV14. Values represent percent activation relative to RhoAV14 transfected cells and are the mean of three independent experiments. Asterisk denotes statistically significant repression (p ⬍ 0.05, Student’s t test). ing the LIM-only protein FHL2 is an SRF target gene and further implicate FHL2 in the regulation of SRFdependent gene control. Functional Clusters of SRF Target Genes Our expression profiling approach uncovered a total of 59 genes whose expression is induced by both SRFVP16 and serum. Functional clustering of these genes suggests a role for the transcription factor SRF in several different processes, including cytoskeletal organization, apoptosis, and myogenesis, processes in which SRF has been previously implicated. Furthermore, we identify more SRF target genes involved in wound healing (CTGF, ET-1, KRT-17) in addition to Egr-1. This suggests that SRF regulates gene expression in response to wounding, which is in agreement with the finding that serum stimulation activates a transcriptional program related to the physiology of wound repair in fibroblasts (Iyer et al., 1999). Interestingly, several of the new SRF target genes, including CTGF and ET-1, were also identified by Iyer et al.. Our study now implicates SRF in the regulation of these genes. The role of SRF in wound healing in vivo will be addressed by conditional deletion of Srf in the mouse skin. Regulation of FHL2 Expression Several SRF target genes, including Fhl2, are preferentially expressed in muscle, emphasizing the pivotal role FHL2, a Rho-Dependent Antagonist of SRF 877 Figure 6. FHL2 Antagonizes MAL/RhoA-Mediated Activation of SM Promoters and Competes with MAL for SRF Binding (A–E) 293T cells were transiently transfected with the indicated luciferase promoter constructs, 1.4 g FHL2 expression plasmid, 100 ng RhoAV14, and 25 ng MAL as indicated. Values represent fold activation compared to vector control-transfected cells and are the mean of three independent experiments. Asterisk denotes statistically significant repression (p ⬍ 0.05, Student’s t test). (F and G) Undifferentiated 100 Srf⫺/⫺ ES cells were transiently transfected with SRF expression plasmid (100 ng). Cotransfection with MAL (20 ng), FHL2 (1.4 g), and RhoAV14 (100 ng) expression plasmids was performed as indicated. mRNA was prepared 48 hr after transfection and subjected to quantitative RT-PCR analysis. Values were normalized to the average activation by MAL as deduced from four independent experiments. Asterisk denotes statistically significant repression (p ⬍ 0.05, Student’s t test). (H) EMSA was performed using a 32P-labeled CArG box oligonucleotide, 293T cell extracts expressing MAL⌬N, and in vitro-translated FHL2. DNA complexes of SRF and SRF/MAL⌬N are indicated. of SRF in muscle differentiation and function. We identified Fhl2 to be robustly induced by both SRF-VP16 and serum in ES cells. Additional experiments con- firmed Fhl2 as a direct SRF target gene that is activated by SRF upon RhoA signaling. Several other musclerestricted SRF target genes, as well as the Srf gene Molecular Cell 878 itself, were shown to be regulated by the RhoA/actin signaling (Miano, 2003; Sotiropoulos et al., 1999). Since RhoA activity increases during myogenesis (Sordella et al., 2003), RhoA-stimulated and SRF-dependent expression of FHL2 might represent a mechanism to ensure FHL2 expression in cardiac muscle. However, since RhoA/actin signaling regulates SRF target genes in different muscle lineages, additional mechanisms must exist that confer cardiac specific FHL2 expression. In this regard, a conserved binding site for the homeodomain transcription factor Nkx2.5, an SRF partner protein and one of the earliest markers of the cardiac lineage, is present in the Fhl2 promoter (Scholl et al., 2000), suggesting regulation of FHL2 expression by SRF and Nkx2.5. Regulation of SRF Activity by Cofactor Interactions Interaction with transcriptional cofactors is widely used to regulate SRF activity. The best-studied complex is the one formed between SRF and TCFs on IEG promoters (Buchwalter et al., 2004). In addition, increasing evidence highlights the importance of interactions between SRF and cofactors on the promoters of muscle-specific genes. Myocardin/MRTFs constitute a family of potent coactivators of SRF. The ability of the SRF coactivator MRTF-A/MAL to activate SRF-regulated transcription is dependent upon RhoA-mediated nuclear translocation (Miralles et al., 2003). Similarly, we demonstrated recently that RhoA regulates the transcriptional activity of FHL2 by inducing its nuclear translocation (Müller et al., 2002). Accordingly, activation of RhoA and subsequent nuclear localization of FHL2 are required for FHL2-mediated inhibition of SRF target promoters and the observed antagonism of FHL2 and MAL. Together, these findings suggest that FHL2, like MAL, links RhoA activity to SRF-dependent gene expression. Since MAL remains nuclear after the shutdown of target gene expression (Miralles et al., 2003), FHL2 might participate in the downregulation of MAL activity after RhoA stimulation. Myocardin/MRTFs are unable to discriminate between the promoters of cardiac and SM genes (Wang et al., 2001, 2002). Therefore, mechanisms must exist that interfere with Myocardin/MRTF-activated SM gene expression in the heart. Given the abundant expression of FHL2 in the heart, it is conceivable that FHL2 functions as an inhibitor of SM genes in cardiac muscle. In addition to FHL2, other SRF corepressors are known. GATA transcription factors repress myocardin activity at some SRF target genes including SM22␣ and ANF (Oh et al., 2004). HOP-mediated repression in the heart is not restricted to SM-specific SRF targets and, therefore, likely regulates the balance between cardiomyocyte proliferation and differentiation (Chen et al., 2002; Shin et al., 2002). Recently, a role for the TCF Elk-1 in the repression of SRF-dependent SM genes was reported (Wang et al., 2004). Since Elk-1 is a target of MAPK signaling, it will be interesting to determine how both MAPK and RhoA signaling are integrated at different SRF-responsive promoters. Interestingly, in addition to FHL2-mediated inhibition of the RhoA pathway, FHL2 has been shown to inhibit MAPK signaling (Purcell et al., 2004), suggesting a regulatory contribution of FHL2 to both these signaling pathways that target SRF. Mechanism of FHL2-Mediated Target Gene Repression FHL2 inhibits the expression of a specific subset of SRF target genes after RhoA/MAL-mediated activation. Our results suggest that FHL2 and MAL compete for the same binding site on SRF. Since RhoA stimulation strongly increases FHL2 expression and nuclear localization, we hypothesize that FHL2 displaces MAL at certain SRF-dependent promoters in response to RhoA activation, leading to subsequent reduction in the expression of these genes. A similar mechanism has recently been described for the antagonism between Myocardin and Elk-1 (Wang et al., 2004). We found that the HDAC inhibitor trichostatin A did not relieve FHL2mediated repression suggesting that HDAC repressor complexes are not involved (U.P., unpublished data). The molecular basis for the promoter selectivity of FHL2mediated repression is currently unknown. However, as we observe this inhibition in the heterologous 293T cell system, we speculate that DNA sequence elements present in specific subsets of SRF-responsive promoters may mediate accessibility of SRF to FHL2. Since direct DNA binding of FHL2 has not been reported, such a sequence element would likely be either bound by a factor that specifically recruits FHL2 to promoters, e.g., SM promoters, or by a factor that specifically inhibits FHL2 binding to other SRF target genes, e.g., cardiac muscle and growth-related genes. Future work is needed to identify the components of FHL2 containing complexes at SRF target promoters. Taken together, our data suggest that SRF and FHL2 are components of an autoregulatory feedback loop that antagonizes RhoA/MAL-mediated activation of smooth muscle SRF target genes. Such mechanisms might orchestrate the lineage-specific expression of SRF target genes in different muscle tissues. Experimental Procedures Plasmids The plasmids SRF-VP16, SRF⌬M-VP16, wtSRF, RhoAV14, tk80-luc, SRE2-luc, and c-fos-luc (Schratt et al., 2002), as well as the plasmids pCMXPL2, pCMX-FHL2, pCMX-Flag-FHL2, pCMXGal-FHL2-C7A, pGEX-FHL2, pGEX-FHL2 LIM0-2, pGEX LIM3-4, pGEX LIM3, and pGEX LIM4 (Müller et al., 2000, 2002) have been described previously. pCMX-FHL2-C7A was derived from pCMXGal-FHL2-C7A. For the GST-expression plasmids, the ORFs of mFHL1, mFHL3, mFHL4, and mACT were cloned into pGEX4-T-1 (Amersham). FlagFHL2 and Flag-LIM0-2 were generated by cloning FHL2 or LIM0-2 into pcDNA6/Flag. The plasmids SM22␣-luc, ANF-luc, MRTF-A, HASRF, SMA-luc, SMASRE1pm-luc, and SMASRE2pm-luc have been described (Chang et al., 2003; Wang et al., 2002). MAL⌬N was cloned by deleting the 80 N-terminal amino acids of MRTF-A. The FHL2 promoter reporter constructs, FHL2 and FHL2mutated, were generated by PCR amplification of human genomic DNA and cloned into pGL3basic (Promega). The CArG box was mutated to the sequence CCATATAACT (mutated bases are shown in bold) resulting in FHL2mutated. Cell Culture and Transfection The ES cell lines E14 Srf⫹/⫹, 99 Srf⫺/⫹, 81 Srf⫺/⫺, 100 Srf⫺/⫺, and 100-2 Srf⫺/⫺rescue were cultured as described previously (Weinhold et al., 2000). Transient transfection of ES cells was performed using Lipofectamine 2000 (Invitrogen) as reported (Schratt et al., 2002). Serum stimulation of ES cells was done as described previously (Schratt et al., 2001), except that cycloheximide was added 30 min prior to stimulation. FHL2, a Rho-Dependent Antagonist of SRF 879 For monolayer ES cell differentiation, cells were washed twice with ES media lacking LIF and cultured in the absence of LIF. For differentiation in the presence of Retinoic Acid (RA), 5 ⫻ 10⫺7 M alltrans RA (Sigma) was added to ES media lacking LIF. 293T cells were cultured in DMEM containing 10% FCS, penicillin, streptomycin, and L-Glutamin (GIBCO-BRL). For reporter assays, 293T cells were transfected using Lipofectamine 2000 (1:2 ratio of DNA to Lipofectamine 2000). Luciferase activity was determined 48 hr after transfection. NIH3T3 cells were cultured in DMEM containing 10% FCS, penicillin, streptomycin, and L-Glutamin. For reporter assays, NIH3T3 cells were transfected using DOTAP (Roche) and luciferase activity was assayed 24 hr after transfection. Microarray Analysis Total RNA was isolated from ES cells using the RNeasy Mini Kit (Qiagen). After DNase treatment, RNA was reverse transcribed using Oligo-dT and Superscript II (Invitrogen) followed by second-strand synthesis. Double-stranded cDNA was transcribed in vitro using the Ambion T7 MegaScript Kit in the presence of biotin-labeled ribonucleotides (Enzo). Biotin-labeled cRNA was fragmented at 95⬚C for 35 min and subsequently hybridized onto AffymetrixMurine Genome U74A Gene Arrays (12,500 probe sets). We note that the U74 release of Affymetrix Gene Arrays contained a significant proportion of nonfunctional probe sets (ca. 2600). For further details, see http:// www.affymetrix.com/support/technical/product_updates/mgu74_ product_bulletin.affx. The raw expression data was scaled to obtain equal mean intensities for each individual array, and absent genes (A-calls) were omitted. Scaling factors were generally less than 1.2. Data normalization and analysis was performed using the Gene Cluster Software 2.0 (http://www.broad.mit.edu/cancer/software/genecluster2/gc2.html). For the SRF-VP16 data, genes were only considered if they showed at least a 3-fold change in expression in all of the four replicates performed. Comparative Sequence Analysis of Noncoding DNA in Orthologous Gene Loci Orthologous genes in man and mouse were defined by protein sequence comparisons following the procedure outlined by Tatusov et al. (1997). We computed pairwise best BLAST hits for all human and murine protein entries of ENSEMBL release 17. Upstream regions of putative orthologs were scanned for local sequence similarities. 15 kB of DNA sequences upstream of the start of translation (ATG) were retrieved for all human and murine transcripts. Mammalianwide repeats and low-complexity regions were excluded from the analysis. Significant local sequence similarities were detected by computing suboptimal local alignments. Further details can be found elsewhere (Dieterich et al., 2003). Subsequently, we predicted putative SRF binding sites by scanning all conserved noncoding blocks for the consensus SRF binding motif, the CArG box CC(A/T)6GG. One mismatch to the motif consensus was allowed in either species. Quantitative RT-PCR Total RNA preparation, cDNA synthesis, and quantitative PCR using the SYBR Green technology (Perkin Elmer) were performed as described previously (Weinhold et al., 2000). See Supplemental Data for primer sequences. Chromatin Immunoprecipitation Chromatin immunoprecipitation (ChIP) assays were performed using a ChIP assay kit (Upstate Biotechnology Inc.) as described previously (Schratt et al., 2004). Samples were derived from E14 Srf⫹/⫹ ES cells, which were differentiated under monolayer conditions for 8 days in the absence of LIF (d8). SRF-DNA complexes were immunoprecipitated using 2 l polyclonal anti-SRF antibody (Santa Cruz). For FHL2 ChIPs, undifferentiated (d0) or differentiated (d4) E14 Srf⫹/⫹ ES cells were transiently transfected with 1.5 g HA-SRF and 5 g pCMX-Flag-FHL2 in the absence or presence of 0.5 g RhoAV14. d0 ES cells were fixed 2 days after transfection. Transfected d4 ES cells were differentiated for an additional 4 days and fixed on day 8 (d8). Flag-FHL2-DNA complexes were immunoprecipitated using 2 l M2 anti-Flag antibody (Sigma). For primer sequences see Supplemental Data. Standard PCR was performed using Platinum Taq DNA Polymerase (Invitrogen) with an annealing temperature of 65⬚C and 35 cycles. For the SM22␣ PCR, the annealing temperature was 60⬚C. For the Fhl2, SMA, and CTGF PCRs, the Advantage-GC Genomic PCR Kit (Clontech) was used. Western Blotting and Coimmunoprecipitation Cells were lysed in lysis buffer A, and Western blotting was performed according to standard procedures using polyclonal antiFHL2 (Müller et al., 2002) or monoclonal anti-␣-actinin antibody (Clone BM-75.2; Sigma). For coimmunoprecipitation, 293T cells were transfected with 1.5 g HA-SRF and 5 g pCMX-Flag-FHL2 in the absence or presence of 0.5 g RhoAV14. Three days after transfection, cells were lysed in GST-Buffer. Protein extracts were precleared using goat antimouse IgG beads (Sigma), incubated with M2 anti-Flag antibody (Sigma) or an IgG control antibody (Sigma), and immunoprecipitated with goat anti-mouse IgG beads. Eluates were analyzed by Western blotting using the polyclonal anti-SRF (Santa Cruz) and M2 antiFlag antibody. Rat embryonic hearts (Wistar rats, E19–E21; Charles River) were dissected according to standard procedures and homogenized in lysis buffer B. Protein extracts were diluted in GST-buffer, precleared using Protein A Agarose beads (Pierce) and incubated with polyclonal anti-SRF or rabbit IgG antibody (Sigma) crosslinked to Protein A Agarose beads (Pierce Seize X Protein A Immunoprecipitation Kit). Proteins were eluted and analyzed by Western Blotting using the polyclonal anti-SRF and the polyclonal anti-FHL2 antibody. See Supplemental Data for buffer compositions. GST Pull-Down Expression of the GST fusion proteins (Pharmacia) and the in vitro transcription-translation reactions (Promega) were performed according to the manufacturers’ protocols. GST pull-down assays using in vitro-translated human SRF or SRF Core were performed as described previously, except that binding reactions were at 37⬚C (Müller et al., 2000). Electrophoretic Mobility Shift Assays Electrophoretic mobility shift assays (EMSAs) were performed as described previously (Weinhold et al., 2000). A 32P-labeled DNA fragment containing the c-fos CArG box in the context of mutated TCF site was used as DNA binding probe. Acknowledgments We thank E. Olson and R. Schwartz for plasmids. U.P. received a Ph.D. scholarship by the Boehringer Ingelheim Fonds. This work was supported by the DFG (SFB 446/B7) and the Fonds der Chemischen Industrie to A.N. and by the DFG (SFB388/C9) and the MildredScheel-Stiftung (10-2019-BU I) to R.S. F.G. was supported by NIH grant GM58801. We thank T. Golub and J. Staunton for help with microarray analysis. Received: April 28, 2004 Revised: October 18, 2004 Accepted: November 22, 2004 Published: December 21, 2004 References Arsenian, S., Weinhold, B., Oelgeschlager, M., Ruther, U., and Nordheim, A. (1998). Serum response factor is essential for mesoderm formation during mouse embryogenesis. EMBO J. 17, 6289–6299. Buchwalter, G., Gross, C., and Wasylyk, B. (2004). Ets ternary complex transcription factors. Gene 324, 1–14. Chan, K.K., Tsui, S.K., Lee, S.M., Luk, S.C., Liew, C.C., Fung, K.P., Waye, M.M., and Lee, C.Y. (1998). Molecular cloning and characterization of FHL2, a novel LIM domain protein preferentially expressed in human heart. Gene 210, 345–350. Chang, D.F., Belaguli, N.S., Iyer, D., Roberts, W.B., Wu, S.P., Dong, Molecular Cell 880 X.R., Marx, J.G., Moore, M.S., Beckerle, M.C., Majesky, M.W., et al. (2003). Cysteine-rich LIM-only proteins CRP1 and CRP2 are potent smooth muscle differentiation cofactors. Dev. Cell 4, 107–118. Chen, F., Kook, H., Milewski, R., Gitler, A.D., Lu, M.M., Li, J., Nazarian, R., Schnepp, R., Jen, K., Biben, C., et al. (2002). Hop is an unusual homeobox gene that modulates cardiac development. Cell 110, 713–723. cytoskeletal organization and focal adhesion assembly in embryonic stem cells. J. Cell Biol. 156, 737–750. Schratt, G., Philippar, U., Hockemeyer, D., Schwarz, H., Alberti, S., and Nordheim, A. (2004). SRF regulates Bcl-2 expression and promotes cell survival during murine embryonic development. EMBO J. 23, 1834–1844. Dieterich, C., Wang, H., Rateitschak, K., Luz, H., and Vingron, M. (2003). CORG: a database for COmparative Regulatory Genomics. Nucleic Acids Res. 31, 55–57. Shin, C.H., Liu, Z.P., Passier, R., Zhang, C.L., Wang, D.Z., Harris, T.M., Yamagishi, H., Richardson, J.A., Childs, G., and Olson, E.N. (2002). Modulation of cardiac growth and development by HOP, an unusual homeodomain protein. Cell 110, 725–735. Fimia, G.M., De Cesare, D., and Sassone-Corsi, P. (1999). CBPindependent activation of CREM and CREB by the LIM-only protein ACT. Nature 398, 165–169. Sordella, R., Jiang, W., Chen, G.C., Curto, M., and Settleman, J. (2003). Modulation of Rho GTPase signaling regulates a switch between adipogenesis and myogenesis. Cell 113, 147–158. Fimia, G.M., De Cesare, D., and Sassone-Corsi, P. (2000). A family of LIM-only transcriptional coactivators: tissue-specific expression and selective activation of CREB and CREM. Mol. Cell. Biol. 20, 8613–8622. Sotiropoulos, A., Gineitis, D., Copeland, J., and Treisman, R. (1999). Signal-regulated activation of serum response factor is mediated by changes in actin dynamics. Cell 98, 159–169. Hill, C.S., Wynne, J., and Treisman, R. (1995). The Rho family GTPases RhoA, Rac1, and CDC42Hs regulate transcriptional activation by SRF. Cell 81, 1159–1170. Treisman, R. (1995). Journey to the surface of the cell: Fos regulation and the SRE. EMBO J. 14, 4905–4913. Tatusov, R.L., Koonin, E.V., and Lipman, D.J. (1997). A genomic perspective on protein families. Science 278, 631–637. Iyer, V.R., Eisen, M.B., Ross, D.T., Schuler, G., Moore, T., Lee, J.C., Trent, J.M., Staudt, L.M., Hudson, J., Jr., Boguski, M.S., et al. (1999). The transcriptional program in the response of human fibroblasts to serum. Science 283, 83–87. Wang, D., Chang, P.S., Wang, Z., Sutherland, L., Richardson, J.A., Small, E., Krieg, P.A., and Olson, E.N. (2001). Activation of cardiac gene expression by myocardin, a transcriptional cofactor for serum response factor. Cell 105, 851–862. Kong, Y., Shelton, J.M., Rothermel, B., Li, X., Richardson, J.A., Bassel-Duby, R., and Williams, R.S. (2001). Cardiac-specific LIM protein FHL2 modifies the hypertrophic response to beta-adrenergic stimulation. Circulation 103, 2731–2738. Wang, D.Z., Li, S., Hockemeyer, D., Sutherland, L., Wang, Z., Schratt, G., Richardson, J.A., Nordheim, A., and Olson, E.N. (2002). Potentiation of serum response factor activity by a family of myocardinrelated transcription factors. Proc. Natl. Acad. Sci. USA 99, 14855– 14860. Manabe, I., and Owens, G.K. (2001). CArG elements control smooth muscle subtype-specific expression of smooth muscle myosin in vivo. J. Clin. Invest. 107, 823–834. McLoughlin, P., Ehler, E., Carlile, G., Licht, J.D., and Schafer, B.W. (2002). The LIM-only protein DRAL/FHL2 interacts with and is a corepressor for the promyelocytic leukemia zinc finger protein. J. Biol. Chem. 277, 37045–37053. Miano, J.M. (2003). Serum response factor: toggling between disparate programs of gene expression. J. Mol. Cell. Cardiol. 35, 577–593. Miralles, F., Posern, G., Zaromytidou, A.I., and Treisman, R. (2003). Actin dynamics control SRF activity by regulation of its coactivator MAL. Cell 113, 329–342. Morlon, A., and Sassone-Corsi, P. (2003). The LIM-only protein FHL2 is a serum-inducible transcriptional coactivator of AP-1. Proc. Natl. Acad. Sci. USA 100, 3977–3982. Müller, J.M., Isele, U., Metzger, E., Rempel, A., Moser, M., Pscherer, A., Breyer, T., Holubarsch, C., Buettner, R., and Schüle, R. (2000). FHL2, a novel tissue-specific coactivator of the androgen receptor. EMBO J. 19, 359–369. Müller, J.M., Metzger, E., Greschik, H., Bosserhoff, A.K., Mercep, L., Buettner, R., and Schüle, R. (2002). The transcriptional coactivator FHL2 transmits Rho signals from the cell membrane into the nucleus. EMBO J. 21, 736–748. Oh, J., Wang, Z., Wang, D.Z., Lien, C.L., Xing, W., and Olson, E.N. (2004). Target gene-specific modulation of myocardin activity by GATA transcription factors. Mol. Cell. Biol. 24, 8519–8528. Purcell, N.H., Darwis, D., Bueno, O.F., Müller, J.M., Schüle, R., and Molkentin, J.D. (2004). Extracellular signal-regulated kinase 2 interacts with and is negatively regulated by the LIM-only protein FHL2 in cardiomyocytes. Mol. Cell. Biol. 24, 1081–1095. Scholl, F.A., McLoughlin, P., Ehler, E., de Giovanni, C., and Schafer, B.W. (2000). DRAL is a p53-responsive gene whose four and a half LIM domain protein product induces apoptosis. J. Cell Biol. 151, 495–506. Schratt, G., Weinhold, B., Lundberg, A.S., Schuck, S., Berger, J., Schwarz, H., Weinberg, R.A., Ruther, U., and Nordheim, A. (2001). Serum response factor is required for immediate-early gene activation yet is dispensable for proliferation of embryonic stem cells. Mol. Cell. Biol. 21, 2933–2943. Schratt, G., Philippar, U., Berger, J., Schwarz, H., Heidenreich, O., and Nordheim, A. (2002). Serum response factor is crucial for actin Wang, Z., Wang, D.Z., Hockemeyer, D., McAnally, J., Nordheim, A., and Olson, E.N. (2004). Myocardin and ternary complex factors compete for SRF to control smooth muscle gene expression. Nature 428, 185–189. Weinhold, B., Schratt, G., Arsenian, S., Berger, J., Kamino, K., Schwarz, H., Rüther, U., and Nordheim, A. (2000). Srf(⫺/⫺) ES cells display non-cell autonomous impairment in mesodermal differentiation. EMBO J. 19, 5835–5844. Wixler, V., Geerts, D., Laplantine, E., Westhoff, D., Smyth, N., Aumailley, M., Sonnenberg, A., and Paulsson, M. (2000). The LIM-only protein DRAL/FHL2 binds to the cytoplasmic domain of several alpha and beta integrin chains and is recruited to adhesion complexes. J. Biol. Chem. 275, 33669–33678. Accession Numbers The GEO accession numbers for the U74A microarray data sets reported in this paper are GSE1948 and GSE1949.