KLF3 regulates muscle-specific gene expression and synergizes with serum response factor on KLF binding sites

Abstract

This study identifies KLF3 as a transcriptional regulator of muscle genes and reveals a novel synergistic interaction between KLF3 and serum response factor (SRF). Using quantitative proteomics, KLF3 was identified as one of several candidate factors that recognize the MPEX control element in the Muscle creatine kinase (MCK) promoter. Chromatin immunoprecipitation analysis indicated that KLF3 is enriched at many muscle gene promoters (MCK, Myosin heavy chain IIa, Six4, Calcium channel receptor alpha-1, and Skeletal alpha-actin), and two KLF3 isoforms are upregulated during muscle differentiation. KLF3 and SRF physically associate and synergize in transactivating the MCK promoter independently of SRF binding to CArG motifs. The zinc finger and repression domains of KLF3 plus the MADS box and transcription activation domain of SRF are implicated in this synergy. Our results provide the first evidence of a role for KLF3 in muscle gene regulation and reveal an alternate mechanism for transcriptional regulation by SRF via its recruitment to KLF binding sites. Since both factors are expressed in all muscle lineages, SRF may regulate many striated- and smooth-muscle genes that lack known SRF control elements, thus further expanding the breadth of the emerging CArGome.

FIG. 1.

KLF3 binds the MCK promoter and other muscle gene promoters in skeletal myocytes. (A) Peptides corresponding to KLF3 are enriched in MPEX versus MPEX-mt DNA affinity-purified samples. Factors binding the MPEX site in the MCK promoter were selectively enriched from skeletal myocytes and identified by ICAT-based quantitative proteomics; the experimental details are described in our previous study (25). Peptides were analyzed by microcapillary liquid chromatography-electrospray ionization-tandem mass spectroscopy (μLC-ESI-MS/MS), and protein identifications were assigned using the SEQUEST algorithm to search a mouse protein sequence database. ICAT-labeled cysteine residues are underlined, and the dots represent sites of tryptic cleavage. The PeptideProphet probability scores for each peptide were >0.9. The relative abundance of each peptide in heavy (isolated from MPEX beads) versus normal (isolated from MPEX-mt beads) ICAT-labeled samples was calculated using XPRESS and is expressed as a ratio. (B) KLF3 in skeletal myocytes binds the MPEX sequence. Labeled mouse MPEX probe was mixed with 2 μg of skeletal myocyte nuclear extracts and analyzed via gel shift interference assay. KLF3-specific antisera reduced formation of the indicated complexes, whereas nonimmune sera (cytotoxic T lymphocytes [CTL]) had no effect. (C) KLF3 occupies muscle gene promoters in skeletal myocytes. ChIP assays were performed using MM14 skeletal myocytes and KLF3-specific antisera or preimmune sera. Immunoprecipitated chromatin was analyzed by qPCR using primers specific for the promoters of MyHCIIa, Six4, MCK, CaChR, and Skαact. The data are represented as fold enrichment (enrich.) of the indicated promoter region by KLF3-specific antisera relative to preimmune sera (CTL). Each bar represents the average and standard deviation (SD) from 3 independent ChIP experiments, with 3 replicate PCRs per experiment. The number of KLF3 consensus motifs (CACCC) and KLF3-binding MPEX motifs (based on gel shift analysis) (Fig. 3D) is shown for the region encompassed by the PCR primers and within 500 bp 5′ or 3′ of the primers.

FIG. 2.

KLF3 expression is initiated during skeletal myocyte differentiation. (A) KLF3 transcripts increase during myocyte differentiation (Diff.). RNA was isolated from undifferentiated skeletal myoblasts (0 h) and myocytes differentiated for 12, 24, 48, and 72 h. qRT-PCR was performed using primers specific for KLF3 mRNA or 18S rRNA. The data are represented as the fold change in KLF3/18S RNA relative to myoblasts. Student's t test P values were 0.01 for 0 h versus 24 h and 0.05 for 24 h versus 48 h (n = 4). The error bars represent SD. (B) KLF3 protein increases during myocyte differentiation. Cytoplasmic extracts were made from undifferentiated skeletal myoblasts and myocytes differentiated for 1 day (early diff.) and 4 days (late diff.) and subjected to Western analysis using antisera to KLF3 (top panel). The predicted size of mouse KLF3 is 38 kDa. The bands in the lower gel represent Ponceau S staining for total protein as a loading control.

FIG. 3.

Multiple KLF binding motifs are important for the activity of the MCK promoter in skeletal myocytes. (A) Sequence conservation of KLF binding motifs in the MCK promoter. The position of KLF binding motifs in the MCK enhancer (−1256 to −1050) and proximal promoter (−358 to +1) are shown (dark bars), with sequence alignment of MPEX and C(A/C)CACCC boxes within the MCK promoters from multiple mammalian species. Forward and reverse sequences are indicated, and bases that differ from the mouse sequences are underlined. (B) CAC1 and CAC2 are important for MCK promoter activity in skeletal myocytes. Skeletal myocytes were transfected with constructs containing the CAT reporter under the control of either the 358-bp MCK proximal promoter, the 80-bp MCK minimal promoter, or the equivalent constructs containing the MCK enhancer, and the PAP reference plasmid. The activities of the wild-type constructs compared to constructs containing a deletion of CAC1 or a mutation in CAC2 are shown. The data are plotted as the mean value and standard deviation of the CAT/PAP ratio determined for each culture dish, and the activity of each corresponding wild-type construct is set at 100. (C and D) KLF3 recognizes divergent MPEX sequences, but not the divergent CAC2 motif present in nonmouse species. Labeled mouse MPEX probe was mixed with 2 μg of nuclear extracts from COS-7 cells overexpressing FLAG-KLF3 and analyzed via gel shift assays. The KLF3-specific band (supershifted by antibodies to FLAG) (lane 2) is indicated. In panel C, oligonucleotides containing the MPEX and CAC2 sequences from the mouse, human, cat, dog, and bovine MCK promoters were tested alongside a mutant MPEX sequence (MT) as competitors for KLF3 binding. In panel D, oligonucleotides containing the wild-type mouse MPEX sequence (WT) or with single-base-pair changes (underlined) at each position were tested alongside a mutant MPEX sequence (MT) as competitors for KLF3 binding. (The oligonucleotides containing changes in bases 2 [C→T] and 4 [C→T] are the human and dog versions of MPEX, respectively, tested in panel C, lanes 4 and 6.) A summary of the KLF3 recognition sequence within MPEX is shown, with less stringent base requirements indicated in lowercase letters.

FIG. 4.

KLF3 and SRF synergize in transactivating the MCK promoter, and the synergy is independent of CArG sites. (A) Diagrams of the constructs used in panel B and Fig. 5 and 6. CAT reporter constructs contain the MCK enhancer linked to the proximal promoter (e−358MCKCAT), the MCK proximal promoter (−358MCKCAT), the MCK minimal promoter (−80MCKCAT), or the MCK minimal promoter containing mutations in MPEX [(MPEX-mt)−80MCKCAT], CAC2 [(CAC2-mt)−80MCKCAT], or both sites [(MPEX/CAC2-mt)−80MCKCAT]. (B) KLF3 and SRF synergize in a CArG-independent fashion. COS-7 cells were transfected with e−358MCKCAT, −358MCKCAT, or −80MCKCAT. Each reporter construct was transfected alone or with 0.4 μg of KLF3 expression vector, 1 μg of SRF expression vector, or both. The data are plotted as the mean value and standard deviation of relative CAT activity, with the activity of the reporter construct alone set at 100.

FIG. 5.

Synergy is specific to KLF3 and SRF, requiring the N terminus and zinc fingers of KLF3 and the MADS domain and TAD of SRF. (A) SRF does not synergize with KLF4. COS-7 cells were transfected with −80MCKCAT alone or with 5 μg of KLF4 expression vector, 1 μg of SRF expression vector, or both. (B) KLF3 does not synergize with MEF2. COS-7 cells were transfected with −80MCKCAT alone or with 0.4 μg of KLF3 expression vector, 0.5 μg of MEF2C expression vector, or both. (C) The MADS box and TAD of SRF are important for synergy with KLF3. COS-7 cells were transfected with (MPEX-mt)−80MCKCAT alone or with 0.4 μg of KLF3 expression vector, 1 μg of expression vectors containing various truncated versions of SRF, or both. The diagram of SRF shows the MADS box and C-terminal TAD. (D) Full-length KLF3 is required for synergy with SRF. COS-7 cells were transfected as for panel C, except that expression vectors containing various truncated versions of KLF3 were tested with and without full-length SRF. The diagram of KLF3 shows the N-terminal RD and C-terminal zinc fingers. For panels A to D, the data are plotted as the mean value and standard deviation of relative CAT activity, with the activity of the reporter construct alone set at 100.

FIG. 6.

KLF3-SRF synergy is stronger with CAC2 than with MPEX in the MCK promoter. (A and B) KLF3 binds preferentially to CAC2 over MPEX and CAC1. Labeled mouse MPEX or CAC1 probe was mixed with 2 μg of nuclear extracts from COS-7 cells overexpressing FLAG-KLF3 and analyzed via gel shift assays. The KLF3-specific complex (supershifted by antibodies to FLAG [A, lane 2] or prevented by antibodies to FLAG [B, lane 2]) is indicated. Decreasing concentrations of oligonucleotides containing mouse CAC1, CAC2, or MPEX (100-, 50-, and 25-fold molar excess over probe) were tested as competitors. CAC2 competed more strongly than MPEX or CAC1 for KLF3 binding (boxed bands in lanes 6 to 8). (C) KLF3 transactivates CAC2 more strongly than MPEX, and KLF3-SRF synergy is stronger with CAC2 versus MPEX. COS-7 cells were transfected with various reporter constructs (diagrammed in Fig. 4A) with or without KLF3 and SRF expression vectors, as in Fig. 4. The levels of activity over reporter constructs alone are indicated for KLF3-plus-SRF-transfected cells. The data are plotted as the mean value and standard deviation of relative CAT activity, with the activity of the reporter construct alone set at 100.

FIG. 7.

KLF3 physically associates with SRF. (A to C) COS-7 cells were transfected with various combinations of FLAG-KLF3, FLAG-KLF3(Δ8-89), FLAG-KLF3(Δ8-119), SRF, SRF(Δ8-133), or SRF(Δ8-177), and cytoplasmic extracts were immunoprecipitated with FLAG antibodies. Proteins coimmunoprecipitating with FLAG-KLF3 were subjected to Western analysis using antisera to SRF. The predicted size of full-length SRF is 65 kDa. All truncated proteins were the expected size and expressed at levels similar to that of the equivalent full-length protein, except for SRF(Δ8-177), which was expressed at ∼3-fold-higher levels than full-length SRF (data not shown). (A) KLF3 interacts with SRF. Lanes 1 to 3, 0.5% input lysates from cells that were mock transfected or transfected with SRF or FLAG-KLF3 plus SRF; lane 4, FLAG-immunoprecipitated proteins from SRF-transfected cells; lane 5, FLAG-immunoprecipitated proteins from cells transfected with FLAG-KLF3 plus SRF; lane 6, GATA-2 IgG-precipitated proteins from cells transfected with FLAG-KLF3 plus SRF (negative control). A diagram of molecular interactions is shown. Prot, protein. (B) Sequences C terminal to the RD of KLF3 are important for interaction with SRF. Lanes 1 to 3, 0.5% input lysates from cells that were transfected with FLAG-KLF3 plus SRF, FLAG-KLF3(Δ8-89) plus SRF, or FLAG-KLF3(Δ8-119) plus SRF; lanes 1* to 3*, FLAG-immunoprecipitated proteins from cells transfected with each of the above combinations. (C) The MADS domain of SRF is required for interaction with KLF3. Lane 1, 0.5% input lysates from cells that were transfected with FLAG-KLF3 plus SRF, FLAG-KLF3 plus SRF(Δ8-133), or FLAG-KLF3 plus SRF(Δ8-177); lane 2, FLAG-immunoprecipitated proteins from cells transfected with each of the above combinations. The predicted sizes of full-length and truncated forms of SRF are shown.

FIG. 8.

Model of KLF3-SRF synergy on CACCC boxes. (A) Established models of KLF3 and SRF activities. SRF contacts CArG sites in its target genes via α-helix I of its MADS domain. α-Helix II of SRF recruits the MRTF coactivators, and the TAD of SRF may recruit other families of coactivators to promote target gene expression. KLF3 contacts CACCC boxes in its target genes via its three C-terminal zinc fingers, and the N-terminal RD (aa 1 to 90) recruits corepressors, such as CtBP2 and FHL3 to repress target gene expression. (B) Proposed model of KLF3-SRF synergy. KLF3 recruits SRF to CACCC boxes, possibly through α-helix I of the SRF MADS domain, which has been shown to mediate both DNA binding and interaction with a number of accessory factors. Association with SRF changes the conformation of the KLF3 N terminus (aa 1 to 90) to allow recruitment of coactivators instead of corepressors. The TAD of SRF may also aid in the recruitment of coactivators.