The Smooth Muscle Cell-restricted KCNMB1 Ion Channel Subunit Is a Direct Transcriptional Target of Serum Response Factor and Myocardin*

Large conductance calcium-activated potassium (MaxiK) channels play a pivotal role in maintaining normal arterial tone by regulating the excitation-contraction coupling process. MaxiK channels comprise α and β subunits encoded by Kcnma and the cell-restricted Kcnmb genes, respectively. Although the functionality of MaxiK channel subunits has been well studied, the molecular regulation of their transcription and modulation in smooth muscle cells (SMCs) is incomplete. Using several model systems, we demonstrate down-regulation of Kcnmb1 mRNA upon SMC phenotypic modulation in vitro and in vivo. As part of a broad effort to define all functional CArG elements in the genome (i.e. the CArGome), we discovered two conserved CArG boxes located in the proximal promoter and first intron of the human KCNMB1 gene. Gel shift and chromatin immunoprecipitation assays confirmed serum response factor (SRF) binding to both CArG elements. A luciferase assay showed myocardin (MYOCD)-mediated transactivation of the KCNMB1 promoter in a CArG element-dependent manner. In vivo analysis of the human KCNMB1 promoter disclosed activity in embryonic heart and aortic SMCs; mutation of both conserved CArG elements completely abolished in vivo promoter activity. Forced expression of MYOCD increased Kcnmb1 expression in a variety of rodent and human non-SMC lines with no effect on expression of the Kcnma1 subunit. Conversely, knockdown of Srf resulted in decreases of endogenous Kcnmb1. Functional studies demonstrated MYOCD-induced, iberiotoxin-sensitive potassium currents in porcine coronary SMCs. These results reveal the first ion channel subunit as a direct target of SRF-MYOCD transactivation, providing further insight into the role of MYOCD as a master regulator of the SMC contractile phenotype.

Smooth muscle cells (SMCs) 2 are of crucial importance in maintaining normal structural and contractile integrity of various organs and tissues throughout the vertebrate body. Unlike skeletal and cardiac muscle cells, which largely undergo irreversible terminal differentiation, SMCs have the inherent ability to alter their differentiated state in response to diverse stimulatory cues. This process of SMC phenotypic modulation is a hallmark of several human pathological conditions, including vascular occlusive disease, asthma, intestinal and bladder obstruction, and Alzheimer disease. SMC phenotypic modulation is often defined in terms of unique molecular signatures of gene expression involved with contraction and cytoskeletal architecture as well as extracellular matrix remodeling (1,2). For example, vascular SMCs display reduced levels of contractile filaments following injury to the vessel wall, adopting the so-called synthetic phenotype characterized by an overabundance of rough endoplasmic reticulum (3). On the other hand, recent studies have shown an exaggerated SMC contractile phenotype thought to contribute to disease progression (4,5).
The majority of SMC-restricted genes contain one or more conserved CArG elements that bind the widely expressed transcription factor, SRF (serum response factor) (6, 7). SRF controls disparate programs of gene expression, including those linked to differentiation and growth, through its association with more than 60 cofactors. One such cofactor is MYOCD (myocardin), which together with SRF and a variety of chromatin remodeling factors constitutes a molecular switch for cardiac and SMC-restricted gene expression (8 -10). Consistent with a role in directing the expression of SMC-specific genes, mice lacking Myocd throughout the embryo die at embryonic day 10.5, presumably because of defective dorsal aortic SMC differentiation (11). Moreover, restricted inactivation of Myocd to neural crest-derived cells reveals diminished expression of several SMC-restricted genes as well as ultrastructural evidence for a synthetic SMC phenotype in blood vessels (12). A complementary ultrastructure study showed that MYOCD gain of function directs de novo SMC myofilament formation in cultured cells with no evidence of cardiac sarcomerogenesis (13).
In addition to SMC contractile filament genes, MYOCD and SRF orchestrate expression of regulatory genes involved with SMC contraction. The telokin promoter, which drives expression of a gene encoding the carboxyl-terminal end of SM myosin light chain kinase, contains a single CArG element that is responsive to both SRF and MYOCD (14,15). The SM myosin light chain kinase gene itself was shown to harbor conserved CArG elements also reactive to SRF-MYOCD (16,17). One of myosin light chain kinase's substrates, the 20-kDa myosin light chain, exhibits increased phosphorylation with MYOCD overexpression coincident with SMC-like contractile activity (13). Further, the calcium transport pump, sarcoplasmic reticulum calcium-ATPase, contains a conserved CArG box (18) and is induced upon forced expression of MYOCD (19). Thus, MYOCD triggers expression of both structural and regulatory genes to enable SMC contractile force generation.
A critical prerequisite for muscle contraction is the proper expression and activity of ion channels, none of which have yet to be defined as direct targets of SRF-MYOCD. SMCs are known to express an array of ion channels, including the voltagedependent L-type calcium channels and the MaxiK channels (20). MaxiK channels are composed of an ␣ gene (Kcnma1) encoding a pore-forming subunit and one of four ␤ genes (e.g. Kcnmb1) encoding cell-restricted modulatory subunits (21)(22)(23). KCNMB1 confers heightened channel sensitivity to calcium and voltage in vascular SMCs, which provides for efficient fine-tuning of vascular tone (24,25). Although the functionality of KCNMB1 has been studied intensively, much less is known regarding its transcriptional regulation. Here, we show that transcription of the KCNMB1 gene is a function of SRF-MYOCD acting over two conserved CArG elements.

EXPERIMENTAL PROCEDURES
Animals-Tissues were rapidly harvested from 2-month-old C57BL/6 male mice or castrated male swine and rinsed in PBS prior to processing for RNA extraction and quantitative analysis of Kcnmb1 expression. All animal protocols were approved by local institutional animal care and use committees.
In Situ Hybridization-Mouse organs were fixed in 10% neutral buffered formalin, paraffin-embedded, and sectioned at 6-m thickness under RNase-free conditions. A mouse Kcnmb1 riboprobe was PCR-amplified (see supplemental Table  1 for primers) and cloned into pBluescript. As a control, we also included a riboprobe to Sm22␣ (Tagln). Linearized antisense and sense riboprobes were purified and labeled with digoxigenin utilizing the MEGAscript Kit (Ambion Inc.). All slides were deparaffinized and rehydrated in RNase-free water. Slides were pretreated for 15 min at room temperature in 10 g/ml proteinase K, washed in 0.2% glycine, postfixed for 10 min in 4% paraformaldehyde, and acetylated for 30 min in 0.1 M triethanolamine, 0.25% acetic anhydride. Slides were then incubated at 65°C for 1.5 h in prehybe solution (diethylpyrocarbonatetreated water containing 50% formamide, 5ϫ saline sodium citrate, 1% SDS, 50 g/ml yeast tRNA, and 50 g/ml heparin) and overnight in the same solution containing 10% dextran sulfate and 100 ng/l digoxigenin-labeled probe. Washes were conducted at 65°C with varying concentrations of saline sodium citrate, followed by a final wash in maleic acid buffer (MABT). Slides were then blocked for 30 min in 1% milk in MABT and incubated for 2 h in 1:1000 sheep anti-digoxigenin antibody/ MABT (Roche Applied Science), sequentially washed in MABT and chromogenic buffer for 10 min each, and then stained from 1 to 5 days with nitro blue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate, toluidine salt (Roche Applied Science) in alkaline phosphate buffer. Sections were counterstained with nuclear fast red and permanently covered for photomicroscopy.
Vascular Injury Models-10 -12-week-old male C57BL/6 mice underwent left carotid artery ligation as described (26). Four days after ligation, animals were perfused with 3.0 ml of 0.9% NaCl to clear blood from major arteries. Injured arteries and control uninjured arteries were rapidly dissected for total RNA extraction. Balloon angioplasty (1.4ϫ overinflation) was performed on the left circumflex and left anterior descending coronary arteries of castrated male swine as described (27). Coronary arteries were harvested at 2 h and 2 days following balloon catheter injury and quickly frozen in liquid nitrogen. Medial cells were isolated by laser capture microdissection (Arcturus, Pixcell IIe), and total RNA was isolated using a Picopure RNA isolation kit (Arcturus) for quantitative RT-PCR.
Cell Culture-10T1/2, COS7, HeLa, HEK293, rat aortic SMCs (RASMCs), and PAC1 SMCs were maintained in 10-cm plates containing Dulbecco's modified Eagle's medium (high glucose) and 10% fetal bovine serum without antibiotics or antimycotics. Growing and differentiated BC 3 H1 cells were studied as described previously (28). Human coronary artery smooth muscle cells (HCASMCs) were maintained in growth medium 231 and supplements as provided by the manufacturer (Invitrogen). Porcine coronary artery smooth muscle cells (PCASMCs) were isolated and grown to postconfluence, serum-starved for 6 days to up-regulate expression of SMC-specific marker genes, and then treated with or without 10 ng/ml platelet-derived growth factor BB (ϩBB) for 24 h as described (29). RASMCs were growth-arrested at 40% confluence for 48 h before treatment with either vehicle or PDGF-BB (30 ng/ml; Upstate). For adenovirus studies, cells were seeded and allowed to adhere overnight and transduced the following day with either cytomegalovirus promoter-driven myocardin (Ad-Myocd; cardiac long form), cytomegalovirus promoter-driven myocardin-related transcription factor A (Ad-MRTF-A; a kind gift from Dr. Paul Herring), or U6 promoter-driven short hairpin RNA to SRF (Ad-shSRF). Equal amounts of controls were adenovirus carrying cytomegalovirus-driven lacZ (for Ad-Myocd or Ad-MRTF-A) or U6-driven short hairpin-enhanced green fluorescent protein (shEGFP; for Ad-shSRF). Cells were washed and refed new medium the next day, and protein or RNA was harvested 48 -72 h later.
RT-PCR-Total RNA was extracted from mouse tissues or different cells by TRIzol or Pico-pure RNA (for Fig. 2D) isolation kit and reverse transcribed to cDNA with a first strand cDNA synthesis kit (GE Healthcare) or the iScript cDNA synthesis kit (Bio-Rad; for Fig. 2C). Semiquantitative PCR or SYBR Green-based real-time PCR (MyIQ; Bio-Rad) was conducted to measure mRNA expression levels across a variety of experimental conditions as indicated under "Results." The primers used to amplify the various target genes are listed in supplemental Table 1.
Bioinformatics-Orthologous KCNMB1 genomic sequences from several vertebrate species were downloaded from the University of California Santa Cruz genome browser (30) and analyzed either with the FASTA algorithm for sequence homologies or the FINDPATTERNS algorithm that rapidly searches large genomic sequences for all 1,216 permutations of the CArG box using the Genetics Computer Group software package (Accelrys, San Diego CA). Comparative genomic analysis was done with the visualization tools for alignments (VISTA) algorithm (available from the VISTA web site) at a threshold of 75% homology over at least 100 base pairs of sequence. Sequence logos were generated with an on-line tool (available from the WebLogo web site). Detailed protocols for facile use of these bioinformatic tools are available upon request.
Plasmid Construction and Mutagenesis-Human short and long KCNMB1 promoters (Fig. 4A) encompassing CArG elements were PCR-amplified from genomic DNA derived from HeLa cells using high fidelity polymerase (Roche Applied Science) with restriction enzyme-clamped primers (supplemental Table 1) and cloned into XhoI-HindIII sites of the pGL3-Basic luciferase reporter plasmid (Promega). All CArG mutants were constructed through PCR-based mutagenesis as per the manufacturer's directions (QuikChange; Stratagene). Big-dye sequencing was done by the Cornell University Life Sciences Core Laboratories Center to confirm nucleotide sequence fidelity.
Western Blot-Indicated cells were washed with phosphatebuffered saline, and protein was extracted in cold lysis buffer containing 1% protease inhibitor mixture (Sigma) as described (13). Protein concentration was determined by a detergentcompatible protein assay (Bio-Rad). Equal amounts of protein (50 g) were resolved by SDS-PAGE, transferred onto nitrocellulose membranes, blocked with 5% nonfat milk for 1 h, and then incubated with antibodies to KCNMB1 (sc33608, Santa Cruz Biotechnology, Inc. (Santa Cruz, CA)), MYOCD (sc-34238, Santa Cruz Biotechnology, Inc.), MYC to detect Myctagged MRTF-A and MRTF-B, FLAG to detect FLAG-tagged Myocd, or control ␣-tubulin (TUBA, T5168, Sigma) overnight at 4°C, followed by incubation with secondary antibody for 1 h. Specific signals were revealed by enhanced chemiluminescence reagent (Pierce). At least one additional experiment was performed for Western blotting studies.
Transfection and Luciferase Assay-PAC1 SMC or COS7 cells were dispersed in 24-well plates, allowed to adhere overnight, and grown to 70% confluence before co-transfecting each of the indicated plasmids (primers for generating each luciferase reporter are provided in supplemental Table 1). The plasmid form of Myocd used in these experiments was the original 807-amino acid variant (8). To correct for varying transfection efficiencies, a Renilla reporter gene was included as an internal control. Transfections were done by calcium phosphate co-precipitation (31) for 15-20 h before adding fresh medium for an additional 24 h, after which cell lysates were prepared for a dual luciferase assay, as described by the manufacturer (Promega). All transfections were performed in triplicate and repeated in at least one independent experiment. Data were analyzed with GraphPad Prism Software (version 4.0, GraphPad Software Inc.) and expressed as the normalized -fold increase over controls Ϯ S.D.
DNA Binding Experiments-Gel shifts were performed as described previously (32). Briefly, in vitro translated SRF was incubated with 50,000 cpm of 32 P-labeled probes (supplemental Table 1) in binding buffer containing 40 mM KCl, 0.4 mM MgCl 2 , 5 mM HEPES, 4 mM EDTA, 2 mM spermidine, and 0.2 mM dithiothreitol plus 4% glycerol, 16 g of bovine serum albumin, and 0.125 g of poly(dI-dC). DNA and protein complexes were fractionated on 5% non-denaturing polyacrylamide gels, vacuum-dried, and exposed to Kodak X-AR film at Ϫ70°C. Competition experiments were performed by the addition of a 100-fold molar excess of non-radioactive double-stranded wild type probe or the relative point mutated probe (supplemental Table 1). Supershift assays were conducted by a 20-min incubation with either rabbit anti-SRF (sc-335, Santa Cruz Biotechnology, Inc.) or control rabbit anti-MEF2A (sc-313, Santa Cruz Biotechnology, Inc.) following incubation of probe and SRF in binding buffer. ChIP assays were performed with EZ-ChIP in HCASMCs, as specified by the manufacturer (Millipore), using the same SRF antibody as in gel shifts. Primers for amplifying fragments of the KCNMB1 locus containing CArG elements (C3 and C5) and exon 4 (devoid of CArG elements) are included in supplemental Table 1.
Transgenic Mouse Studies-We cloned a short and long form of the human KCNMB1 promoter (see Fig. 4A) upstream of a nuclear targeted lacZ reporter gene and tested activity in cultured PAC1 SMCs by staining for ␤-galactosidase. This analysis revealed higher activity from the long form of human KCNMB1 comprising the 5Ј promoter and a fragment of the first intron (data not shown). The latter linearized plasmid was microinjected into fertilized oocytes (SJL/C57BL/6 hybrid) by the University of Rochester Transgenic Core Facility to generate embryonic day 12.5 founder mice. Dissected embryos were processed for lacZ staining and genotyped (see supplemental Table  1 for primers) as described (33). Embryos were either sectioned for microscopy or "cleared" through a graded series of methanol washes (25-100%) followed by overnight washing in 100% methanol and the addition of a 2:1 (v/v) solution of benzyl benzoate and benzyl alcohol.
Current was recorded before and after the addition of the selective MaxiK channel blocker, Iberiotoxin (Ibtx; 100 nM) to the superfusate. Currents were low pass-filtered with a cut-off frequency of 1,000 Hz, digitized at 2.5 kHz, and stored on a computer. Data acquisition and analysis were accomplished by using pClamp 8.0 software (Axon Instruments).

Kcnmb1 mRNA Is Restricted to SMC-enriched Tissues-Pre-
vious expression studies using RNA blotting or a lacZ reporter knocked into the Kcnmb1 locus suggested Kcnmb1-restricted expression to vascular and visceral SMCs (22,23,34). To further confirm and extend these findings, we performed quantitative RT-PCR and in situ hybridization assays in various mouse organs. Relatively high expression of Kcnmb1 was seen in such SMC-rich tissues as bladder and uterus with moderate levels in aorta, lung, and spleen (Fig. 1A). Conversely, Kcnmb1 mRNA showed lower expression in skeletal muscle, liver, and heart. Using a specific antisense riboprobe, we found low level Kcnmb1 transcripts in medial SMCs of the aorta and coronary vessels of the heart (Fig. 1B, a and c); no detectable signal was seen when a sense riboprobe to Kcnmb1 was applied to these tissues (Fig. 1B, b and d). We also observed Kcnmb1 mRNA in microvessels of adult skeletal muscle as well as developing heart and dorsal aorta of embryonic day 13.5 mice (supplemental Fig.  1). The expression pattern of Kcnmb1 is similar to that of Tagln (also known as SM22 ␣), although the latter shows much higher level of expression (supplemental Fig. 1). These results further establish the restricted expression of Kcnmb1 mRNA for adult SMC-containing tissues.
Expression of Kcnmb1 mRNA Is Sensitive to SMC Phenotypic Modulation-Expression data reported here and elsewhere (34) indicate that Kcnmb1 is an authentic adult SMC marker and thus should be considered as part of the molecular signature of SMC-restricted gene expression. To determine whether expression of Kcnmb1 is modulated in a manner similar to SMC cytocontractile genes during phenotypic modulation, we examined its expression in four distinct model systems. We first used the BC 3 H1 cell line, which was previously shown to exhibit a reversible SMC-like phenotype (28). When BC 3 H1 actively grow, Kcnmb1 and other SMC markers (e.g. Myocd) are expressed ( Fig. 2A) (35). Upon serum withdrawal, BC 3 H1 display lower expression of Kcnmb1 and other SMC markers as levels of the skeletal myogenic regulatory factor Myog (myogenin) increase ( Fig. 2A). In the second model, we tested the effect of PDGF-BB on Kcnmb1 expression in cultured RASMCs and PCASMCs because this growth factor is known to promote a less contractile SMC phenotype while increasing proliferation and migration (36). Kcnmb1 mRNA decreased in a time-dependent manner along with Myocd and Myh11, whereas Map2k3 showed increases in expression (Fig.  2B). A similar decline in Kcnmb1 mRNA (but not Kcnma1 mRNA) was evident in cultured PCASMCs treated with PDGF-BB (supplemental Fig. 2). In the third system, we used the mouse carotid artery ligation model to examine whether expression of Kcnmb1 mRNA was sensitive to an in vivo condition of SMC phenotypic modulation. Consistent with PDGF stimulation, which is known to occur in the vessel wall following arterial insult (37), Kcnmb1 transcript levels were downregulated 4 days after ligation injury. In contrast, Kcnma1 mRNA showed no decrease in expression after ligation injury (Fig. 2C). To extend these in vivo findings to a larger mammalian model of vascular injury, we performed balloon angioplasty of the left anterior descending and left circumflex coronary arteries of male swine. Significant decreases in Kcnmb1 mRNA were observed as early as 2 h postangioplasty and persisted for an additional 2 days (Fig. 2D). Taken together, these results demonstrate that Kcnmb1 is part of the SMC genetic program subject to down-regulation following growth factor stimulation or vascular injury.
Comparative Genomic Analysis Reveals Multiple CArG Elements in the Human KCNMB1 Gene-The restricted expression of Kcnmb1 mRNA and its down-regulation in models of SMC phenotypic modulation, suggested that Kcnmb1 transcription may be under the control of the SRF-MYOCD transcriptional switch. To begin exploring this possibility, we carefully examined orthologous species of Kcnmb1 through a bioinformatics approach used previously to further define the mammalian CArGome (38). Interestingly, the ϳ11.5-kb human KCNMB1 gene is found within the 370-kb intron 1 of KCNIP1, which encodes for a neuronal calcium-binding protein that alters type A potassium channels (39) (Fig. 3A). VISTA analysis reveals high homology across the four exons of KCNMB1 between human, monkey, dog, mouse, and rat (Fig.  3B). We found a total of five CArG boxes within 2 kb of the annotated transcription start site of human KCNMB1 (supplemental Fig. 3). Three of the CArG elements (C1, C2, and C4) are not conserved in the orthologous mouse gene. Within the first untranslated exon is a non-consensus CArG element that is 100% conserved across human, monkey, horse, dog, rat, mouse, and armadillo orthologs of KCNMB1; we shall refer to this CArG box as C3 (Fig. 3C and supplemental Fig. 3). In addition, we discovered another non-consensus CArG element within the first intron of KCNMB1 that is conserved in mouse but shows some sequence divergence across species; we shall hereafter refer to this CArG box as C5, (Fig. 3C and supplemental  Fig. 3). The single nucleotide polymorphism tracker at the UCSC Genome Browser (available at the UCSC web site) revealed no annotated single nucleotide polymorphisms within 50 bp of either C3 or C5 of the human KCNMB1 gene. This bioinformatic analysis provided good evidence for a CArG-dependent ion channel subunit gene.
KCNMB1 Contains Functional CArG Elements Responsive to SRF and MYOCD-Having identified five CArG elements in close proximity to the regulatory region of KCNMB1, we tested whether KCNMB1 promoter activity was elevated in cultured SMCs and responsive to SRF/MYOCD. To this end, we PCR-cloned into a luciferase reporter the human KCNMB1 promoter and 5Ј-untranslated exon encompassing the first four CArG elements (designated short) or the same promoter region plus a portion of the first intron containing C5 (designated long; Fig. 4A). The short KCNMB1 reporter showed ϳ4-fold increase in basal luciferase activity in two distinct SMC lines (PAC1 and A7r5), whereas little to no activity was found in several non-SMC lines (Fig. 4B). Both long and short versions of the KCNMB1 promoter were activated by SRF-VP16 and MYOCD, providing evidence for functional CArG elements around the KCNMB1 locus (Fig. 4, C-E). Gel shift assays showed SRF forming a nucleoprotein complex over C3 and C5 (Fig. 5A). Specificity for SRF binding these CArG elements was demonstrated by the absence of a complex upon the addition of cold oligonucleotides, the persistence of a nucleoprotein complex with mutant sequences, and the supershifting of the nucleoprotein complex with an antibody to SRF but not MEF2A (Fig. 5A). To further confirm the association of SRF with C3/C5, ChIP assays were carried out in growing HCASMCs. Consistent with the gel shift data, ChIP assays unveil clear enrichment of DNA containing C3 or C5 following immunoprecipitation with SRF antibody but not when CArG-less exon 4 sequence of the KCNMB1 gene was evaluated (Fig. 5B). To determine whether SRF-bound CArG boxes are important for promoter activity in vitro, we performed mutagenesis of each of the five CArG elements and tested the ability of MYOCD to transactivate the long form of KCNMB1. The results reveal that individually mutating C1, C2, and C4 resulted in modest decreases in KCNMB1 promoter activity. However, mutation of C3 or C5 resulted in greater attenuation in MYOCD-dependent transactivation. Mutation of both C3 and C5 completely abrogated MYOCD transactivation of the KCNMB1 promoter (Fig. 6). MRTF-A and to a lesser extent MRTF-B also stimulated KCNMB1 promoter activity Map2k3 mRNA is shown to be up-regulated in response to PDGF-BB treatment. Data reflect -fold change relative to vehicle conditions for each time point. All genes are normalized to 18 S rRNA expression using the 2 Ϫ⌬⌬Ct method. Note that the y axis is plotted on a logarithmic scale. C, quantitative RT-PCR analysis of Kcnmb1 or Kcnma1 expression in 10 -12-week-old male C57BL/6 mice 4 days following left carotid artery ligation (n ϭ 9). Changes in gene expression were determined by normalizing to 18 S in each artery and then normalizing the ligation artery value to the control artery value (dotted line set to 1). * Kcnmb1 is significantly reduced compared with Kcnma1 (paired t test, p ϭ 0.030). D, quantitative RT-PCR (as in B) of medial tissue obtained via laser capture microdissection either 2 h or 2 days following control or balloon catheter injury (BCI) of porcine coronary arteries. *, p Ͻ 0.05 versus respective control. NOVEMBER 27, 2009 • VOLUME 284 • NUMBER 48 in a CArG-dependent manner (Fig. 7). Taken together, these results validate the in vitro importance of two CArG boxes (C3 and C5) in KCNMB1 that bind SRF and respond to MYOCD transactivation.

SRF-Myocardin Regulation of KCNMB1
CArG Element-dependent KCNMB1 Promoter Activity in Vivo-To find out whether the human KCNMB1 promoter confers transcriptional activity in a living animal, we generated transgenic mouse founder embryos (embryonic day 12.5) carrying the long form of the KCNMB1 promoter with wild type or mutant C3 and C5 sequences linked to a lacZ reporter gene. In five independent wild type founders, we observed lacZ staining throughout the myocardium (Fig. 8, A and D, and supplemental  Fig. 4), consistent with our in situ hybridization data (supplemental Fig. 1) and a previous report documenting expression of the endogenous Kcnmb1 transcript in the developing mouse heart (40). We further observed in three independent founders, KCNMB1 promoter activity in the dorsal aorta (Fig. 8, A and E) (data not shown), where endogenous Kcnmb1 mRNA levels could also be seen (supplemental Fig. 1). Interestingly, KCNMB1 promoter activity was also seen in the developing neural tube, but no detectable staining was ever seen in the developing gut region (Fig. 8, A and C). In contrast to lacZ staining with the wild type KCNMB1 promoter, mutating both conserved CArG elements (C3 and C5) resulted in no detect-able lacZ staining in any of 22 independent transgenic founder embryos (Fig. 8B) (data not shown). These findings complement our in vitro studies by validating the necessity of C3 and C5 in directing restricted KCNMB1 promoter activity in embryonic mice.
Endogenous KCNMB1 Expression Control through SRF-MYOCD-Results thus far establish a critical role for two SRFbinding CArG elements in directing the activity of the KCNMB1 promoter in vitro and in vivo as well as MYOCD-dependent transactivation of the same promoter in vitro. To ascertain whether SRF-MYOCD is important in driving endogenous expression of the KCNMB1 gene and protein, we performed in vitro gain and loss of function studies. We first transduced Ad-Myocd in different cell lines and examined Kcnmb1 mRNA expression by semiquantitative RT-PCR. MYOCD increased Kcnmb1 mRNA expression in 10T1/2 fibroblasts, BC 3 H1 cells, HeLa cells, HEK-293 cells, and HCASMCs (Fig.  9A). Similarly, MRTF-A dose-dependently increased transcript levels of Kcnmb1 and the gold standard marker for SMC lineages, Myh11 (41) (Fig. 9B). Consistent with the RT-PCR data, KCNMB1 protein was elevated upon graded overexpression of MYOCD in PAC1 SMCs and BC 3 H1 cells (Fig. 9C). MYOCD had no effect on Kcnma1 mRNA, suggesting differential regu- The y axis denotes the percentage homology of exons (light blue peaks, untranslated; darker blue peaks, protein coding) and non-protein coding sequences (pink peaks with at least 75% homology, middle line of each VISTA plot) over 100 bp in intergenic and intronic regions between human KCNMB1 and the orthologous genes from rhesus monkey (Mmul), dog (Cfa), mouse (Mmu), and rat (Rno). C, sequence logos and location within locus (denoted by arrows) of the two conserved CArG elements in human, monkey, dog, mouse, and rat KCNMB1 genes. The logos represent the frequency of each nucleotide across the five species' CArG elements as described (69). C3, CArG element number 3 located at ϩ251 (from annotated transcription start site of human KCNMB1; see supplemental lation of MaxiK channel subunits (Fig. 9D). A short hairpin RNA to Srf (42) knocked down endogenous Srf levels in BC 3 H1 (Fig. 9E). As expected, a known SRF target gene, Acta2 (43), was markedly reduced upon Srf knockdown; decreases in Kcnmb1 mRNA were also seen under the same conditions (Fig. 9E). To examine the requirement for SRF in Kcnmb1 expression more rigorously, we performed quantitative RT-PCR. These results also showed SRF-dependent expression of Kcnmb1 mRNA in BC 3 H1 cells (Fig. 9F). Collectively, these results show that endogenous Kcnmb1, but not Kcnma1, is regulated by MYOCD and SRF, most likely through direct interaction with C3 and C5.

Ectopic Myocardin Increases MaxiK Current in Coronary Smooth
Muscle-To investigate whether MYOCD-induced KCNMB1 mRNA/protein expression results in altered MaxiK channel activity, whole-cell I K was measured in the presence or absence of the selective MaxiK channel blocker Ibtx (100 nM). PCASMCs were transduced with Ad-Myocd or empty adenovirus or maintained as non-transduced controls. A whole-cell voltage clamp was used to measure I K during 500-ms step potentials from Ϫ60 to ϩ120 mV before and after Ibtx. Currents from non-transduced control cells were no different from the empty adenovirus control cells; therefore, data were combined into one control group. Fig. 10, A-D, displays representative traces from Ad-Myocd and control cells and induction of I K by Ad-Myocd. Fig. 10E reveals current-voltage (I-V) relationships of Ibtx-sensitive current (i.e. the MaxiK channel component of I K , in both Ad-Myocd and control cells). Both representative traces and I-V relationships document increased Ibtx-sensitive current in Ad-Myocd-transduced cells compared with controls. These results provide evidence for a functional role of MYOCD in eliciting strong potassium currents in cultured SMCs.

DISCUSSION
SRF and MYOCD constitute a molecular switch for the biochemical, structural, and physiological SMC differentiated phenotype. Although many SRF-MYOCD target genes encoding elements of the SMC contractile apparatus exist, there have been no reports of ion channel subunit genes directly regulated by this transcriptional complex. Here, we have extended the mammalian CArGome to the SMCrestricted KCNMB1 subunit of the MaxiK ion channel and find that this gene is down-regulated similarly to other SMC-specific genes in various models of SMC phenotypic modulation. We show that two conserved CArG elements (C3 and C5) located in the first exon and intron of KCNMB1 bind SRF and are vital for MYOCDdependent transactivation. Transgenic mouse studies verify the necessity of these two CArG elements in mediating KCNMB1 promoter activity in embryonic cardiomyocytes and vascular SMCs. We further demonstrate endogenous Kcnmb1 mRNA induction upon forced expression of MYOCD in various nonmuscle cell lines. Conversely, reducing levels of Srf elicits decreases in expression of Kcnmb1. Finally, MYOCD gain-offunction studies disclose increases in iberiotoxin-sensitive potassium currents in cultured SMCs.
The results reported here showing reductions in Kcnmb1 mRNA expression following growth factor stimulation and arterial injury are consistent with this ion channel subunit being altered in other instances of SMC phenotypic modulation. For example, in an angiotensin II-infused rat model of hypertension, MaxiK currents were reduced concomitantly with Kcnmb1 mRNA expression (44). The latter results are in line with the hypertensive phenotype observed in Kcnmb1 knock-out mice (34). SMCs derived from coronary vessels of  (C5). B, KCNMB1 promoter (short) was transfected into indicated cell lines, and base-line luciferase activity was measured as the normalized -fold increase to the control pGL3 basic reporter (set to 1). C, COS7 cells were transfected with long or short KCNMB1 reporter in the presence of either empty vector (EMSV) or SRF-VP16. Reporter activity is defined here as normalized luciferase to the internal Renilla control reporter. MYOCD-dependent activation of both KCNMB1 reporters is shown in COS7 (D) or PAC1 SMCs (E). Luciferase activity was normalized to the internal Renilla control.
aged Fisher 344 rats display attenuated MaxiK currents and Kcnmb1 mRNA expression (45). Levels of KCNMB1 protein are also depressed in retinal microvascular SMCs of streptozo-tocin-induced diabetic rats (46). In contrast to lower levels of Kcnmb1 reported here and elsewhere (29,44,46), we did not observe obvious changes in expression of the Kcnma1 subunit during SMC phenotypic modulation, suggesting that each MaxiK channel subunit is under distinct transcriptional control mechanisms (see below).
Altered Kcnmb1 expression appears to coincide with changes in levels of another SMC-restricted ion channel subunit, the L type voltage-gated calcium channel (Cacna1c), whose expression is reduced in cultured SMC and within the arterial wall following balloon angioplasty (47,48). Prior work has delineated distinct regions of the Cacna1c gene that direct basal SMC promoter activity in vitro, although no muscle-restricted transcription factor binding sites have yet been identified and experimentally validated (49,50). Interestingly, calcium current through the CACNA1C channel induces mRNA expression of Myocd and several contractile genes in cultured SMCs (51). The latter findings support the emerging concept of excitation-transcription coupling within vascular SMC (52,53). It will be important to define the transcriptional code underlying current-induced alterations in Myocd expression (54) and explore the relationship between calcium current and expression of the Kcnmb1 subunit.
Surprisingly little information exists with respect to musclerestricted ion channel genes and their transcriptional control through myogenic regulatory factors. The skeletal muscle type 1 sodium channel harbors a myogenin-binding E-box that directs expression of this ion channel by relieving an upstream repressor (55). Recently, the cardiac muscle-specific channel, hyperpolarization-activated cyclic nucleotide-gated potassium channel 4, was shown to be a direct target of the MEF2 (myocyte-specific, enhancer-binding factor 2) transcription factor (56). In Drosophila, the Kcnma1 gene (slowpoke) is controlled by multiple promoters, including one directing expression of this channel in muscle fibers and tracheal cells through sequences containing MEF2 and E-box binding sites (57). The orthologous mouse Kcnma1 gene is also under the control of several promoters that contain E-boxes, MEF2 sites, and putative CArG boxes (58), some of which are conserved in the human ortholog (59). However, as yet, there are no data supporting a critical role for any of these binding sites or their DNA-binding transcription factors in the regulation of Kcnma1 expression. Indeed, MYOCD overexpression studies performed in this report showed no effect on mRNA expression of KCNMA1 in human coronary artery SMCs, consistent HCASMCs. Equal aliquots of chromatin lysates were immunoprecipitated with antibodies to either SRF or a control rabbit IgG. DNA associated with the immunoprecipitates was isolated and used as templates for PCR amplification of either KCNMB1 promoter (C3) or intron 1 region (C5). As a control, primers to exon 4 of KCNMB1 (without CArG boxes) were used following SRF immunoprecipitation. Results were repeated in two independent studies. with the absence of conserved CArG boxes in the immediate vicinity of this ion channel subunit's promoter. Further, we did not observe appreciable changes in Kcnma1 mRNA expression in several models of SMC phenotypic modulation. Although these negative results are consistent with independent models of SMC phenotypic change (44,46), other studies have demonstrated attenuated expression of Kcnma1 in vitro and in vivo (60 -62). We suspect that different time points of analysis and distinct model systems underlie these disparate findings.
In contrast to the multiple promoters proposed to regulate expression of KCNMA1 in both invertebrates and vertebrates, the human KCNMB1 gene contains a single promoter with one major transcription start site (22). Despite the reporting of several MEF2 and E-box elements in the KCNMB1 promoter (22), there has been virtually no information as to how this SMCrestricted gene is regulated transcriptionally either in vitro or in vivo. In fact, the only published report of authentic KCNMB1 promoter activity is a recent one suggesting a role for Sp1 in the regulation of rabbit Kcnmb1 (63). Comparative sequence analysis, however, proves that the Sp1 binding site is not conserved in human and mouse orthologous promoters. Thus, it is unclear at this time whether Sp1-mediated expression of Kcnmb1 is unique to the rabbit gene or more broadly applicable to other mammalian orthologs of this ion channel subunit.
Several lines of evidence provided here support an essential role for SRF in the SMC-restricted regulation of KCNMB1 both in vitro and in vivo. First, quantitative RT-PCR and in situ FIGURE 7. KCNMB1 promoter is responsive to MRTF transactivation. COS7 cells were co-transfected as in Fig. 6, using the indicated expression and reporter plasmids. Luciferase activity was normalized as in Fig. 6. The inset represents Western blots using antibodies to FLAG (for ectopic MYOCD detection) or MYC (for ectopic MRTF-A and MRTF-B detection) in cell extracts from parallel plates transfected with equal amounts of each expression plasmid. Increasing the input of MRTF-B did not result in any change in the transactivation of the KCNMB1 promoter (data not shown). WT, wild type. hybridization assays show the highest expression of Kcnmb1 mRNA in mouse vascular and visceral SMC tissues, where SRF protein is abundantly expressed. Second, bioinformatic analyses demonstrate the existence of two conserved CArG boxes, C3 in the untranslated first exon and C5 in the first intron, which bind SRF and are requisite for promoter activity in cultured SMCs as well as dorsal aortic SMCs of the mouse embryo. Additional non-conserved CArG boxes have modest effects on KCNMB1 promoter activity in vitro but appear to be insufficient to drive promoter activity in transgenic mice. Finally, Srf knockdown in an SMC-like cell line elicits corresponding decreases in expression of Kcnmb1 mRNA. The last result agrees with a report showing reduced expression of Kcnmb1 in the embryonic heart of Srf knock-out mice (40). Although previous reports have documented functional CArG boxes in the promoters to calcium (18) and chloride (64) transporters, the present results are the first to demonstrate functional CArG elements in an SMC-restricted ion channel subunit. Whether the CArG elements controlling KCNMB1 gene expression functionally regulate KCNIP1, whose first intron harbors KCNMB1, is presently unknown.
The virtual absence of Kcnmb1 mRNA expression in the adult heart ( Fig. 1) (22) suggests that its expression and promoter activity in embryonic heart is ephemeral. This is consistent with a number of other SMC-specific genes that display transient expression and promoter activity in the developing myocardium, including the SMC isoforms of calponin and ␣-actin, the smoothelin A isoform, SM22 ␣, dystrophin, and the cysteine-rich protein 1 gene, all of which contain conserved, SRF-binding CArG elements in the vicinity of their promoter regions (7). Given the expression of such a repertoire of SMCspecific genes in early cardiomyogenesis, it is intriguing to consider the possibility that early cardiomyocytes exhibit contractile activity more closely resembling adult SMCs than adult cardiac muscle. This hypothesis will require a comprehensive analysis of the functional characteristics of embryonic car-  48 h, and endogenous Kcnmb1 and Myh11 mRNA expression was assayed by RT-PCR. C, Western blots of endogenous MYOCD and KCNMB1 protein in the indicated cell lines following exposure to increasing amounts of Ad-Myocd. D, the indicated cell lines were transduced with Ad-Myocd as in A, and levels of Kcnma1 and Kcnmb1 mRNA were measured by RT-PCR; brain and bladder mRNA samples were included as a positive control for Kcnma1 expression. BC 3 H1 cells were transduced with Ad-shSRF (ϩ) or Ad-shEGFP (Ϫ) for 72 h and subjected to semiquantitative RT-PCR (E) or quantitative RT-PCR (F) for the indicated target genes. The knockdown of Kcnmb1 in F is presented as the percentage decrease from shEGFP control set to 1. All expression data are representative of two or more independent experiments. diomyocytes. Alternatively, expression of SMC-specific transcripts in the developing heart may merely reflect the abundant expression of SRF during early cardiomyogenesis (65) and its ability to override negative signaling and/or repressors that otherwise are active in the normal adult heart to block SMCrestricted gene expression.
In order for SMCs to contract appropriately, there must be tight control of both the expression and activity of ion channels that integrate electrical current with the cytocontractile machinery. We recently reported that forced expression of MYOCD was sufficient to orchestrate the assembly of SMC myofilaments leading to agonist-induced contraction in a cell type that normally does not display such phenotypes (13). The results of the present study provide the first evidence for MYOCD directly regulating expression of an ion channel subunit critical for normal SMC contractile activity. Moreover, patch clamp studies indicate that MYOCD can induce iberiotoxin-sensitive potassium currents in cultured coronary artery SMC that otherwise are essentially devoid of such current activity. The latter results are consistent with those of Jiang et al. (22), who showed that ectopic expression of KCNMB1 augmented the effects of the pore-forming ␣ subunit (KCNMA1) on calcium/voltage sensitivities. Given the direct activation of Kcnmb1 by MYOCD and the former's compromised expression in various disease processes (this report) (44 -46), it will be instructive to establish whether MYOCD expression is similarly attenuated in such vascular diseases as hypertension and diabetic retinopathy. Based on published reports showing reductions in Myocd mRNA expression following arterial injury (66) or diet-induced atherosclerosis (29), it is possible that restoring MYOCD expression will normalize Kcnmb1 expression and MaxiK currents, thus minimizing disease progression. One potentially exciting way of normalizing MYOCD levels in the perturbed arterial wall will be to overexpress microRNA-145 which was recently shown to be a positive regulator of Myocd expression in vascular SMCs and stem cells with the potential to differentiate into SMCs (67,68).
In summary, results of this report offer the first example of an SRF-MYOCD target gene encoding an ion channel subunit. The molecular insight into the transcriptional basis for KCNMB1 expression documented here provides a foundation to begin developing novel therapies for the potential treatment of hypertension, vascular occlusive disease, and diabetic retinopathy, all of which are characterized by aberrant expression and activity of this MaxiK channel subunit.