Myomaxin Is a Novel Transcriptional Target of MEF2A That Encodes a Xin-related α-Actinin-interacting Protein*

The physiological targets regulated by MEF2 in striated muscle are not completely known. Several recent studies have identified novel downstream target genes and shed light on the global transcriptional network regulated by MEF2 in muscle. In our continuing effort to identify novel, downstream pathways controlled by MEF2, we have used mef2a knock-out mice to find those genes dependent on MEF2A transcriptional activity. Here, we describe the characterization of a direct, downstream target gene for the MEF2A transcription factor encoding a large, muscle-specific protein that localizes to the Z-disc/costameric region in striated muscle. This gene, called myomaxin, was identified as a gene markedly down-regulated in MEF2A knock-out hearts. Myomaxin is the mouse ortholog of a partial human cDNA of unknown function named cardiomyopathy associated gene 3 (CMYA3). Myomaxin is expressed as a single, large transcript of ∼11 kilobases in adult heart and skeletal muscle with an open reading frame of 3,283 amino acids. The protein encoded by the myomaxin gene is related to the actin-binding protein Xin and interacts with the sarcomeric Z-disc protein, α-actinin-2. Our findings demonstrate that Myomaxin functions directly downstream of MEF2A at the peripheral Z-disc complex in striated muscle potentially playing a role in regulating cytoarchitectural integrity.

The mechanisms by which MEF2 (myocyte enhancer factor-2) transcriptional activity is modulated, via specific signal transduction pathways and protein-protein interactions, to regulate gene expression has been extensively characterized (1)(2)(3). These findings have led to a better understanding of the involvement of MEF2 in a variety of developmental, physiological, and pathological pathways in muscle (4 -6). Despite this extensive information regarding MEF2 function a gap remains in our understanding of the downstream cellular processes regulated by MEF2 in muscle.
The number of MEF2-regulated genes has been increasing, since recent reports have described novel, downstream target genes responsive to MEF2 activity, such as CHAMP (7), HRC (8), mef2c (9,10), BOP (11), and Srpk3 (12). An elegant comprehensive genomic profiling analysis in combination with chromatin immunoprecipitation (ChIP 2 -on-Chip) has revealed a broad array of direct target genes for MEF2 during muscle differentiation (13). Likewise, DNA microarray approaches have identified MEF2-regulated genes that function in signal transduction pathways (14) and ion channel regulation (15) in skeletal muscle and heart, respectively, further highlighting the diversity of genes controlled by this transcription factor.
We recently reported that the myospryn gene is directly regulated by MEF2A (16). myospryn encodes an ␣-actininand dysbindin-interacting protein localized to the peripheral Z-disc region of striated muscle known as the costamere (16,17). Myospryn is the mouse ortholog of CMYA5, a cardiomyopathyassociated gene of unknown function, whose transcripts are down-regulated in mef2a null mice (16,18). Given the intriguing finding of a "cardiomyopathy-associated" gene dysregulated in mef2a null hearts, which have severe cardiac abnormalities (18), we set out to determine whether additional cardiomyopathy-associated (CMYA) genes existed and, if so, whether these are also sensitive to MEF2A dosage.
Using an in silico approach, we identified four such additional cardiomyopathy-associated genes named CMYA1, CMYA2, CMYA3, and CMYA4 in the NCBI data base. These additional CMYA genes are partial human sequences identified by genome scale expression profiling of cardiac muscle (Incyte Genomics) and deposited in the public data base. Interestingly, the full-length mouse orthologs have been reported for human CMYA1, -2, and -4, but those studies were published either prior to the identification of the CMYA genes or failed to point out the similarity to the CMYA genes. The mouse ortholog for CMYA1, -2, and -4, is Xin (19), Myomegalin (20), and Unc-45 (21), respectively. Xin is an actin-binding protein required for cardiac morphogenesis in the developing chick heart (19,22). Myomegalin is a Golgi/centrosomal and sarcomere-localized protein that binds phosphodiesterase 4D3 (20). Unc-45 functions as a molecular chaperone for myosin and is modified by E3/E4-multiubiquitylation (21,23). All of the mouse orthologs for the CMYA genes are expressed predominantly or exclusively in striated muscle.
A partial human cDNA for CMYA3 has been described and tentatively named Xirp2 (22). This particular sequence encodes a small portion of Xirp2 that binds F-actin through a novel 16-amino acid repeat sequence named the "Xin repeat" because of its similarity to a repetitive motif present in the muscle protein Xin (19,22). A recent paper has also reported the up-regulation of CMYA3 transcripts in hearts from mice exposed to angiotensin II, a known inducer of cardiac hypertrophy (24). In this paper, we describe the cloning and characterization of the full-length mouse ortholog of CMYA3/Xirp2, which we have named Myomaxin (myogenic MEF2-activated Xin-related protein). We show that myomaxin is dramatically down-regulated in mef2a knock-out hearts and that it is a direct MEF2 target gene. Furthermore, Myomaxin is localized at the periphery of the Z-disc in striated muscle, where it interacts with sarcomeric ␣-actinin in two separate subdomains within the actin-binding 16-amino acid repeat region. These findings indicate that Myomaxin may play a role in the regulation of muscle cytoarchitecture downstream of MEF2A.

MATERIALS AND METHODS
In Silico Analysis and Bioinformatics-Given the similarity of myospryn to CMYA5 (16), we searched the NCBI data base for additional genes containing the name "cardiomyopathy-associated." The mouse CMYA3 sequence was identified as one of four such genes whose sequences were deposited by Incyte Genomics. Continued searching of the data base failed to identify additional CMYA genes, indicating that only five novel gene sequences were deposited by Incyte Genomics.
Transient Transfections, Coimmunoprecipitations, and Western Blots-COS-7 cells were maintained in DMEM containing 10% FBS, 2 mM glutamine, and penicillin/streptomycin. For reporter assays, COS cells in 6-well plates, each well containing ϳ500,000 cells, were transfected with 0.2 g of the wild type or mutant forms of the myomaxin promoter in pGL3-Basic (Promega) and 0.2 g of human MEF2A in pcDNA1 (Invitrogen). Cells were transfected using FuGENE 6 transfection reagent (Roche Applied Science) and were harvested after 36 -48 h. Luciferase assays for the myomaxin promoter in C2C12 cells were performed in the same manner as for COS cells with the exception that 6-well plates were seeded at a density of 300,000 cells/well. Luciferase assays were performed using the Dual Luciferase reporter assay system (Promega), and protein concentrations were normalized via Bradford assay. For co-immunoprecipitations, COS cells in 100-mm dishes (at 25% confluence) were transfected with 20 g of expression plasmids for full-length Myomaxin and 5 g for truncated forms of Myomaxin and ␣-actinin, using FuGENE 6 reagent (Roche Applied Science). Myomaxin proteins were fused with an amino-terminal Myc epitope, and ␣-actinin was fused with an amino-terminal FLAG epitope. Forty-eight hours after transfection, cells were harvested in ELB buffer (50 mM HEPES, pH 7.0, 250 mM NaCl, 5 mM EDTA, 0.1% IGEPAL, 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, and protease inhibitor mixture (Complete, Roche Applied Science)). Extracts were immunoprecipitated for 2 h at 4°C using protein A/G-agarose and 1 l (or 1 g) of monoclonal anti-Myc antibody (Santa Cruz Biotechnology, Inc., Santa Cruz, CA). Subsequently, the pellet was washed with ELB buffer and subjected to SDS-PAGE, followed by transfer to Immobilon polyvinylidene difluoride membrane and immunoblotting using anti-FLAG antibodies (Sigma).
Gel Shift and Chromatin Immunoprecipitation Assays-For gel shift assays, in vitro translated MEF2A was produced in rabbit reticulocyte lysates (Promega) and incubated in the presence of radiolabeled, double-stranded oligonucleotides corresponding to each of the three MEF2 sites in the myomaxin promoter as described previously (16). For ChIP analysis, C2C12 myoblasts were seeded at a density of 1.5 ϫ 10 6 in 10-cm dishes induced to differentiate. On day 7, cells were crosslinked with 37% formaldehyde, incubated for 10 min at room temperature, and subsequently harvested in cold PBS containing protease inhibitors (Roche Applied Science). ChIP assays were carried out according to the manufacturers' suggestions (Upstate Biotechnology) with the following modifications. Cells harvested from one 10-cm dish were sonicated in 800 l of SDS lysis buffer containing protease inhibitors. Each 800-l sample was subsequently split into 200-l samples and diluted 10-fold in ChIP dilution buffer containing protease inhibitors. Immunoprecipitations were performed using 4 g of anti-MEF2A antibody (C-21; Santa Cruz Biotechnology). The rat, mouse, and human myomaxin promoter sequences were aligned using the MatInspector program (available on the World Wide Web at www.genomatix.de). Primers used for amplification of the Ϫ75 MEF2 site were as follows: 5Ј-GTGG-CTATGGAAGGGAGTTTGAGTG-3Ј and 5Ј-CCAGCTC-CACCGAAGAGTCTATGAG-3Ј.

RESULTS
In Silico Analysis and Identification of Myomaxin-Given the surprising finding that the MEF2-regulated myospryn gene is the mouse ortholog of human CMYA5 (cardiomyopathy-associated gene 5) we used an in silico approach to identify additional "cardiomyopathy-associated" genes and to test whether these genes are also dysregulated in mef2a knock-out hearts. By searching the NCBI nucleotide data base using the key words "cardiomyopathy-associated," we found four such additional sequences named CMYA1 (cardiomyopathy-associated gene 1), CMYA2, CMYA3, and CMYA4 (Fig. 1A). All four of these human expressed sequence tags, including CMYA5, were iden-tified by expression profiling of human cardiac muscle and deposited in the NCBI data base (Incyte Genomics, Palo Alto, CA). To determine whether the CMYA1, -2, -3, and -4 genes are similarly dysregulated in mef2a knock-out hearts, we performed RT-PCR with cDNA from postnatal day 5 hearts of mef2a mutant mice using mouse-specific primers for each of these genes. As shown in Fig. 1B, each CMYA gene was downregulated to some extent (right), but the expression of CMYA3 exhibited the most dramatic down-regulation in mef2aϪ/Ϫ hearts to nearly undetectable levels (left). Similar amplification levels for each of the CMYA genes tested were observed between wild type and heterozygote whole heart cDNA (data not shown). These results indicate that CMYA3 is highly dependent on MEF2A transcriptional activity and have thus named the mouse ortholog, myomaxin (myogenic MEF2-activated Xin-related protein).
Regulation of Myomaxin Transcription by MEF2A-To test whether myomaxin represents a direct transcriptional target for MEF2A, we identified the 5Ј-end of the gene using a combination FIGURE 1. Bioinformatics identification of CMYA3 and analysis of CMYA transcripts in wild type and mef2a mutant hearts. A, the NCBI public data base was searched for CMYA genes using the key words "cardiomyopathyassociated." A total of four such additional genes were identified: CMYA1, -2, -3, and -4. B, expression levels of the CMYA genes in mef2a knock-out hearts. Total RNA was isolated from postnatal day 5 (P5) hearts of wild-type, heterozygote, and mef2a mutant mice, and transcripts were amplified by semiquantitative RT-PCR. RT-PCR reactions were performed for 25 cycles and were fractionated on a 1% agarose gel. Similar amplification levels for each of the CMYA genes tested were observed between wild type and heterozygote cDNA. The CMYA genes are designated by their published names: Xin (CMYA1), Myomegalin (CMYA2), Myomaxin (CMYA3), and Unc-45 (CMYA4). ␣-Myosin heavy chain (␣-MHC) and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) transcripts, which are not dysregulated in mef2a null hearts, were used as controls. No transcripts were detected in the absence of reverse transcriptase. WT, wild type; KO, mef2a mutant.
of the expressed sequence tags data base, 5Ј-rapid amplification of cDNA ends, primer extension, and RNase protection (data not shown). Based on this information, we isolated 5.0 kb of mouse genomic sequence upstream of the first exon and cloned this fragment into the pGL3BASIC-luciferase reporter. MEF2A responsiveness of this 5Ј region was tested in transfected COS cells. This 5.0-kb fragment alone exhibited background levels of activity in COS cells. However, when co-transfected with a MEF2A expression vector, the 5.0-kb fragment was induced nearly 5.0-fold ( Fig.  2A), indicating that myomaxin is activated by MEF2A.
Given the ability of MEF2A to transactivate this upstream fragment, we examined this region for MEF2 binding sites and identified three putative MEF2 sequences exhibiting similarity to the consensus MEF2 site, (C/T)TA(A/T) 4 TA(G/A). To determine whether MEF2A could bind these sequences, we performed gel mobility shift assays. The MEF2 sequences located at Ϫ75 (CTAAAAATAG) and Ϫ2937 (TTATTTT-TAA) relative to the putative transcription start site bound to in vitro translated MEF2A (Fig. 2C), but binding to the Ϫ3096 site (TTATTATTAA) was detectable only upon longer exposure FIGURE 2. Direct activation of myomaxin promoter by MEF2A and muscle specificity of the myomaxin proximal promoter. A, MEF2A activates the myomaxin promoter. A MEF2A expression vector was co-transfected into COS cells with the wild-type 5.0-kb myomaxin promoter or a series of mutant promoter constructs. The -fold activation is relative to the luciferase activity (normalized to 1) of the wild type or mutant reporter transfected with pcDNA vector backbone. Values obtained for the mutant MEF2 sites are significant at p Ͻ 0.05. B, sequence of the mouse myomaxin proximal promoter. The sequence of 1.5 kb upstream of the putative transcription start site is depicted. The Ϫ75 MEF2 site and a putative E-box sequence are underlined. C, MEF2A binds two MEF2 sites present in myomaxin promoter. In vitro translated MEF2A was subjected to gel shift assays. Protein-DNA complexes were formed with each of the MEF2 sites within the myomaxin promoter with the exception that binding the most distal MEF2 site (Ϫ3096) was detected only upon longer exposure. MEF2A was unable to bind mutant MEF2 sites. D, the proximal myomaxin promoter is sufficient for MEF2A-dependent activation. A 1.5-kb myomaxin-luciferase reporter construct was co-transfected into COS cells with a MEF2A expression vector. MEF2A activated the wild-type (WT) but not the promoter with mutant Ϫ75 MEF2 site. E, the 1.5-kb upstream sequence is sufficient to confer muscle-specific gene transcription. The 1.5-kb myomaxin-luciferase vector was transfected into COS and C2C12 myoblasts, and luciferase units were measured. This fragment was sufficient to confer muscle-specific activity. The Ϫ75 MEF2 mutant exhibited basal activity in C2C12 myoblasts. DECEMBER 22, 2006 • VOLUME 281 • NUMBER 51 (data not shown). MEF2A failed to bind to the Ϫ75 and Ϫ2937 MEF2 sites harboring mutations within the core A-T-rich region of the element known to disrupt interactions with MEF2 (Fig. 2C). These results confirm that MEF2A protein directly binds to two MEF2 sites within the myomaxin promoter.

Myomaxin Is a Xin-related MEF2A Target Gene
To determine whether the myomaxin promoter was being activated through one or more of the MEF2 sites, mutations that disrupt MEF2 binding were introduced within each of the core sequences in the context of the 5.0-kb proximal promoter. As shown in Fig. 2A, mutations within the two upstream MEF2 sites, Ϫ2937M and Ϫ3096M, had no significant effect on MEF2dependent transactivation of the myomaxin promoter. In contrast, a mutation within the Ϫ75 MEF2 site (Ϫ75M), was activated less than 2-fold by MEF2A ( Fig. 2A). To further demonstrate that the two upstream MEF2 sites are not necessary for transactivation, we generated a deletion construct lacking the two most upstream sites and determined its ability to be transactivated by MEF2A. This 5Ј deletion removed all but the proximal 1.5 kb relative to the transcription start site (Fig. 2B). This construct was induced to a similar level as the 5.0-kb upstream fragment (Fig. 2D), demonstrating that the upstream MEF2 sites are not necessary and that the Ϫ75 MEF2 site is sufficient to activate the myomaxin promoter. A mutation of the Ϫ75 site in the context of the smaller 1.5-kb fragment also significantly reduced MEF2A-dependent transcription to less than 2-fold (Fig. 2D).
To determine whether the 1.5-kb fragment was sufficient to confer muscle specificity, we tested this reporter construct in the C2C12 muscle cell line and showed that this minimal region was highly active in C2 myoblasts as compared with COS cells (Fig. 2E). Moreover, the Ϫ75 site is essential for this muscle-specific responsiveness, since a mutation within this proximal MEF2 site nearly completely abolishes muscle-specific reporter activity (Fig. 2E). These results demonstrate that the Ϫ75 MEF2 binding site is necessary and sufficient for myomaxin transcriptional regulation.
Given the importance of the Ϫ75 site for MEF2-dependent transactivation of the myomaxin promoter, we searched for possible evolutionary conservation of the proximal MEF2 site. Indeed, the Ϫ75 site and neighboring sequences are conserved between rodents and humans (Fig. 3A). To determine whether the Ϫ75 MEF2 site is directly bound by MEF2A in vivo, we performed ChIP analysis on the myomaxin promoter. Nucleosomes were isolated from differentiated C2C12 myotubes and subjected to immunoprecipitation using anti-MEF2A antibodies (C-21; Santa Cruz Biotechnology). We designed primers that would amplify mouse genomic sequences surrounding the evolutionarily conserved Ϫ75 MEF2 site (Fig. 3A). As shown in Fig. 3B, we were able to demonstrate that the Ϫ75 MEF2 site in the mouse myomaxin promoter is bound by MEF2A in differentiated myotubes (left). Importantly, ChIP assays performed with IgG and two additional unrelated antibodies, anti-GAL4 (Santa Cruz Biotechnology) and anti-green fluorescent protein (Santa Cruz Biotechnology), failed to immunoprecipitate the Ϫ75 MEF2 site (Fig. 3B, right panel). Collectively, these results demonstrate that myomaxin is a direct transcriptional target for MEF2A.
Cloning Full-length Myomaxin-We isolated the full-length mouse Myomaxin cDNA from a Marathon skeletal muscle cDNA library (Clontech) using high fidelity PCR (Roche Applied Science) with gene-specific primers designed against predicted exons in the public data base. These PCR products were sequenced and assembled to generate the full-length cDNA (GenBank TM accession number EF122140). Conceptual translation of the Myomaxin cDNA revealed an open reading frame of 3,283 amino acids with a predicted size of 350 kDa (Fig.  4A). BLAST analysis using the open reading frame of Myomaxin confirms that it is the mouse ortholog of the partial human CMYA3/Xirp2 sequence and also exhibits modest sim- ilarity to mouse Xin within the actin-binding 16-amino acid repeat region (amino acids 309 -1,599) (Fig. 4A).
Using a combination of public (NCBI and Ensembl) and private (Celera) data bases, we identified the entire mouse genomic locus for myomaxin. Mouse myomaxin localizes to chromosome 2 and spans a region covering nearly 100 kilobases (kb) (Fig.  4B). The myomaxin gene has six small exons preceding an unusually large seventh exon, over 9,000 base pairs in length. Surprisingly, the other novel MEF2 target gene we identified recently, myospryn, also encodes a large muscle-specific protein with a large coding exon of over 9,000 nucleotides. Although the myomaxin and myospryn genes do not exhibit any sequence similarity to each other, the resemblance in gene structure is intriguing.
Muscle-specific Expression of Myomaxin-Northern analysis of mouse tissues, using a probe derived from the middle portion of the mouse Myomaxin cDNA, revealed a transcript migrating above the 9.5 kb RNA marker and highly restricted to adult heart and skeletal muscle (Fig. 5A, top). Expression was not detected in any other adult mouse tissue. Moreover, myomaxin transcripts were not detected in either mouse aorta or stomach, representative tissues enriched for smooth muscle (Fig. 5A, bottom), indicating a striated muscle-specific expression of the myomaxin gene.
myomaxin expression was also examined in the C2C12 myoblast cell line using RT-PCR. myomaxin transcripts were detected in both proliferating C2C12 myoblasts and in differentiating myotubes (Fig. 5B). We observed a modest increase in myomaxin expression at the onset of myogenic differentiation, but surprisingly, the most mature myotubes (day 7) exhibited a robust up-regulation in myomaxin transcription (Fig. 5B). Examination of myomaxin transcripts in neonatal rat cardiomyocytes also showed that myomaxin is expressed in these cells and is up-regulated by hypertrophic stimuli, such as the ␤-adrenergic agonist, phenylephrine, and serum (Fig. 5C). Altogether, these results indicate that myomaxin is a striated muscle-specific gene induced by signals that promote muscle differentiation and growth. Subcellular Localization of Myomaxin and Co-localization with Sarcomeric ␣-Actinin-To determine the subcellular localization of Myomaxin in striated muscle, polyclonal antibodies were raised against a GST-Myomaxin fusion protein (see "Materials and Methods") and used for immunohistochemistry on cryosections from adult mouse hearts. As shown in Fig. 6A, Myomaxin antibodies revealed a robust, striated pattern (right), whereas preimmune IgG exhibited a diffuse, low level background signal (left). As an initial demonstration of specificity, Myomaxin antibodies were tested by immunoblot analysis. Anti-myomaxin antibodies detected specific signals in heart and skeletal muscle but not liver extracts (Fig. 6B, left). An essentially identical pattern was observed with two additional Myomaxin antibodies raised against different regions of the protein. This pattern is also observed in neonatal cardiomyo-  A, an adult mouse multiple-tissue Northern blot (Clontech) was probed with an NcoI-EcoRI fragment (corresponding to nucleotides 8,218 -9,051) of Myomaxin cDNA sequence. A single myomaxin transcript, greater than the 9.4-kb RNA standard, was detected exclusively in heart and skeletal muscle. Lower panel, RT-PCR analysis of myomaxin in adult mouse heart, stomach, and aorta. Expression is observed only in the heart and not in representative smooth muscle tissues. B, expression of myomaxin in C2C12 muscle cell line. Myomaxin transcripts are detected in proliferating C2C12 myoblasts. Expression of the myomaxin gene shows a modest increase during early stages of myogenic differentiation (days 1, 3, and 5) with robust induction at day 7. C, expression of myomaxin in neonatal rat ventricular cardiomyocytes. Myomaxin is expressed in isolated primary rat cardiac muscle cells and induced under conditions that stimulate hypertrophy: atrial natriuretic factor (ANF), brain natriuretic peptide (BNP), phenylephrine (PE), and serum (10% FBS). GAPDH, glyceraldehyde-3-phosphate dehydrogenase, internal control. cyte and C2C12 myoblast extracts. 3 The upper band represents full-length Myomaxin, since it comigrates with in vitro translated full-length Myomaxin cDNA (Fig. 6B, right). The lower molecular mass bands may represent degradation products or alternative isoforms of Myomaxin. To further demonstrate specificity of this signal, we performed blocking experiments on immunoblots and muscle cryosections. By Western analysis, preincubation of Myomaxin antibodies with recombinant protein effectively competed the immunoreactive bands (not shown). In muscle cryosections, incubation of the anti-Myomaxin antibodies with recombinant Myospryn (a protein localized to the Z-disc/costamere) (16) at 10-and 50-fold molar excess or no blocking protein did not eliminate the striated signal (Fig. 6C, top). In striking contrast, incubation of the anti-Myomaxin antibodies with a specific, GST-Myomaxin recombinant protein completely blocked the striated pattern at 10-and 50-fold molar excess (Fig. 6C, bottom). The striated pattern observed with the Myomaxin antibodies suggested possible Z-disc localization. To rule out potential cross-reactivity of these antibodies to other proteins known to be localized at the Z-disc, we performed immunocytochemistry on COS cells transfected with representative Z-disc proteins. As shown in Fig. 7, the anti-Myomaxin antibodies specifically detected COS cells transfected with Myomaxin but not those cells expressing ␣-actinin, muscle LIM protein, or T-cap/telethonin, indicating that the striated signal in muscle cryosections corresponds to Myomaxin.
To more precisely determine that the striated signal actually represented Z-disc localization, we performed double label immunohistochemistry with anti-␣-actinin-2 (Sigma). As shown in Fig. 8A (top), anti-sarcomeric ␣-actinin-2 (left; green) exhibits a periodic, Z-line staining pattern as expected. Anti-Myomaxin (middle; red) displays similar periodic striations in these myofibers. When these images are superimposed (right), the periodic striations are now yellow, indicative of co-localization of both proteins to the Z-disc. We also examined transverse sections of skeletal muscle using the anti-Myomaxin antibodies. These immunostaining results revealed specific reactivity along the periphery of myofibers indicating localization at the sarcolemmal region (Fig. 6A, bottom). Taken together, the immunostaining data from longitudinal and transverse sections indicate Myomaxin localization at the level of the Z-disc in the costameric region of striated muscle.
Myomaxin Interacts Directly with the Sarcomeric Z-disc Protein ␣-Actinin-Based on the above co-localization studies, we sought to demonstrate a direct interaction between Myomaxin and ␣-actinin-2. For these studies, COS cells were transfected with full-length MYC-Myomaxin (amino acids 1-3,283) and FLAG-␣-actinin-2. These extracts were immunoprecipitated with anti-MYC antibody and subsequently used in Western blot analysis with anti-FLAG antibodies. As shown in Fig. 8B  (left), a band corresponding to ␣-actinin-2 could be immunoprecipitated from extracts containing full-length Myomaxin (amino acids 1-3,283) but not in those extracts transfected with pCDNA3 vector.
We then sought to delineate the minimal region in Myomaxin that interacts with ␣-actinin by generating several overlapping FLAG-tagged fragments spanning the entire open reading frame of Myomaxin and testing these by co-immunoprecipitation (co-IP). For these deletion experiments, ␣-actinin-2 was expressed as an N-terminal Myc-tagged fusion protein. The results from these co-IP experiments are summarized in Fig. 8B (right). Those polypeptide fragments either N-terminal or C-terminal to the actin binding repeat region, such as Pre-Xin, Frag1A, and FragB, were unable to interact with ␣-actinin. In addition, a fragment within the actin-binding repeat region, XinC, was unable to interact. However, two subdomains embedded within the actin-binding repeat region interacted effectively with ␣-actinin (Fig. 8B, right). Potential interactions between ␣-actinin and Myomaxin fragments, XinB and Frag2A, could not be tested, since these constructs did not express in COS cells.

DISCUSSION
An in silico approach was used to identify additional "cardiomyopathy-associated" or CMYA genes. Four human expressed sequence tags named CMYA1, -2, -3, and -4 were identified, and each of these was examined for their possible dysregulation in mef2a knock-out hearts. Of the four novel CMYA genes identified, the mouse ortholog of CMYA3 was the only gene to be significantly down-regulated in mef2a null hearts. The full-length mouse cDNAs have been cloned for all the CMYA genes except for CMYA3. We report the cloning of the full-length mouse ortholog of CMYA3, named myomaxin, and demonstrate that it is a direct MEF2 target gene. The myomaxin gene product harbors an actin-binding 16-amino acid repeat region and exhibits similarity within this region to the muscle protein Xin. Myomaxin has no obvious homology to any other protein family in the data base. myomaxin now represents another novel MEF2A-dependent gene down-regulated in mef2a knock-out hearts whose gene product localizes to the peripheral Z-disc region (or costamere), where it interacts with sarcomeric ␣-actinin. Our recent findings point to the potential importance of MEF2A regulating the Z-disc/costamere in striated muscle by controlling the expression of genes whose products are targeted to these specialized structures.
Responsiveness of Myomaxin to MEF2A, Myogenic Differentiation, and Hypertrophy-Our studies have established for the first time that myomaxin is a direct transcriptional target for MEF2A in striated muscle. The myomaxin promoter contains three MEF2 cis-elements within 5.0 kb of the transcription start site. Mutation and deletion analysis of each of the three MEF2 cis-elements has revealed that the Ϫ75 MEF2 binding site is the only functionally important site for full activation of the myomaxin promoter. The Ϫ75 MEF2 site is evolutionarily conserved in rodents and humans, it is bound by MEF2A in vivo, and it is also the site that has the most dramatic effect on transcriptional activity. To date, myomaxin represents the most down-regulated gene we have identified in mef2a knock-out mice, since its expression is virtually undetectable in these mutant hearts. It is tempting to speculate that the nearly complete absence of myomaxin expression may contribute to the severity of the myofibrillar defects in mef2a mutant hearts.
It was previously demonstrated that the myomaxin related gene, Xin, is also regulated by the MEF2C transcription factor (19). Consistent with previous results, our RT-PCR analysis revealed that Xin gene expression is dependent on MEF2 activity, since it is down-regulated in MEF2A-deficient hearts. However, it is not down-regulated to the extent by which myomaxin is affected, demonstrating that Xin is less sensitive to the loss of MEF2A activity. The different responsiveness of myomaxin and Xin gene expression in mef2a null hearts may reflect the ability of MEF2A to interact selectively with co-factors within the respective regulatory regions or specific combinations of MEF2 proteins, which function as homo-and heterodimers, within each promoter region. It remains to be determined whether myomaxin gene expression is dysregulated in mef2c mutant hearts.
In the C2C12 myogenic differentiation model, we observed that myomaxin is expressed in both proliferating myoblasts and differentiated myotubes. Since MEF2A is not expressed in proliferating C2C12 myoblasts, the expression of myomaxin in myoblasts is probably regulated by MEF2D, the only MEF2 factor expressed in undifferentiated muscle cells (26). We noted a modest induction of myomaxin at the onset of myogenic differentiation, but it was not until very late in this process that we regions may dictate ␣-actinin binding specificity. This seems unlikely, because the interrepeat sequences in the XinA and PostXin fragments that bind to ␣-actinin are not any more similar to each other than to XinC. A final possibility is that secondary structure or a combination of secondary structure and repeat sequences may determine binding interaction with ␣-actinin. Interestingly, secondary structure analysis predicts the formation of numerous ␣-helices within the actin-binding repeat fragments, XinA and PostXin, whereas the XinC subdomain does not harbor any predicted ␣-helical structure.
Relationship to the Intercalated Disc Protein, Xin-Myomaxin exhibits modest similarity to Xin within the 16-amino acid repeat region but no obvious homology outside this region. It is within this region that Xin and the human ortholog of Myomaxin, Xirp2, interact with actin. Given that Myomaxin interacts with ␣-actinin within the actin-binding repeat region, it would also be of interest to determine whether Xin also interacts with ␣-actinin. So why is it that striated muscle has two related cytoplasmic proteins that serve to bind F-actin? This may be explained partly by the different subcellular localizations of Myomaxin and Xin. Immunolocalization studies have revealed that Xin resides within the adherens junction of the intercalated disc in cardiac muscle and the functionally equivalent myotendinous junction in skeletal muscle (28). In contrast, Myomaxin is precisely localized at the peripheral Z-disc/costameric region, a subcellular structure, coincidentally, completely devoid of Xin immunoreactivity (28). These results suggest that two related actin-binding proteins have evolved to perform unique functions, distinct from binding actin, which enables them to coordinate the assembly and organization of the actin cytoskeleton and associated protein complexes within their respective subcellular locations. Indeed, Xin has been demonstrated to exist in a complex with N-cadherin and ␤-catenin, two proteins found in the adherens junctions (28).
Potential Function of Myomaxin-The importance of the Z-disc and peripheral structures in the coordination of striated muscle contraction is well established (29 -31). In recent years, many novel, muscle-specific proteins residing within these structures have been identified, thereby adding to the complexity of this subcellular region. The diversity of protein-protein interactions further adds an additional layer of specificity to this region.
The observation that myomaxin is down-regulated in mef2a knock-out hearts suggests that it functions in a regulatory pathway required for the structural assembly and integrity of the cytoskeleton in striated muscle. Also of note is the up-regulation of myomaxin under conditions that stimulate muscle hypertrophy, a condition that involves cytoskeletal remodeling events. This raises the possibility that perturbations in the expression and/or function of myomaxin may be a contributing factor in the pathogenesis of muscle disease. Future studies investigating the in vivo function of Myomaxin will provide insights into the role of this protein in striated muscle biology and contribute to an understanding of the myogenic pathways regulated by MEF2A.