α-Actinin 4 Potentiates Myocyte Enhancer Factor-2 Transcription Activity by Antagonizing Histone Deacetylase 7*

Histone deacetylase 7 (HDAC7) is a member of class IIa HDACs that regulate myocyte enhancer factor-2 (MEF2)-mediated transcription and participate in multiple cellular processes such as T cell apoptosis. We have identified α-actinin 1 and 4 as class IIa HDAC-interacting proteins. The interaction domains are mapped to C terminus of α-actinin 4 and amino acids 72-172 of HDAC7. A point mutation in HDAC7 that disrupts its association with MEF2A also disrupts its association with α-actinin 4, indicating that MEF2A and α-actinin 4 binding sites largely overlap. We have also isolated a novel splice variant of α-actinin 4 that is predominantly localized in the nucleus, a pattern distinct from the full-length α-actinin 4, which is primarily distributed in the cytoplasm and plasma membrane. Using small interfering RNA, chromatin immunoprecipitation, and transient transfection assays, we show that α-actinin 4 potentiates expression of TAF55, a putative MEF2 target gene. Loss of MEF2A interaction correlates with loss of the ability of α-actinin 4 to potentiate TAF55 promoter activity. Ectopic expression of α-actinin 4, but not the mutant defective in MEF2A association, leads to disruption of HDAC7·MEF2A association and enhancement of MEF2-mediated transcription. Taken together, we have identified a novel mechanism by which HDAC7 activity is negatively regulated and uncovered a previously unknown function of α-actinin 4.

The ␣-actinins are an actin-binding protein family that consists of four members including muscle-specific ␣-actinin 2 and 3 and the ubiquitously expressed ␣-actinins 1 and 4 (1). All four proteins share extensive sequence homology with a conserved organization of functional domains that include an N-terminal actin binding domain composed of 2 calponin homology (CH) 4 domains, a central rod domain consisting of four spectrin repeats (SR), 2 EF-hand calcium binding domains, and a C-terminal calmodulin-like domain (2,3).
Despite intense study, the cellular functions of ␣-actinin 4 remain unclear. ␣-Actinin 4 has been shown to bind F actin to modulate cytoskeleton organization and cell motility (4). In addition, another report suggested that ␣-actinin 4 participates in apoptosis (5). Although ␣-actinin 4 has been shown to associate with transcription factors, there is no functional data suggesting that it is involved in transcriptional regulation (6,7). Mutations of ␣-actinin 4 are linked to familial focal segmental glomerulosclerosis (8) and deletion of ␣-actinin 4 in mice causes severe glomerular disease (9). These observations suggest that ␣-actinin 4 plays a role in regulating multiple cellular processes and animal development.
Although predominately localized in the cytoskeleton, ␣-actinin 4 is also found in the nucleus of certain cell types (4) or translocates into the nucleus in response to extracellular stimuli. For example, treatment with phosphatidylinositol 3-kinase inhibitors or actin depolymerization can lead to nuclear accumulation of ␣-actinin 4 (4). However, the mechanism by which ␣-actinin 4 shuttles between the nucleus and the cytoplasm remains unclear, and the functional significance of nuclear ␣-actinin 4 has not been demonstrated.
Histone deacetylases (HDACs) are well known for their function as epigenetic regulators of the transcription machinery. In mammals, 18 different HDACs have been identified and grouped into three distinct classes based on sequence homology, subcellular localization, and the chemistry of their enzymatic activity. Class II HDACs can be subdivided into class IIa that includes HDAC4, -5, -7, and -9 (will be referred as class II hereafter) and class IIb, HDAC6 and -10 (10). Class II HDACs play a pivotal role in cell differentiation and animal development, partly due to their association with MEF2 transcription factors (10,11). Class II HDACs are unique in their ability to shuttle between the nucleus and the cytoplasm. Consequently, nucleocytoplasmic shuttling of class II HDACs is an important regulatory mechanism that controls the activity of MEF2 (10,11). During muscle differentiation and thymocyte develop-ment, class II HDACs are sequestered in the cytoplasm (12,13). As a result, transcriptional repression by HDACs is relieved, leading to up-regulation of MEF2 target genes, such as muscle creatine kinase and Nur77 and subsequent muscle differentiation and thymocyte development, respectively. Thus far, the mechanism accounting for de-repression of MEF2 activity (by class II HDACs) has been largely attributed to the activity of calmodulin kinase and chromosome region maintenance 1 (CRM1) proteins that promote cytoplasmic retention of class II HDACs.
In this study we have identified ␣-actinin 1 and 4 as HDAC7interacting proteins and defined their novel function. We also isolated a novel splice variant of ␣-actinin 4. We demonstrate that ␣-actinin 4 is capable of interacting with class II HDACs and potentiates transcription activity by MEF2. Taken together, our data support a model in which ␣-actinin 4 potentiates transcriptional activity of MEF2 in part by antagonizing HDAC7 activity.

EXPERIMENTAL PROCEDURES
Construction of HeLa Yeast Two-hybrid Library-HeLa cell cDNAs were generated by RT-PCR according to manufacturer's protocol (Invitrogen). cDNAs sized above 1 kilobase were isolated from gels and purified. EcoRI linker was added to both ends of the cDNAs, and the resulting cDNAs were ligated with EcoRI-digested pGAD vector (Stratagene). Ligation reactions were transformed into Escherichia coli (Stratagene). More than 1 ϫ 10 6 independent clones were obtained.
Yeast Two-hybrid Screens-Yeast two-hybrid (Y2H) screens were carried out using the standard lithium acetate method. HeLa Y2H library and pGBT9-HDAC7 (2-533, S178E/S344E/S479E) were co-transformed into yeast strain Y190. Approximately 3 ϫ 10 6 yeast transformants were screened and selected on yeast minimal medium ϪLeu-Trp-His plates containing 40 mM 3-aminotriazole (Sigma). After 7 days, colonies were picked and confirmed by ␤-galactosidase assays. For confirmation, plasmids were recovered from yeast and retransformed into yeast along with the bait construct. Positive clones were then subjected to sequencing. In addition to clones encoding ␣-actinin 1 and 4, we also isolated several cDNA clones encoding C-terminal binding proteins (CtBP) proteins, consistent with previous reports (14). Standard Y2H assays were carried out according our published protocol (15).
RT-PCR-RNA isolation from MCF7, HeLa, and HEK293 cells was performed using the RNeasy Mini RNA isolation kit (Qiagen). All procedures were performed according to the manufacturer's protocol. DNase I digestion was performed on the RNeasy column following the manufacturer's suggestions using RNase-and DNase-free DNase I (Qiagen). The isolated RNA was used in a semi-quantitative RT-PCR reaction using the One-Step RT-PCR kit (Invitrogen) according to the manufacturer's protocol. 200 ng of RNA template was used in each reaction. PCR amplification was repeated for 45 cycles. The final primer concentration in each reaction was 0.2 M. The primers used are as follows: 5Ј-GCATGGTGCAACTCC-CACCTG-3Ј (forward), 5Ј-CCCCCGCTCTGGCTTAGG-3Ј (reverse, full-length-specific), 5Ј-GATATGACCACCAG-GAGCAG-3Ј (reverse, isoform-specific), and 5Ј-GGGTGTTG-ACATTGTGGGAGTCG-3Ј (reverse, both full-length and isoform).
Plasmid Construction-Full-length cDNA of actinin ␣4 was generated by PCR using a ZAP-expressed phage vector (a generous gift from Dr. Hirohashi) as a template. ␣-Actinin 4 (isoform) cDNA was PCR-amplified using Y2H clone as a template. Truncated and deleted ␣-actinin 4 cDNAs were PCR-amplified using full-length ␣-actinin 4 or its isoform as a template. The cDNAs were cloned into pCMX-PL1-HA or CMX-PL1-FLAG vectors (16) digested with EcoR1 and NheI site. For the glutathione S-transferase (GST) constructs, ␣-actinin 4 was PCRamplified and subcloned into pGEX4T vector. MEF2 expression plasmids and MEF2 reporter construct have been previously described (17). HDAC7 point mutation and deletions were generated by PCR reactions. TAF55 reporter constructs were gifts from Dr. Cheng-Ming Chiang (18).
Antibodies-␣-Actinin 4 antiserum was generated using GST-␣-actinin 4 (isoform) fusion proteins. Antibodies were purified by sequential purification through GST and GST-␣actinin 4 affinity columns. Anti-HA, anti-MEF2, and anti-FLAG antibodies were purchased from Santa Cruz and Sigma. HDAC7 antibodies were generated with peptide encompassing amino acids 115-129 of mouse HDAC7. HDAC7 antibodies were purified from peptide affinity chromatography (Affinity BioReagents). Anti-HDAC7 antibodies did not cross-react with HDAC4 or HDAC5 (data not shown).
GST Pulldown Assays-GST fusion proteins GST-␣-actinin 4 and GST-HDAC7 were expressed in E. coli DH5␣ strain, affinity-purified, and immobilized on glutathione-Sepharose 4B beads. GST pulldown assays were carried out with purified, immobilized GST fusion proteins incubated with whole cell extracts expressing FLAG-or HA-tagged proteins. Binding reactions were carried out at 4°C for 1 h with NETN buffer (20 mM Tris-HCl, pH 8.0, 100 mM NaCl, 1 mM EDTA, 0.1% Nonidet P-40, 10% glycerol, 2 M phenylmethylsulfonyl fluoride, 1 mM dithiothreitol) followed by extensive washes. The bound fractions were separated on SDS-PAGE gel and subjected to immunoblotting with anti-HA or anti-FLAG antibodies. For pulldown assays, 10% of the input is shown.
Immunoprecipitation-Cells were grown on 10-cm plates and transfected with appropriate plasmid (10 g of total DNA) with Lipofectamine 2000 (Invitrogen). After 48 h, the cells were washed in 1ϫ PBS and resuspended in NETN buffer with protease inhibitors. After incubating on ice for 2 h, the lysed cells were centrifuged at 4°C at 10,000 RPM for 10 min; supernatant was collected and kept at Ϫ80°C. These lysates were incubated with appropriate antibody for 4 h at 4°C, and then protein A/G beads were incubated with pre-clear whole cell extracts for 2 h at 4°C. The immunopellets were washed 3-4 times followed by Western blots probed with the appropriate antibody. Ten percent of the input is shown on Western analyses.
Confocal Microscopy-Transfected cells were fixed in 3.7% paraformaldehyde in PBS for 30 min at room temperature and permeabilized in PBS with the addition of 0.1% Triton X-100 and 10% goat serum for 10 min. The cells were washed 3 times with PBS and incubated in a PBS, goat serum (10%) plus 0.1% Tween 20 solution (ABB) for 60 min. Incubation with primary antibodies was carried out for 120 min in ABB.
The cells were washed 3 times in PBS, and the secondary antibodies were added for 30 -60 min in the dark at room temperature in ABB. Coverslips were mounted to slides using Vectashield mounting medium with DAPI (4Ј,6-diamidino-2-phenylindole; H-1200, Vector Laboratories, Inc.). All confocal images were acquired using a Zeiss LSM 510 inverted laser-scanning confocal microscope. A 63ϫ numerical aperture of 1.4 oil immersion planapochromat objective was used for all experiments. For endogenous and transiently transfected ␣-actinin 4, images of Alexa Fluor 594 were collected using a 633-nm excitation light from a He/ Ne2 laser, a 633-nm dichroic mirror, and 650-nm long pass filter. For endogenous HDAC7, images of Alexa Fluor 488 were collected using a 488-nm excitation light from an argon laser, a 488-nm dichroic mirror, and 500 -550-nm band pass barrier filter. All 4Ј,6-diamidino-2-phenylindolestained nuclear images were collected using a Coherent Mira-F-V5-XW-220 (Verdi 5W) Ti-Sapphire laser tuned at 750-nm, a 700-nm dichroic mirror, and a 390 -465-nm band pass barrier filter. The primary antibodies used were purified ␣-HDAC7 rabbit polyclonal (Affinity BioReagents) and ␣-FLAG mouse monoclonal antibodies (Sigma). The secondary antibodies used were from Molecular Probes (␣-mouse or ␣-rabbit Alexa Fluor 488, ␣-rabbit Alexa Fluor 594, and mouse Alexa Fluor 594).
Transient Transfection and Luciferase Assay-Transient transfections and luciferase assays were performed in 48-well culture plates. HeLa and CV-1 cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 50 units/ml penicillin G, and 50 g of streptomycin sulfate at 37°C in 5% CO 2 . Cells were transfected with a muscle creatine kinase promoter-containing reporter construct (17) and CMX-␤-gal expression plasmids. Lipofectamine 2000 was used as a transfection reagent. The amount of DNA was kept constant by the addition of pCMX vector. After 5 h the medium was replaced, and the cells were harvested after 48 h of transfection and luciferase assay, and ␤-galactosidase activity was measured by using the luciferase assay system (Promega). Each reaction was performed in triplicate. Results shown indicate luciferase activity normalized to ␤-galactosidase levels. Transient transfection assays for TAF55 were carried out using as reporter constructs as described (18).
Total RNA extraction was prepared using the RNeasy commercial kit (Qiagen). Semi quantitative RT-PCR was performed using the SuperScript One step RT-PCR kit with Platinum Taq. 500 ng of RNA was used for each reaction. The primers used for RT-PCR to examine the effects of ␣-actinin 4 siRNA are: TAF55, 5Ј-GAAGGGCAGTACAGTCTGGT-3Ј (forward) and 5Ј-GGA-TACAAGCATCTGACA-3Ј (reverse); glyceraldehyde-3-phosphate dehydrogenase, 5Ј-TGGTCACCAGGGCTGCTT-3Ј (forward), 5Ј-AGCTTCCCGTTCTCAGCCTT-3Ј (reverse). The primers to detect expression of the full-length and isoform of ␣-actinin 4 are described above. The intensity of the signals was quantified using the VERSADOC 3000 (Bio-Rad).
To verify that the cloned ␣-actinin 4 (isoform) cDNA was generated from mRNA, we carried out RT-PCR to examine the presence of this spliced isoform. Human ␣-actinin 4 is pre-dicted to consist of 21 exons. Based on the isolated cDNA sequence, exons 3-12 are spliced out in the isoform. Primers that anneal to full-length ␣-actinin 4 mRNA (lanes 1, 3, and 5) or its isoform (lanes 2, 4, and 6) were used for RT-PCR using RNA prepared from MCF-7 (lanes 1 and 2), HeLa (lanes 3 and 4), or HEK293 (lanes 5 and 6) cells as templates (Fig. 1B). We found that MCF-7, HeLa, and HEK293 cells express both fulllength ␣-actinin 4 and the spliced isoform. Using a pair of primers capable of annealing to both full-length ␣-actinin 4 and the spliced mRNA variant, we were able to detect the presence of both the full-length ␣-actinin 4 and its splice variant (lanes 7-9). Sequence analyses of the eluted, purified PCR product also indicated that the sequence of the PCR product matched that of the isolated ␣-actinin 4 (isoform) cDNA (data not shown). These data verified the presence of the novel isoform of ␣-actinin 4 in MCF-7, HeLa, and HEK293 cells. We further tested whether the spliced mRNA variant is indeed translated. A full-length or isoform HA-␣-actinin 4 expression plasmid was transiently transfected into HeLa cells. Whole cell lysates were prepared followed by Western blotting with anti-␣-actinin 4 or anti-HA antibodies. Fig. 1C demonstrates that in the absence of transfected plasmid, in addition to the full-length ␣-actinin 4, anti-␣-actinin 4 antibodies detected another minor protein species, which migrated faster than full-length ␣-actinin 4 (lanes 1 marked with an asterisk). In cell lysates expressing HA-␣-actinin 4 isoform or full-length, we detected both endogenous and transfected isoform (lane 2) and full-length proteins (lane 3). As expected, Western blotting with anti-HA antibodies detected ␣-actinin 4 isoform (lane 5) and the full-length protein (lane 6). These data indicate that the cloned ␣-actinin 4 (isoform) comigrates with the endogenous ␣-actinin 4 (lanes 1  and 2), indicating that the isoform is endogenously expressed in HeLa cells. Furthermore, the low expression of ␣-actinin 4 (isoform) protein observed is consistent with our RT-PCR data, demonstrating that full-length ␣-actinin 4 protein is expressed in excess to the isoform in HeLa cells.
Because HDAC7 (S178E/S344E/S479E) was used as bait to screen for HDAC7-interacting proteins, we tested whether ␣-actinin 4 is capable of interacting with wild-type HDAC7 by Y2H assays. We found that wild-type and mutant HDAC7 (S178E/S344E/S479E) interacted equally well with ␣-actinin 4 ( Fig. 2A). To investigate whether endogenous ␣-actinin 4 associates with HDAC7, we carried out co-immunoprecipitation with extracts prepared from HeLa and CV-1 cells and found that endogenous HDAC7 and ␣-actinin 4 associate in these cells (Fig. 2B). We then tested the interaction of other ␣-actinins and HDAC7 by co-immunoprecipitation experiments. HA-HDAC7 expression plasmid was co-transfected with expression plasmids encoding FLAG-␣-actinin 1, FLAG-␣-actinin 4, or FLAG-␣-actinin 4 (isoform) into HeLa cells. Immunoprecipitation and immunoblotting were carried out using anti-FLAG and anti-HA antibodies. As shown in Fig. 2C, HA-HDAC7 was immunoprecipitated in the presence of ␣-actinin 1 and ␣-actinin 4. Furthermore, both full-length and ␣-actinin 4 (isoform) were co-precipitated. These data indicate that HDAC7 is capable of interacting with both ␣-actinin 1 and 4 in mammalian cells.
To understand the molecular basis of the interaction between HDAC7 and ␣-actinin 4, we mapped the interaction domains by yeast two-hybrid and GST pulldown assays. Deletion and truncation fragments of HDAC7 fused with yeast Gal4 DNA binding domain (pGBT9-HDAC7) were co-transformed into yeast with yeast Gal4 activation domain fused with ␣-actinin 4 (pGAD-␣-actinin 4). ␤-Galactosidase assays were carried out to examine the interaction. We found that HDAC7 fragments containing amino acids 2-172 and 72-254 were sufficient to interact with ␣-actinin 4, suggesting a minimal interaction domain localized within amino acids 72-172 of HDAC7 (Fig. 3A). The observation that both ␣-actinin 4 and its isoform interact with HDAC7 suggests that amino acids 644 -911 are sufficient for the association with HDAC7 (Fig. 1A). To further map the interaction, we generated deletion and truncation constructs from amino acids 644 of full-length ␣-actinin 4. Fig. 3B shows that amino acids 832-911 of ␣-actinin 4 were sufficient to interact with HDAC7.
Amino acids 72-172 of HDAC7 share extensive sequence homology with other members of the class II HDACs, indicating that ␣-actinin 4 likely associates with other members of class IIa HDACs. To test this possibility, we carried out GST pulldown assays with the class II HDACs. Immobilized GST-␣actinin 4 (isoform) fusion protein was incubated with whole cell extracts expressing HA-tagged HDAC1, -4, -5, and -7 and Western blotting with anti-HA antibodies. Fig. 3G shows that ␣-actinin 4 interacts with HDAC4 and HDAC5 but not HDAC1. These data indicate that ␣-actinin 4 interacts specifically with class II HDACs both in vivo and in vitro.
␣-Actinin 4 has been detected in the nucleus, cytoplasm, and plasma membrane depending on the cell type and culture conditions (4). The identification of the ␣-actinin 4 (isoform) raised the possibility that this isoform has a distinct subcellular localization from full-length ␣-actinin 4. Therefore, confocal microscopy experiments were carried out to examine the subcellular distribution of ␣-actinin 4 (isoform) in HeLa cells. A FLAG-tagged full-length ␣-actinin 4 expression plasmid was transfected into HeLa cells, and immunostaining was carried FIGURE 3. Mapping of the interaction domains between ␣-actinin 4 and HDAC7. A, mapping of ␣-actinin 4 binding site in HDAC7 by Y2H assays. B, mapping of the HDAC7 binding site of ␣-actinin 4 by Y2H assays. C, mapping of the HDAC7 binding site in ␣-actinin 4 by GST pulldown assays. HEK293 cells were transfected with FLAG-HDAC7, and whole cell extracts were prepared. Truncated or deleted fragments of ␣-actinin 4 (isoform) were fused with GST and expressed in bacteria. Immobilized GST-␣-actinin 4 fusion proteins were incubated with whole cell extracts expressing FLAG-HDAC7 for pulldown assays. D, HEK293 cells were transfected with FLAG-␣-actinin 4, and whole cell extracts were prepared. GST-HDAC7 fusion proteins were prepared and incubated with whole cell extracts for GST pulldown assays. E, amino acids 1-832 of full-length ␣-actinin 4 do not bind HDAC7. Full-length or truncated (2-839) ␣-actinin 4 expression plasmid, FLAG-␣-actinin 4, was transfected into HeLa cells. Whole cell extracts were prepared for GST pulldown assays followed by Western blotting with anti-FLAG antibodies. F, amino acids 2-449 of the isoform fail to bind HDAC7. G, ␣-actinin 4 interacts with class IIa HDACs but not HDAC1. HEK293 cells were transfected with HA-HDAC1, HA-HDAC4, HA-HDAC5, or HA-HDAC7. Immobilized, purified GST-␣-actinin 4 (isoform) proteins were incubated with the whole cell extracts. After extensive washing, bound fractions were resolved on SDS-PAGE and Western blotting with ␣-HA antibodies. out using anti-FLAG and anti-HDAC7 antibodies to probe transfected ␣-actinin 4 and endogenous HDAC7, respectively. Fig. 4A shows that full-length ␣-actinin 4 was distributed throughout the whole cell including the plasma membrane and colocalized with ␣-actinin 4 (panels a, d, and e). In untransfected HeLa cells HDAC7 was primarily localized in the nucleus (panels c, e, and f ). In transfected cells, however, a significant fraction of HDAC7 was re-distributed to the cytoplasm (panel c). We also examined the subcellular localization of ␣-actinin 4 (isoform). To our surprise, FLAG-␣-actinin 4 (isoform) was found predominately to localize in the nucleus of HeLa cells (Fig. 4B). We also observed significant colocalization of transfected ␣-actinin 4 (isoform) and endogenous HDAC7. These data demonstrate that the subcellular localization of ␣-actinin 4 (isoform) is very different from that of the full-length ␣-actinin 4 and that overexpression of the full-length ␣-actinin 4 can sequester endogenous HDAC7 from its normal localization.
Class II HDACs associate with and repress transcription by MEF2 (10,11,17). It is possible that ␣-actinin 4 modulates transcriptional regulation by MEF2 through its association with class II HDACs. To test this possibility, we performed transient transfection assays using a reporter plasmid harboring a MEF2 binding site fused to the firefly luciferase gene (17). A constant amount of HDAC7 expression plasmid was co-transfected with the reporter construct and increasing amounts of full-length ␣-actinin 4 or its isoform. Fig. 5A shows that in CV-1 cells, overexpression of HDAC7 inhibits the transcription activity of the reporter gene (lanes 1 and 2). However, co-expression of full-length ␣-actinin 4 (lanes 3-5), isoform (lanes 6 -8), or ␣-actinin 1 (lanes 9 -11) with HDAC7 was capable of relieving inhibition by HDAC7 in a dosage-dependent manner. Similarly, we observed relief of HDAC7-mediated transcription repression by both full-length ␣-actinin 4 and its isoform in HeLa cells (Fig. 5B).
We also observed a slight activation of the reporter activity at the highest concentration of ␣-actinin 4 (Fig. 5, lanes 5 and 8), suggesting ␣-actinin 4 is capable of potentiating MEF2 activity. To test this possibility, we co-transfected ␣-actinin 4 and MEF2C expression plasmids along with the reporter construct in the absence of exogenous HDAC7 expression plasmid.   HeLa cells (Fig. 6B). We further tested whether the HDAC7 interaction domain was essential for ␣-actinin 4 to activate the MEF2 reporter. Fig. 6C demonstrates that in CV-1 cells co-transfection of ␣-actinin 4 (isoform) activated the MEF2 reporter construct, whereas a mutant lacking the HDAC7 interaction domain (2-449) did not. A similar trend was observed in HeLa cells (Fig.  6D). These data suggest that the same domain of ␣-actinin 4 is critical for HDAC7 interaction and transcription activation of a MEF2 reporter construct.
To verify whether our findings apply to endogenous MEF2 target genes, we examined whether ␣-actinin 4 is capable of modulating expression of TAF55 by siRNA knockdown experiments. TAF55 has been proposed to be a MEF2 target gene (18). To test whether expression of endogenous TAF55 is regulated by ␣-actinin 4, we introduced siRNA against ␣-actinin 4 or a control siRNA into HeLa cells and measured expression of TAF55 by RT-PCR. Fig. 7A demonstrates that knockdown of endogenous ␣-actinin 4, but not the control siRNA, significantly decreased the expression of TAF55. As a control, glyceraldehyde-3-phosphate dehydrogenase expression was included. To test whether ␣-actinin 4 associates with TAF55 promoter, we carried out ChIP assays. Fig. 7B demonstrates that MEF2, HDAC7, and ␣-actinin 4 associate with the DNA sequence flanking nt Ϫ264 to ϩ7 upstream of the transcription start site of the TAF55 promoter (top panel). As a control, MEF2, HDAC7, and ␣-actinin 4 do not associate with nt Ϫ2898 to Ϫ2720 upstream of the transcription start site (bottom panel). We further investigated whether MEF2, ␣-actinin 4, and HDAC7 regulate TAF55 expression by transient transfection assays. A reporter construct harboring TAF55 basal promoter including the putative MEF2 binding site was co-transfected with MEF2C, HDAC7, and/or ␣-actinin 4 into HeLa cells. As a control, a deletion construct in which the MEF2 binding site was removed was included for parallel analyses. Fig. 7C shows that MEF2C, HDAC7, and/or ␣-actinin 4 did not have effects on the reporter activity of the control construct (lanes 1-10). In contrast, whereas MEF2C and ␣-actinin 4 potentiated the reporter activity (lanes 12, 14 -20), HDAC7 inhibited the reporter activity (lanes 11 and 13). Taken together, these results indicate that MEF2C, HDAC7, and ␣-actinin 4 control TAF55 expression through association with its promoter.
We further tested whether the ability of ␣-actinin 4 to potentiate MEF2 activity required physical association. To do so we generated an ␣-actinin 4 mutant in which amino acids 441-479 were deleted. This mutant no longer interacted with HDAC7 ( Fig. 9A) or MEF2A (Fig. 9B) but was still localized in both the nucleus and cytosol (Fig. 9C). Furthermore, this mutant dramatically lost its ability to potentiate TAF55 promoter activity in transient transfection assays (Fig. 9D). These observations suggest that the ability of ␣-actinin 4 to activate MEF2 activity correlates with its ability to interact with MEF2 and/or HDAC7. If ␣-actinin 4 competes with HDAC7 for binding to MEF2, we anticipate that overexpression of ␣-actinin 4 would disrupt the association of HDAC7 with MEF2. To test this possibility, constant amounts of YFP-HDAC7 and FLAG-MEF2A expression plasmids were co-transfected into HeLa cells with an increasing concentration of HA-␣actinin 4 (isoform). If ␣-actinin 4 sequestered HDAC7, less HDAC7 would be co-precipitated with MEF2A when ␣-actinin 4 was co-expressed. As shown in Fig. 9E, in the absence of HA-␣-actinin 4, anti-FLAG antibodies co-precipitated with YFP-HDAC7 (lanes 1 and 4). However, in the presence of increasing concentration of HA-␣-actinin 4, less YFP-HDAC7 FIGURE 7. ␣-Actinin 4 is required for TAF55 expression. A, knockdown of ␣-actinin 4 decreased expression of TAF55. ␣-Actinin 4 or a control siRNA was transfected into HeLa cells. Total RNA was prepared for RT-PCR reactions to examine the levels of ␣-actinin 4, TAF55, and glyceraldehyde-3-phosphate dehydrogenase (GAPDH). The percentage remaining after knockdown is shown from three independent experiments. B, association of MEF2, ␣-actinin 4, and HDAC7 with TAF55 promoter. ChIP assays were performed to detect association of MEF2, ␣-actinin 4, and HDAC7 with TAF55 promoter. Top, PCR product flanking nt Ϫ264 to ϩ7; bottom, PCR product flanking nt Ϫ2898 to Ϫ2720 (control). Mock, pre-immune serum. ChIP assays were repeated three times. A representative data is shown. C, transient transfection assays were performed to evaluate the effects of MEF2C, ␣-actinin 4, and HDAC7 on TAF55 reporter with or without a MEF2 binding site. shown. E, a sequence alignment of HDAC4, HDAC5, and HDAC7. mHDAC, mouse HDAC; hHDAC, human HDAC. F, isolation of HDAC7 mutant defective in its interaction with ␣-actinin 4. MEF2A and ␣-actinin 4 binding sites largely overlap. Note that HDAC7 (L112A) specifically loses its interaction with MEF2A and ␣-actinin 4 but not 14-3-3⑀. WT, wild type. was co-precipitated (lanes 2 and 3 and lanes 5 and 6). In contrast, ␣-actinin 4 (⌬441-479) did not have any effect on the association between HDAC7 and ␣-actinin 4. These data indicate that ␣-actinin 4 has an inhibitory effect on the association of MEF2 with HDAC7 and that MEF2 and/or HDAC7 interaction domain is critical for this inhibitory activity.

DISCUSSION
Increasing evidence has indicated that some cytoskeletal proteins, including the focal adhesion Lim-domain protein Trip6 not only act in the cytoplasm but may also be active in nuclear events such as transcriptional regulation (20). In this study we demonstrate that ␣-actinin 4, a well characterized actin-binding protein involved in cross-linking actin filaments (21), is capable of potentiating transcription activation by MEF2C. We have also isolated a splice variant of ␣-actinin 4 that is predominantly localized in the nucleus of HeLa cells. Taken together, our data support a model in which ␣-actinin 4 potentiates transcriptional regulation by MEF2 transcription factors.
Several lines of evidence strongly support the notion that ␣-actinin 4 is a transcriptional co-regulator. First, transient transfection data indicate that ␣-actinin 4 potentiates transcriptional activity by MEF2. Second, ectopic expression of ␣-actinin 4 decreases the interaction of MEF2A and HDAC7. Third, knockdown of ␣-actinin 4 decreases expression of TAF55. Fourth, MEF2C, ␣-actinin 4, and HDAC7 associate with the TAF55 promoter. These observations suggest that nuclear ␣-actinin 4 is capable of modulating the activity of a subset of transcription factors. Therefore, how nucleocytoplasmic shuttling of ␣-actinin 4 is regulated becomes an important issue. Although ␣-actinin 4 does not contain any distinct nuclear localization signal, previous studies have demonstrated that ␣-actinin 4 is capable of localizing in the nucleus depending on cell type and extracellular signals (4). The observation that ␣-actinin 4 (isoform) is predominantly localized in the nucleus suggests that amino acids 89 -478 of the full-length ␣-actinin 4 have a negative effect on nuclear localization. Alternately, it is possible that nuclear localization of the ␣-actinin 4 (isoform) is simply due to diffusion because of its small size. It has been previously shown that treatment with a phosphatidylinositol 3-kinase inhibitor promoted nuclear accumulation of ␣-actinin 4 (4). Along with our findings, these data raise the possibility that phosphatidylinositol 3-kinase phosphorylates and controls the subcellular localization of ␣-actinin 4, thus regulating transcription and possibly other nuclear processes. Further investigation will be necessary to elucidate the  1 and 4) or presence of increased amounts of full-length HA-␣-actinin 4 or its mutant (lanes 2 and 3). Cell extracts were precipitated with anti-FLAG antibodies followed by Western blotting with anti-HDAC7 antibodies (lanes 5 and 6). mechanisms underlying the nucleocytoplasmic trafficking of ␣-actinin 4.
Our current data indicate that transcriptional regulation by MEF2 involves a protein-protein interaction network among MEF2, HDAC7, and ␣-actinin 4. Mapping the interaction domains indicates that ␣-actinin 4 binds to amino acids 2-86 of MEF2A (Fig. 8, B-D), which is highly conserved among MEF2 transcription factors. Notably, our published data indicated that HDAC7 binds to amino acids 1-86 of MEF2A (17), suggesting that MEF2 cannot bind HDAC7 and ␣-actinin 4 simultaneously. Indeed, we demonstrate that ␣-actinin 4 competes with HDAC7 for binding to MEF2A (Figs. 9, E and F ). Similarly, MEF2 and ␣-actinin 4 binding sites on HDAC7 largely overlap (Fig. 8E). Therefore, HDAC7 is unlikely to directly associate with MEF2 and ␣-actinin 4 simultaneously. Our data indicate that the ability of ␣-actinin 4 to potentiate MEF2 correlates with its ability to interact with MEF2 and/or HDAC7, suggesting a competition model in which MEF2 may directly recruit ␣-actinin 4 to displace HDAC7 from MEF2. Alternatively, HDAC7 may recruit ␣-actinin in response to stimuli followed by association of ␣-actinin 4 with MEF2 and activation of transcription. Further investigation will be required to distinguish between these two possibilities.
Overexpression and mutations of ␣-actinin 4 are linked to human cancers and kidney diseases, respectively (4,8,9). Although the precise roles of ␣-actinin 4 in tumorigenesis are controversial, ␣-actinin 4 has been proposed to play a role in cell motility and cancer invasion (4,(22)(23)(24)(25). Our findings that ␣-actinin 4 has a role in transcriptional regulation provides a possible, additional mechanism by which mutant ␣-actinin 4 might contribute to the pathogenesis of these diseases.
We demonstrate that overexpression of ␣-actinin 4 re-distributes the subcellular localization of endogenous HDAC7, disrupts the interaction of HDAC7 with MEF2A, and activates MEF2 transcription. However, ␣-actinin 4 (isoform, ⌬441-479), which does not interact with HDAC7, failed to effect MEF2-mediated transcription. These observations suggest that the ability of ␣-actinin 4 to activate MEF2-mediated transcription partly depends on the ability of ␣-actinin 4 to bind HDAC7. Amino acids 449 -521 harbor a calmodulin-like domain. In this study we show that this calmodulin-like domain is critical for interacting with amino acids 72-172 of HDAC7. Interestingly, previous reports have shown that calmodulin binds HDAC4 (26) and HDAC5 (27) through the corresponding region of HDAC7, and the association of calmodulin with class II HDACs resulted in their phosphorylation. It is likely that the association of ␣-actinin 4 with HDAC7 has a similar effect to that of calmodulin binding. The calmodulin-like domain (amino acids 449 -521) in ␣-actinin 4 (isoform) is highly conserved among members of ␣-actinin family. It is very likely that HDAC7 also interacts with other members of this family. Indeed, our data show that HDAC7 interacts with ␣-actinin 1 as well. These observations raise the possibility that muscle-specific actinins such as ␣-actinin 2 and 3 might also interact with class II HDACs. Because both class II HDACs and ␣-actinins play a role in myocyte differentiation (28), their physical interactions are likely significant. Future experiments will investigate the functional significance of the interaction between class II HDACs and the ␣-actinins during muscle differentiation.