Nuclear Rho Kinase, ROCK2, Targets p300 Acetyltransferase*

Rho-associated coiled-coil protein kinase (ROCK) is an effector for the small GTPase Rho and plays a pivotal role in diverse cellular activities, including cell adhesion, cytokinesis, and gene expression, primarily through an alteration of actin cytoskeleton dynamics. Here, we show that ROCK2 is localized in the nucleus and associates with p300 acetyltransferase both in vitro and in cells. Nuclear ROCK2 is present in a large protein complex and partially cofractionates with p300 by gel filtration analysis. By immunofluorescence, ROCK2 partially colocalizes with p300 in distinct insoluble nuclear structures. ROCK2 phosphorylates p300 in vitro, and nuclear-restricted expression of constitutively active ROCK2 induces p300 phosphorylation in cells. p300 acetyltransferase activity is dependent on its phosphorylation status in cells, and p300 phosphorylation by ROCK2 results in an increase in its acetyltransferase activity in vitro. These observations suggest that nucleus-localized ROCK2 targets p300 for phosphorylation to regulate its acetyltransferase activity.

ROCK is thought to be mediated primarily through the regulation of actin cytoskeleton dynamics (9,10). ROCK phosphorylates and activates LIM kinase, which in turn phosphorylates cofilin and inhibits its actin-depolymerizing activity, leading to increased actin polymerization (10). ROCK also phosphorylates the myosin-binding subunit (MBS) of myosin phosphatase and inactivates the phosphatase activity, which results in increased myosin phosphorylation and contractility (10).
Two members of the ROCK family, ROCK1 and ROCK2, have been identified. There is 65% identity in their overall amino acid sequences, and they share several common domains (11). Although ROCK1 and ROCK2 appear to have different activities on a cellular level (12), there has been little difference identified in phenotypes of ROCK1-and ROCK2-null mice, suggesting possible functional redundancy between the two ROCK isoforms (10,13). ROCK has three main domains critical for its functions: the N-terminal catalytic domain (CAT), a coiled-coil domain in the middle portion that mediates homodimerization, and the C-terminal Rho-binding (RB) and pleckstrin homology (PH) domains (RB/PH domain) (11, 14 -16). The RB/PH domain acts as an autoinhibitory domain by binding the CAT and suppressing its kinase activity in vitro. Consistently, when expressed in cells, the RB/PH domain alone can act as a dominant negative mutant (15,17). Conversely, various ROCK mutants lacking the RB/PH domain but retaining the CAT act as a constitutively active mutant (14,18). Such active ROCK mutants are naturally created during apoptosis: ROCK1 and ROCK2 can be cleaved by caspase-3 and granzyme B, respectively. These cleavages cause the loss of the C-terminal autoinhibitory domain containing the PH domain and the generation of active forms of ROCKs that contribute to the generation of cytoplasmic apoptotic features, such as apoptotic bleb (19 -23).
Besides its roles in cytoskeleton reorganization, cytoplasmic Rho signaling is also transmitted to the nucleus to regulate the functions of various transcription factors, including SRF, estrogen receptor, NF-B, GATA-4, and CREB, and a coactivator, FHL2 (1, 24 -26). These activities of Rho signaling appear to be mediated in part by ROCK (26,27). In addition, other kinases, such as c-Jun N-terminal kinase (JNK) (28,29) and p38 MAP kinase (25,30), may also be involved, although precise mechanisms for their regulation of transcription have not been fully understood. In certain cases, activation of Rho signaling may trigger translocation of transcriptional coactivators into the nucleus (26,31). Additionally, in cells undergoing epithelial-mesenchymal transition following transforming growth factor ␤ treatment, ROCK reportedly translocates into the nucleus (32,33) where it appears to inhibit the activity of Cdc25A phosphatase to regulate cell cycle progression (33).
In contrast to conventional Rho signaling initiated within the cytoplasm, several studies reported the existence of nucleus-localized Rho * This work was supported in part by Grant 3609 American Heart Association, Texas affiliate, Grant 0455058Y, Muscular Dystrophy Association (to Y. H.), and by the Naito Foundation (to T.T.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1  family members and their regulators; RhoA and RhoB have been detected in the nucleus by biochemical fractionation and immunofluorescence (34 -37). Two Rho guanine nucleotide exchange factors, Net1 and ECT2, are present primarily in the nucleus (38,39). Exogenously expressed Rac1 and Cdc42 are detected mainly in the nucleus (40). These findings are in line with the potential presence of nuclear Rho signaling, although more studies are needed to establish their nucleusspecific functions. Two paralogous acetyltransferases, p300 and CBP (p300/CBP), are recruited to specific gene promoters by associating with sequence-specific transcription factors and modulate the promoter activities by acetylating both histones and nonhistone substrates (41,42). In addition to acetyltransferase activities, p300/CBP have intrinsic transcriptional activation domains at their N and C termini (43,44). Several studies suggest that these activities of p300/CBP are subject to modulation through post-translational modifications in response to extracellular signals or through protein-protein interactions with viral and cellular proteins (26,(45)(46)(47)(48)(49)(50).
To better understand the regulatory mechanisms affecting p300 function, we sought new p300-interacting proteins using the yeast twohybrid system and identified ROCK2 as a candidate p300-interacting protein. Since p300 and ROCK2 have been previously localized predominantly in the nucleus and cytoplasm, respectively (5,41,51), our finding raises a critical question of which subcellular compartment might be the site of their interaction. Interestingly, by an independent approach, we identified ROCK2 as associated with a second nuclear protein, MLH1 (MutL homologue 1), which was involved in DNA base mismatch repair. Furthermore, by subcellular fractionation, ROCK2 was localized in the nucleus at a concentration comparable with the cytoplasm, and within the nucleus, ROCK2 was detected in the soluble nucleoplasmic fraction as well as the chromatin and nuclear matrix fractions. 3 Here, we show by several approaches that p300 associates with ROCK2 both in vitro and in intact cells. ROCK2 phosphorylates p300 in vitro, and nucleus-restricted expression of a constitutively active ROCK2 mutant induces p300 phosphorylation in cells. We find that acetyltransferase activity of p300 is dependent on its phosphorylation status in cells, and p300 phosphorylation by ROCK2 causes an increased acetyltransferase activity in vitro. These observations suggest that nuclear ROCK2 associates with p300, phosphorylates p300, and enhances its acetyltransferase activity through a phosphorylation-dependent mechanism.
Yeast Two-hybrid Screening-The yeast two-hybrid screening was performed using the Matchmaker two-hybrid system (Clontech) according to the manufacturer's protocol. The cDNA encoding the CH3 domain of human p300 (amino acids 1540 -1930) was cloned in frame into the EcoRI-BamHI site of pGBT9. The cDNA libraries of whole mouse embryo (embryonic day 9.5 and 10.5) cloned into pVP16 vector are generous gifts from Dr. Stanley Hollenberg (Oregon Health Sciences University).
Antibodies Tissue Culture and Transient Transfection-HeLa, 293T, and U2OS cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum. Transfection was performed by the BES-buffered saline protocol as described (47). Luciferase reporter gene assay was performed as described (47).
Preparation of Recombinant Protein and Protein-Protein Interaction Assays-GST fusion proteins were prepared in Escherichia coli (BL21-CodonPlus (DE3)-RP; Stratagene) by the method described (47). In vitro translated [ 35 S]methionine-labeled proteins were prepared using the TNT coupled transcription-translation system (Promega). For GST-p300 pull-down experiments with endogenous ROCK2, whole cell extracts were prepared by incubating cells in the NETN buffer (20 mM Tris-HCl, pH 8, 100 mM NaCl, 1 mM EDTA, 0.5% Nonidet P-40) supplemented with freshly prepared protease inhibitors (1 mM phenylmethylsulfonyl fluoride, 1 mM dithiothreitol, 2 g/ml pepstatin A, and aprotinin). Five hundred-microgram extracts were incubated with ϳ100 ng of GST fusion proteins. For co-immunoprecipitation studies, 1 mg of whole cell extracts or nuclear extracts were incubated with 4 g of either anti-p300 (N-15) or control normal mouse IgG overnight, followed by incubation of protein A-agarose (Sigma) for 2 h at 4°C. After washing four times, bound proteins were detected by Western analysis with anti-ROCK2 antibody (H-85).
Immunofluorescence Analysis-Cells were fixed with methanol or 4% paraformaldehyde in phosphate-buffered saline (PBS) for 10 min and were permeabilized for 10 min in HBS-PBS (5% horse serum, 1% bovine serum albumin and 0.1% saponin in PBS). After three washings with PBS, the cells were incubated with primary antibodies for 1 h at room temperature and then with Cy3-or fluorescein isothiocyanate-conjugated secondary antibodies (Sigma) in HBS-PBS for 1 h at room temperature with vigorous washings after incubation with each antibody. DAPI staining was performed to show chromatin DNA. To test the specificity of ROCK2 (H-85) and ROCK1 (H-85) antibodies, an excess of GST-ROCK2-H85 or GST-ROCK1-H85 protein (40-fold excess over the antibodies in molar ratio) was preincubated with the primary antibodies. Subcellular fractionation for immunofluorescence studies was performed as described (56). Briefly, cells were incubated with CSK buffer (10 mM PIPES, pH 6.8, 300 mM sucrose, 100 mM NaCl, 3 mM MgCl 2 , 1 mM EGTA, 0.5% Triton X-100, and the proteinase inhibitors) for 3 min on ice to give insoluble chromatin plus nuclear matrix fractions. Chromatin was removed by digesting genomic DNA with RNasefree DNase I (400 units/ml; Roche Applied Science) in the digestion buffer (CSK plus 50 mM NaCl and proteinase inhibitors) for 1 h at 37°C. Cells were further treated with ammonium sulfate to a final concentra-tion of 0.25 M for 5 min at room temperature and incubated with 2 M NaCl for 5 min at room temperature, with washings with the digestion buffer after each treatment, to give nuclear matrix. Cells were stained after each treatment, using polyclonal rabbit anti-ROCK2 (H-85) and mouse monoclonal p300 (NM11) antibodies. Cells were examined using an Axioplan2 microscope (Carl Zeiss Inc.).
Subcellular Fractionation-Subcellular fractionation was performed as described (53,57) with modifications. Cells were homogenized in buffer A (10 mM Tris-HCl, pH 7.4, 1.5 mM MgCl 2 , 10 mM KCl, 0.1 mM EDTA, 0.1 mM EGTA, 0.1% Triton X-100, and the proteinase inhibitors) with a Dounce homogenizer. The homogenates were loaded onto 1 M sucrose (final concentration at 270 mM) and centrifuged at 2,000 ϫ g for 10 min. The precipitates contained nuclei. The supernatant fraction was further centrifuged at 20,000 ϫ g for 20 min, and the resulting supernatant was used as cytoplasmic fraction. The precipitates containing nuclei were resuspended in buffer A and centrifuged at 2,000 ϫ g for 5 min. The pellet that contains almost only nuclei under phase-contrast microscopy was resuspended in CSK buffer, and the soluble nucleoplasmic proteins were extracted by centrifugation at 20,000 ϫ g for 20 min. The pellet that contained a mixture of insoluble chromatin and nuclear matrix was incubated in digestion buffer containing RNase-free DNase I and the proteinase inhibitors for 1 h at 37°C. Then 1 M ammonium sulfate was added to the sample to a final concentration of 0.25 M for 5 min on ice. After centrifugation at 2,000 ϫ g for 10 min, the supernatant was used as a chromatin fraction. The pellet was further incubated with 2 M NaCl in digestion buffer for 10 min on ice and then centrifuged. The resulting pellet was solubilized in the urea buffer (10 mM Tris-HCl, pH 8.0, 8 M urea, 100 mM sodium phosphate, and the proteinase inhibitors) and used as the nuclear matrix fraction. Protein concentration was determined by the Bradford method (Bio-Rad).
Phosphorylation Assay-Phosphorylation assay was performed as described (54) with some modifications. For the immunocomplex kinase assay, endogenous ROCK2 was immunoprecipitated with 2 g of anti-ROCK2 antibody (H-85) or control IgG from 1 mg of the cytoplasmic or nucleoplasmic fraction of HeLa cells. After washing twice in NETN buffer and twice in 1ϫ kinase buffer (50 mM Tris-HCl, pH 7.4, 5 mM MgCl 2 , 1 mM EDTA, 1 mM EGTA), the immunoprecipitates were preincubated with or without 10 M Y27632 for 10 min at room temperature. A kinase reaction was performed in the presence of 500 ng of GST-MBS-CT (C terminus of MBS) (53) as a substrate and 1 Ci of [␥-32 P]ATP (6000 Ci/mmol; Amersham Biosciences) for the indicated time at 30°C. Reaction mixtures were resolved by SDS-PAGE, dried, and exposed to x-ray film. To test kinase activity of exogenously expressed CAT-nls, 293T cells were transiently transfected with expression vectors for Myc-tagged CAT-nls or empty vector (pCMV-Mycnuc), and whole cell extracts were subjected to immunoprecipitation with anti-Myc antibody, followed by the kinase reaction as described above.
Gel Filtration Chromatography-Gel filtration of HeLa nuclear extracts was performed using Superose 6, HR 10/30 column (Amersham Biosciences) as described (58). Trichloroacetic acid protein precipitation was performed following fractionation, and the protein samples were separated by SDS-PAGE and analyzed by Western blot with anti-ROCK2 (H-85) and -p300 (N-15) antibodies by reprobing the same membrane. Molecular weight was estimated using the molecular weight marker kit for gel chromatography (Sigma) as well as reprobing the membrane with other reference antibodies, including the N-CoR (1.5-2 MDa) (59) and SRC-1 (ϳ600 kDa) (58) (not shown). The signals were quantified using Densitometer and ImageQuant version 5.2 (Amersham Biosciences). Nuclear extracts prepared by two different methods, one with CSK buffer (57) and the other described by Dignam et al. (60), were used, with essentially the same results (not shown).

RESULTS
ROCK2 Associates with p300-To isolate potential regulators of p300, we performed yeast two-hybrid screening using mouse embryonic cDNA libraries (62) and human p300 cysteine-histidine rich domain 3 (CH3) as a bait. The CH3 domain of p300/CBP has been shown to mediate many important protein-protein interactions (41,42). After screening 2 million clones, we isolated 15 clones that remained positive after the second screening. One positive clone encodes part of the coiled-coil domain (amino acids 673-813) of ROCK2. This interaction was specific, since two control bait vectors, the empty vector and an irrelevant cDNA (Ha-ras oncogene), did not show interaction with ROCK2 in the yeast two-hybrid system. The interaction between ROCK2 and p300 was further examined by GST pull-down studies by incubating in vitro-translated 35 S-labeled ROCK2 with GST-p300 (CH3 domain), which demonstrated their strong association (Fig. 1A, left, lane 3). Endogenous ROCK2 from whole cell extract also associated, albeit weakly, with GST-p300 (Fig.  1A, right, lane 3) but not with GST alone (lane 2). Consistently, when endogenous p300 was immunoprecipitated from whole cell extracts, endogenous ROCK2 was successfully detected in the p300 immunoprecipitate (Fig. 1B, left, lane 3). Conversely, the association between ROCK2 and p300 was readily detected following immunoprecipitation of ROCK2 using nuclear extracts in order to minimize potential confounding effects of ROCK2 within cytoplasm (Fig. 1B, right, lane  3). These data demonstrate that ROCK2 associates with p300 both . After washing, immunoprecipitates were subjected to Western analysis using anti-ROCK2 or p300 antibody. 10% input (whole cell extract) was loaded (lane 1). B (right), HeLa nuclear extract (NE) was used to immunoprecipitate endogenous ROCK2, followed by Western blot with anti-p300 antibody. 10% input (nuclear extract) was loaded (lane 1).
in vitro and in vivo, and the endogenous proteins associate at their native levels of expression.
Domains Of ROCK2 and p300 Required for Their Association-To determine ROCK2-interacting domains within p300, in vitro translated 35 S-labeled full-length ROCK2 was incubated with various p300 fragments, expressed as GST fusion proteins. We found that only p300 fragment 4, containing the CH3 domain, associated with ROCK2 ( Fig.  2A), consistent with our prior use of the p300 CH3 domain as the bait that isolated ROCK2 in the yeast two-hybrid screen. In a reciprocal experiment, GST-p300 (fragment 4) was incubated with various deletion mutants of ROCK2 prepared by in vitro translation (Fig. 2B). We found that both ROCK2 fragments containing the RB/PH domain (mutants 3 and 4) strongly bound p300 (lanes 3 and 4), whereas all three mutants lacking this domain (mutants 1, 2, and 5) bound p300 only marginally. Thus, although multiple domains of ROCK2 could potentially associate with p300 at various degrees, the RB/PH domain was the predominant site necessary under these conditions for ROCK2 association with p300.
Nucleus-localized ROCK-To further address nuclear localization of ROCK2, we next took an immunofluorescence approach. Although it has long been thought that ROCK is localized primarily in the cytoplasm (5, 51), recent reports described that ROCK might translocate into the nucleus, at least in response to a certain growth factor (32,33). Consistent with the known cytoplasmic localization, we found that several ROCK2 antibodies detected predominantly cytoplasmic ROCK2 (not shown). However, two ROCK antibodies, ROCK2-H-85 and ROCK1-H-85, that were generated against parts of the coiledcoil domains of ROCK2 and ROCK1, respectively (ROCK2 amino acids 775-860 and ROCK1 amino acids 755-840), detected ROCK2 and ROCK1 in both the cytoplasm and the nucleus (Fig. 3A). Two different fixation methods (4% paraformaldehyde and 100% methanol) consistently showed nuclear ROCK using these antibodies, and similar results were obtained also in diverse other cell types, including HaCaT (human keratinocyte), NMuMG (mouse mammary epithelial cell), C2C12 (myoblast), U2OS (osteoblast), and NIH3T3 (mouse fibroblast) cells (not shown). To ensure specificity of the observed ROCK signals, we employed the respective H-85 peptide blockers in immunofluorescence analysis. We found that staining by the two H-85 antibodies was specifically blocked by each respective peptide but not by the reciprocal H-85 peptide (Fig. 3A). Specificities of the two H-85 antibodies were further evaluated by Western blot analysis, using the GST-H-85 fusion proteins prepared in E. coli (Fig.  3B, left, GST-ROCK1 and GST-ROCK2) as well as the full-length ROCK proteins exogenously expressed by transfection in mammalian cells (Fig. 3B, right). By both methods, each of the two H-85 antibodies selectively detected only the respective ROCK isoform.
To determine whether nucleus-localized ROCK2 possesses phosphotransferase activity, kinase reactions were performed using ROCK2 immunoprecipitated separately from the cytoplasmic or nucleoplasmic fraction with MBS as substrate (53) (Fig. 3C). Phosphorylation of MBS by nuclear ROCK2 was comparable with that by the cytoplasmic counterpart (lanes 2 and 5). This activity of nuclear ROCK2 was also inhibited similarly to that of cytoplasmic ROCK2 by a low concentration of the established ROCK inhibitor Y27632 (10 M; lanes 3 and 6) (63). Thus, nuclear ROCK2 has similar properties to cytoplasmic ROCK2 in molecular weight, kinase activity, and sensitivity to the pharmacological inhibitor.
ROCK2 Is Present as a Large Nuclear Complex and Partially Cofractionates with p300-It has been previously reported that the rat ROCK2 homologue, ROK␣, is present in an ϳ600-kDa multimeric complex in rat brain homogenates (17). Likewise, p300 exists as large protein complexes (58,64). To study whether nucleus-localized human ROCK2 also is present in a large complex, we size-fractionated HeLa nuclear extract by gel filtration as described (58). We found that both nuclear ROCK2 and p300 formed large complexes. Although the two sets of complexes had peaks at different sizes (ϳ1.1 MDa for ROCK2 and ϳ700 kDa for p300 complexes) (Fig. 4), they partially cofractionated, as would be required if a subpopulation of nuclear ROCK2 coexists with p300 within the same complex, a prediction suggested by the observed association of endogenous ROCK2 with p300 using co-immunoprecipitation (Fig. 1).
Co-localization of Nuclear ROCK2 with p300-Given the physical association between ROCK2 and p300 in the nuclear extract as well as their partial co-fractionation on gel filtration, we next asked whether ROCK2 and p300 might colocalize within the nucleus by immunofluorescence analysis. Untreated control cells showed relatively diffuse signals for both ROCK2 and p300 across the entire area of the nucleus (Fig. 5), as expected from the presence of nuclear ROCK2 and p300 in the soluble nucleoplasmic fraction by subcellular fractionation. 3 This broad pattern of immunostaining might obscure the detection of ROCK2 and p300 in more specific locations. Therefore, we next treated cells with CSK buffer prior to fixation to remove soluble cytoplasmic and nucleoplasmic proteins and to selectively retain insoluble structures, including chromatin and nuclear matrix (57). Interestingly, CSK-treated cells revealed significantly, but not completely, overlapping signals of ROCK2 and p300 within the nucleus (Fig. 5, CSK), suggesting their potential co-localization in either chromatin or nuclear matrix or both. Further treatment with DNase I resulted in the loss of chromatin as verified by the complete disappearance of DAPI staining, leaving the lamin B-positive nuclear structures (Fig. 5, CSK ϩ DNase I) (57). It is noteworthy that after these sequential treatments of cells, the integrity of lamin B-positive nuclear structures (nuclear envelope and matrix) (57) remained well preserved (Fig. 5, Lamin B). Whereas lamin B staining was broadly detected throughout the nucleus following the treatment with CSK plus DNase I, ROCK2 and p300 were detected in discrete subregions of the nucleus, in which ROCK2 partially colocalized with p300 (Fig. 5, CSK ϩ DNaseI). These data are consistent with the idea that the association of nuclear ROCK2 and p300 might occur not only in the soluble nucleoplasm ( Fig. 1) but also in the nuclear matrix and possibly chromatin. Whole cell extracts were subjected to Western analysis with anti-ROCK2, -ROCK1, or -␤-tubulin antibody. C, nuclear ROCK2 has similar properties to cytoplasmic ROCK2 in phosphotransferase activity, sensitivity to a ROCK inhibitor, and molecular size. Cytoplasmic (CP) and nucleoplasmic (NP) fractions were prepared from HeLa cells. Each fraction was subjected to immunoprecipitation (IP) with either control IgG (lanes 1 and 4) or anti-ROCK2 antibody (lanes 2, 3, 5, and 6), followed by kinase reactions using GST-MBS as a substrate, either with or without the ROCK inhibitor (10 M Y27632). Part of the immunoprecipitates were used for Western analysis to verify comparable amounts of ROCK2 immunoprecipitated (bottom). . Nuclear ROCK2 is present in a large complex that partially cofractionates with p300 in a gel filtration analysis. HeLa nuclear extracts were subjected to Superose 6 gel filtration, and alternate fractions were subjected to Western analysis with anti-ROCK2 (H-85) or -p300 (N-15) antibody (bottom). Band intensity was determined by densitometry, and bands with maximum intensity were taken as 100% for each protein analyzed (top).

ROCK2 Phosphorylates p300 in Vitro and Promotes p300
Phosphorylation in Cells-To address the possible significance of the association between nuclear ROCK2 and p300, we studied whether p300 served as a substrate for phosphorylation by ROCK2. We found that affinitypurified p300 was rapidly phosphorylated by the ROCK2 catalytic domain (GST-CAT) in vitro (Fig. 6A). To address whether p300 phosphorylation is induced by ROCK2 in intact cells, we next expressed the constitutively active ROCK2 mutant, CAT, by transfection. However, we found that CAT was expressed predominantly in the cytoplasm (Fig.  6B). Because our data suggested that the nucleus was the site of the association between ROCK2 and p300, we generated another ROCK2 mutant, CAT-nls, by adding heterologous nuclear localization signals from SV40T at the C terminus of CAT, in order to limit ROCK2 CAT expression to the nucleus. Transfection of cells with CAT-nls successfully restricted its expression to the nucleus (Fig. 6B). CAT-nls immunoprecipitated from the transfected cells showed a phosphotransferase activity toward MBS, whereas no phosphorylation was observed with cells transfected with an empty vector (Fig. 6B, bottom). Interestingly, when CAT-nls was expressed in cultured cells, a fraction of p300 migrated more slowly in SDS-PAGE (Fig. 6C, lane 2) than cells transfected with empty vector (lane 1), as shown by Western analysis. The more slowly migrating p300 represents phosphorylated p300, because pretreatment of cell extracts with phosphatase abolished the mobility shift (Fig. 6C, lane 3). Thus, these data suggest that ROCK2 phosphorylates p300 in vitro and promotes p300 phosphorylation in cells.
ROCK2 Phosphorylation of p300 Increases Its Acetyltransferase Activity-Acetyltransferase activities of p300/CBP appear to change in parallel with their phosphorylation status in cultured cells (45,65,66). To directly test whether p300 phosphorylation might influence its acetyltransferase activity, we affinity-purified recombinant human p300 expressed by baculovirus in Sf9 insect cells and studied its histone acetyltransferase (HAT) activity before and after treatment with calf intestine alkaline phosphatase. We found that calf intestine alkaline phosphatase-treated p300 had a significantly lower HAT activity than untreated p300 (Fig. 7A). Thus, p300 HAT activity was phosphorylationdependent, although a kinase responsible for its phosphorylation in Sf9 cells is not known. Given the observed phosphorylation of p300 by ROCK2 as well as the proteins' physical association, we next studied whetherROCK2couldinfluencep300HATactivityinaphosphorylationdependent manner. Preincubation of p300 with the ROCK2 catalytic domain did not affect p300 HAT activity in the absence of a phosphate donor, ATP (Fig. 7B, lane 2). However, p300 HAT activity was signifi-FIGURE 5. Nuclear ROCK2 is partially colocalized with p300 in insoluble nuclear structures. Double immunofluorescence using rabbit polyclonal anti-ROCK2 antibody and mouse monoclonal anti-p300 antibody was performed in nontreated (top row, control), CSK-treated (middle row), or CSK plus DNase I-treated (bottom row) HeLa cells. Lamin B staining was used to examine the integrity of nuclear structures after CSK or CSK plus DNase I treatment (right six panels). The absence of DAPI staining after CSK ϩ DNase I treatment confirms the loss of chromatin by DNase I (bottom row). cantly increased by ROCK2 if ATP was provided, as indicated by increased acetylation of all histones as well as autoacetylation of p300 (lane 3). Consistent with the requirement for ATP, the ROCK2 effect depended on its intact kinase activity, since the ROCK inhibitor Y27632 was sufficient to block the ability of ROCK2 to enhance p300 HAT activity (lane 4). These results suggest that phosphorylation of p300 by ROCK2 enhances its HAT activity in vitro.
Enhanced p300-dependent Transcription by the Constitutively Active ROCK2-To address whether the physical association and phosphorylation of p300 by ROCK2 might be accompanied by a functional change of p300, we studied p300-dependent transcription in cells expressing either the constitutively active ROCK2 (CAT-nls) or its kinase-deficient counterpart (CAT-KD-nls) in the nucleus. We found that nuclear expression of CAT-nls increased the p300-dependent transcription by Gal-p300 without changing the Gal-p300 expression level, whereas CAT-KD-nls and cytoplasmic expression of CAT were inactive in this capacity (Fig. 8). These findings are consistent with the idea that nuclear ROCK2 activates p300-dependent transcription through its kinase activity.

DISCUSSION
Significant progress has been made in our understanding of cytoplasmic ROCK functions. These ROCK functions have been determined by studies that largely rely on various dominant active and negative ROCK mutants and pharmacological inhibitors, such as Y27632 (63), and more recently by siRNA (12,67) and targeted gene disruption (13,68,69). One potential caveat of these approaches, however, is that they do not necessarily take into consideration multiple cellular functions that ROCK may have at different subcellular locations. Because cytoplasmic ROCK was generally the only previously recognized species, results of these studies were thought to indicate cytoplasmic ROCK functions. Here, we demonstrate that ROCK2 is detected not just in the cytoplasm but also in the nucleus of growing cells. We show that nuclear ROCK2 associates with p300 both in vitro and in intact cells by several complementary approaches and that ROCK2 phosphorylates p300 in vitro and enhances p300 phosphorylation in cells. Remarkably, phosphorylation of p300 by ROCK2 leads to an increased acetyltransferase activity of p300 in vitro. These data demonstrate a novel function of nuclear ROCK2 involved in the regulation of p300 acetyltransferase.
Nuclear ROCK2-We have previously shown by subcellular fractionation that ROCK is present in the nucleus of growing cells at a concentration comparable with that of cytoplasmic ROCK2. 3 The present study extends these findings by further characterizing nuclear ROCK2; nuclear ROCK2 was indistinguishable from the cytoplasmic counterpart in its immunoreactivity to multiple ROCK2-specific antibodies (Fig. 3 and data not shown), molecular size in Western blot analysis, catalytic activity, and sensitivity to the Y27632 inhibitor (Fig. 3). Only a single human ROCK2 gene has been identified, on chromosome 2p24 (70), and hence, both cytoplasmic and nuclear ROCK2 are derived from this same gene. We have noted the presence of several nuclear transport signals within the ROCK2 amino acid sequences and found that at least one such nuclear export signal was functionally active in exporting ROCK2 from the nucleus (not shown). We speculate that these nuclear transport signals might play an important role in sorting ROCK2 to the nucleus or the cytoplasm. How these nuclear transport signals might be regulated to sort ROCK2 to different subcellular sites is not known, but this regulation might involve post-translational modifications of ROCK2. Alternatively, nuclear ROCK2 might represent a splice variant that lacks nuclear export signals but carries intact nuclear FIGURE 7. p300 histone acetyltransferase (HAT) activity is dependent on its phosphorylation by ROCK2 in vitro. A, p300 HAT activity is dependent on its phosphorylation status. Recombinant p300 was expressed in Sf9 cells by baculovirus and affinity-purified. p300 was either pretreated with calf intestine alkaline phosphatase (CIAP; lane 2) or left untreated (lane 1) and used for an in vitro HAT assay, using histones (H1, H2A, H2B, H3, and H4) as substrates. 14 C-labeled acetylated proteins were analyzed by autoradiography after SDS-PAGE. CBB, Coomassie Brilliant Blue. B, phosphorylation of p300 by ROCK2 enhances p300 HAT activity. p300 was preincubated with GST-ROCK2 CAT protein, either in the absence of ATP (lane 2) or in the presence of ATP (lane 3) or ATP plus Y27632 (lane 4), and then HAT assay was performed. localization signals (NLS). Further studies are clearly needed to address these issues.
The finding that only certain ROCK antibodies (i.e. H-85) detected nuclear ROCK, whereas other antibodies detected predominantly cytoplasmic ROCK in immunofluorescence, may suggest that nuclear ROCK might exist in a different conformation than the cytoplasmic ROCK, such that some ROCK epitopes (i.e. those in the H-85 regions) might be exposed only in the nuclear environment. In an analogous situation, it has been proposed that nuclear actin may assume a different conformation than cytoplasmic actin, based on distinct staining patterns of actin in the two subcellular compartments revealed by an antiactin monoclonal antibody (71). Alternatively, differential modifications or occlusion by ROCK-bound proteins depending on subcellular locations might account for differences among these epitopes.
It has been previously shown that the rat ROCK2 homologue ROK␣ may form a homotetramer with a molecular size of 600 kDa in rat brain homogenates (17). We detected an even larger nuclear ROCK2 complex of ϳ1.1 MDa, using HeLa nuclear extracts, although we additionally did observe ROCK2 in the fraction of ϳ600 kDa as a minor component (Fig. 4). Exact reasons for the different sizes of the reported rat ROK␣ complex and our ROCK2 complex are not known, but this may be ascribable to different tissues and species (rat brain versus HeLa human cervical cancer cell line) or to the fact that our ROCK2 complex is derived from nuclear extracts instead of whole brain homogenates. Given that nuclear ROCK2 has distinct functions, such as regulation of Cdc25A expression 3 and p300 (this paper), it would not be surprising if nuclear ROCK2 complex has subunits different from those of cytoplasmic ROCK2 complex, which could explain the different sizes of the two ROCK2 complexes. Identification of other subunits of the nuclear ROCK2 complex should provide further insight into the roles and regulation of nuclear ROCK2.
Candidate Regulators and Targets of Nuclear ROCK2-How may the activity of nuclear ROCK2 be regulated? We have recently found that nuclear ROCK2 up-regulates the cell cycle regulator Cdc25A to promote S-phase entry: this property of nuclear ROCK2 is inhibited by phosphorylation at specific TQ (threonine/glutamine) motifs, probably by the checkpoint kinase, ATR (ataxia telangiectasia and Rad3-related), in response to replication stress. 3 To our knowledge, this is the first demonstration of phosphorylation as the regulatory mechanism for a nuclear ROCK2 function, and it will be important in the future to determine, analogously, what inputs and modulators govern the regulation of p300 by ROCK2.
Several candidates are prompted by published studies. ROCK was originally isolated using the GTP-bound active form of RhoA as a probe (5,6,51), and GTP-bound RhoA has been shown to enhance the phosphotransferase activity of purified ROCK in vitro (5,6). Although Rho GTPase functions primarily in the cytoplasm, Rho, as well as its guanine nucleotide exchange factors Net1 and ECT2, has been detected also in the nucleus (34 -39), raising the possibility that Rho GTPase might activate nuclear ROCK2. Additionally, certain lipids, especially arachidonic acid, have been shown to activate ROCK2 in vitro, probably through binding to the inhibitory PH domain, in a manner independent of RhoA (72,73). Since arachidonic acid is thought to be present in the nucleus (74,75), it might serve as an activator of nuclear ROCK2. It has been recently shown that a cyclin-dependent kinase inhibitor, p21, directly binds and inhibits ROCK in the cytoplasm (76,77), which raises a question of whether p21 is a regulator of ROCK2 inside the nucleus as well. Since our studies revealed a specific interaction of p300 with the C-termi-nal autoinhibitory RB/PH domain, this interaction might represent an additional regulatory mechanism for nuclear ROCK2 by this acetyltransferase.
Our data suggest that p300 acetyltransferase is a candidate substrate for phosphorylation by nuclear ROCK2. What are other possible substrates for nuclear ROCK2? Cytoplasmic ROCK regulates actin cytoskeleton reorganization, through phosphorylation of LIM kinase and myosin phosphatase. Interestingly, accumulating data demonstrate that LIM kinase, cofilin, and actin, as well as myosin phosphatase and myosin, all are localized also in the nucleus (78 -80). Nuclear actin may play a critical role in gene transcription by all three classes of RNA polymerases and chromatin remodeling (81,82). It has been proposed that the nuclear actomyosin motor might be involved in elongation of RNA transcripts (83) and that nuclear actin may exist not only as a monomer but also polymers (81,82). These observations invite speculation that ROCK might play a role in the control of nuclear actomyosin by phosphorylating key substrates, such as LIM kinase and myosin phosphatase, within the nucleus.
Modulation of p300 Acetyltransferase Activity by Nuclear ROCK2-It has been proposed that acetyltransferase activities of p300/CBP are influenced by their phosphorylation status; the HAT activity of CBP, a p300 paralogue, was enhanced by its phosphorylation by Cdk2-cyclin E complex in vitro (45), although phosphorylation residues responsible for the increased HAT activity of CBP still remain to be defined. Phosphorylation of p300 at Ser 89 by protein kinase C␦ appears to inhibit its HAT activity (49), whereas p300 Ser 1834 phosphorylation by Akt can activate its HAT activity (66). In addition to the HAT activity, phosphorylation can also promote p300 degradation (84) or enhance p300/CBP transcriptional activity (41,50,85). Our data show that nuclear ROCK2 associates with p300 at its CH3 domain adjacent to the HAT domain and activates p300 HAT activity in vitro, contingent on phosphorylation. The increased p300 HAT activity by ROCK2 was observed toward all histones as well as p300 itself (autoacetylation) in vitro. However, histones might not be the physiological targets for ROCK2-dependent acetylation by p300, since we have not detected a significant change in histone acetylation in cells expressing constitutively active ROCK2 in the nucleus. Many diverse sequence-specific transcription factors have been identified that are acetylated by p300 (42), and further studies are needed to identify the in vivo acetylation substrates for p300 in the ROCK2-p300 pathway. The enhanced p300-dependent transcription by ROCK2 (Fig. 8) suggests in vivo significance of the observed physical and functional interactions between p300 and ROCK2. ROCK2 might activate the p300dependent transcription by augmenting the HAT activity (86) or the functions of p300 transcription domains (43).
In summary, our data demonstrate that ROCK2 is localized in the nucleus, associates with and phosphorylates p300, and increases its HAT activity in vitro. Besides gene transcription, p300 has been implicated in a variety of other cellular activities, including the cell cycle, DNA repair, and apoptosis, and the importance of its acetyltransferase function in these activities is documented amply (42,87). p300 behaves as a tumor suppressor in certain circumstances, and somatic mutations of the p300 gene have been observed in several types of cancers (42). Thus, understanding the regulatory mechanism of p300 acetyltransferase activity is of critical importance, and our results provide a new insight by identifying nuclear ROCK2 as a novel candidate regulator of p300 acetyltransferase. Physiological cues that regulate this nuclear ROCK2 activity as well as in vivo p300 acetylation targets in this pathway remain to be determined. ROCK inhibitors have been extensively used in both basic research and clinical settings. Given that nuclear ROCK2 can also be a target for these inhibitors, it is imperative to be aware of their effects not only toward cytoplasmic ROCK but also nuclear ROCK and to further deepen our understanding of the functions and regulation of nuclear ROCK.