Expression of the human myotonic dystrophy kinase-related Cdc42-binding kinase gamma is regulated by promoter DNA methylation and Sp1 binding.

Myotonic dystrophy kinase-related Cdc42 binding kinases (MRCKs) are family members most related to the myotonic dystrophy kinase (DMPK), RhoA-binding kinase (ROK), and citron kinase. Two highly conserved members, MRCKalpha and -beta, have been previously identified and characterized. We now describe a novel isoform, MRCKgamma, which is functionally and structurally related to members of this kinase family. We show these kinases to have marked similarities in their genomic organization, substrate phosphorylation, and catalytic autoinhibition. Unlike MRCKalpha and -beta, which are expressed ubiquitously, MRCKgamma mRNA was only expressed in heart and skeletal muscle. In cultured cells, MRCKgamma showed differential expression with high levels of expression only in certain cell lines. DNA analysis showed that lack of expression is correlated with promoter DNA methylation. We have mapped the methylation sites in the MRCKgamma promoter. Significantly, agents that suppressed DNA methylation caused increases in the expression of the kinase in low-expressing cells, further supporting the notion that promoter DNA methylation plays an important role in the expression of MRCKgamma. Analysis of the MRCKgamma promoter has also revealed two proximal Sp1 sites that are essential for transcriptional activity. We conclude that both promoter DNA methylation and Sp1 binding are important regulators for MRCKgamma expression.

The Rho GTPases regulate cell morphology, cell growth, and cell polarity (1,2). Numerous downstream effectors have been reported in the last decade; they comprise both kinases and non-kinases. Rho kinase ROK 1 interacts specifically with the GTP-bound form of RhoA and organize actin bundling in cultured cells, whereas p21-activated kinases (PAKs) are Rac1 and Cdc42 binders signaling for actin disassembly and focal adhesion dissolution, through their complexes with PAK-interacting exchange factor PIX, G-protein receptor kinase-interacting protein GIT, and paxillin (3,4). The non-kinase Rac1/ Cdc42-binding Wiskott Aldrich's syndrome protein WASP and related n-WASP are also downstream targets and are essential for actin polymerization, through their interaction with Arp2/3 (5). A well conserved motif for the interaction of Rac1/Cdc42 (Cdc/Rac1-interactive binding, CRIB motif) has been identified when comparisons were made with a number of the Rac1/ Cdc42-binding proteins such as PAK, ACK, and WASP (6,7). Members of the myotonic dystrophy kinase-related Cdc42binding kinase family also possess the CRIB motif, which has preferential binding to Cdc42 (8). However, their catalytic domain is closely related to that of Rho kinase, which specifically binds GTP-RhoA. Although the exact function of these kinases remains to be characterized, the catalytic domain homology suggests a similar function to Rho kinases ROKs, which are known to enhance the phosphorylation status of myosin light chain-2 (MLC-2), directly by phosphorylation or indirectly by phosphorylation and inactivation of myosin-targeting subunits of myosin phosphatase (9 -11). MRCKs, resembling ROKs, can phosphorylate and activate LIM kinases, which phosphorylate and inactivate cofilin, thereby facilitating actin polymerization events (12). At the cellular level, MRCK has been shown to be involved in regulating cell morphology by enhancing Cdc42induced membrane extensions (8), and the dominant negative form of MRCK␣ can block nerve growth factor-induced neurite outgrowth in PC12 cells (13). The Drosophila autologue Genghis Khan (GEK) has also been implicated in actin polymerization events during development (14).
Two genes encoding different isoforms of MRCK have been reported in mammals (8). Both isoforms can exist as tetrameric forms through intermolecular interaction of their extended coiled-coil domains (15). A region in the distal coiled-coil region (CC2/3) was found to be essential for kinase inhibition, but the exact location has not been mapped previously. Activation of the kinase was observed upon binding of phorbol ester to the neighboring cysteine-rich domain, presumably by releasing the constraint of the inhibitory effect on the catalytic activity (15). Similar oligomeric structures with distinctive features have also been reported for the related Rho kinases ROKs (16,17) and the myotonic dystrophy kinase (DMPK) (18). Both MRCK␣ and MRCK␤ are ubiquitously expressed in various mammalian tissues and are abundantly expressed in all cell lines studied (18,19). Interestingly, MRCK␣ is present as multiple species through differential splicing (20), mainly at an internal variable splice site, which is located between the inhibitory region and the phorbol-binding cysteine-rich domain. Whether or not these sequence diversities play any roles in MRCK function remains to be determined.
The characterization of yet another member, MRCK␥, has not been reported. We now describe the biochemical and functional characterization of MRCK␥. The expression of this kinase show marked variation in some cell lines, and this is dependent on the methylation status of its promoter DNA as well as on Sp1 binding.

MATERIALS AND METHODS
Cell Culture, Transfection, and Cell Staining-HeLa, Hct116, and MRC5 cells were cultured in minimum Eagle's medium, MCF7 and MKN28 cells were grown in RPMI medium, and COS-7 cells were maintained in Dulbecco's modified Eagle's medium. All media were supplemented with 10% fetal bovine serum, and cell cultures were maintained in humidified 5% CO 2 . Subconfluent cells plated on culture dishes for 24 h were transfected with respective DNA constructs using Lipo-fectAMINE (Invitrogen) according to the recommended protocol. For immunostaining experiments, transfected cells were fixed with 4% paraformaldehyde and stained with anti-FLAG antibody (M2; Sigma).
Immunoprecipitation and Kinase Assays-COS-7 cells expressing HA⅐MRCK␣⅐CAT or HA⅐MRCK␥⅐CAT alone or co-expressed with GST⅐ MRCK␥⅐KIM were lysed in lysis buffer containing 25 mM HEPES, pH 7.3, 150 mM NaCl, 1.5 mM MgCl 2 , 0.2 mM EDTA, 20 mM ␤-glycerol phosphate, 1 mM sodium vanadate, 5% glycerol, 0.5% Triton X-100, and 1ϫ complete protease inhibitor mixture (Roche Applied Science), and immunoprecipitations were performed essentially as described (15) using either gluthathione-Sepharose (Amersham Biosciences) or anti-HA antibody (12CA5; Roche Applied Science). Kinase assays with 10 g of substrates ((histone H1, myelin basic protein, GST fusion of myosin light chain-2 (MLC-2), or GST fusion of the phosphorylation inhibitory motif of myosin phosphatase MBS85 (GST⅐PIM)) and 0.1 g of the immunoprecipitated kinase were carried out as described previously (15). The reactions were stopped by the addition of sample buffer, and the proteins were resolved by 11% SDS-PAGE, dried, and autoradiographed.
Western Blot Analysis-Protein expression was detected by Western immnunoblotting. Cells were harvested in lysis buffer, and 100 g of proteins were separated by 7.5% SDS-PAGE. MRCK␣, -␤, and -␥ proteins were detected by specific antibodies raised against MRCK␣, -␤, and -␥, respectively.
In Vitro GTPase Binding Assay-GST fusion proteins (50 ng) of MRCK␣ and MRCK␥ containing the CRIB domain were resolved by SDS-PAGE and transblotted onto a polyvinylidene difluoride membrane. Immobilized proteins were first renatured for 3-4 h at 4°C in renaturing buffer (phosphate-buffered saline containing 1% bovine serum albumin, 0.1% Triton X-100, 0.5 mM MgCl 2 and 5 mM dithiothreitol) and subjected to probing with solution containing either [␥-32 P]GTP-labeled GST⅐Cdc42 or TC10 as described previously (21). Probed filters were washed three times and autoradiographed.
Northern Blot Analysis-Blots containing mRNA from various human tissues were obtained from Clontech and were hybridized with a C-terminal [ 32 P]-labeled 3-kb NheI/NotI fragment from full-length MRCK␥ (nucleotides 1610 -4661).
Cell Treatment with 5-Aza-CdR and Trichostatin A and Reverse Transcriptase-PCR-HeLa cells were plated onto 90-mm 2 dishes 18 -24 h prior to experiments. 5-Aza-2Ј-deoxycytidine (5-Aza-CdR; Sigma) was freshly added onto cells every 24 h at a final concentration of 1 M for 48 h. 100 ng/ml trichostatin A (Sigma) was added at the last 12 h of the cell treatment with 5-Aza-CdR. Total RNA from treated or untreated cells with drugs was obtained using the Qiagen RNeasy® RNA isolation kit. cDNA was reverse-transcribed from the total RNA using M-MuLV reverse transcriptase (New England Biolabs). PCR was then performed using Hotstar Taq and the CRIB domain primers. The PCR conditions were as follows: 95°C for 15 min; 40 cycles of 94°C for 30 s, 50°C for 30 s, and 72°C for 30 s; and a final extension of 72°C for 5 min. Primers for glyceraldehyde-3-phosphate dehydrogenase (5Ј-TCC ATC ACC ATC TTC CAG-3Ј (forward) and 5Ј-ATG AGT CCT TCC ACG ATA CC-3Ј (reverse)) were used for similar PCR as control. For amplification of the internal splice site, primers 5Ј-TAG CAG GAA GAC CTT CTG TAG-3Ј/ 5Ј-TGG AGC TGC AGT CAG CGC T-3Ј were used. The PCR products were subcloned into pDrive vector and sequenced.
Determination of Transcriptional Initiation Site-RT-PCR was performed using Hct116 poly(A) ϩ RNA to determine the transcriptional initiation site. Several primers close to the putative initiation site of MRCK␥ (including primer 1, 5Ј-GAA GGT GAC AGC GGG GAGG-3Ј; primer 2, CCA GGT GAG GGC GCG CTG-3Ј; primer 3, 5Ј-CGG CGC CCA GGT GAG GGC-3Ј; and primer 4, 5Ј-GGC TCC CAT TGG CCG GCG-3Ј) were synthesized to pair up with the reverse primer 5Ј-CTG CTG AGC TCG TGG TGC A-3Ј. Various annealing temperatures for each primer pair were used for optimizing reaction conditions using genomic DNA as template, and the right-sized products from RT-PCR were subcloned into pDrive vector for DNA sequencing.
Luciferase Reporter Assay-Hct116 cells were transfected with various MRCK␥ luciferase reporter gene constructs, and cells were harvested for luciferase assays using the single luciferase assay system (Promega). pXJ40-HA-Sp1 was also co-transfected with pGL3-2 to determine the effect of Sp1 on the promoter activity. A cytomegalovirus promoter-␤-galactosidase reporter (pCMV-␤-galactosidase) construct was also included to monitor transfection efficiency.
Preparation of Whole Cell Extracts and Electromobility Shift Assays (EMSA)-Preparation of whole cell extracts from HeLa, MKN28, and COS-7 transfected cells was performed according to Manley et al. (22), and EMSA was performed essentially as described previously (23). The oligonucleotide used for EMSA was the consensus Sp1 binding site (underlined) in MRCK␥ promoter (5Ј-CCC CGG CCC GCC CCC GCA CTT-3Ј; see Fig. 5B). The oligonucleotide was 5Ј-end-labeled with [␥-32 P]ATP and annealed with its complementary strand (5Ј-AAG TGC GGG GGC GGG CCG GGG-3Ј). A DNA fragment of 205 bp (Ϫ263 to Ϫ59) with sequences spanning the tandem Sp1 binding sites was also used for EMSA. The 32 P-radiolabeled primer pair (5Ј-CAG GTA CCG CGC CTC CTG GGT CA-3Ј/5Ј-GCT AGA GGT GAA GGT GGG-3Ј) was used for PCR of the 205-bp DNA fragment. Unlabeled competitors were added in 100-fold excess to the reaction mix prior to the addition of radiolabeled probe. For supershift assays, 2 g of the polyclonal Sp1 antibody (Santa Cruz Biotechnology) were added to the binding reaction prior to the addition of radiolabeled probe. The DNA-protein complex was then size-fractionated by using 5% PAGE in 0.5ϫ Tris-borate EDTA buffer at 150 V at 4°C. The gel was dried and autoradiographed.
Preparation of Genomic DNA and McrBC Digestion-The genomic DNA from various cell lines was prepared using a DNeasy® tissue kit (Qiagen) according to the manufacturer's instructions. 250 ng of genomic DNA were cleaved with McrBC (New England Biolabs) in a final reaction volume of 25 l at 37°C for 1 h. The heat-inactivated cleavage mixture (2.5 l) was used for PCR using primers 5Ј-GTC TGA AGC CAC CAA TAT GTC-3Ј and 5Ј-CTG CTC GGC TAC AGT CTG G-3Ј. Bisulphite Modification and Sequencing-For mapping of methylated cytosine residues, genomic DNA samples from Hct116 and HeLa cells were modified by bisulphite reaction using the CpGenome TM DNA modification kit (Intergen Inc.), and the final products were amplified with primers 5Ј-GGG TTT AAT TTA AGA GTT ATT TTG-3Ј (forward)/ 5Ј-CTA TCA CCT TCC TCC-3Ј (reverse), which were specific for the modified MRCK␥ sequence. PCR was performed with the BD Advantage TM 2 system (BD Biosciences) and started at 95°C for 1 min followed by 5 cycles of 95°C for 30 s, 65°C for 30 s, and 68°C for 1 min, with a decrease of 2°C in annealing temperature per cycle. After another 35 cycles of 95°C for 30 s, 55°C for 30 s, and 68°C for 1 min and a final extension of 68°C for 3 min, the PCR products were cloned into pDrive vector, and 20 separate clones for each cell lines were sequenced for analysis.

MRCK␥ Is a Novel Member of the MRCK Serine/Threonine
Kinase Family-We have previously reported two members, MRCK␣ and MRCK␤, in this class of Rho GTPase-binding serine/threonine kinases. Sequences of a related kinase have been reported in the databases (e.g. accession number XM_290516), but hitherto, no reports of its biochemical and functional characterization have been reported. We have now isolated and characterized the human cDNA of this novel kinase. The sequence codes for a new member of the MRCK family, which we have termed MRCK␥ (Fig. 1A). The organization of domains in MRCK␥ is similar to the other MRCKs, and comparisons with MRCK␣ revealed the following sequence identity (in percent) in the various domains: kinase (73%), cysteine-rich (54%), pleckstrin homology (47%), Citron homology (42%), and CRIB (53%). The coiled-coil domain in MRCK␥ is much shorter (mainly in the proximal coiled-coil 1 (CC1) region) and more diverse (29%). Within the coiled-coil domain, there is a highly conserved motif (66% identical to MRCK␣), termed the KIM because of its unique property of interacting with, and inhibiting activity of, the kinase (Figs. 1, A and B, and 2A) (see also Ref. 15). MRCK␥ has a molecular mass of about 160 kDa, when compared with 180 kDa for the other known family members, the main difference being due to the shorter CC1 domain. Phylogenetic analysis of the MRCK family showing its relationship to the mammalian and invertebrate counterparts is presented in Fig. 1C. The gene is located at chromosome 11q13. Analysis of the genomic sequence reveals the coding sequence to reside within 36 exons (Fig. 1, A and B) with an organization that is markedly similar to that of MRCK␣ (19) and MRCK␤ (Ref. 20 and data not shown). However, the overall size of the gene is only about 30 kb, which is far more compact than the 250 kb reported for MRCK␣ and the 130 kb reported for MRCK␤.
Biochemical Characterization, Expression, and Cellular Localization of MRCK␥-The marked homology of the kinase domain of MRCK␥ with those of MRCK␣/␤ and Rho kinases ROK inferred similar substrate specificity. A GST kinase domain fusion protein of MRCK␥ was tested for this. Like MRCK␣, MRCK␥ kinase phosphorylated the known substrates, MLC-2 and a fragment from the myosin binding subunit of myosin phosphatase MBS85 (GST⅐PIM; Fig. 2A) but not histone H1 nor myelin basic protein. The conserved KIM motif of MRCK␥ (amino acid residues 677-765 in CC2/3) also bound the kinase domain of MRCK␣ and MRCK␥ and inhibited their catalytic activity (Fig. 2A). These results are consistent with MRCK␥ sharing similar substrate specificity and conserved auto-inhibitory mechanism with the other MRCK kinases.
The CRIB domain responsible for Cdc42/Rac1 interaction in MRCK␥ is far less conserved when compared with other MRCK counterparts (Fig. 2B). We therefore tested the binding of GST⅐MRCK␥⅐CRIB to RhoA, Rac1, Cdc42 as well as the related TC10 GTPase. Although this CRIB domain can bind Cdc42, it binds more strongly to TC10 (Fig. 2B). It binds Rac1 weakly but not RhoA (data not shown). It seems that MRCK␥ differs from the other MRCK isoforms in that it may interact preferentially with Rho GTPases other than Cdc42 and Rac1. Apart from TC10, the identity of any other GTPase binding and their physiological roles remain to be determined.
Northern blot analysis of MRCK␥ revealed that a 6-kb message was expressed in human heart and skeletal muscle (Fig.  2C). By immunological analysis, the protein was found to be also highly expressed in a number of cell lines, including MKN28 cells (Fig. 3A). RT-PCR of MKN28 cells showed a differential splicing event at the internal variable region within exons 21-23 (Fig. 1B), with the major product containing exons 21/22/23 and minor products containing alternative exons 21/ 22a/23 or 22/23 alone (Fig. 2D). Extensive splicing events in this region have been documented for MRCK␣ (20), and MRCK␥ has adopted a similar but simpler processing. However, other splicing events occurring at the CRIB domain and C-terminal of MRCK␣ were not observed with MRCK␥. When expressed in HeLa cells, FLAG-tagged MRCK␥ was mainly cytoplasmic with a higher density at the leading edges (Fig. 2E).
Distinctive Expression Pattern of MRCK␥ in Various Cell Lines Determined by Promoter Activity-We used specific antibodies to the different isoforms of MRCKs to evaluate their relative expression in the soluble fractions derived from a variety of human and mammalian cell lines. Although the protein expression of MRCK␣ and MRCK␤ was ubiquitous, that of MRCK␥ was more restricted. High expression was detected in MKN28, HCT116, and MCF7 cells but not in HeLa, MRC5, and COS-7 cells in the panel of mammalian cells tested (Fig. 3A). RT-PCR analysis of MRCK␥ mRNA expression (Fig. 3B) showed this to be correlated with protein expression, high and poor transcription being responsible for the high and very low protein expression of MRCK␥ in MKN28 and HeLa cells, respectively. We then investigated whether promoter DNA methylation, a key regulator in transcriptional control, contributed to the variation in transcription observed in these cells. Genomic DNA isolated from the low-expressing HeLa and MRC5 cells (but not from high expressers MKN28 and Hct116 cells) was extremely sensitive to restriction enzyme cleavage by the methylation-requiring nuclease McrBC at the putative promoter region of MRCK␥ and resulted in failure to obtain an intact PCR product within this region, indicating hypermethylation in the presumed promoter of these cells (Fig. 3C) (see also Refs. 24 and 25). From the pattern of expression and the characteristics of the putative promoter, we conclude that MRCK␥ expression is regulated by transcription and is likely to be under the control of promoter DNA methylation.
Regulation of Expression of MRCK␥ by DNA Methylation-We first attempted to determine the transcriptional initiation site for MRCK␥. The use of either conventional primer extension assay of mRNA from MKN28 cells or PCR from human cap-site cDNA libraries failed to produce recognizable products. This failure to detect the start site may possibly stem from low mRNA expression in the available human tissues or as a result of the very GC-rich region around this site (Fig. 4B). We therefore used PCR primers to tentatively identify this site. As the primer 5Ј-CCA GGT GAG GGC GCG CTG-3Ј (Fig. 4A, primer 2) gives more intense PCR product than 5Ј-CGG CGC CCA GGT GAG GGC-3Ј (primer 3 with common sequence underlined) from cDNA derived from poly(A) ϩ -RNA of MKN28 cells, we infer that the former primer encompasses the start site. This site is located 171 bp upstream of the ATG translational start codon (Fig. 4B). As is typical of many housekeeping genes, the DNA sequence is GC-rich and lacks a TATA box consensus sequence. An inverted CAAT sequence was located Kinase assays with histone H1 (H1), myelin basic protein (MBP), GST⅐myosin light chain (GST⅐MLC-2), and GST⅐PIM were used as substrates, which were marked with asterisks. b, the inhibitory effects of MRCK␥⅐KIM on the catalytic activities of MRCK␣ and MRCK␥. HA⅐MRCK␣⅐CAT and HA⅐MRCK␥⅐CAT constructs were expressed alone (lanes 1 and 3) or co-expressed with GST⅐MRCK␥⅐KIM construct (lanes 2 and 4). Immunoprecipitations were carried out using anti-HA antibody (lanes 1 and 3) or glutathione-Sepharose beads (lanes 2 and 4), and the immunoprecipitates recovered were assayed for kinase activity using GST⅐MLC-2 as substrate. Blotted filters were also immunostained with anti-HA and anti-GST antibodies after autoradiography. c, KIM motif alignment. KIM motifs of various MRCKs from human and Drosophila were aligned with the Clustal method (DNASTAR). Conserved residues are boxed in black, and the numbers indicate the positions of residues. B, sequence alignment of the CRIB sequences of human hMRCK␣, hMRCK␤, and hMRCK␥, Fugu fMRCK␣, Drosophila GEK, and Caenorhabditis elegans ceMRCK was performed as described in the legend for Fig. 1 at position Ϫ45. Multiple GC-rich Sp1 binding sites were found in the 1-kb region of the promoter; in particular, a tandem sequence GCCCGCCCCCGC with consensus to Sp1/EGR1 binding sites is located at positions Ϫ121 and Ϫ89. These sites were confirmed to bind Sp1 in a gel shift assay (Fig. 5B). As promoter DNA methylation is expected to be within the 1-kb region of the promoter, genomic DNA from HeLa cells and MKN28 cells (low-and high-expressing cells, respectively) was used in a bisulphite conversion assay to detect methylation sites. Subclones of the PCR products of this converted DNA showed heavy methylation in the region of Ϫ126 to Ϫ555 in the promoter of HeLa (Fig. 4, B and C) but not MKN28 cells (data not shown). As there are no more CG dinucleotide residues beyond position Ϫ555 nor any detectable CpG in the 3Ј GC-rich region from ϩ89 to Ϫ126, these hypermethylation sites may represent the only possible sites in the MRCK␥ promoter. Significantly, when HeLa cells were treated with agents that inhibit DNA methylation, a concomitant increase in MRCK␥ expression was observed (Fig. 4D). This provides further evidence that promoter DNA methylation is essential for regulating the expression of MRCK␥.
The Proximal Sp1 Binding Sites in MRCK␥ Promoter Are Essential for Transcriptional Activity-To test whether the putative promoter region has any transcriptional activity, a luciferase vector pGL3 with 997 bp around the promoter region of MRCK␥ (from Ϫ844 to ϩ152) and a series of deletion mutants were used to locate important regions for gene transcription. Deletion of 5Ј 386 bp (from Ϫ844 to Ϫ357; Fig. 5A, pGL3-1 to pGL3-4) caused a 20 -55% decrease in transcriptional activity, suggesting a putative positive regulatory site(s) within this region. However, no matches to any known transcription factor binding sites are detectable using the BIMAS Proscan program. Further deletion from Ϫ357 to Ϫ60 (Fig. 5A, pGL3-5) resulted in almost complete loss of transcriptional activity. A number of Sp1 binding sites, especially a tandem sequence with Sp1/EGR1 consensus, were present between the CpG island and CAAT box, suggesting that these may be the important positive regulators for the MRCK␥ promoter. Furthermore, truncation of the GC-rich 3Ј end sequence (Fig. 5A, pGL3-1⌬3Ј) resulted in significant increases in transcription activity, indicating that this GC-rich region may form secondary structures that negatively regulate transcription. However, deletion mutants of the putative inverted CAAT box (Fig. 5A, pGL3-2⌬BssHII and pGL3-3⌬BssHII) did not show any significant effects, and the identity and importance of this CAAT sequence remains to be investigated.
To determine whether the putative Sp1 binding sites are capable of interacting with their cognate binding partner, a 205-bp fragment (nucleotides Ϫ264 to Ϫ60) encompassing the multiple Sp1 binding sites was analyzed by EMSA. This revealed a pattern of major DNA-binding proteins resembling that obtained with the control lysate from COS-7 cells overexpressing Sp1 (Fig. 5B). To confirm the identity of the binding proteins, a specific oligonucleotide corresponding to one of the tandem Sp1-binding repeat sequences was used for the gel shift assay. Again, major binding resembling that of control Sp1 lysate was detectable in both MKN28 and HeLa cell lysates (Fig. 5B). In addition, both major binding proteins corresponding to Sp1 could be supershifted in the presence of Sp1 antibody; this is a further confirmation that Sp1 is the major DNA-binding protein to this region of the promoter. Moreover, overexpressing Sp1 further increased the promoter activity of MRCK␥ (Fig. 5C), strongly supporting the notion that this transcription factor has a positive regulatory role for the expression of MRCK␥.

DISCUSSION
Here we describe a novel human kinase, MRCK␥ kinase, that shows a marked similarity to the related MRCK␣ and MRCK␤. This includes the number and the arrangement of the functional domains, the genomic structure, and splicing events. The internal splice sites in the variable region of MRCK are also conserved in MRCK␥, but the splicing events in MRCK␥ are far less extensive and generate fewer splice products. The MRCK␥ gene, however, does not contain tandem CRIB domains as reported for MRCK␣. The kinase domain of MRCK␥ shares 73% identity with that of MRCK␣ and MRCK␤ and has similar substrate specificity. All members of the MRCK family, resembling ROK, are capable of phosphorylating both MLC-2 and myosin phosphatase regulatory subunit (10 -11), unlike Citron kinase, which acts on MLC-2 but not myosin phosphatase (26). Furthermore, the coiled-coil domains of MRCKs that are less conserved (29%) contain within their CC2/3 regions a highly homologous sequence that is known to exhibit kinase auto-inhibition (15). We show this conserved sequence in MRCK␥, the KIM, to be able to directly interact with and inhibit the MRCK kinase domain. It appears that MRCK␥ and the other MRCK members utilize a similar mechanism in regulating its catalytic activity.
The CRIB motif was originally characterized as a Cdc42/Racinteractive motif (6, 7). The CRIB motif of ACK, WASP and, to a certain extent, MRCK␣ and MRCK␤ has been shown to interact specifically with Cdc42. The CRIB motif of MRCK␥ binds only weakly to Cdc42 and even more weakly to Rac1, suggesting that it may interact specifically with some other Rho GTPases. Indeed, we have demonstrated here that it has greater binding to TC10, another highly related Rho GTPase. The biological significance of this interaction and the possible interaction with some other yet unidentified GTPases require further investigations.
As deduced from phylogenetic analysis of the CRIB domain from various MRCKs (27), MRCK␥ has been suggested to be the older of the mammalian MRCK kinases. MRCK␥ shows an overall divergence from the other members, with marked differences in the coiled-coil domain as well as moderate differences in other regions. This divergence and its more compact genomic makeup lend support to the hypothesis that MRCK␥ has an earlier evolutionary history than its mammalian counterparts.
MRCK␥ expression is restricted, being detectable in heart and skeletal muscle but not other human tissues tested, unlike the other MRCKs, the expression of which is ubiquitous. The expression pattern of MRCK␥ in the various cultured cells is also peculiar. Whereas high expression was found in some cells such as MKN28, MCF7, and HCT116 cells, only very low expression was detectable in other cells such as HeLa, MRC-5, and COS-7 cells. Most significantly, this pattern of expression can be correlated with the sensitivity of the cellular genomic DNA to cleavage by the methylation-sensitive nuclease McrBC in the putative promoter region of MRCK␥. This is consistent with promoter DNA methylation playing a role in the regulation of MRCK␥ expression. This notion finds further support from the subsequent mapping and identification of a CpG island in the MRCK␥ promoter. Moreover, MRCK␥ expression was increased in HeLa cells when these were treated with agents that interfere with DNA methylation. The 5Ј-untranslated region is extremely GC-rich (89%) and can also impose a negative regulation as its deletion resulted in significant in-creases in promoter activity. This unique feature of being extremely GC-rich may also explain the failure to identify the transcriptional initiation site by conventional methods such as primer extension as it is possible that the secondary structure formed in this region may perturb the extension reaction, which was optimally carried out at lower temperatures.
Analysis of the MRCK␥ promoter has also revealed the importance of the multiple Sp1 binding sites. Sp1 is a general transcriptional activator in many gene promoters, although it has also been amply documented to be involved in gene repression through its expression, protein modifications, and interaction with other transcriptional components (28 -30). Here we report a positive regulatory role of Sp1 binding in the activation of the MRCK␥ promoter. Interestingly, these multiple Sp1 sites are in close proximity to the CpG island, and it is possible that the Sp1 binding may well be influenced by the neighboring methylation event. Such a prospect has been documented for the human Leukosialin gene, in which methylation of the pro- FIG. 5. Reporter gene analysis of MRCK␥ proximal promoter and identification of binding proteins. A, luciferase (Luc) reporter constructs containing the MRCK␥ promoter or 5Ј-deleted constructs were co-transfected with pCMV-␤galactosidase into Hct116 cells. Luciferase and ␤-galactosidase assays were performed on cell extracts after 24 h transfection. Left panel, a schematic diagram of the various pGL3-MRCK␥ promoter constructs. Right panel, luciferase activities in relative light units (RLU) were normalized for ␤-galactosidase activity and represent the means and standard errors of at least three different experiments. Luciferase activities are also expressed as a fraction of pGL3-2 activity (taken as 1) for comparison. B, EMSA analysis was performed with radiolabeled oligonucleotide probes derived from the putative Sp1 binding site from MRCK␥ promoter (right panel) and a radiolabeled 205-bp MRCK␥ DNA fragment spanning the putative Sp1 binding sites (left panel). Whole cell extracts from MKN28 cells, HeLa cells, and COS-7 cells transfected with pXJ40-HA-Sp1 were used. Competition was performed in the presence (ϩ) and absence (Ϫ) of a 100-fold excess of unlabeled competitors, and 2 g of the Sp1 antibody were used for the supershift assay. The arrows show the supershifts and the major binding of Sp1. C, Hct116 cells were transfected with pGL3 basic vector/pCMV-␤-galactosidase (␤-Gal) (control), pGL3-2 promoter/pCMV-␤-galactosidase construct or triple-transfected with pXJ40-Sp1. At 24 h after transfection, cells were lysed for luciferase assays. Luciferase activities were normalized for ␤-galactosidase activity and represent the means of three independent experiments. The asterisks indicate a significant increase in luciferase activity with a p value Ͻ0.01 (**). Error bars indicate the standard error of measurements. moter led to the binding of methyl-CpG-binding protein MeCP2 that represses Sp1 activation (31). In the p21 cip1 promoter, methylation of the adjacent CpG sites affects Sp1/Sp3 binding and promoter activity (32). Perhaps in the MRCK␥ promoter, the neighboring methylation event may result in recruitment of other transcriptional repressors such as MeCP2 that can affect its Sp1 binding and subsequent transactivation.
MRCK-related kinases are involved in cytoskeletal reorganization. The mammalian MRCKs may play a role in filopodia formation and motility (8), and the Drosophila GEK counterpart has been shown to participate in actin polymerization events in development (9). Like other mammalian MRCKs, MRCK␥ is a cytosolic kinase enriched in the leading edges of cultured cells. However, its expression is restricted, being detectable in heart and skeletal muscle but not other human tissues. Our studies on its promoter have shown transcription of MRCK␥ to be subject to both negative and positive regulation. Given the role of the MRCK family in actin-myosin events underlying crucial cellular events and its presence only in muscle-containing tissues, MRCK␥ certainly merits further investigations, particularly in relation to the other mammalian MRCKs. It may well be that deliberate changes in the transcriptional activity of MRCK␥ at some stage play a role in signaling for dynamic cellular events such as cell polarity and cell migration (1,2).