Identification and Characterization of Zipper-interacting Protein Kinase as the Unique Vascular Smooth Muscle Myosin Phosphatase-associated Kinase*

Excitation-contraction coupling in smooth muscle involves activation of myosin light chain (MLC) phosphorylation, which increases activity of the myosin actin-activated ATPase, resulting in contraction. Phosphorylation of MLC phosphatase (SMPP-1M) by Rho-associated kinase or endogenous SMPP-1M-associated kinase inhibits SMPP-1M, enhancing MLC phosphorylation and contraction. However, the precise identity of SMPP-1M-associated kinase remains unclear. Biochemical evidence strongly supports the idea that SMPP-1M-associated kinase is related to the human serine/ threonine leucine zipper-interacting protein kinase (hZIPK), which is important in cell apoptosis, and the SMPP-1M-associated kinase has therefore been called ZIP-like kinase (MacDonald, J. A., Borman, M. A., Murani, A., Somlyo, A. V., Hartshorne, D. J., and Haystead, T. A. J. (2001) Proc. Natl. Acad. Sci. U. S. A. 98, Whether the vascular smooth muscle SMPP-1M-associated kinase is a truncated version of hZIPK, native hZIPK, or a unique homologue of hZIPK is un-clear. Here we show that only native hZIPK mRNA and protein are detectable in human vascular smooth muscle cells (VSMCs). High stringency screening of a human aortic cDNA library for the SMPP-1M-associated kinase identified 18 positive clones, all of which proved to be clones of hZIPK. PCR-based studies of VSMC RNA revealed native hZIPK transcripts but no evidence for splice variants of hZIPK or a ZIP-like kinase. Northern blotting studies of multiple vascular and non-vascular tissue RNAs, including human bladder RNA, showed only 2.3 kb of mRNA predicted for full-length hZIPK. Immunoblotting showed native full-length 52-kDa hZIPK expression in VSMCs. Full-length and N-terminal hZIPK bound the C-terminal domain (amino acids 681– 847) of the myosin binding subunit (MBS) of SMPP-1M. hZIPK immunoprecipitated with the MBS of SMPP-1M and dominant negative RhoA inhibited the hZIPK-MBS interaction. These data identify hZIPK as the unique SMPP-1-associated kinase expressed in human vesicu-lar smooth muscle and support a role for Rho in promot-ing the hZIPK-MBS interaction.

Blood vessel tone regulates blood pressure and flow and is itself dynamically regulated by the contractile state of vascular smooth muscle cells (VSMCs) 1 in the blood vessel wall. Contraction and relaxation of VSMCs is determined by the phosphorylation state of myosin light chains (MLCs), a process that is tightly regulated by the opposing activities of myosin light chain kinase and myosin phosphatase (SMPP-1M) (1, 2). Myosin phosphatase is the critical enzyme that dephosphorylates MLC, leading to cell relaxation (3).
In recent years, accumulating evidence supports the view that myosin phosphatase activity is highly regulated. Nitrovasodilators, via cGMP and cGMP-dependent protein kinase I␣, lead to activation of PP1M and cell relaxation (4 -8). Vasoconstrictors, conversely, increase MLC phosphorylation by at least two pathways, namely activation of MLCK (9,10) and inhibition of SMPP-1M. Vasoconstrictors can inhibit SMPP-1M by activation of the potent PP1M inhibitor CPI-17 (11,12) or via RhoA-mediated SMPP-1M phosphorylation (13). RhoA, when activated by vasoconstrictors, binds and activates its effector Rho-kinase, which leads to SMPP-1M phosphorylation (13). Although the mechanism by which RhoA and Rho-kinase are targeted to SMPP-1M on contractile fibers is unclear, both RhoA and SMPP-1M have been shown recently to interact with the actin-binding protein M-RIP (14).
Recently, MacDonald et al. identified a SMPP-1M-associated kinase and named it ZIP-like kinase. ZIP-like kinase was isolated from bladder smooth muscle as a 32-Da phosphoprotein that co-purified with SMPP-1M, phosphorylated SMPP-1M at inhibitory residues, and was activated by the smooth muscle contractile agonist carbachol (16). Furthermore, introduction of ZIP-like kinase into rabbit ileal smooth muscle led to calciumindependent contractions (20).
Despite data supporting a physiological role for a ZIP-like kinase in smooth muscle, the precise identity of this kinase and its presence and function in VSMCs remains unclear. Sequencing of ZIP-like kinase-derived peptides revealed high homology to the 52-kDa zipper-interacting protein kinase (hZIPK), raising the possibility that ZIP-like kinase could be a splice variant of hZIPK, a separate kinase with high homology to hZIPK, or a degradation product of hZIPK (16). hZIPK was identified originally as a protein involved in programmed cell death that interacts with the transcription factor ATF4 and mediates apoptosis when overexpressed (21). hZIPK has also been shown to have a potential role in regulating VSMC contraction. hZIPK was found to phosphorylate MLC at both Ser 19 and Thr 18 in a calcium-independent manner, leading to cell contraction (22).
The following study was initiated to determine the precise identity of the PP1M-associated kinase(s) in VSMCs. Through a combination of mRNA and protein analysis, we have found evidence to strongly support the proposition that "ZIP-like kinase" in VSMCs is actually derived from hZIPK, which is present in VSMCs and interacts with SMPP-1M in a RhoAregulated manner.

MATERIALS AND METHODS
Antibodies-Sources of antibodies were as follows. The rabbit polyclonal anti-ZIP kinase (amino acids 279 -298) was from Calbiochem, the anti-FLAG-M2 antibody was from Sigma, the anti-myosin phosphatase polyclonal antibody came from Covance (Berkeley, CA), and normal mouse IgG and normal rabbit IgG were from Santa Cruz Biotechnology Inc., Santa Cruz, CA.
Cell Culture and Transfection-Human embryonic kidney 293 (HEK293) cells were purchased from the American Type Culture Collection (Manassas, VA). Immortalized aorta smooth muscle cells, coronary smooth muscle cells, pulmonary artery smooth muscle cells, and radial artery smooth muscle cells were developed in our laboratory from human tissues by the explant method. These cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum. HEK293 cells were transfected by the calcium phosphate method.
Plasmids-To generate plasmids expressing hZIPK, cDNA-fragments were amplified by a polymerase chain reaction from pACT2 (Clontech) containing hZIPK fragments and cloned into pT7Blue-3 (Novagen, Madison, WI). pT7Blue3-hZIPK was digested with EcoRI and BamHI, and the fragments were cloned into the pFLAG-CMV4 vector (Sigma). GST fusion proteins of full-length and C-terminal hZIPK were generated as follows. DNA fragments corresponding to full-length and C-terminal hZIPK were amplified by PCR using the primer pairs of 5Ј-GGGAATTCTATGTCCACGTTCAGGCAGGAG-3Ј (5Ј) and 5Ј-GGGT-CGACCTAGCGCAGCCCGCACTCCACG-3Ј (3Ј) for the former and 5Ј-GGGAATTCAGCGGCCGCAAGCCCGAGCGGC-3Ј (5Ј) and 5Ј-GGGTC-GACCTAGCGCAGCCCGCACTCCACG-3Ј (3Ј) for the latter. The PCR products were cloned into pT7Blue-3 and digested with EcoRI and SalI. These fragments were cloned into pGEX4T-3 and pGEX4T-1 vectors (Amersham Biosciences). To generate the plasmid expressing the GST-N-terminal region of hZIPK, NotI-digested pGEX4T-3 hZIPK was treated with a Klenow fragment (New England Biolabs, Beverly, MA) to blunt the ends, followed by self-ligation. pEF-BOS-mouse ZIPK was a gift from S. Akira, Department of Biochemistry, Hyogo College of Medicine. pCXN 2 -IRES-EGFP (23) expression vectors were derived from pCAGGS and contain the internal ribosomal entry site (IRES) and the coding region of the enhanced green fluorescent protein (EGFP) at the 3Ј-end of the multiple cloning site. The cDNA fragments encoding the small GTPase RhoA, RhoN19 (dominant negative form), and RhoQ63L (constitutively active form) were amplified by polymerase chain reaction and subcloned into pCXN 2 -IRES-EGFP.
Human Aorta cDNA Library Screening-A human aorta cDNA library containing 3.5 ϫ 10 6 independent clones (Clontech) was screened by the colony hybridization technique (24). 32 P-labeled EcoRI-NotI fragments of hZIPK cDNA were made from the cDNA clone AI660136 (Genome System) encoding the N terminus of hZIPK and used as a probe. The library was transferred to Hybond N synthetic nylon membranes (Amersham Biosciences) and then pre-hybridized and hybridized in 5ϫ Denhardt's solution, 5ϫ SSC, 0.1 M sodium phosphate, 0.5% SDS, and 100 g/ml denatured salmon sperm DNA under high (hybridization temperature 65°C) and low (hybridization temperature 58°C) stringency conditions.
RNA Expression Analysis by Northern Blotting-Total RNA from a series of mouse tissues and poly(A) ϩ RNA from cultured cells and human bladder was electrophoresed in an agarose gel containing 2% formaldehyde and 20 mM MOPS and blotted onto Hybond N synthetic nylon membrane (Amersham Biosciences). A multiple tissue Northern blot that has poly(A) ϩ RNA from a series of human tissues was purchased from Clontech. The membranes were pre-hybridized and hybridized in ExpressHyb hybridization solution (Clontech) with [ 32 P]dCTPlabeled probe. The cDNA probes used in this study are described as follows. A 0.9-kb EcoRI-NotI fragment encoding the 5Ј-end of the human ZIPK cDNA and a 1.4-kb EcoRI-BamHI fragment encoding full-length human ZIPK were made from pFLAG-CMV4-ZIPK. A 1-kb SalI-NotI fragment encoding the 5Ј-end of mouse ZIPK and a 1.3-kb SalI-SalI fragment encoding full-length mouse ZIPK were made from pEF-BOS-ZIPK.
Preparation of GST Fusion Proteins for Binding Studies-BL21 cells were transformed with GST-ZIPK plasmids, described above, and GST-MBS plasmids expressing the C-terminal 183 aa of MBS or the Cterminal 183 aa in which the four leucine residues of the leucine zipper domain have been mutated to alanines (L1007A, L1014A, L1021A, and L1028A) were made as described previously (14). The transformed cells were grown in Luria-Bertani media at 30°C until the absorbance at 600 nm reached 0.6 -0.8. Protein expression was induced by the addition of 0.1 mM isopropyl-␤-D-thiogalactopyranoside for 3 h, and GST fusion proteins were isolated by using glutathione-Sepharose 4B beads.
Immunoprecipitation, Immunoblotting, and Cell Staining-Cells were washed with ice-cold Tris-buffered saline (25 mM Tris-HCl, pH7.5, and 150 mM NaCl), and lysed in lysis buffer (20 mM Tris-HCl, pH7.5, 137 mM NaCl, 1% Triton X-100, 2 mM EDTA, 10% glycerol, 25 mM ␤-glycerophosphate, protease inhibitor mixture set III (Calbiochem), and protease inhibitor mixture tablets (Roche Applied Science). Lysates were cleared by centrifugation at 15,000 ϫ g for 10 min. Aliquots of total cell lysate were subjected to immunoblotting with antibodies as indicated in Figs. 1 and 5-7. The remaining cell lysate was subjected to either immunoprecipitation using antibodies as indicated in Figs. 6 and 7 and protein A-and G-Sepharose (Amersham Biosciences) or GST fusion protein interaction studies by incubation with the indicated GST fusion protein. This treatment was followed by SDS-PAGE and transfer to polyvinylidene difluoride membranes. The blocked membranes were incubated with the primary antibody at 4°C overnight and then probed with the secondary antibody linked to peroxidase. Immunoreactive bands were visualized by a chemiluminescence system (Amersham Biosciences) and subjected to densitometric quantitation. For immunofluorescence labeling, cells were washed with phosphate-buffered saline, fixed by 4% paraformaldehyde at room temperature, and then permeabilized with 0.1% Triton X-100. Permeabilized cells were incubated with an anti-FLAG antibody followed by Alexa 488 goat antimouse IgG (Molecular Probes, Eugene, OR).
Statistical Analysis-Data are presented as means Ϯ S.E. Statistical difference was evaluated by Student's t test. A p value of Ͻ 0.01 was regarded as significant.

RESULTS
Immunoblotting Studies-A 32-kDa SMPP-1M-associated kinase was detected previously by immunoblotting of rabbit bladder lysates with the anti-hZIPK antibody (16). We have used this antibody in immunoblotting studies of proteins expressed by native vascular smooth muscle cells and in studies of HEK293 cells expressing the cDNA for hZIPK (Fig. 1). In both VSMC cells and in heterologous expression studies of HEK293 cells the most prominent protein band detected by the anti-ZIPK antibody was 52 kDa (Fig. 1, upper arrows). A 32-kDa band was also detected in both cell types (arrowhead) but was less prominent, especially in human VSMC lysates. The inclusion of protease inhibitors did not alter the amount of 32-kDa protein detected (ϩ lanes, Fig. 1). In the HEK293 cells expressing hZIPK, another protein of ϳ48 kDa was also prominent. Detection of this 48-kDa protein was also not affected by the inclusion of a mixture of protease inhibitors.
cDNA Library Screening for ZIPK and ZIP-like Kinase-A human aorta cDNA library containing 3.5 ϫ 10 6 independent clones was screened for the presence of ZIPK and ZIP-like kinase by the colony hybridization technique. A 981-bp fragment of the 5Ј-end of hZIPK was used as the probe. This region of hZIPK cDNA was chosen for the probe design based on a previous report of homology between peptide fragments of ZIPlike kinase and the amino-terminal kinase domain of ZIPK (16) (Fig. 2A). The human aorta cDNA library was screened under both high and low stringency conditions. Using high stringency conditions, we obtained 18 positive clones. These positive clones were fully sequenced, and all 18 positive clones were identified as either full-length or fragments of hZIPK. Under low stringency conditions, 41 positive clones were obtained, and 31 were identified as full-length or fragments of hZIPK. We also cloned myosin light chain kinase (eight clones), the kinase domain of which has 48% amino acid sequence identity and 59% nucleotide sequence identity (25), as well as part of the ZIPK gene on chromosome 19 (two clones). ZIP-like kinase was not identified by either of the library screens.
PCR Studies to Search for ZIP-like Kinase and Detect Potential ZIPK Splice Variants-We designed degenerate PCR primers corresponding to the three peptide sequences derived originally from ZIP-like kinase with minor or uncertain sequence differences from hZIPK (Fig. 2B) (16). All DNA fragments amplified from the human aorta library using these degenerate primers were sequenced (Fig. 2C) but proved to be nonspecific, and no sequences corresponded to those of hZIPK or ZIP-like kinase.
To determine whether ZIP-like kinase is a splice variant of hZIPK, we searched for hZIPK splice variants in the human aorta cDNA library. hZIPK mRNA is derived from eight exons on chromosome 19. We designed PCR primers corresponding to each exon and two putative exons designated potential exons A and B, one 5Ј to exon1 (potential exon A) and the other between exon 2 and exon 3 (potential exon B) (Fig. 3A) based on predicted potential splice sites and variants (the NCBI LocusLink program). PCR was performed with 11 primer pairs designed to amplify both the known exons and potential exons A and B. Only the hZIPK exons were amplified. A number of DNA fragments were sequenced and proved to be of 100% identity to hZIPK (Fig. 3B). These results make the existence of a potential splice variant in human aorta of hZIPK unlikely.
Northern Blotting Studies-We investigated the expression of mRNA encoding ZIP-like kinase in various human tissues and cultured cells by performing Northern blot analyses. In all of these human tissues and cultured cells, a 5Ј-probe, based on hZIPK, was used at moderate stringency to detect hZIPK and any related transcripts. In all tissues and cells examined, a 2.3-kb band consistent with the predicted size of the hZIPK transcript was present, but no evidence of any other transcripts was noted (Fig. 4, A and B). ZIP-like kinase was co-purified with SMPP-1M originally from bladder smooth muscle (16). We therefore obtained human bladder tissue, isolated the mRNA, and performed separate Northern blot experiments (Fig. 4B, right panel). In human bladder, only a 2.3-kb band consistent with the predicted size of the hZIPK transcript was detected, and no evidence of any other transcript was noted. Re-probing of these blots with full-length hZIPK probe led to detection of the same bands and no others, indicating that these 2.3-kb bands are hZIPK mRNA (data not shown). We also investigated hZIPK and ZIP-like kinase expression in various mouse tissues, including brain, aorta, heart, lung, liver, spleen, kidney, uterus, and bladder. Among the mouse tissues, an ϳ2.3-kb band was detected. No other smaller mRNA species were detected (data not shown). We therefore could find no evidence that suggested the existence of an mRNA for a ZIP-like kinase in human and mouse tissues or in cultured cells.
Interaction of hZIPK with MBS-We investigated whether full-length hZIPK interacts with MBS. FLAG-tagged hZIPK was well expressed in HEK293 cells and distributed uniformly throughout the cytoplasm in these cells (Fig. 5, A and B).  (Fig. 5C). The binding of native ZIPK from human vascular smooth muscle cells to these GST-MBS fusion proteins was similarly observed (Fig. 5E). In reciprocal studies, full-length ZIPK and N-terminal ZIPK both bound MBS, whereas GST-C-terminal ZIPK did not, supporting the idea that the amino-terminal half of ZIPK binds to MBS (Fig. 5D). Next, MBS and hZIPK were each immunoprecipitated from cells to test for an interaction between the two proteins in vivo. When MBS was immunoprecipitated from HEK293 cells, FLAG-hZIPK was present in the immunopellet (Fig. 6A), and, reciprocally, the immunoprecipitation of FLAG-hZIPK led to the recovery of MBS in the immunopellet (Fig. 6B). These data indicate that hZIPK and MBS are complexed in the cell. Finally, we studied whether the small GTPase RhoA might regulate the interaction we observed between hZIPK and MBS. In GST pull-down assay studies, no differences were detected in the level of ZIP kinase bound by MBS from native Rho with FIG. 2. Cloning of human ZIPK and related kinases from the human aortic vascular cDNA library. A, diagram of EcoRI-NotI fragments encoding the N-terminal kinase domain of ZIPK used as the probe to screen the human aortic cDNA library at both low and high stringency. Probe design was based on the previous report of homology between peptide fragments of ZIP-like kinase and the amino-terminal kinase domain of ZIPK (16). B, summary of clones isolated from the cDNA library screen with both low and high stringency hybridization conditions. C, degenerate PCR Studies to detect ZIP Kinase and ZIP-like kinase. Amino acid sequences of peptides derived from the putative ZIP-like kinase (16) and aligned with the amino acid sequence of hZIPK (GenBank TM accession number AB022341) are shown. Degenerate PCR primers corresponding to these amino acid sequences were designed (arrows) and used in PCR reactions with the human aorta library DNA as the template. The PCR products of primer pair IA and IIA shown were used as templates for PCR reactions with primer pair IB and IIB, respectively. D, PCR products of primer pair IB and IIB, as well as primer pair III, were size-fractionated by agarose gel electrophoresis (shown in the far left lanes in each panel). Four fragments from primer pair I and II and two fragments from primer pair III were gel-extracted and reamplified and are shown in the right section of the gel for each primer pair. The reamplified fragments were used for DNA sequencing. All cDNA recovered proved to be nonspecific, and no sequences corresponded to hZIPK or ZIP-like kinase were detected. either RhoN19 (dominant negative) or RhoQL (constitutively active) (data not shown). MLC phosphorylation by ZIP kinase also was assayed by isolating ZIP kinase from cells expressing wild type Rho, RhoN19, or RhoQL, but differences in the level of MLC phosphorylation were not detected in this assay (data not shown). When hZIPK was immunoprecipitated, however, co-immunoprecipitation of MBS was significantly diminished in the presence of a dominant negative mutant of RhoA, RhoN19 (Fig. 7). The constitutively active Rho protein, RhoQL, did not increase the level of ZIPK bound to MBS above the level seen with native (wild type) Rho in these studies (Fig. 7). These immunoprecipitation data support the belief that active Rho promotes the hZIPK-MBS interaction characterized in these studies.

DISCUSSION
Because the regulatory pathways that modulate SMPP-1M activity remain incompletely understood, we undertook this study to identity the endogenous SMPP-1M-associated kinase(s). A 32-kDa ZIP-like kinase had been found previously to co-purify with and inhibit SMPP-1M activity, whereas 52-kDa hZIPK had been shown to phosphorylate MLC and cause cell contraction (16,22,25). We set out to clone the SMPP-1Massociated ZIP-like kinase from vascular tissue, employing two methods to screen a highly representative human aorta cDNA library for evidence of ZIP-like kinase. In the first approach, we performed colony hybridization using a probe against the 5Ј-half of hZIPK, which, because of high homology, would be expected to identify both ZIP-like kinase and hZIPK. We also used a reverse transcription PCR approach with this human aorta library as the template and both degenerate primers and primers designed to identify hZIPK splice variants. In our studies, only hZIPK but not any ZIP-like kinase was present in this library.
As alternative approaches, we also used both the 5Ј-catalytic domain and full-length hZIPK as probes in both vascular and non-vascular cellular RNA Northern hybridization experiments. A variety of human tissues as well as human blood vessel, heart, and VSMCs from a variety of human blood vessel types were studied. Importantly, human bladder tissue was also studied because the ZIP-like kinase was originally purified from bladder smooth muscle tissue (16). In all of these tissues and cells only a single transcript of the expected 2.4 kb for hZIPK was detectable, supporting further the proposal that hZIPK is the sole SMPP-1 M-associated kinase expressed in human tissues. . These filters, along with 1.5 g of poly(A) ϩ RNA from various human tissues, were hybridized with the radiolabeled 5Ј-end of hZIPK and exposed to a PhosphorImager (Amersham Biosciences). B, Northern filters prepared from human primary vascular cell lines and human bladder tissue RNA. Poly(A) ϩ RNAs (2 g) from various cultured cells and human bladder were loaded in each lane. The filters were hybridized with the radiolabeled 5Ј-end of hZIPK as a probe and exposed to a PhosphorImager. AoSMC, primary human aortic smooth muscle cell; CoSMC, human coronary arterial smooth muscle cell; PuSMC, human pulmonary arterial smooth muscle cell; RaSMC, human radial artery smooth muscle cell; HUVEC, human umbilical vein endothelial cells.
We also tested for the binding of hZIPK to MBS in several ways. Heterologous expression of hZIPK demonstrated that hZIPK binds directly to the coiled coil-containing domain of MBS in GST pull-down protein-protein interaction studies and to complexes with MBS in co-immunoprecipitation experiments, strongly supporting the idea that hZIPK interacts with MBS in the cell. Furthermore, disruption of RhoA activation by overexpression of a dominant negative RhoA mutant inhibited this hZIPK-MBS interaction in cells. We did not detect an effect of RhoN19 on hZIPK-MBS binding using GST fusion proteins or in a MLC phosphorylation assay using immunoprecipitated hZIPK derived from cells expressing wild type or mutant Rho protein. These negative data support either the reduced sensitivity of these assays compared with co-immunoprecipitation or the need for additional cellular proteins for Rho to regulate the hZIPK-MBS interaction. Taken together, the data support the hypothesis that the interaction between hZIPK and MBS may be regulated by RhoA and that hZIPK may play a role in PP1M regulation by RhoA.
Studies in both animals and humans have shown that SMPP-1M activity is regulated in health and in disease. Hyperactivity of the RhoA pathway leading to SMPP-1M inhibition and blood vessel contraction has been shown in hypertensive states, and enzyme inhibitors that prevent SMPP-1M inhibition are promising new treatments for cardiovascular disorders (26,27). Our data establish that hZIPK is the SMPP-1M-associated kinase and support the hypothesis that the hZ-IPK-SMPP-1M interaction is regulated by RhoA. Future studies will be directed at defining the role of hZIPK in the modulation of SMPP-1M activity by vasoconstrictor agonists and other potential mechanisms through which hZIPK regulates vascular smooth muscle cell function.  7. Effect of small GTPase RhoA on the interaction between human ZIPK and MBS. HEK293 cells were cotransfected with hZIPK and wild type (WT) RhoA or either dominant negative (N19) or constitutively active (QL) RhoA. Immunoprecipitation (IP) with normal mouse IgG (control; left) or anti-FLAG antibody for ZIP kinase (right) was followed by immunoblotting (IB) with anti-MBS antibody and densitometric quantitation. The signal intensity of Rho wild type (Rho WT) was arbitrarily designated as unity. *, p Ͻ 0.01(n ϭ 3 to 5).