Rho-associated kinase and zipper-interacting protein kinase, but not myosin light chain kinase, are involved in the regulation of myosin phosphorylation in serum-stimulated human arterial smooth muscle cells

Myosin regulatory light chain (LC20) phosphorylation plays an important role in vascular smooth muscle contraction and cell migration. Ca2+/calmodulin-dependent myosin light chain kinase (MLCK) phosphorylates LC20 (its only known substrate) exclusively at S19. Rho-associated kinase (ROCK) and zipper-interacting protein kinase (ZIPK) have been implicated in the regulation of LC20 phosphorylation via direct phosphorylation of LC20 at T18 and S19 and indirectly via phosphorylation of MYPT1 (the myosin targeting subunit of myosin light chain phosphatase, MLCP) and Par-4 (prostate-apoptosis response-4). Phosphorylation of MYPT1 at T696 and T853 inhibits MLCP activity whereas phosphorylation of Par-4 at T163 disrupts its interaction with MYPT1, exposing the sites of phosphorylation in MYPT1 and leading to MLCP inhibition. To evaluate the roles of MLCK, ROCK and ZIPK in these phosphorylation events, we investigated the time courses of phosphorylation of LC20, MYPT1 and Par-4 in serum-stimulated human vascular smooth muscle cells (from coronary and umbilical arteries), and examined the effects of siRNA-mediated MLCK, ROCK and ZIPK knockdown and pharmacological inhibition on these phosphorylation events. Serum stimulation induced rapid phosphorylation of LC20 at T18 and S19, MYPT1 at T696 and T853, and Par-4 at T163, peaking within 30–120 s. MLCK knockdown or inhibition, or Ca2+ chelation with EGTA, had no effect on serum-induced LC20 phosphorylation. ROCK knockdown decreased the levels of phosphorylation of LC20 at T18 and S19, of MYPT1 at T696 and T853, and of Par-4 at T163, whereas ZIPK knockdown decreased LC20 diphosphorylation, but increased phosphorylation of MYPT1 at T696 and T853 and of Par-4 at T163. ROCK inhibition with GSK429286A reduced serum-induced phosphorylation of LC20 at T18 and S19, MYPT1 at T853 and Par-4 at T163, while ZIPK inhibition by HS38 reduced only LC20 diphosphorylation. We also demonstrated that serum stimulation induced phosphorylation (activation) of ZIPK, which was inhibited by ROCK and ZIPK down-regulation and inhibition. Finally, basal phosphorylation of LC20 in the absence of serum stimulation was unaffected by MLCK, ROCK or ZIPK knockdown or inhibition. We conclude that: (i) serum stimulation of cultured human arterial smooth muscle cells results in rapid phosphorylation of LC20, MYPT1, Par-4 and ZIPK, in contrast to the slower phosphorylation of kinases and other proteins involved in other signaling pathways (Akt, ERK1/2, p38 MAPK and HSP27), (ii) ROCK and ZIPK, but not MLCK, are involved in serum-induced phosphorylation of LC20, (iii) ROCK, but not ZIPK, directly phosphorylates MYPT1 at T853 and Par-4 at T163 in response to serum stimulation, (iv) ZIPK phosphorylation is enhanced by serum stimulation and involves phosphorylation by ROCK and autophosphorylation, and (v) basal phosphorylation of LC20 under serum-free conditions is not attributable to MLCK, ROCK or ZIPK.

Smooth muscle MLCK is capable of phosphorylating T18 of LC 20 in addition to S19 [5,6]; however, this has only been demonstrated in vitro at high kinase concentrations (a high MLCK: myosin II ratio), and abundant evidence indicates that it does not occur in situ [7,8].
In some instances, however, LC 20 is phosphorylated at both T18 and S19 in vascular smooth muscle tissue. For example, in renal afferent arterioles of the rat, the pathophysiological stimulus endothelin-1 induces diphosphorylation of LC 20 at T18 and S19, whereas the physiological stimulus angiotensin II induces exclusively monophosphorylation at S19 [9]. Diphosphorylation is associated with a decrease in the rate of dephosphorylation of LC 20 by myosin light chain phosphatase (MLCP) and, therefore, prolongation of the contractile response [10]. LC 20 diphosphorylation has also been observed frequently in cultured smooth muscle and endothelial cells [11][12][13]. The observation that LC 20 diphosphorylation occurs in response to treatment with membrane-permeant phosphatase inhibitors indicates that blocking phosphatase activity unmasks basal activity of endogenous kinase(s) that phosphorylate LC 20 at T18 and S19 [14]. These findings raise the question: which kinase(s) is responsible for phosphorylation of LC 20 at T18 and S19? Candidate kinases include Rho-associated coiled-coil kinase (ROCK) [15], zipper-interacting protein kinase (ZIPK) [16][17][18][19][20], integrin-linked kinase (ILK) [21,22] and citron kinase (a Rho-dependent kinase related to ROCK) [23]. In this study, we followed the time courses of phosphorylation of LC 20 and of proteins implicated in the regulation of myosin light chain phosphorylation, the myosin targeting subunit of MLCP (MYPT1), prostate-apoptosis response-4 (Par-4) and ZIPK itself, in serum-stimulated human cultured vascular smooth muscle cells from coronary artery and umbilical artery. We then investigated the effects of MLCK, ROCK and ZIPK knockdown and pharmacological inhibition on the seruminduced phosphorylation of LC 20 , MYPT1, Par-4 and ZIPK.

Human arterial smooth muscle cell cultures
Human coronary artery smooth muscle cells (CASMC: CC-2583) and human umbilical artery smooth muscle cells (UASMC: CC-2579), purchased from Lonza (Basel, Switzerland), were cultured in smooth muscle growth medium (Lonza SmGM-2: CC-3182) containing 5% fetal bovine serum (FBS), growth factors (hEGF, insulin and hFGF-β), gentamicin and amphotericin-B at 37˚C in a humidified incubator with 5% CO 2 . These cells maintain their morphological and phenotypic characteristics for up to 10 passages and were, therefore, used for experiments within 10 passages. Cells were plated at a density of 2 X 10 5 cells/ml into 4-well plates (Nunc, Thermo Scientific, Waltham, MA) with smooth muscle growth medium containing 5% FBS. Prior to treatment, cells were grown to 70% confluence and starved in FBSfree medium overnight. Cells were then stimulated with 5% FBS for times indicated in the figures (typically 0, 15 s, 30 s, 1 min, 2 min, 5 min, 30 min, 1 h and 2 h) to observe their various responses to FBS. The treated cells were lysed immediately in Laemmli sample buffer and heated at 90˚C for 5 min. To investigate the effects of kinase inhibitors, cells were starved in FBS-free medium for 8 h to induce cell quiescence and then incubated with various inhibitors for � 45 min prior to stimulation with 5% FBS for 2 min and then lysed in Laemmli sample buffer.
Negative control siRNA (Ambion negative control #2, part #AM4613) contains sequences that do not target any gene product.

Phosphate affinity SDS-PAGE (Phos-tag SDS-PAGE)
The phosphorylation of LC 20 and ZIPK was analyzed by phosphate-binding tag SDS-PAGE [35]. For LC 20 phosphorylation analysis, samples were resolved in SDS gels containing 12% acrylamide, 40 μM Phos-tag reagent and 0.1 mM MnCl 2 at 20 mA/gel. Separated proteins were transferred to PVDF membranes (Roche Applied Science, Penzberg, Germany) overnight at 28 V and 4˚C in 25 mM Tris-HCl, pH 7.5, 192 mM glycine, 10% (v/v) methanol. Proteins were fixed on the membrane by treatment with 0.5% glutaraldehyde in phosphate-buffered saline (137 mM NaCl, 2.68 mM KCl, 10 mM Na 2 HPO 4 , 1.76 mM KH 2 PO 4 ) for 30 min. For ZIPK phosphorylation analysis, samples were separated in SDS gels containing 6.5% acrylamide, 55 μM Phos-tag reagent and 0.1 mM MnCl 2 (or 1-2 mM EDTA) at 20 mA/gel. Separated proteins were transferred to nitrocellulose membranes in 25 mM Tris-HCl, pH 7.5, 192 mM glycine, 20% (v/v) methanol at 28 V for 18 h at 4˚C. Membranes were then blocked with 5% non-fat dry milk in TBST for 1 h at 20˚C, washed with water and TBST and incubated for 2 h with primary antibody (anti-ZIPK) in Can Get Signal (Immunoreaction Enhancer Solution 1; Toyobo Life Science Department, Osaka, Japan) prior to further washing in water and TBST and incubation for 1 h with secondary antibody in 1% (w/v) non-fat dry milk in TBST. Immunoreactive bands were detected with SuperSignal West Femto Maximum Sensitivity Substrate (Thermo Fisher Scientific).

Data analysis
Values are presented as mean ± SEM, with n indicating the number of independent experiments performed. Statistically significant differences were determined by Student's unpaired ttest or Dunnett's post hoc test for multiple comparisons as indicated in the text and figure legends. Statistically significant differences were indicated by p < 0.05.

Serum-induced phosphorylation of LC 20
Analysis of LC 20 phosphorylation in serum-starved human coronary and umbilical arterial smooth muscle cells by Phos-tag SDS-PAGE, which separates unphosphorylated, mono-and diphosphorylated LC 20 species [10], revealed a significant basal level of phosphorylation: 0.55 ± 0.05 mol P i /mol LC 20 (n = 7) in CASMC and 0.85 ± 0.02 mol P i /mol LC 20 (n = 6) in UASMC (time zero samples in Fig 1B and 1D, respectively). Most of the basal phosphorylation under these conditions was at a single site, although diphosphorylated LC 20 was also detected (time zero samples in Fig 1A and 1C). Serum stimulation induced a rapid increase in phosphorylation to a peak of 1.12 ± 0.08 mol P i /mol LC 20 (n = 7) in CASMC (Fig 1A and 1B) and 1.33 ± 0.02 mol P i /mol LC 20 (n = 6) in UASMC (Fig 1C and 1D) at 2 min after serum addition. In the case of CASMC, this level of LC 20 phosphorylation was sustained for a prolonged period of time; the value after 2 h serum stimulation was 0.97 ± 0.03 mol P i /mol LC 20 (n = 7), not significantly different from the value at 2 min (p = 0.09; Student's unpaired t-test) (Fig 1B), whereas it declined significantly over this period in UASMC to 1.02 ± 0.04 mol P i /mol LC 20 (n = 6), significantly lower than the value at 2 min (p < 0.001; Student's unpaired t-test) ( Fig  1D). Analysis of LC 20 diphosphorylation using phosphospecific antibodies that recognize LC 20 only when phosphorylated at both T18 and S19 confirmed a low level of basal diphosphorylation under serum-free conditions, which increased rapidly in both CASMC (Fig 2A) and UASMC ( Fig 2B) in response to serum stimulation and then significantly declined in UASMC (p < 0.02 versus peak value; Student's unpaired t-test) to steady-state levels significantly above resting values ( Fig 2B); this is also demonstrated by the diphosphorylated band in the Phos-tag gels in Fig 1A and 1C. There was no significant decline in diphosphorylation of LC 20 over the 2-h incubation period in CASMC (Fig 2A; p = 0.13 versus peak value; Student's unpaired ttest).

Serum-induced phosphorylation of MYPT1
Analysis of MYPT1 phosphorylation at T696 and T853 by western blotting with phosphospecific antibodies revealed a low basal level of phosphorylation at both sites under serum-free conditions in CASMC (Fig 3A and 3B) and UASMC (Fig 3C and 3D). Serum stimulation induced a rapid increase in phosphorylation at both sites in both cell types, which peaked at 1-2 min and declined steadily with time, in most cases returning to basal levels within 2 h of exposure to serum. Representative western blots are shown in panels A and C above cumulative quantitative data for each of the LC 20 bands. Panels B and D show the time courses of total phosphorylation stoichiometry, calculated as follows: mol P i /mol LC 20 = (1P + 2x2P) / (0P + 1P + 2P). Values indicate the mean ± SEM (n = 7 for panels A and B; n = 6 for panels C and D). Significant differences from the value at time zero are indicated: � p < 0.0001, † p = 0.0001 (B), � p < 0.0001, † p = 0.0023, # p = 0.0004 (D) (Dunnett's post hoc test). https://doi.org/10.1371/journal.pone.0226406.g001

Serum-induced phosphorylation of Par-4
Analysis of Par-4 phosphorylation at T163 by western blotting with phosphospecific antibodies revealed a low basal level of phosphorylation under serum-free conditions in CASMC ( Fig 4A) and UASMC (Fig 4B). Serum stimulation induced a rapid increase in phosphorylation in both cell types, which peaked at 30 s and declined towards basal levels over the 2-h incubation period.

Serum-induced phosphorylation of Akt
We also investigated the time courses of serum-induced phosphorylation events distinct from the signaling pathways involved in LC 20 phosphorylation, specifically the kinases Akt/PKB, ERK1/2 and p38 MAPK as well as the kinase substrate, heat shock protein HSP27. PI3K/Akt signaling in VSMCs is involved in the etiology of atherosclerosis [36,37]. Phosphorylation at T308 and S473 of Akt/PKB is required for full activation [38]. Analysis of Akt phosphorylation at S473 by western blotting with phosphospecific antibodies revealed a low basal level of phosphorylation under serum-free conditions in CASMC (S1A Fig) and UASMC (S1B Fig). Serum stimulation induced a slower rate of increase in phosphorylation in both cell types compared to the phosphorylation events described above, peaked at 5-30 min and remained elevated over the 2-h incubation period.

Serum-induced phosphorylation of ERK1/2
The ERK MAPKs, ERK1/p44 and ERK2/p42, are activated by a wide variety of growth factors and mitogens via phosphorylation at the T and Y residues (T202 and Y204 in ERK1 and T185 and Y187 in ERK2) within the activation loop TEY sequence [39]. ERK1 and ERK2 have been implicated in cell migration [40]. Analysis of ERK1 and ERK2 phosphorylation by western

Serum-induced phosphorylation of p38 MAPK
p38 MAPK has also been implicated in cell migration through activation by several growth factors, cytokines and chemotactic substances [40] and is activated by phosphorylation of the T and Y residues in the TGY motif of the activation loop (T180 and Y182 of human p38MAPKα) [41]. Analysis of p38 MAPK phosphorylation at T180 and Y182 by western blotting with phosphospecific antibodies revealed a low basal level of phosphorylation under serum-free conditions in CASMC (S4A  induced a relatively slow rate of increase in phosphorylation in both cell types, which peaked at 5 min and declined towards basal levels over the 2-h incubation period.

Serum-induced phosphorylation of HSP27
The small heat shock protein HSP27 acts as an ATP-independent chaperone in protein folding, has been implicated in cell migration and cytoskeletal architecture, and is regulated by phosphorylation at multiple sites [42], including S82, which induces polymer assembly and actin binding [43]. Analysis of HSP27 phosphorylation at S82 by western blotting with phosphospecific antibodies revealed a low basal level of phosphorylation under serum-free conditions in CASMC (S4C

Knockdown of MLCK, ROCK and ZIPK
Transfection with siRNA was used to investigate the involvement of MLCK, ROCK and ZIPK in serum-induced phosphorylation events. Transfection of CASMC with siRNA targeting MLCK reduced MLCK protein to~55% of control levels (cells transfected with negative control siRNA) but had no effect on ROCK1 or ZIPK levels ( Fig 5).
Transfection of CASMC and UASMC with siRNA targeting ZIPK reduced ZIPK protein tõ 55% and~50% of control levels (cells transfected with negative control siRNA), respectively (Fig 6), confirming our previous observations [33]. Consistent with our previous findings [33], ZIPK knockdown was accompanied by an increase in ROCK1 protein to~165% of control levels in CASMC and to~133% of control levels in UASMC (Fig 6).

Effect of MLCK knockdown and inhibition on phosphorylation of LC 20
To investigate the role of MLCK in LC 20 phosphorylation following serum stimulation, we first examined the levels of smooth muscle MLCK in CASMC and UASMC. As shown in Fig  7A, MLCK of the expected molecular weight (130 kDa) for smooth muscle MLCK was clearly detected in CASMC by western blotting, but we could not detect the larger non-muscle isoform of 210 kDa [4]. We were unable to detect a consistent, clear signal for MLCK in UASMC, however, and therefore further experiments were carried out with CASMC. MLCK knockdown had no significant effect on LC 20 diphosphorylation at T18 and S19 in response to serum treatment for 2 min, as shown by SDS-PAGE and western blotting with anti-2P-LC 20 (Fig 7B). Phos-tag SDS-PAGE and western blotting with anti-LC 20 confirmed the lack of effect of MLCK knockdown on serum-induced LC 20 phosphorylation: phosphorylation at one and two sites and overall phosphorylation stoichiometry were unaffected by MLCK knockdown (Fig 7C and 7D).
The MLCK inhibitors, ML7 and wortmannin, also had no significant effect on LC 20 phosphorylation at T18 and S19 in response to serum treatment. SDS-PAGE and western blotting with anti-2P-LC 20 indicated that serum-induced LC 20 diphosphorylation was unaffected by ML7 ( Fig 8A). Phos-tag SDS-PAGE analysis confirmed that there was no significant difference between serum-induced mono-or diphosphorylation of LC 20 in the absence or presence of ML7 ( Fig 8B). Likewise, SDS-PAGE and western blotting with anti-2P-LC 20 indicated that serum-induced LC 20 diphosphorylation was unaffected by wortmannin (Fig 8C), which was confirmed by Phos-tag SDS-PAGE analysis (Fig 8D). In control experiments, we verified the membrane permeability and efficacy of ML7 by demonstrating inhibition of U46619 (thromboxane A 2 mimetic)-induced contraction of de-endothelialized rat caudal arterial smooth muscle, as described previously [44]. We also verified that wortmannin (an inhibitor of PI 3-kinase as well as MLCK) is membrane permeant and effective since it inhibited phosphorylation of Akt, a PI 3-kinase substrate (S5 Fig).
Finally, chelation of Ca 2+ with EGTA to remove extracellular Ca 2+ and deplete intracellular stores had no significant effect on serum-induced LC 20 phosphorylation at T18 and S19. From western blotting with anti-diP-LC 20 , there was no significant difference between the normalized signal intensities for diP-LC 20 from cells treated with serum in the presence of Ca 2+ or EGTA (Fig 9A). From Phos-tag SDS-PAGE and western blotting with anti-LC 20 , LC 20 phosphorylation stoichiometry was identical in the presence of Ca 2+ and EGTA (Fig 9B). EGTA   treatment affected the mobility of LC 20 species during Phos-tag SDS-PAGE (Fig 9B) but did not affect the phosphorylation stoichiometry. We are confident that the high concentration of EGTA used (5 mM) does effectively deplete the endogenous calcium stores in smooth muscle cells since we have shown that a lower EGTA concentration (2 mM) abolished both caffeineinduced and U-46619-induced contraction of de-endothelialized rat caudal arterial smooth muscle and U-46619-induced LC 20 phosphorylation, which occurred exclusively at S19 [44].

Effect of ROCK knockdown on phosphorylation of LC 20 , MYPT1 and Par-4
ROCK-knockdown and control cells were serum starved overnight, treated with 5% FBS for 2 min and lysed for SDS-PAGE and western blotting. Down-regulation of ROCK correlated with significant reductions in serum-induced diphosphorylation of LC 20 (Fig 10), of MYPT1 at T696 and T853, and of Par-4 at T163 in both CASMC and UASMC (Fig 11).

Effect of ROCK inhibition on phosphorylation of LC 20 , MYPT1 and Par-4
Pre-treatment of cells with the ROCK inhibitor GSK429286A (1 μM) had a pronounced inhibitory effect on serum-induced LC 20 diphosphorylation in both CASMC and UASMC (Fig 12A  and 12D). Phos-tag SDS-PAGE confirmed that LC 20 phosphorylation stoichiometry was reduced by ROCK inhibition in CASMC and UASMC (Fig 13A and 13B). ROCK inhibition by GSK429286A also markedly reduced the serum-induced increase in phosphorylation of MYPT1 at T853 and of Par-4 at T163, but had no significant inhibitory effect on MYPT1 phosphorylation at T696 (Fig 12A and 12D). Similarly, ROCK inhibition by H1152 (0.2 μM) in CASMC had no effect on MYPT1 phosphorylation at T696 but markedly reduced MYPT1 phosphorylation at T853 and Par-4 phosphorylation at T163 (Fig 12B, 12C and 12D).

Effect of ZIPK knockdown on phosphorylation of LC 20 , MYPT1 and Par-4
ZIPK-knockdown and control cells were serum starved overnight, treated with 5% FBS for 2 min and lysed for SDS-PAGE and western blotting. LC 20 diphosphorylation was reduced tõ 60% of control in CASMC and UASMC (Fig 11A, 11B and 11D). Phos-tag SDS-PAGE confirmed that ZIPK knockdown caused a small, but significant reduction in serum-induced LC 20 phosphorylation in both CASMC and UASMC (Fig 10). ZIPK knockdown in both CASMC and UASMC increased MYPT1 phosphorylation at T696 and T853 and Par-4 phosphorylation at T163 (Fig 11A, 11B and 11D). The increases in MYPT1 and Par-4 phosphorylation could be accounted for by the increase in ROCK protein levels induced by ZIPK knockdown shown earlier (Fig 6).

Effect of ZIPK inhibition on phosphorylation of LC 20 , MYPT1 and Par-4
ZIPK inhibition by HS38 [28] attenuated the increase in LC 20 diphosphorylation induced by treatment of CASMC and UASMC with 5% FBS for 2 min, but had no effect on the phosphorylation of MYPT1 at T696 or T853 or of Par-4 at T163 (Fig 12A and 12E). Phos-tag SDS-PAGE confirmed that HS38 reduced the serum-induced phosphorylation stoichiometry of LC 20 in CASMC and UASMC (Fig 13A and 13C).

Serum-induced phosphorylation of ZIPK and the effect of ROCK and ZIPK knockdown and inhibition
ZIPK phosphorylation, reflecting activation of the kinase [45], was assessed by Phos-tag SDS-PAGE to separate phosphorylated and unphosphorylated forms of the enzyme, which were then detected by western blotting with a pan-ZIPK antibody. Under serum-free conditions, ZIPK was partially phosphorylated in CASMC and UASMC with low levels of phosphorylation at one, two, three and four sites (time zero samples in Fig 14A and 14B, respectively). Serum treatment induced a rapid increase in ZIPK phosphorylation, particularly of the diphosphorylated protein, which was maintained for � 1 h in CASMC (Fig 14A) but transient in UASMC (Fig 14B). Knockdown of ROCK1 had a significant inhibitory effect on serum-induced ZIPK phosphorylation, while ZIPK siRNA treatment reduced ZIPK levels but did not affect the protein's phosphorylation in response to serum treatment (Fig 15A, lefthand panel). Chelation of divalent cations with EDTA resulted in a single band of ZIPK   corresponding to the migration of the unphosphorylated protein (Fig 15A, right-hand panel) confirming that the slower migrating bands observed in the presence of MnCl 2 (Fig 15A,    and UASMC (Fig 15C) suggesting that ROCK phosphorylates ZIPK, and ZIPK can autophosphorylate. The combination of ROCK and ZIPK inhibitors prevented ZIPK phosphorylation induced by serum stimulation and almost abolished the basal level of ZIPK phosphorylation in the absence of serum stimulation of UASMC (Fig 15D).

Effects of kinase knockdown and inhibition on LC 20 phosphorylation under serum-free conditions
As shown in Fig 1, there is a basal level of LC 20 diphosphorylation in serum-starved cultured human vascular smooth muscle cells. We used a combination of siRNA-mediated knockdown and pharmacological inhibition to evaluate the roles of MLCK, ROCK and ZIPK in the regulation of LC 20 phosphorylation in the absence of serum stimulation. Phos-tag SDS-PAGE and western blotting with anti-LC 20 showed no effect of MLCK knockdown on LC 20 phosphorylation under serum-free conditions (Fig 16A). The MLCK inhibitors, ML7 and wortmannin, also had no significant effect on LC 20 phosphorylation at T18 and S19 under serum-free conditions (Fig 16B and 16C, respectively). Furthermore, chelation of Ca 2+ with EGTA had no effect on LC 20 phosphorylation at T18 and S19 in the absence of serum stimulation (Fig 16D). Finally, ROCK1 knockdown ( Fig 17A) or inhibition by GSK429286A (Fig 17B) or ZIPK knockdown (Fig 17A) or inhibition with HS38 ( Fig 17B) under serum-free conditions did not have a significant effect on LC 20 phosphorylation under serum-free conditions.
ROCK has been shown to phosphorylate the myosin targeting subunit of MLCP (MYPT1) at T696 and T853 (human MYPT1 numbering; GenBank ID number: 289142212) resulting in decreased phosphatase activity [46,48,[65][66][67][68][69][70]. T696 can be phosphorylated by several other protein kinases [54,69] and it appears to be constitutively phosphorylated, i.e. significantly phosphorylated in the absence of stimulation [71]. T853 is more commonly phosphorylated in response to stimulation [44,71]. ROCK can also phosphorylate Par-4 at T163, which abolishes the interaction of Par-4 with MYPT1 and exposes T696 and T853 of MYPT1 to endogenous kinases, resulting in inhibition of MLCP activity, increased LC 20 phosphorylation and contraction [72]. While this regulatory mechanism was demonstrated in cultured cells, its operation in vascular smooth muscle tissue has been questioned [73]. ZIPK (DAPK3) is a member of the death-associated protein kinase (DAPK) family, which includes DAPK1, DAPK2, DAP-related protein kinase 1 (DRAK1) and DRAK2 [74,75]. Unlike other DAPK family members, ZIPK lacks a calmodulin-binding site and its activity is independent of Ca 2+ [76]. Seven phosphorylation sites have been identified in ZIPK: T180, T225, T265, T299, T300, T306 and S311 [77]. Phosphorylation at T299, T306 and T311 has little effect on ZIPK activity, but phosphorylation at T180 (within the kinase activation loop), T225 (in the substrate-binding groove) and T265 (in kinase subdomain X) is essential for full kinase activation [45]. ROCK also phosphorylates ZIPK at multiple sites leading to its activation [45,[77][78][79], and our results provide additional evidence for the integration of ZIPK and ROCK signaling networks at the cellular level.
Arterial smooth muscle cells undergo dedifferentiation when cultured, which involves transition from a non-proliferative, contractile phenotype to a proliferative, motile phenotype [80]. These changes reflect similar changes occurring in arterial smooth muscle cells in situ that contribute to cardiovascular diseases such as atherosclerosis [81]. The major goals of this study were to characterize the time courses of serum-induced phosphorylation events relevant to contractility, migration and cytokinesis of human cultured arterial smooth muscle cells [48], with an emphasis on LC 20 diphosphorylation and the kinases responsible, directly and indirectly, for its phosphorylation. We used two experimental approaches to investigate ROCK and ZIPK as likely candidates in the diphosphorylation of LC 20 as well as MYPT1 and Par-4 phosphorylation in response to serum treatment: (i) siRNA-mediated knockdown of ROCK and ZIPK, and (ii) inhibition of kinase activity by well-characterized and selective inhibitors (GSK429286A and H1152 for ROCK and HS38 for ZIPK).
Exposure of serum-starved cells to serum resulted in the rapid phosphorylation of LC 20 at T18 and S19 (Figs 1 and 2). Similar time courses of serum-induced phosphorylation of MYPT1 at T696 and T853 (Fig 3) and of Par-4 at T163 (Fig 4) were observed, both of which result in inhibition of MLCP activity [72,82], suggesting that the rapid diphosphorylation of LC 20 is a consequence of activation of a kinase(s) that directly phosphorylates LC 20 with the concomitant inhibition of MLCP activity. Furthermore, LC 20 phosphorylation at both T18 and S19 reduces the rate of LC 20 dephosphorylation by MLCP compared to that of LC 20 phosphorylated exclusively at S19 [10]. This helps to explain the observed sustained phosphorylation of LC 20 .
The most obvious candidate kinase for the serum-induced phosphorylation of LC 20 was MLCK itself. While S19 is the site of phosphorylation responsible for smooth muscle contraction, MLCK is well known to be capable of phosphorylating T18 as well [5]. As noted in the Introduction, however, this has only been demonstrated in vitro with purified MLCK at supraphysiological concentrations relative to the LC 20 substrate [5,6] and there is plenty of evidence to support the conclusion that in situ and under physiological conditions MLCK does not phosphorylate T18 [7,8,83]. Nevertheless, we investigated the potential role of MLCK in serum-induced LC 20 phosphorylation in CASMC and found that: (i) western blot analysis confirmed the expression of smooth muscle MLCK in CASMC (Fig 7A), (ii) MLCK knockdown had no effect on LC 20 phosphorylation at T18 or S19 (Fig 7B, 7C and 7D), although we were only able to reduce MLCK levels by~45% (Fig 5), (iii) the MLCK inhibitors ML7 and wortmannin also had no effect on LC 20 phosphorylation at T18 or S19 (Fig 8), and (iv) removal of extracellular Ca 2+ and depletion of endogenous Ca 2+ stores with EGTA had no effect on serum-induced LC 20 phosphorylation at T18 or S19 (Fig 9). MLCK activity is absolutely dependent on Ca 2+ [14]. We conclude, therefore, that MLCK does not play a role in LC 20 phosphorylation at T18 or S19 in response to serum treatment of CASMC. blotting with anti-2P-LC 20 : representative western blots are shown in the upper panel with cumulative quantitative data in the lower panel. "ns" denotes p > 0.05 (p = 0.8633; n = 11; Student's unpaired t-test). (D) Effect of chelation of Ca 2+ with EGTA on LC 20 phosphorylation under serum-free conditions. CASMC were incubated in medium containing Ca 2+ (2.5 mM) or EGTA (5 mM) under serum-free conditions prior to Phos-tag SDS-PAGE and western blotting with anti-LC 20 . A representative western blot is shown in the upper panel with quantitative data depicted in the lower panel. "ns" denotes p > 0.05 (Student's unpaired t-test).
ROCK knockdown (by 73-77%: Fig 6) reduced the serum-induced diphosphorylation of LC 20 at T18 and S19 and the phosphorylation of MYPT1 at T696 and T853 and of Par-4 at T163 (Fig 11). ROCK inhibition also reduced LC 20 diphosphorylation, Par-4 phosphorylation at T163 and MYPT1 phosphorylation at T853 but, unlike the effect of ROCK knockdown, ROCK inhibition did not affect serum-induced phosphorylation of MYPT1 at T696 (Figs  12A-12D and 13A and 13B). This suggests that ROCK knockdown down-regulates a distinct kinase that is responsible for the phosphorylation of MYPT1 at T696 whose activity is unaffected by GSK429286A. Indeed, we have shown previously that ROCK knockdown affects the expression of numerous cellular proteins [33]. ZIPK knockdown (by 45-50%) had a lesser, albeit significant effect on LC 20 diphosphorylation but actually increased MYPT1 and Par-4 phosphorylation (Fig 11A, 11B and 11D). The latter observation can be explained by an increase in ROCK levels due to ZIPK knockdown ( Fig 6C). Depletion of ZIPK may also alleviate Par-4-mediated "lockdown" of MYPT1 inhibitory phosphorylation sites, thereby allowing for increased access to ROCK. ZIPK inhibition reduced the serum-induced increase in LC 20 diphosphorylation but had no effect on MYPT1 or Par-4 phosphorylation (Figs 12A, 12E and 13A and 13C).
As noted above, ZIPK activity is regulated by phosphorylation at multiple sites. Phos-tag SDS-PAGE is a very useful electrophoretic technique that separates phosphorylated forms of a protein based on the interactions of the phosphate moieties with an immobilized ligand in an SDS-polyacrylamide gel such that the higher the phosphorylation stoichiometry the slower the migration rate [35]. We have previously adapted this technique to the analysis of various phosphoproteins, e.g., MYPT1 [84], and in the current work to the phosphorylation of ZIPK. Using Phos-tag SDS-PAGE, we demonstrated that serum treatment of CASMC and UASMC induced a rapid increase in ZIPK phosphorylation at up to four sites (Fig 14). ROCK knockdown inhibited serum-induced ZIPK phosphorylation (Fig 15A), confirming ROCK-mediated phosphorylation of ZIPK originally identified by Haystead's group [78], while ZIPK knockdown reduced ZIPK levels but did not have a significant effect on ZIPK phosphorylation in response to serum treatment ( Fig 15A). Inhibitors of ROCK (GSK429286A) and ZIPK (HS38) individually reduced the serum-induced phosphorylation of ZIPK, with ROCK inhibition having a larger effect (Fig 15B and 15C). ROCK and ZIPK inhibitors together abolished seruminduced ZIPK phosphorylation (Fig 15D). We conclude that ROCK is the predominant kinase responsible for ZIPK phosphorylation in response to serum treatment with additional contribution from ZIPK autophosphorylation.
We used the same strategy combining kinase knockdown and pharmacological inhibitors to investigate the potential roles of MLCK, ROCK and ZIPK in LC 20 phosphorylation under serum-free conditions (Figs 16 and 17) and concluded that none of these kinases could account for the basal level of LC 20 phosphorylation that is observed in the absence of serum stimulation.
Our principal conclusion from these studies is that ROCK plays a major role in seruminduced phosphorylation of LC 20 at T18 and S19, phosphorylation of MYPT1 at T853 and of Par-4 at T163, whereas ZIPK plays a lesser role in serum-induced diphosphorylation of LC 20 and does not phosphorylate MYPT1 or Par-4 (Fig 18). Finally, our results indicate the importance of considering ZIPK and ROCK as integrated signaling molecules when interpreting the results of experiments that use genetic knockdown and/or pharmacological inhibition of the kinases.