Chemical Genetics of Zipper-interacting Protein Kinase Reveal Myosin Light Chain as a Bona Fide Substrate in Permeabilized Arterial Smooth Muscle*

Zipper-interacting protein kinase (ZIPK) has been implicated in Ca2+-independent smooth muscle contraction, although its specific role is unknown. The addition of ZIPK to demembranated rat caudal arterial strips induced an increase in force, which correlated with increases in LC20 and MYPT1 phosphorylation. However, because of the number of kinases capable of phosphorylating LC20 and MYPT1, it has proven difficult to identify the mechanism underlying ZIPK action. Therefore, we set out to identify bona fide ZIPK substrates using a chemical genetics method that takes advantage of ATP analogs with bulky substituents at the N6 position and an engineered ZIPK capable of utilizing such substrates. 32P-Labeled 6-phenyl-ATP and ZIPK-L93G mutant protein were added to permeabilized rat caudal arterial strips, and substrate proteins were detected by autoradiography following SDS-PAGE. Mass spectrometry identified LC20 as a direct target of ZIPK in situ for the first time. Tissues were also exposed to 6-phenyl-ATP and ZIPK-L93G in the absence of endogenous ATP, and putative ZIPK substrates were identified by Western blotting. LC20 was thereby confirmed as a direct target of ZIPK; however, no phosphorylation of MYPT1 was detected. We conclude that ZIPK is involved in the regulation of smooth muscle contraction through direct phosphorylation of LC20.

activation of myosin light chain kinase (MLCK) 6 and subsequent phosphorylation of the 20-kDa myosin regulatory light chains (LC 20 ) at Ser-19 to initiate actomyosin cross-bridge cycling and force development (1). On the other hand, relaxation is induced by dephosphorylation of LC 20 by myosin light chain phosphatase (MLCP), a type 1 protein serine/threonine phosphatase (2,3). Contraction of multiple smooth muscle tissues has frequently been observed in the absence of an increase in cytosolic free Ca 2ϩ concentration in response to a variety of stimuli (4). This phenomenon, commonly referred to as Ca 2ϩ sensitization, involves alteration of the MLCK:MLCP activity ratio in favor of the kinase, which can be achieved by the following mechanisms: (i) activation of MLCK by a mechanism not involving Ca 2ϩ /calmodulin, e.g. phosphorylation of MLCK by proline-directed kinases (5,6); (ii) an increase in Ca 2ϩ -independent LC 20 kinase activity (7); and (iii) inhibition of MLCP either directly by phosphorylation of inhibitory residues (Thr-697 and/or Thr-855) in the myosin phosphatase targeting subunit 1 (MYPT1) of MLCP (8 -10) or indirectly by phosphorylation of the protein kinase C-potentiated inhibitory protein for heterotrimeric MLCP of 17 kDa (CPI-17) at Thr-38 (11,12). The observation that intact and permeabilized smooth muscle tissues exhibit Ca 2ϩ -independent contractile responses to inhibitors of type 1 protein phosphatases (13), which correlate with diphosphorylation of LC 20 at Thr-18 and Ser-19 (14), has led to interest in the protein kinase(s) that phosphorylate these sites in a Ca 2ϩ -independent manner. Two major candidate kinases have emerged: integrin-linked kinase (ILK) and zipperinteracting protein kinase (ZIPK) (15), both of which have been shown to phosphorylate LC 20 at Thr-18 and Ser-19 in vitro (7, 16 -19). Evidence from Triton-skinned vascular smooth muscle supports a role for direct diphosphorylation of LC 20 by ILK (14,20). ILK and ZIPK also phosphorylate MYPT1 and CPI-17 in vitro, with Thr-697 being the predominant site in MYPT1 (18,(21)(22)(23) and Thr-38 in CPI-17 (16,24,25).
ZIPK is a Ser/Thr protein kinase that has been implicated in the regulation of a number of biological processes, including cell motility, apoptosis, and smooth muscle contraction (26,27). ZIPK (also known as DAPK3) is a member of the deathassociated protein kinase (DAPK) family. In contrast to other DAPKs, ZIPK does not possess a calmodulin-binding domain, and its activity is regulated independent of intracellular [Ca 2ϩ ]. The addition of constitutively active recombinant ZIPK to permeabilized smooth muscle tissue demonstrates profound Ca 2ϩ -independent contraction, which correlates with an increase in phosphorylation of both LC 20  and MYPT1 (Thr-697) (7,18,28,29). However, it is unclear from these findings whether the contractile effects are due to direct phosphorylation of LC 20 and/or inhibition of MLCP via MYPT1 and/or CPI-17 phosphorylation.
The unambiguous assignment of in vivo substrates to specific protein kinases such as ZIPK has proven a difficult task for several reasons, including: the large number of protein kinases, many of which exhibit overlapping substrate specificities; the common catalytic requirement for ATP; the organization of protein kinases into multiprotein complexes that often contain several kinases; and the difficulty in demonstrating that a given kinase has been purified free of contaminating kinases (30). The identification of kinase substrates in situ commonly involves the use of selective inhibitors of the kinase of interest. Unfortunately, selective inhibitors of ZIPK are not currently available. Small molecule inhibitors of ZIPK (ML-7, ML-9, and staurosporine) do exist (28), but they are also active against a number of other protein kinases found in smooth muscle. For example, ML-7 and ML-9 are commonly used inhibitors of MLCK (31), and staurosporine is a general kinase inhibitor (32). More recently, pseudosubstrate peptides (SM1 and AV25), based on the autoinhibitory region of MLCK, have been shown to inhibit ZIPK activity in vitro (29). We showed previously that Ca 2ϩindependent, microcystin (phosphatase inhibitor)-induced contraction of Triton-skinned rat caudal arterial smooth muscle strips is not blocked by pretreatment with AV25, suggesting that ZIPK activity is not required for the observed Ca 2ϩ -independent contraction or LC 20 diphosphorylation (17). In the case of Triton-skinned ileal smooth muscle strips, however, more recent studies have revealed contractile potentiation and possible activation of ZIPK upon the addition of the AV25 peptide (33). Unexpected variability in the effects of peptide inhibitors in situ, therefore, limits their usefulness in identifying substrates of ZIPK.
In this study, we adopted a different experimental approach to identify direct substrates of ZIPK in a smooth muscle tissue and thereby to infer physiological function. The chemical genetics approach (34 -36) employs an engineered protein kinase that can utilize as substrate (phosphoryl donor) an ATP analog that is not a substrate of the wild-type kinase. The addition of such an engineered ZIPK in the presence of the [␥-32 P]ATP analog to rat caudal arterial smooth muscle will then allow the identification of direct substrates of ZIPK, thereby providing insights into the mechanisms by which ZIPK promotes increased contractile force. The data presented herein identify LC 20 , but not MYPT1, as a direct substrate of ZIPK in smooth muscle. Furthermore, ZIPK can affect the phosphorylation of MYPT1 by another kinase(s) via a mechanism that does not require ZIPK activity.
Homology Modeling-A model of the tertiary structure of ZIPK 7 in complex with ATP was generated by homology modeling with SWISS-MODEL (38) using DAPK1 in complex with ATP as a backbone structure. The structural coordinates for DAPK1 8 were downloaded from the PDB database (PDB ID: 3F5G) (39), and Cn3D v4.1 software was used for the structural rendering.
Step one involved transfer of the ␥-32 P-labeled phosphoryl group from ATP to NDPK in a reaction mixture (150 l) containing 200 units of NDPK, 800 Ci of [␥-32 P]ATP, 150 mM NaCl, 5 mM MgCl 2 , and 20 mM HEPES, pH 7.4. The reaction mixture was incubated for 5 min at 30°C after which 32 P-labeled NDPK was purified from ADP and the remaining [␥-32 P]ATP by desalting through Microspin G50 columns.
Step two involved transfer of the radiolabel from NDPK to N 6 -phenyl-ADP. This reaction was performed by the addition of 1 nmol (10 l) of N 6 -phenyl-ADP to the 32 P-labeled NDPK followed by incubation for 20 min at 30°C. As the final step, the [␥-32 P]N 6 -phenyl-ATP was separated from NDPK on a Micron-YM10 column.
Tissue Preparation and Force Measurements-Caudal arteries were removed from male Sprague-Dawley rats (300 -350 g) that had been anesthetized and euthanized according to protocols approved by the University of Calgary Animal Care and Use Committee. The arteries were cleaned of excess adventitia and adipose tissue, cut into helical strips (1.5 ϫ 6 mm), and demembranated (skinned) with 1% (v/v) Triton-X-100.
Smooth muscle strips were mounted on a Grass isometric force transducer (FT03C), and force was recorded as described previously (17). The Ca 2ϩ -free solution (pCa 9) contained: 4 mM K 2 EGTA, 5.83 mM MgCl 2 , 75.6 mM potassium propionate, 3.9 mM Na 2 ATP, 16.2 mM phosphocreatine, 30 units/ml creatine kinase, and 20 mM TES, pH 6.9. Experimental tissues were treated with recombinant ZIPK (10 M) in pCa 9 solution for 1 h. For experiments designed to identify direct substrates of ZIPK, tissue baths also contained 100 Ci of [␥-32 P]N 6 -phenyl-ATP. For experiments requiring the removal of endogenous ATP, muscle strips were washed three times for 5 min each with pCa 9 solution containing 0 ATP and then supplemented with 4 mM N 6 -phenyl-ATP and incubated for 1 h in the presence of ZIPK-L93G (10 M). At the end of contractions, the muscle strips were flash-frozen in 10% trichloroacetic acid/10 mM DTT/acetone followed by three 10-s washes in 10 mM DTT/ acetone. Tissues were then lyophilized overnight.
For analysis of LC 20 phosphorylation, samples were resolved by Phos-tag TM SDS-PAGE as described previously (41,42). After electrophoresis, proteins were transferred to polyvinylidene difluoride (PVDF) membranes at 25 V for 16 h at 4°C. Proteins were fixed on the membrane with 0.5% glutaraldehyde in phosphate-buffered saline (136.9 mM NaCl, 2.7 mM KCl, 10 mM Na 2 HPO 4 , and 1.76 mM KH 2 PO 4 ) for 20 min and then washed with TBST. Nonspecific binding sites were blocked with 5% (w/v) nonfat dry milk in TBST. Membranes were washed with TBST and incubated overnight with anti-LC 20 at a 1:500 dilution in 1% (w/v) nonfat dry milk in TBST. Membranes were incubated for 1 h with HRP-conjugated secondary antibody (dilution 1:10,000) and developed with ECL reagent.
HPLC Fractionation and Phosphoamino Acid Analysis of MYPT1 Phosphopeptides-Phosphorylated MYPT1 was electrophoresed in a 10% SDS-PAGE at 200 V for 45 min. [ 32 P]MYPT1 was visualized by Coomassie Blue staining and autoradiography. Gel slices containing [ 32 P]MYPT1 were excised from the gel and washed with 10% acetic acid/50% methanol. Trypsin (0.5 g in 50 mM ammonium bicarbonate, 100 l) was added to the gel slice, and the mixture was incu-bated at 37°C overnight. The solution was recovered, and at least 75% of the radioactivity was recovered. For fractionation of [ 32 P]MYPT1 phosphopeptides by HPLC, tryptic digests were acidified with trifluoroacetic acid (0.5% v/v) and applied to a reverse-phase C18 column (Acclaim 300, 3 m, 4.6 ϫ 250 mm) equilibrated in 0.1% trifluoroacetic acid. The column was washed with 0.1% trifluoroacetic acid, and peptides were eluted with a linear gradient of acetonitrile (0 -60% (v/v) over 100 min, 0.5 ml/min). Fractions (0.5 ml each) were collected and phosphopeptides identified by Č erenkov radiation. Fractions containing the major peaks of radioactivity were subjected to phosphoamino acid analysis. Lyophilized tryptic peptides were resuspended in 6 M HCl (250 l), incubated for 2 h at 110°C, dried, and resuspended in 5-10 l of electrophoresis buffer (water:acetic acid:pyridine at 94.875:5:0.125) containing phospho-Ser, phospho-Thr, and phospho-Tyr standards (40 ng of each). Samples were spotted onto cellulose thin-layer plates (10 ϫ 20 cm). Amino acids were electrophoresed at 1200 V at 10°C in electrophoresis buffer for 35 min, and the plates were dried for 10 min in a fume hood and baked at 110°C for 10 min. Amino acid standards were visualized by spraying the plates with 0.25% ninhydrin dissolved in acetone and baking at 110°C for 15 min. 32 P-Labeled amino acids were visualized by autoradiography.
Data Analysis-Values are presented as the mean Ϯ S.E., with n indicating the number of animals (tissue experiments) or the number of different preparations of ZIPK (in vitro kinase assays). Data were analyzed by Student's t test, with p Ͻ 0.05 considered to indicate statistically significant differences.

RESULTS
To assess ZIPK-dependent changes in the phosphorylation status of LC 20 and MYPT1, we examined the phosphorylation of these potential substrates in response to the addition of wildtype and kinase-dead ZIPK to Triton-skinned rat caudal arterial smooth muscle strips under Ca 2ϩ -free conditions (pCa 9). Control Ca 2ϩ -induced contractions were first elicited by the replacement of pCa 9 solution with pCa 4.5 solution. As demonstrated previously (e.g. see Ref. 17), the Triton-skinned tissue contracted in response to the increase in [Ca 2ϩ ] and relaxed upon return to Ca 2ϩ -free solution (Fig. 1A). This contraction is due exclusively to LC 20 phosphorylation at Ser-19 catalyzed by MLCK (Fig. 1B) as demonstrated previously (e.g. 14). Application of constitutively active ZIPK-WT (10 M) (Fig. 1A, trace a), but not the kinase-dead ZIPK-KD (10 M) (Fig. 1A, trace b), elicited a Ca 2ϩ -independent contraction. As shown previously (14), microcystin, an inhibitor of type 1 and 2A protein Ser/Thr phosphatases, elicited a sustained contraction under Ca 2ϩ -free conditions (Fig. 1A, trace c), which correlated with a significant increase in LC 20 diphosphorylation (Fig. 1B). The contractile response to ZIPK-WT (Fig. 1A, trace a) was accompanied by an increase in the phosphorylation of LC 20 at Thr-18 and Ser-19 (Fig. 1B). LC 20 monophosphorylation increased from 0 to 10.3 Ϯ 2.6% of total LC 20 , and the amount of diphosphorylation increased from 0 to 24.4 Ϯ 9.2% of total LC 20 (Fig. 1B). Interestingly, ZIPK-KD also induced mono-and diphosphorylation of LC 20 , albeit to lower levels than observed with ZIPK-WT (7.8 Ϯ 4.2% of total LC 20 for monophosphorylated LC 20 and 1.9 Ϯ 0.9% of total LC 20 for diphosphorylated LC 20 ). This level of phosphorylation was clearly insufficient to evoke a contractile response (Fig. 1A, trace b).
Small increases in MYPT1 phosphorylation were also detected at the known inhibitory sites, Thr-697 (Fig. 1C) and Thr-855 (Fig. 1D), in response to ZIPK-WT or ZIPK-KD. ZIPK-WT-induced MYPT1 phosphorylation at Thr-697 and Thr-855 increased 2.4 Ϯ 0.7-fold and 2.1 Ϯ 0.6-fold, respectively, compared with the pCa 9 control. Tissues treated with ZIPK-KD exhibited similar increases in MYPT1 phosphorylation (2.4 Ϯ 0.7-fold and 1.6 Ϯ 0.4-fold increases for the Thr-697 and Thr-855 sites, respectively). Much higher levels of MYPT1 phosphorylation at both Thr-697 and Thr-855 were measured in response to microcystin treatment (Fig. 1, C and D). We concluded from these experiments that: (i) ZIPK-WT induces an increase in phosphorylation of LC 20 at Thr-18 and Ser-19 and of MYPT1 at Thr-697 and Thr-855; (ii) ZIPK-KD induces a smaller increase in LC 20 diphosphorylation but similar increases in MYPT1 phosphorylation at Thr-697 and Thr-855; and (iii) the overall increase in LC 20 phosphorylation in response to ZIPK-WT, but not ZIPK-KD, is sufficient to evoke a contractile response. In vitro kinase assays confirmed that ZIPK-KD is devoid of catalytic activity (supplemental Fig. 1), suggesting that this mutant kinase can affect the activity of another distinct kinase that is capable of phosphorylating LC 20 and MYPT1, e.g. ILK.
To further elucidate the mechanism by which the addition of ZIPK-KD causes an increase in MYPT1 phosphorylation at both inhibitory sites, ZIPK-KD was added to the tissue in the presence of a nonselective kinase inhibitor (staurosporine), a Rho-associated kinase inhibitor (Y-27632), or a protein kinase C inhibitor (GF-109203X). Phosphorylation at the Thr-697 site, when compared with control tissue ( Fig. 2A), was significantly decreased following treatment with staurosporine (decreased to 18.4 Ϯ 1.3% of control); however, Y-27632 and GF-109203X had no effect (107.4 Ϯ 4.6 and 98.2 Ϯ 5.7%, respectively). Phosphorylation at the Thr-855 site (Fig. 2B) was significantly decreased by treatment with staurosporine (decreased to 9.6 Ϯ 0.8% of control), Y-27632 (decreased to 15.7 Ϯ 3.0%), and GF-109203X (decreased to 42.8 Ϯ 2.9%). Taken together, these findings suggest that ZIPK interaction with the contractile filament allows for kinase-dependent phosphorylation of Thr-697 and Thr-855. The results suggest that Thr-855 phosphorylation is a result of Rho-associated kinase (ROK) and/or protein kinase C (PKC) activities and that Thr-697 phosphorylation occurs through another unidentified kinase, most likely ILK, which is resistant to Y-27632 and GF-109203X but sensitive to staurosporine (17).
Prior to this study, phosphorylation of MYPT1 at Thr-855 by ZIPK had not been examined. We sought to determine whether the Thr-855 site was a target of ZIPK in vitro. Indeed, ZIPK was able to catalyze phosphorylation of synthetic peptides corresponding to either the Thr-697 (Fig. 3A) or Thr-855 (Fig. 3B) phosphorylation sites of MYPT1. K m values were determined from Lineweaver-Burk secondary plots to be 21.0 Ϯ 5.6 M (Thr-697) and 13.1 Ϯ 5.4 M (Thr-855). HPLC analysis of tryptic digests of phosphorylated ( 32 P-labeled) wild-type and mutant MYPT1 proteins (Fig. 3C) showed a decrease in phosphorylation of MYPT1 mutants S696A/T697A and S854A/ T855A relative to the wild type, confirming that phosphorylation of MYPT1 does occur within these sites of MYPT1. To determine whether phosphorylation occurred on the Ser or Thr residues, phosphoamino acid analysis was completed (Fig. 3D) and showed phosphorylation of both Ser and Thr in WT, S696A/T697A, and S854A/T855A, whereas only Ser residues were phosphorylated in T697A/T855A and only Thr residues in S696A/S854A. Taken together, these data provide the first in vitro evidence that ZIPK can phosphorylate not only Thr-697 of MYPT1 but can also target Ser-696, Ser-854, and Thr-855.
To identify the direct substrates of ZIPK, we employed the chemical genetics approach of Shokat and colleagues (34 -36). This method involves engineering the ATP-binding pocket of the protein kinase to accept ATP analogs that are not utilized as substrates by the wild-type kinase. This strategy has been used successfully to examine the function of a variety of protein kinases, including but not limited to v-Src (43,44), ERK2 (40), JNK (45), Pho85 (a yeast cyclin-dependent kinase) (36), Cdk1 (Cdc28) (46), PKA (47), and Raf1 (30). A sequence comparison of ZIPK and DAPK1 with other Ser/Thr kinases examined previously by this approach (ERK2 (NCBI NP_002736), Cdk2 (GenBank CAA43985), and PKA (NCBI NP_002721)) was performed (Fig. 4A). The structure of ZIPK in complex with ATP has not yet been solved; however, protein kinases possess highly conserved catalytic domains, and the nearest relative of ZIPK, DAPK1, exhibits Ͼ90% sequence conservation in this domain. Fortunately, the structure of DAPK1 in complex with ATP has been published (PDB ID: 3F5G), and this was aligned with a structural solution of ZIPK in complex with pyridone (PDB ID: 3BQR) to reveal the conserved "gatekeeping" residues in the ATP-binding pocket (Fig. 4B). Furthermore, a homology model of ZIPK (Fig. 4C) was prepared to identify residues within the ATP-binding pocket of ZIPK that would be suitable for mutation to accommodate ATP analogs with bulky substituents at the N 6 position. Three amino acid residues within the ATPbinding pocket of ZIPK were identified as candidates for mutation: Leu-68, Ile-77, and Leu-93 (Fig. 4D). From this in silico analysis, the Ile-77 and Leu-93 residues in ZIPK were found to correspond to residues selected previously for mutation in both Cdk2 and PKA (Fig. 4A). Single, double, and triple mutations were generated at these locations, and ZIPK-L68G, -I77G, -L93G, -L68G/I77G, -L68G/L93G, -I77G/L93G, and -L68G/ I77G/L93G proteins were expressed and purified.
Initially, we sought to determine whether the expressed mutant proteins retained kinase activity. Kinase assays showed that ZIPK-L93G retained the highest kinase activity (data not shown). Then, using ZIPK-L93G, we determined whether commercially available ATP analogs (N 6 -benzyl-ATP, N 6 -1-methylbutyl-ATP, N 6 -2-methylbutyl-ATP, N 6 -2-phenylethyl-ATP, and N 6 -phenyl-ATP) (Fig. 5A) could occupy the ATP-binding pocket. To address this issue, ZIPK activity assays were performed to assess the ability of the various ATP analogs (at 50, 100, and 400 M) to competitively inhibit the phosphorylation of a known ZIPK substrate (the peptide KKRQS-RRSTQGVTLT, corresponding to the region around Thr-697 of MYPT1 with the addition of N-terminal Lys residues) using [␥-32 P]ATP as the phosphoryl group donor (23). The data revealed that ZIPK-WT was unaffected by the addition of any of the ATP analogs (50 M), confirming the inability of the wildtype kinase to bind these analogs (Fig. 5B). However, phosphorylation of the MYPT1 peptide by ZIPK-L93G was blocked by all ATP analogs except N 6 -1-methylbutyl-ATP (Fig. 5C). In the absence of ATP analog, ZIPK-L93G phosphorylated the MYPT1 peptide with a specific activity of 26.4 pmol of P i incorporated/min/g of ZIPK. However, in the presence of 50 M ATP analog, the extent of MYPT1 peptide phosphorylation catalyzed by ZIPK-L93G decreased to specific activities of 5.7 Ϯ 0.4 pmol/min/g (N 6 -benzyl-ATP), 4.8 Ϯ 0.8 pmol/min/g (N 6 -2-methylbutyl-ATP), 1.8 Ϯ 0.1 pmol/min/g (N 6 -2-phenylethyl-ATP), and 7.0 Ϯ 1.3 pmol/min/g (N 6 -phenyl-ATP), whereas phosphorylation was unaffected (32.7 Ϯ 4.9 pmol/ min/g) in the presence of N 6 -1-methylbutyl-ATP. These data suggest that N 6 -benzyl-ATP, N 6 -2-methylbutyl-ATP, N 6 -2phenylethyl-ATP, and N 6 -phenyl-ATP can occupy the ATPbinding pocket of ZIPK-L93G, whereas N 6 -1-methylbutyl-ATP cannot. Structural modeling of ZIPK with the various ATP analogs confirmed that N 6 -1-methylbutyl-ATP was the only analog tested that did not correctly orientate into the ATP-binding pocket (data not shown). In the four cases in which ZIPK-L93G activity was inhibited by the addition of ATP analog, structural analysis revealed that the added substituent on the N 6 position of ATP orientates into the space occupied previously by Leu-93. However, in the case of 1-methylbutyl-ATP, the 1-methylbutyl group is directed toward residues Ile-77 and Glu-94 and therefore is unable to occupy the newly engineered binding pocket of ZIPK-L93G.
To determine whether ZIPK-L93G could utilize the various ATP analogs in a phosphotransfer reaction, kinase assays were performed using ZIPK, purified LC 20 protein and ATP analog in the absence of ATP. Reaction mixtures were examined by Phos-tag SDS-PAGE, which allows the separation of un-, mono-, and diphosphorylated LC 20 (Fig. 5D). The results indicate that ZIPK-L93G can utilize each of the ATP analogs to phosphorylate LC 20 . However, the phosphorylation profile of ZIPK-L93G with N 6 -phenyl-ATP most closely resembled ZIPK-WT with ATP in that diphosphorylation was observed. Therefore, N 6 -phenyl-ATP appears to function as the most suitable phosphoryl donor with the engineered ZIPK mutant. To determine whether the L93G mutation had any effect on ZIPK substrate specificity, ATP binding, or catalytic efficiency, we measured the K m , V max , k cat , and specificity constant (k cat / K m ) of ZIPK-WT and ZIPK-L93G with ATP or N 6 -phenyl-ATP and LC 20 protein substrates (Table 1). ZIPK-WT and ZIPK-L93G displayed similar K m values for ATP (11.2 Ϯ 1.4 and 9.4 Ϯ 0.9 M, respectively), indicating that mutation within the binding pocket of ZIPK did not affect the affinity for ATP. Mutation did, however, increase the V max (from 36.4 Ϯ 0.6 to 88.0 Ϯ 0.4 nmol of P i incorporated/min/mg), the number of substrate molecules turned over per enzyme molecule/min (k cat ) (from 1.2 Ϯ 0.02 to 4.5 Ϯ 0.04 min Ϫ1 ), and the enzyme efficiency (k cat /K m ) (from 0.10 to 0.48 min Ϫ1 /M). Under the same conditions, ZIPK-WT could not use N 6 -phenyl-ATP as a substrate, as expected. Because the structural topologies of ATP-binding pockets of different Ser/Thr kinases are almost indistinguishable, it is likely that other endogenous kinases will not be able to utilize N 6 -phenyl-ATP either. The K m of ZIPK-L93G for N 6 -phenyl-ATP (16.1 Ϯ 0.7 M) was slightly higher than that for ATP (9.4 Ϯ 0.9 M). Also increased were the V max (241 Ϯ 4 nmol of P i incorporated/min/mg), k cat (11.4 Ϯ 0.6 min Ϫ1 ), and k cat /K m (0.71 min Ϫ1 /M) values. These data suggest that the L93G mutation introduced little steric hindrance within the ATP-binding pocket, which in turn did not affect the binding efficacy for ATP per se. At saturating concentrations of ATP, ZIPK-WT and ZIPK-L93G displayed similar kinetic properties: K m for LC 20 protein substrate (0.42 Ϯ 0.07 and 0.51 Ϯ 0.05 M, respectively), V max (1060 Ϯ 47 and 1320 Ϯ 34 nmol of P i incorporated/ min/mg, respectively), k cat (0.38 Ϯ 0.02 and 0.30 Ϯ 0.06 min Ϫ1 , respectively), and k cat /K m (0.90 and 0.59 min Ϫ1 /M, respectively). These results confirm that mutation of Leu-93 to Gly within the ATP-binding pocket of ZIPK did not affect the protein/peptide-binding site of the kinase. As a final validation that the L93G mutation did not affect ZIPK substrate specificity, in vitro kinase assays were carried out with purified ZIPK-WT, ZIPK-L93G, or ZIPK-KD, known in vitro substrates of ZIPK (LC 20 , MYPT1, and CPI-17), and [␥-32 P]ATP or [␥-32 P]N 6 -phenyl-ATP. The total protein load was visualized by Coomassie Brilliant Blue staining following SDS-PAGE, and 32 P incorporation was identified by autoradiography. All three substrate proteins, LC 20 , MYPT1, and CPI-17 (Fig. 6), were phosphorylated by ZIPK-WT with ATP, ZIPK-L93G with ATP, and ZIPK-L93G with N 6 -phenyl-ATP. As shown in earlier experiments, ZIPK-WT with N 6 -phenyl-ATP, ZIPK-KD with ATP, and ZIPK-KD with N 6 -phenyl-ATP were unable to phosphorylate any of the substrates. In addition, ZIPK-L93G was able to effectively phosphorylate isolated whole smooth muscle myosin (supplemental Fig. 2). For quantification, an equal volume of each reaction mixture was removed at selected times of incubation and spotted onto P81 paper, and 32 P incorporation (experimental minus no ZIPK control) was determined from scintillation counting (Fig. 6A). Taken together, these assays provided further evidence that ZIPK substrate specificity is unaffected by the L93G mutation.
To identify bona fide targets of ZIPK, we examined protein phosphorylation during contraction of Triton-skinned rat caudal arterial smooth muscle strips induced under Ca 2ϩ -free conditions by the addition of exogenous ZIPK-L93G in the presence of ATP and [␥-32 P]N 6 -phenyl-ATP. The application of ZIPK-L93G (10 M) induced a Ca 2ϩ -independent contraction (Fig. 7A, trace a). Treatment with [␥-32 P]N 6 -phenyl-ATP in pCa 9 solution without ZIPK did not alter the contractile state of the tissue (Fig. 7A, trace b). Direct substrates of ZIPK, identified by an increase in 32 P incorporation in ZIPK-L93G-treated tissue (Fig. 7A, trace a) compared with control (Fig. 7A, trace b), were detected by 7.5-20% gradient SDS-PAGE and autoradiography. Three direct targets of ZIPK, proteins of 17, 20, and 40 kDa, were identified (Fig. 7B). To verify that endogenous kinases could not utilize [␥-32 P]N 6 -phenyl-ATP, tissues were treated with Ca 2ϩ (pCa 4.5) or microcystin (1 M) in the presence of [␥-32 P]N 6 -phenyl-ATP. No phosphorylation occurred in response to pCa 4.5 or microcystin, indicating that MLCK and a basally active kinase (e.g. ROK or ILK) cannot utilize [␥-32 P]N 6 -phenyl-ATP. To confirm that the 20-kDa substrate of ZIPK-L93G in the [␥-32 P]N 6 -phenyl-ATP-treated muscle strips was indeed LC 20 , proteins were extracted from the tissue and separated by Phos-tag SDS-PAGE (Fig. 7C). Autoradiography revealed two 32 P-labeled bands, which were identified as monoand diphosphorylated LC 20 by Western blotting.
Identical experiments were carried out with the ZIPK-WT and [␥-32 P]ATP pairing, and autoradiography revealed only LC 20 phosphorylation. 32 P-Labeled bands were not detected at 17 or 40 kDa (supplemental Fig. 3). As indicated by the kinetic data (Table 1), ZIPK-L93G with N 6 -phenyl-ATP is more catalytically efficient (i.e. has a larger k cat /K m value) than ZIPK-WT with ATP. The reason why these minor substrates were detected in the experiments of Fig. 7 is most likely because of the kinetic differences between ZIPK-WT and ZIPK-L93G. Attempts to identify the two minor substrates of ZIPK (17 and 40 kDa) by mass spectrometry were unsuccessful. We concluded, therefore, that LC 20 is the major substrate of ZIPK in rat caudal arterial smooth muscle during Ca 2ϩ -independent contraction.
MYPT1 was not identified as a direct substrate of ZIPK in the experiments of Fig. 7 and supplemental Fig. 3. To rule out the possibility that the method employed did not provide sufficient sensitivity to detect MYPT1 phosphorylation, we developed an alternative approach to examine whether MYPT1 was a direct target of ZIPK. Following a Ca 2ϩ -induced contraction of Triton-skinned rat caudal arterial smooth muscle strips to verify viability, ATP was removed by a series of incubations in Ca 2ϩfree solution devoid of ATP. The successful removal of ATP was confirmed by the failure of the addition of Ca 2ϩ in the presence of N 6 -phenyl-ATP to elicit a contraction (Fig. 8A,  trace a). Washout of N 6 -phenyl-ATP and replacement of ATP restored the ability of the tissue to contract in response to Ca 2ϩ . If ATP was maintained throughout the protocol, Ca 2ϩ -induced contractions could be evoked consistently (Fig. 8A, trace b). The addition of N 6 -phenyl-ATP (4 mM) and ZIPK-L93G (Fig.  8B, trace a) allowed direct targets of ZIPK to be examined by Western blotting with phosphospecific antibodies. The N 6 -phenyl-ATP could act as a selective donor for phosphorylation events driven by ZIPK-L93G but not by endogenous kinases. Under these conditions, we predicted that no contractile force would be generated because myosin II cannot utilize N 6 -phenyl-ATP as a substrate, i.e. activated myosin cannot hydrolyze N 6 -phenyl-ATP to provide the energy required for cross-bridge cycling. Control experiments utilized identical conditions but with the omission of ZIPK-L93G (Fig. 8B, trace  b). These experiments provided additional evidence that ZIPK directly phosphorylates LC 20 , as Western blot analysis confirmed an increase in both mono-and diphosphorylated species (Fig. 8C). The amount of monophosphorylation increased from 0 to 6.0 Ϯ 1.8% of total LC 20 , and the amount of diphosphorylation increased from 0 to 8.3 Ϯ 2.8% of total LC 20 . No signifi-

TABLE 1 Kinetic properties of purified ZIPK-WT and ZIPK-L93G
Kinetic constants (K m , V max , k cat , and k cat /K m ) were determined for purified, recombinant ZIPK-WT and ZIPK-L93G. The results (mean Ϯ S.E.) are representative of three independent ZIPK preparations with assays completed in triplicate. ND, not determined due to lack of enzymatic activity. cant changes in phosphorylation of MYPT1 at the inhibitory sites, Thr-697 (Fig. 8D) and Thr-855 (Fig. 8E), were detected, supporting the conclusion that MYPT1 is not a direct substrate of ZIPK.

DISCUSSION
Ca 2ϩ -independent protein kinases can contribute directly to LC 20 mono-and diphosphorylation and are thought to play a role in smooth muscle Ca 2ϩ sensitization. Previous reports have implicated several Ca 2ϩ -independent kinases, including ROK, MAPKAPK-2, MAPKAPK-1b (RSK-2), ILK, and ZIPK, in the phosphorylation of LC 20 in vitro (discussed in Ref. 17). Current evidence, however, suggests that ZIPK, ILK, or a combination of the two kinases is responsible for the Ca 2ϩ -independent LC 20 phosphorylation observed in vascular smooth muscle during Ca 2ϩ sensitization (17), although direct evidence is still required. As a potential regulator of smooth muscle contraction, ZIPK has been observed to phosphorylate LC 20 in vitro (7,18,19). Moreover, the addition of constitutively active recombinant ZIPK to mesenteric or ileal smooth muscle (7,18,28) and caudal artery (this study (Fig. 1)) induced an increase in diphosphorylation of LC 20 at Thr-18 and Ser-19. ZIPK was found associated with MLCP and could inhibit its activity by phosphorylation of MYPT1 at Thr-697 (18,22,23). ZIPK may also regulate MLCP activity indirectly, because it can phosphorylate CPI-17 at Thr-38 in vitro (25). However, it is unclear from these data whether the increase in phosphorylated LC 20 is due to direct phosphorylation of LC 20 by ZIPK or via an indirect mechanism involving phosphorylation of MYPT1 or CPI-17 and subsequent inhibition of MLCP.
This study was designed to examine whether ZIPK is directly responsible for the phosphorylation of LC 20 and/or MYPT1 that has been observed previously upon the addition of constitutively active recombinant ZIPK protein to permeabilized tissue (7,18,23,28,29,33) or isolated cells (48). In a general sense, the identification of cognate substrates for protein kinases has been hampered by problems with the unambiguous assignment of substrates to a particular kinase in situ and in vivo. The traditional profiling of protein kinase substrate pools has relied heavily on the use of pharmacological inhibitors that block the phosphoryl transferase activity of the kinase, dominant-negative expression models, and/or genetic deletion models. Unfortunately, the interpretation of pharmacological or genetic knock-out phenotypes for ZIPK is difficult because of functional redundancies and/or alterations in signaling constancy (15). The ability of protein kinases with overlapping substrate specificities to compensate for one another is an important limiting factor. Specific small molecule inhibitors of ZIPK have not been developed and validated in smooth muscle tissues. Although two closely related pseudosubstrate peptide inhibitors of ZIPK (SM1 and AV25) have been shown to selectively inhibit ZIPK under Ca 2ϩ -free conditions (17,29), confounding results obtained from their use in situ (33) have hindered their broad application in defining ZIPK function. Because of these challenges, we decided to employ the chemical genetics approach pioneered by Shokat and colleagues (34 -36) with the aim of engineering a ZIPK that could utilize specific ATP analogs while preserving its ability to faithfully phosphorylate its cognate substrate pool.
Our initial analysis of several engineered ZIPKs containing mutations in key gatekeeping residues at the mouth of the ATP-binding pocket, in combination with several ATP analogs containing bulky substituents at the N 6 position, provided biochemical confirmation of several key enzymatic properties: (i) ZIPK-L93G retains the kinetic and substrate specificity proper-ties of the wild-type kinase; (ii) this mutant is capable of binding selected ATP analogs with high affinity; (iii) ZIPK-L93G can utilize the ATP analogs to phosphorylate a protein substrate; (iv) this mutant has an unaltered affinity for the protein substrate; (v) N 6 -phenyl-ATP is the most suitable ATP analog of those tested; and (vi) wild-type ZIPK and other endogenous protein kinases are unable to utilize the N 6 -phenyl-ATP analog as a substrate for phosphoryl transfer. With these requirements achieved, the pairing of ZIPK-L93G and N 6 -phenyl-ATP was used in situ to identify direct substrates of ZIPK in Tritonskinned vascular smooth muscle tissue. Importantly, our study is the first to provide evidence that ZIPK can induce Ca 2ϩindependent contraction through its ability to directly phosphorylate LC 20 at Ser-19 and Thr-18.
Although ZIPK-L93G can be expressed transiently in isolated tissue or dispersed cells, the need to provide the ATP analog to the system requires permeabilization (or Triton skinning) and is a significant limitation of the method. Important smooth muscle components not tightly bound to the contractile filaments may be lost. For example, in the case of rat caudal artery, we have demonstrated that CPI-17 is lost from the muscle strip during skinning (49). Furthermore, because of the effects of the detergent on membrane structure, Triton skinning eliminates the response to agonists that act via membranebound receptors. In addition, it is unlikely that the ATP analog could be regenerated by creatine kinase once hydrolyzed, in which case the ATP analog concentration may become limiting in the tissue over time. However, the fact that N 6 -Phe-ATP is not hydrolyzed by the actomyosin ATPase and is, therefore, used only as a substrate of the kinase makes this unlikely. These caveats notwithstanding, the approach does offer significant advantages in that the direct substrate(s) of a particular protein kinase can be identified. In the future it may be possible to induce the expression of ZIPK-L93G in tissue and provide a cell-permeable, small molecule inhibitor engineered to complement the mutation introduced into ZIPK, as accomplished previously with other protein kinases (50,51). MLCP activity is regulated by phosphorylation of the Thr-697 inhibitory site of MYPT1 (discussed in Refs. 2 and 3), and several protein kinases, including ROK, ZIPK, myotonic dystrophy protein kinase, ILK, Raf-1, and p21-activated protein kinase, are known to direct this phosphorylation. In addition, another inhibitory residue of MYPT1, Thr-855, is thought to be a specific target of ROK (3). Previous studies have shown an increase in phosphorylation of MYPT1 at Thr-697 when permeabilized ileal smooth muscle was incubated with ZIPK-WT (18,28,29), but phosphorylation at Thr-855 was not examined. Our in vitro studies (Fig. 3) demonstrated that ZIPK-WT was capable of phosphorylating both inhibitory sites, Thr-697 and Thr-855. Additionally, site-directed mutations of MYPT1 (Thr to Ala) were able to decrease phosphorylation relative to WT-MYPT1, further confirming the ability of ZIPK to phosphorylate MYPT1 in vitro at both inhibitory sites. When ZIPK-WT was provided to skinned rat caudal arterial smooth muscle strips, an increase in MYPT1 phosphorylation at both Thr-697 and Thr-855 was observed. Previous studies did not examine MYPT1 phosphorylation changes in response to the addition of ZIPK-KD; it was assumed that, because this mutant is devoid of FIGURE 7. Identification of direct substrates of ZIPK in Triton-skinned rat caudal arterial smooth muscle. A, following control Ca 2ϩ -induced contraction-relaxation cycles, Triton-skinned rat caudal arterial smooth muscle strips were treated with [␥-32 P]N 6 -phenyl-ATP (prepared as described under "Experimental Procedures") with (a) or without (b) ZIPK-L93G (10 M), both in the presence of ATP. B, at the end of the incubation times in A, tissues were quick-frozen, proteins were separated by SDS-PAGE (7.5-20% polyacrylamide gradient gels) and visualized with Coomassie Brilliant Blue staining (CBB), and 32 P-labeled proteins were detected by autoradiography (ARG). The results are representative of five independent experiments. MC, microcystin. C, LC 20 phosphorylation was analyzed by Phos-tag SDS-PAGE and Western blotting with anti-LC 20 (IB). 32 P incorporation into LC 20 was detected by autoradiography. The results are representative of four independent experiments. kinase activity, there would be no phosphorylation. However, we made the unexpected observation in the present study that the addition of this kinase-dead mutant of ZIPK could, in fact, elicit increased MYPT1 phosphorylation at Thr-697 and Thr-855, as well as mono-and diphosphorylation of LC 20 in situ, but to levels insufficient to elicit a contractile response. Furthermore, use of the ZIPK-L93G/N 6 -phenyl-ATP pair demonstrated that MYPT1 was not a direct substrate of ZIPK in situ. We interpreted these results as follows: ZIPK-KD, through protein-protein interaction, unmasks the activity of another kinase, likely ILK, which phosphorylates MYPT1 at Thr-697 and Thr-855, and as a consequence of MYPT1 phosphorylation and inhibition, LC 20 phosphorylation at Thr-18 and Ser-19 is augmented. In this case, however, the level of LC 20 phosphorylation achieved is insufficient to elicit a contraction.
The addition of exogenous ZIPK and its subsequent association with the contractile filament might induce changes within the contractile architecture to allow the phosphorylation sites of MYPT1 to become more accessible to other kinases. Indeed, recent investigations by Vetterkind et al. (52,53) have revealed a unique regulation of ZIPK in vascular smooth muscle that involves the prostate apoptosis response 4 (Par-4) protein and have proposed a model whereby Par-4 binding to MYPT1 blocks access to the inhibitory phosphorylation sites (53). It is possible that the addition of ZIPK to caudal arterial smooth muscle may alter this Par-4-MYPT1 interaction. We have identified that pretreatment with staurosporine, a nonselective kinase inhibitor, prior to the addition of ZIPK-KD, abolished the increase in MYPT1 phosphorylation at both inhibitory sites and confirmed a role for another protein kinase in this phenom-FIGURE 8. Effect of ZIPK-L93G in the presence of N 6 -phenyl-ATP on LC 20 and MYPT1 phosphorylation in Triton-skinned rat caudal arterial smooth muscle. A, trace a, following an initial control Ca 2ϩ -induced contraction-relaxation cycle, ATP was washed out of the system. The effective removal of ATP was verified by the lack of a contractile response to the addition of N 6 -phenyl-ATP at pCa 4.5. Subsequent replenishment with ATP restored Ca 2ϩ -induced contraction. Trace b, repeated Ca 2ϩ -induced contractions were recorded when ATP was maintained throughout the experiment. B, following initial control Ca 2ϩ -induced contraction-relaxation cycles, ATP was washed out of the system. N 6 -phenyl-ATP (4 mM) was then added in the presence (trace a) or absence (trace b) of ZIPK-L93G (10 M) at pCa 9. Tissues were then quick-frozen for analysis of LC 20 phosphorylation by Phos-tag SDS-PAGE and Western blotting with anti-LC 20 (IB), and band intensities were quantified by scanning densitometry (C). Values indicate mean Ϯ S.E. (n ϭ 6). MYPT1 phosphorylation was analyzed by Western blotting with phosphospecific antibodies, anti-phospho-Thr-697-MYPT1 (D), and anti-phospho-Thr-855-MYPT1 (E). MYPT1 phosphorylation levels were normalized to actin, and the results are expressed relative to the control (No Kinase). #, significantly different from control (Student's t test, p Ͻ 0.05). enon (Fig. 2). Treatment with a PKC inhibitor (GF-109203X) or a ROK inhibitor (Y-27632) reduced phosphorylation at Thr-855 without having any effect on phosphorylation at Thr-697. These data suggest that ZIPK was not a significant contributor to Thr-855 phosphorylation in this vascular smooth muscle. Instead, it is likely that the p116 RIP /M-RIP protein, which has been shown to provide scaffolding of MYPT1 and RhoA/ROK (54), may be present to direct ROK-dependent phosphorylation of the Thr-855 site. The signaling outcome at MYPT1 (i.e. the balance of Thr-697 and Thr-855 phosphorylation) may vary in different tissue types, because a Par-4/ZIPK contribution to Thr-855 phosphorylation was previously noted in the ferret portal vein (53). Because our in vitro assays confirmed that ZIPK-KD retained no catalytic activity, phosphorylation at Thr-697 within the rat caudal artery in situ was most likely a result of endogenous ZIPK, myotonic dystrophy protein kinase, ILK, Raf-1, and/or p21-activated protein kinase activity.
Multiple signaling pathways, which could act in a cooperative manner to initiate contraction, most likely contribute to the Ca 2ϩ sensitization phenomenon. The kind of stimulus as well as the type of smooth muscle cell/tissue are two components that could dictate which downstream pathway is activated and the ultimate physiological effect. Differences in signaling pathways would allow for the functional differences observed in various smooth muscle beds. For example, differences in the ILK contribution to Ca 2ϩ sensitization have been reported between the esophagus, which is normally relaxed and exhibits brief and powerful phasic contraction, and the lower esophageal sphincter, which maintains spontaneous myogenic tone (55). It is also possible that different Ca 2ϩ -independent MLCKs allow for differences in LC 20 phosphorylation, which may influence the velocity and type of contraction produced (i.e. transient versus sustained). Finally, the relative amounts of the various signaling components may also influence the extent to which a particular pathway is active. Recently, RT-PCR and Western blot analyses revealed higher levels of RhoA, ROK␣, LC 20 , phospho-MYPT1, and CPI-17 in internal anal sphincter (tonic), lower levels in rectal smooth muscle (tonic and phasic), and the lowest level in anococcygeus smooth muscle (phasic) (56).