Cyclic nucleotide-dependent vasorelaxation is associated with the phosphorylation of a small heat shock-related protein.

Activation of cyclic nucleotide-dependent signaling pathways leads to the relaxation of various smooth muscles. One of the major phosphorylation events associated with cyclic nucleotide-dependent vasorelaxation in bovine trachealis and carotid artery smooth muscle is the phosphorylation of two 20-kDa phosphoproteins with pI values of 5.7 and 5.9 (previously designated pp8 and pp3, respectively). The present studies sought to determine the identities of pp3 and pp8 in vascular smooth muscle. The phosphopeptide maps for the pp8 and pp3 proteins were similar. Preparative two-dimensional gel electrophoresis and amino acid sequencing of a peptide fragment of the pp3 protein revealed a sequence identical to a 20-kDa heat shock-related protein (HSP20) previously purified from skeletal muscle. Western blot and immunoprecipitation analysis with anti-HSP20 antibodies demonstrated that the pp3 and pp8 proteins are phosphorylated forms of HSP20. In addition, HSP20 could be phosphorylated in vitro by both cAMP-dependent protein kinase and cGMP-dependent protein kinase. These data suggest that the phosphorylation of the heat shock-related protein HSP20 is associated with cyclic nucleotide-dependent relaxation of vascular smooth muscle.

Activation of cyclic nucleotide-dependent signaling pathways leads to the relaxation of vascular smooth muscle. Isoproterenol, prostacyclin, and forskolin stimulate the adenylate cyclase/ cAMP pathway and activate cAMP-dependent protein kinase (PKA) 1 (1). Nitric oxide, atrial natriuretic peptide, sodium nitroprusside, and nitroglycerin stimulate the guanylate cyclase/ cGMP pathway and activate cGMP-dependent protein kinase (PKG) (2).
A number of investigators have sought to identify the protein targets for PKA and PKG that are involved in smooth muscle relaxation. Two sites of particular interest have emerged. First, the phosphorylation of myosin light chain kinase by PKA decreases its sensitivity to activation by Ca 2ϩ -calmodulin, leading to a decrease in the phosphorylation of the myosin light chains (3). Second, PKA and/or PKG activation leads to changes in the activities of one or more Ca 2ϩ channels and/or Ca 2ϩ pumps, thereby reducing the intracellular Ca 2ϩ concentrations (4,5). However, there is no simple correlation between the extent of myosin light chain phosphorylation and the state of contraction of vascular smooth muscle (6,7). On the other hand, cyclic nucleotide-dependent relaxation of intact muscle strips can occur under conditions where the Ca 2ϩ concentration is fixed at either a low or a high concentration (8 -12).
The relaxation of several different types of smooth muscles by forskolin or sodium nitroprusside is associated with an increase in the extent of phosphorylation of two 20-kDa proteins with pI values of 5.9 and 5.7, previously described as proteins 3 (pp3) and 8 (pp8), respectively (13,14). In addition, the extent of phosphorylation of these two proteins increases during forskolin or sodium nitroprusside-induced vasorelaxation under circumstances where the intracellular Ca 2ϩ concentrations are low and fixed (12). Finally, there is no increase in the extent of the phosphorylation of these two proteins in human umbilical artery smooth muscle, a muscle that is uniquely refractory to cyclic nucleotide-dependent vasorelaxation (15). Taken together, these data suggest that an increase in the phosphorylation of these two 20-kDa phosphoproteins plays a role in cyclic nucleotide-dependent vasorelaxation.
In the present study, preparative gel electrophoresis was employed to isolate one of these 20-kDa proteins (pp3) from bovine carotid smooth muscle. Amino acid sequencing of a peptide fragment of the pp3 protein revealed a sequence identical to a small heat shock-related protein (HSP20) recently isolated from rat and human skeletal muscle (16). Using highly specific antibodies to HSP20, in situ and in vitro phosphorylation studies of HSP20 show that the two 20-kDa phosphoproteins are phosphorylated forms of HSP20 and that HSP20 is a substrate for both PKA and PKG. These data indicate that there is a correlation between the ability of an intact vascular smooth muscle to undergo relaxation and an increase in the extent of the phosphorylation of a small heat shock-related protein, HSP20.

EXPERIMENTAL PROCEDURES
Materials-The catalytic subunit of PKA and endoproteinase Lys-C were purchased from Promega (Madison, WI). The [ 32 P]orthophosphate and [␥-32 P]ATP were from Amersham Corp. The cAMP-dependent protein kinase inhibitor peptide was from Peninsula Labs (Belmont, CA). Serotonin and protein A-Sepharose beads were from Sigma. Forskolin, leupeptin, and aprotinin were from Calbiochem. Electrophoresis reagents and the DC protein assay kit were from Bio-Rad. Rabbit anti-HSP20 antibody was produced against purified HSP20 as described * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
§ Supported by an American Heart Association Georgia Affiliate grant-in-aid, National Institutes of Health Grant DK43405, and a Department of Veterans Affairs Merit Review.
previously (16). Rabbit anti-␣B-crystallin was from Chemicon (Temecula, CA), and mouse anti-HSP27 was a generous gift from Dr. Michael Welsh (University of Michigan, Ann Arbor, MI). All other reagents were of analytical grade.
Preparation of Vascular Smooth Muscle Strips-Intact bovine carotid arteries were obtained from an abattoir. The adventitia was dissected free, and the endothelial layer was gently denuded. The arteries were opened longitudinally, and thin transverse strips were cut. The contractile viability of the vessels was confirmed by parallel muscle-bath experiments as described previously (17).
Whole Cell Phosphorylation-Strips of bovine carotid artery smooth muscle were equilibrated in a bicarbonate buffer (120 mM NaCl, 4.7 mM KCl, 1.0 mM MgSO 4 , 1.0 mM NaH 2 PO 4 , 10 mM glucose, 1.5 mM CaCl 2 , and 25 mM Na 2 HCO 3 ) bubbled with 95% O 2 , 5% CO 2 for 1 h at 37°C. The strips were then rinsed and incubated in a low phosphate buffer consisting of 10 mM Hepes, pH 7.4, 140 mM NaCl, 4.7 mM KCl, 1.0 mM MgCl 2 , 1.5 mM CaCl 2 , 10 mM glucose, and 0.3 mM NaH 2 PO 4 for 15-30 min. 250 Ci/ml [ 32 P]orthophosphate was added 3 h before the addition of a vasorelaxant (10 M forskolin or 10 M sodium nitroprusside). The incubation was terminated by immersing the muscle strips in a dry ice/acetone slurry and then crushing the tissue with mortar and pestle under liquid N 2 . The powder was resuspended in homogenization buffer (20 mM Hepes, pH 7.4, 20 mM sucrose, 100 mM NaF, 15 mM EDTA, 2 mM EGTA, 1 mM phenylmethylsulfonyl fluoride, and 1.3% SDS) and boiled for 5 min. Protein concentrations were normalized using the Bio-Rad DC protein assay kit.
Two-dimensional Gel Electrophoresis-Two-dimensional electrophoresis was performed using vertical slab isoelectric focusing gels with the modification described by Hochstrasser et al. (18). Briefly, the proteins in the samples were acetone-precipitated and reconstituted in 9 M urea and 2% CHAPS. The samples were protein-normalized, and 100 g of protein was adjusted to a final concentration of 9 M urea, 2% CHAPS, 100 mM dithiothreitol, 15% glycerol, and 5% Ampholine (5 parts pH 6 -8, 3 parts pH 5-7, 2 parts pH 3-10). The first dimensions were focused for 10,000 V-h and then run on a 12% SDS-PAGE second dimension (19). The gels were stained with Coomassie Brilliant Blue, and the dried gels were exposed to Kodak XAR-5 film.
For peptide sequencing, the second dimension gels were transferred to Immobilon-P SQ (Millipore, Bedford, MA) and stained with Ponceau Red to visualize the spots for isolation and sequencing. Eight to ten gels were transferred, and the pp3 protein was submitted to the Microchemical Facility, Winship Cancer Center, Emory University for amino acid sequencing. The protein was digested using endoproteinase Lys-C and separated by reversed phase microbore HPLC, and selected fragments were sequenced.
Peptide Mapping-Peptide mapping was performed according to the method of Cleveland et al. (20). The spots corresponding to the 20-kDa proteins pp3 and pp8 were cut from two-dimensional gels and rehydrated with 125 mM Tris, pH 6.8, 0.1% SDS, for 1 h. The rehydrated gel pieces were placed in the wells of a 15% SDS-PAGE gel and overlaid with 10 mg of Staphylococcus aureus V8 protease in 125 mM Tris HCl, pH 6.8, 0.1% SDS, and 15% glycerol. The gel was run at 150 V until the dye front reached the end of the gel. The gels were dehydrated in graded methanol (to 100% methanol), dried, and exposed to Kodak XAR-5 film.
Immunoprecipitation-Strips of bovine carotid artery smooth muscle were homogenized in TBS (0.5 g of tissue/ml of buffer), and then the samples were centrifuged at 10,000 ϫ g for 15 min. The soluble proteins were then diluted 10-fold with TBS. The anti-HSP20 antiserum was added to the supernatants (1:100 dilution). The samples were shaken gently for 14 h at 4°C. Protein A-Sepharose beads (0.1 volume) were added, and the samples were incubated for an additional 3 h at 4°C. The beads were washed 3 times with TBS, 0.5% Tween 20. A final wash of 10 mM Tris, pH 7.4, was then done. The immunoprecipitated samples were phosphorylated in vitro (see below) or reconstituted in 9 M urea, 2% CHAPS, 100 mM dithiothreitol, 15% glycerol, and 5% Ampholine (5 parts pH 6 -8, 3 parts pH 5-7, 2 parts pH 3-10) and separated by two-dimensional mini-gel electrophoresis. Quantitative Analysis-Densitometric analysis of the phosphoproteins was performed using a CCD camera interfaced with a two-dimensional analyzer software package (Inovision Software, Bioimage Corp., Ann Arbor, MI). The results are depicted as the mean Ϯ S.E.. Statistical analysis was by one-way analysis of variance followed by the Tukey test, and p Ͻ 0.05 was considered significant.

RESULTS
The Phosphorylation of HSP20 in Intact Strips of Bovine Carotid Artery Smooth Muscle-Treatment of strips of bovine carotid artery smooth muscle with either the adenylate cyclase activator forskolin (10 M) or the guanylate cyclase activator sodium nitroprusside (10 M) resulted in a significant increase in the phosphorylation of two 20-kDa proteins with pI values of 5.7 and 5.9 (previously designated pp8 and pp3, respectively (Fig. 1)). Sodium nitroprusside elicited a 2.7-fold increase in pp8 phosphorylation and a 1.8-fold increase in pp3 phosphorylation (Table I). Forskolin stimulated a 5.2-fold increase in pp8 phosphorylation and a 3.0-fold increase in pp3 phosphorylation (Table I). Also, as described previously (14) the phosphorylation of a third 20-kDa protein with a pI of 6.0 (designated pp4) decreased following treatment with either forskolin or sodium nitroprusside (Fig. 1). The increases in the phosphorylation of pp3 and pp8 are the major phosphorylation changes observed during cyclic nucleotide-dependent vasorelaxation of carotid artery smooth muscle as determined with whole cell phosphorylation and two-dimensional gel electrophoresis (14).
To assess the relationship of proteins pp3 and pp8, we performed S. aureus V8 limited proteolysis (20) of pp3 and pp8 phosphorylated in response to either forskolin or sodium nitroprusside. Digests of the two 20-kDa proteins gave similar phosphopeptide maps (Fig. 2). These data suggested that the pp3 and pp8 proteins are structurally related and are phosphorylated within closely related peptide sequences in response to both forskolin and sodium nitroprusside.
Identification of 20-kDa Phosphoproteins-To identify the  for 10 min, B), or the adenylate cyclase activator forskolin (10 M for 10 min, C). Arterial homogenates were separated by two-dimensional electrophoresis, and autoradiographs were obtained. The pH gradient is indicated at the bottom of the gels, and molecular mass markers (kDa) are identified at the left. Treatment with sodium nitroprusside or forskolin resulted in an increase in the phosphorylation of two 20-kDa phosphoproteins (pp3 and pp8 (arrows)). The phosphorylation of another 20-kDa protein (pp4) was decreased with sodium nitroprusside or forskolin treatment. The 20-kDa myosin light chain (m) and the intermediate filaments desmin and vimentin (IF) are also identified based on their previously described migration pattern on two-dimensional gels (13). These blots are representative of 10 separate experiments. 20-kDa proteins, the pp3 phosphoprotein was isolated from preparative two-dimensional gels. A Lys-C peptide digest was resolved on HPLC, and a single major peptide peak was isolated and submitted for sequencing. Amino acid analysis of the Lys-C peptide fragment revealed a sequence of HFSPEEIAVK. This sequence contained an 80% sequence identity with the small heat shock protein ␣B-crystallin (HFSPEELKVK) and was completely identical to another protein recently purified from skeletal muscle (HSP20 (16)).
To confirm that HSP20 was phosphorylated during cyclic nucleotide-dependent relaxation, immunoblots with anti-HSP20 antibodies were performed. Intact strips of bovine carotid artery smooth muscle were labeled with [ 32 P]orthophosphate and then stimulated with 10 M forskolin for 10 min. The strips were homogenized, and the 10,000 ϫ g supernatant proteins were separated by two-dimensional gel electrophoresis. The proteins were transferred to Immobilon, and an autoradiograph was developed. The membranes were subsequently probed with rabbit anti-HSP20 antibodies. Both pp8 and pp3, which demonstrated increases in phosphorylation with cAMPdependent vasorelaxation, were immunoreactive with anti-HSP20 antibodies (Fig. 3). A more basic non-phosphorylated protein was also immunoreactive with anti-HSP20 antibody. The non-phosphorylated immunoreactivity comigrated with purified rat HSP20 (data not shown). Immunoreactivity for the non-phosphorylated protein decreased with forskolin stimulation. In addition, the pp4 protein was recognized by the HSP20 antiserum and, similar to pp4 phosphorylation, immunoreactivity decreased with stimulation by forskolin. Antibodies against closely related ␣B-crystallin and another small heat shock protein (HSP27) did not cross-react with any of the HSP20 immunoreactive proteins (data not shown).
Finally, to confirm the identities of pp3 and pp8 intact strips of bovine carotid arteries were stimulated with 10 M forskolin for 10 min, and muscle proteins were immunoprecipitated with anti-HSP20 antibodies (Fig. 4). The anti-HSP antibodies immunoprecipitated two proteins with similar molecular masses and isoelectric points to the pp3 and pp8 proteins phosphorylated in response to forskolin stimulation (Fig. 4).
In Vitro Phosphorylation of HSP20 -The phosphorylation of HSP20 increased under conditions where intracellular cAMP and/or cGMP concentrations were elevated. Increases in cAMP and cGMP are thought to effect cellular responses via the activation of PKA and PKG, respectively (1,2). To determine if PKA and PKG phosphorylate HSP20 in vitro, HSP20 was immunoprecipitated from cytosolic protein fractions of carotid artery smooth muscle and phosphorylated in vitro with purified PKG or the catalytic subunit of PKA (Fig. 5). HSP20 was phosphorylated in vitro by both PKG and PKA and produced two-dimensional patterns similar to the in situ phosphorylation observed after sodium nitroprusside or forskolin treatment of intact strips of carotid artery. In vitro phosphorylation with PKA also variably led to an additional minor, more acidic isoform. No phosphorylation of HSP20 was observed in immunoprecipitated fractions in which PKG or PKA was not added. Thus, while another small HSP (␣B-crystallin) has been shown to contain "autokinase" activity (21), these data suggest that HSP20 does not co-immunoprecipitate with autokinase activity.

DISCUSSION
Cyclic nucleotide-dependent vasorelaxation is associated with an increase in the phosphorylation of two 20-kDa phosphoproteins, pp8 and pp3 (14,15). These two phosphoproteins share partial sequence identity and are immunoreactive with a recently identified small heat shock-related protein (HSP20). They appear to represent phosphorylated forms of the same protein. The phosphorylation of HSP20 increases with cyclic TABLE I Effects of sodium nitroprusside and forskolin on in vivo phosphorylation of pp8 and pp3 in intact bovine carotid artery smooth muscle strips Data are means Ϯ S.E. for the volume of each spot calculated in arbitrary units by ImageQuant densitometry performed on autoradiographic film exposed for 12 h at Ϫ70°C with an intensifying screen. The phosphorylation of the reference B protein as previously described (13,14) was used as an internal standard. The data are compiled from seven individual experiments.  2. Phosphopeptide maps of the pp3 and pp8 20-kDa proteins. Strips of bovine carotid artery smooth muscle were prelabeled with [ 32 P]orthophosphate and treated for 10 min with either 10 mM forskolin (F) or 10 mM sodium nitroprusside (S). The spots corresponding to the 20-kDa phosphoproteins pp8 and pp3 were excised from the two-dimensional gels. Proteolytic digests of the proteins were performed according to the method of Cleveland et al. (20), the peptides were separated on a 15% SDS-PAGE gel, and autoradiographs were obtained. The pattern of the phosphorylated peptide fragments from the pp3 protein was similar to the pattern for the pp8 protein. The autoradiograph is representative of three separate experiments. The corresponding anti-HSP20 immunoblots (C, D) demonstrate immunoreactive proteins corresponding to the pp8, pp3, and pp4 phosphoproteins. In addition, the immunoblot also identifies a non-phosphorylated protein immunoreactive with anti-HSP20 (*). These blots are representative of three separate experiments.
nucleotide-dependent vasorelaxation under physiologic Ca 2ϩ conditions and under conditions where the intracellular Ca 2ϩ is low and fixed (12,14). However, the phosphorylation of HSP20 does not increase in a muscle that is uniquely refractory to cyclic nucleotide-dependent vasorelaxation, human umbilical artery smooth muscle (15). These data suggest that the phosphorylation of HSP20 is an important event in cyclic nucleotidedependent relaxation of vascular smooth muscle.
The phosphopeptide maps of the two isoforms of HSP20 are similar for muscles that were stimulated to relax with either forskolin or sodium nitroprusside. These results indicate that HSP20 is phosphorylated on similar sites by both PKA and PKG. HSP20 can also be phosphorylated in vitro by the purified cyclic nucleotide-dependent protein kinases PKA and PKG. The in vitro phosphorylation of HSP20 by PKA and PKG leads to phosphoproteins of similar mobility on two-dimensional gels, again suggesting that HSP20 represents a common substrate for both PKA and PKG. Indeed, the protein sequence for HSP20 (16) contains a consensus sequence (RRAS) for both PKA and PKG phosphorylation.
An additional 20-kDa phosphoprotein (pI 6.0) previously referred to as phosphoprotein 4 (pp4 (14)) was also immunoreactive with anti-HSP20 antibodies. However, the phosphorylation of this protein increases with stimuli that induce contraction of the vascular smooth muscle and decreases with stimuli that induce relaxation (14). Treatment of carotid artery smooth muscle with phorbol esters also elicits an increase in pp4 phosphorylation (14). Thus, pp4 may represent a population of HSP20 that is phosphorylated by protein kinase C or another unidentified kinase. Alternatively, it is possible that pp4 represents a different protein that shares homology with HSP20.
Heat shock proteins represent a family of phylogenetically well conserved proteins whose expression is induced by cellular stress (for review, see Ref. 22). Many HSPs (including HSP20) are also expressed constitutively and thus appear to play a role in normal cellular behavior. ␣B-crystallin, HSP27, and HSP20 are all members of the low molecular weight HSP family ("small HSPs"). The small HSPs share considerable sequence homology and often copurify in large macromolecular aggregates (16,23). HSP20 was originally identified as a by-product of the purification of HSP27 (16). While HSP20 is ubiquitously distributed, it is found in much higher levels in skeletal, smooth, and heart muscle (16). The prevalence of HSP20 in muscle tissue supports a role for HSP20 in contractile physiology. Unlike ␣B-crystallin and HSP27, the amount of HSP20 does not increase after heat shock in rat skeletal muscle (16). However, HSP20 does redistribute from a cytosolic to an insoluble fraction and dissociates from an aggregated form after cellular stress (16). Thus, HSP20 does share some of the functional properties of the other small HSPs.
While the precise functions of the HSPs are not known, many HSPs act as "molecular chaperones" and assist in the assembly, disassembly, stabilization, and internal transport of intracellular proteins. Recent studies suggest that small HSPs are important regulatory components of the actin-based cytoskeleton (24), and phosphorylation of HSP27 has been implicated in regulating the contraction of rectal sphincter smooth muscle (25). Other investigations have suggested that small HSPs interact with intermediate filaments (26), which in turn may play a regulatory role in vascular smooth muscle contraction and relaxation (27).
Although the specific role that the phosphorylation of HSP20 plays in vasorelaxation is not known, the phosphorylation of HSP20 may alter its ability to associate either with components of either the actin-myosin contractile domain or with the intermediate filament domain of smooth muscle cells (28,29). The localization of another small HSP (␣B-crystallin) with the Z band of cardiac muscle along with the evidence that the dense bodies are the structural counterpart of the Z band in smooth muscle raises the possibility that small HSPs act at the level of the dense bodies (30). Since the dense bodies are the sites at which both the actin-myosin contractile filament and the intermediate filament domains are anchored, cyclic nucleotide-dependent relaxation may lead to a simultaneous reorganization of each fibrillar domain and release the muscle from the socalled "latch" state (31). FIG. 4. In situ phosphorylation and immunoprecipitation with anti-HSP20 antibodies. Intact strips of bovine carotid artery smooth muscle were preincubated with [ 32 P]orthophosphate and then incubated in the absence (A) or presence (B, C) of forskolin (10 M for 10 min). The strips were homogenized and immunoprecipitated with anti-HSP20 antibodies (A, B) or protein A-Sepharose beads alone (C). The immunoprecipitated proteins were separated on two-dimensional gels, and autoradiographs were obtained. The autoradiographs demonstrate two phosphoproteins with similar mobility to pp8 and pp3. There was an increase in the phosphorylation of the pp8 and pp3 proteins with forskolin stimulation (B). No phosphoproteins were identified on blots in which the protein A-Sepharose beads were not pre-incubated with antibodies (C). These blots are representative of three separate experiments.
FIG. 5. Immunoprecipitation and in vitro phosphorylation of HSP20 from bovine carotid artery smooth muscle. Homogenates of bovine carotid artery smooth muscle were immunoprecipitated with anti-HSP20 antibody followed by phosphorylation of the immunoprecipitates in the presence of buffer alone (A), PKG (B), or the catalytic subunit of PKA (C). The proteins were separated by two-dimensional gel electrophoresis followed by autoradiography. Both PKA and PKG phosphorylate HSP20 in vitro (pp8 and pp3). In vitro phosphorylation with PKA also led to the phosphorylation of an additional, more acidic isoform of HPS20. However, associated endogenous kinase activity did not immunoprecipitate with HSP20 since there was no phosphorylation in the absence of PKA or PKG (A). The autoradiographs are representative of three separate experiments.