Calponin and Mitogen-activated Protein Kinase Signaling in Differentiated Vascular Smooth Muscle*

Contraction of smooth muscle cells is generally assumed to require Ca2+/calmodulin-dependent phosphorylation of the 20-kDa myosin light chains. However, we report here that in the absence of extracellular calcium, phenylephrine induces a contraction of freshly isolated ferret aorta cells in the absence of increases in intracellular ionized calcium or light chain phosphorylation levels but in the presence of activation of mitogen-activated protein kinase. A protein at 36 kDa co-immunoprecipitated with the mitogen-activated protein kinase and was identified as the actin-binding protein, calponin, by immunoblot. An overlay assay further confirmed an interaction between the kinase and calponin, even though the kinase did not phosphorylate calponin in vitro. Calponin also co-immunoprecipitated from smooth muscle cells with protein kinase C-ε. High resolution digital confocal studies indicated that calponin redistributes to the cell membrane during phenylephrine stimulation at a time when mitogen-activated protein kinase and protein kinase C-ε are targeted to the plasmalemma. These results suggest a role for calponin as a signaling molecule, possibly an adapter protein, linking the targeting of mitogen-activated protein kinase and protein kinase C-ε to the surface membrane.

Although the importance of [Ca 2ϩ ] i and LC20 1 phosphorylation in the regulation of smooth muscle contraction is recognized, certain incongruous observations have indicated that additional factors may also regulate, particularly, sustained contractions of smooth muscle. During the initiation of smooth muscle contraction, a transient rise in [Ca 2ϩ ] i generally occurs that in turn activates the Ca 2ϩ /calmodulin-dependent myosin light chain kinase, causing phosphorylation of LC20 and a consequent increase in myosin ATPase activity and crossbridge cycling velocity. The increase in intracellular calcium as well as LC20 phosphorylation is usually observed to decline with time during sustained agonist-induced contractions (1)(2)(3)(4). An hypothesis has been put forward to explain this dissociation of [Ca 2ϩ ] i and LC20 phosphorylation from force, which sug-gests that noncycling or slowly cycling cross-bridges termed "latch bridges" are responsible for maintaining force (5); however, it has been difficult to reconcile all reported data with this hypothesis, and the precise mechanism of regulation of latch bridges has been difficult to define. Others have suggested that the maintenance may be regulated by the actin-binding proteins, calponin and caldesmon (reviewed in Ref. 6).
A considerable and growing body of evidence has suggested that factors in addition to the level of LC20 phosphorylation are responsible for regulating sustained contraction in smooth muscle. The involvement of a second kinase, PKC, was first suggested by the observation that phorbol esters, known to activate PKC, induce slow sustained contractions in several types of vascular smooth muscle (4,(7)(8)(9)(10)(11). In some cases the phorbol ester-induced contractions were observed in the absence of changes in [Ca 2ϩ ] i or phosphorylation of the myosin light chain kinase sites on LC20 (4,10,11). ␣-Agonist-induced PKC-dependent contractions have also been shown to be, in part, Ca 2ϩ -independent in permeabilized ferret aorta cells (12). Studies reported by Khalil et al. (13) have documented a cytosol-to-plasmalemmal translocation of the Ca 2ϩ -independent isoform PKC-⑀ during ␣-agonist-induced cell shortening in these cells. Ferret aorta cells contain both PKC-⑀ and PKC-, another Ca 2ϩ -independent PKC isoform, but only the addition of exogenous PKC-⑀ and not PKC-induced contraction in these cells (14). These data suggest that the Ca 2ϩ -independent contraction is associated with activation of PKC-⑀.
Nonetheless, the signaling pathway by which PKC-⑀ translocation to the plasma membrane activates the contractile filaments remains unclear. The actin-binding proteins calponin and caldesmon can both be phosphorylated in vitro by PKC, but their direct involvement in the signaling cascade has not been demonstrated (reviewed in Ref. 6). Khalil et al. found MAPK to transiently co-distribute with membrane-associated PKC but then subsequently to be targeted to the contractile apparatus; thus, they suggested that MAPK links PKC activation to the measured contractile activity (15). Reports that caldesmon is phosphorylated during contraction (16) and that a caldesmon peptide antagonist contracted permeabilized cells (17) suggest that caldesmon may play a role in this pathway. Furthermore, the identification of the phosphorylation sites on caldesmon as MAPK sites (18) support a role for MAPK in this signaling cascade. The late targeting of MAPK to the contractile filaments was found to be dependent on tyrosine phosphorylation of MAPK, but the early targeting of MAPK to the plasmalemma occurred independent of tyrosine phosphorylation of MAPK (15), and its mechanism is unknown.
In the studies presented herein we provide evidence that in the absence of extracellular Ca 2ϩ , a sustained ␣-agonistinduced contraction is associated with phosphorylation of a p46 MAPK, but occurs in the absence of a detectable increase in the level of LC20 phosphorylation. Additionally, we present evidence supporting a role for calponin as a signaling molecule linking the targeting of PKC and MAPK to the plasmalemma.

MATERIALS AND METHODS
Tissue Preparation-Ferrets were anesthetized with chloroform in a ventilation hood, and the aorta were quickly removed to an oxygenated physiological saline solution (PSS) (4). The endothelium was removed by gentle abrasion with a rubber policeman. Circular muscle strips were prepared as described previously (4) and attached to a force transducer. Muscle strips were incubated in PSS for at least 1 h and then challenged with 24 mM KCl PSS (made by replacing NaCl for KCl mole for mole) to test for viability. Muscles were frozen with Freon, precooled in liquid N 2 at the desired time points following agonist stimulation, and then stored at Ϫ80°C until used.
LC20 Phosphorylation-Phosphorylated and unphosphorylated forms of the 20,000-Da myosin light chain were determined by twodimensional polyacrylamide gel electrophoresis according to Jiang et al. (4). After homogenization, samples were transferred to an isoelectric focusing gel in which an 80%:20% mixture of pH 4.4 -5.4 and pH 3-10 ampholytes were used. After SDS gel electrophoresis, all gels were processed for silver staining. Gels were scanned using a laserjet scanner. The area of the phosphorylated and unphosphorylated spots were determined using ImageQuant software for image analysis. Myosin phosphorylation levels were calculated by dividing the area of the phosphorylated spot by the area of the phosphorylated plus unphosphorylated spots.
Immunoprecipitation-Samples were prepared as described for Western blot. Prior to electrophoresis, samples were incubated with a protein A-agarose slurry to preclear the sample of nonspecific binding to protein A. Additionally, after the pre-clear, antibody-free beads were run in parallel as a negative control for nonspecific binding. The MAPK antibody (ERK-1, Upstate Biotechnology Inc.) and the GRB-2 antibody (Santa Cruz) covalently bound to protein A-agarose beads were added to each sample. This mixture was incubated overnight at 4°C in a cold room. For PKC-⑀ immunoprecipitation, following the preclear, 2 g of PKC-⑀ antibody (Santa Cruz) was added to the sample and allowed to bind for 1 h at 4°C. Then 50 l of 10% protein A-agarose was added and allowed to incubate overnight in the cold room. For both protocols the beads were washed five or six times with 200 l of fresh homogenization buffer and resuspended in a final volume of 60 l. To this, 60 l of 2 ϫ Lamelli sample buffer was added. The samples were boiled for 5 min and centrifuged, and the supernatant was collected and run on SDSpolyacrylamide gels.
Overlay Assay-Western blots were incubated with 10 g/ml ERK-1 K67R expressed in Sf9 cells (19) for 1 h in a solution containing 10 mg/ml bovine serum albumin, 10 g/l leupeptin, and 10 g/ml aprotinin in Tris-buffered saline consisting of 50 mM Tris pH 7.4 and 0.5 M NaCl. Blots were then washed twice in Tris-buffered saline and fixed in 0.5% formaldehyde. Blots were neutralized in 2% glycine, washed three times in phosphate-buffered saline and processed in antibodies as for Western blots.
In Vitro Phosphorylation-Porcine stomach caldesmon was purified according to Bretscher (20), and chicken gizzard calponin was a generous gift from Dr. J. Haeberle (University of Vermont). Caldesmon and calponin were phosphorylated under similar conditions with purified recombinant MAPK (ERK-1) that was activated by a constitutively active GST-MEK (21). Calponin was used at a 3-fold molar excess versus caldesmon to enhance the probability of observing the incorporation of phosphate.
Single Cell Isolation and Digital Imaging-Smooth muscle cells were enzymatically isolated using a previously described procedure, developed specifically to retain pharmacological responsiveness and contractility of the cells (17). Contractile ability and responsiveness to the ␣-agonist phenylephrine was confirmed in all experiments. At appropriate time points cells were fixed with 4% paraformaldehyde. Subsequently, the cells were permeabilized with 0.1% Triton X-100, blocked with 10% goat serum,, and reacted with a mouse anti-human calponin (Sigma) (1:1000) followed by Texas Red secondary antibody (Calbiochem) and mounted with Fluorosave (Calbiochem, San Diego, CA) before analysis.
Images were obtained using a Nikon Diaphot 300 inverted microscope equipped with a Nikon ϫ100 oil immersion objective (NA 1.3). Filters used were 560 Ϯ 20 nm (excitation), 595 nm (dichroic), and 630 Ϯ 30 nm (emission) for Texas Red. Images were recorded with a liquid cooled CCD camera (Photometrics CH250) via Photometrics Microsoft-compatible Image-processing Software (PMIS™) attached to a MS-DOS-based microcomputer. The digitized images were then processed and analyzed on a SPARC station 5 computer using deconvolution and analysis algorithms written in the Khoros image processing environment and based on Bayesian Maximal Entropy theory, using a priori information as described previously (22). This method gives a resolution of less than 140 nm in the x-y axis.
A modified version of a previously described ratio analysis (22) of fluorescence intensities was performed to determine the relative distribution of various labeled proteins within each cell and to normalize for possible differences in staining between cells. A ratio (R) for a central optical section was calculated (after deconvolution) by determining the mean pixel intensity (total intensity, divided by the number of pixels) of the outer 20% of the cell (surface cortex) and dividing this value by the mean pixel intensity of the remaining area (cell core). The part of the section containing the nuclear area was avoided when calculating R values.

RESULTS
Contraction Occurs in the Absence of Changes in LC20 -As has been reported previously from this laboratory, in the presence of PSS containing 2.5 mM [Ca 2ϩ ] e , force, [Ca 2ϩ ] i , and LC20 phosphorylation levels increase in ferret aortic strips at 37°C (Fig. 1A) upon addition of 10 Ϫ5 M of the ␣-agonist phenylephrine. A sustained increase in isometric force was observed, whereas the increase in intracellular [Ca 2ϩ ] i and LC20 phosphorylation were observed to be largely transient during this time period. Isometric force reached a maximum level by 4 min and remained at this level throughout the period of activation. The resting level of LC20 phosphorylation was 8.2 Ϯ 1.7 mol P i /mol protein (n ϭ 5). The level of LC20 phosphorylation reached a maximum of 44 Ϯ 4.9 mol P i /mol protein (n ϭ 4) after 1 min of stimulation with 10 Ϫ5 M phenylephrine and declined toward basal levels reaching 16 Ϯ 2.2 mol P i /mol protein (n ϭ 6) at 10 min. Following stimulation with 10 Ϫ5 M phenylephrine, the concentration of intracellular ionized calcium rose sharply from a basal level of 179 Ϯ 18.5 nM to 502 Ϯ 42.5 (n ϭ 3) at 30 s, decreased to 225 Ϯ 37.1 nM(n ϭ 3) by 2 min, and decreased to 200 Ϯ 34.6 nM (n ϭ 3) by 10 min.
In contrast to the above results, when force, [Ca 2ϩ ] i , and LC20 phosphorylation were measured during activation with 10 Ϫ5 M phenylephrine in the absence of [Ca 2ϩ ] e (2 mM EGTA) (Fig. 1B) Specific Activation of ERK-1 or p46 JNK-Subcellular redistributions of PKC-⑀ and MAPK have been shown to occur during phenylephrine-induced calcium-independent contractions in this tissue (15,24). Whether all members of the MAPK family or a specific MAPK are involved has not yet been determined. Fig. 2A shows Western blots of MAPKs in whole cell ferret aorta homogenates. Four different antibodies were tested here to determine which isoforms of MAPK are present in this tissue. Three anti-ERK antibodies were used, a polyclonal ERK-1 antibody (Upstate Biotechnology Inc.), a monoclonal ERK-2 antibody (Transduction Laboratories), and a monoclonal Pan ERK antibody (Transduction Laboratories), reported to recognize MAPKs in the 54-and 90-kDa ranges in addition to ERK-1 and -2. The polyclonal "ERK-1" antibody was the same antibody previously used by Khalil et al. (15,26) to demonstrate the biphasic translocation of MAPK in ferret aorta cells. All anti-ERK antibodies detected a p42 and a p44 ERK and possibly a p38 isoform in ferret aorta homogenates. The Pan ERK antibody also detected proteins at 54 and ϳ87 kDa. Because members of the JNK family of MAPKs are found in the 50-kDa range as well as 46-and 87-kDa ranges, it is possible that some of the bands recognized by the Pan ERK antibody in Fig. 2 are part of the JNK family. Gerthoffer et al. (25) have also reported the presence of 38-, 50-, and 57-kDa proteins in tracheal smooth muscle recognized by a MAPK antibody that are capable of phosphorylating myelin basic protein. If a p46 JNK were present in ferret aorta, it would possibly be obscured by the presence of ERK-1, thus a p46 specific JNK-1 antibody (Santa Cruz) was also used ( Fig. 2A), demonstrating that p46 JNK is present in ferret aorta. Additionally, we found that the ERK-1 antibody previously used to image the targeting of MAPK in the ferret aorta also reacts with expressed JNK-1 protein (Santa Cruz) (Fig. 2B). Thus, a number of members of the MAP kinase family are present in this vascular smooth muscle. Exactly what role(s) or specificities can be assigned to each type of MAP kinase remains to be determined.
MAPK activation was monitored by probing Western blots with an anti-phosphotyrosine antibody. Interestingly, in these freshly isolated tissues, in the absence of serum or agonist stimulation, only a single major band at 45-46 kDa was detected (Fig. 2C). This band increased in intensity after addition of phenylephrine. On over-exposed blots, a second band could also be detected at p38. Stimulation with a phorbol ester produced no greater signal at p46 but an increased signal at 38 kDa (Fig. 2C), and bands at higher molecular masses appeared. These data clearly show, however, that the major band tyrosine phosphorylated in response to the ␣-agonist is at 46 kDa, presumably either ERK-1 or JNK-1 MAPK.
Calponin and MAPK Co-immunoprecipitate-Immunoprecipitation of MAPK was performed using the polyclonal ERK-1 antibody that was previously used to demonstrate the transient plasmalemmal translocation followed by contractile filament targeting of MAPK in these cells (15). MAPK was identified as immunoprecipitating at 46 kDa by Western blot of the immunoprecipitate using the same ERK-1 antibody (Fig. 2D). The major protein occurring at 36 kDa was identified by Western blot as calponin, an actin-binding protein (Fig. 2D). As a negative control, an antibody to GRB-2 was also used in immunoprecipitation experiments on ferret aorta whole cell homogenate. The antibody recognizes and immunoprecipitates GRB-2 from these samples, but no co-immunoprecipitating calponin is detectable, supporting the specificity of the interaction between calponin and MAPK.
No detectable increase in calponin co-immunoprecipitating with MAPK was seen during agonist stimulation, but at 10 min after the addition of PE, the amount of calponin in the immunoprecipitates decreased to an average of 24% of the control level (n ϭ 4). Actin was also identified in the MAPK immunoprecipitates; raising the question of whether calponin was directly binding to MAPK or indirectly, by binding to actin in the immunoprecipitates.
PKC-⑀ Immunoprecipitates with Calponin-Even though PKC-⑀ and MAPK appear to co-localize on high resolution fluorescence micrographs after 4 min of stimulation of tissue with PE (15), we were unsuccessful in detecting PKC-⑀ in the MAPK immunoprecipitates, which suggests that they do not directly interact. Therefore, immunoprecipitation with an anti-PKC-⑀ antibody after 4 min of stimulation with phenylephrine was performed on ferret aorta homogenates to further investi- gate whether PKC-⑀ associates with MAPK. MAPK was not observed in the immunoprecipitate, but again calponin was determined to be present by Western blot (data not shown), and the amount of calponin in the immunoprecipitate increased by 2 min to an average of 456% of the unstimulated levels (n ϭ 3).
Calponin Directly Binds Both MAPK and PKC-⑀-To confirm that calponin binds directly to MAP kinase rather than simply being associated with actin in the immunoprecipitates, we performed an overlay assay using ERK-1 K67R kinase. As shown in Fig. 2E, MAPK binds directly to isolated purified calponin, actin, caldesmon, and, as a positive control, myelin basic protein. Additionally, a band at the molecular mass of CaP in whole cell homogenates (Fig. 2E, arrow) also bound MAPK. As a negative control, MAPK did not bind to calmodulin (data not shown). These data indicate that the calponin can directly bind MAPK and argue against its being simply a contaminant in the immunoprecipitate. Additionally, the binding of MAPK to actin directly in the overlay assay is consistent with the targeting of MAPK to contractile filaments in situ.
Calponin has previously been reported to be a substrate of PKC-⑀ (14); however, an interaction between CaP and MAPK has not been previously reported. Therefore, we tested whether CaP could be used as a substrate by MAPK and found that it was not significantly phosphorylated (Fig. 2F). In contrast, under identical conditions, using one-third the concentration of caldesmon, phosphorylation by MAPK was readily detectable (Fig. 2F). These findings are consistent with there being no (S/T)P sequences in CaP.
Calponin Is Targeted to the Plasmalemma with MAPK and PKC-⑀ in Intact Cells-Calponin has previously been reported to undergo a dynamic subcellular redistribution during contraction/relaxation cycles in ferret portal vein cells (22), but it is not known if this occurs in other cell types. To determine if calponin is in the appropriate spatio-temporal location to interact with MAPK and PKC-⑀ in the intact ferret aorta cell, high resolution digital imaging studies were performed using a monoclonal antibody to calponin. In resting cells (Fig. 3A, top) calponin was distributed on filamentous bundles in the core of the cell. Following 2 min of stimulation with an ␣-agonist, calponin began to redistribute toward the subplasmalemmal cell cortex. After 4 min (Fig. 3A, middle) of stimulation with Immunoprecipitates were immunoblotted with antibodies to ERK-1, CaP, and actin. E, overlay assay. Whole cell homogenates and pure proteins (calponin, myelin basic protein, actin, and caldesmon) were incubated with recombinant ERK-1 then immunoblotted with an anti-ERK-1 antibody to determine which proteins bound ERK-1. F, in vitro phosphorylation with MAPK. Caldesmon and calponin were phosphorylated by purified MAPK (ERK-1) that was activated by constitutively active GST-MEK. phenylephrine, calponin redistributed to the submembranous cortex where it remained through 10 min of stimulation (Fig.  3A, bottom). These results were quantitated (Fig. 3B) by the measurement of a surface-to-cytosolic ratio for the distribution of fluorescently labeled CaP, as described previously (22). In Fig. 3C, the time course of the redistribution CaP is compared with that of MAPK and PKC-⑀ and cell shortening as previously determined for this cell type (26). These data show that through 4 min of stimulation, calponin, MAPK, and PKC-⑀ follow an indistinguishable time course for targeting to the plasmalemma. DISCUSSION In the present study we have demonstrated that in the absence of a change in [Ca 2ϩ ]i or LC20 phosphorylation, phenylephrine can cause a contractile response in ferret aorta smooth muscle. Others have suggested that inhibition of LC20 phosphatase can allow agonist-induced contraction of smooth muscle to occur in the absence of changes in [Ca 2ϩ ] i (27). However, inhibition of the phosphatase would result in sustained increases of LC20 phosphorylation, and if this were the mechanism of the contraction, contractile force should parallel LC20 phosphorylation levels. Clearly, the results reported here are not consistent with such a mechanism. The fact that the EC 50 for phenylephrine in the presence of [Ca 2ϩ ] e was determined to be significantly different from the EC 50 for phenylephrine in the absence of [Ca 2ϩ ] e is consistent with there being two different mechanisms involved in the presence and absence of [Ca 2ϩ ] e . In the presence of Ca 2ϩ , activation of myosin light chain kinase is expected to contribute to the generation of tone, but in the absence of Ca 2ϩ , other mechanisms must be involved. Past studies have implicated PKC-⑀ and MAPK in this alternative pathway (6), and this is consistent with the detection of MAPK activation under these Ca 2ϩ -free conditions (Fig. 2C).
The signaling pathways involving MAP kinase in differentiated nonproliferating cells, such as contractile smooth muscle, are less well understood than those in proliferating cells. This laboratory (15) has previously reported that phenylephrineinduced Ca 2ϩ -independent contraction of ferret aorta cells is completely abolished by PKC inhibitors and significantly inhibited by tyrosine kinase inhibitors suggesting that PKC activation and tyrosine phosphorylation of MAPK are involved in a signaling cascade leading to Ca 2ϩ -independent smooth muscle contraction. The signals that link PKC activation to MAP kinase activation in differentiated smooth muscle remain unclear. Khalil et al. (15) described an initial translocation of MAPK to the membrane after 4 min of phenylephrine stimulation. This translocation was totally inhibited by PKC antagonists and was previously shown to coincide with translocation of PKC-⑀ to the plasma membrane. Tyrosine phosphorylation had been thought to act as a targeting mechanism for translocation of kinases to the surface membrane. However, in this same study, Khalil et al. showed that tyrosine kinase antagonists did not block and actually enhanced translocation of MAPK to the plasmalemma. Additionally, on imaging with antiphosphotyrosine antibodies, tyrosine phosphorylation was only detectable after MAPK had translocated to the membrane. In contrast, tyrosine kinase inhibition blocked the later targeting to the contractile filaments, suggesting a possible mechanism for the second phase of MAPK targeting in these cells, but the mechanism by which MAPK localization to the plasmalemma occurs was unknown.
We were surprised to find that calponin co-immunoprecipitates with MAPK in these cells and furthermore to find that CaP directly binds MAPK as demonstrated by overlay assay. Interestingly, MAPK was found not to phosphorylate CaP as a result of this interaction. The fact that the amount of CaP in the MAPK immunoprecipitates decreased 4-fold below control levels at a late (by 10 min) time point of agonist-activation but that the amount of calponin in the PKC-⑀ immunoprecipitates increased at early time points is consistent with a model where calponin is associated with MAPK in the resting cells, but upon PE-induced activation of PKC-⑀, calponin also associates with PKC-⑀ and acts as an adapter protein that may facilitate the translocation of MAPK to the membrane with PKC-⑀.
Calponin was originally suggested, based on its in vitro actin binding and myosin ATPase-inhibiting properties, to play a role in directly inhibiting the interaction of actin and myosin (reviewed in Ref. 6). Some results reported in permeabilized cells from this and other laboratories are consistent with this interpretation (28,29). However, other published results are not (30). The imaging studies reported here indicate that CaP leaves the actin filaments before detectable shortening occurs, and it is unclear how to reconcile these findings with a mechanism involving direct modulation of myosin ATPase activity at the myofilament level. In contrast, the fact that the redistribution of calponin coincides with that of PKC-⑀ and MAPK in these cells suggests a signaling role for CaP and provides a possible mechanism for the targeting of MAPK to the surface membrane.