Actions downstream of cyclic GMP/protein kinase G can reverse protein kinase C-mediated phosphorylation of CPI-17 and Ca²⁺ sensitization in smooth muscle.

Ca(2+) sensitivity of smooth muscle contraction is modulated by several systems converging on myosin light chain phosphatase (MLCP). Rho-Rho kinase is considered to inhibit MLCP via phosphorylation, whereas protein kinase C (PKC) induced sensitization has been shown to be dependent on phosphorylation of the inhibitory protein CPI-17. We have explored the interaction of cGMP-dependent protein kinase (PKG) with Ca(2+) sensitization pathways using permeabilized mouse smooth muscle. Three conditions giving approximately 50% of maximal active force were compared in small intestinal preparations: 1). Ca(2+)-activated unsensitized muscle (pCa 5.9 with Rho kinase inhibitor Y27632); 2). Rho-Rho kinase-sensitized muscle (pCa 6.1 with guanosine 5'-3-O-(thio)triphosphate); and 3). PKC-sensitized muscle (pCa 6.0 with Y27632 and PKC activator phorbol 12,13-dibutyrate). 8-Br-cGMP relaxed the sensitized muscles but had marginal effects on unsensitized preparations, showing that PKG reverses both PKC and Rho-mediated Ca(2+) sensitization. CPI-17 was present in permeabilized intestinal tissue. In PKC-sensitized preparations, CPI-17 phosphorylation decreased in response to 8-Br-cGMP. The rate of PKC-mediated phosphorylation in the presence of the MLCP inhibitor microcystin-LR was not influenced by 8-Br-cGMP. PKC-induced Ca(2+) sensitization also was reversed in vascular smooth muscle tissues (portal vein and femoral artery). We conclude that actions downstream of cGMP/PKG can reverse PKC-mediated phosphorylation of CPI-17 and Ca(2+) sensitization in smooth muscle.

Smooth muscle contraction involves a rise in intracellular [Ca 2ϩ ]. The Ca 2ϩ -calmodulin complex activates myosin light chain kinase, which phosphorylates the 20-kDa myosin regulatory light chains (MLC 20 ). 1 This phosphorylation is counter-acted by myosin light chain phosphatase (MLCP), which dephosphorylates MLC 20 . The extent of phosphorylation and contraction is modulated by the relative activities of myosin light chain kinase and MLCP (1). These enzymes are influenced by a signaling network of processes in the smooth muscle cell (2,3). The inhibition of MLCP leads to Ca 2ϩ sensitization, i.e. increased MLC 20 phosphorylation and force at a given intracellular Ca 2ϩ concentration. The activation of MLCP leads to Ca 2ϩ desensitization, i.e. decreased MLC 20 phosphorylation and force at a given intracellular Ca 2ϩ concentration. Alterations in Ca 2ϩ sensitivity also can be achieved by modulation of myosin light chain kinase, e.g. by the action of calmodulin-dependent kinase II (4). In addition, contraction can be regulated by MLC 20 phosphorylation-independent mechanisms, e.g. acting on thin filaments (2,3).
MLCP (PP1M) belongs to the PP1 group of phosphatases and is a heteromeric enzyme consisting of three subunits: a catalytic subunit (PP1M C ); a myosin-targeting subunit (MYPT1); and the M20 subunit (5). MLCP has been reported to be inhibited directly by phosphorylation of its myosin-targeting subunit, MYPT1, on residue Thr-695 performed by Rho kinase (6), MYPT1-associated kinase (sometimes referred to as ZIP-like kinase) (7), and integrin-linked kinase (8). Phosphorylation of MYPT1 at Thr-34 by protein kinase C (PKC) has been shown in vitro (9); the relevance of this result and the effect on the holoenzyme are unknown. MLCP is blocked and inhibited by the binding of the phosphorylated form of protein kinase Cpotentiated inhibitory protein for heterotrimeric myosin light chain phosphatase of 17 kDa (CPI-17). CPI-17 is a major target for phosphorylation by PKC, which phosphorylates the Thr-38 residue. This process has been shown to be crucial for PKCmediated increase in Ca 2ϩ sensitivity of force (Ca 2ϩ sensitization) in smooth muscle (10,36,43). In vitro, Thr-38 on CPI-17 is also phosphorylated by Rho kinase (11), MYPT1-associated kinase (12), protein kinase N (13), integrin-linked kinase (14), p21-activated protein kinase (15), cAMP-activated protein kinase A (16), and cGMP-activated protein kinase (PKG) (17). Hence, several known mechanisms modulating Ca 2ϩ sensitivity in smooth muscle converge on MLCP and CPI-17. The physiological importance of these mechanisms in vivo is not known. Also, it is difficult to appreciate the role of protein kinase A and PKG phosphorylation of CPI-17, because these two kinases are mediators of desensitizing pathways in smooth muscle. CPI-17 also has been shown to have roles in other tissues, e.g. cerebellar signaling (18) and platelet secretion (19).
Cyclic guanosine monophosphate (cGMP) and cGMP-activated kinase (PKG) are major mediators of nitric oxide induced relaxation of smooth muscle. Their effects are because of both decreased intracellular Ca 2ϩ concentration and a decreased sensitivity to Ca 2ϩ (20 -22). The decreased Ca 2ϩ sensitivity in response to cGMP has been shown to be mediated by PKG (22) * This work was supported in part by grants from the Swedish Medical Research Council (04x-8268), The Crafoord Foundation, the Swedish Heart-Lung Foundation, and the Medical Faculty, Lund University. 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  and is associated with the activation of MLCP (23,24). One mechanism contributing to decreased Ca 2ϩ sensitivity in response to cGMP and PKG activation includes the phosphorylation of the small monomeric G-protein RhoA. Phosphorylation of RhoA prevents its membrane binding and activation of Rho kinase (25). PKG also has been shown to phosphorylate the protein telokin, promoting its relaxant effect via MLCP (26). Another proposed pathway for PKG-induced relaxation involves the phosphorylation of the heat shock protein 20 (Hsp20), which has been suggested to act independently of MLC 20 phosphorylation (27,28). PKG is targeted to MYPT1 via a leucine zipper interaction, and dissociation of this complex has been reported to result in a loss of cGMP-induced Ca 2ϩ desensitization (29,30), a mechanism suggested to be dependent on MYPT1 isoforms (31). PKG can phosphorylate MLCP; however. this is not on the inhibitory site and PKG phosphorylation of MLCP does not alter the ability of MLCP to dephosphorylate MLC 20 (32). A more indirect mechanism has been suggested, possibly via a change in affinity of MLCP for phospholipid membranes (32). In vitro, PKG can phosphorylate CPI-17 on Thr-38, which would be associated with increased Ca 2ϩ sensitivity if it were to occur in vivo (17), as discussed above. Recently, it was reported that nitric oxide-mediated relaxation of intact vascular smooth muscle is associated with decreased phosphorylation of CPI-17 (33), raising the question of whether cGMP and PKG can promote smooth muscle relaxation by increasing the rate of dephosphorylation of CPI-17.
The aim of this study was to explore whether cGMP/PKG can relax PKC-sensitized, Rho kinase-independent smooth muscle contraction and whether phosphorylation on residue Thr-38 of CPI-17 is influenced. We provide evidence that actions downstream of cGMP/PKG can reverse PKC-mediated phosphorylation of CPI-17 and Ca 2ϩ sensitization in smooth muscle.

EXPERIMENTAL PROCEDURES
Smooth Muscle Preparations-Young adult female NMRI mice weighing approximately 30 g were used (B&K Universal, Sollentuna, Sweden). The animals were killed by cervical dislocation, and the femoral artery, portal vein, and the distal part of the small intestine were isolated. The longitudinal (external) muscle layer was peeled off from the small intestine, and preparations were made ϳ4 ϫ 1 mm and 0.05 mm thick. The portal vein was dissected free of connective tissue and divided longitudinally in two strips, and the femoral artery was dissected free of connective tissue and cut into 2-mm-long sections using forceps and microscissors under the microscope.
Force Recordings from Permeabilized Preparations-The portal vein and intestinal strips were wrapped at both ends with aluminum foil. The sections of the femoral artery were mounted on two stainless steel wires (diameter 30 m). The wires were bent 90°outside each end of the vessel. The wires again were bent 90°at 1 mm from the vessel and wrapped with aluminum foil. All of the preparations were mounted horizontally between the extended arm of a force transducer (AE 801, SensoNor a.s., Horten, Norway) and a tungsten wire enabling length adjustment. The experiments were performed at room temperature (22°C) in the surface tension bubble of 140 l of solution in small plastic cups. The solution was stirred continuously throughout the experiments using small stirring magnets except during the ␤-escin permeabilization procedure. The strips first were equilibrated for ϳ45 min in HEPES-buffered physiological saline solution (PSS) (see below) before a test contraction using 80 mM KCl was made. Thereafter, the preparations were permeabilized using ␣-toxin or ␤-escin as described below.
Staphylococcus aureus ␣-Toxin Permeabilization-The preparations were permeabilized using the staphylococcus ␣-toxin essentially as described previously (34). The preparations were relaxed in Ca 2ϩ -free HEPES-buffered PSS for 10 min. The preparations then were transferred to the relaxation solution, and after 5 min, they were treated with relaxation solution with S. aureus ␣-toxin (5000 -10,000 units/ml, Calbiochem) for 60 min. Finally, the preparations were kept 20 min in ␣-toxin-free relaxation solution containing 10 M of the calcium ionophore A23187.
␤-Escin Permeabilization-The ␤-escin permeabilization procedure was preformed essentially as described by Otto et al. (35). The preparations were relaxed in Ca 2ϩ -free HEPES-buffered PSS for 10 min, transferred to the relaxation solution for 5 min, and treated with relaxation solution containing 50 M ␤-escin for 30 min. The preparations then were kept 15 min in ␤-escin-free relaxing solution containing 10 M of the calcium ionophore A23187. CPI-17 Protein Analysis-CPI-17 content and phosphorylation were determined in small intestinal muscle preparations subjected to the same ␤-escin permeabilization procedure and experimental protocol as shown in the force experiments. The preparations were frozen using clamps precooled in liquid N 2 and immediately homogenized in SDS sample buffer (see below) in glass homogenizers at 4°C, boiled for 1 min, centrifuged briefly, and separated on 15% polyacrylamide gels. The samples from the different treatment conditions were run in identical concentrations on two gels in parallel, one that was used to determine the total amount of CPI-17 and the other that was used to determine the amount of phosphorylated CPI-17. After transfer to the polyvinylidene difluoride membranes (Bio-Rad), one membrane was incubated with a rabbit polyclonal antibody against CPI-17 (Upstate Group, Inc.) and the other was incubated with a polyclonal antibody, raised in goat, using a peptide of CPI-17 phosphorylated at Thr-38 (sc-17560, Santa Cruz Biotechnology) (for the specificity of these antibodies, see Ref. 36). The antibody binding was detected using the enhanced chemiluminescence method (ECL, Santa Cruz Biotechnology). The ECL intensity of the gels was analyzed using Quantity One software (Bio-Rad). The extent of CPI-17 phosphorylation was determined as the ratio of the ECL reaction from phosphorylated CPI-17 relative to total CPI-17. To obtain a measurement of the content of CPI-17 in intact and in ␤-escin-permeabilized small intestinal preparations relative to intact aorta, Coomassie blue-stained gels and Western blots were performed in parallel. The CPI-17 fluorescence of the ECL reaction was expressed relative to the actin intensity on the Coomassie blue-stained gel.
Statistics-All of the values are given as the mean Ϯ S.E. of n observations. All of the curve fitting and statistical analysis was performed using SigmaPlot 8.0 and SigmaStat 3.0 software (SPSS Inc.).

RESULTS
The smooth muscle strips from ileum and portal vein permeabilized with S. aureus ␣-toxin exhibited a force at saturating [Ca 2ϩ ] of ϳ0.4 milli Newton. This was equivalent to ϳ 100% of the initial high K ϩ (80 mM KCl) induced responses in the intact muscle. The corresponding values for the femoral artery were ϳ0.2 milli Newton and ϳ120%. The force at saturating [Ca 2ϩ ] in ␤-escin-permeabilized intestinal muscle strips was ϳ80% of the high K ϩ -induced force.
Dose-response curves for ␤-escin-permeabilized small intestinal muscle strips to Ca 2ϩ in the presence of 0.5 M calmodulin were sigmoidal with a half-maximal response at a pCa of ϳ6.0. If these ␤-escin-permeabilized strips were subjected to an intermediate Ca 2ϩ concentration (pCa 5.9) giving ϳ70% of maximal force and then challenged with 30 M of the Rho kinase blocker Y27632, they relaxed slightly by ϳ15% (Fig. 1). At pCa 6.1, the force was ϳ20% maximal force and slightly inhibited by Y27632 (relaxation ϳ 10%). In absence of Y27632, at this Ca 2ϩ level, GTP␥S (10 M) increased the force markedly to ϳ60% of the maximum (pCa 4.5). The GTP␥S responses were abolished completely if 30 M Y27632 was present or added. The protein kinase C activator PDBu (1 M) increased the force at a constant suboptimal Ca 2ϩ concentration in the presence of 30 M Y27632, both in ␤-escinand ␣-toxin-permeabilized small intestinal muscle strips as shown in Fig. 2.
The effects of 8-Br-cGMP were compared in ␤-escin-permeabilized small intestinal strips under three different conditions, all giving ϳ50% of the maximal force: 1) unsensitized condition (pCa 5.9 with Rho kinase inhibitor, 30 M Y27632); 2) Rho-Rho kinase-sensitized condition (pCa 6.1 with 10 M GTP␥S); and 3) PKC-sensitized condition (pCa 6.0 with Rho kinase inhibitor, 30 M Y27632, and PKC activator, 1 M PDBu). The initial force values under these conditions are shown in Fig. 3, panel A. The extent of relaxation after the addition of 100 M 8-Br-cGMP is shown in Fig. 3, panel B. The effect of 8-Br-cGMP on unsensitized muscle (condition 1, above) (pCa 5.9 with Rho kinase inhibitor Y27632) was marginal. As shown in Fig. 3, panel B, 8-Br-cGMP significantly reduced tension in muscles sensitized with GTP␥S or PDBu.
To explore whether the relaxation of PKC-induced responses by 8-Br-cGMP also was present using a different permeabilization method of the small intestinal strips, we performed experiments using the ␣-toxin-permeabilized tissue. In ␣-toxinpermeabilized muscle strips contracted at pCa 6.0, treated with 30 M Y27632, and sensitized with 1 M PDBu (same protocol as in Fig. 2, panel B), 1 M 8-Br-cGMP induced the relaxation (Fig. 4, panel A). There was no further relaxation if 100 M 8-Br-cGMP was added.
To examine whether the 8-Br-cGMP-induced relaxation of PKC-sensitized smooth muscle was present also in other types of smooth muscle, the experiments on PKC-sensitized muscle (condition 3, above) were performed on ␣-toxin-permeabilized preparations of the portal vein (Fig. 4, panel B) and femoral artery (Fig. 4, panel C). The muscles were activated at a [Ca 2ϩ ], giving ϳ50% of maximal force (pCa 6.0 for portal vein and pCa 6.3 for femoral artery). In all of the three tissues examined, PDBu induced an increase in force in the presence of Y27632, which could be relaxed by 1 M 8-Br-cGMP. The femoral artery was more responsive to PDBu and 1 M PDBu resulted in a maximal contraction that could not be influenced by 8-Br-cGMP. However, as shown in Fig. 4, panel C, a lower concen-tration of PDBu (30 nM) induced a potentiation of force that was inhibited fully by 8-Br-cGMP. Fig. 5 summarizes these results and shows that 8-Br-cGMP relaxes PKC-sensitized preparations of these different muscle types. The time to reach halfmaximal relaxation in response to 8-Br-cGMP was 46 Ϯ 5 s in the small intestine, 105 Ϯ 9 s in the portal vein, and 424 Ϯ 11 s in the femoral artery (n ϭ 4).
Western blot protein analysis was performed on ␤-escinpermeabilized small intestinal muscle strips to examine whether CPI-17 was present and whether its state of phosphorylation was altered in response to 8-Br-cGMP. The CPI-17/ actin ratio in intact ileum preparations compared with intact aorta was 0.22 Ϯ 0.06 (n ϭ 6). There was no detectable difference in CPI-17 content after the ␤-escin permeabilization procedure (data not shown). The preparations were subjected to the same protocol as shown in Fig. 4, adding 1 or 100 M 8-Br-cGMP at 5 min after addition of PDBu. To determine CPI-17 phosphorylation, ␤-escin-permeabilized preparations were frozen rapidly 10 min after the addition of 8-Br-cGMP. Control samples did not receive cyclic nucleotide but were frozen at the same point in time, 15 min after the addition of PDBu. Relative CPI-17 phosphorylation was decreased significantly after the addition of 1 or 100 M 8-Br-cGMP compared with controls (Fig. 6, panels A and B). In experiments following the same protocol, with the exception that 1 M microcystin-LR was added together with PDBu, 8-Br-cGMP failed to decrease CPI-17 (Thr-38) phosphorylation. Instead, we noted a slight but not significant increase in CPI-17 (Thr-38) phosphorylation compared with microcystin-LR-treated controls not receiving 8-Br-cGMP (Fig. 6, panel C). To examine whether 8-Br-cGMP reduced CPI-17 phosphorylation via the inhibition of PKC, we used the protocol above in ␤-escin-permeabilized muscle at a high M) with and without 8-Br-cGMP were introduced 5 min after the addition of Y27632. The preparations were frozen immediately before and 3 and 10 min after exposure to PDBu. CPI-17 phosphorylation increased with time after the addition of PDBu in the presence of Y27632 and microcystin-LR. We observed no difference in CPI-17 (Thr-38) phosphorylation over time between 8-Br-cGMP-treated preparations and untreated controls as illustrated in Fig. 7. DISCUSSION We show that 8-Br-cGMP can relax PKC-mediated Ca 2ϩsensitized contraction of the smooth muscle. This relaxation is coupled with increased dephosphorylation of CPI-17 (at residue Thr-38) and can occur in a situation when Rho kinase activity is blocked. We conclude that PKG can activate MLCP and desensitize contraction to Ca 2ϩ via a mechanism independent of the Rho-Rho kinase pathway.
Activation of Rho-Rho kinase is one of the main mechanisms for increased Ca 2ϩ sensitivity of smooth muscle following receptor activation (37). In this study, this pathway has been shown to have a significant role in Ca 2ϩ sensitization in several smooth muscle tissues including the small intestine ( Fig. 1) (38,39). Phosphorylation of the MYPT1 of MLCP has been identified as one possible mechanism of Rho kinase-induced inhibition of MLCP (6). However, the phosphorylation of MYPT at the relevant site (Thr-695) does not always change during Rho-Rho kinase-induced Ca 2ϩ sensitization (40). Other targets for Rho kinase have been identified recently, e.g. CPI-17 (36).
PKG has been shown to interact with the Rho-Rho kinase pathway by phosphorylating RhoA on Ser-188, thereby inhibiting its membrane interaction and its ability to activate Rho kinase (25). This site also can be phosphorylated by protein kinase A (41, 42). Our result that 8-Br-cGMP relaxes the ten- The muscles were activated as shown in Fig. 4, and the extent of relaxation is expressed relative to the force prior to the addition of the cyclic nucleotide (n ϭ 4 -6).
sion of GTP␥S-sensitized muscle to a much larger extent than unsensitized muscle is consistent with this model where PKG can act via RhoA. However, as discussed below, the RhoA phosphorylation mechanism cannot explain our result that 8-Br-cGMP can relax PKC-sensitized muscle in the presence of a large dose of the Rho kinase blocker Y27632 (30 M).
The CPI-17 content expressed relative to calponin has been shown to vary among smooth muscles (44). The contents in rat vas deferens and urinary bladder were ϳ20% of that in the aorta. We show that the CPI-17/actin ratio in mouse small intestinal smooth muscle was ϳ22% of that in the aorta, which is consistent with findings that CPI-17 is present in visceral smooth muscle, but in lower concentrations than in large arteries (44). The functional role of CPI-17 in intestinal smooth muscle is not known. Interestingly, the expression of CPI-17 has been shown recently to decrease following interleukin-1 treatment of rat small intestinal muscle (45). The contractility was altered, suggesting that CPI-17 might be involved in the regulation of gastrointestinal motility and that this protein might be altered in inflammatory conditions of the bowel. CPI-17 is a comparatively small protein, which is completely lost after extensive Triton X-100 permeabilization and partly lost after extensive ␤-escin permeabilization, and maintained in ␣-toxin-permeabilized tissue (10). We have used a gentle ␤-escin permeabilization procedure of the small intestinal muscle and show that the CPI-17 content is maintained and similar to that of the intact preparations. Further, the PKC (PDBu) potentiated contractions were of similar magnitude as in the ␣-toxin-permeabilized tissue. We show that CPI-17 phosphorylation increases during PDBu-induced Ca 2ϩ sensitization in the permeabilized mouse small intestinal smooth muscle. These results show that CPI-17 is present and that the PKC/ CPI-17 pathway for Ca 2ϩ sensitization is operating in parallel with the Rho-Rho kinase pathway in the small intestinal smooth muscle.
The main effects of cGMP are considered to be mediated by PKG. This cyclic nucleotide has been suggested to be able to cross-activate protein kinase A at high concentrations (46 -48). In longitudinal smooth muscle from the mouse small intestine, this kind of cross-talk can contribute to the relaxant effect of 8-Br-cGMP at higher concentrations (22). However, at 1 M, these effects seem to be minimal (22) and we suggest that cGMP via PKG is the operating mechanism for reversing PKCinduced Ca 2ϩ sensitization of contraction in smooth muscle.
One mechanism for CPI-17 inhibition of MLCP seems to be the binding of phosphorylated CPI-17 to the catalytic subunit of MLCP (PP1M C ) (49). Because the similarities in the region surrounding the phosphorylatable residues exist between the regulatory light chain and CPI-17, it has been suggested that CPI-17 acts as an inhibiting pseudosubstrate for MLCP (49). This is consistent with a mixed, competitive and non-competitive inhibition (49). The negative charge, introduced by the phosphorylation of CPI-17 on Thr-38, influences the conformation of CPI-17, although the phosphorous group is required for full inhibitory action (50). In histamine-activated intact swine carotid media, the nitric oxide donor, sodium nitroprusside, reduced force and CPI-17 phosphorylation (33). This finding suggests that alterations in CPI-17 phosphorylation can be a mechanism involved in cGMP/PKG-induced relaxation of intact smooth muscle. However, in intact muscle, CPI-17 phosphorylation could be influenced by several mechanisms including altered PKC activity due to changes in intracellular [Ca 2ϩ ], PKG phosphorylation of RhoA, diminishing Rho kinase activity, and possibly other mechanisms, e.g. cGMP/PKG actions influencing CPI-17 dephosphorylation. In this study we used permeabilized tissue, clamped [Ca 2ϩ ], and Rho kinase inhibition and report that 8-Br-cGMP/PKG inhibits PKC-potentiated force, concomitant with dephosphorylation of CPI-17. The direct measurements of PKC-induced CPI-17 phosphorylation show that 8-Br-cGMP/PKG does not inhibit PKC. Thus, the reduction of CPI-17 phosphorylation by 8-Br-cGMP reflects an increased rate of dephosphorylation. We show the relaxant effect of 8-Br-cGMP/PKG on PKCinduced contractions in a tissue where both Rho-Rho kinase and PKC sensitize the muscle to Ca 2ϩ (Figs. 1 and 2). The relative ratio of CPI-17/MYPT1 correlates with the extent of PKC (PDBu-activated) potentiation of contraction (44). In this aspect, the small intestine has a comparatively low CPI-17 content and PKC responses. Therefore, it is possible that the effects of 8-Br-cGMP/PKG on PKC-sensitized muscle vary among smooth muscle tissues. To evaluate this possibility, we examined whether 8-Br-cGMP also could relax PKC-sensitized muscles with higher CPI-17 content, the portal vein, and the femoral artery (44). In both these tissues, 8-Br-cGMP/PKG induced the relaxation of Rho-Rho kinase-independent PKCsensitized muscle. This finding shows that this cGMP/PKG relaxation mechanism is functional in vascular as well as in visceral smooth muscle. The femoral artery exhibited prominent PKC-induced responses, most probably reflecting a very high CPI-17 content. At high [PDBu] (1 M), the responses could not be reversed by 8-Br-cGMP. However, at a lower PKC activation level (30 M PDBu), 8-Br-cGMP induced relaxation. This finding is consistent with a model where actions downstream of 8-Br-cGMP/PKG decrease the CPI-17-P/MYPT1 ratio. In this model, high total CPI-17 content, low MYPT1 content, and high activity of PKC would slow this relaxant effect of cGMP/PKG. CPI-17 has been shown to be dephosphorylated in vitro by phosphatases of types PP2A and PP2C but not by type PP1, e.g. MLCP (PP1M) (51). We show that microcystin-LR inhibits the PKG-influenced dephosphorylation of CPI-17, which suggests that PP2A is involved in CPI-17 dephosphorylation, because microcystin-LR primarily acts on PP2A and PP1 (52). Small changes in the CPI-17 structure can remove its resistance to dephosphorylation by the PP1-type MLCP (53). PKG is associated closely with the MLCP complex, and PKG binding to MLCP has been reported to be a necessary step in cGMP/PKGinduced Ca 2ϩ sensitization (29). PKG has been shown to phosphorylate the carboxyl terminus of MYPT1 at Ser residues, which does not influence directly the ability of MLCP to dephosphorylate MLC 20 but could be involved via a more indirect mechanism (32). One possibility is that PKG alters the structure of CPI-17 or MLCP and thereby increases the CPI-17 dephosphorylation activity of MLCP.
In conclusion, PKG can reverse Ca 2ϩ sensitization by increasing the rate of CPI-17 dephosphorylation. This relaxation mechanism would have a prominent role in agonist-induced contractions involving the activation of PKC. Considering the large number of kinases acting on CPI-17, this mechanism could be a more general component of nitric oxide-mediated relaxation.