Signal transduction by G-proteins, Rho-kinase and protein phosphatase to smooth muscle and non-muscle myosin II

Abstract

We here review mechanisms that can regulate the activity of myosin II, in smooth muscle and non-muscle cells, by modulating the Ca²⁺ sensitivity of myosin regulatory light chain (RLC) phosphorylation. The major mechanism of Ca²⁺ sensitization of smooth muscle contraction and non-muscle cell motility is through inhibition of the smooth muscle myosin phosphatase (MLCP) that dephosphorylates the RLC in smooth muscle and non-muscle. The active, GTP-bound form of the small GTPase RhoA activates a serine/threonine kinase, Rho-kinase, that phosphorylates the regulatory subunit of MLCP and inhibits phosphatase activity. G-protein-coupled release of arachidonic acid may also contribute to inhibition of MLCP acting, at least in part, through the Rho/Rho-kinase pathway. Protein kinase C(s) activated by phorbol esters and diacylglycerol can also inhibit MLCP by phosphorylating and thereby activating CPI-17, an inhibitor of its catalytic subunit; this mechanism is independent of the Rho/Rho-kinase pathway and plays only a minor, transient role in the G-protein-coupled mechanism of Ca²⁺ sensitization. Ca²⁺ sensitization by the Rho/Rho-kinase pathway contributes to the tonic phase of agonist-induced contraction in smooth muscle, and abnormally increased activation of myosin II by this mechanism is thought to play a role in diseases such as high blood pressure and cancer cell metastasis.

Myosin II, the major molecular motor of muscle and most non-muscle cells, is regulated not only by fluctuations in cytoplasmic calcium ([Ca²⁺]_i), but also by other important signalling mechanisms. Thus, whereas excitation- contraction coupling in vertebrate striated muscles is under membrane potential control and contraction is initiated by binding of Ca²⁺ to a thin (actin) filament-associated protein, troponin, contractility of smooth muscle is regulated not only by such electromechanical coupling and [Ca²⁺]_i, but also by membrane potential-independent, pharmacomechanical coupling (Somlyo & Somlyo, 1968; reviewed in Somlyo & Somlyo, 1994; Somlyo et al. 1999a). Furthermore, both smooth muscle and non-muscle myosin II are regulated by phosphorylation/dephosphorylation of the myosin regulatory light chain (RLC) by, respectively, Ca²⁺-calmodulin-regulated myosin light chain kinase (MLCK) and myosin phosphatase (MLCP, also known as SMPP-1M); actin activation of myosin II is the result of RLC phosphorylation on Ser 19 (Adelstein & Conti, 1975; reviewed in Hartshorne, 1987; Tan et al. 1992; Somlyo & Somlyo, 1994; Gallagher et al. 1997). Therefore, it was to be expected that the Ca²⁺-independent mechanisms that regulate smooth muscle myosin II will also regulate non-muscle myosin II and non-muscle motility (Somlyo & Somlyo, 1994). The small GTPase, RhoA, and its upstream activators and downstream effectors play a major role in these processes and are the subject of this brief review.

Increased RLC (Ser 19) phosphorylation of smooth and non-muscle myosin II can be effected not only by increasing [Ca²⁺]_i and, thereby, the activity of MLCK, but also by inhibiting MLCP. This was the basis of the suggestion (Somlyo et al. 1989), verified experimentally (Kitazawa et al. 1991), that contractions induced at constant [Ca²⁺]_i by certain agonists and by GTPγS were due to inhibition of a myosin phosphatase. The inhibitory signal for ‘Ca²⁺ sensitization’ is communicated by RhoA to a Rho-kinase that phosphorylates the M_110-130 regulatory subunit and inhibits the catalytic activity of MLCP, resulting in increased RLC phosphorylation, contraction and cell motility (Fig. 1). Experimental evidence that led to these conclusions is summarized below.

Figure 1. Regulation of myosin II in smooth and non-muscle cells.

Pathways indicated in red activate myosin II, resulting in contraction, cell migration and cancer metastasis. Pathways that reduce myosin II activity are shown in blue. The major, Ca²⁺-independent pathway that increases myosin II activity is through activation of Rho-kinase by RhoA^.GTP, phosphorylation of the regulatory subunit of myosin phosphatase (MLCP) by Rho-kinase and some other kinases (see text), resulting in inhibition of MLCP activity and increased myosin RLC phosphorylation. Increases in arachidonic acid, due to a variety of stimuli, can also activate Rho-kinase and, at least in vitro, also inhibit MLCP activity by dissociating the regulatory (M_110-130) from the catalytic (PP-1C) subunit. The third pathway shown that enhances myosin II activity is through phosphorylation of CPI-17 by protein kinase C (PKC) leading to direct inhibition of PP-1C by the phosphorylated CPI-17 (CPI-17.P). Each of these mechanisms requires the presence of an active kinase that can phosphorylate Ser19 of RLC and increase myosin II activity by inhibiting MLCP even at constant [Ca²⁺]_i (‘Ca²⁺ sensitization’). These mechanisms operate in parallel with and independently of the activation of MLCK by Ca²⁺ released from the sarcoplasmic reticulum/endoplasmic reticulum (SR/ER) by InsP₃ or by Ca²⁺ influx (not shown). Abbreviations: PLC, phospholipase C; PiP₂, phosphatidylinositol-bis-phosphate; InsP₃, inositol 1,4,5-trisphosphate; DAG, diacylglycerol; PKC, protein kinase C; GEF, guanine nucleotide exchange factor; MLCK, myosin light chain kinase; Cam, calmodulin; RLC, regulatory light chain; C3, Clostridium botulinum, exoenzyme that ADP-ribosylates Asn 41 of RhoA, inhibits RhoA activity and blocks its translocation to the membrane; Y-27632, selective inhibitor of Rho-kinase.

Although G-protein-coupled regulation of a myosin phosphatase was recognized over a decade ago (Somlyo et al. 1989), the regulated phosphatase (MLCP) that dephosphorylates the RLC of intact myosin was identified only recently. It consists of a 110–130 kDa regulatory (M_110-130), an ∼37 kDa catalytic (PP-1C; variously described as β or δ) subunit, and a 20 kDa subunit of unknown function (Alessi et al. 1992; Shirazi et al. 1994; Shimizu et al. 1994; Haystead et al. 1995; reviewed in Hartshorne et al. 1998). It is also present in non-muscle (e.g. Nakai et al. 1997; Murányi et al. 1998; Suzuki et al. 1999; Essler et al. 1999; reviewed in Hartshorne et al. 1998), including human prostate cancer cells (Somlyo et al. 1999b). Selectivity of PP-1C for myosin is conferred by the M_110-130 subunit that potentiates dephosphorylation of myosin by PP-1C, in vitro (Alessi et al. 1992; Shirazi et al. 1994; Shimizu et al. 1994; reviewed in Hartshorne et al. 1998) and in permeabilized smooth muscle (Gailly et al. 1996). Phosphorylation of M_110-130 in its C-terminal half by Rho-kinase (Kimura et al. 1996) and some other kinase(s) on the same site (Feng et al. 2000) inhibits MLCP activity. Certain N-terminal fragments of M_110-130 can cause Ca²⁺ sensitization, probably through competitive inhibition of the endogenous protein (Zhou et al. 1999).

Agonists acting on receptors coupled to G_q induce both Ca²⁺ sensitization and activation of phospholipase C (PLC)-mediated hydrolysis of phosphatidylinositol-bis-phosphate to inositol-1,4,5-trisphosphate (InsP₃) and diacylglycerol to, respectively, cause Ca²⁺ release and activate protein kinase C (PKC); the concurrence of these effects implicated Gα_q in both processes (Somlyo & Somlyo, 1994). However, Ca²⁺ release can be dissociated from Ca²⁺ sensitization (Kobayashi et al. 1991), and InsP₃ does not Ca²⁺ sensitize smooth muscle. These results exclude PLC products as major Ca²⁺-sensitizing effectors of Gαq, consistent with the view that conventional and novel PKCs (for historical antecedents, see Andrea & Walsh, 1992; Somlyo et al. 1999a) play only a small and transient (Iizuka et al. 1999) or no role in G-protein-coupled Ca²⁺ sensitization (Jensen et al. 1996; Walker et al. 1998; Strassheim et al. 1999). Furthermore, some very potent Ca²⁺-sensitizing agonists, such as U-46619, cause little or no detectable release of intracellular Ca²⁺ (Bradley & Morgan, 1987; Himpens & Somlyo, 1988; Himpens et al. 1990), as would be expected of receptors coupled to Gα_q/11. Subsequent studies of non-muscle cells revealed that RhoA can be activated by α subunits of other trimeric G-proteins, Gα_12,13, via linked guanine nucleotide exchange factors (GEFs; Hart et al. 1998; Kozasa et al. 1998; Gohla et al. 1999). The receptors activated by U-46619 are coupled to G_12,13 that, unlike Gα_q/11, do not activate PLC, suggesting that the disproportionately high Ca²⁺-sensitizing (compared with Ca²⁺ releasing) effect of U-46619 (Himpens et al. 1990) is mediated by the Gα_12,13 family. A Gα-GAP (GTPase-activating protein), p¹¹⁵ Rho-GEF, interacts with Gα₁₃ and is probably the upstream ‘convector’ between this trimeric G-protein and RhoA (Kozasa et al. 1998; Hart et al. 1998). It now appears that several trimeric G-proteins, including Gα_q, Gα_12,13 and Gα_i-2, can activate RhoA, depending on the receptors and cell types involved (Katoh et al. 1998; Croxton et al. 1998; Klages et al. 1999; Hirshman & Emala, 1999).

RhoA, a monomeric G-protein, is, like most GTPases, active when it contains bound GTP and inactive when the bound nucleotide is GDP. RhoA.GTP does not directly inhibit MLCP, as indicated by its lack of effect on extensively permeabilized (with Triton X-100) smooth muscle (Gong et al. 1996), and Ca²⁺ sensitization requires its translocation to a relatively intact plasma membrane (Gong et al. 1997; Fujihara et al. 1997; Taggart et al. 1999). In resting smooth muscle (Gong et al. 1997) as in other cells (Bourmeyster et al. 1992; Abo et al. 1994; Bokoch et al. 1994), the cytosolic, inactive forms of RhoA and Rac are complexed with RhoGDI (guanine nucleotide dissociation inhibitor). The hydrophobic geranyl-geranylated tail of RhoA inserts into a hydrophobic cavity of GDI and binding is reinforced by protein-protein interactions between the N-terminus of GDI and a highly conserved (among Rho-family proteins) epitope of RhoA (Fig. 2; Gosser et al. 1997; Longenecker et al. 1999 and references therein). Cytosolic RhoA.RhoGDI is activated by Rho-GEFs (guanine nucleotide exchange factors) that stimulate nucleotide exchange on RhoA (GTP replaces GDP), followed by dissociation of RhoA.GTP from the complex and translocation to the plasma membrane, while GDI is retained in the cytosol (Gong et al. 1997; Fujihara et al. 1997; Read et al. 2000; reviewed in Cherfils & Chardin, 1999).

Figure 2. Structural features of the complex between RhoA and RhoGDI.

A, ribbon drawing of the crystal structure of RhoA highlighting the switch 1 (blue) and switch 2 (magenta) regions (Wei et al. 1997). B, crystal structure of the RhoA-RhoGDI complex (Longenecker et al. 1999) reveals the mutual disposition of the individual structures of RhoA and RhoGDI. The length of the complex is approximately 77 Å (7.7 nm). The surface of RhoGDI is coloured according to resonance shifts observed in solution by NMR (Gosser et al. 1997). RhoGDI residues affected by addition of Cdc42 (unprenylated) are coloured yellow, and those affected by addition of a prenylated peptide are red. C and D, models (orange) of the N-terminal domain of RhoGDI (residues 24–68) and of the C-terminal extension of RhoA (residues 181–190) display additional structural features of the complex. The conservation of amino acids among the Rho-family (rho, rac, Cdc42) is depicted on the surface of RhoA as a gradation of blue, where dark blue represents 100 % conservation. The large conserved surface is consistent with the ability of RhoGDI to bind each of the above Rho-family GTPases. A ribbon drawing of RhoGDI is coloured as in B. The views in B and D are rotated about the vertical axis relative to A and C.

Bacterial exoenzymes that ADP-ribosylate Asn 41 (C3, EDIN) or monoglucosylate Thr 37 (C. difficile toxin B) of RhoA (reviewed in Sehr et al. 1998) block its biological activity; ADP-ribosylation also prevents its translocation to the plasma membrane (Gong et al. 1997; Fujihara et al. 1997; Croxton et al. 1998). These agents inhibit Ca²⁺ sensitization of RLC phosphorylation and force in smooth muscle (Noda et al. 1995; Itagaki et al. 1995; Gong et al. 1996; Fujihara et al. 1997; Otto et al. 1996) and stress fibre formation in non-muscle cells (reviewed in Hall, 1994).

The Ca²⁺-sensitizing effector of RhoA.GTP is a serine/ threonine-kinase (we refer to it generically as ‘Rho-kinase’) that contains a Rho-binding domain, has two identified isoforms and is activated by Rho.GTP (Leung et al. 1995; Ishizaki et al. 1996; Matsui et al. 1996). Both α and β isoforms are present in smooth muscle (Yoshii et al. 1999), with ROKα (ROCK-II) predominant in gizzard (Feng et al. 1999). The detailed mechanism of Rho-kinase activation by RhoA.GTP is not known. However, the requirement for RhoA to translocate to the plasma membrane to cause Ca²⁺ sensitization and the lack of Ca²⁺ sensitization by a recombinant, non-prenylated (E. coli-expressed) RhoA.GTP that does not bind to the membrane (Gong et al. 1996), suggest that activation of Rho-kinase by RhoA.GTP, in analogy with the Ras-activated Raf-kinase, occurs upon recruitment of both proteins to the plasma membrane (Leung et al. 1995). Activated Rho-kinase phosphorylates the regulatory subunit of MLCP and inhibits myosin phosphatase activity (Kimura et al. 1996). MLCP can also be phosphorylated on the same inhibitory site (Feng et al. 2000) and inhibited by other, yet unidentified kinases (Trinkle-Mulcahy et al. 1995; Ichikawa et al. 1996; reviewed in Hartshorne et al. 1998).

In addition to RhoA.GTP, arachidonic acid, a Ca²⁺-sensitizing agent released by certain agonists (Gong et al. 1995; Gailly et al. 1997), can also activate Rho-kinase (Fu et al. 1998; Feng et al. 1999) and may contribute an ancillary pathway of RhoA-mediated Ca²⁺ sensitization. In vitro, arachidonic acid dissociates the regulatory from the catalytic subunit of MLCP (Gong et al. 1992) leading to a several-fold reduction in myosin-targeted phosphatase activity.

The Ca²⁺-sensitizing Rho/Rho-kinase pathway can be inhibited at several stages: at the receptor by antagonists to the activating agonist, by hydrolysis of bound GTP to GDP facilitated by GAPs, by complexation of free RhoA with GDI (for smooth muscle, see Longenecker et al. 1999), through deactivation/inhibition of Rho-kinase, and by site-specific dephosphorylation of M_110-130. RhoA-induced Ca²⁺ sensitization can also be inhibited by Rnd1, a physiological RhoA antagonist that inhibits Ca²⁺ sensitization induced by carbachol, GTPγS, and recombinant RhoA (Loirand et al. 1999). Rnd1 is a prenylated Rho-related protein that is constitutively GTP bound and associated with the cell membrane; its expression is stimulated by progesterone or oestrogen. It is, therefore, important to realize that contractions initiated, maintained and terminated through the RhoA/Rho-kinase pathway are the result of a multiple-step process having complex kinetics.

Feedback inhibition of Rho/Rho-kinase-mediated Ca²⁺ sensitization may occur through inhibition of MLCK, because GTPγS and carbachol also increase MLCK phosphorylation of an inhibitory site (Tang et al. 1993). This may be due to MLCK's inhibitory site being dephosphorylated by MLCP or by another protein phosphatase that is inhibited by a G-protein-coupled mechanism, or to phosphorylation and inhibition of MLCK by PAK, a serine/threonine kinase activated by other Rho-subfamily GTPases, Cdc42 and Rac (Sanders et al. 1999). Val 12 Cdc42 and L61 Rac1 (constitutively active mutants) can inhibit Ca²⁺ sensitization of force induced by constitutively active Val 14 RhoA in permeabilized smooth muscle (M. Gong, P. Read, R. Nakamoto, A. V. Somlyo & A. P. Somlyo, unpublished observation). It remains to be determined whether these effects of Rac and Cdc42 are due to activation of PAK.

The Rho/Rho-kinase pathway plays an important physiological role in intact smooth muscle. The tonic phase of agonist-induced contractions, previously ascribed solely to maintained Ca²⁺ influx, is inhibited by a Rho-kinase inhibitor (Uehata et al. 1997; Fu et al. 1998), by a cell permeant chimera (DC3B) of C3 that selectively inactivates RhoA (Fujihara et al. 1997) and by the somewhat less selective toxin B, without any accompanying decrease in [Ca²⁺]_i (Lucius et al. 1998). These results suggest that the slow (tonic) phase of contractions induced by carbachol in intestinal smooth muscles, unaccompanied by increased [Ca²⁺]_i (Himpens & Somlyo, 1988), reflects the time course of Ca²⁺ sensitization by RhoA/Rho-kinase-mediated inhibition of myosin phosphatase. Identification of physiological and pathological functions of this pathway is greatly facilitated by use of the highly selective, cell permeant Rho-kinase inhibitor Y-27632 (Uehata et al. 1997; Fu et al. 1998; Yoshii et al. 1999).

Rho-kinase regulates myosin II in both smooth muscle and non-muscle by inhibiting dephosphorylation of myosin RLC, but the activity of MLCP can also be modulated by at least two other mechanisms. Protein kinase C (PKC), activated by phorbol esters or diacylglycerol (Jensen et al. 1996; Walker et al. 1998 and references therein), and a constitutively active PKC (Ikebe & Brozovich, 1996) can enhance contraction at constant [Ca²⁺]_i by inhibiting myosin phosphatase directly or by phosphorylating an inhibitor (Somlyo et al. 1989; Itoh et al. 1993). PKC phosphorylates CPI-17, a potent inhibitor (when phosphorylated) of PP-1C (Eto et al. 1995; Kitazawa et al. 1999; Senba et al. 1999). Rho-kinase and PKC can inhibit MLCP, through convergent mechanisms, in both smooth muscle (Jensen et al. 1995) and non-muscle cells (Strassheim et al. 1999), with PKC playing a minor and transient (Iizuka et al. 1999) role in contractile regulation, its extent possibly depending on the agonist and/or cell type involved. Phorbol ester-induced Ca²⁺ sensitization is not inhibited by Y-27632 (Fu et al. 1998). Conversely, cyclic nucleotide (cAMP or cGMP)-activated kinases, possibly through the activity of phosphorylated telokin, accelerate dephosphorylation of RLC (Wu et al. 1998 and references therein) leading to muscle relaxation. RLC phosphorylation and contraction can be induced in the absence of Ca²⁺ by phosphatase inhibitors, such as microcystin or calyculin (e.g. Gong et al. 1995; Walker et al. 1998; Kolodney et al. 1999; Weber et al. 1999), and are inhibited by a promiscuous kinase inhibitor, staurosporin, but not by Y-27632 (Kureishi et al. 1999; Iizuka et al. 1999). The Ca²⁺-independent kinases(s) mediating such phosphorylation remain to be identified (Somlyo, 1999). Direct, in vitro phosphorylation of RLC by Rho-kinase (Amano et al. 1996) plays no significant role in vivo, as in the absence of [Ca²⁺]_i to activate MLCK even massive stimulation of the Rho/Rho-kinase pathway by GTPγS induces minimal and very slow or no RLC phosphorylation and contraction of smooth muscle (Somlyo et al. 1989; Iizuka et al. 1999; Swärd et al. 2000). G-protein-coupled Ca²⁺ sensitization requires the presence of an active, but not necessarily Ca²⁺-activated, myosin light chain kinase (Iizuka et al. 1999), but the Ca²⁺-independent kinase (Weber et al. 1999; reviewed in Somlyo, 1999a) is not Rho-kinase (Kureishi et al. 1999; Iizuka et al. 1999).

Pathological activity of Rho-kinase in smooth muscle has been implicated in experimental hypertension (Uehata et al. 1997) and asthma (Chiba et al. 1999). The Rho-kinase inhibitor Y-27632 reduces blood pressure of hypertensive rats, without affecting normal blood pressure (Uehata et al. 1997). In (experimental) asthma muscarinic Ca²⁺ sensitization of intrapulmonary bronchial smooth muscle is enhanced and expression of RhoA is increased (Chiba et al. 1999). Tumour necrosis factor, implicated in asthma, also increases Ca²⁺ sensitivity of bronchial smooth muscles (Parris et al. 1999), but the participation of Rho in this process is not yet known. Thrombin-stimulated vascular smooth muscle migration (Seasholtz et al. 1999) and endothelial contraction (Essler et al. 1999) are also inhibited by C3 and by Y-27632, suggesting that the Rho/Rho-kinase pathway may play a role in atheromatous plaque formation and post-angioplasty re-stenosis.

The importance of Rho family proteins in cytoskeletal organization of non-muscle cells and the specific role of RhoA.GTP in stimulating stress-fibre formation have been known for some time (Paterson et al. 1990; reviewed in Hall, 1994; Lim et al. 1996; Narumiya et al. 1996), although the connection between these effects and inhibition of myosin phosphatase was recognized only as the result of observations on smooth muscle (reviewed in Somlyo & Somlyo, 1994; Somlyo et al. 1999a). An ever increasing literature now documents the important role of the Rho/Rho-kinase-myosin phosphatase inhibition pathway in a variety of non-muscle cells, including platelets (Suzuki et al. 1999), neuronal (Majumdar et al. 1998) and endothelial cells (Vouret-Craviari et al. 1998; Essler et al. 1998). A highly important pathological consequence of inhibition of myosin phosphatase by Rho-kinase is its enhancement of the motility of malignant cells: their migration, transcellular invasion and metastasis, effects that can be inhibited by the Rho-kinase inhibitor Y-27632 (Itoh et al. 1999; Somlyo et al. 1999b).

We conclude that the RhoA/Rho-kinase pathway plays major physiological and pathophysiological roles through modulation of the activity of smooth and non-muscle myosin II, by inhibiting, as originally suggested (Somlyo et al. 1989), the dephosphorylation of myosin II.