The degree of phosphorylation of myosin light chain 20 (MLC20) is a major determinant of force generation in smooth muscle. Myosin phosphatases (MPs) contain protein phosphatase (PP) 1 as catalytic subunits and are the major enzymes that dephosphorylate MLC20. MP regulatory targeting subunit 1 (MYPT1), the main regulatory subunit of MP in all smooth muscles, is a key convergence point of contractile and relaxatory pathways. Combinations of regulatory mechanisms, including isoform splicing, multiple phosphorylation sites, and scaffolding proteins, modulate MYPT1 activity with tissue and agonist specificities to affect contraction and relaxation. Other members of the PP1 family that do not target myosin, as well as PP2A and PP2B, dephosphorylate a range of proteins that affect smooth muscle contraction. This review discusses the role of phosphatases in smooth muscle contractility with a focus on MYPT1 in uterine smooth muscle. Myometrium shares characteristics of vascular and other visceral smooth muscles yet, during healthy pregnancy, undergoes hypertrophy, hyperplasia, quiescence, and labor as physiological processes. Myometrium presents an accessible model for the study of normal and pathological smooth muscle function, and a better understanding of myometrial physiology may allow the development of novel therapeutics for the many disorders of myometrial physiology from preterm labor to dysmenorrhea.
- myosin light chain 20
- myosin phosphatase regulatory targeting subunit 1
- protein phosphatase 2A
- protein phosphatase 2B
- protein phosphatase 1
- ion channel
regulation of protein phosphorylation by kinases and phosphatases is key to cell activity. Whole cell phosphoproteome studies show that serine and threonine residues constitute as much as 98% of the thousands of residues phosphorylated (78, 87, 140, 173, 199, 229). Compared with phosphatases, kinases dominate in the PubMed (http://www.ncbi.nlm.nih.gov/pubmed) literature on smooth muscle, with more than six times more publications. Recent studies, however, demonstrate that phosphatases play a significant role in fine-tuning smooth muscle contractility.
Smooth muscle contraction depends on the interaction of myosin with actin filaments to generate force (68, 86, 100, 177, 197, 222). Essential to this interaction is the phosphorylation of myosin regulatory light chain 20 (MLC20) at Ser19 by myosin light chain kinase (MLCK), which is activated by Ca2+ signaling (32, 92, 93; reviewed in Refs. 2, 5, 91, 100, 252). MLC20 Ser19 phosphorylation is the principal determinant of contractile amplitude and duration (5). MLC20 can be diphosphorylated at Ser19 and Thr18, and Thr18 phosphorylation may sustain contractile force by reducing the rate of MLC20 dephosphorylation (228). However, Thr18 phosphorylation alone does not affect contractile force (191, 228).
MLC20 phosphorylation is modulated by smooth muscle serine/threonine myosin phosphatases (MPs), which dephosphorylate MLC20 at Ser19 and Thr18, thereby causing relaxation (72, 93). MPs are active in their default state (62); however, they are carefully regulated and can be inhibited by multiple active processes. Importantly, most of the major kinase pathways involved in contractility regulate MP to control the rate of MLC20 dephosphorylation and modulate contractile force.
Other non-MP members of the protein phosphatase (PP) 1 family, as well as PP2A and PP2B, regulate MLC20 phosphorylation by modulating MP regulatory pathways and ion channels, which affect MP activity and Ca2+ signaling, respectively. They may also have a role in contraction by acting on other myosin-actin-interacting proteins.
This review discusses how MP regulates myometrial contractility and the role of serine/threonine phosphatases in regulating smooth muscle contraction, with a focus on the MP-containing MP regulatory targeting subunit 1 (MYPT1).
Myometrium is a relatively accessible human model that can reveal details of smooth muscle function, in particular species-specific processes. Myometrium shares characteristics, such as the core contractile molecules, with vascular and visceral smooth muscles. Unique to myometrium is its normal physiological hypertrophy and hyperplasia during pregnancy (reviewed in Ref. 214), transformations normally associated with pathologies, including arteriosclerosis and asthma, in other adult smooth muscles. Another unique aspect of myometrium is that it remains in a relatively noncontractile state for most of pregnancy to facilitate fetal development. This suggests the involvement of phosphatases in keeping the contractile machinery refractory to contractile stimuli. Activation of myometrium occurs at the onset of labor, when the uterus must produce forceful rhythmic phasic contractions for a successful birth. Mechanisms controlling phosphatase activity are likely to be involved in the switch to labor. Mistiming of labor or defective myometrial contractility causes life-threatening pathologies, including preterm birth and postpartum hemorrhage (143, 158). Dystocia (dysfunctional uterine contractions with no or only slow progression of the fetus down the reproductive tract) is also a major cause of cesarean section (49).
Serine/threonine phosphatase family members.
Three families of serine/threonine phosphatases have been identified: phosphoprotein phosphatases (PPPs), metal-dependent phosphatases, and aspartate-based phosphatases (210). Of these, only the PPP family has characterized roles in smooth muscle contractility.
Within the PPP family, seven closely related phosphatases are known: PP1, PP2A, PP2B, PP4, PP5, PP6, and PP7 (210). All PPP proteins are holoenzymes, which contain a catalytic subunit and at least one regulatory subunit (210). Only PP1, PP2A, and PP2B (Fig. 1) have defined roles in smooth muscle contractility, while PP4, PP5, PP6, and PP7 are poorly characterized and are reviewed elsewhere (9, 41, 60, 84, 88).
PP1 consists of a 35- to 38-kDa catalytic subunit, known as PP1c, which may exist as five isoforms: PP1cα1, splice variant PP1cα2, PP1cδ, PP1cγ1, and splice variant PP1cγ2 (40, 53, 204). The catalytic subunit can bind to >100 regulatory subunits, which modulate the activity of PP1c and localize it within the cell (31, 39, 210). All MP holoenzymes contain PP1 and a regulatory subunit (4, 30, 67, 217, 234).
PP2A can exist as a dimer or a trimer (38, 94, 124, 210, 283). The 36-kDa catalytic subunit PP2Ac has two isoforms, PP2Acα and PP2Acβ (43, 44). These isoforms form heterodimers (283), mainly with the 65-kDa regulatory A subunit PR65, and are expressed as two isoforms, PR65α and PR65β (80). Trimers are formed when PP2Ac-PR65 dimers bind a further regulatory B subunit (242). At least 21 different gene products and isoforms of the B subunit have been identified (94, 124).
PP2B, also known as calcineurin, is a Ca2+-dependent phosphatase (79, 113, 202, 210, 225). It consists of the 60-kDa catalytic subunit calcineurin A, which has at least six isoforms: Aα1, splice variant Aα2, Aβ1, splice variants Aβ2 and Aβ3, and Aγ (141, 152). These associate with a 19-kDa regulatory subunit called calcineurin B, expressed as at least three isoforms: B1, a splice variant of B1, and B2 (33, 244). Calcineurin B possesses four Ca2+-binding sites (3, 113). Binding of Ca2+ to calcineurin B results in minimal activation of PP2B but allows Ca2+-bound calmodulin to activate calcineurin by causing conformational changes that relieve autoinhibition of calcineurin A (113, 224, 225, 272).
Complexity in phosphatase signaling arises from the number of combinations of catalytic subunit with different regulatory subunits. The latter are predominantly responsible for regulating phosphatase activity, substrate specificity, and cellular localization (reviewed in Ref. 251). However, phosphatase regulation is not always exclusive to the regulatory subunits; for example, PP1α and PP2A catalytic subunits can be inhibited by phosphorylation at Thr320 and Thr304, respectively (35, 130).
Compartmentalization of PPP.
PPs can be localized by scaffolding proteins (195). A kinase-anchoring proteins (AKAPs) are the major family of proteins that localize phosphatases in skeletal and cardiac muscles (101, 139, 195, 216), and this may be the case in smooth muscle. AKAPs are critical in normal smooth muscle function, as AKAP150 (AKAP79) knockout mice are hypotensive, with disrupted regulation of the L-type Ca2+ channel (163).
Immunoprecipitation and localization experiments demonstrated that AKAP150 binds PP2A and PP2B in smooth muscle (12, 157, 163). Potentially many more AKAP species may also localize PP1, PP2A, and PP2B to smooth muscle targets such as ion channels. Future studies are needed to identify candidates and determine whether changes in expression of AKAP proteins occur in smooth muscle to affect contractility by altering phosphatase localization.
Role of PP1 in Smooth Muscle Contractility
The specific MP holoenzyme containing MYPT1 is the key PP1 regulating smooth muscle contraction. For clarity, we refer to MP holoenzyme as MP-MYPT1 and address subunits their specific names.
The five MP complexes, MYPT1 (4), MYPT2 (67), MYPT3 (217), MBS85 (234), and transforming growth factor β1-inhibited membrane-associated protein (TIMAP) (30), differ in their regulatory subunits. MYPT1 is present in all smooth muscles and is the most highly expressed subunit. MYPT2 and MYPT3 are expressed in skeletal muscle and heart, respectively (67, 217). MBS85 is expressed in phasic vascular smooth muscles and may contribute to the phasic phenotype (276). TIMAP is expressed in endothelial cells (30). The presence of all five MPs was reported in the human uterus (121, 122). However, only MYPT1 has a defined physiological role in regulating smooth muscle contraction.
MYPT1 gene and isoforms.
The human MYPT1 regulatory subunit is expressed from the gene PPP1R12A, located at chromosome 12q15-q21.2 (232). The full-length human transcript is 5,726 bp, expressed as a 1,030-amino acid protein (NM_002480.2) (Fig. 2). Human PPP1R12A contains 27 exons and is expressed as ≥6 splice variants.
In rats and mice, alternative splicing of exon 13 and/or 14 produces five MYPT1 isoforms, which differ in size by 50–100 amino acids at a “central insert” region of the protein (48). Similar alternative splicing of human PPP1R12A at exon 15, identical to rat exon 13, also occurs (NM_001244992.1). A recent in vitro study showed that expression of a different central insert in avian MYPT1 changed the rate of MYPT1 phosphorylation; however, total phosphorylation was unaffected, suggesting phosphorylation of the same site(s) at different speeds (274). Thus it is tempting to speculate that the mammalian MYPT1 central inserts also regulate MYPT1 phosphorylation.
MYPT1 isoforms also differ in their variable expression of a COOH-terminal leucine zipper (LZ) domain (36, 105). This domain regulates the sensitivity of MYPT1 isoforms to serine phosphorylation (see mypt1 serine phosphorylation) (226, 274). The rat and human LZ domains are identical, and isoforms lacking this domain (MYPT1LZ−) contain an additional 31-bp exon [exon 23 in mice (66)], which encodes a premature stop codon (105). In humans, mRNA sequences of MYPT1LZ− isoforms have not been published; however, PCR and Western blot studies indicate human expression of MYPT1LZ− and MYPT1LZ+ isoforms in myometrium (238). Our alignments of human MYPT1LZ+ (NM_002480.2) with mouse MYPT1LZ− (NM_027892) sequences indicate that this exon is located between human MYPT1 exons 25 and 26. Many permutations of the different central inserts and LZ domains are possible, but the isoforms have mainly been independently studied. A systematic approach that addresses the distribution and combinations of the central inserts and LZ domains is required in humans.
MYPT1 has two other subunits: the PP1 catalytic subunit PP1cδ and the M20 regulatory subunit with no known smooth muscle function (4). PP1cδ and M20 interact with MYPT1 at different regions. An RVxF motif at amino acids 35–38 binds PP1cδ (239). PP1cδ further interacts with eight ankyrin repeats spaced among amino acids 40–296 (98, 187), while M20 binds to the MYPT1 COOH terminus at amino acids 934–1006 (97).
MYPT1 contains many serine and threonine phosphorylation sites, including Ser432, Ser445, Ser472, Ser473, Ser601, Ser668, Ser692, Ser695, Thr696, Ser852, Thr853, and Ser910 (62, 240, 266, 270, 274, 275). Phosphorylation at Ser668, Ser692, Ser695, Thr696, Ser852, and Thr853 has known or proposed roles in smooth muscle contractility, while MYPT1 can be phosphorylated at Ser445, Ser472, and Ser910 to inhibit cell adhesion (275) and at Ser432, Ser473, and Ser601 during mitosis (240, 270).
Regulation of MP during contraction.
MP-MYPT1 can be inhibited during contractions, facilitating MLC20 phosphorylation and increasing contractile force (Fig. 3A). This occurs by MYPT1 phosphorylation at Thr696 and/or Thr853, increased MP-MYPT1 membrane localization, and phosphorylation of MP-MYPT1 inhibitory proteins such as 17-kDa PKC-potentiated inhibitor protein (CPI-17) and phosphatase holoenzyme inhibitor 1 (PHI-1). Agonist and/or Ca2+ signaling can activate these pathways. When Ca2+-independent pathways inhibit MP-MYPT1 by these processes and an active MLC20 kinase is present, the increased contractile force produced due to increased MLC20 phosphorylation is known as Ca2+ sensitization, as the increase in force occurs without an increase in intracellular Ca2+ levels (221). Combinations of the MP-MYPT1 inhibitory mechanisms vary between smooth muscles and can be differentially regulated by signaling molecules, allowing intricate control of MP-MYPT1 activity.
MYPT1 THREONINE PHOSPHORYLATION.
Phosphorylation of chicken, rat, or pig MYPT1 in vitro at Thr696 and/or Thr853 reproducibly inhibited MP-MYPT1 activity by ∼80% (62, 73, 106, 133, 153, 154, 266). Rho-associated protein kinase (ROCK) isoforms 1 and 2 (ROCK1/2), integrin-linked kinase (ILK), and zipper-interacting kinase (ZIPK) phosphorylate MYPT1 at Thr696 (62, 133, 154), while ROCK1/2 and ZIPK phosphorylate MYPT1 at Thr853 (25, 62, 76). In vitro, ZIPK phosphorylated MYPT1 6.7-fold more efficiently at Thr696 than at Thr853, while ROCK1/2 phosphorylated MYPT1 3-fold more efficiently at Thr853 than at Thr696 (76).
MYPT1 threonine phosphorylation is thought to inhibit MP-MYPT1 by causing a conformational change in which the phosphorylated residues in MYPT1 bind to the active site on PP1c (106). In vitro studies showed that chicken and pig MYPT1 phosphorylation is up to 37-fold more potent in inhibiting MP-MYPT1 activity at Thr696 than at Thr853 (106, 153).
MYPT1 phosphorylated at Thr696 and/or Thr853 has been found under resting conditions in vascular, airway, gastrointestinal, and myometrial muscle preparations or cells in species including rat, rabbit, and human (6, 17, 47, 83, 90, 95, 99, 110, 117, 144, 149, 153, 178, 179, 207, 241, 253, 260, 264). This MYPT1 phosphorylation occurs due to basal activity of procontractile kinases and can differ between smooth muscles (17, 178, 194). These differences allow muscle-specific control of MP-MYPT1 activity at rest to control basal muscle tone. Furthermore, by partially inhibiting MP-MYPT1 at rest, the basal activities of these kinase(s) affect the contractile force produced during contractions in which MP-MYPT1 is not further inhibited (6).
MYPT1 phosphorylation at Thr696 and/or Thr853 can be further increased during contractions. This occurs mainly through activation of G protein-coupled receptor (GPCR) kinase pathways by contractile agonists (221). Increased phosphorylation of MYPT1 at Thr853, but not Thr696, has been shown during contraction in many smooth muscle preparations and/or cells, including rat and rabbit vascular and gastrointestinal smooth muscles (47, 85, 96, 99, 110, 149, 166, 189, 253, 260, 264). However, some studies have found increased MYPT1 phosphorylation at Thr696 and Thr853 during contractions in rat, rabbit, and pig vascular and/or gastrointestinal smooth muscles, as well as human myometrium (90, 138, 153, 164, 193, 256). Additionally, the amount of MYPT1 threonine phosphorylation has been shown to be agonist-dependent (110, 264). For example, in endothelium-denuded rabbit vas deferens, 1 μM endothelin-1 (ET-1) increased MYPT1 Thr853 phosphorylation 3.5-fold, while 154 mM KCl or 30 μM phenylepherine (PE) increased MYPT1 Thr853 phosphorylation 2-fold (110). Taken together, these data are consistent with ROCK being the major, but not always the only, kinase that phosphorylates MYPT1 during smooth muscle contraction. Differences between MYPT1 phosphorylation at Thr696 and Thr853 can represent methodological differences or species-, tissue-, and/or agonist-specific processes during contractions.
In a species- and tissue-specific manner, RhoA, the primary smooth muscle activator of ROCK1/2 (reviewed in Ref. 190), can be activated by membrane depolarization and/or Ca2+ signaling (129, 203, 208, 245, 254), causing increased MYPT1 phosphorylation at Thr696 and/or Thr853 (203, 208, 254, 273). This may contribute as much as 50% of the total force generated by depolarization-induced contractions (203). This can be species-specific, as the ROCK inhibitor Y-27632 dose dependently inhibited phasic contraction of electric field-stimulated rat, but not guinea pig, ureters (209). In rat aorta, Ca2+ activated RhoA by first activating class 2 phosphoinositide 3-kinase (PI3K) (254). In cultured rat aorta smooth muscle cells, knockdown of class 2 PI3K using small interfering RNA (siRNA) decreased RhoA activity, MYPT1 phosphorylation at Thr853, and contraction induced by the Ca2+ ionophore ionomycin (273). PI3K class 2 was expressed in multiple rat arteries and veins, suggesting that this mechanism of RhoA-ROCK activation may be common in vascular smooth muscles (208).
In spontaneous and oxytocin-induced contractions of full-term nonlaboring human myometrium, phosphorylation of MYPT1 at Thr696 and Thr853 was increased approximately twofold at peak contraction compared with the relaxed state between contractions (90). This MYPT1 phosphorylation positively correlated with increased MLC20 monophosphorylation at Ser19 and diphosphorylation at Ser19 and Thr19, which is consistent with studies in other muscles (47, 110, 138, 149, 189, 193, 264). Phosphorylation of myometrial MYPT1 at Thr696 and Thr853 in spontaneous and oxytocin-induced contractions suggests that this may be a physiological process that is critical to generating strong myometrial contractions by pushing the MLCK vs. MP-MYPT1 balance toward MLCK activity. MLC20 Thr18 and Ser19 diphosphorylation may also reduce the rate of MLC20 dephosphorylation, further altering the balance to maximize MLC20 phosphorylation and increase contractile amplitude.
Employing the ROCK inhibitor G-1153, Hudson et al. (90) demonstrated that, during contraction of full-term nonlaboring human myometrial preparations, MYPT1 phosphorylation was ROCK1- and ROCK2-dependent at Thr853 and partially ROCK1- and ROCK2-dependent at Thr696. In support of this finding, in primary cultured human pregnant myometrial cells, ROCK1 and ROCK2 knockdown by siRNA significantly reduced basal MYPT1 Thr853 phosphorylation but only caused a nonsignificant trend of decreased Thr696 phosphorylation (90). In rats, ROCK1/2 mRNA was increased at full term or during labor (167, 230, 231); in mice, ROCK1/2 protein was increased during gestation (201). In humans, increased myometrial ROCK1/2 during pregnancy was reported (146); however, subsequent studies showed that myometrial ROCK1/2 mRNA and protein remained unchanged during human pregnancy and labor (65, 120, 123, 200). Human myometrial ROCK1/2 activity during pregnancy has not been measured. These findings suggest that regulation of the myometrial Rho-ROCK pathway is species-specific. The additional kinases responsible for MYPT1 Thr696 phosphorylation in human myometrium were not identified by Hudson et al. but may include ILK and/or ZIPK; however, their expression has yet to be determined.
Increased RhoA activity was not found in comparisons of full-term laboring vs. nonlaboring human myometrial samples that had been immediately snap-frozen following collection (123). Phasic processes associated with contractions can be masked by this collection process, as samples may be taken at labor but while the uterus is relaxed between contractions (184). Taken together, these data suggest three possibilities regarding regulation of ROCK and the resultant MYPT1 phosphorylation: 1) RhoA-independent mechanisms cause activation of ROCK during contraction, 2) RhoA could be phasically activated during contractions, and this could involve Ca2+ signaling, and 3) ROCK in myometrium is constitutively active and is not affected by contraction, but MYPT1 phosphatases that dephosphorylate MYPT1 at Thr696 and Thr853 are inhibited during contractions (see below). To elucidate these possibilities, future studies should directly investigate the activity of ROCK and MYPT1 phosphatase(s) during contractions.
MYPT1 MEMBRANE LOCALIZATION.
Changes in subcellular MP-MYPT1 localization have been demonstrated in hamster vascular resistance arteries, rat cerebral arteries, and ferret portal vein (23, 164, 212). In these muscles, immunofluorescence studies demonstrated that MYPT1 was dispersed throughout the cell at rest but cell membrane localization of MYPT1 increased during contraction (23, 155, 164, 212). Membrane localization spatially separates MP-MYPT1 from MLC20 to reduce MLC20 dephosphorylation (164). In hamster resistance arteries and ferret portal vein, MYPT1 membrane translocation was inhibited by Y-27632 or transfection of cells with N19RhoA, a dominant-negative RhoA mutant, indicating that it was ROCK-dependent (23, 212). The extent of MYPT1 membrane localization was agonist-specific (212). This is probably due to differences in agonist-induced ROCK activation. Shin et al. (212) proposed that MYPT1 membrane localization requires MYPT1 phosphorylation. Subsequent studies in smooth muscle have shown data supporting a role of MYPT1 Thr853 phosphorylation in membrane localization. 1) In rat cerebral arteries treated with the contractile agonist thromboxane A2 mimetic U-46619, immunofluorescence experiments showed increased membrane localization of MYPT1 phosphorylated at Thr853, while localization of MYPT1 phosphorylated at Thr696 was unchanged (164). 2) MYPT1 Thr853 phosphorylation was shown to dissociate MP-MYPT1 from myosin in vitro (247). However, while MYPT1 Thr853 phosphorylation may increase membrane localization, it should be noted that phosphorylation is not essential for MP-MYPT1 to interact with phospholipids in vitro (160). MP-MYPT1 might also be translocated by other proteins that preferentially interact with MYPT1 phosphorylated at Thr853. MYPT1 has been shown to be associated with the membrane in nonlaboring and laboring human myometrium (238). It is probable that increased MYPT1 Thr853 phosphorylation causes increased MP-MYPT1 membrane localization during contractions.
MYPT1 INHIBITOR PROTEINS.
In smooth muscle, MP-MYPT1 is inhibited by the endogenous 17-kDa phosphoprotein CPI-17 (58, 271) and the functionally similar, but less efficacious, 23-kDa PHI-1 (56). CPI-17 and PHI-1 proteins are active in their phosphorylated forms and inhibit MP-MYPT1 independently of MYPT1 threonine phosphorylation (56, 77). CPI-17 may be phosphorylated at Thr38 by PKC (58), ROCK (115, 176), ILK (45), and protein kinase N1 (PKN1) (77), while PHI-1 may be phosphorylated at Thr57 by PKC (56), ROCK, and ILK (45). In vitro phosphorylation of CPI-17 and PHI-1 increased their affinities for MP-MYPT1 by ∼1,000- and 50-fold, respectively (56, 77). Binding of CPI-17 to the MP-MYPT1 PP1 catalytic site has been suggested to require conformational changes and interaction of the phosphorylated residues with the PP1 catalytic site (55). CPI-17 is differentially expressed in many smooth muscle tissues in the rabbit, being 10-fold higher in tonic than phasic smooth muscles (263), suggesting that CPI-17 is the preferred mechanism for MP-MYPT1 inhibition in tonic smooth muscle. Furthermore, in endothelium-denuded femoral artery, PE caused CPI-17 phosphorylation before MYPT1 threonine phosphorylation, indicating that CPI-17 can be more rapidly activated to increase contractile force (47).
Only CPI-17 has been investigated in human myometrium. Ozaki et al. (175) reported increased mRNA and protein of CPI-17 and PKCβ in human full-term nonlaboring myometrium compared with nonpregnant myometrium. Additionally, the PKC activator phorbol 12,13-dibutyrate caused contraction without raising intracellular Ca2+ levels in nonlaboring myometrium (175). However, CPI-17 phosphorylation was not directly measured, and laboring samples were not investigated. Therefore, although CPI-17 activation is consistent with these findings, other PKC mechanisms promoting contraction could not be ruled out. Lartey et al. (123) investigated CPI-17 phosphorylation, as well as PKN1 activity, in nonpregnant, nonlaboring, and laboring human myometrial samples. They found increased PKN1 expression and CPI-17 phosphorylation in full-term nonlaboring myometrium compared with nonpregnant samples but no significant difference between full-term nonlaboring and laboring samples. It is tempting to speculate that this CPI-17 phosphorylation reflected basal levels controlling resting MP-MYPT1 activity and that CPI-17 phosphorylation will be phasically increased during myometrial contractions to further inhibit MP-MYPT1 and increase contractile force.
MP regulation during relaxation.
MP-MYPT1 activity can be increased to cause smooth muscle relaxation (59, 108, 126, 159, 267, 268). The cAMP and cGMP pathways are major regulators of MP-MYPT1 activity during relaxation (Fig. 3B) (59, 108, 126, 159, 267, 268). These pathways relax smooth muscles by two mechanisms: 1) they decrease intracellular Ca2+ levels, and 2) they cause Ca2+ desensitization (reviewed in Ref. 148). Ca2+ desensitization involves the relaxation of smooth muscle independently of Ca2+ withdrawal, so that muscles can relax even when intracellular Ca2+ levels are high (221). This can occur through increase in MP-MYPT1 activity (59, 108, 126, 267, 268) or regulation of the actin cytoskeleton, which occurs primarily through phosphorylation of heat shock protein (HSP) 20 (HSP20) (14, 196, 243, 262). Increased MP-MYPT1 activity is promoted during relaxation by MYPT1 serine phosphorylation, MYPT1 threonine dephosphorylation, telokin activation, and CPI-17 inactivation; these processes can vary between smooth muscle tissues.
The human myometrium expresses components of pathways that cause Ca2+ desensitization. These include cyclic nucleotide-dependent MP-MYPT1 regulatory proteins such as telokin (145, 219) and actin regulatory proteins such as HSP20 (243). However, Ca2+ desensitization has not been studied in myometrium. It is probable that myometrial MP-MYPT1 will be regulated by cyclic nucleotide signaling to facilitate Ca2+ desensitization.
MYPT1 SERINE PHOSPHORYLATION.
PKG and PKA can phosphorylate all smooth muscle MYPT1 isoforms in vitro and in vivo at Ser668, Ser692, Ser695, Ser852, Thr696, and Thr853 (73, 159, 223, 266, 274). The rate of PKG-mediated serine phosphorylation of chicken MYPT1LZ+ isoforms is 75-fold greater than that of MYPT1LZ− isoforms in vitro (274), and expression of the MYPT1LZ+ isoform has been positively correlated with sensitivity to cGMP-induced relaxation and MLC20 dephosphorylation in rat aorta, portal vein, and uterine arteries, mouse duodenum, and chicken aorta and gizzard (34, 66, 89, 103, 105, 132, 185, 274). These studies showed that expression of the MYPT1LZ+ isoform affects levels of MYPT1 serine phosphorylation by PKG in vivo and suggest that MYPT1 serine phosphorylation is associated with relaxation. However, it has been consistently found that in vitro MYPT1 phosphorylation by PKA and/or PKG does not change MP-MYPT1 activity relative to MP-MYPT1 containing nonphosphorylated MYPT1 (73, 154, 159, 160, 266).
It was first proposed that MYPT1 phosphorylation at Ser695 or Thr696 and at Ser852 or Thr853 causes steric inhibition of the adjacent site (266). This was based on in vitro experiments where MYPT1 phosphorylation by PKA or ROCK prevented further MYPT1 phosphorylation upon subsequent incubation with the alternative kinase (266). However, using Western blotting, Nakamura et al. (159) subsequently demonstrated that MYPT1 can be diphosphorylated at Ser695 and Thr696. They showed that in vitro MYPT1 phosphorylation by PKG prevented Thr696 phosphorylation by ROCK, but MYPT1 phosphorylation by ROCK did not significantly change subsequent phosphorylation by PKG when measured via levels of 32P incorporation. These data suggest that MYPT1 Ser695 phosphorylation was still accessible to PKA following ROCK phosphorylation of Thr696. Grassie et al. (73) recently reconciled this conflict, showing in vitro that MYPT1 phosphorylated by ROCK at Thr696 was not subsequently phosphorylated at Ser695 by PKA. However, PKA phosphorylated both sites, with Ser695 phosphorylation preceding and occurring at twice the rate of Thr696 phosphorylation. This was shown by TLC and Western blotting using phospho-specific MYPT1 antibodies that were validated against phosphorylated MYPT1 mutants containing combinations of alanine mutations at these serine and threonine sites (73). Consistent with this finding, two previous studies detected MYPT1 Thr696 phosphorylation in Western blots following in vitro MYPT1 phosphorylation by PKA (111, 266). Grassie et al. also confirmed that the MYPT1 Ser852 and Thr853 sites were inhibited by phosphorylation of the adjacent site identically to the MYPT1 Ser695 and Thr696 sites. The rate of Ser852 phosphorylation by PKA was not determined because of lack of a phospho-specific antibody (73).
It should be noted that there is no known procontractile kinase that phosphorylates MYPT1 at serine residues or at threonine residues following phosphorylation of the adjacent serine residue, suggesting that these phosphorylation events are exclusive to PKA and PKG. Thus, procontractile kinases monophosphorylate MYPT1 at Thr696 and/or Thr853, while PKA and PKG monophosphorylate MYPT1 at Ser695 or Ser852 or sequentially diphosphorylate MYPT1 at Ser695 and Thr696 and/or Ser852 and Thr853. The role of MYPT1 diphosphorylation by PKA and PKG is not known. However, it may function as a negative-feedback mechanism to prevent excessive MYPT1 threonine dephosphorylation during relaxation, thereby allowing the muscle to maintain a basal tone while becoming relaxed by cyclic nucleotides.
Evidence for bidirectional inhibition of adjacent-site phosphorylation of MYPT1 at Ser695 and Thr696 and/or Ser852 and Thr853 in muscle preparations has been found. In permeabilized rabbit ileum, 8-bromo-cGMP (8-BrcGMP) increased MYPT1 Ser695 phosphorylation, and this was associated with decreased MYPT1 Thr696 phosphorylation upon subsequent addition of MYPT kinase compared with similarly 8-BrcGMP-treated samples (266). In rat caudal artery, forskolin increased MYPT1 Ser695 phosphorylation and prevented contraction and MYPT1 Thr696 phosphorylation upon subsequent treatment with U-46619 (73). U-46619 pretreatment caused contraction, increased MYPT1 Thr696 phosphorylation, and inhibited Ser695 phosphorylation upon subsequent forskolin treatment; however, forskolin still caused relaxation. In endothelium-intact rabbit cerebral arteries, U-46619 caused contraction, activated RhoA, and, specific to cerebral arteries, activated the nitric oxide (NO) pathway (164). Western blotting showed that U-46619 increased MYPT1 phosphorylation 1.6-fold at Thr696, 2.5-fold at Thr853, and 2.5-fold Ser695. Pretreatment of strips with the NO synthase inhibitor NG-nitro-l-arginine methyl ester followed by U-46619 prevented MYPT1 phosphorylation at Ser695, increased MYPT1 phosphorylation 2.5-fold at Thr696 and 8-fold Thr853, and increased contractile force. Since Grassie et al. (73) showed that these MYPT1 Thr696 and Thr853 phospho-specific antibodies also recognized MYPT1 diphosphorylated at Ser695 and Thr696 or Ser852 and Thr853, respectively, it is likely that the signal detected in samples treated with U-46619 alone by Neppl et al. (164) included MYPT1 diphosphorylated at these adjacent sites by PKG. As adjacent serine-threonine site phosphorylation reverses MP-MYPT1 inhibition yet U-46619 caused contraction in cerebral arteries, it is probable that a pool of MYPT1 was phosphorylated only at Thr696 and/or Thr853 by procontractile kinases.
In endothelium-denuded PE-contracted rabbit femoral arteries containing MYPT1 phosphorylated at Thr696 and/or Thr853, the NO donor sodium nitroprusside (SNP) increased MYPT1 Ser695 phosphorylation, suggesting that a different pool of MYPT1 was phosphorylated at Ser695 (111). In these time-course experiments, MLC20 phosphorylation was decreased to minimum levels by 30 s, with ∼70% of this decrease occurring in the first 15 s. MYPT1 Ser695 phosphorylation increased 21% at 15 s and 56% at 30 s relative to its maximum level, suggesting that Ser695 phosphorylation is not necessary to initiate relaxation. Pretreatment of these arteries with SNP before PE stimulation significantly decreased, but did not prevent, force production. MYPT1 Ser695 phosphorylation was elevated to maximum levels by SNP pretreatment. Upon subsequent PE treatment, Thr696 and Thr853 phosphorylation was decreased by ∼20% relative to control at 10 s and 5 min. However, while MYPT1 Thr853 phosphorylation was significantly reduced compared with control, it was still nearly twofold increased at 5 min after PE treatment relative to preparations before PE treatment. This study may also have been confounded by detection of MYPT1 diphosphorylated at the adjacent serine-threonine sites with their antibodies; however, it suggests that, following treatment with contractile agonists or antagonists, a pool of MYPT1 remains unphosphorylated in this tissue and can be phosphorylated by subsequent stimuli. In permeabilized rabbit ileum, 8-BrcGMP was equally efficient in reversing Ca2+ sensitization caused by addition of truncated MYPT1 mutants containing Ser695 and Thr696 or Ala695 and Thr696 sites (106), further supporting the view that MYPT1 Ser695 phosphorylation is not required for relaxation. These studies suggest that MYPT1 phosphorylation at Ser695 and/or Ser852 primarily functions as a longer-term adaptation to maintain relaxation. The ratio of MYPT1 phosphorylated at these serine sites to nonphosphorylated MYPT1 may dictate the levels of MYPT1 monophosphorylation at Thr696 and/or Thr853 and contractile force caused by procontractile stimuli in a tissue-specific manner, and this ratio is probably affected by expression of MPYTLZ+ isoforms.
Two PKA and/or PKG sites at Ser692 and Ser668 have been detected in vitro by Edman sequencing and mass spectroscopy, respectively (266, 274). Yuen et al. (274) singly mutated a MYPT1LZ+ isoform to produce four mutants containing Ala668, Asp668, Ala695, or Asp695. The rate of phosphorylation by PKG was ∼6.5-fold higher in Asp695 mutants and 2-fold higher in Ala695 mutants than in either Ser668 mutant. The authors suggested that Ser668 becomes phosphorylated at a faster rate than Ser695. Using two-dimensional TLC, Grassie et al. (73) demonstrated that PKA phosphorylated an unidentified serine residue on MYPT1 and suggested that it might be Ser692. However, the findings by Yuan et al. suggest that this may represent Ser668 phosphorylation. Since these sites are not adjacent to procontractile kinase phosphorylation sites, it is likely that they will still be accessible in threonine-phosphorylated MYPT1; thus they could be involved in restoring threonine-phosphorylated MP-MYPT1 activity to cause relaxation. Furthermore, since Ser668 phosphorylation appears faster than Ser695 phosphorylation (274), it may regulate MP-MYPT1 activity to cause rapid MLC20 dephosphorylation during relaxation.
Serine phosphorylation may inhibit membrane localization of MYPT1 to increase colocalization of MP-MYPT1 and myosin. An in vitro study showed that ∼70% of nonphosphorylated chicken gizzard MYPT1 associated with phospholipid vesicles and that this decreased to 13% following phosphorylation by PKG (160). Using two truncated mutants of MYPT1 containing amino acids 1–674 or 667–1004, they showed that the COOH-terminal MYPT1 mutant was primarily responsible for the MYPT1-phospholipid interaction and contained the PKG phosphorylation site involved; however, the precise site is unknown. Phosphorylation of this site in vivo may translocate membrane-localized MP-MYPT1 to myosin to cause relaxation. If it occurs, this mechanism of increasing MP-MYPT1 activity may not have been detected in the in vitro studies, which measured MP-MYPT1 activity following phosphorylation by PKA (73, 154, 159, 160, 266). Studies will need to confirm this dissociation in vivo and identify the exact sites, as this will determine if all membrane-bound MYPT1 or pools of MYPT1 differentially phosphorylated at the adjacent serine and threonine sites can be regulated by this mechanism.
cAMP, by activating PKA, is a key relaxant of pregnant human myometrium. Agents that increase intracellular cAMP levels, including forskolin, cAMP-phosphodiesterase (PDE) type 4 inhibitors, and 8-bromo-cAMP, dose dependently inhibit human full-term nonlaboring myometrial contractile force by up to 100% (13, 64, 128, 142, 172, 218, 243, 248). It is probable that myometrial cAMP signaling causes MYPT1 serine phosphorylation. Importantly, this could maintain quiescence by preventing the inhibitory MYPT1 threonine phosphorylation that is associated with myometrial contraction.
The pregnant human myometrium is relatively insensitive to relaxation by cGMP analogs. Two studies using 8-BrcGMP showed dose-dependent relaxation of human myometrium at full term, with 40% and 60% maximum inhibition of contraction (54, 168). This suggests that myometrial MYPT1 may be regulated by cGMP, but to a lesser extent than cAMP. The NO-soluble guanylyl cyclase-cGMP-PKG1 pathway, which causes MYPT1 serine phosphorylation in nonmyometrial smooth muscles (164), is unlikely to cause significant MYPT1 serine phosphorylation in myometrium. Myometrial PKG1 levels are decreased during human pregnancy (42), and inhibitors of soluble guanylyl cyclase had no effect on full-term pregnant myometrial relaxation caused by NO donors or the PDE5 inhibitor sildenafil citrate, suggesting that relaxation caused by these agents is independent of cGMP (26, 104). Instead, myometrium may employ a unique mechanism of cGMP regulation of MYPT1 that does not involve PKG1. Myometrium is the only smooth muscle that expresses PKGII, and it coimmunoprecipitated with MYPT1 in caveolae (238). Tichenor (238) proposed that activation of particulate guanylyl cyclase in caveolae would activate PKGII, leading to relaxation through MYPT1 phosphorylation at the cell membrane. This is consistent with data showing compartmentalization of cGMP signaling in pregnant guinea pig myometrium, where activation of particulate, but not soluble, guanylyl cyclase increased cGMP levels in caveolae and caused relaxation (27). However, this MYPT1 is dissociated from myosin and is possibly threonine-phosphorylated. Nevertheless, this pool of MYPT1 is a prime candidate to translocate from the membrane to myosin to increase MLC20 dephosphorylation, and thus myometrium is an ideal tissue to examine the effects of MYPT1 serine phosphorylation on membrane localization in vivo.
MYPT1 THREONINE DEPHOSPHORYLATION.
Experimental evidence indicates the existence of smooth muscle MYPT1 phosphatases, which dephosphorylate MYPT1 (159). Purified PP1, PP2A, PP2B, and the metal-dependent phosphatase PP2C have been shown to dephosphorylate MYPT1 in vitro (233); however, the identities of physiological smooth muscle MYPT1 phosphatase(s) remain elusive. MYPT1 threonine dephosphorylation is independent of MYPT1 serine phosphorylation (159). It was shown in vitro that MYPT1 can be autodephosphorylated by MP-MYPT1; however, this was prevented by diphosphorylation of the adjacent serine-threonine sites (73).
In homogenized extracts of rabbit femoral artery, 8-BrcGMP increased the rate of dephosphorylation of Thr696 in MYPT1 phosphorylated at Thr696 and MYPT1 diphosphorylated at Ser695 and Thr696, while MYPT1 Ser695 dephosphorylation was significantly slower and was unaffected by 8-BrcGMP (159). This suggests that unknown phosphatase(s), and not autodephosphorylation, is the major regulator(s) of MYPT1 threonine dephosphorylation. In rabbit femoral artery and rat ileum muscle preparations, ∼30% reduction of MYPT1 Thr696 phosphorylation and 40% reduction of MYPT1 Thr853 phosphorylation were observed following 8-BrcGMP or SNP treatment (106, 111, 159). In contrast, in intact rat cerebral artery precontracted with U46619, Neppl et al. (164) showed that MYPT1 Thr696 and Thr853 phosphorylation was unchanged following 8-BrcGMP-induced relaxation. Time courses in endothelium-denuded rabbit femoral artery contracted with PE indicated that Thr696 and Thr853 dephosphorylation followed relaxation, suggesting that it may be a longer-term adaptation to increase MYPT1 activity and maintain relaxation (111). These studies could possibly be confounded by recognition of MYPT1 diphosphorylated at adjacent serine-threonine sites by PKG with MYPT1 Thr696 and/or Thr853 phospho-specific antibodies. Therefore, these antibodies may mask larger decreases in levels of MYPT1 phosphorylated at Thr696 and/or Thr853 only. These studies thus indicate that cGMP-induced decreases in MYPT1 threonine phosphorylation may differ between smooth muscle tissue types. These differences may be reconciled by differential expression and activity of MYPT1 kinases or MYPT1 threonine phosphatases. Therefore, the identities, expression, and regulation of smooth muscle MYPT1 threonine phosphatases need to be elucidated.
In full-term nonlaboring human myometrial preparations, MYPT1 threonine sites were dephosphorylated to basal levels between contractions (90). This indicates the presence of an active MYPT1 threonine phosphatase. A cyclic nucleotide-sensitive MYPT1 phosphatase expressed in myometrium would increase MP-MYPT1 activity to maintain relaxation during gestational uterine quiescence.
Telokin can regulate MP-MYPT1 activity without altering MYPT1 Thr696 and Thr853 phosphorylation (107). Telokin is a 17-kDa isoform of the COOH terminus of MLCK, driven by its own smooth muscle-specific promoter within the MLCK gene (69, 82). Telokin is highly expressed in visceral and phasic vascular smooth muscle in rabbits, rats, and mice (37, 81, 82, 267). Unlike other MLCK isoforms, telokin does not possess kinase activity or the ability to bind calmodulin (220). Telokin is activated by phosphorylation, and PKG and PKA phosphorylate telokin at Ser13 (134, 267). Phosphorylated telokin regulates myosin and MP-MYPT1 activity (108, 213, 215). cGMP-induced relaxation in ileum smooth muscle is reduced by ∼50% in telokin knockout mice (108).
Khromov et al. (107) demonstrated that telokin reactivated MP-MYPT1 that had been threonine-phosphorylated in mouse ileum smooth muscle. How telokin reactivates MP-MYPT1 has yet to be elucidated, but Khromov et al. proposed two mechanisms: 1) telokin may bind to MYPT1 to directly reactivate it, and/or 2) telokin may bind to myosin and scaffold interactions between MP-MYPT1 and phosphorylated myosin.
Studies have shown telokin expression in nonpregnant rabbit, rat, mouse, and human uterus (69, 82, 145, 219). In human myometrium, telokin levels were increased in full-term nonlaboring compared with nonpregnant samples (145, 219). However, telokin levels in full-term laboring myometrium were not reported. As telokin is regulated by PKA and telokin levels are increased with pregnancy, it is an ideal candidate to cause Ca2+ desensitization during myometrial quiescence. In myometrium, it is likely that telokin will be phosphorylated by PKA to increase MP-MYPT1 localization to myosin during uterine quiescence and that this will be reversed at labor to facilitate contraction.
MYPT1 may become activated by dephosphorylation of CPI-17 during NO-induced relaxation. In histamine-contracted endothelium-denuded rat vascular carotid and PE-contracted endothelium-denuded rabbit femoral arteries, SNP caused CPI-17 dephosphorylation, which paralleled or preceded MLC20 dephosphorylation, and MLC20 dephosphorylation preceded the acute phase of relaxation (59, 111). CPI-17 dephosphorylation was transient, as CPI-17 phosphorylation recovered during the sustained phase of relaxation (59, 111). The phosphatase(s) responsible for this CPI-17 dephosphorylation is unknown but may include non-MP PP1 holoenzymes or PP2A (see Smooth muscle PP1 targets independent of MYPT1 and Role of PP2A in Smooth Muscle Contractility). Studies of myometrial CPI-17 are needed to address whether CPI-17 dephosphorylation occurs during phasic contraction-relaxation cycles and myometrial quiescence.
Scaffolding of MYPT1.
Scaffolding proteins, including P116 myosin interacting protein (M-RIP), prostate apoptosis response-4 (Par-4), and probably HSP27, regulate MP-MYPT1 activity. M-RIP, first identified as a MYPT1-binding protein in yeast two-hybrid screening experiments, is a 116-kDa protein that is expressed in brain, heart, lung, kidney, and smooth muscle (70, 227). In mice, M-RIP levels differ among gastrointestinal smooth muscles (17). M-RIP can bind myosin, MYPT1, RhoA, and α-smooth muscle actin simultaneously at different sites (114, 150, 151, 227). M-RIP localizes the MYPT1LZ+, but not the MYPT1LZ−, isoform to myosin, which increases MLC20 dephosphorylation without increasing MP-MYPT1 activity (114, 150, 227). By binding MYPT1LZ+ isoforms, M-RIP may modify smooth muscle sensitivity to serine-phosphorylated MYPT1.
Knockdown of M-RIP in rat aorta A7r5 cells using siRNA did not affect basal Thr853 phosphorylation but prevented increased MYPT1 Thr853 phosphorylation following lysophosphatidic acid (LPA) stimulation (198). This suggests that M-RIP expression increases the rate of MYPT1 phosphorylation by ROCK, even though M-RIP does not bind ROCK, and it is not required for MYPT1 phosphorylation (114, 255). Furthermore, M-RIP may cause preferential phosphorylation of the MYPT1LZ+ isoform over the MYPT1LZ− isoform by ROCK.
M-RIP may also mediate links between the contractile machinery and contractile stimulation by the immune system (246). In human aortic smooth muscle cells and endothelial cells, the immune factor TNFα caused a greater than sixfold upregulation of the NUAK family SNF1-like kinase 2 (NUAK2) (269), which interacted with MYPT1 and M-RIP in HeLa cells (246). This interaction stabilized actin fibers and inhibited MP-MYPT1 activity through an unknown mechanism (246). Inflammation is associated with the initiation of labor (71), and it is possible that NUAK2 and M-RIP interactions may represent an important regulatory pathway not yet described in myometrium.
Par-4 is a 38-kDa protein found in many rat tissues, including vascular, airway, and gastrointestinal smooth muscles (22). Par-4 interacts with MYPT1 fragments containing the LZ domain in vitro. In rat A7r5 cells, Par-4 overexpression increased total PP1 activity and decreased MLC20 phosphorylation, consistent with Par-4 increasing MP-MYPT1 activity (249). Phosphorylation of Par-4 caused its dissociation from MYPT1, and Par-4 phosphorylation increased following LPA treatment. This dissociation was disrupted in cells expressing a Par-4 mutant in which Thr155, a ZIPK and PKA phosphorylation site, was mutated to alanine. siRNA knockdown of Par-4 increased resting MYPT1 Thr696 and MLC20 Ser19 phosphorylation; however, LPA-increased MYPT1 Thr696 and MLC20 Ser19 phosphorylation was attenuated (249).
In rat A7r5 cells, PGF2α stimulation temporarily increased localization of Par-4 with ZIPK, and this was antagonized by a Par-4 decoy peptide, and ZIPK coimmunoprecipitated with Par-4 (250). In ferret portal vein muscle preparations, pretreatment with Par-4 decoy peptides or partial knockdown of Par-4 levels by 20% with morpholino oligonucleotides decreased PGF2α-induced force by ∼20% (250). In the Par-4-knockdown experiments, MYPT1 Thr853 phosphorylation was significantly reduced by 10% following PGF2α treatment. Vetterkind et al. (249) proposed a “molecular padlock” mechanism, whereby Par-4 bound to MYPT1 blocks inhibitory MYPT1 phosphorylation and increases MP-MYPT1 activity, while Par-4 simultaneously localizes ZIPK, which can phosphorylate Par-4 to cause dissociation from MYPT1. This process frees MYPT1 to allow subsequent phosphorylation at Thr696 and Thr853 by ZIPK. In this manner, Par-4 increases MP-MYPT1 activity or inhibition, depending on the expression and activity of kinases that interact with and/or are blocked by Par-4.
Thus, M-RIP and Par-4 regulate MLC20 dephosphorylation through MP-MYPT1 localization and/or MYPT1 phosphorylation. These proteins place MP-MYPT1 into a “standby mode,” in which MP-MYPT1 activity is increased to cause relaxation while scaffolding of kinases makes MYPT1 more responsive to phosphorylation, facilitating rapid MP-MYPT1 inhibition and contraction.
HSP27 may also regulate MP-MYPT1. HSP27 can be phosphorylated at Ser15, Ser78, and/or Ser82 by procontractile kinases, including MAPK, PKC, and PKD (51, 119, 127). Phosphorylated HSP27 promotes smooth muscle contractions (20). HSP27 regulates contractions by affecting actin filament stability and by mediating interactions among PKC, ROCK, and RhoA and their translocation to the membrane (15, 19, 118, 181, 182, 186).
In rabbit colon smooth muscle cells, Patil and Bitar (180) showed that HSP27 immunoprecipitated with Thr696-phosphorylated MYPT1 in membrane fractions that were isolated by ultracentrifugation. It is probable that this pool of MYPT1 was also phosphorylated at Thr853; however, neither this nor HSP27 phosphorylation status was assessed. The authors suggested that HSP27 might scaffold a MYPT1-ROCK-HSP27 complex to facilitate MYPT1 phosphorylation. However, the findings are also consistent with HSP27 translocating MYPT1 to the membrane or the MYPT1-HSP27 interaction being exclusive to the membrane. Studies to determine whether HSP27 enhances MYPT1 phosphorylation by scaffolding procontractile kinases or affects its translocation will reveal the contribution of HSP27 in regulating MP-MYPT1 during contractions.
Future studies will need to determine whether M-RIP and/or Par-4 is expressed in myometrium, as these could change MP-MYPT1 activity during myometrial quiescence or activation. Whether HSP27 regulates myometrial MYPT1 and the mechanisms involved could have important implications in the regulation of myometrial MP-MYPT1 activity. MacIntyre et al. (135) showed that pregnant human myometrial expression of αB-crystallin, which regulates HSP27, was downregulated at labor. Immunofluorescence demonstrated that, in full-term nonlaboring myometrium, αB-crystallin reduced the association of HSP27 with actin filaments (135). It is possible that, by regulating HSP27, αB-crystallin could reduce MP-MYPT1 translocation and/or prevent the contraction-associated increase of MYPT1 threonine phosphorylation, thereby maintaining activated MP-MYPT1 close to its substrate.
Regulation of MYPT1 expression during pregnancy.
Myometrial MYPT1 protein levels are twofold higher in mice at day 13 of pregnancy than in nonpregnant animals (21, 131). This contrasts with invariant human myometrial levels of MYPT1 between nonpregnant and full-term nonlaboring myometrium (120). This probably represents species differences; however, increases in MYPT1 levels earlier in human pregnancy might be masked by a return to nonpregnant levels at full term. This possibility could be explored by examining total MYPT1 levels in preterm nonlaboring myometrium.
In pregnant mice, smoothelin-like protein 1 (SMNTL1), may regulate MYPT1 and inhibit progesterone receptor-B (PR-B) signaling (Fig. 4) (21, 131). Knockout of mouse SMNTL1 increased myometrial MYPT1 by sixfold at day 13 of pregnancy (131), and mRNA for the contractile-associated proteins cyclooxogenase-2 (Cox-2), oxytocin receptor, and connexin-43 was decreased during pregnancy (21), suggesting a role for SMNTL1 or PR-B in regulating myometrial MYPT1 expression. SMNTL1 was phosphorylated at Ser301 by PKA, and phosphorylation was necessary for SMNTL to translocate to the nucleus, where it may inhibit PR-B (24, 131). Fetalvero et al. (63) showed that prostacyclin, a myometrial relaxant that increases production of cAMP, paradoxically upregulates myometrial contractile-associated proteins, including connexin-43, in a cAMP-dependent manner. The above-mentioned role of SMNTL1 may provide a possible mechanism for this process. SMNTL1 also inhibits MLC20 dephosphorylation in vitro, and immunoprecipitation and immunolocalization experiments showed that SMNTL1 binds MYPT1 in mouse myometrium (131, 265). Lontay et al. (131) proposed that nonphosphorylated SMTNL1 inhibits MP-MYPT1 and that phosphorylation causes its dissociation from MYPT1 and MP-MYPT1 activation. Thus PKA may cause myometrial relaxation by increasing MP-MYPT1 activity and increase expression of contraction-associated proteins to facilitate labor.
Smooth muscle PP1 targets independent of MYPT1.
In addition to MP-MYPT1, other PP1 complexes regulate smooth muscle contractility by targeting ion channels, changing cGMP levels, and regulating the MP-MYPT1 pathway (Fig. 5). PP1 regulates at least two K+ channels. In mouse tracheal smooth muscle, PP1 increases the large-conductance, voltage- and Ca2+-gated K+ (BKCa) channel open probability by dephosphorylating Ser695 (282). In mouse colon, PP1 inhibits the A-type K+ channel 19-pS through its fast inactivation mechanism (7, 236). The human umbilical vein L-type Ca2+ channel was reported to be activated by PP1 (74). However, this study conflicted with a follow-up study in the same cell type (75). This channel is also regulated by PP2A and PP2B.
In gastric smooth muscle cells, cGMP levels can be modulated by PP1. Using tautomycin and immunoprecipitation, Murthy (156) showed that PP1 interacts with and dephosphorylates PDE5 in cells stimulated with the NO donor S-nitrosoglutathione. Activation of PKC reduced the interaction of PP1 with phosphorylated PDE5, and this decrease was reversed by PKC inhibitors.
PP1 holoenzymes can dephosphorylate CPI-17. Eto et al. (57) showed that CPI-17 was dephosphorylated by the PP1 glycogen-targeting holoenzyme in vitro. Additionally, using sequential immunoprecipitation, they showed that CPI-17 and inhibitor-2 associated with different PP1 holoenzymes. They proposed that individual PP1 inhibitor proteins were specific to a subset of PP1 holoenzymes and that insensitive PP1 holoenzymes may regulate these inhibitors. Consistent with this mechanism, they showed that PP1 antagonists increase CPI-17 phosphorylation in permeabilized rabbit femoral artery. This study suggests that non-MP PP1 holoenzymes can regulate MP activity through PP1 inhibitor protein dephosphorylation. Differences in expression or combinations of non-MP PP1 holoenzymes are other possibilities for muscle-specific control of MP-MYPT1.
Role of PP2A in Smooth Muscle Contractility
PP2A regulates multiple proteins involved in smooth muscle contraction (Fig. 5). Okadaic acid (OA), a PP1 and PP2A antagonist, which is selective for PP2A at <1 μM (18, 61), has been extensively used to identify smooth muscle PP2A targets. However, few details are known about the protein phosphorylation sites targeted by PP2A.
PP2A can inhibit L-type Ca2+ channel activity in vascular smooth muscle (75, 161). In inside-out patches of human umbilical vein cells, addition of PP2A inhibited channel open probability but had no effect on channel availability. Navedo et al. (161) showed that PP2A regulates persistent Ca2+ sparklet sites. These sites are areas of sustained Ca2+ influx that are localized to discrete regions within the cell and are caused by phosphorylation of the L-type Ca2+ channel by PKC (162). PP2A inhibitors caused Ca2+ sparklets at sites that had little or no basal sparklet activity but had no effect on highly active sites (161). Navedo et al. proposed that this was due to compartmentalized differences in levels of stimulatory kinase compared with inhibitory phosphatases.
The BKCa channel can be activated by PP2A in bovine trachea, rat mesenteric arteries, and human myometrium in a tissue-specific manner. In trachea, PP2A inhibition prevented a cGMP-specific increase in channel open probability (279). However, this result conflicts with a more recent study that showed that PP2A did not affect BKCa channel activity in trachea (282). In mesenteric arteries, OA caused a 17-fold reduction in 11,12-epoxyeicosatrienoic acid-stimulated increased BKCa channel open probability (46). This increase was associated with increased cAMP, but not cGMP, levels by 11,12-epoxyeicosatrienoic acid treatments. In human nonpregnant myometrial cells, three types of regulation of the BKCa channel by PP2A were shown by inside-out patch-clamp experiments (280). In ∼63% of cells, BKCa channel open probability was inhibited by cAMP and cGMP and was insensitive to OA. In 27% of cells, cGMP, as well as cAMP to a lesser extent, increased channel open probability, which was inhibited by OA. In <10% of cells, cyclic nucleotides and OA did not have an effect on the BKCa channel. It was proposed that these differences were due to alternative expression of BKCa channel isoforms. In support of this concept, BKCa channel isoforms determine whether cAMP or cGMP is stimulatory or inhibitory (237, 278). In inside-out patch-clamp experiments, BKCa channel activity was increased by PKG and PKA in enzymatically isolated pregnant human myometrial cells (281), suggesting that PP2A regulates this channel in pregnant myometrium.
PP2A may inhibit CPI-17, since isolated PP2A can dephosphorylate CPI-17 in turkey gizzard smooth muscle extracts (233). Eto et al. (57) challenged whether this happens in vivo, as CPI-17 phosphorylation was not changed by the PP2A inhibitors OA and fostriecin in permeabilized rabbit femoral artery. However, more recent studies have implicated PP2A in regulating CPI-17. In rabbit intestinal smooth muscle, ET-1 activated PKC but did not increase CPI-17 phosphorylation (83). Addition of OA to ET-1-stimulated preparations increased CPI-17 phosphorylation threefold, and this was antagonized by coincubation with the PKC inhibitor bisindolylmaleimide. Since PKC was activated by ET-1, this suggests that PP2A dephosphorylated CPI-17 and was not inhibiting the PKC pathway. In endothelium-denuded canine basilar artery, OA increased CPI-17 phosphorylation (170). This was inhibited by the PKC inhibitor GO6976, but not the ROCK inhibitor Y-27632, suggesting that PP2A was regulating PKC activity. Whether CPI-17 is a genuine target of PP2A in smooth muscle in vivo requires further elucidation. It is probable that tissue- and/or agonist-specific differences in CPI-17 regulation by PP2A and/or PP1 may explain these conflicting findings.
PP2A can also inhibit PKC, since in canine basilar artery, PKCα phosphorylation at Ser657 and its activity were increased in a dose-dependent manner by treatment with 0.01–1 μM OA (169). A subsequent study using immunoprecipitation assays showed that PKCα and PP2A interacted (170). This study found that stretch inhibited PP2A activity, as well as the PP2A-PKC interaction, in a ROCK-dependent manner. This is an interesting finding, as it suggests that ROCK regulates multiple downstream phosphatases and could therefore control phosphorylation of many targets that are not necessarily phosphorylated by ROCK. Stretch can cause myometrial contractions (258); thus it may involve ROCK regulation of PP2A to activate contractile proteins.
PP2A inhibits cAMP-specific PDE4D5, increasing smooth muscle cAMP levels (157). In vitro, PP2A dephosphorylated PDE4D5 isolated from rabbit stomach smooth muscle cells (157). In these cells, concurrent stimulation with the contractile agonist CCK and the cAMP agonist forskolin significantly increased PDE4D5 activity above forskolin treatment alone. Addition of OA to the forskolin treatment increased PDE4D5 activity to levels that were not significantly different from simultaneous CCK and forskolin treatment. PKC inhibitors reversed the increase in PDE4D5 activity caused by CCK and forskolin, and 32P labeling showed that PP2A was phosphorylated by PKC in vitro. These findings further suggest that PKC can regulate smooth muscle PP2A activity. It is probable that there are agonist- or tissue-specific differences in PP2A-PKC regulation.
HSP27 has been suggested to be a target of PP2A in smooth muscle. In primary cultured cells of rat aorta, Berrou and Bryckaert (16) showed that 0.2–1 μM OA causes a dose-response increase in HSP27 phosphorylation.
Calponin and caldesmon can be inhibited by PP2A-mediated dephosphorylation in vitro (183, 261), but PP2B and PP1 had little effect on their dephosphorylation, and calponin was approximately sevenfold more sensitive to PP2A than was caldesmon. Caldesmon is phasically phosphorylated in spontaneously contracting full-term nonlaboring human myometrial preparations (184), suggesting that PP2A may regulate caldesmon phosphorylation during myometrial contractions.
All the PP2A targets discussed above cause smooth muscle relaxation and, when phosphorylated, facilitate contraction. Hence, PP2A inhibitors are expected to cause smooth muscle contraction. Paradoxically, inhibitor-based studies have shown that ≤1–3 μM OA caused relaxation in rat, guinea pig, dog, pig, and bovine vasculature, rat myometrium, and bovine trachea smooth muscles (1, 10, 11, 18, 28, 29, 102, 109, 211, 235). In myometrium, only >10 μM OA produced contraction or temporary contraction and then relaxation, and these contractions were attributed to PP1 inhibition (10, 28, 29).
OA can cause relaxation of canine basilar artery by increasing PKC-mediated phosphorylation of MLC20 at Thr9, which inhibits myosin ATPase activity (169–171). While these studies explored MLC20 Thr9 phosphorylation in canine basilar artery only, it is possible that this is a common pathway in muscles that are relaxed by PP2A inhibition. Circumstantial evidence also suggests that relaxation caused by PP2A inhibitors involves PKC. In guinea pig hepatic portal vein, OA-induced relaxation is significantly reduced in Triton X-100-skinned muscles, but not in α-toxin-permeabilized muscles (257). PKC levels are decreased by 90% in Triton X-100-skinned muscles compared with intact muscles; however, PKC levels in α-toxin-permeabilized muscle were not reported (112). Taken together, these results suggest that the failure of OA to cause relaxation in Triton X-100-skinned muscles may be due to loss of PKC. However, unknown 67-to 200-kDa PP2A target(s), which can be lost through the larger pore sizes produced by Triton X-100 skinning (257), could also cause OA-induced smooth muscle relaxation.
The paradox between the known relaxant role of PP2A in smooth muscle and relaxation by OA suggests that PP2A may be spatially and/or temporally regulated in smooth muscle. Compartmentalization of PP2A could allow pools of PP2A to be regulated independently and facilitate agonist and/or tissue-specific regulation of PP2A toward its targets.
In humans, expression of PP2A is invariant between full-term nonlaboring and laboring myometrium (12). However, it is possible that expression of PP2A holoenzymes and scaffolding proteins changes in myometrium. A role for myometrial PP2A in the regulation of many of these smooth muscle targets needs to be tested.
Role of PP2B in Smooth Muscle Contractility
PP2B regulates smooth muscle genes and proteins (Fig. 5). PP2B activates the nuclear factor of activated T cells (NFAT) pathway to remodel vascular phenotype, and proteins expressed by this pathway, including K+ channels, lead to changed smooth muscle contractility (8, 24, 147, 165, 206). PP2B also regulates cAMP response element-binding protein through HSP90-dependent interactions with the mineralocorticoid receptor (24). Studies of the effect of PP2B on contractile pathways have mostly focused on ion channels.
PP2B inhibits L-type Ca2+ channels. In human umbilical vein, Ca2+ influx from L-type Ca2+ channels may activate PP2B, which subsequently inhibits channel availability (205). In rat cerebral arteries, PP2B inhibitors increase Ca2+ sparklets at sites with no or only low basal levels of sparklet activity but not in areas of high sparklet activity (161). This finding is identical to that for PP2A inhibitors, suggesting that PP2B and PP2A are targeting the same PKC site(s). Since high sparklet activity causes sustained Ca2+ influx, this suggests that any PP2B feedback mechanism is overwhelmed by higher PKC activity at these sites. Differences in L-type Ca2+ channel regulation by PP1, PP2A, and PP2B suggest that while overlap in dephosphorylation sites occurs, additional phosphatase-specific dephosphorylation sites may exist. The role of each specific phosphatase in regulating L-type Ca2+ channel current will need to be clarified.
PP2B can tissue specifically inhibit inositol trisphosphate (IP3) receptors, preventing Ca2+ release from the sarcoplasmic reticulum (136). In guinea pig colonic smooth muscle, PP2B-bound IP3 receptors and PP2B inhibitors increased IP3-induced Ca2+ release (136). However, inhibition of PP2B in guinea pig aorta and portal vein cells had no effect on IP3-induced Ca2+ release (136, 137).
ATP-sensitive K+ (KATP) channels can be inhibited by PP2B. In rat aortic smooth muscle cells, Wilson et al. (259) used whole cell patch clamp to show that channel activity was decreased by increasing concentrations of intracellular Ca2+ and that PP2B inhibitors increased channel activity. Orie et al. (174) expressed the KATP channel Kir6.1 in HEK-293 cells, and in vitro radioimmunoassay experiments showed that a PKA phosphorylation site in the COOH terminus was dephosphorylated by calcineurin Aα. In rat aorta cells, the cAMP effector protein exchange protein activated by cAMP (Epac) can activate PP2B, causing dephosphorylation of the channel (192). The authors speculated that since Epac is less sensitive than PKA to cAMP, it might act as a buffer against increased channel activity when cAMP levels are high. Taken together, these studies suggest that different pathways can activate PP2B to negatively regulate KATP channel activity during depolarization and cAMP-induced relaxation.
Ca2+-activated Cl− channels may be further activated by PP2B upon Ca2+ stimulation. In rabbit coronary artery cells, cyclosporin A, a PP2B inhibitor, decreased Cl− channel current when cells were dialyzed with 350 or 500 nM extracellular Ca2+ (125). However, cyclosporin A had no effect on currents at 1 μM extracellular Ca2+. At this higher Ca2+ concentration, Ca2+/calmodulin-dependent protein kinase II (CaMKII) inhibition increased channel activity. The authors proposed that PP2B is activated at low intracellular Ca2+ concentrations to promote channel activity, while at higher intracellular Ca2+ levels, CaMKII, which is less sensitive to Ca2+, becomes the primary regulator of the channel. In this manner, opposing PP2B and CaMKII activities may switch channel activity to promote or inhibit depolarization, respectively. The channel responsible for the Ca2+-activated Cl− currents was unknown at the time of these studies, but it is likely that the recently identified transmembrane protein (TMEM) 16A (TMEM16A) and/or its paralog TMEM16B (52) may be involved. Future studies are needed to determine which phosphorylation sites are regulated by PP2B.
These studies suggest that PP2B may act as a negative regulator of specific ion channel activity during sustained contractile or relaxatory signaling processes. Also, it may promote depolarization by inhibiting K+ outflow and increasing Cl− efflux.
In rat and human myometrium, PP2B may dephosphorylate PLCβ3. In rats, PP2B inhibitors significantly reduced dephosphorylation of PLCβ3 by myometrial extracts in vitro (50). In human myometrial cells, the PP2B inhibitor cypermethrin, but not OA, at a concentration that inhibits PP1 and PP2A, prevented dephosphorylation of PLCβ3 at Ser1105 following stimulation of cells with calcitonin gene-related peptide (277). Cypermethrin had no effect when Ser1105 phosphorylation was increased with oxytocin treatment; however, OA delayed PLCβ3 Ser1105 dephosphorylation. This indicated that different stimuli cause dephosphorylation of PLCβ3 by different phosphatases. The mechanisms of this process were not explored, but the data suggest that phosphorylation at additional sites may change the sensitivity of PLCβ3 to PP2B and PP1/PP2A.
Species-specific changes in PP2B expression occur in myometrium. In humans, total PP2B levels and PP2B localization to the plasma membrane declined with labor (12, 116), leading to the hypothesis that PP2B may be involved in uterine quiescence (12). However, in cultured nonpregnant human myometrial cells, oxytocin-activated calcineurin activated the NFAT pathway and upregulated regulator of G protein signaling 2, regulator of calcineurin 1, and Cox-2 (188). These data suggest that calcineurin may promote the labor phenotype and may have a second role in regulating myometrial quiescence. In rats, PP2B is upregulated at day 21 compared with day 19 (50). This suggests that there are species-specific differences in the regulation of contractile proteins and/or expression of NFAT pathway-regulated genes and that PP2B might be more important in initiating labor in rats than in humans.
Smooth muscle phosphatases modulate protein phosphorylation to decrease and even increase contractile force. MP-MYPT1 containing PP1 catalytic, MYPT1 regulatory, and M20 subunits is the major smooth muscle phosphatase. It controls contractile amplitude and duration through MLC20 dephosphorylation.
Multiple mechanisms, including expression of specific isoforms, phosphorylation, inhibitors, and scaffolding proteins, underpin the fine regulation of MP-MYPT1 activity to enable the complex range of tissue- and cell-specific responses to contractile and relaxatory agonists. Inhibition of MP-MYPT1 activity disables it from dephosphorylating MLC20 and stopping contraction and can increase contractile force. This is achieved by MYPT1 phosphorylation at Thr696 and/or Thr853 and can occur through Ca2+-dependent and/or Ca2+-independent signaling pathways. In myometrium, the production of strong contractile force during labor most likely requires MP-MYPT1 inhibition. Myometrial MP-MYPT1 inhibition may be phasic to increase contractile force, while preventing potentially damaging outcomes of tonic myometrial contractions. MP-MYPT1 activity causes relaxation, and cyclic nucleotides are key activators of MP-MYPT1. The pathways that can increase MP-MYPT1 activity include MYPT1 serine phosphorylation, MYPT1 threonine dephosphorylation, telokin activation, and dephosphorylation of inhibitor proteins. These mechanisms are ideal candidates to maintain myometrial quiescence for most of gestation.
Other non-MP PP1 phosphatases, along with PP2A and PP2B, also regulate contractile activity. These phosphatases may affect MLC20 phosphorylation by regulating MP-MYPT1 pathways and Ca2+ signaling. They can also control proteins that affect myosin-actin interactions. Specific inhibition of PP2A causes relaxation in contrast to its expected role, suggesting that important inhibitory feedback mechanisms are activated or PP2A inhibits unknown prorelaxatory proteins. PP2A and PP2B are expressed in myometrium and can directly regulate myometrial proteins involved in contractility. PP2B also upregulates the expression of contraction-associated proteins and, therefore, may be important in initiating labor.
The smooth muscle role of protein phosphatases and their molecular identities require further elucidation, and this will likely reveal novel contractile pathways and proteins against which drugs can be developed to treat smooth muscle pathologies, a pressing need in the setting of the uterus, where defects in physiology are a leading cause of maternal and neonatal death worldwide.
No conflicts of interest, financial or otherwise, are declared by the authors.
T.A.B., N.E.-F., R.S., and E.-C.C. are responsible for conception and design of the review; T.A.B. and J.P. prepared the figures; T.A.B. and E.-C.C. drafted the manuscript; T.A.B., J.P., N.E.-F., R.S., and E.-C.C. edited and revised the manuscript; T.A.B. and E.-C.C. approved the final version of the manuscript.
- Copyright © 2013 the American Physiological Society