Regulation of cardiac excitation and contraction by p21 activated kinase-1

Abstract

Cardiac excitation and contraction are regulated by a variety of signaling molecules. Central to the regulatory scheme are protein kinases and phosphatases that carry out reversible phosphorylation of different effectors. The process of β-adrenergic stimulation mediated by cAMP dependent protein kinase (PKA) forms a well-known pathway considered as the most significant control mechanism in excitation and contraction as well as many other regulatory mechanisms in cardiac function. However, although dephosphorylation pathways are critical to these regulatory processes, signaling to phosphatases is relatively poorly understood. Emerging evidence indicates that regulation of phosphatases, which dampen the effect of β-adrenergic stimulation, is also important. We review here functional studies of p21 activated kinase-1 (Pak1) and its potential role as an upstream signal for protein phosphatase PP2A in the heart. Pak1 is a serine/threonine protein kinase directly activated by the small GTPases Cdc42 and Rac1. Pak1 is highly expressed in different regions of the heart and modulates the activities of ion channels, sarcomeric proteins, and other phosphoproteins through up-regulation of PP2A activity. Coordination of Pak1 and PP2A activities is not only potentially involved in regulation of normal cardiac function, but is likely to be important in patho-physiological conditions.

1. Introduction

Studies in the heart first elucidated effects and key proteins in major signaling by adrenergic and cholinergic pathways. Transduction of these signals involves reversible phosphorylation of proteins controlling ion channels, exchangers and pumps, sarcomeric and cytoskeletal function, energy metabolism, and gene transcription and translation. Kinase cascades that phosphorylate these proteins are relatively well worked out compared to the phosphatases that dephosphorylate these proteins. Moreover, modulation of the activity of phosphatases in the integrative biology of control of cardiac dynamics remains poorly understood. There is some understanding of control of phosphatases such as calcineurin by Ca²⁺/calmodulin and protein phosphatase 1 (PP1) by PKA and inhibitor 1, but there is little understanding of signaling to protein phosphatase 2A (PP2A), a major cardiac phosphatase. We review here emerging evidence indicating that p21 activated kinase-1 (Pak1), a serine/threonine protein kinase directly activated by Cdc42 and Rac1 is an important signaling molecule potentially involved in major cardiac processes regulating excitation and contraction in healthy and disordered hearts.

1.1. Discovery

Studies on the Rho family of small GTPases including RhoA, Cdc42 and Rac1, which have diverse functions in mammalian cells, such as modulation of ROS (reactive oxygen species) (Knaus et al., 1991; Mizuno et al., 1992), regulation of cell proliferation and differentiation, apoptosis, and cytoskeletal reorganization (Hall, 1992; Ridley, 1995) led to the identification of Pak1. Pak1 was discovered in rat brain as a major binding partner for Cdc42 and Rac1 by protein overlay assay. Both the kinase activity and autophosphorylation of Pak1 in vitro increase significantly upon binding to the small G proteins (Manser et al., 1994).

Pak1 is a member of a family of serine/threonine protein kinases exhibiting direct activation by Cdc42 and Rac1. Pak1, 2 and 3 belong to group I of the Paks and share substantial sequence homology with each other, especially in their catalytic domains (Jaffer and Chernoff, 2002) (Fig. 1). Actually, in early studies Paks had been isolated from rabbit reticulocyte with unusually low kinase activities. When treated with proteases, the kinase activities increased significantly (Tahara and Traugh, 1981). A protease activated kinase (PakI) phosphorylates myosin regulatory light chain from smooth and skeletal muscle cells (Tuazon et al., 1982; Tuazon and Traugh, 1984). Unlike PKA and PKG, the Pak activities are independent of cyclic nucleotides (Tahara and Traugh, 1981). Molecular cloning of the protease activated kinases indicates that they are identical with the small G protein activated kinases (Jakobi et al., 1996).

Fig. 1.

The group I p21 activated kinases include Pak1 (α-Pak), Pak2 (γ-Pak) and Pak3 (β-Pak). DI—dimerization domain, PBD—p21 binding domain, KI—kinase inhibitory domain. —proline-rich sequence, aa—amino acids. Both the regulatory and the catalytic domains of group I Paks are highly homologous to each other.

Pak1 induces the same cytoskeletal changes as its upstream activators Cdc42 and Rac1, such as dissolution of stress fibers and focal adhesion complexes, formation of filapodia, lamellipodia and membrane ruffles in mammalian cells (Manser et al., 1997; Sells et al., 1997). This has led to speculation for a potential role of Pak1 in cardiac remodeling (Sussman et al., 2000).

1.2. Structure

Pak1 protein is divided into an N-terminal regulatory domain and a C-terminal catalytic domain, each is about the half of the enzyme in amino acid sequence (Fig. 1). Even though Pak1 has very low sequence similarity with PKA, the prototype of all the serine/ threonine protein kinases, Pak1 and PKA still share substantially similar kinase motifs in their catalytic domains. As with many other serine/threonine protein kinases, such as Mek1, PKA and PKC, Pak1 contains an activation loop in its catalytic domain and an autophosphorylation site inside the loop (Lei et al., 2000; Manser et al., 1997). The N-terminal half of Pak1 contains a few regulatory sequences. The p21 binding domain (PBD) is upstream of and partially overlaps with a kinase inhibitory domain that imposes an inhibitory effect on the catalytic activity by stabilizing an autoinhibitory configuration of the kinase. Sequence upstream of the PBD is involved in Pak1 dimerization (DI). The boundary between DI and PBD is not clear (Jaffer and Chernoff, 2002; Lei et al., 2000) (Fig. 1).

There are five proline-rich motifs scattered over the N-terminal region of Pak1. Some of these proline-rich motifs are followed by one or two auto-phosphorylation sites at the serine residues (Jaffer and Chernoff, 2002; Manser et al., 1997). Nck binds to the first proline-rich motif that provides a potential link of Pak1 activity to cell growth signals (Bokoch et al., 1996; Galisteo et al., 1996). Pak1 interacts with a G protein exchange factor (GEF) pix through a proline-rich region downstream of the kinase inhibitory sequence. It also contains a unique dynein light chain interacting sequence (Lu et al., 2005; Vadlamudi et al., 2004) that is absent in either Pak2 or Pak3.

1.3. Auto-phosphorylation and dimerization

Pak1 contains seven serine/threonine auto-phosphorylation sites most of which are localized in the N-terminal half of the enzyme (Fig. 2). Threonine 423 is the only auto-phosphorylation site in the kinase domain and is localized inside the activation loop (Manser et al., 1997). Auto-phosphorylation at the activation loop of Pak1, a common feature of many other serine/threonine protein kinases, is required for kinase activation. As occurs in Mek-1, PKC, and other protein kinases, substitution of the threonine in the activation loop with aspartic or glutamic acid, renders Pak1 constitutively active (Manser et al., 1997).

Fig. 2.

Pak1 has multiple auto-phosphorylation sites which are mostly clustered in the N-terminal half of the kinase. Auto-phosphorylation of T423 localized at the catalytic domain is critical for the kinase activation. Substitution of T423 for a glutamic acid renders the kinase constitutively active.

Group 1 Paks demonstrate dimerization as a regulatory mechanism, which is common to protein kinases related to yeast Ste20 and Cla4. For example, MST1 forms a homodimer and contains both a dimerization and an auto-inhibitory region (Creasy et al., 1996). Studies from X-ray crystallography indicate that Pak1 forms a homodimer in an anti-parallel configuration (Lei et al., 2000). Formation of the homodimer is essential for auto-inhibition involving amino acids in both regulatory and catalytic domains of the kinase (Lei et al., 2000). This is mirrored by the observation that phosphorylation at multiple auto-phosphorylation sites contributes to kinase activation (Chong et al., 2001; Gatti et al., 1999). Differences in amino acids from the wild type sequence may also have the same effect on the kinase activity. In Pak3, single amino acid mutations spread over different regions of the kinase produce a similar phenotype of mental retardation (Bienvenu et al., 2000; Gedeon et al., 2003). The effect of the mutations on auto-inhibition and the kinase activity remains unclear. Auto-phosphorylation attenuates auto-inhibition and dimer formation in a multiple-step process that may eventually lead to breakdown of the dimer, a structure preventing the access of other protein substrates to the catalytic domain of Pak1 (Parrini et al., 2002). Interestingly, in another study, phosphorylation in the activation loop appears to promote formation of dimer (Pirruccello et al., 2006). Assuming auto-phosphorylation occurs through an intermolecular mechanism (phosphorylation between two Pak1 monomers), Pak1 monomers maintain the capacity to physically interact with each other until the last autophosphorylation site is occupied by a phosphate group. No matter whether auto-phosphorylation occurs through an inter or intramolecular mechanism, a protein substrate must compete with the auto-phosphorylation site(s) for the catalytic center of Pak1.

Phosphorylation at the N-terminal serine sites also regulates Pak1 partnering with other cellular proteins. When serine 21, the first auto-phosphorylation site is substituted by an aspartic acid, Nck binding to the amino acid sequence upstream of the serine 21 is significantly attenuated (Zhao et al., 2000). Pak1 has demonstrated translocation associated with kinase activation. In ventricle myocytes and sino-atrial (SA) node pacemaker cells, the intracellular localization of the endogenous Pak1 has defined patterns. The constitutively active Pak1 localizes differently in these cells (Ke et al., 2004; Ke et al., 2007a).

1.4. Regulation by upstream signals

Soon after discovery of Pak1, it was reported that Pak1 activities are functionally linked to the inhibitory G proteins in leukocytes (Knaus et al., 1995). This has been supported in a later study that in fibroblasts, LPA (lysophosphatidic acid)-induced cell spreading is blocked by pertussis toxin and the blockage can be overcome by expression of constitutively active Cdc42 and Rac1 (Ueda et al., 2001). Bradykinin and acetylcholine, which stimulate Gi-coupled receptors, produce the same phenotype as Cdc42 and Rac1 in 3T3 and N1E-115 cells (Kozma et al., 1995; Kozma et al., 1997). Receptors for bradykinin and acetylcholine are coupled to Gi (Liebmann et al., 1990; Migeon et al., 1995). FLJ00018 is a G protein exchange factor (GEF) potentially involved in activation of Cdc42/Rac1 by Gi-coupled receptors (Ueda et al., 2008).

Sphingosine-1 directly activates both Pak1 and Pak2 bypassing their upstream regulatory proteins (Bokoch et al., 1998; Roig et al., 2001). Auto-phosphorylation induced by sphingosine-1 occurs at the same sites as those on Pak1 stimulated by Cdc42 and Rac1 (Roig et al., 2001). The mechanism whereby Pak is activated by the small organic molecule remains unclear. Sphingosine-1 may interact with Pak1 at the p21 binding domain (Zenke et al., 1999). However, in other studies, sphingosine-1 and Cdc42/Rac1 appear to have cooperative effect on activation of Pak1 (Chong et al., 2001). Recent studies indicate that Pak1 isolated from bovine ventricle muscle is activated by C2 and C6 ceramides, which are structurally very similar to sphingosine-1 (Ke and Solaro, 2008).

1.5. Cytoskeletal effects

Cell remodeling produced by Pak1 and its upstream signals has suggested its potential role in cardiac hypertrophy and dilatation. However, the molecular mechanism underling the cytoskeletal effect produced by Paks remains uncertain. In HeLa cells, constitutively active Pak1 induces loss of stress fibers and dissolution of focal adhesion complexes (Manser et al., 1997; Sells et al., 1997). On the other hand, expression of the constitutively active Pak1 also induces formation of lamellipodia, filopodia and membrane ruffles (Sells et al., 1997; Sells et al., 1999). Lamellipodia and filopodia are structures that contain filamentous actins.

Formation and extension of F-actin are regulated by protein phosphorylation and external signals and the control of actin polymerization by Pak1 may involve cofilin. Cofilin is a cytoskeletal regulatory protein that promotes depolymerization of actin filaments. Phosphorylation of cofilin at Ser 3 by Lim kinase removes the inhibitory effect of cofilin on actin polymerization (Arber et al., 1998; Yang et al., 1998). Lim kinase is phosphorylated and activated by GST-Pak1, suggesting a signaling pathway from Pak1 to cofilin that leads to polymerization of actin (Edwards et al., 1999). This signaling pathway predicts a stimulatory effect of Pak1 on growth of actin filaments, which is not supported by the effect of Pak1 on dissolution of stress fibers in HeLa cells (Manser et al., 1997). Formation of lamellipodia and filopodia is also regulated by microtubules (Gordon-Weeks, 2004). Some protein kinases and protein phosphorylation appear to have a negative effect on formation of microtubule networks (Hlavanda et al., 2007).

The role of Pak1 on cytoskeletal structure and function also involves signaling to myosin regulatory light chain (MLC₂₀). Phosphorylation of myosin regulatory light chain regulates assembly of myosin II (Smith et al., 1983). A recent report indicates that overexpression of the cardiac-specific form of myosin light chain kinase, which is Ca²⁺/camodulin independent, induces cell remodeling in cardiac myocytes (Chan et al., 2008). This is supported by the observation that expression of constitutively active Pak1 in HeLa cells induced both cell remodeling and dephosphorylation of myosin regulatory light chain (Manser et al., 1997; Sanders et al., 1999). GST-Pak1 also phosphorylates myosin light chain kinase in vitro that may interfere with its interaction with Ca²⁺/camodulin and thus inhibit the kinase (MLCK) activity (Sanders et al., 1999). Myosin regulatory light chain is the sole substrate for MLCK. Therefore, dissolution of stress fibers appears due to a Pak1-MLCK-myosin regulatory light chain (MLC₂₀) pathway. However, the effect of GST fusion protein which is about 26 kD, on Pak1 activation, dimerization and substrate specificity are not well characterized. Constitutively active Pak1 does not phosphorylate myosin light chain kinase and has no effect on phosphorylation of MLC₂₀ in smooth muscle cells (Ke et al., 2007b).

Pak1 is highly expressed in muscles and is abundant in different regions of the mouse heart (Fig. 3). It is localized to Z-disc, cell and nuclear membrane and intercalated disc of rat ventricular myocytes (Ke et al., 2004). In guinea pig SA node pacemaker cells, Pak1 is expressed in different patterns. In central pacemaker cells, expression of Pak1 is evenly distributed in different parts of the cells. In peripheral pacemaker cells, Pak1 expression shows striations. A mixed pattern is observed in an intermediate type of the cell (Ke et al., 2007a) (Fig. 4). Although there are indications that Pak1 may have strong effects on cardiac remodeling, direct evidence is still not available. In neonatal cardiomyocytes, Cdc42 increases the length/width ratio of the cells (Nagai et al., 2003). In Rac1 transgenic mice, cardiac hypertrophy occurs accompanying with altered intracellular localization of Pak1. Expression of the constitutively active Pak1 in rat neonatal cardiomyocytes increased the cell length by 50% (Y. Ke, unpublished observation). Heart specific knock out of Rac1 abolishes angiotensin II induced cardiac hypertrophy (Satoh et al., 2006).

Fig. 3.

A. A mouse heart section showing posterior view. Pak1 is highly expressed in different regions of the heart including in large blood vessels. B and C. Pak1 and PP2A demonstrate the same localization in mouse myocardium. The mouse heart was sliced in 10 µm sections and stained with Pak1 antibody (rabbit) and antibody against the catalytic subunit of PP2A (mouse). Pak1 (B) and PP2Ac (C) were detected by FITC and Rhodamine conjugated secondary antibodies respectively. Images of B and C were taken from a ventricle region in image A marked by a white box and a black arrow. The magnification of objective in A is 10× and that in B and C is 60×.

Fig. 4.

Examples of distinct patterns of expression of Pak1 in SAN cells demonstrated by confocal microscopy. In central pacemaker cell (A), expression of Pak1 demonstrates a “diffused pattern” without striation. In peripheral pacemaker Cell (C), Pak1 has a striated expression. The intermediate cell (B) has a mixed pattern. The figure is reproduced from Cir. Res. by permission (Ke et al., 2007a).

2. Regulation of excitation and contraction by Pak1 in the heart

2.1. Regulation of protein phosphorylation and activities by β-adrenergic signaling pathways

As reviewed in a later section, a prominent mechanism by which Pak1 affects cardiac function involves activation of PP2A and reversal of effects of β-adrenergic stimulation. The β-adrenergic signaling pathways play central roles in regulation of excitation and contraction/relaxation dynamics in the heart. Intracellular accumulation of cyclic AMP is enhanced by agonists of β-adrenergic receptors through the stimulatory heterotrimeric G proteins (G_s) that activate adenylate cyclase. cAMP releases the inhibitory effect of the regulatory subunit (R subunit) on the catalytic subunit of cAMP dependent kinase (PKA). Then the catalytic subunit of PKA is separated from the R subunit, autophosphorylated and becomes active towards multiple regulatory proteins in cardiac cells including cardiac troponin I (cTnI), myosin-binding protein C (MyBP-C), phospholamban, ryanodine receptors (RyR) and L-type Ca²⁺ channels.

2.1.1. cTnI

cTnI is an inhibitory element in the troponin/tropomyosin complex containing a Ca²⁺ binding protein troponin C, an anchor protein troponin T, and tropomyosin that binds to F-actin and directly regulates the reactions of myosin cross-bridges with the thin filament. Phosphorylation of cTnI at serine 23, 24 at the unique N-terminal extension reduces affinity of troponin C for Ca²⁺ and decreases myofilament Ca²⁺ sensitivity of tension development (Robertson et al., 1982). During cross-bridge cycling, detachment of the myosin head from the thin filament is facilitated by PKA phosphorylation of cTnI (Biesiadecki et al., 2007; Kentish et al., 2001; Strang et al., 1994). These effects of cTnI phosphorylation are important elements in the enhanced relaxation kinetics and abbreviation of the contraction/relaxation cycle that matches cardiac dynamics to the increased heart rate. In hearts of a transgenic mouse expressing only the slow skeletal troponin I (Tg-ssTnI), which lacks PKA phosphorylation sites, the myofilaments demonstrate increased Ca²⁺ sensitivity and the transgenic heart has impaired relaxation (Fentzke et al., 1999) and blunted response to β-adrenergic stimulation. PKA phosphorylation on cTnI also enhances the length dependent activation of myofilaments (Konhilas et al., 2003), a mechanism critical to the Frank–Starling relation of the heart.

2.1.2. Phospholamban (PLB)

PLB is a small proteo-lipid forming a pentamer of about 25 kD that interacts and inhibits the activities of SERCA2a, a pump responsible for re-uptake of Ca²⁺ from cytosol to sarcoplasmic reticulum during diastole. In coordination with cTnI phosphorylation, phosphorylation of phospholamban at serine 16 by PKA releases its inhibitory effect on SERCA2 and promotes the rate of cardiac muscle relaxation (Koss and Kranias, 1996; Kranias and Solaro, 1982). Ablation of PLB in a mouse model induced enhanced myocardial contractility with diminished response to β-adrenergic stimulation (Luo et al., 1994). Cross-breeding this PLB “knock-out” mouse with the Tg-ssTnI mouse generated hearts with no abbreviation of the contraction/relaxation cycle upon stimulation with isoproterenol (Wolska et al., 2002)

2.1.3. Ryanodine receptor (RyR)

The RyR is a giant molecule forming a homotetramer with a molecular weight of about 2.5 MDa. In cardiac muscle, voltage-initiated Ca²⁺ current induces Ca²⁺ release from RyR receptor (Ca²⁺ induced Ca²⁺ release or CICR). Phosphorylation of RyR by PKA at serine 2809 attenuates FKBP12.6 binding to RyR and sensitizes the channel (Marks, 2001; Marx et al., 2000]). β-blockade reverses hyperphosphorylation of RyR by PKA and increase level of the SR Ca²⁺ store (Reiken et al., 2001). RyR receptor associates with multiple signaling molecules including phosphatases PP1 and PP2A (Marks, 2001).

2.1.4. L-type Ca²⁺ channel, (dihydropyridine receptor; DHPR)

L-type Ca²⁺ channels produce large and long-lasting Ca²⁺ currents, which are different from currents from the T-type Ca²⁺ channel, and which are critical to CICR. The voltage gated Ca²⁺ current carried by the L-type Ca²⁺ channels triggers release of Ca²⁺ through RyRs. These Ca²⁺ channels are regulated by PKA-dependent phosphorylation of serine 1928 located within the intracellular tail region of the Cav1.2 (L-type Ca²⁺ channel α1C subunit) (Leach et al., 1996). The L-type Ca²⁺ current is responsible for the plateau phase of the action potential in working cardiac myocytes and is a determinant of diastolic depolarization of the pacemaker action potentials in SA nodal cells (Eisner et al., 2003; Kamp and Hell, 2000; Vinogradova et al., 2005). The PKA effect on L-type Ca²⁺ channels is reversed by PP2A in the heart (Table 1).

Table 1.

Regulation of (cardiac) ion channel activities by β-adrenergic stimulation and by PP2A.

Ion channels	Regulation by β-adrenergic stimulation	Regulation by PP2A	Reference
L-type Ca2+ channels	Yes	Yes	Carter et al., 2006; Chen et al., 2002; Collis et al., 2007; Davare et al., 2000; duBell et al., 2002; Gao et al., 1997; Gerhardstein et al., 1999; Groschner et al., 1996; Hall et al., 2006; Ji et al., 1999; Klein et al., 2003; Kamp and Hell, 2000; Perets et al., 1996; Vinogradova et al., 2006; Walsh and Wang, 1996;
SR Ca2+ release channels or Ryanodine receptors (RyR)	Yes	Yes	Bers, 2006; Carter et al., 2006; Ferrier et al., 1998; Hain et al., 1995; Marks, 2001; Marx et al., 2002; Sobie et al., 2006; Uehara et al., 2002; Yano et al., 2005
Delayed rectifier potassium channel	Yes	Yes	Huang et al., 1994; Ke et al., 2007a; Peralta, 1995
Inwardly rectifier potassium channel	Yes	Yes	Barros et al., 1993; Inoue and Imanaga, 1995; Zitron et al., 2004
Acetylcholineactivated channels (Ik(Ach); Ik(Ado))	Yes	Yes	Luo et al., 1998; Mullner et al., 2000; Nikolov and Ivanova- Nikolova, 2004
ATP-sensitive potassium channel	Yes	Yes	Lin et al., 2000; Light et al., 1996
Ca2+-activated K+ currents (BK channel)	Yes	Yes	Hayashi et al., 2004; Lin et al., 2006; Widmer et al., 2003; Zhou et al., 1996
G protein activated potassium channel	Yes	Yes	Mullner et al., 2000; Nikolov and Ivanova-Nikolova, 2004
HERG	Yes	Unknown	Choe et al., 2006; Cui et al., 2000; Kagan et al., 2002; Kiehn et al., 1998; Thomas et al., 2003
KCNQ1-KCNE1, I(Ks) (slowly activating potassium current) channel	Yes	Yes	Chen et al., 2005; Marx et al., 2002
Kv11.1 K+ channel	Yes	Unknown	Tutor et al., 2006
cardiac Na/K ATPase	Yes	Yes	Bhasin et al., 2007; Pavlovic et al., 2007
Connexin 43	Yes	Yes	Ai and Pogwizd, 2005; Duncan and Fletcher, 2002
Sodium/calcium exchange (NCX)	Yes	Yes	Reppel et al., 2007; Schulze et al., 2003;
Sodium channel	Yes	Unknown	Becchetti et al., 2002; Frohnwieser et al., 1997; Lu et al., 1999; Mullner et al., 2000; Murphy et al., 1996; Schreibmayer et al., 1994
Cl-channel (SR)	Yes	Unknown	Kawano et al., 1992
Cl-channel	Yes	Yes	Bahinski et al., 1989; Chow and Barrett, 2007; Luo et al., 1998; Nagel et al., 1992; Zhang et al., 1994; Zitron et al., 2008

The β-adrenergic cascade had been considered as the single most important signaling process in the heart. Nevertheless, emerging evidence supports the hypothesis that a parallel anti-adrenergic signaling pathway exists and balances the functional effects of β-adrenergic stimulation.

2.2. The under-estimated role and regulation of phosphatases in control of cardiac function

For many years phosphatases were considered as the “passive” enzymes that execute dephosphorylation with a constitutive activity (Solaro, 2000). In the case of β-adrenergic stimulation, the dominant idea was that modulation of cAMP levels and PKA activity mainly controlled levels of phosphorylation of key regulatory proteins (Fig. 5). In this mechanism the anti-adrenergic effect in the heart is through down regulation of β-adrenergic signaling cascade including inhibition of adenylate cyclase activity by Gi (Fig. 5). However the following representative studies demonstrated the existence of additional mechanisms controlling protein phosphorylation: i) When cardiomyocytes were treated with agents activating Gi coupled muscarinic receptors or adenosine receptors, the intracellular level of cAMP remained unchanged, whereas phosphorylation of cTnI and PLB was significantly reduced (Gupta et al., 1993; Gupta et al., 1994). ii) In endothelial cells, both the target(s) of pertussis toxin and cAMP play the same role in promotion of barrier integrity (Patterson et al., 1995; Lum et al., 1999). iii). Recently, Liu and Hofmann (2002, 2003) reported that inhibition of phosphorylation of cTnI and PLB by Gi may be mediated by P³⁸ MAP kinase and PP2A in cardiomyocytes. iv). In a rat model of long term β-adrenergic stimulation, the activities of both PP1 and PP2A increased (Boknik et al., 2000). v). In rabbit with non-ischemic heart failure, increased dephosphorylation of connexin 43 was reported to be associated with enhanced PP2A co-localization with the ion channel (Ai and Pogwizd, 2005).

Fig. 5.

A conventional scheme of regulation of cardiac function by β-adrenergic signaling cascades. Cardiac troponin I (cTnI), phospholamban (PLB), ryanodine receptor (RyR), L-type Ca²⁺ channel (DHPR), etc are phosphorylated by PKA. PP2A is responsible for dephosphorylation of these substrates and is generally considered constitutively active. Adenylate cyclase is stimulated by Gs and inhibited by Gi. +P = phosphorylation. −P = dephosphorylation.

PP1 and PP2A are major protein phosphatases in the heart and account for more than 90% of dephosphorylation of the phosphoproteins (Luss et al., 2000). In ventricular myocytes, PP2A has the same pattern of localization as Pak1 (Fig. 3). PP2A is often responsible for removal of phosphate groups at PKA phosphorylation sites in cardiac muscle and dephosphorylates cTnI (MacDougall et al., 1991), myosin-binding protein C (Ke et al., 2004), phospholamban (MacDougall et al., 1991), inwardly rectifier potassium channels (Barros et al., 1993) and L-type Ca²⁺ channels (Davare et al., 2000) (Table 1). PP2A has a stimulatory effect on generation of Ca²⁺ sparks (Terentyev et al., 2003) and regulates transcription factors such as CREB (Wadzinski et al., 1993) and NFκB (Yang et al., 2001), which are involved in cardiac remodeling (Fentzke et al., 1998; Frantz et al., 2003). These findings indicate that PP2A may function to balance the effects of β-adrenergic stimulation in the heart.

2.3. Role of PP2A in end-stage heart failure and in regulation of ion channel activity

A hallmark of congestive heart failure is reduced protein phosphorylation at PKA sites in cTnI and phospholamban associated with impaired relaxation of cardiac muscle (de Tombe and Solaro, 2000). Investigation of the etiology of heart failure in the past has focused on cAMP-mediated signaling cascades. Reduced expression of β-adrenergic receptors, change of the ratio of β-1/β-2 adrenergic receptors, and increased Gi activity with reduced accumulation of cAMP may all account for down regulation of PKA activity and reduced phosphorylation of both cTnI and PLB (Bishop et al., 1998). However, emerging evidence indicates that altered activity of PP2A may play an equally important role during the progression of heart failure (Boknik et al., 2000). Activation of PP2A by Pak1 increases myofilament sensitivity to Ca²⁺ in cardiac myocytes (Ke et al., 2004). The enhanced sensitivity of the myofilaments to Ca²⁺ that occurs with dephosphorylation of the myofilaments (van der Velden et al., 2003; Wolff et al., 1996) is especially significant in that mutations genetically linked to hypertrophic myopathies induce enhanced myofilament Ca²⁺ sensitivity (Brixius et al., 2002; Muthuchamy et al., 1999).

Altered PP2A activity may also affect regulation of multiple ion channels (Table 1). Studies of Davare et al., (2000) indicated that PP2A is associated with class C L-type Ca²⁺ channels and antagonizes PKA effects on channel activity (Davare et al., 2000). While in patients with atrial fibrillation, the expression of PP2A is reduced with decreased L-type Ca²⁺ channel open probability in atrial myocytes (Klein et al., 2003). Regulation by both PKA and PP2A is a mechanism common to many ion channels in cardiac and other mammalian cells (Table 1).

2.4. Regulation of PP2A by protein/protein interactions and by Pak1

One of the exciting directions of research into control of cardiac contractility is investigation of the role of Pak1 in control of PP2A. PP2A is a multifunctional protein consisting of a heterotrimer A, B, and C subunits. Subunit A is a scaffolding protein and forms a core enzyme with the C subunit exhibiting catalytic activity. At least 18 different B subunits have been identified in mammalian cells and they are coded by four unrelated families of genes, B (PR55), B′ (PR61), B″ (PR72/130) and B‴(PR93/110) (Janssens and Goris, 2001; Kamibayashi et al., 1994). In cardiac cells, different B isoforms demonstrate different patterns of intracellular localizations (Gigena et al., 2005). The catalytic subunit (C) is highly conserved and can be modified by tyrosine phosphorylation at Y307 and methylation at L309 in response to extracellular signals (Chen et al., 1992; Chen et al., 1994; Janssens and Goris, 2001).

The activity of PP2A is subjected to regulation by viral gene products. SV40 small t antigen and the small/middle T antigens of polyoma virus directly bind to PP2A. The small and middle T antigens bind to a region in PR65/A displacing the B subunit and inhibit the enzymatic activity (Walter et al., 1990; Pallas et al., 1990). Cells expressing small and middle T antigen are growth-stimulated with elevated MAP kinase activities (Sontag et al., 1993). PP2A may also be regulated by cellular proteins through a direct protein–protein interaction. Association of Tap42 with PP2A regulates yeast cytoskeletal reorganization during the cell cycle (Wang and Jiang, 2003). Regulation of PP2A by casein kinase is also through protein/ protein interaction without transfer of phosphate groups between two proteins (Heriche et al., 1997).

Investigations into the role of Paks related to myosin II activity illustrate the daunting task of identifying physiological substrates for Paks. Pak2 phosphorylation of myosin regulatory light chain from endothelial cells predicts an increase of myosin II activity (Zeng et al., 2000). However, subsequent studies reported that the same kinase phosphorylates myosin light chain kinase and induces dephosphorylation of myosin regulatory light chain leading to decreased activity of myosin II (Goeckeler et al., 2000). Studies in cardiac myocytes indicate that Pak1 interacts physically with and up-regulates the activities of PP2A.

An interaction between Pak1 and PP2A was first identified in rat brain suggesting a possible signaling pathway for Paks (Westphal et al., 1999). Both Pak1 and Pak3 physically associate with the catalytic subunit of PP2A. Cross-linking of PP2A with its partners in a protein complex shifts Pak1 and Pak3 to higher molecular weights on SDS gel analysis. (Westphal et al., 1999). Constitutively active Pak1 also associates with PP2A when purified from cultured mammalian cells (Ke et al., 2004; Ke et al., 2007b). PP2A carries out dephosphorylation from the auto-phosphorylation sites on Pak (Zhan et al., 2003). In cardiomyocytes, Pak1 associates with PP2A, induces post-translational modification of PP2A and dephosphorylation of cTnI at the PKA sites (Ke et al., 2004; Sheehan et al., 2007), suggesting that Pak1 and PP2A form a regulatory module (Fig. 6).

Fig. 6.

Pak1 forms a signaling module with PP2A. Auto-phosphorylation occurs when it is activated by upstream signals such as Cdc42/Rac1 and sphingosine-1. The associated protein PP2A becomes auto-dephosphorylated and the phosphatase activity on cardiac regulatory proteins increases.

2.5. Possible link of Pak activity with cardiac arrhythmia

The potential role of Pak1 in regulation of ion channel activities was demonstrated by studies of Pak1 function in isolated guinea pig sino-atrial (SA) node preparation and single SA node pacemaker cells (Ke et al., 2007a) (Fig. 7). We first perfused isolated guinea pig hearts with either recombinant adenovirus (AdPak1) expressing constitutively active Pak1 or the control virus AdLacZ, and then monitored the electrical activities of the SA node in the presence of increasing doses of isoproterenol. In SA node expressing active Pak1, responses to isoproterenol stimulation manifested by the frequency of firing electrical impulse were significantly depressed compared to those of controls. We also characterized the beating rate of cultured SA nodal cells infected with AdPak1 or AdLacZ. Increase of beating rate in SA node cells expressing the active Pak1 was diminished when the cells were treated with isoproterenol. While in cells infected with AdLacZ, the enhanced beating rate increased by 40–50% in the presence of 5 nM of isoproterenol. Whole cell patch clamping demonstrated that both L-type Ca²⁺ current and the delayed rectifier K current had a repressed response to isoproterenol in cells expressing the active Pak1. In the SA node cells, constitutively active Pak1 redirects intracellular localization of PP2A (Ke et al., 2007a).

Fig. 7.

(A and B), Representative L-type Ca²⁺ current recordings from cultured SAN cells infected with Ad-LacZ or Ad-Pak1 in the absence and presence of 100 nmol/L ISO for 5 min. Currents were recorded during 200-ms step depolarizations from a holding potential of −50 mM to a range of potentials between −40 and +50 mV. (C and D), Current-voltage relationship of lea in cells infected with Ad-LacZ or Ad-Pak1, in the absence and presence of ISO. (E), Active Pak1 inhibits phosphorylation of cTnI in adult rat cardiac myocytes. Myocytes were first infected with either AdPak1 or AdLacZ at moi of 100 and incubated in ³²P to label the nucleotide pool. Lane 1, Basal phosphorylation of cTnI, ventricular myosin light chain 2 (MLC2v) in myocytes infected with AdLacZ. Lane 2, Positive control demonstrating phosphorylation of cTnI when myocytes without viral infection were treated with 1 µmol/L isoproterenol. Lane 3, Demonstration of reduction in phosphorylation of cTnI in myocytes infected with AdPak1. The same results were obtained in 5 separate experiments. F, Ca²⁺-dependent isometric tension development of detergent-extracted single myocytes cultured in the presence of AdLacZ or AdPak1. Myocytes were first infected with either AdPak1 or AdLacZ at moi of 100, and membranes were extracted in Triton X-100. Isometric tension of single myocytes infected with AdPak1 (●) and AdLacZ (○) was measured as a function of Ca²⁺. The half-maximally activating free Ca²⁺ (EC₅₀) was significantly lower (P<0.05) in AdPak1-infected myocytes (0.87 ± 0.14 µmol/L; n = 4) than in AdLacZ myocytes (1.43 ± 0.08 µmol/L; n = 3). The maximum tension (Hill coefficient) was 32.2 ± 9.8 mN/mm² (3.9 ± 0.8) for AdPak1 myocytes and 23.1 ± 5.4 mN/mm² (3.8 ± 0.6) for AdLacZ controls (P> 0.05). Data are presented as mean ± SEM. The Figure is reproduced from Circ. Res. (Ke et al., 2004 and Ke et al., 2007a) with permission.

Based on these results, we suggest that there is a dynamic balance between kinase and phosphatase activity in control of L-type Ca²⁺ channel as well as delayed rectifier K⁺ channel activity in cardiac cells. The balance between these kinase and phosphatase actions may be of importance in controlling cardiac pacemaker activity in response to autonomic and humoral stimulation. The importance of this balance is highlighted by the recent report by Vinogradova et al. of a high basal PKA-dependent phosphorylation in SA node pacemaker cells that drives rhythmic internal Ca²⁺ store oscillations and spontaneous beating of these cells (Vinogradova et al., 2005). Alteration of Pak1 activity may lead to the alteration of the dynamic regulatory processes and balance between kinase and phosphatase activity in cardiac cells, and thus result in change in various ion channel activity, the later is clearly arrhythmogenesis. It will be important to further investigate the regulatory role of Pak1 on ion channels in the heart under physiological and pathological conditions.

Our observation in the SA node indicates that activation of Pak1 may also be a potential therapeutic strategy for prevention and inhibition of tachycardia triggered by catecholamines in different pathological conditions of the heart. Pak1 is highly expressed in atria from different mammalian species. Yet, the potential role of Pak1 in regulation of electrical activities in atrium remains unclear. Important questions regarding the role of Pak1 in atrial fibrillation are the following: i) How is Pak1 activation related to regulation of major ion channel currents in atrial myocytes? ii) In a heart with overall increase or decrease of Pak1 activities is induction of atrial fibrillation facilitated or inhibited? The approach to these questions is couched in two apparently mutually exclusive hypotheses regarding the arrhythmic mechanisms induced by re-entry or by acquired automatic activities of non-pacemaking cells; both may be related to the anti-(β)-adrenergic activity of Pak1.

The effect of Pak1 on myofilament phosphorylation is also a potential factor contributing to generation of arrhythmias. Myofilaments have the capacity to act as a “trigger” for arrhythmias because they serve as a reservoir that binds and releases Ca²⁺ during muscle contraction, which is in turn regulated by protein phosphorylation (Allen and Kentish, 1985; Allen and Kentish, 1988; Ter Keurs et al., 2006; Venetucci et al., 2008). TnC is a major site for Ca²⁺ binding to myofilament. The affinity of TnC for Ca²⁺ is regulated by factors such as protein phosphorylation, pH, and sarcomere length. Abnormal release or binding of Ca²⁺ by myofilaments may significantly disturb the Ca²⁺ transient and create conditions for arrhythmias (Ter Keurs et al., 2006). As phosphorylation at serine 23, 24 decreases the affinity of troponin C for Ca²⁺ (Robertson et al., 1982), it is likely that Pak1 may regulate Ca²⁺ exchange with myofilaments through PP2A dependent dephosphorylation of TnI at the PKA sites. Therefore, an increase of Pak1 activity may increase the buffering capacity of myofilament as the affinity of cTnC for Ca²⁺ increases.

2.6. Potential role of Pak1 in ischemia

There is little systematic investigation of the role of Pak1 in ischemia/reperfusion damage and on preconditioning of the heart (here, preconditioning refers to a protective effect of a mild ischemia preceding a more severe one). This is due in part to the controversial nature of results from Pak1 functional studies. Therefore, a potential role of Pak1 related to ischemic heart diseases can only be inferred from its potential up and downstream signaling molecules.

Pak1 activation induces a dephosphorylation of cTnI and myosin-binding protein C, resulting in an increase of the Ca²⁺ sensitivity in myofilament tension development (Ke et al., 2004). Phosphorylation of cTnI regulates cardiac conditions such as heart failure and ischemic heart diseases. In transgenic mouse hearts expressing slow skeletal cardiac troponin I (ssTnI) that is unphosphorylatable by PKA, relaxation of the heart is impaired with reduced kinetics of myofilament cross-bridge cycling. On the other hand, the systolic function is better preserved after ischemia/ reperfusion in the heart expressing the slow skeletal troponin I (Arteaga et al., 2005). The transgenic heart also maintains a healthy systolic function in left ventricle during respiratory hypercapnia (Urboniene et al., 2005), suggesting that reduced phosphorylation of cTnI and enhanced myofilament Ca-sensitivity confer heart more resistance to hypoxia.

Pak1 may be a key factor downstream of multiple protective and preconditioning signals, such as adenosine, bradykinin and acetylcholine (Table 2). Pak1 is a downstream effector of the inhibitory G protein (Gi) in mammalian cells (Knaus et al., 1995; Ueda et al., 2001). Inhibition of Gi activity completely abolishes cardiac preconditioning, suggesting that the beneficial effects of the preconditioning factors are through Gi (Schultz et al., 1998). Pak1 associates with NADPH oxidase (Ding et al., 1996; Prigmore et al., 1995) and may be responsible for the stimulatory effect of Rac1 in generation of reactive oxygen species (ROS) (Hordijk, 2006). ROS is considered as a preconditioning factor (Saini et al., 2004).

Table 2.

A list of protective and preconditioning factors that are also potential activators for Pak1.

Protective or preconditioning factors	Regulation (stimulation) of Pak1 Activities	Effect on cTnI phosphorylation	References
Adenosine	Yesa	Dephosphorylation	Cohen et al., 2000; Gupta et al., 1993; Liu and Hofmann, 2003
Acetylcholine	Yes	Dephosphorylation	Critz et al., 2005; Gupta et al., 1994; Kozma et al., 1997
Bradykinin	Yes	Dephosphorylationa	Critz et al., 2005; Kozma et al., 1995; Parratt et al., 1995
Gi	Yes	Dephosphorylation	Chen and Xia, 2000; Knaus et al., 1995; Liu and Hofmann, 2003; Ohyanagi and Iwasaki, 1996; Schultz et al., 1998; Ueda et al., 2001;
Sphingosinephosphate	Yes	Dephosphorylationa	Jin et al., 2002; Ke and Solaro, 2008; Xia, 2007
Sphingosine kinase	Unknown	Unknown	Jin et al., 2004; Pchejetski et al., 2007; Vessey et al., 2006
C2 ceramide	Yes	Dephosphorylationa	Lecour et al., 2006a; Lecour et al., 2006b; Furuya et al., 2001; Ke and Solaro, 2008
C6 ceramide	Yes	Dephosphorylationa	Ke and Solaro, 2008
ROS	Regulated by Rac1	Unknown	Oldenburg et al., 2003, Oldenburg et al., 2004; Vanden Hoek et al., 1998; Yaguchi et al., 2003

^aKe, Y., unpublished observation.

The ischemic heart also has an enhanced response to β-adrenergic stimulation which is often arrhythmogenic and lethal (Schomig et al., 1995; Zicha et al., 2006). β-adrenergic stimulation increases heart rate as well as the demand for energy supply while de novo generation of ATP is restricted due to the oxygen shortage in ischemic heart (Das and Maulik, 1996; Hjalmarson, 1998). PKA activity may also inhibit generation of reactive oxygen species (ROS) (Maj et al., 2004; Piccoli et al., 2006) that are cardioprotective (Oldenburg et al., 2004; Oldenburg et al., 2003; Vanden Hoek et al., 1998; Yaguchi et al., 2003). Paradoxically, β-adrenergic stimulation is a preconditioning factor while it also increases ischemia and reperfusion damage in the heart (Spear et al., 2007). The intracellular mechanism for these puzzling findings is unclear. Activation of PP2A by Pak1 could be a novel protective mechanism in as much as PP2A reverses PKA phosphorylation in heart and antagonizes the effects of β-adrenergic stimulation (Davare et al., 2000; Firulli et al., 2003; Jaquet et al., 1995). Therefore, a significant undertaking is the generation of strong experimental evidence to clarify the role of Pak1 during ischemia–reperfusion injury (Table 2).

3. Concluding remarks

There is ample evidence indicating that activation of Pak1 is functionally equivalent to inhibition of β-adrenergic signaling in cardiac excitation and contraction (Ke et al., 2004; Ke et al., 2007a; Lei et al., 2007; Sheehan et al., 2007) (Fig. 7, Fig. 8). It requires more investigations into Pak1 regulation of Ca²⁺ flowing “in” and “out” of sarcoplasmic reticulum (SR). It is a paradox that phosphorylation of RyR by PKA has no effect on generation of Ca²⁺ sparks (Sobie et al., 2006). On the other hand, PP2A appears to increase the frequency of Ca²⁺ sparks transiently (Terentyev et al., 2003). Sheehan et al., recently reported an inhibition of spark frequency in heart cells expressing caPak1 (Sheehan et al., 2009). A mouse model of heart specific expression of the active Pak1 may provide more definite answer to the question. It has been frustrating to nail down the physiological substrates for Paks. Without doubt, a bona fide physiological substrate of Pak1 is Pak1 itself. Auto-phosphorylation of Pak1 at multiple sites must “disturb” its partners in vivo. Both constitutively active Pak and the kinase deficient mutant sometimes produce the same phenotypes in different mammalian cells (Kiosses et al., 1999; Obermeier et al., 1998), which raises the possibility that cellular function of Pak1 can be manifested independent of phosphorylation of a downstream target. Some compounds that demonstrate negative inotropic effects in the heart may function through regulation of Pak1 activities (Ke and Solaro, 2008). Therefore, Pak1 could be the target of novel therapeutics for some cardiovascular diseases. Ventricle tachycardia (VT) triggered by catecholamine contributes substantially to sudden death in patients with end-stage heart failure (Chakko et al., 1989; Pogwizd and Bers, 2004). Modulation of Pak1 activity may provide a novel therapy for the patients. In an aging population, cardiac diseases are often complicated with other adverse conditions. It has been known that β-blockers have detrimental effects in asthma patients (Latimer and Ruffin, 1990; Nagy et al., 1989). An alternative therapy other than β-blockage will be hailed by those who suffer from both asthma and heart failure or ischemic heart diseases.

Fig. 8.

Regulation of cardiac function by Pak1 mediated signaling cascades: current view. Cardiac troponin I (cTnI), phospholamban (PLB), ryanodine receptor (RyR), L-type Ca²⁺ channel (DHPR), etc. are phosphorylated by PKA. PP2A is responsible for dephosphorylation from these substrates. PP2A is regulated by Pak1 that links to Gi activity through the small GTPases Cdc42/Rac1. GPCR—G protein coupled receptor. +P = phosphorylation; −P = dephosphorylation.

Abbreviations

AC

denylate cyclase

Ad

adenovirus

BK channel

Ca²⁺-activated K⁺ currents

B(PR55)

PP2A regulatory subunits

B′(PR61)

PP2A regulatory subunits

B″(PR72/130)

PP2A regulatory subunits

B‴(PR93/110)

PP2A regulatory subunits

cAMP

cyclic AMP

Cla4

a Saccharomyces cerevisiae Cdc42p-activated kinase

cTnI

cardiac troponin I, the inhibitory element of troponin complex

Cav1.2

a L-type Ca channel

CICR

Ca²⁺ induced Ca²⁺ release

CREB

cAMP-responsive binding protein, a transcription factor

DHPR

dihydropyridin receptor, L-type Ca channel

DI

dimerization domain of Pak1

F-actin

filamentous actin, polymerized actin

FKBP12.6

a FK binding protein, accessory protein of ryanodine receptors

GPCR

G protein coupled receptor

GEF

G protein exchange factor

Gi

the inhibitory large G protein

Gs

the stimulatory large G protein

GST

glutathione S-transferase

HERG

the human either-a-go-go-related gene or gene product, a potassium channel

Ik(ach)

acetylcholine-gated potassium current

Ik(ado)

adenosine (ado) induced muscarinic potassium current

I(ks)

slow delayed rectifier potassium current

ISO

isoproterenol

KCNQ1-KCNE1

an I(ks) channel

Kv11.1

(ERG1) K+ channels

LPA

lysophosphatidic acid

MAP kinase

mitogen activated protein kinase

MLC₂₀

myosin regulatory light chain

MLCK

myosin light chain kinase

Mst1

mammalian Ste20-like kinase

MyBP-C

myosin binding protein C

NCX

sodium/calcium exchanger

NFκB

kappa immunoglobulin enhancer-binding protein, a transcription factor

Niell5

a cell line derived from neurolastoma

Pak

p21 activated kinase

PBD

p21 binding domain

PKA

cAMP dependent protein kinase

PLB

phospholamban

PP1

protein phosphatase 1

PP2A

protein phosphatase 2A

ROS

reactive oxygen species

RyR

ryanodine receptor

SA node

sino-atrial node

Serca2

a sarco/endoplasmic reticulum Ca²⁺-ATPase isoform

SR

sarcoplasmic reticulum

ssTnI

slow skeletal troponin I

SV40

simian virus 40

3T3

a cell line derived from mouse fibroblasts

Ste20

a yeast protein kinase homologue of Cla4

Tg

transgenic

TnC

Troponin C

VT

ventricular tachyarrhythmia