Reviews |
From the Molecular Pharmacology Group (M.D.H., G.S.B.), Division of Biochemistry and Molecular Biology, Institute of Biomedical and Life Sciences, University of Glasgow, Scotland, United Kingdom; and Department of Pharmacology and Toxicology (D.H.M.), Queens University, Kingston, Ontario, Canada.
Correspondence to Prof Miles Houslay, Molecular Pharmacology Group, Division of Biochemistry and Molecular Biology, Institute of Biomedical and Life Sciences, University of Glasgow, Glasgow G12 8QQ, Scotland, UK. E-mail M.Houslay{at}bio.gla.ac.uk
This Review is part of a thematic series on Phosphodiesterases, which includes the following articles:
Compartmentation of Cyclic Nucleotide Signaling in the Heart: The Role of Cyclic Nucleotide Phosphodiesterases
Overview of PDEs and Their Regulation
Regulation of Phosphodiesterase 3 and Inducible cAMP Early Repressor in the Heart
cAMP-Specific Phosphodiesterase-4 Enzymes in the Cardiovascular System: A Molecular Toolbox for Generating
Compartmentalized cAMP Signaling
cAMP and cGMP Signaling Cross-Talk: Role of Phosphodiesterases and Implications for Cardiac Pathophysiology
PDE5 and Regulation of Vessel and Heart Function
David A. Kass Editor
| Abstract |
|---|
|
|
|---|
Key Words: phosphodiesterase-4 cAMP cardiomyocytes vascular smooth muscle cells compartmentation ß2 adrenoceptor ß-arrestin
| Introduction |
|---|
|
|
|---|
How does cAMP exert such a myriad of sophisticated actions on particular cell types? Individual cells are complex entities whose components are intricately organized in 3D space. This extends to the machinery involved in the control of cAMP levels and the generation of specific responses at discrete intracellular loci.1,2,1214
cAMP is generated by adenylyl cyclase isoforms, the majority of which are embedded in the cell surface plasma membrane.15 These are activated by transmembrane receptors, such as the ß-adrenergic receptor16 that, on agonist occupancy, couple to the stimulatory G protein Gs, thereby activating adenylyl cyclase. This confers cAMP generation to the cytosol surface of the plasma membrane, from which emanates a cloud of cAMP. However, because cells are polar, then different Gs-coupled receptors and adenylyl cyclase isoenzymes may be restricted to plasma membrane subdomains,15,17 so as to provide distinct "point sources" of cAMP generation at the plasma membrane. This offers potential for compartmentation (compartmentalization) of cAMP signaling in cells, a notion that was first mooted by Buxton and Brunton in milestone studies on cardiac myocytes18 and substantiated by others in such cells.1926 However, as the free diffusion of cAMP is rapid (130 to 700 µm2 sec1), the cell interior will quickly be equally distributed with cAMP. Furthermore, without any means of degrading cAMP then, after adenylyl cyclase activation, the cell interior would rapidly be saturated with cAMP. The ability to generate and shape cAMP gradients within the cell depends on the degradation of cAMP to 5'-AMP, which is achieved by cAMP phosphodiesterases (PDEs).27 Cytosol PDE activity would allow the formation of gradients of cAMP that depended on the source of cAMP generation by adenylyl cyclases located within subdomains of the plasma membrane. This adenylyl cyclasePDE-dependent formation limits the potential for shaping gradients of cAMP in cells, thereby channeling cAMP signaling along specific conduits. To achieve this process, a further level of sophistication is engineered into the system, namely the ability to spatially restrict cAMP gradients at specific intracellular sites by targeting PDEs to specific intracellular sites and signaling complexes within cells.12,14,2830
Tethering of PDEs allows these enzymes to form and shape localized cAMP gradients that can be visualized with genetically encoded sensors.20,24,31,32 Up until recently the general notion was that tethered PDEs provided a barrier to free diffusion of cAMP from restricted microenvironments surrounding the site of generation at the plasma membrane. However, it is clear that such a model imposes severe limitations on the degree of control that can be exerted, not the least of which, that it confines spatial control to two compartments, namely those defined as being either inside or outside the PDE "barrier." Recent experimental evidence, using siRNA knockdown of cAMP-specific PDE4, has suggested an additional scenario.31 This scenario envisages spatially confined populations of tethered PDEs generating localized "sinks" or "black holes," down which cAMP "disappears" as it is converted into 5'-AMP (Figure 1). Spatially constrained PDE subpopulations coupled with free diffusion of cAMP will allow a myriad of localized gradients of cAMP to be generated and shaped in cells. Such a sophisticated system provides a means of generating a multitude of microenvironments in the cell interior that are under precise control of specific, tethered PDE subpopulations. Each of these PDE subpopulations can be envisaged as regulating distinct cAMP-controlled processes through either protein kinase A (PKA) and its associated substrates or EPAC (exchange protein activated by cAMP) and its associated Rap1/2 effectors, with specific phenotypes therefore being associated with displacing specific PDEs from individual locales. Thus targeted PDEs are fundamental to the generation and control of compartmentalized cAMP signaling processes. These findings clearly underpin the complex nature of cAMP actions in the heart, where there are a plethora of distinct cAMP-regulated systems requiring distinct regulation of cAMP inputs that appear to be regulated by distinct PDEs. Such a system offers potential for sophisticated regulation by manipulating PDE tethering. Indeed, the cell typespecific expression of components of the proteins that form the cAMP signaling toolbox allows specific tailoring of the compartmentation of cAMP signaling. Additionally, it can be envisaged that alterations in either PDE isoform profile or tethering mode could contribute to certain pathologies.
|
There are 11 different PDE families, of which 8 encode a plethora of isoenzymes able to degrade cAMP.27 The use of selective inhibitors,27 small interfering RNA (siRNA)-mediated gene knockdown,33 dominant negative constructs,3335 and targeted gene knockouts3641 has identified nonredundant, functional roles for an increasing number of PDEs.
Here we focus on the insight that investigation of the PDE4 family has provided into cAMP signal compartmentation, with special emphasis on cells of the cardiovascular system. Indeed, analysis of PDE4 enzymes has provided the paradigm for intracellular targeting of cAMP degradation30 and highlights the fundamental role that individual PDE4 isoforms are poised to play in tailoring compartmentalized cAMP signaling. This, undoubtedly, is a key reason why the complex 4 gene PDE4 family has been highly conserved through evolution.
| "PDE4-ology" |
|---|
|
|
|---|
PDE4A is located at Chr19p13.2, PDE4B at Chr1p31, PDE4C at Chr19p13.1, and PDE4D at Chr5q12. These genes span approximately 50 kb and comprise approximately 20 exons, the core catalytic unit of which is encoded by 7 exons.4446 Additional exons encode regulatory regions and the N-terminal regions that uniquely identify individual isoforms. Various studies have linked the PDE4D gene to stroke47 and to changes in bone mineral density48; other studies have linked the PDE4B gene to schizophrenia.49
Unique to the PDE4 family are the highly conserved regulatory regions upstream conserved region 1 (UCR1) and upstream conserved region 2 (UCR2).50 Each is encoded by 3 separate exons, with UCR1 being formed from some 55 amino acids and UCR2 being formed by some 76 amino acids. PDE4 isoforms are subcategorized into 4 groups based on their UCR1/UCR2 complement. Thus "long" isoforms have UCR1 and UCR2, "short" isoforms lack UCR1, and "super-short" isoforms have just a truncated UCR2, whereas "dead-short" isoforms lack UCR1 and UCR2 and have an inactive catalytic unit that is both N- and C-terminally truncated (Figure 2).51,52 All PDE4 genes encode long isoforms, although only certain PDE4 genes encode types of short forms (Figure 2).
|
UCR1 and UCR2 likely interact in PDE4 long isoforms through ionic interactions53 to form a regulatory module that directs the functional outcome of phosphorylation by PKA and extracellular signal-regulated kinase (ERK).5461
UCR1 is joined to UCR2 by LR1, a region that is encoded by 2 exons, is approximately 22/24 amino acids in length, and shows profound heterogeneity between subfamilies. LR2, which joins UCR2 to the catalytic unit, is encoded by a single exon and, as with LR1, shows no similarity between PDE4 subfamilies and varies in length between 10 and 28 aa. Their functional significance remains to be ascertained.
The final exon encodes part of the core catalytic unit as well as the C-terminal tail unique to each PDE4 subfamily. Indeed, the difference in primary sequence of this region has been exploited by us to make antisera specific to each PDE4 subfamily.
The most 5' isoform for each PDE4 gene, seemingly, has dual 5' exons encoding its unique N-terminal region, whereas other isoforms have a single 5' exon encoding their unique N-terminal region. Specific promoters found immediately 5' to the N-terminal coding exon control the expression of individual isoforms.6265 Such minimal PDE4 promoters appear to lack a canonical TATA box but contain CpG-rich islands and a series of perfect stimulating protein 1 (Sp1) consensus binding sites that drive basal promoter activity. Undoubtedly, regions 5' to this confer cell typespecific expression and further regulation.
It has also been shown that an alteration in the degree of histone acetylation of the PDE4D1/2 intronic promoter regulates the extent to which these variants are expressed in VSMCs.66 Histone acetylation is among the numerous epigenetic factors that control expression of many genes,67 and it will be interesting to determine whether other PDE4 isoforms are similarly regulated. Additional control of PDE4 isoform expression can occur through regulation of mRNA stability by cross-talk with the ERK pathway,68 the action of which has been implicated in cardiac hypertrophy.69
At a genomic level, the sequence of coding exons is highly conserved among species, indicating that the complexity of organization and plethora of PDE4 isoforms must provide a functional advantage to have survived evolutionary pressures in such an intact state. If changes in the PDE4 isoform and tethering-protein profiles change in pathological states, then this is likely to have a profound effect of compartmentation of cAMP signaling.
| The Long, Short, and Super-Short of PDE4 Isoforms |
|---|
|
|
|---|
The mitogen-activated protein kinase ERK regulates numerous aspects of cardiomyocyte and VSMC functions, both in health and disease, including their hypertrophic responses.69,7173 The catalytic unit of all PDE4 subfamilies, save for PDE4A, contains a serine within a consensus site (PQSP) that allows phosphorylation by ERK in vivo, altering activity54,55,57,61 and expression.68 As with all authentic ERK substrates, the catalytic unit of PDE4 enzymes contains both a KIM docking site (VxxKKxxxxxxLL), located on an exposed ß-hairpin loop some 122 amino acids N-terminal to the target serine, and an ERK specificity motif (FQF), located on an exposed
-helix some 18 amino acids C-terminal to the target serine.57 It is the presence of absence of UCR1/UCR2 that determines the functional outcome of ERK phosphorylation of PDE4, with long isoforms being inhibited, short isoforms being activated, and super-short isoforms being weakly inhibited. Thus cAMP signaling can be either positively or negatively coupled to ERK activation in specific intracellular locales dependent on the complement of short and long isoforms expressed. Such cross-talk can be reprogrammed by changes in the PDE4 isoform expression profile as seen in monocyte to macrophage differentiation,74 and it will be of interest to see whether changes in cross-talk occurs in VSMC differentiation, where the PDE4 long/short profile changes.66
Interestingly, ERK inhibition of PDE4 long isoforms can be negated by PKA phosphorylation.55 This can lead to a situation where ERK-induced PDE4 inhibition can raise cAMP levels, causing PKA to become activated and phosphorylate the long PDE4, thereby ablating the inhibitory effect of ERK phosphorylation. Thus, as a consequence of ERK activation, long PDE4 isoforms may cycle through inhibition followed by activation, thereby causing either a transient, programmed rise in cAMP levels in their immediate locale or even oscillations.
More recently it has been demonstrated61 that the N-terminal portion of the PDE catalytic unit (Ser239 in PDE4D3) can be phosphorylated by an unknown kinase that acts downstream of phosphatidylinositol 3-kinase and is activated by oxidative stress. Phosphorylation at this site alone in PDE4D3 has no effect on catalytic activity. However, oxidative stress also activates ERK and it is when PDE4D3 is phosphorylated both by ERK (Ser579) and the unknown kinase (Ser239) that the function of this kinase is uncovered as reprogramming the effect of inhibitory ERK phosphorylation to now cause activation. Indeed, this now mimics the "loss of UCR1," seen in short isoforms, where ERK phosphorylation of the PDE4 catalytic unit confers activation.57
This unknown kinase is activated by reactive oxygen (ROS), and it may be linked to stress-induced reprofiling evident in cardiovascular disease. Clearly it will be important to identify it.
| The Motor in the Middle |
|---|
|
|
|---|
-helices folded into 3 subdomains. These subdomains come together to form a deep pocket containing the cAMP binding active site, which contains tightly bound Zn2+ and loosely bound Mg2+, essential for catalytic activity. This pocket has a volume of 440 Å3, which contains the 232 Å3 cAMP molecule. PDE4 activation by PKA phosphorylation is influenced by [Mg2+] and the dominant "connections" that hold Mg2+ and are links to amino acids on helices 10/11. These connections, together with their connecting loop, fold over the surface of the catalytic center so as to create a "tweezer-like" motif that grips the Mg2+. It is possible that UCR/UCR2 may direct actions to Mg2+ at the catalytic center via helices 10/11 or others that either interact directly with them or indirectly cause conformational changes in them. Analyses of the PDE4 catalytic unit structure, proposed catalytic mechanism, binding of selective inhibitors and "inside-out" signaling where inhibitor binding might transmit changes to the molecule surface are discussed elsewhere.28,43,76
| Finding the Perfect Partner and Identifying Targeting "Zip Codes" |
|---|
|
|
|---|
The paradigm for this notion came from studies on the PDE4A1 super-short isoform.77 Uniquely, PDE4A1 is exclusively membrane-associated and requires detergents to effect its release. PDE4A1 is uniquely characterized by its 25-aa N-terminal region, the removal of which generates a soluble, fully active species.7782 Thus all of the information essential for membrane targeting is held within its unique N-terminal region. Consistent with this, chimeric species, made with various soluble, cytosolic proteins transformed them to membrane-bound species that localized within cells as did PDE4A1.
A key feature of individual PDE4 isoforms is their ability to be targeted to specific sites/signaling complexes within cells, leading to the notion of a "PDE4 interactome" (Figure 3).
|
The N-terminal region of the long PDE4D5 isoform contains binding sites for the signaling scaffold proteins ß-arrestin33,34,83,84 and RACK1,8587 which we discuss in detail below.
The N-terminal regions of the long PDE4A4/5 and PDE4D4 isoforms contain proline-rich sequences that confer interaction with SH3-domains of certain proteins, such as the tyrosyl kinases Lyn, Fyn, and Src.8891 Differences in specificity of interaction are seen between PDE4A4/5 and PDE4D4 because of different sequences surrounding their distinct proline- and arginine-rich sequences.
PDE4A4/5 can bind to the immunophilin AIP/XAP2/ARA9, which is known to interact with the aryl hydrocarbon receptor (AHR), a transcription factor required for normal cardiac development92; AHR is upregulated in cardiomyopathy93; and the genetic deletion of AHR leads to cardiac hypertrophy, hypertension, and fibrosis.94 Aryl hydrocarbon receptor interacting protein (AIP) interacts not only with the PDE4A4/5 N-terminal region to give isoform specificity but also interacts with UCR2 to elicit an inhibitory effect on PDE4A4/5 activity.95 This paradigm is a clear indication of how proteinprotein interaction may regulate PDE4 catalytic activity. However, many PDE4-interacting proteins, such as RACK1, ß-arrestin, and SH3 domaincontaining proteins do not exert any profound effect on catalytic activity.
Interestingly, other proteins have been shown to interact with UCR2, namely the scaffold proteins, myomegalin,96 myeloid translocation gene protein,97 and DISC1.49 However, it remains to be seen whether they interact with additional sites on PDE4 and whether they alter PDE4 activity.
The long PDE4D3 isoform interacts with the PKA anchor protein mAKAP,98100 which is induced in cardiac hypertrophy101 and serves to relocate PDE4D3 to the perinuclear region of hypertrophic cardiac myocytes.99 PKA phosphorylation of Ser13, within the unique N-terminal region of PDE4D3, increases interaction with mAKAP.98 The activities of mAKAP-associated PDE4D3 and PKA are intertwined with PDE4D3 being phosphorylated at 2 sites by PKA,56,5860 namely Ser54 in UCR1, causing activation, and Ser13, causing increased binding to mAKAP.
Mapping sites of interaction between proteins has, traditionally, been an arduous process involving truncation and mutation approaches. However, we have recently pioneered peptide array technology as a rapid means of ascertaining interaction sites.85 Here a recombinant interactor protein is used to probe a library of overlapping, immobilized 25-mer peptides that scan the entire sequence of a particular protein. This allows rapid identification of regions that may contribute to the proteinprotein binding. An interacting peptide was used as a template to generate a library of progeny where individual amino acids in the 25-mer parent are replaced by alanine, for example, in a corollary to scanning mutagenesis. Thus amino acids of putative importance to binding can be identified and used to direct mutagenesis approaches using intact proteins in 2-hybrid, pull-down, coimmunoprecipitation, and colocalization approaches. Where structural information is also available, this can additionally be used to identify surface residues that likely form a binding site, thereby further facilitating mutagenesis strategies.
| Cellular Function and Phenotype Conferred by PDE4D5 Association With ß-Arrestin |
|---|
|
|
|---|
|
|
In various cell types including cardiomyocytes PKA phosphorylation of the ß2AR partially attenuates coupling to Gs and switches on coupling of the phosphorylated ß2AR to the guanine nucleotide regulatory protein, Gi (Figure 5).12,103 Through this process, the PKA phosphorylated ß2AR elicits the SRC-dependent activation of ERK.
Recruitment of the ß-arrestinPDE4 complex serves to orchestrate a dual desensitization event. Thus the recruited ß-arrestin physically interdicts signaling between receptor and Gs, initiating a reduction in adenylyl cyclase activation and a subsequent decrease in cAMP production. However, simultaneously, ß-arrestinbound PDE4D5 provides a localized sink for cAMP degradation, which acts to downregulate ß2AR associated PKA activity and, thereby, dampen signaling to ERK through Gi. Such a regulatory function has been demonstrated in neonatal cardiomyocytes as well as model cell lines.12,34,83,103,104
It has also been demonstrated that cAMP pools generated by stimulation of different Gs coupled receptors in rat ventricular myocytes are shaped by the differential coupling of each receptor type to different PDE families.19 However, the extent to which G proteincoupled receptorspecific cAMP "pools" are influenced by recruited PDE4D5 in complex with ß-arrestin remains to be ascertained.
In cardiac myocytes, PDE4D5 is preferentially associated with ß-arrestin and selectively recruited to the ß2AR on agonist challenge, despite the fact that PDE4D5 expression was some 5 times lower than that of PDE4D3.34,83 As discussed below, PDE4D5 interacts preferentially with ß-arrestin because of an additional binding site unique to this isoform.83,85
In cardiac myocytes, chemical ablation of PDE4 activity by the specific inhibitor, rolipram enhances both PKA phosphorylation of the ß2AR and the switching of its signaling to ERK activation.34,83 However, rolipram inhibits all PDE4 isoforms similarly and so cannot identify control by any one PDE4 isoform. That selective silencing of all PDE4D isoforms by siRNA-mediated knockdown mimicked such actions of rolipram33 identifies the importance of this subfamily in modulating ß2AR signaling but gives no insight into which particular isoform and whether targeting is required. Isoform-specific, siRNA-mediated knockdown subsequently identified PDE4D5 as the relevant species.33 However, although this technological approach indicates the role of PDE4D5, it gives no insight into whether the entire cellular PDE4D5 pool is of importance or whether a subpopulation is important, namely one that is tethered specifically to ß-arrestin. Such an analysis demanded a new technological approach, and for this we used overexpression of a catalytically inactive PDE4D5 (Asp556Ala), which was introduced into cardiac myocytes by adenoviral-mediated gene delivery.34 This catalytically inactive PDE4D5, when overexpressed, serves to displace endogenous PDE4D5 from ß-arrestin and prevent agonist-mediated delivery of active PDE4D5 to the ß2AR, providing a dominant negative action.33,34,85 Dominant negative PDE4D5 amplifies PKA activity at the plasma membrane but not in the cytoplasm. This mimics the phenotype engendered by treatment with either rolipram or PDE4D5 knockdown. Final verification that the PDE4D5 phenotype resulted from its preferential association with ß-arrestin resulted from the demonstration that a discrete mutation in the N-terminus of catalytically inactive PDE4D5 construct, made so as to compromise its ability to bind ß-arrestin; prevented such a species from displacing endogenous active PDE4D5 from ß-arrestin; and failed to elicit a dominant negative effect.85 This dominant negative approach, undertaken on cardiac myocytes, provided the first indication that a cellular phenotype could be assigned to an individual PDE isoform.
Clearly, a dominant negative strategy provides a means of dissecting out functional roles for anchored subpopulations of PDE4 isoforms that cannot be determined using either active site-directed inhibitors or siRNA knockdown. The identification of small molecules that disrupt targeting of specifically anchored PDE4 isoforms may provide for novel therapeutic agents that are not plagued by the various side effects seen with active site-directed PDE4 inhibitors. Such targeting disruptors can be expected to allow diminution of PDE4 activity at a highly specific spatial locale.
| Molecular Determinants Mediating PDE4D5ß-Arrestin Interaction |
|---|
|
|
|---|
PDE4D5 can also form a complex with the WD repeat scaffold protein RACK1. This interaction is unique to PDE4D5 because, as with ß-arrestin, it involves a binding site located in the PDE4D5-unique N-terminal region.86,87 Indeed, overlapping binding sites for both ß-arrestin and RACK1 in the unique N-terminal region of PDE4D5 coupled to distinct second sites of interaction on the catalytic unit confers their mutually exclusive binding to PDE4D5.85 Thus ß-arrestin and RACK1 independently sequester PDE4D5, ensuring fidelity of signaling through these distinct scaffold proteins.
Insight into the location and nature of the binding sites for ß-arrestin and RACK1 on PDE4D5 was garnered using peptide array technology (Figure 6).85 This allowed us to demonstrate that both RACK1 and ß-arrestin to PDE4D5 could bind to amino acids between residues 22 and 45 within the N-terminal portion of PDE4D5. Using alanine substitution arrays to scan this region, specific amino acids were identified as involved in determining the binding of either ß-arrestin (E27, D28, L29) or RACK1 (N22, P23, W24, V30, K31), exclusively, or were found in common as important for the binding of each of these signaling scaffold proteins (L33, R34). Analysis of the 3D structure of this portion of the PDE4D5 N-terminal region showed these residues to be surface exposed and that the concomitant binding of both ß-arrestin and RACK1 to PDE4D5 was not possibly attributable to the proximal and overlapping nature of their respective binding sites. Thus in any one cell, there are likely to be specific, spatially distinct subpopulations of PDE4D5 because of the association of PDE4D5 with scaffolding proteins such as ß-arrestin and RACK1.
|
Peptide array analysis also facilitated resolution of the amino acids that contribute to the common ß-arrestin binding region within the conserved catalytic unit of all PDE4s.85 The amino acids identified (F670, F672, L674, and L676) are all surface exposed and located on helix-17, which appears to be attached to the compared core catalytic unit by a mobile hinge region (Figure 6).85 Interestingly, 2 of the amino acids implicated (F670, F672) in ß-arrestin binding also form part of the ERK specificity/docking binding motif on PDE4 enzymes,57 which would preclude PDE4 isoforms from binding directly to both ß-arrestin and ERK. As ERK can phosphorylate and deactivate PDE4 long forms, it is tempting to speculate that by preventing ERK docking to PDE4D5, ß-arrestinbound PDE4D5 ensures only activated enzyme is found in this complex and so recruited to the ß2AR on agonist activation.
This work highlights the power of peptide array technology for the rapid and informative definition of proteinprotein interactions.
| Cardiomyocyte PDE4s |
|---|
|
|
|---|
Although several studies have described altered cardiomyocyte PDE3A expression in certain types of heart disease,111,112 no studies have systematically addressed whether the PDE4 expression profile is altered in heart failure. Given the distinct spatial and temporal regulation of cAMP levels afforded by distinct PDE4 subtypes, their altered expression in heart disease would be predicted to, potentially, markedly alter cardiac function. In contrast, significant advances have been made relative to the manner by which these enzymes allow compartmentalized cAMP signaling in cardiomyocytes. Recent elegant molecular studies have determined that individual PDE4 variants can be tethered to distinct sarcoplasmic reticulum (SR) proteins, allowing them to either directly or indirectly control SR function.12,19,20,22,25,31,34,38,108,113 Indeed, tethering of PDE4 variants to specific signaling complexes likely represents the molecular basis for much of the selective actions seen with PDE4 inhibitors, compared with PDE2 and PDE3-selective inhibitors in cardiomyocytes. Moreover, it is possible that loss of tethering for certain PDE4D variants may lead to cardiac failure, as seen in PDE4D-null mice and, perhaps, in humans (see below).38
PDE4B and PDE4D localize to sarcomeric M- and Z-line structures, respectively, in neonatal rat ventricular cardiomyocytes. Consistent with this, in fully differentiated adult cardiomyocytes, PDE4 activity is high at the transverse (T) tubule/SR junctional space of cardiomyocytes, the area regulating excitationcontraction coupling.25,114 Notwithstanding that cardiomyocytes express several long PDE4D isoforms (PDE4D3, PDE4D5, PDE4D8, PDE4D9), to date most studies have limited their analysis to how PDE4D3, and PDE4D5 tethering contributes to compartmented cAMP signaling in cardiomyocytes. Thus PDE4D3 associates with the ryanodine receptor 2 (RyR2)38 and A-kinase anchoring proteins (AKAPs)99,115,116 in cardiomyocytes, so as to spatially and temporally regulate cAMP. PDE4D5 is the key isoform interacting with ß-arrestin,33,34,83,85 which allows it to preferentially regulate ß2AR signaling12,34 but can also interact with the signaling scaffold protein, RACK1.8587,117,118
| PDE4DRyR2 Interaction |
|---|
|
|
|---|
12-kDa FK506-binding protein FKBP12.6. In addition to FKBP12.6, several other proteins interact with the RyR2 within a large macromolecular signaling complex in cardiomyocytes, which may include PKA, CaMKII, PP1, PP2A, mAKAP, spinophilin, PR130, sorcin, triadin, junctin, calsequestrin, and Homer.123
Until very recently, analyses of PDE4D- and PDE4B-null mice had revealed no significant pathological role for these enzymes. Indeed, most of the attention had focused on either PDE3B within a phosphatidylinositol 3-kinase
complex or on PDE3A, the more abundant isoform expressed in cardiomyocytes.124,125 However, recently, PDE4D-null mice were reported38 to display a very late, age-dependent, cardiac phenotype composed of a progressive cardiomyopathy and an increased incidence of exercise-induced arrhythmias. This phenotype is similar to that reported when a RyR2 defect in patients produces heart failure and sudden cardiac death.126 At a functional level, the PDE4D-null mouse phenotype was associated with RyR2 hyperphosphorylation and a reduced capacity of hyperphosphorylated RyR2 to gate Ca2+. It was suggested that hyperphosphorylation of RyR2 was attributable to the absence of PDE4D within the RyR2-based macromolecular complex of PDE4D-null animals.38,127 The authors suggested that, because the phenotype was suppressed in mice engineered to lack one of the potential PKA phosphorylation sites within RyR2 (S2808) and because less PDE4D3 was associated with RyR2 in human myocardium from heart failure patients, their findings were consistent with the hypothesis that PDE4D deficiency may contribute to heart failure and arrhythmias by promoting defective regulation of the RyR2 channel in humans. In addition, the authors also speculated that prolonged PDE4 inhibitor use might predispose patients to unexpected cardiac events. In this context, because no fewer than 70 individual mutations that alter the biophysical properties of the RyR2 have been described, because phosphorylation of the RyR2 is catalyzed by numerous kinases in addition to PKA, and because PDE4D activity and targeting are also each dynamically regulated by multiple factors, it is likely that further work in human cardiomyocytes will be required to fully assess the significance of the effect in humans. Additional work is needed to explore this proposal in studies using animals with different genetic backgrounds and, importantly, with conditional knockouts, so as to exclude any phenotypic input resulting from loss of PDE4D during development. Also, because the PDE4D knockout was generated by deletion of a catalytic exon such an approach might lead to the generation of truncated proteins that could interact with signaling complexes to exert dominant negative actions independent of loss of PDE4 activity. In this context, it should be noted that no significant cardiac side effects have been reported in clinical trials of PDE4 inhibitors designed for use in treating chronic obstructive pulmonary disease or asthma in humans, and no cardiac toxicology has been reported in animal studies performed using various PDE4-selective inhibitors in development.43,128130 Thus it is important to extend studies of PDE4 action in the heart to understand fully its role in health and disease.
| PDE4D3mAKAP |
|---|
|
|
|---|
Although little is known concerning the number of PDE4 variants that interact with cardiomyocyte AKAPs, recent studies have identified some of the proteins that also populate the mAKAP signaling complex. Indeed, the EPAC, a cAMP-activated Rap-GEF,4 as well as ERK5, may be found together in certain complexes.100 Interestingly, the presence of both PKA and EPAC within such a complex may represent a situation in which local cAMP concentrations can differentially control the activity of 2 cAMP effectors with distinct sensitivities to activation by cAMP. Indeed, whereas cAMP activation of mAKAP-associated PKA served to phosphorylate and thereby activate PDE4D3, reducing local cAMP concentrations, maximal activation of mAKAP-associated ERK5 served to suppress PDE4D3 activity, thus allowing for activation of EPAC. The mechanism by which ERK-mediated phosphorylation inhibits PDE4D3 involves phosphorylation of this long isoform at Ser579.54,55,57
Although it has been suggested that PDE4D3 represents the adaptor protein that recruits EPAC1 to the mAKAP complex,100 this has yet to be established. Indeed, PDE4D3 interacts with mAKAP via its unique 18-aa N-terminal domain.99 Thus, unless EPAC also interacts with this small PDE4D3-specific domain, which seems unlikely as it might then be expected to compete with mAKAP for binding to PDE4D3, we would expect that EPAC binds to a different site. Indeed, this appears to be the case as we (M. Houslay, H. Bos, M. Lynch, G. Baillie, unpublished results, 2006) can show that there is a common binding site on PDE4 isoforms for EPAC. Thus various other PDE4 isoforms may also be able to recruit EPAC to their site of anchorage within the cell. If multiple PDE4 isoforms were able to recruit EPAC to such signaling complexes, this would further increase the need to determine whether the PDE4 isoform expression profile is impacted in heart failure.
| Phenotypic Modulation of VSMCs |
|---|
|
|
|---|