Albert Smolenski, UCD Conway Institute, UCD School of Medicine and Medical Science, University College Dublin, Belfield, Dublin 4, Ireland. Tel.: +353 1 716 6746; fax: +353 1 716 6701. E-mail: firstname.lastname@example.org
Summary. Endothelial prostacyclin and nitric oxide potently inhibit platelet functions. Prostacyclin and nitric oxide actions are mediated by platelet adenylyl and guanylyl cyclases, which synthesize cyclic AMP (cAMP) and cyclic GMP (cGMP), respectively. Cyclic nucleotides stimulate cAMP-dependent protein kinase (protein kinase A [PKA]I and PKAII) and cGMP-dependent protein kinase (protein kinase G [PKG]I) to phosphorylate a broad panel of substrate proteins. Substrate phosphorylation results in the inactivation of small G-proteins of the Ras and Rho families, inhibition of the release of Ca2+ from intracellular stores, and modulation of actin cytoskeleton dynamics. Thus, PKA/PKG substrates translate prostacyclin and nitric oxide signals into a block of platelet adhesion, granule release, and aggregation. cAMP and cGMP are degraded by phosphodiesterases, which might restrict signaling to specific subcellular compartments. An emerging principle of cyclic nucleotide signaling in platelets is the high degree of interconnection between activating and cAMP/cGMP-dependent inhibitory signaling pathways at all levels, including cAMP/cGMP synthesis and breakdown, and PKA/PKG-mediated substrate phosphorylation. Furthermore, defects in cAMP/cGMP pathways might contribute to platelet hyperreactivity in cardiovascular disease. This article focuses on recent insights into the regulation of the cAMP/cGMP signaling network and on new targets of PKA and PKG in platelets.
Cyclic AMP (cAMP) has been known as powerful inhibitor of platelet aggregation since the 1960s. Initial studies showed inhibitory effects of prostaglandin E1 on platelet activation , which were soon linked to cAMP . Later, endothelium-derived prostacyclin was discovered as the main physiologic stimulator of cAMP production in platelets . Nitric oxide (NO) and cyclic GMP (cGMP) were observed to inhibit platelet functions around 1980 , and endothelial release of NO was linked to cGMP-dependent platelet inhibition . Defects in platelet cyclic nucleotide signaling might play a role in common diseases such as ischemic heart disease, heart failure, and diabetes, where reduced sensitivity of platelets to the inhibitory effects of NO contributes to platelet hyperreactivity . In addition, a number of rare genetic abnormalities in platelet cyclic nucleotide signaling have been described. Defects in prostacyclin signaling reduce cAMP levels, resulting in hyperreactive platelets and a prothrombotic state . Hypersensitivity of prostacyclin signaling caused by gain of function mutations in a Gs protein results in elevated cAMP levels and induces a bleeding phenotype, presumably via exaggerated inhibition of platelet functions . cAMP-elevating and/or cGMP-elevating agents have shown clinical benefit as platelet inhibitors. For example, dipyridamole in combination with low-dose aspirin is an approved therapy for stroke prevention . Dipyridamole elevates cAMP levels in platelets by several mechanisms, including inhibition of phosphodiesterase (PDE)-mediated breakdown. Cilostazol is a specific inhibitor of PDE3 that has been Food and Drug Administration-approved for the treatment of intermittent claudication, and that appears to be effective in reducing the risk of restenosis after coronary angioplasty . Cilostazol has also been used successfully for the secondary prevention of ischemic stroke . Recently, a group of compounds that activate cGMP production by soluble guanylyl cyclase have been shown to reduce thrombus formation in animal models . The exact mechanisms involved in cAMP/cGMP-mediated platelet inhibition and the wiring of the cyclic nucleotide signaling network are only partly understood. cAMP and cGMP are able to block many aspects of platelet activation, including early activatory signals such as release of Ca2+ from intracellular stores and G-protein activation, and adhesion, granule release, aggregation, and apoptosis [11,12]. In general, platelet inhibition by cyclic nucleotides does not appear to be restricted to any particular activatory pathway. cAMP and cGMP block platelet activation mediated by ligands of G-protein-coupled receptors, such as thrombin, ADP, and thromboxane, and by collagen, von Willebrand factor (VWF), and fibrinogen. This article will provide an overview of the various components of the cAMP/cGMP signaling network in platelets, including novel insights into mechanisms of regulation and new target molecules.
Regulation of cAMP synthesis
Levels of free cytosolic cAMP are controlled by their synthesis through adenylyl cyclases (ACs) (Fig. 1). Platelet proteome analysis suggests that platelets express a number of different membrane-bound AC isoforms, including AC3, AC6, and AC7 . ACs are activated by G-protein-coupled receptor signaling. Binding of prostacyclin (prostaglandin I2) to its receptor on the platelet surface (IP receptor)  triggers the activation of intracellular receptor-linked stimulatory G-protein α-s (Gs) subunits. Gs is turned into its active GTP-bound form, Gs–GTP, and then binds to AC and stimulates the synthesis of cAMP from ATP. Gs-mediated activation of AC is turned off by regulator of G-protein signaling 2, a specific GTPase-activating protein that helps to hydrolyze Gs–GTP back to inactive Gs–GDP . Other receptors linked to Gs-mediated AC stimulation in platelets include A2A and A2B receptors for adenosine [16,17] and VPAC1 receptors for pituitary AC-activating and vasoactive intestinal peptides . Platelet activators such as ADP and thrombin block AC function through inhibitory Gα-i (Gi) proteins, resulting in a drop in cAMP levels during platelet activation [19,20]. The specific roles of individual AC isoforms and their possible regulation by other G-protein-independent factors have not been studied in platelets to date.
Regulation of cGMP synthesis
cGMP production in platelets depends on a single enzyme, the soluble NO-sensitive guanylyl cyclase (sGC or NO-GC), which is composed of two subunits, α1 and β1 . NO originates from endothelial cells containing endothelial NO synthase (eNOS), and thorough analyses of mouse and human platelets have appeared to rule out the presence of any significant NO synthase enzyme within platelets [22–24]. Recent data from eNOS knockout animals indicate that other, as yet unidentified, sources of NO might play a role in platelet inhibition . NO permeates the plasma membrane of platelets and activates cytosolic sGC to generate cGMP (Fig. 1), resulting in an approximately 10-fold increase in cGMP levels. A number of recent studies of sGC-deficient platelets have demonstrated that most effects of NO in platelets are indeed mediated by sGC and cGMP, at least in mice. Absence of sGC abolished cGMP synthesis, resulting in an almost complete loss of NO effects, both in a constitutive sGC-deficient mouse model and in platelet-specific sGC-deficient mice [11,26,27]. Using very high concentrations of the NO donor sodium nitroprusside (above 100 μm), one group observed residual inhibition of collagen-induced platelet aggregation and ATP release in isolated sGC-deficient platelets, which was attributed to cGMP-independent NO-mediated protein nitrosylation . However, in a different study, neither 800 μm sodium nitroprusside nor high concentrations of other NO donors affected collagen-induced aggregation of isolated sGC-deficient platelets . sGC function can be activated by VWF, thrombin, or collagen (Fig. 1). A small (mostly about two-fold) increase in cGMP levels after exposure of platelets to VWF, calcium ionophore, H2O2 [22,28,29] thrombin or collagen [27,30,31] has been described, although these findings are not supported by all studies [23,32,33]. It should be noted that current methods for determining intracellular cyclic nucleotide levels in whole cell lysates are limited by the rapid PDE-mediated breakdown and the potential subcellular compartmentalization of cAMP and cGMP (see below). Alternative pathways for sGC activation might involve phosphorylation. The VWF-induced increase in cGMP levels is absent in mice lacking either the serine/threonine kinases Akt1 and Akt2 or the tyrosine kinase Lyn [34,35]. Tyrosine phosphorylation of the α1 and β1 subunits of sGC has been observed in VWF-treated and collagen-treated platelets [22,36]. The physiologic role of these alternative and possibly NO-independent pathways of sGC activation remains to be determined. sGC was recently suggested to be involved in platelet activation in a platelet-specific sGC-deficient mouse model. Aggregation and ATP release of isolated sGC-deficient platelets were reduced at low concentrations of collagen and thrombin, and tail bleeding times and thrombus formation were slightly increased in these animals . However, in constitutive sGC-deficient mice, strongly reduced bleeding times were described . Thus, a potential role for sGC in platelet activation will need to be clarified by further studies. Recently, platelet-specific sGC-deficient mice were used to establish a new role for NO and cGMP in the inhibition of platelet apoptosis . Defective platelet sGC function has been described in patients with ischemic heart disease, heart failure, and diabetes . Studies of patients with obesity suggest that not only might the function of platelet sGC be disturbed, but additional defects in cAMP synthesis and in downstream cAMP and cGMP targets might contribute to platelet hyperreactivity in cardiovascular and metabolic disease . In these patients, impaired NO-mediated inhibition of platelet aggregation has been linked to reduced levels of cGMP synthesis in the presence of oxidative stress . sGC is known to be inactivated by oxidation, which might eventually lead to the loss of the NO-binding heme group from the enzyme. Oxidized or heme-free sGC has been shown to play an important role in the pathogenesis of cardiovascular disease, and new compounds have been developed that are able to activate heme-free sGC . In addition, stimulators of sGC are available that act in synergy with NO to increase sGC activity. The sGC stimulator BAY 41-2272 is a potent new platelet inhibitor , although some of its effects might be mediated by blocking of PDE5-mediated degradation of cGMP .
PDE-mediated degradation of cyclic nucleotides
Platelet cyclic nucleotide levels are controlled by PDE-mediated degradation providing a negative feedback loop for cyclic nucleotide signaling. Platelets have been shown to express PDE2A, PDE3A, and PDE5A . PDE2 and PDE3 are able to degrade cAMP and cGMP; however, inhibitor studies have suggested that PDE2 and PDE3 mainly regulate cAMP in platelets [41,42], whereas PDE5 specifically degrades cGMP. cGMP has a regulatory effect on all three platelet PDEs (Fig. 1). cGMP stimulates PDE2 and PDE5 activity by binding to specific cyclic nucleotide-binding domains, called GAF domains. cGMP inhibits PDE3 activity by competing with cAMP binding at the catalytic site. PDE3A helps to maintain low basal levels of cAMP in platelets [41–43]. PDE3A activity is upregulated by about 50% during thrombin activation involving protein kinase C (PKC)-mediated and possibly also protein kinase B-mediated phosphorylation [44,45]. PKC-mediated phosphorylation of PDE3A on Ser428 results in binding of 14-3-3, although this does not appear to be required for increased activity. PDE3A is also activated by cAMP-dependent protein kinase (protein kinase A [PKA])-mediated phosphorylation, suggesting a negative feedback loop for cAMP signaling [44,46]. cGMP-mediated inhibition of PDE3A has been implicated in elevation of cAMP levels leading to cross-activation of PKA, which could play a role in NO-induced inhibition of platelet shape change . The main function of PDE5 is to provide negative feedback on cGMP levels. Not only is PDE5 activated by cGMP, but additional phosphorylation of PDE5 by cGMP-dependent protein kinase (protein kinase G [PKG]I) results in activation of PDE5 catalytic activity and long-term desensitization of an NO-induced cGMP response . PDE5 might be involved in compartmentalization of cGMP signaling in specific subellular regions (see below) .
cAMP/cGMP-dependent protein kinases
PKA and PKG translate cAMP/cGMP levels into protein phosphorylation patterns. Human platelets contain micromolar concentrations of PKA and PKG . PKA is a heterotetramer composed of two catalytic subunits and two regulatory subunits. When cAMP binds to the regulatory subunits, the catalytic subunits dissociate from the complex and phosphorylate their substrates, thereby suppressing platelet activation. The main isoforms of regulatory and catalytic subunits expressed in human platelets are RIα, RIβ, RIIβ, Cα, and Cβ, resulting in the formation of PKAI and PKAII holoenzymes [13,24]. The specific roles of PKAI and PKAII in platelets have not been investigated. Recent data suggest that platelets lacking RIIβ of PKA might be suppressed in their ability to become activated, presumably because of the unregulated, inhibitory activity of the catalytic subunit . Another way of activating PKA involves the release of the catalytic subunit of PKA from a nuclear factor-κB–IκB complex . Thrombin and collagen are able to trigger the release of the catalytic subunit from IκB, resulting in the phosphorylation of the substrate proteins vasodilator-stimulated phosphoprotein (VASP) and Rap1GAP2 . This alternative mechanism for PKA activation might represent a negative feedback loop for thrombin-induced and collagen-induced platelet activation (Fig. 1).
In contrast to PKA, the regulatory cGMP-binding and catalytic domains of PKG are combined within one molecule that dimerizes via its N-terminal regulatory region [52–54]. The main isoform expressed in human platelets is PKGIβ. Knockout of the PKGI gene in mice led to a prothrombotic phenotype. The effects of a cGMP analog on platelet shape change, granule release and aggregation were abolished in PKGI-deficient platelets, whereas the effects of a cAMP analog were maintained . In addition, PKGI-deficient mouse platelets adhered more strongly to injured vascular surfaces . The inhibitory effects of NO donors on fibrinogen binding were lost in PKGI-deficient platelets, whereas the effects of a prostacyclin analog were maintained, indicating that most effects of endogenous cGMP are mediated by PKGI. Studies in PKGI-deficient human platelets confirm the role of PKGI in cGMP-mediated inhibition of Ca2+ release from intracellular stores . In 2003, a paper by Li et al. sparked a series of investigations into a potential role for PKGI in platelet activation. Platelets from PKGI knockout mice were observed to exhibit impaired spreading on VWF, and the bleeding times of these animals were slightly increased . In experiments with membrane-permeable cGMP analogs, low levels of PKGI stimulation were suggested to contribute to platelet activation, whereas prolonged stimulation of PKGI would result in platelet inhibition . These data were challenged by other groups, who showed that short-term treatment of platelets with low concentrations of cGMP analogs may lead to unspecific effects [33,57]. Another issue with the proposed biphasic model of PKGI function in platelets is the lack of consistent data on possible targets of cGMP or PKGI that might mediate platelet activation. Initial suggestions that PKGI might activate p38 mitogen-activated protein kinase (MAPK) signaling  could not be confirmed [59,60]. On the other hand, a recent investigation by an independent group showed small but significant stimulatory effects of low concentrations of NO donors on Ca2+ release from intracellular stores induced by low concentrations of thrombin . Taken together, these findings show that a potential role of PKGI during a specific early phase of platelet activation has not yet been established.
Cyclic nucleotide signaling is compartmentalized in many cell types, and some evidence for localized cAMP and cGMP function in platelets has been provided [49,62]. Localized cAMP signaling is often coordinated by A-kinase anchoring proteins (AKAPs) that bind PKA and other signaling compoments such as PDEs, resulting in the formation of subcellular cAMP signaling compartments. Recently, with a mass spectrometry-based screening approach, seven AKAPs (AKAP1, AKAP2, AKAP7, AKAP9, AKAP10, AKAP11, and microtubule-associated protein 2) were identified in platelets , although these initial findings need to be verified with independent methods. Transcriptome analysis suggests that platelets might also express AKAP5, AKAP8, AKAP8L, and AKAP13 . The PKGI substrate protein IRAG represents the only known G-kinase anchoring protein (GKAP) in platelets .
Substrates of cAMP/cGMP-dependent protein kinases
cAMP and cGMP pathways inhibit platelet activation, adhesion, granule release, and aggregation. The phosphorylated substrates of PKA and PKGI link cAMP/cGMP signaling to the functional outcomes of blocked platelet functions. Few PKA and PKGI substrate proteins have been identified to date, and the total numbers and identities of all phosphorylated substrates remain unknown. PKA and PKGI substrates in platelets can be broadly grouped into two main categories: signaling regulators and actin-binding proteins (ABPs) (Fig. 1). An interesting point about cyclic nucleotide signaling in platelets is that, in many cases, cAMP and cGMP signals appear to converge at the level of substrate proteins, as PKA and PKGI activation tends to result in the phosphorylation of the same proteins (Table 1). Important tools for assessing protein phosphorylation in intact platelets are phosphorylation site-specific antibodies. Only a few antibodies against PKA/PKG phosphorylation sites are currently available for platelet studies (Table 1). The substrate proteins included in this review are proteins for which phosphorylation in response to cAMP/cGMP activation was shown in intact platelets with either phosphorylation site-specific antibodies or 32P labeling.
Table 1. Substrates of cAMP-dependent protein kinase (protein kinase A [PKA]I/II) and cGMP-dependent protein kinase (protein kinase G [PKG]I) in platelets
Proposed role of phosphorylation
GP, glycoprotein; HSP27, heat shock protein 27; IP3-R, inositol 1,4,5-trisphosphate receptor; LASP, Lim and SH3 domain protein; PDE, phosphodiesterase; TRPC6, transient receptor potential channel 6; VASP, vasodilator-stimulated phosphoprotein. The PKA and PKGI columns indicate whether substrates have been shown to be phosphorylated by PKA or by PKGI. Question marks indicate that no data on phosphorylation in intact platelets have been published. P-AB indicates the availability of phosphorylation site-specific antibodies against the indicated sites.
One of the first substrates of PKA and PKGI to be identified in platelets was the small G-protein Rap1B . The phosphorylation site was localized to the C-terminus of the protein at Ser179, within a membrane-binding region . Rap1B is a potent regulator of integrin activity, and rap1b−/− mice display impaired platelet aggregation and prolonged tail bleeding times . Little is known about the functional consequences of Rap1B phosphorylation. The phosphorylation does not affect the ability of Rap1 to bind GTP or its GTPase activity . The phosphorylation kinetics are much slower (minutes) than the rapid switch between GTP-bound and GDP-bound states (seconds). The only known functional consequence of Ser179 phosphorylation is a redistribution of Rap1 from the plasma membrane to the cytosol . Live microscopy of DsRed-tagged wild-type Rap1B in HeLa cells shows plasma membrane staining, whereas a phosphomimetic S179E mutant Rap1B localizes to the cytoplasm (O. Danielewski and A. Smolenski, unpublished data). Thus, phosphorylation of Rap1B on Ser179 appears to impact on the subcellular localization of Rap1B.
Considering the vital role of Rap1 in integrin signaling, the function and regulation of its activity are of central importance. Studies of Rap1–GTP levels in human platelets have shown that prostacyclin blocks thrombin-induced Rap1–GTP formation . NO donors and PKG-activating cGMP analogs block thrombin-induced, collagen-induced and ADP-induced Rap1 activation, and this effect was shown to involve PKGI . It is important to note that human platelets do not appear to express any Epac, which is a cAMP-dependent activator of Rap1 . Thus, cAMP signaling in platelets is mainly inhibitory towards Rap1. To identify the possible target of PKA and PKGI that could regulate Rap1 activity in platelets, Schultess et al. screened platelet mRNA for the expression of specific guanine nucleotide exchange factors (GEFs) and GTPase-activating proteins (GAPs) of Rap1. This resulted in the identification of Rap1GAP2 as the only GAP of Rap1 in platelets . Further studies revealed that Rap1GAP2 could be phosphorylated on Ser7 by PKA and PKGI, and a phosphorylation site-specific antibody was used to verify phosphorylation in platelets treated with activators of cAMP and cGMP signaling [72,73]. A 14-3-3-binding site was mapped to the neighboring Ser9 of Rap1GAP2 . 14-3-3 proteins are small, phosphoserine/threonine-binding proteins that function as scaffolds and, despite possessing no catalytic activity, may regulate key signaling components . There are seven highly conserved isoforms in humans, six of which are expressed in platelets . Elevation of cyclic nucleotides, and therefore activation of PKA and PKGI, results in phosphorylation of Ser7 and dissociation of 14-3-3 from Rap1GAP2. 14-3-3 binding appears to dampen the function of Rap1GAP2, whereas detachment of 14-3-3 triggers increased activity, resulting in markedly reduced cell adhesion in transfected cells . Interestingly, platelet activation with agonists such as ADP or thrombin enhanced the binding of Rap1GAP2 to 14-3-3, indicating that this interaction is regulated by both activatory and cyclic nucleotide-dependent inhibitory pathways. Targeting of Rap1 via Rap1GAP2 might explain some of the inhibitory effects of cyclic nucleotides on integrin activation, platelet adhesion, and aggregation. Furthermore, PKA and PKG block the release of Ca2+ from intracellular stores, which might contribute to the inhibition of Rap1 activation, as one of the GEFs of Rap1 in platelets, CalDAG-GEFI, is activated by Ca2+ . The activation of the Rap1B-related protein Rap2B is also effectively blocked by cAMP pathways, although the PKA substrates involved have not been determined .
Another small G-protein that is regulated by cAMP and possibly also by cGMP is RhoA. RhoA is involved in myosin light chain phosphorylation, actin remodeling, integrin activation, and platelet aggregation. Both prostacyclin and direct PKA activators block the formation of RhoA–GTP, whereas PKG activators have a less pronounced effect . Inhibition of RhoA might be mediated by phosphorylation of the heterotrimeric Gα13 (G13), which activates Rho-GEFs. G13 can be phosphorylated by cAMP pathways in platelets, and phosphorylation of G13 on Thr203 reduces RhoA activation [79,80]. However, no phosphorylation site-specific antibodies against Thr203 of G13 have been described, and phosphorylation by PKGI has not been tested. Yet another small G-protein that has been shown to be regulated by cyclic nucleotides is Rac1. Rac1 is involved in lamellopodia formation and platelet granule release and aggregation. cAMP and cGMP were shown to block the formation of Rac–GTP , but the PKA and PKGI substrates mediating Rac inhibition are not known.
Elevation of intracellular Ca2+ concentrations plays an important role in platelet activation. Activation of phospholipase Cβ and phospholipase Cγ isoforms leads to the hydrolysis of phosphatidylinositol 4,5-bisphosphate, generating IP3, which in turn triggers the release of Ca2+ from intracellular stores via IP3-R channels. cAMP and cGMP pathways strongly inhibit the elevation of cytosolic Ca2+ concentrations in platelets, including all types of Ca2+ oscillations that have been observed under flow conditions . At least some of these effects are thought to be mediated by direct phosphorylation of IP3-R . All three isoforms of IP3-R are expressed in platelets, and all can be phosphorylated by PKA and PKGI on as yet unknown sites .
A type I IP3-R-associated protein called IRAG (IP3-R-associated cGKI substrate protein, also known as MRVI1) has been clearly linked to PKGI-mediated inhibition of Ca2+ release in platelets [64,84]. The main phosphorylation sites of IRAG are Ser664 and Ser677, and phosphorylation site-specific antibodies have been used to show phosphorylation in intact platelets in response to activators of cGMP signaling . In a mouse model expressing a mutant form of IRAG that does not bind to IP3-R, inhibition of platelet collagen-induced and thrombin-induced aggregation by cGMP analogs or by NO donors was significantly impaired . The inhibitory effects of NO donors on thrombus formation in the intact carotid artery were abolished in the IRAG mutant, whereas prostacylin and cAMP effects were maintained. Complete deletion of IRAG expression in mice resulted in hyperreactive platelets with a significantly enhanced aggregation response towards thrombin, collagen, and U46619, a thromboxane mimetic . Studies of these IRAG-deficient platelets confirmed that IRAG is involved in NO/cGMP-mediated inhibition of thrombin-dependent, collagen-dependent and U46619-dependent integrin αIIbβ3 activation, aggregation and ATP release . The effects of a cAMP analog on integrin activation were not altered in IRAG-deficient platelets; however, prostacyclin and cAMP effects were not studied extensively . PKGI associates with IP3-RI/IRAG in platelets . In addition, the type I IP3-R was shown to bind PDE5 in platelets, thus providing a link between PKGI and PDE5 . Interestingly, only a fraction of platelet PDE5 was reported to associate with IP3-R, and PKGI might preferentially phosphorylate and activate this IP3-R-associated PDE5, resulting in local control over cGMP-dependent inhibition of Ca2+ release . These results represent initial evidence for compartmentalized cGMP signaling in platelets; however, other data suggest that most platelet PDE5 is activated and phosphorylated by cGMP and PKGI .
Ca2+ levels are further regulated by the transient receptor potential channel 6 (TRPC6), which was shown to be a substrate of PKA and PKGI in platelets . TRPC6 could play a role in store-operated Ca2+ entry, but the consequences of phosphorylation are not clear. TRPC6 was reported to form a complex with type II IP3-R in platelets .
Other signaling proteins
PDE5A is preferentially phosphorylated by PKGI on Ser92, resulting in long-term activation of PDE activity and degradation of cGMP . Similarly, PDE3A is phosphorylated by PKA on Ser312, and this is associated with an increase in catalytic activity, resulting in cAMP degradation . Another signaling target of cyclic nucleotide pathways in platelets is p38 MAPK. p38 MAPK activation is blocked by cAMP and cGMP [59,60]; however, the PKA or PKGI substrates mediating p38 MAPK inhibition are unknown. The glycoprotein (GP)Ib complex is required for platelet adhesion, and GPIbβ was shown to be phosphorylated by PKA on Ser166 . A phosphorylation site-specific antibody against Ser166 has been described . The functional consequences of GPIbβ Ser166 phosphorylation are unclear. Both inhibition and activation of cell adhesion have been observed [88,89]. Ser166 was also suggested to play a role in binding of 14-3-3 to the GPIb complex [90,91]; however, other 14-3-3-binding sites have been described, and the significance of Ser166 phosphorylation in 14-3-3 binding has been questioned . The role of cGMP/PKGI in GPIb phosphorylation has not been investigated. Findings from a study using isolated platelets suggest that NO might be less efficient in inhibiting adhesion by the GPIb complex itself, but might rather block secondary integrin activation . A potential substrate of PKA and PKGI in platelets that is commonly mentioned is thromboxane receptor-α. However, in 32P-labeled platelets, no significant incorporation of phosphate into the thromboxane receptor in prostaglandin-treated or forskolin-treated platelets could be detected , and no other data on phosphorylation of the receptor in intact platelets have been reported. Mass spectrometry-based screening approaches have led to the identification of other putative PKA and PKG substrates in platelets that need to be verified with independent methods .
The regulation of the actin cytoskeleton is a major function of cAMP/cGMP signaling in platelets. One of the first substrates of PKA and PKGI to be identified in platelets was VASP . VASP is expressed at a high concentration of ∼ 25 μm in platelets , and the main phosphorylation sites are Ser157 and Ser239 . Analysis of phosphorylation kinetics has indicated that Ser157 might be preferentially phosphorylated by PKA, whereas Ser239 is preferred by PKGI . Phosphorylation site-specific antibodies have been generated against phosphorylated Ser157 and phosphorylated Ser239, and these antibodies have been used extensively as markers of cyclic nucleotide activity in platelets and other cells [97,98]. The 16C2 mAb against phosphorylated Ser239 has been used in a flow cytometry assay for monitoring the Gi function of the P2Y12 ADP receptor in patients . Mouse models deficient in VASP expression have shown that platelet VASP is involved in platelet activation and that VASP mediates NO-dependent inhibition of platelet adhesion to the vessel wall [100,101]. Furthermore, VASP appears to be involved in the inhibitory effects of PKA and PKGI on fibrinogen binding and platelet aggregation, but not in the inhibition of Ca2+ release or granule secretion . VASP is clearly an important regulator of actin dynamics, but the molecular consequences of VASP phosphorylation in platelets are not well defined. Phosphorylation inhibits binding of VASP to F-actin, and reduces F-actin bundling in vitro . VASP phosphorylation has also been shown to be involved in focal adhesion dynamics  and in regulating the rigidity of the actin cytoskeleton . PKA-mediated phosphorylation of VASP on Ser157 might be controlled locally by integrin β3 .
Other actin-associated proteins that are phosphorylated in human platelets include Lim and SH3 domain protein (LASP), heat shock protein 27 (HSP27), filamin-A (ABP-280), and caldesmon. LASP is phosphorylated by PKA and PKGI on Ser146, resulting in reduced binding of LASP to F-actin and to focal adhesions . During thrombin activation, LASP is phosphorylated on Tyr171, probably by Src kinase . Phosphorylation of HSP27 by PKA and PKGI on Thr143 attenuates HSP27-dependent actin polymerization . Upon ADP treatment of platelets, HSP27 is phosphorylated on additional serines (Ser15, Ser78, and Ser82) by p38 MAPK-dependent pathways , suggesting that, as with LASP, multiple phosphorylation events contribute to the regulation of HSP27. Another actin-binding substrate of PKA is filamin-A. Phosphorylation of filamin-A on Ser2152 protects filamin-A protein against degradation [111,112]. The relevance of filamin-A stabilization for platelet function is unclear. Interestingly, filamin-A is required for the maintenance of platelet membrane stability at high shear levels by binding to the GPIb complex . Caldesmon is an ABP that has been shown to be phosphorylated by prostacyclin-induced signaling in human platelets . The functional consequences of caldesmon phosphorylation for the actin cytoskeleton in platelets have not been studied. A potential substrate of PKA and PKGI that has been mentioned in previous reviews is myosin light chain kinase (MLCK). However, the only study of MLCK phosphorylation showed phosphorylation of purified MLCK by purifed PKA in vitro, but not in intact platelets .
Although cAMP and cGMP have been known to play a powerful role in platelet regulation for many years, the molecular patterns and details mediating this effect are only beginning to emerge (Fig. 1). Most, if not all, of the inhibitory functions of cAMP and cGMP in platelets can be attributed to phosphorylation of substrate proteins by PKA and PKGI. Groups of different substrate proteins with related functions contribute to the inhibitory actions of PKA and PKGI. Currently known substrates can be broadly classified into regulators of signaling and/or regulators of actin dynamics. Signaling regulation involves small G-proteins of the Ras and Rho families, such as Rap1, RhoA, and Rac. Dynamic changes in the architecture of the actin cytoskeleton are involved in many platelet responses, including shape change, adhesion, granule release, and aggregation. The effects of PKA and PKGI on ABPs might complement the effects on small G-proteins, resulting in the net outcome of inhibited platelet adhesion and aggregation. Initial evidence for compartmentalized cAMP/cGMP signaling in platelets has emerged from studies of the IP3-RI complex involved in the regulation of Ca2+ release from intracellular stores. Regulation of Ca2+ levels is likely to have a broad impact on many pathways, including the activation of G-proteins and the granule release reaction. Some of the identified PKA and PKGI substrates, that is, Rap1GAP2, PDE3A, and HSP27, appear to harbor additional phosphorylation sites, which are targeted by activatory pathways. Multiple phosphorylation events need to be translated into appropriate functional outcomes. This might, at least in some cases, be achieved by differential binding of 14-3-3, as shown for Rap1GAP2 and PDE3A [44,73].
Adequate control of platelet reactivity requires a careful balance between activatory and inhibitory signaling pathways. Platelet activators are known to counteract the cAMP/cGMP system at various levels. For example, Gi-coupled receptor signaling attenuates cAMP synthesis, activation of PDE3A results in the degradation of cAMP , and platelet activatory signaling interferes with cyclic nucleotide signaling at the level of PKA/PKG substrate proteins . Thrombospondin-1 was suggested to mediate platelet activation by blocking cyclic nucleotide signaling at the level of sGC, PDE3A, PKA, and PKGI [116,117]. On the other hand, platelet activators may cross-activate inhibitory cyclic nucleotide signaling, as shown for VWF-induced activation of sGC [22,28] and for thrombin-induced and collagen-induced activation of PKA .
Major open questions in platelet cAMP/cGMP signaling that need to be addressed are: (i) the mechanisms and significance of new pathways of sGC activation; (ii) compartmentalization of cAMP/cGMP signaling and the role of AC isoforms, AKAPs, and GKAPs; (iii) the specific role of PKAI and PKAII isoforms; (iv) the proposed role of cGMP in platelet activation by low concentrations of platelet agonists; (v) the identity of all PKA and PKGI substrates and the coordination of their actions; and (vi) the contribution of defects in cAMP/cGMP pathways to platelet hyperreactivity in cardiovascular disease. Further studies of the cAMP/cGMP signaling network in platelets might lead to the identification of novel markers of platelet function and reactivity, and possibly new therapeutic targets.
The author would like to thank K. Gegenbauer and A. Hampson for help with the preparation of the manuscript, and three anonymous reviewers for valuable comments and suggestions. This work was supported by Science Foundation Ireland (08/IN.1/B1855) and the UCD School of Medicine and Medical Science.
Disclosure of Conflict of Interests
The authors state that they have no conflict of interest.