Compartmentation and compartment-specificregulation of PDE5 by protein kinase G allowsselective cGMP-mediated regulation ofplatelet functionsLindsay S. Wilson*, Hisham S. Elbatarny†, Scott W. Crawley‡, Brian M. Bennett†, and Donald H. Maurice*†§

Departments of *Pathology and Molecular Medicine, †Pharmacology and Toxicology, and ‡Biochemistry, Queen’s University, Kingston, ON, Canada K7L 3N6

Edited by Joseph A. Beavo, University of Washington School of Medicine, Seattle, WA, and approved July 1, 2008 (received for review May 16, 2008)

It is generally accepted that nitric oxide (NO) donors, such assodium nitroprusside (SNP), or phosphodiesterase 5 (PDE5) inhib-itors, including sildenafil, each impact human platelet function.Although a strong correlation exists between the actions of NOdonors in platelets and their impact on cGMP, agents such assildenafil act without increasing global intra-platelet cGMP levels.This study was undertaken to identify how PDE5 inhibitors mightact without increasing cGMP. Our data identify PDE5 as an integralcomponent of a protein kinase G1� (PKG1�)-containing signalingcomplex, reported previously to coordinate cGMP-mediated inhi-bition of inositol-1, 4, 5-trisphosphate receptor type 1 (IP3R1)-mediated Ca2�-release. PKG1� and PDE5 did not interact in sub-cellular fractions devoid of IP3R1 and were not recruited to IP3R1-enriched membranes in response to cGMP-elevating agents.Activation of platelet PKG promoted phosphorylation and activa-tion of the PDE5 fraction tethered to the IP3R1-PKG complex, aneffect not observed for the nontethered PDE5. Based on thesefindings, we elaborate a model in which PKG selectively activatesPDE5 within a defined microdomain in platelets and propose thatthis mechanism allows spatial and temporal regulation of cGMPsignaling in these cells. Recent reports indicate that sildenafil mightprove useful in limiting in-stent thrombosis and the thromboticevents associated with the acute coronary syndromes (ACS), situ-ations poorly regulated with currently available therapeutics. Wesubmit that our findings may define a molecular mechanism bywhich PDE5 inhibition can differentially impact selected cellularfunctions of platelets, and perhaps of other cell types.

calcium � PKG

B lood platelets prevent blood loss and promote wound healingby aggregating in response to injury and promoting throm-

bus formation. Although these processes are of significantadaptive advantage, chronic platelet activation results in alteredblood flow patterns and promotes the formation of thrombus-based arterial occlusions. Arterial occlusions contribute to thepathogenesis of acute coronary syndromes (ACS) (1), a spec-trum of conditions including unstable angina and myocardialinfarctions. Although drug-eluting stents have largely mitigatedthe problem of in-stent restenosis after percutaneous coronaryinterventions, in-stent thrombosis, which can occur anytimeafter stenting, often presents catastrophically triggering death oracute myocardial infarctions (2–6). Although anti-plateletagents, including aspirin or thienopyridines, can reduce throm-bosis in ACS or at stents, their weak potencies rarely eliminateplatelet-mediated re-occlusions in response to strong plateletactivation signals, including those associated with thrombolysis(7, 8). The anti-platelet actions of the selective cyclic nucleotidephosphodiesterase 5 (PDE5) inhibitors, including sildenafil ci-trate (Viagra), was suggested to represent a therapeutic optionin controlling ACS and in-stent thrombosis (9–11). Although theanti-platelet actions of sildenafil support the use of PDE5

inhibitors as anti-thrombotic agents, other reports suggest thatsildenafil might have proaggregatory effects (12). To date, nobasis for these seemingly contradictory findings were offered.

A significant literature supports the concept that cellularcyclic nucleotide signaling is compartmented. Indeed, distinctmacromolecular cAMP-signaling complexes have been shown toallow resolution of the distinct spatial and temporal effects ofseveral cAMP-mediated actions in cells (13, 14). Although mostcAMP-signaling complexes contain protein kinase A (PKA), andare defined based on the identity of the A-kinase anchoringprotein (AKAP) tethering the PKA, others contain the ex-change-protein activated by cAMP (EPAC) (14–16). Morerecently, a critically important role for integration of specificcyclic nucleotide phosphodiesterases (PDEs) into these com-plexes has emerged as a mechanism, allowing their coordinatedactions in cells. Although much less extensively studied, growingevidence also supports an important role for compartmentationof cGMP-based cellular signaling. Indeed, putative PKG-bindingproteins (GKAPs) have been identified in certain cells (17, 18)and selective subcellular distribution of cGMP-activated kinases(PKG) and cGMP-hydrolyzing PDEs has also been reported(19–21). For example, a model in which distinct PDEs selectivelyregulate either plasma membrane, or cytosolic, cGMP ‘‘pools,’’has emerged from recent experiments using heterologouslyexpressed cGMP biosensors (19, 21). Herein we report thatPDE5 forms an intrinsic and critically important regulatorycomponent of a previously identified IP3R1-based, PKG-containing signaling complex in human platelets (22, 23). Wepropose that these data may further our understanding of theeffects of PDE5 inhibitors reported in recent literature andsupport the concept that PDE5-inhibitors may be useful agentsin inhibiting platelet activation.

ResultsSodium Nitroprusside and Sildenafil Synergize to Increase PlateletcGMP and Inhibit Aggregation. NO donors (i.e., sodium nitroprus-side, SNP) inhibit platelet activation by increasing cGMP andinhibitors of the dominant platelet cGMP phosphodiesterase,PDE5, potentiate these effects (26–29). In our experiments, SNP(1–100 �M) inhibited platelet aggregation in a concentration-dependent manner; a selective PDE5 inhibitor, sildenafil (100nM), potentiated this effect (Fig. 1A). Sildenafil alone did not

Author contributions: L.S.W., S.W.C., and D.H.M. designed research; L.S.W. and H.S.E.performed research; B.M.B. contributed new reagents/analytical tools; L.S.W., H.S.E.,S.W.C., B.M.B., and D.H.M. analyzed data; and L.S.W. and D.H.M. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

§To whom correspondence should be addressed. E-mail: mauriced@post.queensu.ca.

This article contains supporting information online at www.pnas.org/cgi/content/full/0804738105/DCSupplemental.

© 2008 by The National Academy of Sciences of the USA

13650–13655 � PNAS � September 9, 2008 � vol. 105 � no. 36 www.pnas.org�cgi�doi�10.1073�pnas.0804738105

Dow

nloa

ded

by g

uest

on

July

26,

202

1

inhibit platelet aggregation (Fig. 1 A). SNP (10 �M) caused arapid (3.3 � 0.4-fold after 1 min) and transient (1.8 � 0.4-foldafter 3 min) increase in platelet cGMP but sildenafil (100 nM)did not increase platelet cGMP levels (0.96 � 0.2-fold controlvalues after 3 min). Together, SNP (10 �M) and sildenafil (100nM) synergistically increased cGMP (24 � 3-fold after 3 min).

Sildenafil Inhibits Thrombin-Induced Ca2� Release. Thrombin gen-erates intra-platelet Ca2� transients by promoting opening ofIP3R1 channels and releasing endoplasmic reticulum (ER) Ca2�

stores (30). PKG activation inhibits this action of thrombin (23)and PKG-mediated phosphorylation of IP3R1, and of an IP3R1-associated PKG-substrate protein (IRAG), coordinates the in-hibition (22, 23). In agreement with previous reports (23), SNP

inhibited thrombin-induced Ca2� release in human platelets(Fig. 1B). Thus, at a concentration that inhibited platelet aggre-gation by approximately 50% (10 �M), SNP inhibited thrombin-induced Ca2� release by 39 � 4% (Fig. 1B). In marked contrastto its effect on platelet aggregation, sildenafil alone significantlyinhibited thrombin-induced platelet Ca2� release. In fact, silde-nafil (100 nM) had a more marked effect than SNP (10 �M) inthis regard (Fig. 1B). Because thrombin did not alter plateletcGMP levels, or influence the ability of sildenafil to alter plateletcGMP (Table 1), a comparison of the effects of SNP andsildenafil are inconsistent with a direct correlation betweenchanges in global intra-platelet cGMP and inhibition of Ca2�

release. Rather, these data are consistent with the concept thatsildenafil acted locally to regulate this intra-platelet event.

Previous reports showed that NO donors could alter cAMPlevels in cells expressing the cGMP-sensitive cAMP-hydrolyzingPDEs, PDE2, or PDE3. Under the conditions of our study,neither SNP nor sildenafil altered human platelet cAMP andthrombin did not alter this fact (Table 1). To test directly whetherPKA was involved in SNP-, or sildenafil-induced inhibition ofthrombin-mediated Ca2� release, we inhibited PKA in someexperiments. Although the PKA-activator, 6BzcAMP (30 �M),inhibited thrombin-induced Ca2� release, and the cell-permeable PKA inhibitory peptide, myristoylated PKI (My-PKI), reversed this effect, My-PKI did not attenuate the abilityof SNP, or sildenafil, to inhibit thrombin-induced Ca2� release(Fig. 1B). Together these data demonstrate that the cAMP-PKAsystem did not coordinate the effects of SNP, or sildenafil, onCa2� transients in human platelets.

PDE5 Is Resident in an ER cGMP-Signaling Complex. Because silde-nafil inhibited thrombin-induced Ca2� transients without in-creasing global intra-platelet cGMP, we tested the hypothesisthat PDE5 might be acting locally at the platelet ER. Moreprecisely, we determined whether PDE5 formed a functionalpart of the IP3R1-IRAG-PKG1�-signaling complex (23). Dif-ferential centrifugation of platelet lysates identified PKG1�(�25%), PDE5 (�45%), and IP3R1 (�100%) in particulatecellular fractions (Fig. 2A). Interestingly, selective immunopre-cipitation of PKG1�, PDE5, or IP3R1, allowed coimmunopre-cipitation of all three proteins (Fig. 2C); a finding consistent withthe idea that PDE5 was integral to the IP3R1-IRAG-PKG1�intra-platelet complex. Consistent with our finding that PKAinhibition did not impact sildenafil-induced inhibition of Ca2�

transients, neither the platelet cAMP PDE (PDE3A), nor PKAwere recovered in immune complexes containing the cGMP-signaling proteins. To ensure specificity of our immunoprecipita-tions, all experiments contained a control immunoprecipitationwith rabbit IgG (control IPs) (Fig. 2C). Also, immune complexeswere routinely probed for, and found deficient in, abundant ERproteins such as BiP (data not shown). We were unable to securean aliquot of an anti-IRAG antibody and could not directly test forassociation between IRAG and PDE5 in our studies.

PKG Phosphorylates and Activates PKG-Associated PDE5 in Vitro.Because PDE5 was a known PKG substrate (31, 32), andphosphorylation activated this enzyme, we determined whether

Fig. 1. Regulation of platelet functions. (A) PRP was incubated with SNP (10�M), sildenafil (100 nM), or both agents (30 s, 37°C while stirring at 1,000 rpm).Subsequently, platelet aggregation was promoted [ADP (2 �M), 3 min, 1000rpm]. *, significant difference (P � 0.05) against ‘‘Basal’’; **, significantdifference (P � 0.05) against SNP. Aggregation are expressed as mean � SEM.(n � 5). (B) Fluo-4-loaded platelets were incubated with or without My-PKI (10�M, 5 min) and then with 6BzcAMP (30 �M), SNP (10 �M), sildenafil (100 nM)either singly or in combination (3 min). Subsequently, thrombin (0.4 units/ml)was added and [Ca2�] measured at 520 nm (n � 4). *, significant differences(P � 0.05) between thrombin alone and thrombin with these agents; **,significant difference (P � 0.05) between SNP and SNP/sildenafil.

Table 1. Impact of SNP and sildenafil treatment on intracellular platelet cGMP and cAMP levels

Additions

cGMP measurement, pmol cGMP/mg protein cAMP measurement, pmol cGMP/mg protein

Without thrombin 0.4 unit/ml thrombin Without thrombin 0.4 unit/ml thrombin

Basal 0.27 � 0.03 0.19 � 0.02 3.14 � 0.41 2.68 � 0.4610 �M SNP 3 min 1.07 � 0.35* 1.90 � 0.21* 2.97 � 0.54 2.50 � 0.28100 nM sildenafil 3 min 0.33 � 0.08 0.34 � 0.09 2.39 � 0.32 2.00 � 0.5410 �M SNP�100 nM sildenafil 3 min 5.80 � 1.19** 5.25 � 1.3** 3.36 � 0.80 2.41 � 0.13

*P � 0.05 against basal; **, P � 0.05 against SNP.

Wilson et al. PNAS � September 9, 2008 � vol. 105 � no. 36 � 13651

PHA

RMA

COLO

GY

Dow

nloa

ded

by g

uest

on

July

26,

202

1

PKG-associated PDE5 was a PKG substrate. Thus, in vitro kinaseassays with PKG1�-immunoprecipitates allowed phosphoryla-tion of several proteins with electrophoretic mobilities consistentwith IP3R1 (�250 kDa), IRAG (�120 kDa), and PDE5 (�95kDa) (Fig. 2B). The approximate 250-kDa and 95-kDa phos-phoproteins were recognized by IP3R1- or PDE5-selective an-tisera, respectively (Fig. 2C). Also, in vitro kinase assays withPKG1�-immunoprecipitates showed that cGMP (50 �M) andATP (250 �M) resulted in robust phosphorylation of PDE5 atS102 (Fig. 3A), and a significant level of activation of the tetheredPDE5 (�3-fold, Fig. 3B). Addition of cGMP alone to PKG-immune complexes did not cause PDE5-S102 phosphorylation orPDE5 activation (Fig. 3 A and B). In marked contrast, in vitrokinase assays of anti-PDE5 immunoprecipitates did not resultin PDE5 phosphorylation at S102, nor PDE5 activation (Fig. 3 Aand B).

PKG-Associated PDE5 Is Selectively Activated by PKG in Platelets.Although previous work indicated that PDE5 was activated uponcGMP-binding to a PDE5 GAF-A domain, or PKG phosphor-ylation of PDE5 at S102 (31–33), these studies were silent on therelative importance of these mechanisms in cells. To address thisissue, we compared the phosphorylation and activation of thePKG-associated and non-PKG-associated forms of PDE5 in8BrcGMP (1 mM, 15 min)-treated platelets. Strikingly,8BrcGMP treatment of platelets markedly increased the S102

phosphorylation status and activity of the PKG-associated formof PDE5, but not that which was not associated (Fig. 4 and Table2). Consistent with the idea that the phosphorylated PDE5 wasresident within the IP3R1/IRAG/PKG1� complex, IP3R1 wasrecovered in the anti-PKG immune complexes but not in thoserepresenting bulk PDE5 (Fig. 4). Similarly, when anti-IP3R1immune complexes were obtained from control or 8BrcGMP-treated platelets, only the IP3R1-associated PDE5 was activatedby 8BrcGMP (Fig. 5). An identical pattern of PDE5 activationwas obtained when PDE5 was isolated using the method used

originally to isolate and characterize the IP3R1/IRAG/PKG1�complex in platelets (Fig. 5). Taken together, these data wereconsistent with the novel idea that only PKG-associated PDE5

Unbound BoundA B

170IP3R

(kDa)

PDE5

130

IRAG

PDE5

PKG

IP R 95

72

PDE5IP3R

Immunoprecipitatingantisera

ImmunoprecipitatingantiseraC

IP3R

PDE5

Immunoblotantisera

Immunoblotantisera IP3R

PDE5PDE5

PKG

PDE5

PKG

PDE5

PDE3A

PKAc

PDE3A

PKAc

Fig. 2. Identification of a cGMP signaling complex. (A) Washed platelets were fractionated, and PDE5, PKG, or IP3R1 were detected by immunoblot analysis.(B) Incorporation of 32P into anti-PKG immunoprecipitate proteins after incubation with cGMP (50 �M) and ATP (250 �M [�-32P]ATP) for 30 min. 32P incorporationin the immune complex, or released by the treatment, were resolved by SDS/PAGE and detected by autoradiography (n � 3). (C) Precleared platelet lysates (1.5mg) were incubated with either antisera against PDE5 (1 �g), PKG (1 �g), IP3R1 (1 �g), or mouse IgG (1 �g) and Protein A/G beads (16 h, 4°C). Immune complexesand aliquots of cell lysates (30 �g) were each resolved by SDS/PAGE and immunoblotted for PDE5 (1:5,000), PKG (1:5,000), IP3R1 (1:1,000), PDE3A (1:2,000), orPKAc (1:1,000).

250µµM ATP -

TotalLysate PKG IP

+PDE5 IP

- +

A

(P)S102-PDE5

50µM cGMP + + + +

PDE5

PKG

PKG IP

*PDE5 IP

300

n

14

16

18 *

n 200

250

B

mol

/mg/

min

8

10

12

ol/m

g/m

in

150

200

pm

2

4

6

pmo

50

100

0DMSO DMSOSildenafil Sildenafil

50µM cGMP 50µM cGMP+ 250µM ATP

DMSO DMSOSildenafil Sildenafil

50µM cGMP 50µM cGMP+250µM ATP

0

Fig. 3. In vitro phosphorylation of PKG-bound PDE5. Anti-PKG or anti-PDE5immunoprecipitates were incubated with either cGMP (50 �M) or both cGMP(50 �M) and ATP (250 �M) (30 min, 30°C). (A) Immune complexes wereresolved by SDS-PAGE and immunoblotted with a S102-specific phospho-PDE5antibody (1:1,000). (B) Effect of treatments in (A) on PDE5 activities in theimmunoprecipitates. *, significant difference (P � 0.05) between PKG-immunoprecipitate treated with or without ATP. Immunoblots and PDE5activity values are from the same experiment and are representative of threeexperiments.

13652 � www.pnas.org�cgi�doi�10.1073�pnas.0804738105 Wilson et al.

Dow

nloa

ded

by g

uest

on

July

26,

202

1

was subject to PKG-catalyzed phosphorylation and activation incells treated with 8BrcGMP. Consistent with this, addition ofalkaline phosphatase to anti-PKG immune complexes dephos-phorylated and further reduced their PDE5 activity. In contrast,phosphatase treatment was without effect on the PDE5 activitypresent in anti-PDE5 immunoprecipitates (Table 2).

In addition to being selectively activated by PKG in cells, ourdata indicate that the IP3R1-IRAG-PKG1�-associated PDE5had a lower specific activity than that isolated from the bulkcytosol (Fig. 6). Indeed, the IP3R1-IRAG-PKG1�-associatedPDE5 was only approximately 14% as active as that isolated inanti-PDE5 immunoprecipitates (i.e., �21 vs. �140 pmol/mg/min, respectively). Because we routinely assessed PDE5 activityin our studies under nonsaturating substrate conditions, furtherstudies will be required to assess the kinetic basis of the lowerspecific activity of the tethered PDE5 in platelets.

Impact of PKG-Activation on IP3R1-IRAG-PKG-PDE5 Complex Dynamicsin Platelets. To determine whether PDE5 and PKG formed astable complex in platelets, or only interacted within the ER-based IP3R1-signaling complex, we assessed whether these pro-teins could be coimmunoprecipitated in the IP3R1-devoid cyto-solic fractions. Our data are inconsistent with the idea that PDE5and PKG form a stable complex in platelet cellular fractionsdevoid of IP3R1. Indeed, although PKG, or PDE5, were eachindividually immunoprecipitated from platelet supernatants, inthe four separate experiments in which it was tested, theseproteins never coimmunoprecipitated from this fraction [sup-porting information (SI) Fig. S1]. Because PDE5, PKG, or IP3R1was not enriched in IP3R1-containing subcellular fraction in cellsincubated with a cGMP analogue, or with an NO donor, our dataare inconsistent with the idea that cGMP elevation triggersmovement of these proteins between these fractions (Fig. S1). Ofnote, we did observe that cGMP-elevating agents increased thefraction of IP3R1 and PKG that could be coimmunoprecipitated(Fig. S1). Although it is likely that this effect will impact thedynamics of PKG-mediated phosphorylation of proteins in thissignaling complex, and perhaps more broadly in the ER, furtherstudies will be required to directly test this hypothesis and to fullyelucidate its molecular basis.

DiscussionIn this article, we confirm that NO donors inhibit ER Ca2�

release (23), that this effect correlates positively with their abilityto increase intra-platelet cGMP, that human platelets contain aprotein complex consisting of IP3R1, IRAG, and PKG1�, andshow that the activation of PKG in this complex promotes IP3R1phosphorylation. In addition, our data identify a potentiallyimportant role for cGMP hydrolysis by PDE5 in coordinatingthis event, a step that had previously been ignored (23). Thus, wereport that PDE5 resides in the IP3R1-based complex and showthat PDE5 inhibition can play a determinant role in cGMP-basedcontrol of Ca2� release in platelets without increasing globalintra-platelet cGMP. We suggest that our data are inconsistentwith the idea that global levels of cGMP are an estimate of theinhibition of Ca2� release caused by agents acting throughcGMP, but rather with the idea that localized changes in cGMP,perhaps at the ER, may more closely correlate with their effects.In addition to showing that sildenafil impacts platelet function

8BrcGMP (1mM) - --+ + +Lysate PKG IP PDE5 IP

(P)S102-PDE5

PDE5

PKG

IP3R

Fig. 4. PKG-associated PDE5 is selectively phosphorylated by PKG in intactplatelets. Washed human platelets were incubated with, or without, 8BrcGMP(1 mM) and processed for immunoblot analysis as described in the legend toFig. 2. Data are representative of observations made in three experiments.

Table 2. Impact of 8BrcGMP treatment of human platelets on PKG- and PDE5-immunoprecipitated PDE5 activities

Sample AdditioncGMP PDE activity (�sildenafil),

pmol�mg�1�min�1

cGMP PDE activity (�sildenafil),pmol�mg�1�min�1

cGMP PDE activity (�1 unitCIAP), pmol�mg�1�min�1

Total lysate None 31.0 � 1.1 3.3 � 0.8Total lysate 8BrcGMP 40.0 � 6.2 1.2 � 0.6PKG IP None 4.6 � 0.7 0.9 � 0.8 0.9 � 0.7#PKG IP 8BrcGMP 7.6 � 0.9* �0.5 � 0.8 0.0 � 0.7#PDE5 IP None 626.3 � 13.4 25.9 � 1.1 572.7 � 22.4PDE5 IP 8BrcGMP 660.9 � 21.6 80.8 � 5.9 567.4 � 27.1

*, P � 0.05 against basal PKG-IP; #, P � 0.05 against basal cGMP PDE activities.

Fig. 5. Particulate, IP3R1-associated PDE5, is selectively activated by PKG.Control or 8BrcGMP-treated human platelet lysates were fractionated andprocessed as described in the legend to Fig. 2. Data are cGMP PDE activitiesobtained in representative IP3R1-immunoprecipitates or cGMP-agarose pre-cipitates from three experiments. *, significant difference (P � 0.05) betweencGMP PDE activity in IP3R1-immunoprecipitates, or cGMP-agarose precipi-tates, derived from the 100,000 � g pellet of 8BrcGMP-treated plateletscompared with this fraction in control platelets.

Wilson et al. PNAS � September 9, 2008 � vol. 105 � no. 36 � 13653

PHA

RMA

COLO

GY

Dow

nloa

ded

by g

uest

on

July

26,

202

1

without increasing platelet cGMP, we identify a scaffold onwhich this effect is coordinated. We propose that the presenceof PDE5 within this complex likely allows local actions of cGMPto be regulated in a more dynamic manner; a propositionconsistent with the potent inhibition of thrombin-induced re-lease of Ca2� transients induced by sildenafil in the absence ofsignificant changes in platelet cGMP.

Of critical importance and described in this report was thedemonstration that activation of platelet PKG selectively pro-moted the phosphorylation/activation of the IP3R1-IRAG-PKG1� complex-associated PDE5. Indeed, no evidence of PKG-mediated phosphorylation/activation of the bulk PDE5 in thecytosol was observed. Although unexpected, our data are con-sistent with the idea that only PKG-associated PDE5 is subjectto this mode of regulation in cells. Indeed, we report that PKGand PDE5 could only be coimmunoprecipitated from plateletfractions enriched in IP3R1, but not from the bulk cytosol, thefraction in which PDE5 and PKG were most abundant. Inaddition to reporting that PKG and PDE5 are only ever incomplex at membranes enriched with IP3R1, we also found noevidence to support the idea that PKG, or PDE5, could betrafficked to the IP3R1-enriched membranes in response toincreases in cellular cGMP, or activation of PKG.

As presented in this manuscript, we found little evidence thatPDE5 could be activated directly after binding of cGMP to thisenzyme in cells, or in PKG-based immunoprecipitates. However,previous reports have elegantly shown that the rate of PDE5hydrolysis of cGMP renders a direct test of this hypothesisvirtually impossible in intact cells, or in concentrated PDE5-containing immune complexes (31–34). We submit that thediscovery of GAF-selective cGMP analogues will likely benecessary before this hypothesis can be rigorously tested in acellular context.

This study shows that the differential regulation of the activ-ities of compartmented vs. noncompartmented PDEs has thepotential to allow selective effects in cells. Thus, we propose amodel (Fig. S2) in which compartmented PDE5 exhibits low

catalytic activity and that this enzyme is only fully active subse-quent to its phosphorylation by PKG. In contrast, we proposethat the noncompartmented platelet PDE5 is constitutively moreactive than its compartmented counterpart and that this enzymeis not further activated by PKG-mediated phosphorylation. Wepropose that this model, combined with recent finding in whichPDE2, or PDE5, were each shown to selectively regulate plasmamembrane or cytosolic, pools of cGMP, respectively (21), shouldspur further investigations aimed at achieving the greatestdegree of selectivity possible with PDE inhibition.

The catastrophic consequences of coronary artery disease-associated ACS, and the recent reports identifying very signif-icant rates of ‘‘late’’ and ‘‘very late’’ in-stent thrombotic eventsin otherwise healthy individuals have each spurred efforts toidentify more effective agents than those anti-thrombotic agentscurrently available. Indeed, although it is generally acknowl-edged that drugs such as aspirin and the thienopyridines caneffectively reduce thrombosis in ACS, they are significantly lesseffective at inhibiting the strong platelet activations associatedwith thrombolysis or in-stent thrombosis (7, 8); the latter rep-resent a significant cause of acute myocardial infarctions andsudden cardiac death (2–5). In this context, recent attention hasshifted to using selective PDE5 inhibitors such as sildenafilcitrate (Viagra) for prevention of thrombosis (10, 11). Althoughthe use of PDE5 inhibitors for anti-platelet therapeutics isconsistent with the widely accepted idea that increases inintracellular cGMP result in inhibition of platelet aggregation,scattered recent reports have suggested that sildenafil mighthave proaggregatory effects and that these prothrombotic effectscould limit their utility (12). Based on the findings reported here,we suggest that further analysis of the effects of selectiveinhibition of the IP3R1-IRAG-PKG1�-associated PDE5, and ofthe fraction not associated with this complex, may illuminatethese conflicting reports. In conclusion, our study is consistentwith an important role for PDE5 in shaping and maintainingdistinct cGMP pools in platelets and demonstrates that cGMPcompartmentation in platelets depends on both selective target-ing and the differential regulation of this enzyme. Our data shedlight on the molecular mechanism by which sildenafil, and perhapsother potent and selective PDE5 inhibitors, could reduce humanplatelet activation and support the notion that they may proveuseful in reducing unwanted thrombotic episodes.

Experimental ProceduresMaterials. Pharmacological agents were from Sigma–Aldrich, Biolog, Calbio-chem, or Fisher Scientific. Radiolabeled nucleotides were from Perkin Elmer.Sildenafil was isolated as described in ref. 24.

Platelet Function Studies. Platelet-rich plasma (PRP) was prepared by centrif-ugation of heparinized (15 units/ml) blood (284 � g, 15 min at 25°C) and usedin aggregation studies as described in ref. 25. Platelets were washed in aCa2�-free buffer (0.35% BSA, 137 mM NaCl, 2.7 mM KCl, 11.9 mM NaHCO3, 1mM MgCl2, 0.26 mM EGTA, 3 mg/ml apyrase ,and 5 mM Pipes, pH 6.5) and thenin this buffer containing 2 mM CaCl2 and 5 mM Hepes (pH7.4) before beingloaded with Fluo-4 A.M. (3 �M, 30 min) for use in Ca2�-release studies.Drug-induced Ca2� transients were measured with a spectrophotometer.Cyclic nucleotides were measured by RIA from trichloroacetic acid (6% vol/vol)precipitates of platelets (108) after incubation with pharmacological agents asdescribed in ref. 26.

Protein–Protein Interaction Studies. Platelets were lysed in a Tris-based buffer(pH 7.4) [1 mM EDTA, 100 mM NaCl, 5 mM MgCl2, 5 mM benzamidine, 1 �l/mlaprotinin, 5 �l/ml bestatin, 2 �g/ml leupeptin, 10 mM phenylmethylsulfonylfluoride (PMSF), 10 mM sodium �-glycerophosphate, 10 mM sodium pyro-phosphate, 10 mM NaF, and 10 mM sodium vanadate], and precleared withrabbit IgG (1 �g)/protein A/G-agarose (3 h, 4°C). Precipitates were generatedfrom precleared platelet lysates (1.5 mg), or isolated subcellular fractionprepared by centrifugation, using PKG (Stressgen), PDE5 [a gift from S. S.

Total Total

IB: PDE5

Lysate IP: PDE5 Lysate IP: PKGA

IB: PKG

B

)

140

mo

l/min

/mg

100

120

E a

ctiv

ity

(p

40

60

80

*

PD

0

20

40

0

PDE5 ipsildenafilDMSO

PKG ip

DMSO sildenafil

Fig. 6. Low specific activity of compartmented PDE5. Precleared plateletlysates were processed as described in the legend to Fig. 2. Immunoblottingdata (A) and cGMP PDE activities with cGMP (1 �M) (B) are representative ofseven experiments. *, significant difference (P � 0.05) between the cGMP PDEactivity in anti-PKG and anti-PDE5-derived immune complexes.

13654 � www.pnas.org�cgi�doi�10.1073�pnas.0804738105 Wilson et al.

Dow

nloa

ded

by g

uest

on

July

26,

202

1

Visweswariah (Indian Institute of Science, Bangalore, India)], IP3R1 (Neuro-Mab), phospho-S102 PDE5 (Fabgennix) or control rabbit-antisera (1 �g), andprotein A/G 40 �l or 8-AET-cGMP agarose (25 �l). Immune complex-associatedproteins were detected by SDS/PAGE/immunoblot analysis and cGMP PDEactivities were measured using a fixed concentration of cGMP (1 �M), asdescribed in ref. 14. For in vitro PKG kinase assays, immune complexes wereincubated in a buffer (50 �M cGMP, 20 mM TES, 2 mM MgCl2, 10 mM NaF, and10 mM Na2VO4) supplemented with 250 �M ATP or [�-32P ATP] (30 min, 30°C).Reaction products were analyzed by immunoblot or PDE assays as describedabove.

Statistical Analysis. Values are presented as Mean � SEM from at least threeindependent experiments. Effect of agents on aggregation, cGMP or cAMPlevels, Ca2� transients, or PDE activities were independently tested for signif-icance using a two-tailed Student’s t test with P � 0.05 considered significant.

ACKNOWLEDGMENTS. This work was supported by Heart and Stroke Foun-dation of Ontario and Canadian Institute of Health Research Grants (to D.H.M.and B.M.B.). D.H.M. is a Career Investigator with the Heart and Stroke Foun-dation of Ontario.

1. Fuster V, Badimon L, Badimon JJ, Chesebro JH (1992) The pathogenesis of coronaryartery disease and the acute coronary syndromes. N Engl J Med 326:242–250.

2. Cook S, et al. (2007) Incomplete stent apposition and very late stent thrombosis afterdrug-eluting stent implantation. Circulation 115:2426–2434.

3. Stone GW, et al. (2007) Offsetting impact of thrombosis and restenosis on the occur-rence of death and myocardial infarction after paclitaxel-eluting and bare metal stentimplantation. Circulation 115:2842–2847.

4. Webster MW, Ormiston JA (2007) Drug-eluting stents and late stent thrombosis. Lancet370:914–915.

5. Doyle B, et al. (2007) Outcomes of stent thrombosis and restenosis during extendedfollow-up of patients treated with bare-metal coronary stents. Circulation 116:2391–2398.

6. Windecker S, Meier B (2007) Late coronary stent thrombosis. Circulation 16:1952–1965.7. Coller BS (1990) Platelets and thrombolytic therapy. N Engl J Med 4:33–42.8. Chen LY, et al. (1995) Aspirin does not potentiate effect of suboptimal dose of the

thrombin inhibitor inogatran during coronary thrombolysis. Cardiovasc Res 30:866–874.9. Schmidt U, et al. (2001) Cessation of platelet-mediated cyclic canine coronary occlusion

after thrombolysis by combining nitric oxide inhalation with phosphodiesterase-5inhibition. J Am Coll Cardiol 37:1981–1988.

10. Lewis GD, et al. (2006) Sildenafil improves coronary artery patency in a canine modelof platelet-mediated cyclic coronary occlusion after thrombolysis. J Am Coll Cardiol47:1471–1477.

11. Rehse K, Scheffler H, Reitner N (1999) Interaction of Viagra with the NO donorsmolsidomine and RE 2047 with regard to antithrombotic and blood pressure loweringactivities. Arch Pharmacol 332:182–184.

12. Li Z, et al. (2003) A stimulatory role for cGMP-dependent protein kinase in plateletactivation. Cell 112:77–86.

13. Dodge KL, et al. (2001) mAKAP assembles a protein kinase A/PDE4 phosphodiesterasecAMP signaling module. EMBO J 20:1921–1930.

14. Netherton SJ, Sutton JA, Wilson LS, Carter RL, Maurice DH (2007) Both protein kinaseA and exchange protein activated by cAMP coordinate adhesion of human vascularendothelial cells. Circ Res 101:768–776.

15. Raymond DR, Wilson LS, Carter RL, Maurice DH (2007) Numerous distinct PKA-, orEPAC-based, signalling complexes allow selective phosphodiesterase 3 and phospho-diesterase 4 coordination of cell adhesion. Cell Signal 19:2507–2518.

16. Dodge-Kafka KL, et al. (2005) The protein kinase A anchoring protein mAKAP coor-dinates two integrated cAMP effector pathways. Nature 437:574–578.

17. Vo NK, Gettemy JM, Coghlan VM (1998) Identification of cGMP-dependent proteinkinase anchoring proteins. Biochem Biophys Res Commun 246:831–835.

18. Romorini S, et al. (2004) A functional role of postsynaptic density-95-guanylate kinase-associated protein complex in regulating Shank assembly and stability to synapses.J Neurosci 24:9391–9404.

19. Piggott LA, et al. (2006) Natriuretic peptides and nitric oxide stimulate cGMP synthesisin different cellular compartments. J Gen Physiol 128:3–14.

20. Dolci S, et al. (2006) Subcellular localization and regulation of type-1C and type 5phosphodiesterases. Biochem Biophys Res Commun 341:837–846.

21. Castro LR, Verde I, Cooper DM, Fischmeister R (2006) Cyclic guanosine monophosphatecompartmentation in rat cardiac myocytes. Circulation 113:2221–2228.

22. Schlossmann J, et al. (2000) Regulation of intracellular calcium by a signalling complexof IRAG, IP3 receptor and cGMP kinase Ibeta. Nature 404:197–201.

23. Antl M, et al. (2007) IRAG mediates NO/cGMP-dependent inhibition of platelet aggre-gation and thrombus formation. Blood 109:552–559.

24. Francis SH, Sekhar KR, Rouse AB, Grimes KA, Corbin JD (2003) Single step isolation ofsildenafil from commercially available Viagra tablets. Int J Impot Res 15:369–372.

25. Elbatarny HS, Maurice DH (2005) Leptin-mediated activation of human platelets:Involvement of a leptin receptor and phosphodiesterase 3A-containing cellular sig-naling complex. Am J Physiol 289:E695–E702.

26. Maurice DH, Haslam RJ (1990) Molecular basis of the synergistic inhibition of plateletfunction by nitrovasodilators and activators of adenylyl cyclase: Inhibition of cyclicAMP breakdown by cyclic GMP. Mol Pharmacol 37:671–681.

27. Haslam RJ, Dickinson NT, Jang EK (1999) Cyclic nucleotides and phosphodiesterases inplatelets. Thromb Haemost 82:412–423.

28. Jang EK, Azzam JE, Dickinson NT, Davidson MM, Haslam RJ (2002) Roles for both cyclicGMP and cyclic AMP in the inhibition of collagen-induced platelet aggregation bynitroprusside. Br J Haematol 117:664–675.

29. Dunkern TR, Hatzelmann A (2005) The effect of sildenafil on human platelet secretoryfunction is controlled by a complex interplay between phosphodiesterases 2, 3 and 5.Cell Signal 17:331–339.

30. Sargeant P, Sage SO (1994) Calcium signalling in platelets and other nonexcitable cells.Pharmacol Ther 64:395–443.

31. Corbin JD, Turko IV, Beasley A, Francis SH (2000) Phosphorylation of phosphodiester-ase-5 by cyclic nucleotide-dependent protein kinase alters its catalytic and allostericcGMP-binding activities. Eur J Biochem 267:2760–2767.

32. Rybalkin SD, Rybalkina IG, Feil R, Hofmann F, Beavo JA (2002) Regulation of cGMP-specific phosphodiesterase (PDE5) phosphorylation in smooth muscle cells. J Biol Chem277:3310–3317.

33. Rybalkin SD, Rybalkina IG, Shimizu-Albergine M, Tang XB, Beavo JA (2003) PDE5 isconverted to an activated state upon cGMP binding to the GAF A domain. EMBO J22:469–478.

34. Mullershausen F, et al. (2003) Direct activation of PDE5 by cGMP: Long term effectswithin NO/cGMP signaling. J Cell Biol 160:719–727.

Wilson et al. PNAS � September 9, 2008 � vol. 105 � no. 36 � 13655

PHA

RMA

COLO

GY

Dow

nloa

ded

by g

uest

on

July

26,

202

1