Regulator of G-protein signaling 18 integrates activating and inhibitory signaling in platelets

In healthy vasculature, endothelial cells lining the blood vessels constantly produce and release prostacyclin (PGI₂) and nitric oxide (NO) into the vessel lumen. The interaction of endothelial factors with platelets plays a fundamental role in controlling hemostasis and in maintaining platelets in a resting state. Platelet inhibition by both PGI₂ and NO has been well established. The signaling pathways of both molecules result in an elevation of cyclic nucleotides that activate cyclic AMP-dependent protein kinase A (PKA) and cyclic GMP-dependent protein kinase I (PKGI). These in turn phosphorylate an unknown number of substrate proteins, resulting in reduced release of calcium ions (Ca²⁺) from intracellular stores and reduced activation of G-proteins such as Rap1, ultimately leading to a block of platelet adhesion, granule release, and aggregation. PKA and PKGI have overlapping substrate specificities, which may explain the synergistic role of the 2 pathways. Few substrates have been established in platelets, among them Rap1GAP2, a GTPase-activating protein (GAP) of the small G-protein Rap1, as we have shown previously.¹  Other substrates include vasodilator-stimulated phospho-protein (VASP), heat-shock protein 27 (HSP27), and LIM and SH3 domain protein (LASP), all of which regulate actin dynamics.^2,3  The IP₃-receptor and the IP₃-receptor-associated G-kinase substrate (IRAG) are the only PKA and/or PKGI substrates that have been shown to mediate cAMP/cGMP effects on intracellular Ca²⁺ release.^2-4  Limited data are available on the specific substrates and signaling events that translate PKA/G substrate phosphorylation into platelet inhibition.

Conversely, binding of thrombin, ADP, and thromboxane A2 to G-protein coupled receptors (GPCRs) leads to platelet activation. GPCR signaling is terminated by the family of regulator of G-protein signaling (RGS) proteins, the function of which has mainly been attributed to their ability to interact with G-α subunits of heterotrimeric G proteins. RGS proteins are GAPs that enhance the rate of GTP hydrolysis of G-α subunits. Thirty-seven RGS domain–containing proteins have been identified in humans,⁵  and in a recent transcriptome analysis, at least 16 were shown to be expressed in human platelets.⁶  The functional significance of RGS proteins in platelets has recently been validated in a mouse model expressing a mutant G-α-i2 (Gi2) that is unable to interact with RGS.⁷  RGS18 belongs to the subfamily of B/R4 RGS proteins that contain a conserved central RGS domain and a short N and C terminus. These N- and C-terminal regions may be crucial for the association with specific GPCRs but also with other signaling factors.^8,9  First identified in 2001, RGS18 is abundantly expressed in platelets and megakaryocytes.^10-13  Interaction partners include G-α-q (Gq) and Gi1,2,3, but not Gz, Gs, or G12.^10,11  Platelet activation by GPCR ligands requires both Gq and Gi signaling.¹⁴  Gq activates PLC-β, leading to the production of inositol-1,4,5-trisphosphate (IP₃) and the release of Ca²⁺ from intracellular stores. Gi inhibits adenylyl cyclase and activates PI3K pathways,¹⁵  and RGS18 may be involved in the regulation of these pathways. To date, no other interaction partners of RGS18 have been described.

In the present study, we show that RGS18 is a target of PKA- and PKGI-mediated platelet inhibition, as well as of GPCR-mediated platelet activation. We show that RGS18 interacts with the 14-3-3γ protein via phosphorylated residues S49 and S218. 14-3-3s are abundantly expressed dimeric proteins that operate as serine/threonine–binding modules with a wide range of functions.¹⁶  There are 7 highly conserved isoforms described in mammals, 5 of which are expressed in platelets.¹⁷  PKA- and PKGI-mediated phosphorylation of RGS18 on serine 216 (S216) has a dominant regulatory function, leading to detachment of 14-3-3γ, which renders RGS18 more efficient in its ability to inhibit Gq-mediated Ca²⁺ release.

A polyclonal Ab against RGS18 was developed using full-length recombinant GST-tagged human RGS18 purified from Escherichia coli BL21 cells. An Ab against phosphorylated S216 of RGS18 was produced using a phosphorylated peptide conjugated to keyhole limpet hemocyanin. Immunization of rabbits and subsequent Ab purification were performed by ImmunoGlobe Antikörpertechnik (Himmelstadt, Germany). The Ab against phosphorylated S7 of Rap1GAP2 has been described previously.¹⁸  The following commercially available Abs were used in this study: mouse anti-RGS18 (1H6; Sigma-Aldrich), rabbit pan 14-3-3 (K19; Santa Cruz Biotechnology), mouse 14-3-3γ (3F8; Abcam), mouse 14-3-3β (H8; Santa Cruz Biotechnology), mouse 14-3-3ϵ (8C3; Santa Cruz Biotechnology), mouse 14-3-3ζ (Abcam), FLAG tag (M2; Sigma-Aldrich), Myc tag (9E10; Santa Cruz Biotechnology), and mouse anti-GST (clone GST-2; Sigma-Aldrich). HRP-coupled donkey anti–rabbit and donkey anti–mouse Abs were from Jackson ImmunoResearch Europe and were used as secondary Abs for immunoblot analysis visualized by enhanced chemiluminescence (Thermo Scientific). All statistical analyses were performed using Prism Version 5 software (GraphPad).

Full-length human RGS18 cDNA was obtained from the Missouri S&T cDNA Resource Center (Rolla, MO). RGS18 was FLAG-tagged at the N-terminus and expressed using mammalian expression vector pcDNA4/TO (Invitrogen). mCherry-tagged RGS18 was obtained through subcloning into mammalian expression vector pmCherryC1 (Invitrogen), and the GST fusion protein was generated using pGEX-4T3 vector (GE Healthcare). All RGS18 constructs were subcloned using XhoI/BamHI sites of their respective vectors. Site-directed mutagenesis was performed by PCR amplification using mutagenic primer pairs, Pfu DNA polymerase (Fermentas), digestion with DpnI (Fermentas), and transformation into TOP10 bacteria (Invitrogen). Human 14-3-3γ cDNA was a kind gift of James McRedmond (Conway Institute, University College Dublin, Dublin, Ireland). 14-3-3γ was myc-tagged and cloned into mammalian expression vector pcDNA4TO via BamHI and XhoI. For GST-tagged 14-3-3γ, 14-3-3γ was cloned into pGEX-4T3 using EcoRI and XhoI restriction sites. Constructs were verified by DNA sequencing.

GST-14-3-3γ, GST-wt-RGS18, and mutants were expressed in E coli BL21 and affinity purified using glutathione-Sepharose 4B beads (GE Healthcare). The purity of all proteins was examined by SDS-PAGE followed by Coomassie blue staining.

Recombinant GST-RGS18 fusion proteins attached to GSH beads were incubated with the purified C-subunit of PKA (New England Biolabs) in a buffer containing 40μM ATP and 1 μCi (γ-³²P) ATP (PerkinElmer) at 30°C for 3 minutes. Reactions were stopped by the addition of SDS-sample buffer, followed by SDS-PAGE, blotting, and exposure of the membrane to film. Equal loading was verified by immunoblotting using anti-GST Ab. For phosphate-binding tag (Phos-tag; Wako Chemicals)–supplemented SDS-PAGE, gels were prepared as described in the manufacturer's protocol. Briefly, 50μM Phos-tag and 100μM MnCl₂ (Sigma-Aldrich) were added to the separating gel solution before polymerization. To remove Mn²⁺ ions before Western blotting, gels were incubated in transfer buffer containing 1mM EDTA for 10 minutes at room temperature.

HEK293T cells were cultured using DMEM supplemented with 10% FCS and 1% penicillin/streptomycin at 37°C and 5% CO₂. Cells were transfected using Metafectene (Biontex) or Fugene (Promega) according to manufacturer's instructions. Venous blood was drawn from healthy volunteers taking no medications who gave their informed consent according to the Declaration of Helsinki. Platelet isolation was performed as described previously.¹⁹  Platelets were stimulated at 37°C using 0.1 U/mL of thrombin (Roche) for 30 seconds, U-46619 (Cayman Europe) as a thromboxane A2 mimetic at 1μM for 1 minute, ADP (Sigma-Aldrich) at 10μM for 1 minute, forskolin at 10μM for 30 minutes, prostacyclin (Cayman Europe) at 1μM for 5 minutes, sodium nitroprusside (SNP) at 10μM for 10 minutes, the cAMP analog 5,6-di-chlorobenzimidazole riboside-3,5-cyclic monophosphorothioate, Sp-isomer (Biolog) at 0.3mM for 30 minutes or GMP analog 8-(4-chlorophenylthio)guanosine-3,5-cyclic monophosphate (Biolog) at 0.3mM for 20 minutes.

Cell lysis, immunoprecipitation, and pull-down assays were performed as indicated using 5 μL of anti-FLAG M2 Affinity Gel (Sigma-Aldrich), 20 μL of anti-RGS18 polyclonal Ab, or 5 μL of a glutathione Sepharose 4B suspension (GE Healthcare) saturated with GST or GST-14-3-3γ, respectively. For peptide competition assays, lysates were supplemented with 100μM of either TNLRRRSRSFTVN or TNLRRRSR(pS)FTVN acetylated peptides (Schafer-N) before pull-down assays. For dephosphorylation experiments, samples were treated with λ protein phosphatase as described in the manufacturer's protocol (New England Biolabs).

For mass spectrometry (MS) analysis, samples were subjected to SDS-PAGE and subsequently to Coomassie blue staining of gels. Protein bands of interest were excised and subjected to in-gel digestion, as described previously.²⁰  Briefly, gel bands were washed and cysteine residues were reduced and alkylated using dithiothreitol and iodoacetamide. Samples were digested with trypsin and run on a Thermo Scientific LTQ Orbitrap XL mass spectrometer connected to an Eksigent Nano LC.1DPLUS chromatography system. A high-resolution MS scan was performed using the Orbitrap to select the 5 most intense ions before MS/MS analysis using the ion trap. The MS/MS spectra were searched against the UniProt/SwissProt human database (January 2009) using Bioworks Browser 3.3.1 SP1, a proteomics analysis platform. All MS/MS spectra were sequence-database searched using the algorithm TurboSEQUEST. The following search parameters were used: “precursor-ion mass tolerance of < 10 ppm,” “fragment ion tolerance of 0.5 Da,” and “fully tryptic peptide termini,” with “cysteine carboxyamidomethylation” and “methionine oxidation” specified as variable modifications and a maximum of 3 missed cleavages. Protein identifications were accepted if they could be established at greater than 99.0% probability as validated by the Protein Prophet algorithm²¹  and contained at least 2 identified peptides.

HEK293T cells were transfected as described in “Cell preparation, transfection, lysis, immunoprecipitation, and pull-down experiments” using empty pmCherry vector or mCherry-RGS18 constructs. Cells were lifted by incubation in PBS at room temperature for 2 minutes, resuspended in DMEM supplemented with 10% FCS, and subsequently stained with 0.25μM Fluo-4 (Invitrogen) at 37°C for 25 minutes. The cells were then washed twice with serum-free medium. Flow cytometry and Ca²⁺ measurements were performed using an Accuri C6 flow cytometer, as described previously.²²  Fluo-4– and mCherry-positive cells were identified by excitation with a 488-nm laser. The fluorescence was collected using 530 ± 15 (Fluo-4) and 675LP (mCherry) filters. Single cells showing both signals were selected for monitoring intracellular Ca²⁺ levels after stimulation with 0.1 U/mL of thrombin (Roche). Cells were stimulated after 1 minute in the cytometer and monitored for a total of 3 minutes. Changes in fluorescence, reflecting increases in intracellular Ca²⁺ concentration, were monitored over time. Samples were run as triplicates. Data from each sample were normalized to Fluo-4 loading according to F/F0, with F being fluorescence collected with the 530 ± 15 band-pass filter at any given time point and F0 being the fluorescence collected with the same filter before the addition of thrombin. Data from each experiment were further normalized to the maximum relative fluorescence after thrombin treatment of control (mCherry) transfected cells to account for variations in thrombin sensitivity of the cells during separate experiments.

In previous studies, we generated a phospho-specific Ab against the PKA/PKG phosphorylation site (pS7) of Rap1GAP2.¹⁸  In platelets with elevated cGMP- or cAMP levels, we observed several bands in addition to Rap1GAP2. For example, treatment with the NO-donor SNP induced the phosphorylation of Rap1GAP2, as described previously.¹⁸  Three prominent bands appeared on this blot in addition to Rap1GAP2 at approximately 65, 50, and 30 kDa (Figure 1A). Interestingly, these proteins were not detected in resting platelets and followed similar time dependence as the Rap1GAP2 phosphorylation. We were able to identify the band at 50 kDa as VASP (data not shown) and established the Ab-recognition site as its PKA/G phosphorylation site pS239.²³  Therefore, we hypothesized that the additional bands detected might represent genuine PKA and PKGI substrates. Using 2 independent purification approaches, immunoprecipitation with pS7 Ab and lysate fractionation followed by MS, we were able to identify the band at 30 kDa as the RGS18 protein. To determine whether RGS18 is indeed phosphorylated by PKA and recognized by the pS7-Rap1GAP2 specific Ab, cDNA of human RGS18 was cloned into a bacterial expression vector and purified GST-RGS18 was incubated with the catalytic subunit of PKA in the presence or absence of ATP. The pS7 Ab indeed recognized the PKA-phosphorylated form of RGS18 (Figure 1B).

We next wanted to identify the exact site on RGS18 that is phosphorylated by PKA and PKGI. Comparing the amino acid sequences of RGS18 and Rap1GAP2 revealed a region close to the C-terminus of RGS18 with high similarity to the phosphorylated peptide against which the pS7-Rap1GAP2 Ab was raised (Figure 1C). In addition, the scansite program (http://scansite.mit.edu/)²⁴  predicted S216 on RGS18 as a putative PKA phosphorylation site. To test this prediction, we generated point mutants of RGS18 with S216 mutated to alanine. As a control, we mutated the neighboring S218 residue to alanine. In vitro phosphorylation experiments using purified proteins and isotope labeling confirmed that S216 is indeed the main PKA phosphorylation site on RGS18 (Figure 1D). Next, we generated a phosphospecific Ab against pS216 of RGS18. In vitro phosphorylation using GST-wt-RGS18, S216A, and S218A mutants confirmed the specificity of this Ab (Figure 1E). Treatment of washed platelets with PGI₂, forskolin, SNP, a cAMP analog, or a cGMP analog led to S216 phosphorylation of RGS18 (Figure 1F). In contrast to this, thrombin, ADP, or U46619, a thromboxane mimetic, did not alter the phosphorylation state of S216 (Figure 1F). From these data, we conclude that PKA and PKGI phosphorylate endogenous RGS18 on S216 in human platelets.

ClustalW2 multiple sequence alignment²⁵  suggests that the S216 PKA phosphorylation site is embedded within a short, highly conserved stretch of 7 amino acids (RRRSRSF; Figure 2A). Furthermore, the second serine residue in this sequence (S218) is predicted to be a mode I 14-3-3–binding site.²⁴  14-3-3 proteins are small, phosphoserine/threonine-binding proteins that function as scaffolds and regulate key signaling components.²⁶  To assess whether RGS18 can bind 14-3-3 at endogenous level in human platelets, we performed immunoprecipitations from human platelet lysate using a rabbit Ab generated against GST-RGS18 and immunoblotted with mouse Abs specific for the γ, β, ζ, and ϵ isoforms (Figure 2B and data not shown), which are believed to be the most abundantly expressed in platelets.^17,27  Only 14-3-3γ was observed to interact consistently with RGS18 in thrombin-stimulated human platelets (Figure 2B lane 4). Control immunoprecipitation from thrombin-treated platelets showed the absence of 14-3-3γ (Figure 2B lane 1).

To identify the 14-3-3–binding site in RGS18, we expressed FLAG-wt-RGS18 and mutants in HEK293T cells and performed 14-3-3γ pull-down assays. Both wt-RGS18 and S216A-RGS18 strongly bound to GST-14-3-3γ, whereas no binding was detected in GST-only control experiments (Figure 2C). As predicted, mutation of S218 to alanine in RGS18 decreased 14-3-3 binding. Remarkably, mutation of S216 to glutamate, which mimics PKA/G phosphorylation, dramatically decreased binding of RGS18 to 14-3-3γ. To examine whether the interaction of RGS18 with 14-3-3γ is phospho-serine dependent, we performed GST-14-3-3γ pull-down assays from human platelets and subsequently treated samples with λ protein phosphatase. In contrast to the co-immunoprecipitation experiment (Figure 2B), an interaction between GST-14-3-3 and endogenous RGS18 was observed in the absence of thrombin (Figure 2D). This is most likely because of the high sensitivity of GST pull-down assays, which use an excess of GST-fusion protein to probe for binding. In phosphatase-treated samples, the interaction of 14-3-3γ and RGS18 was strongly reduced (Figure 2D). In addition, we performed GST-14-3-3γ pull-down assays from human platelets in the presence of peptides mimicking the phosphorylated or nonphosphorylated S218-binding site on RGS18. Only the phosphorylated peptide reduced the association of RGS18 and 14-3-3γ in pull-down assays (Figure 2E). These data confirm that 14-3-3γ interacts with RGS18 in a phosphorylation-dependent manner and that RGS18 exhibits a basal level of phosphorylation in resting platelets.

To further analyze the regulation of interaction between 14-3-3γ and RGS18, we performed co-immunoprecipitation studies in transfected cells. As observed in pull-down experiments, wt-RGS18 and S216A mutant were strongly associated with 14-3-3γ, whereas mutation of S216 to glutamate again strongly decreased binding (Figure 3A S216E). We did not study mutations of S218 to glutamate or aspartate because these mutations have been well established as not being able to mimic phosphorylated 14-3-3–binding sites.²⁸  Mutation of S218 to alanine greatly decreased binding of 14-3-3 to RGS18; however, there was some residual binding detected, which prompted us to search for a potential secondary 14-3-3–binding site on RGS18. In RGS18, only the sequence surrounding S49 (KRNRLpSLL) displayed convincing evidence as a second potential 14-3-3 site, because amino acids at positions pS −5, −4, −2, and +1 match residues most commonly found in 14-3-3 motifs.²⁸  Furthermore, S49 is highly conserved across species and, most importantly, S49 has already been described as being phosphorylated endogenously in human platelets in response to thrombin receptor–activating peptide stimulation.²⁹  In co-immunoprecipitation studies, S49A mutation alone had a minor but significant effect on 14-3-3γ binding to RGS18 (Figure 3B densitometry from 7 independent experiments). These data indicate that RGS18 contains two 14-3-3–binding sites, one at pS218 and a second one at pS49. To examine the phosphorylation of RGS18 on S218 and S49, we used Phos-tag–supplemented gels.³⁰  Phos-tag, in combination with Mn²⁺, displays preferential trapping of phosphorylated proteins in SDS-PAGE. This slows down the running velocity of phosphorylated proteins compared with their dephosphorylated counterparts: that is, phosphorylated proteins tend to appear as higher–molecular weight bands in Western blots.³⁰  To obtain the basal, dephosphorylated state, wt-RGS18 expressing HEK293T lysate was treated with λ phosphatase and then Phos-tag SDS-PAGE and Western blotting were performed. Phosphatase treatment resulted in the appearance of a low–molecular weight band of RGS18 (Figure 3C lane 1 b). In contrast, 3 distinct bands of higher apparent molecular weight appeared in untreated wt-RGS18 samples, suggesting phosphorylation-induced shifts (Figure 3C lane 2 s1, s2, and s3). To investigate whether any of these shifts was caused by phosphorylation of the newly identified RGS18 phosphorylation sites, we analyzed lysates from S49A-, S216A-, S218A-, and S49A/S218A mutant-expressing cells. Mutation of S49 resulted in selective loss of the s3 band, whereas mutation of S216 resulted in selective loss of the s2 band (Figure 3C). Mutation of S218 induced a loss of s2 and s3 and the appearance of the dephosphorylated form b, whereas additional mutation of S49 did not result in a different pattern (Figure 3C). For an unequivocal interpretation of these patterns, it will be necessary to develop phosphorylation site–specific Abs against pS49 and the combination of pS216-pS218. However, 3 conclusions can be drawn from our results: (1) S49 phosphorylation contributes to the s3 shift, (2) S216 phosphorylation is involved in the s2 shift, and (3) S218 phosphorylation contributes to the s1 shift of RGS18. Most likely, the s3 band represents a combination of phosphorylated S49 and S218. These data confirm that S49 and S218 are constitutively phosphorylated in HEK293T cells, thus mediating binding of 14-3-3 to RGS18 (Figure 3A-B).

We next examined lysates of human platelets by Phos-tag SDS-PAGE and Western blotting. Platelets were stimulated with either forskolin or thrombin before lysis. An aliquot of nonstimulated platelet lysate was treated with λ phosphatase. Analogous to HEK293T cells, a low/basal band appeared (Figure 3D lane 1 long exposure b). Nonstimulated samples appeared at higher molecular weights relative to phosphatase-treated samples (Figure 3D lane 2 s1), indicating a phosphorylation-induced shift. Forskolin treatment of platelets resulted in the appearance of an s2 band (Figure 3D lane 3 s2). Thrombin treatment induced an s3 band of the highest apparent molecular weight (Figure 3D lane 4 s3). Together with the data obtained using RGS18 mutants in HEK293T cells (Figure 3C), we conclude that: (1) RGS18 is constitutively phosphorylated on S218 in platelets (the s1 band); (2) PKA activation leads to the phosphorylation of S216 (the s2 band), which confirms previous data shown in Figure 1F; and (3) thrombin treatment results in the phosphorylation of S49 of RGS18 (s3 band), which results in enhanced binding of 14-3-3 to RGS18 (Figures 2B and 4A). We conclude that phosphorylated S218 and S49 are the main 14-3-3–binding sites of RGS18 in platelets.

Because the glutamate mutation, mimicking phosphorylation of S216, resulted in decreased binding of RGS18 to 14-3-3γ in pull-down assays and in co-immunoprecipitation studies in HEK293T cells, we sought to determine the role of S216 phosphorylation for 14-3-3γ binding in human platelets. As shown in Figure 1F, S216 on RGS18 can be phosphorylated by stimulation of platelets with PGI₂, forskolin, SNP, cAMP analog, or cGMP analog, but not with the GPCR agonists thrombin, ADP, and U46619. We therefore treated platelets with forskolin and SNP to induce S216 phosphorylation (Figure 4A-C middle panels) and performed GST-14-3-3γ pull-down experiments. Indeed, forskolin and SNP strongly reduced binding of RGS18 to 14-3-3γ in human platelets (Figure 4A-C). In contrast, thrombin, ADP, or U46619 potently stimulated the association of RGS18 and 14-3-3γ. When platelets were treated with forskolin or SNP before thrombin, ADP, or U46619 stimulation, 14-3-3γ binding was reduced (Figure 4A-C). Reduced 14-3-3γ binding was correlated with increased phosphorylation of RGS18 on S216. Densitometric analysis of data from repeat experiments confirmed that the effects of forskolin and SNP on basal and thrombin-, U46619-, or ADP-stimulated binding of 14-3-3γ to RGS18 were significant (Figure 4D-F). These data, together with the 14-3-3–binding studies of RGS18 mutants (Figure 3A-B) and Phos-tag analyses (Figure 3C-D), suggest that the increased interaction of RGS18 and 14-3-3γ occurs because of increased phosphorylation of S49. These data also suggest that PKA/PKG-mediated S216 phosphorylation has a dominant inhibitory effect on the interaction of RGS18 and 14-3-3γ.

We next assessed the functional consequences of the RGS18-14-3-3γ interaction on downstream signaling events. RGS18 has been shown previously to activate the GTPase function of Gi and Gq.^10-13  Gq signaling results in the release of Ca²⁺ from intracellular stores. To examine RGS18 effects on Gq signaling, we transfected pmCherry-tagged RGS18 constructs into HEK293T cells. We chose 2 different RGS18 constructs, one containing a point mutation of S216 to alanine, which binds 14-3-3 constitutively (Figure 3A-B), and one in which 14-3-3 binding cannot be inhibited by phosphorylation of S216 (Figure 4). The second construct had both 14-3-3–binding sites mutated to alanine (S49A/S218A), corresponding to 14-3-3–deficient RGS18 (Figure 3A-B). Both RGS18 constructs were able to bind Gq in co-immunoprecipitation studies (data not shown). After transfection, cells were stained with Fluo-4, a membrane-permeable dye that greatly increases its fluorescence after Ca²⁺ binding. Gq-mediated release of Ca²⁺ from intracellular stores was triggered using thrombin. Thrombin is known to activate Gq-mediated Ca²⁺ release in platelets via PAR1 receptors,³¹  and PAR1-dependent intracellular Ca²⁺ release in HEK293 cells has been described previously.³²  The thrombin-induced increase in Fluo-4 fluorescence was monitored by flow cytometry based on the method of Vines et al.²²  Only cells that displayed mCherry transfection were considered. As expected, both forms of RGS18 were able to inhibit Gq-mediated Ca²⁺ release relative to control mCherry-transfected cells. Remarkably, RGS18 deficient in 14-3-3 (the S49A/S218A mutant) was significantly more efficient in inhibiting Gq-mediated Ca²⁺ release compared with 14-3-3–bound RGS18 (P < .01 by ANOVA and the Bonferroni posttest; Figure 5). These findings suggest that 14-3-3 attenuates the function of RGS18, whereas removal of 14-3-3 enables RGS18 to inhibit Gq-mediated Ca²⁺ signaling more efficiently.

The cyclic nucleotide signaling system is an essential regulator of platelet reactivity. Endothelial prostacyclin and nitric oxide trigger the synthesis of cAMP and cGMP in platelets, leading to activation of PKA and PKGI. The phosphorylation of specific PKA and PKGI substrates links cyclic nucleotides to inhibition of platelet function. Only few PKA and PKGI substrate proteins that may be capable of mediating these effects, among them Rap1GAP2, are known to date.¹⁸  In the present study, we describe RGS18 as new substrate of PKA and PKGI and establish functional outcomes of RGS18 phosphorylation at the molecular and cellular levels. Cross-reactivity of a phosphorylation site–specific Ab provided initial evidence for a new 30-kDa substrate of PKA and PKGI in platelets, and we were able to identify this protein by MS as RGS18. Mapping studies revealed S216 of RGS18 as the main PKA and PKGI phosphorylation site, and we monitored endogenous phosphorylation of S216 in response to activators of cAMP/cGMP signaling in intact platelets using a newly developed phospho-specific Ab. These experiments clearly establish RGS18 as new target of cAMP/cGMP pathways in platelets and provide another example of the convergence of PKA and PKGI signaling in platelets. The phosphorylated S216 residue is embedded in a highly conserved, RGS18-specific region of 7 amino acids close to the C-terminus, which we show to be involved in binding of 14-3-3γ to RGS18. We confirmed the interaction of RGS18 and 14-3-3γ at the endogenous level in human platelets and in transfected cells and mapped phosphorylated S218 on RGS18 as the primary 14-3-3–binding site. In addition, we identified phosphorylated S49 as a secondary 14-3-3–binding site. Both 14-3-3–binding sites are localized outside of the catalytic RGS domain of RGS18. This is in agreement with other examples of 14-3-3 interactions in which the 14-3-3 dimer straddles either side of a folded functional domain.²⁸  Treatment of platelets with activatory GPCR ligands such as thrombin, ADP, and thromboxane analog further increases binding of 14-3-3 to RGS18, which is most likely mediated by additional phosphorylation of the S49-binding site as indicated by Phos-tag Western blotting (Figures 3C-D and 6). This is supported by the findings of Garcia et al,²⁹  who observed that PAR-1 receptor signaling induces the phosphorylation of S49 of RGS18. Therefore, pS49 might represent the gatekeeper 14-3-3–binding site of RGS18, which does not have a high affinity for 14-3-3 in itself, but significantly augments 14-3-3 binding in the presence of pS218.^33,34  The identity of the kinase(s) that phosphorylates S49 and S218 remains to be determined.

PKA and PKGI phosphorylation of S216 on RGS18 has a dominant-negative effect on binding of 14-3-3 to RGS18 in human platelets. Both basal and thrombin-, thromboxane A2-, or ADP-stimulated binding of 14-3-3 to RGS18 are abolished by S216 phosphorylation. As to the direct effect of S216 phosphorylation on 14-3-3 binding, 2 mechanisms can be envisioned: (1) that pS216 interferes directly with the interaction of 14-3-3 and RGS18 or (2) that pS216 induces the dephosphorylation of pS49 or pS218 or both. Preliminary evidence using Phos-tag gels suggests that S216 phosphorylation might induce some decrease in S49 and S218 phosphorylation (data not shown), although the exact consequences of S216 phosphorylation will need to be determined in future studies. PKA/PKG-induced detachment of 14-3-3 from its binding partner has been described previously for the GTPase-activating protein Rap1GAP2 in platelets. PKA- and PKGI-mediated phosphorylation of S7 of Rap1GAP2 triggered the removal of 14-3-3 from its binding site at phosphorylated S9, whereas thrombin and ADP treatment enhanced the phosphorylation of S9 and the association of Rap1GAP2 and 14-3-3.¹⁸  Another example of the inhibition of 14-3-3 binding by the phosphorylation of the −2 serine relative to the actual 14-3-3–binding site has been described for Cdc25B³⁵  and Cdc25C³⁶  involving CdkI and Cdc2 kinases.^35,36 

To understand the consequences of 14-3-3 binding for RGS18 function, we investigated downstream signaling events. In initial studies, we verified the interaction of RGS18 with Gq and Gi (data not shown) and confirmed that RGS18 blocks the Gq-mediated release of Ca²⁺ from intracellular stores. Removal of 14-3-3 from RGS18 significantly potentiated RGS18-mediated inhibition of Ca²⁺ release, suggesting increased turnover of Gq-GTP to inactive Gq-GDP. The exact mechanism of inhibition of RGS18 by 14-3-3 remains to be determined. Studies of other RGS family proteins have shown that 14-3-3 binds to RGS3, RGS4, RGS5, RGS7, RGS9-2, and RGS16.^37-41  The general consequences of 14-3-3 binding are controversial: either no significant effects or inhibition of RGS activities have been observed.^37-40  14-3-3 might interfere with the binding of RGS proteins to their Gα targets, and recent data on RGS3 suggest that 14-3-3 binding induces a conformational change in RGS3 that is likely to affect the interaction of RGS3 and Gα.^42,43  Our studies on the interaction between RGS18 and Gi2 and Gq by co-immunoprecipitation from transfected cells showed a trend toward increased binding of the 14-3-3–deficient RGS18 mutant to Gi2 and Gq compared with wt-RGS18; however, this effect was not statistically significant (data not shown). Live microscopy of fluorescently labeled wt-RGS18 and 14-3-3–deficient mutants in transfected cells did not reveal any major difference in subcellular distribution (data not shown). Because RGS proteins are increasingly recognized as having multiple binding partners,⁴⁴  14-3-3 could potentially regulate RGS18 by interfering with the binding of other currently unknown RGS18-interacting proteins. The PKA and PKGI substrate Rap1GAP2 is regulated in a manner similar to RGS18. Binding of 14-3-3 keeps Rap1GAP2 in an inactive state. Removal of 14-3-3 from Rap1GAP2 by PKA- and PKGI-mediated phosphorylation reduces cell adhesion, implicating increased GAP activity.¹⁸  Negative regulation of 14-3-3 binding might represent a common theme of cAMP/cGMP function in platelets that needs to be investigated in other PKA and PKGI substrates.

In summary, the results of the present study suggest that phosphorylation of RGS18 on S216 by PKA and PKGI effects the detachment of 14-3-3, resulting in a potentiation of RGS function (Figure 6). Consequently, Gq signaling is attenuated, leading to reduced receptor-mediated release of Ca²⁺ from intracellular stores. This mechanism might contribute to the inhibitory actions of cAMP and cGMP in platelets. Furthermore, phosphorylation of RGS18 on S49 during platelet activation stimulates the interaction of RGS18 and 14-3-3, resulting in reduced RGS18 function. Therefore, Gq signaling can be maintained during platelet activation, leading to enhanced release of Ca²⁺ from intracellular stores. The regulated interaction between RGS18 and 14-3-3 might represent a switch in the control of calcium signaling in platelets.

Access to and use of the University College Dublin Conway Institute Mass Spectrometry Resource instrumentation is gratefully acknowledged.

This work was supported by a Principal Investigator Program grant from Science Foundation Ireland (08/IN.1/B1855 to A.S.) and by the International Society for Advancement of Cytometry Scholar program.

Contribution: All authors performed the experiments and analyzed the data; and K.G. and A.S. designed the research and wrote the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Albert Smolenski, UCD Conway Institute, UCD School of Medicine and Medical Science, University College Dublin, Belfield, Dublin 4, Ireland; e-mail: albert.smolenski@ucd.ie.