Stoichiometric and Reversible Phosphorylation of a 46-kDa Protein in Human Platelets in Response to cGMP-and CAMP-elevating Vasodilators*

cGMP- CAMP-dependent catalyze up to 1.4 mol of of

Royal Society, London, September 14-15 (abstr.)). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "adver- or CAMP (i.e. prostaglandin E1 (PGE,), prostacyclin) inhibit platelet aggregation at an unknown early point of the activation cascade (l-3). In intact platelets, stimulation of the phosphorylation of certain proteins by these vasodilators appears to be mediated by the activation of cGMP-and CAMPdependent protein kinase (cGK and cAK), respectively (3,4).
Our laboratory has characterized in some detail the phosphorylation of a 50-kDa membrane-associated protein from human platelets since this protein was phosphorylated in intact platelets in response to cGMP-and CAMP-elevating vasodilators and in platelet membranes by endogenous cGK and cAK (4). We recently purified this human platelet 50-kDa vasodilator-stimulated phosphoprotein (VASP) to apparent homogeneity (5). VASP was purified predominantly as a dephosphoprotein with an apparent molecular mass of 46 kDa by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), and phosphorylation shifted the apparent molecular mass of VASP from 46 to 50 kDa (5).
We now report the stoichiometry of phosphorylation of purified VASP by cGK and cAK, the production of a specific antiserum against VASP, and use of this antiserum for studying some kinetic and stoichiometric aspects of VASP phosphorylation in intact washed human platelets.

Phosphorylation
Experiments with Purified VASP-Recently, our laboratory established a procedure to purify the 46-kDa vasodilator-stimulated phosphoprotein from human platelets and also demonstrated that purified VASP is a substrate for purified cGK and cAK (5). We have now investigated the stoichiometry of VASP phosphorylation by cGK and the catalytic subunit of cAK. Under the conditions used (0.96 pM VASP, 0.51 pM protein kinase, and 18 I.LM ATP), both cGK and the catalytic subunit of cAK rapidly phosphorylated VASP, resulting in a maximum phosphate incorporation of about 1.2-1.4 mol/mol of VASP within 15 min (Fig. 1B and Fig. 2). In several separate experiments using two different VASP preparations, the maximum phosphate incorporation was 1.3 +-0.3 mol/mol of VASP using either cGK or cAK (Table I). This stoichiometry is based on the VASP protein concentration determined with the Bio-Rad protein assay system using catalase as standard.
If bovine serum albumin was used as standard in this protein assay system, ' Portions of this paper (including "Experimental Procedures," Figs. 2, 5, and 7-10, and    of the 46-kDa species of VASP was observed with only cGK, but not cAK (Fig. 1B).
The shift in the apparent molecular mass of VASP after phosphorylation was also detected by a radioimmunolabeling (Western blot) method using a rabbit antiserum against VASP (Fig. 1C). Both Coomassie Blue staining (Fig. 1A) and radioimmunolabeling ( Fig. 1C) also demonstrated that VASP was purified predominantly as a 46-kDa protein and that it could be completely converted to the 50-kDa protein by cGKor cAK-catalyzed phosphorylation. Regulation of VASP Phosphotylation by Cyclic Nucleotide-elevating Vasodilators and Cyclic Nucleotide Analogs-Certain properties of VASP (mobility change on SDS-PAGE after phosphorylation) and the availability of an antiserum which recognized both the phospho and dephospho forms of VASP enabled us to study the regulation of VASP phosphorylation in intact washed human platelets using a radioimmunolabeling procedure.
With total homogenates prepared from untreated human platelets, the antiserum prepared against purified VASP labeled primarily a 46-kDa protein and, to a much lesser degree, a 50-kDa protein (Fig. 3, lane 0 min). Longer film exposures of the blots shown in Fig. 3 also revealed labeling of a 80.kDa protein. This 80.kDa protein appears to be unrelated to VASP since its labeling disappeared from sera of later bleedings from the same rabbit, whereas strong labeling of the 46. and 50-kDa proteins persisted (data not shown). When platelet homogenates and purified VASP standards were analyzed on the same blot, the 46. and 50-kDa proteins labeled in platelet homogenates had the same mobility on SDS-PAGE as the 46-and 50-kDa species of purified VASP (data not shown).
Treatment of washed human platelets with 100 PM SNP or 10 @M PGE, caused a time-dependent increase of the 50.kDa protein and a reciprocal decrease of the 46-kDa protein (Figs. 3,4,and 6). Quantitation of the relative amounts of the 46and 50-kDa proteins was performed by counting the radioactivity bound to either the 46. or 50.kDa area of the blot and expressing this radioactivity as a percentage of the total radioactivity bound to both the 46-and 50-kDa blot areas. With the conditions used, radioactivity bound to both the 46. and 50.kDa blot areas did not vary by more than 15% within a set of blots and was proportional to the amount of platelet protein when the protein analyzed per gel lane was equivalent to 0.2-1.0 x 10' platelets (data not shown).
The time-and concentration-dependent effects of SNP on cyclic nucleotide levels and VASP phosphorylation are demonstrated in Figs. 4 and 5. Within 3-4 min, SNP converted up to 45-60% of VASP from the 46-kDa protein to the 50-kDa protein, and this conversion was preceded by an increase in platelet cGMP content from 2.2 to 20.7 pmol/lOg cells. These cGMP levels correspond to intracellular cGMP concentrations of 0.44 and 4.1 PM, respectively, assuming a mean platelet volume of about 5 fl (9) Although SNP effects on platelet cGMP levels and VASP phosphorylation did not appear to reach plateau levels with the SNP concentrations studied, it is of interest that small effects on cGMP levels and VASP phosphorylation were already detectable with 0.1-1.0 FM SNP (Fig. 5). PGE, (10 pM) converted up to 60% of VASP to the 50-kDa protein within l-2 min, and this conversion was preceded by a rapid 6.5-fold increase in platelet CAMP content from 21.8 (corresponding to 4.4 /*M) to 142 (corresponding to 28.4 PM) pmol/ lo9 cells, as demonstrated in Fig. 6. PGE, had very little, if any, effect on platelet cGMP content (Figs. 6 and '7). Effects on platelet CAMP content and VASP phosphorylation were detectable with 0.01-0.1 pM PGE, and were maximal with 1.0-10.0 FM PGE, (Fig. 7). Platelet cyclic nucleotide content and the level of the 50-kDa protein of VASP eventually slowly declined despite the continued presence of SNP or PGE, (Figs. 4 and 6).

Reversibility of Effects of Vasodilators on Platelet Cyclic
Nucleotide Content and VASP Phosphorylation-Removal of SNP from the platelet suspension by washing resulted in a rapid decline of the elevated cGMP content to near basal levels, followed by a reverse conversion of VASP from the 50-kDa protein to the 46-kDa protein, as shown in Fig. 9. Similarly, removal of PGEi from the platelet suspension caused the return of the elevated CAMP content to basal levels, followed by a complete reverse conversion of VASP from the 50-kDa protein to the 46-kDa protein (Fig. 10).

DISCUSSION
Recent purification (5) of a 46-kDa platelet protein (referred to as VASP) previously shown (4) to be phosphorylated in response to cGMP-and CAMP-elevating vasodilators made possible this study of certain kinetic and stoichiometric aspects of the phosphorylation of this protein in either its purified or endogenous state in intact human platelets. Both the catalytic subunit of cAK and cGK rapidly phosphorylated VASP, resulting in a maximum phosphate incorporation of 1.2-1.4 mol/mol of VASP ( Fig. 1 and Table 1). This phosphorylation altered the mobility of VASP on SDS-PAGE, shifting its apparent molecular mass from 46 to 50 kDa. We interpret these phosphate incorporation results and the finding that transient phosphorylation of the 46-kDa species of VASP is observed with cGK, but not with cAK (Fig. l), as evidence for the presence of two distinct phosphorylation sites in VASP. These two distinct sites are phosphorylated by both cAK and cGK, but cAK appears to phosphorylate the site responsible for the shift in apparent molecular mass more efficiently than the second site since transient phosphorylation of the 46-kDa species of VASP is absent or negligible with cAK. In contrast, cGK appears to phosphorylate these two sites with similar efficiency since transient phosphorylation of the 46-kDa species of VASP is observed. Other data published previously (4,5) suggest that the two distinct phosphorylation sites are both serine residues located in close proximity since tryptic fingerprinting and phosphoamino acid analysis of the 46-and 50-kDa species of VASP phosphorylated by either cGK or cAK only produced one major phosphopeptide and only phosphoserines. However, this point will have to be confirmed by microsequencing of the phosphorylation sites. Under our in vitro phosphorylation conditions (0.96 pM VASP, 0.51 pM cGK or cAK, and 18 ELM ATP), much lower concentrations of protein kinases and substrates were used than those present in intact human platelets. Recent studies" using specific antisera against cGK, the catalytic subunit of cAK, and VASP suggest an intracellular protein concentration of 3.5 PM catalytic subunit of cAK (assuming a molecular mass of 40 kDa), 6 fiM cGK (assuming a molecular mass of 74 kDa), and 8-16 PM VASP (assuming a molecular mass of 46 kDa), whereas the intracellular concentration of ATP is generally assumed to be in the millimolar range. The results indicating rapid and stoichiometric phosphorylation of VASP in vitro (Fig. 2) and the very high concentrations of protein kinases and their substrates in intact platelets suggest that intact human platelets have the capacity for a rapid and stoichiometric phosphorylation of VASP via activation of either cGK or cAK. To test this with intact platelets, an antiserum was prepared which recognized both the phospho (50 kDa) and dephospho (46 kDa) forms of VASP (Fig. 1C). The properties of this antiserum made possible an analysis of the state of phosphorylation of VASP by the Western blot method. A similar method has been used to study the phosphorylation of type II regulatory subunit of cAK in intact cells (8).
Initial studies demonstrated that purified VASP and VASP of untreated washed platelets consisted primarily of the 46-kDa protein (Figs. 1 and 3). For practical reasons, we use the term 50-kDa protein interchangeably with phospho-VASP and the term 46-kDa protein interchangeably with dephospho-VASP.
However Therefore, our study of VASP phosphorylation in intact platelets principally analyzes the phosphorylation/dephosphorylation of the site which is responsible for the shift in apparent molecular mass of VASP. SNP and PGEl were capable of producing a 65-70% conversion of VASP to the 50-kDa phospho form, a conversion which was preceded by an elevation of the intracellular concentration of either cGMP or CAMP, respectively . Since SNP and PGEl had no significant effects on the level of CAMP or cGMP, respectively, the phosphorylation of VASP induced by SNP appears to occur via activation of cGMP, and the phosphorylation induced by PGEl via activation of cAK.
It is of interest to note that whereas the intracellular concentration of cGMP under basal conditions (-0.44 PM) is more than an order of magnitude lower than that of the cGMP-binding sites of cGK (-12 pM, assuming two cGMPbinding sites/cGK monomer), the basal CAMP concentration  is of the same order of magnitude as the CAMPbinding site concentration of cAK (-7 pM, assuming two CAMP-binding sites/regulatory subunit and equimolar concentrations of the regulatory and catalytic subunits). This may explain the observation that the onset of VASP phosphorylation induced by SNP or 8-Br-cGMP is slower than that of VASP phosphorylation induced by PGE, or 6,and 8). Presumably, a sufficient intracellular concentration of cGMP or a cGMP analog has to accumulate in order to occupy at least some of the available free cGMPbinding sites on intracellular cGMP-binding proteins (cGK and cGMP-regulated phosphodiesterases) before activation of cGK is observed. In contrast, relatively small changes in the CAMP level may already be sufficient for full activation of cAK. Therefore, it should also be considered that SNP may cause VASP phosphorylation in intact human platelets via activation of cAK. SNP-increased cGMP could inactivate cGMP-inhibited CAMP phosphodiesterase (the major low K, CAMP phosphodiesterase in human platelets (lo)), leading to elevated CAMP levels and activation of cAK. However, we consider this possibility unlikely in our experiments since 1) SNP had no significant effects on platelet CAMP concentration (Figs. 4 and 5), and 2) the pattern of protein phosphorylation induced by SNP or 8-Br-cGMP in intact human platelets was quite different than that induced by PGEl or CAMP analogs (4). The considerably lower lipophilic properties of 8bromo derivatives of cyclic nucleotides, compared to 8-chlorophenylthio derivatives (ll), certainly contribute to the relatively slow onset of 8-Br-cGMP effects also. The extent of VASP phosphorylation induced by SNP or PGEl may also be limited by the stability of these compounds in solution since the stable prostacyclin analog I loprost was capable of shifting more than 90% of VASP to the 50-kDa phospho form (data not shown).
Removal of SNP or PGEl from the platelet suspension resulted in a prompt return of the cGMP or CAMP concentration, respectively, to near basal levels, followed by a conversion of 50-kDa phospho-VASP to 46-kDa dephospho-VASP ( Figs. 9 and 10). These results demonstrate that intact human platelets have not only a significant capacity for cyclic nucleotide synthesis and cyclic nucleotide-dependent protein phosphorylation, but also powerful activities catalyzing the degradation of cGMP and CAMP and the dephosphorylation of proteins such as VASP.
In conclusion, we interpret these data as evidence for near stoichiometric and reversible phosphorylation of VASP which occurs in intact human platelets in response to cyclic nucleotide-elevating vasodilators via activation of either cGK or cAK. The reasonably good correlation between the concentra-3092 Vasodilator-stimulated Phosphoprotein in Human Platelets tion-and time-dependent effects of cyclic nucleotide-elevating vasodilators on platelet aggregation and VASP phosphorylation suggests that VASP phosphorylation may be an important component for the inhibitory effects of these vasodilators on platelet activation. This is also supported by the findings that CAMP-and cGMP-elevating vasodilators have synergistic effects with respect to both the inhibition of platelet activation (12, 13) and VASP phosphorylation.4 However, more than one pathway for the inhibition of platelet activation may exist (3). Vasodilators also appear to regulate the phosphorylation of other platelet proteins such as glycoprotein IbB (14), cGMP-inhibited CAMP phosphodiesterase (15), and a ras-related GTP-binding protein (16, 17) via activation of cAK. The precise mechanism for the inhibition of platelet aggregation by cyclic nucleotide-elevating vasodilators has not been determined, although it appears to occur at an early step of the activation cascade, possibly at the level of phospholipase C (1,3,(18)(19)(20). We hope that our ongoing biochemical and cell biological studies with human platelets will elucidate the function of CAMP-and cGMP-mediated phosphorylation of VASP and contribute to the understanding of the mechanism of action of cyclic nucleotide-elevating vasodilators.