Heat Shock Protein 27 Is a Substrate of cGMP-dependent Protein Kinase in Intact Human Platelets

Phosphorylation of heat shock protein 27 (Hsp27) in human platelets by mitogen-activated protein kinase-activated protein kinase (MAPKAP) 2 is associated with signaling events involved in platelet aggregation and regulation of microfilament organization. We now show that Hsp27 is also phosphorylated by cGMP-dependent protein kinase (cGK), a signaling system important for the inhibition of platelet aggregation. Stimulation of washed platelets with 8-para-chlorophenylthio-cGMP, a cGK specific activator, resulted in a time-dependent phosphorylation of Hsp27. This is supported by the ability of cGK to phosphorylate Hsp27 in vitro to an extent comparable with the cGK-mediated phosphorylation of its established substrate vasodilator-stimulated phosphoprotein. Studies with Hsp27 mutants identified threonine 143 as a yet uncharacterized phosphorylation site in Hsp27 specifically targeted by cGK. To test the hypothesis that cGK could inhibit platelet aggregation by phosphorylating Hsp27 and interfering with the MAPKAP kinase phosphorylation of Hsp27, the known MAPKAP kinase 2-phosphorylation sites (Ser15, Ser78, and Ser82) as well as Thr143 were replaced by negatively charged amino acids, which are considered to mimic phosphate groups, and tested in actin polymerization experiments. Mimicry at the MAPKAP kinase 2 phosphorylation sites led to mutants with a stimulating effect on actin polymerization. Mutation of the cGK-specific site Thr143alone had no effect on actin polymerization, but in the MAPKAP kinase 2 phosphorylation-mimicking mutant, this mutation reduced the stimulation of actin polymerization significantly. These data suggest that phosphorylation of Hsp27 and Hsp27-dependent regulation of actin microfilaments contribute to the inhibitory effects of cGK on platelet function.

The activation of human platelets and vessel wall-platelet interactions are processes tightly regulated under physiological conditions and often impaired in thrombosis, arteriosclerosis, hypertension, and diabetes. Agonists such as thrombin, thromboxane, vasopressin, and ADP activate platelets and cause shape change, aggregation, and degranulation. Platelet activation is inhibited by a variety of agents, including aspirin and Ca 2ϩ antagonists as well as cGMP-and cAMP-elevating agents such as NO and prostaglandin I 2 , respectively (for review, see Ref. 1). The inhibitory effects of cGMP and cAMP are principally mediated by cGMP-and cAMP-dependent protein kinases (cGK and cAK, respectively), with some cross-talk existing between the two systems. For example, cGMP stimulates the hydrolysis of cAMP via cGMP-regulated phosphodiesterases (2,3). The molecular mechanisms of platelet inhibition by cGMP signaling distal to cGK activation are only partially understood (4). Studies using cGK-deficient mice demonstrated defective cGMP-mediated inhibition of platelet aggregation (5). Several proteins have been reported to be phosphorylated in response to cGK activation either in vitro or in intact cells, including cGMP-specific phosphodiesterase (6), myosin light chain kinase (7), the inositol 1,4,5-trisphosphate receptor (8), an inositol 1,4,5-trisphosphate receptor-associated cGMP kinase substrate (9), G-substrate (10), Na ϩ /K ϩ -ATPase (11), and endothelial NO synthase (Ref. 12, for review, see Ref. 13). None of these proteins, however, could be established as a downstream effector of cGK in platelets. Recently, it was assumed that at least part of the inhibitory response mediated by cGK depends on the phosphorylation of the thromboxane receptor (14). These experiments, however, were performed using HEL cells. The only known in vivo substrates of cGK involved in platelet inhibition are the vasodilator-stimulated phosphoprotein VASP, associated with focal adhesion (15), and the small GTP-binding protein rap 1b (16).
To identify additional intracellular targets for cGK, we used two-dimensional gel electrophoresis of radiolabeled human platelets in combination with nano-electrospray ionization mass spectrometry (nano-ESI-MS). By applying this method, we identified heat shock protein 27 (Hsp27) 1 as a substrate of cGK I in intact platelets. Additionally, we suggest that phosphorylation of Hsp27 may contribute to the inhibitory actions of cGMP by regulating actin polymerization.
cGK I␣ and the catalytic subunit of cAK type II were purified from bovine lung and bovine heart, respectively (17). cGK I␤ and cGK II were expressed in and purified from the baculovirus-Sf9 cell system (18).
Isolation of Platelets-Freshly donated blood from healthy volunteers (50 ml) was collected in acid-citrate dextrose and centrifuged for 10 min at 300 ϫ g to yield platelet-rich plasma. Platelet-rich plasma was centrifuged for 20 min at 500 ϫ g and the pellet was resuspended and washed once in an isotonic buffer containing 10 mM Hepes (pH 7.4), 137 mM NaCl, 2.7 mM KCl, 5.5 mM glucose, and 1 mM EDTA at a density of 1 ϫ 10 9 cells/ml. After resuspension, platelets were allowed to rest at 37°C for 15 min. 32 P Labeling of Platelets-Platelet preparation was carried out essentially as described above. After washing, 1 ml of platelets at a concentration of 1 ϫ 10 9 /ml was incubated with 500 Ci of [ 32 P]orthophosphate (HCl-free) for 1.5 h at 37°C. Platelets were then centrifuged at 500 ϫ g for 7 min and resuspended in 1 ml of isotonic buffer. Aliquots of 100 l (corresponding to 200 g of protein) were used for activation with 500 M 8-pCPT-cGMP for 30 min at 37°C. After stimulation, platelets were briefly centrifuged (500 ϫ g) to yield a pellet.
Two-dimensional Gel Electrophoresis-Isoelectric focussing for twodimensional gel electrophoresis was performed using the Multiphor II system from Amersham Pharmacia Biotech (Uppsala, Sweden) according to the instructions of the manufacturer. The platelet pellet (about 200 g of protein) was solubilized for 15 min by sonication in 220 l of lysis buffer containing 7 M urea, 2 M thiourea, 4% (w/v) CHAPS, 15 mM dithiothreitol (electrophoresis grade), 0.5% carrier ampholytes, pH 3-10. Pellet homogenate was loaded on a 13-cm immobilized IPG strip, pH 3-10, using a reswelling cassette (custom-built). Focussing was carried out for 1 h at 150 V, 1 h at 600 V, and 25 h at 3500 V.
After equilibration in 50 mM Tris, pH 8.9, 6 M urea, 30% glycerol, 2% SDS, strips were immediately applied to a vertical 10% SDS gel without stacking gel. Electrophoresis was carried out at 8°C with a constant current of 30 mA per gel. The gels of radioactively labeled platelet proteins were fixed in 30% ethanol, 10% acetic acid and exposed. Radioactive spots were cut out, collected, and concentrated in a Pasteur pipette according to Gaevert et al. (19).
Mass Spectrometry-The concentrated gel piece was washed sequentially for 10 min in tryptic digestion buffer (10 mM NH 4 HCO 3 ) and digestion buffer:acetonitril, 1:1. These steps were repeated three times and led to a shrinking of the gel. It was reswollen with 2 l of protease solution (trypsin at 0.05 g/l) in digestion buffer and incubated overnight at 37°C. The supernatant was collected and dried down to 1 l.
Electrospray ionization mass spectrometry (ESI-MS) was carried out using a TSQ 7000 triple quadrupole mass spectrometer (Finnigan MAT, Bremen, Germany) equipped with a nanospray source of 0.6 to 1.1 kV constructed in-house. Mass spectra were acquired with a scan speed of 1000 Da/s. Argon at a pressure of 3 mTorr was used as collision gas. For the fully automated interpretation of fragment ion spectra, the SE-QUEST TM algorithm (version B22) was employed.
Western Blot Analysis of p38 and Hsp27-Washed, intact human platelets (100 l) at a concentration of 1 ϫ 10 9 cells/ml were incubated at 37°C by adding 2 units/ml thrombin for 2 min or by adding 500 M 8-pCPT-cGMP for the times indicated in the figures. After treatment, platelets were briefly centrifuged (500 ϫ g) to yield a pellet, which was immediately boiled in Laemmli SDS stop solution and separated by SDS-PAGE on a 10% gel. After blotting on polyvinylidene difluoride membrane and blocking with 3% nonfat dry milk in 10 mM Tris (pH 7.5), 100 mM NaCl, 0.1% Tween 20, the membrane was first incubated with a polyclonal antibody against dual phosphorylated p38 (1:500) followed by incubation with horseradish peroxidase-coupled goat anti-rabbit IgG (1:5000) and detection by ECL.
For Hsp27 detection, a two-dimensional SDS gel was blotted on nitrocellulose, blocked with 1% hemoglobin in phosphate-buffered saline, and incubated first with an anti-Hsp27 rabbit polyclonal antibody (1:1000) followed by incubation with horseradish peroxidase-coupled goat anti-rabbit IgG (1:5000) and detection by ECL.
In Vitro Phosphorylation of Hsp27-Hsp27 and its mutants S15D, S78D,S82D, S15D,S78D,S82D, S15D,S78D,S82D,T143E, and T143E (each 0.5 M) were incubated at 30°C in a total volume of 20 l with 10 mM Hepes (pH 7. Preparation of Pyrene Actin-G-actin was prepared from pig skeletal muscle according to the procedure of Pardee and Spudich (22). For labeling with N-(1-pyrenyl)iodoacetamide, G-actin was dialyzed 3 times for 12 h against G-buffer (2 mM Tris, pH 8.0, 0.2 mM ATP, 0.2 mM CaCl 2 ). To polymerize G-actin to F-actin, 100 mM KCl and 1 mM MgCl 2 was added for 1 h at room temperature. N-(1-Pyrenyl)iodoacetamide (at a 2-fold molar excess) was dissolved in dimethyl sulfoxide and added slowly with gentle stirring to the F-actin solution. The solution was kept at room temperature in the dark for 20 h. After labeling, F-actin was dialyzed 5 times against G-buffer with 0.5 mM dithioerythritol at 4°C to form G-actin. To remove any residual F-actin, the solution was centrifuged for 1 h at 100,000 ϫ g in a swing-bucket rotor and the supernatant was used in the polymerization experiments. The degree of labeling was determined by UV spectroscopy at 344 nm assuming an extinction coefficient of 2.2 ϫ 10 4 M Ϫ1 cm Ϫ1 , and was found to be 70 -80%.
Fluorometric Measurement of Actin Polymerization-For standard assays, pyrene-labeled G-actin in G-buffer at a final concentration of 2 M and various amounts of Hsp27 were mixed in a total volume of 500 l in a solution of 20 mM Tris (pH 7.6), 0.05 mM NaN 3 , 0.002 mM phenylmethylsulfonyl fluoride, 0.5 mM dithioerythritol, 10 mM MgCl 2 , 30 mM NH 4 Cl. The solutions were mixed with 1 l of 1 M MgCl 2 and 12.5 l of 2 M KCl to start actin polymerization. Polymerization was measured by the enhancement of pyrene-actin fluorescence using the luminescence spectrophotometer LS50 (PerkinElmer Life Sciences). Excitation was measured at 365 nm with a 2.5-mm slit width, and emission was detected at 407 nm with 2.5-mm slit width.

Phosphorylation of Hsp27 in Intact Human Platelets Treated
with the cGK-specific Stimulus 8-pCPT-cGMP-To identify substrates of cGK in intact human platelets, cells were labeled with [ 32 P]orthophosphate, stimulated with 500 M of the specific cGMP-dependent protein kinase activator 8-pCPT-cGMP, and proteins of the resulting platelet lysate were separated by two-dimensional gel electrophoresis. Fig. 1 shows low basal phosphorylation of three proteins with an approximate molecular mass of 27 kDa in resting platelets (control). Phosphorylation of the two more acidic protein spots was significantly increased after stimulation with 8-pCPT-cGMP. To identify these proteins, the three spots were excised from several twodimensional gels, concentrated, digested with trypsin, and the resulting peptides were analyzed by electrospray ionizationtandem mass spectrometry. All spots contained Hsp27, suggesting that the three spots either represent the mono-, bis-, and tris-phosphorylated isoforms of the protein with a 8-pCPT-cGMP-induced increase in the amount of the bis-and trisphosphorylated forms or indicate some different post-translational modifications (23). To confirm the identification of Hsp27, human platelets were labeled with [ 32 P]orthophosphate, stimulated with 500 M 8-pCPT-cGMP for 10 min, and proteins of the homogenate were separated by two-dimensional gel electrophoresis. The proteins were transferred to nitrocellulose and positions of the phosphoproteins were determined by autoradiography (Fig. 2, lower panel). The membranes were probed with a rabbit polyclonal antibody against Hsp27. Two radioactive protein spots that demonstrated increases in phosphorylation after cGK activation were immunoreactive with anti-Hsp27 antibody (Fig. 2, upper panel). Three more basic proteins, most likely representing additional nonphosphorylated or weakly phosphorylated isoforms of Hsp27, were also immunoreactive and decreased in amount during stimulation.
In Vitro Phosphorylation of Hsp27-To determine whether cGK phosphorylates Hsp27 in vitro, purified, recombinant Hsp27 was incubated with the three cGK isoforms I␣, I␤, and II or the catalytic subunit of cAMP-dependent protein kinase in the presence of [␥-32 P]ATP. An autoradiogram of a representative SDS-PAGE gel is shown in Fig. 3. Incorporation of phosphate was observed after 30 min with all of the four kinases, albeit at different levels, with cGK causing less phosphate incorporation than the C subunit. In a control experiment, VASP, a well known substrate for cAK and cGK (15), was equally phosphorylated by all four kinases.
No Stimulatory Effect of cGK on p38 MAPK Phosphorylation-It is known that Hsp27 is phosphorylated in human platelets directly by MAPKAP kinase 2 after stimulation of the platelets with thrombin and subsequent activation of the p38 MAPK cascade (24 -28). To exclude any direct stimulation of p38 MAPK or MAPKAP kinase 2 by 8-pCPT-cGMP and any indirect effect of cGK on Hsp27 phosphorylation via p38 MAPK, human platelets were stimulated with 500 M 8-pCPT-cGMP or, as a positive control, with 2 units/ml thrombin, and p38 MAPK activation was monitored by a specific antibody that recognizes the active, bis-phosphorylated form of p38 MAPK. In contrast to the control experiment with thrombin treatment, where p38 MAPK was rapidly phosphorylated and activated in platelets after 2 min, the stimulation with 8-pCPT-cGMP did not lead to increased p38 phosphorylation at any of the times analyzed (Fig. 4). Similar negative results were obtained when we investigated the ability of cGK to directly phosphorylate and activate MAPKAP kinase 2 (data not shown).
Identification of Phosphorylation Sites-It has been shown that Hsp27 is phosphorylated in vitro and in vivo by MAPKAP kinase 2 at Ser 15 , Ser 78 , and Ser 82 , with this latter residue being the most prominent in vitro phosphorylation site (29,30).
Experiments with cAMP-dependent protein kinase revealed phosphorylation of Ser 15 and Ser 86 of mouse Hsp25 in vitro, albeit with low efficiency (31). Interestingly, our sequence analysis of Hsp27 identified an additional putative phosphorylation site of Hsp27 for cAK and cGK at threonine 143 (Arg-Lys-Tyr-Thr 143 -Leu). To study this potential phosphorylation site, we constructed two mutants in which threonine 143 was replaced by a phosphate-mimicking glutamic acid: Hsp27-T143E and Hsp27-S15D,S78D,S82D,T143E. In addition, we investigated three Hsp27 mutants reported previously: Hsp27-S15D, Hsp27-S78D,S82D, and Hsp27-S15D,S78D,S82D (21). Analysis of these earlier mutants by in vitro phosphorylation experiments confirmed the results obtained previously with MAP-KAP kinase 2 showing complete absence of phosphate incorporation after substitution of all three known serine phosphorylation sites in mutant Hsp27-S15D,S78D,S82D (Fig. 5). In contrast, this mutant was still phosphorylated by both cGK and cAK (8 Ϯ 0.8 and 20 Ϯ 0.5% of wild-type phosphorylation, respectively). Only after mutation of threonine 143 to glutamic acid (Hsp27-S15D,S78D,S82D,T143E), phosphate incorporation was abolished (Fig. 5, Table I). To confirm this result, the threonine phosphorylation by cGK and cAK was further analyzed by phosphorylating wild type Hsp27 and Hsp27-T143E with the two kinases. A 50% reduction in phosphate incorporation was observed for the threonine mutant providing further evidence that threonine 143 represents an important phosphorylation site for cGK and cAK in Hsp27 (Fig. 6A). For quanti- fication, the areas of the gel corresponding to the autoradiogram in Fig. 5 were collected for liquid scintillation counting. These data are summarized in Table I. Ser 15 is probably not  phosphorylated by cGK since the mutants Hsp27-S78D,S82D and Hsp27-S15D,S78D,S82D showed similar phosphate incorporation. Interestingly, cAK also appears not to phosphorylate Ser 15 , although mimicking Ser 15 phosphorylation (Hsp27-S15D) increased incorporation of phosphate by cAK about 2-fold compared with wild-type Hsp27. This enhanced phosphate incorporation after Ser 15 mutation was also observed with MAPKAP kinase 2, albeit to a lesser extent: S15D phosphorylation increased to 118 Ϯ 9% of wild-type phosphorylation (Fig. 5, Table I). In the presence of cGK, wild-type Hsp27 is phosphorylated 24.8 Ϯ 3% with respect to wild-type phosphorylation by MAPKAP kinase 2 (Table I).
We next examined whether the phosphorylation of Hsp27 at threonine 143 (Hsp27-T143E) by cGK and cAK might influence MAPKAP kinase 2 phosphorylation. However, neither the ki-netics of MAPKAP kinase 2 phosphorylation nor the overall phosphate incorporation by MAPKAP kinase 2 was influenced by the T143E mutation (Fig. 6B).
Actin Polymerization-It has been suggested that small heat shock proteins are important regulatory components of the actin-based cytoskeleton (32) and that phosphorylation of Hsp27 might be implicated in regulating actin polymerization (33,34). Therefore, we compared the stimulatory effects on actin polymerization of the unphosphorylated wild-type Hsp27 with the phosphorylation-mimicking mutants. When preincubated with labeled G-actin, wild-type recombinant Hsp27 at all concentrations tested (up to 1 M) did not alter the polymerization of actin (Fig. 7A). In contrast, wild-type Hsp27 phosphorylated by MAPKAP kinase 2, as well as the MAPKAP kinase 2 phosphorylation-mimicking mutant Hsp27-S15D,S78D,S82D, both revealed a faster and stronger actin polymerization (124 Ϯ 6 and 124 Ϯ 1.8% of the control, respectively; Fig. 7B). This polymerization was significantly reduced (115 Ϯ 2.6% of control) by introducing the fourth cGK/cAK phosphorylation-mimicking site at threonine 143 (Hsp27-S15D,S78D,S82D,T143E) (Fig. 7C). Analysis of the single and double mutants Hsp27-S15D and Hsp27-S78D,S82D showed a slight

TABLE I Phosphate incorporation of wild-type and mutant Hsp27
Wild-type Hsp27 and the mutants S15D,S78D,S82D, S15D,S78D,S82D and S15D,S78D,S82D,T143E (each 0.5 M) were incubated with 0.2 units of MAPKAP kinase 2 or 0.05 M cGK and C-subunit in the presence of [␥-32 P]ATP at 30°C for 30 min in a total volume of 20 l. The proteins were resolved by SDS-PAGE and visualized by autoradiography. The corresponding areas of the gel were excised for liquid scintillation counting. Values presented are mean Ϯ S.E. from triplicate studies. Wild-type phosphorylation (set at 100%) corresponds to 0.6 mol of phosphate/mol of Hsp27.

DISCUSSION
Hsp27 was first identified as a substrate of the p38 MAPK/ MAPKAP kinase 2 pathway (24 -28, 35). In human platelets, activation of the p38/MAPKAP kinase 2 pathway after thrombin stimulation leads to a marked shift from the 27-kDa unphosphorylated form to at least three major phosphorylated forms (24,28). The sites phosphorylated by MAPKAP kinase 2 in vivo were identified as Ser 15 , Ser 78 , and Ser 82 (29,30). The present study characterizes Hsp27 as a substrate additionally of cGK in vivo, a protein involved in platelet inhibition (5).
Analysis of phosphate incorporation of different Hsp27 mutants after phosphorylation by cGK in vitro revealed an unknown phosphorylation site at threonine 143, as well as phosphorylation of residues Ser 78 and Ser 82 . Phosphorylation of human Hsp27 and its mouse homologue Hsp25 by different protein kinases has been addressed in several studies. Recently, Maizel et al. (36) showed that recombinant Hsp25 is efficiently phosphorylated by protein kinase C-␦ at Ser 15 and Ser 86 (analogous to Ser 82 of the human sequence). Earlier studies by Gaestel et al. (31) also identified both protein kinase C-␣ and cAMP-dependent protein kinase as being capable of phosphorylating Hsp25 in vitro at Ser 15 and Ser 86 . (Ser 78 , representing the third phosphorylation site in human Hsp27, is not conserved in rodent species.) In addition, another small heat shock protein in muscle (Hsp20) is phosphorylated on Ser 16 during activation of the cAMP-dependent pathway and could be phosphorylated in vitro by both cAK and cGK (37,38). The phosphorylation is associated with changes in the macromolecular association of Hsp20 (39).
Since the cGMP-and cAMP-dependent protein kinase pathways are involved in platelet inhibition, it was tempting to speculate that the phosphorylation at threonine 143 might negatively influence Hsp27 phosphorylation by MAPKAP kinase 2. In contrast to our hypothesis, however, we could not detect any changes in the rate or extent of Hsp27-T143E phosphorylation by MAPKAP kinase 2 in comparison with wild-type Hsp27 phosphorylation.
In our experiments concerning the influence of Hsp27 and its mutants on actin filament organization, we demonstrate that recombinant wild-type Hsp27 and the Hsp27-T143E mutant had no effect on actin polymerization. Interestingly, recombinant Hsp27, phosphorylated by MAPKAP kinase 2, and the phosphorylation-mimicking mutant Hsp27-S15D,S78D,S82D both increased actin polymerization by ϳ23%. In general, these results show the same tendency as the data described by  Benndorf et al. (34), where phosphorylation of native Hsp25 leads to a transition from inhibition to neutral effects, while in our experiments phosphorylation of recombinant Hsp27 shifts from neutral to stimulating effects. Most interestingly, additional mimicry of phosphorylation by changing threonine 143 to glutamic acid in the Hsp27-S15D,S78D,S82D mutant significantly reduces the stimulation of actin polymerization and could explain in part the inhibitory effect of cGK on thrombininduced platelet activation since actin polymerization is required for agonist-induced shape change.
The single-and double-mutated Hsp27 forms S15D and S78D,S82D induced a 10% decrease in actin polymerization. This is probably due to the different oligomerization of the mutants. Hsp27-S15D and Hsp27-S78D,S82D both showed large, round particles (6 tetramers), whereas wild-type Hsp27 phosphorylated by MAPKAP-kinase 2 as well as the mutant Hsp27-S15D,S78D,S82D formed mostly single tetramers, which are thought to be responsible for stabilization of actin filaments (21).
Apart from this, phosphorylation of Hsp27 by cGK might influence its chaperone function. For example, Zhue et al. (32,40) observed that thrombin activation leads to co-precipitation of platelet factor XIII and two yet unidentified proteins with Hsp27. This association between any of these protein and Hsp27 might be influenced by cGK phosphorylation. Heat shock protein 60 (Hsp60) has been shown to function as a molecular chaperone for histone 2B (H2B) when both proteins are in their dephosphorylated form. Phosphorylation by cAK of both Hsp60 and H2B causes dissociation of H2B from Hsp60 and loss of H2B from the plasma membrane (41).
Certainly, phosphorylation of Hsp27 by cGK is not the only cGK-dependent pathway involved in platelet inhibition. For example, phosphorylation of Rap 1B by cGK (and cAK) in intact platelets is associated with the inhibition of thrombin-induced PLC␥ activity (16,42). Phosphorylation of the major cGK substrate in human platelets, the 50-kDa protein VASP, is assumed to be involved in the regulation of the fibrinogen receptor glycoprotein IIb-IIIa (43). Recent experiments with VASP mutants imitating defined phosphorylation states revealed that phosphorylation of VASP down-regulates its in vitro Factin binding and actin polymerization promoting activity (44). The mechanism of cGK action also includes inhibition of inositol 1,4,5-trisphosphate-mediated Ca 2ϩ mobilization from intracellular stores and the secondary, store-related calcium influx (45,46). Phosphorylation of the inositol 1,4,5-trisphosphatereceptor, however, is well established in vitro (8,47) but remains to be demonstrated for human platelets.
In summary, the present findings demonstrate that Hsp27 is an important substrate for cGK in intact human platelets. The different effects of mutating the phosphorylation sites for MAP-KAP kinase 2 and cGK on the stimulation of actin polymerization could explain the stimulatory role of MAPKAP kinase 2 and in part the inhibitory role of cGK in platelet activation. This further supports the notion that different enzymes can phosphorylate Hsp27 in human platelets at different sites, and, by this mechanism, positively or negatively regulate platelet activation.