Loss of Serotonin Transporter Function Alters ADP-mediated Glycoprotein αIIbβ3 Activation through Dysregulation of the 5-HT2A Receptor*

Reduced platelet aggregation and a mild bleeding phenotype have been observed in patients chronically taking selective serotonin reuptake inhibitors (SSRIs). However, it remains unclear how SSRIs, which inhibit the plasma membrane serotonin transporter (SERT), modulate hemostasis. Here, we examine how sustained inhibition of SERT activity alters serotonergic signaling and influences platelet activation and hemostasis. Pharmaceutical blockade (citalopram dosing) or genetic ablation (SERT−/−) of SERT function in vivo led to reduced serotonin (5-hydroxytryptamine (5-HT)) blood levels that paralleled a mild bleeding phenotype in mice. Transfusion of wild-type platelets to SERT−/− mice normalized bleeding times to wild-type levels, suggesting that loss of SERTs causes a deficiency in platelet activation. Although SERT−/− platelets displayed no difference in P-selectin or αIIbβ3 activation upon stimulation with thrombin, ADP-mediated αIIbβ3 activation is reduced in SERT−/− platelets. Additionally, synergistic potentiation of αIIbβ3 activation by ADP and 5-HT is lost in SERT−/− platelets. Acute treatment of wild-type platelets with 5-HT2A receptor (5-HT2AR) antagonists or SSRIs revealed that functional 5-HT2ARs, not SERTs, are necessary for the synergistic activation of αIIbβ3 by dual 5-HT/ADP stimulation. Pharmacological studies using radiolabeled guanosine 5′-3-O-([35S]thio)triphosphate and [3H]ketanserin revealed that platelets isolated from SERT−/− or citalopram-treated mice have reduced activation of G-proteins coupled to 5-HT2ARs and receptor surface expression. Taken together, these data demonstrate that sustained SERT loss of function reduces 5-HT2AR surface expression that is critical for the synergistic activation of αIIbβ3 by 5-HT and ADP. These results highlight an antiplatelet strategy centered on blocking or desensitizing 5-HT2AR to attenuate ADP-mediated αIIbβ3 activation.

In the periphery, serotonin (5-HT) 3 is produced by enterochromaffin cells in the gastrointestinal tract, released into the plasma, and quickly taken up by platelets via the plasma membrane serotonin transporter (SERT). Following uptake, 5-HT is stored in dense granules by the actions of the vesicular monoamine transporter 2 (1)(2)(3)(4). Chronic inhibition of SERT through selective serotonin reuptake inhibitors (SSRIs) (e.g. citalopram and paroxetine) leads to dramatically reduced platelet 5-HT granule content (5,6), altering peripheral 5-HT homeostasis and potentially modifying multiple physiological processes including hemostasis (7)(8)(9)(10). Clinically, increased bleeding risk has been observed in patients taking SSRIs, and platelet aggregation is disrupted (5,11). Here, we have characterized a similar effect in two distinct mouse models of lost SERT function, suggesting that sustained loss of SERT function influences hemostasis.
Platelet dense granules contain 5-HT along with other platelet agonists including adenosine diphosphate (ADP), thromboxane (TXA2), and histamine. Appropriate platelet activation depends on the timely release of these factors (4,5,12). Platelet aggregation is crucial early in thrombus formation (4,5,13). Aggregation, which is the bridging of platelet-platelet contacts, requires a conformational alteration in the glycoprotein ␣IIb␤3, leading to its activation and fibrinogen binding. 5-HT has been shown to enhance aggregation in a 5-HT 2A receptor (5-HT 2A R)-dependent manner (4, 14 -17). The 5-HT 2A R is the only serotonergic receptor found on platelets and potentiates platelet responses to weak agonists like ADP (18). Subthreshold concentrations of two different platelet agonists can exert a synergistic effect on platelet activation. One example includes dual ADP and 5-HT activation leading to increases in cytosolic [Ca 2ϩ ] (13). However, the role of 5-HT during in vivo hemostasis remains unclear, particularly in the context of chronic SERT inhibition.
To elucidate the underlying mechanisms of SSRI effects on platelet aggregation, a better understanding of acute versus chronic inhibition of SERT function during platelet activation is required. Acute and chronic blockage of SERT function results in distinct scenarios regarding the effects on 5-HT homeostasis. Acute inhibition of SERT blocks the amount of 5-HT carried into the cytosol during platelet activation, whereas chronic blockage of SERT slowly depletes granule 5-HT, resulting in loss of 5-HT secretion and 5-HT signaling during platelet activation. Some have shown that acute SERTmediated 5-HT uptake modulates platelet function (20,21) in part through the transamination of small GTPase proteins important for platelet ␣-granule exocytosis. However, the effects of chronic loss of SERT function on platelet activation remain unresolved.
In this study, we investigated the mechanisms by which chronic inhibition of SERT alters hemostatic function using two independent models (SERT Ϫ/Ϫ and 6-day citalopram treatment). We hypothesized that the bleeding effects noted with the disruption of SERT function are due to altered 5-HT 2A R signaling during platelet activation. Indeed, we found that mice lacking SERT function display a bleeding phenotype that can be rescued by the addition of wild-type platelets. Furthermore, we found that ADP-mediated ␣IIb␤3 activation was reduced in SERT Ϫ/Ϫ and citalopram-treated platelets due to loss of 5-HT 2A R signaling and surface expression.

Decreased 5-HT Content in Platelets Isolated from SERT ؊/؊ and Citalopram-treated Mice Parallels Bleeding Phenotype
Mice Treated with Citalopram for 6 Days Have Reduced Whole Blood 5-HT Levels and Increased Tail Bleed Time-To determine whether SSRIs alter blood 5-HT homeostasis, mice were exposed to citalopram-treated water for 6 days, 1 day beyond the lifetime of a circulating platelet in a mouse (22). We used high performance liquid chromatography (HPLC) to measure whole blood levels of 5-HT and its major metabolite 5-hy-droxyindoleacetic acid (5-HIAA). Whole blood from wild-type mice undergoing citalopram (Cit) treatment showed reduced 5-HT levels as compared with water-treated controls (Fig. 1A). No significant difference in 5-HIAA levels was found between water-and Cit-treated samples (Fig. 1B). To determine whether citalopram treatment alters hemostasis, we performed a tail bleed assay. We found that citalopramtreated mice exhibited a significantly longer tail bleed time as compared with water-treated controls (Fig. 1C). These findings indicate a phenotypic association between reduced 5-HT levels and increased bleeding time in mice treated with SSRIs.
Reduced 5-HT Whole Blood Levels and Increased Tail Bleed Times in SERT Ϫ/Ϫ -We observed very little measurable 5-HT in SERT Ϫ/Ϫ platelets (Fig. 1D). Similar to whole blood samples isolated from citalopram-treated mice, there was no significant difference in 5-HIAA between SERT ϩ/ϩ and SERT Ϫ/Ϫ mice (Fig. 1E). We found that bleeding time was significantly increased in SERT Ϫ/Ϫ mice as compared with SERT ϩ/ϩ (Fig.  1F). These data show that SERT function modulates whole blood 5-HT levels and that SERT inhibition is associated with increased bleeding time.
was observed between genotypes ( Fig. 2A). To determine whether addition of SERT ϩ/ϩ platelets rescued the SERT Ϫ/Ϫ bleeding phenotype, mice were transfused with gel-filtered platelets resuspended in saline via jugular vein injection. A tail bleed was performed 3 min after transfusion of SERT ϩ/ϩ platelets to SERT ϩ/ϩ and SERT Ϫ/Ϫ mice (Fig. 2B). In control experiments, saline-transfused SERT Ϫ/Ϫ mice showed a significantly longer tail bleed time as compared with SERT ϩ/ϩ mice. We found that addition of gel-purified platelets (2 ϫ 10 7 in 200 l of saline) to SERT ϩ/ϩ mice did not significantly alter tail bleed time. The difference in SERT ϩ/ϩ and SERT Ϫ/Ϫ bleeding time was abolished following platelet transfusion, indicating that hemostasis was rescued in SERT Ϫ/Ϫ mice. These findings support the conclusion that the bleeding phenotype of the SERT Ϫ/Ϫ mice results from a defect in platelet function.
SERT Ϫ/Ϫ Platelets Have Reduced ADP-mediated ␣IIb␤3 Activation-We next tested ADP platelet activation in SERT Ϫ/Ϫ platelets. We found that 10 M, but not 1 M, ADPmediated ␣IIb␤3 activation is significantly reduced in the SERT Ϫ/Ϫ platelets (Fig. 3C). No difference in P-selectin expression was observed in SERT Ϫ/Ϫ or SERT ϩ/ϩ platelets (Fig. 3D). These data demonstrate that loss of SERT function leads to a deficiency in the ADP-mediated inside-out signaling that triggers activation of ␣IIb␤3.

5-HT 2A R Stimulation Enhances ADP-mediated ␣IIb␤3 Activation
To examine the role of 5-HT in ADP-mediated platelet activation, we distinguished the role of two plasma membrane proteins that are responsive to 5-HT in platelets: SERT and the 5-HT 2A R. We performed a dose-response curve with citalopram and paroxetine, two structurally distinct SSRIs, and found that they block intact 5-[ 3 H]HT platelet uptake with an IC 50 of 1.5 ϫ 10 Ϫ11 and 1.3 ϫ 10 Ϫ11 , respectively (Fig. 4A). Platelet activation with 1 and 10 M ADP was examined with the addition of either 5-HT 2A R antagonists (ketanserin and methysergide) or SERT inhibitors (citalopram and paroxetine) 5 min before ADP activation. Both ketanserin and methysergide treatment significantly reduced ADP-mediated activation at 1 M (Fig. 4B). Acute inhibition of 5-HT uptake by SERT with citalopram (10 nM) or paroxetine (10 nM) had no effect on ADPmediated ␣IIb␤3 activation despite effectively blocking SERTmediated 5-HT uptake. These data reveal a role for 5-HT 2A R activation, but not 5-HT uptake, in ADP-mediated ␣IIb␤3 inside-out activation.

ADP Levels and ADP Receptor Expression Are Retained in SERT
Ϫ/Ϫ Platelets-We first explored whether the loss of ADP-mediated ␣IIb␤3 activation was due to a deficit in ADPdependent signaling. We examined ADP receptor (P2Y 1 and P2Y 12 ) expression levels via [ 3 H]ADP binding that was competitively blocked with ADP receptor-specific antagonists. This assay was optimized in SERT ϩ/ϩ platelet membrane preparations and shown as percentage of total [ 3 H]ADP binding blocked in the presence of 2,2Ј-pyridylisatogen tosylate, a P2Y antagonist (Fig. 5A). Competitive inhibition with MRS2179 (P2Y 1 -specific antagonist) and ticagrelor (P2Y 12 -specific antagonist) was used to determine relative receptor binding in SERT ϩ/ϩ and SERT Ϫ/Ϫ platelet membrane preparations (  H]ADP binding to P2Y 1 in platelet membranes. We found no statistically significant difference in either P2Y 12 or P2Y 1 binding between SERT ϩ/ϩ and SERT Ϫ/Ϫ membrane preparations. Additionally, we measured whole blood ADP levels using an ELISA and observed no difference between SERT ϩ/ϩ and SERT Ϫ/Ϫ preparations (Fig. 5E). These data demonstrate that the ADP whole blood levels and receptor numbers are not different in SERT Ϫ/Ϫ platelets.

5-HT 2A R G-protein Activation and Surface Expression Levels
Are Reduced in SERT Ϫ/Ϫ Platelets-Based on our findings, the 5-HT 2A R plays an important role in ADP-mediated ␣IIb␤3 activation, and this effect is reduced in SERT Ϫ/Ϫ platelets. Therefore, we purified platelet membranes and measured [ 35 S]GTP␥S incorporation to 5-HT 2A R following activation with 1-[2,5-dimethoxy-4-iodophenyl]-2-aminopropane (DOI), a 5-HT 2A R-specific agonist. We observed a significant reduction in [ 35 S]GTP␥S incorporation to SERT Ϫ/Ϫ membrane preparations as compared with SERT ϩ/ϩ ( Fig. 5F; n ϭ 8; p ϭ 0.0002, Mann-Whitney test). To determine whether this reduction in G-protein activation was caused by reduced 5-HT 2A R surface expression, we used radiolabeled [ 3 H]ketanserin to measure surface levels of the 5-HT 2A R on intact platelets. A saturation curve was performed to examine the number of surface receptors (B max ) of 5-HT 2A R (Fig. 5G). SERT Ϫ/Ϫ platelets showed significantly reduced B max for 5-HT 2A R surface levels as compared with those observed in SERT ϩ/ϩ platelets (Fig.  5H). There was no significant difference in K d between FIGURE 3. Platelet ADP-mediated ␣IIb␤3 activation is reduced in SERT ؊/؊ mice. A, JON/A binding in SERT Ϫ/Ϫ (n ϭ 6) and SERT ϩ/ϩ (n ϭ 6) mice with 10 and 200 nM thrombin shows no difference based on genotype (Tukey's post-test following two-way ANOVA: thrombin, p Ͻ 0.0001). B, P-selectin binding did account for some variation (two-way ANOVA: thrombin, p Ͻ 0.0001; genotype, p ϭ 0.032) but did not show any difference following post-test (Tukey's multiple comparison test). C, ADP-mediated JON/A binding is significantly reduced in the SERT Ϫ/Ϫ platelets at 10 M ADP (10 M ADP-treated SERT ϩ/ϩ versus 10 M ADP-treated SERT Ϫ/Ϫ , p ϭ 0.0354 (*)) but not at 1 M ADP (two-way ANOVA revealed an overall effect of ADP (p ϭ 0.0001) and an interaction effect (p ϭ 0.026); additionally, 1 M ADP-treated SERT ϩ/ϩ versus 1 M ADP-treated SERT Ϫ/Ϫ showed a significant effect (p ϭ 0.999) using Tukey's multiple comparison test). D, there is no difference in P-selectin expression in SERT Ϫ/Ϫ and SERT ϩ/ϩ mice (two-way ANOVA: ADP, p ϭ 0.0473; Tukey's multiple comparison test, not significant). Data are shown with median, range, and individual points. Error bars represent the range of the data set. gMFI, geometric mean fluorescence intensity; PE, phycoerythrin. SEPTEMBER 16, 2016 • VOLUME 291 • NUMBER 38

Citalopram-treated Platelets Recapitulate the SERT ؊/؊ Platelet Phenotype: Citalopram-treated Platelets Have Reduced ADP-mediated ␣IIb␤3 Activation and Reduced 5-HT 2A R Surface Expression
To determine whether we could recapitulate the loss of 5-HT 2A R surface expression using an alternative model of lost SERT function, we treated mice chronically with citalopram for 6 days. Similar to SERT Ϫ/Ϫ platelets, citalopram-treated platelets showed reduced ADP-dependent ␣IIb␤3 activation compared with water-treated controls (Fig. 6A). In control watertreated mouse platelets, ␣IIb␤3 activation by ADP was enhanced by 5-HT, whereas 5-HT potentiation of ADP signaling was absent in citalopram-treated mouse platelets (Fig. 6B). Furthermore, 5-HT 2A R levels were reduced in platelets isolated from citalopram-treated mice (Fig. 6C). These findings establish that 6-day treatment with SSRI is sufficient to reduce platelet expression of 5-HT 2A R and decrease the capacity of platelets to be synergistically activated by ADP and 5-HT. Additionally, these findings demonstrate that the loss of 5-HT 2A R surface expression and platelet function in SERT Ϫ/Ϫ mice is not due to loss of the SERT protein but rather results from sustained loss of SERT function. Finally, to test whether premature exposure to 5-HT could alter ADP-mediated ␣IIb␤3 activation, wildtype platelets were incubated with serotonin (10 M) for 15 min, washed, and resuspended. They were then activated with 10 M ADP, and JON/A binding was measured (Fig. 6D). We found that platelets pretreated with 5-HT for 15 min lost ADPmediated ␣IIb␤3 activation similarly to both SERT Ϫ/Ϫ and citalopram-treated platelets. These results suggest that the preactivation or possible desensitization of the 5-HT 2A R reduces platelet ADP-mediated ␣IIb␤3 activation.

Discussion
Our research has demonstrated how loss of SERT function leads to platelet dysfunction, providing mechanistic insight into the increased bleeding times previously observed in patients taking SSRIs (4,23). We characterized a bleeding phenotype in two independent models of lost SERT function, genetic and pharmacologic. This effect was rescued by addition of purified platelets with intact SERT function, indicating that the actions of SSRIs are mediated primarily via alteration of platelet function. Sustained loss of SERT function reduces 5-HT 2A R surface expression, leading to reduced ADP-mediated ␣IIb␤3 activation. Therefore, our findings support a novel mechanism for the reduced aggregation in SSRI-treated patients:  92). B, JON/A binding following 15-min ADP activation was measured using geometric mean fluorescence intensity (gMFI) following a 5-min preincubation with either 5-HT 2A receptor antagonist ketanserin or methysergide or selective serotonin reuptake inhibitor citalopram or paroxetine on the same platelet preparation (n ϭ 6). Both ketanserin and methysergide treatment significantly reduced ADP ␣IIb␤3 activation at both 1 (ketanserin, p ϭ 0.0409 (*); methysergide, p ϭ 0.0041 (**)) and 10 M (ketanserin, p ϭ 0.0187 (*); methysergide, p Ͻ 0.0001 (****); Dunnett's test with two-way repeated measures ANOVA: interaction, p ϭ 0.0103; ADP, p Ͻ 0.0001; inhibitors, p Ͻ 0.0001; subjects (matching), p Ͻ 0.0001). Non-parallel experiments examined JON/A binding following incubation with 5-HT alone (C), ADP 1 M ϩ 5-HT (p Ͻ 0.0001 (***)) (D), or ADP/5-HT following a 5-min incubation with ketanserin (Sidak's test following two-way ANOVA) (E). Data are shown with median, range, and individual points. Error bars represent the range of the data set. There are three possibilities by which SSRIs may mediate their effects on platelets: 1) acute blockage of SERT function and 5-HT uptake, 2) depletion of granule 5-HT and loss of 5-HT secretion during activation, and 3) increased extracellular 5-HT levels. Many studies concluding that acute uptake of 5-HT contributes to platelet function used SSRIs as tool compounds to acutely inhibit 5-HT uptake. However, these studies utilized SSRIs at high concentrations (Ͼ10 M) that are known to exert off-target effects (e.g. the 1 receptor) (11,24). Comparatively, our study uses concentrations of SSRIs that block platelet-mediated SERT uptake but are far below the reported concentrations resulting in off-target effects (10 nM; Fig. 4). Here, we observed no change in ADP-mediated ␣IIb␤3 activation following acute SSRI treatment, suggesting that acute 5-HT uptake is not required for inside-out activation of ␣IIb␤3 by weak agonists. However, we observed reduced ADP-mediated ␣IIb␤3 activation following chronic SSRI treatment.

5-HT 2A R Surface Expression Reduced with Lost SERT Function
Others have suggested that acute uptake of 5-HT by SERT leads to a receptor-independent signaling pathway via the posttranslational modification of intracellular proteins (20). This process, known as serotonylation, involves the covalent attachment of 5-HT molecules to proteins mediated by the enzyme transglutaminase, resulting in transamination of small GTPases important for platelet ␣-granule secretion (20,25). In our experiments, we demonstrated no difference in P-selectin antibody as a measure for ␣-granule exocytosis in SERT Ϫ/Ϫ platelet following thrombin activation (Fig. 3B). These data would suggest that ␣-granule exocytosis is intact following the chronic loss of SERT function. Instead, we observed that chronic loss of SERT function led to reduced 5-HT 2A R signaling and reduced ADP-mediated ␣IIb␤3 activation, which is necessary for proper aggregation in vivo.
In conjunction with depleting platelet granule 5-HT levels, one would expect that chronic SERT inhibition would increase plasma concentrations of 5-HT as suggested by the effects of SSRIs in the CNS (26,27). However, because 5-HT 2A R stimulation by 5-HT alone does not lead to platelet activation, loss of SERT function likely leads to local increases in plasma 5-HT levels within the portal vein and indirectly triggers internalization of the 5-HT 2A R. It has been demonstrated that 5-HT induces internalization of 5-HT 2A R in a ␤-arrestin-dependent manner within 30 min of initial exposure (28,29). 5-HT 2A R  SEPTEMBER 16, 2016 • VOLUME 291 • NUMBER 38 desensitization has been shown in the brain following prolonged exposure to elevated levels of serotonin in citalopramtreated mice (30,31). However, it has yet to be validated whether extracellular plasma 5-HT levels increase following SSRI treatment. Evaluating changes in plasma (i.e. extracellular) 5-HT levels as compared with platelet (i.e. intracellular) 5-HT levels following SSRI treatment has been challenging. Measuring plasma 5-HT levels has proven difficult due to very low concentration by extraordinarily efficient liver metabolism as well as the free diffusion of 5-HT out of circulation (32)(33)(34). Furthermore, levels of plasma 5-HT can vary greatly, possibly due to contamination by platelet granule release during plasma purification (1,2,34,35). Reported plasma 5-HT levels fluctuate between investigators and preparations, but our data clearly demonstrate that 5-HT homeostasis is altered by loss of SERT function (36 -38). Furthermore, our data demonstrate that improper exposure of platelets to 5-HT likely desensitizes the receptor, leading to reduced ADP-mediated ␣IIb␤3 activation (Fig. 6D). Our findings indicate that chronic loss of SERT function reduces 5-HT 2A R surface expression and signaling in each of two models of altered 5-HT homeostasis.
Clinically, targeted inhibition of the 5-HT 2A R has been efficacious (17). APD791, an inverse agonist of 5-HT 2A R, is currently in clinical trials and attenuated recurrent thrombosis irrespective of the time of treatment (16,17). No increase in bleeding time in the presence of APD791 was observed as compared with other antiplatelet therapies. Additionally, APD791 was able to block 5-HT-dependent platelet activation over a short time scale (2 h) (16,17). Our data provide mechanistic insight into the reduced aggregation previously observed in SSRI-treated patients and support explorations of peripherally restricted, well tolerated SSRIs or 5-HT 2A R antagonists as an antiplatelet therapeutic approach. The therapeutic implications of this work suggest a novel approach targeting the 5-HT signaling systems via SERT inhibition, leading to reduced 5-HT 2A R-mediated platelet aggregation. Animals and Housing-All mice were group-housed (two to five mice per cage) in temperature-and humidity-controlled conditions under a 12-h light/dark cycle with food and water available ad libitum. All studies were performed in accordance with humane guidelines established by the Vanderbilt Institutional Animal Care and Use Committee under an approved protocol (M/12/121). Age-and sex-matched mice were used in all experiments (8 -20 weeks of age). All experiments were run with either SERT ϩ/ϩ (wild-type) or SERT Ϫ/Ϫ (serotonin transporter knockout) homozygous mice of both sexes in a C57BL6 background.

Materials-Thrombin
Administration of Citalopram-Citalopram was administered acutely (addition of drug during platelet activation) or chronically (via drinking water treatment). Citalopram-treated drinking water was prepared based on average weight of the mice and an average consumption of ϳ7 ml/day/mouse for a dosing of 15 mg/kg/day. Citalopram-prepared water was added to a water dispenser, and mice were given full access to either non-treated control or citalopram-treated water. Mice were exposed to citalopram-treated water for 6-days, 1 day beyond the lifetime of a circulating platelet in a mouse (22), to model chronic loss of SERT function in the periphery but minimize central nervous systems effects. After 6 days, mice were euthanized, and experiments were performed as indicated.
Tail Bleed-Experiments were carried out as described previously (48). Briefly, mice were maintained under anesthesia (2% isoflurane and 1 ml/min oxygen; JD Medical Distributing Co., Inc., Phoenix, AZ), and a transverse incision was made over a lateral tail vein. The tail was immersed in normal saline (37°C) in a hand-held test tube. The time from the incision to the cessation of bleeding was recorded as the bleeding time. Maximum time allowed for cessation of bleeding was 12 min before manually stopping bleeding.
Blood Collection and Platelet Purification-We used isoflurane because the effect is considered negligible on platelets in most species (19). Cardiac puncture was performed in euthanized mice using a 25-gauge needle/1-ml syringe containing anticoagulant. For platelet isolation, whole blood was diluted 1:1 in Tyrode's buffer (10 mM HEPES, 11.9 mM NaHCO 3 , 127.2 mM NaCl, 5 mM KCl, 0.4 mM NaH 2 PO 4 , 1 mM MgCl 2 , 5 mM glucose, pH 7.4), layered onto 2 ml of Fico/Lite platelets (Atlanta Biologicals, Inc., Lawrenceville, GA), and spun for 15 min at 350 ϫ g. Three antiplatelet activation agents used during our purification include acid-citrate-dextrose, an additional calcium chelator; apyrase, an enzyme that degrades ADP to prevent premature ADP activation during preparations; and PGE 1 , which stimulates adenyl cyclase activity in platelets and increases cyclic AMP concentrations. Platelets were washed in Tyrode's buffer containing acid-citrate-dextrose, 0.2 unit/ml apyrase, and 5 g/ml PGE 1 (10 min at room temperature) and spun at 1000 ϫ g for 10 min to obtain a washed platelet pellet. Platelets were resuspended in Tyrode's buffer and allowed to equilibrate for 30 min before experimentation. Platelets were counted on a Coulter counter and diluted in Tyrode's buffer with 0.1% BSA to the indicated concentrations.
Platelet Transfusion-Platelet transfusions were performed by jugular vein injection of 200 l of 1 ϫ 10 8 /ml gel-filtered platelets. A 3-min incubation time was allowed before tail bleed time was determined. Tail bleed was performed as described above.
ADP, 5-HT, and 5-HIAA Levels-Whole blood was collected in 3.2% sodium citrate and stored at Ϫ80°C until analyzed. 5-HT and 5-HIAA levels were measured by HPLC by the Vanderbilt Neurochemistry core. Briefly, 5-HT was determined by a specific HPLC assay utilizing an Antec Decade (oxidation, 0.4) electrochemical detector. Samples were injected using a Water 2707 autosampler onto a Phenomenex Kintex (2.6-m, 100-Å) C 18 HPLC column (100 ϫ 4.60 mm). 5-HT eluted with a mobile phase consisting of 89.5% 0.1 M TCA, 10 Ϫ2 M sodium acetate, 10 Ϫ4 M EDTA, and 10.5% methanol, pH 3.8. HPLC control and data acquisition were managed by Millennium 32 software. Daily calibration curves were generated by supplementing whole blood with stock 5-HT solutions to yield final concentrations of 1, 3, 6, 12, 25, 50, 75, and 100 ng/ml using 3,4-dihydroxybenzylamine as an internal standard. An ADP ELISA kit was purchased from Abcam (ab83359). ELISA was performed as described by the manufacturer.
Gel-filtered Platelet Flow Cytometry-After purifying platelets, platelet counts were determined and normalized to 1 ϫ 10 7 /ml. 25 l of platelets were added and mixed with 5 l each of both anti-mouse P-selectin and anti-JON/A that binds a conformation of ␣IIb␤3 (EMFRET Analytics & Co. KG). Platelets were allowed to incubate with the antibodies for 15 min before activation. Antagonists were allowed to incubate with the platelet antibody mixture for 5 min before activation. Activation was stopped by addition of 100 l of 2% paraformaldehyde in phosphate-buffered saline (PBS) (0.138 M NaCl, 0.0027 M KCl, pH 7.4) 15 min after activation followed by addition of 300 l of SEPTEMBER 16, 2016 • VOLUME 291 • NUMBER 38 PBS after 15 min. Samples were analyzed at the Nashville Veterans Affairs Medical Center Flow Cytometry Resource Center (17,48) using geometric mean fluorescence intensity, which gives the mean fluorescence intensity of each fluorescence channel for the sample (17,48).

5-HT 2A R Surface Expression Reduced with Lost SERT Function
Radioligand Binding-Platelet counts were normalized before analysis using a Coulter Counter at 5-3 ϫ 10 8 cells/ml. A saturation curve was done with increasing concentrations of [ 3 H]ketanserin (0.625-20 nM) incubated with 50 l of platelets in suspension and methysergide at 20 M to determine nonspecific binding. B max values were calculated using a nonlinear regression analysis for one-site specific binding (GraphPad Software, Inc., La Jolla, CA). G-protein activation was measured via [ 35 S]GTP␥S incorporation. Mouse platelets were pooled (n ϭ 2) per assay. Platelets were lysed by addition of double distilled H 2 O and placed at Ϫ20°C for a minimum of 3 h. Samples were thawed and spun at 20,000 ϫ g for 20 min to collect membranes. Membranes were resuspended in membrane storage buffer (10 mM HEPES, pH 7.4, 1 mM EDTA). Samples were incubated in 50 M GDP and 0.5 mM dithiothreitol for 15 min. 1 M DOI and 0.2 nM [ 35 S]GTP␥S were added to samples and incubated for 60 min at room temperature. Counts were standardized by protein concentration of membrane preparation. For nonspecific [ 35 S]GTP␥S incorporation, excess unlabeled GTP␥S (100 M) was added. 5-HT 2A receptor-specific [ 35 S]GTP␥S was measured following normalization of SERT ϩ/ϩ DOI-specific counts to non-DOI-specific counts. The assay was terminated by filtration through polyethyleneimine-coated GF/B Whatman filters using a Brandel Cell Harvester (Brandel, Gaithersburg, MD), and final counts were measured using a scintillation counter. Finally, competitive binding of [ 3 H]ADP to platelet membrane preparations was used to determine relative percentages of P 2 Y 1 and P 2 Y 12 receptors. [ 3 H]ADP at 20 nM was added to platelet membrane preparations (5-20 g total). Platelets were then incubated with the indicated concentration of either P2Y 12 (ticagrelor), P2Y 1 (MRS2179), or dual P2Y 1 and P2Y 12 antagonist (2,2Ј-pyridylisatogen tosylate) to determine receptor-specific binding.
Statistics-All data were analyzed in Prism 4.0c (GraphPad Software, Inc.). Outliers were determined using ROUT methods with Q ϭ 1%. A non-parametric Mann-Whitney test was performed to account for non-determination of equal variance (included in results). All figures display the median and the range. Median values are reported in the text. Student's t tests or two-way analysis of variance (ANOVA) followed by Tukey's post-tests were performed as appropriate (indicated in results).