Temporal phosphoproteomics reveals key regulators of procoagulant COAT platelet generation

Key Points

• •Differential phosphorylation of ion channels, inhibitory proteins, and GTPases precedes the procoagulant response and may initiate it.
• •Differential phosphorylation of calcium and talin-1 regulators may promote high cytosolic calcium levels and integrin αIIbβ3 inactivation.

Visual Abstract

Abstract

Procoagulant platelets play a major role in stabilizing the primary hemostatic plug formed by aggregated platelets. The intracellular signaling pathways driving the shift from an aggregatory to a procoagulant COAT platelet phenotype upon collagen plus thrombin stimulation are only partially elucidated. Aiming at identifying differentially phosphorylated proteins and potential regulators of the procoagulant response, we mapped time-dependent changes in the phosphoproteome of aggregatory vs procoagulant COAT platelets using a liquid chromatography-mass spectrometry–based temporal phosphoproteomics approach. Upon stimulation with convulxin (an agonist of the collagen receptor glycoprotein VI) and thrombin, we observed that all platelets rapidly increased their overall phosphorylation during the first minute. Afterward, aggregatory platelets maintained or further increased their phosphorylation levels, whereas procoagulant COAT platelets exhibited a progressive and marked decrease in phosphorylation. Notably, during the first minute after activation, before the phenotypic development of the procoagulant response, several ion channels (calcium, sodium, hydrogen, and chloride), proteins involved in receptor negative feedback loops, cyclic nucleotide–dependent regulatory proteins, GTPases, and key kinases were already differentially phosphorylated in procoagulant COAT vs aggregatory experimental conditions. At the onset of the procoagulant response (1-3 minutes after activation), some calcium channels, phosphatidylinositol transfer proteins, α2-adrenergic receptors, and proteins involved in intracellular and extracellular mechanisms regulating integrin αIIbβ3 functionality were also differentially phosphorylated. This study highlights the usefulness of assessing phosphorylation dynamics to identify very early phosphorylation events and to understand the rapid phenotypic switch from aggregatory to procoagulant COAT platelets. This approach identified several candidate proteins and mechanisms likely to initiate and modulate the procoagulant response.

Introduction

Platelets play a key role in hemostasis by facilitating the formation of a stable hemostatic plug.¹ Within seconds upon simultaneous activation by collagen and thrombin, all platelets release granular contents and activate the fibrinogen receptor integrin αIIbβ3 (glycoprotein IIb/IIIa [GPIIb/IIIa]), acquiring their aggregatory properties, which support primary clot formation. Starting 1 to 2 minutes after activation by collagen and thrombin, a subset of aggregatory platelets downregulates integrin αIIbβ3 and develops a procoagulant phenotype (called procoagulant COAT platelets), sustaining thrombin generation on their surface and focal fibrin deposition to stabilize the clot.2, 3, 4 Abnormal procoagulant platelet generation contributes to the hemostatic imbalance leading to bleeding or thrombotic events.^5,6

For a platelet to acquire the procoagulant phenotype, the reverse mode of the sodium-calcium exchanger (NCX) has to be engaged, providing an additional pathway for calcium entry, eventually resulting in a high and sustained cytosolic calcium concentration.⁷ This overloads the mitochondrial calcium buffering capacity, triggering mitochondrial permeability transition pore opening and leading to mitochondria depolarization.8, 9, 10 The resulting calcium release into the cytosol leads to calpain activation, decreases aggregatory properties, and triggers the translocation of negatively charged phospholipids on the platelet surface via activation of the scramblase TMEM16F.^4,10 Concomitantly, procoagulant COAT platelets also acquire a coating with α-granule proteins, which confer procoagulant and adhesive properties.^2,11 Although this sequence of events leading to procoagulant COAT platelet formation is well-known, the mechanisms upstream to NCX reverse mode activation and initiating the procoagulant transition are not entirely understood.

Platelet functions are rapidly modulated by (de-)phosphorylation of proteins, which plays a key role in platelet signal transduction.^3,12,13 Analysis of protein phosphorylation (phosphoproteomics) has been used to investigate resting^14,15 and preactivated platelets¹⁶ and mechanisms reversing their activation.¹⁷ Phosphoproteomics has also been used to explore downstream signaling upon GPVI activation in healthy donors^13,18 and patients with Scott syndrome.¹² Moreover, the study of temporal phospho-regulation upon ADP stimulation¹⁹ modulated by iloprost²⁰ has demonstrated the possibility of assessing timing and sequence of phosphorylation events in platelet activation. However, to our knowledge, time-dependent changes in the phosphoproteome of aggregatory platelets differentiating into procoagulant COAT platelets have not been investigated to date.

Thus, the aim of this study was to identify phosphorylation events differing during the development of procoagulant COAT platelets from aggregatory ones. To do so, we employed temporal liquid chromatography-mass spectrometry–based phosphoproteomics on activated platelets undergoing aggregatory or procoagulant transformation. Using this novel approach, we identified several early differentially phosphorylated proteins that (to our knowledge) are highlighted for the first time as potential candidate regulators of the platelet procoagulant transition. Our observations and current evidence provide the rationale for further investigations into the role of these phosphoproteins.

Methods

Complete information on materials, methods, and analytical workflow is detailed in the supplemental Data.

Experimental design

As extracellular calcium is necessary to shift an aggregatory phenotype to a procoagulant one,^3,21 washed human platelets (supplemental Figure 1) from healthy donors (n = 5) were activated with convulxin (CVX) plus thrombin (THR) in absence or presence of extracellular calcium, inducing the generation of >80% aggregatory and >90% procoagulant COAT platelets, respectively (supplemental Figure 2). Activated platelets were lysed at baseline and at 9 different time points after activation up to 8 minutes as shown in Figure 1. This experimental design, by comparing activation profiles developing simultaneously under 2 different conditions, enabled us to capture early differences in the phosphoproteome of platelets undergoing the procoagulant COAT transition vs platelets remaining in the aggregatory state.

Figure 1.

Experimental design and phosphoproteomic workflow. Washed platelets (light blue) were stimulated with CVX + THR for 8 minutes in absence or presence of 5mM extracellular calcium, thus creating conditions in which mainly aggregatory (dark blue, >80%) or procoagulant COAT platelets (red, >90%) were generated (refer to supplemental Figure 2 for quantification of the subpopulation generated). The experimental design with subsampling at various time points enables temporal analysis of the phosphoproteome, capturing early phosphorylation events before and during the phenotypic shift from aggregatory to procoagulant COAT platelets. A standard phosphoproteomic workflow was then applied, consisting of platelet lysis, protein digestion, and protein labeling (using tandem mass tag [TMT]) of each sample. Then, 10 samples (baseline plus 9 indicated time points) were multiplexed for phosphopeptides enrichment. MS quantitative analysis was performed on each TMT 10-plex mix for total proteome and phosphoproteome. Further data processing is described in the supplemental Methods. LC-MS/MS, liquid chromatography–tandem mass spectrometry. Figure created with biorender.com. Aliotta A. (2026) https://biorender.com/rtnerq1.

Phosphoproteomic workflow

The several steps of the phosphoproteomic analysis are displayed in Figure 1. Briefly, the proteins from platelet lysates were denatured and linearized, purified on beads, and digested with trypsin. The peptide mixtures of each sample for a given donor, condition, and time point were labeled using tandem mass tag 10-plex (Thermo Fisher Scientific, Waltham, MA). Thus, 10 batches (5 donors and 2 conditions) of 10 multiplexed samples (each time point) were generated. The phosphopeptides of each 10-plex were enriched by immobilized metal affinity chromatography. Phospho-enriched peptide mixtures were separated by reversed-phase liquid chromatography and analyzed by mass spectrometry for identifying the proteins and quantifying their phosphorylation levels. Log₂-transformation of peak intensity and several normalization steps accounting for total material injected and basal phosphorylation intensity were performed. Finally, only the phosphorylation sites quantified in all samples were retained for analysis.

Definition of the phosphorylation events of interest

A clustering analysis was performed to highlight different phosphorylation patterns during platelet activation in aggregatory and procoagulant COAT conditions. To select relevantly regulated phosphosites, the mean difference in the log₂-transformed phosphorylation intensity (relative fold change) was calculated in procoagulant COAT vs aggregatory conditions at all time points for each phosphosite. The phosphosites were considered relevant when the relative fold change was ±0.5 (ie, an absolute increase to >1.4 and a decrease to <0.7) and if the P value was < .05 when using Mann-Whitney test. “Hypophosphorylation” and “hyperphosphorylation” refer to lower or higher phosphorylation in procoagulant COAT relative to aggregatory condition, respectively. “Dephosphorylation” and “phosphorylation” refer to changes relative to baseline.

Results

Time-dependent changes in the phosphoproteome of aggregatory and procoagulant COAT platelets

We investigated time-dependent changes in the phosphoproteome of platelets from healthy donors (n = 5) during 8 minutes after stimulation with CVX + THR (Figure 1). Of 13 263 phosphosites identified in the complete phosphoproteome data set, 1404 were consistently quantified in the 5 donors and both phenotype-generating conditions at all time points (supplemental Table 1B). A clustering analysis identified different phosphorylation patterns in the first minute of activation in both conditions (Figure 2A): 2 clusters with marked phosphorylation increase (clusters 1 and 2), 3 clusters with steady phosphorylation (clusters 3, 4, and 5), and 1 cluster with decreased initial phosphorylation (cluster 6). Notably, 2 minutes after activation, a pronounced dephosphorylation was observed in the procoagulant COAT condition for all clusters except cluster 4, regardless of the initial amplitude of phosphorylation. A similar dephosphorylation pattern occurring around 2 minutes after activation was observed for each subset of 1175 serine-, 191 threonine-, and 38 tyrosine-phosphorylated phosphosites (Figure 2B-D). Notably, analysis of the proteome of aggregatory and procoagulant COAT platelets showed that the dephosphorylation in procoagulant COAT platelets is not due to a change in protein levels (data not shown).

Figure 2.

Global phosphorylation dynamics of aggregatory and procoagulant COAT platelets. Platelets were stimulated by convulxin plus thrombin in aggregatory (AGG) and procoagulant COAT platelet–generating conditions and sampled at different time points for 8 minutes (Figure 1). The 1404 phosphosites consistently quantified in the 5 donors, in both conditions, and at all time points were analyzed. (A) The heat map displays 6 clusters of distinct phosphorylation profiles. Hyperphosphorylation and hypophosphorylation are represented in yellow and purple, respectively, at different time points by displaying log₂ intensity RFC to baseline (set as 0, turquoise color). On the right side, the mean log₂ intensity RFC in the AGG (black lines) and procoagulant COAT (red lines) phenotype-generating condition is displayed for each cluster. (B-D) The mean and standard deviation of the median phosphorylation log₂ intensity in AU is displayed over time for the AGG (black squares and lines) and procoagulant COAT (red circles and lines) conditions for serine (B; 1175 phosphosites), threonine (C; 191 phosphosites), and tyrosine (D; 38 phosphosites) residues. Statistical significance is defined as ∗ P < .05 and ∗∗ P value < .01, Mann-Whitney test. AU, arbitrary units; RFC, relative fold changes.

Temporal analysis of phosphorylation events linked to the aggregatory-procoagulant dichotomy

To correlate the potential regulatory role of differential phosphorylation events with the procoagulant response, the temporal phospho-regulation in procoagulant COAT vs aggregatory conditions was dissected into 3 phases (supplemental Table 1B). The first minute was defined as “initiation phase” (preceding the procoagulant response); the time window from 1 to 3 minutes after activation as “development phase” (onset of procoagulant COAT platelet generation), and the period from 3 to 8 minutes as “consolidation phase” (stable procoagulant COAT platelet generation). The last timeframe is not addressed in this article, because our primary aim was to highlight early regulators of the procoagulant response.

Despite observing a similar global phosphorylation profile in aggregatory and procoagulant COAT conditions in the initiation phase (0-1 minutes after activation; Figure 2), analysis of individual phosphosites showed that out of 1404 phosphosites, 43 (corresponding to 40 proteins) were already differentially phosphorylated, transiently or in a sustained manner, in procoagulant COAT vs aggregatory conditions. This corresponded to 38 hypophosphorylated and 5 hyperphosphorylated phosphosites depicted in Figure 3A and supplemental Figure 3. An interactome based on published data of 40 differentially phosphorylated proteins was drawn and 17 proteins already known to interact with each other were identified with a confidence score of >0.5 according to the STRING interaction database (https://string-db.org/; supplemental Figure 4). Among these differentially phosphorylated proteins, there are (1) kinases (SRC and PRKCD); (2) calcium, sodium, hydrogen, and chloride channels and regulators (SLC9A1, SLC4A2, STIM1, and PANX1); (3) cyclic nucleotide regulators (PDE3A and ADCY6); (4) cytoskeleton interactors (cortactin [CTTN] and filamin A [FLNA]); and (5) other regulatory proteins (ARRB1, PECAM1, and FYB1) known to interact with each other. Notably, among the proteins not considered by the interactome, there are several of particular interest, such as ARHGAP6, MRVI1, and DMTN (refer to “Discussion”).

Figure 3.

List of phosphosites potentially involved in the procoagulant response. (A) Initiation phase (0-1 min after activation). All 43 phosphosites differently phosphorylated in procoagulant COAT vs aggregatory conditions are listed here. (B) Development phase (1-3 minutes after activation). All 15 hyperphosphorylated phosphosites, the 15 most hypophosphorylated as well as 2 transiently hypophosphorylated phosphosites are listed. Protein classification of Huang et al¹⁰⁰ was used. Notably, the class 17 is in squared brackets as its function is putative. At given time points, statistically significant (P < .05, Mann-Whitney test, n = 5) and differentially phosphorylated phosphosites in procoagulant COAT vs aggregatory conditions are displayed in red if the mean relative log2 fold change was >0.5, in pink if >0 to ≤0.5, in pale blue if greater than or equal to −0.5 to <0, and in blue if less than −0.5. For each phosphosite, possible implication of the protein in the procoagulant response, according to the current understanding of its function in platelets or in other cells, were added. For additional information, refer to supplemental Table 1C for panel A and supplemental Table 1D for panel B. ER, endoplasmic reticulum; GPCR, G protein-coupled receptor; PAR1, protease-activated receptor 1; P2Y12, purinergic ADP receptor; SOCE, store-operated calcium entry.

In addition, during the development phase (1-3 minutes after activation), we observed 15 hyperphosphorylated and many hypophosphorylated phosphosites, of whom the 15 most prominent and 2 transiently hypophosphorylated are represented in Figure 3B and supplemental Figures 5 and 6.

Discussion

Our results show that (1) procoagulant COAT platelets undergo a global dephosphorylation (Figure 2), which is largely independent from proteolysis (both observations are in agreement with previous findings)^3,12,22; and (2) several differential phosphorylation events already occur before the emergence of the procoagulant phenotype, which becomes visible ∼2 minutes after activation.^2,3,7,23 Notably, in the first minute of activation, no differential phosphorylation patterns were observed for canonical signaling proteins of platelet activation, such as PKC, SYK, and AKT (supplemental Table 1B). This is most likely due to their central role during the initial global platelet activation and their engagement immediately downstream of receptor activation, which makes them unsuitable for modulating the procoagulant differentiation that occurs later. Therefore, we hypothesize that several other differential phosphorylation events, shown in Figure 4, may initiate and sustain the switch from aggregatory to procoagulant phenotype in light of their timing and the current knowledge on the protein functional role.

Figure 4.

Proteins and phosphosites with relevant diverging phosphorylation profiles in the initiation and development phases. The phosphosites are presented in the order they are addressed in the “Discussion.” The phosphorylation profile (log₂ intensity relative to baseline) is shown for the procoagulant COAT (red line and circles) and aggregatory (AGG, black line and squares) platelet–generating condition as mean ± standard deviation (n = 5). A Mann-Whitney test was performed for each time point (procoagulant COAT vs AGG conditions). Statistical significance is defined as ∗ P < .05 and ∗∗ P < .01. Refer to supplemental Table 1C-D for additional information.

Dichotomous ion fluxes drive the procoagulant response

To date, NCX reversing is the earliest event known to commit platelets to the procoagulant phenotype, by promoting a high and sustained cytoplasmic calcium influx, a defining feature of procoagulant COAT platelets.^3,7 In our data set (Figures 3 and 4; supplemental Figures 3 and 5-6), we observed several differential phosphorylation events that could induce and maintain the NCX reverse mode of function and other events that may directly promote high cytosolic calcium concentrations.

Generating a local increase in cytoplasmic sodium to regulate NCX functional mode

It is known that a high intracellular sodium level shifts NCX to its reverse mode, which extrudes cytoplasmic sodium in exchange for calcium.^7,24 During the initiation phase, we observed a hypophosphorylation of the sodium/hydrogen antiporter SLC9A1 (NHE1) at S796 in procoagulant COAT compared with aggregatory conditions (Figure 4). S796 is part of the binding site for carbonic anhydrase II.²⁵ The latter increases SLC9A1 activity by generating protons and has been shown to specifically regulate the procoagulant response.²⁶ Work by Golaszewska et al showed that the adrenaline-(co)induced procoagulant response could be strongly reduced upon SLC9A1 inhibition,²⁷ suggesting a similar role in CVX + THR–induced procoagulant response. Interestingly, SLC9A1 is phosphorylated at S796 after amotosalen/ultraviolet A treatment of platelet concentrates,²⁸ which have a decreased ability to generate procoagulant COAT platelets.²⁹ Based on these data, we propose that the decreased phosphorylation of SLC9A1 at S796 may increase intracellular sodium in exchange for excess of protons generated by platelet metabolism,30, 31, 32 thus priming NCX reverse mode.

Reducing calcium extrusion via PMCA

The plasma membrane calcium ATPase (PMCA) is an ATP-activated calcium channel extruding calcium to maintain cytoplasmic calcium homeostasis.³³ PMCA is downregulated in procoagulant platelets and its inhibition increases phosphatidylserine (PS) exposure.^33,34 During the development phase, ATP2B1 (PMCA isoform 1) experienced a rapid and transient dephosphorylation at S1155 in procoagulant COAT condition (Figure 4). We therefore propose that PMCA dephosphorylation at S1155 may be an early inhibitory process that inhibits PMCA activity and ensures an elevated cytosolic calcium, thereby supporting the procoagulant state.

Regulating calcium fluxes via phosphatidylinositol transfer proteins

PITPNM2 (phosphatidylinositol transfer protein membrane associated 2, Nir3) and C2CD2L (C2 calcium-dependent domain containing 2, also known as TMEM24, another phosphatidylinositol relocator) regulate calcium fluxes between the plasma membrane, endoplasmic reticulum, and mitochondria.35, 36, 37, 38 Notably, neither has been investigated in platelets thus far. Both were transiently hyperphosphorylated during the development phase in procoagulant COAT platelets (PITPNM2 at S305 and C2CD2L at S662 in its C-terminal region) compared with aggregatory ones (Figure 4), which is consistent with a role in coordinating intracellular calcium shifts between compartments, as described in other cell types.^36,39, 40, 41

Additional calcium-regulating proteins

Our data indicate that other early differential phosphorylation events may increase cytosolic calcium, committing activated platelets to become procoagulant (Figure 4). The dephosphorylation of STIM1 S575 (driver of store-operated calcium entry),⁴² also observed by Babur et al upon GPVI stimulation,¹⁸ and of DMTN S16/S26 (regulator of calcium flux from the dense tubular system [DTS]) in the procoagulant COAT platelets could contribute to an increased calcium entry into the cytosol and release from internal stores, respectively.⁴³ Finally, the striking dichotomous phosphorylation of the chloride/bicarbonate exchanger SLC4A2 at S144, starting 1 minute after activation (Figure 4), may promote chloride entry (required for full PS exposure) and sustain SLC9A1-dependent cytosolic sodium increase.^26,44

Aggregatory platelets overcoming intrinsic inhibitory barriers become procoagulant

The differential phosphorylation events described in the following sections may reflect an imbalance in inhibitory barriers of aggregatory platelets allowing them to become procoagulant.

Dysregulation of cyclic AMP/GMP

Cyclic adenosine-monophopsphate (cAMP) and cyclic guanosine-monophosphate (cGMP) maintain platelets in a quiescent state by activating PKA and PKG to inhibit platelet activation pathways (eg, calcium release from the DTS).^45,46 Cytoplasmic levels of cAMP and cGMP are the result of a fine-tuned balance between synthesis by adenyl cyclase (ADCY) and guanylyl cyclase (GC), respectively, and their degradation by phosphodiesterases (PDE).⁴⁷

During the initiation phase, we observed the dephosphorylation of PDE3A at S408 (Figure 4), hypophosphorylation/dephosphorylation of PDE3A at S438, S520 in the development phase, and a slight hypophosphorylation at S428 only significant at 8 minutes (supplemental Table 1B). These observations fit with the notion that PDE3A (at S438) is regulated by PKA,⁴⁸ and inhibition of PKA via the inactivation of ADCY facilitates procoagulant COAT platelet formation.^27,49,50 Based on these data and observations in cardiomyocytes,⁵¹ we hypothesize that the lower phosphorylation of PDE3A could reflect a mechanism reducing calcium re-entry in DTS via inhibition of SERCA2 (sarcoplasmic reticulum Ca²⁺ ATPase 2), hence maintaining high cytosolic calcium.^48,52

The downstream signaling of the α2-adrenergic receptors (ADRA2A) lowers cAMP level, which facilitates platelet activation by other agonists, increasing aggregation and the procoagulant response.^27,53,54 During the development phase, a peculiar transient hyperphosphorylation of the ADRA2A at S346 was observed in the COAT condition (Figure 4). This differential phosphorylation may induce differential cAMP levels in the procoagulant COAT platelets vs aggregatory ones.

MRVI1 (murine retrovirus integration site 1 homolog), also known as IRAG1 (inositol 1,4,5-triphosphate receptor associated 1), is regulated by PKG and prevents calcium release from the DTS.^55,56 During the initiation phase, we observed a marked dephosphorylation of MRVI1 at S689, which started 1 minute after activation and persisted in the procoagulant COAT platelets, whereas in aggregatory platelets, it was phosphorylated back to the resting levels (Figure 4). This suggests that the MRVI1-dependent inhibitory barrier for calcium release is lowered in the procoagulant COAT platelets and could therefore contribute to the high cytosolic calcium required for the procoagulant response.

Differential phosphorylation of GPVI signaling inhibitors

The hypophosphorylation of the GPVI signaling inhibitor PECAM1 at S726 in the regulatory C-terminal domain57, 58, 59 and the dephosphorylation of the LAT competitor LAT2⁶⁰ (LAB) S135 may inhibit their negative regulatory functions, thereby prolonging GPVI downstream signaling in the procoagulant COAT platelets (Figure 4). We hypothesize that differential phosphorylation of these components of GPVI feedback mechanisms primes platelets to become procoagulant upon CVX + THR activation.

The role of GTPases and cytoskeleton remodeling in the procoagulant response

Rho-GTPases have been increasingly linked to the procoagulant response.61, 62, 63 For instance, the RhoA pathway participates in various aspects of the development of procoagulant activity, such as translocation of coagulation factor XIII, binding of factor V on the surface of procoagulant COAT platelets,^62,64 and RAP1-dependent exposure of PS.⁶³

Starting in the initiation phase, we observed the dephosphorylation of several GTPase regulators such as ARHGAP6 (S772, S820), RALBP1 (S48, S62), RAB8A/B (T72), and AGFG1 (S181) (Figure 4; supplemental Figure 3). Of major interest is ARHGAP6, a platelet specific GTPase-activating protein inactivating RhoA, which exhibits a differential phosphorylation as early as 10 second after activation and is among the most dephosphorylated proteins in the development phase (S820; Figure 4).^65,66 Although the specificity of these regulators are (mostly) elucidated, the functional role of their (de-)phosphorylation is currently unknown.⁶⁵ The strong dephosphorylation of ARHGAP6 in the procoagulant COAT condition suggests its role in preserving RhoA activation, which is required for cytoskeleton remodeling and procoagulant activity.⁶⁷

FLNA and the myosin phosphatase subunit PPP1R12A (MYPT1) are involved in RhoA-dependent platelet shape change whereas CTTN is involved in the CDC42/Rac1-dependent shape change.68, 69, 70 The early dephosphorylation of FLNA at S2163, CTTN at S47/Y421, and the hyperphosphorylation of PPP1R12A at S871 in the development phase (Figure 4) may modulate the balance between actin contractility and myosin phosphatase activity possibly necessary for procoagulant cytoskeleton remodeling.70, 71, 72

Kinases driving the procoagulant response

Our data set highlights several differential phosphorylations of kinases possibly mediating key changes in intracellular signaling pathways, which could drive the aggregatory to procoagulant transformation.

SRC plays a dual role in platelet activation. SRC is recruited downstream of GPVI and GPIb activation^73,74 whereas a pool of SRC is constitutively associated with the cytoplasmic tail of integrin β3 (ITGB3) to positively regulate outside-in signaling.^4,75,76 We observed a slightly delayed phosphorylation in the initiation phase at S17 and a sustained hypophosphorylation at T72 in the development phase (Figure 3; supplemental Figures 3 and 6). The differential phosphomodulation of SRC at its N terminus (involved in membrane targeting, interaction with integrin αIIbβ3, and initiation of signaling pathways) and its central role in the interactome of the initiation phase (supplemental Figure 4) may suggest a possible pivotal role in the procoagulant phenotypic switch.⁷⁷

LYN, a member of the SRC family kinases, is known to have antithetical roles downstream of GPVI vs protease-activated receptor (PAR) stimulation with regard to dense-granule secretion and platelet aggregation, and is part of a PECAM1-dependent inhibitory pathway.78, 79, 80 We observed a transient hyperphosphorylation of LYN Y397 followed by dephosphorylation (Figure 4). This peculiar regulation of LYN downstream of GPVI/PAR may be linked to the procoagulant response, possibly contributing to diverging signaling leading to integrin αIIbβ3 inactivation.

Upon mitochondrial depolarization, procoagulant COAT platelets produce reactive oxygen species.⁷ The hyperphosphorylation of the oxidant stress–activated serine/threonine kinase STK24 (MST3, STE20) T190 is therefore intriguing (Figure 4). Indeed, STE20-like kinase, a similar protein, has been shown to be activated and play a role in granules secretion and calcium mobilization upon GPVI or PAR-stimulation.⁸¹ STK24 could therefore link a functional response to the ongoing oxidative stress in procoagulant COAT platelets.

The mitogen-activated protein kinases MAP4K4 has recently been shown to be involved in overall platelet activation.⁸² The hyperphosphorylation of MAP4K4 at Y182 could indicate an active role in the procoagulant response (Figure 4).

Mechanisms supporting fibrinogen receptor αIIbβ3 inactivation

Downregulation of integrin αIIbβ3 is one of the hallmarks of procoagulant COAT platelet transition from aggregatory ones.^4,11 Several differentially phosphorylated proteins observed in our dataset may regulate the inactivation of the fibrinogen receptor.

Intracellular phosphorylation events possibly driving integrin αIIbβ3 downregulation

The mechanism activating integrin αIIbβ3 is an “inside-out” signaling process primarily mediated by talin-1 (TLN1) binding to the cytoplasmic tail of integrin β3 (ITGB3).^83,84 It is known that the intermediate form of active calpain (78 kDa) generated in the first few seconds upon platelet stimulation participates in the activation of integrin αIIbβ3 through limited cleavage of talin, and that the fully autolyzed form (76 kDa), appearing several minutes later, mediates a partial downregulation of αIIbβ3 by cleaving β3, talin, and SRC, thus disrupting integrin-cytoskeletal interactions.^4,85,86 Our observations suggest that phosphorylation events precede and possibly complement calpain-induced αIIbβ3 downregulation.

FYB (FYN binding protein, ADAP, SLAP-130) bridges TLN1 and kindlin-3 to ITGB3, thus facilitating integrin αIIbβ3 activation and irreversible fibrinogen binding.⁸⁷ In the initiation phase, after an initial dephosphorylation in both conditions, we observed a marked rephosphorylation of FYB Y571 in the aggregatory condition starting 1 minute after activation (Figure 4). This differential phosphorylation may regulate FYB localization and its association with TLN1, modulating αIIbβ3-mediated thrombus stabilization.⁸⁸ Moreover, after an initially marked phosphorylation in both conditions, we observed an early hypophosphorylation of TLN1 at S429 in procoagulant COAT platelets (Figure 4). Because phosphorylation at S425/S429 protects TLN1 from ubiquitylation,⁸⁹ dephosphorylation at S429 could promote its degradation, contributing to the downregulation of αIIbβ3 characteristic of procoagulant COAT platelets.^3,11,23

Several phosphosites (T767, S778, T779; supplemental Table 1B) located in the cytoplasmic tail of β3 were hypophosphorylated in procoagulant COAT vs aggregatory condition already 2 to 3 minutes after activation. This coincides with the development phase of procoagulant COAT platelets and is coherent with the observation that tyrosine dephosphorylation of β3 sensitizes it to calpain-mediated inactivation.⁹⁰

Finally, RASA3, a GTPase-activating protein, negatively regulates platelet activation by inhibiting Rap1.⁹¹ This small GTPase directly regulates integrin αIIbβ3 activation by facilitating its interaction with TLN1.^91,92 RASA3 S833 was dephosphorylated in aggregatory platelets compared with procoagulant ones during the development phase (Figure 4). We hypothesize that this selective dephosphorylation in aggregatory platelets may enhance Rap1-dependent TLN1 function, sustaining integrin αIIbβ3 activation,⁹³ whereas the maintained phosphorylation in procoagulant COAT platelets would allow integrin αIIbβ3 downregulation.

Altogether, these observations suggest that early modulations within the FYB-Rap1-TLN1-ITGB3 axis could play a pivotal role in cytosolic integrin αIIbβ3 downregulation in procoagulant COAT platelets.⁹⁴

Extracellular modulation of integrin αIIbβ3

PDIA6 (platelet disulfide isomerase family A member 6) belongs to the thiol isomerase family that is secreted and recruited to the platelet surface upon CVX, collagen, or THR activation.⁹⁵ PDIA6 regulates the extracellular part of ITGB3 and recent data indicated its participation in the disengagement of fibrinogen from activated integrin αIIbβ3.^96,97 We observed that PDIA6 was markedly phosphorylated at S156 (next to a thioredoxin domain) in the procoagulant COAT condition during the development phase (Figure 4). We therefore hypothesize that the differential phosphorylation of PDIA6 might represent a new mechanism of integrin αIIbβ3 inactivation complementary to the current TLN1-dependent intracellular model.⁴

Limitations

Our approach may not identify all transitional states from aggregatory to procoagulant COAT platelets. Nonetheless, it highlighted divergent phosphorylation patterns as early as 10 seconds after activation. These early events would not have been detected using conventional methods for isolating different phenotypes, such as cell sorting.⁹⁸

A relevant limitation of our work is that the functional impact of several differential phosphorylation events, and even the role of some proteins identified by our phosphoproteomics approach, are unknown. This greatly hinders the interpretability of the data. For instance, there is limited published knowledge to interpret the functionality of the hyperphosphorylation of PCNP S87, TNIK S610, MYCT1 S108, UBE2O S515, and GAPDH T211 as well as the hypophosphorylation of SAMD14 S256, F11R S287, and CD84 Y316 in procoagulant COAT vs aggregatory conditions (Table 1, supplemental Figure 3).

To select the strongest changes in procoagulant COAT vs aggregatory platelets, we used a stringent refinement of the phosphosites. Although this approach provided an exhaustive list of candidate proteins for further study, it eliminated potentially interesting data (eg, DAPP1 Y139; supplemental Table 1B).⁹⁹

The major limitation of our work is its hypothesis-generating character. However, this study used a robust temporal phosphoproteomic approach to delineate phosphorylation events and pathways associated with the emergence of the aggregatory and procoagulant COAT platelet phenotypes. We correlated these events with current evidence, highlighting their potential role in the procoagulant transformation of aggregatory platelets, and we therefore provide a strong rationale for future mechanistic investigations. In order to confirm the diverging phosphorylation profiles, spatio-temporal regulation, and functional relevance of the phosphoproteins discussed here, more studies are warranted, in particular experiments using pathway modulators combined with phospho-specific analysis, and site-directed mutant models.

Conclusions

Our work highlights the usefulness of temporal phosphoproteomics for identifying previously undescribed candidate regulators of the dichotomous aggregatory-procoagulant platelet response. We summarized our mechanistic hypotheses by proposing a model (Figure 5) of aggregatory platelets acquiring specific (“preprocoagulant”) phosphorylation features possibly underlying procoagulant transformation. Then, in the resulting “early stage” procoagulant COAT platelets, additional phospho-regulated mechanisms could support the terminal differentiation. Further studies of the candidate regulators and pathways proposed could lead to the identification of novel druggable targets modulating platelet procoagulant response to rebalance hemostasis.

Figure 5.

Model of diverging phosphorylation in key signaling pathways necessary for the procoagulant transformation. The differential phosphorylation patterns observed in this study could affect several mechanisms likely to facilitate the differentiation of AGG platelets into procoagulant COAT ones. The main mechanisms highlighted in the initiation phase are (1) increased sodium and calcium mobilization (eg, DMTN, SLC4A2, SLC9A1, SRC, and STIM1), (2) reduced inhibitory barriers (MRVI1, PDE3A, LAT2, and PECAM1), (3) differential cytoskeleton regulation (ARHGAP6, CTTN, and FLNA, and later PPP1R12A) and in the development phase (4) sustained calcium mobilization (ADRA2A, ATP2B1, C2CD2L, and PITPNM2) and (5) inactivation of integrin αIIbβ3 (FYB, ITGB3, PDIA6, RASA3, and TLN1). For clarity, only the phosphorylation events (white boxes with black outlines) displayed in Figure 4 are represented (refer to “Discussion” section for more details). Hyperphosphorylation (orange) and hypophosphorylation (blue) refer to the comparison of procoagulant COAT vs AGG platelets. CAII, carbonic anhydrase II; DAG, diacylglycerol; IP3-R, receptor of inositol triphosphate; mPTP, mitochondria permeability transition pore; Rac1, Ras-related C3 botulinum toxin substrate 1; ROCE, receptor-operated calcium entry. Figure was generated using SMART elements (smart.servier.com).

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