Differential effects of physiological agonists on the proteome of platelet‐derived extracellular vesicles

Arterial thrombosis manifesting as heart attack and stroke is the leading cause of death worldwide. Platelets are central mediators of thrombosis that can be activated through multiple activation pathways. Platelet‐derived extracellular vesicles (pEVs), also known as platelet‐derived microparticles, are granular mixtures of membrane structures produced by platelets in response to various activating stimuli. Initial studies have attracted interest on how platelet agonists influence the composition of the pEV proteome. In the current study, we used physiological platelet agonists of varying potencies which reflect the microenvironments that platelets experience during thrombus formation: adenosine diphosphate, collagen, thrombin as well as a combination of thrombin/collagen to induce platelet activation and pEV generation. Proteomic profiling revealed that pEVs have an agonist‐dependent altered proteome in comparison to their cells of origin, activated platelets. Furthermore, we found that various protein classes including those related to coagulation and complement (prothrombin, antithrombin, and plasminogen) and platelet activation (fibrinogen) are attributed to platelet EVs following agonist stimulation. This agonist‐dependent altered proteome suggests that protein packaging is an active process that appears to occur without de novo protein synthesis. This study provides new information on the influence of physiological agonist stimuli on the biogenesis and proteome landscape of pEVs.


INTRODUCTION
Arterial thrombosis, manifesting as myocardial infarction and ischemic stroke remains the leading cause of death and disability globally.
Platelets play a fundamental role in mediating both hemostasis and pathological thrombus formation [1].Typically, platelets circulate in a quiescent state, but react rapidly in response to vascular injury to seal the injured vessel and ensure hemostasis.Importantly, it is now appreciated that thrombi have a heterogenous architecture which reflects the activation state of platelets within the thrombus [2][3][4][5][6].Indeed, platelet thrombi comprise two zones, an inner core and outer zone.The inner core is composed of highly activated platelets typically proximal to the region of vascular injury.These inner core platelets are tightly packed with the formation of this region being largely dependent on thrombin activity.In contrast, the outer zone is comprised of loosely packed platelets and is largely sensitive to Gi-linked signaling, such as induced by ADP [2][3][4][5][6].
Whilst the differential response of platelets to soluble agonist activation, and consequent thrombus formation has been extensively studied [2][3][4][5][6], less is known on how different platelet agonists regulate the formation of platelet-derived extracellular vesicles (EVs).EVs play a major role in cell-cell communication and signaling through the transport of microRNAs (miRNA) [7][8][9], receptors [10,11], and proteins [12] to cells either by membrane-based interactions, or through uptake into the target cell.
Though EVs are secreted from numerous cells, platelet-derived EVs (pEVs) are the most abundant EVs in the circulation in healthy humans [13], and increase in number in patients with cardiovascular disease [14,15].pEVs have a diameter range of approximately 100−250 nm; are negatively charged at the zeta potential; possess surface anionic phospholipids, cell-derived antigens, cytokines, and matrix metalloproteinases; and contain various mRNAs and miRNAs [16].Because pEVs are derived from platelets, they possess markers on the membrane surface, such as CD41 and CD62P, that mediate the targeting of pEVs to local inflammation and tumor sites [17].Due to their bioactive cargo, cellular origin, and innate capacity for circulation, their advantages for use in drug-loaded therapy are significant.pEVs deliver proteins, lipids, nucleic acids, and other biologically active molecules to target cells to regulate cellular functions.
Significantly, there is a growing body of evidence now implicating pEVs in mediating a range of thrombotic and inflammatory disorders.
Indeed, pEVs have been demonstrated to play a multifaceted role, both positive and negative, in a wide range of diseases [18].For example, pEVs have been shown to enhance metastasis through upregulation of matrix metalloproteinases in prostate [19] and lung [20] cancers promoting cancer cell invasion.Conversely, there is evidence that pEVs can infiltrate solid tumors to release miRNAs that inhibit tumor growth [8], highlighting the heterogeneous nature and effects of pEVs.pEVs are also implicated in inflammatory diseases such as rheumatoid arthritis, where they induce an inflammatory response through interleukin-1 [21,22]; however, as with cancer, pEVs have also been shown to have anti-inflammatory effects in plasmacytoid dendritic cells [23,24].In thrombosis and hemostasis, pEVs were first described as procoagulant "platelet-dust," [25] and have subsequently been shown to play important roles in hemostasis, as evidenced by bleeding phenotypes in patients who do not produce pEVs [26], though this procoagulant nature remains uncertain as it may be influenced by pEV isolation techniques [13,27].Cardiovascular diseases such as atherosclerosis are associated with increased numbers of circulating pEVs where they are implicated in atherogenesis through transport of miR-223 [28], as well as through interactions with immune cells enhancing recruitment to the plaque [15,29].
Given the wide range of roles that pEVs have in regulating health and the many disease states in which pEVs are involved, it stands to reason that the contents of pEVs must be heterogeneous to accommodate for the wide range of functions.Although various mechanisms associated with EV biogenesis are described (reviewed Dixson et al. [30]), including endosomal maturation [31], plasma membrane budding, and cytokinesis [32], the cell state influences which cellular components are ultimately packaged in EVs [30].For example, processes influencing cytoskeletal rearrangement, calpain activation [15], and the activation state of the cell impact the cellular context and cargo in EVs.
Previous literature has also suggested that the isolation method can influence the size, number and form of EVs [33], with some studies suggesting contradictory roles for pEVs in coagulation based on isolation method [13,27,34].Similarly, proteomic analyses of pEVs have revealed an association between platelet protein content and pEV size [35], and platelet agonist used to stimulate pEV production [36][37][38].
Of note, Milioli et al. [36] used quantitative proteomics to show that protein abundance in pEVs is influenced by agonist potency.In this study, proteins related to platelet activation were more abundant in pEVs produced by strong agonists (thrombin + collagen) than weaker agonists (ADP) compared to the control group (thrombin-generated pEVs) [36].While the previous work in this field has been highly informative as to how the pEV proteome can be altered by the technique used to generate pEVs [36][37][38][39], a knowledge gap remains in addressing how different platelet agonists impact the proteome composition of pEVs compared to resting and activated platelets, and how proteins are sorted and packaged from platelets into pEVs.Furthermore, little is known about how different platelet activation pathways induced by specific platelet agonists alter the protein cargo of pEVs.
Here, we utilized isolated human platelets stimulated with physiological platelet agonists which represent the heterogeneity of platelet activation within a thrombus to generate distinct pEV populations.
Agonists included collagen (mediates initial platelet adhesion to damaged endothelium) [40], thrombin (essential for potent and prolonged platelet activation in the thrombus core) [4], and ADP (representing a weak soluble agonist important for platelet activation in the thrombus shell) [4].Additionally, EVs were generated from a dual-stimulation condition combining collagen and thrombin, which represents the highly activated platelets at the heart of the thrombus core.Our study reveals that pEVs encapsulate protein cargo that is agonist-and EV-dependent.Importantly, we show that platelets actively package proteins from the activated platelet into the released pEVs, providing insight into how platelets release EVs in response to specific physiological stimuli.

Generation of platelet extracellular vesicles from human washed platelets
All experiments using human donors were approved by the Alfred Health Human Ethics Committee (project number: 627/17).Human washed platelets were obtained from five individual donors as described previously [41].After a resting period of 30 min, washed platelets were stimulated with either ADP (20 µM, Sigma-Aldrich), collagen (10 µg/mL, Chrono-log), thrombin (1 U/mL, Sigma-Aldrich) or a combination of collagen and thrombin (10 µg/mL, 1 U/mL) for 30 min at 37 • C. Following stimulation, platelets were removed by centrifugation (12,000 × g, 3 min) and the supernatant containing pEVs was collected.pEVs were pelleted by centrifugation (10,000 × g, 1 h), the supernatant was discarded and pEVs were resuspended in 0.2 µm filtered phosphate buffered saline (PBS) and stored at −80 • C prior to experimentation.Protein quantity of platelets and amount per sample was calculated using a micro-BCA kit (Thermo Scientific).

Flow cytometry
Platelet activation and EV release were confirmed using flow cytometry.Washed platelets were adjusted to a fixed platelet concentration of 5 × 10 7 platelets/mL and incubated with fluorescently labelled antibodies against the activated conformation of α IIb β 3 (PAC-1 FITC,

Significance Statement
In this study, we examined if the proteome of platelet-derived extracellular vesicles (pEVs) is influenced by the platelet agonist used to induce platelet activation.The agonists chosen were representative of the varied activation pathways of platelets within a thrombus.Importantly, we found significant differences in the pEV proteome compared to the proteome of platelets as their cell of origin, depending on the agonists used to generate the pEVs.A holistic overview of the pEV proteome revealed an upregulation of various classes of proteins including proteins involved in positive and negative regulation of complement and coagulation.
Differential proteome analysis of pEVs displayed specific packaging/sorting dependent on the agonist used to induce platelet activation.Furthermore, we highlight that the proteome of pEVs appears to be altered in the absence of de novo protein synthesis.This study provides new information on the biogenesis and proteome landscape of pEVs impacted by physiological agonist stimuli and improves our understanding, which could be leveraged to produce EVs with specific or enhanced biological functions.
5 µg/mL, BD), or a marker of platelet degranulation, P-selectin (CD62P, 0.3 µg/mL, BD) and incubated in the dark for 15 min [42].Following incubation, platelets were stimulated under the same conditions as during EV generation.Activated platelets were diluted in PBS and analyzed by flow cytometry (FACS Fortessa, BD).

EV image profiling
pEVs were profiled using the ONI EV Profiler Kit using antibodies against CD9, CD63, and CD81 to capture and immobilize pEVs.pEVs were labelled with fluorescent antibodies against CD9, CD63, and CD41 according to the manufacturer's instructions and imaged using dSTORM on the ONI Nanoimager [43,44].Images were analyzed using the CODI cloud platform (https://alto.codi.bio/;ONI) and the percentages of EVs with single, dual, or triple expression of CD9, CD63, and CD41 were calculated.
Unassigned precursor ions charge states and slightly charged species were rejected and peptide match disabled.Selected sequenced ions were dynamically excluded for 30 s.
For the generation of a platelet-specific spectral library, we employed pooled peptide mixtures from unstimulated or agoniststimulated human platelets or from pEV samples.Briefly, desalted peptide elutions were lyophilized by vacuum-based SpeedVac for 1 h and reconstituted in 25 mM ammonium formate, pH 10 for high pH reversed-phase (RP) microscale fractionation as per manufacturer's instruction (#84868, Thermo Fisher Scientific).For each sample group (platelet or pEV) a total of 9 fractions were pooled into 6 samples using alternating combinations and lyophilized by SpeedVac.Peptide samples were reconstituted in 0.07% trifluoracetic acid (TFA) and nano-MS/MS performed as detailed.
The mass spectrometry proteomics data (platelets, pEVs, and spectral library) have been deposited to the ProteomeXchange Consortium via the MASSive partner repository with the identifier MSV000093007.

Data processing and bioinformatics
Peptide identification and quantification were performed as described previously [48,49] using MaxQuant (v1.6.14) with its built-in search engine Andromeda [50].[51].Proteins identified in at least 70% of one biological group were selected for downstream analysis.

Statistical analysis
Data were analyzed using GraphPad Prism v8.4.3, with all data pretested for normality.If the data were non-parametric, a Kruskal-Wallis with a Tukey's post-hoc test or Mann-Whitney U analysis was performed.If parametric, one-way ANOVA with a Tukey's post-hoc test or unpaired t-test was performed.All data presented as mean ± standard deviation (mean ± SD).In all analyses, *p < 0.05 is considered statistically significant.

Production of pEVs from isolated platelets via three distinct platelet agonists
Human resting washed platelets were isolated from human whole blood and stimulated with physiological platelet agonists ADP (20 µM), collagen (10 µg/mL), thrombin (1 U/mL), or a combination of thrombin and collagen (1 U/mL, 10 µg/mL, respectively) which stimulate platelet activation through distinct pathways.Platelet activation was confirmed using flow cytometry as an increase in PAC-1 fluorescence (binds to the activated GPIIb/IIIa receptor) (Figure 1A) and a CD62P antibody (binds to P-selectin mobilized to the platelet surface on degranulated platelets) (Figure 1B).The hierarchy of agonist potency was ADP < collagen < thrombin ≤ thrombin/collagen (Figure 1A,B).pEVs were captured using a cocktail of anti-tetraspanin antibodies commonly expressed on EVs (CD9, CD63, and CD81) and were labelled with antibodies against the EV markers CD61, CD9, and CD41 (which labels the GPIIb subunit of the platelet-specific integrin GPIIb/IIIa) and profiled using super resolution microscopy (Figure 1C).As CD61 and CD9 are abundant on pEVs, we do not expect antigen detection to be impacted by the capture antibodies.The majority of EVs profiled were triple positive for CD63 + /CD41 + /CD9 + , with relatively few single positive EVs detected.In ADP-stimulated pEVs, a large proportion of double positive CD41 + /CD9 + pEVs were found, which was not observed when collagen or thrombin were used to stimulate platelets.
The total amount of protein in the pEV protein pellet following isolation was measured using a µ-BCA kit.There was no significant difference in the amount of protein that was detected comparing pEVs generated by different agonists; however, the overall trend in rank order of agonist potency could be observed, with ADP producing the least amount of pEV protein and thrombin/collagen the most (Figure 1D).

Proteome profiling of activated platelets and their EVs
Next, we subjected pEVs from activated platelets to single-pot, solidphase-enhanced sample preparation coupled with nanoLC-MS-based proteomics analysis to ascertain their proteome landscape (Figure 2A).
For comparative analysis, we also obtained proteome profiles of resting platelets and their activated counterparts using the same proteomics pipeline.By employing stringent peptides/proteins identification criterion (1% false discovery rate), we quantified over 1766 proteins in EVs and 1957 proteins in platelets (Figure 2B, Table S1).The platelet agonist thrombin stimulates platelets via its protease activity toward specific platelet membrane receptors.We did not see a detectable change in the proteome of platelets or platelet-derived EVs (Figure 2B, Table S1).For downstream analysis, we selected proteins that were quantified in at least 70% (>3/5 replicates) of each agonist stimulation group, which resulted in 1192 proteins with intensities subjected to variance-stabilizing normalization (Figure S1A and Table S2).
The platelet and derived pEV proteomes reported here are highly dynamic, spanning over nine orders of magnitude measured by their protein abundance (LFQ intensities) (Figure 2C).Unsupervised hierarchical clustering of protein expression profiles (Figure 2D) revealed that pEV proteomes were distinct to their platelet cell of origin proteomes.Moreover, Pearson's correlation coefficient of protein intensities between replicates in each group was over 0.8, indicating distinct and reproducible measurements (Figure S1B).Furthermore, we found that the majority of core EV proteins [55] including proposed universal markers of small EVs/exosomes, Syntenin 1 (SDCBP1) and MISEV (minimal information for studies of extracellular vesicles; MISEV2018 [56]) indicated that EV-typical proteins were enriched in pEVs compared to platelets (Figure 2E).On the other hand, the majority of cellular proteins (except EIF2H and PDAP1) that are proposed as exclusion proteins and typically not found in the EV proteome were enriched in platelets.Thus, our data show that we were able to ascertain a specific and meaningful EV proteome for pEVs.

Pair-wise comparison of platelets and their EVs highlight conserved EV function
To gain a holistic insight into EV biology upon platelet activation, we first performed a pair-wise analysis of pEV versus platelet proteomes (Figure 3, Table S2).Principal component analysis further highlights distinct proteomes of platelets and their derived EVs (Figure 3A).
We identified 113 proteins significantly enriched in EVs compared to platelets (fold change > 1.5 or < −1.5, p < 0.05) (Figure 3B).Conversely, we identified 56 proteins with higher abundance in platelets compared to EVs.This resulted in a total of 169 DAPs between platelets and their EVs (Table S2).Proteins enriched in pEVs include FLNA, ACTA1, and GSN while SLC25A6 was enriched in platelets (Figure 3C).
Rank-based GSEA analysis of DAPs revealed that proteins ubiquitously enriched in EVs were implicated in processes such as complement and coagulation cascade (Figure 3D,E and Table S3).pEVs were enriched for proteins such as antithrombin (AT3), the plasminogen zymogen (PLG), and factor II (prothrombin) implicated in the coagulation pathway, corroborating the procoagulant role of pEVs, as depicted in Leading-edge analysis (Figure 3E).We highlight proteins implicated in "complement and coagulation cascade" that are enriched in EVs (Figure 3F, Figure S2A).Thus, our data highlight biological processes conserved in EVs released by activated platelets.

Agonist-specific packaging of platelet EVs
Next, to interrogate agonist-specific packaging of pEVs, we identified DAPs (p < 0.05, fold change >1.5 or <−1.5) comparing the proteomes of different EVs released by platelets activated by different agonists (Figure 4, Table S4, Figure S2B).Pearson's correlation coefficient of protein intensities between replicates in each group was over 0.8, indicating reproducible measurements (Figure 4A).Importantly, pEV proteomes following thrombin and thrombin/collagen treatment clustered together, whereas pEVs released from platelets following ADP or collagen activation clustered separately.This is reflective of the agonist potency observed with platelet activation (Figure 1A,B) and suggests that the potent agonist, thrombin, is more influential in dictating a more specific pEV proteome.
Using K-means clustering, we identified protein clusters that were specifically enriched in EVs in response to different agonist stimulation  S5-S6).
Thus, our data show that different agonists can alter the pEV proteome, resulting in heterogeneous pEV populations with potentially distinct biological functions.

Activated platelets selectively sort (package) their cargo into EVs
Given that EVs were collected from platelets within 30 min of agonist stimulation, we hypothesized that platelets package their cargo into EVs, as opposed to de novo protein synthesis during EV release.Indeed, proteins whose abundance was lower in activated platelets versus resting were concomitantly enriched in EVs versus activated platelets (Figure 5).We report similar observations of cargo sorting in all agonist stimulations (Figure 5A-D).This suggests that platelets, upon stimulation, package their proteins into EVs for release.Given that EVs have distinct proteomes to platelets (Figure 3A), our data sug-gest an active cargo selection mechanism operating during EV release as opposed to random sampling.We highlight two proteins (CD82 and PPP2CA) whose abundance in activated platelets inversely correlated with their elevated abundance in released EVs (Figure 5E).Conversely, proteins that were in similar abundance between resting and activated platelets (Figure S3A) or upregulated in higher abundance in activated platelets versus resting platelets (Figure S3B) displayed low abundance in pEVs.

DISCUSSION
Platelet activation in a thrombus occurs through a diverse mix of physical and biochemical stimuli.Within a thrombus, platelet activation is heterogeneous, resulting in distinct zones with different activation states [2][3][4][5][6].The strongly activated platelets in the core of the thrombus are primarily stimulated through collagen and thrombin, while the less activated platelets around the thrombus shell are primarily regulated by weaker agonists such as ADP.It has been reported that pEVs produced from platelets stimulated with different agonists have distinct proteomes; however, the literature has focused on assessing the difference between pEVs generated by specific agonists [36,37,39], or through shear-induced platelet activation [38], but has not reported a comparison between platelets and the proteome of the pEVs that these platelets generate.Previous work from Millioli et al. [36] focused on providing an overview of how agonist potency may influence the pEV proteome.Specifically, this study compared the biological processes associated with ADP-, collagen-, and thrombin/collagen-generated pEVs to thrombin-generated pEV controls to show a positive correlation between platelet activation proteins and agonist potency, while there was a negative correlation between platelet degranulation proteins and agonist potency.This work shows clearly that the proteome of pEVs is directly related to the platelet agonist used for pEV generation, but does not show a comparison to resting or activated platelets.
In our current study we have extended on this work by including the proteome of the cells of origin, resting and activated platelets, allowing us to compare the proteomes of resting platelets, activated platelets, and pEVs from the same donor on the same day, and specifically to explore protein packaging into pEVs.Additionally, we included a holistic analysis of pEV proteomes to provide insights into the potential role of pEVs in thrombosis.
Here, we used resting platelets isolated from healthy human volunteers to produce pEVs in vitro, allowing us to produce pEV populations free from contamination by other blood cells or plasma proteins.A holistic overview of pEVs showed that pEV proteomes were distinct from the proteome of platelets, the cells of origin from which these EVs were generated.GSEA analysis showed that pEVs were enriched using R package Differential Enrichment analysis of Proteomics data (DEP) [51].Adjusted p values are reported in Table S2.  in proteins related to the coagulation cascade, as had been reported previously [13,37].While the major reported role for pEVs in the literature is centered on their procoagulant function, it is intriguing to note that 22 proteins involved in the complement and coagulation pathways were enriched in pEVs compared to resting and activated platelets, including enrichment in both positive (F2, F5), and negative regulators of coagulation (PROS1, AT3 and PLG), highlighting a potential dual role for pEVs in regulating coagulation (Figure S2).Additionally, this finding also reflects the dual role of platelets in coagulation and fibrinolysis, whereby strongly activated procoagulant platelets, such as those seen within the platelet core, express a pool of plasminogen for conversion to the fibrinolytic plasmin [57].Simultaneously, these platelets release plasminogen activator inhibitor (PAI-1) from their α-granules to negatively regulate fibrinolysis and promote thrombus stability [58].
Further investigation into these enriched proteins showed stronger enrichment in pEVs derived from collagen-or thrombin-stimulated platelets, reflecting the state of procoagulant platelets in the thrombus core.Furthermore, there was enrichment in the key complement proteins C5 and C3, which play a significant role in the interplay between thrombosis and complement activation [59].As the current literature has focused predominantly on the role of procoagulant pEVs, further research into a potential role in anticoagulation or fibrinolysis would be intriguing, and a better understanding of the process by which pEVs are formed and package proteins may assist in identifying a population of anticoagulant pEVs.
Analysis of proteins enriched in EVs compared to activated platelets revealed several biological processes which were heavily influenced by platelet agonist.KEGG pathway analysis revealed several proteins which were associated with coagulation and hemostasis that clustered in relation to the agonist used to stimulate the platelets.The three significant clusters were C1 (C1; Figure 4A-D) which included the fibrinogen subunits (FGA, FGB, and FGG essential for platelet aggregation) and EMILIN1 (associated with platelet aggregation and clot retraction) [60] found in relatively low abundance in pEVs produced from thrombin stimulation, including thrombin/collagen stimulation, when compared to ADP-and collagen-produced pEVs.As EVs are thought to reflect in some way the activation state of their cells of origin, one explanation for this may be that platelets stimulated with thrombin efficiently release their granular components (Figure 1B) including fibrinogen stored within their α-granules [61].This is supported by C4 proteins involved in platelet activation, exocytosis and membrane reorganization such as PLEK [62], and ARHGAP17 [63,64], which associated more strongly with thrombin stimulation.This is contrasted against C3 proteins with relatively low expression in ADPinduced pEVs such as prothrombin (F2), the precursor to thrombin stored in platelet α-granules [65], CD9 which forms a complex with GPIIb/IIIa [66] to regulate inside-out signaling [67], and GPV (GP5), part of the GPIb-IX-V complex involved in platelet adhesion to von Willebrand Factor (vWF).Intriguingly, KEGG analysis identified a separate cluster of proteins (C6) more abundant in pEVs derived from collagen-stimulated platelets (C6; Figure 4C,D) involved in platelet activation.Whereas fibrinogen and EMILIN1 are important for platelet aggregation and clot retraction, C6 platelet activation proteins LCP2 (also known as SLP76) [68,69], SNAP23 [70], and the negative regulator of platelet activation VASP [71] were more important in collagenand GPVI-induced platelet signaling pathways, further highlighting the influence that the individual platelet agonist has on the pEV proteome.
A key outstanding question pertains to the potential functional differences between these distinct pEV populations, and their relevance to human health and disease.As discussed above, although there was overlap in the proteomes of all pEV types studied, there was still clear clustering, which could be associated with platelet agonist.Given the relative importance of platelet agonists within the thrombus, these differences may be indicative of their potential function.For example, a study comparing pEVs isolated from healthy human donors to pEVs isolated from patients undergoing cardiac surgery found that pEVs from cardiac patients induced more rapid thrombin generation than those from healthy donors, as well as enhancing venous thrombosis in vivo [72].In contrast, a study of pEVs released after limb ischemia exerted a cardioprotective effect following heart ischemia/reperfusion in rats [73].The different environments in which these pEVs were generated in vivo seem likely to influence their function.From the current study this could be reflected in the upregulation of pro-and anti-coagulant proteins (F2, AT and PLG).
It is likely that the differences in proteome observed between activated platelets and pEVs is due to active protein packaging rather than de novo protein synthesis.This hypothesis is supported by the short stimulation period (30 min), and the finding that proteins enriched in pEVs were relatively downregulated in activated platelets compared to resting platelets (Figure 5).Various mechanisms of how cargo is packaged into different EVs have been suggested, although exact mechanisms remain elusive, particularly in microvesicles compared to exosomes.Furthermore, the mechanisms of cargo sorting are dependent on cell and EV type [74][75][76], including modes of biogenesis and cell context [30,77].Much of the focus in this area has been on the role of post-translational modifications (PTMs) and subcellular localization into the budding EVs, either due to cellular activation, or for regulating cellular homeostasis [78,79].This process may also be aided by plasma membrane anchor proteins localizing cytoplasmic proteins to the budding EV [80].Exosome biogenesis and cargo research has revealed an important role for endosomal sorting, which appears to be at least partially shared by microvesicles [81].Studies in tumor cell-derived EVs have revealed a key role for the GTP-binding protein ARF6 in regulating membrane budding and EV release, and suggested that some proteins typically trafficked through ARF6-regulated endosomal sorting are contained within tumor EVs [82].It is interesting to note, however, that other proteins such as transferrin also trafficked through ARF6 mediated pathways were absent from tumor EVs suggesting that some proteins are selectively packaged in, while others selectively sorted out of, EVs.In platelets, little is known of the specific mechanisms by which protein cargo is sorted into pEVs, though the data presented here, and by others [36,39] suggest that the pEV cargo likely reflects the activation state of the platelet, and can be modulated Understanding the mechanisms regulating the formation of the pEVs produced in this study could potentially be leveraged to produce pEVs that are cardioprotective, but not pro-thrombotic.In our study, we used KEGG pathway analysis to look for insight into what function our pEV populations may have.Aside from proteins, the other major cargo that pEVs carry are miRNAs [7,8], and, similar to our findings, miRNA released from platelets (most likely contained in pEVs) appears to be dependent on platelet agonist [84].A comparison of the protein content and the miRNA content could provide further insight into the roles of pEVs in health and disease and guide future experiments in looking for functional differences of pEV populations derived from differently activated platelets.A limitation of our study is the relatively small sample size (comprising of samples derived from platelets from 5 individual donors) examined.Although the paired nature of our data allows for higher statistical power, future experiments should also aim to validate these findings with a larger sample size.
pEVs are the most abundant vesicles in the blood and exhibit many of the functional characteristics of platelets.This study has explored the impact of different physiological stimuli on altering the proteomic landscape of pEVs and platelets to reveal new insights into the process of stimuli-induced encapsulation for pEVs.Given their emerging therapeutic potential in bioactive cargo delivery, including proteins and drugs [85], these findings highlight the potential for tuning the cargo of pEVs for therapeutic functions such as anticoagulation or fibrinolysis, or as potent effectors of tissue regeneration and cell function.
Indeed, pEVs have recently been shown to target inflammation areas as an effective drug-loaded delivery system [86], and act as immune cell extracellular traps to form thrombus scaffolds [87].Therefore, as the mode of platelet activation has a direct impact on the pEV proteome and function, the elucidation of the mechanisms by which specific proteins are packed into pEVs will be important in unlocking the development of therapeutic pEVs with distinct biological functions, and may be promising therapies for thrombotic and cardiovascular diseases.

F I G U R E 4 F I G U R E 5
Agonist-specific packaging of platelet EVs.(A) Heatmap of Pearson correlation matrix of protein abundance between proteomes.(B) Heatmap of differentially abundant proteins (p.adjusted < 0.05, log2 FC > 1.5) with k-means clustering identifying distinct protein clusters.Differential enrichment analysis, and the raw p-values adjusted to correct for multiple testing using Benjamini-Hochberg method, was performed using R package Differential Enrichment analysis of Proteomics data (DEP) [51], and reported in Table S4.(C) KEGG pathways and (D) Ontology terms (Biological processes) enriched in each cluster (p.adjusted < 0.05) are indicated.Activated platelets selectively encapsulate (package) their cargo into EVs.(A-D) Scatter plots of proteins showing their relative abundance in "EVs versus activated platelets" and "activated platelets versus resting platelets."(E) Bar plots depicting abundance (centered) for indicated proteins.
through the stimulant used to induce pEV formation.Here we show that protein is sorted from the resting platelet into pEVs following activation; however, we have not assessed the PTM status of these proteins.Future experiments investigating the PTM status of pEV proteins may give insight into the mechanisms by which proteins are sorted into pEVs following platelet activation.Harnessing the biogenesis of EVs for bioengineering and therapeutic cargo loading to modify composition and function of derived EVs is now a rapidly emerging aspect of the field of EV-based therapeutics[77,83].
Tandem mass spectra were searched against as variable modifications.Data were processed using trypsin/P as the proteolytic enzyme, with up to two missed cleavage sites allowed.The search tolerance and fragment ion mass tolerance were set to 7 ppm and 0.15 Da, respectively, at less than 1% false discovery rate on peptide spectrum match (PSM) level employing a target-decoy approach at peptide and protein levels.A label-free quantification (LFQ) algorithm in MaxQuant was used to obtain normalized quantification intensity values.Data were analyzed using R package Differential Enrichment analysis of Proteomics data (DEP)