Platelet procoagulant phenotype is modulated by a p38-MK2 axis that regulates RTN4/Nogo proximal to the endoplasmic reticulum: utility of pathway analysis

Abstract

Upon encountering physiological cues associated with damaged or inflamed endothelium, blood platelets set forth intracellular responses to ultimately support hemostatic plug formation and vascular repair. To gain insights into the molecular events underlying platelet function, we used a combination of interactome, pathway analysis, and other systems biology tools to analyze associations among proteins functionally modified by reversible phosphorylation upon platelet activation. While an interaction analysis mapped out a relative organization of intracellular mediators in platelet signaling, pathway analysis revealed directional signaling relations around protein kinase C (PKC) isoforms and mitogen-activated protein kinases (MAPKs) associated with platelet cytoskeletal dynamics, inflammatory responses, and hemostatic function. Pathway and causality analysis further suggested that platelets activate a specific p38-MK2 axis to phosphorylate RTN4 (reticulon-4, also known as Nogo), a Bcl-xl sequestration protein and critical regulator of endoplasmic reticulum (ER) physiology. In vitro, we find that platelets drive a p38-MK2-RTN4-Bcl-xl pathway associated with the regulation of the ER and platelet phosphatidylserine exposure. Together, our results support the use of pathway tools in the analysis of omics data sets as a means to help generate novel, mechanistic, and testable hypotheses for platelet studies while uncovering RTN4 as a putative regulator of platelet cell physiological responses.

INTRODUCTION

Blood platelets are the primary cellular mediators of hemostasis and contribute to inflammation and thrombosis in vascular disease (34, 46). For over the past 20 years, biochemical and molecular studies of platelets have established critical roles for a number of intracellular signaling systems driving cell biological outputs underlying platelet function (reviewed in Refs. 4, 5, 32, 45). Traditionally, mechanistic findings regarding the molecular bases of platelet physiology are built on hypothesis testing around insights afforded from the evolving literature and studies of other molecular regulatory systems generalized over a variety of other cell types. While such endeavors to define the molecular events of platelet regulation continue to be fruitful, they may be, in part, skewed by the bias of investigators or the surreptitious availability of pharmacological agents or genetically manipulated mice and typically neglect to account for how the majority of the >5,000 proteins that constitute the platelet proteome potentially contribute to platelet function (21, 33, 59, 76).

Concurrent with efforts to understand platelet physiology at the molecular level, a range of systems biology and omics-based projects provide an ever-growing variety of increasingly rich data sets associated with platelet physiological function (19, 21, 76). Mass spectrometry-driven proteomics experiments have been especially critical in cataloging the expression and modification of platelet proteins under resting, stimulated, and disease conditions (16, 19, 22). Surveying the results of platelet proteomics studies, it is immediately apparent that platelets express abundant cytoskeletal, metabolic, and signaling proteins as well as proteins specific to platelet function and hemostasis (19, 22). More rigorous computational analyses of the platelet proteome focused on mapping physical and functional protein-protein interactions (PPIs) have additionally established that entire (sub)networks of proteins critical to G protein-coupled receptor, intracellular calcium, cyclic nucleotide and phospholipid signaling are also present in platelets and likely critical to platelet function (19, 21, 33, 76). Specialized informatics platforms such as PlateletWeb (19) now allow investigators to access and take advantage of a variety of omics data sets aimed at addressing specific questions in the context of platelet physiology; however, other systems modeling tools that take into account how exact protein phosphorylation modifications and other events are causally linked to one another remain underutilized or unavailable in platelet studies.

To better understand the molecular physiology of platelets and the biochemical pathways and systems regulating platelet function, we interrogated a set of platelet proteins functionally modified by reversible tyrosine and serine/threonine phosphorylation using a complementary set of informatics and pathway analysis tools to highlight potentially unrecognized, causal events underlying platelet biology. Using high-detail pathway information from Pathway Commons together with pathway tools such as CausalPath and ChiBE (12, 25), we find that, in addition to noting established signaling steps in platelet function, pathway resources can help to uncover novel, testable, and potentially critical relations underlying the regulation of platelet form and function. In this study, to demonstrate the utility of pathway tools in platelet studies, we present modeling and in vitro experimental findings detailing the sequential p38-MK2 phosphorylation of the Bcl-xl sequestering protein RTN4 (also termed reticulon-4, RTN-4S, RTN-XS, Nogo, ASY; referred to as RTN4 here going forward) in regulating molecular events in platelets at the level of the endoplasmic reticulum (ER) associated with establishing and organizing platelet procoagulant activities.

MATERIALS AND METHODS

Reagents.

All reagents were from Sigma except as noted. Human fibrinogen was from Enzyme Research. Antibodies and other labeling agents were sourced as follows: p38 (sc-7972), Nogo/RTN4 (sc-271878), protein disulfide isomerase (PDI) (sc-20132), and mouse IgM (sc-3881) were from Santa Cruz Biotechnology; p-p38 (no. 4511), p-MK2 (no. 3007), MK2 (no. 3042), RXXpS*/T* (no. 9614), and Bcl-xl (no. 2764) were from Cell Signaling; p-Bcl-xl (44-428G) and anti-mouse IgM heavy chain horseradish peroxidase (62-6820) were from ThermoFisher; α-tubulin (T6199) and TRITC-phalloidin (P1951) were from Sigma; goat anti-mouse IgM Alexa Fluor 488 (A21042), goat anti-rabbit IgG Alexa Fluor 546 (A11010), and annexin V-Alexa Fluor 488 (A13201) were from Invitrogen (ThermoFisher Scientific). The P2Y1 antagonist MRS 2179, the P2Y12 antagonist AR-C 66096, the p38 inhibitor SB202190, and the MK2 inhibitor PF3644022 were from Tocris.

Interactome and neighborhood analysis.

To generate protein-protein interaction (PPI) networks, we used ChiBE to query Pathway Commons. We used the “paths-between” query on the Simple Interaction Format (SIF) network, selecting “interacts-with” as the relation type, with length limit 1. For the neighborhood analysis of MK2 (MAPKAP2) and RTN4, we queried the Pathway Commons database using ChiBE in SIF mode selecting relation types “controls-state-change-of” and “in-complex-with” using length limit 1.

Pathway analysis.

We used CausalPath (14) to identify the pathway fragments that can directly explain correlated changes in the phosphoproteomic data set from Beck et al. (16). CausalPath first processes the Pathway Commons database (25) using the BioPAX-pattern library (11), searching for graphical patterns that capture potential binary cause-effect relations between protein activity, abundance, and modifications (14). CausalPath then selects the subset of the causal relations that fit the given data. In addition to Pathway Commons data, we performed a manual literature curation for some of the phosphorylation site effects that were not covered by Pathway Commons. We used CausalPath in two modes: 1) default mode; and 2) by relaxing site-matching constraints. The site-matching constraint requires the phosphorylation site position to be mentioned in the pathway model and to match with the site position identified and measured in the phosphoproteomic data set of interest. CausalPath is available at https://github.com/PathwayAndDataAnalysis/causalpath. CausalPath result networks are rendered with the ChiBE visualization tool (12, 13), which is available at https://github.com/PathwayCommons/chibe.

Platelet preparation.

Washed human platelets were prepared from venous blood drawn from a rotating pool of >20 healthy, adult (>18 yr old) male and female volunteers by venipuncture into 1:10 sodium citrate (3.8%), as previously described (6, 10). Written informed consent was obtained from the subjects and the protocol was approved by Oregon Health & Science University Institutional Review Board. Blood was centrifuged at 200 g for 20 min to obtain platelet-rich plasma (PRP), and platelets were isolated from PRP by centrifugation at 1,000 g for 10 min in the presence of prostacyclin (0.1 μg/ml). Platelets were resuspended in modified HEPES/Tyrode buffer and washed once via centrifugation before resuspension in modified HEPES/Tyrode buffer. Following platelet preparation, platelet samples from different donors were not mixed or pooled for any of the experiments reported in this study.

Immunoprecipitation.

Washed human platelets (5 × 10⁸/ml) were pretreated with inhibitors for 10 min before stimulation with agonists as indicated. Following stimulation, platelets were lysed with the addition of 0.1% Triton X-100 together with protease and Sigma phosphatase inhibitor cocktail 2 and 3 and sonication. Lysates were precleared with protein L agarose (4°C, 1 h) before the addition of RTN4 antibodies or mouse IgM (2 μg, 4°C, overnight). Captured proteins were precipitated with protein L agarose (Santa Cruz Biotechnology) (4°C, 2 h). Agarose beads were washed three times with HEPES/Tyrode (with 0.1% Triton X-100) before elution in Laemmli sample buffer and Western blot analysis for captured proteins. For Bcl-xl coimmunoprecipitation, platelets were incubated with 2 mM DTBP cross-linker as previously described (71) (30 min, room temperature) before lysis by sonication into 0.1% Triton X-100, immunoprecipitation, and Western blot analysis as previously described (57). Band densitometry measurements were performed in ImageJ, and data were analyzed by one-way ANOVA with post hoc Tukey’s comparison test. P < 0.05 was considered statistically significant for all tests. Statistical analyses were performed using GraphPad PRISM (San Diego, CA) as previously described (57).

Platelet adhesion assays.

Platelet spreading, differential interference contrast (DIC), and super resolution structured illumination microscopy (SR-SIM) microscopy experiments were carried out as previously described (6, 52). Briefly, for platelet spreading experiments, 12-mm no. 1.5 glass coverslips (Fisher Scientific) or coverglass-bottom dishes (MatTek) were coated with human fibrinogen (50 μg/ml) followed by surface blocking with filtered fatty acid-free BSA (5 mg/ml). Vehicle (0.1% DMSO) or pharmacological inhibitors were added to platelets in solution (2 × 10⁷/ml) for 10 min before seeding onto immobilized surfaces at 37°C for 45 min followed by washing with PBS. Adherent platelets were fixed with 4% paraformaldehyde (PFA) at room temperature for 10 min before mounting on glass slides with Fluoromount G (Southern Biotech). Platelets were imaged using Kohler-illuminated Nomarski DIC optics with a Zeiss ×63 oil immersion 1.40 numerical aperture (NA) plan-apochromat lens on a Zeiss Axio Imager M2 microscope using Slidebook 5.5 image acquisition software (Intelligent Imaging Innovations, Denver, CO) as previously described (6).

Fluorescence microscopy.

Following platelet spreading and PFA fixation as above, adherent platelets were permeabilized with a blocking solution (1% BSA and 0.1% SDS in PBS), as previously described (52). Platelets were then stained with indicated primary antibodies overnight at 4°C at a 1:100 dilution in blocking buffer. Alexa Fluor secondary antibodies (1:500) or TRITC-phalloidin (1:500) were added in blocking buffer (2 h). Coverslips were mounted with Fluoromount G on glass slides. Platelets were imaged using a Zeiss Axio Imager M2 microscope. Adherent platelets were also imaged using super resolution structured illumination microscopy (SR-SIM) with a Zeiss ×100 oil immersion 1.46-NA alpha plan-apochromat lens on a Zeiss Elyra PS.1 microscope as previously described (7).

Annexin V staining of adherent platelets.

Following preparation as described above, washed human platelets were pretreated with inhibitors as described and then incubated on fibrinogen-coated coverglass dishes in the presence or absence of bovine thrombin (0.25 U/ml) for 45 min at 37°C. Next, 5 μl annexin V-Alexa Fluor 488 and 2.5 mM CaCl₂ were added to the 100 μl platelets for an additional 30 min. Following washing in HEPES/Tyrode buffer supplemented with 2.5 mM CaCl₂, platelets were immediately imaged by fluorescence microscopy and DIC optics at ×63 magnification as described above.

Platelet secretion assay.

Platelet secretion was measured using a plate reader-based assay of ATP released from platelet dense granules, measured as the light output generated by an ATP-luciferin-luciferase reaction as previously described (49). Briefly, human washed platelets (2 × 10⁸/ml) were incubated with inhibitors as indicated or vehicle alone (15 min, 37°C). Platelets were then incubated with orbital shaking in a white flat bottom 96-well plate (Corning Costar, Tewksbury, MA) in the presence of platelet agonist for 30 s at 37°C. Detection reagent Chronolume (Chrono-Log) was added to the wells and sample luminescence was detected using an Infinite M200 spectrophotometer (TECAN, Mannerdorf, Switzerland). A grouped analysis was performed using two-way ANOVA with post hoc Tukey’s comparison test in GraphPad PRISM.

RESULTS

Interactome analysis of the platelet phosphoproteome.

Over the past several years, proteomics studies have detailed the expression and functional modification of platelet proteins in a number of contexts (19, 22, 39). Recently, a systematic analysis of differentially abundant phosphoproteins in stimulated platelets revealed regulated nodes and functional subnetworks around platelet cytoskeletal dynamics, degranulation, and aggregation as well as more novel nodes suggesting roles for ubiquitin modifications and small GTPase regulatory systems in platelet regulation (16). To determine whether a protein-protein interaction (PPI) analysis of proteins that undergo significant changes in phosphorylation can likewise provide novel insights into specific mechanistic steps underlying platelet activation, we first carried out a ChiBE-based query of databases associated with Pathway Commons—the largest detailed human pathway database indexing more than 4,000 pathways and 1.3 million genetic and protein associations and biochemical modifications, including site-specific protein phosphorylation events from a number of databases such as PhosphoSitePlus (36). As seen in Fig. 1, a PPI network analysis of dynamically phosphorylated platelet proteins connects a number of established effectors of platelet function, including myosin light chain phosphatase (indicated by gene name PPP1R12A), filamin (FLNA), and protein kinase C (PKC) isoforms (PRKCD, PRKCQ). Other more novel signaling systems of emerging interest to platelet physiological studies were also apparent, including higher density nodes around the ArfGAP AGFG1, the Rac/Cdc42 GEF Cool-1/-PIX (ARHGEF7), and the mitogen-activated protein (MAP) kinase p38 (MAPK14). However, while this PPI analysis begins to organize some novel nodes and terminal effectors into the logic of platelet regulation, it does not immediately offer directional, mechanistic details for hypothesis-driven studies of platelet function.

Fig. 1.

Interactome analysis of the regulated platelet phosphoproteome. A protein-protein interaction (PPI) network of dynamically regulated platelet phosphoproteins identified by Beck et al. (16) was generated from an interaction query of Pathway Commons using ChiBE. Node colors indicate the relative intensity of the reported phosphorylation change (red = increase; blue = decrease). Selected proteins of interest to this study, including the MAP kinase p38 (MAPK14), the MAP kinase-activated protein kinase MK2 (MAPKAPK2), and RTN4 are indicated with stars.

Pathway and causality analysis of platelet protein phosphorylation.

In addition to building interaction networks, more specific pathway analysis tools can offer further mechanistic insights into signaling networks from omics data sets (25, 30, 76). We next analyzed the set of platelet proteins modified by phosphorylation described above with CausalPath (14) to identify potential cause-effect relations using an integration of publicly available pathway data provided by Pathway Commons (25, 30). As seen in Fig. 2A, a pathway model built from causal enzyme-substrate relationships recapitulated details of pathways in platelet activation downstream of platelet receptors, which feed into PKC, MAPK, and other prominent signaling systems and effectors. Notably, some nodes of this resulting model such as p38 MAPK (MAPK14) and PKC (PRKCD, PRKCQ) are centrally integrated into intracellular signaling networks associated with a diversity of platelet responses. For instance, p38 MAPK is phosphorylated and activated downstream of ASK1 (MAP3K5) activation as well as activities associated with PLCγ3, Syk, and the tyrosine kinase adaptor/scaffold protein Nck2. Following phosphorylation and activation, p38 is predicted to regulate a diverse set of substrates associated with MAPK signaling, intracellular physiology, and platelet hemostatic and procoagulant responses, including the MAP kinase-activated protein kinase MK2 (MAPKAPK2), GSK3β, and ADAM17. In addition to MAPKs and PKC isoforms, platelet activation upregulates the phosphorylation of Syk and Syk effectors associated with ITAM signaling and the LAT and PI3K signalosome (BTK, Vav3, PLCB3, DAPP1). More terminal targets downstream of several signaling systems with roles in platelet activation are also apparent, including cytoskeletal regulators and elements such as myosin light chain kinase (MYLK), myosin phosphatase (PPP1R12A), cortactin (CTTN), moesin (MSN), zyxin (ZYX), and α-adducin (ADD1) as well as cytoskeletal-associated proteins with uncharacterized roles in platelets, including the centrosomal protein CEP131, Merlin (NF2), WAS/WASL-interacting protein WIPF1, and the actin organizing protein girdin (CCDC88A). Dynamic, feedback-like modifications on receptors and proteins associated with platelet activation as well as immunologic and metabolic processes are also apparent, including phosphorylation of thromboxane A2 receptor (TBXA2R), the disintegrin ADAM17, chemokine receptor (CXCR4), glucocorticoid receptor (NR3C1), insulin receptor substrate (IRS1), IL-6 receptor (IL6ST), and the caspase recruitment adaptor protein CARD9, which connects ITAM and Toll-like receptor (TLR) signaling in myeloid cells (35). Interestingly, mitochondria-associated proteins, including dynamin-1-like protein (DNM1L) and the neurite outgrowth inhibitor protein RTN4, were also integrated into select pathways. Other proteins with more traditional roles in the nucleus were modified within the pathway context of platelet activation, including NFATC1, NFATC2, and DDX5. Together, this analysis suggested a concerted association between a number of platelet cytoskeletal, secretory, and inflammatory outputs associated with PKC, MAPK, and other signaling systems, offering a means and rationale to build and test novel hypotheses potentially relevant to platelet physiology not immediately apparent by other analyses.

Fig. 2.

Pathway and causality analysis of the activated platelet phosphoproteome. Results from CausalPath identifying pathway fragments and associated changes in platelet protein phosphorylation to show potential cause-effect relations as visualized with ChiBE. The unrestricted model in A is based on generalized enzyme:substrate relations (i.e., phosphorylated/activated p38 is known to be involved in the phosphorylation and activation of MK2), independent of phosphorylation site localization details. The stringent model in B requires the matching of enzymatic events with phosphorylation site details (i.e., phosphorylation of p38 at Thr₁₈₀,Tyr₁₈₂ is known to be involved in MK2 Thr₃₃₄ phosphorylation and activation). This model highlights a putative p38→MK2→RTN4 axis associated with platelet activation (light blue background). As indicated in the Legend, nodes represent proteins (conventionally labeled with gene names), and edges represent either causal phosphorylation (green) or dephosphorylation (red) events. Protein phosphorylation sites are shown with smaller “p” circles, where a green border indicates an activating site and red border indicates inactivating site. The background color of phosphorylation sites indicates their differential measurement from data, red indicating an increase and blue indicating a decrease. The grouped nodes in a compound node indicate that all members have the same graph topology and are grouped for complexity management. Relative targets of the p38 inhibitor (SB202190) and MK2 inhibitor (PF3644022) used in this study are indicated in the context of this model.

Next, to extract more specific mechanistic information, we included a phosphorylation site matching requirement between detected phosphorylation sites and the sites in the pathway model, to generate a higher-confidence subgraph of the pathway network in Fig. 2A. As seen in Fig. 2B, in addition to highlighting cascades terminating in NFATC1, NFATC2, cortactin, and CXCR4 phosphorylation, this analysis predicted a specific MAPK signaling axis (ASK1 Ser₁₀₃₃→p38 Thr₁₈₀,Tyr₁₈₂→MK2 Thr₃₃₄) resulting in the phosphorylation of RTN4 Ser₁₀₇. Together, these models demonstrate that pathway analysis of phosphoproteomics data can predict causal, mechanistic information around regulators and effectors of platelet activation.

p38 and MK2 activation and function in platelets.

The biology associated with p38 MAPK signaling in platelets is complex, as a variety of agonists and physiological stimuli are well known to upregulate the phosphorylation of p38 and p38 substrates in platelets, yet specific functions for platelet p38 remain unclear (2). While p38 Thr₁₈₀,Tyr₁₈₂ phosphorylation has been extensively documented in platelets under a number of experimental conditions, limited details are known regarding the phosphoregulation of the p38 substrate MK2 in platelets. As a first step in determining whether the causal pathway model above reflects and predicts signaling relations in platelets, we next examined platelet p38-MK2 activation and signaling in vitro using biochemical and cell biological methods. Washed human platelets (5 × 10⁸/ml) were pretreated with pharmacological inhibitors of p38-associated signaling pathways or vehicle alone (0.1% DMSO) and stimulated with thrombin (0.5 U/ml, 5 min) before lysis in Laemmli sample buffer, separation by SDS-PAGE, and Western blot analysis with p38 Thr₁₈₀,Tyr₁₈₂ and MK2 Thr₃₃₄ phosphospecific antisera. As seen in Fig. 3, A and B, thrombin stimulation significantly upregulated p38 as well as MK2 phosphorylation (P = 0.0036 and P = 0.0001, respectively), as determined by Western blot analysis of platelet lysates with p38 and MK2 phosphorylation site-specific antisera. Pretreatment of platelets with the p38 inhibitor SB202190 [2 μM, previously reported to inhibit p38 activity in platelets (42)] significantly abolished the p38-driven phosphorylation of MK2 Thr₃₃₄ in response to thrombin stimulation with minimal effects on the upstream phosphorylation of p38 itself at Thr_180,Tyr₁₈₂. Likewise, pretreatment of platelets with the MK2 inhibitor PF3644022 significantly limited the phosphorylation of MK2. As seen in Fig. 3, A and B, pretreatment with of platelets with PF3644022 also appeared to have some minor although not statistically significant inhibitory effects on p38 phosphorylation in response to thrombin, consistent with previous reports in other cell types suggesting a feedback interaction between MK2 and p38 activation (44, 73).

Fig. 3.

p38 and MK2 phosphorylation and function in platelets. A: replicate samples (n = 3) of washed human platelets (5 × 10⁸/ml) were pretreated with the p38 inhibitor SB202190 (2 μM), the MK2 inhibitor PF3644022 (2 μM), a combination of the P2Y1 and P2Y12 antagonists MRS 2179 (10 μM) and AR-C 66096 (10 μM), or vehicle alone (0.1% DMSO) before stimulation with bovine thrombin (0.5 U/ml, 5 min), lysis into Laemmli sample buffer, SDS-PAGE, transfer to nitrocellulose, and Western blot analysis of phospho-p38 Thr₁₈₀,Tyr₁₈₂ (p-p38) and phospho-MK2 Thr₃₃₄ (p-MK2) immunoreactivity. Total α-tubulin levels serve as a control for equal protein loading. Tick marks indicate the relative positions of 40-kDa molecular mass marker (for p-p38) 50-kDa molecular mass marker (for p-MK2 and α-tubulin). B: densitometry analysis of p38 (black bars) and MK2 (gray bars) phosphorylation (arbitrary units; AU). *P ≤ 0.05 relative to thrombin stimulated platelets. C: washed human platelets (2 × 10⁸/ml) were incubated with vehicle (0.01% DMSO), SB 202190 (10 µM), PF 3644022 (10 µM), or Ro 31-8220 (5 µM) before activation with thrombin (1 U/ml) for 30 s. Platelet dense granule secretion was measured as function of ATP release generated by an ATP-luciferin-luciferase reaction. Values are mean ± SE of raw luminescence; (n = 4). *P ≤ 0.05, platelets compared with vehicle in the presence of thrombin (AU). D: washed human platelets (2 × 10⁷/ml) were treated with SB202190 (2 µM), PF3644022 (2 µM), the lysine acetyltransferase inhibitor C646 [10 µM, a previously described inhibitor of platelet spreading on fibrinogen (8)], or vehicle alone (0.1% DMSO) before incubation on fibrinogen-coated coverglass (45 min, 37°C), fixation, staining and imaging by DIC, and fluorescence microscopy (n = 3). Scale bar = 10 µm.

In vivo, in addition to serving as a chemotactic agent that activates platelets at select sites of injury or inflammation in the vasculature, ADP is secreted by platelets and binds to purinergic P2Y1 and P2Y12 G protein-coupled receptors (37). Consequently, ADP amplifies platelet responses in response to physiological agonists such as thrombin downstream of protease-activated receptors (PARs) to support a number of cell biological outputs associated with p38 activation, including “inside-out” activation of integrins, thromboxane generation, procoagulant phosphatidylserine (PS) exposure, thrombin generation, and other platelet functional outputs (37). Accordingly, we also examined the phosphorylation states of p38 and MK2 in platelets under P2Y1/P2Y12-inhibited conditions. As seen in Fig. 3, A and B, combined pretreatment of platelets with the P2Y1/P2Y12 antagonists MRS 2179 (10 μM) and AR-C 66096 (10 μM) significantly inhibited the phosphorylation of p38 as well as MK2 in response to thrombin. Together, these results show that following stimulation with thrombin, platelets phosphorylate MK2 Thr₃₃₄ in a manner dependent on p38 activity as well as purinergic signaling.

Previous studies of platelet MAPKs have hypothesized roles for p38 activity in platelet secretion downstream of phospholipase A (cPLA2) phosphorylation, as p38 inhibition or ASK1 genetic deletion inhibits secretion in response to various platelet agonists (40, 42, 51). To investigate roles for p38 and MK2 in platelet secretion, we pretreated washed human platelets with SB202190, PF3644022, vehicle alone (0.1% DMSO), or the PKC inhibitor Ro 31-8220 before stimulation and measured ATP released from platelet dense granules with an assay of ATP-luciferin-luciferase reactivity. As seen in Fig. 3C, PKC and p38 inhibition both significantly impaired platelet secretion in response to thrombin stimulation; however, MK2 inhibition had no significant effect on secretion in response to thrombin (Fig. 3C). Roles for p38 activity in platelet cytoskeletal dynamics remain more controversial, as some studies have reported that p38 inhibition interferes with actin regulation in platelets while others conclude that p38 has no major role in the regulation of the platelet cytoskeleton (17, 47, 61). Hence, we next examined the ability of platelets to adhere to and spread on fibrinogen and form actin-rich lamellipodia under basal as well as p38- and MK2-inhibited conditions to assay platelet cytoskeletal function. As seen in Fig. 3D, pretreatment of platelets with the lysine acetyltransferase inhibitor C646, a previously described inhibitor of platelet actin cytoskeletal dynamics and lamellipodia formation (8), readily prevented the spreading of platelets on a surface of fibrinogen. Conversely, inhibition of either p38 or MK2 had no effect on the ability of platelets to adhere to and spread on immobilized fibrinogen and establish phalloidin-positive actin structures, suggesting that p38 as well as MK2 are not critical to events driving platelet cytoskeleton regulation downstream of the fibrinogen receptor integrin α_IIbβ₃. Together, these results show that platelets phosphorylate MK2 in a p38-dependent manner and that p38 activity has roles in platelet secretion independent of MK2 activation, likely through MK2-diverging substrates such as cPLA2 that are associated with eicosanoid metabolism.

RTN4 expression, localization, and phosphorylation in platelets.

Following from functional studies of RTN4 phosphorylation by MK2 in cell line models (58) as well as >350 separate records of RTN4 Ser₁₀₇ phosphorylation from proteomics studies curated in PhosphositePlus (36), a causality and pathway analysis of differentially phosphorylated platelet proteins (16) through Pathway Commons revealed RTN4 as a specific, putative, and differentially active target of p38→MK2 signaling in platelets (Fig. 2). While proteomics studies have detected RTN4 protein in human platelets (16, 19, 22), roles for RTN4 in platelet biology remain unknown. In mammalian cells, a number of RTN4 splice variants are expressed in a cell-specific manner (54) to effect specific cell biological functions, including the regulation of ER morphology, cell migration, and cell fate (55, 72). Accordingly, we next examined RTN4 expression in human platelet lysates and immunoprecipitates by Western blotting relative to nucleated MDA-MB-231 human breast adenocarcinoma epithelial cells and human umbilical vein endothelial cells (HUVECs). As seen in Fig. 4A, MDA-MB-231 cells, HUVECs, and washed human platelets all express abundant levels of the predominant 55-kDa Nogo-B/RTN4B isoform of RTN4. Following lysis and preclearing with protein L agarose, RTN4-immunoreactive proteins matching the molecular weight of expressed RTN4 isoforms were readily immunoprecipitated from platelets with RTN4 antisera but not with nonspecific mouse IgM (Fig. 4A).

Fig. 4.

RTN4 expression, localization and phosphorylation in platelets. A: MDA-MB-231 cell lysates, HUVEC lysates, human platelet lysates, and RTN4 and nonspecific IgM immunoprecipitates (IP) from platelets were separated by SDS-PAGE, transferred to nitrocellulose and examined for RTN4 immunoreactivity by Western blotting (WB). Positions of molecular weight (MW) markers are indicated. Representative results are shown (n = 3). B: replicate samples (n = 3) of washed human platelets were incubated on fibrinogen-coated coverglass in the absence or presence of 1 U/ml thrombin before fixation, staining for RTN4 (green) and PDI (red), and imaging by SR-SIM at wide field (scale bar = 10 µm) and ×100 (scale bar = 2 µm) magnification. C: replicate samples (n = 3) of washed human platelets were pretreated with SB202190 (SB; 2 µM), PF3644022 (PF; 2 µM), or vehicle alone (veh; 0.1% DMSO) before stimulation with thrombin (0.5 U/ml, 5 min) and lysis. RTN4 and nonspecific IgM immunoprecipitates (IP) were examined for total RTN4 and RTN4 phospho Ser₁₀₇ (RXXpS) immunoreactivity by Western blotting (WB). Tick marks indicate position of 50-kDa molecular mass marker.

To better understand potential roles for RTN4 in platelet function, we next examined the intracellular localization of RTN4 in adherent platelets by super resolution-structured illumination microscopy (SR-SIM). As seen in Fig. 4B, in platelets spread on a surface of fibrinogen, RTN4 displayed a specific, reticular staining pattern similar to that observed in other cell types that partially colocalized with the platelet endoplasmic reticulum (ER) and dense tubular system (DTS) marker protein disulfide isomerase (PDI) (Fig. 4B). The addition of thrombin to adhering platelets, which activates phospholipases and mobilizes calcium from the ER downstream of G protein-coupled receptors, condensed the ER-like RTN4 staining pattern together with PDI, suggesting that the RTN4 protein is regulated at the platelet ER in a functional manner in the context of platelet activation (Fig. 4B).

To determine if an upregulation of RTN4 Ser₁₀₇ phosphorylation following platelet activation is readily detected by more common biochemical methods, we assayed RTN4 immunoprecipitates from resting and thrombin-stimulated platelets for RXX-phosphoserine motif immunoreactivity by Western blotting. Notably, RTN4 Ser₁₀₇ is found within an acidophilic kinase RXXS substrate motif (. . . A₁₀₁PERQPS₁₀₇WDP₁₁₀ . . .) previously described using similar antisera in a number of phosphoproteomics studies from other cell and tissue types (36). As seen in Fig. 4C, RTN4 was readily immunoprecipitated from resting and thrombin-stimulated platelets while nonspecific mouse IgM alone did not precipitate detectable RTN4 beyond background levels. Western blot of RTN4 immunoprecipitates with RXXpS motif antisera showed positive, overlapping immunoreactivity for thrombin-stimulated platelets that was not detectable on RTN4 from resting platelets (Fig. 4C). To examine whether RTN4 Ser₁₀₇ is phosphorylated in a p38- and MK2-dependent manner in accordance with the site-matched pathway model from causality analysis above, we also pretreated platelets with the p38 and MK2 inhibitors SB202190 (2 μM) or PF3644022 (2 μM), respectively, before stimulation with thrombin, lysis, RTN4 immunoprecipitation and Western blot analysis. As seen in Fig. 4C, pretreatment with SB202190 or PF3644022 completely abrogated RXXpS immunoreactivity associated with RTN4 immunoprecipitated from thrombin-stimulated platelets. Together, these results show that platelets express RTN4, which colocalizes to the platelet ER/DTS upon platelet activation, and that RTN4 is phosphorylated in platelet at an RXXS motif corresponding to RTN4 Ser₁₀₇ in a p38- and MK2-dependent manner in response to thrombin stimulation.

Bcl-xl associates with RTN4, p38 and MK2 in platelets.

As specific roles for RTN4 as well as MK2 in platelets have not yet been examined, we next queried Pathway Commons for proteins expressed in platelets in the physical and functional vicinities of MK2 and RTN4 to visualize a RTN4 + MK2 neighborhood with ChiBE. As seen in Fig. 5A, this neighborhood model places MK2 between p38 and RTN4 and notes that MK2 may change the functional state of a number of downstream targets, including RTN4 (58) as well as Bcl-xl (BCL2L1) (75), a prosurvival Bcl-2 family member well known in the regulation of mitochondria permeabilization, ER mitochondria communication, and apoptosis-related pathways in platelets regulating procoagulant phosphatidylserine exposure (63). This model also noted that Bcl-xl has previously been described as an RTN4-interacting protein from studies of apoptosis in cell culture (71). Accordingly, we next examined the interaction of RTN4 and Bcl-xl in platelets by coimmunoprecipitation. As seen in Fig. 5B, immunoprecipitation of RTN4 from resting platelet lysates readily captured RTN4 as well as coprecipitating Bcl-xl. Although not identified in platelet proteomics analyses (16), pathway and neighborhood analyses also noted an interaction between MK2 and Bcl-xl (Fig. 5A), as MK2 has been reported to phosphorylate Bcl-xl Ser₆₂ in vitro (75). We next examined the phosphorylation of Bcl-xl Ser₆₂ in platelets with phosphospecific antisera, finding that like RTN4, Bcl-xl is also phosphorylated in thrombin-stimulated platelets in a manner significantly inhibited by the inclusion of SB202190 or PF3644022 (P = 0.0025 and 0.0038, respectively) (Fig. 5C).

Fig. 5.

Proximity of Bcl-xl to p38, MK2, and RTN4 in platelet function. A: signaling and protein binding relations in the neighborhood vicinity of MK2 and RTN4 in Pathway Commons, as queried and rendered by ChiBE. Directed relations (arrows) indicate that the source protein controls a state change of the target protein. A state change can be a modification on the protein, on its location, or a change to the complex that the target protein is a member. Undirected edges (connecting lines) indicate that two proteins appear as members in the same complex. B: replicate samples (n = 3) of washed human platelets (1 × 10⁹/ml, 1 ml) were incubated with DTBP cross-linker for 30 min at room temperature before lysis and processing for immunoprecipitation (IP). Following IP with RTN4 or control IgM antisera, samples were analyzed for RTN4 capture and Bcl-xl coimmunoprecipitation by Western blotting (WB). Tick marks indicate relative positions of 30- and 50-kDa molecular mass markers for Bcl-xl and RTN4, respectively. C: replicate samples (n = 3) of washed human platelets (5 × 10⁸/ml) were pretreated with SB202190 (SB; 2 µM), PF3644022 (PF; 2 µM), or vehicle alone (veh.; 0.1% DMSO) before stimulation with thrombin (0.5 U/ml, 5 min). Following lysis into Laemmli sample buffer, samples were examined for Bcl-xl phospho-Ser₆₂ (p-Bcl-xl) immunoreactivity by Western blotting. Total Bcl-xl levels serve as a control for equal protein loading. Tick marks indicate relative positions of 30-kDa molecular mass marker.

p38-MK2 regulates the intracellular organization of RTN4 in platelets.

Next, to examine the intracellular association or colocalization of RTN4 together with Bcl-xl in platelets, platelets were treated with the p38 and MK2 inhibitors SB202190 or PF3644022, respectively, or vehicle alone (0.1% DMSO) before adhering to fibrinogen in the absence or presence of thrombin and visualization by immunofluorescence SR-SIM (Fig. 6A). As seen in Fig. 6A, Bcl-xl showed a diffuse localization throughout platelets, partially colocalizing with RTN4-stained reticular structures. Similar to the results above (Fig. 4B), thrombin stimulation promoted the condensation of RTN4 together with Bcl-xl in the region of the platelet granulomere (Fig. 6A). Strikingly, p38 or MK2 inhibition had profound effects on RTN4 localization and organization under control conditions, as the ER-like RTN4-positive elements in platelets adherent to fibrinogen were severely diffused under p38- as well as MK2-inhibited conditions relative to vehicle alone. Similarly, RTN4 as well as Bcl-xl collapse was dramatically less in response to thrombin in platelets pretreated with SB202190 or PF3644022 (Fig. 6A).

Fig. 6.

p38-MK2-RTN4-Bcl-xl axis in procoagulant platelet function. A: washed human platelets were pretreated with SB202190 (2 µM), PF3644022 (2 µM), or vehicle alone (0.1% DMSO) before incubation on fibrinogen-coated coverglass in the absence or presence of thrombin (0.5 U/ml). Following fixation in PFA, samples were processed for SR-SIM visualization of RTN4 (green) and Bcl-xl (red) at wide field (×63, scale bar = 10 µm) and ×100 magnification (scale bar = 2 µm). Results are representative of n = 3 experiments. B: washed human platelets (2 × 10⁷/ml) were pretreated with SB202190 (2 µM), PF3644022 (2 µM), the intracellular calcium chelator BAPTA (40 µM), or vehicle alone (0.1% DMSO) before incubation on fibrinogen-coated cover glass in the absence or presence of 0.25 U/ml thrombin (37°C, 45 min). Following an additional 30 min in the presence of annexin V-Alexa Fluor 488 and 2.5 mM calcium, platelets were visualized for PS exposure and general morphology by DIC and fluorescence microscopy. Scale bar = 10 µm. Results are representative of n = 3 experiments.

The platelet ER/DTS serves as a reservoir for intracellular calcium mobilization and has critical roles in the initial steps of platelet activation as well as later phases where Bcl-xl-regulated apoptotic pathways promote platelets to take on a procoagulant phenotype through PS exposure (53, 64, 74). Previous studies have demonstrated that pharmacological inhibition of p38 activity or knockout of the p38-activating MAP3K ASK1 has no significant effect on intracellular calcium mobilization (17, 51). Despite the striking alterations in RTN4-positive ER morphology under p38- and MK2-inhibited conditions, we similarly detected no alterations in intracellular calcium mobilization in Alexa Fluor 488-Fura-2-loaded platelets in response to thrombin under p38- or MK2-inhibited conditions (data not shown).

We recently found that calcium-dependent signaling events, specifically the activation of PKC isoforms and PKC substrate phosphorylation, can localize proximal to the platelet ER/DTS in a manner that may have roles in regulating the intracellular spatiality of platelet signaling and activation processes (52). While previous studies have suggested that p38 has no role in regulating phosphatidylserine exposure in response to strong platelet agonists (i.e., dual PAR + glycoprotein VI stimulation), p38 inhibitors have been noted to prevent procoagulant blebbing of platelets adherent to fibrinogen as well as PS exposure in other contexts (26, 66). Moreover, the morphology of procoagulant platelets under MK2-inhibited conditions has not yet been described. We next imaged thrombin-stimulated platelets adherent to fibrinogen for PS exposure through annexin V staining and microscopy. As seen in Fig. 6B, treatment of adherent platelets with 0.25 U/ml thrombin for 45 min, followed by the addition of annexin V and 2.5 mM CaCl₂, promoted a characteristic rounded or “ballooning”-like morphology (3) and upregulated PS exposure as determined by annexin V stain. Pretreatment of platelets with either SB202190 or PF3644022 limited the number of platelets exhibiting a rounded, PS-positive morphology associated with a procoagulant phenotype, as p38 or MK2 inhibition promoted an increase in PS-positive spread platelets, suggesting either a disruption of intracellular processes that promote procoagulant activity or kinetic delays in platelet proapoptotic signaling. Only background levels of annexin V staining were observed on platelets adherent to fibrinogen in the absence of thrombin stimulation or in the presence of the intracellular calcium chelator BAPTA (Fig. 6B). Together with the causal pathway model and other data detailed above, our results suggest that p38 and MK2 target RTN4 as well as Bcl-xl proximal to the platelet ER/DTS to help orchestrate platelet cell biological responses associated with platelet activation.

DISCUSSION

Over the past decade, proteomics and informatics studies have begun to build a wealth of data associated with platelet biology in experimental and physiological contexts. In general, analyses of omics data associated with platelet physiology rely on PPI-driven tools such as STRING or PlateletWeb that cluster physical and functional protein associations to reveal network-level details underlying platelet function (19, 70). While such analyses are valuable in demonstrating the presence and organization of regulatory (sub)networks in platelets, mechanistic insights are more limited. Here, we use a cause-effect-oriented pathway analysis approach to interrogate a set of platelet phosphoproteins regulated in response to platelet stimulation, highlighting potentially novel signaling pathways in platelet activation programs. Unlike PPI methods, this approach considers site-specific protein phosphorylation and other events driving the activation (or inhibition) of effectors that, in turn, modify and regulate other target proteins within a signaling pathway model (14). As an example of the utility of this approach, we identify a novel arm of a MAPK signaling network in platelets and provide mechanistic, biochemical, and cell biological evidence that p38 MAPK targets RTN4 to regulate the cellular physiology of the endoplasmic reticulum in platelet activation programs.

The causal pathway model resulting from our analysis of regulated platelet phosphoproteins noted the MAP kinase p38 (MAPK14) as a putative central, highly connected node in platelet signaling networks. Numerous studies have determined that p38 is phosphorylated and activated in platelets in a variety of contexts, yet specific roles for p38 in platelet function remain enigmatic (2). Importantly, megakaryocyte/platelet-specific knockout models have recently demonstrated a role for platelet p38 in thrombotic and inflammatory pathologies associated with ischemia and myocardial infarction (65). Earlier knockout studies of p38 function in platelets were hindered by embryonic lethality; however, heterozygous p38^+/− mice exhibited reduced thrombus formation in a ferric chloride-induced model of carotid arterial injury (62). Likewise, genetic deletion the p38-activating MAP3K ASK1 (MAP3K5) limits platelet p38 activation, thromboxane generation, secretion, and thrombus formation (51). Studies taking advantage of specific p38 inhibitors similarly support roles for p38 in thromboxane generation via phosphorylation and activation of the cytosolic phospholipase cPLA2 (18, 42). However, roles for p38 in platelet hemostatic function remain less apparent and controversial (17, 43, 47). Using pharmacological inhibitors against p38 and MK2, we provide evidence that p38 has roles in platelet secretion independent of MK2 activation, which regulates cell physiological events at the platelet ER through the phosphorylation of RTN4 and RTN4 interactions with the antiapoptotic Bcl-2 family member Bcl-xl.

While a number of putative p38 substrates are expressed in platelets (19, 22), little is known regarding mechanistic p38 targets in platelet activation programs beyond cPLA2 (28). In addition to cPLA2, the MAP kinase-activated protein kinase MK2 has been assumed to serve as a p38 substrate with roles in platelets via the phosphorylation of the heat shock protein Hsp27 (23, 41, 56); however, evidence for MK2 phosphorylation in platelets has only been recently demonstrated (65). In other cell types, p38 phosphorylates and activates MK2 in response to specific cellular stresses, leading to the phosphorylation of MK2 substrates such as RTN4 and Bcl-xl. Interestingly, RTN4 is also known as a Bcl-xl-interacting and sequestration protein with roles in regulating ER morphology and physiology as well as ER-mitochondria communication underlying apoptotic regulation. In platelets, the endoplasmic reticulum (often referred to as the dense tubular system, or DTS) is highly specialized and interconnected with the platelet open canicular system, microtubule cytoskeleton, and other ultrastructural components. Here we show that in response to thrombin stimulation, platelets upregulate the phosphorylation of RTN4 Ser₁₀₇ as well as Bcl-xl Ser₆₂ in a manner dependent on p38 and MK2 activity that may organize the ER/DTS relative to the mitochondria and other intracellular targets to facilitate localized calcium signaling events that form the basis for platelet activation processes.

Given the requisite roles of RTN4 in regulating ER morphology and physiology in other cell types (38) together with the role of Bcl-xl in mediating platelet PS exposure through apoptosis-related pathways (27, 60), the p38-MK2-RTN4 axis afforded by causal pathway analysis offered an intriguing target for in vitro studies. Following activation with procoagulant stimuli, platelets elevate cytosolic calcium through a variety of mechanisms in a manner requisite for PS exposure and procoagulant activity (64). However, the signaling mechanisms regulating and driving platelet PS exposure and procoagulant activity are unknown. Here, we show that in addition to preventing RTN4 and Bcl-xl phosphorylation in response to thrombin, inhibition of p38 and MK2 dramatically relaxed and expanded the characteristic reticular staining pattern of RTN4 in adherent platelets and altered the morphology and phosphatidylserine surface distribution of procoagulant platelets. These results suggest that p38 signaling may have specific roles in specifying the proaggretory or procoagulant fates of platelets within the context of a growing thrombus in a manner hypothesized by recent studies (50, 68). Such prohemostatic roles for p38 via MK2 and other targets would help to understand the multifaceted and disparate roles for p38 reported in the literature. In addition to better placing p38 into the context of platelet activation programs, our model and results may also help to better understand the role of p38 signaling in other contexts where RTN4 regulates vascular remodeling (1), endothelial cell function (48), pulmonary hypertension (69), and a growing list of other physiological and metabolic processes generally rooted in ER-mitochondria communication. Indeed, future efforts to investigate and target other components of the ER-mitochondria axis beyond RTN4 such as mitofusin-2 (29) and PACS-2 (9, 67) may also help to advance studies of platelets for basic science and translational efforts.

Beyond roles in organizing platelet signaling in hemostasis, thrombosis, and immunity, p38 provides an interesting target for extending platelet lifetime in storage for transfusion. Indeed, a number of studies have noted that p38 is activated in stored platelets and that p38 inhibition can decrease PS exposure in aging platelets (24) as well as in response to UV treatment (27). Interestingly, oxidative stress activates ADAM17/TACE (a p38 target noted in our pathway analysis in Fig. 2) and induces target receptor shedding in platelets in a p38-dependent fashion, and p38 inhibition has been suggested to increase platelet lifetime (20). Given the predicted diverging and more specific roles of MK2 downstream of p38 described herein, MK2 inhibition may serve as a more optimal strategy to extend platelet lifetime. Even more specialized RTN4 inhibitors that disrupt ER organization in cancer cells in a manner similar to that observed in platelets with MAPK inhibition in this study may also help to limit platelet PS exposure and extend the viability of platelets in storage (15).

In addition to the MAPK-RTN4 axis described herein, causal pathway analysis highlighted several other targets and systems of interest to platelet physiological function for future studies (Fig. 2). For instance, DAPP1, a recently described component of the platelet PI3K signalosome, was noted as a functionally phosphorylated target in the context of platelet activation (31). Furthermore, a number of receptors associated with platelet activation and inflammatory signaling are suggested to be phosphorylated in an “inside-out” signaling-like manner, including thromboxane receptor TBXAR2 as well as the disintegrin ADAM17, chemokine receptor (CXCR4), glucocorticoid receptor (NR3C1), insulin receptor substrate (IRS1), and IL-6 receptor (IL6ST). Along these lines, an ASK1-p38 axis was recently demonstrated to have a role in phosphorylating the P2Y12 receptor to support signaling processes through ADP that sustain Akt activation in platelets (40). While our work highlights some potential novel signaling steps in platelet regulation for future studies, it should be noted that pathway models such as those generated herein are dependent on several developing factors, including advancing proteomics tools, informatics technologies, data curation methods, and the careful description and cataloging of gene/protein functions. Nonetheless, despite any current limitations, as omics tools and informatics databases continue to mature, causal pathway tools will become an invaluable tool in organizing, modeling, and discovering novel signaling routes and targets in platelets and other physiologically relevant cell and tissue systems.

GRANTS

This work was supported by the Knight Cardiovascular Institute and grants from the American Heart Association (17SDG33350075 to J. E. Aslan and 13EIA12630000 to O. J. T. McCarty), National Institutes of Health (5U41-HG-006623 to E. Demir and R01-HL-101972 and R01-GM-116184 to O. J. T. McCarty), and Army Research Office-DARPA Big Mechanism Program (W911NF-14-C-0119 to E. Demir). The super resolution microscopy (SR-SIM) studies were supported by the M. J. Murdock Charitable Trust. Z. Rub was supported by a Leadership, Innovation and Liberal Arts Center (LILAC) Summer Internship from Bryn Mawr College (Bryn Mawr, PA).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

Ö.B., J.P., and J.E.A. conceived and designed research; Ö.B., A.T.P.N., R.A.R., J.P., Z.T.R., A.E.B., and J.E.A. performed experiments; Ö.B., A.T.P.N., R.A.R., and J.E.A. analyzed data; Ö.B., A.T.P.N., R.A.R., J.P., A.M., O.J.T.M., E.D., and J.E.A. interpreted results of experiments; Ö.B., A.T.P.N., and J.E.A. prepared figures; Ö.B., A.T.P.N., and J.E.A. drafted manuscript; Ö.B., A.T.P.N., R.A.R., A.M., L.L.D., O.J.T.M., E.D., and J.E.A. edited and revised manuscript; Ö.B., A.T.P.N., R.A.R., J.P., Z.T.R., A.M., L.L.D., O.J.T.M., E.D., and J.E.A. approved final version of manuscript.