Significance of neutrophil microparticles in ischaemia‐reperfusion: Pro‐inflammatory effectors of endothelial senescence and vascular dysfunction

Abstract Endothelial senescence is an emerging cause of vascular dysfunction. Because microparticles are effectors of endothelial inflammation and vascular injury after ischaemia‐reperfusion, we examined leucocyte‐derived microparticles of spleen origin as possible contributors. Microparticles were generated from primary rat splenocytes by either lipopolysaccharide or phorbol‐myristate‐acetate/calcium ionophore, under conditions mimicking innate and adaptive immune responses. Incubation of primary porcine coronary endothelial cells with either type of microparticles, but not with those from unstimulated splenocytes, leads to a similar threefold raise in senescence‐associated β‐galactosidase activity within 48 hours, indicating accelerated senescence, to endothelial oxidative stress, and a fivefold and threefold increase in p21 and p16 senescence markers after 24 hours. After 12‐hour incubation, the endothelial‐dependent relaxation of coronary artery rings was reduced by 50%, at distinct optimal microparticle concentration. In vitro, microparticles were pro‐thrombotic by up‐regulating the local angiotensin system, by prompting tissue factor activity and a secondary generation of pro‐coagulant endothelial microparticles. They initiated an early pro‐inflammatory response by inducing phosphorylation of NF‐κB, MAP kinases and Akt after 1 hour, and up‐regulated VCAM‐1 and ICAM‐1 at 24 hours. Accordingly, VCAM‐1 and COX‐2 were also up‐regulated in the coronary artery endothelium and eNOS down‐regulated. Lipopolysaccharide specifically favoured the shedding of neutrophil‐ and monocyte‐derived microparticles. A 80% immuno‐depletion of neutrophil microparticles reduced endothelial senescence by 55%, indicating a key role. Altogether, data suggest that microparticles from activated splenocytes prompt early pro‐inflammatory, pro‐coagulant and pro‐senescent responses in endothelial cells through redox‐sensitive pathways. The control of neutrophil shedding could preserve the endothelium at site of ischaemia‐reperfusion–driven inflammation and delay its dysfunction.


| INTRODUC TI ON
Endothelial damage is a prime sensor of ischaemia-reperfusion and a potential inducer of pro-coagulant and pro-inflammatory responses via leucocyte and platelet recruitment associated with oxidative stress, all characterizing ischaemia-reperfusion injury (IRI). IRI is a major cause of graft damage characterized by oxidative stress and inflammation. Upon restoration of blood flow, accumulation of reactive oxygen species (ROS) and the release of cytokines prompt the endothelial up-regulation of adhesion molecules, and the consecutive monocyte and neutrophil recruitment and extravasation into the post-ischaemic tissues. 1 Chemokines also favour neutrophil and monocyte recruitment in the early stages of reperfusion of transplanted organs, 2 thereby amplifying IRI-induced inflammation. 3 Consequently, graft damage would favour immediate or acute rejection peri-transplantation, as typified in pancreatic islet transplantation which is associated with a particular form of IR termed instant blood-mediated inflammatory reaction (IBMIR). 4,5 The noxious interaction between damaged endothelial cells (ECs) and leucocytes demonstrated ex vivo, and in transplantation 6 or thrombosis animal models, 7 suggests that IRI-driven initial graft damages would favour immediate or acute peri-transplantation rejection through vascular dysfunction.
Circulating endothelial cells and endothelial-derived microparticles (EMPs) are a signature of endothelial damage 8 post-IRI in humans. Elevated plasma levels of EMPs were associated with cardiac rejection and the duration of cold-ischaemia. 9 Conversely, in patients with renal failure, post-transplantation EMP levels decrease, indicating endothelial recovery. 10 Microparticles (MPs), also referred to as microvesicles, are vesicles shed from the plasma membrane of activated cells released in body fluids. 11 MPs harbour cell membrane proteins and contain active lipids, proteins and mRNA, making them pro-inflammatory and pro-coagulant cellular effectors. They also identify the cell lineage they were stemmed from. 12 16 and worsened outcome in stem cell transplantation. 17 Pro-inflammatory MPs from platelets, lymphocytes and monocytes would favour leucocyte recruitment at the inflamed endothelium surface known to expose adhesion molecules, like ICAM-1, VCAM-1 and E-selectin, thereby accelerating IRI. [18][19][20] Interestingly, a recent report describes a cardio-splenic axis that augments infarct size during post-ischaemic reperfusion via leucocyte activation and the recruitment of spleen neutrophils at site of the ischaemic heart, 21 eventually favouring local endothelial dysfunction.
The impact of immune cells during IR remains unexplored with respect to endothelial senescence and vascular dysfunction.
Endothelial senescence shifts the endothelium to a pro-coagulant and pro-inflammatory status with major TF activity, endothelial dysfunction and pro-senescent EMP shedding, 22,23 and favours the development of a senescence-associated secretory phenotype (SASP), mainly consisting of cytokines and metalloproteases. 24 To decipher the inflammation-driven endothelial damages during IRI, we used the rat spleen as a source of immune cell lineages and assessed the effects of MPs generated from splenocytes (SMPs), under conditions mimicking post-ischaemic stress on endothelial accelerated senescence and dependent vascular function by in vitro and ex-vivo SMP-mediated crosstalk models.

| Animals
Male Wistar rats (300 g, Janvier Labs) were maintained on a stand-

| Rat splenocyte isolation and culture
After sacrifice, spleens were withdrawn, weighed and washed in phosphate-buffered saline (PBS) before homogenization under sterile conditions and further filtered through 100-µm sterile cell strainers. Following 300 g centrifugation for 5 minutes, the cell pellet was could preserve the endothelium at site of ischaemia-reperfusion-driven inflammation and delay its dysfunction.

K E Y W O R D S
endothelial senescence, inflammation, ischaemia-reperfusion, microparticles, transplantation, vascular dysfunction re-suspended in a 5-mL ammonium-chloride-potassium erythrocyte lysis buffer for 5 minutes, centrifuged (300 g, 5 minutes), washed in PBS and re-suspended in FBS-supplemented complete RPMI-1640 medium and seeded into T75 culture flask at 5.10 6 cells/mL.

| Generation, isolation, quantification and characterization of splenocyte-derived MPs
Splenocytes were stimulated by 5 µg/mL of lipopolysaccharide (LPS) (0127: B8; Sigma) or a combination of phorbol 12-myristate 13-acetate (PMA) (25 ng/mL; Enzo) and A12387 calcium ionophore (HBSS) (supporting information). SMP measurement was performed by pro-thrombinase assay after capture onto Annexin V-coated micro-wells using a microplate spectrophotometer and their concentration referred to as phosphatidylserine (PhtdSer) equivalent, by reference to a standard curve made with synthetic vesicles of known amounts of PhtdSer. In this assay, PhtdSer. is the rate-limiting factor of prothrombin conversion into thrombin. 25 The size distribution analysis of SMPs was performed by Tunable Resistive Pulse Sensing technology (TRPS) using the Izon qNano system and Izon control suite 3.3 software (Izon Science Ltd). Characterization of the SMP cell origin was performed by capture onto biotinylated antibodies directed against leucocyte CDs before quantification by pro-thrombinase assay (supporting information). The SMP concentration was obtained by subtracting the OD values measured using istotype control biotinylated IgGs.

| Isolation and treatment of Coronary Artery Primary Endothelial Cells
Pig hearts were collected from the local slaughterhouse (COPVIAL, Holtzheim, France) and ECs freshly isolated from the left circumflex coronary arteries (supporting information).

| Measurement of apoptosis and senescenceassociated β-galactosidase activity
Apoptosis and SA-β-gal were measured by flow cytometry using propidium iodide and AnnexinV (PI/AV) double labelling, or the C12FDG fluorogenic cell-permeable substrate. SA-β-gal was revealed on the EC monolayer by microscopy using the X-gal chromogenic substrate (supporting information).

| Western blot
Experiments were performed as described in supporting information. Auto-fluorescence gains were set at the first logarithmic decade (supporting information).
Conversion of Factor X into Xa by TF was revealed by the cleavage of a specific chromogenic substrate (CS11, 0.1 mmol/L, Hyphen Biomed). Variations in absorbance were recorded using a microplate spectrophotometric reader in kinetic mode set at 405 nm (Molecular Device). Data are expressed as fM of active TF per 5. 10 4 living ECs by reference to a standard curve established with known amounts of highly purified lipidated recombinant human TF (ADF Biomedical).

| Vascular reactivity
Vascular reactivity of coronary artery rings was assessed as previously described. 27 Briefly, the coronary artery was cut into Relaxations are expressed as a percentage of the contraction to U46619.

| Immunofluorescence microscopy of coronary arteries
Pig coronary arteries were incubated with SMPs for 24 hours before OCT inclusion and immunostaining (supporting information).

| Selective depletion of splenocyte-derived microparticles
Immuno-magnetic depletion of SMP of specific cell origins was achieved using biotinylated antibodies insolubilized on streptavidincoated beads (50 μL beads-300 μL of SMPs) and directed against either rat neutrophil CD11b, monocytes/macrophages CD11b/c or endothelial CD31 (DynabeadsM-280, Thermo Fisher). Control isotype IgG-coated beads served as control. Immuno-depletion was achieved by the addition of 1 μg of a chosen antibody-coated bead suspension to the concentrated SMPs and the mixture further incubated for 30 minutes at 4°C, before a 2-minute contact with the magnet. A double depletion step allowed 75%-80% of specific SMP depletion by pro-thrombinase capture assay.

| Statistical analysis
Data, expressed as mean ± standard error mean (SEM), were analysed using GraphPad Prism5 ® . Statistical analysis was performed using one-way ANOVA test followed by the Tukey's post hoc or Mann-Whitney tests. A P < .05 was considered significant. All measurements were from at least three separate individuals.

| Differential impact of LPS or PMA/I treatments on splenocyte cell subsets
The leukocyte cell phenotype pattern differed with the splenocyte treatment, indicating distinct triggering of adaptive and innate immunity by PMA/I and LPS. PMA/I mainly altered lymphocyte subsets with enhanced proportions of B cells, CD8 + -T lymphocytes and CD4 + -CD25 + T regulatory cells and decreased CD4 + T Lymphocytes.

| LPS or PMA/I induces the generation of splenocyte-derived microparticles that interact with young endothelial cells
Compared to vehicle, a significant SMP release was measured by prothrombinase assay after 24 hours, by, respectively, 2.5-and 3.5-fold in response to LPS (SMP LPS ) or PMA/I (SMP PMA/I ), together with a twofold and threefold increase in splenocyte apoptosis ( Figure S2A,B).
Interestingly, kinetics of PKH26-labelled SMPs integration into target plasma membrane of ECs were similar for all SMPs, 28% of fluorescent P1ECs being stained after 6 hours, the proportion increasing thereafter to a 82% plateau at 24 hours, by flow cytometry ( Figure 1C).

| SMP LPS or SMP PMA/I are specific inducers of premature senescence in target P1ECs with no proapoptotic effect
SMP LPS or SMP PMA/I but not SMP CTL (P = .1002) markedly increased SA-β-gal activity measured by flow cytometry in target P1ECs after 48 hours. Significant elevations were observed in response to 10-30 nmol/L of either SMP LPS or SMP PMA/I , the higher concentration inducing a SA-β-gal activity similar to that observed in P3 senescent cells ( Figure 1D). Simultaneously, the absence of apoptosis, even after 30 nmol/L SMPs treatment, suggested a specific pro-senescent effect, as confirmed by PI/AV labelling and Western blot analysis of cleaved caspase-3. Conversely, 100 µmol/L H 2 O 2 prompted maximum EC apoptosis (33.67 ± 2.57% vs 9.34 ± 0.72% in untreated P1ECs, Figure S3A,B).
The pro-senescent feature of SMP LPS or SMP PMA/I was also evidenced on EC monolayers by transmission microscopy, using the X-Gal blue SA-β-gal substrate. A significant threefold increase in cells bearing SA-β-gal activity was induced after 48 hours, values reaching those observed in senescent P3 cells ( Figure 1E). In addition, the expression of senescence markers increased as early as 24 hours after incubation with SMP LPS or SMP PMA/I . A fourfold up-regulation of p53, and a fivefold and threefold increase in down-stream p21 and p16 proteins was respectively observed in P1ECs. The absence of SMP CTL -driven up-regulation of p16 (P = .7261) confirms the specific pro-senescent properties of SMP LPS or SMP PMA/I (Figure 1F-H).
The SMP-driven pro-senescent and specific effect was confirmed by the fact that concentrations of truly soluble LPS (0.5-5 µg/ mL) or PMA/I (2.5-25 ng/mL/0.1-1 µmol/L) remained unable to promote senescence. Furthermore, no soluble moiety from splenocyte supernatant could account for the induction of premature senescence, as incubation of P1ECs with SMP-free conditioned medium did not lead to senescence ( Figure S4A,B).

| SMP LPS or SMP PMA/I induce oxidative stress in target P1ECs
As ROS are strong inducers of senescence, we examined their variations in 10 nmol/L SMP LPS or SMP PMA/I -induced premature senescence. An early significant twofold ROS increase was induced after 6 hours in P1ECs cytoplasm and mitochondria (Figure 2A  P1ECs. ACE and AT1R were significantly up-regulated, respectively, by fourfold and twofold ( Figure 3D,E), suggesting a thrombogenic effect as reported. 22 To confirm the SMP-mediated endothelial dysfunction, SMP LPS or SMP PMA/I were incubated during 12 hours with porcine coronary artery rings before assessment of bradykinin endothelial-induced relaxation. Both SMP LPS and SMP PMA/I blunted relaxation in a dose-dependent manner, SMP PMA/I being more efficient with a 50% inhibition measured in response to a 10 nmol/L concentration vs 30 nmol/L for SMP LPS (Figure 3F,G). Of note, SMP CTL had no effect, even at 30 nmol/L. The MP-driven specific inhibition of relaxation was also confirmed by the fact that truly soluble 5 µg/mL of LPS or 25 ng/mL/1 µmol/L of PMA/I remained unable to alter bradykinin-induced vaso-relaxation ( Figure S5A-C). Furthermore, SMPs lysates had no effect, indicating the pivotal role of the SMP membrane proteolipid structure for endothelial targeting (data not shown).

| SMP LPS or SMP PMA/I prompt MAPKs and PI3 kinase pathways
As MAP kinases were involved in senescence signalling, 22  Data are expressed as mean ± SEM of experiments performed at least on three separate occasions. *P < .05, **P < .01 51 ± 3.6% respectively, as revealed by SA-β-gal activity measured after 48 hours ( Figure 7A).
Altogether, data suggest that SMP LPS or SMP PMA/I induce premature ECs senescence through an early activation of the redox-sensitive MAP-Kinase and the PI3 kinase/Akt pathways.

| Neutrophil-and monocyte-derived SMPs contribute to early endothelial senescence
Because the SMP LPS showed a specific profile with 65% elevation in neutrophil and monocyte cell origin compared to SMP PMA/I (Table S1), we further characterized their specific contribution to SMP-driven senescence. As we previously demonstrated that EMPs from senescent cells are pro-senescent, 22 endothelial CD31 + -SMP LPS were depleted before specific removal of either neutrophilor monocyte-derived SMP LPS (Table S2).

| D ISCUSS I ON
We report herein that MPs generated from LPS-or PMA/calcium ionophore-treated splenocytes induce endothelial dysfunction and premature senescence via early NADPH oxidase-dependent inflammatory responses. SMPs trigger NF-κB and both MAPK and PI3 kinase/Akt pathways. Endothelial senescence was characterized by drastic eNOS down-regulation, oxidative stress, up-regulation of TF expression and enhanced activity, and the generation of secondary pro-coagulant EMPs. The MP-driven endothelial dysfunction was confirmed by the collapse of endothelial-dependent vascular relaxation. Importantly, SMPs were strictly pro-senescent with no proapoptotic potential, indicating a specific effect.

| Splenocytes as a valuable source of immune cell-derived MPs
Because spleen was reported to exacerbate the inflammatory response during myocardial infarction and worsen the infarct size, 21 we reasoned that SMPs would constitute a IRI-driven inflammatory signature. Indeed, in rat models, an intraperitoneal sub-septic LPS dose leads to transient cytokine secretion associated to early monocyte and neutrophil spleen recruitment within 3 hours. 28 Consistently, in our primary splenocyte model, the SMP pattern is

| SMP LPS are early contributors to senescence
SMPs shared common endothelial pro-senescent effects, with distinct impacts on key protein expression or activity. In ECs, SMP LPS were more effective in the up-regulation of senescent markers p16 and p21 and prompted higher pro-coagulant and pro-inflammatory typical responses like TF and VCAM-1 up-regulation, with an earlier IκBα activation, whereas MAP kinases, PI3 kinases and Akt were mostly phosphorylated at comparable time by both SMPs.
Interestingly, SMP LPS or SMP PMA/I acted as differential vascular effectors in the blunting of endothelial-dependent relaxation, SMP PMA/I being more efficient. Immuno-depletion assays confirmed that neutrophil-and monocyte-derived SMP LPS are key to premature endothelial senescence.

| Key role of spleen neutrophils in the generation of pro-senescent microparticles
Several previous reports, although not clearly deciphering the respective contribution of microparticles from exosomes, strongly suggest paracrine effects of circulating neutrophil microparticles (CNMPs) on various endothelial territories. In models of sepsis and inflammatory disorders including atherothrombosis, CNMPs disrupt the endothelial barrier and up-regulate pro-inflammatory mediators such as IL-6, IL-8 and ROS. [33][34][35][36] During inflammation, neutrophil migration from tissue compartments may partly account for the rapid shift of blood neutrophil count (10%-25% in mice, 50%-70% in human) 37

| SMP LPS or SMP PMA/I as early pro-inflammatory effectors of premature endothelial senescence triggered by innate immune cells
The pro-senescent effect of both SMP LPS and SMP PMA/I was TAK kinase-dependent with an early activation of NF-κB only 1 hour after challenge, suggesting a SMP-mediated pro-inflammatory pathway. Our observation of an early activation of p38 and JNK MAP kinases confirms the key role of pro-inflammatory pathways and mediators in the SMP-driven endothelial senescence. Accordingly, p38 MAPK activation was reported a redox-sensitive contributor to senescence in HUVEC after serum starvation, using p38 silencing, 42 and to the release of pro-inflammatory EMPs from human aortic ECs in response to TNFα. 43 Because NF-κB signalling is involved in both endothelial SASPs and in the anti-inflammatory control of senescence by caveolae initiating after 2 days and increasing thereafter in response to TNFα, 44,45 kinetics of SMP-driven SASPs remain to be established beyond 48 hours to identify mechanisms of prolonged endothelial senescence. Altogether, monocyte and neutrophil MPs appear paracrine inductors of early endothelial inflammation, which in turn amplifies dysfunction and senescence, possibly via the secondary shedding of pro-inflammatory EMPs already known as pro-senescent effectors. 22,23,46 In our model, as the depletion of monocyte-and neutrophil-derived SMP LPS maintains ECs at a pre-senescent P2 stage, these MPs are likely initiators of the senescence responses. Nevertheless, the contribution of SMP PMA, mainly of lymphocyte origin remains to be explored.

| SMP LPS or SMP PMA/I as pro-oxidant and procoagulant endothelial effectors of coronary artery dysfunction
ROS are known mediators of both premature and replicative senescence in coronary and aortic ECs. 22,23,47 Furthermore, EMPs released by premature or replicative senescent cells are autocrine inducers of superoxide anion via NADPH oxidase and cyclooxygenase, amplifying the endothelial dysfunction. 22,23 Similarly, using a pharmacological approach, we herein demonstrate that both SMP LPS and SMP PMA/I are pro-senescent and promote a major early oxidative stress in ECs via NADPH oxidase and cyclooxygenase-2. Our data do not confirm the ROS accumulation and enhanced NO production combined to unchanged eNOS activity previously reported in the Eahy.926 lineage treated by THP-1-derived MPs. 48 Discrepancies might rely on the initial stress (VP-16-induced apoptosis) or on cell type.
Ageing is associated with progressive endothelial senescence and dysfunction and enhanced circulating MP levels, 49,50 while hampering graft survival. 51,52 Our ex vivo data showed that pro-senescent SMPs blunt bradykinin-induced relaxation of coronary arteries, indicating the collapse of endothelial NO-mediated vaso-protection owing to reduced eNOS expression and up-regulated COX-2, whereas SMP CTL were ineffective. Taken together, our data confirm and extend previous reports indicating that the endothelium-dependent vaso-relaxation is a prime target of circulating pathological MPs isolated from patients with acute myocardial infarction. 53 Other authors and our team have reported that angiotensin II triggers endothelial senescence and MP shedding 23 through AT1R, enhanced ACE activity and redox-sensitive pathways. 23,47 In accordance, we show herein that SMP-induced premature endothelial senescence is characterized by up-regulated AT1R and ACE. In the absence of AT1R mRNA quantification, the present data should however be balanced in view of the well-known broad specificity of the antibodies against the AT1R. 54 Nevertheless, we demonstrate an early enhancement in TF activity, in compliance with the reported F I G U R E 7 SMP LPS induce endothelial senescence through MAP-Kinase and PI3 kinase pathways, and via neutrophil-derived MPs. A, P1ECs were incubated either with 100 μmol/L H 2 O 2 , or a selective inhibitor of PI3 kinase (PI3Ki, LY294002, 10 μmol/L), p38 MAPK (p38i, SB203580, 10 μmol/L) or ERK ½ (ERKi, PD98059, 10 μmol/L) for 1 h prior to the addition of 10 nmol/L SMP LPS or SMP PMA/I for 48 h before the determination of SA-β-gal activity. B, Effect of the removal of either CD31+/CD11b+-MPs or CD31+/ CD11b+/CD11b/c+-MPs on the endothelial pro-senescent effect of SMP LPS . P1ECs were incubated with 10 nmol/L of SMP LPS and with non-depleted suspensions for 48 h before the determination of SA-β-gal activity. C, Characterization of the cell origin of SMP shed from splenocytes stimulated by LPS or PMA/calcium ionophore. After 24-h stimulation, SMPs were measured in cell supernatants by pro-thrombinase assay and expressed as nmol/L PhtdSer per 100.10 6 cells. Characterization of the SMP cell origin was performed by capturing SMPs onto biotinylated antibodies directed against leucocyte CDs before quantification by pro-thrombinase assay. Data are expressed as mean ± SEM of experiments performed at least on three different cell cultures. AV: Annexin V. *P < .05, **P < .01 TF-associated endothelial dysfunction. 22,55 As the TF gene is early expressed in response to cytokines, it is tempting to anticipate that SMP-driven endothelial senescence is linked to a rapid TF-driven pro-coagulant switch of the endothelium. Interestingly, no PAR-1mediated inflammatory response could be identified in spite of PAR-1 up-regulation after 24 hours. Possible explanations would be (a) an earlier burst of PAR-1 glycosylation, (b) a counter-regulation of the PAR-1-induced chemokine and cytokine up-regulation initiated by the suppressor of cytokine signalling 1 (SOCS-1), as in human endometrial ECs. 56,57 Because VCAM-1 and ICAM-1 were up-regulated in isolated ECs after 24 hours, or in the EC lining the rings, it can be anticipated that in the vessel, an acute generation of leukocyte MPs favours endothelial inflammation and consecutive senescence. Early consequences of IR remain difficult to decipher in-vivo.
SMPs therefore appear suitable tools in alternate approaches to assess IRI-driven cellular activation and for the monitoring of drugs targeting neutrophil and monocyte endothelial interactions, eventually triggered by NETosis. 66 Our spleen MP-mediated endothelial senescence model is a reliable tool in deciphering mechanisms of ageing-induced inflammatory responses as well as those of accelerated endothelial senescence during transplantation-associated IR. 67 Endothelial cyto-protection of highly vascularized grafts such as pancreatic islets, and the control of pro-senescent MP shedding might limit IR in the early stages of transplantation, preserves endothelium at sites prone to early vascular injury and delay its dysfunction.

CO N FLI C T O F I NTE R E S T
The authors of this manuscript have no conflict of interest to disclose.

AUTH O R CO NTR I B UTI O N S
AE performed crosstalk experiments, analysed data and wrote the paper. RA performed ex vivo vascular response investigations; MA, AWQ and HE contributed to flow cytometry analysis; FZ measured microparticles, MK, GK, LA and SK contributed to the initial experimental design; VS discussed some aspects of the manuscript; LK supervised the experimental design and discussed the manuscript; and FT conceived and supervised the study, and wrote and revised the manuscript.

DATA AVA I L A B I L I T Y S TAT E M E N T
All data that support the findings of this study are available within the article and its supporting information file and from the corresponding author upon reasonable request.