The characteristics of all subjects are shown in Table 1. We obtained blood samples from 43 patients with pSS fulfilling American-European Consensus Group criteria 5, 20 with SLE fulfilling American College of Rheumatology criteria 6 and 24 with RA fulfilling American College of Rheumatology criteria 7 in the Department of Rheumatology of Bicêtre University Hospital. The study was approved by the local research ethics committee, and informed written consent was obtained from all patients.

Among the 43 pSS patients, extra-glandular involvement as previously defined 8 was present in 17 patients: lung involvement (n = 3), neurological involvement (n = 4), active synovitis (n = 2), myositis (n = 2), vasculitis (n = 2), renal involvement (n = 1), and lymph node enlargement (n = 3). Five patients had malignant hemopathy, three with marginal zone lymphoma (one current, two previous) and two current multiple myeloma, and two had monoclonal gammopathy of undetermined significance (MGUS). Seven pSS patients received immunosuppressive drugs (rituximab, n = 3; rituximab plus methotrexate, n = 1; cyclophosphamide plus melphalan, n = 1; methotrexate, n = 2).

For patients with SLE, disease activity was measured by the SLE Disease Activity Index (SLEDAI) on the day of blood testing 9. Eleven patients received immunosuppressive drugs (mycophenolate mofetyl, n = 7; azathioprine, n = 2; rituximab, n = 2; prednisone >10 mg daily, n = 4). Four patients presented with a secondary anti-phospholipid syndrome according to international criteria 10. Patients with acute or chronic infections or with primary anti-phospholipid syndrome were excluded from the study.

For RA patients, disease activity was measured by the Disease Activity Score for 28 joints (DAS28) on the day of blood testing. Immunosuppressive agents were given to 19 RA patients (methotrexate, n = 16; anti-TNFα agents, n = 5; leflunomide, n = 2); no patient received steroids more than 10 mg daily.

As controls, after informed consent was obtained, we used a group of 44 healthy controls (HCs) who presented no inflammatory, neoplasic, autoimmune or metabolic diseases.

Cardiovascular risk factors (diabetes, smoking, arterial hypertension, hyperlipidemia) were noted in three patients with pSS, one with SLE, and six with RA. Of note, at the time of blood testing, no patient or controls presented signs of acute thrombosis or infection known to modify the plasma level of MPs.

According to a standardized procedure 1112, after collection of citrated fresh blood samples, MPs were isolated by double centrifugation at 1500 g for 15 minutes and 13,000 g for 2 minutes at room temperature and immediately stored at -80°C for further analysis. This procedure has been previously validated as mainly yielding MPs and excluding larger apoptotic bodies, eliminated by the two centrifugation steps 12.

An aliquot of plasma obtained before the second centrifugation was kept for assessment of secretory phospholipase A2 (sPLA2) activity and sCD40L and soluble P-selectin (sCD62P) content.

Circulating MPs were captured onto insolubilized annexin V and were called total MPs because annexin V-positive MPs represent the large majority of the MP population. Capture with monoclonal antibodies (mAbs) (mAb against human platelet anti-glycoprotein Ib (GPIb) and CD11a) was performed for platelet and leukocyte MP isolation, respectively. Then quantification of these captured MPs was performed with a functional prothrombinase assay in which concentrations of purified clotting factors and calcium (factor Xa, factor Va, prothrombin and CaCl₂) were determined to ensure that PS was the rate-limiting parameter of the generation of thrombin from prothrombin 1112. Thus, thrombin generation is dependant on the PS content, which is proportional to the immobilized MPs. Results are expressed as nanomolar PS equivalent (nM PS Eq) by reference to a standard curve constructed with liposomes of known PS concentrations 13. For capture by CD11a and GPIb, background values obtained with irrelevant immunoglobulin (Ig) Gs of corresponding isotypes were subtracted from those measured with specific mAbs. Different affinities of MPs for annexin V and mAbs prevent direct comparison or addition between levels of MPs measured with use of these ligands.

Flow cytometry experiments were adapted from Combes and colleagues 14 and Robert and colleagues 15. All analyses were performed by use of a fluorescent-activated cell sorter (FACS; EPICS XL; Beckman Coulter, Roissy, France) and RXP-software analysis (Beckman Coulter). Forward scatter and side scatter were set as a logarithmic gain, and Megamix (Biocytex, Marseille, France), containing a mix of fluorescent microbeads of various diameters (0.5, 0.9 and 3.0 μm), was used for initial settings and before each experiment to measure MPs, as an internal control. Gates were then set to include events between 0.5 and 1.0 μm with exclusion of background corresponding to debris usually present in buffers.

We incubated 40 μL of platelet-free plasma MPs for 30 minutes in the dark at room temperature with annexin V-fluorescein isothiocyanate (FITC; Beckman Coulter, Roissy, France), specific antibodies or isotype-matched irrelevant control (10 μL) conjugated with phycoerythrin after gentle shaking, and 400 μL of annexin V buffer (containing calcium ions) or PBS1X was added before immediate acquisition. Two negative controls of annexin V ligation were used: MPs incubated with annexin V in a calcium-free buffer (to prevent annexin V ligation to PS) or without annexin V in a specific annexin V buffer to estimate the auto-fluorescence.

MP subpopulations were determined according to the expression of membrane-specific antigens from platelets and leukocytes by use of anti-CD61 and anti-CD45 mAbs (Beckman Coulter, Roissy, France), respectively. Staining with isotype-matched irrelevant mAbs (Beckman Coulter, Roissy, France) at the same concentration and under the same conditions was used as a control.

Before acquisition, a known number of 3 μm calibrated microbeads (Sigma Aldrich, Saint Louis, MO, USA) was placed in each tube and run concurrently with the MP samples in the FACS, thus allowing for quantitative determination of MPs (annexin V-positive or from different origin). The absolute number of MPs per millimeter plasma was then determined by counting the proportion of beads and the exact volume of plasma from which MPs were analyzed. The analysis was stopped when a fixed number of microbeads (10,000) were counted. Results are expressed as number of MPs per microliter by the formula N = (MP × beads per tube/volume of plasma)/number of counted beads.

We assessed the functional activity of sPLA2 because plasma sPLA2 is able to hydrolyze phospholipids such as PS or phosphatidylcholine present in MPs 16. Plasma sPLA2 activity, expressed as nanomoles per minute per millilitre (nmol/min per mL), was measured by selective fluorometric assay as previously described 17.

Platelet activation was assessed by measurement of sCD40L and sCD62P concentrations in plasma by use of the human sCD40L Quantikine kit and a human sCD62P Immunoassay (R&D Systems, Lille, France), respectively, following the manufacturer's instructions.

Anti-Ro/SSA and anti-La/SSB antibodies and IgG anti-double-stranded DNA (dsDNA) antibodies were determined by counter-immunoelectrophoresis and ELISA, respectively, as described previously 18. The serum β2 microglobulin level was determined by nephelometry (Array 360 system, Beckman Coulter, Roissy, France) as previously described 8.

For biological anti-phospholipid investigations, anticardiolipin and anti-β2GPI antibody levels were assessed by ELISA (Biorad, Marne la Coquette, France and INOVA Diagnostics, San Diego, CA, USA, respectively).

Other biological tests were performed as routine in the Departments of Hematology and Biochemistry of our hospital (leukocyte and platelet counts, fibrinogen and C-reactive protein levels).

Characteristics of patients are expressed as number and percentage and median and range. Results for MP levels are expressed as mean ± standard error of the mean. Comparisons of mean MP levels, sPLA2 activity, sCD40L and sCD62P concentrations between different groups of subjects (independent analysis) were analyzed by non-parametric Mann-Whitney U test. Spearman's rank correlation coefficients were calculated to investigate the relation between MP counts and clinical and biological parameters. A P < 0.05 was considered statistically significant. Statistical analysis involved use of GraphPad Prism 5 software (GraphPad Software Inc., San Diego, CA, USA).

MPs detectable by capture onto annexin V were measured in 43 pSS, 20 SLE and 26 RA patients and 44 HCs. The level of total MPs was significantly higher in patients with pSS (8.49 ± 1.14 nM PS Eq), SLE (7.3 ± 1.25 nM PS Eq), and RA (7.23 ± 1.05 nM PS Eq) than in HCs (4.13 ± 0.2, P < 0.004; Table 2, Figure 1a). This increase involved particularly platelet-derived MPs (Table 2, Figure 1b). However, pSS, SLE and RA patients did not differ in level of total or platelet MPs.

The level of leukocyte-derived MPs was higher in patients with pSS than in HCs (5.78 ± 0.37 versus 3.92 ± 0.21 nM PS Eq, P < 0.0001), with no difference between HCs and patients with SLE or RA (3.89 ± 0.4 and 4.28 ± 0.9, P = 0.46 and P = 0.18, respectively; Table 2, Figure 1c). Moreover, the level of leukocyte-derived MPs in pSS was significantly higher than that in SLE or RA (P = 0.003 and P = 0.015, respectively; Figure 1c).

The number of patients with cardiovascular comorbidities was low, yet after excluding these subjects, the results of statistical analyses remained unchanged (data not shown). Moreover, in pSS patients, the MP levels was the same in patients with hemopathy (lymphoma, multiple myeloma or MGUS; n = 7) and the others (n = 36; total MPs: 8.6 ± 1.3 vs 7,8 ± 2, P = 0.93). Likewise, MPs level was not different in SLE patients with (n = 4) or without secondary APLS (n = 16; total MPs: 7.7 ± 1.3 vs 5.5 ± 1.2, P = 0.48).

Of note, the level of leukocyte-derived MPs and absolute leukocyte count were not correlated, nor was the level of platelet MPs and platelet count in each group of patients correlated (data not shown).

We assessed the plasma levels of total, leukocyte and platelet MPs by concomitant capture assay and flow cytometry in 17, 8 and 15 subjects, respectively. Results are in Table 2, and a representative staining is in Figure 2. Results from the two measurements showed a significant positive correlation for total MPs (r = 0.72, P = 0.001) as well as for platelet MPs (r = 0.76, P = 0.04) and for leukocyte MPs (r = 0.54, P = 0.04).

As MPs can reflect the state of cellular stimulation, we hypothesized that the level of MPs could be associated with AID activity. Platelet MP levels were significantly lower in pSS patients with extra-glandular involvement (3.88 ± 2.3 nM PS Eq) than in those with only glandular involvement (5.5 ± 1.45 nM PS Eq, P = 0.02) with a similar tendency for total (7.93 ± 2.25 versus 8.86 ± 1.2 nM PS Eq, P = 0.06) and leukocyte MPs (5.06 ± 0.5 versus 6.25 ± 0.5 nM PS Eq, P = 0.08; Figure 3).

Serum β2 microglobulin level, a B-cell activation marker associated with extra-glandular involvement 8, was inversely correlated with level of annexin V-positive MPs (r = -0.48, P = 0.002) and platelet MPs (r = -0.37, P = 0.03; Figures 4a to 4c).

Interestingly, we found similar results for SLE patients: a significant negative correlation between level of platelet MPs and level of anti-double-stranded DNA IgG (r = -0.65, P = 0.003; Figure 4d) and a negative correlation, although not significant, between level of platelet MPs and the SLEDAI score (r = -0.46, P = 0.056). For RA patients, level of leukocyte-derived MPs and the DAS28 showed a significant negative correlation (r = -0.6, P = 0.005; Figure 4e).

Patients with AIDs receiving (n = 52) or not receiving (n = 37) immunosuppressive drugs or biological agents did not differ in level of MPs (total MPs: 8.4 ± 0.9 vs 7.1 ± 1.0, P = 0.13).

We hypothesized that because MPs contain accessible anionic phospholipids such as PS, they could be consumed by sPLA2. This enzyme is increased in level and activity in some inflammatory diseases and catalyzes hydrolysis of aminophospholipids, including PS, as well as phosphatidylcholine and phsophatidylethanolamine, all contained in microvesicles 16. Plasma sPLA2 activity was significantly higher in patients with pSS (P = 0.028), SLE (P = 0.036), and RA (P = 0.005) than in HCs (Table 2). Interestingly, the level of total MPs and activity of sPLA2 showed a significant inverse correlation for all patients with AIDs (r = -0.37, P = 0.0007 and r = -0.36, P = 0.002 for total and platelet MPs, respectively; Figure 5). Moreover, sPLA2 activity was significantly higher in the 14 pSS patients with extra-glandular involvement than in those with only glandular involvement (56.7 ± 3 versus 47.3 ± 5.2 nM/min/mL, P = 0.01). Conversely, MP level was not correlated with level of C-reactive protein, a classical marker of systemic inflammation.

As we found increased levels of platelet-derived MPs in the three studied AIDs and because platelet activation has been poorly assessed in pSS, we assessed plasma levels of sCD40L and sCD62P, two biomarkers of platelet activation (Table 2). The concentration of sCD40L was significantly higher in pSS and RA patients than in HCs and higher, but not significantly, in SLE patients than in HCs (P = 0.006, P < 0.0001 and P = 0.09, respectively). sCD62P levels were significantly higher in patients with pSS, RA and SLE than in HCs (P < 0.0001, P = 0.0003 and P < 0.0001, respectively). We found no association or correlation of level of these biomarkers with disease activity.