The clinical and functional relevance of microparticles induced by activated protein C treatment in sepsis

Circulating MPs were from patients diagnosed with severe sepsis (American College of Chest Physicians criteria) [19], who also fulfilled the National Institute of Clinical Excellence (England and Wales) criteria [20] for rhAPC (Drotrecogin alfa (activated)) (Xigris^®, Eli Lilly, Houten, Netherlands) treatment. The type of organisms isolated included pneumococcus, enterococcus, enterobacter, coagulase-negative staphylococcus and staphylococcus aureus. Patients who received a 96-hour continuous infusion of rhAPC (24 μg/kg/hr) were contrasted with an equal number of patients who were eligible but not treated because of concerns over bleeding risks. These concerns included gastrointestinal bleeding within six weeks (2), platelets < 30 × 10⁹/L (6), internal bleeding (3), intracranial pathology (1), chronic severe liver disease (4), recent major surgery (7) and trauma (2). None of these patients were on heparin prophylaxis because of bleeding concerns and all rhAPC-treated patients received low molecular weight heparin prophylaxis before and after but not during rhAPC infusion. The study protocol was approved by the Local Research Ethics Committee and the Research and Development departments of the Royal Liverpool University Hospital and Guy's and St. Thomas' Hospitals. Informed consent was obtained from patients or when patients were unable to consent, assent was sought from their next-of-kin for enrolment into the study and publication of results. Written consent was also obtained from six healthy normal donors who provided blood samples for MP isolation. A copy of the written consent is available for review by the Editor-in-Chief of this journal. Blood samples were collected into 0.105 M trisodium citrate with and without 0.1 M benzamidine. From each patient, this was six blood samples. In the rhAPC-treated group, this was before rhAPC initiation and then at 24, 48, 72 and 96 (during rhAPC infusion) and 120 hours (post-rhAPC treatment). In the non-rhAPC treated group, corresponding time points for blood sampling were also used.

CD13 (aminopeptidase N) is a trans-membrane protease present in endothelial cells and recognised as a marker for MPs arising from these cells. EPCR, the cellular receptor for protein C and activated protein C, is primarily localized on endothelial membrane and has been demonstrated to be expressed on endothelial MPs. Using fluorescence-activated cell sorting (FACS) [17], CD13 and EPCR were enumerated from a MP-specific gate based on size and calculated in relation to 3 μm latex (polystyrene) beads. MPs were isolated by initial centrifugation of plasma at 5,000 g for 10 minutes to remove cells, followed by further centrifugation of the supernatent at 18,000 g for 30 minutes twice at 4°C. EPCR and APC on MPs were measured by ELISA, as previously described [17]. In brief, MPs were captured by anti-EPCR-RCR2 and detected using chromogen S2366 or anti-EPCR RCR49-biotin for quantifying APC and EPCR, respectively. All assays were performed blinded to knowledge of rhAPC treatment or its timing. For experiments assessing functional effects, platelet-derived MPs were removed as these are not induced by APC [17] by first labelling with CD41 (platelet glycoprotein IIb)-phycoerythrin (PE) for 15 minutes and mixed with anti-PE magnetic beads (Miltenyi Biotech, Bisley, Surrey, UK) for 20 minutes on ice. Wash-through was collected for CD41-negative MPs with negativity confirmed by FACS. Clinical data were collected, including baseline characteristics and outcome at 28 days.

In addition to the clinical studies of MP number, functional assays were also performed to assess the effect of rhAPC-expressing MPs on endothelial cell function. Details of reagents and methods used to determine endothelial gene expression are available in Additional file 1. As previously reported [21, 22], a 1.5-or greater fold-change in hybridization intensity was used to denote significant regulatory change in gene expression. The effect of MPs on endothelial permeability was analysed in a dual chamber system using Evans Blue-labelled bovine serum albumin (BSA) [23]. In brief, EAhy926 cells (endothelial cell line) were plated on 6.5 mm diameter Transwell polycarbonate membranes of 3 μm pore size (Costar, Corning, NY, USA). Upper and lower chambers were filled with 100 μL and 500 μL of growth medium, respectively. Cells were grown for two days to obtain monolayers that were "sub-confluent". As described by Feistritzer [23], the sub-confluent model enables any improvement in endothelial barrier integrity to be detected; that is, reduction in leakiness by the agonist. Blocking antibodies and inhibitors were added 30 minutes before APC or MP. Cells were exposed to APC or MP for three hours and permeability assessed afterwards by adding 0.67 mg/ml Evans Blue in medium containing 4% BSA to the top chamber. After 10 minutes, an aliquot was taken from the lower chamber and diluted 1:2 in medium for measurement at 650 nm. Further studies on the influence of zymogen protein C on APC binding to MPs [24] was based on the binding assay of Regan et al.[25] (Supplementary Materials and methods).

Data were expressed as the mean and standard deviation (SD) of at least four independent experiments. For MP quantification in Figure 1A, B, data were expressed as the median with interquartile ranges (IQR). In data that were not normally distributed, Mann-Whitney U tests with Bonferroni correction were used.

In initial experiments on six rhAPC treated patients, FACS demonstrated a significant increase in CD13 and EPCR expressing MPs (P < 0.05) within 24 hours of the treatment (Table 1). Due to sample volume limitations from the critically ill patients and concordant information from CD13 and EPCR expression in the available cases, further analyses (50 patients) utilised only CD13 positivity as a marker of EPCR-expressing MPs. Table 2 highlights the characteristics of these groups along with the APACHE II scores, sepsis aetiology, vasopressor treatment, DIC scores, ITU stay and survival data, which were comparable. A trend towards lower platelet counts was, however, noted in the non-rhAPC treated group.

Figure 1A shows that the number of CD13+ MPs was higher (P = 0.1) in patients that underwent rhAPC infusion and survived (22.7 × 10⁶/ml, IQR 18.5 to 31.5) compared to rhAPC treated non-survivors (10 × 10⁶/ml, IQR 8.2 to 28.9). The level of CD13+ MPs in the non-rhAPC treated survivor group were not significantly different statistically from non-rhAPC treated non-survivor patients (P = 0.19).

The amount of APC on the generated MPs was measured by ELISA (Figure 1B). The levels were significantly higher during rhAPC treatment compared with individual, pre-treatment controls and unrelated, non-rhAPC treated patient controls (Figure 1B). In all the 25 rhAPC-treated patients, the level of APC detected in MPs was between 120 and 350 ng/ml. Daily monitoring showed no significant variation in levels. Levels were higher from the rhAPC survivors (202.6 ng/ml, IQR 194.5 to 251.7) compared to the rhAPC treated non-survivors (187.5 ng/ml, IQR 160.2 to 198.4) (P = 0.04). These data were compared with our in vitro studies which showed that the proportion of rhAPC that is bound to MPs following APC stimulation is approximately 6 to 10% (Additional file 2). To assess if APC on MPs could be due to passive binding to already circulating MPs, rhAPC was added to plasma samples from non-rhAPC-treated septic patients for 24 hours. The results showed less than 20 ng/ml APC in these cases (data not shown), indicating that findings during rhAPC treatment reflect APC bound to circulating MP-associated EPCR. Specific EPCR ELISA indicated a significant increase of EPCR levels on MPs during rhAPC treatment (160 ng/ml) in comparison to pre-treatment levels (80 ng/ml) (Figure 1C). This was also demonstrated to be not because of passive binding of sEPCR to MPs, as sEPCR levels in MP-free plasma were unchanged from pre- to rhAPC-treatment samples.

The functional aspects of MP released during rhAPC treatment were next analysed. Figure 1D shows a substantial loss in thrombin generation by these MPs compared with pre-treatment stage. Mean thrombin generation fell from approximately 294 ng/ml to 184 ng/ml during rhAPC infusion. This was reversible by PC blockade or the APC inhibitor, a₁AT. By contrast, patients not treated with rhAPC did not express this anticoagulant property (data not shown). This indicates that a functionally active circulating form of APC exists on MP-EPCR during clinical treatment with rhAPC. APC levels on MPs were observed to be similar in the absence or presence of PC. This indicates that APC on patients' MPs (n = 4) are not easily displaceable as levels in the separated supernatant remained unchanged even at very high PC concentrations (Figure 2).

Since APC on MP-EPCR in plasma retain anticoagulant activity, we postulated that they could modulate genes involved in apoptosis and inflammation. As the majority of circulating MPs are platelet-derived and platelets do not express EPCR [26], these were depleted to obtain an EPCR-enriched MP pool. The efficiency was 95 to 98% depletion and the vast majority of these CD41-depleted MPs were CD13 and EPCR positive (data not shown). When these MPs (approximately 0.15 × 10⁶) were tested on gene arrays, those isolated during rhAPC treatment induced the expression of anti-apoptotic genes including A20, Bcl-x and the vascular endothelial growth factor (VEGF) receptor 2/kinase insert domain-containing receptor (KDR) (Figure 3). In most cases, PC blockade reversed these specific expressions (Additional file 3). Exceptions to the expression of several anti-apoptotic genes were matrix metalloproteinase (MMP)-2, and MMP-1 to a lesser extent (PC blockade remained above the 1.5 cut-off level for significant regulatory change). Most transcript effects of APC were reversible by PAR1 antagonism with the exception of KDR. This could be because of PAR1-independent APC effects or insufficient PAR1 inhibition under these conditions.

rhAPC-induced MPs could also inhibit apoptosis to a similar level as free-APC through a PAR-1 dependent mechanism (Figure 4). This cytoprotective property can be shown to be absent with PAR1 antagonism. Panel 8 (in Figure 4) shows that PC blockade did not, however, completely abrogate cytoprotective signals in all cells, suggesting that other cytoprotective mechanisms may exist in the context of rhAPC treatment. Another plausible explanation is that the MPs used were not 100% EPCR and APC positive. Nonetheless, these results indicate that circulating APC on MP-EPCR from rhAPC-treated patients have anti-apoptotic properties that are comparable to the free APC effect. Furthermore, this was not observed with MPs from non-rhAPC treated samples.

To further investigate a role for APC-MP on inflammation, the effect of free or MP-associated APC on endothelial permeability was examined. Using MPs from healthy donors as controls, we compared circulating APC-MP from three patients before and during rhAPC infusion. The same number of circulating MPs from healthy individuals or patients were applied; that is, 1.4 × 10⁵ MPs per well. Figure 5 shows that circulating MPs from the rhAPC treatment period reduced permeability by approximately 35 to 40%, similar to free APC. This protective effect was again APC and PAR1 dependent. Notably, MPs taken from the patients before rhAPC treatment or those taken from healthy individuals had a minor protective effect that did not reach statistical significance. This was also neither APC- nor PAR1-dependent.