Sepsis is characterized by a robust cellular response to invading organisms. In the case of Gram-negative bacteria it is mostly governed by lipopolysaccharide (LPS), the main outer surface component, which is a strong stimulator of the immune system . Patients with sepsis frequently show severe thrombocytopenia because of sequestration of activated platelets in the microvasculature [1, 2]. In vitro studies [2-8] extensively examined platelet activation induced by a wide range of concentrations of LPS from Escherichia coli strains under rather variable conditions. Surprisingly, the platelet responses gave controversial data, and it was also debated whether LPS directly modulates platelet activation or only in a leukocyte-dependent manner (reviewed in ). Various toll-like receptors (TLR 1, 2, 4, 6, 8 and 9) were detected on the platelet surface, but TLR4 was the main receptor for direct interactions of LPS with platelets [2, 3, 6, 9].
Certain clinically relevant Gram-negative bacteria produce a highly heterogeneous mixture of total LPS with varying numbers of repeating polysaccharides . It contains diverse proportions of S-LPS, which comprises lipid A moiety, core-oligosaccharide, and O-polysaccharides, and the rough form of LPS (Re-LPS) lacking O-specific chains. Smooth LPS (S-LPS) from E. coli showed discrepancies in activating platelets [4-8], while Re-LPS from Salmonella minnesota (Re595) stimulated TLR4/MD2 expressing mast cells , and human monocytes and neutrophils with a potency similar to its parent S-LPS . Thus, we studied for the first time whether Re-LPS (Re595) from S. minnesota alters the levels of various platelet activation markers in platelet-rich plasma (PRP) samples, and compared these data with experiments with the same type of S-LPS from E. coli (O111:B4) used earlier [4-8].
First, to evaluate the biological activity of these LPS forms, we tested their procoagulant properties in a recalcified, LPS-stimulated mononuclear cell suspension. Re-LPS and S-LPS (10 μg mL−1) significantly (P < 0.05) decreased the mean clotting time (56.0 s and 60.8 s, respectively) compared with the non-activated control (111.7 s) (Fig. 1A). LPS interacted directly with platelets as FITC-labeled Re-LPS binding resulted in significantly increased FL-1 mean fluorescence intensity (MFI) of the positive cells (MFI, 49.7 ± 8.3; P = 0.01) compared with control platelets with unlabeled Re-LPS (10.2 ± 2.4). An even larger increase in MFI (112.4 ± 31) was observed with TRAP (10 μm), as similarly shown  where CD62P was also implicated in LPS binding (Fig. 1B). This ‘valid’ fluorescence was abolished by a 20-fold molar excess of unlabeled Re-LPS (Fig. S1A), and considerably decreased by anti-TLR4 and anti-P-selectin antibodies (Fig. S1B), in agreement with previous results .
Next, we examined whether Re-LPS and S-LPS affected platelet aggregation (Fig. 1C-D). Neither LPS form alone (0.1–10 μg mL−1) induced platelet aggregation, even with pretreatment with LPS for 10–60 min, or when only added to the PRP sample prior to recording . In contrast, 1 μg mL−1 of Re-LPS but not S-LPS augmented submaximal TRAP-induced (5 μm) platelet aggregation. The expression of CD62P and CD40L as sensitive activation markers  was also analyzed after activation with up to 10 μg mL−1 of Re-LPS and S-LPS in parallel. No increase in the level of P-selectin-positive platelets was induced by either LPS form compared with the untreated sample (9.0 ± 4.0%, 9.5 ± 5.0% vs. 7.6 ± 3.0%; data not shown). On the other hand, a statistically significant elevation in CD40L expression (2.7 ± 1.9%; P < 0.05) with high MFI values (172.5 ± 80.0; P < 0.001) was detected but only at 10 μg mL−1 of Re-LPS vs. the negative control (1.3 ± 0.2%; MFI = 21.0 ± 8.0). In addition, TRAP also caused a substantial increase in CD40L expression (33.6 ± 4.9%; MFI = 33.6 ± 4.7). Interestingly, S-LPS did not raise the level of surface CD40L (1.4 ± 0.43%; MFI = 22.0 ± 3.0), in contrast to earlier reports [6, 8] (Fig. 1E).
We found a significant increase (P < 0.05) in platelet-derived microparticle (PMP) levels induced by Re-LPS (10 μg mL−1) (450 ± 170 PMPs μL−1 plasma); however, S-LPS did not alter the PMP number (273 ± 144 PMPs μL−1 plasma vs. 227 ± 95 PMPs μL−1 plasma in unstimulated sample). These levels were below those we observed in TRAP-treated (5 μm) samples (970 ± 145 PMPs μL−1 plasma) (Fig. 1F). We also wondered if Re-LPS at even lower concentrations could potentiate PMP generation in the presence of TRAP by synergism. PMP levels were significantly enhanced already at 1 μg mL−1 of Re-LPS during co-activation (1384 ± 299 PMPs μL−1 plasma; P < 0.05 compared with the sample with TRAP alone). Of note, as concluded before , S-LPS had neither synergistic nor inhibitory effects on the PMP level under the same conditions (983 ± 120 PMPs μL−1 plasma). Finally, S-LPS at a high concentration was unable to prevent the effects of Re-LPS (Fig. S1C). This suggests that these forms of LPS interact differently with platelets.
It was previously established that, in contrast to S-LPS that needs CD14 to bind to TLR4 for its functions , Re-LPS interacts with TLR4 without CD14 [10, 13]. As platelets do not express CD14 , we checked whether soluble CD14 is required for Re-LPS to influence platelet activation. Gel-filtered platelets with undetectable levels of plasma proteins were activated with added fibrinogen (2 g L−1) by Re-LPS (1 μg mL−1) with TRAP (5 μm) in the absence or presence of LPS binding protein (LBP) and soluble CD14 (data not shown). Platelet aggregation was not further enhanced when platelets were also pretreated with LBP and soluble CD14 at Re-LPS plus TRAP stimulation. Thus, Re-LPS did not require soluble CD14 to activate platelets. In summary, although both LPS forms are biologically active, Re-LPS, but not S-LPS, directly modulates platelet activation.