Toll-Like Receptor Signalling Is Not Involved in Platelet Response to Streptococcus pneumoniae In Vitro or In Vivo

Streptococcus (S.) pneumoniae strains vary considerably in their ability to cause invasive disease in humans, which is at least in part determined by the capsular serotype. Platelets have been implicated as sentinel cells in the circulation for host defence. One of their utensils for this function is the expression of Toll-like receptors (TLRs). We here aimed to investigate platelet response to S. pneumoniae and a role for TLRs herein. Platelets were stimulated using four serotypes of S. pneumonia including an unencapsulated mutant strain. In vitro aggregation and flow cytometry assays were performed using blood of healthy volunteers, or blood of TLR knock out and WT mice. For in vivo pneumonia experiments, platelet specific Myd88 knockout (Plt-Myd88-/-) mice were used. We found that platelet aggregation was induced by unencapsulated S. pneumoniae only. Whole blood incubation with all S. pneumoniae serotypes tested resulted in platelet degranulation and platelet-leukocyte complex formation. Platelet activation was TLR independent, as responses were not inhibited by TLR blocking antibodies, not induced by TLR agonists and were equally induced in wild-type and Tlr2-/-, Tlr4-/-, Tlr2/4-/-, Tlr9-/- and Myd88-/- blood. Plt-Myd88-/- and control mice displayed no differences in bacterial clearance or immune response to pneumonia by unencapsulated S. pneumoniae. In conclusion, S. pneumoniae activates platelets through a TLR-independent mechanism that is impeded by the bacterial capsule. Additionally, platelet MyD88-dependent TLR signalling is not involved in host defence to unencapsulated S. pneumoniae in vivo.


Validation of anti-TLR2 antibodies
Anti-TLR2 antibodies T2.5, TLR2.45, TL2.1 (kindly provided by HBT, Uden, The Netherlands) were tested for their ability to inhibit TLR2 function by 30 minutes pre-incubation of the antibodies in heparinized human whole blood and stimulation with 300ng/mL of TLR2 ligand PAM3CSK4 added by an equal of the ligand in RPMI1640 medium supplemented with 0.1% human albumin and overnight incubation at 37°C and determination of released TNFα using ELISA (BD Biosciences Pharmingen (San Diego, CA).
Mice were housed in a the animal facility of the Academic Medical Center with a 12 hour day-night rhythm, food and water ad libitum, temperature and moisture control, and daily checks. Upon arrival in the facility, mice were acclimatized for at least 7 days before use in experiments. Mice were euthanized by cervical dislocation after anesthesia with (0.12mg/g body weight) ketamine and (0.3ug/g body weight) dexmedetomidine intraperitoneally. Mice were monitored minimally once daily during experiments. Human endpoint for infection experiments was if mice were segregated from the others and unable to lift themselves from supine position. No mice reached the human endpoint before the experimental endpoint. The Institutional Animal Care and Use Committee of the Academic Medical Center approved all experiments (Permit Number DIX21BR and DIX101643).

Experimental study design
Pneumonia was induced by intranasal inoculation with ΔcpsD39 (2 x 10 7 colony forming units (CFU) in 50 μL isotonic saline) using previously described methods [21,29]. Mice were euthanized 16 hours after induction of pneumonia (N = 7/9 mice per group). Bacterial quantification and storage of organs were performed as described [21,29], platelet counts and activation (by expression of P-selectin as described above) were determined in citrated whole blood by flow cytometry. Mouse tumour necrosis factor (TNF-)α, interleukin (IL-)6, IL-1β, keratinocyte chemoattractant (KC), PF4, soluble (s)P-selectin, E-selectin (R&D Systems) and thrombin-antithrombin complexes (TATc; Bio-connect, Huissen, the Netherlands) were measured by ELISA. Four-micrometer sections of the left lung lobe, spleen and liver were stained with hematoxylin and eosin (H&E). To make sure sections were representative of the entire lung, sections were first carefully cut into the middle part of the fixated lung and assessed by a blinded pathologist before scoring. Slides were coded and scored by a pathologist blinded for group identity for the following parameters: infiltrative surface (expressed as the percentage of total lung surface), bleeding, infiltration, interstitial inflammation, endothelialitis, bronchitis, oedema, pleuritis and presence of thrombi. All parameters were rated separately from 0 (condition absent) to 4 (most severe condition). The total histopathological score was expressed as the sum of the scores of the individual parameters.

Statistical analysis
All analyses were done using GraphPad Prism version 5.01 (GraphPad Software, San Diego, CA). Comparisons between groups (8 mice per group) were tested using the Mann-Whitney U test as data was non-parametric." P-values < 0.05 were considered statistically significant.

Results
Unencapsulated, but not encapsulated, S. pneumoniae induces platelet aggregation Unencapsultated ΔcpsD39 S. pneumonia serotype 2 induced platelet aggregation in human platelet rich plasma consistent with the observations of Keane et al [30] that unencapsulated S. pneumoniae causes platelets to aggregate. The encapsulated D39 serotype 2, as well as S. pneumoniae serotype, 3 and 4 (6303 and TIGR4 respectively) failed to induce platelet aggregation ( Fig 1A). The finding that capsulated S. pneumoniae did not induce aggregation was consistent with the lack of aggregation in the presence of a recombinant preparation of S. pneumoniae capsule (CPS2) (Fig 1B). Platelet aggregation by ΔcpsD39 was activation dependent, required fibrinogen binding to GPIIbIIIa and FcγRII occupation as it could be inhibited by PGE1, the GPIIbIIIa antagonist Abciximab and the FcγRII antagonist AT10 (Fig 1C). In this respect the platelet aggregation induced by ΔcpsD39 is in perfect agreement with previous reports [30,31].
Platelets express Toll like receptors 2 and 4, which have been previously described to be functional [19,30,[32][33][34]. Using Flow cytometry, we could also detect Toll like receptor 2 and 4 on human platelets ( Fig 1D). However, blocking of TLR2 did not affect aggregation ( Fig  1C) which is in contrast to the TLR2-dependent S. pneumoniae-induced platelet activation described by Keane et al [30]. Moreover, stimulation with the purified TLR2 agonists Pam3CSK4 and LTA and the TLR4 agonist LPS failed to induce any response even at high concentrations (5 μg/mL; Fig 1E). We confirmed the capacity of the used TLR2 antibody T2.5 to inhibit TLR2 responses in other assays. First we showed that T2.5 is a superior TLR2 blocking antibody compared to other TLR2 antibodies in a whole blood assay (S1 Fig). Additionally, TLR2 activation by S. pneumoniae ΔcpsD39 is inhibited by T2.5 ( Fig 1F). These results indicate that S. pneumoniae may aggregate platelets in a TLR2 independent manner. Prestimulation with S. pneumoniae fails to modulate platelet aggregation to subthreshold concentrations of TRAP Previous studies have described a role for LPS in platelet 'priming', where LPS pretreatment induced platelet hypersensitivity to subthreshold concentrations of classical platelet agonists (24;25). However, we failed to observe any priming effect of pre-incubation of platelets with either S. pneumoniae or TLR agonists before stimulation with subthreshold concentration TRAP (Fig 2A and 2B). The priming effect of LPS described by Montrucchio (24) was monocyte-dependent; we therefore repeated these experiments in the presence of isolated PBMC's. Still, no platelet hypersensitivity to subthreshold TRAP was found ( S2 Fig). S. pneumoniae D39, ΔcpsD39, TIGR4 and 6303 induce platelet degranulation Platelet activation by different agonists can induce a variety of responses. We therefore focused on platelet granule release. Alpha granule degranulation was detected by CD62p (P-selectin) surface expression and dense granule release was detected based on surface expression of CD63 [35]. Whole blood stimulation by S. pneumoniae D39, 6303, TIGR4 and ΔcpsD39 all resulted in platelet CD62p and CD63 exposure, ΔcpsD39 being the most potent activator (Fig 3).
S. pneumoniae did not activate platelets via TLR2 or 4, as pre-incubation with α-TLR2 and α-TLR4 did not inhibit CD62p expression by S. pneumoniae (Fig 4A). Opposed to aggregation, FcγRII and GPIIbIIIa inhibition did not block CD62p expression by S. pneumoniae D39 or ΔcpsD39, but PGE1 did (Fig 4A). Platelet surface expression of CD62p or CD63 was not induced by direct TLR agonists LTA, Pam3CSK4 or LPS (Fig 4B and 4C).

Whole blood S. pneumoniae incubation results in platelet-leukocyte complex formation
To determine whether S. pneumoniae whole blood stimulation results in formation of plateletleukocyte complexes, platelet markers CD61 and CD62p were measured on neutrophils, monocytes and lymphocytes (shown for CD61 in Fig 5). All S. pneumoniae strains tested induced some platelet-neutrophil complexes; ΔcpsD39 being the most potent ( Fig 5A). Platelet-monocyte complex formation occurred readily upon stimulation with all S. pneumoniae serotypes tested (Fig 5B), platelet-lymphocyte complexes were not induced (Fig 5C). TLR2 and TLR4 were not directly involved in platelet-leukocyte complex formation as it was not induced by the TLR agonists LTA, Pam3CSK4 or LPS (shown for neutrophils, monocytes and lymphocytes in Fig 5D-5F).
Wild-type mouse platelets respond to S. pneumoniae D39 and ΔcpsD39 in a similar manner as platelets from Tlr2 -/-, Tlr4 -/-, Tlr2/4 -/-, Tlr9 -/and Myd88 -/mice In order to test the contribution of TLR2 and 4 signalling in platelet responses to S. pneumoniae without the use of antibodies or synthetic agonists, we conducted similar whole blood stimulation experiments in mouse blood comparing wild-type platelets with platelets of Tlr2 -/-, Tlr4 -/and Tlr2/4 -/strains using CD62p expression as readout for platelet activation. Recently, a functional role for platelet TLR9 was described [18]. We therefore included Tlr9 -/mouse blood to investigate a possible role for TLR9 in this model. As a final control, we performed the stimulation experiments with blood obtained from Myd88 -/mice, blocking downstream signalling of all TLR receptors except for TLR3 [17,36]. Tlr2 -/-, Tlr4 -/-, Tlr2/4 -/-, Tlr9 -/and Myd88 -/platelets all showed enhanced CD62p expression to a similar extent as wild-type platelets upon stimulation with S. pneumoniae D39 or ΔcpsD39, implicating that there is no role for TLR signalling in direct platelet response to S. pneumoniae (Fig 6A  and 6B). Platelet MyD88 is not involved in host defence and response to ΔcpsD39 in vivo It is known that platelets especially exert proinflammatory and immune modulatory effects in the lungs [37]. To determine the impact of platelet specific TLR signalling during pneumonia in vivo, Plt-Myd88 -/and littermate control mice were inoculated with 2 x 10 7 CFU ΔcpsD39 via the airways. We chose to conduct these experiments with ΔcpsD39 which is cleared in an almost completely MyD88 dependent manner [38], and the strain that was the most potent inducer of platelet activation and platelet-leukocyte formation in our in vitro experiments. As Percentages were determined using isotype control antibodies to set the gate. TRAP was used as a positive control and induced CD62p-and CD63 expression on 87% and 56% of platelets respectively; PBS induced CD62p-and CD63 expression on 10% and 2% of platelets. Histograms are representative of 2 independent experiments using different donors. control mice clear this unencapsulated S. pneumoniae strain within 24 hours, we therefore sacrificed the mice after 16 hours when bacterial loads are still present. No differences were detected in bacterial burdens in the lungs, blood, spleen or liver between control and Plt-Myd88 -/mice ( Fig 7A). Additionally, no differences were found in platelet counts (Fig 7B) or platelet activation measured by platelet surface CD62p (P-selectin) expression, PF4 and platelet and endothelial cell activation marker sP-selectin (Fig 7C-7E). (Activated) platelets are considered to play an essential role in coagulation by providing a phospholipid surface for the assembly of activated clotting factors [39]. To obtain insight in the role of MyD88 dependent platelet signalling in systemic coagulation activation during ΔcpsD39 pneumonia, we measured TATc levels in plasma of infected Plt-Myd88 -/and control mice. No differences were detected between the groups (Fig 7F). Lastly, platelet MyD88 signalling had no influence on endothelial cell activation during ΔcpsD39 pneumonia as E-selectin levels did not significantly differ (Fig 7G).
Platelets secrete inflammatory mediators upon activation like Platelet Factor 4 and RANTES [6] and platelets in complex with leukocytes can influence leukocyte effector function [4]. During ΔcpsD39 pneumonia however, cytokine production in the lungs did not differ between control and Plt-Myd88 -/mice (Fig 8A-8D); plasma cytokine levels were below detection. Platelets have been both associated with enhanced histopathological damage during inflammatory challenges [40], and the protection of vascular integrity during inflammation [14,15,41]. However, no differences for inflammation parameters or infiltrated lung surface were found between Plt-Myd88 -/and control mice, as reflected by the semi-quantitative scores of lung histopathology slides (Fig 8E and 8F). Additionally, no bleeding was found in the lungs of either Plt-Myd88 -/or control mice.

Discussion
S. pneumoniae represents a major health burden worldwide [42]. Recently, platelets have been implicated as major players in infection and immunity [6] and we have specifically shown this for S. pneumoniae in vivo [14]. Platelets are activated during sepsis, directly by an invading pathogen or indirect via injured endothelium and host coagulation activation [4]. In this paper we demonstrate that S. pneumoniae directly activates platelets in a TLR independent fashion. Human whole blood S. pneumoniae incubation results in platelet-leukocyte complex formation. Whole blood was stimulated with S. pneumoniae D39, ΔcpsD39, TIGR4 or 6303. Following 30 minutes of incubation leukocytes subsets were stained and analysed for surface expression of CD61. Percentages were determined using isotype control antibodies to set the gate. Neutrophil-CD61 is depicted in (A), monocyte-CD61 in (B) and lymphocyte-CD61 in (C). TRAP was used as a positive control and PBS as a negative control. Histograms are representative of 2 independent experiments using different donors. In a similar fashion, neutrophil-platelet (D) and monocyte-platelet (E) and lymphocyte-platelet (F) complex formation was analysed following stimulation with TLR agonists LTA, Pam3CSK4 and LPS. Platelet activation by all serotypes tested resulted in surface expression of CD62p and CD63 and platelet-leukocyte complex formation; ΔcpsD39 additionally induced platelet aggregation. In accordance, Plt-Myd88 -/mice were unaffected during ΔcpsD39 pneumonia.
The pneumococcal capsule inhibits mucosal clearance, facilitates binding to the epithelial surface and inhibits complement-and phagocyte-mediated immunity [1]. Besides reduction of exposure to several antibodies, capsular polysaccharide was suggested to prevent interaction between Fcγ receptors to the Fc component of IgG bound to pneumococci [1,43]. This could be why the only pneumococcal strain capable of inducing (FcyRII dependent) platelet  aggregation was unencapsulated ΔcpsD39. Our results are in conflict with an earlier report showing that both encapsulated and unencapsulated S. pneumoniae induced platelet aggregation via TLR2 mediated signalling. The strains we tested however did not induce platelet aggregation unless in its mutated unencapsulated form (ΔcpsD39). ΔcpsD39 did not induce aggregation in a TLR2 dependent manner, as we could not inhibit the reaction by adding TLR2 blocking antibodies. In addition, direct TLR2 stimulation by TLR2 agonists LTA and Pam3CSK4 failed to induce platelet activation.
In contrast to the results found on platelet aggregation, we found that all strains of S. pneumoniae can induce platelet degranulation and complex formation and that the pneumococcal capsule only partly reduced this. In line with platelet aggregation, this was TLR independent, as blocking TLR-antibodies did not inhibit this and platelets from WT and Tlr2 -/-, Tlr4 -/-, Tlr2/ 4 -/-, Tlr9 -/and Myd88 -/mice showed similar results. In contrast to platelet aggregation, this was not FcyRII dependent, as blocking FcγRII antibodies had no effect and mice (which lack FcγRII [44]) also show platelet degranulation and complex formation. It seems other (FcγRII independent) mechanisms are involved in platelet degranulation and complex formation, possibly GpIb, PECAM-1 or complement receptors [45][46][47].
Our results differ from a previously published paper by Keane et al., which found that encapsulated D39 (serotype 2) and TIGR4 (serotype 4) could induce platelet aggregation. Moreover, they found that anti-TLR2-antibodies could inhibit aggregation, whereas we found no role for TLR2 in interactions between platelets and S. pneumoniae. Several other papers have also reported functional roles for platelet TLRs in vivo [13,19,[48][49][50], however controversy still surrounds the functionality of these receptors in in vitro assays. Stimulation of platelets with TLR2 or 4 ligands sometimes did [30,51,52] or did not [13,34,53] induce aggregation, did [34,51] or did not [13,19,33] induce CD62p (P-selectin) expression, did [52,54] or did not [33] induce Ca 2+ mobilisation or thrombin generation [55]. We were unable to induce platelet activation by direct TLR2 agonists LTA and Pam3CSK4 or TLR4 agonist LPS, in a variety of functional assays. Possible differences between previous studies and ours (as well as differences between previous papers) remain difficult to clarify, but could well encompass technical issues such as culture method, amount and species of bacteria, quality of antibodies, PRP spinning protocols or platelet isolation methods, presence of plasma or different aggregometers.
Opposed to direct activation two groups found a priming effect on platelets of LPS alone [34], or in co-incubation with monocytes [53], whereafter platelets were 'hyperexcitable' and aggregated by addition of subthreshold levels of classic platelet agonists. Nevertheless, TLR agonists or encapsulated S. pneumoniae strains did not modulate the platelet response to subthreshold concentrations of TRAP in our hands.
The results presented have been generated in both murine and human blood. Although there are great similarities between mice and humans [56] differences obviously exist [57]. Therefore, caution must be taken when extrapolating results generated in mice. In the present study, in vitro data in human and murine blood showed similar results, with respect to the lack of involvement of TLR2 in activation of platelets by S. pneumoniae. Moreover, a recent study showed similar effects of platelets on host response in human sepsis patients as previously found in mice [58].
In vitro human experiments were performed using different donors. It has previously been reported that gender [59] and polymorphisms [60] can influence TLR expression and function. We observed similar results in 3 donors, but we cannot exclude effects of polymorphisms in this setting.
Platelets have been shown to be important in the host defence to S. pneumoniae pneumonia [14] and the unencapsulated serotype 2 strain D39 (ΔcpsD39) is cleared in a MyD88 dependent manner [38]. MyD88 dependent TLR signalling in platelets is not involved herein, as bacterial clearance was similar in Plt-Myd88 -/and control mice during ΔcpsD39 pneumonia. In a gram negative pneumosepsis model using K. pneumoniae, we also observed no or minor contribution of platelet MyD88 dependent signalling. [61]. Moreover, platelet MyD88 deletion had no influence on platelet counts, platelet activation or coagulation activation.
Platelet TLR4 has been reported to modulate TNF-α production to bacterial lipopolysaccharide (LPS) [48]. Platelet activation during infection could additionally influence cytokine levels by release of cytokines from their own granules or by influencing leukocyte effector function [4,6]. TNF-α and other cytokine levels were however similar in lungs of Plt-Myd88 -/and control mice in our experiments. While platelets are additionally known to regulate lung architectural changes and vascular integrity during inflammation [40,41], platelet activation via MyD88 dependent TLR signalling seems not involved as we found no histopathological differences between the groups in our pneumonia model. Differences with previous findings and the current could be explained by differences in type of bacteria used (gram positive or negative), dosis and model (inflammation vs. infection experiments).
The described activation patterns provide additional evidence that platelets function as circulatory sentinel cells in our immune system to detect and battle S. pneumonia as reported [62]. However, in this work we also show that S. pneumoniae apparently activates platelets by a mechanism that is independent of TLR signalling in platelets.