TLR8 and complement C5 induce cytokine release and thrombin activation in human whole blood challenged with Gram‐positive bacteria

We recently showed that TLR8 is critical for the detection of Gram‐positive bacteria by human monocytes. Here, we hypothesized that TLR8 and complement together regulate antibacterial responses in human blood. Anticoagulated blood was treated with selective inhibitors of TLR8 and/or complement C5, and then challenged with live Streptococcus agalactiae (Group B streptococcus, GBS), Staphylococcus aureus, or Escherichia coli. Cytokine production, plasma membrane permeability, bacterial survival, phagocytosis, and activation of coagulation was examined. GBS and S. aureus, but not E. coli, triggered TLR8‐dependent production of IL‐12p70, IL‐1β, TNF, and IL‐6 in fresh human whole blood. In purified polymorphonuclear neutrophils (PMN), GBS and S. aureus induced IL‐8 release in part via TLR8, whereas PMN plasma membrane leakage and extracellular DNA levels increased independently of TLR8. TLR8 was more important than C5 for bacteria‐induced production of IL‐12p70, IL‐1β, and TNF in blood, whereas IL‐8 release was more C5 dependent. Both TLR8 and C5 induced IL‐6 release and activation of prothrombin cleavage, and here their combined effects were additive. Blocking of C5 or C5aR1 attenuated phagocytosis and increased the extracellular growth of GBS in blood, whereas TLR8 inhibition neither reduced phagocytosis nor intracellular killing of GBS and S. aureus. In conclusion, TLR8 is more important than C5 for production of IL‐12p70, IL‐1β, and TNF upon GBS and S. aureus infection in blood, whereas C5 is central for IL‐8 release and phagocytosis. Both TLR8 and C5 mediate IL‐6 release and activation of coagulation during challenge with Gram‐positive bacteria in blood.


INTRODUCTION
Sepsis is defined as a life-threatening organ dysfunction caused by a dysregulated host response to infection. 1 Despite much effort, no specific therapy has been developed. 2,3 When the host loses local role in the early stages of sepsis pathophysiology, and leads both to hyperinflammation and immunosuppressive counter-reactions. 6 It is therefore vital to understand the mechanisms of how pathogens are recognized and how the innate defense system is regulated in order to identify new targets for treatment of sepsis.
The TLR family and the complement system are key sensor and effector systems of innate immunity. The TLRs recognize microbial components and elicit potent proinflammatory responses when activated. TLR8 function as an endosomal sensor of the ssRNA degradation products uridine and short ssRNA oligomers, which bind cooperatively to the preformed TLR8 dimer and activate signaling. 7 We and others have earlier shown that TLR8 contributes to sensing of bacteria such as GBS, S. aureus, S. pyogenes, and E. coli. [8][9][10][11] Using a selective small-molecule inhibitor of TLR8, 12 we recently showed that TLR8 is a dominant sensor of the Gram-positive bacteria GBS, S. aureus, and S. pneumoniae in monocytes, whereas it appears considerably less important for the detection of Gram-negative species. 13 The complement system is a part of the innate defense system were it facilitates phagocytosis via C3b opsonization, induces inflammation via anaphylatoxins such as C3a and C5a, and lyses sensitive bacteria via terminal complement complex formation. 14 Systemic activation of complement can be harmful, triggering excessive inflammation and organ damage in sepsis. 14,15 The C5a-C5aR1 signaling axis can be central in such processes. 16 Blocking complement together with CD14, a coreceptor for TLR2 and TLR4, efficiently attenuates the cytokine induction by E. coli in human blood. 17,18 Combined C5-and CD14-blockade can be protective during experimental sepsis in large mammals, [19][20][21] and attenuates inflammation in human whole blood challenged with S. aureus. 22 Further, the complement system interacts with the clotting mechanism in several ways, such as the C5a-mediated induction of tissue factor (TF) expression and activity on several cell types, triggering the intrinsic clotting pathway. 14 Sepsis-induced disseminated intravascular coagulation (DIC) is primarily TF driven, 23 and in the human blood model where the thrombin inhibitor lepirudin is used, S. aureus activates prothrombin cleavage mainly via C5a-induced TF activity on monocytes. 24 In this infection model all leukocytes, complement factors, and parts of the coagulation system are present and functional under normal physiologic conditions. Therefore, it mimics normal blood stream infections, and is more relevant than simpler cell monoculture infection models. We here examined the effects of selective TLR8 inhibition in whole blood, alone and in combination with blockade of C5, during bacterial infection.

Materials
The TLR8 antagonists CU-CPT9a and CU-CPT9b are previously described, 12

Bacteria
The

Whole blood model and cytokine analysis
The whole blood model was performed as described previously. 27  the inhibitors a few min prior to the addition of ligands or bacteria. The TLR8 ligand pU was premixed with poly-L-arginine to allow complex formation before addition to blood. After incubation for 120-240 min, aliquots were sampled and EDTA was added. Plasma was isolated by centrifugation of blood at 1700 × g for 15 min, and stored at −20 • C until analyses. Cytokine analysis was done using Bio-Plex assays and analysis on a Bio-Plex TM 200 System using a Bio-Plex Pro TM Wash Station, as described by the manufacturer (Bio-Rad).

Leukocyte viability
To

Neutrophil isolation and infection
Human polymorphonuclear neutrophils (granulocytes; PMN) were obtained from fresh peripheral blood drawn by venipuncture as described. 28

Bacterial survival
Anticoagulated blood was incubated with CU-CPT9b or DMSO control for 90-120 min. Subsequently, live bacteria were added and mixed by pipetting. Immediately (T0) and after 240 min of incubation on a tube roller at 37 • C, 20 µl blood was lysed by dilution into 180 µl ice cold sterile water with 0.3% saponin. The lysed blood samples were further diluted in a 10-fold series, and 30 µl of each dilution was plated in triplicates on blood agar. After incubation at 37 • C over night, the number of CFU were counted and the corresponding CFU concentration blood was calculated. Similar experiments were performed using lepirudin plasma.
The number of intracellular viable bacteria in WBC was determined by a previously described method, 29 using some modifications. After bacterial challenge for 240 min, 1 ml blood was diluted in 9 ml of RBC lysis buffer using 15 ml tubes (Nunc) at RT for 10 min. The lysed blood was centrifuged at 300 × g for 10 min at 4 • C, and the plasma/lysis phase containing extracellular bacteria was carefully removed. The serially diluted, and plated for CFU quantification, as described above.  test. Significant differences between the control/vehicle-and the treatment-conditions are indicated with the symbol "*", whereas horizontal bars indicate comparisons of specific treatments. Significance levels correspond to *P < 0.05, **P < 0.01, and ***P < 0.001.

Safety and ethics
The

Efficacy of the TLR8 inhibitor CU-CPT9b in blood
To determine the minimum concentration of CU-CPT9b required to block TLR8 function, anticoagulated blood was pretreated with the inhibitor or the control compound at different concentrations for 120 min, and subsequently the TLR8 agonists pU or CL075, the TLR4 agonist LPS, or vehicle control (PBS) were added. Cytokine analyses in plasma after 240 min of treatment revealed that the TLR8 inhibitor efficiently blocked the CL075 induced production of TNF, IL-1 , and IL-6 at the lowest dose tested (2.5 µM), whereas the pU induced response was fully blocked at 5-10 µM (Supplemental Fig. 1A). IL-8 showed a similar tendency but was weakly induced. The LPS-induced cytokine release was not affected by CU-CPT9b, whereas the control compound (Supplemental Fig. 1B) did not influence the TLR8-induced responses (Supplemental Fig. 1A). Hence, we considered 5 µM CU-CPT9b to be optimal for inhibition of TLR8 activation in blood, which is similar to our recent findings with the analogue CU-CPT9a in cultures of purified monocytes. 13

TLR8-inhibition attenuates cytokine induction by GBS and S. aureus in whole blood
To determine the impact of TLR8 for the sensing of viable bacteria in blood, we pretreated blood with 5 µM of CU-CPT9b or the control reagent for 120 min, and subsequently added pU, LPS, FSL-1 (TLR2/6 agonist), or live GBS, S. aureus or E. coli at different doses. Blood plasma was separated 240 min after the initiation of the challenge and cytokine levels were examined. As expected, TLR8 blockade almost eliminated the cytokine induction by pU, whereas it had no effect on the cytokine induction by LPS or FSL-1 (Fig. 1). More importantly, CU-CPT9b clearly reduced the production of cytokines after GBS and S. aureus challenge. The effect was most pronounced for IL-12p70, IL-1 , and TNF, but also significant for IL-6 for the lower bacterial concentrations. There were no major effects of CU-CPT9b treatment on the release of IL-8 or MCP-1 (data not shown). Moreover, cytokine induction by E. coli was mainly TLR8 independent, which is in agreement with our recent findings with this strain in monocyte cultures. 13 We also included 2 clinical isolates of E. coli that to some extent activate TLR8 in monocytes, but no clear effect of CU-CPT9b on cytokine induction in blood was found (data not shown). Over all, cytokine induction by live GBS and S. aureus challenge of human blood is partially TLR8 dependent, whereas E. coli is sensed mainly in a TLR8-independent manner.

Monocyte and PMN viability and release of DNA and IL-8 by PMN
The integrity of the plasma membrane of monocytes and PMN during GBS and S. aureus challenge of blood for 240 min was examined by PI staining and flow cytometry analysis. The membrane integrity of PMN, but not monocytes, was compromised during challenge with the highest bacterial load, but TLR8 inhibition did not influence this ( Fig. 2A). To further examine whether TLR8 has a functional role in other forms of necrotic cell death. 30 Anyway, both effects occurred independently of TLR8 signaling. In contrast, IL-8 production by PMN in response to both bacteria was partially reduced by TLR8 inhibition, most strongly during GBS challenge (Fig. 2B). This suggests that TLR8 directly contribute to the sensing of bacteria in PMN. Since C5 blocking does not interfere with C3b-dependent opsonization, this seems to be a better and more selective target than C3.

Effects of combined TLR8 and C5 inhibition in blood upon GBS and S. aureus challenge
In agreement with our hypothesis, the cytokine induction by the Grampositive bacteria in blood is both TLR8 and complement dependent.
Next, we examined if combined inhibition of TLR8 and C5 could further attenuate the cytokine responses. E. coli was excluded from this experiment, since it mainly triggered CD14-dependent and TLR8independent cytokine production. CU-CPT9b markedly decreased the induction of IL-1 and TNF during 120 min of challenge with GBS and S. aureus, with the strongest effect for GBS (Fig. 3A). C5 inhibition had similar but weaker effects, whereas the combined inhibition of C5 and TLR8 did not significantly reduce the IL-1 and TNF levels more than TLR8 inhibition alone. On the other hand, IL-8 induction by both bacteria during 120 min of challenge was strongly dependent on C5, but not TLR8. Also, combined inhibition did not reduce IL-8 more than C5 inhibition alone (Fig. 3A). Induction of IL-6 during 120 min of challenge with GBS was dependent on both C5 and TLR8, whereas S. aureus triggered early IL-6 release mainly via C5. Combined inhibition did not reduce the IL-6 release significantly more than single inhibition.
After 240 min of challenge IL-12p70 was induced, mainly in a TLR8dependent manner (Fig. 3B). Moreover, the effect of C5 inhibition on the other cytokines was diminished, except for IL-6, whereas the effect of CU-CPT9b had increased and partially attenuated also the IL-8 production for the low bacterial dose. For the high bacterial dose, no significant reduction in IL-8 levels was obtained with the inhibitors (Fig. 3B).
IL-6 was most strongly reduced when combining the C5-and TLR8inhibitors. Taken together, cytokine induction in blood by the Grampositive bacteria is dependent on both TLR8 and C5, as hypothesized.
We observed an increased impact of TLR8 and a reduced effect of C5 at the later time point. Combined blockade of C5 and TLR8 had additive effects, most notably on IL-6 production.

Impact of TLR8 and C5 on growth and survival of bacteria in blood
To examine whether inhibiting TLR8 or C5 influenced the bacterial survival in blood, an aliquot of blood samples was lysed, diluted, and plated on blood agar. The viability (CFU/ml) of GBS strongly increased during 240 min of challenge, whereas the S. aureus viability declined (Fig. 4A). Inhibition of C5 alone or in combination with TLR8 further increased the number of viable GBS in blood for the highest bacterial dose, whereas CU-CPT9b alone did not have any effect (Fig. 4A).

Effect of TLR8 and C5 on coagulation activation by bacteria
Coagulation disturbances are believed to be of major importance in sepsis. Therefore, we examined the effects of inhibiting TLR8 and C5 With CU-CPT9b alone, there was also a tendency towards reduction of PTF 1+2 levels. Combined C5-and TLR8-inhibition gave an additive effect, and strongly reduced the initiation of coagulation triggered by the bacteria.
Taken together, our data indicate that TLR8 is important for the production of central proinflammatory cytokines in human blood exposed to live GBS and S. aureus, whereas E. coli is sensed mainly in a TLR8independent manner. TLR8 in PMN contribute to IL-8 release during Gram-positive challenge, although in whole blood IL-8 release is mainly C5 driven. IL-6 production and initiation of coagulated is dependent on both TLR8 and C5, and combined inhibition is an efficient means of attenuating these responses. Phagocytosis and intracellular bacterial killing do not require TLR8 signaling, whereas signaling via C5aR1 enhance phagocytosis and appears especially important for limiting the extracellular growth of GBS.

DISCUSSION
We here examined the importance of TLR8 and C5 for the activation of antibacterial responses in a human whole blood model of bacteremia. The recently developed small molecule inhibitor of TLR8 12 is highly efficient and selective also in human blood. Inhibiting TLR8 reveals its importance for the production of IL-12p70, IL-1 , and TNF during S. aureus and GBS challenge in blood, and is in accordance with our recent findings of a dominating role of TLR8 in the sensing of Gram-positive bacteria in monocytes, mediating strong induction of IFN , 8,9 IL-12p70, IL-1 , TNF, IL-6, and IL-10. 13 The Gram-positive bacteria induced IL-12p70 production in whole blood almost exclusively via TLR8. This is also seen in

F I G U R E 3 Cytokine induction by GBS and S. aureus in whole blood is variably dependent on TLR8 and C5.
Anticoagulated blood was treated with TLR8 inhibitor CU-CPT9b or anti-C5 (eculizumab; aC5), individually or in combination, and a nonspecific antibody in combination with a control reagent served as control. Subsequently, the blood was challenged with live GBS or S. aureus (5 × 10 6 bacteria/ml or 5 × 10 5 bacteria/ml). Cytokine levels in plasma were examined after (A) 120 min and (B) 240 min of infection. In noninfected blood the mean cytokine level (pg/ml) after 120/240 min of incubation were: IL-1 (2/11), TNF (138/231) IL-6 (7/14), IL-8 (866/2489), IL-12p70 (−/1). The graphs show means + SEM (N = 6) purified monocytes, where TLR8 agonists induce much more IL-12p70 compared with TLR2-or TLR4-agonists. 13 Induction of IL-12p70 and IFN via TLR8 requires IRF5 nuclear translocation, which is not triggered by surface TLR signaling. 8,9,13 TLR2 and TLR4 trigger robust production of IL-1 , TNF and IL-6 by monocytes, 13 probably via canonical NF-kB-and MAPK-pathways. IRF5 activation is not essential, but still potentiates the induction of IL-1 , TNF, and IL-6 in TLR8 signaling. 31 Human PMN express all TLRs except TLR3 and TLR7. 32 Synthetic TLR8 ligands induce high levels of IL-8 release in highly pure PMN cultures, whereas the induction of other cytokines is much lower. 33 We here show that TLR8 contributes to the release of IL-8 by PMN F I G U R E 4 C5 cleavage but not TLR8 signaling attenuates the growth GBS in human blood, whereas both processes are dispensable for controlling the growth of S. aureus. (A) Anticoagulated blood was treated with inhibitors of TLR8 or C5, individually or in combination, or with PBS as a vehicle control (no inhibition). Live GBS or S. aureus (5 × 10 6 bacteria/ml or 5 × 10 5 bacteria/ml) were added, and blood aliquots were sampled immediately (T0) or after 240 min of incubation, lysed and plated on blood agar to calculate the numbers of CFU. The graph shows means + SEM (N = 6).
(B) Effect of TLR8 inhibition on intracellular bacterial survival. Anticoagulated blood was treated with TLR8-inhibitor CU-CPT9b or DMSO as vehicle control. Subsequently, live bacteria were added (5 × 10 6 or 5 × 10 5 bacteria/ml), and blood aliquots were sampled immediately (T0) or after 240 min of incubation. WBC were isolated, washed, and lysed to determine the number of intracellular viable bacteria. The graphs show means + SEM (N = 4) F I G U R E 5 Prothrombin cleavage induced by GBS and S. aureus is partially dependent on TLR8 and C5. Anticoagulated blood was treated with TLR8 inhibitor CU-CPT9b or aC5, individually or in combination, or with a control mAb combined with control reagent (Control). Live GBS or S. aureus (5 × 10 6 bacteria/ml or 5 × 10 5 bacteria/ml) were incubated for 240 min, and the plasma levels of prothrombin fragment 1+2 (PTF 1+2 ) were determined with ELISA. In noninfected blood the mean level of PTF 1+2 after 240 min of incubation was 17.5 nM. The graph shows means + SEM (N = 5) during Gram-positive infections, but this did not occur in monocyte cultures. 13 In whole blood, bacteria triggered IL-8 release mainly in a C5-dependent manner, although TLR8 contributed nonredundantly in some conditions. Bacterial infection also resulted in increased permeability of the PMN plasma membrane and increased the levels of extracellular DNA. This could be a result of accidental or programmed necrosis, or it might be due to NET formation, which can be of vital or suicidal type. 30 TLR signaling may influence the PMN life span by triggering of NET formation, but TLR signaling in PMN also attenuates spontaneous apoptosis. 32 However, our findings that may be related to these processes occurred independently of TLR8.
Cytokine induction by GBS challenge in blood appears more TLR8dependent compared with S. aureus. TLR2 also contributes to S. aureus detection in blood, 22 and TLR2 signaling reduces TLR8-mediated IRF5activation. 8 These findings are consistent with a more important role of CD14 on S. aureus induced cytokines, as CD14 is a cofactor in TLR2-and TLR4-signaling that enhances cytokine responses. 34 In contrast, the majority of GBS strains produce little TLR2 activating lipoproteins, 35 13 In the current study, also the E. coli isolates failed to mediate nonredundant cytokine production via TLR8 in blood, whereas the reasons for this discrepancy between these model systems are unclear.
Blocking C3 or C5 alone reduced IL-8 release by E. coli, but had otherwise minor effects on cytokine induction by this bacterium. C5a was previously identified as central for E. coli mediated IL-8 production. 17 With GBS and S. aureus challenge, C3-or C5-inhibition reduced cytokine production more than with E. coli, and the C5a-C5aR1 interaction is known as important for S. aureus induced responses. 24 As expected, the Gram-positive bacteria induce cytokines in blood via both TLR8-and C5-dependent mechanisms, and combined inhibition gave an additive effect, most clearly seen for IL-6.
The number of viable GBS strongly increased during 240 min of incubation in blood, whereas the S. aureus numbers declined. This is in line with previous studies with these bacteria. 24,37,38 Hence, phagocytes in blood more easily resolve infections with S. aureus than GBS. C5 inhibition further increased the growth of GBS, probably because C5aR1 signaling enhances phagocytosis. Similarly, distinct complement inhibitors from scabies increased the growth of S. pyogenes in blood. 39 Phagocytosis of S. aureus was also reduced by C5or C5aR-inhibition, in agreement with our previous findings. 40 Still, C5a-C5aR1 signaling gradually becomes dispensable for phagocytosis of this bacterium as time progresses, and C5 inhibition did not impair the resolution of infection in blood. Although costimulation with a TLR2 agonist increases the uptake of S. aureus by monocytes but not PMN in blood, 8 anti-CD14 has only modest effects. 40 Sensing of bacteria via TLR8 occurs downstream of phagocytosis and bacterial degradation, and it is delayed relative to TLR2 activation. TLR8 signaling may therefore occur too late to influence the phagocytic process.
TLR8 inhibition neither affects the intracellular killing of the bacteria, even though TLR8 can trigger bactericidal effector mechanisms such as degranulation and respiratory burst in PMN in vitro. 33  Coagulation limits the spreading of invading bacteria, whereas DIC is a serious complication in sepsis. E. coli challenge triggers monocyte TF expression and coagulation in a C5a-C5aR1-dependent manner, 17 and inhibition of C5 cleavage attenuated coagulopathy in a baboon model of E. coli sepsis. 19 Activation of coagulation by S. aureus is C5 dependent, 24 and we here show that both C5 and TLR8 enhance prothrombin cleavage upon GBS and S. aureus infection, with an additive effect when these inhibitors are combined.
Numerous selective and nonselective anti-inflammatory reagents and strategies have been tested as adjuvant sepsis treatments, including TLR4 antagonist (eritoran), corticosteroids, anti-TNF strategies, as well as modulators of coagulation. 44 Although novel therapies have shown benefits in animal sepsis models, they have so far failed to translate to the clinical setting. 45 Anti-inflammatory intervention clearly has 2 sides: although it may reduce the harmful effects of excessive inflammation, it may also compromise the host's defense system against infections. Therefore, treating infections with immunomodulatory agents is a fine-tuned act of balance where patient stratification, timing and likely a combination of approaches is needed. We