A versatile assay to determine bacterial and host factors contributing to opsonophagocytotic killing in hirudin-anticoagulated whole blood

Most bacteria entering the bloodstream will be eliminated through complement activation on the bacterial surface and opsonophagocytosis. However, when these protective innate immune systems do not work optimally, or when bacteria are equipped with immune evasion mechanisms that prevent killing, this can lead to serious infections such as bacteremia and meningitis, which is associated with high morbidity and mortality. In order to study the complement evasion mechanisms of bacteria and the capacity of human blood to opsonize and kill bacteria, we developed a versatile whole blood killing assay wherein both phagocyte function and complement activity can easily be monitored and modulated. In this assay we use a selective thrombin inhibitor hirudin to fully preserve complement activity of whole blood. This assay allows controlled analysis of the requirements for active complement by replacing or heat-inactivating plasma, phagocyte function and bacterial immune evasion mechanisms that contribute to survival in human blood.

Blood is normally sterile, but in cases when epithelial barriers are compromised and the immune system is not optimally equipped to fight pathogens, bacteria can be present in the blood, which is called bacteremia. Bacteria have evolved various mechanisms that prevent opsonophagocytosis, contributing to their ability to colonize their host, but also occasionally resulting in severe infections. Overall, Gram-positive bacteria are protected from complement-mediated lysis by the presence of a thick outer cell wall consisting of peptidoglycan, which prevents the bacterial membrane from lysis by the pore-forming membrane attack complex 1 . Conversely, Gram-negative bacteria, which are characterized by an outer membrane surrounding the bacterial cell wall, are vulnerable to complement-mediated killing due to assembly and insertion of the membrane attack complex on the bacterial surface 2 . Several bacterial species express a polysaccharide capsule, that protects them from recognition by opsonizing antibodies and in Gram-negative bacteria such as Haemophilus influenzae from insertion of the membrane attack complex 3 .
Besides a protective capsule, which can be found on both Gram-positive and Gram-negative bacteria, many invasive bacteria are able to hijack human complement regulatory proteins, thereby decreasing complement activation on their bacterial surface. For instance, Streptococcus pneumoniae, H. influenzae, Escherichia coli and Neisseria meningitidis are able to bind human factor H 4-7 , which decreases alternative complement activation and thereby reduces C3 opsonization.
In order to study the complement evasion mechanisms of bacteria, or the capacity of complement to opsonize and kill bacteria, most in vitro studies performed to date are using serum, plasma or baby rabbit complement containing active complement for complement opsonization. For opsonophagocytosis, isolated phagocytes or phagocyte-like cell lines such as HL-60 are used [8][9][10][11] . However, this is by no means representative to the real live situation in whole blood. For instance, the isolation of neutrophils leads to priming, which affects the ability of the neutrophils to form reactive oxygen species and changes their responses to cytokines 12 . In addition, serum has altered levels of coagulation proteins compared to plasma in whole blood. An example is plasminogen 13 , which can bind to the bacterial surface of S. pneumoniae and is involved in bacterial virulence 14,15 . Another example is fibrinogen, shown to bind to Streptococcus pyogenes M protein, which decreases C3b deposition and opsonophagocytosis 16,17 .
To circumvent these limitations in order to study complement-mediated opsonophagocytosis of bacteria, we explored the possibility to use whole blood directly after venous puncture for use in opsonophagocytosis assays. Here, we describe a versatile and easy to perform whole blood killing assay in which both phagocyte function and complement activity can be monitored and modulated. We used a selective thrombin inhibitor hirudin, which preserved complement activity of whole blood, in contrast to lithium heparin, sodium heparin, EDTA or sodium citrate.

Material and Methods
Ethics statement. After informed consent, a venous blood specimen was collected from the median cubital vein of healthy volunteers (age, 20-40 years; both males and females). Collection of blood was approved by the Ethics Committee of the Radboud University, Nijmegen, the Netherlands and experiments were carried out in accordance with local guidelines and regulations and complies with the Declaration of Helsinki and the Good Clinical Practice guidelines.
Bacterial growth conditions. Streptococcus pneumoniae strain TIGR4 18 22 were used in this study. H. influenzae was grown under shaking conditions at 37 °C in brain heart infusion (BHI) broth (Becton Dickinson) supplemented with 10 μ g/mL haemin (Sigma-Aldrich) μ g/mL β -nicotinamide adenine dinucleotide (Merck) (sBHI). S. pneumoniae was grown under static conditions at 37 °C with 5% CO 2 in Todd-Hewitt broth supplemented with 5 g/L yeast extract. N. meningitidis was grown on blood agar plates and collected directly from overnight plates. K. pneumoniae, S. aureus, E. coli and P. aeruginosa were grown under shaking conditions at 37 °C in Luria-Bertani (LB) broth.
Whole blood killing assay. After informed consent was obtained, a venous blood specimen was collected from the median cubital vein of healthy volunteers (age, 20-40 years; both males and females) into S-Monovette r-Hirudin tubes (Sarstedt). Blood was kept at room temperature on a roller bench until used.
For the whole blood killing assay, 100 μ L of hirudin-anticoagulated blood was added per well in a 96-well plate. Bacterial suspensions in PBS, containing 1.10E5 colony forming units (CFU), were added in a maximum volume of 5 μ L and immediately mixed with the blood. The 96-well plate was incubated for the indicated time at 37 °C under continuous shaking. The number of bacterial CFU was determined at start and after incubation by plating serial 10-fold dilutions. The percentage of bacteria that survived was calculated.
For plasma inactivation, 100 μ L of hirudin-anticoagulated blood was added per well in a 96-well plate and centrifuged at 1000 × g for 5 min. Plasma was removed and heat-inactivated for 20 min at 56 °C. Blood cells were washed by adding 100 μ L PBS and centrifuged with 1000 × g for 5 min. PBS was removed and heat-inactivated plasma was mixed with the pelleted cells and used for the killing assay. To examine the effect of plasma alone on bacterial clearance, 200 μ L hirudin-anticoagulated blood was centrifuged 1 min at 16.000 × g and 100 μ L plasma was used for the killing assay in the absence of blood cells. For 50%, 25% and 10% active plasma, 50 μ L, 25 μ L and 10 μ L active plasma was mixed with 50 μ L, 75 μ L and 90 μ L heat-inactivated plasma, respectively, and was mixed with the pelleted blood cells and used for the killing assay. For plasma replacement, 100 μ L Scientific RepoRts | 7:42137 | DOI: 10.1038/srep42137 of hirudin-anticoagulated blood was added per well in a 96-well plate and centrifuged at 1000 × g for 5 min. Plasma was removed and cells were washed by adding 100 uL PBS and centrifuged at 1000 × g for 5 min. PBS was removed and pooled hirudin-anticoagulated plasma was mixed with the pelleted cells and used for the killing assay. For C6-depleted serum (CompTech) and C6-deficient patient serum 23 , serum was diluted in PBS to 10%. Reconstitution of C6 was performed by supplementing 6.4 μ g/mL purified C6 (CompTech) in 10% serum because manufacturer's product description states full reconstitution of serum was achieved with 64 μ g/mL.

Results and Discussion
Hirudin-anticoagulated blood is optimal for complement preservation. We used Streptococcus pneumoniae as model organism to set-up a whole blood killing assay because this bacterium is causing bacteremia in immune competent individuals 25,26 . In order to survive in blood, this bacterium has developed various mechanisms that inhibit recognition by the immune system 27 . For efficient opsonophagocytic killing, C3b opsonization of the bacterial surface of S. pneumoniae is required 28 . To determine which anticoagulants preserved complement C3b opsonization capacity, we determined IgG, IgM and C3 binding to the bacterial surface of S. pneumoniae after 30 minutes with 10% human serum or 10% human plasma anticoagulated with hirudin, lithium heparin, sodium heparin, EDTA or sodium citrate.
Binding of IgG to the bacterial surface of S. pneumoniae incubated with 10% hirudin or EDTA anticoagulated human plasma was slightly increased compared to 10% human serum, whereas no significant differences for IgM were observed. More striking were the differences in C3 opsonization. Here, hirudin anticoagulated plasma showed the highest C3 opsonization of S. pneumoniae, whereas all other anticoagulants showed a significant decrease in C3 opsonization. Complement activity was preserved for at least 2 hours when blood was kept at room temperature (Fig. 1D). From these data, we conclude that hirudin anticoagulated plasma is superior in preserving complement activity.
Previously, Ison et al. determined killing of Neisseria meningitidis in citrate and heparin-anticoagulated whole blood 29 . In this study, heparin-anticoagulated whole blood was superior in killing N. meningitidis serogroup A compared to citrate-anticoagulated whole blood. In subsequent experiments, the same group compared this whole blood killing assay to serum bactericidal assay with blood from vaccinated children and consistently showed increased sensitivity for the whole blood killing assay 30,31 . Also, they showed a reduction of survival of N. meningitidis in the whole blood killing assay with increasing age of patients 32 . Whole blood killing of N. meningitidis has also been performed with hirudin-anticoagulated whole blood. Welsch et al. showed efficient killing of N. meningitidis serogroup B with whole blood from adults 33 . A slightly modified whole blood killing assay, with 25% heat-inactivated serum, showed increased killing with post-immunization serum compared to pre-immunization serum 34 . Comparisons in whole blood killing between huridin and other anticoagulants have not been studied previously.
The differences in complement activity preservation can largely be explained by the function of the different anticoagulants. Lithium heparin and sodium heparin induce a conformational change of antithrombin III to accelerate the inhibition of thrombin and factor Xa, thus preventing thrombin activation and the generation of fibrin. However, heparin is known to bind different proteins in the complement cascade 35 , as well as calcium and magnesium ions 36 , thereby affecting complement activity. Sodium citrate prevents blood from clotting through chelation of calcium ions by forming calcium citrate and EDTA scavenges bi-valent cations, such as calcium and magnesium, both are also required for complement activation. In contrast, hirudin (also known as lepirudin) is a highly specific thrombin inhibitor that does not interfere with complement activation 37 .
Hirudin has previously also been used in whole blood stimulation assays 33,[38][39][40] . This enables to determine the contribution of cross-talk between complement and other factors such as cytokine release 38 , oxidative burst 40 and phagocytosis 39 .
Even though thrombin is not directly involved in complement activation, there are some reports where it has shown to modulate complement activity. For instance, in C3− /− mice, thrombin was overexpressed and showed to cleave C5 into C5a and C5b 41 . In these studies, hirudin reduced acute lung inflammatory injury in C3− /− mice, but had no effect in C3+ /+ , indicating that thrombin-mediated cleavage of C5 only contributed to acute lung inflammatory injury when C3 is absent.
Whole blood killing assay. Many bacterial pathogens such as S. pneumoniae, Staphylococcus aureus, Klebsiella pneumoniae N. meningitidis and H. influenzae frequently cause invasive disease, including sepsis [42][43][44] . When present in the blood, bacteria need to withstand the bactericidal activity of the complement system, and phagocytosis by peripheral blood neutrophils. We used hirudin anticoagulated blood to determine the survival of invasive bacterial pathogens in blood. For these experiments, we used S. pneumoniae strain TIGR4, originally isolated from the blood of a 30-year-old male 18,45 .
The whole blood killing assay is an easy-to-use opsonophagocytic assay to determine survival of bacterial pathogens in blood. Bacteria are added to 100 μ L hirudin anticoagulated blood in a 96-wells round bottom plate and incubated at 37 °C while shaking to prevent sedimentation. Different inoculums (10 3 -10 5 CFU/100 μ L blood) were tested and all showed a decrease in CFU counts over time (data not shown). For subsequent experiments, 10 5 CFU/100 μ L blood were used.
We determined killing of S. pneumoniae in whole blood and observed significant killing already after 1 hour, which increased further in time ( Fig. 2A). In order to determine the role of complement activity and phagocyte function, we performed the whole blood killing assay with either heat-inactivated plasma (see Material and Methods section for procedure) or with only plasma containing active complement. Whereas S. pneumoniae was killed in blood, no killing was observed with plasma only, indicating that phagocytes are required for efficient killing (Fig. 2B). When heat-inactivated plasma was mixed with blood cells, S. pneumoniae was able to grow very rapidly. This indicates that active complement is required for effective opsonophagocytosis as well, but also shows that whole blood contains sufficient nutrients for fastidious growth. The contribution of active complement in S. pneumoniae opsonophagocytosis is known for a long time [46][47][48] , and our results are consistent with these studies.
We determined whole blood killing after 1 hour for different pathogens that cause bacteremia [42][43][44] . Survival of S. pneumoniae strain TIGR4 was 20%. Similar survival was found for K. pneumoniae (16%) and P. aeruginosa (7%), whereas survival for S. aureus (85%) or H. influenzae serotype B (41%) were higher. Survival of H. influenzae serotype A (2%), E. coli (0.03%) and N. meningitidis (0.1%) was much lower. Survival of NTHi was strain dependent, 9% for R2886, but only 0.2% and 0.3% for strains 3655 and 11P6H, respectively. These strain-dependent differences in survival are probably due to variance in complement resistance since we have  10E7) were incubated for 30 minutes in HBSS3+ containing 10% plasma anticoagulated with hirudin, lithium heparin, sodium heparin, EDTA or sodium citrate or serum from the same donor and binding of (A) IgG, (B) IgM, and (C) C3 was determined by flow cytometry (n = 3). One-way analysis of variance (ANOVA) with Dunnett's Multiple Comparison Test was used for statistical analysis. * p < 0.05, ** p < 0.01. (D) Hirudin anticoagulated blood was immediately (0 h) or after 2 hours rolling on a roller mixer (2 h) centrifuged and plasma was stored. Bacteria (1.10E7) were incubated for 30 minutes in HBSS3+ containing 10% of the plasma that was immediately or after 2 hours stored, and binding C3 was determined by flow cytometry (n = 3). A one-tailed student t-test was used for statistical analysis. NS = not significant.
previously shown that survival in pooled human serum was much lower for NTHi strains 3655 and 11P6H as compared to strain R2866 4 .
In order to compare complement-mediated killing and opsonophagocytic-dependent killing for Gram negative and Gram positive bacteria, we determined survival of S. pneumoniae and NTHi strain 3655 with heat-inactivated plasma, plasma and whole blood. Both plasma and whole blood showed significant killing of NTHi strain 3655, whereas this was only the case with whole blood for S. pneumoniae (Fig. 2D). These data clearly indicate that killing of Gram negative, unencapsulated, NTHi strain 3655 was largely dependent on complement-mediated killing, whereas killing of S. pneumoniae was dependent on complement activation and opsonophagocytosis.

Modulation of bacterial, cellular and humoral factors contributing to whole blood killing. With
this whole blood killing assay, bacterial factors as well as host cellular and humoral factors can be modulated to determine their contribution to opsonophagocytic killing. For instance, blocking complement receptor 3 (CR3) with α -CD11b antibody 44a decreased killing of S. pneumoniae (Fig. 3A), indicating that recognition of C3b on the bacterial surface by phagocytes contributes to killing. The contribution of the CR3 in opsonophagocytosis is S. pneumoniae by neutrophils and macrophages is widely investigating [49][50][51] , and our results are consistent with these studies.
In addition, treatment of blood with cytochalysin D, an inhibitor for actin polymerization, also decreased killing of S. pneumoniae (Fig. 3A), indicating that killing was dependent on phagocytosis.
Recently, Corriden et al. showed that tamoxifen augmented neutrophil-mediated killing of S. aureus, E. coli and Pseudomonas aeruginosa through enhancing several pro-inflammatory pathways in human neutrophils, including chemotaxis, phagocytosis and neutrophil extracellular trap (NET) formation 52 . Here, we show that adding 10 μ M 4-hydroxytamoxifen significantly augmented killing of S. pneumoniae in whole blood (Fig. 3B).
Killing of S. pneumoniae, but also other pathogens, is affected by the presence of opsonizing antibodies and the overall complement activity. To determine the role of complement activity, we used whole blood of which plasma was removed by centrifugation and replaced with 50%, 25% or 10% plasma containing active complement (see Material and Methods section for procedure). Replacement of the total amount of active plasma with 50% active plasma clearly decreased the killing capacity, which was even more apparent for 25% and 10% active plasma (Fig. 3C), indicating that decreasing the level of active complement reduces the capacity to clear S. pneumoniae from blood in a dose dependent manner. Previously, we have used the whole blood killing assay to assess the contribution of human factor H in controlling complement activity and killing of S. pneumoniae by replacing plasma with factor H-depleted serum and supplementation with different concentrations of purified human factor H. In this assay, we showed that increasing human factor H to blood increased survival of S. pneumoniae, whereas decreasing factor H levels increased killing 53 . Adding 100 μ g/mL factor H to whole blood decreased killing of S. pneumoniae significantly (Fig. 3D), which is in accordance with the findings that higher factor H levels decreased bacterial killing 53 . Binding of factor H to the bacterial surface was shown to protect many bacteria from complement-mediated opsonization 4,5,7,54-58 . Pneumococcal surface protein C (PspC) of S. pneumoniae is known to bind human factor H. In order to determine the role for factor H binding in whole blood killing we determined survival of a Δ pspC mutant and found that this mutant, as expected, had a decreased survival in whole blood (Fig. 3E). Overall, this demonstrates several possibilities in studying functions of complement in bacterial clearance using the whole blood killing assay.
While performing our whole blood killing experiments, we observed large inter-patient differences in S. pneumoniae survival (% survival 0.03-2.00) (Fig. 3F). To determine whether these differences can mainly be attributed to differences in plasma content or phagocyte function, we used whole blood from which plasma was removed by centrifugation and replaced it with pooled plasma in which the concentrations of opsonizing antibodies and the complement activity are constant (see Material and Methods section for procedure). In this assay, killing of S. pneumoniae was more consistent (% survival 0.39-1.30), compared to survival in blood from the four single donors (% survival 0.03-2.00) (Fig. 3F). These data indicate that mainly differences in plasma components (opsonizing antibodies and complement activity) between these four donors attribute to the inter-donor variation in whole blood killing capacity. This approach can also be used to determine vaccine-induced protection. Previously, Welsch et al. showed that supplementation of whole blood with 25% heat-inactivated post-vaccination serum increased killing of N. meningitidis compared to pre-immunization serum 34 . This approach enables comparison of whole blood killing of pathogens with different serum samples in combination with a single fresh blood donor. To mimic this in our whole blood killing assay, we replaced plasma with 10% C6-depleted serum and determined survival of N. meningitidis serogroup B strain H44/76 after 30 minutes. The presence of 10% heat-inactivated C6-depleted serum showed 9.1% survival (Fig. 4A), indicating that N. meningitidis serogroup B strain H44/76 is killed by complement-independent mechanisms, which has been described in literature previously. For instance, N. meningitidis serogroup C was killed by antibody-dependent cell-mediated antibacterial activity 60 as well as opsonin-independent phagocytosis 61 . Opacity (Opa) proteins have been implicated to be important in opsonin-independent phagocytosis of N. meningitidis 62,63 through neutrophil surface receptors CD66 and CR3 64,65 , whereas macrophages bind unopsonized N. meningitidis almost exclusively via the class A macrophage scavenger receptor 66 . Although not investigated in detail, we show that N. meningitidis serogroup B strain H44/76 is efficiently killed through complement-independent mechanism.
To study the contribution of complement in addition to complement-independent mechanisms, we used 10% C6-depleted serum in the whole blood killing assay. When 10% C6-depleted serum was used, survival was significantly lower (4.0%) compared to heat-inactivated serum (Fig. 4A). Since C6-depleted serum is not able to form a membrane attack complex, this increased killing is likely due to complement-dependent opsonophagocytosis. In order to restore terminal complex activity, we supplemented C6-depleted serum with C6 and observed a significant increase in killing, implicating that formation of membrane attack complex, next to opsonin-dependent and opsonin-independent killing, contributed to overall clearance of N. meningitidis from whole blood. Previously, we have described a patient with a novel heterozygous missense mutation in the C6 gene. Next to this novel heterozygous C6 mutation, a known heterozygous splice site variation was also identified, resulting in a C6 molecule that is 14% shorter due to a premature stop codon, but can still be build into the terminal complement complex, can kill bacteria, and is hemolytically active 67,68 . But, both mutations resulted in a lower (5%) C6 protein level. Normal immunoglobulin levels (IgG/IgA/IgM/IgE) and other complement factors (C3, C4) were found. When 10% C6-deficient patient serum was used, survival was low (2.3%) (Fig. 4B), which was consistent with results obtained with C6-depleted serum. Survival was significantly lower when C6 was reconstituted (0.7%) (Fig. 4B), indicating that C6 supplementation increased bacterial killing in whole blood.
Altogether, these results with C6-depleted serum and C6-deficient patient serum obtained similar results; decreased killing as compared to C6-reconstituted serum, which is consistent with the clinical phenotype of these patients.
Monocytes and neutrophils contribute to opsonophagocytosis of S. pneumoniae. In order to address which cell type was predominantly responsible for opsonophagocytosis of S. pneumoniae in whole blood, we labelled S. pneumoniae with CFSE as previously described 24 . CFSE-labelled bacteria were added to whole blood in the absence or presence of cytochalysin D to block phagocytosis. Especially monocytes (74%) and neutrophils (72%) were found to bind and phagocytose S. pneumoniae, which was only 9% for lymphocytes (Fig. 5A). Cytochalysin D decreased S. pneumoniae association to monocytes, neutrophils and lymphocytes to 24%, 42% and 5%, respectively, indicating that half of the cells in the control condition actually phagocytosed S. pneumoniae, whereas the other half of the CFSE-labelled bacteria were cell-associated. When the total percentage of S. pneumoniae association with cells was determined, most of them, 85%, were neutrophils, 10% monocytes and 5% lymphocytes (Fig. 5B), indicating that neutrophils are the most important cell type for opsonophagocytosis of S. pneumoniae in whole blood. These results are consistent with previous literature where phagocytosis experiments showed efficient uptake of opsonized S. pneumoniae by both macrophages and neutrophils [69][70][71] . The important role for neutrophils in opsonophagocytic killing and protection against pneumococcal disease is supported by in vivo models wherein neutrophils were depleted 72,73 .

Conclusion
The use of hirudin-anticoagulated whole blood enabled us to study the contribution of both bacterial and host factors in the killing of several pathogens, including S. pneumoniae, K. pneumoniae, S. aureus and H. influenzae. Complement activity preservation of hirudin was superior compared to lithium heparin, sodium heparin, EDTA or sodium citrate. Altogether, we describe a versatile assay to determine bacterial and host factors affecting opsonophagocytic killing of bacteria in hirudin-anticoagulated whole blood as a model for bacteremia.