Inhibition of staphylothrombin by dabigatran reduces Staphylococcus aureus virulence


Thomas Vanassche, Center for Molecular and Vascular Biology (CMVB), University of Leuven, O&N1 – 9th Floor, Herestraat 49, B-3000 Leuven, Belgium.
Tel.: +32 16 34 57 75; fax: +32 16 34 59 90.


Summary. Background: Staphylocoagulase and von Willebrand binding protein (VWbp) bind to prothrombin to form the staphylothrombin complex that converts fibrinogen into fibrin. Objectives: To study the role of staphylothrombin and its inhibition by dabigatran on Staphylococcus aureus virulence. Methods: We studied the effect of staphylothrombin inhibition on bacterial attachment to polystyrene surfaces, leukocyte activation and bactericidal activity for S. aureus ATCC 25923, S. aureus Newman, and staphylocoagulase- and VWbp-negative S. aureus Newman mutants in the presence or absence of prothrombin and fibrinogen. We measured the abscess size after subcutaneous (s.c.) injection of S. aureus ATCC 25923 and S. aureus Newman, as well as an S. aureus Newman mutant strain lacking staphylocoagulase and VWbp, in mice treated with either dabigatran or placebo. Results: Staphylothrombin-mediated fibrin increased the association of S. aureus to polystyrene surfaces and reduced the bactericidal activity of leukocytes. The absence or inhibition of staphylothrombin decreased the bacterial association, enhanced leukocyte activation and reduced bacterial survival in vitro. Abscess size was smaller in mice treated with dabigatran or infected with a coagulase-negative mutant. Conclusion: Inhibition or the absence of staphylothrombin reduced S. aureus virulence in in vitro and in vivo models.


Staphylococcus aureus is a frequent and versatile pathogen. Although it is part of the commensal skin flora in about 30% of healthy individuals [1], S. aureus causes a broad spectrum of clinical infections, ranging from localized skin and soft tissue infections to life-threatening conditions such as infective endocarditis and sepsis. Staphylococcus aureus is the leading cause of hospital-acquired infections, and the rapid spread of antibiotic-resistant strains imposes major public health and economical challenges [2,3].

The hallmark of S. aureus infection is the formation of abscesses, consisting of bacteria and an intense inflammatory infiltrate composed mainly of polymorphonuclear cells (PMN) [4]. To persist in the presence of these massive numbers of activated leukocytes, S. aureus has evolved an impressive arsenal of virulence factors to bypass both the innate and adaptive immunity, including factors that inhibit complement activation, opsonization and phagocytosis [5,6]. One of the earliest features of staphylococci linked to its disease-causing potential was the ability to clot human blood [7], and until today, the coagulase activity is used in routine laboratory testing to distinguish S. aureus from less virulent coagulase-negative staphylococci [8].

The ability to trigger coagulation is the result of the secretion of bacterial proteins capable of directly activating the host’s prothrombin. The best characterized coagulase is staphylocoagulase, a protein secreted by nearly all isolates of S. aureus. Staphylocoagulase directly binds human prothrombin to form the staphylocoagulase–prothrombin complex. The resulting conformational change of prothrombin creates an enzymatically active site capable of converting fibrinogen into fibrin [9]. In addition to staphylocoagulase, von Willebrand binding protein (VWbp) also acts as a coagulase that binds to and activates prothrombin in a similar conformational, non-proteolytic way [10]. The active complex of the host’s prothrombin with either staphylocoagulase or VWbp is referred to as ‘staphylothrombin’.

Although the potential of S. aureus to coagulate blood has been known for nearly a century [7], and in spite of detailed insights into the underlying molecular mechanism by which staphylocoagulase and VWbp conformationally activate prothrombin [9,11], the role of S. aureus-induced coagulation in virulence has long remained unclear. From previous studies, experimental data have yielded conflicting results. Although some early studies showed a decreased virulence of coagulase-negative mutants [12–16], these studies used non-specific mutants, likely to have an inherently lower virulence compared with the parent strain. More recent studies using site-specific mutants failed to show an effect on S. aureus virulence in an experimental infective endocarditis model [17], but showed a reduced virulence of matched coagulase-negative mutants in a model of pulmonary infection [18], and, more recently, in a sepsis model of S. aureus infection [19]. Furthermore, the inhibition of staphylocoagulase both with inhibitory RNA [20] and neutralizing antibodies [19] has resulted in a reduced virulence in an in vivo S. aureus infection model.

Because the direct conformational activation of prothrombin by staphylocoagulase and VWbp bypasses the tight physiological regulation of prothrombin activation, classic anticoagulant drugs such as vitamin K antagonists and heparins have no effect on staphylothrombin function [21]. As the binding of staphylocoagulase to (pro)thrombin blocks the thrombin exosite 1 [11], the bivalent direct thrombin inhibitors such as hirudin and its derivatives cannot bind to the staphylothrombin complex and consequently do not inhibit staphylothrombin. The absence of a potent pharmacological staphylothrombin inhibitor has hampered research on the role of coagulase activity in S. aureus virulence.

In contrast to the bivalent thrombin inhibitors, the newer small molecule thrombin inhibitors argatroban and dabigatran that target only the catalytic site of thrombin inhibit staphylothrombin with a high affinity [22,23]. Using dabigatran to pharmacologically target staphylothrombin activity, we studied the role of fibrin deposition by S. aureus in bacterial attachment and on the persistence of S. aureus in the presence of human leukocytes in vitro, and investigated the role of staphylothrombin in a murine model of subcutaneous abscess formation.

Materials and methods

List of bacterial strains

Staphylococcus aureus ATCC 25923 (S. aureus Seattle 1945) and S. aureus Newman are reference strains that were originally isolated from clinical infections. The matching S. aureus Newman mutants S. aureus Newman Δcoa−/−, S. aureus Newman ΔVWbp−/−, and the double mutant S. aureus Newman Δcoa−/−,ΔVWbp−/−were recently described; as well as S. aureus Newman Δcoa−/−,ΔVWbp−/− carrying a plasmid encoding staphylocoagulase and VWbp [19]. All strains were stored in BHI (Brain Heart Infusion) with glycerol at −80 °C. Before use, strains were allowed to grow overnight in tryptic soy broth (TSB) at 37 °C in aerobic conditions.

Dabigatran inhibits staphylothrombin

The chromogenic thrombin substrate CS-01(38) (400 μm; HYPHEN BioMed, Neuville-sur-Oise, France) was added to either human prothrombin (25 μg mL−1 in phosphate-buffered saline [PBS]; Enzyme Research Laboratories, Swansea, UK) or murine prothrombin (6.25 μg mL−1 in PBS) (Nuclilab, Ede, the Netherlands) in the presence of dabigatran (Boehringer Ingelheim, Pharma GmbH & Co. KG, Biberach, Germany) at concentrations of 0, 100, 200, 400, 800 and 1600 nm. After 10 min, staphylocoagulase (Synapse, Maastricht, the Netherlands) was added (10 nm). Substrate conversion was measured in an ELISA reader at 405 nm at 30-s intervals over 15 min. The maximal change in absorbance (A405 nm min−1) was normalized using enzyme activity in the absence of dabigatran as 100%.

To calculate the Ki, murine staphylothrombin was incubated with dabigatran at concentrations of 0, 50, 100, 200, 400 and 800 nm. The maximal rate of substrate conversion was analyzed at varying substrate concentrations (200, 400, 800 and 1600 μm), and the Ki was calculated from the resulting substrate-velocity curves by non-linear fit for competitive inhibition using Prism software version 5.0d (GraphPad Software, San Diego, CA, USA).

Relative contribution of staphylocoagulase and VWbp towards total staphylothrombin activity using human and murine prothrombin

To quantitatively assess the relative contribution of staphylocoagulase and VWbp towards the total staphylothrombin activity, we measured the maximum rate of substrate conversion after incubation of 25 μL of a suspension (OD600 of 1.5) of S. aureus Newman and matching staphylocoagulase-, VWbp- and a double mutant with either human or murine prothrombin (both at 6.25 μg mL−1), with or without dabigatran (500 nm). The maximal change in absorbance (A405 nm min−1) was normalized using the enzyme activity of S. aureus Newman with human prothrombin in the absence of dabigatran as 100%.

Attachment to polystyrene plates

Overnight cultures of S. aureus strains ATCC 25923, S. aureus Newman, and both single and double staphylocoagulase- and VWbp-knockout strains were resuspended in TSB to 1.0 McFarland, and further diluted 1/100 in either TSB or TSB mixed with thawed pooled human plasma (vol/vol 1/1), with or without dabigatran (500 nm). For the other experiments, the strains were diluted 1/100 in TSB with fibrinogen (2 mg mL−1), human prothrombin (2 μm), or both, with or without inhibitor. Inhibitors used were lepirudin (Pharmion, Tiel, the Netherlands) (50 μg mL−1), dabigatran (500 nm) or argatroban (1 μm) (GlaxoSmithKline, Zeist, the Netherlands), depending on the experiment. After overnight incubation in polystyrene 96-well plates, wells were washed and air-dried.

Wells were stained with 150 μL of crystal violet during 15 min. After washing and air-drying, plates were examined by light microscopy and read in an ELISA reader at 550 nm after solubilization of the crystal violet with ethanol. As a control condition in order to assess the direct effect of fibrin on crystal violet staining, human α-thrombin (1 U mL−1) was added to the wells instead of the bacteria and adherent fibrin layers were stained using the same protocol (data shown in Fig. S1).

For the quantification of viable bacteria, wells were filled with 200 μL of sterile PBS and sonicated for 10 min. Serial dilutions were plated on TSA agar and colony-forming units (CFUs) were counted after overnight aerobic incubation.

For the semi-quantitative measurement of viable bacteria, 100 μL of 1 mg mL−1 methylthiazolyldiphenyl-tetrazolium bromide (MTT; Sigma-Aldrich NV, Bornem, Belgium) was added to the wells for 3 min. After washing, the resulting MTT formazan was solubilized with a 10% dimethylsulfoxide (DMSO) in acidified ethanol and plates were read in an ELISA reader at 550 nm with a reference at 630 nm (A550–630 nm).

Leukocyte activation by S. aureus

The total peripheral leukocyte fraction was isolated by dextran-sedimentation of fresh citrated blood of healthy volunteers, followed by osmotic lysis of red cells. Leukocytes were suspended in PBS, counted, and adjusted to 10 000 μL−1. The resulting PMN fraction was approx. 75%–80% of total WBC count.

In 96-well plates, we added 20 μL of a 1.0 McFarland suspension of S. aureus Newman, S. aureus Newman Δcoa−/−, S. aureus Newman ΔVWbp−/−, S. aureus Newman Δcoa−/−,ΔVWbp−/− and the plasmid-carrying double mutant strain to 100 μL of TSB containing either fibrinogen (2 mg mL−1) or fibrinogen and prothrombin (2 μm), with and without dabigatran (500 nm) or Gly-Pro-Arg-Pro peptide (GPRP; Sigma-Aldrich NV) (0.5 mg mL−1). After 60 min of incubation at 37 °C, we added 50 μL of the leukocyte suspension to each well and further incubated for 60 min, after which the peroxidase detection substrate (OPD in citrate buffer at pH 5.0 with H2O2) was added. The reaction was stopped after 5 min by adding H2SO4 (2.0 m), and plates were read at 490 nm with a reference at 630 nm. As a control, Triton X-100 was added to a series of wells to evaluate myeloperoxidase (MPO) activity after cell lysis (maximum stimulation) with and without additional thrombin (1.0 U mL−1), to exclude a direct effect of fibrin formation on the measurement of MPO activity (data shown in Fig. S2).

Leukocyte killing of S. aureus

To study whether staphylothrombin-mediated fibrin deposition affected leukocyte killing of S. aureus, bacteria were cultured in plasma with or without dabigatran (500 nm) and bacterial viability was assessed 1 h after coincubation with isolated human leukocytes by conversion of a chromogenic substrate.

For this purpose, S. aureus strain ATCC 25923, S. aureus Newman and the respective mutant strains were diluted in TSB to 1.0 McFarland. In a 24-well plate, 15 μL of the bacterial suspension was mixed with 285 μL of plasma with or without dabigatran (500 nm) and incubated for 60 min at 37 °C on a plate shaker (200 rpm). The total peripheral leukocyte fraction was isolated as described above, and 100 μL of leukocyte suspension (20 000 μL−1) or PBS (control) was added to the wells. After 120 min of coincubation in the same conditions, leukocytes were lysed by adding 30 μL of Triton X-100 10%. Preliminary experiments had shown that this had no effect on bacterial viability (data not shown). To assess bacterial viability, 20 μL of MTT (1 mg mL−1) was added and color conversion was allowed over 5 min after which the reaction was stopped and MTT formazan was solubilized as described above. To enhance solubilization, plates were sonicated during 15 min, and A500–630 nm was determined using an ELISA reader. The percentage of bacterial survival was calculated as follows: 100%*(A550–630 nm with leukocytes)/(A550–630 nm without leukocytes) for plasma (control) and plasma with dabigatran.

In vivo subcutaneous abscess model

We studied abscess size as a function of time (0–6 days) after the subcutaneous (s.c.) injection of a suspension of S. aureus and sterile dextran beads (Cytodex; Sigma-Aldrich NV) in mice treated with either dabigatran or placebo.

Female Swiss mice (10 weeks old) were fed a normal rodent chow supplemented with dabigatran etexilate (10 mg g−1 chow) or matching placebo starting 2 days before inoculation. For pharmacological analysis, mice were bled by retro-orbital puncture on day 2, 500 μL of blood was collected on citrate (1/10 vol/vol), and prothrombin time (PT) and activated partial thromboplastin time (APTT) were measured on a BCS-XP coagulation analyzer (Siemens, Hamburg, Germany) using manufacturer reagents. EDTA-anticoagulated murine plasma samples were obtained for the measurement of dabigatran concentrations.

The s.c. abscess model is described elsewhere [24]. Briefly, mice were anesthetized using Nembutal, shaven and injected s.c. in both flanks with 100 μL of a suspension containing 107 CFU of S. aureus ATCC 25923, S. aureus Newman or S. aureus Newman Δcoa−/−,ΔVWbp−/− mixed with sterile dextran beads in sterile PBS. In previous experiments, we had established that a sterile suspension of dextran beads as such did not yield any palpable abscesses. Abscess diameter (d) was measured daily on anesthesized mice, and the abscess volume was calculated assuming spherical abscesses with r = d/2. On day 6, animals were killed and the lesions were excised, fixed in 4% paraformaldehyde and embedded in paraffin. In a number of animals, the kidneys and spleen were removed in sterile conditions, homogenized and plated for bacterial quantification. All animal experiments were approved by the ethical committee of the University of Leuven.

Fibrinogen immunohistochemistry staining

Sections were deparaffinized and incubated in antigen retrieval buffer (pH 6.0; Dako, Heverlee, Belgium) for 20 min at 95 °C. After rinsing, the slides were incubated in 0.3% hydrogen peroxide in methanol for 20 min and then in 20% normal rabbit serum (Dako) for 45 min. The primary antibody (goat anti-mouse fibrinogen, Nordic, 1/1000 in TNB buffer) was applied to the slides for overnight incubation at 4 °C. Rabbit anti-goat immunoglobulins 1/100 (Sigma-Aldrich, Bornem, Belgium) in 10% normal mouse serum (Dako) were used as the secondary antibody (45 min). Goat-PAP (1/50, Sigma) was applied for 45 min, and antigen-antibody binding was detected with the DAB substrate system. The slides were briefly immersed in hematoxylin for counterstaining and evaluated by light microscopy.

Statistical analysis

All calculations were done using GraphPad Prism 5.0d (GraphPad Software) . Values were compared using one-way anova with Tukey’s Multiple Comparison post-testing when multiple values were compared, or Student’s t-test when comparing two values. Error bars represent mean ± standard error of the mean (SEM). A P-value of < 0.05 was considered significant.


Dabigatran inhibits staphylothrombin

In the presence of dabigatran, staphylothrombin activity towards the chromogenic substrate was inhibited in a dose-dependent manner with an IC50 of 219 nm (95% confidence interval [CI] 150–319) and 114 nm (95% CI 91–142) for murine and human staphylothrombin, respectively (Fig. 1A). The Ki of dabigatran for murine staphylothrombin was measured as 107 nm (95% CI 74.7–141.2 nm).

Figure 1.

 (A) Dabigatran inhibits staphylothrombin. Human and murine staphylothrombin activity as measured by the conversion of the chromogenic substrate CS-01(38) was reduced by dabigatran in a concentration-dependent way. (B) The relative contribution of staphylocoagulase and von Willebrand binding protein (VWbp) towards total staphylothrombin activity using human and murine prothrombin. Staphylothrombin activity was measured after incubation of Staphylococcus aureus in the presence of human (left panel) and murine (right panel) prothrombin, showing the relative contribution of staphylocoagulase and VWbp towards total staphylothrombin activity (black bars). Regardless of the source of prothrombin, dabigatran fully inhibited staphylothrombin activity by staphylocoagulase, VWbp or both (gray bars). P-value: < 0.0001****; < 0.001***; 0.01 to 0.05*; > 0.05NS.

Relative contribution of staphylocoagulase and VWbp towards total staphylothrombin activity using human and murine prothrombin

When using murine prothrombin, the total staphylothrombin activity of wild-type S. aureus Newman was lower as compared with human prothrombin (38% vs. 100%). Using both murine and human prothrombin, the double knockout mutant had a very low rate of substrate conversion, similar to the control conditions, and expression of both staphylocoagulase and VWbp by means of a plasmid restored full staphylothrombin activity. The Δcoa−/− strain (which only has VWbp activity) and ΔVWbp−/− (only staphylocoagulase activity) had intermediate levels of total staphylothrombin activity compared with the parent strain, with the Δcoa−/− mutant showing a larger decrease in staphylothrombin activity than the ΔVWbp−/− mutant (29.1% vs. 69.9%). This was true both for human and murine prothrombin. In the presence of dabigatran (500 nm), substrate conversion was completely inhibited in all strains (Fig. 1B).

Staphylothrombin enhances S. aureus association to polystyrene surfaces

Using crystal violet staining, the addition of 50% plasma to the medium increased the retention of S. aureus strain ATCC 25923 on the polystyrene surface (A550 nm of 2.06; 95% CI 1.95–2.17 vs. 0.19 in the absence of plasma; 95% CI 0.18–0.19). However, when dabigatran was added to the TSB/plasma mixture, the amount of staining was not increased when compared with TSB alone (0.22; 95% CI 0.19–0.25) (Fig. 2B). Although fibrin in itself was found to marginally increase crystal violet staining, additional control experiments confirmed that the observed large increase in staining was because of the presence of S. aureus (data not shown). Similarly, the addition of plasma also increased the retrievable CFUs more than 100-fold when compared with TSB alone (4.01 × 106 vs. 2.49 × 104 CFU μL−1). The addition of either dabigatran (3.55 × 104 CFU μL−1) or argatroban (1.08 × 105 CFU μL−1), but not lepirudin (2.89 × 106 CFU μL−1), reduced the number of retrievable CFUs to control levels (Fig. 2C).

Figure 2.

 The association to polystyrene wells. Photomicrograph of crystal violet staining at 20× magnification (A). The association of Staphylococcus aureus ATCC 25923 to polystyrene wells as measured by crystal violet staining was increased in the presence of plasma (A, middle panel) compared with TSB alone (A, left panel). When dabigatran was added to the plasma, crystal violet staining was reduced to control levels (A, right panel). (B) shows quantification of crystal violet staining after solubilization in ethanol. When bacteria are quantitatively plated, the presence of plasma increased the bacterial counts a hundredfold, which was inhibited by both dabigatran and argatroban, but not hirudin (C). The bacterial association was increased only in the combined presence of both fibrinogen and prothrombin (D). The increase in retention was lost in the staphylocoagulase–von Willebrand binding protein (VWbp) double mutant, and was less pronounced in both the coagulase-negative and the VWbp-negative mutants, compared with the parent strain (E, black bars). Pharmacological inhibition of staphylothrombin abrogated the increase in S. aureus in the parent strain and in the VWbp-negative mutant, but had no further effect on the mutants lacking staphylocoagulase (E, gray bars). Fg, fibrinogen; ProTr, prothrombin; A550, absorbance at 550 nm; CFU, colony-forming units; TSB, tryptic soy broth. P-value: < 0.0001****; < 0.001***; 0.001 to 0.01**; > 0.05ns.

We analyzed a crystal violet staining after S. aureus growth in TSB that was supplemented with fibrinogen, prothrombin, or fibrinogen and prothrombin together, in order to identify the role of the individual components of the staphylocoagulase–prothrombin complex and its substrate fibrinogen. Staining for S. aureus was increased only when both fibrinogen and prothrombin were present (2.83; 95% CI 2.46–3.21 compared with 0.22; 95% CI 0.01–0.44), but not when either one of them alone was present (not significant [NS] compared with the control). Again, in the presence of both fibrinogen and prothrombin, dabigatran, but not lepirudin, significantly reduced the bacterial association (Fig. 2D). Similar results were obtained using MTT as opposed to crystal violet to quantify S. aureus retention (data not shown).

In the presence of plasma, the combined Δcoa−/−,ΔVWbp−/− knockout strain showed the largest reduction in association with polystyrene surfaces when compared with the wild type (ΔA500–630 nm of 0.06 ± 0.03 vs. 0.89 ± 0.05). Both the staphylocoagulase mutant and the VWbp mutant also showed an intermediate reduction in association (ΔA500–630 nm of 0.38 ± 0.03 and ΔA500–630 nm of 0.57 ± 0.08, respectively). Dabigatran reduced the retention of the parent strain and the ΔVWbp−/− strain, but had no additional effect on the Δcoa−/− and double mutant strains (Fig. 2E).

Staphylothrombin reduces leukocyte activation

Coincubation of human leukocytes with S. aureus Newman induced the release of MPO (A490–630 nm of 1.13 [95% CI 0.76–1.50] vs. 0.32 [95% CI 0.20–0.46] in the absence of S. aureus). We observed a significant reduction in MPO release when S. aureus Newman was allowed to grow in the presence of both fibrinogen and prothrombin (56.1%, < 0.001), but not in the presence of fibrinogen alone (110.1%, NS). This reduction in MPO release was not noted in the presence of the fibrin polymerization inhibitor GPRP (95.5%, NS), nor in the presence of dabigatran (117.2%, ns) (Fig. 3A). Similarly, this reduction in leukocyte activation was abrogated by knockout of staphylocoagulase or staphylocoagulase and VWbp (94.5 ± 10.3% and 100.2 ± 6.0%), and could be restored by the transformation of the double mutant with the staphylocoagulase and VWbp carrying plasmid. Compared with the wild type, the VWbp-lacking mutant showed a smaller reduction in MPO release (77.3 ± 6.4%; Fig. 3B).

Figure 3.

 Staphylothrombin and leukocyte activation. Leukocyte MPO release after exposure to Staphylococcus aureus Newman was lowered in the presence of both fibrinogen and prothrombin, but not in the presence of fibrinogen alone. Both dabigatran that inhibits staphylothrombin and GPRP that blocks fibrin polymerization restored MPO release by leukocytes exposed to S. aureus Newman in the presence of fibrinogen and prothrombin (A). The reduced activation of leukocytes by S. aureus Newman in the presence of both fibrinogen and prothrombin depends on staphylocoagulase, and, to a lesser extent, von Willebrand binding protein (VWbp). Knockout of staphylocoagulase restored MPO release by leukocytes to control levels, whereas the knockout of VWbp alone led to an intermediate phenotype. Transformation of the double knockout with the staphylocoagulase, VWbp-carrying plasmid restored the parent phenotype (B). MPO, myeloperoxidase; GPRP, Gly-Pro-Arg-Pro peptide; TSB, tryptic soy broth; Fg, fibrinogen; ProTr, prothrombin. P-value: < 0.001***; 0.001 to 0.01**; 0.01 to 0.05*.

Staphylothrombin activity and bacterial survival in the presence of leukocytes

When grown in plasma, the bacterial survival after coincubation with leukocytes was 87.5% (95%CI 79.1–96.0) for S. aureus ATCC 25923 and 98.2% (88.1–108.4) for S. aureus Newman. When compared with the parent strain, both the Δcoa−/− and the ΔVWbp−/− mutant, as well as the double mutant strain, demonstrated a significant reduction in survival when exposed to leukocytes (81.4 ± 2.1%, 84.5 ± 4.9% and 70.0 ± 4.8%, respectively; < 0.05 for all). The insertion of a plasmid encoding staphylocoagulase and VWbp into the double mutant strain restored survival (93.7%, 95% CI 83.4–103.9, NS compared with wild type).

Dabigatran significantly reduced the survival of S. aureus ATCC 25923 (65.8% vs. 87.5% with and without dabigatran, respectively; < 0.0001) and of S. aureus Newman (70.0% vs. 98.2%, < 0.0001). Although to a lesser extent, dabigatran further reduced the survival of the Δcoa−/− and ΔVWbp−/− mutants (69.3% vs. 81.4% and 68.2% vs. 84.5%; both < 0.05), but had no additional effect on the survival of the Δcoa−/−,ΔVWbp−/− mutant (64.3% vs. 70.0%, P = NS). Insertion of the staphylocoagulase, VWbp-encoding plasmid restored the effect of dabigatran on survival in the Δcoa−/−,ΔVWbp−/− mutant (70.8% vs. 93.7% [< 0.01]) (Fig. 4).

Figure 4.

Staphylococcus aureus survival in the presence of leukocytes. After incubation in plasma, survival in the presence of leukocytes was lower in the Δcoa−/−, the ΔVWbp−/− and the Δcoa−/−,ΔVWbp−/− mutant when compared with the parent strain, and insertion of a staphylocoagulase/von Willebrand binding protein (VWbp)-encoding plasmid in the double mutant restored survival (white bars). Dabigatran (black bars) reduced the survival of S. aureus ATCC 25923, S. aureus Newman and the plasmid-carrying strain. Dabigatran further reduced the survival of the Δcoa−/− and the ΔVWbp−/− mutant, but had no additional effect on survival in the Δcoa−/−,ΔVWbp−/− mutant. P-value: < 0.0001****; 0.001 to 0.01**; 0.01 to 0.05*; > 0.05ns.

Role of staphylothrombin in a mouse s.c. abscess model

Treatment with chow supplemented with dabigatran etexilate resulted in dabigatran plasma concentrations of 427 ng mL−1 (95%CI 369–485 ng mL−1), and prolonged PT and aPTT (23.3 ± 1.34 vs. 12.7 ± 0.76 s, P = 0.0001 and 81.6 ± 3.85 vs. 25.0 ± 1.30 s, < 0.0001, respectively). There were no clinical signs of bleeding at any time point. There were no differences in the amount of food consumed in the dabigatran and control groups.

After the s.c. injection of S. aureus ATCC 25923, all mice developed evaluable s.c. lesions. Mice treated with dabigatran had significantly smaller abscess sizes (8.8 vs. 45.1 mm3, < 0.0001 on Day 2). The differences remained significant until Day 5, after which the abscess volume regressed (Fig. 5A). A similar reduction in abscess size in dabigatran-treated mice was observed when S. aureus Newman was injected. When mice were infected with the Δcoa−/−,ΔVWbp−/− double mutant, almost no abscesses were observed, in contrast to the parent strain. Dabigatran had no further effect on the small abscesses caused by the knockout strain (Fig. 5B). Homogenates from the kidneys and spleen remained sterile both in control mice as in dabigatran-treated mice.

Figure 5.

 After infection with both Staphylococcus aureus ATCC 25923 (A) and S. aureus Newman (B), abscess size was significantly smaller in dabigatran-treated mice compared with controls. The S. aureus Newman Δcoa−/−,ΔVWbp−/− mutant strain showed a reduced capacity to elicit abscess formation (B). Immunohistochemistry staining for fibrin/fibrinogen (C, representative images are shown). In both dabigatran-treated and control mice, we observed a fibrin/fibrinogen capsule around the abscess lesions (C, white arrows). Fibrin/fibrinogen staining in the center of the lesions appeared more intense in the control group (C, black arrows). The arrowheads show the dextran beads. The dotted boxes refer to the area shown in higher magnification.

Immunohistochemistry staining for fibrinogen showed a strong signal surrounding the abscess in both control- and dabigatran-treated animals (Fig. 5C, panels A and C, white arrows). In contrast to the dabigatran-treated animals, abscesses from control mice demonstrated a more pronounced staining at the center of the abscesses (Fig. 5C, black arrows).


The observation that the new thrombin inhibitor dabigatran also inhibits S. aureus coagulase activity [23] offered new opportunities to investigate the role of staphylothrombin in the virulence of S. aureus in in vitro and in vivo models.

Using the staphylocoagulase and VWbp mutant strains, we showed that staphylocoagulase contributed to a larger extent than VWbp to the total staphylothrombin activity. Dabigatran inhibited staphylothrombin activity of both thrombin-activating proteins.

We studied the role of staphylothrombin-dependent fibrin deposition on the attachment of S. aureus to polystyrene well plates, a classic assay for biofilm formation. We observed an increase in the bacterial association only in the combined presence of a coagulase (staphylocoagulase and/or vWpb), prothrombin and the substrate fibrinogen. The absence of either of these, or inhibition of staphylothrombin by dabigatran, abrogated this increased bacterial retention, whereas lepirudin, a thrombin inhibitor that does not inhibit staphylothrombin, had no effect.

Staphylococcus aureus expresses a high number of fibrin(ogen) binding proteins, including clumping factor A and B (ClfA and ClfB), fibronectin-binding protein A (FnBPA), extracellular adhesion protein (Eap), extracellular fibrinogen-binding protein (Efb) and bone sialoprotein binding protein (Bbp) [25–29]. Mutants lacking either FnBPA or ClfA showed a strongly decreased in vitro binding to immobilized fibrinogen, which was associated with a reduced infectivity. Conversely, gene transfer and the expression of S. aureus ClfA and FnBPA increased the virulence of non-fibrinogen binding bacterial strains [17,30,31]. We did not study the relative contribution of the different fibrin(ogen)-binding proteins to the increase in S. aureus and the role of passive entrapment of S. aureus in the fibrin matrix. However, in view of the reported efficiency and importance of S. aureus–fibrinogen interactions, the ability to generate a three-dimensional fibrin network may offer a physical substrate for attachment and growth, similar to bacterial biofilms, which play a role in the pathogenic potential of various bacteria, including S. aureus [32–34].

This embedding in a matrix of host proteins may have other benefits to the pathogen, such as shielding from the immune system. In the present experiments, staphylothrombin-mediated fibrin was able to reduce the bactericidal leukocyte activation after exposure to S. aureus. This attenuation of leukocyte activation was observed only for staphylocoagulase-positive strains, in the combined presence of fibrinogen and prothrombin, and was inhibited by dabigatran. Inhibition of fibrin polymerization by GPRP also restored the impaired leukocyte activation, suggesting that insoluble staphylothrombin-mediated fibrin is required for this effect.

We measured the release of MPO as a marker of leukocyte activation, which has a bactericidal effect by catalyzing the formation of hypochlorous acid [35]. Impaired leukocyte activation may translate into a survival benefit for S. aureus. Indeed, strains expressing either staphylocoagulase or VWbp demonstrated a higher survival in the presence of leukocytes compared with the double mutant when co-incubated in plasma, whereas the parent strain showed the highest survival. The high survival of the wild type could be fully restored by transformation of the double knockout with the plasmid carrying both staphylocoagulase and VWbp. Again, staphylothrombin inhibition by dabigatran restored the bactericidal effect of leukocytes and decreased bacterial survival to levels similar to that of the double knockout.

To study the role of staphylothrombin and its pharmacological inhibition in vivo, we used a mouse s.c. abscess model. Treatment with dabigatran resulted in plasma levels of dabigatran that were well above the Ki for murine staphylothrombin. Compared with controls, dabigatran-treated mice had significantly smaller abscesses when infected with the S. aureus Newman or ATCC 25923 strain. Dabigatran not only inhibits staphylothrombin but also murine thrombin, which may also be involved in local inflammatory reactions. However, the observation that the Newman Δcoa−/−,ΔVWbp−/− mutant, which is defective in coagulase activity, developed no or only very small lesions suggests that the development of s.c. abscesses indeed mainly depends on staphylothrombin activity and that the observed reduction in abscess size is mainly as a result of staphylothrombin inhibition rather than thrombin inhibition.

Our observations are in agreement with the previous findings of Cheng and coworkers. Using a mouse model of S. aureus sepsis, they demonstrated that fibrinogen, prothrombin and staphylocoagulase are present in and around S. aureus abscesses, and showed that an antibody-based inhibition of staphylocoagulase reduced sepsis mortality and peripheral abcedation [19]. They concluded that staphylothrombin-dependent fibrin deposition in a pseudocapsule in the center of the abscess may shield the bacteria from the immune system [36]. This is in accordance with our finding that compared with dabigatran-treated mice, control mice had stronger fibrinogen/fibrin staining in the center of the abscesses.

Recurrent infection and persistence of S. aureus in abscesses in spite of high concentrations of leukocytes are a typical feature of S. aureus disease. Various virulence factors of S. aureus have been identified that reduce the activation or the bactericidal activity of leukocytes [6,37]. Encapsulation of S. aureus by pathogen-induced fibrin deposition may be an additional mechanism by which S. aureus evades the host’s defense system. Increased local levels of fibrinogen and the local activation of the coagulation cascade are part of a non-specific innate immunity reaction to various noxious stimuli, such as trauma and infection, and the presence of fibrinogen and its interaction with leukocytes is required by the host for the successful clearance of bacteria [38]. It is interesting to note that in addition to procoagulant proteins, S. aureus secretes additional factors that specifically block this fibrin(ogen)–leukocyte interaction [39]. This may explain how S. aureus bypasses the control of these protective actions and rather uses them to promote its virulence.


In summary, the present data show that staphylothrombin activity is important for S. aureus colonization and infectivity. In addition to providing a physical substrate for the attachment of bacteria, staphylothrombin-mediated fibrin may help to shield bacteria from immune cells, reducing the activation of leukocytes and subsequently increasing bacterial survival. These effects were reversed by pharmacologically inhibiting staphylothrombin. These in vitro findings correlate with our in vivo finding that treatment with dabigatran reduced the abscess size in a subcutaneous infection model. The impact of pharmacological inhibition as an adjunctive therapeutic strategy in the treatment of S. aureus infections warrants further investigation.


This work was supported by a grant of the Research Foundation Flanders (FWO) and by a research grant from Boehringer-Ingelheim.

Disclosure of Conflict of Interests

J. van Ryn is an employee of Boehringer-Ingelheim, the other authors state that they have no conflict of interest.