Jay L. Degen

The blood-clotting protein fibrin(ogen) plays a critical role in host defense against invading pathogens, particularly against peritoneal infection by the Gram-positive microbe Staphylococcus aureus. Here, we tested the hypothesis that direct binding between fibrin(ogen) and S. aureus is a component of the primary host antimicrobial response mechanism and prevention of secondary microbe dissemination from the peritoneal cavity. To establish a model system, we showed that fibrinogen isolated from Fibγ^Δ5 mice, which express a mutant form lacking the final 5 amino acids of the fibrinogen γ chain (termed fibrinogenγ^Δ5), did not support S. aureus adherence when immobilized and clumping when in suspension. In contrast, purified wildtype fibrinogen supported robust adhesion and clumping that was largely dependent on S. aureus expression of the receptor clumping factor A (ClfA). Following peritoneal infection with S. aureus USA300, Fibγ^Δ5 mice displayed worse survival compared to WT mice coupled to reduced bacterial killing within the peritoneal cavity and increased dissemination of the microbes into circulation and distant organs. The failure of acute bacterial killing, but not enhanced dissemination, was partially recapitulated by mice infected with S. aureus USA300 lacking ClfA. Fibrin polymer formation and coagulation transglutaminase Factor XIII each contributed to killing of the microbes within the peritoneal cavity, but only elimination of polymer formation enhanced systemic dissemination. Host macrophage depletion or selective elimination of the fibrin(ogen) β2-integrin binding motif both compromised local bacterial killing and enhanced S. aureus systemic dissemination, suggesting fibrin polymer formation in and of itself was not sufficient to retain S. aureus within the peritoneal cavity. Collectively, these findings suggest that following peritoneal infection, the binding of S. aureus to stabilized fibrin matrices promotes a local, macrophage-mediated antimicrobial response essential for prevention of microbe dissemination and downstream host mortality.

Significance: A thrombin-fibrin pathway drives rapid and robust bacterial killing of Staphylococcus aureus following peritoneal infection. Elimination of coagulase activity results in S. aureus protection from antimicrobial activity and a corresponding reduction in host survival. These findings provide a molecular basis for the prevalence of coagulase negative staphylococci in peritonitis and highlight a caution when considering anticoagulant use in patients susceptible to peritoneal infection.

Abstract 103

Abstract 858 Multiple sclerosis (MS) is an inflammatory disease characterized by autoimmune demyelination of neurons and axonal damage within the central nervous system, leading to relapsing/remitting neurological deficits, including sensory and motor deficits. MS plaques are characterized by blood-brain barrier disruption leading to perivascular deposition of fibrin that correlates with areas of microglia activation and myelin damage. Consistent with multiple studies showing that hemostatic factors can serve as important modulators of the inflammatory response in vivo, fibrin(ogen) has been identified as a novel regulator of microglial activation and differentiation. Furthermore, fibrin(ogen) has been found, via the α_Mβ₂ binding motif on the γ chain, to play a role in microglial activation in experimental autoimmune encephalomyelitis (EAE), a murine model of MS. However, previous work by our lab, as well as others, suggests that thrombin-mediated proteolysis plays a greater role in driving neuroinflammatory disease than by merely supporting fibrin formation. Our working hypothesis is that thrombin-mediated proteolysis drives autoimmune neuroinflammation through both fibrin(ogen)-dependent and fibrin(ogen)-independent pathways. To test this hypothesis, we have initiated studies of EAE in mice lacking selected thrombin substrates, including protease activated receptors (PARs) and factor XIII (fXIII). Clinical evaluation of loss of motor function in EAE-challenged cohorts of PAR-1-deficient, PAR-4-deficient and WT control mice revealed a similar outcome in all three cohorts, suggesting PARs may have either a modest or a secondary role in EAE-induced neuroinflammatory disease. However, mice with a constitutive deficiency of another thrombin substrate, the fibrin cross-linking transglutaminase fXIII, were found to have a significant amelioration in motor function loss when challenged with EAE. To further explore this phenotype, we have utilized cystamine, a transglutaminase inhibitor, as a pharmacologic complement to the fXIII knockout mice. Mice with cystamine placed in their drinking water during the EAE challenge period also experienced significantly less disease than control mice that did not receive cystamine. This data suggests that (pro)thrombin contributes neuroinflammatory disease by both supporting local fibrinogen polymerization and by activation of fXIII leading to fibrin stabilization. Furthermore, this line of research suggests that more detailed studies of fXIII may reveal novel strategies for limiting or reversing the devastating pathologies associated with multiple sclerosis.

Abstract 2237 Chronic inflammation has been recognized as a major factor in the development and progression of multiple cancers. A prime example of this is the strong association between colitis and colon cancer. However, the specific factors that regulate disadvantageous immune processes in the context of inflammation-associated cancers remain poorly defined. Growing evidence suggests that hemostatic system components, traditionally associated with the maintenance of vascular integrity and prevention of blood loss, also directly regulate inflammatory processes. Furthermore, thrombin and several thrombin targets (e.g., PARs, fibrinogen, factor XIII) have been shown to regulate tumor cell proliferation and apoptosis, support metastasis, and protect tumor cells from innate immune surveillance mechanisms in other experimental contexts. A logical extension of these findings is the hypothesis that thrombin, as a master regulator of both inflammatory processes and tumor cell biology, is a major determinant of the progression of inflammation-driven cancers such as colitis-associated colon cancer (CAC). To test this hypothesis, we induced CAC in mice carrying prothrombin levels 50% of normal (fII+/−) and WT mice in parallel using an established two step protocol consisting of azoxymethane (AOM) and dextran sodium sulfate (DSS) exposure. The modest diminution in prothrombin levels imposed by the fII+/− genotype resulted in a dramatic diminution in the number of colonic adenomas formed after AOM/DSS challenge relative to WT mice. In order to determine if the diminution in adenoma formation observed in fII+/− mice was coupled to thrombin function, wildtype mice challenged with AOM/DSS were treated with daily i.p. injections of hirudin, a direct thrombin inhibitor, or saline carrier. Similar to the finding in mice with a genetically-imposed diminution in circulating prothrombin, hirudin treatment significantly blunted adenoma formation. To determine if reduction of thrombin generation improved the inflammation preceding the development of colonic adenomas, we used a novel, highly-specific factor XI antisense oligonucleotide “gapmer” (ISIS Pharmaceuticals) to inhibit hepatic factor XI synthesis prior to DSS challenge. Gapmer-mediated diminution of fXI levels to ∼15% of normal resulted in a dramatic improvement in colitis related symptoms. Gapmer-treated mice had less intestinal bleeding and weight loss associated with DSS challenge relative to mice treated with a control oligonucleotide. Consistent with these gross observations, microscopic analyses of colonic tissue showed that fXI gapmer treatment significantly limited mucosal ulceration. Factor XI gapmer treatment also significantly diminished local levels of several inflammatory cytokines known to play a role in colon cancer progression (i.e., IL-6, IL-1β, IL-12). These results demonstrate that thrombin is a crucial driver of the pathogenesis of colitis-associated colon cancer and suggest that therapies directed at thrombin or thrombin generation could treat or prevent inflammation-driven colon cancer. As pathological inflammation has been estimated to account for as many as 1 in 5 cancer-related deaths, thrombin-directed therapies could have broad applicability to multiple malignancies.

Abstract 821 Staphylococcus aureus is a common and re-emerging gram-positive bacteria that is a leading cause of both community and hospital-acquired infections. Multiple bacteria-derived products appear to drive pathogenicity, but some of the most potent virulence factors identified are those that engage host hemostatic system components. To directly investigate the hypothesis that host prothrombin and key thrombin substrates significantly contribute to S. aureus virulence in the context of acute bacteremia, we evaluated the survival profile of mice with either genetically-imposed deficiencies or functional alterations in prothrombin and thrombin targets following intravenous infection with S. aureus. Mice expressing 10% of normal levels of prothrombin (termed fII^flox/- mice) were found to have a major survival advantage following intravenous infection with ∼2.5 × 10⁸ colony forming units (cfu) of wild-type Newman-strain S. aureus. This difference in survival did not appear to be tied to platelet function as mice that were either pharmacologically depleted of platelets or deficient in the platelet integrin receptor subunit α_IIb did not display a survival advantage when similarly challenged with intravenous S. aureus. Fibrinogen was found to be at least one thrombin substrate critical to supporting S. aureus virulence, as the genetic elimination of fibrinogen resulted in dramatically prolonged survival following acute S. aureus bacteremia relative to fibrinogen-sufficient animals. One mechanism driving this prolongation in survival was linked to S. aureus binding of host fibrinogen via the bacterial fibrinogen receptor, clumping factor A (ClfA). Mice expressing a mutant form of fibrinogen that retains normal clotting function but lacks the binding motif for ClfA located on the fibrinogen g chain (termed Fibγ^Δ5) exhibited a profound survival advantage to wild-type animals over a wide bacterial dose range (e.g., LD₅₀ to an LD₁₀₀ for control animals). Additional studies revealed that the survival advantage in Fibγ^Δ5 mice correlated with reduced bacterial burdens in the hearts and kidneys, diminished histological evidence of cardiac damage, and significantly lower levels of plasma markers of cardiac damage (i.e., lactose dehydrogenase and cardiac troponin I) relative to infected control mice. To determine whether the observed survival advantages for fII^flox/- and Fibγ^Δ5 mice were strictly linked to ClfA, identical survival studies were performed with ClfA-deficient S. aureus. Consistent with the concept that ClfA is a potent virulence factor, mice of each genotype survived longer after infection with ClfA-deficient S. aureus than with the same dose of ClfA-positive bacteria. However, the survival advantage observed in Fibγ^Δ5 mice over wild-type animals was largely, but not entirely eliminated when challenged with i.v. ClfA-deficient S. aureus. The fact that Fibγ^Δ5 mice continued to display a modest, but significant, survival advantage relative to control animals when infected with bacteria lacking ClfA suggests that the loss of C-terminal residues of the gamma chain provides an additional, but still unidentified, benefit to the host. These data indicate that the host prothrombin and fibrinogen as well as the bacterial fibrinogen receptor ClfA are, cooperatively, major determinants of S. aureus virulence in the context of bacteremia, and that strategies designed to disrupt S. aureus engagement of these proteins may limit disease progression and improve host survival in S. aureus bacteremia/sepsis.

Abstract 3188 Poster Board III-125 The pervasive gram-positive bacterial pathogen Staphylococcus aureus has evolved and maintained a number of proteins that directly engage the host hemostatic system, including factors that directly interact with the host coagulation factor fibrinogen. Here, we investigate the hypothesis that binding of host fibrin(ogen) through the bacterial fibrinogen receptor, clumping factor A (ClfA), significantly contributes to virulence in the context of acute bacteremia. To directly test this view, we compared survival following intravenous injection of ClfA-postive S. aureus into fibrinogen-deficient and control mice. The genetic elimination of fibrinogen resulted in a dramatic prolongation in survival relative to fibrinogen sufficient control animals. To determine if this prolongation in survival was related to the ability of the pathogen to directly bind fibrinogen, we performed identical survival studies in mice expressing a mutant form of fibrinogen that retains normal clotting function but lacks the binding motif for ClfA located on the fibrinogen γ chain (termed Fib γ^Δ5). A profound survival advantage was observed for Fib γ^Δ5 mice relative to wild-type animals over a wide bacterial dose range corresponding to an LD₅₀ to an LD₁₀₀ for control animals. This difference in survival appeared to be linked to ClfA, as control and Fib γ^Δ5 mice revealed an identical survival profile when infected with ClfA-deficient S. aureus. Analysis of bacterial burden after infection in various organ systems revealed that ClfA-positive S. aureus rapidly marginalized from the circulation and colonized target organs in mice of both genotypes. The highest numbers of colony forming units (cfu) were found in the heart and kidney of challenged mice, but the cfu values observed in both of these tissues were significantly reduced in Fib γ^Δ5 mice, relative to control animals at 24 and 48 hours after initial infection. The cfu values also tended to be reduced in Fib γ^Δ5 mice over wild-type controls in the liver, lung, spleen and blood, but this difference did not achieve statistical significance. Complementary studies revealed a comparable systemic inflammatory response in control and Fib γ^Δ5 animals following initial infection; plasma levels of Il-6 and fibrinogen (an established acute phase reactant) were similarly elevated in both genotypes following infection. Notably, increased histological evidence of cardiac damage secondary to S. aureus bacteremia in wild-type mice relative to Fib γ^Δ5 animals correlated with significantly higher plasma levels of markers of cardiac damage (i.e., lactose dehydrogenase and muscle creatine kinase enzyme activity) 48 hours after infection. These data suggest that bacterial engagement of host fibrinogen via the bacterial fibrinogen receptor ClfA significantly contributes to S. aureus virulence following intravenous infection. Strategies designed to disrupt this interaction may be of significant clinical benefit to limit disease progression in S. aureus sepsis and could be achieved without necessarily compromising host hemostatic function.

Abstract 4021 Poster Board III-957 TGF-β1 is a multifunctional cytokine with profound effects on several biological processes, including malignancy, immunity, wound repair, and tissue fibrosis. In addition to local generation of TGF-β1, it also circulates in plasma where it has the potential to produce systemic effects. The source(s) of plasma TGF-β1 have not, however, been defined, and a very wide range of normal values have been reported (∼1.0 to 50.0 ng/mL). Since platelets contain 40-100 times more TGF-β1 than other cells, release of platelet TGF-β1 during blood drawing and/or sample preparation can influence the results. Citrated plasma prepared from blood samples obtained from C57Bl/6 mice by the retrobulbar capillary technique showed evidence of variable release of the platelet-specific α granule proteins platelet factor 4 (PF4) and thrombospondin-1 (TSP-1), and the degree of release correlated with the total TGF-β1 levels measured by an ELISA (Table 1). Adding PGE₁ to the anticoagulant reduced but did not completely eliminate the release of platelet granule proteins. In contrast, plasma prepared from free-flowing blood obtained using ultrasound guidance to percutaneously enter the left ventricle inflow tract with a 27 gauge needle showed minimal or no evidence of release of platelet granule proteins and had a total TGF-β1 level of 2.1 ± 1.2 ng/mL. To more firmly assess the role of platelets in circulating TGF-β1, we measured the platelet, plasma, and serum TGF-β1 levels in mice with a megakaryocyte-specific conditional knockout of TGF-β1 (achieved by crossing mice carrying a floxed TGF-β1 allele with transgenic mice expressing a Cre-recombinase under the control of a PF4 promoter) and littermate control mice (WT; WT/ PF4Cre⁺; TGF-β1^flox/flox/PF4Cre^-). Compared to the controls, TGF-β1^flox/flox/PF4Cre⁺ mice exhibited an ∼80 to 90% reduction of TGF-β1 in both platelets and serum as well as an ∼45% reduction in plasma TGF-β1 (Table 2). As a complementary test of the contribution of platelets to plasma TGF-β1, we generated profoundly thrombocytopenic mice by i.v. injection of the hamster anti-mouse αIIbβ3 mAb 1B5 (1.0 mg/kg). 24 hours later the platelet count decreased by >98% and plasma TGF-β1 levels in these thrombocytopenic animals were reduced by more than 70% (0.6 ± 0.2 ng/mL) compared to pretreatment values (3.2 ± 0.4 ng/mL; p=0.002) or values after giving saline or control polyclonal hamster IgG (3.2 ± 2.0 ng/mL; p=0.002). We conclude that platelets are the major source of plasma TGF-β1 in mice. In addition, based on the rapid decrease of plasma TGF-β1 levels after the induction of thrombocytopenia, we infer that plasma TGF-β1 has a relatively short survival time. Our data have potential implications for the systemic contribution of platelet-derived TGF-β1 in regulating various normal and disease states, including wound healing and organ fibrosis.

The conversion of soluble fibrinogen to an insoluble fibrin matrix following thrombin cleavage of the Aα and Bβ chains is critical to hemostasis. However, even soluble fibrinogen holds potential biological significance apart from fibrin polymer formation; for example, fibrinogen contributes significantly to overall blood rheology and can engage a number of integrins (e.g., αIIbβ3), enzymes (e.g., fXIII) and matrix proteins (e.g. fibronectin). To provide a system to study mice carrying fibrinogen but no capacity for fibrin formation, we employed a gene-targeting strategy to eliminate the Aα chain thrombin cleavage site and replace it with a sequence recognized by the highly-selective alternate protease, enterokinase (P6-P1: ADDDDK). Based on the prevailing view that thrombin cleavage of the Aα chain is essential to polymer formation, we hypothesized that this mutant fibrinogen, termed FibEK, would be locked in the soluble form in vivo, but would be readily clotted in vitro by exogenous enterokinase. The FibEK allele was transmitted through the germline and supported the expected level of mRNA expression and plasma protein production. Like previously established fibrinogen-null mice, FibEK mice uniformly developed to term. However, survival beyond the perinatal period was vastly superior in mice carrying fibrinogen-EK relative to fibrinogen-null animals in the same (C57Bl/6) genetic background; ~90% of FibEK mice survive to adulthood, whereas less than 30% of fibrinogen-null mice survive the perinatal period. Consistent with our initial hypothesis, FibEK plasma was found to be unclottable following the addition of excess exogenous murine thrombin at 37°C and complementary fibrinopeptide release assays by HPLC revealed that murine thrombin is completely incapable of releasing fibrinopeptide A (FpA) from fibrinogen-EK. Murine thrombin did support quantitative FpB release from fibrinogen-EK, albeit at a slower rate than wild-type fibrinogen. No thrombin-induced fibrin polymer formation could be appreciated by standard turbidity assays in incubation mixtures containing FibEK plasma at 37°C, even after 24 hrs. In contrast, wild-type plasma supported rapid polymer formation, with turbidity peaking within minutes of thrombin addition. Interestingly, despite the complete absence of FpA release, both plasma and purified fibrinogen-EK supported the formation of small and irregular clots (characterized by thin fibrils in scanning EM) following long incubations at reduced temperatures (i.e., 24 hr at 22 °C). Unlike fibrinogen-null mice, platelet-rich plasma prepared from FibEK mice supported robust ADP-induced platelet aggregation. However, FibEK animals essentially phenocopy fibrinogen-null mice in their inability to efficiently clear the microbial pathogen S. aureus. Following intraperitoneal infection with 10⁹ CFU of S. aureus, control mice successfully cleared >99% of the bacteria within 1 hour, whereas the number of bacteria retrieved within peritoneal lavage fluid of both FibEK and FibAα−/− mice remained similar to the original input CFU. In summary, FibEK mice provide a unique opportunity to further explore the biochemistry of thrombin-mediated clot formation in vitro and in vivo. These animals will also be a vital tool in defining the specific contribution of fibrin matrices to broad array of physiological and pathological processes in vivo.

Thrombin is central in thrombus formation as both a positive mediator of thrombus formation through the proteolytic activation of PARs, fibrinogen, fXI and other prothrombotic substrates, and a negative modulator of the coagulation cascade through the activation of protein C. Detailed structure-function studies have revealed that thrombin can be redesigned to favor either procoagulant or anticoagulant substrates. The introduction of W215A/E217A substitutions in the murine thrombin active site (fIIWE) results in a pronounced “specificity switch” that reduces catalytic efficiency with fibrinogen by at least 3-orders-magnitude while only modestly reducing activity for protein C activation. To evaluate the effects of fIIWE activity in vivo, we have used a gene-targeting strategy to generate mice carrying the W215A/E217A mutations in the endogenous murine prothrombin gene. The mutant allele was transmitted through the germline and was found to support the expression of normal levels of hepatic fII mRNA and plasma fII in both heterozygous and homozygous neonates. Unlike fII knockout mice, homozygous fIIWE mice were observed at term with the expected Mendelian frequency. Nevertheless, homozygous fIIWE offspring uniformly succumbed to spontaneous bleeding events within days of birth. Heterozygous fIIWE/WT animals generally survived to adulthood, were capable of carrying multiple liters to term, and unchallenged mice displayed a hematological profile similar to wildtype mice. However, consistent with a predicted anticoagulant phenotype, adult fIIWE/WT heterozygotes exhibited significantly delayed thrombus formation following ferric chloride injury of mesenteric arterioles and extended bleeding times following tail tip excision relative to control mice expressing wildtype fII. Given that activated protein C has been shown to be efficacious in the treatment of sepsis, we explored whether the shift in thrombin specificity in heterozygous fIIWE/WT mice would confer the benefit of rendering animals tolerant to acute septic challenges. Kaplan-Meier analyses following intravenous administration of S. aureus revealed that fIIWE/WT mice exhibited a significant survival advantage over littermate wildtype animals challenged in parallel and tracked over a 7-day observation period. Notably, extended thrombus formation and bleeding times as well as resistance to sepsis was not simply a function of half normal wildtype fII expression. When these analyses were performed in animals carrying one wildtype allele and one null mutation allele, results were similar to wiltype. These studies further underscore the interplay between the hemostatic and inflammatory systems in vivo and highlight the possible therapeutic utility of recombinant (pro)thrombin derivatives with selected alterations in substrate specificity.