Plasminogen Is a Critical Host Pathogenicity Factor for Group A Streptococcal Infection

See allHide authors and affiliations

Science  27 Aug 2004:
Vol. 305, Issue 5688, pp. 1283-1286
DOI: 10.1126/science.1101245


Group A streptococci, a common human pathogen, secrete streptokinase, which activates the host's blood clot–dissolving protein, plasminogen. Streptokinase is highly specific for human plasminogen, exhibiting little or no activity against other mammalian species, including mouse. Here, a transgene expressing human plasminogen markedly increased mortality in mice infected with streptococci, and this susceptibility was dependent on bacterial streptokinase expression. Thus, streptokinase is a key pathogenicity factor and the primary determinant of host species specificity for group A streptococcal infection. In addition, local fibrin clot formation may be implicated in host defense against microbial pathogens.

Plasmin, the major serine protease that degrades the fibrin blood clot, is generated through cleavage of the proenzyme plasminogen (PLG) by the plasminogen activators urokinase (uPA) and tissue-type plasminogen activator (tPA). Human PLG can also be activated by streptokinase (SK) secreted by group A streptococci (GAS), leading to the clinical use of SK for therapeutic fibrinolysis of pathologic thrombi associated with myocardial infarction, stroke, and venous thrombosis (1). PLG deficiency results in the rare genetic disorder ligneous conjunctivitis, characterized by thick fibrinous pseudomembranous deposits on the conjunctiva and other mucous membranes (2). PLG null mice develop spontaneous fibrin deposition in multiple tissues, leading to retarded growth, rectal prolapse, and decreased survival (3, 4). Plg null mice also exhibit eye findings similar to those for the human disease (5). In addition to GAS, a number of other human pathogens also express a plasminogen activator, including staphylokinase of Staphylococcus aureus (6) and the Pla protein of Yersinia pestis, the agent responsible for the plague (7). Other microbes, such as Borrelia burgdorferi, use a host plasminogen activator as a pathogenicity factor (8, 9). GAS is the most common cause of clinically significant bacterial pharyngeal and skin infections. A subset of these infections are highly invasive into soft tissues or organs and result in rapidly progressive, life-threatening conditions, such as necrotizing fasciitis (10). Like many microbial pathogens, GAS is highly host-specific and naturally only infects humans. This restricted host specificity is correlated with the limited ability of many candidate pathogenicity factors, including SK and surface receptors for PLG, to interact with ligands or protein substrates from species other than humans (1116). Studies of streptococcal pathogenicity and the role of SK and surface-bound PLG-binding proteins have been hampered by this limitation, as well as by species-specificity in the interactions among host fibrinolytic system components, including tPA and PLG (17).

We generated a “humanized” transgenic mouse expressing human PLG under control of the mouse albumin gene regulatory sequences within a bacteria artificial chromosome (BAC) transgene [(18) and fig. S1A]. Two independent founder lines, AlbPLG1 and AlbPLG2, directed highly liver-specific PLG expression (fig. S1B). The level of PLG detected in the plasma of AlbPLG1 corresponded to ∼17% of the PLG level in control human plasma (16.7 ± 1.78) with a much lower level (<1%) for AlbPGL2. Consistent with previous reports (11, 15), human PLG derived from the transgene also exhibited increased susceptibility to activation by human tPA and SK compared with mouse PLG (fig. S1, C and D) (17, 19).

Mice that are Plg–/– are viable but exhibit retarded growth, diffuse fibrin deposition in multiple tissues, and frequent rectal prolapse (3, 4). Introduction of the human PLG transgene resulted in reversal of the widespread tissue fibrin deposition observed in the livers of Plg–/– mice and near-complete rescue of the weight loss and rectal prolapse phenotypes (Fig. 1, A to D, and fig. S2). The fibrinolytic system is thought to play a critical role in controlling the thrombotic complications associated with bacterial sepsis (20), and mice deficient in α2-antiplasmin exhibit enhanced fibrin clearance after endotoxin [lipopolysaccharide (LPS)] administration (21). We also tested the response of Plg–/– mice to LPS (Fig. 1E). Here too, human PLG directed by the AlbPLG1 transgene could substitute for the orthologous mouse protein and rescued the Plg–/– mice from markedly reduced survival.

Fig. 1.

Rescue of Plg–/– phenotype by human PLG transgene. (A to C) Liver sections immunostained for fibrin(ogen). Scale bar, 100 μM. (D) Hepatic deposits of fibrin(ogen) per high power field (200×) were quantified by blinded analysis of liver samples from three Plg+ (Plg+/+ or Plg+/–), three Plg–/–, and five Tg+Plg–/– mice at 10 weeks of age. (E) Survival rate of Plg+, Plg–/–, and Tg+Plg–/– mice 3 days after intraperitoneal injection of 13.5 μg LPS per gram body weight. **P < 3 × 10–5.

Mice are generally highly resistant to skin infection by most human pathogenic GAS, with intradermal injection of 108 CFU of one such strain resulting in only ∼20% mortality at 6 days in outbred CD1 mice (19). The GAS strain UMAA2616 exhibits enhanced virulence in a mouse skin infection model because of site-directed mutation in the regulatory locus, csrRS (22). Using this strain at a dose of 2 × 105 CFU, we observed 20% mortality at 9 days in control littermate mice on the C57BL/6J background. However, introduction of human PLG expressed by the AlbPLG1 transgene markedly increased mortality to 75% (Fig. 2A). MGAS166, the wild-type clinical isolate from which the hyper-virulent UMAA2616 strain was derived, caused no mortality at 10 days in control mice injected with up to 7 × 106 CFU and only 30% mortality at 7 days after a dose of 3 × 107 CFU. Corresponding mortalities in Tg+ mice ranged from 35 to 100% (Fig. 2, B to D). Thus, human PLG plays a critical role in the pathogenicity of these human GAS isolates in this mouse model, a function that is not provided by mouse PLG. The AlbPLG2 transgene directs only very low levels of human PLG expression (<1%). Although a trend to increased mortality was seen in animals carrying the AlbPLG2 transgene, this effect was not statistically significant (Fig. 2E). Thus, there appears to be a minimal threshold level of human PLG that is required for GAS pathogenicity.

Fig. 2.

Increased susceptibility to GAS conferred by human PLG. Survival curves for Tg mice are indicated by dashed lines and for Tg+ mice by solid lines. P values were determined by log-rank test. (A) Survival curve of AlbPLG1Tg and Tg+ mice after infection with 2 × 105 CFU UMAA2616. Data are pooled from five independent injections with at least five mice per group for each experiment (a total of 42 Tg+ and 56 Tg mice). (B to D) Survival curve of Tg and Tg+ mice after inoculation with strain MGAS166 at three different doses, 3 × 105, 7 × 106, and 3 × 107 CFU. Each group contains at least six mice. (E) Survival curve of AlbPLG2 Tg and Tg+ mice after inoculation with 2 × 105 CFU UMAA2616. Data are pooled from two independent experiments. Each group consisted of 6 to 10 mice (a total of 16 Tg+ mice and 13 Tg mice). (F) Survival curve of AlbPLG1 Tg and Tg+ mice after inoculation with the SK-deficient strain 2 × 105 CFU UMAA2641. Data are pooled from two independent experiments. Each group consisted of 4 to 6 mice (a total of 12 Tg+ mice and 10 Tg mice). (G) Survival curves of AlbPLG1 Tg and Tg+ mice after subcutaneous injection with the 108 CFU PAM-expressing strain AP53 and an isogenic derivative of the AP53 strain deficient in PAM expression. Data are pooled from two independent experiments (with a total of 12 mice in each group). (H) Survival curves of AlbPLG1Tg and Tg+ mice after infection intravenously with 2 × 106 CFU UMAA2616. Data are pooled fromthree independent experiments with 5 mice per group for each experiment (a total of 15 Tg+ and 15 Tg mice).

Enhanced virulence is observed in GAS skin infection in the mouse after preincubation with human plasma or coinjection with purified human PLG (14). Increased invasiveness is also seen when GAS is preincubated with SK and human fibrinogen and PLG in vitro (19). Species specificity has also been demonstrated for PLG receptors on the bacterial surface and for interactions with fibrinogen (11-13, 15). To test the relative contribution of SK to the overall host species specificity of human pathogenic streptococci, we inoculated animals with an isogenic streptococcal strain deleted for the SK gene (ska, UMAA2641) (14). The increased susceptibility of Tg+ mice to the wild-type GAS isolate UMAA2616 was largely abrogated by deletion of ska. The 75% mortality observed at 9 days with strain UMAA2616 (Fig. 2A) contrasted with the 27% mortality using its ska-deleted derivative (Fig. 2F). Although not statistically significant (P > 0.2), a trend toward increased mortality in Tg+ mice infected with the SK strain compared with their Tg littermate controls is consistent with an additional contribution of other host species–specific PLG-dependent factors.

We next examined the potential role of the bacterial surface protein (PAM) that is expressed by a subset of GAS associated with skin infections. PAM specifically binds human PLG or plasmin with high affinity, although exhibiting considerably weaker binding to mouse PLG (12). The PAM-expressing GAS strain AP53 produced low levels of SK and exhibited low virulence in Tg mice (∼20% mortality) (Fig. 2G). In contrast, Tg+ mice were much more susceptible to the AP53 strain, with mortality increased to ∼80%. An isogenic derivative of the AP53 strain deficient in PAM expression, and thus unable to bind PLG, showed minimal virulence in both Tg+ and wild-type mice, which suggested that the amount of SK secreted by this strain was not sufficient to allow bacterial dissemination in the absence of PAM. Thus, the ability to focus PLG at the bacterial surface provides an additional mechanism whereby GAS, and potentially other pathogens, can exploit the host fibrinolytic system to facilitate establishment of infection and subsequent invasion.

A number of mechanisms have been proposed to explain how the recruitment and activation of PLG by an invasive bacterial pathogen might enhance virulence (16). It is likely that the host inflammatory response to bacterial infection results in local thrombosis and microvascular occlusion. This response may serve to seal off the infectious nidus and to prevent systemic spread. We hypothesized that GAS hijacks the host fibrinolytic system in order to circumvent this local defense and to reopen the vascular tree to systemic spread. Consistent with this model, we observed an increase in bacterial colonies in the spleens of Tg+ mice injected with 2 to 3 × 107 CFU MGAS166 (Fig. 3A) or 108 CFU AP53 (Fig. 3B) compared with Tg controls, presumably representing increased systemic spread. This enhanced dissemination may also explain the observed increase in mortality in these animals (Fig. 2. A to D, and G).

Fig. 3.

Fibrin as a barrier to GAS dissemination. (A) Spleen and skin lesions were harvested at 64 hours and bacterial colonies quantified. Each data point represents one mouse. Although potentially partially masked by increased mortality in Tg+ mice, a significant difference in mean bacterial colonies (CFU) in the spleen was still observed between Tg+ (◼) and Tg (⚫) mice (P values determined by Mann-Whitney test). (B) Similar data were obtained for Tg (circles) and Tg+ (squares) mice infected with strain AP53 (filled symbols) or the isogenic PAM mutant (open symbols). Spleens were dissected after 24 hours, and the number of CFUs was determined. Enhanced splenic spread is observed in Tg+ mice infected with the PAM+ strain compared with the PAM strain. Greater levels of spread were also observed in Tg+ mice compared with Tg mice for both the PAM+ and PAM strains (P values determined by Mann-Whitney test). (C) Representative hematoxylin and eosin–stained sections (left) from the injection site, 64 hours after inoculation with 2 to 3 × 107 CFU UGAS166. An acutely inflamed dermis (id) is evident, along with about half of a bacteria-induced abscess (ab) surrounded by a rimlike infiltrate of neutrophils (arrows), and with acute inflammation and central necrosis of the overlying epidermis (ep) (outlined with arrowheads) in both Tg+ and Tg mice. Scale bar: 100 μM. (D) Survival curve of Ancrod-treated (solid line) and control mice (dashed line) after injection with 9 × 106 CFU of strain MGAS166 (six mice per group).

If GAS-associated fibrinolysis is required primarily to facilitate bacterial access to the vasculature, then pathogenic GAS introduced via the intravenous route should no longer depend on human PLG function. Indeed, no difference in mortality was observed between Tg+ and Tg mice injected intravenously with 2 × 106 CFU of UMAA2616 (Fig. 2H), in contrast to the marked difference in mortalities seen when the same genotypes were injected subcutaneously (Fig. 2A).

Although these data suggest that GAS-induced PLG activation is required to overcome local microvascular occlusion, we observed no difference in bacterial colony counts from homogenized local skin lesions (Fig. 3A), and histological analysis of skin surrounding the inoculation site failed to identify an obvious difference in fibrin deposition or vascular occlusion between Tg+ and Tg animals (Fig. 3C and fig. S2C). Thus, bacterial-induced fibrin dissolution may be transient, or localized to specific invasive sites, and not evident on gross histologic examination.

To further explore the role of fibrin deposition in host defense against the dissemination of SK-expressing streptococcal pathogens, we examined mortality after GAS injection in C57BL/6J mice treated with the snake venom Ancrod, which proteolytically degrades plasma fibrinogen (23). Ancrod-treated animals demonstrated an approximately 60% reduction in plasma fibrinogen. This decrease in fibrinogen was also associated with a marked increase in mortality after a subcutaneous inoculum of 9 × 106 CFU MGAS166, from 30% at 6 days in control animals to 100% in Ancrod-treated mice (Fig. 3D).

Thus, activation of host PLG by SK leads to accelerated clearance of host fibrin and is a central mechanism for GAS invasion and spread. It is likely that similar interactions are central to the invasive program of other unrelated, plasminogen activator–associated pathogens that occupy diverse microenvironmental niches, such as S. aureus, B. burgdorferi, and Y. pestis (69). The remarkable species specificity of SK for host PLG probably resulted from host and pathogen coevolution (15). Similar coevolution probably also explains the recently reported specificity of the Neisseria meningitides for human CD46 (24).

Our findings raise the possibility that polymorphic variation in the level of plasminogen, and potentially other fibrinolytic components, within human populations could represent a significant susceptibility factor for bacterial infection. Another virulence determinant was recently shown to form complexes with fibrinogen that induce vascular leakage, potentially enhancing the severity of GAS infection (25). These observations highlight the potential role of infectious disease as a critical force in the evolution of the hemostatic system and the unusual species specificity of many coagulation factor interactions.

Supporting Online Material

Materials and Methods

Figs. S1 and S2

Table S1


References and Notes

View Abstract

Navigate This Article