Although the cutaneous form of Bacillus anthracis infection is most common worldwide, inhalational and gastrointestinal anthrax, as well as the recently recognized soft tissue infection in injection drug users, termed injectional anthrax, are more lethal [1,2,3]. Death in these latter three is almost always preceded by progressive hemodynamic instability and shock. Compared to more commonly encountered types of bacteria, shock with anthrax appears to be potentially more resistant to conventional types of hemodynamic support. In the 2001 US outbreak of anthrax, all patients who developed shock died despite aggressive intensive care unit support. While the number of patients involved in this outbreak was small, a mortality rate of 100% is much greater than typically reported for other types of treated bacterial shock [4]. Similarly, patients who developed shock in the recent UK anthrax outbreak among injection drug users have also been reported to be more difficult to manage with standard hemodynamic support [5].

While B. anthracis produces several components which may play an important role in shock, its two exotoxins, lethal toxin (LT) and edema toxin (ET), have received particular attention. Past and recent work directed at the cardiovascular effects of LT and ET provides important insights into the pathogenesis and potential management of shock with anthrax. Here, we review some of that work, and also briefly consider other components of B. anthracis that might contribute to shock.

Lethal and edema toxins are A-B type exotoxins composed of 2 proteins each, the A component being either lethal factor (LF) and/or edema factor (EF), respectively, and the B component, common to both, being protective antigen (PA). Following infection, PA released into the circulation binds to host cell surface receptors such as tumor endothelial marker 8 (TEM8) or capillary morphogenesis gene 2 (CMG2) [6,7,8,9]. Possibly most highly expressed in endothelial cells, both receptors have been demonstrated in a range of tissues including heart, lung, small intestine, spleen liver, kidney, skeletal muscle, and skin [10,11].

By itself, PA does not appear to have pathologic effects on the host [12]; its importance lies in its ability to facilitate intracellular delivery of the toxic anthrax factors (LF and EF). Upon binding to receptors, the 83 kD PA molecules undergo furin cleavage into 20 kD portions (freed into extracellular space), and 63 kD subunits that remain cell surface bound and combine to produce heptamers [13,14]. One to three LF and/or EF subunits competitively bind to these heptamers forming complexes, which undergo endocytosis and progressive acidification, after which EF and LF are released intracellularly [15].

Lethal factor is a zinc metalloprotease that inactivates mitogen-activated protein kinase kinases (MAPKK) 1–4, 6 and 7. While LF has been demonstrated to cause lysis of macrophages in vitro [6,16] and inhibit important host cell functions (e.g., innate and adaptive immunity and apoptosis), how it contributes to death associated with LT is still unclear. Edema factor is a calmodulin-dependent adenyl cyclase that increases intracellular cAMP concentration and impairs host defenses, including inhibition of phagocytosis [6,17,18]. It causes edema when injected subcutaneously into experimental animals [19]. Recent reviews as well as other contributions in this issue provide comprehensive overviews of the likely intracellular effects of LF and EF [16,18,20].

While older studies suggested that LT was the key virulence factor required for the lethal effects of B. anthracis, more recent studies have emphasized the potential importance of ET [12,21,22,23]. Both LT and ET are capable of producing cardiovascular effects leading to shock, and given their different actions, the combination of both toxins in active infection has potentially additive effects on increasing the severity of anthrax-associated shock.

Lethal toxin has been a focus of research since its original description as a bacterial component closely associated with death due to anthrax [24,25]. Shortly after the original description of LT in the 1950’s, a small number of studies in guinea pigs, rats and rhesus monkeys administered LT as a bolus, suggested that this toxin caused extravasation of fluid, hemoconcentration and shock [26,27,28,29]. However, based on limited measures of cardiac and pulmonary function, it was concluded that this shock was a terminal event related in part to respiratory failure [8]. Over the 30 years following these early investigations, no additional studies characterizing the cardiopulmonary effects of LT were reported. However, in contrast to findings from these early studies with LT, during the 2001 US anthrax outbreak, shock occurred well before death in non-survivors and did not initially appear to result from respiratory failure [1]. Similarly, shock has been a primary event noted in the UK anthrax outbreak in injection drug users [5].

In vivo studies from several laboratories over the past 5 to 10 years, while employing differing species, toxin challenges and methods of measure, together suggest that LT does indeed produce shock and organ injury directly, and that the underlying physiologic mechanisms are likely multifactorial. Based on the pattern of histological injury noted in mice challenged with LT as an intraperitoneal bolus, Moayeri et al. concluded that lethality with LT was related to tissue hypoxia such as would occur with organ hypoperfusion [30]. In rats with indwelling central venous and arterial catheters, we found that 24 h LT infusions produced progressive hypotension that was significantly greater in animals that later died [31]. These changes were associated with evidence of hemoconcentration (i.e., increased hemoglobin concentrations) and tissue hypoperfusion (i.e., increased lactate levels). In subsequent studies with sedated and mechanically ventilated canines with invasive hemodynamic monitoring, 24-h LT infusions also caused progressive shock and reductions in central venous pressure over the subsequent 72 h observation period (Figure 1) [23]. These findings are consistent with a potential effect of LT on the peripheral vasculature; several groups have reported that LT can disrupt endothelial barrier function. Stimulation of endothelial cell apoptosis, alteration of actin fibers and cadherins and mast cell activation are invoked as possible mechanisms causing endothelial dysfunction from in vitro experiments [32,33,34,35,36]. In addition, work in a zebrafish model has implicated LT disruption of MAPKKs in endothelial dysfunction [37,38]. The time course of endothelial changes from in vitro studies is consistent with the gradual onset of clinical changes noted in the canine model. Interestingly though, in rat and canine models, LT challenge was not associated with pleural effusions like those noted in humans with anthrax infection [1]. This finding suggests that other factors, possibly working in combination with LT, are necessary for this manifestation of active infection.

In addition to its potential effects on the peripheral vasculature, there is evidence to suggest that LT may directly depress myocardial function. We found that in sedated, mechanically ventilated canines, progressive hypotension occurred, and left ventricular ejection fraction (LVEF) measured with echocardiography also progressively decreased from 48 to 96 h after initiating a 24 h LT infusion in doses producing low (n = 6) or high (n = 9) lethality rates compared to a control group (n = 9) (Figure 1) [23]. Pulmonary artery occlusion pressure was not reduced, suggesting that decreases in preload were not the primary basis for hypotension or reduced LVEF. Heart rate did increase over time, and cardiac output was maintained (Figure 1). Other labs have also noted myocardial effects of LT. Watson et al assessed myocardial function with echocardiography in rats administered injections of LT producing hypotension and lethality [39,40]. They noted that 1–2 h after challenge, LT increased velocity of propagation and left ventricular diastolic and systolic areas [39]. In another report, the same authors noted that by 18 h after challenge, LT also increased left ventricular systolic area and decreased velocity of propagation, circumferential fiber shortening and left ventricular ejection fraction [40]. In a subsequent abstract and review, the group went on to describe an experiment employing pressure-volume loop measures in canines where, compared to controls, LT injection in 2 animals was associated with reduced stroke volume, end systolic pressure and ejection fraction, and increased left ventricular end diastolic pressure (LVEDP), in an overall pattern consistent with severe heart failure by 96 h after challenge [41]. Consistent with these findings, Moayeri et al. reported that intraperitoneal LT challenge in mice was associated with reduced LVEF and myocardial changes on electron microscopy as early as 6 h [42]. Although the mechanisms underlying the potential myocardial effects of LT are not clear, inhibition of MAPK ERK1/2 has been shown to produce stress-induced apoptosis and heart failure, whereas augmentation is protective [43,44,45]. Also, a recent in vitro study has suggested that LT may depress cardiomyocyte function via an NADPH oxidase-dependent mechanism [46].

Despite the cardiac effects of LT noted in in vivo models, using an isolated perfused rat heart system observed over a 4 h period we were only able to document a direct effect of LT on myocardial function with toxin doses substantially higher than those producing shock and lethality in vivo [22]. This, in combination with the in vivo studies, suggests that the effects of LT on myocardial function may take at the least several hours to develop. It is also possible that peripheral vascular changes and resulting hypotension may aggravate myocardial changes seen in in vivo models. In our canine model, although pulmonary arterial occlusion pressure (PAOP) was not reduced, central venous pressure (CVP) was, and fluid loading increased LVEF to some degree [23].

Finally, LT’s contribution to the development of shock may also involve its apparent immunosuppressive effects. Cardiovascular dysfunction during severe bacterial infection is thought to relate in part to the damaging effects of inflammatory mediators (e.g., cytokines, nitric oxide, oxygen free radicals) produced by the host during the innate immune response. One report suggested that macrophage damage by LT might produce increased circulating inflammatory cytokine levels, and that macrophage inhibition was beneficial with lethal doses of LT [48]. Subsequent studies however have not found a clear association between the cardiovascular dysfunction produced by LT and excessive inflammatory mediator production [8,12,30,31]. Such findings are consistent with LF’s known effect on suppressing stress kinase pathways, which are typically important contributors to the inflammatory response during infection [16,20,49]. In fact, we found in a rat model that pretreatment of animals with a sub-lethal dose of LT actually blunted the inflammatory response stimulated by either lipopolysaccharide (LPS) or intratracheal E. coli challenges (Figure 2) [47]. Notably however, while pretreatment with LT increased survival with LPS, it reduced survival with E. coli. Therefore, as has been suggested by other investigators, LT may also participate in the pathogenesis of shock and organ injury during anthrax infection by suppressing host microbial clearance and contributing to the very high bacterial loads noted in subjects dying with infection [49].

ET has received considerably less attention than LT since mutant deletion studies as well as direct comparisons of the two toxins suggested that the latter played a more central role in anthrax pathogenesis [50]. A study in rabbits suggested that LT may be produced in greater proportion than ET during active infection [51]. We reported that when administered as 24 h infusions, on a molar dose basis, LT was 5 to 20 fold more lethal than ET in both rats and canines [12,23]. In addition, efforts to generate large quantities of purified EF from B. anthracis cultures proved difficult, resulting in low yields and high levels of contaminating proteins [52]. Despite such data, there are still reasons to believe that ET plays an important role in the pathogenesis of shock during anthrax. Furthermore, recent developments allowing the generation of recombinant EF from cultures of E. coli have demonstrated consistently large quantities of >97% pure EF [53], mitigating previous concerns that the degree of purity of some edema factor preparations may skew the interpretation of cell culture or in vivo data.

From its first description and as its name implies, ET was noted for its ability to promote local extravasation of fluid [25]. Clinically, the notable extravascular fluid collections, such as the pleural effusions observed during the 2001 US outbreak, suggested that ET might play an important role in systemic infection [54]. Some patients in the recent UK outbreak also demonstrated striking soft tissue edema, not only locally at the sites of initial infection, but at more distant sites as well [5]. Finally, we noted that mortality from LT challenge increased in both rat and canine models when LT was co-administered with equimolar but nonlethal doses of ET [12,23].

While the mechanisms underlying shock with LT appear to be multifactorial and thus have been difficult to delineate, those underlying ET’s possible role may be easier to understand. ET has potent calmodulin dependent adenylcyclase activity, estimated to be 1000 times greater than that associated with eukaryotic adenylcyclases [17]. In vitro, ET increases intracellular cAMP to very high levels; cAMP is a key in mediator of vasodilation of both arteries and veins [55,56]. We noted rapid reductions in both central venous and arterial blood pressure in canines in patterns that persisted for up to 72 h after challenge with 24 h ET infusions, using doses causing high or low mortality (Figure 3) [23]. In a rat model, non-lethal and lethal ET infusions also produced rapid, persistent reductions in blood pressure [12]. Notably in both rat and canine studies, ET infusion was not associated with evidence of hemoconcentration (i.e., circulating hemoglobin concentrations were not increased) or extravascular fluid collections (e.g., pleural effusions or ascites), suggesting that reductions in central venous pressure (CVP) might in part represent relaxation of venous capacitance vessels. In both models, ET caused tachycardia that appeared to be out of proportion to other hemodynamic changes, but was consistent with increased cAMP in cardiac pacemaker cells (Figure 3) [57,58]. Echocardiography also demonstrated evidence of reduced preload in rats administered a rapid ET challenge by other investigators, although no extravasation of fluid was noted on lung pathology.

In a perfused rat heart model that is free of preload and after-load influences, we found that ET challenge, in addition to producing significant tachycardia, caused substantial increases in coronary flow rate consistent with a direct vasodilatory effect of the toxin [22]. Interestingly, in this isolated system, ET also had inotropic effects (Figure 4). While these effects were transient with higher ET doses, they persisted at lower ones. Consistent with its predominant mechanism of action, ET produced increases in both myocardial tissue and effluent cAMP levels (Figure 5). Adefovir, a nucleoside agent which has been shown to inhibit EF adenylcyclase activity in vitro [59], inhibited ET associated increases in heart rate (HR), coronary flow and cAMP levels (Figure 5). Given its effect on intracellular cAMP, ET may be a potential cause of the vasopressor resistant shock noted in patients in both the U.S. and U.K. outbreaks [5], since such agents can blunt the vasoconstrictive effects of agents like norepinephrine and phenylephrine.

Additionally, we noted in our canine model that even as shock developed, ET challenge resulted in ten-fold increases in urine output, along with decreased serum sodium (Figure 6) [23]. Since increased tubular cAMP stimulated by other types of toxins (e.g., cholera toxin) is associated with inappropriate sodium and water losses, it is possible that similar mechanisms may aggravate shock in patients with anthrax [60,61,62]. Of note, hyponatremia has been a frequently reported abnormality in patients presenting with anthrax [63]. Furthermore, in a murine model, ET was associated with adrenal gland necrosis, which could also aggravate sodium and water losses [21].

Like LT, ET may also have immunosuppressive effects. Challenge with ET alone does not stimulate the typical inflammatory mediator release noted with most types of bacterial infection [47]. Furthermore, ET has been shown to alter several different aspects of host defense, including reduced activation and function of antigen-presenting cells, increased release of cytokines from dendritic cells, and impaired chemotaxis and differentiation of T lymphocytes [49].

Although on a molar dose basis ET is less lethal than LT and the production of LT may be greater than ET during infection, the two toxins together may have synergistic effects [12,51]. In mouse, rat and canine models, nonlethal ET doses have been shown to add to the lethality of LT dose [12,23,64]. The basis for such synergy is not clear. ET has been shown to up-regulate the expression of PA receptors on macrophages and dendritic cells in vitro, thereby increasing the rate of toxin internalization [65]. ET has also been shown to potentiate the inhibitory effects of LT on T-cell and dendritic cell function and chemotaxis [66,67].

Although LT and ET have been the focus of research as pathogenic products of B. anthracis, other components of this organism such as bacterial cell wall and several metalloproteinases may also be important. In contrast to lethal or shock-inducing doses of LT or ET, challenge with live B. anthracis organisms does stimulate inflammatory mediator release [68,69,70,71]. As with other types of sepsis, this inflammation may contribute to shock and organ injury with anthrax.

There is evidence suggesting that B. anthracis cell wall may stimulate this inflammatory response. Human peripheral blood mononuclear cells, possibly via stimulation of TLR2/6 heterodimers, release TNFα, IL-1β and IL-6 in response to challenge with whole B. anthracis cell wall [72,73]. Anthrax infection in vitro also results in NOD-2 dependent IL-1β release [74]. We noted dose dependent increases in lethality, lactate, circulating inflammatory cytokine, chemokine and nitric oxide levels and thrombocytopenia when we challenged rats with whole anthrax cell wall [75]. Peptidoglycan appears to be the component of the cell wall largely responsible for its immuno-stimulatory effects, and in fact, peptidoglycan is shed by replicating bacteria [76,77]. This ability of anthrax cell wall to produce a robust intravascular inflammatory response and participate in the pathogenesis of shock could be important; patients and animals dying with anthrax have very high bacterial loads providing a reservoir of active cell wall constituents [73,77].

Besides LF, B. anthracis also produces proteases that may be important in anthrax infection. The delta Ames (pXO1⁻ and pXO2⁻) anthrax strain produced metalloproteases belonging to the M4 thermolysin and M9 bacterial collagenase families in culture studies [78]. Mice develop hemorrhagic tissue injury in response to purified preparations of these proteases. Moreover, animals challenged with Sterne strain spores showed better survival if treated with chemical inhibitors or immune serum against the M4 and M9 proteases of B. anthracis [78]. Additional studies of homologue proteases have implicated them in other pathophysiologic effects, including the interruption of tight junctions in Vibrio cholerae infections, and the cleavage of antitrypsin-1α, TNF-α and IL-2 on human T cell surfaces in Legionella infections [79,80].

Both LT and ET have the ability to produce significant cardiovascular dysfunction at either the peripheral vascular or myocardial levels, or both. The ability of ET to also potentially interfere with renal sodium retention could aggravate the direct cardiovascular effects of either toxin. Importantly, both ET and LT likely exert their cardiovascular effects via very different and potentially additive mechanisms; some in vivo data suggests that such additive effects are indeed the case [12,23]. Therefore, utilization of adjunctive agents with the ability to inhibit the two toxins together may very well be needed in patients progressing to shock despite appropriate antibiotic therapy and aggressive hemodynamic support. Historically, antibodies to individual components of the bacteria were used even before antibiotic therapy was available [81]. Two PA directed antibody preparations, one polyclonal (Anthrax Immune Globulin, Cangene Corporation, Winnipeg, ME, Canada) and one monoclonal (Raxibacumab, Human Genome Sciences, Inc., Rockville, MD), have been made available in recent outbreaks or added to the U. S. Strategic National Stockpile [82,83,84,85]. However, while in vitro or in vivo data has supported the effectiveness of these agents, their clinical efficacy in humans must be clarified. A variety of other types of inhibitors have also been proposed [8,86]. Just as important as determining the effectiveness of adjunctive therapies directed against LT and ET is defining how other components of the bacteria may contribute to cardiovascular dysfunction during infection.