5.4.1. Macrophages

From the histopathological observations outlined in sections 5.1 to 5.3, it would appear that phagocytosis plays a more important role in the dissemination of B. anthracis after entry into the body by aerosol, as compared with cutaneous (and, presumably, alimentary tract) exposure. The possible involvement of macrophages in the uptake of B. anthracis in the intestine was alluded to in section 5.3.

It seems clear that spores lodged in the alveoli following inhalation are phagocytosed by, and germinate within, alveolar macrophages as they move to the lymph nodes (Ross, 1957; Guidi-Rontani et al., 1999; Dixon et al., 2000). As reviewed by Guidi-Rontani & Mock (2002), phagocytosis is mediated by receptors for the Fc portion of IgG. The ability of the emergent vegetative cells to multiply has been demonstrated in cultured macrophages (Dixon et al., 2000) but seemingly not all agree that germinated spores multiply within macrophages (Moayeri & Leppla, 2004). Guidi-Rontani & Mock (2002) postulate that B. anthracis spores may possess a unique system for detecting specific germinants within the macrophage, although putative germinants remain to be identified. As reviewed by Moayeri & Leppla (2004), lethal and oedema toxins (section 5.5.3), expressed at the spore stage and by newly germinated spores, play an early role in survival of the germinated cell within the macrophage and apparently contribute to its death.

In addition to the role of macrophages in facilitating spore germination and possibly outgrowth, various laboratories have documented a defensive role for macrophages in protecting the host from infection. Guidi-Rontani et al. (2001) showed that, although spores could germinate in macrophages, growth of the germinated spore did not occur in their in vitro system. Welkos et al. (2002) and Cote et al. (2004) observed an apparent protective sporicidal activity by macrophages in vitro and in vivo. Piris-Giminez et al. (2004) described the bacterial activity of a macrophage phospholipase against B. anthracis. The roles that spore dose, mode of infection and other factors play in the relative extent to which macrophages abet and facilitate the infection of the host versus protecting the host from infection require further study. The in vitro macrophage systems used to date are of necessity artificial; studies that track infection in real time in vivo are needed.

The involvement of phagocytosis in anthrax acquired through the epithelium is less clear at present. The role of the capsule of B. anthracis (see section 5.5.1) has long been regarded as being to protect the bacterium from phagocytosis (Sterne, 1959) and the observation of little evidence of phagocytosis by Bloom et al. (1947) suggests that the vegetative cells themselves make their way to the lymph nodes via the lymphatics in these cases. Researchers in the first half of the 20th century (reviewed by Sterne, 1959) believed that the capsule inhibits phagocytosis by neutralizing anthracidal substances present in normal serum and leucocytes. Whether or not the capsule is also essential to protecting the emerging cells when phagocytosed spores germinate within macrophages is discussed in section 5.5.1.

Complement C5-derived chemoattractants may be important in the recruitment of macrophages (Welkos et al., 1989).

Vegetative cells survive within the host as extracellular bacilli avoiding phagocytosis by means of the protective capsule (see section 5.5.1).

That macrophages may have a central role in the lethality of anthrax was demonstrated in experiments showing that laboratory mice were rendered insensitive to anthrax lethal toxin when their native macrophage population had been depleted by silica injection. The mice could be made sensitive again by injection of a toxin-sensitive macrophage cell line (Hanna et al., 1993). In contrast, Cote et al. (2004) showed that depletion of macrophages by other means increased the susceptibility of mice to infection. Supplementing the macrophage population in mice by treating them with additional exogenous macrophages enhanced their resistance to infection (Cote et al., 2004, 2006). Animals harbouring lethal toxin-resistant macrophages are, however, also susceptible to the toxin (Leppla, personal communication, 2004). The point at which these toxin-related observations manifest themselves in the actual infection has yet to be elucidated (see also section 5.5.3).

The role of macrophages in the pathogenesis of anthrax is now the subject of sophisticated research (Chai et al., 2004; Bergman et al., 2005).

5.4.2. Neutrophils (polymorphonuclear leukocytes)

Histopathology indicates that anthrax bacilli elicit a neutrophil response, and that both the lethal and oedema toxins enhance migration of neutrophils (Wade et al., 1985). Perversely, however, one role of the oedema toxin component of anthrax toxin (section 5.5.3) is to prevent mobilization and activation of neutrophils and thereby to suppress their phagocytosis of the bacteria (Leppla et al., 1985; O’Brien et al., 1985; Leppla, 1991). A recent report (Zenewicz et al., 2005) states that neutrophils are essential for early control of vegetative bacterial growth, although others (Cote & Welkos, 2005) considered that they played a relatively minor role in the early host response to anthrax spores.

5.5.1. The polypeptide capsule

The role of the poly-γ-D-glutamic acid capsule has long been regarded as being to protect the bacterium from phagocytosis (see section 5.4.1; Sterne, 1959). Early researchers believed that it inhibits phagocytosis by neutralizing anthracidal substances in serum and leucocytes. Ezzell & Abshire (1996) showed that the capsule appeared early during the process of germination of spores, which would suggest it has a protective role for emerging cells when phagocytosed spores germinate within macrophages. However Guidi-Rontani et al. (2001), using fluorescence imaging analysis to follow germination of B. anthracis spores and study survival of the germinated entities within primary mouse macrophages, concluded that the capsule played no role in this process. The observation of Meynell & Lawn (1965) was that, in vitro in the dividing vegetative cells, capsulation occurred at the end of exponential growth and that the cells divided equatorially with the capsules partitioned among the progeny.

Elucidation of the chemical nature of the capsule dates back to 1933 when Tomcsik & Szongott obtained a nitrogenous, polysaccharide-free, capsular material from B. anthracis. This was identified by Ivanovics & Bruckner (1937a, 1937b) as poly-D-glutamic acid, and how the glutamic acid residues were linked became the subject of a substantial number of studies (reviewed by Zwartouw & Smith, 1956a, and Fouet & Mesnage, 2002), with the final conclusion by Zwartouw & Smith (1956a) that they were linked by their γ-carboxyl groups. They also established that the antiphagocytic activity of the capsule was due to the multivalent negative charge of the polyglutamate ion.

The synthesis of the polypeptide through transamination between α-ketoglutaric acid and L-aspartic acid was worked out by Herbst (1944) and Housewright & Thorne (1950). The size of the polyglutamaic chains varies with the growth conditions, being 20–50 kDa when produced in vitro but in the order of 215 kDa in vivo (reviewed by Fouet & Mesnage, 2002). The biochemical structure of the capsule formed by B. anthracis is indentical to that formed by B. licheniformis with the same L-glutamic acid precursor (Thorne, 1956; Leonard & Housewright, 1963).

Ivanovics (1939) injected isolated capsular material intravenously into mice and rabbits and found that it was rapidly excreted by the kidneys without being broken down. He considered that this explained why the capsular material was harmless in itself and also why it was poorly antigenic. Weak immunogenicity of the surrounding capsule presumably aids B. anthracis in evading host immune response to its presence; at the same time, the capsule prevents antibodies to deeper surface antigens reaching those antigens. Tomcsik & Ivanovics (1938) did, however, succeed in developing antibodies in rabbits and recorded that they resulted in “certain immunity” in the rabbits and passively immunized mice. In more recent studies, opsonic IgG antibodies to the poly-γ-D-glutamic acid conjugated to proteins have been produced, and the suggestion made that the conjugates might be contributory as vaccine additives (Schneerson et al., 2003; Wang et al., 2004). A more definitive claim that the capsule could induce protective antibodies has been made by Chabot et al. (2004, 2005). Wang & Lucas (2004) immunized various strains of inbred mice and concluded that the antibody response patterns defined the capsule as a thymus-independent type 2 antigen.

The capsule does not form under normal aerobic culture and, in in vitro cultures, requires an atmosphere of elevated CO₂ together with the presence in the medium of serum and/or bicarbonate, or both (section 6.1, 6.3.1.6; Annex 1, section 3.7). Meynell & Meynell (1964) showed that sera from different species had marked differences in capsule-promoting activity and that the contribution of the sera was not nutritional but rather that it absorbed a dialysable inhibitor of capsule production present in the medium. The inhibitor was thought to be a fatty acid interfering with the assimilation of CO₂. It could also be absorbed by 0.2% (w/v) activated charcoal which was as efficient as serum in promoting capsulation. They showed that there was a relationship between pH and the threshold concentration of HCO₃- at which the capsule would be formed under any one atmospheric CO₂ concentration. They also noted that glucose suppressed capsulation. Later (Meynell & Meynell, 1966), using mutants with different nutritional requirements for capsulation, they concluded that, while HCO₃- made capsular synthesis possible, it was not required for the formation of the capsule itself. They also concluded that the inhibitory fatty acid(s) act by interfering with uptake or utilization of HCO₃-. Finally they showed that capsule formation was not inhibited by tetracycline and therefore was not synthesized like a protein.

Most of the research of recent years on the pathogenesis of anthrax has focused on the structure and function of the toxin, and the detailed role of the capsule is a topic still to be revisited properly with today’s molecular tools. Welkos (1991) showed that, in inbred strains of mice, cap+/tox- strains do possess a degree of virulence but, in the absence of toxin, the pathogenesis of these strains was unexplained. That there is strain variation in virulence indicates that there may be more to this than simple elaboration of the capsule, and the possibility that the phenomenon may be related to some other pXO2 gene(s), possibly together with chromosomal gene(s), has not been ruled out (Welkos, personal communication, 2004).

5.5.2. Polysaccharide

As reviewed by Smith & Zwartouw (1956), the production of polysaccharide material by B. anthracis was noted by several authors in the 1940s and 1950s. At least some of this was cell-wall associated. It precipitated with antipneumococcus type XIV serum and was present in both virulent and avirulent strains, and completely inactive in tests for aggressive activity. The conclusion was that it plays no important role in the virulence of B. anthracis.

5.5.3. The anthrax toxin complex

The toxin complex, which consists of three synergistically acting proteins, Protective Antigen (PA, 83kDa), Lethal Factor (LF, 90 kDa) and Oedema Factor (EF, 89 kDa), is produced during the log phase of growth of B. anthracis. LF in combination with PA (lethal toxin) and EF in combination with PA (oedema toxin) are now regarded as responsible for the characteristic signs and symptoms of anthrax. A considerable amount is known now about all three toxin components; they have been sequenced and the crystal structures of all three have been worked out with the roles of the domains within each molecule broadly elucidated (Petosa et al., 1997; Pannifer et al., 2001; Drum et al., 2002; Lacy & Collier, 2002).

According to the currently accepted model (Fig. 7), PA binds to receptors on the host’s cells and is activated by a host-cell surface furin-like protease which cleaves off a 20 kDa (PA₂₀) piece leaving exposed an EF/LF receptor-binding site. This “activated” PA₆₃ fragment combines with six other PA₆₃ fragments to form ring-shaped heptamers that bind EF and LF competitively and forms channels through which the complex is internalized by endocytosis (Lacy & Collier, 2002). Following acidification of the endosome, the LF and EF are released into the host-cell cytosol. The protein tumour endothelial marker 8 (TEM8) and capillary morphogenesis protein 2 (CMG2) have been shown to function as cellular receptors, although it is not clear which is relevant in vivo (Bradley et al., 2001; Scobie et al., 2003; Moayeri & Leppla, 2004). According to Ezzell et al. (1992, 2005), the cleavage of PA₈₃ into PA₆₃ and PA₂₀ occurs in the blood through the action of a serum protease, and the binding of EF and LF occurs here rather than at the cell surface just prior to endocytosis.

The literature now generally treats the combinations of PA with LF, and of PA with EF, as two separate toxins, referred to respectively as lethal and oedema toxins. They fall into the A-B class of toxins (binary toxins) with the enzymatically active “A” and binding “B” domains as discrete structures (Lacy & Collier, 2002).

EF is an adenylate cyclase which, by catalysing the abnormal production of cyclic-AMP (adenosine monophosphate) (cAMP), produces the altered water and ion movements that lead to the characteristic oedema of anthrax. High intracellular cAMP concentrations are cytostatic but not lethal to host cells. EF has been shown to impair neutrophil function and incapacitates phagocytes and cytokine pathways (reviewed by Moayeri & Leppla, 2004) and its role in anthrax infection may be to prevent activation of the inflammatory process. Recent presentations (Firoved et al., 2005; Quesnel-Hellmann et al., 2005) reported that oedema toxin (EF + PA) disrupted cytokine networks during infection, and that purified oedema toxin administered to BALB/cJ mice intravenously resulted in circulatory and cardiac dysfuntions, tissue damage and focal haemorrhaging, accumulation of several cytokines in the sera and death.

LF is a highly specific zinc-dependent protease that cleaves the amino termini of six mitogen-activated protein kinase kinases (MAPKK), and thereby disrupts pathways in eukaryotic cells concerned with regulating activity of molecules through phosphorylation cascades (Duesbury et al., 1998; Vitale et al., 1998, 2000). The affected signalling pathways are involved in cell growth and maturation as well as cellular stress responses; the manner in which the disruption of these pathways leads to the known effects of lethal toxin has yet to be fully elucidated but may act by impairing the ability of endothelial cells to initiate an immune response by destabilizing cytokine IL-8 mRNA (Batty et al., 2005) or through other modes of inhibition of cytokine gene expression (Tonello et al., 2005).

Advancing knowledge of the structure of the toxin components and the stages of interaction of these with the host cell, from binding of PA to internalization and catalytic actions of EF and LF in the cytosol, are reviewed by Lacy & Collier (2002).

On the basis of mouse and tissue culture models, the primary cell type affected in anthrax pathogenesis has for some years been regarded as the macrophage (Friedlander, 1986; Hanna et al., 1993; see also section 5.4.1). Low levels of lethal toxin cleave MAPKK-3. High levels of lethal toxin lead to lysis of the macrophage by an as yet unknown mechanism (Pannifer et al., 2001). The relationship between macrophage cell death and death of the infected host is under investigation, although it seems clear that macrophage lysis is not required for lethal toxin lethality (Moayeri et al., 2003). It is emerging that the effect of lethal toxin on other cell types may also be integral to the pathogenesis of anthrax. For example, lethal toxin impairs host B and T cell immune responses by its action on dendritic cells (Moayeri & Leppla, 2004). The action of lethal toxin is further discussed in section 5.5.4, and the current wisdom is that death is the result of the toxin acting through a non-inflammatory mechanism that involves hypoxic injury but not macrophage sensitivity to toxin (Moayeri et al., 2003).

The inverse relationship between susceptibility to infection and susceptibility to the toxin was discussed in section 3.1.

5.5.4. Toxin and terminal haemorrhage

The characteristic terminal haemorrhage to the exterior from the orifices of the animal at death – an essential part of the organism’s cycle of infection (Figs 1&2) – is caused by the action of the toxin on the endothelial cell-lining of the blood vessels, histologically visible as necrosis, which results in their breakdown and bleeding. Anthrax toxin acts directly on the intact membranes of endothelial cells, making them permeable to plasma and causing intravascular thrombosis (Beall & Dalldorf, 1966; Dalldorf, 1967; Dalldorf et al., 1971). This is manifested in petechiae in the visceral organs and haemorrhages leading to bleeding from the orifices. Kirby (2004) found that lethal toxin (PA+LF), but not oedema toxin (PA+EF), induces apoptosis in cultured human endothelial cells. He hypothesized that lethal toxin action on endothelial cells, primarily through inhibition of the extracellular signal-regulated kinase (ERK) pathway, contributed to vascular pathology and haemorrhage during systemic anthrax. The disruption of key functions of the endothelium by lethal toxin has been studied further more recently (Warfel et al., 2005).

The degree of endothelial damage appears related to the number of bacilli. In early infection of the lymph nodes, only oedema is present. As the number of bacilli rises, concomitant toxaemia increases to a critical level, at which the toxin causes endothelial cell injury with resulting thrombosis and haemorrhage.

5.5.5. S-layer

In common with many, but not all, bacteria, B. anthracis forms the proteinaceous paracrystalline sheath over the peptidoglycan cell surface known as an S-layer. Very few bacteria share with B. anthracis both an S-layer and a capsule (Fouet & Mesnage, 2002) and this may be pertinent to the pathogenicity of B. anthracis. S-layers have been shown in other bacterial species to be virulence factors, and this may be the case with B. anthracis (Fouet et al., 1996; Fouet & Mesnage, 2002).

The S-layer of B. anthracis comprises two components, a 94-kDa protein termed Sap (surface array protein) and a 94-kDa protein known as EA1 (extractable antigen 1) (Ezzell & Abshire, 1988; Fouet & Mesnage, 2002). Sap is cell-associated but also released into the surrounding medium in cell or culture suspensions; EA1 is cell-associated and only released if the cells are washed with sufficient vigour (Fouet & Mesnage, 2002; Williams & Turnbough, 2004).

The presence of an S-layer is not required for capsulation of B. anthracis bacilli, but the presence of all three is necessary for maximal resistance to complement pathway-mediated defences (Fouet & Mesnage, 2002). This may be related to protection of the emerging germinating cell which, at early stages, is not wholly surrounded by capsule (Ezzell & Abshire, 1996).

5.5.6. Exosporium

In common with spores of all Bacillus species, the spores of B. anthracis comprise a core surrounded by peptidoglycan cortex, in turn surrounded by the multiple-layered proteinaceous spore coats. These apparently form the outermost layer for some Bacillus species but, in the case of the B. cereus group, there is an additional loose-fitting, two-layer exosporium comprising about 2% of the mass of the spore. The inner layer has a hexagonal crystal structure, while the outer layer consists of hairlike filaments – the so-called “hirsute nap”. About 50% of the exosporium is protein; the precise number and identities of the component proteins are not yet known, although several have been described.

The function of the exosporium is not known, but it may be the basis of the hydrophobicity of spores and, thereby, play a role in adhesion both in the environment and in vivo. It is speculated that it is involved in the interaction between spores and uptake by, or germination within, macrophages, although it is also stated that spores devoid of the exosporium are as infectious as those that have it. (For relevant references, see Steichen et al., 2003; Liu et al., 2004; Redmond et al., 2004).

5.5.7. Other potential virulence factors

A number of general and specific observations suggest that there is more to the virulence of B. anthracis than simply its ability to elaborate the toxin and the capsule. The following are examples of these:

• While the toxin and capsule components, the genes encoding these and the plasmids carrying these genes are seemingly identical in different B. anthracis strains, the strains themselves possess different LD₅₀s.
• Strains cured of one or other of the plasmids, and therefore able to produce toxin or capsule, but not both, retain a level of virulence (Welkos, 1991; see section 5.5.1).
• There is substantial evidence that some tox+/cap+ strains can overwhelm immunity induced by toxin-based vaccines more readily than other strains.
• There is some evidence that B. anthracis can elaborate a thiol-activated cytolysin, anthrolysin o (Shannon et al., 2003; Mosser et al., 2005; Thomason et al., 2005). Analogous cytolysins in other pathogens are established virulence factors.
• While B. subtilis spores injected into the footpad of a mouse are cleared rapidly, those of a B. anthracis strain cured of both pXo1 and pXo2 persist, implying chromosomal involvement in the ability of B. anthracis to persist (Pezard et al., 1991).
• Evidence exists of chromosomal loci with a role in the anthrax infectious process (Stepanov, Mikshis & Bolotnikova, 1996; Stepanov et al., 1996, 1999).
• The regulator atxA, known to be the anthrax toxin activator and responsible for up-regulation of the capsule genes, has been shown to be involved in the regulation of numerous other genes on pXo1, pXo2 and the chromosome. These genes may have direct or indirect roles in the pathogenesis of B. anthracis (Bourgogne et al., 2003).
• Strain comparisons of germination within macrophage cultures suggested a requirement for multiple germinant receptor genes located on both the chromosome and pXo1 for host-specific germinants and initiation of the infectious cycle (Hanna, 2001).
• Compared to B. subtilis, B. anthracis harbours a large number of genes predicted to be important for amino acid and peptide utilization, and which may provide an advantage during its life-cycle within an animal host. Moreover, several chromosomally encoded proteins have been noted which may contribute to pathogenicity, including haemolysins, phospholipases and iron uptake entities (Read et al., 2003).
• Germination operon gerX is located on plasmid pXo1 and is unique in being the only germination operon known to be located on a virulence plasmid (Guidi-Rontani & Mock, 2002). The gerX-encoded proteins may be virulence factors, contributing to the pathogenesis of B. anthracis.
• A recent claim has been made (Heninger et al., 2005) that overwhelming murine septicaemia can be induced independent of toxin function.