Ever since Pasteur developed the first anthrax vaccine containing attenuated live organisms, clinical anthrax vaccines have largely been empirically derived [1,2,3]. There are two main types of anthrax vaccine available for use in humans; cell-free filtrates of antigenic proteins derived from cultures of avirulent B. anthracis strains and live attenuated spore-based vaccines. The live vaccines were developed from avirulent, non-encapsulated Sterne strains (ST-1 and ST-3) and predominantly used in Russia [2,4]. These vaccines may be administered by scarification of the skin, aerosol or subcutaneous injection, and require annual boosters [5,6]. Live vaccines have a low reported rate of adverse events [5], however, there are concerns regarding the use of live spores, as well as a dearth of data regarding the immunogenicity of these vaccines [6]. This means that currently, there are two anthrax vaccines widely used in humans, the UK licensed anthrax vaccine precipitated (AVP), and the US licensed anthrax vaccine adsorbed (AVA or Biothrax), both of which are based upon culture filtrates of B. anthracis [1,7,8]. The AVA vaccine is produced from a filtrate of the non-encapsulated, nonproteolytic strain V770-NP1-R, which is adsorbed onto aluminium hydroxide [9]. The UK vaccine is produced from an alum precipitated filtrate of the avirulent, non-encapsulated Sterne strain 34F2 [1,8]. Protective immunity is induced by administering the vaccines subcutaneously, in a series of up to six initial doses [7,8], followed by annual booster vaccinations. This extensive vaccination regimen, in combination with reported adverse reaction rates of 11% for the UK vaccine [10], and up to 60% for the US vaccine [8], has prompted the need for rationally designed, effective vaccines with low reactogenicity [1].

Both the AVA and AVP vaccines contain Protective Antigen (PA) and variable amounts of the two enzymatically active toxin subunits, Lethal Factor (LF) and Edema Factor (EF), which combine to form binary exotoxins [3]. In addition to their importance as key virulence factors for B. anthracis, these toxins form the basis of next generation anthrax vaccines currently under development. Human clinical trials have shown that vaccines based upon the protein sub-units are capable of eliciting T cell and antibody immune responses while avoiding the adverse reactions associated with older filtrate based vaccines [11,12,13].

As PA was the earliest focus of work to identify an immunologically protective antigen and remains the most extensively characterized of the anthrax toxins, it is perhaps unsurprising that the vast majority of approaches described to date have concentrated upon PA as the sole vaccine immunogen [14,15,16,17]. AVA and recombinant protective antigen (rPA) based vaccines have been found to induce anti-PA antibodies in a variety of animal species [18], antibodies which were also correlated with survival in animals challenged with anthrax infection [19], and could be passively transferred to naive animals [20]. In addition to anti-PA IgG, immunisation of rabbits and guinea pigs with AVA or rPA also induced the production of toxin neutralising antibodies specific to PA, levels of which were predictive of survival following anthrax infection [21,22,23,24,25]. This indicated that the presence of toxin neutralising antibodies, which could also be passively transferred to naive animals [25], conferred protective immunity to anthrax. Furthermore, Weiss et al. discovered that in rabbits immunised with PA based vaccines the toxin neutralising antibody titres were better predictors of survival following anthrax challenge than total anti-PA antibody titres [26].

Despite this, human and animal vaccination studies have indicated that not only PA, but also LF is capable of conferring protective immunity [3,27,28]. Human vaccinees, following immunisation with either the AVA or AVP vaccines, show antibody responses to both PA and LF [29,30], while both PA and LF specific antibodies have been detected in sera taken from naturally infected anthrax patients [29,31]. A case study involving a naturally acquired cutaneous anthrax infection notably demonstrated detectable anti-LF antibodies in the absence of anti-PA antibodies [32]. This was supported by the recent work by Brenneman et al. which found that in cutaneous anthrax patients, the majority of toxin specific IgG antibodies were directed against LF, and were induced considerably earlier than either PA or EF specific antibodies [33]. Animal studies have found that one year after PA based vaccinations there was minimal protection against infection [18,34], providing evidence that PA based immunity may not be long lasting. Crowe et al. also recently discovered that while the overwhelming majority of AVA vaccinated humans in a study cohort demonstrated PA-specific antibody responses, many of the responses were not actually associated with a detectable toxin neutralising effect [35]. As a link has been established between the passive protection afforded to mice against both toxin challenge [36,37] and anthrax infection [38], by toxin neutralising monoclonal antibodies generated from AVA vaccines, it suggests that current vaccines, such as AVA, which focus upon induction of anti-PA antibody responses, may produce a variable quality of immunoprotection. Antibodies directed against LF have proven to be toxin neutralising, both in vitro and in vivo in rats [39] and mice [36]. The growing body of work on LF, identifying B cell epitopes [40,41], providing evidence that LF can boost the magnitude of PA-specific antibody responses in mice following co-administration [42,43], and that LF truncate proteins confer protection against B. anthracis aerosol challenge [17,44], adds weight to the evidence that responses against LF may be more important mediators of protective immunity than previously thought, with the potential for inclusion of LF derived epitopes in a sub-unit vaccine.

In contrast to the focus on antibody-based immunity which has driven studies of B. anthracis antigens to date, there is a paucity of work on T cell immunity, with those few studies which exist mainly concentrating on responses to PA [45,46,47]. The discovery that guinea pigs which survived B. anthracis exposure exhibited extremely low anti-PA and LF antibody titres indicated that protection was not necessarily associated with the induction of antibody responses [29]. However, it is only recently that the critical role cellular immunity plays in the clearance of B. anthracis has been investigated more fully, questioning the primary importance of humoral immunity in providing protection against anthrax infection [48]. Our own recent research has shown that LF specific IFNγ producing CD4+ T cells play an important role in generating long lasting immunity to anthrax [49]. In individuals exposed to anthrax spores, circulating T cells reactive to both LF and PA have been identified [50], and low-level anthrax exposure has been found to lead to PA and LF specific cellular responses in the absence of detectable antibodies [51]. The importance of considering the cellular immune response to both the PA and LF immunogens echoes a growing trend in the vaccinology field that a focus upon eliciting antibody responses should not be the sole approach for pathogens that encompass an intracellular component to their pathogenesis [52]. It is thus crucial that we understand how these exotoxins impact on adaptive immunity, as well as their potential in eliciting more effective T cell immunity.

The entry of B. anthracis exotoxins into host cells is mediated by PA, which binds to two cell surface receptors; anthrax toxin receptor 1 (ANTXR1), also known as tumour endothelial marker 8 (TEM8) and anthrax toxin receptor 2 (ANTXR2), also known as capillary morphogenesis gene 2 (CMG2) [53,54,55,56]. Both of these receptor proteins, whose normal physiological function in the host concerns capillary development, bind to PA through the cation-dependent metal ion-dependent adhesion site (MIDAS), within their von Willebrand A (vWA) domains, which are also referred to as “integrin-like” domains [57,58]. Cohen et al. have recently demonstrated in vitro that α4β1- and α5β1-integrin complexes are capable of binding to PA and mediating transport of the anthrax toxins into the cell [59]. Whilst the discovery of a third PA receptor on the host cell may prove to have interesting implications for efforts to develop monoclonal antibodies, blocking PA binding to host cell receptors, currently ANTXR2/CMG2 remains the most physiologically relevant receptor during anthrax infection. Not only did Cohen et al. establish that the integrin receptors bound in vitro to PA with an affinity comparable to that of ANTXR1/TEM8, much lower than ANTXR2/CMG2 [59], but also the in vivo toxicity of anthrax toxin is predominantly mediated by ANTXR2/CMG2 rather than ANTXR1/TEM8, due to both the high affinity PA displays for the ANTXR2/CMG2 receptor and the higher levels of this receptor expressed upon cell populations within the host [60,61,62]. PA forms heptameric structures to which the EF and LF proteins bind through a homologous N-terminus region, forming Edema Toxin (ET), or Lethal Toxin (LT), which are transported into the cytosol. The structure and function of PA and EF have been extensively reviewed elsewhere; in this issue we will concentrate primarily on LF and its mode of action.

In contrast to the LT induced activation of caspase-1, which can take up to 60 min [115], MKK cleavage is extremely rapid, occurring within 20 min of LT addition to cells in vitro [116]. The MAPK cascades are critical in controlling T-cell activation, as these pathways are involved in the activation of T cell receptor (TCR) dependent nuclear factor of activated T cells (NFAT) family members. Along with activating transcription factor 2 (ATF2), the NFAT family members act synergistically with nuclear subunit activation protein 1 (AP-1) as transcription factors regulating the expression of inducible genes vital to the adaptive immune response. The regulation of NFAT, AFT2 and AP-1 activation is controlled at multiple levels by MAPK phosphorylation, the action of LT upon the MKKs thus blocks JNK, ERK and p38 mediated T cell proliferation [117,118]. The NFAT transcription factors are also involved in the regulation of genes encoding IL-2 and the IL-2 Receptor (IL-2R), which are essential for the clonal expansion of antigen specific T cells. LT induced impairment of T cell proliferation is traditionally associated with the reduced production of the Th1 and Th2 cytokines IL-2, IFNγ, TNFα and IL-5 and the downregulation of the activation markers, CD69 and CD25 [117,119]. The blockade of T cell activation being potent enough to inhibit the co-stimulation ex vivo of LF exposed CD4+ T cells, by α-CD3 and α-CD28 antibodies [118]. Due to its role in the perturbation of PKA dependent intracellular signalling pathways, EF also has an indirect effect upon NFAT transcriptional regulation, interfering with the calcium dependent signalling, but only when present at much higher concentrations than LF [117].

The antagonistic effect of EF upon PKA strengthens the effects of LF on MAPK cascades, as EF is capable of both phosphorylating a negative regulatory residue of Raf, which lies upstream of the MAPK pathway, and phosphorylating a positive regulatory residue of Rap1 which inhibits Ras initiation of the MAPK cascade [120]. The synergistic effect of LF and EF upon the MAPK pathways suppresses T cell chemotaxis in response to CXCL12 [120]. This may be mediated through inhibiting the phosphorylation of both regulatory components such as paxillin and structural components such as actin, of the cell cytoskeleton [121,122]. The elevation of intracellular cAMP by EF also appears to skew the differentiation of naïve CD4+ T cells towards a Th2 subset, inhibiting the TCR dependent activation of Akt1, a protein essential for the development of a Th1 subset, whilst enhancing the activation of the guanine nucleotide exchanger Vav1 and the stress kinase p38 which are involved in Th2 differentiation [123]. A predominantly Th2 response has also been reported following stimulation of PA-specific T cells from AVA immunised rhesus macaques typified by increased levels of IL-4 and IL-2 but not IFNγ, IL-6, IL-1β or TNFα mRNA levels [18]. Boyaka et al. have also found that PA specific CD4+ T cells from mice immununised with rPA secrete low levels of Th1 associated cytokines, but increased levels of characteristic Th2 cytokines IL-4, IL-5 and IL-10 [124]. It is important to note however, that both studies utilised adjuvants which may have had an impact upon the induction of distinct T cell subpopulations, as both the alum adjuvant in AVA vaccines and the cholera toxin adjuvant used by Boyaka et al. have been implicated in biasing the CD4 T cell response towards a Th2 population [125,126,127]. Along with the LF and EF effect upon DC cytokine release, which promotes preferential Th2 differentiation, this represents both direct and indirect consequences of anthrax toxins upon the lineage commitment of naïve CD4+ T cells.

The importance of an IFNγ CD4+ T cell response in mediating protection from anthrax infection in vivo [48], in addition to the increased survival and B. anthracis bactericidal activity of IFNγ activated macrophages [128], and the disproportionate focus of LT upon suppressing IFNγ release by NK cells [129], suggests that survival of B. anthracis within the host depends in part, upon the suppression of a Th1 associated IFNγ release. Conversely this also indicates that the induction of an IFNγ producing memory T cell population should be an important goal of any future B. anthracis vaccine design.

Recent research has suggested a role for natural killer (NK) cells in the production of IFNγ following exposure to anthrax spores [130], and natural killer T (NKT) cells in the generation of PA-specific, LT neutralizing antibodies by B cells [131]. NKT-derived IL-4 and IFNγ influences the production of polyclonal PA-specific IgG1, which Devera et al. found to be more protective in vivo than polyclonal PA-specific IgG2b or IgG2c [132]. However, it must be noted that Abboud et al. found that PA-specific IgG2a isotypes were superior to both IgG2b and IgG1 in terms of in vivo protection [133]. Devera et al. acknowledge that due to the absence of IgG2a isotypes within their work, IgG1 may have assumed a greater importance than it might otherwise warrant, although the passive transfer of monoclonal antibodies in the model of Abboud et al. could equally have led to a greater role for IgG2a than may yet be demonstrated within a natural anthrax infection.

The CD1d restricted type I invariant NKT (iNKT) cells provide B cell help, in addition to the production of the pro-inflammatory cytokines IFNγ, TNFα and GM-CSF. Surprisingly iNKT cells appear to express higher levels of the PA receptors TEM-8/ANTRX1 and CMG-2/ANTRX1 than other hematopoietic cells [134]. As NKT cells represent less than 1% of the circulating T lymphocytes in humans (although this percentage varies substantially between individuals) this may denote a disproportionate focus upon this cell population by anthrax toxins. This preferential targeting of iNKT cells is supported by the downregulation of the activating receptor NKG2D on iNKT cells but not NK cells in the presence of LT [134]. LT is capable of mediating a variety of effects upon iNKT cells through the cleavage and inactivation of MKK2 which impacts upon the phosphorylation of downstream ERK1 and 2 [134,135]. Inhibition of the ERK pathway impairs CD1d mediated antigen presentation to iNKT cells, in addition to an observed reduction in endosomal trafficking of antigen loaded MHC class II molecules; this has major implications for the induction of both NKT and CD4+ Th1 adaptive immune responses to B. anthracis [92,135].

The inhibitory effects of both LT and ET upon expression of the activation markers CD25 and CD69 and the secretion of the pro-inflamatory cytokines IL-2, TNFα, and IFNγ by human T cells has been described in vitro [117,119]. Murine lymphocytes have also demonstrated impaired TCR mediated cell activation and CD4+ T cell production of the cytokines IL-3, IL-4, IL-5, IL-6, IL-10, IL-17, TNFα, IFNγ and GM-CSF following exposure to LT and ET [118]. However, the cellular immunity identified within naturally infected humans indicates that, although in vitro exposure to ET has been implicated in immune deviation towards both the Th2 and Th17 pathways [123,136], the human immune response against LF encompasses a broad range of cytokines associated with Th1, Th2, Th9 and Th17 subsets, indicating little or no helper T cell polarization [114]. This lack of skewing following repeated exposure to LF antigens echoes the work of the Peakman lab, who found that there was an equal balance of Th1 and Th2 responder cells in the recall immunity to anthrax [137]. Martchencko et al. [138] observed that anthrax toxin sensitivity, which varies between cell lines generated from different humans, appears to be strongly correlated with variation in the expression of the host cell surface receptor ANTXR2/CMG2. A single-nucleotide polymorphism (SNP), affecting the uptake of anthrax toxins, was identified in the protein coding region of ANTXR2/CMG2. Interestingly, the polymorphism, which is commonly found in African and European populations, decreased the cellular entry and subsequent lethality of anthrax toxin in murine cells, although this was not translated into a statistically significant difference between the in vitro human cell populations, in terms of toxin sensitivity. It remains unclear, at present, what role the genetic diversity of the host plays in the course of a B. anthracis infection, however two recent large scale studies have indicated differences in the generation of toxin specific antibodies between AVA vaccinees of different ethnicities. Crowe et al. reported that African American individuals displayed lower titres of toxin neutralising antibodies following vaccination [36], while Marano et al. found that, during the immunisation schedule, anti-PA IgG levels were significantly lower in African Americans when compared to Americans of European descent (although this effect was not observed at a time point seven months post-vaccination) [139]. As Ingram and Baillie observed, a range of genetic factors may govern this variation in response to anthrax derived proteins, including the expression of cellular receptors like ANTXR2/CMG2, polymorphisms in both cytokine and cytokine receptor genes, and variations in Toll like receptor (TLR) genes or HLA haplotype [140].

Extensive allelic polymorphisms observed within the HLA region influence the repertoire of antigenic epitopes presented by MHC class II molecules to CD4 T cells; in order to fully elucidate the cellular immune response to anthrax it is therefore important to define these epitopes. Most vaccine strategies against anthrax have concentrated upon PA, however LF has been identified as a major target of T cell immunity in humans [49]. As the amount of LF released by B. anthracis is one-sixth of that of PA [141], these findings indicate that LF may contain proportionally more epitopes which are of relevance in the T cell immune response. The elicited human T cell responses indicated that the key LF epitopes, LF_574–593, LF_654–673, LF_674–693 and LF_714–733, are concentrated in domain IV (Figure 1). It is interesting to note that this reveals a very different pattern to that of the B cell epitope mapping performed by Nguyen et al., who found that the majority of structural B cell epitopes clustered in domains I and III of LF [41].

The binding groove created by domains II, III and the catalytically active center of the LF toxin, domain IV, is responsible for holding the 16 amino acid long peptide which makes up the tail of the MAPK kinase family [63], and it has been observed that mutations in the sequence coding for domain IV eliminates the peptidase activity of LF, abrogating its toxicity [142]. The putative zinc binding site, which lies between the amino acid residues LF₆₈₆ and LF₆₉₂ [73,77] was a feature of the response to LF_674–693; these represent important residues for zinc binding and catalysis as well as cytotoxicity, which are conserved across this class of metalloprotease [72]. The immunodominant epitopes identified within domain IV therefore appear to comprise essential residues of LF which are critical for efficient catalytic activities and the execution of substrate cleavage.

The cleavage of MAPK kinases by LT inhibits the evolutionarily conserved MAPK signal transduction pathway and has multiple downstream consequences for both the innate and adaptive immune systems. The resulting immune subversion might be expected to impact upon the T cell memory of B. anthracis, however it is clear that the involvement of factors such as the Nlrp1 directed inflammasome allow the immune system to develop antigen-specific responses to LF. We propose that in addition to resolving infection, the cell infiltration and cytokine milieu seen in early inflammation are crucial in driving antigen presentation and T cell priming which results in broad Th cell subset responses to LF, which can be detected several years after the initial infectious event. The identification of T cell epitopes from the LF protein may therefore be beneficial in developing a future protective sub-unit based vaccine.