|Home | About | Journals | Submit | Contact Us | Français|
Anthrax lethal toxin (LT) and edema toxin (ET) are the major virulence factors of anthrax and can replicate the lethality and symptoms associated with the disease. This review provides an overview of our current understanding of anthrax toxin effects in animal models and the cytotoxicity (necrosis and apoptosis) induced by LT in different cells. A brief reexamination of early historic findings on toxin in vivo effects in the context of our current knowledge is also presented.
The three toxin polypeptides encoded by the Bacillus anthracis pXO1 plasmid combine in binary combinations to form two toxins, lethal toxin (LT) and edema toxin (ET), which are responsible for the lethality and symptoms associated with anthrax disease. Although the pXO2 plasmid-encoded capsule is an important virulence factor for the establishment of disease, the symptoms associated with anthrax are the result of toxin production following septicemia. Thus, antibiotics which clear bacteria from infected hosts cannot protect against toxin effects.
Beyond the replication of disease symptoms and lethality by the toxins, proof of their essential and singular role in anthrax also comes from the multitude of animal studies which have shown that immunization against the protective antigen (PA) component of anthrax toxin and antibodies against this protein are sufficient for protection from lethality.
PA is the receptor binding component in LT and ET, for which it facilitates the delivery, respectively, of lethal factor (LF) or edema factor (EF) into cells. PA binds to two cellular receptors, is cleaved and oligomerizes to form binding sites for the EF and LF enzymatic moieties (Figure 1). Detailed analyses of PA interaction with cellular receptors, endocytosis and LF/EF translocation through the PA heptamer pore are not the subjects of this review, and can be found in reviews by Young and Van der Goot as well as Collier in this volume.
LF is a zinc metalloproteinase (Klimpel et al., 1994) which cleaves the N-terminus of a number of mitogen activated protein kinase kinases (MAPKKs, or Meks) (Duesbery et al., 1998;Pellizzari et al., 1999;Vitale et al., 1998;Vitale et al., 2000). LF cleavage of Meks disrupts the ERK1/2, JNK/SAPK and p38 signaling pathways. These pathways play important roles in numerous cellular functions ranging from proliferation and cell cycle regulation, to immune modulation and survival against toxic insults. The pharmacological inhibition of the Mek pathways has yielded a large volume of data implicating them in these functions in different cell types, and as expected, LF treatment of cells often replicates these inhibitor effects in the same cells.
EF is a potent calmodulin-dependent adenylate cyclase (Leppla, 1982) and cAMP production by this toxin also affects cell signaling pathways. The literature on regulation of cell function by this important second messenger is even more substantial and it is expected that many cAMP-mediated effects already described in different cell types would also occur universally in response to ET as almost all tested cells have functional PA receptors.
It is interesting that both anthrax toxins target such ubiquitous cell signaling pathways with wide-reaching effects in almost every cell type. The potential modulation of so many different pathways by these toxins makes the identification of the relevant in vivo targets a crucial goal for the understanding of their role in human disease.
This review will provide an overview of our current understanding of anthrax toxin toxic effects in vivo as well as primarily focusing on cytotoxic effects for different cells. A brief reexamination of the early historic findings on toxin in vivo effects in the context of our current knowledge is also presented. The important nontoxic immunomodulatory effects of the toxins are covered in detail by other reviews in this volume.
Studies with highly pure Bacillus anthracis-derived, endotoxin-free LT in animal models confirm discoveries made in the 1960s (see later section on historic studies). LT induces an atypical vascular collapse in mice and rats without classic hallmarks of endotoxic shock, marked by absence of thrombosis or cytokine involvement (Cui et al., 2004;Moayeri et al., 2003). In contrast, ET used at doses that are lethal for mice is associated with hemorrhaging lesions in many organs accompanied by both hypotension and bradycardia (Firoved et al., 2005). In fact, early in vivo studies which often used PA or LF preparations contaminated with EF also did note capillary thrombosis (Beall and Dalldorf, 1966). ET also induces shock in rats in a manner distinctly different than that caused by LT, accompanied by hemorrhage (Kuo et al., 2008), increased heart rate (Cui et al., 2007;Watson et al., 2007a), and reduced cardiac output without altered myocardial contractility (Watson et al., 2007a;Watson et al., 2007b). ET effects on the heart are likely secondary to general loss of circulatory fluids in many tissues.
The striking rapid death LT induces certain inbred rats was first observed decades ago (Beall et al., 1962). As little as 7.0 μg of each toxin component (injected IV) can kill numerous strains while other strains are completely resistant to doses as high as 250 μg. All thus far tested rat strains can be divided into absolute sensitive and resistant groups (M. Moayeri, unpublished data). Sensitive rat strains (including Fischer, Sprague Dawley and Brown Norway) can die as rapidly as 37 minutes (Ezzell et al., 1984;Gupta et al., 2008) when given a saturating dose of LT (100 μg each of PA and LF, IV) and never take longer than 60 minutes to succumb. In contrast, inbred mice die over days (48–120 hours).
Not surprisingly, there are no histopathological changes in LT-treated rats, and almost all histopathological observations in LT-treated mice are simply secondary to vascular insufficiency. While the absence of classic anthrax hemorrhagic pathology previously attributed to LT’s effect is a distinct difference from anthrax disease, other LT resultant effects such as pleural edema and rapid shock closely parallel what is seen with infection in humans and numerous animal models. Despite early hypotheses on LT-induced endotoxic-like shock occurring due to macrophage lysis and cytokine (TNF-α/IL-1β) induction by LT (Hanna et al., 1993), we now know cytokine shock is not required for LT-mediated vascular collapse in rodents and mice with resistant macrophages also succumb to LT (Cui et al., 2004;Moayeri et al., 2003). The rapidity of the hemodynamic changes in the rat model also supports a unique LT-mediated event which differs from most classic shock states. Amazingly, we still do not know how LT induces vascular collapse in animals. Endothelial cells may be targeted but in vivo evidence for their involvement is lacking. Recent data indicates the heart may actually be the primary target for LT in rats and the pulmonary edema associated with LT treatment may simply result from severe heart failure (Kuo et al., 2008;Watson et al., 2007a;Watson et al., 2007b). Our recent electron microscopy analyses show striking early LT-mediated pathological changes in murine heart myocytes and endothelium as early as 6 hours after toxin treatment, accompanied by changes in major cardiac markers (Moayeri et al., 2009). These changes, not detectable by light microscopy analyses, suggest an early targeting of the murine heart and endothelial cells by LT. It will be important to find out of electron microscopy can also detect in vivo toxin targeting of endothelial cells in other tissues.
Over recent years, a Sprague Dawley rat infusion model has provided information on the novel vascular shock induced by LT, underlining its differences from cytokine-mediated endotoxic shock (Cui et al., 2004;Cui et al., 2006;Sherer et al., 2007). LT infusion causes blood pressure and heart rate changes in these rats, but cytokines, nitric oxide or histopathological changes are absent. Classic shock therapies such as fluid support either worsen outcome in LT-mediated rat shock (Sherer et al., 2007) or have no beneficial effect, as seen with norepinephrine treatment (Li et al., 2009). Once again, these findings support the concept that LT-related shock is independent of classic vasopressor-mediated in vivo changes and essentially different than most bacterial-induced shock.
Two issues arise in the use of LT-sensitive rodents as models for toxin-mediated effects in vivo. First, mice succumb to toxin-deficient encapsulated B. anthracis strains (by multiple routes) similarly to wild type strains (Heninger et al., 2006), making these animals a poor model for B. anthracis bacterial infection studies assessing the role of toxin. However, toxin knockouts in the non-ecapsulated Sterne strain (pXO1+ pXO2−) are attenuated (Pezard et al., 1991). While rodents are clearly not the best animal model for the study of anthrax infection, purified LT can reproduce major symptoms or anthrax such as pleural edema in these animals, making them valuable for assessment of in vivo toxin mechanisms (Ezzell et al., 1984;Firoved et al., 2005;Moayeri et al., 2003).
The second important issue when using a reductionist approach in studying toxin effects in vivo is the question of dose relevance. The amounts of LT used as a bolus in mice are based on older data collected on terminal levels of toxin in various animals (Turnbull, 1990). Until recently, however, ET was often only measured as trace levels in infected animals and the levels in terminal infection were unknown. A recent detailed analysis of in vivo toxin production levels in anthrax infected rabbits found a range of LF (10–35 μg/ml) by 48 hours and a 5:1 ratio of LF:EF in most samples (Dal Molin et al., 2008). This finding actually confirms what was reported for in vitro toxin production (PA:LF:EF, 20:5:1) (Sirard et al., 1994) as well as findings in another in vivo study of Ames spore infection in rabbits (Mabry et al., 2006). Based on the 5:1 ratio of LF:EF found in real infection models, it is likely lethal doses of ET for mice are actually not be achieved in infection (Firoved et al., 2005). However the pathology associated with lower nonlethal doses of ET which does include nonlethal hemorrhaging in multiple organs and extreme changes in adrenal glands are more likely the relevant ET effects in infection (Firoved et al., 2005). Clearly, further investigation of the timing and levels of toxins produced in early stages of infection will be important
Finally, the wide range of mouse and rat strain susceptibilities to LT and ET, which cannot be attributed to receptor expression, suggest an interesting role for genetic background in susceptibility to these toxins (see Tables 1–3). Such genetic-based toxin sensitivity differences may also have important consequences in the human population. BALB/cJ mice succumb to LT (LD100 is 35 μg, IV, 100 μg IP), while the AKR/J, NOD/J and DBA/2J strains are resistant to these doses of toxin. The C57BL/6J strain which has intermediate resistance to LT relative to BALB/cJ, is highly susceptible to ET compared to most other mice. In the case of LT, there appear to be multiple loci involved in determining susceptibility to toxin in mice, only one of which has been identified. The Nalp1b (also known as Nlrp1b) locus which determines macrophage sensitivity to LT can contribute to increased sensitivity to toxin, but only in certain strains (Moayeri et al., 2004)(see next section). Recombinant inbred mice derived from DBA/2J and Balb/cJ mice suggested at least three loci controlling murine sensitivity (McAllister et al., 2003). The identification of genes contributing to LT and ET sensitivity in animals may be important in understanding the widely differing levels of disease found in humans exposed to similar spore infections.
Macrophages from select inbred mouse strains undergo a rapid lysis in response to LT, while other strains harbor LT-resistant macrophages (Friedlander, 1986;Friedlander et al., 1993) (see later sections). LT induced vascular shock kills mice with LT resistant macrophages as well as macrophage-deficient mice (Moayeri et al., 2003;Moayeri et al., 2004). The strong IL-1β and cytokine burst resulting from macrophage lysis (and only occurring in mice with sensitive macrophages) can exacerbate LT-induced vascular collapse (Moayeri et al., 2004), but only in some strains. Certain inbred mice such as those in the C3H background (independent of their Tlr4 status) have macrophages that rapidly lyse following LT treatment, but the mice are resistant to toxin doses that kill Balb/cJ mice. C57BL/6J mice, however, can be converted to the same sensitivity of Balb/cJ mice if harboring macrophages that lyse (Moayeri et al., 2004). Recombinant inbred mice derived from DBA/2J and Balb/cJ mice support these findings and suggest at least three loci controlling murine sensitivity (McAllister et al., 2003). Genetic background may provide protection against the cytokine burst seen in mice harboring LT-sensitive macrophages (Moayeri et al., 2004) and it is tempting to hypothesize that genes involved in regulation of endocrine responses may be involved in sensitivity differences (Moayeri et al., 2005). In addition to the LT macrophage sensitivity locus, Nalp1b (Boyden and Dietrich, 2006) another chromosome 11 susceptibility locus originally hypothesized to be inducible nitric oxide synthase (iNOS)(McAllister et al., 2003) was ruled out through mouse knockout studies (Moayeri et al., 2004). The identification of the genetic loci influencing host resistance is important to understanding the role of toxin sensitivity differences in human disease, because thus far tested human macrophages are resistant to Nalp1 mediated macrophage lysis (see later sections).
Unlike mice, intermediate sensitivity states are absent in rats (our unpublished data). Rats either succumb to toxin after injection of 10 μg PA + 10 μg LF by 60–90 minutes or are completely resistant to 25-fold higher doses. In rats survival is thought to be controlled by a single locus (Nye et al., 2007), and a correlation to macrophage sensitivity has been noted, although the susceptibility locus has not yet been mapped. The rapid death in rats is both similar (induction of pleural edema) and very different (death as rapidly as 37 minutes) than lethality in mice (which occurs over days). Thus care should be taken in extrapolating findings from the rat toxin lethality model to mice.
While the advances made in purification of high quality recombinant LF and PA from B. anthracis have allowed for the assessment of LT effects in vivo without the endotoxin issues that plagued E.coli-derived preparations, moderate differences in specific activity of toxin preparations can still alter both in vitro and in vivo analyses of toxin effects. For LT the most sensitive assay for testing potency of LF and PA preparations is the minutes to death in the Fischer rat, much in the manner the assay was used for verification of toxin purity in the 1960s. While most labs use macrophage lysis assays to assess potency of toxin preparations, that assay is not sensitive enough to pick up differences that severely alter responses in animals (i.e., a one well difference in a two-fold dilution curve is invariably associated with loss of LT lethal activity at receptor saturating toxin concentrations in mice). Our laboratory has accumulated extensive data on the potency (as time to death, TTD) of toxin preparations in the Fischer rat model, and how it correlates to LT effects in other species such as mice. 10 μg LT (10 μg LF + 10 μg PA) generally kills a rat in 55–90 min and a toxin preparation with a lower specific activity may prolong TTD in rats and not kill mice at receptor-saturating doses. More importantly, in vitro studies with low activity preparations lead to changes in the dominant pathway of cell death, or altered immune responses (see later sections). Thus, in our view, the many highly variable results obtained both in vitro and in vivo for a good number of reports on LT can be directly linked to the potency of the toxin used which can often translate to orders of magnitude difference in “dose” of toxin treatment. Hence one laboratory kills a sensitive macrophage with 200 ng/ml of each toxin component in under 90 min and another requires 5 hours. The ramifications of simply using higher doses of toxin to compensate for low specific activity include considerations of receptor clearance effects by PA which further complicate analyses.
Perplexingly, all B. anthracis-derived recombinant EF preparations (produced commercially or in various laboratories) to date have specific activities that are orders of magnitude lower than those for EF produced in E. coli or from the original Sterne and Ames B. anthracis strains (unpublished observations). The basis for the variability in EF preparations is unrelated to protein breakdown or sequence alterations and is currently unknown.
LT, through its effects on the p38 pathway, inhibits glucocorticoid receptor (GR)-mediated activation, but not repression of transcription factors (Webster et al., 2003;Webster and Sternberg, 2005). The importance of the hypothalamic-pituatary-adrenal (HPA) axis and the (GR) response in inflammation during infection is well established and inhibition of GR activation by LT has been demonstrated in mice (Webster et al., 2003). Thus, endocrine response differences to underlying infections may explain variation in mouse sensitivities observed for one strain of mouse in different animal facilities. Supporting this hypothesis, LT resistance in mice can be reversed by adrenalectomy or glucocorticoid feeding/injections (Moayeri et al., 2005). Stress and animal housing conditions must be considered when performing LT studies in mice. Proposed LT mouse susceptibility loci may involve endocrine-related pathways (McAllister et al., 2003;Moayeri et al., 2004). As the preferred treatment for shock in the clinic often involves steroid-based control of inflammatory responses, special attention to potential exacerbation of LT-mediated effects may be warranted.
Sublethal doses of ET which do not induce pathological changes can sensitize LT-resistant DBA/2J mice (Firoved et al., 2007). We now know that LPS associated with the EF used in these studies may exacerbate this sensitization effect by ET in the context of the high cAMP response (unpublished work). ET’s striking effects on the adrenal gland (Firoved et al., 2005) and the resulting endocrine perturbation known to alter LT sensitivity in murine models (Moayeri et al., 2005) may account for sensitization. In the rat infusion model, ET was shown to increase heart rate (unlike LT, which decreases heart rate), but the toxins cooperate together to worsen shock (Cui et al., 2007). Care should be taken in interpreting any results of combinatorial ET/LT in vivo studies, as the actual relevant doses of each toxin produced in different animal models through various stages of disease may have very different resultant effects. A better understanding of the contribution of each toxin to the other may be possible through comparing the effects of toxin-isogenic mutants in animal models.
While most studies of the immunomodulatory effects of the toxins are performed in vitro using cell lines or primary cells, there are a small number of studies in which immunomodulatory effects were also verified in vivo. These studies are covered by other reviews in this volume.
The discovery that LT induces a rapid and striking lysis of murine macrophages from some inbred mice (Friedlander, 1986;Friedlander et al., 1993) resulted in a singular pursuit of this cell type as the target of LT function for over a decade. The discovery of LT’s cellular substrates has now led to increased focus on the likely more relevant noncytotoxic immunomodulatory effects this toxin induces in many cell types (see the review of Tournier et al. in this volume). Despite the fact that the classic rapid lysis of murine macrophages has little relevance to human anthrax disease (human macrophages do not undergo LT-induced rapid lysis) the bulk of the scientific literature on LT’s cytotoxic effects still focuses on murine macrophage lysis. Thus, this review will also provide a detailed review of LT cytotoxicity in this particular cell type. It is far more likely, however, that LT effects on another target are responsible for its lethality to the host. Ironically, a classic inverse relationship that has been noted between host macrophage sensitivity to LT and susceptibility to spore infection is often ignored (Lincoln et al., 1967;Welkos et al., 1986).
In recent years many reports of LT ‘cytotoxicity’ to other cell types have left the impression that this toxin induces cell death in many targets. LT’s inactivation of the Mek1/2 pathway, however, is likely to lead to cell cycle arrest in many of these cells, and an accompanying inhibition of proliferation that is often mistaken for cell death in assays that are performed over days. Thus, it is important to review the literature on LT-mediated effects on other cell types very carefully.
Unsurprisingly, ET is not cytotoxic to any cell type.
Macrophages from inbred mice and rats can be divided into LT-sensitive (examples for mice include Balb/cJ, C3H/HeJ, CBA/J, FVB/NJ, SWR/J and for rats Brown Norway and Fischer) and LT-resistant groups (examples for mice include DBA/2J, AKR/J, SJL/J, A/J, C57BL/6J and for rats Lewis and Wistar-Kyoto) (Nye et al., 2007;Roberts et al., 1998). Sensitive macrophages succumb to LT by 90 minutes if subjected to saturating fully active toxin concentrations (1 ug/ml each component). “Resistant” macrophages that do not succumb to the rapid lysis by LT undergo a slower apoptotic death (Muehlbauer et al., 2007;Park et al., 2002) which can even be observed in LT-sensitive macrophages when sublytic doses of toxin or low activity toxin preparations are used (Park et al., 2002;Popov et al., 2002b). The ability to undergo apoptosis, therefore, appears to be universal to all macrophages, but masked by the rapid lysis/necrosis pathway which dominates in LT-sensitive macrophages. We discuss each type of macrophage death in the following sections (SUMMARIZED IN FIGURE 1).
The identification of Nalp1b as the LT sensitivity locus has been the most important leap in the understanding of LT-mediated rapid macrophage lysis (Boyden and Dietrich, 2006). Initially mapped to a single locus on mouse chromosome 11, Ltxs1 (Roberts et al., 1998) and first misidentified as Kif1c (Watters et al., 2001), the Nalp1b gene was definitively implicated as the sensitivity allele through a series of experiments. Expression of a Nalp1bS containing BAC (bacterial artificial chromosome) in Nalp1bR containing resistant mouse macrophages demonstrated that this single gene was dominant in conferring sensitivity and Nalp1bS silencing protected macrophages from rapid lysis. To date, five polymorphic Nalp1b alleles have been described in mice (Boyden and Dietrich, 2006).
Nalp1b is a member of the NLR (Nod-like receptor) protein family. These proteins function in a complex known as the “inflammasome” which is responsible for recruitment and proteolytic activation of caspase-1, which in turn activates the pro-inflammatory cytokines IL-1β and IL-18 (Martinon and Tschopp, 2007). NLRs act as intracellular pattern recognition sensors for danger signals in the cytoplasm (Martinon and Tschopp, 2007;Tschopp et al., 2003). Inflammasome-mediated caspase-1 activation is not only involved in the innate immune response leading to cytokine processing and release, but also cell death in response to many stimuli (Yu and Finlay, 2008).
Early in vivo studies showed that rapid transcription-independent IL-1β release only occurred in LT-injected mouse strains harboring sensitive macrophages (Moayeri et al., 2004;Moayeri et al., 2003). We now know that LT activates caspase-1 in LT-sensitive (Nalp1bS) but not LT-resistant (Nalp1bR) macrophages (Boyden and Dietrich, 2006;Cordoba-Rodriguez et al., 2004) and that this activation is required for rapid lysis (Muehlbauer et al., 2007;Wickliffe et al., 2008b). The Nalp1bS inflammasome is similar to other inflammasomes in requiring an upstream potassium efflux event (Fink et al., 2008;Petrilli et al., 2007;Wickliffe et al., 2008b) and different in that its formation also requires proteasome activity (Fink et al., 2008;Squires et al., 2007;Wickliffe et al., 2008b). Late addition of proteasome inhibitors even after Mek cleavage has occurred can rescue macrophages as long as caspase-1 activation has not occurred (Alileche et al., 2006;Squires et al., 2007;Wickliffe et al., 2008b). Furthermore, the Nalp1bS inflammasome complex also differs from most inflammasome complexes in lacking the adaptor protein ASC (Nour et al., 2009). Caspase-1 activation alone is clearly insufficient for LT-mediated macrophage lysis, as it can be induced in any macrophage, and must occur in the context of a Nalp1bS inflammasome for rapid lysis to ensue (Wickliffe et al., 2008b). We do not know what LT-induced events activate the Nalp1bS inflammasome, or how caspase-1 activation leads to rapid macrophage lysis and what role, if any, Mek cleavage plays in this death pathway. Earlier studies showed no correlation between Mek cleavage and macrophage sensitivity to LT (Pellizzari et al., 2000;Pellizzari et al., 1999). It is possible, however, that the cellular events downstream of Mek cleavage (which occurs equally in Nalp1bS and Nalp1bR macrophages) are sensed differently by Nalp1bS and Nalp1bR proteins. In such a scenario Meks may indeed be the only LF substrates. Alternatively, unidentified LT substrates including potential components of the inflammasome pathway may be involved in initiating cell death. The cleavage of these substrates may also lead to their targeting to the proteasome.
In a recent report NOD2 has was found to be associated with the Nalp1-caspase1 complex and a potential role for NOD2 was suggested for LT-mediated caspase-1 activation (Hsu et al., 2008). Because LT effects were only tested in Nalp1bR macrophages in which LT does not activate caspase-1, it is difficult to interpret the role of LT in the B. anthracis-induced IL-1β release observed in these macrophages. The response to bacterial infection in the LT-resistant macrophages clearly depended on NOD2 and Nalp1bR and was attenuated when LT-deficient mutant bacteria were used, but this finding could have more relevance to the general effects of LT deficiency on bacteria (i.e., growth, division, etc.). It remains to be seen if NOD2 is required for LT-mediated caspase-1 activation in Nalp1bS macrophages.
While the recent discovery of an inflammasome activation requirement for rapid LT-mediated macrophage death has redirected research on this topic, it is useful to revisit the numerous reports on LT effects in sensitive macrophages for potential links to the inflammasome-mediated cell death.
Early reports on LT-mediated superoxide production and antioxidant protection against LT (Hanna et al., 1994) suggest a role for reactive oxygen species (ROS) in LT-mediated cell death. ROS which have been linked to inflammasome activation (Cruz et al., 2007;Meissner et al., 2008) may also control LT-mediated caspase-1 activation. ROS production may be linked to the mitochondrial membrane potential loss that occurs in response to LT (Alileche et al., 2006). Mitochondrial Bcl-2 family proapoptotic Bnip3 proteins have been implicated in both apoptotic and rapid necrotic LT-mediated macrophage death pathways and resistance in Nalp1bS macrophages is correlated with induction of these proteins (Ha et al., 2007a) (see later section). It is unknown if LT-mediated mitochondrial potential loss occurs due to plasma membrane perturbations or if LT directly targets mitochondria.
Among the many treatments reported over the years that protect against LT-mediated macrophage lysis only those which do not effect LF binding, entry and translocation (thus not effecting Mek cleavage), may provide clues to the mechanism of cell death. For this reason the many studies on compounds protecting cells by preventing LF delivery to the cytosol are not discussed here. Among the many treatments protecting against LT-lysis at a step downstream of Mek cleavage, proteasome inhibitors have been the most interesting. The historic observation of macrophage protection by proteasome inhibitors (Tang and Leppla, 1999) has now clearly been linked to LT-mediated inflammasome and caspase-1 activation (Fink et al., 2008;Squires et al., 2007;Wickliffe et al., 2008b). The identity of the proteasome substrate(s) involved in LT-mediated lysis and Nalp1b inflammasome function is unknown, but this target is subject to the N-end rule of protein breakdown, as type-2 destabilizing amino acid derivative inhibitors of this pathway (e.g., phenylalanine amide) are protective and can also synergize with the aminopeptidase inhibitor bestatin methyl ester (Wickliffe et al., 2008a). The only currently identified N-end rule substrate targeted for proteasome breakdown by LT is c-IAP1, a member of the inhibitor of apoptosis protein (IAP) family, although its breakdown is not required for activation of the Nalp1b inflammasome (Wickliffe et al., 2008a) and may be more important to the slower apoptosis induced by LT (see next section).
Recent studies show that heat shock protects against LT lysis at a step downstream of Mek cleavage. Heat shock may be a universal inhibitor of caspase-1 activation through trapping pro-caspase-1 in a large complex as both LT-mediated Nalp1b inflammasome activation and LPS/nigericin-mediated Nalp3 (Nlrp3 or cryopyrin) inflammasome activation can be inhibited by heat shock with no effect on caspase-1 enzymatic activity (Levin et al., 2008).
A series of studies have shown protection with phosphatase inhibitors, calcium channel blockers/antagonists, phospholipase A2 inhibitors, genistein, neomycin, protein kinase C inhibitors, and phospholipase C inhibitors (Bhatnagar et al., 1989;Bhatnagar et al., 1999;Kau et al., 2002;Panchal et al., 2007;Shin et al., 1999;Shin et al., 2000). While in the case of phosphatase inhibitors Mek targeting may be affected, many of these studies need to be revisited to see how the compounds impact the inflammasome pathway. It bears mentioning that the inhibitor dosing used in some these older studies is often not relevant to their physiological effects.
Finally, despite the many reports on transcriptional and translational responses to LT treatment of sensitive macrophages, continued protein synthesis is not needed for LT-induced inflammasome-mediated lysis (Pellizzari et al., 1999;Alileche et al., 2006;Levin et al., 2008). The targets reported in these studies to be up or downregulated often represent the cell’s protective responses to severe stress and cell death rather than LT-specific events. Perplexingly, the time points utilized in many of these studies are often when a substantial percentage of cells are already dead, even by each study’s own death response curves, (Comer et al., 2005b;Kuhn et al., 2006;Sapra et al., 2006), and even when time points are selected where a smaller number of cells are dead, the lytic event has clearly already commenced (Chandra et al., 2005). The only study investigating responses at very early times after LT treatment and prior to any cell death found altered expression of genes under GSK-3β(glycogen synthase kinase 3β) regulation and inhibition of this kinase sensitized both Nalp1bS and Nalp1bR macrophages (Tucker et al., 2003). The upregulation of genes controlled by GSK-3β may be a survival response in the apoptotic death pathway LT induces in “resistant” macrophages (see next section), as GSK-3βinhibition is normally a protective event, but it is difficult to speculate on how this kinase plays a role in caspase-1 mediated lysis. GSK-3βinhibition has also been linked to protection from Mek-cleavage mediated cell arrest induction by LT (Ha et al., 2007b).
Almost all tested macrophages from numerous species which do not succumb to LT through the rapid caspase-1 lysis are subject to toxin-induced apoptosis.
Nalp1bR macrophages succumb to LT over a 16- to 72-hour period through an apoptotic pathway dependent on Mek cleavage (Muehlbauer et al., 2007;Park et al., 2002) and LPS can sensitize them to an even more rapid 8-hour death (Park et al., 2002). The apoptotic death by LT was shown to require inhibition of the p38 protective pathway (Park et al., 2002). Surprisingly, LPS-primed Nalp1bS macrophages in the same study were also shown to undergo apoptosis after LT treatment, possibly because the LF/PA63 mixture used had a low specific activity allowing observation of this apoptosis in the absence of rapid caspase-1 mediated lysis (Park et al., 2002). A third caspase-independent necrotic mechanism of macrophage involving TNF-α has also been suggested for Nalp1b R macrophages sensitized to LT (Kim et al., 2002). Various microbial products were shown to result in a slow LT-mediated necrotic death, which in contrast to the LPS-mediated LT sensitization to apoptosis (Park et al., 2002), did not require classic apoptotic caspase activation (Kim et al., 2002). Interestingly, LT-resistant human macrophages were also sensitized through a TNF-α dependent pathway, but no effect was seen on Nalp1bS murine macrophages (Kim et al., 2002). It is attractive to speculate that macrophage sensitization to LT may occur in real infections through apoptosis or a novel slow necrosis different than the murine caspase-1 dependent rapid lysis and this cell death could have consequences for the infectious process in humans.
Subpopulations of Nalp1bS macrophages become resistant to LT after multiple passages (Ha et al., 2007a) or by treatment with sublytic toxin doses (Salles et al., 2003). Nalp1bS macrophages in the phenotypically “resistant state” have reduced expression of two related proteins, mitochondrial Bcl2/adenovirus E1B-interacting proteins Bnip3 and Bnip3L (Ha et al., 2007a). The p38 shutdown required for LT killing of LPS-sensitized cells (Park et al., 2002) is actually essential to LT-mediated induction of resistance in Nalp1bS macrophages such that p38 inhibitors can mimic LT-induced resistance. It is believed the p38 shutdown leads to reduction of the Bnip3 proteins. Bnip3 overexpression in Nalp1bS macrophages imparts resistance, while siRNA mediated shutdown of expression sensitizes. It is unclear if Bnip3 controls caspase-1 mediated cell death in Nalp1bS macrophages or if it primarily serves a protective role in resistance. Bnip3 proteins may alter mitochondrial protective responses downstream of Nalp1b-mediated activation of caspase-1 or alternatively, influence early events which prevent LT-mediated mitochondrial targeting upstream of potassium flux, inflammasome activation, and caspase-1 activity. Bnip3 have been implicated in both apoptosis and a necrosis pathway involving mitochondrial membrane permeabilization, ROS production and sudden loss of plasma membrane integrity reminiscent of the lysis seen in Nalp1bS macrophages (Vande Velde et al., 2000). The role of these proteins in LT-mediated cell death certainly deserves further study.
While human macrophages thus far tested to date from different donors are resistant to LT-mediated caspase-1 lysis (our unpublished data), human monocytic cell lines (U-937, HL-60 and THP-1) can be sensitized to LT-mediated apoptosis following activation. LT normally induces cell division arrest in these cells over a course of days, without cell death (Kassam et al., 2005), but activation signals make them sensitive to a slow caspase-1-independent apoptosis that is distinctly different than the necrotic death induced by TNF-α mediated “sensitization” of resistant mouse macrophages (Kassam et al., 2005). These findings are not limited to monocytic cell lines, as activated human peripheral blood mononuclear cells undergo apoptosis following inhibition of cell division (Popov et al., 2002a). As will be seen in the next section, dendritic cell (DC) sensitivity to LT is also influenced by activation state (Reig et al., 2008), although maturation signals actually protect in that case.
Human monocytic cell lines, like many cell types, require Mek1 activity for growth. Many reports of “cytotoxicity” by LT in a variety of cell types actually measure the cell arrest induced by LT through inhibition of the Mek1/2 pathway relative to untreated controls. Ha and colleagues have shown that LT induces cell arrest in human monocytic cells through breakdown of cyclin D1/D3, a process which can be reversed by activation of the phosphatidylinositol 3-kinase/Akt pathway. Akt provides protection from cell arrest by inhibition of GSK-3β (Ha et al., 2007b). One can hypothesize that LT-mediated cell cycle arrest may be the basis for the pathological events linked to LT vascular collapse in species such as mice which succumb over a period of days (Moayeri et al., 2003).
Reports on LT effects on DC are highly varied. Human DC (Maldonado-Arocho et al., 2006), murine Nalp1bR DC (Tournier et al., 2005) and even murine Nalp1bS DC (Chou et al., 2008) have been reported to be resistant to LT toxicity over treatment periods as long as 48 hours. On the other hand, rapid proteasome and caspase-1 dependent lysis of DC harboring the Nalp1bS allele and slow apoptosis of murine Nalp1bR DC and human DC has also been reported (Alileche et al., 2005;Muehlbauer et al., 2007). Conflicting reports also exist on the role of activation state in DC sensitivity. One group found that activation signals do not affect DC sensitivity, demonstrating that bone-marrow derived DC (immature) and spleen-derived activated DC had equal LT sensitivity (Alileche et al., 2005). Another group showed DC activation protects against apoptosis of Nalp1bR DC (Reig et al., 2008). Furthermore, DC from mice with the Nalp1bS allele in which caspase-1 was knocked out (eliminating the rapid necrosis pathway and allowing the apoptosis pathway to dominate) were also protected against LT by maturation signals. Although Mek1/2 cleavage is not affected by maturation state, inactivation of these kinases by LT is required for DC apoptotic death, since maturation also protected against Mek1/2 inhibitor-mediated cell death (Reig et al., 2008).
It is unclear if the variability in DC literature reports is linked to the different activation states resulting from the isolation method and tissue source of these cells or other confounding factors. Nevertheless, care should be taken in interpretation of LT effects on DC, not only in the context of cytotoxicity effects, but especially as relating to immunomodulatory functions of LT.
As previously mentioned, induction of cell cycle arrest by LT is a common occurrence in any cell type that requires the Mek1/2 pathway or downstream ERK activity for proliferation. While some studies are careful to make the distinction between cell arrest and cell death, reports on many cell types do not, which may explain the discrepancies in findings for some cell types. The endothelial cell is attractive as a potential relevant in vivo target of LT because of the unique vascular collapse induced by this toxin and the association of the two anthrax toxin receptors with vascular function or angiogenesis. However, the initially reported apoptotsis of HUVECs (Kirby, 2004) may have reflected cell arrest as numerous laboratories have been unable to induce HUVEC cell death with LT. Of course genetic differences in the human source of these cells may also play a role in cell sensitivity. In the initial study, viable cell numbers were measured relative to untreated cells grown in parallel for the same number of days, such that proliferation arrest would not be distinguishable from apoptotic death. FITC-ZVAD staining, however, accompanied by protection using caspase inhibitors supported an LT-mediated apoptosis in these cells. Some investigators have clearly described LT-mediated proliferation arrest in HUVEC (Huang et al., 2008) and SV40-transformed mouse endothelial cells (Depeille et al., 2007), while others studies on HUVEC and human lung microvascular endothelial cells are unclear as to whether apoptosis or cell arrest is being measured (Paddle et al., 2006;Pandey and Warburton, 2004). A number of laboratories have specifically noted the inability to induce apoptosis in HUVECs (Batty et al., 2006;Warfel et al., 2005), but do report proliferation arrest (Batty et al., 2006) and LT-induced loss of barrier function in the absence of any endothelial cell death (Warfel et al., 2005). Thus it remains unclear whether HUVECs actually undergo apoptosis in response to LT, and even then, these cells may not even be reflective of LT-mediated effects on endothelial cells in vivo. LT-induced loss of barrier function, however, would clearly play a very important role in vascular collapse induced by this toxin in vivo (Warfel et al., 2005). The rapidity with which LT induces vascular leakage in a modified Miles-assay in mice also seems to support altered endothelial barrier function independent of cell death (Gozes et al., 2006).
LT is not cytotoxic to neutrophils and lymphocytes, but proliferation arrest has been reported in these cells through Mek pathway shutdown over a period of days (Comer et al., 2005a;Fang et al., 2005;Xu et al., 2008). Conflicting results on epithelial cell sensitivity to LT have also been reported (Lehmann et al., 2009;Paddle et al., 2006;Pandey and Warburton, 2004).
With the exception of one report of cytotoxic effects in zebrafish (Voth et al., 2005) and a recent study suggesting that ET can induce cell cycle arrest in macrophages through activation of nuclear GSK-3β (Larabee et al., 2008), ET has been shown to be nonlethal to all cells on which it has been tested. These include, but are not limited to DC (Chou et al., 2008;Tournier et al., 2005), macrophages (Comer et al., 2006), T-cells (Paccani et al., 2005), neutrophils (Crawford et al., 2006) and human microvascular endothelial cells (Hong et al., 2007). cAMP increases generally do not induce cell death and thus these findings are expected.
The effects of anthrax toxins on various immune cells are reviewed in by Tournier et. al in the current volume and will not be presented in great detail here. A very brief overview is presented for most cell types (SUMMARIZED IN FIGURE 1).
Almost all of LT’s immunomodulatory effects are inhibitory for the innate immune response and are a direct result of the shutdown of the Mek1/2-Erk1/2, Mek 4/7-SAPK-JNK and Mek3/6-p38 signaling pathways (and thus can be replicated with specific Mek activity inhibitors). In macrophages and dendritic cells these effects include impairment of cytokine responses, antigen presentation function, bactericidal ability and phagocytosis. In neutrophils conflicting reports on altered superoxide responses and NADPH oxidase activity as well as chemotaxis have been reported. In T-lymphocytes classic cytokine/chemokines pathways and chemotaxis are disrupted. In B cells proliferative responses and IgM production are inhibited. While LT targeting of the adaptive immune response by manifesting inhibitory effects on B-cells or T-cells is unlikely to play a role in an acute disease like inhalational anthrax, it is clearly possible that B. anthracis benefits from targeting adaptive immunity in non-acute, cutaneous anthrax infections.
ET manifests its many immunomodulatory effects through increase of cellular cAMP and the resulting activations of the protein kinase A (PKA)/cAMP response element binding protein (CREB) and exchange protein directly activated by cAMP (Epac)/Ras-associated protein-1(Rap1) pathways. As this important second messenger has both stimulatory and inhibitory effects on an incredible range of cellular functions, it is expected that ET will eventually be found to have the same wide range of effects. It appears that while LT inactivation of Mek pathways is almost invariably suppressive to innate immune response pathways, cAMP activation of the PKA-dependent CREB pathway or the PKA-independent EPAC pathway results in complicated pro- and anti-inflammatory effects in monocytes, DC and lymphocytes and opposing effects on chemotaxis in different cell types.
LT and ET effects in macrophages, DC, neutrophils and lymphocytes are reviewed in detail by other authors in this volume and we refer you to the review by Tournier et. al for the references pertaining to the findings on various immune cells.
Although the effects of LT on endothelial cells have not been directly investigated in vivo, numerous studies have shown that LT affects angiogenesis and tumor vasculature (Alfano et al., 2008;Depeille et al., 2007;Duesbery et al., 2001;Liu et al., 2008), supporting a targeting of the endothelium in the host. In a study directly linking Mek1/2 inhibition in vivo to vascular dysfunction, LT was shown to induce endothelial cell-death independent permeability in zebrafish vasculature with associated pericardial edema in a manner duplicated by Mek1/2 inhibitors (Bolcome, III et al., 2008). It is difficult to understand how Mek cleavage is linked to resistance differences in mice and rats, however, unless cleavage of these kinases in endothelial cells from resistant animals results in very different effects. The role of Mek cleavage in endothelial cell dysfunction deserves further study.
Perhaps the most important in vitro finding on LT-mediated effects on endothelial cells is the cell-death independent loss of barrier function induced by toxin (Warfel et al., 2005) (a cell-death independent finding replicated in the previously mentioned zebrafish model (Bolcome, III et al., 2008)). LT can also modulate endothelial cell responses to LPS and cytokine stimuli, examples of which include enhancement of TNF-induced VCAM-1 expression and monocyte adhesion to endothelial cells (Steele et al., 2005;Warfel and D’Agnillo, 2008) and inhibition of IL-8 production through transcript destabilization (Batty et al., 2006).. These types of effects are likely not relevant to LT-mediated lethality or vascular collapse (which are cytokine independent), but play an important role in bacterial infections which are accompanied by strong cytokine production. Classic immune cell-mediated vascular damage is likely responsible for the vasculitis associated with anthrax infections (Fritz et al., 1995;Grinberg et al., 2001;Vasconcelos et al., 2003;Zaucha et al., 1998) that is absent in LT treated animals (Cui et al., 2004;Moayeri et al., 2003). Bacterial infection induced vasculitis can be significantly affected by cytokine responses and receptor expression changes (such as that observed with VCAM-1) in endothelial cells. Paradoxically, a number of the thus far in vitro reported LT effects on endothelial cells, such as the inhibition of IL-8 (Batty et al., 2006) or other chemokines/chemokines receptors (van Sorge et al., 2008), or tissue factor expression (Rao et al., 2004) would theoretically be inhibitory to classic shock vasculitis but could contribute substantially to immune cell recruitment in the innate immune response. In a manner opposing LT effects, ET increases transendothelial electrical resistance of endothelial monolayers (Tessier et al., 2007). The toxin is believed to induce vascular leakage in rabbits through the action of neurokinins, prostanoids and histamine rather than alterations in endothelial cell permeability (Tessier et al., 2007).
LT induced platelet aggregation resulting in suppression of human blood clotting, altered P-selectin expression, and endothelial cell interaction have been reported in one study (Kau et al., 2005). The absence of fibrin deposits in mice treated with LT makes it less likely that platelets play a role in the vascular insufficiency induced by LT alone, despite a toxin-induced thrombocytopenic effect (Moayeri et al., 2003). Thus it remains to be seen if LT effects on platelets occur in mice. LT-mediated platelet dysfunction, however, would have great implications for LT’s alteration of the host response to bacterial infection. ET has been reported to suppress thrombin-induced platelet aggregation and clotting function (Alam et al., 2005), likely contributing to the hemorrhaging associated with ET in animal models (Firoved et al., 2005).
A recent study of LT effects in human lung epithelial cells found LT-mediated decreases in transepithelial electrical resistance and loss of barrier function associated with altered cellular tight junctions and focal adhesions over days of toxin treatment (Lehmann et al., 2009). LT-mediated effects were due to increases in F-actin levels in cells, and could be reversed by transfection of cells with uncleavable Mek proteins, implicating these kinases directly in the cytoskeletal changes induced by toxin. If toxin-mediated remodeling of the lung epithelium and functional consequences can be verified in vivo, it will certainly have important implications for anthrax infection pathogenesis. However, it is unlikely these effects contribute to LT-mediated lethality, especially in the rapid death induced in the rat model, but they may be suggestive of similar cytoskeletal-mediated changes linked to barrier function alterations in endothelial cells.
When reviewing the individual studies on LT, ET, and B. anthracis infection effects in various cell types, the curious observations emerge that LT appears to shut down the cytokines induced by B. anthracis, and that the numerous ET-induced cAMP-mediated effects theoretically negate LT-mediated effects. As an example, ET activates CREB, which is a macrophage survival factor downstream of p38 (Park et al., 2005). LT in turn inhibits the p38 pathway through Mek3/6 cleavage and many of LT’s effects in cells are due to blocking this pathway. ET activation of CREB has been shown to promote survival of LT-treated macrophages that normally would undergo apoptosis (Park et al., 2005).
The difficulty of designing experiments to study the combined effects of LT and ET cannot be underestimated, since it is very challenging to accurately model the complex events that occur in vivo during an infection. The relative amounts of each component, timing and site of production, and factors effecting binding and uptake (such as receptor clearance) are unknown for most stages of anthrax infection. Thus, the doses and treatment times of each toxin component used in co-treatment experiments are unlikely to accurately mimic events in an actual infection. Furthermore, interpretation of findings on ET and LT combined effects may be complicated by ET’s upregulation of PA receptors on monocytic cell populations (Maldonado-Arocho et al., 2006).
In the limited number of studies that have investigated the combined effects of the toxins in T-cells and DCs, LT and ET were found to both cooperatively synergize in manifesting some inhibitory effects, but also to have inverse effects on select pathways, likely reflective of the pro- and anti-inflammatory effects of cAMP signaling (Chou et al., 2008;Paccani et al., 2005;Tournier et al., 2005).
The anthrax toxins are believed to play a role in the early stages of infection and the fate of germinated bacteria, prior to eliciting any systemic toxic effects. Reports suggest that toxin action is required for release of the large numbers of dividing intracellular bacteria from cells (Dixon et al., 2000) and that toxin is necessary for spore survival against macrophage killing (Guidi-Rontani et al., 2001). Because most spore studies are performed in LT-sensitive macrophages, it is difficult to extrapolate from these results to LT-resistant human macrophages. The real situation in vivo may be a complicated balance between macrophage killing of spores, germination, and dissemination after macrophage killing. In a study using a toxin-receptor negative variant of the LT-sensitive RAW264.7 macrophage cell line injected in mice, these cells were able to provide better protection against a challenge with Ames spores in animals than their receptor-expressing counterparts, suggesting a protective role for macrophages that bacteria can overcome with toxin (Cote et al., 2008). In a way, these studies provide an indirect comparison of LT-sensitive and LT-resistant macrophages for their ability to protect against spores, and it appears that LT-resistance of macrophages (in this case through absence of receptor) would allow better killing of spores and control of infection. Paradoxically, however, macrophage sensitivity to LT has been shown to have an inverse correlation to spore sensitivity in animals (Lincoln et al., 1967;Welkos et al., 1986), suggesting LT-mediated macrophage killing may be detrimental to establishing infection in some species. While translocation of LT produced by germinating spores from phagolysosomes into cytosol (through unknown mechanisms in the absence of receptor) has been reported to kill LT-sensitive macrophages and allow the escape of germinated bacteria (Banks et al., 2005), it is still unclear if these findings have relevance to LT-resistant human macrophages. It will be important to decipher LT’s role in controlling early infection steps including germination and survival in DC/macrophages in hosts harboring LT-resistant Nalp1b (or Nalp1) alleles.
The volume of literature on anthrax and anthrax toxins between 1953 and 1968 is almost as remarkable as the absence of publications in this field from 1969 through the early 1980s. The research work done in the earlier period on the anthrax toxins can now be recognized as remarkable not only for its intensity, but also for the ability of researchers to deduce many key features of the toxins, in spite of not having modern tools such as recombinant DNA technology and powerful protein purification methods. In a period where animal welfare regulations were far more lax, and prior to the advent of cell biology, almost all early studies on anthrax toxin function were performed in vivo. Interestingly, in a manner that parallels the more recent period, once the lethal activity of LT was discovered, almost all work focused on this toxin, with little attention being directed to ET. In this section we will revisit the toxin studies from this early period and review important observations in the context of the material reviewed above.
The anthrax toxins were discovered in 1954 when Harry Smith and colleagues observed that bacteria-free filtered serum from B. anthracis infected guinea pigs was lethal when injected in mice and guinea pigs (Smith et al., 1955b;Smith et al., 1955a;Smith and Keppie, 1954). The filtrates also induced edema in guinea pigs and rabbits when injected intradermally. The recognition that an exotoxin was responsible for the lethal effects of anthrax disease led to purification of two separate toxin activities which required a third common protein (now known as PA) to manifest their in vivo functions (Smith and Stanley, 1962;Stanley et al., 1960;Stanley and Smith, 1961). The combinatorial toxins to this day remain named after the early observations made about their in vivo effects (lethality and edema).
The initial reports of toxin-induced vascular collapse and shock in mice and guinea pigs were key aspects of the discovery of the toxin (Smith et al., 1955a). The discovery of the rapid death of the Fischer rat and the extensive pleural edema associated with LT-mediated rat death came after numerous improvements in the purification of the toxins and was the basis for much of the early in vivo investigations (Beall et al., 1962;Stanley et al., 1960;Stanley and Smith, 1961;Thorne et al., 1960). Early electron microscopy studies in rats using partially purified LT (which likely had some EF present) showed the elevation of cytoplasmic processes of lung endothelial cells away from basement membranes without destruction of endothelial cells, leading to a conclusion that these pathologies were strictly secondary mechanical manifestations of pulmonary edema (Beall and Dalldorf, 1966). Capillary thrombosis in the lungs was likely due to contaminating ET in early preparations (based on our current understanding of these toxins’ histopathology in animals). Interestingly light microscopy studies did not show the changes in rat endothelium, much like the LT-induced changes we have failed to observe in vivo over years of performing histopathological analyses in rodents. Recently, however, our EM analyses of the hearts of mice treated with LT show striking effects on cardiac capillary endothelium (Moayeri et al., 2009). Even with all the emphasis on the lung, early studies recognized that LT induced vascular permeability in rat skin and peritoneum as well and that this toxin caused a general effect on the vasculature (Beall and Dalldorf, 1966). Thus investigators concluded that “anthrax toxin acts directly on the membranes of the capillary endothelial cells.”
Interestingly, one of the first primary uses of the Fischer rat model was as an assay for toxin purification and specific activity. Most of the early papers that followed the 1962 discovery of this rapid death always used the minutes to death of the Fischer rat model to assess purity and potency of purified toxins and to assign units of activity to preparations (including (Beall et al., 1962;Buzzell, 1967;Fish et al., 1968b;Haines et al., 1965)). As discussed in previous sections, our laboratory has now returned to this method to assess toxin preparation quality.
In the 1960s some investigators began doing side-by-side spore infection and toxin-infusion or injection studies in rats, rabbits, guinea pigs, monkeys and chimpanzees in order to definitively implicate toxin in the pathophysiology associated with spore infection. In all these models, investigators using a variety of physiological measurements noted that the effects seen in terminal anthrax disease and with LT treatment of animals were virtually identical, but simply occurred earlier with toxin treatment (Klein et al., 1962;Klein et al., 1966;Lincoln et al., 1967;Walker et al., 1967). Much can be learned from the many simple observations made in these studies: One of the most interesting findings from these studies was the inverse relationship seen among animal species and strains between the ability to establish anthrax infection and susceptibility to LT, first based on striking differences between the NIH black and Fischer rats. Although the NIH black rat was very resistant to toxin treatment, it was highly sensitive to spore infections, with very high numbers of germinated bacteria in circulation at the same times that the Fischer rat had very low numbers. Yet the much lower number of bacteria in the Fischer rat was sufficient for a more rapid death in this highly LT-sensitive animal (Klein et al., 1963;Lincoln et al., 1967). Investigators interpreted this as reflecting a two phase disease process: a first phase of resistance to germination and outgrowth, and a second stage of succumbing to toxins. We can now suggest that the basis for these differences was the sensitivity of the DC and macrophages in Fischer rats, which would not allow the germination in or transport of the bacteria by these cells, resulting in a relative “resistance” to spore infections.
Another repeated findings that fascinated early investigators of the LT-induced shock state, and is often-mentioned by Dr. Smith in talks reviewing toxin effects, was the extreme hyperglycemia associated with LT-mediated shock (Eckert and Bonventre, 1963;Klein et al., 1966;Slein and Logan, Jr., 1960;Walker et al., 1967). This finding may fit with what is now known about LT’s potent inhibitory effects on GR function both in vitro and in vivo (see sections on LT in vivo effects). This receptor is essential for glucogenesis and many cellular functions controlling glucose levels in vivo.
While investigators often revisit the early literature on LT-induced shock, one aspect that is often ignored is the striking and controversial work suggesting that LT targeting of the central nervous system (CNS) could explain the induction of pleural edema, vascular dysfunction and respiratory arrest. In these early reports, Rhesus monkeys and chimpanzees challenged with toxin showed electroencephalogram (EEG) abnormalities a few hours before respiratory cessation and myocardial failure. Spore infections showed the same findings in these animals, and antisera against toxin could reverse these effects if administered within 8 hours of toxin challenge. PA alone and LF alone did not cause the same changes. LT preparations injected directly into the spinal fluid of monkeys, caused 6–10 minute deaths, following rapid drops of left ventricular heart function (Vick et al., 1968;Walker et al., 1967). However, histopathological effects in the CNS were not found, indicating the blood-brain barrier was unaffected by the toxins (Bonventre et al., 1967). Interestingly, abnormal EEG activity was also found associated with the rapid 58 minute death in rats (Fish et al., 1968a). In both the monkey and rat studies, EKG, heart rate and blood pressure were normal until minutes before death (Fish et al., 1968a;Vick et al., 1968). Knowing the timeline of LF entry and delivery to the cytosol, it is hard to imagine how a 6 -10 minute death would result from the toxin’s intracellular actions. The negative controls for these experiments, however, seem extensive and accurate. LT targeting of the CNS is an area that should be revisited, especially when considering the rapidity of death in the rat model.
Early suggestions of LT effects on cardiac function came not only from the monkey studies described above, but from studies where injection of B. anthracis culture supernatants in dogs affected multiple cardiovascular functions such as blood pressure, cardiac contractile force, cardiac output and heart rate (Liu and Williams, 1971). The effects of LT on the heart are today an active area of study. Norepinephrine as a cause for the pleural edema was also postulated because culture supernatants from B. anthracis cause vascular changes similar to this mediator (Williams et al., 1967) and it was released after LT injections, but at very low levels (Slein and Logan, Jr., 1960).
Investigators in the field of anthrax toxin research would benefit from a regular re-reading of the older studies in this field, if only to gain an appreciation for the number of astute observations on toxin function that were deduced strictly from in vivo studies. In revisiting this literature, there are a large number of observations on the in vivo effects of toxin which could be important and useful in guiding investigators today.
The discovery of the cellular pathways targeted by LT and ET has allowed researchers to predict and provide a better understanding of LT and ET’s cytotoxic and immunomodulatory effects on many cell types in vitro. In a way, the anthrax toxins can be thought of as highly specific “drugs” with potencies exceeding those of nearly all small molecule compounds. One would have predicted that the in vivo action of these toxins would be more focused, limited, and therefore more easily defined than for traditional drugs. However the involvement of the Mek pathways and cAMP in multiple cellular functions in virtually every cell in the body makes it far more difficult to identify the crucial and relevant in vivo targets for LT and ET in real infections. It is an easier first step to demonstrate the sequalae of Mek inhibition or cAMP signaling in response to the toxins in a particular cell type in vitro and in fact, researchers would be hard pressed to find a cell type unaffected by these toxins in vitro. It is an entirely different problem to find relevance for the findings in vivo. Thus far, the links between affected cellular pathways and the roles the toxins play in infected hosts remain mostly unknown. Ironically, the thus far most often studied anthrax toxin effect, the rapid lysis of murine macrophages, likely has little relevance to anthrax disease.
The challenge now facing anthrax toxin researchers is to extend their analyses so as to understand how the toxins contribute to human infection. There are many important questions needing analysis which come to mind, but some include the following. Is Mek cleavage essential or sufficient for LT-induced lethality? If not, what other LF substrates mediate its in vivo effects? If essential, in which target cells does MEK cleavage matter? How does LT induce vascular collapse? Is there an evolutionary advantage to harboring a sensitive Nalp1b allele, and are studies on the unique LT-mediated mouse macrophage lysis helpful in understanding the disease process in human hosts? Does the Nalp1b allele status determine the response in other cell types such as endothelial cells? What are the relevant intersections of the pathways perturbed by LT and ET during infection, especially in the context of the multitude of other pathways altered during bacterial infection? How many of the regularly reported immunomodulatory findings in vitro can be correlated to relevant cellular events during anthrax infection?
Like most problems in biology, the problem here lies is the complexity of the human organism, and the difficulty of understanding it from a “systems biology” view. While there is great value in testing toxin effects in many in vitro cell systems, these can provide only a certain amount of information that is of value in understanding an infectious process in mammals. Reductionist approaches using toxin treatment in animal models, while useful, will also only go so far. In the end the true role of these toxins in disease must be studied in the context of infections in different animal species, and especially in non-human primates. This would be the only system in which the complex interplay between LT and ET, already shown to act in contradictory directions in many signaling pathways, can be understood in the context of other virulence factors and the sequalae of bacterial infection.
Before the major advances in molecular and cellular biology, anthrax toxin research began in animals, with simple observations of toxin-mediated pathologies relying on crude toxin preparations prepared directly from infected hosts, and a large volume of what is known today about the actual functions of the toxins was initially observed and documented in these animal systems, even if it was not fully understood. Perhaps with the many additional tools now available to us, future studies on anthrax toxin should again focus on in vivo models of infection, as frequently pointed out by Harry Smith (Smith, 2002), whose seminal work started us on this journey (Smith et al., 1955a;Smith and Keppie, 1954).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.