The toxins are A/B type, formed by the association of three components. Components A bear the enzymatic activity. Edema factor (EF, 89 kDa) for ET is a calmodulin-dependent adenylate cyclase, that increases intracellular cAMP concentrations [4,16], while lethal factor (LF, 90 kDa) for LT is a zinc-dependent metalloprotease cleaving specifically the N-terminus of most mitogen-activated protein kinase kinases (MAPKK or MEK) [4,17]. This MEK cleavage disrupts signaling cascades essential in cell proliferation, cell cycle regulation and immune function, such as ERK 1/2, JNK/SAPK and p38, signaling pathways (review in [4,7]). The B component, involved in binding cell receptor, is common for both toxins: it is named protective antigen (PA, 83 kDa), after its immunogenic properties (review in [9,18]). The mechanisms of cell penetration by toxins have been thoroughly deciphered over the past twenty years and can be separated into three major steps: receptor binding, internalization, membrane translocation (reviewed in [9,18]). PA₈₃ binding to specific cell receptors (ANTXR1 or TEM-8 for Tumor Endothelial Marker-8 and ANTXR2 or CMG-2 for Capillary Morphogenesis Protein-2) [19] allows its cleavage by a furin-like protein into two subunits of 63 and 20 kDa, the latter being released (Figure 1). A third co-receptor named LDL receptor protein (LRP)-6 has been proposed [20], while its effective role has been discussed and may be cell-type specific [21,22]. Interestingly, a group has demonstrated that anthrax toxin receptors interact with LRP6 to control Wnt signaling [23].

PA₆₃ subunits associate in heptamers to form a pre-pore, redistributed to the cell surface in lipid rafts. This allows the binding of EF or LF components in a stoichiometric ratio of 3 LF/EF to 7 PA. The toxin-receptor complex is then internalized by clathrin-dependent endocytosis [9,24,25]. A pH decrease within early endosomes results in the translocation of these factors in multivesicular bodies (MVB), finally merging with the intracellular membrane of late endosomes. This last step facilitates the translocation of LF into the cytoplasm, while EF stays associated to the membrane of the late endosome [26]. Its perinuclear localization generates intracellular cAMP gradients from the cell nucleus to the periphery. Interestingly, mouse mutants for each toxin receptor have been produced, showing that CMG-2 plays a unique role as a major receptor of anthrax toxin in vivo [27].

The cytoskeleton is a cellular scaffolding system whose functions include broad fundamental cell processes such as morphology and plasticity maintenance, movement, signal transduction, membrane and organelle trafficking. The cytoskeleton consists of three filament systems integrated into a complex network regulated by associated proteins: (i) actin microfilaments; (ii) microtubules; and (iii) intermediate filaments. As anthrax toxin effects have been described only on actin filaments and microtubules, we will restrict our description on these two systems.

Microfilaments are formed by a globular protein called actin, which exists in unpolymerized (G-actin) and polymerized (F-actin) forms. F-actin is composed of two parallels strands of actin monomers. The dynamic and spatial organization of the actin cytoskeleton is regulated at multiple levels by a variety of proteins that control numerous processes such as nucleation, polymerization, stabilization, branching and cross-linking of actin filaments [28,29,30]. Moreover, these regulatory and structural proteins enable cells to form and remodel functional structures like stress fibers, lamellipodia, filopodia, phagosomes and endocytic vesicles [30,31,32]. These macromolecular scaffolding structures are regulated by a complex interplay of Rho GTPases, kinases and phosphatases, which are again affected by bacterial pathogens or toxins [33,34]. The Rho GTPases are master regulators of cytoskeletal dynamics and cell shape [35], immune responses [36,37] and phagocytosis [38]. The three best-known members of the Rho GTPases family include RhoA, Cdc42 (Cell division Cycle 42) and Rac1, all of which can act as actin regulator switches by cycling between an inactive GDP- and active GTP-bound states [28,34]. In their active GTP-bound state, Rho GTPases interact with downstream effectors involved in the dynamic rearrangement of actin and microtubule filaments. Activation of RhoA induces the formation of stress fibers that are contractile bundles of polymerized actin containing myosin motor proteins. Rac1 regulates the formation of membrane ruffles or lamellipodia and Cdc42 controls the formation of filopodia or finger-like cell protrusions at the cell periphery [31,34,35].

Microtubules form well-organized hollow tubes composed of covalent association of alpha/beta tubulin heterodimers. These protofilaments radiate from the microtubule-organizing center (MTOC) located at the centrosome in the cytoplasm, in order to allow directional flow of proteins and organelles to a specific location [39,40]. The complex organization of microtubules is responsible for cell polarity, leading to MTOC’s movement toward the site of phagocytosis [41]. The Rho GTPases and many microtubule-associated proteins (MAPs), motor proteins like dyneins, and kinesins also interact with microtubules [37,40].

As the cytoskeleton intervenes in many dynamic cell processes, it has to be controlled by many regulatory proteins and is the target of multiple signaling pathways. As a result, the cytoskeleton network is a prime target for pathogens and their virulence factors.

The effects of LT on the cytoskeleton have been relatively recent discoveries. Most of the studies have focused upon actin in endothelial and epithelial cells, as well as immune cells like macrophage or neutrophils (Figure 2).

ET induces significant morphological and cytoskeletal changes in different mammalian cells [63], including macrophages [10,64,65], primary human microvascular endothelial cells (HMVEC) [12] and neutrophils [57] (Figure 2). Some morphological differences have been observed between cells after ET treatment: (i) formation of filopodial protrusions in mammalian cells [63,64] versus reduction in macrophage [10]; (ii) rounded morphology in macrophages and mammalian cells [10,63,64] versus flattened morphology in human microvascular endothelial cells (HMVEC) [12]. In any case, every cell type had reduced spread morphology, a lowered F-actin content and actin redistribution to the cell margin in a time-dependent manner [10,57,63,64].

Some results support a role for the cAMP-dependent Protein Kinase A (PKA) pathway in the reorganization of actin network [10,57,63,64]. PKA is a serine-threonine kinase that phosphorylates many cytoskeletal proteins, including actin, microtubules and intermediate filaments. The role of Exchange Protein Activated by cAMP (Epac) is less clear, as one study on mammalian cells suggest a lack of Epac activation [63], while another suggests an activation of Epac and Ras-proximate-1 (Rap1) pathway alone [12], or associated with PKA signaling [10]. The disruption in the balance of Epac/PKA activity seems to be responsible for these cytoskeleton’s effects. As for LT effects, it clearly appears that ET-related cytoskeleton disruption and intracellular signaling are cell type-dependent. So far, no studies have explored the potential effect of ET on microtubules.

Because of its biochemical adenylate cyclase activity, ET may be compared to toxins with identical functions: ExoY of Pseudomonas aeruginosa [66] or adenylate cyclase toxin of Bordetella pertussis [67]. B. pertussis, the causative agent of whooping cough, produces toxins including an adenylate cyclase (CyaA) which also possesses hemolytic activity. It appears clearly that CyaA causes morphological changes [68,69,70], including membrane blebbing in erythrocytes [70]. Recent work by Kamanova et al. demonstrated that the adenylate cyclase activity of CyaA causes transient and selective inactivation of the GTPase RhoA in mouse macrophages and activation of cofilin, leading to massive actin rearrangements. These latter effects wane faster at high toxin concentration and are accompanied by a formation of membrane extensions referred as lamellipodia or membrane ruffling [71]. Consequently, cAMP signaling of CyaA toxin rapidly halts complement-mediated phagocytosis of macrophages [71] and human neutrophils [69,72,73].

Soluble adenylate cyclase ExoY of P. aeruginosa, injected by a type III secretion system in a host also generates a cAMP pool in the cytosol and mimicked using a forskolin activated soluble adenylyl cyclase I/II (sACI/II) by several authors ([74] and reviewed in [75]). Activation of sACI/II or directly ExoY generates a large cAMP increase that induces endothelial cell retraction [74,75], with an intact cortical F-actin network and a decrease of MLC phosphorylation [74] and probably a role in bleb-niche formation in epithelial cells [76].

Phagocytosis is one of the first innate defense mechanisms involved for pathogen scavenging. Moreover, phagocytosis activates the adaptive immune system by antigen presentation. As B. anthracis escapes so efficiently from the innate and adaptive immune response, phagocytosis disruption by anthrax toxins does not come as a surprise.

In vivo, LT inhibits primary peritoneal macrophages fluorescent microspheres phagocytosis after LT-intraperitoneal injection [77], even though cultured nonhuman primate alveolar macrophages are still capable of spore phagocytosis after LT treatment [78]. Recently, Kau et al. demonstrate that LT suppresses the phagocytosis of J774 macrophage cells at low doses without influencing the MAPK pathways during early infection [79], implicating another pathway. Nevertheless, these effects are unclear and do not affirm that phagocytosis is inhibited by LT.

The effect of ET on human neutrophils [80] or human macrophages [10] shows a reduction, even an inhibition, of the phagocytic capacity of these cells. This is largely due to an inability to undergo actin remodeling, required for phagocytic activities. However, it clearly appears that aspects of ET-related actin cytoskeleton on phagocytic capacities of B. anthracis Ames spores depend on the cell type [10].

The cytoskeleton disruption by anthrax toxins and potential consequences on phagocytosis is likely important during later stages of infection and would benefit pathogen survival. The ability of these toxins to reduce or inhibit phagocyte functions represents a dangerous immuno-evasive mechanism where the extracellular vegetative cells can rapidly multiply and be eliminated with difficulty.

Corruption of vascular integrity by ET has been known since its discovery in the mid 1960s, as it was named after its effects observed following subcutaneous injection [81]. These observations could be related to clinical records in oro-pharyngeal anthrax, an uncommon form of the disease characterized by impressive neck swelling [82]. Interestingly, subcutaneous versus intravascular injection of ET are not symmetrical, as they do not lead to the same outcome. On one hand a local reversible edema is only observed, while on the other hand the toxin is lethal [83]. This observation has one corollary: toxins act differentially when in the blood lumen or out of the vessel. When both anthrax toxins enter the bloodstream, they corrupt vascular homeostasis leading to a cardio-vascular collapse (review in [8,84]). From a pathogen’s standpoint both toxins facilitate pathogen spread, as they breach the vascular barrier between the blood and tissues leading to bacilli proliferation in multiple organs (review in [84]). These results suggest that ET may have been selected throughout evolution for its effects at the last stage of the disease when it is released into the bloodstream. Recent studies have confirmed that LT opens the blood-brain barrier favoring meningitis, one of the most frequent complications of systemic anthrax [85,86].

Surprisingly, anthrax toxins target the vascular endothelial system but are still poorly described mechanistically (let alone the cardiac effects). LT disrupts in vitro vascular endothelial cells by increasing actin stress fibers and disorganizing VE-cadherin at the adherents junctions (AJ) gasket [87] and protein zona occludens-1 at tight junctions (TJ) [85]. In vivo LT increases vascular permeability in zebrafish [88]. ET has also significant disrupting effects on cadherin in AJ [89]. Strikingly, mechanistic insights have been brought by a group using Drosophila as a model to deconstruct host-pathogen interactions [90,91]. They have very elegantly shown that ET and LT vascular effects are caused through a coordinate disruption of Rab11/sec15 exocyst [89]. The exocyst complex regulates docking of specific cargo vesicles to AJ. By deregulating AJ, both toxins corrupt cytoskeleton organization and vascular permeability. Another effect of toxin on vascular permeability is the formation of transendothelial macro-aperture (TEM) [92]. Very recently ET has been shown to induce TEMs, thus increasing peripheral vascular permeability [93]. More interestingly, the cytoskeleton senses the formation of TEM through a protein called Missing in Metastasis (MIM) that senses de novo membrane curvature, ultimately driving Arp2/3-dependent actin polymerization [93]. These very well designed studies clearly demonstrate the role of cytoskeleton in maintaining vascular integrity that is challenged by ET.