Histotoxic Clostridial Infections

ABSTRACT

The pathogenesis of clostridial myonecrosis or gas gangrene involves an interruption to the blood supply to the infected tissues, often via a traumatic wound, anaerobic growth of the infecting clostridial cells, the production of extracellular toxins, and toxin-mediated cell and tissue damage. This review focuses on host-pathogen interactions in Clostridium perfringens-mediated and Clostridium septicum-mediated myonecrosis. The major toxins involved are C. perfringens α-toxin, which has phospholipase C and sphingomyelinase activity, and C. septicum α-toxin, a β-pore-forming toxin that belongs to the aerolysin family. Although these toxins are cytotoxic, their effects on host cells are quite complex, with a range of intracellular cell signaling pathways induced by their action on host cell membranes.

GENERAL CHARACTERISTICS OF HISTOTOXIC CLOSTRIDIAL INFECTIONS

Many pathogenic clostridial species cause potentially fatal soft tissue infections in humans and animals because of their ability to produce extracellular protein toxins (1, 2). The ability of these pathogens to produce spores that are resistant to environmental stress is an important factor in the epidemiology of these diseases. Disease pathogenesis involves the growth of the clostridial pathogen in the tissues and extensive tissue destruction, which is the result of the action of extracellular toxins. Typical histotoxic clostridial diseases include human gas gangrene or myonecrosis and blackleg in cattle. Although there are several clostridial species that are responsible for these syndromes (Table 1), this review will focus on histotoxic infections caused by Clostridium perfringens and Clostridium septicum, primarily because these are the species that have been the subject of the most extensive molecular and functional studies.

TABLE 1.

Histotoxic clostridial infections

Syndromea	Causative agent	Host	Major toxinsb	References
Traumatic gas gangrene	C. perfringens type A	Humans	α-toxin	12, 17
			Perfringolysin O (θ-toxin)
Traumatic gas gangrene	C. novyi	Humans	α-toxin	151
	C. septicum		α-toxin	129
	C. histolyticum		Lethal factor	152
	C. sordellii		Lethal toxin (TcsL)	153
			Hemorrhagic toxin (TcsH)
Nontraumatic gas gangrene	C. septicum	Humans	α-toxin	3, 4
	C. fallax	Humans and animals	Not known	154
Necrotizing pancreatitis	C. perfringens type A	Humans	α-toxin	155
Clostridial myonecrosis	C. perfringens types A, B, C, and D	Sheep, cattle, and other animals	α-toxin	29
			β-toxin
			ε-toxin
Blackleg	C. chauvoei	Cattle, sheep, and other ruminants	Cytolysin A (CctA)	156

^a The terms “gas gangrene” and “clostridial myonecrosis” are used interchangeably in this review.

^b Note that the different α-toxins referred to in this column are not necessarily structurally or functionally related.

Although the histotoxic clostridia cause severe, rapidly developing infections, these potent toxigenic bacteria still must be regarded as opportunistic pathogens. All of these infections require predisposing conditions to cause disease, often a traumatic wound that enables the entry of spores from the soil or gastrointestinal tract to gain access to ischemic internal organs or soft tissues of the body. Alternatively, in nontraumatic myonecrosis caused by C. septicum, gastrointestinal malignancy is commonly associated with the onset of disease, with tumor development leading to ulceration and necrosis of the gastrointestinal mucosa and entry of the gastrointestinal microbiota into the circulation and subsequent infection at distal sites (3, 4). Clostridial necrotizing soft tissue infections, especially those caused by C. perfringens, Clostridium sordellii, and Clostridium novyi, also have resulted from the subcutaneous injection of contaminated black tar heroin (5–7).

PATHOGENESIS OF CLOSTRIDIAL MYONECROSIS CAUSED BY C. PERFRINGENS AND C. SEPTICUM

As previously described (8), the pathogenesis of these infections involves three distinct stages, irrespective of whether the source of infection involves a traumatic wound. Stage 1 involves an interruption to the blood supply such that the redox potential in the tissues drops to a level that facilitates spore germination and/or the growth of the infecting clostridial cells. Stage 2 involves bacterial growth and establishment of the conditions for toxin production. In C. perfringens, regulation of toxin production primarily utilizes the VirSR two-component signal transduction system (9) and an accessory growth regulator-like quorum sensing system (10, 11), as well as other regulatory networks. Stage 3 encompasses the toxin-mediated cell and tissue damage that leads to necrosis, systemic toxicity, and clinical disease (8).

C. perfringens type A is the major cause of traumatic gas gangrene, although other histotoxic clostridia may also be responsible for this sporadic, but fulminant and often fatal, disease (Table 1) (1). The major toxin produced by gas gangrene strains of C. perfringens is α-toxin, a zinc metallophospholipase that has both phospholipase C and sphingomyelinase activity (12). α-Toxin was the first bacterial toxin to be shown to have enzymatic activity (13). Two independent studies using different experimental approaches have shown that α-toxin is essential for C. perfringens-mediated myonecrosis. First, immunization studies in mice using recombinant α-toxin variants purified from Escherichia coli, and therefore devoid of any other C. perfringens toxins, showed that the C-terminal domain of α-toxin was immunoprotective (14). Second, mutation of the α-toxin structural gene (plc or cpa) abrogated the ability of the bacterium to cause clostridial myonecrosis in a murine model. Virulence was restored by complementation in trans with a recombinant plasmid containing the wild-type plc gene (15), thereby fulfilling molecular Koch’s postulates and providing proof of the essential role of this toxin in disease.

The other major toxin produced by C. perfringens type A is perfringolysin O, or θ-toxin, a pore-forming toxin that is a member of the cholesterol-dependent cytolysin family (16, 17). Although mutation of the perfringolysin O structural gene, pfoA, did not eliminate the ability to cause disease (18, 19) subsequent studies of plcpfoA double mutants (20) provided evidence that perfringolysin O has a synergistic effect with α-toxin on disease pathogenesis (20, 21).

Recent studies have involved the concurrent analysis of the transcriptomes of both the host and C. perfringens in a murine myonecrosis infection (22). The results showed that many host genes involved in the innate immune response to infection were upregulated in C. perfringens-infected muscle tissues. In the C. perfringens cells, upregulated genes included those encoding potential adhesins and proteins associated with the cell envelope.

C. septicum is the major causative agent of nontraumatic gas gangrene, often associated with a gastrointestinal malignancy (4). The major toxin produced by C. septicum is also called α-toxin, but it is not related to C. perfringens α-toxin. C. septicum α-toxin is an aerolysin-like pore-forming toxin that is secreted as an inactive prototoxin that is cleaved near the C terminus by membrane-bound furin or furin-like proteases to form the active toxin (23). The toxin then oligomerizes on the membrane and subsequently inserts to form a membrane pore, which ultimately leads to cell lysis (24). Genetic studies have shown that α-toxin is essential for C. septicum-mediated myonecrosis in mice (25). Mutation of the α-toxin structural gene, csa, leads to a loss of virulence, which is restored by complementation with the wild-type csa gene.

STRUCTURE AND FUNCTION OF C. PERFRINGENS α-TOXIN

α-Toxin is a 43-kDa single polypeptide that is a hemolytic, cytotoxic, dermonecrotic, and lethal extracellular toxin (26–29). Moreover, it induces contractions of blood vessels and smooth muscle, aggregation of platelets, superoxide production, and cytokine release. Importantly, together with perfringolysin O, it is responsible for the hallmark histopathological feature of C. perfringens-mediated myonecrosis, the paucity of the polymorphonuclear leukocyte (PMNL) influx into the infected lesion (15, 19–21, 30).

Determination of the crystal structure of α-toxin revealed that it has an α-helical N-terminal domain (amino acids 1 to 246) that is responsible for its enzymatic activity, a flexible linker (amino acids 247 to 255), and a C-terminal (amino acids 256 to 370) β-sandwich membrane-binding domain (31) (Fig. 1). The two domains also are linked by hydrophobic surface interactions between their neighboring faces. The N-terminal domain has significant sequence similarity to phospholipase C enzymes from Bacillus cereus (32), Clostridium bifermentans (33), and Listeria monocytogenes (34). It comprises nine densely packed α-helices and resembles the structure of B. cereus phospholipase C (31, 35), which has enzymatic activity, but is neither toxic nor hemolytic. There are three loops in the N-terminal domain: loop 1 (residues 70 to 90), loop 2 (residues 135 to 150), and loop 3 (residues 205 to 215). These loops serve a vital role in the direct interaction between the toxin and the membrane-binding component. The C-terminal domain comprises eight-stranded antiparallel β-sheets (31). This domain plays a role in membrane binding and is required for hemolytic and toxic activity (28, 36). It is absent from nontoxic phospholipase C enzymes (31). It has structural similarity to the C-terminal domain of pancreatic lipase, the N-terminal domain of soybean lipoxygenase, and the C2 phospholipid-binding domains of eukaryotic intracellular proteins, such as synaptotagmin I, which are involved in signal transduction and vesicular transport (37, 38).

FIGURE 1.

Structure of α-toxin. The structure of α-toxin was obtained from the Protein Data Bank (code 1QM6) and was created using MacPyMOL. A ribbon representation is shown, colored from blue (N-terminal) to red (C-terminal). Two zinc metal ions in the active site are shown as red spheres.

The active site in the N-terminal domain is situated within a solvent-accessible cleft that contains two zinc ions and an exchangeable divalent cation. His-148 and Glu-152 dock with one zinc ion, which is critical for the catalytic site of the toxin. His-11 and Asp-130 tightly dock with the other zinc ion, which is needed for maintenance of the structure. His-68, His-126, and His-136 dock with an exchangeable divalent cation, which is required for binding to target cell membranes (39). The external loop regions (residues 70 to 90, 135 to 150, 205 to 215, 265 to 275, 290 to 300, and 330 to 340) bind to phospholipids in the host cell membranes (31). Thr-74 is present in one of these loops, and although the substitution derivative T74I is still active and can cleave a water-soluble substrate, its membrane-damaging activity toward red blood cells and phosphatidylcholine-cholesterol liposomes is decreased compared to wild-type toxin (40, 41). However, the structure of the T74I variant is identical to that of the wild-type toxin (41). The T74I substitution results in altered recognition of the glycerol moiety in the phospholipid, without affecting the function of the active site, supporting the hypothesis that the amino acid 70 to 90 loop is important for phospholipid recognition (42, 43).

Computational modeling of α-toxin and phosphatidylcholine revealed that Tyr-57 and Tyr-65, which exist at an external site near the catalytic cleft and are in proximity with the 70 to 90 loop, are responsible for substrate interaction (44). The replacement of these Tyr residues with Ala did not have any effect on phospholipase C and sphingomyelinase activity but had a marked effect on the membrane-damaging activity (hemolysis of sheep red cells and disruption of liposomes). These residues exist in or near the entrance of the catalytic cleft and are essential for the membrane-damaging effect of α-toxin, because they are required for entry of the catalytic cleft into the hydrophobic region of plasma membranes (44).

Several studies have shown that the calcium-binding C-terminal domain is responsible for membrane binding (28, 31, 36, 37). Two acrylodan-labeled C-terminal domain derivatives, S263C and S365C, bound to lipid bilayers and displayed a chromogenic shift, demonstrating entry of the C-domain of the toxin into the hydrophobic regions of the liposomes (36). The data provide evidence that the C-terminal domain of the toxin binds to the membrane and that the amino acid 70 to 90 loop and Tyr-57 and Tyr-65 serve an important role in the recognition of target membranes (Fig. 2). Other studies have shown that substitution of the C-terminal domain residues Asp-269, Asp-336, Tyr-275, and Tyr-331, which were predicted to be involved in membrane binding (31), were required for hemolytic and cytotoxic activity (45).

FIGURE 2.

Binding model of α-toxin and membrane phospholipids. The structure of α-toxin was obtained from the Protein Data Bank (code 1QM6) and was created using MacPyMOL. Phospholipids are shown as light gray spheres for head groups and light gray lines for tail groups. The head group of a phospholipid, from the outer membrane into the active site of α-toxin, is displayed. Amino acids that may play a role in membrane interaction are shown.

INTERACTION OF C. PERFRINGENS α-TOXIN WITH HOST CELLS

Membrane-Damaging Activity on Phospholipid Bilayers

The effect of α-toxin on host tissues is caused by the phospholipase C- and sphingomyelinase-mediated cleavage of phospholipids in the host cell membranes. The resultant disruption of the phospholipid bilayer together with the end products of these cleavage reactions, diacylglycerol and ceramide, respectively, activates lipid metabolism in host cell membranes and plays a role in host cell damage (27, 28).

The composition of the phospholipid bilayer modulates the extent of the α-toxin-mediated membrane disruption (46). In another study that measured the release of carboxyfluorescein from various liposomes, α-toxin-mediated release, or liposomal leakage, decreased with increasing chain length of the hydrocarbon residues in the phospholipid bilayer (40). It was concluded that membrane disruption was related to the fluidity of the membrane and double bonds in the acyl chain. Subsequently, several phosphatidylcholines (C18:0/C18:1) with a double bond in the sn-2 fatty acyl chain were prepared (47). The phase transition temperature value was lowest when the cis-double bond was positioned at the center of the C18:1 sn-2 acyl chain and increased continuously as the double bond was moved toward either end of the acyl chain. α-toxin-mediated carboxyfluorescein release from various liposomes was highest when the cis-double bond was positioned at the center of the sn-2 fatty acyl chain, resulting in a bell-shaped curve (47). Accordingly, the membrane-disrupting activity of the toxin is strongly associated with the membrane fluidity of target membranes.

There is evidence that α-toxin also promotes perfringolysin O-induced membrane damage in phospholipid bilayers. Hydrolysis of phosphatidylcholine residues of liposomes that contained cholesterol increased the amount of free cholesterol in the lipid bilayer and enhanced perfringolysin O-dependent cytolytic activity (48). These data may at least partly explain the synergistic effects of α-toxin and perfringolysin O that were previously observed in C. perfringens infections of mice (20).

Activation of Phospholipid Metabolism and Hemolysis

α-toxin provokes contraction of isolated rat ileum and aorta tissues in a dose-dependent manner through stimulation of the phospholipid metabolic pathway (49, 50). In isolated rat organs, α-toxin activates the phosphatidylinositol metabolic pathway and then the arachidonic acid cascade, thereby promoting thromboxane production. Contraction caused by the toxin was evoked by thromboxane A2 generation induced by arachidonic acid (49, 50). The metabolism of arachidonic acid results in the production of bioactive lipid mediators, including leukotrienes, thromboxanes, and prostaglandins (27). These bioactive lipid mediators play key roles in enhancing the local inflammatory reaction and lead to vasoconstriction, which is responsible for inducing anoxia in the early stages of bacterial infections. These effects presumably contribute to the capacity of C. perfringens to proliferate in the host and cause disease.

Treatment of rabbit neutrophils with α-toxin promotes the formation of diacylglycerol, which is followed by adhesion of the cells to extracellular matrix proteins and the generation of superoxide ions (Fig. 3B) (51). α-Toxin induces diacylglycerol generation through activation of intrinsic phospholipase C enzymes by a pertussis toxin-sensitive GTP-binding protein. Moreover, it increases the phosphorylation of neurotrophic tyrosine kinase receptor type 1 (TrkA), leading to 3-phosphoinositide-dependent protein kinase 1 activation (52). Therefore, α-toxin independently causes stimulation of intrinsic phospholipase C and protein kinase 1 activity through TrkA receptor activation. Both phenomena synergistically augment the activity of protein kinase C θ, leading to the formation of superoxide ions via activation of the mitogen-activated protein kinase (MAPK) cascade (Fig. 3B) (52). In human lung carcinoma (A549) cells, α-toxin mediates the activation of the p38 MAPK, NF-κB, and extracellular signal-regulated kinase 1/2 (ERK1/2) through TrkA activation, which leads to interleukin-8 (IL-8) release (53) (Fig. 3B).

FIGURE 3.

Relationship between phospholipid metabolism and the biological activities of α-toxin. (A) Signaling events involved in α-toxin-induced hemolysis of sheep or rabbit erythrocytes. (B) Signaling events involved in α-toxin-activated generation of O₂^– in neutrophils and in α-toxin-mediated release of IL-8 from A549 cells. Abbreviations: SM, sphingomyelin; PIP2, phosphatidylinositol 4,5-bisphosphate; PLC, phospholipase C; DG, diacylglycerol; PC, phosphatidylcholine; SMase, sphingomyelinase; PLD, phospholipase D; PA, phosphatidic acid; PKCθ, protein kinase Cθ; PI3K, phosphatidylinositol 3-kinase; PIP3, phosphatidylinositol 3,4,5-trisphosphate; CDase, ceramidase; CER, ceramide; SPH, sphingosine; S1P, sphingosine 1-phosphate; DGK, DG kinase; PDK1, phosphatidylinositide dependent kinase 1; DRM, detergent-resistant membrane; TrkA, neurotrophic tyrosine kinase receptor type1; MAPKK, mitogen-activated protein kinase kinase; Erk1/2, extracellular signal-regulated kinase 1/2.

Treatment of rabbit erythrocytes with α-toxin induces the biphasic formation of phosphatidic acid (54) (Fig. 3A). α-Toxin-induced formation of phosphatidic acid is activated in the earlier stage (30 s) by intrinsic phospholipase C and in the later stage (ca. 25 min) by intrinsic phospholipase D. Both processes involve stimulation of the G-protein-signaling pathway by toxin-mediated membrane phospholipid cleavage (55) (Fig. 3A).

Sheep erythrocytes have higher amounts of sphingomyelin but do not contain phosphatidylcholine. Cleavage of sphingomyelin by α-toxin leads to increased levels of ceramide and sphingosine 1-phosphate (56), which is essential for toxin-induced hemolysis (57) and activates the sphingomyelin metabolism system (Fig. 3A) (56). The resultant cleavage of N-nervonoic sphingomyelin (C_24:1-SM) occurs in a detergent-resistant membrane component (57) (Fig. 3A). Pertussis toxin blocks the production of C_24:1-ceramide caused by α-toxin and blocks red cell hemolysis, which indicates that intrinsic sphingomyelinase, which can cleave C_24:1-SM to C_24:1-ceramide, is regulated by a pertussis toxin-sensitive G-protein in plasma membranes. In addition, the activation of Rho protein by α-toxin increases sphingosine kinase activity. Subsequently, the ceramides are quickly metabolized to sphingosine, indicating that the toxin also activates ceramidases (57). These results indicate that the cleavage of C_24:1-SM to sphingosine 1-phosphate by α-toxin in lipid rafts is essential for toxin-induced hemolysis of sheep red cells (Fig. 3A).

Interaction of α-Toxin with Gangliosides

Amino acid sequence analysis of botulinum toxin and tetanus toxin showed that the ganglioside-GT1b- and lactose-binding site is determined by the existence of the conserved peptide motif, H.....SXWY.....G (ca. 30 amino acids [aa]), with the Tyr and Trp residues being particularly critical (58). A similar motif (H.....SWY.....G, amino acids 68 to 93) is present in α-toxin (53), in the region (amino acids 55 to 93) that is located at the interface between the N-terminal and C-terminal domains (42). It has been demonstrated that the 70 to 90 loop in α-toxin plays a role in binding to the cell surface (43) (Fig. 2). Computational docking studies indicated that Trp-84 binds to the sialic acid residue on ganglioside GM1a via an aromatic stacking interaction and a hydrogen-bonding interaction and that Tyr-85 associates with the galactosamine residue on GM1a via an α-glycosidic bonding interaction (59). Replacement of Trp-84 and Tyr-85 in α-toxin, which correspond to the ganglioside recognition site of botulinum and tetanus toxins, significantly reduced binding to ganglioside GM1a-containing liposomes. W84A and Y85A derivatives had markedly decreased capabilities to stimulate TrkA (59). Ganglioside GM1a assembles with a TrkA molecule on the cell surface and stimulates TrkA and ERK1/2 (60). It has been reported that ganglioside GM1 clustering leads to the condensation of TrkA in lipid rafts and the stimulation of downstream signaling targets (61). α-Toxin-mediated diacylglycerol production leads to diacylglycerol flip-flop movement that affects the dynamics of cell membranes (Fig. 4). Therefore, it serves in GM1a clustering and intrinsic phospholipase C-γ1 activation through TrkA, triggering various signal transduction activities (62–64).

FIGURE 4.

Model of α-toxin-induced membrane dynamics and clustering of the GM1a and TrkA complex. α-Toxin binds to GM1a and, through the phospholipase C activity of the toxin itself, causes diacylglycerol (DG) production at the outer membrane. The flip-flop movement of DG affects membrane dynamics, facilitating clustering of GM1a and TrkA activation by phosphorylation. Activation of TrkA results in the activation of PLCγ-1, leading to the increased production of DG. Finally, DG formation results in the enhanced activation of TrkA, triggering activation of signal transduction pathways, which induce neutrophil activation. Abbreviations: PIP₂, phosphatidylinositol 4,5-bisphosphate; PLC, phospholipase C; DG, diacylglycerol; PC, phosphatidylcholine; TrkA, neurotrophic tyrosine kinase receptor type 1.

α-toxin causes higher cytotoxicity in a ganglioside-deficient cell line (Don Q) (65) and can be internalized by dynamin-mediated endocytosis (66, 67). The toxin binds to caveolin-1 both on the cell membranes and in endosomal vesicles, and it also induces ceramide production in the cell membrane (28), which generates negative membrane curvature and facilitates endocytosis (68). Internalized α-toxin is transported to the early and late endosomes and lysosomes. Lysosomes are damaged during progressive α-toxin internalization into Don Q cells. In addition, the internalization of α-toxin is also mediated through the stimulation of MEK/ERK signaling (66, 67). It has been reported that the toxin causes reactive oxygen species generation via PKC, MEK/ERK, and NF-κB signaling activities in Don Q cells (66, 67). Therefore, α-toxin perturbs the host cell signaling cascade and acts not only on the cell membrane, but also in the cytoplasm.

Effect of α-Toxin on Hematopoietic Cells

The first line of defense of the innate immune system involves neutrophils, which play an important role in the phagocytosis and subsequent elimination of pathogenic bacteria (69, 70). Normally, the number of neutrophils is sustained in a steady state through granulopoiesis, which is accelerated to replenish neutrophils during bacterial infection (71–73). Pattern recognition receptors, such as Toll-like receptors (TLRs), are responsible for the recognition of structural components of microorganisms (74, 75). TLR2 and TLR4 have been identified as being pivotal for the recognition of Gram-positive or Gram-negative bacteria by recognizing cell wall components such as peptidoglycan and bacterial endotoxin (lipopolysaccharide), respectively (76–79). During bacterial infection, these bacterial components stimulate the production of granulocyte colony-stimulating factor (G-CSF), which is a glycoprotein that promotes the proliferation, survival, and differentiation of neutrophils and their progenitor cells (80) from endothelial cells and monocytes, resulting in the acceleration of granulopoiesis (81). G-CSF-deficient and G-CSF receptor-deficient mice exhibit chronic neutropenia and reduced infection-driven granulopoiesis, leading to impaired ability to eliminate infecting bacteria (82, 83). Thus, granulopoiesis is precisely regulated to replenish neutrophils during bacterial infection. However, C. perfringens can still cause life-threatening infections, and the mechanisms by which C. perfringens avoids these host defenses are still not completely understood.

C. perfringens-mediated myonecrosis progresses so rapidly that death, which is associated with muscle destruction, shock, and multiple organ failure, sometimes precedes an accurate diagnosis (84). C. perfringens infection is characterized by a greatly reduced PMNL influx to the site of the infection (19, 85–87), suggesting that the bacteria disrupt neutrophil-based innate immunity to evade host defenses. α-Toxin mediates the formation of platelet-leukocyte aggregates, leading to vascular occlusion, and induces a marked reduction in microvascular perfusion (45, 86, 88), which is thought to impede neutrophil extravasation (89). Recently, it was shown using a mouse model of infection that C. perfringens infection reduced mature neutrophils in bone marrow, in peripheral blood, and in C. perfringens-infected muscle, in an α-toxin-dependent manner (90). α-Toxin inhibits neutrophil differentiation and impairs the replenishment of differentiated mature neutrophils in the peripheral circulation, resulting in reduced recruitment of neutrophils to the C. perfringens infection site. The half-life of peripheral neutrophils is less than 12 hours in vivo (91), meaning that the number of mature neutrophils required to counter a bacterial infection is sustained through continuous granulopoiesis. Therefore, blockage of both granulopoiesis and PMNL diapedesis into the tissues, as a result of platelet aggregation, leads to a reduction in cell numbers in a short period of time. Thus, C. perfringens infection impairs the host immune system, which could explain the both scarcity of PMNL at the site of infection and a propensity for polymicrobial infections in many patients (Fig. 5).

FIGURE 5.

Inhibition of granulopoiesis and erythropoiesis in a C. perfringens-infected host. Infection with C. perfringens diminishes mature hematopoietic cells in the bone marrow, leading to impairment of replenishment of differentiated neutrophils and erythrocytes in the peripheral circulation and resulting in host innate immune deficiency. The major virulence factor of C. perfringens, α-toxin, plays an important role in this phenomenon by interfering with cell differentiation of myeloblasts and erythroblasts.

α-Toxin is essential for disease pathogenesis (18–21, 30), and blockage of neutrophil differentiation is dependent on its phospholipase C and sphingomyelinase activities (90). In human lymphocytes, S. aureus sphingomyelinase disrupts lipid rafts, which are cholesterol-rich plasma membrane microdomains (92). Lipid rafts function as platforms of signaling molecules involved in the regulation of cell differentiation in many cell types (93, 94). Ceramide, the hydrolysis product of sphingomyelinase, also is known to be a lipid messenger implicated in various cellular responses, including apoptosis, differentiation, and cell growth (95). Recently, C. perfringens α-toxin was reported to perturb the integrity of lipid rafts, leading to the blockage of neutrophil differentiation (90). The detailed molecular mechanism behind this blockage has not been elucidated, but it may be important to focus on the relationship between the role of sphingomyelinase and cell differentiation to shed further light on host-pathogen interactions in clostridial myonecrosis.

Effect of α-Toxin on Erythropoiesis

α-Toxin has been reported to lyse erythrocytes from various species. It activates T-type Ca²⁺ channels and increases intracellular Ca²⁺, which plays an important role in the hemolysis of horse erythrocytes (96). The toxin activates the sphingomyelin metabolic system, leading to the hemolysis of sheep erythrocytes (56, 57), and the lysis of rabbit erythrocytes involves α-toxin-mediated activation of endogenous phospholipase C activity (54, 55). Recently, Bacillus anthracis lethal toxin was reported not only to induce hemolysis, but also to inhibit erythroid differentiation of cord blood-derived CD34⁺ hematopoietic stem cells, leading to the suppression of erythropoiesis (97). The toxin-induced suppression of erythropoiesis is potentially part of a lethal toxin-mediated pathophysiology in anemia and hypoxia. Similarly, it was reported that mature erythroblasts were preferentially decreased when isolated mouse bone marrow cells were treated with α-toxin, and subsequent experiments revealed that the blockage of erythroid differentiation was involved in the reduction in mature erythroblasts by α-toxin (98). Massive intravascular hemolysis and severe anemia have been reported in C. perfringens-infected patients (99), although it is a relatively uncommon event. The detailed mechanisms by which C. perfringens infection causes severe anemia have not been elucidated, but blockage of erythropoiesis by α-toxin may be important (Fig. 5).

Experiments carried out using an inactive α-toxin H148G variant revealed that its enzymatic activity was required for the α-toxin-mediated reduction of erythroblast levels (98). Furthermore, the cell surface expression of a lipid raft marker, GM1, decreased in association with erythroid differentiation, and α-toxin affected lipid raft integrity in erythroid progenitors, leading to the impairment of erythroid differentiation. During erythroblast enucleation, the clustering of lipid rafts at the furrow between incipient reticulocytes and pyrenocytes is observed, and an inhibitor of lipid rafts impedes enucleation (100). Therefore, lipid rafts play an important role in the regulation of erythropoiesis, but the detailed mechanism by which the disturbance in lipid raft integrity by α-toxin impairs erythroid differentiation remains unclear.

STRUCTURE AND HOST CELL INTERACTIONS OF C. PERFRINGENS PERFRINGOLYSIN O

Perfringolysin O (θ-toxin) is a pore-forming hemolytic toxin that belongs to the cholesterol-dependent cytolysin (CDC) family (16, 101), which includes streptolysin O from Streptococcus pyogenes, listeriolysin O from L. monocytogenes, pneumolysin from Streptococcus pneumoniae (17), and related cytolysins produced by several other clostridia (102). The perfringolysin O structural gene, pfoA, is located on the chromosome (103, 104). Perfringolysin O is elaborated by most, but not all, strains of C. perfringens (104–107), and mature perfringolysin O (53 kDa, 473 aa) (108) is secreted as a water-soluble monomer and binds to host cell membranes in a cholesterol-dependent manner (102). Binding to the cell membrane initiates oligomerization and the formation of a pore complex that can contain up to 50 monomeric subunits and is responsible for cell lysis.

Perfringolysin O monomers have a hydrophilic, β-sheet-rich structure that can be divided into four distinct domains (109, 110) (Fig. 6). Domain 1 contains a seven-stranded antiparallel β-sheet flanked by helices and loops. Domain 2 forms an elongated β-sheet structure. Domain 3 comprises α-helices and β-sheets, and domain 4 (the C-terminal region) is divided into two distinct β-sandwich regions. Domain 4 also contains a highly conserved 11-aa Trp-rich motif (ECTGLAWEWWR) or undecapeptide loop that is near the C-terminus. This structure subsequently was shown to be similar to that of members of the eukaryotic membrane attack complex/perforin family (111).

FIGURE 6.

Structure of perfringolysin O. The structure of perfringolysin O was obtained from the Protein Data Bank (code 1PFO) and was created using MacPyMOL. A ribbon representation is shown, colored from blue (N-terminal) to red (C-terminal).

Perfringolysin O and most of the other cholesterol-dependent cytolysins bind directly to cholesterol in the host cell membrane. Contrary to what was thought for many years, the undecapeptide is not the receptor-binding motif (112). Instead, there is a Thr-Ile pair located in a nearby loop (L1) of domain 4 that is responsible for binding to cholesterol (113). The role of the undecapeptide is to assist in anchoring the toxin monomer to the membrane and to couple membrane insertion with oligomerization (112). The next step in the oligomerization process is a conformational change of two α-helical bundles in domain 3 to form two longer amphipathic β-hairpins; this extended β-hairpin structure inserts into the host cell membrane, initially to form toxin dimers and then to form a large oligomeric β-barrel pore that contains 35 to 45 monomers. The biological effect of large pore formation is to induce cell lysis by osmotic stress (114). In addition, perfringolysin O can interfere with cell signaling events by activating TLR4 and inducing inducible nitric oxide synthase expression and tumor necrosis factor-α (TNFα) secretion in bone marrow-derived macrophages (115) and by stimulating the degradation of Ubc9, thereby reducing the posttranslational SUMOylation of host proteins (116). Therefore, perfringolysin O has effects on both membrane integrity and internal cell signaling.

Perfringolysin O works synergistically with α-toxin to cause gas gangrene by promoting leukostasis and intravascular coagulation (19–21, 117). The result is decreased blood flow (87, 88) and the accumulation of PMNL and macrophages in the blood vessels around the site of infection, which leads to vascular leukostasis and contributes to the paucity of leukocyte infiltration to the site of infection (20, 21). Similarly, intramuscular injection of a crude C. perfringens toxin preparation causes and promotes the formation of intravascular platelet aggregates (87). Leukocytes and fibrin accumulate rapidly in the aggregate, leading to obstruction of the blood vessels and the production of thrombi that reduce local blood flow (87, 88).

Perfringolysin O upregulates the expression of adherence molecules, including platelet-activating factor, CD11b, and endothelial adherence molecules (118–120). In addition, it stimulates the expression of intercellular adherence molecule 1 and platelet-activating factor in endothelial cells (19, 118, 119, 121). The toxin triggers platelet/leukocyte and platelet/platelet aggregation by activating the platelet fibrinogen receptor gpIIb/IIIa synergistically with α-toxin (86, 122). Moreover, it has been reported to impair the polymerization of actin filaments in leukocytes and the migration of neutrophils in response to chemoattractants (118, 123). These events lead to the formation of intravascular platelet/leucocyte/fibrin aggregates, and the adherence of the aggregates to the vascular endothelial cells results in vascular injury, vessel obstruction, hypoxia, and tissue destruction (1, 30, 124). In addition, a direct cytotoxic effect of perfringolysin O on endothelial cells has been demonstrated, which also impairs leukocyte transmigration (30, 88). Disruption of the endothelium contributes to progressive edema (30). α-Toxin impairs granulopoiesis, leading to the reduction of neutrophils, and a direct cytolytic effect of perfringolysin O on leukocytes has been reported (123). Therefore, perfringolysin O works synergistically with α-toxin to reduce the level of leukocytes. Finally, C. perfringens cells can escape from the phagosome of macrophages and survive, and perfringolysin O is an important factor in this escape process (125, 126). These data are in agreement with genetic studies that showed that α-toxin and perfringolysin O act synergistically in the disease process (19, 20).

In the latter stage of clostridial gas gangrene, cardiovascular collapse, tachycardia, and multiorgan failure are observed in patients (1). Toxins such as perfringolysin O and α-toxin are released into the blood and can act at a distance from the site of infection. Perfringolysin O reduces systemic vascular resistance, increases cardiac output, and decreases the heart rate without a drop in mean arterial pressure (117, 127). In addition, it promotes the release of inflammatory cytokines, such as TNFα, IL-1, IL-6, platelet-activating factor, and prostaglandin I2, by acting synergistically with α-toxin, which contributes to the symptoms of toxic shock, including hypotension, hypoxia, and reduced cardiac output (30, 128).

FUNCTION AND HOST CELL INTERACTIONS OF C. SEPTICUM α-TOXIN

C. septicum is the primary causative agent of nontraumatic clostridial myonecrosis and enumerates an α-toxin that is essential for disease and is responsible for reduced PMNL influx into the infected lesions (25). C. septicum α-toxin is a member of the aerolysin family of pore-forming toxins (129, 130). These toxins include aerolysin from Aeromonas hydrophila, ε-toxin and C. perfringens enterotoxin (CPE) from C. perfringens, and parasporin toxins produced by Bacillus thuringiensis. These toxins all form small, often heptameric, β-barrel pores in host cell membranes (130). Although the crystal structures of several members of the aerolysin family have been determined, there is no such structure available for α-toxin. However, molecular modelling of α-toxin shows that it does not have the small-lobe lectin-binding domain (D1) of aerolysin, known as the aerolysin pertussis toxin domain (131), but has homologous domains to the other three aerolysin domains (Fig. 7) (132).

FIGURE 7.

Crystal structure of aerolysin and molecular model of C. septicum α-toxin. Domains are indicated. Note that domain 1 (D1) of aerolysin is not present in α-toxin. AT, C. septicum α-toxin. Reprinted with permission from reference 137. Copyright (2006) American Chemical Society.

α-Toxin is synthesized as a preprototoxin, with a 31-aa leader sequence that is cleaved upon secretion from the bacterial cell and a 45-aa C-terminal extension that is cleaved by host proteases such as furin prior to pore formation (23, 102, 133). The resultant cleaved propeptide prevents premature oligomerization (134). The 46-kDa prototoxin binds to glycophosphatidylinositol-anchored protein receptors located in lipid rafts in the host cell membrane (135). After protease cleavage on the cell membrane, the 43-kDa mature toxin undergoes a conformational change and oligomerizes to form a heptameric prepore complex that is required for pore formation (24). Cleavage of the C-terminal extension enables the formation of a transmembrane β-barrel structure, with each monomer contributing one extended β-hairpin loop to the β-barrel that forms the transmembrane pore (130, 132).

The analysis of substitution derivatives of α-toxin has yielded insights into the functional domains of the toxin (132, 136–138). Y155F and S189C α-toxin derivatives had reduced toxicity for CHO cells but were still cleaved by host proteases and could bind to glycophosphatidylinositol-anchored receptors and form oligomers. A S189C/S238C derivative retained its ability to bind, act as a substrate for furin cleavage, and form oligomers but was not toxic (136). Like aerolysin, α-toxin has a Trp-rich motif (amino acids 302 to 312) in its C-terminal domain (138). Several of the residues in this motif, including three Trp residues, are essential for binding and/or cytotoxic activity (137, 138). Fluorimetric studies using Cys-substituted derivatives led to the identification of the amphipathic transmembrane β-hairpin (Lys-203 to Gln-232) that spans the host cell membrane. Deletion of this region eliminated the toxin’s ability to form pores and kill cells, but the resultant protein could still bind, undergo cleavage, and form oligomers (132). Subsequently, saturation Ala or Cys substitutions were made across the remaining residues in the α-toxin molecule, and the resultant proteins were purified and analyzed for their biological and biophysical properties (137). The results showed that the only α-toxin Cys residue, Cys-86 in domain 1, was required for receptor binding. In summary, derivatives altered in receptor binding all had substitutions in domain 1, but proteins with altered oligomerization properties had changes in domains 1 and 3. Several of these mutants have been examined in the mouse myonecrosis model (139). The results showed that both the amphipathic β-hairpin and Cys-86 were essential for disease, providing evidence that receptor binding and pore formation are important in pathogenesis.

Pore formation is essential for C. septicum-mediated myonecrosis (129, 140), but pore-mediated osmotic cell lysis is not the only consequence of intoxication by α-toxin. Treatment of C2C12 mouse myoblast cells with α-toxin leads to a rapid Ca²⁺ influx into the cell but does not induce apoptosis. Instead, α-toxin induces programmed cellular necrosis (140), which is consistent with the fact that it is more active against nucleated cells (141, 142). α-Toxin-mediated Ca²⁺ influx resulted in the activation of proteolytic calpain activity, a reduction in lysosomal integrity, leading to cathepsin release, mitochondrial dysfunction and ATP depletion, and release from the nucleus of the histone-binding protein HMGB1 (140). Induction of this programmed necrotic pathway was dependent on the ability to form pores; treatment with a substituted α-toxin lacking the β-hairpin loop did not induce a Ca²⁺ influx or programmed necrosis. All of these changes are consistent with α-toxin-mediated pore formation causing the induction of the programmed cellular necrosis pathway (140). Other studies have shown that α-toxin treatment of Madin-Darby Canine Kidney (MCDK) cells causes a rapid K⁺ efflux, ATP depletion, and caspase-3-independent necrosis and nonapoptotic cell death (129). Subsequently, it was shown that pore formation by α-toxin led to ERK, c-Jun N-terminal protein kinase, and p38 phosphorylation and activation of their resultant MAPK pathways and the resultant dose-dependent release of TNFα (143). C. septicum α-toxin also leads to a significant reduction in vascular perfusion of muscle tissues, as shown by comparative intravital microscopy studies using culture supernatants from an isogenic series of C. septicum strains (88).

CONTROL AND TREATMENT OF HISTOTOXIC CLOSTRIDIAL INFECTIONS

The treatment of myonecrotic infections caused by C. perfringens and C. septicum is complicated by the rapid onset and progression of disease. Death can result within 24 hours of initial symptoms. Standard treatment regimens involve surgical debridement of the infected tissues, which needs to occur as early as possible (2), and treatment with penicillin and clindamycin (144). The effectiveness of the penicillin on its own has been questioned (145), and there is experimental evidence that clindamycin and metronidazole, alone or in combination with penicillin are more effective in mice (145, 146). Often, surgical removal of an infected limb is the only option available in life-threatening infections.

In other studies, mice lacking TNF-α were protected from C. perfringens α-toxin-induced lethality (147), demonstrating that TNF-α release is essential for the lethal effect of the toxin. Treatment with erythromycin blocked α-toxin-induced release of TNF-α from peripheral neutrophils and the activation of TrkA and ERK1/2, demonstrating that toxin-induced TNF-α release is related to stimulation of the ERK/MAPK signaling pathway through TrkA. Anti-TNF-α antibodies, but not anti-IL-1β or anti-IL-6 antibodies, blocked the death of mice and intravascular hemolysis by α-toxin. Compounds that block the release of TNF-α therefore may be valuable in treating patients with C. perfringens infection. Recently, experiments with C. perfringens infections in mice provided evidence that treatment with opioids such as buprenorphine stop the development of disease (148), but this treatment option has not been tested in clinical settings.

Several studies have investigated the use of recombinant α-toxin variants to vaccinate against C. perfringens-mediated gas gangrene, primarily from a military or biosecurity perspective or for elderly patients that are at risk. Initial studies showed that the isolated nontoxic C-terminal C2 lipid-binding domain of α-toxin (amino acids 247 to 370) was highly immunoprotective in mice (14). Mice immunized with this formaldehyde-treated recombinant antigen were protected from the effects of α-toxin and from challenge with a 10× LD₅₀ dose of C. perfringens. By contrast, mice immunized with the equivalent N-terminal enzymatic domain (amino acids 1 to 249) produced immunoreactive antibodies, but these antibodies did not protect against the toxin or disease. These results were confirmed in a subsequent, more detailed, vaccination study that also showed that mice vaccinated with the C-terminal domain and then challenged in a murine infection model had reduced thrombosis and an increased PMNL influx into the site of infection (149). To further define the immunoprotective region of the C-terminal domain, the potency of several recombinant fragments (amino acids 251 to 370, 281 to 370, and 311 to 370) fused to glutathione S-transferase (GST) were examined in an active vaccination model (150). Vaccination with the 251 to 370 and 281 to 370 GST-fusions protected against the toxicity of α-toxin and against C. perfringens challenge, whereas the amino acid 311 to 370 derivative yielded only partial protection. Despite these three promising studies, no commercial C. perfringens vaccine is currently available.