|Home | About | Journals | Submit | Contact Us | Français|
Sulfur mustard (SM), a chemical weapon first employed during World War I, targets the skin, eyes, and lung. It remains a significant military and civilian threat. The characteristic response of human skin to SM involves erythema of delayed onset, followed by edema with inflammatory cell infiltration, the appearance of large blisters in the affected area, and a prolonged healing period. Several in vivo and in vitro models have been established to understand the pathology and investigate the mechanism of action of this vesicating agent in the skin. SM is a bifunctional alkylating agent which reacts with many targets including lipids, proteins, and DNA, forming both intra- and intermolecular cross-links. Despite the relatively nonselective chemical reactivity of this agent, basal keratinocytes are more sensitive, and blistering involves detachment of these cells from their basement membrane adherence zones. The sequence and manner in which these cells die and detach is still unresolved. Much has been discovered over the past two decades with respect to the mechanisms of SM-induced cytotoxicity and the intracellular and extracellular targets of this vesicant. In this review, the effects of SM exposure on the skin are described, as well as potential mechanisms mediating its actions. Successful therapy for SM poisoning will depend on following new mechanistic leads to develop drugs that target one or more of its sites of action.
Sulfur mustard (SM), or mustard gas, is a bifunctional alkylating agent (bis[2-chloroethyl] sulfide) (see Fig. 1 for structure). It was first synthesized in the early 19th century by Despretz (1822) and later by Guthrie (1860) and Niemann (1860), during which time its distinctive mustard and garlic-like odor and blistering action on the skin were noted. Meyer (1886) subsequently devised a process for the synthesis of pure SM that was later used by the Germans for its preparation as a chemical warfare agent. Its first use on the battlefield was during the German attack near the Belgian town of Ypres, in 1917. As a consequence, the French named it Yperite (Dacre and Goldman, 1996). SM was also referred to as “lost,” an acronym from the first letters of the chemists Lommel and Steinkopf, who investigated the military use of the compound (Kehe and Szinicz, 2005), “Yellow Cross,” because of the markings on German munitions containing SM, and HD (from Hunstoffe, distilled), the designated acronym used by the U.S. military (Dacre and Goldman, 1996). It is distinguished from other chemical warfare agents by its persistence in the environment, latency of action, and debilitating effects which incapacitate its victims. Chemically related vesicants include 2-chlorethylethyl sulfide (CEES) or half-mustard, a monofunctional alkylating agent with attenuated activity, and methylbis (2-chlorethyl) amine or mechlorethamine (HN2), a nitrogen analog of SM that is used in the treatment of leukemia (Mustargen-Merck, 2002) (Fig. 1).
The primary targets of SM are the skin, cornea, and respiratory tissues. Signs of injury follow an asymptomatic period of variable length, depending on the level of exposure and the target organ. The asymptomatic period is briefest for the eye and is followed by redness and irritation which can progress to corneal damage with photophobia, blepharospasm, and temporary blindness (Papirmeister et al., 1991). Airway exposure primarily targets the nasal, laryngeal, and tracheobronchial mucosa, with early mild irritation giving rise to epistaxis, laryngeal injury, and bronchial edema and damage (Papirmeister et al., 1991). Respiratory barrier function is compromised, and at higher concentrations, the lower airway epithelium becomes involved, with corresponding impairment of gas exchange (Papirmeister et al., 1991). Initial contact with the skin is typically not associated with discomfort. Erythema, the first sign of exposure, develops after a latency of 2–24 h and may be associated with intense itching. Vesicles filled with a pale yellow fluid appear several hours later, and with time, they may coalesce to form pendulous blisters. Blisters subsequently burst, resulting in formation of a necrotic layer or eschar on the affected skin surface. Typically, the wounds resolve over the course of 10–50 days, leaving pigmentation changes that may persist for months or years (Balali-Mood and Hefazi, 2005; Chiesman, 1944). Figure 2 depicts the time course of pathophysiological responses following cutaneous SM exposure. Multiple rounds of blistering and healing in the same individual have also been reported (Balali-Mood and Hefazi, 2005).
Inflammation is thought to play a significant role in SM-induced pathology with initial exposure characterized by an accumulation of both granulocytes and macrophages at sites of injury (Dannenberg et al., 1985; Guignabert et al., 2005; McClintock et al., 2002; Millard et al., 1997; Papirmeister et al., 1991; Smith et al., 1995). Exposure to high levels of SM can also result in systemic toxicity; notably bone marrow depression, consequent immune suppression, and increased susceptibility to infection (Balali-Mood and Hefazi, 2005). Chronic conditions arising from acute or long-term low-level exposure to SM have been documented involving the skin, eyes, lung, nervous system, immune system, and gastrointestinal tract (Balali-Mood and Hefazi, 2005; Ghanei et al., 2006; Khateri et al., 2003; Sidell and Hurst, 1997). Occupational exposures to SM that have occurred during its manufacture have also been associated with various forms of cancer (Easton et al., 1988; Inada et al., 1978; Pechura and Rall, 1993; Wada et al., 1968; A. Weiss and B. Weiss, 1975; Yanagida et al., 1988). The present review surveys the current knowledge of the biological effects of SM in the skin, with a focus toward understanding mechanisms involved in cutaneous blister formation.
Insight into the cutaneous histopathological effects of SM comes from early scientific experiments of human subjects (Ginzler and Davis, 1943; Henriques et al., 1943; Pappenheimer, 1926; Renshaw, 1946; Warthin and Weller, 1919) and more recent observations of victims of the Iran-Iraq war (Mehzad, 1988; Requena et al., 1988). The earliest changes in the epidermis are noted during the erythematous phase of injury, during which time a few scattered basal keratinocytes were identified as undergoing degenerative alterations including nuclear swelling, dispersion of chromatin, and cytoplasmic vacuole formation. With sufficient exposure, the lesion progresses and involves greater numbers of keratinocytes located predominately in the basal layer. The appearance of pyknotic nuclei is followed by disintegration of the cytoplasmic membrane. Foci of necrotic cells then appear which coalesce to form microblisters and eventually to macroblisters. Reports are generally consistent with a locus of blistering at the epidermal-dermal junction (Papirmeister et al., 1991). In some instances, necrotic lesions have been identified where bulla formation was bypassed, and the full thickness of the epidermis was affected (Henriques et al., 1943).
Damage to the dermal layer of the skin has also been described in humans exposed to SM (Ginzler and Davis, 1943; Henriques et al., 1943; Warthin and Weller, 1919). Most affected are cells of the adnexal structures, in particular, epithelial cells surrounding hair follicles. Capillary reactions are also observed, with hyperemia coinciding with the erythematous phase of SM-induced injury. At higher doses, SM causes more serious injury to the capillaries. Damage to individual dermal fibroblasts is noted at very early time points; however, the frequency of these effects varies in vesicating lesions. Inflammatory cells are also present in the skin of SM-exposed individuals. Mononuclear cells are detected very early in the erythematous phase of injury. This increases in intensity with time and is followed by an influx of neutrophils into the tissue prior to vesication.
Various animal models have been utilized to investigate mechanisms mediating SM-induced toxicity in the skin. Common species employed include rabbits, guinea pigs, pigs, and mice (Casillas et al., 1997; Isidore et al., 2007; Lindsay et al., 2004; Monteiro-Riviere and Inman, 1995; Smith et al., 1997a; Tewari-Singh et al., 2009). SM was applied to the skin as a vapor using a vapor cup or in liquid form, diluted in an appropriate vehicle, such as dichloromethane. In each of these models, four characteristic stages of SM-induced injury are observed: latency, erythema, vesiculation, and necrosis. Although gross vesiculation characteristic of human exposures is not evident in animal models, microblisters are detectable (Smith et al., 1997a). Several in vitro models, including biopsied human skin, various forms of engineered human skin equivalent (MatTek, LifeCell, Skin2, others) (Blaha et al., 2000; Greenberg et al., 2006; Hayden et al., 2005, 2009), and isolated human and mouse epidermal keratinocytes (Shakarjian et al., 2007; Smith et al., 1990), have also been established. These in vivo and in vitro models display varying degrees of utility in terms of recapitulating the pathophysiological stages of SM-induced injury and repair, understanding molecular mechanisms of toxicity, or screening for the discovery of useful countermeasures.
A number of groups have investigated the effects of SM using the pig skin model, which shares structural and functional similarities to human skin, including density of hair follicles, and the presence of sweat glands (Bartek et al., 1972; Monteiro-Riviere and Inman, 1997; Monteiro-Riviere and Riviere, 1996; Sabourin et al., 2002). Using the isolated perfused porcine skin flap model, Monteiro-Riviere and Inman (1995, 1997) demonstrated dose-dependent gross blister formation in response to SM (0.04–2 mg in 200 μl). Epidermal changes induced by SM are generally localized to the stratum basale, the innermost keratinocyte layer, comprised of self-renewing cells which maintain attachments via hemidesmosomes to the basement membrane. Significant mitochondrial destruction and nuclear pyknosis are also observed in response to SM with occasional cytoplasmic vacuoles and lipid inclusion bodies in affected basal keratinocytes. These changes precede signs of vesication. SM-induced cellular changes are focal in nature and coincide with sites of microblister formation. Transmission electron microscopy has revealed separation between the laminin-rich lamina lucida and the collagen IV–rich lamina densa of the basement membrane. Further definition of the locus of disruption has been provided by immunohistochemical analysis. Antibodies against the hemidesmosomal marker, bullous pemphigoid antigen (BP230), stain the roof of blisters, while type IV collagen is evident at the base of the blister. Whereas most laminin immunoreactivity is localized at the base of the blister, a smaller amount is associated with the stratum basale. These findings suggest that the plane of SM-induced epidermal-dermal separation is beneath the hemidesmosomes, within the upper portion of the lamina lucida (Monteiro-Riviere and Inman, 1995).
A mouse ear model was initially used by Brinkley et al. (1989) to assess SM-induced dermal alterations and then further characterized and developed by Casillas et al. (1997). Termed the mouse ear vesicant model (MEVM), liquid SM (0.04–0.64 mg in 5 μl) is applied to the medial surface of one ear, with the other ear serving as the vehicle control. After a characteristic latency period, a dose-dependent increase in the frequency and severity of histopathological markers including edema, epidermal necrosis, and the formation of microblisters on medial and lateral surfaces is observed. Immunostaining and electron microscopy has revealed that epidermal-dermal separation occurs at the level of the upper lamina lucida, with focal cleavage into the lower portion of the stratum basale as indicated by the detection of hemidesmosomal components on both dermal and epidermal sides of the cleavage site (Monteiro-Riviere et al., 1999). This locus of separation is attributed to the unique anatomical complexity of the ear relative to other epidermal sites.
More recently, a newer method for assessing SM-induced skin injury using nude mice grafted with human skin has been described (Greenberg et al., 2006). Previous variations of this model used biopsied partial thickness human skin as graft tissue (Papirmeister et al., 1984; Vogt et al., 1984). In these systems, SM was found to induce edema and microvesication, but only at relatively high doses (Papirmeister et al., 1991). The newer method utilizes engineered human skin formed by growing keratinocytes at an air-liquid interface on a fibroblast-populated collagen matrix. Though devoid of adnexal structures, these grafts appear to retain many of the morphological and functional features of normal human skin (Kolodka et al., 1998; Smith et al., 1997a). Using this model, two distinct stages of SM-induced epithelial damage have been identified by Greenberg et al. (2006). The first, a prevesication stage, observed 6-h postexposure, is largely independent of dose. It is characterized by the appearance of discrete clusters of pyknotic epithelial cells located exclusively in the stratum basale. The second stage, observed at 24 h, is dose dependent and involves significantly greater number of cells in the basal and suprabasal layers of the epidermis. Typical of this stage is the formation of microblisters through separation along the basement membrane zone. At moderate doses of SM, a polymorphonuclear cellular infiltrate is also evident in the dermis at 6 h, with focal infiltration into the epidermis at 24 h. Moreover, while the distribution of type VII collagen and laminin-332 immunoreactivity is unchanged during prevesication, a discontinuous pattern is noted after 24 h, a time coincident with microvesication.
In conclusion, application of SM to pig, mouse ear, and engineered human skin graft models reveals similar pathological sequelae and demonstrates a locus of vesication at the level of the lamina lucida of the basement membrane. These changes are consistent with those observed in humans after SM exposure; however, they do not precisely mimic clinical SM exposure, and thus, opportunities remain for improved methods development.
Because the prevesication stage of SM-induced injury is characterized by only mild-to-moderate inflammatory cell infiltration, and more pronounced inflammation is not observed until later in the pathologic process, it has been suggested that inflammation plays a minor role in the primary events mediating cutaneous injury and vesication (Papirmeister et al., 1991). In more recent studies, however, it has been argued that inflammation is in fact significant in the early vesication event and that inflammatory cells and mediators may actually contribute directly to the formation of the primary lesion (Cowan and Broomfield, 1993).
An influx of neutrophils into the forearm of skin of human subjects has been observed as early as 30 min after exposure to SM (Warthin and Weller, 1919). Similarly, in the rabbit model, increases in granulocytes and mononuclear cells have been reported within 2 h of SM administration, persisting for 24 h (Dannenberg et al., 1985). Leukocyte emigration into the papillary dermis and epidermis has also been reported in mice and in human skin explants after SM exposure (Lindsay and Rice, 1996; Wormser et al., 2005). Myeloperoxidase activity, an indicator of neutrophil influx, increases within 9 h, preceding macrophage migration into the skin of hairless mice after CEES exposure (Tewari-Singh et al., 2009). In the MEVM, pretreatment of mice with neutrophil-depleting monoclonal antibodies significantly reduces late-stage necrosis induced by SM, suggesting a role for these cells in skin toxicity (Levitt et al., 2004). In contrast, while there is evidence of dermal mast cell degranulation in SM-treated tissues and increases in dermal mast cell number after CEES treatment has been reported (Levitt et al., 2004; Rikimaru et al., 1991; Tewari-Singh et al., 2009), treatment of wild-type and mast cell–deficient (W/Wv) mice yielded similar changes in ear weight and extravasation, suggesting a lack of involvement of mast cells in the edematous stage of SM-induced skin pathology (Levitt et al., 2004).
Several in vivo studies have documented increased expression of proinflammatory cytokines in the skin following SM exposure. Using in situ hybridization techniques, increases in interleukin (IL)-1β, IL-8, monocyte chemoattractant protein (MCP)-1, and growth related gene mRNA were noted as early as 2 h after application of liquid SM to rabbit skin (Tsuruta et al., 1996). In mouse ear, IL-1β, IL-6, tumor necrosis factor (TNF)-α, and granulocyte monocyte-colony stimulating factor have been reported to be elevated within 6 h (Ricketts et al., 2000; Sabourin et al., 2000; Wormser et al., 2005). Increases in IL-1α protein and IL-1β, TNF-α, macrophage inflammatory protein, MIP-2, and MCP-1 mRNA have also been detected in the dorsal skin of hairless mice after exposure to SM vapor (Ricketts et al., 2000; Sabourin et al., 2003). Similarly, following vapor cup exposure of weanling pigs to SM, increases in relative mRNA levels of IL-1β, IL-6, IL-8, and TNF-α were noted (Sabourin et al., 2002). In cultured human keratinocytes, SM stimulates the release of IL-1β, IL-6, IL-8, and TNF-α at 100–300μM, doses relevant to in vivo exposure (Arroyo et al., 2000). Cultured skin fibroblasts have also been shown to express IL-6 in response to SM (Arroyo et al., 2001). These cytokines are thought to be key to inflammatory cell recruitment and activation at sites of injury, initiating a second phase of soluble mediator release. Cytokine expression is controlled by several signaling molecules, including the transcription factors nuclear factor-kappaB (NF-κB) (Ghosh et al., 1998) and activator protein-1 (AP-1) (Zenz et al., 2008). NF-κB has been reported to be activated after SM exposure (Atkins et al., 2000; Minsavage and Dillman, 2007; Rebholz et al., 2008) and both AP-1 and NF-κB after CEES exposure (Pal et al., 2009).
Arachidonic acid and its cyclooxygenase and lipooxygenase products are important inflammatory mediators that have also been observed in the skin after SM exposure (Blaha et al., 2000; Dachir et al., 2004; Lefkowitz and Smith, 2002; Rikimaru et al., 1991; Tanaka et al., 1997). Several of these mediators increase capillary permeability facilitating the influx of additional inflammatory substances including complement components, kinins, and fibrin into the dermal interstitium (Rikimaru et al., 1991). Cyclooxygenase-2 (COX-2), the rate-limiting enzyme in prostaglandin biosynthesis, has also been identified in the epidermis of SM-treated mice (Nyska et al., 2001). Findings that nonsteroidal anti-inflammatory agents (NSAIDs) reduce skin injury suggest that these mediators are important in SM toxicity (Casillas et al., 2000). That COX-2 is involved in toxicity is also supported by studies showing that the extent of ear swelling and histopathological signs of lesion severity are markedly reduced in COX-2 null mice treated with SM or in wild-type mice treated with celecoxib, a COX-2–specific inhibitor (Wormser et al., 2004). In contrast, loss of COX-1, the constitutive isoform of the enzyme, has no effect on cutaneous injury induced by SM. Taken together, these studies suggest an involvement of inflammatory mediators in SM cutaneous pathology. However, it remains to be determined which of these are important in the vesication process.
Much has been theorized regarding the potential for SM and its analogs to produce oxidative and electrophilic stress, processes often associated with inflammation, and the role that this plays in toxicity (Dacre and Goldman, 1996; Paromov et al., 2007). Under homeostatic conditions, a net reducing environment is maintained in tissues by the presence of glutathione (GSH), which serves as a buffer against cytotoxic electrophiles and reactive oxygen species (ROS). The propensity of SM to react with sulfhydryls is thought to lead to a concentration-dependent depletion of reducing equivalents within cells. Recent findings also suggest that SM and related vesicants can interact with key intracellular reductases to generate mustard-free radicals (Brimfield et al., 2009). In addition, inflammatory cells, which infiltrate into the skin in response to SM-induced injury, generate additional ROS that contribute to oxidative stress (Dröge, 2002). This raises the possibility that the toxicity of mustard alkylating agents involves oxidative stress. In this regard, lipid peroxides formed by the reaction of ROS with membrane lipids, have been reported to be elevated systemically after percutaneous intoxication of rats with SM (Vijayaraghavan et al., 1991). There is also evidence of increases in lipid peroxidation in A431 cells after HN2 or CEES (Pino et al., 2007), in cultured human keratinocytes after SM (Steinritz et al., 2009), and on the dorsal skin of hairless mice after CEES exposure (Pal et al., 2009). Support for a role of superoxide anion in SM-induced injury comes from findings that administration of superoxide dismutase reduces cutaneous toxicity in a guinea pig model (Eldad et al., 1998). Human skin cell lines pretreated with the GSH-depleting agent, buthionine sulfoximine, display enhanced toxicity to SM (Simpson and Lindsay, 2005). Conversely, the cytotoxic actions of SM are reduced in primary keratinocytes pretreated with sulforaphane, a cytoprotective agent which increases intracellular GSH levels by activating the transcription nuclear factor (erythroid-derived 2)-like 2 (Nrf2) (Gross et al., 2006). Direct evidence for the GSH-depleting action of chloroethyl alkylating agents has been described in lymphocytes treated with CEES (Han et al., 2004). It has been suggested that GSH depletion by SM can lead to the production of quinone-generated free radicals in melanocytes (Smith, 1999). Findings that pharmacological inhibition of these quinone radicals protects G361 melanocytes against SM-induced toxicity provides support for this concept (Smith and Lindsay, 2001). Based on these findings, it is tempting to speculate that unregulated oxidative/electrophilic stress contributes significantly to the cutaneous vesicating action SM; however, this remains to be determined.
Evidence is also accumulating that reactive nitrogen species (RNS) including nitric oxide (NO), may also contribute to SM-induced toxicity. NO is generated from L-arginine via the enzyme, nitric oxide synthase (NOS). Three isoforms of the enzyme have been identified, including two constitutive isoforms, endothelial NOS (eNOS) and neuronal NOS (nNOS), and an inducible NOS isoform, (iNOS). NO is a potent oxidizing agent. It can also react rapidly with superoxide anion generating a more long-lived RNS, peroxynitrite (Virág et al., 2002). iNOS has been reported to be upregulated by SM in vivo, in the guinea pig skin back model (Nyska et al., 2001). Using an in vitro scratch wound model, Ishida et al. (2008) found that iNOS induction is accompanied by wound closure in human keratinocytes. Moreover, knockdown of iNOS by a small interfering RNA inhibits wound closure. A noncytotoxic concentration of SM (20μM) acted similarly to the iNOS knockdown, inhibiting iNOS induction in the scratched monolayer while also blocking reepithelialization. These data suggest that SM may delay wound healing by blocking iNOS induction. Using 100 and 300μM concentrations, Steinritz et al. (2009) demonstrated that SM induces expression of iNOS, as well as eNOS within 6 h in HaCaT cell monolayers. These changes coincided with nitrotyrosine modifications of cellular proteins, which is a biochemical marker for peroxynitrite generation. Further studies are necessary to explore the precise role of RNS in the cutaneous actions of SM.
Aqueous solutions of SM are highly reactive, displaying a half-life of only 24 min in physiological solutions at room temperature (Bartlett and Swain, 1949). Rates of hydrolysis increase with increasing temperature and decreasing chloride ion concentration, resulting in its conversion to thiodiglycol and HCl. The initial reaction involves the formation of a cyclic ethylene sulfonium ion intermediate followed by electrophilic attack on the target molecule (Fig. 3). One mole equivalent of H+ and Cl− are liberated in each reaction. Reactive groups on target molecules include sulfhydryls, phosphates, ring nitrogens, and carboxyl groups. Thus, DNA, RNA, proteins, carbohydrates, and lipids are targets for alkylation by SM (Debouzy et al., 2002; Mol et al., 2008; Noort et al., 2002; Papirmeister et al., 1991). Because it is a bifunctional alkylating agent, SM can form not only monofunctional adducts but also intra- and intermolecular cross-links.
The most extensively investigated molecular target of SM alkylation is DNA. Its significance was initially established in bacteria where inhibition of DNA replication was noted at doses relevant to in vivo mammalian toxicity (Papirmeister et al., 1991). Subsequent use of 35S-labeled SM led to the identification of alkylation sites on DNA. Approximately 65% and 17% of the DNA alkylation products are monofunctional adducts on the N7 of guanine and the N3 of adenine, respectively, and approximately 17%, N7 guanine bifunctional cross-links. In mammalian cells, it has been estimated that 25% of the DNA cross-links occur between complementary DNA strands, while the remainder are intrastrand (Walker, 1971). The fact that monofunctional analogs such as CEES exhibit reduced potency relative to SM suggests that DNA cross-linking is important in toxicity. However, the observation that CEES is a biologically active vesicant suggests that bifunctional cross-links are not required for skin injury.
Evidence suggests that the extent of SM-induced cytotoxicity due to DNA alkylation and cross-linking is influenced not only by the capacity of cells to repair DNA but also by the specific repair mechanism activated. Studies using Escherichia coli and mouse embryo fibroblasts have revealed that the nucleotide excision repair (NER) and base excision repair (BER) pathways play contrasting roles in SM toxicity (Matijasevic and Volkert, 2007). Thus, while E. coli lacking NER capability exhibit greater sensitivity to the cytotoxic actions of SM than wild-type E. coli, their susceptibility to CEES is minimally altered, underscoring the importance of NER in repairing cross-linked DNA. Both E. coli and mouse embryo fibroblasts deficient in BER are also less susceptible to SM. In contrast, an intact BER pathway protects cells from the monofunctional methylating agent, methyl methanesulfonate, and pharmacological inhibition of BER increases the sensitivity of skin cells to CEES (Jowsey et al., 2009; Matijasevic and Volkert, 2007). Interestingly, hypothermia protects the skin and keratinocytes, as well as mouse embryo fibroblasts, from the damaging effects of SM, a response markedly reduced in cells deficient in BER (Matijasevic and Volkert, 2007; Mi et al., 2003; Sawyer and Risk, 1999; Sawyer et al., 2002). These findings suggest that BER may play a distinct role in the response to mono- and bifunctional alkylating agents.
DNA alkylation, and the formation of apurinic sites during repair processes, can result in single- and double-strand DNA breaks. This leads to activation of poly(ADP-ribose) polymerase (PARP), a family of nuclear cell signaling enzymes involved in poly-ADP ribosylation of DNA-binding proteins (Shall and de Murcia, 2000). While low levels of PARP activation signal repair, excessive activity can deplete cells of NAD+ and adenine triphosphate (ATP) resulting in cytotoxicity. Whether this results in apoptosis or necrosis depends on the cell type and other factors (Nicotera and Melino, 2004; Rosenthal et al., 2001). Using a hairless guinea pig model, Kan et al. (2003) observed the appearance of apoptotic basal keratinocytes within 6 h of SM exposure, which was followed by necrosis 24 h later, a time when there is significant cell lysis and neutrophilic infiltration. This suggests that in the skin, SM produces a temporal continuum of apoptosis followed by necrosis. The role of PARP in SM-induced cytotoxicity has also been addressed using a transgenic mouse model (Rosenthal et al., 2001). Fibroblasts isolated from mice lacking PARP-1, the most abundant PARP isoform, are much more likely to undergo apoptosis, when compared to wild-type cells, which preferentially undergo necrotic cell death. Unlike results in fibroblasts, immortalized keratinocytes derived from wild-type or PARP-1−/− mice only exhibit apoptosis following SM treatment. These data suggest that while PARP may determine the mode of SM-induced death in some cell types, apoptosis appears to predominate in mouse keratinocytes. Using the HaCaT human keratinocyte cell line, Kehe et al. (2008) demonstrated that SM readily stimulates PARP-1 activity and produces a dose-dependent continuum of cell death from apoptosis to necrosis. Furthermore, treatment of the cells with 3-aminobenzamide, an inhibitor of PARP, causes a discernible inhibition of necrosis. These results contrast to the observations of Rosenthal et al. (2001) described above and suggest that PARP may contribute, at least in part, to necrosis in keratinocytes. While reasons for this discrepancy are unclear, cell type, culture conditions, and rates of proliferation may all be factors (Paromov et al., 2007). Differences may also be due to the use of immortalized cells which exhibit altered responses to DNA-damaging agents when compared to primary cells (Petit-Frère et al., 2000).
Mechanisms underlying SM-induced apoptosis have been explored using primary cultures of human keratinocytes. An increase in the pro-apoptotic protein p53, and a decrease in the anti-apoptotic protein Bcl-2 have been observed in keratinocytes after SM exposure (Rosenthal et al., 1998, 2000). These findings are consistent with previous reports in an in vivo weanling pig skin model (Smith et al., 1997b). Treatment of human keratinocytes with 100–300μM SM also leads to activation of caspase-8, which initiates the Fas-dependent death receptor pathway, as well as caspase-9, which initiates the mitochondrial apoptotic pathway (Rosenthal et al., 2003). These two pathways converge to activate caspase-3, the central executioner protease (Zimmermann et al., 2001). Transfection of immortalized keratinocytes with a dominant-negative Fas-associated death domain results in a blunted caspase response following SM treatment (Rosenthal et al., 2003). Microvesication and tissue injury produced by SM treatment of transfected cells after grafting on to athymic nude mice are also reduced. A recent report has also demonstrated protective effects of inhibitors of caspase-8 or caspase-9 on SM–induced apoptosis in cultured human skin (Mol et al., 2009).
Changes in intracellular calcium levels are known to activate the mitochondrial pathway of apoptosis. A key regulator of calcium-dependent proteins is calmodulin. SM causes a time-dependent induction of calmodulin in human keratinocytes (Simbulan-Rosenthal et al., 2006). Moreover, depletion of keratinocyte calmodulin using antisense probes attenuates SM-induced activation of caspases and nuclear fragmentation. Bad, a pro-apoptotic Bcl-2 family member present in an inactive phosphorylated form in viable cells, is also activated by SM. Furthermore, cyclosporine A, a selective inhibitor of calcineurin, a Bad phosphatase, inhibits SM-induced apoptosis. These results suggest that calcium-dependent activation of Bad may be a mechanism by which SM induces apoptosis in keratinocytes.
One form of cellular demise common to epithelial cells is detachment-initiated apoptosis, also referred to as anoikis (Chiarugi and Giannoni, 2008; Frisch and Francis, 1994). Keratinocytes rely on signals derived from the surrounding extracellular matrix (ECM) for survival. For instance, matrix proteins, such as laminin-332 interact directly with integrins α3β1 and α6β4, essential components of hemidesmosomes found on the basolateral keratinocyte surface (Schneider et al., 2006). Integrin-associated molecules including paxillin, caveolin, Shc, integrin signaling kinase (ILK), focal adhesion kinase (FAK), and various growth factor receptors are thought to transduce anchorage-dependent survival signals (Chiarugi and Giannoni, 2008; Frisch and Screaton, 2001). In the absence of these signals, keratinocytes undergo anoikis via Fas-dependent or mitochondrial pathways of apoptosis (Chiarugi and Giannoni, 2008). Evidence suggests that loss of survival signals contributes to SM-induced epidermal injury and that cell detachment from the basal lamina precedes cytotoxicity. SM can alter the dynamics of cytosolic proteins that control the attachment of cells to the basement membrane. For example, SM has been reported to modify intracellular actin microfilaments and keratin intermediate filaments which are known to be important in maintaining epithelial cell connections with the basal lamina (Hinshaw et al., 1999). In fact, modifications of actin microfilament architecture and cell morphology have been observed within 3 h of exposure of human keratinocytes to SM. These changes are associated with a significant decrease in keratinocyte adherence without evidence of cytotoxicity (Hinshaw et al., 1999). In addition, SM has been reported to cause rapid decreases in expression of keratins 5 and 14, intermediate filaments present in undifferentiated keratinocytes (Werrlein and Madren-Whalley, 2000). Experiments with animals (Gunhan et al., 2004), keratinocyte cultures (Dillman et al., 2003; Mol et al., 2008), and purified proteins (Hess and FitzGerald, 2007) have all demonstrated that keratins 5 and 14 become alkylated by SM and its analogs, HN2 and CEES. Sites of alkylation may be similar to the dominantly acting mutations in keratins 5 and 14 that are responsible for the human blistering disorder, epidermolysis bullosa simplex, in which basal epidermal cells are also targeted (Fuchs, 1997). The keratin cytoskeleton of basal keratinocytes is linked to the hemidesmosome and, through plectin, makes connections with the β4 cytoplasmic tail of integrin α6β4, thereby strengthening adhesion to the basement membrane via laminin-332 (Giancotti and Tarone, 2003). Alkylation of keratins 5 and 14 by SM may cause aggregation of the intermediate filament network resulting in basal cell separation from the basement membrane.
In addition to targeting epidermal cells, SM can directly alkylate ECM proteins in the skin, a process that can interfere with the ability of basal keratinocytes to maintain vital connections with the basement membrane. This is supported by findings that SM reduces the ability of naive human keratinocytes to deposit laminin at the dermal-epidermal interface (Gentilhomme et al., 1998). Moreover, treatment of fibronectin and laminin with SM interferes with adherence of keratinocytes to these matrix proteins, a response blocked by SM scavengers (Zhang et al., 1995a). Similarly, experiments in our laboratories have demonstrated reduced adherence of keratinocytes to HN2- or CEES-treated laminin-332 (Shakarjian et al., 2008). SM and HN2 also reduces tissue immunreactivity of laminin-332, as well as integrin α6β4 and BP230, two hemidesmosomal components that are critical for keratinocyte adherence (Kan et al., 2003; Smith et al., 1997b, 1998; Werrlein and Madren-Whalley, 2000; Zhang and Monteiro-Riviere, 1997). Interestingly, each of these ECM proteins, like keratins, has been implicated in human blistering disorders involving separation of the epidermis at the dermal-epidermal junction (Pulkkinen et al., 1998; Yancey, 2005). These findings suggest that SM can alter the interaction of basal cells with matrix proteins that are critical for basement membrane adherence.
Although the precise sequence of events leading to blister formation after dermal exposure to SM have not been established, an involvement of proteases is considered important in the vesication process (Papirmeister et al., 1985). Proteases have been implicated in several subepidermal blistering diseases in humans including epidermolysis bullosa aquisita (Shimanovich et al., 2004), dermatitis herpetiformis (Airola et al., 1995; Oikarinen et al., 1983), pemphigus vulgaris (Koch et al., 1997), and bullous pemphigoid (Liu et al., 1998). A number of animal models have been utilized to investigate the contribution of proteases to SM toxicity. SM has been reported to increase protease and antiprotease activities within 24 h in rabbit skin explants (Higuchi et al., 1988). Matrix metalloproteinase (MMP) activity has also been detected in culture fluids from SM-treated rabbit skin (Woessner et al., 1990). Plasminogen activator, as well as histamine, increases in an organ culture of full-thickness human skin after treatment with SM or HN2 (Rikimaru et al., 1991). Histamine is a major product of mast cells and a marker of mast cell degranulation. Mast cell degranulation also releases proteases including chymase and tryptase (Prussin and Metcalfe, 2006). Elastase, tryptase, calpain, and gelatinase (MMP-2 and MMP-9) activity is increased within 24 h after SM treatment of mouse ears (Powers et al., 2000). Increases in gelatinases are of particular interest because of their ability to cleave basement membrane components and disrupt the dermal-epidermal junction (Malemud, 2006). Using a weanling pig model, Sabourin et al. (2002) found increases in relative mRNA of MMP-9, but not MMP-2, 24 h post-SM treatment. Similarly, our laboratories found that latent gelatinase activity is rapidly increased after SM exposure and remains elevated for at least 7 days in the mouse ear model (Shakarjian et al., 2006). This increase in activity appears to be due to a rise in MMP-9, but not MMP-2, protein levels. Likely sources of gelatinases include not only infiltrating neutrophils but also epidermal keratinocytes and dermal fibroblasts. A recent report by Ries et al. (2008) suggests that epidermal-dermal cross talk may contribute to the increase in cutaneous MMP-9 in response to SM.
A critical role of MMP-9 has also been described in a murine model of bullous pemphigoid initiated by administration of antibodies to the hemidesmosomal protein bullous pemphigoid antigen 180 (BP180, also known as collagen XVII) (Liu et al., 1998, 2000, 2005). Mice with a targeted deletion in the gene for MMP-9 fail to develop blistering in this model (Liu et al., 1998). By destroying α1 protease inhibitor (α1PI) and allowing elastase to degrade BP180, MMP-9 plays an essential role in disease pathology in this model (Liu et al., 2000). The importance of neutrophil elastase and MMP-9 in bullous pemphigoid, as well as epidermolysis bullosa acquisita, has been confirmed experimentally using human tissue (Shimanovich et al., 2004). Thus, early elevations of elastase and MMP-9 after cutaneous exposure to SM, combined with their ability to cause dermal-epidermal separation, suggest that these proteases are potential effectors of SM-induced vesication.
Laminin-332, like BP180, is a matrix protein essential for junctional integrity; laminin-332 mutations are associated with severe forms of junctional epidermolysis bullosa (Igoucheva et al., 2007). As described above, laminin-332 immunoreactivity is reduced in SM-induced skin lesions, and evidence indicates that this matrix protein is a target of degradation. Lindsay and Rice (1995) discovered partially degraded laminin in extracts of SM-treated minipig skin, suggesting an involvement of matrix-degrading proteases in blister formation. Subsequently, Chakrabarti et al. (1998) isolated a serine protease from SM-treated human epidermal keratinocytes with laminin-cleaving activity. Further investigation revealed that SM and HN2, in contrast to nonvesicating alkylating agents, induce multiple laminin-332 cleaving activities in these cells. The use of class-specific protease inhibitors has demonstrated that these consisted of serine proteases and MMP's (Jin et al., 2004, 2008). Although neutrophil elastase and MMP-2 degrade the gamma2 chain of laminin-332, a lack of species matching has made these results inconclusive (Giannelli et al., 1997; Mydel et al., 2008). Current knowledge indicates that mammalian tolloid metalloproteinase (Veitch et al., 2003), BMP-1 (Amano et al., 2000), and MMP-19 (Sadowski et al., 2005) cleave the gamma2 chain, while MMP-2 and MMP-14 cleave the α3 chain, of laminin-332 (Veitch et al., 2003). Keratinocytes are known sources of these enzymes. The potential role of membrane-type metalloproteinases, such as MMP-14, in vesication induced by SM has been investigated in cultured human skin explants (Mol et al., 2009). Findings that selective inhibitors of TNF-α–converting enzyme (TACE) and furin, proteins essential for the processing of membrane-type MMPs and A Disintegrin and Metalloproteinases into their catalytic forms, protect cultured human skin against injury, suggest an involvement of these proteases in SM toxicity. It is hypothesized that proteases such as MMP-14 and TACE/ADAM17 participate in the vesication response to SM through their actions on BP180 and laminin-332. Examination of the involvement of these and other proteolytic enzymes should provide further insight into the role of active matrix degradation in the pathogenesis of SM-induced dermal-epidermal separation.
Several recent studies have applied differential gene expression profiling techniques to ascertain a more genome-wide view of the skin's response to SM. Using the MEVM, Rogers et al. (2004) identified 19 genes specifically upregulated in mouse skin 24 h after exposure to SM. Major gene categories activated include apoptosis, transcription factors, cell cycle, oncogenes, and inflammation. An effect of SM dose is observed in terms of the number and the category of genes upregulated. Mouse ears exposed to SM alone or in combination with the antivesicants dimercaprol, octyl homovanillamide, or indomethacin have also been analyzed (Dillman et al., 2006). Genes categories altered by SM include cell cycle and growth, inflammatory and immune response, cytoskeletal and cell adhesion, and signal transduction. A correlation has been reported between the changes in several of these genes and the ability of antivesicants to reduce SM-induced ear edema. Significant changes in gene expression have also been reported up to 7 days following exposure of mice to SM (Gerecke et al., 2009). Gene ontology analysis suggests that the most significantly altered biological processes are immune and inflammatory responses and that the specific genes expressed differ as a function of time after exposure. Transcriptional responses of porcine skin to SM and thermal burns have also been compared (Price et al., 2009; Rogers et al., 2008). Several overlapping biological functions have been identified between the two types of skin injury, suggesting the potential for applying pharmacological agents known to be effective against burns to SM-injured skin. Clearly, there is potential for biomarker identification and high throughput analytical tools to be developed for further microarray studies. The significance of individual genes and pathways will become more apparent as PCR and protein expression validation methods are applied. However, at the present time, microarray studies have yet to provide a precise understanding of the complex mechanisms involved in the vesication process.
To elucidate molecular changes that occur after SM exposure, Mol et al. (2008) treated human keratinocytes with 14C-SM and followed adducted proteins by two-dimensional electrophoresis. Nineteen modified proteins, the majority, variants of keratins 5, 6, 14, 16, and 17, have been identified by matrix-assisted laser desorption/ionization-time of flight mass spectrometry. Two high–molecular weight cross-linked aggregates of keratins were also identified. Three additional proteins identified as SM alkylation targets were actin, stratifin, an adaptor protein that can interact with BP180 (Li et al., 2007), and galectin-7, a protein involved in cell adhesion and proliferation (Klíma et al., 2008). Differential protein expression analysis using two-dimensional electrophoresis demonstrated significant changes in patterns between 8- and 42-h posttreatment. A number of protein species unique to SM-treated keratinocytes were detected, the majority being fragments of keratins 14, 16, and 17. Findings that the appearance of these fragments is mitigated by preincubating keratinocytes with caspase inhibitors suggests an active apoptotic process. Two additional proteins were identified as a phosphorylated variant of HSP27 and ribosomal protein P0. HSP27 may be involved in activation of the p38 MAP kinase pathway and has been implicated in pemphigus vulgaris (Berkowitz et al., 2008). Ribosomal protein P0, essential for proper protein synthesis, was found in its inactive, unphosphorylated form, suggesting that SM impairs ribosomal function. Thus, the proteomic approach has shown value by illuminating known and novel targets of SM that may be important in the cell's response to this agent.
SM remains a significant threat to the military and civilian populations. Despite extensive research aimed at understanding the molecular mechanisms of SM action, the pathways leading from exposure to vesication remains unclear, and effective medical countermeasures have yet to be developed. Human cutaneous exposures have defined four stages of injury from SM: latency, erythema, vesiculation, and necrosis. Although SM is a relatively nonselective alkylating agent, keratinocytes of the stratum basale appear to be the most sensitive to its cytotoxic actions, and blistering involves the detachment of these cells from the supporting basal lamina of the basement membrane of the epidermal-dermal junction. The reason for the sensitivity of this epidermal layer remains unresolved. It has been established that basal keratinocytes differ from neighboring epidermal cells in their self-renewing nature and greater capacity for proliferation. A different spectrum of transcription factors also predominate in these cells which direct expression of a unique protein repertoire including cytoskeletal proteins, such as keratins 5 and 14 (as compared to the keratins 1 and 10 found in differentiating keratinocytes), and the protein constituents of the hemidesmosome, a structure unique to basal cells, and includes BP180, BP230, and the α6 and ß4 integrin chains (Koster, 2009; Xu et al., 2004). It is also possible that basal keratinocytes respond to DNA damage differently than their more differentiated counterparts in terms of capacity for repair. Residing on the basement membrane, basal keratinocytes may also be more susceptible to ECM-degrading enzymes. Further examination of the unique characteristics of basal keratinocytes will likely provide clues for the selectivity of SM.
A scheme summarizing current understanding of the potential mechanisms contributing to the dermal toxicity of SM is depicted in Figure 4. Because of its ability to alkylate a wide variety of small and large molecules, SM is likely to have multiple targets, including DNA, protein, and small molecules that contribute to its vesicating action. DNA damage can lead directly to cytotoxicity, or it may trigger other changes resulting in cellular dysfunction, cell death or tissue repair. For example, whereas overactivation of PARP following exposure to high concentrations of SM can trigger apoptotic or necrotic cell death, depending on the extent of ATP depletion, milder activation can trigger DNA repair mechanisms that may prove to be either beneficial or detrimental to cell survival. Although less well studied, intracellular and extracellular protein alkylation by SM might also contribute to vesication. Current data suggest that alkylation of cytoskeletal, cell anchoring–related, and ECM proteins may weaken keratinocyte-basement membrane interactions, ultimately leading to epidermal-dermal separation and anoikis. However, one cannot exclude the possibility that less abundant protein targets are also significant to the pathogenic process. Finally, intracellular GSH likely exemplifies a small molecule target and the depletion of this important reducing agent following exposure to SM can initiate oxidative stress leading to lipid peroxidation and other oxidative cellular damage.
A variety of proinflammatory proteins are induced following dermal exposure to SM including enzymes that produce RNS and ROS, chemokines, cytokines, and proteases. These most likely act in concert to promote tissue injury and amplify the inflammatory response. At present, the precise role of these different inflammatory mediators and cells in the toxicity of SM is unknown. Evidence suggests that disruption of cell-matrix interactions is key to the process, and it remains to be determined if inflammatory cells participate in this pathogenic step.
Studies on the mechanism of action of SM have suggested several potential therapeutic approaches. For example, the observation that SM induces inflammation and the release of cytotoxic/proinflammatory mediators suggests that anti-inflammatory drugs may be efficacious in mitigating the cytotoxic actions of SM. In this regard, both administrations of NSAIDs including indomethacin as well as COX-2 inhibitors have proven to be efficacious against SM-induced injury in a variety of in vivo models (Babin et al., 2000; Casillas et al., 2000; Yourick et al., 1995; Zhang et al., 1995b). Neurogenic inflammation is also thought to be an important component of tissue damage induced by SM. Nonmyelinated sensory C-fibers function as dual sensory afferents, transmitting sensory information to the central nervous system and also releasing nociceptive and inflammatory neuropeptides, such as substance P (Szallasi and Blumberg, 1999). Many of these actions are mediated by transient receptor potential V1 (TRPV1) in the skin. Capsaicin, a constituent of hot pepper, has been reported to exert its actions via TRPV1, and several capsaicin analogs significantly reduce SM-induced inflammation in the mouse ear model (Casillas et al., 2000; Sabourin et al., 2003). Targeting specific proinflammatory cytokines and chemokines may also prove clinically efficacious in treating SM poisoning. This is supported by findings that antibodies to TNF-α inhibit SM damage in the MEVM (Wormser et al., 2005).
A variety of enzyme inhibitors have also been tested as SM therapeutics. For example, recent studies have shown that PARP inhibitors abrogate SM toxicity in the MEVM (Smith, 2009). Similarly, inhibition of caspases that mediate apoptosis have been shown to suppress the toxic and vesicating actions of SM (Mol et al., 2009; Rosenthal et al., 2003). Moreover, blocking the activity of matrix-degrading enzymes reduces injury from SM exposure in human skin organ cultures (Mol et al., 2009).
Another interesting approach to treating SM toxicity is the use of iodine. Wormser and colleagues recently isolated a peptide factor from iodine-treated burns that may partially mediate iodine's actions (Brodsky et al., 2008). The peptide, an 11 amino acid fragment of histone H2A, reduces the severity of SM-induced ear swelling in the MEVM, and transfection of a plasmid encoding this factor into HaCaT cells significantly increases their resistance to the cytotoxic effects of SM.
Vesicants such as SM remain a significant health threat. Thus, it is imperative to identify targets for therapeutic intervention in the skin, as well as the cornea and lung. This can best be accomplished by further investigation into the mechanism of cytotoxic actions of SM. Of particular importance is elucidating the roles of oxidative and nitrosative stress in toxicity and the specific contribution of inflammatory mediators. Leads derived from microarray analyses and proteomics are likely to be useful in identifying new targets for countermeasure development.
National Institutes of Health (AR055073, CA100994, CA132624, CA093798, ES004738, GM034310, EY009056, AI051214, AI084137, and ES005022); National Institutes of Health Counter ACT Program through the National Institute of Arthritis and Musculoskeletal and Skin Diseases (Award No. U54AR055073). This work's contents are solely the responsibility of the authors and do not necessarily represent the official views of the federal government.