Evidence for Widespread Epithelial Damage and Coincident Production of Monocyte Chemotactic Protein 1 in a Murine Model of Intestinal Ricin Intoxication

Abstract

The development of small-animal models is necessary to understand host responses and immunity to emerging infectious diseases and potential bioterrorism agents. In this report we have characterized a murine model of intestinal ricin intoxication. Ricin administered intragastrically (i.g.) to BALB/c mice at doses ranging from 1 to 10 mg/kg of body weight induced dose-dependent morphological changes in the proximal small intestine (i.e., duodenum), including widespread villus atrophy and epithelial damage. Coincident with epithelial damage was a localized increase in monocyte chemotactic protein 1, a chemokine known to be associated with inflammation of the intestinal mucosa. Immunity to intestinal ricin intoxication was achieved by immunizing mice i.g. with ricin toxoid and correlated with elevated levels of antitoxin mucosal immunoglobulin A (IgA) and serum IgG antibodies. We expect that this model will serve as a valuable tool in identifying the inflammatory pathways and protective immune responses that are elicited in the intestinal mucosa following ricin exposure and will prove useful in the evaluation of antitoxin vaccines and therapeutics.

The development and testing of effective vaccines and therapeutics against potential bioterrorism agents pose major challenges to the biomedical scientific community (2). Foremost is the fact that human exposure to these so-called select agents is rare and often is poorly documented in the clinic. Consequently, an understanding of the molecular basis of both pathophysiology of and protective immunity to this diverse collection of viruses, microbial pathogens, and toxins must rely on the use of well-established animal models. This is especially true in the case of ricin toxin, a potent ribosome-inactivating protein from the castor bean (Ricinus communis) that has already proven to be an effective murder weapon and bioterrorism agent (25).

While ricin (∼64 kDa) is one of the simplest members of the A-B family of toxins, it is also one of the most promiscuous (32, 37). The toxin's single A subunit (RTA) is an N-glycosidase that selectively depurinates a conserved adenine residue within 28S rRNA and in this respect is indistinguishable from shiga toxin (11). The toxin's B subunit (RTB) is a bivalent lectin that mediates toxin attachment to terminal galactose residues on glycoproteins and glycolipids (4). The ubiquitous nature of ricin's receptors, combined with its universally conserved enzymatic substrate, enables ricin to intoxicate virtually all known cell types. Not surprisingly, ricin can be lethal to humans following injection, inhalation, or ingestion (3, 6, 26).

Although a recent expert panel workshop sponsored by the National Institutes of Health deemed the adulteration of food and water supplies to be the most likely mechanism by which ricin would be disseminated as a bioterrorism agent, very little is known about the effects of ricin on the gastrointestinal mucosa (1). Ingestion of whole castor beans results in severe abdominal pain, vomiting, diarrhea, and (depending on the number of beans and degree of mastication) death (3, 6, 26). Experimentally, Sekine and colleagues (38) demonstrated that rats exposed intragastrically (i.g.) to ricin develop pronounced lesions in the proximal small intestine as early 2 h postchallenge, with the median lethal dose estimated to be 10 mg/kg of body weight. In addition to the direct cytotoxic effects, it is postulated that ricin can also initiate an inflammatory response that exacerbates pathogenesis. Shiga toxins, including ricin, trigger a so-called “ribotoxic stress response” in human intestinal epithelial cell lines and in macrophage cell lines that culminates in the secretion of proinflammatory cytokines and chemokines (12, 23, 40). However, the proinflammatory cytokines/chemokines induced following ricin intoxication of the intestinal mucosa in vivo have not been identified.

In an effort to better understand the pathophysiology associated with ricin exposure, we have developed and characterized a mouse model of intestinal ricin intoxication. Our results indicate that ricin elicits dose- and time-dependent morphological changes in the proximal small intestine that coincide with the local production of proinflammatory chemokines. Protection against ricin intoxication was achieved by immunizing mice i.g. with ricin toxoid and correlated with elevated levels of antitoxin mucosal immunoglobulin A (IgA) and serum IgG antibodies. This mouse model provides the research community with a valuable tool with which to begin to dissect the inflammatory pathways and protective immune responses that are elicited in the intestinal mucosa following ricin exposure.

MATERIALS AND METHODS

Chemicals and reagents.

Ricin (Ricinus communis agglutinin II) was purchased from Vector Laboratories (Burlingame, CA). Phenylmethylsulfonyl fluoride and bovine serum albumin were purchased from Sigma Company (St. Louis, MO). Tween 20 was obtained from Bio-Rad (Torrance, CA), and protease inhibitor cocktails were purchased from Calbiochem-EMD Biosciences (La Jolla, CA). Paraformaldehyde (16%) was purchased from Electron Microscopy Sciences (Fort Washington, PA), and Bouin's fixative was obtained from Krackeler Scientific (Albany, NY).

i.g. ricin challenge and tissue collection.

All animals used in this study were housed under conventional, specific-pathogen-free conditions and were treated in strict compliance with guidelines established by the Institutional Animal Care and Use Committee at the Wadsworth Center. Female BALB/c mice ages 6 to 8 weeks were purchased from Taconic Laboratories (Germantown, NY). Animals weighing 18 to 22 g were fasted for 1 h prior to being administered azide-free ricin (final volume, 0.4 ml) i.g. by means of a 22-gauge, 1.5-in. blunt-end feeding needle (Popper Scientific, New Hyde Park, NY). Food was provided ad libitum 1 h after challenge. At designated time points, animals were sacrificed by CO₂ asphyxiation, followed by cervical dislocation. The entire small intestine was surgically removed, beginning at the ileocecal junction, and then laid out on a moistened paper towel. Alternating segments (0.25-cm) of the duodenum were immersed in Bouin's fixative and subsequently embedded in paraffin by the Wadsworth Center Animal Histopathology core facility or immersed in ice-cold cell lysis buffer (Cell Signaling, Beverly, MA) supplemented with protease inhibitors and homogenized on ice using a Tekmar Tissuemizer (Fisher Scientific) tissue homogenizer. The following protease inhibitors (Calbiochem) were used: 150 nM aprotinin, 1 μM leupeptin, 50 μM 4-(2-aminoethyl) benzenesulfonyl fluoride hydrochloride, 1 μg/ml bestatin, and 0.5 mM phenylmethylsulphonyl fluoride. Homogenates were centrifuged (10,000 × g, 10 min) at 4°C, and the resulting supernatants were passed through a QiaShredder instrument (QIAGEN, Valencia, CA) to remove any residual debris. The amounts of protein in intestinal homogenates, as determined by the bicinchoninic acid assay (Pierce Chemical), were equalized in all samples by the addition of cell lysis buffer. Homogenates were frozen at −20°C as 0.2-ml aliquots and thawed immediately prior to use in the BD cytometric bead array (CBA) assay.

Histological and proinflammatory cytokine analysis of mouse duodenum.

Paraffin sections of mouse duodena were stained with hematoxylin and eosin (H&E) and visualized by light microscopy using a Zeiss Axioskop II microscope equipped with a charge-coupled-device camera. Tissue sections were scored for ricin intoxication using a 12-point histological grading system (21, 34) based on the severity and extent of alterations in villus shape (width and height), lamina propria edema, interepithelial swelling, and the presence of cellular infiltrate in the intestinal lumen. Tissue section samples were coded and blinded prior to being provided to investigators for scoring.

Proinflammatory cytokine (gamma interferon, interleukin 1 [IL-1], IL-6, IL-12p70, monocyte chemotactic protein 1 [MCP-1], and tumor necrosis factor alpha) levels in intestinal homogenates were determined using the BD CBA mouse inflammation kit (BD Biosciences, San Jose, CA). The cytokine concentrations in intestinal homogenates were calculated from standard curves generated using purified cytokines provided by the manufacturer. Statistical analysis of differences between CBA values from treatment groups of mice was conducted using analysis of variance (Excel 2003, Microsoft Corporation, Redmond, WA).

Production of RT.

Ricin toxoid (RT) was produced as described by Yan and colleagues (44). Ricin (1 mg per ml in phosphate-buffered saline [PBS]) was dialyzed in a Slide-a-lyzer dialysis unit (molecular weight cutoff, 10,000; Pierce Chemical) against 4% paraformaldehyde for 18 h at 47°C, followed by 30 h at 42°C. Dialysis was then continued against 4 liters of 0.1 M glycine for 4 days in the cold room to quench residual paraformaldehyde in the RT preparations. RT preparations (1 mg/ml) were stored at −80°C and were thawed immediately prior to use.

i.g. immunization of mice with RT and measurement of serum and fecal antibodies by ELISA.

Groups of BALB/c mice, ages 6 to 8 weeks, were immunized three times i.g. with RT (200 μg per animal per immunization) at 2-week intervals). Serum and fecal pellets were collected as described previously (28) 2 days before the first immunization and 7 days after each immunization. The antiricin IgA and IgG antibody titers in serum and the antiricin IgA antibody titers in fecal pellets were determined by enzyme-linked immunosorbent assay (ELISA), as done previously (29). Briefly, NUNC Maxisorb F96 microtiter plates (Krackeler Scientific, Albany, NY) were coated overnight at 4°C with 0.1 μg of ricin (or RTA or RTB) per well in a volume of 0.1 ml in PBS (pH 7.4). Microtiter plates were washed with PBS-Tween 20 (0.05% [vol/vol]), blocked with goat serum (2% [wt/vol] in PBS-Tween 20), and overlaid with serum or fecal extracts diluted in block solution. Secondary goat anti-mouse IgG- and IgA-specific antibodies labeled with horseradish peroxidase were obtained from Southern Biotech (Birmingham, AL). ELISA plates were developed with a one-component TMB colorimetric substrate (Kirkegaard and Perry, Gaithersburg, MD) and were read using a SpectraMax 250 microtiter plate reader equipped with Softmax software (Molecular Devices, Union City, CA). Mouse monoclonal IgGs and IgAs specific for RTA or RTB were used as controls for these assays (29, 30). ELISAs were done a minimum of two times in duplicate for each sample analyzed. Averages and standard errors (SE) between duplicate samples were calculated using Softmax and Excel 2003.

RESULTS AND DISCUSSION

Histopathology in the murine duodenum associated with ricin intoxication.

To better define the effects of ricin on the intestinal mucosa, we examined by light microscopy H&E-stained sections of duodena from groups of mice that had been challenged i.g. with ricin (5 to 10 mg/kg). As early as 3 h after toxin challenge, we observed mild interepithelial swelling and villus deformation (data not shown). By 18 h after challenge, histopathologic changes in the mucosa were widespread and included blunting of intestinal villi, swelling of interepithelial spaces, mucus secretion, and separation of the epithelium from the lamina propria (Fig. 1a to e). We also observed the sloughing of enterocytes from the tips of villi and occasional cellular infiltrate consisting of polymorphonuclear leukocytes in the intestinal lumen (Fig. 1f). The observed histopathology persisted for the duration of the experiment (24 h), with no evidence of epithelial restitution. Intestinal ulcerations were not observed at the time points examined.

FIG. 1.

Histological changes in the murine duodenum associated with ricin intoxication. BALB/c mice were administered PBS (control) (a and c) or ricin (b and d to f) at the indicated doses by gavage and were sacrificed 18 h later. Intestinal segments (1 to 2 cm) were immersed in Bouin's fixative and embedded in paraffin. Tissue sections (5 μm) were stained with H&E and viewed by bright-field microscopy. (a) Low-power (×10) image of a cross section of the duodenum of a control, PBS-fed mouse. The villi appear as long, slender, finger-like projections extending into the intestinal lumen. (b) Low-power image of a cross-section of the duodenum from a mouse treated with ricin (10 mg/kg). The villi appear blunted and swollen, especially at the tips. The swelling (arrowheads) is likely due to edema. (c) High-power (×40) image of two villi from the duodenum of a control, PBS-treated mouse. (d) High-power image of a villus from the duodenum of a mouse treated with ricin (5 mg/kg). There is extensive interepithelial swelling (arrowheads) along the basolateral aspects of enterocytes. Excessive mucus, which appears as “webbing” in the lumen surrounding the affected villus, is also evident. (e) High-power image of two villi from a mouse exposed to ricin (10 mg/kg). At the tips of the villi, the epithelium has separated from the lamina propria (arrowheads), and the basal aspects of many individual enterocytes have degenerated or have been completed destroyed. (f) Cellular infiltrate, consisting of polymorphonuclear leukocytes (arrows), was occasionally evident in the intestinal lumen of ricin-treated animals.

To enable us to quantitatively measure ricin intoxication of the intestinal mucosa, we adapted a tissue scoring system based on the severity and extent of alterations in villus shape (width and height), lamina propria edema, interepithelial swelling, and the presence of cellular infiltrate in the intestinal lumen (see Materials and Methods). Using this scoring system, we observed a dose-dependent increase in intestinal lesions (Fig. 2). It should be noted that current guidelines established by the Wadsworth Center Institutional Animal Care and Use Committee prevented us from performing time-to-death experiments in this study. However, based on previous studies done with rats, we estimate that the oral median lethal dose for mice is approximately 10 mg/kg, with death likely occurring 36 to 48 h postchallenge (10, 38).

FIG. 2.

Histological changes in the murine duodenum associated with ricin intoxication. Paraffin sections of duodena collected from mice 24 h after the animals were challenged i.g. with ricin at the indicated doses (0 to 10 mg/kg) were scored by light microscopy. The tissues were scored based on villus morphology, interepithelial swelling, and cellular infiltrate in the lumen. At least 40 sections from 2 mice were examined at each dose. Average scores with SE are shown.

It is remarkable that damage to the intestinal mucosa was evident only at doses of ricin equal to or greater than 2.5 mg/kg. While these results are consistent with those obtained with rats by Sekine and colleagues (38), the apparent resilience of the intestinal epithelium in relation to ricin remains an enigma. The most likely explanation is that the amount of ricin that is effectively available to adhere to the enterocytes is limited, because the toxin is easily entrapped in intestinal mucus and/or adsorbed onto particles in the lumen (17). Other factors in intestinal secretions may also interfere with toxin absorption. For example, we have recently shown that the oligosaccharide side chains on secretory IgA (SIgA) can serve as receptors for ricin and competitively inhibit toxin attachment to the luminal surfaces of human intestinal epithelium (27). Nonetheless, even small amounts of toxin can have detrimental consequences to intestinal epithelial cells. Application of ricin to the apical surfaces of polarized epithelial cell monolayers grown in vitro results in rapid toxin internalization, transcytosis, and arrest in protein synthesis (29, 42).

Ricin intoxication correlates with elevated MCP-1 levels in intestinal mucosa.

To determine whether ricin intoxication is accompanied by a local increase in proinflammatory cytokines, we challenged mice i.g. with ricin at a range of doses (1 to 10 mg/kg) for various lengths of time (0, 5, 12, and 24 h) and then measured cytokine levels in intestinal homogenates by CBA (see Materials and Methods). There was no detectable increase in local gamma interferon, IL-1, IL-6, IL-12p70, or tumor necrosis factor alpha levels following ricin challenge at any of the doses or time points examined (data not shown). In contrast, we observed a dose- and time-dependent increase in MCP-1 in intestinal homogenates from mice treated with ricin (Fig. 3A and B). Interestingly, a comparison of Fig. 2 and 3 reveals that the severity of intestinal damage following ricin exposure correlates with MCP-1 levels.

FIG. 3.

Dose- and time-dependent increases in MCP-1 levels in the duodena of mice challenged with ricin. The duodena of BALB/c mice challenged i.g. with ricin were homogenized and assayed for MCP-1 by CBA. (A) Dose-dependent MCP-1 production. Tissues were collected 24 h after groups of mice (n = 5/group) had been challenged with the indicated doses of ricin. Average values with SE are shown. (B) Time-dependent MCP-1 production. Groups of mice (n = 6) were challenged with ricin (5 mg/kg), and tissues were then collected at the indicated time points. Average values with SE are shown. The amount of total protein in each intestinal homogenate sample was determined using the bicinchoninic acid assay (Pierce Chemical). Statistical analysis of differences between groups of mice was determined using an independent t test.

MCP-1, also known as CCL2, has been previously implicated in mediating intestinal inflammation (35) and could possibly be involved in ricin-mediated tissue damage. MCP-1 is a chemoattractant for lymphocytes, monocytes, and macrophages and can stimulate macrophages to undergo respiratory burst activity (5, 7, 9, 24, 36). Moreover, MCP-1 is released by intestinal epithelial cell lines following bacterial infection (16, 18) and exposure to microbial toxins, including Bacteroides fragilis enterotoxin (21) and Clostridium difficile A toxin (22). MCP-1 expression is regulated in part by the mitogen-activated protein kinase p38 (43). Ricin activates p38 mitogen-activated protein kinase in a variety of cell types, including human monocytes/macrophages and intestinal epithelial cells (8, 12, 23). Preliminary studies from our laboratory suggest that toxin-mediated tissue damage is attenuated in MCP-1 knockout mice compared to control animals (N. Mantis and J. Yoder, unpublished data). While these studies would implicate MCP-1 as a possible mediator of tissue damage following ricin exposure, we expect that other cytokines/chemokines are elicited upon toxin exposure. For example, Thorpe and colleagues have shown in vitro that epithelial cell lines exposed to shiga toxin or ricin secrete IL-8 and growth-regulated oncogene alpha (40, 41).

i.g. immunization of mice with RT stimulates a systemic and mucosal antibody response that protects animals against toxin challenge.

While vaccination of mice with RT has been shown to confer immunity to both systemic and aerosol ricin challenge (20, 44), it has not been examined whether vaccination with RT confers gastrointestinal immunity to ricin. To examine this possibility, groups of mice were immunized i.g. with formaldehyde-inactivated RT three times at biweekly intervals. Serum and fecal pellets were collected 7 days after each immunization and analyzed for antiricin antibodies by ELISA, as done previously (29). i.g. immunization of mice with RT stimulated antiricin IgG and IgA antibodies in serum and IgA antibodies in intestinal secretions (Fig. 4 A and B).

FIG. 4.

Antiricin IgG and IgA titers in serum and fecal samples from mice immunized i.g. with RT. BALB/c mice (n = 6 per group) were immunized i.g. with PBS (control) or RT three times (days 0, 14, and 36), and serum and fecal samples were collected 7 days later. Antiricin-specific antibody titers were determined using a ricin-specific ELISA and were expressed as n-fold increases over baseline titers measure in PBS-immunized littermates, as described in Materials and Methods. (A) Antiricin-specific IgG levels in serum. (B) Antiricin-specific IgA levels in serum and fecal extracts.

To test whether RT-immunized mice were immune to ricin, we challenged groups of animals i.g. with a sublethal dose of toxin and examined the duodena 24 h later for lesions and MCP-1 levels. Groups of immunized and nonimmunized animals were challenged with ricin (5 mg/kg) and scored for evidence of local ricin intoxication 24 h later. Mice sham immunized with PBS and then challenged with ricin developed lesions in the duodenum similar to those described above (e.g., widespread villus atrophy, swelling of interepithelial spaces, and separation of the epithelium from the basal lamina) (Fig. 5A). This group of animals also had elevated levels of MCP-1 in intestinal homogenates (Fig. 5B). In contrast, mice immunized with RT and challenged with ricin had no measurable lesions in the duodenum, and levels of MCP-1 in intestinal homogenates were at background levels (Fig. 5A and B). These data demonstrate for the first time that i.g. immunization of mice with RT confers protection against i.g. ricin intoxication and suggest that immunity to ricin is mediated by serum IgG and/or mucosal IgA antibodies.

FIG. 5.

Mice immunized i.g. with RT were protected against subsequent ricin challenge. Groups of mice were immunized i.g. five times at approximately 2-week intervals with PBS or RT and then challenged i.g. with ricin (5 mg/kg). Twenty-four hours after challenge, the animals were sacrificed, and the duodena were analyzed for histological changes and MCP-1 levels, as described in the legends to Fig. 2 and 3, respectively. (A) Mice immunized with PBS and challenged with ricin (PBS/ricin) had significantly (P = 0.01) more lesions than control animals (PBS/PBS) or animals immunized with RT and then challenged with ricin (RT/ricin). (B) MCP-1 levels were significantly (P = 0.02) elevated in PBS/ricin animals, compared to RT/ricin and PBS/PBS animals. Statistical significance was determined using an independent t test.

There is accumulating evidence that SIgA may be necessary to confer complete protection against ricin intoxication in mucosal compartments (15, 19, 20, 33, 44). This is best exemplified by the fact that serum IgG antibodies elicited by parental vaccination with RT provide only partial protection against an aerosol challenge (i.e., mice demonstrated widespread inflammation and necrotic lesions within the upper and lower respiratory tracts and remained debilitated for 10 to 14 days following challenge (15, 33). In contrast, full protection was achieved when mice were immunized mucosally to stimulate SIgA antibodies in bronchoalveolar lavage fluids (13, 14, 20, 44). We have recently produced a collection of monoclonal IgA and IgG antibodies (29, 30) that when coupled with the mouse model described herein will enable us to sort out the relative contributions of the different antibody specificities and isotypes that mediate immunity to ricin in mucosal compartments. Identifying the roles of secretory IgA and IgG in mediating mucosal immunity to ricin has important implications for the delivery of a ricin vaccine to humans (31, 39).

In conclusion, we have developed a mouse model of ricin intoxication, as well as quantitative measures of local tissue damage and inflammation elicited upon toxin exposure. We have also demonstrated for the first time that immunity to ricin in the intestinal tract can be elicited by vaccination, although the specific roles of IgA and IgG in this protection remain to be elucidated. This model will enable our laboratory, as well as other laboratories, to begin to test the hypothesis that inflammatory cytokines and/or chemokines contribute directly to the tissue damage associated with ricin exposure. As mentioned above, we can also use available monoclonal IgA and IgG antibodies to determine the mechanisms of mucosal immunity and test whether these are the same as or different from those in the systemic compartment.