Dextran Sulfate Sodium Colitis Facilitates Colonization with Shiga Toxin-Producing Escherichia coli: a Novel Murine Model for the Study of Shiga Toxicosis

Shiga toxin-producing Escherichia coli (STEC) bacteria are globally important gastrointestinal pathogens causing hemorrhagic gastroenteritis with variable progression to potentially fatal Shiga toxicosis. Little is known about the potential effects of E. coli-derived Shiga-like toxins (STXs) on host gastrointestinal immune responses during infection, in part due to the lack of a reproducible immunocompetent-animal model of STEC infection without depleting the commensal microbiota.

ABSTRACT

Shiga toxin-producing Escherichia coli (STEC) bacteria are globally important gastrointestinal pathogens causing hemorrhagic gastroenteritis with variable progression to potentially fatal Shiga toxicosis. Little is known about the potential effects of E. coli-derived Shiga-like toxins (STXs) on host gastrointestinal immune responses during infection, in part due to the lack of a reproducible immunocompetent-animal model of STEC infection without depleting the commensal microbiota. Here, we describe a novel and reproducible murine model utilizing dextran sulfate sodium (DSS) colitis to induce susceptibility to colonization with clinical-isolate STEC strains. After exposure to DSS and subsequent oral STEC challenge, all the mice were colonized, and 66% of STEC-infected mice required early euthanasia. Morbidity during STEC infection, but not infection with an isogenic STEC mutant with toxin deleted, was associated with increased renal transcripts of the injury markers KIM1 and NGAL, histological evidence of renal tubular injury, and increased renal interleukin 6 gene (IL-6) and CXCL1 inflammatory transcripts. Interestingly, the intestinal burden of STEC during infection was increased compared to its isogenic Shiga toxin deletion strain. Increased bacterial burdens during Shiga toxin production coincided with decreased induction of colonic IL-23 axis transcripts known to be critical for clearance of similar gastrointestinal pathogens in mice, suggesting a previously undescribed role for STEC Shiga toxins in suppressing host immune responses during STEC infection and survival. The DSS+STEC model establishes infection with clinical-isolate strains of STEC in immunocompetent mice without depleting the gastrointestinal microbiota, enabling characterization of the effects of STXs on the IL-23 axis and other gastrointestinal pathogen-host interactions.

INTRODUCTION

Shiga toxin-producing Escherichia coli (STEC) bacteria are globally important gastrointestinal (GI) bacterial pathogens responsible for sporadic epidemic outbreaks of hemorrhagic diarrhea with progression to potentially fatal systemic Shiga toxicosis in 5 to 15% of infected patients, especially young children (1). Shiga-like toxin 1 (STX1) and Shiga-like toxin 2 (STX2) isoforms are highly ribotoxic, particularly targeting renal glomerular endothelium in human kidneys and renal tubular epithelium in rodents due to species-specific localization of the toxin globotriaosylceramide receptor CD77 (2, 3). The toxins are genetically mobile virulence factors encoded in lysogenized lambda-like bacteriophages, resulting in a growing array of pathogenic E. coli strains acquiring the ability to secrete Shiga toxins (STXs). Newly emerging strains with previously uncharacterized combinations of virulence factors are of particular concern (4, 5). While the molecular outcomes of Shiga toxin ribotoxicity have been characterized in sensitive cell lines in vitro, the impacts of Shiga toxins on host gastrointestinal immune responses during STEC infection remain unclear due to the lack of appropriate in vivo models (6, 7).

STEC strains contain genomic pathogenicity islands, like the locus of enterocyte effacement or the locus of adhesion and aggregation, that encode proteins enabling close associations with gastrointestinal epithelial cells, as well as proteins that suppress or modulate local acute inflammatory responses (8, 9). The gastrointestinal tract contains diverse, spatially segregated immune cells that coordinate localized responses to potential pathogens via recruitment of phagocytes and modulation of epithelial barrier defense functions, with distinct variations in cellular populations and phenotypes between anatomical regions of the gut (10). The sensitivities of epithelial, monocytic, and lymphocytic subpopulations to Shiga toxicosis remain uncharacterized in vivo, and cell culture results vary depending on the origin of the cell line and culture conditions (11, 12). Accordingly, any potential benefit to the bacteria conferred by Shiga toxin production during STEC colonization and the impacts of the toxins on host defense mechanisms during STEC infection are unclear. Citrobacter rodentium, a nontoxigenic murine pathogen that forms attaching-and-effacing lesions similar to those of many pathogenic STEC strains, is known to induce robust responses along the interleukin 23 (IL-23) axis within the gastrointestinal tract. IL-23 from mucosal dendric cells initiates IL-23 receptor (IL-23R) signaling on receptor-expressing cells, which is necessary for effective survival and clearance of C. rodentium infection (13). IL-23R-stimulated upregulation of IL-17, IL-22, and other cytokines from regional lymphocytes is critical for phagocyte recruitment and epithelial barrier repair (14, 15). The similarities between colonization characteristics of C. rodentium and those of clinically relevant STEC strains suggest that IL-23 axis responses could also be critical for host clearance of STEC (16). Infection of germ-free mice by STEC strains with genetic ablation of STX production induces modest numbers of CD4⁺ Th17 lymphocytes, the key effectors of IL-23R-stimulated adaptive immunity, but the impact of Shiga toxins produced by STEC on the IL-23 axis response is unknown (17).

A significant barrier to study of host-pathogen interactions during STEC infection is the lack of a reproducible murine model of infection by clinical-isolate STEC strains with progression from gastrointestinal inflammation to systemic Shiga toxicosis. Systemic circulation of Shiga toxins induces renal tubular injury in mice, but standard laboratory strains of naive mice are resistant to gastrointestinal colonization by STEC (18, 19). Previous approaches to induce susceptibility to STEC colonization include severe protein restriction, high-dose antibiotic treatment, and germ-free conditions (17, 20–22). Colonization may follow, but these models have proven difficult to reproduce, the commensal microbiome is grossly ablated, and germ-free conditions are not generalizable due to altered host responses to pathogens in the absence of host-microbiota interactions (2). Colonization of naive mice with a strain of C. rodentium transduced to express STX2d was a major advance in the field and is a murine model of infection with an attaching-and-effacing pathogen that produces STX2d (23). Infection with C. rodentium (STX2d⁺) results in colitis and toxin-induced renal tubular injury but is limited by lack of E. coli-specific non-STX virulence factors and regulatory elements. STEC bacteria produce virulence factors other than Shiga toxins, and lysogenized lambda phage-encoded proteins other than STXs regulate strain-specific virulence gene expression by STEC in the anaerobic intestinal environment (24, 25). Emerging hybrid STEC strains from environmental, animal, and clinical samples have acquired previously uncharacterized mobile virulence factors (26–28), underscoring the need for a reproducible mouse STEC infection model capable of utilizing clinically isolated strains.

To overcome this critical barrier, we leveraged an observation in the literature describing significant outgrowths of commensal E. coli in C57BL/6 mice exposed to dextran sulfate sodium (DSS) (29). DSS administration in drinking water is a well-characterized colitis model in rodents in which intestinal epithelial injury and colitis severity can be manipulated reproducibly by DSS dosage (30). The proportion of Escherichia sp. in the fecal microbiota of C57BL/6 mice, determined by 16S rRNA sequencing, increased from <1% prior to short-term DSS exposure to approximately 20% after short-term DSS exposure (29). This observation suggested that short-term, mild DSS pretreatment may alter the intestinal environment sufficiently to permit STEC colonization in mice. Here, we report a novel murine DSS+STEC model that is the first model of STEC infection with clinical-isolate strains in immunocompetent mice without depletion of the microbiota. The model develops moderate colitis and STX2-induced renal tubular injury in the absence of bacteremia, similar to conditions seen in STEC-infected patients (1). STX2 production resulted in increased STEC burdens and decreased colonic IL-23 axis transcripts in the DSS+STEC model, demonstrating its application to examining previously uncharacterized host-pathogen interactions.

RESULTS

After a series of pilot experiments evaluating DSS dosage and timing, the optimal conditions supporting STEC colonization in 6-week-old C57BL/6 mice consisted of exposure to 2.5% (wt/vol) DSS in drinking water for 5 days, followed by challenge via oral gavage with 1 × 10⁹ to 5 × 10⁹ CFU of STEC bacteria on days 5 and 7 after starting DSS. STX2 isoforms are known to exert greater toxicity in mice than STX1 isoforms; thus, the clinical-isolate STEC strains TW14359 (STX2a and STX2c) and 86-24 (STX2a) were chosen to evaluate colonization and subsequent disease in the DSS+STEC model (31). E. coli TUV86-2 [STEC(ΔSTX2)] is an STEC STX2a deletion strain that was previously constructed from strain 86-24, providing an isogenic ΔSTX2 control for use in infection experiments (32). Control mice were gavaged with sterile phosphate-buffered saline (PBS) during DSS colitis and otherwise treated identically to the STEC-challenged mice.

STEC challenge following the 5-day DSS exposure resulted in high and sustained colonization by E. coli strains 86-24 (STEC) and TUV86-2 [STEC(ΔSTX2)] (Fig. 1A). STEC colonization was determined by quantitative PCR (qPCR) of fecal DNA for STEC-specific eae, the gene encoding intimin. One copy of eae is present per STEC genome, allowing the eae copy number per gram to approximate number of CFU per gram of STEC. Following the first exposure, both the STX2a-producing strain 86-24 and the STEC(ΔSTX2) strain TUV86-2 established colonization, with 10⁸ STEC eae copies/g present in fecal samples 1 day after gavage. Repeat oral gavage with the bacterial strains 2 days following the first challenge successfully increased STEC burdens, with both STX2a-producing strain 86-24 and ΔSTX2 strain TUV86-2 achieving >10⁸ eae copies/g 1 day following the second exposure (Fig. 1A). STEC burdens were several orders of magnitude greater than in a previously published study of BALB/c mice fasted for 18 h prior to infection with strains 86-24 and TUV86-2 (33). Initial colonization was not impacted by STX2 production, but burdens of STX2-producing STEC were significantly higher than those of STEC(ΔSTX2) 4 and 5 days following infection (Fig. 1B to D). The increased burden of STEC during DSS colitis suggests that DSS exposure and STX2 production altered the host colonic microenvironment to favor STEC survival.

FIG 1.

STX2 production increases STEC burdens during DSS colitis-facilitated colonization of C57BL/6 mice. C57BL/6 mice were orally gavaged with 3.4 × 10⁹ to 5.5 × 10⁹ CFU of STX2a-producing STEC strain 86-24 (STEC) or 3.3 × 10⁹ to 4.6 × 10⁹ CFU of isogenic STX2 deletion strain TUV86-2 [STEC(ΔSTX2)] 0 and 2 days after 5-day exposure to 2.5% (wt/vol) DSS. Controls were gavaged with sterile PBS. (A) STEC eae copy numbers per gram of feces determined by qPCR of fecal DNA for eae. OG, oral gavage; BDL, below detectable limit. (B to D) Box plots of eae copies per gram of feces on the indicated days postinfection. Limit of detection, 9.22 × 10³ eae copies/g feces. STEC, n = 12; STEC(ΔSTX2), n = 12; PBS, n = 9. Means ± standard errors (SE) are shown. Statistical significance was determined by two-tailed Student t test (**, P < 0.01).

Oral gavage time points and the dose and duration of DSS treatment used in the DSS+STEC model were determined via dose-response pilot experiments exploring DSS concentrations (1.0 to 3.5% [wt/vol]) and duration (DSS for 4 to 5 days with or without low-dose DSS maintenance dosing) to identify conditions that induced mild or moderate, but not severe, colitis. Gram-negative STEC can grow on MacConkey's medium under aerobic conditions, and MacConkey's medium with sorbitol has been used for partial selection of STEC from clinical samples (34). Prior to DSS exposure, bacteria capable of growing on MacConkey's medium were isolated at low abundance from C57BL/6 mouse feces, with 1.8 × 10⁵ ± 0.7 × 10⁵ CFU/g recovered after overnight aerobic incubation at 37°C. Following 5 days of treatment with filter-sterilized 2.5% (wt/vol) DSS ad libitum in drinking water, colony numbers on MacConkey's medium increased by an order of magnitude to 2.1 × 10⁶ ± 0.8 × 10⁶ CFU/g (see Fig. S1 in the supplemental material). These results were consistent with the prior study (29) and suggested that DSS-induced dysbiosis opened a niche for Gram-negative bacterial expansion in the GI tracts of C57BL/6 mice.

Five days of 2.5% DSS resulted in reproducible but transient weight loss and colitis (not shown), with approximately 15% peak weight loss followed by gradual recovery of body weight (see Fig. S2A in the supplemental material). Single oral gavage challenge of mice with STEC strain TW14359 (STX2a and STX2c) 5 or 7 days after the first exposure to DSS established robust colonization (see Fig. S2B and C in the supplemental material) but was less reproducible with respect to kidney injury markers (data not shown). These pilot experiments demonstrated a fairly wide time window of susceptibility to colonization and also infection with a different clinical STEC isolate. No gender differences were observed with respect to DSS-induced colitis or susceptibility to STEC colonization under any of the conditions used.

DSS-treated mice infected using two oral gavage bacterial challenges were evaluated for morbidity following infection with STEC strain 86-24 or STEC(ΔSTX2) strain TUV86-2 compared to PBS-gavaged control mice. Based on weight loss and clinical appearance, 66% of STEC 86-24-infected mice were euthanized between 3 and 5 days after infection, with PBS controls and STEC(ΔSTX2) TUV86-2-infected mice experiencing overall minimal morbidity and consequently higher survival rates (Fig. 2A). The rate of weight loss between days 3 and 4 postgavage was greater in STEC 86-24-infected mice than in PBS controls and STEC(ΔSTX2) TUV86-2-infected mice, corresponding to the time of euthanasia for the most severely affected mice (Fig. 2B).

FIG 2.

STX2-producing-STEC infection causes increased mortality and weight loss. C57BL/6 mice were orally gavaged with STX2a-producing STEC strain 86-24 (STEC) or isogenic STX2 deletion strain TUV86-2 [STEC(ΔSTX2)] 0 and 2 days after 5-day exposure to 2.5% (wt/vol) DSS. Controls were gavaged with sterile PBS. The mice were euthanized based on clinical appearance and weight loss. (A) Survival curves. Statistical significance was determined by Mantel-Cox test. OG, oral gavage. (B) Rate of daily weight loss between 3 and 4 days after the first gavage. STEC, n = 12; STEC(ΔSTX2), n = 12; PBS, n = 9. Means ± SE are shown. Statistical significance was determined by one-way analysis of variance (ANOVA) with Tukey posttest (*, P < 0.05).

Bacteremia is not observed in STEC-infected patients (1), and similarly, bacterial invasion of the caudal mesenteric lymph node and liver was extremely low or undetectable in STEC strain TW14359-infected mice or PBS controls following the 5-day DSS plus two gavage STEC exposure protocol (see Fig. S3A and B in the supplemental material). This occurred despite a peak burden of >10⁸ copies of eae/g of feces in STEC strain TW14359-infected mice (data not shown). Similarly, cardiac-blood cultures were negative postinfection (see Fig. S3C in the supplemental material). To confirm this observation, mice exposed to sufficient DSS to induce severe colitis followed by STEC challenge developed positive liver and mesenteric lymph node cultures (1.2 × 10⁴ to 1.6 ×10⁴ CFU/g tissue on day 2 postinfection) and bacteremia (50% positive cardiac-blood cultures on days 4 and 7 postinfection), indicating loss of intestinal epithelial barrier function (see Fig. S4 in the supplemental material). Collectively, the data are consistent with the conclusion that sepsis from bacterial invasion did not contribute to morbidity in the DSS+STEC model.

Since the primary disease outcome of systemic Shiga toxicosis in mice is renal tubular injury, STEC-infected mice were stratified by euthanasia time point to determine if early euthanasia was associated with renal tubular injury during STX2-producing STEC infection. Molecular markers of renal injury are regularly used in toxin and ischemia models as relatively sensitive indicators of early renal cellular damage prior to the onset of histologically evident structural changes or clinical blood biomarkers of renal injury (35). Tissues were obtained from mice either at euthanasia based on morbidity criteria (days 3 to 5 postinfection) or on day 7 for surviving mice. Compared to PBS control mice, renal transcripts for injury markers KIM1 and NGAL were elevated in kidneys from euthanized mice infected with STEC 86-24 and unchanged in STEC(ΔSTX2) TUV86-2-infected mice (Fig. 3A and B). Histological evaluation of hematoxylin and eosin (H&E)-stained formalin-fixed kidneys from STEC 86-24-infected mice showed evidence of tubular injury characterized by tubular dilation, tubular epithelial flattening, loss of tubular brush borders, and tubular epithelial vacuolization (Fig. 3C). Blinded scoring of tubular injury revealed that euthanized STEC-infected mice developed histologically apparent tubular damage, whereas none of the control or STEC(ΔSTX2) TUV86-2-infected mice had visible evidence of tubular injury (Fig. 3D). Periodic acid-Schiff (PAS)-stained kidney sections were evaluated for the presence of brush borders in the convoluted tubules of the outer renal cortex (see Fig. S5 in the supplemental material). Data were recorded as the percentage of tubules lacking an intact brush border for each sample to account for the presence of healthy brush border-negative distal convoluted tubules interspersed among brush border-positive proximal convoluted tubules throughout the renal cortex. Euthanized STEC 86-24-infected mice had a significantly higher percentage of brush border-negative cortical tubules than STEC(ΔSTX2) TUV86-2-infected mice and PBS controls (Fig. 3E), indicating proximal tubular epithelial damage and loss of brush border expression.

FIG 3.

STX2a-induced morbidity is associated with renal tubular injury. Mice were orally gavaged with STX2a-producing STEC strain 86-24, STEC(ΔSTX2) strain TUV86-2, or sterile PBS 0 and 2 days following 5-day exposure to 2.5% (wt/vol) DSS. Kidney and blood samples were collected at the time of sacrifice, with surviving mice sacrificed 7 days postinfection. The STEC-infected mice were stratified by euthanasia time point to model clinical STEC infection outcomes of progression to systemic disease versus resolution of colitis. (A and B) Fold change of total renal transcripts of molecular markers of renal injury KIM1 and NGAL normalized to PBS transcript content determined by RT-qPCR. (C) Representative images (×150 magnification; insets, ×600 magnification) of H&E-stained renal cortex from DSS+PBS, DSS+STEC(ΔSTX2), and DSS+STEC mice showing renal tubular dilation and epithelial flattening in STEC-infected mice (arrows). The images are from mice euthanized 3 or 4 days after infection. (D) Tubular injury scoring. The statistical significance of the histological scoring was determined by Kruskal-Wallis test. (E) Percentages of renal tubules lacking intact PAS⁺ brush borders. (F) Serum BUN at time of sacrifice. PBS controls, n = 9; STEC Live, n = 4; STEC Euth (euthanized), n = 8; STEC(ΔSTX2), n = 12. Means ± SE are shown in the bar graphs. Statistical significance was determined by one-way ANOVA with Tukey posttest (*, P < 0.05; **, P < 0.01; ***, P < 0.001).

Serum blood urea nitrogen (BUN) of euthanized STEC 86-24-infected mice was increased compared to PBS-gavaged controls (Fig. 3F), whereas serum creatinine was not elevated in mice from any group (data not shown), indicating renal injury in the presence of STX2 that had not progressed to acute renal failure. There was no histological evidence of glomerular injury or thrombotic microangiopathy in any of the mice, consistent with prior reports in other STEC and endotoxin-free STX murine models (23, 36). The incidence of renal injury in the DSS+STEC model appeared to correlate with increased exposure to STX2-producing STEC. Exposures to 0.7 × 10⁸ to 3.5 × 10⁸ CFU of STEC TW14359 via a single oral gavage challenge resulted in lower rates of morbidity and renal injury during DSS colitis, with only 3 out of 17 infected mice requiring early sacrifice (see Fig. S6 in the supplemental material).

Renal injury in purified STX rodent and nonhuman primate models is associated with increases in renal inflammatory cytokine transcripts (37–39). Renal transcripts of CXCL1 (keratinocyte chemoattractant [KC]) chemokine and IL-6 cytokine genes were significantly elevated in euthanized STEC 86-24-infected mice, but no elevations in either transcript were apparent in kidneys from STEC(ΔSTX2) TUV86-2-infected mice or PBS controls (Fig. 4A and B). In contrast, renal CCL3 transcripts encoding the proinflammatory C-C chemokine macrophage inflammatory protein 1α were unchanged during STEC strain 86-24 colonization (Fig. 4C). These results suggest that the toxin or other soluble bacterial virulence factors do not necessarily induce broad-based inflammation, but rather, may target specific immune defense mechanisms during renal toxicity.

FIG 4.

STX2a production induces renal acute inflammatory IL-6 and CXCL1 transcripts. Mice were orally gavaged with STX2a-producing STEC strain 86-24, STEC(ΔSTX2) strain TUV86-2, or sterile PBS 0 and 2 days following 5-day exposure to 2.5% (wt/vol) DSS. Kidney samples were collected at the time of sacrifice, with surviving mice sacrificed 7 days postinfection. Shown are fold changes of total renal transcripts of CXCL1 (A), IL-6 (B), and CCL3 (C). All fold changes were normalized to the PBS control renal transcript content determined by qPCR. PBS controls, n = 9; STEC Live, n = 4; STEC Euth, n = 8; STEC(ΔSTX2), n = 12. Means ± SE are shown. Statistical significance was determined by one-way ANOVA with Tukey posttest (***, P < 0.001).

An advantage of the DSS+STEC model was the observed increase in STEC bacterial burdens during STX2 production (Fig. 1), suggesting that STX2 production altered gastrointestinal host responses to favor STEC survival during colitis. In the presence of STX2, STEC infection led to worsened colitis scoring (Fig. 5A) characterized by significantly increased ulceration of the gastrointestinal epithelium 7 days postinfection compared to colons from mice infected with STEC(ΔSTX2) TUV86-2 (Fig. 5B). This may reflect altered epithelial barrier repair in the presence of toxin and could contribute to increased STEC burdens. Epithelial hyperplasia (Fig. 5C) and inflammatory infiltrates (Fig. 5D) were not significantly changed in the presence of toxin.

FIG 5.

STX2-producing-STEC infection increases gastrointestinal epithelial ulceration. Mice were orally gavaged with STX2-producing STEC or STEC(ΔSTX2) 0 and 2 days following 5-day exposure to 2.5% (wt/vol) DSS. Colons were removed 7 days after infection for fixation in NBF and histological processing. (A and B) Histological scoring of H&E-stained colon samples collected from mice sacrificed 7 days postinfection for total colitis (A), structural damage (B), epithelial hyperplasia (C), and inflammatory infiltration (D). STEC, n = 7; STEC(ΔSTX2), n = 10. Statistical significance was determined by one-tailed Mann-Whitney test (*, P < 0.05).

The DSS+STEC model was leveraged to evaluate critical components of the local host immune response that have not previously been evaluated during infection with clinically isolated strains of STEC. Attaching-and-effacing gastrointestinal pathogens induce specialized host gastrointestinal immunological responses along the interleukin 23 axis that are necessary for host survival and successful clearance of the pathogen (13, 14, 17). In addition, IL-23 and IL-23R-induced IL-22 are both critical cytokines in epithelial barrier function and repair (13, 15). In the DSS+STEC model, colonic transcripts for the heterodimeric subunits of IL-23 (IL23A and IL12B) were elevated during STEC(ΔSTX2) strain TUV86-2 infection but not STX2-producing STEC 86-24 infection (Fig. 6). Transcripts for IL17A and IL-22, cytokines produced by Th17 lymphocytes following IL-23R signaling, were also increased during STEC(ΔSTX2) strain TUV86-2 infection relative to DSS+PBS controls, but not during infection with the toxin-producing strain. While the reduction in IL-17A (P = 0.06) and IL-22 (P = 0.09) transcripts during STEC infection did not reach statistical significance relative to STEC(ΔSTX2) infection, the decreases were consistent with the overall pattern of IL-23 axis transcript suppression during STX2-producing infection. Colonic transcripts of CXCL1, a chemokine associated with acute DSS colitis, were unchanged between groups. Notably, decreased colonic IL-23 axis transcripts occurred despite a greater burden of STX2-producing STEC in the animals (Fig. 1). Attempts to assay colonic cytokine protein levels from the murine tissues were unsuccessful, but STX-sensitive human THP-1 cells were evaluated for lipopolysaccharide (LPS)-stimulated IL-23 production during Shiga intoxication in vitro. While STXs increased transcriptional upregulation of CXCL8 and IL23A transcripts after 12 h of LPS stimulation (Fig. 7A to C), secretion of IL-23 heterodimer into the supernatant was significantly reduced by STX exposure without affecting IL-8 protein secretion (Fig. 7D and E). Decreased IL-23 cytokine production by Shiga-intoxicated THP-1 cells in vitro was consistent with reduced IL-23R-stimulated IL17A and IL-22 colonic transcripts in the DSS+STEC model. Interestingly, STX-induced increases in THP-1 IL23A transcripts suggest that the transcriptional decreases in colonic IL23A and IL12B colonic transcripts in the DSS+STEC model may underestimate the effect of STX2 on colonic IL-23 protein production due to a previously undescribed decoupling of IL-23 transcriptional upregulation and protein secretion. Collectively, the data indicate that the host IL-23 networked response may be a particular target of STX2 toxicity.

FIG 6.

STX2a production inhibits upregulation of colonic IL-23 axis-related transcripts induced by STEC infection. Mice were orally gavaged with STX2a-producing STEC strain 86-24, STEC(ΔSTX2) strain TUV86-2, or sterile PBS on day 0 and day 2 following a 5-day exposure to 2.5% (wt/vol) DSS. Colons were removed 7 days after the first oral gavage for total RNA isolation and RT-qPCR. Fold changes were normalized to the average fold changes in PBS controls for targets CXCL1, IL23A, IL12B, IL17A, and IL-22. STEC, n = 4; STEC(ΔSTX2), n = 10; PBS, n = 8. Means ± SE are shown. Statistical significance was determined by one-tailed Student t test (*, P < 0.05; **, P < 0.01).

FIG 7.

Shiga toxins inhibit IL-23 secretion by differentiated THP-1 cells. Following differentiation via exposure to PMA for 72 h, THP-1 cells were stimulated with 100 ng/ml LPS alone or concurrently with 0.1 ng/ml STX1 or 10 ng/ml STX2. Cell viability was >90% after 12-h exposure at these toxin concentrations. (A to C) Fold changes of CXCL8 (A), IL23A (B), and IL12B (C) transcripts relative to baseline values measured by RT-qPCR of cell lysates. (D and E) Concentrations of secreted IL-8 (D) and IL-23 (E) heterodimers from culture supernatants after 12 h of stimulation measured by ELISA. Statistical significance was determined by one-way ANOVA with Tukey posttest (n = 3 or 4/group/time point) (*, P < 0.05; **, P < 0.01; ***, P < 0.001).

DISCUSSION

Colonization of an immunocompetent mouse strain with clinical-isolate strains of STEC following DSS exposure is a novel model that reproducibly establishes gastrointestinal infection by STEC with toxin-induced renal inflammation and tubular injury without severe depletion of the gastrointestinal microbiota. It is notable that not all the mice succumbed to infection, similar to outcomes observed in humans and less severe toxin models. Increasing the overall burden of STX2-producing STEC via infection with two oral gavages increased the incidence of morbidity and renal tubular injury in the DSS+STEC model, suggesting that increased exposure to STX2 and other STEC virulence factors led to higher rates of disease. We further showed how the DSS+STEC model can be applied for study of host defense mechanisms in an in vivo context. Compared to the STX2a deletion mutant strain, exposure to STX2a during infection resulted in increased intestinal epithelial damage, increased STEC burdens, and inhibition of colonic IL-23 axis transcriptional upregulation. These results demonstrate the utility of the model to delineate the impacts of STXs on host responses during infection that promote pathogen survival. We also observed intestinal colonization using this approach with two clinical-isolate STEC strains, suggesting that DSS-induced STEC susceptibility may be broadly applicable and could be of use to work out the differences and similarities of the newly emerging hybrid strains with respect to colonization, virulence factors, pathology, and therapeutic opportunities.

Determining the mechanistic causes of DSS-induced susceptibility to STEC colonization may provide insights regarding clinical STEC infection with heterogeneous progression to systemic Shiga toxicosis. DSS damages the colonic epithelium and thus impairs epithelial barrier defenses, including antimicrobial protein secretion, mucus production from goblet cells, and epithelial secretion of microRNA (miRNA) sequences that normally prevent colonization by pathogenic bacteria (13, 40). Thus, a niche for E. coli growth during DSS colitis could be a direct result of the loss of host defenses that actively inhibit STEC colonization. Alternatively, alteration of the gastrointestinal microbiota during DSS colitis contributes to facilitating E. coli growth by altering nutrient or biochemical conditions within the colonic microenvironment (29, 41). E. coli is known to outcompete other commensal bacteria during acute alterations in the gut environment, and increased availability of free sialic acid and nitrates during colitis contributes to outgrowths of commensal E. coli in mice (29, 42, 43). Study of the mechanism of DSS-induced susceptibility to STEC colonization in mice could potentially provide insight into host factors associated with risk of developing severe hemorrhagic colitis during STEC infection or progression to systemic Shiga toxicosis due to increased colonization by or delayed clearance of STEC.

Increased morbidity in the DSS+STEC mice was associated with histologic evidence of renal tubular injury, increased renal transcripts of the molecular injury markers KIM1 and NGAL, and increased renal inflammatory IL-6 and CXCL1 transcripts, as would be expected during STX2-induced pathology. Renal injury was not observed in DSS-pretreated mice infected with the STEC(ΔSTX2) TUV86-2 strain despite similar initial levels of colonization, indicating that STX2 production by STEC was necessary for renal injury. Bacteremia is not a recognized component of systemic Shiga toxicosis in patients (1), a condition that is recapitulated in the DSS+STEC model. There is a dose response to STX2-producing STEC in the model, with increased STEC burden and potential STX2 exposure via multiple challenges resulting in a higher incidence of increased morbidity. While histologic evidence of STEC attaching-and-effacing lesions or STX localization within the murine intestinal epithelial cells was not determined, it has previously been shown that close association of STX-producing bacterial pathogens and the host intestinal epithelium is required for systemic injury (23, 44). The DSS+STEC model recapitulates key components of clinical STEC infection by establishing gastrointestinal infection by clinical-isolate strains of STEC during colitis, followed by variable progression to STX-induced systemic disease without bacteremia in immunocompetent mice.

STX2a-producing-STEC infection in the DSS+STEC model resulted in increased STEC burdens and decreased host colonic IL-23 axis transcriptional responses compared to infection with the toxin deletion mutant strain. This is a novel finding, suggesting that STX2 targets critical host immune responses to an attaching-and-effacing pathogen, and one result may be to inhibit STEC clearance. STX2 activity may prolong infection by inhibiting IL-23- and IL-22-induced epithelial barrier functions during colitis or by reducing recruitment of phagocytes induced by IL-17 signaling downstream of IL-23R stimulation (45). Pathology evaluation of colons from DSS+STEC mice suggested that repair of epithelial ulcerations was inhibited during STX2-producing-STEC infection without altering the severity of inflammatory cell infiltration, although changes in the quantity or spatial distribution of inflammatory infiltrate phenotypes were not characterized or quantified. Decreased LPS-induced IL-23 secretion by macrophage-like human THP-1 cells during Shiga intoxication supported colonic transcriptional data from the DSS+STEC model, but further in vivo characterization of primary cellular and tissue responses to Shiga toxins and STEC is required to define the mechanism of IL-23 axis suppression by STX2 due to the differences in cell types, microbial stimulation, and duration of stimulation between the two systems. The current study also did not include STEC-producing STX1 and/or multiple STX2 isoforms, so the impact of these toxins on colonization and host-pathogen interactions in this context is unknown. The DSS+STEC mouse model is a novel tool that can provide molecular and cellular insights into the role of pathogen-host interactions within the gastrointestinal tract during infection, filling a key need in the field of STEC and Shiga toxin research.

MATERIALS AND METHODS

Mice.

Male and female 6-week-old C57BL/6 mice were purchased from Charles River Laboratories (Wilmington, MA) and housed in disposable cages during experiments (4 mice per cage). Daily monitoring was performed, and weights were recorded. Mice were euthanized if weight loss exceeded 20% of the body weight at the time of the first oral gavage or if the mouse was moribund. The mice were housed and used in accordance with the guidelines and approved protocols of the Institutional Animal Care and Use Committee and the Institutional Biosafety Committee of Boston University School of Medicine.

Bacterial strains.

STEC strain TW14359 (STX2a⁺ STX2c⁺ STX1⁻) was originally isolated from a Michigan patient infected by consumption of contaminated spinach (46). TW14359 was stored in LB broth at −80°C prior to use in experiments. STEC strain 86-24 (STX2a⁺ STX1⁻), originally isolated from a patient in Washington, and its isogenic STX2a deletion mutant strain TUV86-2 were kindly provided by John Leong (Tufts University School of Medicine, Boston, MA) (32). The presence of stx2 (wild-type strains), lack of stx1, and deletion of stx2 (TUV86-2) were confirmed in each strain via qPCR using targeted primers: Stx2a F, GCA TCC AGA GCA GTT CTG C; Stx2a R, GCG TCA TCG TAT ACA CAG GAG; stx1 F, CAA GAG CGA TGT TAC GGT TTG; and stx1 R, GTA AGA TCA ACA TCT TCA GCA GTC.

DSS exposure.

Colitis-grade DSS (MP Biomedicals, Santa Ana, CA) was dissolved in distilled water (dH₂O) at a concentration of 2.5 g of DSS per 100 ml distilled, deionized water (2.5% [wt/vol] DSS). Each DSS suspension was passed through a sterile 0.22-μm polyethersulfone (PES) membrane filter (Sigma-Aldrich, St. Louis, MO) prior to use in experiments. Fifty milliliters of sterilized 2.5% (wt/vol) DSS solution was added to sterile disposable water bottles on day 0 of experiments. The DSS was discarded and replaced with freshly dissolved and sterilized 2.5% (wt/vol) DSS every 48 h. After 5 days, the 2.5% DSS was removed and the mice were given access to water ad libitum. The mice were randomly reassigned between cages immediately prior to oral gavage to minimize any potential cage-dependent effects of DSS treatment. The mice were weighed daily and observed (blinded) for signs of clinical morbidity (e.g., huddling, ruffled fur, or sluggish movement). Mice were euthanized if they met approved euthanasia criteria. Mice losing >5% of their starting body weight within 5 days of starting DSS treatment were excluded from experiments prior to oral gavage (3 mice total).

Oral gavage.

Frozen stocks of STEC strains TW14359, 86-24, and TUV86-2 were used to inoculate 5 ml of LB broth (Sigma-Aldrich) for overnight incubation at 37°C with shaking at 200 rpm in a floor incubator. The resulting cultures were streaked on LB agar plates (Sigma-Aldrich) and incubated at 37°C overnight. The following day, individual colonies were selected to inoculate 5-ml aliquots of LB broth for incubation at 37°C with 200-rpm shaking for 14 to 16 h. Cultures were spun at 1,000 × g for 10 min to pellet bacteria, the supernatant was aspirated, and the pellet was then resuspended in 500 μl of sterile PBS (GE Life Sciences, Pittsburgh, PA). Bacterial suspensions were drawn into a 1-ml syringe for oral gavage. Food was withheld from the mice 30 min prior to oral gavage. The mice were anesthetized using isoflurane gas (Henry Schein, Melville, NY) in a drop jar and then orally gavaged with 100 μl of bacterial suspension via a gavage needle inserted esophageally to the level of the stomach. Gavage doses were approximately 1 × 10⁹ to 5 × 10⁹ CFU of bacteria. Control mice were anesthetized and gavaged with 100 μl of sterile PBS. The gavage suspension was serially diluted in sterile PBS, plated on LB agar, and incubated at 37°C overnight to determine the number of CFU administered.

Organ and blood cultures.

DSS-exposed mice were orally gavaged with STEC strain TW14359 on days 5 and 7 after starting DSS. The mice were sacrificed 0, 2, 4, or 7 days following the first oral gavage for sample collection. The distal right middle liver lobe was removed under sterile conditions and placed in a preweighed sterile microcentrifuge tube on ice. The cecum was then reflected cranially, and the caudal mesenteric lymph node was located for removal to a separate preweighed tube on ice. The heart was exposed via diaphragmatic incision for cardiac blood collection. Right ventricular blood was aspirated using a 25-gauge needle on a 1-ml syringe and plated on blood agar for overnight aerobic incubation at 37°C. The organs were weighed and then homogenized in 10 μl of PBS/mg of tissue using a sterile syringe plunger. The resulting homogenate was spun at 300 × g for 5 min at 4°C, and the supernatant was serially diluted and plated on blood agar. The plates were incubated under aerobic conditions at 37°C overnight for determination of the number of CFU per gram of tissue. The limit of detection of organ cultures was 1 × 10³ CFU/g of tissue.

Fecal cultures.

During DSS exposure, daily fecal samples were placed in sterile preweighed tubes, weighed, and then homogenized in 10 μl of sterile PBS per mg of feces using a sterile toothpick. Serial dilution of feces was performed in sterile PBS for plating on MacConkey agar plates with sorbitol (BD Biosciences, San Jose, CA). The number of CFU per gram of feces was determined following overnight aerobic incubation at 37°C.

Fecal qPCR.

Following oral gavage, fecal samples were collected daily and stored at −80°C. Fecal DNA was isolated using QIAamp DNA stool minikits (Qiagen, Valencia, CA) following the manufacturer's instructions. The DNA mass was quantified via 260-nm absorbance on a Nanodrop 2000c (Thermo Scientific, Waltham, MA) and diluted to 20 ng/μl in Tris-EDTA buffer (Fisher Scientific, Waltham, MA). DNA isolation from weighed naive mouse fecal pellets spiked with serial dilutions of an STEC culture in LB broth was performed to generate standards. The number of CFU per gram of fecal DNA standards was determined by plating serial dilutions of STEC culture used to spike naive fecal pellets on LB agar overnight at 37°C. qPCR was performed using Quantifast SYBR green qPCR kits (Qiagen, Valencia, CA) on a StepOnePlus (Applied Biosystems, Foster City, CA) following the manufacturer's instructions, with 50 ng of DNA for each PCR. A standard curve of DNA isolated from a series of STEC CFU-per-gram standards was run on each plate to quantify colonization, with 1 CFU/g being equal to 1 copy of eae/g. The primers used were eae F, AAA GCG GGA GTC AAT GTA ACG, and eae R, GGC GAT TAC GCG AAA GAT AC.

Colon lengths.

Following sacrifice, the abdomen was longitudinally incised and the colon was removed via sharp dissection at the cecocolic junction and the pelvic inlet for measurement.

Serum BUN and creatinine measurements.

Blood samples were collected via facial vein just before euthanasia and allowed to clot at room temperature for 30 min. Samples were spun at 1,000 × g for 10 min, and serum was removed for storage at −80°C. Serum BUN and creatinine concentrations were determined using a Quantichrom urea assay kit (Bioassay Systems, Hayward, CA) and a Quantichrom creatinine assay kit (Bioassay Systems), following the manufacturer's instructions. The optical density (OD) was measured with a VersaMax microplate reader (Molecular Devices LLC, San Jose, CA).

Tissue RT-qPCR.

Kidneys and segments of proximal and distal colons of euthanized mice were removed and fixed in RNAlater (Ambion, Waltham, MA) at 4°C for 48 to 96 h. The tissues were then stored at −80°C in RNAlater until use. The tissues were disrupted and homogenized using 4-mm stainless steel beads in a TissueLyzer II (Qiagen). RNA was then isolated using RNeasy Plus minikits (Qiagen) in a Qiacube (Qiagen). The RNA concentration was quantified via 260-nm absorbance on a Nanodrop 2000c. Reverse transcription (RT) reactions using Quantitect RT kits (Qiagen) in a Veriti 96-well thermal cycler (Applied Biosystems, Foster City, CA) were performed with 500 ng RNA per RT reaction. The resulting cDNA was diluted to 14.3 ng/μl in Tris-EDTA for use in qPCRs using Quantifast SYBR green qPCR kits in a Step One Plus. A total of 35.7 ng of cDNA was used in each PCR. HPRT was used as an endogenous control for all targets to account for variation in DNA content between samples. The primers used for qPCR of total renal cDNA were as follows: HPRT F, TGG GCT TAC CTC ACT GCT TTC; HPRT R, CCT GGT TCA TCA TCG CTA AT; NGAL F, CCC TGT ATG GAA GAA CCA AGG A; NGAL R, CGG TGG GGA CAG AGA AGA TG; KIM1 F, GGA GAT ACC TGG AGT AAT CAC ACT G; KIM1 R, TAG CCA CGG TGC TCA CAA GG; CXCL1 F, TCA AGA ACA TCC AGA GCT TGA AG; CXCL1 R, GGA CAC CTT TTA GCA TCT TTT GG; IL-6 F, TCC AAT GCT CTC CTA ACA GAT AAG; IL-6 R, CAA GAT GAA TTG GAT GGT CTT G; CCL3 F, TGT TTG CTG CCA AGT AGC CAC ATC; CCL3 R, AAC AGT GTG AAC AAC TGG GAG GGA. The primers used for qPCR of total colonic cDNA were as follows: HPRT F, TGG GCT TAC CTC ACT GCT TTC; HPRT R, CCT GGT TCA TCA TCG CTA AT; CXCL1 F, TCA AGA ACA TCC AGA GCT TGA AG; CXCL1 R, GGA CAC CTT TTA GCA TCT TTT GG; IL23A F, GGT GGC TCA GGG AAA TGT; IL23A R, GAC AGA GCA GGC AGG TAC AG; IL12B F, ACA TCT ACC GAA GTC CAA TGC A; IL12B R, GGA ATT GTA ATA GCG ATC CTG AGC; IL17A F, GCT CCA GAA GGC CCT CAG A; IL17A R, CTT TCC CTC CGC ATT GAC A; IL-22 F, TCC GAG GAG TCA GTG CTA AA; IL-22 R, AGA ACG TCT TCC AGG GTG AA.

Renal pathology.

Transverse sections of central portions of kidneys from euthanized mice were fixed in 10% neutral buffered formalin (NBF) within 5 min of harvest (Thermo Scientific) for paraffin embedding, sectioning, mounting, and staining by the Boston University School of Medicine Experimental Pathology Laboratory Services Core facility. H&E-stained slides were evaluated blindly for evidence of tubular damage, as evidenced by tubular dilation, tubular epithelial flattening, and sloughing of tubular cells. Scoring was as follows: 0, no evidence of tubular damage; 1, <5% of tubules damaged; 2, 5 to 15% of tubules damaged; 3, 15 to 25% of tubules damaged; 4, >25% of tubules damaged. PAS-stained slides were evaluated blindly for renal tubular brush borders. The outer cortices of kidneys were evaluated at ×400 magnification, with at least 5 fields used to determine the brush border status of 200 tubules per sample.

Colon pathology.

The distal portion of the proximal colon and the proximal portion of the descending colon were fixed in 10% neutral buffered formalin for processing, as described for kidney samples. H&E slides were blindly scored for colitis using scoring systems based on the methods used to score DSS colitis and C. rodentium models (47). Scoring of epithelial hyperplasia (0, normal; 1, <10% increase in crypt depth; 2, 10 to 25% increase in crypt depth; 3, 25 to 50% increase in crypt depth; 4, >50% increase in crypt depth), inflammatory cell infiltration (0, no inflammatory infiltrates; 1, infiltrates limited to mucosal layer; 2, infiltrates limited to mucosa and submucosa; 3, submucosal infiltrates with focal transmural infiltration; 4, pronounced transmural infiltration), and structural damage (0, epithelium intact; 1, erosions of epithelium present; 2, focal ulcerations of epithelium present; 3, multifocal ulcerations present with granulation tissue) was recorded. A total colitis score was calculated as the sum of individual scores (0-to-11 scale).

Cell culture.

Human monocytic leukemia THP-1 cells (ATCC, Manassas, VA; TIB-202) were grown in RPMI medium (GE Healthcare, Marlborough, MA) supplemented with 10% fetal bovine serum (Gibco, Waltham, MA), 50 U/ml penicillin (Gibco), 50 μg/ml streptomycin (Gibco), and 1× GlutaMAX (Gibco) at 37°C with 5% CO₂ in a humidified incubator.

THP-1 stimulation during Shiga intoxication.

At passages 5 to 7, 2 × 10⁵ to 3 × 10⁵ THP-1 cells were plated in 24-well plates with RPMI medium supplemented with 10% fetal bovine serum, 50 U/ml penicillin, 50 μg/ml streptomycin, 1× GlutaMAX, and 100 nM phorbol-12 myristyl acetate (PMA) (Abcam, Cambridge, MA) for 72 h to induce differentiation into a macrophage-like state. The cells were washed twice with PBS and then cultured with serum-free RPMI medium with 50 U/ml penicillin, 50 μg/ml streptomycin, and 1× GlutaMAX for 24 h. Following serum starvation, the cells were washed twice with PBS and then cultured in 500 μl of RPMI medium supplemented with 10% fetal bovine serum, 50 U/ml penicillin, 50 μg/ml streptomycin, 1× GlutaMAX, 100 ng/ml of LPS from E. coli O127:B8 (Difco, San Jose, CA) with or without 0.1 ng/ml STX1 (Tufts University, Boston, MA) or 10 ng/ml STX2 (Tufts University, Boston, MA). The culture supernatants were collected after 12 h and flash-frozen for storage at −80°C. The supernatants were then thawed on ice for enzyme-linked immunosorbent assays (ELISAs) in 96-well Nunc plates (Thermo Scientific) using a human Ready-Set-Go IL-23 ELISA kit (eBioscience, Waltham, MA) and a human IL-8 Ready-Set-Go ELISA kit (eBioscience) according to the manufacturer's instructions.

Statistical analysis.

Data were analyzed using GraphPad PRISM as described in the figure legends.