Two days prior to infection, CR(ΦStx_2dact) and CR(ΦStx_2dact::kan^R), described previously (Mallick et al., 2012, 2014), were streaked from frozen glycerol stock onto LB agar containing the appropriate antibiotic (i.e., 10 μg/ml chloramphenicol for CR(ΦStx_2dact), and 10 μg/ml chloramphenicol and 25 μg/ml kanamycin CR(ΦStx_2dact::kan^R) for single colonies. The following day, a single colony was inoculated into 40 ml LB broth (without antibiotic) in a 50 ml conical tube (VWR, catalog number 89039-662) and incubated at 37°C with 5% CO₂, without shaking but with the cap loosened to allow for gas exchange. Cultures were incubated typically for 10–12 h, to achieve an OD₆₀₀ ~ 0.6–0.7. The culture was centrifuged at 10294 × g for 12 min at room temperature. The bacterial pellet was resuspended in 500 μl of Dulbecco's Phosphate-buffered saline DPBS (Life technologies, catalog number 14190-144) and then centrifuged at room temperature at 5226 × g for 2 min. Bacterial pellets were resuspended in 60 μl of DPBS. Six micro liter of a resuspended inoculum, encompassing ~ 1 × 10⁸ colony-forming units (cfu), was carefully pipetted onto a ~35 mg piece of Teklad 2918 irradiated rodent chow, visually ensuring that the inoculum was fully absorbed into the food (~1 min). For smaller inoculums, overnight cultures were appropriately diluted in PBS and into the food as described above. To titer the dose, a sample of the rodent chow inoculated in parallel was resuspended in 1 ml of DPBS, vortexed to disperse chow and bacteria, serially diluted, and plated on LB agar containing the appropriate antibiotic (i.e., 10 μg/ml chloramphenicol for CR(ΦStx_2dact), and 10 μg/ml chloramphenicol and 25 μg/ml kanamycin CR(ΦStx_2dact::kan^R).

All procedures involving live animals were pre-approved by Tufts University Institutional Animal Care and Use Committee (IACUC). Five to six week old C57BL/6 mice were purchased from Jackson Laboratory and housed for 1–2 weeks in the animal facility at Tufts University School of Medicine prior to the start of infection. One day prior to infection, mice were weighed, cohoused on raised wire flooring to prevent coprophagy (Thoren caging systems: CC90F01), and food-restricted for 12 h. After food restriction, each mouse was individually placed into an empty cage and presented with a ~35 mg piece of inoculated rodent chow. After each mouse consumed the entire food inoculum, mice were weighed and returned to the original cage (without raised wire flooring) with access to unlimited food and water. For dose-response studies, mice were returned to caging with raised wire flooring to prevent coprophagy for the duration of the infection. Infection was followed by monitoring of CR(ΦStx_2dact) or CR(ΦStx_2dact::kan^R) titers in the feces, as follows. Fecal samples, were weighed, suspended in DPBS at a volume of 10 μl per 1 mg of stool, and suspensions were serially diluted and plated on agar containing the appropriate antibiotic (10 μg/ml chloramphenicol for CR(ΦStx_2dact), and 10 μg/ml chloramphenicol and 25 μg/ml kanamycin CR(ΦStx_2dact::kan^R). Mice were weighed and observed for clinical distress (ruffled fur, paucity of movement, and shivering) each day. Mice with greater than 12% body weight loss with or without signs of distress, or 10% body weight loss with signs of distress, were euthanized by CO₂ inhalation followed by cervical dislocation. In addition, in some experiments, colonic and luminal colonization was measured by sacrificing mice at day 3, 4, or 5 post-infection. Each colon was dissected and stool contents were removed by gently pushing the stool through the colon with the smooth edge of a scalpel. The colonic luminal contents were collected by gently flushing the colons with 1 ml of DPBS. Colonic tissues were homogenized in 500 μl of DPBS using a Fisher Scientific PowerGen 125 Homogenizer. To determine the titer of CR(ΦStx_2dact), colonic homogenates and luminal wash were serially diluted and plated on LB agar containing the appropriate antibiotic, as described above. For each weight loss experiment, the data passed the KS normality test.

The growth conditions of the inoculum are likely to influence the course of infection and disease. For example, host-adapted C. rodentium present in the feces of infected mice are in a hyper-infectious state, i.e., highly transmissible and capable of accelerated colonic colonization, compared to C. rodentium grown in LB (Wiles et al., 2005). EHEC grown in gnotobiotic piglets also exhibit enhanced virulence-related phenotypes (Brady et al., 2011). Thus, in considering ways to develop a consistent and predictable CR(ΦStx_2dact) infection model, we first revisited the preparation of the inoculum. The previous described CR(ΦStx_2dact) infection model, utilized bacteria that had been grown overnight to stationary phase (Mallick et al., 2012), a growth phase associated with global physiologic changes (Battesti et al., 2011), some of which alter infectivity and/or virulence (Molofsky and Swanson, 2004; Sonenshein, 2005). In addition, the CR(ΦStx_2dact) cultures had previously been shaken, which facilitates aerobic growth and is predicted to sensitize C. rodentium to acid killing (Smith and Bhagwat, 2013). Hence, in our current study, we grew CR(ΦStx_2dact) in a 5% CO2 atmosphere without shaking, a condition predicted to temper aerobic growth. These cultures were grown to an OD600 = ~0.6–0.7.

In the previously described CR(ΦStx_2dact) model (Mallick et al., 2012), bacteria were delivered to the intestinal tract by oral gavage, a method that not only risks aspiration and esophageal and/or gastric puncture, but also induces a host stress response that might alter the course of infection (Hoggatt et al., 2010; Walker et al., 2012). Methods for intestinal delivery of pharmacological agents by feeding have been established that minimize this stress response (Walker et al., 2012). Further, delivery of Listeria monocytogenes in food rather than by gavage altered the fate of bacteria in the host and the virulence factors required for infection. (Bou Ghanem et al., 2012). Hence, 6–8 week old C57BL/6J mice were inoculated by ingestion of 1 × 10⁸ CFU of CR(ΦStx_2dact) absorbed into a small (~35 mg) piece of rodent chow (Figure 1 and Materials and Methods).

To determine if inoculation of mice with CR(ΦStx_2dact) in this manner resulted in efficient intestinal colonization, we used fecal shedding CR(ΦStx_2dact) as a non-terminal indicator of colonic colonization: on successive days following infection, we plated for viable CR(ΦStx_2dact) on selective media. High levels of fecal shedding, i.e., ~10⁷ cfu/g of feces, were observed on day 3 post-infection, increasing to >10⁸ cfu/g of feces on days 4 and 5 post-infection (Figure 2A). High levels of colonization were also observed in colonic luminal washes and colonic homogenates following sacrifice. These titers of CR(ΦStx_2dact) were roughly equal to the peak fecal shedding, colonic colonization, and luminal washes previously reported (Mallick et al., 2012, 2014) by intra-gastric inoculation. These data indicate that ingestion of food inoculated with mid-log phase bacteria grown without shaking results in efficient intestinal colonization.

The production of Stx by C. rodentium has been associated with slightly higher levels of intestinal colonization following inoculation by oral gavage (Mallick et al., 2012, 2014). Therefore, we also infected mice with CR(ΦStx_2dact::kan^R), a CR(ΦStx_2dact) mutant that does not produce Stx. Fecal shedding of CR(ΦStx_2dact::kan^R) was indistinguishable from that of CR(ΦStx_2dact) (Figure 2A), and CR(ΦStx_2dact::kan^R) titers were similar to those of CR(ΦStx_2dact) in luminal washes and colonic homogenates. Thus, at least for this small sample size (n = 5), Stx did not detectably promote colonization in this model (Figure 2).

Mice are typically gavaged with 10⁸ C. rodentium to initiate productive infection. To determine if inoculation of C57/BL6J mice with lower doses of CR(ΦStx_2dact) by our feeding model would result in productive infection and disease, we fed groups of 10 mice doses of CR(ΦStx_2dact) spanning four orders of magnitude, from 3.5 × 10⁸ down to 3 × 10⁴. To avoid coprophagy, which due to the ingestion of bacteria shed in feces would confound accurate quantitation of the oral dose, the mice were housed on elevated wire floors for the duration of the infection. Fecal shedding of CR(ΦStx_2dact) revealed that although the kinetics of intestinal colonization appeared to be slightly delayed in mice receiving the lowest dose, all groups of mice were colonized by day 5 post-infection and reached >10⁸/g stool by day 7 (Figure 3A). Similarly, all groups of mice lost weight, the kinetics of which varied with dose: the first indication of significant weight loss occurred on post-infection day 7 for the group of mice infected with 7.5 × 10⁷ or 3.5 × 10⁸ CR(ΦStx_2dact), day 8 for the group infected with 3.6 × 10⁶, and day 9 for the group infected with 3.0 × 10⁴ (Figure 3B). Finally, all groups of mice became moribund and required euthanasia (Figure 3C). Consistent with the trends observed for the kinetics of intestinal colonization and weight loss, mice infected with lower doses exhibited a somewhat longer interval between inoculation and required euthanasia: by day 7 post-infection, the group of mice infected with 3.5 × 10⁸ CR(ΦStx_2dact) were all euthanized, whereas for groups of mice infected with 7.5 × 10⁷, 3.6 × 10⁵, and 3.0 × 10⁴ bacteria, the day at which 100% were euthanized was day 8, 9, and 10, respectively. We conclude that feeding of C57/BL6J mice with as few as 3.0 × 10⁴ CR(ΦStx_2dact) results in productive and lethal infection.

Previous work revealed that the kinetics of disease following gavage inoculation with stationary phase CR(ΦStx_2dact) varied widely, e.g., resulting in required euthanasia as early as 4 days and as late as 9 days post-infection within a given experiment (Mallick et al., 2012). In contrast, the trend of delayed colonization and disease kinetics observed for mice given different doses of CR(ΦStx_2dact) by feeding suggested that the time course of disease in this model may be more consistent than by oral gavage. To carefully assess the time course of disease, in multiple experiments mice were weighed daily after inoculation with mouse chow containing the relatively high dose of 1 × 10⁸ CR(ΦStx_2dact), or, as a control to evaluate Stx-dependent disease, CR(ΦStx_2dact::kan^R). (For simplicity of the protocol, and in contrast to the dose-response studies described above, mice were not housed on wire floors after inoculation, raising the possibility that the dose may be somewhat higher due to coprophagy.) In our first experiment, all five mice infected with CR(ΦStx_2dact) crossed a threshold of 4.5% body weight loss on day 5 post infection, whereas mice infected with CR(ΦStx_2dact::kan^R) did not suffer weight loss (Figure 4A; Table 1, Experiment 1). In addition and in contrast to mice infected with the Stx-deficient strain, which uniformly survived infection, all mice infected with CR(ΦStx_2dact) required euthanasia on day 6 (Figure 4; Table 1). To assess the reproducibility of the course of disease in this model, we repeated the experiment three times for a total of four independent experiments encompassing 20 mice (Table 1). Although the kinetics of weight loss and required euthanasia for Experiments 2 through 4 were not as uniform as in Experiment 1, cumulatively 70% of mice reached the threshold of >4.5% body weight loss on day 5 post-infection, and all mice required euthanasia on day 6 or 7 (Table 1). We conclude that this infection protocol, in contrast to the gavage model using stationary phase CR(ΦStx_2dact), results in a highly synchronous course of disease.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.