Oral Administration with Live Attenuated Citrobacter rodentium Protects Immunocompromised Mice from Lethal Infection

ABSTRACT

Enteropathogenic Escherichia coli (EPEC) and enterohemorrhagic E. coli (EHEC) are important causative agents for foodborne diseases worldwide. Besides antibiotic treatment, vaccination has been deemed as the most effective strategy for preventing EPEC- and EHEC-caused foodborne illnesses. Despite substantial progress made in identifying promising antigens and efficacious vaccines, no vaccines against EPEC or EHEC have yet been licensed. Mice are inherently resistant to EPEC and EHEC infections; infection with Citrobacter rodentium (CR), the murine equivalent of EPEC and EHEC, in mice has been widely used as a model to study bacterial pathogenesis and develop novel vaccine strategies. Mirroring the severe outcomes of EPEC and EHEC infections in immunocompromised populations, immunocompromised mouse strains such as interleukin-22 knockout (Il22^−/−) are susceptible to CR infection with severe clinical symptoms and mortality. Live attenuated bacterial vaccine strategies have been scarcely investigated for EPEC and EHEC infections, in particular in immunocompromised populations associated with severe outcomes. Here we examined whether live attenuated CR strain with rational genetic manipulation generates protective immunity against lethal CR infection in the susceptible Il22^−/− mice. Our results demonstrate that oral administration of live ΔespFΔushA strain promotes efficient systemic and humoral immunity against a wide range of CR virulence determinants, thus protecting otherwise lethal CR infection, even in immunocompromised Il22^−/− mice. This provides a proof of concept of live attenuated vaccination strategy for preventing CR infection in immunocompromised hosts associated with more severe symptoms and lethality.

INTRODUCTION

Enteropathogenic Escherichia coli (EPEC) and enterohemorrhagic E. coli (EHEC) are among leading etiological agents for foodborne diseases, which have remained an enormous economic burden and a significant public health issue around the world (1–6). These sophisticated human pathogens utilize a type III secretion system (T3SS) to intimately adhere to intestinal epithelial cells and form the characteristic attaching/effacing (A/E) lesions, resulting in diarrheal symptoms and self-limiting infections in immunocompetent populations (1, 2, 7). EPEC is a frequent cause of infantile acute diarrhea in the lower middle-income countries (LMICs), whereas Shiga toxin-producing EHEC causes a wide spectrum of diseases ranging from mild diarrhea to severe clinical complications including hemorrhagic colitis and hemolytic uremic syndrome (HUS) (1–5). Citrobacter rodentium (CR) shares approximately 66.7% of encoded genes and most pathogenic mechanisms including the T3SS and A/E lesions with human pathogens, thus being considered a murine equivalent of EPEC and EHEC (8–10). CR infection in mice has been widely used as a small animal model to study EPEC/EHEC pathogenic mechanisms and host immune responses against enteric bacterial infections, since EPEC and EHEC fail to colonize and infect mice (8–12). A growing amount of evidence demonstrates that while immunocompetent people suffer mild symptoms and recover quickly from EPEC and EHEC infections, those infections in immunocompromised populations may result in more severe symptoms, even lethality (13, 14). Consistently, immunocompetent mouse strains display little to no mortality to CR infection; in contrast, certain immunocompromised mice such as interleukin-22 knockout (Il22^−/−) and C3HeJ mice are susceptible to CR infection with substantially elevated mortality rates (8–10, 15, 16). IL-22 plays a critical role in inflammation, tissue protection, regeneration, and antimicrobial defense (17). In particular, human populations with IL-22 deficiency, such as autoimmune polyendocrine syndrome type patients (carrying autoantibodies against IL-22), are more susceptible to various infections than healthy controls with normal IL-22 levels (18–20). Of note, the pathogenic mechanisms of A/E pathogen infections have been actively investigated in immunocompetent hosts; however, the crucial host–pathogen interactions during infections in immunocompromised hosts, in particular those associated with severe symptoms and mortality, remain elusive.

The severe outcomes in immunocompromised mice during CR infection indicate that some disease phenotypes and complex virulence regulatory mechanisms could be masked by the mild symptoms and self-limiting infection of CR under immunocompetent conditions. Our recent study using CR infection in Il22^−/− mice as an immunocompromised infection model reveals that E. coli secreted protein F (EspF), a multifunctional effector secreted via T3SS (21–26), is crucial for CR infection-caused severe morbidity and mortality in Il22^−/− mice via damaging tight junction (TJ) strands and disrupting colon epithelial barrier function (16). Chromosomal deletion of espF (encoding EspF), but not other TJ-damaging effector genes espG (encoding E. coli secreted protein G) and map (encoding mitochondrial associated protein), completely attenuates CR-induced lethality in Il22^−/− mice (16). In contrast, similar to wild-type CR, EspF knockout (ΔespF) strain exhibits almost identical colonization, proliferation, and clearance in the infected mice, and causes colonic inflammation, indicative of infection-elicited host immune responses. The strikingly attenuated virulence of ΔespF CR, even in the immunocompromised Il22^−/− mice, hints at the potential for this strain as a prototype to develop live attenuated vaccines that mimic natural infections for preventing A/E pathogen infections associated with severe outcomes in immunocompromised hosts; in particular, no vaccines against EPEC or EHEC have been licensed yet so far (27).

While mimicking natural infection and eliciting stronger immune response, live attenuated bacterial vaccines may harbor the risk of residual virulence, especially for immunocompromised hosts. We recently showed that during A/E pathogen infections, UshA, a genotoxin with DNA cleavage activity, is injected via the T3SS into host cells, where it elicits DNA damage and promotes far-reaching impact on accelerating colonic tumor formation (28). In this study, we further attenuated the DNA damaging capacity of ΔespF CR by generating ΔespFΔushA CR to avoid the potential of CR infection for tumorigenesis, examined whether infection with ΔespFΔushA provides protective immunity, and assessed its impact on CR infection in the susceptible Il22^−/− mice as well as immunocompetent C57BL/6J mice. Our results demonstrate that the oral administration of live avirulent CR strain offers significant protective effects against the otherwise lethal CR infection, even in immunocompromised hosts.

RESULTS

Deletion of EspF and UshA attenuates CR infection-caused morbidity and mortality in Il22^−/− mice.

To eliminate UshA-conferred long-term impact of CR infection on colon tumorigenesis, we generated ΔespFΔushA strain where the ushA gene was chromosomally deleted in the parental ΔespF CR (Fig. 1A). While UshA deletion was previously reported to compromise growth of nonpathogenic E. coli in minimal media (29), ΔespF, ΔushA, or ΔespFΔushA strain, in comparison to wild-type CR, exhibited no growth defects when cultured in rich medium LB broth or Dulbecco's Modified Eagle Medium (DMEM) that mimics host conditions (Fig. 1B). These results suggest that double knockout of EspF and UshA does not impact CR metabolism and growth in vitro. To assess the impacts of EspF and UshA on CR virulence, we infected Il22^−/− mice with wild-type, ΔespF, ΔushA, and ΔespFΔushA CR strains. Although the colonization and proliferation of these strains in the infected animals were almost identical, as analyzed by viable CR recovered from the feces postinfection (Fig. 1C), infection with wild-type CR resulted in consistent body weight loss, elevated clinical scores, and 100% lethality within 14 days postinoculation (dpi), whereas ΔespF CR failed to do so (Fig. 1D to F), consistent with our previous observations (16). Interestingly, infection with ΔushA CR caused comparable body weight loss, severe diarrhea, and mortality to wild-type CR in Il22^−/− mice (Fig. 1D-F), indicating a dispensable role of UshA in CR virulence. In contrast, Il22^−/− mice displayed substantial resistance to ΔespFΔushA CR infection, with transient body weight loss, mitigated diarrhea severity, and abolished mortality (Fig. 1D–F). These results suggest that ΔespFΔushA CR shares comparable virulence to its parental ΔespF strain during lethal infection in Il22^−/− mice and that UshA deletion doesn’t affect the replication, colonization, and virulence of CR in vitro and in vivo. Thus, ΔespFΔushA CR is an avirulent strain with substantially attenuated pathogenicity in the immunocompromised Il22^−/− mice.

FIG 1.

EspF and UshA deletion attenuates CR infection-caused morbidity and mortality in Il22^−/− mice. (A) Upper, diagrams show Citrobacter rodentium (CR) espF and ushA genes and PCR products amplified using the gene-specific primer pairs. Bottom, representative image of amplified PCR products using the indicated primer pairs from the CR strains as indicated. (B) Growth curves of wild-type (WT), ΔespF, ΔushA, and ΔespFΔushA CR in LB medium or Dulbecco's modified Eagle medium (DMEM), at 1:100 dilutions from the overnight cultures. OD₆₀₀, optical density measured at a wavelength of 600 nm. (C) Colonization kinetics of CR in Il22^−/− mice inoculated with 2 × 10⁹ CFU of each CR strain as indicated. (D–E) Body weight loss (D) and clinical scores (E) of Il22^−/− mice at indicated periods postinoculation with the indicated CR strains. (F) Kaplan-Meier analysis of the survival rate in Il22^−/− mice inoculated with the indicated CR strains. Data are mean ± s.e.m., with specific n numbers indicated, and representative of at least two independent experiments. ***, P < 0.001 and ****, P < 0.0001, for ΔespF CR versus WT CR in blue and ΔespFΔushA CR versus WT CR in red, respectively.

Infection with ΔespFΔushA strain elicits robust systemic and humoral immunity against CR.

CR infection induces CR-specific antibodies in infected hosts, thus facilitating the clearance of this bacterial pathogen (30, 31). Despite the markedly attenuated virulence, the ΔespFΔushA strain shared similar CR burdens and clearance kinetics as its wild-type equivalent in Il22^−/− mice (Fig. 1C to F), which led us to examine whether ΔespFΔushA CR infection elicited CR-specific antibody response in the survived animals. To this end, Il22^−/− mice were orally inoculated with either live ΔespFΔushA CR or inactivated ΔespFΔushA strain pretreated with formalin (Fig. 2A and B); the serum and fecal samples were collected from these animals after the clearance of CR, i.e., at 26 dpi, to analyze CR-specific antibody responses. As illustrated by enzyme-linked immunoassays (ELISAs), infection with live ΔespFΔushA CR, compared to the inactivated control, in Il22^−/− mice caused augmented levels of antibodies that specifically recognize the whole bacteria of CR (Fig. 2C–I). In addition, live bacterial infection elicited systemic immune responses, in particular the CR-specific antibodies including IgG (as well as the IgG₁, IgG_2a, IgG_2b, and IgG₃ subtypes), IgM, and IgA, which were all significantly elevated in the serum samples collected from Il22^−/− mice infected with live ΔespFΔushA CR, in comparison to those from inactivated control-inoculated animals (Fig. 2C–I). Moreover, the inactivated CR barely triggered fecal IgA response in the infected animals, whereas the CR-specific fecal IgA titers were notably elevated in Il22^−/− mice even after ΔespFΔushA CR clearance (Fig. 2J). Besides the whole bacteria, the soluble fractions from sonicated CR solution were utilized in ELISAs to examine whether the dynamic host–pathogen interactions during live ΔespFΔushA CR infection facilitate the presentation and recognition of native antigens within CR by host immune systems. Indeed, the levels of serum IgG, IgM, and IgA, as well as fecal IgA specific for CR soluble fractions, were substantially elevated in Il22^−/− mice infected with live ΔespFΔushA CR, in contrast to those in the animals challenged with inactivated control (Fig. 3A–D). Consistently, the SDS-PAGE separation followed by immunoblot assays revealed that a wide range of proteins, in particular those with molecular weights ranging from 10–15 kilo-Dalton (kDa) and 25–55 kDa, in the whole cell lysates of wild-type CR were detected by the serum antibodies derived from Il22^−/− mice infected with live ΔespFΔushA CR (Fig. 3E). In striking contrast, the separated CR lysates were scarcely recognized by the serum antibodies collected from inactivated control-infected animals (Fig. 3E). These results demonstrate that oral administration of live ΔespFΔushA CR, but not inactivated strain, elicits robust and systemic immune responses in the survived Il22^−/− mice, generating a collection of antibodies that recognize the antigens on the surface and inside of CR.

FIG 2.

ΔespFΔushA CR infection elicits specific antibody responses in Il22^−/− mice. (A) A schematic of infection experiment in Il22^−/− mice with ΔespFΔushA Citrobacter rodentium (CR) or formalin-inactivated ΔespFΔushA CR. The serum and fecal samples were collected at 26 days postinoculation (dpi). (B) Representative images of ΔespFΔushA CR and formalin-inactivated CR on a MacConkey agar plate after overnight culture. (C–I) CR-specific immunoglobulin G (IgG) (C), IgG₁ (D), IgG_2a (E), IgG_2b (F), IgG₃ (G), IgM (H), and IgA (I) levels in the serum samples collected from Il22^−/− mice infected with ΔespFΔushA CR or formalin-inactivated CR at 26 dpi. (J) CR-specific IgA levels in the feces collected from Il22^−/− mice infected with ΔespFΔushA CR or formalin-inactivated CR at 26 dpi. OD₄₅₀, optical density measured at a wavelength of 450 nm. Data are mean ± s.e.m., with specific n numbers indicated. *, P < 0.05; **, P < 0.01; ***, P < 0.001; and ****, P < 0.0001.

FIG 3.

Live ΔespFΔushA CR infection triggered antibody responses recognize biomolecules inside of CR. (A–C) Citrobacter rodentium (CR) lysate-specific immunoglobulin G (IgG) (A), IgM (B), and IgA (C) levels in the serum samples collected from Il22^−/− mice infected with ΔespFΔushA CR or formalin-inactivated ΔespFΔushA CR at 26 days postinoculation (dpi). (D) CR lysate-specific IgA levels in the feces collected from Il22^−/− mice infected with ΔespFΔushA CR or formalin-inactivated CR at 28 dpi. OD₄₅₀, optical density measured at a wavelength of 450 nm. (E) CR whole cell lysates were separated by SDS-PAGE, followed by immunoblot (IB) using three serum samples collected from Il22^−/− mice infected with the indicated CR strains at 26 dpi, with DnaK as a loading control. Data are mean ± s.e.m., with specific n numbers indicated. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001.

Oral administration of ΔespFΔushA CR effectively protects Il22^−/− mice from lethal infection.

To assess whether ΔespFΔushA CR-elicited antibody responses in Il22^−/− mice are sufficient to confer protective immunity, Il22^−/− mice inoculated primarily with live ΔespFΔushA CR or inactivated control were subjected to a secondary infection with wild-type CR (Fig. 4A). Consistent with our previous observations (Fig. 1F), the gradual colonization and clearance of ΔespFΔushA CR (Fig. 4F) caused transient body weight loss in Il22^−/− mice (Fig. 4G); both live and inactivated ΔespFΔushA CR-infected animals survived. At 28 dpi, they were rechallenged with wild-type CR (Fig. 4A). While CR infection causes measurable inflammation and minimal morphological changes in the colons of immunocompetent animals, CR challenge in immunocompromised Il22^−/− mice results in more severe inflammation and substantial shortening in colon lengths (16). Indeed, secondary infection with wild-type CR triggered discernible morphological damage in the colons of Il22^−/− mice that were primarily inoculated with the inactivated control (Fig. 4B), with substantially shortened colon lengths (Fig. 4C). In contrast, the colons derived from Il22^−/− mice subjected to primary ΔespFΔushA and secondary wild-type CR infections, exhibited almost normal morphology and lengths compared to those from uninfected Il22^−/− animals (Fig. 4B and C). In line with the wild-type CR-caused colonic tissue damage (Fig. 4B and C), markedly elevated CR burdens were detected in the liver and the spleen of Il22^−/− mice primarily infected with inactivated control (Fig. 4D and E), whereas the increased CR burdens in these peripheral organs were abolished in the animals preinfected with live ΔespFΔushA CR (Fig. 4D and E). As expected, the CR burdens gradually increased and peaked at 6 days after secondary wild-type CR infection in Il22^−/− mice preinoculated with inactivated strain (Fig. 4F), and these animals suffered from substantial body weight loss (Fig. 4G) and resulted in 91% lethality (Fig. 4H). In contrast, the maximal burdens were extensively reduced, and wild-type CR was cleared as rapidly as 2 days post-secondary infection in Il22^−/− mice orally administered with live ΔespFΔushA CR (Fig. 4F). Crucially, neither body weight loss nor diarrheal symptoms occurred in the live ΔespFΔushA CR-administered Il22^−/− mice post-secondary challenge with wild-type CR, and all these animals survived for at least 20 days (Fig. 4G and H). These results suggest that the robust and systemic immune responses elicited by live ΔespFΔushA CR are sufficient to protect immunocompromised Il22^−/− mice from an otherwise lethal infection with wild-type CR.

FIG 4.

ΔespFΔushA CR infection completely protects Il22^−/− mice from lethal CR infection. (A) A schematic of infection experiment in Il22^−/− mice infected primarily with ΔespFΔushA Citrobacter rodentium (CR) or formalin-inactivated ΔespFΔushA CR, followed by a secondary oral challenge with wild-type (WT) CR at 28 days postinoculation (dpi). (B) Representative images of the cecum and the colon derived from Il22^−/− mice, primarily infected with ΔespFΔushA CR or formalin-inactivated CR and secondarily infected with WT CR, at 8 days post WT CR infection, or uninfected Il22^−/− animals. (C) Colon lengths of Il22^−/− mice infected as in (B). Each data point represents a single animal (n = 4). (D–E) CR burdens in the liver (D) and the spleen (E) derived from Il22^−/− mice infected with ΔespFΔushA CR or formalin-inactivated CR and secondarily infected with WT CR, at 8 days post WT CR infection. Each data point represents a single animal (n = 4). (F) Colonization kinetics of CR in Il22^−/− mice at indicated periods postinoculations with the indicated CR strains as in A. (G) Body weight loss of Il22^−/− mice at indicated periods postinoculations with the indicated CR strains. (H) Kaplan-Meier analysis of the survival rate in Il22^−/− mice at indicated periods postinoculations with the indicated CR strains. Data are mean ± s.e.m. with specific n numbers indicated. Data in B are representative of three independent experiments and in F–H are summarized from two independent experiments. *, P < 0.05; **, P < 0.01; ****, P < 0.0001.

ΔespFΔushA CR administration provides protection to CR infection in wild-type C57BL/6J mice.

We further assessed the impact of live ΔespFΔushA strain-elicited immune responses on protecting CR infection in immunocompetent animals. Wild-type C57BL/6J mice inoculated primarily with live ΔespFΔushA CR or inactivated control were subjected to a secondary infection with wild-type CR (Fig. 5A). As expected, live ΔespFΔushA CR minimally affected the body weight and survival of the infected wild-type C57BL/6J mice (Fig. 5B and C), despite a gradual colonization and clearance of CR (Fig. 5D). At 28 dpi, these mice were rechallenged with wild-type CR (Fig. 5A). Consistent with previous reports that immunocompetent mouse strains display little to no mortality with mild symptoms to CR infection, secondary wild-type CR infection did not affect the body weight and lethality of wild-type C57BL/6J mice preinoculated with either live or inactivated CR strain (Fig. 5B–C). However, the CR burdens peaked at 10 days and were gradually cleared within the 4th week after secondary wild-type CR infection in the mice preinoculated with inactivated CR, whereas the maximal burdens of wild-type CR were substantially attenuated, followed by a rapid clearance at 2 days post-secondary infection in wild-type C57BL/6J mice orally administered with live ΔespFΔushA CR (Fig. 5D), similar to our findings in the immunocompromised Il22^−/− mice (Fig. 4F). Hence, these results demonstrate that live ΔespFΔushA strain-elicited immune responses dampen the burdens and facilitate the clearance of CR infection, thus offering protection in immunocompetent wild-type C57BL/6J animals as well.

FIG 5.

ΔespFΔushA CR infection protects wild-type C57BL/6J mice from CR infection. (A) A schematic of infection experiment in wild-type (WT) C57BL/6J mice infected primarily with ΔespFΔushA Citrobacter rodentium (CR) or formalin-inactivated ΔespFΔushA CR, followed by a secondary oral challenge with WT CR at 28 days postinoculation (dpi). (B) Body weight changes of WT C57BL/6J mice at indicated periods postinoculations with the indicated CR strains. (C) Kaplan-Meier analysis of the survival rate in WT C57BL/6J mice at indicated periods postinoculations with the indicated CR strains. (D) Colonization kinetics of CR in WT C57BL/6J mice at indicated periods postinoculations with the indicated CR strains. Data are mean ± s.e.m. with specific n numbers indicated and summarized from two independent experiments. ****, P < 0.0001.

DISCUSSION

Nonantibiotic strategies to treat EPEC and EHEC infections have emerged as a crucial option to limit the increase in antibiotic resistant strains (6), or the only option for disease where antibiotic therapies are not recommended (32), in particular where antibiotic treatment of EHEC infection could lead to bacterial membrane damage and increased production and release of the Shiga toxins (33). In the past decades, significant efforts have been made to identify efficacious vaccines against A/E pathogens, particularly EHEC, which include Shiga toxin-based vaccines (34, 35), bacterial ghost-based vaccines (36, 37), protein-based vaccines (38–40), peptide-based vaccines (41, 42), plant-based vaccines (43, 44), DNA-based vaccines (45–48), polysaccharide-based vaccines (49, 50), adjuvant enhanced vaccines (41, 51), and others. These subunit and whole inactivated vaccines offer great safety; however, the potential epitope alternation by inactivation process and the lack of dynamic host–pathogen interaction-conferred native antigen presentation during natural infections may reduce the protective effects against pathogen infections. Notably, the T3SS machinery and the Shiga toxins have long been deemed as promising antigens for the most effective vaccination strategy to prevent A/E pathogen-caused foodborne diseases (27); however, no vaccines against EPEC or EHEC have been licensed so far.

The risk of residual live attenuated vaccines, especially for immunocompromised people, is a common concern for vaccine safety (52). Moreover, clinical trials using wild-type EHEC (O157: H7 strain) cannot be performed on human volunteers, due to the severity of disease (27). While causing self-limiting infections and mild symptoms in immunocompetent hosts, A/E pathogens are known to result in severe symptoms and even mortality under immunocompromised conditions (15, 16). Hence, there are only a few previous studies using live attenuated bacterium-based vaccines for preventing EPEC and EHEC infections (53, 54). Of note, the accumulating knowledge about critical host–pathogen interactions during A/E pathogen infections in immunocompromised hosts suggests the potential feasibility of live attenuated vaccine strategy. In particular, our recent study showed that ΔespF strain completely diminished wild-type CR-caused lethality in the immunocompromised Il22^−/− mice, although deletion of EspF did not affect the colonization, proliferation, and clearance of CR (16). Considering the potential far-reaching impact of CR infection on colon tumorigenesis, ΔespF strain has been further genetically manipulated to delete ushA gene encoding UshA, a newly discovered genotoxin conserved in A/E pathogens (28). Phenocopying its parental ΔespF strain, ΔespFΔushA CR infection fails to cause mortality in the immunocompromised Il22^−/− mice. Indeed, mimicking natural infection, oral administration of live ΔespFΔushA CR, compared to the inactivated control, elicits more robust systemic and humoral immunity with elevated levels of CR-specific antibodies. Moreover, inoculation of inactivated control produces antibodies recognizing several CR antigens; by contrast, live ΔespFΔushA CR-administrated Il22^−/− mice generate antibodies specific for a wide range of antigens on the surface and inside of CR, likely due to the preservation of native antigens and the antigen presentation facilitated by dynamic host–pathogen interactions during live infection. Consequently, oral administration of live ΔespFΔushA CR elicits perfect protective immunity protecting the immunocompromised Il22^−/− mice from lethal infection with wild-type CR, especially the accelerated CR clearance at two days post-secondary infection. As expected, in contrast to subunit vaccines based on only one or few virulence factors, oral infection with ΔespFΔushA CR in Il22^−/− mice promotes an efficient immune response against an array of known and yet unidentified virulence determinants. Thus, our results provide a proof of concept of live attenuated vaccination strategy for preventing A/E pathogen infections in immunocompromised hosts associated with more severe symptoms and lethality.

MATERIALS AND METHODS

Ethics statement.

All animal experiments were performed according to the protocol approved by the Johns Hopkins University’s Animal Care and Use Committee and in direct accordance with the NIH guidelines for housing and care of laboratory animals. Wild-type (WT) and Il22^−/− (IL-22 knockout) mice in C57BL/6J background (16) were maintained in a specific pathogen-free facility and given autoclaved food and water ad libitum.

Reagents and antibodies.

Luria-Bertani (LB) broth, Dulbecco's modified Eagle medium (DMEM), LB plates, and MacConkey plates were purchased from VWR (Radnor, PA). Formalin (F8775) and sulfuric acid (258105) were obtained from Sigma-Aldrich (St. Louis, MO). Antibodies used were DnaK (8E2/2) antibody from Thermo Fisher Scientific (Halethorpe, MD) and horseradish peroxidase (HRP) conjugated antibodies specific for mouse IgG, IgG₁, IgG_2a, IgG_2b, IgG₃, IgM, and IgA from Santa Cruz Biotechnology (Dallas, TX) and SouthernBiotech (Birmingham, AL).

Bacterial strains and growth conditions.

Wild-type (DBS 100 strain), ΔespF, and ΔushA Citrobacter rodentium (CR) strains were described previously (16, 28). The ΔespFΔushA CR strain was generated with the “scar”-free in-frame deletion method based on the suicide vector pRE112 via allelic exchange, as described previously (16, 28). CR strains (listed in Table 1) were validated by genomic PCR using gene-specific primer sets with detailed sequences in Table 2. CR strains were grown from single colonies on LB plates in LB broth at 37°C overnight with shaking. For growth curve measurement, the overnight CR culture was diluted at 1:100 and grown in LB broth or DMEM at 37°C with shaking, and 100 μL of bacterial culture was taken at indicated time periods to read OD₆₀₀ on a 96-well plate using a POLARStar Omega Plate Reader (BMG Labtech, Cary, NC). Formalin-inactivated CR was prepared as previously described (55). Briefly, after centrifugation at 4°C, the pelleted CR culture was washed twice with ice-cold sterile phosphate-buffered saline (PBS) pH 7.4 and resuspended in PBS containing formalin (3%, vol/vol), followed by incubation at 4°C for 16 h.

TABLE 1.

Citrobacter rodentium strains used in this study

Strain or genotype	Deficiency	Source or reference
Wild-type CR, DBS 100	None	16
ΔespF	EspF	16
ΔushA	UshA	28
ΔespFΔushA	EspF, UshA	This study

TABLE 2.

PCR primers used in this study

Primer ID	Sequences
espF_Up_150F	5′-AATCCTCTCGACCGCAAATCATTTC-3′
espF_Gene_217R	5′-ATCTCAGTCTTTACCCCCTATTGC-3′
espF_Dn_58R	5′-ATTATGTGAGGGGATAGTAGTGAGTTATTTTCT-3′
ushA_Up_210F	5′-ACGGCGCAGCGGAGCCGGGAAT-3′
ushA_Gene_350R	5′-CATGATTGCCGACGGCCATTGCGTCATAG-3′
ushA_Dn_280R	5′-CGATCTGGCCGCCATTCTGGACGCA-3′

CR infection in mice.

Eight-week-old wild-type C57BL/6J and Il22^−/− mice were fastened for 6 h before oral inoculation with 200 μL of PBS containing 2 × 10⁸ CFU of indicated CR strains. Mice were monitored for body weight, fecal CR burden, and morbidity for 28 days postinoculation. The disease clinical scores were calculated as the sum of weight loss and diarrhea, as previously described (56).

CR burden analysis.

For fecal CR burden analysis, stool was collected from live animals at various times postinoculation; for CR dissemination analysis, the liver and the spleen were collected from euthanized mice at indicated time periods postinoculation. The stool and tissue were homogenized, diluted in PBS, plated on MacConkey agar plates, and incubated overnight at 37°C. CR CFU were enumerated the following day and normalized to the stool or tissue weight (16).

Enzyme-linked immunoassays (ELISAs).

Antibody titers against CR were determined by ELISA, as previously described (30). Briefly, 96-well high-binding polystyrene plates (VWR) were coated per well with 100 μL of wild-type CR solution (2 × 10⁹ CFU/mL in water) or CR soluble supernatants generated from CR solution (2 × 10⁹ CFU/mL in water) post sonication and centrifugation 38,000 × g at 4°C for 30 min. Plates were air-dried at room temperature overnight, fixed in 0.15% glutaraldehyde in 0.15 M phosphate, pH 7.0 for 5 min, and treated with 0.15 M glycine in 15 mM phosphate buffer, pH 7.0 at room temperature for 10 min, followed by blocking with PBS containing 5% nonfat dry milk and 0.5% Tween 20 at 4°C overnight. After washing, serially diluted mouse serum or fecal solutions were incubated for 2 h at room temperature, followed by incubation with HRP-conjugated antibodies as indicated. Captured HRP was visualized with tetramethyl benzidine-H₂O₂ in acetate buffer, and reactions were stopped with 2 N sulfuric acid and read at optical density 450 nm.

Immunoblot.

Immunoblot assay was conducted as previously described (56). In brief, wild-type CR cells were harvested and lysed in 0.4 mL of lysis buffer (50 mM Tris-HCl [pH 8.0], 150 mM NaCl, 10% SDS) for 30 min. The lysates were centrifuged at 10,000 × g at 4°C for 10 min, followed by a separation by SDS-PAGE under reduced and denaturing conditions. The resolved protein bands were transferred onto nitrocellulose membranes, immunoblotted with the indicated mouse serum samples and appropriate secondary antibodies, probed by the Super Signaling system (Thermo Fisher Scientific) according to the manufacturer's instructions, and imaged using a FluorChem E System (Protein Simple, Santa Clara, CA).

Colon tissue analysis.

Colon tissue harvest was performed as previously described (56). Briefly, after euthanizing mice, the cecum and the colon were removed under aseptic conditions and subjected to imaging and length measurement using a FluorChem E System (Protein Simple).

Statistical analysis.

Statistical analysis was performed using GraphPad Prism version 9.0.1 (GraphPad Software, San Diego, CA). Standard errors of means (s.e.m.) were plotted in graphs. Statistical significance was determined by Student's t test, one-way ANOVA with Bonferroni post hoc test, or log-rank (Mantel-Cox) test (survival curves). Significant differences were considered: ns, nonsignificant difference; * at P < 0.05; ** at P < 0.01; *** at P < 0.001; and **** at P < 0.0001.