All manipulations of E. coli strains were conducted in approved and certified Biosafety Level 2 facilities at the University of Texas Medical Branch (UTMB), and experiments were performed in accordance with standard operating practices. The animal studies were carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. The protocol (IACUC #0709042B) was approved by the Institution for Animal Care and Use Committee at UTMB.

All bacterial strains used in this study were stored in 50% glycerol at −80°C. Liquid cultures were generated by inoculating Luria-Bertani (LB) broth, with or without 50 μg/mL streptomycin. Liquid cultures were grown overnight at 37°C with agitation. The prototypical E. coli O157:H7 strain 86-24 was used for animal challenge studies. E. coli O157:H7 wildtype strain EDL933 was utilized for gene amplifications and bacterial adhesion assays. E. coli strain DH5α (Life Technologies) was used for propagation of strains bearing pVAX1 plasmid and was routinely grown in LB broth and agar containing kanamycin (Sigma, 50 μg/mL) for plasmid selection.

We used NetMHC-II (Nielsen and Lund, 2009), NetCTL (Larsen et al., 2007) and ABCPred (Saha and Raghava, 2006) to predict the MHC class II, MHC class I and linear B-cell binding epitopes, respectively, in EHEC O157:H7 specific proteins (further details in García-Angulo et al., 2014; Kalita et al., 2014). Program outputs reported the binding affinity (IC₅₀) of individual epitopes to various HLA alleles. Based on these IC₅₀ values, all possible 9-mer epitopes from each protein were predicted as weak (IC₅₀ > 500 nM) or strong binding (IC₅₀ > 50 nM). We selected only strong binding peptides having a predicted IC₅₀ value < 50 nM (used as threshold). To assess the potential to induce a Th2-skewed protective response, we focused on NetMHCII output consisting of six HLA-DP, six HLA-DQ, and 14 HLA-DR alleles. We derived two parameters from the NetMHCII output: a) average MHC Affinity (Avg-MA) for all epitopes in a given protein sequence and b) mature epitope density (MED_Th2) score using the formula below, where epitope length is core 9-mer (Santos et al., 2013). (1)MEDTh2=PredictionsChances=number of predicted epitopes X (50-Avg-MA)protein sequence length-epitope length* + 1 This MED_Th2 score reflects aggregate T-cell (Th2) epitope content. Higher scores indicate a better prediction for the protective nature of the protein. Also, the number of alleles bound by each protein was evaluated in order to target a larger population coverage. A heatmap of Avg-MA against each HLA-allele for all protein sequences was generated (Supplemental Figure 1). A clustered image map (CIM) of a normalized matrix was created that correlates Avg-MA of each allele to different proteins. For each protein, mean and standard deviation were calculated from their Avg-MA for all alleles. Z-score transformation was calculated for each of the alleles by subtracting each Avg-MA value by the row mean and dividing by the row standard deviation (Kalita et al., 2013). Hierarchical clustering of HP proteins was performed using an average-linkage clustering algorithm based on their MED_Th2 scores.

Vaccine candidates were amplified from E. coli EDL933 genomic DNA. Forward (Fw) and reverse (Rv) primer sequences contained HindIII and XhoI restriction sites, respectively (Supplemental Table 1). The 5′ end of Fw primers were designed with a Kozak consensus sequence (ACCATGG) to enhance transcription. Genes were amplified with Phusion® High Fidelity Polymerase (New England Biolabs) and ligated into the eukaryotic expression vector pVAX1 (Invitrogen, Life Technologies). Plasmids containing desired candidates were verified by directional sequencing and transformed into competent E. coli DH5α for propagation. For immunization studies, plasmids were purified using the Endotoxin-free Giga Kit (Qiagen) according to manufacturer's instructions. DNA samples were quantified using a Epoch Microplate Spectrophotometer (BioTek) and stored at −20°C.

Six to eight-week-old female BALB/c mice were obtained from Charles River Laboratories and housed in a specific pathogen-free barrier under biosafety level 2 conditions and allowed to acclimate for 5 days prior to vaccination. Mice were divided into 5 groups (n = 10 each), including pVAX (vector), lomW (pVAX-10), escJ (pVAX-41), escC (pVAX-56), and combination (pComb). Mice were anesthetized using isoflurane inhalation and administered a prime and two boosts (days 0, 14, and 28) intranasal (i.n.) immunization of ~60 μg DNA in Tris-EDTA. Prime vaccinations were administered along with Cholera Toxin (CT) as adjuvant (1 μg/uL). In the case of the combination vaccine, ~20 μg of each plasmid were mixed and administered as a single vaccine. Fecal and sera samples were collected prior to vaccination for determination of baseline antibody titers. Fecal samples were collected following final boost to monitor mucosal antibody titers. Briefly, fecal pellets were weighed and diluted to 1 g/mL in PBS. After homogenization by vortexing, fecal samples were then centrifuged at 4000 rpm for 10 min. Supernatants were stored at −20°C prior to IgA measurement. Sera samples were collected 2 weeks after prime and second boost vaccination to monitor changes in antibody levels. Sera was collected via retro-orbital bleeding and incubated at room temperature for 30 min to allow clotting. Sera was separated from whole blood by centrifugation at 10,000 rpm for 10 min. Supernatants were collected and stored at −80°C prior to enzyme-linked immunosorbent assay (ELISA).

Two weeks after the second boost, all mice were challenged with a dose of 5 × 10⁹ CFU of streptomycin resistant E. coli O157:H7 strain 86-24 via gavage (400 μL). Food was restricted 12 h before infection but was administered ad libitum throughout the remainder of the study. Two hours prior to challenge, mice were injected intraperitoneally with cimetidine (50 mg/kg, Sigma) to reduce stomach acidity. Fecal samples were collected daily for 7 days to assess bacterial shedding. Fecal pellets were homogenized in phosphate-buffered saline (PBS), serially diluted, and plated on MacConkey agar plates containing streptomycin (25 μg/mL) and incubated at 37°C. To enumerate bacterial colonization in gastrointestinal tract, mice were euthanized, and ceca and large intestines were removed. Organs were homogenized in 1 mL PBS, serially diluted and plated on MacConkey agar containing streptomycin.

Total IgG and IgA responses were determined using Ready-set-Go! ELISA kits (EBioscience) and were performed according to manufacturer's instructions. To determine immunoglobulin levels, polystyrene 96-well high-binding ELISA plates (Nunc, Denmark) were coated overnight with capture IgG or IgA antibody at 4°C. The plates were washed 2x with PBS containing 0.05% Tween 20 (PBS-T) prior to blocking in 2X Assay Buffer. For IgG, the serum samples were diluted (1:1000 and 1:10,000) in 1X Assay Buffer. Similarly, for IgA levels, samples were diluted (1:2 and 1:4) in 1X Assay Buffer. Following incubation, horseradish peroxidase (HRP)-conjugated goat anti-mouse IgG, or goat anti-mouse IgA was diluted in 1X Assay Buffer (1:250) and added to ELISA plates to determine IgG and IgA concentration, respectively. Plates were incubated with agitation for 3 h at room temperature, followed by washing. A total of 100 μL of tetramethylbenzidine (TMB) was added to each well and incubated at room temperature for 15 min. The reaction was stopped using 100 uL of 2N H₂SO₄ and plate was read at 450 nm (Biotech Microplate Spectrophotometer).

Caco-2 cells (ATCC® HTB-37™) were maintained at 37°C with 5% CO₂ in complete HTB-37 medium. Complete HTB-37 media consisted of Eagle's Minimum Essential Medium (EMEM, GIBCO) supplemented with 2 mM glutamine, 1 mM sodium pyruvate, 1X non-essential amino acids, penicillin-streptomycin (100 U/ml, 100 μg/ml), and 10% fetal bovine serum. For adhesion assays, 12-well plates were seeded with 10⁵ cells per well and incubated as described above to achieve 80% confluence. Approximately 1 h prior to infection, the monolayer was washed twice with 1 ml PBS prior to addition of 1 ml medium containing no supplements. Fresh bacterial culture of E. coli O157:H7 strain EDL933 was grown in LB overnight at 37°C prior to infection. Bacterial culture was diluted in LB (1:100) and incubated at 37°C, shacking, until culture reached an OD₆₀₀ of 1.0. Culture was pelleted at 5000 × g for 5 min, resuspended in PBS (Ca²⁺ and Mg²⁺ free), and plated for input bacterial load. Remaining bacteria was incubated with immune or naïve sera (5 and 10%) for 45 min at 37°C with agitation. At this time, media was removed and replaced with 1 ml fresh media containing 10⁷ bacterial cells (multiplicity of infection [MOI], 100). Inoculated monolayers were incubated for 3 h at 37°C with 5% CO₂. After incubation, cells were washed three times with PBS prior to addition of 200 μl of 0.1% Triton X-100 in PBS. Wells were incubated at 37°C until cell monolayer detached from the plate. Monolayers were homogenized by pipetting, then samples were serially diluted and plated onto LB agar. The percentage of bacteria recovered was calculated as the number of CFU/ml recovered divided by the input CFU/ml to account for slight variances in input between groups.

Statistical significance between control and vaccinated groups was assessed using GraphPad software. One-way analysis of variance (ANOVA) and Student t test were used to analyze the data for colonization and antibody response, respectively. Adhesion assay experiments were repeated in triplicate. Bacteria recovered were normalized to the mean percentage of the bacteria inoculated and the groups were compared using one-way ANOVA followed by Kruskal-Wallis posthoc test. P < 0.05 were considered significant.

Figure 1A provides a schematic representation of Th2-oriented epitope prediction and selection of seventeen, potentially protective, high-priority (HP) vaccine candidates in EHEC O157:H7 (García-Angulo et al., 2014). The distribution of high-binding epitopes across numerous HLA alleles is depicted in a heatmap (Supplemental Figure 1). This heatmap highlights allele coverage, with proteins exhibiting increased allele coverage, and thus more likely to induce an effective immune response in a heterogenous population. While the majority of high priority (HP) proteins have high MED_Th2 scores (above 10, Figure 1B), there is selective binding across the array of HLA alleles. For example, while lomW showed a high predictive binding affinity toward DQ5 and DRB6 alleles, the escJ epitopes may bind more strongly to DP4 as compared to other gene products. The data in panel 1B was used to derive a dendrogram (Figure 2) to further highlight protein clustering based on MED_Th2 score. Based on our previous study that tested a pool of three randomly selected HP candidates, immunization with truncated escC resulted in the most significant reduction in bacterial colonization. However, that study utilized only the second half of the gene as a vaccine target instead of the entire gene, because we had problems to clone it. In our present study, we were successfully able to clone the full-length escC gene. The goal of this study was to examine the protective capacity of whole escC gene together with two closely related genes (lomW and escJ). In summary, candidates were selected based on MED_Th2 scores, allele coverage, physiochemical features, and predicted function.

BALB/c mice were immunized with the DNA vaccine candidates as desbribed in Material and Methods. Two weeks after the last immunization, animals were challenged with a dose of 5 × 10⁹ CFU of streptomycin-resistant E. coli O157:H7 strain 86-24 via gavage. Seven days post-challenge, large intestines and ceca were collected to enumerate bacterial colonization. The bacterial load in the gastrointestinal tract (Figure 3) indicates bacterial reduction in mice immunized with the three tested candidate groups lomW (pVAX-10), escJ (pVAX-41), escC (pVAX-56) compared to pVAX-only immunized group. When a combination of all three (pComb) candidates were administered, the reduction in colonization was minimum. Immunization with lomW (pVAX-10) resulted in the greatest reduction in bacterial colonization in the large intestine (p = 0.0423) (Figure 3A). A similar trend is observed in cecum colonization, despite no statistical significance (Figure 3B).

Fecal samples collected 2 weeks post-immunization were used to measure sIgA production. Mice immunized with lomW produced the highest levels of total sIgA when compared to unimmunized mice (baseline), mice immunized with pVAX1 alone, or any of the other immunization groups, though no statistical significance was observed (Figure 4A). Similarly, escC was also shown to induce increased sIgA production. Furthermore, sera collected from immunized mice 2 weeks after the last immunization was used to measure total IgG antibodies. Unlike sIgA, there appears to be no observable differences in immunoglobulin production between the immunization groups (Figure 4B).

To further characterize the antibodies produced after vaccination, we analyzed the capacity of immune sera to prevent the adherence in vitro of E. coli O157:H7 to human intestinal epithelial cells (Caco-2). Wild-type E. coli O157:H7 strain EDL933 was incubated with pooled sera (n = 3) from immunized mice (5 and 10% concentration) at an MOI of 100 prior to infection of Caco-2 cells. We observed the most significant reduction in bacterial adherence from lomW (p = 0.0466) and escC (p = 0.0029) immunized mice sera at 10%, and from lomW (p = 0.0466) and pComb (p = 0.0143) at 5% concentration (Figures 5A,B). We noticed a decrease in the percent bacterial adherence from all groups at 5% compared to 10% sera. Also, we noticed a reduction in bacterial adherence by sera from escJ and pComb compared to control groups at both concentrations, but this reduction did not reach statistical significance (Figures 5A,B). The ability of sera to inhibit bacterial adherence in vitro to intestinal epithelial cells suggest the possibility of some specificity of the sera while recognizing surface-exposed proteins present on EHEC O157:H7 wild-type strain EDL933.

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.