The bacterial strains and plasmids used in this study are described in Table 1. The asd deleted S. Typhimurium mutant were cultivated in Luria-Bertani medium or on Luria-Bertani agar plates and grown at 37°C with DAP (Sigma-Aldrich, St. Louis, MO, USA). The bacterial strains harboring the ghost gene cassette were grown on NB agar supplemented with 0.2% L-arabinose.

The 846 and 450 bp DNA fragments encoding FedF and FedA proteins, respectively, were amplified in JOL505 using polymerase chain reaction (PCR) with primer pairs (FedA_F:5′-ccgcgaattccagcaaggggatgttaaat-3′ and FedA_R: 5′-ccgcaagcttga tgattacttgtaagta-3′, FedF_F:5′-ccgcgaattcgcgtctactctacaagta-3′ and FedF_R:5′-ccgcaagcttttactgtatctcgaaaacaa-3′). The PCR products digested by EcoRI and HindIII were cloned into the overexpression plasmid pET28a, producing pET28a-fedF and pET28a-fedA, respectively. Each plasmid was transformed into BL21 (DE3) pLysS (Jawale and Lee, 2014). Briefly, the N-terminally 6x-His-tagged FedF or FedA expressed in each E. coli BL21 strain was purified following the protocol previously described (Jawale and Lee, 2014). The concentration of purified protein was quantified using Bradford reagent (Bio-Rad Laboratories, Hercules, CA, USA). To insert fedF and fedA individually into the pJHL184, the fimbrial gene fragments originating from pET28a-fedF and pET28a-fedA were individually subcloned into EcoRI/HindIII-digested pJHL184 to generate pJHL184-fedF and pJHL184-fedA, respectively. The resultant plasmids were initially introduced into Δasd E. coli χ6212 (JOL232) to reduce instability of the plasmids in the balanced lethal system (Galán et al., 1990), and then the plasmids were electroporated into the ΔasdΔlonΔcpxR S. Typhimurium strain JOL912 which was prepared by deletion of the asd gene in ΔcpxR Δlon S. Typhimurium strain JOL911 using allelic exchange method (Hur and Lee, 2010). The transformants were designated JOL1460 for the strain carrying FedF and JOL1464 for the strain carrying FedA. JOL912 harboring only the ghost plasmid pJHL184, JOL1400 was constructed following the process and used as the empty vector control.

The JOL1460 and JOL1464 strains were inoculated into 200ml of nutrient broth (NB) under the lysis gene E repressing condition containing 0.2% L-arabinose and were incubated at 28°C with slow agitation. When the cultures reached logarithmic phase, the cells were harvested, and then washed three times with NB to eliminate arabinose. The cells were resuspended in 100 ml of NB and incubated at 42°C for 48 h with 200 rpm agitation to induce E gene-mediated lysis. The lysed cells were collected by centrifugation at 13,000 rpm for 20 min after the lysis process and washed twice with autoclaved phosphate-buffered saline (PBS) (pH 7.4). The pellets were stored at -70°C until further use. The morphological alteration of the lysed JOL1460 and JOL1464 was visualized by using scanning electron microscope (SEM) as previously described (Jawale et al., 2014).

Western blot analysis was performed to validate the expression of the individual recombinant heterologous antigens, FedF and FedA fused to downstream of the six-histidine tag of the pJHL184 from the inactivated Salmonella carrier. Following a published protocol (Jawale and Lee, 2014), the lysed cells of JOL1460 and JOL1464 were prepared. The lysed samples were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) on 15% gels. Subsequently, the proteins separated on the gel were electrotransferred laterally onto membranes of polyvinylidene fluoride (Millipore, Billerica, MA, USA) and blocked with 3% bovine serum albumin. The protein expressed in the S. Typhimurium ghosts was detected using a primary anti-His antibody (1:5,000) and secondary an HRP-labeled anti-mouse IgG (1:8,000). The immunoreactive bands were developed by the West-OneTM Western Blot Detection System (iNTRON, KOR).

All experimental and animal management procedures described in this study were approved (CBNU2015-00085) by The Chonbuk National University Animal Ethics Committee in accordance with the guidelines of the Korean Council on Animal Care. The thirty 5-week-old female BALB/c mice which have been widely used to assess the immunogenicity of bacterial vaccine (Lauvau et al., 2001), were randomly divided into two groups. The mice in group A were injected intramuscularly with 1.0 × 10⁸ of mixture formula of the lysed JOL1460 and JOL1464 cells in 100 μl of sterile PBS at week 0 and week 2, respectively, following the previously published protocol (Hur and Lee, 2015). Non-immunized mice in group B were intramuscularly inoculated with 100 μl of sterile PBS. At week 0, 2, 4, and 6 post-immunization (PI), blood samples were collected from the mice via the infraorbital vein to evaluate immunoglobulin (Ig) G, G1, and G2a. To assess secretory IgA (sIgA) produced in the mice, vaginal washes were obtained from the mice using sterile PBS. All samples were kept at -70°C until used.

To evaluate humoral immunity elicited by the immunization, the presence of total IgG, IgG1, IgG2a, and sIgA antibodies against FedF and FedA fimbrial antigens, respectively, in sera and vaginal washes was determined by indirect enzyme-linked immunosorbent assay (ELISA) as previously reported (Hur and Lee, 2015). Purified FedF and FedA proteins (0.5 μg/ml) were used as coating antigens. The final concentration of serum IgG and sIgA were analyzed by a standard curve of purified mouse immunoglobulins (Southern Biotechnology, Birmingham, AL, USA). The ratio of IgG2a to IgG1 isotypes was determined by dividing values of optical density (OD) at 470 nm for IgG2a by the OD values for IgG1.

An alteration in T cell subsets of the immunized mice was assessed using fluorescence-activated cell sorting (FACS) (Jawale and Lee, 2014). Spleen cells were isolated from the immunized and non-immunized mice at day 7 PI and 1 × 10⁶ of the cells were seeded into a 96-well cell culture plate. The splenic cells were stained with the surface markers containing anti-mouse CD3a-PE and anti-mouse CD4-perCP-vio700 (Miltenyi Biotec, Bergisch Gladbach, Germany). After the stained cells were washed twice in FACS buffer (Miltenyi Biotec, Bergisch Gladbach, Germany), the stained CD3+ and CD3+CD4+ splenic cells were sorted with an MACSQuant^® Analyzer (Miltenyi Biotec, Bergisch Gladbach, Germany). The altered FACS profiles of CD3+ and CD3+CD4+ splenic T cell subsets in the immunized mice were analyzed in comparison to those from non-immunized mice.

The magnitude of splenocyte proliferation following in vitro antigen stimulation were assessed by incorporation of MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) which is only converted to blue formazan dye by actively proliferating cells (Denizot and Lang, 1986). Splenocytes were isolated from the immunized and non-immunized groups at week 2 PI. 1 × 10⁶ of the splenic cells were stimulated with 300 ng/ml of the purified FedF or FedA protein in triplicate, and then placed in a humidified 37°C, 5% CO₂ incubator for 48 h. After incubation for 48 h, 1 mg/ml of MTT was added to the stimulated cells, and the mixed culture was incubated for 4 h at 37°C. The MTT solution was converted into blue formazan dye only when the stimulated cells are actively proliferating. After the incubation, the resulting formazan precipitation was dissolved in 100 μl dimethyl sulfoxide (DMSO). The absorbance spectrum at 490 nm of MTT formazan solutions was measured by a spectrophotometer.

To evaluate production of immunomodulatory cytokines elicited by the vaccine candidates, expression of IL-4 and IFN-γ mRNA secreted by T-helper-2 (Th-2) (Swain et al., 1990) and Th-1 clones (Scott, 1991), respectively, were measured in the splenocytes pulsed with each antigen. Splenocytes were aseptically isolated from the immunized mice and the non-immunized mice at week 2 PI. 1 × 10⁶ of splenocytes were stimulated with 500 ng/ml of either FedF or FedA antigens in a 96-well plate and were incubated with RPMI 1640 medium (GIBCO, cat. no.11875093) containing 5% FCS (GIBCO, cat. no. 10099141) at 37°C in a 5% CO₂ incubator for 48 h. Total RNA from the stimulated cell cultures was extracted using GeneAll^® Hybrid-RTM (GeneAll Biotechnology, Seoul, Korea) and was converted into cDNA with the ReverTra Ace^® qPCR RT Kit (FSQ-101, TOYOBO, Japan). The level of IL-4 and IFN-γ mRNA expression was determined by real-time reverse transcription-polymerase chain reaction (RT-PCR) with SYBR^® Green Realtime PCR Master Mix (QPK-201, TOYOBO, Japan) following the manufacturer’s instructions. The primer sets were derived from the conserved regions of mouse β-actin (as an internal standard), IL-4 and IFN-γ (Overbergh et al., 1999). The relative mRNA quantification of each cytokine was analyzed using the threshold method, using ΔC_T values calculated based on the internal standard. The relative fold change of the cytokines expressed in the stimulated cells was presented as 2^-(ΔΔCT), compared to those of the non-immunized group (Schmittgen and Livak, 2008).

BALB/C mice did not show any symptoms of ED when they are orally challenged with a wild-type F18ab+STEC strain, JOL654, even though natural transmission of ED occurs via an oral route. Instead, we found that intraperitoneal injection of the challenge strain caused similar symptoms of ED in the mice. The median lethal dose of JOL654 in mice (LD50, 2 × 10⁷ CFU) was evaluated using the Reed–Muench method (Reed and Muench, 1938). At 4 weeks after the last immunization, both the immunized and non-immunized groups were intraperitoneally injected with 100 μl of JOL654, which contained a titer of 2 × 10⁷ CFU. Mortality, clinical sign, and body weight were monitored in the challenged mice during the week after inoculation.

Non-parametric Mann–Whitney test was applied to evaluate the difference in immune responses elicited by the immunized and non-immunized groups. All data are expressed as means ± standard deviation. Differences were considered statistically significant when P-values were ≤ 0.05.

The fedF and fedA genes were each introduced into the heterologous antigen delivery site in the plasmid pJHL184. The resultant plasmids, pJHL184-fedF and pJHL184-fedA were individually transformed into the attenuated S. Typhimurium strain JOL912 to construct JOL1460 and JOL1464, respectively. The convergent promoter components induced expression of the E gene in pJHL184 under a temperature increase of up to 42°C and the removal of 0.2% L-arabinose. After 48 h at 42°C, the magnitude of lysis was determined by counting the number of viable cells plated from the culture of JOL1460 and JOL1464 grown at 42°C. No viable colony was detected on the LB plate supplemented with arabinose after overnight incubation at 27°C. Further, the transmembrane tunnel generated on the surface of lysed JOL1460 and JOL1464 was identified by the scanning electron microscopy (SEM) techniques. The cell surface appeared to be partially collapsed by releasing cytoplasmic contents through the pores on the cell membrane (Figures 1B,C) compared to the morphology of normal cells (Figure 1A). While the E gene-mediated lysis was activated, secretion of the fimbrial proteins into the cytoplasmic space was initiated. Subsequently, the proteins fused with the ompA signal sequence were secreted and stayed in the lysed Salmonella. Each fimbrial protein expressed in JOL1460and JOL1464 was validated by immunoblot analysis. The protein fused to OmpA expressed were detected at ∼40 kDa for FedF, and ∼24 kDa for FedA in the pellet of JOL1460 and JOL1464, respectively (Figure 2).

The titers of serum IgG and sIgA antibodies against FedF and FedA antigens were quantified in serum and vaginal wash samples, respectively. FedF-specific IgG and sIgA was markedly elevated (P < 0.05), whereas the IgG level specific to FedA was only significantly increased in mice at week 4 and 6 PI (P < 0.05) (Figure 3). At week 4 PI, the level of IgG against both FedF and FedA increased approximately 11 and 12.6 times, respectively, compared to levels in the non-immunized mice (Figure 3). The immunized mice generated 12.8 times more sIgA antibodies against FedF compared to those of the control at week 2 PI. The titers of serum IgG isotypes, IgG2a and IgG1, indicating activation of T helper (Th)-1 and Th-2 cells, respectively, were measured in the mice. The titer of IgG isotypes was presented as the OD values at 405 nm. The level of both IgG isotypes specific to FedF was nearly 10 times higher than those specific to FedA from week 4 to week 6 PI (Figures 4A,B). Additionally, the ratio of IgG2a to IgG1 was evaluated to elucidate relative contribution of Th-2 and Th-1 cells, respectively, to the immune responses. At week 4 and 6 PI, the ratio of IgG2a to IgG1 in the mice ranged from 0.9 to 1.76, which indicated that balanced Th-1 and Th-2-related immune responses were induced in the immunized mice (Figure 4C).

The MTT cell proliferation assay, the splenocytes stimulated in vitro with either FedA or FedF showed enhanced formazan absorbance values at 490 nm following the addition of MTT solution into the cell cultures (Figure 5). The absorbance value in FedF- and FedA-stimulated cells was augmented 1.53 ± 0.01 and 2.63 ± 0.08 times, respectively, in the comparison with those of the non-immunized mice at week 2 PI. The changes in T cell subsets was evaluated by detecting CD3⁺ and CD3⁺CD4⁺ surface marker expression in splenic T cells isolated from the immunized mice at week 1 PI using flow cytometry analysis. The FACS data revealed overall elevation of CD3⁺ total T cell and CD3⁺CD4⁺ T cell subsets in the immunized mice (Figure 6C). The significant increase of CD4⁺ T cells in the CD3⁺ gated splenocytes (2.9 ± 1.58 percent) were observed in the immunized mice (Figure 6B) compared to those of non-immunized mice (Figure 6A) (P < 0.05).

To evaluate IL-4 and IFN-γ cytokine production modulating the immune responses, we observed fold-changes in mRNA gene expression levels in vitro in restimulated splenic T cells collected from the immunized mice using qRT-PCR. A statistically significant augmentation (P < 0.05) of IL-4 cytokine was observed in splenocytes restimulated in vitro with FedF antigen (Figure 7). In the splenocytes restimulated with FedA antigen, the mRNA level of IL-4 cytokine was 3.54 ± 1.09 fold higher than that in the non-immunized group. The level of IFN-γ cytokine was enhanced in both splenocytes stimulated in vitro with either FedF or FedA compared to those of the control group.

To evaluate the protective efficacy conferred by the immunization, all mice in group A (n = 7) and B (n = 7) were intraperitoneally injected with a lethal dose of the virulent F18⁺STEC strain JOL654 at week 6 PI. The survival rates and weight loss in the mice were monitored in immunized and non-immunized animals for nine days after challenge. In both immunized and non-immunized mice, weight loss was observed until day 2 post challenge. While the weight of immunized animals was fully recovered by day 4 post-challenge, the weight of the non-immunized animals declined until day 9 post-challenge (Figure 8B). Particularly, body weight of the non-immunized mice was significantly dropped compared to the immunized mice at day 5, 7, and 8 post-challenge (P < 0.01). A clear difference in survival rate was observed between the immunized group and the control group. All immunized mice survived the entire observation period, whereas 28% of the non-immunized mice in the control group died within 24 h after challenge (Figure 8A). The clinical signs in the mice such as diarrhea, hunched posture, and hair elections were also observed in non-immunized mice.