This study was carried out in accordance with the recommendations of Arizona State University Institutional Animal Care and Use Committee. The protocol (#1168R) was approved by the Arizona State University Institutional Animal Care and Use Committee. Six-week-old female BALB/c mice (Charles River Laboratories, Wilmington, MA, USA) and 4-week-old female CBA/J mice (Jackson Laboratories, Bar Harbor, ME, USA) were obtained for infection experiments. Mice were acclimated for 7 days before experiments began. During the experiments, animals were monitored twice daily by our team, animal caretakers, and further inspected by a veterinarian.

Strains and plasmids used are listed in Table 1. Strains were stored as stock cultures at −80°C in peptone-glycerol medium. Unless otherwise specified, strains were grown in lysogeny broth (LB) containing 0.1% glucose. Salmonella enterica serovar Typhimurium attenuated strain χ9558, derived from the virulent S. Typhimurium strain UK-1 (χ3761), using strategies that enhance safety and immunogenicity (21, 22), was used to deliver ECP. Urosepsis strain CFT073 (23) was used in animal challenge experiments. ExPEC strains CFT073, JJ1886, UTI89, and RS218, non-pathogenic E. coli strains HS-4, J198, and Nissle 1917, laboratory E. coli MG1655, and S. Typhi χ3444, S. Typhimurium χ3761, and S. Paratyphi χ8387 strains were used for in vitro assays.

Recombinant attenuated Salmonella vaccine strains and challenge strain CFT073 were grown statically overnight in LB. The next day, the culture was inoculated 1:100 into fresh LB. RASV strains were grown with aeration at 37°C to OD₆₀₀ of ~0.85. CFT073 was grown with aeration at 37°C to OD₆₀₀ of ~0.85 (sepsis challenge) or statically overnight (intraurethral challenge). Strains were harvested by centrifugation at 24°C and resuspended in PBS.

Fragment ecpABCD, a 5622 bp portion of the ecp operon comprising ecpRABCDE, was PCR-amplified from genomic DNA of E. coli strain E24377A using KlenTaq LA DNA Polymerase (DNA Polymerase Technology, Inc., Saint Louis, MO, USA) and primer sets: P1: ecpA (BsrGI)-F: 5′ TAGTAATGTACATGAAAAAAAAGGTTCTGGCAATAG 3′ and P2: ecpD (HindIII)-R: 5′ CCCAAGCTTGGGTTAGTTAATGTTACGCCACCGTCGCC 3′ (Figure S1 in Supplementary Material). The amplified fragment included the 5′ region of the first gene, ecpA, from its start codon through the stop codon of the last gene, ecpD. To introduce the enzyme recognition site for BsrGI into vector pYA3337, we amplified the vector plasmid using primer sets: P3: pYA3337 (HindIII) F 5′ CCACAAGCTTGGCTGTTTTGGCGGATGAGA 3′ and P4: pYA3337 (BsrGI) R 5′ CCTATGTACATGTTTCCTGTGTGAAATTG 3′. The amplified fragment included at the 5′ region, the enzyme restriction site HindIII, and at the 3′ region, the enzyme restriction site BsrGI. PCR products were cloned into the pCR-XL-TOPO vector according to the manufacturer’s instructions (Invitrogen).

PCR products of asd-positive vector pYA3337 (BsrG1-positive) and ecpABCD fragment cloned into the pCR-XL-TOPO were cut using HindIII and BsrGI enzymes, respectively. DNA bands of the digested ecpABCD_{(HindIII, BsrGI)} and pYA3337_{(HindIII, BsrGI)} were purified from an agarose gel and ligated together using T4 DNA ligase (New England Biolabs, Ipswich, MA, USA) to generate plasmid pYA4428. Plasmids were verified by PCR, on agarose gel, restriction digestion analysis with HindIII and BsrGI, and sequencing.

The recombinant plasmid was first introduced into E. coli strain χ6212 commonly used for synthesis of foreign proteins (26). The purified plasmid obtained from χ6212 was then electroporated into competent cells of asd-negative Salmonella vaccine strain χ9558 to obtain the balanced-lethal construct. Selection for transformants was achieved by growth on LB agar plates and verified by PCR amplification.

Bacterial ECP synthesis was evaluated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and Western blot with rabbit anti-EcpA and -EcpD antibodies. The E. coli strain E24377A was used as a positive control and χ9558 containing ecpABC, ecpABCD, or ecpRABCDE were tested to determine which portion of the operon maintains ECP synthesis. χ9558 containing ecpABC or ecpRABCDE were constructed using the aforementioned methods. Surface display of ECP was visualized by transmission electron microscopy (TEM) using rabbit anti-EcpA and -EcpD antibodies and goat anti-rabbit IgG with 10 nm colloidal gold (MP Biomedicals, Santa Ana, CA, USA) at a concentration of 1:250 (9).

Mice were orally (behind the incisors) administered 20 µl using a pipette containing either PBS (unvaccinated), 10⁹ CFU of χ9558(pYA3337) that carries the asd-plasmid with no ecp genes, or 10⁹ CFU χ9558(pYA4428) that carries the asd-plasmid with ecpABCD (Figure S2 in Supplementary Material). Before immunization, mice were deprived of food and water for 4 h and resupplied 30 min after vaccination. Serum was obtained at days 20 and 41 (BALB/c mice) and days 20 and 30 (CBA/J mice) post-immunization from blood collected from the submandibular vein. Vaginal wash samples were obtained at days 28 and 41 (BALB/c mice) and day 28 (CBA/J mice) by repeated flushing of the vaginal tract and aspiration of 50 µl of PBS. Serum IgG and vaginal IgA and IgG responses against ECP proteins and Salmonella LPS were determined by ELISA as described previously (14). Briefly, E. coli antigens were PCR amplified and cloned into pET-101/D-TOPO vectors (Invitrogen) and expressed in E. coli strain BL21 as His-tagged proteins. Proteins were purified using ProBond Ni-NTA resin columns (Invitrogen). Endotoxin removal spin columns (Pierce Biotechnology, Rockford, IL, USA) were used to remove any remaining LPS from purified proteins. Commercial S. Typhimurium LPS (Sigma) was used as a source of LPS. Plates were coated at a concentration of 2.0 µg/ml of antigen and incubated overnight at 4°C. The remaining steps were performed at room temperature. Plates were washed with PBS containing 0.05% Tween-20 and blocked for 1 h with SEA BLOCK (Thermo Scientific). Serum or vaginal wash samples were added at 1:50 or 1:10, respectively. Samples were diluted twofold down the plate, incubated for 1 h, and washed. Goat anti-mouse IgG for serum and day 41 vaginal wash samples from BALB/c mice (1:5,000; Southern Biotech, Birmingham, AL, USA) or goat anti-mouse IgA for all vaginal wash samples (1:5,000; Southern Biotech) was added and plates were incubated for 1 h. Plates were washed, and streptavidin, alkaline phosphatase conjugate (1:2,000; Southern Biotech) was added and incubated for 1 h. After incubation and washing, p-nitrophenyl phosphate (Thermo Scientific) was added according to the manufacturer’s protocol. The reaction was stopped with 2 N NaOH and plates were read at 405 nm. The endpoint titer was set as the reciprocal of the highest dilution that gave an OD₄₀₅ twice that of the unvaccinated control.

Antibodies elicited against ECP were further evaluated by Western blot. Serum collected on day 41 was pooled in equal amounts from 10 mice per group. From unvaccinated, χ9558(pYA3337) and χ9558(pYA4428) immunized BALB/c mice pooled serum was used as a substitute for rabbit anti-EcpA to probe against purified EcpA by Western blot as described above.

Bacterial inhibition was tested using bacterial strains (Table 1) in pooled (n = 8/group, in equal volumes) serum or vaginal wash samples obtained on day 41 from BALB/c mice. Bacterial colonies from a fresh LB agar plate were suspended in M9 minimal media until OD₆₀₀ reached 0.1. The suspension was diluted in M9 media (1 × 10² CFU), mixed with an equal volume of pooled serum or vaginal wash samples, and incubated at 37°C for 6 h. After incubation, the mixture was serially diluted and plated on MacConkey agar to determine viable counts. Samples were tested in duplicate in two independent experiments.

Mouse models of human sepsis and UTI were used to evaluate the protective ability of RASV immunization. For the mouse sepsis model, BALB/c mice were intraperitoneally challenged with 100 µl of PBS containing 10⁸ CFU of CFT073 on day 41 postvaccination. Previously, 100% lethality was observed in a mouse sepsis model using 10⁸ CFU of CFT073 by 7 days postchallenge (34). In order to quantitatively determine bacterial loads in internal organs, previous studies have used early endpoints of 24–48 h postchallenge with CFT073 (35, 36). Therefore, mice were euthanized before lethality at 24 h postchallenge, and blood, liver, and spleen were collected, serially diluted, and plated on MacConkey agar for enumeration of E. coli.

For the mouse UTI model, CBA/J mice were intraperitoneally anesthetized with a cocktail of ketamine (100 mg/kg), xylazine (10 mg/kg), and acepromazine (2.5 mg/kg) and the bladder was emptied by gentle pressure on the abdomen. The median infection dose for transurethral inoculation of CFT073 was determined previously as 10⁶ CFU/mouse (35). To ensure infection, mice were inoculated with 50 µl of PBS containing 10⁸ CFU of CFT073 via transurethral catheterization using a sterile polyethylene catheter (Intramedic, Becton Dickinson, Sparks, MD, USA) on day 28 postvaccination. At 48 h postchallenge, mice were euthanized and bladder, kidney, liver, and spleen samples were collected, serially diluted, and plated on MacConkey agar for enumeration of E. coli.

An ANOVA followed by Tukey’s test for multiple comparisons was used to compare between groups for ELISAs, in vitro sera and vaginal wash assays, and the sepsis and UTI models. Fisher’s exact test (two-tailed) was used to compare treatments for the proportion of tissues positive for E. coli in the UTI mouse model. Analyses were carried out in GraphPad Prism 6.0. P values <0.05 were considered significant.

As screened on the SDS-PAGE gel stained with Coomassie blue and Western blot (Figure S3 in Supplementary Material), ECP production was detected in χ9558(pYA4428) but not in χ9558(pYA3337). EcpA and EcpD were displayed on the surface of χ9558(pYA4428) as shown by TEM (Figure S4 in Supplementary Material).

To assess immune responses to vaccination, serum and vaginal wash samples from 8 to 10 individual mice per treatment were evaluated by ELISA for anti-Salmonella LPS and anti-E. coli (EcpA and EcpD) antibodies. For BALB/c mice on day 20, only one mouse vaccinated with χ9558(pYA4428) elicited anti-EcpA and EcpD IgG antibodies, and no IgG antibodies were observed for LPS (Figure 1). On day 28, no IgA antibodies were detected in vaginal wash samples or if elicited were below the limit of detection in the ELISA. On day 41, some mice vaccinated with χ9558(pYA3337) or χ9558(pYA4428) elicited IgG and IgA antibodies to all three antigens, except that no IgG antibodies were detected against EcpA from vaginal wash samples of χ9558(pYA3337) immunized mice. Although some RASV immunized mice had elevated antibody titers compared with unvaccinated mice, no significant differences were observed between vaccinated and unvaccinated mice or between vaccination groups.

For CBA/J mice on day 20, serum and vaginal wash samples from ten individual mice per group were tested by ELISA. Only one χ9558(pYA4428) vaccinated mouse but not χ9558(pYA3337), elicited anti-EcpA IgG antibodies, one χ9558(pYA3337) and two χ9558(pYA4428) immunized mice elicited anti-EcpD IgG antibodies, and no anti-LPS IgG antibodies were observed (Figure 2). Similar to unvaccinated mice, anti-LPS, anti-EcpA, and anti-EcpD IgA antibodies were not detected in vaginal washes of vaccinated mice on day 28. On day 30, one χ9558(pYA3337) and two χ9558(pYA4428) immunized mice elicited anti-LPS IgG antibodies, two χ9558(pYA3337) and one χ9558(pYA4428) immunized mice elicited anti-EcpA IgG antibodies, and three χ9558(pYA3337) and two χ9558(pYA4428) immunized mice elicited anti-EcpD IgG antibodies.

To determine whether antibodies against ECP were also detected by Western blot, pooled serum collected from BALB/c mice (n = 10/group) was used to probe purified EcpA. Western blot analysis showed no reaction to purified EcpA when probed with serum extracted from unvaccinated or χ9558(pYA3337) immunized mice (Figure S5 in Supplementary Material). A positive reaction was observed for serum from χ9558(pYA4428) immunized mice. Thus, we confirmed our ELISA results for χ9558(pYA4428) but could not confirm our results for χ9558(pYA3337) by Western blot.

Pooled mouse serum and vaginal wash samples from 10 mice by group were mixed with bacterial strains and incubated for 6 h to determine whether antibodies or other antimicrobial products elicited from RASV immunization influenced bacterial levels. These assays were used to test inhibitory activity of serum and vaginal wash samples against multiple pathogenic bacterial strains including ExPEC and Salmonella strains not tested in vivo and non-pathogenic E. coli strains that represent organisms inherent in the gastrointestinal tract of humans. A limited amount of sera was collected during the study, and therefore, heat-inactivation of sera was not used as a control. Generally, sera from mice vaccinated with χ9558(pYA4428) had the strongest inhibition against strains (Figure 3). The urosepsis strain JJ1886 was significantly decreased in serum samples of both χ9558(pYA3337) and χ9558(pYA4428) vaccinated mice compared with unvaccinated mice (Figure 3A). A significant decrease was also observed in serum of χ9558(pYA4428) compared with χ9558(pYA3337) immunized mice. The neonatal meningitis-causing E. coli strain RS218 was significantly decreased in serum from χ9558(pYA4428) immunized mice but not χ9558(pYA3337) compared with unvaccinated mice. A significant decrease in levels of RS218 was detected in serum of χ9558(pYA4428) compared with that of χ9558(pYA3337) immunized mice. No significant differences were observed for urosepsis strain CFT073 or laboratory E. coli strain MG1655 between treatments.

Cervicovaginal lavage samples from healthy women have been shown to inhibit growth of E. coli ex vivo (37). To determine if RASV immunization influences bacterial inhibition, various ExPEC, Salmonella, and non-pathogenic E. coli strains were tested for growth in vaginal wash samples (Figure 3B). Two of three ExPEC strains (CFT073 and JJ1886) were significantly reduced in vaginal washes from χ9558(pYA4428) immunized mice but not χ9558(pYA3337) compared to unvaccinated mice. One of three Salmonella strains (S. Typhi χ3444), two of three commensal non-pathogenic E. coli strains (HS-4 and J198), and laboratory E. coli strain MG1655 had significantly decreased levels in vaginal washes from χ9558(pYA4428) vaccinated mice but not in that of χ9558(pYA3337) compared to unvaccinated mice. No significant differences were detected between χ9558(pYA3337) vaccinated and unvaccinated mice or between any of the treatments for ExPEC strain UTI89, S. Typhimurium χ3761, S. Paratyphi χ8387, and E. coli Nissle 1917.

After vaccination, all mice survived and no clinical signs of disease due to Salmonella infection (diarrhea, weight loss, etc.) were observed. In the mouse model of sepsis, BALB/c mice were intraperitoneally challenged with 10⁸ CFU of CFT073 to assess the efficacy of vaccine treatment against systemic infection. BALB/c mice were selected based on past sepsis studies using this genetic background to evaluate ExPEC vaccines (36, 38). The only significant difference relating to bacterial loads in the mouse sepsis model was in the spleen of mice vaccinated with χ9558(pYA3337) (Figure 4).

In the UTI mouse model, CBA/J mice were intraurethrally challenged with 10⁸ CFU of CFT073 to determine the impact of vaccination on bacterial loads in urinary system organs (bladder and kidney) and other internal organs (liver and spleen) (Figure 5). CBA/J mice were selected based on an established UTI protocol (39) to assess the efficacy of vaccine treatment against UTI, and to assess whether a different genetic background resulted in a response similar to RASV immunization of BALB/c mice. In general, mice vaccinated with RASV strains had numerically lower bacterial loads in organs. Significantly lower bacterial loads than unvaccinated mice were observed in the bladder for both χ9558(pYA3337) and χ9558(pYA4428). Also, χ9558(pYA4428) but not χ9558(pYA3337) vaccinated mice had significantly fewer number of E. coli-positive bladder samples than unvaccinated mice. In the kidney, liver, and spleen, bacterial loads for both χ9558(pYA3337) and χ9558(pYA4428) were not significantly different than unvaccinated mice. However, for the liver, χ9558(pYA4428) vaccinated mice had significantly fewer number of E. coli-positive samples than χ9558(pYA3337), but was not significantly different from unvaccinated mice.