A wild-type Korean epidemic virulent strain A17-DW-005 (SG005) of S. enterica ser. Gallinarum with resistance toward ciprofloxacin (CIP), streptomycin (STR), gentamicin (GEN), nalidixic acid (NAL), sulfisoxazole (FIS), and colistin (COL) was originally isolated from the liver of a 10-day-old broiler, which came from a broiler farm in Jeonbuk province in 2017. This strain was classified as type 1 by pulsed-field gel electrophoresis analysis (PFGE). Since type 1 has the highest proportion of all types, this strain was judged to be an epidemic strain. The suicide plasmid pRE112 and its host Escherichia coli (E. coli) χ7,213 were kept in our laboratory. pBluescript II SK (+) and pET-30 b (+) stored in our laboratory were used as intermediate vectors in this study. E. coli and S. enterica ser. Gallinarum were grown at 37°C in Luria–Bertani (LB) broth or on LB agar. When required, antibiotics were added to culture media at the following concentrations: ampicillin (Amp) at 100 μg/ml, kanamycin (Km) at 50 μg/ml, chloramphenicol at 25 μg/ml, and DL-α,ε-diaminopimelic acid (DAP) at 50 μg/ml. LB agar containing 5% sucrose was used for sacB gene-based counterselection in allelic exchange experiments. A rough attenuated S. enterica ser. Gallinarum live vaccine SG9R (Intervet International, Boxmeer, The Netherlands) was used as control vaccine in this study.

Ross broiler chickens were obtained from the YangJi Company (Cheonan, Choongnam, South Korea). Serum samples collected from all chickens in the present study were tested for Salmonella specific antibodies using a commercial ELISA kit (Biochek, cat# CK117, Crabethstraatt, Netherland). Moreover, the antibodies level of Avian influenza and Newcastle disease were detected by hemagglutination-inhibition (HI) test virus. The test results were all negative. 10% of the chickens were randomly selected and euthanized. The chickens were dissected to check whether there were any lesions in the internal organs. Liver and fecal samples were aseptically collected for routine Salmonella and avian pathogenic Escherichia coli isolation and culture, and the follow-up Salmonella infection experiment was performed after the complete negative was confirmed. The chickens were used after adaptation for a week. To ensure the best environmental conditions, the conditions of the isolator (temperature, humidity, and ventilation) were continuously monitored. The chickens were taken care of and handled by a well-trained staff.

All experimental and animal management procedures were undertaken in accordance with the requirements of the Animal Care and Ethics Committee of Jeonbuk National University. The animal facility at Jeonbuk National University is fully accredited by the National Association of Laboratory Animal Care (approval number: JBNU 2020-0162).

Construction of deletion strains was performed by the allelic exchange method using the suicide vector pRE112 as described previously (Supplementary Figure 1) (17). The 1,000-bp upstream fragment of the waaJ gene was amplified from the genomic DNA of S. enterica ser. Gallinarum using a pair of primers (W1 and W2) by PCR (Supplementary Table 1). The PCR product was cloned into the XbaI and BamHI sites of the pBluescript II SK(+) vector, resulting in pSK-waaJ-up. The 1,000-bp downstream fragment of the waaJ gene was then PCR-amplified using a pair of primers (W3 and W4) and cloned into the BamHI and KpnI sites of pSK-waaJ-up to obtain pSKΔwaaJ. The 2,000-bp fragment, which included the upstream and downstream fragments of the waaJ gene from the XbaI- and KpnI-digested plasmid pSKΔwaaJ, was ligated into plasmid pRE112 to yield the suicide plasmid pREΔwaaJ. The transfer of recombinant suicide plasmids into SG005 was accomplished by conjugation using E. coli χ7213 (pREΔwaaJ) as the plasmid donor. Strains containing single-crossover plasmid insertions (SG005waaJ::pREΔwaaJ) were isolated on plates containing chloramphenicol. The first homologous recombination was confirmed with the primers W5 and W6. A loss of the suicide vector after the second recombination between homologous regions (i.e., allelic exchange) was selected for by using the sacB-based sucrose sensitivity counterselection system. The second homologous recombination was confirmed with the primers W6 and W7. Primers W8 and W9 were used to confirm whether the full waaJ gene was removed from the genome. Primers W10 and W11 were used to confirm whether the suicide plasmid was removed from the genome (Supplementary Figure 2). This deletion strain was designated as SG005ΔwaaJ. Using the same method, deletion of spiC was introduced into SG005, and this deletion strain was designated as SG005ΔspiC. Notably, in the construction of the suicide plasmid pREΔspiC, pET-30 b (+) was used as an intermediate vector. Similarly, the deletion of spiC was introduced into SG005ΔwaaJ, and the double deletion strain was designated as SG005ΔwaaJΔspiC.

The complemented strains were constructed previously described with some modifications (18–20). The gene waaJ was PCR-amplified from the wild type strain SG005 using the primers pBR322-waaJ-F and pBR322-waaJ-R (Supplementary Table 1), which were designed based on the coding sequence of waaJ gene. After amplification, the DNA fragment was digested by BamHI and SalI and ligated to the plasmid pBR322. The resulting vector pBR322-waaJ was introduced into SG005ΔwaaJ by electroporation to produce the complemented strain SG005ΔwaaJ (pBR322-waaJ). Electroporation on the electroporator (Bio-Rad Laboratories, Hercules, CA, USA) was performed using optimized parameters (2.5 kV, 200 Ω, 25 μF, 5 ms). Similarly, the complemented strains SG005ΔspiC (pBR322-spiC) and SG005ΔwaaJΔspiC (pBR322-waaJ-spiC) were constructed. It should be pointed out that when constructing SG005ΔwaaJΔspiC (pBR322-waaJ-spiC), the waaJ gene and spiC gene were inserted into the plasmid pBR322 in tandem.

For the in vitro growth analysis of SG005, SG005ΔwaaJ, SG005ΔspiC, and SG005ΔwaaJΔspiC, a single colony of each strain was inoculated into 5 ml of LB broth and cultured at 37°C with shaking at 200 rpm for approximately 15 h. Subsequently, the optical density (OD₅₉₀) of each strain was measured by a Genesys 10S UV-Vis spectrophotometer (Thermo Fisher Scientific, Waltham, USA), and cultures were diluted in 20 ml of LB broth; then, the OD₅₉₀ was determined to achieve an initial concentration (OD₅₉₀ = 0.03) at the starting time point (0 h). The cultures were incubated at 37°C with shaking at 200 rpm, and the OD₅₉₀ was determined at 1, 2, 3, 4, 5, 6, 7, and 8 h. Each strain was tested in triplicate in three independent experiments. The biochemical properties of the strains were tested according to the API20E handbook (Biomerieux, SA, France).

O9 antigen is one of the antigens produced on S. enterica ser. Gallinarum. S. enterica ser. Gallinarum with smooth phenotype have O9 antigen and therefore can be agglutinated by O9 antibody. Briefly, colonies from LB agar plates were mixed with 20 μl of phosphate-buffered saline (PBS) and 20 μl of the O9 factor rabbit antiserum (SSI^® SALMONELLA ANTISERA, Denmark) on a glass slide. Mixtures were rotated for 1 min and observed for agglutination. The acriflavine agglutination test was performed to detect the rough phenotype characteristic of the deletion strains (21). Briefly, colonies of a 24-h culture prepared from the deletion strains on LB agar plates were mixed with 20 μl of 0.2% acriflavine on a glass slide. Mixtures were rotated for 1 min and observed for agglutination.

Auto-aggregation refers to the phenomenon that bacteria aggregate into clusters by themselves during the cultivation process. The stronger is the hydrophobicity of the cell surface, the higher is the auto-aggregation ability of the cell. The auto-aggregation assay was performed based on a previously described method with some modifications (22). Briefly, SG005, SG005ΔwaaJ, SG005ΔspiC, and SG005ΔwaaJΔspiC were statically grown in 5 ml of LB broth at 37°C for 24 h in 15-ml conical tubes. The upper 0.5 ml was carefully removed to measure its OD₅₉₀ (recorded as OD_{590 prevortex}). Then, the remaining culture in the tube was mixed by vortexing to resuspend the aggregated cells, and 0.5 ml of the suspension was removed, and its OD₅₉₀ was measured (recorded as OD₅₉₀ postvortex). The “percent aggregation” was calculated using the formula: 100% × (OD_{590 postvortex} – OD_{590 prevortex})/OD_{590 postvortex}.

To determine the stability of the deletion strains, the liquid culture of each strain was continuously passaged 60 times every 12 h. Every 10 generations, DNA was extracted for PCR identification.

The sensitivity of SG005, SG005ΔwaaJ, SG005ΔspiC, and SG005ΔwaaJΔspiC to ultraviolet (UV) light and various disinfection agents was evaluated as described with some modifications (23, 24). Bacteria were grown to OD₅₉₀ = 1.0 at 37°C with shaking at 200 rpm and were diluted in PBS to 10⁹ colony forming units (CFU)/ml. One milliliter of suspension was then transferred to a sterile Petri dish and exposed to UV radiation (Sankyo denki UV G30TB: 30 W, distance: 45 cm) for 1 min. For testing oxidative stress and alkali and acid tolerance, each bacterial suspension and the stressor were mixed in equal volume and incubated at room temperature (RT, 25°C): 100-μM hydrogen peroxide (H₂O₂) for 5 min, 12.5-mM sodium hydroxide (NaOH) for 5 min, and 50-mM citric acid for 20 min. Finally, the number of viable bacteria was determined by plating serial dilutions onto LB agar plates. The experiment was performed three times.

To determine the 50% lethal dose (LD₅₀), SG005, SG005ΔwaaJ, SG005ΔspiC, and SG005ΔwaaJΔspiC were grown for 12 h at 37°C in LB broth. One milliliter of broth from each grown bacterium was inoculated into 100 ml of LB broth. After the inoculated strains were grown to OD₅₉₀ = 1.0, the bacterial cells were recovered by centrifugation for 10 min at 2,400 g at RT. The harvested cells were suspended in PBS to a concentration of 10¹¹-10¹² CFU/ml. The suspended strains were serially diluted by 10-fold. The 7-day-old chickens were intramuscularly (IM) injected or orally inoculated with each dose of these strains (10¹-10¹¹ CFU). Six chickens were used for each dose. Feed and water were not provided to the tested chickens for 4 h before and until 1 h after the treatment. The treated chickens were observed twice a day for 2 weeks after administration. Clinical scores were determined and recorded using a system as we have previously described (25). Briefly, disease severity was scored as follows: 0, normal; 1, depression and ruffled feathers; 2, depression, ruffled feathers, and respiratory distress; 3, abovementioned clinical signs plus anorexia, emaciation, and green-yellowish diarrhea; and 4, death. When the birds showed a clinical score of 3 (humane endpoint), they were euthanized immediately by cervical dislocation performed by trained veterinarians. LD₅₀ was calculated by the method of Reed and Muench (26).

Bacterial persistence and clearance were analyzed by counting the number of the bacteria per gram of liver or spleen. Seven-day-old chickens were randomly divided into four groups (n = 12 for each group): SG005ΔwaaJ group, SG005ΔspiC group, SG005ΔwaaJΔspiC group, and PBS control group. The first three groups were injected intramuscularly with 1 × 10⁸ CFU of SG005ΔwaaJ, SG005ΔspiC, and SG005ΔwaaJΔspiC, respectively. The last group was inoculated with an equal volume of PBS in the same way. Three chickens from each group were euthanized at 2, 6, 10, and 14 days post-inoculation (dpi), and liver and spleen samples were aseptically removed. The samples were weighed and suspended in 1 ml of PBS and then individually homogenized. The homogenates were diluted serially and subsequently plated on MacConkey agar (Difco Laboratories, Detroit, MI, USA) at 37°C. After overnight cultivation, the bacterial number was counted.

Sixty 7-day-old chickens were divided into five groups (n = 12 for each group) to evaluate the protective efficacy of the deletion strains. The group 1 (PBS negative control) and group 2 (PBS positive control) were intramuscularly inoculated with 500 μl of PBS (pH 7.2). The group 3 and group 4 were intramuscularly inoculated with 10⁸ CFU of SG9R and SG005ΔwaaJΔspiC, respectively. The group 5 was orally inoculated with 10⁸ CFU of SG005ΔwaaJΔspiC. After 4 weeks, except for the group 1, chickens in other groups were challenged intramuscularly with 10⁸ CFU of virulent SG005. Deaths and clinical signs were recorded daily for 2 weeks. At the end of this period, all surviving chickens were euthanized. Clinical scores were determined and recorded following the method previously described (25).

A laboratory-made indirect enzyme-linked immunosorbent assay (ELISA) using heat-killed whole S. enterica ser. Gallinarum bacteria as the coating antigen was applied to quantify IgG in the serum. Briefly, bacterial antigen (10⁶ CFU/well) was coated in a 96-well plate and incubated overnight at 4°C. After blocking, the sera samples collected on days 14, 21, and 28 were diluted (1:400) with 2% bovine serum albumin (BSA) and added to the wells. The wells were incubated with a 1:8,000 dilution of affinity-purified, peroxidase-labeled goat anti-chicken IgG (H+L) (KPL Inc., MD, USA) for 1 h. OD₄₅₀ was measured with a microplate reader VICTORTM X4 (PerkinElmer Inc., Waltham, MA, USA) after the reaction was stopped with 4.5 N H₂SO₄. The OD₄₅₀ average value of negative serum samples plus the standard deviation (SD) of three times was used as the cut-off value (27).

The serum plate agglutination (SPA) test was used to assess the DIVA capability of SG005ΔwaaJ, SG005ΔspiC, and SG005ΔwaaJΔspiC. Seven-day-old chickens were intramuscularly inoculated with each deletion strain. The sera were prepared on 14, 21, and 28 dpi by collecting blood from wing veins. The blood was allowed to coagulate for 1 h at RT, centrifuged for 10 min at 600 g, and then, the clean phase of serum was collected. Serum samples (30 μl) were mixed with an equal volume of S. enterica ser. Pullorum/Gallinarum standard antigen (Green Cross Veterinary Products, Yongin, South Korea), and the agglutination reaction was observed.

Statistical analysis was performed using SPSS version 21.0 (SPSS Inc., Chicago, IL, USA). The data were analyzed by one-way analysis of variance (ANOVA). Differences were considered statistically significant at: ^*P < 0.05, ^**P < 0.01, ^***P < 0.001.

The deletion of the waaJ gene in SG005 was confirmed by PCR using the primers W6 and W7 with the expected amplicon sizes of 2,490 and 1,476 bp for the wild type and deletion type of the waaJ gene, respectively. The deletion of the spiC gene in SG005 was confirmed by PCR using the primers S6 and S7 with the expected amplicon sizes of 2,003 and 1,619 bp for the wild type and deletion type of the spiC gene, respectively (Supplementary Figure 3). In the three deletion strains, the corresponding target gene and suicide plasmid could not be detected by PCR (Supplementary Figure 4). The sequencing results also showed that the deletion strains SG005ΔwaaJ, SG005ΔspiC, and SG005ΔwaaJΔspiC were constructed successfully (Supplementary Text 1).

Growth curve analysis revealed no significant differences between the wild type and each deletion strain when cultured in LB broth at 37°C (Figure 1A). Results of biochemical tests including 2-nitrophenyl-β D-galactopyranoside, L-arginine, L-lysine, L-ornithine, trisodium citrate, sodium thiosulfate, urea, L-tryptophane, L-tryptophane, sodium pyruvate, gelatin, D-glucose, D-mannitol, inositol, D-sorbitol, L-rhamnose, D-sucrose, D-melibiose, amygdalin, and L-arabinose were the same between wild type and each deletion strain, suggesting that deletions in these two genes did not alter the biochemical characteristics of SG005 (Supplementary Table 2).

SG005 and SG005ΔspiC were agglutinated with O9 factor rabbit antiserum but not with acriflavine, indicating that they had a smooth phenotype. However, SG005ΔwaaJ and SG005ΔwaaJΔspiC were not agglutinated with O9 factor rabbit antiserum but were agglutinated with acriflavine, indicating that they had a rough phenotype (Figure 1B). The complemented strains SG005ΔwaaJ (pBR322-waaJ), SG005ΔspiC (pBR322-spiC) and SG005ΔwaaJΔspiC (pBR322-waaJ-spiC) were all agglutinated with O9 factor rabbit antiserum but not with acriflavine. Therefore, complementary analysis proved that the deletion of the waaJ gene was the direct cause of the bacteria changing from smooth to rough.

Auto-aggregation was tested in LB broth (Figure 1C), and the aggregation percent was calculated using OD₅₉₀ measurements from these cultures. Both SG005 and SG005ΔspiC showed ~5% aggregation, whereas SG005ΔwaaJ and SG005ΔwaaJΔspiC demonstrated 84 and 79% aggregation, respectively (Figure 1D).

The deletion Strains were serially passaged 60 times in LB medium, and the absence of waaJ and spiC genes was confirmed by PCR, indicating that each deletion strain had good genetic stability (Figure 2).

After UV irradiation, the survival rates for SG005, SG005ΔwaaJ, SG005ΔspiC, and SG005ΔwaaJΔspiC were 48, 37, 45, and 37%, respectively (Figure 3A). A significant difference (P < 0.001) was seen between ΔwaaJ strains (SG005ΔwaaJ and SG005ΔwaaJΔspiC) and non-ΔwaaJ strains (SG005 and SG005ΔspiC), indicating that the deletion of the waaJ gene increased the sensitivity of the strain to UV. In terms of oxidative stress, SG005ΔwaaJ and SG005ΔwaaJΔspiC with survival rates of 52 and 50% showed reduced resistance compared to SG005 and SG005ΔspiC with survival rates of 59 and 60% (P < 0.001) in H₂O₂ (Figure 3B). For alkali endurance, the survival rates for SG005, SG005ΔwaaJ, SG005ΔspiC, and SG005ΔwaaJΔspiC were 68, 51, 67, and 49%, respectively. A significant difference was noted between ΔwaaJ strains and non-ΔwaaJ strains (P < 0.001; Figure 3C). For acid endurance, the survival rates for SG005, SG005ΔwaaJ, SG005ΔspiC, and SG005ΔwaaJΔspiC were 34, 31, 35, and 31%, respectively. Statistical analysis revealed a difference between ΔwaaJ strains and non-ΔwaaJ strains (P < 0.05; Figure 3D). However, deletion of spiC gene had no effect on bacterial resistance to these environmental stresses.

To confirm the safety of the deletion strains, the virulence of wild and deletion strains was evaluated in 7-day-old chickens. The IM LD₅₀ of SG005ΔwaaJ and SG005ΔspiC was about 10⁸-fold higher than that of the wild strain. The IM LD₅₀ of SG005ΔwaaJΔspiC was about 10¹⁰-fold higher than that of the wild strain. The oral LD₅₀ of the three deletion strains SG005ΔwaaJ, SG005ΔspiC, and SG005ΔwaaJΔspiC was about 10⁵-fold higher than that of the wild strain SG005 (Table 1).

The results of bacteria persistence and clearance in organs were shown in Figure 4A. SG005ΔwaaJΔspiC colonized and persisted in the liver and spleen for at least 10 days. Within these 10 days, the number of SG005ΔwaaJΔspiC gradually decreased. At 14 dpi, SG005ΔwaaJΔspiC could not be isolated from the liver and spleen. However, SG005ΔwaaJ and SG005ΔspiC could still be isolated, indicating that the double deletion strain could be cleared more quickly than the single deletion strains.

Chickens that were intramuscularly inoculated with 10⁸ CFU of SG9R and SG005ΔwaaJΔspiC showed 83 and 92% protection against S. enterica ser. Gallinarum challenge, respectively. Chickens that were orally inoculated with 10⁸ CFU of SG005ΔwaaJΔspiC showed 83% protection. All chickens died in the PBS-positive control group (Figure 4B). Depression, anorexia, and diarrhea were observed in the PBS-positive control group, whereas no clinical signs were observed in the SG005ΔwaaJΔspiC group.

Statistical analysis showed a significant difference in the humoral immune response in chickens immunized with the deletion strains compared with the PBS control group. Serum IgG levels at each post-immunization time point were significantly higher in the immunized groups than in the PBS control group. The IgG level of the SG005ΔwaaJΔspiC group immunized by intramuscular injection oral administration was not significantly different from that of the SG9R group at 3 and 4 weeks after inoculation (P > 0.05; Figure 4C).

Antibodies to S. enterica ser. Gallinarum were detected by the SPA test using the SG005 antigen. Sera collected from SG005ΔwaaJ- or SG005ΔwaaJΔspiC-immunized chickens at different time intervals were identified as SPA-negative (Figure 4D). However, sera collected from SG005- or SG005ΔspiC-immunized chickens were identified as SPA-positive. Overall, SG005ΔwaaJ and SG005ΔwaaJΔspiC showed evident distinguishable capability.