Live-attenuated Salmonella enterica serotype Choleraesuis vaccine with regulated delayed fur mutation confer protection against Streptococcus suis in mice

Background Recombinant Salmonella enterica serotype Choleraesuis (S. Choleraesuis) vaccine vector could be used to deliver heterologous antigens to prevent and control pig diseases. We have previously shown that a live-attenuated S. Choleraesuis vaccine candidate strain rSC0011 (ΔPcrp527::TT araC PBADcrp Δpmi-2426 ΔrelA199::araC PBADlacI TT ΔasdA33, Δ, deletion, TT, terminator) delivering SaoA, a conserved surface protein in most of S. suis serotypes, provided excellent protection against S. suis challenge, but occasionally lead to morbidity (enteritidis) in vaccinated mice (approximately 1 in every 10 mice). Thus, alternated attenuation method was sought to reduce the reactogenicity of strain rSC0011. Herein, we described another recombinant attenuated S. Choleraesuis vector, rSC0012 (ΔPfur88:: TT araC PBADfur Δpmi-2426 ΔrelA199:: araC PBADlacI TT ΔasdA33) with regulated delayed fur mutation to avoid inducing disease symptoms while exhibiting a high degree of immunogenicity. Results The strain rSC0012 strain with the ΔPfur88::TT araC PBADfur mutation induced less production of inflammatory cytokines than strain rSC0011 with the ΔPcrp527::TT araC PBADcrp mutation in mice. When delivering the same pS-SaoA plasmid, the intraperitoneal LD50 of rSC0012 was 18.2 times higher than that of rSC0011 in 3-week-old BALB/C mice. rSC0012 with either pS-SaoA or pYA3493 was cleared from spleen and liver tissues 7 days earlier than rSC0011 with same vectors after oral inoculation. The strain rSC0012 synthesizing SaoA induced high titers of anti-SaoA antibodies in both systemic (IgG in serum) and mucosal (IgA in vaginal washes) sites, as well as increased level of IL-4, the facilitator of Th2-type T cell immune response in mice. The recombinant vaccine rSC0012(pS-SaoA) conferred high percentage of protection against S. suis or S. Choleraesuis challenge in BALB/C mice. Conclusions The live-attenuated Salmonella enterica serotype Choleraesuis vaccine rSC0012(pS-SaoA) with regulated delayed fur mutation provides a foundation for the development of a safe and effective vaccine against S. Choleraesuis and S. suis.


Background
Streptococcus suis is a pandemic pathogen responsible for a wide range of invasive diseases such as pneumonia, meningitis and bacteraemia in both humans and pigs [1,2]. S. suis type 2 (SS2) is the most frequently and virulent isolated from both humans and pigs among all serotypes reported to date [1,3]. The surface-anchored protein (Sao) is a highly conserved membrane-anchored protein and proved to be a immunogenic vaccine candidate [4]. However, Sao formulated with Emulsigen-Plus® provides only partial protection to mice against SS2 infection [3]. In our previous study, a recombinant attenuated Salmonella enterica serotype Choleraesuis vaccine strain rSC0016 carrying saoA gene, provided full protection to mice against SS2 challenge [5]. From the above, an effective delivery system such as live Salmonella enterica serotype Choleraesuis play a crucial role to the effectiveness of Sao.
The use of intracellular Salmonella enterica as a vehicle to deliver heterologous protective antigens against pathogens is an attractive strategy. Curtiss et al. developed the RDAS (Regulated Delayed Attenuated Strategies), which enable live salmonella vaccine effectively colonize lymphoid tissues during the invasion stage because of its wild-type aggressiveness and then be full attenuated by silencing the virulence factor, while stimulate both strong cellular and humoral immunity in the immunized mice [6]. Several ways were used to implement this strategy (RDAS). One way is the reverse synthesis of lipopolysaccharide O-antigen by pmi mutation [7]. Another way is to replace the upstream regulatory and promoter sequences of virulence genes with a tightly regulated araC P BAD activator-promoter [8]. This strategy has been successfully used for S. Typhimurium and S. Typhi [6][7][8]. With this strategy, we construct a regulated delayed S. Choleraesuis vaccine strain rSC0011 with ΔP crp527 ::TT araC P BAD crp and pmi mutations [9]. rSC0011 delivering S. suis antigens were effective to induce protective immunity against SS2 in mice, but it occasionally caused enteritidis.
We sought to improve our S. Choleraesuis candidate vector vaccine by using alternative mutation or introducing new mutation to decreasing its potential to induce enteritidis and enhance immunogenicity. Fur is an important regulatory protein in Salmonella. In the presence of iron, Fur acts as a repressor of iron-controlled genes and mounts an adaptive acid tolerance response [8]. Synthesis of Fur in a vaccine strain during growth confers acid tolerance and maintains iron homeostasis. A decrease of Fur synthesis in Salmonella leads to acid sensitivity and iron acquisition [9]. Curtiss et al. reported that a S. Typhimurium strain with an arabinose regulated delayed fur mutation is highly immunogenic [6]. In these consideration, an arabinose regulated delayed fur mutation (ΔP fur88 :: TT araC P BAD fur) was introduced into a S. Choleraesuis vaccine strain with multiple preexist mutations (Δpmi-2426 ΔrelA199::araC P BAD lacI TT ΔasdA33) to generate strain rSC0012. A plasmid pS-SaoA [5], encoding saoA from SS2, was transformed into this strain. We evaluated the virulence, immunogenicity and protection against challenge with virulent SS2 or S. Choleraesuis C78-3.

Construction and characterization of the S. Choleraesuis vaccine strain rSC0012
Fur is a ferric uptake regulator that is involved not only in iron metabolism, uptake, and transport, but also invasion and survival of S. Typhimurium in the hosts [10][11][12]. The absence of Fur attenuates S. Typhimurium [6,13]. To improve the safety and increase the immunogenicity of S. Choleraesuis vector, a new strain, rSC0012, was generated with an arabinose regulated fur, ΔP fur88 :: TT araC P BAD fur (Fig. 1a).
The phenotypes of the mutations ΔP fur88 ::TT araC P BAD fur and ΔrelA::araC P BAD lacI TT were confirmed by western blot analysis (Fig. 1b). The level of Fur synthesis decreased with arabinose dilution (Fig. 1b). The presence of mutation ΔrelA::araC P BAD lacI TT in rSC0012(pS-SaoA) were confirmed by the increased synthesis of SaoA (Fig. 1b) due to the derepression of P trc promoter on plasmid in rSC0012, which resulted from reduced LacI production whose production was controlled by arabinose. The pmi gene encodes 6phosphomannose isomerase that interconverts fructose-6-phosphate and mannose-6-phosphatein in Salmonella [7]. Because mannose is required for O-antigen synthesis, the Δpmi mutation enables the strain rSC0012 to display a smooth LPS pattern in nutrient broth in the presence of mannose and a rough pattern in the absence of mannose (Fig. 1c). The ΔasdA mutation enables the strain rSC0012 to have an obligate requirement for DAP [14], which can be complemented with a vector harboring the asd gene then eliminates the need for antibiotic resistance genes for plasmid maintenance [15]. The growth rates of rSC0012 with DAP, rSC0012(pS-SaoA), and rSC0012(pYA3493) were similar (Fig. 1d). rSC0012 could grow only with DAP (Fig. 1d).
Antigen synthesis and plasmid stability in S. Choleraesuis rSC0012 Stable maintenance of plasmids and the production of heterologous antigens are critical to ensure efficacy of recombinant live vaccines. The SaoA protein is a highly conserved surface protective antigen among S. suis serotypes [2,5]. Using live attenuated S. Choleraesuis vector delivering SaoA antigen from S. suis will allow to develop a bivalent vaccine against both S. Choleraesuis and S. suis. The stabilities of pS-SaoA and pYA3493 in rSC0012 were evaluated by continuous culturing for 50 generations. The stabilities of both Asd + plasmids, pS-SaoA and pYA3493, were 100% in rSC0012 (data not shown). All rSC0012 colonies examined (100 clones/generation) by endonuclease digestion possessed the Asd + plasmid pS-SaoA or pYA3493. The 34-kDa SaoA protein was detected in cells obtained from both the first and 50th generations of rSC0012(pS-SaoA) (data not shown), indicating the stability of plasmid and stable synthesis of SaoA.
In liver, the titers of rSC0012(pYA3493) and rSC0012(pS-SaoA) were similar to those of vaccine strains rSC0018(pYA3493) and rSC0018(pS-SaoA) at 3, 14, 21, 28 days after inoculation (Fig. 3c). At 3d, 7 d, 14 d, 21 d and 28 d post-inoculation, the titers of rSC0012(pS-SaoA) were significantly lower than those of rSC0011(pS-SaoA), same for two control vector, the numbers of rSC0012(pYA3493) were significantly lower than those of strains rSC0011(pYA3493).respectively(P < 0.01; Fig. 3c), whereas the strain rSC0012(pYA3493) was the fastest to be cleared in liver (Fig. 3c). These results indicated that the ΔP fur88 :: TT araC P BAD fur mutation impaired the colonization in the liver of S. Choleraesuis vaccine strains. In a summary, the ΔP fur88 :: TT araC P BAD fur mutation reduced the colonization ability of S. Choleraesuis vaccine strains in mice spleen and liver but not in the Peyer's patches.
Antibody responses in mice immunized with S.

IFN-γ or IL-4 production induced by S. Choleraesuis strains
To further evaluate the effect of the ΔP fur88 :: TT araC P BAD fur and ΔP crp527 ::TT araC P BAD crp mutations in a strain with multiple preexist mutations on Th1/Th2 immune responses, the levels of IFN-γ and IL-4 in the spleen tissues of mice 7 days and 14 days after booster immunization were measured. The results showed that the titers of IFN-γ and IL-4 induced in mice immunized with rSC0012, and rSC0011 with either pYA3493 or pS-SaoA were significantly higher than those induced in mice inoculated by strain rSC0018 with either pYA3493 or pS-SaoA at both 7 and 14 days after booster (Fig. 5a, b; *, P < 0.01,**, P < 0.05). Although rSC0011(pS-SaoA) induced a slightly higher level of IFN-γ in mice than rSC0012(pS-SaoA) did, no significant difference was observed in 3-week-old mice at both 7 and 14 days after booster (Fig. 5a-c). However, rSC0011(pYA3493) elicited a higher level of IFN-γ than rSC0012(pYA3493) did at 7 Fig. 3 Colonization of Salmonella Choleraesuis rSC0012 in BALB/c mice at diferent time points. Mice were orally inoculated with 1.0 ± 0.3 × 10 9 CFU of the indicated strains. The numbers of bacteria loads in the Peyer's patches (a), spleen (b) and liver (c) of mice at 3 d, 7 d, 14 d, 21 d and 28 d after inoculation were plotted. Bars represent the arithmetic mean ± standard deviations from two separate experiments each with 5 mice per group. *, P < 0.05; **, P < 0.01, for rSC0018 compared to rSC0012 or to rSC0011 with either pYA3493 or pS-SaoA; #, P < 0.05, ##, P < 0.01, for rSC0011 compared to rSC0012 or to rSC0011 with either pYA3493 or pS-SaoA; $$, P < 0.01, for C78-3 compared to rSC0011, rSC0012 and rSC0018 with either pYA3493 or pS-SaoA, as indicated. The data were collected from two independent experiments
Comparison of the protective immunity induced by S.

Choleraesuis strains
To evaluate the protective immunity conferred by rSC0012(pS-SaoA), the mice in each immunized group were challenged orally with 50 × LD 50 of the virulent S. Choleraesuis C78-3 strain or 20 × LD 50 of the virulent SS2 strain at 14 days post-boost immunization. After challenge with C78-3, the results revealed 100% protection in mice immunized with either strain rSC0011(pS-SaoA) or strain rSC0012(pS-SaoA), suggesting full protection. Mice immunized with the rSC0018(pS-SaoA) strains resulted in 20% survival. In contrast, all the mice in the PBS group succumbed to the challenge after 4 days. There were no significant difference between the groups immunized with rSC0012(pS-SaoA) and rSC0011(pS-SaoA), though both displayed significantly higher levels of protection than the group immunized with rSC0018(pS-SaoA) (Fig. 7a). After the SS2 challenge, immunization with the rSC0011(pS-SaoA), rSC0012(pS-SaoA) and rSC0018(pS-SaoA) strains resulted in 80% survival, 95% survival and 16.7% survival with the lethal SS2 challenge, respectively. In contrast, all the mice in the PBS group succumbed to the challenge after 2 days.

Discussion
The ultimate goal of an engineered live vaccine strain relies on achieving the proper balance between immunogenicity and attenuation [18,19]. Achieving that goal will restrict unacceptable reactogenicity to avoid overexcitation of inflammatory responses, but sufficient metabolic activity should be maintained to enable the live vaccine to reach deep lymphatic tissues and induce protective immunity [20]. The development of bacterial vaccine relies on a combination of defined mutations [19,21]. In order to enhancing the immunogenicity of vaccine strains or to disarm them, multiple independent defined mutations were introduced into Salmonella to generate new recombinant attenuated vaccine strains [6,19,22]. By coalescent proper mutations, vaccine strains can be befittingly designed to avoid unacceptable reactogenicity and enhanced immunogenicity [17]. Our previous live attenuated S. Choleraesuis vaccine vector rSC0011 occasionally caused enteritidis in mice [9]. One (See figure on previous page.) Fig. 4 Antibody responses in ten mice. Serum IgG responses to SaoA (a), S. Choleraesuis OMPs (b) and vaginal wash IgA responses to SaoA (c) were measured by ELISA at weeks 3 and 5. Each triangle represents one mouse. Error bars represent variation between mice. Significant differences were indicated. *, P < 0.05; ** P < 0.01, for Salmonella carrying pS-SaoA compared to each other; #, P < 0.05; ##, P < 0.01, for the titers of antibody at 3 weeks after immunization were compared to those at 5 weeks after immunization. No immune responses were detected to antigen tested in mice immunized with PBS or in pre-immune sera from vaccinated mice (reciprocal titer <1:50). The assay was performed in duplicate and repeated at least 3 times of the ways to address this problem is by incorporating a sopB mutation [5,19]. In this paper, another way to improve the S. Choleraesuis vector was reported. Fur is an important regulatory proteins of Salmonella, which has been implicated in the acid tolerance response since fur mutants are acid sensitive and cause altered expression of several acid shock proteins [10,23]. A S. Typhimurium strain with an arabinose regulated fur mutation is adequately attenuated and highly immunogenic [6,12]. However, S. Typhimurium belong to group B, while S. Choleraesuis group C. Whether the ΔP fur88 ::TT araC P BAD fur mutation that results in the proper balance between attenuation and immunogenicity of S. Typhimurium is also appropriate for attenuated S. Choleraesuis vaccine has not been reported previously.
We corroborated the S. Choleraesuis strain rSC0012 (ΔP fur88 ::TT araC P BAD fur Δpmi ΔrelA::araC P BAD lacI TT ΔasdA) displayed a regulated decrease of Fur production in the absence of arabinose. In a previous publication, S. Typhimurium strain with a single ΔP fur88 ::TT araC P BAD fur mutation has shown higher virulent than S. Typhimurium with a single ΔP crp527 ::TT araC P BAD crp mutation by an oral immunization. Unlike report from Curtiss et al. in S. Typhimurium [6], our studies did not showed that the ΔP fur88 ::TT araC P BAD fur mutation has higher virulent than ΔP crp527 ::TT araC P BAD crp mutation in S. Choleraesuis. In fact,the LD 50 of rSC0012(pS-SaoA) was 38-fold higher than rSC0011(pS-SaoA) for 3-weeksold mice with intraperitoneal injection. This result suggest that the ΔP fur88 ::TT araC P BAD fur mutation does different attenuation with S. Choleraesuis or S. Typhimurium,which may due to its complex role as a transcriptional activator of virulence in different strains [24].
The ΔP fur88 ::TT araC P BAD fur mutation can modify the iron regulated outer membrane protein, then strain with ΔP fur88 ::TT araC P BAD fur induced lower inflammation than the isogenic strain with ΔP crp527 ::TT araC P BAD crp mutation or wild type strain in mice. Theses results suggest that the S. Choleraesuis with ΔP fur88 ::TT araC P BAD fur mutation exhibits a lower tendency to trigger excessive inflammation while attenuated sufficiently.
Through the oral vaccination route, a vaccine strain will endure the challenges of acid, bile, and antimicrobial Cytokines levels in ten mice immunized with the S. Choleraesuis vaccines. IFN-γ (a) and IL-4 (b) in spleens 7 d after the booster dose were assayed with an EILSA kit. A PBS control were also included. (c) the ratio of IFN-γ to IL-4. The significant differences between groups of each strain was indicated. The assay was performed in duplicate and repeated at least 3 times. *, P < 0.05; ** P < 0.01, for strains rSC0011, rSC0012, rSC0018 with either pYA3493 or pS-SaoA compared each other; #, P < 0.05; ##, P < 0.01, for the strains rSC0011, rSC0012, rSC0018 with pS-SaoA compared to these strains with pYA3493 respectively at 7 d and 14 d post immunization peptides existing in the gastrointestinal tract. Once inside Peyer's patch, it will face macrophagocytes and T cells during the process transferring to deep lymphatic tissues [13]. Fur is essential to Salmonella for accessorial an adaptive acid tolerance response [10]. Although both rSC0011 and rSC0012 had similar levels of Peyer's patch colonization by oral immunization,we observed that the rSC0012 was cleared more rapidly than the rSC0011 in deeper lymphatic tissues, spleen, and liver of mice. These results suggest that strain rSC0012 was more attenuated than rSC0011 in deep lymphoid tissue, which maybe due to an increase in acid sensitivity cause by loss of Fur makes the cell more susceptible to killing by macrophages.
Both rSC0012 and rSC0011 induced higher levels of IgA, IgG, IFN-γ, and IL4 responses with great colonization compared to strain rSC0018 with less colonization in mice. In general, the live Salmonella vaccines with RDAS display superior colonization level in lymphoid tissues during the invasion stage, leading to enhanced protection by effectively colonizing lymphoid tissues [6,25]. However, there are do exist high levels of colonization but low immunogenicity. A strain with the regulated delayed rfc mutation exhibits superior colonization and yet does not Fig. 6 Induction of inflammatory cytokines at 6 h and 12 h post-infection in ten mice immunized with different Salmonella strains. Analysis of RNA transcript levels by qPCR showed that all mutant strains induced less inflammatory cytokine production than the wild-type strain C78-3 at both 6 h (a and c) and 12 h (b and d) post-infection. &, P < 0.05, C78-3 ΔP fur88 TT araC P BAD fur compared to C78-3; #, P < 0.05, ##, P < 0.01, rSC0012(pS-SaoA) compared to rSC0011(pS-SaoA); *, P <0.05, **, P < 0.01, strains with ΔP fur88 TT araC P BAD fur mutation compared to C78-3 stimulate higher heterologous protection than a Δrfc strain without RDAS [26]. Thus, in addition to colonization level, other determining factors may exist to induce the enhanced protection achieved by regulated delayed attenuation. From the above, Selecting the proper mutation is critical for vaccine development with RDAS. This study confirmed above statement. Although the colonization of rSC0012(pS-SaoA) in mice was less than that of rSC0011(pS-SaoA), rSC0012(pS-SaoA) stimulated stronger serum IgG and mucosal IgA responses than the rSC0011(pS-SaoA). This phenomenon suggests that strain with regulated delayed fur mutation may stimulate stronger antibody response with fewer bacteria than strain with regulated delayed crp mutation.
Both rSC0012 and rSC0011 aroused a Th1 cellmediated response, as ostensived by the significant up-regulation of imprint Th1 cytokines, IFN-γ, the stimulator of Th1-type T cell immune response. It could partially be that the intracellular characteristics of Salmonella enterica cause them to be detected on the surface of APC through MHC-I molecules. This result is consistent with previous studies with S. Typhi [6]. Strain rSC0012 induced higher Th2 cell-mediated response than strain rSC0011, as aroused by the significant up-regulation of imprint Th2 cytokines, IL-4 and greater humoral responses than rSC0011 in mice. This phenomenon may be related to the secreting of heterologous antigen. The more heterologous antigens were secreted during infection, the more likely to trigger an IL-4-dominant response [27]. We found a larger secretion of the SaoA from rSC0012, which could be presented by MHC-II molecules. An additional factor may be due to an increase in acid sensitivity cause by loss of Fur makes the rSC0012 more susceptible to killing by macrophages, thus weakening rSC0012's ability to survive in macrophages. The presence of S. suis-specific IgA serves to promptly deliver the antigen to Peyer's patch dendritic cells or phagocytes and also promptly excite the adaptive immunity during secondary exposure [16]. Generally speaking, the more attenuated the vaccine strain is, the lower its immunogenicity is [17,22], while, the more attenuated rSC0012 strain induced significantly higher IgA and IgG antibody responses to SaoA than the rSC0011 at 3 weeks after the initial immunization in juvenile mice. This may be the result of a equilibrium between attenuation and the ability to participation immune components and activate a controlled over-inflammatory response by the finely crafted strain [19].
The results presented herein highlight that strain rSC0012(pS-SaoA) with regulated delayed fur mutation has retained vaccine efficacy and adequate immunogenicity whilst being safer than previous strain rSC0011(pS-SaoA). Furthermore, the inclusion of the ΔP fur88 ::TTar-aCP BAD fur mutation may be able to decrease inflammation caused by live-attenuated Salmonella enterica serotype Choleraesuis vaccine, which has been an imperfection for other live-attenuated Salmonella enterica serotype Choleraesuis vaccine which transformed from wild type virulent strain. Both S. Choleraesuis and S. suis are major swine pathogens. Currently, there exist license live vaccines against S. Choleraesuis for swine, including Entersol® Salmonella T/C and SC-54 manufactured by Boehringer Ingelheim and Arugs SC/ST, respectively [28]. However, there is no licensed vaccine against S. suis. Preventing the diseases caused by SS2 in swine with a combined vaccine is a long-sought goal.

Conclusions
Our results have shown the strains rSC0012(pS-SaoA) with regulated delayed fur mutation could confer higher protection against challenges with lethal doses of SS2 or Fig. 7 Protection in mice.Groups of 40 mice were orally immunized twice at 3-weeks intervals with indicated strains. Half of the mice were challenged orally with 50×LD 50 of C78-3 and the other half were intraperitoneally injected with 20×LD 50 of SS2 at 2 weeks after the 2nd immunization S. Choleraesuis C78-3. Thus, the use of attenuated S. Choleraesuis to develop a vaccine against S. suis will have the great potential to ease the burden of both pathogens. The regulated delayed fur mutation in the novel vaccine rSC0012 resulted in a well-safety, highly immunogenic, and effective vaccine in mice, this study has paved the way for testing in piglets. Our findings will aid the optimal of a S. Choleraesuis vaccine vector capable of eliciting a suitable immune response against other pathogens.

Animals
Three-week-old female BALB/c mice were purchased from Animal Center of Yangzhou University, and kept 1 week before inoculation. All animal experiments were authorized by the Jiangsu Administrative Committee for Laboratory Animals (permission number SYXK-SU-2007-0005) and accorded to the Jiangsu Laboratory Animal Welfare and Ethics guidelines of the Jiangsu Administrative Committee of Laboratory Animals. All surgery was performed under anesthesia intraperitoneally injected with sodium pentobarbital, 40 mg per kilogram mouse weight. All the animals were humanely euthanized after the study by inhalation of CO 2 ,while injection with sodium pentobarbital, 40 mg per kilogram mouse weight, and all efforts were made to minimize suffering.

Strains, plasmids, and culture conditions
The strains, plasmids, used in this study are described in Table 2. C500, an approved live S. Choleraesuis vaccine strain attenuated by thallium compound in China, was used as an attenuation control [32]. The genetic characterization of this strain has been reported [29]. S. suis serotype 2 (SS2, CVCC3928) and S. Choleraesuis C78-3 (CVCC79103) were purchased from China Institute of Veterinary Drug Control. Plasmid pYA3493 is an Asd + vector with a P trc promoter. The asd gene from Salmonella was used as a unique plasmid marker to be used in asd mutants to constitute a balanced-lethal system [14]. LB medium [5],Nutrient broth (NB) and MacConkey agar (Difco) were used for phenotype characterization. When required, media were supplemented with 2,6-diaminopimelic acid (DAP;50 μg/mL), chloramphenicol (Cm; 25 μg/mL),L-arabinose (0.2% wt/ vol), D-mannose (0.2% wt/vol) or sucrose (5% wt/vol). The empty plasmid vector pYA3493 and expression vector pS-SaoA were described on previous studies [5]. The saoA gene is under the control of the P trc promoter ( Table 2). S. Choleraesuis vaccine vector strain rSC0012 harboring plasmid pS-SaoA (expression vector) or pYA3493 (control vector) were grown in LB broth with both 0.2% arabinose and 0.2% mannose. Selenite broth was used for enrichment of S. Choleraesuis from mice tissues. Strains were prepared as previously described [5,6,9,12]. Bacterial growth was monitored with a spectrophotometer at OD 600 and by direct plating for colony counts.

Construction of S. Choleraesuis mutant strains
Four mutations ΔP fur88 ::TT araC P BAD fur, Δpmi, ΔrelA::araC P BAD lacI TT, and ΔasdA were introduced into S. Choleraesuis C78-3 by conjugation with E. coli χ7213 harboring aforementioned suicide vectors as previously described [33]. The suicide vectors used are listed in Table 2 and Fig. 1a. To construct mutation ΔP fur88 ::TT araC P BAD fur, a 1335-bp TT araC P BAD cassette were used to replace the 239-bp promoter sequence of the fur gene to achieve arabinose-regulated Fur synthesis (Fig. 1a). The araC P BAD cassette contains a transcription terminator (TT) sequence to prevent araC transcription reading through adjoining genes. Plasmid for ΔP fur88 ::TT araC P BAD fur were confirmed by DNA sequencing. All the primers used have been reported [21].

Characterization of S. Choleraesuis mutations in vitro
All mutations were confirmed by colony PCR using homologous primers [12]. The ΔasdA mutation was verified by growth with or without DAP in LB broth [14,15]. Lipopolysaccharide (LPS) profiles were examined by silver staining in 12% sodium dodecyl sulfatepolyacrylamide (SDS) gel for the Δpmi mutation [33,34]. The ΔP fur88 ::TT araC P BAD fur deletion-insertion mutation was verified by reduced production of Fur protein as arabinose concentrations decreased with the increased bacterial growth by western blot using anti-Fur antiserum [12]. The production of SaoA was verified by western blot using anti-SaoA antiserum, respectively [9, 18,19,20,].

Salmonella subcellular fractionation
To evaluate the subcellular localization of synthesized SaoA in the live attenuated S. Choleraesuis vaccine, cultures were grown in NB to an OD 600 of 0.8 and centrifuged at 13,200×g for 5 min to collect supernatant and pellet. The culture supernatant was saved for later analysis of the bacterial secreted proteins. Periplasmic fractions were prepared using the lysozyme-osmotic shock method as previously described [27,35]. Equal volumes of supernatant,periplasmic, cytoplasmic samples were separated by SDS-PAGE. Proteins were then transferred to polyvinylidene fluoride membrane for western blot analysis using anti-SaoA antiserum [5]. The gel band were analyzed with ImageJ software (NIH) [31].

Preparation of SaoA and Salmonella outer membrane proteins (SOMPs)
His-tagged SaoA fusion protein and Salmonella outer membrane proteins (SOMPs) were prepared as previous studies [5].

Determination of virulence in mice
Three-week-old female BALB/c mice were obtained from Animal Center of Yangzhou University. Mice were fasting for 6 h before inoculation. Recombination S. Choleraesuis vector strains were cultured in LB broth with D-0.2% mannose and 0.2% L-arabinose for 12 h at 37°C as standing cultures. The cultures were diluted 1:50 in the same media and cultured at 37°C to an OD 600 of 0.9. Bacteria were collected by centrifugation at 13,200×g for 5 min at room temperature and resuspended in PBS to densities suitable for the inoculation. Serial dilutions of the S. Choleraesuis strains were plated onto LB agar supplemented with 0.2% D-mannose and 0.2% L-arabinose to measure the actual densities. Groups of five mice were inoculated with different doses in 20 μl (oral immunization) or 100 μl (intraperitoneal immunization). The mice were observed for 4 weeks for death. The LD50s were calculated according the method of Reed and Muench [10].

Immunization of mice
Bacteria were grown and collected as above. Serial dilutions of the S. Choleraesuis vaccine strains were plated onto LB agar supplemented with 0.2% D-mannose and 0.2% L-arabinose to determine the actual dose. Threeweek-old female BALB/c mice were orally inoculated with 1 ± 0.2 × 10 9 CFU of S. Choleraesuis vaccine strains containing either pS-SaoA or pYA3493. Mice were boosted inoculated with the same dose of the same strain after 3 weeks. About 50 μL of whole blood was collected by tail vein 3 weeks after primary inoculation and 2 weeks after boosting. Serum was separated from the whole-blood samples and stored at − 70°C. Vaginalwash samples in mice were collected at the indicated time and stored at − 70°C [9,25,30,36]. This experiment was performed in triplicate with each group receiving a similar dose of the vaccine strains.

Tissue collection after Salmonella infection of mice
Three-week-old female BALB/c mice were divided into 5 groups with 10 mice in each group. Groups of mice were orally inoculated with 1 ± 0.3 × 10 9 CFU of Salmonella strains. Spleen and intestinum tenue of the mice were collected at 6 h and 12 h postinfection. The tissues were frozen with liquid nitrogen and then transferred to − 70°C.

Enzyme-linked immunosorbent assay (ELISA)
Serum IgG antibody production against S. Choleraesuis outer membrane proteins (OMPs) and SS2 SaoA,and vaginal-wash IgA antibody production against SaoA in mice were evaluated by Enzyme-linked immunosorbent assay (ELISA) [5]. Cytokines in tissues were analyzed by sandwich ELISA using commercial kits (BD Biosciences) according to manufacturer's instructions. The results from the two experiments were pooled for statistical analysis.

Quantitative real -time PCR (qPCR) for cytokines
For RNA isolation, gut tissues were homogenized and suspended in TRIzol® (Thermo Fisher Scientific,USA). Tubes were vortexed for 3 min to disrupt the tissues. Chloroform was added to TRIzol® -treated samples and the samples centrifuged at 13200×g for 10 min. The aqueous phase was separated out, and the RNA precipitated using precooled isopropanol. For quantitative realtime PCR, 1 μg of RNA was then reverse transcribed to cDNA. The primers were designed using Primer Blast (NCBI net) and synthesized by TSINGKE Biological Technology Co., Ltd. The sequences of the primers are listed in Table 3. Each sample were amplified using 7500 Fast Real -Time PCR Instrument (ABI,US) using Fast SYBR Green Master Mix (Thermo -Fisher Scientific).
The results were using internal reference GAPDH as control for normalization, and the 2^-ΔΔCt method was used to estimate the relative expression level of the mRNAs of target genes.

Statistical analysis
Statistical analyses on ELISA were presented as the geometric means and standard deviations for all assays. A Mann-Whitney U Test (GraphPad Software, Inc.) was applied to contrasting the distribution of the S. Choleraesuis in tissues of mice. The Kaplan-Meier method (SPSS Software) was used for obtain the survival fractions following i.p. challenge of immunized mice. A P value of 0.05 was considered statistically significant.