Levilactobacillus brevis 23017 (CCTCC AB 2018164), Levilactobacillus plantarum 27197, and Levilactobacillus casei 27382 were isolated and identified by our laboratory to have excellent probiotic activity (Cui et al., 2018; Shi et al., 2019). The standard strains Levilactobacillus acidophilus ATCC4356, Levilactobacillus rhamnosus LGG, Lactococcus lactis MG1363, Bacillus subtilis WB800N, and C. perfringens strain C57-1 (contain alpha toxin) were stored in our lab. The E. coli CVCC230 isolate was purchased from the China Institute of Veterinary Drug Control (Beijing, China). MRS medium was used to cultivate Levilactobacillus at 37°C. At 30°C, L. lactis MG1363 was cultivated in an M17 medium. The LAB strains were incubated in a constant-temperature incubator for 18–24 h until the OD₆₀₀ value reached 1 and were used for subsequent experiments. C. perfringens C57-1 was cultured anaerobically in cooked meat medium at 37°C in a constant-temperature incubator overnight, then adjusted the concentration to 10⁸ CFU/200 μL. B. subtilis WB800N and E. coli CVCC230 were cultured in LB medium at 37°C in a constant-temperature incubator overnight for use in PCR.

We used the PDB database,¹ to look up the C-terminal domain of E. coli OmpA, mapped its protein molecular structure with Pymol,² and analyzed surface electrostatic potential.

Figure 1A shows the process of plasmid construction. The C-terminal domain of E. coli OmpA was named OACD. E. coli CVCC230 strain was used as the PCR template to amplify OACD as the anchoring protein with the primers in Supplementary Table S1, the N-terminus contains the EcoR I site and the C-terminus contains the Xho I site. The amplified OACD fragment was linked to the pMD™19-T vector by the cloning kit (3,271, Takara, Beijing, China), followed by inserted into the expression plasmid pET-24a at the EcoR I (1040A, Takara, Beijing, China) and Xho I (1094A, Takara, Beijing, China) sites using T4 DNA ligase (2011A, Takara, Beijing, China) according to the instructions. The constructed plasmid was pET24a-OACD. The plasmid pET28a-RFP has been saved in our laboratory. At the same time, the RFP fragment was inserted into the N-terminal of OACD at the Nde I (1161A, Takara, Beijing, China) and EcoR I sites, named pET24a-RFP-OACD. The recombinant plasmids were transformed into E. coli Rosetta will select positive clones for recombinant protein expression. The positive clones were expanded and cultured in 5 mL LB liquid medium at 37°C with shaking at 180 rpm. Until their OD600 value reached 0.6, followed by a 13 h induction by isopropyl β-D-1-thiogalactopyranoside (IPTG, 1 mM) (I6070D, Biotopped, Beijing, China) at 22°C with shaking at 180 rpm. The absorbances were measured at 600 nm with a spectrophotometer (SpectraMax ABS & ABS PLUS, Molecular Devices, Shanghai, China). The induced bacteria were collected through centrifugation and broken by ultrasound at 200 W for 20 min in ice-water. After sonication and centrifugation of the expressed bacteria, the obtained supernatant contains the soluble protein.

As previously mentioned (Fu et al., 2018), BLPs were prepared using chemical pretreatment. Freshly cultured LAB strains were suspended in 10 mL of 10% trichloroacetic acid and boiled in a water bath for 45 min. Next, the particles were washed three times with 25 mL PBS to remove nucleic acids and proteins. After the last wash step, the particles were resuspended with PBS and stored at −80°C. The BLPs, which were made from L. lactis MG1363, L. brevis 23017, L. plantarum 27197, L. acidophilus ATCC4356, L. rhamnosus LGG, and L. casei 27382, were, respectively, named BLP1363, BLP23017, BLP27197, BLP4356, BLPLGG, and BLP27382. To verify whether RFP can bind to BLPs, the prepared BLPs were added to the supernatant containing the RFP-OACD fusion protein and shaken slowly at 100 rpm for 30 min at 37°C. To remove unbound proteins, the particles were washed with PBS three times. Finally, the binding product was resuspended with sterile PBS.

According to the reference (Gordillo et al., 2020), we can verify whether RFP-OACD successfully binds to BLPs by directly observing whether there was a red light on the surface of BLPs by ultra-high Resolution (Deltavision OMX SR, GE, Fairfield, CT, USA). To this end, we aspirated 2 μL fusion product onto microscope slides and promptly covered them with coverslips. We sealed the perimeter of the coverslips with colorless nail polish. Finally, the resulting slides were observed at ultra-high Resolution.

According to the literature (Raya-Tonetti et al., 2021), we explored the stability of OACD bound to BLPs. The products of RFP-OACD bound to BLP23017 were, respectively, washed with different molar concentrations of NaCl (0.2 M, 0.5 M, and 1 M), urea (0.5 M, 1 M, and 2 M), and buffers of different pH values (4, 7, and 9). Finally, each fusion product was fixed at OD600 of 1, and the RFU of each sample was measured by a fluorescence spectrophotometer (F-7100, HITACHI, Tokyo, Japan). Also, we determined the binding ability of RFP-OACD to BLP23017 at different temperatures (4°C, 30°C and 37°C). The experiment was conducted in three replicates.

Bioinformatics analysis is increasingly being used to create multiple-epitope vaccines that are non-allergenic, non-toxic, and free of unwanted peptide fragments. In our previous experiments, bioinformatics analysis was used to predict the dominant antigenic epitopes of several proteins of C. perfringens to construct the multi-epitope group antigen CPMEA. It includes dominant antigenic epitopes such as Net_B146-322, alpha toxin_284-398, and Zmp_698-1022, targeting different toxins secreted by C. perfringens (Guo et al., 2024). These predicted dominant epitopes were linked with five peptide antigens and five B-cell epitopes, which have been shown to be immunogenic in existing studies (Duff et al., 2019; Aldakheel et al., 2021), to provide protection against various toxins. We used the restriction sites of Nde I and Sal I to replace the RFP in pET24a-RFP-OACD with CPMEA. Finally, pET24a-CPMEA-OACD was transformed in E. coli Rosetta by heat shock. Subsequently, positive clones were screened and induced for expression using the 2.2 mentioned method. After sonication and centrifugation, the supernatant containing CPMEA-OACD was purified using the His-tag Protein Purification Kit (P2229S, Beyotime, Shanghai, China). Then, the purified supernatant was mixed with BLP23017 to prepare COB17.

To demonstrate that OACD and PA might co-display multiple antigens on the BLP surface, we amplified the flagellin protein Hag of Bacillus subtilis and linked it to PA. This flagellin protein of Bacillus subtilis possesses potent immunoadjuvant functions. Figure 1B shows the process of plasmid construction. Bacillus subtilis WB800N was used as the PCR template to amplify the Hag gene, the N-terminus contains the Nde I site and the C-terminus contains the Xho I site. PET24a-PA with Nde I and Sal I (1080A Takara, Beijing, China) sites has been successfully constructed in previous studies. Because Sal I and Xho I are isocaudomers, the enzymatic sites of Nde I and Xho I/Sal I can be utilized to insert Hag into pET24a-PA to eventually obtain the recombinant plasmid pET24a-Hag-OACD. Positive clones were expressed with the method mentioned in 2.2, then purified using the His-tag Protein Purification Kit. The purified soluble CPMEA-OACD protein and Hag-PA protein were well mixed and bound to BLP23017 by shaking the mixture slowly at 100 rpm for 30 min at 37°C. To remove unbound proteins, the particles were washed with PBS three times. Finally, the binding product was resuspended with sterile PBS. Identification of two proteins bound to BLP by SDS-PAGE analysis of the resuspended binding product.

SDS-PAGE and Western blotting were used to analyze the expression and purification of CPMEA-OACD and Hag-PA proteins, as well as the proteins bound to BLPs. Briefly, the recombinant protein was transferred to a polyvinylidene difluoride (PVDF) membrane following SDS-PAGE. The membrane was blocked with 5% powdered skim milk at 37°C for 2 h. Then, wash three times with PBS and incubate with the primary antibody at 4°C overnight. Western blot analysis of CPMEA-OACD using the anti-α-toxoid antibody diluted at 1:1000 as the primary antibody (Ys-2273R, YaJi biological, Shanghai, China) and a horseradish peroxidase (HRP)-conjugated goat anti-rabbit IgG antibody (Sigma, United States) diluted at 1:2000 as the secondary antibody. The western blot analysis of Hag-PA using the His Tag Mouse Monoclonal Antibody (HRP Conjugated) (1:2000; AF2873, Beyotime, Shanghai, China). After washing, the enhanced chemiluminescence (ECL) (BL520A, biosharp, Beijing, China) was used to detect the conjugated horseradish peroxidase. The relative expression of specific bands was analyzed and quantified using ImageJ Software (GeneGnome XRQ, Syngene, United Kingdom).

4–6 weeks old female Kunming mice with weight of 18–22 g were purchased from Changsheng Biological Company (Liaoning, China). The mice were maintained in a controlled environment and had free access to rodent food and tap water during a 12-h cycle of light and darkness. Before the formal experiment, the mice were allowed to adapt to the laboratory environment for 3 days. To explore the best route to inoculate COB17, the mice were randomly divided into five groups (n = 8), and each group was immunized by different immune routes. On the d 0, the control and infection groups were subcutaneous injected with PBS 200 μL. The next three groups were immunized with the same doses of COB17 but via different routes, subcutaneous immunization (50 μg/200 μL); intranasal immunization (50 μg/100 μL), the COB17 was slowly dropped into the nasal cavity of mice and inhaled naturally (50 μL/nostril), and 30 min between immunizations of both nostrils; and oral immunization (50 μg/200 μL). Blood and feces from mice were collected on d 7, 25, and 37 after immunization. The blood was obtained from the tail vein and centrifuged at 1,800 × g for 10 min, and the serum was taken for the cytokines and antibodies test. Briefly, fecal samples as weighted pellets were added to PBS containing 0.1% sodium azide (1 mL/100 mg fecal sample). The pellet was vortex mixed and centrifuged at 13,000 × g for 10 min at 4°C. Supernatants were collected and stored at −70°C for the determined of IgA antibodies according to the method of reference (Kim et al., 2021). On the 31st d of immunization, mice other than those in the control group were challenged with C. perfringens type A (10⁸ CFU/200 μL) by intraperitoneal injection. After the infection, weight changes and survival rates of mice were observed and recorded for 7 days. Finally, all mice were euthanized by cervical dislocation to collect jejunum and spleen tissues. The spleen tissues were stored in −80°C for RNA extract. Intestinal mucus from the jejunum of mice was collected for total secretory immunoglobulin A (SIgA) content. The collected ileum tissues were fixed in 10% formalin pathological section. The experiment was conducted in three replicates. The animal study protocol was approved by the Institutional Review Board of the Northeast Agricultural University for animal experiments (NEAUEC-20200303).

Based on Zhang et al.’s method (Zhang et al., 2022), specific antibodies against C. perfringens type A (IgA, IgG) were detected by indirect ELISA. Briefly, 96 well plates (31,121, Labselect, Tianjin, China) were coated with 3.4 μg/mL CPMEA 100 μL per well overnight at 4°C, then blocked for 2 h at 37°C with 5% skim milk. The serum of mouse collected on d 7, 25, and 37 was diluted 1:200 and then serially diluted at a 2-fold ratio, added serum to 96 well plates, and incubated at 37°C for 1.5 h. Peroxidase-bound anti-mouse IgG (A0216, Beyotime, Shanghai, China) or IgA (RS030211, Immunoway, Texas, United States) antibodies were added and incubated at 37°C for 1 h. The reaction was carried out with a TMB (PR1210, Solarbio, Beijing, China) substrate reagent. The enzyme reaction was stopped after 20 min by the addition of 2 M H₂SO₄, and the optical density was measured at 450 nm with a spectrophotometer (Thermo Fisher Scientific, New York, NY, USA). High-affinity antibody responses were measured by indirect ELISA as described above after addition of urea (4 mM) to remove antibodies bound to the antigen with low affinity (Kim et al., 2021). The presence of a high affinity antibody indicates that the antibody is more specific and can validate the accuracy of the specific antibody results. The mucosal SIgA was detected by ELISA kit according to the instructions (D721136, Sangon Biological Technology, Shanghai, China). The experiment was conducted in three replicates.

Based on the method of Shi et al. (2022), the detection of spleen tissue cytokines was implemented with a little modification. The spleen tissues of mice were collected aseptically, and total RNA was extracted from the spleen using the RNA extraction kit (ER101-01, TransGen Biotech, Beijing, China). Then reverse transcribed into cDNA using ReverTra Ace® (TRT-101, TOYOBO, Shanghai, China), following the manufacturer’s protocol, each 40 μL reverse transcription reaction contained 2 μg RNA template. The cDNA was used for quantification and expression with 2 × Hi SYBR Green qPCR Mix (A2250B, HaiGene Biotech Co., Ltd., Harbin, China). Furthermore, the Real-Time Quantitative PCR reactions were carried out using the Applied Biosystems® 7,500 Real-Time PCR Systems (Analytik Jena AG, Jena, Germany) according to the instructions. Actin β was taken as the reference gene. Primers for this study were synthesized with the company (Comate Bioscience Co., Ltd., Changchun, Jilin, China) and provided in Supplementary Table S1 according to the reference (Cheng et al., 2014; Chen et al., 2016; Pan et al., 2022; Zhang and Yi, 2022). The relative mRNA levels were quantified using the 2^-ΔΔCT method (Livak and Schmittgen, 2001). The experiment was conducted in three replicates.

Serum IL-4 (EK0405), IL-5 (EK0408), TNF-ɑ (EK0527), and TGF-β1(EK0515) cytokine levels were detected by ELISA kit according to the instructions (Boster Biological Technology, China). The experiment was conducted in three replicates.

The method of toxin neutralization test in vitro was determined according to the method that has been reported (Goossens et al., 2016). The supernatant of C. perfringens C57-1 cultured anaerobically in cooked meat medium was cultured overnight on the blood plate at 37°C. It was observed that the internal complete hemolytic ring caused by θ-toxin and the external incomplete hemolytic ring caused by α-toxin appeared on the blood plate. We mixed the supernatant of C. perfringens C57-1 with the serum of the mice immunized in equal volumes and incubated at 37°C for 2 h. The mixture was dropped on the blood plate and evaluated for hemolytic situation after overnight culture. The experiment was conducted in three replicates.

According to the previous established pathological tissue staining protocol (Trajković et al., 2007), after 7 d of C. perfringens C57-1 challenged, the mouse ileum were collected and fixed in 10% paraformaldehyde at room temperature for at least 48 h for histopathological evaluations. Pathological sections were prepared by a commissioned company (Wuhan Servicebio Technology Co., Ltd., Wuhan, Hubei, China).

All data were statistically analyzed using GraphPad Prism 9 software. Multiple comparisons, compare the mean of each column with the mean of every other column, of one-way ANOVA were used to analyze the significance of differences between groups. Survival analysis was performed using the Kaplan–Meier method, and differences in the survival rates were assessed by the log-rank test. Results are the averages from independent experiments with standard deviations indicated by error bars. Results were expressed as mean with SD. The asterisks above bars refer to differences compared to the control group; the two groups connected by a horizontal line with asterisks above them represent the differences compared between these two groups. Statistical significance of p < 0.05 indicates a significant difference, and p < 0.01 indicates a highly significant difference. The primers designed in this study were done with oligo7 software.

Figures 2A,B depicts the structure of the C-terminal of E. coli OmpA. The C-terminal domain of OmpA consists mainly of two long α chains and four β chains, which is known as “β/α/β/α-β” fold. β4 chain and other β chains are reverse parallel, with short helix α1 connecting β1 and α2, and the extension formed by α4 and α5 is fixed by a disulfide bond. According to the literature, we indicated the presumed PG binding point with an arrow. The results of surface electrostatic potential showed that the environment of the groove indicated by the arrow was very hydrophilic, and it was speculated that the hydrophilic groove could accommodate PG chains.

We identified if OACD can serve as an innovative anchoring protein for displaying antigens on the BLP surface based on the flowchart in Figure 2C. Firstly, RFP protein and RFP-OACD fusion protein were constructed and expressed, and the purified proteins were used to bind to BLPs. As shown in Figure 2D, we characterized protein expression by SDS-PAGE. After three washes of PBS, the precipitate after RFP without OACD bound to BLPs had no color, and the precipitate after RFP-OACD bound to BLPs was red. This result indicates that the binding of RFP and BLPs is dependent on OACD (Figure 2E). In addition, the phenomenon observed by fluorescence microscopy further confirms that RFP was bound to BLPs by OACD (Figure 2F).

The products of RFP-OACD bound to BLP23017 were washed with buffer solutions having varying molar concentrations of NaCl, urea, and pH in order to determine the stability of the OACD-BLP23017 surface display system. As shown in Supplementary Figure S1A, the samples were washed with NaCl buffer solutions of different concentrations, and OACD still adheres to the surface of BLPs. Similarly, when washing with urea of different concentrations, OACD still adheres heavily to BLP, and higher concentrations of urea may be required for separation. At pH = 4, it was observed that fluorescence intensity was stronger. As the pH value increased, the fluorescence intensity gradually decreased. At the same time, we also bound the fusion protein to BLPs at different temperatures (4°C, 30°C, 37°C). At 30°C and 37°C, the binding ability of OACD to BLPs was not significantly different. At low temperatures, the binding capacity is significantly reduced. Overall, these results confirm that the binding of OACD to BLP is relatively stable. In summary, these results confirm that the OACD-BLP23017 surface display system is stable.

BLPs were produced and bound to RFP-OACD in order to investigate the possibility of using OACD as an anchoring protein for the presentation of antigens on various LAB surfaces. As shown in Supplementary Figure S1B, the binding product of RFP-OACD bound to BLPs after washing and centrifugation showed a red precipitate, indicating that OACD can bind to BLP27197, BLPLGG, BLP4356 and BLP27382 prepared from different LABs.

We created the pET24a-CPMEA-OACD recombinant plasmid in order to create the COB17 vaccine. Under the induction of IPTG at 22°C overnight, CPMEA-OACD achieved optimal expression. SDS-PAGE (Figure 3A) and western blot (Figure 3B) detected similar bands with predicted molecular weights of CPMEA-OACD. Western blot analysis used the anti-α-toxoid antibody (1:1000) as the primary antibody and a horseradish peroxidase (HRP)-conjugated goat anti-rabbit IgG antibody (1:2000) as the secondary antibody. Next, CPMEA-OACD is bound to BLP23017. SDS-PAGE confirmed that the fusion protein was successfully bound to BLP23017, which demonstrates the successful construction of COB17 (Supplementary Figure S2).

To demonstrate that OACD and PA might co-display multiple antigens on the BLP surface, we constructed pET24a-Hag-PA. SDS-PAGE (Figure 3C) and Western blot (Figure 3D) detected similar bands with predicted molecular weights of Hag-PA. The purified CPMEA-OACD protein and Hag-PA protein were co-incubated with BLP23017 and analysis by SDS-PAGE of bound BLP particles. The results showed that the co-incubation group of both proteins with BLP showed similar bands with predicted molecular weights of CPMEA-OACD and Hag-PA proteins. It confirmed the successful binding of the CPMEA-OACD protein and Hag-PA protein with BLP23017 (Supplementary Figure S2). Finally, multiple antigens were successfully displayed on the surface of the same BLP by OACD and PA.

The challenge assay using C. perfringens C57-1 was used to evaluate the immune-protective effect of COB17 in 8 mice per group (Figure 4A). Figure 4B demonstrates the survival rate, in which all mice in the infection group died on the first day. In the three immunized groups, the survival rates were significantly higher compared to the infection group (p < 0.05). The immune protection rate was 75% (6/8) in the CPMEA-OACD-BLP23017 (s.c.) group, 50% (4/8) in the CPMEA-OACD-BLP23017 (i.n.) group, and 75% (6/8) in the CPMEA-OACD-BLP23017 (i.g.) group. Figure 4C demonstrates the percent change in body weight. All mice in the immunized group had significantly lower body weights compared to the control group on the seventh day after the infection (p < 0.01). The mice in the CPMEA-OACD-BLP23017 (i.g.) group showed a minimal weight loss trend, and the CPMEA-OACD-BLP23017 (i.n.) group showed the largest trend. The above results showed that CPMEA-OACD-BLP23017 had a good immune protective effect on C. perfringens C57-1 infected mice, and the oral immunization group had a better immune effect.

To evaluate the antibody levels in the serum of COB17-induced mice, specific IgG and high-affinity IgG antibodies were determined in serum samples. As shown in Figures 4D,E, the specific IgG antibodies and the levels of high-affinity IgG antibodies were significantly higher in all vaccination groups than in the control group (p < 0.01). The above results also showed that serum IgG antibody levels were higher in the subcutaneous injection group.

The serum-specific IgA and SIgA antibody levels were determined to evaluate the level of mucosal immunity induced by COB17 in mice. Figure 4F showed that the specific IgA antibodies were significantly higher in all vaccination groups than in the control group (p < 0.01). Figure 4G shows that the total SIgA of the three immunized groups of mice was higher than that of the control group. The above results also showed higher levels of mucosal immunity in the oral immunization group.

To assess the changes in immune-related factors in serum, we measured levels of IL-4, IL-5, TNF-α, and TGF-β1 by ELISA kits. Figure 5 shows that the levels of IL-4, IL-5, and TNF-α in each immunization group were higher than the control group, while the results in TGF-β1 were the opposite. The results indicate that the oral COB17 vaccine significantly regulates the expression of immune-related factors in serum.

To determine the changes in immune-related cytokines in splenocytes, we performed real-time qPCR analysis of the mRNA expression. Figure 6 depicts the levels of each cytokine in the CPMEA-OACD-BLP23017 (i.g.) and CPMEA-OACD-BLP23017 (i.n.) groups were higher than the control group. The CPMEA-OACD-BLP23017 (s.c.) group had higher levels of IL-1β, IL-4, IL-6, and IL-12 than the control group. These results indicate that the oral COB17 vaccine significantly increased the expression of immune-related cytokines in splenocytes compared to subcutaneous and intranasal drops.

To demonstrate that mice immunized with COB17 can be neutralize to alpha toxin infection in addition to C. perfringens type A infection, we performed alpha toxin neutralization assays in vitro using sera collected from the immunized mice. As shown in Figure 7A, an external hemolytic ring was visible on the blood plate after the control group sera were co-incubated with the alpha toxin, which size does not differ much from the size of the external incomplete hemolytic ring caused by the alpha toxin. The CPMEA-OACD-BLP23017 (i.n.) group showed a slight larger hemolytic ring compared to the other immunized groups, indicating that this group of mice was less protected against alpha toxin challenge compared to the other immunized groups. In conclusion, the mice in the three immunization groups were able to neutralize the challenge of alpha toxin.

To observe the intestinal damage of C. perfringens type A in mice, the ileum was collected and prepared into pathological sections. As shown in Figure 7B, the ileum in the control group shows a normal morphology with neat and complete villi and no inflammatory cell infiltration. The submucosa and epithelial cells of the ileum in the CPMEA-OACD-BLP23017 (s.c.) group were normal, with only slight inflammatory cell infiltration. The ileum of the CPMEA-OACD-BLP23017 (i.n.) group had submucosal edema with the detachment of epithelial cells and severe inflammatory cell infiltration. The ileum in the CPMEA-OACD-BLP23017 (i.g.) group had normal mucosal epithelial cell structure, a regular and neat arrangement, and slight inflammatory cell infiltration. The results showed that COB17 can help prevent the intestinal damage induced by C. perfringens type A in mice.