According to the whole genome sequence of N. seriolae ZJ0503 (accession no. NZ_JNCT01000022), sequence analysis was performed with the BLAST program using NCBI (http://www.ncbi.nlm.nih.gov/BLAST/). The amino acid sequence for NsICL was deduced, and the physical and chemical properties were predicted using ExPASy software (http://www.expasy.org/). The three-dimensional structure of the protein was predicted using the SWISS-MODEL website (https://swissmodel.expasy.org/) and modified by PyMOL software. Multiple sequence alignment analysis of amino acids from different bacterial species was performed using ClustalX 2.0 and GeneDoc software. The phylogenetic tree was constructed using the neighbor-joining method with MEGA 5.0 software.

The wild-type strain of N. seriolae ZJ0503 was taken from a sick Trachinotus ovatus (golden pomfret) in Yangjiang, China in 2005. It is preserved in the Guangdong Provincial Engineering Research Center for Aquatic Animal Health Assessment and is cultivated using brain heart infusion medium (BHI). The deletion strain and complemented strain were constructed using the plasmid pRE112, which is preserved in our lab. The healthy hybrid snakehead were purchased from a fishery in Guangzhou, China, and the size was 13 ± 2 cm in length and 25 ± 5 g in weight. This study was approved by the Animal Research and Ethics Committee (AREC) of Guangdong Ocean University, China. All animal experiment operations were carried out following the university's regulations for animal experiments, and the animal facility was managed in line with the National Institutes of Health (NIH) guidelines for the care and use of laboratory animals (NIH Publication No. 8023, revised 1978).

The extracellular products of N. seriolae ZJ0503 were obtained via the cellophane overlay method. N. seriolae ZJ0503 was cultured on optimized medium agar plates at 28 °C for 72 h to obtain well-grown bacterial colonies, and a single colony was prepared for bacterial suspension. Then, took 100 μl N. seriolae suspension and evenly spread it on the cellophane covered BHI plates, and cultivated at 28 °C for 3–5 days. N. seriolae cells grown on the cellophane sheet were stripped from the plates, and the extracellular products were subsequently washed off with sterilized PBS. The harvested suspension was centrifuged at 8,000 g at 4 °C for 20 min, and the supernatant containing extracellular products was filter sterilized with a 0.2 μm membrane filter. Then, the sterilized supernatant was transferred into a dialysis tube (3.5 kMW) and dialysed in ultrapure water at 4 °C for 16–24 h. During dialysis, the ultrapure water was changed 3–4 times. The purified supernatant was transferred into a centrifuge tube after dialysis and frozen at −80 °C. Finally, it was lyophilized using a vacuum freeze dryer to obtain the protein dry powder which was identified via shotgun mass spectrum (MS).

This study is based on the whole genome sequence of N. seriolae ZJ0503 (accession no. NZ_JNCT01000022). PCR technology is used to amplify the upstream and downstream regions of the NsICL gene respectively. The primers used for this amplification were ICL-UF/ICL-UR for the upstream fragment and ICL-DF/ICL-DR for the downstream fragment (Supplementary Table 1). The two fragments were linked by overlapping PCR. The particular amplification procedure was as follows: 94 °C for 5 min, followed by 35 cycles of 94 °C for 30 s, 64 °C for 30 s, 72 °C for 30 s, and final elongation at 72 °C for 5 min. Next, the overlapping sequences were added to the pRE112 plasmid using Mlu I and Xba I restriction sites to create the deletion plasmid pRE112-ΔICL. Additionally, the NsICL gene was inserted into the pRE112 plasmid to obtain the complemented plasmid pRE112-cΔICL, and all recombinant plasmids were verified using specific primers (112-F/R) as detailed in Supplementary Table 1.

To obtain the deletion strain (NS-ΔICL) and the complementary strain (NS-cΔICL), the deletion plasmid pRE112-ΔICL and the complemented plasmid pRE112-cΔICL were transformed into the N. seriolae ZJ0503 by electroyransformation with a micro pulse (Etta biotech) according to the previously described method (19). This strategy utilized the sucrose-sensitivity conferred by the sacB gene present on the pRE112 vector, which allows for positive selection of double-crossover events and subsequent plasmid curing on media containing sucrose (20). First, 100 μL of N. seriolae cells (1 × 10⁵ cfu/mL) were mixed with 1 μg of the recombinant plasmid on ice for 30 min, then placed into a 96-well plate with 100 μL of mixed solution in each well. The electrotransformation method as follow: voltage 200 V, frequency 30, interval 1,000 ms, duration 60 ms. After electrotransformation, added 100 μL 28 °C preheated BHI liquid medium and incubated for 2 h at 28°C. The recovered bacteria were plated on BHI solid medium containing 20 mg/mL chloramphenicol for screening, and suspected positive clones were verified by PCR using primers 112F/R. Finally, the positive clones were inoculated into BHI liquid medium containing 10% sucrose to counter-select against bacteria that still retained the plasmid, based on the lethal effect of the sacB gene product in the presence of sucrose. After more than 30 generations of continuous culture, the stable hereditary strains NS-ΔICL and NS-cΔICL were obtained, and their stability was verified using ICL-F2/ICL-R2 primers.

The deletion strain NS-ΔICL and the wild-type strain were cultivated to the logarithmic growth phase, and 1 mL bacterial suspension of each strain was pipetted onto the slide to prepare a bacterial smear. The bacterial smears were subjected to Gram staining using a standard protocol. Briefly, the slides were stained with crystal purple for 1 min and subsequently rinsed with sterile water. Then, iodine solution was applied as a mordant for 1 min, followed by another rinse. Decolorization was performed by applying 95% ethanol and gently agitating the slide for approximately 1 min. Immediately after decolorization, the slide was washed with sterile water and blotted dry with absorbent paper. Finally, the samples were counterstained with safranin for 1 min, rinsed thoroughly with water, and air-dried prior to observation under an ordinary optical microscope.

To investigate the growth characters of NS-ΔICL, we monitored the growth rate of three strains (NS-ΔICL, NS-cΔICL and ZJ0503). Single colonies were picked from the plates and put into BHI liquid medium. Each strain was set up with three replicates, and the OD value of each strain was measured and recorded at 12 h intervals. The growth curves of three strains were established with OD value as ordinate and culture time as abscissa.

The challenge tests and fish survival experiments were conducted on hybrid snakehead (weight: 25 ± 5 g, length: 13 ± 2 cm) to evaluate the pathogenicity of NS-ΔICL and NS-cΔICL. To prepare for the challenge experiment, the hybrid snakehead were fed for 7 days. Subsequently, 30 fish per group (3 repetitions) were injected intraperitoneally with 100 μL of N. seriolae NS-ΔICL, NS-cΔICL, and ZJ0503 at concentrations of 104, 105, 106, 107, and 108 cfu/mL. In this study, we have verified that the differences in cfu/ml among the parent strain ZJ0503, mutant strain NS-ΔICL, and complemented strain NS-cΔICL under the same OD value are negligible, so the use of OD value to estimate bacterial concentration is reliable. After the challenge, the mortality rate was recorded daily for 14 days, and the median lethal bacterial dose (LD₅₀) was calculated by probit analysis using the SPSS statistical software package.

Healthy fish were randomly divided into 2 groups (designated as the NS-ΔICL group and the PBS group), with 3 repetitions per group, each containing 70 fish. Additionally, the immunized dosage of NS-ΔICL vaccine was determined by the aforementioned LD₅₀ assay, and the LD₁₅ (5.45 × 10³ cfu/fish) of the NS-ΔICL was chosen according to the calculations in Supplementary Tables 2, 3. The NS-ΔICL group was injected in the abdomen with 100 μL of NS-ΔICL suspension (5.45 × 10³ cfu/fish), while the PBS group was given 100 μL of sterile phosphate-buffered saline (PBS). All fish were fed under the condition of 28 ± 0.5 °C for 35 days post-vaccination (d.p.v.).

After 35 days of immunization with the attenuated vaccine NS-ΔICL or sterile phosphate-buffered saline (PBS), 2 groups (3 replicates/group) of surviving fish were challenged with 30 fish in each replicate. The bacterial suspension of wild-type strain N. seriolae ZJ0503 with a LD₅₀ of 4.32 × 10⁴ cfu/fish was prepared according to the result of LD₅₀ experiment. The 4.32 × 10⁴ cfu/fish of N. seriolae ZJ0503 was injected intraperitoneally into immunized fish at 100 μL/fish. Subsequently, mortality was recorded for 14 days after challenge. The relative percentage survival (RPS) was calculated using the following formula: RPS = [1 – (% mortality of immunized group/% mortality of control group)] × 100.

Following our previous study (21), at 1, 4, 7, 14, 21, 28, and 35 d.p.v., blood samples from 3 randomly selected fish in each group were collected from sterile syringes. After that, the professional protease detection kits provided by Nanjing Jiancheng Institute of Biological Engineering in China were used to carry out the activity levels of lysozyme (LZM), peroxidase (POD), alkaline phosphatase (AKP), and acid phosphatase (ACP) in serum samples.

Similar to our previous study, the immunoglobulin M (IgM) titers in the serum were measured by enzyme-linked immunosorbent assay (ELISA). To expand, the N. seriolae bacterial suspension (1 × 108 cfu/mL) treated after 30 s of ultrasound was wrapped in a dose of 100 μL/well on the 96-well microplate, and carried out IgM detection on the 1, 4, 7, 14, 21, 28, and 35 days after immunity. Then, serum samples were added to each well and blocked with 2% BSA. Antibody combined with the antigen which was detected by using the rabbit anti-hybrid snakehead IgM antibody previously prepared by our lab. Microplates were incubated with goat anti-rabbit IgG HRP conjugate (BOSTER Biological Technology, China). The reaction was developed with a chromogenic reagent tetramethylbenzidine (Nanjing Jiancheng Bioengineering Institute, China) and stopped by 2.0 mol/L H₂SO₄. Absorbance at 450 nm was measured using a microplate reader (Bio-Rad, USA) (21).

The kidney, spleen, and liver, which are immune-related organs in fish, were isolated from 3 randomly selected fish in each group at 1, 4, 7, 14, 21, 28, and 35 days after immunization. And the real-time fluorescence quantitative PCR technology was used to detect the expression levels of immune-related genes such as the main tissue compatibility complex class I α (MHCIα), differentiation cluster 4 (CD4) and Interleukin-8 (IL-8). Gene specific quantitative primes to detect aforementioned genes mRNA transcript expression are listed in Table 1. The qRT-PCR was carried out with the following program: 95 °C for 5 min, 95 °C for 15 s, 60 °C for 30 s, 40 cycles. All tests treat the β-actin gene as an internal reference gene, and each sample has 3 technical repetitions. The mRNA transcript levels of the immune-related genes were normalized to β-actin using the Ct 2^−ΔΔCt method.

All statistical analyses were generated in SPSS 21.0 software (IBM, Chicago, IL, USA). Data were presented as the means ± standard error of the mean (SEM) of triplicate vaccination. Differences between the means of the NS-ΔICL group and PBS control group were analyzed by Student's t-test. To account for multiple comparisons across time points, p-values were adjusted using the Holm-Bonferroni method. Results were considered significant at an adjusted p < 0.05.

Bioinformatic analysis revealed that the NsICL gene is 1,290 bp in length, and is predicted to encode a protein of 429 amino acids with a molecular weight of 46.72 kDa and an isoelectric point (pI) of 4.91. Domain prediction indicated that the protein belongs to the isocitrate lyase superfamily. Further analysis predicted that a region spanning nucleotides 56–362 bp in the NsICL gene encodes a functional domain characteristic of this family, and it is predicted that the amino acid sequence (189–194) of “KKCGHL“ constitutes a key part of the active site (Supplementary Figure 1). The three-dimensional structure of the NsICL protein was predicted using SWISS-MODEL (Supplementary Figure 2) and then compared with that of M. tuberculosis. The models showed significant overlap in their three-dimensional conformations and high structural similarity. This suggests that the function of NsICL may be similar to M. tuberculosis. Multiple sequence alignment demonstrated high conservation of ICL among Nocardia, Rhodococcus, and Corynebacterium (Supplementary Figure 3), while phylogenetic analysis clustered N. seriolae NsICL with other Actinobacterales homologs (Supplementary Figure 4). These results confirmed NsICL as a conserved metabolic enzyme and a potential virulence target, justifying its selection for gene knockout.

The extracellular products of N. seriolae were obtained, and the secreted proteins were identified using shotgun MS. Results showed the peptide sequences of NsICL (VEGDTSVANWLAPIVADAEAGFGGALNAYELQK) were detected with confidence greater than or equal to 99%, which proved that NsICL was a secreted protein of N. seriolae.

After immunization with the NS-ΔICL vaccine, serum samples from fish in the PBS group and the NS-ΔICL group were collected at 1, 4, 7, 14, 21, 28, and 35 d.p.v., respectively. At 21 d.p.v., AKP activity in serum stimulated by NS-ΔICL was significantly higher than that in the PBS group (Figure 1A). During 14–35 d.p.v., the serum ACP activity of NS-ΔICL group was significantly higher than PBS group and peaked at 21 d.p.v. (Figure 1B). During 4–35 d.p.v., POD activity in the NS-ΔICL group was significantly higher than that of PBS group (Figure 1C), reaching the highest level at 14 d.p.v. During 7–35 d.p.v., LZM activity in the serum stimulated by NS-ΔICL was significantly higher than that in the PBS group (Figure 1D). The specific antibodies IgM were detected by ELISA at 1, 4, 7, 14, 21, 28, and 35 d.p.v., respectively. The results showed that IgM levels were significantly higher in the NS-ΔICL group than in the PBS group during the period of 4~35 d.p.v. and peaked at 14 d.p.v. (Figure 1E).

At the transcriptional level, we further analyzed the expression of three immune-related genes (MHCIα, CD4 and IL-8) in the liver, spleen, and kidney by qRT-PCR. The results revealed that the expression of the MHCIα gene reached its peak at 21 d.p.v. in the liver (Figure 2A) and kidney (Figure 2G), while it peaked at 28 d.p.v. in the spleen (Figure 2D). The expression patterns of the CD4 and IL-8 genes were similar in the liver and kidney; they both peaked at 21 d.p.v. (Figures 2B, C, H, I). The expression patterns of the CD4 and IL-8 genes in the spleen were also similar, both peaking at 28 d.p.v. (Figures 2E, F). It is interesting to note that the expression of the MHCIα and IL-8 genes in the kidney not only peaked at 21 d.p.v., but was also significantly higher at 35 d.p.v. than in the PBS group (Figures 2C, I).

After 35 days of immunization, the cumulation mortality rates of NS-ΔICL and PBS were 10.19% and 1.42%, respectively. A challenge experiment with wild-type strain N. seriolae ZJ0503 was conducted at 35 d.p.v. to detect the vaccine efficacy. After the challenge, the cumulative survival rates of fish in NS-ΔICL group and PBS group were measured (Figure 3). The survival rates of fish in the PBS group and the NS-ΔICL group were 29.52% and 83.81%, respectively. The RPS of fish vaccinated with NS-ΔICL was 77.03%, indicating that the NS-ΔICL vaccine provided a good immune protective effect against N. seriolae infection. All the symptoms observed on diseased fish were consistent with those reported for fish nocardiosis.