Background

Streptococcus iniae is a prominent warm-water pathogen and is known to infect a wide range of fish species. The infected fish exhibit varied clinical symptoms of erratic swimming, lethargy, meningoencephalitis, exophthalmia, enteritis and septicaemia [1, 2]. According to the previous reports, S.iniae can cause severe diseases in Trachinotus ovatus and result significant economic losses [3, 4]. Meanwhile, S.iniae is a zoonotic bacteria that poses a threat to public health [5].

Different from other streptococci of the same genus, the serotype of S.iniae cannot be classified by the traditional method of Lancefield [6]. Random amplified polymorphic DNA (RAPD) or repetitive primer polymerase chain reaction (rep-PCR) method can be used to distinguish the serotypes of S.iniae [7, 8]. Through analysis of rep-PCR and RAPD, 29 isolates of S.iniae were divided into two genotypes and a correlation between the genotype and the virulence had been identified [9]. Fuller et al. reported that S.iniae virulence is associated with distinct genetic profile and demonstrated differences between pathogenic and nonpathogenic isolates [10]. Rep-PCR analysis of 14 S.iniae strains from diseased O.niloticus presented genetic heterogeneity and were divided into six genotypes banding patterns [11]. The routine methods for preventing bacterial infections are the use of antibiotics and chemotherapeutants, but antibiotic usage in the aquaculture industry is largely uncontrolled. Abuse and/or excessive use of antibiotics can lead to the emergence of antimicrobial-resistant bacteria [12]. However, there is little information about the mechanisms behind antibiotic resistance and virulence differentiation among diverse S.iniae strains, with the lack of information hampering the effective treatment of the disease.

Comparative genomics analysis is a helpful way to identify genome-wide genetic variants of bacteria that may be associated with host and geographic origin but also to better understand their potential pathogenicity and antibiotic resistance. Comparative genomic analysis of S. agalactiae isolates with distinct clinical origins or host associations has provided insight into potential mechanisms of evolution, virulence, and host adaptation [13]. Fanelli et al. identified a large number of antibiotic, heavy metals and virulence determinants by reporting the whole genome sequencing and genomic characterization of two Arcobacter butzleri strains isolated from shellfish [14]. Similarly, the virulence and antimicrobial resistance-associated genomic determinants of two Salmonella Typhimurium strains were reported through comparative genomics approaches [15]. Although some complete genome sequences of S.iniae isolated have been sequenced, to our knowledge, there have been no reports on the comparative genomics analysis for understanding the genetic basis of pathogenicity and multidrug resistance. In our previous work, we isolated 17 strains of S.iniae from T.ovatus and differentiated them into two genotypes using RAPD and rep-PCR methods [16]. Among them, BH15-2 was divided as designated genotype A in RAPD analysis and genotype 1 in rep-PCR analysis, while BH16-24 was classified as genotype B and genotype 2. Here, we compared the phenotypes (growth, drug-resistance, and virulence variation) and genome of BH15-2 and BH16-24, which may explain the mechanism of their differences in biological properties.

Results

Clinical symptoms of infected fish and biochemical characterization of the strain

In this study, the natural infected fish were observed with symptoms of meningoencephalitis, enteritis, and hemorrhage of pterygiophore, operculum and liver (Fig. 1A-E). After 24 h incubation at 28 ℃ on BHI agar plates, the pathogenic bacteria formed circular, buff, smooth surface, intact edge, convex colonies with a diameter of approximately 0.5-1.0 mm (Fig. 1F). Gram staining showed that the strains are Gram-positive coccus appearing in short chains under an optical microscope. Biochemical analyses revealed complete homogeneity in BH15-2 and BH16-24. According to the growth curves, the growth rate of BH16-24 was significantly faster than that of BH15-2 after 6 h of culture (p < 0.05) (Fig. 2).

Fig. 1
figure 1

Clinical signs of diseased T.ovatus. (A) pterygiophore hemorrhage, (B) meningoencephalitis, (C) operculum hemorrhage (D) liver hemorrhage (E) enteritis (F) Bacterial colony on BHI agar

Full size image
Fig. 2
figure 2

Growth curve of BH15-2 and BH16-24 strains cultured in BHI broth

Aliquots of cell culture were taken at various time points and measured for cell density at OD600, *indicates P < 0.05,**indicates P < 0.01

Full size image

The results of virulence test

The infection results from BH15-2 and BH16-24 were shown in Table 1. Cumulative mortality rates of BH15-2 and BH16-24 after post infection were illustrated in Table 2. At 28 days post-infection (dpi), BH15-2 induced higher levels of mortality and revealed a greater virulence than BH16-24. Concretely, the LD50 value of BH15-2 was 4.0 × 102 CFU/g, while that of BH16-24 was 1.2 × 105 CFU/g (body weight). The virulence of BH15-2 was 1000 times than BH16-24. BH15-2 caused 75%, 75%, 87.5% and 100% cumulative mortality, while BH16-24 caused 6.25%, 6.25%, 12.5% and 62.5% at 1.5 × 105, 1.5 × 106, 1.5 × 107, 1.5 × 108 cfu/ml concentrations in T. ovatus respectively. In the control groups, there was no mortality during the observation period. Most deaths occurred within 48 h p.i. and did not show any clinical signs of disease prior to death. But bacterial isolated were retrieved from the brain and liver of the challenged fish and found to be the same species.

Table 1 Total mortality of T. ovatus challenged by intraperitoneal injection with 100 µl of S.iniae strains BH15-2 and BH16-24
Full size table
Table 2 Results of antimicrobial susceptibility of BH15-2 and BH16-24.
Full size table

Drug resistance

The antibiogram study results are shown in Table 2. Both strains were sensitive to tetracycline, erythromycin, norfloxacin, azithromycin, ciprofloxacin, chloramphenicol, deoxytetracycline, and intermediate resistant toward to neomycin, gentamycin. BH16-24 was resistant or intermediate resistant toward all the β-lactam antibiotics tested in this study, i.e., penicillin, ampicilin, cefazolin, ceftriaxone Sodium, and ceftazidime, while BH15-2 was sensitive to these antibiotics. In addition, BH16-24 was resistant to sulfonamide antibiotics (sulfasomizole, compound sulfamethoxazole), polymyxin B, aminoglycoside antibiotic (spectinomycin), rifampin, whereas BH15-2 was susceptibility to these antibiotics. Therefore, BH16-24 is a multidrug-resistant (MDR) bacterium. The calculated MAR index of BH15-2 was 0, while BH16-24 was 0.37.

Comparative Genomics

General features of whole genome sequencing results

The general features of BH15-2 and BH16-24 genomes are summarized in Table 3. The genome size of BH15-2 was 2,175,659 bp with a GC content of 36.80%, while BH16-24 is 2,153,918 bp long with a GC content of 36.83%. Both genomes were optimized to assemble a circular genome with 0 gap. The genome of BH15-2 contained 2090 genes, 60 tRNAs and 15 rRNAs, while the genome of BH16-24 had 2039 genes 68 tRNAs and 18 rRNAs. Figure 3 shows the assembling results of BH15-2 and BH16-24 strains. The circular genomes of the two strains exhibited the coding sequence (CDS), repetitive sequences, GC content, number of RNA and GC skew (Fig. 3), where the outer 2 and 3 circles represented the CDS on the positive strand and negative stands. The genome sequences of BH15-2 and BH16-24 have been summited to the National Center for Biotechnology Information database with the accession number CP132229 and CP132230, respectively. Bioinformatics analysis indicated that BH15-2 and BH16-24 contains 71 and 21 unique genes, respectively, with a total of 2018 common genes.

Table 3 General features of BH15-2 and BH16-24.
Full size table
Fig. 3
figure 3

Schematic circular diagrams of the S.iniae BH15-2 (A) and BH16-24 (B) genomes

Circle 1 (from outside to inside): scale (in kb); circles 2 and 3: genes in positive strand and negative strand; circle 4: repetitive sequences; circle 5: tRNA (in blue) and rRNA (in purple); circles 6: GC content; circles 7: GC skew

Full size image

GIs analysis

The genomes of BH15-2 and BH16-24 were screened for horizontally acquired DNA using IslandPath-DIMOB, SIGI-HMM, IslandPick, and Islander methods integrated with the IslandViewer server (Fig. 4). In the genome of BH15-2, 17 presumed genome islands (GIs) ranging from 4912 to 50,177 bp were detected. The largest GI consisted of 50,177 bp with 79 predicted gene coding regions, of which 71 genes were unique in BH15-2 genome. And the shortest GI consisted of only 5 predicted gene. A total of 275 genes were predicted into GIs. For strain BH16-24, 21 presumed GIs ranging from 4716 to 38,833 bp were detected. The largest GI consisted of 38,833 bp and predicted to encode 45 genes. A total of 263 genes were predicted into GIs. It was found that most of the GIs of the two strains are the same by comparing the GIs of BH15-2 and BH16-24. However, based on the large inversion fragment between BH15-2 and BH16-24, the positions of several GIs are very different. The two strains each have a unique gene island, composed of unique genes (circled in red in Fig. 4). Both strains have no unique genes encoding virulence or drug resistance related genes.

Fig. 4
figure 4

Circular visualization of the predicted Genomic Islands (GIs) on BH15-2 (A) and BH16-24 (B) strains. The analysis was conducted in IslandViewer 4. The interactive visualization of the distinct islands across the genomes is shown with blocks colored according to the predictor tool as described: red represents the predicted by at least one method, blue represents that the results of IslandPath-DIMOB predicted, yellow represents the predicted results of SIGI-HMM

Full size image

Prophage analysis

Prophage analysis of BH15-2 and BH16-24 showed that the two strains harbored 11 identical incomplete and questionable prophages (Supplementary Figure S1). But due to the inversion fragment, several of them are in different positions. Further, BH15-2 harbored an intact prophage and is also an active prophage predicted by Prophage Hunger (Fig. 5), which was absent in BH16-24. The intact prophage is 46 kb size located at 897,486 bp – 943,491 bp, contained a putative attL site and an attR site. And it encoded 60 prophage-related proteins and 17 hypothetical proteins, including phage tail protein, phage protein, siphovirus Gp157 family protein, ImmA/IrrE family metallo-endopeptidase, XRE family transcriptional regulator and endolysin.

Fig. 5
figure 5

Genomic organization of coding sequences (CDS) of prophage in BH15-2 (show annotated only)

Att: Attachment site; PLP: Phage-like protein; Ter: Terminase; Por: Portal protein; Sha: Tail shaft; Coa: Coat protein; Tra: Transposase

Full size image

Large-scale genomic inversion fragment

The genomic synteny analysis of the whole S.iniae genome of BH15-2, BH16-24, and other three strains SF1, YSFST01-82 and ISET0901 which separately isolated from Oreochromis niloticus, and Paralichthys olivaceus was performed using Mauve software (The strain information is listed in Table 4). The result showed that all strains except BH16-24, were quite similar with respect to genome structure, with the exception of some small inversion. Thus, all strains except BH16-24 shared a similar synteny with each other. The BH16-24 genome displayed a large-scale inversion including 1.28 Mbp occurred across the origins/terminus axis (ori/ter axis), which located in the region from 502,513 bp-1,788,813 bp (Fig. 6).

Table 4 S.iniae genome sequences included in genomic comparison
Full size table
Fig. 6
figure 6

Mauve comparison diagrams of the BH15-2, BH16-24, SF1, YSFST01-82 and ISET0901. The scale represents the coordinates of each genome. Different color blocks represent local collinear blocks (LCBs), which are conserved segments in five genomes. Within LCBs, the white area represents low similarity regions or regions unique to one genome but absent in another. LCBs above the black horizontal central line are in forwarding orientation and below this are in reverse orientation. Colored lines show the rearrangement of LCBs among the genomes

Full size image

Antibiotic resistance genes in the inversion fragment

The large inversion fragment encodes 1242 genes. Among them, many antibiotic resistance genes including efflux system genes, ABC transporter and β-lactamase-encoding were screened out, such as penicillin-binding protein 2B (pbp2B), penicillin-binding protein 1 A (pbp1A), penicillin-binding protein 2X (pbp2X), aminoacyltransferase (femA), DNA gyrase subunit A (gyrA), oligopeptide-binding protein (oppA), aminoglycoside phosphotransferase (aph), dihydrofolate reductase (dfrA), macrolide export ATP-binding/permease protein (macB) (Table 5).

Table 5 The main antibiotic resistance genes encoded by the large inversion fragment
Full size table

Virulence related genes in the inversion fragment

According to the annotation results of the VFDB (Virulence Factors of Pathogenic Bacteria) database, the two strains contain 336 putative virulence related genes, of which 221 virulence genes are in the inversion fragment. Among them, several important virulence factor such as capsule (cpsA, cpsB, cpsC, cpsD), CAMP factor (cfa/cfb), C5a peptidase (scpA, scpB, scpI), phosphoglucomutase (pgmA), beta-hemolysin/cytolysin (cylG, cylA, cylA, cyll), hemolysin A/B (hlyA, hlyB), laminin-binding protein (lmb), arein the inversion fragment (Table 6).

Table 6 The main virulence genes encoded by the large inversion fragment
Full size table

Discussion

There have been many studies to distinguish the genotypes of S.iniae using molecular methods. However, the multidrug resistance and virulence differentiation characters and the underlying mechanisms with different genotypes have been poorly explored to date. The investigation on the molecular mechanisms of drug resistance and virulence variation is essential to the prevention of further spreading of these multidrug resistance strains or the occurrence of new resistant strains. The results showed great significance to explain the multidrug resistant and virulence differentiation of S.iniae strains. As reflected by our results, BH16-24 is a native multidrug resistant and low virulence strain. Further genomic comparisons between BH15-2 and BH16-24 revealed valuable information on the possible multidrug resistance and virulence differentiation among S.iniae strains. To our knowledge, this is the first report on the differences in biological characteristics and genomes among different genotypes of S.iniae.

Fish infected of S.iniae often show a variety of clinical signs, such as anorexia, lethargy, erratic swimming, and visceral hemorrhage [19]. Similarly in the present study, clinical signs including meningoencephalitis, enteritis, hemorrhage of pterygiophore, operculum, liver hemorrhage were also observed. In our previous study, the two genotypes strains of S.iniae showed identical phenotypic features [16]. However, the growth ability of the two strains was different, where the growth rate of BH16-24 was significantly faster than that of BH15-2 at the exponential stage according to the growth curve.

S.iniae recovered from farmed fish were genetically distinct from wild reef fish and exhibited a trend toward higher minimal inhibitory concentrations against several antibiotics [20]. Consistently, our results suggest that BH15-2 was susceptible to most of the tested antibiotics except neomycin and gentamycin. In contrast, BH16-24 exhibited multidrug resistance, including high resistance against penicillin, ceftazidime, sulfasomizole, compound sulfamethoxazole, spectinomycin, polymyxin B, rifampin and intermediate against ceftriaxone sodium, ampicillin, cefazolin, neomycin, and gentamycin. In the previous study, S.iniae obtained from fish were susceptible to most of the antibiotics [4, 21]. However, according to recent reports, more and more resistant strains of S.iniae was found to be resistant to many important antibiotics including amoxicillin, penicillin, ampicillin, gentamycin,spectinomycin, amikacin, neomycin, enrofloxacin, lincomycin, and sulfamethoxazole [17, 22, 23]. The emergence of multidrug resistance strain is worrisome. In addition, both BH15-2 and BH16-24 remained highly susceptible to tetracycline, erythromycin, norfloxacin, azithromycin, ciprofloxacin, chloramphenicol and deoxytetracycline, which is indicative that these antibiotics might be useful in controlling the disease in the future.

The relationship between genotype and virulence of S.iniae isolates was investigated. Fuller et al. reported that S.iniae virulence is associated with distinct genetic profile and demonstrated differences between pathogenic and nonpathogenic isolates [10]. The isolates belong to genotype 1 in rep-PCR analysis showed a high virulence in the flounder, while the isolated belonging to genotype 2 were relatively low in virulence [9]. Similarly, among 14 isolates of S.iniae with six clonal patterns, two clones have one fold lower in pathogenicity challenge than others [11]. 11 S.iniae isolates from diseased wild and farmed fish showed significant differences in virulence and persistence, with a certain correlation to genogroup [24]. In this study, the pathogenicity of the two strains were compared by intraperitoneal injection in T.ovatus. Similar results were obtained in our analyses, where the virulence of genotype 1 strain BH15-2 showed much greater virulence than that of genotype 2 strain BH16-24.

Genetic variability not only depends on point mutations but also largely on horizontal genes transfer and intra-genomic rearrangements, which may disrupt chromosome organization [25]. During cell division, the symmetry of the origin and terminus loci play a role in the precise choreography of replicated chromosome separation [26]. Genome arrangement may affect gene expression and is thought to be related to the diversity of phenotypes seen in organisms [27]. Studies have shown that after the common ancestor branch of Streptococcus, chromosome reversal across the replication axis often occurs in a single streptococcal species [28]. A previously study demonstrated that unbalanced genome is prone to generate DNA rearrangements in the M3 strain of S.pyogenes, which is caused by the loss or acquisition of phages [29]. Moreover, results from a previous study suggest that a large-scale genomic rearrangement may resulted in biological discrepancies between a native avirulent and highly virulent S.suis strains [30]. Compared with its parental strain GX005, YM011 had a 0.4 M large inversion fragment which may result in abnormal expression of some genes including drug resistance genes and virulence factors, eventually leads to virulence attenuation [31]. By comparing the whole genomes of the two strains with other S.iniae which were also isolated from diseases fish, a large-scale inverted fragment about 1.28 Mbp was found as the major difference in BH16-24. The genomic recombination in BH16-24 leads to abnormal expression of some resistance and virulence genes in the fragment, eventually leads to multidrug resistance and virulence attenuation. Thus, our findings are consistent with the notion that inversion events of intergenic regions correlate to phenotypic variation [32].

β-lactam antibiotics are commonly being used in aquaculture in some parts of the world to treat bacterial infections [33]. Penicillin-binding proteins (PBPs) are membrane proteins involved in the biosynthesis of peptidoglycans in bacterial cell walls. The β-lactam antibiotics participate in the synthesis of peptidoglycans by binding to the active site of PBPs, thereby disrupting the formation of normal cell walls and inducing cell death through bacteriolysis [34]. The major PBPs causing for the β-lactam antibiotics resistance of S. pneumoniae are PBP2x, PBP2b and PBP1a [35]. In our study, BH16-24 was resistant to the two β-lactam antibiotics i.e., penicillin, ceftazidime, and intermediate resistant toward the other three β-lactam antibiotics used in this study, i.e., ampicillin, cefazolin, ceftriaxone sodium, whereas BH15-2 was susceptibility to these antibiotics. Pbp1a, pbp2x, and pbp2b gene of BH16-24 is located on an inverted fragment, and its abnormal expression changes the amount of PBP protein, which may lead to BH16-24 resistance to β-lactam antibiotics. In addition, the present of the RNA polymerase beta subunit (rpoB) with mutations have been reported in rifamycin resistance in Brucella melitensis, Mycobacterium tuberculosis, and other microorganisms [36, 37]. The DNA gyrase (gyrA) have been suggested to be involved in fluoroquinolone resistant isolates of S.agalatiae [38]. The dihydrofolate reductase enzyme encoded by the dfr gene promotes bacterial resistance to trimethoprim [39, 40]. Aminoglycoside antibiotics are transported by the oligopeptide transport system, thus when the gene for oppA protein was deleted, sensitivity to aminoglycoside antibiotics was greatly decreased [41]. Aminoglycoside phosphotransferases which encoded by aph gene are bacterial enzymes responsible for the inactivation of aminoglycoside antibiotics by O-phosphorylation [42]. In this study, genes such as gryA, gryB, rpoB, dfrA, oppA, aph were in the inverted fragment, which may be related to the resistance of BH16-24 to sulfonamide antibiotics, aminoglycoside antibiotic, rifampin. Moreover, the efflux pumps play an important role in conferring resistance by actively excreting the harmful antibiotic drugs from the bacteria [43]. Efflux pumps exists in almost all bacterial species. It can not only excrete a wide range of antibiotics, but also reduce intracellular antibiotic concentration and promote mutation accumulation [43, 44]. Here, we identified several efflux-pump related genes from different efflux-pump families among the genomes. The abnormal expression of these resistance genes on the inverted fragment may lead to multiple drug resistance of BH16-24.

To date, several virulence-associated factors (VAFs) have been characterized which are closely related to the pathogenesis of S.iniae infection. The most critical VAF validated in S.iniae including polysaccharide capsular (cps), phosphoglucomutase (pgmA), M-like protein (simA), beta-hemolysin/cytolysin (cyl), C5a peptidase (scp). CPS of S.iniae is a major virulence factor that provides resistance to the bactericidal activity of phagocytes and stimulates prolonged inflammatory responses, which including cpsA, cpsB, cpsC, cpsD, cpsE, cpsG, cpsJ [45, 46]. The streptococcal CpsA protein was reported to associated with important virulence determinants, including cell wall processing [47], polysaccharide synthesis [48], and reaction to antimicrobial stress [49]. The absence of cpsD could reduce the ability of S.iniae to survive phagocytosis and escape the immune system [50]. CpsJ influences the synthesis of CPS and loss of this protein showed lower virulence in a channel catfish infection model [51]. The pgm gene play an pivotal role in normal cell wall morphology, surface capsule expression, resistance to innate immune clearance mechanisms, therefore it is necessary for the virulence in S.iniae [52]. M-like protein contributes to cellular adherence and invasion and provides resistance to phagocytic killing based in vitro cell analysis [53]. The virulence factor Beta-hemolysin/cytolysin is encoded by cyl gene, capable of exerting cytolytic, proapoptotic, proinvasive, proinflammatory, or antiphagocytic effects on a variety of target cells [54]. In our study, both genomes harbor 336 putative genes involved in virulence, among which 221 virulence genes are located on the inversion fragment. Among the 221 virulence genes, we observed genes encoding CPS, phosphoglucomutase, M-like protein, beta-hemolysin/cytolysin, C5a peptidase. The abnormal expression of these virulence-related genes on the inverted fragments contributed to the reduced virulence of BH16-24.

Prophage can enhance bacterial adherence to animal cells, encode a series of bacterial toxins, and affect bacterial biofilm formation, which is closely related to bacterial virulence [55,56,57]. As reported by Wang et al., the phage was the major reason of causing different levels of virulence between S.agalactiae strains [58]. Prophage analysis of BH15-2 and BH16-24 showed that both strains harbored 11 identical incomplete and questionable prophages, although several of them are in different positions due to the inversion fragment. Moreover, there is a 46 kb length, intact phage sequence which only existed in the BH15-2 genome. The GC contents of the prophage fragment (35.46%) deviate from the host genomes (36.80%). The prophage encoded 77 proteins, and 17 genes encoded hypothetical proteins, the other 60 genes encoded phage hit proteins such as phage lysin, phage tail protein, and phage integrase. However, the function of the prophage in BH15-2 is still unknown, and whether it is related to its virulence needs further experiments to verify.

Conclusions

In summary, we compared the biological characteristics such as growth, virulence, drug resistance, and whole genome sequence of two different genotypes of S.iniae. Comparative genomic studies of BH15-2 and BH16-24 showed that the main difference is a 1.28 Mbp inversion fragment. The inversion fragment may lead to abnormal expression of drug resistant and virulence genes, which is believed to be the main reason for the multiple resistance and weakened virulence of BH16-24. Aside from the differences in genomic rearrangement, BH15-2 harbored a novel intact prophage which is absented in BH16-24. There was finished concordance between genotypic evidence and biological characteristics. Further research is needed on how the genomic rearrangements affect the gene expression, drug resistance, and pathogenicity of S.iniae.

Materials and methods

Bacterial strains

The S.iniae strain BH15-2 and BH16-24 were originally isolated from the livers of moribund cultured golden pompano on two separate farms of China in 2015 ang 2016, respectively [16]. The fish from the two outbreaks displayed similar clinical signals, and the cumulative mortality rate was approximately 20–30%. Briefly, the stored strains were removed from − 80 ℃ refrigerator and streaked onto the BHI plate, cultured at 30 ℃ for 24–48 h. Then picked up a single colony and inoculated into 10 ml of BHI medium, cultivated at 30 ℃ by shaking.

Growth analysis

To measure the growth level of bacteria in BHI broth, overnight cultures of BH15-2 and BH16-24 were inoculated into BHI with an initial OD600 of 0.01 in a ratio of 1:50, respectively. The cultures were collected every 2 h of intervals and the optical density was measured at 600 nm from 0 to 18 h of growth at 30 ℃ with shaking in 180 r/min. Data were expressed as mean ± standard deviation (SD). Statistical analyses were performed using Student’s t-test using SPSS 21.0 and p < 0.05 was considered significant.

Comparison of the virulence between BH15-2 and BH16-24

Adult golden pompano with a mean weight of 50 g, were purchased from a local fish farm and maintained in a 16 m3 tank with aeration and sand-filtered seawater supply. The fish (n = 150) were acclimated for 2 weeks at 28–30 ℃ and checked randomly to confirm that no bacterial infected. Fish were fed twice daily with commercial fish expanded pellets (Guangdong Yuehai Feed Group), and waste was removed daily. The bacterial concentration determined by plating 10-fold serial dilutions onto BHI agar plates. Suspensions from 1.5 × 108 to 1.5 × 105 CFU/ml were prepared by serial 10-fold dilution. Fish were divided into nine groups with 16 fish per group. Before experimental treatment or organ extraction, fish were euthanized in 100 mg/L MS-222 (Sigma, USA). Eight groups were injected intraperitoneally (i. p.) with 0.1 ml of diluted bacterial cell suspension of the strain BH15-2 and BH16-24 at the final concentration of 1.5 × 108, 1.5 × 107, 1.5 × 106 and 1.5 × 105 CFU/ml, respectively. The control group were i. p. with the same amount of sterilized saline. The mortalities were recorded every 24 h interval for 28 days post-infection. The bacteria were reisolated from the brain, kidney and spleen tissues of all dead fishes at the end of the experiment and identified.

Comparison of the antibiogram between BH15-2 and BH16-24

The antibiogram study of the bacterium was determined on BHI plates according to the disc diffusion method, and the diameters of the inhibition zones were measured using Vernier calipers. The tested antibiotic impregnated discs were summarized in Table 2. Resistant, intermediate, and susceptible phenotype determinations were based on manufacturer guidelines (Hangzhou Binhe Microorganism Reagent Co., Ltd., China). Multiple antibiotic resistance (MAR) index of the two strains against the tested antibiotics was calculated by following the procedure described by Krumperman [59].

Genome sequencing and annotation

The genomes of BH15-2 and BH16-24 were sequenced by PacBio sequencing at the Beijing Biomarker Bioinformatics Technology Co., Ltd. For genome assembly, the filtered subreads were assembled by Canu v1.5 software [60], and then circlator v1.5.5 was taken to cyclizing assembly genome [61].

Genome component prediction

For genome component prediction, coding genes prediction was performed by Prodigal v2.6.3 [62]. The GenBlastA v1.0.4 program was used to scan the whole genomes after masking predicted functional genes [63]. Putative candidates were then analyzed by searching for non-mature mutations and frame-shift mutations using GeneWise v2.2.0 [64]. Transfer RNA (tRNA) genes were predicted with tRNAscan-SE v2.0 [65], Ribosome RNA (rRNA) genes were predicted with Infernal v1.1.3 [66]. Repetitive sequences were predicted using RepeatMasker v4.0.5 [67]. CRT v1.2 was used for CRISPR identification [68]. Circos v0.66 was used to draw genomic circles [69].

Gene functions

For functional annotation, the predicted proteins were blast (e-value: 1e− 5) against Nr (Non-Redundant Protein Database databases), Swiss-Prot, TrEMBL, KEGG (Kyoto Encyclopedia of Genes and Genomes), eggNOG, GO (Gene ontology). The pathogenicity and drug resistance of pathogenic bacteria were analyzed using VFDB and ARDB (Antibiotic Resistance Genes Database).

Comparative genomics analysis

Genomic synteny was analyzed using Mauve v2.3.1 [70]. GIs of BH15-2 and BH16-24 were determined with IslandViewer 4 [71]. PHASTER was used to identify prophage sequences [72]. Prophage Hunger was used to predict the intact prophage [73].