The Attenuation Mechanism and Live Vaccine Potential of a Low-Virulence Edwardsiella ictaluri Strain Obtained by Rifampicin Passaging Culture

Abstract

The rifampicin-resistant strain E9-302 of Edwardsiella ictaluri strain 669 (WT) was generated by continuous passage on BHI agar plates containing increasing concentrations of rifampicin. E9-302 was attenuated significantly by 119 times to zebrafish Danio rerio compared to WT in terms of the 50% lethal dose (LD₅₀). Zebrafish vaccinated with E9-302 via intraperitoneal (IP) injection at a dose of 1 × 10³ CFU/fish had relative percentage survival (RPS) rates of 85.7% when challenged with wild-type E. ictaluri via IP 14 days post-vaccination (dpv). After 14 days of primary vaccination with E9-302 via immersion (IM) at a dose of 4 × 10⁷ CFU/ml, a booster IM vaccination with E9-302 at a dose of 2 × 10⁷ CFU/ml exhibited 65.2% RPS against challenge with wild-type E. ictaluri via IP 7 days later. These results indicated that the rifampicin-resistant attenuated strain E9-302 had potential as a live vaccine against E. ictaluri infection. A previously unreported amino acid site change at position 142 of the RNA polymerase (RNAP) β subunit encoded by the gene rpoB associated with rifampicin resistance was identified. Analysis of the whole-genome sequencing results revealed multiple missense mutations in the virulence-related genes esrB and sspH2 in E9-302 compared with WT, and a 189 bp mismatch in one gene, whose coding product was highly homologous to glycosyltransferase family 39 protein. This study preliminarily explored the molecular mechanism underlying the virulence attenuation of rifampicin-resistant strain E9-302 and provided a new target for the subsequent study of the pathogenic mechanism of E. ictaluri.

Introduction

Enteric septicemia of catfish (ESC) is a common disease of cultured channel catfish Ictalurus punctatus. At 22-28°C, ESC is acute septicemia with rapid development, and death usually occurs four days after infection [1]. Therefore, ESC inflicts considerable economic damage on the channel catfish aquaculture industry each year. The main bacterial pathogen causing ESC is Edwardsiella ictaluri, a facultative, intracellular gram-negative bacterium belonging to the Enterobacteriaceae and Edwardsiella genus, and a common bacterial pathogen in fish. E. ictaluri mainly infects Siluriformes, such as channel catfish Ictalurus punctatus [2, 3], Brazilian catfish Pseudoplatystoma corruscans [4], and yellow catfish Pelteobagrus fulvidraco [5, 6]. However, there have also been reports on mortality in non-catfish species, including Nile tilapia Oreochromis niloticus [7], rainbow trout Oncorhynchus mykiss [8], riverine ayu Plecoglossus altivelis [9], and zebrafish Danio rerio [10]. Especially in China, the frequent occurrence of bacterial diseases due mainly to E. ictaluri has led to serious economic losses in aquaculture and caused massive mortality of yellow catfish accompanied by splitting head (hole-in-the-head) disease [5], severe ascites, and enteric septicemia.

Although antimicrobial agents are still a common means of preventing and controlling outbreaks of bacterial fish diseases like edwardsiellosis, they have little effect as sick fish have a decreased feeding rate. In addition, the intensive use of antibiotics in aquaculture applies a selective pressure that creates reservoirs of drug-resistant bacteria [11] and pollutes the aquaculture environment. Therefore, immunoprophylaxis is the way forward for future drug development as new approaches are urgently needed to replace antibiotics to control explosive epidemic bacterial diseases. For this purpose, vaccination, which stimulates the immune system, can be an effective, alternative prevention strategy [12].

Studies have made it clear that mutagenesis using rifampicin is one of the most successful chemical mutagenesis strategies against gram-negative bacteria. Furthermore, the potential of live attenuated vaccines against fish pathogens based on rifampicin resistance has been demonstrated. Rifampicin is a broad-spectrum antibiotic drug belonging to the rifamycin family. Its action site is the β subunit of bacterial DNA-dependent RNA polymerase, which can inhibit bacterial DNA transcription at the initiation stage of transcription, so as to achieve sterilization [13]. In 1999, Klesius and Shoemaker successfully developed a live attenuated vaccine of E. ictaluri, which was created by multiple passages of a wild-type strain isolated from diseased channel catfish on increasing concentrations of rifampicin [1]. This vaccine was later patented [14]. In addition to E. ictaluri, the strategy based on the ability of rifampicin to induce rough mutants was successfully applied to Flavobacterium columnare [15] and Brucella abortus [16].

Antibiotic resistance due to chromosomal mutations is often accompanied by a fitness cost as antibiotic-resistant strains show reduced growth rates, decreased competitiveness, and attenuated virulence compared with the phenotype of antibiotic-sensitive parental wild strains [17]. Previous studies have shown that most rifampicin-resistant fish pathogens exhibit decreased virulence [18-20], but there has been no study on the correlation between fitness cost and rifampicin resistance. Some researchers believe that the fitness cost of resistance is the most important factor promoting the evolution of resistance in pathogen populations [21]. Hence, studying fitness cost can help in understanding the evolution of antibiotic resistance in bacteria.

In our study we sought to utilize this strategy on the highly pathogenic E. ictaluri isolated from diseased yellow catfish. The specific objectives were to: (i) generate rifampicin-resistant strains of E. ictaluri isolated from yellow catfish and screen out an attenuated strain using different kinds of fish, (ii) compare the phenotypic differences between the parent strain and the resistant strain to explore fitness cost of the resistant one, (iii) determine the virulence of the resistant strain in zebrafish, (iv) determine if immunization of zebrafish with the rifampicin-resistant strain confers a protective immune response against wild-type E. ictaluri challenge, and (v) investigate the possible mechanism of attenuation and explore potential virulence factors of E. ictaluri based on whole-genome sequencing analysis.

Materials and Methods

Bacterial Strains and Culture Conditions

Table S1 shows the name and origin of all strains used in our study, which were confirmed to be E. ictaluri isolates by sequencing. Bacteria strains were stored at -80°C as glycerol stocks and cultured on Brain Heart Infusion (BHI) agar plates or BHI broth with shaking (200 ×g) at 28°C.

Diversity Analysis of E. ictaluri

Ten housekeeping genes (adk, atpD, dnaJ, gapA, glnA, Y-hsp60, phoR, pyrG, rpoA, tuf) were selected as targets for PCR assays. The housekeeping genes and the primers used for multilocus sequence typing (MLST) according to the public databases for MLST of Edwardsiella (Edwardsiella spp. | PubMLST) are shown in Table S2 (slightly modified). The above E. ictaluri isolates from different origins (Table S1) were expanded to the logarithmic growth phase. Genomic DNA was extracted from logarithmic-phase bacterial cultures using a Bacterial DNA Kit (China) according to the manufacturer’s instructions. The concentration and quality of extracted genomic DNA was measured using a NanoDrop 1000 spectrophotometer (Thermo Fisher Scientific, Inc., USA), and DNA was stored at -20°C until used as a template for amplification.

PCR amplification was carried out in a 50 μl reaction mixture containing 19 μl sterile ultrapure water, 2 μl of each pair of specific upstream and downstream primers, 2 μl of template DNA, and 25 μl of 2×Es Taq MasterMix (containing Es Taq DNA Polymerase, 3 mM MgCl₂ and 400 μM each dNTP) (Dye) (CWbio, China). The optimized amplification program used in a thermal cycler (iCycler, Bio-Rad Laboratories, USA) was initial denaturation at 94°C for 2 min, 35 cycles containing denaturation at 94°C for 30 s, annealing at 55-65°C for 30 s, and elongation at 72°C for 30 s followed by a final extension of 72°C for 2 min. The PCR products were detected by 1% agarose gel electrophoresis and purified using a FastPure Gel DNA Extraction Mini Kit (Vazyme, China). The purified products were ligated into vector pMD-18T and transformed into competent Escherichia coli DH5α cells. Positive clones were selected by coating ampicillin-resistant plates and sent out to Tsingke Biotechnology Co., Ltd., China, for bidirectional sequencing.

The sequences of ten housekeeping genes of each strain were submitted to the database, and the allelic numbers with the best match for each gene were obtained. The sequence type (ST) was obtained according to standard procedure. In addition, the allelic tandem sequences of E. ictaluri isolates with known STs (Table S3) were downloaded from the database. MEGA v.11.0 software was used to align the tandem sequences and construct the dendrogram, which was used to analyze the genetic and evolutionary relationships between the strains stored in our laboratory and the isolates from different hosts in different years in other regions. The best model was selected from 24 replacement models according to the Bayesian Information Criterion (BIC), and the Maximum Likelihood method of the Tamura 3-parameter model was used to construct the phylogenetic tree. The bootstrap test was used (1000 replicates) to test the quality of the tree.

Generation of Rifampicin-Resistant Strains by Subculture

Rifampicin (analytical purity, AR; Kehbio, China) was dissolved in methanol (AR) to a concentration of 5 mg/ml, which was sterilized with a sterile 0.22 μm syringe filter and stored at -20°C until use. The generation of rifampicin-resistant strains was based on an existing procedure [1] with minor modifications. Two virulent strains of E. ictaluri isolated from diseased yellow catfish, strain 668 and strain 669, were used as the parent strains to induce the development of rifampicin-resistant strains. The minimal inhibitory concentration (MIC) of rifampicin against strain 668 and strain 669 was 2 μg/ml, and the initial concentration of rifampicin that allowed the growth of E. ictaluri was 1.5 μg/ml. After 20 passages (the concentrations of rifampicin in BHI agar plates were 1.5, 3, 6, 8, 10, 12, 16, 20, 25, 30, 50, 70, 90, 120, 150, 180, 210, 240, 270, and 300 μg/ml, respectively) of E. ictaluri strain 668 or strain 669 in BHI agar containing progressive concentrations of rifampicin, and the rifampicin-resistant strains were able to grow in BHI agar containing 300 μg/ml. Strains were identified by sequencing and stored at -80°C until use.

Screening of an Attenuated Rifampicin-Resistant Strain

Fish

Blue gourami Trichogaster trichopterus and zebrafish were purchased from a local aquarium, and Nile tilapia were purchased from Yingshan Tilapia Breeding Base. Fish were temporarily raised for 2 weeks before the experimental operation, and 5% of fish were randomly selected for detection of E. ictaluri in the liver, kidney, and spleen. Only fish without detection of E. ictaluri could be used in subsequent experiments. The keeping and experimentation temperature regarding the fish was set at 26 ± 2°C.

Primary Screening of Attenuated Rifampicin-Resistant Strains

Blue gourami have often been used as an experimental model to study the mechanisms of bacterial pathogenesis and evaluate the efficacy of vaccines in previous studies [12, 22, 23]. Blue gourami (4 cm average length, 20 fish per group) were IP injected with the same high dose (5 × 10⁷ CFU/fish) of sterile phosphate-buffered saline (PBS)-resuspended strain 668 and strain 669 and their corresponding rifampicin-resistant strains, while the control group was IP injected with the same volume of sterile PBS. The strains with fewer deaths were selected as the attenuated rifampicin-resistant strains obtained from the primary screening.

Secondary Screening of Attenuated Rifampicin-Resistant Strains

Nile tilapia (6 cm average length, 20 fish per group) were IP injected with the same high dose (5 × 10⁷ CFU/fish) of sterile PBS-resuspended strains obtained from primary screening and their corresponding wild-type strains, while the control group was IP injected with the same volume of sterile PBS. The strain with the fewest deaths was selected as the final attenuated rifampicin-resistant strain and named E9-302.

Fitness Costs due to Resistant Mutations

Comparison of rpoB Sequence

Since rifampicin inhibits bacterial RNA synthesis by specifically targeting a small but highly conserved pocket in the RNA polymerase (RNAP) subunit encoded by rpoB [24], 95% of rifampicin resistance mutations are present in the rpoB [25]. Difference in rpoB sequence between wild-type strain and resistant strain was the most direct fitness cost associated with rifampicin resistance [26]. Comparison of rpoB was completed by subsequent whole-genome sequencing.

Cell Proliferation Profile of E9-302 Compared to WT

A slower growth rate has been reported as a fitness cost of antibiotic-resistant bacteria [27, 28]. Cell proliferation assays were performed according to published procedures [29] with slight modifications. Briefly, both WT and E9-302 were cultured overnight in BHI at 28°C, 200 rpm. The optical density (OD) of each strain at 570 nm was determined and adjusted to OD_{570 nm} = 1. Seven dilutions of two bacterial solutions (1:10, 1:20, 1:40, 1:80, 1:160, 1:320, 1:640) and a blank control of fresh BHI medium were set in sterile 96-well microtiter plates in triplicates per strain and incubated at 28°C after adding the bacterial solution in proportion to each dilution gradient. The OD of the 96-well plate measured at 570 nm soon after the addition of the bacterial solution was considered as the initial OD value for 0 h and the OD_{570 nm} was measured every hour. Relative increased OD value was calculated using the following formula: Increased OD_{570 nm} value = OD_{570 nm} value after incubation - OD_{570 nm} value at 0 h of the incubation. The increased OD_{570 nm} value after incubation was then plotted against incubation time.

In Vitro Competition Assay

The cost of drug-resistant mutations was reported to be determined by direct competition against drug-sensitive parental strains [30]. The in vitro competition assays were performed according to published procedures [30] with slight modifications: After adjusting the OD of WT and E9-302 at 600 nm to OD_{600 nm} = 1, an equal volume of each strain was mixed to form 1 ml of bacterial broth and transferred to antibiotic-free BHI medium at a ratio of 1:100 for incubation at 28°C for 14 h. Then, 1 ml of the culture was centrifuged at 2380 ×g for 5 min and resuspended in sterile PBS. The bacterial suspension was coated on antibiotic-free BHI agar plates and the number of colonies was counted as the total number of bacterial cells. The number of colonies obtained by counting the same volume of bacterial suspension coated on rifampicin-resistant BHI agar plates was taken as the number of E9-302 colonies, and the number of WT colonies = total number of bacterial cells - the number of E9-302 colonies. The assays were performed in triplicate with three independent cultures. Serial dilutions of each culture were plated and counted three times to take the mean value. The difference in fitness cost between two competing strains at time t was calculated using the following formula:

$$S_t=\ln \left[{\left(\frac{r_t/S_t}{r_{t-1}/S_{t-1}}\right)}^{\frac{1}{8}}\right]$$ (1)

Where r_t and S_t represent the absolute number of rifampicin-resistant strain E9-302 and rifampicin-sensitive strain WT at time t, respectively, r_t−1 and S_t−1 represent the absolute number of rifampicin-resistant strain E9-302 and rifampicin-sensitive strain WT at the previous time point, respectively. S_t is called the selection coefficient at time t. r_t/r_t−1 and S_t/S_t−1 represent the growth rate of rifampicin-resistant strain E9-302 and rifampicin-sensitive strain WT, respectively. Hence, S can represent the natural logarithm of the quotient of the growth rate of competing strains.

According to the research, S is positive if rifampicin resistance increases bacterial fitness, S is equal to 0 if there is no difference in fitness between two competing strains, and S is negative if rifampicin resistance incurs fitness cost.

Virulence Determined in Zebrafish

Zebrafish were used to determine the 50% lethal doses (LD₅₀) of the WT and E9-302. Zebrafish (length, 3 ± 0.2 cm) were fed a basal diet and temporarily kept for 7 days before the experiment. Groups of 20 fish were IP injected using microinjectors (Hamilton, Switzerland) of a 50 μl range with 5 μl of WT and E9-302 at different concentration gradients (10² ~ 10⁷ CFU/fish) with three parallel groups. The blank control was IP injected with the same volume of sterile PBS. Survival of zebrafish was observed and recorded for 7 days, and LD₅₀ was estimated with the Spearman-Kärber method [31] modified from Dias [32].

Vaccination Trials in Zebrafish

Zebrafish (length, 3 ± 0.2 cm) were divided into 12 parallel experimental groups of 20 fish each. In groups 1-3 and groups 4-6, fish were inoculated with E9-302 or PBS via IP injection, respectively. In groups 7-9 and groups 10-12, fish were inoculated with E9-302 or PBS via immersion (IM). For the IP injection group, the inoculation dose was 1 × 10³ CFU/fish, that was the experimental fish were IP injected using microinjectors as mentioned before with 5 μl of 2 × 10⁵ CFU/ml E9-302, while the control fish were IP injected with the same volume of sterile PBS. At 14 days post-vaccination, the vaccinated fish from each group (30 fish) were challenged via IP injection with 1 × 10⁴ CFU/fish of parental E. ictaluri WT. For the IM group, zebrafish were immersed in 4 L of sterile PBS resuspension of E9-302 with a concentration of about 4 × 10⁷ CFU/ml for 40 min, while the control fish were immersed in sterile PBS with the same volume for the same length of time, and continuous aeration was maintained in all groups. A booster vaccination via IM was conducted upon 14 days after primary IM. The experimental settings were consistent with the primary IM except for the IM concentration of 2 × 10⁷ CFU/ml. Thirty fish from each treatment group (groups 1-3, groups 4-6, groups 7-9, and groups 10-12) were randomly selected and challenged with 1 × 10⁴ CFU/fish of parental E. ictaluri WT upon 7 days after booster vaccination. The mortality of each group was recorded for both IP infection group and IM group after challenge, and the results were expressed as relative percentage survival (RPS) as previously described [33]. RPS was computed according to the following formula:

$$\text{RPS = [1-(vaccinated mortality/control mortality)]}\times 100\%.$$ (2)

Whole-Genome Sequencing

The bacterial whole-genome sequencings of WT and E9-302 were completed by a combination of next-generation sequencing and SMRT, i.e., Illumina Hiseq sequencing combined with single-molecule PacBio sequencing at the Majorbio Biotech Co., Ltd. (China). Briefly, next-generation sequencing started with the purification of genomic DNA, genomic DNA fragmentation using Covaris to construct a genomic sequencing library, and Bridge PCR followed by IlluminaHiseq sequencing. As for SMRT, the purified genomic DNA was fragmented using the G-tube method to construct the SMRT Bell library, which was then annealed and bound to the bottom polymerase of zero-mode waveguides (ZMW). The libraries were quantified by Qubit and the inserted fragment size was measured by Agilent 2100, and finally sequenced using the PacBio platform. The differences in genes related to virulence between WT and E9-302 were compared and analyzed to examine the underlying genomic basis of attenuation.

Results

Diversity Analysis of E. ictaluri

To produce an MLST assay we used ten housekeeping genes (Table S2), which were bidirectionally sequenced and then submitted to the database to obtain the allelic numbers with the best match for each gene. The sequence type (ST) of each strain was computed according to the standard procedure provided by the database (Table 1). The results showed that the STs of E. ictaluri isolated from different origins were consistent, as all were ST26, showing the high degree of genetic homogeneity among E. ictaluri in China. Accordingly, the results also made it possible to develop widely applicable vaccines against edwardsiellosis in China.

Table 1.

STs of E. ictaluri isolates in our study.

Strain	Allelic numbers with the best match for each gene										ST
	adk	atpD	dnaJ	gapA	glnA	Y-hsp60	phoR	pyrG	rpoA	tuf
656	9	10	10	10	8	2	12	12	6	7	26
658	9	10	10	10	8	11	12	12	6	7	26
663	9	10	10	10	8	1	12	12	6	9	26
665	9	10	10	10	8	11	12	12	6	7	26
668	9	10	10	10	8	11	12	12	6	7	26
669	9	10	10	10	8	11	12	12	6	9	26

The phylogenetic tree of fifteen E. ictaluri isolates (Tables S1 and S3) was constructed using MEGA v.11.0 software based on tandem sequence (adk- atpD- dnaJ- gapA- glnA- Y-hsp60- phoR- pyrG- rpoA- tuf) with the Maximum Likelihood method and Tamura 3-parameter model (Fig. 1). The genetic correlations among E. ictaluri isolates investigated in this study showed that the isolates were largely clustered according to sequence type (ST). The sequence type of the E. ictaluri prevalent in China was mainly ST26, and isolates in China were distantly related to those from other countries and regions.

Fig. 1. Phylogenetic tree based on tandem sequence (adk- atpD- dnaJ- gapA- glnA- Y-hsp60- phoR- pyrGrpoA- tuf) showing relationships of different STs of fifteen E. ictaluri isolates in this study.

Screening and Obtaining of an Attenuated Vaccine Candidate

The strain E9-302 was a rifampicin-resistant mutant, which was produced by the highly pathogenic E. ictaluri strain 669 (WT) isolated from yellow catfish by continuous passages on BHI agar plates containing increasing concentrations of rifampicin. Blue gourami and Nile tilapia as the experimental subjects were IP injected with the same dose of high-concentration sterile PBS resuspension of each alternative strain to perform virulence screening. After two rounds of screening, we selected the strain with the fewest deaths as the attenuated rifampicin-resistant vaccine candidate and named it E9-302. Figs. 2 and 3 showed survival curves of WT and E9-302 in two rounds of virulence screening, respectively.

Fig. 2. Survival curve of blue gourami IP injected with equal high dose (5 × 10⁷ CFU/fish) of WT and E9-302.

Fig. 3. Survival curve of Nile tilapia IP injected with equal high dose (5 × 10⁷ CFU/fish) of WT and E9-302.

Fitness Cost Analysis of E9-302

Sequencing Results of rpoB

The rpoB sequences of WT and E9-302 were both 4,029 bp in length, coding for 1,342 amino acids. Sequence alignment between rpoB of WT and that of E9-302 revealed the three missense mutations (Table 2).

Table 2.

List of mutations between rpoB of WT and that of E9-302.

Codon in WT	Codon in E9-302	Nucleotide (bp)		Deduced amino acid change
		From	To	From	To
GAG	GGG	424	426	Glutamic acid/E	Glycine/G
AGC	CGC	1525	1527	Serine/S	Arginine/R
CTG	CGG	1531	1533	Leucine/L	Arginine/R

Cell Proliferation Profile of E9-302 Compared to WT

When the initial amount of bacteria was either at 1:10 or 1:40, the relative increase of the OD_{570 nm} values of WT from 0 h to 6 h was higher than that of E9-302, but not significant (Fig. 4). Similar patterns were observed as well when the beginning amount of both bacteria in each well was at other dilutions (Fig. S1).

Fig. 4. Cell proliferation rate of E. ictaluri E9-302 compared to that of E. ictaluri WT.

In Vitro Competitiveness of E9-302 Compared to WT

The initial bacteria amount of WT and E9-302, the bacteria amount after 14 h of incubation, and the S value were shown in Table 3. The S value, which can be interpreted as the natural logarithm of the quotient of growth rates of competing strains, was used to judge fitness differences. Data were shown as the mean of three independent parallel experiments. As shown in Table 3, the fitness difference S < 0 indicated that resistance to rifampicin reduced the fitness of bacteria.

Table 3.

Comparison of in vitro competitiveness between E9-302 and WT.

Strain	Initial bacteria amount (CFU ml-1)	Bacteria amount after incubation (CFU ml-1)	S
WT	3.23 × 108	6.95 × 107	-0.05
E9-302	2.40 × 108	3.46 × 107

Assessment of Virulence

Zebrafish have been developed as a powerful laboratory model to study the pathogenesis of bacterial fish pathogens [34-36], including E. ictaluri [37-39]. As shown in Table 4, E. ictaluri E9-302 was attenuated with an LD₅₀ of 1.04 × 10⁵ CFU/fish, a 119-fold increase over the parental strain WT, indicating that the virulence of E9-302 was significantly weakened. The dying zebrafish exhibited clinical symptoms associated with edwardsiellosis such as body spinning, head up and tail down in the water, blood spots on the body surface, and E. ictaluri was detected in internal organs.

Table 4.

LD₅₀ of E9-302 and WT.

Dose of challenge (CFU fish-1)	Number of death/ Total		Cumulative mortality (%)
	WT	E9-302	WT	E9-302
1.0 × 107	20/20	17/20	100%	85%
1.0 × 106	20/20	14/20	100%	70%
1.0 × 105	18/20	9/20	90%	45%
1.0 × 104	15/20	6/20	75%	30%
1.0 × 103	11/20	2/20	55%	10%
1.0 × 102	5/20	0/20	25%	0%
LD50a	8.73 × 102	1.04 × 105	—	—

^aZebrafish (length, 3 ± 0.2 cm) were IP injected with WT and E9-302 at different concentration gradient (10² ~ 10⁷ CFU fish^-1) with three parallel. Data were presented as mean from triplicates. LD₅₀ was computed with the Spearman–Kärber method [31] modified from Dias [32].

Immunoprotective Effects of E9-302 in Zebrafish

The zebrafish were vaccinated with E9-302 via IP injection or IM and challenged with 1 × 10⁴ CFU/fish of wild-type E. ictaluri strain WT at 14 days after injection or 7 days after booster IM. When zebrafish vaccinated E9-302 via IP injection were challenged with WT at 14 dpv, the RPS of vaccinated fish was 85.7% (Table 5), and when zebrafish vaccinated E9-302 via IM with booster IM were challenged with WT at 7 days after booster IM, the RPS of vaccinated fish was 65.2% (Table 5), indicating E9-302 provided protection against the parental E. ictaluri strain WT, and thereby showing its potential as a vaccine against edwardsiellosis.

Table 5.

RPS results for E9-302 with different vaccination strategies in zebrafish.

Group	Vaccination dose of IPa (CFU fish-1)	Cumulative mortality of IPb (%)	RPS of IPc (%)	Primary vaccination dose of IMd (CFU ml-1)	Booster vaccination dose of IMe (CFU ml-1)	Cumulative mortality of IMf (%)	RPS of IMg (%)
PBS control	Nh	70	N	N	N	76.7	N
E9-302	1 × 103	10	85.7	4 × 107	2 × 107	26.7	65.2

^aThe fish in each group (20 fish group^-1) was vaccinated with E9-302 at a dose of 1 × 10³ CFU/fish via IP injection.

^bZebrafish in the immunized group and control group were challenged with 1 × 10⁴ CFU/fish E. ictaluri strain WT at 14 days after IP injection and the cumulative mortality of each group was recorded 12 days after challenge.

^cZebrafish in the immunized group and control group were challenged with 1 × 10⁴ CFU/fish E. ictaluri strain WT at 14 days after IP injection and the RPS was calculated according to the formula exhibited in 2.7.

^dThe fish in each group (20 fish per group) was immersed in sterile PBS resuspension of E9-302 at a concentration of 4 × 10⁷ CFU/ml for 40 min with continuous aeration, which was considered primary vaccination via immersion (IM).

^eThe fish in each group (20 fish per group) were re-immersed in sterile PBS resuspension of E9-302 at a concentration of 2 × 10⁷ CFU/ml for 40 min with continuous aeration at 14 days after primary IM, which was considered booster vaccination via IM.

^fZebrafish in the immunized group and control group were challenged with 1 × 10⁴ CFU/fish E. ictaluri strain WT at 7 days after booster IM and the cumulative mortality of each group was recorded 12 days after challenge.

^gZebrafish in the immunized group and control group were challenged with 1 × 10⁴ CFU/fish E. ictaluri strain WT at 7 days after booster IM and the RPS was calculated according to the formula exhibited in Materials and Methods.

^hNot assayed or not applicable.

Genomic Analysis of E9-302 Compared to WT

The whole-genome sequencing comparison showed that in addition to the differences in the gene rpoB as mentioned before, the genes esrB, sspH2, and a gene of unknown function (named Gt39-like in this paper), were also mutated. There was one missense mutation in esrB, a 27 bp sequence insertion and multiple missense mutations in sspH2, and a 189 bp mismatch in gene Gt39-like (Table 6). The gene esrB encodes EsrB, a response regulatory (RR) element in the two-component EsrA-EsrB system, and the gene sspH2 encodes the E3 ubiquitin ligase SspH2.

Table 6.

List of genetic differences between E9-302 and WT.

Gene	Codon in WT	Codon in RE	Nucleotide (bp)a		Deduced amino acid change
			From	To	From	To
esrB	CTG	CGG	568	570	Leucine/L	Arginine/R
sspH2	ACG	ACA	1759	1761	Threonine/T	Threonine/T
	GCA	GCG	1765	1767	Alanine/A	Alanine/A
	AGT	AGC	1783	1785	Serine/S	Serine/S
	GTC	ATC	1795	1797	Valine/V	Isoleucine/I
	CAC	AAC	1825	1827	Histidine/H	Asparagine/N
	CCG	CAG	1834	1836	Proline/P	Glutamine/Q
	CTC	TTA	1840	1842	Leucine/L	Leucine/L
	TCA	TCG	1846	1848	Serine/S	Serine/S
	GCG	GTA	1849	1851	Alanine/A	Valine/V
	GGG	AGT	1852	1854	Glycine/G	Serine/S
	CTA	TTG	1870	1872	Leucine/L	Leucine/L
	CTA	CTG	1879	1881	Leucine/L	Leucine/L
	GAT	GAG	1882	1884	Aspartic acid/D	Glutamic acid/E
	CTA	CTG	1885	1887	Leucine/L	Leucine/L
	GTA	ATA	1903	1905	Valine/V	Isoleucine/I
	AAG	ACG	1909	1911	Lysine/K	Threonine/T
	TTT	GAT	1918	1920	Phenylalanine/F	Aspartic acid/D
	−b	GA∼AAb	-	-	-	E∼K
	TTC	GTC	1921	1923	Phenylalanine/F	Valine/V
	GAA	GAG	1927	1929	Glutamic acid/E	Glutamic acid/E
	GTC	GTT	1936	1938	Valine/V	Valine/V
	CTT	CTG	1942	1944	Leucine/L	Leucine/L
	CAG	CAA	1951	1953	Glutamine/Q	Glutamine/Q
	GAA	AAA	1966	1968	Glutamic acid/E	Lysine/K
	GAA	GAG	1975	1977	Glutamic acid/E	Glutamic acid/E
	CTA	CTG	1978	1980	Leucine/L	Leucine/L
	ACC	AGC	1981	1983	Threonine/T	Serine/S
	GGT	GGA	1984	1986	Glycine/G	Glycine/G
	ACT	GCC	1987	1989	Threonine/T	Alanine/A
	GGA	GAG	1996	1998	Glycine/G	Glutamic acid/E
	CAA	CGA	2002	2004	Glutamine/Q	Arginine/R
	GCA	GTA	2014	2016	Alanine/A	Valine/V
	AGA	CAC	2020	2022	Arginine/R	Histidine/H
	ACA	ACC	2026	2028	Threonine/T	Threonine/T
	TTA	TCA	2032	2034	Leucine/L	Serine/S
	AAA	CAA	2041	2043	Lysine/K	Glutamine/Q
	GCT	GCC	2047	2049	Alanine/A	Alanine/A
	GTG	GTA	2059	2061	Valine/V	Valine/V
	AGT	ACC	2065	2067	Serine/S	Threonine/T
	CAA	CAG	2080	2082	Glutamine/Q	Glutamine/Q
	AGG	GGG	2086	2088	Arginine/R	Glycine/G
	GAG	GAA	2107	2109	Glutamic acid/E	Glutamic acid/E
	CCA	CCG	2110	2112	Proline/P	Proline/P
	GGC	AAC	2119	2121	Glycine/G	Asparagine/N
	AAA	ATA	2128	2130	Lysine/K	Isoleucine/I
	AGA	AAA	2134	2136	Arginine/R	Lysine/K
	TCG	GCG	2140	2142	Serine/S	Alanine/A
	AAA	GAA	2143	2145	Lysine/K	Glutamic acid/E
	CAG	GAC	2146	2148	Glutamine/Q	Aspartic acid/D
	CAA	AAA	2152	2154	Glutamine/Q	Lysine/K
	ACG	GCG	2155	2157	Threonine/T	Alanine/A
	GAC	AAC	2164	2166	Aspartic acid/D	Asparagine/N
	ATG	ATT	2173	2175	Methionine/M	Isoleucine/I
	CAG	GCG	2200	2202	Glutamine/Q	Alanine/A
	TCT	TAT	2206	2208	Serine/S	Tyrosine/Y
	CTA	CTG	2218	2220	Leucine/L	Leucine/L
	AAA	GAA	2221	2223	Lysine/K	Glutamic acid/E
	CTC	GGC	2236	2238	Leucine/L	Glycine/G
	ATA	CTA	2245	2247	Isoleucine/I	Leucine/L
	GAG	GAA	2254	2256	Glutamic acid/E	Glutamic acid/E
	GGC	GGT	2266	2268	Glycine/G	Glycine/G
	GCT	GCG	2275	2277	Alanine/A	Alanine/A
	GAG	GAA	2290	2292	Glutamic acid/E	Glutamic acid/E
	AAG	AAA	2293	2295	Lysine/K	Lysine/K
	CAC	CGC	2314	2316	Histidine/H	Arginine/R
	CTA	CTG	2320	2322	Leucine/L	Leucine/L
	TTA	TTG	2332	2334	Leucine/L	Leucine/L
	CTT	CTG	2335	2337	Leucine/L	Leucine/L
	GGC	GCC	2338	2340	Glycine/G	Alanine/A
	AGG	GGG	2341	2343	Arginine/R	Glycine/G
Gt39-like	CAG	CA−c	700	702	Glutamine/Q	Mismatch

^aThe gene differential loci were subject to the loci in the WT sequencing results.

^bIn contrast to WT, there was an insertion of a 27 bp nucleotide sequence. GAGGTCTATGGAGGCGATGAAGATAAA in the sspH2 gene of the E9-302 sequence, which coded the inferred amino acid sequence EVYGGDEDK.

^cCompared with WT, E9-302 had a deletion of base G at gene locus 701 in the sequencing of the gene Gt39-like, resulting in a 189 bp sequence mismatch.

The symbol – indicated that there was no corresponding sequence, site information, or inferred amino acid change in the strain WT or E9-302 sequencing results.

The bold letters represented differences in amino acids inferred from different bases.

As for the gene Gt39-like with unknown function in WT, BLAST alignment showed that the amino acid sequence (488 AA) was highly homologous with the amino acid sequences of glycosyltransferase family 39 protein (GT39) of Salmonella enterica (NCBI Accession: WP_183051533.1), Yokenella regensburgei (NCBI accession: WP_040903585.1), and Klebsiella michiganensis (NCBI Accession: WP_004131974.1).

Discussion

Multilocus sequence typing (MLST) is a molecular typing method for bacteria based on nucleic acid sequence sequencing. Since it was first established and applied in 1998 for Neisseria meningitides [40], MLST has come to be widely used in other pathogens or environmental microorganisms due to its high sensitivity and specificity, and is now recognized as a more accurate and critical means to detect bacterial epidemiology, study evolution, and assess genetic diversity [41]. The typing results of the E. ictaluri isolates preserved in our laboratory were consistent, with all being classified as ST26. In addition, the plasmid profiles of E. ictaluri isolates from different provinces and cities in China stored in our laboratory all showed two plasmids (data not shown), which was consistent with the conclusion of Liu et al. [5]. These findings indicated a high degree of genetic homogeneity of E. ictaluri isolates in China and offered the possibility to develop efficient vaccines against edwardsiellosis with wide applicability in China. In this study, the strategy for exploiting rifampicin-resistant attenuated pathogens was applied to the highly pathogenic E. ictaluri isolated from diseased yellow catfish. After multiple rounds of screening, an attenuated rifampicin-resistant strain of E. ictaluri named E9-302 was successfully obtained, which was 150 times more resistant to rifampicin than its parental strain (WT). Meanwhile, the commercial ESC vaccine Aquavac-ESC (RE-33) was only 6.4-fold resistant to rifampicin compared to its parent strain EILO [1]. Virulence assessment suggested a significant decrease in virulence with a 119-fold reduction in the 50% lethal dose (LD₅₀) of E9-302 to zebrafish compared to highly pathogenic WT, which was in agreement with previous findings showing reduced virulence [18] of rifampicin-resistant pathogenic strains. The attenuated strain E9-302, inoculated into zebrafish in different ways, had a protective effect against challenge with the parental wild-type E. ictaluri strain WT, producing 85.7% (by IP route) and 65.2% (by IM plus booster IM route) RPS respectively, which was good evidence of the potential of E9-302 as a live attenuated vaccine against edwardsiellosis. Klesius et al. immersed the channel catfish in 1 × 10⁵ CFU/ml of RE-33 for 2 min, and the fish were challenged by immersion in 1 to 2 × 10⁷ CFU/ml of the parental strain (EILO) for 60 min upon 14 days post-vaccination, and the RPS of vaccinated fish was 51.7% [1]. However, since the fish species, immunization concentration, immunization time, and challenge mode were different, whether E9-302 will also provide higher protection to the same size of channel catfish by the same route merits further study.

Mutations in antibiotic resistance are often accompanied by virulence reduction and other fitness costs [33, 42, 43]. In clinical medicine, the fitness cost of antibiotic resistance is the basis for studying methods to limit the spread of drug-resistant mutants, which is helpful in researching the evolution of bacterial antibiotic resistance [21]. At present, no investigation is being undertaken on the fitness cost of rifampicin-resistant fish pathogens, and this is the first study on the correlation between fitness cost and rifampicin resistance. In addition to the decreased virulence, our results also showed that the cell proliferation rate of E9-302 was lower than that of WT (Fig. 4). Studies have shown that one of the fitness costs of antibiotic-resistant strains is reduced growth rate [27, 28], and this conclusion is also supported by the fact that the growth rate of rifampicin-resistant Streptococcus iniae strain D3-r25 was significantly lower than that of its parent S. iniae strain D-WT [44]; and the cell proliferation rate of novobiocin-resistant S. iniae strain ISNO was apparently lower than that of its virulent parental S. iniae strain ISET0901 [45]. The difference in fitness S calculated from in vitro competition experiments is less than zero, also indicating that resistance to rifampicin incurs a fitness cost. As the broad-spectrum antibiotic rifampicin specifically targets RNAP, changes in the gene encoding the drug target enzyme RpoB represent the most direct fitness cost of rifampicin resistance. The differences in the gene rpoB of rifampicin-resistant and sensitive strains of both Mycobacterium kansasii (three mutations in codons 513, 526, and 531) [46] and Escherichia coli (seventeen mutations in three distinct clusters in the center of the gene rpoB) [47] also validated this view. Here, we compared the rpoB sequences of E9-302 and WT on the basis of whole-genome sequencing and found a total of three missense mutations (Table 2). Among them, the 509th and 511th amino acid sites of RNAP β subunit encoded by the gene rpoB were in rifampicin resistance determination region (RRDR) cluster II, and the 511th amino acid residue was one of the amino acid residues directly associated with rifampicin binding, verifying the direct fitness cost from rifampicin resistance. Additionally, the 142nd amino acid with missense mutation has not been reported in the literature. The above results indicated that, in the absence of antibiotics, E9-302 was in an inferior position compared with WT in terms of growth rate and competitiveness. Changes in the gene rpoB on chromosome did confer fitness costs on antibiotic-resistant E9-302, suggesting that the growth and transmission of E9-302 might likewise be limited in the natural environment.

One of the common approaches to developing live attenuated vaccines is to obtain less virulent or even non-virulent mutants from highly pathogenic wild-type strains through multiple rounds of gradient increasing rifampicin concentration. As mentioned earlier, many successful cases have demonstrated that this approach is effective and feasible for generating attenuated mutants, but the mechanism leading to attenuation of virulence is still unclear. There may be various attenuation mechanisms of the same antibiotic on different bacteria. Here, we conducted whole-genome sequencing for WT and E9-302, aiming to preliminarily explore the molecular mechanism of rifampicin-attenuated E. ictaluri.

In our study, compared with WT, whole-genome sequencing analysis revealed a missense mutation in the E9-302 esrB gene, which encodes EsrB, a response regulatory element of the EsrA-EsrB two-component system. EsrA-EsrB of Edwardsiella is homologous to SpiR-SsrB/SsrA-SsrB of Salmonella. In Salmonella, the two-component system SsrA-SsrB is encoded by Pathogenicity Island 2 (SPI-2) [48]. From the homology of Salmonella and Edwardsiella, it can be inferred that EsrA-EsrB of Edwardsiella also plays a regulatory role in the pathogenic process. At present, the regulation of major virulence factors by the two-component system EsrA-EsrB of Edwardsiella has been studied; for instance, the gene esrB regulates the secretion of the essential secretory proteins EseB, EseC, and EseD of T3SS [49], and the effector protein EvpP of T6SS in E. tarda [50] also strongly induces the expression of T3SS/T6SS in E. piscicida [51]. In summary, the EsrA-EsrB two-component system has regulatory effect on both T3SS and T6SS of the host “weapons” of Edwardsiella pathogen infection. It was speculated that the function of the gene esrB with a missense amino acid mutation in this study might be affected, which then led to a drop in virulence. The regulatory effect of the gene esrB on the secretion system and other virulence factors in E. ictaluri needs to be further verified by gene knockout and site-directed mutagenesis.

Ubiquitination is an integral signal for the host immune response to pathogenic pathogens [52, 53]. It is precisely because of its crucial role that the host ubiquitination pathway becomes a target of pathogen hijacking. Pathogens secrete effector proteins that mimic eukaryotic ubiquitin ligase E3 or encode novel E3 ligases (NELs) that are different in sequence and structure from any eukaryotic ubiquitin ligase E3 [54]. They can also encode deubiquitinating enzymes to manipulate the host ubiquitination process [55], thereby facilitating bacterial entry into the host cell, setting up a replication environment and accelerating the infection process. In this study, the gene sspH2 in E9-302 had multiple missense mutations (Table 6), which likely affected its normal function. The E. ictaluri gene sspH2 is homologous to the gene sspH2 encoding protein SspH2 of the E3 ubiquitin ligase family in Salmonella. In Salmonella, the T3SS2 encoded by pathogenicity island 2 (SPI-2) is activated following invasion of host cells, which is essential for intracellular survival and proliferation [56]. SspH1, SspH2, and SlrP, the important effector proteins of T3SS2, are three members of the Salmonella NELs, all of which contribute to virulence in pathogen infection [57-59]. It is worth noting that SspH2, encoded by gene sspH2, could assist pathogens to invade host cells and advance themselves intracellularly to disrupt host immune responses [60-67]. Based on the protein structure and homology, it is hypothesized that E. ictaluri SspH2 may also have a regulatory or even disruptive effect on the host immune response, and the conjecture will be subsequently verified by a combination of multiple techniques including deletion strain construction, as well as the combination of dual luciferase reporter genes and immunoprecipitation.

Glycosylation modifications of bacterial proteins have a major impact in many of their life activities including adhesion, colonization, pathogenicity and immune escape [68-70]. In another gene named Gt39-like with differences in this study, E9-302 was missing a base compared with WT, resulting in a mismatch of the subsequent 189 bp sequence (Table 6). Alignment revealed that the amino acid sequences encoded by this gene were highly homologous to that of GT39 of S. enterica, Y. regensburgei, and K. michiganensis. It was found through comparison with the CAZy database that the protein encoded by the gene Gt39-like belonged to dolichol-phosphate-mannose-protein mannosyltransferase (PMT), a relatively conserved GT39 protein in all organisms. PMT acts as a catalyst for the initial step of protein mannosylation by catalyzing the transfer of mannosyl residues to serine or threonine residues [71], and plays an indispensable part in protein mannosylation. Studies have shown that PMT plays an instrumental role in maintaining the virulence of Mycobacterium tuberculosis [72] and contributes to the adherence and invasive ability of Campylobacter jejuni [73]. Up to the present, protein mannosylation in prokaryotes has been poorly studied. The product encoded by the gene named Gt39-like may also have a part in virulence and adhesion of E. ictaluri, which needs to be verified by subsequent knockout.

In this study, the strategy for exploiting rifampicin-resistant attenuated pathogens was applied to the highly pathogenic E. ictaluri isolated from diseased yellow catfish. An attenuated strain E9-302 with a 119-fold decrease in LD₅₀ in virulence to zebrafish in comparison to the parental strain WT was obtained. The zebrafish inoculated with strain E9-302 in different ways showed protection against challenge with the parental wild-type E. ictaluri strain WT, producing 85.7% (by IP route) and 65.2% (by IM plus booster IM route) RPS respectively, suggesting the potential of E9-302 as a live attenuated vaccine against edwardsiellosis. The virulence assay combined with the comparison of cell proliferation rate and in vitro competition ability of the parental strain WT and E9-302 proved for the first time that the acquisition of rifampicin resistance gave fish pathogens fitness costs in terms of reduced pathogenicity, slower cell proliferation, and decreased competitiveness. Furthermore, the differences in the gene rpoB sequences between the two on the basis of whole-genome sequencing emphasized the generation of fitness cost again. Additionally, whole-genome sequencing revealed that mutations in the gene esrB, the gene sspH2 encoding E3 ubiquitin ligase SspH2, and the gene named Gt39-like encoding the glycosyltransferase family 39 protein, were likely to be associated with virulence attenuation in E9-302. For these reasons, this work provides new ideas and targets for subsequent studies on the pathogenesis of E. ictaluri.

Fig. 5. Survival curves of the IP immunized group, the IM immunized group, and respective control groups after the strain WT challenge.