Tricarboxylic Acid Cycle and One-Carbon Metabolism Pathways Are Important in Edwardsiella ictaluri Virulence

Edwardsiella ictaluri is a Gram-negative facultative intracellular pathogen causing enteric septicemia of channel catfish (ESC). The disease causes considerable economic losses in the commercial catfish industry in the United States. Although antibiotics are used as feed additive, vaccination is a better alternative for prevention of the disease. Here we report the development and characterization of novel live attenuated E. ictaluri mutants. To accomplish this, several tricarboxylic acid cycle (sdhC, mdh, and frdA) and one-carbon metabolism genes (gcvP and glyA) were deleted in wild type E. ictaluri strain 93-146 by allelic exchange. Following bioluminescence tagging of the E. ictaluri ΔsdhC, Δmdh, ΔfrdA, ΔgcvP, and ΔglyA mutants, their dissemination, attenuation, and vaccine efficacy were determined in catfish fingerlings by in vivo imaging technology. Immunogenicity of each mutant was also determined in catfish fingerlings. Results indicated that all of the E. ictaluri mutants were attenuated significantly in catfish compared to the parent strain as evidenced by 2,265-fold average reduction in bioluminescence signal from all the mutants at 144 h post-infection. Catfish immunized with the E. ictaluri ΔsdhC, Δmdh, ΔfrdA, and ΔglyA mutants had 100% relative percent survival (RPS), while E. ictaluri ΔgcvP vaccinated catfish had 31.23% RPS after re-challenge with the wild type E. ictaluri.


Introduction
Channel catfish, Ictalurus punctatus, farming is the largest aquaculture industry in the United States, and enteric septicemia of catfish (ESC), caused by Edwardsiella ictaluri, is the most prevalent disease affecting this industry. Although RometH 30, TerramycinH, and AquaflorH are approved antibiotics to treat infections in commercial catfish by oral delivery in medicated feed, effectiveness is limited because fish develop anorexia at early stages of the infection. Also, antibiotic resistant E. ictaluri strains can emerge [1]. Therefore, vaccination is the preferred method for prevention of ESC.
Live attenuated vaccines can provide effective protection against certain diseases if they can express protective antigens without causing disease in the host [2]. In E. ictaluri, some candidate live attenuated vaccines that have been developed include chondroitinase [3] and auxotrophic (aroA and purA) [4,5] mutants. However, none of these vaccine candidates are in commercial production. The commercial vaccine Aquavac-ESC (RE-33) was developed by selecting for rifampin resistance [6]. However, antibiotic resistance is not a desired trait for a vaccine. In addition, the genetic basis for attenuation in RE-33 is undefined [7], although it is known that RE-33 expresses shortened LPS O side chains [8]. Despite the availability of Aquavac-ESC, ESC is still the most prevalent disease in the catfish industry [9,10]. E. ictaluri is considered a facultative intracellular pathogen, and it is capable of surviving inside channel catfish neutrophils and macrophages [11,12]. Although E. ictaluri is effectively phagocytosed by catfish neutrophils, it is only killed by neutrophils to a limited extent [11,13]. A recent study by Karsi et al. [14] showed that genes encoding tricarboxylic acid (TCA) cycle enzymes, glycine cleavage system, a sigmaE regulator, the SoxS oxidative response system, and a plasmid-encoded type III secretion system (TTSS) effector are important for survival in neutrophils [14]. The same study discovered that some neutrophil-susceptible E. ictaluri strains were highly attenuated and demonstrated very good potential as live attenuated vaccines. In particular, strains with insertion mutations in genes encoding TCA cycle enzymes succinate dehydrogenase (sdhC) (EiAKMut5) and malate dehydrogenase (mdh) (EiAKMut12) generated better protection than the available commercial vaccine when juvenile catfish were vaccinated by immersion [14]. Similarly, E. ictaluri glycine dehydrogenase (gcvP) mutants (EiAKMut02 and EiAKMut08) were also completely attenuated and had better vaccine efficacy than the commercial vaccine [14]. Glycine dehydrogenase is part of the glycine cleavage system pathway, which is part of one-carbon (C1) metabolism. Therefore, the objective of this research was to introduce in-frame deletions in E. ictaluri sdhC, mdh, and frdA genes (encoding enzymes in the TCA cycle) and gcvP and glyA genes (encoding C1 metabolism proteins) to determine their roles in E. ictaluri virulence.

Ethics statement
All fish experiments were conducted in accordance with a protocol approved by the Institutional Animal Care and Use Committee (IACUC) at Mississippi State University.
Bacterial strains, plasmids, and growth conditions Bacterial strains and plasmids used in this work are listed in Table 1. E. ictaluri was grown at 30uC using brain heart infusion (BHI) broth and agar (Difco, Sparks, MD). Escherichia coli were grown at 37uC using Luria-Bertani (LB) broth and agar (Difco). E. coli CC118 lpir and SM10 lpir/S17-1 lpir were used for cloning gene deletions into suicide plasmid pMEG-375 and transferring recombinant pMEG-375 or pAKgfplux1 into E. ictaluri. Ampicillin was used at 100 mg/ml to maintain pMEG-375 and pAKgfplux1. Colistin was used at 12.5 mg/ml for counter selection against E. coli SM10 lpir following conjugation. E. ictaluri strains were cultivated for 18 h (stationary phase) for all fish challenges.

Construction and bioluminescence tagging of in-frame deletion mutants
The method of overlap extension PCR [15] was used to generate in-frame deletions of E. ictaluri sdhC, mdh, frdA, gcvP, and glyA. Four primers were designed for each gene including forward (lflp), internal-reverse (lfrp), internal forward (rflp), and reverse primers (rfrp) ( Table 2). Restriction sites were included in forward and reverse primers. Genomic DNA was isolated from E. ictaluri using a Wizard Genomic DNA Kit (Promega, Madison, WI) and used as template in PCR. Upper fragments were amplified by forward and internal-reverse primer sets while reverse and internal-forward primer sets were used to amplify lower fragments.
The resulting upper and lower PCR products were gel extracted using a QIAquick Gel Extraction Kit (Qiagen, Valencia, CA), mixed in a 1:1 ratio, and then re-amplified using the forward and reverse primers. The resulting in-frame deleted fragment was purified by using a QIAquick Gel Extraction Kit (Qiagen, Valencia, CA). The purified PCR product was digested with appropriate restriction enzymes (Promega) ( Table 1) and cleaned using a Wizard SV Gel and PCR Clean-Up Kit (Promega).
The suicide plasmid pMEG-375 was purified from an overnight E. coli culture by a QIAprep Spin Miniprep Kit (Qiagen) and cut with restriction enzymes respective to the inserts, producing compatible ends. The purified PCR product with in-frame deletion was ligated into pMEG-375 vector using T4 DNA Ligase (Promega) at 4uC overnight, generating pEiDsdhC, pEiDmdh, pEiDfrdA, pEiDgcvP, and pEiDglyA (Table 1). Insert in each plasmid was confirmed by restriction enzyme digestion as well as sequencing.
The suicide plasmids with in-frame deleted genes were transferred into E. coli SM10 lpir/S17-1 lpir by electroporation and mobilized into E. ictaluri 93-146 by conjugation [16]. The recipient bacteria were spread on BHI plates containing colistin (12.5 ug/ml) and ampicillin (100 ug/ml) to select E. ictaluri with integrated vector by single crossover through allelic exchange. Ampicillin resistant colonies were propagated on BHI plates to allow for the second crossover allelic exchange, followed by streaking on BHI plates with 5% sucrose, 0.35% mannitol, and colistin to select for loss of pMEG-375 with sacB gene. Potential mutant colonies were tested for ampicillin sensitivity to ensure loss of the plasmid. Deleted regions were amplified from the resulting ampicillin sensitive colonies and confirmed by sequencing. After confirmation, EiDsdhC, EiDmdh, EiDfrdA, EiDgcvP, and EiDglyA mutants were labeled with bioluminescence using pAKgfplux1 as described in Karsi and Lawrence [16]. Mutant virulence and ability to protect against E. ictaluri infection Experimental infections were conducted in 40-L challenge tanks supplied with flow-through dechlorinated municipal water. Water temperature was maintained at 25uC (62) throughout the experiments. Twenty-eight specific pathogen free (SPF) catfish fingerlings (14.260.35 cm, 25.4561.82 g) were randomly allocated into seven groups (4 fish/group). Five treatments were injected with E. ictaluri mutants, one group was injected with wild type E. ictaluri strain 93-146, and the last group served as negative control (phosphate-buffered saline (PBS). Fish were anesthetized in water containing 100 mg/L MS222 and injected with approximately 1610 4 colony forming units (CFU) in 100 ml PBS.
Bioluminescent imaging (BLI) was conducted using an IVIS 100 Imaging System to measure number of photons emitted by bioluminescent bacteria in fish [17]. Briefly, catfish were anesthetized in water containing 100 mg/L MS222 and transferred immediately to the photon collection chamber for image capture. Total photon emissions from the whole fish body were collected at an exposure time of one min. Following BLI imaging, fish were returned to well-aerated water for recovery. BLI was conducted at 2, 4, 8, and 24 h post-infection, and subsequent daily intervals until 168 h. Bioluminescence was quantified from the fish images using Living Image Software v 2.5 (Caliper Corporation., Hopkinton, Massachusetts), and mean photon counts for each treatment were used in statistical analysis.
To determine the ability of mutants to protect against E. ictaluri infection, the juvenile catfish vaccinated with mutants (virulence challenge) were immersion challenged [14] with 4.8610 7 bioluminescent wild type E. ictaluri at 4 weeks post-vaccination. Photon emissions from fish were collected at 2, 4, 8, 24, 48, 72, and 96 h post-infection using an IVIS 100 as described above, and statistical analysis was performed on the mean photon counts.

Mutant ability to protect against ESC induced mortalities
Approximately 420 eight-month-old SPF channel catfish fingerlings (17.6160.63 cm, 47.4765.31 g) were stocked into 21 tanks at a rate of 20 fish/tank. Each treatment had three replicate tanks. Treatments consisted of EiDsdhC, EiDmdh, EiDfrdA, EiDglyA and EiDgcvP (vaccination), wild type E. ictaluri (positive control), and BHI (sham control). Channel catfish were vaccinated by immersion in water containing approximately 4.3610 7 CFU/ml of water for 1 h, followed by gradual removal of bacteria. Mortalities were recorded for 21 days following vaccination. At 21 days postvaccination, both vaccinated and non-vaccinated treatments were immersion exposed to wild type parent E. ictaluri 93-146 (approximately 3.06610 7 CFU/ml), and fish mortalities were recorded daily for 14 days. Relative percent survival (RPS) was

Statistical analysis
Photon counts were transformed by taking the base 10 logarithm to improve normality. One-way ANOVA was conducted using SPSS V19 (IBM Corp., Armonk, NY) to compare mean photon counts at each time point (p,0.05). Pairwise comparison of the means was done using Tukey procedure. Data was then retransformed for interpretation.

Mutant virulence and ability to protect against E. ictaluri infection
BLI results revealed that bioluminescence (quantified as photon counts) from the catfish infected with EiDsdhC, EiDmdh, EiDfrdA, EiDgcvP, and EiDglyA mutants were low at 2, 6, and 12 h postinfection. However, bioluminescence for mutants EiDmdh, EiDfrdA, and EiDgcvP increased from 24 h to 72 h and then decreased thereafter. Bioluminescence for EiDsdhC followed the same pattern as the other mutants, except the signal peaked at 120 h. However, in mutant EiDglyA, very low bioluminescence was detected at all time points. In fish infected with wild type E. ictaluri, bioluminescence increased until all fish died (Fig. 2). Average photon counts in the fish infected with 93-146 at 72 h post-infection were approximately 7-fold higher than the average of all fish infected with mutant strains, and it was 2,265-fold higher at 144 h. At this time point, fish infected with wild type E. ictaluri strain died, while bioluminescence from fish infected with mutant strains was in decline (Fig. 2). Photon counts were 118-and 5,329-fold higher in wild type E. ictaluri compared to EiDglyA at 72 h and 144 h postinfection, respectively (Fig. 2). In the wild type infected treatment, two fish died at 144 h post-infection, and the remaining two fish died at168 h. Mean photon counts between all mutants (except EiDfrdA) and wild type E. ictaluri were significantly different (p,0.5) at 24 h and thereafter. Mean photon counts for wild type E. ictaluri were significantly higher than EiDfrdA at 48 h and thereafter.
When mutant challenged fish were immersion exposed to wild type E. ictaluri at 4 weeks post-vaccination, photon counts were significantly lower (p,0.5) at each time point for the vaccinated fish compared to the sham-vaccinated control (Fig. 3). Average photon counts in sham-vaccinated fish at 6 h post-infection were 4-fold higher than the average of all five mutant-vaccinated fish treatments, which increased to 14-fold at 96 h. At this time, bioluminescence in EiDsdhC and EiDmdh vaccinated fish was declining, while bioluminescence in EiDfrdA, EiDgcvP, and EiDglyA vaccinated fish was increasing (Fig. 3).At 96 h post-infection, all fish in the sham vaccinated group died. In summary, BLI demonstrated that all mutants are significantly attenuated compared to wild type E. ictaluri, and all mutants except EiDglyA provided significant protection against E. ictaluri infection.

Discussion
The primary objective of this study was to construct live attenuated E. ictaluri strains based on mutations in genes encoding enzymes in the TCA cycle (mdh, schC, and frdA) and enzymes involved in C1 metabolism (gcvP and glyA). Additional aims included assessing the mutant strains' virulence in catfish and ability to protect against wild type E. ictaluri infection. We constructed in-frame deletion mutants to avoid polar effects of the mutations and to avoid insertion of antibiotic resistance genes, which is undesirable in vaccine strains. Splicing overlap extension combined with allelic exchange is an effective method for gene deletion in E. ictaluri and has been reported previously [19,20]. We utilized bioluminescence imaging to assess virulence of mutants, which allows better quantification compared to percent mortalities. It also enables sensitive detection of subclinical infection and mutants' abilities to invade and establish infection. Mutant strains EiDsdhC, EiDmdh, EiDfrdA, and EiDgcvP were clearly able to establish infection because bioluminescence was detected after 12 h post-infection. However, channel catfish injected with the mutant strains started clearing the bacteria after 72 h post-infection. Thus, our results showed that although EiDsdhC, EiDmdh, EiDfrdA, and EiDgcvP do not cause mortalities, Table 3. Properties of the E. ictaluri TCA cycle and C1 metabolism genes and percentage of gene deleted.  they are able to invade and establish infection before being cleared. Because of mutants' abilities to survive and replicate in fish up to 72 h post-infection, we expected them to generate an immune response and protection against wild type E. ictaluri. On the other hand, the EiDglyA mutant did not replicate well in the host, and we anticipated much less systemic protection from this mutant. By contrast, wild type E. ictaluri increased in quantity until mortality occurred. Our current study corroborated an earlier study showing that 1610 9 photons 21 cm 22 steradian 21 seems to be a critical threshold for bacterial tissue concentrations where mortality is imminent [21]. Ultimately, prevention of mortalities is used as a common measure of vaccine efficacy. Thus, we used percent survival to evaluate efficacy of our candidate vaccines in catfish fingerlings using immersion exposure, which is a practical route of vaccination of catfish fry in catfish production systems. Results for mutant strains EiDsdhC, EiDmdh and EiDgcvP were similar to our previous study that evaluated vaccine efficacy of E. ictaluri sdhC, mdh, and gcvP transposon insertion mutants [21]. In our previous study, sdhC and mdh insertion mutants gave 100% protection against E. ictaluri infection, and a gcvP insertion mutant gave 89.15% survival in catfish fingerlings. Our current results with deletion mutants show that attenuation is not due to polar effects of the insertion mutations. The deletion mutants have an additional advantage in that they do not carry antibiotic resistance genes. The current study is the first to report vaccine efficacy of E. ictaluri EiDglyA and EiDfrdA mutants; both provided significant protection against mortalities by immersion vaccination.
We also evaluated vaccine efficacy of our candidate mutant strains using a more sensitive measure than percent survival; namely, we evaluated the ability of the mutant strains to prevent invasion of virulent E. ictaluri as monitored using BLI. Vaccination in this trial was by injection, which is not a practical route of vaccination for commercial catfish production, but it does allow accurate vaccine dose delivery. Protection results by injection vaccination were very similar to results obtained by immersion Figure 3. Bioluminescence imaging of juvenile catfish after immersion exposure to wild type E. ictaluri. Fish were challenged with E. ictaluri mutants as described in the virulence trial, and at 4 weeks post-vaccination they were challenged with bioluminescent wild type E. ictaluri. A, BLI imaging of catfish. B, Total photon emissions from each fish. Each data point represents the mean photon emissions from four fish. Star indicates significant difference between the E. ictaluri mutants and wild type. doi:10.1371/journal.pone.0065973.g003 vaccination, except that EiDglyA vaccination provided better protection by immersion vaccination than injection (Fig. 4). It is possible that immersion vaccination using EiDglyA may activate mucosal immunity better, preventing wild type E. ictaluri septicemia. We saw the opposite trend when fish were vaccinated with the EiDgcvP mutant, which protects fish better when vaccination is applied by injection rather than immersion.
Succinate dehydrogenase (SDH) is part of the aerobic respiratory chain in the TCA cycle, oxidizing succinate to fumarate while reducing ubiquinone to ubiquinol [22]. It is closely related to fumarate reductase, which catalyzes the reverse reaction. Succinate dehydrogenase and fumarate reductase can replace each other [22,23]. Although SdhC has similar function, hydrophobicity, and protein size to the membrane-binding subunit fumarate reductase (FrdC), sdhC and frdC do not share significant sequence identity [24]. The organic acids formate and succinate have a protective effect in stationary phase cells against killing effects of antimicrobial peptide BPI, which appears to disrupt the bacterial respiratory chain [25]. Maintenance of protective levels of formate and succinate requires the activity of formate dehydrogenase and succinate dehydrogenase, respectively.
In E. coli and Salmonella, succinate dehydrogenase is known to contribute to pathogenicity. Recently, it was shown that a full TCA cycle is required for Salmonella enterica virulence, and a sdhDCA mutant is attenuated in an oral mouse infection model [26], which is similar to our finding. In Helicobacter pylori, fumarate reductase was found to be essential for colonization of mouse gastric mucosa [27]. In Salmonella enterica, deletion of sdhCDA caused partial attenuation, and complete attenuation was achieved when both sdhCDA and frdABCD were deleted [28]. Our results indicated that deletion of only the E. ictaluri sdhC gene and deletion of only frdA resulted in full attenuation in catfish fingerlings. However, our previous results showed that catfish fry are more sensitive to E. ictaluri than catfish fingerlings (unpublished data), so further testing in catfish fry is warranted. Regardless, the data show that succinate dehydrogenase and fumarate reductase play an important role in pathogenesis. The other mutant that was tested in this study was mdh, which encodes malate dehydrogenase. Our results show that mdh is also important in E. ictaluri virulence, which was consistent with findings in Salmonella using the mouse oral challenge model, where a mdh mutant was found to be highly attenuated [26]. The glycine cleavage system is a loosely associated four subunit enzyme complex that catalyzes the reversible oxidation of glycine to form 5, and 10-methylenetetrahydrofolate, which serves as a one carbon donor. It is one of two sources of C1 units; serine hydroxymethyltransferase is another source, and it is considered a more important source. Expression of the glycine cleavage enzyme system is induced by glycine [29,30], and gcv mutants are unable to use glycine as a C1 source and excrete glycine [31]. We have previously shown that E. ictaluri gcvP is required for virulence [14]. This is the first report that glyA is required for E. ictaluri, virulence, and to our knowledge, this is the first report that serine hydroxymethyltransferase is associated with virulence in any bacterial species.
Although BLI for real-time monitoring of E. ictaluri infection in live fish was shown by our group [17], this is the first time we report the use of BLI to quantify the degree of E. ictaluri attenuation in channel catfish. It appears that BLI could be used for vaccine evaluation by using a relatively low number of fish (four fish in this work). Also, use of BLI provides a more sensitive measure of vaccine protection than percent mortalities.
In summary, our results showed that the EiDsdhC, EiDmdh, EiDfrdA, EiDgcvP, and EiDglyA mutants were significantly attenuated and provided protection against ESC under controlled laboratory conditions. Thus, EiDsdhC, EiDmdh, EiDfrdA, and EiDgcvP mutants have potential for use as live attenuated vaccines for catfish fingerlings. The E. ictaluri DglyA mutant was found to be incapable of persisting in catfish when injected, which might be the reason for lower protection than when it is used in immersion vaccination. Based on these results, testing of these vaccine candidates in catfish fry is warranted.