The Aspartate-Semialdehyde Dehydrogenase of Edwardsiella ictaluri and Its Use as Balanced-Lethal System in Fish Vaccinology

asdA mutants of Gram-negative bacteria have an obligate requirement for diaminopimelic acid (DAP), which is an essential constituent of the peptidoglycan layer of the cell wall of these organisms. In environments deprived of DAP, i.e., animal tissues, they will undergo lysis. Deletion of the asdA gene has previously been exploited to develop antibiotic-sensitive strains of live attenuated recombinant bacterial vaccines. Introduction of an Asd+ plasmid into a ΔasdA mutant makes the bacterial strain plasmid-dependent. This dependence on the Asd+ plasmid vector creates a balanced-lethal complementation between the bacterial strain and the recombinant plasmid. E. ictaluri is an enteric Gram-negative fish pathogen that causes enteric septicemia in catfish. Because E. ictaluri is a nasal/oral invasive intracellular pathogen, this bacterium is a candidate to develop a bath/oral live recombinant attenuated Edwardsiella vaccine (RAEV) for the catfish aquaculture industry. As a first step to develop an antibiotic-sensitive RAEV strain, we characterized and deleted the E. ictaluri asdA gene. E. ictaluri ΔasdA01 mutants exhibit an absolute requirement for DAP to grow. The asdA gene of E. ictaluri was complemented by the asdA gene from Salmonella. Several Asd+ expression vectors with different origins of replication were transformed into E. ictaluri ΔasdA01. Asd+ vectors were compatible with the pEI1 and pEI2 E. ictaluri native plasmids. The balanced-lethal system was satisfactorily evaluated in vivo. Recombinant GFP, PspA, and LcrV proteins were synthesized by E. ictaluri ΔasdA01 harboring Asd+ plasmids. Here we constructed a balanced-lethal system, which is the first step to develop an antibiotic-sensitive RAEV for the aquaculture industry.


Introduction
Aspartate b-semialdehyde deshydrogenase (Asd; EC 1.2.1.11), a highly conserved homodimeric enzyme encoded by the asd gene, is involved in the conversion of b-aspartyl phosphate to aspartate bsemialdehyde. Asd is an enzyme common to the biosynthesis of the essential amino acids lysine, threonine, methionine, and isoleucine. It also performs a key step in the production of diaminopimelic acid (DAP), a required component for the peptidoglycan synthesis of Gram-negative and some Gram-positive bacterial cell walls [1,2,3,4] and an immediate precursor to lysine. asd mutants have an obligate requirement for DAP, and in the absence of DAP they undergo lysis. This has been demonstrated by gene-knockout studies with Legionella pneumophila [5], Salmonella Typhimurium [6] and Streptococcus mutans [7].
The Asd enzyme is also found in plants, where lysine is synthesized via the DAP pathway [8,9]. In contrast, mammalian cells neither synthesize nor use DAP as a substrate in any metabolic pathway, and lysine is not synthesized since it is an essential amino acid that is obtained from dietary sources [5,10,11]. Also lysine, threonine, methionine, and isoleucine are essential amino acids in the diet of teleostei fish [12,13,14,15,16,17], suggesting the absence of both the DAP/lysine synthesis pathway and Asd enzyme in fish cells.
Since DAP is absent from mammalian tissues, deletion of the asd gene has been exploited to develop a balanced-lethal system for vaccine delivery vehicles using a cloned asd gene as a selective marker in place of antibiotic-resistance markers, which are totally impractical in vivo [6]. Introduction of an Asd + plasmid into asd mutants makes the bacterial strain plasmid-dependent. This dependence on the Asd + plasmid vector creates a balanced-lethal complementation between the bacterial strain and the recombinant plasmid [18]. Asd + vectors introduced into live recombinant attenuated Salmonella vaccines have been used to deliver heterologous antigens [19]. The construction of live attenuated recombinant bacterial vaccines not only require the absence of antibiotic-resistance markers in their recombinant plasmid, but also in their chromosomal deletions.
Edwardsiella ictaluri, a Gram-negative bacterial pathogen, is the cause of enteric septicemia in catfish, which causes losses estimated at $50-80 million annually [20]. The current USDA licensed vaccine, live E. ictaluri AQUAVAC-ESCH (Intervet Inc.), has been selected by multiple passages in increased concentrations of the antibiotic rifampicin [21,22,23]. The selected spontaneous mutant strain presented an attenuated phenotype missing part of the lipopolysaccharide (LPS) [20,24]. Although there are FDA and USDA regulations against the use of antibiotic resistance in live attenuated bacterial vaccines for birds, mammals, and humans, the catfish industry currently allows antibiotic-resistant vaccine strains. Despite the fact that the current vaccine against enteric septicemia in catfish is antibiotic resistant, by using this vaccine we have learned that E. ictaluri live attenuated vaccines can be easily delivered to young fish and stimulate both humoral and cellular immunity of long duration [25,26]. These results provide guidance to design live attenuated antibiotic-sensitive vaccines for the catfish aquaculture.
As a first step in developing an antibiotic-sensitive live recombinant E. ictaluri vaccine strain (RAEV), we adapted suicide vector technology [27] to E. ictaluri to construct defined unmarked chromosomal deletion mutations, for instance the asd deletion. Two E. ictaluri asd genes were identified, a functional asdA and a non-functional asdB pseudogene. The asdA gene was deleted by using the described suicide vector technology. Using Asd + expression vectors [19], we developed a balance-lethal system compatible with E. ictaluri native plasmids, to express and secrete heterologous proteins through the type II secretion system. The virulence of the E. ictaluri DasdA mutant, harboring an AsdA + expression vector, was evaluated in vivo in the catfish (Ictalurus punctatus) and in the zebrafish (Danio rerio) host models. Here we report the first balanced-lethal vector-host system in E. ictaluri, a key in constructing antibiotic-sensitive live RAEV for the catfish industry.

Sequence analysis
To develop a balanced-lethal system we first characterized the asd genes present in E. ictaluri. The genome of E. ictaluri has two asd gene sequences, asdA (gene ID 7960734) and asdB (gene ID 7959931). Sequence and structural alignment between functional representative bacterial Asd proteins reveals that 22 amino acid residues (,6%) are strictly conserved out of 367 residues in E. ictaluri AsdA (Fig. 1). E. ictaluri AsdA has 28%, 81%, 82%, 84%, and 97% amino acid similarity to the Asd of Streptococcus mutans, Salmonella enterica, Escherichia coli, Yersinia (Y. pestis and Y. ruckeri), and E. tarda, respectively. The overall domain organization of E. ictaluri AsdA is similar to other Gram-negative Asd-family members, presenting an N-terminal domain comprising the NAD binding site and a C-terminal catalytic domain (Fig. 1). The same set of key functional groups in the active sites (Cys-135, Gln-162, Glu-241, Arg-267, and His-274) are conserved in E. ictaluri AsdA and likely have the same catalytic mechanism as other Asd enzymes (Fig. 1).
The sequence and structural alignment between representative bacterial AsdB proteins reveals that 52 amino acid residues (,15%) are strictly conserved out of 336 in E. ictaluri AsdB (Fig. 2). The E. ictaluri AsdB has 30%, 32%, 40%, 75%, and 99% amino acid similarity to the AsdB of Streptococcus mutans, Mycobacterium marinum, Vibrio cholerae, Y. pestis, and E. tarda, respectively. In contrast to AsdA, the overall domain organization of E. ictaluri AsdB is similar to other Gram-positive Asd-family members. However, E. ictaluri AsdB lacks key functional groups in the active sites (Cys-135, Gln-162, and Arg-267) and likely has no catalytic activity.
The guanine plus cytosine (G+C) content found in the E. ictaluri asdA gene was 62%, significantly higher than the 54% of G+C found in the Escherichia coli asdA gene. Overall DNA comparison of the asdA gene showed that the E. ictaluri asdA gene shared 72% identity with the Escherichia coli asdA gene.
In terms of phylogeny, the bacterial Asd family is subdivided into two structural branches consisting of the enzymes from Gramnegative and Gram-positive bacteria [28] (Fig. 3). The E. ictaluri AsdA enzyme belongs to the Gram-negative branch, in contrast to AsdB that belongs to the Gram-positive branch (Fig. 3). Edwardsiella species comprise a linage that diverged from the ancestral trunk before the divergence of some other enteric bacteria, such as Salmonella and Escherichia [29,30]. The phylogenetic position of the E. ictaluri AsdA enzyme corresponds with the E. ictaluri genome phylogenetic position (Fig. 3). Inside of the AsdB branch, a nonfunctional AsdB branch composed of Edwardsiella and Yersinia AsdB sequences was identified (Fig. 3), indicating that these nonfunctional AsdB proteins may have a common origin.

Construction and characterization of asdA mutants
The construction of E. ictaluri DasdA mutants was performed first by using pEZ101, a pR112 (Cm) base suicide vector (Tables 1 and  2). pEZ101 was conjugated from Escherichia coli x7213 to E. ictaluri J100 and E. ictaluri J102 using the methods described for E. ictaluri [31] and Escherichia coli [32]. The selection of transconjugants was carried out in BHI agar supplemented with Col, DAP, and Cm. We did not recover transconjugants by using pEZ101. Therefore, we constructed and used pEZ102, a pMEG-375 (Cm, Amp) base suicide vector ( Table 2). The selection of transconjugants was carried out in BHI agar supplemented with Col, DAP, Amp or Cm. Transconjugants were recovered in the presence of Amp, but not in the presence of Cm. Transconjugants Amp r , harboring pEZ102 (Amp, Cm), were sensitive to Cm. We determined that E. ictaluri is highly sensitive to Cm. Small colonies (.0.5 mm) harboring pEZ102 were recovered in a Cm concentration below 1 mg/ml. Using BHI agar supplemented with Col, DAP, and Cm (1 mg/ml), transconjugants were not recovered using pEZ101 (Cm) or pEZ102 (Amp, Cm). Certainly, these results indicate that Cm selection and Cm-base suicide vectors are not useful to genetically manipulate E. ictaluri.
Single colonies of E. ictaluri transconjugants harboring pEZ102 (Col r , Amp r ), were grown in BHI, TSB or LB supplemented with DAP and Col at 28uC for 6 h with aeration (180 r.p.m.). The selection was performed in BHI, TSA and LB agar plates supplemented with DAP, Col, and 5% sucrose at 28uC for 4-5 days. BHI sucrose selection agar did not provide selection, due to E. ictaluri overgrowth. TSA and LB sucrose selection agar presented a satisfactory selection. Positive mutants were screened for Col r , Amp s , and DAP -. Several E. ictaluri DasdA mutants were recovered from TSA and LB sucrose-selection agar plates. The genotype was verified by PCR, and the phenotype by growth in presence of DAP and no growth in absence of DAP (Fig. 4). The biochemical profile, evaluated by API20E, did not present any difference between the wild type and DasdA01 mutant strains. E. ictaluri strains were identified as Edwardsiella sp (code 4004000). These results confirmed that the AsdB present in E. ictaluri is nonfunctional, since deletion of asdA is enough to preclude cell growth in the absence of DAP. Thus, asdB can be considered a pseudo gene in E. ictaluri.
We evaluated reutilization of DAP by the E. ictaluri DasdA01 mutants released from lysed DasdA cells grown in absence of DAP. Washed cells of E. ictaluri J112 DasdA01 were diluted from 10 1 to 10 10 CFU/ml in BHI Col. The estimated minimum number of E. ictaluri DasdA01 cells needed to support growth in absence of DAP was 1.3610 8 -2.7610 8 CFU/ml. This is because of DAP-less death and reuse of DAP to permit growth on media without DAP.
The amount of DAP in the cell wall of Escherichia coli has been estimated at ,3.5610 6 molecules [33]. Based on the results obtained for the minimum number of E. ictaluri DasdA01 cells  needed to support growth in the absence of DAP, and the calculated amount of DAP molecules per cell of Escherichia coli, we estimated that the minimum number of DAP molecules to support growth is ,4.5610 14 -9.5610 14 molecules of DAP/ml in the growth media. We evaluated the growth of E. ictaluri DasdA01 in 10 10 to 10 20 molecules of DAP/ml in BHI Col. E. ictaluri DasdA01  did not grow in concentrations below 10 14 molecules of DAP/ml. Our previous estimation about the minimum number of DAP molecules required to support growth was confirmed, indicating that the amount of DAP in the cell wall of E. ictaluri is similar to Escherichia coli.
It has been reported that lysine, threonine, methionine, and isoleucine are essential amino acids in the diet of teleostei fish [12,13,14,15,16,17], suggesting the absence of the DAP/lysine synthesis pathway in fish cells. We tested the growth of E. ictaluri J112 DasdA01 in different catfish broths (1% of catfish liver, spleen, kidney and meat in BHI) in presence and absence of DAP. E. ictaluri J112 DasdA01 was not able to grow in fish broth not supplemented with DAP. E. ictaluri J100 wild-type, used as control, grew in all fish broth conditions (data not shown). This result supports the idea that as mammalian cells, fish cells neither synthesize nor use DAP as substrate in any metabolic pathway.
Complementation of E. ictaluri asdA gene and E. ictaluri DasdA01 mutant The structural analysis of E. ictaluri AsdA indicated that the overall domain organization is similar to other AsdA-family members and has the same set of key active-site functional groups and therefore the same catalytic mechanism as other Asd enzymes ( Fig. 1). To evaluate the likely broad functionality of E. ictaluri AsdA enzyme, asdA mutants of Escherichia coli, Salmonella enterica (serovars Typhimurium, and Typhi), Y. pestis, and E. ictaluri, were complemented with the E. ictaluri asdA gene. Because overproduction of AsdA enzyme increases generation times [19,34] and synthesis of Asd enzyme is proportional to the copy number of the complementing plasmid, asdA mutants were complemented with E. ictaluri asdA gene with (P asdA-asdA) and without its promoter (SD-asdA), this last to decrease Asd synthesis, cloned into p15A ori plasmid (pACYC184; Table 2).
Escherichia coli, S. enterica and Y. pestis DasdA mutants complemented with E. ictaluri SD-asdA presented similar growth rates compared to wild type (Fig. 5), indicating full complementation. E. ictaluri DasdA01 mutants complemented with SD-asdA presented a significantly lower growth rate than the wild type (Fig. 5). This could be due to overproduction or underproduction of Asd. It has been reported that SD-asd constructions do not enable DasdA strains to survive in absence of DAP if the origin of plasmid replication (ori) is from pSC101 or p15A. In other words, with these lower-copy-number replicons, the amount of Asd enzyme synthesized is insufficient to enable growth in absence of lysis [19]. To evaluate if the decrease in the generation time of the SD-asdA complemented E. ictaluri DasdA01 strain was due to overproduction or underproduction of AsdA, complementation with P asdA -asdA (pEZ142) was performed. Complementation of E. ictaluri DasdA01  mutants with pEZ142 decreased the growth rate even more than complementation with SD-asdA. These results suggest that the decreased growth rate in the E. ictaluri DasdA01 complemented with its own asdA gene is due to overproduction of AsdA. There are differences in the SD regions that could justify part of the difference in the growth rate of E. ictaluri DasdA01 complemented with its own SD-asd gene. The SD region of E. ictaluri asdA gene has an optimal spacing (6 nt) between the SD region and the ATG initiation codon of the mRNA [35] in contrast to the other bacterial species complemented with E. ictaluri SD-asdA (Fig. 5).
Complementation of E. ictaluri DasdA01 mutants by Grampositive AsdB enzyme was also evaluated. Streptococcus mutans asdB region (including the full promoter), cloned into pYA575 [36] and S. mutans SD-asdB, cloned into pYA248 [18] complemented E. ictaluri DasdA01 mutants. However these strains presented lower growth rates than the wild type (Fig. 6). E. ictaluri DasdA01 mutants complemented with SD-asdB (pYA248), presented the lowest growth rate, suggesting that S. mutans AsdB is probably required in higher levels to fully complement E. ictaluri or S. mutans AsdB do not interact efficiently with E. ictaluri aspartokinase enzymes to transfer the b-aspartyl phosphate to Asd.
Complementation by Asd + vectors to develop a balanced-lethal system in E. ictaluri The asdA gene from E. ictaluri complemented S. enterica DasdA mutants, in addition the Asd enzymes from E. ictaluri and S. enterica share 81% similarity. Therefore, we used the Asd + vectors utilized in live recombinant attenuated Salmonella vaccines [19] to develop a balanced-lethal system in E. ictaluri. The Asd + vectors utilized in this study possess only the SD-asdA gene from S. Typhimurium with a modified start codon from ATG to GTG. E. ictaluri DasdA01 mutants were complemented with the asdA gene from S. Typhimurium (Fig. 7). The growth rate of E. ictaluri DasdA01 complemented with different copy number of Asd + vectors was similar to the wild type in all cases (Fig. 7). The Asd + vectors were compatible with the native plasmids of E. ictaluri (Fig. 7) and stable for at least 80 generations. These results show the first balancedlethal system in E. ictaluri.

Expression of genes encoding GFP protein in the AsdA + vector
The synthesis of heterologous proteins, for instance GFP, cloned into Asd + vectors was evaluated in E. ictaluri DasdA01 to potentially develop live E. ictaluri recombinant vaccines. First, the synthesis of heterologous proteins was evaluated by using the GFP + Asd + vector pYA3994 (Table 2). E. ictaluri DasdA01 mutant strains harboring the GFP + Asd + vector grew in absence of DAP and synthesized GFP + as expected (Fig. 8). The GFP + Asd + vector was compatible with the native plasmids of E. ictaluri in the relaxed conformation (Fig. 8). The GFP + Asd + vector was stable in E. ictaluri DasdA01 strains for at least 80 generations. The expression of LcrV and PspA heterologous proteins using AsdA + vectors was also evaluated (see below).

Secretion of heterologous proteins
Secretion of the heterologous antigens by live attenuated recombinant bacterial vaccines has been shown to enhance immunogenicity against the heterologous antigen [37]. The synthesis and secretion of heterologous proteins was evaluated by using the proteins derived from Gram-positive and Gramnegative bacterial strains. PspA-Rx1 from Streptococcus pneumoniae was utilized as a Gram-positive representative and LcrV from Yersinia pestis was utilized as a Gram-negative representative. The heterologous antigens, PspA-Rx1 and LcrV fused to b-lactamase signal sequence, were expressed from the Asd + vectors pYA4088 and pYA3841, respectively ( Table 2). Both heterologous proteins were secreted through the type II secretion system. No difference in the growth rate was observed between the recombinant E. ictaluri and the wild-type strain J100.
Virulence of E. ictaluri DasdA01 strain complemented with the AsdA + plasmid vector in catfish host and zebrafish host models The idea to develop a balanced-lethal system in a pathogenic bacterial strain is to synthesize heterologous antigens, without the use of antibiotic-resistant genes, in either the plasmid or in the bacterial chromosome. This is the first step towards developing live recombinant bacterial vaccines. The ideal balanced-lethal system should present nearly the same level of virulence as the wild-type strain with regard to invasion and colonization of lymphoid tissues. We evaluated the virulence of E. ictaluri DasdA01 mutants with and without the balanced-lethal system in the catfish and zebrafish hosts (Tables 3 and 4). We used pYA3493 AsdA + since this vector has been used successfully in live recombinant Salmonella vaccines [19,38,39,40]. E. ictaluri DasdA01 was attenuated at the high dose of 10 8 CFU, but still produced some mortality in catfish (Table 4). E. ictaluri DasdA01 at a high dose (10 8 CFU) was not attenuated in zebrafish and all the fish died (Table 3). However, at lower doses (10 7 -10 4 ) E. ictaluri DasdA01 was totally attenuated in zebrafish (Table 3). E. ictaluri DasdA01 harboring the Asd + vector pYA3493 increased the LD 50 one log-fold, from 10 4 CFU to 10 5 CFU in orally infected catfish, and two logfold, from 10 3 CFU to 10 5 CFU, in zebrafish (Tables 3 and 4). Catfish i.p. infected with E. ictaluri DasdA01 harboring the Asd + vector pYA3493 presented the same level of virulence as E. ictaluri wild type (Table 4). From moribund orally infected catfish, E. ictaluri DasdA (pYA3493) AsdA + was recovered from the head-kidney, spleen and liver, indicating that E. ictaluri DasdA (pYA3493) AsdA + colonized these lymphoid tissues.

Discussion
To develop a balanced-lethal system in E. ictaluri, we first characterized the asdA and asdB genes present in the genome of E. ictaluri (Fig. 1). Deletion of the asdA gene precluded the growth of E. ictaluri in absence of DAP (Fig. 4), indicating that asdB does not encode for a functional protein related to DAP synthesis. This is consistent with the bioinformatic analysis (Fig. 2), which showed that the AsdB enzyme lacked several key amino acid residues at the catalytic active site.
The phylogeny of Asd has two branches, AsdA related with Gram-negatives and AsdB related with Gram-positives [28]. We found a particular group of non-functional AsdB genes in Edwardsiella and Yersinia. The common origin of AsdB in these bacteria suggests that the genes might have lost their activity through evolution, and that asdB could be considered a pseudogene in Edwardsiella and Y. pestis.
Suicide vector technology has been successfully used in several enteric bacteria to develop antibiotic-sensitive mutants [27]. Using this technology it was possible to construct defined deletion mutations in the absence of antibiotic-resistance markers for the first time in E. ictaluri (Fig. 3). During this process, we determined that E. ictaluri is extremely sensitive to Cm, even in the presence of the cat gene. The cat gene confers high-level resistance to Cm in most bacterial species. It codes for an enzyme called chloramphenicol acetyltransferase which inactivates Cm by covalently linking one or two acetyl groups, derived from acetyl-S-coenzyme A, to the hydroxyl groups on the chloramphenicol molecule [41]. This might indicate that chloramphenicol acetyltransferase is not functional or inefficient in E. ictaluri. Further studies are required to answer this.
The current live attenuated E. ictaluri vaccine is a rifampicinresistant strain [22]. Antibiotic resistance in live attenuated bacterial vaccines present a threat to both the animal and to human health, due to the horizontal transmission of genes, in this case by transduction. Recently lytic bacteriophages have been isolated from catfish ponds against E. ictaluri [42]. This suggests that temperate phages for E. ictaluri that can establish lysogeny might be present in these environments and could spread rifampicin resistance to native environmental bacterial flora. Here we have described a methodology to genetically engineer E. ictaluri without the use of antibiotic-resistance genes in the final strain. This advancement opens up the field of E. ictaluri live attenuated vaccine development and will provide opportunities for further research into the pathogenesis of this important organism.
Although, E. ictaluri DasdA01 is complemented with its own asdA gene, the complemented strain did not grow at the same rate as the parental wild-type strain, presenting a higher growth rate. To achieve the right amount of native AsdA in E. ictaluri using Asd + vectors requires further studies. However, E. ictaluri DasdA01 was fully complemented by the Salmonella SD-asdA gene, allowing the development of a balanced-lethal system.
One of the major difficulties in the construction of a balancedlethal system in E. ictaluri is the incompatibility of the Asd + vectors with cryptic plasmids present in the bacterial strain. E. ictaluri possesses two native autonomous small plasmids, pEI1 and pEI2 [43], that have been implicated in virulence [44]. The Asd + expression vectors were compatible with pEI1 and pEI2 native plasmids of E. ictaluri, indicating that the origin of replication of these plasmids, ColE1 ori and ColE2 ori-like, respectively [43], are compatible with p15A ori, pBR ori and pUC ori. E. ictaluri was described by Hawke in 1979 [45], and recently sequenced (NCBI's Entrez Genome database NC_012779). Most of its genes encode for putative functions. E. ictaluri possesses the machinery for the type II secretion system in its genome. Therefore we evaluated the secretion of proteins by using a blactamase signal sequence at the N-terminal end of a recombinant protein [19], a signal required for a protein to be secreted through the system mentioned above. Recombinant proteins, cloned in the AsdA + vector and using the b-lactamase signal sequence, were secreted in a similar fashion (Fig. 8) as for a Salmonella recombinant vaccine [37], suggesting that the type II secretion system in E. ictaluri is fully functional.  Salmonella DasdA mutants are totally attenuated in mice orally infected with 10 8 CFU per dose [19]. E. ictaluri DasdA01 mutants were not fully attenuated in catfish i.p. or orally infected (Table 4). Zebrafish i.m. infected with doses of 10 8 CFU succumbed to E. ictaluri DasdA01 mutant infection (Table 3). Lower doses of E. ictaluri DasdA01 mutants (10 6 -10 4 CFU) were totally attenuated (Table 3). It has been reported that E. ictaluri contain toxins, like hemolysin [46,47]. We believe that the mortality caused by E. ictaluri DasdA01 mutants, either in catfish or zebrafish, is due to a toxic shock-like effect caused by the toxins realized after this DAP dependent mutant lyse in vivo. These toxins probably are not LPS related, since fish and amphibians are resistant to the toxic effects of LPS [48,49]. E. ictaluri DasdA01 (pAsdA + ) was attenuated by one log-fold in catfish animal host model (orally infected), and two log-fold in zebrafish. The next step in the construction and design of a live recombinant E. ictaluri vaccine is the attenuation of the bacterial strain without altering colonization of lymphoid tissues and immunogenicity. From moribund orally infected catfish, E. ictaluri DasdA (pAsdA + ) were recovered from the  Table 3. Survival of zebrafish (D. rerio) infected with wild-type and E. ictaluri DasdA01 with and without Asd + vectors. head kidney, spleen and liver, indicating that E. ictaluri asdA (pAsdA + ) colonize lymphoid tissues. The increase in attenuation in catfish orally infected with E. ictaluri DasdA (pAsdA + ) could be used together with other genetic modifications to attenuate E. ictaluri in regard to constructing a live RAEV.
In summary, we have described methods to genetically engineer E. ictaluri without the use of antibiotic-resistant genes in the final strain. This opens up the field of RAEV development and will provide opportunities for further research into E. ictaluri pathogenesis. We have developed an antibiotic-sensitive recombinant E. ictaluri strain, using suicide vector technology [27] and Asd + expression vectors [19]. This first balanced-lethal vector-host system in E. ictaluri is key in constructing antibiotic-sensitive live RAEV for the catfish industry.

Ethics statement
All research involving fish was conducted as per Protocol #09-1042R, approved by the Arizona State University Institutional Animal Care and Use Committee.

Sequence analysis
Nucleotide Basic Local Alignment Search Tool (BLAST) was performed based on the sequences of the putative asd genes present in the genome sequence of E. ictaluri 93-146 accessed from NCBI's Entrez Genome database (NC_012779).
Asd sequences used were obtained from NCBI's Entrez Protein database. Amino acid sequence alignments were performed using the CLC Free Workbench software tool (v. 6.1 CLC bio A/S, Aarhus, Denmark). Protein structural-based alignments were performed by using the web-based interface for ESPript v.2.2 located at http://espript.ibcp.fr/ESPript/cgi-bin/ESPript.cgi [51]. Phylogenetic position of E. ictaluri AsdA protein was performed with CLC Free Workbench version using the unweighted pair group method with arithmetic mean (UPGMA). Bootstrap analysis was performed with 100 resamplings.

Construction and characterization of asdA mutants
The recombinant suicide vector pEZ102 (Table 2) carrying the linked flanking regions (59 361 bp and 39 422 bp) to generate an in-frame deletion of the asdA gene was constructed as described in [52]. The DasdA01 defined deletion mutation encompasses a 1,104 base pair deletion including the ATG start codon but not including the TAG stop codon. Primers (primer 1) 59-ACATG-CATGCAATGCCGTCAACGCCGCAGAAT-'3 and (primer 2) 59-CCGCTCGAGATGCACTCCTGCCTTGGATGGTGA -'3 were designed to amplify the upstream asdA flanking region (361 bp). A SphI site was included in the primer 1 (underlined) and a XhoI site was included in primer 2 (underlined). The downstream asdA flanking region (422 bp) was amplified by primers (primer 3) 59-CCGCTCGAGTGAGGCTACTGCTCTAGCCCGTGC -'3 and (primer 4) 59-TCGTCTAGAGCCAGATAGATTT-GATGTTGTCTCTGCTGC -'3. A XhoI site was included in primer 3 (underlined) and XbaI site was included in primer 4. The flanking regions were amplified from E. ictaluri J100, ligated, cloned into pRE112 and pMEG-375, and then digested with SphI and XbaI. The resulting plasmids were designated pEZ101 and pEZ102, respectively. To construct the E. ictaluri DasdA01 mutant, the suicide plasmid was conjugationally transferred from Escherichia coli x7213 [53] to E. ictaluri wild-type strains J100 and J102. Strains containing single-crossover plasmid insertions (E. ictaluri asdA::-pEZ102) were isolated on BHI agar plates containing Col, Amp, and DAP. Loss of the suicide vector after the second recombination between homologous regions (i.e., allelic exchange) was selected by using the sacB-based sucrose sensitivity counterselection system [27]. The colonies were screened for Amp s , Col r and for growth only in presence of DAP. DAPcolonies were screened by PCR using primer 1 and 4. Biochemical profiles of E. ictaluri strains were determined using the API 20E system (bioMériux, Marcy I'Etoile, France).

Complementation of asdA gene
The asdA gene of E. ictaluri, with and without its promoter, was cloned into a pAYCY184 vector [54] by inactivating the Tet cassette at the BamHI and XbaI restriction sites. The primers used to amplify asdA with its promoter (P asdA -asdA) were 59 -TCGTCTA-GATCTTGTAAGTTTGAGGATTA -39 (upstream) and 59 -CGGGATCCTCAGCATGCGGCGCAACGGCTC -39 (downstream). An XbaI and BamHI site were included in these primers, respectively (underlined). To amplify the E. ictaluri Shine-Dalgarno (SD)-asdA promoter-less the upstream primer 59 -TCGTCTA-GAAGGCAGGAGTGCATATGAAAAA -39 was used with the downstream primer previously described. An XbaI site was included in this primer (underlined). The E. ictaluri promoter-less asdA includes the SD AGGA region, 6 bp upstream from the ATG start codon (SD-asdA). The resulting plasmids, pEZ140 (SD-asdA) and pEZ146 (P asd -asdA) were used to complement different DasdA mutant strains. Also asd from Streptococcus mutans, cloned into pYA575 [36] and pYA248 [18], was used to evaluate complementation of E. ictaluri DasdA01 mutants.
To create a balanced-lethal system in E. ictaluri, several Asd + expression vectors harboring the SD-asdA gene sequence from Salmonella Typhimurium UK-1 with different origins of replication, (Table 2) [19] were transformed into E. ictaluri DasdA01 to evaluate their complementation and stability. The growth rate of the complementing strains was evaluated in the absence of DAP. Plasmid stability was evaluated for fifty generations as described by Konjufca et al. [55].

Expression of heterologous antigens by E. ictaluri DasdA01
Asd + expression vectors encoding different heterologous proteins (Table 2) were transformed into E. ictaluri DasdA01 to evaluate the expression and secretion of foreign proteins. First, the green fluorescent protein (GFP) was used to evaluate protein synthesis in the E. ictaluri DasdA01 strain. The vector pYA3994 AsdA + GFP + without a peptide secretion signal sequence was transformed into E. ictaluri DasdA01 ( Table 2). The synthesis of GFP was evaluated by fluorescent microscopy. The synthesis of LcrV and PspA was evaluated by western blot and the secretion was evaluated by subcellular fractionation [37].

Edwardsiella subcellular fractionation
Cultures were grown in BHI at 28uC static to an OD 600 of 0.6 and centrifuged at 5,865 g for 10 min. Periplasmic fractions were prepared by a modification of the lysozyme-osmotic shock method [58] as previously described [37]. The supernatant fluid was saved for analysis of secreted proteins. Equal volumes of periplasmic, cytoplasmic, and supernatant fractions and total lysate samples were separated by SDS-PAGE for western blot analysis.

Determination of LD 50 in zebrafish animal host
Zebrafish infections were performed by the methodology described by Petri-Hanson et al. [59] with modifications. The temperature of the water was 2661uC and the fish were acclimated during 2 weeks prior to the start of the experimentation. Adult zebrafish (average weight, 0.5 g) were sedated in 100 mg/L tricaine methanesulfonate (MS-222, Sigma) and then injected intramuscularly (i.m.). Groups of zebrafish (typically 15 fish per group) were injected i.m. with 10 ml of the bacterial suspension (10 3 -10 9 CFU) into each fish. A 3/10-cc U-100 ultrafine insulin syringe with a 0.5-in.-long (ca. 1-cm-long) 29gauge needle (catalog no. BD-309301; VWR) was used to inject the fish. Two sets of controls were used: fish that were injected with 10 ml of sterile phosphate-buffered saline containing 0.01% gelatin (BSG) [60] and fish that were not injected. Moribund fish demonstrating clinical signs were euthanized, necropsied, and bacteria isolated as previously described [59]. The fish were fed twice daily with TetraMin Tropical Fish Flake Feed. During the experiments, the fish were observed daily, and every other day water quality was monitored for pH and NO 2 with standard kits. The LD 50 was calculated by the method of Reed-Muench [61]. Fish care and use was performed in accordance with the requirements of the Arizona State University, Institutional Animal Care and Use Committee.

Determination of LD 50 in catfish animal host
Specific-pathogen-free channel catfish (Ictalurus punctatus) fingerlings were used with a mean weight of 18.561.3 g. The animals were randomly assigned to treatment groups of 6 to 8 fish each in 100 liter tanks. Each tank was equipped with a self-contained, recirculating, biofiltered, mechanical filtered, and U.V. water treated system with 12 h of illumination daily. The water temperature was 2861uC during the 2 weeks of acclimatization and during the experiments. The fish were fed daily with commercial Aquamax grower 400 (Purina Mills Inc., St. Louis, MO). During the experiments, the fish were observed daily, and every other day water quality was monitored for pH and NO 2 with standards kits. Catfish were infected with 10 3 to 10 9 CFU of E. ictaluri strains (fish were not fed until 1 h after infection) orally and intra peritoneal (i.p.). The fish were anesthetized with tricaine methanesulfonate (MS-222, Sigma; 100 mg/L of water) prior to handling. The LD 50 was calculated by the method of Reed-Muench [61]. Moribund animals were necropsied to evaluate presence of E. ictaluri in kidney, spleen and liver. Fish care and use was performed in accordance with the requirements of the Arizona State University, Institutional Animal Care and Use Committee.

Bacteria preparation
Bacterial strains were grown overnight in standing cultures that were diluted 1:20 in prewarmed BHI broth and grown with mild aeration (180 r.p.m.) at 28uC to an OD 600 of 0.8 to 0.9 (,10 8 CFU/ml). Bacteria were sedimented 10 min by centrifugation (5,865 g) at room temperature and resuspended in BSG [60] to densities appropriate for the inoculation.