Identification of Specific In Vivo-Induced (ivi) Genes in Yersinia ruckeri and Analysis of Ruckerbactin, a Catecholate Siderophore Iron Acquisition System

Abstract

This work reports the utilization of an in vivo expression technology system to identify in vivo-induced (ivi) genes in Yersinia ruckeri after determination of the conditions needed for its selection in fish. Fourteen clones were selected, and the cloned DNA fragments were analyzed after partial sequencing. In addition to sequences with no significant similarity, homology with genes encoding proteins putatively involved in two-component and type IV secretion systems, adherence, specific metabolic functions, and others were found. Among these sequences, four were involved in iron acquisition through a catechol siderophore (ruckerbactin). Thus, unlike other pathogenic yersiniae producing yersiniabactin, Y. ruckeri might be able to produce and utilize only this phenolate. The genetic organization of the ruckerbactin biosynthetic and uptake loci was similar to that of the Escherichia coli enterobactin gene cluster. Genes rucC and rupG, putative counterparts of E. coli entC and fepG, respectively, involved in the biosynthesis and transport of the iron siderophore complex, respectively, were analyzed further. Thus, regulation of expression by iron and temperature and their presence in other Y. ruckeri siderophore-producing strains were confirmed for these two loci. Moreover, 50% lethal dose values 100-fold higher than those of the wild-type strain were obtained with the rucC isogenic mutant, showing the importance of ruckerbactin in the pathogenesis caused by this microorganism.

Intensive aquaculture has led to an increase in the problems associated with infectious diseases in farmed fish. Yersiniosis, or enteric redmouth disease, is among the most widespread and causes economic losses all over the world. Its causative agent is the gram-negative bacterium Yersinia ruckeri, which mainly infects salmonids. This is a highly homogeneous species, but several serogroups have been identified (41), O1 being the most virulent. Carrier fish are the prime source of infection by shedding bacteria from their intestine, and transmission is also favored by the capacity of most wild strains to survive and remain infective in the aquatic environment (54), sometimes forming biofilms (13). Although fish are generally treated with inactivated whole-cell vaccines, outbreaks still occur under severe stress or poor hygiene conditions. Moreover, commercial vaccines do not confer protection against a new serogroup recently isolated in southern England (3).

Despite the importance of this syndrome, there is little information about the precise pathogenic mechanisms by which this bacterium is able to defeat the host defenses and cause disease. Until now, the only molecule whose involvement in virulence has been proved is the protease Yrp1 (20, 21), but it is only produced by Azo⁺ strains (47). Previously, Romalde et al. (42) showed that the extracellular products of Y. ruckeri, which included lipase, protease, and cytotoxic and hemolytic activities, reproduce some characteristic symptoms of this pathology, such as hemorrhage in the mouth and intestine. Besides, they proved the ability of Y. ruckeri to adhere to and effectively invade fish cell lines cultured in vitro. Other reports describe the production of a phenolate siderophore by this pathogen (40), as well as four iron-regulated outer membrane proteins (16). All these molecules are known to be important virulence factors in other microorganisms and could therefore constitute an interesting field of research.

Iron acquisition is a prerequisite for a successful colonization and invasion of many microbial pathogens (for a review, see references 19 and 59). Siderophores are small iron-chelating molecules and can be divided into three major classes, catecholates, hydroxamates, and heterocyclic compounds. In contrast to other members of the family Enterobacteriaceae, pathogenic Yersinia species (mouse-virulent Y. pestis, Y. pseudotuberculosis, and Y. enterocolitica biogroup 1B) do not produce either catecholate or hydroxamate siderophores. However, they are provided with uptake systems for both of them. The only siderophore that these strains are able to produce and utilize is the heterocyclic compound yersiniabactin, which has been related to pathogenicity.

Laboratory media appear to be a limited tool when trying to study the molecular mechanisms of disease, since it is very difficult to mimic the complex and changing environment of the host. In an attempt to overcome these limitations, Mahan et al. (32) developed a genetic selection device called in vivo expression technology. This technique is actually a promoter trap that uses the host as a selection medium, allowing the identification of genes expressed during infection but not under laboratory conditions. The in vitro expression approach has been successfully applied in other microorganisms (1) and has led to the discovery of novel virulence genes whose mutation produce an attenuation of the parental strain, for instance, the putative iron transport system SitABCD of Salmonella enterica serovar Typhimurium (29) or a transcriptional regulator of Pseudomonas aeruginosa (56). Originally, it was based on the in vivo complementation of purine auxotrophy (32), but later a new system based on antibiotic resistance was developed (33). This adaptation facilitates the use of this technique in many other pathogen-host systems with no need for identification of attenuating auxotrophies. For this reason, the vector pIVET8 (33) has been chosen to select gene fusions expressed specifically in fish tissues. This plasmid has promoterless cat and lacZY genes, which permit the selection of active promoters either in vivo or in vitro. In addition, the use of a suicide plasmid not only eliminates the high-copy-number effect of a replicative vector, but also avoids the risk of gene disruption of transposon insertion fusions.

In this paper, we report the optimization of an in vivo selection system for a fish-pathogenic species as well as the identification and sequencing of 14 ivi (in vivo-induced) genetic loci. It also shows a deeper analysis of those related to iron acquisition and tests the involvement in virulence of two of them. Moreover, it confirms the production and utilization of a virulence-involved catecholate siderophore by Y. ruckeri.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

The bacterial strains and plasmids used in this study are listed in Table 1. For the routine culture of Escherichia coli strains, 2×TY (tryptone/yeast) broth and agar were used, whereas Y. ruckeri strains were grown on either nutrient broth (NB), nutrient agar (NA), or eosin methylene blue (EMB) agar (Pronadisa). If required, the following compounds were added to the media: ampicillin (100 μg/ml), rifampin (50 μg/ml), kanamycin (25 μg/ml), streptomycin (50 μg/ml), 2,2′-dipyridyl (100 μΜ), or FeCl₃ (100 μΜ). Incubations were at 37°C for E. coli and 18 or 28°C for Y. ruckeri, with orbital shakers at 250 rpm. For the detection of siderophore production, chrome azurol S (CAS) plates and low-iron minimal medium M9 were prepared as described by Barghouthi et al. (4) and Romalde et al. (40), respectively. All glassware used to prepare these two media was first treated with 1 M HCl to remove iron traces and rinsed with water.

TABLE 1.

Bacterial strains and plasmids used in this study

Strain or plasmid	Characteristics	Source or reference
Yersinia ruckeri
150	Trout isolate, virulent strain	47
150R	Rifr derivative of 150	20
150R ivi I-XIV	Strains containing ivi fusions expressed only in the host	This study
150RrucC	RifrrucC::pIVET8 Apr	This study
150RrupG	RifrrupG::pIVET8 Apr	This study
Escherichia coli
DH5αλpir	F′ endA1 hsdR17 (rk− mk+) supE44 thi-1 recA1 gyrA (Nalr)	58
SM10λpir	thi thr leu tonA lacY supE recA::RP4-2-Tc::Mu Km::λpir	50
S17-1λpir	λ(pir) hsdR pro thi RP4-2 Tc::Mu Km::Tn7	50
S17-1λpir iviI-iviXIV	E. coli strains carrying iviI-XIV plasmids	This study
MT1694	HB101 derivative containing pRK2013	37
Plasmids
pIVET8	AproriR6K mob+ promoterless cat-lacZY	33
pRK2013	Tra+ Kmr	22
pLPY3	pIVET8::BglII (entC), Apr	This study
pLPY4	pIVET8::BglII (fepG), Apr	This study

Genetic techniques.

Routine DNA manipulation was performed as described by Sambrook et al. (44). Phage T4 DNA ligase and calf intestinal alkaline phosphatase were purchased from Roche Ltd., restriction enzymes were from Amersham Pharmacia Biotech Ltd., and oligonucleotides were from Sigma Aldrich Co.

A genomic library was constructed by ligating BglII-digested and dephosphorylated pIVET8 with Y. ruckeri 150 chromosomal DNA partially digested with Sau3AI and size fractionated in a sucrose gradient (20). The fractions selected for ligation had an average fragment size of 3 to 5 kb. The resulting ligation mixture was transformed into E. coli DH5αλpir. Pooled plasmid DNA was extracted and used to transform E. coli SM10λpir, from which the library was transferred to Y. ruckeri by filter mating (20). The transconjugants were pooled and conserved in 25% glycerol at −70°C.

Experimental infections.

Rainbow trout (Oncorhynchus mykiss) weighing from 10 to 15 g were kept in 60-liter tanks at 18 ± 1°C in continually flowing dechlorinated water. Six groups of 30 fish were injected intraperitoneally with doses of 10⁴ transconjugant cells in 100 μl of phosphate-buffered saline. At 24 and 48 h postinfection, fish were given intramuscular injections of 50 μl of 6-mg/ml chloramphenicol; 72 h postinfection, the animals were sacrificed and dissected, and the liver, spleen, and intestine were homogenized in NB with a stomacher. Afterwards, the pool of bacteria was grown in NB supplemented with ampicillin plus rifampin and used for a second round of infection in the same conditions.

Analysis of ivi clones.

Lac⁻ clones recovered in EMB plates after the second round of infection were selected for further analysis. Plasmid DNA was recovered from Y. ruckeri chromosome by triparental mating (37). Briefly, cultures of the Y. ruckeri fusion-containing strain, an E. coli π⁺ recipient (S17-1λpir), and an E. coli strain harboring a helper plasmid (pRK2013) that encodes Tra were mixed in equal amounts (100 μl of overnight cultures), pelleted, plated onto a 2×TY plate, and incubated at 28°C for 6 h. Transconjugants were selected in plates with both ampicillin and streptomycin at the concentrations mentioned above. Plasmid DNA obtained from the resulting strains was digested with BamHI to differentiate allelic groups. The nucleotide sequences upstream of the cat and bla genes was sequenced with the initial primers catseq-2 (5′-CGGTGGTATATCCAGTG-3′, corresponding to nucleotides 31 to 15 of the cat gene) and blaseq (5′-GGGAATAAGGGCGACAC-3′, corresponding to nucleotides 36 to 20 of the bla gene). Subsequent primers were derived from ivi sequences. For gene rucC, a degenerate reverse primer called rucCRP (5′-ACNACCCAYTCNCCRTT-3′) was designed from homologous proteins in order to obtain more of the coding sequence (N is A, C, G, or T; Y is C or T; R is A or G). It was used to perform a PCR with forward primer rucCFP (5′-TCACGGGCGGTAGAAAC-3′) (T_a = 47°C, t_e = 1 min). An amplicon of about 600 bp was obtained and sequenced.

DNA sequencing was performed by the dideoxy chain termination method with the DR Terminator kit (Applied Biosystems) according to the manufacturer's instructions in an ABI Prism 310A automated DNA sequencer from Perkin-Elmer. Sequences were compared to those in the databases with the Blastx program. For promoter activity studies, the iviII and iviIII clones were grown in M9, M9 supplemented with 2,2′-dipyridil, or M9 supplemented with FeCl₃ at either 18 or 28°C. Samples from stationary-phase cultures were collected, and β-galactosidase activity was measured as described by Miller (34). After an analysis of variance test, P values of <0.05 were considered significant.

Construction and analysis of insertion mutants.

Internal fragments of the predicted open reading frames (ORFs) of rucC (449 bp) and rupG (604 bp) were amplified by PCR with the following primers: forward primers rucC1 and rupG1 (5′-GCCAAGATCTCGTAGTTTGCCGAACGA-3′, nucleotides 361 to 377 in bold type, and 5′-GCCAAGATCTGCTACGATGGCGCTGAT-3′, nucleotides 238 to 255 in bold type), respectively, and reverse primers rucC2 and rupG2 (5′-CGTTAGATCTTGAATGAGTGATGGCTG-3′, nucleotides 809 to 793 in bold type, and 5′-CGTTAGATCTGTGCGGCCAGTGCAATA-3′, nucleotides 841 to 824 in bold type), respectively. The reaction was performed in a Perkin-Elmer 9700 GeneAmp thermocycler with an initial denaturation cycle at 94°C for 5 min, followed by 25 cycles of amplification (denaturation at 94°C for 30 s, annealing at 46 and 49°C for rucC and rupG, respectively, for 30 s, and extension at 72°C for 1 min), and a final 7-min elongation step at 72°C. All primers contained a BglII site (italic) and four additional bases at their 5′ end. The amplicons generated were digested with BglII and ligated into pIVET8 previously digested with the same enzyme and dephosphorylated.

The ligation mixture was used to transform electrocompetent cells of E. coli SM10λpir. Clones containing inserts were used to conjugate with Y. ruckeri 150R to obtain the mutants as previously described (20). These mutations were, then confirmed by Southern blot analysis after digestion of genomic DNA from mutant and wild-type strains with BamHI. The amplicons mentioned above were used as probes. Mutant clones were tested for their production of siderophores in CAS-agar plates as well as by the Arnow assay (2). Afterwards, 50% lethal dose (LD₅₀) tests of both wild-type and mutant strains were carried out as described by Fernández et al. (20) to determine the involvement in virulence of the genes tested. Briefly, groups of 10 fish (Oncorhynchus mykiss) weighing between 8 and 10 g were kept in 60-liter tanks at 18 ± 1°C. They were challenged by intraperitoneal injection of 0.1 ml of serial 10-fold dilutions as described by Fernandez et al. (20). The LD₅₀ determinations were calculated by the method of Reed and Muench (38).

Nucleotide sequence accession numbers.

The sequence accession numbers for the iviI to iviXIV genes in the GenBank database are AY576529 to AY576542, respectively.

RESULTS

Isolation of transcriptional fusions induced in vivo.

Fragments with an average size of 3 to 5 kb were ligated into the single BglII site of pIVET8. The ligation mixture was then electroporated into E. coli DH5αλpir, and at least 67,000 individual transformants were pooled. Randomly selected clones were checked, showing that about 85% of the population contained inserts. Plasmid DNA obtained from the pool was transformed into E. coli SM10λpir, and the library was then transferred to Y. ruckeri 150R by means of filter mating. Since Y. ruckeri lacks the Pi protein necessary for the replication of pIVET8, transconjugants should have integrated the plasmid in their chromosome by a single-crossover recombination, with the cloned bacterial DNA as a source of homology (Fig. 1). About 300,000 individual exconjugants were obtained in 10 independent conjugation experiments.

FIG. 1.

Schematic representation of in vitro expression selection in fish. Approximately 10⁴ cells of Y. ruckeri transcriptional fusions were injected intraperitoneally in rainbow trout (O. mykiss). At 24 and 48 h postinfection, chloramphenicol (6 mg/ml) was injected intramuscularly, and the fish were sacrificed 24 h later. Bacterial cells were recovered from fish tissue, and a new round of infection was carried out. Then, cells were plated on EMB, and Lac⁻ colonies were identified. Triparental mating was used for plasmid rescue from the ivi clones, and sequencing of the cloned fragments was carried out with primers blaseq and catseq-2 from the contiguous bla and cat genes, respectively.

These 10 independent pools were used for experimental infections as described above. The ability of chloramphenicol to select active promoters in vivo was tested by the increase in Lac⁺ colonies after two rounds of infection, which is known as the red shift (32, 33). The average percentage of Lac⁻ colonies in the transconjugants pool was 57.5%, whereas it fell to 1.5% after the second passage through fish. In the 10 experiments, a total of 70 ivi clones were isolated.

Analysis of ivi clones.

Clones were rescued from the Y. ruckeri chromosome through triparental mating with E. coli S17-1λpir as a recipient strain. Afterwards, plasmid DNA from the different clones was obtained and analyzed after digestion with BamHI. This allowed classification into 14 allelic groups. The sequence of fragments 400 to 600 bp upstream of cat gene (Fig. 1) was determined and compared with the sequences in the databases in all six reading frames by the Blastx program (Table 2). Not surprisingly, all of them were novel sequences, since only a few genes have been sequenced in this pathogen.

TABLE 2.

In vivo-induced Y. ruckeri genes^a

Gene	Database homology (accession no.)	Predicted function	Amino acid identity (%)
iviI	Fct/Erwinia chrysanthemi (Q47162)	Ferrichrysobactin receptor precursor	52
iviII	EntC/MenF/Chromobacter violaceum (NP_901155)	Isochorismate synthase	42
iviIII	FepC/Y. enterocolitica (AAD29087)	Putative transport protein	80
iviIV	Hypothetical protein/E. coli (NP_753126)	Putative hemagglutinin or hemolysin	49
iviV	ShlB/Serratia marcescens (P15321)	Hemolysin activator protein	48
iviVI	No significant similarity
iviVII	TonB complex protein/Y. pestis (NP_670792)	Transport of small molecules	74
iviVIII	No significant similarity
iviIX	TctC/P. luminescens (NP_930410)	Tricarboxylic transport	74
iviX	Hypothetical protein/E. coli (NP_755736)	Putative inner membrane protein	60
iviXI	No significant similarity
iviXII	TraI/Serratia entomophila (NP_938092)	Relaxase	57
iviXIII	TadD-like protein/Y. pestis (NP_670787)	Required for aggregation	54
iviXIV	YdhC-like/Y. pestis (NP_405925)	Putative drug resistance protein	54

^aOnly those open reading frames in the same strand as cat-lacZY were taken into account. Genes listed in this table are the ones immediately upstream of the cat gene in each clone.

Four of these clones (iviI, iviII, iviIII, and iviVII) contained putative ORFs which showed high homology with genes involved in iron uptake through siderophores in other microorganisms. They were analyzed more deeply in order to characterize this system in Y. ruckeri as well as its involvement in virulence.

iviIV, iviV, iviXII, and iviXIII had significant similarity to genes whose products have been related to virulence in other bacteria, a putative hemagglutinin or hemolysin, a hemolysin activator protein, a putative type IV secretion system, and proteins involved in tight adherence, respectively. Thus, the deduced iviIV translational product had significant homology with members of the ShlA/HecA/FhaA exoprotein family, such as a putative E. coli protein (NP_753126), a putative Y. pestis hemagglutinin-like secreted protein (NP_669014), and Erwinia chrysanthemi hemolysin/hemagglutinin-like protein HecA (39), with 49, 46, and 33% identity, respectively. The iviV locus showed 48, 48, and 38% identity with hemolysin secretion/activation proteins such as Serratia marcescens ShlB (36), Photorhabdus luminescens PhlB (7), and Edwardsiella tarda EthB (26). iviXII showed 57 and 53% identity with the relaxase TraI from Serratia entomophila (27) and S. enterica serovar Typhimurium (31) and 31% with DotC from Legionella pneumophila (55). The iviXIII ORF had the highest homology with proteins involved in tight adherence such as a hypothetical TadD-like protein from Y. pestis (NP_670787) and TadD from Actinobacillus actinomycetemcomitans (30) and Pasteurella multocida (NP_245783), having 54, 47, and 46% amino acid identity, respectively.

Other ivi genes (iviIX, iviX, and iviXIV) had similarity with genes not related to virulence or putative translational products not studied yet. Thus, iviIX showed high homology with the TctC protein of the tripartite tricarboxylate transporter family, such as those of P. luminescens (NP_930410), S. enterica serovar Typhimurium (NP_461712), and Pseudomonas putida (NP_743576), with 74, 77, and 71% identity, respectively. iviX had a significant similarity with hypothetical transmembrane proteins from E. coli (NP_755736), Shigella flexneri (NP_708914), and S. enterica serovar Typhimurium (NP_462152), having 60, 60, and 59% identity, respectively. The iviXIV locus was similar to several putative drug resistance proteins such as those from E. coli (17), P. luminescens (NP_929842), and Y. pestis (NP_405925), with 55, 54, and 45% amino acid identity, respectively.

Finally, there were three clones (iviVI, iviVIII, and iviXI) with no significant homology to known sequences on the minus strand.

Genetic organization of Y. ruckeri genes involved in siderophore-mediated iron uptake.

As mentioned above, some of the in vivo-induced genes (iviI, iviII, iviIII, and iviVII) isolated in Y. ruckeri might be involved in iron acquisition by a catechol siderophore. Following an initial screening of the genes directly upstream of the reporter cat-lacZY cassette, further sequencing analysis was carried out in order to identify the presence and relative location of other genes essential in siderophore-mediated iron uptake.

Partial sequence analysis upstream of the cat gene (Fig. 2) in the fragment cloned in iviI showed the presence of a gene (rupA) encoding a putative protein with significant similarity to TonB-dependent receptors. It had the highest homology, 52%, with the ferrichrysobactin receptor Fct from Erwinia chrysanthemi (45), which, despite being a catechol siderophore receptor, showed greater similarity with receptors of hydroxamate chelators than with those of other catecholates such as E. coli enterobactin. So does the Y. ruckeri putative receptor, which is 46 and 42% homologous to the ferrioxamine B receptor FoxA from Y. enterocolitica (5) and ferrichrome-iron receptor FhuA from E. coli (15), respectively.

FIG. 2.

Genetic organization of iviI, iviII, and iviIII clones from Y. ruckeri containing the ruckerbactin biosynthetic uptake cluster, and comparison with the E. coli enterobactin gene cluster. Thick arrows represent genes identified in both microorganisms.

The putative translational product of the sequence just upstream of cat in iviII (Fig. 1) showed high homology with isochorismate synthases, the enzymes that catalyze the first step in the biosynthetic pathway of catechol siderophores. The cloned fragment plus an additional 600 bp amplified by PCR allowed us to identify the initial 338 amino acids of a protein (RucC) highly homologous to EntC from Chromobacterium violaceum (NP_901155), Brucella suis (NP_699223), and E. coli (AAB40793), with 50, 46, and 44% identity, respectively. The sequence upstream of rucC was obtained, revealing the presence of a gene homologous to fepB (rupB), and the deduced initial 115 amino acids of this protein showed 58, 58, and 56% identity with FepB from C. violaceum (NP_901909), S. enterica serovar Typhimurium (NP_455170), and E. coli (NP_308658), respectively. The FepB protein binds the siderophore and delivers the ligand to the corresponding inner membrane permease complex.

The initial analysis of iviIII revealed the presence of a putative ORF encoding one of the components of an ABC transporter system responsible for the mobilization of the iron-siderophore complex from the periplasm into the cytoplasm. The product of this gene (rupC) had the highest homology with FepC from Y. enterocolitica (46) and C. violaceum (NP_901904) and Vibrio cholerae ViuC (60), with 80, 63, and 58% identity, respectively. The deduced sequence of the initial 146 amino acids contained one of the two conserved amino acid motifs characteristic of the ATP-binding component of this type of transporter, known as Walker motif A, GPNACGKS (amino acids 48 to 55). It had only one residue different from that of E. coli (A instead of G) and was identical to that of Y. enterocolitica. The second ORF encoded a protein (RupG) homologous to FepG from Y. enterocolitica (46), E. coli (NP_286316), and C. violaceum (NP_901905) with 60, 48, and 46% identity, respectively.

The next ORF (rupD) encoded a protein similar to the third component of this type of transporter, FepD. It showed 80, 53, and 53% identity with FepD from Y. enterocolitica (46), E. coli (NP_286317), and S. flexneri (NP_706446), respectively. These two proteins, FepG and FepD, form the integral membrane permease complex that transports the siderophore across the inner membrane. It is likely that these three genes (rupDGC) are transcribed as a single operon encoding the ruckerbactin transport complex. Upstream of fepD (Fig. 2) but in the opposite orientation, another ORF was found. It has only been sequenced partially, showing 65% amino acid identity with the initial 61 amino acids of a probable POT family transport protein from C. violaceum (NP_901907) and P43 from S. flexneri (NP_706447) and E. coli (49). This kind of protein has recently been proven to be involved in enterobactin secretion to the extracellular environment and has been designated EntS (24).

The iviVII predicted ORF would encode a protein belonging to the TonB complex, necessary to provide the energy for iron transport at the outer membrane. The percentage of identity was 74, 69, and 68% to ExbB from Y. pestis (35), E. coli (18), and S. flexneri (51), respectively. In summary, sequence analysis of these four clones revealed the presence of genes responsible for ruckerbactin biosynthesis (rucC), secretion (rucS), reception (rupA), periplasmic binding (rupB), transport to the cytoplasm (rupDGC), and energy transduction (exbB). In addition, it disclosed the hypothetical chromosomal location of some of these genes, showing the great similarity to that of enterobactin in E. coli (Fig. 2).

In vitro regulation of rucC and rupDGC by iron and temperature.

The four ivi clones containing genes involved in siderophore-mediated iron uptake were grown in EMB plates with either iron or 2,2′-dipyridyl. All of them were Lac⁻ in the iron-rich medium, but only three of them (iviI, iviII, and iviVII) had strong β-galactosidase activity in the iron-depleted medium. On the contrary, iviIII showed only weak activity in the presence of the chelator.

To study these two patterns of iron regulation more deeply, β-galactosidase activity assays were carried out with the rucC(iviII), involved in the biosynthesis of ruckerbactin, and rupDGC(iviIII), the membrane transport components, clones. The results (Fig. 3A, a and b) showed that both operons are regulated by iron availability and temperature. Expression was induced under low-iron conditions, as expected, but in the rupDGC clone (P = 1.31 e⁻⁵), the induction was much weaker than in rucC (P = 5.4 e⁻⁶). Besides, the expression levels when the growth temperature was 18°C were higher than at 28°C (Fig. 3A, a and b) (P values of 0.009 and 0.04 for rupG and rucC, respectively).

FIG. 3.

Anaysis of rucC and rupDGC loci from the ruckerbactin cluster of Y. ruckeri. (A) β-Galactosidase activity of rupDGC (a) and rucC (b) lacZ fusions was measured in cells grown at 18°C (solid bars) and 28°C (open bars) in minimal medium (M9) containing: FeCl₃ (100 μM) (column 1), M9 (column 2), or 2,2′dipyridyl (100 μM) (column 3). The results are the averages of three independent experiments. Note that the data are plotted in different scales. (B) Siderophore production in CAS-agar plates by strains 150RrucC (rucC insertion mutant) (lane 1), 150R (wild type) (lane 2), and 150RrupG (rupG insertional mutant) (lane 3).

Analysis of rucC and rupG mutants and presence of these genes in different Y. ruckeri strains.

Isogenic rucC and rupG mutant strains were obtained, checked by Southern blot analysis, and tested for the production of siderophores in CAS-agar plates. Chromosomal DNA from the wild-type and mutant strains digested with BamHI was analyzed by hybridization with the PCR amplicons from rupG and rucC as probes. Both mutant strains showed a different hybridization pattern from the wild-type strain with the respective probes, confirming that gene interruption had occurred as expected. The CAS-agar plate assay for the production of siderophores showed that the rucC mutant strain did not produce them at all, whereas the rupG mutant had a much larger halo than the wild-type strain (Fig. 3B), indicating ruckerbactin overproduction. Identical results were obtained with the Arnow assay, confirming the catecholic nature of the siderophore activity detected in CAS-agar plates. Thus, the rucC mutant did not have any catecholate in its supernatant after growth in low-iron conditions, whereas the wild-type and rupG mutant strains produced even more of this kind of molecule.

Both mutant strains showed retarded growth compared to the wild-type strain in iron-depleted medium and were unable to reach the same cell density. The maximum optical density was 50% of that of the wild-type strain in the presence of the iron chelator 2,2′-dipyridyl. Nevertheless, the mutants behaved differently when assayed for their virulence for fish. The LD₅₀ obtained with the rucC mutant was 100-fold higher than with the wild-type strain (Table 3). In contrast, the LD₅₀ of the rupG mutant was only slightly higher than that of the wild-type strain (Table 3).

TABLE 3.

Quantitative infection after intraperitoneal injection of rainbow trout with the parental (150R) and mutant (150RrucC and 150RrupG) strains^a

Inoculum (100 μl)	No. of fish dead/10 in group
	150R		150RrucC		150RrupG
	A	B	A	B	A	B
107	10	10	10	8	10	10
106	10	10	1	2	9	10
105	10	8	0	1	8	4
104	1	1	1	0	0	0
103	1	2	0	0	0	0
LD50	2.45 × 104	2.83 × 104	2.45 × 106	2.74 × 106	4.87 × 104	1.47 × 105

^aLD₅₀ was calculated for each experiment 7 days after the infection by the method of Reed and Muench (38). Groups of 10 fish were injected intraperitoneally with doses ranging from 10³ to 10⁷ cells. Columns A and B correspond to dead fish from two independent experiments.

Twelve Y. ruckeri strains (146, 147, 148, 149, 150, 3585, 4319, 1386, 955, 956, A100, and A102) (21, 47) were positive for the production of siderophores in CAS-agar plates. In addition, PCR assays for the detection of rucC and rupG confirmed the presence of these genes in all the strains tested, even though three different plasmid profiles, represented by strains 955, 956, and the rest of the strains, respectively, were analyzed.

DISCUSSION

Bacterial gene expression is, in many cases, a response to specific environmental stimuli. During the pathogen-host interaction, a group of bacterial genes contribute to facilitate colonization, invasion, and definitive infection progression. The expression of some of these genes is enhanced or occurs exclusively in vivo. An in vitro expression strategy based on plasmid pIVET8 (33) was employed as a selection system for promoters induced during the infection of fish by Y. ruckeri. This vector allowed the selection of active promoters in vivo by chloramphenicol treatment and the isolation of those inactive in vitro with the lacZY cassette as a reporter. The most limiting factor in this technique is the determination of the appropriate conditions for antibiotic selection in the host due to the pharmacokinetics of the drug. As this is the first time an in vitro expression system has been used in fish, parameters such as the number of bacteria for intraperitoneal injections, the antibiotic dose for selection, and the organs used for the recovery of bacteria were first established. Before and after two rounds of fish infection, screening on EMB plates indicated the effective selection of active promoters in vivo because of the significant increase in the proportion of Lac⁺ colonies, a phenomenon known as the red shift (32, 33).

Fourteen loci were selected, and most of them encoded proteins which had homology to known virulence factors from other pathogens. Others were similar to proteins whose involvement in virulence was not known or unclear.

Two loci (iviIV and iviV) encode proteins that belong to the two-partner secretion family that includes several cytolysins and adhesins (28). One (iviV) is homologous proteins involved in the activation and secretion of Serratia-type hemolysins, which form pores not only in erythrocytes but also in fibroblasts and epithelial cells and have been related to invasiveness (25). Both hemolytic and cytotoxic activities have been described in the exocellular products of Y. ruckeri (42) that were able to reproduce some of the symptoms of the disease, such as hemorrhages in the gills and intestine. The other (iviIV) encodes a putative pore-forming toxin of this type, but its function as a hemolysin or as an adhesin is still to be determined.

Another interesting locus is the one encoding a protein similar to TadD (iviXIII), which is involved in the tight adherence properties of A. actinomycetemcomitans and is part of a gene cluster responsible for the synthesis, assembly, and transport of Flp pili. These proteins have been proposed to constitute a new subfamily of type IV secretion systems (6) and seem to be involved in virulence, since a tadD mutant of Pasteurella multocida was reported to be attenuated for virulence in mice (23). Perhaps this system is responsible for the previously described biofilm-forming capacities of some Y. ruckeri strains (13), being necessary to survive in the environment. On the other hand, the in vivo activation of these genes suggests their putative role in adherence to host tissues.

Type IV secretion systems built from core components of conjugation machines and transport proteins or protein-DNA complexes have been demonstrated to be involved in virulence in other microorganisms, such as Agrobacterium tumefaciens (43), Helicobacter pylori (14), and Legionella pneumophila (61). iviXII shows homology with the lipoproteins TraI and DotC, involved in secretion systems of this type. Several studies show that the Tra operon is often induced by extracellular or growth phase-dependent signals (for a review, see references 8 and 57). Recently, a conjugative transfer system has also been described in Y. enterocolitica (53).

Metabolic genes induced by the presence of a particular substrate could be selected as ivi genes if the inductor molecule is present in the host and not in the culture medium. This could be the case with the tctC homologue. TctC is part of a tripartite tricarboxylate transporter system that has not been well characterized yet. S. enterica serovar Typhimurium tctC mutants showed poor growth in the presence of citrate, one of the binding substrates of this protein (52). The isolation of an orthologous tctC in Y. ruckeri as an iviIX gene and its induction in vivo suggest that it plays an important role in obtaining a carbon and energy source in the host.

iviX and iviXIV are putative transmembrane proteins whose involvement in transport of molecules such as proteins and metals or in mechanisms of drug resistance is still to be analyzed.

Counterparts of the loci fct (TonB-dependent receptor), entC (isochorismate synthase), fepDGC (ABC transport system), fepB (periplasmic iron-siderophore binding protein), entS (involved in secretion of the siderophore), and exbB (TonB complex component), all involved in iron piracy, were identified in Y. ruckeri. This confirms that this microorganism produces a catechol siderophore (ruckerbactin), as previously suggested by Romalde et al. (40). This system turned out to be similar to that of E. coli enterobactin not only in terms of amino acid sequence homology but also in the chromosomal arrangement of some of these genes. In addition, high homology was found between the ruckerbactin receptor (RupA) and the ferrichrysobactin receptor (Fct) from E. chrysanthemi (45), showing more similarity with hydroxamate receptors than with other catecholate receptor proteins such as FepA from E. coli. This suggests that ruckerbactin is more similar to chrysobactin than to enterobactin in chemical structure.

However, this differs from other pathogenic yersiniae that synthesize only the heterocyclic compound yersiniabactin in spite of its ability to transport and utilize other siderophores belonging to the catecholate and hydroxamate families. For instance, they are endowed with a gene cluster encoding the proteins needed for enterochelin uptake (fepA, fepDGC, and fepB). However, the genes required for enterochelin biosynthesis and secretion seem to have been deleted. All these data, together with the involvement of protease Yrp1 in virulence (20), show that, in terms of pathogenicity mechanisms, Y. ruckeri is closer to E. chrysanthemi than to other yersiniae.

The absence of a halo in CAS-agar plates around the rucC mutant colonies indicates that this is the only high-affinity iron-scavenging system present in this microorganism. In contrast, the rupG mutant strain showed a larger halo than the wild-type strain, indicating overexpression of the siderophore biosynthetic pathway. As expected, rucC and rupG were both upregulated under iron-limited conditions in vitro. This induction was temperature dependent, being higher at 18°C, the infection temperature, than at 28°C, the optimal growth temperature, as occurs with Y. ruckeri protease Yrp1 (20) and other genes from fish pathogens (11, 48). Nevertheless, induction of rupG was weaker than of rucC, in agreement with data published by Christoffersen et al. (9) for E. coli, where expression levels of rupDGC were significantly lower than those of the other enterobactin transcription units. Siderophore production seems to be a conserved character in this species, because 12 strains from different origins were positive in the CAS-agar plate assay. Additionally, PCR analysis showed that rucC and rupG were present in Y. ruckeri strains with three different plasmid profiles, which suggests but does not prove that the ruckerbactin gene cluster is not plasmid borne.

This study has proved that ruckerbactin is one of the iron uptake systems that this pathogen utilizes in the host and is also involved in virulence. First, it confirms that genes involved in the siderophore pathway are upregulated during the infection of fish. On the other hand, the LD₅₀ of a rucC mutant strain was 100-fold higher than that of the wild-type strain. Strikingly, the rupG mutant strain LD₅₀ was similar to that of the wild-type strain despite its retarded growth in vitro in iron-depleted medium, like the rucC mutant. Perhaps this mutant has altered overexpression in other systems able to supply iron in the host, with different sources such as hemoglobin. The connection between siderophores and iron acquisition in the virulence of bacteria is well established (19, 59), and studies on in vivo induction of genes related to iron in fish-pathogenic microorganisms has already been described. Thus, iron limitation during in vivo growth induces the expression of outer membrane proteins in Vibrio salmonicida (10). In the same way, studies carried out in Vibrio anguillarum 01 showed that iron acquisition by siderophore is involved in the virulence of this bacterium (12).

In summary, this study has led to the identification of 14 transcription units specifically induced in the fish pathogen Y. ruckeri during infection. Four of these loci turned out to be part of a catechol-type siderophore system. The chemical structure and biosynthetic route of this molecule, called ruckerbactin, are still to be determined. Most of the remaining genes were identified as putative virulence factors according to their sequence homology with known virulence factors from other pathogens. This work may be the basis for future analysis of these genes in order to have a general picture of the main pathogenicity determinants in this bacterium.