The bacterial strains and plasmids used in this work are listed in Table 1. V. anguillarum strains were grown at 25°C or 15°C in tryptic soy agar (TSA) or broth (TSB) (Cultimed) supplemented with 1% NaCl. Escherichia coli strains were grown in Luria-Bertani broth or agar (Cultimed) at 37°C. When required, antibiotics were added at the following final concentrations: kanamycin 50 μg ml^–1, ampicillin sodium salt 60 μg ml^–1 or 100 μg ml^–1, and gentamycin 15 μg ml^–1.

In-frame deletions of araC1, araC2, h-ns, and toxR-S were constructed by allelic exchange in V. anguillarum RV22 strain in a ΔvabF background (impaired to synthesize vanchrobactin) as previously described (Balado et al., 2006). The flanking regions of each gene were amplified by PCR and cloned into the vector pWKS30; the resulting constructions were ligated into the suicide vector pNidKan. The resulting plasmid was conjugated with RV22ΔvabF strain and selected based on ampicillin and kanamycin resistance. A second event of recombination was performed, and the mutants were selected based on sucrose (15%) resistance. A PCR was performed to confirm the allelic exchange event. This process led to the formation of V. anguillarum mutant strains RV22ΔvabFΔaraC1, RV22ΔvabFΔaraC2, RV22ΔvabFΔh-ns, and RV22ΔvabFΔtoxR-S. For araC1 mutant complementation, araC1 was amplified by PCR, cloned into the vector pSEVA651 in E. coli S17-1 λpir, and mobilized to the appropriate mutant strain by conjugation. To restore the original phenotype, the WT genes h-ns and toxR-S were cloned into the suicide vector pNidKan and the complementation was accomplished as indicated above for the construction of mutants. The oligonucleotides used are listed in Supplementary Table 1.

Growth ability assays were performed in CM9 medium supplemented with 10 μM FeCl₃ to achieve iron excess or with 25 or 75 μM 2,2′-dipyridyl (TCI) to achieve iron deficiency. V. anguillarum strains RV22ΔvabF, RV22ΔvabFΔaraC1, RV22 ΔvabFΔaraC2, RV22ΔvabFΔh-ns, RV22ΔvabFΔtoxR-S, and complemented strains were grown overnight in TSB-1. Each culture was adjusted to an OD₆₀₀ = 0.5 and a 1:50 dilution was inoculated in CM9 medium. The resulting cultures were incubated at 15°C with shaking at 120 rpm. Growth was recorded after 48 h.

Bacterial cultures grown in CM9 medium with 25 μM 2,2′-dipyridyl and at an OD₆₀₀ ∼ 0.8 were used to measure siderophore production with the chrome azurol-S (CAS) liquid assay (Schwyn and Neilands, 1987). Briefly, supernatants were obtained by pelleting bacterial cells using centrifugation and equal volumes of these supernatants were incubated with the CAS reagent at room temperature for 15 min. The quantification was performed by measuring A₆₃₀ in a spectrophotometer (Hitachi).

The regions immediately upstream of Photobacterium damselae subsp. piscicida (DI21 strain) araC1 and frpA genes were amplified by PCR and fused to a promoterless lacZ gene in the low copy number plasmid pHRP309. This process leads to the construction of plasmid pLP28 and plasmid pLP9 carrying constructs araC1_pdp::lacZ (ParaC1_pdp) and frpA_pdp::lacZ (PfrpA_pdp), respectively. Plasmids pMB277 and pMB276 carrying the lacZ fusions of V. anguillarum araC1_ang::lacZ and frpA_ang::lacZ (ParaC1_ang and PfrpA_ang promoters) were previously constructed (Balado et al., 2018). The constructs were mobilized from E. coli S17-1 λpir to V. anguillarum (RV22ΔvabF and its derivative mutants) and P. damselae subsp. piscicida (DI21 strain) by conjugation. The presence of the promoters was confirmed by PCR. The resulting V. anguillarum strains and P. damselae subsp. piscicida carrying the promoter fusions were grown under weak iron restriction using CM9 medium supplemented with 25 μM 2,2′-dipyridyl. When the bacterial cultures reached an OD₆₀₀ = 0.3, the β-galactosidase activities were measured by the method of Miller (1992). The results shown are means of three independent experiments.

Overnight cultures of RV22ΔvabF and RV22ΔvabFΔaraC1 were grown in TSB-1 and adjusted to OD₆₀₀ = 0.5. A 1:50 dilution was inoculated in 10 ml CM9 medium supplemented with 25 μM 2,2′-dipyridyl. As the bacterial cultures reached an OD₆₀₀ = 0.8, they were pelleted at 4,000 rpm, for 30 min at 4°C. The pellet was resuspended in 5 ml of 10 mM Tris-HCl and 0.3% NaCl, pH 8.0. Cellular disruption was accomplished by sonication on ice (five cycles of 30 s). Then, the samples were centrifuged at 4,000 rpm for 30 min at 4°C to eliminate the cellular debris. For the isolation of membrane proteins, 1% sarkosyl was added to the supernatant and incubated at room temperature for 30 min. The samples were centrifuged at 40,000 rpm for 30 min at 4°C and the pellet was resuspended in 20 μl of water. The samples were mixed 1:1 with SDS-PAGE loading buffer and loaded into a 12% polyacrylamide gel. After the separation by SDS-PAGE, the proteins were transferred onto a PVDF membrane as previously described. The membrane was blocked in blocking buffer (5% skim milk in TBST, Tris Buffered Saline with Tween 20) for 1 h at room temperature with shaking. The membrane was incubated overnight in the primary antibody solution against the target protein (1:10,000 dilution of the anti-FrpA antibody). After rinsing the membrane with TBST, it was incubated with the secondary antibody (1:10,000 anti-rabbit IgG HRP conjugate antibody) for 1 h at room temperature with shaking. The signal was detected using the Clarity^TM Western ECL substrate (Bio-Rad). Rabbit polyclonal antibodies against the external loop 6 (short peptide PGGFSPAPRSSGDKNGYSP) of FrpA (anti-FrpA) were purchased from GenScript.

Experimental infections were performed using Senegalese sole (Solea senegalensis) fingerlings of approximately 15 g of weight. Fish were divided into four groups of 30 animals, one per tested strain, and were maintained into 50-L seawater tanks at 18°C with aeration and water recirculation. Colonies from a fresh 24-h plate were resuspended in saline solution (0.85% NaCl) to achieve an OD₆₀₀ of 0.5. Fish were intraperitoneally injected with 100 μl of the bacterial suspension (2–3 × 10³ CFU/fish). The number of bacterial cells injected was determined by plating serial dilutions on TSA-1. A control group was injected with saline solution. Mortalities were followed for 12 days after injection and dead events were daily registered. Statistical differences in survival curves were determined using the Kaplan–Meier method with Mantel-Cox log-rank test using SPSS (version 20; IBM SPSS Inc., Chicago, IL). p-values were significant when P was < 0.05. The protocols for animal experimentation follow the current legislation and have been approved by the Bioethics Committee of the University of Santiago de Compostela.

To analyze the diversity of piscibactin gene promoters, we performed BlastN searches in the nucleotide collection (nr/nt) and whole-genome shotgun (wgs) NCBI databases using as a query the nucleotide sequence of V. anguillarum RV22 irp-HPI between 250 bp upstream pbtA and the frpA stop codon (sequence ID. AEZB01000030, from position 30,437 to 35,095). Homologous sequences were clustered in promoter types by similarity; thus, each type included sequences sharing 100% of coverage and≥99.5% of nucleotide identity. Representative sequences of each type were downloaded from NCBI and aligned using MUSCLE (MEGA X suite). Phylogenetic trees were inferred using p-distances (Transitions + Transversions) and Neighbor-Joining method. All ambiguous positions were removed for each sequence pair (pairwise deletion option). There was a total of 3,377 positions in the final dataset. Sequence alignments, nucleotide pairwise p-distances, and evolutionary analyses were conducted in MEGA X (Kumar et al., 2018).

To analyze the role of the two AraC-like regulators encoded by irp-HPI in piscibactin production, in-frame deletion mutants for araC1 and araC2 genes (Figure 1) were constructed in a V. anguillarum RV22ΔvabF background, a strain that produces only piscibactin as siderophore (Balado et al., 2018). Then, the RV22ΔvabF parental strain and its derivative ΔaraC1 and ΔaraC2 mutants were challenged to grow under iron excess (CM9 supplemented with FeCl₃ 10 μM) and under iron-deprivation conditions by adding the iron chelator 2,2′-dipyridyl at 25 μM (weak iron-deprivation) or at 75 μM (strong iron-deprivation).

Under iron excess, or weak iron-deprivation conditions, no significant differences were observed between parental and mutant strains. However, under strong iron-deprivation conditions, the ΔaraC1 defective mutant was impaired in growth (Figure 2). The ΔaraC1 mutant showed a phenotype like that observed for the Δirp1 defective mutant that is unable to produce piscibactin (Balado et al., 2018). Under this condition, the parental strain and the ΔaraC2 defective mutant showed indistinguishable growth ability. Subsequent evaluation of siderophore content in the cell-free culture supernatants showed a strong reduction in the ΔaraC1 defective mutant (Figure 2). When the ΔaraC1 mutant was complemented with a functional version of araC1, the parental phenotype was restored. This result suggests that araC1 could encode a transcriptional regulator essential for siderophore piscibactin synthesis. Consequently, araC1 and araC2 were renamed as piscibactin transcriptional regulator pbtA (piscibactin regulator A) and pbtB, respectively.

The piscibactin TonB-dependent outer membrane transporter frpA gene is located in the piscibactin operon upstream the genes encoding the biosynthetic functions (Balado et al., 2018; Figure 1). To determine whether inactivation of pbtA blocks the piscibactin siderophore system, the presence of FrpA in the V. anguillarum ΔvabF mutant (used as parental strain) and in its derivative pbtA defective mutant was determined by Western blot (Figure 3). While a unique protein band of ca. 70 kDa was detected (which is congruent with the 68-kDa molecular weight of FrpA) in the outer membrane sample (Figure 3A) of the V. anguillarum parental strain (RV22ΔvabF) and in the corresponding Western blot using anti-FrpA (Figure 3B), only a residual amount of FrpA was detected in the sample of the outer membrane proteins and Western blot of V. anguillarum pbtA defective mutant (Figure 3B). These findings greatly suggest that inactivation of the AraC-like transcriptional regulator PbtA disables iron uptake via the siderophore piscibactin.

To evaluate a putative role of the global regulators H-NS and ToxR-S in piscibactin production, single h-ns or toxR-S mutants were constructed. Then, growth ability and siderophore production were assayed in each mutant and compared to the parental strain RV22ΔvabF (Figure 2). While the phenotype of the ΔtoxR-S mutant (RV22 ΔvabFΔtoxR-S) was indistinguishable from that of the parental strain, the Δh-ns mutant showed a slight reduction of growth ability under strong iron-restricted conditions. The diminution of growth observed in the Δh-ns mutant correlates with a decrease in piscibactin production. Finally, when the Δh-ns mutant was complemented with a functional h-ns gene, the parental phenotype was restored. These results suggest that H-NS is required for a maximum piscibactin production. By contrast, ToxR-S did not exhibit a role in siderophore production.

To evaluate the role of PbtA and H-NS in V. anguillarum virulence, groups of 30 sole fingerlings were inoculated with a dose of 2–3 × 10³ CFU per fish of either V. anguillarum parental strain or one of its derivatives ΔpbtA or Δh-ns mutant strains. A control group that was inoculated with saline solution did not show any signs of infection and no mortality was observed. The survival curves of each group of fish are shown in Figure 4. While the fish group challenged with the parental strain (RV22ΔvabF) showed a 60% mortality 7 days after infection, the V. anguillarum pbtA defective strain (ΔpbtA) showed a significant reduction of virulence since it caused approximately 20% mortality. Notably, survival curves of pbtA defective mutant and the ΔvabFΔirp1 double mutant, impaired to produce siderophores (neither piscibactin nor vanchrobactin) (Balado et al., 2018), were statistically indistinguishable. In addition, inactivation of h-ns also resulted in a significant reduction of virulence for fish.

Sequences upstream of pbtA (formerly araC1) and frpA (Figure 1) were previously characterized as the main promoter regions that control the expression of piscibactin synthesis and transport genes (Balado et al., 2018). Thus, to evaluate the role of the putative transcriptional regulator PbtA in the expression of piscibactin genes, the transcriptional activity of pbtA and frpA promoters (PpbtA and PfrpA, respectively) was measured in a pbtA^– background (ΔpbtA mutant). At both temperatures tested, 15°C or 25°C, inactivation of pbtA results in the loss of PfrpA activity (Figure 5). Notably, the expression pattern of PpbtA still follows a temperature-dependent pattern in a pbtA^– background, showing identical activity levels as the parental strain at the same temperatures. On the other hand, pbtB (formerly araC2) deletion did not alter the transcriptional levels of PpbtA or PfrpA. PpbtA and PfrpA promoter activity evaluation in the pbtA^– complemented strain could not be done since the plasmid used to complement (pSEVA651) and the plasmid used to obtain the LacZ fusions (pHRP309) both confer gentamicin resistance. However, as was shown above, complementation of ΔpbtA mutant with the wild-type gene restored a phenotype identical to the parental strain.

Transcriptional activity of PpbtA and PfrpA was also evaluated in the h-ns and toxR-S defective mutants. Unexpectedly, while the activity of the pbtA promoter in the h-ns mutant was indistinguishable from that of the parental strain at both temperatures tested, the activity of frpA promoter (PfrpA) in a Δh-ns background was reduced ca. 60% at 15°C and ca. 20% at 25°C (Figure 4). The evaluation of piscibactin promoters activity in the ΔtoxR-S mutant showed that the expression levels of PpbtA and PfrpA are increased two-fold and 30%, respectively, at 25°C, when compared to the parental strain (Figure 5). No significant changes in expression were observed at 15°C. Complemented strains, either of h-ns or toxR-S, showed expression patterns indistinguishable from those of the parental strain.

Several versions of the irp-HPI genomic island are found within Vibrionaceae showing identical gene structure and an overall nucleotide diversity (p-distance) of 0.3 substitutions per site (Figure 6). Not surprisingly, nucleotide diversity is much higher in intergenic regions than within protein-coding sequences. Alignment of representative irp-HPI genomic islands from different species showed that the sequences immediately upstream of pbtA ATG start codon (PpbtA region) (Supplementary Figure 1) and the pbtB-frpA intergenic region (PfrpA region) (Supplementary Figure 2) showed major differences between Vibrio species and thus different types of piscibactin promoters PpbtA and PfrpA could be defined according to their similarity. Notably, the distribution of each piscibactin promoter type (PpbtA and PfrpA sequences) does not match with irp-HPI phylogenetic lineages (Figure 7). The most variable sequences are found in the pbtB-frpA intergenic region (Supplementary Figures 3,4). Although all pbtB-frpA sequences share a conserved region of ca. 100 bp located immediately upstream of frpA start codon, the region downstream of pbtB stop codon shows higher differences since deletions and/or insertions events would have occurred (Supplementary Figure 2). Thus, the irp-HPI genomic islands of species like V. anguillarum, V. ordalli, and V. qinghaiensis contain a long pbtB-frpA intergenic region with a size of ca. 360 bp, which includes a low complexity sequence between positions 137 and 167 with six repeats of an AAAAT motif (Figure 7 and Supplementary Figure 2). By contrast, P. damselae, V. ostreicida, V. sonorensis, and V. cholerae harbor shorter intergenic sequences with nucleotide lengths between 100 and 140 bp due to the lack of the segment immediately downstream of pbtB stop codon (Figure 7 and Supplementary Figure 1). In addition, there are some intermediate versions such as those found in V. mimicus and V. neptunius (Supplementary Figure 2). The high variability found upstream of pbtA (PpbtA) and in the pbtB-frpA intergenic region (PfrpA) sequences could imply the existence of different expression patterns among those Vibrio spp. harboring the irp-HPI genomic island.

To compare the expression pattern of different versions of the irp-HPI genomic island found in different Vibrionaceae, the expression pattern of the irp-HPI present in Photobacterium damselae subsp. piscicida (irp-HPI_pdp) was compared with the expression pattern of the V. anguillarum island (irp-HPI_ang). Particularly, irp-HPI_pdp harbors the short version of frpA promoter (PfrpA_pdp) described above (Figures 6, 7). LacZ fusions of the sequences immediately upstream of pbtA and frpA from P. damselae subsp. piscicida (denoted as pbtA_pdp and frpA_pdp, respectively), homologous to their counterparts of V. anguillarum (Figures 1, 7), were obtained and the transcriptional activity of each promoter (PpbtA_pdp and PfrpA_pdp) was assayed at 15 and 25°C under low iron availability (Figure 8A). The expression pattern of each piscibactin promoter from P. damselae subsp. piscicida (Figure 8A) showed extensive differences with piscibactin promoters from V. anguillarum (Figure 8C). Thus, PpbtA_pdp promoter is expressed in P. damselae subsp. piscicida although it showed almost the same transcriptional activity at 15°C and at 25°C (Figure 8A), which suggests that piscibactin genes, in contrast with the behavior described in V. anguillarum (Balado et al., 2018), do not show a temperature-dependent expression in P. damselae subsp. piscicida. In addition, PpbtA_pdp showed a threefold lower activity in P. damselae subsp. piscicida (Figure 8A) than its counterpart PpbtA_ang in V. anguillarum (Figure 8C). Unexpectedly, while PfrpA_ang reached ca. 50,000 β-galactosidase units at 15°C, the activity displayed by PfrpA_pdp was almost undetectable (<750 U), suggesting that the short pbtA-frpA intergenic region found in irp-HPI_pdp does not contain a transcriptional promoter.

To ascertain whether the differences observed between the expression patterns of each version of irp-HPI promoters would be due to the differences in the promoter versions or to other factors encoded in the respective recipient genomes, a heterologous evaluation of each LacZ fusion was done (Figures 8B,D). The results showed that both versions of the irp-HPI promoters, either PpbtA_pdp and PfrpA_pdp or PpbtA_ang and PfrpA_ang, have a 2.5-fold less transcriptional activity in P. damselae subsp. piscicida (Figures 8A,B) than PpbtA_ang and PfrpA_ang in V. anguillarum (Figure 8C). The expression level achieved by PpbtA_ang in P. damselae subsp. piscicida was equivalent to that shown by the native evaluation of PpbtA_pdp. Notably, while the transcriptional activity of PpbtA_ang and PfrpA_ang measured within P. damselae subsp. piscicida did not show differences between temperatures (Figure 8B), the PpbtA_pdp promoter showed a three-fold higher transcriptional activity at 15°C than at 25°C when its activities were measured within V. anguillarum (Figure 8D). By contrast, PfrpA_pdp did not show any activity within V. anguillarum, which greatly suggests that the short version of pbtB-frpA intergenic sequence present in irp-HPI_pdp does not contain a transcriptional promoter (Figure 8A).