All the strains used in this study are listed in Table 1. Their sequences are available in the GenBank data base. Genome comparisons to determine protein homology, common genes, and genomic organization were performed using the RAST software (Aziz et al., 2008) and BLASTp.

Phylogenetic studies of individual proteins were performed using the platform Phylogeny.fr (http://www.phylogeny.fr/) (Dereeper et al., 2008). Phylogenetic studies of complete genomes were performed using Composition Vector Tree (http://tlife.fudan.edu.cn/cvtree/). These phylogenetic studies were carried out as recommended by the user manual (http://tlife.fudan.edu.cn/cvtree/help/index.html) with a K-peptide length of 6, as recommended by the authors for prokaryote organisms (Xu and Hao, 2009).

The pyoverdine peptide chain was predicted using the non-ribosomal peptide synthase substrate predictor NRPSsp (http://nrpssp.com/execute.php), NRPSpredictor2 (http://nrps.informatik.uni-tuebingen.de), and PKS/NRPS programs (http://nrps.igs.umaryland.edu/) for pyoverdine synthetases.

TonB-dependent transporters were identified by selecting TonB annotated proteins from the genomes of the microorganisms used in this study using the RAST Annotation Server (http://rast.nmpdr.org/rast.cgi) and the set of sequences deposited in Genbank (https://www.ncbi.nlm.nih.gov/genbank/). We selected TonB-transporter proteins bearing the characteristic domains of siderophore transporters by application of InterPro (http://www.ebi.ac.uk/interpro/) and Blastp web sites that provide functional analysis. Exclusive TonB-dependent transporters were those presenting < 80% identity in the amino-acid sequence with others present in P. putida strains.

Cells were pre-grown in M9 minimal media (Abril et al., 1989) with glucose (5 mM) as the sole carbon source. The overnight cultures were washed three times and re-suspended to reach OD_660nm of 0.1 in M9-glucose. Salt conditions were adjusted to 0.05 or 0.9% w/v of NaCl in the low and high osmolarity media respectively. When indicated the iron chelator EDDHA was added to reach a final concentration of 20 μM. The visual and spectrophotochemical detection of pyoverdine production was performed as indicated by Bultreys et al. (2001). The production of pyoverdine-type siderophores was visualized as maximum absorption at 365, 375, and 410 nm.

Supernantants of cultures grown with 5 g/L of casamino acids (CAA) supplemented with MgSO₄ 7H₂0 (0.25 g/L) and K₂HPO₄ (1.18 g/L) were analyzed by IEF according to the method described by Koedam et al. (1994). Pyoverdine was separated in polyacrylamide (5%) gels (125 × 65 × 0.4 mm) containing 2% ampholines (Byolyte 3/10 from Bio-Rad) to develop a pH gradient from 3.5 to 9.3 during electrophoresis in model 111 Mini IEF Cell. Preparation of the gels and electrophoresis conditions were those recommended by the manufacturer. Samples of 1 μl of the 20-fold-concentrated (through lyophilisation) CAA-culture supernatants were loaded. Samples were run at 4°C for 1.5 h at constant power (12W) with voltage from 200 V at the beginning to 1000 V at the end of the electrophoresis. Immediately after the run, the fluorescent bands of pyoverdines were visualized under UV light at 365 nm.

Siderophore-mediated iron uptake was performed as described by Fuchs et al. (2001) and adapted by Hoegy and Schalk (2014). Briefly, labeled ferrisiderophores were prepared by mixing a 4μl⁵⁵FeCl₃ 250 μM (Amersham) with 4 μl of a 1 mM pyoverdine solution obtained from the cultures of the different strains. After 30 min of incubation at room temperature, this mixture was diluted with Tris-HCl pH 8 (50 mM) to obtain a 10 μM siderophore-⁵⁵Fe complex.

Pseudomonas strains to be used as siderophore uptaking cells were grown in CAA medium for 24 h at 30°C. Then, bacterial cells were washed twice in CAA medium to remove the native siderophore and were re-suspended in Tris-HCl pH 8 (50 mM) to reach an optical density of 1 at 600 nm.

Five microlitres of siderophore-⁵⁵Fe complex were mixed with 500 μl of the cell suspension and incubated for 30 min at 30°C. Then, 100 μl of the labeled bacterial cells were withdrawn and rapidly filtered through 0.45 μm membranes (Whatmann GFB filters). The cells remaining on the filters were thoroughly washed twice with Tris-HCl pH 8 (50 mM) and the radioactivity was measured to determine the amount of labeled iron incorporated during the incubation time Gamma emission was counted using a PerkinElmer gamma counter. Iron uptake of heterologous pyoverdine was expressed in percentage terms compared with the homologous pyoverdine uptake.

Statistics were evaluated with the Statgraphics Plus v.5.1. software (Statistical Graphics Corp., Herndon, VA, United States). Data are means of three independent determinations and were subjected to one- or two-way analysis of variance (ANOVA) followed by Tukey posthoc test (p ≤ 0.01) when a significant difference was detected.

The genes which are typically involved in pyoverdine synthesis, transport, and perception are conserved in P. putida strains (Figure 1). All the genomes analyzed contain the genes that encode PvdH, PvdP, PvdM, PvdN, PvdO that are involved in the pyoverdine chromophore maturation and PvdD, PvdJ, PvdI which are involved in the synthesis of the peptidic part of the pyoverdine (the non-ribosomal peptide synthetase module). They also contained pvdE, a gene that encodes a protein for the export of the pyoverdine precursor peptide to the periplasm, and pvdR, pvdT, and opmQ and fvpA. PvdRT and OpmQ mediate the transport of the newly synthesized pyoverdine across the outer membrane and the recycling of pyoverdines. The FvpA receptor is a TonB-transporter family protein involved in pyoverdine-Fe sensing and internalization (Imperi et al., 2009; Cornelis, 2010; Schalk and Guillon, 2013; Gasser et al., 2015; Chen et al., 2016). fpvI that encodes for a RNA polymerase ECF-type sigma factor is also present in all the strains at the same locus. Although all the genes are present in the analyzed genomes, their organization varies depending on the P. putida strain (Figure 1).

P. putida strains can be grouped into 5 different clades on the basis of the percentage of identity shared by the proteins encoded in the pyoverdine locus (each clade was defined when the overall amino-acid identity was higher than 80%; Tables 1, 2, Figure 1). Strains in Clade I can be classified into two sub-clades, each one sharing an identity higher than 95%. (Tables 1, 2). Subclade A includes the clinical isolates P. putida HB13667, HB3267 and the waste-water strain P. putida JLR11, (Tables 1, 2). Subclade B included the waste-water strains P. taiwanensis SJ9 and P. monteilii SB3101; the rhizosphere strain S16 and the soil and naphthalene degrader strain KG-4. The aquatic strain GB-1 and the clinical strain NBRC 14164T form a less homogeneous sub-clade; the amino-acid identity of the pyoverdine locus among the strains of this subclade did not reach 90% (Table 2). In all the strains forming Clade I, the pyoverdine genes are located in two loci in different locations in the chromosome (Table 1, Figure 1).

Clade II can also be divided into two subclades (Tables 1, 2). The first one includes the clinical strains P. putida H8234 and JCM 18798; P. putida SJ3, a waste-water and caprolactam degrader strain, with the soil isolate and naphthalene degrader P. putida OUS82 (Tables 1, 2). Overall amino-acid identity amongst the pyoverdine locus of the four strains within the subclade is higher than 95%. The second subclade includes the clinical strain P. putida HB4184 and the rhizobacterium B001 (Table 2). In all the strains forming Clade II, the pyoverdine genes are located in a single locus in the chromosome (Table 1, Figure 1).

Clade III and IV are very homogeneous (Tables 1, 2). Clade III comprises strains that showed more than 98% identity in the pyoverdine-related proteins. As in Clade II, the pyoverdine genes are located in a single locus in all the strains forming Clade III, Table 1, Figure 1). Clade IV, includes strains with an overall identity >96%, and comprises only water isolates, such as the toluene degrading strains P. putida F1 and DOT-T1E, the naphthalene degrader ND6, the triclosan degrader TR01, and the polyhydroxyalkanoate synthesizer LS46 (Tables 1, 2). In all these strains the pyoverdine genes are located in two different loci (Table 1, Figure 1), with the exception of P. putida DOT-T1E that presents three loci associated with pyoverdine synthesis (Table 1).

The endophitic strain W619, with strain SQ1 isolated from lake sediment and the Antarctic isolate ATH-43 are found in Clade V (Tables 1, 2). In Clade V, the pyoverdine genes are located in a single locus in the chromosome (Table 1, Figure 1).

Two bacteria, traditionally considered close relatives of the P. putida group (Molina et al., 2016), were included in this study; the entomopathogen P. entomophila L48 (Vodovar et al., 2006), and the fish pathogen P. plecoglossicida DSM 15088 (Nishimori et al., 2000). None of them presented an identity higher than 68% at this locus when compared to the rest of the P. putida strains (Table 2), indicating the different evolution of these strains in terms of siderophore production and perception.

Bioinformatic prediction of the composition of the peptide sequences of different pyoverdines from several fluorescent Pseudomonas was performed by Ye et al. (2013). This study supports our clade classification. Furthermore, we have predicted the pyoverdine peptide sequences of the strains not previously included in Ye et al. (2013) and our results show that all the strains included in a clade have the same peptide sequence (Table 1). The consensus sequence for the pyoverdine peptide obtained for Clade II strains (Table 1) is Asp-5hOrn-Lys-Thr/Gly- a peptide that was not previously predicted by Ye et al. (2013) and that is clearly different from the sequences of other clades.

To experimentally confirm the in silico results, we carried out IEF siderotyping assays on representative strains of each of the clades defined in this work (except strains from Clade V that were not available in our lab). We found that strains belonging to one clade had similar pyoverdine production profiles and that the profile of strains within a clade was different from that of strains belonging to other clades (Figure 2).

Our bioinformatic and experimental results suggest that the average identity of genes forming part of the pyoverdine locus, together with bioinformatics using the predictor NRPSsp, can easily discriminate the type of pyoverdine produced by the different strains within the species P. putida, instead of more laborious techniques such as siderotyping.

In silico analysis of pyoverdine loci grouped the strains of P. putida HB13667 and HB3267 in Clade I with 98% sequence identity; both had been isolated from human blood. Although P. putida NBRC14164T was also grouped in Clade I, it belonged to a different subclade sharing a lower amino-acid identity (84%) with HB13667 and HB3267. P. putida H8234, H4184, and JCM18798 (with an overall identity of 86%) grouped into Clade II and were clearly separated from the other three clinical strains identity below 65% (Table 2). This suggests that siderophore-related proteins of the clinical isolates have at least two different phylogenic origins.

Interestingly, the pyoverdine-related proteins of isolate HB4184, isolated from human sputum, showed an identity close to 99.0% with P. putida B001 (Table 2), a rhizobacterium that was isolated from shore sand, oligotrophic and high salinity environment (Park et al., 2011; Table 2). To investigate if pyoverdine production in P. putida HB4184 was induced under high salt concentrations, we analyzed their production in a medium with the standard salt concentration used in laboratory media (0.05% w/v NaCl) and in the same medium but with 0.9% w/v NaCl content, in the presence or absence of the iron chelator EDDHA. Pyoverdine is secreted to the media in poor iron conditions; consequently, production of siderophores under iron rich conditions was barely perceptible in all the strains tested regardless of the salinity of the media (Figure 3). In the absence of free iron ions (in the presence of the iron chelator EDDHA) the production of a pyoverdine-type siderophore was increased in both media. This increase was especially significant in the environmental strains P. putida KT2440 and BIRD1 under standard laboratory medium. Production of pyoverdine was induced in the presence of high salt concentrations in all the clinical strains tested, but not in the environmental strains. This increase was significantly higher in strain P. putida HB4184 (Figure 3). This experiment suggests that clinical strains are better adapted to uptake iron under high salinity conditions than the environmental strains, with P. putida HB4184 being especially adapted to this condition.

Analysis of the homology amongst P. putida proteins involved in the synthesis and maturation of the chromophore part, in the transport of the peptidic part of pyoverdine or the complete siderophore, which are encoded by some genes of the pyoverdine locus, showed that they are highly conserved in all the P. putida studied strains (with an identity in the range between 75 and 90%). However, the identity percentage diminished when the pyoverdine receptor protein (FpvA) or the non-ribosomal peptide synthesis module PvdD, PvdJ, PvdI were analyzed (identity below 65% in most cases; data not shown). A similar observation was reported in P. aeruginosa strains by Smith et al. (2005). These authors proposed that the FpvA protein may be driving diversity at the pyoverdine locus. The outer membrane receptor FpvA, involved in the internalization of the pyoverdine-Fe complex following extracellular iron chelation, is specific for one pyoverdine type (Meyer et al., 2002; Smith et al., 2005; Nader et al., 2011). In P. aeruginosa three structurally different types of pyoverdine (types I, II, III) are produced, and each one is recognized at the outer membrane by the specific receptors, FpvAI, FpvAII, and FpvAIII, respectively (Cornelis et al., 1989).

Classification of P. putida strains on the basis of the homology only amongst FpvA proteins gave a similar grouping of strains than when we used the complete pyoverdine locus (Table 1, Figure 4). Strains included in Clade I and II appear to belong to the classical FpvA receptor group (FpvA-I) where the FpvA type protein is similar to the FpvA of P. aeruginosa that recognizes the pyoverdine type I (Nader et al., 2011; Table 1, Figure 4). However, FpvA sequences of strains previously classified as Clade II are phylogenetically more distant to the sequences of strains belonging to Clade I; therefore we suggest that they may constitute different subgroups (Table 1, Figure 4).

Strains belonging to Clade III presented homology with an alternative receptor of type I pyoverdine named FpvB (Ghysels et al., 2004; Table 1, Figure 4). P. putida FpvA proteins belonging to Clade III are phylogenetically separated from other members of the group that are not P. putida species, such as Pseudomonas protegens Pf-5 (Hassan et al., 2010), Pseudomonas chlororaphis subsp. aureofaciens 30–84 (Loper et al., 2012), and P. protegens CHA0 (Ramette et al., 2011; Table 1, Figure 4). In the P. putida strains used in this study, the fpvA gene codifying the FpvB receptor is part of the pyoverdine locus (Figure 1) whilst in P. aeruginosa this gene is located elsewhere in the chromosome (Ghysels et al., 2004).

FpvA proteins of strains in Clade IV form a highly homogeneous group, which is related to the FpvA-III receptor from the opportunistic human pathogen P. aeruginosa PAO1 (Szmolka et al., 2012; Figure 4).

The FpvA protein of the endophitic strain W619 and strains SQ1 and ATH-43 (Clade V) form part of a different group of receptors, closer to that of Pseudomonas sp. GM84, a strain isolated from rhizosphere/endosphere of Populus deltoides (Brown et al., 2012).

Interestingly, P. putida JLR11, HB13667, HB3267 FpvA proteins were at least 98% identical, suggesting a common origin of this transporter in these three strains (Table 1, Figure 4). FpvA from HB4184 shows high homology (98%) with the rhizobacterium P. putida B001, isolated from a sandy shore soil from the Yellow Sea.

To sum up, the results described above indicate that phylogenetic studies based on the FpvA protein sequence of different P. putida strains yielded the same five clades that have been defined on the basis of overall protein homology with the pyoverdine-related genes (Table 1, Figure 4).

Pseudomonas strains are also able to obtain iron through the acquisition of siderophores produced by other microorganisms (xenosiderophores) by TonB-dependent transporters located in their outer membrane (Schalk and Guillon, 2013). The function and number of TonB-dependent transporters in a particular strain have been related to their ability to adapt to changing environments and to colonize new niches (Cornelis, 2010; Llamas et al., 2014). TonB-dependent transporters are made up of three domains: the first one, a β-barrel with 11 large extracellular loops and periplasmic turns. The lumen of this barrel is occluded by a globular domain, which is the second domain, called the “plug” and the third domain consists of a long periplasmic N-terminal extension that is not involved in the transport of iron siderophores but participates in the regulation of transcription (Noinaj et al., 2010).

Analysis of the genome sequence of the P. putida strains revealed the presence of 7–14, depending on the strain, TonB-dependent transporters with these three domains, (Table 1, Supplementary Table 1). Remarkably, strains belonging to subclade A of Clade II and those belonging to subclade C or Clade I possess the highest number of these transporters (from 19 to 25) (Table 1, Supplementary Table 2). The presence of a higher number of TonB-dependent transporters than in the other P. putida strains seems to indicate that strains belonging to these subclades may have more versatility to acquire iron sources.

Most of the TonB-dependent transporters of the clinical strains P. putida HB13667, HB3267, and HB4184 are similar to those present in other P. putida strains (Supplementary Table 2), indicating than they are able to recognize the same type of xenosiderophores; this kind of transporter is considered to be non-exclusive. However, P. putida H8234 possesses two exclusive TonB-dependent transporters that have no homologs in any other P. putida strains we have studied. These transporters, namely L483_22175 and L483_11595, have their closest homologs in transporters identified in P. aeruginosa strains. L483-22175 presents 83% identity with P. aeruginosa isolates identified in cystic fibrosis patients (De Vos et al., 2001) and belongs to the FpvAII group (Figure 4) whilst L483_11595 belongs to the FpvB group (Figure 4) and its closest homolog (77% identity) was found in opportunistic human pathogens, i.e., P. aeruginosa strains PAO and LESB58. These results indicate that the P. putida isolate H8234 is able to recognize different siderophores from those recognized by other P. putida strains.

To investigate the incorporation of different pyoverdines produced by several Pseudomonas strains, we measured the incorporation of P. putida siderophores bound with ⁵⁵Fe into several P. putida strains. Siderophores produced by P. putida strains belonging to the same clade incorporated 100% of the complex siderophore-⁵⁵Fe, with the only exception of P. putida HB4184 that was only able to uptake the 48.2% of siderophores produced by P. putida H8234. These two strains belong to different subclades, and differences at the level of pyoverdine-related proteins may account for this differential uptake. Cross-incorporation of siderophores between strains belonging to different clades was inferior to 4%.

As previously mentioned, the outer membrane TonB-dependent transporters are involved in the internalization of the xenosiderophore-Fe complexes and therefore we tested if siderophores produced by P. aeruginosa strains can be taken up by P. putida strains. In these assays we used the strains P. aeruginosa PAO1 that produces pyoverdine type I, P. aeruginosa 27,853 that produces pyoverdine type II, and P. aeruginosa Pa6 that produces pyoverdine type III. The level of siderophore (probably pyoverdine) cross-incorporation was below 10% in all cases, with the exception of P. putida DOT-T1E that was able to incorporate 18.8% of the siderophore-Fe complexes of P. aeruginosa strain PAO1, 21.7% of strain 27,853 and 73.8% of strain Pa6 (Table 3).

In agreement with our phylogenetic studies of the FpvA protein, strains with FpvA proteins included in the FpvA-I and FpvB type of receptors (Figure 4), known to recognize pyoverdine type I (Cornelis et al., 1989; Ghysels et al., 2004) and belonging to Clade I, II, and III recognized the siderophores produced by the P. aeruginosa PAO1 (which mainly produces pyoverdine type I) better than siderophores produced by the other two P. aeruginosa strains (Table 3). Also in agreement with our classification, P. putida DOT-T1E (belonging to Clade IV and with FpvA protein included in the FpvA- III group; Figure 4), uptook siderophores more specifically produced by P. aeruginosa strain Pa6 (which mainly produces pyoverdine type III).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.