The evolution of three siderophore biosynthetic clusters in environmental and host-associating strains of Pantoea

Craig D. Soutar1 · John Stavrinides1

Abstract

For many pathogenic members of the Enterobacterales, siderophores play an important role in virulence, yet the siderophores of the host-associating members of the genus Pantoea remain unexplored. We conducted a genome-wide survey of environmental and host-associating strains of Pantoea to identify known and candidate siderophore biosynthetic clusters. Our analysis identified three clusters homologous to those of enterobactin, desferrioxamine, and aerobactin that were prevalent among Pantoea species. Using both phylogenetic and comparative genomic approaches, we demonstrate that the enterobactin-like cluster was present in the common ancestor of all Pantoea, with evidence for three independent losses of the cluster in P. eucalypti, P. eucrina, and the P. ananatis—P. stewartii lineage. The desferrioxamine biosynthetic cluster, previously described and characterized in Pantoea, was horizontally acquired from its close relative Erwinia, with phylogenetic evidence that these transfer events were ancient and occurred between ancestral lineages. The aerobactin cluster was identified in three host-associating species groups, P. septica, P. ananatis, and P. stewartii, with strong evidence for horizontal acquisition from human-pathogenic members of the Enterobacterales. Our work identifies and describes the key siderophore clusters in Pantoea, shows three distinct evolutionary processes driving their diversification, and provides a foundation for exploring the roles that these siderophores may play in human opportunistic infections.

Keywords Siderophores · Pantoea · Desferrioxamine · Aerobactin · Enterobactin · Horizontal transfer

Introduction

Both free-living and host-associating microorganisms produce and secrete low molecular weight molecules known as siderophores, which chelate ferric iron (Neilands 1995). Siderophore production is directed by a cluster of genes that code for enzymes involved in synthesis, regulators, a transporter used for siderophore export, and a highly selective membrane-associated ATP-dependent receptor that is responsible for the uptake of the siderophore–iron complexes that form in the environment (Köster 2001; Crosa and Walsh 2002; Challis 2005). The enzymes encoded by the gene cluster may constitute a multi-modular enzymatic complex referred to as a non-ribosomal peptide synthetase (NRPS), which systematically assembles the natural product (Frueh et al. 2008; Hider and Kong 2010). Siderophores formed by NRPSs include some catecholates (e.g., enterobactin), hydroxamates (e.g., coelichelin), and nitrogen heterocycles (e.g., yersiniabactin) (Kadi and Challis 2009). However, some siderophores, such as aerobactin and desferrioxamines B, G1, and E are synthesized via NRPS-independent pathways (Miethke and Marahiel 2007; Kadi and Challis 2009). These diverse siderophores enable microbes to exploit a variety of nutrient-restricted or nutrient-limited niches, including host organisms.
Animal pathogenic members of the Enterobacterales produce a diversity of siderophores that are deployed during pathogenesis to compensate for the defense-related iron sequestration mechanisms of their hosts, thereby allowing them to initiate infection (Weinberg 1993; Ratledge and Dover 2000). Yersinia enterocolitica requires the siderophore yersiniabactin for full virulence (Heesemann 1987), while avian pathogenic Escherichia coli strain, E058, requires the hydroxamate siderophore, aerobactin, for virulence (Ling et al. 2013). The human pathogen, Salmonella enterica serotype Typhimurium, has been found to produce a derivative of the siderophore enterobactin called salmochelin, which is resistant to uptake by the siderophore-scavenging host defense protein, lipocalin-2 (Raffatellu et al. 2009). Salmonella mutants deficient in either biosynthesis or export of salmochelin had reduced virulence in mice (Crouch et al. 2008). Likewise, gut-associated E. coli strains were more likely to produce multiple siderophores compared to plantassociated and environmental isolates (Searle et al. 2015), reinforcing their central role in host colonization and pathogenesis. Siderophores, however, also play an essential role during plant pathogenesis. The enterobacterial plant pathogen, Erwinia amylovora, the causative agent of fire blight of apple and pear, is dependent on the hydroxamate siderophore desferrioxamine E, which not only improves iron acquisition, but also provides the bacteria with protection from host-induced oxidative stress incurred during pathogenesis (Dellagi et al. 1998). Another plant pathogen, Dickeya dadantii (formerly Erwinia chrysanthemi) produces chrysobactin and achromobactin, both of which are necessary for systemic dissemination in plant hosts (Dellagi et al. 2009). Siderophores are therefore important factors that facilitate the association of bacteria with their animal and plant hosts.
Members of the enterobacterial genus Pantoea live freely in the environment, but also exploit a wide range of hosts including plants and humans (Nadarasah and Stavrinides 2014; Walterson and Stavrinides 2015). Despite the important role of siderophores in host-specific virulence for the Enterobacterales, little is known about the type and distribution of siderophore biosynthetic clusters across Pantoea. For plant pathogenic strains of Pantoea, siderophores such as aerobactin and desferrioxamine E are essential determinants facilitating host association (Berner et al. 1988; Burbank et al. 2015). Aerobactin was found to be necessary for full plant virulence in Pantoea stewartii subsp. stewartii DC283, as mutants with disrupted aerobactin biosynthetic or receptor genes exhibited reduced motility on both the plant surface and within the plant, resulting in impaired biofilm formation and host colonization (Burbank et al. 2015). Siderophore production by Pantoea eucalypti M91 enhanced iron acquisition and growth of Lotus japonicus plants under iron-limited conditions (Campestre et al. 2016), and siderophore-producing endophytic Pantoea ananatis strains appear to be highly competitive, preferentially colonizing various tissues of host rice plants including roots, leaves, and grains (Loaces et al. 2011). In contrast to these plant strains, there is currently no information on the siderophores used by Pantoea species that associate with animal hosts, including opportunistic P. ananatis, Pantoea agglomerans, and Pantoea septica, all of which are recurrently isolated from patients in the nosocomial environment (Brady et al. 2010; De Maayer et al. 2012; Sengupta et al. 2016). To address this knowledge gap, we surveyed siderophore biosynthetic gene clusters in Pantoea and explored their distribution and evolution. Our analysis identified at least three different siderophore biosynthetic gene clusters, which exhibit disparate evolutionary histories that include vertical inheritance, loss, as well as horizontal exchanges with human-pathogenic species.

Materials and methods

Databases and data collection

Siderophore biosynthetic gene cluster prediction was carried out on a small subset of complete and draft genomes (Walterson and Stavrinides 2015) using standalone antiSMASH with the “–clusterblast –subclusterblast –knownclusterblast –smcogs –inclusive” switches (Weber et al. 2015). This preliminary dataset included representatives of P. agglomerans, P. ananatis, Pantoea brenneri, Pantoea calida, Pantoea dispersa, Pantoea eucalypti, P. septica, P. stewartii, and Pantoea vagans, along with reference outgroup strains of Cronobacter, Dickeya, Erwinia, Klebsiella, and Pectobacterium (Supplementary Table S1). The output of this analysis was used to identify predicted biosynthetic clusters that were shared among groups of Pantoea strains. Cluster boundaries were identified by comparative analysis with BLAST and MIBiG (Medema et al. 2015), referencing known siderophore clusters.
The distribution of desferrioxamine, aerobactin, and enterobactin-like clusters was assessed against a more comprehensive dataset composed of 10 species of Erwinia, 12 species of Pantoea, along with representative Citrobacter, Enterobacter, Klebsiella, Rahnella, Salmonella, Serratia, Shigella, and Yersinia strains (Table 1) using an inhouse Perl-based pipeline that annotated genomes using GenemarkS (Besemer 2001), and identified and extracted homologs of the query using standalone BLAST and a BLAST parser script. Whole gene clusters and individual genes were compared by BLAST using a minimum 60% nucleotide sequence identity and 60% query coverage, with a minimum E-value of 2e-97. Housekeeping genes used for multi-locus sequence analysis (MLSA), cpn60, fusA, leuS, infB, recA, rpoB, and rplB, along with dfoS and foxR homologs were extracted from all genomes using the inhouse Perl-based pipeline. Gene clusters and MLSA loci have been deposited in Genbank under accession numbers MH015021–MH015174.

Alignments and phylogenetic trees

Alignments for all trees were generated using Clustal Omega version 1.2.1 using default parameters (Sievers et al. 2011). Neighbor-joining and maximum likelihood trees were constructed in either MEGA version 6.06 or 7.0.26 (Tamura et al. 2013; Kumar et al. 2016) with 1000 bootstrap replicates. Modeltest, as implemented in MEGA, was used to identify the best substitution model (as indicated for each tree). Alignments for the enterobactin cluster consisted of concatenated EntA, EntB, EntE, EntC, FepB, FepD, FepG, FepC, EntF and Fes proteins, the aerobactin alignment consisted of concatenated IucA, IucB, IucC, and IucD, and the desferrioxamine tree consisted of DfoJ, DfoA, and DfoC. Individual phylogenies were constructed for the desferrioxamine-associated DfoS and FoxR proteins as indicated, and multi-locus species trees were generated using concatenated cpn60, fusA, leuS, infB, recA, rpoB, and rplB genes. FASTA-formatted sequences (aligned and non-aligned) for all phylogenetic trees are available as supporting data.

Results

Three siderophore clusters are prevalent across Pantoea strains

To identify siderophore biosynthetic gene clusters, Pantoea and other representative enterobacterial genomes were analyzed using the antiSMASH pipeline (Weber et al. 2015). Clusters corresponding to enterobactin, aerobactin, desferrioxamine, turnerbactin, yersiniabactin, amychelin, griseobactin, and colibactin were identified, with matches having 2–95% overlap in gene composition to database reference clusters, as reported by the antiSMASH pipeline (Supplementary Table S2). Additional strain genomes were selected for further comparative analysis (Table 1) based on initial antiSMASH findings. We focused on enterobactin, aerobactin, and desferrioxamine, which we noted as being the more prevalent across the Pantoea genomes (Table 2). Predicted clusters were analyzed and compared further with BLAST, revealing that they were highly conserved across Pantoea strains at the nucleotide level. Enterobactin clusters shared at least 79% nucleotide identity, the aerobactin clusters at least 75% identity, and the desferrioxamine clusters at least 77% identity. An evaluation of the distribution of these three predicted clusters across 40 strains representing 12 Pantoea species groups revealed that both enterobactin and desferrioxamine had a considerably broader distribution than aerobactin. No siderophore cluster was present across all Pantoea species surveyed, although several species carried multiple clusters. All P. agglomerans, P. anthophila, and P. vagans strains analyzed possess both desferrioxamine and enterobactin clusters, P. ananatis and P. stewartii carry both desferrioxamine and aerobactin clusters, and P. septica has both enterobactin and aerobactin clusters (Table 2). The ten species of the closely related Erwinia that were analyzed contained the desferrioxamine biosynthetic gene cluster, but lacked the enterobactin cluster, and only Erwinia iniecta possessed the aerobactin biosynthetic cluster. More divergent enterics, including representative Shigella, Salmonella, Serratia, and Enterobacter species, carried both the enterobactin and aerobactin clusters, but lacked the desferrioxamine cluster (Table 2). Representative Klebsiella species carry the enterobactin cluster, while only one of the two species contained the aerobactin cluster. All the Yersinia species analyzed possessed the desferrioxamine cluster, although two strains also carried the aerobactin cluster, and the Citrobacter and Rahnella species analyzed also carried a single siderophore cluster (Table 2). Given the distribution of these three clusters across the Enterobacterales, we analyzed their evolution further to attempt to identify how these siderophore clusters may have originated in Pantoea.

The enterobactin gene cluster was present in the common ancestor of all Pantoea strains

The Pantoea enterobactin cluster was initially identified as turnerbactin by antiSMASH (Crosa and Walsh 2002; Han et al. 2013), but subsequent comparative analyses suggested that the cluster was compositionally more consistent with that of the enterobactin cluster of Salmonella enterica serovar Typhimurium (Crouch et al. 2008) and the serratiochelin cluster of Serratia plymuthica V4 (Seyedsayamdost et al. 2012). The Pantoea cluster and the serratiochelin cluster both lacked the ferric enterobactin transport-coding gene, fepE, and the biosynthetic gene, entD, found in the enterobactin cluster (Crouch et al. 2008) (Fig. 1a, Fig S1).
However, both the enterobactin and the Pantoea clusters did not have the export/uptake-associated, schP, or the predicted biosynthetic gene, schG found in serratiochelin (Seyedsayamdost et al. 2012) (Supplementary Fig. S1; Supplementary Table S3). We refer to the Pantoea cluster hereonin as “enterobactin-like”.
The enterobactin-like cluster of Pantoea, which shares 12 of the 14 genes described for the Salmonella enterobactin cluster, was found in all Pantoea genomes except P. ananatis, P. stewartii, P. eucalypti, and P. eucrina. The cluster from P. anthophila 11-2 differed from the other Pantoea clusters by one gene, carrying 11 of the 14 genes described for Salmonella enterobactin. Relative to the Salmonella cluster previously reported by Crouch et al. (Crouch et al. 2008), most Pantoea clusters have two additional genes that encode a predicted hot dog superfamily protein (thioesterases and dehydratases) and a predicted MbtH-like family protein, named for their similarity to the MbtH protein from the mycobactin biosynthesis gene cluster of Mycobacterium tuberculosis (Quadri et al. 1998) (Supplementary Table S3). The clusters in all strains of P. dispersa, P. rwandensis, and P. rodasii do not contain the gene encoding the predicted MbtH-like family protein. In addition, the enterobactin-like clusters across Pantoea are in the same genomic context, as indicated by homologous flanking genes (Fig. 1a). In P. ananatis, P. stewartii, P. eucalypti, and P. eucrina, the same flanking genes are conserved and present, but the cluster itself is absent (Fig. 1b). A gene genealogy of concatenated enterobactin genes (using only representative strains from each species) is largely congruent to the MLSA species tree, with only the position of P. septica being discordant (Fig. 1b). P. septica forms a sister group to all other Pantoea species groups in the MLSA tree, but is a sister taxon to the P. agglomerans, P. vagans, P. anthophila, and P. brenneri branch in the cluster tree, both with strong support (Fig. 1b). Enterobactin clusters were also identified in Klebsiella, Enterobacter, Shigella, Salmonella, Citrobacter, Serratia, and the outgroup Pseudomonas (Table 2). The positions of these taxa in the phylogenetic analysis were consistent between the two trees.

Aerobactin was horizontally acquired independently at least twice

The aerobactin biosynthetic gene cluster, composed of the four biosynthetic genes, iucABCD, and the predicted receptor gene, iutA, has been previously reported in P. stewartii (Burbank et al. 2015) (Fig. 2a). The aerobactin cluster was identified in 3 of the 12 Pantoea species surveyed, namely all strains of P. stewartii, P. ananatis, and P. septica, as well as in other enterobacteria including Serratia, Enterobacter, Salmonella, Shigella, and only one of the ten Erwinia species surveyed, Erwinia iniecta (Table 2). A more divergent aerobactin cluster is present in Grimontia hollisae (formerly Vibrio hollisae) (Thompson et al. 2003), which provided an appropriate outgroup for our analyses (Fig. 2b). The genetic composition and organization of all the aerobactin clusters were largely consistent across taxa, although that of G. hollisae contained a predicted ferric iron reductase, fhuF, between iucD and iutA (Suzuki et al. 2006) that is not present in any of the enterobacterial clusters. In addition, the clusters we identified in Pantoea are flanked upstream by a predicted major facilitator superfamily (MFS) transporter (Fig. 2a), which is absent in clusters of E. iniecta, both Yersinia species, and G. hollisae. The gene genealogy generated using the four concatenated aerobactin biosynthetic genes using representative strains was incongruent to the MLSA species tree (Fig. 2b). Notably, the clusters from the Pantoea species did not group together, with P. septica grouping with Enterobacter cloacae, while P. ananatis and P. stewartii grouped with Salmonella and Shigella, which is suggestive of at least two horizontal transfer events. Horizontal exchanges are further supported by the position of E. iniecta, which rather than forming a sister group to the Pantoea lineages formed a monophyletic group with Serratia and Yersinia species with strong support (Fig. 2b).

Desferrioxamine was transferred from ancestral Erwinia to ancestral Pantoea

The desferrioxamine biosynthetic gene cluster in Pantoea and Erwinia has been described previously (Smits and Duffy 2011). The cluster is composed of the three predicted biosynthetic genes dfoJAC, the predicted MFS transporter, dfoS, and the predicted siderophore receptor, foxR (Smits and Duffy 2011), which is consistent in biosynthetic gene content with the prototypical cluster identified in Streptomyces coelicolor A3(2) (Bentley et al. 2002) (Fig. 3). The desferrioxamine biosynthetic cluster was identified in 7 of the 12 Pantoea species surveyed in this study, with three different cluster configurations observed (Table 2) (Fig. 3). The first configuration, observed in P. ananatis and P. stewartii, includes dfoJACS and a divergently transcribed foxR gene. The second configuration, seen in P. eucrina, includes dfoJACS with foxR being in the same orientation as the rest of the cluster. The third configuration is found in P. agglomerans, P. vagans, P. eucalypti, and P. anthophila, in which there is no foxR homolog adjacent to dfoJACS. Within Erwinia, three different configurations of the desferrioxamine cluster were identified (Fig. 3). The first configuration, exemplified by the clusters of E. amylovora, E. pyrifoliae, E. tasmaniensis, and E. piriflorinigrans carries the dfoJAC genes with foxR in reverse orientation, but lacking the predicted MFS transporter, dfoS (Fig. 3). The second configuration, found in E. mallotivora, E. billingiae, E. typographi, E. toletana, and E. iniecta, has dfoJACS, but with foxR being species tree elsewhere in the genome. This latter configuration is similar to that of P. agglomerans/ vagans/ eucalypti/ anthophila as well as in Rahnella aquatilis. The third configuration, found in E. oleae, has dfoJACS and foxR all in the same orientation and is similar to the P. eucrina cluster.
We carried out a phylogenetic analysis of the DfoJAC proteins to validate the evolutionary history of the desferrioxamine biosynthetic gene cluster previously reported (Smits and Duffy 2011). A comparison of our rooted DfoJAC tree with that of our rooted MLSA species tree shows that all Pantoea desferrioxamine clusters nest inside the Erwinia lineage, which is defined by the two basal groups composed of E. mallotivora, E. billingiae, and E. typographi, as well as E. toletana and E. iniecta (Fig. 3). Although the two trees were incongruent, we noted that individual clades of both Pantoea and Erwinia in the DfoJAC tree that corresponded to known species relationships were conserved, with a few exceptions. For example, P. ananatis and P. stewartii group together in both trees, as do E. amylovora, E. pyrifoliae, E. tasmaniensis, and E. piriflorinigrans; however, these two clades formed sister groups, rather than nesting within lineages corresponding to their own respective genera (Fig. 3). Likewise, P. agglomerans, P. vagans, P. eucalypti, P. anthophila, and P. eucrina, which share a recent common ancestor, form a single clade in both trees. The grouping of Pantoea desferrioxamine clusters within the Erwinia lineage is suggestive of transfer of the DfoJAC cluster from Erwinia into Pantoea, while the conservation of species relationships within individual clades indicates that these transfers occurred between ancestral lineages of Erwinia and Pantoea prior to the divergence of the species groups.
Because the basal position of Erwinia in the DfoJAC tree suggests transfer of the cluster from Erwinia into Pantoea, we attempted to validate this further by analyzing the evolutionary history of one of the other cluster-associated genes, dfoS. We first identified all dfoS homologs in Erwinia and Pantoea genomes, and noted that Pantoea species that possess the desferrioxamine cluster carry two dfoS copies, while all Erwinia strains carried only one copy (Fig. 4). A phylogenetic analysis of dfoS yielded two major groups: one composed of all Pantoea strains, and the other composed of all the Erwinia strains (Fig. 4). The second, cluster-associated dfoS homolog of Pantoea grouped in together with the Erwinia strains.
The Pantoea dfoS genes that were not associated with the cluster formed a clade corresponding to the Pantoea species tree, and all were flanked by the same group of homologs pykA, msbB, yebA, znuA, and znuC in their respective genomes (Fig. 4). This dfoS ortholog appears to be the native copy that is present in all Pantoea strains and appears to have been inherited vertically. Similarly, the dfoS alleles of Erwinia formed a single clade and were also flanked by the same pykA, msbB, yebA, znuA, and znuC genes (except in E. oleae) indicating this is the native Erwinia dfoS allele. Notably, some dfoS alleles are cluster associated, while others are not. The dfoS alleles that are associated with the Pantoea desferrioxamine clusters grouped within this Erwinia lineage, indicating that the cluster-associated dfoS alleles in Pantoea are also of Erwinia origin and are therefore xenologous. This is strong evidence for acquisition of the dfoJAC genes together with the cluster-associated dfoS by Pantoea ancestors from Erwinia ancestors. Two copies of dfoS genes were also found within the outgroup, Rahnella aquatilis, one associated with the desferrioxamine cluster and the other outside the cluster (Fig. 4). The cluster-associated dfoS gene of R. aquatilis is also flanked by the same five genes found across the majority of Pantoea and Erwinia strains, suggesting this is the ancestral state.
To better understand the evolution of the cluster, we examined the foxR gene, which was associated with several of the desferrioxamine clusters. We searched for all foxR genes across Pantoea and Erwinia genomes, including in strains that lack the cluster, and found that all strains have only one foxR allele. A FoxR genealogy revealed that the majority of foxR alleles of Pantoea formed a single clade that reflected the species tree, and were also found in the same genomic context, flanked by the yeaQ, treA, ycdW, and ycdX genes (Fig. 5), supporting vertical inheritance.
The exceptions were P. eucrina, P. stewartii, and P. ananatis whose foxR gene, which is linked to the desferrioxamine cluster, grouped with all the Erwinia alleles, further reinforcing horizontal acquisition with the cluster (Fig. 5). Notably, P. eucrina, P. ananatis, and P. stewartii, which have only the cluster-associated Erwinia allele of foxR, have lost the native foxR copy that is flanked by the yeaQ, treA, ycdW, and ycdX genes in the other Pantoea strains (Fig. 5). In Erwinia, foxR genes not linked to the cluster are flanked by at least two of the conserved genes, while in strains where foxR is associated with the cluster these same flanking genes can be identified, but foxR is absent in between them (Fig. 5). This supports the acquisition of the Erwinia desferrioxamine cluster along with dfoS and foxR allele by some ancestral lineages of Pantoea.

Discussion

A survey of the siderophore biosynthetic clusters across Pantoea identified gene clusters homologous to enterobactin, desferrioxamine, and aerobactin. The clusters in Pantoea matched antiSMASH reference clusters to varying percentages, although our analysis showed that gene clusters identified as turnerbactin were more similar to the enterobactin biosynthetic gene cluster (Supplementary Fig. S1). At least four genes are shared between enterobactin and turnerbactin, which is consistent with our current understanding of the evolution of siderophore clusters; namely, that those of the same structural type share similar biosynthetic steps and, thus, similar genes (Challis 2005). Although the biosynthetic clusters we identified in Pantoea are homologous to previously identified siderophore gene clusters, the specific chemical structure of the siderophores produced by Pantoea strains may differ. For example, the Pantoea enterobactinlike cluster did not contain the biosynthetic entD gene found in the enterobactin biosynthetic cluster of Salmonella or the biosynthetic schG gene found in the Serratia serratiochelin biosynthetic cluster (Supplementary Fig. S1). It is therefore possible that the enterobactin-like cluster of Pantoea directs the synthesis of a siderophore that is structurally different from either enterobactin or serratiochelin. In contrast, both the predicted desferrioxamine and aerobactin clusters of Pantoea possess all the genes found in their respective prototypical gene clusters, such that the siderophore structure may be more similar to the chemical architectures described for other species groups.
Each of the three biosynthetic clusters exhibited a unique and distinct evolutionary history. The enterobactin-like cluster was widely distributed across Pantoea species and the phylogeny was congruent with the MLSA species tree (Fig. 1b), which is consistent with the vertical inheritance of the cluster. This was supported by the conservation of the flanking genes, which suggests that the common evolutionary ancestor of all Pantoea species possessed the enterobactin-like cluster. The absence of the cluster in P. eucalypti, P. eucrina, and sister taxa P. ananatis and P. stewartii is therefore attributable to three possible independent loss events, with the conservation of the flanking genes of the enterobactin-like cluster in these species suggesting only the cluster itself has been lost (Fig. 1b). In P. ananatis, a single remnant (entH) still remains between the conserved flanking genes. Moreover, because all P. stewartii and P. ananatis strains lack the cluster, this loss event is evolutionarily old and occurred in the common ancestor prior to the divergence of these two species. All of these four species carry other siderophore biosynthetic clusters, indicating either selectively neutral loss of enterobactin, or selection for the replacement of enterobactin by an alternative siderophore that may be involved in niche-specific adaptation for these particular species groups.
Aerobactin had a considerably more limited distribution than enterobactin, being present only in P. septica, P. ananatis, and P. stewartii, suggesting horizontal acquisition that was supported by our phylogenetic analysis. Furthermore, because the Pantoea biosynthetic clusters grouped separately in the iucABCD tree (Fig. 2b), two independent horizontal acquisition events from two different sources have occurred. The P. septica cluster is most closely related to clusters found in Enterobacter cloacae, while those of P. ananatis and P. stewartii group with Shigella boydii and Salmonella enterica. P. septica is the cause of various infections in humans and is therefore frequently obtained from clinical samples, including sputum and blood (Walterson and Stavrinides 2015), possibly suggesting a role in host colonization. Aerobactin has been found to be critical for host association and virulence, such as for hypervirulent Klebsiella pneumoniae ST86 (K2 serotype), which produces multiple siderophores including aerobactin, enterobactin, salmochelin, and yersiniabactin (Russo et al. 2015). When aerobactin biosynthesis is compromised, K. pneumoniae exhibits decreased virulence and survival in mouse infection models (Russo et al. 2015). The inhibition of production of the other siderophores either alone or in combination did not reduce virulence, indicating aerobactin is a critical virulence factor for hypervirulent K. pneumoniae (Russo et al. 2015). In addition, disruption of aerobactin biosynthesis decreases growth and/or survival of K. pneumoniae in human ascites fluid and serum ex vivo. The acquisition of aerobactin by P. septica may reflect an adaptation for exploiting eukaryotic hosts. Both P. ananatis and P. stewartii also gained aerobactin, an evolutionary event that appears to have occurred prior to speciation. The ancestor of these two species also lost the enterobactin-like biosynthetic gene cluster prior to speciation, although the timing of these two events is unclear. The gain and loss of specific siderophores may be selectively neutral, or may be reflective of adaptation given that aerobactin has been shown to be important for both plant and animal pathogenicity (Russo et al. 2014; Burbank et al. 2015). P. ananatis contains plant pathogenic strains, but also strains that are opportunistic pathogens of humans (De Baere et al. 2004; Coutinho and Venter 2009). Likewise, P. stewartii, a well-documented plant pathogen that causes Stewart’s wilt in maize, requires aerobactin for plant pathogenesis (Roper 2011; Burbank et al. 2015); however, P. stewartii also colonizes the flea beetle as part of its life cycle (Correa et al. 2012), although it is not clear whether aerobactin is important for the establishment of the bacterium within its eukaryotic vector. The acquisition of aerobactin by P. ananatis, P. stewartii, and P. septica from known animal pathogenic species may suggest that this siderophore may have been important to the evolution of host association in these species.
The desferrioxamine gene cluster was widely distributed across most Pantoea and Erwinia species, with previous research suggesting that it was present in the common ancestor prior to the divergence of the two genera (Smits and Duffy 2011). When we rooted our dfoJAC genealogy with an outgroup, we found that the branching pattern was incongruent with the species tree, with two distinct clades of Pantoea species falling inside an Erwinia group that is formed by a basal lineage of E. iniecta and E. toletana. This is strong evidence that the dfoJAC clusters of Pantoea were acquired from Erwinia, and the congruence of the Pantoea and Erwinia clades to known species relationships reflects that these transfers occurred prior to the diversification of Pantoea and Erwinia lineages. The phylogenetic pattern we observed resembles the model of ancient horizontal gene transfer proposed by Keeling and Palmer (Keeling and Palmer 2008) in which individual monophyletic lineages are preserved in the gene tree, but their location relative to the other species within the species tree are different. This was supported by a phylogenetic analysis of the dfoS and foxR genes, which were associated with only some of the desferrioxamine clusters. Non-cluster-associated dfoS alleles were found in all Pantoea species and formed a monophyletic group in the DfoS tree that is congruent with the Pantoea species tree (Fig. 3), indicating that this ortholog was likely vertically inherited and represents the ancestral Pantoea allele. The cluster-associated dfoS alleles of Pantoea nest together within the Erwinia alleles, indicating that the cluster-associated dfoS alleles of Pantoea originated in Erwinia and were inherited together with the biosynthetic cluster. Consistent with this, Pantoea species that carry the cluster carry two dfoS alleles: the Erwinia allele that is associated with the desferrioxamine cluster and their ancestral copy. What does the foxR say? Our analysis of the foxR gene also supported transfer of the desferrioxamine cluster from Erwinia into Pantoea. The foxR alleles associated with the Pantoea desferrioxamine clusters group together with the Erwinia foxR alleles, indicating that the foxR allele was transferred to these groups along with the cluster. Although we were expecting these strains to have two copies of the foxR gene as with dfoS, their native Pantoea foxR allele has been lost from within a region that is conserved across all Pantoea strains. All other Pantoea strains retain their ancestral foxR gene, and the evolutionary history of these alleles is consistent with vertical inheritance as shown by a single, strongly supported clade that mirrors the species tree.
The gain and loss of siderophore biosynthetic gene clusters may also be influenced by microbial social dynamics. Specifically, the Black Queen Hypothesis (Morris et al. 2012) states that gene loss can actually benefit an organism by preventing the expenditure of resources on production of the encoded products. However, this organism must retain the ability to obtain the product produced by other members of its community and likewise some members of the community must retain the genes required for production. In clinical populations of Pseudomonas aeruginosa, mutants which no longer possess the ability to produce the siderophore pyoverdine frequently arise and proliferate; however, these mutants retain the pyoverdine receptor that allows them to exploit the pyoverdine produced by other members of the population (Andersen et al. 2015). This concept could help partially explain the presence of ancestral foxR receptor genes in Pantoea species that do not possess the desferrioxamine biosynthetic gene cluster. Possessing the foxR receptor alone likely allows these Pantoea species to pillage desferrioxamine that has been secreted by other organisms, such as Erwinia. Although there is evidence of non-orthologous gene displacement of the ancestral foxR in Pantoea species that have acquired a desferrioxamine cluster that includes foxR, in the remainder of Pantoea species the foxR allele appears ancestral (Fig. 5). Additionally, the siderophore receptor genes fepA and iutA of enterobactin and aerobactin, respectively, were not found in any Pantoea species that did not possess the entire biosynthetic cluster for the corresponding siderophore. Furthermore, the biosynthetic gene clusters of enterobactin, desferrioxamine, and aerobactin were present in all strains of the Pantoea species that possessed them (Table 2). It is therefore unclear whether the Black Queen Hypothesis applies given that we would expect to find some strains within a Pantoea species that have a given biosynthetic cluster and some that do not. For example, within species groups of the bacterial genus Vibrio, only 40% of strains possess siderophore biosynthetic capability, while the remaining strains retain the siderophore receptor only (Cordero et al. 2012).
The host-associating capabilities of different Pantoea species remain largely unexplored, yet there is strong evidence for horizontal acquisition of candidate virulence determinants, including type III secretion systems (Kirzinger et al. 2015), type VI secretion systems (De Maayer et al. 2011), as well as other genetic determinants, including biosurfactants (Smith et al. 2016), and antibiotics (Walterson et al. 2014). Our analysis of three siderophore biosynthetic clusters across species of Pantoea revealed evidence of horizontal transfers, including between human-pathogenic species of Pantoea and other human-pathogenic enterics. Because of the very prominent role of siderophores in host association among the Enterobacterales, the gain of siderophores by these human-pathogenic Pantoea species groups from other human pathogens suggests that the roles siderophores play, particularly in opportunistic infections in humans, warrant further exploration.

References

Andersen SB, Marvig RL, Molin S et al (2015) Long-term social dynamics drive loss of function in pathogenic bacteria. Proc Natl Acad Sci 112:10756–10761. https ://doi.org/10.1073/pnas.15083 24112
Bentley SD, Chater KF, Cerdeño-Tárraga A-M et al (2002) Complete genome sequence of the model actinomycete Streptomyces coelicolor A3(2). Nature 417:141–147. https: //doi.org/10.1038/41714 1a
Berner I, Konetschny-Rapp S, Jung G, Winkelmann G (1988) Characterization of ferrioxamine E as the principal siderophore of Erwinia herbicola (Enterobacter agglomerans). Biol Met 1:51– 56. https ://doi.org/10.1007/BF011 28017
Besemer J (2001) GeneMarkS: a self-training method for prediction of gene starts in microbial genomes. Implications for finding sequence motifs in regulatory regions. Nucleic Acids Res 29:2607–2618. https ://doi.org/10.1093/nar/29.12.2607
Brady CL, Cleenwerck I, Venter SN et al (2010) Emended description of the genus Pantoea, description of four species from human clinical samples, Pantoea septica sp. nov. Pantoea eucrina sp. nov., Pantoea brenneri sp. nov. and Pantoea conspicua sp. nov., and transfe. Int J Syst Evol Microbiol 60:2430–2440. https ://doi.org/10.1099/ijs.0.01730 1-0
Burbank L, Mohammadi M, Roper MC (2015) Siderophore-mediated iron acquisition influences motility and is required for full virulence of the xylem-dwelling bacterial phytopathogen Pantoea stewartii subsp. stewartii. Appl Environ Microbiol 81:139–148. https ://doi.org/10.1128/AEM.02503 -14
Campestre MP, Castagno LN, Estrella MJ, Ruiz OA (2016) Lotus japonicus plants of the Gifu B-129 ecotype subjected to alkaline stress improve their F e2 + bio-availability through inoculation with Pantoea eucalypti M91. J Plant Physiol 192:47–55. https ://doi.org/10.1016/j.jplph .2016.01.001
Challis GL (2005) A widely distributed bacterial pathway for siderophore biosynthesis independent of nonribosomal peptide synthetases. ChemBioChem 6:601–611. https ://doi.org/10.1002/ cbic.20040 0283
Cordero OX, Ventouras L-A, DeLong EF, Polz MF (2012) Public good dynamics drive evolution of iron acquisition strategies in natural bacterioplankton populations. Proc Natl Acad Sci 109:20059–20064. https ://doi.org/10.1073/pnas.12133 44109
Correa VR, Majerczak DR, Ammar ED et al (2012) The bacterium Pantoea stewartii uses two different type III secretion systems to colonize its plant host and insect vector. Appl Environ Microbiol 78:6327–6336. https ://doi.org/10.1128/AEM.00892 -12
Coutinho TA, Venter SN (2009) Pantoea ananatis: an unconventional plant pathogen. Mol Plant Pathol 10:325–335. https :// doi.org/10.1111/j.1364-3703.2009.00542 .x
Crosa JH, Walsh CT (2002) Genetics and assembly line enzymology of siderophore biosynthesis in bacteria. Microbiol Mol Biol Rev 66:223–249. https: //doi.org/10.1128/MMBR.66.2.223-249.2002
Crouch MLV, Castor M, Karlinsey JE et al (2008) Biosynthesis and IroC-dependent export of the siderophore salmochelin are essential for virulence of Salmonella enterica serovar Typhimurium. Mol Microbiol 67:971–983. https ://doi.org/10.1111/j.1365-2958.2007.06089 .x
De Baere T, Verhelst R, Labit C et al (2004) Bacteremic infection with Pantoea ananatis. J Clin Microbiol 42:4393–4395. https ://doi.org/10.1128/JCM.42.9.4393-4395.2004
De Maayer P, Venter SN, Kamber T et al (2011) Comparative genomics of the type VI secretion systems of Pantoea and Erwinia species reveals the presence of putative effector islands that may be translocated by the VgrG and Hcp proteins. BMC Genom 12:576. https ://doi.org/10.1186/1471-2164-12-576
De Maayer P, Chan WY, Rezzonico F et al (2012) Complete genome sequence of clinical isolate Pantoea ananatis LMG 5342. J Bacteriol 194:1615–1616. https ://doi.org/10.1128/JB.06715 -11
Dellagi A, Brisset MN, Paulin JP, Expert D (1998) Dual role of desferrioxamine in Erwinia amylovora pathogenicity. Mol Plant Microbe Interact 11:734–742. https ://doi.org/10.1094/MPMI.1998.11.8.734
Dellagi A, Segond D, Rigault M et al (2009) Microbial siderophores exert a subtle role in arabidopsis during infection by manipulating the immune response and the iron status. Plant Physiol 150:1687–1696. https ://doi.org/10.1104/pp.109.13863 6
Frueh DP, Arthanari H, Koglin A et al (2008) Dynamic thiolationthioesterase structure of a non-ribosomal peptide synthetase. Nature 454:903–906. https ://doi.org/10.1038/natur e0716 2
Han AW, Sandy M, Fishman B et al (2013) Turnerbactin, a novel triscatecholate siderophore from the shipworm endosymbiont Teredinibacter turnerae T7901. PLoS One 8:e76151. https :// doi.org/10.1371/journ al.pone.00761 51
Heesemann J (1987) Chromosomal-encoded siderophores are required for mouse virulence of enteropathogenic Yersinia species. FEMS Microbiol Lett 48:229–233. https ://doiorg/10.1111/j.1574-6968.1987.tb025 47.x
Hider RC, Kong X (2010) Chemistry and biology of siderophores. Nat Prod Rep 27:637–657
Kadi N, Challis GL (2009) Chap. 17 siderophore Deferoxamine biosynthesis. A substrate specificity assay for nonribosomal peptide synthetaseindependent siderophore synthetases involving trapping of acyladenylate intermediates with hydroxylamine. Methods Enzymol 458:431–457
Keeling PJ, Palmer JD (2008) Horizontal gene transfer in eukaryotic evolution. Nat Rev Genet 9:605–618. https ://doi.org/10.1038/ nrg23 86
Kirzinger MWB, Butz CJ, Stavrinides J (2015) Inheritance of Pantoea type III secretion systems through both vertical and horizontal transfer. Mol Genet Genomics 290:2075–2088. https :// doi.org/10.1007/s0043 8-015-1062-2
Köster W (2001) ABC transporter-mediated uptake of iron, siderophores, heme and vitamin B12. Res Microbiol 152:291–301. https ://doi.org/10.1016/S0923 -2508(01)01200 -1
Kumar S, Stecher G, Tamura K (2016) MEGA7: molecular evolutionary genetics analysis version 7.0 for bigger datasets. Mol Biol Evol 33:1870–1874. https ://doi.org/10.1093/molbe v/ msw05 4
Ling J, Pan H, Gao Q et al (2013) Aerobactin synthesis genes iucA and iucC contribute to the pathogenicity of avian pathogenic Escherichia coli O2 strain E058. PLoS One. https ://doi. org/10.1371/journ al.pone.00577 94
Loaces I, Ferrando L, Scavino AF (2011) Dynamics, diversity and function of endophytic siderophore-producing bacteria in rice. Microb Ecol 61:606–618. https ://doi.org/10.1007/s0024 8-010-9780-9
Medema MH, Kottmann R, Yilmaz P et al (2015) Minimum information about a biosynthetic gene cluster. Nat Chem Biol 11:625– 631. https ://doi.org/10.1038/nchem bio.1890
Miethke M, Marahiel MA (2007) Siderophore-Based Iron Acquisition and Pathogen Control. Microbiol Mol Biol Rev 71:413–451. https ://doi.org/10.1128/MMBR.00012 -07
Morris JJ, Lenski RE, Zinser ER (2012) The black queen hypothesis: evolution of dependencies through adaptative gene loss. MBio 3:1–7. https ://doi.org/10.1128/mBio.00036 -12.Copyr ight
Nadarasah G, Stavrinides J (2014) Quantitative evaluation of the host-colonizing capabilities of the enteric bacterium Pantoea using plant and insect hosts. Microbiol (U K) 160:602–615. https ://doi.org/10.1099/mic.0.07345 2-0
Neilands JB (1995) Siderophores: structure and function of microbial iron transport compounds. J Biol Chem 270:26723–26726
Quadri LEN, Sello J, Keating TA et al (1998) Identification of a Mycobacterium tuberculosis gene cluster encoding the biosynthetic enzymes for assembly of the virulence-conferring siderophore mycobactin. Chem Biol 5:631–645. https ://doi. org/10.1016/S1074 -5521(98)90291 -5
Raffatellu M, George MD, Akiyama Y et al (2009) Lipocalin-2 resistance confers an advantage to Salmonella enterica serotype Typhimurium for growth and survival in the inflamed intestine. Cell Host Microbe 5:476–486. https ://doi.org/10.1016/j. chom.2009.03.011
Ratledge C, Dover LG (2000) Iron metabolism in pathogenic bacteria. Annu Rev Microbiol 54:881–941. https ://doi.org/10.1146/ annur ev.micro .54.1.881
Roper MC (2011) Pantoea stewartii subsp. stewartii: lessons learned from a xylem-dwelling pathogen of sweet corn. Mol Plant Pathol 12:628–637. https ://doi.org/10.1111/j.1364-3703.2010.00698 .x
Russo TA, Olson R, MacDonald U et al (2014) Aerobactin mediates virulence and accounts for increased siderophore production under iron-limiting conditions by hypervirulent (hypermucoviscous) Klebsiella pneumoniae. Infect Immun 82:2356–2367. https ://doi.org/10.1128/IAI.01667 -13
Russo TA, Olson R, MacDonald U et al (2015) Aerobactin, but not yersiniabactin, salmochelin, or enterobactin, enables the growth/ survival of hypervirulent (hypermucoviscous) Klebsiella pneumoniae ex vivo and in vivo. Infect Immun 83:3325–3333. https: // doi.org/10.1128/IAI.00430 -15
Searle LJ, Méric G, Porcelli I et al (2015) Variation in siderophore biosynthetic gene distribution and production across environmental and faecal populations of Escherichia coli. PLoS One. https: //doi. org/10.1371/journ al.pone.01179 06
Sengupta M, Banerjee S, Das NK et al (2016) Early onset neonatal septicaemia caused by Pantoea agglomerans. J Clin Diagn Res 10:DD01–DD02. https: //doi.org/10.7860/JCDR/2016/19613.780 7
Seyedsayamdost MR, Cleto S, Carr G et al (2012) Mixing and matching siderophore clusters: structure and biosynthesis of serratiochelins from Serratia sp. V4. J Am Chem Soc 134:13550–13553. https: // doi.org/10.1021/ja304 941d
Sievers F, Wilm A, Dineen D et al (2011) Fast, scalable generation of high-quality protein multiple sequence alignments using Clustal Omega. Mol Syst Biol. https ://doi.org/10.1038/msb.2011.75
Smith DDN, Nickzad A, Déziel E, Stavrinides J (2016) A novel glycolipid biosurfactant confers grazing resistance upon Pantoea ananatis BRT175 against the social amoeba Dictyostelium discoideum. mSphere 1:e00075–e00015. https ://doi.org/10.1128/ mSphe re.00075 -15
Smits THM, Duffy B (2011) Genomics of iron acquisition in the plant pathogen Erwinia amylovora: Insights in the biosynthetic pathway of the siderophore desferrioxamine E. Arch Microbiol 193:693– 699. https ://doi.org/10.1007/s0020 3-011-0739-0
Suzuki K, Tanabe T, Moon YH et al (2006) Identification and transcriptional organization of aerobactin transport and biosynthesis cluster genes of Vibrio hollisae. Res Microbiol 157:730–740. https: //doi. org/10.1016/j.resmi c.2006.05.001
Tamura K, Stecher G, Peterson D et al (2013) MEGA6: molecular evolutionary genetics analysis version 6.0. Mol Biol Evol 30:2725–2729. https ://doi.org/10.1093/molbe v/mst19 7
Thompson FL, Hoste B, Vandemeulebroecke K, Swings J (2003) Reclassification of Vibrio hollisae as Grimontia hollisae gen. nov., comb. nov. Int J Syst Evol Microbiol 53:1615–1617. https ://doi. org/10.1099/ijs.0.02660 -0
Walterson AM, Stavrinides J (2015) Pantoea: Insights into a highly versatile and diverse genus within the Enterobacteriaceae. FEMS Microbiol Rev 39:968–984. https: //doi.org/10.1093/femsre/fuv02 7
Walterson AM, Smith DDN, Stavrinides J (2014) Identification of a Pantoea biosynthetic cluster that directs the synthesis of an antimicrobial natural product. PLoS One 9:e96208. https ://doi. org/10.1371/journ al.pone.00962 08
Weber T, Blin K, Duddela S et al (2015) AntiSMASH 3.0-A comprehensive resource for the genome mining of biosynthetic gene clusters. Nucleic Acids Res 43:W237–W243. https: //doi.org/10.1093/ nar/gkv43 7
Weinberg ED (1993) The development of awareness of iron-withholding defense. Perspect Biol Med 36:215–221. https ://doi. org/10.1353/pbm.1993.0063