By means of baiting with the conserved proteins from several characterized T6SS in other bacteria, between two and four T6SS loci were identified in the complete or draft genome sequences of five Pantoea and four Erwinia strains (Figure 1). These strains were the Eucalyptus pathogen P. ananatis LMG 20103 31, the biological control strains P. vagans C9-1 32 and P. agglomerans E325 (Smits and Duffy, unpublished results), the insect-associated strains Pantoea sp. aB-valens 29 and Pantoea sp. At-9b 30 which represent novel species of the genus Pantoea, the fire blight pathogens E. amylovora CFBP 1430 22 and E. pyrifoliae DSM 12163^T 33 and the apple epiphytes E. tasmaniensis Et1/99 34 and E. billingiae Eb661 25. Two orthologous loci are found in all Pantoea and Erwinia strains, while a more restricted distribution could be observed for the T6SS-3 locus (Figure 1). Additional T6SS loci were identified in the genome sequences of E. billingiae Eb661 and Pantoea sp. aB-valens, respectively.

The T6SS-1 loci range in size from 28.8 and 40.0 kb and are built around a well-conserved and syntenous core which includes orthologs of all sixteen conserved T6SS proteins identified by Boyer et al. 15 (Figure 2 and 3). Additional genes encode a serine/threonine protein kinase (PpkA) and serine/threonine phosphatase (PppA). These proteins interact with the Fork Head-Associated (FHA) domain protein (COG3456) for the post-translational regulation of the T6SS 35. In both Pantoea and Erwinia strains the conserved core T6SS proteins are arranged in two syntenous clusters (block I and III - Figure 2 and 3). These are separated by a variable region (block II), which is linked to the hcp gene (COG3157). A further variable region occurs next to the vgrG gene (block IV). In all sequenced Pantoea and Erwinia genomes, with the exception of E. pyrifoliae DSM 12163^T, this variable region is flanked at the 3' end by a gene encoding a re-arrangement hot spot (rhs) element 36. In P. agglomerans E325, E. amylovora CFBP 1430, E. pyrifoliae DSM 12163^T and E. tasmaniensis Et1/99, an additional copy of the vgrG gene occurs within this variable region. A non-conserved region harboring two hypothetical genes, which is not associated with a vgrG or hcp gene could be observed between COG3520 and clpV in E. tasmaniensis Et1/99 and E. pyrifoliae DSM 12163^T (Figure 3).

The T6SS-2 loci of the Pantoea strains and E. billingiae Eb661 are structurally and genetically conserved and consist of eight syntenous genes which encode proteins that are highly conserved among the sequenced strains (Figure 4). Only four of these belong to conserved T6SS proteins outlined by Boyer et al. 15, suggesting the T6SS-2 loci encodes a partial and potentially non-functional T6SS. An average amino acid identity of 52% could be observed between the proteins encoded in the T6SS-2 locus and their respective paralogs in the T6SS-1 locus for each of the strains. This suggests that the T6SS-2 locus could have arisen through a partial duplication of the T6SS-1 locus. This is supported by phylogenetic analysis with the IcmF (COG3523) protein sequences (Figure 5), where the T6SS-2 loci branch from the T6SS-1 loci. The E. amylovora CFBP 1430 and E. pyrifoliae DSM12613^T T6SS-2 loci encode only four proteins, lacking the genes encoding the serine/threonine kinase and phosphatase required for posttranslational regulation, while an FHA domain (COG3456) protein is encoded in these loci. The icmF gene in the E. pyrifoliae DSM12613^T T6SS-2 locus is inactivated by a frameshift 33, while the E. tasmaniensis Et1/99 T6SS-2 locus is missing the FHA domain gene, indicating further gene loss in these T6SS loci (Figure 4).

The Pantoea and Erwinia T6SS-3 loci are between 18.8 and 34.6 kb in size and encompass 19 to 32 protein coding sequences. In contrast to the T6SS-1 and -2 loci, the T6SS-3 locus is not universal to all Pantoea and Erwinia strains (Figure 1). The phylogeny based on the IcmF protein shows a greater evolutionary distance between the Pantoea and Erwinia T6SS-3 loci, which are interspersed with those of other Enterobacteriaceae (Figure 5), indicating that this locus may have been acquired through horizontal gene transfer. This can also be correlated with greater diversity in gene content and order between the various Pantoea and Erwinia T6SS-3 loci (Figure 6 and 7). Thirteen of the T6SS conserved core proteins identified by Boyer et al. 15 are encoded within the T6SS loci. Genes encoding orthologs of the COG3913, COG4455 and FHA domain (COG3456) proteins are absent. No PpkA or PppA orthologs are encoded in the T6SS-3 loci either, suggesting that this locus is not post-translationally regulated. As in the case of the T6SS-1 loci, syntenous blocks of core genes can be observed in the T6SS-3 loci (block I, III and V - Figure 6 and 7). Here, the hcp gene is included in a conserved block (Figure 6 and 7). The non-conserved regions in the T6SS-3 loci are generally associated with a vgrG gene. An additional complete and two partial vgrG genes can be found in the T6SS-3 locus of E. amylovora CFBP 1430. An additional non-conserved region, encoding two predicted proteins, PANA_4135 and PANA_4136 can be observed in the P. ananatis LMG 20103 T6SS-3 locus (Figure 7).

A further T6SS locus was identified in Pantoea sp. aB-valens (Figure 8). This 19.6 kb locus encodes thirteen conserved core proteins. The IcmF phylogeny shows this T6SS locus to be distantly related to all other Pantoea and Erwinia T6SS loci. Similarly, an additional partial T6SS locus is encoded on the E. billingiae Eb661 genome, which contains four T6SS core proteins. A non-conserved region associated with the vgrG gene is likewise present. This E. billingiae Eb661 T6SS locus is flanked at both the 5' (EbC_39320 - phage integrase) and 3' end (EbC_393430-39600 - hypothetical phage genes) by genes of an integrated phage element. A phage origin has previously been hypothesized for the T6SS 17.

The amino acid sequences of all VgrG proteins encoded in the T6SS-1, T6SS-3 and additional T6SS loci in E. billingiae and Pantoea sp. aB-valens were analyzed for the presence of conserved and evolved domains. The N-terminal region of all T6SS-1, T6SS-3 and additional T6SS VgrG proteins is conserved among all Pantoea and Erwinia species and contains a conserved VgrG (TIGR3361 37) and phage Gp5 (pfam04717 38) domain (Figure 9). However, as has been observed in other bacteria, the VgrG proteins differ considerably in length between the various Pantoea and Erwinia species. This can be attributed to variable C-terminal regions, which suggest that they represent evolved VgrG proteins. Seven of the thirteen Pantoea and Erwinia T6SS-1 VgrG proteins contain such C-terminal extensions (Figure 10). Analysis of these C-terminal extensions showed that several of them contain conserved domains. The P. agglomerans E325 T6SS-1 VgrG2 protein (Pagg_1105) contains a peptidoglycan (PG)-binding domain (Pfam09374 38 - Cdsearch score: 45) as well as COG3926 and COG5526 lysozyme domains 3940, which were also found in the lytic bacteriophage φ8, suggesting a bacteriolytic function for this VgrG protein. A similar function can be expected for the Pantoea sp. aB-valens T6SS-1 VgrG (PANABDRAFT_2668) which contains a β-N-acetyl-glucosaminidase domain (COG4193 40 - Cdsearch score: 58.5) in its C-terminal extension. The E. tasmaniensis Et1/99 T6SS-1 VgrG2 (ETA_06370) C-terminal extension shows structural homology to the Streptomyces sp. N174 chitosanase (1_chk_A - Hhpred score: 190) 41.

The C-terminal extensions of VgrG proteins encoded in the T6SS-3 loci of all Pantoea and Erwinia species contain a conserved domain of unknown function (COG4253), which is absent in the T6SS-1 loci VgrG proteins (Figure 9). This domain has also been identified flanking the N-terminal region of the S. enterica subsp. arizonae IIIa VgrG proteins, while it is not present in the VgrG proteins of other Salmonella serovars 14. A role for the COG4253 domain, which links the N-terminal VgrG transporter and C-terminal extension, in modulating the function of the VgrG between secretion and virulence has been suggested 15. However, with the exception of the P. agglomerans E325 T6SS-3 VgrG protein, which carries a C-terminal extension subsequent to the COG4253 region, no further C-terminal extensions are present in the Pantoea and Erwinia T6SS-3 VgrG proteins. An alternative function in host cell adhesion has also been suggested 16. The latter function would imply that the COG4253-positive T6SS-3 and COG4253-negative T6SS-1 VgrG proteins could have different targets and serve different functions. A phylogeny based on the T6SS-1 and T6SS-3 VgrG proteins shows that they branch into two distinct clades suggesting distinct evolutionary backgrounds for these paralogous proteins (Figure 10). The VgrG protein found in the additional T6SS locus in E. billingiae Eb661 also contains a conserved COG4253 domain and clusters with the T6SS-3 VgrG proteins, while the additional Pantoea sp. aB-valens VgrG paralog clusters with the other VgrG proteins lacking this domain.

While the T6SS-1 and T6SS-3 loci share conserved and syntenous cores among the various Pantoea and Erwinia strains, considerable variability in the vgrG and hcp regions can be observed. The G+C deviation across these T6SS loci was determined. This showed that for the T6SS-1 and T6SS-3 loci, there is a much lower G+C content in the variable regions associated with the hcp and vgrG genes compared to the conserved core regions (Figures 2 and 4). In the T6SS-1 loci, the conserved core regions had an average G+C content of 60.17% across all strains, while the hcp regions (average G+C% = 42.30) and vgrG regions (average G+C% = 47.91) had a much lower G+C content. The substantial G+C deviations, variability in the gene content of the hcp and vgrG regions and the differential homology of proteins encoded in these regions to proteins in bacteria in other genera indicates that these regions represent variable islands within the T6SS loci. Similar G+C deviations could be observed for the vgrG regions in the additional Pantoea sp. aB-valens and E. billingiae Eb661 T6SS loci, which further supports that these regions serve as "hot spots" for rearrangement 36. These hot spots are frequently associated with Rhs proteins which are capable of displacing its C-terminal tip and replacing it with a non-homologous alternative. By this means, the Rhs protein can drive the sequential insertion of heterogeneous C-terminal sequences into the hot spot 36. As the additional E. billingiae Eb661 T6SS locus occurs within an integrated phage element, we postulate that transducing phages may play a role in the horizontal acquisition of non-conserved genes in the vgrG and hcp regions. Similar vgrG/hcp islands have also been identified in a number of Pseudomonas species 42. These islands are associated with "orphan" vgrG and hcp paralogs separately located from the remainder of the T6SS loci. In contrast, this is the first observation of hcp and vgrG islands associated with the conserved core T6SS loci. Analysis of the T6SS loci of other bacteria (data not shown), however, shows that this phenomenon is not restricted to Pantoea and Erwinia.

The proteins encoded in the variable hcp and vgrG islands in the T6SS-1 and T6SS-3 loci in the different Pantoea and Erwinia species were analyzed for sequence similarity and structural homology to known proteins and the presence of conserved domains. The majority of proteins encoded on the islands showed homology to proteins of unknown function. However, a number of island proteins share high sequence identity and contain conserved domains which suggest they may represent T6SS effectors with putative functions in host-microbe and inter-bacterial interactions (Additional File 1 Table S1).

The amino acid sequence of ETA_06430 encoded in the E. tasmaniensis Et1/99 T6SS-1 vgrG island shares homology with a phospholipase in V. cholerae HE48 (VCHE48_2681 - 40% aa identity) while that of E. billingiae Eb661 EbC_4130 encoded on the T6SS-3 vgrG island shows homology to a lipase/esterase in Yersinia bercovieri ATCC 43970 (Yberc0001_35290 -64% aa identity). The former also shows structural homology to a P. aeruginosa lipase (1ex9_A - Hhpred score: 115) while the latter shows structural homology to a lipase in Penicillium expansum (3g7n_A: Hhpred score: 157), supporting that these two proteins represent lipases. Similarly, EbC_39360 encoded in the additional T6SS of E. billingiae Eb661 shares weak structural homology with a lipase in Archaeoglobus fulgidus (2zyr_A - Hhpred score: 81.1). Lipases hydrolyze long-chain triglycerides into fatty acids and glycerol and have been shown to represent major virulence factor in both animal and plant pathogens 4344. Genes encoding lipases were also identified in the vgrG islands of P. aeruginosa and transcriptome analysis showed that their expression is co-regulated with that of the T6SS 42. These lipases may therefore represent T6SS-secreted effectors in P. aeruginosa as well as the Erwinia species. Several genes in the vgrG regions of both Pantoea and Erwinia T6SS-1 and T6SS-3 loci may encode proteases. The P. agglomerans E325 Pagg2200 and E. tasmaniensis Et1/99 ETA_6240 proteins share 28% and 32% aa identity respectively with a putative zinc protease in Acinetobacter baumanii (AbauB_010100012633) and show weak structural homology to a zinc metalloproteases in Geobacter sulfurreducens (3c37_A - Hhpred score: 47.3 and 51.3, respectively). Furthermore, the P. agglomerans E325 Pagg_1085, E. tasmaniensis Et1/99 ETA_6210 and E. amylovora CFBP 1430 EAMY_3018 protein sequences show weak structural homology to the secreted cysteine protease stathopain (1cv8_A - Hhpred score: 29.8 34.2 and 34.2, respectively), which plays a role in skin infection by Staphylococcus aureus 45. EbC_39340 and EbC_39350 localized in the vgrG island in the additional T6SS locus of E. billingiae Eb661 encode proteins with weak structural homology to the secreted Clostridium perfringens sialidase NanI (2bf6_A - Hhpred score: 46 and 44.3, respectively) which is involved in the removal of sialic acids from host glycoconjugates with an important role in bacterial nutrition and pathogenesis 46. PANA_4143 associated with the T6SS-3 in P. ananatis LMG 20103, shares high sequence homology with the M23 peptidase of Klebsiella sp. 92-3 (HMPREF9538_05689: 78% aa identity) and extensive structural homology to the secreted chitinase G of Streptomyces coelicolor of (1chk_A - Hhpred score: 203). Chitinases degrade chitin, the carbohydrate polymer found in insect shells and the cell walls of fungi, suggesting the T6SS of P. ananatis LMG 20103 may secrete an effector with an antifungal or insecticidal function 47.

Two proteins, PANA_2363 and Pagg_2194, encoded in the T6SS-1 and T6SS-3 hcp islands of P. ananatis LMG20103 and P. agglomerans E325, respectively contain conserved peptidoglycan binding (Pfam01474 - PG_binding_1) and LysM (Pfam01476) domains which are typically associated with proteins involved in bacterial cell wall degradation. The latter also shows sequence homology to a putative lytic enzyme in Acinetobacter calcoaceticus RUH2202 (HMPREF0012_02474 - 34% aa identity). Similarly, E. pyrifoliae DSM 12163^T Epyr_0675 shares sequence homology with a Edwardsiella tarda FL6-60 lysozyme (ETAF_ple052 - 32% aa identity) and also contains a conserved peptidoglycan binding domain (Pfam09374 - PG_binding_3) and a domain of unknown function (DUF847) frequently observed in lysozymes. The E. billingiae Eb661 EbC_05851 protein also shares sequence homology with the Pseudomonas phage PaP1 endolysin (PaP1_gp072 - 46% aa identity). The presence of domains conserved in bacterial cell wall degrading enzymes and the sequence homology to these enzymes indicate that these proteins may play a similar role to the Tse1 and Tse3 lytic enzymes in P. aeruginosa in the degradation of the bacterial cell wall and may thus have a bactericidal function 1248. Furthermore, PANA_4136 encoded in the additional non-conserved region in the P. ananatis LMG 20103 T6SS-3 locus shows extensive sequence homology to S-pyocin domain containing proteins in E. coli 3030-1 (EC30301_3278 - 58% aa identity) and Yersinia pseudotuberculosis IP 31758 (YpsIP31758_0897 - 53% aa identity) and shows weak structural homology to the colicin S4 of Escherichia coli (3few_X - Hhpred score: 45.5). Similarly, weak structural homology to colicin S4 can be observed for the T6SS hcp island P. vagans C9-1 Pvag_1032, Pantoea sp. At-9b Pat9b_2515 and Pantoea sp. aB-valens PANABDRAFT_2446 proteins (3few_X - Hhpred score: 60, 58.5 and 58.4, respectively). Colicins and pyocins are bacteriocins that are involved in killing closely related bacterial species 49. The presence of bacteriocin-like proteins in the T6SS loci of these Pantoea species agrees with the finding of a potential function for the T6SS in antibiosis and competition 1112.

Analysis of the G+C contents of the Pantoea and Erwinia T6SS-1 vgrG genes shows that the N-terminal regions, which contain the conserved VgrG and Gp5 domains, have an average G+C content of 63.68%, which is similar to the conserved core of the T6SS-1 loci. By contrast, the C-terminal extensions have an average G+C content of 46.59%. This is similar to the non-conserved vgrG island, suggesting that the C-terminal extensions form part of the vgrG islands. In the T6SS-3 loci, the conserved N-terminal region of the vgrG genes has an average G+C content of 59.01% while the vgrG islands have an average G+C content of 48.38%.

Some of the proteins encoded in the vgrG islands show sequence homology and contain conserved domains found in the C-terminal extensions of evolved VgrG proteins. The E. pyrifoliae DSM 12163^T T6SS-1 Epyr_00675 amino acid sequence shares 75% aa identity with the C-terminal region of the P. agglomerans E325 T6SS-1 evolved VgrG (Pagg_1105). The P. ananatis LMG 20103 T6SS-1 PANA_2363 and the C-terminal region of the evolved T6SS-1 VgrG of P. agglomerans E325 (Pagg_1105) also each contain a PG-binding domain and a LysM domain (Pfam01476) found in lytic enzymes, and show structural homology a cell wall degrading lysozyme in Pseudomonas phage Phikz (2kbh_A - Hhpred score: 128 and 44.5, respectively). BlastP analysis with the proteins encoded in the vgrG islands against the NCBI protein database showed that proteins encoded in the vgrG islands of several Pantoea and Erwinia species show substantial sequence identity to the C-terminal extensions in VgrG in bacteria belonging to other genera (Additional File 1 Table S1). The putative lytic enzymes encoded in the T6SS-1 vgrG islands of P. ananatis LMG 20103 (PANA_2363) and E. billingiae Eb661 (EbC_05851) share homology with the C-terminal extensions in the VgrG proteins of V. cholerae TMA21 (VCB_002278 - 41% aa identity) and V. cholerae CT 5369-63 (VIH_000452 - 49% aa identity), respectively. The putative zinc proteases encoded in the P. agglomerans E325 T6SS-3 vgrG island (Ppag_2200) and the E. tasmaniensis Et1/99 T6SS-1 vgrG island (ETA_06240) share 43 and 46% aa identity with the C-terminal extension of a Burkholderia sp. 383 VgrG protein (Bcep18194_C7612). Similarly, the putative E. tasmaniensis Et1/99 lipase (ETA_06430) shows 42% aa identity to the Burkholderia glumae BGR1 VgrG (Bglu_2g02560). The sequence homology and the presence of shared conserved domains between the C-terminal extensions of evolved VgrG proteins and putative effector proteins encoded in the vgrG islands, suggest that VgrG proteins may carry effector proteins encoded in the vgrG islands. The COG4253 conserved domain found at the C-terminal end of all the T6SS-3 VgrG proteins may be involved in the anchorage of the VgrG transporter to the effector proteins, thereby modulating the function of the VgrG protein as a structural component of the T6SS apparatus and the transport of effector proteins 15. This domain is absent in the T6SS-1 loci, and the means by which effectors become associated with the VgrG transport region would need to be determined.

In the fish pathogen Edwardsiella tarda, the secreted effector protein EvpP has been shown to interact with the Hcp protein 50. Furthermore, Hcp orthologs with C-terminal extensions have been identified in S. enterica, which may represent evolved Hcp proteins 14. It is therefore plausible that, as is the case for the Pantoea and Erwinia VgrG proteins, the Hcp proteins may transport effector proteins encoded on the hcp islands. By this means, various effector proteins could be tagged to the VgrG and Hcp proteins, thereby forming different VgrG-effector and Hcp-effector combinations, which may perform different biological functions. The VgrG or Hcp proteins could transport bacteriocins or pyocins, which would allow the bacterium to target other bacteria competing for an ecological niche. A chitinase effector carried by the VgrG protein could play an antifungal or insecticidal role in the Pantoea and Erwinia species. These and other putative pathogenicity factors encoded by hypothetical proteins in the hcp and vgrG islands translocated by the VgrG and Hcp proteins may therefore enable Pantoea and Erwinia species to target a range of bacterial, invertebrate, vertebrate and or plant hosts.

The T6SS loci in five Pantoea and four Erwinia species, for which complete or draft genome sequences are available, were identified: Pantoea ananatis LMG 20103 31, Pantoea vagans C9-1 32, Pantoea agglomerans E325 (Smits and Duffy, unpublished results), Pantoea sp. aB-valens 29, Pantoea sp. At-9b 30, Erwinia amylovora CFBP 1430 22, Erwinia pyrifoliae DSM 12163^T 33, Erwinia tasmaniensis Et1/99 35 and Erwinia billingiae Eb661 25. Identification of the T6SS loci was done by BlastP analysis 51 with the conserved core proteins identified by Boyer et al. 15 against local protein databases created for the Pantoea and Erwinia strains. Proteins neighboring Pantoea and Erwinia T6SS clusters were compared using BlastP against the NCBI protein database 52 to identify to full extent of the T6SS loci. Sequence manipulations were conducted with multiple subroutines of the LASERGENE package (DNASTAR, Madison, WI, USA).

Phylogenetic analyses were done using the procedures outlined by Bingle et al. 53. A ClustalW alignment with default parameters was performed with the IcmF (COG3523) amino acid sequences, as this represents the only protein conserved among all Pantoea and Erwinia T6SS loci. A phylogenetic tree was constructed with the Molecular Evolutionary Genetics Analysis (MEGA) v. 5.0.3 software package 54, using the neighbor-joining method, with Poisson correction, complete gap deletion and bootstrapping (n = 1,000) parameters. The IcmF amino acid sequences from the T6SS loci of several closely related organisms as identified by BlastP analysis against the NCBI protein database 52 were included in the tree. This same procedure was employed to construct phylogenetic trees for Gyrase B (GyrB) and the VgrG proteins.

The average G+C contents for the conserved core regions and hcp and vgrG islands of the T6SS loci were determined using the Bioedit v.7.0.5.3 package 55. The G+C content was determined using 50 bp windows in 10 bp steps. Similarly, the G+C contents for the vgrG genes were determined for the conserved N-terminal region, which included the conserved Vgr and Gp5 domains, and for the C-terminal extension which was considered as all nucleotides located at the 3' end of the Gp5 domain. Proteins encoded in the vgrG and hcp islands were identified using the FgenesB 56 and Orf finder 57 web servers. The amino acid sequences for the proteins encoded in the hcp and vgrG islands were analyzed for sequence identity by BlastP analysis against the NCBI protein database 5152 the presence of conserved domains by Blast analysis against the Conserved Domain Database (CDsearch) 5859. Structural homology of the vgrG and hcp island proteins to those for which the chemical structure has been determined was performed using the HHpred server (http://toolkit.tuebingen.mpg.de/hhpred) 6061.