Structural genes and phylogenetic tree of the T6SS in the genus Xanthomonas
The T6SS of P. aeruginosa has been well studied both at the bioinformatic and experimental levels, and its importance in pathogenicity and in the interaction with other bacteria has been demonstrated1,15. We used the T6SS genes of Pseudomonas aeruginosa PAO1 as references to identify the T6SS genes in 44 out of the 60 evaluated genomes of the genus Xanthomonas. Our BLASTP and ORTHOMCL analyses (see methods) show that the reconstructed T6SS clusters in Xanthomonas contain between twelve and sixteen genes. For further analyses, we only considered the genomes of species that had at the minimum set of genes required for T6SS functionality48. Therefore, we selected 14 representative genomes within the Xanthomonas genus (Additional file: Table S4 bold letter) to perform phylogenetic analyses that offer insights into the evolutionary paths and diversity in organization of the T6SS in this group of plant pathogens (Figure 1). We identified each of the genes encoded in the T6SS in these genomes by OrthoMCL and BLASTP (Additional file: Table S5), and detected a subset of the genes in additional genomes (Xanthomonas cassavae str. CFBP 4642, Xanthomonas perforans str 91-18 and Xanthomonas axonopodis str.29) using EDGAR2.0 49 (Additional file: Table S6). These analyses suggest that the T6SS is present in a widespread array of species in Xanthomonas, pinpointing at the biological importance of this cluster in this genus.
Among the Xanthomonas species with putative functional T6SS, we found three different types of clusters (referred to I, II and II in Figure 1, additional file: Table S4, Table S5). Remarkably, the T6SS cluster are not the result of simple duplication events. For example, the Xoo strains have two clusters, one of which follows the phylogeny of the species, being distantly related form the phaseoli clade. However, the second cluster forms a monophyletic cluster with one of the copies from X. euvesicatoria and X. perforans. This implies that the clusters have a recent common origin and could have been horizontally acquired. We observed conserved synteny in the organization of Cluster III of Xv, Xeu, Xfa1 and Xcc3 (all members of axonopodis clade28,29). This suggests that a common origin of the clusters precedes the divergence of the pathovars. In summary, there is evidence for both vertical and horizontal inheritance of the T6SS clusters in Xanthomonas.
In agreement with previous reports 23,24, transcriptional regulators (LysR, TssB and TssA) were detected at the boundaries of some of the reconstructed T6SS clusters (I and III). The presence of noncoding RNAs in Cluster I suggests additional post-transcriptional regulation. Notably, we detected transcriptional regulators of the LysR family as a new class of regulators for Cluster I. The identified regulators may conditionally act individually or in combination to regulate their target clusters. No regulators were identified for type II clusters. Surprisingly, transcriptional regulators of the AraC family, previously propopsed as a characteristic feature of type III clusters23, were not detected in the clusters of Xfa and Xcc.
Pseudogenes 50 were noted in the T6SS clusters of some of the genomes analyzed (Figure 1, additional file: Table S5). We considered some proteins with an early stop codon that were divided in two parts as well as other genes that had shorter versions or frameshifts (Figure 1). fha, clpV and impH were among the truncated genes in the clusters I of Xvm0 and XvmN. Since Fha is a target for TagF involved in posttranslational regulation of P. aeruginosa and A. tumefaciens51, this process may be affected in these strains and they might have an alternative type of regulation, which would need to be experimentally determined. Other divided genes of the T6SS clusters include: icmF in Xvm0 and XvmN (cluster I), ompA in Xvm0 (cluster I), and clpV in Xpm (cluster III) (Figure 1). Experimental procedures are necessary to determine the function of these genes in the Xanthomonas T6SS. Sixteen Xanthomonas strains did not contain the core genes of the T6SS; some were Xanthomonas campestris pv. campestris (strain 8004, strain ATCC33913 and strain B100), Xanthomonas campestris pv. armoraciae str. 756C and Xanthomonas albilineans and were therefore considered as T6SS-depleted.
Xpm, Xeu and Xcc3 show similarity in the organization of the T6SS gene clusters
A comparative analysis was performed between Xpm and its closest relatives with completely sequenced genomes, Xeu and Xcc328. Figure 2 shows the presence of two T6SS clusters (containing the same genes in different order) of Xeu, In contrast, Xpm and Xcc3 only have a single T6SS cluster (cluster III), orthologous to cluster III of Xeu.
To determine if the T6SS is conserved among members of the Xpm pathovar, or is exclusively present in the CIO151 strain, we searched for the T6SS genes in the genomes available for this pathovar32. We performed a BLASTN analyses with each T6SS gene of CIO151 against the other 64 genomes of manihotis pathovar (data not shown). We found the 16 T6SS genes are conserved among the analyzed genomes, with 75% to 100% identity at the nucleotide level. We found that clpV is divided in two parts due to a stop codon in the middle of the two AAA-ATPase domains. Nonetheless, because the stop codon does not truncate any of the protein domains (Additional file: Table S7), the two ClpV fragments may interact to create a functional protein complex (Figure 3 and 4).
The cluster of orthologous groups (COG), together with the results obtained from our phylogenetic results discussed above, suggest Horizontal Gene Transfer (HGT) events as the probable origin of the T6SS in Xanthomonas. Thus we searched for genomic islands (GI), insertion sequences (IS), deviations in the GC content, and tRNA genes surrounding the T6SS clusters12. The positions and types of identified GI and IS were not exactly the same in clusters I and clusters III of the three organisms (Xpm, Xeu and Xcc3, Figure 2 and additional file: Table S8). However, GIs ere consistently located between the vgrG and clpV genes. The IS within the T6SS clusters of Xeu and Xpm belong to the IS3 family, while the IS in the Xcc3 cluster is part of the IS4 family31 (Figure 2 and additional file: Table S8). In contrast, no significant difference in GC content (with respect to the average GC content of the corresponding genomes) was detected int the T6SS cluster of Xeu, Xcc3, and Xpm. Hence, there are some characteristics of HGT for these clusters, but if these events occurred, there has been enough time to adapt these clusters to the rest of the genome.
Mutations in vgrG, clpV and hcp decrease the virulence of Xpm on susceptible cassava plants
To determine whether the T6SS contributes to the virulence of Xpm, we constructed and then inoculated vgrG, clpV, icmF and hcp single deletion mutants into susceptible cassava plants. Initially, in vitro growth assays were performed for each mutant strain, to determine whether the introduced mutations decrease the general fitness of the pathogen (Figure 3). The mutants were complemented with the wild type genes in all cases to restore the genotype. The resulting vgrG, clpV and hcp mutant strains did not show significant differences in fitness with respect to the wild type control (paired t-test p-values=0.14, 0.48 and 0.15, respectively). icmF mutants showed a small decrease in growth in vitro at 18 hours with respect to the wild type strain (paired t-test p-value=0.02) but the effect was undetectable at later time points.
These results suggest that while mutations in vgrG, clpV and hcp do not affect the ability of Xpm to grow in vitro, mutation in icmF has a slight effect on the growth of this pathogen.
We then tested the mutants for their ability to cause disease in susceptible cassava plants. The vgrG, clpV and hcp mutant strains were able to produce symptoms on leaves of susceptible cassava. Therefore, we conclude that these genes are not required for full pathogenicity (Figure 4). However, a decrease in symptoms was observed for CIO151ΔvgrG, CIO151ΔclpV, and CIO151Δhcp strains at 15 days post-inoculation by two different methods of inoculation (Fig. 4a and Fig. 4b). For the hole-inoculation method, the lesion area was measured with the program ImageJ46 and statistically significant differences when compared to the wild type were observed (ANOVA p-value>0.01), as show in Figure 4c. The phenotype was complemented when mutants were transformed back with their respective wild type gene (Fig. 4c). Together, these results suggest that the genes vgrG, clpV and hcp are required for maximal aggressiveness of Xpm on susceptible cassava plants.
clpV deletion decreases Xpm motility
The T6SS has been implicated in pathogenicity15,52,53, motility54–56 and interaction with other bacteria21,57. Motility contributes to virulence in the genus Xanthomonas58–60. We therefore tested the T6SS mutants for swimming motility on petri dish with a low proportion of agar (0,3% agar) after 24 h and 48 h. Motility was significantly different after 48 for the CIO151ΔclpV mutant with respect to the (p-value=0.0007). Notably, the motility of this strain was indistinguishable from that of the CIO151 strain, after transformation with the wild type clpV gene (Fig. 5c), which demonstrates that the observed differences was due to the truncation/deletion of this gene. For the other evaluated mutants, there were no differences in motility with respect to the wild type (p-value>0.05; Figure 5a, 5b and 5d). In general, these results suggest that ClpV activity impacts motility, whereas VgrG, IcmF and Hcp are dispensable for this function in Xpm.