Introduction

Bacterial wilt of forage grasses, caused by Xanthomonas translucens pv. graminis (Xtg), is one of the main diseases of forage grasses in temperate grasslands [1]. Xtg is a member of the recently-defined clade Xt-III of X. translucens, which also contains four other forage grass-infecting pathovars (pv. arrhenatheri, phlei, phleipratensis and poae) and is distinguished from clades Xt-I and Xt-II, which contain mainly cereal-infecting pathovars such as pv. translucens and cerealis, respectively [2, 3]. While most X. translucens pathovars are generally restricted to a small host range, Xtg is able to infect many grass genera, including Agrostis, Alopecurus, Dactylis, Deschampsia, Festuca, Lolium, Phalaris, Phleum, Poa, and Trisetum [1, 4]. It is also the most widespread, being found in most of Europe, America, and New Zealand [4,5,6].

In contrast to cereal crops such as wheat, where X. translucens is considered a seedborne pathogen, no seed or soil transmission of Xtg has been observed on forage grasses so far [7, 8]. Here, the bacteria are mainly spread by contaminated mowing tools and penetrate via plant wounds. While some X. translucens pathovars such as pv. undulosa are restricted to the non-vascular tissue, all forage grass infecting X. translucens pathovars spread in the xylem, resulting in a disruption of the flow of water and nutrients [7, 9]. The disease affects mainly Italian ryegrass (Lolium multiflorum Lam.), which is cut several times during the growing season for harvesting. After penetration, the bacteria spread in the xylem, causing a disruption of water and nutrient absorption. The symptoms manifest as curling and wilting of the leaves and tillers, and eventually the death of the plant [9, 10]. Infection by the pathogen can lead to serious yield losses, estimated to reach up to 40% under unfavourable conditions [11]. To control the disease in an efficient manner, it is essential to understand the mechanisms involved in the pathogenicity of Xtg.

In plant pathogenic bacteria, successful infection of the plant host generally involves virulence factors that allow the bacteria to adhere to the plant surface, invade the host tissue, acquire nutrients, and suppress plant defence mechanisms. In xanthomonads, some of the most important virulence factors are effectors, in particular type III effectors (T3E), that are secreted into the plant host cells by the type III secretion system (T3SS). These T3E play a major role in pathogenicity by specifically targeting different pathways in the host cell. A total of 53 classes of T3E have been defined, generally referred to as “Xanthomonas outer proteins” (Xop) [12]. Other T3E can cause a hypersensitive response in their plant host and are referred to by this characteristic, like AvrBs1 and AvrBs2. Additionally, transcription activator-like effectors (TALEs) are specific T3E that consist of repetitive sequences that allow them to bind to specific nucleotide sequences in the host genome and manipulate the expression of specific genes [13, 14]. However, while TALEs are characteristic to the Xanthomonas genus, no TALEs have ever been identified in Xtg [3, 15].

Other secretion systems also play an important role in the infection of Xanthomonas spp., including the type II, IV and VI secretion systems (T2SS, T4SS and T6SS, respectively). The T2SS is mostly involved in the secretion of plant cell wall degradative enzymes in the apoplast, allowing the bacteria to invade the plant host, as well as acquire the resulting degradation products as a source of nutrients [16]. Two types of T2SS can be found in Xanthomonas spp., the xps T2SS, which is conserved across the genus, and the xcs T2SS, which is only found in some species [17]. To date, no xcs T2SS was found in X. translucens [3].

The T4SS and T6SS have a similar function and are involved in defence against microbial predators and competition with other microorganisms [18, 19]. Three subtypes of the T6SS have been found in Xanthomonas spp., subtypes 1, 3 and 4, and subtype 3 is further divided into subgroups 3*, 3** and 3***. Previously, no T4SS has been identified in Xtg, and T6SS genes were identified only in strains Xtg2, Xtg9, Xtg10 and NCPPB 3709 [3, 20].

To reliably identify virulence factors in plant pathogenic bacteria, complete and high-quality genome sequences are essential, as they provide an accurate representation of the genes present in each strain. Such genome sequences enable comparative genomics, which can reveal conserved or divergent genes of a pathovar and help define genes that are crucial to pathogenicity. Previously, draft genome sequences of Xtg strains CFBP 2053, ICMP 6431, NCPPB 3709, Xtg2, Xtg9 and Xtg29 allowed a first comparative analysis, which revealed specific features of the pathovar, such as the lack of a flagellum and a distinct type IV pilus [20]. Hypothetical proteins specific to Xtg were also identified, including genes with predicted functions in nutrient acquisition, regulatory mechanisms, virulence, adhesion and motility. Effector proteins that were specific to Xtg included a YopT-like cysteine protease and a XopJ class effector protein, which were tested for their role in virulence. However, single and double knock-out mutants for these genes showed no significant difference in virulence with the wild type [20].

At that time, genomic resources for X. translucens pathovars consisted of highly fragmented and incomplete draft genome sequences based on short read sequencing, which might have prevented the identification of important genes. Furthermore, these genome assemblies fail to assemble complex repetitive sequences such as TALEs, which may have hindered their identification. Long read sequencing technologies now allow to assemble complete bacterial genome sequences, allowing for a comprehensive overview of the gene content within these genomes, including TALEs [21]. Recently, high-quality, complete genome sequences were produced for all pathotype strains of X. translucens, which allowed for a more comprehensive comparative analysis [3]. This revealed that the Xtg pathotype strain LMG 726 lacked many virulence features, such as a T4SS and T6SS, and had the smallest set of T3E. However, the inclusion of only one strain of Xtg did not allow for an in-depth comparative analysis. In this study, we aimed at producing high-quality complete genome sequences for four additional strains of Xtg, perform a comprehensive comparative genomic analysis of the pathovar, and identify unique features that could play a role in its virulence and host specificity.

Materials and methods

Bacterial strains, growth conditions and DNA extraction

The Xtg strains NCPPB 3709, Xtg2, Xtg9 and Xtg29 were selected to be sequenced in this study (Table 1). These strains were grown at 28 °C on YDC agar medium (2% dextrose, 1% yeast extract, 2% CaCO3, 1.5% agar) for 48 h. Bacteria were then dissolved in 10 mL washing buffer (50 mM TRIS-HCl pH 8.0, 50 mM EDTA pH 8.0, 150 mM NaCl). Genomic DNA was then extracted with the NucleoSpin® Microbial DNA kit (Macherey Nagel, Duren, Germany), according to the manufacturer’s recommendations.

Sequencing, genome assembly and annotation

Library preparation and DNA sequencing was performed at the Functional Genomics Center Zurich. Libraries were prepared and multiplexed with the PacBio SMRTbell® Express Template Prep Kit 2.0 (PacBio, Menlo Park, CA, USA) according to the published protocol [22]. After shearing and size selection to 7–10 kb, 1 µL of the non-size-selected library was spiked in the final solution, for a final volume of 10 µL, to allow for sequencing of potential plasmids. Libraries were then sequenced on the PacBio Sequel II platform.

Genomic sequences were de novo assembled with flye 2.7 using the “--plasmids --iterations 2” parameters [23]. Since the assembly of strain NCPPB 3709 did not result in a complete circular chromosome, we used canu with default parameters for this strain [24]. The assembled circular chromosomes were rotated to start with the dnaA gene sequence for comparability. Genomes were functionally annotated with Prokka 1.14.6 with default parameters [25].

Phylogeny and comparative analysis

To determine phylogeny within clade Xt-III, we used all available complete genome sequences of the clade, as well as the genome sequences of pv. translucens strain DSM 18974 and pv. cerealis strain CFBP 2541 as representatives of clades Xt-I and Xt-II, respectively (Table 1). Average nucleotide identity (ANI) and alignment fraction (AF) were calculated using the anib subcommand of pyani 0.2.12 with default parameters [26], and a phylogeny dendrogram was constructed using Ward’s hierarchical clustering method in R version 4.2.0 [27]. To identify genomic rearrangements and conserved genomic regions within Xtg, the genomic structure of the five Xtg genome sequences were compared with Mauve v20150226 [28].

For a comparative analysis of the gene content of Xtg compared to other X. translucens strains, all currently available complete genome sequences of the X. translucens species were retrieved and annotated with Prokka (Table 1, Table S1). The gene content within the X. translucens species was compared with Roary 3.13.0 using default parameters [29]. This allowed to predict singletons, genes that are exclusive to Xtg, by selecting genes that were only found in Xtg strains in our dataset. The pangenome and the predicted singleton contents of Xtg strains were visualized in Venn diagrams with jvenn [30]. Clusters of orthologous groups (COG)-based functional characterization of the gene content of Xtg was performed using eggNOG-mapper v2 [31, 32]. As the COG analysis revealed a high number of transposable elements in Xtg, these were further characterized using the mobileOG-db tool in the Proksee online software, with default parameters [33, 34]. Additionally, putative horizontal transfer events were identified using the Alien Hunter tool in Proksee [35]. The presence of signal peptides or non-classical secretion were predicted using SignalP 6.0 and SecretomeP 2.0, respectively [36, 37].

Minor pilins identified in the singleton analysis were investigated with InterProScan by identifying the characteristic pilin N-terminal transmembrane domain, as well as with the PilFind online tool [38, 39]. To compare the two filamentous hemagglutinin genes identified with similar previously-identified genes, the nucleotide sequence of these genes in the Xtg29 draft genome sequence in which they were identified was extracted [20]. These sequences were then used as query in a BLASTn analysis using the nucleotide sequences of the genes identified in this study or the whole genome sequences of Xtg strains as subject. The DNA sequences of these gene clusters were compared with Clinker v0.0.27 [40].

Presence of T2SS, T3SS, T4SS and type III effectors was determined as detailed by Goettelmann et al. [3]. Presence of a T6SS was determined using SecReT6 3.0 with default settings [41]. The presence of TALEs was determined using AnnoTALE 1.5 [42].

Results

Features of the Xtg genome sequences

Each strain was assembled to one circular complete chromosome. The chromosome sequences were of comparable size, ranging from 4,673,077 bp for Xtg29 to 4,734,182 bp for Xtg9. The sequence coverage ranged from 64- to 89-fold for all the sequenced strains (Table 1). Additionally, a circular contig sequence of 44,971 bp was assembled in strain NCPPB 3709. A comparison by BLASTn to the NCBI non-redundant nucleotide sequences database showed homology to plasmid sequences. This contig was thus considered to represent a plasmid of this strain.

Table 1 Xanthomonas translucens strains used for comparative genomic analysis and characteristics of the genome assemblies. The pv. translucens and pv. cerealis pathotype strains DSM 18974 and CFBP 2541 were included as representatives of clades Xt-I and Xt-II, respectively
Full size table
Fig. 1
figure 1

Average nucleotide identity (ANI)-based phylogeny of Xanthomonas translucens clade Xt-III, including all complete genome sequences currently available for the clade, as well as DSM 18974 and CFBP 2541 as representatives of clades Xt-I and Xt-II, respectively. Phylogeny was constructed with Ward’s hierarchical clustering method. ANI is depicted as a gradient from blue (< 96%) to white (98%) to red (100%). Distance in the dendrogram represents dissimilarity between nodes. Orange: clade Xt-I, blue: clade Xt-II, pink: clade Xt-III

Full size image
Fig. 2
figure 2

Pairwise comparisons of the chromosome structure within Xanthomonas translucens pv. graminis using all five currently available complete genome sequences of the pathovar. The LMG 726 genome sequence at the top was used as a reference. Colours represent locally colinear blocks (LCB), regions that show no rearrangement across all the compared genome sequences. Synteny is displayed by a line that links orthologous LCBs. LCBs found on the bottom part represent regions that are found in reverse orientation compared to the reference

Full size image

Xtg displays a high genomic plasticity

Phylogeny based on ANI revealed that all strains of Xtg were very similar, with 99.9–100% ANI (Fig. 1). The alignment fraction was > 95% between Xtg strains, further supporting their high similarity (Figure S1). Comparison of the genomic structure within Xtg showed 65 to 124 locally collinear blocks (LCB), genomic regions that show no rearrangements between the compared strains (Fig. 2). This showed that although Xtg strains are very close genetically, a high level of rearrangements can be seen within the pathovar.

Fig. 3
figure 3

 A) Pangenome (i.e., total gene content) of the five Xanthomonas translucens pv. graminis strains investigated, and B) content in predicted singletons (i.e., genes that were exclusive to X.t. pv. graminis strains). Numbers indicate the number of genes shared by specific sets of strains. Bottom bars indicate the number of genes that were shared by five, four, three or two strains or that were specific to one strain

Full size image
Table 2 Gene content of the Xanthomonas translucens strains compared in this study
Full size table

The number of genes found in each strain of Xtg was similar, ranging from 4,408 in Xtg9 to 4,445 in NCPPB 3709 (Table 2). The pangenome, i.e., the total number of genes, of the five Xtg strains used in this study consisted of 4,931 genes, while the core genome, i.e., the set of genes shared by all strains, consisted of 3,739 genes (Fig. 3A). Moreover, a total of 1,409 predicted singletons, genes that were not found in any other pathovar of the X. translucens species, were identified, with 536 being shared by all Xtg strains (Fig. 3B). A total of 433 predicted singletons were found to be strain-specific, while 131 were shared by two strains, 164 were shared by three strains, and 145 were shared by four strains. The total number of predicted singletons per strain ranged from 856 in Xtg9 to 922 in Xtg29 (Table 2).

Fig. 4
figure 4

Clusters of orthologous groups (COG)-based functional characterization of A) the total gene content of the five Xanthomonas translucens pv. graminis strains investigated, and B) predicted singletons, genes that were exclusive to X. t. pv. graminis strains

Full size image

The functional characterization based on COG (Fig. 4) showed that many genes within Xtg could not be functionally annotated (13–15%) or were of unknown function (15–16%). The largest category with a known function was replication, recombination, and repair, with 733 to 781 genes of this category per strain (16–18%). Furthermore, this category was also largely represented in predicted singletons (48–53%), ranging from 424 genes of this category in NCPPB 3709 to 453 in Xtg9. Investigation of these genes revealed that most of them were annotated as transposases, suggesting a high transposase activity in Xtg. Further investigation of mobile genetic elements in the Xtg genome sequences showed an overrepresentation of integration/excision elements, corresponding with the previously identified transposases, with 413 to 457 of such elements found per strain, compared to 24 to 123 in other strains of the species (Table 2). The number of other types of genetic mobile elements was similar between Xtg and other pathovars. Additionally, Xtg strains displayed a higher number of putative horizontal gene transfer events, with 99 to 124 predicted putative events, compared to 49 to 62 in the rest of the species (Table 2).

Fig. 5
figure 5

Presence of the xps and xcs type II secretion systems (T2SS), type III secretion system (T3SS), type IV secretion system (T4SS), subtype 3*** type VI secretion system (T6SS-i3***), and subtype 4 type VI secretion system (T6SS-i4) gene clusters, as well as putative type III effectors in the genome sequences of Xanthomonas translucens strains. Green circles indicate presence and white circles indicate absence. Numbers denote the number of putative effectors of each family

Full size image

Unique features of Xtg

Investigation of secretion systems showed that all Xtg strains have the xps T2SS and the T3SS but not the xcs T2SS and T4SS (Fig. 5). Furthermore, a T6SS-i4 was found in NCPPB 3709, Xtg2 and Xtg9, but no T6SS could be found in LMG 726 and Xtg29. The type III effector (T3E) repertoire of Xtg was comparable among the strains of the pathovar, though only one effector of the XopP family was found in LMG 726 and Xtg29, compared to two in the other strains, and a XopL family effector was identified in LMG 726 only. When comparing to other pathovars, differences could be seen in the number of effectors of some families, such as AvrBs2 and XopF effectors, with only one found in Xtg, compared to two in most other pathovars, though only one XopF effector was found in LMG 730 and LMG 843 as well (Fig. 5). No XopG effector could be identified in Xtg, while all other pathovars possessed one. Moreover, aside from one gene found in LMG 726, no XopL effector was found in the other four Xtg strains, when other pathovars possessed one to three effectors of this family. While all other pathovars of clade Xt-III possessed exactly two XopX effectors, three could be found in Xtg, as was the case for pv. translucens (Xt-I) and pv. cerealis (Xt-II). Finally, three genes of the XopE family were identified in all Xtg strains, while only two were found in CFBP 2541, one in LMG 730, LMG 843 as well as DSM 18974, and none were found in LMG 728 and LMG 727. Overall, the Xtg strains displayed the smallest T3E set, ranging from 20 to 21 per strain, compared to 24 to 26 in other pathovars of clade Xt-III and 31 in pv. translucens and pv. cerealis. No TAL effector could be identified in any Xtg strain.

Among the 856 to 922 predicted singletons observed in Xtg strains, 627 to 647 were functionally annotated (Table 2). After excluding transposases, 181 to 222 annotated predicted singletons could be identified, of which 148 were shared by all strains (Table S2). Among these predicted singletons, genes that could have a role in the pathogenicity of Xtg were further investigated. As secreted or cell surface proteins are often involved in interaction with other organisms, we focused on genes that harbored a predicted secretion signal or a putative non-classical secretion. Among these, genes encoding for a filamentous hemagglutinin family outer membrane protein, a YadA family autotransporter adhesin, ten TonB-dependent receptors (TBDRs), and 11 plant cell wall degradative enzymes, including two subtilases, two esterases, a serine hydrolase, a glycerophosphodiester phosphodiesterase, a pectate lyase, a phospholipase, an alpha-galactosidase, a beta-galactosidase, and an aminopeptidase, were found. Seven additional TBDRs that were not shared by all strains were also found. Of the 17 identified TBDRs, five were annotated as diverse iron-siderophore receptors, five as oar-like proteins, three as vitamin B12 transporters, and one as a colicin receptor. However, while TBDRs were overrepresented in Xtg, ranging from 74 to 77 per strain, compared to 51 to 67 in other pathotype strains of the species (Table 2), many genes with similar annotations were also found among the TBDR genes that were conserved between pathovars, suggesting a functional redundancy of these TBDRs.

Additionally, a second gene predicted as encoding a filamentous hemagglutinin gene was found neighbouring the first one, being shared by all strains except NCPPB 3709. Further investigation of the locus containing the two genes revealed the presence of transposases in between them, suggesting it may be a single gene that was truncated by the insertion of these transposable elements. A BLASTn comparison with similar partial filamentous hemagglutinin genes that were previously identified in a draft genome sequence of Xtg29 revealed they corresponded to the same locus. One of the previously identified genes corresponded to the 5’-most gene that was identified in the singleton analysis, while the second gene corresponded to an additional smaller gene downstream of the two other genes we identified, which was present in all strains except NCPPB 3709 (Fig. 6). A BLASTx comparison of the hypothetical full-length gene, consisting of the combined partial sequences, revealed a high identity to other filamentous hemagglutinin genes from Xanthomonas spp., further confirming this locus corresponds to a single gene that was truncated.

Fig. 6
figure 6

Pairwise comparisons of the filamentous hemagglutinin locus within Xanthomonas translucens pv. graminis. Colors denote similar genes. Numbers below strain names refer to the position of the gene cluster in this genome sequence

Full size image

Furthermore, a cluster containing five minor pilins of the type IV pilus (T4P) fimT, pilV, pilW, pilX, and pilE, as well as the anti-retraction factor pilY1 was also found. This cluster is also present in other X. translucens pathovars, but the cluster found in Xtg is highly divergent from the one found in other pathovars, and these genes were thus predicted as singletons by the roary pipeline (Fig. 7A). An in-depth analysis of the minor pilins revealed that the pilX gene lacked the characteristic N-terminal methylation domain and class III signal peptidase cleavage site in any X. translucens strains investigated, suggesting it may not be functional. In contrast, these domains were present in pilE, pilV and fimT for all strains. When investigating pilW, the gene was predicted as encoding a putative pilin only in Xtg, pv. phleipratensis strain LMG 843 and pv. poae strain LMG 728 (Fig. 7B). However, the conserved N-terminal phenylalanine residue was present only in LMG 843 and was replaced by a leucine in LMG 728 and all Xtg strains.

Fig. 7
figure 7

 A) Pairwise comparisons of the minor pilins gene cluster between strains of Xanthomonas translucens (i.e. X.t. pv. translucens DSM 18974, pv. cerealis CFBP 2541, pv. phlei LMG 730, pv. phleipratensis LMG 843, pv. arrhenatheri LMG 727, pv. poae LMG 728, and pv. graminis LMG 726, Xtg2, Xtg9, Xtg29 and NCPPB 3709). Colors denote similar genes. The pilY1 gene is represented in different colours in pv. graminis strains compared to other strains as its sequence was too divergent and they were considered as different genes by the software. Numbers below strain names refer to the position of the gene cluster in this genome sequence. B) Alignment of the amino acid sequences of pilW. Only the pv. graminis strain LMG 726 is represented as the sequences were identical in all five pv. graminis strains. Colours represent amino acids that were conserved among at least six out of seven strains (red) or amino acids with similar properties that were found in at least six out of seven strains (green). Highlighted in blue is the class III signal peptidase cleavage site that is characteristic to pilins

Full size image

Discussion

Comparative analysis of five Xtg strains revealed a very high genome plasticity within the pathovar compared to all other pathovars of X. translucens. While the strains showed 99.9 to 100% nucleotide identity, ANI compares conserved sequences, thus the genome sequences may still differ in lost or acquired sequences, as well as in genome structure. Indeed, the analysis of conserved genomic regions showed a high level of rearrangement between the five strains with 65 to 125 LCBs. Additionally, regions without any link to a similar block can be seen, which may be due to the presence of duplicated sequences, which are not considered by the MAUVE software.

This is the first report of such a high level of rearrangements within Xtg, which could be observed thanks to the generation of complete genome sequences that allowed a clear observation of the genome organization in each strain. In comparison, a similar analysis with data published by Shah et al. [45] and using complete genome sequences of four pv. translucens and eight pv. undulosa strains with ANI values as low as 98.9 and 99.5%, respectively, showed a considerably lower amount of rearrangements, with 53 to 67 and 39 to 67 LCB (38, data not shown). This is in line with a previous comparative analysis within the X. translucens species that showed that clade Xt-III, which contains Xtg, was the clade showing the most rearrangements [3]. Additional genome sequences for other pathovars of clade Xt-III would help to determine whether this high level of plasticity is exclusive to Xtg or whether it is a characteristic of the entire clade.

The singleton analysis revealed 1,409 genes that were exclusive to the Xtg strains, with 536 of them being found in all five strains. Functional characterization revealed that most of these predicted singletons were annotated as transposases, and further analysis of mobile genetic elements also revealed a high amount of insertion/excision elements in the genomes of Xtg strains, corresponding to the identified transposases. Such a high number of transposable elements in Xtg could explain the high genomic plasticity that was observed, as they are genetic elements that are mobile within the genome and can facilitate gene duplications, chromosomal recombinations and horizontal transfer [46, 47]. A high genome plasticity has previously been found in many Xanthomonas spp., and was suggested to play an important role in host adaptation [48]. A comparably large number of insertion sequences was reported in X. oryzae pv. oryzae, where these numerous mobile elements were suggested to play a role in race differentiation [49]. Insertion elements can also cause the loss of gene function, and virulence can emerge from the inactivation of avirulence genes [50, 51]. Furthermore, transposable elements have been suggested to play a role in the movement of “passenger genes” in Xanthomonas spp., including virulence genes, contributing to their diversification and spread [52, 53]. As such, mobile genetic elements can play a crucial role in pathogen adaptation, and the large amount of such elements in Xtg compared to other pathovars of the species could, at least partially, explain its high virulence and broad host range. Investigating all the identified transposable elements, their site of insertion, and the genes flanking them would be crucial in understanding their role in pathogenicity and adaptation of Xtg and should be tackled in the future.

The total number of genes found in each strain of the pathovar was comparable, confirming the high number of genes already identified in the Xtg pathotype strain LMG 726. When compared with other pathotype strains, LMG 726 exhibited the highest number of genes of the species, as well as the highest number of genes that were exclusive to the strain [3]. While in this previous comparative analysis, as much as 950 genes were found to be exclusive to LMG 726, this number was influenced by the strains included in the comparison. The finding of 1,409 genes exclusive to Xtg, with 536 shared by all strains, using all available complete genome sequences of X. translucens strains is most likely a better estimation of the number of genes specific to Xtg. However, this unusually high amount seems to mainly be explained by the abundance of transposases in the pathovar. Nonetheless, many predicted singletons that could play a role in the pathogenicity and host range of Xtg have been uncovered.

First, a cluster containing five minor pilins and the anti-retraction factor pilY1 was found to greatly differ between Xtg and other pathovars of the species. These pilins are components of the T4P, which mediates surface adhesion, biofilm formation and twitching motility [54, 55]. The T4P consists of a filamentous polymer of the major pilin PilA, and minor pilins may be incorporated into the pilus. Further investigation of minor pilins of this cluster revealed that only pilE, pilV and fimT may be functional in all pathovars of the species. For pilW, the N-terminal cleavage and methylation domain that is characteristic to pilins and required for their processing was present only in pvs. graminis, poae, and phleipratensis, though the conserved phenylalanine that undergoes methylation after cleavage was only present in pv. phleipratensis and was substituted by a leucine in the other two pathovars. However, it was previously shown that although the phenylalanine residue is highly conserved in pilins, its substitution by another amino acid did not affect secretion, cleavage and methylation of the pilin and subsequent T4P assembly [56]. It is thus plausible that pilW is still functional in pvs. graminis and poae. The six identified genes, as well as the major pilin PilA were previously shown to differ significantly in Xtg29 compared to strains of other X. translucens pathovars, while the genes encoding the main components of the T4P machinery were highly conserved [20]. That this divergent cluster is conserved within Xtg could indicate a role in the virulence of the pathovar. Indeed, the T4P is an extracellular feature of the bacteria that may be detected by plant defense mechanisms. The high sequence variation in the minor pilins, as well as the major pilin PilA may allow Xtg to avoid plant defenses while mediating surface adhesion and biofilm formation.

Among other predicted singleton genes, two filamentous hemagglutinin genes were identified. However, further investigation revealed that they likely correspond to a single gene in which multiple transposases were inserted. Moreover, these genes corresponded to the two partial filamentous hemagglutinin genes that were previously reported as predicted singletons of Xtg [20]. The N-terminus of filamentous hemagglutinins possesses a signal peptide, as well as a two partner system domain that are involved in their secretion to the periplasm and the apoplasm, respectively [57, 58]. Whether the gene encoding for the N-terminal part of the protein is still functional remains to be determined, though it is unlikely, as essential domains may be found in the C-terminal part of the gene. In X. axonopodis pv. citri, filamentous hemagglutinins were shown to be crucial to virulence by promoting plant surface adhesion and biofilm formation but acted as elicitors of plant defense in X. campestris pv. vesicatoria [59, 60]. Thus, the loss or truncation of this gene in Xtg may help it evade recognition by the plant host.

Additionally, 17 predicted singleton genes coding for TBDRs were found, with ten being shared by all Xtg strains. TBDRs are known for their role in iron and vitamin B12 uptake and are mostly regulated by the ferric uptake regulator gene [61]. Mutants of this regulator were unable to induce disease symptoms in X. o. pv. oryzae which was hypothesized to be due to an inability to cope with the oxidative stress conditions encountered during the infection, emphasizing the role of TBDRs in virulence [62]. Moreover, in X. campestris pv. campestris, some TBDRs were found to be activated by HrpG and HrpX, regulatory proteins of the type III secretion system, and the SuxA TBDR was found to play a major role in pathogenicity by acting on sucrose import [63]. A comparative analysis across 226 eubacterial genome sequences showed that TBDRs were overrepresented in Xanthomonas spp. compared to other bacteria, ranging from 36 TBDRs in X. oryzae pv. oryzae to 68 in X. axonopodis pv. citri [63]. In this study, similar numbers of TBDRs were found in all strains of X. translucens, with an overrepresentation in Xtg. The identified predicted singletons were mainly annotated with functions related to iron and vitamin B12 uptake, and similar genes were found to be shared across the species. It is thus unclear if these 17 predicted singletons serve a specific function in the pathogenicity of Xtg, or whether they are functionally redundant with other TBDRs.

A total of 11 plant cell wall degradative enzymes were also found in singletons that were shared by all Xtg strains and were predicted to be secreted. Such enzymes are often secreted by the T2SS and involved in the degradation of the plant cell wall [16, 18]. These include cellulases, xylanases, pectate lyases, and polygalacturonases, which degrade the main polysaccharide cell wall constituents, as well as various proteases and lipases. X. oryzae pv. oryzae mutants lacking a T2SS were shown to be virulence deficient and multiple enzymes secreted by the T2SS were shown to be necessary for virulence [64,65,66]. As Xtg mutants lacking a T3SS showed a drastic reduction in symptoms, but still survived in the plant, it was hypothesized that other virulence factors may be important in Xtg [15]. Therefore, the T2SS and the enzymes it secretes may play a crucial role in the in planta growth of Xtg.

Finally, a gene encoding for a YadA family autotransporter adhesin was predicted as a singleton shared by all Xtg strains. Such adhesins showing a high similarity to YadA from Yersinia spp. have been shown to mediate cell to cell aggregation, biofilm formation, and adhesion to the host cell surface [67, 68]. In Xanthomonas oryzae pv. oryzae, mutations in the genes encoding for the YadA-like adhesins XadA and XadB were shown to result in reduced surface attachment and virulence after surface inoculation [69, 70]. In Xtg, this adhesin may therefore play a similar role in attachment to the host.

An in-depth analysis of the secretion systems within Xtg showed that a T6SS-i4 was present in strains Xtg2, Xtg9 and NCPPB 3709 and not in LMG 726 and Xtg29, confirming previous reports investigating these strains [20]. It was previously hypothesized that the T4SS and T6SS could act as elicitors of plant defense, and that their absence in LMG 726 could help it evade plant defenses [3]. However, the presence of a T6SS in Xtg2, Xtg9 and NCPPB 3709 may disprove this hypothesis. Nonetheless, its absence in Xtg29 and LMG 726 indicates that it is not crucial to the pathogen’s survival and virulence. Conversely, the xps T2SS and the T3SS were found to be conserved, suggesting they may be crucial to the pathovar.

The T3E repertoire within the pathovar was largely conserved, except for only one XopP family effector being present in strains Xtg2, Xtg9 and NCPPB 3709, compared to two in LMG 726 and Xtg29, as well as one XopL family effector being present only in the LMG 726 strain. This XopL effector could have been conserved in LMG 726, or it might have been recently acquired through horizontal transfer. Nonetheless, as it is unique to LMG 726, it is likely not crucial to the lifecycle of Xtg. Rather, the absence of XopL effectors may have been beneficial to other strains of Xtg, as they all lacked such effectors, while all other pathovars of the species possessed at least one and up to three. Similarly, several effectors appear to have been lost in Xtg, such as AvrBs2, XopF, or XopG, as the pathovar possesses fewer of these effectors than other pathovars. Furthermore, a comparable loss of a XopP effector may have occurred in strains LMG 726 and Xtg29. Interestingly, Xtg possesses the smallest set of T3E in the species, which may be explained by such a loss of effectors during evolution. Finally, no TALEs were identified, further confirming the absence of such effectors in the pathovar [3, 15]. This suggests that these T3E were either unnecessary or even deleterious to the pathovar. As these effectors could trigger plant defence, their loss may have contributed to the success of Xtg as a pathogen of many grass species. On the other hand, three effectors of the XopE family were present in Xtg, while only two, one, or none were found in other pathovars. Additionally, three XopX effectors were found in Xtg, while other pathovars of clade Xt-III possessed only two. XopE effectors belong to the HopX family, which contains transglutaminases and possess a catalytic site that may be involved in proteolysis [71]. The XopE2 effector was shown to suppress hypersensitive response in Nicotiana spp. and to inhibit pattern-triggered immunity [72, 73]. XopX effectors were found to be required for virulence in X. euvesicatoria on tomato and pepper by interfering with plant immune responses [74, 75]. This suggests that XopE and XopX effectors may play a role in the pathogenicity of Xtg by suppressing plant defences.

Phylogeny based on ANI showed that all five available Xtg strains were highly identical, with 99.9 to 100% ANI. A previous analysis of the genetic diversity of Xtg using AFLP markers also showed that multiple strains from Switzerland, France and Belgium displayed a very low diversity [43]. In other pathovars of the species, a higher diversity was observed, with only 99% ANI within pv undulosa and pv. translucens [45]. Although these strains were from a more diverse origin than in Xtg, with strains from North and South America as well as Iran, there was still a high diversity when comparing strains of a similar origin. This further highlights the particularity of the low diversity observed within Xtg. Still, as aside from one Norwegian strain, all Xtg strains compared in this study originate from Switzerland, obtaining additional complete genome sequences from strains of more diverse origins would help to better determine the genomic diversity existing within the pathovar.

Overall, many common features of X. translucens were lacking in Xtg and seem to have been lost over the course of evolution, as they were present in the rest of the species. No T4SS was found in the pathovar and a T6SS was found only in strains NCPPB 3709, Xtg2 and Xtg9. It had the smallest set of T3E and no TALEs were identified. Previous research also showed that although a nearly complete flagellar gene cluster is found in strain NCPPB 3709, such a cluster was absent in all other strains investigated [20]. In this regard, the high mobile genetic element activity observed in Xtg could have played an important role in this loss of virulence features through transposable element insertion within a gene, or by facilitating chromosomal region deletions [50, 51, 76]. Furthermore, a highly divergent T4P was found, both in terms of the amino acid sequence of the pilins, as well as in the presence or absence of functional minor pilins, compared to the rest of the species. This loss or divergence of main virulence features in Xtg may be responsible for its ability to infect a large range of plant hosts, by allowing the bacteria to evade plant defense mechanisms.

Conclusion

Our study provided new high-quality complete genome sequences for X. translucens Xtg, allowing for an in-depth comparative genome analysis within the pathovar. This revealed that Xtg exhibits a remarkable genome plasticity, likely related to an unusually high amount of transposases, which could explain its success as a critical pathogen of a large range of forage grasses. A set of potential virulence factors of the pathovar were identified, and a comprehensive analysis of these genes will contribute to better understand its virulence and host range, providing a basis for the development of new cultivars of forage grasses with increased resistance to the disease.