Comparative genomics reveals the emergence of copper resistance in a non‐pigmented Xanthomonas pathogen of grapevine

Abstract Xanthomonas citri pv. viticola (Xcv) is the causal agent of bacterial canker in grapevine. The pathogen is restricted to India, where it was first reported in the 1970s, and Brazil. In the present study, we report the first complete genome sequence of Xcv LMG965, which is a reference pathotype strain. We also report genome sequences of additional isolates from India and comparative genome‐based studies of isolates from Brazil. Apart from revealing the monophyletic origin of the pathovar, we could also confirm a common frameshift mutation in a gene that is part of the Xanthomonadin pigment biosynthetic gene cluster in all the isolates. The comparative study also revealed multiple intrinsic copper resistance‐related genes in Brazilian isolates, suggesting intense selection, possibly because of heavy and indiscriminate usage of copper as an antimicrobial agent in the orchards. There is also the association of a Tn3‐like transposase in the vicinity of the copper resistance genes, indicating a potential for rapid diversification through horizontal gene transfer events. The findings, along with genomic resources, will allow for systematic genetic and functional studies of Xcv.


INTRODUCTION
Grapevine bacterial canker of grapes (Vitis vinifera L.) is caused by Xanthomonas citri pv.viticola (Xcv) (da Gama et al., 2018).The disease leads to the development of cankers on almost all plant parts, including leaves, berries, petioles and twigs.Apart from Vitis vinifera, the pathogen is also reported to infect Azadirachta indica (neem), Phyllanthus maderaspatensis, cashew plants, Mangifera indica (mango) and plants of Amaranthus and Glycine genus (Chand & Kishun, 1990;da Gama et al., 2018;Nayudu, 1972;Peixoto et al., 2007).The disease was first reported in India in 1972 with disease outbreaks in the 1980s (Chand & Kishun, 1990).The disease was endemic to India until 1998, when Vitis vinifera plants in Brazil were reported to be developing a similar disease, and the concerning bacterial strains were found to be comparable to the Indian pathotype strain, NCPPB 2475 (Trindade et al., 2007).There is also a report of its presence from Thailand in the Asian region, and the disease is not yet reported from the European Union and other grape-growing areas (Health et al., 2021;Midha & Patil, 2014).Recently, the pathogen was assessed by the European Food Safety Authority as a potential quarantine threat (Health et al., 2021).
The formation of yellow-coloured colonies due to the presence of Xanthomonadin pigment is a characteristic feature of the Xanthomonas group of phytopathogens (He et al., 2020).However, few pathogens have acquired mutations in the Xanthomonadin biosynthesis gene cluster, like frameshift mutation, deletion and insertion, to lose their pigmentation (Midha & Patil, 2014).X. citri pv.viticola (Xcv) is one such pathovar that forms white colonies.Previously (Midha & Patil, 2014), a frameshift mutation in a gene encoding phosphotransferase/dehydratase, resulting in a truncated protein was reported to be the reason behind this phenotype.Interestingly, in another pathovar, X. citri pv.mangiferaeindicae (Xcm) of this species, that infects mango, the pigment cluster is deleted (Midha & Patil, 2014).Xanthomonadin pigment is reported to be important for the protection of the bacterium from photodamage but not required for virulence (He et al., 2020).It is possible that after acquiring a frameshift mutation in one of the genes, over time, the pathovar loses the pigment cluster.Hence, there is a need to confirm the mutation status over decades in multiple strains of diverse geographic locations.However, such studies are lacking in Xcv, Xcm, and in other Xanthomonas pathogens that produce white-coloured colonies.At the same time, it is also important to understand why few of the Xanthomonas pathogens are acquiring mutations in the Xanthomonadin gene cluster or pigment genes.
Copper-based bactericides have been frequently used for plant disease management for years.Extensive use of these bactericides has resulted in the resistance to copper compounds in phytopathogens (Behlau et al., 2011;Cooksey, 1990;Xiaojing et al., 2022).Copper sprays are also used for the control and prevention of grapevine bacterial canker (Malavolta Jr et al., 1999;Marques et al., 2009).There has been a constant increase in copper tolerance over the years 1998-2006 in X. citri pv.viticola strains (Marques et al., 2009).In earlier studies, multiple genetic mechanisms of many copper-resistant strains from different Xanthomonas spp.have been reported (Basim et al., 2005;Behlau et al., 2011;Giovanardi et al., 2016;Lee et al., 1994).Three of them are plasmid-borne, while one is chromosomally encoded (Basim et al., 2005;Huang et al., 2021).Huang et al. characterised Xanthomonasassociated copper resistance gene clusters into three groups, however, the genomic determinants of copper resistance/homeostasis in X. citri pv.viticola are not yet reported.Hence, it will be interesting and important to investigate the origin and evolution of copper resistance loci over time in different geographic regions.The complete genome resource, along with genome sequences of multiple strains from diverse regions, will help in the systematic study of pathogenicity and management.

Complete genome features of Xcv LMG965
The circular representation of the chromosome and plasmid of Xcv LMG965 was created using the CGview comparison tool (CCT) (Grant et al., 2012).The coding sequences (CDSs) were categorised into clusters of orthologous groups (COGs) using CCT (Grant et al., 2012).Insertion sequence (IS) elements were predicted using ISsaga 2.0 (Varani et al., 2011).Transcription activator-like effectors (TALEs) were predicted and assigned to TALE classes using the AnnoTALE version 1.5 (Grau et al., 2016).
Xanthomonadin pigment cluster analysis XAC_RS20620 (gene encoding phosphotransferase) from X. citri pv.citri 306 (NC_003919) was used to retrieve homologous sequences from the Xcv strains using standalone tBLASTn.The nucleotide sequences were annotated with the help of ORF Finder to obtain amino acid sequences.Both the amino acid and nucleotide sequences were aligned using MUSCLE inbuilt in MEGA X v10.2.6 (Kumar et al., 2018).

RESULTS
Complete genome sequence of the pathovar reference strain of X. citri pv.viticola The complete genome sequence of the pathovar reference strain first isolated from India is depicted in Figure 1.Hybrid assembly with Illumina short reads and Oxford nanopore reads of LMG965 resulted in a circular chromosome of 5.11 Mb and a plasmid of 16.3 kb (Figure 1).The complete genome of Xcv LMG965 encodes 4270 coding sequences (CDSs) and 53 tRNAs with two rRNA operons.These coding F I G U R E 1 (A) Circular representation of the chromosome of Xanthomonas citri pv.viticola (Xcv) LMG965.The outermost ring depicts coding sequences on the forward strand assigned to clusters of orthologous group (COG) categories.The following two rings show coding sequences and RNAs on forward and reverse strands, respectively.The next ring shows coding sequences on the reverse strand assigned to COG categories.Furthermore, the two innermost rings depict GC content and GC skew (+ and À).The colour coding for all the features is given in the figure.IS elements and transcription activator-like effector (TALE) coding gene locations are labelled on the outermost ring.(B) Circular representation pLMG965-1 plasmid of Xcv LMG965.The two outermost rings depict CDSs on the forward and reverse strands, followed by two innermost rings depicting GC content and GC skew (+ and À).The colour coding for all the features is given in the figure.CDSs are annotated on the outermost ring.
sequences were further classified into COG categories (Figure 1A).The COG categories with maximum representation were amino acid transport and metabolism (E), cell wall/membrane/envelope biogenesis (M), signal transduction mechanisms (T), carbohydrate transport and metabolism (G), Translation, ribosomal structure and biogenesis (J) and replication, recombination and repair (L).A large number of CDSs were assigned to general function prediction only (R) and function unknown (S) categories (Figure 1A).The genome carries 26 IS elements, further classified into IS element families.The maximum number of IS elements were assigned to IS3 ssgr IS51 and IS3 ssgr IS407 IS element families.IS elements were also assigned to ISL3, IS1595 ssgr IS1595, ISNCY, IS1595 ssgr ISNha5, Tn3 and IS5 ssgr IS5 families (Table S1, Figure 1A).
The complete genome also carries one transcriptional activator-like effector (TALE) that was assigned to TalJE class by AnnoTALE version 1.5 (Figure 1A, Table S1) (Grau et al., 2016).TALEs are 33-35 amino acid-long repeats directed to bind host DNA in a sequence-specific manner.They play an imperative role in virulence by manipulating host gene expression.Usually, in the case of X. citri, TALEs are also found to be associated with the plasmids, however, for Xcv LMG965, no TALEs were encoded on the plasmid (da Silva et al., 2002;Rana et al., 2022).The plasmid pLMG965-1 encodes various type IV secretion systemrelated proteins, partition, replication and mobilisation proteins, toxin/antitoxin system and hypothetical proteins (Figure 1B).The pLMG965-1 shares homology with plasmids of X. citri pv.fuscans strain CFBP6166 (NZ_CP021005) and CFBP6975 (NZ_CP021009) with more than 90% identity.

Genome-based phylogeny and taxonomy of Xcv isolates from India and Brazil
Four genomes of X. citri pv.viticola reported previously were retrieved from the NCBI GenBank database for comparative genome analysis (de Farias et al., 2020;Ferreira et al., 2019;Lima et al., 2017).We also generated genome sequences of two additional isolates from India.The genome size of LMG966 and LMG967 is 5.11 and 5.10 Mb, with 58 and 59 contigs, respectively.Their GC content is 64%.Detailed genome assembly statistics and metadata of the strains used in the study are given in Table 1.The core gene phylogeny of Xcv strains with X. citri pathovar representatives revealed that all the Xcv strains were forming a separate clade in close association with X. citri pv.martyniicola LMG9049, X. citri pv.vitiscarnosae LMG939 and X. citri pv.vitistrifoliae LMG940 (Figure 2).X. citri pv.vitiscarnosae LMG939 and X. citri pv.vitistrifoliae LMG940 infect plants belonging to the Vitaceae family T A B L E 1 Genome assembly and metadata statistics of the strains used in the study.S1).

A common frameshift mutation in Xanthomonadin biosynthetic gene cluster
The frameshift mutation in a gene encoding phosphotransferase was reported in the type strain of Xcv, LMG965 (Midha & Patil, 2014).Here, we looked for the above-mentioned frameshift mutation in the rest of the Xcv strains from India and Brazil.All other Xcv harbour the same frameshift mutation, resulting in a truncated protein of 122 amino acids from 143 amino acids which further revealed the deletion of four nucleotide bases from the 348-351 position (Figure 3).The Xanthomonadin gene cluster analysis revealed that the nonpigmented colony phenotype had been maintained over the years in the Xcv strains from different geographical locations.

Genome variation in Indian and Brazilian Xcv isolates
Interestingly, a genome size difference was observed between the Indian and Brazilian strains.Indian strains' genome size is approximately 5.11 Mb, whereas the genome size of Brazilian strains ranges from 5.31 to 5.48 Mb (Table 1).Furthermore, the pangenome analysis revealed the number of core genes for Xcv strains was 4164, 86.4% of the total pangenome size (4820 genes).The number of unique genes was very few per strain, LMG965 (19), LMG966 (4), LMG967 (6), CCRXcv80 (67), CCRXcv17 (7), CCRXcv117 (26) and CFBP7764 (71).Furthermore, the unique gene repertoire of Indian strains was compared with that of Brazilian strains.31 genes were unique to Indian isolates, which encode for type 4 secretion system-related genes, toxin/antitoxin proteins, hypothetical proteins and recombination/replication-related proteins.On close inspection, it was found that these genes are associated with the plasmid pLMG965-1.Hence, genes encoded on the plasmid of LMG965 were present in draft genomes of both the other Indian strains showing that this plasmid is prevalent in Indian isolates and missing or has been lost from the Brazilian population.
On the other hand, Brazilian isolates have 47 unique genes, which are mostly encoding CzcABC heavy metal efflux RND transporters, cation transporter, copper resistance-related genes, transposases and hypothetical proteins.The Brazilian strains carry genes involved in heavy metal resistance and transport that were missing from the Indian isolates.LMG965 and CCRXcv117 were taken as reference genomes to represent the unique gene content of Indian and Brazilian isolates, respectively (Table S1).
The SmmD/CusAB gene cluster, which forms a channel through the periplasm (copper/silver efflux pump) present in S. maltophilia and reported from pLH201.1 plasmid of X. citri pv.citri LH201, was also found in the Brazilian Xcv strains (Crossman et al., 2008;Richard et al., 2017;Routh et al., 2011).Another important cluster, czcABCD involved with the heavy metal (cobalt/zinc/cadmium) efflux associated with the copper-resistant xanthomonads was found in all the Brazilian Xcv strains (Richard et al., 2017) (Figure 4c).These three gene clusters previously described to be encoded on copper-resistant plasmids are situated on the same contigs in the case of CCRXCV17 and CCRXcv117, whereas they were distributed amongst different contigs in the case of CFBP7764 and CCRXcv80.The GC content of NZ_PESX01000012 of Xcv strain CCRXcv117 carrying the heavy metal resistance genes is 60.7%, which is lower than the GC content of the genome (approximately 64%).All the copper resistance and related genes have variable GC content from as low as 58.2%-70.3%.The contig has TnpT, TnpS and TnpA in the vicinity of the czcABCD operon, together they constitute an important part of Tn3-like transposon reported from Xanthomonas oryzae (Niu et al., 2015).Previously (Richard et al., 2017), copper-resistant gene cluster was reported to be associated with Tn3-like transposase situated on plasmids in X. citri pv.citri strains.On comparison with the pLH201.1 from Xcc LH201 carrying the Tn3 transposon-associated copper resistance genes, a particular region of the plasmid, TnpLH201.1 (formerly characterised) shares homology with the NZ_PESX01000012 contig of CCRXcv117.CCRXcv117 shows a relationship with the copper resistance genes associated with the TnpLH201.1 region (Figure 4c).However, arsenic resistance-related genes, as well as a single Ile tRNA gene associated with the TnpLH201.1 that is required for the transcription of the associated genes on the transposon, were absent from CCRXcv117.The conjugative apparatus composed of 16 tra genes reported to be associated with the copper-resistant plasmid, pLH201.1 is absent in the CCRXcv117 (Richard et al., 2017).ICEfinder analysis revealed that region 77,573-139,450 of NZ_PESX01000012 (CCRXcv117) is predicted to be a putative integrative and conjugative element with type 4 secretion system-related genes.This region is in the vicinity of the copper resistance operon of the CCRXcv117 situated at the same contig.
Interestingly, there is another set of copLAB operons reported from some copper-sensitive Xanthomonas strains (Behlau et al., 2011).Apparently, these genes are not involved in copper resistance but rather are associated with copper homeostasis, hence, termed cohLAB (Behlau et al., 2011).Homologous of these genes were found in all the Xcv strains, both from India and Brazil (Figure 4B).Homologues of cohL, cohA and cohB shared 98.3%, 99.5% and 98.6% nucleotide identity, respectively, with the cohLAB operon of X. citri pv.citri 306.The gene encoding cohB carries a frameshift mutation in the CFBP7764.This cohLAB operon was situated on different contigs from the copLABMGF operon in the Xcv strains (Figure 4B).The cohLAB of CCRXcv117 shares very less nucleotide identity with copLAB genes of the same strain T A B L E 2 BLASTn percentage Identity values between copper resistance genes copL, copA, copB, copM, copG, and copF of the Brazilian Xcv strains with previously reported copper resistance operons/extracted from genomes available on NCBI.
The strains belonging to these pathovars are closely related and form monophyletic clades, indicating rapid diversification in a short time (Bansal et al., 2017).The Xcv isolates from India and Brazil were forming a distinct monophyletic clade from the rest of the X. citri pathovars suggesting their distinct course of evolution.
The phylogenetic divergence in the Indian and Brazilian clades is also supported by the increased genome size of the Brazilian isolates when compared to the Indian isolates and the existence of two subclades corresponding to Indian and Brazilian isolates.This can also be attributed to the fact that the Indian isolates are decades old (1969)(1970)(1971)(1972), whereas Brazilian strains were isolated recently (2009)(2010)(2011)(2012).
The increased genome size of Brazilian strains hints at the acquisition of important genomic features that might help in the adaptations and survival in different geographic conditions and new agricultural conditions.The increased genome size of Brazilian strains revealed the acquisition of genomic determinants for copper resistance.As grapes are highly susceptible to fungal and microbial diseases, there is heavy and indiscriminate usage of copper-based antimicrobials, which might have led to such a rich copper resistance repertoire in Brazilian Xcv strains.The copLABMGF operon is so far reported to be associated only with the plasmids in the genus Xanthomonas (Behlau et al., 2011;Huang et al., 2021;Richard et al., 2017).However, in the case of Stentrophomonas strains, it was chromosomally encoded.S. maltophilia is found in water, soil, plants and humans, where it might function as a reservoir and carrier for such accessory genes required for survival.The contig carrying copLABMGF was carrying the same gene repertoire as that of Tnp201.1, a Tn3-like transposase from copper-resistant plasmids in xanthomonads.However, the absence of arsenic resistance-related genes suggests selective pressure to reduce the cost of maintenance of additional or unrelated accessory genes.Furthermore, the tRNA gene associated with the plasmids carrying TnpLH201.1 was also absent from the Xcv strains, similar to the chromosomally integrated transposases (Richard et al., 2017).The 16 tra genes apparatus required for the conjugational transfer of copper-resistant plasmid, pLH201.1 was also absent in the Xcv strains (Richard et al., 2017).All these factors, along with the observation of an ICE element in the vicinity of copLABMGF operon and other copper-resistance-related genes, hint at the possible integration of copper-resistant genes in the genome of the Brazilian Xcv strains.
Till now, Tnp201.1 has been majorly found to be associated with the plasmids in xanthomonads (Richard et al., 2017).However, it was also found to be integrated with the genome of Stenotrophomonas sp.LM091, which was isolated from the citrus plant (Richard et al., 2017).The high similarity of copper resistance clusters of Xcv strains with plasmid-borne copper resistance systems, the presence of Tn3-like transposase, and variable GC content hints at the acquisition of copper resistance genes from other microbes.They could be either integrated in the chromosome or might belong to a separate plasmid unknown until now.Since the analysis was based on the draft genome assemblies, which fail to capture mobile genetic elements such as plasmids, there is no direct evidence that the copLABMGF is not, in fact, part of a plasmid in the case of Xcv strains.Further, complete genome-based studies will help in elucidating the origin and the location of these genes.
While copper resistome reflects environmental selection pressure in the Xcv strains, another piece of evidence in this regard is the presence of a common frameshift mutation in the Xanthomonadin pigment biosynthetic gene in all the Xcv strains.This suggests that the non-pigmented feature has been maintained over the decades in the Xcv strains from different geographical locations.The finding of a common frameshift mutation in a pigment biosynthetic gene cluster indicates the unique evolution of the pathogen and possible environmental-based selection pressure, which might be because of the way grapevines are cultivated as trellis in orchards.Similarly, in X. citri pv.mangiferaeindicae, the mango pathogen, the pigment cluster is deleted (Midha & Patil, 2014).However, the pigment locus is not investigated in multiple strains of diverse geographic regions to confirm the mutation or deletion of the cluster.Further molecular and genetic studies into the ecology of non-pigmented Xanthomonas are required.

CONCLUSION
Apart from the complete genome sequence of the reference strain of Xcv, multiple genome sequences from India along with Brazil have allowed us to understand the evolutionary pattern of Xcv.Interestingly, the geographic-specific variation is mediated by a plasmid, as can be seen in Indian isolates, and by genome expansion at the chromosome level, as seen in Brazilian isolates.Further characterisation of unique genes at the molecular and functional levels will help in identifying genes required for pathogenicity and virulence.A common frameshift mutation in all the Xcv isolates supports the monophyletic origin of the pathovar confirmed through phylogenomic investigation.Acquisition of multiple loci for copper resistance indicates high selection pressure for copper resistance over the course of time and usage of copper-based antimicrobial compounds.At the same time, the intrinsic nature of copper resistome, unlike plasmid or extrachromosomal in other Xanthomonas, strengthens this observation.The findings, along with genomic resources, will allow the development of markers in the identification and management of this important pathovar of a major fruit plant.Furthermore, the comparative genomic study has provided insights into the origin and nature of the mutation, leading to the variant white colony phenotype of Xcv apart from its potential for rapid evolution, as seen in the genomic repertoire of resistance mechanisms of copper resistance and homeostasis.Further genetic and functional studies will help in unravelling the copper resistance mechanism of the Brazilian Xcv strains.
) similar to Xcv and were also isolated fromIndia during 1961-1962(Bansal et al., 2017).The monophyletic clade of Xcv strains was further forming two subclades, where Indian strains were clustered together in one subclade, and Brazilian strains were clustered in another subclade, indicating further diversification.The Xcv strains shared approximately 99.9% orthoANI values with each other and approximately 98.7% orthoANI values with X. citri pv.citri (Xcc) strain 306 (Table likelihood core gene phylogeny of X. citri pathovar representatives.X. perforans DSM18975 T is taken as an outgroup.Indian Xcv strains are depicted in purple-coloured boxes, and Brazilian Xcv strains are depicted in yellow-coloured boxes.Bluecoloured circles represent internal nodes, and bootstrap values are mentioned in rectangular boxes at the nodes.

F
I G U R E 3 Frameshift mutation location in the Xanthomonas citri pv.viticola (Xcv) strains and Xcc 306 (A) Protein sequence alignment of phosphotransferase.Conserved and variable residues are highlighted in red and blue, respectively, and (B) nucleotide sequence alignment.Deleted residues are represented in green colour.

F
I G U R E 4 Genomic loci involved in copper resistance/homeostasis gene clusters.(A) Copper resistance operon organisation in the Brazilian Xanthomonas citri pv.viticola (Xcv) strains and related xanthomonads.(B) Copper homeostasis operon organisation in all the Xcv strains and Xcc 306.(C) Comparison of Tnp201.1 with NZ_PESX01000012 of CCRXv117.Genes are labelled along with their GC content for CCRXcv117.Arrows represent genes.The same genes across the strains are highlighted in the same colour.Numbers mentioned in the connecting regions between genes represent the percentage of amino acid identity.HP denotes hypothetical protein.
a Amino acid identity values are given.COMPARATIVE