Genomic Features and Insights into the Taxonomy, Virulence, and Benevolence of Plant-Associated Burkholderia Species

The members of the Burkholderia genus are characterized by high versatility and adaptability to various ecological niches. With the availability of the genome sequences of numerous species of Burkholderia, many studies have been conducted to elucidate the unique features of this exceptional group of bacteria. Genomic and metabolic plasticity are common among Burkholderia species, as evidenced by their relatively large multi-replicon genomes that are rich in insertion sequences and genomic islands and contain a high proportion of coding regions. Such unique features could explain their adaptability to various habitats and their versatile lifestyles, which are reflected in a multiplicity of species including free-living rhizospheric bacteria, plant endosymbionts, legume nodulators, and plant pathogens. The phytopathogenic Burkholderia group encompasses several pathogens representing threats to important agriculture crops such as rice. Contrarily, plant-beneficial Burkholderia have also been reported, which have symbiotic and growth-promoting roles. In this review, the taxonomy of Burkholderia is discussed emphasizing the recent updates and the contributions of genomic studies to precise taxonomic positioning. Moreover, genomic and functional studies on Burkholderia are reviewed and insights are provided into the mechanisms underlying the virulence and benevolence of phytopathogenic and plant-beneficial Burkholderia, respectively, on the basis of cutting-edge knowledge.


Introduction
The genus Burkholderia belongs to the subphylum of β-proteobacteria and encompasses Gram-negative bacterial species with high genetic versatility and adaptability to various ecological niches [1]. More specifically, they are able to occupy ecosystems as diverse as soils, plants, and even animal and human bodies as serious disease-causing agents [2][3][4]. Recently, Burkholderia organisms have drawn increasing attention as abundant and fundamental components of various ecosystems. The first member of the genus was isolated in 1942 from carnations displaying wilt and root-rot symptoms. It was initially named Phytomonas caryophylli and later renamed Pseudomonas caryophylli [5]. In 1950, another species was isolated from onions showing sour skin rot and named Pseudomonas cepacia [6,7]. Some members, originally included in the genus Pseudomonas, were later reclassified in the new Burkholderia genus due to improved taxonomic tools, the availability of techniques for DNA-DNA hybridization, as well as for 16S rRNA analysis [8]. Since then, many other species have been, and are still being, described and added to the genus.
Beukes et al. [21] addressed these complex phylogenetic relationships and confirmed the existence of generic boundaries among Burkholderia sensu lato bacteria. Although the analyzed genomes were similar in size and number of encoded genes, the genomes of Burkholderia sensu stricto showed higher G+C contents. The latter study includes a large-scale phylogenetic analysis on 106 common genes of Burkholderia sensu lato, specifically selected to limit phylogenetically irrelevant data. The results support the subdivision of Burkholderia sensu lato into five distinct lineages: Burkholderia sensu stricto, Paraburkholderia, Caballeronia, Robbsia, and Paraburkholderia rhizoxinica [21]. More recently, whole-genome analysis has been employed in the context of an international joint effort to investigate the taxonomic status of selected species of Burkholderia sensu lato [22]. Based on the phylogenetic analysis of conserved genes and average nucleotide and amino acid identities, two novel genera were proposed within the Burkholderia sensu lato assemblage. The first, Mycetohabitans, encompasses fungal symbiotic species with relatively small genomes. The second, Trinickia, includes soil and plant-associated species. Furthermore, the comparison of core-genomes revealed the presence of functional genus-specific genes. A historical overview of the major rearrangements and updates in Burkholderia taxonomy is summarized in Table 1.
It is clear that the Burkholderia sensu lato group has undergone several taxonomic changes and could be subjected to further adjustments in the future. The increased availability of genomic information, as well as the advancements in bioinformatics analysis, have enhanced our understanding of the evolutionary and taxonomic relationships between clinically, environmentally, industrially, and agriculturally important Burkholderia species. All this considered, the large group of Burkholderia sensu lato is currently composed of six genera, i.e., Burkholderia sensu stricto, Caballeronia, Paraburkholderia, Robbsia, Mycetohabitans, and Trinickia. The six genera were established after phylogenetic analysis of the main representative Burkholderia sensu lato strains using the neighbor-joining method, based on 16S rRNA sequences retrieved from the SILVA database (https: //www.arb-silva.de/), as shown in Figure 1.
With regard to the scope of the current review, phytopathogenic Burkholderia species are classified in the Burkholderia sensu stricto group, closely related to the Burkholderia cepacia complex; the former Burkholderia andropogonis is currently a separate genus as Robbsia andropogonis; and Burkholderia caryophylli is now classified as Trinickia caryophylli. On the other hand, most of the plant-beneficial Burkholderia species were transferred to the Paraburkholderia and Caballeronia genera. Phylogenetic tree constructed by the neighbor-joining method using 16S rRNA sequences of representative strains of Burkholderia sensu lato assemblage. The 16S rRNA sequences were retrieved from the SILVA database (https://www.arb-silva.de/). Species names were corrected to match the current taxonomic positions and the accession numbers for each sequence are given in parenthesis following the species name. The updated genera and their phylogenetic relationships are also shown. Ralstonia pickettii (AY741342) was used as an outgroup. Phylogenetic tree constructed by the neighbor-joining method using 16S rRNA sequences of representative strains of Burkholderia sensu lato assemblage. The 16S rRNA sequences were retrieved from the SILVA database (https://www.arb-silva.de/). Species names were corrected to match the current taxonomic positions and the accession numbers for each sequence are given in parenthesis following the species name. The updated genera and their phylogenetic relationships are also shown. Ralstonia pickettii (AY741342) was used as an outgroup.

2011
A second genus (Caballeronia) was suggested Based on phylogenetic analysis of multiple genes and comparative genomics; however, the evidence was not sufficient to confirm the new grouping [15] 2014 The genus Paraburkholderia was proposed Based on the analysis of conserved sequence in/dels [18] 2016 Inclusion of several species in the Paraburkholderia genus and establishment of the Caballeronia genus Eleven species were reclassified as Paraburkholderia and 14 species were transferred to the newly established Caballeronia genus [19] 2017 Burkholderia andropogonis was separated in a newly proposed genus as Robbsia andropogonis Based on multilocus sequence, 16S rRNA gene phylogeny, and average nucleotide identity analyses, as well as tetranucleotide signature frequency and percentage of conserved proteins [20] 2017 Confirmation of the genetic boundaries among the 4 established groups and suggestion of a fifth division: Paraburkholderia rhizoxinica Five groups (Burkholderia sensu stricto, Paraburkholderia, Caballeronia, Robbsia, Paraburkholderia rhizoxinica) were separated based on maximum likelihood phylogenies using the amino acid and nucleotide sequence of 106 conserved proteins [21] 2018 Two novel genera (Mycetohabitans and Trinickia) were proposed Based on whole-genome comparative study and phylogenetic analysis of conserved genes [22]

Genomic Features of the Plant-Associated Pathogenic and Beneficial Burkholderia
Because of the clinical importance of the B. cepacia complex, special consideration is devoted to these bacteria by genomic studies and, therefore, their genomes are well-characterized [9]. Besides common genomic features shared by most organisms of the large Burkholderia sensu lato group, functional features specific to the plant-associated groups are present and allow them to occupy various ecological niches with different lifestyles, as pathogens, legume nodulators, or biocontrol agents. Common features of Burkholderia species include relatively large multi-replicon genomes, the presence of genomic islands, and multiple insertion sequences conferring high genome plasticity [9,10].

Genome Size
Generally, Burkholderia organisms have larger genomes (7-8 Mbp on average) than other bacteria; however, the genome size greatly varies among different Burkholderia groups. The smallest genome among the free living Burkholderia (3.75 Mbp) is found in the fungal endosymbiont, Mycetohabitans rhizoxinica (basonym Burkholderia rhizoxinica), whereas the largest genome (11.5 Mbp) is found in the soil bacterium Paraburkholderia terrae (basonym Burkholderia terrae). It is suggested that genome size can provide insights into the evolutionary history and the relationships between Burkholderia species. It is also noted that the particular lifestyle of Burkholderia species might reflect selective pressure for a specific genome size [23].
Two main models for genome evolution in Burkholderia species were proposed: (i) plasmids might undergo rearrangements following parallel transfer, becoming established as additional chromosomes or megaplasmids and allowing for bacterial acquisition of essential genes; and (ii) loss of genomic information following rearrangements [24]. An example of the latter mechanism could be M. rhizoxinica, the obligate endosymbiont of the zygomycete Rhizopus microsporus, which is a causal agent of rice seedling blight development by production of rhizoxin [25]. The genome of M. rhizoxinica, containing one chromosome, a megaplasmid, and a plasmid, is remarkably small (relative to other Burkholderia), most likely because of rearrangements and deletion of genomic information, especially in genes coding for mobile genetic elements. The endosymbiotic lifestyle reduced the bacterial capacity to adapt to varying habitats and made efficient sensing superfluous [26,27]. On the other hand, a large part of the chromosomal coding region is devoted to the production of the secondary metabolite, rhizoxin [26,27].
Regarding the plant symbiotic Burkholderia, which can also be free-living in the bulk soil or water, a characteristic large genome size could explain their versatile lifestyle and the capacity to perform many plant-beneficial functions [28].
In another comparative genomic study of phytopathogenic Burkholderia species, Seo et al. [29] analyzed the genome sequences of different strains of B. glumae, B. gladioli, and B. plantarii and found genome size variations not only between species but also between different strains of the same species. For B. glumae, the smallest genome size,~4.9 Mbp, was exhibited by the AU6208 strain, whereas the largest,~7.2 Mbp, was found in the BGR1 strain. The relatively small genome size of B. glumae AU6208 could be attributed to genome rearrangements or deletions, as a result of adaptation to different hosts. Notably, although pathogenic to rice, AU6208 was originally isolated from infant patients, unlike other strains that were all isolated from rice as the original host. The genome size of B. gladioli BSR3 is 9 Mbp and that of B. plantarii is~8 Mbp.

Multi-Replicon Nature
A multi-replicon genome structure is found in most Burkholderia species with various numbers of chromosomes and plasmids. Martínez-Aguilar et al. [28] studied the multi-replicon genomes of plant-associated diazotrophic Burkholderia species and found that three strains of Paraburkholderia unamae and Paraburkholderia silvatlantica (formerly Burkholderia spp.) contain genomes of four replicons. The largest and smallest replicon sizes are~3.31 and~1.11 Mbp, respectively. In addition, the genomes of three strains of Paraburkholderia tropica (formerly Burkholderia tropica) contain five replicons ranging from~3.24 to~0.53 Mbp. The authors suggested that such multi-replicon large genomes could explain the capacity of these diazotrophic Burkholderia species to efficiently colonize the rhizosphere as well as endophytic environments.
A multi-replicon genome structure is also reported among the plant-pathogenic Burkholderia species. The genomes of the representative strains B. glumae BGR1 and B. gladioli BSR3 contain two chromosomes and four plasmids, while the B. plantarii ATCC43733 genome contain two chromosomes and one plasmid [29]. Agnoli et al. [30] found that deletion of the third chromosome from the genomes of several BCC species results in the attenuation of virulence and the loss of many functions such as the ability to utilize several substrates, antifungal activity, and biosynthesis of extracellular polymeric substances. Intriguingly, bacterial viability is not significantly affected, despite the massive loss of genomic information (~1 Mbp). This finding indicates that multi-replicon genomes are essential for adaptation but not necessarily for bacterial growth and survival.
While studying the pan-and core-genomes of Burkholderia species, Ussery et al. [14] found that only few hundred genes are conserved among all genomes to form the Burkholderia core genome, whereas more than 40,000 gene families are present in the Burkholderia pan genome. A more recent study of the available genome sequences, focusing on 110 genome sequences, shows that around 35,680 genes belong to pan-gene and 717 to core-gene families [23]. In the latter study, the number of shared genes has significantly dropped upon adding genes from different replicons to the analysis, which indicates the low percentage of genes shared by different genomic replicons of Burkholderia. In the same context, Seo et al. [29] analyzed the genomes of 106 Burkholderia species, representing the different groups, and found that 78,782 genes correspond to pan-genomes, while 587 are core genes conserved in all tested species. By considering the genomes of several strains of three plant pathogenic Burkholderia (B. glumae, B. gladioli and B. plantarii), 12,758 pan-genome genes are present in the tested strains, while 1908 genes are found conserved as the core-genome [29].

Genomic Islands and Multiple Insertion Sequences
Along with the above-mentioned features of Burkholderia genomes, other unique features are related to the genetic plasticity of these bacteria, such as the presence of genomic islands and multiple insertion sequences. Genomic islands refer to foreign DNA regions that have been horizontally acquired and became integrated within the genomes of several Burkholderia species. They have important evolutionary roles and can specify several accessory functions [31]. Multiple approaches have been used to identify these foreign DNAs, such as the analysis of G+C content, which has been found to differ in genomic islands with respect to the rest of the genome [32,33]. Genomic islands vary in their structure and content between Burkholderia species and account for more than 10% of the genomes in most species [24]. Genomic islands linked to pathogenicity and transmission have been extensively characterized. However, the coding capacity of genomic islands is not only associated with pathogenicity and virulence but also with adaptive traits such as symbiosis and metabolism [31]. An example of this is the environmental species Paraburkholderia xenovorans (formerly Burkholderia xenovorans), in which more than 20% of total genes are a recent acquisition and mainly involved in niche adaptation [34].
The genomes of Burkholderia species also contain many diverse insertion sequences that are mainly attributed to genomic rearrangements and replicon fusion and are related to transcriptional regulation [10,24,35,36]. This finding indicates that the presence of multiple insertion sequences in Burkholderia genomes has a role in phenotypic, as well as genetic diversity, by activating and/or inactivating adjacent genetic loci. Target genes may be inactivated or, in the case of adjacent promoter-less genes, activated by the promoters provided by insertion sequences [37]. Furthermore, the presence of multiple copies of the same insertion sequence may drive genetic rearrangements by fusion of two replicons, inversion, deletion, and duplication of the intervening region [38,39]. Accordingly, the transposition of insertion sequences is a very important evolutionary mechanism for the emergence of adaptation capacity and variability in Burkholderia species. A clear example of insertion sequence-mediated genomic alterations is provided by the genome of B. mallei, which is suggested to arise from several deletions and inversions of the Burkholderia pseudomallei genome. Evidence supporting such an assumption includes the presence of insertion sequences bordering most of the breakdown points and the presence of regions from chromosome 1 of B. pseudomallei on chromosome 2 of B. mallei [40]. In the phytopathogen B. glumae, the insertion sequences IS1417, IS1418, and IS1419 are characterized and found to be prevalent among different field strains. Apparently, these sequences may have been horizontally acquired by B. glumae from other related species, as they are also found in species such as B. gladioli and B. plantarii, with IS1417 being the most prevalent [41].

Plant Pathogenic Burkholderia
The Burkholderia group comprises several etiological agents of plant diseases causing significant economic losses by affecting the production of various crops. The major phytopathogenic species of Burkholderia group are B. glumae, B. gladioli, B. plantarii, R. andropogonis, and Trinickia caryophylli (formerly Burkholderia caryophylli) [42]. Although the mechanisms involved in pathogenicity and adaptation have not been fully elucidated, genomic studies and functional analyses, accompanied by phenotypic characterization, have enhanced our understanding of the virulence factors involved [29].
B. glumae, causing bacterial panicle blight and seedling rot of rice, was first reported in Japan, and is currently considered a major threat to rice cultivation worldwide [43,44]. In rice, the seed borne nature of B. glumae facilitates disease spread, which is estimated to cause the loss of up to 75% of heavily infested fields [45]. B. glumae is also found to be responsible for wilting symptoms in other hosts including pepper, sunflower, and tomato [46].
B. plantarii was first reported to cause rice seedling blight in Japan and has been isolated from several other host plants [42,51]. Both B. plantarii and B. gladioli have also been reported to cause, in rice, a similar disease to that caused by the related species B. glumae [51,52].
R. andropogonis (formerly known as B. andropogonis) was first reported to be the causative agent of strip disease in sorghum and leaf spot in velvet bean, and since then it has been described to cause disease in a wide range of host plants [42,[53][54][55].
T. caryophylli (formerly B. caryophylli) is responsible for a damaging disease of carnation in Japan, causing significant losses and was also found to cause onion rot [56,57].
The well-characterized fungal endosymbiont M. rhizoxinica (formerly B. rhizoxinica) could also be considered as a phytopathogen due to its cooperation with the phytopathogenic fungus Rhizopus microsporus in the development of rice seedling blight [42]. M. rhizoxinica controls the vegetative growth of the fungus [58]. However, although the main virulence factor of rice seedling blight was originally thought to be produced by the fungus, it was later revealed that M. rhizoxinica is responsible for rhizoxin production [59].

Virulence Factors in Phytopathogenic Burkholderia
The first isolated members of Burkholderia group were phytopathogens with unique genomic features, as described above, enabling them to develop an arsenal of virulence factors, thereby causing diseases in various host plants [29,42]. The pathogenicity of phytopathogenic members of the Burkholderia group is complex and several virulence factors have been found to contribute to disease development. The virulence factors and their genetic regulation in B. glumae have been extensively studied due to the significant economic impact of bacterial panicle blight and bacterial grain rot on rice cultivations worldwide [44,60]. The major virulence factors include phytotoxins and other virulence-related enzymes, bacterial motility, secretion systems, and exopolysaccharides (EPSs). Some of these virulence factors are common among phytopathogenic Burkholderia, whereas others are species-specific.

Phytotoxins
Phytotoxins are plant-deleterious compounds produced by pathogens and one of the major responsible factors for the development of disease symptoms [61]. Toxoflavin, the most important and studied phytotoxin of phytopathogenic Burkholderia species, was first described as an azapteridine antibiotic in Burkholderia cocovenenans [62], later found in B. glumae [63], and more recently in the related species B. gladioli [64]. Toxoflavin from B. glumae is associated with the inhibition of sprout and root elongation in rice seedlings, as well as severe damage in toxin-treated panicles [65]. Moreover, Jeong et al. [46] showed that toxoflavin is also involved in development of severe wilting symptoms in several other crops, indicating that a broad range of host plants are affected. Toxoflavin is also toxic to mice and exhibits antibacterial and antifungal activities [66]. The wide spectrum of toxoflavin activities might be explained by its role as an active electron carrier between NADH and oxygen, involved in the production of reactive oxygen species such as hydrogen peroxides [67].
The genetic regulation of toxoflavin biosynthesis and transport is well characterized in B. glumae. A tox operon composed of five genes (toxA, toxB, toxC, toxD, and toxE) is responsible for toxin biosynthesis and is regulated by a LysR-type regulator, ToxR [68,69]. In a later study, another operon composed of four genes (toxF, toxG, toxH, and toxI) is responsible for toxoflavin transportation [70]. In the same study, the regulation of both tox operons is further elucidated. Upon activation by toxoflavin binding (as a co-inducer), ToxR regulates the expression of both operons. Moreover, the other regulator, ToxJ, is also involved in the activation of both tox operons. The activation of ToxJ is mediated by a quorum sensing (QS) system via the luxI homolog, tofI, and the luxR homolog, tofR, via AHL (N-acyl homoserine lactone) QS signals. That is, bacterial cell density is required for active toxoflavin production, so that bacterial cells can start toxoflavin-mediated attack of host cells only once a sufficiently high bacterial population has been reached [70]. An illustration of both tox operons and a proposed model of regulation is shown in Figure 2. The complete genome sequence of B. gladioli BSR3 reveals the presence of the toxoflavin operons, as well as relevant regulatory genes [72]. To investigate the genetics of toxoflavin production and regulation in B. gladioli, Lee et al. [73] conducted a comparative genomic analysis of several strains of B. gladioli, including the non-toxoflavin-producing strain KACC11889 and B. glumae BGR1. The results indicate that the toxoflavin-producing strains of B. gladioli shared with B. glumae the genes for toxoflavin production and regulation, except that toxJ is located on another chromosome in B. gladioli. In addition, the genome of the non-toxoflavin-producing strain, B. gladioli KACC11889, contains the operons for biosynthesis and transport of toxoflavin, whereas the toxoflavin regulatory QS genes, tofI and tofR, are not found. Thus, the loss of these regulatory genes results in the inhibition of toxoflavin The complete genome sequence of B. gladioli BSR3 reveals the presence of the toxoflavin operons, as well as relevant regulatory genes [72]. To investigate the genetics of toxoflavin production and regulation in B. gladioli, Lee et al. [73] conducted a comparative genomic analysis of several strains of B. gladioli, including the non-toxoflavin-producing strain KACC11889 and B. glumae BGR1. The results indicate that the toxoflavin-producing strains of B. gladioli shared with B. glumae the genes for toxoflavin production and regulation, except that toxJ is located on another chromosome in B. gladioli. In addition, the genome of the non-toxoflavin-producing strain, B. gladioli KACC11889, contains the operons for biosynthesis and transport of toxoflavin, whereas the toxoflavin regulatory QS genes, tofI and tofR, are not found. Thus, the loss of these regulatory genes results in the inhibition of toxoflavin production, confirming the suggestion of a similar mechanism of toxoflavin regulation in B. gladioli and B. glumae [73].
Although toxJ regulatory genes are identified in the genome of the related species B. plantarii, toxoflavin biosynthesis and transport genes are not identified [29]. B. plantarii produces another phytotoxin, a non-benzenoid aromatic tropolone, which is responsible for seedling blight symptoms in rice, as confirmed by the replication of the characteristic bacterial-induced blight symptoms upon treatment of rice seedlings with exogenous tropolone [51,74]. Tropolone is also a potential iron chelator and has a wide range of biological roles such as antibacterial, antifungal, and insecticidal activities [75,76].
B. plantarii produces AHL-QS signal molecules resembling those of B. glumae, as the luxI family member, plaI, and the luxR family member, plaR, are highly similar to the B. glumae AHL-QS system genes, tofI and tofR. Through the generation of plaR knockout mutants, the role of the AHL-QS regulatory system in the virulence of B. plantarii to rice is established [77]. It was later revealed that treatment of B. plantarii with AHL-QS inhibitors resulted in the suppression of tropolone production by blocking the expression of plaI [78]. Therefore, a resemblance between the regulation of tropolone biosynthesis and that of toxoflavin, in B. glumae, could be inferred from the major role played by QS-related genes. This possibility is further supported by the presence of QS-AHL signal synthase-regulator loci within the tropolone biosynthesis operon (Seo et al., 2015 [29]). More recently, Miwa et al. [79] reported that a two-component signal transduction system regulates both tropolone biosynthesis and virulence in B. plantarii. Defects in any of the three genes encoding the two-component system (sensor histidine kinase and response regulators) resulted in the inhibition of toxin production and attenuation of bacterial virulence in rice. Three genes (i.e., troR1, troK, and troR2) are suggested to form an operon for tropolone synthesis, due to their adjacent chromosome position and their simultaneous requirement for toxin production [79]. A schematic organization of the two-component system genes controlling tropolone production is shown in Figure 3A. Further investigations are required to unravel the gene cluster organization and the regulation of tropolone biosynthesis, the main virulence factor in B. plantarii, and the possible regulatory roles of tropolone. B. plantarii produces AHL-QS signal molecules resembling those of B. glumae, as the luxI family member, plaI, and the luxR family member, plaR, are highly similar to the B. glumae AHL-QS system genes, tofI and tofR. Through the generation of plaR knockout mutants, the role of the AHL-QS regulatory system in the virulence of B. plantarii to rice is established [77]. It was later revealed that treatment of B. plantarii with AHL-QS inhibitors resulted in the suppression of tropolone production by blocking the expression of plaI [78]. Therefore, a resemblance between the regulation of tropolone biosynthesis and that of toxoflavin, in B. glumae, could be inferred from the major role played by QSrelated genes. This possibility is further supported by the presence of QS-AHL signal synthaseregulator loci within the tropolone biosynthesis operon (Seo et al., 2015 [29]). More recently, Miwa et al. [79] reported that a two-component signal transduction system regulates both tropolone biosynthesis and virulence in B. plantarii. Defects in any of the three genes encoding the twocomponent system (sensor histidine kinase and response regulators) resulted in the inhibition of toxin production and attenuation of bacterial virulence in rice. Three genes (i.e., troR1, troK, and troR2) are suggested to form an operon for tropolone synthesis, due to their adjacent chromosome position and their simultaneous requirement for toxin production [79]. A schematic organization of the two-component system genes controlling tropolone production is shown in Figure 3A. Further investigations are required to unravel the gene cluster organization and the regulation of tropolone biosynthesis, the main virulence factor in B. plantarii, and the possible regulatory roles of tropolone.
The other phytotoxin relevant for plant disease is rhizoxin, produced by M. rhizoxinica, the fungal endosymbiont of R. microspores. This bacterial endosymbiont alters the vegetative growth and sporulation of the host fungus by preventing the production of endosymbiont-free fungal spores [25,26]. The endosymbiont-produced rhizoxin is an effective inhibitor of eukaryotic cell growth by acting on β-tubulin and consequently blocking cell mitosis [80]. Besides its role as a virulence factor responsible for seedling blight symptoms in rice, various additional physiological functions have been attributed to rhizoxin, including reproduction in the host fungus and antibiotic activity against other microbes [25,81].
The availability and analysis of the complete genome sequence of M. rhizoxinica have allowed for the clarification of rhizoxin biosynthesis and the identification of the responsible gene cluster (rhi) [26,59]. The rhi locus is found on the large chromosome, flanked by transposase genes, indicating that it may represent a potentially mobile region of the genome [26,27]. It is also suggested that the rhizoxin gene cluster might have been transferred from the megaplasmid to the chromosome after horizontal acquisition, as concluded from the chromosomal location of the gene cluster, near the replication terminus, and the comparison of its G+C content with that of the rest of the chromosome [26]. Notably, in the comparative genomic analysis conducted by Seo et al. [29], genes involved in rhizoxin biosynthesis are also detected in other phytopathogenic Burkholderia spp., such as B. plantarii ATCC43733 and B. glumae PG1, for which the ability to produce rhizoxin has not been demonstrated. The organization on the rhi gene cluster is shown in Figure 3B.  The other phytotoxin relevant for plant disease is rhizoxin, produced by M. rhizoxinica, the fungal endosymbiont of R. microspores. This bacterial endosymbiont alters the vegetative growth and sporulation of the host fungus by preventing the production of endosymbiont-free fungal spores [25,26]. The endosymbiont-produced rhizoxin is an effective inhibitor of eukaryotic cell growth by acting on β-tubulin and consequently blocking cell mitosis [80]. Besides its role as a virulence factor responsible for seedling blight symptoms in rice, various additional physiological functions have been attributed to rhizoxin, including reproduction in the host fungus and antibiotic activity against other microbes [25,81].
The availability and analysis of the complete genome sequence of M. rhizoxinica have allowed for the clarification of rhizoxin biosynthesis and the identification of the responsible gene cluster (rhi) [26,59]. The rhi locus is found on the large chromosome, flanked by transposase genes, indicating that it may represent a potentially mobile region of the genome [26,27]. It is also suggested that the rhizoxin gene cluster might have been transferred from the megaplasmid to the chromosome after horizontal acquisition, as concluded from the chromosomal location of the gene cluster, near the replication terminus, and the comparison of its G+C content with that of the rest of the chromosome [26]. Notably, in the comparative genomic analysis conducted by Seo et al. [29], genes involved in rhizoxin biosynthesis are also detected in other phytopathogenic Burkholderia spp., such as B. plantarii ATCC43733 and B. glumae PG1, for which the ability to produce rhizoxin has not been demonstrated. The organization on the rhi gene cluster is shown in Figure 3B.
The phytopathogen R. andropogonis produces the enol-ether amino acid phytotoxin, rhizobitoxine, that is responsible for leaf chlorosis in host plants [82]. Notably, rhizobitoxine has also been associated with a positive role in the symbiosis of legume nodulators. The phytotoxic activity of rhizobitoxine is related to inhibition of the methionine and ethylene biosynthetic pathways. The genes for rhizobitoxin biosynthesis (rtx operon) are identified in the legume nodulator, Bradyrhizobium elkanii [83,84]. However, further work is required to describe the biosynthetic genes and their regulation in R. andropogonis.
Another example of the versatility of Burkholderia species and their contribution to vital ecological aspects is the phytotoxin produced by another Burkholderia endo-symbiont. In this case, the endo-symbiont is associated with the pine wood nematode, Bursaphelenchus xylophilus, causing pine wilt disease, a devastating threat to pine trees worldwide [85]. The pine wood nematode endo-symbiont Burkholderia arboris produces pyochelin, which is a bacterial siderophore with several bioactivities due to its capacity to generate reactive oxygen species [86]. However, B. arboris-produced pyochelin is responsible for damaging plant tissues and is linked to the development of pine wilt disease, as the treatment of pine callus with purified pyochelin confirms its strong phytotoxicity [87]. Further studies are required to unveil the phytotoxic mechanism of pyochelin.
The different phytotoxins produced by the phytopathogenic species of Burkholderia group, their effect, and mode of action are summarized in Table 2. * M. rhizoxinica is considered as a phytopathogenic species, as it is responsible for the production of rhizoxin, the major virulence factor for rice seedling blight.

Secretion Systems
The ability of Gram-negative bacteria, including Burkholderia spp., to control the translocation of potent effectors from their sites of synthesis to the exterior of the cell, or even to other surrounding organisms, is a very important tool for communication and survival [88]. Six types of bacterial secretion systems have been discovered and differentiated based on their structure, function, or type of secreted effectors [89]. Species possessing more than one type of secretion system are also common [90].
Phytopathogenic Burkholderia spp. exhibit a wide spectrum of secretion systems, playing essential roles in adaptation and virulence. Kang et al. [91] confirmed the association of Type 2 secretion system (T2SS) and T3SS with functions related to host plant colonization and bacterial growth, as indicated by the reduced virulence and retarded growth of T2SS-and Hrp T3SS-deficient mutants in host plants. Comparative genomic analysis of phytopathogenic Burkholderia reveals that the genes for T3SS are highly conserved in several strains of B. glumae, B. gladioli, and B. plantarii. However, within species-variation is observed in T1SS and T4SS [29].
Phytopathogenic Burkholderia organisms have different groups of T6SSs. Group 1 is conserved in all species [92]. T6SS is involved in the delivery of effectors to both eukaryotic and prokaryotic cells. Thus, it is suggested to be important for the competition with other surrounding microbes and for virulence to host plants [93].
Although several functions have been reported for different bacterial secretion systems, for many reasons, including the remarkable diversity and limited expression of effectors as well as the multifarious nature of the secretion systems, not all secreted effectors have been identified and their exact mechanisms of action are still elusive [91,94,95].

Other Virulence Factors
Several other factors are associated to the virulence of phytopathogenic Burkholderia species. Virulence-related enzymes such as lipases, proteases, and cellulases, as well as exopolysaccharides (EPS), contribute to the virulence of several species. For full virulence, the degradation of plant cell wall polymers by hydrolytic enzymes is crucial. Lipase is an important virulence factor for many bacterial pathogens, as it is responsible for the hydrolysis of the carboxyl ester bonds present in triacylglycerols, which liberate fatty acids and glycerol. Lipase could also play a role in the degradation of the epicuticle wax layer in plant tissues [96][97][98]. In B. glumae, lipase is also regulated by the AHL-QS system and is essential for panicle blight symptoms in rice [99]. The gene encoding lipase (LipA) is found in the three phytopathogenic species B. glumae, B. gladioli, and B. plantarii [29].
Pectin, mainly composed of polygalacturonic acid, is one of the major constituents of the plant cell wall. Polygalacturonases produced by phytopathogens can hydrolyze the pectic glycosidic bonds, depolymerizing pectin and causing rotting of plant tissues. B. glumae produces two different polygalacturonases (encoded by pehA and pehB) that are possibly regulated by QS and secreted by a T2SS. The pehA locus is present in B. glumae but not detected in B. gladioli and B. plantarii [29].
The EPSs have been shown to contribute to the pathogenesis of several plant pathogens by causing occlusion in the host vascular systems, leading to plant wilting [100]. Although the role of EPSs in the epidemiology of rice seedling blight has not yet been clearly established, they could explain the wide range of crops showing wilting symptoms in response to B. glumae infection [46]. In T. caryophylli, EPSs are suggested as important virulence factors, responsible for bacteria-plant interactions [101]. Further investigations are required to elucidate the contribution of EPSs to the virulence of phytopathogenic Burkholderia.

Plant-Beneficial and Symbiotic Burkholderia Species
If, on the one hand, pathogenic Burkholderia species are the subject of intensive research, mainly because of their economic implications, on the other hand, plant-beneficial and symbiotic species have attracted the attention of scientists due to their impressive ability to promote plant physiology and growth. The main intriguing findings about this group of bacteria are the discovery of nitrogen fixation ability [102] and the identification of plant endosymbionts and legume nodulators [103,104]. As a result of multiple taxonomic rearrangements and updates (see Section 2), the majority of plant-beneficial and symbiotic strains increasingly clustered together in separate clades, mainly within the Paraburkholderia group, based on the presence of several common features such as nitrogen fixation, metabolic versatility, and plant growth promoting ability [105]. Comparative genomic analyses between beneficial plant-associated Burkholderia species revealed a high degree of genetic synteny in their large multireplicon genomes [28]. Previous studies focusing on genomics and metabolism have reported the capacity of this Burkholderia subgroup to support healthy plant growth.

Benevolence Factors in Plant-Beneficial and Symbiotic Burkholderia
Beneficial plant-associated species of the Burkholderia sensu lato group have been reported to carry out several functions that are beneficial to plants, such as colonization of plant surfaces and rhizosphere, nitrogen fixation, production of siderophores, solubilization of phosphates, metabolism of volatile compounds, degradation of pollutants, production of plant growth-promoting phytohormones, reduction of plant ethylene levels, and protection from biotic and abiotic stress [106][107][108].

Colonization of Plant Tissues and the Role of EPSs
The ability of plant-beneficial bacteria to colonize the surface of vegetal tissues and the rhizosphere is important for the promotion of healthy plant growth. In this respect, the production of EPSs is essential not only to enhance bacterial tolerance to environmental stress but also for efficient colonization, by adhesion to the root surface, and for the aggregation of root-adhering soil particles [100,109]. Bacterial EPSs also play a key role in the interaction between the plant and the endophytic or associated bacteria. It has been suggested that bacteria-plant symbiosis in the β-rhizobia is initiated by the bacterial production of an infection thread allowing bacteria to invade and colonize plant roots. Jones et al. [110] reported that, unlike wild-type bacteria, mutants deficient in EPS production are unable to initiate a symbiotic relationship with a host plant. EPS-producing bacteria regulate the expression of plant defense-related genes to prepare plants for colonization. These findings suggest that bacterial EPSs could serve as signals regulating gene expression in plants [110]. Several members of plant-beneficial Burkholderia, such as Paraburkholderia tropica, Paraburkholderia caribensis, Paraburkholderia phytofirmans, and Paraburkholderia kururiensis, have been reported to synthesize EPSs, the structure of which has been elucidated [111][112][113][114].

Nitrogen Fixation in Diazotrophic and Legume Nodulator Burkholderia
In Burkholderia bacteria, biological nitrogen fixation has been found in both plant endosymbiotic diazotrophic non-legume nodulator species, such as P. phytofirmans and Paraburkholderia unamae, and in legume nodulator species such as Paraburkholderia tuberum and P. phymatum [102,115].
The nitrogenase enzyme responsible for nitrogen fixation is composed of two structurally conserved components. Component I is encoded by nifD and nifK and component II is encoded mainly by nifH, forming the basic regions of the nif operon [116][117][118]. The nifHDK operon was first reported in the Burkholderia group in an endosymbiotic species of the arbuscular mycorrhizal fungus, Gigaspora margarita [118]. The presence of nifHDK coupled with nitrogen fixation has been confirmed in several other species in the Burkholderia group [102]. Moreover, the fix genes are also essential for nitrogen fixation, especially in microaerobic conditions. The nifHDK is controlled by the transcriptional regulator NifA. NifA and nifHDK, along with the fix genes, are regulated through the two-component signal transduction system through FixL and FixJ [119,120].
Legume nodulation is defined as the induction of nodule formation in plants by certain bacteria species, which fix atmospheric nitrogen for plants and receive carbons in exchange [121]. Previously, legume nodulation was thought to be an exclusive feature of the Rhizobiaceae members of α-Proteobacteria [122]. However, with the identification of Burkholderia legume nodulators, this hypothesis was disproved, and the term β-rhizobia was proposed to describe the legume nodulators in the β-proteobacteria [103,[123][124][125][126][127]. Since then, several other Burkholderia species have been reported as legume nodulators [15].
The nodulation genes (nod), encoding proteins for production, transport, and regulation of the nod factor, along with the nif genes for nitrogen fixation, are considered as genes essential for the β-rhizobia symbiosis and are reported to be directly involved in Burkholderia nodulation ability [11]. The phylogenetic analyses conducted on the symbiosis-related genes, nodC and nifH, have evidenced that symbiosis is an ancient function in Burkholderia and these genes are vertically inherited and not the result of recent horizontal acquisition [104]. In contrast, in a recent study by Estrada-de los Santos et al. [22], phylogenies constructed by using nifD and nifH for nitrogen fixation and nodABCD for nodulation provided evidence that the nif genes have been acquired via horizontal gene transfer [22]. In the same study, the nif genes are widely present among Burkholderia sensu lato bacteria, especially in the Paraburkholderia group of plant-associated legume nodulators. However, nif genes are not found in Caballeronia, Robbsia or the newly assigned genus of fungal endosymbionts, Mycetohabitans.
With regard to the nod genes, which are only found in the group of nodulators, phylogenetic analysis has divided the nodulators into two clades based on the respective host plants: the mimosoid nodulator and the papilionoids nodulator clades. The divergence in the nod genes among the Paraburkholderia nodulators seems to be related to host plant specificity [120]. A more in-depth evaluation of nod gene organization supports the distinction between mimosoid and papilionoids nodulators. In the mimosoid-nodulators, the nod operon was organized as a single copy of nodBCIJHASU, differing from the related α-rhizobia and the papilionoid-nodulators [120]. The genetic organization of the nod and nif/fix gene clusters of representative species of α-rhizobia, mimosoid and papilionoid Paraburkholderia nodulators is shown in Figure 4. The papilionoid nodulators' nod genes could have been acquired from the related α-rhizobia, whereas a different evolutionary origin is suggested for the nod genes of mimosoid nodulators [22]. With regard to the nod genes, which are only found in the group of nodulators, phylogenetic analysis has divided the nodulators into two clades based on the respective host plants: the mimosoid nodulator and the papilionoids nodulator clades. The divergence in the nod genes among the Paraburkholderia nodulators seems to be related to host plant specificity [120]. A more in-depth evaluation of nod gene organization supports the distinction between mimosoid and papilionoids nodulators. In the mimosoid-nodulators, the nod operon was organized as a single copy of nodBCIJHASU, differing from the related α-rhizobia and the papilionoid-nodulators [120]. The genetic organization of the nod and nif/fix gene clusters of representative species of α-rhizobia, mimosoid and papilionoid Paraburkholderia nodulators is shown in Figure 4. The papilionoid nodulators' nod genes could have been acquired from the related α-rhizobia, whereas a different evolutionary origin is suggested for the nod genes of mimosoid nodulators [22].

Plant Growth Promotion
Plant growth promotion is common among members of the beneficial plant-associated Burkholderia, which possess several growth promoting factors involved not only in nitrogen fixation but also in ACC (1-aminocyclopropane-1-carboxylate) deaminase activity, phosphate solubilization, as well as siderophore and phytohormone production [105].
In plant tissues, ACC is an intermediate precursor in the production of the plant growth inhibitor, ethylene, in developing or stressed plants. The bacterially produced ACC deaminase enzyme catalyzes the conversion of ACC to ammonia and α-ketobutyrate, thereby reducing the ethylene levels in plant tissues, which in turn results in plant growth promotion, mainly manifesting as root elongation [128]. Production of ACC deaminase is a common feature in beneficial plantassociated Burkholderia species and the relevant encoding gene (acdS) is highly conserved and widely distributed among the rhizospheric, endophytic, and nodulator groups [129]. Furthermore, ACC deaminase activity is confirmed by comparing tomato plant responses to inoculation with an ACC deaminase-producing wild-type P. unamae strain and an acdS knockout mutant, which clearly demonstrates the plant growth-promoting properties of ACC deaminase-producing P. unamae [129]. In another study, an asdS deletion mutant of P. phytofirmans, devoid of ACC deaminase activity, lacked the ability to promote the root elongation of canola seedlings [130].
Phosphorus is an essential macronutrient for plant growth. Soil mineral and organic phosphates can be solubilized and become available for plants by the action of beneficial soil microbes [131]. A phosphate solubilization ability is reported in several species of plant-associated Paraburkholderia [132][133][134]. Among the species tested by Estrada-de los Santos et al. [22], only the soil bacterium Trinickia dabaoshanensis (formerly Paraburkholderia dabaoshanensis) can solubilize phosphate. The

Plant Growth Promotion
Plant growth promotion is common among members of the beneficial plant-associated Burkholderia, which possess several growth promoting factors involved not only in nitrogen fixation but also in ACC (1-aminocyclopropane-1-carboxylate) deaminase activity, phosphate solubilization, as well as siderophore and phytohormone production [105].
In plant tissues, ACC is an intermediate precursor in the production of the plant growth inhibitor, ethylene, in developing or stressed plants. The bacterially produced ACC deaminase enzyme catalyzes the conversion of ACC to ammonia and α-ketobutyrate, thereby reducing the ethylene levels in plant tissues, which in turn results in plant growth promotion, mainly manifesting as root elongation [128]. Production of ACC deaminase is a common feature in beneficial plant-associated Burkholderia species and the relevant encoding gene (acdS) is highly conserved and widely distributed among the rhizospheric, endophytic, and nodulator groups [129]. Furthermore, ACC deaminase activity is confirmed by comparing tomato plant responses to inoculation with an ACC deaminase-producing wild-type P. unamae strain and an acdS knockout mutant, which clearly demonstrates the plant growth-promoting properties of ACC deaminase-producing P. unamae [129]. In another study, an asdS deletion mutant of P. phytofirmans, devoid of ACC deaminase activity, lacked the ability to promote the root elongation of canola seedlings [130].
Phosphorus is an essential macronutrient for plant growth. Soil mineral and organic phosphates can be solubilized and become available for plants by the action of beneficial soil microbes [131]. A phosphate solubilization ability is reported in several species of plant-associated Paraburkholderia [132][133][134]. Among the species tested by Estrada-de los Santos et al. [22], only the soil bacterium Trinickia dabaoshanensis (formerly Paraburkholderia dabaoshanensis) can solubilize phosphate. The genetic background and mechanisms of phosphate solubilization have not been clearly characterized yet.
Plant growth regulators such as the auxin class of phytohormones are known to promote plant growth, particularly by controlling growth and development of plant roots [135]. Indole-3-acetic acid (IAA) is the most common and naturally occurring auxin and its presence has been extensively documented in several soil and rhizosphere bacteria. A clear example of IAA-producing species, among beneficial plant-associated Burkholderia, is the well-known plant growth-promoting P. phytofirmans, in which whole-genome sequence analysis reveals the presence of two independent putative IAA synthesis pathways, the indole-3-acetamide and the tryptophan side-chain oxidase pathway [136,137]. More recently, Zúñiga et al. [138], reported that IAA also plays a role in plant-microbe interactions (cell-cell communication) and that P. phytofirmans is not only a producer of IAA but can also degrade excess exogenous IAA that could suppress plant growth. The production of IAA by P. phytofirmans and its association with plant growth promotion and colonization were further confirmed by Naveed et al. [139] on maize plant, who found it to be dependent on L-Tryptophan.

Other Benevolence Factors in Plant-Associated Burkholderia
In addition to the above described beneficial effects of plant-associated Burkholderia, other plant-beneficial roles have been described. Several species exert antagonistic activities against phytopathogens including fungi, bacteria, and even nematodes. Broad-spectrum antibiotic activity is reported in a strain of Burkholderia ambifaria, exhibiting an effective control over several soil fungal pathogens [140]. Antifungal activity is reported in several other species, e.g., Paraburkholderia bryophila, Paraburkholderia megapolitana, Paraburkholderia ginsengiterrae, and Paraburkholderia panaciterrae [141,142]. Along with several other beneficial plant-associated Burkholderia, the latter species have also been found to produce siderophores, allowing them to compete for the available iron and dominate over pathogenic microbes [106]. In addition, the induction of resistance against biotic and abiotic stress was also reported, mainly by P. phytofirmans, as bacterial treatment triggers resistance against gray mold in grapevine and to improve plant tolerance to cold stress [143,144].
Another type of Burkholderia-beneficial interactions with plants is the intimate association of the plant leaf-nodulating Candidatus Burkholderia. Although the attempts to cultivate this group of bacteria remain unsuccessful, culture-independent molecular analyses are helpful to reveal the taxonomic position of these bacteria within the Burkholderia assemblage and the importance of this association to host plant [145]. The association of Ca. Burkholderia with leaves of Ardisia, Pavetta, Psychotria, and Serricanthe genera is vertically transmitted (i.e., transmitted from one generation to the next through seeds by incorporation to the reproductive stages), very specific and critical for the host plant development as the heat-treated seeds that germinate into aposymbiotic seedlings cannot reach maturity [146,147]. The genomic and functional studies on Candidatus Burkholderia kirkii and Candidatus Burkholderia crenata provide insights into secondary metabolism and possible beneficial roles to host plant such as the biosynthesis of antibiotic and insecticidal compounds for protection [145,148,149]. Future work is required to further characterize the mechanisms and roles of the unique symbiotic leaf-nodulating Ca. Burkholderia, which differs from the known symbiosis.
The main plant-associated phytopathogenic and beneficial species of the Burkholderia group, as well as their major virulence or benevolence factors, are summarized in Table 3. Mimosa nodules nod, nif, siderophore [22,166] Member that have been isolated from human clinical samples are excluded from the list of plant-associated beneficial species. Nod, can form nodules in legumes; nif, nitrogen fixation; EPSs, exopolysaccharides; T3SS, Type 3 secretion system; T6SS, Type 6 secretion system; IAA, indole acetic acid.

Conclusions and Future Perspectives
Burkholderia organisms are ubiquitous in diverse environmental niches and display a high adaptability to changing environments. Unique genomic features and a large encoding capacity provide this group of bacteria with the flexibility required for adaptation to various lifestyles. Recent advancements in sequencing techniques and bioinformatics tools have led to a remarkable increase in the availability of genome sequences, improving our comprehension of the relationships between different Burkholderia species, their taxonomy, the mechanisms underlying their versatility and the potential factors involved in bacterial virulence or benevolence. The phytopathogenic Burkholderia group encompasses several important etiological agents of plant disease, mainly causing significant loss of rice crops. Comparative genomic analyses combined with functional and phenotypic studies have generated new knowledge of disease-causing mechanisms and involved virulence factors. However, further research efforts are needed to better characterize the virulence factors of phytopathogenic Burkholderia, their genetic background, and regulation.
Regarding the beneficial plant-associated Burkholderia, although several studies have reported efficient plant growth-promoting and potential biocontrol activities of several members of Burkholderia spp., their use in agriculture has been limited due to concerns related to their similarity with opportunistic human pathogens. However, based on current knowledge, including refined taxonomy, as well as the absence of pathogenic hallmarks in their genomes, a clear distinction between these bacteria and human pathogens has been established, suggesting that, in the coming years, members of the beneficial Burkholderia spp. could be utilized as suitable alternatives to agrochemicals to promote plant growth, manage plant diseases, and for biotechnological applications

Conflicts of Interest:
The authors declare no conflict of interest.