Grudgingly sharing their secrets: new insight into the evolution of plant pathogenic bacteria


The conventional wisdom regarding bacterial plant pathogens is that strains found causing disease on plants are specialized and distinct from saprophytic strains found in the environment (soil, water, decaying plant material) that have no apparent role in disease. In this issue of New Phytologist, Monteil et al. (pp. 800–811) show that environmental populations are likely to have had a role in the emergence of Pseudomonas syringae pv. tomato (Pto), directly as reservoirs of pathogenic or potentially pathogenic strains and indirectly as sources of novel genetic variation that is introduced to pathogenic strains via recombination. They found strong signatures of recombination in P. syringae when disease-associated strains of Pto were analyzed together with environmental strains collected from alpine snow and creeks. Thus, P. syringae is likely to have diverse and recombinogenic environmental populations that produce clonal strains associated with disease epidemics. Analogous population structures have been demonstrated for environmental human pathogens, including P. aeruginosa (Pirnay et al., 2002). An extended life cycle of P. syringae, that incorporates movement within the freshwater cycle as well as enrichment and diversification in nonagricultural ecosystems, was proposed by Morris et al. (2008). Incorporating the results of Monteil et al., one can hypothesize an ‘evolutionary cycle’ for P. syringae involving both environmental and plant-associated populations (Fig. 1). A genetically diverse population exists in the environment, where variable conditions in space and time do not favor any one genotype and recombination among strains creates a web of phylogenetic relationships (Fig. 1A). Pseudomonas viridiflava, a close relative of P. syringae, also has diverse and recombinogenic populations in weed communities (Goss et al., 2005). These environmental populations may be key to understanding the emergence of pathogens, because there is increasing evidence that the ecology of plant and animal pathogens, across variable environments, has shaped their evolution as pathogens (Pallen & Wren, 2007; Morris et al., 2009).

Figure 1.

A hypothetical evolutionary cycle for Pseudomonas syringae plant pathogens. Population structures are illustrated through phylogenetic networks; circles represent multi-locus haplotypes in rough proportion to their frequency and connecting lines represent phylogenetic relationships. (A) Diverse environmental populations recombine frequently creating a web of ancestral relationships among genes in these populations. (B) Highly pathogenic clones originate from environmental reservoirs via an unknown number of genetic changes and proliferate in agricultural environments. (C) Agricultural populations are characterized by a less diverse clonal population structure, and strong selection for virulence and aggressiveness. (D) Recombination among agricultural pathogens occasionally produces new strains with increased fitness. Here, recombination or horizontal gene transfer between the red species and the blue species creates a novel purple strain. (E) Dispersal of pathogenic strains back to environmental reservoirs creates the opportunity for recombination of genes from agricultural strains into environmental populations.

Occasionally, perhaps rarely, an environmental strain comes in contact with a plant host that it can infect and in which it can proliferate. Like P. viridiflava, environmental strains of P. syringae may be opportunistic pathogens that occasionally cause outbreaks of disease under conducive conditions, particularly when plant hosts are stressed. A disease epidemic creates the opportunity for selection to act on new mutations and may lead to fixation of advantageous mutations in the population. Thus, a new crop strain emerges as a clone rather than as a diverse population (Fig. 1C). Pseudomonas syringae was previously inferred to have a clonal core genome based on samples from disease epidemics (Sarkar & Guttman, 2004; Cai et al., 2011). In human pathogens, loci under adaptive selection are associated with high rates of recombination, indicating that recombination has contributed to the emergence of new pathogenic strains (Didelot & Maiden, 2010). This is likely for plant pathogens as well, although most examples to date have been inferred to be the result of horizontal gene transfer among clonal populations. Phyllosphere and rhizosphere interactions among pathogens and nonpathogens may facilitate the acquisition of adaptive genes by recombination or horizontal gene transfer (Fig. 1D). Streptomyces turgidiscabies, causative agent of potato scab, contains the largest known pathogenicity island (PAI) in gram-positive bacterial plant pathogens (PAISt). The PAISt contains virulence genes including thaxtomin A, which imparts ability to cause potato scab, and has been transferred among plant pathogenic and saprophytic Streptomyces species (Huguet-Tapia et al., 2011).When the PAISt was experimentally integrated into the nonpathogen S. diastochromogenes, a plant-pathogenic phenotype resulted (Kers et al., 2005). Two aggressive tomato pathogens, Pto and Xanthomonas gardneri, share an array of type III effectors that are absent from other xanthomonads that cause bacterial spot disease on tomato (Potnis et al., 2011). The closest known relative of X. gardneri is X. campestris pv. campestris, a cruciferous and weed pathogen, suggesting evolution of this tomato pathogen via horizontal gene transfer. Another bacterial spot species, X. vesicatoria, appears to have acquired genes from multiple origins. Xanthomonas vesicatoria shares a common ancestor with X. gardneri and X. campestris pv. campestris for important pathogenicity genes such as Type II and III protein secretion clusters and some core effectors including avrBs2 (Fig. 2a; Potnis et al., 2011). By contrast, the core effector xopL in X. vesicatoria is phylogenetically most closely related to X. citri and the two bacterial spot pathogens, X. euvesicatoria and X. perforans (Fig. 2b). Xanthomonas vesicatoria also has phylogenetically distinct core effectors such as xopN (Fig. 2c), xopX, and xopK. These bacterial spot pathogens have clearly taken multiple evolutionary paths to effectively and aggressively colonize tomato. These paths may have included host range changes and even speciation, and will be better understood in the context of diverse communities of saprophytic and epiphytic bacteria.

Figure 2.

Changing phylogenetic relationships across three Xanthomonas vesicatoria (Xv1111) core effectors. Species are color-coded according to phylogenetic relationships to highlight their relative positions in each maximum likelihood tree. The effector avrBs2 in Xv1111 (bold black) is closely related to X. gardneri (Xg102, black) avrBs2, but this relationship is not observed for effectors xopL and xopN. Other abbreviations are: X. euvesicatoria (Xe, dark red), X. perforans (Xp, red), X. citri (Xc, orange), X. oryzae pv. oryzae (Xoo, brown), and X. campestris pv. campestris (Xcc, blue). Sequences are from and Potnis et al. (2011). Branch lengths are substitutions per site. Asterisks indicate bootstrap values of 100.

‘Incorporating the results of Monteil et al., one can hypothesize an “evolutionary cycle” for Pseudomonas syringae involving both environmental and plant-associated populations.’

Pathogenic strains that have adapted to the agricultural environment may disperse back to environmental populations with genes that could facilitate the emergence of other strains that acquire these genes (Fig. 1E). Horizontal gene transfer is known to introduce gene functions important for the agricultural environment into emerging strains. Copper sprays are routinely used to control bacterial pathogens in crops. In most cases, genes responsible for copper resistance (CuR) are encoded on a large conjugative plasmid with conservation of sequence identity and organization of the CuR operon in the pathogens Xanthomonas axonopodis citrumelonis and X. vesicatoria, and the epiphyte Stenotrophomonas maltophilia. Successful in vitro transfer of the CuR genes from CuR S. maltophilia to the copper sensitive pathogen X. citri (Behlau et al., 2012), suggests that plant-associated epiphytes could serve as a gene reservoir. Another example comes from X. perforans, which produces bacteriocins that inhibit growth of the close relative, X. euvesicatoria, and has displaced X. euvesicatoria from Florida tomato fields (Tudor-Nelson et al., 2003; Hert et al., 2005). Bacteriocin production is a trait that could be beneficial to bacteria in both agricultural and natural environments.

These examples suggest that the findings of Monteil et al. are not likely to be specific to P. syringae. Broad sampling of plant pathogenic bacteria may produce different interpretations of population structure and evolution than narrow sampling of diseased crops. The clonality of plant pathogenic bacteria in agroecosystems has limited our ability to detect selection on specific genes because all loci are essentially linked and thus selective sweeps affect entire genomes. Broad sampling efforts that include environmental strains should be valuable for studying the molecular evolution of bacterial plant pathogens, because the recombination in these populations breaks up linkage across the genome (e.g. Araki et al., 2006). Monteil et al. found key type III effectors of crop pathogens in their pooled environmental sample, and identified loci putatively under selection based on patterns of genetic variation. Our incomplete understanding of population structure and diversity may have caused systematic underestimation of the evolutionary potential of bacterial plant pathogens. Monteil et al.'s findings are a call to look beyond the agricultural environment to understand the evolution and emergence of bacterial plant pathogens.