Recombination of ecologically and evolutionarily significant loci maintains genetic cohesion in the Pseudomonas syringae species complex

Abstract

Background: Pseudomonas syringae is a highly diverse bacterial species complex capable of causing a wide range of serious diseases on numerous agronomically important crops. We examine the evolutionary relationships of 391 agricultural and environmental strains using whole-genome sequencing and evolutionary genomic analyses.

Results: We describe the phylogenetic distribution of all 77,728 orthologous gene families in the pan-genome, reconstruct the core genome phylogeny using the 2410 core genes, hierarchically cluster the accessory genome, identify the diversity and distribution of type III secretion systems and their effectors, predict ecologically and evolutionary relevant loci, and establish the molecular evolutionary processes operating on gene families. Phylogenetic and recombination analyses reveals that the species complex is subdivided into primary and secondary phylogroups, with the former primarily comprised of agricultural isolates, including all of the well-studied P. syringae strains. In contrast, the secondary phylogroups include numerous environmental isolates. These phylogroups also have levels of genetic diversity typically found among distinct species. An analysis of rates of recombination within and between phylogroups revealed a higher rate of recombination within primary phylogroups than between primary and secondary phylogroups. We also find that "ecologically significant" virulence-associated loci and "evolutionarily significant" loci under positive selection are over-represented among loci that undergo inter-phylogroup genetic exchange.

Conclusions: While inter-phylogroup recombination occurs relatively rarely, it is an important force maintaining the genetic cohesion of the species complex, particularly among primary phylogroup strains. This level of genetic cohesion, and the shared plant-associated niche, argues for considering the primary phylogroups as a single biological species.

Keywords: Comparative genomics; Pseudomonas syringae; Recombination; Species definition.

Fig. 1

Rarefaction curves for the core (a) and accessory (b) genome of P. syringae, as estimated using PanGP. a Families present in 95% (soft-core genome) and 100% (hard core genome) of P. syringae strains exponentially decays as each new genome is added to the analysis. b The total number of gene families identified continues to increase indefinitely as each new genome is added to the analysis when singleton gene families (families that are only present in one strain) are included, suggesting that P. syringae has an open pan-genome

Fig. 2

Core (a) and pan (b) genome phylogenies of Pseudomonas syringae strains. The core genome, maximum-likelihood tree was generated from a core genome alignment of the 2410 core genes present in at least 95% of the P. syringae strains analyzed in this study. The pan-genome tree was generated by hierarchical clustering of the gene content in each strain using the Jaccard coefficient method for calculating the distance between strains and the Ward hierarchical clustering method for clustering. Strain phylogroups, hosts of isolation, and whether the strain is a type or pathotype strain are shown outside the tree

Fig. 3

Prevalence of different forms of type III secretion systems (T3SSs) and phytotoxin biosynthesis genes in each of the P. syringae phylogroups. A given T3SS was considered present if all full-length, core, structural genes of the T3SS were present in the genome, while phytotoxins were considered present if more than half of the biosynthesis genes for a given phytotoxin were present in the genome

Fig. 4

Prevalence of all known type III secreted effectors (T3SEs) in each of the P. syringae phylogroups analyzed in this study. T3SEs were identified using a tblastn of 1215 experimentally verified or computationally predicted effector sequences from the BEAN 2.0 database and were considered present if a significant hit was found in the genome (E value < 1⁻⁵). Gray scaling indicates the prevalence of each T3SE family within the respective phylogroups

Fig. 5

Recombination analysis between P. syringae strains from different phylogroups (PGs). Pairwise phylogroup recombination events were normalized based on the pan-genome size, the number of strains, and the total branch length for each phylogroups pair. a Regression analysis of recombination rates and corresponding non-synonymous substitution rates (Ka). There is a significant negative log linear relationship between recombination rates and Ka for strains within the same phylogroup and between different primary phylogroups (F = 49.51, df = 30, p < 0.0001, r² = 0.6227); however, the inverse relationship exists when comparing more distantly related strains from different secondary phylogroups and strains from primary and secondary phylogroups (F = 10.58, df = 32, p = 0.0027, r² = 0.2485) b Regression analysis of recombination rates and corresponding synonymous substitution rates (Ks). The same significant negative (F = 54.53, df = 30, p < 0.0001, r² = 0.6451) and positive (F = 11.40, df = 32, p = 0.0019, r² = 0.2627) log linear relationships were observed for strains within the same phylogroup and between different primary phylogroups, and more distantly related strains from different secondary phylogroups and strains from primary and secondary phylogroups, respectively c Hierarchical clustering of homologous recombination frequency between phylogroups of the P. syringae species complex. Pairwise distances between phylogroups were calculated using the Jaccard coefficient method, based on the normalized pairwise recombination rates. Note that phylogroup 10 (PG10) is a primary phylogroup that is more closely related to phylogroups 1, 2, 3, 4, 5, and 6. Agricultural vs. Environmental labeling indicates that the bulk of the strains in these phylogroups come from these sources

Fig. 6

Relationships between inter-phylogroup recombination, virulence-association (“ecologically significant” loci), and positive selection (“evolutionarily significant” loci) for genes in P. syringae based on chi-squared proportions tests. Bars represent the percentage of total genes in each category and absolute values are inside each bar. There is no significant association between positively selected and virulence-associated genes (a). However, there is a significant positive association between gene families that have undergone inter-phylogroup recombination with virulence-associated gene families (b), positively selected gene families (c), and the small collection of gene families that are both virulence-associated and positively selected (d). The Venn diagram (e) depicts the number gene families undergoing inter-phylogroup recombination, the number of gene families that are virulence associated, and the number of gene families that are positively selected, as well as the significance of the overlap between these families