Comparative genome analysis provides insights into the evolution and adaptation of Pseudomonas syringae pv. aesculi on Aesculus hippocastanum

Abstract

A recently emerging bleeding canker disease, caused by Pseudomonas syringae pathovar aesculi (Pae), is threatening European horse chestnut in northwest Europe. Very little is known about the origin and biology of this new disease. We used the nucleotide sequences of seven commonly used marker genes to investigate the phylogeny of three strains isolated recently from bleeding stem cankers on European horse chestnut in Britain (E-Pae). On the basis of these sequences alone, the E-Pae strains were identical to the Pae type-strain (I-Pae), isolated from leaf spots on Indian horse chestnut in India in 1969. The phylogenetic analyses also showed that Pae belongs to a distinct clade of P. syringae pathovars adapted to woody hosts. We generated genome-wide Illumina sequence data from the three E-Pae strains and one strain of I-Pae. Comparative genomic analyses revealed pathovar-specific genomic regions in Pae potentially implicated in virulence on a tree host, including genes for the catabolism of plant-derived aromatic compounds and enterobactin synthesis. Several gene clusters displayed intra-pathovar variation, including those encoding type IV secretion, a novel fatty acid biosynthesis pathway and a sucrose uptake pathway. Rates of single nucleotide polymorphisms in the four Pae genomes indicate that the three E-Pae strains diverged from each other much more recently than they diverged from I-Pae. The very low genetic diversity among the three geographically distinct E-Pae strains suggests that they originate from a single, recent introduction into Britain, thus highlighting the serious environmental risks posed by the spread of an exotic plant pathogenic bacterium to a new geographic location. The genomic regions in Pae that are absent from other P. syringae pathovars that infect herbaceous hosts may represent candidate genetic adaptations to infection of the woody parts of the tree.

Figure 1. Disease symptoms of Pae on horse chestnut.

(A) Bleeding canker on stem of European horse chestnut caused by E-Pae and (B) leaf spots (arrows) on Indian horse chestnut caused by I-Pae.

Figure 2. Evolutionary relationship of P. syringae pv. aesculi to other strains of P. syringae.

Phylogenetic relationships were estimated from concatenated sequences from seven housekeeping genes (3129 bp) using a Bayesian Markov chain Monte Carlo method (See MATERIALS AND METHODS). Values in brackets indicate numbers of strains of the same pathovar with identical sequences (e.g., four strains of Pae). Red branches indicate the clade comprised of four pathovars that infect a woody host. Stars mark internal branches supported by posterior probability values of at least 0.98. The scale bar represents 0.02 nucleotide substitutions per site. Details are shown only for the clade designated as group 3 by , which corresponds to genomospecies 2 ; group 2 contains genomospecies 1 strains including P. syringae pv. syringae, group 1 contains genomospecies 3 strains including P. syringae pv. tomato, and group 4 contains probable genomospecies 4 strains including P. syringae pv. oryzae.

Figure 3. An E-Pae encoded pathway for the catabolism of plant-derived aromatic compounds.

Shown is a 20 kb section of a 46 kb E-Pae contig (GenBank: ACXT01000012) which contains putative genes encoding enzymes for the catabolism of benzoate via the catechol branch of the β-ketoadipate pathway (Indicated by red arrows). Full details of the predicted genes based on blastp searches are shown in Table 1. Regions of sequence identity with other P. syringae genomes (with a significance threshold of 1e-10) are indicated by the green bars. Grey arrows indicate uncharacterized proteins.

Figure 4. E-Pae-encoded pathways for the catabolism of plant-derived aromatic compounds and enterobactin biosynthesis.

Shown is a 27 kb E-Pae contig (Genbank: ACXT01000075) which includes genes encoding the protocatechuate 4,5-dioxygenase pathway (yellow arrows) as well as a pathway for enterobactin biosynthesis (blue arrows). Full details of the predicted genes based on blastp searches are shown in Table 2. Regions of sequence identity with other P. syringae genomes (with a significance threshold of 1e-10) are indicated by the green bars. Uncharacterized or hypothetical proteins are indicated by grey arrows.

Figure 5. An E-Pae-encoded pathway for the biosynthesis of fatty acids.

Shown is a cluster of genes in E-Pae implicated in fatty acid biosynthesis (indicated by the red arrows) with homology to Serratia proteamaculans, but which is absent in I-Pae and other P. syringae pathovars. The seven genes (A–G) occupy the entire 6.8 kb contig (GenBank: ACXT01000043). Full details of the predicted genes based on blastp searches are shown in Table 4.

Figure 6. An E-Pae-encoded pathway for the utilization of sucrose.

Shown is a cluster of genes in E-Pae which is implicated in the uptake and utilization of sucrose but which is not found in I-Pae. Details of the predicted genes based on blastp searches are shown in Table 5. This gene cluster is homologous to a region of Pph 448A (96–98% nucleotide sequence identity). In E-Pae the six principal genes (A–F) are on a 7.4 kb section of a 32.1 kb contig (GenBank: ACXT01000147.1) whereas the transcriptional regulator (G) is found at the beginning of a 12 kb contig (GenBank: ACXT01000532.1); (the first 67 codons are missing from the start of the contig). Genes in E-Pae are indicated by the blue arrows; genes in Pph 1448A are represented by the green arrows.

Figure 7. Plasmid profiles of Pae.

Agarose gel electrophoresis was carried out as described by to compare the number and size of native plasmids present within the genomes of each of the four Pae strains; P. syringae pv. phaseolicola strain 1448A was included for comparison. M represents marker plasmids from Escherichia coli strain 39R861 . Note that E-Pae strains 2250 and P6623 have two similarly sized plasmids of ca. 70 kb.