Phylogenomics of Reichenowia parasitica, an alphaproteobacterial endosymbiont of the freshwater leech Placobdella parasitica

Abstract

Although several commensal alphaproteobacteria form close relationships with plant hosts where they aid in (e.g.,) nitrogen fixation and nodulation, only a few inhabit animal hosts. Among these, Reichenowia picta, R. ornata and R. parasitica, are currently the only known mutualistic, alphaproteobacterial endosymbionts to inhabit leeches. These bacteria are harbored in the epithelial cells of the mycetomal structures of their freshwater leech hosts, Placobdella spp., and these structures have no other obvious function than housing bacterial symbionts. However, the function of the bacterial symbionts has remained unclear. Here, we focused both on exploring the genomic makeup of R. parasitica and on performing a robust phylogenetic analysis, based on more data than previous hypotheses, to test its position among related bacteria. We sequenced a combined pool of host and symbiont DNA from 36 pairs of mycetomes and performed an in silico separation of the different DNA pools through subtractive scaffolding. The bacterial contigs were compared to 50 annotated bacterial genomes and the genome of the freshwater leech Helobdella robusta using a BLASTn protocol. Further, amino acid sequences inferred from the contigs were used as queries against the 50 bacterial genomes to establish orthology. A total of 358 orthologous genes were used for the phylogenetic analyses. In part, results suggest that R. parasitica possesses genes coding for proteins related to nitrogen fixation, iron/vitamin B translocation and plasmid survival. Our results also indicate that R. parasitica interacts with its host in part by transmembrane signaling and that several of its genes show orthology across Rhizobiaceae. The phylogenetic analyses support the nesting of R. parasitica within the Rhizobiaceae, as sister to a group containing Agrobacterium and Rhizobium species.

Figure 1. Transmission electron micrograph showing the rod-shaped morphology and several cross-sections of Reichenowia parasitica.

The micrograph shows the inside of an epithelial cell of the mycetome from Placobdella parasitica at 5640x magnification, with some bacterial cells (red arrowheads), secretory esophageal cells (e), nuclei (n) and a mitochondrion (m) marked.

Figure 2. Main workflow followed in this study.

Figure 3. Single most parsimonious tree (length = 408,192 steps, consistency index = 0.647 and retention index = 0.648) recovered from the phylogenetic analysis of the 358 orthologues across 51 taxa.

The topology is identical to the maximum likelihood tree recovered by RAxML. Values above the nodes are standard bootstrap re-sampling and partition bootstrap values, respectively, and below the nodes are likelihood bootstrap values. Solid black circles denote nodes with bootstrap support ≥90% for all three support measures.

Figure 4. Comparison of Clusters of Orthologous Groups (COG's) between animal and non-animal endosymbionts.

The 358 R. parasitica orthologues, as well as the genomes of species of Agrobacterium, Mesorhizobium, Wigglesworthia, Buchnera and Escherichia were used as queries against the database. The different colors denote separate functional groups to which the genes are linked. In both of the phylogenetically related groups (left: Reichenowia, Agrobacterium and Mesorhizobium, and right: Wigglesworthia, Buchnera and Escherichia) the topmost wheels represents animal-inhabiting endosymbionts, whereas the bottommost wheels represent non-animal endosymbionts. When compared to the non-animal endosymbionts, the animal endosymbionts each show a decrease in the proportion of genes related to 1-K (transcription), and an increase in the proportion of genes related to 1-J (translation, ribosomal structure and biogenesis), 2-O (posttranslational modification, protein turnover, chaperones), and 3-F (nucleotide transport and metabolism).

Figure 5. Estimation of the genome size of Reichenowia parasitica based on Newton-Rhapson estimation on a non-linear general logistic equation.

Blue diamonds denote the general logistic equation with the asymptotic end-point being predictive of full genome size. Red squares denote the average contig size at 16.5%, 33%, 66% and 100% of the total bacterial pyrosequencing fragment pool, respectively. The estimated end-point and thus the full genome size is predicted at 2.84 Mbp.