Genome sequence analysis indicates that the model eukaryote Nematostella vectensis harbors bacterial consorts

Abstract

Analysis of the genome sequence of the starlet sea anemone, Nematostella vectensis, reveals many genes whose products are phylogenetically closer to proteins encoded by bacteria or bacteriophages than to any metazoan homologs. One explanation for such sequence affinities could be that these genes have been horizontally transferred from bacteria to the Nematostella lineage. We show, however, that bacterium-like and phage-like genes sequenced by the N. vectensis genome project tend to cluster on separate scaffolds, which typically do not include eukaryotic genes and differ from the latter in their GC contents. Moreover, most of the bacterium-like genes in N. vectensis either lack introns or the introns annotated in such genes are false predictions that, when translated, often restore the missing portions of their predicted protein products. In a freshwater cnidarian, Hydra, for which a proteobacterial endosymbiont is known, these gene features have been used to delineate the DNA of that endosymbiont sampled by the genome sequencing project. We predict that a large fraction of bacterium-like genes identified in the N. vectensis genome similarly are drawn from the contemporary bacterial consorts of the starlet sea anemone. These uncharacterized bacteria associated with N. vectensis are a proteobacterium and a representative of the phylum Bacteroidetes, each represented in the database by an apparently random sample of informational and operational genes. A substantial portion of a putative bacteriophage genome was also detected, which would be especially unlikely to have been transferred to a eukaryote.

Fig 1

GC contents of N. vectensis (A) and H. magnipapillata (B) contigs. Only contigs from scaffolds with at least one annotated gene are included. In panel A, the curves corresponding to Pseudomonas and Flavobacteriales contigs are linked to the right vertical axis.

Fig 2

Distributions of intron numbers in the complete set and the bacterium-like subset of genes in the N. vectensis genome database. Only genes with less than 31 predicted introns are shown.

Fig 3

Intron validation for genes located in bacterium-like scaffolds. The numbers of introns and genes in each set are indicated. In addition, 14 genes were misclassified during this analysis, as all their BLASTP matches originated from protein repeats or similar short domains.