A natural barrier to lateral gene transfer from prokaryotes to eukaryotes revealed from genomes: the 70 % rule

Abstract

Background: The literature harbors many claims for lateral gene transfer (LGT) from prokaryotes to eukaryotes. Such claims are typically founded in analyses of genome sequences. It is undisputed that many genes entered the eukaryotic lineage via the origin of mitochondria and the origin of plastids. Claims for lineage-specific LGT to eukaryotes outside the context of organelle origins and claims of continuous LGT to eukaryotic lineages are more problematic. If eukaryotes acquire genes from prokaryotes continuously during evolution, then sequenced eukaryote genomes should harbor evidence for recent LGT, like prokaryotic genomes do.

Results: Here we devise an approach to investigate 30,358 eukaryotic sequences in the context of 1,035,375 prokaryotic homologs among 2585 phylogenetic trees containing homologs from prokaryotes and eukaryotes. Prokaryote genomes reflect a continuous process of gene acquisition and inheritance, with abundant recent acquisitions showing 80-100 % amino acid sequence identity to their phylogenetic sister-group homologs from other phyla. By contrast, eukaryote genomes show no evidence for either continuous or recent gene acquisitions from prokaryotes. We find that, in general, genes in eukaryotic genomes that share ≥70 % amino acid identity to prokaryotic homologs are genome-specific; that is, they are not found outside individual genome assemblies.

Conclusions: Our analyses indicate that eukaryotes do not acquire genes through continual LGT like prokaryotes do. We propose a 70 % rule: Coding sequences in eukaryotic genomes that share more than 70 % amino acid sequence identity to prokaryotic homologs are most likely assembly or annotation artifacts. The findings further uncover that the role of differential loss in eukaryote genome evolution has been vastly underestimated.

Fig. 1

Identification of clades and sister groups. For each tree, largest possible clade(s) and their respective sister group(s) are identified for different taxonomic groups (e.g., eukaryotes or bacilli). One (a, b) or more (c, d) clades can be present for a single taxonomic group, with close (a), divergent (b, c), or both close and divergent sister groups

Fig. 2

Phylogenomic dissection of major prokaryotic groups. All largest possible clades are plotted for each taxonomic group. y-axis: average sequence identity between a clade and its sister group (I _C-S); x-axis: number of taxa (species in bacteria or genomes in archaea). A horizontal reference line is drawn corresponding to the average of the singleton I _C-S greater than or equal to their third quartile. a–f Bacterial groups. g–h Archaeal groups

Fig. 3

Phylogenomic dissection of eukaryotes. All largest possible eukaryotic clades are plotted. y-axis: average sequence identity between a clade and its sister group (I _C-S); x-axis: number of species. A horizontal reference line is drawn corresponding to the average of the singleton I _C-S greater than or equal to their third quartile. a All clades. b Clades of plastid origin (shown in green in a) are selectively removed

Fig. 4

Eukaryotes have relatively fewer non-singleton high-identity clades (HICs). Taxonomic groups are plotted according to their ratio of non-singleton HICs to singleton HICs against their number of taxa. Red: eukaryotes with all clades (Fig. 3a) or with clades of plastid origin removed (Fig. 3b); blue: prokaryotic groups based on the original eukaryotic-prokaryotic clusters (Fig. 2; Additional file 3: Figure S2); green: prokaryotic groups based on clusters generated using the same clustering procedure as for eukaryotes (Additional file 8: Figure S3)

Fig. 5

Close-up of the distribution of small-sized high-identity clades (HICs). HICs with up to one-third of the total taxa are shown for each group in Figs. 2 and 3 (with x-axis plotted to the same scale for each group). a–h Prokaryotic groups. i All eukaryote clades. j Eukaryotes with clades of plastid origin (shown in green in i) selectively removed. The seven proteins having >70 % sequence identity to prokaryotic homologs but appearing in more than one eukaryotic genome are annotated as (from left to right): fructose-bisphosphate aldolase, unknown (carbohydrate transport and metabolism), homocitrate synthase, component of cytochrome b6f complex, ribulose-phosphate 3-epimerase, pyridoxal biosynthesis, and adenosylhomocysteinase

Fig. 6

Distribution of clades in the phylogenomic space. a Seven representative clades are plotted in the phylogenomic space with clade-sister identity as the y-axis and clade size as the x-axis. b Phylogenetic trees corresponding to the seven clades illustrate the effects of lineage diversification (a–d), sequence divergence (a–g), and differential gene loss (d–g)