Comparative analysis of RNA families reveals distinct repertoires for each domain of life

Abstract

The RNA world hypothesis, that RNA genomes and catalysts preceded DNA genomes and genetically-encoded protein catalysts, has been central to models for the early evolution of life on Earth. A key part of such models is continuity between the earliest stages in the evolution of life and the RNA repertoires of extant lineages. Some assessments seem consistent with a diverse RNA world, yet direct continuity between modern RNAs and an RNA world has not been demonstrated for the majority of RNA families, and, anecdotally, many RNA functions appear restricted in their distribution. Despite much discussion of the possible antiquity of RNA families, no systematic analyses of RNA family distribution have been performed. To chart the broad evolutionary history of known RNA families, we performed comparative genomic analysis of over 3 million RNA annotations spanning 1446 families from the Rfam 10 database. We report that 99% of known RNA families are restricted to a single domain of life, revealing discrete repertoires for each domain. For the 1% of RNA families/clans present in more than one domain, over half show evidence of horizontal gene transfer (HGT), and the rest show a vertical trace, indicating the presence of a complex protein synthesis machinery in the Last Universal Common Ancestor (LUCA) and consistent with the evolutionary history of the most ancient protein-coding genes. However, with limited interdomain transfer and few RNA families exhibiting demonstrable antiquity as predicted under RNA world continuity, our results indicate that the majority of modern cellular RNA repertoires have primarily evolved in a domain-specific manner.

Figure 1. Venn diagram of RNA family distribution.

Taxonomic information attached to EMBL-derived Rfam annotations reveals that the majority (99%) of RNA families are domain-specific, with only seven RNA families universally conserved (across the three domains of life plus viruses; Table S1 in Text S1). Numbers within dashed circles indicate viral RNA families.

Figure 2. RNA-based processes traceable to the Last Universal Common Ancestor.

Universal Rfam families that show evidence of vertical inheritance (Table S1 in Text S1) are all associated with the processes of translation (rRNAs, tRNAs, RNase P) and protein export (SRP RNA). A previous study examining the antiquity of protein coding genes identified only 37 universally distributed proteins which show evidence of vertical inheritance. The majority of these vertically inherited proteins are associated with translation and protein export; numbers of such proteins associated with each of the depicted processes is given in grey (original data are from Harris [21]). The proteins associated with RNase P are not universally conserved, with archaeal and eukaryotic RNase P proteins being unrelated to their bacterial counterparts . While tRNA synthetases are universal, they have undergone ancient horizontal gene transfer events , which complicates establishing the timing of their origin.

Figure 3. Reconstruction of broadly distributed RNA repertoires for each domain, plus interdomain RNA families.

Colored bars at far right indicate normalized taxonomic abundance of each Rfam for major taxonomic groupings within each domain. Horizontal traces (see text, Table S1 in Text S1) for interdomain families, are depicted as follows: general transfer patterns are given by dashed arrows; proposed HGT patterns for individual families are depicted by number (inset). For Rfam families present in more than one domain (far left and inset), bars indicate normalized taxonomic abundance by domain (color scheme at bottom left). Asterisks indicate additional broadly-distributed bacterial candidates identified using GEBA tree topology (see text). Note that the Rfam rRNA families in Rfam 10.0 are based on conserved subsequences, and are not as comprehensive as other resources (see main text) and are included here for consistency. The universally-distributed rRNAs are the small subunit (16/18S) rRNA, large subunit (23/28S) rRNA and 5S rRNA (see Table S1 in Text S1). The 5.8S rRNA of eukaryotes is known to be homologous to the 5′ end of bacterial and archaeal 23S rRNA , , so its inclusion as a eukaryote-specific family in Rfam is in this respect artefactual.

Figure 4. Rfam-based functional classification of RNA families.

The tree depicts classification of the higher level data structures within Rfam, and is not a phylogeny. Numbers of sequences and families in Rfam 10 that fall into each functional classification are shown as bar charts. Domain-level taxonomic distribution for each functional category is shown by black (present) and white (absent) boxes, right. The grey box indicates that H/ACA family RNAs are known from archaea , , but are not in Rfam 10.