Evolutionary Time-Scale of the Begomoviruses: Evidence from Integrated Sequences in the Nicotiana Genome

Pierre Lefeuvre,Gordon W Harkins,Jean-Michel Lett,Mark W Chase,Benoit Moury,Darren P Martin,Rob W Briddon,Sergios-Orestis Kolokotronis

doi:10.1371/journal.pone.0019193

Abstract

Despite having single stranded DNA genomes that are replicated by host DNA polymerases, viruses in the family Geminiviridae are apparently evolving as rapidly as some RNA viruses. The observed substitution rates of geminiviruses in the genera Begomovirus and Mastrevirus are so high that the entire family could conceivably have originated less than a million years ago (MYA). However, the existence of geminivirus related DNA (GRD) integrated within the genomes of various Nicotiana species suggests that the geminiviruses probably originated >10 MYA. Some have even suggested that a distinct New-World (NW) lineage of begomoviruses may have arisen following the separation by continental drift of African and American proto-begomoviruses ∼110 MYA. We evaluate these various geminivirus origin hypotheses using Bayesian coalescent-based approaches to date firstly the Nicotiana GRD integration events, and then the divergence of the NW and Old-World (OW) begomoviruses. Besides rejecting the possibility of a<2 MYA OW-NW begomovirus split, we could also discount that it may have occurred concomitantly with the breakup of Gondwanaland 110 MYA. Although we could only confidently narrow the date of the split down to between 2 and 80 MYA, the most plausible (and best supported) date for the split is between 20 and 30 MYA – a time when global cooling ended the dispersal of temperate species between Asia and North America via the Beringian land bridge.

Highlights

Virus populations are generally dynamic ensembles of individual lineages, each with distinctive genetic diversities, geographical distributions and host ranges
Due to the strong possibility that recombination confounded our analysis to a greater degree than that of Duffy and Holmes [11], we suggest that their short-term substitution’’ (STS) rate estimate is more accurate than ours and, that the date of the OW-NW begomovirus split inferred from their STS rates, between 722– 376 670 YA (95% highest probability density (HPD)), is probably more reliable than ours
One important finding is that the age of the OW-NW begomovirus most recent common ancestor (MRCA) cannot be accurately inferred based on short-term nucleotide substitution rates estimated from sequences sampled only in the last 30 years

Summary

Introduction

Virus populations are generally dynamic ensembles of individual lineages, each with distinctive genetic diversities, geographical distributions and host ranges. Different approaches can be used to infer the rate at which mutations arise (the mutation rate) and become fixed within a population (the substitution rate) during virus evolution (reviewed in [6,7]) These include: (i) the direct estimation of mutation and substitution rates using experiments where virus genomes with either known or accurately inferable sequences are used to initiate infections and are compared with virus sequences subsequently sampled from these infections over time-periods ranging from days to years [8,9,10]; (ii) the direct estimation of substitution rates from natural sequences sampled over extended time periods by correlating degrees of divergence from ancestral sequences with sampling dates [11,12]; (iii) the indirect estimation of substitution rates by associating the phylogenetic divergence of virus lineages that have different host or geographical ranges with either the known divergence dates of their host species (so-called virus-host co-divergence) or the dates when the geographical ranges of virus lineages diverged (for example, when one of the lineages was either introduced into a new isolated environment such as an island or a different continent) [13,14,15]; and (iv) the indirect estimation of substitution rates based on the relationships of contemporary virus sequences to those of ‘‘fossil’’ virus sequences that have become integrated into host genomes [7,16,17,18,19]

Methods

Results

Conclusion