Modelling mutational and selection pressures on dinucleotides in eukaryotic phyla –selection against CpG and UpA in cytoplasmically expressed RNA and in RNA viruses

BMC Genomics

Research Article

J Kenneth Baillie¹  Peter Simmonds²  Wenjun Xia³  Ken McKinnon³ 

[1] Division of Genetics and Genomics and Roslin Institute, University of Edinburgh, Easter Bush, EH25 9RG, Edinburgh, UK;Division of Infection and Immunity, Roslin Institute, University of Edinburgh, Easter Bush, EH25 9RG, Edinburgh, UK;School of Mathematics, University of Edinburgh, James Clerk Maxwell Building, The King’s Buildings, EH9 3JZ, Edinburgh, UK;

关键词: Dinucleotide;    Methylation;    Exon;    RNA virus;    Markov modelling;   

DOI  :  10.1186/1471-2164-14-610

 received in 2013-06-20, accepted in 2013-09-04,  发布年份 2013

来源: Springer

【 摘 要 】

BackgroundLoss of CpG dinucleotides in genomic DNA through methylation-induced mutation is characteristic of vertebrates and plants. However, these and other eukaryotic phyla show a range of other dinucleotide frequency biases with currently uncharacterized underlying mutational or selection mechanisms. We developed a parameterized Markov process to identify what neighbour context-dependent mutations best accounted for patterns of dinucleotide frequency biases in genomic and cytoplasmically expressed mRNA sequences of different vertebrates, other eukaryotic groups and RNA viruses that infect them.ResultsConsistently, 11- to 14-fold greater frequencies of the methylation-associated mutation of C to T upstream of G (depicted as C→T,G) than other transitions best modelled dinucleotide frequencies in mammalian genomic DNA. However, further mutations such as G→T,T (5-fold greater than the default transversion rate) were required to account for the full spectrum of dinucleotide frequencies in mammalian sequence datasets. Consistent with modeling predictions for these two mutations, instability of both CpG and CpT dinucleotides was identified through SNP frequency analysis of human DNA sequences. Different sets of context-dependent mutations were modelled in other eukaryotes with non-methylated genomic DNA. In contrast to genomic DNA, best-fit models of dinucleotide frequencies in transcribed RNA sequences expressed in the cytoplasm from all organisms were dominated by mutations that eliminated UpA dinucleotides, observations consistent with cytoplasmically driven selection for mRNA stability. Surprisingly, mRNA sequences from organisms with methylated genomes showed evidence for additional selection against CpG through further context-dependent mutations (eg. C→A,G). Similar mutation or selection processes were identified among single-stranded mammalian RNA viruses; these potentially account for their previously described but unexplained under-representations of CpG and UpA dinucleotides.ConclusionsMethods we have developed identify mutational processes and selection pressures in organisms that provide new insights into nucleotide compositional constraints and a wealth of biochemical and evolutionarily testable predictions for the future.

【 授权许可】

CC BY   
© Simmonds et al.; licensee BioMed Central Ltd. 2013