ANRIL/CDKN2B-AS shows two-stage clade-specific evolution and becomes conserved after transposon insertions in simians

Sha He,Yize Li,Weiling Gu,Hao Zhu

doi:10.1186/1471-2148-13-247

Abstract

BackgroundMany long non-coding RNA (lncRNA) genes identified in mammals have multiple exons and functional domains, allowing them to bind to polycomb proteins, DNA methyltransferases, and specific DNA sequences to regulate genome methylation. Little is known about the origin and evolution of lncRNAs. ANRIL/CDKN2B-AS consists of 19 exons on human chromosome 9p21 and regulates the expression of three cyclin-dependent kinase inhibitors (CDKN2A/ARF/CDKN2B).ResultsANRIL/CDKN2B-AS originated in placental mammals, obtained additional exons during mammalian evolution but gradually lost them during rodent evolution, and reached 19 exons only in simians. ANRIL lacks splicing signals in mammals. In simians, multiple transposons were inserted and transformed into exons of the ANRIL gene, after which ANRIL became highly conserved. A further survey reveals that multiple transposons exist in many lncRNAs.ConclusionsANRIL shows a two-stage, clade-specific evolutionary process and is fully developed only in simians. The domestication of multiple transposons indicates an impressive pattern of “evolutionary tinkering” and is likely to be important for ANRIL’s structure and function. The evolution of lncRNAs and that of transposons may be highly co-opted in primates. Many lncRNAs may be functional only in simians.

Highlights

Many long non-coding RNA genes identified in mammals have multiple exons and functional domains, allowing them to bind to polycomb proteins, DNA methyltransferases, and specific DNA sequences to regulate genome methylation
The macaque genome was chosen because it is neither so close to the human genome that the resulting covariance models (CMs) would be overly specific nor so distant from the human genome that the structure and sequence information contained in human exons would be weakened or blurred, ensuring that truly putative exons could be identified in the target genomes
To confirm that the failure to detect exons in non-placental mammals and vertebrates was not due to the evolutionary distance between human/macaque and these organisms, we built CMs based on the identified exons in rabbit and horse and re-searched these CMs against the genomes of opossum and chicken

Summary

Introduction

Many long non-coding RNA (lncRNA) genes identified in mammals have multiple exons and functional domains, allowing them to bind to polycomb proteins, DNA methyltransferases, and specific DNA sequences to regulate genome methylation. X-chromosome inactivation in female placental mammals, which causes the products of genes on the X chromosome to have equal dosages in males and females, is controlled by a set of long non-coding RNA (lncRNA) genes, including Xist, Tsix, Jpx, and Tsx (reviewed in [1,2]). The imprinted expression of some mammalian genes is controlled by lncRNAs (reviewed in [3,4]). Many lncRNAs are located antisense to the genes they regulate; typical examples include AIRN, H19, and Kcnq1ot, which control the imprinted expression of Igf, Igf2r, and Kcnq1 [17,18]. It is relatively easy to experimentally identify antisense lncRNAs; few in-depth analyses have been performed

Methods

Results

Discussion

Conclusion