Genome-wide evaluation and discovery of vertebrate A-to-I RNA editing sites

https://doi.org/10.1016/j.bbrc.2011.07.075Get rights and content

Abstract

RNA editing by adenosine deamination, catalyzed by adenosine deaminases acting on RNA (ADAR), is a post-transcriptional modification that contributes to transcriptome and proteome diversity and is widespread in mammals. Here we administer a bioinformatics search strategy to the human and mouse genomes to explore the landscape of A-to-I RNA editing. In both organisms we find evidence for high excess of A/G-type discrepancies (inosine appears as a guanosine in cloned cDNA) at non-polymorphic, non-synonymous codon sites over other types of discrepancies, suggesting the existence of several thousand recoding editing sites in the human and mouse genomes. We experimentally validate recoding-type A-to-I RNA editing in a number of human genes with high scoring positions including the coatomer protein complex subunit alpha (COPA) as well as cyclin dependent kinase CDK13.

Highlights

► We employ a computational pipeline to predict A-to-I RNA editing recoding sites. ► We reveal high excess of A/G-discrepancies affecting non-synonymous codon sites. ► Most recoding RNA editing events are likely subject to a low level of modification. ► We validate novel recoding events in several human genes including COPA and CDK13.

Introduction

The post-transcriptional modification of adenosine to inosine in RNA molecules is a widespread mechanism in multicellular animals for generating RNA and protein variation [1], [2]. In primates, Alu-repeats have been shown to constitute a major target of adenosine deamination [3], [4]. Furthermore, since inosine is interpreted as guanosine during translation, the site-selective alteration of single adenosines within protein-coding sequences can critically modulate protein function as a result of non-synonymous codon changes that cause amino acid substitutions [2].

A-to-I editing is mediated by adenosine deaminases acting on RNA (ADAR), of which ADAR1 and ADAR2 are thought to mediate all known editing events [1]. They recognize partially double-stranded (ds) RNA targets through several dsRNA binding domains. However, the exact mode of interaction and how site-selectivity is achieved are unknown. Repeat element mediated editing such as in Alus exhibits low specificity and is mainly driven by the strong ds-character of the RNA secondary structure. In contrast, in recoding editing single adenosine residues in pre-mRNA molecules are targeted with high specificity. It is currently not possible to predict an RNA editing site based on RNA sequence or predicted RNA secondary structure. In fact, most characterized recoding targets have been identified serendipitously.

Still, the increasing number of validated RNA editing target sequences and secondary structures reveal a bias toward certain molecular characteristics. For example, the local sequence environment influences whether editing occurs by ADAR1 or ADAR2, respectively, due to enzyme-specific preferences for certain bases preceding and following the targeted A. Another critical feature is a partially base-paired RNA secondary structure involving sequences flanking the editing site(s) [1]. Also, such sequences are more highly conserved across species since their involvement in forming a functional RNA secondary structure exerts increased selection pressure [5].

Even though above features contribute to establishing an editing target, none is sufficient to allow for straight-forward prediction of bona fide editing sites. Despite these limitations, several new targets of editing have recently been identified using bioinformatics strategies that employ a combination of these molecular features as filter criteria [6], [7], [8]. In addition, high-throughput sequencing has also helped to validate potential cases of editing [9], [10]. However, the ad hoc nature of previously adopted bioinformatics strategies and the observation that very little if any overlap between predicted candidate lists emerged strongly limits the genome-wide evaluation of the prevalence and importance of A-to-I editing.

We developed the bioinformatics-based search tool RNA Editing Dataflow System (REDS) that allows for a more comprehensive screening, which, combined with standard or high-throughput experimental validation, would facilitate mapping the A-to-I RNA editing landscape and defining the overall impact of editing on gene expression. We recently explored the feasibility of our search strategy using a subset of human genes and experimentally confirmed several predicted novel editing sites [7]. Here, we applied our search strategy to the genomic scale to analyze the overall landscape of base discrepancies in two species followed by experimental validation of novel recoding editing sites in the human transcriptome.

Section snippets

Evidence for abundant, site-selective recoding A-to-I editing in human and mouse

In eukaryotes, A-to-I RNA editing is the only known mechanism for generating inosine residues in RNA molecules. First we asked whether an excess of A/G discrepancies versus other types of base differences between cDNA and genomic DNA is detectable, even when excluding repetitive element mediated editing. Such a finding would support the hypothesis that many more editing sites within protein-coding sequences remain to be identified and may provide an estimate on the total number of existing

Databases and REDS

Human, mouse and zebrafish genomic DNA sequences, mRNA data files, SNP data, as well as table annotations were retrieved using the UCSC genome browser ftp site (assembly March 2006) [11]. REDS consists of three consecutive computational stages (Supplemental Fig. S1). Stage 1 aligns expressed sequences (UCSC mRNA database) from a given species to the corresponding genomic sequence. Coding sequences are translated, allowing for determination of non-synonymous codon positions within ORFs.

Acknowledgments

We thank Dr. William Coleman for mouse specimens. We are grateful to Adrienne LaFleur, Mark Strohmaier and Emaan Abdul-Majid for technical assistance.

References (18)

  • S. Farajollahi et al.

    Molecular diversity through RNA editing: a balancing act

    Trends Genet.

    (2010)
  • R. Beck et al.

    The COPI system: molecular mechanisms and function

    FEBS Lett.

    (2009)
  • B.L. Bass

    RNA editing by adenosine deaminases that act on RNA

    Annu. Rev. Biochem.

    (2002)
  • A. Athanasiadis et al.

    Widespread A-to-I RNA editing of Alu-containing mRNAs in the human transcriptome

    PLoS Biol.

    (2004)
  • E.Y. Levanon et al.

    Systematic identification of abundant A-to-I editing sites in the human transcriptome

    Nat. Biotechnol.

    (2004)
  • B. Hoopengardner et al.

    Nervous system targets of RNA editing identified by comparative genomics

    Science

    (2003)
  • D.R. Clutterbuck et al.

    A bioinformatic screen for novel A-I RNA editing sites reveals re-coding editing in BC10

    Bioinformatics

    (2005)
  • W.M. Gommans et al.

    Screening of human SNP database identifies recoding sites of A-to-I RNA editing

    RNA

    (2008)
  • E.Y. Levanon et al.

    Evolutionarily conserved human targets of adenosine to inosine RNA editing

    Nucleic Acids Res.

    (2005)
There are more references available in the full text version of this article.

Cited by (26)

  • Using mouse models to unlock the secrets of non-synonymous RNA editing

    2019, Methods
    Citation Excerpt :

    For example, A-to-I editing of KCNA1 converts an isoleucine to valine in the voltage-gated potassium channel KV1.1, generating a channel with an inactivation rate ∼20 times greater than in the non-edited channel [14,15]. Large-scale transcriptome information has advanced the RNA editing field by providing wide-angle knowledge regarding the identity of RNA editing substrates [11,13,16–19], tissue- and disease-specific RNA editing profiles [10,20–23], and the conservation of edited substrates [14,24–26]. Heterologous expression systems and animal models, particularly mouse, are useful for determining the function of RNA editing-dependent protein isoforms.

  • The molecular link between inefficient GluA2 Q/R site-RNA editing and TDP-43 pathology in motor neurons of sporadic amyotrophic lateral sclerosis patients

    2014, Brain Research
    Citation Excerpt :

    These results indicate that ADAR2 downregulation is a selective, profound and disease-specific pathological change in motor neurons from sporadic ALS patients. Recently, by using global genome wide and RNA sequencing analysis, it is now recognized that ADAR2 is involved in the editing of a large number of RNAs including GluA2 (Cattenoz et al., 2013; Li et al., 2009; Maas et al., 2011; Peng et al., 2012; Zhu et al., 2013). The reduction of ADAR2 in motor neurons from ALS patients may induce various defects, including the potential alteration in the role for RNA splicing (Solomon et al., 2013) and control of neuronal excitability (Li et al., 2013).

  • Base modification RNA Editing: Information recoding on the fly

    2012, Seminars in Cell and Developmental Biology
  • Somatic hypermutation and the discovery of A-to-I RNA editing sites?

    2011, Biochemical and Biophysical Research Communications
View all citing articles on Scopus
View full text