Editing inducer elements increases A-to-I editing efficiency in the mammalian transcriptome
Daniel et al. Genome Biology
Editing inducer elements increases A-to-I editing efficiency in the mammalian transcriptome
Chammiran Daniel 0
Albin Widmark 0
Ditte Rigardt 0
Marie Öhman 0
0 Department of Molecular Biosciences, The Wenner-Gren Institute, Stockholm University , Svante Arrheniusväg 20C, 10691 Stockholm , Sweden
Background: Adenosine to inosine (A-to-I) RNA editing has been shown to be an essential event that plays a significant role in neuronal function, as well as innate immunity, in mammals. It requires a structure that is largely double-stranded for catalysis but little is known about what determines editing efficiency and specificity in vivo. We have previously shown that some editing sites require adjacent long stem loop structures acting as editing inducer elements (EIEs) for efficient editing. Results: The glutamate receptor subunit A2 is edited at the Q/R site in almost 100% of all transcripts. We show that efficient editing at the Q/R site requires an EIE in the downstream intron, separated by an internal loop. Also, other efficiently edited sites are flanked by conserved, highly structured EIEs and we propose that this is a general requisite for efficient editing, while sites with low levels of editing lack EIEs. This phenomenon is not limited to mRNA, as non-coding primary miRNAs also use EIEs to recruit ADAR to specific sites. Conclusions: We propose a model where two regions of dsRNA are required for efficient editing: first, an RNA stem that recruits ADAR and increases the local concentration of the enzyme, then a shorter, less stable duplex that is ideal for efficient and specific catalysis. This discovery changes the way we define and determine a substrate for A-to-I editing. This will be important in the discovery of novel editing sites, as well as explaining cases of altered editing in relation to disease.
RNA editing; Adenosine deamination; Glutamate receptor; ADAR; EIE; miRNA
Background
Complex organisms require a great diversity of gene
products for proper development and function,
particularly in the brain. This is achieved by the use of
numerous co- or post-transcriptional processes, such as
alternative splicing, alternative polyadenylation, and
RNA editing. Adenosine-to-inosine (A-to-I) RNA editing
is a highly conserved RNA modification process that
occurs in all metazoan lineages [
1
]. Inosine base pairs with
C and is interpreted as G by the cellular machineries.
Hence, A-to-I RNA editing can be designated as an
A-to-G conversion and, if situated in exonic sequence, it
has the potential to alter codons and consequently
contribute to the expression of additional protein isoforms
(reviewed in [
2
]). A-to-I conversions within introns and 3′
UTRs can also have an impact on the transcriptome, e.g.,
by creating new splice sites and changing miRNA target
recognition. A-to-I editing is essential to the organism
and aberrant editing has been linked to a variety of
different human diseases: amyotrophic lateral sclerosis (ALS)
and other neurological disorders, several types of cancer,
and autoimmune disorders such as the Aicardi-Goutières
syndrome (AGS) [
3–6
]. To understand what determines
the level of editing in different substrates and under
different circumstances, we need to know the mechanism of
substrate recognition. It is, however, still largely unclear
what factors determine the efficiency of editing.
A-to-I RNA editing is performed by the adenosine
deaminases that act on RNA (ADAR) enzymes that
recognize adenosines located in double-stranded RNA
(dsRNA) to be deaminated into inosines [
7
]. ADAR
proteins are evolutionarily conserved in metazoans and
mammals have two enzymatically active ADAR enzymes,
ADAR1 and ADAR2 [
8–10
]. In some cases, the
substrate selectivity of the two enzymes overlaps, but more
commonly the targets are specific for either enzyme
[
11–13
]. ADAR1 and ADAR2 share certain domain
structures, such as the deaminase domain and the
double-stranded RNA binding domains (dsRBDs).
However, the numbers of dsRBDs differ between the two
enzymes (ADAR1 contains three while ADAR2 contains
two) as well as the spacing between them. The dsRBDs
recognize one face of the sugar backbone of an A-form
helix, such as the RNA duplex, spanning two minor
grooves and an intervening major groove [
14
]. Thus,
there is little sequence specificity via interaction of the
dsRBDs and theoretically they can interact with any
double-stranded RNA longer than 16 nucleotides (nt).
However, sequence-specific interactions between the two
dsRBDs of human ADAR2 at the GluA2 stem loop at
the R/G site have been reported based on the NMR
structure [
15
]. Interestingly, it has recently been shown
that the deaminase domain also requires a
doublestranded structure in order to interact with the substrate
and perform the catalysis [
16, 17
].
In general, there are two categories of A-to-I RNA
editing determined by the structure of the RNA. Long
double-stranded structu (...truncated)