Adenosine deaminases that act on RNA (ADARs) are
RNA editing enzymes that convert adenosines to inosines
within cellular and viral RNAs. Certain glutamate receptor
(gluR) pre-mRNAs are substrates for the enzymes in vivo.
For example, at the R/G editing site of gluR-B, -C, and
-D RNAs, ADARs change an arginine codon (AGA) to a glycine
codon (IGA) so that two protein isoforms can be synthesized
from a single encoded mRNA; the highly related gluR-A sequence
is not edited at this site. To gain insight into what features
of an RNA substrate are important for accurate and efficient
editing by an ADAR, we performed a phylogenetic analysis
of sequences required for editing at the R/G site. We observed
highly conserved sequences that were shared by gluR-B,
-C, and -D, but absent from gluR-A. Surprisingly, in contrast
to results obtained in phylogenetic analyses of tRNA and
rRNA, it was the bases in paired, helical regions whose
identity was conserved, whereas bases in nonhelical regions
varied, but maintained their nonhelical state. We speculate
this pattern in part reflects constraints imposed by ADAR's
unique specificity and gained support for our hypotheses
with mutagenesis studies. Unexpectedly, we observed that
some of the gluR introns were conserved beyond the sequences
required for editing. The ∼600-nt intron 13 of gluR-C
was particularly remarkable, showing >94% nucleotide
identity between human and chicken, organisms estimated
to have diverged 310 million years ago.