Several mammalian neuropeptide genes are readthrough candidates

1
150 fly genes appear to undergo stop codon
readthrough
We searched for regions showing clear
evolutionary signatures of protein-coding
sequences immediately downstream of D.
melanogaster stop codons.
We identified 150 strong candidates, where we
believe translational readthrough is the best
inference from the available evidence.

• We took steps to eliminate obvious alternative
  explanations
• We eliminated recent nonsense mutations (stop
  codon in D. melanogaster only)
• We searched for high-scoring splice acceptor
  sites in the readthrough regions. Canonical
  splicing mechanisms may explain only a few of the
  observed examples.

• Other properties of the group suggest a conserved
  mechanism
• 95 of the putative readthrough stop codons are
  perfectly conserved across all aligned species,
  while it is common for normal stop codons to
  wobble or move around
• The readthrough stop codons are enriched for
  conserved RNA structures in the vicinity, as
  predicted by EvoFold (J.S. Pedersen)

• The readthrough genes do not appear to be
  selenoproteins
• We found no convincing examples of SECIS
  elements governing selenocysteine recoding
• Most (68) of the putative readthrough stop
  codons are TGA, but 32 are not

• The readthrough genes are enriched for nervous
  system function
• Enriched Gene Ontology terms
• neurogenesis (p1.8e-8)
• transmission of nerve impulse (p4.8e-6)
• Enriched tissues (from ImaGO in situ data)
• ventral nerve cord (p1.6e-5)
• central nervous system (p4.2e-5)

We hypothesize A?I editing as a possible
causative mechanism, based on tissue and RNA
structure enrichment.
Translation of the potassium channel CG12904
appears to bypass two stop codons
Several mammalian neuropeptide genes are
readthrough candidates
OPRK1 (Kappa-type opioid receptor)
The fly likely has many more polycistronic
transcripts than currently known
Polycistronic transcripts are single processed
transcripts containing several disjoint ORFs that
are each translated into separate polypeptides.
We searched for candidate polycistronic
transcripts by identifying high-scoring ORFs
within annotated UTRs of existing transcripts.
Evidence for several programmed translational
frameshifts
We rediscovered 85 of 115 annotated dicistronic
transcripts (73) and predict an additional 135
ORFs in 123 genes.
We searched for candidate translational
frameshifts by looking for adjacent windows that
score highly in different reading frames. We
found four examples where a programmed
frameshift appears to be the best explanation
based on available data.
A candidate translational frameshift has a
striking association with a highly conserved RNA
structure