Two nucleotides immediately upstream of the essential A(6)G(3) slippery sequence modulate the pattern of g insertions during sendai virus mRNA editing

Editing of paramyxovirus P gene mRNAs occurs cotranscriptionally and functi ons to fuse an alternate downstream open reading frame to the N-terminal ha lf of the P protein, G residues are inserted into a short G run contained w ithin a larger purine run (A(n)G(n)) in this process, by a mechanism whereb y the transcribing polymerase stutters (i.e., reads the same template cytos ine more than once). Although Sendai virus (SeV) and bovine parainfluenza v irus type 3 (bPIV3) are closely related, the G insertions in their P mRNAs are distributed differently. SeV predominantly inserts a single G residue w ithin the G run of the sequence 5' AACAAA AAAGGG, whereas bPIV3 inserts one to six G's at roughly equal frequency within the sequence 5' AUUAAA AAAGGG G (differences are underlined). We have examined how the cis-acting editing sequence determines the number of G's inserted, both in a transfected cell system using minigenome analogues and by generating recombinant viruses. W e found that the presence of four rather than three G's in the purine run d id not affect the distribution of G insertions. However, when the underline d AC of the SeV sequence was replaced by the UU found in bPIV3, the editing phenotype from both the minigenome and the recombinant virus resembled tha t found in natural bPIV3 infections (i.e., a significant fraction of the mR NAs contained two to six G insertions). The two nucleotides located just up stream of the polypurine tract are thus key determinants of the editing phe notype of these viruses. Moreover, the minimum number of A residues that wi ll promote SeV editing phenotype is six but can be reduced to five when the upstream AC is replaced by UU, A model for how the upstream dinucleotide c ontrols the insertion phenotype is presented.