Analysis of the gene start and gene end signals of human respiratory syncytial virus: quasi-templated initiation at position 1 of the encoded mRNA.

The gene start (GS) and gene end (GE) transcription signals of human respiratory syncytial virus (RSV) strain A2 were analyzed in helper-dependent monocistronic and dicistronic minireplicons which were complemented by a standard RSV strain. The GS signal, which is the start site for mRNA synthesis, is highly conserved for the first nine genes: 3'-CCCCGUUUA(U/C) (negative sense). This conserved version of the signal was analyzed by "saturation" mutagenesis, in which all 10 positions, as well as one downstream and one upstream position, were changed one at a time into each of the other three nucleotides. Most of the positions appear to contribute to the signal: positions 1, 3, 6, 7, and, in particular, 9 were the most sensitive, whereas position 5 was relatively insensitive. The effect of nucleotide substitution in the first position of the signal was examined further by cDNA cloning and sequence analysis of the residual mRNA which was produced. For the two mutants examined (1C to U, and 1C to A), the site of initiation was unchanged. However, the mRNAs were dimorphic with regard to the assignment of the 5'-terminal nucleotide: two-thirds contained the predicted mutant substitution, and one-third contained the parental assignment. Intracellular minigenome contained only the mutant assignment, indicating that the heterogeneity was at the level of transcription by the RSV polymerase. This suggests that the templated mutant assignment at position 1 can sometimes be overridden by an innate preference for the parental assignment, a phenomenon which we dubbed quasi-templated initiation. The GS signal of the L gene, encoding the 10th RSV mRNA, contains three differences (3'-CCCUGUUUUA) compared to the conserved version. It was shown to be equal in efficiency to the conserved version. This was unexpected, since the saturation mutagenesis described above indicated that U in place of A at position 9 should be highly inhibitory. Instead, the A at position 10 of the L GS signal was found to be critical for activity, indicating that an essential A residue indeed was present in both versions of the GS signal but that its spacing differed. The GE signal, which directs termination and polyadenylation, has more sequence diversity in nature than does the GS signal. The naturally occurring GE signals of strain A2 were compared by their individual incorporation into a dicistronic minigenome. They were similar in the ability to produce translatable mRNA except in the cases of NS1 and NS2, which were approximately 60% as efficient.