Promoter-proximal poly(A) sites are processed efficiently, but the RNA products are unstable in the nucleus.

The presence of two polyadenylation signals in the primary transcript of the human immunodeficiency virus type 1 (HIV-1) provirus leads to a requirement for regulation of 3'-end processing. To ensure that viral genome replication and gene expression occur, polyadenylation must occur at the poly(A) site transcribed from the 3' long terminal repeat (LTR) but not the 5' LTR. Models that have been proposed to explain this regulation include (i) inhibition of the 5' site as a result of proximity to the promoter and (ii) enhancement of the 3' site by U3 sequences that are transcribed upstream of only the 3' poly(A) site. In previous studies designed to investigate these models, a reduction in the levels of steady-state RNA was observed when the HIV-1 poly(A) site was placed within 500 nucleotides of the cap site. Although these findings were interpreted to be the result of promoter proximity effects on 3'-end processing, in vitro studies demonstrated that the HIV-1 poly(A) site was equally functional in promoter-proximal and promoter-distal positions. These results led to the hypothesis that, in vivo, the poly(A) site is fully active at this close distance but the short transcripts produced are highly unstable in the nucleus and undergo rapid degradation, precluding their appearance as abundant mRNAs in the steady-state pool. To investigate the biogenesis of these short RNAs in vivo, experiments were performed to examine directly the nuclear processing rates of the HIV-1 poly(A) site in intact cells. By using recombinant adenoviruses as expression vectors, it is now demonstrated conclusively that the HIV-1 poly(A) site is indeed processed at equivalent levels independent of its distance from the promoter. Although transcripts containing the promoter-proximal poly(A) site are processed efficiently, most undergo degradation in the nucleus instead of nucleocytoplasmic transport.