Post-transcriptional regulation in HIV-1 and human fibronectin: Multiple cis-acting elements

For many years, the fact that the primary transcript of human immunodeficiency virus type 1 (HIV-1) is inefficiently spliced was largely taken for granted and the underlying mechanism remained unexplored. We set out to define the cis-acting elements involved in splicing regulation of the HIV-1 tat/rev intron. This task was facilitated by the use of a plasmid designed for transient expression in COS-7 cells of a synthetic gene that contained the first intron of human beta-globin. By exchanging either the 5' splice site (5' ss) or the 3' splice site (3' ss) of beta-globin with various portions of the HIV-1 genome spanning the tat/ rev splice sites, we could draw several conclusions.;By trimming away flanking sequences, we demonstrated that the tat/rev 3' ss is intrinsically inefficient. Mutagenesis studies revealed that both the branchpoint and polypyrimidine tract elements were suboptimal. The tat/rev 5' ss, on the other hand, was shown to be utilized efficiently. The efficiency of the tat/rev 5' ss was not affected by inclusion of the tat/ rev intron sequences. Therefore, we concluded that the only requirement for inefficient splicing of the tat/rev intron is the suboptimal nature of the branchpoint and polypyrimidine tract elements.;Deletion studies revealed that two cis-acting elements within the downstream exon modulate splicing of the tat/ rev intron. These elements include a positive element, the exon splicing enhancer (ESE), and an adjacent negative element, the exon splicing silencer (ESS). The ESE was mapped to a 10 nt purine-rich element that is functionally interchangeable with other purine-rich ESEs. The ESS exhibited context dependence in that its extent of splicing inhibition was sensitive to mutations that improve the branchpoint or polypyrimidine tract.;During the course of our investigation of the HIV-1 ESS in a heterologous context, we made the unexpected discovery of novel splicing regulatory elements within the EDA exon of the human fibronectin gene. As in HIV-1, these elements include adjacent positive and negative elements. Although the fibronectin ESE is not purine-rich, we demonstrated that it is functionally analogous to the ESE of HIV-1. The fibronectin ESS is relatively large. Evolutionary alignment and mutagenesis studies implicate secondary structures as determinants of splicing inhibition.;Finally, using chimeric beta-globin/HIV-1 introns as above, we tested a model for nuclear retention of unspliced pre-mRNA which proposes that the RNA becomes trapped in "handicapped" splicing complexes as a consequence of the pairing of a strong splice site with a weak one. We present evidence that the inefficient tat/rev 3' ss is insufficient to cause nuclear retention; additional tat/rev intron sequences are required to prevent the cytoplasmic accumulation of unspliced RNA.