Functional sites in the 5' region of human immunodeficiency virus type 1 RNA form defined structural domains.

The 5' region of HIV-1 RNA contains functional elements involved in key steps of the retroviral cycle, such as genomic RNA transcription, splicing, translation, dimerization or initiation of reverse transcription. In the present work, we investigated the conformation of the first 500 nucleotides covering the RNA leader and the 5' gag coding sequences of HIV-MAL, using chemical probing. We provide detailed information on almost each nucleotide at one of their Watson-Crick positions and on position N-7 of purines. Experiments were conducted on two in vitro transcribed RNA fragments (1 to 707 and 1 to 311). A secondary structure model was derived by combining the experimental data, computer predictions and sequence comparison. Under conditions favoring dimerization (high salt concentration), HIV-1 RNA folds into independent structural domains that can be related to defined functional regions. The first domain corresponds to TAR forming a stable stem-loop. Intrinsic structural features are found to stabilize the TAR hairpin loop. The second domain (nucleotides 56 to 299) contains the PBS sequence, which is located in a stable subdomain constrained by a four stem junction (nucleotides 139 to 218). Although the MAL isolate has an insertion near the PBS, probably resulting from the duplication of a 23-nucleotide sequence, the structural organization of this subdomain is conserved in all other HIV-1 isolates. The third domain (nucleotides 300 to 404) contains the splice donor site, packaging and dimerization elements and the AUG initiation codon of gag. A major result is the structural versatility of this region. Two mutually exclusive structures, both equally in agreement with probing data, could modulate the different functions involving this domain. The reduced accessibility of the gag translational initiation site possibly accounts for the low efficiency of the in vitro translation of the dimer. Finally, the 5' gag coding sequences form a metastable domain.