Terminal nucleotides of the preintegrative linear form of HIV-1 DNA deduced from the sequence of circular DNA junctions.

The termini of integrated retroviral DNAs are characterized by highly conserved dinucleotide sequences: 5'TG...//...CA3'. For the avian and murine C-type retroviruses, the dinucleotide sequences reside two deoxynucleotides (usually AA and TT) from the LTR ends of unintegrated viral DNA (5' AATG...//...CATT3'). The number and identity of terminal deoxynucleotides in unintegrated HIV-1 linear DNA that extend beyond the conserved dinucleotides have not been determined. However, they had been presumed to consist of a single nucleotide (5'CTG...//...CAG3') on each end, based on inspection of the nucleotide sequence between the end of the supposed primer sites for retroviral DNA synthesis and the conserved proviral termini. We have utilized the polymerase chain reaction (PCR) to amplify segments representing the joined ends of linear DNA present in covalently closed circular HIV-1 DNA molecules isolated from infected T cells. These fragments were cloned and the nucleotide sequence of the LTR-LTR circle junction of several independent clones determined. Based upon the results, we predict that, like the avian and murine viruses, HIV-1 linear DNA contains two nucleotides beyond the conserved dinucleotides: 5'ACTG...//...CAGT3'. Models for the origin of these termini and other observed junction sequences are proposed.