Target-sequence preferences of HIV-1 integration complexes in vitro.

Integration of reverse transcribed retroviral cDNA is not restricted to particular host DNA sequences. However, the frequency of integration into a particular phosphodiester bond is influenced by the local sequence. Here we examine the target-sequence preferences of purified HIV integrase and viral nucleoprotein complexes (preintegration complexes) isolated from freshly infected cells. We find that the three-base sequence including the integration site is not the major factor determining the frequency of integration, since identical triplets embedded in different sequences are used with very different efficiencies. However, there is a statistically significant bias against integration upstream of a pyrimidine nucleotide. The target-sequence preferences of purified integrase and preintegration complexes are very different. Strong integration sites on opposite DNA strands occur in pairs separated by five residues when preintegration complexes are used but not with purified integrase. These studies highlight the difference between the two sources of HIV integration activity and may provide the basis for a simple assay for the correct assembly of viral nucleoprotein complexes.