Sequence determinants of breakpoint location during HIV-1 intersubtype recombination

Retroviral recombination results from strand switching, during reverse transcription, between the two copies of genomic RNA present in the virus. We analysed recombination in part of the envelope gene, between HIV-1 subtype A and D strains. After a single infection cycle, breakpoints clustered in regions corresponding to the constant portions of Env. With some exceptions, a similar distribution was observed after multiple infection cycles, and among recombinant sequences in the HIV Sequence Database. We compared the experimental data with computer simulations made using a program that only allows recombination to occur whenever an identical base is present in the aligned parental RNAs. Experimental recombination was more frequent than expected on the basis of simulated recombination when, in a region spanning 40 nt from the 5′ border of a breakpoint, no more than two discordant bases between the parental RNAs were present. When these requirements were not fulfilled, breakpoints were distributed randomly along the RNA, closer to the distribution predicted by computer simulation. A significant preference for recombination was also observed for regions containing homopolymeric stretches. These results define, for the first time, local sequence determinants for recombination between divergent HIV-1 isolates.