Background:  HIV-1 drug-resistant mutants emerge from diverse virus populations, but little is known regarding the contributions of various genetic mechanisms to the origin and maintenance of this diversity. To study this issue, we investigated HIV-1 sequence variation in plasma of chronically infected drug-naïve and drug resistant patients.

Methods:  Individual HIV-1 pro-pol sequences were obtained using single-genome sequencing from 9 naïve and 6 drug-resistant patients; > 15 amplicons per sample were obtained and analyzed phylogenetically. Recombination was investigated using linkage disequilibrium methods and the 4 gamete model of Hudson.

Results:  HIV-1 populations were highly diverse, (average pair-wise distance, 0.009 to 0.02 per site), with little change in population structure over several years. Population variability was independent of viral RNA level over a 1000-fold range, suggesting a number of infected cells sufficient to avoid a genetic bottleneck, even at low viremia. Variation was seen in 91% of the pro-pol positions. In naïve patients, average pair-wise distance analysis revealed that variation at these positions occurred randomly; modeling studies revealed HIV-1 populations in vivo closely approximated populations with mutations distributed according to Poisson statistics. Additional analyses of co-variation of mutations revealed strong evidence for frequent recombination in HIV-1 populations in both naïve and drug-resistant populations. The mean number of recombination intervals per sequence was about 10, similar to values reported for infected cell lines. The lack of a strong association between the level of viremia and predicted numbers of recombination events implied that the number of infected cells is sufficiently large to permit frequent recombination, even at relatively low RNA levels (3000 copies/mL). We also analyzed sequential samples from recently infected individuals and identified the emergence of several new variants and of recombinants between them. Using this dataset, we estimated that 1-3 recombination events occurred per replication cycle. Based on estimates that progeny of doubly infected cells have about 10 crossovers per genome, we can crudely estimate that a substantial proportion (> 20%) of infected cells in vivo are infected with 2 or more proviruses. 

Conclusions:  Frequent recombination among HIV-1genomes indicates that infection occurs in a large well-mixed population with frequent double infection.

Keywords: Recombination; population genetics; drug resistance