Multi-Scale Evolution of Virulence of HIV-1

HIV-1 is a rapidly replicating retrovirus that faces two distinct fitness landscapes: within-host HIV-1 faces viral competition for host cells and for escape from the immune system, and between hosts HIV-1 faces a transmission bottleneck in which the majority of new infections are started by a single virus strain. Possibly as a result of these conflicting selective pressures, the rate of evolution of HIV-1 tends to be greater within-host than between hosts. A current hypothesis for this difference in evolutionary rates is that the HIV-1 latent reservoir acts to archive virus for later transmission. We offer a related but complimentary hypothesis: while some of the viruses’ life history traits are under selective pressure within-host, traits that are responsible for the efficiency of transmission to a new host are not under direct selection within-host and thus are subject to drift. Combined with the necessity of transmission through an extremely severe, competitive bottleneck, this results in the preferential transmission of founder-like viral lineages. As further evidence of the conflict between transmission fitness and within-host fitness, experimental evidence demonstrates that subtypes A and D are 100-fold more fit than subtype C in in vitro fitness competitions, yet subtype C dominates the global spread of new infections. It is unclear whether this discrepancy is caused by differences in within- and between-host fitness, or primarily reflects differences in in vitro versus in vivo fitness measures. To address this question, data from a four-year, 8000 participant study in Uganda and Zimbabwe were analysed for evidence of in vivo fitness differences between subtypes A, C and D. Analyzing this dataset along with simulated participant data, we conclude that either more frequent data sampling, or an even larger study, would be necessary to capture the early within-host dynamics sufficiently for a comparison across subtypes. Similar to subtypes A and D, subtype B is estimated to have an eight- to ten-fold in vitro fitness advantage over subtype C. Since frequent data collection over the early course of infection is necessary to quantify in vivo viral fitness, another approach to this question is to use data collected for simian/human immunodeficiency virus (SHIV). We develop a non-linear mixed-effects model for a meta-analysis of 143 non-human primates from over 20 sources to study in vivo fitness differences between SHIV subtypes B and C. Results suggest that subtype C has a lower replicative fitness but higher burst size than subtype B