Quantifying Selection against Synonymous Mutations in HIV-1 env Evolution
-
Fig 1Time series of frequencies of synonymous (A) and nonsynonymous (B) single nucleotide variants (SNVs) in env, C2-V5, from patient p10 (17). While many nonsynonymous SNVs fix, few synonymous SNVs do so even though they are frequently observed at high frequencies. Colors indicate the position of the site along the C2-V5 region (blue to red). (Inset) The fixation probability Pfix of a neutral SNV that reached 50% frequency is one half.
-
Fig 2Fixation and loss of SNVs. Panel A shows how quickly synonymous SNVs are purged from the populations. Specifically, the figure shows the fraction of SNVs that are still observed after Δt days, conditional on being observed in one of the three frequency intervals (which are shown in different colors). In each frequency interval, the fraction of synonymous SNVs that ultimately survive is the fixation probability Pfix conditional on the initial frequency. The neutral expectation for Pfix = ν0 is indicated by dashed horizontal lines. Panel B shows the fixation probability of synonymous SNVs as a function of ν0. Polymorphisms within C2-V5 fix less often than expected for neutral SNVs (indicated by the black dashed diagonal line). This suppression is not observed in other parts of env or for nonsynonymous SNVs. The horizontal error bars indicate the bin sizes, and the vertical ones indicate the standard deviation after 100 patient bootstraps of the data. Data in this figure are taken from references 17, 18, and 19.
-
Fig 3Permissible synonymous mutations tend to be unpaired. (A) Distribution of SHAPE reactivities among sites at which synonymous SNVs fixed, sites at which SNVs reached frequencies above 15% but were subsequently lost, and sites at which high-frequency SNVs were never observed (all categories are restricted to the regions V1-V5 [within 100 bp of V1-V5]). Sites at which SNVs fixed tend to have higher SHAPE reactivities, corresponding to less base pairing, than those at which SNVs are lost. Sites at which no SNVs are observed show an intermediate distribution of SHAPE values. (B) Fixation probability of synonymous SNVs in C2-V5 separately for variable regions V3-V5 and the connecting conserved regions C2-C4 that harbor RNA stems. As expected, the fixation probability is lower inside the conserved regions. Data in this figure are taken from references 10, 17, 18, and 19.
-
Fig 4Distribution of selection coefficients on synonymous sites. (A) The depression in Pfix depends on the deleterious effect size of synonymous SNVs. This parameter also reduces synonymous diversity, measured by the probability of a SNV to be found at intermediate frequencies Pinterm (top inset). The area observed in the data is also shown (bottom inset). (B) To assess the parameter space that affects synonymous fixation and diversity, we ran 3,000 simulations with random combinations of parameters for deleterious effect size, fraction of deleterious synonymous sites, average escape rate, rate of introduction of new epitopes, population size, mutation rate, and recombination rate (see Materials and Methods). The density of simulations that reproduce the synonymous diversity and fixation patterns observed in the data are shown. These simulations demonstrate that deleterious effects are around 0.002 and that a large fraction of the synonymous mutations need to be deleterious. The marginal distributions of each parameter are given in Fig. S5 in the supplemental material.
- Copyright © 2013, American Society for Microbiology. All Rights Reserved.
