In vivo mutation rates and the landscape of fitness costs of HIV-1

Abstract

Mutation rates and fitness costs of deleterious mutations are difficult to measure in vivo but essential for a quantitative understanding of evolution. Using whole genome deep sequencing data from longitudinal samples during untreated HIV-1 infection, we estimated mutation rates and fitness costs in HIV-1 from the dynamics of genetic variation. At approximately neutral sites, mutations accumulate with a rate of 1.2 × 10^-5 per site per day, in agreement with the rate measured in cell cultures. We estimated the rate from G to A to be the largest, followed by the other transitions C to T, T to C, and A to G, while transversions are less frequent. At other sites, mutations tend to reduce virus replication. We estimated the fitness cost of mutations at every site in the HIV-1 genome using a model of mutation selection balance. About half of all non-synonymous mutations have large fitness costs (>10 percent), while most synonymous mutations have costs <1 percent. The cost of synonymous mutations is especially low in most of pol where we could not detect measurable costs for the majority of synonymous mutations. In contrast, we find high costs for synonymous mutations in important RNA structures and regulatory regions. The intra-patient fitness cost estimates are consistent across multiple patients, indicating that the deleterious part of the fitness landscape is universal and explains a large fraction of global HIV-1 group M diversity.

Keywords: Evolution; HIV-1; fitness landscape.; mutation rate.

Figure 1.

Mutation rate estimates. (A, B) Accumulation of divergence at approximately neutral sites for transitions and transversions, respectively (EDI: estimated date of infection). The slope of the individual regression lines in panels A and B provide estimates of the in vivo mutation rates. (C) Schematic representation and quantification of the mutation rates. Error bars for the estimates, indicated in parenthesis as uncertainties over the last significant digit, are standard deviations over 100 patient bootstraps.

Figure 2.

Average intra-patient fitness cost across quantiles of global HIV-1 group M diversity. (A) Divergence (measured as 1 − frequency of the ancestral state) saturates fast in the conserved parts of the genome (dark blue to cyan), more slowly in regions of intermediate conservation (green and yellow) and keeps increasing at the least conserved sites (red dots). The solid lines show fits of Eq. (2) to the binned data with fitness cost s as free parameter while the mutation rate is fixed at 1.2 × 10⁻⁵ per site per day (black line). (B) The “Sat” line shows fitness cost estimated for the blue, cyan, green, yellow, and red curves of panel A (indicated by arrows of the same colors). The most conserved quantile (dashed dark blue line in panel A) is not shown because saturation happens too rapidly to obtain an accurate fit. The “Pooled” line refers to harmonic averages of site-specific cost estimates. The ranges of entropy values contributing to each data point are indicated by horizontal lines, while the vertical error bars refer to the standard deviation of 100 bootstraps over patients: note that while error bars are small, there is substantial variation of fitness costs across sites within each diversity group.

Figure 3.

Fitness costs along the HIV-1 genome. (A) Fitness costs of synonymous and non-synonymous mutations in gag, pol, vif, vpu, env, and nef as a geometric sliding average with a window size of 30 bases. Estimates in gp120 are expected to be less accurate due to consistent difficulties amplifying this part of the genome. (B) Fitness costs in selected regions of the genome that contain important regulatory elements. Blue dots show estimates for individual bases, blue lines indicate running averages with a window size of eight bases and red lines are running averages excluding bases where mutations cause amino acid changes. PBS: tRNA primer binding site. U5: unique 5′ region. SL 1–4 PSI: stem loops of the PSI packaging signal. (c) PPT: (central) poly purine tract. A1, D2: splice sites.

Figure 4.

Distributions of fitness costs within coding regions. (A) Synonymous mutations, (B) mutations that are synonymous in one gene but affect another protein in a different reading frame, and (C) non-synonymous mutations (includes codons in gag, pol, vif, vpu, vpr). Half of non-synonymous mutations are very costly (>10 percent), while most synonymous mutations have a relatively small cost (<1 percent). The extremal bins include all points beyond the axis boundary. Fitness costs are measured in 1/day.

Figure 5.

CTL selection blurs the relationship between fitness costs and diversity. (A) Each dot represents a site in nef: red (blue) dots are associated (not associated) with HLA types (Carlson et al. 2012). Dots surrounded by a green circle are associated with low viral load (Bartha et al. 2013). Intrapatient fitness costs are anticorrelated with subtype diversity (Spearman $\rho =-0.59$). The majority of sites in nef with high diversity despite high fitness costs—top right corner—are associated with either HLA types or with low viral load, while few sites in the lower left corner are associated with HLA variation. Panel B quantifies this trend by plotting the fraction of HLA associated sites in bins of increasing diversity and fitness costs (bin boundaries are denoted by straight grey lines in panel A, α  = 2). This figure uses data from subtype B patients only.

Figure 6.

Pre-existing drug resistance mutations carry a high cost. Each point shows the average frequency of minor amino acids in individual patients. The bottom row indicates in how many out of ten patients each mutation is not observed, the top panel shows the estimated fitness costs associated with the mutations. The following mutations were never found at frequencies above 0.1 percent in any patient, indicating a large fitness cost: PI: L24I, V32I, I154VTAM, L76V, N88S, L90M; NRTI: M41L, K70ER, L74VI, Y115F, T215YF, K219QE; NNRTI: L100I, K103N, V106AM, E138K, V179DEF, Y188LCH, M230L; INI: E92Q, N155H. Most mutations are observed in no patient or only in a few patients, indicating high fitness costs.