[Paper Review] In-vivo mutation rates and fitness landscape of HIV-1

Fabio Zanini, Vadim Puller+3 more|arXiv (Cornell University)|Mar 21, 2016

HIV Research and TreatmentImmunology and Microbiology1 references3 citations

TL;DR

This study estimates in vivo HIV-1 mutation rates and fitness costs using longitudinal deep-sequencing data from untreated patients. By modeling the balance between mutation and selection at neutral and non-neutral sites, it reveals a mutation rate of 1.2×10⁻⁵ per site per day and shows that ~50% of nonsynonymous mutations have fitness costs >10%, with costs varying across genomic regions, especially in regulatory and structural RNA elements.

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 temporal dynamics of genetic variation. At approximately neutral sites, mutations accumulate with a rate of 1.2 x 10^-5 per site per day, in agreement with the rate measured in cell cultures. The rate from G to A is largest, followed by the other transitions C to T, T to C, and A to G, while transversions are more rare. At non-neutral sites, most mutations reduce virus replication; using a model of mutation selection balance, we estimated the fitness cost of mutations at every site in the HIV-1 genome. About half of all nonsynonymous mutations have large fitness costs (greater than 10\%), while most synonymous mutations have costs below 1\%. The cost of synonymous mutations is especially low in most of gag and pol, while much higher costs are observed in important RNA structures and regulatory regions. The intrapatient fitness cost estimates are consistent across multiple patients, suggesting that the deleterious part of the fitness landscape is universal and explains a large fraction of global HIV-1 group M diversity.

Motivation & Objective

• To quantify in vivo mutation rates and fitness costs of HIV-1 mutations in untreated patients.
• To overcome limitations of in vitro assays and population-level sequence data by using longitudinal intrapatient variation.
• To estimate fitness effects at single-nucleotide and amino acid resolution across the HIV-1 genome.
• To determine whether the fitness landscape of HIV-1 is universal across patients, independent of immune escape dynamics.
• To link sequence conservation in global HIV-1 group M diversity to direct fitness cost estimates from within-host evolution.

Proposed method

• Used whole-genome deep sequencing of HIV-1 RNA from 9 untreated patients across 6–12 time points per patient.
• Estimated mutation rates via linear regression of variant frequency over time at approximately neutral sites, using time-binned frequency data.
• Applied a two-state model (consensus vs. derived mutation) to sites with stable consensus over time to estimate selection coefficients via nonlinear least squares fitting.
• Used weighted averaging of SNP frequencies across patients, with weights based on template input to correct for sampling bias and sequencing error.
• Calculated site-specific fitness costs as μ/ x̂, where μ is the mutation rate away from consensus and x̂ is the equilibrium frequency of the derived allele.
• Constructed bootstrap distributions to estimate uncertainty in mutation rates and fitness cost estimates.

Experimental results

Research questions

• RQ1 What is the in vivo mutation rate of HIV-1, and how does it compare to in vitro measurements?
• RQ2 What is the spectrum of mutations (e.g., transitions vs. transversions) in HIV-1 under natural infection?
• RQ3 How do fitness costs of mutations vary across the HIV-1 genome, and are they consistent across different patients?
• RQ4 To what extent do synonymous mutations contribute to fitness costs, and where are these costs highest?
• RQ5 Is the deleterious part of the HIV-1 fitness landscape universal across hosts, independent of immune selection?

Key findings

• The in vivo mutation rate of HIV-1 is 1.2×10⁻⁵ per site per day, consistent with rates measured in cell culture.
• The mutation spectrum shows G→A is the most frequent, followed by C→T, T→C, and A→G, with transversions being rare.
• Approximately 50% of nonsynonymous mutations have fitness costs exceeding 10%, indicating strong purifying selection.
• Most synonymous mutations have fitness costs below 1%, but costs are significantly higher in functional RNA structures and regulatory regions.
• Fitness costs in gag and pol are particularly low for synonymous mutations, except in key RNA elements like the RRE and psi.
• Intrapatient fitness cost estimates are consistent across multiple patients, suggesting a universal fitness landscape that explains a large fraction of global HIV-1 group M diversity.