Within-patient HIV mutation frequencies reveal fitness costs of CpG dinucleotides, drastic amino acid changes and G→A mutations

HIV has a high mutation rate and exhibits remarkable genetic diversity. However, we know little about the fitness cost of HIV mutations in vivo. We calculated the mean frequency of mutations at 870 sites of the pol gene in 160 patients, allowing us to determine the cost of different types of mutations. We found that non-synonymous mutations that lead to drastic amino acid changes are three times more costly than those that do not, mutations that create new CpG dinucleotides are up to four times more costly than those that do not, and G→A mutations are more than twice as costly as A→G mutations. In addition, within-patient mutation frequencies are highly correlated with substitution frequencies across the global HIV pandemic. We anticipate our new frequency-based approach will provide insights into the fitness landscape and evolvability of not only HIV, but a variety of microbes.