Hepatitis C virus (HCV) contains a variety of naturally occurring polymorphisms in the nonstructural protein 5A (NS5A) that reduce susceptibility to currently approved NS5A inhibitors. Previous investigations showed the Q80K mutation in the NS3 region of HCV has a common origin in the world population around 1930 in genotype 1a strongly coupled to substitutions A915/T, S147N or V29A. We sought to reconstruct the evolutionary history of the NS5A resistance-associated substitutions (RASs).
We collected more than 200,000 HCV sequences from global databases. Sequences were aligned using MAFFT v.7.54b and visually inspected using AliView v.1.18. After discarding sequences from the same patient and sequences that do not cover NS5A by 75%, we were left with 1,613 sequences. RASs were defined as substitutions in the NS5A gene previously identified based on literature sources. We inferred a phylogenetic tree for genotype specific datasets (1a, 1b, and 3a) under approximate maximum likelihood as implemented in FastTree2. Resulting phylogenies were rescaled to units of time using the R package ape. Next, we reconstructed the ancestral sequences at the nodes of the trees using MG94xREV codon model. Finally, NS5A RASs were mapped onto the resulting trees to recover patterns of ancestry.
We found, in contrast to Q80K in NS3, that most NS5A RAVs do not have a common origin, but rather each has evolved independently many times. They are widely dispersed amongst the phylogenetic trees of each genotype. Of the 22 types of RASs found in the total population, only 6 had origins that were not at the tips of the tree, each which had less than 5 RAS descendants. All of these RASs had their origins at tips of the tree, indicating recent origins. While the Q80K polymorphism has a highly localized origin in North America, we find that NS5A RASs have no strong pattern with respect to geography, with the overall prevalence ranging from 0.06% to 5%.
The inferred distribution of RASs in the NS5A region and frequency of their origin suggest that, unlike Q80K, there is a low fitness barrier without the need for co-evolution of compensatory mutations. A low fitness barrier may allow rapid selection of de novo resistance to NS5A inhibitors during therapy.