Primer ID Informs Next-Generation Sequencing Platforms and Reveals Preexisting Drug Resistance Mutations in the HIV-1 Reverse Transcriptase Coding Domain.

Sequencing of a bulk polymerase chain reaction (PCR) product to identify drug resistance mutations informs antiretroviral therapy selection but has limited sensitivity for minority variants. Alternatively, deep sequencing is capable of detecting minority variants but is subject to sequencing errors and PCR resampling due to low input templates. We screened for resistance mutations among 184 HIV-1-infected, therapy-naive subjects using the 454 sequencing platform to sequence two amplicons spanning HIV-1 reverse transcriptase codons 34-245. Samples from 19 subjects were also analyzed using the MiSeq sequencing platform for comparison. Errors and PCR resampling were addressed by tagging each HIV-1 RNA template copy (i.e., cDNA) with a unique sequence tag (Primer ID), allowing a consensus sequence to be constructed for each original template from resampled sequences. In control reactions, Primer ID reduced 454 and MiSeq errors from 71 to 2.6 and from 24 to 1.2 errors/10,000 nucleotides, respectively. MiSeq also allowed accurate sequencing of codon 65, an important drug resistance position embedded in a homopolymeric run that is poorly resolved by the 454 platform. Excluding homopolymeric positions, 14% of subjects had evidence of ≥1 resistance mutation among Primer ID consensus sequences, compared to 2.7% by bulk population sequencing. When calls were restricted to mutations that appeared twice among consensus sequence populations, 6% of subjects had detectable resistance mutations. The use of Primer ID revealed 5-15% template utilization on average, limiting the depth of deep sequencing sampling and revealing sampling variation due to low template utilization. Primer ID addresses important limitations of deep sequencing and produces less biased estimates of low-level resistance mutations in the viral population.