Targeted deep sequencing of HIV-1 using the IonTorrentPGM platform

Abstract

The characterization of mixed HIV-1 populations is a key question in clinical and basic research settings. This can be achieved through targeted deep sequencing (TDS), where next-generation sequencing is used to examine in depth a sub-genomic region of interest. This study explores the suitability of IonTorrent PGM(LifeTechnologies) for the TDS-based analysis of HIV-1 evolution. Using laboratory reagents and primary specimens sampled at pre-peak viremia the error rates from misincorporation and in vitro recombination were <0.5%. The sequencing error rate was 2- to 3-fold higher in/around homopolymeric tracts, and could be discerned from true polymorphism using bidirectional sequencing. The limit of detection of complex variants was further lowered by using haplotyping. The application of this system was illustrated on primary samples from an individual infected with HIV-1 followed from pre-peak viremia through six months post-acquisition. TDS provided an augmented view of the extent of genetic diversity, the covariation among polymorphisms, the evolutionary pathways, and the boundaries of the mutational space explored by the viral swarm. Based on its performance, the system can be applied for the characterization of minor viral variants in support of studies of viral evolution, which can inform the rational design of the next generation of vaccines and therapeutics.

Keywords: HIV-1; IonTorrent; Molecular evolution; Next-generation sequencing; Targeted deep sequencing.

Figure 1

Sequencing error rate based on E. coli CAT. a) At each sequenced position, the frequency of each base differing from the reference sequence (shown in the abscissa) is depicted. The inset shows the distribution of the error rate, and the single position where the error rate was greater than 0.5% (#1) is highlighted. b) For each base sequenced at each position, the frequency derived from forward and reverse reads is shown. c) Detail of the frequency of C at position #1 computed based on forward and reverse reads. See text for details.

Figure 2

Sequencing error rate based on the gp120-V2 region of a pre- peak viremia sample. a) At each sequenced position, the frequency of each base differing from the reference sequence (shown in the abscissa) is depicted. The inset shows the distribution of the error rate, and the three positions where the error rate was greater than 0.5% (#1, #2, and #3) are highlighted. b) For each base sequenced at each position, the frequency derived from forward and reverse reads is shown. c-e) Details of the frequency of G, A and G at positions #1, #2, and #3, respectively, computed based on forward and reverse reads. See text for details.

Figure 3

Sequencing error rate based on the gp120-C3 region of a pre- peak viremia sample. a) At each sequenced position, the frequency of each base differing from the reference sequence (shown in the abscissa) is depicted. A G₄ homopolymeric tract responsible for sequencing error at position #3 is underlined. The inset shows the distribution of the error rate, and the three positions where the error rate was greater than 0.45% (#1, #2, and #3) are highlighted. b) For each base sequenced at each position, the frequency derived from forward and reverse reads is shown. c-e) Details of the frequency of G, A and A at positions #1, #2, and #3, respectively, computed based on forward and reverse reads. See text for details.

Figure 4

TDS-based longitudinal analysis of the HIV-1 gp120-V2 region from pre-peak viremia through six months post-infection. a) The frequency of viral variants detected at each time-point are shown. b) Sequences of detected viral variants are depicted, and differences from the T/F virus (variant #1) are highlighted. Codon numbers in the HIV-1 env gene, relative to the HXB2 coordinate nomenclature, are indicated. See text for details.

Figure 5

TDS-based longitudinal analysis of the HIV-1 gp120-C3 region from pre-peak viremia through six months post-infection. a) The frequency of viral variants detected at each time-point is shown. b) Sequences of detected viral variants are depicted, and differences from the T/F virus (variant #1) are highlighted. Codon numbers in the HIV-1 env gene, relative to the HXB2 coordinate nomenclature, are indicated. c) The figure shows a likely reconstruction of the evolution of the different variants. Color-coded diamonds and square boxes represent the corresponding variants. Large diamonds represent major variants, detectable by traditional SGA-based analyses, and small squares represent minor variants (frequency <5%) that would have been missed by traditional SGA-based analyses. Solid blue arrows depict single-base substitution events, red arrows depict multiple G-to-A substitution events, and dotted lines represent alternative single-base substitution or recombination events. The date of first detection of the variants is shown. See text for details.