In-Depth Characterization of Full-Length Archived Viral Genomes after Nine Years of Posttreatment HIV Control

Abstract

In the search for control of human immunodeficiency virus type 1 (HIV-1) infection without antiretroviral therapy, posttreatment controllers (PTCs) are models of HIV remission. To better understand their mechanisms of control, we characterized the HIV blood reservoirs of 8 PTCs (median of 9.4 years after treatment interruption) in comparison with those of 13 natural HIV infection controllers (HICs) (median of 18 years of infection) and with those of individuals receiving efficient antiretroviral therapy initiated during either primary HIV infection (PHIs; n = 8) or chronic HIV infection (CHIs; n = 6). This characterization was performed with single-genome amplification and deep sequencing. The proviral diversity, which reflects the history of past viral replication, was lower in the PTCs, PHIs, and aviremic HICs than in the blipper HICs and CHIs. The proportions of intact and defective proviruses among the proviral pool in PTCs were not significantly different from those of other groups. When looking at the quantities of proviruses per million peripheral blood mononuclear cells (PBMCs), they had similar amounts of intact proviruses as other groups but smaller amounts of defective proviruses than CHIs, suggesting a role of these forms in HIV pathogenesis. Two HICs but none of the PTCs harbored only proviruses with deletion in nef; these attenuated strains could contribute to viral control in these participants. We show, for the first time, the presence of intact proviruses and low viral diversity in PTCs long after treatment interruption, as well as the absence of evolution of the proviral quasispecies in subsequent samples. This reflects low residual replication over time. Further data are necessary to confirm these results. IMPORTANCE Most people living with HIV need antiretroviral therapy to control their infection and experience viral relapse in case of treatment interruption, because of viral reservoir (proviruses) persistence. Knowing that proviruses are very diverse and most of them are defective in treated individuals, we aimed to characterize the HIV blood reservoirs of posttreatment controllers (PTCs), rare models of drug-free remission, in comparison with spontaneous controllers and treated individuals. At a median time of 9 years after treatment interruption, which is unprecedented in the literature, we showed that the proportions and quantities of intact proviruses were similar between PTCs and other individuals. Unlike 2/7 spontaneous controllers who harbored only nef-deleted proviruses, which are attenuated strains, which could contribute to their control, no such case was observed in PTCs. Furthermore, PTCs displayed low viral genetic diversity and no evolution of their reservoirs, indicating very low residual replication, despite the presence of intact proviruses.

Keywords: HIV reservoirs; defective proviruses; next-generation sequencing; posttreatment HIV controller; posttreatment HIV controllers; provirus; proviruses; ultradeep sequencing.

FIG 1

Correlation of the genetic diversity indexes. Viral diversity was calculated by several indexes and based on 24,091 nondefective RT sequences from 32 individuals: the mean branch length, the mean number of alleles per nucleotide site, the mean entropy, and the mean p-distance. A random sampling of 20 sequences without replacement was performed to calculate these indexes, except for the mean branch length; the final results are the mean values of 1,000 repetitions. Correlations between all diversity indexes and between each diversity index and the HIV DNA viral load (log copies/10⁶ PBMCs) were calculated using Spearman’s coefficients and considered significant when the P value was <0.05.

FIG 2

Viral diversity of HIV DNA based on the RT gene estimated by different indexes. A 700-bp amplicon on the RT gene was amplified in HIV DNA genomes and sequenced using a 454 GS Junior. Sequences were aligned with Clustal, and phylogenetic trees were constructed with PhyML. Viral diversity was calculated by several indexes and based on 24,091 nondefective RT sequences. The indexes used were the branch length of the phylogenetic tree (samples with fewer than 4 nondefective haplotypes were excluded) (A), the entropy (B), the number of alleles per nucleotide site (C), and the p-distance (D). A random sampling of 20 sequences without replacement was performed for indexes B to D; the final results are the mean values of 1,000 repetitions. The mean value for each group is depicted with a colored horizontal line. The diversity was compared among groups using Wilcoxon tests. Significant differences (P < 0.05) are depicted with continuous lines above the graph, while trends (P < 0.05 and P < 0.08) are depicted with dotted lines. HICs were divided into two subgroups: HIC-a (aviremic) individuals had more than 50% of their viral loads from the 5 years before the sample date below 50 copies/mL and none above 400 copies/mL, while HIC-b (blippers) individuals did not meet those criteria.

FIG 3

Identification of intact sequences and the different types of defective sequences and their proportions. HIV DNA genomes were amplified from the 5′LTR to the 3′LTR after limiting dilution and sequenced using a MiSeq instrument. Complete genomic sequences were assembled using Spades. (A) Process of identification of genetically defective and intact proviruses. A sequential approach was used. Premature stop codons due to hypermutations and frameshifts were detected using pairwise alignment implemented in Goalign. MSD defects were visually identified by alignment of genomic sequences and comparison to HIV compendium sequences. (B) Mean proportions of each type of HIV DNA genome in each group of patients: N is the number of patients in the group, and n is the mean number of full-length sequences per individual.

FIG 4

Proportions of each type of full-length HIV DNA genome. HIV DNA genomes were amplified from the 5′LTR to the 3′LTR and sequenced using a MiSeq instrument. Complete genomic sequences were assembled using Spades. Genetic defects were sequentially identified as follows: large deletions (D), APOBEC-induced hypermutations (E), frameshifts (F), and defects in the Ψ/MSD regions of the 5′LTR (G). Sequences with a defect in the nef gene only were considered attenuated strains (C), and the remaining sequences were inferred to be intact (A). The mean value for each group is depicted with a colored horizontal line. The proportions of each type of provirus were compared among groups using Wilcoxon tests. Significant differences (P < 0.05) are depicted with continuous lines above the graphs, while trends (P values between 0.05 and 0.08) are depicted with dotted lines. (A) The median proportions (P) of intact proviruses were 2% in CHIs, 0% in HICs, 22% in PHIs, and 4% in PTCs. (B) The median proportions of defective proviruses (sum of large deletions, hypermutations, frameshifts, and 5′LTR defects) were 96% in CHIs, 67% in HICs, 75% in PHIs, and 96% in PTCs. (C) The median proportions of proviruses with defects in the nef gene only (attenuated strains) were 0% in CHIs, 6% in HICs, 2% in PHIs, and 0% in PTCs. (D) The median proportions of proviruses with large deletions were 82% in CHIs, 56% in HICs, 49% in PHIs, and 60% in PTCs. (E) The median proportions of proviruses with hypermutations were 11% in CHIs, 0% in HICs, 4% in PHIs, and 5% in PTCs. (F) The median proportions of proviruses with frameshifts were 0% in CHIs, 0% in HICs, 8% in PHIs, and 4% in PTCs. (G) The median proportions of proviruses with 5′LTR defects were 0% in all four groups.

FIG 5

Changes in the proportions of intact full-length HIV DNA genomes for individuals with a second sample available 4 to 6 years after the initial sample. The median proportion of intact proviruses was compared between the two time points using a Wilcoxon test.

FIG 6

Phylogenetic trees of all nearly full-length HIV DNA genomic sequences for 3 PTCs. Individual trees of proviruses from the 3 PTCs who had two blood samples taken after TI. HIV DNA genomes were amplified from the 5′LTR to the 3′LTR and sequenced using a MiSeq instrument. Complete genomic sequences were assembled using Spades and aligned with MAFFT, and the phylogenetic tree was inferred with PhyML. Black squares represent sequences from the first samples, and red squares those from the second samples. Blue stars indicate identical sequences assumed to be clones. 180001 (A); 180002 (B); 180003 (C).