Genetically-barcoded SIV facilitates enumeration of rebound variants and estimation of reactivation rates in nonhuman primates following interruption of suppressive antiretroviral therapy

Abstract

HIV and SIV infection dynamics are commonly investigated by measuring plasma viral loads. However, this total viral load value represents the sum of many individual infection events, which are difficult to independently track using conventional sequencing approaches. To overcome this challenge, we generated a genetically tagged virus stock (SIVmac239M) with a 34-base genetic barcode inserted between the vpx and vpr accessory genes of the infectious molecular clone SIVmac239. Next-generation sequencing of the virus stock identified at least 9,336 individual barcodes, or clonotypes, with an average genetic distance of 7 bases between any two barcodes. In vitro infection of rhesus CD4+ T cells and in vivo infection of rhesus macaques revealed levels of viral replication of SIVmac239M comparable to parental SIVmac239. After intravenous inoculation of 2.2x105 infectious units of SIVmac239M, an average of 1,247 barcodes were identified during acute infection in 26 infected rhesus macaques. Of the barcodes identified in the stock, at least 85.6% actively replicated in at least one animal, and on average each barcode was found in 5 monkeys. Four infected animals were treated with combination antiretroviral therapy (cART) for 82 days starting on day 6 post-infection (study 1). Plasma viremia was reduced from >106 to <15 vRNA copies/mL by the time treatment was interrupted. Virus rapidly rebounded following treatment interruption and between 87 and 136 distinct clonotypes were detected in plasma at peak rebound viremia. This study confirmed that SIVmac239M viremia could be successfully curtailed with cART, and that upon cART discontinuation, rebounding viral variants could be identified and quantified. An additional 6 animals infected with SIVmac239M were treated with cART beginning on day 4 post-infection for 305, 374, or 482 days (study 2). Upon treatment interruption, between 4 and 8 distinct viral clonotypes were detected in each animal at peak rebound viremia. The relative proportions of the rebounding viral clonotypes, spanning a range of 5 logs, were largely preserved over time for each animal. The viral growth rate during recrudescence and the relative abundance of each rebounding clonotype were used to estimate the average frequency of reactivation per animal. Using these parameters, reactivation frequencies were calculated and ranged from 0.33-0.70 events per day, likely representing reactivation from long-lived latently infected cells. The use of SIVmac239M therefore provides a powerful tool to investigate SIV latency and the frequency of viral reactivation after treatment interruption.

Fig 1. Insertion of genetic barcode into SIVmac239.

(A) A 34 base cassette (yellow) bearing a stretch of 10 random bases was inserted between vpx and vpr of wild-type SIVmac239 to generate the genetically barcoded virus SIVmac239M. The MluI restriction site used is outlined in black, and the sequences of the barcode flanking regions are colored in blue and green to depict the two possible insertion orientations. (B) Representative sequences in single stock aliquots depicting the bimodal distribution of authentic barcodes (green) versus barcodes containing PCR error (red).

Fig 2. SIVmac239M in vitro and in vivo replication.

(A) CD8+ T-cell-depleted PBMCs infected with either SIVmac239 (teal) or SIVmac239M (maroon) were monitored over 13 days and samples collected were assayed by ELISA specific for reverse transcriptase (RT) over time. (B) Infectivity was tested in vivo in two rhesus macaques, ZK37 (light blue) and ZK56 (purple) following intravenous infection. Viral RNA copies were measured in plasma over 100 days post infection with a lower limit of detection of 15 copies/mL.

Fig 3. Evaluation of clonotypes found in stock and monkeys.

(A) Each individual clonotype was plotted by its rank order in the stock versus the number of animals in which it was found. A linear correlation was generated with an R² value of 0.77. Density of clonotypes at any single point are colored using a log scale heat map where red points depict 2 logs of clonotypes and dark blue points represent single clonotypes. (B) The number of barcodes was plotted against the number of monkeys in which the barcodes were found. Of the 9,336 total stock barcodes, 7,991 were found systemically in at least one animal, and 1,345 were not found in any of the 26 animals. (C) The mean relative frequency of each individual barcode (grey circles) was plotted against the number of monkeys in which the barcode was found. The relative frequencies of the barcodes demonstrate the comparative homogeneity of all clonotypes across all animals.

Fig 4. Plasma viral loads in animals in study 1.

Four rhesus macaques, KMB (purple), MK9 (pink), KZ2 (navy), and KTM (blue) were infected with SIVmac239M on day 6 and treated with cART for 82 days (study 1). Viral RNA copies were measured in plasma over 150 days.

Fig 5. Plasma viral loads, PBMC CA-vRNA and CA-vDNA in animals in study 2.

(A) 6 rhesus macaques, DEJX (red), DFGV (yellow), DEJW (blue), H090 (purple), DEPI (light green), and H105 (dark green) were infected with SIVmac239M and cART was initiated on day 4 post-infection. Pairs of macaques were removed from therapy on days 305, 374, and 482 post-infection. Viral RNA copies were measured in plasma collected up to 550 days. Bars over the viral load data indicate duration of therapy for each animal. (B-C) Cell-associated viral DNA (B) and RNA (C) was measured in PBMCs over the duration of cART therapy in animals infected with SIVmac239M. Measurements obtained are shown overlaying the detected plasma viral loads. Open symbols represent measurements that were below the limit of detection.

Fig 6. Longitudinal rebound clonotypes.

The relative proportion of the clonotypes in the six animals in study 2 were identified at multiple time points following therapeutic interruption. Each color represents a unique barcode. Although individual barcode numbers are not shown for simplicity, no rebound clonotypes were found in more than one animal. White portions of each column denote the area below the limit of detection for each sample based on template input quantities. Time points denoted with an asterisk indicate the time-point corresponding to the highest viral titers in the exponential growth phase.

Fig 7. Pairing of relative viral loads and time of reactivation.

Approximate day of reactivation of cells harboring clonotypes was estimated in study 2 animals following interruption of long-duration therapy for animals DEJX (A), DFGV (B), DEJW (C), H090 (D), DEPI (E), and H105 (F). Total viral load at the time point selected is comprised of the relative proportions of clonotypes determined by sequencing analysis. Each individual clonotype’s growth rate was estimated during the maximum exponential phase of the total viral load curve, and the slope of the growth of each clonotype was extended below the limit of detection to estimate the approximate time of reactivation. The grey line represents the plasma concentration of one viral copy in the total plasma, and the tick marks represent the theoretical reactivations based on calculated reactivation rate for each animal.