Evaluating Clonal Expansion of HIV-Infected Cells: Optimization of PCR Strategies to Predict Clonality

Abstract

In HIV-infected individuals receiving suppressive antiretroviral therapy, the virus persists indefinitely in a reservoir of latently infected cells. The proliferation of these cells may contribute to the stability of the reservoir and thus to the lifelong persistence of HIV-1 in infected individuals. Because the HIV-1 replication process is highly error-prone, the detection of identical viral genomes in distinct host cells provides evidence for the clonal expansion of infected cells. We evaluated alignments of unique, near-full-length HIV-1 sequences to determine the relationship between clonality in a short region and clonality in the full genome. Although it is common to amplify and sequence short, subgenomic regions of the viral genome for phylogenetic analysis, we show that sequence identity of these amplicons does not guarantee clonality across the full viral genome. We show that although longer amplicons capture more diversity, no subgenomic region can recapitulate the diversity of full viral genomes. Consequently, some identical subgenomic amplicons should be expected even from the analysis of completely unique viral genomes, and the presence of identical amplicons alone is not proof of clonally expanded HIV-1. We present a method for evaluating evidence of clonal expansion in the context of these findings.

Fig 1. Clonal prediction scores of 1 kb amplicons spanning the HIV-1 genome.

(a) Schematic of algorithm to calculate clonal prediction score (CPS), which quantifies the proportion of unique sequences in an alignment that are correctly identified as unique using the amplicons produced by a specific primer set. (b) CPS of 1 kb-wide amplicons spanning the HIV-1 genome for six Acute treated–DNA subjects. (c) CPS of 1 kb-wide amplicons spanning the HIV-1 genome, averaged over all subjects in five different sample categories (Table 1). (d) Overlay of the five plots in part c. Schematics of the HIV-1 genome are aligned to the charts in parts b through d to highlight viral gene locations. The amplicons in parts b through d are defined with reference to the HXB2 genome.

Fig 2. Relationship between amplicon length and CPS.

Summary statistics describing all amplicons of a given length, spanning the HXB2 reference genome at 10 bp intervals. Amplicons with undefined CPS were not included in these summary statistic calculations. (a) Average, median, and minimum of the CPS values for every amplicon of a specified length spanning the viral genome. Summary statistics are shown for each subject and grouped by sample type. (b) Proportion of all of the amplicons of a specified length that have CPS values above 80.

Fig 3. Relationship between amplicon length and PCR coverage.

Percentage of sequences in each alignment that would be detectable by PCR, averaged over all amplicons of a specified length spanning HXB2 positions 2000 through 8000 at 10 bp intervals. Results are shown for 31 subjects and grouped by sample type (Table 1).

Fig 4. CPS of previously published amplicons.

(a) CPS for each of the primer sets listed in Table 2, averaged over all subjects within each sample type (Table 1). (b and c) Evaluation of previously published phylogenetic trees in the context of the CPS of the primer sets used to generate those trees. Part b shows proviral DNA samples and part c shows plasma RNA samples. Black diagonal lines show the relationship between the number of sequences collected and the number expected to be unique for each primer set listed in Table 2, as an estimate of the background signal level (assuming no clonality); the slopes of these lines are equal to the CPS values in part a divided by 100. The dotted red lines were calculated as the black lines plus or minus one standard deviation in CPS. Each plotted point indicates the actual number of total sequences and unique sequences present in a previously published phylogenetic tree. Phylogenetic trees containing both proviral DNA and plasma RNA sequences were counted separately for the two sequence types and plotted separately in parts b and c. Points plotted far below the black line (green-shaded region) indicate trees with more clonality than would be expected by chance from a sample of unique HIV-1 genomes. *References in which hypermutated sequences were not explicitly included in phylogenetic trees.