Metagenomic Sequencing of HIV-1 in the Blood and Female Genital Tract Reveals Little Quasispecies Diversity during Acute Infection

Abstract

Heterosexual transmission of human immunodeficiency virus type 1 (HIV-1) is associated with a significant bottleneck in the viral quasispecies population, yet the timing of that bottleneck is poorly understood. We characterized HIV-1 diversity in the blood and female genital tract (FGT) within 2 weeks after detection of infection in three women enrolled in a unique prospective cohort in South Africa. We assembled full-length HIV-1 genomes from matched cervicovaginal lavage (CVL) samples and plasma. Deep sequencing allowed us to identify intrahost single-nucleotide variants (iSNVs) and to characterize within-sample HIV-1 diversity. Our results demonstrated very little HIV-1 diversity in the FGT and plasma by the time viremia was detectable. Within each subject, the consensus HIV-1 sequences were identical in plasma and CVL fluid. No iSNV was present at >6% frequency. One subject had 77 low-frequency iSNVs across both CVL fluid and plasma, another subject had 14 iSNVs in only CVL fluid from the earliest time point, and the third subject had no iSNVs in CVL fluid or plasma. Overall, the small amount of diversity that we detected was greater in the FGT than in plasma and declined over the first 2 weeks after viremia was detectable, compatible with a very early HIV-1 transmission bottleneck. To our knowledge, our study represents the earliest genomic analysis of HIV-1 in the FGT after transmission. Further, the use of metagenomic sequencing allowed us to characterize other organisms in the FGT, including commensal bacteria and sexually transmitted infections, highlighting the utility of the method to sequence both HIV-1 and its metagenomic environment.IMPORTANCE Due to error-prone replication, HIV-1 generates a diverse population of viruses within a chronically infected individual. When HIV-1 is transmitted to a new individual, one or a few viruses establish the new infection, leading to a genetic bottleneck in the virus population. Understanding the timing and nature of this bottleneck may provide insight into HIV-1 vaccine design and other preventative strategies. We examined the HIV-1 population in three women enrolled in a unique prospective cohort in South Africa who were followed closely during the earliest stages of HIV-1 infection. We found very little HIV-1 diversity in the blood and female genital tract during the first 2 weeks after virus was detected in the bloodstream. These results are compatible with a very early HIV-1 population bottleneck, suggesting the need to study the HIV-1 population in the female genital tract before virus is detectable in the bloodstream.

Keywords: bottleneck; female genital tract; human immunodeficiency virus; metagenomic.

FIG 1

Viral load patterns in plasma and CVL fluid. The plots indicate the HIV-1 load in plasma (solid lines) and the HIV-1 RNA quantification in CVL fluid (dashed lines) for the 3 subjects in this study over time. Viral loads in plasma are not directly comparable to viral loads in CVL fluid due to differences in specimen collection, processing and HIV-1 quantification. Samples used for sequencing in this study are indicated by large circles (plasma) and squares (CVL fluid).

FIG 2

Metagenomic sequencing and analysis approach. The schematics indicate laboratory sequencing methods, metagenomic analysis of microbial content, and HIV-1-specific analysis. The asterisks mark steps in which two independent preparations were performed to ensure reproducibility.

FIG 3

Phylogenetic tree of HIV-1 consensus sequences from each sample. Sequences from subject A are labeled in blue, those from subject B are labeled in purple, and those from subject C are labeled in magenta. Reference subtype C sequences from South Africa are labeled in green, other subtype C sequences are labeled in yellow, one subtype B sequence is labeled in orange, and one subtype A sequence is labeled in red. Sequence names for samples in this study indicate the subject, day of sampling, and sample type. Reference sequences are named by subtype, country of origin, year, and GenBank accession number. Nodes with at least 80% support (out of 1,000 bootstraps) are labeled with the bootstrap values.

FIG 4

HIV-1 genome coverage for each sample. The plots indicate the sequencing depth across the HIV-1 genome for each sample, which was calculated as a sliding average with a bin width of 100 nt and a sliding window of 10 nt. The coverage represents the sum across both independently prepared sequencing libraries. The dashed lines represent the mean coverage.

FIG 5

Comparison of SGA and metagenomic sequencing for iSNV detection. For each iSNV in the validation sample, Fisher’s exact test was used to compare the frequency of the iSNV detected by SGA with the frequency of the iSNV detected by metagenomic sequencing. The plot shows a frequency distribution of the resulting P values for the observed comparisons, as well as P values for simulations performed to assess whether the differences between the two methods were more than could be expected from sampling variance.

FIG 6

Frequencies of iSNVs in subject A (a) and subject B (b). Each row represents one sample, and each column represents one iSNV position; invariant positions are not shown. iSNV frequencies are indicated by color. No chart is shown for subject C because no iSNVs were detected. Table S2 contains further information, including the exact position of each iSNV and its consensus and variant alleles.

FIG 7

gag and env variants in subject A. (a) (Top) Schematic showing the genome positions of the deletion in the PTAP region of gag and the complex variant in env gp120 (prior to the V1 loop) in subject A, each marked with an asterisk. (Bottom) Nucleotide positions are indicated relative to the HXB2 reference sequence. Red indicates mismatches from subject A’s consensus sequence, and the red vertical line represents the 12-amino-acid deletion. The italicized Ns indicate potential N-linked glycosylation sites. (b) Frequency of each variant in each sample from subject A. (Left) Frequencies of the gag deletion. (Right) Frequencies of the env variant. For gag, the plot shows the lower limit of the frequency of the deletion in each sample. Because the deletion occurred in an area of genome duplication, not all reads could be unambiguously mapped; the upper limit of the frequency of the deletion in each sample was approximately 25%. For env, all the reads could be unambiguously mapped, and the plot shows the frequency of the variant in each sample. Each sample is identified by the number of days (D) after the first positive HIV-1 fingerstick.

FIG 8

Metagenomic profiles of CVL fluid and plasma over time. (a) Results of metagenomic classification, indicating the log proportions of genera frequently found in CVL fluid after removal of human, HIV-1, and contaminating Burkholderiales reads. The columns represent genera, and the rows represent samples. Each sample is identified by subject identifier–number of days after the first positive HIV-1 fingerstick (D)–sample type. (b) PCA of centered log ratio-transformed genus abundance proportions for each sample.