Human immunodeficiency virus (HIV) type 1 proviral hypermutation correlates with CD4 count in HIV-infected women from Kenya

Abstract

APOBEC3G is an important innate immune molecule that causes human immunodeficiency virus type 1 (HIV-1) hypermutation, which can result in detrimental viral genome mutations. The Vif protein of wild-type HIV-1 counteracts APOBEC3G activity by targeting it for degradation and inhibiting its incorporation into viral particles. Additional APOBEC cytidine deaminases have been identified, such as APOBEC3F, which has a similar mode of action but different sequence specificity. A relationship between APOBEC3F/G and HIV disease progression has been proposed. During HIV-1 sequence analysis of the vpu/env region of 240 HIV-infected subjects from Nairobi, Kenya, 13 drastically hypermutated proviral sequences were identified. Sequences derived from plasma virus, however, lacked hypermutation, as did proviral vif. When correlates of disease progression were examined, subjects with hypermutated provirus were found to have significantly higher CD4 counts than the other subjects. Furthermore, hypermutation as estimated by elevated adenine content positively correlated with CD4 count for all 240 study subjects. The sequence context of the observed hypermutation was statistically associated with APOBEC3F/G activity. In contrast to previous studies, this study demonstrates that higher CD4 counts correlate with increased hypermutation in the absence of obvious mutations in the APOBEC inhibiting Vif protein. This strongly suggests that host factors, such as APOBEC3F/G, are playing a protective role in these patients, modulating viral hypermutation and host disease progression. These findings support the potential of targeting APOBEC3F/G for therapeutic purposes.

FIG. 1.

Sequence context of identified G-to-A hypermutation in the first 75 nucleotides of the env ORF from 13 dramatically mutated proviruses. The top panel displays the nucleotide sequence, while the bottom panel shows the corresponding amino acid sequence. The consensus sequence for both the DNA and protein was generated based on the 240 proviral sequences examined in this publication. The sequences are numbered from the HIV-1 env start. The proviral sequence was compared to the corresponding nonhypermutated plasma RNA sequence, where available, or else to a clade-specific consensus sequence. Darkened nucleotide letters in the sample sequences indicate changes from G to A in either a GG or GA context (i.e., proposed APOBEC3G or APOBEC3F hypermutation). Darkened amino acid letters in the sample sequences indicate where proposed hypermutation at the nucleotide level caused an amino acid change at the protein level.

FIG. 2.

Proviral adenine proportion. (A) Distribution of proviral adenine proportion in a 590-nucleotide fragment spanning vpu and the 5′ end of env for 240 HIV-1 isolates. The adenine proportion is based on proviral sequences from PBMC DNA isolated from HIV-infected women from two Kenyan cohorts. The values greater than 1 standard deviation above the mean (x̄ ± 1 S.D.) and the outliers are indicated. (B) Adenine proportion in proviral HIV-1 sequence and plasma-derived viral RNA sequence in 18 matched samples. The same 590-nucleotide HIV-1 genomic region was examined for both proviral and viral RNA sequences. The four sequences indicated with the asterisk had hypermutated proviral vpu/env sequences.

FIG. 3.

Clones and directly sequenced PCR product are compared to a population-specific, clade-specific consensus sequence. The nucleotides are numbered across the bottom of each panel, indicating position within the 590-nucleotide region. ML103 was selected as a representative nonhypermutated sequence. Changes in the patient sequence compared to the consensus sequence are indicated by colored bars, as indicated in the legend. The representations were generated with the Highlighter tool available from http://www.hiv.lanl.gov/content/sequence/HIGHLIGHT/highlighter.html. Black circles below hypermutation sites were added manually to ease visualization. The sequences are ordered from top to bottom from the most similar to the consensus sequence to the least similar. Identical sequences are shown once, with the number of times the sequence was retrieved listed beside the clone name. Patients with an asterisk showed intraclonal diversity.

FIG. 3.

Clones and directly sequenced PCR product are compared to a population-specific, clade-specific consensus sequence. The nucleotides are numbered across the bottom of each panel, indicating position within the 590-nucleotide region. ML103 was selected as a representative nonhypermutated sequence. Changes in the patient sequence compared to the consensus sequence are indicated by colored bars, as indicated in the legend. The representations were generated with the Highlighter tool available from http://www.hiv.lanl.gov/content/sequence/HIGHLIGHT/highlighter.html. Black circles below hypermutation sites were added manually to ease visualization. The sequences are ordered from top to bottom from the most similar to the consensus sequence to the least similar. Identical sequences are shown once, with the number of times the sequence was retrieved listed beside the clone name. Patients with an asterisk showed intraclonal diversity.

FIG. 4.

Dinucleotide context of G-to-A hypermutation in 13 patients. The proportion of G-to-A hypermutation in each proviral sequence occurring at the dinucleotides GG, GA, GC, and GT is indicated. Hypermutation context was determined through comparison to a clade-specific consensus sequence, using Hypermut 2.0.

FIG. 5.

Association of higher CD4 count with hypermutated sequences. CD4 counts are compared in patients with hypermutated provirus and patients without hypermutated provirus. (A) Hypermutation was defined as significant general APOBEC cytidine deaminase activity, as determined by Hypermut 2.0. (B) Hypermutation was defined as sequences with mutations in Vpu and Env, proviral adenine proportion greater than 1 standard deviation above the mean, and significant Hypermut 2.0 hypermutation. Groups are compared by a Mann-Whitney test. The height of the bar indicates the mean, and the error bar represents the standard error of the mean.

FIG. 6.

Correlation of adenine proportion with CD4 count. CD4 counts were available for 208 subjects. Adenine proportion was measured from the 590-nucleotide vpu/env HIV-1 proviral region. These two measures were significantly positively correlated (P = 0.041).