Associations between the level of gag proviral hypermutation and markers of disease progression. Each point represents one individual, with the percentage of gag sequences that were identified as hypermutated indicated on the x axis. Plasma viral load (a) and CD4 count (b) measured in chronic infection are indicated on the y axis. Associations were evaluated by using the Spearman rank correlation test, with results indicated above the plots. The two individuals with extensive hypermutation (at least 40%) are labeled.

Virus replication and hypermutation in PBMC cultures of vif variants. (a) Plot of p24 pg/ml versus day of culture for each vif variant in the first set of cultures (PBMC from donor A, infected at an MOI of 0.02). Q23Δvif (red line) is the negative control; Q23/Q23vif (yellow) is the vif defective clone with its own vif reinserted; Q23/QA039vif (green) and Q23/QB368vif (purple) contain vif variants from individuals with extensive hypermutation; and Q23/QA268vif, Q23/QC440vif, and Q23/QF307vif are vif variants from other individuals without extensive hypermutation. (b) Example of a gag sequence from the Q23 culture (top) and two gag sequences from the QA039 culture (middle and bottom) for donor A. For each sequence, a region of both the forward and reverse sequencing chromatograms is shown. The positions in which G-to-A changes or mixed G-A peaks were identified are indicated in boxes. For QA039 hypermutated sequence 1, three positions with mixed peaks are indicated in the first two boxes. For QA039 hypermutated sequence 2, two G-to-A substitutions are indicated in the first and third boxes. Cultures were repeated for a subset of the vif variants in PBMC from two different donors infected at an MOI of 0.1, and plots of p24 versus day of culture are shown for donor B (c) and donor C (e). (d) Two hypermutated sequences from the QA039 culture for donor B are shown, with mixed peaks highlighted in the box. (f) One sequence each from the Q23, QA039, and Q23Δvif cultures for donor C. The box highlights a region in which two mixed peaks were found in QA039 and three in Q23Δvif.

Definition of hypermutation by two methods. In each plot, the APOBEC3G-specific score for each sequence is plotted against its general G→A score (see Materials and Methods). In panel a, sequences identified as hypermutated by k-means cluster analysis are indicated as gray squares, while those not hypermutated are black circles. In panel b, sequences identified as hypermutated by Hypermut 2.0 are indicated as gray squares, while those not hypermutated are black circles.

Alignment of Vif sequences. This alignment includes sequences from the two individuals with extensive hypermutation (QA039 and QB368) at both early (E) and chronic (C) infection time points, three other randomly chosen individuals in the present study (QA268, QC340, and QF307), and a subtype A reference strain from Kenya (Q23). All sequences are subtype A based on phylogenetic analysis (not shown) and are aligned against the subtype A consensus sequence (A con). Amino acid substitutions are indicated with capital letters, and positions with a mixture of amino acids are indicated with lowercase letters. The two polymorphisms of interest are highlighted in gray. *, Stop codon.

Evaluation of hypermutation throughout infection in QA039. The plot indicates this individual's viral load through time. Diamonds represent follow-up visits, and those at which sequences were examined are marked with large circles. The early infection and chronic infection time points used in the initial screen for hypermutation are marked by asterisks. The table below the graph shows the total number of gag proviral sequences that were examined at each time point, as well as the number of sequences that were found to be hypermutated.