G-to-A hypermutation has been sporadically observed in human immunodeficiency virus type 1 (HIV-1) proviral sequences from patient peripheral blood mononuclear cells (PBMC) and virus cultures but has not been systematically evaluated. PCR primers matched to normal and hypermutated sequences were used in conjunction with an agarose gel electrophoresis system incorporating an AT-binding dye to visualize, separate, clone, and sequence hypermutated and normal sequences in the 297-bp HIV-1 protease gene amplified from patient PBMC. Among 53 patients, including individuals infected with subtypes A through D and at different clinical stages, at least 43% of patients harbored abundant hypermutated, along with normal, protease genes. In 70 hypermutated sequences, saturation of G residues in the GA or GG dinucleotide context ranged from 20 to 94%. Levels of other mutants were not elevated, and G-to-A replacement was entirely restricted to GA or GG, and not GC or GT, dinucleotides. Sixty-nine of 70 hypermutated and 3 of 149 normal sequences had in-frame stop codons. To investigate the conditions under which hypermutation occurs in cell cultures, purified CD4 ⁺ T cells from normal donors were infected with cloned NL4-3 virus stocks at various times before and after phytohemagglutinin (PHA) activation. Hypermutation was pronounced when HIV-1 infection occurred simultaneously with, or a few hours after, PHA activation, but after 12 h or more after PHA activation, most HIV-1 sequences were normal. Hypermutated sequences generated in culture corresponded exactly in all parameters to those obtained from patient PBMC. Near-simultaneous activation and infection of CD4 ⁺ T cells may represent a window of susceptibility where the informational content of HIV-1 sequences is lost due to hypermutation.

Background. Transmitted human immunodeficiency virus type 1 (HIV-1) drug resistance (TDR) mutations can become replaced over time by emerging wild-type viral variants with improved fitness. The impact of class-specific mutations on this rate of mutation replacement is uncertain. Methods. We studied participants with acute and/or early HIV infection and TDR in 2 cohorts (San Francisco, California, and São Paulo, Brazil). We followed baseline mutations longitudinally and compared replacement rates between mutation classes with use of a parametric proportional hazards model. Results. Among 75 individuals with 195 TDR mutations, M184V/I became undetectable markedly faster than did nonnucleoside reverse-transcriptase inhibitor (NNRTI) mutations (hazard ratio, 77.5; 95% confidence interval [CI], 14.7–408.2; P < .0001), while protease inhibitor and NNRTI replacement rates were similar. Higher plasma HIV-1 RNA level predicted faster mutation replacement, but this was not statistically significant (hazard ratio, 1.71 log₁₀ copies/mL; 95% CI, .90–3.25 log₁₀ copies/mL; P = .11). We found substantial person-to-person variability in mutation replacement rates not accounted for by viral load or mutation class (P < .0001). Conclusions. The rapid replacement of M184V/I mutations is consistent with known fitness costs. The long-term persistence of NNRTI and protease inhibitor mutations suggests a risk for person-to-person propagation. Host and/or viral factors not accounted for by viral load or mutation class are likely influencing mutation replacement and warrant further study.

Recombination is an important way to generate genetic diversity. Accumulation of HIV-1 full-length genomes in databases demonstrated that recombination is pervasive in viral strains collected globally. Recombinant forms achieving epidemiological relevance are termed circulating recombinant forms (CRFs). CRF12_BF was up to now the only CRF described in South America. The objective was to identify the first CRF in Brazil conducting full genome analysis of samples sharing the same partial genome recombinant structure. Ten samples obtained from individuals residing in Santos, Brazil, sharing the same recombination pattern based on partial genome sequence data, were selected from a larger group to undergo full length genome analysis. Near full length genomes were assembled from overlapping fragments. Mosaic genomes were evaluated by Bootscan, alignment inspection, and phylogenetic analysis using neighbor joining and maximum likelihood. Full genomes were also analyzed by split decomposition. We were able to identify five mosaic genomes. Two of these structures were represented by at least three samples derived from epidemiologically unlinked individuals. These structures were named CRF28_BF and CRF29_BF and are the second and third CRFs composed exclusively by subtypes B and F as well as the second and third CRFs encountered in South America. Other recombinant forms studied here resembled CRF28_BF and CRF29_BF. Our results suggest that a diverse population of related recombinants, including CRFs may play an important part in the Brazilian and South American epidemic.

This study characterized HIV-1 among antiretroviral-naïve populations presenting recent infection (RI) or long-standing infection (LSI). Sera collected from January 1999 to December 2001 at an anonymous HIV testing site in Santos, Brazil, were submitted to serologic testing algorithm for recent HIV seroconversion (STARHS). The STARHS methodology uses a combination of a sensitive and a less sensitive version of an anti-HIV enzyme immunoassay (EIA), and specimens found to be positive on the sensitive EIA and negative on the less sensitive EIA are considered to represent RI. HIV-1 V3 and pol regions of those with RI and LSI were compared. Antiretroviral resistance was defined solely by genotypic analysis. Ninety samples were evaluated representing those taken from an original cohort of 345 individuals, for whom adequate samples were available. Of 90 HIV-positive individuals, 25 presented RI. Cumulatively, 36.8% of those with RI and 25% of those with LSI presented resistance to at least one antiretroviral class. In the pol and V3 regions, 47% and 53% of those with RI presented clade B viruses and B/F recombinant viruses, respectively, whereas 56.2%, 41.7%, and 2.1% of those with LSI harbored clades B, B/F, and clade C viruses, respectively. Primary resistance and the prevalence of B/F recombinants was high in this population. Monitoring HIV-1 genetic diversity is important for developing vaccines and treatment strategies.