Recombinant viruses and early global HIV-1 epidemic

Abstract

Central Africa was the epicenter of the HIV type 1 (HIV-1) pandemic. Understanding the early epidemic in the Democratic Republic of the Congo, formerly Zaire, could provide insight into how HIV evolved and assist vaccine design and intervention efforts. Using enzyme immunosorbent assays, we tested 3,988 serum samples collected in Kinshasa in the mid-1980s and confirmed seroreactivity by Western blot. Polymerase chain reaction of gag p17, env C2V3C3, and/or gp41; DNA sequencing; and genetic analyses were performed. Gene regions representing all the HIV-1 group M clades and unclassifiable sequences were found. From two or three short gene regions, 37% of the strains represented recombinant viruses, multiple infections, or both, which suggests that if whole genome sequences were available, most of these strains would have mosaic genomes. We propose that the HIV epidemic was well established in central Africa by the early 1980s and that some recombinant viruses most likely seeded the early global epidemic.

Figure 1

Neighbor-joining tree of 56 Zaire strains from the mid-1980s, 197 Democratic Republic of the Congo (DRC) strains from 1997, and subtype-specific reference strains. The number of nucleotides in the final alignment was 304 bp after gap stripping. The model of evolutionary change used in the tree was the Transversional Model (TVM) +g (0.8575), where g is the shape parameter of the γ distribution (heterogeneity among sites) and TVM is the model of substitution whereby A↔G = C↔T and the other four rates, all transversions, are unique. Our 1980s C2V3C3 sequences are indicated by a five-digit code, and their branches are red; the 1997 sequences have the subtype (as indicated in GenBank) as the first letter, followed by a one- or two-digit number or the designation unk = unclassifiable, and these branches are black; the reference strains have the subtype as the first letter followed by the reference name as follows: A.SE7253, A.92UG037, A2.CDK, A2.Cy017; B.HXB2, B.MN; C.ETH2220, C.IN21068; D.84ZR085, D.NDK, MAL; E.CD402, E.CM240, E.TH253; F1.FIN9363, F1.93BR020; F2.CMMP255, F2.CMMP257, G.NG083, G.SE6165; H.90CF056, H.VI991, J.SE91733, J.SE92809; K.96CMMP253, and K.97REQTB; these branches are purple. Subtype characterizations of the phylogenetic clusters in the tree are indicated in bold, lineages consisting of unclassifiable sequences are designated as U.

Figure 2

Distribution of subtypes and recombinant viruses. The pie chart represents 66 strains for which sequences from at least 2–3 gene regions were available for comparison; the subtypes in the pie chart represent concordant phylogenies suggestive of possible "pure" subtypes; the CRF01 and unique recombinant viruses are indicated in the pie chart. Table 2 summarizes subtypes of unique recombinant viruses.

Figure 3

Hypothetical model of HIV-1, group M evolution. A. Star phylogeny representing the evolution of the ancestral HIV-1, group M virus that was able to adapt in humans and was transmitted among rural populations in central Africa from approximately the 1930s (22). Over time, the viruses would have become increasingly genetically distinct from each other and the original parental strain. The dotted circle denotes the beginning of migration from these remote areas to cities in central Africa (approximately 1960–1970). B. Recombinant lineages (outside the dotted circle), represented by multicolored lines indicating mosaic viruses or genetic mixes of the circulating strains, would have been the result of population migration, urbanization, patterns of sexual activity, and medical practices (two of the oldest, fully characterized sequences, MAL [1985] and Z321 [1976], were both recombinant viruses from Zaire). Recombinant viruses would have continued to be generated and transmitted until introduced into high-risk populations, such as commercial sex workers, taxi drivers, commercial truck drivers, or long-distance truck drivers, and then rapidly transmitted within and between these social networks. Such high-risk social networks throughout central Africa were responsible for the rapid expansion of a relatively small number of evolving viruses, including recombinant strains, locally, regionally, and eventually globally. These epidemiologic groupings are represented as clusters of highly related strains at the end of a few HIV-1 lineages. Panel C shows what phylogenetic analysis of global HIV-1 strains collected in the early 1990s, when sequence characterization first began, and after being exported out of central Africa, would have looked like. The clusters of related sequences from founder viruses, which were disseminated globally, would have appeared as subtypes or clades, arbitrarily labeled a–e. Occasionally, strains that were not widely expanded were identified and designated as unclassifiable. From this hypothetical modeling, and the high numbers of recombinant strains, it seems unlikely that only pure subtypes were exported from this region of Africa to establish mini-epidemics in other countries. Therefore, at least some of what we currently define as pure subtypes most likely arose from recombinant genomes originally generated somewhere in central Africa.