Nearly 35 million people across the globe are infected with the human immunodeficiency virus type 1 (HIV-1) and an additional 2.5 million new infections occur annually 
. The current pandemic is the result of viruses that are genetically diverse and have been divided into 9 different subtypes (A–D, F–H, J, K) and at least 28 circulating recombinant forms (www.hiv.lanl.gov
). Subtype B viruses predominate in North America and Western Europe, and as a result, research efforts have focused primarily on this subtype. However, almost 25 million individuals in sub-Saharan Africa have been infected with non-subtype B viruses via heterosexual transmission, leading to a significant gap in our understanding of the most predominant HIV-1 variants worldwide. In order to achieve protection against infection with a globally effective vaccine, the origin and nature of variants that establish a new infection must be better defined.
The mechanisms by which HIV-1 is transmitted sexually across a mucosal barrier remain poorly understood. Characterization of viral sequences in newly infected individuals has produced mixed results. For instance, studies of individuals newly infected with subtype B have demonstrated a relatively homogenous viral population, suggesting infection by a single variant 
. In contrast, studies involving subtype-B virus transmission in a population harboring sexually-transmitted diseases and subtype-A virus in a sex-worker cohort suggest multiple viral variants can be transmitted and are capable of establishing infection in a new host 
. Thus, the viral correlates of transmission are likely to be influenced by subtype, study population, and route of infection. Since only the newly transmitted virus population was characterized in the majority of these studies, no information was available regarding its relationship to the virus circulating in the transmitting partner. Nevertheless, these studies indicate that either a single viral variant or multiple variants can be transmitted and establish a new HIV infection.
A previous study in this laboratory examined 8 epidemiologically linked heterosexual transmission pairs participating in a HIV-discordant couple cohort in Lusaka, Zambia 
. Seven of these pairs harbored subtype C viruses, while the remaining pair was infected with subtype G. In each transmission pair, comparison of the virus population derived from the chronically infected (donor) and newly infected partners (recipient) revealed a severe genetic bottleneck during heterosexual transmission of HIV, which was characterized by low sequence diversity in the recipient. Moreover, the monophyletic nature of each recipient virus population relative to the donor quasispecies indicated that each was derived from a single variant within the donor.
Here we have expanded our studies of HIV-1 transmission to include 12 additional transmission pairs from the Lusaka cohort and 8 pairs from a similar cohort in Kigali, Rwanda 
. The predominant circulating subtypes in Lusaka and Kigali are subtype C and subtype A respectively, providing a unique opportunity in which to investigate the correlates of heterosexual transmission in regions where the two most predominant HIV-1 subtypes circulate 
. Each cohort is comprised of over 1,000 cohabitating HIV-discordant couples enrolled in a prospective prevention study in which participants return at 3-month intervals for preventive counseling and condom provision. Despite these interventions, a low frequency of HIV-1 transmission still occurs 
. In the present study, plasma samples from HIV seronegative partners were tested at each visit for the appearance of antibodies to HIV-1 as well as for the presence of p24 antigen to identify acutely infected individuals in whom virus is present in the peripheral blood but antibody levels are still undetectable 
. Moreover, the studies presented here utilized end-point dilution PCR (or single genome amplification (SGA)) to compare partial env
gene sequences within the quasispecies of the donor and recipient. To investigate transmission in a setting analogous to those that are associated with transmission of multiple variants, we included a limited number of epidemiologically unlinked transmission events, in which the seronegative partner was infected by someone other than their spouse. The results indicate that, in the majority of cases, HIV-1 infection is initiated by a single viral variant from a complex quasispecies in the transmitting partner. The marked reduction in genetic diversity that is observed during HIV-1 transmission appears to be imposed in large part by the mucosal barrier, since this extreme genetic bottleneck can be mitigated when inflammatory infections are present in the recipient partner.