The fact that CRF02_AG seems largely predominant over other circulating HIV strains in an African area with extremely high HIV genetic diversity may have several explanations. First, the recombinant form may have some biological advantage over the parental strains, including a possibly higher replicative fitness and/or transmission capacity. Second, the recombinant strain could have been introduced first in that particular area and consequently get established in a population before other subtypes entered the scene (founder-effect) . In the case of the epidemiological spread of CRF02_AG in West Central Africa, the founder hypothesis is probably a less likely explanation. Several studies on the prevalence of HIV-1 subtypes in the Democratic Republic of Congo (DRC) have shown that subtypes A and G are relatively prevalent in this area [34–37]. Moreover, it is likely that at least a limited spread of subtype A and G viruses must have preceded the creation and spread of CRF02_AG in West Central Africa .
In the present study, we explored whether the replicative fitness of CRF02_AG was related to the epidemiological spread of this virus in extended areas of West and Central Africa. We showed that CRF02_AG primary isolates had a higher replicative fitness compared to subtype A and G isolates, in a cellular model for HIV pathogenesis (i.e. activated T-cells) and HIV transmission (i.e. MO-DC). The higher relative replicative fitness of CRF02_AG viruses was evident for isolates from patients with low (<200 cells/μL) and with higher (>200 cells/μL) CD4+ T cell counts, and it was found to be independent of the viral co-receptor use. An independent study investigating the same hypothesis was published recently and also showed an increased replicative capacity of CRF02_AG viruses compared to subtype A and G isolates, using basic virus growth kinetics as a measure of replication capacity . In contrast to our study, Konings et al.  studied only thirteen HIV-1 isolates and presented limited data on viral load and CD4+ T cell counts. Furthermore, the growth competition assays used in our study are able to discriminate minor differences in replication capacity and also provide the internal control lacking in monoinfection kinetic assays, as used by Konings et al. [38–40].
The viral load in the donor and the integrity of mucosal tissues in the acceptor are amongst the most important determinants upon HIV transmission [41, 42]. Previous observations by Fischetti et al. , showed significantly higher viral loads in asymptomatic CRF02_AG infected individuals compared to patients infected with non-CRF02_AG strains. A direct correlation between viral load and replicative capacity in activated T cells was repeatedly shown [30, 31]. Taken together with our observations, these data suggest that patients infected with CRF02_AG strains may more easily transmit virus, because of a higher viraemia, which could be a consequence of the higher replicative fitness in activated T cells. This interpretation is consistent with the observation by Ariën et al. , who previously showed that group M viruses in general have a much higher in vitro replicative fitness than group O or HIV-2 viruses, corresponding to the relative spread of these viruses in the pandemic as a whole and in West Africa (where they all co-circulate) in particular.
One could argue that the observed higher relative fitness of CRF02_AG strains versus subtype A and G isolates in the present study simply reflects a more advanced disease stage of patients infected with CRF02_AG or a differential viral co-receptor tropism. However, we have shown that CRF02_AG with either X4 or R5 co-receptor tropism and derived from patients with more or less advanced disease (based on CD4+ T cell count) are on average more fit than subtype A or G viruses (Figure 2). In addition, our data suggests that the replicative fitness of CRF02_AG in MO-DC was slightly, but not significantly higher than the parental subtypes (A and G) (Figure 3). There is substantial evidence that DC play an important role during HIV transmission and it could be speculated that a slight advantage in replicative fitness in dendritic cells may have an important impact on transmission at the population level. On the other hand, the number of competitions performed in DC may just have been too low to result in a significant difference.
Studies by Ball et al.  and Ariën et al.  showed that viruses of subtypes B and C were equally fit in Langerhans' dendritic cells, while subtype C isolates were out competed by any other group M virus in activated PBMC. It is not completely clear yet how HIV replicative fitness in T cells and dendritic cells relate to transmission and epidemiological spreading. It is also possible that the focus on replicative capacity in DC as a measure of transmission efficiency may be too limited, since other cell types at the mucosal interface are likely involved in transmitting HIV. Hence a better model to study HIV transmission is desirable and should include Female Genital Tract (FGT) epithelia and other important target cells, such as T cells and macrophages, in addition to DC [41, 42]. We are currently elaborating on such models in order to study early events during HIV transmission and replicative fitness.
The CRF02_AG genome is a mosaic of subtype A (gag, vpr and parts of pol, env and nef) and subtype G (LTR, rev, tat and parts of pol, env, and nef). An important question that needs to be answered is which part of the viral genome may be responsible for the increased replicative fitness of CRF02_AG. Unfortunately, our experimental set up did not allow us to study the contribution of individual genes to the overall replicative fitness of a virus isolate. Therefore, future studies should try to elucidate the role of those genes that have a mosaic appearance for their impact on the fitness of the recombinant virus. It is clear that recombination occurs often in dual- or super infected individuals, generating ISR. It could be speculated that those ISR that generate viable progeny subsequently undergo severe selection pressure by the innate and adaptive host immune responses and that only the most successful/fit ISR may eventually be able to spread epidemically and become a CRF.
More detailed analyses of HIV samples from West Africa have shown that CRF02_AG has already undergone further recombination [34, 44]. Clearly, viral recombination is inevitable with the continuous intermixing of HIV subtypes and will have its impact on the evolution of the HIV epidemic. It is important to envisage that a CRF that we label as very fit today may be out competed by a new and even more fit recombinant virus tomorrow.
In conclusion, our data on a small, but carefully selected sample from a Cameroonian cohort clearly suggests that the prevailing CRF02_AG recombinant may be favoured in his spread over "parental" subtype A and G viruses as a result of a higher replicative fitness in T cells and likely also in dendritic cells. More extensive and in-depth studies are needed to confirm this preliminary evidence and to unravel the molecular mechanisms underlying the predominance of CRF02_AG in large parts of West Africa.