All other Central African human strains clustered between and together with African simian strains in four different human subtypes. tree with PAML using a variant of the Tamura-Nei model allowing for different parameters in the LTR and env revealed no statistical difference in the estimates of the branch lengths (data not shown).
In addition, several simian clades had no human representative. When the latter data set was tested for substitution saturation using Dambe (Xia 2000), it was clear that no saturation could be observed when transitions and transversions were plotted versus evolutionary distance (fig. The plot shows that transitions and transversions increase linearly with increasing divergence between different PTLV strains.
Using the available strains, the HTLV-If subtype appears to have emerged within the last 3,000 years, and the HTLV-Ia, HTLV-Ib, HTLV-Id, and HTLV-Ie subtypes appear to have diverged between 21,100 and 5,300 years ago. PTLV-I has been associated with both malignant lymphoma and leukemia in humans (Yoshida, Miyoshi, and Hinuma 1982) and nonhuman primates (Miyoshi et al. The transition/transversion ratios used were scored using Puzzle, version 4.0 (Strimmer and von Haeseler 1997): 4.44 for the LTR alignment and 5.57 for the env analysis.
Interspecies transmissions, most probably simian to human, must have occurred around that time and probably continued later. To test the robustness of the NJ and mpars tree topologies, 1,000 bootstrap replicates were performed.
Since the molecular clock was not rejected and no evidence for saturation was found, a constant rate of evolution at these positions for all 33 HTLV-I and STLV-I strains was reasonably assumed. Since African STLV-I strains have been found to cluster tightly with African HTLV-I subtypes, except for HTLVIa, we were able to investigate the possible selective pressure due to intra- and interspecies transmission using synonymous versus nonsynonymous substitution ratios.
The spread of PTLV-I in Africa is estimated to have occurred at least 27,300 6 8,200 years ago. Materials and Methods Phylogenetic Analysis Phylogenetic trees were generated from the multiple alignments (made in Geneworks 2.5.1, Oxford Mo lecular Systems, United Kingdom) of the long-terminalrepeat (LTR) and env regions separately, using neighborjoining (NJ), maximum-parsimony (mpars), and maximum-likelihood (ML) (under the Tamura-Nei substitution model) methods implemented in the software package PAUP*, version 4.0b4a (Swofford 1998).
Phylogenetic analysis of the LTR region was performed on a 510-nt fragment and separately on a 522nt fragment of the gp21 of the env region using all available known EMBL/Gen Bank database STLV-I strain sequences (33 strains for LTR and 45 strains for env). 1997), only a few strains of each subtype, representing the highest divergence within these subtypes, were chosen to illustrate their relationship to the simian strains.
Transitions and transversions are then scored after applying Kimuras two-parameter method to correct for multiple hits.