Large-scale prediction shows that the dominant structure of the HIV-1 domain closed by the U5-AUG duplex contains the alternative SD hairpin, and the domain variant without SD is rare.

Using several models of the HIV-1 5' leader, it was shown that the domain containing the structural elements that regulate the processes of dimerization and genome packaging, as well as the initiation of reverse transcription, is closed by the U5-AUG duplex. However, there is no consensus in the literature on the structure of the upper part of this domain. Currently, the model proposed by Keane et al. in 2015 is dominant, although the question of whether it is general structure or specific to the experimental HIV-1 genome NL4-3 of subtype B remains open. To clarify this issue, we conducted large-scale in silico studies on the secondary structure of the domain closed by the U5-AUG duplex in 2754 HIV-1 genomes of different subtypes. Our investigation showed that the proportion of HIV-1 genomes in which the structure of the domain under study is similar to that in Keane et al. model is low. It forms mainly in HIV-1 genomes of subtype B with the frequency of 3.8 % in the optimal foldings or foldings with the energy increment of the lowest change in free energy (ΔΔG)<1.0 kcal/mol. In particular, certain base changes in common SD hairpin or base changes stabilizing Psi hairpin contribute to the formation of this domain variant. The dominant structure of the domain closed by the U5-AUG duplex is similar to that in Wilkinson et al. model (2008) but with the alternative SD hairpin. We found also new variants of this domain, which occur in foldings with ΔΔG<1.0 kcal/mol and may co-exist with dominant structure. However, it is possible that the variants of the domain closed by the U5-AUG duplex similar to Wilkinson et al. or Keane et al. models are formed only in the early stages of HIV-1 replication, while in the late stage (in the presence of nucleocapsid protein) the domain adopts structure similar to that in Sakuragi et al. (2012) model and the initiation of the reverse transcription occurs just in this structure. Extreme conservation of GACGC-GCGUC duplex, proposed in Sakuragi et al. model, supports this assumption.