The kissing-loop domain (KLD) encompasses a stem-loop, named kissing-loop or dimerization initiation site (DIS) hairpin (nucleotides [nt] 248 to 270 in the human immunodeficiency virus type 1 strains HIV-1 Lai and HIV-1 Hxb2 ), seated on top of a 12-nt stem-internal loop called stem-loop B (nt 243 to 247 and 271 to 277). Destroying stem-loop B reduced genome dimerization by ϳ50% and proviral DNA synthesis by ϳ85% and left unchanged the dissociation temperature of dimeric genomic RNA. The most affected step of reverse transcription was plus-strand DNA transfer, which was reduced by ϳ80%. Deleting nt 241 to 256 or 200 to 256 did not reduce genome dimerization significantly more than the destruction of stem-loop B or the DIS hairpin. We conclude that the KLD is nonmodular: mutations in stem-loop B and in the DIS hairpin have similar effects on genome dimerization, reverse transcription, and encapsidation and are also "nonadditive"; i.e., a larger deletion spanning both of these structures has the same effects on genome dimerization and encapsidation as if stem-loop B strongly impacted DIS hairpin function and vice versa. A C258G transversion in the palindrome of the kissing-loop reduced genome dimerization by ϳ50% and viral infectivity by ϳ1.4 log. Two mutations, CGCG2613UUAA261 (creating a weaker palindrome) and a ⌬241-256 suppressor mutation, were each able to reduce genome dimerization but leave genome packaging unaffected.The kissing-loop domain (KLD) encompasses a stem-loop, named kissing-loop hairpin (nucleotides [nt] 248 to 270 in human immunodeficiency virus type 1 strain HIV-1 Lai and HIV-1 Hxb2 genomic RNA), seated on top of a short steminternal loop called stem-loop B (nt 243 to 247 and 271 to 277) (18). The apical loop of the kissing-loop hairpin contains an almost invariant hexameric autocomplementary sequence (ACS) (see reference 17 and references therein), also called a palindrome. The palindrome is seen as the dimerization initiation site (DIS) of genomic RNA (13, 15, 31); thus, the kissingloop hairpin is also called the DIS hairpin. The level of genomic RNA dimerization within isolated HIV-1 viruses is influenced by the DIS hairpin (6, 9, 17) and p55Gag processing (8).In the kissing-loop model of HIV-1 genome dimerization (13, 15, 31), stem-loop B has ill-defined roles (15, 17); one might be to properly orient the DIS hairpin within the covalently linked 9,000-nt-long tangle of secondary and tertiary structure (18). Experimentally, substantial deletions within stem-loop B or the DIS hairpin have identical impacts on viral infectivity and genomic RNA encapsidation (18). This raises the possibility that the KLD might be nonmodular, i.e., a highly integrated structure whereby stem-loop B and the DIS hairpin may have similar, if not identical, physiological impacts. To establish this, it is necessary to show that stem-loop B mutations inhibit genomic RNA dimerization and proviral DNA synthesis, two processes affected by the DIS hairpin (6,9,17,25).In this paper, we identify a crucial role of stem-loo...

The genome of all retroviruses consists of two identical RNAs noncovalently linked near their 5' end. In vitro synthesized RNAs from human immunodeficiency virus type 1 (HIV-1) can form loose or tight dimers depending on whether their respective kissing-loop hairpins (nts 248-270 in HIV-1Lai) bond via their hexameric autocomplementary sequences (ACS), also called palindromes, or via the ACS and stem sequences [Laughrea, M., and Jetté, L. (1996) Biochemistry 35, 1589-1598]. To understand the role of the ACS in HIV-1 replication and in the formation and stability of HIV-1 RNA dimers, we replaced the central CGCG261(or tetramer) of the HIV-1Lai ACS by two other HIV-1 tetramers (UGCA/UGCG), four non-HIV-1 tetramers [GUAC, UUAA (respectively found in HIV-2Rod and SIVmnd), GGCC and AGCU (absent from HIV and SIV viruses)], or GGCG, a nonpalindromic tetramer. The infectivity of GGCC, GUAC, and UGCA viruses was unchanged or insignificantly decreased; the infectivity of AGCU and UGCG viruses was decreased by 80%; the infectivity of UUAA and GGCG viruses was decreased by 92-98%. Thus, the four non-HIV-1 palindromes yielded phenotypes ranging from wild-type to as defective as a virus bearing a nonpalindrome. Studies of in vitro synthesized HIV-1 RNAs were generally consistent with in vivo results, specifically: (i) loose dimerization of GGCC and GUAC RNAs, but not of UUAA and AGCU RNAs, was influenced by the 3' DLS (a sequence located downstream of the 5' splice junction) in a way expected for a wild-type ACS; (ii) the 3' DLS strongly reduced tight dimerization of UUAA and AGCU RNAs, but not of GGCC and GUAC RNAs. We conclude that HIV-1 is sensitive to the ACS sequence without discriminating against all nonnative ACS: GGCC/GUAC, but not AGCU/UUAA, are good substitutes for the prevalent CGCG/UGCA native tetramers and better substitutes than the very rare UGCG native tetramer. The correlation between in vivo and in vitro results suggests that in vitro assays measure parameters of in vivo relevance. Deletion of CUCGG247 (the 5' strand of stem-loop B) decreased the replicative capacity by more than 99.9% and metamorphosed the 3' DLS into an inhibitor of the loose dimerization of HIV-1 RNA.

Genomic RNA isolated from HIV-1 variously mutated in nucleocapsid protein (NC) was characterized by nondenaturing gel electrophoresis. Mutations in the C-terminal, the N-terminal, and the linker regions had no effect on genomic RNA dimerization [they are R7R10K11S, P31L, R32G, S3(32-34), and K59L], while a C36S/C39S mutation in the distal zinc knuckle (Cys-His box or zinc finger) inhibited genome dimerization as much as disrupting the kissing-loop domain. The four mutations which inhibited tRNA(Lys3) genomic placement (i.e., the in vivo placement of tRNA(Lys3) on the primer binding site) had no effect on genome dimerization. Among five mutations which inhibited genome packaging, four had no effect on genome dimerization. Thus the N-terminal and linker regions of NC control genome packaging/tRNA(Lys3) placement (two processes which do not require mature NC) but have little influence on genome dimerization and 2-base extension of tRNA(Lys3) (two processes which are likely to require mature NC). It has been suggested, based on electron microscopy, that the AAGCUU82 palindrome crowning the R-U5 hairpin stimulates genomic RNA dimerization. To test this hypothesis, we deleted AGCU81 from wild-type viruses and from viruses bearing a disrupted kissing-loop hairpin or kissing-loop domain; in another mutant, we duplicated AGCU81. The loss of AGCU81 reduced dimerization by 2.5 +/- 4%; its duplication increased it by 3 +/- 6%. Dissociation temperature was left unchanged. We reach two conclusions. First, the palindrome crowning the R-U5 hairpin has no impact on HIV-1 genome dimerization. Second, genomic RNA dimerization is differentially influenced by NC sequence: it is Zn finger dependent but independent of the basic nature of the N-terminal and linker subdomains. We propose that the NC regions implicated in 2-base extension of tRNA(Lys3) are required for a second (maturation) step of tRNA placement. Genome dimerization and mature tRNA placement would then become two RNA-RNA interactions sharing similar NC sequence requirements.