**3. Structural Conformers of the Wild-Type HIV-1 RRE**

Since the discovery of Rev [48,49] and the RRE [26–28,50] about 30 years ago, a wealth of studies on the HIV-1 Rev-RRE system have advanced our understanding of the structural details of the system. A consensus secondary structure of the RRE, however, has been lacking. Although the minimal size of a functional RRE was initially reported to be 234 nt [28], later studies [23,31,32] indicated that a fully functional RRE is ~350 nt. Early studies on RRE secondary structure were performed with the subtype B HXB2 234 nt RRE [29,30] which was reported to assume a 5 SL structure. As NL4-3 increasingly become the standard HIV molecular clone studied in laboratories, subsequent structural studies of the RRE were performed mostly on this isolate. Interestingly, studies on the 351 nt NL4-3 RRE also supported a 4 SL conformer [23,35]. It was then speculated that the presence of a longer stem-loop I used might have facilitated the alternative structure. However, contrary to this speculation, Watt et al. [31] subsequently showed that the full-length RRE of RNA purified from HIV-1 NL4-3 virions adopts a 5 SL conformation. This study, therefore, showed that at least, in NL4-3 RRE, the longer stem-loop I was not the determinant of the alternative structure. This finding was further supported by another study that used SHAPE to show that an in vitro transcribed 232 nt NL4-3 RRE formed the alternative 4 SL structure [47]. Besides HXB2 and NL4-3, the secondary structure of another HIV-I molecular clone (ARV-2/SF2) RRE (354 nt), differing from NL4-3 sequence at 13 nucleotides, has also been reported to adopt a 5 SL like conformation. This variant differs from the canonical 5 SL structure in that the nucleotides to the left of the top stem region of SL I/I' base pair with nucleotides from the central loop, bridging SL-IV and SL-V and forming a stem region which opens into individual SL-IV and SL-V.

Since chemical probing techniques used in these studies provide ensemble-average structural information, it could not be excluded that RRE structural heterogeneity might give rise to such discrepancies. In-gel probing, in contrast, offers an alternative strategy to examine conformationally heterogeneous RNAs, providing that they can be separated by non-denaturing strategies. As an example, Kenyon et al. [51] applied in-gel SHAPE to define the structure of monomeric and dimeric species of the HIV-1 packaging signal RNA, supporting a structural switch model of RNA genomic dimerization and packaging [51]. In light of (i), HIV-2 RRE conformational heterogeneity (see later) and (ii), the observation that a single nucleotide alteration sufficed to stabilize the 5 SL HIV-1 RRE conformer [47], we showed that by extended non-denaturing polyacrylamide gel electrophoresis that the HIV-1 RRE could be resolved into two closely migrating species (Figure 3A) [52]. Subsequent in-gel SHAPE verified the slower migrating RNA as the 4 SL conformer and the faster migrating RNA as the 5 SL conformer (Figure 3B,C, respectively), suggesting that in vivo, the wild-type HIV-1 RRE could exist in a conformational equilibrium.

**Figure 3.** The HIV-1 RRE exists in a conformational equilibrium. (**A**) Following extended non-denaturing PAGE, slow and fast migrating RRE conformers were observed. Subjecting these RNAs to in-gel SHAPE defines these as 4 SL (**B**) and 5 SL conformers (**C**). Note that, despite their conformational heterogeneity, the topology of SL-II, the primary Rev binding suite, is preserved. Modified from Sherpa et al. [52]. The 232 nt HIV-1 RRE RNAs appended with a 3 structure cassette were prepared for analysis by in vitro transcription.

To investigate the function of these alternate HIV-1 RRE conformers, Sherpa et al. [52] created stable 4 SL and 5 SL RRE variants by in vitro mutagenesis to determine their Rev-RRE activity. The RRE Mutant M1 was predicted to disrupt base pairing at the base of the combined SL-III/IV structure in the 4 SL conformer but maintain base pairing in SL-IV in the 5 SL variant and, thus, likely to adopt only the latter structure. Conversely, the mutant RRE M3 was expected to disrupt the base pairing in both SL-III and SL-IV of the 5 SL structure but keep the combined SL-III/IV intact and, thus, likely adopt only a 4 SL conformation. As illustrated in Figure 4A,B, these predictions were borne out experimentally since the initial RRE population was resolved into two stable conformers. The SHAPE analysis indicated that the mutant RREs preserved the structure predicted by mutagenesis. More importantly, electrophoretic mobility shift experiments indicated that Rev binding to the mutant RREs was largely unaffected, demonstrating that their global topology had not been affected by mutagenesis. Growth competition assays were next performed in a T-cell line (SupT1) to assess whether these RRE

conformers conferred a selective growth advantage. In this heteroduplex tracking analysis strategy [53], viruses with a different RRE were added to the same culture of SupT1 cells at an equal multiplicity of infection (MOI) and allowed to replicate and spread throughout the cultures for several days, after which cellular DNA was recovered. Relative amounts of integrated pro-viral DNA produced by each virus was measured using a PCR-based heteroduplex tracking. Since sequence mismatches within the wt/M1 and wt/M3 heteroduplexes caused them to migrate differently from each other and the perfectly wt/wt homoduplex, each heteroduplex could be resolved by native gel electrophoresis. Figure 4C indicates that the stable 5 SL RRE conformer displayed a selective growth advantage over its stable 4 SL counterpart as well virus encoding the wt RRE. Supporting this observation, HIV gag/pol expression assays demonstrated that the stabilized 5 SL conformer was functionally superior to wt and the 4 SL RRE. Thus, structural plasticity of the RRE promoting similar [54] or different levels of Rev-RRE function demonstrates how the Rev-RRE regulatory axis might function as a "replication rheostat" rather than a simple on/off switch.

**Figure 4.** Alternative HIV-1 RRE conformers promote different rates of virus replication. (**A**) Conformer construction (see text). (**B**) Chemical acylation (SHAPE) confirms a 5 SL conformation of mutant M1 and a 4 SL conformation of mutant M3. (**C**) Heteroduplex tracking analysis. In both M1/wt and M1/M3 mutant co-infections, the stabilized 5 SL M1 conformer displays a replicative growth advantage. The full experimental background is provided in Reference [52].
