NMR Detection of Structures in the HIV-1 5´-Leader RNA that Regulate Genome Packaging

The 5´-leader is the most conserved region of the HIV-1 genome and is responsible for regulating multiple activities during viral replication, including RNA encapsidation during virus assembly (1, 2). Like all retroviruses, HIV-1 packages two copies of its genome, enabling strand-transfer mediated recombination during reverse transcription and promoting genetic evolution under environmental and chemotherapeutic pressure (3). Dimeric genomes are trafficked from the cytoplasm to plasma membrane assembly sites by a small number of viral Gag proteins, where additional Gag proteins assemble and budding occurs (4–8). Dimerization and packaging are mediated by interactions between the nucleocapsid (NC) domains of the viral Gag polyproteins and RNA elements within the 5´-leader of the genome, and there is evidence that these activities and translational control are mechanistically coupled (7, 9–13).

Understanding of the mechanisms that regulate 5´-leader activities is limited, due in part to incomplete knowledge of the leader structure (14). Recombinant leader-containing RNAs have been probed by nucleotide accessibility mapping, mutagenesis, and biochemical approaches, and although there is general consensus that transcriptional activation, primer binding, dimerization, and splicing activities are promoted by discrete hairpin structures (10, 15–23) (TAR, PBS, DIS and SD hairpins, respectively; Fig. 1), there is less agreement regarding the structures that regulate packaging (14). In particular, residues overlapping the gag start codon (G328-A356, AUG) that are critical for genome packaging (24) and RNA dimer stability (25) have been proposed to form a hairpin (Fig. 1B), to base-pair with residues of the upstream Unique-5´ element (U5, Fig. 1C) (10, 20), or to adopt other conformations (14). In vivo nucleotide reactivity mapping has supported multiple AUG models, without consensus (14, 21–23). NMR is potentially well suited to probe RNA structure, but signal degeneracy and relaxation problems have thus far limited applications to relatively small oligonucleotides (typically fewer than 50 residues) (26, 27). Here we describe an NMR approach that enabled direct detection of structures formed by AUG and other packaging elements within the intact, 230 kDalton dimeric 5´-leader, as well as the identification of conformational changes that regulate dimerization, NC binding and packaging.

A 356 nucleotide HIV-1_NL4-3 5´-leader RNA that includes the entire 5´-Untranslated Region (5´-UTR) and the first 21 residues of gag (5´-L) was prepared by enzymatic ligation of non-labeled 5´-RNA (residues 1–327) and ¹³C-enriched AUG fragments (residues 328–356) (Fig. 1) (28). The 5´-leader exists predominantly as a monomer at low ionic strength (Fig. 1D), and under these conditions the AUG residues of 5´-L gave rise to ¹H-¹³C correlated HMQC (28) NMR spectra similar to those observed for the isolated AUG fragment, which is known to form a hairpin (29) (Fig. 1E,F). NMR line widths of AUG in the intact 5´-leader and the isolated AUG RNA were also similar, indicating that the hairpin is structurally mobile and does not form A-minor-like contacts (30).

Significant changes in the ¹H-¹³C HMQC spectra of AUG-labeled 5´-L were observed upon incubation at physiological ionic strength (PI buffer: 10 mM Tris-HCl, 140 mM KCl, 10 mM NaCl, 1 mM MgCl₂, pH 7.0), which correspond to a reversible equilibrium shift from a predominantly monomeric species to a mixture of monomeric and dimeric species (Fig. 1G,H; dimer dissociation constant K_d = 0.9 ± 0.1 µM). Signals of the AUG hairpin in 5´-L exhibited uniformly reduced intensities as a function of incubation time, and new signals were observed for the 3´-residues of AUG (A345-A356) (Fig. 1H). Similar NMR spectral changes were observed for an isolated AUG oligo-RNA upon titration with an oligo-U5 fragment (Fig. 1I). The NMR chemical shifts and narrow line widths observed for A345-A356 indicate that these residues are unstructured and mobile in the dimeric form of the leader ([5´-L]₂). No ¹H-¹³C HMQC signals were detected for G328-G344 of [5´-L]₂ due to severe line broadening (discussed below). These findings are consistent with a phylogenetically-derived structural model (10, 20), in which the 5´-residues of AUG base pair with U5 and the 3´-residues are disordered (Fig. 1C).

To directly probe for U5:AUG base pairing, we developed an NMR approach that involves replacement of a short stretch of adjacent base pairs by A-U base pairs (lr-AID; long-range probing by Adenosine Interaction Detection). The substituting element (ideally [U_iU_jA_k]:[U_lA_mA_n], but other sequences can suffice) (Fig. 2A) affords an upfield-shifted A_m-C2-⁻¹H NMR signal (~6.5 ppm), enabling direct detection of cross-strand A_k-H2 and -H1´ ¹H-¹H nuclear Overhauser effects (NOEs) without the need for heteronuclear spectral editing. The HIV-1_NL4-3 leader naturally contains one [UUA:UAA] element in the TAR hairpin ([U12-A14]:[U45-A47]), and to preclude signal overlap, A46 was mutated to G (a naturally occurring substitution in 7% of the reported HIV-1 TAR sequences (31)) (Fig. 2B–D). NOESY spectra (28) obtained for the lr-AID modified dimeric 5´-leader ([5´-L^{lr-AID-U5:AUG}]₂, C110-G112 and G338-G339 substituted by UUA and AA, respectively; Fig. 2A) exhibited well-resolved A338-H2 signals with frequencies similar to those observed for an isolated lr-AID modified U5:AUG oligoribonucleotide (Fig. 2E,F). Other outlier signals in the NOESY spectrum were unaffected by the lr-AID substitutions. All of the expected A338-H2 inter-adenosine NOEs were observed, including cross-strand NOEs with A112-H1´ and -H2. The A338-H2 T₁ and T₂ ¹H NMR relaxation rates (8.7 sec and 9.4 msec, respectively) are consistent with restricted rotational motion and indicate that U5:AUG resides within a globular region of the 5´-L structure.

The relationship between U5:AUG formation and dimerization was investigated by site-directed mutagenesis. Mutations in AUG designed to stabilize the hairpin and disrupt base pairing with U5 (5´-L^AUG-HP) favored the monomer, whereas mutations that promote U5:AUG base pairing and destabilize the hairpin ([5´-L^U5:AUG]₂) favored the dimer (Fig. S1). Deletion of AUG (5´-L^ΔAUG) also favored the monomer (Fig. 3A), whereas incubation of 5´-L^ΔAUG with a 17-nucleotide AUG oligoribonucleotide (G328-G344; AUG-17) promoted dimerization (Fig. 3B). These findings indicate that dimerization is suppressed when AUG exists in a hairpin conformation (AUG^HP), that U5:AUG formation promotes dimerization, and that dimerization is induced by intramolecular U5:AUG base pairing rather than intermolecular tethering.

NMR chemical shifts and NOEs associated with A268-H2 in 5´-L, 5´-L^ΔAUG, and an isolated DIS oligo-RNA were similar, indicating that the attenuated dimerization activity of the AUG^HP form of 5´-L is not due to re-folding of the DIS (32). Sequence complementarity between U5 and the GC-rich loop of the DIS suggested that dimerization may instead be inhibited by U5:DIS base pairing. Consistent with this hypothesis, mutations in 5´-L designed to enhance U5:DIS base pairing promoted monomer formation (Fig. 3C,D), whereas mutations to disrupt U5:DIS base pairing induced dimerization of 5´-L^ΔAUG (Fig. S2). In addition, NMR data obtained for an lr-AID substituted 5´-L^ΔAUG RNA with 5´-L^ΔAUG-like dimerization and NC binding properties (Fig. S4) exhibited cross-strand NOEs and NMR chemical shifts consistent with the predicted U5:DIS interface (Fig. 3E–G).

To determine if U5:AUG formation influences NC binding, isothermal titration calorimetry (ITC) experiments were performed with 5´-L and mutant 5´-L^AUG-HP and 5´-L^U5:AUG RNAs. Under conditions of the ITC experiments, 5´-L^AUG-HP and 5´-L^U5:AUG exist predominantly as monomers and dimers, respectively, and native 5´-L exists as a 70:30 monomer:dimer equilibrium mixture (Fig. 4A). All three RNAs gave rise to two-component NC binding isotherms that included an initial exothermic event associated with high-affinity NC binding, and a subsequent endothermic event attributed to NC-induced RNA unfolding at high NC:RNA ratios (33) (Fig. 4B and Fig. S3). Whereas 5´-L^AUG-HP binds 7 ± 1 NC molecules with high affinity (K_d = 87 ± 3 nM), [5´-L^U5:AUG]₂ binds 32 ± 2 NC molecules with high affinity (K_d = 71 ± 3 nM) (Fig. 4B). The ITC profile observed for native 5´-L was similar to that obtained for a 70:30 mixture of 5´-L^AUG-HP and 5´-L^U5:AUG (Fig. 4A,B), validating the use of 5´-L^AUG-HP and 5´-L^U5:AUG as models for the native monomer and dimer, respectively. Importantly, the NC binding and dimerization properties of 5´-L^{lr-AID-U5:AUG} and 5´-L^{lr-AID-U5:DIS} were similar to those of the 5´-L^U5:AUG and 5´-L^AUG-HP controls, respectively (Fig. S4), further indicating that the lr-AID substitutions did not alter the structure of the RNA. Substitution of the dimer-promoting GC-rich loop of the DIS hairpin by a GAGA tetraloop prevented dimerization of 5´-L^U5:AUG but did not affect its NC binding properties (Fig. S5), indicating that enhanced NC binding by the U5:AUG form of the 5´-leader is due to intramolecular conformational changes associated with U5:AUG base pairing and not to dimerization per se.

The above findings suggested that in vivo RNA packaging should be dependent on U5:AUG formation, and we therefore measured packaging efficiencies of vector RNAs containing 5´-L^AUG-HP and 5´-L^U5:AUG mutations in competition experiments. An HIV-1_NL4-3 helper construct that expressed the native 5´-leader and all viral proteins except Env was co-expressed with test vector RNAs with 5´-L^AUG-HP or 5´-L^U5:AUG mutations (Fig. 4C). Consistent with the structural and NC binding studies, the 5´-L^U5:AUG RNAs were packaged nearly as avidly as the native construct (Fig. 4C; lane 5) whereas the 5´-L^AUG-HP RNAs exhibited severe packaging defects (Fig. 4C; lane 6).

Previously observed packaging defects associated with mutations in AUG (24, 34) can be attributed to defects in U5:AUG dependent exposure of NC binding sites rather than inhibition of a cis-packaging mechanism (34). The inability of helper RNAs to rescue packaging of RNAs with AUG mutations (24) is likely due to a defect in U5:AUG dependent exposure of the DIS since rescue requires DIS-mediated heterodimer formation. The relationship between U5:AUG formation and dimerization also explains why mutations in AUG can lead to the production of virions containing genomes that are sensitive to dissociation by mild denaturants (25). Hybrid 5´-leader structures containing both U5:AUG and AUG^HP features, predicted from in vivo ribose reactivity measurements (22), were not observed by NMR but can be explained by the presence of the AUG^HP/U5:AUG equilibrium. Elevated U5 nucleotide reactivities observed by traditional chemical probing, which are incompatible with an exclusive U5:AUG structure (21), may also be explained by the observed AUG^HP/U5:AUG equilibrium. The finding that the gag start codon is exposed in a mobile hairpin in the monomeric leader and sequestered in the dimer is consistent with observations that dimerization attenuates both the chemical reactivity of the gag start codon and the in vitro translational activity of the genome (16).

In summary, we have developed an approach that extends the size of RNAs that can be structurally probed by NMR to ~230 kDaltons. The lr-AID method avoids NMR relaxation problems associated with heteronuclear editing, probes interactions over a narrow distance range (≤ ~5 Å), provides both chemical shift and distance information for structural analysis, and is readily implemented using standard 2D NMR experiments and conservative base pair mutagenesis. Our findings support a translation/packaging RNA structural switch mechanism, in which the dimer-promoting GC-rich loop of the DIS hairpin is sequestered by base pairing with U5 in the AUG^HP form of the 5´-leader, and is displaced by AUG upon U5:AUG formation (Fig. 4D). Conformational changes associated with U5:AUG base pairing simultaneously sequester the gag start codon and expose the DIS and high affinity NC binding sites, thereby attenuating translation and promoting the packaging of a dimeric genome.