Secondary structure of the HIV reverse transcription initiation complex by NMR

Abstract

Initiation of reverse transcription of genomic RNA is a key early step in replication of the human immunodeficiency virus (HIV) upon infection of a host cell. Viral reverse transcriptase (RT) initiates from a specific RNA-RNA complex formed between a host transfer RNA (tRNA^Lys₃) and region at the 5'end of genomic RNA; the 3'end of the tRNA acts as a primer for reverse transcription of genomic RNA. We report here the secondary structure of the HIV genomic RNA-human tRNA^Lys₃ initiation complex using heteronuclear nuclear magnetic resonance methods. We show that both RNAs undergo large-scale conformational changes upon complex formation. Formation of the 18 base pair primer helix with the 3' end of tRNA^Lys₃ drives large conformational rearrangements of the tRNA at the 5' end, while maintaining the anticodon loop for potential loop-loop interactions. HIV RNA forms an intramolecular helix adjacent to the intermolecular primer helix. This helix, which must be broken by reverse transcription, likely acts as a kinetic block to reverse transcription.

Introduction

RNA structure guides key functions in HIV infection and replication. The viral genome is RNA, and RNA-RNA and RNA-protein interactions pervade the viral life cycle. The first step in viral replication after entry into an infected cell is reverse transcription, which is catalyzed by a viral enzyme, reverse transcriptase (RT)¹. RT must initiate from a specific RNA assembly and then navigate the complex secondary structure of genomic RNA while copying RNA to DNA². RT is a major target of therapeutic intervention in treatment of Acquired Immunodeficiency (AIDS).

HIV RT is a virally encoded RNA/DNA directed DNA polymerase, which consists of two subunits, p66 and p51. RT has both polymerase activity, as well as an RNase H domain, which cleaves RNA-DNA hybrids. The general mechanism of reverse transcription has been determined³. RT initiates reverse transcription from a specific RNA complex formed between a host tRNA^Lys₃^{4; 5} and a specific region (primer binding site or PBS) of the HIV genome near the 5'-end of HIV RNA^{6; 7; 8} (shown schematically in Fig. 1a). This initiation complex is preassembled in the virion, and initiation occurs by addition of deoxynucleotides to the free 3'OH of the tRNA^Ly3₃ (nts)^{9; 10}. RT copies the 5'-end of viral RNA into DNA, using its RNAase H activity to digest the RNA template. The DNA product is complementary to the 3' long terminal repeat (LTR) region of genomic RNA; interaction of the two strands allows reverse transcription to continue through the viral RNA 3' to 5'. Short stretches of genomic RNA remain that are resistant to RNase H cleavage, which then prime synthesis by RT of positive-strand DNA. Complementary regions of the single-stranded DNA hybridize and lead to eventual synthesis of fully double-stranded DNA that is competent for integration into the host genome.

Figure 1.

(A.) Schematic of the reverse transcription initiation complex within the HIV genome. The RNA genome of HIV RNA has the standard structure for retroviral genomes, with long terminal repeat regions (U5, U3, R), and the primer binding site (PBS) near the 5'end. Host tRNA^Lys₃ interacts with the PBS through an 18 base-pair interaction, providing the 3'-OH group to initiate reverse transcription by viral reverse transcriptase (RT).

(B.) RNA oligonucleotides corresponding to HIV-1 genomic RNAs used in the current study (left) a 99 nt and (right) 69 nt RNA with numbering according to the Mal isolate. Additional nucleotides added at the 5'end for transcription with T7 RNA polymerase are indicated as outlines. The primer binding site region of 18bp of complementarity with tRNA is highlighted in red, whereas the A-rich loop is in yellow. (c) Secondary structure of human tRNA^Lys₃ with modified nucleotides; Coloring as in (b).

Initiation of reverse transcription in HIV is directed from a specific RNA-RNA complex formed between host tRNA^Lys₃ and HIV genomic RNA¹¹. The PBS forms an 18 bp duplex with the 3' end of tRNA^Lys₃ but sequences outside this region of tRNA-HIV RNA complementarity are also critical for viral replication. Attempts to force HIV to use other tRNAs by making the PBS sequence complementary to the 3'-most 18 nts of the tRNA lead to rapid reversion of the virus back to using tRNA^Lys₃ by mutation of the PBS^{12; 13; 14; 15; 16}. Thus, sequences outside the simple PBS-tRNA base-pairing region are required for efficient initiation. Chemical probing of the viral RNA focused on a ca. 100 nt region whose reactivity to chemical probes changed upon binding of the viral RNA^{6; 7}. Mutations within this region can change both the efficiency of tRNA-viral RNA complex formation and the rate of initiation of reverse transcription in vitro^{8; 17} and in vivo^{18; 19}. Different viral isolates have sequence changes in this region^{20; 21; 22}.

Biochemical experiments have defined the general RNA-RNA interactions that direct formation of the initiation complex. The RNA-RNA complex that forms between host tRNA^Lys₃ and HIV PBS is >40 kDa in size, and global features of its secondary structure have been determined by chemical and enzymatic probing^{7; 8; 23}. Modified nucleotides in tRNA^Lys₃ facilitate proper initiation⁶, and a specific loop-loop interaction between an A-rich loop region of the viral RNA and the U-rich anticodon of tRNA^Lys₃ has been proposed to be critical for initiation complex formation and activity, with a modified 2-thio-uridine at position 34 in the anticodon stabilizing this interaction^{6; 24; 25}. A secondary structure for the tRNA-HIV RNA complex has been proposed, but no structural data beyond the experiments discussed above are available.

NMR spectroscopy allows determination of RNA secondary and tertiary structures^{26; 27; 28; 29}. Base-pair formation can be monitored through the exchangeable imino proton resonances; these protons are protected from solvent exchange upon base-pair formation, and appear in a distinct ¹H frequency in an NMR spectrum. Isotopic labeling of RNAs with ¹³C and ¹⁵N facilitate assignment and analysis of RNA NMR spectra^{26; 27; 30}. High-resolution structures of RNAs up to 100–120nts have been painstakingly achieved by using a combination of distance constraints and residual dipolar couplings obtained through heteronuclear NMR^{31; 32; 33; 34}. Large RNA complexes suffer from line broadening caused by slow molecular tumbling, although TROSY methods greatly improve linewidths for RNA base ¹H resonances²⁷. Secondary structure and overall folds for large RNAs (>25 kDa) can be achieved by analysis of experiments on imino ¹H resonances^{35; 36}.

Here we use heteronuclear NMR methods to probe the structure of the tRNA^Lys₃-HIV RNA complex. Biochemical analyses, coupled with prior experiments, defined the region of HIV RNA required for interaction with human tRNA^Lys₃. We show that that large scale conformational changes occur in both the HIV genomic RNA and host tRNA^Lys₃ upon formation of a binary RNA complex; the expected 18 base pair duplex that forms the central feature of the PBS is observed, as is the predicted downstream duplex within HIV genomic RNA. These results support prior models of the initiation complex and are a starting point for further mechanistic and structural studies of HIV reverse transcription.

Results

We first defined a system to probe structurally the initiation complex in vitro. Prior groundbreaking work by Marquet and co-workers defined the essential features of the HIV RNA-RNA initiation complex using biochemical and chemical probing (discussed above). Their results defined a 300 nt fragment of the HIV-1 RNA genome, spanning the PBS region that was required for complex formation^{6; 7; 8; 23; 37}. These experiments more specifically defined a ca. 100 nt region (120–220 in the Mal isolate) that formed a specific complex with modified native tRNA^Lys₃. This was our starting point for biophysical experiments. We initially explored the conformation of a 99nt HIV fragment of PBS (Fig. 1b), spanning this region, by NMR. Size-exclusion chromatography demonstrated a monomeric conformation for this RNA. However, the imino proton NMR spectrum (not shown) of this RNA was broad and showed evidence for multiple conformers, and no evidence for the formation of the A-rich loop, which has distinct chemical shifts determined by our prior structure determination³⁸. We concluded that the 99mer RNA has a misfolded conformation in solution and in addition undergoes hydrolysis at a UpA step between positions 199 and 200. We thus abandoned this construct for further biophysical study.

The segment of HIV viral RNA between nts 130 to 200 directs tRNA^Lys₃ hybridization. A 69mer RNA (Fig. 1b) that corresponds to this region did not hydrolyze under standard conditions, and NMR spectroscopy suggested a single conformation for this RNA. We prepared unlabeled and ¹³C, ¹⁵N-labeled versions of the 69mer, and confirmed a general secondary structure as proposed from chemical probing experiments. The NMR data showed that the A-rich loop, whose structure was previously determined by NMR on the isolated domain oligonucleotide³⁸, was formed in this construct.

Likewise, we have examined previously the secondary and tertiary structure of human tRNA^Lys₃ (Fig. 1b) using NMR spectroscopy³⁵. Using ¹⁵N, ¹³C labeled tRNA transcripts (described in Materials & Methods), we showed that the correct tRNA secondary and tertiary, L-shaped fold is formed in the presence of 5–10mM MgCl₂. We assigned the imino ¹H resonances for the four helical stems of the tRNA, confirming their formation. These data also confirmed the formation of the key base triple interactions: A9--U12-A23 and C13-G22-G46 as well as the U8-A14 tertiary pair. Thus, modified nucleotides are not required for correct folding of this tRNA.

To observe whether the HIV RNA constructs could form a specific initiation complex with tRNA^Lys₃, we performed chemical probing footprinting experiments. Natural, fully modified tRNA^Lys₃ was purified from bovine liver as described previously³⁹, and was used as a control in comparison to transcribed tRNA^Lys₃. RNAs were hybridized at 1:1 stoichiometry at 90°C for 3 minutes in low salt (10mM Na phosphate or cacodylate, pH 7), followed by cooling to 60°C, addition of 100mM NaCl, and slow cooling to room temperature. MgCl₂ was only added at room temperature, to avoid hydrolysis of the RNAs. Hybridization of tRNA and HIV RNA was readily confirmed by native gel electrophoresis.

RNA structure in the free and complexed HIV 69mer was probed using chemical reactivity with dimethyl sulfate, which reacts with the N1 position of A and N3 position of C⁴⁰. In the absence of tRNA, the 69nt RNA shows chemical reactivity within the A-rich loop region between A164–167 (Fig. 2). Upon addition of 1 molar equivalent of either natural tRNA^Lys₃ or transcribed tRNA^Lys₃, similar protections are observed in the A-rich loop region, indicating formation of RNA structure in this region. These results confirmed that tRNA transcripts or native tRNAs form similar complexes with the 69nt HIV genomic RNA construct. We thus used the transcribed tRNA, which is readily labeled with ¹⁵N and ¹³C, for subsequent NMR experiments on the RNA-RNA complex.

Figure 2. Chemical probing experiments on HIV initiation complexes with 69nt model HIV1 genomic RNA.

Accessibility of adenosine N1 and cytosine N3 positions to reaction with dimethyl sulfate (DMS) was detected in the presence or absence of 1:1 stoichiometry of either pure bovine native tRNA^Lys₃ (nat) or T7 RNA polymerase transcript (T7). Reactivity with DMS was detected using primer extension with reverse transcription using a DNA primer in the absence (−) and presence (+) of dimethyl sulfate reaction. Reactions were performed at room temperature as described.

Despite the large molecule weight of the 1:1 tRNA^Lys₃-HIV₆₉ RNA complex (MW 47,000), the imino ¹H NMR spectrum was well resolved and interpretable. Large changes in the imino spectrum of tRNA (and HIV 69 mer) is observed upon 1:1 complex formation (Fig. 3), indicating large scale structural rearrangements in the tRNA. Strikingly, resonances that originate from the tRNA acceptor stem (e.g. G6), D stem, and tRNA tertiary structure (amino resonances, U8, U55)³⁵ are lost or shifted in the RNA-RNA complex. A large number of new resonances that originate from imino protons within the tRNA are observed. These qualitative observations are suggestive of large changes in the tRNA structure upon complex formation.

Figure 3. Imino ¹H NMR spectra of folded human tRNA ^Lys₃ transcript (bottom) compared to that of 1:1 complex of human tRNA^Lys₃ with 69nt HIV1 genomic RNA (top).

Spectra were acquired at 800 MHz in 10mM MgCl₂, 100mM NaCl, 10mM Na phosphate, pH 6.5 at 25°C. Assignments of resonances in folded tRNA^Lys₃ were reported previously.

To confirm this observation and focus on changes in only tRNA^Lys₃, we prepared 100% ¹³C, ¹⁵N-labeled tRNA^Lys₃ and formed a 1:1 complex with unlabeled HIV genomic RNA (69mer). We have previously assigned the exchangeable ¹H spectrum for folded tRNA^Lys₃³⁵. Within the 1:1 RNA-RNA complex, drastic changes in the spectrum for tRNA^Lys₃ are observed (Fig. 4a). The D-stem, with its set of tertiary interactions involving G10-C25, C11-G24, U12-A23, and U8, undergoes large chemical shift changes in the complex. Likewise, shifts are observed in the T- and acceptor stem resonances G6-U67, C2-G71, U55 in the T loop (Fig. 4b). These changes are consistent with the proposed hybridization of 18 nts at the 3'end of tRNA^Lys₃ (located in the T- and acceptor stems) with the complementary PBS sequence in HIV RNA. Only minor shifts were observed in the anticodon stem loop resonances. The results suggest changes in the D, T and acceptor stems of the tRNA upon binding to HIV genomic RNA.

Figure 4. Changes in tRNA^Lys₃ conformation upon initiation complex formation.

(A.) ¹⁵N-¹H TROSY spectrum of uniformly ¹³C, ¹⁵N labeled human tRNA^Lys₃ either alone (red) or in 1:1 complex with unlabeled HIV-1 69nt RNA (black). The comparison of the two spectra highlights the large changes in tRNA structure that occurs upon complex formation. (B.) Secondary structure of human tRNA^Lys₃; major perturbations (pink circles) of the folded tRNA spectrum upon complex formation with HIV RNA occur in the acceptor, T and D-stems, whereas resonances for the anticodon stem are relatively unperturbed. The spectra were acquired at 800 MHz in 10mM MgCl₂ 50mM NaCl, 10mM Na phosphate, pH 6.5 at 25°C using a T1-relaxation optimized TROSY pulse sequence as described in the text for improved signal-to-noise.

To assign in detail the secondary structure of the tRNA-HIV complex, the following general strategy was taken. First, imino proton resonances were assigned to either tRNA or HIV RNA by preparing two individual samples with one component uniformly labeled with ¹³C, ¹⁵N nucleotides, and the partner unlabeled (Fig. 5a, b). An unlabeled complex sample was prepared for homonuclear NOESY experiments to map imino-imino NOEs and allow assignment of helical elements. Finally, intramolecular (tRNA-tRNA or HIV-HIV) helical elements were assigned using HNN COSY approaches. We explore each step in detail below.

Figure 5. ¹H-¹⁵N TROSY experiments for the HIV initiation complex at 1:1 tRNA^Lys₃-HIV RNA stoichiometry.

(A.) Spectrum for initiation complex formed with uniformly labeled ¹³C, ¹⁵N-tRNA^Lys₃ and unlabeled HIV-1 69nt RNA. (B.) Spectrum for initiation complex formed with uniformly labeled ¹³C, ¹⁵N-HIV-1 69nt RNA and unlabeled tRNA^Lys₃. Spectra were acquired at 800 MHz ¹H frequency T1-relaxation optimized TROSY at 25°C. Spectral assignments of the imino resonances as discussed in the text are indicated.

Based on the changes in the tRNA data, we assumed that the 18bp PBS helix was formed. We first assigned the imino ¹H resonances for base pairs in the 18 bp PBS helix. We performed homonuclear NOESYs at several mixing times (50, 100, 200ms) to observe imino-imino ¹H NOES characteristic of A-form RNA helices. The NOESY experiments confirm that 18 bp helix is indeed formed in our tRNA-HIV complex (Fig. 6). Connectivity of imino resonances in the NOESY was observed for 16 continuous base pairs from C_H196-G_T59 until G_H181-C_T74. These assignments were confirmed by synthesis of an oligonucleotide containing the 18 base pair helix in isolation, capped at one end by a UUCG tetraloop (Supplementary Fig. 1). Assignment of the base pairing imino ¹H spectrum of this RNA agreed with those of the same helix in the HIV-tRNA complex. This further suggests that the PBS helix forms without extensive tertiary contacts to the rest of the RNA in the binary complex.

Figure 6. NOESY spectrum of unlabeled 1:1 complex of human tRNA^Lys₃ and HIV1 69nt RNA.

The region of imino ¹H-¹H NOES is shown; data were acquired at 800 MHz with 100ms mixing time at 25°C. NOE connectivities are observed that allow assignment of 16 of the 18 base pairs formed between the tRNA and HIV RNA.

To assign remaining features of the complex structure, we used the heteronuclear NMR experiments^{27; 41} on complexes with one component of the complex isotopically labeled and the other unlabeled (see Materials and Methods). First, assignment of the 18 bp helix was confirmed by the sorting of imino protons from either tRNA or HIV RNA, as determined by the ¹⁵N-TROSY experiments on complexes with either uniformly ¹⁵N-labeled tRNA or HIV RNA. Intramolecular base pairing within an individual RNA component was identified by direct measurement of ¹⁵N-¹⁵N scalar couplings across base pairs using HNN-COSY experiments⁴².

We next confirmed formation of an intramolecular RNA-RNA helix within HIV RNA in the initiation complex. The helical connectivity extends from base pairs C132-G175 until U140-A168. The existence of this helix is supported by imino ¹H NOESY crosspeaks, and the observed correlation peaks in the HNN NOESY experiment using an initiation complex formed with ¹³C,¹⁵N-labeled HIV RNA and unlabeled tRNA^Lys₃ (Fig. 7a). Several imino resonances arise from the HIV RNA that we have not assigned, indicating possible additional pairing within the large hairpin loop formed by the HIV RNA.

Figure 7. H-NN COSY experiments showing through-space ¹⁵N-¹⁵N scalarcouplings across base-pairing hydrogen bonds.

(A.) H-NN COSY spectrum of the initiation complex formed with uniformly labeled ¹³C, ¹⁵N-tRNA^Lys₃ and unlabeled HIV-1 69nt RNA. (B.) H-NN COSY spectrum for initiation complex formed with uniformly labeled ¹³C, ¹⁵N-HIV-1 69nt RNA and unlabeled tRNA^Lys₃. Both experiments only show correlations for intramolecular base pairing for the ¹⁵N-labeled component of the complex. Data were acquired at 25°C at 800 MHz using a T1-relaxation optimized (BEST-HNN COSY) version that uses band selective pulses centered on the imino proton resonances as described in the materials and methods. (C). Secondary Structure of the HIV initiation complex determined here by NMR spectroscopy. HIV 1 genomic RNA (Mal-1 isolate) is in black, tRNA^Lys₃ in red. The key regions of the structure, including 18 bp PBS helix, HIV genomic RNA intramolecular helix, and the conservation of the tRNA anticodon stem-loop are highlighted. Potential U-rich anticodon loop interaction with A-rich sequence in the HIV RNA is shaded. Arrows indicate pausing sites observed in kinetic investigations of the HIV-1 initiation complex with reverse transcriptase. These strong pauses are at positions +3, +6 and +16.

We determined rearrangements in the tRNA^Lys₃ secondary structure within the initiation complex. tRNA^Lys₃ was uniformly ¹³C, ¹³N labeled and HIV RNA was unlabeled. HNN COSY experiments confirmed the presence of extensive intratRNA base pairing within the initiation complex (Fig. 7b). As shown above, the 3' portion of the acceptor stems and T-stems have rearranged to form the 18 bp helix. This explains the large changes in chemical shift observed for imino resonances for nucleotides G59-A76. In addition, significant changes are also observed for resonances from nucleotides in the D stem and 5' side of the acceptor stem; this suggests rearrangements driven by loss of secondary and tertiary structure in the tRNA upon binding. Interestingly, resonances from the anticodon stem did not undergo significant changes upon complex formation, suggesting that this structural feature, which is independent of the surrounding secondary and tertiary structure, is maintained in the initiation complex. This observation is confirmed by the more detailed analysis presented below.

The secondary structure of the tRNA^Lys₃ in the initiation complex was confirmed by NOESY experiments. The helices of the anticodon stem (U27-A43 to G30-C40) and D-stem (G10-C25 to C13-G22) are formed in the complex; the chemical shifts of the anticodon stem residues are similar to those in the tRNA, whereas significant shifts are observed in D-stem resonances, reflecting the disruption of tertiary interactions in free tRNA^Lys₃ upon initiation complex formation. The 5'-end of the acceptor stem now pairs with a region formerly in the T-loop. A helix interrupted by a single-nucleotide bulge is formed from nucleotides G1 to A9 pairing with nucleotides G45 to U54. This pairing is confirmed by both HNN-COSY crosspeaks and imino-imino proton NOEs.

Discussion

The NMR results presented here allowed definition of the secondary structure for the initiation complex (Fig. 7c). We first defined a system for NMR using biochemical approaches, building on the pioneering studies of Marquet and co-workers^{7; 39}. We focused on the PBS region of HIV-1-RNA, initially defined as a 300 nt region of the HIV genome. The initiation complex between HIV RNA and tRNA^Lys₃ was formed biochemically by hybridization of the two RNAs using heat annealing. Longer constructs (>90nt) of HIV RNA gave poor biophysical behavior on their own, likely due to multiple RNA conformations and potential aggregation; efficient complex formation with these longer RNAs requires fully modified natural tRNA^Lys₃⁶. However, a 69 nt fragment of the HIV 1 Mal isolate primer binding site is well-behaved biophysically and allows efficient hybridization of either naturally modified or unmodified transcript human tRNA^Lys₃. Correct complex formation, including interaction of the HIV A-rich loop, was observed with 1:1 stoichiometry of 69nt HIV RNA and unmodified tRNA^Lys₃ using chemical probing. These experiments defined a tractable, biochemically relevant system for subsequent NMR studies of the initiation complex.

The 50 kDa tRNA^Lys₃-HIV RNA complex was a challenging target for NMR study. Here, we used both homonuclear and heteronuclear NMR to define the secondary structure of the initiation complex (Fig. 7c). Imino proton resonances were assigned and intermolecular hydrogen bonding and NOEs used to define base pairing interactions. NMR spectra were simplified and imino ¹H resonances were assigned to either the HIV RNA or tRNA component of the complex through selective ¹⁵N, ¹³C- labeling of either the HIV RNA or tRNA. Imino resonances were then assigned using a combination of sorting of imino proton resonances to the two component, HNN COSY data to define base pairing and imino-imino ¹H NOEs to define patterns of secondary structure. Where possible, assignments were confirmed by synthesis and NMR analysis of shorter oligonucleotides corresponding to domains of the initiation complex. A total of 38 of 42 observed imino proton resonances were assigned using this approach.

The secondary structure of the initiation complex determined by NMR shows multiple features that may modulate its function (Fig. 7c). As expected, nucleotides 179–196 of HIV genomic RNA pairs with nts 69–76 at the 3'end of tRNA. This stable intermolecular 18 base pair primer helix must bind within the active site cleft of reverse transcriptase, positioning HIV nucleotide G178 as the template for the first round of DNA synthesis. Downstream from the 18bp primer helix, there is a 3bp stretch of single stranded nucleotides. Then, HIV RNA forms an intramolecular helix as proposed and demonstrated by Marquet et al.⁷ with base pairs from nts C132-G175 to G137-C170. We could not assign any additional significant secondary structure for the HIV RNA after this region.

tRNA^Lys₃ undergoes significant remodeling of its secondary structure upon initiation complex formation. The 3'end of tRNA forms the 18 bp primer helix, which deprives the 5'end of the acceptor stems and T-stems of their pairing partners. These regions refold to pair with each other in the initiation complex; nts G1 to A9 form an imperfect duplex with nucleotide 45–54. The anticodon and D-stems remained base paired as in the tRNA secondary structure, although chemical shift and NOE data support that tertiary interactions that occur in folded tRNA within the D-stem do not occur in the initiation complex.

The secondary structure that we observe by NMR confirms prior predictions made by a variety of biochemical and modeling experiments²³. The 18 base pair helix between the tRNA and HIV is a central feature of the secondary structure, as demonstrated by many biochemical and viral replication experiments. Formation of this thermodynamically stable intermolecular helix drives structural rearrangements in both the HIV genomic RNA and tRNA^Lys₃. These conformational rearrangements in vivo are likely facilitated by the nucleocapsid protein and other chaperones^{43; 44}.

The tRNA rearrangement that occurs upon formation of the initiation complex further explains the choice of tRNA^Lys₃ as a primer. The stable pairing between the T and acceptor stems leads to a stable refolding of the tRNA sequence upon initiation complex formation. The refolded tRNA structure maintains the U-rich anticodon stem loop to make potential contacts with the A-rich loop within the HIV RNA. We do not observe imino proton resonances consistent with anticodon-A-loop interaction, but chemical probing data, coupled with recent NMR experiments by Agris and co-workers²⁵, suggests that such a pairing occurs; the absence of modified nucleotides in our tRNA construct likely increases the solvent exchange rate for these potential base pairs. The A-loop region in the complex structure is presented as single-stranded nucleotides available for additional pairing. The stability of this loop-loop interaction is likely such that the imino resonances from the anticodon loop are broadened by solvent exchange.

Upon initiation complex formation, the HIV genomic RNA rearranges to form a large hairpin loop structure with an intramolecular helix starting just downstream from the 18bp PBS helix. This secondary structure element, first proposed by Marquet and coworkers⁷, represents a block to reverse transcriptase^{20; 45}, which must melt this helix during the early rounds of transcription. This helix is site of two reverse transcription pauses during minus strand strong stop DNA synthesis at positions +3 and +6²⁰. (see Fig. 7c); rates of reverse transcription are 3400-fold slower for the first 15 nts compared to subsequent nucleotides, with limited pausing also after the potential A-rich sequence-anticodon loop interactions. The RNA interactions observed in our structure present a plausible explanation for the slow elongation rates during the early phase of reverse transcription, and may explain the limited extension observed in the preformed initiation complex within the HIV virion.

The initiation complex must be recognized by HIV reverse transcriptase. The 18bp helix would fit within the cleft of HIV RT, where other nucleic acid helical structures (DNA-DNA and DNA-RNA duplexes) have been observed to bind^{1; 46}. This would place the downstream HIV helix adjacent to the active site. How HIV RT recognizes the initiation complex and perhaps reorganizes the tRNA-HIV RNA complex structure observed here awaits more detailed structural investigation of the initiation complex.

Recent single-molecule fluorescence experiments on HIV RT bound to the initiation complex have underscored the role of RNA structure in modulating protein-RNA dynamics⁴⁵. Using fluorescence resonance energy transfer (FRET), Legrice, Zhuang and co-workers found dynamic switching between correct, active initiation complex orientation within the binding cleft of RT, and an inactive orientation in which the +1 position is flipped 180° and located near the RNAase H domain. The presence of the intramolecular HIV helix that we observe here by NMR stabilizes RT in the incorrect orientation, explaining the poor kinetics of elongation through this helix; once RT traverses the +6 position, the active orientation of RT is greatly favored.

The results presented here represent a first level of understanding of the initiation complex. Our data highlight the power of modern NMR experiments to map global RNA folds, and demonstrate how RNA structure may control the activity of reverse transcription. More detailed structural data on the RNA initiation complex using NMR and crystallography are required to provide a tertiary structure of both the RNA-RNA complex and its interaction with RT.

Materials and Methods

RNA synthesis

RNA samples were prepared by in vitro transcription using T7 polymerase as described^{47; 48}. HIV RNA was purified using Superdex 200 (26/60) gel filtration as described^{48; 49}. Natural tRNA^Lys₃ was purified from 10 kg of bovine livers using modifications of published protocols^{6; 39}. Unmodified human tRNA^Lys₃ was prepared by in vitro transcription and purified using preparative polyacrylamide gel electrophoresis as described previously⁵⁰. Uniformly ¹³C, ⁵N-labeled RNAs were transcribed using ¹³C, ¹⁵N-labeled ribonucleoside triphosphates prepared in house^{51; 52}. All RNA samples were concentrated by Vivaspin (cutoff 5,000Da) and kept in 10 mM phosphate buffer, pH 6.5. Molar extinction coefficients were calculated from the A260 values at 20□C by extrapolation from the values at 90□C. All RNAs were characterized for the oligomerization state using analytical gel filtration chromatography and native gel electrophoresis to confirm monomeric species, or bimolecular RNA-RNA complexes.

Chemical Probing

Chemical probing was performed using dimethyl sulfate reaction⁵³ with 20 nM RNA complexes at 25°C in 100mM NaCl, 10mM MgCl₂ 10mM Na phosphate pH 6.5; the 1:1 binary complexes of HIV 69 nt RNA with either native bovine tRNA^Lys₃ or transcribed human tRNA^Lys₃ were incubated with a 1:600 dilution of DMS for 7 minutes at 25°C and quenched by the addition of 4 volumes of 94% EtOH, 63 mM βME, 94 mM NaOAc (pH 5.5), 54 mM Tris-HOAc (pH 7.5), 30 μg/mL carrier tRNA and the reactions were precipitated by spinning at 16,000 × g for 15 minutes, washed with 70% EtOH and re-suspended in water. A 18nt DNA primer was hybridized to the 5'-end of HIV genomic RNA complementary to the PBS. The DNA primers are diluted to 1 μM in 225 mM HEPES-KOH (pH 7.0) and 450 mM KCl and added to an equal volume of resuspended RNA, mixed and then heated to 90°C for 1 min and then slow cooled to 42°C. The primed RNA was then incubated with AMV-RT in 130 mM Tris-HCl (pH 8.5), 10 mM MgCl₂, 10 mM DTT, 6 μCi α-³²P-dTTP, 110 μM dATP, dCTP, dGTP, and 6 μM dTTP at 50°C for 30 minutes. The reaction was stopped by the addition of 1/10th volume of 3 M NaOAc (pH 5.2) and then phenol:chloroform extracted before being EtOH precipitated. The cDNA pellet was then suspended in a small volume (10 μL) of denaturing urea loading buffer, heated to 90°C for 5 minutes and immediately loaded onto a pre-run 15% denaturing polyacrylamide sequencing gel. Subsequent to electrophoresis, the gel was transferred to Whatman filter paper, dried and exposed to film or phosphorimager to detect the radioactive bands.

NMR Spectroscopy

All NMR spectra were collected at 25°C on a Varian INOVA 800 NMR spectrometer equipped with triple resonance x,y,z-axis gradient probes within the Stanford Magnetic Resonance Laboratory (SMRL). ¹H and ¹⁵N assignments were obtained using standard homonuclear and heteronuclear methods optimized for RNA structure determination (RNAPack, http://www.varianinc.com)^{27; 41} following detailed published procedures, except where outlined below. NMR data were processed with VNMR (http://www.varianinc.com) and spectra were analyzed with SPARKY⁵⁴.

Several RNA samples were prepared: unlabeled HIV1 99mer genomic RNA, unlabeled HIV1 69mer genomic RNA, unlabeled human tRNA^Lys₃, uniformly ¹³C, ¹⁵N-labeled 69mer HIV1 genomic RNA and uniformly ¹³C, ¹⁵N-labeled tRNA^Lys₃. 1:1 stoichiometric complexes of tRNA^Lys₃-69mer HIV genomic RNA samples were formed by heat annealing the two RNAs at 95°C for 3 min in 10mM Na Phosphate, pH 6.5, incubation at 60°C for 5 min with adjustment of the monovalent ion concentration to 100mM NaCl, and slow cooling to room temperature; where indicated, MgCl₂ was added at room temperature to avoid hydrolysis. Samples were 0.5–0.8 mM for unlabeled RNAs and 0.25–0.4mM for labeled RNAs in 250μL in Shigemi NMR tubes.

To assign exchangeable resonances of individual RNAs and RNA-RNA complexes, 2D NOESY (30, 50, 75, 100 and 200 ms mixing times; 2.0 s delay time) were obtained using shifted laminar pulses for solvent suppression (SSNOESY)⁵⁵. 1H-15N TROSY, 3D 1H −15N NOESY-HSQC, and 2D HNN-COSY were collected in 90% H2O/10% 2H2O at 298K or 303K. For the large (50 kDa) RNA-RNA complexes, T1-optimized spectroscopy was used to obtain improved signal-to-noise for ¹H-¹⁵N TROSY and HNN COSY experiments⁵⁶; these experiments use selective pulses centered on the imino proton region of the spectrum to enhance population differences with neighboring amino proton resonances, thus facilitating T1 relaxation through dipolar mechanisms. For these experiments, recycle delay was set to 0.5 s for optimal signal, and good-quality HNN-COSY spectra were obtained in 2–8 hrs.

Supplementary Material

01. Supplementary Figure 1. NOESY spectrum of a model oligonucleotide that corresponds to the 18bp primer helix.

The secondary structure and sequence of the oligonucleotide is shown; the first base pair was changed to allow transcription. The NOESY spectrum shows imino-imino ¹H NOEs obtained using shifted laminar pulse solvent suppression and a mixing time of 200ms at 800 MHz and 25°C. Assignment of imino-imino NOES is shown, and the region assigned is highlighted in red in the oligonucleotide. The numbering uses that of the equivalent positions in the tRNA and HIV RNA (see Fig. 6 & 7c).