Probing nucleation, reverse annealing, and chaperone function along the reaction path of HIV-1 single-strand transfer

Abstract

Reverse transcription of the HIV-1 genome involves several nucleic acid rearrangement steps that are catalyzed (chaperoned) by the nucleocapsid protein (NC), including the annealing of the transactivation response region (TAR) RNA of the genome to the complementary sequence (TAR DNA) in minus-strand strong-stop DNA. It has been extremely challenging to obtain unambiguous mechanistic details on the annealing process at the molecular level because of the kinetic involvement of a complex and heterogeneous set of nucleic acid/protein complexes of variable structure and variable composition. Here, we investigate the in vitro annealing mechanism using a multistep single-molecule spectroscopy kinetic method. In this approach, an immobilized hairpin is exposed to a multistep programmed concentration sequence of NC, model complementary targeted-oligonucleotides, and buffer-only solutions. The sequence controllably “drags” single immobilized TAR hairpins among the kinetic stable states of the reaction mechanism; i.e., reactants, intermediates, and products. This single-molecule spectroscopy method directly probes kinetic reversibility and the chaperone (catalytic) role of NC at various stages along the reaction sequence, giving access to previously inaccessible kinetic processes and rate constants. By employing target oligonucleotides for specific TAR regions, we kinetically trap and investigate structural models for putative nucleation complexes for the annealing process. The new results lead to a more complete and detailed understanding of the ability of NC to promote nucleic acid/nucleic acid rearrangement processes. This includes information on the ability of NC to chaperone “reverse annealing” in single-strand transfer and the first observation of partially annealed, conformational substates in the annealing mechanism.

Reverse transcription of the HIV-1 RNA genome involves several critical nucleic acid rearrangement steps that are chaperoned (catalyzed) by the HIV-1 nucleocapsid protein, NC (1, 2). Mechanistic investigations of these rearrangements have been challenging because of the extreme structural and kinetic heterogeneity of these processes, which are believed to involve a heterogeneous distribution of nucleic acid/protein complexes of variable composition and unknown secondary structure (3). These challenges have been particularly well documented for the NC annealing of the transactivation response region (TAR) RNA of the HIV-1 genome to the complementary sequence (TAR DNA) in minus-strand strong-stop DNA (4). Liu et al. (5) recently investigated the kinetics of an in vitro model (equation i of Fig. 1) for this multistep reaction; i.e., the annealing of two hairpins to produce the thermodynamically favored duplex. In the experiments of Liu et al., isolated TAR hairpins were immobilized on a biologically compatibilized coverslip that was located within a flow system that supplied “fresh” unaggregated solution of complementary DNA or RNA hairpin. Real-time fluorescence single-molecule spectroscopy (SMS) of oligonucleotides labeled with donor (D) and acceptor (A) dyes was used to simultaneously monitor the annealing kinetics, the secondary structure of the reactants, and the state-of-aggregation of the hairpins, in situ. Internal distances between labeled sites in reacting hairpin pairs were probed by time-resolved fluorescence resonance energy transfer (i.e., single-molecule FRET) between the donor and acceptor dyes.

Fig. 1.

Hypothetical kinetic scheme for NC-chaperoned annealing of Cy3-labeled TAR DNA to its Cy5-labeled complements, with Cy3 as fluorescence donor (D) and Cy5 as acceptor (A). T, TAR DNA; C, complementary cTAR DNA or TAR RNA; N, nucleocapsid protein, NC. In this scheme, N binds to T and C, leading to a partially melted structure, namely the Y form of T (T′) and C(C′). The subscripts i, j, k, and l are used to describe the number of NC bound to nucleotides. Two partially melted hairpins form an encounter complex that leads to the formation of nucleation complexes. The annealing can go through either zipper nucleation or loop nucleation, therefore forming zipper nucleation complexes (Z) or loop nucleation complexes (L), both leading to the formation of fully annealed duplexes.

Based on the data of Liu et al. (5) and previous observations, these authors proposed a hypothetical mechanism for the annealing reaction that is summarized by equations ii and iii of Fig. 1. The reacting hairpins in this mechanism undergo a rapid NC-induced equilibrium between closed and partially melted “Y”-shaped conformations, which is supported by previous experiments (6, 7). The chaperone (catalytic) activity of NC is envisioned as arising from essentially two independent effects: an NC-induced partial melting of the Watson–Crick pairing of the reactant hairpins (8–11) and an NC-induced decrease in the energy cost of bringing the hairpins together to form the encounter complex, due presumably to a screening of the negative charges on the hairpins and perhaps through specific interactions (12–16). The nucleation of annealing occurs in an encounter complex that is comprised of partially melted TAR DNA and TAR RNA hairpins and several bound molecules of NC (5). Two nucleation-complex isomers are envisioned for the annealing process, depending on whether the annealing occurs in the region of the L1L2 internal loops or, alternatively, in the region of the L4 internal bulge and hairpin loop (HL) of TAR DNA. The L1∼L4 and HL regions are indicated in equation i of Fig. 1.

Here, the annealing reaction between TAR DNA and various DNA or RNA oligonucleotides (Fig. 2) is probed by a new SMS approach in which single-molecule kinetic data are acquired while the immobilized TAR DNA hairpins are exposed to a time-programmed concentration sequence of different targeted oligonucleotides with and without the NC chaperone present in the solution. This procedure chemically “drags” individual pairs of reacting hairpins through the reactant states and the intermediate states, and back again (5). This approach offers information on the different stages of the annealing mechanism, especially the putative nucleation complexes. The experiments described in this article have been built on previous simpler SMS TAR DNA annealing studies with DNA oligonucleotides (5).

Fig. 2.

Structures of various oligonucleotides used in this study. The secondary structures were predicted by the program mfold (www.bioinfo.rpi.edu/applications/mfold/dna/form1.cgi) (21).

Results and Discussion

The multicomponent, oligonucleotide SMS experiments that are the focus of this article can be put in context by first considering the previously investigated annealing of “full-length” hairpin complementary TAR (cTAR) DNA to an immobilized TAR-DNA hairpin (equation i of Fig. 1) (5, 17). Single-molecule kinetic results for this reaction are shown in the first column of Fig. 3. Here, the irreversible NC-catalyzed annealing reaction was initiated by exposing a dilute, immobilized sample of the Cy3-TAR DNA to a freshly mixed solution of Cy5-cTAR DNA and NC at zero time, t = 0, in analogy to a stopped-flow experiment. The freshly mixed solution was prepared in situ in a mixing chamber that combined various concentrations of acceptor-labeled short oligonucleotides solutions in Hepes buffer and a 889 nM solution of NC. Before t = 0, only a buffer solution was flowed into the flow cell. The acceptor and donor intensities, I_A(t) and I_D(t), were recorded by sample scanning confocal microscopy at various stages of the reaction correspondingly, and the donor and acceptor images were analyzed. For each immobilized hairpin, an apparent FRET efficiency, E_A, was determined as follows:

By measuring E_A for each hairpin, at various times, t, after introducing the complementary Cy5-hairpin solution FRET trajectories (Fig. 3A) were recorded to monitor the instantaneous distance between the 5′ end of the immobilized Cy3-TAR DNA hairpin and the 3′ end of the Cy5 hairpins.

Fig. 3.

Room temperature single-molecule kinetic measurement on the annealing of Cy3-TAR DNA to 2 nM Cy5-cTAR, 25 nM Cy5-zipper DNA, 25 nM loop DNA, and 50 nM Cy5-zipper RNA at 0.2 mM Mg²⁺ and 889 nM NC. (A) E_A of each single molecule during the reaction, where each colored line corresponds to a single molecule. (B) Ensemble mean E_A increases during the annealing reaction. (C) Number of product (N_P, shown as % among total number of molecules) molecules during the annealing reaction (blue), with the number of reactant (green) (N_R) molecules decreasing by the same amount. Both of the increase/decrease trends over time can be fitted with a single exponential (dotted lines). The data were collected from 100 single molecules for cTAR annealing kinetic measurement, 140 for zipper annealing and 200 for loop annealing.

FRET trajectories for individual reacting pairs (Fig. 3A) show the previously reported behavior of exhibiting discrete switching at the time of annealing from a low FRET value due to an isolated Cy3-TAR DNA hairpin to a high value due to an annealed Cy3-TAR DNA/Cy5-cTAR pair. FRET is efficient for the annealed pair because of the close proximity of the 5′ end of TAR DNA and the 3′ end of cTAR. Previously reported histograms of the FRET from many hairpins at different times during annealing show the expected dual peaks corresponding to the two-state annealing reaction (17). Very few of the hairpins reveal reverse annealing (high-to-low FRET changes), indicating that the annealing reaction highly favors the full duplex product state. This also demonstrates that photobleaching (18–20), which is also manifested by a similar high-to-low FRET change, is not a significant complication in these experiments. Some hairpins do not anneal during the time period because of nonidealities such as unlabeled cTAR and clustering or imperfect immobilization. Because of these imperfections, the mean FRET value as a function of time (Fig. 3B) does not perfectly reflect the true FRET changes because it reaches an asymptotic value of 0.86, whereas the E_A of the annealed form is ≈1. Fig. 3C portrays the number of surviving reactant hairpins N_R (green curve) and the number of annealed product N_P (blue curve) pairs, respectively. These were determined by counting the number of hairpins that were below/above an E_A threshold of 0.4, as described in ref. 17.

The second and third columns in Fig. 3 correspond to annealing experiments with shorter DNA oligonucleotides that are targeted for “zipper” (L1L2 stem loops) and “loop” [L4, hairpin loop (HL)] regions of TAR DNA, as shown in Fig. 2. SMS data on annealing of both targeted oligonucleotides show evidence of reversible annealing that leads at later times to an equilibrium distribution of annealed and unannealed TAR. For example, the individual E_A trajectories show much more high-to-low FRET transitions than the TAR DNA/cTAR case. The 〈E_A〉 curves and the number of reactant and product data are also consistent with an equilibrium mixture at long times, with an apparent dissociation constant, K_d, of ≈10 nM and ≈16 nM for the zipper and loop DNA, respectively. One can estimate K_d from the ≈70% and ≈60% annealing percentage at 25 nM concentration for zipper and loop DNA, respectively. The annealed adducts of TAR DNA with the target oligonucleotides are arguably models for intermediates in the annealing reaction of full-length cTAR or TAR RNA and, indeed, in minus-strand transfer itself. We envision that the annealed form of TAR DNA with zipper is a model for the putative “Z” nucleation complex in Fig. 1, and that, in turn, the annealed adduct of TAR DNA with the loop oligonucleotide is a model for the “L” nucleation complex in Fig. 1. To simplify the nomenclature in this article, the equilibrium form of the annealed adduct of two nucleic acids will be denoted as follows, nucleic-acid-1/nucleic-acid-2, e.g., TAR DNA/zipper DNA. The annealing of TAR DNA with TAR RNA, zipper RNA, and loop RNA are more difficult to study accurately because of their higher tendency to form aggregates and stick to surfaces but generally exhibit similar kinetic behavior to the DNA oligonucleotides. As an example, the fourth column of Fig. 3 shows the kinetics of TAR DNA/zipper RNA annealing.

To characterize the nucleic acid rearrangement pathways available to the TAR DNA/targeted-oligonucleotide adducts, we subjected these models for nucleation complexes to a sequence of solutions containing buffer only, buffer plus NC, and NC plus cTAR. Typical results are shown in Fig. 4 for the loop DNA oligonucleotide. In the first time epoch of the experiment, immobilized TAR was exposed to a loop DNA plus NC solution, leading to an annealing equilibrium. This was followed by a period in which the loop DNA plus NC solution was rapidly replaced with a buffer-only solution. The major effect of the buffer-only period was to freeze the concentration of the TAR/loop DNA adduct, even though the equilibrium constant strongly favors dislocated adducts in the absence of oligonucleotides in solution. In contrast, when NC was added to the solution (in the third epoch), the concentration of the adduct decreased relatively rapidly and continuously.

Fig. 4.

Example of room-temperature, single-molecule kinetic measurement of the annealing reaction between immobilized Cy3-TAR DNA and 25 nM Cy5-loop DNA in the presence of NC (889 nM). The sample was then washed out by buffer and NC sequentially, reannealed with 25 nM Cy5-labeled loop DNA (reannealing), and reacted with 25 nM Cy5-cTAR DNA (cTAR). All solutions contained 0.2 mM Mg²⁺. (Top) Number of annealed (blue) and unannealed TAR DNA (green) molecules at different reaction stages. (Middle) Single-molecule E_A trajectories, where each colored line corresponds to a single molecule. (Bottom) Mean E_A of the single molecules in Middle. The final annealed percentages in each reaction stage are also shown in the corresponding graphs.

These data, therefore, demonstrate that NC not only catalyzes the forward annealing process but also catalyzes the reverse annealing process. This is expected according to microscopic reversibility for an NC-chaperoned process. Because NC interacts more strongly with single-stranded DNA than with double-stranded DNA, the combined NC effects strongly suggest that the transition state for the annealing reaction possesses more “single-stranded character” than the reactants, products, and even stable intermediates along the reaction path. This is qualitatively consistent with the proposed “Z” nucleation complex structure in Fig. 1. The final epoch in Fig. 4 involves an NC-promoted strand displacement of loop DNA by cTAR. This step in the sequence is a simple assay on the “activity” of the immobilized TAR hairpins, allowing for a validation of the entire procedure. Because efficient cTAR annealing was observed, it is unlikely that the TAR DNA hairpins were damaged or poorly immobilized by the programmed sequence of reagents. Very similar results were observed for the annealing of TAR DNA to zipper DNA, zipper RNA, and loop RNA.

Various observables and derived kinetic parameters for the annealing reaction of TAR DNA with the various nucleotides are listed in Table 1. Some clear trends are apparent in the data. For example, NC is clearly required for both the annealing and reverse annealing reactions to be rapid. In addition, the reverse annealing of the full-length oligonucleotides (cTAR and TAR RNA) is much slower than that of the short oligonucleotides, consistent with the idea that the annealed adduct of the TAR DNA with the short oligonucleotides is indeed a model for the nucleation complex for the full-length annealing reactions. In other words, considerably more base pairs must be broken to achieve the proposed transition state for reverse annealing of the full-length oligonucleotides. Presumably, the reverse annealing of the TAR DNA/zipper DNA adduct is analogous to the slow reverse reaction process in Fig. 1 that converts “Z” to “TN···CN,” demonstrating reversibility in this step of the mechanism.

Table 1.

Rate constants for annealing (k_a) and reverse annealing (k_r) at 0.2 mM Mg²⁺ in the presence or absence of 889 nM NC

Parameter*	cTAR DNA	Zipper DNA	Loop DNA†	TAR RNA	Zipper RNA	Loop RNA†
ka (105 s−1·M−1)
−NC	0	0	0	0	0	0
+NC	8‡	10	1	2	0.1	0.06
kr (10−4 s−1)
−NC	0	1§	0.1§	0	5	0
+NC	0	18	4	0	5	3

*k_a denotes the annealing rate constant, and k_r denotes the reverse annealing rate constant. The annealing reaction kinetic curve is measured in the presence (+NC) or absence (−NC) of NC and fitted by a single exponential function. The single exponential rate constant k is k = k_r + k_a · [NA] (NA = reacting nucleic acid). k_r is determined by washing out the annealed product with NC (+NC) or buffer only (−NC) solution and fitting the kinetic curve by a single exponential function with a rate constant k_r. k_a is determined from k and k_r as k_a = (k− k_r)/[NA].

^†The loop annealing reaction kinetics is potentially complicated by steric interference of the biotin/streptavidin attachment group (denoted by ″B″ in Fig. 1). To avoid this complication, loop annealing kinetic were measured with an inverted TAR DNA construct, with the B group attached to the T64. Although there does not appear to be appreciable steric interference for the TAR DNA-loop DNA annealing (5), there is measurable steric interference of the TAR DNA-loop RNA annealing.

^‡This value is from our previously published data (17).

^§This value is based on the slow decay component of N_P. The fast decay component is associated with reverse annealing of one-armed annealed TAR DNA/zipper or loop DNA adduct.

Further evidence for the reversibility of this pathway was obtained by forming an equilibrium mixture of the TAR DNA/zipper DNA adduct by NC annealing and then rapidly replacing the NC plus zipper DNA solution with an NC plus cTAR (dye-labeled) solution. The zipper DNA hairpins were observed to be efficiently replaced by the cTAR (data not shown). Presumably, this net “strand displacement” reaction occurred because of NC-induced annealing between cTAR and the unannealed TAR DNA. (The unannealed TAR DNA presumably resulted from NC-induced reverse annealing of the TAR DNA/zipper DNA adduct.) Various strand displacement reactions were undertaken as summarized in Table 2. In every case where a stable annealed product between TAR DNA and a full-length oligonucleotide was expected, the strand displacement reaction was too slow to observe. For example, it was observed that the cTAR in a TAR DNA/cTAR adduct were not efficiently displaced by a concentrated mixed solution of NC and zipper DNA (and also not washed out by NC only). Correspondingly, for all adducts of TAR DNA with short oligonucleotides, efficient strand displacements of the short oligonucleotide by another oligonucleotide were observed.

Table 2.

Summary of strand-displacement experiments

Reaction sequence
Efficient strand displacement
Cy3-TAR/Cy5-zipper DNA + Cy5-cTAR
Cy3-TAR/Cy5-loop DNA + Cy5-cTAR
Cy3-TAR/Cy5-zipper RNA + Cy5-cTAR
Cy3-TAR/Cy5-loop RNA + Cy5-cTAR
Cy3-TAR/Cy5–14-mer DNA + cTAR
Inefficient strand displacement
Cy3-TAR/cTAR + Cy5-cTAR
Cy3-TAR/cTAR + Cy5-loop DNA
Cy3-TAR/cTAR + Cy5-cTAR + Cy5–14-mer DNA

The annealing and replacement experiments were run in buffer containing 0.2 mM Mg²⁺ and 889 nM NC, 25 nM Cy5-zipper/loop nucleotide, and Cy5-cTAR DNA with the exception that the Cy5–14-mer DNA measurement was run with 50 nM Cy5–14-mer DNA and 2.5 nM unlabeled cTAR.

Further insights into the annealing mechanism were obtained by a different type of application of strand displacement, involving a short Cy5-labeled DNA oligonucleotide that is complementary to L3L4 (see 14-mer in Fig. 2). We used this 14-mer DNA as a “dynamic probe” for base paring in the TAR DNA L3L4 region during the irreversible annealing of cTAR (Fig. 5). Because of the small number of base pairs in the TAR/14-mer adduct, its reverse annealing rate is very rapid. This was demonstrated by the single-molecule FRET trajectories in Fig. 5 Upper. Starting with a TAR DNA sample with no NC (epoch I), a solution of NC and Cy5-labeled 14-mer was added at the beginning of epoch II. Complete annealing was observed within the mixing time of the flow system, followed by rapid reversible on–off events in E_A trajectories due to rapid annealing/reverse annealing. The rapid NC-induced annealing is also reflected in the rapid rise of the mean FRET as a function of time curve (Fig. 5 Lower, epoch II). The most interesting part of Fig. 5 is the epoch III. At the beginning of epoch III, cTAR DNA was added to the NC and Cy5-labeled 14-mer solution, and a rapid irreversible displacement of the 14-mer by cTAR was observed at the usual time scale for TAR DNA/cTAR annealing. The fact that the irreversible annealing of nonlabeled cTAR DNA with TAR occurs with a concomitant displacement of the 14-mer is highly consistent with the hypothetical mechanism in Fig. 1, which assumes that full annealing occurs soon after the rate-limiting nucleation step.

Fig. 5.

Room-temperature, single-molecule E_A trajectories (Upper) and ensemble mean E_A (Lower) during the annealing reaction of Cy5–14-mer DNA annealed to Cy3-TAR DNA and then replaced by nonlabeled cTAR DNA. Three time epochs during the reaction are plotted: I, no annealing occurred in the presence of 50 nM Cy5–14-mer DNA only; II, reversible annealing reaction occurred in the presence of 50 nM Cy5–14-mer DNA and 889 nM NC; and III, the annealed Cy5–14-mer DNA was displaced by nonlabeled cTAR DNA in the presence of 2.5 nM nonlabeled cTAR DNA, 50 nM Cy5–14-mer DNA, and 889 nM NC. All reactions were run in the presence of 0.2 mM Mg²⁺.

Although the main features of the SMS data are well explained by the mechanism in Fig. 1, certain small but discernable features in the data suggest that the actual mechanisms may be somewhat more complex. For example, consider the interruption of the annealing equilibrium of the TAR DNA/zipper DNA adduct with a buffer-only washout step, as shown in Fig. 6. As expected from the mechanism in Fig. 1; the population of the adduct decayed only very slowly after the buffer entered the cell; i.e., the reverse annealing step requires NC to occur rapidly. The E_A trajectories also exhibit the expected underlying behavior; i.e., very few transitions between the low and high FRET levels after the buffer solution enters the sample cell. However, unexpectedly, immediately after the TAR/zipper DNA adduct was exposed to a buffer-only solution, a small but statistically significant drop was observed in the number of TAR DNA/zipper DNA adducts (N_P). A corresponding, unexpected small decrease in 〈E_A〉 was also observed. Both effects were reproduced in several trials. We tentatively assign the small, rapid drop in N_P to rapid reverse annealing of a conformational substate of the TAR/zipper adduct that is only partially annealed and, as a result, can be washed out by the incoming buffer solution. A good prospect for this substate is a “one-arm” annealed form of the TAR DNA/zipper DNA adduct, as outlined in the equation at the top of Fig. 6. The “one-arm” annealed form is analogous to the previously reported rapidly reversible annealing of TAR DNA with a 13-mer short DNA oligonucleotide (5), which is complementary to only one arm of the “Y” form of TAR. This nonideality does not appear to be a factor for the TAR DNA/loop DNA adduct or for the full-length TAR DNA/cTAR reaction itself. Of course, all of the above observations are subject to our time resolution (>120 sec). The faster events beyond the time resolution will be missed, and the actual mechanism could be more complex.

Fig. 6.

Comparison between single-molecule measurements on buffer washout (0.2 mM Mg²⁺) after immobilized Cy3-TAR DNA plus Cy5-zipper (Left) and immobilized inverted Cy3-TAR DNA plus Cy5-loop (Right) DNA annealing reaction. (Upper) Number of annealed (blue) and unannealed molecules during washout. (Lower) Single-molecule E_A trajectories. The equation shows a hypothetical scheme for the formation of a one-arm annealed TAR DNA/zipper DNA adduct.

Conclusions

An SMS approach has been developed for analyzing the mechanism of protein-induced nucleic acid rearrangements that involve multiple oligonucleotides and are highly heterogeneous and complex. The approach involves exposing one of the oligonucleotides (which is immobilized) to a sequence of solutions containing the complementary oligonucleotide, other target oligonucleotides, buffer only, and the chaperone protein in various combinations. This procedure effectively drags the nucleic acid/protein systems from reactants through key intermediates and finally toward the rearranged products while monitoring the conformational states and dynamics of the system with single-molecule FRET. Here, this sequence has been applied to investigate the NC protein-chaperoned annealing of TAR DNA to cTAR DNA/TAR RNA, which is an in vitro model for the minus-strand transfer step in HIV-1 reverse transcription. The results strongly suggest that the nucleation event for annealing of transactivation response region TAR DNA to cTAR DNA involves base pair formation in different regions of TAR DNA. In addition, the results clearly demonstrate that NC induces reversible annealing at various stages along the reaction path of the annealing reaction.

Experimental Methods

SMS data were recorded by repetitive confocal scanning imaging as described in refs. 5 and 17. HIV-1 NC was synthesized as described in ref. 5. Various functionalized DNA hairpins (all purchased from TriLink BioTechnologies, San Diego, CA) and RNA hairpins (purchased from Dharmacon, Lafayette, CO) were used without further purification as described in ref. 5. TAR DNA was immobilized on the coverslip of an assembled flow chamber. Three syringe pumps delivered three solutions, NC, target complementary hairpins (containing Mg²⁺), and buffer solution (containing buffer A and Mg²⁺), separately. All of the solutions contained buffer A [40 mM NaCl/25 mM Hepes, pH 7.3/glucose oxygen scavenger system (5)].

Abbreviations

cTAR

complementary TAR

SMS

single-molecule spectroscopy

TAR

transactivation response element.