Binding of RNA template to a complex of HIV-1 reverse transcriptase/primer/template

Abstract

HIV-1 reverse transcriptase (RT) catalyzes the synthesis of DNA from DNA or RNA templates. During this process, it must transfer its primer from one template to another RNA or DNA template. Binary complexes made of RT and a primer/template bind an additional single-stranded RNA molecule of the same nucleotide sequence as that of the DNA or RNA template. The additional RNA strand leads to a 10-fold decrease of the off-rate constant, k_off, of RT from a primer/DNA template. In a binary complex of RT and a primer/template, the primer can be cross-linked to both the p66 and p51 subunits. Depending on the location of the photoreactive group in the primer, the distribution of the cross-linked primers between subunits is dependent on the nature of the template and of the additional single-stranded molecule. Greater cross-linking of the primer to p51 occurs with DNA templates, whereas cross-linking to p66 predominates with RNA templates. Excess single-stranded DNA shifts the distribution of cross-linking from p66 to p51 with RNA templates, and excess single-stranded RNA shifts the cross-linking from p51 to p66 with DNA templates. RT thus uses two primer/template binding modes depending on the nature of the template.

Replication of the RNA genome of the HIV-1 is carried out by reverse transcriptase (RT), the product of the viral pol gene. RT uses the genomic single-stranded (+) RNA (ssRNA) as a template to synthesize a double-stranded DNA (dsDNA) copy. The active form of RT is composed of two subunits of 66 and 51 kDa (p66 and p51, respectively) (1, 2). Site-directed mutagenesis has led to the assignment of DNA- and RNA-dependent DNA polymerase, and ribonuclease H (RNase H) activities found in the p66/p51 complex to the p66 subunit, with one active site per dimer (3).

RT is a remarkable enzyme in that it uses RNA primers, such as tRNA^3,lys and the polypurine tract (4), as well as DNA primers for DNA synthesis. The template can be either RNA or DNA. RT also mediates the transfer of primer strands from one molecule to another. The first strand transfer occurs when the nascent DNA strand reaches the 5′ end of the genomic (+) RNA, producing the “strong stop (−) DNA” (5, 6). The primer DNA strand is then relocated on the RNA genome, and the complementary synthesis of (−) DNA resumes. This strand transfer could involve a RNA/DNA/RNA intermediate, but such a complex has not been demonstrated (7, 8). The second strand transfer occurs when RT is bound to dsDNA (5, 9) to drive DNA replication to completion prior to the integration of the final dsDNA provirus. In this second strand transfer, a single-stranded DNA (ssDNA) invades RT bound to a DNA/DNA duplex to mediate bidirectional priming. Complete DNA synthesis requires strand displacement to obtain a full length, functional dsDNA proviral copy (10). Again, a putative DNA/DNA/DNA intermediate could be involved (9).

Although there is considerable information on the mechanism of nucleotide polymerization (11–15), little is known about the mechanism by which RT binds to primer/templates and the process by which strands are transferred from one template to another. It seems reasonable that such highly specific transfer must be mediated in part by the RT itself. Such a reaction would undoubtably involve a ternary complex of RT, a primer/template complex, and an additional strand to which the primer could be transferred (the acceptor strand).

In this communication, we demonstrate the formation of a ternary complex consisting of RT, a primer/template, and an additional ssRNA molecule homologous to the template. In addition, we show that RT interacts with a primer/template in different ways depending on the nature of the template and of the additional single-stranded molecule.

MATERIALS AND METHODS

Proteins, Reagents, Expression, and Purification of Enzymes.

Recombinant wild-type HIV-1 RT was overproduced in Escherichia coli and purified to apparent homogeneity. The enzyme had a specific activity of 9,500 units/mg protein. One unit of enzyme catalyzes the incorporation of 1 nmol of dTMP in 10 min at 37°C. All RT concentrations were determined spectrophotometrically (13). Oligonucleotides were 5′-³²P-labeled using T4 polynucleotide kinase. p-Azidophenacyl bromide was from Sigma. The RNA template and DNA oligonucleotides containing a phosphorothioate linkage were obtained from Oligos Etc. (Guilford, CT). DNA primers were synthesized by the Biopolymers Laboratory, Harvard Medical School. RNA and DNA primers were purified by electrophoresis through an 18% polyacrylamide gel. ³²P-labeled nucleotides were from Amersham. 3′-Deoxynucleoside 5′-triphosphates were from Pharmacia.

Synthesis and Purification of Primers Containing a p-Azidophenacyl Photoactive Probe.

Upon UV irradiation at 365 nm, aromatic azido groups react with protein side-chains to form a covalent bond. Attachment to DNA primers of a photoreactive azidophenacyl moiety was as described (16). In brief, 2 mg of phosphorothioate-containing oligonucleotides were dissolved in 0.3 ml of 150 mM potassium phosphate buffer (pH 7.0) mixed with 0.75 ml of p-azidophenacyl bromide (5 mg/ml in methanol) and incubated for 3 hr at room temperature. Reaction products were precipitated with ethanol, redissolved in water, and lyophilized. The modified oligonucleotides were purified using C₁₈ reverse-phase HPLC. Annealing of the primer/templates was as described (13). The phosphorylation of the 5′ end of each oligonucleotide was only slightly affected by the photoprobe at the 5′ end of oligo-3 (see Fig. 4). The labeling efficiency for this oligonucleotide was only 5% lower than for oligo-1 and -2.

Figure 4.

Cross-linking of RT to primers (21-mers) containing a p-azidophenacyl photoreactive group at specific positions along the DNA primer. (A) Location of the photoreactive groups on primer/templates. Based on an approximate value of 10–11 bases per helical turn, the three different azidophenacyl oligonucleotides probe opposite sides of the duplex. Only two bases out of the 10 base 5′ overhang of the 31-mer template are represented. The photoreactive group is attached to the 8th, 15th, and 20th phosphate internucleotidic bond of the 5′ end labeled 21-mer primer, giving oligo-1, -2, and -3, respectively. (B) Electrophoretic analysis of the cross-linked products using RT and 5′-³²P end labeled oligo-1, -2, and -3. RT was allowed to bind to 5′-³²P-labeled oligo-1, -2, or -3 annealed to a 31-mer DNA or RNA template in RT buffer and 50 mM KCl. The mixture was UV-irradiated, heated in SDS/PAGE loading buffer, and loaded onto SDS/polyacrylamide gel containing 7 M urea. The figure is an autoradiograph of the gel. Loaded sample volume of reaction using oligo-3 was five times greater than for oligo-1 and -2. Estimated sizes corresponding to the bands are indicated (72 and 57 kDa for labeled p66 and p51, respectively, based on molecular weight markers). (C) Graphical representation of the radioactivity measurements of the bands shown in B. The percentage of radioactive label UV-attached to the p66 is represented as a function of the distance in nucleotides from the 3′ end of the primer. Radioactive signals were measured using a Fuji imaging apparatus.

RT-Mediated Band-Shift Analysis.

RT (50 nM) was allowed to bind to unlabeled primer/template (500 nM) for 10 min at 0°C in the presence of RT buffer (10 mM Tris⋅HCl, pH 8.0/10 mM MgCl₂/0.05% Triton X-100) and 5% glycerol. 5′-³²P-labeled acceptor ssRNA or ssDNA was added to a 1 μM final concentration and incubated at room temperature for the indicated time. The products were analyzed by nondenaturing 6% PAGE at 250 V in the cold room with a probe limiting the gel temperature to 15°C. The gel was run in TBE buffer (89 mM Tris/HCl, pH 8.0/89 mM sodium borate/2 mM EDTA) containing 10 mM MgCl₂. Bands were visualized by autoradiography and radioactivity was measured using a Fuji imaging apparatus.

Active-Site Titration and Off-Rate Constant (k_off) of RT from Primer/Templates.

Active-site titrations indicate the amount of RT which is catalytically competent for DNA polymerization relative to the total amount of enzyme. Active-site titrations were performed using either a fast-quench machine (Kin Tek, University Park, PA) (13) or manually (15). Only one saturating primer/template concentration was used in duplicate experiments to determine both the burst amplitude and the k_off of RT from various primer/templates. RT (50 nM) was incubated in RT buffer at room temperature with 250 nM primer/template for 2 min before an excess of a single nucleoside 5′-triphosphate (500 μM in 5 mM MgCl₂) was added to the reaction. Aliquots (5 μl) were withdrawn, quenched at various times (5–30 s) in 80% formamide, 20 mM EDTA, and analyzed using denaturing 18% polyacrylamide gel electrophoresis. Linear time courses did not pass through the origin (12, 13, 17), and the extrapolated intercept with the y axis gave the burst amplitude of the first phase. Both the amplitude of the initial burst and the slope giving the second phase steady-state constant were measured. These values correspond to the active enzyme E_a in the reaction mixture and the k_off (12, 13, 17). The percentage of active enzyme is expressed as the ratio of active enzyme E_a over total enzyme E₀ as determined spectrophotometrically.

DNA/Protein Cross-Linking Assays Using Photoprobe-Coupled Oligonucleotides.

p-Azidophenacyl modified primers were labeled and annealed in a similar fashion to that of unmodified primers. RT (250 nM) was mixed (1:1 vol/vol) with an equimolar amount of 5′-³²P-labeled oligo-1, -2, or -3 annealed to a DNA or RNA template on parafilm laid on a flat, cold (4°C) surface in a 10 μl final volume in 12.5% glycerol, 5 μg/ml BSA, 5 mM MgCl₂, 25 mM potassium phosphate buffer (pH 7.4), 1 mM DTT, and KCl as indicated. The mixture was irradiated (λ = 365 nm) with a two-bulb 15-W UV lamp at an approximate distance of 2 cm to yield a UV-irradiation dose of 0.1–0.2 W/cm² (18). The mixture was heated in SDS/PAGE loading buffer containing 80% formamide, 4% SDS, and 1 mM 2-mercaptoethanol at 95°C for 3 min and loaded onto a SDS/polyacrylamide gel containing 7 M urea. Radioactive signals were measured using a Fuji imaging apparatus, and the gel was exposed to x-ray films. Measurements of radioactive label incorporated indicated that ≈10% of the input protein was labeled.

Three-Dimensional (3D) Computer Modeling of RT.

The coordinates of RT were obtained from the Brookhaven Protein Database and corresponded to data for the protein backbone (accession no. 1RTH) (19), and for the DNA backbone (accession no. 1HMI) (20). The three dimensional model was displayed using the program o (21). The surface potential of RT was calculated using the program grasp (22).

RESULTS

First, we have investigated the ability of RT to form a ternary complex made of RT, primer/template, and exogenous ssDNA or ssRNA. Second, we have carried out experiments to show how primer/templates are bound by RT in the presence or absence of exogenous ssDNA or ssRNA.

Binding of ssRNA to a RT/Primer/Template Complex.

We have used a mobility-shift gel assay to monitor the binding of RT to a primer/template. The mobility of the primer/template through a nondenaturing polyacrylamide gel is retarded in the presence of RT. To determine if RT is able to bind an additional ssDNA or ssRNA molecule, we first saturated the enzyme with an unlabeled primer/template made of a 21-mer DNA primer annealed to either a 31-mer DNA or an RNA template. Then, we added a 5′-³²P-labeled 31-mer oligonucleotide of either ssDNA or ssRNA, referred to as exogenous ssDNA or RNA. After an incubation of 10 and 40 min at 23°C the reaction was analyzed by nondenaturing PAGE to determine if any of the radioactively labeled ssDNA or ssRNA migrated with the RT.

As shown in Fig. 1A, exogenous ssRNA molecules bound to the preformed RT/primer/template complex when they had the same nucleotide sequence as that of the template in the RT complex (arrow head 4 in lanes 7, 8, 11, 12, 14, and 15). The ssRNA bound to the complex regardless of whether the template was DNA (lanes 7, 8) or RNA (lanes 11, 12, 14, and 15). The amount of ssRNA associated with the RT/primer/template complex was ≈0.8 molecule per complex as determined by quantitation using photostimulable plates. The ssRNA bound to the preformed complex (arrowhead 4) migrated slower than did control RT/primer/templates (arrowheads 2 and 3), indicating that it was composed of RT, primer/template, and exogenous ssRNA (lanes 7, 8, 11, 12, 14, and 15).

Figure 1.

Mobility-shift gel analysis of RT bound to excess primer/template in the presence of exogenous single-stranded nucleic acids. (A) RT was bound to a 10-fold excess of 21-mer primer annealed to a 31-mer template and then incubated in the presence of a 20-fold excess of 5′-³²P-labeled 31-mer exogenous ssRNA or ssDNA as indicated by R or D, or of a 20-fold excess of 5′-³²P-labeled 21-mer exogenous ssDNA primer, as indicated by P, below each lane. Lanes 1–4: the primer/template (labeled DNA_NSH) had no sequence homology with the 5′-labeled exogenous ssDNA or ssRNA. Lanes 5–17: the 5′-labeled exogenous ssDNA or ssRNA had the same nucleotide sequence as that of the template. Arrowheads 1–4 indicate radioactive complexes of decreasing mobility. Arrowheads 2 and 3 indicate radioactive bands that comigrated with controls made of RT bound to a labeled DNA primer/DNA template and DNA primer/RNA template, respectively. Arrowheads 1 and 4 indicate novel complexes that are discussed in the text. D, R, P: 5′-labeled exogenous ssDNA template, ssRNA template, and DNA primer, respectively. Incubation time with exogenous templates was for 10 min (lanes 1, 3, 5, 7, 9, 11, 14, and 16) and 40 min for all other lanes before loading the gel. (B) RT was bound to a 10-fold excess of the same primer. template as above, labeled as indicated by ∗, and then incubated for 20 min in the presence of a 20-fold 5′-³²P-labeled (R*) or cold 31-mer exogenous ssRNA (R). Lanes 1–3: labeled primer annealed to DNA template (lane 1) and RNA template (lane 2 and 3). Lanes 4–6: labeled RNA template. 5′-³²P-labeled 31-mer exogenous ssRNA was present in lanes 3 and 4, and cold 31-mer exogenous ssRNA in lane 5. The amount of radioactivity in lanes 1, 2, 3, 4, 5, and 6 relative to lane 2 was 0.9, 1.0, 1.8, 2.0, 1.1 and 1.0. D, R, and P stand for DNA, RNA, and DNA primer, respectively.

When the primer/template lacked homology with the exogenous ssRNA, no radioactivity was associated with the complex (Fig. 1A, lanes 3 and 4). The lesser amount of radioactivity appearing in the more rapidly migrating species observed in lanes 3 and 4 was due to binding of ssRNA to a fraction of RT with no bound primer/template (arrowhead 1).

In contrast to the binding of ssRNA to the RT complex, ssDNA having the same sequence as that of the template in the RT complex barely bound to the complex (lanes 5, 6, 9, and 10). The lesser amount of radioactivity appearing in these complexes migrated together with a control complex consisting of RT and primer/DNA template (arrowhead 2), suggesting that a small fraction of exogenous ssDNA had exchanged with the template of the preformed RT/primer/template complex.

When the 5′-³²P-labeled exogenous molecule was a ssDNA of a sequence unrelated to the primer/template, no radioactive signals were detected in the vicinity of the RT/primer/template complex (lanes 1 and 2). Likewise, addition of 5′-³²P-labeled ssDNA having the sequence of the primer did not yield any high molecular weight RT complex (lanes 16 and 17). However, the small amount of radioactivity seen in lanes 16 and 17 (arrowhead 3) presumably, corresponds to a RT/primer/RNA template complex formed when a fraction of the 5′-labeled exogenous primer anneals to free RNA template present in the medium. The position of the resulting complex is useful to confirm that this RT/primer/RNA template (arrowhead 3) migrates faster than the complex made of RT, primer/template, and exogenous ssRNA (arrowhead 4).

Since this latter complex (arrowhead 4) might be relevant to the first primer transfer reaction, we designed an experiment to confirm the presence of two ssRNA molecules per complex. In this experiment, RT is bound to an excess of primer/RNA template, in which either the primer or the template is 5′ labeled. The resulting RT/primer/template complex is subsequently incubated with an excess of either 5′-labeled or cold exogenous ssRNA to determine if the exogenous template is taken up by the complex or displaces the initial RNA template.

As shown in Fig. 1B, the amount of radioactivity found in the initial complex increased 2-fold in the presence of 5′-labeled exogenous ssRNA (lane 3 and 4), whereas it remained the same when the exogenous ssRNA was not labeled (lane 5) or omitted (lane 6) relative to controls (lanes 1 and 2). A 10-fold or 20-fold excess of 5′-labeled exogenous ssRNA gave the same results, indicating that this site was probably saturable. Under our experimental conditions, RNase H activity does not seem to affect the stability of the complexes: the amount of radioactivity in the same in lanes 2, 5, and 6 (Fig. 1B, and data not shown).

We conclude that RT bound to its primer/RNA template can bind an additional ssRNA, but not ssDNA, of the same sequence as that of the template. When bound to its primer/DNA template, RT can accommodate a ssRNA, but not ssDNA, of the same sequence as that of the template.

ssRNA Bound to RT/DNA/DNA Complex Decreases the Off-Rate of RT.

It was of interest to assess if RT could catalyze nucleotidyl transfer in the presence of an additional ssRNA molecule. In addition, we wished to determine if the presence of ssRNA in the RT/primer/template complex changed the binding affinity of RT for the primer/template.

To address these issues, we examined the incorporation of a single nucleotide into preformed RT/primer/templates in the presence of ssRNA under conditions where the exogenous RNA complementary to the DNA primer would bind to the RT/primer/template complex (Fig. 1). When RT is allow to bind to an excess of primer/template, the time course of single nucleotide incorporation is biphasic (12, 13, 17). The first, rapid phase of product formation corresponds to the reaction rate occurring before or during formation of the phosphodiester bond. The amplitude of this first phase is directly related to the concentration of enzyme active sites, E_a. The second, slow phase corresponds to the overall rate of enzyme cycling to produce a new, catalytically competent enzyme/primer/template complex. This second rate constant is rate of dissociation (k_off) of the polymerase from its primer/template under the conditions of this assay (12, 13, 17). Therefore, a single nucleotide incorporation assay provides a convenient way to determine both the proportion of active enzyme E_a/E₀ in a given preparation as well as the k_off of the polymerase from its primer/template. We used such an assay to measure both the proportion of active sites E_a/E₀ on the preformed RT/primer/template complexes as well as the k_off of RT from its primer/template in the presence of an excess of exogenous homologous ssDNA or RNA.

No differences were found in the proportion of active sites whether or not an additional ssDNA or ssRNA was present in the incorporation reaction (Fig. 2A). Thus, although exogenous ssRNA associates with the complex (Fig. 1), its presence does not lock the complex into a form catalytically incompetent for DNA synthesis.

Figure 2.

Pre-steady-state analysis of single nucleotide incorporation into primer/templates in the presence of exogenous homologous ssRNA or DNA. The 21-mer primer annealed to a 31-mer DNA or RNA template was the same as that of Fig. 1. (A) The k_off of RT from a DNA primer annealed to DNA or RNA templates in the presence of exogenous homologous ssDNA or RNA. (B) Active-site titration of RT for nucleotide polymerization.

The off-rate of RT from a preformed complex was 10-fold higher when DNA rather than RNA was the template (Fig. 2B). But even with DNA templates, in the presence of an additional ssRNA molecule k_off decreased 10-fold to the same value as when RNA was the template. Interestingly, even though ssRNA does bind to the RT/primer/RNA template complex, its addition does not further stabilize the complex. As expected from the previous section, additional ssDNA had almost no effect on the off-rate of RT from any complex.

Design of a Primer/Template Probe to Map the Protein-Nucleic Acid Contacts.

As shown in the previous section, a ssRNA molecule homologous to the template affects the binding properties of RT to a primer/template containing a DNA but not an RNA template. The presence of the ssRNA exogenous molecule might have led to the repositioning or displacement of the primer/DNA template in a similar way as when RNA was the template. In both cases, DNA primers annealed to a DNA or RNA template would not bind to RT in the same manner. To test this hypothesis, we used a high-resolution mapping technique (18) to determine whether primer/templates are positioned differently on the RT molecule depending on the nature of the template or the exogenous homologous ssRNA molecule.

The spatial coordinates of a binary complex of RT with a short duplex (20) show that RT can be viewed as a right hand holding the nucleic acid substrate in a large crevice, which extends from the “palm” subdomain of RT to the RNase H active site (Fig. 3). Most of the bottom of this crevice is made of the interface of p66 and p51, so that each strand of the duplex contacts alternatively p66 and p51 sites along the nucleic acid helix (Fig. 3).

Figure 3.

3D model of a duplex DNA bound to the RT p66 (green)/p51 (white) heterodimer taken from ref. (20). An extension with four phosphodiester bonds at the 5′ end of the primer and the 3′ end of the template was modeled using the program o (21). The duplex is a hybrid A/B-type form of helix. Phosphate atoms representing the sites of attachment of the single photoreactive group in oligo-1, -2, and -3 are shown in red on the yellow DNA primer. The template strand is shown in orange. See text and Fig. 4 for description of the system.

On the basis of this model, we designed a primer/template system consisting of the same primer/template as for the mobility-shift experiments, but in which a photoreactive group is coupled to phosphorothioates at defined positions along primer strand (Figs. 3 and 4A). The chemistry of such amino acid azidophenacyl cross-linker has been described (18). The position of the phosphorothioate linkage fixes the distance between the probe and the primer 3′ end. Three different primer/template probes were designed: Oligo-1, -2, and -3, carry a single photoprobe at phosphates located 8, 15, or 20 nucleotides, respectively, from the 3′ end of the 21 nucleotide long primer (Figs. 3 and 4A). Consequently, once the 3′ end of the duplex is positioned at the polymerase active site, these photoreactive groups point toward either p66 or p51 along the double helix. With such a system, each side of the DNA/DNA or DNA/RNA helix can be probed independently for interaction with either p51, p66, or both subunits. This model predicts that oligo-1 should label p51 more than p66, oligo-2 should label p66 only, and oligo-3 should label both subunits (Fig. 3).

Mapping the Primer/Template Position on RT.

Using the three primers described above, each of which contains a single photoprobe, we determined the distribution of cross-linking to each of the two subunits of RT. These contacts were compared between RNA and DNA templates. RT was incubated with primer/template in which the photoprobe-containing primer was ³²P-labeled at the 5′ end, the mixture was UV-irradiated, the products separated on a denaturing polyacrylamide gel, and radioactive signals corresponding to the primer attached to either p51 or p66 were measured.

When the photoprobe was located at phosphate position 8 (oligo-1), both the p51 and p66 subunits were cross-linked to the radioactively labeled primer strand as evidenced by the appearance of radioactivity associated with 57- and 72-kDa species (Fig. 4B), precisely the molecular weights expected for p51 and p66 cross-linked to the 21-mer primer, respectively. The p51 subunit reacted with approximately 75% of the bound oligo-1 primer regardless of whether DNA or RNA was used as the template. (Fig. 4C).

However, when the photoprobe was located at phosphate position 15 (oligo-2), the distribution of the labeled subunits was significantly different. In the presence of a DNA template, ≈30% of the label was contained in p66 whereas in the presence of an RNA template, 60% of the label was on p66. When the photoprobe was located at phosphate position 20 (oligo-3), a 6-fold lower overall labeling of both subunits was observed, and the distribution of label between p51 and p66 was comparable.

In view of the effect of ionic strength on strand transfer (23, 24), cross-linking reactions were performed in the presence of 50 mM or 150 mM KCl. KCl enhanced the differential labeling of the subunits between DNA and RNA templates: in the presence of standard RT buffer containing 50 mM KCl, oligo-2 resulted in 30 and 60% of cross-linking to p66 (Fig. 4) on DNA and RNA templates, respectively, whereas these values were 25% and 80% in the presence of 150 mM KCl (see Discussion and Fig. 5).

Figure 5.

Cross-linking to p66 and p51 subunits of 5′-³²P-end labeled primer in the presence of exogenous homologous ssDNA or ssRNA. Oligo-2 was used as the photoreactive primer probe. RT was allowed to bind to 5′-³²P-labeled oligo-2 annealed to a 31-mer DNA or RNA template in RT buffer containing 150 mM KCl in the presence or absence of exogenous homologous ssDNA or ssRNA. The mixture was UV-irradiated, heated in SDS/PAGE loading buffer, and loaded onto SDS/polyacrylamide gel containing 7 M urea. (A) Autoradiograph of the gel. Signals correspond to cross-linked products consisting of RT and 5′-³²P-end labeled oligo-2 on DNA (lanes 1–3) and RNA (lanes 4–6) templates. The presence or absence of exogenous ssDNA or ssRNA template is indicated below each lane. Estimated sizes corresponding to the bands are indicated (72 and 57 kDa for labeled p66 and p51, respectively, based on molecular weight markers). (B) Quantitation of the radioactivity present in the lanes of panel A. Each individual gel lane has one corresponding column located below on the graphic, which indicates the average relative distribution of label found on p66 for three independent experiments.

Based on the structural model of RT (Fig. 3) and for a right-hand helix, oligo-1 should label more p51 than p66, oligo-2 p66 only, and oligo-3 both subunits. Precisely this pattern was observed when RNA was the template. However, when a DNA template is present oligo-2 predominantly labels the p51 subunit (Fig. 4 A and B), a result difficult to reconcile with the 3D-model of RT and a primer/template (Fig. 3).

A trivial explanation would be that a portion of the duplex primer/template bound to the RT in the opposite orientation and that the blunt end of the duplex would be positioned at the active site. To rule out this possibility we examined the binding of a primer/template that had a 5′ overhang of one nucleotide at one end and a 5′ overhang of 10 nucleotides at the other. Simultaneous single-turnover incorporation of one nucleotide at each 3′ terminus revealed that all incorporation of nucleotide (>95%) occurred at the terminus bearing the 10-nucleotide overhang while the proportion of active sites was the same as with the previous primer/template (data not shown). Thus, the length of the single-stranded template is the major determinant to position the 3′ end of the primer/template in the active site.

We conclude that, although the 3′ end of the primer is positioned at the RT active site, the duplex DNA is either not positioned on RT as would be predicted from the 3D model, or the DNA is distorted upstream of the eighth phosphate bond from the 3′ end of the primer.

Influence of an Acceptor Single-Stranded Molecule on the Position of the Primer/Template.

To examine the relationship between a primer/template bound to RT and an exogenous RNA molecule of the same sequence as the template, we have used the primer containing a photoreactive group at the phosphate position 15 nucleotide from the 3′ terminus of the molecule. In this experiment we have determined the effect of exogenous homologous ssRNA and DNA. Using mobility-shift experiments, we had shown that exogenous ssRNA homologous to the template forms a stable complex with RT/primer/template only when the template is RNA.

RT was incubated with primer/template in which the photoprobe-containing primer (oligo-2) was 5′-³²P-end labeled, the mixture was UV-irradiated, the products separated on a denaturing polyacrylamide gel, and radioactive signals corresponding to the primer attached to either p51 or p66 were measured. In the absence of exogenous DNA or RNA, 25% of the cross-linked-labeled primer was on p66 when DNA was the template (Fig. 5, lane 1) whereas >80% of the cross-linked label was on p66 when RNA was the template (lane 4), in agreement with the previous section.

When exogenous ssRNA homologous to the template was added to the cross-linking reaction, the distribution of label between p51 and p66 was greatly changed in the presence of a DNA template (Fig. 5, lane 3). An excess of ssRNA template was able to induce a shift of labeling toward the p66 subunit on DNA templates, from 25% to 60% (Fig. 5, lanes 1 and 3). However, ssRNA had no effect on the distribution of the cross-linked labeled primer on RNA templates (Fig. 5, lanes 4 and 6).

Because ssDNA did not bind to any complex, it was expected to have no effect on the distribution of the cross-linked labeled primer. Indeed, labeling of the RT subunits showed no change upon addition of exogenous homologous ssDNA when the template was DNA (Fig. 5, lanes 1 and 2). Surprisingly, exogenous homologous ssDNA shifted the distribution of the cross-linked labeled primer from p66 to p51 when RNA was used as the template (Fig. 5, lanes 4 and 5, see Discussion).

We conclude that RT binds to a primer/RNA template in a different manner than it binds to a primer/DNA template; i.e., RT has two primer/template binding modes, and that the presence of exogenous homologous single-stranded molecules can switch the binding of primer/template from one mode to another.

DISCUSSION

The efficiency with which reverse transcription takes place suggests that specific mechanisms must exist to mediate the two strand transfers that occur during the first and second jumps. The economy of proteins involved in reverse transcription suggests that RT itself is the key player. Because the actual process of strand transfer involves a search for homology and subsequent homologous base pairing, it is not unlike similar reactions mediated by proteins of recombination such as the RecA protein of E. coli.

If RT is to mediate the recombination-like event it must establish a complex consisting of the donor primer/template and the acceptor template, and such a complex has been proposed (7, 8). During the first jump or strong stop strand transfer the proposed complex would consist of a donor DNA primer/RNA template and an acceptor RNA template. During the second jump, the complex would consist of a donor DNA primer/DNA template (or possibly a tRNA template) and an acceptor DNA.

How might RT accommodate such a recombination-like intermediate? Based on structural data, Arnold et al. (20) proposed that a dsDNA would occupy a large crevice extending from the polymerase active site to the RNase H active site. However, a reservation to the relevance of this binary complex is the fact that it was obtained in complex with an antibody, and the template 5′ overhang is of only one nucleotide. Although several 3D structures of RT have been published, this is the only structure of a RT/primer/template binary complex to date. Nevertheless, the surface potential of the RT molecule make the accommodation of an additional nucleic acid possible (Fig. 6), and the p51 subunit has been shown to be involved in tRNA binding (25).

Figure 6.

Surface potential of RT as determined using grasp (22) and data from ref. 20. Positive and negative areas are colored blue and red, respectively. The orientation of the RT molecule is the same as in Fig. 3.

This present work has shown that a ssRNA molecule forms a stable complex with RT and a DNA primer annealed to a DNA or RNA template. The requirement for complementarity between the ssRNA and the DNA/RNA primer/template is a feature of the first jump in which the strong stop DNA is transferred to a homologous sequence (the R sequence) on the viral RNA genome. Thus, the results presented here suggest that the first strand transfer occurs with an intermediate made of both donor and acceptor templates that can be formed without ongoing polymerization.

During the second jump, the acceptor molecule is a ssDNA. The results presented here with model oligonucleotides failed to detect any stable complex with exogenous ssDNA. This may be due to the fact that our model primer/templates do not mimic correctly the in vivo situation. Although we do not detect a stable complex, we do observe a shift of the position of a primer/RNA template on RT when ssDNA is present (Fig. 5, lanes 4 and 5). This apparent contradiction might indicate that the complex formed is not stable enough to be detected with a mobility-shift gel assay.

Interestingly, DNA duplexes are able to accommodate a ssRNA molecule. However, our results do not determine if the ssRNA molecule displaces the DNA template in a process apparent to homologous recombination. This situation does not mimic any known strand transfer intermediate of biological relevance. However, it might be relevant to recombination if the DNA template is interrupted to provide an alternate template. More importantly, once the second strand transfer has been performed, synthesis of (+) strand DNA occurs from the primer binding site sequence using the (−) strand DNA as a template. At that point, RT must synthesize DNA on a DNA template as well as perform strand displacement from the U5 to the U3 sequence of the viral DNA (24). Because RT has low processivity on DNA templates (23), we propose that template switching to an intact (+) RNA genome might provide an alternate and more efficient way of synthesizing the large terminal repeat.

How does RT bind these various nucleic acid substrates? One clue can be derived from the processivity of nucleotide polymerization of RT on the various templates. Processivity of nucleotide polymerization is defined as the average number of nucleotides that are incorporated in a single polymerization cycle before the enzyme falls off the nascent DNA strand. RT is ≈100-fold more processive on RNA than on DNA templates (23), and this difference can mostly be accounted for by the difference between the k_off from these two templates (11, 12). It is thus logical to expect that RT would make more extensive contact with primer/RNA templates (low k_off) than with primer/DNA templates (high k_off).

Our studies show that the organization of the DNA and RNA strands on the RT molecule depends on the nature of the template. Hydroxyl-radical footprinting of RT on DNA templates has established that DNA primer/templates adopt a duplex A-form in the RT active site (26) instead of the expected B-form. Whatever A-, B-, or hybrid A/B-form is adopted by the nucleic acid duplex according to Fig. 3, the structural differences induced by these helixes cannot account for the template-mediated changes of nucleic acid position on RT as determined by the cross-linking experiments. The position of a DNA/RNA duplex is consistent with the binary complex structure of Fig. 3, but results using the DNA duplex are not. We propose that a DNA duplex would not be bound as depicted in Fig. 3: either it would not enter the large cleft, or the primer/template would be shifted within the cleft. Several lines of evidence support our hypotheses. First, the DNA duplex cross-links almost exclusively to p51. Second, exogenous single-stranded molecules are able to induce a shift from one mode of binding to the other. Third, there is a strong correlation between high k_off values of RT and cross-linking of the primer to the p51 subunit.

Is there a potential ssRNA binding site on the p51 subunit? Examination of the surface potential of HIV-1 RT heterodimer suggests that p51 has a significant positive charge that may be involved in nucleic acid binding (Fig. 6). Our data suggest that ssRNA might well be accommodated in the seam that is visible on the p51 subunit. Although the seam on p51 is too narrow to accommodate a DNA duplex, its positive charge may be sufficient to orient a primer/DNA template on the p51 subunit from at least eight bases upstream the 3′ end of the primer.

ABBREVIATIONS

RT

reverse transcriptase

dsDNA

double-stranded DNA

RNase H

ribonuclease H

ssRNA

single-stranded RNA

ssDNA

single-stranded DNA

3D

three dimensional

k_off

off-rate constant