Hydrolysis of RNA/DNA hybrids containing nonpolar pyrimidine isosteres defines regions essential for HIV type 1 polypurine tract selection

Abstract

Both x-ray crystallography and chemical footprinting indicate that bases of the HIV type 1 (HIV-1) polypurine tract (PPT)-containing RNA/DNA hybrid deviate from standard Watson–Crick base pairing. However, the contribution of these structural anomalies to the accuracy of plus-strand primer selection by HIV-1 reverse transcriptase is not immediately clear. To address this issue, DNA templates harboring single and pairwise non-hydrogen-bonding isosteres of cytosine (2-fluoro-4-methylbenzene deoxyribonucleoside) and thymine (2,4-difluoro-5-methylbenzene deoxyribonucleoside) were synthesized and hybridized to PPT-containing RNA primers as a means of locally removing hydrogen bonding and destabilizing paired structure. Cleavage of these hybrids was examined with p66/p51 HIV-1 reverse transcriptase and a mutant carrying an alteration in the p66 RNase H primer shown to specifically impair PPT processing. Analog insertion within the PPT (rG):(dC) and central (rA):(dT) tracts repositioned the RNase H domain such that the RNA/DNA hybrid was cleaved 3–4 bp from the site of insertion, a distance corresponding closely to the spatial separation between the catalytic center and RNase H primer grip. However, PPT processing was significantly impaired when the junction between these tracts was substituted. Substitutions within the upstream (rA):(dT) tract, where maximum distortion had previously been observed, destroyed PPT processing. Collectively, our scanning mutagenesis approach implicates multiple regions of the PPT in the accuracy with which it is excised from (+) U3 RNA and DNA, and also provides evidence for close cooperation between the RNase H primer grip and catalytic center in achieving this cleavage.

Replication of HIV requires converting single-stranded viral RNA into a double-stranded DNA copy suitable for integration. This process takes place in multiple steps, each catalyzed by HIV reverse transcriptase (RT). Whereas minus-strand DNA synthesis is primed by a host-derived tRNA primer annealed to the RNA genome at the primer binding site, plus-strand synthesis initiates from the 3′ and central polypurine tracts (PPTs) derived from cleavage the RNA genome after minus-strand synthesis has occurred. These identical, purine-rich sequences are selected for plus-strand priming in part by precise RNase H-mediated cleavage at their 3′ termini. Initiation from the 3′ PPT represents an especially critical stage in virus replication, because incorrect priming would result either in truncation of the 5′ LTR and deletion of one or more transcriptional control elements, or extension of the preintegrative DNA and impaired integration (1, 2). Accordingly, PPT processing constitutes a potentially fruitful target for antiviral therapy, and the mechanistic basis by which this occurs is the subject of the current study.

Although the all-purine nature of the PPT sequence renders it moderately resistant to ribonuclease H (RNase H)-mediated hydrolysis (3, 4), recent data implicate structural features of a PPT/(–)DNA hybrid in its selection by orienting reverse transcriptase (RT) such that the RNase H domain is correctly positioned over the biologically relevant cleavage site. Sarafianos et al. (5) reported that nucleic acid in a cocrystal of HIV type 1 (HIV-1) RT and a PPT-containing RNA/DNA hybrid exhibits severe distortion at several positions, comprising a collection of weakly paired, unpaired and mispaired bases (Fig. 1A). Our subsequent chemical footprinting studies (6) indicated that template thymines in the same region, as well as the template thymine immediately 3′ to the PPT, do not assume standard Watson–Crick base pairing, even in the absence of RT. The observation that these naturally occurring PPT distortions are separated by ≈12 bp was particularly intriguing, because this distance approximates the spatial separation of the thumb subdomain and RNase H catalytic center of the heterodimerassociated p66 subunit (5, 7, 8). From these combined studies, we speculated that A-tract-induced PPT deformations helped sequester RT via an interaction with a structural motif at the base of the p66 thumb, thereby positioning the PPT-U3 junction in the RNase H catalytic center.

Fig. 1.

(A) Combined crystallographic and KMnO₄ footprinting data for the HIV-1 PPT and experimental rationale for nucleoside analog substitution. Solid bars represent loosely paired bases, among which two unpaired bases (circled) were identified by x-ray crystallography of the nucleoprotein complex. KMnO₄-sensitive template thymines in the absence of HIV-1 RT are shaded. From left to right, RNA sequences within the upstream and central (rA):(dT) and (rG):(dC) tracts are underlined. (B) Structures of the thymine and cytosine mimics dF and dD, respectively.

To examine whether alterations to nucleic acid geometry at different regions of the HIV-1 PPT influence processing, it was necessary to devise an approach that changed local flexibility while preserving base stacking and minimizing steric clashes resulting from “simple” base mismatches. Recently, a pyrimidine mimic, 2,4-difluoro-5-methylbenzene deoxynucleoside (dF), which is isosteric with thymine (9) but has severely reduced hydrogen bonding capacity, was used to understand the role of hydrogen bonding and steric effects in the fidelity of DNA synthesis (10–15) and in DNA mismatch repair (16, 17). A second pyrimidine mimic, 2-f luoro-4-methylbenzene deoxynucleoside (dD), a non-hydrogen-bonding cytosine isostere, was reported by Kool and coworkers (16) (Fig. 1B). Because the (–) DNA complement of the HIV-1 PPT RNA primer is exclusively pyrimidine (3′-TTTTCTTTTCCCCCC-5′), this substitution strategy offered the unique opportunity of locally perturbing PPT duplex stability by introducing single or pairwise dT → dF and dC → dD substitutions while retaining the sequence and all-purine nature of the plus-strand primer. The consequences of these substitutions on hydrolysis at the PPT-U3 junction by p66/p51 HIV-1 RT and an RNase H primer grip mutant (p66^Y501A/p51) specifically impaired in PPT processing (18, 19) are the subject of this study.

We demonstrate that locally removing hydrogen bonding in the (rG):(dC) or either of the (rA):(dT) tracts [referred to hereafter at the central and upstream (rA):(dT) tracts] alters both the precision and efficiency of PPT processing. In addition, nucleic acid pairing stability at the junction between the (rG):(dC) and central (rA):(dT) tracts appears critical for correct PPT function, because dC → dD substitutions here virtually eliminated processing, whereas substitutions 1 bp removed on either side had little effect. Finally, a recurring theme in this study is repositioning of the RNase H catalytic center 3–4 bp downstream of the site of analog substitution, placing the localized destabilization this induces within the p66 RNase H primer grip, a motif implicated in guiding the RNA strand of the hybrid into the catalytic center (5). From these combined studies, we propose a model for HIV-1 PPT recognition requiring that a motif at the base of the p66 thumb and the RNase H primer grip interact with regions of the RNA/DNA hybrid that have an unusual geometry and are separated by ≈7–8 bp. These combined interactions result in placement of the RNase H catalytic center over the biologically relevant processing site. Within the RNase H primer grip motif, studies with mutant RT suggest an important role for Tyr-501 in stabilizing the interaction.

Materials and Methods

Synthesis of Analog dD and Its Phosphoramidite Derivative. Synthesis details are provided in Supporting Text, which is published as supporting information on the PNAS web site, www.pnas.org. Reagents were purchased from Sigma unless noted otherwise. Solvents were made anhydrous by distillation. ¹H,¹³C NMR spectra were taken on a Varian NMR spectrometer (500 MHz). High-resolution mass spectroscopy spectra were performed by the mass spectrometry facility of the University of California (Riverside).

Oligodeoxyribonucleotides and Oligoribonucleotides. Forty-five-nucleotide DNAs containing dD or dF substitutions were synthesized at 1 μmol scale on an ABI model 394 synthesizer. DNA sequences and locations of substitutions are illustrated in the figures and described in the text. dF phosphoramidite was purchased from Glen Research (Sterling, VA). Stepwise coupling yields for incorporation of dF or dD were >98% determined by trityl cation monitoring. Deprotection and cleavage of oligonucleotides from the CPG support were carried out by incubation in concentrated ammonia for 12 h at 55°C. Oligonucleotides were purified by preparative polyacylamide gel electrophoresis and quantified by UV absorbance. The intact incorporation of new analog dD into DNA was confirmed by electrospray mass spectrometry. A complementary 30-nt RNA containing the HIV-1 3′ PPT and flanking sequences was purchased from Dharmacon Research (Boulder, CO).

Polypurine Tract Selection. PPT selection was evaluated as described (19) by using recombinant RT and the above-mentioned synthetic RNA and DNAs. Briefly, substrates were generated by 5′ end-labeling the 30-nt RNA and annealing it to each of the singly or doubly substituted 45 nt DNAs, as well as to an unsubstituted control DNA. Reactions were initiated by adding wild-type p66/p51 HIV-1 RT, or mutant p66^Y501A/p51 (19) to RNA/DNA hybrids in reaction buffer (10 mM Tris·HCl, pH 8.0/80 mM NaCl/5 mM DTT/6 mM MgCl₂) at 37°C, with enzyme and RNA/DNA present at final concentrations of 50 and 200 nM, respectively. Reactions were terminated after 10 min by adding an equal volume of a urea-based gel-loading buffer. Cleavage products were fractionated by high-voltage electrophoresis through 15% (wt/vol) polyacrylamide gels containing 7 M urea and visualized by autoradiography.

Melting Temperatures. Melting temperatures of hybrid duplexes were determined spectrophotometrically by using a Beckman DU640C Spectrophotometer equipped with a T_m Analysis accessory. Solutions (1 μM) of RNA/DNA in 10 mM Tris·HCl, pH 8.0/80 mM NaCl/5mMDTT/6 mM MgCl₂ were evaluated. E₂₆₀ was measured at 0.2°C intervals from 25°C to 90°C. The T_m of each hybrid was calculated by the “first derivative” method described by the manufacturer.

Results

Single dC → dD and dT → dF Substitutions in the HIV-1 PPT. Initially, the consequence of individual dC → dD and dT → dF substitutions of the (rG):(dC) and two (rA):(dT) tracts, respectively, on processing of the HIV-1 PPT (Fig. 1 A) was examined. Template nucleotide +1 was also substituted with dF, because this base is sensitive to KMnO₄ oxidation (6), suggesting a local structural distortion. The results of this experiment are presented in Fig. 2. In keeping with published reports (19–21), the wild-type RNA/DNA hybrid was hydrolyzed primarily at the PPT-U3 junction (defined as –1) and to a lesser extent at positions –2 and –3 (Fig. 2, lane a). Replacing template nucleotide +1(dT) with dF relaxed cleavage specificity, promoting minor cleavage between +1 and +3 (Fig. 2, lane + 1). However, an adjacent dC → dD substitution at position –1 had a significantly stronger effect, causing HIV-1 RT to select the PPT-U3 junction and a second site at +3 with equal efficiency (Fig. 2, lane b). Partitioning of RT with an equivalent spatial relationship of 3–4 bp is also evident for –2 and –3 dC → dD substitutions, inducing cleavage at template nucleotides +2 and +1, respectively (Fig. 2, lanes c and d). However, this cleavage pattern is not observed with substrates containing dC → dD substitutions at other regions of the (rG):(dC) tract (Fig. 2, lanes e–g). dT → dF substitutions of the central (rA):(dT) tract did not severely compromise PPT processing, but continued to induce novel cleavage 3–4 bp from the site of analog insertion, albeit to a lesser extent (Fig. 2, lanes h–k). Finally, although equivalent substitutions at positions –12 through –14 of the upstream (rA):(dT) tract are compatible with PPT processing (Fig. 2, lanes l–m), a single –15 replacement reduced this to ≈10% of wild type (lane o). The latter result is consistent with data obtained by using a single-cycle virus replication assay (22), wherein a 71% reduction in virus titer was observed on introduction of a single rA → rC substitution mutation at position –15 of the 3′ PPT (J. Julias, personal communication).

Fig. 2.

Cleavage of mono-substituted HIV-1 PPT variants by wild-type HIV-1 RT. Cleavage at the PPT-U3 junction is designated –1. Lane +1, dT → dF substitution at template nucleotide + 1; lane a, wild-type PPT; lanes b–g, dC → dD substitutions of template nucleotides –1 through –6, respectively; lanes h–k, dT → dF substitutions of template nucleotides –7 through –10, respectively; lanes l–o, dT → dF substitutions of template nucleotides –12 through –15, respectively. The sequence of the PPT RNA/DNA hybrid and position of analog insertion are indicated.

Collectively, the data of Fig. 2 suggest that the interactions at either extremity of the HIV-1 PPT are particularly sensitive to alterations in hydrogen bonding or stability. In addition, the ability to induce novel cleavage 3–4 bp from the site of analog insertion suggests that the locally destabilized RNA/DNA hybrid might interact with a structural motif of HIV-1 RT an equivalent distance from the RNase H catalytic center.

Dual dC → dD and dT → dF Substitutions. Processing profiles of PPT variants containing dual, overlapping dC → dD or dT → dF substitutions of their DNA template, designed to further increase the flexibility of the RNA/DNA hybrid, are shown in Fig. 3. The –1/–2 and –2/–3 substitutions in the (rG):(dC) tract were capable of completely displacing the RNase H catalytic center from the PPT-U3 junction to a novel site 3–4 bp from the position of analog insertion (Fig. 3, lanes b and c). Aberrant cleavage at an equivalent distance accompanied a –3/–4 substitution (Fig. 3, lane d), whereas a –4/–5 substitution yielded the correct profile (Fig. 3, lane e). In contrast, altering hydrogen bonding at positions –5/–6, i.e., at the boundary of the (rG):(dC) and central (rA):(dT) tract, rendered the PPT virtually refractory to cleavage (Fig. 3, lane f). This effect was clearly local, because processing was restored after a dual –7/–8 dT → dF substitution (Fig. 3, lane g). Additional substitutions through this tract induced a gradual reduction in processing at the PPT-U3 junction, but preserved aberrant cleavage within the PPT at a distance defined by the position of analog insertion (Fig. 3, lanes h and i). Finally, regardless of the position of dT → dF substitution in the upstream (rA):(dT) tract, PPT processing was severely inhibited (Fig. 3, lanes j–l). Interestingly, this gradual loss of processing function accompanying dual substitutions between positions –9/–10 and –14/–15 corresponds to the region of the hybrid shown previously by KMnO₄ footprinting to deviate from standard Watson–Crick base pairing (6). A thermal melting profile was obtained for each doubly substituted hybrid to exclude the possibility that gross structural deformations resulted from analog insertion. The T_m of the unsubstituted duplex was 73.9°C and was reduced maximally to 58.9°C for a –3/–4 dC → dD substitution (data not shown). From this analysis, we conclude that the PPT-containing hybrids were not inherently unstable at the temperature at which processing was evaluated (37°C).

Fig. 3.

Cleavage of dual-substituted HIV-1 PPT variants by wild-type HIV-1 RT. Lane a, wild-type PPT; lanes b–f, dC → dD substitutions of template nucleotides –1/–2 through –5/–6, respectively; lanes g–i, dT → dF substitutions of template nucleotides –7/–8 through –9/–10, respectively; lanes j–l, dT → dF substitutions of template nucleotides –12/–13 through –14/–15, respectively.

Analog-Induced Nucleic Acid Melting and the RNase H Primer Grip. The results shown in Figs. 2 and 3 indicate that the efficiency and precision of HIV-1 PPT processing are influenced by introducing non-hydrogen-bonding pyrimidine analogs into the template. With respect to specificity, a recurring theme within the (rG):(dC) and central (rA):(dT) tracts was repositioning of the RNase H catalytic center ≈3–4 bp from the site of analog insertion. These findings imply that a structural motif of the RNase H domain interacts with locally destabilized nucleic acid, thereby relocating the catalytic center this distance downstream. Based on the recent crystal structure of Sarafianos et al. (5), a logical candidate for this is the RNase H primer grip, a motif of p66 shown to contact nucleic acid 4–9 bp upstream of the scissile phosphate (Fig. 4). In fact, melted base pairs and an unpaired base are evident 4–6 bp from the scissile phosphate in the RT-RNA/DNA structure (5). From these combined observations, we hypothesized that the RNase H primer grip mediates selection of the PPT-U3 junction by accommodating a region of the RNA/DNA hybrid ≈5 bp upstream, thereby positioning the catalytic center over the biologically relevant cleavage site. Although this notion will be discussed in more detail later, one means by which our hypothesis could be tested would be to examine the influence of analog insertion on PPT processing by an enzyme specifically altered in its RNase H primer grip. These experiments are described below.

Fig. 4.

Proposed model for disposition of motifs of HIV-1 RT on the PPT. The model is designed after studies of Sarafianos et al. (5) and Kvaratskhelia et al. (6), showing PPT “unzipping” (between –8 and –14) and KMnO₄-sensitive template thymines, respectively. RNase H primer grip residues of p66 have been positioned on the duplex relative to His-539 of the catalytic center at the scissile bond, i.e., the PPT-U3 junction. Positioning of the thumb subdomain and polymerase active site is proposed from crystallographic data of RT complexes with either duplex DNA (7, 8) or RNA/DNA (5).

PPT Processing by an HIV-1 RNase H Primer Grip Mutant. Recently, we described a mutant of HIV-1 RT, p66^Y501A/p51, in which a highly conserved residue of the RNase H primer grip (19) was altered. Although retaining RNase H activity on a random RNA/DNA hybrid, this mutant was specifically impaired in PPT processing, implying an important role for Tyr-501 in this event. The use of PPT-containing RNA/DNA hybrids whose conformational flexibility was altered stepwise throughout the duplex appeared to be an ideal approach to understand RNase H primer grip function. Thus, an equivalent analysis was performed with p66^Y501A/p51 RT, the results of which are presented in Figs. 5 and 6.

Fig. 5.

Cleavage of mono-substituted HIV-1 PPT variants by p66^Y501A/p51 HIV-1 RT. Cleavage at the PPT-U3 junction is designated –1. Lane +1, dT → dF substitution at template nucleotide + 1; lane a, wild-type PPT; lanes b–g, dC → dD substitutions of template nucleotides –1 through –6, respectively; lanes h–k, dT → dF substitutions of template nucleotides –7 through –10, respectively; lanes l–o, dT → dF substitutions of template nucleotides –12 through –15, respectively.

Fig. 6.

Cleavage of dual-substituted HIV-1 PPT variants by p66^Y501A/p51 HIV-1 RT. Lane 1, wild-type PPT; lanes a–e, dC → dD substitutions of template nucleotides –1/-2 through –5/-6, respectively; lanes f–h, dT → dF substitutions of template nucleotides –7/-8 through –9/-10, respectively; lanes i–k, dT → dF substitutions of template nucleotides –12/-13 through –14/-15, respectively.

As shown (19), p66^Y501A/p51 RT cleaved the RNA/DNA hybrid at position +5, and to a lesser extent at +2, more efficiently than the at PPT/U3 junction (Fig. 5, lane a). A +1dT → dF substitution further reduced processing at the PPT 3′ terminus, whereas it induced cleavage at position +3. However single dC → dD substitutions introduced at positions –1 and –2 eliminated correct processing in favor of the novel site 3–4 bp removed (Fig. 5, lanes b and c). Substitutions at positions –3 and –4 resulted in authentic PPT cleavage, whereas they reduced the spurious event at position +2, effectively enhancing cleavage specificity (Fig. 5, lanes d and e). Structural features on one side of the junction between the (rG):(dC) and central (rA):(dT) tracts were clearly critical for correct PPT function, because –5 and –6 substitutions selectively inhibited cleavage at the PPT-U3 junction, whereas processing was restored after a single –7 substitution (Fig. 5, lane h). Additional dT → dF substitutions of the central and upstream (rA):(dT) tracts result in a progressive decrease in hydrolysis at the PPT/U3 junction (Fig. 5, lanes i–o).

Processing of PPTs doubly substituted in the (rG):(dC) tract by p66^Y501A/p51 RT (Fig. 6) bears a remarkable similarity to the profiles obtained with wild-type enzyme. In fact, the localized melting created by –1/–2, –2/–3, –3/–4 and –4/–5 substitutions appears strong enough to sequester the entire enzyme population and induce cleavage 3–4 bp downstream (Fig. 6, lanes a–d). Correct and exclusive PPT cleavage by mutant RT on substrate carrying a –3/–4 or –4/–5 substitution (Fig. 6, lanes c and d) effectively shows that the structural consequences of this RNase H primer grip mutation can be complemented by locally changing the stability of the substrate. Impaired PPT hydrolysis when positions –5 and –6 of the (rG):(dC) tract were altered and its restoration when the site of substitution was relocated by 1 bp into the central (rA):(dT) tract are again evident, although we observed only low-level activity with the latter substrate (compare Fig. 6, lanes e and f). Thereafter, PPT processing was virtually eliminated, but was accompanied by restoration of cleavage at the non-PPT sites.

Discussion

This study was predicated on observations that (i) the HIV-1 PPT is accurately selected for cleavage when embedded within a considerably larger RNA/DNA hybrid (23) and (ii) in the absence of RT, two regions of the hybrid deviating from standard Watson–Crick geometry are evident from chemical footprinting (6). Together, these observations implicate structural features of the PPT-containing duplex as active participants in its recruitment of, and processing by, HIV-1 RT. Because the PPT comprises three homopolymeric tracts and is preceded by a (rU):(dA) tract whose alteration affects virus replication (24, 25), the geometry adopted both within these tracts and at their junctions may be pivotal to the accuracy with which the plus-strand primer is processed. Lastly, a contribution of the p66 RNase H primer grip (5) to the accuracy of PPT selection has been suggested from our recent mutagenesis studies (19).

To better understand this interplay between the protein and nucleic acid components, a strategy that alters local nucleic acid stability stepwise throughout the PPT while minimizing steric clashes and preserving the sequence context of the RNA primer would be advantageous. We report here the use of non-hydrogen-bonding thymine and cytosine isosteres (dF and dD, respectively) to study HIV-1 PPT interactions via their single and dual placement throughout the (–) DNA template. Local “melting” induced by dF and dD incorporation is shown here either to inhibit PPT processing when placed at different positions of the RNA/DNA hybrid or to sequester a structural motif of the RNase H domain, thereby relocating the active site to 3–4 bp downstream. As discussed below, we believe these combined observations provide a plausible hypothesis for the mechanism by which the wild-type PPT is recognized.

Fig. 4 presents a model positioning HIV-1 RT on the PPT-containing duplex in a mode compatible with cleavage at the PPT-U3 junction. As a point of reference, His-539 of the RNase H active site is positioned at the scissile bond, and other RT domains are disposed throughout the hybrid as indicated by Sarafianos et al. (5). In our model, the RNase H primer grip spans ≈3 bp on either side of the (rG):(dC)–(rA):(dT) junction, whereas residues of the minor groove binding tract of the p66 thumb span the junction between the upstream (rA):(dT) and abutting (rU):(dA) tracts. The “unzipped” portion of the PPT is positioned between these protein motifs.

The (rU):(dA)-(rA):(dT) junction and adjacent (rA):(dT) tract correlate well with the portion of the PPT-containing duplex that would be contacted by the minor groove binding tract of the p66 thumb (26), a motif implicated in correct tracking of the enzyme over nucleic acid. Previous mutagenesis studies (27) have suggested that this motif functions as a “sensor” of duplex configuration, detecting base pair alterations introduced by lesions. Localized destabilization (enhanced melting) of this PPT region could therefore compromise thumb contacts. Moving further into the PPT, the “unzipped” portion (5) would be positioned between the DNA polymerase and RNase H catalytic centers of RT, corresponding to the region that assumes a 40° bend during the obligatory transition from A- to B-form geometry determined for all nucleic acid-containing RT structures (5, 7, 8). We might therefore predict that further destabilizing this region would impair this transition in nucleic acid geometry, as supported by our observation that dual dT → dF substitutions from positions –9 to –15 severely affect PPT cleavage efficiency (Figs. 3 and 6).

The (rG):(dC)–(rA):(dT) junction also appears to be extremely sensitive to local perturbations in nucleic acid geometry, because a dual –5/–6 substitution inhibited cleavage by wild-type RT (Fig. 3, lane f) and a single substitution at either position is sufficient to affect the primer grip mutant (Fig. 5, lanes f and g). Surprisingly, relocating the position of nucleic acid destabilization 1 bp into the central (rA):(dT) tract restored cleavage, indicating that one side of the junction is particularly prone to destabilization. Alternatively, we must also consider that the junction between these two tracts may induce an irregular geometry, similar to that that observed by Dickerson and colleagues (28, 29) with a (dG):(dC)–(dA):(dT) junction. This structural anomaly might then be recognized by the RNase H primer grip, which is positioned exactly over the (rG):(dC)–(rA):(dT) junction in Fig. 4. Although the crystal structure of Sarafianos et al. (5) does not completely resolve the G:C-containing portion of the PPT, these authors have also proposed that unusual features of this junction might affect nucleic acid trajectory and guiding of the RNA strand into the RNase H active site. In addition, the NMR structure of a DNA version of the HIV-1 PPT demonstrates broadening of resonances at or near the (dG:dC)-(dA:dT) junction, which these authors (30) attributed to structural stress possibly caused by docking of the two tracts.

If a structural irregularity in the PPT (rG):(dC) tract interacts with the p66 RNase H primer grip as a means of promoting accurate cleavage, how would this RT motif function? Here, the use of non-hydrogen-bonding pyrimidine mimics is particularly informative, because these agents are capable of locally destabilizing nucleic acid helices. A proposed role for the RNase H primer grip is, by interacting with the DNA strand, to impose the appropriate trajectory of its RNA complement into the RNase H catalytic site. Artificially separating strands of the RNA/DNA hybrid, as we have achieved here with dC → dD substitutions, appears capable of facilitating this step, promoting a critical protein–nucleic acid interaction that relocates the RNase H catalytic center 3–4 bp from the site of substitution. In the context of the wild-type element, this model infers an unusual geometry ≈4 bp upstream of the PPT-U3 junction. The combined data with wild-type and mutant RTs lend support to this notion, because dC → dD substitutions at positions –5 and –6 are detrimental to PPT function. However, loss of PPT processing accompanying these substitutions, rather than continued repositioning of the RNase H catalytic center, suggests that localized destabilization induced by a –5/–6 dC → dD substitution is incompatible with hybrid geometry between the (rG):(dC) and central (rA):(dT) tract, affecting either bending at their junction or the rigidity known to be imposed by A tracts (5).

Finally, although we have reported that altering Tyr-501 of the RNase H primer grip affects PPT cleavage specificity, the structural basis underlying this observation remains to be resolved (19). p66^Y501A/p51 cleaves between the GA ribonucleotide pair 5 nt downstream of the PPT-U3 junction, consistent with cleavage preferences of murine leukemia virus (MLV) RT-associated RNase H noted by Rattray and Champoux (31). It is intriguing that, to a lesser extent, single dC → dG substitutions at positions –2 to –5 in the PPT-containing hybrid also direct wild-type RT to cleave at this location (Fig. 1, lanes c–f), suggesting that position +5 may represent a favored secondary cleavage site of some importance in PPT selection. Such a notion is supported by recent reports indicating that, in Moloney MLV, cleavage within U3 near the PPT-U3 junction may facilitate initiation of plus-strand DNA synthesis (32).

Although speculative, the data of this communication suggest that Tyr-501 may participate in stacking interactions as a means of stabilizing region(s) within a hybrid substrate that must be locally melted or distorted for the RNA strand to gain access to the catalytic center. Although the mechanism would be different, this type of interaction would resemble stabilization of flipped bases by DNA methyltransferases (33) or glycosidases (34). Lowering the energy barrier for this process via the introduction of non-hydrogen-bonding bases into the PPT (–) DNA template would effectively compensate for the function contributed by Tyr-501, as observed here experimentally.