Investigating HIV-1 polypurine tract geometry via targeted insertion of abasic lesions in the (-)-DNA template and (+)-RNA primer.

A variety of biochemical and structural studies indicate that two regions of the human immunodeficiency virus type 1 (HIV-1) polypurine tract (PPT)-containing RNA/DNA hybrid deviate from standard Watson-Crick geometry. However, it is unclear whether and how these regions cooperate to ensure PPT primer selection by reverse transcriptase-associated ribonuclease H and subsequent removal from nascent (+)-DNA. To address these issues, we synthesized oligonucleotides containing abasic lesions in either the PPT (+)-RNA primer or (-)-DNA template to locally remove nucleobases, although retaining the sugar-phosphate backbone. KMnO(4) footprinting indicates such lesions locally alter duplex structure, whereas thermal melting studies show significantly reduced stability when lesions are positioned around the scissile bond. Substituting the (-)-DNA template between positions -15 and -13 altered cleavage specificity, whereas equivalent substitutions of the (+)-RNA had almost no effect. The unpaired base of the DNA template observed crystallographically (-11C) could also be removed without significant loss of cleavage specificity. With respect to the scissile -1/+1 phosphodiester bond, template nucleobases could be removed without loss of cleavage specificity, whereas equivalent lesions in the RNA primer were inhibitory. Our data suggest an interaction between the p66 thumb subdomain of HIV-1 reverse transcriptase, and the DNA template in the "unzipped" portion of the RNA/DNA hybrid could aid in positioning the ribonuclease H catalytic center at the PPT/U3 junction and also provides insights into nucleic acid geometry around the scissile bond required for hydrolysis.

In retroviruses and long terminal repeat (LTR) 1 -containing retrotransposons, second or (ϩ)-strand synthesis requires (a) specific cleavage between the 3Ј end of the (ϩ)-strand, polypurine tract (PPT), and the 3Ј unique sequence (U3) of the LTR; (b) DNA-dependent DNA synthesis from the newly created primer; and (c) removal of the PPT primer from nascent (ϩ)-DNA (1) (see Fig. 1A). Each of these steps can be accurately recapitulated in vitro when PPT sequences are embedded within a larger RNA/DNA hybrid (2)(3)(4)(5)(6), i.e. where the nucleic acid termini cannot influence enzyme orientation, implicating structural features of the PPT in both resistance to internal reverse transcriptase-associated ribonuclease H (RT/RNaseH) cleavage and specific processing at its 3Ј terminus from (ϩ)-RNA and (ϩ)-DNA. Early NMR studies, using a short fragment of this hybrid, identified a 15 o bend at the HIV-1 PPT/U3 junction (7), which may contribute to the accuracy of cleavage. Subsequent crystallographic studies with HIV-1 reverse transcriptase (RT) bound to a PPT-containing hybrid indicated a pattern of weakened base pairing centered ϳ13 bp upstream of the PPT/U3 junction (8), a notion supported by chemical footprinting of the duplex in the absence of RT (9). Interestingly, the distance between the PPT/U3 junction and the upstream region of weakened base pairing is close to the spatial separation between the p66 thumb subdomain and RNase H catalytic center of HIV-1 RT (8,10,11). These combined observations suggested that an induced fit between the HIV-1 RT thumb subdomain and the upstream portion of the PPT might position the RNase H catalytic center over the PPT/U3 junction. As a consequence, this would render intervening regions inaccessible and RNase H-resistant. Implicating regions upstream of the PPT/U3 junction in enzyme positioning (9,12) is supported by our studies of PPT recognition in the LTR-containing retrotransposon Ty3. Although the Ty3 PPT lacks the hallmark rA⅐dT and rC⅐dG homopolymeric stretches characteristic of retroviral (ϩ)-strand primers (13), altering nucleic acid geometry 10 -12 bp upstream of the PPT/U3 junction also affects the precision of PPT cleavage (14). 2 One approach to studying PPT cleavage specificity is by introducing structural changes in a manner preserving sequence and spatial context. As an example, we substituted the thymine analog 2,4-difluoro-5-methylbenzene and cytosine analog 2-fluoro-4-methylbenzene throughout the HIV-1 PPT (Ϫ)-DNA template to remove hydrogen bonding and locally increase flexibility (12,15). This approach indicated that the RNase H catalytic center could be relocated ϳ4 bp 3Ј of the site of analog insertion, indirectly implicating the "RNase H primer grip" of the p66 subunit (8) in positioning the scissile bond in the RNase H-active center. A similar approach with the Ty3 PPT relocated its RNase H catalytic center ϳ12 bp downstream of the site of analog insertion (14), suggesting that alternative structural elements of the retrotransposon enzyme may participate in PPT recognition. The converse strategy, i.e. decreasing local flexibility, was evaluated by introducing locked nucleic acid analogs (16 -18) into the HIV-1 PPT (Ϫ)-DNA template, indicating that regions at both the 5Ј and 3Ј end of the RNA/ DNA hybrid are critical for correct enzyme processing. This study (6) also supports a finding of Schultz et al. (19), suggesting two PPT cleavage modes, namely the PPT/U3 junction and ϳ5 bp into the U3 region, the latter of which might contribute to efficient (ϩ)-strand synthesis. Related studies evaluating the interaction of protein subdomains with nucleic acid exiting the DNA polymerase catalytic center (20 -22) illustrate the value of nucleoside analogs in dissecting protein-nucleic acid interactions.
The "unzipped" portion of the HIV-1 PPT reported by Sarafianos et al. (8) was of particular interest to us, because template base Ϫ11 and primer base Ϫ13 (defining position Ϫ1 as the base pair 5Ј to the PPT/U3 junction, Fig. 1B) are unpaired, thereby enhancing the flexibility of this region. At the same time, we and others have noted that template base ϩ1 of the HIV-1 and murine leukemia virus PPT is surprisingly tolerant to substitution, accepting substitution with non-hydrogenbonding pyrimidine isosteres and base mismatches with minimal effect on the accuracy and overall rate of hydrolysis (2,9,13). The latter observations might indicate that, as the RNA/ DNA hybrid enters the RNase H catalytic center, the DNA strand is displaced, allowing "docking" of the scissile phosphodiester bond in the active site, a model consistent with the role proposed for the p66 RNase H primer grip (8). A detailed understanding of how duplex geometry influences the accuracy of PPT cleavage is important in understanding this critical step in the reverse transcription cycle.
In this communication, we examined the effect of introducing abasic lesions (23) into the unzipped region of both the HIV-1 (Ϫ)-DNA template and (ϩ)-RNA primer between positions Ϫ15 and Ϫ11, as well as around the PPT/U3 cleavage junction. Although there is little information on abasic lesions within RNA/DNA hybrids, structural studies with duplex DNA suggest that local elimination of the base does not affect the sugarphosphate backbone (24,25) but, in general, increases flexibility (25). An abasic site can also affect whether its unpaired complement assumes an intra-or extrahelical configuration. Collectively, our data suggest positions Ϫ15, Ϫ14, and Ϫ13 of the HIV-1 PPT (Ϫ)-DNA template are important modulators of cleavage specificity, either through directly contacting the thumb subdomain or imparting local flexibility to nucleic acid in its vicinity. The previously reported (8) unpaired template base (Ϫ11C) could also be removed without affecting PPT processing. In contrast, abasic lesions in the (ϩ)-RNA primer between positions Ϫ15 and Ϫ11 are tolerated with very little alteration in cleavage specificity. With respect to structural requirements at the PPT/U3 junction, we show here that, although eliminating primer bases 5Ј and 3Ј of the scissile bond destroys cleavage, removing their complement on the (Ϫ)-DNA template is only partially inhibitory, indicating that, consistent with studies on human RNase H1 (26), a DNA base at the scissile bond is not essential for hydrolysis.

EXPERIMENTAL PROCEDURES
Oligonucleotide and Enzyme Preparation-Substituted 40-nt oligodeoxynucleotides and 30-nt oligoribonucleotides were synthesized at a 1 mol scale on a PE Biosystems Expedite 8909 oligonucleotide synthesizer by standard phosphoramidite chemistry and deprotected according to the manufacturer's specifications. r-Spacer (abasic RNA lesions) and d-Spacer (abasic DNA lesions) phosphoramidites were purchased from Glen Research, Sterling, VA. Stepwise coupling yields for incorporation of each analog was Ͼ98%, determined by trityl cation monitoring. Oligonucleotides were purified by preparative polyacrylamide gel electrophoresis and quantified spectrophotometrically (260 nm), assuming a molar extinction coefficient equal to the sum of the constituent bases. A wild type 30-nt RNA containing the HIV-1 3Ј PPT and flanking sequences at its 5Ј and 3Ј termini was purchased from Dharmacon Research (Boulder, CO). DNA/RNA hybrids were prepared by heating equimolar amounts of RNA and DNA to 95°C in 10 mM Tris-HCl, pH 7.8, 25 mM NaCl for 5 mins, followed by slow cooling to 4°C. The samples were stored at Ϫ20°C. p66/p51 HIV-1 RT was purified according to published procedures (27).
Thermal Melting Profiles-Equimolar amounts of 30-nt RNA primer and 40-nt, (Ϫ)-DNA templates were annealed by heating to 90°C and slow cooling in degassed 10 mM Na 2 HPO 4 /NaH 2 PO 4 , pH 7.0, 80 mM NaCl. Non-denaturing gel electrophoresis was used to determine that complete hybridization had been achieved. For measurement of melting temperatures (T m ), 10 g/ml solutions of each substrate were analyzed in a Beckman DU 640 spectrophotometer. E 260 was measured at 0.2°C intervals from 30 to 85°C. The T m of each hybrid was calculated by the "first derivative" method described by the manufacturer.
KMnO 4 Footprinting of PPT Variants-RNA/DNA hybrids containing abasic lesions in the template or primer were incubated at room temperature for 5 min in 20 mM Tris-HCl, pH 8.0, 100 mM NaCl, and 100 M MgCl 2 . The total reaction volume was 20 l. Reactions were initiated by adding 2 l of freshly prepared 25 mM KMnO 4 solution and terminated after 30 s with 2 l of 14 M ␤-mercaptoethanol. After ethanol precipitation and dessication, the samples were treated with 100 l of 1 M piperidine for 30 min at 90°C. Piperidine was removed by vacuum desiccation. Nucleic acids were washed three times with 50 l of water and vacuum-dried after each resuspension. Samples were finally resuspended in 89 mM Tris borate, pH 8.3, 2 mM EDTA, and 95% formamide containing 0.1% bromphenol blue and xylene cyanol and analyzed by electrophoresis through 15% denaturing polyacrylamide gels.
RNase H-mediated PPT Processing-Specific hydrolysis at the PPT 3Ј terminus was evaluated as previously described (6) using p66/p51 HIV-1 RT and 30-nt RNA/40-nt DNA hybrids containing abasic lesions. 5Ј end labeling of the PPT-containing RNA was performed with T4 polynucleotide kinase and [␥-32 P]ATP. PPT cleavage was initiated by adding HIV-1 RT to RNA/DNA hybrids in 10 mM Tris-HCl (pH 8.0), 80 mM NaCl, 5 mM dithiothreitol, and 6 mM MgCl 2 at 37°C, with enzyme and RNA/DNA hybrid present at final concentrations of 50 and 200 nM, respectively. Hydrolysis was terminated at the times indicated in the Fig. 3 legend by adding an equal volume of 95% (v/v) formamide containing 0.1% (w/v) bromphenol blue and xylene cyanol, and the products fractionated by high voltage electrophoresis through denaturing 15% (w/v) polyacrylamide/7 M urea gels, visualized by autoradiography, and quantified following phosphorimaging.

RESULTS
Thermal Stability of Abasic PPT Duplexes-A portion of the HIV-1 PPT-containing hybrid used in our studies is illustrated in Fig. 1B, highlighting anomalous base pairing observed in the RT-RNA/DNA co-crystal between positions Ϫ15 and Ϫ9 (defining Ϫ1 as the first base pair 5Ј to the scissile phosphate) (8). For the present study, abasic lesions were introduced into the (Ϫ)-DNA template and (ϩ)-RNA primer between positions Ϫ15 and Ϫ11. The second region selected for analysis was the Ϫ1/ϩ1 scissile bond corresponding to the PPT/U3 junction, where lesions were introduced on either side of the junction and downstream at position ϩ2. To determine how these substitutions affected duplex stability, the melting temperature (T m ) of each hybrid was determined, the results of which are presented in Table I.
For PPT variants containing substitutions between positions Ϫ15 and Ϫ11, most melted at temperatures similar to that of the wild type hybrid (67.1°C), regardless of whether the nucleobase was eliminated from the (Ϫ)-DNA template or (ϩ)-RNA primer. The maximum decrease in T m was observed with duplexes containing the unpaired bases defined by Sarafianos et al. (8) i.e. template base Ϫ11 (T m ϭ 65.6°C) and primer base Ϫ13 (T m ϭ 65.0°C). From these results, we concluded that the stability of the unzipped portion of the HIV-1 PPT was not seriously affected by base elimination. In contrast, removing nucleobases around the PPT/U3 cleavage junction was significantly more destabilizing, depressing the T m by as much as 12.3°C (Ϫ1RAb). The destabilizing effect was evident as far as position ϩ2, suggesting that abasic lesions inserted at the PPT/U3 junction influenced neighboring base pairs, possibly through the alteration of stacking interactions. For a duplex of the length used in our studies (30 bp), a 10°C reduction in T m can be likened to introducing three base mismatches. Thus, if duplex geometry at the unsubstituted PPT/U3 junction is altered, as suggested by NMR (7) and chemical footprinting studies (9), this appears to be further destabilized as a consequence of nucleobase elimination.
Susceptibility of Template Thymines to KMnO 4 Oxidation-To assess how targeted nucleobase elimination altered PPT structure, susceptibility of template thymines between positions Ϫ15 and ϩ1 to KMnO 4 oxidation was investigated (see Fig. 1B). This strategy can determine whether thymines are unpaired (28,29) or are structurally distorted but exhibit weak hydrogen bonding (30). Template thymine reactivity in response to abasic insertions in both strands of the RNA/DNA hybrid was determined. Fig. 2A presents thymine sensitivity when template nucleobases between positions Ϫ15 and Ϫ12 were removed. Clearly, removing any thymine will eliminate that particular hydrolysis product. Eliminating nucleobases Ϫ15, Ϫ14, and Ϫ13 slightly increased reactivity of adjacent thymines ( Fig. 2A, lanes 1-3), which in conjunction with the T m data of Table I, suggests local alteration in base stacking rather than extensive disruption of the duplex. Eliminating template base Ϫ12T had little effect on thymine reactivity at positions Ϫ13, Ϫ14, and Ϫ15 ( Fig. 2A,  lane 4). In contrast, Ϫ12T was highly susceptible to oxidation following removal of the unpaired nucleobase, Ϫ11C ( Fig. 2A, lane 5), suggesting disruption of the original Ϫ12T/Ϫ11rG mispair suggested by Sarafianos et al. (8) (Fig. 1B) when the stacking environment of the template is altered. The notion that eliminating base Ϫ11C is more destabilizing is supported by the observation that this substitution also affects Ϫ10T reactivity. With this exception, template substitutions between position Ϫ15 and Ϫ12 did not alter KMnO 4 sensitivity within the adjacent r(A) 4 ⅐d(T) 4 tract between Ϫ10T and Ϫ7T.
Reactivity of template nucleobase ϩ1T in response to introducing abasic lesions at positions Ϫ1, ϩ1, and ϩ2 is shown in Fig. 2B. As expected, no product is evident for substrate ϩ1DAb (Fig. 2B, lane 8). Also, altered migration of the ϩ1 hydrolysis product on substrate ϩ2DAb (Fig. 2B, lane 9) is accounted for by the fact that the DNA template was 5Ј endlabeled, thus the cleavage product lacks one base. Substitutions Ϫ1DAb and ϩ2DAb resulted in enhanced ϩ1T reactivity (Fig. 2B, lanes 7 and 9, respectively). Table I indicates that the T m for substrates containing lesions at these positions was reduced by ϳ10°C, which is equivalent to loss of hydrogen bonding over three base pairs. The data of Fig. 2B and Table I thus indicate that abasic template lesions are more destabilizing when positioned at the PPT/U3 junction. Fig. 2, C and D, indicate template thymine reactivity in response to removing nucleobases from the (ϩ)-RNA primer. In general, substitutions Ϫ15RAb to Ϫ11RAb increased KMnO 4 sensitivity between bases Ϫ15T and Ϫ12T, the effect being more pronounced at thymines opposite the site of nucleobase removal (Fig. 2C, lanes 1-5). In an analogous manner to its template counterpart, primer substitution Ϫ11RAb increased Ϫ12T and Ϫ10T reactivity (Fig. 2C, lane 5), suggesting local disruption of the RNA/DNA hybrid. As might be expected, removing primer nucleobase ϩ1 enhanced KMnO 4 sensitivity of template thymine ϩ1 (Fig. 2D, lane 8). Enhanced ϩ1T reactivity was also evident following removal of primer nucleobase Ϫ1 (Fig. 2D, lane 7). Finally, ϩ1T reactivity of substrate ϩ2RAb was similar to that of the unsubstituted duplex (Fig.  2D, lanes 9 and 6, respectively), suggesting that stacking of template nucleobases at the scissile bond was unaffected by removing primer nucleobase ϩ2.
Removing Template Nucleobases between Positions Ϫ15 and Ϫ11 Affects PPT Cleavage-Cleavage at the PPT/U3 junction in response to eliminating template nucleobases between positions Ϫ15 and Ϫ11 is shown qualitatively in Fig. 3A and quantitatively in Fig. 3B (for the former, lane d of each hydrolysis profile was selected). Using a 5Ј end-labeled RNA primer, the wild type RNA/DNA hybrid (Fig. 3, A and B, panels i) was hydrolyzed predominantly at the Ϫ1/ϩ1 junction and, to a lesser extent, between positions Ϫ2 and ϩ6. No template substitution gave rise to aberrant cleavage within the PPT itself (data not shown). Removing template nucleobase Ϫ15 induced relaxed cleavage specificity, with the consequence that specific cleavage at the PPT/U3 junction was reduced (Fig. 3, A and B, panels ii). This effect was even more pronounced with substrate Ϫ14DAb, where the PPT/U3 junction was the least-preferred cleavage site (Fig. 3, A and B, panels iii). Although substrate Ϫ13DAb was hydrolyzed with improved specificity at the Ϫ1/ϩ1 junction, positions ϩ5 and ϩ6 were still the favored sites (Fig. 3, A and B, panels iv). The data of Fig. 3, A and B, panels ii-iv, suggest alternative scenarios, namely that (a) specific contacts between the DNA strand between positions Ϫ15 and Ϫ13 and a structural motif of HIV-1 RT have been disturbed or (b) enhanced duplex flexibility upon nucleobase removal is incompatible with its trajectory between the catalytic centers of HIV-1 RT. Each of these possibilities will be discussed later.
In contrast, we observed enhanced specificity for cleavage at the PPT/U3 junction with substrates Ϫ12DAb and Ϫ11DAb (Fig. 3, A and B, panels v and vi, respectively), which are mispaired and unpaired, respectively, in the HIV-1 RT/ PPT structure described by Sarafianos et al. (8). Interestingly, a Ϫ11DAb substitution increased reactivity of template nucleobase Ϫ12T to KMnO 4 oxidation ( Fig. 2A, lane 5) to a level equal to or exceeding that of any template thymine rendered unpaired by removal of its primer counterpart, suggesting that Ϫ12T is selectively unpaired when template nucleobase Ϫ11 is removed. In all structures of HIV-1 RT containing nucleic acid (8,10,11), the duplex undergoes an A-to B-form transition downstream of the DNA polymerase catalytic center, an event accompanied by a 40 o bend centered ϳ11 bp from the RNase H catalytic center. Locating the RNase H catalytic center over the PPT/U3 junction, as depicted in the model of Fig. 1B, would center this bend around position Ϫ11. Thus, local weakening of duplex architecture via targeted nucleobase removal (i.e. substrates Ϫ12DAb and Ϫ11DAb) might lower the energy required to achieve this bend, providing an improved "fit" between enzyme and substrate and, as a consequence, greater cleavage specificity.
Eliminating Primer Nucleobases between Positions Ϫ15 and Ϫ11-Crystallography of the upstream portion of the HIV-1 PPT RNA/DNA hybrid has revealed an unusual stacking pattern in the RNA strand at the sequence 5Ј-A-G-A-3Ј (31), which these authors have suggested contributes to deformation of the duplex between positions Ϫ15 and Ϫ9 in the RT-RNA/DNA co-crystal (8). To examine whether altering the stacking environment of primer bases influences selection at the PPT/U3 junction, as well as to investigate a need for the unpaired ribonucleotide Ϫ13A (Fig. 1B), abasic lesions were likewise introduced into the RNA primer between positions Ϫ15 and Ϫ11. The results of this analysis are presented in Fig. 4.
Unlike their DNA complement, primer nucleobases Ϫ15, Ϫ14, and Ϫ13 could be eliminated with only a marginal impact on PPT cleavage specificity (Fig. 4, A and B, panels ii-iv). If nucleobase removal simply weakened duplex architecture, this would predict that lesions in the same position of the DNA template and RNA primer would have the equivalent effect on processing. However, the ability to alter PPT cleavage when template nucleobases Ϫ15, Ϫ14, and Ϫ13 are Note that for substitutions at position ϩ1 and between positions Ϫ11 and Ϫ15, the substituted base is absent from the chemical cleavage profile. Also, altered migration of the ϩ1 product in lane 9 is due to the fact that the DNA template is 5Ј endlabeled, thus the cleavage product contains the abasic site. Sensitivity of template thymines to KMnO 4 oxidation in response to abasic primer lesions at the 5Ј end of the PPT (C) and around the PPT/U3 junction (D). removed, although this is unaffected by removing the equivalent primer nucleobases, suggests that the former are likely involved in specific protein/nucleic acid contacts. Furthermore, if the stacking pattern at the sequence 5Ј-A-G-A-3Ј described by Kopka et al. (31) contributed to duplex deformation, we had anticipated that PPT selection might be influenced by eliminating primer nucleobases positions Ϫ12 and Ϫ11. In contrast, we observed only a 2-fold decrease in cleavage at the PPT/U3 junction on substrates Ϫ12RAb (Fig.  4, A and B, panels v) and Ϫ11RAb (Fig. 4, A and B, panels vi). At the same time, none of these primer substitutions resulted in enhanced cleavage within the PPT (data not shown), suggesting the 5Ј-A-G-A-3Ј sequence does not make a major contribution to PPT deformation and recognition by HIV-1 RT.
Targeted Removal of Nucleobases at the PPT/U3 Junction-Studies with substrates containing mispaired bases (9) or nonhydrogen-bonding pyrimidine isosteres (12) in the (Ϫ)-DNA template have demonstrated that alterations in nucleic acid geometry are surprisingly well tolerated at the HIV-1 PPT/U3 junction without major implications for cleavage specificity. Relocating the RNase H catalytic center of HIV-1 RT over the PPT/U3 junction, as indicated in Fig. 2, also suggests few, if any, contacts to the (Ϫ)-DNA template opposite the scissile bond. We therefore assessed whether retaining only the sugarphosphate backbone of the DNA template and RNA primer around the scissile bond was compatible with hydrolysis by introducing abasic lesions at positions Ϫ1, ϩ1, and ϩ2 of each strand.
PPT selection in response to eliminating template nucleobases is presented in Fig. 5. Although the rate of cleavage at the PPT/U3 junction was reduced, the hydrolysis profiles derived from substrates Ϫ1DAb and ϩ1DAb indicated that template nucleobases were not absolutely necessary 5Ј and 3Ј of the scissile bond of the PPT primer (Fig. 5, A and B, panels ii and iii, respectively); this experiment was performed in parallel with abasic template lesions between Ϫ15 and Ϫ11, thus the same control digest was included for reference. For these two substrates, minor changes in cleavage specificity ahead of the PPT/U3 junction were evident, e.g. hydrolysis at position ϩ1 was reduced on substrate Ϫ1DAb, and although this is restored on substrate ϩ1DAb, the latter had very little ϩ2 hydrolysis product. Significant cleavage of substrate ϩ1DAb at position ϩ5 is also evident. Finally, although correct cleavage is observed on substrate ϩ2DAb, the major hydrolysis product now corresponds to hydrolysis between positions ϩ1 and ϩ2 (Fig. 5,  A and B, panels iv).
Analysis of hybrids containing lesions in the RNA primer between positions Ϫ1 and ϩ2 is shown in Fig. 6. In this case, although hydrolysis at positions ϩ5 and ϩ6 is unaffected, removing primer nucleobase Ϫ1G resulted in loss of cleavage at the PPT/U3 junction (substrate Ϫ1RAb; Fig. 6, A and B, panels ii). Note here that the Ϫ1 and Ϫ2 hydrolysis products migrate together, because the former lacks the nucleobase. Removing primer nucleobase ϩ1A (substrate ϩ1RAb) likewise eliminates Ϫ1/ϩ1 cleavage, although permitting cleavage at ϩ5 and ϩ6 (Fig. 6, A and B, panels iii). In this case, the PPT/U3 hydrolysis product does not contain an abasic lesion and thus migrates with the expected mobility. The data of Fig. 6, A and B, panels ii and iii, therefore indicate that, although the sugarphosphate backbone corresponding to the scissile Ϫ1/ϩ1 phosphodiester bond of the PPT primer is preserved in the RNase H active center, critical contacts with residues of the active site are affected when the nucleobase is removed. Potential sites of contact affected will be discussed in the following section. Finally, Fig. 6, A and B, panels iv, indicate normal cleavage at the PPT/U3 junction when nucleobase ϩ2C was removed. The profiles of substrates ϩ2RAb and ϩ2DAb are particularly interesting, because the latter induced elevated cleavage at the ϩ1/ϩ2 junction (Fig. 5, A and B, panels iv) and increased KMnO 4 sensitivity of template thymine ϩ1, whereas the RNase H hydrolysis profile and KMnO 4 reactivity of ϩ1T of substrate ϩ2RAb are similar to the unsubstituted hybrid.  Fig. 3. B, quantification of hydrolysis data, determined as in the legend to Fig. 3.

DISCUSSION
Examining how nucleic acid geometry contributes to recognition and cleavage of the HIV-1 PPT to provide the (ϩ)-strand primer has been addressed in this communication via targeted insertion of abasic lesions, which eliminate the nucleobase, although leaving the sugar-phosphate backbone intact. In the co-crystal of HIV-1 RT and a PPT-containing RNA/DNA hybrid (8), the RNase H active site was located several base pairs upstream of the PPT/U3 junction. Thus, interpretation of our results has required constructing a model that relocates protein/nucleic acid contacts to approximate this catalytic center positioned for hydrolysis at the scissile Ϫ1/ϩ1 phosphodiester bond (Fig. 7). When the highly conserved His-539 of the RNase H active center is placed at this junction, extrapolating from co-crystals of RT with duplex DNA (10,11) and RNA/DNA (8) suggests helix ␣-H of the p66 thumb subdomain would make multiple contacts with the sugar-phosphate backbone of the (Ϫ)-DNA template between positions Ϫ16 and Ϫ13, most likely involving residues Trp-266, Lys-263, Gly-262, Lys-259, Gln-258, and Asn-255. At the same time, helix ␣-I would be predicted to contact the (ϩ)-RNA primer at positions Ϫ12 and Ϫ10 via Ser-280, Ala-284, and Gly-285 and Thr-286. This helix-turn-helix motif of the p66 thumb could therefore be envisaged as asymmetrically "grasping" the destabilized region of the RNA/DNA hybrid between positions Ϫ15 and Ϫ9. In addition, at the PPT/U3 junction, the model of Fig. 7 implicates p66 residues Arg-448, Asn-474, Gln-475, Gln-500, and His-539 in base, sugar, and phosphate contacts to (ϩ)-RNA primer between positions Ϫ2 and ϩ2. Using this model as a reference, the consequences of nucleobase removal at both regions of the RNA/DNA hybrid will be discussed. At the same time, it is important to take into consideration a 40 o bend the hybrid will adopt over ϳ5 bp (between positions Ϫ13 to Ϫ9) as it undergoes a transition from A-to B-form geometry (8,10,11).
With respect to template substitutions Ϫ15DAb, Ϫ14DAb, and Ϫ13DAb, nucleobase removal will alter intrahelical stacking, which Fig. 7 suggests could affect multiple contacts with p66 helix ␣-H. Experimentally, we observed that specific cleavage at the PPT/U3 junction diminishes with these three mutant substrates, and is accompanied by an increase in hydrolysis between positions ϩ1 and ϩ6, the latter observation suggesting relaxed specificity. The model of Fig. 7 also proposes fewer protein contacts with both the sugar-phosphate backbone and template nucleobases Ϫ12 and Ϫ11; in keeping with this notion, substrates Ϫ12DAb and Ϫ11DAb exhibit specific, and possibly enhanced, cleavage at the PPT/U3 junction. In addition to fewer protein contacts, the 40 o bend the hybrid adopts in the presence of HIV-1 RT will be centered close to positions Ϫ12/-11. Removing these nucleobases might therefore be seen as a means of aiding the A-to B-form transition. In this respect, it is interesting to note that removing template nucleobase Ϫ12 has the most significant consequence for PPT architecture (Fig.  2A, lane 5) but is less deleterious with respect to PPT processing (Fig. 3, A and B, panels v).
With regard to the PPT (ϩ)-RNA primer, nucleobase removal between positions Ϫ15 and Ϫ11 has very little affect on the specificity of PPT cleavage. Although a slight reduction in PPT/U3 cleavage was observed on substrate Ϫ13RAb, this was still the primary site of hydrolysis. If removing nucleobases between positions Ϫ15 and Ϫ13 nonspecifically destabilized the RNA/DNA hybrid, we might have expected the same result for PPT/U3 hydrolysis regardless of whether the nucleobase was removed from the template or primer. The sensitivity of PPT processing to altering specifically the (Ϫ)-DNA template thus supports our contention that substitutions Ϫ15DAb, Ϫ14DAb, and Ϫ13DAb alter important protein/nucleic acid contacts with the p66 thumb.
Positioning the catalytic His-539 of the HIV-1 RNase H domain at the scissile Ϫ1/ϩ1 phosphodiester bond allows contacts within the active site to be more accurately defined and compared with a model proposed for Escherichia coli RNase H. NMR analysis (32) and studies with nucleoside analogs (33) suggest Gln-72 of E. coli RNase H is hydrogenbonded to the 2Ј-OH group of the nucleoside two bases 5Ј to the scissile bond. In our model, the HIV-1 RT counterpart, Gln-500, could fulfill this role, contacting ribose of primer base Ϫ2G. Additionally, Cys-13 of E. coli RNase H has been implicated in an interaction with the 2Ј-OH of the ribonucleotide at position ϩ1. For HIV-1 RT, Arg-448 could perform the same function. Building upon these reference points, Fig.  7 predicts Gln-475 of HIV-1 RT contacts nucleobase Ϫ2G and the ribose of Ϫ1G, whereas Asn-474 contacts the phosphate backbone of the Ϫ1/ϩ1 phosphodiester bond, and Arg-448 would make additional contact with the ribose of base ϩ2C. Surprisingly, although the RNase H primer grip of HIV-1 RT (8) contacts the DNA template several bases upstream, few (if any) template contacts are made opposite the scissile bond. Because previous studies have indicated that base pairing at position ϩ1 is not absolutely required for hydrolysis (12), the necessity of template bases opposite the scissile bond was investigated here. Although we observed reduced hydrolysis, Fig. 7 indicates template nucleobases Ϫ1 and ϩ1 can, in fact, be removed. Thus, once the RNA primer is correctly positioned in the RNase H active site through the contacts suggested above, transient opening of the duplex at positions Ϫ1 and ϩ1 may occur to facilitate hydrolysis, which is formally analogous to a model forwarded by Nakamura et al. (32). Although the preference for cleavage at position ϩ1 following removal of template nucleobase base ϩ2 was unexpected, the data of Fig. 2 indicate this substitution greatly enhanced reactivity of ϩ1T to KMnO 4 oxidation, suggesting that loss of stacking induced by removing template nucleobase ϩ2 allows both the Ϫ1/ϩ1 and ϩ1/ϩ2 phosphodiester bonds of the primer access to the catalytic center. Although not shown experimentally, it is also conceivable that removing template nucleobase ϩ2 induces its complement to assume an extrahelical configuration, which might also allow the ϩ1/ϩ2 phosphate junction to be accessed by the catalytic center.
For the RNA primer, removing nucleobases Ϫ1 and Ϫ2, although retaining the sugar-phosphate backbone, clearly inhibits hydrolysis at the PPT/U3 junction (Fig. 6, A and B, panels ii and iii). From studies with E. coli RNase H, Uchiyama et al. (34) suggest an outer sphere complex is formed between the catalytic Mg 2ϩ ion and the 2Ј-OH group of the nucleoside 5Ј to the scissile bond. We therefore interpret loss of PPT/U3 cleavage with substrate Ϫ1RAb as loss or alteration of this critical interaction. With respect to the nucleobase 3Ј of the PPT/U3 junction, Haruki et al. (35) have exploited RNA/DNA hybrids containing phosphorothioate substitutions to propose that the pro-R p -oxygen of the phosphate group 3Ј to the scissile bond cooperates during catalysis with the catalytic His-124 of E. coli RNase H, analogous to the model of substrate-assisted catalysis proposed for type II restriction endonucleases (36). Fig. 7 suggests that eliminating primer nucleobase ϩ1A could directly alter contacts with Arg-448 or indirectly influence backbone geometry to antagonize its cooperation with His-539. Finally, the wild type hydrolysis profile derived from substrate ϩ2RAb indicates that critical primer contacts do not extend beyond the nucleoside 3Ј to the scissile bond.
Although the model we have proposed here will require further validation, an induced fit would seem a plausible mechanism to correctly position HIV-1 RT on its cognate substrate for accurate cleavage at the PPT/U3 junction. Flexibility resulting from weakened base pairing centered ϳ13 bp upstream of the Ϫ1/ϩ1 phosphodiester bond would promote a primary interaction with helices ␣-H and ␣-I of the p66 thumb, which indirectly positions the RNase H catalytic center over the Ϫ1/ϩ1 scissile bond. NMR studies are presently underway to validate this hypothesis.