Interaction of the Ty3 Reverse Transcriptase Thumb Subdomain with Template-Primer*

Amino acid sequence alignment was used to identify the putative thumb subdomain of reverse transcriptase (RT) from the Saccharomyces cerevisiae long terminal repeat-containing retrotransposon Ty3. The counterpart to helix αH of human immunodeficiency virus type 1 (HIV-1) RT, which mediates important interactions with a duplex nucleic acid ∼3-6 bp behind the DNA polymerase catalytic center, was identified between amino acids 290 and 298 of the Ty3 enzyme. The consequences of substituting Ty3 RT Gln290, Phe292, Gly294, Asn297, and Tyr298 (the counterparts of HIV-1 RT Gln258, Leu260, Gly262, Asn265, and Trp266, respectively) for both DNA polymerase and RNase H activities were examined. DNA-dependent DNA synthesis was evaluated on unmodified substrates and on duplexes containing targeted insertion of locked nucleic acid analogs and abasic lesions in either the template or primer. Based on this combined strategy, our data suggest an interaction of Ty3 RT Tyr298 with primer nucleotide -3, Gly294 with primer nucleotide -4, and Asn297 with template nucleotide -6. Substitution of Ala for Gln290 was well tolerated, despite the high degree of conservation at this position. Mutations in the thumb subdomain of Ty3 also affected RNase H activity, suggesting a closer spatial relationship between its N- and C-terminal catalytic centers compared with HIV-1 RT.

Crystallographic studies of human immunodeficiency virus type 1 (HIV-1) 1 reverse transcriptase (RT) with duplex DNA (1,2) and an RNA/DNA hybrid (3) have identified several protein motifs essential for efficient nucleic acid polymerization and hydrolysis as well as translocation of the replication machinery over a variety of conformationally distinct nucleic acid duplexes. In addition to the DNA polymerase and RNase H catalytic centers, the ␤12-␤13 hairpin (or "primer grip") of the p66 DNA polymerase domain maintains the primer terminus in an orientation appropriate for nucleophilic attack on the incoming deoxynucleoside triphosphate (4 -7). Contacts with the template are mediated via elements of the p66 fingers and palm subdomains collectively designated the "template grip" (8 -10). Participation of the RNase H primer grip in imposing the correct trajectory on the RNA strand of RNA/DNA hybrid for hydrolysis at the RNase H catalytic center has been suggested (11)(12)(13). Finally, a network of contacts between residues of helices ␣H and ␣I at the base of the p66 thumb with the primer backbone 3-6 bp behind the polymerase active center may compose a "translocation track" and also assist the transition in nucleic acid geometry from the A to B form downstream of the DNA polymerase catalytic center, an event accompanied by considerable duplex bending (14 -20). The combined biochemical and biological data available from mutagenesis studies lend strong support to the proposed functions of these protein motifs.
Information gleaned from such studies has accrued almost exclusively from analysis of retroviral enzymes, in particular those of human, avian, and murine origin. In contrast, there is a paucity of detailed biochemical information on equivalent events mediated by RT from long terminal repeat (LTR)-containing retrotransposons such as Ty1 and Ty3 of Saccharomyces cerevisiae and Ty5 of Saccharomyces paradoxus. Such an analysis is particularly important in view of differences at several steps of the LTR-containing retrotransposon reverse transcription cycle. Examples of such differences include (a) initiation of (Ϫ)-strand DNA synthesis from an internal site of the cognate tRNA primer in Ty5 (21), (b) use of a bipartite primer-binding site to initiate (Ϫ)-strand DNA synthesis in Ty3 (22), (c) self-primed initiation of (Ϫ)-strand DNA synthesis (23), (d) a role for RNA branching and debranching in Ty1 (Ϫ)strand DNA transfer (24,25), and (e) divergence in both size and sequence of the (ϩ)-strand polypurine tract (PPT) primers of Ty1 and Ty3 (26,27). The recent availability of active recombinant Ty1 (28,29) and Ty3 (30 -32) RTs now allows cis-acting signals of LTR-containing retrotransposons and protein motifs of their cognate RTs to be compared and contrasted with their retroviral counterparts.
In this study, secondary structure prediction programs and amino acid sequence alignments were used to identify the Ty3 RT equivalent of HIV-1 RT helix ␣H. Crystallographic analyses of HIV-1 RT have indicated that Gln 258 , Gly 262 , and Trp 266 of this ␣-helix are in close contact with the sugar phosphate backbone of the primer, playing an important role in translocation and as a "sensor" of nucleic acid configuration (1-3, 33, 34). These studies have also implicated p66 Asn 265 , in this case via an interaction with the sugar phosphate backbone of the template. Mutations at these positions of the HIV-1 enzyme have pleiotropic effects, including alterations in DNA binding, processivity of DNA synthesis, and frameshift fidelity (18 -20) and PPT usage (15). Selected residues within this putative ␣-helix of Ty3 RT were mutated and evaluated with respect to the DNA polymerase and RNase H functions of the purified enzymes. In addition to studying duplex DNA and RNA/DNA hybrids, polymerization efficiency on duplex DNA containing targeted insertion of locked nucleic acid analogs (35) or abasic lesions (36,37) in the template or primer was examined. Our data indicate that Gly 294 , Asn 297 , and Tyr 298 of Ty3 RT may mediate the same role as their HIV RT counterparts (Gly 262 , Asn 265 , and Trp 266 , respectively). Alterations in templateprimer usage were noted when Ty3 RT Phe 292 (HIV-1 RT Leu 260 ) was altered, whereas alanine substitution at the conserved Gln 290 (HIV-1 RT Gln 258 ) was tolerated with minimal alteration to DNA polymerase and RNase H functions. Collectively, our data provide the first comprehensive biochemical study of an important structural motif of the Ty3 RT DNA polymerase domain.

EXPERIMENTAL PROCEDURES
Oligonucleotides-Standard DNA oligonucleotides and those containing locked nucleic acid analogs or abasic lesions were purchased from Integrated DNA Technologies (Coralville, IA). RNA oligonucleotides for analysis of RNase H activity were purchased from Dharmacon (Boulder, CO).
Preparation and Purification of Ty3 RT Mutants-Point mutations in the DNA polymerase domain of Ty3 RT expressed on plasmid p6HTy3RT (30) were introduced using the QuikChange XL site-directed mutagenesis kit (Stratagene). Briefly, sense and antisense mutagenic oligonucleotides were purchased for each point mutation and annealed to the denatured p6HTy3RT plasmid. Extension and incorporation of the mutagenic primers during temperature cycling were performed with PfuTurbo DNA polymerase. The reaction was treated with DpnI endonuclease to digest the parental DNA template and then transformed into Escherichia coli XL10-Gold ultracompetent cells. Ty3 RT mutants were purified by a combination of metal chelate and size exclusion chromatography according to Lener et al. (31) and stored in a 50% glycerol-containing buffer at Ϫ20°C. Protein concentration was determined using the Quick Start TM Bradford protein assay (Bio-Rad).
DNA Polymerase Activity-RNA-dependent DNA polymerase activity was evaluated on a 152-nucleotide (nt) RNA template (prepared by in vitro transcription), corresponding to nucleotides 2225-2363 of the Ty3 genome plus 14 additional nucleotides, hybridized to a 32 P-5Ј-endlabeled 21-nt DNA primer. Template-primer was annealed by incubation at 95°C in 10 mM Tris-HCl (pH 7.8) and 100 mM NaCl and slow cooling to room temperature. A reaction mixture containing 50 nM template-primer and each dNTP at a final concentration of 250 M was prepared in buffer comprising 10 mM Tris-HCl (pH 7.8), 80 mM NaCl, 5 mM dithiothreitol, 0.01% (v/v) Triton X-100, and 9 mM MgCl 2 . DNA synthesis was initiated at 30°C by addition of wild-type or mutant RT to a final concentration of 100 nM in a final reaction volume of 10 l and terminated after 10 min by mixing with an equal volume 89 mM Tris borate (pH 8.3), 2 mM EDTA, and 95% (v/v) formamide containing 0.1% (w/v) bromphenol blue and xylene cyanol. Polymerization products were resolved by high voltage denaturing 10% polyacrylamide gel electrophoresis and visualized by phosphorimaging. Processivity during RNAdependent DNA polymerase activity was evaluated using the substrate indicated above, but with the RT concentration increased to 200 nM. After incubating RT and template-primer for 5 min at 30°C, DNA synthesis was initiated by addition of dNTPs and a heparin trap to final concentrations of 250 M and 2 mg/ml, respectively. The efficiency of the competitor was determined by adding it prior to Ty3 RT and templateprimer, which resulted in complete inhibition of DNA synthesis (data not shown). DNA-dependent DNA polymerase activity was evaluated on HhaI-digested single-stranded M13mp18 DNA (U. S. Biochemical Corp.) to which the 32 P-end-labeled M13 universal primer was hybridized. (Unlike the majority of restriction enzymes, HhaI cleaves singlestranded DNA.) Assay conditions were similar to those described above, with the exception that 1.5 nM template-primer was used.
Determination of Dissociation Rate Constants by Nucleotide Incorporation-Wild-type and mutant Ty3 RTs (100 nM) were preincubated for 5 min at 30°C with 8 nM template-primer (40-mer DNA hybridized to a 32 P-5Ј-end-labeled 30-mer DNA primer) in 10 mM Tris-HCl (pH 7.8), 80 mM NaCl, 5 mM dithiothreitol, 0.01% (v/v) Triton X-100, and 9 mM MgCl 2 . Heparin was subsequently added to a final concentration of 2 mg/ml, and incubation was continued for varying times before 250 M dNTP was added. One minute following dNTP addition, the reaction was stopped with an equal volume of 89 mM Tris borate (pH 8.3), 2 mM EDTA, and 95% (v/v) formamide containing 0.1% (w/v) bromphenol blue and xylene cyanol. Starting polymerization with the dNTP mixture at varying times following heparin addition reveals the fraction of enzyme bound to the template at a particular time point. The fraction of enzyme bound to the template as a function of time was plotted and fit using KaleidaGraph with a single exponential function: f(t) ϭ exp(Ϫk off ⅐t), where k off is the dissociation rate constant and t is the incubation time with the trap.
DNA Polymerase Activity on Substrate Analogs-A 40-nt DNA template containing an abasic lesion was hybridized to a 32 P-end-labeled 18-nt DNA primer such that the lesion was positioned 3, 4, 5, or 7 nucleotides from the primer 3Ј-OH (defining position Ϫ1 as the first template-primer base pair in the catalytic center). Template-primer (2.5 nM) was preincubated at 30°C with 15 nM Ty3 RT in the reaction mixture described above. Synthesis was initiated by addition of dATP to a final concentration of 8 M, which permitted addition of a single nucleotide. Single nucleotide extension on duplexes containing locked nucleic acid (LNA) substitutions of the primer or template was evaluated in a similar manner or in the presence of all four dNTPs.
RNase H Activity-Nonspecific RNase H activity was determined on a 5Ј-end-labeled 40-nt RNA template annealed to a 30-nt DNA primer (31). A reaction mixture containing 50 nM template-primer was prepared in buffer containing 10 mM Tris-HCl (pH 7.8), 80 mM NaCl, 5 mM dithiothreitol, and 9 mM MgCl 2 . Hydrolysis was initiated by addition of enzyme to a final concentration of 100 nM in a final volume of 10 l and terminated after 30 min by mixing with an equal volume of 89 mM Tris borate (pH 8.3), 2 mM EDTA, and 95% (v/v) formamide containing 0.1% (w/v) bromphenol blue and xylene cyanol. Hydrolysis products were resolved by high voltage denaturing 15% polyacrylamide gel electrophoresis and visualized by phosphorimaging. For Ty3 PPT selection, a 46-nt (Ϫ)-strand DNA template (corresponding to nucleotides 4780 -4809 of the Ty3 genome) was hybridized to a 29-nt PPT-containing RNA including 13 nt 3Ј to the PPT/U3 cleavage site. The duplex was annealed as described above, and a reaction mixture containing 50 nM template-primer was prepared in 25 mM Tris-HCl (pH 7.8), 9 mM MgCl 2 , 80 mM NaCl, 5 mM dithiothreitol, and 0.01% (v/v) Triton X-100. Hydrolysis was initiated by addition of RT to a final concentration of 150 nM in a 80-l volume. The reaction mixture was incubated at 30°C. Hydrolysis products were evaluated by high voltage electrophoresis and autoradiography.
Sequence Alignment and Structure Modeling-The RT sequences of chosen retroviruses and LTR-containing retrotransposons were multialigned using the ClustalW function of MacVector. The alignment was done in respect to well defined amino acid motifs 1-7 from the RT fingers and palm subdomains (38). The secondary and tertiary structure prediction was accomplished using MetaServer (available at bioinfo.pl). Two templates with the highest 3D-Jury score (Protein Data Bank codes 1HMV and 1MU2 for HIV-1 and HIV-2 RTs, respectively) were chosen to predict the tertiary structure of Ty3 RT using the MODELLER program of the same server.

RESULTS
Mutagenesis Strategy-The lack of detailed structural information for Ty3 RT necessitated applying amino acid sequence alignment and structural prediction programs to locate the counterpart of HIV-1 RT helix ␣H. Within the DNA polymerase active center of Ty3 RT, the proposed -Tyr-Leu-Asp-Asp-active-site motif between residues 211 and 214 was recently verified biochemically by site-directed mutagenesis (39). Multiple alignments involving amino acid sequences of both retroviral and LTR-containing retrotransposon RTs were conducted (Fig. 1A). Using this strategy, seven conserved segments defined previously by Xiong and Eickbush (38) were identified within Ty3 RT. As revealed by crystallographic data for retroviral RTs (1, 2, 40 -42), these conserved segments are located within the fingers and palm subdomains, with the last segment (motif 7 or the ␤12-␤13 hairpin of HIV-1 RT) connecting the palm and thumb subdomains. C-terminal to this segment in Ty3 RT (amino acids 288 -300) is a region with homology to HIV-1 RT helix ␣H (residues 256 -268). Secondary structure prediction analysis was also used to confirm that this sequence of Ty3 RT could assume a helical configuration. Based on these combined approaches, we have tentatively designated Ty3 RT residues 290 -298 as the structural counterpart to HIV-1 RT helix ␣H. The same structural prediction analysis was used to identify the Ty3 counterpart of helices ␣I and ␣J in the thumb of HIV-1 RT (Fig. 1C). Because Gln 258 , Gly 262 , and Trp 266 of HIV-1 RT occupy the same face of helix ␣H and are involved in multiple contacts with the sugar phosphate backbone of the primer (Fig. 1B) (1-3), their Ty3 counterparts (Gln 290 , Gly 294 , and Tyr 298 , respectively) were selected for mutagenesis. In addition, Ty3 Asn 297 was selected because its HIV-1 equivalent (Asn 265 ) contacts the sugar phosphate backbone of the template. Finally, Ty3 Phe 292 was altered to alanine based on the high degree of conservation at this position of LTR-containing retrotransposon RTs.
DNA Polymerase Activity of Thumb Mutants-Mutations in the Ty3 RT thumb were initially characterized with respect to their DNA-and RNA-dependent DNA polymerase activities. Activity was determined in the context of multiple rounds of DNA synthesis or by inclusion of the competitor heparin, restricting this to a single binding event, the results of which are presented in Fig. 2. Under conditions allowing multiple rounds of synthesis, DNA-dependent DNA synthesis activity and pausing patterns of mutants Q290A, N297A, Y298F, and Y298W ( Fig. 2A, lanes 1, 6, 8, and 9, respectively) were similar to those of wild-type Ty3 RT (lane w). All Phe 292 substitutions and mutant Y298A caused a slight diminution in DNA polymerase activity ( Fig. 2A, lanes 2-4 and lane 7, respectively). In contrast, the relatively modest G294A substitution resulted in enhanced pausing at multiple positions on the DNA template, some of which were unique to this mutant ( Fig. 2A, lane 5). In the presence of heparin, only mutants Q290A and Y298F displayed activity similar to that of wild-type Ty3 RT (Fig. 2B,  lanes 1, 8, and w, respectively), whereas other mutants showed varying levels of activity. A significant reduction in polymerase activity was observed with mutants F292A, G294A, Y298A, and Y298W. Such a reduction may reflect decreased affinity for template-primer or, alternatively, template-specific features such as secondary structure. A second processivity assay with a shorter DNA template was performed and likewise demonstrated that the processivity of these mutants was still affected (data not shown).
RNA-dependent DNA polymerase activity in the absence of heparin (Fig. 2C) was essentially the same pattern for all mutants, with the exception that enhanced pausing was not evident with mutant G294A on an RNA template (lane 5). Under conditions permitting a single binding event, the activity of mutants Q290A, Y298F, and Y298W was equivalent to or slightly less than that of Ty3 RT (Fig. 2D, lanes 1, 8, 9, and w, respectively). All remaining mutants displayed severe polymerization and processivity defects under these conditions. Dissociation Rate Constants-As a more quantitative measure of interactions involving Ty3 RT and duplex DNA, dissociation rate constants were determined for mutants carrying alanine substitutions (Fig. 3). For each mutant, a time course is presented, and dissociation rate constants have been tabulated in the inset. The value for mutant Q290A (0.098 s Ϫ1 ) is close to that of wild-type Ty3 RT (0.055 s Ϫ1 ), which is in keeping with the data of Fig. 2. Such values are within the general range we (43) and others (44) described previously for p66/p51 HIV-1 RT. The most significant effect was observed for mutant G294A, which dissociated ϳ8-fold more rapidly from template-primer (0.425 s Ϫ1 ), followed by F292A (0.247 s Ϫ1 ), Y298A (0.224 s Ϫ1 ), and N297A (0.150 s Ϫ1 ).
Single Nucleotide Extension on Duplex DNA Containing LNA Substitutions-In the absence of crystallographic data, an indirect approach was taken to investigate specific contacts between duplex DNA and the Ty3 thumb subdomain. DNA duplexes containing LNA substitutions were synthesized and used in a single nucleotide extension assay. These 2Ј-O,4Ј-Cmethylene-linked bicyclic analogs (Fig. 4A) have the property of locking the deoxyribose ring in the C-3Ј-endo configuration, thereby increasing the local organization of the phosphate backbone (45). LNA modifications were localized within the region of the primer and template predicted to be involved in contacts with the Ty3 RT thumb subdomain (positions Ϫ3 to Ϫ7, defining position Ϫ1 as the first base pair in the catalytic center) (Fig. 4 B). In the context of duplex DNA, the LNA methylene linker is oriented into the minor groove (46), which could potentially disrupt contacts between Ty3 RT and the FIG. 2. A and B, DNA-dependent DNA polymerase activity of Ty3 RT mutants. Activity was determined in either the absence (A) or presence (B) of the competitor heparin. The amino acids selected for mutagenesis are indicated above each panel, below which is the residue to which they were altered. Lane w, wild-type Ty3 RT. P, unextended primer. The reaction was performed with an HhaI-digested single-stranded M13 DNA template to which a 5Ј-radiolabeled M13 universal primer was annealed. C and D, RNA-dependent DNA polymerase activity evaluated in the absence and presence of heparin, respectively. The nucleic acid substrate used was an in vitro transcribed 152-nt RNA template (comprising sequence 2225-2363 of the Ty3 genome) to which a 5Јradiolabeled DNA primer was annealed. Lanes c, no enzyme. sugar phosphate backbone. Fig. 4C indicates that wild-type Ty3 RT was significantly inhibited when the LNA primer substitution was positioned close to the DNA polymerase catalytic center, i.e. nucleotides Ϫ3 and Ϫ4. Compared with the unsubstituted duplex, the inhibitory effect decreased when primer nucleotide Ϫ5 was substituted, and modest stimulation was noted when nucleotides Ϫ6 and Ϫ7 were replaced. Such a result suggests that primer nucleotides Ϫ3 to Ϫ5 comprise a region of close contact with the enzyme. In contrast, LNA template substitutions at positions Ϫ3 to Ϫ5 did not inhibit primer extension, but induced the opposite effect, with the efficiency of nucleotide addition increasing as the substitution was relocated from position Ϫ3 to Ϫ5. However, replacement of template nucleotides Ϫ6 and Ϫ7 inhibited polymerization, indicating potential involvement of the sugar phosphate backbone of template bases Ϫ6 and Ϫ7 in contacting Ty3 RT. This asymmetric LNA-induced pattern of inhibition for both the primer and template co-localizes with interactions defined crystallographically for HIV-1 RT (2,34).
Although the inhibitory effect of the primer between positions Ϫ3 and Ϫ5 and template modification at positions Ϫ6 and Ϫ7 was consistent with predicted LNA-induced structural interference, stimulation of DNA synthesis when template nucleotides Ϫ3, to Ϫ5 were replaced was unexpected. One possible explanation for this might be local deformation of the DNA duplex. Incorporation of the LNA analog into the duplex has been demonstrated to change the local conformation, inducing widening of the minor groove (45). The structural consequence of LNA substitution should be particularly pronounced in this assay, as the substrate contains dA tracts previously shown to induce minor groove narrowing and possibly alter enzyme contacts (47,48). Incorporation of an LNA analog into the DNA duplex would therefore be predicted to alleviate this effect, as observed experimentally in Fig. 4C.
Activity of Ty3 Mutants on DNA Containing LNA Primer Substitutions-One way to exploit targeted LNA insertions is whether the structural interference they induce is relieved in the context of a particular RT mutant, which would suggest a point of contact between the wild-type residue and the LNA analog. The histograms of Fig. 5 compare the efficiency of single nucleotide addition of each Ty3 mutant on duplexes containing LNA primer substitutions with the activity of the same mutant on unsubstituted DNA. The method of evaluation is analogous to that recently adopted by Zang et al. (49) to study nucleotide addition opposite guanine analogs containing N-2 adducts. The inhibitory effect of LNA substitution at positions Ϫ3 and Ϫ4 was not observed with Ty3 RT mutants F292A, F292W, F292Y, and G294A (Fig. 5, panels iii-vi, respectively), which previously showed decreased processivity and templateprimer affinity (Fig. 2). We attribute such tolerance to LNA modification to relaxed interactions between the thumb subdomain and the sugar phosphate backbone of the primer. For mutants Q290A, N297A, Y298A, Y298W, and Y298F (Fig. 5, panels ii and vii-x, respectively), primer modifications caused more diverse effects. As for wild-type RT, an LNA analog at position Ϫ3 and Ϫ4 decreased primer extension efficiency for most of the mutants. The exception was mutant Y298A (Fig. 5,  panel viii), where a substrate containing a substitution at position Ϫ3 was more efficiently extended than both a substrate containing a substitution at position Ϫ4 and the unmodified duplex (Figs. 4C and 5, panel i). This phenotype clearly differs from that of mutants Y298W and Y298F, demonstrating that removing the bulky aromatic side chain at position 298 compensates for LNA-induced interference, implicating Tyr 298 in contacting the DNA primer at position Ϫ3. Furthermore, mutant Q290A was stimulated on a primer containing a substitution as position Ϫ5 (Fig. 5, panel ii), suggesting that this may be a point of contact with Gln 290 . With the exception of wild-type Ty3 RT and mutant Y298F, priming efficiency on duplexes containing substitutions at positions Ϫ6 and Ϫ7 was enhanced, in particular with mutants that showed a more pronounced defect in processivity or template-primer binding (Fig. 2, B and D). This result suggests that the LNA-induced deformation introduced into the DNA duplex partially compensates for the inability of mutant enzymes to correctly bind template-primer (for example, by increasing accessibility to the minor groove).
Activity of Ty3 Mutants on DNA Containing LNA Template Substitutions-As shown in Fig. 6, LNA substitution at positions Ϫ3 to Ϫ5 of the template increased priming efficiency for wildtype Ty3 RT and its variants compared with the unmodified template. Conversely, modification at positions Ϫ6 and Ϫ7 decreased priming, with the former being slightly more inhibitory. Elimination of the inhibitory effect on substrates containing substitutions at positions Ϫ6 and Ϫ7 with Phe 292 mutants (in particular, mutant F292A) suggests a more global effect, possibly through destabilizing thumb architecture (Fig. 1C). In contrast, inhibition of DNA synthesis was relieved for mutant N297A specifically on the template containing a substitution at position Ϫ6, suggesting that the wild-type residue contacts the sugar phosphate backbone of this nucleotide, i.e. although introducing an LNA analog interferes with this interaction, decreasing the size of the side chain of Asn 297 suppresses the inhibition.
Relatively low stimulation of priming was noticed for mutant F292A on a substrate containing a substitution at position Ϫ4 (Fig. 6, panel iii). In our structural model of Fig. 1C, the side chain of Phe 292 is oriented toward the hydrophobic core of the thumb and may participate in stabilizing the local structure. Replacing this residue with Ala could introduce a structural deformation to induce nonspecific interference with the DNA template containing the LNA analog at position Ϫ4.
Single Nucleotide Extension on Duplex DNA Containing Abasic Primer Lesions-Modification of duplex DNA with LNA analogs was designed to probe Ty3 RT interactions with the sugar phosphate backbone of the minor groove. An alternative assay using abasic lesions (50 -55) was applied to examine potential interactions with nucleobases of the duplex. Single nucleotide extension was evaluated on the substrate in Fig. 4B whose primer in this case contained an abasic lesion at position Ϫ3, Ϫ4, Ϫ5, or Ϫ7. As with LNA insertions, the activity of each mutant on the substituted duplex was compared with that of the same mutant on the wild-type duplex. For wild-type RT, primer extension was reduced on substrates PϪ3Ab and PϪ4Ab, whereas near wild-type levels were observed for substrate PϪ5Ab, and full activity was achieved with substrate P-7Ab (Fig. 7, panel i). Thus, although the data with wild-type Ty3 RT in Fig. 7 (panel i) do not directly indicate which thumb residues contact a nucleic acid, we predict that productive primer extension requires an interaction with primer bases Ϫ3 and Ϫ4. Similar results were obtained with mutant Q290A (Fig. 7, panel ii). Also, although the priming efficiency of mutants F292A and N297A was lower than that of wild-type RT (Fig. 7, panels iii and v, respectively), the profiles on each substituted duplex follow a pattern similar to that of wild-type Ty3 RT, supporting the notion that these amino acids do not mediate critical primer nucleobase contacts between positions Ϫ3 and Ϫ7. In contrast, a different pattern was observed for mutant G294A, where activity comparable with that on the unsubstituted duplex was restored with substrates P-4Ab and P-5Ab (Fig. 7, panel iv). To understand this result, modeling of HIV-1 RT complexed with duplex DNA was performed, where Gly 262 (the counterpart of Ty3 RT Gly 294 ) was substituted with Ala. This exercise indicated that the methyl group introduced into the side chain was pointed toward the stacking interface between nucleobases Ϫ4 and Ϫ5 (data not shown). Removing the primer nucleobase might then be considered as a means of suppressing steric interference introduced by a Gly-to-Ala substitution, arguing that the Gly 294 main chain of Ty3 RT mediates contacts with primer nucleobase Ϫ4.
The most difficult effect of nucleobase elimination to explain was the priming efficiency of mutant Y298A, where removing the aromatic side chain inhibited priming on substrates P-4Ab and P-5Ab (Fig. 7, panel vi). Clearly, this phenotype cannot be explained in terms of suppression of steric interference. Based on studies with LNA-substituted primers, Tyr 298 was proposed to contact the sugar phosphate backbone at position Ϫ3 (Fig. 5,  panel vii) and potentially stabilize the local A-like conformation of the DNA duplex. However, in the experiment in Fig. 7, removing a nucleobase altered the stacking environment in its immediate vicinity. Because our proposal based on Fig. 1B implicates Gly 294 in contacting the sugar phosphate backbone at position Ϫ4, a Y298A substitution might indirectly affect contacts involving neighboring nucleobases by changing stacking interactions.
Mutations in the Thumb Subdomain Affect RNase H Activity-Altering protein motifs contacting a nucleic acid near the DNA polymerase catalytic center would be predicted in the first instance to inhibit this enzymatic function. However, nucleic acid positioning within the C-terminal RNase H domain and RNase H function might be indirectly affected, a notion supported by our previous studies of mutants of the HIV-1 RT thumb (15). Amino acid sequence alignment of retroviral and LTR-containing retrotransposon RTs (56) also indicated that the latter lack a connection subdomain, suggesting a closer spatial relationship between the two catalytic centers. We therefore evaluated the RNase H activity of our Ty3 RT mutants on a nonspecific 40-nt RNA/30-nt DNA hybrid and a second substrate mimicking the junction between the Ty3 PPT and downstream U3 RNAs. Because accurate cleavage at the PPT/U3 junction is necessary for subsequent integration of the double-stranded DNA element, PPT processing determines whether RNase H cleavage specificity is altered. The results of our RNase H analysis are presented in Fig. 8.
Under conditions allowing multiple binding events, wildtype RT and mutants Q290A, N297A, Y298A, and Y298F displayed comparable activity (Fig. 8A, lanes w, 1, and 6 -8, respectively). Interestingly, although the relatively modest Y298F substitution had a negligible effect, the F292Y substitution severely impaired RNase H activity (Fig. 8A, lanes 8 and  4, respectively). Also, although Ala could be substituted for Tyr 298 without loss of RNase H activity (Fig. 8A, lane 7), the equivalent substitution of Phe 292 severely affected activity (lane 3). Based on our model in Fig. 1C, such results are consistent with the notion that Ty3 RT Phe 292 is involved in maintaining overall thumb architecture. Finally, although mutant G294A retained DNA-and RNA-dependent DNA polymerase activities (Fig. 2, A and C, lanes 5), its RNase H activity was more severely affected (Fig. 8A, lane 5).
In Fig. 8B, RNase H cleavage specificity was evaluated via hydrolysis of an RNA/DNA hybrid mimicking the Ty3 PPT/U3 junction. Because the RNA strand of this hybrid contains an additional 13 nt downstream of the junction, this substrate allows simultaneous evaluation of both specific and nonspecific hydrolyses (32). As an example, although low level hydrolysis occurred between positions ϩ1 and ϩ10, wild-type RT and mutants Q290A and Y298A cleaved preferentially at the PPT/U3 junction (Fig. 8B, lanes w, 1, and 7, respectively). A minor but reproducible effect of the Y298A mutation was elevated cleavage within the PPT at position Ϫ2. Although the significance of this novel cleavage is not immediately apparent, it is noteworthy that Ty3 proviral DNA, unlike other retroelements, has a 2-bp terminal extension. In general, the efficiency of PPT/U3 cleavage by the remaining mutants was similar to that on the nonspecific RNA/DNA hybrid. The exception was mutant Y298F (Fig. 8B, lane 8), which was less efficient at PPT/U3 cleavage, although (a) hydrolysis of the same RNA/ DNA hybrid between positions ϩ1 and ϩ10 was unaffected, and (b) this mutant efficiently cleaved a nonspecific RNA/DNA hybrid (Fig. 8A, lane 8). A similar result was obtained with HIV-1 RT carrying a mutation at Trp 266 , which we have proposed here is the counterpart to Ty3 RT Tyr 298 (15). DISCUSSION There is general consensus that the thumb subdomain of several DNA polymerases makes numerous contacts with the duplex product of DNA synthesis ϳ3-8 bp behind the catalytic site (57). Studies with the Klenow fragment of E. coli (58) and HIV-1 RT (14 -20) have suggested that a helix-turn-helix motif of the thumb tracks the minor groove of the nascent duplex, serving as an important modulator of both the processivity and fidelity of DNA synthesis. Using a combination of amino acid sequence alignment and homology modeling programs, we have tentatively identified Ty3 RT Gln 290 , Gly 294 , Asn 297 , and Tyr 298 as counterparts of HIV-1 helix ␣H Gln 258 , Gly 262 , Asn 265 , and Trp 266 , respectively (Fig. 1A). The predicted function of this helix-turn-helix motif in stabilizing the Ty3 ternary complex was evaluated via changes in processivity and dissociation rate constants for alanine-scanning mutants. Subsequently, the effect of targeted placement of either conformationally restrained LNA analogs or abasic lesions into the template and primer was employed to localize contacts between these residues of Ty3 RT and duplex DNA.
Primer extension in the presence of heparin (Fig. 2, B and D) highlighted a severe defect in processivity for mutants F292A, G294A, and Y298A; a moderate effect for mutants F292W, F292Y, N297A, and Y298W; and a wild-type phenotype for mutants Q290A and Y298F. The same reaction without heparin (Fig. 2, A and C) highlighted differences among Ty3 RT mutants most seriously affected in the processivity assay, indicating that the G294A substitution most dramatically affected ternary complex stability on a DNA duplex. Consistent with this trend, the largest increase in the dissociation rate was observed for mutant G294A (ϳ8-fold), followed by mutants F292A and Y298A (ϳ5-fold). The effects of Ty3 RT substitutions G294A and Y298A demonstrated here are consistent with analogous mutations in HIV-1 RT (G262A and W266A, respectively), which likewise exhibited elevated dissociation rate constants, together with reduced processivity and frameshift fidelity (19).
Comparing processivity and dissociation rate constants suggests an important function for Ty3 RT Phe 292 , Gly 294 , and Tyr 298 and, to lesser extent, Asn 297 in stabilizing the ternary complex. However, these approaches do not discriminate between direct contacts with duplex DNA and their alternative function of stabilizing local protein structure. To address this question, nucleoside analog interference mapping was exploited to probe potential contacts between the predicted Ty3 RT helix-turn-helix motif and duplex DNA immediately adjacent to the DNA polymerase catalytic center. For wild-type Ty3 RT, single LNA substitutions induced a different pattern of priming inhibition depending on whether the template or primer was substituted (Fig. 4C). LNA interference was noted at primer positions Ϫ3 and Ϫ4 (and to a lesser extent, Ϫ5) and template positions Ϫ6 and Ϫ7. Moreover, duplexes containing dual LNA insertions indicated that template interactions did not extend beyond position Ϫ7 (data not shown). The interference patterns obtained with our analog interference strategy co-localized with interactions defined for HIV-1 RT interaction with the sugar phosphate backbone of duplex DNA (2,34), illustrating the value of this approach when crystallographic data are unavailable.
Although the analog-induced inhibition pattern highlights a general region of interaction with the enzyme, suppression of LNA interference by Ty3 thumb subdomain mutants aids in defining specific interactions. For example, mutant Y298A efficiently extended a DNA primer containing the LNA analog at position Ϫ3 (Fig. 5, panel viii). In contrast, wild-type RT and mutants Y298W and Y298F were significantly inhibited by the same primer substitutions (Figs. 4C and 5, panels i, ix, and x, respectively). This result suggests interference between a bulky substituent at position 298 and the methylene-linked sugar at position Ϫ3 of the primer (Fig. 4A). Modeling studies suggested that substituting alanine for Tyr 298 compensated for such steric hindrance (data not shown). Single nucleotide extension by mutant N297A uncovered a second example of specific suppression of LNA-induced interference (Fig. 6, panel  vii). In this case, the inhibitory effect of LNA insertion into position Ϫ6 of the DNA template was eliminated.
Interference suppression was also observed with mutants F292A, F292W, and F292Y (Fig. 5, panels iii-v, respectively). However, in this case, priming inhibition was relieved equally with primers containing LNA substitutions at positions Ϫ3 to Ϫ5, potentially revealing a long-range effect of substituting Phe 292 . In our homology modeling (Fig. 1C), the side chain of Phe 292 (like its HIV-1 equivalent, Leu 260 ) projects into the thumb rather than toward the nucleic acid. It is therefore reasonable to assume that these residues may mediate hydrophobic interactions with other helical elements of the thumb subdomain. Thus, the "long-range" suppression observed when Phe 292 is replaced may be an indirect consequence of altering thumb architecture on contacts between neighboring residues and duplex DNA. An alteration in thumb architecture would also be consistent with inhibition of RNase H activity with mutants F292A, F292Y, and F292W (Fig. 8). Because this alteration is not expected to be limited to the interaction of a single residue with duplex DNA but instead affect the local environment (for example, destabilization of a structural element), the trajectory of the heteroduplex within the binding cleft of Ty3 RT may be distorted. Such changes and their implications for RNase H activity have been documented previously (3,15). Single nucleotide extension reactions with substrates containing strategically placed abasic analogs in the templateprimer duplex supported our contention that bp Ϫ3, Ϫ4, and, to a lesser extent, Ϫ5 participate in important contacts within the Ty3 RT nucleic acid-binding cleft. These results are consistent with crystallographic data determined for the complex between HIV-1 RT and double-stranded DNA (2). The inhibitory effect of nucleobase removal at positions Ϫ4 and Ϫ5 of the primer noted for most mutants was suppressed by mutant G294A (Fig.  7, panel iv), where modeling suggested that the extra methyl group introduced by Ala would most likely be oriented toward the stacking interface between nucleobases Ϫ4 and Ϫ5, which would cause steric interference (data not shown). Thus, removing one of these nucleobases might be predicted to compensate for this interference. With respect to the unmodified duplex, the structural hindrance hypothesized here may alter duplex trajectory within the RT binding cleft, resulting in loss of RNase H activity with this mutant (Fig. 8), although an increased dissociation rate is an alternative explanation. Although a role for Ty3 RT Gln 290 was not apparent from our results, mutagenesis data at Phe 292 , Gly 294 , Asn 297 , and Tyr 298 are consistent with our amino acid sequence alignment and homology modeling, revealing a thumb architecture similar to FIG. 8. RNase H activity of Ty3 RT mutants. A, activity determined on a nonspecific RNA/DNA hybrid, where Ty3 RT cleaves at positions corresponding to 13 and 21 nt behind the DNA polymerase catalytic center (defined as positions Ϫ13 and Ϫ21, respectively); B, activity determined on a Ty3 PPT-containing RNA/ DNA hybrid. The position of specific cleavage at the PPT/U3 junction is indicated. Lane notations are explained in the legend to Fig. 2. Lanes c, no enzyme; lanes w, WT Ty3 RT. that previously defined for the complex between HIV-1 RT and double-stranded DNA (Fig. 1C).
Finally, crystallization of monomeric Moloney murine leukemia virus RT (42) has suggested a lack of conservation with residues of HIV-1 RT proposed by biochemical and structural studies to mediate translocation (14,15,18,33,34,59), in particular, Gln 258 , Gly 262 , and Trp 266 , whose Ty3 RT counterparts have been the subject of this study. Although it is conceivable that the thumb subdomain of the murine enzyme has a unique architecture, the data of this study suggest that the monomeric Ty3 (30) and heterodimeric HIV-1 enzymes exploit common structural motifs for translocation of the polymerization machinery.