Restoration of tRNA3Lys-primed(-)-strand DNA synthesis to an HIV-1 reverse transcriptase mutant with extended tRNAs. Implications for retroviral replication.

The mechanism for the initiation of reverse transcription in human immunodeficiency virus type 1 (HIV-1) was studied utilizing a unique reverse transcriptase (RT) mutant altered in its noncatalytic p51 subunit. This mutant (p66/p51Delta13) retains full DNA- and RNA-dependent DNA polymerase activity but has reduced affinity for tRNA3Lys, the cognate HIV primer. When the ability to support(-)-strand DNA synthesis on a viral RNA template was evaluated, this mutant initiated from an 18-nucleotide (nt) oligoribo- or oligodeoxyribonucleotide primer complementary to the primer binding site (pbs). However, it failed to do so from natural and synthetic versions of tRNA3Lys. tRNA-primed(-)-strand synthesis could, however, be rescued by substituting the 76-nt tRNA3Lys with 81- and 107-nt tRNA-DNA chimeras, i.e. tRNA3Lys extended by 5 and 31 deoxyribonucleotides complementary to the viral genome upstream of the pbs. These findings imply that through interactions involving its p51 subunit, RT may be required to disrupt additional tRNA-viral RNA duplexes outside the pbs to proceed into productive(-)-strand DNA synthesis. Alternatively, specific interactions between tRNA3Lys and HIV-1 RT may be necessary for efficient initiation of(-)-strand DNA synthesis.

Initiation of reverse transcription in retroviruses is a complex process requiring several key elements. These include virus-coded reverse transcriptase (RT), 1 a specific host tRNA, viral genomic RNA, and auxiliary components such as the nucleocapsid protein. Convergent evolution of retrovirus and host has most likely increased the specificity of retroviral reactions requiring host molecules. Thus, placement of a specific host tRNA on the retroviral genome and utilization of the tRNA-viral RNA complex by a specific RT reflect highly organized and distinct events. Recent studies provide increasing evidence of extensive interplay between elements near the 5Ј end of the retroviral genome and its cognate tRNA primer in controlling initiation of (Ϫ)-strand DNA synthesis (1). In addi-tion to the 18-base pair intermolecular duplex of primer binding site (pbs) and tRNA 3Ј nucleotides (Fig. 1, region A), biochemical and genetic evidence (2)(3)(4)(5) suggests formation of a viral RNA⅐tRNA complex involving the tRNA T⌿C arm and a heptanucleotide sequence in the 5Ј unique (U5) sequence of viral RNA (Fig. 1, region B). Initially demonstrated for Rous sarcoma virus (2), sequence analysis indicates that this complex may be generally applicable (5). Chemical footprinting of the leader-stem of HIV-1 genomic RNA and tRNA 3 Lys (the HIV replication primer) has revealed another site of intermolecular contact, i.e. between the tRNA U-rich anticodon loop and the A-rich U5-IR loop of viral RNA immediately preceding the pbs (Fig. 1, region C1) (6,7). In contrast, Kohlstaedt and Steitz (8) suggest that the tRNA anticodon loop interacts with viral RNA sequences immediately 3Ј to the pbs (Fig. 1, region C2). Despite apparent discrepancies, these findings collectively implicate higher order RNA structures (9) in the "preinitiation" complex, which must be disrupted subsequent to the onset of (Ϫ)-strand DNA synthesis. Involvement of tRNA primers in such structures might ensure that following sequestration into the virion, they are directed to the pbs rather than other regions of the genome with which they display reduced homology.
We have demonstrated that activities of p66/p51 HIV-1 RT can be modulated through minor truncations in the C-terminal "connection" subdomain (10) of its noncatalytic p51 subunit (11). In particular, reconstituted heterodimer whose p51 subunit lacks residues Gln 428 -Phe 440 (p66/p51⌬13) shows wild type polymerase activity from a heteropolymeric templateprimer while showing severely reduced affinity for free tRNA 3 Lys . These findings are consistent with crystallographic data (10) proposing that the p51 connection subdomain interacts with the T⌿C and D arms of pbs-bound tRNA, whose 3Ј terminus is extended into (Ϫ)-strand DNA at the p66 polymerase catalytic center. The notion that tRNA 3 Lys and a templateprimer duplex can interact with different subdomains of RT is suggested by mutagenesis studies on the "primer grip" (12). Altering Trp 229 in the ␤12-␤13 hairpin of this p66 motif results in normal tRNA binding but severely reduced affinity for template-primer (13). 2 Our results with RT mutant p66/p51⌬13, coupled with a potential contribution of tRNA-viral RNA loop-loop interactions to initiation of (Ϫ)-strand DNA synthesis, prompted us to evaluate how an enzyme with reduced affinity for free tRNA 3 Lys interacted with and utilized the pbs-bound intermolecular duplex. We demonstrate here that although p66/p51⌬13 RT efficiently extends 18-nt DNA and RNA primers hybridized to a pbs-containing viral RNA template, it fails to do so when these are replaced by natural or synthetically prepared tRNA 3 Lys . However, if the 76-nt tRNA is replaced with 81-or 107-nt tRNA-DNA chimeras (i.e. tRNA extended by 5 or 31 deoxynucleotides at its 3Ј terminus), the ability of p66/p51⌬13 RT to support (Ϫ)-strand strong stop DNA synthesis is restored. This is not the case with a tRNA primer whose 3Ј terminus contains a single deoxynucleotide. These findings suggest that the tRNA binding deficiency of p66/p51⌬13 RT is further manifested in an inability to disrupt loop-loop interactions of the tRNA-viral RNA preinitiation complex. Properties of additional deletion mutants provide further insights into structural elements of the p51 connection subdomain, which might participate in tRNA binding.
Determination of RNA-and DNA-dependent DNA Polymerase Activity-A 90-nt RNA template or 71-nt DNA template (0.2 pmol) was annealed to a [ 32 P] 5 Ј end-labeled 36-nt DNA primer (0.1 pmol) by heating at 85°C for 2 min and slowly cooling to 37°C in a buffer of 25 mM Tris/HCl, pH 7.5, and 25 mM KCl (14). HIV-1 RT and MgCl 2 were then added to final concentrations of 42 nM and 10 mM, respectively. After 30 s at 37°C, dNTP/ddNTP elongation combinations allowing primer extension by 4 or 19 nucleotides (13,14) were added to the reaction mixture (final concentrations 50 and 500 M, respectively). The final reaction volume was 10 l. DNA synthesis was terminated 10 min later by the addition of urea-based gel loading buffer, and samples were fractionated by denaturing high resolution gel electrophoresis. For heparin challenge experiments, competitor was added to a final concentration of 2.0 mg/ml. Synthetic and Natural tRNA 3 Lys -Synthetic 32 P-internally labeled tRNA 3 Lys was prepared by in vitro transcription of a 256-base pair cassette containing the tRNA gene linked to a bacteriophage T7 promoter (15). Natural (i.e. fully modified) tRNA 3 Lys was prepared from bovine liver as described previously (6).
Oligonucleotide-and tRNA-primed (Ϫ)-Strong Stop DNA Synthesis-A pbs-containing RNA template, flanked by U5 and repeat sequences of the HXB2 genome (16), was prepared from AccI-cleaved pHIV-PBS by in vitro transcription (17). 18-nt oligodeoxy-and oligoribonucleotides complementary to the pbs, as well as synthetic and natural tRNA 3 Lys , were evaluated as primers. Template and primer were mixed at a 2:1 ratio in annealing buffer (50 mM Tris/HCl, pH 7.5, 100 mM KCl). The mixture was heated to 85°C for 2 min then slowly cooled to allow formation of secondary structures and annealing of tRNA primers to the template. Template-primer was preincubated with wild type or mutant RT (42.5 nM) and 10 mM MgCl 2 prior to the addition of dNTPs to 100 M. The final reaction volume was 20 l. DNA synthesis was allowed to proceed 30 min at 37°C. Products from (Ϫ)-strand DNA synthesis were internally labeled with [␣-32 P]dATP and [␣-32 P]dCTP (5 Ci of each; ICN) when comparing the efficiency of synthetic and natural tRNA 3 Lys as primers. Alternatively, end-labeled DNA and RNA primers or internally labeled synthetic tRNA 3 Lys were employed. Reaction products were fractionated by high voltage electrophoresis through denaturing 6% polyacrylamide gels, visualized by autoradiography, and quantified by a PhosphorImager (Molecular Dynamics).
RNA-dependent DNA Synthesis from tRNA 3 Lys -DNA Chimeras-In a first step, internally labeled synthetic tRNA 3 Lys was annealed to a 60-nt pbs-containing DNA template (representing nucleotides ϩ604 to ϩ663 of HIV-1 HXB2 ) as described above. Klenow fragment (50 units, Boehringer Mannheim) and dNTP mixtures allowing addition of 1, 5, or 31 deoxynucleotides to tRNA 3 Lys were added to the template-primer duplex, followed by incubation at 37°C for 1 h. The 77-, 81-, and 107-nt tRNA 3 Lys -DNA chimeras were freed from the DNA template by denaturing 10% polyacrylamide gels and purified from crushed gel slices (18). These tRNA-DNA chimeras were subsequently hybridized to the pbscontaining RNA template and assayed for their ability to support (Ϫ)strand strong stop DNA synthesis as described above.
Determination of RNase H Activity-RNase H activity was determined qualitatively using a heteropolymeric 90-nt RNA template 36-nt DNA primer whose template was radiolabeled at its template 5Ј terminus with [␥-32 P]ATP and polynucleotide kinase (12). Template hydrolysis was determined in the absence of DNA synthesis. In order to visualize hydrolysis intermediates, samples were analyzed after incubating wild type or mutant RT for 5 s or 5 min with the RNA-DNA hybrid. Hydrolysis products were evaluated by high voltage denaturing electrophoresis and autoradiography. Product lengths were determined by co-electrophoresis of DNA size markers.

DNA Polymerase Activities of Selectively Deleted
Mutants-DNA-and RNA-dependent DNA polymerase activities of the selectively deleted mutants were assessed by "programmed" synthesis on defined, heteropolymeric template-primer combinations (13,14). Mutant p66/p51⌬19 RT displayed a more distributive mode of DNA-dependent synthesis in the absence of the competitor heparin ( Fig. 2A, v). At the same time, its rapid dissociation from template-primer was evident following heparin challenge (restricting each enzyme to a single round of synthesis), where little to no nascent DNA was detected. The behavior of p66/p51⌬19 RT is in keeping with previous reports using single-stranded M13 DNA as template (11). The remaining enzymes supported efficient DNA-dependent DNA synthesis in the absence and the presence of heparin ( Fig. 2A, i-iv), suggesting that stepwise removal of up to 13 C-terminal residues from p51 influences neither the affinity of the reconstituted heterodimer for a DNA template-DNA primer nor the structural integrity of its DNA polymerase catalytic center. In the absence of heparin challenge, RNA-dependent DNA synthesis activity of p66/p51⌬19 RT was again severely compromised, whereas p66/p51⌬5, p66/p51⌬9, and p66/p51⌬13 RT were unaffected (Fig. 2B, i-v). Lower affinity of RT for the RNA template and rapid dissociation resulted in low level synthesis upon heparin challenge. Although the yield was low, equivalent levels of Pϩ4 product were derived from each enzyme, with the exception of p66/p51⌬19. tRNA 3 Lys -primed Synthesis of (Ϫ)-Strand Strong Stop DNA-With the exception of p66/p51⌬19 RT, the data of Fig. 2 indicate that heteropolymeric template-primers are efficiently utilized by mutants containing a truncated p51 subunit. However, since previous data indicated reduced affinity of mutant p66/ p51⌬13 for synthetic tRNA 3 Lys in gel-mobility shift assays (11), we determined how this and other mutants supported DNA synthesis from a pbs-bound tRNA primer.
The 497-nt pbs-containing HIV-1 RNA template of Fig. 3A contains repeat, U5, pbs, and 5Ј noncoding sequences (17) predicted to adopt extensive intramolecular base pairing (7,19), in addition to intermolecular structures with the anticodon and T⌿C loops of tRNA 3 Lys (3)(4)(5)(6)(7)(8). Mutants p66/p51⌬5 and p66/ p51⌬9 were capable of extending both synthetic and natural variants of tRNA 3 Lys into full-length (Ϫ)-strand strong stop DNA, although with lower efficiency than the parental enzyme (Fig. 3B, lanes 1-3). The efficiency of (Ϫ)-strand DNA synthesis from natural tRNA was ϳ10-fold greater than from its synthetic counterpart, which is in keeping with predictions of Isel et al. (6,7) from chemical probing studies of tRNA-viral RNA duplexes. In contrast, the ability of mutants p66/p51⌬13 and p66/p51⌬19 RT to support DNA synthesis from either natural or synthetic tRNA 3 Lys was severely compromised (Fig. 3B, lanes  4 and 5). A 100-fold drop in (Ϫ)-strand strong stop DNA synthesis was estimated between reactions catalyzed by wild type enzyme and mutant p66/p51⌬13, whereas product was completely absent from those catalyzed by p66/p51⌬19. The experiment of Fig. 2B utilizes internally labeled nascent (Ϫ)-strand DNA and therefore failed to address whether mutants p66/ p51⌬13 and p66/p51⌬19 RT could initiate (Ϫ)-strand DNA synthesis. In order to address this, the analysis was repeated with internally labeled, synthetically prepared tRNA 3 Lys (Fig.  3C). The virtual absence of stalled (Ϫ)-strand products in the vicinity of the tRNA primer in reactions catalyzed by mutants p66/p51⌬13 and p66/p51⌬19 suggests a deficiency prior to initiation rather than during elongation. Reducing the salt concentration from 100 to 5 mM allowed limited tRNA-primed (Ϫ)-strand synthesis by mutant p66/p51⌬13 but still substantially less than with the wild type enzyme under the same conditions (data not shown).
Oligonucleotide-primed Synthesis of (Ϫ)-Strand Strong Stop DNA-Lack of p66/p51⌬13-catalyzed DNA synthesis from the pbs-bound tRNA 3 Lys primer (Fig. 3) suggested the inability to: (i) bind and utilize tRNA 3 Lys as a primer, in keeping with previous data (11) or (ii) initiate RNA-dependent DNA synthesis from an RNA primer on an RNA template. In order to distinguish between these possibilities, we evaluated the ability of p66/p51⌬5, p66/p51⌬9, and p66/p51⌬13 RT to support (Ϫ)-strand DNA synthesis from 18-nt RNA and DNA primers hybridized to the pbs-containing viral RNA genome. The results of these experiments are illustrated in Fig. 4, indicating that both oligonucleotide primers are efficiently extended by p66/p51⌬13. As described in an earlier study (20), increased stalling was evident in oligodeoxynucleotide-primed reactions (Fig. 4C).
Data of Fig. 4C clearly illustrate that the inability of p66/ p51⌬13 RT to initiate (Ϫ)-strand DNA synthesis from pbsbound tRNA cannot be attributed to the nature of the primer 3Ј terminus or features of the viral RNA template. This could imply that recognition and destabilization of an intermolecular duplex between the viral RNA template and the T⌿C loop of tRNA primer by RT (6, 7) may be a prerequisite to efficient initiation of (Ϫ)-strand DNA synthesis. tRNA 3 Lys -DNA Chimeras as Primers for (Ϫ)-Strand DNA Synthesis-Inefficient utilization of tRNA 3 Lys and efficient use of oligonucleotides as primers of (Ϫ)-strand DNA synthesis by p66/p51⌬13 (Figs. 3 and 4, respectively) pointed increasingly toward an inability of the enzyme to interact with and destabilize a higher order intermolecular structure assumed by the tRNA-viral RNA duplex, as suggested by several groups (2)(3)(4)(5)(6)(7)(8).
To test this hypothesis, we evaluated whether our mutants supported synthesis of (Ϫ)-strand strong stop DNA from a series of tRNA-DNA chimeras in "two-step" DNA synthesis experiments. This procedure, outlined schematically in Fig. 5A, involves limited extension of the tRNA primer on a DNA tem- plate with the Klenow fragment of DNA polymerase, recovery and hybridization of the tRNA-DNA chimera to the viral RNA template, and subsequent extension by wild type and mutant RT. A similar approach was adopted by Wöhrl et al. (21) to analyze the p66 and p51 subunits of equine infectious anemia virus RT. Although Klenow fragment inefficiently catalyzed tRNA extension on the short DNA template, it avoided potential degradation of the tRNA primer by the RNase H activity of RT. Fig. 5B indicates that wild type RT supports (Ϫ)-strand DNA synthesis from a 77-nt tRNA-DNA chimera, i.e. the 76-nt tRNA containing dC at its 3Ј terminus. Although the activity of p66/ p51⌬5 and p66/p51⌬9 RT with this primer is reduced, the levels are in keeping with the data of Fig. 3 (B and C). In contrast, p66/p51⌬13 RT is unable to overcome the barrier to initiation of (Ϫ)-strand DNA synthesis. If, however, the 77-nt primer is replaced by an 81-nt tRNA-DNA chimera, (i.e. tRNA 3 Lys containing the 3Ј extension dC-dT-dC-dT-dA) p66/ p51⌬5, p66/p51⌬9, and p66/p51⌬13 RT support levels of (Ϫ)strand DNA synthesis comparable with that of the parental heterodimer, whereas mutant p66/p51⌬19 remains inactive (Fig. 5C). Region C1 of Fig. 1 may be disrupted with tRNA 3 Lys extended at its 3Ј terminus by 5 deoxynucleotides (81-nt tRNA-DNA chimera). However, extension of tRNA 3 Lys by a single deoxynucleotide (77-nt tRNA-DNA chimera) would be predicted to have only a minimal effect on the loop-loop interactions between tRNA 3 Lys and the viral RNA template. DNA synthesis by p66/p51⌬13 RT was also restored when a 107-nt tRNA-DNA chimera (tRNA 3 Lys extended by 31 deoxynucleotides at its 3Ј terminus) was hybridized to the viral RNA genome (Fig. 5D). However, significantly increased pausing by all RTs was observed in reactions primed by the Tϩ31 chimera compared with those primed by the shorter chimeras or tRNA 3 Lys . The 107-nt tRNA-DNA chimera would be predicted to anneal to regions C1 and B of Fig. 1, which have been proposed to interact with tRNA 3 Lys via its anticodon (6, 7) and T⌿C loops (2-5), respectively. The complex involving region C2 of Fig. 1 may also be destabilized by the use of a Tϩ31 chimeric primer due to a complete unwinding of the U5-IR stem-loop.
Deletions in p51 RT Alter RNase H Function of Heterodimerassociated p66 -The structure of unliganded p66/p51 HIV-1 RT containing a full-length copy of p51 (22) suggests that the extreme C terminus of this subunit is adjacent to its thumb subdomain and proximal to the RNase H domain of p66. Thus, as the p51 C terminus is gradually removed, the geometry of these two elements may be subtly altered. In order to test this hypothesis, we analyzed the RNase H activity of each mutant on a defined heteropolymeric template-primer, and visualized the products by autoradiography. The results in Fig. 6 lend credence to our hypothesis. Fig. 6A provides a schematic representation of the RNase H assay. Enzyme located over the primer 3Ј OH will cleave the RNA template nucleotide Ϫ17, which defines the spatial separation of the DNA polymerase and RNase H domains (8,12,22). In our system, endonucleolytic cleavage results in release of a 71-nt product. Subsequent directional processing as far as FIG. 3. tRNA 3 Lys -primed (؊)-strand strong stop DNA synthesis. A, schematic representation of the exogenous DNA synthesis reaction. The 18-nt viral pbs sequence is bold and is preceded by 173 nt from the repeat and U5 region of the RNA genome. Intact (Ϫ)-strand strong stop DNA is evidenced by a 250-nt tRNA-DNA chimera. B, tRNA-primed DNA synthesis activity of selectively deleted HIV-1 RT mutants. In these experiments, nascent DNA was internally labeled with 32 P. Lane notations S and N refer to DNA synthesis primed with synthetic (i.e. unmodified) and natural (i.e. fully modified) tRNA 3 Lys , respectively. position Ϫ8 yields a 62-nt product. These two activities and their uncoupling were recently described by Ghosh et al. (14). In Fig. 6B, the Ϫ17 hydrolysis product can be visualized following short incubation with wild type RT and with prolonged incubation is replaced with the Ϫ8 product (Fig. 6B, lanes 1). Although the rate of hydrolysis is slightly slower for the selectively deleted mutants p66/p51⌬5 and p66/p51⌬9, their cleavage specificity is retained. In contrast, the RNase H activity of p66/p51⌬13 RT is quite different. Following short incubation, additional Ϫ21 and Ϫ25 hydrolysis products were evident, reflecting a relaxed specificity of the RNase H domain. Furthermore, with prolonged incubation, hydrolysis was only observed as far as template nucleotide Ϫ17. Hydrolysis products representing Ϫ17, Ϫ21, and Ϫ25 intermediates were also produced by mutant p66/p51⌬19 although greatly reduced. Thus, in an analogous manner to the findings of Ghosh et al. (14) directional processing activity of p66/p51⌬13 RT appears to have been lost. However, although Ghosh et al. (14) achieved this phenotype by altering the p66 RNase H domain, the data of this communication illustrate that the same can be achieved by subtle alterations to the geometry of an intact RNase H domain.

DISCUSSION
Genetic (2)(3)(4)(5) and biochemical data (6 -8) point increasingly toward higher order RNA structures outside the pbs controlling initiation of retroviral replication. Although alternative tRNAviral RNA duplexes have been proposed, each implicates looploop interactions involving bases of the anticodon (6 -8) or T⌿C arms (2-5) of the tRNA primer. For HIV, the importance of these structures has been investigated with a unique RT mutant lacking 13 C-terminal p51 residues (p66/p51⌬13) whose DNA polymerase activity on heteropolymeric RNA and DNA templates was unaffected. However, unlike wild type enzyme, this mutant failed to initiate (Ϫ)-strand DNA synthesis from tRNA 3 Lys annealed to an HIV-1 pbs-containing RNA template. This feature was specifically tRNA 3 Lys -dependent, because p66/ p51⌬13 could extend pbs-bound RNA or DNA primers to fulllength (Ϫ)-strand strong stop DNA. This mutant will, however, catalyze synthesis of (Ϫ)-strand DNA from tRNA 3 Lys containing deoxynucleotide extensions of 5 and 31 nt at its 3Ј terminus, which increase homology with the viral genome to 23 and 49 nt, respectively. Restoration of DNA synthesis by the two chimeras suggests disruption of tRNA-viral RNA structures by the mu- tant. Alternatively (although less likely), this could reflect increased affinity of the mutant for an extended nucleic acid duplex of chimeric primer and RNA template.
The connection subdomain at the p51 C terminus has been proposed to accommodate the anticodon and D stems and loops of tRNA 3 Lys (10). However, recent data suggest that only residues in the p66 subunit interact directly with tRNA 3 Lys (23). In all available structures, residues 421-426 of the p51 C termi-nus are found as a short ␣ helix 3 and form a key part of the interface between the p66 RNase H domain and the p51 connection subdomain (8,12,22). Only in crystals of the unliganded HIV-1 RT with a full 440-residue p51 are residues 427-433 visible and shown to extend toward the p51 connection and 3 D. Rodgers and S. Harrison, personal communication. palm subdomains, opposite the nucleic acid binding cleft. Although the p51 subunit may not be in direct contact with primer tRNA (23), truncating its C terminus could alter the subunit interface and disrupt the structural framework required for a proper interaction of the p66 subunit with tRNA 3 Lys . These alterations must be subtle, because (i) p66/ p51⌬13 accommodates and extends oligonucleotide DNA and RNA primers as efficiently as the parental enzyme (Figs. 2 and 4) and (ii) the tRNA priming deficiency is rectified with p66/ p51⌬9, which restores only four C-terminal residues to p66/p51⌬13.
The same nucleotides of the anticodon loop of tRNA 3 Lys involved in specific binding to RT (U 32 -U 37 ) (24) interact with the A-rich U5-IR loop of viral RNA (Fig. 1, region C1) (6,7). By binding to tRNA, wild type RT positioned over the primer terminus may destabilize this loop-loop structure, as well as other tRNA-viral RNA interactions to initiate DNA synthesis. p66/p51⌬13 RT, whose affinity for the free tRNA primer is reduced (11), appears unable to disrupt these tRNA-viral RNA strucutres and subsequently initiate (Ϫ)-strand DNA synthesis from pbs-bound tRNA. Annealing of 81-(Tϩ5) and 107-nt (Tϩ31) tRNA-DNA chimeras to viral RNA would selectively disrupt two proposed structures, i.e. those between the anticodon loop/U5-IR loop (Fig. 1, region C2) (6,7) and the T⌿C loop/U5-IR stem (Fig. 1, region B) (4), respectively. These tRNA-DNA chimeras were efficiently utilized as primers by p66/p51⌬13 RT, suggesting that constraints preventing this mutant from extending tRNA 3 Lys or a 77-nt (Tϩ1) tRNA-DNA chimera are released. Thus, disruption of the anticodon loop-U5-IR loop structure may be required for p66/p51⌬13 to productively use the tRNA primer.
As predicted by computer-generated nucleic acid folding programs, extending tRNA by 5 nt is unlikely to disrupt a pro-posed interaction between its T⌿C loop and the U5-IR stem of the viral RNA (Fig. 1, region B), suggesting this intermolecular interaction is less critical for HIV replication than was recently demonstrated for Rous sarcoma virus (2,3,5). In support of this notion, inspection of several HIV-1 genomes suggests that homology to the tRNA primer can often extend into viral sequences directly 3Ј to the pbs. This would have the consequence of making the T⌿C loop unavailable for hybridization to the U5-IR stem while having little effect on the anticodon loop-U5-IR loop interaction. However, it is possible that alternative structures have evolved to control initiation of replication in different retroviruses. For example, the U5-IR loop of other retroviral genomes have considerably lower probability for a loop-loop interaction with anticodon bases of their cognate tRNA primer (4). Furthermore, an interaction between the D arm of tRNA i Met and genomic sequences 50 -100 nt downstream of the pbs has also been proposed to play a role in initiation of (Ϫ)-strand synthesis in the retrotransposon Ty (25). Thus, although different structures are invoked, recognition of an extensively base paired tRNA-viral RNA preinitiation complex appears a mechanism common to retroviruses and retroviruslike elements. The results of this communication also appear to contradict the data of Kohlstaedt and Steitz (8), who suggested that the interaction of an A-rich segment of the viral genome 3Ј to the pbs and the tRNA anticodon loop provides an important control mechanism. A possible reason for this discrepancy might lie in the fact that these authors used tRNA hybridized to a short DNA template rather than a larger portion of viral RNA genome, which may prevent formation of stable intermolecular base paired structures proposed by others (2-7). However, it is equally conceivable that retroviral initiation proceeds through several conformationally distinct nucleoprotein complexes prior to productive elongation, one of which involves the A-rich loop 3Ј to the pbs. Clearly, additional studies are required to more precisely evaluate conformational changes in the initiation complex and their susceptibility to therapeutic attack.
An alternative mechanism for the restoration of tRNAprimed DNA synthesis to p66/p51⌬13 by tRNA-DNA chimeras could be that extended complementarity to the viral RNA template renders initiation events independent of structural features of the tRNA primer. In recent work (26), we demonstrated that p66/p51 HIV-1 RT accommodates 23 or 24 base pairs of the template-primer duplex during DNA-dependent DNA synthesis. A hybrid between 81-nt tRNA-DNA chimera and viral genome may allow the 23 base pairs of templateprimer duplex to be better accommodated by the replicating enzyme. Inefficient use of tRNA or a 77-nt (Tϩ1) tRNA-DNA chimera as primer might then suggest that the duplex region of these template-primers were too short to stably bind p66/ p51⌬13. However, this hypothesis would be inconsistent with our finding that 18-nt DNA or RNA primers were more efficiently extended into (Ϫ)-strand DNA than tRNA 3 Lys or any chimera with extended homology to the RNA template.
Finally, data presented here provide a good example of how subunit and subdomain interactions influence the multiple activities of this highly versatile retroviral enzyme. Although the RNase H domain of mutant p66/p51⌬13 is intact, loss of directional processing activity suggests that the architecture of the C-terminal p66 domain can be subtly altered by a minor truncation of the p51 subunit. The same architectural alteration may also affect the interaction with tRNA 3 Lys . Although the complete 440-residue p51 subunit is used here, heterodimeric RT has been prepared from recombinant microorganisms where a bacterial protease cleaves p66 in the vicinity of the p51-p66 junction rather than directly between Phe 440 FIG. 6. RNase H activity of selectively deleted HIV-1 RT mutants. A, schematic representation of the RNase H assay. Template and primer nucleotides occupied by RT (filled ellipsoids) in the absence of DNA synthesis are those determined by a combination of molecular modeling and DNase I footprinting (26). Initial endonucleolytic hydrolysis of the template at position Ϫ17 generates a 71-nt RNA. Subsequent directional processing activity to template position Ϫ8 reduces the size of this fragment to 62 nt (14). B, RNase H activity of selectively deleted mutants. Hydrolysis products were visualized after 5 s (left) and 5 min (right) of incubation with wild type or mutant enzyme. Lanes 1, wild type p66/p51; lanes 2, p66/p51⌬5; lanes 3, p66/p51⌬9; lanes 4, p66/ p51⌬13; lanes 5, p66/p51⌬19. The sizes of hydrolysis products were determined by co-electrophoresis of DNA sequencing ladders. and Tyr 441 (10,12). The possibility therefore exists that the DNA polymerase and RNase H activities of such enzymes may be altered and escape detection by conventional analyses, which often measure the total amount of product rather than its nature.