Mechanistic Studies of Early Pausing Events during Initiation of HIV-1 Reverse Transcription*

We have investigated the role of sequences that surround the primer binding site (PBS) in the reverse transcriptase-mediated initiation of (−) strand DNA synthesis in human immunodeficiency virus type 1. In comparisons of reverse transcription initiated from either the cognate primer tRNALys.3 or a DNA primerd-Lys.3, bound to PBS sequences, we observed that a +3 pausing site occurred in both circumstances. However, the initiation reaction with tRNALys.3 was also characterized by a pausing event after incorporation of the first nucleotide. Alteration of sequences at the 5′-end instead of the 3′-end of the PBS resulted in elimination of the +3 pausing site, suggesting that this site was template sequence-dependent. In contrast, the pausing event at the +1 nucleotide position was still present in experiments that employed either of these mutated RNA templates. The mutations at the 5′-end of the PBS also caused a severely diminished rate of initiation and the strong arrest of reactions at the +1 stage when tRNALys.3 was used as primer. Therefore, we propose that the +1 pausing event is an initiation-specific event in regard to reactions primed by tRNALys.3 and that sequences at the 5′-end of the PBS may facilitate the release of reverse transcription from initiation to elongation.

Retroviruses employ specific tRNA molecules as primers to initiate the synthesis of (Ϫ) strand DNA (1)(2)(3)(4)(5). These primer tRNAs bind to an 18-nt 1 segment of viral genomic RNA termed the primer binding site (PBS). Although such binding is principally mediated by a stretch of 18 nt close to the 3Ј-end of tRNA, other sequences within viral genomic RNA and the tRNA primer are also involved and can modulate the initiation of (Ϫ) strand DNA synthesis, as shown in both retroviral (6 -10) and other (11)(12)(13)(14)(15)(16) systems. In HIV-1, this includes a stretch of four nucleotides, i.e. 622 AAAA 625 (the A-rich loop) that interacts directly with positions 33 USUU 36 of the anticodon loop of primer tRNALys. 3. This A-rich loop is important for initiation of (Ϫ) strand DNA synthesis and generation of progeny virus (17).
In HIV-1, the initiation stage of synthesis of (Ϫ) strand DNA, primed by tRNALys.3, can be distinguished from subsequent strand elongation in regard to both the binding and kinetic properties of reverse transcriptase (RT) (18 -20). These reports showed that initiation was characterized by both early short (Ϫ) strand DNA products, resulting from pausing at the ϩ3 or ϩ5 nt positions, and rapid dissociation of RT from the initiation complex. The secondary structure formed between tRNALys.3 and the viral RNA template may play an important role in initiation of reverse transcription.
To pursue this subject, we developed an in vitro reverse transcription system in which low concentrations of dNTPs (i.e. 160 nM) were used to enhance our ability to detect very early pause sites, e.g. ϩ1 and ϩ3, in reactions primed with the tRNALys.3 cognate primer. To study the roles of sequences flanking the PBS, we generated a series of mutated HIV-1 RNA templates that contained mutations at both the 5Ј-and 3Ј-ends of the PBS that may potentially disrupt the secondary structure of complexes of tRNALys.3 and viral RNA template. Our data provide the first evidence for a ϩ1 pausing event in reverse transcription initiated from primer tRNALys. 3. We have also demonstrated on the basis of mutagenesis studies that the ϩ3 pausing site depends on the nature of sequences at the 5Ј-end of the PBS and that deletion of an A-rich loop in this region or of substitutions at the 5Ј-end of the PBS may make it difficult for primer tRNALys.3 to be extended beyond the ϩ1 pausing site.
In Vitro Reverse Transcription-These reactions were performed as described using RNA template that was generated through use of an Ambion Mega-scripts kit (Austin, TX) (see Fig. 1A) (17). tRNALys.3 prepared from human placenta (24), or a synthetic DNA primer complementary to the PBS, i.e. D-Lys.3 (5Ј-GTCCCTGTTCGGGCGCCA-3Ј) (positions 653-636), was annealed onto RNA template by denaturing at 85°C for 5 min and annealing at 55°C for 10 min in a reaction mixture containing 83 mM Tris-HCl (pH 7.5) and 125 mM KCl. Reverse transcription reactions were performed in a volume of 20 l containing 1 pmol of primer:RNA template complex, 50 mM Tris-HCl (pH 7.5), 75 mM KCl, 5 mM MgCl 2 , 10 mM DTT, dNTPs, 20 units of RNA guard ribonuclease inhibitor (Amersham Pharmacia Biotech) and reverse transcriptase at 37°C for various times, after which reactions were terminated by adding EDTA to a final concentration of 50 mM. In some experiments, reverse transcription reactions were performed in the presence of HIV-1 nucleocapsid (NC) protein (30 pmol for each reaction) as described (21). The cDNA products were fractionated on 8% denaturing polyacrylamide gels containing 7 M urea. The RT preparations used included wild-type HIV-1 enzyme (p66/51) prepared as described (25) and mutated HIV-1 RT containing a mutation at codon 478 (i.e. E478Q), responsible for defective RNase H activity (26).

RESULTS
The Initiation of HIV-1 (Ϫ) Strand DNA Synthesis from tRNALys.3 Is Characterized by Rate-limiting Pause Sites at nt Positions ϩ1 and ϩ3-Initiation of HIV-1 reverse transcription is a distinct step from that of elongation (18 -20). To further investigate this subject, we developed an in vitro reaction system that employed a PBS that contained only G, A, and C among the 5 nt at its 5Ј-end (Fig. 1A). Therefore, when dCTP, dTTP, and dGTP were included in these reactions, only early products of reverse transcription were generated. Reactions were also performed with only dCTP (Fig. 1A, lane 1) or both dCTP and dTTP (Fig. 1A, lane 2) to provide information on the positions of the first and second bands seen on gels. The results of Fig. 1A show that reaction products were observed at both the ϩ1 and ϩ3 positions in addition to a final expected product at the ϩ5 site (lane 3). These reactions did not pause at either the ϩ2 or ϩ4 positions, indicating that they were rate-limiting only after the first and third nt were added. To prove that the above-mentioned pause sites (i.e. ϩ1 and ϩ3) were not due to an absence of dATP in the reactions, experiments performed with all four dNTPs were terminated at different times (5, 15, or 45 min), with the result that the ϩ1 and ϩ3 pause sites were still present (Fig. 1A, lanes 4, 5, and 6).
We next observed that the ϩ1 and ϩ3 pause sites were also present when increased concentrations of dNTPs, i.e. 80 nM, 160 nM, 320 nM, 640 nM, 1.28 M, and 2.56 M were employed in reactions that yielded higher levels of (Ϫ) strand DNA products (Fig. 1B). However, the ϩ1 nt pausing site became increasingly faint with addition of higher concentrations of dNTP, whereas the ϩ3 nt pause site did not diminish in intensity; this suggests that both pause sites are integral features of HIV-1 reverse transcription reactions and are seen most clearly in reactions performed at dNTP concentrations of 160 nM (Fig. 1B, lane 2).
The ϩ3 Pause Site Is Dependent on Sequences at the 5Ј-End but not the 3Ј-End of the PBS-To study the role of sequences at the 5Ј-end of the PBS, we generated a mutated RNA template termed HIV/del-A that contained a deletion of the A-rich loop ( 622 AAAA 625 ) and a mutated RNA template HIV/HUA described above (Fig. 2A). The latter construct contained only T, A, and C within the 13 nt at the 5Ј-end of the PBS; hence, when only dATP, dTTP, and dGTP are included in reactions with the HIV/HUA template, extension should only proceed to the ϩ13 stage. In reactions performed with 45 ng of HIV-1 RT, tRNALys.3 as primer, and wt RNA genome, i.e. HIV/WT, three bands at positions ϩ1, ϩ3, and ϩ5 were clearly observed ( Fig.  2A, lane 9). However, when either HIV/del-A or HIV/HUA served as RNA template, only one band at position ϩ1 was seen, even when three different dNTPs were included (lanes 1-6); furthermore, this band was weaker than that seen at the same position with the wt RNA template HIV/WT. Considering the (Ϫ) strand DNA products at the ϩ3 and ϩ5 positions in the case of wt RNA template HIV/WT, deletion of the A-rich loop or substitutions within the region nt 624 -635 led to both greatly diminished efficiency of initiation, as well as an arrest of ex-tension at the ϩ1 nt position.
To investigate the effects of HIV/del-A and HIV/HUA on the ϩ3 pausing event, higher quantities of RT, i.e. 405 ng, were used to extend reactions beyond the ϩ1 stage. In this circumstance, extended products (both ϩ3 and ϩ5) were observed in reactions performed with the HIV/del-A mutated RNA template (Fig. 2B, lanes 1-3). However, when HIV/HUA was used, we detected extended products at the ϩ5 and ϩ13 sites and no pausing at the ϩ3 site (lanes 4 -6). Therefore, the presence of the A-rich loop (622-625) and maintenance of sequences at positions 624 -635 are necessary for a release from the ϩ1 nt pause site to occur; in addition, the strong pausing at the ϩ3 position is dependent on nt sequences 624 -635 at the 5Ј-end of the PBS.
We also investigated the 3Ј-end of the PBS in the initiation of (Ϫ) strand DNA synthesis through use of appropriate deleted RNA templates, i.e. HIV/LD1, HIV/LD2, and HIV/LD3, containing deletions at wt positions 654 -671, 672-691, and 692-707, respectively (Fig. 3). When tRNALys.3 was used as primer, similar band patterns were observed in reactions that used either wt RNA template (HIV/WT) or the mutated RNA templates (HIV/LD1, HIV/LD2, and HIV/LD3), although reactions proceeded less efficiently with the latter constructs (Fig. 3).
Reverse Transcription Does Not Pause at the ϩ1 Site when Initiated from a DNA Primer-As shown above, reverse transcription initiated from tRNALys.3 still paused at the ϩ1 position, even when sequences at both the 5Ј-and 3Ј-ends of the PBS were changed. When the first nt at the 5Ј-end of the PBS was changed from G to T (i.e. template HIV/HUA), the addition of the first dATP still represented a rate-limiting step ( Fig. 2A, lanes 4 -6). To study whether this was unique to RNA primers, reactions were also performed with an 18nt DNA primer, i.e. D-Lys.3, bound to the PBS.
When only dCTP was included in the reaction mixture, only one band was seen at position ϩ1 (Fig. 4A, lane 1). When both dCTP and dTTP were present, a band corresponding to nt position ϩ2 was observed, and that at the ϩ1 position had disappeared (lane 2). When dCTP, dTTP, and dGTP were included, two bands corresponding to the ϩ3 and ϩ5 nt pause sites were observed (lane 3). When all four dNTPs were present, extension occurred to beyond the ϩ5 position, although a strong pause site at nt position ϩ3 was still present, regardless whether reactions were run for 5, 15, or 45 min (Fig. 4A, lanes  4, 5, and 6). Thus, when a DNA oligomer was utilized as primer, reactions did not pause at the ϩ1 site, but they did pause at position ϩ3.
Wild-type HIV-1 RT (p66/51) possesses RNase H activity. Because the latter might interfere in assays that used a DNA primer, experiments were performed with a mutant RT (i.e. E478Q RT), defective in RNase H activity. The results of Fig.  4B show that reactions performed with this enzyme (i.e. RNase H-) did not pause at the ϩ1 site, although pausing still occurred at the ϩ3 position (lane 3). When all four dNTPs were included, further extension of (Ϫ) strand DNA was observed. However, reactions were still partially arrested at the ϩ3 position, even  1, 4, and 7), 15 min (lanes 2, 5, and 8), and 45 min (lanes 3, 6, and 9) in the presence of dCTP (␣-32 P), dTTP, and dGTP for each of the HIV/del-A and HIV/WT templates or dATP (␣-32 P), dTTP, and dGTP for the HIV/HUA template. In the case of HIV/HUA, reactions can reach the ϩ13 stage even when dCTP is absent, because the 13 nt upstream of the PBS do not include any Gs.  1, 4, 7, and 10), both dCTP (␣-32 P) and dTTP (lanes 2, 5, 8, and 11), or each of dCTP (␣-32 P), dTTP, and dGTP (lanes 3, 6, 9, and 12). over periods as long as 45 min (lanes 4 -6). Therefore, RNase H activity was not responsible for reaction release after addition of the first nt when a DNA primer was used.
As a control for our experiments with the 18 nt DNA primer, we also employed an 18 nt RNA primer complementary to the PBS. Consistent with results obtained with tRNALys.3 as primer, we found that these reactions also paused at the ϩ1 nt position (data not shown). Hence, the ϩ1 nt pausing event is associated with use of RNA primers during initiation of synthesis of (Ϫ) strand DNA.
As stated, the ϩ3 pause site was dependent on the nature of the template but not the primer (i.e. DNA or RNA) used in these reactions. To further verify such template dependence, experiments were performed with the DNA primer D-Lys.3 and the mutated viral RNA templates HIV/del-A and HIV/HUA. The results of Fig. 5 show that extension from the DNA primer occurred normally in reactions performed with either wt (HIV/ WT) or the mutated (HIV/del-A, HIV/HUA) RNA templates. However, when HIV/HUA was used, no strong pausing was observed at the ϩ3 position (lane 6), consistent with results obtained with the cognate primer, tRNALys.3 (Fig. 2B).
NC Protein Helps Reverse Transcription Escape from the ϩ1 Pausing Event-To study whether NC protein could affect the ϩ1 and ϩ3 pausing events in reverse transcription reactions primed with tRNALys.3, 30 pmol of NC protein was added to our 20-l reaction system, and reactions were terminated at various times (1, 4, 16, 32, and 64 min) (Fig. 6, lanes 1-5). A similar time-course without NC protein was also performed as a control (Fig. 6, lanes 6 -10). The data show that inclusion of NC resulted in significantly less pausing at the ϩ1 nt position and did not affect the ϩ3 nt pause site. Furthermore, more ϩ5 nt product was generated in the presence of NC protein, suggesting that NC had contributed to more efficient elongation of reverse transcription.

. Reverse transcription reactions (45 ng of HIV-1 RT) were primed with D-Lys.3 DNA using either mutated (HIV/del-A or HIV/HUA) or wild-type (HIV/WT) RNA as a template at 37°C
in the presence of 160 nM dNTPs and were run for 15 min. The order of lanes 1-9 is the same as that of Fig. 2A.

Pausing at A-rich Loop Positions during Synthesis of (Ϫ)
Strand DNA-When a still higher dNTP concentration was used, i.e. 2.56 M, four additional pause sites were observed at positions ϩ11, ϩ12, ϩ13, and ϩ14, corresponding to the four As found within the A-rich-loop ( 622 AAAA 625 ). These four bands diminished in intensity when reactions were incubated for longer times, e.g. 45 min (Fig. 7, lanes 4-6).
To confirm that these four bands were due to the presence of the A-rich loop in wt genomic RNA, the mutated template, HIV/del-A, deleted of these As, was studied. The results of Fig.  7 show this resulted in elimination of these bands (Fig. 7, lanes  7-9). When a different RNA template was used, i.e. HIV/A2 (deleted of the four As at positions 622-625 but containing As instead of Gs at positions 620 and 621, thus partially reconstituting a A-rich region), the result was reappearance of the pause sites at positions ϩ11, ϩ12, and ϩ13 (lane 10), although the pause site at the ϩ11 site gradually disappeared as reactions were incubated for longer times (e.g. 45 min) (lanes 10 -12). Therefore, the A-rich loop is apparently responsible for specific pause sites during synthesis of (Ϫ) strand DNA. DISCUSSION Biochemical analysis has shown that the initiation of HIV-1 reverse transcription can be distinguished from subsequent elongation (18 -20). The biological relevance of this observation is suggested by the fact that RT can lose its ability to discriminate against a nonself tRNA primer when the latter was extended by two nt (27). Second, in vitro labeling revealed that primer tRNALys.3 was extended by two nt within virus particles that had engaged in synthesis of (Ϫ) strand DNA (28). In our system, initiation has been characterized on the basis of several early pause sites (e.g. ϩ1 and ϩ3) that add substantially to our understanding of early events in reverse transcription.
This is the first demonstration that pausing at the ϩ1 nt site represents a rate-limiting step in reverse transcription reactions performed with an HIV-1 RNA template and the cognate primer tRNALys.3. This is not an unexpected finding because HIV-1 RT possesses both RNA-dependent DNA polymerization and DNA-dependent DNA polymerization activities. During reverse transcription, the enzyme is bound to either RNA-DNA or DNA-DNA hybrids during RNA-dependent DNA polymerization and DNA-dependent DNA polymerization, respectively, except at the initiation of synthesis of (Ϫ) strand strong-stop DNA ((Ϫ) ssDNA), when the enzyme is bound to a RNA-RNA hybrid and employs tRNALys.3 as primer for production of cDNA. After initiation takes place, the role of primer is effectively replaced by the newly made DNA, from which further extension will occur (5). Therefore, the initiation of (Ϫ) ssDNA synthesis is a distinct stage of reverse transcription, especially in regard to incorporation of the first dNTP.
When the first C deoxyribonucleotide from the dNTP pool is added to the 3Ј-OH of the A ribonucleotide at the 3Ј-end of tRNALys.3, displacement of a ribonucleotide-ribonucleotide pair (A-U) must occur in favor of a newly formed deoxyribonucleotide-ribonucleotide pair (dC-G) at the RT polymerization active site. Due to the absence of a 2Ј-OH residue in dC, the two  lanes 1-5) or by heat annealing (؊NC) (lanes  6 -10). Reactions were primed with tRNALys.3 using wild-type (HIV/ WT) RNA template at 37°C in the presence of 160 nM dCTP (␣-32 P), dTTP, and dGTP and were terminated at different times (1,4,16,32, and 64 min). nucleotide pairs (A-U and dC-G) may assume different conformations. Therefore, a structural rearrangement of the polymerization active site is required for the RT enzyme to adapt to the new dC-G pair and to add the next deoxyribonucleotide (dT) to the one-base extended primer. It is generally believed that a conformational change of RT must precede the chemical step, resulting in a rate-limiting event (29 -32). Our results show that this rate-limiting step results in the pause site at the ϩ1 position. Because the RT enzyme was originally bound to a RNA-RNA helix, it may require a conformational rearrangement to switch its binding mode from a RNA-RNA duplex to a DNA-RNA hybrid during synthesis of (Ϫ) ssDNA. Our results demonstrate that this conformational rearrangement may begin to occur as soon as the first dC has been added. In contrast, the use of the DNA primer, D-Lys.3, necessitates that the duplex to which RT is bound will always be a DNA-RNA hybrid, thus permitting the enzyme to catalyze the extension reaction without a change in conformation. Consequently, reactions primed by the DNA primer, D-Lys.3, do not need to pause after addition of the first dC.
Other studies have also indicated that a conformational rearrangement might be required for RT to proceed after addition of the first nt. Direct evidence for this comes from studies of the relationship between the RNase H cleavage site and the polymerization active site (26). Generally, a constant distance of 18 nt must exist between the RNase H cleavage site and the nascent primer 3Ј terminus. However, a distance of 19 nt instead of 18 nt was observed between these sites after incorporation of the first dC at the 3Ј-end of primer tRNALys.3 (26). Because the spatial relationship between the functional RNase H cleavage site and the polymerization active site serves as an indication of RT conformation, the 19-nt distance suggests a novel conformation for RT after incorporation of the first nt. Other evidence for this has been provided by studies of the binding and kinetic properties of HIV-1 RT during initiation and elongation that showed that RT dissociated approximately 200 times faster from the initiation (RNA template/tRNALys.3) than from the elongation complex (19). Further supporting this notion, mutagenesis studies of the HIV-1 RT palm subdomain have shown that RNA and DNA primers may be differentially recognized by RT, i.e. RT may be associated with RNA versus DNA primers in different conformations (33). This also helps to account for the importance of the ϩ1 pause site demonstrated in this paper.
We have also noted that synthesis of (Ϫ) strand DNA paused at the ϩ3 nt position, regardless whether tRNALys.3 or a DNA primer was employed. This suggests that the secondary structure of the complex formed between the viral RNA template and the primer must play a role in the specification of this pause site. The PBS and its flanking sequences constitute a highly ordered secondary structure. The binding of the tRNALys.3 primer to the PBS results in a complex with altered secondary structure in the region of the PBS, in which the three nt (GAC) at the 5Ј-end of the PBS are looped out and the eight nt just at the 5Ј-end of the PBS bind to other complementary sequences to form a stable stem structure (8,9). Therefore, when extension from the primer reaches the ϩ3 stage (dG), the stem structure has to be disrupted before the fourth nucleotide (dC) can be added, resulting in pausing at the ϩ3 nt position (Fig. 8A). A similar secondary structure is probably formed when the DNA primer, D-Lys.3, is bound to the PBS, such that pausing at the ϩ3 stage also occurs in this circumstance. This hypothesis is supported by our experiments in which a mutated RNA template, HIV/HUA, containing substitutional changes at the 5Ј-end of the PBS (624 -635) was employed (Fig.  2B). In this situation, the stem structure shown in the wt RNA template (Fig. 8A) was disrupted (Fig. 8B). As a consequence, we found that pausing no longer occurred at the ϩ3 position. In addition, reactions that employed the mutated RNA template HIV/del-A (Fig. 2B) paused less frequently at the ϩ3 nt site than did those performed with wild-type RNA template (HIV/ WT). This result might also be attributable to a destabilization of the stem-loop caused by deletion of the A-rich loop, because mutations in this region can cause disturbances in secondary structure of complexes between tRNALys.3 and viral genomic RNA (34). Therefore, template structure may be responsible for the type of pausing seen at the ϩ3 position during initiation of synthesis of (Ϫ) strand DNA.
Our data add significantly to the notion that the initiation of HIV-1 reverse transcription represents a distinct phase in RT reactions (18 -20). First, we have documented that a pause site at the ϩ1 position is rate-limiting. The failure of other groups to have previously detected this site might be due to their use of higher concentrations of dNTPs (e.g. 50 M). Also, the tRNALys.3 employed in our experiments is from human placenta rather than from other animal species. Because all retroviruses use tRNAs as primers, the ϩ1 pausing site may be common to all RTs.
Second, the transition from initiation to elongation probably occurs at the ϩ1 site as well as at the previously observed ϩ3 and ϩ5 positions. On the basis of our studies, the ϩ3 pause site strongly depends on the secondary structure of the viral RNA template. Conceivably, the ϩ1 position may represent the actual point of transition from initiation to elongation, and the ϩ3 pause site may be involved in the arrest of reactions at other early stages. Finally, the A-rich loop, as well as other sequences at the 5Ј-end of the PBS, may be involved in both the efficiency of initiation as well as release from the ϩ1 pause site.
The ϩ1 and ϩ3 pausing events in viral reverse transcription may play a regulatory role similar to that observed in the case FIG. 8. Schematic illustration of the effects of secondary structure between tRNALys.3 and viral RNA template on pausing events during initiation of HIV-1 reverse transcription. A, the three nt (GAC) at the 5Ј-end of the PBS exist as a bulge, whereas the fourth nt (G) is a start site in the stem structure. Therefore, when reactions are extended to the third base, the hydrogen bonds between G-C must be disrupted before the fourth nucleotide (dC) can be incorporated. Consequently, these reactions must pause at the ϩ3 nt position. B, replacement of the sequence 5Ј-AAUCUCUAGCAG-3Ј at the 5Ј-end of the PBS with an irrelevant sequence, i.e. 5Ј-GAACACCCAA-CAUU-3Ј (i.e. HIV/HUA), results in disruption of the stem structure, eliminating the ϩ3 nt pause site. of E. coli RNA polymerase, for which initiation of gene transcription begins in an abortive mode and pauses near the start site, and for which a regulatory subunit, , is required for transcription to proceed to elongation (35,36). In HIV-1, the NC protein may play a role analogous to that of the factor, because NC can help to overcome the ϩ1 pausing event (Fig. 6). NC protein is also able to unwind the secondary structure of template RNA, facilitating the synthesis of long products of reverse transcription (37), and can stimulate the annealing of complementary RNA and/or DNA sequences to stabilize hybrids (38). We believe that NC helps to form and stabilize complexes between primer tRNALys.3 and viral genomic RNA and that the ϩ3 nt pause site is a consequence of the secondary structure of the complex thus formed. In this context, the ϩ3 nt pause site should be observed both when NC protein is present or absent, as observed in the results of Fig. 6.
The role of the A-rich loop in HIV-1 reverse transcription and viral replication is controversial (39). Mutations in the A-rich loop resulted in both severely diminished initiation efficiency of HIV-1 reverse transcription and defective transition from initiation to elongation (Refs. 17 and 18 and this report). However, our use of viral RNA template deleted of the A-rich loop, HIV/ del-A, resulted in only a modest decrease in amount of final (Ϫ) strand DNA product (171 nt) (Fig. 7). Consistently, viruses containing a deletion of the A-rich loop were only moderately impaired in replicative capacity (17). The present studies show that the A-rich loop is responsible for four strong pause sites, ϩ11 through ϩ14, that limit processivity and that deletion of these four As eliminated these pause sites and yielded longer (Ϫ) strand DNA products. Thus, the initial deficit in viral replication caused by deletion of the A-rich loop may have been partially compensated by elimination of the pause sites at positions ϩ11 through ϩ14.