Contacts between Reverse Transcriptase and the Primer Strand Govern the Transition from Initiation to Elongation of HIV-1 Reverse Transcription*

HIV-1 reverse transcriptase (RT) utilizes RNA oligomers to prime DNA synthesis. The initiation of reverse transcription requires specific interactions between HIV-1 RNA, primer tRNA3 Lys , and RT. We have previously shown that extension of an oligodeoxyribonucleotide, a situation that mimicks elongation, is unspecific and differs from initiation by the polymerization rate and dissociation rate of RT from the primer-template complex. Here, we used replication intermediates to analyze the transition from the initiation to the elongation phases. We found that the 2′-hydroxyl group at the 3′ end of tRNA had limited effects on the polymerization and dissociation rate constants. Instead, the polymerization rate increased 3400-fold between addition of the sixth and seventh nucleotide to tRNA3 Lys. The same increase in the polymerization rate was observed when an oligoribonucleotide, but not an oligodeoxyribonucleotide, was used as a primer. In parallel, the dissociation rate of RT from the primer-template complex decreased 30-fold between addition of the 17th and 19th nucleotide to tRNA3 Lys. The polymerization and dissociation rates are most likely governed by interactions of the primer strand with helix αH in the p66 thumb subdomain and the RNase H domain of RT, respectively.

HIV-1 reverse transcriptase (RT) utilizes RNA oligomers to prime DNA synthesis. The initiation of reverse transcription requires specific interactions between HIV-1 RNA, primer tRNA 3 Lys , and RT. We have previously shown that extension of an oligodeoxyribonucleotide, a situation that mimicks elongation, is unspecific and differs from initiation by the polymerization rate and dissociation rate of RT from the primertemplate complex. Here, we used replication intermediates to analyze the transition from the initiation to the elongation phases. We found that the 2-hydroxyl group at the 3 end of tRNA had limited effects on the polymerization and dissociation rate constants. Instead, the polymerization rate increased 3400-fold between addition of the sixth and seventh nucleotide to tRNA 3 Lys . The same increase in the polymerization rate was observed when an oligoribonucleotide, but not an oligodeoxyribonucleotide, was used as a primer. In parallel, the dissociation rate of RT from the primer-template complex decreased 30-fold between addition of the 17th and 19th nucleotide to tRNA 3 Lys . The polymerization and dissociation rates are most likely governed by interactions of the primer strand with helix ␣H in the p66 thumb subdomain and the RNase H domain of RT, respectively.
Reverse transcription is the central event of the retrovirus life cycle (1)(2)(3). Reverse transcriptase (RT) 1 , the viral enzyme that converts the homodimer of single-stranded genomic RNA into double-stranded DNA, is a multifunctional enzyme that possesses RNA-and DNA-dependent DNA polymerase activities, as well as an RNase H function (reviewed in Ref. 3). In addition, complete DNA synthesis requires two strand transfers, which are also mediated by RT (4 -6).
Like most replicative DNA polymerases, RT uses RNA primers to initiate DNA synthesis. RT initiates synthesis of the (Ϫ) strand DNA from the 3Ј end of a specific cellular tRNA species that is complementary to the primer binding site (PBS) of the genomic RNA (reviewed in Refs. [7][8][9]. The primer for (ϩ) strand DNA synthesis is a polypurine tract that is resistant to the RNase H activity of RT (10). Evidence for specific initiation of (Ϫ) strand DNA synthesis in retroviruses and retrotransposons has recently accumulated. Complex species-specific interactions have been identified in vitro between the primer tRNA and genomic RNA of avian retroviruses (11,12), human immunodeficiency virus type 1 (HIV-1) (13,14), and the yeast retrotransposon Ty1 (15). In HIV-1, in addition to the interaction of the 18 3Ј-terminal nucleotides of the replication primer tRNA 3 Lys with the PBS, the anticodon loop, the 3Ј strand of the anticodon stem, and part of the variable loop of the tRNA interact with the genomic RNA (13,14,16). The interaction between the tRNA anticodon loop and HIV-1 RNA is required for optimal replication and reverse transcription in infected cells (17)(18)(19)(20). Post-transcriptional modifications of tRNA 3 Lys are required for stabilizing the primer-template interactions (13,14) and for efficient initiation of reverse transcription (21,22). Extension of tRNA 3 Lys , which corresponds to the initiation phase, requires the homologous primer, template, and RT for efficient polymerization, whereas extension of an oligodeoxyribonucleotide, which mimicks the elongation phase, is unspecific (22)(23)(24)(25).
A detailed comparison of the initiation and elongation complexes of HIV-1 reverse transcription points to a dramatic difference in processivity (26). In contrast to the processive elongation phase, DNA synthesis is distributive during initiation (22,26). The origin of this difference is 2-fold. First, RT dissociates much faster from the initiation complex than from the elongation complex. Second, the polymerization rate is approximately 50-fold faster in the elongation phase compared with initiation of reverse transcription (26).
Whether these differences are due to the presence of the 2Ј-hydroxyl group at the 3Ј end of tRNA 3 Lys at the DNA polymerase catalytic site during the initiation phase or to other differences between the HIV-1 RNA-tRNA 3 Lys ⅐RT and HIV-1 RNA-DNA⅐RT complexes is presently unknown. Similarly, it is not known whether the transition from initiation to elongation is accompanied by a smooth continuous or a sharp discontinuous variation of the polymerization and dissociation rates. Here, we used replication intermediates corresponding to the addition of 1 to 25 deoxyribonucleotides to tRNA 3 Lys to monitor these parameters during addition of the next correct nucleotide by HIV-1 RT. Our results show that the 2Ј hydroxyl group at the 3Ј end of tRNA 3 Lys has limited effects on the polymerization and dissociation rate constants. Instead, the polymerization rate is primarily determined by the nature (RNA versus DNA) of the primer strand contacting the thumb subdomain of RT, whereas dissociation from primer-template is governed by the nature of the primer strand interacting with its RNase H domain.

EXPERIMENTAL PROCEDURES
Template and RT-In all experiments, the template was an RNA corresponding to nucleotides 1-311 of HIV-1 genomic RNA (MAL isolate) synthesized in vitro by transcription with T7 RNA polymerase as described (27). Heterodimeric p66/p51 HIV-1 RT bearing the E478Q mutation abolishing RNase H activity was purified as described previously (28). This RNase H(Ϫ) RT was used instead of the wild type RT to prevent cleavage of the RNA template during formation of tRNA 3 Lys -DNA chimera-HIV-1 RNA⅐HIV-1 RT complexes. This mutant is also devoided of RNase H* activity, which would destroy the tRNA-viral RNA duplex. We previously showed that both RTs are equally efficient in initiating reverse transcription (22). tRNA 3 Lys -DNA Chimera-Natural tRNA 3 Lys was purified from beef liver as described previously (13). Its sequence and post-transcriptional modifications are identical to those of human tRNA 3 Lys . 2 Labeling of tRNA 3 Lys at the 3Ј end with [␣-32 P]ATP was as described (13). Five-to 25-mer oligodeoxyribonucleotides complementary to the viral RNA sequence upstream of the PBS were ligated to the 3Ј end of natural tRNA 3 Lys with T4 DNA ligase using the bridge method (29) (Fig. 1). tRNA 3 Lys with DNA extensions ranging from 1 to 4 nucleotides were synthesized by limited extension of natural tRNA 3 Lys with the Klenow fragment of the Escherichia coli DNA polymerase I using the appropriate oligodeoxyribonucleotide templates and deoxyribonucleoside triphosphate mixtures ( Fig. 1) (21). tRNA 3 Lys -DNA chimeras were precipitated and purified on polyacrylamide denaturing gel.
Kinetic Experiments-Template RNA was hybridized to 32 P-labeled primer as described previously (13). Primer-template (ϳ200 pM) and excess RT (20 nM, unless otherwise stated) were preincubated for 4 min at 37°C in Tris-HCl, pH 8.0, 50 mM KCl, 6 mM MgCl 2 , and 1 mM dithioerythritol. In the first series of experiments, reactions were started by adding the four dNTPs at a final concentration of 50 M together with poly(rA)-(dT) 15 , which had been previously annealed at 70°C for 20 min at a 10:1 ratio (w/w), at a final concentration of 1.66 M of (dT) 15 . Reactions were stopped at increasing time intervals by the addition of EDTA and formamide.
In the next experiments, reverse transcription was initiated by adding the first complementary dNTP at a final concentration of 50 M.
Slow reactions were stopped at increasing time intervals by adding formamide containing 50 mM EDTA to aliquots of the reaction mixture. To follow fast reactions, we used a home-built quenched-flow apparatus that allowed determination of a complete reaction curve in one stroke and reaction times ranging from 2 to 450 ms (30). In these experiments, reactions were stopped by the addition of EDTA and sodium acetate at 25 and 750 mM final concentrations, respectively. Samples were precipitated and redissolved in formamide before analysis by denaturing polyacrylamide gel electrophoresis (26).
Dissociation Rate of Primer-Template⅐RT Complex-To measure dissociation of RT from the ternary complex, the primer-template⅐RT complex was pre-formed as described above. After 4 min, 1 M poly(rA)-(dT) 18 was added to trap RT that dissociated from the ternary complex. Aliquots were removed at increasing time intervals after addition of the trap and mixed with the first complementary dNTP, to extend the remaining ternary complex. Extension was stopped after 90 s with EDTA and formamide. The efficiency of poly(rA)-(dT) 18 to trap free RT was controlled by checking that primer extension did not significantly increase from 15 s to 15 min.
Data Analysis-In all experiments, the reaction products were analyzed on 8% denaturing polyacrylamide gels and quantified with a BioImager BAS 2000 (Fuji) using the whole band analyzer software (Bio Image). Curve-fitting was performed with the SigmaPlot software (Tandel Scientific Software).

RESULTS
Experimental Strategy-We previously showed that the complex formed upon binding of RT to the tRNA 3 Lys -HIV-1 RNA complex was an initiation complex, whereas extension of an 18-mer oligodeoxyribonucleotide bound to the PBS of HIV-1 RNA started in the elongation mode, without any initiation phase (22,26). Detailed analysis of the initiation and elongation complexes revealed significant differences in the polymerization rate (k pol ) of the preformed primer-template⅐RT complexes and in the dissociation rate (k off ) of RT (26). Here, to study the transition from initiation to elongation, we constructed replication intermediates (tRNA 3 Lys ϩ n) corresponding to the addition of n ϭ 1 to 8, 11, 14, 16 to 19, and 25 deoxyribonucleotides to the 3Ј end of 32 P-labeled tRNA 3 Lys (see "Experimental Procedures") ( Fig. 1). These extended tRNAs 2 G. Keith, unpublished results. were annealed to an RNA corresponding to nucleotides 1-311 of the HIV-1 genomic RNA, which contained the PBS (nucleotides 179 -196). We first studied reverse transcription in the presence of the four dNTPs during a single binding event of RT to various tRNA 3 Lys ϩ n-HIV-1 RNA complexes. These experiments allowed us to qualitatively appreciate the magnitude and location of the transition between the initiation and elongation phases of HIV-1 reverse transcription.
Next, for each tRNA 3 Lys ϩ n intermediate, we determined k pol for incorporation of the next nucleotide and k off for dissociation of RT. These parameters could not be obtained directly by monitoring the kinetics of tRNA 3 Lys extension in the presence of the four deoxyribonucleotides triphosphates. It might be argued that the k off and k pol values measured for addition of a single nucleotide to the replication intermediates could differ from those during processive elongation. However, the addition of the first five nucleotides to tRNA 3 Lys is known to be distributive (22). Furthermore, when the primer is DNA, dissociation of RT from the primer-template before polymerization or after nucleotide incorporation takes place essentially at the same rate (26,31,32).
Extension of Replication Intermediates in the Presence of the Four dNTPs during a Single RT Turnover-In the first set of experiments, we monitored extension of tRNA 3 Lys , tRNA 3 Lys ϩ 2, tRNA 3 Lys ϩ 5, tRNA 3 Lys ϩ 8, and tRNA 3 Lys ϩ 18 during a single RT binding event (Fig. 2). For this goal, excess poly(rA)-(dT) 15 was added together with dNTPs to trap the enzyme dissociating from the tRNA 3 Lys ϩ n-HIV-1 RNA complexes. In agreement with our previous studies (22), only short extension products, which corresponded to addition of 3, 4 (faint band), or 5 nucleotides, were observed when the primer was tRNA 3 Lys (Fig. 2, lanes 1). The product pattern did not significantly change dur-ing the time interval examined here, reflecting distributive DNA synthesis and rapid dissociation of RT from the primertemplate complex. Similarly, extension of tRNA 3 Lys ϩ 2 was limited to the addition of 1, 2, or 3 nucleotides, yielding tRNA 3 Lys ϩ 3, tRNA 3 Lys ϩ 4, and tRNA 3 Lys ϩ 5, respectively ( Fig.  2, lanes 2). Almost no extension could be detected when tRNA 3 Lys ϩ 5 was the primer (Fig. 2, lanes 3). By contrast, extension of tRNA 3 Lys ϩ 8 and tRNA 3 Lys ϩ 18 gave very long products. Moreover, a clear time dependence of the product length was apparent. Minus strand strong stop DNA appeared after 5 min of reaction, indicating slow dissociation of RT and processive DNA synthesis (Fig. 2, lanes 4 and  5). Thus, these results clearly localized the transition between initiation of reverse transcription, i.e. the distributive mode of DNA synthesis (22,26), and elongation, i.e. the processive mode of DNA synthesis (22,26), between addition of the sixth and ninth nucleotides to tRNA 3 Lys . Likewise, more (Ϫ) strand strong stop DNA was produced from tRNA 3 Lys ϩ 18 than from tRNA 3 Lys ϩ 8, suggesting slower dissociation of RT from the longer primer.
Addition of the First Six Nucleotides to tRNA 3 Lys Is Slow-The above experiments showed that primer extension and dissociation of RT from the tRNA 3 Lys ϩ n-HIV-1 RNA complexes take place on the same time scale during initiation of reverse transcription. Thus, determination of the authentic polymerization rate constant requires working with a large excess of RT over primer-template, especially since, due to the limited amount of available tRNA-DNA chimera, the primer-template concentration was below the equilibrium dissociation constant of the primer-template⅐RT complex (K d ϳ 3 nM) (26).
The addition of the first six nucleotides to tRNA 3 Lys (i.e. extension of tRNA 3 Lys ϩ n chimera with 0 Յ n Յ 5) was slow, occurring in the second to minute time scale (Fig. 3, A, B, and D). The addition of the first, second, third, fifth, and sixth nucleotides occurred at similar rates, whereas the addition of the fourth was one order of magnitude slower (Fig. 3). As previously observed for addition of the first nucleotide (26), addition of the second to the sixth nucleotides to tRNA 3 Lys was biexponential (Fig. 3B). The fast phase (k pol fast ) corresponded to the authentic polymerization rate (k pol ), whereas the slow one (k pol slow ) was attributed to a rate-limiting conformational change of the primer-template (26). The k pol fast values for addition of the second, third, fifth, and sixth nucleotides to the tRNA 3 Lys were in the range of 0.20 to 0.50 s Ϫ1 (Fig. 3D). The previously determined rate of addition of the first nucleotide (0.22 s Ϫ1 ) was in the same range (26).
The ϩ3 position also corresponded to a change in the relative amplitude of the fast phase ( Fig. 3B and data not shown). During the addition of the first (26), second, and third nucleotides, the fast phase represented more than half of the overall reaction amplitude. In contrast, this phase contributed only 16 to 19% of the overall extension of tRNA 3 Lys ϩ 3, tRNA 3 Lys ϩ 4, and tRNA 3 Lys ϩ 5 (Fig. 3B). As will be discussed below, variations of the fast phase amplitude as well as the very low rate of extension of tRNA 3 Lys ϩ 3 could be accounted for by the secondary and tertiary structure of the tRNA 3 Lys -HIV-1 RNA complex. Addition of the 7th to the 26th Nucleotide to tRNA 3 Lys Is Fast-A dramatic increase of the extension rate was observed starting with tRNA 3 Lys ϩ 6 ( Fig. 3, A and C). The addition of the 7th to the 26th nucleotide was complete in less than 1 s, thus requiring a quenched-flow apparatus to determine the polymerization rate constants (Fig. 3, A, C, and D). For n ϭ 7, 8, 11, 14, and 20, the time course extension of the tRNA 3 Lys ϩ n primer could be fitted with a single exponential. Similarly, extension by HIV-1 RT of an oligodeoxyribonucleotide annealed to a DNA or RNA template usually followed first order kinetics  (26,(31)(32)(33)(34)(35). The extension rate constant of these tRNA 3 Lys ϩ n chimera ranged from 9.7 to 112 s Ϫ1 (Fig. 3, C and D). Again, extension of oligodeoxyribonucleotide primers occurred at similar rates when HIV-1 RT was not blocked by stable secondary structures (26,(31)(32)(33)(34)(35)(36).
Unexpectedly, a satisfactory fit of the time course extension of tRNA 3 Lys ϩ n chimera with n ϭ 6, 18, 19, and 25 required a sum of two first order processes (Fig. 3C). In these cases, the rate constant of the fast process (k pol fast ) ranged from 122 to 750 s Ϫ1 , whereas that of the slow phase (k pol slow ) was between 4.3 and 17.2 s Ϫ1 (Fig. 3D). Since the k pol slow values are at least one order of magnitude higher than the dissociation rate constant (k off ) of RT from the tRNA 3 Lys ϩ n-viral RNA complex (see below), the origin of the slow component of the reaction could not be in multiple RT turnover. Instead, the biphasic nature of the kinetics probably reflects heterogeneity of the tRNA 3 Lys ϩ n-viral RNA⅐HIV-1 RT complexes. It is noteworthy that the k pol slow values measured here were 10-to 250-fold higher than the slow process associated with DNA extension by HIV-1 RT through a stable secondary structure (36). Thus, both k pol slow and k pol fast probably are authentic polymerization rate constants. Indeed, the k pol slow and k pol fast values correspond to the lower and higher limits, respectively, of the k pol of those tRNA 3 Lys ϩ n chimeras whose extension was monoexponential (Fig. 3D). Remarkably, the highest k pol fast value, observed with tRNA 3 Lys ϩ 6, was one order of magnitude higher than the highest k pol values previously determined for HIV-1 RT (31,33,36). This value corresponded to the technical limits of the quenched flow technique (30).
Thus, the polymerization rate increased by 3.5 orders of magnitude between the addition of the sixth and seventh nucleotides (i.e. between extension of tRNA 3 Lys ϩ 5 and tRNA 3 Lys ϩ 6). The mean k pol fast value for addition of the first six nucleotides was 0.24 s Ϫ1 , whereas the mean polymerization rate constant of the seventh to the 26th nucleotides was 110 s Ϫ1 (taking into account both k slow and k fast for biphasic kinetics) (Fig. 3D).

The k pol Transition Does Not Require the Extended tRNA 3 Lys -HIV-1 RNA Interactions-Several lines of evidence support the existence of complex intermolecular interactions between tRNA 3
Lys and HIV-1 RNA, in addition to the annealing of the 18 3Ј-terminal nucleotides of the tRNA to the PBS (13, 14, 16, 18 -20, 37). To test the possible involvement of these interactions in the sharp k pol transition, we compared the kinetics of addition of the sixth and seventh nucleotide to tRNA 3 Lys and 18-mer oligoribonucleotides (ORN) and oligodeoxyribonucleotides (ODN) complementary to the PBS. The additions of these nucleotides to ORN and tRNA 3 Lys were very similar (Table I). Both kinetics were biexponential, and the rate constants of the fast phase increased by 3500-fold between the addition of the sixth and seventh nucleotides to ORN, which is very close to the 3400-fold increase observed with the natural primer. On the other hand, the addition of the sixth and seventh nucleotides to ODN were monoexponential, and addition of the two nucleotides to this primer occurred at the same rate (Table I). Thus, no transition in k pol was observed when an ODN was used as primer.
A Transition in the RT Dissociation Rate Constant-To measure the dissociation rate constant (k off ) of RT from the tRNA 3 Lys ϩ n-HIV-1 RNA complexes, we used excess poly(rA)-(dT) 18 to trap the dissociating RT. This technique was previously used to determine the k off of RT from the unextended tRNA 3 Lys -HIV-1 RNA complex (26). When n ϭ 0 or 5, dissociation of polymerase from the primer-template complexes followed monoexponential kinetics (Fig. 4A). However, with the other tRNA 3 Lys ϩ n chimeras, the ternary complex concentration followed a biexponential decay (Fig. 4). Biphasic decay of the preformed primer-template⅐polymerase complexes is rather frequent and has been observed with HIV-1 RT (38, 39) as well as other DNA polymerases (40). In our case, the relative amplitudes of the two phases and their rate constants did not vary with the concentration of HIV-1 RT or poly(rA)-(dT) 18 (data not shown). Thus, our results might reflect the existence of two interconvertible forms of the preformed ternary complexes, as has been postulated in the case of E. coli DNA polymerase I (41). Alternatively, they might be due to heterogeneity in primer-template complex or in the RT preparation. However, since the relative amplitude of the fast dissociation of RT from the tRNA 3 Lys ϩ n-HIV-1 RNA complexes strongly and monotonously decreased with increasing n values (Fig. 4B), the latter explanation seems unlikely.
When comparing the different tRNA 3 Lys ϩ n-HIV-1 RNA complexes, the rate constant of the fast dissociation phase (k off fast ) varied within two orders of magnitude (Fig. 4B). For small n values (0 Յ n Յ 6), rather large fluctuations of k off fast were observed from one to the next complex (the k off fast value could not be determined for n ϭ 3 because of the very low polymerization rate.). However, no general trend was observed until a pronounced k off fast decrease took place when n increased from 16 to 18 (Fig. 4A). The mean k off fast value for 0 Յ n Յ 16 was 0.21 s Ϫ1 , which corresponded to the previously determined dissociation rate constant of HIV-1 RT from the tRNA 3 Lys -HIV-1 RNA complex (Fig. 4, A and C) (26). By comparison, the mean k off fast value dropped to 0.0078 s Ϫ1 when 18 Յ n Յ 25 (Fig. 4C). This value was higher than the dissociation rate of HIV-1 RT from the HIV-1 RNA primed with an 18-mer DNA complementary to the PBS (26) (0.001 s Ϫ1 ) but was comparable with the fast dissociation phase of HIV-1 RT from poly(rA)-(dT) 16 (39).
When looking at the slow dissociation phase, large fluctuations of k off slow were also found for small n values (Fig. 4D). However, fluctuations in k off fast and k off slow were not parallel. For n increasing from 1 to 5, the decrease in k off fast was concomitant with a k off slow increase. The only feature common to k off slow and k off fast was a marked decrease in the dissociation rates with n increasing from 16 to 19 (Fig. 4D).

DISCUSSION
We previously showed that the initiation and elongation phases of HIV-1 reverse transcription markedly differ in processivity (22,26). Here, to obtain a detailed view of the transition between these two phases, we have constructed a complete set of replication intermediates consisting of natural tRNA 3 Lys with an increasing number of deoxynucleotides attached to its 3Ј end. In a first set of experiments, we showed that the transition between the initiation and elongation phases of HIV-1 reverse transcription, which is accompanied by a dramatic increase in the processivity of DNA synthesis, takes place between the addition of the sixth and the ninth nucleotide to tRNA 3 Lys . In the next experiments, we determined the kinetic parameters of addition of the first 26 nucleotides to tRNA 3 Lys and of dissociation of RT from the corresponding primer-template complexes.
Rate of Nucleotide Addition-The k pol and k pol fast values for the addition of the second, third, fifth, and sixth nucleotide to tRNA 3 Lys are in the range of 0.20 to 0.50 s Ϫ1 , close to the polymerization rate of the first nucleotide that we previously determined (0.22 s Ϫ1 ) (26). Thus, our previous and present work shows that the 2Ј-hydroxyl group at the 3Ј end of tRNA 3 Lys , which is present in the catalytic site of HIV-1 RT during addition of the first deoxyribonucleotide but absent in that site during the subsequent nucleotide additions, is not the cause of the slow extension rate of tRNA 3 Lys . The extremely low k pol fast value for the addition of the fourth nucleotide could be accounted for by the secondary and tertiary structures of the tRNA 3 Lys -HIV-1 RNA complex. As determined from chemical and enzymatic probing experiments, the first 3 nucleotides immediately upstream of the PBS are singlestranded, whereas the 4th to 17th nucleotides are base-paired ( Fig. 1) (13, 14). Due to strong structural constraints in the tRNA 3 Lys -HIV-1 RNA complex (14,16), the viral RNA makes a sharp turn between the third and fourth nucleotides upstream of the PBS. 3 This distortion of the template probably causes the slow extension of tRNA 3 Lys ϩ 3. In turn, this low k pol fast value explains the strong pausing observed three nucleotides upstream of the PBS during synthesis of (Ϫ) strand strong stop DNA (21,22). The structure of the tRNA 3 Lys -HIV-1 RNA complex also explains variations of the amplitude of the fast phase during the addition of the first six nucleotides to tRNA 3 Lys . When the incoming nucleotide is complementary to a singlestranded template nucleotide (i.e. addition of the first three nucleotides), the fast phase is predominant. On the contrary, when the incoming nucleotide is complementary to a template nucleotide that is base-paired in the primer-template complex, k pol fast is not affected, but the amplitude of this phase is significantly decreased. A similar phenomenon was recently observed during elongation of an oligodeoxyribonucleotide primer by HIV-1 RT through a stable stem-loop structure formed by an RNA template (36).
A dramatic increase of the polymerization rate was observed between the addition of the sixth and seventh nucleotides. Unexpectedly, extension of some tRNA 3 Lys ϩ n chimeras follows biphasic kinetics. The most likely explanation for this observation is that some tRNA 3 Lys -HIV-1 RNA complexes might bind in two slightly different modes to the primer-template binding cleft of RT. Biphasic kinetics have not been observed by others during extension of DNA primers (31)(32)(33)(34)(35). However, the number of experimental points collected during the pre-steady state in some of these studies might not be sufficient to detect the superposition of two first order processes, which we distinguished here by collecting up to 30 data points for a single kinetic experiment. Thus, one cannot exclude that biphasic DNA extension is rather common.
As revealed by the comparison of the kinetics of addition of the sixth and seventh nucleotides to tRNA 3 Lys , ORN, and ODN, the extended interactions between tRNA 3 Lys and the viral RNA (13, 14, 16, 18 -20, 37) do not contribute toward the dramatic increase of the polymerization rate between addition of these nucleotides. Indeed, this sharp transition is observed when the 18 nucleotides complementary to the PBS are ribonucleotides (i.e. with tRNA 3 Lys and ORN as primers) but not with the ODN primer. Our previous data suggested that the extended interactions favored extension of the pausing products at positions ϩ3 and ϩ5 during synthesis of (Ϫ) strand strong stop DNA (22). However, these interactions favor formation of the tRNA 3 Lys ϩ n-HIV-1 RNA⅐RT complexes rather than affecting the polymerization rate by the preformed ternary complexes (22,26). This effect was not detected in the present study because of the large excess of RT over primer-template complexes.
Dissociation of HIV-1 RT from the tRNA 3 Lys ϩ n-HIV-1 RNA Complexes-As also observed for extension of some tRNA 3 Lys ϩ n chimeras, dissociation of HIV-1 RT from most of the tRNA 3 Lys ϩ n-HIV-1 RNA complexes was biphasic. The biphasic nature of polymerization (for n Ն 6) as well as that of the dissociation of RT from the primer-template most probably reflects different binding conformations. However, no correlation exists between the relative amplitudes of the fast and slow phases of primer extension and RT dissociation. Thus, our results suggest the existence of several slightly different binding modes of RT to the tRNA 3 Lys ϩ n-HIV-1 RNA complexes. Moreover, they indicate that the rates of primer extension and RT dissociation are not governed by the same RT/nucleic acid contacts.
RT dissociates 200-fold faster from the initiation complex than from the elongation complex (26). Here, we observed that HIV-1 RT dissociates slightly faster from the tRNA 3 Lys ϩ 1-HIV-1 RNA complex than from the tRNA 3 Lys -HIV-1 RNA complex. Thus, fast dissociation of RT from the initiation complex is not due to the presence in the catalytic site of the 2Ј-OH group of the 3Ј-terminal adenosine of tRNA 3 Lys , since this group is displaced from the catalytic site during subsequent nucleotide additions. Regarding the kinetic stability of the tRNA 3 Lys ϩ n-HIV-1 RNA⅐HIV-1 RT complexes, the only feature that can be related to the transition from the initiation to the elongation phases of reverse transcription is a decreased dissociation rate as n increased from 16 to 18. This transition was the only feature common to the fast and slow dissociation phases of RT.
Correlation with RT Structure-We observed a sharp k pol transition between the addition of the sixth and seventh nucleotides when using tRNA 3 Lys and ORN, but not ODN, as primers. Thus, the highly ordered tRNA 3 Lys -HIV-1 RNA complex (14,16) is not required for this transition to occur. Rather, a crucial feature of this transition is an RNA-DNA junction on the primer strand. The structure of HIV-1 RT complexed with an 18-mer DNA primer-19-mer DNA template has been solved by x-ray crystallography (42,43). In this structure, the five base pairs of the primer-template proximal to the polymerase active site adopt A-form geometry, whereas the remaining base pairs adopt B helical form (42,43) (Fig. 5). Interestingly, this structure reveals that the fifth nucleotide and phosphate upstream of the active site on the primer strand are in close contact with the central part of helix ␣H located in the thumb subdomain of the p66 subunit (Fig. 5). Molecular dynamics modeling based on the apoHIV-1 RT (44) and the RT-DNA co-crystal structures (43) suggests that the side chain of Gln-258 makes van der Waals contacts with the primer strand backbone and sugars at the fifth and sixth residues (45). These structures indicate that, when the tRNA 3 Lys ϩ 5-HIV-1 RNA duplex occupies the nucleic acid binding cleft, the 3Ј-terminal adenosine of tRNA 3 Lys and the phosphate at the tRNA-DNA junction are in the immediate vicinity of the central portion of helix ␣H, whereas the first deoxyribonucleotide attached to tRNA 3 Lys occupies this position during addition of the seventh deoxyribonucleotide. The crucial role of this helix in binding DNA primer-template complexes has been clearly demonstrated by site-directed mutagenesis (45)(46)(47). Therefore, the sharp k pol transition is likely dictated by the position of RNA-DNA junction on the primer strand relative to helix ␣H.
In the crystal structure, the A and B helical regions of the DNA primer-DNA template complex are separated by a 41°b end (42,43). During (Ϫ) strand DNA synthesis, the primertemplate complex cannot adopt B-form geometry because of the RNA template. The hybrid formed by the PBS and the 3Ј terminus of tRNA 3 Lys probably adopts classical A-form geometry, whereas the RNA template upstream of the PBS and the complementary deoxyribonucleotides linked to tRNA 3 Lys assumes H-form conformation, in which the template strand has essentially A-form conformation and the primer strand has an intermediate geometry between A-and B-form (48,49). Thus, different interactions of helix ␣H in the thumb of the p66 subunit of HIV-1 RT with A-and H-form geometry primertemplates probably lead to slightly different locations of the primer 3Ј-hydroxyl group in the DNA polymerase catalytic site, which in turn is responsible for the large variation of the polymerization rate.
In parallel with the k pol transition, we observed a decreased dissociation rate of RT from the tRNA 3 Lys ϩ n-HIV-1 RNA complexes when n increased from 16 to 18. In the crystal structure of HIV-1 RT complexed with an 18-mer DNA primer-19-mer DNA template, the phosphate groups 15 to 17 nucleotides upstream from the 3Ј end of the primer contact the RNase H domain of the p66 subunit (Fig. 5) (39,42,43). Contacts between the primer and the RNase H domain of RT are known to be crucial for the stability of the primer-template⅐RT complex. For instance, the equilibrium dissociation constant of HIV-1 RT from poly(rA)-(dT) n decreased by 2 orders of magnitude when n increased from 10 -14 to 16 -20 (50). Thus, our results, together with the x-ray structure (42,43), suggest the lifetime of the RT⅐tRNA 3 Lys ϩ n-HIV-1 RNA complexes increases when the region of the primer strand close to the RNase H domain of RT is DNA. Alternatively, the interaction of a DNA-RNA primer-template instead of an RNA-RNA duplex with the RNase H domain might favor contacts between helix ␣EЈand the ␤5Ј-␣EЈconnecting loop and the template strand around position Ϫ20 relative to the polymerase active site (51,52).
Concluding Remarks-The sharp transition between initiation and elongation of HIV-1 reverse transcription is mainly characterized by a 3400-fold increase of the polymerization rate between the addition of the sixth and seventh nucleotides. The processivity of the elongation phase is further increased by a reduction in the dissociation rate of RT between the addition of the 17th and 19th nucleotides. The increase in k pol and decrease in k off are both dictated by different contacts of the primer strand of RNA-RNA and DNA-RNA primer-template complexes in the binding cleft of HIV-1 RT. Given the conservation of the essential catalytic residues in the DNA polymerase active site (53)(54)(55)(56) and of the interaction between the helical thumb subdomain with the minor groove of the primertemplate complex (43,(57)(58)(59)(60), the slow extension rate of RNA primers might be a general feature of DNA polymerases.