HIV-1 nucleocapsid protein and the secondary structure of the binary complex formed between tRNA(Lys.3) and viral RNA template play different roles during initiation of (-) strand DNA reverse transcription.

In human immunodeficiency virus type 1 (HIV-1), the tRNA(Lys.3) primer and viral RNA template can form a specific complex that is characterized by extensive inter- and intramolecular interactions. Initiation of reverse transcription from this complex has been shown to be distinguished from subsequent elongation by early pausing events, such as at the +1 and +3 nucleotide positions. One major concern regarding the biological relevance of these results is that most kinetic studies of HIV-1 reverse transcription have been performed using tRNA(Lys.3)-viral (v) RNA complexes that were formed by heat annealing. In contrast, tRNA(Lys.3) in viruses is placed onto the primer binding site by nucleocapsid (NC) sequences of the Gag protein. In this study, we have further characterized the initiation features of reverse transcription in the presence of HIV-1 NC protein. In contrast to results obtained with a heat-annealed tRNA(Lys.3).vRNA complex, we found that polymerization reactions catalyzed by HIV-1 reverse transcriptase did not commonly pause at the +1 nucleotide position when a NC-annealed RNA complex was used, and that this was true regardless whether NC was actually still present during reverse transcription. This activity of NC required both zinc finger motifs, as demonstrated by experiments that employed zinc finger-mutated forms of NC protein (H23C NC and ddNC), supporting the involvement of the zinc fingers in the RNA chaperone activity of NC. However, NC was not able to help reverse transcriptase to escape the +3 pausing event. Mutagenesis of a stem structure within the tRNA(Lys.3). vRNA complex led to disappearance of the +3 pausing event as well as to significantly reduced rates of reverse transcription. Thus, this stem structure is essential for optimal reverse transcription, despite its role in promotion of the +3 pausing event.

In human immunodeficiency virus type 1 (HIV-1), the tRNA Lys.3 primer and viral RNA template can form a specific complex that is characterized by extensive inter-and intramolecular interactions. Initiation of reverse transcription from this complex has been shown to be distinguished from subsequent elongation by early pausing events, such as at the ؉1 and ؉3 nucleotide positions. One major concern regarding the biological relevance of these results is that most kinetic studies of HIV-1 reverse transcription have been performed using tRNA Lys.3 -viral (v) RNA complexes that were formed by heat annealing. In contrast, tRNA Lys. 3

in viruses is placed onto the primer binding site by nucleocapsid (NC) sequences of the Gag protein. In this study, we have further characterized the initiation features of reverse transcription in the presence of HIV-1 NC protein.
In contrast to results obtained with a heat-annealed tRNA Lys. 3 ⅐vRNA complex, we found that polymerization reactions catalyzed by HIV-1 reverse transcriptase did not commonly pause at the ؉1 nucleotide position when a NC-annealed RNA complex was used, and that this was true regardless whether NC was actually still present during reverse transcription. This activity of NC required both zinc finger motifs, as demonstrated by experiments that employed zinc finger-mutated forms of NC protein (H23C NC and ddNC), supporting the involvement of the zinc fingers in the RNA chaperone activity of NC. However, NC was not able to help reverse transcriptase to escape the ؉3 pausing event. Mutagenesis of a stem structure within the tRNA Lys. 3 . vRNA complex led to disappearance of the ؉3 pausing event as well as to significantly reduced rates of reverse transcription. Thus, this stem structure is essential for optimal reverse transcription, despite its role in promotion of the ؉3 pausing event.
HIV-1 1 reverse transcription is initiated as its cognate primer tRNA Lys.3 is annealed onto the primer binding site (PBS) of viral RNA template (vRNA). In addition to base pairing between the PBS sequence and the 3Ј-terminal region of tRNA Lys. 3 , the resulting tRNA Lys.3 ⅐vRNA binary complex is also characterized by other extensive but specific inter-and intramolecular interactions (1)(2)(3). The initiation of (Ϫ) strand DNA synthesis from this complex represents a specific stage, during which reverse transcriptase (RT) shows distinct binding and kinetic properties from those of the subsequent elongation stage (4). Using an in vitro reaction system, initiation of reverse transcription can be detected by the formation of short intermediate cDNA products after the tRNA primer is extended by 1, 3, or 5 nt (4 -7). However, in these studies, reverse transcription had been performed with tRNA Lys.3 ⅐vRNA complexes that were formed by heat annealing. In the virus, however, the tRNA primer is placed onto the viral RNA template by nucleocapsid (NC) sequences of the Gag protein. Therefore, it is important to compare the kinetic features of different RNA complexes that have been formed by either heat or NC annealing.
The HIV-1 NC is a small basic protein that can bind to single-stranded nucleic acid. It has two zinc finger motifs, each of which contains zinc ion binding residues, i.e. CCHC, and forms a tight, rigid loop within the protein (8). NC protein exhibits nucleic acid chaperone activity in vitro; it catalyzes the rearrangement of nucleic acid molecules into a thermodynamically stable conformation (9,10). This activity enables NC to participate in reverse transcription at many steps and ensures that highly specific and efficient viral cDNA synthesis will occur. For example, the overall efficiency of negative strand strong-stop DNA ((Ϫ) ssDNA) synthesis can be increased in the presence of NC. NC was found to function in this process in two manners, i.e. by facilitating the formation of an active form of tRNA Lys.3 ⅐vRNA complex (11), or by transiently eliminating template secondary structures, at which the RT enzyme stalls (12)(13)(14)(15).
To assess the roles of NC protein in initiation of reverse transcription, we employed NC to perform primer placement, and then removed it from the system by proteinase K digestion and phenol:chloroform treatment. This system allows reverse transcription to proceed from a NC-derived tRNA Lys.3 ⅐vRNA complex, while avoiding the direct involvement of NC in the reactions at the same time. These results showed that the tRNA Lys.3 ⅐vRNA complex that is formed by NC is already active, and can overcome ϩ1 nt pausing during subsequent polymerization without the further involvement of NC. Therefore, NC, acting as an RNA chaperone, assists in the formation of a more functional RNA complex rather than a more thermostable one, as is obtained by heat annealing of tRNA primer onto the PBS. The zinc finger motifs of NC were found to be important in this regard, as indicated by experiments using H23C NC, which includes a point mutation within the first zinc finger structure, and ddNC, in which both zinc fingers are replaced by Gly-Gly linkages.
We further studied the causes of the ϩ3 pausing event either with or without NC protein in the reactions. To pursue this subject, systematic mutagenesis studies were performed to disrupt a stem structure that is formed by RNA template sequences within the tRNA Lys.3 ⅐vRNA complex. Finally, we have compared early pausing patterns of reverse transcription from tRNA Lys.3 ⅐vRNA complexes derived by heat annealing, NCmediated annealing, and within virions. Our results show that, in contrast to the tRNA Lys.3 ⅐vRNA complexes formed by heat annealing, the RNA complexes that are either derived from virions or formed by NC annealing barely pause at the ϩ1 stage during initiation of reverse transcription.

MATERIALS AND METHODS
Construction of Plasmids-The primers used in our mutagenesis studies are listed in Table I. The mutated HIV-1 RNA templates designated N1-N10 were generated by PCR using primer pairs Bgl-S/ (Nar-A1 to Nar-A10). After digestion with BglII and NarI, the PCR products were inserted into an RNA transcription vector PBS/wild-type that contains a HIV-1 DNA sequence between nt positions 473 and 1417 (16). Because the mutated sequences of N2 and N6 include a BglII restriction site, a partial digestion method was employed for their construction. The N1, N3, N4, N5, and N7 mutations were also introduced into BH10 using Hpa-S/(Nar-A1, Nar-A3, Nar-A4, Nar-A5, and Nar-A7) as primers that include HpaI and NarI restriction sites.
Preparation of Synthetic NC and RNA Template-A series of HIV-1 NC peptides, including the wild-type form of 72 amino acids as well as H23C NC and ddNC, containing mutated zinc fingers, was prepared by solid phase chemical synthesis as described elsewhere (17). RNA transcription plasmids were linearized by BssHII, and used as template in a Mega-Scripts kit (Ambion, Austin, TX) to produce RNA transcripts. The integrity of the RNA transcripts was routinely checked on 5% polyacrylamide gels containing 7 M urea prior to use in reverse transcription reactions.
Viral RNA Isolation-COS-7 cells were grown in Dulbecco's modified Eagle's medium, supplemented with 10% fetal calf serum, and were transfected with HIV-1 DNA constructs by the calcium phosphate method. Forty-eight hours after transfection, culture supernatants (20 ml) were clarified in a Beckman GS-6R centrifuge at 3,000 rpm for 30 min at 4°C and virus particles were then pelleted through a 20% sucrose cushion at 40,000 rpm for 1 h at 4°C, using an SW41 rotor in a Beckman L8-M ultracentrifuge. Virus pellets were suspended in TN buffer, and a 2-l portion was removed for CA-p24 determination by enzyme-linked immunosorbent assay (Abbott Laboratories, Abbott Park, IL). Viral RNA was extracted from viral pellets by Trizol reagent (Life Technologies, Inc., Montreal, Quebec, Canada), and dissolved in diethyl pyrocarbonate-treated double distilled water. Viral RNA equivalent to 200 ng of CA-p24 was used in the subsequent reverse transcription reactions.
In Vitro Reverse Transcription-1 pmol of tRNA Lys.3 that was prepared from human placenta (18) was annealed onto 1 pmol of RNA template by incubation at 37°C for 1 h with a calculated saturating level of NC (i.e. 30 pmol, such that the molar ratio of NC and total number of nucleotides in the template was about 1:8, based on a template of 251 nt in length) in a 10-l reaction mixture containing 50 mM Tris-HCl (pH 7.2), 50 mM KC1, 5 mM MgC1 2 . Alternatively, the placement of tRNA Lys.3 onto RNA template was performed using the same buffer conditions by denaturation of RNA molecules at 85°C for 5 min and annealing at 55°C for 10 min. To determine whether NC protein functioned during initiation of reverse transcription at the primer placement or at the primer extension phase, proteinase K digestion and phenol:chloroform extraction were performed in some cases (11). In the case of RNA isolated from virions, tRNA Lys.3 is already naturally annealed onto the viral RNA template (19); therefore, these tRNA Lys.3 ⅐vRNA complexes were directly subjected to the following in vitro reverse transcription reactions.
Primer tRNA was extended by reverse transcriptase in a volume of 20 l containing 50 mM Tris-HCl (pH 7.2), 50 mM KCl, 5 mM MgC1 2 , 10 mM dithiothreitol, 10 units of RNA-guard (Amersham, Pharmacia Biotech, Montreal, Quebec, Canada), and 160 nM dNTPs at 37°C for 15 min, unless otherwise specified; thereafter, reverse transcription reactions were terminated by adding EDTA (pH 8.0) to a final concentration of 50 mM. The cDNA products were fractionated on 8% denaturing polyacrylamide gels containing 7 M urea. The RTs used in this study were prepared as described (20) and included wild-type HIV-1 enzyme (p66/51), mutated HIV-1 RT containing a mutation at codon 89 (i.e. E89G), or mutated HIV-1 RT containing a mutation at codon 184 (i.e. M184V). An amount of 45 ng of wild-type RT or E89G RT was used in these reactions unless specified. Because M184V RT displays reduced processivity during reverse transcription (21,22), a higher amount, i.e. 250 ng of M184V RT, was employed to achieve levels of cDNA synthesis similar to those attained with wild-type enzyme (data not shown).
Western Blot-Various concentrations of NC were used in combination with 1 pmol of template RNA and 1 pmol of primer tRNA Lys.3 at 37°C for 1 h to promote annealing. Regardless of whether samples were subjected to proteinase K and phenol:chloroform treatment, they were fractionated on 12% (w/v) SDS-polyacrylamide gels and transferred to nitrocellulose filters. After being blocked with 5% (w/v) skim milk, 0.05% Tween 20, PBS at 4°C for 16 h, filters were incubated with rabbit anti-HIV-NC IgG monoclonal antibody at 37°C for 1 h. Following extensive washing with 0.05% Tween 20 and PBS, a secondary antirabbit IgG, which was conjugated to horseradish peroxidase (Amersham Pharmacia Biotech) was added for 1 h at 37°C. After thorough washing, NC was visualized using a ECL chemiluminescence detection kit (Amersham Pharmacia Biotech).

RESULTS
A Role of NC Protein during tRNA Lys. 3 Primer Placement Is to Facilitate the Formation of a Competent Reverse Transcription Initiation Complex-NC is known to elevate the efficiency of (Ϫ) ssDNA synthesis in two ways. 1) it transiently destabi-

5Ј-AGACCAGATCTGAGAATGG-3Ј
468-486 Hpa-S lizes the secondary structure of template that RT may encounter during polymerization (12)(13)(14)(15). 2) It helps the formation of a specific tRNA Lys.3 ⅐vRNA complex that is favored during subsequent reverse transcription (11). To study the roles of NC during initiation of reverse transcription, we used a saturating level of NC in tRNA placement experiments. RT reactions were then allowed to proceed for various periods (1, 4, 16, 32, and 64 min) in either the presence (Fig. 1A, lanes [11][12][13][14][15] or absence of NC, which was achieved by proteinase K and phenol:chloroform treatment prior to reverse transcription (Fig. 1A, lanes 16 -20). Compared with the strong pausing at the ϩ1 nt stage, observed with RNA complexes formed by heat annealing (Fig.  1A, lanes 1-5), the ϩ1 pausing event was dramatically diminished in reactions initiated from a tRNA Lys.3 ⅐vRNA complex preformed by NC. This was true regardless of whether NC was present throughout the reaction or only during the annealing stage.
Similar conclusions were reached by calculating the percentage of each of the pause products relative to the total radioactivity of all pause sites in the reaction. As shown in Fig. 1B, in the initiation reactions performed with heat-annealed RNA complexes, the ratio of the ϩ1 pause product decreased when the RT incubation time increased; however, the percentage of this product never dropped below 60%, even after 64 min of reverse transcription reaction period. When NC was used to place the tRNA primer onto the RNA template, no matter whether it was removed from the subsequent reverse transcription reactions or not, the resulting RNA complexes showed much higher efficiency of cDNA synthesis. As shown in Fig. 1B, only ϳ20% of cDNA products appeared to represent the onebase extension product in these experiments. Therefore, NC is able to help the RNA complex to escape from the ϩ1 pausing event, and this role is played through formation of an active tRNA Lys.3 ⅐vRNA complex and not by the direct involvement of NC in RT reactions.
Another interesting observation in regard to Fig. 1A is that the presence of NC cannot help RT to overcome the ϩ3 pause event. Quantification of Fig. 1A shows that fewer ϩ1 and more ϩ5 nt products were produced when RT reactions were incubated for longer periods, whereas the proportion of ϩ3 nt products remained relatively constant during the time course (Fig.  1B). The presence of NC during the reverse transcription reactions was able to decrease the average percentage of the ϩ3 nt products from 60% to 50%; however, NC did not facilitate escape from the ϩ3 nt pause event as it did for pausing at the ϩ1 nt position (Fig. 1B). This observation suggests that the ϩ3 nt pause during initiation of reverse transcription is biologically relevant.
To confirm that NC protein had indeed been eliminated from our reverse transcription reactions as described above, we used anti-HIV-NC monoclonal antibody to analyze NC proteins after incubation with tRNA and RNA template and subsequent proteinase K and phenol:chloroform treatment. As shown in Fig. 2, no NC protein was detected in Western blots after Protease K digestion and extraction. The sensitivity of the blotting is shown by a control experiment in which 3 pmol of NC protein were loaded directly onto the gel. Previous results had shown that this concentration of NC protein did not affect reverse transcription (11).
The Role of NC Protein in Initiation of Reverse Transcription Is Dependent on Intact Zinc Finger Motifs-To assess the potential roles of zinc finger motifs in the formation of active tRNA Lys.3 ⅐vRNA complex, we next examined two mutated NC proteins in reactions designed to study the initiation of reverse transcription. One is H23C NC, in which a His at the 23 position has been substituted by a Cys; the other is ddNC, in which both of zinc fingers have been replaced by Gly-Gly linkages. Primer tRNA was placed onto viral RNA template through use of various concentrations of NC protein, i.e. 5, 15, 30, and 45 pmol. Thereafter, three of the four dNTPs, i.e. dCTP, dTTP, and dGTP, were added to initiate reverse transcription. The results show that neither H23C NC nor ddNC was able to achieve release from the ϩ1 pause site in these reactions (Fig.  3A, lanes 5-12), and the average ratio of the amount of ϩ1 pausing products to total paused products was 50% in both of these cases, which is higher than the 20% average achieved through use of wild-type NC protein (Fig. 3B). It should be noted that the reactions performed with ddNC yielded a greater number of initiation intermediate products than did reactions performed with wild-type NC. Different types of preparation of NC proteins, including both wild-type and mutated varieties, were used in these experiments and gave rise to similar results each time. Therefore, wild-type zinc fingers are important in order for NC to promote the formation of an active tRNA Lys.3 ⅐vRNA initiation complex.
The ϩ3 nt Pausing Event Is Caused by a Viral RNA Template Stem Structure Located Upstream of the PBS within the tRNA Lys.3 ⅐vRNA Complex-As stated above, the ϩ3 nt pausing   5, and 7) of NC, respectively. Lanes 2 and 3 represent samples after proteinase K and phenol:chloroform treatment; lanes 4 and 5 represent samples without such treatment; in lanes 6 and 7, only 5% of the samples in lanes 4 and 5 were loaded. Western blots were performed as described under "Materials and Methods." persisted during initiation of reverse transcription regardless of whether heat or NC annealing had been performed. This indicates that factors other than NC protein must contribute to this specific initiation event. Our previous work had shown that reactions that used HIV/HUA, an RNA template that includes a substitution at positions 624 -635 at the 5Ј end of the PBS, did not pause at the ϩ3 position (7). In fact, the sequence that is substituted in the HIV/HUA construct participates in the formation of an 8-base pair stem structure located at the fourth nucleotide position upstream of the PBS in the tRNA Lys.3 ⅐vRNA binary complex (1-3). As illustrated in Fig.  4A, this stem is solely composed of template sequences, but it connects sequences involved in two specific intermolecular interactions, i.e. the PBS/3Ј end of tRNA, and the A-rich loop/ anti-codon loop of tRNA. Both of these intermolecular interactions play a role in the determination of tRNA species selection (23)(24)(25)(26) and in initiation of reverse transcription (4,7,(27)(28)(29)(30)(31)(32).
We hypothesized that, when reverse transcription proceeds to this stem structure, RT may first have to disrupt the stem before the incorporation of complementary nucleotides, and that this leads to the reaction pause at the ϩ3 nt position. To pursue this subject, four groups of mutations were introduced into this structure. The first group includes N1 and N2, in which 3 or 5 bases in the left part of the stem were substituted, such that the lower 3 or 5 base pairs were disrupted. The second group of RNA templates, i.e. N3 and N4, involve mutations in the right half of the stem. In N3, an original G, located at the fourth nucleotide upstream of the PBS, was changed to a C, such that the bottom hydrogen bond of the stem was destroyed. In N4, GAU, i.e. the fourth to sixth nucleotides upstream of the PBS, was changed to CGA; therefore, the lower 3 base pairs were disrupted and replaced by two new hydrogen bonds. The third group of mutations includes N5 and N6, which were generated by interchanging the lower 3 and 5 base pairs. We also generated N7 and N8, in which the distance between the stem structure and the PBS was increased by the insertion of CAGs. The structures and sequences of these mutations are shown in Fig. 4 (A and B), respectively.
In the control experiment, wild-type RNA templates were annealed with tRNA Lys.3 by heat annealing, following which one (i.e. dCTP), two (i.e. dCTP, dTTP), three (i.e. dCTP, dTTP, dGTP), or all four of the dNTPs were added. As shown in Fig.  5 (A (lanes 1-4), B, E, and F), the incorporation of one or two dNTPs yielded products at the ϩ1 nt or both the ϩ1 and ϩ2 nt positions. In the presence of each of dCTP, dTTP, and dGTP, pausing at both the ϩ1 and ϩ3 nt positions was seen in addition to the expected five-base extended product. The presence of all three intermediate products, seen with the addition of all four dNTPs, indicates that their presence in lane 3 was not caused by the absence of dATP. Longer cDNA products in lane 4 were labeled at positions ϩ27, ϩ28, and ϩ40 nt. Because low concentrations of dNTPs (i.e. 160 nM) were used in these reactions, it was difficult for longer reverse transcription products, i.e. Ͼ ϩ40 nt, to be seen on the gel.
In the case of N1, the pausing site at the ϩ3 nt position disappeared (Fig. 5A, lanes 5 and 6), whereas pausing at the ϩ5 nt site was diminished but still evident; at the same time, a greater number of longer cDNA products (e.g. ϩ27, ϩ28, and ϩ40 nt) were produced as compared with reactions performed with wild-type template. Even stronger perturbation of the stem structure by the N2 mutation caused both a significantly decreased efficiency of initiation as well as transition from an early to a late phase of initiation, i.e. only a one-base extension product is seen in lanes 7 and 8. To investigate the effects of N2 on the ϩ3 pausing event, higher quantities of RT, i.e. 405 ng, were used to extend reactions beyond the ϩ1 stage. Under this circumstance, we only detected an extended product at the ϩ5 site (lane 9) or at higher positions (lane 10) as expected; no pausing was observed at the ϩ3 site. The N3 and N4 mutations each gave rise to a relatively complicated pausing pattern in initiation reactions. When all four dNTPs were present in these reactions, neither the ϩ3 nt nor the ϩ5 nt products were seen on the gel (Fig. 5B, lanes 6  and 8). When only three dNTPs (i.e. dCTP, dGTP, dTTP) were added, pausing at the ϩ3 nt site disappeared as well. However, cDNA products longer than ϩ16 nt were observed alongside the expected ϩ5 and ϩ7 nt products (Fig. 5B, lanes 5 and 7). Therefore, nucleotide misincorporation and elongation from the misincorporated nucleotide must have occurred at template sequences wherever a U was met. This result was confirmed by use of two mutated forms of RT associated with higher than average base incorporation fidelity, i.e. E89G and M184V (33)(34)(35). During reverse transcription of the N4 RNA template, nucleotide misincorporation was either diminished or eliminated in reactions performed with either the E89G or M184V RT (Fig. 5, C and D). Fig. 5E shows the experiment performed with the N5 and N6 group. Because the stem structure was preserved, the specific intermediate initiation product at the ϩ3 nt position was not affected (lanes 5-8). The efficiency of initiation in reactions performed with these mutated RNA templates was also near wild-type levels. However, this group of mutations significantly diminished the efficiency of the switch from the ϩ3 to the ϩ5 stage; cDNA products beyond ϩ3 nt can hardly be seen on the gel.
In the case of N7, only one CAG repeat was inserted, and the pausing event at the ϩ3 nt position, associated with wild-type template, moved to a higher position at ϩ6 nt (Fig. 5F, lanes 5  and 6). When more insertions were introduced into the template, e.g. N8, the reaction became defective at the initiation stage. As shown in Fig. 5F (lanes 7 and 8), the reactions were arrested after the incorporation of the first nucleotide. We next used higher concentrations of RT (i.e. 405 ng) in reactions performed with N8. Pausing at the ϩ5, ϩ9, and ϩ11 nt positions can be observed when three dNTPs were added into the reactions (lane 9); these represent a common pause position (i.e. ϩ5), the nt position just prior to the stem structure (i.e. ϩ9), and the longest extended position that can be achieved with all three dNTPs (i.e. ϩ11), respectively. When all four dNTPs were present in these reactions, reverse transcription was able to proceed beyond the ϩ11 nt position (lane 10).
Because N3 and N4 included mutations in the right half of the stem, the specific pausing patterns observed may conceivably have been the result of either the altered secondary structures established or, alternatively, of modified sequences. To sort out these possibilities, two additional mutations, i.e. N9 and N10, were constructed to restore the stem base pairing that had been disrupted by N3 and N4 by introducing second site mutations into the left half of the stem of the N3 and N4 constructs. Thus, N9 and N10 keep modified sequences in the right half of the stem, similar to those of N3 and N4, while also maintaining an intact stem structure (Fig. 6, A and B). Reactions employing these two new mutations are shown in Fig. 6C. Strong pausing at the ϩ3 nt site that was eliminated in experiments performed with the N3 and N4 constructs was reestablished when the N9 and N10 templates were used. Thus, the initiation of reverse transcription involves pausing at the ϩ3 nt site when the RNA template used contains an intact stem structure at the 4th nt upstream of the PBS, regardless of whether the stem was formed by wild-type or exogenous sequences.
Because NC protein does not release the pause event at the ϩ3 nt position, we mainly used heat annealing in the above in vitro reverse transcription studies. As a control, we also employed NC protein to place tRNA Lys.3 primer onto the N1-N8 templates, and then checked for the initiation of reverse transcription from these binary RNA complexes. As shown in Fig. 7, all reactions exhibited similar pausing patterns as those performed by the heat annealing method, except that pausing bands at the ϩ1 site were markedly reduced.
Initiation of Reverse Transcription from the Virion-derived tRNA Lys 6, and 8). The final concentration of each of the dNTPs was 160 nM. B, initiation of reverse transcription with mutated N3 and N4 RNA templates. The order of lanes 1-8 is the same as that of A, except that lanes 5 and 6 are reactions that employed N3 template; lanes 7 and 8 are reactions with N4 mutated template. C, initiation of reverse transcription of N4 with E89G mutated RT. The RNA template used in this experiment is N4, and the RTs utilized include 45 ng of wild-type RT (p51/p66) (lanes 1-4) and 45 ng of E89G RT (lanes 5-8). D, initiation of reverse transcription of N4 with M184V mutated RT. The RNA template used in this experiment is N4, and the RTs utilized include 45 ng of wild-type RT (p51/p66) (lanes 1-4) and 250 ng of M184V RT (lanes 5-8). E, initiation of reverse transcription with mutated N5 and N6 RNA templates. The order of lanes 5-8 is the same as that of A, except that lanes 5 and 6 are reactions that used N5 template; lanes 7 and 8 are reactions that used N6 mutated template. F, initiation of reverse transcription with mutated N7 and N8 RNA templates. The order of lanes 1-10 is the same as that of A, except that lanes 5 and 6 are reactions with N7 template; lanes 7-10 are reactions with N8 template.
fected with HIV-1 cDNA constructs containing these mutations. Then, virus particles were harvested, and viral RNA was isolated as described under "Materials and Methods." Initiation from the virion-derived tRNA Lys.3 ⅐vRNA complexes was studied by incubation with 45 ng of RT and 160 nM each of dCTP, dTTP, and dGTP at 37°C for 15 min. Because the sequence of the first five incorporated dNTPs was CTGCT, and because we had used only one type of isotope (i.e. [␣-32 P]dCTP) in the experiment, we reasoned that the expected pausing band at the ϩ3 nt position should be the extension product from the unextended primer, but not from the two-base extended primer, which can also be detected within mature virions (19). The results of Fig. 8 show that both the ϩ3 and ϩ5 nt products were observed with RNA from wild-type virus; in contrast, ϩ3 pausing was hardly seen with RNA from any of the N1, N3, N4, N5, or N7 mutated viruses. These results further support the conclusion that the ϩ3 pausing event during reverse transcription is caused by a stem structure present in the tRNA Lys.3 ⅐vRNA complex.
We also observed that none of the reactions that employed RNA complexes derived from virus particles paused at the ϩ1 stage; moreover, much weaker pausing occurred at the ϩ3 nt stage in reactions performed with wild-type virus-derived RNA complex than in those formed by either heat or NC annealing. Because all sorts of RNA complexes in this experiment were subjected to the same in vitro reverse transcription reaction conditions, the different pausing patterns obtained can only be attributed to the different conformations of the complexes used to initiate the reactions. In general, the NC-annealed RNA complex acts in a manner more similar to that of RNA complexes derived from virus particles than to those formed by heat annealing. Of course, either structural or functional differences may still remain between the tRNA Lys.3 ⅐vRNA complexes that are derived from virions or formed by NC annealing in cell-free assays. DISCUSSION HIV-1 NC is a small basic protein, containing two zinc finger motifs of the CX 2 CX 4 HX 4 C form. NC has been demonstrated to function as an RNA-chaperone protein in a wide variety of cell-free systems. It promotes annealing of complementary nucleic acid sequences (36 -39), it enhances nucleic acid strand transfer from a less stable to a more stable hybrid (37,39), it induces maturation of dimeric retroviral RNA (40), and it stimulates hammerhead ribozyme catalysis (41)(42)(43)(44). During the viral life cycle, NC has been shown to function at multiple steps, including viral RNA packaging and dimerization, virion assembly, reverse transcription, integration, and transcription (8,10,45,46). Because of these features, it has become an attractive target for the design of potential anti-HIV compounds.
Our studies focused on the RNA-chaperone activity of HIV-1 NC protein during the annealing of tRNA Lys.3 to viral RNA template containing the PBS. In a previous study, we found that the tRNA Lys.3 ⅐vRNA initiation complex, formed in the presence of NC, resulted in elevated efficiency of the switch from initiation to elongation in the subsequent synthesis of (Ϫ) ssDNA; the zinc finger motifs of NC were required for this activity (11). In this work, we have shown that the role of NC during initiation of reverse transcription, i.e. the release of pausing from the ϩ1 nt position, was also the result of the activity of NC during the tRNA annealing process. An intact zinc finger structure was indispensable for this specific activity.
On the basis of these observations, the following conclusions regarding the RNA-chaperone activity of HIV-1 NC can be drawn. (i) An RNA complex with the most thermostable structure, i.e. that formed by heat annealing, may not always be the most functional. (ii) The RNA-chaperone mechanism to explain the formation of the most stable RNA complexes includes a transient base pair destabilization, followed by random repairing until the maximal number of base pairs is reached (9,10,47). However, in our system, NC was shown to assist in the formation of a functional RNA complex rather than a thermostable one. This shows that a random or nonspecific model does not fit well with our findings. (iii) Zinc finger motifs were found to be important; this provides further evidence that specific interactions between protein and RNA may involve in the proper folding of the tRNA Lys.3 ⅐vRNA complex. Therefore, the tRNA Lys.3 ⅐vRNA complex formed in the presence of HIV-1 NC protein is a specialized one, which may involve both nonspecific and specific interactions between the protein and the RNA.
Previous studies have shown that it is the basic residues rather than the zinc finger motifs of NC protein that are essential for its RNA-chaperone activity (8,10,39,40,44,48). In agreement, we found that mutations of the zinc fingers did not affect the tRNA primer annealing capability of NC protein (11). However, the tRNA Lys.3 ⅐vRNA binary complex that was formed in the presence of zinc finger mutated NC showed a lower efficiency of initiation of reverse transcription in this work; this is consistent with our previous data, which showed that such mutated NC was less efficient at mediating switch from initiation to elongation (11). Similar conclusions have been drawn from in vivo studies, in which sequences between the two Cys-His boxes of NC were found to be essential for packaging and placement of tRNA Lys.3 , whereas mutations that showed the most deleterious effects in regard to initiation of reverse transcription were those that mapped to either of the two NC Cys-His boxes (49).
The zinc finger-dependent nucleic acid chaperone activity of NCp7 has also been reported in other in vitro systems. For example, although NC (12-53) can promote the efficient annealing of tRNA Lys.3 to the PBS, modification of the first zinc finger in this truncated form of NC led to a large reduction in its primer annealing capacity (50). Zn 2ϩ -binding residues were also found to play important roles during minus and plus strand transfer during HIV-1 reverse transcription, by promoting the unfolding of highly structured RNA and DNA strand transfer intermediates (51). The helix-coil transition of single DNA molecules has been measured through use of an optical tweezer instrument; intact zinc finger structures were shown to be indispensable for two essential activities of NC, i.e. to destabilize double-stranded DNA and to decrease the cooperativity of the helix-coil transition (47). Such prolonged intermediate states between the helix and coil induced by NC may provide an good environment for the specific recognition of nucleotides by zinc finger motifs, which may lead to formation of a reverse transcription competent tRNA Lys.3 ⅐vRNA complex, as shown both in this paper and in previous reports (11).
In contrast to the ϩ1 nt pausing event, the ϩ3 nt position cannot be bypassed by the presence of NC protein during the annealing process. Similar differences in formation of the ϩ1 and ϩ3 nt pause sites were also observed in our previous studies. For example, initiation of reverse transcription paused at the ϩ3 nt site, whenever tRNA Lys.3 or a DNA primer was employed; however, the ϩ1 nt intermediate product disappeared when RT reactions were primed from an oligo DNA (7). In vitro probing data have described the existence of a 8-base pair stem structure, located at the fourth nt position upstream of the PBS within the tRNA Lys.3 ⅐vRNA complex (1-3). Accordingly, we reasoned that reverse transcription must have paused after the third nucleotide was incorporated to dissolve the base pairs of the stem structure. To prove this point, we introduced a series of mutations into the viral RNA template, such that the stem structure was deliberately destabilized. Our results show that reverse transcription of RNA templates containing these mutations no longer paused at the ϩ3 nt position, regardless of whether a heat annealing or NC annealing method was employed.
As stated, the arrest of RT reactions at the ϩ3 nt stage is caused by the existence of the stem within the tRNA Lys.3 ⅐vRNA complex. In addition, the introduction of mutations that altered either the structure (N1, N2, N3, and N4) or the position (N7, N8) of the stem resulted in less efficient initiation of reverse transcription, as shown in Fig. 5 (A, B, and E); this effect was most pronounced in the case of the N2 and N8 mutated templates (Fig. 5, A and E). In contrast, the N5, N6, N9, and N10 templates represent constructs that preserved stem structure, whereas sequences within the stem were changed. The overall efficiency of initiation of RT reactions performed with these templates approached wild-type levels, although the switch from the ϩ3 to the ϩ5 stage was compromised. Therefore, as long as the stem structure was not disturbed, or the distance of the stem from the PBS remained unchanged, the tRNA⅐vRNA complex would remain functional, and an optimal initiation rate of reverse transcription would be achieved. Because of its special position within the tRNA⅐vRNA complex, the stem structure studied may help to promote efficient reverse transcription by stabilizing two adjacent intermolecular interactions between the primer and the template, i.e. the PBS/3Ј end of tRNA and the A-rich loop/anti-codon loop of tRNA; both of these are essential for the efficient initiation of (Ϫ) ssDNA synthesis (4,7,(27)(28)(29)(30)(31)(32).
Although its structure is key, the sequence of the stem can also affect the efficiency of reverse transcription. For example, the sequence of the right half of the stem structure was found to be important in the switch from the ϩ3 nt pause site to a even later initiation intermediate at the ϩ5 nt position. Reverse transcription of the N5 and N6 templates, which exchanged base pair sequences without disturbance of the stem structure, barely proceeded beyond the ϩ3 stage, although the reaction was initiated in near normal fashion (Fig. 5E). The fidelity of reverse transcription was also affected when sequences on the right half of the stem were mutated (Fig. 5B).
The sequence within/around the stem structure may also affect patterns of pausing during initiation of reverse transcription. For instance, in the experiments performed with the N7 and N8 templates, bands at the ϩ5 and ϩ8 nt positions were visible in addition to the expected pause sites at ϩ6 and ϩ9 (Fig. 5F). The question that arises is whether it is the sequence or the stem structure that determines the ϩ3 pause site. Fig. 5 (A and E) answers this because disruption of the stem structure in the RNA templates N1 and N2 eliminated pausing at the ϩ3 site (Fig. 5A). Restoration of stem base pairing by second-site mutations in the N5 and N6 RNA templates led to reappearance of the ϩ3 pause site (Fig. 5E). Thus, formation of the specific pause site at the ϩ3 nt position is a result of the existence of the stem structure rather than of specific RNA sequences.
This conclusion is further supported through experiments performed with constructs N9 and N10, in which the stem structure that had been destabilized by the N3 and N4 mutations was restored by second-site compensatory mutations (Fig.  6, A and B). The results of Fig. 6C show that the ϩ3 pause event that was eliminated in experiments performed with the N3 and N4 templates was reestablished when the N9 and N10 RNA templates were used (Fig. 6C). Thus, the pause event at the ϩ3 nt site is highly specific and is caused by the stem structure located at the fourth nucleotide upstream of the PBS, even though it is also true that other template sequences proximal to the PBS can affect patterns of pausing to a limited extent.
In general, our mechanistic studies of early pausing events and the role of the NC protein in this process contribute further to an understanding of the initiation of reverse transcription. First, we have demonstrated that the ϩ1 pausing event is a distinct rate-limiting step during initiation, because it can only be observed when RNA is used to prime reverse transcription (7), and because the addition of NC protein containing intact zinc finger motifs can help RT to escape this pause site. Second, the presence of the ϩ3 nt pausing event is solely a result of the specific secondary structure of the tRNA Lys.3 ⅐vRNA complex. Therefore, the mechanisms responsible for the formation of these distinct intermediate products during initiation of HIV-1 reverse transcription are different.
Finally, early pause events at the ϩ3 and ϩ5 nt positions were observed when tRNA Lys.3 ⅐vRNA was prepared from virus particles, indicating that similar initiation events also transpire in the virus. However, comparison of relative amounts of each short cDNA product revealed that the initiation complex, isolated from the virus, more closely resemble that annealed by NC than that prepared by heat annealing (Fig. 8). This implies that NC protein plays an important role during initiation of reverse transcription both in vitro and in vivo.