A HIV-1 Minimal Gag Protein Is Superior to Nucleocapsid at in Vitro \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\mathrm{tRNA}_{3}^{\mathrm{Lys}}\) \end{document} Annealing and Exhibits Multimerization-induced Inhibition of Reverse Transcription*

HIV-1 uses \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\mathrm{tRNA}_{3}^{\mathrm{Lys}}\) \end{document} to prime reverse transcription of its viral RNA. In this process, the 3′-end of \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\mathrm{tRNA}_{3}^{\mathrm{Lys}}\) \end{document} must be annealed to the primer binding site of HIV-1 genomic RNA, and the two molecules together form a complex structure. During annealing, the nucleocapsid (NC) protein enhances the unwinding of tertiary structures within both RNA molecules. Moreover, the packaging of \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\mathrm{tRNA}_{3}^{\mathrm{Lys}}\) \end{document} occurs prior to viral budding at a time when NC is still part of the Pr55Gag polyprotein. In contrast, Pr55Gag is able to produce virus-like particles on its own. We have recently shown that an N-terminal extended form of NC (mGag), containing all of the minimal elements required for virus-like particle formation, possesses greater affinity for HIV-1 genomic RNA than does NC alone. We have now studied the \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\mathrm{tRNA}_{3}^{\mathrm{Lys}}\) \end{document}-annealing properties of mGag in comparison to those of NC and report that the former is more efficient in this regard than the latter. We have also tested each of a mutant version of mGag, an extended form of mGag, and an almost full-length form of Gag, and showed that all of these possessed greater tRNA-annealing capacity than did the viral NC protein. Yet, surprisingly, multimerization of Gag-related proteins did not abrogate this annealing process but rather resulted in dramatically reduced levels of reverse transcriptase processivity. These results suggest that the initial stages of reverse transcription may be regulated by the multimerization of Pr55Gag polyprotein at times prior to the cleavage of NC.

During the retrovirus life cycle, viral genomic RNA is retrotranscribed by the reverse transcriptase (RT) 1 enzyme into a double-stranded cDNA, which is then integrated into the host cell genome and forward transcribed to produce viral RNA and proteins. Cellular tRNAs are used as the primer for the reverse transcription reaction. Human immunodeficiency virus, type 1 (HIV-1) uses human tRNA 3 Lys , one of three major tRN Lys isoacceptors, for the priming of reverse transcription (1). tRNA 3 Lys is selectively incorporated into the virus particle and is annealed via 18 bases at its 3Ј-end to a complementary 18-base sequence near the 5Ј-end of viral genomic RNA termed the primer binding site (PBS) (2)(3)(4). Once annealed, tRNA 3 Lys primes the synthesis of minus strand cDNA. For a review, see Ref. 5.
Both tRNA 3 Lys and the PBS have complex three-dimensional structures, and the annealing of the two relevant sequences requires the "unwinding" of local stem-loop structures (6,7). The mature nucleocapsid (NC) protein of HIV-1 has been shown to facilitate this annealing in vitro (8). Presumably, NC enhances local unwinding of tertiary structures in the vicinity of the PBS to allow the annealing of tRNA 3 Lys (see Fig. 1a) (9). NC is a small basic protein that is cleaved from the Gag precursor during viral particle maturation. NC binds to structures in viral RNA termed stem loop 1 (SL1) through SL4 to specifically package HIV-1 genomic RNA (10 -14). NC contains two Zn 2ϩ fingers that are required for both this genomic RNA recognition as well as for its roles in strand transfer (15). However, mutational analyses of NC have not implicated the Zn 2ϩ fingers in the tRNA 3 Lys annealing reaction, although it appears that the basic regions that flank the first Zn 2ϩ finger are important in this regard (16,17).
In current models of HIV-1 viral replication, tRNA 3 Lys is annealed during viral budding prior to proteolytic cleavage of the Gag precursor. Therefore, the activity of NC must occur in the context of the Pr55 Gag polyprotein. Previous research showed that extracts of viral RNA from both wild-type viruses and protease-deficient HIV-1 particles were able to anneal to tRNA 3 Lys (18), a finding consistent with results obtained from studies on murine and avian retroviruses (19,20). NC binds nonspecifically to tRNA 3 Lys in vitro and, consequently, does not play an important role as a mature protein in the specific annealing of tRNA (21). Interestingly, other in vitro experiments have found that both mature NC and immature Gag possess equivalent ability to anneal tRNA 3 Lys to a synthetic 5Ј-HIV-1 genomic RNA molecule (22).
Despite its roles in a variety of HIV-1 life cycle events, much of the Pr55 Gag polyprotein is not required for the assembly of virus-like particles (VLPs), and, indeed, all of the required protein-protein interaction domains required for viral assembly are located in the capsid (CA) and NC regions (23)(24)(25)(26)(27)(28). The minimal Gag domains necessary for VLP formation in vivo are the myristylation signal, the C-terminal domain of CA, the SP1 spacer peptide, NC, and p6 (29,30). It has also been proposed that SP1 and a potential ␣-helix that traverses the CA-SP1 boundary are involved in both Gag multimerization (31,32) and the specificity of HIV-1 genomic RNA packaging (33,34). Indeed, our laboratory has shown that a purified minimal Gag (mGag) protein, which contains all of the Gag domains that support VLP formation, can bind to genomic RNA with a stronger affinity than does NC (35); this suggests that not all genomic RNA recognition by Gag is attributable to sequences in NC (36,37). In addition, mGag was shown to possess an ϳ10-fold greater affinity than NC for a genomic RNA template that included an extended 5Ј-leader region beyond the traditionally recognized "" packaging signal, i.e. the SL5 and early complex platform (ECP) (35). Because tRNA 3 Lys annealing occurs during Gag-mediated VLP assembly, this raises a number of questions. First, do other domains outside of NC but within the Gag precursor also contribute to tRNA 3 Lys annealing? Second, what are the consequences of Gag multimerization on the efficiency of tRNA 3 Lys annealing and the priming of reverse transcription? This report deals with the use of a mGag-purified protein (Fig. 1b), consisting of the myristylation signal, the C-terminal domain of CA, and NC to study the effect of Gag multimerization on tRNA 3 Lys annealing as well as the possibility that Gag sequences besides NC might be involved in tRNA 3 Lys annealing. We now show that mGag is superior to NC in its ability to anneal tRNA 3 Lys to the PBS and also that Gag multimerization does not inhibit annealing of tRNA 3 Lys . In contrast, large multimers of mGag can inhibit reverse transcription of (Ϫ) strong stop DNA.

MATERIALS AND METHODS
Protein Synthesis-The synthesis and purification of the proteins used has been described previously (35). Briefly, protein was expressed in B121 (DE3) cells (Novagen) and purified under denaturing conditions by affinity with nickel agarose. Further purification was performed by anion exchange chromatography (Q-agarose). Protein was refolded by dialysis under reducing conditions, and removal of unfolded protein was accomplished by ultracentrifugation at 100,000 ϫ g for 1 h. Proteins from several preparations were employed in each of the different experiments, and no differences among these various preparations were observed in regard to results obtained.
Synthesis and Folding of RNA-DNA templates were amplified from a vector containing the sequence for HIV-1 HxB2D (GenBank TM K03455) in a PCR in the following manner. A primer with the T7 promoter at its 5Ј-end followed by the sequence complementary to positions ϩ1 to ϩ24 was used as the forward primer with the reverse primers complementary to positions ϩ181 to ϩ199, ϩ360 to ϩ397, and ϩ460 to ϩ486 (see Fig. 1). PCR products were agarose-purified and used as templates in the T7 transcription reaction. RNA was synthesized using the T7 Megashortscript kit (Ambion, Texas) following the manufacturer's recommendations. Competitive RNA was synthesized using the linearized plasmid included in the same kit that encodes a 116-nucleotide-long RNA transcript and transcribes an RNA molecule that contains the sequence of human 18 S rRNA. Reactions were stopped by the addition of 2 units of DNase I and further incubation at 37°C for 15 min. RNA transcripts were purified by denaturing polyacrylamide gel electrophoresis as previously described (35), eluted in 600 mM NaOAc, 1 mM EDTA, and 0.1% SDS. RNA templates were precipitated with ethanol and resuspended in 100 mM NaCl, 20 mM Tris, pH 7.5. RNA templates were refolded by FIG. 1. Genomic RNA templates and proteins used. a, top, shown schematically is the 5Ј-HBX2 HIV-1 genomic RNA structure predicted from M-Fold analysis and previous reports with HIV-1 Mal from nucleotides ϩ1 to ϩ506, where ϩ1 is the capping site. From 5Ј to 3Ј the structures shown are the Tat-associated region (TAR), the poly-adenylation signal (Poly(A)), the primer binding site (PBS), stem loops 1-5 (SL1-5), and the early complex platform (ECP). Bottom, primers used in PCR amplification and the resultant PCR products that were used as genomic RNA templates. b, schematic diagram of proteins used. Numbers correspond to the last amino acids of each peptide within the precursor. A black line denotes amino acids 2-7, which are included in mGag. An asterisk in mGag denotes the position of the M318A mutation. The gray represents the N-terminal histidine tag of the vector used. c, SDS-PAGE analysis of the different proteins used. ϳ5 g of protein were loaded in each lane, separated by electrophoresis, and visualized by staining with Coomassie Brilliant Blue. Two different protein ladders are shown (first and last lanes), and the molecular masses of the bands are given on the side in kDa. The identity of the proteins employed is given below each lane.
In Vitro tRNA 3 Lys Annealing by mGag heating to 85°C and slowly cooling to room temperature. RNA was then stored at Ϫ80°C. Human placental tRNA 3 Lys purified by high pressure liquid chromatography was purchased from Bio S&T (Montreal, Quebec), resuspended in 100 mM NaCl, 20 mM Tris, pH 7.5, and refolded as described above.
Reverse Transcription Elongation Assay-RNA transcripts and tRNA 3 Lys ( Mixtures were incubated at 37°C for 10 min to allow for tRNA 3 Lys annealing following which MgCl 2 was added to a concentration of 3 mM for a final reaction volume of 15 l. The mixtures were then incubated for an additional 30 min at 37°C to allow reverse transcription to proceed. A positive control involved heating the tRNA 3 Lys and template RNA together to 75°C and slowly cooling to room temperature before inclusion in the reaction. A negative control was performed without protein. Reactions were terminated by bringing the final volume to 100 l by addition of 0.1% SDS and 0.3 M NaOAc. RNA was then ethanol precipitated and resuspended in 8 l of denaturing gel loading buffer II (Ambion, Texas). 4 l of each sample was run on a 5% polyacrylamide gel under denaturing conditions, dried, and exposed to Kodak Bio-Max films at Ϫ80°C overnight. Results were analyzed by molecular imaging in comparison with the heat-annealed positive control. This method (shown schematically in Fig. 2a) allows for determinations of relative RT processivity and tRNA 3 Lys annealing efficiency through incorporation of the first 6 nucleotides in newly synthesized (Ϫ) strand cDNA.
Protein Annealing Assay-In this reaction, 25 nM each 1-ECP RNA template and tRNA 3 Lys and 50 nM competitive RNA were first combined with increasing amounts of Gag-related proteins to a total volume of 15 l in a buffer of 50 mM Tris, pH 7.5, 50 mM NaCl, and 3 mM Mg 2ϩ . The reaction mixture was incubated at 37°C for 10 min, after which 50 g/ml RNase-free proteinase K was added in a buffer of 0.01 M Tris, pH 7.5, 0.005 M EDTA, 0.5% SDS in a volume of 100 l. After incubation for 1 h at 37°C, phenol-chloroform extraction was performed, followed by ethanol precipitation. The pellet was resuspended in the same reverse transcription mix described above for the elongation assay and reactions were incubated for 30 min at 37°C. The reaction termination, precipitation, and gel steps were also described above. This method (Fig. 2b) allows us to estimate the ability of a protein to participate in tRNA 3 Lys annealing, because protein is digested, and nucleic acids are phenol-extracted after the annealing step.
Heat Annealing Assay-25 nM each tRNA 3 Lys and 1-ECP template RNA were annealed by heating to 75°C and slowly cooling to room temperature. Protein was then added at the indicated concentrations, and the mixtures were incubated for 10 min at 37°C in the abovedescribed reverse transcription reactions, also containing 50 nM competitive RNA but no Mg 2ϩ in a total volume of 14.1 l. Mg 2ϩ was then added to a final concentration of 3 mM in a volume of 15 l, and reverse transcription was allowed to proceed for 3 min at 37°C. Other procedures were as indicated above for the elongation assay. This method (shown schematically in Fig. 2c) allows us to isolate the ability of a protein to participate in the reverse transcription reaction independently of the ability to promote tRNA 3 Lys annealing. Statistical Analysis-Statistical analysis was performed using GraphPad Prism software, version 4.00 for Windows (San Diego, CA).

RESULTS
mGag Has tRNA Annealing Ability Superior to NC-Previous data from our laboratory showed that mGag is superior to NC in the recognition of HIV-1 genomic RNA (35). We and others have proposed that the CA-SP1 interface plays a role in viral packaging specificity (33,34). Accordingly, we were interested in comparing mGag with NC in ability to anneal tRNA 3 Lys onto the PBS. We generated a HIV-1 probe termed 1-PBS, because it contains nucleotides from positions ϩ1 to ϩ247, and co-incubated this transcript with either NC or mGag at the indicated concentrations together with human placental tRNA 3 Lys . Then, we performed reverse transcription reactions using radiolabeled nucleotides. Inclusion of ddATP in the reactions as a sole source of dATP resulted in a 6-nucleotide polymerization product (38). This method was used to determine the efficiency of tRNA 3 Lys annealing, based on the amount of [␣-32 P]dCTP incorporated into DNA product (see Fig. 2a).
As shown in Fig. 3, mGag was 8 -10-fold superior to NC in its ability to anneal tRNA 3 Lys onto the PBS. Interestingly, however, the use of very high concentrations of mGag resulted in decreased incorporation of the radiolabeled nucleotide; results from experiments performed with these high concentrations of mGag were not included in our K d analyses for mGag (Fig. 3c). Many of the samples yielded a double band, which probably corresponds to incorporation of the first [␣-32 P]dCTP or to partial degradation of the tRNA 3 Lys during reverse transcription. A negative control showed that RT was not able to anneal , genomic RNA, and RT were co-incubated to allow for protein-assisted tRNA 3 Lys annealing. After annealing, Mg 2ϩ was added to initiate reverse transcription. B, protein annealing assay in which protein, tRNA 3 Lys , and genomic RNA were first co-incubated to allow for protein-assisted tRNA 3 Lys annealing. Then, proteinase K (PK) digestion and phenol-chloroform extraction were performed to remove any protein.
Finally, the reverse transcription reaction was initiated using the protein-annealed primer/template. C, heat annealing in which tRNA 3 Lys was first heat-annealed to genomic RNA. In a second step, protein was incubated with the heat-annealed RNAs, and then reverse transcription reaction was initiated.
In Vitro tRNA 3 Lys Annealing by mGag tRNA 3 Lys on its own in the absence of a Gag-related protein. In contrast, a positive control experiment performed with heat annealing showed that RT was enzymatically active when employed in the context of previously placed tRNA 3 Lys . We next wished to determine the effects of the addition of a nonspecific competitive RNA transcript. Hence, similar reactions were carried out using serial dilutions of such a nonspecific transcript and both mGag and NC (data not shown). At ratios of nonspecific competitive RNA to HIV-1 genomic RNA above 2:1, virtually no elongation of the DNA product occurred in the presence of NC, and an ϳ50% reduction occurred with mGag. We therefore used a ratio of 2:1 competitive RNA:viral genomic RNA in all other competitive reactions that were performed. As expected, higher concentrations of NC and mGag were required in the presence of the competitor in order for equivalent levels of annealing to take place. Interestingly, the ability of high protein concentrations to abrogate RT activity was observed for mGag in the absence of competitor RNA but was even more pronounced in cases in which competitor RNA was present. In contrast, this abrogation of RT activity was not seen when NC was used to anneal tRNA onto the PBS (Fig. 3b). The fact that mGag, an N-terminal extended form of NC, was superior to mature NC in ability to extend tRNA 3 Lys suggests that regions within mGag that are outside of NC probably contribute to tRNA 3 Lys annealing, RT processivity, or both.

M318A mGag Has Similar tRNA Annealing Capabilities as Wild-type mGag but Reduced Sensitivity to the Inhibitory Effects of High Protein Concentrations in Initiation of Reverse
Transcription-To investigate the possibility that mGag multimerization at high concentrations might have played an inhibitory role in our in vitro assay, we employed a mutated mGag protein in which the methionine at position 318 had been mutated to an alanine in the C-terminal domain region of CA. This mutation has been shown to both abolish capsid dimerization and to alter virus particle formation (39,40). The same elongation reaction conditions as those described above were employed (Fig. 2a) together with the genomic RNA template 1-ECP (Fig. 1a). The results show that mGag M318A was superior to wild-type mGag when used at high concentrations in the presence of competitive RNA (Fig. 4). Under these reaction conditions, mGag M318A behaved similarly to NC, and increased elongation was seen at higher concentrations. In contrast, in the absence of competition, wild-type mGag was superior to mGag M318A in its ability to promote elongation at low and intermediate protein concentrations (Fig. 4a, compare  lanes 3-5 of the first and second panels). Thus, the mGag M318A mutation, which results in diminished multimerization, appears to have rescued the ability of mGag to anneal tRNA 3 Lys and/or to initiate polymerization, even if it lost a portion of this ability in experiments performed without competition.
High Protein Concentration Does Not Abrogate tRNA Annealing onto the PBS-As shown, large mGag multimers inhibited our in vitro tRNA 3 Lys annealing assay at high protein concentrations. Therefore, we next wished to determine which step of the reaction had been blocked, i.e. either the annealing of tRNA 3 Lys or reverse transcription itself. Accordingly, we performed in vitro reactions as described above under two different sets of conditions. In the first, we performed an annealing reaction using each of the various proteins, tRNA 3 Lys , the 1-ECP RNA template and competitive RNA at the indicated concentrations. After incubation at 37°C for 10 min, we carried out a proteinase K digestion followed by phenol-chloroform extraction to remove any protein. In this circumstance, only protein- Lys heat-annealed to genomic RNA. b, elongation assay performed in the presence of a competitive RNA. Same disposition and protein concentrations as for a. Lane 0 is a negative control without protein. Lane RC is a competitive RNA control, no template RNA. Lane Cϩ is a positive control without protein but with tRNA 3 Lys heat-annealed to genomic RNA. c, the results of a were analyzed by molecular imaging and plotted as a percentage of the positive control (lane Cϩ). These data were analyzed using the equation Y ϭ B max * X/(K d ϩ X) (one-site binding hyperbola) as described under "Materials and Methods." The dissociation constant (K d ) and the regression (R 2 ) for NC and mGag are shown. mGag was shown to be 8 -10-fold superior to NC at promoting the annealing of tRNA 3 Lys onto viral genomic RNA.
In Vitro tRNA 3 Lys Annealing by mGag primed RNA molecules are left and ensuing polymerization is dependent on the amount of annealed tRNA 3 Lys that was formed. Under these conditions, mGag was maximally able to anneal tRNA 3 Lys when used at high concentrations (Fig. 5a). Consistent with previous results, mGag was significantly more efficient in this regard than NC and was slightly more efficient than the mutated mGag M318A at annealing tRNA 3 Lys . These results showed that tRNA 3 Lys annealing is not inhibited by high protein concentrations of mGag.
High Protein Concentrations of mGag Impairs RT-Subsequent to the above, we were interested in knowing whether polymerization might be impaired by high protein concentrations of mGag. Therefore, we pre-annealed tRNA 3 Lys to the HIV-1 template RNA termed 454-ECP by heating and slow cooling to room temperature (Fig. 2c). We next added each of the various proteins, competitive RNA and RT at the concentrations indicated and incubated to allow for protein-RNA binding. Then, Mg 2ϩ was added to initiate reverse transcription (Fig. 2c). In this system, the tRNA 3 Lys is prebound to the RNA template, negating any effects of protein-facilitated tRNA 3 Lys annealing and allowing us to study the effects of protein on polymerization. Fig. 5b shows that the highest concentrations of mGag resulted in virtually no DNA product and that lower protein concentrations yielded better results in this regard. In contrast, the highest concentrations of mature NC gave rise to the highest levels of DNA. With the mutated mGag M318A, decreased efficiency of polymerization occurred at the highest concentrations of protein, but we did not witness the complete abolition of product formation as had been seen with mGag. Hence, the negative effect of mGag at high concentration may have been because of its ability to multimerize and such multimerization does not apparently inhibit Gag-mediated tRNA 3 Lys priming. These results also confirm the superiority of mGag in ability to anneal tRNA 3 Lys . The Impact of RNA Template on tRNA 3 Lys Annealing and RT Processivity-We also wished to investigate the effects of longer HIV-1 genomic RNA templates on tRNA 3 Lys annealing, given that inclusion of SL5 and ECP (Fig. 1a) seemed to promote greater protein affinity for viral genomic RNA. For this purpose, we generated two other HIV-1 RNA transcripts termed 1-SL4 and 1-ECP (Fig. 1a). In elongation RT assays, no signif-icant differences were observed among the three different genomic RNA templates employed (1-PBS, 1-SL4, 1-ECP) (Fig.  6a). Interestingly, mGag showed higher tRNA annealing capacity in protein annealing assays when the 1-PBS genomic RNA template was used as opposed to either 1-SL4 or 1-ECP (Fig. 6b). This suggests that longer RNA templates can reduce the tRNA annealing capabilities of mGag. In contrast, abrogation of RT activity was observed in heat annealing assays at low protein concentrations when either 1-SL4 or 1-ECP was employed as template in comparison with 1-PBS (Fig. 6c). This suggests that inclusion of additional RNA sequences has the ability to enhance abrogation of RT activity in these assays. This may be due to specific RNA sequences and/or structures contained within the template or to changes in the protein: RNA ratio. It is not clear why differences among templates were not observed in ability to anneal tRNA onto viral RNA and to affect reverse transcription processivity in RT elongation assays, whereas we did observe differences when we specifically assessed these activities in other types of assays, i.e. protein annealing and heat annealing (Fig. 6d, see "Discussion").
Annealing and Elongation Properties of mGag2 and Gag⌬p6 -We were interested in understanding the effects of other domains of Pr55 Gag in the tRNA 3 Lys annealing and elongation process. We were particularly interested in SP2, as others have shown that this small spacer can increase some NC activities (41), and in full-length Gag, as others have found no difference in tRNA 3 Lys placement activity on the part of this protein when compared with NC (22). Several investigators have also reported that fulllength Gag is relatively unstable; therefore we used a truncated Gag⌬p6 version (42).
With this objective, we cloned and purified Gag⌬p6 as well as a mGag protein containing SP2, which we termed mGag2. In protein annealing experiments, mGag2 protein did not differ significantly from mGag in its tRNA 3 Lys annealing capacity (Fig. 7, a and c), but, interestingly, mGag2 showed a greater ability to abrogate reverse transcription at low protein concentrations than did mGag in elongation assays (Fig. 7b). It is possible that the SP2 motif, included in mGag2, added another protein-protein interface to mGag, resulting in increased multimerization capabilities of this protein and an increased inhi- In Vitro tRNA 3 Lys Annealing by mGag 5. Comparisons of mGag, mGag M318A, and NC in heat annealing and protein annealing experiments. a, protein annealing assay. Lanes 1-5 correspond to protein concentrations of 10, 7.5, 5, 2.5, and 1 M, respectively. Lane C is a control reaction in which heat annealing was attempted in the absence of protein. mGag and mGag M318A were able to anneal tRNA 3 Lys maximally at their highest concentrations, whereas NC was less efficient at annealing of tRNA 3 Lys than mGag. These results suggest that tRNA 3 Lys annealing is not inhibited by protein multimerization. b, heat annealing assay. The results with Gag⌬p6 were surprising, because the latter protein showed similar tRNA 3 Lys annealing capacities compared with mGag or mGag2 (Fig. 7, a and c), and others have reported that Pr55 Gag has similar tRNA annealing activity as does NC. Unexpectedly as well, Gag⌬p6 in elongation assays showed almost complete abrogation of reverse transcription at almost all concentrations tested. The increased multimerization capacity of this protein appears to be the most likely explanation for these findings. DISCUSSION In this work, we used a purified mGag protein containing all relevant motifs of Pr55 Gag protein for VLP formation, to study the annealing of tRNA 3 Lys onto the PBS and the initiation of reverse transcription. We also employed a CA dimerization mutant, an extended version of mGag including SP2 (i.e. mGag2) and an almost full-length version of Pr55 Gag (i.e. Gag⌬p6). We have shown that NC in the context of the precursor protein is more efficient at the annealing of tRNA 3 Lys than mature NC and that tRNA 3 Lys annealing is not affected by Gag multimerization, which does, however, compromise reverse transcription.
The finding that mGag is superior to NC in promoting the annealing of tRNA 3 Lys was unexpected, because current belief has been that all of the tRNA 3 Lys annealing ability of Gag is contained within the NC domain. Indeed, previous reports found that full-length Pr55 Gag was equivalent to mature NC in this regard (22), but our results with all N-terminal extended versions of NC showed that NC in the context of the precursor has a superior ability at tRNA annealing when compared with mature NC. We have no clear explanation for the different results obtained here versus those of others. The reported instability of Pr55 Gag used in previous work could be one explanation, whereas another might be the different protein purification and solubility criteria used here. In agreement with the latter idea, Gag⌬p6 in elongation assays showed almost complete abrogation of RT, but tRNA annealing was not affected as shown in the protein annealing assay. The known multimerization capacity of this protein appears to be the most possible and logical explanation for the results obtained. Others have also experienced protein-RNA aggregation artifacts/problems but claimed this was only observed exceptionally. In our studies, gel shift assays using mGag after 15 min of centrifugation at 16,000 ϫ g (protein solubility criteria used by others) resulted in protein-RNA complexes that were never able to run into the gel matrix, in contrast to results obtained for NC. This problem was solved by ultracentrifugation of mGag at 100,000 ϫ g (35).
These results suggest that domains outside of NC might improve tRNA 3 Lys annealing. Here we showed that the inclusion of the C-terminal domain CA markedly improves the ability of NC to anneal tRNA 3 Lys to 5Ј-HIV-1 genomic RNA. mGag includes a putative ␣-helix across the CA-SP1 cleavage site that is not present after protease cleavage. This region has been shown to be involved in the specificity of HIV-1 genomic RNA packaging (33,34) and may play a role in HIV-1 protein-RNA interactions required for tRNA 3 Lys annealing. Another possibility is that the multimerization ability of mGag itself might promote a protein conformational change that facilitates tRNA 3 Lys annealing, i.e. NC annealing of tRNA 3 Lys might be enhanced as a protein multimer, which occurs only in the presence of N-terminal extensions (i.e. mGag). Interestingly, we found that the mGag capsid dimer- FIG. 7. Comparison of tRNA annealing and RT elongation activities of mGag, mGag2, and Gag⌬p6. a, protein annealing assay performed in the absence of competitor RNA. b, RT elongation assay performed in the absence of competitor RNA. In both panels, protein concentrations used are shown under the bands (in M). Lane 0 is a negative control without protein, and lane Cϩ is a positive control in which the tRNA was heat-annealed onto the template. c, the results of a were analyzed by molecular imaging and plotted as a percentage of the positive control (Cϩ). The equation employed to analyze these data was Y ϭ B max * X/(K d ϩ X) (one-site binding hyperbola) as described under "Materials and Methods." No significant differences were observed among the various proteins studied (p Ͼ 0.18). ization mutant, M318A, possessed a reduced but almost similar ability to anneal tRNA 3 Lys than wild-type mGag. This suggests that a possible role of multimerization is overlapped by potential annealing sequences found within N-terminal extensions of NC (see below).
Previous results from our laboratory showed that affinity of mGag for HIV-1 genomic RNA was increased when longer genomic RNA templates were used to include an ECP (35). Accordingly, we expected to see a similar increase in annealing ability with the use of longer HIV-1 genomic RNA templates. Nevertheless, we found instead in RT elongations assays that genomic RNA templates that included sequences up to the PBS were equivalent in annealing ability to templates that included sequences through SL4 and the ECP. However, mGag showed higher annealing activity in protein annealing assays with 1-PBS than when 1-SL4 or 1-ECP were used as template, and, paradoxically, the abrogation of RT was clearly reduced when the shorter template was used in comparison to longer ones. One possible explanation is that the reduction of tRNA annealing by mGag when a longer genomic RNA is used, i.e. 1-SL4 or 1-ECP, is because of the fact that at least two strong binding sites for mGag are introduced by the probes (i.e. SL2 and SL3). These two elements could compete for protein binding, reducing the amount of protein available to anneal the tRNA onto the PBS.
But, then, why did we not see any difference among templates in RT elongation assays? Possibly, the tRNA-genomic RNA-mGag interaction might be able to adopt a different conformation when certain elements are present in the probe, most probably SL2 and SL3, making this interaction more stable or efficient than when these elements are not present (i.e. 1-PBS). This particular conformation would also facilitate mGag multimerization, explaining the results obtained with 1-SL4 and 1-ECP in heat annealing assays in comparison to what was observed for 1-PBS. Alternatively, the longer probes might fold differently than 1-PBS, as they lend themselves to long distance base-pairing interactions (43)(44)(45)(46), and this differential RNA folding might explain the results obtained. Thus, a complex underlying mechanism seems to be at work that involves more than RNA coating by protein. In any case, we conclude that 1) tRNA annealing and RT abrogation involve two distinct and independent but related mechanisms, 2) longer RNA probes with complex tertiary structures, or modification of the RNA:protein ratio, is/are able to more easily promote the arrest of reverse transcription, and 3) the mechanism of abrogation of RT activity involves the multimerization ability of the protein and not simply the coating of the RNA, as more protein would be required to coat longer RNA probes.
Our results with the different templates employed (as discussed above) and with the different proteins used (i.e. wildtype mGag, M318A mGag, mGag2, and Gag⌬p6), showing that each of them possessed similar tRNA 3 Lys -annealing capacity but different abilities to abrogate reverse transcription, suggest that the multimerization of Gag-related proteins is what inhibits elongation. If this is true, then PR-mediated maturation of Pr55 Gag might cause a switch from a mode that favors multimerization to one that promotes disruption of the viral particle. Our results at different concentrations of mGag, suggest that Gag multimerization and the formation of VLPs might contribute to the inhibition of RT processivity and that maturation could alter Gag from being an inhibitory multimer to the facilitative role of monomeric NC. Another possibility is that PR cleavage results in the liberation of NC from other Gag motifs that promote Gag multimerization. This would leave NC tightly bound to viral genomic RNA in monomeric fashion, which is facilitative of polymerization.
It is important to note that even if mGag was superior to NC in its tRNA 3 Lys -annealing properties, a ratio of nonspecific RNA: HIV-1 genomic RNA Ͼ2:1 markedly reduced this ability. This is possibly because of a dilutional effect of the protein, which bound nonspecifically to the competitive RNA and hence reduced the amount of free protein available for tRNA 3 Lys annealing. Current estimates are that one molecule of NC can bind to ϳ8 nucleotides (10 -14), consistent with the protein:RNA ratios used in these experiments. This shows that other elements are also involved in the tRNA 3 Lys annealing process in vivo (i.e. Lysyl-tRNA synthetase (47) and Pr160 Gag-Pol (48)). It is also worth noting that initiation of reverse transcription is modulated by interactions of tRNA 3 Lys with other elements in genomic RNA (i.e. A-rich loop and primer activation signal) (6, 49 -52). Our results suggest that mature NC might facilitate the interaction of tRNA 3 Lys with these elements in promotion of initiation of reverse transcription, whereas NC, in the context of Pr55 Gag , might not be able to efficiently facilitate these interactions, as shown by our data. However, the assays employed in this study were not designed to address how these different proteins affect interactions of relevant RNA sequences with RT and tRNA 3 Lys , and this represents ongoing work. Indeed, more precise biochemical tRNA annealing assays are now being developed to better address this point.
In Rous sarcoma virus, a hyperaggregative Gag protein has been identified that does not package viral genomic RNA (53). This finding, together with our own work on Gag multimerization effects on reverse transcription, points to the possibility that attempts to force linkage of Gag molecules into aggregates could lead to inhibition of viral replication.