|Home | About | Journals | Submit | Contact Us | Français|
Assembly of Human Immunodeficiency Virus Type-1 (HIV-1) particles is initiated in the cytoplasm by the formation of a ribonucleoprotein complex comprising the dimeric RNA genome and a small number of viral Gag polyproteins. Genomes are recognized by the nucleocapsid (NC) domains of Gag, which interact with packaging elements believed to be located primarily within the 5´-leader of the viral RNA. Recent studies revealed that the native 5´-leader exists as an equilibrium of two conformers, one in which dimer-promoting residues and NC binding sites are sequestered and packaging is attenuated, and one in which these sites are exposed and packaging is promoted. To identify the elements within the dimeric 5´-leader that are important for packaging, we generated HIV-1 5´-leader RNAs containing mutations and deletions designed to eliminate substructures without perturbing the overall structure of the leader and examined effects of the mutations on RNA dimerization, NC binding and packaging. Our findings identify a 159 residue RNA packaging signal that possesses dimerization and NC binding properties similar to those of the intact 5´-leader and contains elements required for efficient RNA packaging.
Like all retroviruses, the Human Immunodeficiency Virus (HIV) specifically packages two copies of its unspliced RNA genome, both of which are utilized for strand-transfer mediated recombination during reverse transcription.1,2 Although only one DNA allele is generated by this mechanism (and retroviruses are thus considered “pseudodiploid”),3 the production of recombinant proviruses from heterozygous virions appears to serve as a primary means for promoting genetic evolution under conditions of environmental and chemotherapeutic stresses.4–7 Genome selection is mediated by interactions between the nucleocapsid (NC) domains of a subset of the viral structural proteins (called Gag) and RNA elements located primarily within the 5´-leader region of the genome.8–18 Genetic studies indicate that genomes are selected for packaging as dimers,19,20 and there is now compelling evidence that dimerization and selection occur in the cytoplasm subsequent to nuclear export.21–23 HIV-1 genomes are trafficked to the plasma membrane by a small number of Gag proteins, where additional Gag proteins co-localize and virus assembly occurs.24,25 Subsequent to budding, the packaged RNA molecules exist as weakly formed, non-covalently linked dimers that form more stable dimers as the virus ages.26–31
Agents that disrupt the structure of the NC domains of Gag and interfere with genome packaging have potent antiviral properties,32–36 and a better understanding of the structural determinants of genome selection could facilitate the development of new therapeutic approaches for the treatment of AIDS. However, despite considerable effort, current understanding of the protein-RNA interactions37 and mechanisms that promote and/or regulate retroviral genome selection remains limited (for reviews see 8–17,38,39). For some retroviruses, including the Moloney Murine Leukemia Virus (MoMuLV) and Rous Sarcoma Virus (RSV), relatively small regions of the 5´-leader RNAs (fewer than 100 nucleotides) have been identified that are capable of directing the packaging of heterologous RNAs into virus-like particles (VLPs), and this has facilitated high resolution structural studies of the NC:RNA complexes responsible for RNA packaging.40–42 Although atomic-level structures have been determined for the HIV-1 NC protein bound to isolated hairpin and single-stranded RNAs,43–45 these studies did not explain how the virus discriminately packages its dimeric, unspliced genome. To date, no high-resolution structural information is available for a packaging competent HIV-1 RNA.
Efforts to identify a minimal HIV-1 packaging signal have been complicated by a number of factors. First, the 5´-leader contains elements that are important for functions that are unrelated to packaging but critical for replication (including transcriptional activation, primer binding during reverse transcription, splicing, and dimerization) (Fig. 1), and mutagenesis studies designed to test effects on packaging can adversely affect other functions. For example, mutagenesis studies originally suggested that the TAR hairpin is required for genome packaging,46–48 but more recent studies indicate that effects originally attributed to a packaging defect were actually due to dominant negative effects on viral gene expression.49 A second complication is that correct Rev-mediated RNA localization is a prerequisite to lentiviral RNA packaging, so that even RNAs that possess intact packaging signals are inefficiently packaged without Rev.50 A third issue is that HIV-1 VLPs can efficiently assemble in the absence of their native genomes and, in the process, package similar amounts of cellular RNAs.51–55 Another complication relates to the fact that some previous packaging studies were conducted using transfection experiments in which the cellular concentration of mutant vector RNAs was high. At high cellular concentrations, the mutant RNAs can be packaged at high levels, as long as they do not have to compete with RNAs containing the wild-type 5´-leader. In addition, efforts to determine the secondary structure of the HIV-1 5´-leader using chemical probing, mutagenesis and biochemical approaches, and phylogenetic analyses56–68 have led to multiple structural proposals and mechanistic predictions, thus hindering the design of mutations to properly test structure/function hypotheses.39
Recently, using an NMR approach for studying relatively large RNAs, we showed that the HIV-1 5´-leader exists as an equilibrium of two conformationally distinct species, one in which residues spanning the gag start codon (G328-A356; hereafter called “AUG”) form a hairpin and the GC-rich palindromic loop of the dimer promoting DIS hairpin is base paired with an upstream element (U5), and another in which AUG is base paired with U5 and the dimer promoting loop of DIS is exposed.69 U5:AUG formation, proposed originally on the basis of phylogenetic and biochemical analyses,62,70 promotes dimerization by displacing the DIS and simultaneously exposes high-affinity NC binding sites, thereby promoting the packaging of a dimeric genome.69 Having identified the conformer of the 5´-leader that directs packaging, we wished to explore the potential roles of specific elements within the leader on RNA packaging. Our present study identifies a 159 nucleotide segment of the 5´-leader that is sufficient for RNA dimerization and NC binding in vitro and is required for efficient and selective RNA packaging.
A primary goal of these studies was to systematically remove elements from 5´-L (residues 1–356) without altering the folding of other structural elements, as assessed by the NC-binding, dimerization, and NMR spectroscopic properties of the RNA. Under conditions of physiological-like ionic strength (PI buffer = 140 mM KCl, 10 mM NaCl, 1 mM MgCl2, 10mM TRIS-HCl, pH 7.0) and at an RNA concentration of ~ 1.0 µM, 5´-L exists as an equilibrium mixture of two species: a monomeric conformation in which the GC-rich palindrome of the DIS hairpin is sequestered by base pairing with U5 and AUG forms a hairpin, and a dimeric conformation in which AUG base pairs with U5 and the DIS is exposed (Fig. 1c) (dimer dissociation constant Kd = 0.9 ± 0.1 µM).69 Mutations in AUG that favor U5:AUG base pairing promote dimerization and NC binding of 5´-L RNAs in vitro and the packaging of analogous vector RNAs into VLPs (Fig. 1c),69 whereas the complete truncation of the AUG hairpin inhibits dimerization by preventing U5:AUG dependent displacement of the DIS.69 The present studies focused on 5´-L constructs that lacked the 3’-most residues of AUG (A345-A356 deleted; 5´-L344) (Fig. 1b).69 This deletion allows U5:AUG formation while simultaneously precluding formation of the competitive AUG-hairpin structure, thereby shifting the equilibrium in favor of the dimer (Fig. 1c,d).
2D NOESY spectra obtained the fully protonated [5´-L344]2 RNA exhibited severe signal overlap and were unassignable using standard methodologies.69 However, a significant number of resolved adenosine C2-H signals were detectable in 2D NOESY spectra obtained for RNAs in which all non-exchangeable aromatic protons except the C2-H of the adenosines were substituted by deuterium. For example 1H NMR signals with chemical shifts and NOE cross peak patterns matching those observed for the isolated TAR, DIS and Ψ-site oligo-RNA hairpins were readily identified in 2D NOESY spectra obtained for a dimeric 5´-L344 sample that contained protons on the non-exchangeable adenosine C-2 and ribose (A2,R), the guanosine ribose (GR), and the cytosine ribose (CR) carbons and deuterons at all other non-exchangeable aromatic and ribose carbons (A2,R,GR,CR-[5´-L344]2, Fig. 1e) (8, 2, and R = protonated on the C-8, C-2, and ribose carbons, respectively). These outlier signals and their NOE interactions with H1´ protons were monitored to assess potential effects of mutations and deletions on the structure of the RNA.
Additional leader RNAs were prepared that contained deletions of the TAR hairpin (5´-L344-ΔTAR), the upper region of the PBS loop (5´-L344-ΔPBS), the TAR and upper PBS loop (5´-L344-ΔTAR-ΔPBS), and the TAR and Poly(A) hairpins and upper PBS loop (5´-L344-ΔTAR-ΔPoly(A)-ΔPBS) (Fig. 2a). RNA homogeneity was confirmed by denaturing polyacrylamide gel electrophoresis (PAGE, Fig. 2b). Base pairing patterns of the TAR and Poly(A) hairpins and the lower stem of the PBS loop have been supported by several studies,62,71–75 including compensatory mutagenesis-based packaging experiments.76 The PBS deletion construct was designed such that the entire upper region of the loop (residues A132–A216) was substituted by a GAGA tetraloop (a member of the well-characterized GNRA family77–84). Resolved A220 C2-H signals were observed for all constructs in partially deuterated samples (prepared as described above), including those containing both the native and truncated PBS loops (Fig. 2c), indicating that base pairing in the lower stem of the PBS loop is maintained in both the native and mutant 5´-L RNAs. 2D NOESY data obtained for the 5´-L344 mutants did not exhibit line broadening or additional NMR signals (other than those expected for the GAGA tetraloop) that would be indicative of alternative folding. In particular, the outlier adenosine C2-H signals of the most highly truncated construct, 5´-L344-ΔTAR-ΔPoly(A)-ΔPBS, exhibited 1H NMR chemical shifts and 2D NOESY cross peaks similar to those observed for the wild-type 5´-L344 RNA (Fig. 2d). Signals assigned to the TAR, Poly(A) and upper PBS loop structures were missing in spectra obtained for the corresponding deletion mutants, supporting the 1H NMR chemical shift assignments for these elements.
The 5´-L344, 5´-L344-ΔTAR, 5´-L344-ΔPBS, 5´-L344-ΔTAR-ΔPBS and 5´-L344-ΔTAR-ΔPoly(A)-ΔPBS RNAs all exist predominantly as monomers at low ionic strength but form dimers upon incubation in PI buffer (Fig. 1d and Fig 3a), indicating that the TAR, Poly(A), and upper PBS hairpin structures do not play a significant role in dimerization. The potential roles of these structures in NC binding was assessed by isothermal titration calorimetry (ITC).69 As observed previously in ITC studies of native and mutant forms of 5´-L, the binding isotherms obtained for 5´-L344 upon NC titration in PI buffer included an initial exothermic component followed by an endothermic component (Fig.3b). The exothermic events are attributed to high affinity NC binding since similar ITC profiles have been observed for oligo-RNAs that contain a single high affinity NC binding site43–45. The endothermic events occur during the later injections at progressively higher [NC]:[RNA] ratios and are attributed to weak NC-RNA interactions and NC-induced RNA unfolding.69 5´-L344 exists as a dimer under conditions of the ITC experiments, and non-linear least squares fitting using a two-event binding model indicates that [5´-L344]2 binds 15 ± 1 NC molecules per RNA strand with high affinity (Kd = 17 ± 9 nM) (Fig. 3b and Table 1). Both the number of high affinity NC sites and the binding affinity are similar to those measured for 5´-L RNAs containing mutations in AUG that favor U5:AUG base pairing (5´-LU5:AUG; 16 ± 1 NC sites per strand; Table 1).69
The ITC isotherms obtained upon titration of 5´-L344-ΔTAR, 5´-L344-ΔPBS, 5´-L344-ΔTAR-ΔPBS and 5´-L344-ΔTAR-ΔPoly(A)-ΔPBS with NC were similar in appearance to those obtained for 5´-L and 5´-L344 (Fig. 3b,c), except that these truncated RNAs bound fewer NC molecules with high affinity. Deletion of the TAR and PBS loops resulted in losses of 2 ± 1 and 4 ± 1 NC binding sites, respectively, and the simultaneous deletion of both the TAR and PBS loops eliminated 6 ± 1 NC sites (Table 1), indicating that NC binds independently and non-cooperatively to these hairpins. The isolated oligo-TAR RNA was previously shown to interact with NC,85 and an oligo-RNA corresponding to the isolated TAR hairpin was found by ITC to bind a single NC molecule with high affinity (Kd = 110 ± 2 nM) (Heng & Summers, unpublished results). Removal of the Poly(A) hairpin from 5´-L344-ΔTAR-ΔPBS resulted in the loss of a single additional high affinity NC site (Table 1). Thus, TAR, PBS and Poly(A) hairpins appear to contribute approximately 2, 4, and 1 high affinity NC binding sites, respectively. Interestingly, we also observed small but systematic reductions in the magnitude of the endothermic component of the ITC isotherms upon removal of TAR, Poly(A) and PBS elements (Fig. 3b,c), with isotherms for 5´-L344-ΔTAR-ΔPoly(A)-ΔPBS exhibiting only a slight endothermic feature at the higher NC/RNA ratios (Fig. 3c). Thus, the NC-induced RNA unwinding that occurs after saturation of the high affinity NC binding sites appears to diminish with decreasing size of the RNA.
When HIV-1 particles bud from cells in the absence of genomic RNAs with packaging signals, these particles encapsidate “random” host mRNAs roughly in proportion to their intracellular concentrations.86 In contrast, most such adventitiously packaged mRNAs are displaced when RNAs with authentic packaging signals are present, due to the specificity of HIV-1 for its own genome. Here, a packaging assay based on these observations was used to address whether or not the Tar, Poly(A), and PBS loop deletion RNAs studied above possessed information sufficient to mediate authentic HIV-1 RNA packaging. In these experiments, the RNA transcribed from the HIV-1-based helper function plasmid pCMVΔR8.2 (herein called Ψ(−) helper),87 which expresses all virion proteins except env but lacks all HIV-1 5´ leader sequences upstream of the gag start codon, served as an internal control for the extent of packaging of an RNA that lacks all HIV-1 packaging elements. In the present studies, we adjusted transfection levels such that the test and Ψ(−) helper RNAs were transcribed in cells at similar levels, thereby enabling a more quantitative assessment of relative packaging efficiencies, Fig. 4. RNA packaging levels were measured by using RT activities (Fig. 4a) or viral 7SL RNA levels (Fig. 4b) as controls. 7SL RNA is a cellular transcript that is incorporated independent of genomic RNA and can be used as an internal control for quantitating packaging efficiencies.88 Relative efficiencies were determined by comparing packaging levels of test RNAs that contained 5´-leader mutations with those that contained the intact, wild-type leader and gag sequences. All test vectors contained Rev responsive elements (RREs) and were expressed in the presence of Rev, and thus Rev-dependent RNA localization, which is required for efficient packaging,50 was not a variable in these experiments.
As shown in the RNase protection assay in Fig. 4a, when Ψ(−) helper was transfected into 293T cells in the absence of RNAs possessing HIV-1 packaging signals, the transcript was highly expressed in cells (lane 13) and was encapsidated into virions that were released from these cells with modest efficiency (lane 6). In contrast, whereas both transcripts were detected at similar levels within cells (lane 8), when CMVΔR8.2 was co-transfected with an HIV-1 test vector possessing the native NL4-3 5´-leader (Native), RNAs with native NL4-3 packaging elements were significantly enriched among virion RNAs, demonstrating that the leaderless Ψ(−) helper transcripts could not compete for packaging (lane 1). TAR, Poly(A), and PBS loop deletion test RNAs also were co-expressed with Ψ(−) helper. Although these vectors lacked TAR regulatory elements, they were well-expressed under the transfection conditions used, which were selected to ensure that the test vector and Ψ(−) helper RNA levels were similar to one another within cells (lanes 9–12). However, similar to results obtained for the native NL4-3 leader, the Ψ(−) helper RNAs were unable to compete efficiently with these test RNAs (lanes 2–5).
Similar results were obtained in RNase protection assays that quantified vector RNA packaging by normalizing to the co-packaged host RNA 7SL, which serves as a surrogate for virion copy number89 (Fig. 4b). Similar amounts of 7SL RNA-protected products were present in all viral samples from cells transfected with CMVΔR8.2, whether alone or in the presence of test plasmids (lanes 1–6). As expected, HIV-1 specific transcripts were robustly packaged in the sample from cells co-transfected with CMVΔR8.2 plus the parental 5´-L HIV-1 test vector (lane 1), and a significant drop in packaging was observed for the Ψ- transcript produced when CMVΔR8.2 was expressed alone (lane 6). In contrast to this Ψ- transcript, 5´-L RNAs from which either the PBS loop or TAR were deleted (lanes 2–3), those that lacked both TAR and the PBS loop (lane 4), and the TAR, Poly(A), and PBS loop triple-deletion test RNA (lane 5), were all encapsidated at levels similar to those of native 5´-L RNA (lane 1). As quantified in Table 2, only small decreases in overall packaging were observed, with all of these deleted 5´-L RNAs displaying packaging ≥ 80% that observed for the native 5´-leader.
To determine if elements in gag downstream of the AUG are important for packaging, we generated packaging vector variants that retained the 5´ leader sequences studied by NMR above, but that lacked HIV-1 gag-derived sequences downstream of position +350 that were present in the native 5´-L vectors examined above. As shown in RNase protection assay results in Figure 5, co-expression of either the original 5´-L or a truncated 5´-L350 vector lacking gag sequences with the CMVΔR8.2 helper construct yielded indistinguishable RNA levels in cells (lanes 5 and 6) and identical strong packaging levels in virions (lanes 1 and 2; Table 2). These findings demonstrate that sequences in gag are not required for efficient packaging of HIV-1 5´-Leader containing RNAs into VLPs.
The observation that deletion of TAR results in a moderate but reproducible reduction in RNA packaging, whereas deletion of the PBS loop does not affect packaging, is difficult to reconcile with our finding that the PBS upper loop contains a greater number of high affinity NC binding sites. 2D NOESY spectra obtained for [5´-L344-ΔTAR]2 appeared similar to spectra obtained for the parent [5´-L344]2 RNA (except for the missing TAR signals; data not shown), suggesting that the TAR deletion does not affect the overall fold of the dimeric RNA. To determine if the TAR deletion affects the folding of the monomeric form of the leader, we examined the dimerization properties and 2D NOESY spectra of leader RNAs in which AUG was fully truncated (5´-LΔAUG). As indicated above, this RNA lacks the 5´-residues of AUG necessary to displace and expose the DIS hairpin, and like full-length 5´-L RNAs containing mutations that stabilize the AUG hairpin, 5´-LΔAUG exists predominantly as a monomer in PI buffer (Fig. 6a). Unexpectedly, deletion of the TAR hairpin from 5´-LΔAUG (5´-LΔAUG -ΔTAR) strongly promoted dimerization in PI buffer (Fig. 6b). In addition, whereas the outlying adenosine C2-H signals in 2D NOESY spectra obtained for 5´-LΔAUG were similar in appearance to spectra obtained for [5´-L344]2 (Fig. 6b), the spectrum obtained for 5´-LΔAUG-ΔTAR lacked several expected signals, including the A268 C2-H signal associated with the properly folded form of the DIS hairpin (Fig. 6c). These findings indicate that removal of TAR leads to non-native like folding and dimerization of the 5´-leader.
The TAR, Poly(A) and PBS loops have each been implicated in at least some studies to play roles in genome dimerization, NC binding, and packaging. Isolated TAR oligo-RNA hairpins can form dimers in the presence of NC, and electron microscopy (EM) images of NC incubated HIV-1 5´-RNA fragments (nucleotides 1–744) exhibit distinct circular structures that have been interpreted in terms of a TAR-mediated dimer interface.90 Some studies suggested that deletion of TAR promotes the packaging of monomeric RNAs, to an extent similar to that observed upon deletion of the DIS hairpin.91 On the other hand, truncation of the TAR hairpin from a 290 nucleotide 5´-fragment of the HIV-1 leader promoted, rather than inhibited, dimerization of this RNA.92,93 The present studies show that 5´-L344, which spans the entire HIV-1 5´-UTR, forms a stable dimer even in the absence of the TAR hairpin. 2D NOESY spectra are relatively unperturbed (except for the loss of the TAR signals) upon removal of TAR, indicating that dimerization does not result from artificial misfolding of the dimeric leader. The fact that the A46 C2-H signals of TAR are relatively intense in both the monomeric and dimeric forms of 5´-L RNAs (Fig. 1e) suggests that the hairpin is relatively mobile and probably does not participate in long-range interactions of the dimeric 5´-leader (although TAR could participate in intermolecular RNA-RNA interactions in the more stable dimer that is formed in virions during viral maturation91,94).
Deletion of TAR was shown in one study to reduce packaging efficiency by 74% compared to the packaging of a parent RNA derived from the intact viral genome that encoded a stop codon within the capsid domain of gag.47 Although TAR plays a critical role in transcriptional activation via its interactions with the viral Tat protein95,96 these studies concluded that reductions in packaging were too great to be attributed to the relatively modest reductions in cellular RNA levels.47 Mutations that destabilized the stem of TAR reduced packaging to 70% of wild-type levels (after controlling for reductions in cellular RNA levels)48 and altered the ratio of spliced/unspliced viral RNAs that were packaged.97 Interestingly, using a proviral expression system that is largely independent of Tat for transcriptional activation, Rekosh and co-workers found that missense mutations in the upper stem of TAR had little to no effect on packaging, but mutations at the base of the TAR stem could lead to significant reductions in packaging.46 Compensatory mutations in the TAR stem restored packaging to near wild-type levels, indicating that it is the structure of TAR, and not the TAR sequence, that is important for packaging.46 A more recent study showed that the complete removal of TAR had little impact on replication as long as other changes were made to overcome the requirement for TAR-mediated transcriptional activation (packaging efficiencies were not reported).49 These studies also showed that there is evolutionary pressure to restore base pairing at the 5´-terminal end of the viral RNA,49 and supported proposals that packaging defects induced by mutations in TAR result from misfolding of the 5´-leader. Our findings are consistent with this latter hypothesis. Thus, deletion of TAR did not significantly affect the NC binding properties of the 5´-leader RNA and also did not significantly impair vector RNA packaging. The modest packaging defects that we observed upon deletion of TAR are most likely due to misfolding of the monomeric form of the 5´-leader, which we believe leads to a reduction in the cellular levels of the packageable, dimeric form of the RNA.
The Poly(A) hairpin has also been proposed to play roles in both genome dimerization and packaging. Although early computer-modeling studies and EM imaging of the HIV-1 genome suggested that the palindromic 5´-AAGCUU-3’ loop of Poly(A) might form a second dimer linkage structure,98 subsequent mutagenesis studies indicated that the Poly(A) palindrome does not influence RNA dimerization.99 Here we showed that deletion of the TAR and Poly(A) hairpins does not affect the dimerization properties of 5´-L344, in support of the latter conclusion. Poly(A) was originally thought to be important for RNA packaging because removal of the hairpin reduced packaging to ~10% of wild type levels.47,100 Mutations that disrupt the Poly(A) hairpin structure were also shown to inhibit genome packaging and viral replication, whereas compensatory mutations restored packaging to near wild-type levels indicating that it is the structure, rather than specific sequence of the Poly(A) hairpin that is required for packaging.97 However, more recent studies indicated that disruption of the Poly(A) hairpin structure can lead to reductions in cellular levels of the viral RNA due to reductions in RNA stability, and it was suggested that the Poly(A) hairpin may not actually play a significant role in genome packaging.101 As shown in the present studies, removal of Poly(A) from 5´-L344-ΔTAR does not significantly affect its structure, dimerization or NC binding properties. Furthermore, vector RNAs in which the TAR, Poly(A) and PBS loops are deleted were packaged as well as RNAs that contained only TAR and PBS deletions, indicating that Poly(A) does not play a significant role in RNA packaging.
The PBS element contains a sequence that binds human tRNALys, a requirement for initiation of reverse transcription.102–106 Numerous secondary structural models of PBS have been proposed based on mutational studies, enzymatic probing, UV-cross linking, phylogenetic studies and chemical modification assays (reviewed in 13,39,62,70,72,73,75,107). Deletions within the upper loop of the PBS generally do not significantly impair genome packaging, whereas the lower stem loop is required for RNA dimerization and efficient packaging76,108,109. Our PBS deletion mutant was designed to maintain native base pairing in the lower stem of the PBS loop structure. Significantly, the NMR data presented here show that A220 C2-H signal pattern was observed in [5´-L344]2 and all the PBS upper loop truncated constructs, indicating that the lower stem of the PBS element adopts the predicted secondary structure. Removal of the upper PBS loop structure eliminated ~ 4 high affinity NC binding sites but did not impair RNA packaging, suggesting that these NC sites may play roles in tRNA binding and/or reverse transcription but are not critical for packaging.
In vitro biochemical studies and phylogenetic predictions led to the proposal that loop residues of PBS base pair with residues G443-C449 in the matrix (MA) encoding region of gag.110 These putative long range interactions have been supported by chemical probing and modeling studies,62,74,75 but their role in packaging has not been established. Early packaging studies suggested that efficient HIV-1 genome packaging requires residues within the first several hundred nucleotides of gag,111 and could therefore be dependent on Poly(A)-gag interactions. However, the present studies show that RNAs containing the native 5´-leader (nucleotides 1–355, which include residues required for U5:AUG base pairing) but missing nearly the entire gag open reading frame are packaged with the same efficiency as RNAs containing the intact gag gene. Thus, if the Poly(A)-gag interactions exist in vivo, they likely play roles in functions unrelated to RNA packaging.
The present findings also provide insights into the number and general locations of the high-affinity NC bindings sites. Early studies using UV-cross linking indicated that NC is the only major protein that binds to the genomic RNA in HIV-1 virions, and it was estimated that approximately 15 NC molecules are bound per RNA strand.112 In contrast, in vitro cross-linking experiments involving a fragment of the 5´-leader spanning from DIS through residue 170 of gag were interpreted to indicate that only ~2–3 NC or Gag proteins bind to each RNA strand,113 and RNA footprinting studies performed with a 5´-leader construct that included the intact 5´-UTR and 66 nucleotides of gag were similarly interpreted in terms of only two high affinity NC or Gag binding sites.72 In addition, quantitative filter-binding and ITC assays indicated that NC can bind with high affinity (Kd ≈ 55–400 nM) to fragments of the HIV-1 5´-leader between the DIS and AUG hairpins,43–45,64,76 Our ITC data of 5´-L and the truncated mutants titration with NC show that they all have high affinity NC binding sites with Kd in the same range (Table 1). Our current and recent studies69 are most consistent with the earlier work of Darlix,112 and indicate that the 5´-leader contains approximately 16 high affinity NC binding sites per RNA strand. The locations of the high-affinity NC binding sites were probed recently by monitoring the effects of zinc ejection on SHAPE activity.75 These studies suggested that the 5´-leader contains 7 NC binding sites, none of which were proposed to reside within the TAR, PolyA or upper PBS loops.75 Although these results are incompatible with our present findings, we cannot rule out the possibility that structural changes associated with viral maturation, or constraints within the context of the packaged RNA, could lead to altered NC binding.
Sakuragi and coworkers have used a novel approach to identify the minimal determinants of HIV-1 genome dimerization and packaging that involves duplication of the 5´-leader in downstream (ectopic) regions of the RNA, which can lead to the unusual packaging of monomeric RNAs (in addition to dimeric RNAs).114 In this system, RNAs with ectopic sequences lacking the Poly(A) and SD hairpins (to prevent unwanted splicing and polyadenylation) were efficiently packaged as monomers.114 This group subsequently showed that RNAs lacking the native 5´-leader, but containing two downstream copies of a sequence that includes TAR through the first half of gag, but that lacks the Poly(A) and PBS hairpins (a 19-nucleotide deletion in the upper loop to eliminate ectopic initiation of reverse transcription), exclusively packages monomeric RNAs, although virion production and packaging efficiency were reduced to ~20% and ~63% of wild-type levels, respectively.115 Using a similar packaging system that included a much smaller portion of gag, these investigators showed that deletions in the upper loop of the PBS do not significantly affect dimerization or packaging, but deletions in U5 and/or the lower PBS stem lead to significant reductions in packaging.109 Surprisingly, severe modifications and deletions within the DIS hairpin led to relatively modest reductions in packaging levels (to 18–50 % of WT levels).109 A systematic deletion mutagenesis study revealed that relatively efficient packaging of monomeric RNAs can be achieved by RNAs that contain a fully native 5´-leader and an ectopic element that lacks the TAR, Poly(A), upper PBS loop, and SD sequences.107 Since the RNAs employed contained a native 5´-leader sequence, these studies do not rule out the possibility that packaging could depend on the presence of at least one fully native leader RNA per dimer. However, the findings are nevertheless fully compatible with our observation that, under the experimental conditions used here, relatively efficient packaging can be achieved by RNAs that lack the TAR, Poly(A) and upper PBS loops.
In summary, our findings identify 5´-L344-ΔTAR-ΔPoly(A)-ΔPBS as an essential HIV-1 core encapsidation element. This fragment of the 5´-leader exhibits 1H NMR spectral features similar to those of the intact 5´-leader, possesses native-like in vitro dimerization and NC binding properties, and is required for promoting RNA packaging into VLPs. Packaging deficiencies observed here and by others46–48,97 upon deletion of TAR are likely due to misfolding of the monomeric form of the 5´-leader. The clustering of high affinity NC binding sites, coupled with the known self-association propensities of the CA domain of Gag, could function synergistically to promote cooperative assembly of the Gag:RNA complex that is trafficked to the plasma membrane and nucleates virus assembly. Since the minimal packaging signal comprises only 159 nucleotides, high-resolution structural studies by NMR or combined NMR/cyro-electron tomographic approaches116 should now be feasible.
The 5´-L, 5´-LU5:AUG and 5´-LHP-AUG mutants were cloned into PUC19 as described.69 5´-L344 and 5´-L344-ΔTAR were subcloned from the 5´-L plasmid using 5´-CGG TCG AAT TCT AAT ACG ACT CAC TAT AGG TCT CTC TGG TTA GAC CAC ATC TGA GCC TGG GAG CTC TCT GGC-3’ and 5´-CGG TCG AAT TCT AAT ACG ACT CAC TAT AGG CAC TGC TTA AGC CTC AAT AAA GCT TGC CTT GAG ATC-3’ as forward primer, respectively, and 5´-CCC TAG GAT CCC CCG GGC GCA CCC ATC TCT CTC CTT CTA-3’ as their mutual reverse primer. 5´-L344-ΔPBS was made by MegaPrimer mutagenesis method using 5´-L344 as template. The four primers used are 5´-CGG TCG AAT TCT AAT ACG ACT CAC TAT AGG TCT CTC TGG TTA GAC CAG ATC-3’, 5´-CCT GCG TCG AGA GAT CTC CTC TGG TCT CCC AGA GTC ACA CAA CAG ACG GGC ACA CAC TAC TTT GAG CA-3’, 5´-GAG ACC AGA GGA GAT CTC TCG ACG CAG GAC TCG GCT TGC TGG AGA CGG CAA GAG GCG AGG GGC GGC GAC-3’, and 5´-CCC TAG CTA GCT TCT GCC TCC GCT AGT CAA AAT TTT TGG CGT ACT CAC CA-3’. 5´-L344-ΔTAR,-ΔPBS and 5´-L344-ΔTAR-ΔPoly(A)-ΔPBS were then cloned from 5´-L344-ΔPBS plasmid using the forward primer 5´-CGG TCG AAT TCT AAT ACG ACT CAC TAT AGG CAC TGC TTA AGC CTC AAT AAA GCT TGC CTT GAG ATC-3’ and 5'-CGG TCG AAT TCT AAT ACG ACT CAC TAT AGG TGT GCC CGT CGT TGT GTG ACT CTG GTG AGA GCC AGA GGA GA-3', respectively, and the mutual reverse primer 5'-CCC TAG GAT CCC CCG GGC GCA CCC ATC TCT CTC CTT CTA-3'. All of the mutations were sent to the DNA sequencing facility (University of Maryland, Baltimore County) to ensure the desired mutations were obtained.
T-LNL4-3 - - leader region sequences in place of the corresponding sequences in HIV-lac.117 Variants with 5´-LΔTAR, 5´-LΔTAR-ΔPBS or 5´-LΔTAR-ΔPoly(A)-ΔPBS mutations were constructed by introducing mutations into the - -leader region using overlap extension PCR, and using these to replace the corresponding regions of 5´-LNL4-3. Correct structures of all plasmids were confirmed by sequencing. For the initial series of test vectors, a stop codon was introduced after the 85th codon in the gag open reading frame. For gag deletion mutants, PCR mutagenesis was used to delete sequences downstream of the 15th nucleotide in gag and fuse the 15th residue to the Not I restriction site of HIV-lac.117 Riboprobe RNAs to detect helper and test vector RNAs were templated by pEG676-1, a derivative of pBluescript KS (Stratagene), which includes sequences complementary to 286 bases of CMV promoter shared by helper and test vectors plus additional 14 bases sequences unique to the test vectors, preceded by a T7 promoter.
Human 293T cells were cultured and maintained in Dulbecco's modified Eagle's medium (DMEM; Invitrogen) supplemented with 10% fetal bovine serum (Gemini). Cells were grown at 37°C under 5% CO2. To generate virions for RNA packaging analysis, 293T cells were co-transfected with a total of 20 µg of plasmid DNA that included one test vector plasmid and pCMVΔR8.2, a Ψ-negative HIV-1 helper vector driven by the cytomegalovirus (CMV) promoter87, at an approximate 1:3 molar ratio on 100-mm plates. Modest empirically determined variation in co-transfection ratios was introduced for non-native vectors, to achieve similar test/helper expression ratios for all samples. For transfection by PEI, plasmid DNA was mixed with 4 µg of polyethyleneimine (Polysciences) per µg of DNA in 1 ml of 150 mM NaCl by vortexing at medium speed for 10 s. After room temperature incubation for 15 min, the mixture was added dropwise to medium on cells. The transfection mixture was replaced with fresh DMEM medium 24 h after transfection.
Tissue culture medium was harvested at 48 h post-transfection, pooled, and filtered through 0.2-µm-pore-size filters. Virus-containing medium was concentrated by centrifugation through a 2-ml 20% sucrose cushion in phosphate-buffered saline (PBS) for 2 h at 4°C and 25,000 rpm using a Sorvall Surespin 630 rotor in a Sorvall Discovery 90 ultracentrifuge. Viral pellets were suspended in 0.5 ml of Trizol reagent (Invitrogen), and RNA was isolated according to the manufacturer's instructions. Cellular RNA was obtained by harvesting cells 48 h after transfection by scraping cells into 2 ml of Trizol reagent per 100-mm plate, and RNA was isolated according to the manufacturer's instructions.
RNase protection assays were performed and quantified by phosphorimager as previously described.89 Riboprobes templated by pEG676-1 (see above) were used in Figures 4 and and55 experiments. This riboprobe was designed to contain 14 bases unique to the helper as well as 268 shared bases of complementarity, to allow simultaneous detection of both vector and helper RNAs with a single probe. The experiment shown in Figure 4b included a second riboprobe templated by pBru7SL,88 which detects 7SL RNA. The signals in the cluster of bands detected with the 7SL probe were summed: the appearance of multiple bands is typical and results from allelic heterogenity in this multi-copy gene and some probe nicking.88 For each experiment, the amount of riboprobe loaded in the undigested probe lane was 1% of that used in the hybridization reaction for each sample lane. Protein was removed from the aqueous phase by ATrizol (confirmed by spectrophotometric analysis of the RNA).
The PUC19 vectors containing different RNA clones were transformed into DH5α cells. Large scale of DNA template preparation was performed using QIAGEN Plasmid Mega Kit (Qiagen). In vitro transcription by T7 polymerase118 is performed in solution with a mixture of transcription buffer, MgCl2, NTPs, DNA template. The amount of MgCl2 and NTPs were optimized for each RNA sample in order to guarantee best RNA yield. The RNA was then purified using 6% denaturing PAGE gels and recovered via Elutrap Electroelution System (Whatman).
The 55-residue NC protein (from HIV-1 strain NL4–3) was expressed from the bacteria vector pRD2, which was transformed into BL21 (DE3) pLysE. The protein was purified under non-denaturing conditions as described 119.
RNA was prepared in 10 mM Tris-HCl first, heat denatured for 3 minutes, and cooled on ice. 10× ITC salt was then added to RNA sample to reach a final salt concentration of 140 mM KCl, 10 mM NaCl and 1 mM MgCl2. After incubating at 37 °C for 1, 2, 4, 24 and 48 hours, the RNA samples were loaded to 2% agarose gel prestained with ethidium bromide, and run at 110 volts in 1× TB buffer (44.5 mM Tris-boric, pH 7.5) on ice to prevent thermal denaturation during electrophoresis.
ITC experiments were carried out using a VP-Isothermal Titration Calorimeter MicorCalorimeter (VP-ITC) (MicroCal Corp. Northampton, MA; 1989). NC (180 µM) in ITC buffer (10 mM Tris-HCl, pH 7.0, 140 mM NaCl, 10 mM NaCl, 1 mM MgCl2, and 100 µM β-mercaptoethanol) was loaded into the injection syringe. The calorimetry cell was loaded with RNA (0.8 µM). The RNA sample (in 10 mM Tris-HCl) was heat denatured, then cooled on ice. 10× ITC salt was then added to reach a final salt concentration of 140 mM KCl, 10 mM NaCl and 1 mM MgCl2. The RNA sample was then incubated overnight at 37 °C before loading into the calorimetry cell. After thermal equilibration at 30 °C and the initial 60 seconds delay, 28 serial injections of NC protein (10 µl each) were made into calorimetry cell. The thermodynamic parameters of the NC-RNA interaction were determined by a nonlinear-squares fit of the data using two-sites binding mode.
This work was supported by a grant from the National Institute of General Medical Sciences (NIGMS; GM 42561). K.E. was supported by a grant from the NIGMS for enhancing minority access to research careers (MARC U*STAR 2T34 GM008663). We thank HHMI staff at UMBC for technical assistance.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.