Nucleocapsid-RNA Interactions Are Essential to Structural Stability but Not to Assembly of Retroviruses

Abstract

The process of RNA incorporation into nascent virions is thought to be critical for efficient retroviral particle assembly and production. Here we show that human immunodeficiency virus type 1 mutant particles (which are highly unstable and break down soon after release from the cell) lacking nucleocapsid (NC) core protein-mediated RNA incorporation are produced efficiently and can be recovered at the normal density when viral protease function is abolished. These results demonstrate that RNA binding by Gag is not necessary for retroviral particle assembly. Rather, the RNA interaction with NC is critical for retroviral particle structural stability subsequent to release from the membrane and protease-mediated Gag cleavage. Thus, the NC-RNA interaction, and not simply the presence of RNA, provides the virus with a structural function that is critical for stable retroviral particle architecture.

The Gag proteins and viral RNA are the main structural components of retroviral particles (reviewed in reference 49). The Gag proteins have roles in viral assembly and release and are synthesized initially as precursor polyproteins that assemble at the lipid-rich areas of the cell membrane (38, 50). Enveloped RNA viruses bud by taking advantage of the cellular machinery normally used to create vesicles that bud into late endosomal compartments (41, 42). During or after budding, the virion undergoes maturation and the Gag polyprotein is cleaved by the viral protease (PR) into the subenvelope matrix shell (MA), the central capsid shell (CA), the nucleocapsid (NC) core protein, p6, and two small spacer peptides (SP1 and SP2) (24, 49). Formation and release of immature virus particles can be observed in the absence of an active PR; therefore, this proteolytic processing is not required for the assembly process (27, 49). However, maturation of the virion marked by the condensation of its viral core, which is composed of at least CA, NC, and the viral RNA (1, 10, 22), is strictly PR dependent. This event is believed to occur relatively soon after the immature virion has budded from the cell (49).

Substantial in vivo and in vitro data support the view that RNA molecules have a role in virus morphogenesis (10, 11, 22, 36). RNase treatment of viral core preparations disrupts the cores almost entirely (36). In vitro assembly of human immunodeficiency virus type 1 (HIV-1) CA and NC into core structures with the correct conical shape is achieved under physiological conditions only in the presence of RNA molecules, which can be viral or nonviral (22). Cores can also form in the absence of RNA but only under conditions of very high ionic strength (22). These observations are consistent with a role for RNA molecules in virus morphogenesis. It is not clear, however, whether RNA is necessary for the assembly of the viral particle or for its continued stability and mature morphology.

The NC protein is involved in RNA binding, virus assembly, and chaperone activity during reverse transcription (reviewed in references 15 and 44). NC contains two zinc-binding motifs and a high content of basic residues that contribute to RNA binding (15, 43). Deletions in the NC domain or of the entire NC result in severe viral assembly defects (6, 9, 16, 19, 21, 23, 25, 35, 46, 51). In studies of NC deletion mutants, it has been difficult to separate the function of the NC domain of Gag in Gag protein-protein interactions from its functions in NC-RNA interactions during the assembly process, as both are affected by NC deletion. Some investigators detected reduced levels of viral particles for viruses with mutations in the basic residues of NC and attributed this phenotype to deficient assembly (12, 13, 18). As the NC of these mutants has reduced RNA-binding properties, RNA binding by Gag was thought to be necessary for assembly. However, the idea that RNA binding is not critical to assembly is supported by the observation that efficient particle formation and release can still occur when NC is replaced with a non-RNA binding motif, a leucine zipper domain (1, 53). This observation suggests that the protein-protein interactions but not the protein-RNA interactions mediated by NC are critical for virus assembly.

In the absence of RNA molecules containing the viral packaging sequence (Ψ RNA), cellular RNAs become incorporated into virus-like particles, possibly because of the high affinity of the NC protein for nonspecific RNA sequences. Transcripts for tRNAs and ribosomal and housekeeping genes can be detected among these RNAs (2, 4, 7, 8, 21, 36). Mutations in NC basic residues adversely affect the RNA-binding properties of the NC domain and, consequently, viral genomic and spliced RNA incorporation (14, 26, 29, 40, 47). However, transcripts for housekeeping genes can be detected even when the affinity of NC for RNA is substantially reduced (52). Therefore, it is possible that RNA incorporation of cellular RNAs into retroviral particles can occur simultaneously through NC-mediated and NC-independent mechanisms.

The high content of basic residues of all retroviral NC proteins is critical for NC-RNA interactions (14, 17, 31, 47). Of the 55 amino acid residues in HIV-1 NC p7, 17 are basic. In the mutant M1-2/BR, 10 of the 17 positively charged residues are replaced with alanines and viral RNA incorporation into the particle is severely reduced (40). Specific and nonspecific RNA binding of this mutant NC protein is significantly reduced, and the corresponding M1-2/BR virus has also been reported to have a particle assembly defect (13). However, cell membrane assembly of the Gag polyprotein in M1-2/BR was not altered but the majority of viral proteins found outside the cells were no longer particle associated (52). Two possible explanations exist for this observation: abnormal post-cell membrane assembly and budding or normal assembly and budding followed by the instability of the released viral particles (52). Wang and Aldovini favored this second interpretation and speculated that processing of Gag was the event leading to the disruption of these NC mutant particles (52).

In this report we show that particles in which NC-mediated viral RNA incorporation is virtually abolished and Gag processing is prevented are efficiently produced, are stable in the supernatant, and sediment at the correct density for retroviruses. In contrast, only a fraction of particles with these NC mutations and an active PR sediment at the appropriate density. Our data indicate that NC-mediated RNA binding is not required for particle assembly and release but plays a critical role in particle structure by preventing the disassembly of the particle when the Gag precursor is processed into MA, CA, and NC.

MATERIALS AND METHODS

Plasmid constructs.

The constructs used in the experiments described in this report are schematically represented in Fig. 1. The parental viral DNA clone used in these studies is the biologically active plasmid pHXB2gpt, which produces a laboratory-adapted strain of HIV-1 (20). In pM1-2/BR, which was previously described, 10 positively charged residues present in the NC fragment that contains the two zinc-binding motifs of NC were changed to alanines (40). These mutations are also present in pM1-2/BR/PR⁻. PR function was abolished in pHXB2/PR⁻ and in pM1-2/BR/PR⁻ by mutating the aspartic acid in position 25 to a serine according to standard mutagenesis procedures (5) This mutation was previously shown to inactivate the viral PR (45).

FIG. 1.

Particle production by NC mutant constructs. (a) Schematic representation of the constructs. pHXB2gpt and the pM1-2/BR mutant are a wild-type HIV provirus and an HIV provirus with 10 mutations in NC, respectively (40). Constructs pHXB2/PR⁻ and pM1-2/BR/PR⁻ are identical to pHXB2gpt and pM1-2/BR, respectively, with the exception that they also include a D25S mutation in the PR gene. LTR, long terminal repeat; ENV, Envelop; mNC, mutated NC (mutations according to reference 40); mPR, mutated PR (D25S mutation). (b) Analysis of intracellular viral protein accumulation and of pelleted virions. Equal amounts of pHXB2, pM1-2/BR, pHXB2/PR⁻, and pM1-2/BR/PR⁻ cell protein lysates were analyzed by Western blot with a HIV⁺ human antiserum. Virions from transfection supernatants corresponding to equal amounts of p24 were pelleted through a sucrose cushion, and equal volumes were analyzed by Western blot. The p24 contents of the pHXB2, pM1-2/BR, pHXB2/PR⁻, and pM1-2/BR/PR⁻ pelleted virion lysates analyzed by Western blot were 23, 3, 29, and 21 ng, respectively. Mock, sample derived from cells transfected with transfection medium without plasmid DNA.

Cell lines, transfections, and assays.

Transfection in 293T cells was carried out by the calcium phosphate method, using 20 μg of DNA in a 100-mm-diameter petri dish. Cells were maintained in Dulbecco's minimal essential medium high-glucose medium supplemented with 10% fetal bovine serum and in some experiments were switched to serum- and protein-free medium (Cellgro-Free; Cellgro) after transfection. Electron microscopy (EM) was carried out on sections of embedded cells and particles according to standard procedures. For protein analysis, 293T cells transfected with each mutant were lysed in Laemmli buffer (5% glycerol, 1% sodium dodecyl sulfate [SDS], 31.875 mM Tris [pH 6.8], 0.005% bromophenol blue) (34) and subjected to SDS-polyacrylamide gel electrophoresis and Western blot analysis using an HIV-1-positive serum as described previously (40). A quantitative RT-PCR assay was carried out on cellular RNA extracted from 293T cells and DNase treated to eliminate contaminating DNA according to a previously described procedure (40). Supernatants cleared by low-speed centrifugation and filtered though a 0.45-μm-pore-size Millipore filter were utilized for sucrose gradient or sucrose cushion centrifugation according to previously described protocols (3, 13, 16). Viral pellets were resuspended in 0.1% Triton X-100 for p24 enzyme-linked immunosorbent assays (ELISA) (p24 core profile kit; DuPont). p24 ELISA and reverse transcriptase (RT) assays were carried out according to previously published procedures (3). Measurement of the amount of the 55-kDa Gag precursor, which tends to self-aggregate and therefore does not bind to the p24 monoclonal, followed the procedure described by Rose et al. (45), with minor modifications. Samples were sequentially diluted and disrupted in HIV-1 PR buffer (50 mM Na acetate [pH 4.9], 1 M NaCl, 1 mM dithiothreitol, 1 mM EDTA) containing 0.2% Triton X-100. After a freeze-thaw cycle, samples were put on ice for 30 min followed by digestion with 100 ng of recombinant HIV-1 PR (Bachem Bioscience, King of Prussia, Pa.) at 37°C for at least 4 h. The digested samples were checked for complete digestion by Western blot analysis and quantified using a p24 ELISA kit (Dupont NEN, Wilmington, Del.).

RNase treatment.

HXB2/PR and M1-2/BR/PR virions were harvested through a 25% sucrose cushion. The pelleted virions were resuspended in TN buffer (36) containing 0.25% NP-40, and half were left untreated while half were treated with 50 μg of RNase A for 90 min at 37°C. RNase-treated and untreated samples were fractionated in the supernatant and pellet by microcentrifugation. The supernatant and pellet fractions from each sample, corresponding to equal amounts of p24 in the original nonfractionated sample, were analyzed by SDS-polyacrylamide gel electrophoresis and Western blotting. A fraction of the pellet samples corresponding to equal amounts of p24 was used for RNA isolation and RT-PCR with either Gag- or β-actin-specific primers.

Isopycnic sucrose gradient analysis.

Clarified serum-free supernatants (500 μl) were directly layered on the 10 to 60% discontinuous sucrose gradients and were centrifuged in an SW41 rotor at 80,000 × g for 24 h at 4°C (13, 16). A total of 18 fractions of 500 μl each were collected. Gag content in each fraction was assayed by Western blot analysis, RT assays, and p24 ELISA after HIV-1 PR digestion (45).

RT-PCR of particle-associated RNA.

After clarification and evaluation of p24 in the medium of all the transfectants, supernatants corresponding to an equal amount of p24 were centrifuged to pellet the virions. Viral RNA was extracted, and quantitative RT-PCR was performed according to a previously described procedure (39, 40). RNA samples were obtained from three independent transfections of each construct. RNA samples corresponding to 8 ng of p24 were reverse transcribed and subjected to an 18-cycle PCR using SuperScript (Invitrogen, Carlsbad, Calif.). The primers used in the RT-PCR were gag primers specific for HIV MA (5′-GTGAGTACGCCAAAAATTTTG-3′ and 5′-TAATTTTGGCTGACCTGATTG-3′) and nef-specific primers (5′-TAGCTTGCTCAATGCCACAGC-3′ and CCACAGATCAAGGATATCTTG-3′). Human β-actin-specific primers (5′-ATGTTTGAGACCTTCAACAC-3′ and 5′-CACGTCACACTTCATGATGG-3′) were used in an RT-PCR containing a similar amount of RNA and carried out for 30 PCR cycles. The negative controls included a sample from an RT-PCR lacking input RNA and an RT-PCR with RNA extracted from a mock-transfected supernatant. A PCR on an equivalent amount of RNA that did not undergo RT was carried out for each sample to exclude incomplete DNase I treatment.

RNA quantification in viral particles.

Quantification of DNA-free viral RNA samples extracted from viral particles was carried out according to the procedure described by Muriaux et al., with some modifications (36, 52). Briefly, viral particles present in clarified transfection supernatants were concentrated using a Centriprep YM-30 concentrator (Amicon) and purified with a sucrose-step gradient (32). Amounts of viral particles equivalent to 4 μg of p24 were used for the RNA purification, and linear acrylamide (Ambion) at final concentration of 0.02% was added to the samples to facilitate the efficient recovery of the RNA. RNA samples were treated with 1 U of RNase-free DNase RQ1 (Promega)/μl for 1 h at 37°C. The absence of DNA contamination was confirmed by PCR using HIV-1 MA-specific primers and [³²P]dCTP in a 30-cycle reaction. A Ribogreen quantification kit (Molecular Probe) was used to quantitate the RNA according to the procedure suggested by the manufacturer, with minor modifications. Sequentially diluted RNA samples were mixed with an equal volume of Ribogreen reagent in a final volume of 200 μl in a 96-well plate (Strip Plat-8; Costar). Quantification was carried out using low-range rRNA standards that provide a linear readout at the RNA concentrations of the experimental samples. The emission of incorporated Ribogreen was measured at 535 nm, with excitation set at 485 nm on a HTS7000plus Reader (Perkin Elmer). The assay was validated using aliquots of diluted RNA amounts in a range between 10 and 100 ng. The same samples were also used for RNA end labeling. The labeling followed the procedures described by Muriaux et. al. (36) except that 2 μl of cytidine 3′,5′-bis-phosphate (pCp) (NEN) (3,000 Ci/mmol; 1 Ci = 37 GBq) was used in the labeling reaction. Yeast tRNA (brewer’s yeast) dilutions containing 100- to 1-ng amounts of RNA were end labeled as standards.

RESULTS

Virus particle production by NC mutants.

Mutations in the positively charged residues of HIV NC were shown to reduce NC affinity for RNA but did not produce a cell membrane assembly defect (40, 47, 52). However, only ∼10% of the Gag molecules present in the supernatant of this mutant (M1-2/BR) sedimented at the correct density for a retroviral particle and significant amounts of all viral products could be found in the supernatant (52). These studies suggested that the M1-2/BR mutant particles do not have an assembly defect, as was previously suggested (12, 13), but may disassemble soon after release. These disrupted particles may account for the low-density viral proteins detected in the sucrose gradients of the mutant. It is possible that the absence of bound RNA in the Gag complexes present in the “barges” leads to aberrant release of these complexes in the supernatant. We favored the interpretation that RNA binding by NC might be critical to particle stability after budding but not to virus assembly and hypothesized that PR-mediated Gag processing might be the event that triggers particle disassembly in the absence of NC-RNA interaction (52).

To test this idea, we have combined the mutation present in pM1-2/BR with a mutation in the active site of the PR and constructed mutant pM1-2/BR/PR⁻ (Fig. 1a). pHXB2/PR⁻, a control construct with a wild-type gag sequence and a similarly mutated PR gene, was also constructed. The aspartic acid at position 25 was changed to a serine, as this mutation has been shown to eliminate PR function (45). We confirmed that the PR mutations prevented Gag processing and that the Gag proteins that accumulated within the cell are the Gag or the Gag-Pol precursors (Fig. 1b, 3rd and 4th lanes from left). Cells transfected with pHXB2, pM1-2/BR pHXB2/PR⁻, and pM1-2/BR/PR⁻ showed similar transfection efficiencies and accumulated similar levels of intracellular viral protein (Fig. 1b). They also released comparable levels of Gag, suggesting similar levels of particle release (data not shown). When the amount of Gag was evaluated in the pellet after supernatant centrifugation, only about 10% of the Gag protein detectable in the pM1-2/BR supernatant could be recovered in the sedimented particles (Fig. 1b, 7th lane from left) (52); in contrast, the amounts recovered for pHXB2, pHXB2/PR⁻, and pM1-2/BR/PR⁻ were approximately 80% (Fig. 1b, 6th, 8th, and 9th lanes from left). These results suggested that a large fraction of Gag products released from the M1-2/BR mutant was not particle associated. The addition of the PR mutation in pM1-2/BR/PR⁻ increased significantly the percentage of viral proteins that could be recovered from the supernatant in the form of sedimented virions, excluding an effect of these NC mutations on virus assembly. When cells transfected with pM1-2/BR were treated with Ritonavir, an inhibitor of HIV PR activity, a significantly higher percentage of particles could be recovered from the supernatant than when M1-2/BR particles were produced in the absence of Ritonavir, further supporting the role of Gag processing in the instability of M1-2/BR (data not shown).

The budding of wild-type and mutant particles was examined by EM using transfected 293T cells; the results for pHXB2/PR⁻ and pM1-2/BR/PR⁻ are shown in Fig. 2. As in the case of M1-2/BR (52), particles were observed budding from cells in the extracellular space. The abundance of particles seen budding and surrounding individual cells strongly argues against any defect in assembly for pM1-2/BR/PR⁻ (Fig. 2a). The majority of the mutant particles showed a typical immature morphology and a larger size than that typical of most mature particles. As expected due to the PR mutation, no mature capsid cores were observed in any of the mutant particles. Virions with aberrant morphology could be observed in both pM1-2/BR/PR⁻ (Fig. 2a and b) and pHXB2/PR⁻ (Fig. 2c, d, and e), suggesting that aberrant morphology is more likely the result of the mutation in PR than in NC.

FIG. 2.

EM of cell-associated viral particles. Panels show particles from pM1-2/BR/PR⁻ (a and b)- and pHXB2/PR⁻ (c, d, and e)-transfected cells. In panel a the scale is 1:9,000, in panels b, c, d, and e the scale is 1:40,000.

Stability and recovery of NC mutant viral particles.

To more thoroughly characterize and compare the particle phenotypes of the pM1-2/BR and pM1-2/BR/PR⁻ mutants, we subjected transfection supernatants to isopycnic centrifugation (Fig. 3). Two major peaks of Gag-associated p24 were observed in the gradient fractions. The first peak (p1), which was comprised of low-density fractions, contained monomeric and oligomeric Gag polyproteins. The second peak (p2) contained typical retroviral particles, as the density of the fractions in this peak fell into the normal range for the density of retroviral particles (1.15 to 1.17 g/ml). For the wild-type virus HXB2, the amount of Gag-associated p24 present in peak fractions p1 and p2 was approximately 20 and 80%, respectively, of the total Gag-associated p24 present in the supernatant (Fig. 3a). For M1-2/BR, the amounts of Gag-associated p24 present in p1 and p2 were approximately 90 and 10%, respectively (Fig. 3b) (52). Interestingly, the Western blot profile of M1-2/BR showed enrichment for Gag p39 in p2 fractions (Fig. 3b). Gag distribution in the gradient profiles of pHXB2/PR⁻ and pM1-2/BR/PR⁻ was very similar to that seen for HXB2, and the majority of Gag proteins was particle associated and sedimented at approximately 1.17 g/ml (Fig. 3a, c, and d). The Western blot profiles of all gradient fractions of pHXB2/PR⁻ and pM1-2/BR/PR⁻ were virtually indistinguishable (Fig. 3c and d). These results indicate that in absence of PR cleavage the amount of pelletable particles released by pM1-2/BR/PR⁻ in the supernatant is similar to that seen for pHXB2/PR⁻ and that the NC mutations present in M1-2/BR/PR⁻ have no effect on particle assembly and release.

FIG. 3.

Sucrose gradient analyses of HXB2, M1-2/BR, pHXB2/PR⁻, and pM1-2/BR/PR⁻ viral particles. (a) Sucrose gradient fractionation of extracellular Gag and Western blot analysis of sucrose fractions. (b) Gag protein amounts in each fraction were measured by p24 ELISA after HIV-1 PR digestion and were expressed as percentages of the total Gag amounts harvested from the gradient. Density of fractions 11 and 12 (a and b) and 12 to 14 (c and d) was 1.168 to 1.171 g/ml.

RNA incorporation in NC mutants.

We next measured the amounts of incorporated viral Ψ RNA and other mRNAs in sedimented particles (Fig. 4). Efficient Ψ RNA incorporation was detected in viruses produced by pHXB2 and pHXB2/PR⁻ (Fig. 4a). In contrast, viral RNA incorporation was substantially reduced in pM1-2/BR/PR⁻ (Fig. 4a) and to the same extent as that seen in pM1-2/BR (52). This was also true for the incorporation of spliced viral RNAs (Table 1), which are usually incorporated at a lower rate than genomic viral RNA and reflect nonspecific NC RNA binding. The detection of cellular β-actin mRNA in sedimented particles is a qualitative indicator of host cell mRNA incorporation into viral particles; it is likely that a variety of cellular mRNAs are incorporated into retroviral particles (2, 4, 28, 30, 36, 52). β-Actin mRNA could be detected in sedimented particles when the NC protein had been mutated such that both specific and nonspecific RNA binding properties were affected, reflecting most likely its intracellular abundance relative to viral RNA (Fig. 4b). The amount of β actin RNA found in HXB2/PR⁻ was comparable to that seen with HXB2. However, the amount of β-actin RNA found in pM1-2/BR/PR⁻ was consistently two- to threefold less than that found in HXB2 (Fig. 4b). The lower intensity of the β-actin band detected in M1-2/BR/PR⁻ particles compared to HXB2 could reflect the lower affinity of M1-2/BR/PR⁻ NC for this RNA. The amount of β-actin RNA found in M1-2/BR/PR⁻ was lower than the amount found in pelleted M1-2/BR virions (52). It is possible that detection of β-actin RNA in M1-2/BR virions was affected by a higher amount of microvesicles present in the sample, considering that the volume of supernatant necessary to obtain an equal amount of pelleted particles is 10-fold higher than for wild-type virus.

FIG. 4.

RNA incorporation in wild-type and mutant viral particles. (a) RT-PCR of particle-derived RNA. The results of one representative experiment are shown. The percentage of the Ψ RNA content is reported for each construct (HIV-1 MA-specific primers). Panel lanes labeled with + correspond to an RNA sample that was subjected to RT-PCR (results obtained with 1× and 0.5× RNA amounts are shown), while − indicates an RNA sample (1×) that was subjected to PCR only. Mock, sample derived from cells transfected with transfection medium without plasmid DNA. (b) Detection of β-actin RNA in particle RNA carried out by RT-PCR. The numbers in this panel indicate the amounts of β-actin reported as severalfold increases over the amounts detected for HXB2. (c) Isotopic end labeling of RNA extracted from wild-type and mutant particles. Results are shown for reactions carried out with 10, 5, and 2.5 ng of yeast tRNA and for two reactions carried out for each RNA from virions (1× and 0.5× amounts, corresponding to 400 and 200 ng of p24). The gel was exposed for 3 days. RNA samples used for this reaction were the same samples used as described for panel A. (d) A longer exposure (12 days) of selected lanes (1× RNA amount) from the gel shown in panel c. (e) A shorter exposure (3 h) of the bottom part of the gel shown in panel c.

TABLE 1.

RNA content in pelleted particles^a

Constructs	Results (% of wt)		Results (severalfold increase)
	Gag RNAb	Total viralc	β-actin	Total RNAd
pHXB2	100	100	1	1
pM1-2/BR	0.8 ± 0.4e	2 ± 1e	4 ± 2e	0.9 ± 0.09e
pHXB2/PR	107 ± 7	117 ± 11	1	1.1 ± 0.1
pM1-2/BR/PR−	0.3 ± 0.1	6 ± 4	0.4 ± 0.1	1.1 ± 0.1

^aMeans and standard deviations derived from three independent experiments are reported. All evaluations (except those determining total RNA values) were carried out by RT-PCR. Percentage and severalfold-increase values were determined by scanning the intensity of the RT-PCR bands.

^bGag RNA values indicate the amounts of viral genomic RNA.

^cTotal viral values indicate the amounts of virus-specific RNA in particles and were determined using nef primers.

^dValues representing total RNA in particles were estimated for aliquots of pelleted virions containing equal amounts of p24 (analyzed by ELISA after HIV-1 protease digestion). A Ribogreen quantification kit was used to quantitate the total RNA. Background from mock-transfected cells was consistently lower that 5% when analyzed using Ribogreen quantification.

^ePreviously reported by Wang et al. (52) and provided here for comparison with the values obtained for the other mutants.

Measurements of the total amount of RNA in aliquots of pelleted virions containing equal amounts of Gag indicated that RNA in the pelleted HXB2/PR⁻ and M1-2/BR/PR⁻ virions was present in amounts similar to that found in pelleted wild-typeHXB2 virions (Table 1). This result, combined with the result observed for the viral and β-actin RNAs, prompted us to investigate the nature of the RNA present in the total RNA extracted from these particles. Equal amounts of RNA from wild-type and mutant particles were end labeled, and their size distribution was analyzed by gel electrophoresis (Fig. 4c). We found that each of the vast majority of the RNA molecules present in the mutant particles was shorter than 300 nucleotides. In HXB2 the intensity of signal for RNA species larger than 300 nucleotides was less than 2% of the total signal, indicative of a corresponding lower molarity of larger RNAs in the particles. These species were equally represented in the wild type and mutants, and their sizes correspond to ribosomal and 7S RNA, which have been previously shown to be present in wild-type and mutant virions (7, 8, 37). The amount of cellular RNAs incorporated in sedimented M1-2/BR particles was also indistinguishable from that seen with HXB2 (data not shown). We could not detect the 9-kb viral genomic RNA by end labeling, even with an exposure longer than that described for Fig. 4c. However, this viral RNA was detectable by RT-PCR on the same HXB2 and HXB2/PR⁻ samples (Fig. 4a). It is possible that a higher-level input of particle RNA is necessary to visualize the 9-kb viral genomic RNA by end labeling. We conclude that cellular RNAs (and, in particular, small RNAs) represent the majority of the RNA incorporated in retroviral particles and that their levels do not correlate with NC RNA-binding properties, as they were equally represented in HXB2, HXB2/PR⁻, M1-2/BR, and M1-2/BR/PR⁻.

Viral particles from M1-2/BR and M1-2/BR/PR⁻ are produced and incorporate RNA at the same rate. However, a large fraction of the M1-2/BR particles disassemble soon after release, most likely when Gag is processed by PR. Only a small fraction of virions was found after sedimentation, and this fraction was enriched for particles with incomplete Gag processing (Fig. 3b). In contrast, most of the particles produced by M1-2/BR/PR⁻, in which Gag processing did not occur, were recovered after sedimentation. Taken together, these data indicate that the presence of RNA inside a particle is not sufficient per se to provide particle stability after Gag processing, as significant amounts of ribosomal and small RNAs were present within M1-2/BR. Instead, a high-affinity NC-RNA interaction seems critical for particle structure to be retained after Gag processing. This interaction is unlikely in M1-2/BR and in M1-2/BR/PR⁻ because of the NC mutations, which substantially reduce the protein's positive charge (13, 39). The consequences of the loss of high-affinity NC-RNA interactions can be prevented if Gag processing is prevented.

Role of RNA in PR⁻ particle stability.

We reasoned that if a mutation that prevents Gag processing confers particle stability in the absence of high-affinity NC-RNA binding and the RNA found in HXB2/PR⁻ and M1-2/BR/PR⁻ particles is not critical to particle stability, then degradation of particle RNA by RNase treatment should not affect particle stability. We treated virus particles recovered after centrifugation with RNase and measured the amount of particles recovered after treatment. The protein content and RNA content of the recovered particles were analyzed by Western blot analysis and RT-PCR (Fig. 5). The results showed that equal amounts of particles were recovered whether the particles were treated with RNase or not, indicating that the RNA present within these virions is not necessary for particle stability. To confirm that the RNase treatment degraded the RNA present in the particles, we measured the amounts of Gag and β-actin RNA present in the RNase-treated and pelleted particles. While both viral and cellular RNAs could be found in untreated virions, they could not be detected after RNase treatment, confirming the successful digestion of RNA. These results differ from those observed by Merioux et al. with Moloney murine leukemia virus (36), in which RNase treatment of a PR⁻ virus caused a small reduction in the amount of particles recovered after sedimentation. We do not have an explanation for this difference, although it is possible that PR⁻ Moloney murine leukemia virus and PR⁻ HIV differ slightly in their RNA requirements for particle stability. These data further confirm that the particle structure of PR⁻ HIV virions is not dependent on the presence of RNA.

FIG. 5.

RNase treatment of HXB2/PR⁻ and M1-2/BR/PR⁻ viral particles. (A) Western blot analysis of supernatant (S) and pelleted particles (P), with or without RNase treatment, of sedimented HXB2/PR⁻ and M1-2/BR/PR⁻ viral particles. (B) Gag and β-actin RT-PCR on particle RNA extracted from P1, P2, P3, and P4 fractions.

DISCUSSION

Our results demonstrate that the basic charge present in NC is crucial to NC-RNA interactions but not to the Gag-Gag interactions that occur at the cell membrane and are necessary for particle assembly. Furthermore, particle stability is not directly affected by the NC mutations that diminish RNA binding by Gag. If this were the case, M1-2/BR and M1-2/BR/PR⁻ would have very similar phenotypes. Instead, the disassembly of M1-2/BR particles occurs at the time of Gag processing. This may be due to the reduced affinity of the mutated NC for the RNA present in the particle. The protein-protein interactions that occur among M1-2/BR-mutated Gag polyproteins can support the formation and budding of a particle in the absence of Gag binding to NC. However, particle structure is no longer maintained when the Gag polyprotein is cleaved and the interactions between NC and the RNA that lead to core condensation do not occur. The high-affinity RNA-NC binding seems critical to the orderly interactions among RNA, NC, and CA subunits that occur in the core. In its absence, the intraprotein interactions that are known to occur at least for CA are not sufficient to retain particle structure.

The experiments reported here support the hypothesis that the M1-2/BR mutant does not have an assembly defect (52); instead, most of its particles are short lived after budding. These experiments also exclude the less likely interpretation that the soluble HIV proteins found in the supernatant of M1-2/BR result from aberrant secretion of the Gag and Gag-Pol complexes. In light of the data reported here, it is now obvious why the short-lived particles could not be observed in pulse-chase experiments (52), as the time line of these experiments is most likely longer than the life span of these particles.

When NC affinity for RNA is reduced as it is in M1-2/BR, particles are efficiently produced but are highly unstable after release. Finding similar amounts of pelleted virus particles with HXB2/PR⁻ and M1-2/BR/PR⁻ rules out the possibility that the Gag molecules that are not bound to RNA interact less favorably with other Gag molecules at the cell membrane and therefore exclude inefficient assembly of particles when NC-RNA interactions are altered. The analysis of the NC mutations, together with that of the PR mutations in M1-2/BR/PR⁻, indicates that processing of the Gag precursor by the viral PR is the event that leads to particle disruption in M1-2/BR, most likely immediately after budding and cell release, leading to the accumulation of soluble Gag outside the cell. By preventing processing in these mutant retroviral particles, their structures could be retained and analyzed. NC-mediated RNA binding has been proposed to be important for particle assembly (12, 13, 16). Our results argue against RNA being necessary for retroviral assembly but do not rule out the possibility that viral RNA might play some limited role in accelerating or facilitating the assembly process.

Mature and immature particles are routinely found in retroviral cultures. The fraction of M1-2/BR particles that sediments at the correct density has a larger proportion of unprocessed Gag than the total viral protein profile observed in the supernatant of M1-2/BR or the HXB2 wild-type construct. It is possible that these particles represent the fraction of immature particles present in the supernatant at the time of harvesting. This fraction of immature virions may be higher in this mutant than in the wild-type virus, as PR-mediated Gag cleavage might be facilitated by the Gag-RNA interaction (48). Maturation of the particle is linked to postbudding processing of the viral precursors Gag and Gag-Pol and to the condensation of CA and NC on the RNA. Processing of the Gag precursor occurs in an orderly fashion, with initial cleavage between MA and the CA-NC fragment followed by the cleavage of CA and NC (49). We do not know at this point whether the M1-2/BR particles are disrupted as soon as the first Gag subdomain is cleaved off the Gag precursor or when cleavage between CA and NC occurs and complete processing is achieved. Our observation that Gag p39 is enriched in the M1-2/BR particles that sediment at the appropriate density compared to the viral material found in lower-density fractions may indicate that some particle stability can be provided by incomplete processing of the CA-NC fragment (Fig. 3b). The absence of or reduced NC-RNA binding might affect Gag precursor folding and in turn reduce the rate or order of PR processing.

Measurements of total RNA incorporation in pelleted HXB2/PR⁻ and M1-2/BR/PR⁻ revealed that the levels of RNA are comparable. This observation rules out the hypothesis that unstable M-12/BR particles may not contain any RNA (52). If any RNA present in particles is adequate to provide particle stability, then the total RNA found in pelleted M1-2/BR/PR⁻ particles should correlate directly with the percentage of M1-2/BR Gag that can be sedimented. Instead, RNA was found in equal amounts in wild-type and mutant viruses, including M1-2/BR, supporting the possibility that most of the detected RNA can become part of a particle even when NC nonspecific binding is significantly reduced. These RNAs are not sufficient to provide structure stability in M1-2/BR. It is possible that only high-affinity NC-RNA core interactions provide the appropriate structural element necessary for particle stability. It is also possible that an NC with a reduced positive charge can bind with very low affinity to RNA. In this scenario, particles containing RNAs interacting poorly or not at all with NC could disassemble when Gag processing occurs but would be preserved in the absence of Gag processing, providing an explanation for the amount of RNA found in M1-2/BR/PR⁻. This also explains why M1-2/BR immature particles were recovered after sedimentation.

Published observations of HIV and other retroviruses indicate that when present, the Ψ RNA is the RNA that is preferentially bound by NC (references 4, 14, 36, and 43 and references therein). Cellular RNAs that do not have a packaging signal can be bound with relatively high affinity by wild-type NC when RNA species with a packaging signal are not present in the cytoplasm (52). Spliced viral RNAs, which can be found incorporated in the particle at a higher rate than normal in the absence of Ψ RNA and reflect the nonspecific binding activity of Gag, were not detected in M1-2/BR and M1-2/BR-PR⁻, providing direct evidence for reduced nonspecific RNA binding for this mutant. Cimarelli et al. (13) showed that M1-2/BR NC binds nonspecific RNA approximately 20-fold less efficiently than wild-type NC in a Northern-Western assay. In this particular case, nonspecific RNA binding was not evaluated in the presence of competitor Ψ RNA; it is very likely that nonspecific RNA binding would be reduced more than 20-fold if competitor Ψ RNA were present. Gag binding to Ψ RNA is at least 100-fold higher than binding to poly(A) when both RNA species are present simultaneously (33). Because extremely low amounts of Ψ RNA were incorporated in M1-2/BR and M1-2/BR/PR⁻ despite being abundant in the cytoplasm, it is very unlikely that the NCs of these viruses, which have defects in nonspecific RNA binding (13, 40, 47), can actively bind other RNAs at the time of assembly. Yet some abundant host cell mRNAs such as β-actin and rRNA and many small RNAs were found in M1-2/BR/PR⁻ mutant virions. In these NC mutant particles, the relative abundance of certain mRNAs in the cytoplasm seems to be the most significant determinant of its particle incorporation and it is likely that their incorporation is accidental and related to their location near the assembly site rather than being NC mediated. Because the amounts of these RNAs are similar in the wild type and the NC mutants, it is likely that their incorporation occurs in an NC-independent fashion, perhaps even in wild-type virions.

Taken together, the data presented here provide evidence in support of the role played by the viral RNA in particle structure stability but exclude a role for NC-mediated RNA binding in assembly. Our results also indicate that processing is the event that requires appropriate NC-RNA interactions to retain particle stability. Thus, the viral genomic RNA molecule functions both as the genome of the retrovirus and as an essential structural component of mature retroviral virions via its interaction with NC.