|Home | About | Journals | Submit | Contact Us | Français|
Assembly of immature retroviral particles is a complex process involving interactions of several specific domains of the Gag polyprotein localized mainly within capsid protein (CA), spacer peptide (SP), and nucleocapsid protein (NC). In the present work we focus on the contribution of NC to the oligomerization of CA leading to assembly of Mason-Pfizer monkey virus (M-PMV) and HIV-1. Analyzing in vitro assembly of substitution and deletion mutants of ΔProCANC, we identified a “spacer-like” sequence (NC15) at the M-PMV NC N terminus. This NC15 domain is indispensable for the assembly and cannot be replaced with oligomerization domains of GCN4 or CREB proteins. Although the M-PMV NC15 occupies a position analogous to that of the HIV-1 spacer peptide, it could not be replaced by the latter one. To induce the assembly, both M-PMV NC15 and HIV-1 SP1 must be followed by a short peptide that is rich in basic residues. This region either can be specific, i.e., derived from the downstream NC sequence, or can be a nonspecific positively charged peptide. However, it cannot be replaced by heterologous interaction domains either from GCN4 or from CREB. In summary, we report here a novel M-PMV spacer-like domain that is functionally similar to other retroviral spacer peptides and contributes to the assembly of immature-virus-like particles.
Assembly of immature retroviral particles is mediated by mutual interactions of Gag polyproteins. The Gag polyprotein of all retroviruses invariably contains a matrix (MA), a capsid (CA), and a nucleocapsid (NC) protein region. Additional, genus-specific domains may be present. Following assembly, the immature, spherical particle is surrounded by lipid bilayer and released from the host cell by budding through the plasma membrane. During the budding process the virus-encoded protease is activated and cleaves Gag polyprotein, releasing MA, CA, and NC proteins. MA remains associated with the membrane while CA condenses to form a core shell enclosing NC-RNA complex.
The immature particle is stabilized by interactions between various regions throughout the Gag polyprotein. The major protein-protein interaction regions are within the CA that consists of two independently folded domains, an N-terminal domain (NTD) and a C-terminal domain (CTD) linked by a short, flexible linker. In HIV it was shown that whereas the CA NTD is not absolutely required for the assembly, the CA CTD together with a downstream spacer peptide sequence (SP) and NC are essential for the immature particle assembly (2, 4, 29, 37, 58). Preventing the cleavage between CA and SP or its inhibition, e.g., by betulinic acid (PA-457), results in noninfectious particles lacking a properly formed core (25, 30, 43, 59, 63).
The structure reconstruction, obtained by electron cryotomography of immature particles, revealed patches of single continuous hexagonal Gag arrays composed of ordered CA and SP1 domains that are closed by the incorporation of irregular defects (7, 60). The detailed structure of CA lattice in the mature core was studied by cryo-electron microscopy, tomography, and electron crystallography (6, 8, 15, 31). The mature lattices are formed by hexamers of the CA NTD linked by CA CTD dimers.
Though the CA NTD is not essential for immature particle assembly, it is believed that the N-terminal β-hairpin, which is formed only upon proteolytic maturation, can shift the assembly from the immature, spherical particles to the mature, tubular cores (17, 46, 57, 58). All structures of CA NTD known to date show that the N-terminal β-hairpin is stabilized by formation of a salt bridge between two highly conserved residues, the N-terminal proline and aspartate in helix 3 (13, 16, 27, 36, 39).
Unlike the mature-particle-like lattices which can be effectively held together by CA-CA interactions alone, the assembly of immature particles relies on NC to tether Gag molecules together. The NC-NC interactions are most likely augmented by association with nucleic acid. With the exception of spumaviruses, retroviral NCs contain one or two CCHC zinc finger motifs responsible for the specific recognition and incorporation of retroviral genomic RNA into the immature particle. However, they are dispensable for immature virus-like particle (VLP) assembly (for a review, see reference 3). An even more conserved feature of retroviral NC sequences is a discrete patch of numerous basic amino acids, also described as an interaction (I) domain. Mutagenesis of these basic residues revealed their importance for initial Gag-Gag multimerization (5, 12, 50-52, 61, 62). The first seven amino acids of the N-terminal subdomain within HIV-1 NC (MQRGNFR) were shown to be sufficient for VLP reconstitution (50). These results led to a hypothesis that nonspecific RNA binding by the basic residues is required for augmenting Gag-Gag interactions and for providing the driving force during virion assembly (5, 12, 50, 52).
A broad spectrum of different nucleic acids from single-stranded DNA to RNA oligonucleotide was successfully used in in vitro assays, and the results led to an assembly model in which the nucleic acid acts as a scaffold and brings the assembling proteins together in early stages (10, 11, 34, 40, 55). The minimal size of DNA oligonucleotide required for Rous sarcoma virus (RSV) Gag assembly was determined as 16 nucleotides (nt). This sequence is long enough to accommodate just two Gag molecules, suggesting a role of nucleic acid in promoting Gag dimerization (34, 35, 55).
Additional support for this dimerization model was provided by replacing the NC with dimerizing leucine (DLZ) or trimerizing isoleucine (TIZ) zipper motifs (2, 14, 24, 62). Zhang et al. successfully replaced the HIV-1 NC in the Gag polyprotein with DLZ domain of human cyclic AMP response element binding protein (CREB) (62). This chimeric protein efficiently formed VLPs in mammalian cells that were released into the media. Similar results were obtained for the RSV Gag in which the replacement of NC with CREB DLZ restored budding of spherical particles morphologically similar to the wild-type VLPs (24). Accola et al. showed that leucine zipper dimerizing and also trimerizing domains from the yeast transcription factor GCN4 could substitute for NC-SP2 domain in assembly of minimal HIV-1 Gag construct (2). Most recently, Crist et al. replaced NC in HIV-1 Gag polyprotein with GCN4 DLZ and TIZ and studied their effect on assembly in 293T cells (14). The same authors also performed the first analysis of in vitro assembly of recombinant Gag-zipper chimeras (14). Their results demonstrated that while Gag zipper chimeras formed VLPs in mammalian cells, the NC replacement with DLZ and TIZ was not sufficient to support assembly in vitro unless cofactors such as nucleic acids or inositol phosphates were added (14).
Here we probe the role of NC protein in Mason-Pfizer monkey virus (M-PMV) assembly and oligomerization of CA using chimeric protein constructs. We employ a rapid screen for assembly of expressed proteins in Escherichia coli and a previously established in vitro assembly assay. For selected constructs we complement these results by following assembly and budding of viral particles in infected cells. Our results identify a specific “spacer-like” sequence at the NC N terminus consisting of 15 amino acids that together with a nonspecific stretch rich in basic residues (e.g., residues NC16-23 or the first 10 residues from CREB) is essential for M-PMV immature particle assembly. The basic stretch cannot be substituted for a generic dimerization or trimerization domain (e.g., GCN4), indicating that it mediated interactions with RNA rather than protein-protein association. On the other hand the “spacer-like” sequence (NC15) is specific for the virus and cannot be replaced by the HIV-1 spacer sequence. These findings are then compared with new results obtained for the assembly of HIV-1 ΔProCASP1NC.
Plasmids were created by using standard subcloning techniques and were propagated in E. coli DH5α. For in vitro studies, all bacterial vectors were based on the parental M-PMV ΔProCANCpET22b and HIV-1 ΔProCANCpET22b, described earlier (46, 55). Chimeric constructs were created by ligation of amplified PCR fragments encoding a leucine zipper sequence or HIV-1 SP1 with expression vectors. As a template DNA for amplification of dimerization and trimerization leucine zipper domains, vectors encoding appropriate DLZ or TIZ sequence of yeast transcription factor GCN4 (21), kindly provided by Z. Knejzlík (ICT), were used, respectively. Dimerization domain of the human transcription factor CREB was obtained by reverse transcription-PCR (RT-PCR) of total mRNA isolated from 293T cells, using primers 5′ GCGGCCGCCCGAGAGTGTCGTAGAAAGAAG and 3′ AAGCTTTTAATCTGATTTGTGGCAGTAAAGGTC. As a template DNA for amplification of HIV-1 SP1 region, the vector pSAX2, kindly provided by J. Luban, was used. Oligonucleotide-directed mutagenesis using Pfu polymerase and appropriate primers were used for introduction of the stop codon (UAA) or insertion of the NotI restriction site (GCGGCCGC). For preparation of proviral vectors for in vivo studies, helper vectors, carrying M-PMV SacI-Eco72I fragments (nt 1165 to 3275) and HIV-1 SphI-SbfI fragments (nt 2433 to 3828) inserted into pUC19, were prepared. Each chimeric construct was first inserted into a helper vector, and following sequence verification, the SacI-Eco72I fragment was inserted into M-PMV proviral construct pSARM4 and the SphI-SbfI fragment was inserted into HIV-1 proviral construct pSAX2. Further details of the cloning strategy and full sequences of all PCR primers can be obtained from us upon request.
Recombinant proteins were expressed in Escherichia coli BL21, and the purification of M-PMV and HIV CANC related proteins was carried out as described in references 36 and 55 with some modifications. The heparin Sepharose chromatography column (GE HealthCare) was used as the final purification step for ΔProCANC, ΔProCADLZ, and ΔProCATIZ while the proteins ΔProCANC23DLZ, ΔProCANC23TIZ, ΔProCANC23, ΔProCANC15DLZ, ΔProCANC15TIZ, ΔProCASP1DLZ, ΔProCASP1TIZ, ΔProCASP1, ΔProCANC15CREB, ΔProCANC15CREB10, and ΔProCANC15Δ10CREB were purified using Sephadex G-100.
A combined dynamic light scattering (DLS) and 90-degree laser light scattering (LS) flow cell (PDI2020/DLS; Precision Detectors) was coupled to a high-pressure liquid chromatograph (HPLC; Waters 600 HPLC) and a size-exclusion column (Superdex 75, Superdex 200, or TSK 6000) and used for the determination of molecular weights of the produced proteins or assembled products. The column was equilibrated with the assembly buffer (50 mM Tris-HCl, 100 mM NaCl, 1 μM ZnCl2, pH 8; flow, 0.5 ml/min). Purified proteins were centrifuged in a Beckman Airfuge (rotor A-95; 120,000 × g, 15 min), and aliquots of the purified proteins were dialyzed against the assembly buffer with or without RNA. The mixtures were then centrifuged at 120,000 × g for 15 min, and supernatant was loaded on an HPLC column. LS data were evaluated using Precision Analyze software (Precision Detectors) and absorption at 280 nm for concentration measurements. The LS detector was calibrated by standards with known molecular weights (bacteriophage P2 polymerase, bovine serum albumin [BSA], and lysozyme).
An aliquot of 60 μg in the total volume of 100 μl of purified protein was dialyzed against the assembly buffer (50 mM Tris-HCl, 100 mM NaCl, 1 μM ZnCl2, pH 8) overnight at 4°C using 1-kDa Spectrapor dialysis tubing. Nucleic acid (6 μg, DNA oligonucleotides, 67 or 73 nt long [Generi Biotech], or bacteriophage MS2 RNA [Roche Molecular Biochemicals]) was optionally added to the protein prior to dialysis. Assembly products were negatively stained with 2% sodium phosphotungstate (pH 7.3) on carbon-Parlodion-coated grids and studied by electron microscopy (EM) (see details below).
293T cells were grown in Dulbecco's modified Eagle's medium (DMEM; Sigma) supplemented with 10% fetal bovine serum (Gibco) and 1% l-glutamine (PAA Laboratories, Linz, Austria). 293T cells were transfected with the wild-type or mutant proviral DNA using Fugene HD transfection reagent (Roche Molecular Biochemicals). At 24 or 48 h posttransfection, culture media were filtered through a 0.45-μm filter and virions were centrifuged through a 20% sucrose cushion at 210,000 × g for 1 h in a Beckman SW41 Ti rotor. Pellets were resuspended in the protein loading buffer, analyzed by SDS-PAGE, and detected by a rabbit anti-M-PMV CA or a rabbit anti-HIV-1 CA polyclonal antibody on a Western blot.
Bacterial cell pellets (4 h postinduction) or 293T cells expressing the wild-type and chimeric proviral constructs grown on 100-mm plastic dishes were washed with phosphate-buffered saline (PBS) and transferred into a microcentrifuge tube. The cells were fixed with freshly prepared 2.5% glutaraldehyde in 0.1 M cacodylate buffer (pH 7.5), postfixed in 1% osmium tetroxide, dehydrated by applying an ethanol series (30, 50, 70, 80, 90, and 100%), and embedded in fresh Agar100 epoxy resin. Ultrathin sections (70 nm) were cut on an RMC MT 7000 Ultramicrotome and placed onto 200-mesh copper grids. The sections were contrasted with saturated uranyl acetate and lead citrate and studied by transmission electron microscopy (JEOL JEM-1200EX) with a microscope operated at 60 kV.
Several studies have shown that retroviral CANC assembles into tubular, mature-virus-like structures (11, 18, 46, 55). However, by preventing the formation of a β-hairpin at the N terminus of the CA (e.g., by deleting the N-terminal proline or by extending the N terminus), the assembly can be steered from yielding the mature lattices in favor of the immature spherical products (17, 44, 46, 55, 57). M-PMV ΔProCANC, the truncated domain of Gag, has been previously shown to be sufficient for assembly of immature, spherical VLPs in bacterial cells and in vitro (41, 46, 55). This construct was used as the starting point to delineate the contribution of NC interactions to the assembly of immature particles. This approach eliminated any interference from other Gag domains that may be potentially involved in the assembly process. The constructs used in this study are schematically shown in Fig. Fig.1.1. Chimeric proteins in which the entire NC region was replaced by either DLZ or TIZ (both derived from the yeast GCN4 [21, 22, 33]) were cloned and designated ΔProCADLZ and ΔProCATIZ, respectively. The constructs were expressed in E. coli BL21(DE3) and purified to homogeneity (Fig. (Fig.2).2). Their ability to form the expected oligomers, i.e., dimers and trimers, respectively, was assayed using analytical gel filtration coupled to a light-scattering flow cell. Peaks containing proteins with molecular weights corresponding to a dimer of ΔProCADLZ and a trimer-tetramer of ΔProCATIZ (Fig. 3A and B) were detected. The ability of the purified proteins to assemble into regular structures such as VLPs in vitro was subsequently investigated using an established in vitro assembly assay (55). No assembled particles were observed by transmission electron microscopy for either ΔProCADLZ or ΔProCATIZ (data not shown).
The inability of ΔProCADLZ and ΔProCATIZ proteins to assemble into regular structures could be caused by placing the zipper domain directly after the C terminus of M-PMV CA protein. Johnson et al. (24) showed that the RSV Gag polyprotein, in which the NC was replaced by the CREB dimerizing domain, assembled into spherical particles when expressed in insect Sf9 cells. The insertion of the 5-amino-acid flexible linker GSGSG between CA-SP and the dimerizing domain resulted in the formation of tubular particles (24). These data suggested that the spacing between CA and the multimerization domain may be critical. However, a more recent work by Keller et al. clearly demonstrated that in RSV the SP and the closely adjoining sequences (at least 8 residues of CA, all 12 residues of SP, and the first 4 residues of NC) play a critical role in the formation of immature particles (26). Since M-PMV has no similar spacer peptide connecting the CA and NC, we first determined whether the explanation for the failure of the ΔProCADLZ/TIZ proteins to assemble may be due to the absence of a certain specific sequence which corresponds to a structural motif originally residing in the deleted M-PMV NC region. To analyze these possibilities, construct ΔProCANC23, which included the first 23 amino acids of M-PMV NC, was generated. The dimerizing and trimerizing domains were subsequently fused to ΔProCANC23, and the corresponding proteins (ΔProCANC23, ΔProCANC23DLZ, and ΔProCANC23TIZ) were expressed in E. coli cells, purified, and assembled in vitro. All three proteins assembled into spherical particles (Fig. (Fig.4),4), revealing that both zinc fingers are dispensable for the assembly and that the 23 N-terminal amino acids of NC are sufficient to promote the multimerization of M-PMV CA molecules in the presence of nucleic acid. Moreover, DLZ and TIZ domains had no effect on assembly and the shape and size of the resulting particles (Fig. 4C and D). Light-scattering analysis of these proteins (ΔProCANC23, ΔProCANC23DLZ, and ΔProCANC23TIZ) showed polydisperse mixtures of high-molecular-weight oligomers (data not shown).
Next we attempted to design an “SP-like” linker joining M-PMV CA and the dimerizing or trimerizing domains in order to investigate whether these fusions could functionally substitute for NC23. First we determined the longest N-terminal sequence of NC that does not support the assembly of CA and used it in lieu of the spacer. As the contribution of basic residues within NC to the assembly of immature particles is well known (5, 12, 50-52), we focused on the lysine residues within the first 23 N-terminal amino acids, i.e., Lys10, Lys16, Lys18, and Lys20 (Fig. (Fig.5).5). To investigate their contribution to assembly, four constructs, i.e., ΔProCANC19, ΔProCANC17, ΔProCANC15, and ΔProCANC9, containing 3, 2, 1, and 0 lysine residues in the linker, respectively, were prepared. To check whether these shortened NC sequences still promoted CA assembly, we examined thin sections of E. coli cells expressing the corresponding fusion proteins. This analysis showed that the construct with a stop codon introduced in the place of Lys16 (ΔProCANC15) failed to assemble spherical particles, and only tubular structures or protein layers were observed (Fig. (Fig.6).6). Our earlier in vitro studies identified an internal cleavage site of the M-PMV protease exactly within the NC sequence FLNN15*16KNKE (45). Hence, we decided to use the first 15 N-terminal NC amino acids as a natural linker between CA and the leucine zipper domains (ΔProCANC15). To analyze whether the extension of ΔProCANC15 by DLZ or TIZ could rescue the ability of this construct to assemble, constructs ΔProCANC15DLZ and ΔProCANC15TIZ were prepared. The EM analysis of these two proteins both in E. coli and in suspensions of in vitro-assembled proteins revealed that none of these domains could restore assembly, and only amorphous aggregates were observed (data not shown).
As the leucine zipper replacements of NC fused to CA-SP in both RSV ΔPR Gag (24) and HIV-1 Gag (2, 14, 62) were shown to be sufficient for assembly of immature particles, we analyzed the difference between the N-terminal 15-amino-acid-long putative spacer peptide of M-PMV and the spacer originating from HIV-1. The HIV-1 spacer peptide sequence was inserted between M-PMV CA and DLZ (ΔProCASP1DLZ) or TIZ (ΔProCASP1TIZ) (Fig. (Fig.1A).1A). However, no assembly of organized structures was observed either in E. coli or in vitro. From this and the data shown above, it is obvious that at least the first 23 N-terminal amino acids of NC are absolutely necessary for the assembly of M-PMV and that this domain cannot be functionally replaced by DLZ, TIZ, or the HIV-1 spacer sequence SP.
In order to see whether the above results were specific for M-PMV, we analyzed the effect of DLZ and TIZ on the assembly of HIV-1 ΔProCASP1NC. (Note that previously the effects of the leucine zipper domains on HIV-1 [2, 14, 62] and RSV  assembly were studied in the context of the whole Gag.) The HIV-1 chimeric proteins in which the NC was replaced by dimerizing or trimerizing GCN4 leucine zipper domains, resulting in ΔProCASP1DLZ and ΔProCASP1TIZ, respectively, were prepared (Fig. (Fig.1B).1B). Although ΔProCASP1NC and ΔProCASP1 formed tubes both within E. coli (Fig. 7A and B, respectively) and the in vitro system (55), the replacement of NC by leucine zipper domains completely abolished this ability, and only amorphous material was observed for both proteins (Fig. 7F and G). Thus, the addition of oligomerization domains had an effect similar to that observed for M-PMV. However, as shown in previous studies the context of full-length Gag is important for the assembly of immature particles and can completely change the outcome of the above substitutions (2, 14, 24, 62).
Previously, two different leucine zipper domains (CREB and GCN4) were used to replace NC domains within Gag of HIV-1 (2, 14, 62) and RSV (24). Comparison of their amino acid sequences (Fig. (Fig.5)5) revealed a stretch of basic residues (RECRRKKKEY) at the N terminus of the CREB dimerization domain which does not occur in GCN4. To distinguish which part of CREB leucine zipper contributes to the assembly, three CREB-derived domains were fused to M-PMV ΔProCA, ΔProCANC15, and HIV-1 ΔProCASP1 proteins: (i) the whole CREB leucine zipper domain (CREB), (ii) a sequence containing only the N-terminal basic residues (CREB10), and (iii) CREB leucine zipper domain alone, i.e., lacking the sequence encompassing the N-terminal basic residues (Δ10CREB) (Fig. (Fig.1).1). All these proteins were then expressed in E. coli BL21(DE3), purified, and assayed for assembly by EM.
M-PMV ΔProCACREB, ΔProCACREB10, and ΔProCAΔ10CREB did not reveal formation of any spherical particles either in thin sections of bacterial cells expressing these proteins or in negatively stained samples of in vitro assembly reactions (data not shown). Ultrastructural analysis of bacterial cells expressing the M-PMV ΔProCANC15CREB10 revealed formation of mainly spherical particles with a few tubular structures, i.e., a phenotype characteristic of ΔProCANC. Spherical particles and spiral structures formed in E. coli were observed for both ΔProCANC15CREB and ΔProCANC15Δ10CREB (data not shown).
In vitro, only chimeric proteins containing either the complete CREB leucine zipper domain or CREB10 downstream of the NC15 linker (i.e., ΔProCANC15CREB and ΔProCANC15CREB10) formed spherical particles (Fig. 8A and B) similar to those of ΔProCANC (Fig. (Fig.4A).4A). Chimeric protein ΔProCANC15Δ10CRΕΒ lacking the stretch of basic residues formed mainly aggregates (Fig. (Fig.8C).8C). To evaluate whether the sequence of the NC15 linker is essential for CREB-driven assembly, this region was replaced with that of HIV-1 SP1 in the three constructs ΔProCASP1CRΕΒ, ΔProCASP1CRΕΒ10, and ΔProCASP1Δ10CRΕΒ (Fig. (Fig.1A).1A). No particles were observed either in the cells expressing these constructs or in vitro (data not shown). In summary these data suggest that the 15 N-terminal amino acids of M-PMV NC followed by a short sequence containing basic residues originating either from the M-PMV NC (K16NK18EK20) or from the CREB-derived sequence (RECRRKKKEY) are required for M-PMV assembly. The NC15 cannot be functionally replaced by HIV-1 SP1 sequence.
In order to determine whether portions of HIV-1 NC could be replaced with CREB, various fusions of ΔProCASP1 with CREB domains were examined. EM analysis of bacterial cells expressing HIV-1 ΔProCASP1 with fusion to CREB10 basic residues showed assembled tube-like structures (Fig. (Fig.7D)7D) with appearance similar to those of HIV-1 ΔProCASP1NC (Fig. (Fig.7A).7A). The ΔProCASP1CREB formed predominantly tubular and irregular structures (Fig. (Fig.7C).7C). Deletion of the N-terminal basic residues from CREB (ΔProCASP1Δ10CREB) completely abolished assembly as only inclusions containing nonstructured dense material were observed in bacterial cells (Fig. (Fig.7E).7E). The same phenotype was observed for HIV-1 CASP1GCN4 chimeras, i.e., ΔProCASP1DLZ (Fig. (Fig.7F)7F) and ΔProCASP1TIZ (Fig. (Fig.7G).7G). Whereas purified ΔProCASP1NC assembles in vitro mostly into tubular structures, CREB-containing chimeric proteins were very difficult to purify as the proteins tended to precipitate and underwent degradation (data not shown). This was most likely the result of perturbed folding and precluded preparation of a homogeneous population of in vitro-assembled particles using CREB-HIV chimeric proteins. Thus, while the basic residues within CREB can drive assembly of ΔProCASP1 into tubular structures, this sequence is not sufficient to promote formation of immature particles. Clearly, other regions provided by the Gag context are needed.
Recently published results from several laboratories have demonstrated that NC replacement by DLZ/TIZ within the whole Gag of RSV and HIV-1 is sufficient for assembly of immature particles at the plasma membrane (2, 14, 24, 62). Since the immature particles of M-PMV assemble within the cytoplasm, we analyzed a series of M-PMV Gag chimeric polyproteins, in which the NC was replaced by the three CREB-derived domains described above. For this purpose, CREB, CREB10, and Δ10CREB sequences were inserted either between MA-PP-p12-CA-NC15 and p4 of M-PMV Gag or between MA-CA-SP1 and SP2-p6 of HIV-1 Gag (Fig. (Fig.1C).1C). The chimeric Gag proteins were expressed in 293T cells. Similarly to data published for HIV MA-CA-SP1 (14), M-PMV Gag lacking the NC domain did not assemble into spherical particles within the infected cells (data not shown) while all chimeric constructs were released from cells with efficiency similar to that of the wild-type control (Fig. (Fig.9),9), indicating assembly. Although the deletion of the NC region abolished the frameshift sequence and only Gag polyprotein precursor was synthesized, some internal cleavage of both M-PMV and HIV-1 Gag was observed (Fig. (Fig.9).9). The same cleavage pattern was observed in viral constructs lacking the pro region, suggesting that the observed cleavage of chimeric Gag polyprotein is caused by cellular proteases (data not shown).
The morphology of the particles was analyzed by EM (Fig. (Fig.10).10). The numbers and types of resulting structures were scored using 15 to 20 micrographs, and each experiment was performed in triplicate to eliminate possible artifacts due to differences in the transfection efficiency. Fully and partially assembled particles were observed in approximately equal amounts for M-PMV GagCREB (Fig. 10B) and GagCREB10 (Fig. 10C). Although about 10% of the particles revealed the C-type morphology and were assembled at the plasma membrane, the majority of these particles was preassembled within the cytoplasm. In the cells producing M-PMV GagΔ10CREB only about 30% of particles were complete whereas 70% of all particles were only partially assembled (Fig. 10D). The assembly of this chimera probably occurs at the membrane since only partially assembled particles were occasionally observed within the cytoplasm.
HIV-1 chimeric Gag proteins in which the NC domain was replaced either by the complete leucine zipper domain containing the N-terminal basic residues (GagCREB) or by the leucine zipper domain lacking the N-terminal basic sequence (GagΔ10CREB) formed a population of immature particles consisting of regularly shaped spherical VLPs and of larger, irregular VLPs (Fig. 10F and H). Surprisingly, the chimeric HIV-1 Gag in which NC was replaced only by the N-terminal basic part of CREB (GagCREB10) failed to assemble. Layers of accumulated proteins that tended to detach from the plasma membrane and form large vesicles (of more than 500-nm diameter) were observed for this chimeric protein (Fig. 10G).
This study investigated the contribution of NC to the oligomerization of CA and assembly of Gag into immature particles. As dimerization and trimerization have been implicated in Gag assembly (2, 14, 24, 62), we addressed primarily the question whether the dimerization or trimerization motifs could drive the assembly of CA alone as well as within the context of complete Gag. We have created a series of chimeric proteins derived from M-PMV and HIV-1 ΔProCANC and Gag in which NC was replaced by the leucine zipper domains from GCN4 and CREB proteins. The effect of these substitutions was examined by monitoring assembly in vitro and in vivo.
The results provide solid evidence that the very N-terminal part of NC is essential for the assembly of M-PMV immature particles. Similarly to the other retroviruses, both zinc fingers were found dispensable while a minimal domain encompassing the first 23 amino acids of NC (NC23) proved to be sufficient for immature particle assembly both in bacteria and in vitro. The first 17 amino acids (ΔProCANC17) were found essential for assembly into spherical immature particles in bacteria but failed to assemble in vitro. However, the ΔProCANC15 and ΔProCANC9 failed to form spheres and assembled only into tubes resembling those of RSV SP mutants (26). These authors identified an extended SP assembly domain encompassing the last 8 CA residues, the whole SP, and the first 4 residues of NC. This assembly domain was shown to be required for the formation of RSV immature particles since further insertion mutations led to the assembly of tubular lattices (26). Similarly, mutations within SP and in its vicinity have been shown to interfere with HIV-1 and bovine immunodeficiency virus (BIV) particle assembly (1, 19, 20, 29, 32, 38). Although M-PMV lacks a clear SP sequence between CA and NC (45), we suggest that the N terminus of NC (NC15), comprising the first 15 amino acids of NC and likely protruding into the CA, might act as a similar “assembly domain.” Similarly to RSV (26) and BIV (19), this M-PMV-specific sequence cannot be functionally replaced with HIV SP. Neither dimerization, trimerization, a stretch of basic residues alone, nor their combination can substitute for this region. These data suggest a specific role of the N terminus of NC during immature particle assembly.
The reason why the hypothetical “SP-like” sequence is not cleaved off and stays as a part of NC (and presumably also CA) after maturation remains to be elucidated. Although the recombinant M-PMV proteinase produces small amounts of N-terminally truncated forms of NC in vitro (cleavage in positions NC7, NC11, and NC15) detected by mass spectroscopy (45), no other proteolytic cleavage site apart from the one between CA and NC has been detected in the mature virion (23). This suggests that the putative spacer is cleaved during maturation into two moieties which remain attached to C and N termini of CA and NC, respectively. This would lead to its “inactivation” with the effect similar to the removal of HIV-1 or RSV SP during maturation. The reason for such a different mechanism (and the absence of M-PMV spacer peptide in the mature virion) might be related to the different core structures of these retroviruses. The M-PMV core is tubular, in contrast to the conical one of HIV or the angular appearance of RSV core (9).
As mentioned previously, the basic residues downstream of CA are the key players in the assembly similar to those present in the CREB DLZ (RECRRKKKEY) and enabled both M-PMV ΔProCANC15 and HIV ΔProCASP1 to assemble. In contrast the extension of ΔProCANC15 with DLZ/TIZ did not rescue the assembly of M-PMV. The same fusion had an inhibitory effect on the assembly of the otherwise assembly-competent HIV-1 ΔProCASP. The extension of M-PMV ΔProCANC15 by additional NC sequence of 5 amino acids comprising the three lysine residues K16, K18, and K20 was also sufficient for the immature particle assembly. These data are in agreement with the work published by Sandefur et al. (50) in which they showed that the first seven amino acids of the HIV-1 NC (MQRGNFR), containing two arginine residues, are sufficient for VLP reconstitution. The only study focusing on the assembly of Gag-dimerizing chimeras in vitro (14) showed that the presence of DLZ was not sufficient for the assembly of recombinant purified HIV-1 MACASP1-DLZ unless a cofactor, such as nucleic acid, was added. This observation is in agreement with our data as we also show the necessity of basic residues for the assembly. These results suggest that a nucleic acid binding domain, i.e., a stretch of basic residues (originating from either M-PMV NC or CREB) or the positively charged face of MA, is necessary for in vitro assembly of CANC and Gag, respectively. This is in accord with the other studies (5, 12, 50-52, 61, 62) demonstrating the requirement for basic residues for immature particle assembly in the presence of nucleic acids.
In previously published studies the NC substitutions by DLZ or TIZ domains have been made in the context of the whole Gag polyprotein where the leucine zipper domains were positioned directly after the spacer peptides, i.e., SP1 in HIV-1 and SP in RSV (2, 14, 24, 62). In the light of the recent evidence for RSV (26), it is unlikely that the optimal placement of the dimerization domain controls the shape of Gag particles, as believed previously (24). The question of whether the immature particle assembly can be driven by dimeric or trimeric interactions, however, remains open. Given that the M-PMV ΔProCANC15 assembles in vitro when extended with basic residues while the same protein does not assemble when extended with oligomerization domains, it seems unlikely that strict stoichiometric interactions (e.g., dimerization or trimerization) are required. Instead we prefer a model in which initiation of immature particle assembly is driven by interactions between NC and RNA, in agreement with previously published data (5, 11, 12, 40, 50). These interactions with the nucleic acid bring CA molecules together and may trigger an allosteric effect through the SP-like sequence in the C terminus of CA, making it competent for immature assembly. Alternatively, NC-RNA interactions compensate for the entropic penalty associated with assembly and prevent the adoption of mature-assembly-like configuration of CA CTD by a simple steric restriction.
Similarly to the results published for RSV and HIV-1, we found that NC replacement with DLZ in HIV-1 and M-PMV Gag polyproteins did not interfere with assembly. This is in contrast with the in vitro results reported here and elsewhere for CA assembly. Clearly, additional domains within Gag, binding of Gag to the plasma membrane, or other cellular factors were able to augment assembly inside cells.
However, a more detailed analysis of VLP morphology and the assembly locus within the cell revealed remarkable differences, depending on the type of chimera and the virus used (summarized in Fig. Fig.1C).1C). A striking difference was found between CREB chimeras of M-PMV Gag. Chimeric M-PMV Gag with CREB and GagCREB10, both including the 10 basic upstream residues, formed spherical immature particles within the cytoplasm, i.e., followed the typical D pathway. On the other hand, M-PMV GagCREB chimeras lacking the basic N-terminal region resulted in a mix of regular and partially assembled particles, formed mainly at the plasma membrane. Although these proteins contain the signal for transport to the pericentriolar site (residing in the MA domain of Gag) (53, 56), the lack of interaction with nucleic acids disrupted cytoplasmic assembly and the unassembled protein eventually ended underneath the plasma membrane, where matrix interactions with the leaflet might partially compensate for the defect. Assembly of HIV-1 Gag chimeras revealed the necessity for a dimerization domain (additional to that found in CA) since the stretch of basic residues alone failed to assemble. This is rather surprising in the light of data demonstrating the importance of NC basic residues for HIV Gag multimerization despite their dispensability for Gag membrane binding (42). In the HIV case protein-protein interactions clearly dominate over nucleic acid binding, which is essential for M-PMV assembly. The difference may be due to the protein makeup of M-PMV and HIV-1 Gag polyproteins. M-PMV Gag contains the p12 protein that has been shown to be important for cytoplasmic assembly (28, 47-49, 54). Its internal scaffold domain (ISD) was capable of inducing intracytoplasmic assembly of HIV-1 when fused to Gag, which normally assembles at the plasma membrane (47).
Our results show different requirements for assembly of M-PMV and HIV-1 ΔProCANC in vitro compared to the in vivo assembly of Gag. We propose that multimerization mediated by interactions of basic residues with nucleic acids brings CA molecules to sufficient proximity, perhaps to evoke some structural changes required for the initiation of assembly or simply to overcome the entropic penalty. This is consistent with the data demonstrating the requirement of oligonucleotides spanning at least two monomers of RSV CANC (35). In the case of whole Gag, the situation is more complex. In addition to CA and NC other domains of Gag are involved in the process. It is reasonable to hypothesize that M-PMV and HIV-1, i.e., retroviruses with different morphogenetic pathways, evolved different interacting domains to direct and drive either intracytoplasmic or membrane assembly, respectively. In M-PMV Gag, additional interacting motif p12, as an internal scaffolding domain, eliminates the requirement of membrane for Gag assembly as suggested by Sakalian et al. (47-49). In contrast to M-PMV, the MA-membrane interaction provides additional driving force and allows the assembly at the membrane which is the predominant pathway for retroviruses including HIV-1. Such complexity of Gag prevents any further dissection of individual interacting regions by using the chimeric Gag constructs. Our in vitro approach is one way to circumvent this limitation and address the role and evolution of self-assembling domains in retroviruses.
We thank Jeremy Luban from University of Geneva for providing us with HIV-1 vector pSAX2. We are also grateful to Zdeněk Knejzlík from ICT, Prague, for GCN4 expression plasmid. We thank Romana Cubínková for excellent technical assistance and Jana Šípová for critical reading of the manuscript.
The work was supported by the Grant Agency of the Czech Republic, grant 204/09/1388 to M.R.; by the research project 1M0508 and Z40550506 from the Czech Ministry of Education to I.P.; MSM 6046137305, ME 904, National Institutes of Health, United States, grant CA 27834; and EUROCORES Programme EuroSCOPE of the European Science Foundation, European Commission contract no. ERAS-CT-2003-980409.
Published ahead of print on 9 December 2009.