Role of Mason-Pfizer Monkey Virus CA-NC Spacer Peptide-Like Domain in Assembly of Immature Particles

ABSTRACT

The hexameric lattice of an immature retroviral particle consists of Gag polyprotein, which is the precursor of all viral structural proteins. Lentiviral and alpharetroviral Gag proteins contain a peptide sequence called the spacer peptide (SP), which is localized between the capsid (CA) and nucleocapsid (NC) domains. SP plays a critical role in intermolecular interactions during the assembly of immature particles of several retroviruses. Published models of supramolecular structures of immature particles suggest that in lentiviruses and alpharetroviruses, SP adopts a rod-like six-helix bundle organization. In contrast, Mason-Pfizer monkey virus (M-PMV), a betaretrovirus that assembles in the cytoplasm, does not contain a distinct SP sequence, and the CA-NC connecting region is not organized into a clear rod-like structure. Nevertheless, the CA-NC junction comprises a sequence critical for assembly of immature M-PMV particles. In the present work, we characterized this region, called the SP-like domain, in detail. We provide biochemical data confirming the critical role of the M-PMV SP-like domain in immature particle assembly, release, processing, and infectivity. Circular dichroism spectroscopy revealed that, in contrast to the SP regions of other retroviruses, a short SP-like domain-derived peptide (SPLP) does not form a purely helical structure in aqueous or helix-promoting solution. Using 8-Å cryo-electron microscopy density maps of immature M-PMV particles, we prepared computational models of the SP-like domain and indicate the structural features required for M-PMV immature particle assembly.

IMPORTANCE Retroviruses such as HIV-1 are of great medical importance. Using Mason-Pfizer monkey virus (M-PMV) as a model retrovirus, we provide biochemical and structural data confirming the general relevance of a short segment of the structural polyprotein Gag for retrovirus assembly and infectivity. Although this segment is critical for assembly of immature particles of lentiviruses, alpharetroviruses, and betaretroviruses, the organization of this domain is strikingly different. A previously published electron microscopic structure of an immature M-PMV particle allowed us to model this important region into the electron density map. The data presented here help explain the different packing of the Gag segments of various retroviruses, such as HIV, Rous sarcoma virus (RSV), and M-PMV. Such knowledge contributes to understanding the importance of this region and its structural flexibility among retroviral species. The region might play a key role in Gag-Gag interactions, leading to different morphological pathways of immature particle assembly.

INTRODUCTION

Assembly of immature retroviral particles is mediated by oligomerization of the multidomain polyprotein Gag, which invariably contains three structural domains: matrix (MA), capsid (CA), and nucleocapsid (NC). The immature particle has a roughly spherical shape assembled from Gag hexamers, in which the polyproteins are arranged radially with the N-terminal MA on the exterior surface and NC pointing to the center. Activation of viral protease leads to cleavage of the Gag precursor and triggers dramatic reorganization of the immature particle into a mature virion. In the virion, MA remains attached to the viral membrane and the released CA reassembles into a mature core containing the NC-RNA complex. Depending on the retroviral species, the shape of the core may be conical, tubular, spherical, or polyhedral.

The main Gag components responsible for the assembly of immature particles involve CA and the residues downstream of CA, including NC (1). CA, the main structural retroviral protein, consists of two independently folded domains, the N-terminal (NTD-CA) and C-terminal (CTD-CA) domains, which are connected by a short linker. Despite low sequence similarity among retroviral CA proteins, their secondary and tertiary structures are highly conserved (2–9). Both immature and mature viral shells are assembled into curved hexameric protein lattices; however, the spacing and arrangement of the CA domains in these two lattices are different (10–15). The relative positions and orientations, as well as the roles, of NTD-CA and CTD-CA in the assembly of immature and mature hexameric lattices are largely exchanged (16). The dimeric interface in CTD-CA is the only conserved interaction motif in both immature and mature lattices (16). While the immature hexameric lattice is formed mainly by CTD-CTD interactions and by residues downstream of CA, the formation of mature hexamer is driven by NTD-NTD interactions, involving helices 1, 2, and 3 (16, 17). Interactions between NTD-CTD subunits are present only in the mature lattice (18, 19).

Depending on the type of retrovirus, NC contains one or two CCHC zinc finger motifs, which specifically recognize and incorporate viral genomic RNA into the viral particles. NC also contains numerous basic residues shown to be critical for immature particle assembly (20–27). In addition to the role of NC in genomic RNA packaging and facilitating viral assembly (27, 28), it plays a crucial role in RNA dimerization, reverse transcription, and integration (29–32).

Mutational analysis of different retroviral species, such as alpharetroviruses (e.g., Rous sarcoma virus [RSV]), lentiviruses (e.g., HIV-1), gammaretroviruses (e.g., murine leukemia virus [MLV]), and betaretroviruses (e.g., Mason-Pfizer monkey virus [M-PMV]), revealed that the region between CTD-CA and NC is critical for the assembly of immature particles (20, 33–41). This region differs among retroviruses. In HIV-1 and RSV, CA and NC are separated by a short spacer peptide (SP1 and SP, respectively), while in M-PMV and MLV, the CA-NC domains are directly connected to each other. The length and sequence of the SP regions vary. This region consists of 12 amino acids in RSV and 14 amino acids in HIV-1. In both HIV and RSV, a series of mutational studies demonstrated that the region around SP with short extensions into the CTD-CA or NC region is essential for retroviral assembly and proper maturation of viral particles (35, 36, 38, 40–43). MLV has no distinct spacer peptide; however, it encodes a relatively long sequence (41 amino acids) within the C terminus of CA called the “charged assembly helix.” The common feature of the CA-NC interconnecting regions in HIV and RSV and the very end of CA in MLV is their predicted α-helical structure (33, 42). Indeed, the ability of peptides derived from the HIV-1 SP1 region to adopt an α-helical structure was confirmed by nuclear magnetic resonance (NMR) spectroscopy. However, the peptides did not adopt this structure in aqueous solution but did so only in the presence of 30% trifluoroethanol (TFE) (44), a solvent that can artificially affect the three-dimensional (3D) structure of proteins. In aqueous solution, a magic angle spinning (MAS) NMR study of an HIV-1 CA-SP1 fragment revealed that SP1 peptide is in a random-coil conformation and is mobile in the in vitro-assembled CA-SP1 tubular structures (45). Similarly, an NMR study of a 300-residue CA-SP1-NC fragment of HIV-1 Gag revealed that SP1 is disordered in both the presence and absence of nucleic acids (46). Using circular dichroism (CD) spectroscopy, researchers determined the concentration-dependent shift from unstructured peptide to α-helix in aqueous solution for the peptide sequence spanning the last eight amino acids of HIV-1 CA and the entire SP1 (47). In agreement, Bush et al. demonstrated with CD spectroscopy that a 29-mer of RSV SP-derived peptide shifts from a random coil to a helix in a concentration-dependent manner (48).

Cryo-electron tomography studies comparing the structures of immature particles of HIV-1, RSV, M-PMV (49), and MLV (50) showed that the region between CTD-CA and NC is strikingly different in these retroviruses. In RSV and HIV-1, this region has a rod-like structure descending from each hexamer below CTD-CA toward NC, while in M-PMV, no rod-like structure is visible (16, 49). Also, the cryo-electron tomograms of MLV particles assembled in vitro revealed that CTD-CA is connected to NC/RNA via a formation appearing as a rod-like structure (50). Wright et al. developed a model suggesting that these rod-like structures are formed by a “six-helix bundle” that stabilizes immature HIV particles (14). Cryo-electron microscopy (cryo-EM) analysis of HIV Gag-derived immature-like particles showed formation of six extended structures connecting CTD-CA to NC, which strongly supports the six-helix bundle model (51). The work of Bush et al. supported the six-helix bundle model as a structural motif critical for the assembly of immature particles of RSV Gag (48). Based on the concentration-dependent transition of SP-1 derived peptide, Datta et al. hypothesized that SP1 could act as a molecular switch initiating immature virus assembly (47).

M-PMV, as a member of the betaretrovirus family, has neither the cleavable spacer peptide sequence of alpharetroviruses and lentiviruses nor the charged assembly helix sequence of gammaretroviruses. Also, no rod-like structures in the CA-NC spacer region were observed by cryo-electron tomography of M-PMV Gag-derived CA-NC protein assembled in vitro (16). Nevertheless, we show here that, similar to the SP regions of HIV and RSV, the M-PMV region connecting the CA and NC domains (residues CA²¹⁵ to NC¹³) is essential for assembly of immature particles. CD spectroscopy revealed that a short SP-like domain-derived peptide (SPLP) did not adopt any defined secondary structure; a random coil structure prevailed at all tested concentrations. Neither increasing the SPLP concentration nor titrating the peptide with TFE induced significant conformational changes toward helicity. A minor concentration-dependent increase in α-helical content may be consistent with a partial helical arrangement of the coiled structure. Computational modeling of the SP-like domain extended CA hexamers was performed using the 8-Å cryo-EM density maps of M-PMV immature particles. This flexible fitting identified structural features which are strikingly distinct from those of other retroviruses and are proposed to be critical for the M-PMV immature particle assembly.

MATERIALS AND METHODS

DNA constructs.

Standard subcloning techniques were used for all DNA manipulations. All plasmids were propagated in Escherichia coli DH5α. Newly created constructs were verified by DNA sequencing. All bacterial vectors were based on the parental M-PMV vector ΔProCANCpET22b (52), and mutagenesis was carried out using two-step PCR. Introduction of SP-like domain mutations into M-PMV proviral vectors (pSARM4 and pSARM4-EGFP) was carried out in two cloning steps. First, the mutations created by two-step PCR were introduced into a helper vector (MHelppUC19) constructed by inserting M-PMV SacI-Eco72I fragments (nucleotides 1165 to 3275) into pUC19 (53). Following sequence verification, the SacI-Eco72I fragment of MHelppUC19 carrying the appropriate mutation was inserted into the M-PMV proviral constructs pSARM4 and pSARM-EGFP. Further details of the cloning strategy and full sequences of PCR primers can be obtained from the authors upon request.

Cell growth and virus production in tissue culture.

HEK 293T cells were grown in Dulbecco′s modified Eagle medium (DMEM) (Sigma) supplemented with 10% fetal bovine serum (Gibco) and 1% l-glutamine (PAA Laboratories). Transfection of the HEK 293T cells was performed using FuGene HD transfection reagent (Roche Molecular Biochemicals) according to the manufacturer's instructions. At 24 or 48 h posttransfection, virions in the culture medium were harvested, filtered through a 0.45-μm filter, and centrifuged through a 20% sucrose cushion at 210,000 × g for 1 h in a Beckman SW41Ti rotor. M-PMV proteins were detected by Western blotting using a rabbit anti-M-PMV CA polyclonal antibody (54).

Protein expression, radioactive labeling, and quantification of particle release.

HEK 293T cells transfected with the appropriate DNA were starved for 30 min in methionine- and cysteine-deficient DMEM, pulse-labeled for 30 min with 125 μCi/ml of Tran³⁵S-label (MGP, Czech Republic), and chased overnight in complete DMEM. The cells from pulse and pulse-chase experiments were washed with phosphate-buffered saline (PBS), lysed in a lysis buffer (25 mM Tris-HCl [pH 8.0], 50 mM NaCl, 1% Triton X-100, 1% sodium deoxycholate) for 30 min on ice, and clarified by centrifugation at 14,000 × g for 2 min. The culture medium of the chased cells was filtered through a 0.45-μm filter, and SDS was added to a final concentration of 0.1%. Viral proteins from both the cellular lysates and culture medium were immunoprecipitated using a polyclonal rabbit anti-M-PMV CA antibody, followed by incubation with immobilized protein A (Invitrogen). Radiolabeled proteins were separated by SDS-PAGE and detected using a Typhoon PhosphorImager. To quantify the amount of released particles, radiolabeled protein bands of ³⁵S-pulse-labeled Gag (Pr78) and pulse-chase-labeled virion-associated CA (p27) were quantified using ImageQuant TL (Amersham Biosciences). Obtained values for released viral proteins are shown as relative concentrations of CA correlated to the level of intracellular Gag in individual samples.

Quantitative Western blotting.

Virion quantification was carried out as described previously (53). Briefly, the amount of M-PMV CA in individual viral samples was determined by comparing the CA protein band intensities with a standard curve prepared using purified M-PMV CA (54). The samples of particles released into the cultivation medium, obtained as described above, were separated by SDS-PAGE, transferred onto a nitrocellulose membrane, and detected with anti-M-PMV CA antibody using West Femto chemiluminescent substrate and a LAS-2000 imager. Protein band densities were quantified using ImageQuant TL (Amersham Biosciences).

Single-round infectivity assay.

Infectivity was measured as previously described (53, 55). HEK 293T cells were cotransfected with either wild-type (WT) or mutant pSARM4-EGFP (56) and with the pTMO vector (57) containing the M-PMV env gene. After 48 h, virions were harvested, filtered through a 0.45-μm filter, and centrifuged through a 20% sucrose cushion at 210,000 × g for 1 h in a Beckman SW41Ti rotor. Each sample was normalized for CA content by quantitative Western blotting using polyclonal anti-M-PMV CA antibody. Equivalent amounts of viruses were used to infect fresh HEK 293T cells, and the cells were incubated for an additional 48 h. The cells were then fixed with 4% formaldehyde, and the numbers of green fluorescent protein (GFP)-positive (i.e., infected) cells were determined using flow cytometry (BD FACSaria).

Bacterial protein expression.

Luria-Bertani medium containing ampicillin (100 μg/ml) was inoculated with E. coli BL21(DE3) cells carrying the appropriate construct. When the cells reached an optical density at 590 nm of 0.8, protein expression was induced by the addition of isopropyl-β-d-thiogalactopyranoside (IPTG) to a final concentration of 0.4 mM. The cells were harvested at 4 h postinduction.

Protein purification.

The proteins ΔProCANC, CA-3 ΔProCANC, NC1 ΔProCANC, NC2 ΔProCANC, and NC6 ΔProCANC were expressed and purified according to the protocol described in our previous work (52, 58). Briefly, E. coli BL21(DE3) cells were harvested at 4 h postinduction and resuspended in buffer A (50 mM Tris-HCl [pH 8.0], 150 mM NaCl, 1 mM EDTA, 0.5% Triton X-100) with the addition of lysozyme, 0.1% deoxycholate (final concentration), DNase (Novagen), RNase (Roche), Pefablock (Roche), and protease inhibitor cocktail (Sigma). Cell pellets were washed in buffer A with increasing NaCl concentration (150 mM to 1 M). The cleared lysates were dialyzed overnight against buffer E (20 mM Tris-HCl [pH 8.0], 0.1 M NaCl, 50 μM ZnCl₂, 1 mM phenylmethylsulfonyl fluoride [PMSF]) at 4°C and loaded onto a zinc affinity chromatography column. The bound protein was eluted with a 0.1 M to 1 M gradient of NaCl in buffer E. Fractions containing the desired proteins were dialyzed overnight against storage buffer (buffer E with 0.5 M NaCl), concentrated, and loaded onto a Sephadex G-100 column. The purified proteins were concentrated and stored at −70°C. The purity of each protein was analyzed by SDS-PAGE.

In vitro assembly.

Aliquots (60 μg) of the purified proteins were mixed with 6 μg of nucleic acid (bacteriophage MS2 RNA; Roche Molecular Biochemicals) in a total volume of 100 μl storage buffer (20 mM Tris-HCl [pH 8.0], 0.5 M NaCl, 50 μM ZnCl₂, 10 mM dithiothreitol [DTT], 1 mM PMSF). The mixture was dialyzed overnight against assembly buffer (50 mM Tris-HCl [pH 8.0], 100 mM NaCl, 1 μM ZnCl₂) at 4°C. Synthetic peptides (SPLP, SYQQGLAMAAAFSGQTVKDFL; NC^3-21, AFSGQTVKDFLNNKNKEKG; and NC^3-21ctrl, VKNAETKGNLNFKFGKSQD) were alternatively added in a 1:1 molar ratio with the protein. The proteins were incubated with the peptides for 1 h on ice prior to addition of nucleic acid and dialysis.

Electron microscopy analysis.

For thin-section electron microscopic analysis, low-speed pellets of bacterial (4 h postinduction) or HEK 293T (48 h posttransfection) cells expressing the appropriate protein were fixed with 2.5% glutaraldehyde in 0.1 M cacodylate buffer pH 7.5), postfixed in 1% osmium tetroxide, dehydrated by applying an ethanol series, and embedded in fresh Agar100 epoxy resin. Thin sections (70 nm) were cut using an RMC MT 7000 Ultramicrotome. The sections were contrasted with saturated uranyl acetate or lead citrate. Particles formed during in vitro assembly were negatively stained with 2% sodium phosphotungstate (pH 7.3) on carbon-coated grids. Samples were analyzed with a JEOL JEM-1200EX microscope operated at 60 kV.

Peptide synthesis.

The peptide sequence was assembled in an ABI433A solid-phase synthesizer (Applied Biosystems, Foster City, CA, USA) by stepwise coupling of the corresponding 9-fluorenylmethoxy carbonyl (Fmoc)-amino acids to the growing chain on Rink amide resin (0.5 mmol/g). Fully protected peptide resin was synthesized according to a standard procedure involving cleavage of the Nα-Fmoc protecting group with 20% piperidine in dimethylformamide (DMF); coupling was mediated by mixtures of coupling HBTU/HOBt (HBTU, O-benzotriazole-N,N,N′,N′-tetramethyl-uronium-hexafluoro-phosphate; HOBT, 1-hydroxybenzotriazole) reagents in DMF.

CD.

Circular dichroism (CD) spectra of SPLP were collected on a Jasco-815 spectrometer in the 180- to 300-nm spectral range at room temperature. The peptide was measured in water at concentrations of 10 mg/ml (4.26 mM), 5 mg/ml (2.13 mM), 1 mg/ml (0.426 mM), 0.5 mg/ml (0.213 mM), and 0.1 mg/ml (0.0426 mM). The peptide at 0.1 mg/ml was also measured in water-trifluoroethanol (TFE) mixtures (10%, 20%, and 40% [vol/vol] TFE). For concentrations of 10, 5 and 1 mg/ml, we used 6-μm CaF₂ demountable cells, and for concentrations of 0.5 and 0.1 mg/ml, a standard 0.1-cm quartz cell (Hellma) was used. The experimental setup was as follows: 2 scans, 0.5-nm steps, 5-nm/min speed, 32-s time constant, and 1-nm spectral bandwidth. After baseline correction, the final spectra were expressed as molar ellipticity θ (degrees · cm² · dmol⁻¹) per residue. Numerical analysis of the secondary structure and secondary structure assignment were performed using the online circular dichroism analysis program Dichroweb (http://dichroweb.cryst.bbk.ac.uk) (59).

Computational modeling.

The electron density maps of an immature M-PMV particle at 8-Å resolution (16) were obtained from the EMDataBank (http://www.emdatabank.org/) under accession code EMD-2089. The starting model comprised the CA dimer-spanning residues ¹⁵⁶FADF to CSDI²¹⁶ fitted into the density as hexamers.

The downstream SP-like domain (CA residues G²¹⁷ to M²²⁶ and NC residues A¹ to A³) was modeled using a modified homology modeling procedure in Modeler 9.12 (60). Two models were considered (random coil and α-helix), in which CA residues 205 to 217 were restrained to form helix 11 and CA residues 220 to 226 together with the NC residues 1 to 3 were forced into the respective secondary structure. Both models included a disulfide bridge between residues C¹⁹³ and C²¹³, which was added during the homology modeling procedure. Modeled dimers were fitted into the EMD-2089 density map using UCSF Chimera (61), and backbone torsions of selected residues of one chain were manually adjusted to fit to the density map. This chain was copied and fitted to the density map to form a hexamer of dimers. This structure was subjected to 1-ns molecular dynamics density fitting (62) using VMD (63) and NAMD (64). The CHARMM22 force field (65, 66) was used together with secondary structure, chirality, peptide bond, and 6-fold symmetry restraints.

RESULTS

Mutational analysis of the M-PMV SP-like domain.

Our previous work showed that the N-terminal part of M-PMV NC encodes two motifs indispensable for the assembly of immature virus-like particles (VLPs): an assembly domain (NC^1-15) and a basic region (NC^16-20) (Fig. 1A) (20). The basic region contains a sequence of three lysine residues (K¹⁶K¹⁸K²⁰) crucial for the immature particle assembly. However, this sequence is not specific and can be replaced with a stretch of other basic residues (20). Deletion or substitution of the motif encompassing the N-terminal 15 amino acids of NC abrogates the assembly of the immature virus-like particles in in vitro and in vivo systems (20). To verify that the NC^1-15 domain does not serve as a mere “spacer” between the CTD-CA and NC domains, we replaced this region with either HIV-1 SP1 or a GSG linker (Fig. 1B) in both bacterial and mammalian expression vectors. The bacterial vector contains the gene encoding the minimal assembly-competent fragment of M-PMV Gag, ΔProCANC (52, 58), while the mammalian vector encodes the entire M-PMV proviral DNA. EM analysis confirmed that NC^1-15 replacement is fatal for the assembly of immature particles of either ΔProCANC VLPs in E. coli (Fig. 1C to E) or immature viruses in HEK 293T cells (Fig. 1F to H). Protein structures formed of mutant CANC protein in E. coli (Fig. 1D and E) are most probably sheets of misassembled protein. We have not seen the identical phenotype in E. coli so far, except for differently organized and less abundant layers of sheets observed in E. coli expressing the M-PMV ΔProCANC¹⁵ mutant (20).

FIG 1.

Domains of the M-PMV N-terminal NC region. (A) Amino acid sequences of M-PMV assembly domain (NC^1-15) and basic region (NC^16-20). (B) Amino acid sequences of HIV-1 SP-1 and a GSG linker used to replace the M-PMV assembly domain. (C to E) Thin sections of E. coli expressing WT ΔProCANC (C) or its mutants in which the NC^1-15 assembly domain was replaced with the HIV-1 SP-1 sequence (D) or a GSG linker (E). (F to H) HEK 293T cells expressing proviral WT M-PMV (F) or its mutants in which the NC^1-15 assembly domain was replaced with the HIV-1 SP-1 sequence (G) or a GSG linker (H).

The secondary structure prediction (http://bioinf.cs.ucl.ac.uk/psipred/) of the M-PMV CA-NC junction revealed a potential α-helical region spanning amino acid residues from serine S²¹⁹ of CA to leucine L¹³ of NC (Fig. 2A). This suggests that the NC^1-15 assembly domain might be extended upstream to S²¹⁹ of CTD-CA. Based on its similar position with the spacer peptides of HIV and RSV (Fig. 2B), we named this region the spacer peptide-like (SP-like) domain. To study its role in the assembly of M-PMV, we performed a scanning mutagenesis. As the region spanning the M-PMV SP-like domain is rather long (33 amino acids), we decided to prepare 16 double mutants instead of 33 individual mutations. The presence of four alanine residues in the vicinity of the CA-NC cleavage site (GLAM*AAAF) prevented us from mutating this region using a series of alanine pairs (AA). Therefore, we used a combination of neutral (i.e., alanine) and small (i.e., glycine) amino acids. We prepared a series of double mutants (XY/AG), successively replacing every two amino acids (XY) of the region spanning the R²¹¹ residue of CA to the N¹⁷ residue of NC with alanine and glycine residues (AG) (Fig. 2C). For simplified orientation, the double mutants were labeled as presented in Fig. 2C. All mutants were cloned into an M-PMV proviral vector and expressed in HEK 293T cells. Western blot analysis showed that several mutations (CA-5, CA-3, CA-1, CA0, NC1, NC3, NC4, and NC6) abolished viral release (Fig. 2D). To further study the impact of the SP-like domain mutations on M-PMV polyprotein expression, proteolytic processing, and kinetics of particle release, we performed a pulse-chase experiment. HEK 293T cells, transfected with the WT and appropriate SP-like domain mutants, were radioactively pulse-labeled for 30 min and chased overnight. M-PMV proteins were immunoprecipitated using anticapsid polyclonal antibody from the cell lysates and culture medium and analyzed by SDS-PAGE followed by autoradiography. All mutants expressed the Gag-related polyprotein precursors (i.e., Gag, Gag-Pro, and Gag-Pro-Pol) of correct molecular masses and at levels similar to those of the wild-type virus (Fig. 3A). In contrast to the WT, impaired processing of M-PMV polyprotein precursors was observed in cells expressing the mutants CA-5, CA-3, CA-1, CA0, NC1, NC3, NC4, and NC6 during the 16-h chase (Fig. 3B). Correctly sized mature CA protein was observed in the culture media of the CA-7, CA-6, CA-4, CA-2, NC2, NC5, NC7, and NC8 mutants, suggesting their regular particle processing and release. In accord with the Western blot analysis (Fig. 2D), we observed impaired particle release into the culture media for the CA-5, CA-3, CA-1, CA0, NC1, NC3, NC4, and NC6 mutants (Fig. 3C). The efficiency of M-PMV particle release was calculated based on the relative ratios of cell- and virion-associated protein levels (Fig. 3D). According to this quantification, we divided the M-PMV SP-like domain mutants into three categories: (i) WT-like level of particle release, i.e., NC2, NC5, NC7, and NC8; (ii) 20 to 70% of WT release, i.e., CA-7, CA-6, CA-4, and CA-2; and (iii) less than 20% of WT release, i.e., CA-5, CA-3, CA-1, CA0, NC1, NC3, NC4, and NC6 (Fig. 3D).

FIG 2.

Mutations in the spacer peptide-like domain. (A) Amino acid sequence of the WT M-PMV SP-like domain with predicted helical sequence from CA S²¹⁹ to NC L¹³. The position of the predicted helical structure element is shown above the sequence. (B) Schematic diagram of the predicted helical structures in HIV-1 SP-1 and RSV SP. Conserved proline residues upstream of putative helices are labeled in bold. (C) Schematic diagram of the amino acid sequences of the M-PMV SP-like domain mutations used in this study. A series of double mutants (XY/AG) replacing every two amino acids (XY) of the region spanning the R²¹¹ residue of CA to the N¹⁷ residue of NC by with alanine and glycine residues (AG) was prepared. For simplified orientation, the double mutants were numbered according to CA or NC position. (D) Western blot analysis of released WT and SP-like domain mutant M-PMV particles. The VLPs from the culture media were collected by centrifugation through a 20% sucrose cushion at 48 h posttransfection of HEK 293T cells with WT and or mutant proviral DNAs. The viral proteins were analyzed by SDS-PAGE, transferred onto a nitrocellulose membrane, and detected by with rabbit antibodies raised against M-PMV CA.

FIG 3.

Synthesis, processing, and release of WT M-PMV and SP-like domain mutants. HEK 293T cells were transfected with WT or SP-like domain mutant M-PMV proviral DNAs. Viral proteins were metabolically labeled with a [³⁵S]cysteine-methionine mix for 30 min and then chased for 16 h. M-PMV CA-related proteins were then immunoprecipitated from the cells and culture medium and analyzed by SDS-PAGE. (A and B) Intracellular M-PMV proteins Gag (Pr78), Gag-Pro (Pr95), and Gag-Pro-Pol (Pr180) immunoprecipitated from the cell lysate after a 30-min pulse (A) and after a 16-h chase (B). (C) CA-derived proteins of released M-PMV particles were immunoprecipitated from the culture medium 16 h after the chase. (D) Quantification of WT and mutant M-PMV particle release from HEK 293T cells. Band intensities of ³⁵S-pulse-labeled Gag (Pr78) and released CA were calculated. The relative percentage of CA released into the culture medium was corrected for intracellular expression of individual samples.

The SP-like domain affects morphogenesis of the virus.

To study the ability of the M-PMV SP-like domain mutants to form immature particles, we performed transmission electron microscopic (TEM) analysis of thin-sectioned HEK 293T cells expressing the M-PMV mutants. The mutants NC2, NC5, NC7, and NC8, for which regular processing and moderately reduced particle release were observed, produced mature virus-like particles (Fig. 4). A significantly different phenotype was found in cells expressing the release-affected or release-incompetent mutants CA-7, CA-6, CA-5, CA-4, CA-3, CA-2, CA-1, CA0, NC1, NC3, NC4, and NC6. In cells expressing these mutants, no typical D-type intracytoplasmic particle formation was detected. Instead, we observed large electron-dense layers underneath the plasma membrane (Fig. 4). These layers were of variable size, ranging from approximately 100 nm (Fig. 4, NC1, NC4, and NC6) to several micrometers (CA-5, CA-3, CA0, and NC3). Only a few spherical structures (Fig. 4, NC1, NC6, arrows) were detached from the plasma membrane; however, most of these structures remained attached to the membrane, did not affect membrane curvature, and were incapable of initiating the budding process (Fig. 4, CA-5, CA-1, CA0, NC3, NC4, and NC6).

FIG 4.

Transmission EM images of HEK 293T cells expressing WT M-PMV and SP-like domain mutants. At 48 h posttransfection of HEK 293T cells with mutant proviral vectors, the cells were fixed in glutaraldehyde and postfixed in 1% osmium tetroxide. The sections were contrasted with uranyl acetate and lead citrate and analyzed using a JEOL JEM-1200EX analytical transmission electron microscope. Arrows point to spherical structures detached from the plasma membrane. Bars represent 200 or 500 nm, as indicated.

The SP-like domain is essential for infectivity of the virus.

To determine whether the M-PMV SP-like domain mutants that are released from the cells are affected in another step of the virus life cycle, we investigated their infectivity. For this purpose, we prepared reporter constructs in which the mutations CA-7, CA-6, CA-4, CA-2, NC2, NC5, NC7, and NC8 were engineered into pSARM4-EGFP, which expresses enhanced GFP (EGFP) instead of Env (56). HEK 293T cells were cotransfected with the resulting vectors or the WT together with an Env expression vector (pTMO) (57). At 48 h posttransfection, viruses from the culture media were normalized by CA content using semiquantitative Western blot analysis, and the same amounts of viral particles were used for infection of fresh HEK 293T cells. At 48 h postinfection, the cells were analyzed using flow cytometry. The relative infectivity of the wild-type virus (i.e., the percentage of EGFP-positive cells) was regarded as 100%. The infectivities of the SP-like domain mutants were then compared to that of the WT (Fig. 5). Whereas the NC5, NC7, and NC8 mutants exhibited moderately decreased infectivity (60 to 80% of the WT value), the CA-7, CA-6, CA-4, CA-2, and NC2 mutations completely blocked infectivity (Fig. 5).

FIG 5.

Relative infectivity of M-PMV SP-like domain mutants determined by single-round assay. HEK 293T cells were cotransfected with WT or SP-like domain mutant pSARM-EGFP and pTMO vectors. At 48 h posttransfection, the viruses from the culture medium were filtered and normalized by quantitative Western blotting for CA. Equivalent amounts of viruses were used to infect fresh HEK 293T cells. At 48 h postinfection, the cells were harvested and the numbers of GFP-positive cells were determined by flow cytometry (BD FACSaria). The mean percentage of three independent infectivity measurements (with calculated standard deviations) for each mutant relative to the WT is shown.

The SP-like domain is important for in vitro assembly of CANC virus-like particles.

To analyze the impact of SP-like domain mutations on the in vitro assembly of CANC, we selected four mutations, one allowing assembly and particle release (NC2) and three blocking assembly of immature particles and their release (CA-3, NC1, and NC6). These mutations were introduced into ΔProCANC and expressed in E. coli BL21. The mutant ΔProCANC proteins were purified, and their ability to assemble was tested using an in vitro assembly assay. In agreement with the results from tissue culture cells, the NC2 ΔProCANC protein formed spherical particles of a size similar to that of the WT (data not shown). As expected, the ΔProCANC proteins with CA-3, NC1, and NC6 mutations did not assemble into spherical structures but formed various aggregates (data not shown).

Characterization of SP-like domain-derived peptides.

To characterize properties of the SP-like domain peptide segment, we synthesized a series of short peptides corresponding to M-PMV CA²¹⁹-NC¹³, named SPLP (SYQQGLAMAAAFSGQTVKDFL), NC^3-21 (AFSGQTVKDFLNNKNKEKG), and NC^3-21ctrl (VKNAETKGNLNFKFGKSQD, a scrambled sequence of the NC^3-21 amino acids). First, we analyzed whether external addition of SPLP, NC^3-21, or NC^3-21ctrl influenced ΔProCANC protein oligomerization in our in vitro assembly assay. In a competitive in vitro assay, the peptides were added in a 1:1 or 2:1 molar ratio to the ΔProCANC protein and incubated prior to dialysis against the assembly buffer. Peptide addition did not lead to any observable differences in the ability of ΔProCANC to assemble spherical particles (data not shown). These data suggest that, under in vitro assembly conditions, these peptides did not interfere with protein-protein interactions and failed to block the assembly of immature particles.

Circular dichroism (CD) spectroscopic studies of HIV-1 (47) and RSV (48) proteins showed that the ability of SP-derived peptides to adopt an α-helical conformation is dependent on increasing peptide concentration. To analyze whether the SP-like domain of M-PMV undergoes similar concentration-dependent structural changes, CD spectroscopy was applied to study the peptide corresponding to the M-PMV CA²¹⁹-NC¹³ region (SPLP) (Fig. 6A). The CD spectra were measured in an aqueous environment at five different SPLP concentrations: 0.0426 mM (0.1 mg/ml), 0.213 mM (0.5 mg/ml), 0.426 mM (1 mg/ml), 2.13 mM (5 mg/ml), and 4.26 mM (10 mg/ml) (Fig. 6B). All spectra had a similar pattern, with a positive maximum at 190 nm and negative minimum with lower intensity at 216 nm. The data did not show any predominant proportion of α-helical structure at any of the SPLP concentrations tested. Furthermore, we did not observe an SPLP concentration-dependent shift from coil to helix. Numerical spectral analysis showed that SPLP adopts mainly a random-coil structure (from 29 to 37%) at all tested concentrations (Fig. 6C). The minor concentration-dependent increase of α-helical structure, from 18% for 0.1 mg/ml to 26% for 10 mg/ml (Fig. 6C), might be due to a partial helical organization of coiled structure predicted in one of our models (see below). However, in contrast to the case for RSV SP, titrating an aqueous solution of M-PMV SPLP with increasing concentrations of 2,2,2-trifluoroethanol (TFE) (up to 40%) did not induce any conformational changes toward helicity (Fig. 6D and E).

FIG 6.

Circular dichroism spectroscopy of SP-like domain-derived peptide SPLP. (A) Amino acid sequence of the SP-like domain-derived peptide (SPLP). (B) Spectra of SPLP in water at various concentrations ranging from 0.1 mg/ml to 10 mg/ml, measured at 25°C. (C) Quantification of the percentages of secondary structure elements based on the spectral analysis of SPLP shown in panel B. (D) Spectra of SPLP at 0.1 mg/ml in various concentration of TFE from 0% to 40%, measured at 25°C. (E) Quantification of the percentages of secondary structure elements based on the spectral analysis of SPLP shown in panel D. The highest percentage representation of secondary structure is shown in gray.

Modeling of the SP-like domain into 8-Å density maps of immature M-PMV CANC.

Two models adopting different secondary structure (random coil or α-helix) in the SP-like domain (from CA Y²²⁰ to NC A³) were created. Both the coil and helix models fit well with the experimental cryo-EM densities (cross-correlation coefficients of 0.404 and 0.415, respectively) (Fig. 7). The Pro²¹⁸ residue (part of the CA-4 mutation) was expected to terminate helix 11, as inferred from the experimental densities and bioinformatic predictions (Fig. 7). Therefore, in the helix model, a hypothetical helix 12 is formed. The most downstream extremities of the electron densities coincide with the CA-NC scissile bond in the coil and helix models, suggesting ordering in the CA part and disordering in the NC part.

FIG 7.

Computational interpretation of cryo-EM maps for the SP-like domain of immature M-PMV capsid. Isosurfaces of cryo-EM maps contoured at 3σ threshold are taken from EMDataBank (accession number EMD-2089) and overlaid with the computational models. The left panels (A, C, and E) show the helix model; the right panels (B, D, and F) show the coil model. The top (A and B) and bottom (E and F) panels show the view from the inside of the capsid, while the middle panels (C and D) show it from the side. A selected CA monomer is colored cyan, with the modeled region in green and sticks and the Pro²¹⁸ residue shown as red sphere.

The coil and helix models were used to attempt to link the experimental behavior of the SP-like domain mutants to the locations of the residues in the hexameric CTD. Residues Cys¹⁹³ and Cys²¹³ were modeled to form a disulfide bridge, as they were close to each other, maintained the stability of the model, and were compatible with the densities.

DISCUSSION

It has been well documented that the peptide linker connecting the CA and NC domains within Gag is crucial for immature particle assembly in various retroviruses (34–36, 42, 47). We show here that an analogous domain is essential for the assembly of M-PMV. We characterized M-PMV SP-like domain, comprising CA²¹⁵ to NC¹³, by identifying residues critical both for the assembly of immature particles and virus infectivity. For this reason, we prepared 16 double mutants spanning the SP-like regions. The impact of these mutations on the M-PMV life cycle is summarized in Table 1. In 12 mutants (CA-7, CA-6, CA-5, CA-4, CA-3, CA-2, CA-1, CA0, NC1, NC3, NC4, and NC6) intracytoplasmic assembly, as well as proteolytic processing and particle release, was either completely or severely impaired. With the exception of three mutants (NC5, NC7, and NC8), none of the SP-like mutants was infectious. Eight mutations within the M-PMV SP-like domain (CA-5, CA-3, CA-1, CA0, NC1, NC3, NC4, and NC6), as well as replacement of the C-terminal portion of the M-PMV SP-like domain (NC1-15) with HIV-1 SP1 or a GSG linker, led to accumulation of the M-PMV polyproteins underneath the plasma membrane. Although trafficking of the Gag polyprotein to the plasma membrane remained active, the polyproteins only followed the curvature of plasma membrane and were not capable of C-type assembly or budding. Interestingly, a similar phenotype, i.e., accumulation of large electron-dense plaques or patches underneath the plasma membrane, has been observed for HIV-1 SP1 mutants (36, 43). Based on these results, we believe that the SP-like domain is critical for triggering specific interactions of the Gag polyproteins leading to the formation of M-PMV immature intracytoplasmic particles.

TABLE 1.

Summary of phenotypic changes due to mutations of the M-PMV SP-like domain

Characteristic	WT	CA-7	CA-6	CA-5	CA-4	CA-3	CA-2	CA-1	CA0	NC1	NC2	NC3	NC4	NC5	NC6	NC7	NC8
Intracytoplasmic assembly	+++	+	+	−	+	−	+	−	−	−	++	−	−	++	−	++	++
Particle release	+++	++	++	−	++	−	++	−	−	−	+++	−	−	+++	−	+++	+++
Processing	+++	++	++	+	++	+	++	+	+	+	++	+	+	+++	+	+++	+++
Infectivity, % (mean ± SD)	100	0.4 ± 0.6	0.3 ± 0.4	NDa	4.2 ± 0.6	ND	0.3 ± 0.2	ND	ND	ND	0.7 ± 0.1	ND	ND	62 ± 15	ND	79 ± 9	56 ± 6

^aND, not defined.

As outlined in the introduction, the HIV-1 and RSV SP regions appear as a rod-like density descending from the center of immature hexamers below the CTD-CA (49), which is consistent with the six-helix bundle organization as proposed by Wright et al. (14) and confirmed by the high-resolution (9.4-Å) 3D structure of HIV Gag-derived immature-like particles (51). In vitro studies with the HIV SP1- and RSV SP-derived peptides also showed a shift to an α-helical conformation in a concentration-dependent manner (44, 47, 48). Despite the fact that the M-PMV CA-NC connecting region does not have a cleavable SP domain (67), its secondary structure prediction suggests formation of an α-helix, similar to the case for HIV-1 and RSV. However, it does not form any rod-like structure, characteristic of the six-helix bundle as seen in HIV and RSV (16, 49). This is consistent with the fact that in these two viruses, the six-helix bundle may be formed and fixed by a strong helical hydrophobic moment which is missing in the M-PMV SP domain (16, 49). Circular dichroism spectroscopy revealed, however, that increasing concentrations of M-PMV SP-like domain-derived peptide (SPLP) did not lead to the significant increase in helicity observed with HIV SP1- and RSV SP-derived peptides. As we did not observe a shift from coil to helix during the titration of an SPLP solution with TFE, we suppose that SPLP remained in a random-coil conformation that might have adopted a partially helical organization. In contrast to HIV-1 and RSV, in which the α-helicity of SP1- and SP-derived peptides increases proportionally with increasing TFE concentration (44, 47, 48), M-PMV SPLP had a low percentage of α-helical structure even at 40% TFE, which may be consistent with a partial helical content in a coil structure.

The cryo-EM-derived 8-Å electron densities of immature M-PMV (16) show that the parts corresponding to SP-like domains within the hexameric lattice form “kinked-rod”-like structures. To interpret these, we have extended the previous model (16) by a stretch of residues spanning CA G²¹⁷ to NC A³ region and fitted the model flexibly into the 8-Å electron density maps. In this model, helix 11 terminated at G²¹⁷, followed by the kink due to P²¹⁸. Interestingly, the Pro residue occurs at a very similar position also in other retroviral genera (Fig. 2A and B): lentiviruses (HIV), alpharetroviruses (RSV), deltaretroviruses (human T-cell leukemia virus [HTLV]), and gammaretroviruses (MLV). Therefore, we might hypothesize that this kink between the helix 11 and downstream SP region (helix 12) has some functional importance.

From residue Y²²⁰ onward, we modeled two variants, adopting different secondary structures, i.e., α-helix and random coil. Both the helix and coil models fitted well into the density and were also consistent with the CD spectra of the SPLP (see above). In both models there were numerous interactions of the SP-like domain with helix 10 of the neighboring CA dimers. Both models also suggest that the SP-like domain down to the scissile bond is ordered, whereas the N terminus of NC may protrude from the density and be disordered. Although we utilized a C¹⁹³-C²¹³ disulfide bond in our models, it is possible that these cysteines would be reduced, given the experimental reducing conditions. In summary, our computational model based on cryo-EM densities suggests a conformation of the M-PMV SP-like domain (kinked-rod structure) completely different from that observed for HIV or RSV (six-helix bundle).

By serving as a linker between CA, a primary mediator of assembly, and NC, which contains an RNA-binding domain, the SP domain is an important factor during maturation. Structural transition of CA from the immature particle conformation into that of the mature core requires proteolytic liberation of both termini. While the release of the HIV CA N terminus is the second proteolytic step, the release of the CA C terminus is the final step of Gag maturation (68). The formation of an N-terminal β-hairpin, a critical structural element of the mature CA-NTD conformation, not only takes place upon liberation of the N-terminal proline of CA, but also requires proteolytic action at the opposite site of CA, releasing CTD-CA from downstream SP-1 (69). In addition, NC and genomic RNA (gRNA) undergo transition from an immature to a condensed mature conformation upon processing of Gag (70, 71). Interestingly, the cleavage of SP1/NC is the first step of particle maturation, while the C terminus of NC is released (NC/SP2 cleavage) concurrently with the final step of Gag maturation, i.e., the C-terminal cleavage of CA (68). Thus, strictly controlled timing of the maturation process holds CA and NC/RNA in their immature conformations, and sequential release of SP domains controls transformation of all these components into the mature core.

Our data show the importance of the SP-like domain in the assembly of immature M-PMV particles and argue for the absence of a six-helix bundle structure in immature M-PMV. The possible reason for such strikingly different conformations of this crucial domain in HIV, RSV, and M-PMV remains unknown. The densely packed assemblage of M-PMV SP-like domain (49) may be favorable for stabilizing the CA-NC border region, which together with gRNA may serve as a scaffold for intracytoplasmic particle assembly and for protecting it from premature cleavage until it reaches the plasma membrane. This may be a contrast to the flexible six-helix bundle of HIV or RSV (48), for which assembly is facilitated by anchoring the Gag molecules in the plasma membrane.