key: cord-0729939-i267t23m
authors: Wang, Shui-Mei; Chang, Yu-Fen; Chen, Yi-Ming Arthur; Wang, Chin-Tien
title: Severe acute respiratory syndrome coronavirus nucleocapsid protein confers ability to efficiently produce virus-like particles when substituted for the human immunodeficiency virus nucleocapsid domain
date: 2008-07-01
journal: J Biomed Sci
DOI: 10.1007/s11373-008-9265-8
sha: 5943d40911e2da61381868d783fa205c57cd8de3
doc_id: 729939
cord_uid: i267t23m

We replaced the HIV-1 nucleocapsid (NC) domain with different N-coding sequences to test SARS-CoV nucleocapsid (N) self-interaction capacity, and determined the capabilities of each chimera to direct virus-like particle (VLP) assembly. Analysis results indicate that the replacement of NC with the carboxyl-terminal half of the SARS-CoV N resulted in the production of wild type (wt)-level virus-like particles (VLPs) with the density of a wt HIV-1 particle. When co-expressed with SARS-CoV N, chimeras containing the N carboxyl-terminal half sequence efficiently packaged N. However, the same was not true for the chimera bearing the N amino-terminal half sequence, despite its production of substantial amounts of VLPs. According to further analysis, HIV-1 NC replacement with N residues 2–213, 215–421, or 234–421 resulted in efficient VLP production at levels comparable to that of wt HIV-1, but replacement with residues 215–359, 302–421, 2–168, or 2–86 failed to restore VLP production to wild-type levels. The results suggest that the domain conferring the ability to direct VLP assembly and release in SARS-CoV N is largely contained between residues 168 and 421.

The novel infectious disease severe acute respiratory syndrome (SARS) [21] infected individuals in over 30 countries in 2002 and 2003 [50] . The causative agent was identified as the SARS coronavirus (SARS-CoV) [13, 21] . The complete SARS-CoV genome is approximately 30 kb long and contains 10-14 open reading frames [27, 36] . Like many other coronavirus species, the SARS-CoV is an enveloped virus containing three outer structural proteins: spike (S), membrane (M), and envelope (E) [6, 12, 22, 37] . We know that nucleocapsid (N) proteins associated with viral genomic RNA form a viral core within the envelope [29] ; however, SARS-CoV assembly and budding processes are not completely understood. The co-expression of coronavirus M with E [2, 17, 44] or N [19, 35] is sufficient for particle formation. According to these findings, the co-expression of at least two coronavirus structural proteins is required for virus-like particle (VLP) formation. In contrast, retrovirus assembly with polyprotein precursor gag gene expression is sufficient for producing VLPs [14] .

Retroviral gag apparently possesses all required signals for VLP assembly and budding, and functional domains involved in coronavirus VLP assembly and budding are allocated throughout the M, E, and N proteins. Despite being completely unrelated, the SARS-CoV N protein and HIV-1 NC have similar properties: both contain putative proteinprotein interaction domains and both play roles in viral RNA packaging [4, 5, 18, 32, 33, 53] . A domain responsible for Gag-Gag interactions (referred to as an I domain) has been mapped to HIV-1 NC [3, 7] . Likewise, recombinant SARS-CoV nucleocapsid proteins can undergo multimerization via self-association [16, 30, 42, 51] . Since heterologous polypeptides that form interprotein contacts permit efficient VLP production when placed in the HIV-1 NC region [1, 9, 20, 54] , we initially tested whether SARS-CoV N substitution for HIV-1 NC supports chimeric VLP assembly and release, and found that HIV-1 Gag mutants containing SARS-CoV N coding sequences as NC substitutes are capable of VLP assembly. This suggests that the HIV-1 NC assembly domain can be functionally replaced with the SARS-CoV N.

We employed this chimeric VLP assembly system to determine the boundaries of the SARS-CoV N coding sequence involved in conferring the ability to direct VLP assembly and release. Our results indicate that the carboxylterminal half of the SARS-CoV N protein functions in a manner similar to that of leucine zipper motifs in enabling efficient VLP production when substituted for HIV-1 NC.

The parental HIV-1 proviral plasmid DNA used in this study was HXB 2 [34] . The cDNA clone of the SARS-CoV N (SARS coronavirus strain TWC, GenBank accession number AY321118) was provided by the Centers for Disease Control of the Department of Health, Taiwan. A pair of upstream and downstream primers was used to amplify N-coding fragments via PCR-based overlap extension mutagenesis [38] . Primers used to construct the CoN-myc encoding the SARS-CoV N tagged with Myc at the C-terminus were oligonucleotides 5 0 -GCGTGGATCCATG TCTGATAATGGACCC-3 0 (forward) and 5 0 -AGTGTGTC GACCTGAGTTGAATC-3 0 (reverse). The PCR-amplified fragment was digested with BamHI and SalI and ligated into pcDNA3.1/myc-His A (Invitrogen).

To construct GST fusions, amplicons containing SARS-CoV N coding sequences were digested with BamHI and ClaI and fused to the C-terminus of GST, which is directed by a mammalian elongation factor 1a promoter [11] . Primers for cloning GST fusions were GST-CoN, 5 0 -TAAAGGATCCTC TGATAATGGACCC-3 0 (forward), 5 0 -TCATATCGATTTAT GCCTGAGTTGAATC-3 0 (reverse); GST-N1, 5 0 -TAAAGGA TCCTCTGATAATGGACCC-3 0 (forward), 5 0 -GCTCATCG ATTAGCTAGCCATTCGAGC-3 0 (reverse); GST-N2, 5 0 -TG GCTGGATCCGGTGGTGAAACTGCC-3 0 (forward), 5 0 -TC ATATCGATTTATGCCTGAGTTGAATC-3 0 (reverse).

HIV-1 NC deletion mutants DNC and DPC were generated via recombinations of HIV gag mutants containing a BamHI linker inserted at nt 1907, 1918, 1940, or 2076 , nt 1935-TTTAGGATCCAGGCT-2081, DPC, nt 1902-AATTCAGG-GATCCAGGCT-2081, and delNC, nt 1912-CCATAAGG ATCCAGGCT-2081. To replace the HIV-1 NC with SARS-Co N, SARS-CoV N cDNA fragments were PCR-amplified with BamHI-and/or ClaI-flanking primers. Amplicons were gel-purified, digested with BamHI and/or ClaI, and ligated into DNC and DPC or into an HIV-1 gag mutant carrying ClaI and BamHI linker insertions at nt 1920 and 2076, respectively. Primers used for making the designated constructs were: DNC(CoN) and PC(CoN), 5 0 -TAAAGGATCCTCTCTGAT AATGGACC-3 0 (forward), 5 0 -TAAAGGATCCCCTGAGTT GAATCAGC-3 0 (reverse); DNC(N2) and PC(N2), 5 0 -ATGC GGATCCTGCAGGGTGAAACTGCCCTCGC-3 0 (forward), 5 0 -TAAAGGATCCCCTGAGTTGAATCAGC-3 0 (reverse); NC(CoN), 5 0 -AGACATCGATCTGATAATGGACC-3 0 (forward), 5 0 -TAAAGGATCCCCTGAGTTGAATCAGC-3 0 (reverse); NC(N1), 5 0 -AGACATCGATCTGATAATGGAC C-3 0 (forward), 5 0 -TAAAGGATCCCCTGAGTTGAATCAG C-3 0 (reverse); NC(N2), 5 0 -GAATGGATCGATTAGGTGGT GAAACT-3 0 (forward), 5 0 -TAAAGGATCCCCTGAGTTGA ATCAGC-3 0 (reverse); NC(N3), 5 0 -GAATGGATCGATTAG GTGGTGAAACT-3 0 (forward), 5 0 -TGTTTGGATCCCGTC AATGTGCTTG-3 0 (reverse); NC(N4), 5 0 -AGACATCGATG CAAAGTTTCTGGTA-3 0 (forward), 5 0 -TAAAGGATCCCC TGAGTTGAATCAGC-3 0 (reverse); NC(N5), 5 0 -CGCAATC GATGGCCGCAAATTGCA-3 0 (forward), 5 0 -TAAAGGAT CCCCTGAGTTGAATCAGC-3 0 (reverse); NC(N6), 5 0 -AGA CATCGATCTGATAATGGACC-3 0 (forward), 5 0 -CCTTGG ATCCCTGTTGTTCCTTGAGG-3 0 (reverse); NC(N7), 5 0 -AG ACATCGATCTGATAATGGACC-3 0 (forward), 5 0 -GCTCA TCGATTAGCCAATTTGGTCAT-3 0 (reverse).

Plasmid DNAs containing wt (wtZiP) and mutant (KZiP) leucine zipper domains of human CREB [24] were kindly provided by E. Barklis [54] . cDNA fragments of the wt and mutant leucine zipper domains were PCR-amplified, digested, and ligated into DNC, yielding the respective constructs DNC(wtZiP) and DNC(KZiP). The myristylation-minus (Myr -) HIV-1 gag mutant [45] , (defective in membrane association) was used as a control for membrane binding assays. All of the engineered constructs were cloned into the PR-defective HIV-1 proviral expression vector HIVgptD25 [47] . Mutations were confirmed by restriction enzyme digestion or DNA sequencing.

The 293T cells were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal calf serum (GIBCO). Confluent 293T cells were trypsinized and split 1:10 onto 10-cm dishes 24 h prior to transfection. For each construct, cells were transfected with 20 lg of plasmid DNA using the calcium phosphate precipitation method; 50 lm chloroquine was added to enhance transfection efficiency. Unless otherwise indicated, 10 lg of each plasmid was used for co-transfection. Culture supernatant and cells were harvested for protein analysis 2-3 days post-transfection.

At 48-72 h post-transfection, supernatant from transfected 293T cells was collected, filtered, and centrifuged through 2 ml of 20% sucrose in TSE (10 mM Tris-HCl [pH 7.5], 100 mM NaCl, 1 mM EDTA plus 0.1 mM phenylmethylsulfonyl fluoride [PMSF]) at 4°C for 40 min at 274,000g. Pellets were suspended in IPB (20 mM Tris-HCl [pH 7.5], 150 mM NaCl, 1 mM EDTA, 0.1% SDS, 0.5% sodium deoxycholate, 1% Triton X-100, 0.02% sodium azide) plus 0.1 mM PMSF. Next, cells were rinsed with ice-cold phosphate-buffered saline (PBS), collected in IPB plus 0.1 mM PMSF, and microcentrifuged at 4°C for 15 min at 13,700g to remove cell debris. Supernatant and cell samples were mixed with equal volumes of 29 sample buffer (12.5 mM Tris-HCl [pH 6.8], 2% SDS, 20% glycerol, 0.25% bromphenol blue) and 5% b-mercaptoethanol and boiled for 5 min. Samples were resolved by electrophoresis on SDS-polyacrylamide gels and electroblotted onto nitrocellulose membranes. Membrane-bound Gag or chimeric proteins were immunodetected using a mouse monoclonal antibody directed against HIV-1 p24CA [46] . Myc-tagged SARS-CoV N were probed with anti-Myc (Invitrogen) monoclonal antibodies at a dilution of 1:5,000, respectively. The secondary antibody was a sheep antimouse or donkey anti-rabbit horseradish peroxidase-(HRP) conjugated antibody (Invitrogen) at 1:5,000 dilutions. To detect GST and GST fusions, anti-GST HRP conjugate (Amersham) was used at a 1:5,000 dilution. Horseradish peroxidase activity was detected according to the manufacturer's protocol.

The 293T cells either mock transfected or transfected with GST fusion expression vectors were collected, lysed in RIPA buffer (140 mM NaCl, 8 mM Na 2 HPO 4 , 2 mM NaH 2 PO 4 , 1% NP-40, 0.5% sodium deoxycholate, 0.05% SDS) containing complete protease inhibitor cocktail (Roche), and microcentrifuged at 4°C for 15 min at 13,700g (14,000 rpm) to remove cell debris. Aliquots of post-nuclear supernatant (PNS) were mixed with equal amounts of 29 sample buffer and 5% b-mercaptoethanol and held for Western blot analysis. RIPA buffer was added to the remaining PNS samples to final volumes of 500 ll. Each sample was mixed with glutathione agarose beads (30 ll) (Sigma) and rocked for 2 h at 4°C. Bead-bound complexes were pelleted, washed tree times with RIPA buffer, twice with PBS, eluted at 19 sample buffer with 5% b-mercaptoethanol, boiled for 5 min, and subjected to SDS-10% PAGE as described above.

Sucrose density gradient fractionation Supernatant cultures of transfected 293T cells were collected, filtered, and centrifuged through 2 ml 20% sucrose cushions as described above. Viral pellets were suspended in PBS buffer and laid on top of a pre-made 20-60% sucrose gradient consisting of 1 ml layers of 20%, 30%, 40%, 50% and 60% sucrose in TSE that had been allowed to sit for 2 h. Gradients were centrifuged in an SW50.1 rotor at 40,000 rpm (274,000g) for 16 h at 4°C; 500 ll fractions were collected from top to bottom. Sucrose density was measured for each fraction. Proteins in each fraction were precipitated with 10% trichloroacetic acid (TCA) and subjected to Western immunoblotting.

At 48 h post-transfection, 293T cells were rinsed twice, pelleted in PBS, and resuspended in TE buffer (10 mM Tris-HCl [pH 7.5], 1 mM EDTA) containing 10% sucrose and complete protease inhibitor cocktail. Cell suspensions were subjected to sonication followed by low-speed centrifugation. About 200 ll of post-nuclear supernatant (PNS) was mixed with 1.3 ml of 85.5% sucrose in TE buffer, placed at the bottom of a centrifuge tube, and covered with a layer of 7 ml 65% sucrose mixed with 3 ml of 10% sucrose in TE buffer. Gradients were centrifuged at 100,000g for 16-18 h at 4°C. Ten fractions were collected from top-to-bottom of each tube. Proteins in each fraction were precipitated with ice-cold 10% trichloroacetic acid (TCA), rinsed once with acetone, and analyzed by Western immunoblot.

Expression and assembly of chimeric proteins containing a SARS-CoV N coding sequence as a substitute for HIV-1 NC For purposes of roughly defining the self-association involvement of the SARS-CoV nucleocapsid (N) protein, we fused the complete length (GST-CoN), amino-terminal half (GST-N1), or carboxyl-terminal half (GST-N2) of the SARS-CoV N coding sequence to the GST carboxyl terminus; the expression of GST fusions was directed by a mammalian promoter (Fig. 1a) . Each resulting construct was co-expressed with SARS-CoV N (CoN-myc) in 293T cells. GST pull-down assays were used to determine the ability of each GST fusion to interact with SARS-CoV N.

Our results indicate that GST and GST-N1 were incapable of pulling down SARS-CoV N (Fig. 1b, lanes 6 and 8) , while GST-CoN and GST-N2 were capable of doing so (lanes 7 and 9). This suggests that the domain responsible for N-N interaction is largely contained within the carboxyl-terminal region-a finding compatible with those from previous reports [42, 43] .

Next, we tested whether the self-interaction capacity of SARS-CoV N was sufficient for rescuing the assembly defect of HIV-1 NC-deleted mutants. Full-length or carboxyl-terminal halves of SARS-CoV were inserted in frames into the HIV-1 NC-deleted mutants DNC, DPC, and delNC ( Fig. 2a) , yielding chimeras DNC(CoN), DNC(N2), PC(CoN), PC(N2), NC(CoN), and NC(N2). The DNC retained seven amino-terminal NC residues, and the DPCassociated deletions extended to four carboxyl-terminal SP1 residues. Two residues were removed from the SP1-NC junction of the delNC. Based on the possibility that the various NC mutations might impair VLP production to different degrees, we reasoned that the extent to which VLP assembly was restored following N sequence insertion might reveal the self-association capacity of each N sequence.

The backbone for all constructs was the PR-defective HIV-1 expression vector HIVgptD25, referred to in this study as a wild type (wt). Each mutant was transiently transfected into 293T cells and subjected to Western immunoblotting to determine its ability to assemble and release VLPs. As shown in Fig. 2b , chimeric proteins with predicted molecular weights corresponding to Pr55 gag containing a SARS-CoV N coding sequence insertion in the NC region were readily detected. Parental constructs with different extents of NC deletions were markedly defective in VLP production (Fig. 2b, lane 3 and Fig. 2c ); in particular, the DPC VLP assembly was almost abolished. Conversely, inserting the carboxyl-terminal half of SARS-CoV N dramatically restored VLP production to wt level (Fig. 2b, lanes 7-9) , suggesting that the carboxyl-terminal region of N that contains a putative self-association domain is capable of counteracting the assembly defect when placed in the HIV-1 NC region. Furthermore, we noted that DNC(CoN) (with over 400 inserted amino acid residues) was still capable of directing VLP assembly and budding (Fig. 2b, lane 4) . In contrast, PC(CoN) and NC(CoN) were moderately to markedly impaired in terms of VLP production (lanes 5 and 6). DNC(N2), PC(N2) and NC(N2) possessed medium to cell p24CA-associated protein ratios that were even higher than the wt ratio. This result could be due either to increased assembly rate or to increased intracellular mutant degradation. A slight correlation was observed between level of VLPs produced by chimeras containing full-length N and the number of HIV-1 NC residues remaining in the deleted region. Since NC(CoN) has almost all of its NC removed but retains intact SP1-NC and NC-SP2 junctures, we used NC(CoN) or NC(N2) as models for studying the effects of inserted SARS-CoV N on VLP assembly.

The above results suggest that HIV-1 Gag containing a large sequence of approximately 200-400 residues inserted into the NC region is still capable of assembling and releasing VLPs. Since the cultured 293T transfectant supernatant was centrifuged through 20% sucrose cushions for 40 min, we believe the recovered viral proteins in the pelleted medium are virus-associated. However, since the mutants contained large insertions, we performed sucrose density gradient fractionation experiments to determine whether the large insertion mutations had any effect on virus particle density. To make parallel comparisons with Boundaries of the RNA binding domain and self-association domain are indicated [41] . GST-CoN, GST-N1, and GST-N2 expressed GST fusion proteins with full-length, amino-terminal half, and carboxylterminal half of N fused to GST carboxyl terminus, respectively. (b, c) GST pull-down assay. 293T cells were co-transfected with 10 lg of CoN-myc and 10 lg of indicated GST fusion construct. Aliquots of cell lysates preceding (panel b) and following (panel c) GST pull-down were subjected to Western immunoblotting using anti-GST and anti-Myc antibodies as probes wt HIV particle density, NC(CoN) and NC(N2) viral pellets were spun with wt pellets through the same sucrose density gradient. The results show that both Pr55 gag and NC(N2) had peaks in fraction 6 with a density of 1.17 g/ml, consistent with wild-type HIV-1 particle density (Fig. 3 ). An NC(CoN) peak at fraction 7 with a density of approximately 1.19 g/ml suggests that the encoded NC(CoN) was tightly packed when assembled into VLPs.

Mapping the SARS-CoV N domain is required for replacing the HIV-1 NC assembly function Based on our observation that HIV-1 Gag containing SARS-CoV N coding sequences as NC substitutes was capable of VLP assembly, we tried to identify the SARS-CoV N regions responsible for conferring the assembly function. Toward this end we constructed a series of chimeras containing a variety of N coding regions as HIV-1 NC substitutes (Fig. 4a) ; the ability of each chimera to produce VLPs was determined by Western blot analysis. As shown in Fig. 4b , chimeras N1, N2 and N4 produced VLPs efficiently (lanes 4, 5, and 7). In contrast, the VLP quantities produced by chimeras N3, N5, N6, and N7 were smaller than those produced by the wt. Although VLP quantities produced by NC(N1) are markedly higher than those produced by DNC(CoN), NC(N5) and NC(N6), the NC(N1) release efficiency (43%) is relatively lower than the three's. This discrepancy may be due to different steady-state intracellular levels; relatively higher for NC(N1) but lower for DNC(CoN), NC(N5) and NC(N6) Fig. 4b, lanes 13, 15, 19 and 20 ). In addition, we cannot exclude the possibility of increased mutant VLP instability. Note that N1 was capable of efficient VLP production (Fig. 4b, lane 4) even though it could not effectively interact with the full-length N (Fig. 1b) . These results suggest that the N sequence, which is capable of conferring chimeric VLP assembly, is located between residues 168 and 421. Based on evidence showing that the leucine-zipper motif is competent in replacing the HIV-NC assembly function, we built a construct designated DNC(wtZiP) and containing a leucine-zipper motif sequence to serve as a control. We found that DNC(wtZiP) produces VLPs at a level comparable to that of wt (Fig. 4c, lane 4 vs. lane 2) . In contrast, DNC(KZiP) (which contains a NC replacement consisting of a mutated leucine-zipper motif) was incapable of rescuing the DNC assembly defect; VLP-associated chimeric proteins were barely detected (lane 5). Thus, the SARS-CoV N self-association domain is similar to leucinezipper motifs in conferring chimeric ability to direct VLP assembly by promoting protein-protein interactions.

Although SARS-CoV N is genetically unrelated to Gag, it is likely that N-Gag cross-interaction contributes to chimeric VLP assembly. To test this possibility, we expressed SARS-CoV N with wt or individual chimeras containing full lengths, carboxyl-terminal halves, or amino-terminal halves of SARS-CoV N. Our results indicate that VLP-associated SARS-CoV N was undetectable when co-expressed with Gag or NC(N1) (Fig. 4d, lanes 3  and 5) , suggesting that no significant N-Gag interactions occurred and that interaction strength within the SARS-CoV N amino-terminal region is insufficient for NC(N1) to incorporate N into VLPs. Conversely, we readily detected VLP-associated N co-expressed with NC(N2) (lane 6). These results strongly suggest that SARS-CoV N incorporation into VLPs largely depends on intermolecular interactions within the carboxyl-terminal SARS-CoV N, which agrees with our GST pull-down assay results (Fig. 1b) .

Our results also indicate that the N1 sequence as a NC substitute is capable of conferring efficient chimeric VLP assembly (Fig. 4b, d) , suggesting that amino-terminal SARS-CoV N may also possess a potent self-interaction capacity. We therefore reasoned that NC(N1) could efficiently package GST-N1 via N1-N1 interactions if a strong N1 sequence self-association does exist. To test this idea, we co-expressed NC(N1) with GST-CoN, GST-N1, or GSTN2 and used isolated viral pellets for Western immunoblot analysis. Our results indicate that NC(N1) was incapable of packaging GST-CoN, GST-N1, or GST-N2that is, all of the GST fusion signals were barely detected in medium (Fig. 5a) . In contrast, NC(N2) can efficiently package GST-N2 and GST-CoN (Fig. 5b, lanes 4 and 6) , which is compatible with the proposal that the SARS-CoV N self-interaction domain is largely located in the carboxyl-terminal region. According to these data, any N1-N1 interaction strength is insignificant.

Membrane association is required for HIV-1 Gag assembly and budding [8, 31] . Results from a previous study show that HIV-1 NC facilitates Pr55 gag membrane association [39] . Furthermore, most Gag mutants (either multimerization-defective or assembly-defective) demonstrate markedly reduced membrane binding capacities [9, 15, 23] . We performed membrane flotation experiments to test whether the replacement of HIV-1 NC by SARS-CoV N exerts any effect on membrane binding, and to determine if the assembly defect is correlated with reduced membranebinding capacity. Our results indicate that with the exception of DNC(wtZiP), all of the chimeras expressed moderate reductions in membrane binding capacity (Fig. 6) . DNC(wtZiP) showed a membrane binding capacity comparable to wt and exhibited a positive correlation with VLP assembly competence.

Neither SARS-CoV N nor non-myristylated (Myr -) Gag were capable of efficient cell membrane association-that Percentages of p24 gag -associated protein in each fraction were plotted against the fraction's sucrose density is, less than 10% of total expressed protein was membranebound (Fig. 6b) . However, the assembly-competent NC(N2) and NC(N1) and the assembly-defective DNC and NC(N3) displayed similar levels of membrane-binding capacity (50-60% of total membrane-associated protein). Even DNC(KZiP), which is severely defective in VLP assembly, had an encoded protein membrane-associated level of over 40%. The results suggest that NC is not required for efficient Gag membrane association.

Results from several recent SARS-CoV N protein characterization studies suggest the presence of a putative selfinteraction domain within the carboxyl-terminal region [10, 26, 42, 43, 51] . These studies involve in vitro assays using purified recombinant N proteins, co-immunoprecipitation, or hybrid yeast systems. For the present study we established a chimeric VLP assembly process to analyze SARS-CoV N self-interaction capacity. Our evidence-that chimeras containing the SARS-CoV N protein coding sequence as a substitute for HIV-1 NC are capable of producing substantial amounts of VLPs-supports the idea that retroviral gag is capable of tolerating major mutations [48, 49, 54] . The inability of the PC(CoN) chimera to produce VLPs cannot be explained solely by an assembly limitation of extra-large proteins-as is the case with DNC(CoN), which contains the same full-length N yet still produces significant amounts of VLPs.

One possible explanation for this discrepancy is that DNC(CoN) retains the N-terminal region of the I domain, while PC(CoN) lacks the entire I domain and parts of SP1. Researchers who have studied the roles of I and SP1 domains in virus assembly have observed that some mutations in either one disrupt Gag multimerization, thus reducing VLP production [9, 15, 23, 28, 40, 54] . However, we found that PC(N2) produced VLPs at an efficiency level comparable to that of wt (Fig. 2b , lane 8 vs. lane 2). Furthermore, VLPs produced by NC(N2) possess a particle density similar to that of wt HIV-1, indicating that chimeric proteins are packed in a manner similar to that of wt. Combined, these results suggest that SARS-CoV N residues 215-421 confer a potent self-interaction capacity capable of rescuing the assembly defects of NC-deleted mutants. Although the transformation of a severe assembly-defective DPC into an assembly-competent PC(N2) provides compelling evidence that SARS-CoV N possesses a self-interaction domain capable of replacing the HIV-1 NC assembly function, we chose the delNC instead of DPC as a backbone for our SARS-CoV N fragment analysis. Our main reason was that all of the DPC-derived constructs lacked SP1, and therefore had the potential to disrupt an accurate assessment of SARS-CV N sequences involved in VLP assembly.

According to previous in vitro biochemical and biophysical studies, a variety of recombinant SARS-CoV N proteins consisting of the carboxyl-terminal residues 213-422 [42] , 283-422 [25] , 284-422 [51] , 169-422 [43] , or 248-365 [10] are capable of self-associating as either dimers or multimers. This suggests that N residues 248-365 may constitute a minimum required sequence for N self-association. Both of our inserted NC(N2) and NC(N4) sequences containing residues 248-365 were capable of driving efficient VLP production, perhaps by promoting chimeric protein multimerization via the inserted sequence. The inability of NC(N3) to efficiently direct VLP assembly may be due in part to the exclusion of residues 360-365 in the inserted N sequence. According to one previous report, SARS-CoV N residues 343-402 are necessary and sufficient for oligomerizing the carboxyl-terminal region [26] . However, we have consistently observed that NC(N5) cannot direct VLP assembly as efficiently as NC(N2) or NC(N4), suggesting that the region between residues 234 and 302 is important for driving efficient VLP assembly and release.

Results from a crystal structural study of the SARS-CoV N carboxyl-terminal region imply extensive N-N interaction involving residues 270-370 [52] . Accordingly, the ability of chimeras containing a variety of N sequences to act as a HIV-1 NC replacement for directing VLP assembly is compatible with studies indicating that inserted N sequences can promote efficient VLP assembly, providing that such sequences retain most N residues between 270 and 370.

However, it is unlikely that the ability of an inserted N sequence to confer chimeric VLP assembly reflects its capacity for in vitro self-association. NC(N1) (containing N residues 2-213) is capable of producing substantial amounts of VLPs (Fig. 4b) , despite being defective in packaging the co-expressed N protein (Fig. 4d) . Also note that GST-N1 was incapable of efficiently pulling down the N protein (Fig. 1b) or of being packaged into NC(N1) VLPs (Fig. 5) . One possible explanation for this discrepancy is that the N1 sequence may contain a self-association domain that fails to confer sufficient interaction strength for self-association or associating with a full-length N. It has previously been demonstrated that a serine/argininerich motif between N residues 184 and 196 is crucial for N protein self-multimerization [16] . Since CA-SP1 is largely responsible for driving Gag or chimeric protein self-association, protein sequences with small self-interaction capacities may be sufficient for restoring delNC VLP assembly to wt levels when substituted for NC. If true, it is likely that the N serine/arginine-rich motif in the context of NC(N1) makes a significant contribution to VLP assembly. In agreement with this hypothesis, NC(N6) and NC(N7) (both lacking residues 184-196) were found to trigger moderate-to-marked decreases in VLP production compared to NC(N1), suggesting that even though the N selfinteraction domain primarily resides in the carboxylterminal region, part of the amino-terminal N may be involved in N self-association.

Inserting the mutated leucine-zipper domain into the HIV-1 NC deleted region abolished VLP production, suggesting that the inserted sequence may have impaired Gag-Gag interaction to the extent that proper Gag multimerization could not proceed, even though a major CA dimerization domain was retained in the chimera. The N-containing chimeras were not equally effective in rescuing the NC assembly defect. However, the inserted SARS-CoV N sequences did not trigger further restrictions on Gag-Gag interaction-that is, low but detectable amounts of VLPs were observed in most of the chimeras.

Efficient particle production by minimal Gag constructs which retain the carboxyterminal domain of human immunodeficiency virus type 1 capsid-p2 and a late assembly domain

Coronavirus pseudoparticles formed with recombinant M and E proteins induce alpha interferon synthesis by leukocytes

Functional chimeras of the Rous sarcoma virus and human immunodeficiency virus gag proteins

Specific binding of human immunodeficiency virus type 1 gag polyprotein and nucleocapsid protein to viral RNAs detected by RNA mobility shift assays

Retroviral nucleocapsid domains mediate the specific recognition of genomic viral RNAs by chimeric Gag polyproteins during RNA packaging in vivo

The production of recombinant infectious DI-particles of a murine coronavirus in the absence of helper virus

Importance of basic residues in the nucleocapsid sequence for retrovirus Gag assembly and complementation rescue

Myristoylation-dependent replication and assembly of human immunodeficiency virus 1

Human immunodeficiency virus type 1 Gag polyprotein multimerization requires the nucleocapsid domain and RNA and is promoted by the capsid-dimer interface and the basic region of matrix protein

Modular organization of SARS coronavirus nucleocapsid protein

In vitro V(D)J recombination: signal joint formation

Coronavirus particle assembly: primary structure requirements of the membrane protein

Identification of a novel coronavirus in patients with severe acute respiratory syndrome

Assembly and release of HIV-1 precursor Pr55gag virus-like particles from recombinant baculovirus-infected insect cells

Mutation of the SP1 sequence impairs both multimerization and membrane-binding activities of human immunodeficiency virus type 1 Gag

Analysis of multimerization of the SARS coronavirus nucleocapsid protein

Assembly of severe acute respiratory syndrome coronavirus RNA packaging signal into virus-like particles is nucleocapsid dependent

Structure of the N-terminal RNA-binding domain of the SARS CoV nucleocapsid protein

Generation of synthetic severe acute respiratory syndrome coronavirus pseudoparticles: implications for assembly and vaccine production

Nucleic acid-independent retrovirus assembly can be driven by dimerization

A novel coronavirus associated with severe acute respiratory syndrome

The molecular biology of coronaviruses

Characterization of a putative alpha-helix across the capsid-SP1 boundary that is critical for the multimerization of human immunodeficiency virus type 1 gag

Engineered leucine zippers show that hemiphosphorylated CREB complexes are transcriptionally active

SR-rich motif plays a pivotal role in recombinant SARS coronavirus nucleocapsid protein multimerization

Carboxyl terminus of severe acute respiratory syndrome coronavirus nucleocapsid protein: self-association analysis and nucleic acid binding characterization

Roles of matrix, p2, and N-terminal myristoylation in human immunodeficiency virus type 1 Gag assembly

Characterization of the coronavirus M protein and nucleocapsid interaction in infected cells

Characterization of N protein self-association in coronavirus ribonucleoprotein complexes

Myristylation of gag proteins of HIV-1 plays an important role in virus assembly

Charged amino acid residues of human immunodeficiency virus type 1 nucleocapsid p7 protein involved in RNA packaging and infectivity

Nucleocapsid and matrix protein contributions to selective human immunodeficiency virus type 1 genomic RNA packaging

Complete nucleotide sequence of the AIDS virus, HTLV-III

The transmissible gastroenteritis coronavirus contains a spherical core shell consisting of M and N proteins

Characterization of a novel coronavirus associated with severe acute respiratory syndrome

Structural maturation of the transmissible gastroenteritis coronavirus

Molecular cloning: a laboratory manual

The I domain is required for efficient plasma membrane binding of human immunodeficiency virus type 1 Pr55Gag

Mapping and characterization of the N-terminal I domain of human immunodeficiency virus type 1 Pr55Gag

The SARS-CoV nucleocapsid protein: a protein with multifarious activities

The nucleocapsid protein of the SARS coronavirus is capable of selfassociation through a C-terminal 209 amino acid interaction domain

Biochemical and immunological studies of nucleocapsid proteins of severe acute respiratory syndrome and 229E human coronaviruses

Nucleocapsid-independent assembly of coronavirus-like particles by co-expression of viral envelope protein genes

Assembly, processing, and infectivity of human immunodeficiency virus type 1 gag mutants

Analysis of minimal human immunodeficiency virus type 1 Gag coding sequences capable of virus-like particle assembly and release

Assembly and processing of human immunodeficiency virus Gag mutants containing a partial replacement of the matrix domain by the viral protease domain

Characterization of a small (25-kilodalton) derivative of the Rous sarcoma virus Gag protein competent for particle release

Incorporation of chimeric gag protein into retroviral particles

World health report 2004-changing history

Recombinant severe acute respiratory syndrome (SARS) coronavirus nucleocapsid protein forms a dimer through its C-terminal domain

Crystal structure of the severe acute respiratory syndrome (SARS) coronavirus nucleocapsid protein dimerization domain reveals evolutionary linkage between corona-and arteriviridae

Effects of nucleocapsid mutations on human immunodeficiency virus assembly and RNA encapsidation

Analysis of the assembly function of the human immunodeficiency virus type 1 Gag protein nucleocapsid domain

Acknowledgments We thank C.-Y. Chang and Y.-P. Li for reagents and technical assistance, and E. Barklis for plasmids coding the leucine-zipper domain. This work was supported by grants VGH94-314 from the Taipei Veterans General Hospital and NSC94-2320-B-010-035 from the National Science Council, Taiwan, Republic of China.