SARS-CoV envelope protein palmitoylation or nucleocapid association is not required for promoting virus-like particle production

Background Coronavirus membrane (M) proteins are capable of interacting with nucleocapsid (N) and envelope (E) proteins. Severe acute respiratory syndrome coronavirus (SARS-CoV) M co-expression with either N or E is sufficient for producing virus-like particles (VLPs), although at a lower level compared to M, N and E co-expression. Whether E can release from cells or E/N interaction exists so as to contribute to enhanced VLP production is unknown. It also remains to be determined whether E palmitoylation or disulfide bond formation plays a role in SARS-CoV virus assembly. Results SARS-CoV N is released from cells through an association with E protein-containing vesicles. Further analysis suggests that domains involved in E/N interaction are largely located in both carboxyl-terminal regions. Changing all three E cysteine residues to alanines did not exert negative effects on E release, E association with N, or E enhancement of VLP production, suggesting that E palmitoylation modification or disulfide bond formation is not required for SARS-CoV virus assembly. We found that removal of the last E carboxyl-terminal residue markedly affected E release, N association, and VLP incorporation, but did not significantly compromise the contribution of E to efficient VLP production. Conclusions The independence of the SARS-CoV E enhancement effect on VLP production from its viral packaging capacity suggests a distinct SARS-CoV E role in virus assembly.


Background
Coronaviruses are enveloped viruses with 27-32 kb single-strand positive-sense RNA genomes encoding four structural proteins: nucleocapsid (N), spike (S), membrane (M) and envelope (E) [1,2]. Translated on free polysomes, highly basic N interacts with newly synthesized viral genomic RNA to form helical nucleocapsids [3,4]. The M, S and E viral membrane proteins are translated on membrane-bound polysomes, inserted into the endoplasmic reticulum (ER), and transported to the ER-Golgi intermediate compartment (ERGIC), where E and M interact and trigger budding [5,6]. N and S are incorporated into virions via interaction with M, with virions accumulating in large, smooth-walled vesicles that eventually fuse with the plasma membrane and release virions from cells [2,[7][8][9][10][11].
Coronavirus E is a small integral membrane protein consisting of approximately 76 to 109 amino acids and containing a hydrophobic domain. Several researchers have suggested that coronavirus E functions as an ion channel [12,13]. The role of the coronavirus E ion channel in the virus life cycle is not completely clear. The addition of hexamethylene amiloride (HMA, an ion channel inhibitor of mouse hepatitis coronavirus [MHV] and human coronavirus 229E [HCoV229E] ion channel activity in vitro) to culture medium blocks MHV and HCoV229E replication [12], suggesting that the coronavirus E ion channel plays a role in virus replication. Two or three cysteine residues are located on the carboxyl side of the hydrophobic domain in coronavirus E proteins, with some serving as targets for palmitoylation [14][15][16][17], which may contribute to virus assembly in the MHV. E plus M has been shown to be sufficient for VLP formation in the MHV [18], transmissible gastroenteritis virus (TGEV) [19], bovine coronavirus (BCoV) [19], infectious bronchitis virus (IBV) [5], and SARS-CoV [20]. Although M and N co-expression is also sufficient for SARS-CoV VLP production [21,22], VLP yields are further enhanced by E co-expression [23]. According to these data, SARS-CoV E plays a supporting role in promoting virus assembly and/or budding. E and N are thought to be packaged into VLPs through separate and individual associations with M; it remains unknown whether E/N interaction exists, which might contribute to enhanced virion production. MHV and IBV E proteins are capable of release from cells as vesicles [5,24], implying a relationship between E release and its contribution to virus production. Our goal for this study was to determine whether SARS-CoV N is capable of interacting with E and cell release via an association with Econtaining vesicles. Our results indicate no correlation between SARS-CoV E capacity to release or interact with N, or its ability to promote VLP production.

Plasmid construction
Mammalian expression vectors encoding SARS-CoV M, N, S and E were provided by G. J. Nabel [21]. A pair of upstream and downstream primers was used to amplify Ecoding fragments via PCR-based overlap extension mutagenesis [25]. Primers used to introduce an HA or FLAG epitope tag to the E amino or carboxyl terminus are 5′-TTCTGCGATATCGCCACCATGTACCCATACGAC GTGCCTGACTACGCCTACAGCTTCGTGAGCG-3′ (containing a flanking EcoRV restriction site and HA tagcoding nucleotides) and 5′-GCGGATCCTCACTTGTC GTCGTCGTCCTTGTAGTCGCCCACCAGCAGGTCG GGCAC-3′ (containing a flanking BamHI restriction site and FLAG tag-coding nucleotides). The forward primer is 5′-GTCTGAGCAGTACTCGTTGCTG-3 (referred to as the N primer) and reverse primer 5′-GGAAAGGA CAGTGGGAGTGGCAC-3′ (referred to as the anti-N primer). Primers used to construct designated E mutants were EC3A, 5′-CTGAGGCTGGCCGCATATGCCGCG AACATCGTGAACGTGAGC-3′ (reverse); 74LL/AA,5′-GCAGATCTGGATCCTAGTTCACACGGCCGCGTCG GGCACGCC-3′ (reverse); EΔ76V, 5′-CAGATCTGGA TCCTAGTTCACAGCAGGTCGGGCAC-3′ (reverse). The N primer serves as a forward primer while the anti-N primer was used as a reverse primer for the second round PCR amplification. Purified PCR product was digested with BamHI and EcoRV and ligated into the SARS-CoV E expression vector. When constructing N-DsRed, the N primer served as the forward primer, using the SARS-CoV N expression vector as a template and 5′-GCGGATCCTGGGTGCTGTCGGCGCTG-3′ as the reverse primer. Amplified PCR product was digested with BamHI and SalI and ligated into pDsRed-Monomer-N1 (Clontech). The cloned N-DsRed was digested with NheI and BamHI and ligated into pEGFP-N1 (Clonetech), yielding construct N-EGFP.
GST-M was constructed by digesting M expression vector with EcoRV and NotI, ligated into pCDNA3.1myc-HisA (Invitrogen). The resultant construct was then digested with BamHI and Not1, and then the M coding sequence fragment was fused to carboxyl terminus of GST, which is directed by a mammalian elongation factor Ia promoter [26]. GST-N fusions, as described previously [27], were constructed by fusing SARS-CoV N coding sequences to the carboxyl terminus of GST.
Cell culture and transfection 293 T and HeLa cells were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal calf serum (GIBCO). Confluent cells were trypsinized and split 1:10 onto 10 cm dishes 24 h prior to transfection. For each construct, cells were transfected with 20 μg of plasmid DNA using the calcium phosphate precipitation method; 50 μm chloroquine was added to enhance transfection efficiency. Unless otherwise indicated, 10 μg of each plasmid was used for co-transfection. Culture supernatant and cells were harvested for protein analysis 2-3 d post-transfection. For HeLa transfection, plasmid DNA was mixed with GenCarrier (Epoch Biolabs) at a ratio of 1 μg to 1 μl; the transfection procedure was performed according to the manufacturer's protocols.

Laser scanning immunofluorescence microscopy
HeLa cells were split 1:80 onto coverslips 24 h before transfection. Between 18 and 24 h post-transfection, , cells were washed with PBS and permeabilized at room temperature for 10 min in PBS plus 0.1% Triton X-100 following fixation at 4°C for 20 min with methanol/acetone (1:1). Samples were incubated with the primary antibody for 1 h and with the secondary antibody for 30 min. Following each incubation, samples were subjected to three washes (5 to 10 min each) with DMEM/calf serum. Primary antibody concentrations were anti-HA at a dilution of 1:500. A rabbit anti-mouse rhodamine-conjugated antibody at a 1:100 dilution served as the secondary antibody (Cappel, ICN Pharmaceuticals, Aurora, OH). After a final DMEM/calf serum wash, the coverslips were washed three times with PBS and mounted in 50% glycerol in PBS for viewing. Images were analyzed and photographs taken using the inverted laser Zeiss.

Iodixanol density gradient fractionation
Supernatants from transfected 293 T 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 10-40% iodixanol (OptiPrep) gradient consisting of 1.25 ml layers of 10, 20, 30 and 40% iodixanol solution prepared according to the manufacturer's instructions (Axis-Shield, Norway). Gradients were centrifuged in a SW50.1 rotor at 40,000 rpm for 16 h at 4°C; 500 μl fractions were collected from top to bottom and densities were measured for each. Proteins in each fraction were precipitated with 10% trichloroacetic acid (TCA) and subjected to Western immunoblotting.

GST pull-down assay
GST pull-down protocols were as previously described [27]. Briefly, 500 μl of PNS containing complete protease inhibitor cocktail was mixed with 30 μl of glutathione agarose beads (Sigma). All reactions took place at 4°C overnight on a rocking mixer. Immunoprecipitateassociated resin or bead-bound complexes were pelleted, washed tree times with lysis buffer, two times with PBS, eluted with 1X sample buffer, and subjected to SDS-10% PAGE as described above.

SARS-CoV E is capable of associating with N
In a previous study we found that SARS-CoV M coexpression with N leads to VLP formation, and that a FLAG tagged at the SARS-CoV M carboxyl terminus (M-FLAG) significantly inhibits N packaging into VLPs [22]. Since N release into medium depends on an association with M, and since E may promote VLP production, we set out to confirm whether E complements M-FLAG in N packaging into VLPs. Toward this goal we coexpressed N with M-FLAG in the absence or presence of SARS-CoV E tagged with either HA at the amino terminus (HA-E) or with FLAG at the carboxyl terminus (E-FLAG). As shown in Figure 1, N that was barely detectable in medium became readily detected following co-expression with HA-E (lanes 9 versus 10). Since N is incapable of associating with M-FLAG, a simple explanation would be the ability of E to promote N incorporation into VLPs via E/N interaction. E-FLAG was barely HA-E denotes an HA tagged at the SARS-CoV E amino terminus. DNA (5 μg) of each plasmid was used for each transfection, with pBlueScript SK added to maintain a plasmid DNA level of 20 μg. At 48 h post-transfection, supernatants and cells were collected and prepared for protein analysis as described in Materials and Methods. Cell lysate samples (lanes 1 to 5) corresponding to 5% of total, and medium pellet samples (lanes 6 to 10) corresponding to 50% of total, were fractionated by 10% SDS-PAGE and electroblotted onto nitrocellulose filters. SARS-CoV M was probed with rabbit antiserum. SARS-CoV N and E were detected with a mouse anti-N, anti-HA, or anti-FLAG monoclonal antibody. N proteins from medium or cell samples were quantified by scanning N band densities from immunoblots. Rations of N level in media to those in cells were determined for each samples and normalized to that of samples without HA-E or E-FLAG co-expression. detected in medium (Figure 1, lane 7), suggesting that the FLAG tagged at the E carboxyl terminus affected E release or E association with M and/or N. To confirm that E-mediated N release is not due to HA tag, we coexpressed N with E and found that E without HA tag can still promote N release (data not shown), suggesting that HA tag exerts no major impacts on E-mediated N release.
Since SARS-CoV E and M are both capable of release into medium, N detected in medium may be a result of an association with released M-FLAG in the presence of E, but not a result of direct E/N interaction. To rule out this possibility and to confirm that a specific N-E interaction exists, we co-expressed N and either HA-E, E-FLAG, M or S. As expected, N was readily detected when co-expressed with HA-E or with M, but barely detectable in culture medium when expressed alone or coexpressed with S ( Figure 2A, lanes 5 and 8) or E-FLAG (which was also barely detectable in medium) ( Figure 2B, lane 3). These data support our hypothesis that SARS-CoV N is capable of release into medium via association with E, and suggests that FLAG tagged at the SARS-CoV E carboxyl terminus significantly reduces E release capacity in addition to impairing E incorporation into VLPs.
To confirm that N release is associated with E in pelletable particle form, we subjected culture medium from cells expressing E, E plus N, or E plus N and M to iodixanol density gradient fractionation, and found that E primarily sedimented with co-expressed N at a slightly lower density fraction compared to E expression alone ( Figure 2C). Particles formed by E or E plus N exhibited an iodixanol density of 1.13 to 1.16 g/ml, whereas VLPs formed by E, N and M displayed densities ranging from 1.11 to 1.13 g/ml. Combined, these data support the idea that N released into medium is associated with E vesicles when co-expressed with E.
RNA is not required for efficient E/N interaction SARS-CoV N contains an RNA binding domain [30], and results from our previous study indicate that RNA can enhance the N-N interaction [22]. Here our goal was to determine whether the presence of RNA is required for E/N interaction. HA-E or N was co-expressed with GST tagged at the N amino terminus (GST-N). E or N association with GST-N was assessed using a GST pull-down assay in the presence or absence of RNase. GST by itself was not capable of pulling down HA-E or N ( Figure 3A, lanes 5-6). RNase treatment significantly reduced levels of co-pulled-down N ( Figure 3A, lane 9), which is consistent with past results [22]. However, the same RNase condition did not reduce the amount of copulled-down HA-E, suggesting that RNA is not required for E/N interaction. We frequently observed slight increases in pulled-down HA-E following treatment with RNase ( Figure 3A, lane 10 vs. lane 8). Since RNA binds to N readily, it is likely that RNA-bound N molecules may associate less efficiently with E.

E association domain is largely located in the N carboxyl-terminal region
To map the N sequences involved in E association, we constructed a set of GST-N fusions containing multiple N coding sequences fused to the GST carboxyl terminus ( Figure 3B). Each GST fusion construct was transiently co-expressed with E, and culture supernatants and cell lysates were analyzed by Western immunoblotting. As shown in Figure 3C, all GST fusions except for GST-N7 (containing N residues 2 to 86, lane 9) were released into medium when co-expressed with E. None of the GST fusions were readily detected in medium without coexpressed E (data not shown), suggesting that the detected GST-N fusions were released into medium via E association, and that the E binding domains are largely located in the N carboxyl-terminal region (likely involving amino acid residues 87 to 421). Results from a GST pull-down assay ( Figure 3D) provide further support for this conclusion.
Cysteine residues are not required for E release, E/N interaction, or E enhancement of VLP production SARS-CoV and MHV E proteins are post-translationally modified with palmitic acid. Lack of palmitoylation modification can markedly decrease SARS-CoV E membrane association [16] and MHV E stability [17]. Substitution mutations for cysteines-the palmitoylation targets in MHV E can reduce virus yields, implying their importance for MHV virus production [14,17]. Accordingly, we tried to determine whether equivalent cysteines play a role in SARS-CoV E/N interaction and VLP assembly by constructing an E mutant (designated EC3A) with alanine replacements for SARS-CoV E cysteine residues C40, C43 and C44. Supernatant and cell lysate samples containing HA-E were analyzed under non-reducing conditions to confirm the presence of disulfide bonds between SARS-CoV E molecules. Results indicate the presence of an 18 kDa band (equivalent to an HA-E dimer) ( Figure 4A, lanes 2 and 11, arrowheads), which agrees with a previous report [31]. As expected, only the E monomer was detected in EC3A-containing samples under non-reducing conditions ( Figure 4A, lanes 3 and 12).
While most of intracellular E retains its monomeric form, most released E was detected in dimeric form ( Figure 4A, lane 2 vs. lane 11), suggesting that released SARS-CoV E proteins are largely linked by disulfide bonds. Interestingly, EC3A was still efficiently released into medium and apparently capable of promoting N release ( Figure 4A, lanes 14  and 18). The relatively low level of medium HA-E observed in Figure 4A is due to low expression level (top panels, lanes 17 vs. 8). Nevertheless, our data indicate that EC3A produced VLPs as efficiently as wt when co-expressed with both M and N ( Figure 4B, lanes 10-12), suggesting that SARS-CoV E cysteines are not involved in E release, E/N interactions, or VLP assembly. EC3A exhibited greater release efficiency compared to wt, implying that SARS-CoV E cysteines are involved in the E trafficking process. However, immunofluorescence results indicate that the EC3A subcellular distribution pattern was similar to that of the wt. Removal of last amino acid residue from the E carboxyl-terminal tail significantly affects E release and E/N interaction E-FLAG was released at a lower efficiency level compared to HA-E, suggesting that the carboxyl-terminal tail domain may be involved in the E release process. We hypothesized that the E-carboxyl tail is also involved in E/N interaction. To test this idea, we constructed two E mutants: EΔ76V (with the final carboxyl-terminal residue valine removed) and E74LL/AA (with the carboxylterminal dileucine motif 74-LL-75 replaced with alanines). According to our results, (a) the E74LL/AA mutant was capable of release, and (b) a positive correlation exists between the release levels of co-expressed N and E74LL/AA ( Figure 5A, lanes 10-15). In addition, E74LL/AA was capable of producing VLPs as efficiently as wt when co-expressed with M and N ( Figure 5A,  lanes 16 vs. 17 and 18). These results suggest that 74-LL-75 was not functionally involved in E release, E/N interaction, or the promotion of VLP production. Conversely, EΔ76A was not capable of efficiently promoting N release, and exhibited a reduced release capacity ( Figure 5B, lanes 12 to 15). GST pull-down assay data suggest that the EΔ76V mutation markedly affected E/N interaction, but exerted no major impacts on N/M association ( Figure 5C, lanes 16 vs. 18). The level of GST-Nassociated E was significantly lower than that of GST-Massociated E, suggesting that E/N association is not as strong as E/M association ( Figure 5C, lanes 15 vs. 17). While pull-down assay results suggest an association between EΔ76V and M, EΔ76V release was not significantly enhanced following M co-expression ( Figure 5C, lane 18 vs. Figure 5B, lanes 17 and 20). A possible explanation is steric hindrance during VLP assembly. Note that EΔ76V was still capable of promoting VLP production as efficiently as the wt, despite defective packaging into VLPs ( Figure 5B, lane 20). This suggests that the SARS-CoV E contribution to VLP production is independent of its ability to be packaged into VLPs.

None of the mutations significantly affected SARS-CoV E subcellular localization
To determine whether mutations affecting E subcellular localization also partly accounted for altered release efficiency, we co-expressed each mutant with a Golgi labeling marker (pECFP-Golgi). According to a confocal microscopy analysis, wt and the EC3A, E74LL/AA, and EΔ76V mutants largely localized in the Golgi area ( Figure 6A), suggesting that none of the mutations significantly affected E subcellular localization. When co-expressed with EGFP-tagged N, a fraction of N co-localized with EΔ76V in the perinuclear area; this staining pattern was barely distinguishable from that of cells co-expressing HA-E and N-EGFP ( Figure 6B). This implies that even though EΔ76V may associate with N to a certain extent, the association is insufficient for enabling EΔ76V co-precipitation with GST-N ( Figure 5C).

Discussion
To our knowledge, this is the first report of interaction between SARS-CoV E and N proteins. Both SARS-CoV spike (S) and E proteins can be released into medium; however, unlike E, S cannot promote the release of coexpressed N (Figure 2A). This suggests that N is incapable of associating with S, and supports the idea of specific E/N interaction. While RNA can enhance N/N interaction, the presence of RNA is not necessary for E/ N interaction. Genetic analyses suggest that E binding domains are largely located in the N carboxyl-terminal region (Figure 3). SARS-CoV M or E by itself can secrete into medium as vesicles but not virus-like particles (VLPs) which can be formed by M plus N, M plus E, E plus N or M plus N and E. It is likely that there may be a combination of these different VLPs in culture medium when cells are co-transfected with M plus E and N. The iodixanol density gradient fractionation analyses suggest that released E vesicles and E-N VLPs exhibit slightly higher densities compared to those of M-E-N VLPs ( Figure 2C). Since M is the major viral component, the strong presence of M molecules may exert a spatial effect that explains, at least in part, the lower density for M-E-N particles. We consistently observed marked E-N VLP yield enhancement following M co-expression. Since M  is the most abundant viral structural protein capable of recruiting both E and N into VLPs, we do not consider this a surprising result. In addition, M exhibits a noticeably higher N binding capacity than E ( Figure 5C). VLPs formed by M, E and N look more morphologically homogeneous than E-N VLPs, possibly due to additional intermolecular interactions between M/M, M/E and M/N. SARS-CoV E is capable of undergoing oligomerization triggered by both hydrophobic interaction and interchain disulfide bond formation between cysteine residues [16,31]. All three SARS-CoV E cysteine residues have been shown to be post-translationally modified by palmitoylation [16]. Blocking MHV E palmitoylation results in significantly impaired VLP assembly, suggesting that palmitoylated E proteins are essential for murine coronavirus assembly [14]. In the present study we found that changing all three cysteines into alanines (EC3A) did not exert negative effects on SARS-CoV E protein release or VLP assembly (Figure 4). This suggests that E palmitoylation or interchain disulfide bond formation is not required for SARS-CoV E protein release or VLP assembly. Furthermore, we consistently detected higher levels of EC3A compared to wt E in medium, suggesting that EC3A is released more efficiently than wt E. One possible explanation is that post-translational palmitoylation or interchain disulfide bond formation restricts E protein secretion.
Despite a higher release level compared to wt E, EC3A did not produce higher VLP yields than wt E following co-expression with M and N, suggesting no correlation between SARS-CoV E release capacity and its contribution to efficient VLP production. Furthermore, EΔ76A was still capable of enhancing VLP production even though it was defective in both release and N association. Although EΔ76A is capable of M association (as shown by GST assays), it is not efficiently incorporated into VLPs when co-expressed with M. A possible explanation for this discrepancy is that the disruption of cellular compartments allows E/M to associate at a higher capacity, whereas less E/M association occurs in assembly and budding compartments such as ER/Golgi, resulting in smaller amounts of E being packaged into virions. This scenario is compatible with the proposal that E may promote virus assembly and/or budding by facilitating membrane bending and scissions without required packaging into virions. In support of this hypothesis, one research team has reported that E deletion does not significantly affect SARS-CoV replication, but that E-deleted mutants exhibit 100-to 1,000-fold reductions in virus yields associated with decreased efficiency during morphogenesis [32]. Although E/N interactions may not be directly involved in virus assembly and budding, it remains to be seen whether the SARS-CoV E contribution to virus production requires efficient E/M interaction.

Conclusions
Palmitoylation or interchain disulfide bond formation appears to be dispensable for the SARS-CoV E enhancement of VLP yields. The contribution of SARS-CoV E to efficient VLP production is also independent of its release capacity, association with N, or VLP incorporation. Additional experiments are required to clarify the biological relevance of SARS-CoV E/N interaction, and to determine whether E/N interaction also exists in other coronaviruses.
(See figure on previous page.) Figure 6 Subcellular localization of SARS-CoV E and N proteins. (A) Subcellular localization of wt or mutant SARS-CoV E proteins. HeLa cells were co-transfected with the indicated HA-tagged wt or mutant E expression vectors plus pGolgi-ECFP, which encodes a Golgi apparatus labeling marker. (B) Co-localization of SARS-CoV E and N proteins. HeLa cells were co-transfected with SARS-CoV N bearing carboxyl-terminal tagged EGFP (N-EGFP) and HA-tagged wt or Δ76V expression vectors. At 18 h post-transfection, cells were fixed and labeled with a primary anti-HA antibody and a secondary rhodamine-conjugated anti-mouse antibody. Images represent the most prevalent phenotypes. Merged red/blue or red/green fluorescent images are shown in right-hand column panels. Mock-transfected cells failed to yield any signal (data not shown).