Infectious Bronchitis Virus E Protein Is Targeted to the Golgi Complex and Directs Release of Virus-Like Particles

Expression vectors.

The coding region of IBV E was subcloned by PCR from p57-6 (22). The 5′ primer was designed to contain an EcoRI site and the first three codons of IBV E, which are not present in p57-6, and the 3′ primer contained a BamHI site. IBV E was cloned into pBluescript SK (Stratagene, La Jolla, Calif.) behind the T7 promoter, using EcoRI and BamHI to generate pBS/IBV E, and the sequence of the IBV E open reading frame was confirmed by dideoxy sequencing. The pBS/IBV E plasmid was used to express IBV E in vTF7-3-infected cells. IBV M was expressed in vTF7-3-infected cells from pAR/IBVM, which contains a T7 promoter and was generated by subcloning the coding sequence for M from pSV/IBVE1 (22) into pAR2529-X at the XhoI site. Gm1, a Golgi-retained chimeric protein consisting of the ectodomain and cytoplasmic tail of vesicular stomatitis virus (VSV) G protein and the first transmembrane domain of IBV M, was expressed in vTF7-3-infected cells behind the T7 promoter as described (33).

Antibodies.

Synthetic peptides corresponding to the 14 amino-terminal and 14 carboxy-terminal amino acids of IBV E (each with an added cysteine residue) were synthesized, purified, and coupled to keyhole limpet hemocyanin by Boston Biomolecules, Inc. (Boston, Mass.). Polyclonal antibodies recognizing the amino terminus and carboxy terminus of IBV E were made in rabbits by immunizing with these peptides. The rat polyclonal anti-E antibody was made against the carboxy-terminal E peptide. The rabbit anti-E antibodies were affinity purified for use in indirect immunofluorescence by using the Reduce-Imm reducing kit and the Sulfolink kit (Pierce, Rockford, Ill.) as specified by the manufacturer. The affinity-purified polyclonal anti-IBV M antibody used in immunofluorescence has been described previously (22). The polyclonal anti-IBV M antibody used in immunoprecipitations was generated in rabbits against a peptide corresponding to the carboxy-terminal 14 amino acids as described previously (22). The polyclonal antibody against whole IBV virions was made by immunizing rabbits with purified UV-inactivated IBV virions. Virions were prepared as described previously (31), except that the virus was grown in Vero cells. Antibodies to Gm1 were the mouse monoclonal antibody I1, which recognizes the luminal domain of VSV G (18), and a polyclonal antibody, recognizing the cytoplasmic tail, raised in rabbits to the C-terminal 14 amino acids of VSV G protein. The mouse monoclonal antibodies to GM130 and syntaxin 6 were purchased from Transduction Laboratories, (Lexington, Ky.), and the mouse monoclonal anti-mannosidase II antibody was purchased from Berkeley Antibody Co. (Richmond, Calif.). Texas Red-conjugated goat anti-rabbit, anti-mouse, and anti-rat immunoglobulin G (IgG) and fluorescein-conjugated goat anti-rabbit and anti-mouse IgG were from Jackson ImmunoResearch Laboratories, Inc. (West Grove, Pa.).

Immunofluorescence and confocal microscopy.

For localization studies of IBV E and IBV M in recombinant vaccinia virus-infected cells, BHK-21 cells were plated on coverslips in 35-mm dishes 1 day before infection. vvIBVE or vvIBVE plus vvIBVM were adsorbed at a multiplicity of infection of 5 for each virus in 0.5 ml of serum-free DMEM for 30 min at 37°C. At 6 h postinfection, the cells were fixed in 3% paraformaldehyde in phosphate-buffered saline for 20 min at room temperature, permeabilized with 0.5% Triton X-100, and stained as previously described (33), using affinity-purified anti-IBV E raised to the C terminus. For selective permeabilization of the plasma membrane with digitonin, BHK-21 cells were infected with vTF7-3 (adsorption as above except that Opti-MEM [Life Technologies, Rockville, Md.] was used) and then transfected with 5 μg of a plasmid encoding Gm1 using 10 μl of Lipofectin (Life Technologies), as specified by the manufacturer, or infected with vvIBVE as described above. At 6 h postinfection, the cells were transferred to ice, rinsed in KHM (110 mM potassium acetate, 20 mM HEPES [pH 7.2], 2 mM magnesium acetate), and permeabilized for 5 min on ice with 25 μg of digitonin per ml in KHM as described previously (29). Fixation and staining were as described above, with no subsequent permeabilization step. Images were collected with a Pascal 510 confocal laser-scanning microscope (Zeiss) or a Noran OZ confocal laser-scanning microscope using Intervision software on a Silicon Graphics Indy R5000 platform. All images shown are 0.5-μm optical slices in the z axis near the center of the cell.

Metabolic labeling, alkaline carbonate extraction, glycosidase digestion, and immunoprecipitation.

Cells infected with vTF7-3 and transfected with either pBS/IBVE, pAR/IBVM, pAR/VSVG, or pAR/VSVG_soluble as described above were radiolabeled from 3.5 to 4.5 h postinfection with 50 μCi of ³⁵S-Promix (Amersham Pharmacia Biotech, Inc., Piscataway, N.J.) in methionine- and cysteine-free medium. For alkaline carbonate extraction, cells were homogenized in 15 mM NaCl–10 mM Tris-HCl (pH 7.4)–1 mM MgCl₂–8% sucrose by 60 strokes in a tight-fitting Dounce homogenizer. Nuclei were pelleted by centrifugation, and the supernatant was adjusted to 0.1 M Na₂CO₃ (pH 11.5) and incubated 10 min on ice. A control sample was incubated in 0.1 M NaCl under the same conditions. Membranes were pelleted in a Beckman TLA100 rotor at 132,000 × g for 1 h. The supernatant was adjusted to pH 7, 1% Triton X-100, and 0.2% sodium dodecyl sulfate (SDS), and the pellet was resuspended in detergent solution (62.5 mM EDTA, 50 mM Tris [pH 8], 0.4% deoxycholate, 1.0% Nonidet P-40). The supernatant and pellet were immunoprecipitated with the appropriate antibodies as previously described (22). The immunoprecipitates were analyzed by polyacrylamide gel electrophoresis (PAGE) in the presence of SDS and visualized by fluorography. For N-glycanase digestion, cells were lysed in detergent solution with protease inhibitors and immunoprecipitated with the appropriate antibodies as described previously (22). The immunoprecipitates were treated with N-glycanase as described previously (23), separated by SDS-PAGE on a 15% polyacrylamide gel, and visualized by fluorography.

Western blotting.

Vero cells infected with IBV were lysed in 2× SDS sample buffer–5% β-mercaptoethanol, and the lysates were subjected to SDS-PAGE. Proteins were transferred to nitrocellulose, and the membrane was blocked with 5% nonfat milk in 10 mM Tris (pH 7.4)–150 mM NaCl–0.05% Tween 20 for 1 h at room temperature. Incubation with primary antibodies was carried out overnight at 4°C in the same buffer. The membrane was washed extensively before being incubated with peroxidase-conjugated sheep anti-rabbit IgG for 1 h at room temperature in the same buffer, followed by extensive washing. Bound antibody was detected with SuperSignal West Pico chemiluminescent substrate (Pierce).

IBV E is an integral membrane protein.

IBV E was shown to be associated with purified virions (20), and its sequence predicts a single hydrophobic domain. Microsomal membranes from radiolabeled BHK-21 cells expressing IBV E were subjected to alkaline carbonate extraction to assess whether IBV E was an integral membrane protein. After extraction for 10 min at 0°C with 0.1 M NaCl (control) or 0.1 M Na₂CO₃ (pH 11.5), samples were centrifuged and the membrane pellets and supernatants were subjected to immunoprecipitation with anti-E antibodies. The association of IBV E with the pelleted membranes in the presence of 0.1 M Na₂CO₃ indicates that it is an integral membrane protein (Fig. ​(Fig.2).2). Controls were performed using VSV G protein, a known integral membrane protein, and VSV G_soluble, a secreted form of G which lacks the transmembrane domain (data not shown). VSV G was associated with the pelleted membranes in the presence of both 0.1 M NaCl and 0.1 M Na₂CO₃, which is consistent with its integral membrane association. VSV G_soluble pelleted with membranes after 0.1 M NaCl treatment but was extracted from membranes with 0.1 M Na₂CO₃, as expected for a nonintegral, secreted protein.

Open in a separate window

FIG. 2

IBV E is an integral membrane protein. BHK-21 cells infected with vvIBVE were radiolabeled from 3.5 to 4.5 h postinfection, and microsomes were prepared from homogenized cells. The microsomes were extracted with either 0.1 M NaCl (left) or 0.1 M Na₂CO₃, (pH 11.5) (right), and the membranes were pelleted. Pellets (P) and supernatants (S) were immunoprecipitated with anti-IBV E in the presence of detergent, and the immunoprecipitates were analyzed by SDS-PAGE on a 15% polyacrylamide gel and visualized by fluorography.

Membrane topology of IBV E.

To determine the orientation of IBV E in membranes, we used antibodies specific for the N and C termini of the protein. Preliminary indirect-immunofluorescence microscopy suggested that IBV E was localized to an intracellular juxtanuclear compartment when expressed alone from cDNA (see Fig. ​Fig.6).6). BHK-21 cells transiently expressing the E protein were permeabilized with digitonin, which at low concentrations selectively permeabilizes the plasma membrane but leaves intracellular membranes intact (29). To confirm that intracellular membranes were indeed not permeabilized when the cells were treated with digitonin, BHK-21 cells transiently expressing the Golgi-resident Gm1 protein (33) were stained with antibodies specific either to the luminal head domain (Fig. ​(Fig.3b3b and f) or to the cytoplasmic tail domain (Fig. ​(Fig.3a3a and e). The cytoplasmic tail epitope should always be accessible to antibody when the plasma membrane is permeabilized, while the luminal epitope should be accessible only when Golgi membranes are also permeabilized. As expected, the cytoplasmic tail domain of Gm1 was accessible to antibody when the cells were permeabilized with either digitonin (Fig. ​(Fig.3e)3e) or Triton X-100 (Fig. ​(Fig.3a),3a), while the luminal head domain was not accessible to antibody when cells were permeabilized with digitonin (Fig. ​(Fig.3f),3f), consistent with an unpermeabilized Golgi apparatus. The IBV E protein was accessible to C-terminus-specific antibodies in the presence of either digitonin (Fig. ​(Fig.3g)3g) or Triton X-100 (Fig. ​(Fig.3c)3c) but was accessible to N-terminus-specific antibodies only in the presence of Triton X-100 (Fig. ​(Fig.3d).3d). These results are consistent with IBV E possessing a luminal N terminus, a cytoplasmic C terminus, and a single transmembrane domain.

Open in a separate window

FIG. 3

The C terminus of IBV E is cytoplasmic. BHK-21 cells were infected with vTF7-3 and transfected with a plasmid encoding the Golgi resident protein Gm1 (a, b, e, and f) or infected with vvIBV E (c, d, g, and h) as described in Materials and Methods. At 6 h postinfection, cells were either permeabilized with digitonin (e to h) and fixed for immunofluorescence or fixed and permeabilized with Triton X-100 prior to staining (a to d). The cells were stained with antibodies to the cytoplasmic tail of Gm1 (a and e), the luminal head of Gm1 (b and f), the C terminus of IBV E (c and g), or the N terminus of IBV E (d and h). Secondary antibodies were fluorescein-conjugated goat anti-mouse IgG (b and f) and Texas red-conjugated goat anti-rabbit IgG (all other panels).

Open in a separate window

FIG. 6

The IBV E protein is localized to the Golgi complex. BHK-21 cells infected with vvIBVE were fixed for immunofluorescence at 6 h postinfection, permeabilized with Triton X-100, and double labeled with antibodies to IBV E and different endogenous Golgi resident proteins. (a to d) Cells were stained with rabbit anti-IBV E and mouse anti-GM130; (e to h) cells were stained with rabbit anti-IBV E and mouse anti-mannosidase II; (i to l) cells were stained with rabbit anti-IBV E and mouse anti-syntaxin 6. Secondary antibodies were Texas red-conjugated goat anti-rabbit IgG and fluorescein-conjugated goat anti-mouse IgG. In each row, the red image corresponds to IBV E staining and the green image corresponds to the appropriate Golgi marker. The third image in each row (c, g, and k) is a merged image, where yellow represents overlap between the red- and green-staining patterns. The fourth image in each row (d, h, and l) is a phase image of each field of labeled cells. Bar, 10 μm.

Posttranslational modifications of IBV E.

IBV E contains a consensus site for N-linked glycosylation very near its N terminus. Given that the N terminus is luminal (Fig. ​(Fig.3)3) and given the utility of glycosylation in studying trafficking of proteins, we performed N-glycanase digestion to determine if the site is used. BHK-21 cells transiently expressing the IBV E protein were radiolabeled and immunoprecipitated with anti-E antibody. The immunoprecipitates were digested with N-glycanase. The results, shown in Fig. ​Fig.4,4, indicate that IBV E does not undergo N-linked glycosylation, since its electrophoretic mobility does not change after treatment with N-glycanase. As a positive control for N-glycanase digestion, IBV M, which is known to have N-linked sugars (31), was also treated with N-glycanase and shown to increase in mobility under these conditions (Fig. ​(Fig.4).4).

Open in a separate window

FIG. 4

The N-terminal N-linked glycosylation site of IBV E is not used. BHK-21 cells infected with vTF7-3 were transfected with plasmids encoding IBV E or IBV M, radiolabeled from 3.5 to 4.5 h postinfection and lysed in detergent solution. The lysates were immunoprecipitated with appropriate antibodies to E or M. The immunoprecipitates were mock treated (−) or treated with N-glycanase (+), analyzed by SDS-PAGE on a 15% polyacrylamide gel, and visualized by fluorography.

IBV E, like other coronavirus E proteins, contains cysteine residues adjacent to its transmembrane domain, and there is evidence that MHV E is posttranslationally acylated (40). However, we were unable to detect incorporation of [³H]palmitate into the IBV E protein in BHK-21 cells (data not shown).

Release of VLPs from cells expressing IBV E and M.

We determined if VLPs could be generated by expression of IBV E and M, as reported for other coronaviruses (3, 25, 37). IBV E and M were expressed using the recombinant vaccinia viruses vvIBVE and vvIBVM (22). Cells were radiolabeled from 3.5 to 4.5 h postinfection and chased for 3 h. Supernatants were collected, cleared of debris, and immunoprecipitated with antibodies to E and M (Fig. ​(Fig.5a).5a). IBV E was released into the supernatant regardless of whether IBV M was present in transfected cells. However, IBV M was not released into the supernatant unless IBV E was present. These results suggest that E is required for the release of these proteins from cells, because M was not released when expressed alone.

Open in a separate window

FIG. 5

IBV E is released from transfected cells in VLPs that incorporate IBV M if present. (a) BHK-21 cells were mock infected (−), infected with vvIBVE (E), vvIBVM (M), or vvIBVE plus vvIBVM (E+M), radiolabeled from 3.5 to 4.5 h postinfection, and chased for 3 h. Medium was collected, cleared of debris, and immunoprecipitated with anti-E and anti-M antibodies. The immunoprecipitates were analyzed by SDS-PAGE and visualized by fluorography. (b) BHK-21 cells coinfected with vvIBVE and vvIBVM were radiolabeled from 3.5 to 4.5 h postinfection and chased for 3 h. Concentrated supernatants were loaded onto a continuous 20 to 55% sucrose gradient and centrifuged to equilibrium. Fractions were collected and immunoprecipitated with anti-E and anti-M antibodies. The immunoprecipitates were separated by SDS-PAGE and visualized by fluorography. Fraction 1 corresponds to the top of the gradient, and fraction 12 corresponds to the bottom of the gradient. (c) BHK-21 cells were infected with vvIBVE only and treated exactly as described for panel b.

To determine whether the proteins are released in sedimentable particles, as shown for other coronaviruses, the supernatants from cells expressing E or E and M together were loaded onto 20 to 55% sucrose gradients and centrifuged to equilibrium. Fractions were collected and immunoprecipitated with appropriate antibodies. The immunoprecipitates were electrophoresed and visualized by fluorography, and the results are shown in Fig. ​Fig.5b5b (E and M) and c (E alone). The presence of E and M proteins in fractions 4, 5, and 6 (Fig. ​(Fig.5b)5b) shows that the proteins released into the supernatant are present in sedimentable particles. Fraction 5, which contained the majority of the M protein, had a density of 1.11 g/cm³. The E protein was also released from cells in sedimentable particles when expressed alone, as shown in Fig. ​Fig.5c.5c. The peak density for the E protein, both when coexpressed with M and when expressed alone, occurred between fractions 5 and 6 and was measured to be 1.14 g/cm³. These results differ from those reported for MHV E and M proteins, where particles containing both proteins were shown to be denser than those containing MHV E alone (25). The difference might be explained by our observation that the level of IBV E protein released from cells expressing E alone was consistently greater than that of E released from cells expressing both E and M proteins (Fig. ​(Fig.5a).5a). We noted that the release of both types of particles occurred at extremely low efficiency, since the supernatants contained less than 0.01% of cellular E or M proteins.

We also attempted to assess IBV E and IBV M interactions by coimmunoprecipitation from recombinant vaccinia virus-infected cells, but we found no evidence for stable association between the two proteins (data not shown). Also, the rate of IBV M trafficking, measured by monitoring the processing of its N-linked oligosaccharides at different times after synthesis, was unchanged in cells coexpressing IBV E (data not shown). Thus, it appears that IBV E and IBV M do not interact strongly in transfected cells or in detergent lysates.

IBV E is targeted to the Golgi complex.

To establish the intracellular localization of IBV E, BHK-21 cells infected with vvIBVE were analyzed by indirect immunofluorescence using confocal microscopy. A juxtanuclear staining pattern was observed, which was compared with the localizations of three endogenous Golgi marker proteins (Fig. ​(Fig.6).6). The endogenous Golgi proteins that were analyzed were GM130, a cis-Golgi resident (26), mannosidase II, a Golgi stack marker (35), and syntaxin 6, a trans-Golgi network and endosomal resident (5). The red images (Fig. ​(Fig.6a,6a, e, and i) show the intracellular distribution of the IBV E protein, and the green images (Fig. ​(Fig.6b,6b, f, and j) show that of each marker protein. The third image in every row (Fig. ​(Fig.6c,6c, g, and k) corresponds to the overlay of red and green images; yellow represents regions of overlap between the red- and green-staining patterns. The staining pattern of IBV E was most similar to that of mannosidase II, whereas there was somewhat less overlap with either GM130 or syntaxin 6. These results demonstrate that IBV E is targeted to Golgi membranes in the absence of the other IBV proteins and that it reaches the Golgi stacks, where it appears to accumulate. No plasma membrane staining was observed in IBV E-expressing cells.

Topology and intracellular localization of IBV E.

Using indirect-immunofluorescence microscopy in conjunction with the detergent digitonin, which selectively permeabilizes the plasma membrane of cells when used at low concentrations (29), and domain-specific antibodies generated against the IBV E protein, we showed that the carboxyl terminus of E is cytoplasmic. This topology is predicted from the positive-inside rule (38), and the E proteins of all coronaviruses contain a positively charged residue(s) on the C-terminal sides of their predicted transmembrane domains. This topology places IBV E in the type III class of integral membrane proteins (38), since it lacks a cleaved signal sequence. Given this topology of the IBV E protein, it was possible that the protein was N glycosylated at a consensus site near the N terminus. We showed that the glycosylation site is not used, probably because of its extreme proximity to the membrane (27). The topology of IBV E determined here differs from that previously published for TGEV E (9). Godet et al. showed that the C terminus of cell surface TGEV E was extracellular because it was accessible to antibody in nonpermeabilized cells (9).

We showed by indirect immunofluorescence and confocal microscopy that the IBV E protein is localized to the Golgi complex in both transfected and IBV-infected cells. We did not detect IBV E at the cell surface, as described for bovine coronavirus E and TGEV E proteins (1, 9). When we compared the localization of IBV E to that of three marker proteins that reside in different regions of the Golgi complex, we found that the distribution of IBV E most closely overlapped that of mannosidase II, a Golgi stack marker. There was less, but significant, overlap with the trans-Golgi resident syntaxin 6 and the cis-Golgi protein GM130. The proximity of the subcellular distribution of the E protein to the coronavirus budding site is interesting in light of the hypothesis that E plays a role in directing the accumulation of coronavirus envelope proteins at the budding site. The IBV M protein is targeted to the cis-Golgi complex when it is expressed from cDNA, at least in part by information contained in its first transmembrane domain (23, 24, 33). Here we have shown that IBV encodes another envelope protein (E) that possesses Golgi targeting information. It will be interesting to dissect the targeting information in IBV E to determine whether E and M have a common mechanism for Golgi localization.

When the IBV E and M proteins were expressed together in cells, either by coinfection with recombinant vaccinia viruses or by IBV infection, they colocalized when analyzed by indirect immunofluorescence and confocal microscopy. This result is intriguing, since it may point to interactions between the E and M proteins that result in their localization in the same compartment. However, we were unable to demonstrate interactions between IBV E and M directly. We are interested in interactions that the E protein may have with the other IBV proteins and the possible role of these interactions in gathering the envelope proteins at the CGN for budding. We plan to further examine the targeting of the IBV E protein and the effect of its expression on the localization of IBV M at the ultrastructural level in order to address these questions.

IBV VLPs are produced inefficiently.

When the IBV E and M proteins were transiently expressed in cells, they were released into the culture supernatants in sedimentable particles. We also observed that the IBV E protein was released in similar particles when expressed alone. These particles, when centrifuged to equilibrium on sucrose density gradients, had densities of 1.11 g/cm³ (particles containing both E and M) and 1.14 g/cm³ (particles containing E alone). The release of coronavirus E protein from cells in membranous particles has obvious implications for the importance of the role of E in virus assembly. However, at least for IBV proteins in BHK-21 cells, particle release was extremely inefficient, since the amounts of E and M proteins released into the supernatant were less than 0.01% of the cellular E and M levels. Clearly, other IBV components must be required for efficient budding and/or release of virus from cells. The efficiency of VLP formation in cells expressing MHV and TGEV E and M proteins was not reported (3, 25, 37), and so it is not yet clear if the assembly mechanism of IBV differs from that of these other coronaviruses.

Can the E protein drive IBV assembly?

It was proposed that coronavirus M and E proteins are the minimal assembly unit, since expression of these proteins resulted in their release in VLPs (37). This is unusual, since the assembly of many enveloped viruses is nucleocapsid dependent. Recently, it was shown that expression of MHV E alone induces VLPs (25), as we have shown here for IBV E. However, the efficiency of VLP release from cells expressing IBV E with or without M is extremely low. Therefore, it is not clear if IBV VLP production is relevant to virus assembly. IBV E is expressed from a tricistronic RNA (21), and the two open reading frames upstream of IBV E are expressed in IBV-infected cells (19). The protein encoded by open reading frame 3a is extremely hydrophobic and is predicted to be a membrane protein. It remains possible that the proteins encoded by open reading frames 3a and/or 3b are involved in assembly and budding-site selection, and perhaps they function in concert with the IBV E protein in these processes. We are currently investigating this possibility.

We thank M. Delannoy for confocal microscopy expertise and C. Buck for the pSCIIMCS1 vector and helpful advice regarding coinfection with recombinant vaccinia viruses.

This work was supported by National Institutes of Health grant GM42522.