ABSTRACT
Orthopoxviruses produce two, antigenically distinct, infectious enveloped virions termed intracellular mature virions and extracellular virions. Extracellular virions are required for cell-to-cell spread and pathogenesis. Specific to the extracellular virion membrane, glycoproteins A33, A34, and B5 are highly conserved among orthopoxviruses and have roles during extracellular virion formation and subsequent infection. B5 is dependent on an interaction with either A33 or A34 for localization to the site of intracellular envelopment and incorporation into the envelope of released extracellular virions. In this report we show that an interaction between A33 and A34 can be detected in infected cells. Furthermore, we show that a three-protein complex between A33, A34, and B5 forms in the endoplasmic reticulum (ER) that disassociates post ER export. Finally, immunofluorescence reveals that coexpression of all three glycoproteins results in their localization to a juxtanuclear region that is presumably the site of intracellular envelopment. These results demonstrate the existence of two previously unidentified interactions: one between A33 and A34 and another simultaneous interaction between all three of the glycoproteins. Furthermore, these results indicate that interactions among A33, A34, and B5 are vital for proper intracellular trafficking and subcellular localization.
IMPORTANCE The secondary intracellular envelopment of poxviruses at the trans-Golgi network to release infectious extracellular virus (EV) is essential for their spread and pathogenesis. Viral glycoproteins A33, A34, and B5 are critical for the efficient production of infectious EV and interactions among these proteins are important for their localization and incorporation into the outer extracellular virion membrane. We have uncovered a novel interaction between glycoproteins A33 and A34. Furthermore, we show that B5 can interact with the A33-A34 complex. Our analysis indicates that the three-protein complex has a role in ER exit and proper localization of the three glycoproteins to the intracellular site of wrapping. These results show that a complex set of interactions occur in the secretory pathway of infected cells to ensure proper glycoprotein trafficking and envelope content, which is important for the release of infectious poxvirus virions.
INTRODUCTION
Vaccinia virus (VACV), the prototypical member of the Orthopoxvirus genus, was used as a live-attenuated virus vaccine for the eradication of variola virus, the causative agent of smallpox. VACV contains a double-stranded DNA genome of approximately 200 kbp that is predicted to encode more than 200 open reading frames (1–3). Replication occurs entirely in the cytoplasm and results in the production of three morphologically and antigenically distinct forms of the virus: intracellular mature virions (IMV), intracellular enveloped virions (IEV), and extracellular virions (EV) (4, 5). A subset of IMV, the first infectious progeny produced, are trafficked to the trans-Golgi network (TGN), where they are enveloped with two additional membranes to produce IEV (6–9). IEV are transported through the cytoplasm to the cell periphery, where the outermost membrane fuses with the plasma membrane to release a double membraned form, termed EV (10, 11). EV that remain on the cell surface are called cell-associated enveloped virions, while EV that are no longer attached to the cell surface are called extracellular enveloped virions (EEV) (12–14). The production of EV is critical for the rapid cell-to-cell spread and long-range dissemination of orthopoxviruses (3, 12, 14).
Glycoproteins A33, A34, B5, and A56 are exposed on the outer surface of EV, and, with the exception of A56, are required for the efficient production of infectious EV (15–21). Moreover, A33, A34, and B5 are approximately 94 to 97% identical between the Western Reserve strain of VACV (WR) and strains of variola. B5 is a type I transmembrane glycoprotein which has been shown to play an important role in the formation of EV and has been suggested to play a role in EV cell binding (16, 18, 22). Both A33 and A34 are type II transmembrane glycoproteins with homology to C-type lectin-like domains (CTLD) (19, 23, 24). Deletion of either A33R or A34R results in an increased release of EV (17, 25) that is reduced in cell binding and, subsequently, specific infectivity (26, 27). Moreover, all three glycoproteins are exposed on the surface of EV and have specific functions for viral infectivity (13, 15, 17, 25, 28–30). Furthermore, less A33 and B5 are incorporated into the EV envelope in the absence of A34 (27, 31).
The current study investigated an interaction between the two glycoproteins A33 and A34. Here, we detect an interaction between these two proteins in infected cells. Utilizing a set of bimolecular fluorescence complementation (BiFC) constructs, we show that B5 is capable of interacting with the A33-A34 complex to form a three-protein complex. Moreover, our results reveal that the three-protein complex is transient, occurs in the ER, and results in the trafficking of A33, A34, and B5 out of the ER. These results suggest that there is a complex set of interactions between poxvirus glycoproteins that facilitate their exit from the ER and ensures proper targeting to the intracellular site of EV envelopment.
RESULTS
A33 and A34 interact through their ectodomain.Both A33 and A34 have been reported to interact with B5 (22, 26, 27, 29, 31–34). Whereas multiple reports have speculated about an A33-A34 interaction, supporting data are lacking (29, 31, 34). To determine if A34 is capable of interaction with A33, coimmunoprecipitation (co-IP) was performed with epitope-tagged, full-length A34 and A33 (A33 had a C-terminal HA tag and A34 had a C-terminal V5 tag). Constructs were overexpressed in HeLa cells using the VACV T7 expression system (35) in the presence of cytosine arabonoside (AraC) to block expression of postreplicative viral genes. The following day, A34 was immunoprecipitated with an anti-V5 antibody and coimmunoprecipitated proteins were analyzed by Western blotting (Fig. 1A). A band corresponding to A33-HA was precipitated with A34-V5, suggesting that A33-HA and A34-V5 interact in the absence of intermediate and late viral genes. Since previous reports have suggested that A34 and B5 interact in the endoplasmic reticulum (ER) for trafficking to the site of wrapping (32), we theorized that A33 and A34 may also interact in the ER. Therefore, we repeated the experiment in the presence and absence of a protein trafficking inhibitor, brefeldin A (BFA), which inhibits trafficking out of the ER (36, 37). A band corresponding to A33-HA was precipitated with A34-V5 in the presence of BFA (Fig. 1A), demonstrating that the A33-A34 interaction occurs in the ER. Interestingly, in the absence of A33-HA, A34-V5 expression was reduced (Fig. 1B).
Coimmunoprecipitation of A33 and A34. HeLa cells were infected with vTF7-3, or with vTF7-3ΔB5 where indicated, at an MOI of 5 and transfected with the indicated plasmids in the presence and absence (+/−) of AraC and BFA. The following day, cells were lysed and lysates were incubated with anti-V5 followed by either protein G-dynabeads (A) or protein G-agarose (E). Immune complexes (A, C, and E) and cell lysates (B, D, and F) were analyzed by Western blotting with rat HRP-conjugated anti-HA MAb (C and E), rat anti-HA MAb followed by HRP-conjugated donkey anti-rat antibody, mouse anti-V5 MAb followed by HRP-conjugated donkey anti-mouse antibody (B and F), or Alexa Fluor 647-conjugated donkey anti-mouse antibody (D), rat anti-B5 MAb followed by Alexa Fluor 594-conjugated donkey anti-rat, and mouse anti-actin MAb followed by Alexa Fluor 750-conjugated goat anti-mouse antibody. The masses in kilodaltons and positions of marker proteins are shown on the left of the images.
Both A33 and A34 have been shown to interact with B5. B5R has been reported to have a hybrid promoter that allows for both early and late expression during infection (1, 38, 39). Indeed, deep RNA sequencing of cells infected in the presence of AraC could detect ∼1/10 of the total B5R transcripts at early times (1). Although we have never seen detectable levels of B5 in the presence of AraC compared to B5 overexpressed with the T7 system (22, 27, 32, 33, and this study), we considered the possibility that the A33-A34 interaction we detected was being mediated (either directly or indirectly) by the small amount of B5 that is expressed early. To rule out this possibility, we deleted B5R from vTF7-3 and repeated the A33-A34 co-IP in the presence and absence of AraC. As seen previously (Fig. 1A), a band corresponding to A33-HA was precipitated with A34-V5 from cells infected with either vTF7-3 or vTF7-3ΔB5 (Fig. 1C). Importantly, compared to the endogenously expressed B5 in the absence of AraC, there was little B5 detected in the lysates of cells infected with vTF7-3 in the presence of AraC, and no B5 was detected in the lysates of cells infected with vTF7-3∆B5 (Fig. 1D). These results demonstrate that B5 is not required, either directly or indirectly, for A33 and A34 to interact. In the absence of AraC, there was a reduction in the amount of A33-HA precipitated with A34-V5 in cells infected with vTF7-3∆B5 (Fig. 1C), which could be due to interference from other intermediate or late proteins or due to lower expression (Fig. 1D).
A33 and A34 are both type II transmembrane proteins that have short cytoplasmic tails (41 and 18 amino acids, respectively), a single transmembrane domain, and the majority of the protein contained within the C-terminal ectodomains (19, 24). Therefore, we thought it likely that A33 and A34 interact through their ectodomains. To test this idea, we made constructs that expressed only the predicted ectodomain of A33-HA and A34-V5 along with a cleavable signal sequence for translocation into the ER lumen (A3436-168-V5 and A3357-185-HA) for coimmunoprecipitation. An additional A34 construct was made that had the V5 epitope tag on the N terminus (V5-A34). A band corresponding to A33-HA was precipitated with both V5-A34 and A34-V5, demonstrating that addition of the V5 epitope tag to either end of A34 does not alter the interaction (Fig. 1E). In addition, bands consistent with the size of A33-HA or A3357-185-HA were precipitated with A34-V5 and A3436-168-V5, respectively (Fig. 1E). Similarly, an interaction was detected between A3357-185-HA and A3436-168-V5, demonstrating that the ectodomains of these two proteins are sufficient for their interaction. Analysis of the cell lysates confirmed that all of the constructs are expressed, but we were unable to detect A3357-185-HA when coexpressed with A34-V5 (Fig. 1F), which could explain why we were unable to detect an interaction (Fig. 1E).
A33 and A34 interact during infection.The previous results show that A33 and A34 interact when overexpressed in cells. To determine if these glycoproteins interact during infection, a set of recombinant viruses expressing A34R under its normal promoter and containing either a C- or N-terminal V5 epitope tag were generated by inserting either V5-A34R or A34R-V5 into a virus that had the A34R gene deleted (vΔA34R) (15). Similar to previous results, neither vV5-A34R nor vA34R-V5 showed a defect in plaque phenotype compared to WR (data not shown) (27). To examine the A33-A34 interaction during infection, coimmunoprecipitation was conducted on cells infected with vA34R-V5, vV5-A34R, or WR. Coimmunoprecipitation was performed using either an anti-V5 or an anti-A33 antibody. Both V5-A34 and A34-V5 were immunoprecipitated with the anti-A33 antibody, although a reduced amount of A34-V5 was precipitated compared to V5-A34 (Fig. 2A). Similar results were obtained with the anti-V5 antibody, which precipitated A33 in both recombinants, demonstrating that A33 and A34 interact during VACV infection and supporting our transient-transfection results. Importantly, A34-V5, V5-A34, and A33 were expressed to similar levels by the recombinant viruses (Fig. 2B).
Coimmunoprecipitation of A33 from infected cells. HeLa cells were infected with the indicated viruses at an MOI of 5. The following day, cells were lysed and lysates immunoprecipitated with either rabbit anti-A33 antiserum or mouse anti-V5 MAb followed by protein G-agarose (A) or protein G-dynabeads (C). Immune complexes (A and C) and cell lysates (B and D) were analyzed by Western blotting with HRP-conjugated anti-V5 MAb (B, left), rabbit anti-A33 antiserum followed by either HRP-conjugated donkey anti-rabbit antibody (A and B, right) or Alexa Fluor 647-conjugated donkey anti-rabbit antibody (D), mouse anti-V5 MAb followed by HRP-conjugated donkey anti-mouse antibody, and mouse anti-actin MAb followed by Alexa Fluor 750-conjugated goat anti-mouse antibody. The masses in kilodaltons and positions of marker proteins are shown on the left of the images.
To further define the A33 interaction site on A34 during infection, we examined the A33-A34 interaction in cells infected with a previously constructed set of recombinant viruses that express C-terminal truncations in A34 (vA34R1-130-V5, vA34R1-100-V5, vA34R1-70, and vΔA34R) using coimmunoprecipitation (Fig. 2C) (27). A band corresponding to A341-130 was coimmunoprecipitated with A33 from lysates of cells infected with vA34R1-130-V5; however, we were unable to detect an interaction of A33 with A34 for vA34R1-100-V5 and vA34R1-70-V5 (Fig. 2C), which could be due to poor expression (Fig. 2D). These results suggest that the A33 interaction site is contained within residues 36 to 130 of A34 and partially overlaps with the B5 interaction site on A34 (residues 80 to 130) (27).
A33 localizes to the site of IMV wrapping to form IEV in the absence of the A33-A34 interaction site.Previous reports from our lab and others have shown that in the absence of A34, the incorporation of A33 into released EV is reduced (27, 31). In light of our current findings, this suggests the A33-A34 interaction is required for efficient incorporation of A33 into released EV. To understand if the removal of the A33 interaction site from A34 affects localization of A33 to the site of IMV wrapping to form IEV and subsequent EV incorporation, we examined the intracellular localization of A33 and A34 in cells infected with our panel of A34 truncation viruses. Cells on coverslips were infected with vA34R-V5, vA34R1-130-V5, vA34R1-100-V5, vA34R1-70-V5, or vΔA34R and subsequently fixed, permeabilized, and stained with anti-A33 and anti-V5 antibodies (Fig. 3). Cells infected with vA34R-V5 showed colocalization of A33 and A34 to the site of wrapping (arrows), at the cell vertices (concave arrowheads), and in virion-sized particles (VSPs) (arrowheads) (Fig. 3). Similar to what has been described previously (29), during infection with vΔA34R-RFP, few VSPs stained with A33 were detected and A33 staining at the site of wrapping was markedly reduced. For vA34R1-130-V5, which contains the A33 interaction site, A33 was localized similarly to vA34R-V5, with A33 staining at the site of wrapping, at the cell vertices, and at VSPs. However, A34 localization was more diffuse throughout the cell and absent from VSPs, but still appeared to concentrate in the juxtanuclear region and colocalized with A33 at this location. In cells infected with vA34R1-100-V5 and vA34R1-70-V5, A33 was still predominantly localized to the site of wrapping, cell vertices, and VSPs. Interestingly, the A33 localization pattern for these A34 C-terminal truncation viruses looks similar to what has been previously described for B5 (27). Together, these data suggest that A33 is properly localized to the site of wrapping in cells infected with the recombinant viruses, despite the lack of a detectable A33-A34 interaction.
Intracellular localization of A33 and A34. HeLa cells, grown on coverslips, were infected with the indicated viruses at an MOI of 0.5. The next day, cells were fixed, permeabilized, and incubated with rabbit anti-A33 antiserum and mouse anti-V5 MAb followed by Cy2-conjugated donkey anti-rabbit (green) and Alexa Fluor 594-conjugated anti-mouse (red) antibodies. Coverslips were mounted on microscope slides with ProLong Gold Antifade Reagent with DAPI to stain DNA (blue) and imaged via fluorescence microscopy. The overlap of A33 (green) and V5 (red) is shown as yellow. Protein localization at the site of wrapping, cell vertices, and VSPs is shown by arrows, concave arrowheads, and arrowheads, respectively. Red staining in the vΔA34R-RFP panel is RFP fluorescence and not V5 staining. Scale bar 10 μm.
A33, A34, and B5 can form a three-protein complex.B5 and B5-GFP localization to the site of wrapping and incorporation into the EV envelope is dependent on an interaction with either A34 or A33, respectively (26, 31, 32). Our discovery of an A33-A34 interaction (Fig. 1 and 2) suggests that the three glycoproteins A33, A34, and B5 could form a single complex. To test this idea, we utilized bimolecular fluorescence complementation (BiFC) to visualize the A33-A34 complex and probe for a simultaneous interaction with B5. A33 and A34 chimeric fusions were made with either the amino-terminal 158 residues of YFP (YPFN) or the carboxyl-terminal 75 residues of YFP (YFPC) (Fig. 4A). When the chimeric proteins (A33 and A34) appended to the two parts of YFP (YFPN and YFPC) interact, they bring the two fragments of YFP together and form a functional YFP (40). To test the BiFC chimeras, we coexpressed the YFPN and YFPC chimeras and looked for YFP fluorescence (Fig. 4). A33 is known to form homodimers and therefore served as a control for the functionality of our chimeras. As expected, co-expression of YFPN-A33HA with YFPC-A33HA showed fluorescence in the juxtanuclear region, indicating that the two A33 BiFC chimeras are expressed and that they folded properly, dimerized, fluoresced, and trafficked out of the ER (Fig. 4B). Similarly, coexpression of YFPN-A34V5 and YFPC-A34V5 resulted in fluorescence (Fig. 4B), indicating that both A34 BiFC chimeras are expressed and that A34 is capable of forming homodimers. Expression of either YFPN-A34V5 or YFPC-A34V5 with YFPC-A33HA or YFPN-A33HA, respectively, also resulted in BiFC fluorescence, confirming that these two glycoproteins are capable of interaction. Coexpression of the A33 and A34 BiFC chimeras with their untagged YFP partners did not result in detectable BiFC fluorescence, indicating the specificity of the YFP fluorescence to interacting chimeras (Fig. 4B).
BiFC constructs and interactions. (A) Diagram of constructs used for BiFC. Constructs include the cytoplasmic tail (CT), the transmembrane domain (TM), and the ectodomain (ED). N- and C-terminal portions of YFP were appended to the N terminus of A33 and A34. A33 includes an HA epitope tag (hatched) and A34 includes a V5 epitope tag (gray). (B) HeLa cells grown on coverslips were infected with vTF7-3 at an MOI of 0.5 and transfected with the indicated constructs in the presence of AraC. BiFC fluorescence is shown (green). Coverslips were mounted on microscope slides with ProLong Gold Antifade Reagent containing DAPI to stain DNA (blue). Stained coverslips were imaged via fluorescence microscopy. Scale bar 10 μm.
To look for complex formation, BiFC constructs YFPN-A33HA and YFPC-A34V5 were overexpressed with B5 in HeLa cells using the T7 expression system in the presence of AraC. The following day, proteins were immunoprecipitated with an anti-GFP antibody that only recognizes the fully folded form of YFP, and thus would only precipitate an A33-A34 BiFC complex and not the individual YFP fragments. Precipitated proteins were analyzed by Western blotting (Fig. 5A). When B5 was co-expressed with YFPN-A33HA and YFPC-A34V5, a faint band corresponding to B5 was immunoprecipitated (Fig. 5A), suggesting that A33, A34, and B5 are capable of forming a three-protein complex. Importantly, YFPN-A33HA and YFPC-A34V5 were only immunoprecipitated when both constructs were coexpressed and not when they were expressed individually (Fig. 5A), verifying that the GFP antibody only recognizes the fully folded form of YFP and does not precipitate the individual YFP fragments. B5 has been shown to interact with both A33 and A34 independently (22, 26, 29, 31–34). In our system, B5 was only precipitated in the presence of both A33 and A34 BiFC constructs (Fig. 5A), and not when the constructs were individually expressed, further confirming that the GFP antibody only precipitates the A33-A34 BiFC complex.
Coimmunoprecipitation of an A33-A34-B5 complex. HeLa cells were infected with vTF7-3 at an MOI of 5 and transfected with the indicated plasmids in the presence of AraC (A and B) and BFA (C and D). The following day, cells were lysed and lysates incubated with mouse anti-GFP MAb followed by protein G-dynabeads. Immune complexes (A and C) and cell lysates (B and D) were analyzed by Western blotting with rat anti-B5 MAb followed by HRP-conjugated donkey anti-rat antibody, rat anti-HA MAb followed by Alexa Fluor 647-conjugated donkey anti-rat antibody, mouse anti-V5 MAb followed by HRP-conjugated donkey anti-mouse antibody, and mouse anti-actin MAb followed by Alexa Fluor 750-conjugated goat anti-mouse antibody. The masses in kilodaltons and positions of marker proteins are shown on the left of the images.
The previous results suggest that a three-protein complex between A33, A34, and B5 is capable of forming, but is weakly detectable (Fig. 5A). Interactions between B5 and A34 in addition to B5 and A33 have been reported to be necessary for proper trafficking from the ER to the site of wrapping at the trans-Golgi network (TGN) for envelope incorporation (26, 32). Therefore, we hypothesized that before trafficking to the TGN, the three-protein complex transiently interacts in the ER. To better visualize the complex, we conducted the same experiment as described above (Fig. 5A and B) in the presence of brefeldin A (BFA) to block trafficking out of the ER. With BFA treatment, we detected a dramatic increase in the amount of B5 that precipitated with the BiFC complex (Fig. 5C). Importantly, the addition of BFA did not result in a drastic change in the amount of the BiFC constructs that were expressed (Fig. 5B and D). However, we did notice that the expression of B5 was reduced when coexpressed with A34 or A33 and A34 both in the presence and absence of BFA (Fig. 5B and D). These results suggest that the A33-A34-B5 complex occurs in the ER and that the interaction is transient. We also noticed that when the A33 and A34 BiFC constructs were coexpressed, both were immunoprecipitated with the GFP antibody, whether BFA was present or not (Fig. 5C), suggesting that, unlike the three-protein complex, the A33-A34 BiFC complex is not disassociated upon trafficking out of the ER.
The A33-A34-B5 complex is detected in the absence of BiFC.We considered the possibility that the high affinity of the fully folded YFP molecule holds A33 and A34 in a complex upon further trafficking (41, 42) and that the stability of the BiFC interaction in conjunction with BFA treatment, which limits protein trafficking out of the ER, could force an aberrant B5 interaction with the A33-A34 BiFC complex. To verify that a three-protein complex occurs in the absence of the BiFC interaction, we performed a two-step coimmunoprecipitation assay using anti-HA and anti-V5 antibodies sequentially. Constructs were overexpressed in HeLa cells using the T7 expression system in the presence of AraC and BFA. The following day, proteins were immunoprecipitated with an anti-HA antibody conjugated to protein G-dynabeads, bound proteins were eluted from the dynabeads using an HA peptide, and subsequently immunoprecipitated with an anti-V5 antibody before Western blot analysis (Fig. 6A). A band corresponding to B5 was precipitated with the HA and V5 antibodies (Fig. 6A), indicating that B5 does interact with the A33-A34 complex. Due to the separate and successive precipitation of A33-HA followed by A34-V5, it is likely that any B5 precipitated was a result of a simultaneous interaction with both A33 and A34 and demonstrates the existence of a three-protein complex between A33, A34, and B5. Importantly, no B5 was detected in the absence of A33 and A34 (Fig. 6B), confirming the specificity of the three-protein complex (Fig. 6A). Furthermore, a band corresponding to B5 was not precipitated in the presence of A33 and A34 in the absence of BFA (data not shown), confirming that the complex exists in the ER.
Two-step coimmunoprecipitation of an A33-A34-B5 complex. HeLa cells were infected with vTF7-3 at an MOI of 5 and transfected with the indicated plasmids in the presence of AraC and BFA. The following day, cells were lysed and lysates incubated with rat anti-HA MAb followed by protein G-dynabeads. Bound complexes were eluted with an HA peptide and subsequently incubated with mouse anti-V5 MAb followed by protein G-dynabeads. Immune complexes (A) and cell lysates (B) were analyzed by Western blotting with rat anti-B5 MAb followed by HRP-conjugated donkey anti-rat, rat anti-HA MAb followed by Alexa Fluor 647-conjugated donkey anti-rat antibody, mouse anti-V5 MAb followed by HRP-conjugated donkey anti-mouse antibody, and mouse anti-actin MAb followed by Alexa Fluor 750-conjugated goat anti-mouse antibody. The masses in kilodaltons and positions of marker proteins are shown on the left of the images.
Coexpression of B5, A33, and A34 results in their localization to the juxtanuclear region.The previous result detected a three-protein complex with A33, A34, and B5 that occurs in the presence of BFA (Fig. 5C and D) and is weakly detectable in its absence (Fig. 5A and B). We theorized that the complex initially occurs in the ER, disassociates in a post ER compartment, and may be required for glycoprotein localization to the site of wrapping. To begin to understand if complex formation has a role in the localization of A33, A34, and B5, we compared the localization of B5 in the presence and absence of the A33-A34 BiFC constructs, and the localization of the BiFC interaction in the presence and absence of B5. When expressed in the absence of one another, both B5 and the BiFC constructs were predominantly localized diffusely throughout the cell, giving a lacy fluorescence pattern, indicative of the ER localization seen previously (22, 26, 27, 32, 33, 43, 44). However, when B5, A33, and A34 were coexpressed, there was an obvious change in localization, with the B5 and BiFC signal predominantly colocalized to the juxtanuclear region of the cell (Fig. 7A), indicative of the site of wrapping. There was also pronounced colocalization of the constructs in small punctate particles distributed throughout the cell. To determine whether the members of the three-protein complex localized to the trans-Golgi network where wrapping would occur during infection, we also stained for a TGN resident protein, golgin-97 (26, 45). We compared the colocalization of the A33 and A34 BiFC constructs with golgin-97 in the presence and absence of B5. When A33 and A34 were coexpressed in the absence of B5, similar to what was described previously (Fig. 7A), the proteins were diffusely localized throughout the cell, indicative of the ER, and were not predominately colocalized with golgin-97 (Fig. 7B). However, when A33 and A34 were coexpressed in the presence of B5, the proteins were predominantly colocalized with golgin-97 at a juxtanuclear region, indicating that expression of the members of the three-protein complex localizes the proteins to the TGN (Fig. 7B). Overall, these results suggest that coexpression of A33, A34, and B5 is necessary for localization of these proteins to the site of wrapping, which would be required for their incorporation into the viral envelope.
Localization of B5 and BiFC constructs. HeLa cells grown on coverslips were infected with vTF7-3 at an MOI of 0.5 and transfected with the indicated plasmids in the presence of AraC. The following day, coverslips were fixed, permeabilized, and stained with (A) rat anti-B5 MAb followed by Alexa Fluor 594-conjugated donkey anti-rat antibody (red) or (B) mouse anti-golgin 97 MAb followed by Alexa Fluor 594-conjugated donkey anti-mouse antibody (red). Coverslips were mounted on microscope slides with ProLong Gold Antifade Reagent with DAPI to stain DNA (blue). Stained coverslips were visualized via fluorescence microscopy. Overlap of B5 or golgin 97 (red) and BiFC (green) is shown in yellow. Scale bar 10 μm.
To try to understand which of the three glycoproteins might be important for the localization seen in Fig. 7, different combinations of A33, A34, and B5 were overexpressed in HeLa cells using the T7 expression system in the presence of AraC. Cells were subsequently fixed, permeabilized, and stained with anti-B5, anti-HA, and anti-V5 antibodies (Fig. 8A). When individually expressed, both A34 and B5 were localized diffusely throughout the cell, indicative of ER staining. However, when A33 was independently expressed there was an increase in signal at the juxtanuclear region, indicating that A33 on its own is capable of trafficking and accumulating at the site of wrapping (Fig. 8A). These results parallel what has been described previously (46). When we coexpressed in pairwise combinations of A33, A34, and B5, the proteins were diffusely localized throughout the cell (Fig. 8A). However, co-expression of all three proteins resulted in a change in localization, with A33, A34, and B5 predominantly colocalized to the juxtanuclear region of the cell. This localization is similar to the localization visualized using the BiFC constructs (Fig. 7A), verifying that coexpression of B5, A33, and A34 is necessary for localization of the proteins to the site of wrapping.
Localization of A33, A34, and B5. HeLa cells grown on coverslips were infected with vTF7-3 at an MOI of 0.5 and transfected with the indicated plasmids in the presence of AraC. The following day, coverslips were fixed, permeabilized, and stained with (A) rat anti-B5 MAb followed by Alexa Fluor 594-conjugated donkey anti-rat antibody (red), rabbit anti-HA antiserum followed by Cy2-conjugated donkey anti-rabbit antibody (green), mouse anti-V5 MAb followed by Alexa Fluor 647-conjugated donkey anti-mouse antibody (white), or (B) rabbit anti-GFP antiserum (GFP 6556) followed by Cy2-conjugated donkey anti-rabbit antibody (green). Overlap of A33 (green) and B5 (red) is shown as yellow; overlap of A33 and A34 (white) is shown as purple; overlap of A34 and B5 is shown as teal; and overlap of A33, A34, and B5 is shown as blue. Coverslips were mounted on microscope slides with ProLong Gold Antifade Reagent. Stained coverslips were visualized via fluorescence microscopy. Scale bar 10 μm.
It is possible that the addition of the BiFC fragment to A33 and A34 might result in mislocalization of the proteins. To verify that YFPN-A33HA and YFPC-A34V5 localize similarly to A33-HA and A34-V5 (Fig. 8A), we overexpressed these constructs in HeLa cells in the presence of AraC and stained with a polyclonal anti-GFP antibody that recognizes both YFPN and YFPC. Staining with this antibody revealed that YFPN-A33HA and YFPC-A34V5 localized diffusely throughout the cell, while YFPN-A33HA was also present in the juxtanuclear region of the cell (Fig. 8B). This staining is similar to what was seen for the BiFC fluorescence of the homodimer pairs (Fig. 4A) and for both A33-HA and A34-V5 (Fig. 8A).
Together, our data suggest a model where glycoprotein localization and trafficking are dependent on the coexpression of the other glycoproteins (Fig. 9). When individually expressed, A33 is the only glycoprotein capable of export and trafficking to the Golgi, albeit inefficiently. However, this localization is lost when coexpressed with either B5 or A34. Importantly, efficient transport of A33, A34, and B5 from the ER to the TGN requires expression of all three of the glycoproteins (Fig. 9). The means by which this transport is accomplished is likely through a set of glycoprotein interactions, of which interactions have been reported between B5 and A34 and B5 and A33, and in this paper between A33 and A34, and an A33-A34-B5 three-protein complex (Fig. 1 and 5) (22, 26, 27, 29, 31–34). These results suggest that efficient glycoprotein transport is dependent on a complex network of glycoprotein interactions.
Proposed model of glycoprotein localization during infection. At late times of VACV infection, glycoproteins A33 (pink hexagon), A34 (blue rectangle) and B5 (green circle) are cotranslated in the rough ER and trafficked to the site of wrapping at the TGN. At the TGN, they are incorporated into the wrapping membranes of IEV and their ectodomains are exposed after exiting the cell to become EV. Localization of A34 to either the ER or the TGN is dependent on coexpression of A33 and B5 and the reported interactions that occur in the ER among these glycoproteins. The size of the arrow depicts the efficiency with which the glycoproteins are exported from the ER to the TGN.
DISCUSSION
We have uncovered a previously unreported interaction between A33 and A34. Furthermore, we have shown that these two glycoproteins form a complex with a third glycoprotein, B5. Multiple publications have looked at interactions among the IEV/EV proteins to determine their role in IEV formation, but there is still no clear understanding of the events that occur at the TGN to wrap IMV to produce IEV (12, 14, 22, 26, 27, 29, 31–34, 47–51). Pertinent to the work presented here, B5 has been shown to interact with both A33 and A34 (22, 26, 27, 29, 31–34). In the absence of either A34 or B5, the remaining protein is predominantly localized to the ER (27, 31, 52), suggesting that their interaction is required for ER exit and subsequent TGN localization in infected cells. In contrast, expression of B5 in uninfected cells localizes to the Golgi (44, 46), indicating that B5 is capable of ER exit and Golgi localization in the absence of additional viral proteins. Therefore, it seems likely that during infection, B5 interacts with either an early viral protein or a cellular protein that is induced upon poxvirus infection, which prevents the exit of B5 from the ER. This would suggest that the A34-B5 interaction acts to prevent this aberrant interaction and expedite the export of B5 and TGN localization. We considered that the aberrant interaction could be with A33 because B5 and A33 have been shown to interact, but in the absence of A34, A33 localizes to a juxtanuclear compartment in infected cells and is incorporated into EEV, albeit inefficiently (Fig. 3) (49). The cytoplasmic tail of A33 contains a putative diacidic motif which, in other proteins, interact with COPII for ER export (53–55) and may facilitate the ER export and localization to the Golgi of A33 in the absence of the other viral glycoproteins (Fig. 4, 8, and 9) (46). In the absence of A34, the inefficient localization of A33 to the juxtanuclear region may be due to an interaction with B5 that makes them unable to exit the ER. Indeed, the expression of B5 with A33 appeared to reduce the amount of A33 localized to the juxtanuclear region, with more signal found throughout the cell (Fig. 8 and 9). Similarly, coexpression of A34 and A33 changed their localization, with more A34 localizing to a juxtanuclear region and more A33 found throughout the cell. The coexpression of all three glycoproteins afforded the greatest colocalization of their signal to the juxtanuclear region (Fig. 7 and 8), demonstrating the interdependency of these three glycoproteins on proper subcellular localization and ultimately incorporation into the EV membrane (Fig. 9). Furthermore, both A33 and B5 have been reported to interact with F13 (56). As F13 is found exclusively on the cytoplasmic side of the membrane, a direct interaction with these two glycoproteins would likely be through their cytoplasmic tails. Future experiments will determine if F13 is part of the A33-A34-B5 complex and if these proteins are complexed on the surface of EV.
A33 and A34 have several similarities. Both are type II transmembrane proteins with a predicted short N-terminal cytoplasmic tail and a C-type lectin ectodomain (19, 23, 24). In addition, both have been shown to interact with B5 and play a role in EV cell binding (26, 27). Indeed, considering that A33 and A34 reside next to each other in the genome and gene duplication in poxviruses has been shown to be a frequent occurrence (57–60), it is plausible that they arose from a gene duplication event of a common ancestral gene and retain partially overlapping functions during infection. Considering this possibility, it is not surprising that A34 forms homodimers (Fig. 4), as A33 readily forms homodimers through both intermolecular disulfide bonding and an undefined dimerization domain (23, 24, 61). A33 homodimers form in the ER where disulfide bonding takes place. While there is no structural information for A34, the cytoplasmic BiFC tags indicate that homodimerization is between two adjacent molecules in much the same way as for A33. In addition, the interaction between A33 and A34 is likely between homodimers of these two proteins, such that they form a tetramer. The A33-A34 interaction is between two adjacent molecules on the same membrane based on the cytoplasmic BiFC fluorescence, as intermembrane interactions would not bring the two parts of BiFC close enough to interact. Our data suggest that dimerization may be required for the interaction between A33-A34 and B5. The ectodomains of A33 (residues 57 to 185) and A34 (residues 36 to 168) interact (Fig. 1). Similarly, residues 1 to 130 of A34 are sufficient for the interaction (Fig. 2), indicating that the A33 interaction site on A34 is contained within residues 36 to 130. This region overlaps with the B5 interaction site on A34 (residues 80 to 130) (27) and suggests that interaction with this region may be mutually exclusive to either A33 or B5. Therefore, the dimerization of A34 (Fig. 4) may be required to form a complex with both A33 and B5. Alternatively, a single dimer of A33 could interact with both A34 and B5. Further work is clearly needed to understand the stoichiometry of the A33-A34-B5 complex.
In addition to poxviruses, several other viruses are enveloped internally using post-ER membranes, including, bunyaviruses, coronaviruses, herpesviruses, and rubella virus (62–65). The internal site of envelopment requires the accumulation of viral glycoproteins at these sites. Strikingly, all of these viruses express at least 2 glycoproteins that interact for their efficient export out of the ER and subsequent localization to the internal site of envelopment. One of the best studied examples is Gn and Gc encoded by bunyaviruses. Gn and Gc are type I transmembrane glycoproteins that form a heterodimer for their localization to the Golgi. When expressed separately, Gn localizes to the Golgi by targeting and retention signals in its transmembrane domain, whereas Gc localizes to the ER by means of a lysine-based ER retention signal (66–68). It is only when both glycoproteins are coexpressed and interact does Gc localize to the internal site of envelopment. This situation appears to be somewhat mimicked by A33 and A34. When expressed individually, A33 localizes to a juxtanuclear structure characteristic of the Golgi and A34 localizes to the ER. When all three glycoproteins are expressed, they predominantly colocalize in the juxtanuclear region with the TGN marker golgin-97 (Fig. 7). A Golgi-targeting signal in the cytoplasmic tail of B5 has been shown to be responsible for its localization to the Golgi (44). Besides a putative diacidic motif, A33 is not predicted to have any targeting motifs. Similarly, A34 does not have a di-arginine motif that has been shown to retain type II membrane proteins in the ER (69). A34 does have a membrane-proximal di-lysine, but these types of retrieval signals have not been shown to work for type II membrane proteins. In lieu of a bona fide retention/retrieval motif, it is plausible that folding of A34 is inherently slow and accounts for its inability to exit the ER. Interaction among these three proteins initially occurs in the ER (Fig. 1 and 5), therefore coexpression of A33 and B5 may expedite A34 folding and thus ER export. In addition, viral glycoproteins and the receptors they bind are both produced in the ER. Thus, viruses have evolved mechanisms to ensure their glycoproteins do not prematurely bind their receptors in the ER before virions can be produced (70–72). Formation of the A33-A34-B5 complex could be a mechanism to ensure these glycoproteins do not prematurely bind to their receptors and aggregate in the ER. Future studies will focus on determining the exact role of the three-protein complex during viral infection.
MATERIALS AND METHODS
Cells.HeLa and BSC-40 cells were obtained from ATCC and maintained in Dulbecco’s modified Eagle’s media (DMEM) supplemented with 10% fetal bovine serum (FBS).
Viruses.Recombinant viruses vTF7-3, vΔA34R, and viruses containing C-terminal truncations in A34 have been described previously (15, 27, 35). Expression plasmids pBMW-118 and p118-A34R that encode A34 with 500 bp of flanking sequence have been described previously (32, 73). To create pA34R-V5-118 and pV5-A34R-118, overlapping primers that added the coding sequence for the V5 epitope to the C or N terminus of A34R were used to produce a PCR product with 500 bp of flanking sequence, which were subsequently ligated into pBMW-118 (vA34R-V5-118 and vV5-A34R-118, respectively) using standard cloning techniques. To construct vA34R-V5 and vV5-A34R, cells infected with vΔA34R were transfected with either pA34R-V5-118 or pV5-A34R-118 to allow for recombination. Plasmid pBMW-118 contains the coding sequences for HcRed and guanine phosphoribosyltransferase (GPT) under the control of viral promoters. Trans-dominant selection was used to select for single-crossover recombinants that expressed HcRed and were resistant to xanthine-guanine phosphoribosyltransferase selection (6, 74). Single-crossover recombinants were further plaque purified in the absence GPT, screened, and amplified, resulting in recombinant viruses vA34R-V5 and vV5-A34R. The A34R locus was sequenced in all recombinant viruses to verify the sequence. To create vTF7-3ΔB5, cells infected with vTF7-3 were transfected with a plasmid containing mOrange in place of the B5R coding sequence under the natural B5R promoter with 500 bp of flanking sequence to allow for recombination. Plaques expressing mOrange were selected and plaque purified. The entire B5R locus was sequenced to verify the replacement of B5R with mOrange.
Plasmid constructs.Constructs pV5-A34R, pA34R-V5, and pA34R36-168, all in pcDNA3.3 (Invitrogen), have been previously described (27). To construct A33-HA lacking the transmembrane and cytoplasmic tail domains (pA33R57-186-HA), the sequence encoding amino acid residues 57 to 185 followed by an HA epitope tag and a stop codon, which was preceded by the signal peptide sequence of vesicular stromatitis virus G (VSVG), was amplified by PCR and inserted under the control of the T7 promoter in pcDNA3.3. BiFC constructs containing residues 1 to 158 of YFP (pYFPN) and residues 159 to 238 (pYFPC) in pcDNAI/Amp were a kind gift from Catherine Berlot and previously described (75). To construct YFPN-pA33HA/YFPC-pA33HA and YFPN-pA34V5/YFPC-pA34V5, pA33-HA and pA34-V5 were PCR amplified to add an in-frame BamHI restriction site at the 5′ end and a BglII restriction site at the 3′ end. Amplified fragments were cloned using the TOPO TA cloning kit (Invitrogen) and sequenced to verify their integrity. Verified fragments were digested with BamHI and BglII and subcloned into the BglII site of pYFPN and pYFPC. All plasmid constructs were verified by sequencing prior to use.
Immunoprecipitation.For plasmid transfections, HeLa cells were infected with vTF7-3 or vTF7-3ΔB5 at a multiplicity of infection (MOI) of 5 and transfected 2 hpi with the indicated constructs using Lipofectamine (Invitrogen), according to the manufacturer’s instructions. Where stated, 40 μg/ml cytosine arabonoside (AraC; Sigma) and/or 5 μM brefeldin A (BFA; Sigma) was added and present throughout the infection and transfection. For virus infections, HeLa cells were infected with the indicated viruses at an MOI of 5 and incubated overnight at 37°C. The following day, cells were lysed in radioimmunoprecipitation (RIPA) lysis buffer (0.5× phosphate-buffered saline [PBS], 1% Triton X-100, 1% NP-40, 0.1% sodium dodecyl sulfate, 0.2 mM phenylmethlsulfonyl fluoride, and mini-cocktail protease inhibitor tablet [Roche]) as previously described (27). For protein G-agarose, lysates were incubated with mouse anti-V5 monoclonal antibody (MAb) (Invitrogen), mouse anti-A33 MAb (10F10) (76), or rat anti-hemagglutinin (HA) MAb (Roche). Immune complexes were resuspended in protein loading buffer and analyzed by Western blotting as described in the figure legends. For protein G-dynabeads, lysates were incubated with protein G magnetic dynabeads complexed to mouse anti-V5 MAb, or mouse anti-GFP MAb (GFP 1218; Abcam). Immune complexes were resuspended in protein gel sample buffer and analyzed by immunoblot analysis as described in the figure legends. For two-step coimmunoprecipitation, cells were processed as described above and supernatants were incubated with protein G-dynabeads complexed to rat anti-HA MAb for 1 h at 4°C, washed three times with lysis buffer, and eluted with 100 μg HA peptide (Sigma). Supernatants were subsequently incubated with protein G-dynabeads complexed to mouse anti-V5 MAb for 1 h at 4°C, washed three times with lysis buffer, and resuspended in protein gel sample buffer. Samples were visualized by immunoblot analysis as described in the figure legends. The following antibodies were used: rat horseradish peroxidase (HRP)-conjugated anti-HA MAb (Roche), rat anti-HA high affinity MAb (Roche), mouse HRP-conjugated anti-V5 MAb (Invitrogen), mouse anti-V5 MAb (Invitrogen), rabbit anti-A33 antiserum (NR-628; BEI resources), rabbit anti-GFP antiserum (GFP 6556; Abcam), rat anti-B5 MAb (9), and mouse anti-actin MAb (Sigma). HRP-, Alexa Fluor 594-conjugated donkey anti-rat, and Alexa Fluor 647-conjugated donkey anti-mouse, anti-rat, and anti-rabbit antibodies were purchased from Jackson ImmunoResearch Laboratories. Alexa Fluor 750-conjugated goat anti-mouse antibody was purchased from Invitrogen. HRP was detected using chemiluminescent reagents (Pierce) by following the manufacturer’s instructions. The fluorescent and chemiluminescent signal was captured using a Kodak Image Station 40000mmPro (Carestream Health Inc.).
Immunofluorescence.HeLa cells on coverslips were infected with vTF7.3 at an MOI of 0.5 and transfected with the indicated constructs 2 hpi using Lipofectamine according to the manufacturer’s instructions. The following day, cells were fixed in 4% paraformaldehyde and permeabilized in 0.1% Triton X-100. Coverslips were subsequently incubated with rat anti-B5 MAb followed by Alexa Fluor 594-conjugated donkey anti-rat antibody (Jackson), rabbit anti-HA antiserum (Sigma) followed by Cy2-conjugated donkey anti-rabbit antibody (Jackson), mouse anti-V5 MAb followed by Alexa Fluor 647-conjugated donkey anti-mouse antibody, mouse anti-golgin 97 MAb (Invitrogen) followed by Alexa Fluor 594-conjugated donkey anti-mouse antibody (Jackson), or rabbit anti-GFP antiserum (GFP 6556) followed by Cy2-conjugated donkey anti-rabbit. Coverslips were mounted in ProLong Gold Antifade Reagent with 4′,6-diamidino-2-phenylindole (DAPI) (Invitrogen). Cells were imaged using a Leica DMIRB inverted fluorescence microscope with a cooled charge-coupled device (Cooke) controlled by Image-Pro Plus software (Media Cybernetics). Images were compiled and minimally processed using Photoshop (Adobe).
ACKNOWLEDGMENTS
We thank Bernard Moss for providing vΔA34R and vTF7-3. We also thank Catherine Berlot for providing constructs containing the C and N terminus of YFP and Jay Hooper for providing the A33VACV MAb-10F10. The following reagent was obtained through BEI Resources, NIAIID, NIH: polyclonal anti-vaccinia virus (WR) A33R protein, (antiserum, Rabbit), NR-628.
This work was supported in part by NIH grants AI067391 and AI117105. S.R.M. was supported by award number T32AI118689 from the National Institute of Allergy and Infectious Diseases.
FOOTNOTES
- Received 23 December 2019.
- Accepted 23 December 2019.
- Accepted manuscript posted online 15 January 2020.
- Copyright © 2020 American Society for Microbiology.
