We reported previously that replacing the entire HD of IBV E with a heterologous sequence eliminated disruption of the secretory pathway in transfected cells, and dramatically reduced the release of infectious virus from infected cells 
. Total particle release was only modestly affected, however, suggesting that the HD of IBV E is important for preventing damage to virions during egress. Here we have shown that a single amino acid in the HD of IBV E (T16) is critical for disruption of the secretory pathway in cells expressing IBV E, but was not required for VLP production. This result suggests that the alteration to the secretory pathway is uncoupled from the role of E in assembly. Additionally, we generated versions of IBV E that adopted either a transmembrane or membrane hairpin topology. Using these mutants, we showed that a transmembrane topology was required for secretory pathway disruption. The residue equivalent to T16 in SARS-CoV E, N15, is predicted to lie in the pore region of a homo-pentamer 
. Studies on a lysine-flanked peptide of the SARS-CoV E HD showed that N15 was important for the ion channel activity of the peptide in planar lipid bilayers 
. Since we found that a transmembrane topology and T16 are required for disrupting the secretory pathway, and both are predicted to be important for ion channel activity, it is certainly possible that the disruption of the secretory pathway is due to the putative channel activity of IBV E. Alteration of Golgi complex structure and disruption of protein traffic occur when the ion balance at the Golgi complex is disrupted 
. While an active ion channel at the Golgi complex could explain our observations, how altering the ion balance of secretory organelles might facilitate release of infectious particles remains unknown. We speculate that the demands of trafficking large virion cargo require the expansion of the Golgi complex cisternae, which may be achieved by changing the luminal ion concentration. Alternatively, a change in luminal environment may inactivate proteases present in the secretory pathway, thus protecting the virions from degradation that could render them non-infectious. The membrane rearrangements observed in CoV-infected cells are likely due at least partially to a disruption in the luminal microenvironment, although syncytia formation also contributes 
. Expression of the E protein in the absence of infection allowed us to assess its contribution to membrane rearrangements directly.
Many viruses encode small membrane proteins that have ion channel activity 
. As a group these proteins are referred to as viroporins. The best studied viroporin is influenza M2, which forms a tetrameric pH-activated proton channel 
. The M2 channel acidifies the interior of the virion during entry to aid in unpacking the genome 
. For some strains of influenza virus, M2 also plays an important role in the secretory pathway where it raises the pH of the trans
-Golgi to prevent the premature activation of the fusion protein 
. Hepatitis C virus (HCV), like CoVs, assembles intracellularly and must navigate the secretory pathway for release. Interestingly, HCV encodes a proton selective viroporin, p7 
. While the exact role p7 is not fully understood, it is important in the assembly and release of HCV virions, and expression of p7 leads to the alkalinization of secretory organelles 
. It is possible that HCV-p7 and CoV E have analogous roles during infection for altering the secretory pathway to promote the release of virions. Viroporins appear to play important roles in the assembly and trafficking of many viruses; understanding their exact role(s) is important as they represent good targets for therapeutic intervention via small molecule inhibitors.
While T16 in IBV E is required for disrupting the secretory pathway, it is not important for virus assembly as judged by VLP production. Our VLP results also suggest that disruption of the secretory pathway is not required for virus egress, since the T16A mutant produced the same level of VLPs as the wild-type E protein. However, the VLP assay does not allow measurement of infectivity, which was greatly reduced for particles released from cells infected with IBV carrying an E protein with a heterologous HD 
. Another difference between infection and the VLP assay is that more particles are produced in a shorter time during infection, it is likely then that the stress on the secretory pathway is much more robust during infection. Thus, the VLP assay may not accurately reflect virion trafficking during infection. To measure the effect of T16 on virion trafficking, assays that measure both the amount, rate, and route of infectious particle trafficking are necessary. A future goal will be to analyze recombinant viruses carrying mutations at T16 with quantitative trafficking assays.
If CoV E is important for the release of infectious particles, why do some CoVs show only a modest reduction in infectivity when E is deleted 
? Moreover, why do we only observe a measurable disruption in the secretory pathway with IBV E and not the E proteins from other CoVs? The answer to these questions may lie in the exact role(s) that the CoV E protein plays for each virus. While the E proteins from different CoVs share a similar domain structure, there is large variation in their primary sequence. Additionally, the requirement of CoV E for the production of infectious virus is not consistent between different CoVs. The E protein of the TGEV is essential for the production of infectious virus 
. However, a version of MHV lacking the E gene can replicate, albeit at a greatly reduced titer 
. Finally, a recombinant version of SARS-CoV with E deleted shows only a modest reduction in infectivity when passaged in cultured cell lines 
. These results suggest that CoV E may have evolved to perform divergent functions in different CoVs. Somewhat surprisingly then, it was reported that the E protein from several different CoVs, including IBV E, could substitute for MHV E during infection 
. Even more striking, when MHV ΔE was passaged, revertants were recovered with a partial duplication of the M gene (consisting of the N terminus and three transmembrane domains but lacking the C-terminal tail) that were able to largely compensate for the lack of E 
. Taken together, these results show that at least some function(s) of the E protein are conserved among CoVs. However, the requirement for its function(s) may vary significantly due to the compensatory action of other viral proteins or differences in cell and tissue types infected. Of all the CoVs whose E proteins were tested here, IBV is the only one with an avian host. The requirements for assembly and release in avian species may be slightly different than in mammals. We tested whether the disruption of the secretory pathway caused by IBV E occurred in DF-1 chicken fibroblasts (cultured at 39°C), and found that the secretory pathway was disrupted similar to HeLa cells (unpublished data). Another potential difference is the cell type in which each virus replicates. Certainly the requirements for virus egress in different tissues could be an important factor. Another possibility is that the compartmental localization of the E proteins may vary in the absence of the other viral proteins and the impact of each CoV E on the secretory pathway could depend on the Golgi subcompartment in which it is localized. This possibility could be addressed by immunoelectron microscopy on cells expressing the various E proteins. There is a notable difference in the ion specificity and channel behavior among the different E protein channels in planar lipid bilayers 
. Unlike the other CoV E channels characterized, the IBV E channel demonstrated rectification, where ions are moved predominately in one direction 
. Additionally, the IBV E channel is insensitive to the small molecule HMA, unlike the other CoV E proteins tested 
. If the ion specificity or activity varies between the CoV E proteins, it could certainly explain the differences in behavior reported here. The best way to study these differences would require electrophysiological measurements using patch clamp analysis on purified Golgi membranes. This approach would allow the direct measurement of the CoV E protein in its natural membrane with the proper post-translational modifications, but will be very technically challenging. One last point is that the sequences of the CoV E proteins are highly variable. Of note, IBV E is significantly larger and contains more polar residues in its HD than the other CoV E proteins (see ). It will be important to determine how these differences relate to the function of the proteins. This could be addressed by determining how chimeric proteins affect the secretory pathway and virus replication.
Previous reports on CoV E protein topology have suggested that it may exist either as a transmembrane protein or as a membrane hairpin with both the N- and C-termini in the cytoplasm. The ability to adopt multiple membrane topologies could be a mechanism to increase the number of protein functions within the constrictions of genome size. Here, we generated mutant versions of IBV E that adopted either a membrane hairpin or transmembrane topology. We found that the transmembrane version of the protein behaved largely like IBV E, with the exception that it was unable to drive VLP production to the same degree. The membrane hairpin version of IBV E was unable to disrupt the secretory pathway or drive VLP production. These data suggest that IBV E largely functions as a transmembrane protein, with no apparent role for the membrane hairpin. However, such conclusions should be drawn with caution. While we determined that ssFLAG-IBV E behaved largely like ssIBV E, addition of the FLAG tag onto the N-terminus of IBV E could have any number of off-target effects, especially when considering the interaction of the E protein with M. We attempted to generate a membrane hairpin using several different strategies, including altering the charge distribution on either end of the HD, extending the N terminus with different tags, and shortening the C-terminus. Our only successful strategy was adding the FLAG tag onto the N-terminus. It should be noted that all reports of CoV E demonstrating that it adopts a membrane hairpin upon expression have been carried out using N-terminally tagged proteins 
. In fact the most recent data on the topology of SARS-CoV E using the untagged protein and antibodies directed to either terminus show that the predominant topology is Nexo
. What remains unclear is if a membrane hairpin plays a role during infection. It is possible that a portion of the E protein adopts a membrane hairpin topology. We did observe a small difference between ssIBV E and IBV E when we quantified the signal from our selective permeabilization experiment. A small amount of CoV E in the membrane hairpin conformation could play a catalytic role during assembly, and while not necessarily required for assembly, it may increase the efficiency of assembly. This would explain why FLAG-IBV E could not support VLP production on its own. This idea could be addressed by developing infectious clones of IBV carrying the topology mutants of IBV E and examining particle production biochemically and by electron microscopy of infected cells. Also of interest is the mechanism for generation of multiple topologies. A transmembrane topology is likely generated through the canonical signal recognition particle pathway like other type III membrane proteins 
, but the generation of a hairpin could involve a different mechanism. One could speculate that a hairpin could be generated through post-translational insertion, possibly directly into the target membrane 
The IBV E protein is a multifunctional viral protein that plays a role in both the assembly and release of infectious virus. The exact mechanism by which the protein alters the secretory pathway to facilitate infectious particle release is still unknown, but may depend on a single amino acid in the HD. Identification of the mechanism will be a big step in understanding the interplay between the secretory pathway and CoV trafficking. Also of interest is how E protein function varies among CoVs and what underlies any difference(s). Understanding these questions will provide insight into both therapeutic approaches to CoV infection and increase our understanding of how CoVs use the host secretory pathway to their advantage.