Murine hepatitis virus (MHV) is one of the best-studied coronaviruses (CoVs) and serves as a valuable model for other (group 2) CoVs such as severe acute respiratory syndrome (SARS) CoV, the causative agent of the 2003 outbreak of SARS (19
). Human CoVs are the second most frequent cause of the common cold. In addition, CoVs cause economically important diseases of cattle, poultry, and pigs. Since the emergence of SARS-CoV, several new human and animal CoVs have been identified, emphasizing the constant occurrence of new, potentially harmful CoVs and the necessity to further investigate their life cycles and interactions with host cells.
The CoV genome consists of single-stranded RNA of positive polarity and encompasses approximately 30,000 nucleotides. CoVs express subgenomic RNAs that encode accessory and structural proteins. The S, M, and E proteins are associated with cellular membranes, while N resides in the cytoplasm. In MHV, the 180-kDa S protein is cleaved by a cellular protease during virus maturation into two subunits that remain noncovalently associated (12
). SARS-CoV S is cleaved during viral entry (20
). The S protein mediates binding to carcinoembryonic antigen-related adhesion molecule 1 (CEACAM1) and angiotensin-converting enzyme 2 (ACE2), the specific receptors for MHV and SARS-CoV, respectively (16
). Accordingly, antibodies against S can neutralize virus infectivity (17
). Furthermore, the S protein induces fusion of the viral envelope with host cell membranes and cell-cell fusion (34
). Spike proteins are synthesized in the cytoplasm, cotranslationally translocated to the endoplasmic reticulum (ER), and subsequently transported to the Golgi apparatus (56
). A fraction of S is eventually transported to the plasma membrane, where it can mediate cell-cell fusion (42
Viral glycoproteins can induce ER stress during infection as a result of incorrect folding or accumulation in the ER lumen (8
). Cells can respond in several ways to reduce the burden imposed by unfolded proteins in the ER, ways that are collectively known as the unfolded-protein response (UPR). Induction of UPR results in transcriptional activation of genes encoding ER-resident molecular chaperones to increase protein-folding activity and repression of protein synthesis. Three distinct branches that orchestrate the UPR have been identified so far (49
). Binding to unfolded proteins by ER chaperones results in activation of protein kinase R-like ER kinase (PERK), activating transcription factor 6 (ATF6), and inositol-requiring enzyme 1 (IRE1). IRE1 mediates the splicing of mRNA encoding the transcription factor X box-binding protein 1 (XBP1), leading to a frame shift and subsequent translation of functional XBP1 protein (60
). The active transcription factor XBP1 (XBP1s) can in turn stimulate different subsets of genes encoding proteins that promote the folding, transport, and degradation of ER proteins. The IRE1/XBP1- and ATF6-dependent branches are specific for ER stress, while the PERK-dependent pathway is shared with other cellular stress responses (38
). Expression of the ER-resident protein Herpud1 is regulated by both the ER stress-specific and the shared branches of the UPR (38
). Herpud1 balances folding capacity and protein loads in the ER and also plays a role in stabilizing cellular Ca2+
). Expression of Herpud1 is strongly induced by ER stress and can be considered a hallmark of ER stress (28
). ER stress can also induce the expression of innate immunity molecules such as chemokines (22
). Chemokines are excreted by infected cells and attract specialized immune cells to sites of infection. Recent microarray analysis of MHV-infected cells showed induction of several ER stress markers, including Herpud1, and several chemokines, including Cxcl2 (58
). Mouse CXCL2 is a functional counterpart of human interleukin 8 and has been associated with the recruitment of inflammatory cells during various diseases (18
). During MHV infection in vivo, CXCL2 mediates antiviral responses, but it has also been implicated in the development of immunopathology (46
). Cells carry out steady-state production of CXCL2 from basal levels of Cxcl2 mRNA, since it is also involved in other cellular processes. The clear coregulation of ER stress genes and chemokine genes during microarray analysis suggested that MHV infection activates ER stress pathways that in turn induce innate immune responses. In addition, it has been shown that the SARS-CoV spike protein induces ER stress and activation of PERK (8
). However, the ATF6 and IRE1 branches were not activated. The authors concluded that SARS-CoV S protein specifically modulates the UPR to facilitate viral replication.
In this study we show that the MHV spike protein, like that of SARS-CoV, induces ER stress. MHV infection resulted in extensive expression of UPR markers such as Herpud1 and XBP1s, while SARS-CoV infection upregulated only a limited set of UPR genes. Both viruses potently induced Cxcl2 mRNA expression during infection in vitro. Despite the significant increase in the Cxcl2 mRNA concentration, no CXCL2 protein could be detected in the cytoplasm or medium of infected cells. From the data we concluded that CoV infection induces ER stress and triggers the innate immune system. However, at that point, translation of these cellular mRNAs is already severely reduced, resulting in no significant CXCL2 or HERPUD1 protein synthesis in infected cells, despite their increased mRNA concentrations.
In this study, it is demonstrated that lytic MHV and SARS-CoV infection can induce ER stress responses and Cxcl2 mRNA induction. We hypothesize that CoV spike protein-mediated ER stress can initiate the transcription of chemokine genes based on the following observations: (i) MHV and SARS-CoV induced ER stress and upregulated chemokine mRNA levels during infection; (ii) ER stress and chemokine mRNA upregulation were both induced only by spike expression, and their expression occurred simultaneously; (iii) ER-retained MHV spike was sufficient to induce Cxcl2 transcription.
The data reported indicate for the first time that in addition to the innate immune sensors that have been described so far, the overload of the ER with viral glycoproteins could also mediate induction of innate immune responses such as transcriptional activation of chemokine genes. Moreover, our data demonstrate that in productively infected cells, CoV-induced translational attenuation contributes to viral evasion of potentially harmful host proteins, such as chemokines. A comparable situation has been described for foot-and-mouth disease virus infection, where type I interferon mRNA is upregulated during infection, but eventually interferon protein production from the increased mRNA levels is prevented (11
The data presented here indicate two possibilities for the induction of Cxcl2. The first possibility assumes that spike-induced ER stress triggers Cxcl2 induction. The second assumes that spike induces ER stress and Cxcl2 independently of each other. Although a direct link between Cxcl2 induction and ER stress has not been demonstrated in this study, their simultaneous expression and spike dependence, as well as previous reports showing that ER stress can be involved in the induction of innate immunity (22
), favor the first model and suggest a link between ER stress and Cxcl2 induction. In support of this, induction of a UPR by tunicamycin also led to Cxcl2 induction (data not shown).
Although our data firmly established the potential of spike accumulation to induce ER stress, other factors during infection could contribute as well. For example, some CoVs assemble their replication complexes on ER membranes and severely limit the size of the ER, thereby making it vulnerable to stress induction (23
). The development of a CoV deletion mutant lacking the S gene will be required to address the proportional contributions of S accumulation to activation of a UPR and Cxcl2 transcription in more detail. Several viral glycoproteins, such as influenza A virus hemagglutinin, hepatitis B virus MHBst, simian virus 5 HE, and adenovirus E3/19K, can specifically induce ER stress and activate innate immunity during infection (39
). Viruses may benefit from UPR induction, since protein chaperones produced in response to ER stress may aid in the folding of viral proteins, while on the other hand, viral proteins could be degraded as a result of the UPR. Several viruses induce only a subset of the UPR branches and are thought to actively modulate particular parts of the UPR so as to regulate it for their own benefit (62
Upregulation of Herpud1 mRNA concentrations and Xbp1 splicing during MHV infection (Fig. 6A and B
) indicated induction of the IRE1 pathway. SARS-CoV, on the other hand, upregulated Herpud1 mRNA concentrations to a lesser extent than MHV (Fig. 6A
) and did not induce significant Xbp1 splicing (Fig. 6B
). These results are consistent with the recent observation that SARS-CoV spike induced a UPR in 293 cells without Xbp1 mRNA splicing or ATF6 transcriptional activity (8
). These findings indicate that SARS-CoV spike expression induces only a limited set of UPR pathways. This implies that SARS-CoV could suppress certain ER stress regulators, as has been shown for hepatitis C virus (62
). Alternatively, the differences may also be explained by the fact that MHV has a shorter life cycle, due to which it imposes more-strenuous conditions on infected cells than SARS-CoV, since viral proteins are synthesized in a shorter time span.
Activation of PERK, ATF-6, and IRE-1 occurs at a posttranslational level and can directly initiate other cellular responses that do not require protein synthesis for their activation, such as p38 mitogen-activated protein kinase (MAPK) (35
). Both SARS-CoV and MHV induce p38 MAPK, the activation of which stimulates virus replication, progeny virus production, and the onset of CPE (1
). One could hypothesize that the UPR activates cellular pathways, such as p38 MAPK, that have a positive effect on the viral life cycle. Indeed, preliminary experiments using a specific p38 inhibitor showed a significant reduction in the level of MHV production during in vitro infection (G. A. Versteeg et al., unpublished data).
Experimental results addressing the significance of viral UPR induction in vivo remain scarce. So far, induction of ER stress and its influence on pathogenicity in vivo have been shown only during infection with a neurovirulent retrovirus (14
). Infection with a virulent virus strain was accompanied by ER stress marker upregulation in the brain, whereas expression of those markers was absent in mice infected with an avirulent strain (14
). The MHV spike protein has been implicated in (neuro)virulence as well (47
), and it will be interesting to investigate whether differences in ER stress induction may also be involved there.
Both the MHV and the SARS-CoV spike protein are sufficient to induce a UPR and CXC-type chemokine upregulation (this work; see also reference 10
). Yet spike proteins can already be detected at 2 to 3 h before the induction of these cellular responses. Therefore, we propose that it is not mere S expression that triggers the induction but rather the accumulation of S protein in the ER. Expression of ts
379 spike induced substantial amounts of Cxcl2 mRNA, yet always less than those induced by wild-type MHV spike (Fig. 5A
). The spike expression levels of the ts
379 mutant are lower than those of the wild-type virus (37
). This is true at both the permissive and the restrictive temperature, although the difference from the wild type is most extensive at the restrictive temperature. These results correlate with the observations presented in Fig. 5A
and may explain the differences observed in Cxcl2 mRNA upregulation. Very low leakiness and reversion frequency (30
) and an absence of syncytia at the restrictive temperature (Fig. 5B
) make the involvement of leaky spike maturation in Cxcl2 induction highly unlikely.
During productive MHV infection in vitro, total protein synthesis drops steeply around 6 h p.i. (O. Slobodskaya and W. J. M. Spaan, unpublished data), while Cxcl2 and Herpud1 mRNA induction starts at that time. This could well explain the absence of translation of the upregulated Cxcl2 and Herpud1 mRNAs. In contrast to infection, ectopic spike expression induced CXCL2 protein, indicating that viral factors other than spike are responsible for the lack of CXCL2 protein production in infected cells. The lack of secreted CXCL2 protein in the medium of SARS-CoV-infected cells, despite a significant increase in mRNA levels, suggests that similar translational inhibition could occur during SARS-CoV infection as during MHV infection. These data suggest that attenuation of cellular translation might be a general feature of productively CoV infected cells and a strategy to prevent cellular synthesis of antiviral proteins.