Background
The
Vaccinia virus (VACV), the prototype of the
Poxviridae family, is a large double-stranded DNA virus with a brick-shaped enveloped particle and a genome that ranges from 178 to almost 200 kb, depending on the virus’ strain, which codes for approximately 200 ORFs [
1‐
3]. Up to 50% of the virus' genome codes for immune evasion-related and/or host-interaction genes, a feature shared by many other poxviruses. Western Reserve (WR) is a virulent VACV strain originally derived from New York City Board of Health (NYCBH) virus—a vaccine strain used during the smallpox eradication campaign [
4,
5]. Multiple passages of the NYCBH virus in diverse cell types and animals like rabbits and mice induced an array of mutations in the virus genome, making the resulting WR strain considerably pathogenic to a wide range of mammals and unsuitable to be used as vaccine against smallpox [
2,
6]. Nonetheless, the WR strain became a model poxvirus commonly utilized in most studies looking at the biology of poxviruses. The Modified virus Ankara (MVA) strain, on the other hand, is an attenuated strain that is avirulent for humans and other mammals as the virus is unable to fully replicate in most mammalian cells [
7‐
9]. MVA was obtained after more than 570 passages of its parental strain in primary chicken embryo fibroblast cells (CEFs), culminating in the development of an immunogenic and highly attenuated strain. As a vaccine against smallpox, clinically tested, the virus was well tolerated by vaccinees and demonstrated that it can be used to confer cross-protection against Variola virus (the etiological agent of smallpox) [
10]. Complete genome sequence analysis of MVA revealed six major deletion sites on its genome as well as numerous point mutations—a consequence of the cell passaging—encompassing a total reduction of 31 kb on its coding capacity when compared to the parental virus, the Chorioallatoid Ankara strain [
11].
All members of
Poxviridae family replicate entirely in the cytoplasm of the host cell and depend largely on viral proteins for their replication; however, these complex viruses also exploit numerous cellular pathways to ensure their replicative success. The VACV replication occurs in close association with the endoplasmic reticulum (ER) [
12], the largest membranous organelle of most eukaryotic cells, where most nascent proteins are folded. Additionally, the ER is essential for the balance of the intracellular calcium and the organelle plays a key role in the lipids and sterols biosynthesis [
13]. Due to its participation in many important cell processes, the ER is sensitive to perturbations in the cellular homeostasis triggered by stresses from endogenous or exogenous origins. Sources of perturbation includes chemical damage, genetic mutations, nutritional starvation, cell differentiation and, of course, infection by different intracellular pathogens [
14‐
18]. The resultant disturbances in cell homeostasis can alter nascent protein formation, leading to the accumulation of unfolded or misfolded proteins inside the ER, a condition known as ER stress. This condition triggers ER stress responses, which are generally known as unfolded protein responses (UPR).
There are three major signaling pathways that are part of the UPR. These pathways are controlled by the following ER-resident sensor proteins: the inositol-requiring protein 1 alpha (IRE1α); the activating transcription factor 6 alpha (ATF6α) and the protein kinase RNA-like ER kinase (PERK). Together, these pathways are responsible for surveillance of the ER stress. Depending on the cell type and stimuli as well as of the duration of the stress condition, the outcome of these signals may result either in the recovery of protein homeostasis or cell death [
19‐
21]. These sensors are able to attenuate protein translation but also increase the expression of ER chaperones and ERAD components which, in turn, culminate in the increment of the ER capacity and/or reduction of the ER demand, restoring the organelle homeostasis.
Given that endoplasmic reticulum is an organelle exploited by VACV for replication, we examined the ER homeostasis during the virus infection. Here, we describe how virulent or attenuated strains of VACV can affect the UPR signaling pathway and the importance of UPR components for virus multiplication in murine embryo fibroblast cells.
Discussion
Cellular stress responses encompass critical mechanisms that prevent cells from accumulating macromolecular damage so that metabolism equilibrium is attained and efficient host defenses against pathogens are mounted. Counteracting such cell strategies, poxviruses have evolved numerous mechanisms to cope with cellular stress responses [
34]. In this work, we analyzed the modulation of the UPR signaling during VACV infection and the subtle impact of UPR on VACV in vitro infectivity. To that end, we evaluated the unfolding of UPR during infections by replicative and non-replicative VACV strains—WR and MVA, respectively. Upon activation, the ER stress sensor ATF6α transcription factor undergoes dissociation from BiP, exposing a signal to relocate the sensor to the Golgi network where it is cleaved by S1P and S2P proteases [
35] resulting in the release of an amino-terminal fragment. The resultant ATF6α fragment translocates to the nucleus, where it promotes expression of chaperones and genes that code for transcription factors, which play important roles in ER stress, induced apoptosis and proteostasis [
36‐
38]. Our results indicate that VACV infections activate ATF6α-mediated stress-related signaling and that such responses may be beneficial for virus replication.
We used reporter assays to determine nuclear translocation of ATF6α and measure their transcriptional activity upon VACV infection. We detected ER-to-nucleus translocation of ATF6α in either VACV-WR or MVA infected-cells, and this was similar to what was observed in stress induced, tunicamycin-treated positive controls. Likewise, we observed the robust activity of ATF6α-controlled luciferase reporter gene over the course of VACV infections, which was distinguishable from mock-treated cells at late stages of infection. The ATF6α activation upon infection has been observed for other large DNA viruses as well [
18,
39,
40]. Although both VACV-WR and MVA are able to induce ATF6α activity, the kinetics and maximum levels of activation are different, suggesting that activation of the UPR-ATF6α branch at early/intermediate stages of viruses’ infection may be driven by WR-encoded genes that are possibly defective or absent in the MVA genome [
9]. These results may suggest an impact of viral morphogenesis on the activation of ATF6α. The MVA strain infection produces atypical IMVs in non-susceptible cells and cannot undergo subsequent steps in morphogenesis. The blocking in MVA morphogenesis includes the IMV membrane wrapping, which is known to exploit the Ras-related protein Rab-1A [
41]. The puzzle of membranes acquisition by VACV virions has been solved in the last years, shedding light on the crucial role of ER in the VACV life cycle [
42‐
45].
Interestingly, whereas ATF6α was significantly activated during VACV replication, our data suggested that the endoplasmic reticulum (ER) stress-induced activation of IRE1α was attenuated on a specific, crucial point: the IRE1α-mediated XBP1 mRNA splicing was down regulated in VACV-infected cells. During ER stress, IRE1α undergoes dissociation from BiP and BAX inhibitor 1 [
46,
47] triggering its dimerization, autophosphorylation and activation of its endonuclease activity [
48‐
50]. The IRE1α nuclease domain has homology to RNase L and its activation causes splicing of a residual intron (26nt) in the XBP1 mRNA, resulting in a more stable and active form of the XBP1 protein. We observed a drop of IRE1α mediated XBP1 mRNA splicing during VACV infection by RT-PCR-RFLP and qPCR (not shown). This VACV infection-associated decrease in XBP1 mRNA splicing is so remarkable that virus infection is able to counteract the tunicamycin-induced XBP1 splicing, considering that the drug is a potent inducer of UPR activation and IRE1α activation. Importantly, it has been demonstrated that increased ATF6α expression results in diminished IRE1α production in human cells, whereas the shut-off of ATF6α causes increased XBP1 mRNA splicing and IRE1α activity, suggesting that IRE1 is controlled by an ATF6-dependent switch off mechanism during ER stress [
51]. Our results fit well into this model, as the VACV-induced increase in ATF6α activation correlated with a decrease in XBP1 mRNA processing. Furthermore, the XBP1 transcription factor is essential for sustained cytokine production induced by Toll-like receptors [
52] as well other major immune processes [
53‐
56]. Therefore, it seems plausible this virus has adapted evolutionary strategies to down regulate XBP1 mRNA splicing as a way to cope with the host innate responses. Indeed, it has been shown that a VACV protein encoded by the E8L ORF is able to bind to the protein disulfide isomerase-associated 6 (PDIA6) factor, limiting IRE1 activation [
57,
58]. However, whether XBP1 production is directly affected by a virus-coded protein or its diminished expression is an indirect function of the increase in ATF6α activity remains to be determined.
One interesting observation that we have made in this work was that UV-inactivated viruses were not capable to stimulate ATF6α transcriptional activity using a luciferase reporter system, suggesting the requirement of VACV biosynthesis to trigger UPR. However, when infections were carried out with functional viruses but in the presence of a viral DNA replication inhibitor (Ara,C), residual ATF6α transcriptional activity was detected. UV treatment of viruses, especially DNA viruses, leads to a block in both transcription and DNA replication [
59], whereas AraC blocks VACV DNA replication and abrogates late gene expression [
60]. Because we observed continuing ATF6α transcriptional activity in AraC treated cells infected with VACV and no transcription of the luciferase reporter when UV-inactivated viruses were inoculated into cells, we concluded that early viral protein synthesis is able to induce ATF6α activation. Nonetheless, viral DNA replication and subsequent protein synthesis are required for optimal ATF6α activity.
The UPR-targeted genes tested in our study are up regulated during VACV infection in an ATF6α-dependent manner. This conclusion was based upon the fact that molecules such as BiP and PDIA4 had their mRNA levels significantly reduced in ATF6α knockout cells upon infection. Exception was the apoptotic-related CHOP transcription factor, for which mRNA levels increased consistently in the absence of ATF6α. Indeed, Klymenko and co-workers [
61] have demonstrated that expression of ATF6α causes diminished CHOP transcription in ER-stressed alveolar epithelial cells. Nonetheless, the ER stress responses are known to be highly redundant and there is extensive cross talking among different pathways. Therefore, none of these genes are exclusively controlled by ATF6α [
62], although most of them contain UPRE sequences at their promoter sites.
Up to this point we demonstrated that VACV infections induce UPR through the activation of ATF6α, which in turn may modulate other branches of the ER stress signaling, particularly the IRE1 pathway. By asking whether ATF6α activation plays a role in VACV replication, we investigated this event in ATF6α KO cells and compared with WT cells. When we analyzed VACV-WR growth curves in wild typed MEFs or in ATF6α KO fibroblasts we did not see important differences in virus yield (IMV production, which is the virus particle evaluated is such experiments).On the other hand, virus plaque sizes in the absence of ATF6α were consistently smaller than those seen on WT fibroblasts. Moreover, CPE observed on ATF6α KO cells seemed attenuated when compared to normal cells, and cells deficient in ATF6α presented significantly reduced viral gene expression. Interestingly, genetic complementation of the expression of ATF6α transcription factor in deficient cells was able to boost virus yield as well as to rescue CPE to wild type virus patterns (not shown). One possible explanation to why no differences were seen in virus yield in the growth curve experiments is that primary MEFs were otherwise infected, different from the transformed ATF6α knockout cells. Therefore, the intrinsic differences between these two cells could have masked possible differences in virus yield. Nonetheless, we saw no differences in virus morphogenesis either in the presence or in absence of ATF6α.
Collectively, our results indicate that VACV induces and exploits ATF6α signaling and the UPR upon infection in order to maximize its replicative success. Nonetheless, the data shows that ATF6α and target genes are not required for virus replication, but may represent a viral strategy to boost virus yield in infected cells. Indeed, the exploitation of host signaling pathways by VACV in order to potentiate virus replication has been frequently described [
41,
63‐
65]. Similar trend of UPR activation and regulation has been described for other viruses such as
Myxoma virus [
18], another member of
Poxviridae family, as well as for other DNA viruses including members of
Herpesviridae [
39] and
Asfaviridae [
40] families. It is also important to mention that the use of UPR by VACV to achieve maximum viral replication is rather selective. We evaluated VACV replication in PERK knockout cells in comparison to PERK WT cells and observed no differences in virus replication parameters, making this branch of the UPR apparently irrelevant for the VACV replicative success (data not shown). Further experimentation, however, will be necessary to thoroughly explore this hypothesis.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.