ATF6 Signaling Is Required for Efficient West Nile Virus Replication by Promoting Cell Survival and Inhibition of Innate Immune Responses

Cells and virus stocks.

Cell lines consisting of immortalized ATF6 and IRE1 knockout mouse embryonic fibroblasts (MEFs) were a kind gift from Randal Kaufman (University of Michigan) (24, 39) and were maintained in high-glucose Dulbecco's modified Eagle medium (DMEM; Life Technologies) supplemented with sodium pyruvate, 10% (vol/vol) heat-inactivated fetal calf serum (FCS; Life Technologies), penicillin/streptomycin (100 IU/ml and 100 μg/ml, respectively; Life Technologies), 200 μM GlutaMAX (Glx; Life Technologies), 1% (vol/vol) nonessential amino acids (NEAA; Life Technologies), and 1 μM β-mercaptoethanol (Life Technologies). Vero cells were maintained in 5% FCS (Lonza), GlutaMAX, and penicillin/streptomycin as indicated above. All cells were grown at 37°C in a 5% CO₂ incubator. WNV_KUN stocks were propagated from an existing MRM61C secondary stock in Vero C1008 cells at 37°C in DMEM supplemented with 0.2% (wt/vol) bovine serum albumin (BSA) for 32 h. Following infection, the virus-containing supernatant was collected and centrifuged at 4,800 rpm for 5 min to remove cell debris and then aliquots were stored at −80°C. Viral titer was determined by plaque assay as described below.

Antibodies and reagents.

Tunicamycin (containing homologues A, B, C, and D) (Sigma) was dissolved in dimethyl sulfoxide (DMSO) (Sigma) to a concentration of 5 mM. Mouse alpha interferon (IFN-α) was sourced from PBL Interferon and diluted to 100 U/μl in phosphate-buffered saline (PBS) and 0.1% bovine serum albumin (BSA). Mouse anti-NS1, anti-NS5, and envelope monoclonal antibodies were a kind gift from Roy Hall (University of Queensland). WNV_KUN-specific rabbit anti-NS3 polyclonal antiserum has been described previously (40). Rabbit anti-eIF2a and rabbit anti-p-eIF2a (Ser51) polyclonal antibodies were purchased from Life Technologies, and rabbit anti-BiP and rabbit anti-actin polyclonal antibodies were sourced from Sigma. Mouse anti-CHOP antibody (Santa Cruz) was a kind gift from Julie Atkins (La Trobe University), and mouse anti-phospho-STAT1 (Tyr701) was purchased from BD Transduction Laboratories.

Plaque assay.

Vero C1008 cells were seeded in DMEM complete medium in 6-well plates and incubated at 37°C overnight. Virus stocks or supernatant samples were diluted 10-fold in 0.2% BSA/DMEM, and cells were infected with 300 μl of stock dilutions (in duplicate) and incubated at 37°C for 60 min. Two milliliters of a semisolid overlay containing 0.3% (wt/vol) low-melting-point agarose, 2.5% (wt/vol) FCS, P/S, Glx, HEPES, and NaCO₃ was added to cells and solidified at 4°C for 30 min. Cells were incubated at 37°C for 3 days, fixed in 4% (vol/vol) formaldehyde (in PBS) for 1 h, and stained in 0.4% crystal violet (with 20% [vol/vol] methanol and PBS) at room temperature for 1 h. Plaques were manually counted, and numbers of PFU per ml were calculated as a measure of infectious virion secretion.

Virus infections and time course assays.

Knockout and control MEFs were seeded into 6-well or 24-well plates in infection medium (6% FCS, 200 μM Glx, and 1% [vol/vol] NEAA) and cultured overnight. Prior to infection, all cells were counted using a hemocytometer, and parallel lines were infected with quaternary WNV_KUN stocks at a multiplicity of infection (MOI) of 10 in cold infection medium. Infections were synchronized on ice for 1 h, with rocking to ensure complete coverage, and then virus uptake was induced by the addition of prewarmed infection medium to a final volume of 1.5 or 0.4 ml (in 6-well or 24-well plates, respectively). Cells were incubated at 37°C for the time specified before collection of RNA, protein, or infected supernatant samples as described below.

Western blotting.

WNV_KUN-infected cells were aspirated in PBS and then lysed in SDS lysis buffer (0.1% SDS, 1 mM EDTA, 50 mM Tris-HCl [pH 8.0]) containing a protease inhibitor cocktail (Astral Scientific) and the phosphatase inhibitors sodium orthovanadate (25 mM), sodium fluoride (25 mM), and β-glycerophosphate (25 mM) (Sigma). Lysates were diluted in SDS loading buffer (Bio-Rad), boiled at 95°C for 5 min, and separated on a Tris-glycine polyacrylamide gel. Proteins were then transferred to Hi-Bond ECL nitrocellulose membrane (Amersham Biosciences), and the membrane was blocked with 5% (wt/vol) skim milk (Diploma) in Tris-buffered saline (TBS) with 0.05% Tween (TBS-T) or 5% (wt/vol) BSA (Sigma) in TBS-T. Primary antibodies were incubated at 4°C with membrane overnight in blocking solution as described above. Following primary incubation, the membrane was washed in TBS-T and then incubated with secondary antibodies conjugated to Alexa Fluor 647 or Alexa Fluor 488 (Life Technologies) in TBS-T at room temperature for 2 h. The membrane was washed twice in TBS-T and then TBS, and proteins were visualized on the Storm fluorescence scanner (Amersham Biosciences) on either the 635-nm or 430-nm emission channel.

Immunofluorescence and CHOP nuclear scoring.

Cells were infected with WNV_KUN at an MOI of 10 and at 36 h postinfection (p.i.) were fixed and permeabilized on coverslips with either acetone/methanol (1:1) at −20°C for 5 min or 4% (vol/vol) paraformaldehyde followed by 0.1% (vol/vol) Triton X-100. Cells were incubated with primary antibodies diluted in 1% BSA/PBS, washed twice in 0.2% BSA/PBS, and subsequently incubated with secondary antibodies conjugated to either Alexa Fluor 488 or 594 (Molecular Probes) in 1% BSA/DMEM at room temperature for 45 min. Cells were washed in PBS and then mounted onto glass slides using Ultramount mounting medium (Fronine). Immunofluorescent staining was visualized on a Zeiss confocal microscope, and pictures were assembled using Adobe Photoshop. For CHOP nuclear translocation quantitation, 150 infected cells for each cell type were scored for nuclear staining by comparison with cytoplasmic CHOP labeling. This numeration was performed in duplicate in replicate experiments.

RNA extraction and qRT-PCR.

RNA was extracted from WNV_KUN-infected cells with TRIzol reagent (Invitrogen) as indicated by the manufacturers. Total RNA was then treated with RQ1 DNase (Promega) at 37°C for 30 min to remove any contaminating DNA. cDNA was synthesized with Superscript III reverse transcriptase (RT) (Invitrogen) with gene-specific reverse primers at 50°C for 50 min. Following heat inactivation at 65°C, RT reactions were diluted 10-fold in diethyl pyrocarbonate (DEPC)-treated deionized water. cDNA levels were quantified using quantitative PCR (qPCR) with Sybr GreenER (Bio-Rad) on an ICycler PCR cycling machine (Bio-Rad). Primers were designed for the internal control GAPDH (glyceraldehyde-3-phosphate dehydrogenase), CHOP, and IFN-β as well as the WNV_KUN genome (primer sequences are available from the corresponding author upon request). Fold induction of each gene was calculated by comparing threshold cycle values (C_T) to those for the internal control GAPDH and mock-infected cells at each time point.

Cell cytotoxicity assays.

Cells were lysed by treatment with 5% Triton X-100 at 37°C for 1 h. Cell viability was then measured by lactate dehydrogenase activity using a cell cytotoxicity kit (Promega) as recommended by the manufacturers. Briefly, lysed cells were transferred to a new 96-well microtiter plate and incubated with the substrate and reagent for 45 min. The reaction was halted by the addition of a stop solution, and enzyme activity was assessed by colorimetric analysis at 490-nm absorbance on a microplate reader (Anthos Lucy 2 luminometer). Each sample was assayed in triplicate.

WNV_KUN production is decreased in ATF6^−/− cells.

We have previously observed that UPR effectors downstream of ATF6 and IRE1 are upregulated during WNV_KUN replication while PERK signaling appears dispensable (38). To investigate the contributions of each ER stress sensor during virus replication, ATF6^−/− and IRE1^−/− knockout MEFs and matched controls were infected with WNV_KUN and RNA, protein, and infected supernatants were collected at 1, 12, 24, 36, and 48 h postinfection for analysis. As shown in Fig. 1A, viral RNA levels in the ATF6 cell lines peaked at 36 h p.i., with no significant differences between the knockout and control cells. However, viral protein expression was reduced at 12 and 24 h p.i. in the ATF6^−/− cells (Fig. 1C), resulting in a modest but significant reduction in virions secreted at 36 h p.i. (Fig. 1E) (P < 0.05). In contrast, no significant differences were observed between the IRE1 knockout and control cells in RNA, protein, or virion production (Fig. 1B, D, and F), suggesting that IRE1 may be dispensable for efficient WNV_KUN replication. Interestingly, total protein levels (as assessed by cellular actin and BiP content) were decreased in all infected cell lines from 24 h p.i. onwards, although this was more pronounced in ATF6^−/− cells (Fig. 1C and D).

Infected knockout and control cells were also fixed and stained with the WNV_KUN viral protein NS1 and the ER chaperone BiP to assess viral replication complex formation and infection efficiency (Fig. 2). All cell lines had a similar infection rate (data not shown) over 12 to 48 h, negating the possibility that the defects observed in ATF6^−/− cells were due simply to an initially lower percentage of infected cells. NS1 subcellular labeling was comparable, and distinct replication complexes were observed in all samples, although a slight decrease in intensity was detected in the ATF6^−/− cells. In addition, we observed no significant differences in the distribution and localization of the WNV_KUN replication complex formation in each of the cell types (Fig. 2), indicating that intracellular replication of WNV_KUN occurs unimpeded in the presence or absence of ATF6 and IRE1.

eIF2α phosphorylation and downstream CHOP activity are upregulated in WNV_KUN-infected ATF6^−/− cells.

Various studies have shown that the three arms of the UPR are interdependent, with cross talk demonstrated between the acute-phase (PERK- and eIF2α-based responses) and the long-term (ATF6 and IRE1 signaling) effectors. Additionally, some of these effectors (such as p58^IPK) are able to directly dephosphorylate eIF2α (29, 41), thus switching the UPR from the initial response of translation inhibition to stress adaptation, inducing the production of chaperones and degradative enzymes (42). We have observed that WNV_KUN skews the UPR toward ATF6 and IRE1, with negligible eIF2α phosphorylation, as cell lines deficient in PERK permit moderate increases in WNV_KUN viral protein and virion production (38). Thus, we hypothesized that WNV_KUN-induced increases in ATF6 and/or IRE1 signaling downregulate PERK signaling, which, in the absence of either of these sensors, would result in the resumption of eIF2α phosphorylation and global attenuation of transcription.

To investigate this further, ATF6- and IRE1-deficient cells were infected with WNV_KUN for 36 h or stimulated with tunicamycin (an ER stress inducer) for 12 h and assessed for eIF2α phosphorylation. As shown in Fig. 3, phospho-eIF2α was increased by tunicamycin treatment in all cell lines. However, in WNV_KUN-infected cells, the ratio of phospho-eIF2α to total eIF2α was higher in the ATF6^−/− MEFs than in the control cells (Fig. 3A, compare lanes 3 and 6). This was quantified by densitometric analyses, which showed that the proportion of phospho-eIF2α (as standardized to total eIF2α levels) was almost 2-fold higher in the ATF6^−/− cells than in the wild-type (WT) controls. This was not observed in the IRE1 cell lines (Fig. 3B). Interestingly, higher baseline levels of phospho-eIF2α in tunicamycin-treated cells were also evident in the ATF6 and IRE1 knockout cells, suggesting an increased propensity for PERK activation despite infection/drug treatment. As phospho-eIF2α is a potent inhibitor of translation, this may also explain the decreased viral protein levels in ATF6^−/− MEFs during early replication (Fig. 1).

To confirm the increased PERK activation in the ATF6^−/− cells, CHOP transcription was investigated in ATF6 wild-type and knockout cell lines. CHOP is potently induced by ATF4, which is translationally activated by phospho-eIF2α. Correlating with increased phospho-eIF2α, a modest increase in CHOP mRNA levels was evident at 36 h p.i. in WNV_KUN-infected ATF6^−/− cells (Fig. 4A). To confirm downstream transcriptional activity, we assessed CHOP subcellular localization in WNV_KUN-infected cells at 36 h p.i. by immunofluorescence. As shown in Fig. 4B, CHOP translocated to the nucleus in most infected cell types, although with different efficacies compared to that of the tunicamycin-treated control. The percentage of cells with nuclear CHOP was quantified and showed that significantly more ATF6^−/− cells accumulated CHOP in the nucleus than did matched controls (Fig. 4C) (P < 0.05). Additionally, CHOP labeling was much higher in ATF6^−/− cells, a trend which correlated with the increased transcription also observed at this time point.

These observations suggest that CHOP activity is enhanced in ATF6^−/− cells in response to infection, which is presumably regulated by WNV_KUN in wild-type cells to maintain efficient replication. We hypothesized that this increased activity would also lead to premature cell death in infected cells via increased transcription of proapoptotic molecules. To investigate this possibility further, virus-induced cytotoxicity was compared in the ATF6-deficient cells and the matched controls after 48 h of infection. Corresponding with increased CHOP transcription and nuclear localization, ATF6^−/− cells also had a significantly lower percentage survival rate after infection with WNV_KUN (Fig. 4D) (P < 0.05). This result may also explain the lower overall protein levels observed in the time course assays (Fig. 1B), as global translation would likely have been inhibited by increased eIF2α phosphorylation as well as reduced cell viability. Thus, we suggest that CHOP-induced increases in proapoptotic effectors in the WNV_KUN-infected ATF6^−/− cells contributed to the increased cell death observed.

Late-phase IFN signaling is partially restored in ATF6^−/− and IRE1^−/− MEFs.

We have previously shown that chemical or viral induction of the UPR can inhibit type I IFN-mediated STAT1 phosphorylation and nuclear translocation and that this effect is partly mediated by the expression of hydrophobic nonstructural proteins (particularly NS4A) (38). Given that ATF6 and IRE1 downstream signaling is potently activated by WNV_KUN and that ATF6-deficient MEFs show decreased replication, we were interested to investigate if either one of these arms is also required for the UPR-dependent, viral-induced inhibition of IFN signaling. Knockout or control cells were infected with WNV_KUN for 24 h and then stimulated with mouse IFN-α for 1 h, and STAT1 phosphorylation was detected by a Western blot in infected and uninfected cells. As expected, a decrease in phospho-STAT1 was observed in infected control cells after stimulation with IFN-α (Fig. 5A and B, compare columns 2 and 4). However, in both ATF6^−/− and IRE1^−/− WNV_KUN-infected cells, we observed a greater amount of phospho-STAT1 (Fig. 5A and B, compare columns 6 and 8), with the phospho-STAT1 levels being similar in mock and infected cells. Interestingly, higher phospho-STAT1 levels were observed in unstimulated, infected ATF6^+/+ cells than in the knockouts; whether this is due to the increased replication and, hence, immune detection was not able to be ascertained. As MEFs are immunocompetent, we were also interested in determining if this restoration may be due to differences in IFN-β transcription between cell lines. However, as shown in Fig. 5C, no significant variations in IFN-β transcription were observed.

Thus, we suggest that the WNV_KUN-induced defects imposed on IFN-mediated signaling are dependent on the downstream signaling of the ER stress response, particularly those events triggered by ATF6 and IRE1.