Role of Interferon Regulatory Factor 3-Mediated Apoptosis in the Establishment and Maintenance of Persistent Infection by Sendai Virus

Abstract

Infection of cultured cells by paramyxoviruses causes cell death, mediated by a newly discovered apoptotic pathway activated by virus infection. The key proapoptotic protein in this pathway is interferon regulatory factor 3 (IRF-3), which upon activation by virus infection binds BAX, translocates it to mitochondria, and triggers apoptosis. When IRF-3-knockdown cells were infected with Sendai virus (SeV), persistent infection (PI) was established. The PI cells produced infectious SeV continuously and constitutively expressed many innate immune genes. Interferon signaling was blocked in these cells. The elevated levels of IRF-3-driven genes in the PI cells indicated that the amount of residual IRF-3 activated by endogenous SeV was high enough to drive the transcriptional effects of IRF-3 but too low to trigger its apoptotic activity. We confirmed this IRF-3 threshold idea by generating a tetracycline (Tet)-inducible cell line for IRF-3 expression, which enabled us to express various levels of IRF-3. PI could be established in the Tet-off cell line, and as expected, when doxycycline was withdrawn, the cells underwent apoptosis. Finally, we tested for PI establishment in 12 mouse embryo fibroblasts by natural selection. Eleven lines became persistently infected; although seven out of them had low IRF-3 levels, four did not. When one of the latter four was further analyzed, we observed that it expressed a very low level of caspase 3, the final executor protease of the apoptotic pathway. These results demonstrated that SeV PI can arise from infection of normal wild-type cells, but only if they can find a way to impair the IRF-3-dependent apoptotic pathway.

INTRODUCTION

Virus infection elicits complex host responses mediated by both the innate and the adaptive arms of the immune system. However, immediate intrinsic responses of an infected cell play a major role in determining the outcome as well. Consequently, both viral and cellular genes play decisive roles in dictating the fate of the infected cell, including its premature death, its ability to support viral gene expression, its oncogenic transformation, and the establishment of persistent infection (PI) (1, 2). The cellular responses are initiated by the recognition of viral pathogen-associated molecular patterns (PAMPs) by specific cellular receptors. In the case of RNA viruses, viral double-stranded RNA (dsRNA) is often the critical PAMP (3). It is recognized by either membrane-bound Toll-like receptor 3 (TLR3) or the cytoplasmic RNA helicases (RLHs) RIG-I and Mda5 (4). Viruses of different families use these receptors differentially, and there are cell type-dependent specificities as well (5–7). Activation of the cellular receptors by the viral PAMPs triggers many innate immune signaling pathways, which often cause activation of specific transcription factors, such as NF-κB and interferon (IFN) regulatory factor 3 (IRF-3) (8). The activated transcription factors, in turn, induce transcription of hundreds of cellular genes. The proteins encoded by many of these genes, including IFN genes, inhibit virus replication in the infected cell. In addition to gene induction, virus infection can also trigger programmed cell death or apoptosis by activating either the extrinsic apoptotic pathway initiated by caspase 8 or the intrinsic pathway initiated by caspase 9 (9).

We have been studying the cellular response to infection by paramyxoviruses, especially Sendai virus (SeV). These studies have revealed a new apoptotic pathway triggered by the activation of IRF-3 by virus infection (10–12). We have reported that the same apoptotic pathway is activated by other viruses as well and it represents a universal response to dsRNA produced by both RNA and DNA viruses (13). Moreover, the apoptotic pathway is triggered only by the RLH cytoplasmic receptors of dsRNA and not the membrane receptor TLR3. IRF-3 is an extensively studied transcription factor that plays a critical role in the induction of IFN and many IFN-stimulated genes (ISGs) (14). We observed that if the IRF-3-mediated apoptotic pathway is inoperative, SeV infection does not kill infected cells; instead, the cells become persistently infected (12). An unexpected discovery was made when we explored the mechanism of IRF-3-mediated apoptosis (10, 11). It revealed dual independent functions of IRF-3 as a transcription factor and a proapoptotic factor. For both functions, IRF-3 needs to be activated by triggering RLH signaling, but the apoptotic activation requires the presence of additional tumor necrosis factor receptor-associated factor (TRAF) proteins (10). The lack of interdependence of the two properties of IRF-3 has been demonstrated by generating IRF-3 mutants which have one function but not the other. The proapoptotic function of IRF-3 is mediated by its newly discovered BH3 domain, through which it interacts with the proapoptotic protein BAX. Activated IRF-3 translocates to the mitochondria, bringing along activated BAX, which triggers the activation of the mitochondrial apoptotic pathway, releasing cytochrome c to the cytoplasm. Cytoplasmic cytochrome c interacts with apoptotic protease-activating factor 1 (APAF-1) and facilitates its oligomerization and formation of the apoptosome complex, in which pro-caspase 9 is activated by its autocatalytic cleavage. Active caspase 9 cleaves pro-caspase 3 to generate caspase 3, the executor of apoptotic activity. We demonstrated that the mitochondrial function of IRF-3 is activated by any trigger that also activates its transcriptional function. Such triggers include infection by many RNA viruses and DNA viruses and transfection of dsRNA to uninfected cells. We also demonstrated that, by causing premature death of the infected cell, the IRF-3-activated apoptotic pathway promotes the overall antiviral response of cells and mice (13).

Activation of the IRF-3-mediated apoptotic pathway is regulated during virus infection. Phosphatidylinositol 3 (PI3) kinase, which is also activated by SeV infection, inhibits the process at the early phase of infection (12). Consequently, SeV-infected cells die only after 24 h or longer, but if the PI3 kinase/AKT pathway is inhibited, all infected cells die within 6 h of infection. We demonstrated that the preventive PI3 kinase action is mediated by the antiapoptotic protein X-linked inhibitor of apoptosis (XIAP), which binds to the apoptosome and blocks its activation (15). In the absence of PI3 kinase activity or at late times after infection, XIAP is degraded in the infected cells, thus relieving the inhibitory action on the apoptotic machinery. Hence, apoptosis is regulated by at least three cellular pathways, all of which are activated by SeV infection; two of these, activation of IRF-3 and XIAP degradation, are proapoptotic, and the third, activation of PI3 kinase, is antiapoptotic. Interactions and temporal regulation of the three pathways described above determine whether and when the infected cell dies.

An interesting outcome of our recent findings, as outlined above, is that they provided a window of opportunity to study establishment and maintenance of PI at the cellular level. By manipulating the functions of different components of the IRF-3-mediated apoptotic pathway, we could either hasten, slow down, or block the death of the infected cell. A complete block generated persistently infected cells that could produce infectious SeV continuously in culture. In the current study, we have characterized the innate immune status of persistently infected human cells which we have maintained in culture for more than 3 years. PI could also be established in mouse embryo fibroblasts (MEFs) by natural selection. Thus, we demonstrate that PI cells can emerge from infected wild-type (WT) cell cultures without any overt manipulation. In all cases tested so far, the PI cells were defective in the functioning of the IRF-3-mediated apoptotic pathway.

MATERIALS AND METHODS

Cells and reagents.

HT1080 cells, 1K cells (HT1080 cells containing short hairpin RNA against IRF-3 [shIRF-3]), and 1K/SeV cells (1K cells persistently infected with SeV) have been previously described (12). MCF-7 cells persistently infected with SeV (MCF-7/SeV) were generated as described below. All the cell lines were maintained in Dulbecco modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum (FBS), 100 units/ml of penicillin, and 100 μg/ml of streptomycin (complete medium); 1K and 1K/SeV cells were maintained in the presence of G418 (400 μg/ml). Antibodies against human IFIT3/P60, murine Ifit2/P54 (16), and SeV C protein were raised in rabbits by injection of purified full-length proteins by the Hybridoma Core, Lerner Research Institute. Antibodies against phospho-S396–IRF-3, phospho-Y701-STAT1, procyclic acidic repetitive protein (PARP), cleaved PARP, phospho-S32-IκBα (14D4), IκBα, XIAP, and caspase 3 were from Cell Signaling; antibodies against superoxide dismutase 2 (SOD2), IRF-3, and BAX were from Santa Cruz; antibody against A20 was from Imgenex; STAT1 antibody was from BD Transduction Labs; and actin antibody was from Sigma. Anti-whole SeV antibody was a gift from John Nedrud, Case Western Reserve University, Cleveland, OH. Poly(I·C) was obtained from GE Healthcare, and FuGene 6 was obtained from Roche. Poly(I·C) transfection was performed using FuGene 6 according to the manufacturer's protocol. Human beta interferon (IFN-β; Calbiochem) was applied to cells by adding it to the culture medium at a final concentration of 1,000 units/ml.

Generation of cells expressing inducible IRF-3.

N-terminally V5-tagged human IRF-3 was subcloned into the pTRE2hyg vector (Clontech) downstream of the tetracycline (Tet)-responsive element and cotransfected with pTet-Off (Clontech) in 1K cells, and the transfected cells were selected under G418 (400 μg/ml), puromycin (1 μg/ml), and hygromycin (100 μg/ml). Individual clones were screened on the basis of inducible expression of IRF-3, and such a clone (clone 10) was obtained and used for this study. Clone 10 was maintained in the presence of doxycycline (Dox; 1 μg/ml; Clontech). To induce the expression of IRF-3, the cells were washed extensively with phosphate-buffered saline (PBS) and grown in the absence of doxycycline. For obtaining cells expressing various levels of IRF-3, clone 10 cells were maintained in the presence of various concentrations of doxycycline.

Virus infection and titration.

The SeV Cantell strain was obtained from Charles River SPAFAS Inc., and vesicular stomatitis virus (VSV) Indiana was a gift from Amiya Banerjee, Lerner Research Institute, Cleveland Clinic. All virus infections were carried out as described before (12, 13). Briefly, the cells were infected with SeV or VSV in virus infection medium (DMEM supplemented with 2% FBS) at the multiplicity of infection (MOI) indicated in the figure legends. Cells were incubated with virus for 1 h with gentle agitation every 10 min. After 1 h of incubation, the virus was removed and cells were washed twice with DMEM. The cells were then placed in complete medium until they were harvested. To quantify the infectious virus particles, VSV-infected cultures were frozen and thawed at 8 h postinfection, and viral titers were determined by plaque assay on Vero cells, detecting lytic VSV plaques by crystal violet counterstain. This assay does not detect SeV, since this virus does not form lytic plaques.

Generation of persistently infected cells.

1K, MCF-7, and WT MEF cells were infected with SeV at an MOI of 10 (or with virus produced from 1K/SeV, as indicated in the legend to Fig. 1C), and the infected cells were continuously passed every 2 to 3 days to establish persistently infected cell lines. The persistently infected cell lines were tested for the expression of viral proteins by immunostaining using anti-whole SeV antibody or Western blotting for SeV C protein. For generating persistently SeV-infected clone 10 (clone 10/SeV), the cells were infected with SeV at an MOI of 10 and the infected cells were maintained in the presence of doxycycline (1 μg/ml). For inducing IRF-3 expression, the cells were extensively washed with PBS and grown in the absence of doxycycline.

Fig 1.

SeV infection establishes persistence in the absence of IRF-3. (A) Cell lysates from mock-infected or SeV-infected (24 h postinfection) HT1080, 1K/SeV, or 1K cells were analyzed for IRF-3, SeV C protein, and actin by Western blotting. (B) 1K/SeV cells or their matched mock-infected control cells (1K) were immunostained with an antibody against SeV (red fluorescence). Nuclei were visualized by DAPI staining. (C) 1K cells were infected with virus produced by 1K/SeV cells (lane virus:1K/SeV), PI was established, and lysates from these cells, mock-infected 1K cells (lane −), or 1K/SeV cells (lane SeV) were analyzed for SeV C protein.

Western analysis.

Western analysis was performed using previously described procedures (10). Briefly, cells were lysed in 50 mM Tris buffer, pH 7.4, containing 150 mM NaCl, 0.1% Triton X-100, 1 mM sodium orthovanadate, 10 mM sodium fluoride, 10 mM β-glycerophosphate, 5 mM sodium pyrophosphate, and protease inhibitors (Roche); the total protein extracts were analyzed by SDS-PAGE followed by Western blotting.

siRNA experiments.

The small interfering RNAs (siRNAs) against human SOD2 and BAX were obtained from Thermo Scientific (L-009784-00-0005 and L-003308-01-0010, respectively) and were transfected using DharmaFECT 4 reagent. Cells were analyzed as described in the figure legends. A nontargeting siRNA (D-001810-10-05; Thermo Scientific) was used as a control and was transfected using the same protocol described above.

Immunofluorescence.

The persistently infected cells, grown on chamber slides, were fixed with 4% paraformaldehyde, permeabilized with 0.2% Triton X-100, and immunostained with 1:1,000 anti-whole SeV primary antibody, followed by 1:1,000 Alexa Fluor-conjugated secondary antibody, all in 5% goat serum. Objects were mounted with VectaShield/DAPI (4′,6-diamidino-2-phenylindole) and analyzed by fluorescence microscopy.

Microarray.

For mRNA expression profiling of 1K and 1K/SeV cells, total RNA from 3 independent samples was extracted using TRIzol reagent (Invitrogen) and DNase I treatment (DNAfree; Applied Biosystems/Ambion) according to the manufacturers' instructions. RNA was further purified using spin columns (RNeasy minikit; Qiagen) before subjection to mRNA expression microarray analysis by hybridization of cRNA to an Illumina Human Ref-8 bead chip. Results were analyzed using GenomeStudio software (version 2010.2; Illumina, Inc.); cRNA hybridization to chips was performed by the Lerner Research Institute Genomics Core.

RESULTS

Development of PI cell line.

To examine how PI with SeV changes the properties of a cell, we established an SeV-infected cell line. In human HT1080 cells, we knocked down the expression of IRF-3 using shRNAs targeted to the 5′ untranslated region (UTR) and the 3′ UTR of IRF-3 mRNA (12). A clonal cell line (1K) expressing a very low level of IRF-3 (Fig. 1A) was infected with SeV and cultured continuously to establish the PI cell line 1K/SeV. As expected, freshly infected parental HT1080 cells and the 1K/SeV cells expressed abundant SeV C protein and a low level of IRF-3 (Fig. 1A); more than 95% of the 1K/SeV cells were virus infected (Fig. 1B). We have maintained this line for more than 3 years now. SeV produced by the PI line was indistinguishable from the inoculum virus; they had similar infectious unit-to-total particle ratios, and the virus produced by 1K/SeV cells was able to establish PI in 1K cells (Fig. 1C).

In the next experiment, we examined how PI changed cellular gene expression patterns. Microarray profiling was carried out in triplicate to compare the levels of specific mRNA expression between 1K/SeV cells and 1K cells (see Table S1 in the supplemental material). The expression levels of 55 mRNAs were elevated at least twofold in 1K/SeV cells compared to the level of expression in 1K cells (Table 1), and those of 12 mRNAs were reduced at least twofold (Table 2). Among the genes whose expression was elevated, the highest on the list were CCL5 and IFI44, followed by many other ISGs, such as IFIT2, IFIT1, ISG15, and IFIT3. Many of these ISGs are induced not only by IFN signaling but also by RIG-I/Mda5 or TLR3 signaling. SOD2 mRNA was elevated 4.2-fold (Table 1); the corresponding protein was also elevated in 1K/SeV cells compared to the level in 1K cells (Fig. 2A). Activation of RIG-I signaling by poly(I·C) transfection, which is known to activate NF-κB, induced SOD2 protein expression in 1K cells, as expected (Fig. 2A). However, in 1K/SeV cells, the already elevated level of SOD2 was not increased further by RIG-I activation. SOD2 protects cells from the damage done by reactive oxygen species generated by virus infection (17, 18). However, contrary to our expectation, elevated SOD2 expression was not needed for preventing apoptosis of 1K/SeV cells. Knocking down SOD2 expression by cognate siRNA transfection did not cause apoptosis of 1K/SeV cells, as measured by a PARP cleavage assay (Fig. 2B). Another gene whose expression was elevated in 1K/SeV cells was the IRF-3-driven ISG IFIT3, which encodes the protein P60. The level of P60 was high in 1K/SeV cells, and it was further induced by RIG-I activation (Fig. 2C). Surprisingly, P60 was also induced in 1K cells, which express only a low level of IRF-3.

Table 1.

Genes with elevated expression in 1K/SeV cells compared to 1K cells^a

Gene	Fold elevated expression in 1K/SeV cells
CCL5	56.53
IFI44	36.81
NFS1	18.34
IFIT2	15.82
IFIT1	14.77
IFIT3	14.13
ISG15	9.57
CCL3L3	8.05
OASL	6.46
CCL3	5.84
IL1B	5.03
CCL3L1	4.60
SOD2	4.52
ZC3HAV1	4.31
DDX60	3.79
IL-6	3.66
CENTA1	3.53
CCL4L1	3.52
APCDD1L	3.44
SPP1	3.44
GCNT3	3.35
OAS2	3.24
RSAD2	3.22
MMP9	3.22
PMAIP1	3.16
PRSS35	3.15
SAMD9	3.03
ATP6V0D2	3.00
ABCA1	2.84
IFIH1	2.82
ISG20	2.68
HERC6	2.68
RARRES3	2.62
AKR1C3	2.59
IFI6	2.57
GNG11	2.56
NPSR1	2.56
C1orf106	2.51
ITGA1	2.46
RND3	2.44
SP140	2.39
DDX58	2.32
TNFAIP3	2.25
IL-8	2.21
HLA-H	2.21
COL5A1	2.19
PTGS2	2.18
CALB2	2.11
SAMD9L	2.06
FCRLA	2.05
HIST2H2AA3	2.04
HMOX1	2.04
SPOCK1	2.03
DHX58	2.03
ITGA2	2.03
FZD8	2.02
ME1	2.01
CXCL10	2.00

^aRNA expression profiles of three independent samples of 1K and 1K/SeV cells were obtained by hybridizing to an Illumina Human Ref-8 microarray chip. Genes with at least a 2-fold-elevated average expression level in 1K/SeV cells are listed.

Table 2.

Genes with reduced expression in 1K/SeV cells compared to 1K cells^a

Gene	Fold reduced expression in 1K/SeV cells
EEF1A2	5.00
MAGEC2	3.95
ITGB2	3.64
DHRS2	3.55
FLJ35767	3.31
BGN	2.69
TRPM2	2.64
VGF	2.36
SAA1	2.26
MT1E	2.23
HRASLS3	2.08
CYR61	2.08
KRT80	2.04
THBS1	2.03
SHROOM2	2.01
DUSP1	2.01

^aExpression profiles in Table 1 were used to identify genes with at least a 2-fold-reduced average mRNA expression level in 1K/SeV cells compared to 1K cells.

Fig 2.

Elevated expression of SOD2 is not required to protect PI cells from apoptosis. (A) SOD2 protein expression in untreated 1K or 1K/SeV cells or cells transfected with poly(I·C) transfected for the indicated times was analyzed by Western blotting; (B) 1K or 1K/SeV cells were transfected with control (ctrl) or SOD2 siRNA (Dharmacon), and 96 h later the cells were analyzed by Western blotting for PARP cleavage (C-PARP, cleaved PARP), SOD2, and actin. A lysate from SeV-infected (MOI, 10; 24 h postinfection [p.i.]) HT1080 cells was used as a positive control for cleaved PARP (right); (C) induction of P60/IFIT3 protein following poly(I·C) transfection of HT1080, 1K, and 1K/SeV cells was analyzed by Western blotting.

Innate immune status of 1K/SeV cells.

Because several ISGs are constitutively expressed in 1K/SeV cells, we wondered whether they were in a global antiviral state. To test this possibility, HT1080, 1K, and 1K/SeV cells were infected with VSV at an MOI of 0.1 for 24 h. Almost all cells were killed by VSV infection (Fig. 3A, middle); thus, 1K/SeV cells were not in an antiviral state. Moreover, IFN pretreatment could protect HT1080 and 1K cells, but not 1K/SeV cells, from VSV-induced cytopathic effects (Fig. 3A, right). This result was supported by comparing the efficiencies of virus replication in 1K and 1K/SeV cells. Similar levels of infectious virus were produced by the two untreated cell lines; IFN pretreatment reduced VSV replication in 1K cells but did not inhibit VSV replication in 1K/SeV cells, indicating that IFN cannot induce an antiviral state in 1K/SeV cells (Fig. 3B). The last observation suggested that IFN signaling might be not functional in 1K/SeV cells. Indeed, STAT1 was poorly activated by IFN in 1K/SeV cells, as measured by its Tyr phosphorylation; in contrast, STAT1 activation was very strong in the parental 1K cells (Fig. 3C). This was probably caused by SeV proteins blocking IFN signaling in 1K/SeV cells, because prior infection of 1K cells with SeV for 2 h also blocked STAT1 activation by IFN-β.

Fig 3.

Status of innate immune signaling in 1K/SeV cells. (A) HT1080, 1K, or 1K/SeV cells were left untreated or pretreated with IFN-β (1,000 U/ml) and infected with VSV (MOI, 0.1); culture fields were visualized for cell survival at 24 h postinfection; (B) antiviral states of 1K and 1K/SeV cells were compared by infecting them with VSV (MOI, 0.1) before or after IFN-β pretreatment for 8 h; infectious VSV production was determined by plaque assays (presented as numbers of PFU/ml); (C) STAT1 tyrosine 701 phosphorylation was analyzed in uninfected or SeV-infected (MOI, 10; 2 h postinfection) 1K cells or 1K/SeV cells in response to 30 min of IFN-β treatment; (D) IRF-3 serine 396 phosphorylation (p-IRF-3) and induction of A20 protein were analyzed at various times after poly(I·C) transfection of HT1080, 1K, and 1K/SeV cells. The panel for actin is identical to that in Fig. 2C, because the two experiments were done together. (E) IκB-α serine 32 phosphorylation (p-IκB-α) was analyzed before or 2 h after poly(I·C) (pIC) transfection of 1K and 1K/SeV cells. Total IκB-α and SeV C proteins were analyzed by Western blotting.

We next investigated the status of RIG-I signaling in 1K/SeV cells by transfecting poly(I·C). The residual IRF-3 was phosphorylated and activated similarly in 1K and 1K/SeV cells (Fig. 3D, top), indicating that RIG-I signaling to IRF-3 was unimpaired by SeV PI. The same was true for the NF-κB pathway; IκBα was similarly phosphorylated in 1K and 1K/SeV cells upon RIG-I activation (Fig. 3E, top). Consequently, the NF-κB-driven gene product, A20, was strongly induced in both lines by RIG-I activation (Fig. 3D, middle). The results presented above demonstrate that the persistent infection in 1K/SeV cells causes a block in type I IFN signaling but not RIG-I signaling. They also demonstrate that the low level of IRF-3 expressed in 1K/SeV cells was sufficient to induce expression of a few ISGs at a low constitutive level. Moreover, overt external activation of the RIG-I pathway caused further induction of these ISGs but no apoptosis.

Threshold level of IRF-3.

As stated above, in 1K/SeV cells, RIG-I signaling was apparently activating the transcriptional function of the residual IRF-3 but not its apoptotic action. The observation was paradoxical, because activation of the RIG-I pathway is known to cause activation of both functions of IRF-3, its transcriptional activity and its apoptotic activity, even though the two activation pathways are distinct with respect to the requirements for specific signaling proteins and the target residues of IRF-3 that need to be phosphorylated. We wondered whether this difference was due to a need for a higher level of IRF-3 for triggering the apoptotic effect or a selective block of the apoptotic pathway by the viral proteins expressed in the PI cells. To distinguish these possibilities, we set out to generate a cell line in which IRF-3 expression would be driven by a Tet-repressible (Tet-off) promoter and the level of IRF-3 expression would be regulated by various concentrations of Dox in the culture medium. For this purpose, V5-tagged IRF-3 was cloned behind a Tet-responsive promoter and was transfected into Tet repressor-expressing 1K cells. Pools of transfected clones expressed a high level of V5–IRF-3 in the absence of Dox but very little in its presence (Fig. 4A). A tightly regulated cell clone, clone 10, derived from one pool was used for further experiments (Fig. 4B). We used clone 10 to induce different levels of IRF-3-expressing cells. Clone 10 was cultured for several passages in the absence of Dox to induce the maximum level of IRF-3 expression, and then increasing doses of Dox were added to different plates of these cells, causing a graded repression of IRF-3 expression (Fig. 4C, bottom). Activation of RIG-I by SeV infection caused activation of IRF-3 to trigger both the gene induction pathway (Fig. 4C, middle) and the apoptotic pathway (Fig. 4C, top). However, there was a distinct difference in the dose-responses of the two pathways. Even the lowest level of IRF-3 expressed from the incompletely knocked down resident IRF-3 gene and not the Tet-off transgene was sufficient for activating the gene induction pathway maximally (Fig. 4C, left lane), whereas higher levels of IRF-3 expression were required to trigger the apoptotic pathway (Fig. 4C, middle lane). To confirm that the inducible IRF-3-mediated apoptosis was BAX dependent, BAX was knocked down in Dox-untreated clone 10 cells (Fig. 4E). SeV-induced apoptosis, as determined by PARP cleavage, was inhibited in the BAX-knockdown cells (Fig. 4D). We established a PI cell line derived from clone 10 (clone 10/SeV) by infecting it with SeV in the presence of Dox. The majority of the cells were viral antigen positive (Fig. 5A). As expected, removal of Dox caused PARP cleavage (Fig. 5B) and complete apoptosis of these cells within 3 days (Fig. 5C). This cell line will be a valuable tool to study further the role of IRF-3 in preventing maintenance of persistent infection.

Fig 4.

Inducible expression of IRF-3 and its functions in 1K cells. (A) 1K cells were transiently transfected with pTet-Off (Clontech) and V5-tagged human IRF-3 cloned in the pTRE2hyg vector (pTRE2hyg-V5-hIRF-3; Clontech), and 48 h later the cells were either left untreated or treated with Dox (1 μg/ml) and the cell lysates were analyzed for IRF-3 and actin by Western blotting; (B) a stable cell line (clone 10) was generated from the transiently transfected cells described in panel A and maintained in the presence of Dox (1 μg/ml); IRF-3 expression was analyzed in the presence of Dox or after removal of Dox; (C) clone 10 cells were maintained in the absence or the presence of various concentrations of Dox (100 ng/ml, 1 ng/ml, or none) and then infected with SeV (MOI, 10); IRF-3 levels gradually increased with decreasing Dox concentration (bottom), apoptosis was measured from the amount of cleaved PARP production (top), and gene induction was measured from the amount of P60/IFIT3 induction (middle); (D, E) clone 10 cells were maintained in the absence of Dox to induce the expression of IRF-3 and then transfected with siRNAs against BAX or a nontargeting sequence (NT) using DharmaFECT 4 reagent (Thermo Scientific); after 48 h, cells were infected with SeV (MOI, 10), and at 24 h postinfection, cell lysates were analyzed for cleaved PARP (D); BAX knockdown was confirmed by Western blotting (E).

Fig 5.

Inducible expression of IRF-3 causes apoptosis in PI cells. (A) Clone 10/SeV cells were immunostained with an antibody against SeV (green fluorescence). Nuclei were visualized by DAPI staining; (B) persistently infected clone 10 cells (clone 10/SeV) were grown in the absence of Dox for the indicated times, when cell lysates were analyzed for cleaved PARP and actin by Western blotting; (C) clone 10/SeV cells were either grown in the presence of Dox (1 μg/ml) or washed extensively with PBS and grown in the absence of Dox for 72 h, when the culture fields were photographed.

Natural selection for PI in MEFs.

In the next series of experiments, we wanted to explore two additional features of the role of the IRF-3-mediated apoptotic pathway in establishing persistence. We wanted to determine whether persistence can be established in mouse cells as well and whether persistently infected cells can arise from a population of SeV-infected wild-type cells without any overt manipulation of components of the pathway, such as knocking down the expression of IRF-3. We infected 12 WT MEFs (MEF1 to MEF12) which were derived from different WT mice and continued culturing them. All cells died only for MEF2, but proliferating cell lines arose from the other 11 MEFs. They were all persistently infected with SeV and produced more than 10⁶ PFU/ml of infectious SeV in the culture medium. Thus, persistently infected mouse cell lines could be established from MEFs at a high frequency. We then investigated why these cells escaped the apoptotic effects of SeV infection. As expected, the persistently infected MEF lines were all synthesizing high levels of SeV C protein (Fig. 6A and B). Seven of them expressed low or undetectable levels of IRF-3, although the parental uninfected cells had high levels of IRF-3 (Fig. 6A). These cells could apparently survive infection because they expressed little IRF-3. They might have been clonally selected from a heterogeneous MEF population expressing different levels of IRF-3, or virus infection caused effective degradation of IRF-3 in these cells. The more intriguing cases were the persistently infected MEFs, which expressed as much IRF-3 as the parental infected cells (Fig. 6B) but were still not killed by the apoptotic pathway. To identify the cause of their survival, we chose MEF1 as a candidate and determined its levels of different components of the IRF-3-mediated apoptotic pathway. We have reported that, early after infection, XIAP, an antiapoptotic protein, keeps the pathway in check by inhibiting apoptosome assembly; at later times in infection, it is degraded and the brake is released to allow apoptosis (15). We wondered whether degradation of XIAP did not happen in MEF1, but as shown in Fig. 6C, XIAP was equally degraded in persistently infected MEF1 and two other MEFs (MEFs 3 and 4), irrespective of their IRF-3 levels. PI MEF1 cells showed no detectable PARP cleavage; however, they expressed IRF-3-inducible cellular protein P54/Ifit2 and there were abundant levels of SeV C protein (Fig. 7A). A closer examination of PI MEF1 cells revealed that almost all of their IRF-3 was activated, as indicated by a shift in their mobility (Fig. 7B). We have reported that TRAF2, TRAF6, and BAX are essential components of the IRF-3-dependent mitochondrial apoptotic pathway (10). These proteins were present at comparable levels in both uninfected and PI MEF1 cells (Fig. 7B). Bcl-XL, a member of the Bcl-2 family of antiapoptotic proteins, was present at similar levels in both uninfected and PI MEF1 cells (Fig. 7B), suggesting that enhanced expression of antiapoptotic proteins may not be protecting these cells from apoptosis. Both types of cells also contained comparable levels of uncleaved pro-caspase 9 (Fig. 7B). However, this was not true for the executor caspase, caspase 3; the PI cells expressed much less pro-caspase 3 than the parental uninfected cells. Thus, it appears that these cells survived because of a defect in the last step of the apoptotic pathway, which requires the action of caspase 3 (see Fig. 4 in reference15). The phenomenon described above could also be demonstrated in human cells by using cells of the MCF-7 line, a breast cancer cell line that lacks any functional caspase 3 (19). When infected with SeV, MCF-7 cells become persistently infected with all cells expressing viral antigens (Fig. 7C). These results demonstrated that either experimental or natural manipulations of the levels of several components of the pathway, IRF-3, RIG-I, XIAP, or caspase 3, could lead to persistent infection.

Fig 6.

Persistence in MEF cells. Cell lysates were prepared from PI MEF cells and their matched mock-infected controls and analyzed for IRF-3 and SeV C protein (A, B) and XIAP (C) levels by Western blotting. (A) PI MEFs which showed reduced levels of endogenous IRF-3; (B) PI MEFs with unchanged levels of endogenous IRF-3; (C) endogenous XIAP levels from the selected PI MEFs (as indicated).

Fig 7.

PI in the presence of IRF-3 but absence of caspase 3. (A) Lysates from PI MEF1 cells or their matched mock-infected controls were analyzed for cleaved PARP, P54/Ifit2, or SeV C protein; (B) lysates from PI MEF1 cells or their matched mock-infected controls were analyzed for IRF-3, TRAF2, TRAF6, BAX, Bcl-XL, caspase 9, and caspase 3 by Western blotting; (C) persistently infected MCF-7 or 1K cells were immunostained with an antibody against SeV (green fluorescence). Nuclei were visualized by DAPI staining.

DISCUSSION

Our observation that SeV cannot kill an infected cell if the cell does not express IRF-3 has allowed us to study persistent infection. The persistently infected 1K/SeV cells divide and grow normally and continuously produce infectious SeV. Infectious virus produced by 1K/SeV cells was very similar to the original SeV in its ability for replication, gene induction, and induction of apoptosis (results not shown). In the future, these cells can be used for studying virus morphogenesis and egress in the absence of cellular perturbation caused by apoptotic or preapoptotic cells. In the current study, we have, instead, focused on the infected cell itself. By mRNA profiling, we uncovered major changes in cellular gene expression in the PI cells. Among the 55 genes whose expression was altered, the levels of all of the repressed genes and most of the genes with elevated expression were altered by only two- to fourfold, but a few were induced quite strongly. Many of the induced genes are those for known ISGs, chemokines, and cytokines. Transcription of many of them is known to be driven by NF-κB, and transcription of others is induced by members of the IRF family, such as IRF-3, IRF-7, or IRF-9 (as a component of ISG factor 3). These transcription factors are known to be activated by exogenous virus infection, but for the PI cells, it appears that they were activated even by viral gene expression from within. As expected, the elevated mRNAs produced more proteins; for example, in the 1K/SeV cells, the levels of SOD2, A20, and P60 (IFIT3) proteins were all elevated. We investigated the functional significance of one protein with elevated levels, SOD2, which is known to protect cells from the apoptotic damage caused by reactive oxygen species produced upon virus infection. However, elevated SOD2 was not required to prevent apoptosis in PI cells; when its expression was knocked down, there was no noticeable change in the growth pattern of the 1K/SeV cells. It remains to be seen if and how continuous high expression of several ISGs and chemokines changes the properties of the PI cells.

High expression of several ISGs in 1K/SeV cells raised the possibility that they were in a perpetual antiviral state, but challenging them with VSV infection showed that that was not the case. Both 1K cells and 1K/SeV cells were equally hospitable to VSV replication. However, these results should be interpreted with caution. Because different ISGs are directed against different viruses, it is possible that 1K/SeV cells are resistant to other viruses that we have not tested. These results also indicate that the expressed ISGs have no antiviral effects against VSV, which is somewhat unexpected, because mouse Ifit2 has recently been shown to have strong anti-VSV effects in vivo and human IFIT2 was strongly induced in 1K/SeV cells (20). It should be noted, however, that the anti-VSV effect of mouse Ifit2 was highly cell type specific, being manifested only in neurons.

We were interested in determining the effects of the continuous presence of SeV in 1K/SeV cells on innate immune signaling pathways. Even a high dose of IFN-β could not induce ISGs in these cells, demonstrating strong inhibition of type I IFN signaling. This was not surprising because different paramyxoviruses are known to block IFN signaling by inactivating or degrading STAT1 or STAT2; in the case of SeV, viral C and V proteins are responsible for this action (21, 22). Surprisingly, the RLH signaling pathway, which leads to IFN induction by SeV infection, was active in 1K/SeV cells. Paramyxoviruses, in general, block RLH signaling mediated by Mda5 but not RIG-I, because the V proteins of many viruses of this family directly bind to Mda5 (23, 24). We have previously reported that the V proteins of rubulaviruses (Rubulavirus is a genus of the Paramyxoviridae) serve as decoy substrates for TANK-binding kinase 1 (TBK1), the protein kinase that activates IRF-3, thus blocking IFN induction by many pathways that use IRF-3 (25). On the other hand, phosphorylation of V proteins by TBK1 causes their degradation, thus establishing in the infected cell equilibrium between IRF-3 activation and V protein degradation. We observed that in 1K/SeV cells, poly(I·C) transfection, which is known to activate both RIG-I and Mda5 signaling, caused induction of both IRF-3- and NF-κB-driven genes, in spite of the presence of SeV V protein.

One of the most striking findings of the current study is that there is a threshold level of IRF-3 for triggering the apoptotic pathway. This was indicated by the survival of the 1K/SeV cells which expressed a low level of IRF-3. This low level was sufficient to drive the transcriptional activity of IRF-3 but was not sufficient to trigger the apoptotic activity. The IRF-3 threshold idea was validated by developing the Tet-off system for IRF-3 expression. It is clear that a much lower level of IRF-3 could trigger its transcriptional activity, whereas a higher level of expression was needed for the apoptotic pathway. At this point, we can only speculate about the mechanistic basis of this observation. It is possible that there are distinct subcellular pools of IRF-3 which are differentially used by the two pathways. Alternative mechanisms may also play crucial roles. For example, it is likely that the signaling complexes that recruit IRF-3 in the two pathways are distinct. They share many components, but a few proteins, such as TRAF2 and TRAF6, are unique to the apoptotic signaling complex, and they may determine the reduced efficiency of IRF-3 use in the apoptotic pathway (10). We can experimentally test these possibilities in the future using the Tet-off system to manipulate the level and the timing of IRF-3 expression. This system can also be used to express IRF-3 mutants that are capable of triggering one pathway but not the other. For example, if we express the known IRF-3 mutant (S396/398A) that does not get activated as a transcription factor and does not translocate to the nucleus but activates apoptosis (10), we may be able to bring down the threshold level of IRF-3 required for triggering the apoptotic pathway. The clone 10/SeV PI cells can be used in the future to study the temporal regulation of IRF-3-mediated apoptosis by monitoring the progress of the apoptotic pathway after triggering IRF-3 expression.

Our ability to establish PI cultures from infections of almost all WT MEFs provided strong evidence for the possibility of emergence of PI cells from natural infection of WT cells. Because no overt experimental manipulation was required for the establishment of PI MEFs, such cells may well arise in an organism infected with IRF-3-activating viruses. For establishing PI, cell selection was strongly involved because most infected cells died and were eliminated from the culture; but because of their growth advantage, the PI cells eventually dominated the whole culture. In the original population, there might have been clonal variations in the levels of IRF-3 expression, and the cells expressing low levels might have become PI. Alternatively, PI could be the result of a stochastic process. IRF-3 is known to be degraded rapidly in SeV-infected cells using a caspase 8- and proteasome-dependent pathway (26). Thus, there is a race between IRF-3 actions and its degradation. Moreover, the apoptotic pathway is kept in check in the early phase of infection by viral activation of the PI3 kinase pathway, which functions through the antiapoptotic protein XIAP (15). Thus, there is a probabilistic opportunity for most of the IRF-3 to be degraded before the PI3 kinase block of apoptosis is released. Obviously, in different cell types, the relative efficiency of the two competing pathways will be different because of different levels of expression of the individual components of the pathways. Hence, in principle, some cell types may be more prone to PI than others. Ye and Maniatis tested several cell lines for survival after SeV infection (27). Most of the cell lines died, but one MEF line survived and became PI. They observed that IFN expression was blocked at the transcriptional level in their infected MEFs and ectopic expression of IRF-3 led to IFN induction. After the establishment of PI in the MEF, the level of IRF-3 was normal, similar to what we observed in many MEFs.

An unexpected and comforting result came from analyzing our PI MEF1 cells. These cells expressed high levels of IRF-3 but still did not undergo apoptosis. Moreover, the relevant negative regulator of apoptosis, XIAP, was degraded in infected MEF1 cells, as in the other MEFs and human cells, but PI MEF1 cells expressed only a low level of the executor caspase, caspase 3; hence, the apoptotic pathway could not proceed to completion. The low level of caspase 3 may also explain why IRF-3 was not degraded in these cells. Because cells in an uninfected MEF1 cell population express a high level of caspase 3, it is very likely that a few variants expressing low levels of caspase 3 got selected to be persistently infected. We could demonstrate that such selection was reproducible. Fresh infection of MEF1 cells with SeV and selection for PI produced clones similar to the PI MEF1 clones selected originally. Finally, using caspase 3-negative human MCF-7 cells, we could demonstrate complete PI of the whole population. These results demonstrate that the phenomenon is not unique to MEFs. More importantly, one can envision from these results that the functional absence of any component of the IRF-3 apoptotic pathway will lead to the establishment of PI upon SeV infection.