Distinctive Roles for Type I and Type II Interferons and Interferon Regulatory Factors in the Host Cell Defense against Varicella-Zoster Virus

Differential effects of type I and type II IFNs on VZV infection.

VZV replication and cell-cell spread were compared in human embryonic lung fibroblasts (HELFs) that were treated with IFN-α and IFN-γ prior to or after VZV inoculation. Although IFN-α pretreatment (10⁴ U/ml) inhibited replication, VZV titers were only 1 log lower than the titers in untreated HELFs, and increasing the duration of IFN-α pretreatment from 2 h to 24 h did not further diminish virus recovery (Fig. 1A). In contrast, IFN-γ pretreatment (200 U/ml) dramatically reduced VZV titers by 3 to 4 logs compared to those in IFN-α-pretreated and untreated control HELFs. IFN-γ pretreatment also had a duration-related effect, progressing from a 2-log reduction in VZV titers with 2 h of exposure to low or undetectable titers when cell responses were induced for 12 h or 24 h before VZV inoculation. Expression of IFN-γ receptor (IFNGR) on HELFs and melanoma (Mel39) cells was shown by flow cytometry analysis (Fig. 1B).

Removing IFN-α after 2 h or 24 h of pretreatment, at the time of VZV inoculation, reversed its inhibitory effect on replication and allowed more cell-cell spread, as shown by the formation of larger plaques, visible when the monolayers were directly stained with the anti-VZV antibody reagent (Fig. 2A and B). In contrast, VZV spread remained blocked when IFN-γ was removed after 2 h or 24 h of pretreatment prior to inoculation (Fig. 2A). In addition, increasing the VZV inoculum from low (400 PFU/ml) to high (2,000 PFU/ml) enhanced the capacity of the virus to infect cells that were pretreated with IFN-α for 2 h but not those that were pretreated with IFN-γ (Fig. 2A and C). The effect of IFN-α pretreatment was dose dependent, with 10³ U/ml having less of an effect than 10⁴ and 10⁵ U/ml, whereas no plaque formation was detected when HELFs were pretreated with IFN-γ at a low dose of 20 U/ml compared to a higher dose of 100 or 200 U/ml (Fig. 2D). In contrast, when IFNs were added 48 h after VZV inoculation, some viral replication was detected at 72 h and 96 h in both IFN-α- and IFN-γ-treated cells. Nevertheless, VZV titers were reduced compared to those in untreated cells if replication was allowed to progress for 48 h before adding IFN-α or IFN-γ, and VZV replication was even more restricted by IFNs added posttreatment when the inoculum was low (400 PFU/ml) than when the inoculum was high (2,000 PFU/ml) (Fig. 2E). We also investigated IFN-γ effects when the time of addition was varied from 3, 6, 12, or 24 h after VZV inoculation. VZV replication at a titer in the range of 0.5 log higher than the inoculum titer was detected when IFN-γ was added at time points up to 12 h and viral titers were measured at 96 h (Fig. 2F). Adding IFN-γ at 24 h had a modest inhibitory effect on VZV replication, with titers being 1 × 10^3.8 in treated cells and 1 × 10^4.4 in untreated control cells. These observations indicate that antiviral control depends on the early initiation of type I and or type II IFN signaling and that the ISG-mediated host cell defense against VZV infection induced by IFN-γ exposure was more potent than that induced by IFN-α, as shown by irreversible inhibition, independent of the dose and duration of exposure or the VZV inoculum. The differential effects of IFN-α and IFN-γ on VZV replication were also observed in melanoma (Mel39) cells, which expressed IFNGR1, as shown in Fig. 1B). These data suggest that IFN-γ has similar effects on VZV infection, regardless of cell type, if the target cell expresses the IFN-γ receptor.

Comparative transcriptome analysis of fibroblasts treated with IFN-α and IFN-γ.

To identify differences in the host cell environment induced by type I and type II IFNs that might point to differences in the mechanisms of VZV inhibition, we profiled the RNA transcriptomes in HELFs that were treated with IFN-α or IFN-γ, mock treated for 24 h, or infected with VZV. The Illumina NextSeq platform was used to test three biological replicates for each of the four experimental conditions, yielding paired-end/single-end reads that were aligned to the human hg19 genome in the Illumina BaseSpace platform. Since coverage was low in one IFN-γ-treated sample, the data analysis was done with two IFN-γ replicates.

The patterns of differential gene expression between HELFs at 24 h after treatment with IFN-α or IFN-γ and mock-treated HELFs at the same time point are shown in scatter plots of the log fold change (Fig. 3A). IFN-α treatment significantly upregulated 386 genes in HELFs compared to their expression in mock-treated HELFs, whereas 528 genes were upregulated in IFN-γ-treated cells (P ≤ 0.05) (Fig. 3B). Each of the upregulated gene sets was analyzed based on the interaction database of the International Molecular Exchange (IMEx) Consortium using the NetworkAnalyst tool (11). Functional enrichment analyses using the Reactome database identified genes involved in IFN signaling that were detected when HELFs were treated with IFN-α or IFN-γ and compared to mock-treated HELFs, which were then plotted on the network maps shown in Fig. 3C. For example, STAT1 and IRF9 were among the 304 genes that were upregulated in both IFN-α- and IFN-γ-treated cells. Both IFNs also led to the downregulation of 173 genes compared to their expression in untreated HELFs, while 69 and 103 genes were downregulated uniquely in the presence of IFN-α or IFN-γ, respectively (Fig. 3B). Overall, IFN-γ exposure induced or diminished the expression of substantially more cell genes than IFN-α when their basal levels of transcription in HELFs were compared at 24 h posttreatment.

Given the differences between the effects of type I and type II IFNs on VZV infection, we next focused on the cell genes that were differentially regulated by IFN-α and IFN-γ pretreatment for 24 h (Fig. 4A). When we examined the network map generated using the 224 genes that were exclusively upregulated in the IFN-γ-treated cells, we found that among the STAT and IRF family proteins, only IRF1 was differentially expressed (Fig. 4B). Of these, the transcript levels of 23 genes were significantly higher after IFN-α exposure than after IFN-γ exposure, and the transcript levels of 93 were significantly higher after IFN-γ exposure than after IFN-α exposure (P ≤ 0.05). Seventeen of the 93 genes were found to have a direct interaction, based on the network map (Fig. 4C), and IRF1 displayed maximum connectivity in the map, again indicating its central role in the regulation of other genes triggered by IFN-γ. For example, it is well established that IRF1 transcriptionally regulates genes like IDO1 and CD74. The expression levels of the subset of transcripts that were significantly up- or downregulated in IFN-α- versus IFN-γ-treated cells as well as their levels in mock-treated and VZV-infected HELFs are listed in Table S1 in the supplemental material. Examples are IRF1, which showed significantly a higher level of induction by IFN-γ than by IFN-α (P = 0.000001), and IRF7, which had a higher level of induction after IFN-α treatment than after IFN-γ treatment (P = 0.0003) (Fig. 4D). VZV infection did not affect IRF1 transcripts, but IRF7 transcripts were significantly lower in infected cells than in uninfected HELFs (P < 0.005). Therefore, based on the bioinformatics analyses of the transcriptome sequencing (RNA-seq) data, we hypothesized that IRF1 and its downstream transcriptional programming of the cell are likely a key cellular regulator of VZV replication by IFN-γ signaling.

IRF1 is critical for the IFN-γ-mediated host cell defense against VZV.

As the first step in assessing the role of IRF1 in the IFN-γ versus the IFN-α response, we used flow cytometry to evaluate IRF1 protein expression in IFN-stimulated cells. Since IRF1 is proteasomally regulated, cells were treated with 50 μM MG132 for 4 to 6 h before analysis. At 4 h after IFN treatment, activated IRF1 was expressed in ∼40% of IFN-α-treated cells, whereas it was expressed in ∼70% of IFN-γ-treated cells. IRF1 protein expression after IFN-α treatment declined to ∼10% of cells by 8 h and <5% by 12 h (Fig. 5A). In contrast, IRF1 expression was sustained in IFN-γ-treated cells throughout the 24-h period, with >80% of the cells expressing IRF1. In addition, >65% of the IFN-γ-treated cells continued to express IRF1 protein 24 h after IFN-γ was removed, declining to ∼10% only at 48 h (Fig. 5B). Thus, the higher IRF1 transcript levels induced by IFN-γ than by IFN-α at 24 h posttreatment resulted in sustained IRF1 protein expression in most IFN-γ-treated cells.

Next, we tested the hypothesis that IRF1 is critical for the host cell defense against VZV by using a CRISPR-Cas9 lentivirus system. Stable IRF1 knockout (IRF1-KO) HELFs, melanoma (Mel39) cells, and controls were generated using lentiviruses prepared from the transfection of three guide RNAs (gRNAs; gRNAs 1 to 3) targeting IRF1 or a pLentiCRISPRv2 empty vector (gRNA4) (GenScript). When interference with IRF1 expression was evaluated by flow cytometry, only 10 to 15% of the gRNA1-transfected melanoma cells (gRNA1 melanoma cells) expressed IRF1 at 24 h after IFN-γ treatment in the presence of MG132 (Fig. 5C). IRF1 expression in the control gRNA4-transfected cells was similar to that in wild-type (WT) melanoma cells. Since IRF1 knockdown HELFs were difficult to expand in tissue culture and the two IFNs had similar effects on VZV infection in melanoma cells, IRF1-KO melanoma cells were used to evaluate the effects of inhibiting IFN-stimulated expression of IRF1 on VZV replication.

When gRNA1, gRNA4, and WT melanoma cells were untreated or pretreated with either IFN-γ or IFN-α for 24 h and then inoculated with VZV, infection was detected in untreated gRNA1, gRNA4, and WT cells by 48 h postinfection (hpi) (Fig. 5D). While VZV plaques were not detected in IFN-γ-treated WT melanoma cells, VZV replication was observed in IRF1 knockdown gRNA1 cells pretreated with IFN-γ. These effects on VZV replication were evaluated further in untreated and IFN-treated WT cells, gRNA1 knockdown cells, and control gRNA4 cells in a growth curve analysis (Fig. 5E). Cells were treated with either IFN-α or IFN-γ for 24 h prior to VZV inoculation (500 PFU/ml); IFN stimulation was withdrawn at the time of inoculation. VZV titers were similar in untreated WT, gRNA1, and gRNA4 cells. IFN-α pretreatment led to a marginal decrease in viral titers throughout the 6-day period, and a lack of IRF1 expression did not provide any significant advantage to VZV replication in IFN-α-treated cells. As expected, VZV did not replicate in WT or gRNA4 cells pretreated with IFN-γ. In contrast, VZV replication was detected in IFN-γ-pretreated gRNA1 cells. However, titers were 1.5 logs lower in IFN-γ-pretreated gRNA1 cells than in untreated WT or gRNA4 cells, attributable to the residual expression of IRF1 in 10 to 15% of these cells, allowing a limited IFN-γ effect. This restoration of VZV replication in IRF1KO cells, despite IFN-γ pretreatment, implicated IRF1 as being necessary for mediating the antiviral activity of IFN-γ against VZV.

To complete this analysis of the role of IRF1 in regulating VZV-host cell interactions, we investigated whether VZV interferes with IRF1 protein expression. Consistent with the transcriptional analysis, only a small percentage of uninfected HELFs treated with IFN-α expressed IRF1, while most IFN-γ-treated cells expressed IRF1 at the indicated time point (Fig. 5F). In parallel, HELFs inoculated with VZV-green fluorescent protein (GFP) for 24 h were either untreated or stimulated with IFN-α or IFN-γ for an additional 24 h (Fig. 5F). IRF1 protein expression was evaluated in the uninfected and VZV-infected, untreated cells and IFN-treated cells at 48 h postinfection (hpi) and 24 h posttreatment with either IFN-α or IFN-γ in the presence of MG132. Under all conditions, ∼40% of the cells were infected by VZV. As in the uninfected cells, IFN-α did not induce IRF1 expression after 24 h of treatment in either the bystanders or the VZV-infected cells. Of the ∼40% of cells that were infected in the IFN-γ-treated population, ∼10% had low levels of GFP expression, indicating early VZV infection, and expressed IRF1 protein. IRF1 was not detected in ∼30% of cells with high GFP expression levels, indicating that the more heavily infected cells were refractory to IRF1 induction by IFN-γ. In contrast, IRF1 was induced quite markedly in uninfected bystander cells by IFN-γ, which indicates that reduced IRF1 expression was due to a specific block of its induction in VZV-infected cells. These observations are consistent with the finding that replication can occur if IFN-γ is added following VZV inoculation, since the virus is equipped to block IRF1 induction if it enters the cell prior to IFN-γ exposure (Fig. 2E), and suggest that VZV proteins expressed later in the infectious cycle are likely to mediate IRF1 inhibition.

IRF9 is critical for the IFN-α-mediated host cell defense against VZV.

Although the effect of IFN-α was less potent than that of IFN-γ, IFN-α delayed the spread of VZV infection (Fig. 2). However, since both transcriptome and protein analyses showed minimal IRF1 induction by IFN-α at 24 h posttreatment, we sought to define an alternative host defense pathway for the IFN-α-mediated control of VZV. The tripartite ISGF3 complex, consisting of IRF9, pSTAT1, and pSTAT2, is critical for ISG signaling by IFN-α, and the transcription of several ISGs known to depend on this complex was upregulated in IFN-α-treated HELFs (Fig. 3C). Therefore, we examined protein expression and the activation state of the three ISGF3 components in uninfected and VZV-infected HELFs with or without IFN-α treatment. The IRF9 protein was constitutively expressed in uninfected HELFs and was increased by IFN-α stimulation (Fig. 6A). In contrast, basal IRF9 levels were reduced substantially by VZV infection, and the IRF9 response to IFN-α was blocked in VZV-infected cells. The increase in IRF9 expression induced by IFN-α was not inhibited in cells inoculated with nonreplicating, UV-inactivated VZV, and the constitutive expression of IRF9 was not reduced (Fig. 6A), indicating that the inhibition of IFN-α signaling might require the de novo synthesis of VZV proteins in infected cells.

In order to determine if the VZV regulation of IRF9 was specific for infected as opposed to bystander cells, HELFs were inoculated with recombinant VZV expressing GFP for 48 h. IRF9 levels were assessed in the viable infected and uninfected cells sorted by GFP expression and in an uninfected HELF control. The IRF9 protein was reduced only in infected cells, indicating that inhibition of the IFN-α signaling pathway was not secondary to secreted factors made in response to VZV infection (Fig. 6B).

Since the type I IFN response is necessary to restrict VZV lesion formation, we investigated the potential relevance of IRF9 for IFN-α-mediated control using the SCID mouse model of VZV skin pathogenesis. Human skin xenografts were recovered 14 days after mock inoculation or VZV infection and stained for IRF9 and VZV glycoprotein E (Fig. 5C). In mock-infected skin, IRF9 was detected in the cytoplasm of most cells within the basal epithelium and cells lining the hair follicles; nuclear IRF9 was detected in a few cells within the granule and prickle cell layers of the epidermis (Fig. 6C, panels a and b). In contrast, IRF9 expression was increased dramatically in the nuclei of uninfected cells adjacent to VZV lesions in infected xenografts. Nuclear IRF9 expression was prominent both in the basal epithelium and in the hair follicles of these bystander cells, whereas IRF9 was not detected in VZV lesions (Fig. 6C, panels c to f). The pattern of IRF9 nuclear localization in bystander cells was consistent with the activation of IFN-α signaling in VZV-infected skin tissue (2), and the findings show that viral interference with IRF9 expression within infected skin cells occurs during VZV lesion formation in vivo.

STAT1 and STAT2 phosphorylation, nuclear localization, and heterodimerization are not inhibited during VZV infection.

To further document the critical role of IRF9 in the IFN-α-mediated host cell defense, we next investigated VZV effects on the STAT1 and STAT2 components of the ISGF3 complex. Since both STATs are expressed constitutively in most cells and both must be phosphorylated for functional activation during IFN stimulation, we evaluated the phosphorylation status of the two proteins in the presence of VZV infection. As expected, IFN-α induced the phosphorylation of STAT1 and STAT2 in uninfected HELFs, as shown using flow cytometry to detect pSTAT1-Y701 and pSTAT2-Y689 (Fig. 7A). The kinetics of STAT1 and STAT2 phosphorylation in response to IFN-α in VZV-infected cells did not differ from that in uninfected bystander cells or the uninfected HELF control, showing that VZV did not interfere with STAT1 or STAT2 activation (Fig. 7B). In fact, ∼5% of VZV-infected cells expressed pSTAT1 or pSTAT1/2 without IFN-α stimulation.

To determine whether VZV infection interfered with the nuclear transport of pSTAT1-Y701 and pSTAT2-Y689, cytoplasmic and nuclear fractions were prepared from HELFs that were untreated or treated with IFN-α and mock infected or infected with VZV. As shown in Fig. 8A, IFN-α-treated HELFs showed increased levels of pSTAT1 and pSTAT2 both in the cytoplasm and in the nucleus. The levels of nuclear pSTAT1 and pSTAT2 in VZV-infected cells treated with IFN-α did not differ from those in the controls, indicating that VZV did not block the nuclear translocation of phosphorylated STAT proteins. Low levels of pSTAT1 were also detected in the cytoplasm of infected cells in the absence of IFN-α, confirming the flow cytometry data (Fig. 7B). Finally, pSTAT1-pSTAT2 heterodimers were formed in VZV-infected cells as well as in uninfected cells in response to IFN-α, shown using antibody to total STAT2 for immunoprecipitation and probing for pSTAT1 (Fig. 8B); further analysis by mass spectrometry confirmed the presence of STAT1 in the STAT2 complex in both uninfected and VZV-infected cells treated with IFN-α.

Since pSTAT1 homodimers act as transcriptional units to initiate the primary ISG response to IFN-γ signaling, which includes IRF1 upregulation, and IRF1 is part of the secondary response that triggers the upregulation of IRF1-dependent transcripts in response to IFN-γ, we evaluated the phosphorylation state of STAT1 and STAT2 at early (30 min) and late (24 h) time points after IFN-α and IFN-γ exposure. The effect of VZV infection on their activation state was also evaluated (Fig. 8C). As expected, STAT2 was not activated by IFN-γ at the early or late time points, while pSTAT1 was upregulated by both IFNs at 30 min but only by IFN-γ at 24 h. As was observed with IFN-α, VZV did not interfere with STAT1 phosphorylation in the presence of IFN-γ at the early or late time point.

IRF9 is transcriptionally regulated by the VZV IE62 protein.

IRF9 transcripts were reduced significantly in VZV-infected cells compared to uninfected cells, as determined by RNA-seq (P < 0.001) (Fig. 9A). When measured by reverse transcription (RT)-PCR using four primer pairs spanning different regions of the 1,182-bp IRF9 transcript, low basal levels of IRF9 transcripts were detected in uninfected HELFs, as expected, since the IRF9 protein was constitutively expressed in the cytoplasm (Fig. 6A). While IRF9 transcripts were increased in HELFs treated with poly(I·C) as a positive control, IRF9 transcript levels were reduced in VZV-infected cells (Fig. 9B). When the levels of IRF9 transcripts were evaluated in sorted cell populations, IRF9 transcripts were present at levels 50% lower in VZV-infected cells than in bystander or uninfected cells (Fig. 9C). IRF9 expression did not differ in uninfected and VZV-infected cells, when the proteasomal and lysosomal inhibitors MG132 and chloroquine, respectively, were used to interfere with protein degradation, indicating that IRF9 expression was not regulated by accelerated protein turnover in infected cells (Fig. 9D). These results indicate that the impaired IRF9 protein expression in VZV-infected cells was due to reduced IRF9 gene transcription.

Since countering IFN-α signaling is likely to be important at the initial phase of infection, VZV proteins with transactivating functions that are expressed very early in infected cells were evaluated as candidates for IRF9 regulation. These included IE62, the major VZV-transactivating factor; the ORF61 protein; and IE63, which has been proposed to control IRF9 expression in VZV-infected cells (12). IRF9 protein expression was determined after IFN-α treatment of HEK293 cells transfected with plasmids expressing IE62, the IE61 (ORF61) protein, or IE63 and the controls. IE62-transfected cells showed a >60% reduction in the levels of the IRF9 protein compared to cells transfected with IEF61, IE63, and the cDNA control (Fig. 10A). Further, IE62 inhibited the IFN-α-mediated upregulation of IRF9 transcripts in HEK293 cells (Fig. 10B), supporting its contribution to blocking IRF9 expression. When the effects of each of the three immediate early proteins on IRF9 transcription were assessed, only IE62 reduced IRF9 transcription in response to IFN-α (Fig. 10C). Nevertheless, comparing these effects with changes in IRF9 expression in VZV-infected cells indicates that other viral or host factors also contribute to interference with IRF9.

Taken together, these experiments show that the IRF9 component of the ISGF3 complex is critical for the IFN-α-mediated host cell response to VZV, which is further supported by the finding that VZV targets IRF9 protein expression for inhibition by IE62.

Cell culture.

Human embryonic lung fibroblasts (HELFs) were maintained in Eagle's minimum essential medium (Corning, USA) supplemented with 10% fetal calf serum (FCS) and an antibiotic cocktail of penicillin G and streptomycin (Thermo Fisher Scientific, USA) and amphotericin (Corning, USA). Melanoma (Mel39) cells were propagated in the above-described medium supplemented with nonessential amino acids (Corning, USA). Human embryonic kidney (HEK293) cells were maintained in Dulbecco's modified Eagle's medium (Corning, USA) supplemented with 10% FCS and 100 U/ml each of penicillin G and streptomycin. All cells were maintained and grown at 37°C with a 5% CO₂ concentration.

Virus infection and UV inactivation.

The wild-type pOka strain of VZV (28) and the GFP recombinant VZV (GFP-ORF10) (23) were used in the study. For UV inactivation, VZV-infected HELFs were UV irradiated for 25 min under sterile conditions. Virus inactivation was confirmed by virus titration. A duplicate infected plate prepared in parallel was left untreated and used as a control.

VZV plaque assay.

VZV titers were determined by counting the plaques formed in Mel39 cell monolayers. Infected cells were harvested and resuspended in 1 ml of medium. Tenfold serial dilutions of the infected cells were prepared; subsequently, each serial dilution was added to triplicate wells of a 24-well plate and VZV infection was allowed to proceed for 4 days. On day 4, the cells were fixed in 4% paraformaldehyde (PFA), followed by immunohistochemical detection of VZV proteins using an antibody cocktail (Table 1). Plaques from triplicate wells were counted to determine the VZV titer.

Antibodies and reagents.

The antibodies for immunohistochemistry, Western blotting, and flow cytometry analysis used in the study are listed in Table 1. For cell stimulation using purified recombinant interferons, IFN-α and IFN-γ were purchased from PBL Assay Science and used at the concentrations indicated above. MG132 and chloroquine were purchased from Calbiochem.

Transfection, Western blotting, and coimmunoprecipitation.

For transient-transfection experiments, HEK293 cells were plated 24 h prior to transfection. Two micrograms of each of the plasmids expressing VZV proteins IE62, IE63, and ORF61 was transfected using the Fugene6 reagent (Promega, USA). Whole-cell lysates from transfected cells were prepared by lysing cells in radioimmunoprecipitation assay (RIPA) buffer supplemented with a protease inhibitor (Millipore Sigma, USA). The protein lysates were separated on a 4 to 20% SDS-PAGE gel (Bio-Rad, USA), followed by immunoblotting using the antibodies (Table 1) indicated in the figures. For coimmunoprecipitation, cells were lysed from 10-cm culture dishes at 48 h posttransfection and assayed using a Pierce coimmunoprecipitation kit following the manufacturer's instructions (Thermo Fisher Scientific, USA).

Immunofluorescence (IF).

Uninfected and VZV-infected skin sections were prepared from skin xenografts engrafted in SCID mice as described previously (23). Animal use was approved by the Stanford University Administrative Panel on Laboratory Animal Care. Paraffin-embedded skin sections were deparaffinized and rehydrated, followed by antigen retrieval using the citrate method. The sections were probed for VZV infection using anti-gE antibody (Millipore Sigma, USA) and IRF9-H143 (Santa Cruz Biotechnology, USA). The primary antibodies were detected using donkey anti-rabbit immunoglobulin-Alexa Fluor 488 and goat anti-mouse immunoglobulin-Alexa Fluor 594 (Invitrogen, USA). Images were captured using a Leica SP2 confocal microscope at the Stanford Cell Sciences Imaging Facility, Stanford University.

Flow cytometry and cell sorting.

For flow cytometry analysis of uninfected and VZV-infected HELFs, the cells were trypsinized, washed, and resuspended in fluorescence-activated cell sorting (FACS) buffer (0.5% bovine serum albumin plus 0.02% sodium azide in phosphate-buffered saline). The cells were fixed in 1.6% PFA (Electron Microscopy Sciences, USA) for 10 min, washed, and permeabilized for assaying intracellular proteins by incubation in cold methanol on ice for 30 min. Relevant fluorophore-tagged antibodies (Table 1) were added to a final volume of 100 μl, and the cells were stained for 30 min on ice. The cells were then washed in ice-cold FACS buffer and resuspended in 200 μl of 1:5,000 DAPI (4′,6-diamidino-2-phenylindole) diluted in FACS buffer for viability staining. Data acquisition was done on a BD LSR II.UV instrument (BD Biosciences, USA) at the Stanford Shared FACS Facility, Stanford University. BD CompBeads were used to normalize spectral overlap (BD Biosciences, USA). Data analysis was done using FlowJo software (version 10.0.5).

For sorting, uninfected and GFP-expressing VZV-infected cells were trypsinized, washed, and resuspended in FACS buffer containing DAPI to sort viable infected cells. Cells were sorted on the Carmen apparatus at the Stanford Shared FACS Facility, Stanford University. Equal numbers of cells from uninfected, bystander (GFP-negative), and infected (GFP-positive) populations were used for RNA and protein analysis.

RNA isolation and RT-PCR assay.

RNA was isolated from infected and uninfected HELFs using the TRIzol reagent (Thermo Fisher Scientific, USA) per the manufacturer's instructions. Total RNA (1 μg) was used for DNase I (Invitrogen, USA) digestion, and reverse transcription was done using a first-strand cDNA synthesis kit (Thermo Fisher Scientific, USA). Five percent of the first-strand reaction mixture was used for PCR analysis to detect IRF9 and GAPDH (glyceraldehyde-3-phosphate dehydrogenase) transcripts. The sequences for the IRF9 primers (ELIM Biopharm, USA) used were as follows: for primer pair 1, forward primer TGA GAG GGG CAT CCT AGT GG and reverse primer TCT GCT CCA GCA AGT ATC GG; for primer pair 2, forward primer TAC AGA CAC AAC TGA GGC CC and reverse primer TTA CCT GGA ACT TCG GTG GG; for primer pair 3, forward primer GCT GAG CCC TAC AAG GTG TAT and reverse primer AGA TGA AGG TGA GCA GCA GTG; and for primer pair 4, forward primer GAG CTC TTC AGA ACC GCC T and reverse primer GTC TGC TCC AGC AAG TAT CGG. The GAPDH primers have been described previously (28).

CRISPR-Cas9 IRF1KO using lentivirus.

The IRF1 knockout and control cell lines were generated by transfecting IRF1 CRISPR guide RNA (Table 2) cloned in the pLentiCRISPR vector (GenScript, USA) into HEK293 cells using the Lipofectamine 3000 reagent (Thermo Fisher Scientific, USA). Transfected cells were incubated for 48 h at 37°C in 5% CO₂, following which the supernatant containing lentivirus particles was harvested. Target cells (HELFs or Mel39 cells) were transduced in the presence of 4 μg Polybrene (Millipore Sigma, USA) by spinoculation at 1,800 rpm for 45 min at room temperature. The transduced cells were selected by puromycin (Thermo Fisher Scientific, USA) at 10 μg/well of a 6-well plate, and gene knockout was confirmed by FACS analysis for IRF1 expression.

RNA-seq analysis.

Three biological replicates of four conditions including uninfected and untreated HELFs (control), IFN-α- and IFN-γ-treated HELFs, and VZV-infected HELFs were prepared for RNA sequencing. RNA was purified using the TRIzol reagent per the manufacturer's instructions (Thermo Fisher Scientific, USA). RNA library preparation and sequencing on the Illumina NextSeq (Illumina, USA) platform were performed at the Stanford Functional Genomics Facility at Stanford University. The resulting RNA-seq counts per gene were analyzed by closely following the RNAseq123 work flow (29), using tools from the Bioconductor project (30). Filtering, normalization, and modeling were done using methods from edgeR (31) and limma (32). Filtering was performed on log count per million (log CPM)-transformed data to remove low-expression genes across samples. Subsequently, the samples were trimmed mean of M values (TMM) normalized to account for batch effects and the different RNA compositions between the samples (33). The mean variance independence condition for linear modeling was achieved using precision weights, calculated by the voom method (34). Finally, empirical Bayes moderation was used to calculate gene-wise variance estimates by borrowing information from across all genes, yielding the moderated t test statistics used to determine differential expression (35). In addition to meeting the Benjamini-Hochberg-adjusted P value threshold of 0.05, differentially expressed genes from the linear model were considered significant under conservative conditions of a minimum of a 2-fold change in expression between two experimental conditions (36).