Interferon Regulatory Factor 3 Is Necessary for Induction of Antiviral Genes during Human Cytomegalovirus Infection

Abstract

Viral infection activates interferon regulatory factor 3 (IRF3), a cofactor for the induction of interferon-stimulated genes (ISGs). The role of IRF3 in the activation of ISGs by human cytomegalovirus (HCMV) is controversial despite the fact that HCMV has consistently been shown to induce ISGs during infection of fibroblasts. To address the function of IRF3 in HCMV-mediated ISG induction, we monitored ISG expression and global gene expression in HCMV-infected cells in which IRF3 function had been depleted by small interfering RNA or blocked by dominant negative IRF3. A specific reduction of ISG induction was observed, whereas other transcripts were unaffected. We therefore conclude that IRF3 specifically regulates ISG induction during the initial phase of HCMV infection.

Proinflammatory and interferon (IFN)-stimulated genes (ISGs) represent essential components of the innate immune response to viral infection. Upon viral entry into cells, ISG induction occurs in two waves: acute, IFN-independent induction of a subset of ISGs and delayed, IFN-dependent induction via the production of alpha/beta IFN during the initial phase (9). In vitro infection of human cells with the betaherpesvirus human cytomegalovirus (HCMV) has been shown to rapidly elicit ISG transcription (1, 4, 19, 25). This induction is due largely to IFN-independent mechanisms since it also occurs, and is even enhanced, in the presence of protein synthesis inhibitors (4). During many viral infections, IFN-independent ISG induction is mediated by IFN regulatory factor 3 (IRF3), a constitutively expressed transcriptional coactivator that is activated following phosphorylation of carboxy-terminal serine residues by the virus-stimulated kinases IKKɛ and TBK1 (7, 17, 18). Activated IRF3 homodimerizes, complexes with transcriptional coactivators p300 and CBP, and accumulates in the nucleus (21, 24). These complexes bind to positive regulatory domains, leading to increased transcription of a subset of ISGs, including beta IFN (IFN-β), which then initiates IFN-dependent ISG induction via the IFN receptor and JAK/STAT signaling (8, 22).

Several investigators have shown that HCMV stimulates assembly and nuclear accumulation of a transcriptional complex containing IRF3 (2, 5, 13, 16). However, recent observations question whether IRF3 is activated during HCMV infection and whether IRF3 is responsible for IFN-independent ISG induction by HCMV. Abate et al. did not observe IRF3 activation in HCMV-infected fibroblasts but reported a significant induction of ISGs (1). Furthermore, the protein kinase C inhibitor H7 has been shown to prevent ISG induction by HCMV even when IRF3 nuclear localization occurs (14). Employing the IFN-stimulated response element from isg54, Yang and colleagues were unable to pull down protein complexes of sizes comparable to those predicted to contain IRF3 during HCMV infection (23). These observations raise the question of whether or not ISG induction during HCMV infection depends on IRF3 activation.

To address this question, we performed a “loss-of-function” forward genetics approach using small interfering RNA (siRNA) directed against IRF3 and subsequently monitoring ISG induction by HCMV. Primary human foreskin fibroblasts (HFs) were transfected with siRNA targeting IRF3 mRNA or, as a negative control, siRNA directed against an mRNA target from a different virus (Kaposi's sarcoma-associated herpesvirus open reading frame K5 [ORF K5]). Knockdown of IRF3 protein expression was examined using immunoblotting. Figure 1A illustrates the nearly complete absence of IRF3 protein at 7 days posttransfection of IRF3 siRNA compared to the level in untreated cells or cells transfected with control siRNA. Worth noting is the presence of a third, presumably differentially phosphorylated form of IRF3 in our uninfected samples, an observation that seemingly departs from other studies involving this cell type (1, 2). Yet, IRF3 is not constitutively activated in our sample cells based on indirect immunofluorescence assay (Fig. 1B) and ISG transcriptional analysis (see the supplemental material). To examine whether siRNA-transfected cells could be infected with HCMV, we monitored viral gene expression using indirect immunofluorescence. HFs treated with IRF3 or control siRNA were infected for 6 h with HCMV strain AD169 (multiplicity of infection, 3). As shown in Fig. 1B, HCMV ORF UL123 was expressed following transfection of either siRNA. Consistent with previous observations, IRF3 was activated in control siRNA transfectants, as demonstrated by its nuclear accumulation, whereas IRF3 was nearly undetectable in IRF3 siRNA transfectants.

FIG. 1.

siRNA diminishes IRF3 protein expression and nuclear accumulation. HFs were transfected with control or IRF3-specific (Dharmacon catalog no. MU-006875-01) siRNA using Oligofectamine (Invitrogen) or left untreated. (A) At 7 days posttransfection, whole-cell lysates were run on a 7% sodium dodecyl sulfate gel and stained with antibodies for calreticulin (Stressgen catalog no. SPA-601) and IRF3 (Santa Cruz catalog no. SC-9082). (B) Cells were left untreated or infected with HCMV for 6 h at 7 days posttransfection as described in the text and stained using indirect immunofluorescence for HCMV ORF UL123 (Jay Nelson) and IRF3 (BD Pharmingen catalog no. 550428). DAPI, 4^′,6^′-diamidino-2-phenylindole; α, anti.

To examine whether IRF3 knockdown interfered with HCMV-mediated ISG induction, the following ISG transcripts were compared by semiquantitative real-time reverse transcription-PCR (qPCR) as described in references 5 and 12: ISG54 (accession no. NM_001547), ISG56 (NM_001548), ISG60 (NM_001549), IFI-15K (NM_005101), and IFN-β (NM_002176). As expected, HCMV strongly induced these genes in the presence of control siRNA. However, ISG mRNA levels were drastically decreased in HFs treated with IRF3 siRNA despite HCMV infection (Fig. 2). A common difficulty in siRNA experiments is the unplanned stimulation of innate immune genes (20). However, neither siRNA induced this response as indicated by levels of ISG mRNA (Fig. 2). Thus, there was a clear correlation between the presence or absence of IRF3 and the ability or inability of fibroblasts to express ISGs in response to HCMV.

FIG. 2.

Mean relative levels of expression (changes [n-fold]) ± standard deviations of five ISGs versus expression in uninfected, untreated HFs.

To confirm that the observed IRF3-dependent induction of ISGs occurred independently of IFN, we performed the same series of experiments in the presence of cycloheximide to prevent the translation of IFN. IRF3 siRNA severely diminished HCMV-induced ISG transcription in the presence of cycloheximide compared to the control siRNA, indicating that this induction was IFN independent (Fig. 2).

To independently confirm the IRF3 dependence of ISG induction, we employed a dominant negative form of IRF3 (IRF3-ΔN) from which the DNA binding domain was removed (10). IRF3-ΔN was expressed by adenovirus transduction under the control of a tet-regulated promoter. At 24 hours postinfection (hpi), HFs were infected with HCMV, and ISG induction was monitored by qPCR. In the absence of HCMV, recombinant adenovirus infection did not significantly induce ISG expression, whereas HCMV infection of cells containing control adenovirus resulted in ISG induction similar to that with HCMV infection alone (Fig. 2). In contrast, HCMV infection in the presence of Ad-IRF3-ΔN resulted in strongly diminished ISG induction, consistent with results obtained by siRNA transfection. Taken together, these results unequivocally link IRF3 activation to ISG induction by HCMV. Previous transcriptional profiling experiments suggested that IFN-inducible genes represent a major fraction of the cellular response to HCMV (4, 19). It therefore seemed conceivable that IRF3 activation plays a pivotal role in the global transcriptomic changes occurring upon HCMV infection. To examine the contribution of IRF3 to changes elicited by HCMV, we analyzed gene expression profiles of HCMV-infected cells in the presence of IRF3 siRNA using oligonucleotide microarrays (Affymetrix U133 Plus 2.0) displaying probe sets for 47,000 individual transcripts.

HFs were transfected with IRF3 siRNA or control siRNA, and gene knockdown was confirmed with immunoblotting (not shown). At 7 days posttransfection, cells were infected with HCMV as described above or mock infected, and total RNA was harvested at 4, 8, and 24 hpi. mRNA was labeled and hybridized to microarrays according to Affymetrix protocols, and data were analyzed using GeneChip operating software (Affymetrix) and ArrayAssist software (Stratagene). As expected, massive changes in the host cell transcriptome were recorded upon infection of control siRNA-treated HFs (Fig. 3) (see Table S1 in the supplemental material). Interestingly, massive transcriptional changes also occurred upon infection of HFs transfected with IRF3 siRNA (Fig. 3) (see Table S2 in the supplemental material). This finding could indicate that depletion of IRF3 did not influence the majority of HCMV-induced changes. Comparison of IRF3 siRNA-treated and control siRNA-treated cells supports this conclusion since the gene expression profiles were highly correlated (Fig. 3). In uninfected cells, only minor differences were observed between IRF3 siRNA-treated and control siRNA-treated cells, suggesting that, without infection, neither treatment significantly affected transcript levels more than the other. In HCMV-infected cells, however, several transcripts were significantly underrepresented in the IRF3 siRNA-treated samples, particularly at 4 and 8 hpi (Fig. 3). A closer look at these transcripts reveals that they consist predominantly of known ISGs (Table 1). Compared to genes in uninfected cells, these genes are still induced upon IRF3 depletion but to a much lesser extent than in control siRNA-treated cells. The remaining induction of these genes is likely due to residual IRF3 expression rather than IRF3-independent induction, as indicated by the fact that the IRF3 message was not completely shut off but reduced two- to threefold. The resulting residual secretion of IFN-β is likely the reason there was no effect of IRF3 depletion at 24 h, since at this time, IFN-dependent, and thus IRF3-independent, pathways dominate. Together with the qPCR results, global expression analysis thus supports the conclusion that IRF3 is essential for the induction of ISGs in HCMV-infected cells, particularly during the early phase of infection. However, the role of IRF3 is clearly confined to regulating ISGs, whereas HCMV changes the transcriptional profile of many other genes in an IRF3-independent manner.

FIG. 3.

Effect of siRNA on global HCMV-induced changes in transcription. HFs were treated with IRF3 or control siRNA and either infected with HCMV or mock infected for 4 h, 8 h, and 24 h as described in the text. Shown are individual hybridization intensities from a complete probe set of Affymetrix U133 v2.0 GeneChips. Purple, repressed; light blue, upregulated.

TABLE 1.

HCMV-mediated host genes differentially regulated (absolute change [n = fold] of at least twofold) in the presence of siRNA at 4 h and 8 h

^aHighlighted gene titles are known ISGs or immune-related genes.

^bTranscript changes (n-fold) are a quotient derived from probe set hybridization intensities of samples infected with HCMV or not infected (mock) and treated with IRF3-specific or nonspecific (control) siRNA. Highlighted values are statistically significant.

One of the genes underrepresented in HCMV-infected and IRF3 siRNA-treated cells was the prostaglandin synthetase COX2. Previously, it was shown that COX2 inhibitors interfere with the growth of HCMV, a defect that could be restored upon the addition of prostaglandins (26). We were therefore interested in whether the reduction of COX2 induction by IRF3 siRNA had an effect on HCMV replication in fibroblasts. However, analysis of HCMV growth by single-step growth curves revealed the opposite trend, i.e., HCMV appeared to grow better in the absence of IRF3 (see Fig. S1 in the supplemental material). This could indicate that the antiviral effect of most genes regulated by IRF3 negatively influences HCMV infection compared to the growth-promoting effect of some IRF3-regulated genes, such as Cox-2.

Our demonstration that IRF3 is essential for IFN-independent ISG induction by HCMV in HFs implies that the lack of IRF3 activation or binding observed by others may have had a technical basis, since IRF3-dependent ISGs were clearly induced under the conditions employed (1, 23). Glycoprotein B activates intracellular signaling pathways, including IRF3, by an unknown mechanism during viral entry at the step of membrane fusion (2, 15). Thus, our data are consistent with glycoprotein B-mediated ISG induction occurring via IRF3 activation. Using rat vascular smooth muscle cells, Gravel and Servant (8) showed via qPCR that HCMV-induced transcription of ccl5 and CxCl10 (both identified here as IRF3 dependent) was diminished in the presence of stably expressed, dominant negative IRF3 (11). Those authors also demonstrated the dependence of IRF3 phosphorylation on HCMV-induced kinase activity by TBK1 and IKKɛ, further strengthening the essential role of IRF3 in ISG transcription during HCMV infection. As demonstrated here for HCMV, ISG induction was linked to IRF3 activation in Sendai virus-infected cells by similar approaches (6). In contrast to that in HCMV-infected fibroblasts, IRF3 was not activated in rhesus CMV (RhCMV)-infected fibroblasts, and this corresponded to a complete lack of ISG induction, suggesting that RhCMV interferes efficiently with IRF3 activation (5). Interestingly, a more pronounced induction of ISGs and proinflammatory genes was reported for HCMV lacking the major tegument protein pp65 (1, 3). This was linked to interference with the activation of NF-κB (3) or IRF3 (1). While our data show that IRF3 is clearly activated and functional in HCMV-infected cells, they do not rule out the fact that HCMV partially interferes with IRF3 activation. Moreover, activation of IRF3 and NF-κB involves related signaling pathways (7, 18), and these transcription factors cooperate during the induction of ISGs, particularly IFN and proinflammatory genes. Our observation that HCMV tends to grow better upon reduction of the ISG response is consistent with a selective growth advantage for viruses interfering with innate immunity. As with other HCMV-modulated immune pathways, however, it seems that a balance is established that, in the case of HCMV but not RhCMV, favors the activation of this innate response via the induction of IRF3.