Inhibition of Viral Pathogenesis and Promotion of the Septic Shock Response to Bacterial Infection by IRF-3 Are Regulated by the Acetylation and Phosphorylation of Its Coactivators

Whereas TLR3 signals exclusively from the endosomal membrane, TLR4 signals from both the plasma membrane and the endosomal membrane, and the endosomal TLR3 and TLR4 signaling pathways partially overlap (23). The two major transcription factors activated by signaling from both TLRs are NF-κB and IRF-3, which induce transcription of two sets of genes individually or of a third set, which includes IFN-β, by acting together. To examine whether PKC is required for TLR3 signaling, we used the human cell line HT1080 and a universal PKC inhibitor, Gö6976. Microarray gene expression profiling revealed that induction of IRF-3-driven genes (see Fig. S1A, left panel, in the supplemental material), but not NF-κB-driven genes (Fig. S1A, right panel), was strongly inhibited. Western blotting for selective proteins, such as P56/IFIT1 (Fig. S1B), confirmed the conclusion. To distinguish among the various isoforms of PKC, isoform-specific inhibitors were used. A PKC-β-specific inhibitor was as effective as the universal inhibitor (Fig. 1A), indicating that PKC-β was required for IRF-3-driven gene induction. For testing the requirement of HDAC6 in TLR3 signaling, its expression was knocked down by a specific short hairpin RNA (shRNA). It caused a loss of induction of IFIT1 by TLR3 (Fig. 1B), an effect that was shared by many IRF-3-induced genes (Fig. S1C). However, there was no global inhibition of gene induction; induction of an NF-κB-driven gene, A20, by tumor necrosis factor alpha (TNF-α) was unimpaired (Fig. S1D). More strikingly, although the induction of IFIT1 by TLR3 signaling was inhibited in the absence of HDAC6 (Fig. 1B), induction of the same gene by beta interferon (IFN-β) was unimpaired (Fig. 1C). This result indicates that the requirement of HDAC6 is signaling pathway specific. Requirement of HDAC6 was further investigated in the cytoplasmic RIG-I-like helicase (RLH)-mediated transcriptional activation of IRF-3. Induction of IRF-3-dependent genes by RLH activation was inhibited in HDAC6 knockdown human cells (Fig. 1D) and HDAC6^−/− mouse embryonic fibroblasts (MEFs) (Fig. 1E). To expand the significance of our observations, we used primary myeloid cells from wild-type (wt) and HDAC6^−/− mice (24). Splenocytes or bone marrow-derived macrophages (BMDMs) were treated with dsRNA for triggering TLR3 signaling or with LPS for triggering TLR4 signaling. Both pathways induced mouse Ifit1 mRNA in wt BMDMs but not in HDAC6^−/− cells (Fig. 1F). Similar results were obtained in splenocytes when mouse Ifit2 induction was measured at the protein level (Fig.  1G). Although Ifit1 and Ifit2 induction by TLR4 signaling was impaired in HDAC6^−/− cells, the induction of NF-κB-driven genes, such as those for interleukin 6 (IL-6) and TNF-α, was unaffected (Fig.  1H; see also Fig.  S1E in the supplemental material). These results demonstrate that the need for HDAC6 was not only signaling pathway specific but also transcription factor specific.

Once we identified the affected genes to be IRF-3 driven, we inquired at which step of its activation or action PKC-β and HDAC6 were needed. Inhibition of HDAC6 activity by trichostatin A (TSA), a universal HDAC inhibitor, did not impair IRF-3 activation by TLR3 signaling, as measured by its specific phosphorylation, although, as expected, IFIT1 induction was inhibited (Fig.  2A). Similarly, HDAC6 knockdown did not impair nuclear translocation of activated IRF-3 (Fig.  2B). The same was true for cells treated with general or the PKC-β-specific inhibitors (Fig. 2C). Although activated IRF-3 was present in the nuclei of PKC-inhibited cells after TLR3 stimulation, it did not occupy the promoter of the IFIT1 gene, as revealed by chromatin immunoprecipitation (ChIP) assay. Both IRF-3 and polymerase II (Pol II) were present at the IFIT1 promoter region in cells treated with poly(I⋅C) but only in the absence of the PKC inhibitor (Fig.  2D). Similar results were obtained by comparing wt and HDAC6 knockdown cells; upon TLR3-stimulation by poly(I⋅C), IRF-3 and Pol II were bound to the promoter of the IFIT1 gene only in wt cells (Fig.  2E). Remarkably, when cells were stimulated with IFN-β, which uses ISGF3, not IRF-3, to induce the IFIT1 gene, STAT2, a component of ISGF3, and Pol II were bound to the promoter even in the cells lacking HDAC6 (Fig.  2F). Thus, the results from the ChIP assays mirrored those from the gene expression analyses. The results presented in Fig.  2 demonstrated that both PKC-β and HDAC6 were required, not for IRF-3 activation and nuclear translocation, but to form a stable initiation complex at the promoters of target genes. Moreover, the effects appeared to be transcription factor specific and not mediated by any alterations of chromatin structures or the transcription machinery, because the same gene could be induced normally by a different transcription factor, one that recognizes the same promoter sequence.

As a transcription factor, IRF-3 uses CBP as the coactivator. As revealed by their coimmunoprecipitation (Co-IP), the two proteins physically interacted upon IRF-3 activation by TLR3 signaling; this interaction did not occur in cells treated with an HDAC inhibitor (Fig.  3A) or in HDAC6 knockdown cells (Fig.  3B). To identify the substrate of HDAC6 in this context, we did extensive mass spectrometric analysis of IRF-3 isolated from unstimulated and stimulated cells for its possible acetylation and deacetylation; these efforts could not demonstrate that IRF-3 is acetylated under any situation (results not shown). However, in light of the discovery of the need of β-catenin for IRF-3 activity (21), we focused our attention on the acetylation status of β-catenin. We confirmed that β-catenin was needed for IFIT1 induction by TLR3 signaling (Fig.  3C), and as expected, it was strongly acetylated in HDAC6 knockdown cells (Fig.  3D). Moreover, in the absence of β-catenin, there was no interaction between IRF-3 and CBP (Fig.  3E), indicating that β-catenin bridges the two proteins. To examine possible synergistic gene-inducing actions between IRF-3 and β-catenin, cells were treated with various Wnt signaling inducers that are known to increase cellular β-catenin levels (Fig.  3F); all of these treatments augmented IFIT1 induction by TLR3 signaling (Fig.  3G). The above results indicate that both the level and the acetylation status of β-catenin regulate the extent of IRF-3-mediated gene induction by regulating the IRF-3/CBP interaction.

Knowing that β-catenin provides a bridge for IRF-3 to CBP, we investigated whether HDAC6 and PKC-β regulate the IRF-3/β-catenin interaction, the β-catenin/CBP interaction, or both. When HDAC6 expression was knocked down, CBP did not bind to β-catenin (Fig.  4A) but the IRF-3/β-catenin interaction was not affected (Fig.  4B). Similarly, CBP did not interact with β-catenin when PKC-β was inhibited (Fig.  4C), although the IRF-3/β-catenin interaction was maintained (Fig. 4D). These results indicated that the regulation of IRF-3’s action by the two enzymes was exerted at the level of the β-catenin/CBP interaction and not the IRF-3/β-catenin interaction. To demonstrate that β-catenin was the target of HDAC6, we transiently expressed a β-catenin mutant that cannot be acetylated (K49R mutant). Unlike wt β-catenin, the mutant could interact with CBP, even in the absence of HDAC6 (Fig.  4E), and mediate IRF-3-driven gene induction (Fig.  4F). On the other hand, the inhibitor of PKC-β still inhibited the interaction between CBP and the β-catenin mutant (Fig.  4G), suggesting that the target of PKC-β was CBP.

To evaluate the impact of HDAC6-mediated regulation of IRF-3 function in vivo, we chose two pathogenesis models. Bacterial LPS triggers rapid, TLR4-mediated septic shock in mice. As expected, TRIF^−/− mice were much less susceptible to TLR4-mediated septic shock (Fig.  5A) because the relevant signaling pathway of TLR4 requires TRIF. Similarly, IRF-3^−/− mice were resistant to LPS treatment (Fig.  5A). In the reciprocal model, IRF-3 protected mice from pathogenesis caused by Sendai virus infection (Fig.  5B). To test the need of HDAC6 in the above-described in vivo actions of IRF-3, we used HDAC6^−/− mice. Like IRF-3^−/− mice, HDAC6^−/− mice were less susceptible to LPS-induced septic shock (Fig.  5C) and more susceptible to SeV-induced viral pathogenesis (Fig.  5D) than wt mice, suggesting that IRF-3, although present, was nonfunctional in these mice. For both disease models, we observed minor differences in the kinetics of susceptibility between IRF-3^−/− and HDAC6^−/− mice. This was most probably because of the background strain difference between them; the IRF-3 mice were strain Bl6, whereas the HDAC6 mice were strain 129, and the cognate wt mice were used for comparison in each experiment. The above results clearly demonstrated that HDAC6 was required for IRF-3 actions in vivo.

Human fibrosarcoma cell line HT1080 has been reported previously (31). HT1080-derived cells were generated as described below. HDAC6^−/− and matched wt control MEFs were kindly provided by Tso-Pang Yao (Duke University). All the cell lines were maintained in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal bovine serum (FBS), 100 units/ml of penicillin, and 100 µg/ml of streptomycin. Primary bone marrow-derived macrophages (BMDMs) and splenocytes were isolated using a previously described procedure, and the cells were cultured in DMEM supplemented with 10% FBS (14). A V5-tagged human IRF-3 expression vector was described previously (14). A Flag-tagged human wt β-catenin plasmid was obtained from Addgene, and a β-catenin K49R mutant was generated by PCR mutagenesis. Antibodies against human IFIT1/P56 and murine Ifit2/P54 were raised in rabbits by injection of purified full-length proteins by the Hybridoma Core, Lerner Research Institute (32). Antibodies against phospho-S396–IRF-3, β-catenin, and acetylated β-catenin were from Cell Signaling, antibodies against IRF-3, HDAC6, and CBP (C1) were from Santa Cruz, A20 antibody was from Imgenex, V5 antibody was from Invitrogen, and Flag antibody was from Sigma-Aldrich. Poly(I⋅C) was obtained from GE Healthcare, lipopolysaccharide from Escherichia coli O55:B5 and LiCl were obtained from Sigma-Aldrich, trichostatin A (TSA), Gӧ6976, PKC-β inhibitor, Wnt3A, and SB216763 were from Calbiochem. Human IFN-β (Calbiochem) was applied to cells by adding it to the culture media at a final concentration of 1,000 units/ml. Fugene 6 was obtained from Roche.

Western blotting was performed using previously described procedures (14). 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. For coimmunoprecipitation assays, the ExactaCruz system (Santa Cruz) was used according to the manufacturer’s instructions. Briefly, the cell lysates were precleared with preclearing matrix (Santa Cruz) for 1 h, precleared lysates were incubated with antibody-conjugated IP matrix overnight, and the beads were washed with phosphate-buffered saline (PBS) and lysis buffer, boiled with SDS-PAGE buffer, and analyzed by Western blotting.

HT1080 cells were transduced with lentivirus generated using pLKO1 shRNA plasmids (Sigma Mission shRNA), encoding shRNA against human HDAC6 (TRCN0000004843, shHDAC6) or nontargeting shRNA (SHC002, shCon). After 48 h, cells were maintained in puromycin (1 µg/ml)-containing medium. For knockdown of β-catenin, the small interfering RNA (siRNA) against human β-catenin (L-003482-00-0010) was obtained from Thermo Scientific and was transfected using DharmaFECT 4. Cells were analyzed as described in the figure legends. A nontargeting siRNA (Thermo Scientific; D-001810-10-05) was used as a control and was transfected using the same protocol as described above.

For the quantitative profiling of mRNA levels by microarray analysis, HT1080 cells were left untreated or pretreated with 5 µM Gö6976 (Calbiochem) for 1 h, after which 100 µg/ml poly(I⋅C) was added to the culture medium for 6 h. shRNA-expressing cells were left untreated or treated with 100 µg/ml poly(I⋅C) for 6 h. Total RNA was extracted using TRIzol reagent (Invitrogen). After DNase I treatment (DNA free; Applied Biosystems/Ambion), RNA was cleaned up using spin columns (RNeasy minikit; Qiagen) before subjection to mRNA expression microarray analysis by hybridization of cRNA to an Illumina HumanRef-8 BeadChip. Results were analyzed using GenomeStudio software (v2010.2; Illumina, Inc.). cRNA hybridization to chips was performed by the Lerner Research Institute’s Genomics Core. For reverse transcription-PCR (RT-PCR), RT of the RNA was performed with the Superscript III kit (Invitrogen). PCRs were driven by Hot Start Taq polymerase (Denville). The RT-PCR primers used were as follows: for Ifit1, 5′ CAGAAGCACACATTGAAGAA and 3′ TGTAAGTAGCCAGAGGAAGG; for IL-6, 5′ ATGAAGTTCCTCTCTGCAAGAGACT and 3′ CTAGGTTTGCCGAGTAGATCTC; for TNF-α, 5′ TGGAACTGGCAGAAGAGGCACT and 3′ GAGATAGCAAATCGGCTGACGG; and for 18S rRNA, 5′ ATTGACGGAAGGGCACCACCAG and 3′ CAAATCGCTCCACCAACTAAGAACG.

HT1080 or derived cells in 150-mm dishes were left untreated or pretreated for 1 h with 5 µM Gö6976 and then treated with 100 µg/ml poly(I⋅C) added to the media for 2 h. DNA and protein were cross-linked with 1% formaldehyde for 10 min, and chromatin was prepared and sheared (Misonix 3000 microtip) using reagents and directions from the ChIP-IT express kit (Active Motif). The sheared chromatin (25 µg) was subjected to immunoprecipitation with anti-IRF-3 (sc-9082X), anti-Pol II N-20 (sc-899X), or anti-STAT2 C-20 (sc-476X), all from Santa Cruz, or with nonspecific normal rabbit IgG as a control (Sigma). After un-cross-linking and proteinase K digestion of proteins in the precipitation reaction or of input chromatin as a control, coprecipitated genomic IFIT1 promoter DNA was detected by PCR amplification of the 125-bp Interferon Stimulated Response Element (ISRE) plus TATA box region of the IFIT1 promoter. Primers used were 5′ GAATTCCGCTAGCTTTAGTTTCAC and 3′ CCCCAAGACAGTGTTATATAAGGG.

IRF-3^−/− mice (C57BL/6 background) were obtained from Riken Bio Resource Center, Japan (with permission from Tadatsugu Taniguchi, University of Tokyo), TRIF^−/− mice (C57BL/6 background) were obtained from Jackson Laboratories, and HDAC6^−/− mice (129s background) were kindly provided by Timothy McKinsey (University of Colorado, Denver, CO). For LPS treatment, 8- to 10-week-old mice were injected with LPS (1 mg/mouse) intraperitoneally, and their survival was monitored for 1 week. Sendai virus (strain 52) was obtained from ATCC and was grown by Charles River Laboratories. For virus infections, 1,000 PFU of SeV in 35 µl of endotoxin-free PBS was intranasally administered to isoflurane-anesthetized 8- to 10-week-old mice. The mice were monitored daily for their weight loss and disease symptoms. All the animal procedures were performed according to the protocols approved by the institutional animal care and use committee.

All microarray data have been deposited in the NCBI Gene Expression Omnibus (GEO) database (http://www.ncbi.nlm.nih.gov/geo) under accession numbers GSE43217 and GSE43218.