Article24 September 2013Open Access Source Data IFNβ-dependent increases in STAT1, STAT2, and IRF9 mediate resistance to viruses and DNA damage HyeonJoo Cheon HyeonJoo Cheon Department of Molecular Genetics, Lerner Research Institute, Cleveland Clinic, Cleveland, OH, USA Search for more papers by this author Elise G Holvey-Bates Elise G Holvey-Bates Department of Molecular Genetics, Lerner Research Institute, Cleveland Clinic, Cleveland, OH, USA Search for more papers by this author John W Schoggins John W Schoggins Laboratory of Virology and Infectious Disease, Center for the Study of Hepatitis C, Rockefeller University, New York, NY, USA Search for more papers by this author Samuel Forster Samuel Forster Centre for Innate Immunity and Infectious Disease, Monash Institute of Medical Research, Monash University, Clayton, Victoria, Australia Search for more papers by this author Paul Hertzog Paul Hertzog Centre for Innate Immunity and Infectious Disease, Monash Institute of Medical Research, Monash University, Clayton, Victoria, Australia Search for more papers by this author Naoko Imanaka Naoko Imanaka Laboratory of Virology and Infectious Disease, Center for the Study of Hepatitis C, Rockefeller University, New York, NY, USA Search for more papers by this author Charles M Rice Charles M Rice Laboratory of Virology and Infectious Disease, Center for the Study of Hepatitis C, Rockefeller University, New York, NY, USA Search for more papers by this author Mark W Jackson Mark W Jackson Department of Pathology, Case Comprehensive Cancer Center, Case Western Reserve University, Cleveland, OH, USA Search for more papers by this author Damian J Junk Damian J Junk Department of Pathology, Case Comprehensive Cancer Center, Case Western Reserve University, Cleveland, OH, USA Search for more papers by this author George R Stark Corresponding Author George R Stark Department of Molecular Genetics, Lerner Research Institute, Cleveland Clinic, Cleveland, OH, USA Search for more papers by this author HyeonJoo Cheon HyeonJoo Cheon Department of Molecular Genetics, Lerner Research Institute, Cleveland Clinic, Cleveland, OH, USA Search for more papers by this author Elise G Holvey-Bates Elise G Holvey-Bates Department of Molecular Genetics, Lerner Research Institute, Cleveland Clinic, Cleveland, OH, USA Search for more papers by this author John W Schoggins John W Schoggins Laboratory of Virology and Infectious Disease, Center for the Study of Hepatitis C, Rockefeller University, New York, NY, USA Search for more papers by this author Samuel Forster Samuel Forster Centre for Innate Immunity and Infectious Disease, Monash Institute of Medical Research, Monash University, Clayton, Victoria, Australia Search for more papers by this author Paul Hertzog Paul Hertzog Centre for Innate Immunity and Infectious Disease, Monash Institute of Medical Research, Monash University, Clayton, Victoria, Australia Search for more papers by this author Naoko Imanaka Naoko Imanaka Laboratory of Virology and Infectious Disease, Center for the Study of Hepatitis C, Rockefeller University, New York, NY, USA Search for more papers by this author Charles M Rice Charles M Rice Laboratory of Virology and Infectious Disease, Center for the Study of Hepatitis C, Rockefeller University, New York, NY, USA Search for more papers by this author Mark W Jackson Mark W Jackson Department of Pathology, Case Comprehensive Cancer Center, Case Western Reserve University, Cleveland, OH, USA Search for more papers by this author Damian J Junk Damian J Junk Department of Pathology, Case Comprehensive Cancer Center, Case Western Reserve University, Cleveland, OH, USA Search for more papers by this author George R Stark Corresponding Author George R Stark Department of Molecular Genetics, Lerner Research Institute, Cleveland Clinic, Cleveland, OH, USA Search for more papers by this author Author Information HyeonJoo Cheon1, Elise G Holvey-Bates1, John W Schoggins2, Samuel Forster3, Paul Hertzog3, Naoko Imanaka2, Charles M Rice2, Mark W Jackson4, Damian J Junk4 and George R Stark 1 1Department of Molecular Genetics, Lerner Research Institute, Cleveland Clinic, Cleveland, OH, USA 2Laboratory of Virology and Infectious Disease, Center for the Study of Hepatitis C, Rockefeller University, New York, NY, USA 3Centre for Innate Immunity and Infectious Disease, Monash Institute of Medical Research, Monash University, Clayton, Victoria, Australia 4Department of Pathology, Case Comprehensive Cancer Center, Case Western Reserve University, Cleveland, OH, USA *Corresponding author. Department of Molecular Genetics, Lerner Research Institute, Cleveland Clinic, 9500 Euclid Avenue, Cleveland, OH 44195, USA. Tel.:+1 216 444 6062; Fax:+1 216 444 0512; E-mail: [email protected] The EMBO Journal (2013)32:2751-2763https://doi.org/10.1038/emboj.2013.203 PDFDownload PDF of article text and main figures. Peer ReviewDownload a summary of the editorial decision process including editorial decision letters, reviewer comments and author responses to feedback. ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions ShareFacebookTwitterLinked InMendeleyWechatReddit Figures & Info A single high dose of interferon-β (IFNβ) activates powerful cellular responses, in which many anti-viral, pro-apoptotic, and anti-proliferative proteins are highly expressed. Since some of these proteins are deleterious, cells downregulate this initial response rapidly. However, the expression of many anti-viral proteins that do no harm is sustained, prolonging a substantial part of the initial anti-viral response for days and also providing resistance to DNA damage. While the transcription factor ISGF3 (IRF9 and tyrosine-phosphorylated STATs 1 and 2) drives the first rapid response phase, the related factor un-phosphorylated ISGF3 (U-ISGF3), formed by IFNβ-induced high levels of IRF9 and STATs 1 and 2 without tyrosine phosphorylation, drives the second prolonged response. The U-ISGF3-induced anti-viral genes that show prolonged expression are driven by distinct IFN stimulated response elements (ISREs). Continuous exposure of cells to a low level of IFNβ, often seen in cancers, leads to steady-state increased expression of only the U-ISGF3-dependent proteins, with no sustained increase in other IFNβ-induced proteins, and to constitutive resistance to DNA damage. Introduction Type I interferons (IFNs α and β) drive the expression of genes that encode proteins with anti-viral, anti-proliferative, pro-apoptotic, and pro-inflammatory functions. Several important negative feedback mechanisms collaborate to terminate the expression of these genes several hours after IFN stimulation, for example, expression of the potent negative regulator SOCS1 is rapidly induced by IFN (Yoshimura et al, 2007). Although sustained expression of many of the initially induced proteins is deleterious to cell survival (Borden et al, 2007), we have discovered that the increased expression of the proteins encoded by a subset of these genes is sustained for many days (Cheon and Stark, 2009). They provide a selective advantage and cells can tolerate them. In response to IFNβ, signal transducers and activators of transcription (STATs) are phosphorylated on C-terminal tyrosine residues (Y701 for STAT1 and Y690 for STAT2), followed by their combination with IFN response factor 9 (IRF9) to form the tripartite transcription factor IFN-stimulated gene factor 3 (ISGF3), which drives the expression of >100 IFNβ-stimulated genes (ISGs, Borden et al, 2007). Previously, we showed that STAT1 lacking phosphorylation of Y701 (U-STAT1) sustains the response to IFNβ for several days (Cheon and Stark, 2009) and, even in the absence of IFNβ, high levels of U-STAT1 induced the expression of a subset of ISGs, including IFI27, BST2, OAS1, OAS2, OAS3, and STAT1 itself. Many of the encoded proteins have anti-viral activities (Dong et al, 2004; Borden et al, 2007; Neil et al, 2008; Randall and Goodbourn, 2008; Sadler and Williams, 2008; Tan et al, 2008; Brass et al, 2009; Itsui et al, 2009; Schmeisser et al, 2010; Tang et al, 2010; Miyashita et al, 2011; Schoggins et al, 2011; Oudshoorn et al, 2012). We have now elucidated the mechanism and the additional biological consequences of U-STAT1-induced gene expression, finding that IFNβ also induces the expression of un-phosphorylated STAT2 (U-STAT2) and IRF9, which combine with U-STAT1 to form un-phosphorylated ISGF3 (U-ISGF3), a novel transcription factor in which these proteins form a ternary complex without tyrosine phosphorylation. U-ISGF3 in turn maintains the expression of a subset of the initially induced ISGs whose protein products lead to extended resistance to virus infection and DNA damage. Interestingly, expression of the same subset of ISGs is uniquely increased in radiation-resistant cancer cells (Khodarev et al, 2004; Cheon et al, 2011), in cancer cells resistant to a variety of DNA damaging treatments (Gongora et al, 2008; Luszczeck et al, 2010), and in cancer cells from glioblastoma and breast cancer patients who responded poorly to chemo- or radiation therapy (Weichselbaum et al, 2008; Duarte et al, 2012). We show that prolonged exposure of cells to a low level of IFNβ induces a steady state in which only the U-ISGF3-dependent genes are expressed, suggesting that secretion of IFNβ by cancer cells may account for their similar phenotype. Results The expression of anti-viral genes is sustained for several days after IFNβ treatment, along with increased levels of STAT1, STAT2, and IRF9 proteins Over a hundred genes are induced by IFNβ quickly, in response to the tyrosine phosphorylation of STATs 1 and 2 and subsequent formation of ISGF3, but the expression of many genes is downregulated as the level of ISGF3 decreases. However, the expression of many anti-viral genes that are induced initially by IFNβ is sustained and even increased by increased expression of U-STAT1, the levels of which remain high for many days (Cheon and Stark, 2009). As shown in Figure 1A, the expression of four representative anti-viral genes (IFI27, BST2, OAS2, and MX1) is induced greatly after 24 h and sustained at high levels for at least 72 h after a single treatment with IFNβ (3 IU/ml) of two different human non-cancer cell lines, hTERT-HME1 mammary epithelial cells and BJ fibroblasts. To test whether the continued expression of these genes might be due to the presence of a low residual level of phosphorylated STATs 1 and 2, we examined STAT expression and phosphorylation in response to a much higher concentration of IFNβ (50 IU/ml). Even at this high concentration, phosphorylated STATs 1 and 2 were seen only transiently, and we detected little phosphorylated STAT1 or phosphorylated STAT2 after 48 h (Figure 1B). However, the expression of STAT1, STAT2, and IRF9 was increased greatly after 24 h and was sustained for at least 72 h, with kinetics similar to the kinetics of anti-viral gene expression shown in Figure 1A. In contrast to the sustained expression of the four anti-viral genes noted above, the expression of IRF1, an ISG whose expression is driven by ISGF3 and not by U-STAT1, increased transiently and decreased in parallel with the levels of phosphorylated STATs 1 and 2 (Figure 1B), showing that, even if phosphorylated ISGF3 was still present at levels below our ability to detect it, there was not enough to drive the expression of this target gene. The increased levels of STAT1, STAT2, IRF9, and several U-STAT1-induced genes (IFI27, OAS2, and MX1) lasted for at least 12 days after a single treatment with IFNβ (50 IU/ml), while the expression of ISGs (MYD88, IRF1, and IFI16) that are not induced by U-STAT1 (Cheon and Stark, 2009) returned to basal levels after 3 days or sooner (Supplementary Figure S1). In contrast to their prolonged expressions in BJ and hTERT-HME1 cells, the expression of IFI27, OAS2, and MX1 was transient in AG14412 umbilical cord fibroblasts, where IFNβ induced the tyrosine phosphorylation of STATs 1 and 2, did not increase the expression of STAT1 and IRF9 proteins, and increased STAT2 protein expression minimally (Figure 1C). We observed a similarly transient expression of IFI27, OAS2, and MX1 in STAT1-null fibroblasts reconstituted with wild-type STAT1 (Figure 1D). Since STAT1 gene expression in these cells is regulated by the CMV promoter in the vector and not by the endogeneous STAT1 promoter, STAT1 protein expression is not increased in response to IFNβ. These results show that increased levels of STAT1, STAT2, and IRF9 are likely to be important for the prolonged expression of anti-viral genes, while tyrosine phosphorylation of STATs 1 and 2 is important for initial gene expression. Figure 1.The expression of anti-viral genes is sustained for several days after IFN stimulation, along with increased levels of the STAT1, STAT2, and IRF9 proteins. (A) The expression of the anti-viral genes IFI27, BST2, OAS2, and MX1 was analysed by real-time PCR after stimulation of hTERT-HME1 or BJ cells with IFNβ (3 IU/ml). The levels of gene expression were calculated semi-quantitatively by using the ΔΔCt method (see Materials and methods). The data are represented as means of triplicate PCR analyses±standard deviations (s.d.). (B) hTERT-HME1 or BJ cells were treated with IFNβ (50 IU/ml) and the levels of total IRF9 and STATs 1 and 2, or tyrosine-phosphorylated STATs 1 and 2 (PY-701-STAT1 or PY-690-STAT2) were analysed by the western method. (C, D) AG14412 umbilical fibroblasts (C) and STAT1-null fibroblasts transfected with the lentiviral vector containing wild-type STAT1 (D) were used. Left, Cells were treated with IFNβ (50 IU/ml) and the levels of total IRF9, STAT1, and STAT2, or tyrosine-phosphorylated STATs (PY-701-STAT1 or PY-690-STAT2) were analysed by the western method. Right, the expression of the IFI27, OAS2, and MX1 genes was analysed by real-time PCR after stimulation with IFNβ (3 IU/ml). The levels of gene expression were calculated semi-quantitatively by using the ΔΔCt method. The data are represented as means of triplicate PCR analyses±s.d.Source data for this figure is available on the online supplementary information page. Source Data for Figure 1B [embj2013203-sup-0001-SourceData-S1.pdf] Download figure Download PowerPoint High levels of U-STAT1, U-STAT2, and IRF9 are necessary and sufficient for the induction of some anti-viral genes Our previous microarray analysis showed that increased expression of either wild-type or Y701F mutant STAT1 led to increased expression of 30 genes without IFN stimulation in BJ cells, which already express substantial amounts of STAT2 and IRF9, but not in hTERT-HME1 cells, which express little STAT2 and IRF9 (Cheon and Stark, 2009), indicating that U-STAT1 may not induce gene expression without a sufficient amount of STAT2 and IRF9. To investigate the role of STAT2 and IRF9, we stably increased the expression of STAT2 and IRF9, together with STAT1, in hTERT-HME1 cells. Western analyses confirmed that, in the absence of treatment with IFNβ, the highly expressed STATs were not phosphorylated on their tyrosine residues (Figure 2A). The expression of IFI27, OAS2, and MX1 was very low in control hTERT-HME1 cells (Figure 2B, column 1) and their expression was not increased when the levels of U-STAT1, U-STAT2, or IRF9 were increased one at a time without treatment with IFNβ (Figure 2B, columns 2–4). Furthermore, the combined high expression of U-STAT1 plus U-STAT2 or U-STAT1 plus IRF9 still did not increase the expression of these genes in hTERT-HME1 cells (Figure 2B, columns 5 and 6). However, the expression of the target genes increased strongly when the levels of U-STAT2 and IRF9 were increased together (column 7) and increased even more when U-STAT1 expression, already significant, was increased further (column 8). Y701F-STAT1 is a dominant negative protein because it binds to IFN receptor SH2 domains but cannot be phosphorylated on Y701, thus blocking access of the wild-type protein to the receptor. However, high expression of Y701F-STAT1, U-STAT2, and IRF9 together in hTERT-HME1 cells significantly increased the expression of the target genes, IFI27, OAS1, OAS2, MX1, IFIT1, and IFIT3 (Figure 2C), indicating that STAT1 tyrosine phosphorylation was not involved. In contrast, the expression of additional ISGs (MYD88, ADAR, IFI16, and IRF1), which are induced transiently by ISGF3 but not sustained at late times after IFN treatment, was not increased by higher levels of Y701F-STAT1, U-STAT2, and IRF9 (Figure 2D). To exclude the possibility that phosphorylation of endogeneous wild-type STAT1 is involved in the expression of the genes, we also used STAT1-null fibroblasts reconstituted with Y701F-STAT1, where STAT1 cannot be phosphorylated on residue Y701. While IFNβ does not increase the expression of IFI27, OAS2, and MX1 in these cells (Supplementary Figure S2), high expression of STAT2 and IRF9 together with Y701F-STAT1 readily increased those three ISGs (Figure 2E), but not the transiently induced ISGs MYD88, IFI16, and IRF1 (Figure 2F). Figure 2.High levels of STAT1, STAT2, and IRF9 proteins are necessary and sufficient for the induction of some anti-viral genes without IFN-induced phosphorylation. (A) The phosphorylation status of STAT1 (Y701) and STAT2 (Y690) was examined in hTERT-HME1 cells transfected with empty pLV vector or pLV-STAT1, -STAT2, and -IRF9 together (S1S2I9). As a positive control, S1S2I9 cells were treated with 500 IU/ml of IFNβ for 1 h. (B) The expression of the indicated genes was measured by real-time PCR in hTERT-HME1 cells stably transfected with various combinations of STAT1, STAT2, and IRF9 in the lentiviral vector pLV. The levels of gene expression were calculated semi-quantitatively by using the ΔΔCt method. The data are represented as means of triplicate PCR analyses±s.d. ** represents P<0.01 and * represents P<0.05 by two-tailed t-test, compared to vector-transfected controls (column 1). (C, D) The expression of the indicated genes was measured by real-time PCR in hTERT-HME1 cells stably transfected with empty pLV vector (Vec), or pLV-Y701F-STAT1, pLV-STAT2, and pLV-IRF9 together (YF-S1S2I9). The levels of gene expression were calculated semi-quantitatively by using the ΔΔCt method. The data are represented as means of triplicate PCR analyses±s.d. (E, F) STAT1-null fibroblasts were stably transfected with empty pLV vector (Vec), or pLV-Y701F-STAT1, pLV-STAT2, and pLV-IRF9 together (YF-S1S2I9). The expression of the indicated genes was measured by real-time PCR. The levels of gene expression were calculated semi-quantitatively by using the ΔΔCt method. The data are represented as means of triplicate PCR analyses±s.d.Source data for this figure is available on the online supplementary information page. Source Data for Figure 2A [embj2013203-sup-0002-SourceData-S2.pdf] Download figure Download PowerPoint High levels of U-STAT1, U-STAT2, and IRF9 protect cells from virus infection Vesicular stomatitis virus (VSV, a negative ssRNA virus) was less infectious in hTERT-HME1 cells expressing high levels of wild-type STAT1 (WT) or Y701F-STAT1 (YF), together with U-STAT2 and IRF9, than in control cells (Vec, Figure 3A, left panel; Supplementary Figure S3A). The titres of infectious VSV were reduced by high levels of wild-type STAT1/STAT2/IRF9 (WT) or Y701F-STAT1/STAT2/IRF9 (YF) by 51-fold or 33-fold, respectively. In cells overexpressing wild-type STAT1/STAT2/IRF9 (WT), virus replication was inhibited more efficiently in the presence of IFNβ (Supplementary Figure S3B), because increased levels of ISGF3 formed by wild-type STAT1/STAT2/IRF9 sensitize cells to IFNs. However, anti-viral effects in cells overexpressing Y701F-STAT1/STAT2/IRF9 (YF) were not influenced by IFNβ in the media (Supplementary Figure S3B), showing that the Y701F-STAT1/STAT2/IRF9-induced anti-viral effects resulted solely from the high levels of U-STAT1, U-STAT2, and IRF9 proteins rather than the IFN-induced phosphorylation of STATs 1 and 2. Figure 3.High levels of STAT1, STAT2, and IRF9 proteins protect cells from various RNA viruses in an IFN-independent manner. hTERT-HME1 cells transfected with empty vector (Vec), wild-type STAT1/STAT2/IRF9 (WT) or Y701F-STAT1/STAT2/IRF9 (YF) were used. (A) Cells were infected with 0.1 MOI of VSV or 1 MOI of EMCV, and the infected cells and cell-culture media were collected after 10 h (VSV) or 6 h (EMCV). The infectious viral titres in the collected samples were analysed by plaque assays on Vero cells. The data are represented as means of triplicate infections±s.d. An asterisk (*) represents P<0.01, by two-tailed t-test, compared to cells transfected with empty vector (Vec). (B) Cells were infected with VSV or EMCV (10–10−5 MOI). After 48 h, the surviving cells were fixed with methanol (VSV) or 4% paraformaldehyde (EMCV) and stained with crystal violet. (C) Cells were infected with recombinant viral constructs (VSV, PIV3, or YFV) expressing GFP. After 8 h (VSV) or 48 h (PIV3 or YFV), GFP fluorescence was monitored by FACS analyses. The data are represented as mean GFP intensities (MFI)±s.d. of triplicate infections. An asterisk (*) represents P<0.01 and a cross (+) represents P<0.05, by two-tailed t-test, compared to Vec cells. Download figure Download PowerPoint The replication of encephalomyocarditis virus (EMCV, a positive ssRNA virus) was also inhibited, five-fold by high levels of wild-type STAT1/STAT2/IRF9 or four-fold by Y701F-STAT1/STAT2/IRF9, when assayed 6 h after infection (Figure 3A, right panel). We also examined the effect of high levels of wild-type- or Y701F-STAT1/STAT2/IRF9 after several cycles of virus replication. Infected cells are eventually lysed by VSV and EMCV, and we measured the surviving cells after 48 h. hTERT-HME1 cells were infected with 10–10−5 multiplicity of infection (MOI) of VSV or EMCV (Figure 3B). Control cells (Vec) were completely killed at 10−4 MOI of VSV or 10−2 MOI of EMCV, but wild-type STAT1/STAT2/IRF9-transfected cells were 100 times more resistant to VSV and >1000 times more resistant to EMCV in this assay. Y701F-STAT1/STAT2/IRF9-transfected cells showed similar levels of resistance, confirming that the anti-viral effects were induced by high levels of U-STAT1, U-STAT2, and IRF9 proteins independently of virally induced IFN stimulation. We tested additional RNA viruses by infecting the above cells with GFP-tagged VSV, parainfluenza virus type 3 (PIV3), or yellow fever virus (YFV). Consistently, we confirmed using an alternative method, FACS analysis, that high levels of wild-type- or Y701F-STAT1/STAT2/IRF9 reduced VSV replication after 8 h (Figure 3C, P<0.01). The replication of PIV3 (a negative ssRNA virus) was also inhibited significantly by high levels of U-STAT1/U-STAT2/IRF9, by >10-fold 48 h after infection (P<0.01). High levels of U-STAT1/U-STAT2/IRF9 also inhibited significantly the replication of YFV (a positive ssRNA virus), by 30% 48 h after infection with 100 MOI of virus (P<0.01). In summary, our results show that increased levels of U-STAT1, U-STAT2, and IRF9 are able to inhibit infection by several different RNA viruses without IFN treatment. U-STAT1, U-STAT2, and IRF9 form U-ISGF3, which binds to IFN stimulated response elements in target gene promoters We examined whether U-STAT1, U-STAT2, and IRF9 could form a complex without phosphorylation, using co-immunoprecipitation (Co-IP) from hTERT-HME1 cells expressing high levels of these proteins (Figure 4A). Since the interaction between STAT1 and IRF9 in classical ISGF3 was reported to be unstable (Martinez-Moczygemba et al, 1997), we used the cleavable cross-linking reagent dimethyl-3,3′-dithiobis-propinimidate (DTBP). DTBP did stabilize the interaction between U-STAT1 and IRF9 (Figure 4A, lane 2), but we were still able to observe this interaction without cross-linking in the nuclear fractions of hTERT-HME1 cells expressing high levels of U-STAT1, U-STAT2, and IRF9 (lanes 3 and 7). The interactions between STAT1 and STAT2 (lanes 2, 3, and 6) and between STAT2 and IRF9 (lanes 6 and 7) were clearly observed in the nuclear fractions. We performed chromatin-immunoprecipitation (ChIP) assays using hTERT-HME1 cells expressing high levels of U-STAT1, U-STAT2, and IRF9 in the absence of IFN treatment. Sheared chromatin (<1 Kb) was precipitated with antibodies against STAT1, STAT2, or IRF9, and the DNAs were amplified by real-time PCR, using primers spanning the most highly conserved IFN stimulated response elements (ISREs) (striped triangles in Supplementary Figure S4A) in each promoter, identified by using the transcription factor search program TFSEARCH (http://www.cbrc.jp/research/db/TFSEARCH.html). The IRF9 antibody enriched ISRE-containing promoter regions of the IFI27, OAS2, and MX1 genes, by about 3.5-fold, compared to non-immune IgG (Figure 4B, upper panel, P<0.05). Analysis with an STAT1 antibody also showed enhanced binding to the ISREs of the IFI27, OAS2, and MX1 genes, by about three-fold (Figure 4B, upper panel, P<0.05). STAT2 also bound to the same ISREs, with an enrichment of four- to five-fold (Figure 4B, lower panel, P<0.05). Promoter occupancy by U-ISGF3 was not observed in control cells transfected with empty vector (Supplementary Figure S4B). We conclude that the amount of the ternary U-ISGF3 complex is increased in response to high levels of U-STAT1, U-STAT2, and IRF9 without IFN-induced phosphorylation and is present on ISREs in the promoters of U-ISGF3 target genes. However, U-ISGF3 did not bind to ISREs in the promoters of ISGs (MYD88, IRF1, and ADAR) that are not induced by U-ISGF3 (Figure 4C). Figure 4.U-STAT1, U-STAT2, and IRF9 form U-ISGF3, which binds to ISREs on the target gene promoters. hTERT-HME1 cells expressing high levels of U-STAT1, U-STAT2, and IRF9 without IFN stimulation were analysed by co-immunoprecipitation (Co-IP) and chromatin-immunoprecipitation (ChIP) assays. (A) Nuclear proteins were used for Co-IP with normal rabbit IgG or rabbit polyclonal antibodies against STAT1, STAT2, or IRF9. To stabilize protein–protein interactions, the nuclear fraction was treated with the cleavable cross-linker, dimethyl-3,3′-dithiobis-propinimidate (DTBP, lanes 1, 2, and 4). Mouse monoclonal antibodies against STAT1, STAT2, or IRF9 were used for the western method. The purity of the fractions was assessed by determining the levels of GAPDH (a cytoplasmic protein) and HDAC1 (a nuclear protein) in the input lysates by the western method. (B, C) Total protein lysates were cross-linked with 1% formaldehyde and the cell lysates were cross-linked with DTBP. Chromatin was sheared into <1 kb lengths by sonication. Rabbit polyclonal antibodies against STAT1, STAT2, or IRF9, or comparable amounts of normal rabbit IgG, were used for immunoprecipitations. Real-time PCR was performed to amplify the precipitated DNAs with primer pairs spanning ISREs in the promoters of IFI27, OAS2, MX1, MYD88, IRF1, and ADAR genes. MYD88-1 and -2 mean 2 different ISREs in the promoter of MYD88 gene. The amount of amplified DNA was calculated by using the standard curve method. The values (% input) are the percentages of DNA amount in immunoprecipitated samples compared to 2% input DNA. The data are represented as means of triplicate PCR analyses±s.d. ** represents P<0.01 and * represents P<0.05, by two-tailed t-test, compared to the IgG control. ND, not different statistically (P>0.05, by two-tailed t-test).Source data for this figure is available on the online supplementary information page. Source Data for Figure 4A [embj2013203-sup-0003-SourceData-S3.pdf] Download figure Download PowerPoint U-ISGF3-induced genes have distinct ISREs We classified genes into U-ISGF3-induced genes and classical ISGF3-induced genes using microarray data. We identified 150 genes that are upregulated by IFNβ after 6 h, indicating that these genes are likely to have ISREs in their promoters. Indeed, analysis of the structures of transcription factor binding sites revealed a significant enrichment of canonical ISREs within the 150 IFNβ-induced genes compared with putative sites identified in all genes (P<0.001). Among these IFNβ-induced genes, only 29 (20%) were induced by the upregulation of Y701F-STAT1, and we assumed that these are induced by U-ISGF3. Morrow et al (2011) reported that IFNγ induces the expression of anti-viral genes through another form of ISGF3, consisting of PY-STAT1, U-STAT2, and IRF9. Among the remaining 121 IFNβ-induced genes (150 IFNβ-induced genes minus 29 U-ISGF3-induced genes), 73 are induced by IFNγ (6 h). We excluded those genes, assuming that they have ISREs different from those of the genes induced only by IFNβ (48 genes, listed in Supplementary Table S1). To identify differences in ISREs, a guided analysis was performed on the genes induced only by classical ISGF3 and the genes induced by U-ISGF3 (Figure 5A). Genes induced by ISGF3 but not by U-ISGF3 contain ISREs similar to the canonical Transfac-annotated ISRE site. The U-ISGF3-induced genes also have canonical ISRE sites, which have additional conserved sequences in the 5′ and 3′ flanking regions (Figure 5B). The conserved ISRE sequences of the two groups are statistically different, applying symmetrized, position-averaged Kullback-Leibler distance (P<0.05). These results suggest that all IFNβ-induced genes are transcribed when classical ISGF3 binds to canonical ISREs at early times, but that only a subset of these genes, which contain ISREs with variant flanking sequences, can be further induced by U-ISGF3 at late times. Figure 5.U-ISGF3-induced genes
