Virus-induced heterodimer formation between IRF-5 and IRF-7 modulates assembly of the IFNA enhanceosome in vivo and transcriptional activity of IFNA genes.

Transcription factors of the interferon regulatory factor (IRF) family have been identified as critical mediators of early inflammatory gene transcription in infected cells. We have shown previously that IRF-5, like IRF-3 and IRF-7, is a direct transducer of virus-mediated signaling and plays a role in the expression of multiple cytokines/chemokines. The present study is focused on the molecular mechanisms underlying the formation and function of IRF-5/IRF-7 heterodimers in infected cells. The interaction between IRF-5 and IRF-7 is not cooperative and results in a repression rather than enhancement of IFNA gene transcription. The formation of the IRF-5/IRF-7 heterodimer is dependent on IRF-7 phosphorylation, as shown by the glutathione S-transferase pull-down and immunoprecipitation assays. Mapping of the interaction domain revealed that formation of IRF-5/IRF-7 heterodimers occurs through the amino terminus resulting in a masking of the DNA binding domain, the consequent alteration of the composition of the enhanceosome complex binding to IFNA promoters in vivo, and modulation of the expression profile of IFNA subtypes. Thus, these results indicate that IRF-5 can act as both an activator and a repressor of IFN gene induction dependent on the IRF-interacting partner, and IRF-5 may be a part of the regulatory network that ensures timely expression of the immediate early inflammatory genes.


Transcription factors of the interferon regulatory factor (IRF) family have been identified as critical mediators of early inflammatory gene transcription in infected cells. We have shown previously that IRF-5, like IRF-3 and IRF-7, is a direct transducer of virus-mediated signaling and plays a role in the expression of multiple cytokines/chemokines. The present study is focused on the molecular mechanisms underlying the formation and function of IRF-5/IRF-7 heterodimers in infected cells. The interaction between IRF-5 and IRF-7 is not cooperative and results in a repression rather than enhancement of IFNA gene transcription. The formation of the IRF-5/IRF-7 heterodimer is dependent on IRF-7 phosphorylation, as shown by the glutathione Stransferase pull-down and immunoprecipitation assays.
Mapping of the interaction domain revealed that formation of IRF-5/IRF-7 heterodimers occurs through the amino terminus resulting in a masking of the DNA binding domain, the consequent alteration of the composition of the enhanceosome complex binding to IFNA promoters in vivo, and modulation of the expression profile of IFNA subtypes. Thus, these results indicate that IRF-5 can act as both an activator and a repressor of IFN gene induction dependent on the IRF-interacting partner, and IRF-5 may be a part of the regulatory network that ensures timely expression of the immediate early inflammatory genes.
Type I interferons (IFN) 1 play an essential role in the innate immune response against virus infection (1). In uninfected cells, the expression of IFN genes is tightly regulated. Virus infection activates transcription of type I IFN genes, and the cis-acting virus-responsive elements (VRE), located within the 110 nucleotides 5Ј of the transcription initiation site, are suf-ficient for virus-mediated activation (2,3). The VRE of IFNA genes contain purine-rich GAAANN motifs that constitute the specific binding sites (IRF-E and PRDI/III) for the proteins of the interferon regulatory factor (IRF) family, whereas the VRE in the promoter of the IFNB gene contain not only PRDI/ PRDIII elements but also an NF-B (PRDII)-binding site (4).
Nine cellular IRF and three viral homologues (vIRF) have been identified (4 -9). All of the cellular IRF share a region of homology in the amino terminus encompassing a highly conserved DNA binding domain (DBD) that is characterized by five tryptophan repeats (4). Three of the repeats contact DNA recognizing the GAAA or AANNGAAA sequences (10 -12). KSHV-encoded IRF that contain an imperfect DNA binding domain are not able to bind DNA with the same specificity as cellular IRF.
Several of the cellular IRF were implicated in the regulation of type I IFN gene expression in virus-infected cells. IRF-1 was first identified as an activator of the IFNB gene, whereas IRF-2 antagonized the IRF-1-mediated activation and acted as a suppressor (13)(14)(15). Whereas the infected embryonic fibroblasts from IRF-1 Ϫ/Ϫ mice expressed normal levels of IFNA and -B (16,17), it was observed that IRF-1 associates with the VRE of both human IFNA and IFNB promoters in vivo, thus suggesting that IRF-1 may contribute to the transcriptional regulation of the human IFNA genes (18,19). Three members of the IRF family (IRF-3, IRF-5, and IRF-7) were shown to be direct transducers of virus-mediated signaling and to play a crucial role in the expression of type I IFN genes (19 -26). Whereas IRF-3 is constitutively expressed in all cell types (27), constitutive expression of IRF-7 was observed only in lymphoid cells but could be induced by type I IFN in other cell types (20,28). Expression of IRF-5 was detected primarily in B cells and dendritic cells and was further enhanced by type I IFN (5,9,21). In the murine system, induction of type I IFN genes in virus-infected primary embryonic fibroblasts was proposed to proceed by two sequential phases. During the initial phase, which does not require protein synthesis, transcription of IFNB and IFNA4 genes was activated. The second phase, during which the rest of the IFNA subtypes were induced, depended on the IFN-mediated induction of IRF-7 expression (23,29). However, in recently generated mice with a homozygous deletion of the IRF-7 gene, production of IFN␣ in the isolated mouse embryo fibroblasts was completely abolished, whereas the expression of IFN␤ was significantly reduced (30). These results are consistent with the observations in human cells, where expression of IRF-3 conferred only virus-mediated induction of the IFNB gene (31), and IRF-7 was required for expression of IFNA genes. In both the infected mouse and human cells, expression of IFNA could be rescued upon reconstitution of IRF-7 expression (25). Surprising, however, was the observa-tion that in the absence of IRF-7, induction of IFNA genes could be rescued by overexpression of IRF-5 (5,21). However, in contrast to IRF-7, activation of IRF-5 was virus-specific and resulted in the expression of different IFNA subtypes than were identified in infected cells expressing IRF-7 (21).
In uninfected cells, these three IRF proteins are generally present in the cytoplasm. Viral infection induces phosphorylation of IRF-3, IRF-5, and IRF-7 by yet unidentified kinase(s), and these IRF are then retained in the nucleus where they interact with the histone transacetylases p300/CBP or PCAF (32)(33)(34)(35). 2 Of note, IRF-5 contains two nuclear localization signals and translocates to the nucleus in uninfected cells, although less efficiently than in virus-infected cells (5). Interestingly, although a large number of RNA and DNA viruses, as well as double-stranded RNA, induce phosphorylation of IRF-3 and IRF-7, phosphorylation of IRF-5 seems to be virus-specific, and this factor is not induced or phosphorylated by doublestranded RNA (5). Structure-function analysis of IRF-3, IRF-5, and IRF-7 proteins revealed that all of these IRF contain an auto-inhibitory domain, which in transient transfection assays suppresses the transcriptional activity of these factors (5,22,36,37). Furthermore, results from a study with the IRF-7 dominant-negative mutant suggested that IRF-3 and IRF-7 form homo-and heterodimers and that these interactions are critical for the stimulation of the transcriptional activity of endogenous IFNA genes (22,38).
Although the promoters of all IFNA genes are highly homologous, differential expression of these genes, as a function of cell type and inducing agent, has been observed (39,40). IFNA1 was identified as the major subtype expressed in human fibroblasts expressing IRF-3 and ectopic IRF-7 (25) but not IRF-5. In contrast, in cells that express ectopic IRF-5, but not IRF-7, NDV predominantly induced the IFNA8 subtype (21). Analysis of the transcriptional complex/enhanceosome assembled on the promoter of the IFNA1 and IFNB genes revealed that both the IFNA and IFNB enhanceosomes contain IRF-3 and IRF-7, as well as IRF-1 (19,41). In cells that express IRF-5, but not IRF-7, both IRF-3 and IRF-5 were found to be bound to IFNA VRE and the endogenous IFNA promoters (5,21). However, we have observed recently 3 that in monocytes, and particularly in precursor dendritic cells (pDC2) that are high producers of IFN␣, both IRF-7 and IRF-5 are expressed.
The goal of this study was therefore to gain more insight into the molecular mechanisms underlying the expression of human IFNA genes in cells expressing both IRF-7 and IRF-5, and to determine whether the expression of these two IRF in the cell is associated with high levels of IFNA production, as seen in pDC2 cells (42). To this effect, we have generated 2fTGH cell lines expressing ectopic IRF-7 or IRF-5, or both of these proteins. Surprisingly, the simultaneous expression of both IRF-5 and IRF-7 did not result in an enhanced synthesis of biologically active IFN in infected cells; to the contrary, expression of both factors resulted in a distinct repression of IFN. We also reveal that whereas IRF-5 can interact with IRF-7, there is a mutual binding exclusion between these two family members to IFNA promoters from the lysates of infected cells. These results indicate the lack of cooperative interaction between IRF-5 and IRF-7 and enhancement of IFNA production but show that the relative levels of IRF-5 and IRF-7 expressed in infected cells determine the distinct profile of IFNA genes induced.

EXPERIMENTAL PROCEDURES
Cells and Virus-2fTGH/IRF-5 and 2fTGH/IRF-7 overexpressing cells were generated as described previously (21,26). MDBK bovine kidney cells and HeLa cells were obtained from ATCC, and 2fTGH cells, obtained from G. Stark (Cleveland Clinic Foundation, Cleveland, OH), were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum. Sendai virus was purchased from Specific Pathogen-free Avian Supply (Preston, CT), and NDV was purchased from ATCC (VR-699). Infections were conducted with 640 hemagglutinin units/100-mm plate (80% confluency) for a given period.
GST Pull-down Assay-GST fusion constructs were transformed into BL21 bacteria and induced with 0.5 mM isopropyl-1-thio-␤-D-galactopyranoside for 3 h. The GST fusion proteins were purified from bacterial lysates by affinity chromatography on a glutathione-agarose column (Sigma). Coomassie Blue staining of the GST fusion proteins was used to quantify the amount of protein bound to beads. Equal amounts of GST fusion proteins and beads were then applied in each pull-down assay. The 35 S-labeled proteins were synthesized in vitro using the coupled TNT T7 transcription/translation system (Promega) according to the manufacturer's instructions. Each reaction mixture contained 1 g of nonlinearized expression plasmid and was incubated (90 min, 30°C) in the presence of 4 l of Translabel (DuPont) amino acid mixture. GST fusion proteins (0.5 g) bound to glutathione-agarose beads were incubated with 10-l reaction mixture aliquots of 35 S-labeled proteins in 250 l of binding buffer (10 mM Tris-Cl, pH 7.6, 75 mM KCl, 0.1 mM EDTA, 2.5 mM MgCl 2 , 1 mM dithiothreitol, 0.1% Nonidet P-40, 500 g of bovine serum albumin, and 8% glycerol) at 4°C for 2 h. After three washes (10 min at room temperature) with binding buffer, the proteins bound to the beads were solubilized in sample buffer and resolved by 8% SDS-PAGE. The gel was dried and exposed to film. In addition, binding of IRF from whole cell lysates of uninfected or virusinfected 2fTGH/IRF-5 cells was performed using lysis buffer (20 mM Tris, pH 8.0, 1 mM EDTA, 200 mM NaCl, 0.5% Nonidet P-40, and 0.2 mM protease inhibitor mixture). GST fusion protein immobilized on beads and whole cell extract were incubated in 300 l of binding buffer at 4°C for 2 h. The beads were then washed 3 times with binding buffer, and the bound proteins were eluted and resolved by 10% SDS-PAGE, transferred to Hybond transfer membranes (Amersham Biosciences), and incubated with anti-IRF-1 antibodies or anti-IRF-7 antibodies. Immunocomplexes were detected by using the ECL system.
Immunoprecipitation-2fTGH/IRF-5 cells were co-transfected with the IRF-7 expression plasmid to examine interactions between IRF-5/ IRF-7. Cells were uninfected or infected with Sendai virus or NDV for 6 h and then lysed in IP lysis buffer (20 mM Hepes, pH 7.9, 50 mM NaCl, 10 mM EDTA, 2 mM EGTA, 0.1% Nonidet P-40, 10% glycerol, and 0.2 mM protease inhibitor mixture). Extracts (250 g) were incubated with 1 g of anti-IRF-1 antibodies or 1 g of anti-IRF-7 antibodies crosslinked to 30 l of protein A-Sepharose beads, respectively, for 1 h at 4°C. Precipitates were washed 4 times with IP lysis buffer and eluted by boiling the beads for 3 min in 1ϫ SDS loading buffer. Eluted proteins were separated by 10% SDS-PAGE and transferred, and IRF-5 was detected by anti-FLAG antibodies.
Transfections and SAP Assay-In the transient transfection assay, 2 ϫ 10 6 2fTGH or HeLa cells were transfected with a constant amount of DNA (5 g/60-mm plate) by using Superfect transfection reagent (Qiagen). For the SAP assays, equal amounts (2.5 g) of reporter plasmid and IRF-expressing plasmid were co-transfected with the ␤-galactosidase expression plasmid (50 -100 ng). The transfected cells were divided 14 h later and infected with NDV for 16 h. The alkaline phosphatase (SAP) assays were performed as described previously (44,46). Each experiment was repeated 3 times. The ␤-galactosidase expression levels were used to normalize the difference in transfection efficiency.
Oligonucleotide Pull-down Assay and Immunoblot-Doublestranded oligomers corresponding to the IFNA1 and IFNA14 VRE region (Ϫ110 to Ϫ53 bp) were synthesized and biotin-labeled at the 5Ј end of the antisense strand. Two micrograms of the double-stranded oligomers were incubated with streptavidin magnetic beads (200 l) (Dynal Inc.) for 1 h in TEN buffer (20 mM Tris, pH 8.0, 1 mM EDTA, 0.1 M NaCl), and the unbound DNA was removed by extensive washing with same buffer. IRF from whole cell extracts were bound to DNA on magnetic beads as described previously (21). Beads were then washed, and the bound proteins were analyzed on SDS-PAGE. IRF were identified by immunoblot with the respective antibodies as described (5,21). Blots were stripped (200 ml of 10% SDS in 1000 ml, 62.5 mM Tris, pH 6.7, 0.7% 2-mercaptoethanol) by incubating for 10 min at 65°C and then rotating for 10 min at room temperature. Blots were then washed 3-5 times with TBST and re-probed with the respective antibodies.
Reverse Transcription-PCR Analysis and Antiviral Assay-One microgram of total RNA isolated by the cesium chloride method was reverse-transcribed to cDNA with oligo(dT) primers in a 30-l reaction. From this mixture of cDNAs, IFNA was amplified by PCR using primers that detect all IFNA genes, and ␤-actin was amplified as an internal control (25). Individual IFNA subtypes were identified by cloning and subsequent sequencing of the PCR-amplified fragments as described recently (25). A one-way analysis of variance (faculty.vassar.edu/ ϳlowry/VassarStats.html) was used to analyze the significance of observed differences in IFNA subtype expression. Levels of biologically active IFN␣ in the medium were determined by the antiviral assay (47).
Chromatin Immunoprecipitation Assay-The detailed assay procedure was described previously (5,21). Briefly, 2fTGH/IRF-5 or 2fTGH/ IRF-7 cells (6 ϫ 10 7 ) were infected with Sendai virus or NDV for 6 h or left uninfected. The proteins bound to DNA were cross-linked by addition of formaldehyde in aqueous buffer (0.1 M NaCl, 1 mM EDTA, 50 mM Hepes, pH 8.0) to a final concentration of 1% for 30 min at 37°C. The reaction was stopped by addition of 0.125 M glycine. The cell pellets were washed with 5 ml of wash buffer (5 mM PIPES, pH 8.0, 85 mM KCl, 0.5% Nonidet P-40, 0.2 mM PMSF), resuspended in sonication buffer (1% SDS, 10 mM EDTA, 50 mM Tris-HCl, pH 8.0) on ice, and lysed by sonication for 10 s. Samples were diluted 10-fold with dilution buffer (0.01% SDS, 1.1% Triton X-100, 1.2 mM EDTA, 16.7 mM Tris-HCl, pH 8.0, 167 mM NaCl) and pre-cleared with protein A-Sepharose. Following the pre-clearing, equal amounts of proteins (as determined by the Bio-Rad protein assay reagent) were immunoprecipitated with 1 g of anti-IRF-1, anti-IRF-3, anti-IRF-7, or anti-FLAG antibodies for 4 h at 4°C. Immunocomplexes were extensively washed and treated with RNase A (50 g/ml), 0.5% SDS, and proteinase K (500 g/ml). The cross-linked DNA-protein complexes were reverted by heating at 65°C for 6 h, and the DNA was recovered by phenol/chloroform extraction. DNA purified by precipitation with 2 M ammonium acetate/ethanol was used as a template for PCR amplification with universal primers corresponding to the regions of human endogenous IFNA genes that are conserved in all subtypes (5,25).

Interaction between IRF-5 and Other Members of the IRF
Family-Previous work (5,21) has shown that IRF-5 interacts with IRF-3 forming a functional heterodimer involved in the induction of IFNA and IFNB genes in infected cells. However, even in the cells where IRF-3 expression was nearly eliminated, IRF-5 was still capable of inducing low levels of biologically active IFN␣ and was able to bind to the IFNA VRE (21). To what extent IRF-5 interacts with other IRF family members and the functional consequences of these interactions on type I IFN gene expression have not yet been determined. Here we examined whether IRF-5 could associate with IRF-1 or IRF-7, both of which have been shown to be present in the enhanceosome assembled on the promoters of IFNA1 and IFNB genes (19,33). The in vitro interactions between IRF-5 and 35 Slabeled IRF-1, IRF-3, IRF-5, or IRF-7 was examined by the GST pull-down assay. As shown in Fig. 1A, all examined IRF bound efficiently to GST-IRF-5, whereas they did not bind to GST-agarose bead. These results were further confirmed by the GST-IRF-5 pull-down assay with whole cell lysates from uninfected or virus-infected 2fTGH/IRF-7Ϫexpressing cells ( Fig.  1B) (5). Immunoblot analysis of cell extracts has shown that both IRF-5 and IRF-7 were expressed in these cells, although at different levels. Because we have shown previously (5, 21) both physical and functional interactions between IRF-5/IRF-5 and IRF-5/IRF-3, we have focused on the interactions between IRF-5 and IRF-1 or IRF-7. Fig. 1B shows that endogenous IRF-1 binds to GST-IRF-5 both from lysates of uninfected cells and Sendai virus or NDVϪinfected 2fTGH/IRF-7 cells. However, no binding of IRF-1 to GST-agarose beads could be detected. In contrast to the IRF-1 binding profile, IRF-7 bound to GST-IRF-5 much more efficiently from cell lysates of Sendai virus and NDVϪinfected 2fTGH/IRF-7 cells than from the lysates of uninfected cells. The low level of IRF-7 detected as bound to the GST-agarose beads in this experiment was not reproducible.
Taken together, these data suggest that the binding capacity of IRF-1 to GST-IRF-5 is not significantly altered by viral infection. On the other hand, binding of IRF-7 to IRF-5 is enhanced in infected cells suggesting that virus-induced posttranslational modification of IRF-7 plays a role in its ability to interact with IRF-5. We therefore examined the association between IRF-5 and IRF-1 or IRF-7 in cells by immunoprecipitation. In these experiments, 2fTGH/IRF-5 cells were transfected with IRF-7, and 24 h post-transfection, cells were either left uninfected or infected with Sendai virus or NDV for 8 h. Cell lysates were then immunoprecipitated with anti-IRF-1 or shown. Cell extracts were applied to GST-agarose beads or GST-tagged IRF-5Ϫagarose beads, and specifically bound proteins were detected by Western blotting (WB) with anti-IRF-1 or anti-IRF-7 polyclonal antibodies (Ab) as described under "Experimental Procedures." C, heterodimerization of IRF-5 with IRF-1 or IRF-7 in infected and uninfected cells. 2fTGH/IRF-5Ϫexpressing cells were transfected with IRF-7 and then left either uninfected (cont), infected with Sendai virus (SeV), or NDV for 6 h. Whole cell extracts (250 g) were immunoprecipitated (IP) with either anti-IRF-1 or anti-IRF-7 polyclonal antibodies as described under "Experimental Procedures." The immunoprecipitated complexes were separated on SDS-10% PAGE gels, and the presence of IRF-5 in immunoprecipitates was detected by immunoblotting with anti-FLAG monoclonal antibodies. Relative levels of endogenous IRF-1, IRF-5, and transfected IRF-7 in cell lysates are shown (20% input).
anti-IRF-7 antibodies, and the presence of IRF-5 in precipitates was detected by immunoblotting with anti-FLAG antibodies. The relative level of IRF proteins expressed in transfected cells was estimated by immunoblot analysis. Data in Fig. 1C shows that association of IRF-1 with IRF-5 is not dependent on viral infection, whereas IRF-7 binds to IRF-5 only in infected cells. These results would indicate that in infected cells expressing both IRF-5 and IRF-7, heterodimer formation is likely.
Distinct Activation of IFNA1 by IRF-5 and Its Heterodimers-The results shown in Fig. 1 indicate that formation of an IRF-5/IRF-1 heterodimer is not specific to infected cells, whereas the IRF-5/IRF-7 heterodimer forms specifically after virus infection. In order to determine what is the functional outcome of the association between IRF-5 and IRF-1 or IRF-5 and IRF-7, we examined the ability of these IRF family members to activate the IFNA1 promoter using the reporter plasmid containing SAP under the regulation of IFNA1 VRE (Fig. 2). As shown in Fig. 2A, overexpression of IRF-1, IRF-5, or IRF-7 in HeLa cells led to transcriptional activation of the IFNA1 SAP reporter. By immunoblot analysis, all three IRF were expressed to a similar level (data not shown). Whereas IRF-5 activated the SAP reporter in both uninfected and NDVinfected cells, activation by IRF-7 occurred only in infected cells. In contrast to these two IRF, activation by IRF-1 in infected cells was low, as shown previously (15). Consequently, co-transfection of IRF-1 and IRF-5 with the IFNA1 SAP reporter plasmid to HeLa cells infected with NDV yielded similar activation levels to those observed with IRF-5 alone. Although the IFNA1 promoter was activated in uninfected HeLa cells co-transfected with both IRF-5 and IRF-7, upon infection with NDV, the activity of the IFNA1 SAP reporter was suppressed compared with levels induced by either of these factors alone. These results suggest that IRF-5/IRF-7 heterodimer formation in infected cells (as shown in Fig. 1C) does not lead to activation of the IFNA1 promoter.
We have shown previously (19) that IRF-1, IRF-3, and IRF-7 are part of the transcriptional complex-enhanceosome binding to the IFNA1 promoter in infected cells. We therefore investigated the functional differences between IRF-1/IRF-5 and IRF-1/IRF-7 heterodimers. For these experiments, increasing amounts of the IRF-1 expression plasmid were co-transfected together with IRF-5 or IRF-7 and the IFNA1 SAP reporter plasmid to HeLa cells. Levels of protein expressed were determined by immunoblot analysis and observed to be equivalent to the amount of transfected cDNA (data not shown). Results in Fig. 2B demonstrated that increasing levels of IRF-1 repressed the IRF-5-mediated transactivation of the IFNA1 promoter in both uninfected and NDV-infected cells. In comparison, transfection of increasing amounts of IRF-1 with IRF-7 led to a concentration-dependent increase in transcription of the IFNA1 promoter. These results indicate that IRF-1 may compete for binding of IRF-5 to the IFNA1 VRE, whereas increasing levels of IRF-1 enhanced the transcriptional activity of IRF-7 in infected cells suggesting a cooperative interaction between these two factors. Thus, in the context of the IFNA1 promoter, IRF-1 can act both as a repressor or enhancer depending on its interaction with distinct IRF family members.
Co-transfection of IRF-5 and IRF-7 did not activate the expression of the IFNA1 promoter. Because both IRF-5 and IRF-7 were able to transactivate the IFNA1 promoter when expressed alone in the cells, the low transactivation activity of the IRF-5/IRF-7 heterodimers in infected cells was unexpected. To determine whether the transactivating potential of IRF-5 could be rescued, we co-transfected the IFNA1 SAP with a constant amount of IRF-7 and increasing amounts of IRF-5. The results shown in Fig. 2C indicate that the transactivation potential of IRF-5 could be increased when the levels of IRF-5 in the cells were significantly higher than IRF-7 levels, as detected by immunoblot analysis (data not shown). Similarly, activation of the IFNA promoter was restored in cells that expressed higher levels of IRF-7 than IRF-5 (data not shown). These data indicate that the IRF-5/IRF-7 heterodimer is unable to stimulate transcription of the IFNA1 gene in infected cells, yet either of these IRF could stimulate IFNA gene transcription when its relative level of expression predominated.
Competition between Binding of IRF-5 and IRF-7 to IFNA VRE Is Dependent on the Relative Levels of IRF Expressed-The data from the SAP reporter assay revealed that IRF-5/ FIG. 2. Transcriptional activation of the IFNA1 promoter by IRF-5 heterodimers. A, the IRF-5/IRF-7 heterodimers repressed transactivation of the IFNA1 SAP reporter in infected cells. Expression plasmids encoding IRF-1, IRF-5, or IRF-7 were co-transfected with the IFNA1 SAP reporter plasmid together with the ␤-galactosidase-expressing plasmid to HeLa cells, as described under "Experimental Procedures." Cells were infected with NDV for 16 h or left uninfected, and SAP activity was measured as described under "Experimental Procedures." SAP levels were normalized to the constant level of ␤-galactosidase expressed. B, distinct effect of IRF-1 on the transactivation potential of IRF-5 and IRF-7. Constant amounts of the IFNA1 SAP reporter, IRF-5, or IRF-7 (2 g) were co-transfected with increasing amounts of IRF-1 (2, 4, and 8 g) to HeLa cells. Cells were left uninfected or infected with NDV for 16 h, and SAP activity was measured. C, repression of the IFNA1 SAP reporter by IRF-5/IRF-7 heterodimers was overcome in a dose-dependent manner by IRF-5. Constant amounts of the IFNA1 SAP reporter and IRF-7 (2 g) were co-transfected with increasing amounts of IRF-5 (2, 4, and 8 g) to HeLa cells. Cells were left uninfected or infected with NDV for 16 h, and SAP activity was measured.

Virus-induced Heterodimer Formation between IRF-5 and IRF-7
IRF-7 heterodimers repressed IFNA1 gene transcription. We have therefore examined the binding of IRF-5 and IRF-7 to IFNA VRE from lysates of infected cells that express these IRF. Because IRF-5 can form heterodimers with IRF-3 (5), we have also examined whether binding of IRF-5 to IFNA VRE is dependent on the binding of IRF-3. We have shown previously (21) that IRF-3 binds to the IFNA1 but not IFNA14 VRE; therefore, we have examined the binding of IRF-5 (21) or IRF-7 to these two VRE using the oligonucleotide pull-down assay. Lysates of 2fTGH/IRF-5 and 2fTGH/IRF-7 cells infected either with Sendai virus or NDV were utilized for these experiments. The top two panels in Fig. 3A show the binding of IRF-5 and FIG. 3. IRF-5 and IRF-7 compete for binding to IFNA VRE. A, binding of IRF-7 or IRF-5 from cell lysates (250 g) of infected or uninfected 2fTGH/IRF-7Ϫexpressing cells or 2fTGH/IRF-5Ϫexpressing cells to IFNA1 and IFNA14 VRE. 2fTGH/IRF-7 or 2fTGH/IRF-5 cells were then transfected with FLAG-tagged IRF-5 (2fTGH/IRF-7 ϩ IRF-5) or IRF-7 (2fTGH/IRF-5 ϩ IRF-7), respectively. The relative levels of IRF-5, IRF-7, and endogenous IRF-3 in cell lysates were estimated by immunoblot analysis (10% input). The same extracts were used for the oligonucleotide pull-down assay as described under "Experimental Procedures," and the levels of IRF-5, IRF-7, and IRF-3 bound to the respective IFNA1 or -A14 VRE were consequently detected by immunoblotting. Blots were stripped and re-probed with each antibody in the order shown. Ab, antibody; SeV, Sendai virus. B, IRF-7 and IRF-5 compete for binding to IFNA1 VRE. 2fTGH/IRF-7 or 2fTGH/IRF-5 cells were transfected with increasing amounts of FLAG-tagged IRF-5 or IRF-7, respectively, and infected with NDV. The relative levels of IRF-5 and IRF-7 in cell lysates were determined by immunoblotting (10% input). The extracts were used for the oligonucleotide pull-down assay where the levels of IRF-5 and IRF-7 bound to the IFNA1 VRE were detected by immunoblotting. Relative binding levels were quantitated (binding affinity) to show the relative intensity of IRF-5 or IRF-7 bound as a function of the amount of transfected plasmid.

Virus-induced Heterodimer Formation between IRF-5 and IRF-7
IRF-7 to the IFNA1 and -A14 VREs. Binding of IRF-7 to IFNA1 or IFNA14 VRE was detected only from lysates of infected cells and was higher from lysates of Sendai virus-infected cells than NDV-infected cells. In contrast, binding of IRF-5 to IFNA1 and IFNA14 VRE was detected both from lysates of uninfected and infected cells. However, although the binding of IRF-5 from uninfected and Sendai virus infected cells was about the same, it was greatly enhanced after infection with NDV.
We next compared binding of IRF-5 and IRF-7 to IFNA VREs from lysates of 2fTGH cells expressing both IRF-5 and IRF-7. For these studies, we either transfected the 2fTGH/IRF-7 cells with IRF-5 (2fTGH/IRF-7 ϩ IRF-5) or the 2fTGH/IRF-5 cells with IRF-7 (2fTGH/IRF-7 ϩ IRF-5), and the lysates were then prepared from infected and uninfected cells. Results from immunoblot analysis of the input lysates (Fig. 3A) show that the cells expressed the endogenous IRF at higher levels than the transfected IRF; IRF-3 was expressed at about the same levels in both lysates. As shown in Fig. 3A, IRF-3 was bound only to the IFNA1 VRE from lysates of infected cells, which is in correlation with previous reports (19,21). In lysates of cells that expressed higher levels of IRF-7 than IRF-5 (2fTGH/ IRF-7 ϩ IRF-5), binding of IRF-7 to both IFNA1 and IFNA14 VRE could be detected from lysates of infected cells. However, binding of IRF-5 to IFNA1 and -A14 VRE was drastically reduced in these cells. The reduction in IRF-5 binding could not be related to the absence of IRF-3 because binding to both IFNA1 and -A14 VRE was greatly reduced. In the cells expressing higher levels of IRF-5 than IRF-7 (2fTGH/IRF5 ϩ IRF-7), binding of IRF-5 to IFNA1 and -A14 VRE was detected both from the lysates of infected and uninfected cells; however, IRF-5 bound more efficiently from the lysates of NDV-infected cells. The binding of IRF-7 to these VRE was insignificant, and no difference between binding from infected and uninfected lysates could be detected. Taken together, these results indicate that the binding of IRF-5 and IRF-7 to IFNA VRE reflects the relative levels of expression in the cells and that these two IRF probably do not bind to the IFNA VRE as a heterodimer.
In order to determine whether IRF-5 and IRF-7 are competing for binding to the IFNA1 VRE, we used the oligonucleotide pull-down assay to examine the binding profiles of one IRF, in the presence of increasing amounts of the second IRF. As shown in Fig. 3B, increasing amounts of IRF-5 expressed in the 2fTGH/IRF-7 cells competed the binding of IRF-7 to the IFNA1 VRE. In a similar manner, when IRF-7 levels were incrementally increased in 2fTGH/IRF-5 cells, binding of IRF-5 was diminished. Interestingly, the binding affinity of IRF-5 for IFNA1 VRE seems to be higher than that of IRF-7 binding to IFNA1 VRE. By graphing the level of binding of IRF-5 or IRF-7 as a function of the amount of transfected plasmid, it became apparent that the binding of IRF-5 to the IFNA1 VRE occurred at a 2-3-fold lower concentration than binding of IRF-7. These results indicate that IRF-5 and IRF-7 do not bind as a heterodimer; instead their binding is dependent on the relative levels of expression and affinity for the IFNA promoter.
Induction of Biologically Active IFN␣ in Virus-infected 2fTGH Cells Expressing IRF-5 and IRF-7-It was shown previously (21) that induction of IFNA genes by IRF-5 occurred only in cells infected by NDV but not Sendai virus. However, the induction of IFNA genes in cells that express both IRF-5 and IRF- 7 has not yet been examined. We have therefore analyzed the expression of IFNA genes and proteins in 2fTGH cells expressing both ectopic IRF-5 and IRF-7. For these studies, we have used either NDV-infected 2fTGH/IRF-5 cells transfected with IRF-7 or 2fTGH/IRF-7 cells transfected with IRF-5. Fig. 4A illustrates that in the absence of IRF-7, NDV induced high levels of IFN␣ (1500 units/ml) in 2fTGH/IRF-5 cells; yet these levels were drastically reduced when cells also expressed IRF-7 (320 units/ml). Furthermore, whereas Sendai virus did not induce production of IFN␣ in 2fTGH/IRF-5 cells, it induced low levels of IFN␣ (80 units/ml) when the 2fTGH/ IRF-5 cells were also expressing IRF-7. The level of IRF-5 and IRF-7 protein expression in 2fTGH cells, as determined by immunoblot analysis, indicated that the levels in these cells were comparable. Immunoblot analysis of cell lysate from 2fTGH/IRF-7 cells transfected with IRF-5 showed significantly higher levels of IRF-7 than IRF-5. In 2fTGH/IRF-7 cells, IFN␣ was effectively induced both by Sendai virus (800 units/ml) and NDV (350 units/ml) infection. However, in the presence of IRF-5, the levels of IFN␣ produced were significantly decreased in Sendai virus-infected cells (10-fold reduction), whereas in NDV-infected cells, the decrease was only about 2-fold. The relative levels of IFNA mRNA present in Sendai virus and NDV-infected cells correlated with the levels of biologically active IFN␣. These data indicate that in cells expressing both IRF-5 and IRF-7, the production of biologically active IFN␣ was lower than in cells expressing only IRF-5 or IRF-7; expression of IRF-5 significantly decreased IFN␣ production in Sendai virus-infected cells.
We next examined whether co-expression of IRF-5 and IRF-7 also affected the profile of induced IFNA subtypes. We have shown previously that NDV-infected 2fTGH/IRF-5 cells preferentially induced the IFNA8 subtype (Table I, values in parentheses, column 3), whereas 2fTGH/IRF-7 cells preferentially expressed the IFNA1 subtype (Table I,  amplified the IFNA mRNA and cloned the fragments from NDV-infected cells, as described previously (25,40). We have found that this method yields a more accurate identification of the IFNA subtypes expressed than the previously used RNA protection assay (48). Randomly selected clones were then identified by sequencing (21,25). As shown in Table I, 2fTGH/ IRF-5 ϩ IRF-7 and 2fTGH/IRF-7 ϩ IRF-5 transfected cells expressed either IFNA8 or IFNA1, respectively, as the predominant subtype. However, co-expression of IRF-5 or IRF-7 introduced two major changes: 1) although the 2fTGH/IRF-7 cells did not express IFNA8, the presence of IRF-5 in these cells conferred expression of IFNA8 and IFNA2; and 2) in the 2fTGH/IRF-5 cells, co-expression of IRF-7 decreased expression of IFNA8 and IFNA7 while enhancing expression of IFNA1. These data indicate that co-expression of IRF-5 and IRF-7 in infected cells modulates the levels of the IFNA subtypes conferred by either IRF alone.
IRF-5 and IRF-7 Compete for Binding to the Endogenous IFNA Promoters in 2fTGH Cells-We have shown previously (19) that IRF-1, IRF-3, and IRF-7 are components of the enhanceosome binding to IFNA1 VRE in infected cells. Because the binding of IRF-5 and IRF-7 to IFNA1 VRE, as analyzed by the oligonucleotide pull-down assay, shows a competitive binding exclusion, we sought to determine the binding of IRF-5 and IRF-7 to the endogenous IFNA promoters in vivo. To this end, we used the chromatin immunoprecipitation assay (ChIP). The 2fTGH/IRF-5 or 2fTGH/IRF-7 cells utilized for these studies (Fig. 5A) were transiently transfected either with IRF-7 or IRF-5, respectively (Fig. 5B), and then infected with Sendai virus or NDV. Proteins were cross-linked to DNA, and the protein-DNA complexes were precipitated with antibodies specific for IRF-1, IRF-3, IRF-5 (FLAG), or IRF-7 (Fig. 5A), as described under "Experimental Procedures." The analyzed IFNA promoters were first amplified from DNA-protein complexes before immunoprecipitation to ensure that the amount of DNA used for precipitation with different antibodies was constant (data not shown). As described previously (5), the conditions of PCR were optimized to give rise to a linear amplification response of the endogenous IFNA promoter region from the immunoprecipitated DNA-protein complexes. Amplification of IFNA promoters from the DNA-protein complexes immunoprecipitated with IRF-1, IRF-3, or IRF-5 antibodies from 2fTGH/IRF-5 cells is shown in Fig. 5A, top three panels. The fragment containing the IFNA promoter region was amplified only from DNA-protein complexes immunoprecipitated by IRF-1 and IRF-3 antibodies from infected cells, but no amplification was observed in immunoprecipitates from uninfected cells. However, the endogenous IFNA promoter was amplified from complexes immunoprecipitated with anti-FLAG antibodies (IRF-5) in both uninfected and NDV-infected cells

induced by NDV infection in 2fTGH cells expressing ectopic IRF-5 and IRF-7
The amplified IFNA cDNA fragments representing a mixture of IFNA mRNA were cloned as described under "Experimental Procedures." Thirty randomly selected clones were sequenced and analyzed. Values in boldface illustrate major subtype induced by NDV infection in IRF-5-or IRF-7-expressing 2fTGH cells (21). Underlined values are representative of major subtypes induced after co-expression of both IRF family members.

IFNA
IRF-5 ϩ IRF-7 IRF-7 ϩ IRF-5  5. IRF-5 and IRF-7 compete for binding to the endogenous IFNA promoters in 2fTGH cells. A, in vivo binding of IRF-1, IRF-3, IRF-5, and IRF-7 to IFNA promoters in infected 2fTGH/IRF-5 or 2fTGH/IRF-7Ϫexpressing cells. A ChIP of DNA-protein complexes was performed as described under "Experimental Procedures" with anti-IRF-1, IRF-3, or IRF-7 polyclonal antibodies (Ab). IP, immunoprecipitation; SeV, Sendai virus. The IRF-5 binding was detected by precipitation with anti-FLAG monoclonal antibodies. IFNA promoters were amplified from the precipitates by using primers that detect promoters of all IFNA genes. The relative levels of amplified fragments of IFNA promoters are shown. The IFNA promoters were also amplified from the DNA-protein complexes before the precipitation to normalize the amount of DNA used for precipitation (data not shown). B, 2fTGH/IRF-5Ϫexpressing cells or 2fTGH/IRF-7Ϫexpressing cells transfected with IRF-7 and FLAG-tagged IRF-5, respectively, were infected with virus or left uninfected. The ChIP assay was performed as described above and fragments of IFNA promoters were amplified from the precipitates.
Virus-induced Heterodimer Formation between IRF-5 and IRF-7 (Fig. 5A). These data indicate that IRF-1, as well as IRF-3 and IRF-5, are components of the IFNA enhanceosome in NDVinfected cells, but neither IRF-1 nor IRF-3 binds to the IFNA promoters in uninfected cells. In contrast, IRF-5 binds to IFNA promoters in uninfected cells (5,21). In the 2fTGH/IRF-7 cells, the IFNA promoter region was amplified from complexes immunoprecipitated by IRF-7 antibodies in virus-infected cells but not in uninfected cells.
These results indicate that in the presence of high levels of IRF-5, binding of IRF-7 to the IFNA promoter was below the levels of detection. Interestingly, in 2fTGH/IRF-7 cells expressing higher levels of IRF-7 than IRF-5, the IFNA promoter could be amplified from anti-IRF-3, IRF-7, and FLAG immunoprecipitates of DNA-protein complexes from NDV-infected cells. However, the IFNA promoter was amplified only from anti-IRF-3 and IRF-7 immunoprecipitates of DNA-protein complexes from Sendai virus-infected cells. These results confirm the previous observation that Sendai virus infection does not stimulate transcriptional activity of IRF-5 (5,21). Taken together, these data indicate that whereas IRF-3, IRF-5, and IRF-7 bind to the IFNA promoter in NDV-infected cells expressing high levels of IRF-7, IRF-5 has a greater binding affinity than IRF-7 and out-competes the binding of IRF-7 when these factors are expressed in the cells at about the same levels.
Characterization of the IFNA Enhanceosome Complex-In order to determine whether IRF-5 is a component of the enhanceosome containing IRF-3, IRF-7, and IRF-1 which binds to IFNA promoters in cells that express all of these IRF, such as dendritic cells, 3 we performed a sequential ChIP assay (19). The DNA-protein complexes from uninfected or NDV-infected 2fTGH/IRF-5 cells transfected with IRF-7 that express endogenous IRF-1 and IRF-3 (19) were first precipitated with anti-IRF-3 (Fig. 6A), anti-FLAG (Fig. 6B), or anti-IRF-7 (Fig. 6C) antibodies, and the IFNA promoter region was then amplified from these precipitates. The supernatants of these precipitates were then again precipitated with anti-IRF-1, anti-IRF-3, anti-FLAG, or anti-IRF-7 antibodies, and the IFNA promoter region was amplified from the second immunoprecipitates. Results in Fig. 6A show that the IFNA promoters could be amplified from the anti-IRF-3 precipitates but not from the subsequent anti-IRF-1, IRF-3, or IRF-7 precipitates indicating that the complex binding to the IFNA promoter contains IRF-3, IRF-1, and IRF-7 (19). However, the IFNA promoter could be amplified when the second precipitation was performed with anti-FLAG antibodies that detect IRF-5. This result would suggest that in both uninfected and NDV-infected cells, IRF-5 can bind as a homodimer in the absence of IRF-3 and the other IRF family members. The fact that no amplification of the IFNA promoter could be detected from a second precipitation with anti-IRF-3 antibodies indicates that the first precipitation was quantitative.
In the case where the initial precipitation was performed with anti-FLAG (IRF-5) antibodies (Fig. 6B), the IFNA promoter could be amplified from both infected and uninfected cells. Low levels of the IFNA promoter could also be amplified from second precipitations with anti-IRF-1, IRF-3, and IRF-7 antibodies from infected cells, supporting the previous observation that binding of these three IRF is independent of IRF-5 binding. Finally, the IFNA promoter was amplified when the DNA-protein complexes from infected cells were first precipitated with IRF-7 antibodies and from the subsequent precipitates with anti-IRF-3 or IRF-5 antibodies (Fig. 6C). However, in the absence of IRF-7, no binding of IRF-1 to the IFNA promoter was detected. Thus, although we did not detect binding of IRF-7 in the absence of IRF-3 (Fig. 6A), indicating that IRF-7 binds as a heterodimer with IRF-3 (19,38), precipitation with anti-IRF-7 antibodies did not completely remove IRF-3 protein revealing that IRF-3 can also bind in the absence of IRF-7. When the DNA-protein complexes from NDV-infected 2fTGH cells overexpressing IRF-5 and IRF-7 were first precipitated with IRF-1 antibodies, the IFNA promoter was efficiently amplified (data not shown). Interestingly, secondary precipitations with anti-IRF-1, IRF-3, IRF-5, or IRF-7 antibodies revealed that PCR amplification of the IFNA promoter could only be detected from IRF-3 or IRF-5 precipitates indicating again that binding of IRF-3 and IRF-5 but not IRF-7 to this promoter can occur in the absence of IRF-1. Thus, IRF-1 is an important part of the IFNA enhanceosome complex in cells expressing IRF-7 but not in cells expressing IRF-3 and IRF-5. These results would indicate that although IRF-5 can bind to the endogenous IFNA promoter in the absence of IRF-1, IRF-3, or IRF-7, its binding is greatly enhanced in the presence of FIG. 6. In vivo characterization of the endogenous IFNA enhanceosome complex in 2fTGH cells by the ChIP assay. A, the cross-linked DNA-protein complexes from NDV-infected 2fTGH/IRF-5 cells transfected with IRF-7 were first precipitated with anti-IRF-3, IRF-5, or IRF-7 antibodies. The precipitates were collected by centrifugation, and the supernatants were divided and subjected to a second round of precipitation with either IRF-1, IRF-3, IRF-5, or IRF-7 antibodies, respectively. The first and second precipitates were dissolved in 60 l of water, and 15 l of these samples were used for PCR amplification using primers that detect all IFNA promoters, as described under "Experimental Procedures." The amplified IFNA promoter fragments are shown.
Mapping the IRF-5/IRF-7 Interaction Domain-By the GST pull-down and co-immunoprecipitation assays, we have shown that IRF-5 and IRF-7 form heterodimers in infected cells. It appears that these heterodimers have a lower DNA binding capacity than either IRF alone. In order to fully discern the reason why the IRF-5/IRF-7 heterodimer has a lower DNA binding capacity, we have mapped the interacting regions between IRF-5 and IRF-7. To this effect, cell lysates from NDVinfected 2fTGH/IRF-7 and 2fTGH/IRF-5 were incubated with either immobilized amino-terminal or carboxyl-terminal GST-IRF fusion proteins, and the specifically bound proteins were detected by immunoblot analysis with the appropriate serum. As shown in Fig. 7A, IRF-5 interacts with both amino-and carboxyl-terminal GST-IRF-7 fusion proteins, whereas IRF-7 interacts only with the amino terminus of IRF-5 (Fig. 7B). This is in contrast to IRF-3, which interacts with the carboxyl terminus of IRF-5 (Fig. 7B) (5). These results indicate that forma-tion of IRF-5/IRF-7 heterodimers results in masking of the IRF-5 and IRF-7 DNA binding domains. DISCUSSION It has been generally accepted that the transcription factors IRF-3 and IRF-7 play a critical role in the induction of type I IFN genes in infected cells. The expression of IRF-3 is sufficient to support virus-mediated activation of IFNB genes (31), whereas IRF-7 is required for the transcriptional activation of IFNA genes (24,25). A somewhat surprising finding was that ectopic expression of IRF-5 could rescue the transcription of IFNA genes in the absence of IRF-7 (9,21). The expression of IRF-5 and IRF-7 in cells is not exclusive, and the presence of IRF-7 and IRF-5 transcripts can be detected in uninfected monocytes and dendritic cells, as well as precursor dendritic cells (pDC2) that are high producers of IFN␣. 3 Moreover, type I IFN can further stimulate expression of both of these factors. It has therefore become important to determine the contribution of IRF-5 to the activation of IFNA genes in the context of IRF-7. In the present study, we show that although IRF-5 and IRF-7 can form heterodimers, this interaction is not cooperative and results in an inhibition of IFNA gene transcription.
By addressing the molecular mechanism resulting in the IRF-5Ϫmediated suppression of IRF-7 activity, we have characterized the interaction between IRF-5 and other IRF family members known to be involved in the activation of IFN genes, and we have observed that IRF-5 interacts with both IRF-3 and IRF-7 (summarized in Fig. 8). The presence of IRF-5/IRF-7 heterodimers could be detected in infected but not uninfected cells. We have shown previously (5, 21) that NDV but not Sendai virus infection activates IRF-5 by phosphorylation in the carboxyl terminus of the protein, whereas IRF-7 and IRF-3 are activated by both of these viruses. The fact that we could detect binding between IRF-5 and IRF-7 in lysates of cells infected with both Sendai virus and NDV indicates that phosphorylation of IRF-7, but not IRF-5, is required for heterodimer formation. It was shown that IRF-3 and IRF-7 homodimers, as well as IRF-3/IRF-7 heterodimers, could be detected only in infected cells (22,37,38). Consequently, it was proposed that the virus-mediated phosphorylation of IRF-3 and IRF-7 resulted in allosteric changes that allowed for dimerization of these proteins and their interactions with the other members of the IRF family, as well as CBP/p300 and PCAF. Phosphorylation of IRF-3 and IRF-7 is also necessary for their retention in the nucleus (22,37,49). Because the virus-induced phosphorylation of the serine-rich carboxyl-terminal region of IRF-5 (5) does not seem to be required for its ability to form homodimers or heterodimers with other IRF, this indicates that in the IRF-5 polypeptide, unlike the other two IRF (IRF-3 and IRF-7), the interaction domain is not masked. Furthermore, the carboxylterminal phosphorylation of IRF-5 may not be a prerequisite for its nuclear retention because the IRF-5 polypeptide contains two nuclear localization signals that facilitate its nuclear transport even in uninfected cells (5). These data indicate that the IRF-5 polypeptide shows several unique features that distinguish this transcription factor from IRF-3 and IRF-7, and the functions of IRF-5 and IRF-7 are not redundant.
Overexpression of IRF-5 rescued IFN␣ production in the absence of IRF-7, and in the transient transfection assay, IRF-5 enhanced IRF-3Ϫmediated transcriptional activity of the IFNA1 and IFNB promoters in infected cells (5). However, in contrast to the cooperative interaction between IRF-3/IRF-5, the interaction between IRF-5 and IRF-7 resulted in the repression of IFNA1 transcriptional activity. The fact that the basal activity of IFNA promoters was not affected in uninfected cells expressing both IRF-7 and IRF-5 revealed that the IRF-5/IRF-7 heterodimers do not block transcription by interaction FIG. 7. Mapping the IRF-5/IRF-7 interaction domains. A, formation of the IRF-5/IRF-7 heterodimer is mediated through the IRF-5 DNA binding domain. Whole cell lysate (250 g) from NDV-infected 2fTGH/IRF-5Ϫexpressing cells was used for mapping the IRF-7 interaction domain. The relative levels of IRF-5 in cell lysates were determined by immunoblotting (10% input). Extracts were applied to GSTagarose beads (lane 1), amino-terminal IRF-7 fused to GST-agarose beads (lane 2), or carboxyl-terminal IRF-7 fused to GST-agarose beads (lane 3). Specifically bound proteins were detected by immunoblotting with anti-FLAG antibodies (Ab). B, whole cell lysate (250 g) from NDV-infected 2fTGH/IRF-7Ϫexpressing cells was used for mapping the IRF-5 interaction domain. The relative levels of IRF-7 and IRF-3 in cell lysates were determined by immunoblotting (10% input). Extracts were applied to GST-agarose beads (lane 1), amino-terminal IRF-5 fused to GST-agarose beads (lane 2), or carboxyl-terminal IRF-5 fused to GSTagarose beads (lane 3). Specifically bound proteins were detected by immunoblotting with anti-IRF-7 or anti-IRF-3 antibodies.
with basal transcription factors. The block in transcription was limited to infected cells, suggesting that the transactivation potential of IRF-5 and IRF-7 is lost when these factors heterodimerize. This inhibition is likely a result of heterodimer formation that masks the IRF-5 or IRF-7 DNA binding domain and their ability to bind to the IFNA promoters (Fig. 8B). When both IRF-5 and IRF-7 were present in the lysates of infected cells, their binding to IFNA1 and IFNA14 VRE was mutually exclusive, whereas the binding of IRF-3 to IFNA1 VRE was not modulated by the presence of IRF-5 or IRF-7. A similar competition of IRF-5 or IRF-7 binding to the endogenous IFNA promoters was observed in infected cells expressing both IRF-5 and IRF-7 by the ChIP assay. These data further show that in the cells expressing high levels of IRF-5, IRF-5 but not IRF-7 was the major component of the IFNA enhanceosome. In contrast, when the relative levels of IRF-7 were higher than IRF-5, IRF-7 bound to the IFNA promoters, yet the binding of IRF-5 was not completely excluded. It is important, however, to remember that analysis of IRF bound to the endogenous IFNA promoters gives an overall picture of IRF binding to the majority of IFNA promoters. This assay does not allow one to distinguish quantitatively IRF binding to the promoters of individual endogenous genes. Thus, the possibility that binding of IRF-5 and IRF-7 to the endogenous IFNA promoters detected in these assays represents binding to distinct IFNA promoters cannot yet be excluded.
The analysis of the expression of IFNA subtypes in cells overexpressing either IRF-5 or IRF-7 does indicate that IRF-5 and IRF-7 target different IFNA promoters (21,25). We have shown previously that viral infection of 2fTGH/IRF-7 cells induced predominantly expression of IFNA1, and no expression of IFNA2 or IFNA8 genes could be detected in these cells. However, when IRF-5 was co-expressed in these cells, both IFNA2 and IFNA8 genes were expressed. Similarly, IRF-7 expression in 2fTGH/IRF-5 cells resulted in the enhanced expression of IFNA1 and decreased expression of the IFNA8 and IFNA7 genes, whereas the 2fTGH/IRF-5 cells expressed predominantly IFNA8 and only low levels of IFNA1. These results indicate that the relative levels of IRF-5 and IRF-7 in the cells significantly modulate the expressed levels of IFNA subtypes. In addition, we have shown previously that the relative levels of IRF-3 and IRF-7 also affect the expression of IFNA subtypes (25,26). Taken together, these results suggest that despite the close sequence homology between the subtypes of IFNA promoters, the transcriptional activity of these individual promoters shows a distinct activation response to IRF-3, IRF-5, IRF-7, and their combinations. Thus, it is unlikely that the level of IRF-3 and IRF-7 DNA binding alone determines the specificity and activation of individual IFNA genes in infected cells.
Interestingly, in the context of the IFNA1 promoter, IRF-5 can function as an activator when present as homodimers or heterodimers with IRF-3 and as a repressor when interacting with IRF-7. IRF-5 is not the only member of the IRF family that can exert both activation and repression activity. Thus, as shown in this study, high levels of IRF-1 result in a suppression of IFNA1 transactivation activity when complexed with IRF-5, yet IRF-1 and IRF-7 cooperatively activate this promoter. Previously it has been shown that IRF-2 represses the transcriptional activity of IRF-1 (13) and inhibits transcriptional activation of the IFNB gene (14), whereas it activates transcription of the VCAM gene (50) and histone 4 (51,52). Furthermore, IRF-7 was initially discovered for its ability to repress the Q promoter of the EBV-encoded EBNA genes but can effectively activate IFNA and IFNB genes (28). Finally, IRF-8 acts as an activator upon binding to PU.1 (53)(54)(55)(56)(57), yet in association with another member of the ETS family, Tel, it acts as a strong repressor of promoters containing ISRE sites (58). The molecular mechanism by which IRF-5 suppresses IRF-7 activation of the IFNA1 promoter is not yet clearly established. Although IRF-7 interacts with both IRF-3 and IRF-5 in a phosphorylationdependent manner, the interaction with IRF-3 activates the transcription of the IFNA1 promoter, whereas the interaction with IRF-5 represses it. The ability of IRF-5 to interact with IRF-3 and IRF-7 differs in several ways. Whereas IRF-7 and IRF-5 independently bind to the IFNA1 VRE both in vitro and in vivo, there is competition between IRF-7 and IRF-5 binding to this VRE element in vivo. Results from the GST pull-down assay suggest that interaction of IRF-5 with IRF-7 and formation of heterodimers occur through association with the aminoterminal regions of these proteins, thus masking their DNA FIG. 8. IRF-5 dimerization and promoter binding. A, schematic representation of IRF-5 dimerization and binding to the endogenous IFNA promoters in uninfected cells. IRF-5 is present in the cytoplasm and nucleus of uninfected cells due to low levels of constitutive phosphorylation and the presence of an unmasked carboxyl-terminal nuclear localization signals (5). In uninfected cells, IRF-5 can form homodimers (5) and IRF-5/IRF-1 heterodimers; yet only IRF-5 homodimers can bind to the endogenous IFNA promoters. No heterodimerization between IRF-5 and IRF-3 or IRF-7 can be detected in uninfected cells. B, schematic illustration of IRF-5 dimerization and binding to endogenous IFNA promoters in virus-infected cells. Phosphorylation (P) of Ser/Thr residues in the carboxyl terminus leads to conformational changes in IRF-3, IRF-5 (5), and IRF-7 that further relieves carboxyl-terminal autoinhibition and exposes both DBD and interaction domain regions (IAD). The open conformations of IRF-3 and IRF-7 are now able to heterodimerize with IRF-5. Translocation to the nucleus leads to DNA binding at ISRE-and PRD-like (LE) containing promoters. Whereas IRF-5 interacts with IRF-3 through the carboxylterminal interaction domain forming a functional heterodimer, IRF-5/ IRF-7 heterodimerization occurs through the DBD and thus does not lead to a functional complex binding to the IFNA promoters. binding domains (Fig. 8B). Furthermore, whereas IRF-7, IRF-5, and IRF-3 can all bind to the IFNA promoters in vivo, IRF-7 is a part of an enhanceosome that contains IRF-3 and IRF-1, whereas IRF-5 does not seem to be a part of this enhanceosome complex. All of these IRF interact with CBP/p300 and/or PCAF, and these interactions are essential for the transactivation by IRF-3 and IRF-7 (59 -62). The histone transacetylases are also a part of the enhanceosome containing IRF-3, IRF-1, and IRF-7 (19). Recent data (63)(64)(65) indicate that both IRF-3 and IRF-7 are acetylated, and acetylation modulates their function. Thus, there are at least two other possibilities by which IRF-5 can repress activation by IRF-7. The first is by quelching the interaction of IRF-7 with the histone transacetylase PCAF, as was shown for the E1A adenovirus-encoded protein (66). The second is by its ability to recruit HDAC. The selective recruitment of HDAC to the neighborhood of the IRF-3 and IRF-7 enhanceosome may result in an alteration of chromatin structure and block the binding of IRF-7 to the enhanceosome. Although we do not have any evidence that either of these mechanisms is operative, it is not without interest that we have observed association between IRF-5 and histone transacetylases CBP/ p300, PCAF, and the repressors SMRT, Skip, and mSin3a that are components of the HDAC co-repressor complex. 2 The repression of IRF-3-and IRF-7-mediated activation of some murine IFNA promoters by interaction with the homeoprotein Pitx1 binding to an upstream region of murine IFNA promoters was recently observed (67). Whether this protein plays any role in the repression of transcriptional activity of human IFNA genes is not known. Altogether, these data indicate that IRF-5 expression may be a part of the regulatory network that ensures timely expression of the immediate early genes. In infected cells, the expression of IRF-7 is strongly enhanced by type I IFN produced in the context of viral infection and thus the IRF-5mediated repression may be overcome by high levels of IRF-7 and result in the consequent stimulation of IFNA gene expression. Additional experiments are needed to address these possibilities.
In summary, we have described that IRF-5 cannot only activate and cooperate with IRF-3 in the stimulation of IFNA gene transcription, but it can also mediate the suppression of IRF-7-induced expression of IFNA genes in response to viral infection. Thus, it appears that in the context of an individual promoter, IRF-5 can exert both stimulating activity in association with IRF-3 and suppressing function when associated with IRF-7 (Fig. 8). One of the mechanisms by which the suppressive role is acquired is a result of IRF-5/IRF-7 heterodimer formation and inhibition of IRF-5 and IRF-7 binding to the VRE of IFNA promoters. On the other hand, in infected cells, binding of IRF-5 to IFNA promoters appeared to be separate from the binding of the IRF-1, IRF-3, and IRF-7 containing enhanceosome, suggesting that IRF-5 may not be a component of this complex. Considering the fact that an IRF-3/IRF-7-mediated activation is dependent on the recruitment of CBP/ p300, further experiments are underway to determine whether IRF-5-mediated suppression is also coupled to the recruitment of the HDAC co-repressor complex to the IFNA promoter.