Convergence of TNF[alpha] and IFN[gamma] signalling pathways through synergistic induction of IRF-1/ISGF-2 is mediated by a composite GAS/[kappa]B promoter element
Convergence of TNF [alpha] and IFN [gamma] signalling pathways through synergistic induction of IRF-1/ISGF-2 is mediated by a composite GAS/ [kappa]B promoter elementRichard Pine
Public Health Research Institute, New York, NY 10016, USA
Received June 26, 1997;Revised and Accepted September 19, 1997
ABSTRACT
The molecular basis for the well known synergistic biological effects of tumor necrosis factor [alpha] (TNF[alpha]) and interferon [gamma] (IFN[gamma]) is still poorly understood. This report demonstrates that expression of interferon-regulatory factor 1 (IRF-1), also known as interferon-stimulated-gene factor 2 (ISGF-2), is synergistically induced by these cytokines. The induction is a primary transcriptional response that occurs rapidly without a requirement for new protein synthesis. Synergism is mediated by a novel composite element in the IRF-1 promoter that includes an IFN[gamma]-activation site (GAS) overlapped by a non-consensus site for nuclear factor kappa B (NF[kappa]B). These sequences are bound strongly by signal transducer and activator of transcription 1 (STAT-1) and weakly by the p50/p65 heterodimer form of NF[kappa]B, respectively. However, the binding of STAT-1 and NF[kappa]B to the GAS/[kappa]B element in vitro seems to be mutually exclusive and independent. Synergistic induction of IRF-1 is likely to be an important early step in regulatory networks critical to the synergism of TNF[alpha] and IFN[gamma]. The GAS/[kappa]B element may mediate synergistic transcriptional induction of IRF-1 by other pairs of ligands that together activate NF[kappa]B and STAT family members. Other genes are likely to contain this motif and be regulated similarly.
INTRODUCTION
The effects of tumor necrosis factor [alpha] (TNF[alpha]) and interferon [gamma] (IFN[gamma]) on cell growth control are well known, and are related to inflammatory and immunomodulatory properties of these cytokines (reviewed in refs 1 -6 ). In addition to their individual effects, there are notable synergistic responses to the combination of TNF[alpha] plus IFN[gamma] (7 -13 ). Many of these consequences apparently arise from the ability of both TNF[alpha] and IFN[gamma] to change the cellular program of gene expression.
Each of these cytokines has led to a paradigm for signal transduction and transcriptional regulation. In both cases, receptor-ligand interaction initiates a rapid signal transduction cascade which leads to activation and translocation to the nucleus of preexisting cytoplasmic nuclear factor kappa B (NF[kappa]B) or signal transducer and activator of transcription 1 (STAT-1), in response to TNF[alpha] or IFN[gamma], respectively. Each of these two proteins leads to increased transcription of target genes when bound to specific DNA regulatory elements (14 -20 ).
Specificity of the pathway that leads from a receptor to the particular transcription factors that are affected sets the stage for distinct patterns of gene expression induced by different cytokines. However, the final determinants of specificity for transcriptional regulation of gene expression are the particular combinations of regulatory elements in the promoters of different genes and the combinatorial features of transcription factor function (for example, 21 ,22 ). Synergism of TNF[alpha] and IFN[gamma] provides a particularly relevant example of this principle. Synergistic induction of the chemokine IP-10, the cell adhesion molecule ICAM-1, and certain MHC class I antigens results from the combined effects of transcription factors that are induced predominantly or exclusively by one or the other cytokine at the time the synergism occurs. Those transcription factors then bind to distinct, separate sites in the promoters of the genes that encode those proteins (9 ,10 ,13 ). Pairs of transcription factors so far implicated in the synergistic induction include NF[kappa]B plus STAT-1, or NF[kappa]B plus interferon regulatory factor 1 (IRF-1).
IRF-1 has been studied in the contexts of inflammation and immunity and of cell growth control. It was discovered in the course of work on virus induced interferon [beta] (IFN[beta]) gene expression, and independently as interferon-stimulated gene factor 2 (ISGF-2), a transcription factor induced by interferon [alpha] (IFN[alpha]) (23 -25 ). TNF[alpha]-induced accumulation of IRF-1 mRNA and DNA binding activity has also been reported (26 ,27 ). It was ultimately found that IFN[gamma] is a more potent inducer of IRF-1 expression than is IFN[alpha], that IFN[alpha] and TNF[alpha] result in similar induction of IRF-1, and that virus infection is perhaps the least effective inducer of IRF-1 expression. (25 ,28 ; R. Pine, this report and unpublished observations). While several reports supported the hypothesis that IRF-1 regulates virus-induced transcription of the IFN[beta] gene (for example, 29 ,30 ), experiments done with HeLa cells suggested that IRF-1 plays at most only a small role in the production of IFN[beta] (25 ,31 ,32 ). Studies with mice homozygously deleted for IRF-1 subsequently corroborated this conclusion (33 ,34 ). However, IRF-1 does play an important role in resistance to both viral and bacterial infections (31 ,35 ,36 ), consistent with its role in cellular responses to IFNs and TNF[alpha]. Furthermore, IRF-1 can exert an antiproliferative effect on cells and can participate in some apoptotic pathways (37 ,38 ).
As the transcription factor product of a primary response gene, IRF-1 prolongs the expression of other primary response genes or activates secondary response genes. Activation of IRF-1 gene transcription by both IFN[gamma] and IFN[alpha] was found to be mediated by an IFN[gamma] activation site (GAS) in the IRF-1 promoter (28 ). As part of the overall response to IFN[gamma], IRF-1 is required for protein-synthesis-dependent transcriptional induction of the murine GBP gene (39 ) and for synergistic induction of the murine inducible nitric oxide synthetase (iNOS or NOS2) gene by IFN[gamma] plus lipopolysaccharide or TNF[alpha] (35 ,40 ). Maximal TNF[alpha]- induced expression of the VCAM-1 gene requires induced synthesis of IRF-1, which then acts as a secondary regulator (27 ).
Since IRF-1 induced by TNF[alpha] or IFN[gamma] functions in the molecular responses to each cytokine alone, and in the synergistic induction of MHC class I alleles and iNOS by TNF[alpha] plus IFN[gamma], it was of interest to determine if TNF[alpha] plus IFN[gamma] caused synergistic induction of IRF-1. This report shows that transcription of the IRF-1 gene and accumulation of IRF-1 DNA-binding activity are synergistically induced by the combination of TNF[alpha] and IFN[gamma]. A novel composite GAS/[kappa]B element alone can mediate this response. The synergistic induction of IRF-1 by TNF[alpha] plus IFN[gamma] may involve both NF[kappa]B and STAT-1 even though they do not seem to bind simultaneously to the GAS/[kappa]B element in vitro. The results presented here raise the possibility that the composite GAS/[kappa]B element in the IRF-1 promoter might mediate other interactions among pathways that utilize members of the NF[kappa]B family and those that act through STAT proteins. Furthermore, GAS/[kappa]B regulatory elements are likely to occur and function similarly in other genes.
MATERIALS AND METHODS
Reagents
Recombinant human IFN[gamma] and TNF[alpha] were from Amgen and Chiron, respectively. TNF[alpha] and IFN[gamma] were used at 5 ng/ml. All antisera were polyclonal, from rabbits. The anti-ISGF2 antiserum was raised against the human protein purified from HeLa cells, and immobilized on polyvinylidene difluouride membrane (25 ). An irrelevant immune serum was obtained after immunization with duck metallothionein (gift of P. C. Huang) by the same protocol. Anti-STAT-1 and anti-STAT-2 antisera (gift of Chris Schindler) have been previously described (41 ). Antibodies against NF[kappa]B family members and from normal rabbit serum (gift of Hsiou-Chi Liou) have been previously described (42 ). Nitrocellulose was from Schleicher and Schuell. Radioisotopes were from ICN. Poly (dIdC:dIdC) was from Pharmacia. Enzymes were from New England Biolabs or Boehringer Mannheim. All other chemicals were from Boehringer Mannheim, Sigma or Fisher Scientific, except as specifically indicated.
Plasmid constructs and oligonucleotide sequences
The IRF-1-luciferase exon fusion constructs, and the GAS/[kappa]B WT, GAS/[kappa]B 5M and GAS/[kappa]B 3M heterologous promoter constructs have been described before (28 ). The -199/-16, -199/-89 and -89/-16 plasmids contained the respective NarI fragments from the IRF-1 promoter in the heterologous promoter luciferase reporter. The CMV[beta]-GAl plasmid was from California Biotechnology, Inc.
The interferon-stimulated response element (ISRE) oligonucleotide (CTCGGGAAAGGGAAACCGAAACTGAAGCC) and its complement, synthesized with BamHI cohesive termini at the 5' end of each strand, spans from -117 to -89 of the ISG15 promoter (24 ,43 ). The ISRE homology is shown in bold. A non-specific oligonucleotide (CTCTCTGCAAGGGTCATCAGTAC) and its complement, synthesized with HindIII cohesive termini at the 5' end of each strand, includes the distal HNF4 site from the transthyretin promoter (44 ). The GAS/[kappa]B WT oligonucleotide (TACAACAGCCTGATTTCCCCGAAATGACGGC) and its complement, synthesized with HindIII cohesive termini at the 5' end of each strand, spans from -137 to -107 of the IRF-1 promoter. The GAS homology is shown in bold and the non-consensus NF[kappa]B site reverse complement is underlined. The GAS/[kappa]B 5M and GAS/[kappa]B 3M oligonucleotides had TT -> GG or AA -> CC mutations (top strand; -122 and -123, or -115 and -116, respectively). A shorter version of the IRF-1 GAS oligonucleotide (GATTTCCCCGAAAT) and its complement, synthesized with BamHI cohesive termini at the 5' end of each strand, spans from -126 to -113 of the IRF-1 promoter. The GAS homology is shown in bold.
Cell culture and transfection assays
HepG2 cells (ATCC HB 8065) were maintained as subconfluent monolayer cultures in Dulbecco-modified Eagle's medium (BioWhittaker) plus 10% defined supplemented calf serum (HyClone). Cells were transfected in suspension with a calcium phosphate protocol, essentially as described (45 ), except that only 8 µg of DNA (40 µg/ml) was used for ~106 cells. Luciferase reporter constructs that included sequences from the IRF-1 promoter were mixed with an expression vector that encoded [beta]-galactosidase as an internal standard for transfection efficiency. Cells transfected in a single tube were diluted with culture medium then subdivided into 35 mm wells for subsequent untreated control or cytokine treated samples. The monolayers were washed and refed with fresh culture medium ~24 h later, and then received no further treatment or were treated with TNF[alpha], IFN[gamma] or both together ~64 h after transfection for the next 4 h. Extracts were made as recommended by Promega. Reagent from Promega was used to assay extracts for luciferase with an OptoComp I luminometer from MGM Instruments. [beta]-Galactosidase was assayed according to standard procedures (45 ), except that reactions were performed in 96 well plates and optical density was measured with a plate reader at various times without stopping the reactions. Luciferase activity was normalized to [beta]-galactosidase activity, then fold induction was calculated. Each transfection was performed in triplicate, and at least two experiments were done with independent preparations of each plasmid.
Extract preparation and electrophoretic mobility shift assays
HepG2 monolayers at 60-90% confluence were untreated or treated with cytokines as indicated in the figure legends. All extracts were prepared at 0-4oC. Whole cell extracts were prepared essentially as previously described (31 ), except that the proportion of extraction buffer to the number of cells was reduced, so that 80, 250 or 500 µl was used for cells in 35, 60 or 100 mm plates, respectively. The extraction buffer is 0.5% NP-40, 0.3 M NaCl, 10 mM Na4P2O7, 5 mM NaF, 0.1 mM EDTA, 20 mM Na·HEPES, pH 7.9 at room temperature, 10% glycerol; plus 1 mM dithiothreitol, 100 µM Na3VO4, 0.4 mM phenylmethylsulfonyl fluoride, 3 µg aprotinin per ml, 1 µg leupeptin per ml and 2 µg pepstatin per ml, each freshly added before each use (buffer A + 0.3 M NaCl). For preparation of cytoplasmic and nuclear extracts, cell monolayers from 100 mm plates were scraped into phosphate buffered saline, recovered by centrifugation and lysed in 250 µl of buffer A + 0.1 M NaCl. Nuclei were recovered by centrifugation at 1000 g for 5 min, then resuspended in 50 µl of buffer A + 0.3 M NaCl for extraction. After a 30 min incubation, debris was removed by centrifugation at 12 000 g for 10 min and the supernatant was recovered as the nuclear extract. The supernatant from the cell lysate was clarified by centrifugation at 12 000 g for 10 min, and the supernatant was recovered as the cytoplasmic extract.
Approximately 10 000 d.p.m. of oligonucleotide labeled by pulse-chase fill-in of 5' cohesive ends (0.4-2 ng, depending on time since labeling) was used as probe for electrophoretic mobility shift assays (EMSAs) with ~10 µg of extract protein (typically 2-3 µl of extract). In addition to the contribution from the extract, binding reactions included 1* binding buffer (4% ficoll, 0.1 mM EGTA, 1 mM MgCl2, 0.5 mM dithiothreitol and 20 mM Na·HEPES, pH 7.9 at room temperature). The final volume was 12.5 µl. For the ISRE probe, 0.5 µg of pGEM-1 and 1 µg of poly (dIdC:dIdC) were included as non-specific DNA. For the GAS/[kappa]B probe, 0.75 µg of poly (dIdC:dIdC) was used as non-specific DNA. Reactions were incubated for 20-30 min at room temperature. If indicated, 2.5 µl of 1* binding buffer containing 0.5 µl of antiserum or antibody was added after the binding reaction was complete, and then incubation was continued for 40-60 min at 4oC. Samples were electrophoresed on 6% polyacrylamide gels run at 4oC with 20 mM Tris-borate (pH 8.3 at room temperature), 0.4 mM EDTA. Radioactivity in protein-DNA complexes was quantitated by two dimensional gas-ionization beta particle detection (Packard Instant Imager) and visualized by autoradiographic exposure.
Determination of transcription rates
Cells were untreated or treated with cytokines as indicated in the legend to Figure 2. Run-on assays were performed with isolated nuclei to determine relative rates of transcription as previously described (46 ). Radiolabeled RNA was recovered and hybridized to excess plasmid DNA fixed to nitrocellulose. IRF-1 transcription was measured with a cDNA probe (25 ). A [beta]-tubulin pseudogene (47 ) was used as a positive control and internal standard. pGem-1 (Promega) was used as a negative control probe. The results were quantitated directly with a Molecular Dynamics phosphorimager, and visualized by autoradiographic exposure.
RESULTS
IRF-1 expression is synergistically induced by TNF[alpha] plus IFN[gamma]
Since it was already known that transcription of the IRF-1 gene is activated by IFN[gamma], that IRF-1 expression is induced by TNF[alpha], and that TNF[alpha] and IFN[gamma] often act synergistically, the combined effect of TNF[alpha] and IFN[gamma] on induction of IRF-1 was examined (Fig. 1 ). Irrelevant immune antiserum or specific anti-IRF-1 antiserum was included in EMSAs that used the ISG15 ISRE as a probe for binding by IRF family members. TNF[alpha] plus IFN[gamma] led to a synergistic increase in the amount of IRF-1 DNA binding activity after 2 or 4 h of treatment (compare lanes 1-8 and 15-20 with lanes 9-14). Synergism is defined here as a quotient greater than one for the response to the combined treatment divided by the sum of the responses to the respective individual treatments (i.e., the whole is greater than the sum of the parts). These quotients are 1.5 or 1.6 for 2 or 4 h of treatment with TNF[alpha] plus IFN[gamma].
A GAS/[kappa]B promoter element mediates the synergistic transcriptional response to TNF[alpha] plus IFN[gamma]
Transient transfection was used to identify the regulatory elements in the IRF-1 promoter that mediated synergism of TNF[alpha] with IFN[gamma] (Fig. 3 ) Successive 5' deletions made in the context of the IRF-1 transcription start site and first exon fused to a luciferase reporter were examined, and it was found that sequences between -3.4 kb and -160 bp did not contribute to synergism. Constructs with IRF-1 upstream sequence fused to a minimal thymidine kinase promoter-luciferase reporter were then tested to determine the effects of additional 5' and 3' deletions. A significant response to TNF[alpha] was conferred by the -199/-89 fragment, though TNF[alpha] induction was primarily effected by regions downstream from -89. As expected from previous studies with HeLa and K562 cells (28 ,48 ), the -199/-89 fragment gave substantial response to IFN[gamma], which ranged from 50 to 100% of the level obtained with the -199/-16 fragment in this and other experiments. The -89/-16 fragment did not produce a response to IFN[gamma]. All induced responses were similarly elevated compared to those seen with the exon fusion constructs. Thus, there may be a negative regulatory element between -16 and +168 in the IRF-1 gene which affects both TNF[alpha] and IFN[gamma] induction. Alternatively, the interaction of the upstream elements of the IRF-1 promoter with the minimal thymidine kinase promoter may be more effective than the interaction with the native TATA-less IRF-1 basal promoter sequences. Of greatest interest for the present studies, the -199/-89 fragment did mediate a synergistic response to TNF[alpha] plus IFN[gamma], while the -89/-16 fragment did not.
Effect of treatment with TNF[alpha] plus IFN[gamma] on protein binding to the GAS/[kappa]B element
To determine which proteins might mediate the synergistic induction of IRF-1 in response to TNF[alpha] plus IFN[gamma], the GAS/[kappa]B oligonucleotide was used as a probe for EMSA with extracts from cells that were untreated or had been treated with TNF[alpha], IFN[gamma] or both (Fig. 4 ). Two complexes formed with extracts from TNF[alpha]-treated cells (Fig. 4 A, lane 4), and each complex contained both the p50 and p65 subunits of NF[kappa]B, as judged by the effect of antibodies specific for those proteins (Fig. 4 A, lanes 5 and 6). Additional protein(s) are likely to be present in the slower mobility complex, but remain to be identified. Each antibody also reacted essentially completely with those two complexes when extracts from cells treated with both TNF[alpha] and IFN[gamma] were assayed (Fig. 4 A, compare lane 7 to lanes 8 and 9). Antibodies against other human Rel family members, p52, c-Rel and RelB, did not react detectably with either complex regardless of whether cells had been treated with TNF[alpha] only or both TNF[alpha] and IFN[gamma] (data not shown). This result is partially obscured by the intensity of the major IFN[gamma]-induced complex and what may be a minor IFN[gamma]-induced complex that migrates very close to the lower TNF[alpha]-induced NF[kappa]B complex. However, extracts from cells treated with both cytokines or only with IFN[gamma] clearly produced the same pattern in the presence of anti-p50 or anti-p65 antibodies (compare lanes 8 and 9 to lanes 11 and 12). All of the complexes induced by TNF[alpha] plus IFN[gamma] were found to contain either STAT-1, or p50 and p65, by use of the anti-STAT-1 antiserum together with either the anti-p50 or anti-p65 antibodies (Fig. 4 A, lanes 13 and 14). Anti-STAT-1 antiserum alone had no effect on the NF[kappa]B complexes induced by treatment with TNF[alpha] alone (Fig. 4 B, lanes 1-4), or by treatment with TNF[alpha] plus IFN[gamma] (Fig. 4 B, lanes 5 and 6). With extracts from cells treated with TNF[alpha] plus IFN[gamma] (Fig. 4 B, lanes 5 and 6) or with IFN[gamma] alone (Fig. 4 B, lanes 7 and 8), the major induced complex reacted specifically and completely with the anti-STAT-1 antiserum. Together, these results indicate that none of the complexes observed upon assay of extracts from cells treated with both cytokines contained heteromeric factors composed of Rel and STAT subunits.
DISCUSSION
This report presents data showing that IRF-1 DNA-binding activity is synergistically induced by TNF[alpha] plus IFN[gamma], and establishing that transcriptional regulation is the mechanism that underlies this example of synergism between TNF[alpha] and IFN[gamma]. These results and the studies of MHC class I, ICAM-1 and IP-10 regulation (9 ,10 ,13 ) strongly support the idea that synergistic biological effects of TNF[alpha] and IFN[gamma] result from synergistic transcriptional activation of genes that are also regulated by each cytokine alone.
A composite GAS/[kappa]B promoter element mediates the synergistic induction of IRF-1 transcription. Both NF[kappa]B and STAT-1 from cells treated with both cytokines bind to the element, but the binding is mutually exclusive and independent. The details of the protein-protein and protein-DNA interactions that result in synergistic induction of the IRF-1 gene remain to be determined. The regulation of IRF-1 transcription may be related to the synergistic induction by TNF[alpha] and IFN[gamma] of the ICAM-1 or IP-10 genes (9 ,13 ), but is clearly distinct since those promoters include separate NF[kappa]B and STAT-1 binding sites that can be simultaneously occupied.
The function of IRF-1 in protein-synthesis dependent transcriptional activation of TNF[alpha]- or IFN-induced gene expression is most consistent with and has been predicted best from experiments that characterized induction of IRF-1 DNA-binding activity (25 ,27 ). It seems likely that the outcome of interactions between IRF-1 and other transcription factors including ICSBP, HMG I(Y), or ATF-2 (21 ,27 ,49 ) will be affected by this synergistic induction. This possibility should be considered in particular for these or other activities that are themselves modulated by either TNF[alpha] or IFN[gamma]. The data shown here lead to the conclusion that synergistic induction of IRF-1 constitutes a step in a regulatory network. As shown in Figure 6 , the regulation of IRF-1 gene expression and the function of newly synthesized IRF-1 as a transcription factor establish a link between synergistic primary and secondary responses. Synergistic induction of IRF-1 would serve as a preamplifier for the subsequent interaction of NF[kappa]B with IRF-1 in the synergistic induction of MHC class I genes as a protein-synthesis dependent response to TNF[alpha] plus IFN[gamma] (10 ). A similar network effect would pertain to synergistic induction of the iNOS gene (35 ,40 ). Thus, the synergistic induction of IRF-1 by TNF[alpha] plus IFN[gamma] can clearly contribute to the role of this factor in inflammation and immunity. The potential impact of synergistic induction of IRF-1 on apoptotic pathways is not as clear, since pathways that lead to activation of NF[kappa]B or apoptosis in response to TNF[alpha] are distinct (50 ,51 ).
ACKNOWLEDGEMENTS
The author would like to thank H.-S. Liou for providing antibodies against Rel family members, C. Schindler for providing recombinant STAT proteins and antibodies against STAT proteins; D. Thanos for providing recombinant NF[kappa]B proteins; K. Drlica, C. Schindler and P. Tolias for critically reading the manuscript; and A. Canova for expert technical assistance. The author also thanks P. Olson for providing TNF[alpha] and D. Vapnek for providing IFN[gamma]. This work was supported by grants from the Arthritis Foundation and the Lucille P. Markey Charitable Trust.
REFERENCES
1 Aggarwal,B.B. and Natarajan,K. (1996) Eur. Cytokine Net., 7, 93-124.
