VIROLOGY

191,

589-599

(1992)

Characterization interacting

of TH3, an Induction-Specific Protein with the Interferon p Promoter

LUCIE COHEN AND JOHN HISCOTT’ Lady Davis Institute for Medical Research, Sir Mortimer B. Davis-Jewish General Hospital, and Department and Immunology, McGill University, Montreal, Quebec H3T 1E2 Received June 6, 1992; accepted August

of Microbiology

17, 1992

We report the purification and characterization of a unique DNA-binding protein termed TH3 that interacts with the positive regulatory domain (PRD) I and PRDIII domains of the interferon (IFN) /3 promoter. In cells treated with poly rl:rC and cycloheximide, appearance of TH3 DNA-binding activity was inversely proportional to the disappearance of a constitutive complex THl and coincided temporally with induction of IFN-P gene transcription. The TH3 DNA-binding protein is a small 1CkDa polypeptide that appears to be derived from the THl complex; THl in turn is related to interferon regulatory factor (IRF) 2 by immunological cross-reactivity. The TH3 protein appeared to lack the epitope required for recognition by anti-IRF-2 antisera; however, a short microsequence obtained for TH3 overlapped a sequence from the IRF-2 protein. Although TH3 binds to multimers of the AAGTGA hexamer and to PRDI, the TH3 protein alone had a predominantly neutral phenotype on PRDI-dependent transcription in vitro and lacked the negative transcriptional effect attributed to IRF-2. These results raise the possibility that specific proteolysis of a negative regulatory protein involved in silencing the IFN-0 promoter may be an important event leading to transcriptional activation of the interferon gene. 0 1992 Academic Press, Inc.

more and Maniatis, 1990). DNA sequences that regulate IFN-/3 gene transcription are located immediately upstream of the intronless structural gene and consist of multiple overlapping positive and negative regulatory domains essential for virus-induced activation and/or silencing of the promoter (Maniatis et a/., 1991). Three positive regulatory domains (PRDs) contribute to transcriptional activation of IFN-P. One of these elements is the PRDII domain (-66 to -55) originally identified by deletion and mutagenesis analyses (Goodbourn and Maniatis, 1988); PRDII interacts with NF-&/re/ proteins (Cohen et a/,, 1991; Cohen and Hiscott, 1992; Hiscott et a/., 1989; Lenardo et al., 1989; Visvanathan and Goodbourn, 1989). Many studies have now established the NF-KB family of transcription factors as important transcriptional regulators of several immunoregulatoty genes (Baeuerle, 1991). Two other positive regulatory domains, PRDI (-77 to -64) and PRDIII (-94 to -78) participate in virus-mediated activation of the promoter in synergy with PRDII (Fan and Maniatis, 1989; Goodbourn and Maniatis, 1988; Leblanc et al., 1990); both of these elements contain permutations of a hexameric motif -5’AAGTGA ~‘(Fujita et al., 1987). When multiple copies of PRDI, PRDIII, or the hexamer sequence are placed adjacent to a reporter gene, a virus-inducible enhancer element is generated (Fan and Maniatis, 1989; Fujita et a/., 1987; Leblanc et al., 1990). In addition to the AAGTGA hexamerit motif, Type I IFN promoters contain other (GAAANN)like hexameric motifs that upon multimeri-

INTRODUCTION lnterferons (IFNs) are cytokines that possess antiviral, immunomodulatory, and growth regulatory properties (De Maeyer and De Maeyer-Guignard, 1988; Taniguchi, 1988; Vilcek, 1990). The Type I IFNs (IFN-a and IFN-/?) have also provided an important model to investigate the molecular mechanisms of transcriptional activation in response to environmental stimuli. Viruses, double-stranded natural and synthetic RNAs (ds RNA) and certain cytokines are known inducers of Type I IFN gene expression (Vilcek, 1990). Expression of the IFN-P gene is transient and controlled at both transcriptional and post-transcriptional levels (Whittemore and Maniatis, 1990); in virus-infected cells, IFN-P mRNA is detectable within 90 min to 3 hr, reaches a hours after infection, and maximum level at -6-12 then rapidly decreases to undetectable levels. Usually, induction by the synthetic dsRNA poly rl:rC results in accelerated mRNA production since no viral replication is required. Cycloheximide, a protein synthesis inhibitor, can markedly increase the poly rl:rC- or virus-induced levels of IFN-/3 mRNA. This phenomenon, known as superinduction, results from accelerated rate of transcription, decreased rate of postinduction repression, and stabilization of IFN-/I mRNA (Whitte-

’ To whom reprint requests should be addressed at Lady Davis Institute for Medical Research, 3755 Cote Ste. Catherine Road, Montreal, Quebec H3T 1 E2. 589

0042.6822/92

$5.00

CopyrIght 0 1992 by Academic Press, Inc All rights of reproduction I” any form reserved

590

COHEN AND HISCOTT

zation also function as virus-inducible elements (MacDonald et a/., 1990). PRDI, PRDIII, and multimers of AAGTGA constitute high-affinity binding sites for the interferon regulatory factors, IRF-1 and IRF-2 (Fujita et al., 1988; Harada et a/., 1989). IRF-1 is a phosphoprotein of apparent molecular weight 52-56 kDa; the cDNA of the corresponding gene encodes a protein of 329 aa (calculated molecular weight 37.3 kDa), suggesting that IRF-1 is posttranslationally modified to give rise to a 52- to 56-kDa protein (Miyamoto et al., 1988; Pine et a/., 1990). The IRF-2 protein shares 62% amino acid homology with IRF-1 in the amino terminal 154-aa region, whereas the C-terminal portions of the proteins diverge and are only 25% related (Harada et a/., 1989); the amino terminal domain represents the putative DNA-binding domain of these two proteins based upon the high homology in this region, while the carboxy domains lack known function. IRF-1 behaves as a transcriptional activator of Type I IFN genes, while IRF-2 functions as a repressor of transcription. In transfection experiments, expression of mouse and human IRF-1 genes was shown to increase transcription from reporter genes under control of IRFbinding sites (Harada et a/., 1990; Leblanc et a/., 1990; MacDonald et a/., 1990); activation of transcription was abrogated with concomittant expression of IRF-2 (Harada et al., 1990). Similarly, overexpressed IRF-1, but not IRF-2, induced the endogenous IFN-(U and IFN,f3genes in primate fibroblasts or in murine embryonal carcinoma cells. Furthermore, IRF-1 was able to synergize with activated NF-KB to stimulate the IFN-/3 transcription in human cells (Fujita et a/., 1989; Harada et a/., 1990; Leblanc et a/., 1990). Recently, Reis et a/. (1992) reported conclusive evidence for the essential role of IRF-1 in Type I IFN gene expression. In these studies, stable cell lines were selected that contained either sense or antisense IRF-1 cDNA; IRF-1 mRNA was constitutively expressed in either orientation. Antisense expression of IRF-1 abolished IFN-P induction by virus or dsRNA while cells overexpressing IRF-1 produced higher induced levels of IFN-P, compared to control cells. Despite possessing different roles in IFN gene transcription, both IRF proteins display similar DNA-binding specificities for the IFN-P promoter and the genes encoding these proteins are both induced by IFN-P inducers, resulting in de novo synthesis of the proteins following induction (Harada et al., 1989). The initial model of IRF-mediated regulation of IFN transcription proposed that in uninduced cells IRF-2 binding was predominant and caused silencing of the IFN-P promoter, while following viral induction qualitative and quantitative changes in IRF-1 displaced IRF-2, leading to gene activation. In support of this model it was re-

cently shown that a post-translational event(s), presumably a phosphorylation event, was required for activation of the IRF-1 protein (Watanabe et al., 1991). In addition to IRF-2, a second repressor of IFN-P promoter was recently cloned and characterized (Keller and Maniatis, 1991). This protein, called PRDI-BFl, also binds to the PRDI domain of the promoter. Despite similar DNA-binding specificities, the cDNAs encoding these two proteins are not related; PRDI-BFl cDNA encodes for a 88-kDa zinc finger protein, which appears to be involved in postinduction repression of the promoter (Keller and Maniatis, 1991). We previously described the characterization of another protein-termed TH or TH3 protein-interacting with the PRDI and PRDIII domains of IFN-p promoter (Cohen et al., 1991). Binding of TH3 protein was undetectable in uninduced Hela S3 cells, but was inducible in cells treated with poly rl:rC in presence of cycloheximide, indicating that the appearance of the protein did not result from de novo synthesis of TH3. This inducible pattern was distinct from IRF-1 and IRF-2, but was reminiscent of PRDI-BFi, an inducible complex interacting with the IFN-/3 promoter (Keller and Maniatis, 1988). In the present study, we further characterized the nature and the inducibility of TH3. Our results show that TH3 kinetics of induction coincide in time with IFN-P gene activation. TH3 appears to be a product of IRF-2 specifically generated in cells upon induction with poly rl:rC and cycloheximide. These data suggest that inactivation of IRF-2 by specific proteolysis could participate in induction of the IFN-fi promoter. MATERIALS Cell culture

AND METHODS

and conditions

of induction

Hela S3 cells were grown at 37” to a density of 5 X 1O5 cells/ml in RPMI medium containing 10% FCS, 25 ml\/l HEPES (pH 7.3) and supplemented with glutamine and antibiotics. Cells were pretreated with 250 IU/ml of recombinant interferon a2 (r-IFNa2). The inducers were added at the following concentrations: poly rl:rC, 25 pg/ml; cycloheximide, 50 pg; Sendai virus, 1000 HAU/ml, for O-l 0 hr as indicated in individual experiments. Whole cell extract and nuclear extract

preparations

After the various induction periods, cells were harvested and rinsed once with cold phosphate saline buffer; cell pellets were quickly frozen in liquid nitrogen and thawed in three packed cell volumes of whole cell extract buffer (20 mM HEPES (K) pH 7.9, 0.2 mM EDTA, 0.2 mM EGTA, 0.5 mn/l spermidine, 0.15 ml\/l spermine, 10% glycerol, 10 mlVl sodium molybdate,

IFN-@ INDUCTION-SPECIFIC

and 1 mM DlT) containing 0.5 mM PMSF and 1 pg/ml leupeptin, pepstatin, and aprotinin as protease inhibitors. Cell lysis and protein extraction was performed as described previously (Cohen et al., 1991). Briefly, cells were lysed by addition of high salt concentration in one packed cell volume of 2 M KCI and ceil debris removed by ultracentrifugation. For whole cell extracts, the supernatant was collected and diluted to 0.1 M KCI. When purification was performed, proteins in the supernatant were precipitated with ammonium sulfate. Nuclear extracts were prepared according to Shapiro el a/. (1988) and resuspended in nuclear dialysis buffer supplemented with protease inhibitors (0.5 mM PMSF and 0.1 pg/ml of leupeptin, aprotinin, and pepstatin). Protein concentration ranged from 1.5-3 mg/ml for whole cell extracts and from 1O-l 4 mg/ml for nuclear extracts. Protein purification

and sequencing

TH3 protein was purified from poly rl:rC/cycloheximide-induced Hela S3 cells as previously described (Cohen et a/., 1991). Briefly, after ammonium sulfate precipitation, protein extracts were fractionated by gel filtration on a Sephacryl 300 HR resin column (500 ml bed volume) equilibrated with buffer Z (25 mM HEPES (I