Characterization of a Silencer Regulatory Element in the Human Interferon-y Promoter*

Previous analysis of the human interferon-y (IFN-y promoter indicated that the region of DNA from -261 to -216 (designated here as BE (binding element)) possessed silencer activity, as deletion of this region caused an increase in promoter activity. Based on this finding, we have conducted a series of experiments to character- ize BE function and analyze the binding proteins which interact with this region. Transient transfection assays in the Jurkat T cell line revealed that the BE region possesses silencer activity, which is orientation-depend- ent when reinserted 6‘ to the IFN-y core promoter. How-ever, when the BE region was inserted in front of a het- erologous promoter (thymidine kinase (TK)), a mild enhancer activity was observed. Utilizing the electrophoretic mobility shift assay, we have identified two major DNA-protein complexes (designated as S and E com- plexes) which interact with this region. Mutational analysis indicated that the silencer activity observed with the IFN-y promoter correlated with the S complex and the enhancer activity correlated with the E com- plex. Preliminary characterization of these two DNA- protein complexes has demonstrated the presence of multiple proteins in each complex. We have found that the S protein complex has a recognition sequence similar to the nuclear

Interferon-y (IFN-y)' has diverse biological activities in the immune system. It is predominantly produced by activated T cells and large granular lymphocytes, and it is clear that its production in vivo is tightly controlled and restricted (1,2). Although the precise molecular mechanisms underlining the strict control of the IFN-y gene expression have not been fully characterized, a major role of regulated gene transcription in IFN-y production has been well established, and both positive and negative control over IFN-y transcription has been demonstrated (3). Our laboratories have been investigating the regulation of human IFN-y gene expression by analyzing both the gene structure and identifying the cis-acting functional * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
MD 21702-1201. Tel.: 301-846-5700, Fax: 301-846-1673 elements. Our earlier results, based on deletion analysis of the hIFN-y promoter, indicated that there may be at least two enhancer elements and two potential silencer elements located upstream region of the TATA box (3). One of the silencer elements was 36 base pairs in length and located between -251 to -215 (designated as BE, binding element, in this report), as demonstrated by transient transfection studies with human peripheral blood T lymphocytes (3).
Negative regulatory elements have been identified in the promoters of several cytokine genes, including IL-2 (41, and LD78 (13). This suggests that the negative motif plays an important role in cytokine gene regulation. In order to understand the mechanisms by which the IFN-y production is tightly controlled, we selected the BE region for investigation. In this report, we characterize the nuclear protein complexes in Jurkat cells that bind to the BE region specifically and analyze the function of the BE region in transient transfection experiments when linked to the IFN-y and a heterologous promoter.

MATERIALS AND METHODS
Olzgoonucleotides-Oligonucleotides were synthesized by the phosphoramidite method on a DNMRNA synthesizer (Applied Biosystems, model 392, Foster City, CAI. The synthesized oligonucleotides were treated at 50 "C overnight. Complimentary strands were denatured at 80 "C for 5 min and annealed at room temperature. The doublestranded probe wase labeled with [32PldCTP (Amersham Corp.) using the Klenow fragment (Life Technologies, Inc.). Sequences of two specific competitor oligonucleotides are shown as follows: AP2 oligonucleotide (GGTGTGGAAAGTCCCCAGGCTCCCCAGCAC) from the distal 72base pair repeat region of the SV40 gene (15) and YY1 oligonucleotide (ATGCCTTGCAAAATGGCGTTACTGCAG) from the upstream conserved region of the Moloney murine leukemia virus gene (16).
Cell Lines and ReugentsAurkat cells (CD4' human lymphoblast cell line) were cultured in complete medium (RPMI 1640 supplemented with 10% fetal calf serum, 2 m M glutamine, and 1000 unitdm1 penicillin-streptomycin). Purified human peripheral blood T cells (fresh T cells, CD3' > 95%) were cultured in the same medium. W1 antiserum was generously provided by Drs. K. Becker and K. Ozato (Laboratory of Developmental and Molecular Immunity, NICHHD, National Institues of Health, Bethesda, MD). Anti-AP2 antibody was purchased from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). lo8 cells were treated with 500 p l of lysis buffer (50 m~ KC1, 0.5% Nuclear Extract-Nuclear extracts were prepared as follows (17): I x Nonidet P-40, 25 m M Hepes, pH 7.8, 1 m M phenylmethylsulfonyl fluoride, 1 pg/ml leupeptin, 2 pg/ml aprotinin, 100 p~ DTT) on ice for 4 min. After l-min centrifugation at 14,000 rpm, the supernatant was saved as a cytoplasmic extract. The nuclei were washed once with the same volume buffer without Nonidet P-40, then were put into a 300-pl volume of extraction buffer (500 m M KCl, 10% glycerol with the same concentrations of Hepes, phenylmethylsulfonyl fluoride, leupeptin, aprotinin, and DTT as the lysis buffer) and rotated 20 min at 40 rpm. After centrifugation at 14,000 rpm for 5 min, the supernatant used as the nuclear protein extract was harvested, dialyzed against the same buffer with 50 m~ KCl, and stored at -70 "C. The protein concentration was determined by BCA (Pierce Chemical Co.).
Electrophoretic Mobility Shift Assay (EMSAI-The protein-DNA binding reaction (18) was conducted in a 20-1-11 reaction mixture with 0.5 pg of Poly(d1,dC) (Sigma), 2 pg of nuclear protein extract, 5 x lo4 cpm '"P-labeled oligonucleotide probe, and 10 pl of 2 x GS buffer (40 mM Tris, pH 7.4, 120 mM KCl, 8% Ficoll, 4 mM EDTA, 1 mM DTT) or 2 x Y buffer (24% glycerol, 24 mM Hepes, pH 7.9,8 mM Tris-HC1, pH 7.9,2 mM EDTA, 2 mM DTT). In some cases the indicated amount of double-stranded oligomer was added as a cold competitor. This mixture was incubated at room temperature for 10 min before and 30 min after addition of probe, then loaded on a 5% acrylamide gel (National Diagnostics, Atlanta, GA) that had been pre-run a t 210 V for 2 h with 0.5 x TBE buffer. The loaded gel was run at 210 V for 90 min, then dried and placed on Kodak X-Omat film (Eastman Kodak Co.). The film was developed after overnight exposure at -70 "C. W Cross-linking-The bromouridine-substituted BE oligonucleotide was synthesized on the same DNA/RNA synthesizer and labeled with [cu-"'PIdCTP using Klenow fragment (Life Technologies, Inc.). W-crosslinking was carried out according to a published procedure (18) with the following modification; W irradiation was performed on the gel after EMSA with 305-m UV light generated by the W Transilluminator 400 (Stratagene, La Jolla, CA) at a distance of 5 cm for 60 min. The specific bands were cut out and ground, then eluted with TES buffer (10 mM Tris-HC1, pH 8.0, 1 mM EDTA, 0.1% SDS) at 37 "C overnight with 250 rpm rotation. The DNA-protein complexes were precipitated from the elution buffer with 2 volumes of cold acetone, then resolved on a 10% SDS gel at 100 V for 12 h.
Immunoprecipitation-The W-cross-linked DNA-protein complex obtained from the last step was boiled for 5 min in 1 x reducing buffer (62.5 mM Tris, pH 6.8, 10% glycerin, 1% SDS, and 0.5%) 2-mercaptoethanol), then diluted with 500 pl of TNT buffer (20 mM Tris-HC1, pH 7.4, 200 mM NaCl, and 1% Triton X-100). 10 p1 of antibody and 20 pl of protein A-Sepharose CL-4B (100 mg of resin (Pharmacia Biotech Inc.) in 750 111 of fresh TNT buffer) were then added to the DNA-protein complex. After overnight rotation at 30 rpm, the resin was spun down and washed three times with 0.5 ml of TNT buffer, then boiled in 40 p1 of 1 x reducing buffer for 5 min and centrifuged again. The supernatant that contained the immunoprecipitated subunit protein-probe complex was loaded onto a 10% SDS-PAGE mini-gel and electrophoresed at 140 V for 1 h. The gel was dried and exposed to the x-ray film overnight.
Reporter Gene Construction-Two kinds of reporter gene vectors were used in this study: 1) plasmid 108, a p-galactosidase expression vector in which the p-galactosidase gene is controlled by the human IFN-y promoter fragment from -108 to +64 (kindly provided by Dr. Christopher Wilson, Department of Pediatrics and Immunology, University of Washington); 2) plasmid pBCTKp-CAT, a CAT expression vector in which the CAT gene is under control of the herpes simplex virus thymidine kinase (TK) promoter (19). To test silencer activity, we inserted the wild type BE or mutant BE at the Hind111 site of plasmid 108 (upstream of the IFN-y promoter). To test enhancer activity, BE was inserted at the HirzdIII site of the pBCTKp-CAT (upstream of the TK promoter). The cloned inserts were verified by DNA sequencing.
Dansfection Assay-The CD4' human lymphoblastoid Jnrkat T cells were grown in complete medium as described earlier. 8 x lo6 cells were transiently transfected by electroporation in 0.4 ml of complete medium with 20 pg of plasmid vector a t 260 V and 960 microfarads, using a Bio-Rad electroporation device, then diluted into 10 ml of complete medium and incubated at 37 "C for 3 h. Viable cells were isolated by centrifugation over LSM lymphocyte separation medium (Organon Tekinka, Durham, NC) at 2000 rpm for 20 min and cultured for an additional 45 h in the absence or presence of inducers, then harvested for CAT or p-galactosidase analysis. CAT activity was determined by the CAT-enzyme-linked immunosorbent assay (Boehringer Mannheim). p-Galactosidase assay was carried out according to the published method (20). CAT or p-galactosidase activities were normalized based on protein amount loaded at each point, and data from three to four individual experiments were analyzed by the Student's t test.

RESULTS
Activity of the BE Region on the IFN-y Promoter-In order to determine if BE could affect promoter function, we examined the activity of the BE region on the core human IFN-y promoter (Fig. k4). In plasmid 108, the human IFN-y promoter fragment (-108 to +64) expresses inducible activity in Jurkat cells in the presence of phorbol myristate acetate plus ionomycin. Constructs containing the wild type BE, in both proper and reverse A " 108 BEP 108 Bifunctional activities of the BE region. 8 x lofi Jurkat T cells were transfected with 20 pg of the reporter gene vector as described under "Materials and Methods." A, silencer activity of the BE region in the IFN--y promoter. Functional activity of the BE region was examined in the IFN-y promoter in the plasmid 108, 108 (parental vector), BEpl08 (vector with inserted BE in proper orientation), and BErl08 (vector with inserted BE in reverse orientation). After electroporation in the presence of the reporter plasmids, the viable cells were separated from dead cells after 3 h and cultured in the absence of stimulation for 21 h, then in the presence of 10 ng phorbol myristate acetate and 1 pg of ionornycidml for 24 h. Each of the bars represents the mean value ( X f S.D.) of 6-galactosidase activity from four individual experiments. B , enhancer activity of the BE region in the TK promoter. Functional activity of the BE region was examined in the TK promoter in the plasmid pBCTKp-CAT, TK (parental vector), BEpTK (vector inserted BE in proper orientation), and BErTK (vector inserted BE in reverse orientation). After electroporation in the presence of the reporter vector, the viable cells were separated from dead cells after 3 h and cultured in the absence of stimulation for 45 h. Each of the bars represents the mean value ( X f S.D.) of the CAT activity from three individual experiments. orientation, were selected and designated as BEpl08 (BE in proper orientation) and BErl08 (BE in reverse orientation). Results from the P-galactosidase assay showed that the BE fragment, inserted in front of the IFN-y promoter (BEpl08), down-regulated the promoter activity by about 10-fold, indicating that the BE region possesses a silencer activity in the IFN-y promoter. Interestingly, the same BE fragment in reverse orientation (BErl08) did not show any inhibitory effect on the IFN-7 promoter, implying that the silencer activity of the BE region was orientation-dependent.
Activity of the BE Region on a Heterologous Promoter-When the BE fragment was inserted at the 5' end of the TK promoter ( Fig. l B ) , the promoter activity was increased by about 5-fold as measured by CAT activity, and this effect was orientationindependent as both orientations of the BE region (BEpTK and BErTK) had the same effect on the promoter. This result indicated that the BE region also possessed a weak enhancer ac- Fin. 2. Nuclear DNA-binding protein complexes identified in the EMSA assay utilizing GS huffer. A , spt~ific nuelrar protein complrscs binding to t l l c 111.: rrrrion. i\ 'T'-lahrlrd I1E oligonuclrotidr was usrd as a prohr in E5ISA. LnnP I , no comprtitor; / o m 2, 100 XI rscrss unlnhrlrd liE oligonuclrotidr; Inn? 3. 100 LI rscrss SPl hinding oligonuclrotidr. TI. localization of protein binding sites by mutation analysis o f t h r BIS rrgion. Four RE mutants wcrr grnrrntrd hy changing nine nuclrotidrs in srquencr of t.hr HE region (thr mutated svquencrs arr shown in C ) . "P-I,ahrlrd wild typr RE and mutatrd RE wrre usrtl as prohcs in thr EMSA assay as shown at t h r lop o f each lanr.
C , srqurncrs of t h r wild typr and mutatrd RE oligonuclrotidrs. In t h r sequencrs of thr mutntrd HI<, unchanged nucleotidrs arc indicatrd hy t h r dnftrd lincs. tivity and this activity was not dependent on the orientation of the insert. Combined with the above data, our results indicate that the silencer activity of the RE region was specific for the IFN-y promoter.
EMSA of thr BE Rrgion-In order to characterize the nuclear proteins specifically interacting with RE, we performed EMSA using nuclear extract prepared from the human T cell line Jurkat (Fig. 2 4 ) . The EMSA results showed that several DNA protein complexes were formed with the RE probe (Inne I). Two complexes, U and S, showed specific hinding with RE because they could be competed by cold RE oligonucleotide (lone 2 ), hut not by a n u n r e k e d oligonucleotide, SP1 (Innr . ? ). This indicated that U and S are two candidate nuclear protein complexes which may play some role in the function of the RE region.
Mtrtationnl Ann1,vsis of t h t BE Region-Rased on the above results, we next wanted to determine the recognition sequences of the U and S complexes in the BE region. To approach this question, we generated four mutant RE oligonucleotides, designated as M1, M2, M3, and M4, (Fig. 2C) by mutating nine nucleotides a t a time from the 5' to the 3' end. The :'2P-laheled mutant oligonucleotides were utilized a s probes in EMSA (Fig.  2B). The results indicated that M2 and M3 did not form the S complex (lanes . ? and 4 ), but still formed the U complex. There was no change in M1 and M4 hinding capacity for either complex (lancs 2 and 5). These data indicated that the sequences mutated in M2 and M3 are responsihle for the specific interaction between RE and the S complex. In contrast, the U protein binding site could not be determined by this panel of mutant RE oligonucleotides and was not further analyzed.
The S Complex Correlntes with the Silrnrrr Artir~it~-To address whether the S complex correlates with the cis-acting activities of the RE region, we examined thr function of t h r RE mutants M2 and M.3 by inserting them in front of the IF'S-y promoter or TK promoter (Fig. 3 ) . Constructs with insrrts in the proper orientation were selected and designated as M2plO8 and M3p108 in the IFN-y promoter and M2pTK and M3pTK in the TK promoter. The function of thrse mutant IIE rrgions werr examined together with the wild type RE in transirnt transfection of ,Jurkat cells. Results of t h r I3-galactosidnsr assay showed that M2 (M2plOR) and M 3 rM3plOR) no longrr repressed the IFN-y promoter activity (A ). Combinrd with the mutation analysis data from the EMSA assay, thew data indicated that the S complex correlates with the silencer activity.
Idrntificntion of n DNA-Protrin Complcx N't1ir.h Corrrlntrs with thr Enhnnwr Actirlity-As it was important to determine if the DNA protein complex that was associatrd with silcncrr activity also could account for the mild rnhancrr activity observed on the TK promoter. we conducted a srries of cxpcriments utilizing the mutant oligonucleotides descrihrd ahovr. Although t h r M2 and M3 constructs did not show any difference in function with respect to thr IFN-y promotrr and showed no difference in EMSA, different results were obtainrd with respect to the TK promoter (Fig. 3R 1. In t h r TK promoter. the M2 mutant (M2pTK) expressed the same enhancer activity as the wild type BE (REpTK), but the M3 mutant (M3pTK) failed to exhibit any enhancer activity. This result indicated that there might be another complex(s) which was not revealed by the EMSA analysis conducted above and which should hind to the M2, but not the M.7, mutant oligonucleotide. As in vitro DNA-protein interaction is affected hy the hinding buffer, we tried several binding huffrrs with different pH and salt concentrations in order to determine whether we could identifv any new specific DNA-protein complex(s) which may he responsihle for the enhancer activity of the RE region. In Y huffer, which was distinct from the previous buffer with respect to pH and salt concentration, we found a specific DNA-protein complex designated "E" which was competed by cold BE oligonucleotide in a dose-dependent pattern (Fig. 4A. hut was not competed by an unrelated oligonucleotide SPl (A, lonm .5-7). A faster migrating DNA-protein complex, designated L in this figure, was also detected but it could not be reproducihly observed so it was not further investigated.
To determine the DNA binding site of the E complex, we used the same panel of :'"P-Iabeled RE mutant oligonucleotides, M1, M2. M.7, and M4, to probe the Jurkat cell nuclear extracts in the Y huffer ( I l l . The results are shown in Fig. 4R. With the exception of M3, all the other mutant RE oligonucleotides were able to form the E complex, indicating that the DNA sequences corresponding to the M.7 region were involved in the E complrx formation. Combined with the functional data from the TK promoter, these results indicated that the E complex correlated with the enhancer activity of the RE region. Chorocterizotion of the S Complm-"The mutational study suggested that the hinding sequences of the S complex were located in the M2N.7 region of the RE oligonucleotide. A sequence homology search for the known nuclear factors led our attention to the AP2 (15, 21) and the W1 (14, 22) nuclear factors. As shown in Fig. 5 , A and 13  ments utilizingAP2 antihodies, and cotransfection studies with an AP2 expression vector have all failed to demonstrate any interaction of AP2 wit.h the RE refion (data not shown), indicating that subunit a is an AP2-like nuclear protein with respect to DNA binding sequence and molecular mass.
Charactc.ritntion ofthr l? Compl(~.r-As demonstrated ahove, the binding site of the E complex was located in the M.7 region of the BE oligonucleotide. Sequence homology analysis (Fig.  5 R ) revealed t.hat the M.7 region shared a high homology with the binding site of the W1 nuclear factor. The YY1 nuclear factor has heen reported to be a multifunctional nuclear factor (14, 16,[23][24][25], and we wanted to determine i f W l might he a component of the E complex. In a n oligonucleotide competition assay (Fig. 6A ), the E complex was competed specifically hy the W1 oligonucleotidc (Ian(..? ), but not competed by an irrelevant oligonucleotide SP1 (lnnc 4 ). As W1 binding has heen shown to be affected hy salt Concentration ( 2 6 ) . we wanted to know whether salt concentration had any effect on the E complex. A s shown in R, the E complex was affected dramatically by salt concentration of the binding buffer. The E complex was reduced in the presence of 30 mal KC1 (lane 2 ) and disappeared completely with 60 mbr KC1 (lanc .?). This result was consistent with the binding characteristics of the W1 factor as reported by Meier and Groner (26). To further prove that W1 was in the E complex. we carried out supershift analysis with anti-W1 antiserum ( C ) and demonstrated that the E complex was competed by the anti-W1 antiserum (Lon(. 2 ), hut not hy the anti-p50 (a suhunit of NFKR) antiserum, a control antiserum (lnnc. 3 ) . We next performed the W cross-linking analysis of the E complex followed by immunoprecipitation (Fig. 7). The UVcross-linking analysis showed that the E complex consisted of three proteins ( A ) with approximate molecular masses of 96, 65, and 48 kDa (after deduction of the probe molecular mass).
To determine whether the 65-kDa component of the E complex was YY1 (W1 molecular mass is 65 kDa), we performed immunoprecipitation of the UV-cross-linked products using the anti-W1 antiserum. As shown in R, immunoprecipitation with the anti-W1 alltiserum yielded a protein band of molecular mass 65 kDa ( l a w 2 ), whereas the unrelated antiserum did not immunoprecipitate any protein from the UV-cross-linked E complex (Inn(> I ). Rased on these results, we conclude that the E complex consists of at least three suhunit proteins, one of which is the W1 factor. Modulation o f t h c , S a n d E C'ompl~~xc~s-ln ortlrr t o tlvtcv-rninc. if the appearance of the S and E complrxrs corrc.latrs with thc IFN-y transcriptional activation. we investigatrtl thr mrdulation of the complexes after stimulation of human prriphwal blood T cells. First, when we comparrd thv Irv(*Is of S a n d F: complexes in nuclear extracts from frrsh T cc.lls and .Jurkat cells hy EMSA, comparahle levels \vrrv n h s e n w l 1,rtu.c.t-n thv two cells (data not shown ). We nrxt analyzrd thv two complrws in the fresh T cells following comhinrd stimul:ltion for c-ithvr 4 or 24 h with phorhol myristnte acrtatr plus ionomycin or phytohemagglutinin (Fig. 81, which arc known intluccm of IPN-y gene expression in the T cells. The ElISA rrsults int1ir:ltrtl that the 4-h treatment did not change thv Irvrls of r i t hrr th(* 1. : or S complexes ( A ), hut after 24-h trratmc>nt. :I tlis;~ppr:~rancc~ of thv E complex but no significant change of the S romplrs tvns observed rB ). These rcwrlts are consistvnt with th(. fact that IFN-y transcription dors not continuc. i1ftc.r 2/1-11 stirnul:ltion (27) and raises the possihi1it.v that thr S romplrx may supprc-ss IFN-y transcription a t this time. Howrvrr. furthcbr stut1ic.s \vi11 Regulation of IFNy Gene 7Fanscription 257.33

A B 4 Hour 24 Hour
.-nr s-8g' am Fw. 8. Modulation of the S and E complexes in human peripheral blood T cells stimulnted for different times. A , 4-h stimulatinn. T crlls wrrr trratcd for 4 h with the comhined stimulation as indicated at the top of Innrss 2 and 3 . The nuclear extracts from thr treated T cells wrrr assayrd utilizing t h r radiolabrlrd R E olignnuclrntidr in t h r Y hufTrr and GS hufTer. R , 24-h stimulation. Thr nuclear extracts from 24-h trmtrd T cells wrre analyzrd in t h r EMSA assay undrr thr sarnc. conditions as in A. he required to verify the importance of the S complex in the regulation of IFN-y transcription.

DlSCUSSlON
Studies on the regulation of gene transcription often focus on mechanisms of transcriptional activation (28). However, transcriptional repression is also an important factor in the re-lation of many genes, including the lymphokine genes ( 3 , 26). Based on deletion analysis of the IFN-y promoter region, Chrivia rt nl. ( 3 ) developed a model of cytokine gene transcriptional regulation in which enhancer activity may be modified by a silencer element. In this study, we have characterized one of the silencer elements in the IFN-y promoter which is located between nucleotides -251 to -215. We found that this region has silencer activity depending on the orientation and specific for the IFNy promoter, and we have been ahle to correlate the presence of a specific DNA-protein complex with the silencer activity. However, when linked to a heterologous promoter, this region had a mild enhancer activity that was associated with a DNA-binding protein complex which was distinct from the complex associated with silencer activity.
Utilizing different binding buffers, we demonstrated that two different complexes could specifically interact with the BE region. In our studies, utilizing a binding buffer providing a final salt concentration of 60 mM, we identified two specific DNA-protein complexes, U and S. Through mutational analysis of the RE region, we determined the DNA sequences necessary for the S complex formation, hut we were unable to locate the U complex binding site. A sequence homology search indicated that S complex might share a binding sequence with a known nuclear factor AP2 (21, 29). The AP2 nuclear factor is an enhancer-binding protein which can interact with several gene promoters, including human growth hormone. c-myc, H-2K".
SV40, and human metallothionein 11, (15. 21, 29). Using a Jurkat T cell nuclear extract, we were ahle to show the specific competition of the S complrx hy a n AP2-sprcific oli~onuclrotidr in EMSA, suggesting that AP2 may hc. part of the S complrx. However, using a commerciall.v nvnilahlr anti-AP2 nntihody. wr failed to compete or supershift the S complrx in EMSA and also failed to immunoprecipitntr any APZ protrin aftrr L3' crosslinking analysis using the samr antisrrum I d:lt:l not shown I. In addition, cotransfection with an AP2 rxprrssion vrctor failrd to alter silencer activity of the BE rrfinn (data nnt shmvn 1. Thrsr results suggested that the S complex might havr thc same hinding specificity as the AP2 factor, hut suggrsts that AI'% is not part of the S complrx. This conclusion is suppnrtcvl hy :I recent report characterizing the TNF-tr promotrr silrnccr clrment (12). In the report. Fong r t nl. 1121 idcntifird A novc.1 silencer hinding protein that is rrsponsihlr for thr supprrssivr activity of this silencrr element. As t h r factor also $vas similar to the AP2 nuclear factor with rrspect to thr hnmolnk? of DNA hinding sequence, they concluded that thr rrprrssor factor is an AP2 like nuclear protein, hut thry fnilrd to drmonstratr that its binding could he competed hv an AP2 hinding oligonuclrotidr.
The E protein complex, associatrd with thr wrak rnhancrr activity ohserved with a hetrrologous promoter, also had sprcific hinding activity to the RE refion, hut thv huffrr rrquirements were different than those usrd for :Innlysi.q of thr S complex. With the previous hinding huffrr. tvr wrrr unahlr t o detect the E complex, hrcausr the salt in thr prrvious huffcr seriously affected its binding acti\rity. M'hcn ~v e s\vitchvd t o a salt free hinding huffer, "Y huffrr." thc spr>cific binding of thv E complex to the BE renon w a s clrarly demonstrated. Intwrstingly, in the Y huffer, t h r S complrx \vas drastically rrdurrrl. indicating the requircmc.nt of salt for S cnmplcx formation and/or hinding.
Several lines of evidrnce presented hrrc strongly support thr. possihle involvement of the YY1 factor in intrracting with thr BE region. YY1 (also callrd CFI, &, NF-El, or ITClU31'~ is a 65-kDa DNA-hinding protein. h r l o n~n g to t h r GLI-Kri~ppel family. which is expressed in most cells anti highly consrned hrtwecn human and mouse (14, 21. 25). YY1 is a dual functional factor. as i t acts as a repressor in some gcnrs, includinC the aclrnoassociated virus P5 promoter (14). the immunoClnhulin 1 IC) k 3 enhancer (24, 30), the long terminal repeat of hlnlonry murinr leukemia virus (I6.23), the [3-casein grne promoter r26,31), and the rr-actin promoter (32. 3 8 ) . YY1 also functions as a activator in other genes, including thr c-mvc promoter ~.30,88. 34 I. t h r IgFI intronic enhancer (24,30), and the mousr promotrr of rihosomnl proteins L30 and L.32 (25). A recent report demonstratrd that. the YY1 factor can initiate transcription in comhination with TFllR and RNA polymerase ( 3 5 ) . However. t h r Y Y 1 factor has not yet heen reported to he involved in cytokinr genr rxpression. The first indication of a possihle involvemrnt of Y S 1 in t h r complex which hinds to the BE refinn came from the fact that. thr hinding sequence of the E complex had a high srqurncc. homology with theYY1 factorhindingsitr (Fig. 511 I. Whrn wr usrrl the YY1 hinding site as a competitor in EXISA. the E complrx \vas competed hy this oligonucleotide (Fig. 6A I, and this complex was also competed hy the anti-YY1 antisrrum. hut not hy t h r irrclevant antiserum (Fig. 6CI. In addition, thr rcsults from I?' cross-linking indicated that thr E complrx consistcd nf thrrc proteins, one of which had molecular mass of 65 kna, similar to YY1 (Fig. 7A ). To further identify the 65-kDn prntrin :IS \Tl, w r successfully immunoprecipitated the YY1 factor from I?'-cross linked E complex hy anti-YY1 antisrrum (Fig. 711). Takrn together, our data strongly support thr fact that thc Y T I protrin is part of E complex. Our results are supported hy t hr rrport of Meier and Groner (26), in which thry drmonstratrrl that YY1 factor participates in repression of t h r \ k a s c i n g r n r promoter and hinding of the YY1 factor to the promotrr 1)NA is down-regulated with an increasing salt concentration in the assay system. Their results suggested that binding of the Y Y 1 factor to DNA may be enhanced by decreasing ionic strength in the binding buffer system.
Results from the functional studies of the BE region with two different promoters indicate that the BE region possesses silencer function specific for the IFN-r promoter. In the IFN--y promoter, BE exerted a strong silencer activity as only one copy of BE could completely inhibit the inducible IFN-r promoter activity. This silencer activity is strictly dependent on orientation of the BE insert as a n inverted BE region did not exhibit any silencer activity. This feature is in agreement with data obtained on other silencer elements (36). As the mutant forms of BE, M2 and M3, which no longer formed the S complex, did not express silencer activity, we postulate that the silencer activity of BE in the IFN-y promoter is mediated by the S complex. Furthermore, our data suggests that in the S complex, an AP2-like protein may interact with an as yet unidentified 40-kDa protein and mediate the silencer activity.
In contrast to the silencer activity in the IFN--y promoter, BE showed modest enhancer activity when placed in front of the TK promoter. Our data from mutation analysis suggest that the enhancer activity of the BE region was mediated by the E protein complex and that the Y Y 1 factor is part of that complex.
It is not clear what functional role, if any, the E complex plays in the IFN-7 promoter, as we did not observe enhancer activity in the IFN-r promoter with the M2 mutant, which no longer binds to the S complex, but still retains capacity for binding to the E complex. However, although we failed to obtain direct evidence to support a positive role for the E complex in the IFN-7 promoter, we still cannot exclude the possibility that this complex affects IFN--y transcription. It has been well established that the Y Y 1 factor expresses different activity in distinct promoters (32, 33). In the a-actin promoter, the YY1 factor functions as a repressor on interacting with a binding site which is overlapping with a binding site of the serum response factor (32). In that case, although Y Y 1 does not have active suppressive activity on binding to the a-actin promoter, it exerts a suppressive effect through prevention of physical binding of serum response factor to DNA. Since the S complex and the E complex binding sites are overlapping in the BE region, it is possible that these two complexes compete each other at their binding sites. Thus the E complex may have no active influence on the IFN-y promoter when bound to the BE region, but it may function through physical blockage of the S complex binding to the BE region, relieving the active suppressive effect of the S complex on the IFN-r promoter. This hypothesis could be used to explain why we did not detect any positive activity of the E complex with the M2 mutant in the IFN-y promoter. The E complex could bind to the M2 oligonucleotide, but it has no activator effect in the IFN--y promoter because of the promoter dependence of the Y Y 1 factor. More effort, utilizing techniques such as in vivo footprinting, will be required to examine this hypothesis.