The transcriptional regulatory protein, YB-1, promotes single-stranded regions in the DRA promoter.

YB-1 is a member of a newly defined family of DNA- and RNA-binding proteins, the Y box factors. These proteins have been shown to affect gene expression at both the transcriptional and translational levels. Recently, we showed that YB-1 represses interferon-gamma-induced transcription of class II human major histocompatibility (MHC) genes (1). Studies in this report characterize the DNA binding properties of purified, recombinant YB-1 on the MHC class II DRA promoter. The generation of YB-1-specific antibodies further permitted an analysis of the DNA binding properties of endogenous YB-1. YB-1 specifically binds single-stranded templates of the DRA promoter with greater affinity than double-stranded templates. The single-stranded DNA binding sites of YB-1 were mapped to the X box, whereas the double-stranded binding sites were mapped to the Y box of the DRA promoter, by methylation interference analysis. Most significantly, YB-1 can induce or stabilize single-stranded regions in the X and Y elements of the DRA promoter, as revealed by mung bean nuclease analysis. In concert with the findings that YB-1 represses DRA transcription, this study of YB-1 binding properties suggests a model of repression in which YB-1 binding results in single-stranded regions within the promoter, thus preventing loading and/or function of other DRA-specific transactivating factors.

The Transcriptional Regulatory Protein, YB-l, Promotes Single-stranded Regions in the DBA Promoter* (Received for publication, October 19, 1994, and in revised form, December 14, 1994) Gene H. MacDonald:!:, Yoshie Itoh-Lindstrom §, and Jenny P.-Y. Ting YB-l is a member of a newly defined family of DNAand RNA-binding proteins, the Y box factors. These proteins have been shown to affect gene expression at both the transcriptional and translational levels. Recently, we showed that YB-l represses Interferon-v-fnduced transcription of class II human major histocompatibility (MHC) genes (1). Studies in this report characterize the DNA binding properties of purified, recombinant YB-l on the MHC class II DBA promoter. The generation of YB-l-specific antibodies further permitted an analysis of the DNA binding properties of endogenous YB-l. YB-l specifically binds single-stranded templates of the DBA promoter with greater affinity than double-stranded templates. The single-stranded DNA binding sites of YB-l were mapped to the X box, whereas the doublestranded binding sites were mapped to the Y box of the DBA promoter, by methylation interference analysis. Most significantly, YB-l can induce or stabilize single. stranded regions in the X and Y elements of the DBA promoter, as revealed by mung bean nuclease analysis. In concert with the findings that YB-l represses DBA transcription, this study ofYB-l binding properties suggests a model of repression in which YB-l binding results in single-stranded regions within the promoter, thus preventing loading and/or function of other DBAspecific transactivating factors.
YB-1 is a member of a recently defined family of DNAbinding proteins, the Y box factors, also known as the cold shock domain factors (2,3). These proteins represent a multigene family identified in a number of eukaryotic and prokaryotic organisms. Members of this family include human YB-l, dbpA, dbpB, NSEP-l, and BP-8; frog FRGYl, FRGY2, YB3, p56, and p54; rat EFlA; murine MUSY1, MSYl; avian EFIA, RSV-EF-l, chkYB-l; bovine yEFIA#l; and bacterial cspA and cspB (3)(4)(5)(6)(7). Y box proteins are highly conserved, with 97% amino acid homology between rat EFIA and human YB-l. These factors have been shown to regulate gene expression at both the transcriptional and translational levels and several have been suggested to have roles in DNA repair as well as DNA and RNA condensation. Two closely related Y box proteins, YB-l and EFIA, have been shown to regulate transcription. EFlA can activate transcription through the Rous sar-* This work was supported by a National Multiple Sclerosis Grant RG-M85 and National Institutes of Health Grants ROI CA-48185 and CA-37172 (to J. P.-Y. Ting).The costs of publication ofthis 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. coma virus long terminal repeat (8,9), while YB-l has been shown to activate transcription through the HIV, HTL VI, JCV promoters (10,37). Recently, we showed that YB-l represses transcription of human major histocompatibility (MHC)l class II genes (1). Several other Y box proteins (FRGY2, FRGYl, MSYl) were shown to repress translation and protein expression by sequestering mRNA in gametes (11)(12)(13) and somatic cells (14). Y box proteins have a broad specificity for nucleic acids, binding double-stranded DNA, depurinated doublestranded DNA, single-stranded DNA and RNA (3-6, 8, 12, 13, 15-18), Studies to date indicate that the Y box factors are likely to affect gene expression by various different mechanisms.
YB-l was originally cloned by screening a human B cell expression library using a double-stranded oligonucleotide probe spanning the human DRA X and Y elements (15). These authors found that YB-1 binding was strong to the Y box and weaker to the X box. Binding to the Y box was dependent on the inverted CCAAT sequence in the Y box. The X and Y elements are highly conserved in murine and human MHC class II promoters and are necessary for basal, as well as interferon-y (IFN-'Y)-induced transcription. We have since shown that YB-l represses MHC class II gene expression (1). Transfection of cells with a YB-l expression vector repressed endogenous, IFN-'Y·induced class II mRNA and protein expression, as well as IFN-'Y-induced class II DRA promoter-driven reporter gene expression.
Sequences within eukaryotic promoters have been described that are sensitive to reagents that cleave single-stranded DNA and have been implicated in transcriptional regulation, including the promoters of the c-myc, l3-globin, platelet-derived growth factor A, epidermal growth factor receptor, and decorin genes (19)(20)(21)(22)(23), In addition, eukaryotic single-stranded DNAbinding proteins have recently been described such as FSB, STR, MF3, p70, ERDP-l (20,21,24,25) and are implicated in transcriptional regulation. FSB specifically binds singlestranded FUSE, the far upstream element of the c-myc promoter, and transactivates FUSE-CAT reporter constructs (24). In addition, activation of the c-myc gene correlates with the induction of single-stranded sequence in FUSE in vivo.
In this study, the binding properties of YB-l on the MHC DRA promoter are characterized. We demonstrate that YB-l specifically binds single-stranded templates of the DRA promoter with much greater affinity than the double-stranded template. Through methylation interference analysis the contact points of YB-1 in the DRA promoter are identified. The possible mechanisms by which YB-1 represses DRA transcription are explored. Mung bean nuclease analysis reveals that YB-l can stabilize single-stranded regions within the X and Y elements of the DRA promoter. We propose a model of tran- ATCTTGT6Tcc8~1IrTTGCAA6AACCCTTccl1ll_ _cTcAAAATATT~III~6TAAT 1. Summary of probes used. The nucleotide sequence of the DRA promoter is shown with the Servanius (W), X (X1X2), and Y elements shaded. The position of the X, Y, W, X + Y, and WXY probes relative to the promoter is shown. scriptional repression in which YB-I binding results in singlestranded regions in the DRA promoter and as a consequence prevents the binding and/or function of the X and Y box transactivating factors.

MATERIALS AND METHODS
Recombinant Human YB-l-The EcoRI fragment ofthe human YB-l eDNA, originally cloned in the pSFFVneo expression vector (15), was re-cloned into the EcoRI site of the plasmid, pSG5, (Stratagene, La Jolla, CAl. A BamHI restriction enzyme site was introduced by sitedirected mutagenesis (26) 3 bases upstream of the start codon of YB-l to remove the 5'-untranslated sequence. All mutagenesis and cloning were confirmed by sequencing. The BamHI-EcoRI fragment was subcloned into the bacterial expression vector, pGEX2T (Pharmacia Biotech Inc.), to generate a glutathione S-transferase (GST)NB-l fusion protein. The fusion protein was expressed and purified by binding to glutathione-Sepharose beads (Pharmacia) as described (27). Recombinant YE-l (designated rYE-I) was generated by incubating the GSTI YB-l-beads with thrombin (Sigma) at 1 unit/l0 /LI of a 50% bead slurry in 50 mM Tris (pH 7.4), 150 mM NaCI, 2.5 mM CaCI 2 for 1 h at 27 -c. Antibodies-Polyclonal anti-YE-l antisera was generated by multiple immunizations of a New Zealand White rabbit with 100-200 /Lg of rYE-1. Its specificity was determined initially by enzyme-linked immunosorbent assay followed by Western blot analyses, using rYB-l as an antigen in both assays. IgG fractions of the antisera, as well as preimmune sera were purified over protein Nprotein G columns (Pierce) according to the manufacturer's protocol.
Western Blot-Western blots were performed as described (35). Membranes were developed by chemiluminescence following the ECL protocol (Amersham Corp.).
Whole Cell and Nuclear Extracts-The human B cell line Raji the cervical carcinoma line cell, HeLa, and the promyelomonocytic cell iine, D937, were obtained from the ATCC and maintained under recommended conditions. Nuclear extracts were prepared as described previously (28).
Whole cell extracts were prepared by lysing cells in 10 mMHepes (pH 8,0), 1 mM EDTA, 7 mM {3-mercaptoethanol, 50 mM KCI, 0.4% Nonidet P-40, 1 /LM pepstatin, 1 /LM leupeptin, 1 /LM E-64, and 100 /LM phenylmethylsulfonyl fluoride (Boehringer Mannheim). The Iysates were centrifuged for 30 min at 13,000 X g at 4 "C, and the supernatants were used in EMS reactions. Protein concentrations for whole cell and nuclear extracts were determined by the Bio-Rad protein assay (Bio-Rad), Electrophoretic Mobility Shift Assay~EMS assays were carried out as described previously." Probes used in the study are illustrated in Fig.  1. Single-stranded oligonucleotides were end-labeled by T4 polynucleotide kinase (New England Biolabs, Beverly, MA), annealed to the opposite strand, and either gel-purified or purified over Nensorb columns (DuPont NEN). The probe (1 X 10 5 cpm) was incubated with rYE-lor cell extract and 2 /Lg of poly(dI-dC) in 10 mM Tris (pH 7.6), 1 mM EDTA, 0.1% Triton X-I00, 5% glycerol, 80 mM NaCI, 4 mM MgCI 2 , and 10 mM dithiothreitol, at 4 "C for 20 min. In select experiments, the same sequence spanned by the X + Y oligonucleotide ( Fig. 1) was excised from the plasmid pDC.XY50. This excised fragment is designated X + Y D S r e ' pDC.XY50 was constructed by cloning the X + Y ?Iigonucleotide into the XbaI site of the plasmid pDC18. 3 X + Y D S r e was Isolated by a BamHI/HindIII digest and labeled by a fill-in reaction using Sequenase 2, DNA polymerase (D. S. Biochemical Corp.). It is of interest to note that while purified rYE-l bound DNA probes in EMS assays, whole cell bacteria lysate significantly inhibited DNA binding by rYB-1. This could be due to bacterial repressor proteins or competitive binding of rYE-l by bacterial DNA or RNA. For supershift experiments, antibody was preincubated with the EMS reaction mix for 30 min at 4 "C prior to the addition of probe. To determine the specificity of DNA binding, 100-fold molar excess of unlabeled homologous or heterologous DNA was added to the EMS reaction for 15 min at 4 "C prior to the addition of probe.
Sequencing and Footprinting-Methylation interference was performed as described previously (30) with minor modifications. Oligonucleotides spanning either the X or Y elements of the DRApromoter (Fig.  1) were end-labeled on one strand with T4 polynucleotide kinase, annealed to the opposite strand, methylated with dimethyl sulfate for 1.0-1.5 min at 27°C, and purified over Nensorb columns. To form single-stranded probes, the double-stranded, methylated probe was boiled for 10 min and transferred immediately to ice. Double-stranded or single-stranded probes were incubated with rYB-l, as described above for EMS, at the maximum concentrations of YB-l necessary to form a complex with the specific probe being used. Bound and free fractions of probe were separated by nondenaturing gel electrophoresis, recovered from the gel by electroelution, extracted with phenol/chloroform, and cleaved with piperidine. Samples were separated by electrophoresis in a 12% polyacrylamide urea gel.
Mung bean nuclease was used to detect single-stranded regions in double-stranded probes. For mung bean nuclease, an 187-base pair probe ( Fig. 1, WXYJ spanning the DRA promoter was generated by restriction enzyme digest of the plasmid, 5' 6.152 DRA-CAT described previously (31). The plasmid was digested with XbaI, end-labeled with T4 polynucleotide kinase, and digested with BstYI to generate a 187base pair, single end-labeled probe. This fragment was purified by polyacrylamide electrophoresis and electroelution. The probe (4 X 10 4 cpm) was incubated with rYB-l (1 /Lg) in the presence of 1 /Lg of poly(dI-dC) in the EMS buffer in a 1O-/LI reaction at 37 DC for 20 min. The volume of the reaction was expanded five times and 0.10 volume of both a 10 X mung bean nuclease buffer (New England Biolabs) and 10 mM ZnS0 4 were added. Mung bean nuclease reactions were carried out at saturating concentrations of mung bean nuclease activity, as determined by previous titrations of mung bean nuclease on the WXY probe. One unit of mung bean nuclease (New England Biolabs) was added to the reaction and incubated at 37 "C. Ten-/LI aliquots of the reaction were stopped at different time points by incubation for 20 min at 37°C in 240 ml of stop buffer (100 mM Tris (pH 8.0», 100 mM NaCI, 20 mM EDTA, 0.1% SDS, 100 /Lg/ml proteinase K, 200 /Lg/ml glycogen). The samples were then precipitated and separated on a 6% polyacrylamide urea, wedge gel. Sequencing reactions (32) of the WXY probe were run simultaneously with the mung bean nuclease reactions.

RESULTS
YB-I was originally cloned by its binding to a probe containing the DRA X (comprised of Xl and X2) and Y proximal promoter elements, of which all are required for MHC class II transcription (15). We have since shown that YB-I can repress both endogenous DR protein and mRNA levels, as well as DRA promoter-driven, reporter gene transcription (1). In order to examine the mechanism ofYB-I-mediated repression, we produced recombinant YB-I protein (designated rYB-I) and characterized its binding activity on the DRA promoter. We also used the rYB-I to generate YB-I-specific antibodies to identify endogenous YB-I.
Recombinant Human YB-l-rYB-I was generated as described under "Materials and Methods." Bacteria were transformed either with pGEX2T alone or pGEX2T.YB-I and induced with isopropyl-{3-D-thiogalactopyranoside to express GST or the GSTIYB-I fusion protein ( Fig. 2; lanes 2 and 3). GSTI YB-I was purified by binding to glutathione-Sepharose beads (lanes 4 and 6). Thrombin treatment of the GSTIYB-I-beads generated a predominant cleavage product of apparent molecular mass of 33-34 kDa (lane 5). The expected molecular mass of rYB-I is 35.6 kDa.

Pre-1M
Pre-1M Detection of Endogenous YB-1 in N uclear Extra cts-rYE-1 was used to gene rate YE-1-sp ecific an tise ra. Western blot a na lys is of whol e se ra demonstrates th e specificity of t he anti-YB-1 antisera (Fig. 3A ). Thi s antise ra detect ed th e rYB-1 protein migr ating at approxim ately 33-34 kD a . No cross-reactivit y was see n with GST or two ot he r recombinant DNA-binding pr ote ins, NF-YB a nd hXBP, kn own to bind to th e DRA promoter. Th e IgG-enriched a nti-YB recogni zed a band of a pproximately 45 kDa in nucl ear extracts of t wo human cell lin es, Raji and B eLa (Fig . 3B) . Thi s is consisten t with th e reported molecul ar mass of dbpB of 42 kDa in B eLa nucl ear extracts (33). At high er concentrations of polyacrylamide, this band resolved to a doubl et (da ta not shown ). An a dditiona l hi gh molecular weight band was consistently obse rve d (70 kD a ) in th ese extracts a nd may represen t a new memb er of th e Y box family of pr otein s. Th e differenc e betw een th e expecte d molecular ma ss and the observed molecular mass of YB-1 (36 a nd 45 kDa, respecti vely) ma y be du e to post-translational modifications. Tr eatment of t hese ext racts with calf intestinal a lka line phosph atase did not a ffect th e migration pattern of th ese bands (data not shown), indicating that th ese protein s are not lik ely to be phosphorylated .
YB -1 Preferent ially Binds S ingle-strande d DNA-For ease of discu ssion , eac h pr obe is referred to as DS (double-stranded ) or SS (sing le-stra nded); t he SS designation followed by a number 1 indicates th e se nse st ra nd, whereas a number 2 indicates the a ntise nse st rand. EMS a na lys is shows rYE-1 binding to DS and SS oligon ucleotide prob es spa nning th e X a nd Y ele me nts of th e DRA pr omoter (Fig. 4). rYE-1 bound with th e greatest a ffinity to X + YSS l (Fig. 4, lan es 2-7). Slower migrating compl exes wer e form ed at hi gh concen tra tion s ofrYE-l. Th ese resolved to a faster migr a ting species as th e YE-1 concen tration was tit rated down . Th ese slower migrating complexes ma y re present multimers of YE-1, as reported for high concentrations of FRGY1 and FR GY2 (13), or binding to multiple sites on th e pr obe. At 5 ng of rYB-1, significa nt compl ex form ation is still obse r ved on X + Y S S l ' GST, purified as a nega ti ve control, showed no DNA binding (da ta not shown).
In contras t to st rand 1, rYE-1 bound to X + Y S S 2 ' a nd to X + YDS with much lower affinities (Fig. 4, lan es 8 -19 ). To rule out the possibility th at YE-1 binding on th e doubl e-stranded prob e was du e to contaminating sing le-s t ra nded species, th e doubl e- st randed prob e was isolated directl y from a plasmid (Fig. 4, lan es 20 -24 ). Binding to this prob e wa s ext re mely weak, ind icating t hat YB-1 binds poorly to doubl e-st randed DNA containing t he X and Y eleme nts. Preferen tial YE-1 binding to sing lestranded DNA is consistent with th e ability of ot he r Y box pr otein s (dbpB, NSEP-1, BP-8, a nd cspB) to bind singlest ra nded DNA (5,16,17,34).
Unl ab eled hom ologous X + Y S S l DNA compete d the complex form ation , wh ereas an oligonucleotide spa nning strand 1 of a n unrela ted DNA, the W eleme nt of t he DRA promoter, did not. Similarly, oth er single-stranded, het erol ogous oligonucleotides , including strand 1 of t he DRA octa mer eleme nt, str a nd 2 of th e W ele me nt, a nd strand 1 of th e myb binding site of th e c-myc pr omoter did not comp ete for rYE -1 binding (data not shown).
EMS a na lys is usin g whole cell extracts from the U937 cell lin e incubated with t he X + Y S S l pr obe resulted in a complex that co-migr ated with rYE-1 (Fig. 5A, lan es 1 versus 4). In addition to t his band, a slower migrating complex was obse rve d. Both compl exes were specifically competed by homol ogous but not heterologous pr obe, as was rYE-1 (Fig. 5A, lan es 5 a nd 6). Th e a nti-YE-1 antise ra was used to iden tify th e endogenous YB-1 in U937 extracts . Addition of YB-1-sp ecific a ntibodies to t he rYB-1 EMS rea ction resulte d in a slower migr ating, supershifte d complex (Fig. 5B, lan es 3, 5, 7, an d 9). No such complex was observ ed eithe r with t he negative control a ntibodies (lanes FIG. 4. VB-I form s a s t r o n ger co mp lex on s in g le -st r a n d e d DNA w ith a great er affin ity t han doubles t r a n ded DNA. YB-l bin din g prop er ties were a nalyzed on sing le-stra nded and double-stran ded X + Y prob es by EMS . Increasi ng concentrations of rYB-I (500-5 ng) wer e incu bated with X + Y probes a nd se pa ra te d on nondenaturi ng gels. Prob es were: X + Y single st ra nd, strand I (X + Y s s /), X + Y sing le st ra nd, stra nd 2 (X + Y S S 2 ) , X + Y double strand, a nnea led oligonucleot ides (X + Y os ), and X + Y doub le st ra nd prob e isolat ed from pla smid (X + Y" S.r,.)' Put ative YB-I multim er s are marked by t he solid arrow a nd monom ers by the asterisk . Methy lation interfe re nce a na lysis iden tified si ngle-stranded YE-! binding sites on both strands of t he X box in th e X2 eleme nt, with t he strongest pr otection on X S S 1 (Fig. 7A, lan es  1-6, summarized in C). This st rand bias is consistent with th e greatest binding a ffinity of YB-! for X S S 1 observed by EMS (Fig. 6). Four time s more rYE-! was required to generate a rYB-l (n9)~_~_4 , 6, 8 , a nd 10 ) or with a ntibodies in the abse nce of rYB-! (lane 1). Th e su pershi fte d band in th e U937 ext ract is of sim ila r migr ation as rYE -! a nd is antibody-specific (compa re lan es 11, 13 , a nd 15 versus 12, 14, a nd 16 ). Its form ation was a lso depend ent on the conce ntration of antibody used .
When single-st ra nded pr obes spa nning eit her t he X box or the Y box were used , it was found th at rYB-! bound preferentiall y to X S S 1 a nd to YSS2 (Fig. 6). Titration of rYB-! on t hese pr obes shows t hat complex form ation was lost at 40 ng of rYB-! on YSS2' whe reas a strong complex form ati on was obse rve d at t his conce nt ration on X S S 1 ' This may reflect a st ronge r affi nity of YE-! for X S S 1 than YSS2' Both X S S 1 a nd YSS2 complexes were als o compete d by a n homologous cold comp eti tor, bu t not by a n unr elated probe, indica ting specificity of bind ing (da ta not show n). rYE-! a lso form ed a wea k compl ex with X os a nd Y os (Fig. 6, lanes 2 1-24 ).
YB-1 Si ngle-stra nded Bind ing Map s to X2 and Doubl est randed Bi ndi ng to Y-Met hy lation interferen ce analysis wa s used to ide ntify nucl eotid es in volved in YB-! binding to both doubl e-st ra nded a nd si ngle-stranded seque nce spa nni ng t he X box or th e Y box of th e DRA promoter. Piperidine clea vage of methyla ted single-stranded prob es generates an A/G seque nce la dd er as compa re d with only a G ladder followin g cleavage of doub le-stra nded pr obes (for exa mple see Fig. 7A , lan es 1 a nd 4  versus 7 a nd 10 ). the Y/CCAAT sequence prevented in vivo occupancy of adjacent X1X2 elements as determined by genomic footprinting. Taken together we propose that YE-l-induced single-strandedness around the CCAAT element may interfere with the binding of NF-Y/CEF to the CCAAT box and disrupt the assembly of this proximal promoter.
We have also identified YE-l binding sites by methylation interference analysis within the XIX2 element once this sequence has been made single-stranded. The mung bean nuclease data also demonstrate that YE-1 can induce or stabilize singlestranded sites within the X element on a double-stranded probe. The mung bean nuclease-sensitive sites directly flank the methylation interference contact sites. This suggests that YE·l may also bind to double-stranded sequences in or around the X element and induce or stabilize a single-stranded region. This structural change and/or occupancy of the X2 element may additionally prevent binding of X box transactivating proteins, e.g. hXBP, X2EP, and RFX (17,29,36,37).
Single-stranded DNA binding has been described for several Y box proteins (4,5,17,18). The binding of two other Y box proteins, NSEP-l and EP-8 to single-stranded DNA that is pyrimidine-rich has been interpreted to indicate binding to triple helix DNA or H-DNA in the human c-myc and {3-g10bin promoters, respectively. The reagents used here, as well as in these other studies, cannot distinguish between H-DNA and single-stranded DNA. There are several CT-rich stretches in the DRA promoter flanking the X box and the Y box; however, these are most likely too short to form H-DNA. In addition, the binding sites determined by methylation interference are not CT-rich. Alternatively, it is possible that the nuclease-sensitive sites in the c-myc promoter to which NSEP-l binds are in fact single-stranded DNA, as opposed to H-DNA.
In addition to our study, two studies have shown that another Y box protein, FRGY2, represses gene expression. The mechanism has been elegantly determined and appears to be mediated by binding to and preventing translation of mRNA (11). In these studies, mRNA levels were either maintained at a steady state or increased. YE-l, however, appears to be acting at the level of the DR promoter, as opposed to sequestering mRNA, in that DRE mRNA levels are reduced, and the inhibition is specific to the MHC class II promoter-CAT constructs, whereas the heat shock 70-and thymidine kinase-CAT constructs are not affected. Although we have shown here that methylation interference pattern on X S S 2 ' In contrast to the results with the X probes, no interference pattern was observed on the Y S S l nor Y S S 2 probes, despite rYB-l interaction with Y S S 2 seen by EMS (Fig. 6). One obvious explanation is that methylation of G residues did not interfere with protein binding.
The EMS results in Fig. 6 demonstrate that rYB-l binds Y D S , although with a lower affinity than YSS2 (lanes 23 and 24). Previously, double-stranded Y box binding sites have been reported for the Y box protein EFI (8,9). In those reports, methylation interference analyses were carried out using chick embryo nuclear extracts (or fractions of) that most likely contained other Y-binding proteins, such as NF-YA and NF-YB. Purified, recombinant YE-l is used here to address this concern and to define YB-l double-stranded Y box binding. Incubation ofrYB-l with Y D S resulted in a stretch of hypersensitive G and A residues on both strands of the probe (Fig. 7A, lanes 7-12). This observation suggests that rYE-l preferentially binds methylated G and A residues in the context of the CCAAT sequence.
YB-1 Promotes Single-stranded Regions in the DRA Promoter-Due to the high affinity ofYB-l for single-stranded DNA, we examined the possibility that YE-l could induce or stabilize single-stranded regions within the DRA promoter region using mung bean nuclease. Mung bean nuclease is a single-stranded DNA nuclease, with a much lower affinity for double-stranded DNA than the single-stranded nuclease, Sl. Mung bean nuclease analysis of the DRA promoter supports the results from the methylation interference analysis. Incubation of a double-stranded DRA probe with mung bean nuclease at 37°C for 10 min with or without rYE-l did not result in the appearance of a mung bean nuclease-sensitive region (Fig. 7B,  lanes 1 and 2). However, incubation of the probe with mung bean nuclease and rYE-l for longer time periods resulted in regions of enhanced sensitivity around the X and Y boxes relative to the control reactions in the absence of rYE-l (Fig.  7B, lanes 4, 6, and 8 versus 3,5, and 7). The presence ofYE-l in these reactions accelerated or enhanced the appearance of these sites. The nuclease-sensitive sites around the Y box coincide with the hypersensitive sites seen around the Y box by methylation interference (summarized in Fig. 7C), indicating that YE-l either enhances melting of this region or traps the DNA in a melted configuration. The mung bean nuclease-sensitive sites in the X box border the YB-l single-strand contact points determined by methylation interference (Fig. 7C). Although the probe used in the mung bean nuclease assay spans 187 base pairs, the YE-l enhanced single-stranded regions upstream of the TATTA box (marked by the asterisk in B) cluster primarily around the X and Y boxes, suggesting these elements may playa role in the configuration of the promoter. DISCUSSION Previous experiments have shown that YE-1 can suppress DRA promoter function in IFN-'Y-inducible cell lines. This manuscript explores the possible mechanism by which this occurs. The most important finding is that YE-l induces or stabilizes a single-stranded region in the Y box. This region coincides with the binding site of YB-l on a double-stranded probe, as determined by previous DNA binding assays (15). Eased on these findings we propose that YE-l binds to doublestranded sequences flanking the Y box which may be prone to single-strandedness and induces or stabilizes the single strand configuration.
The CCAAT box per se in the Y element binds the NF-Y/CEF family of proteins. Previously, we have shown that the in vivo occupancy of NF-Y is the most critical step in the assembly of the proteins binding the proximal elements (35). Mutations in YB-l binds single-stranded DNA, we cannot rule out that YB-l may also interact with RNA repressing translation.
An additional contribution of this study is the generation of recombinant YB-l-and YB-l-specific antibody. This antibody recognizes not only rYB-l, but has allowed us to identify endogenous YB-l in whole cell extracts as well as in nuclear extracts. This will allow us to further characterize YB-l under varying conditions.
We have shown that YB-l is a transcriptional repressor and propose that this repression is accomplished by promoting single-stranded regions in the Y and X elements. Significant multimer formation was observed with rYB-l, as has been reported for other Y box proteins (13). It is possible that one dimer or multimer of YB-l spans the X and Y elements, with one unit binding strand 1 in the X box while the other unit binds strand 2 in the Y box (Fig. 8). In this situation YB-l may not only preventing binding of the necessary transactivating factors to X and Y, but may also introduce a contortional constraint in this region of the promoter.