fos/jun and octamer-binding protein interact with a common site in a negative element of the human c-myc gene

A negative element has previously been localized to a 57-base pair segment approximately 300 base pairs upstream of the human c-myc promoter P1. Within this element, a 26-base pair region was protected in vitro from DNase I digestion with a HeLa cell nuclear factor(s). Two specific DNA-protein complexes were identified in gel retardation assays using HeLa cell nuclear extracts and an oligonucleotide probe spanning the footprinted region. Exonuclease and chemical foot- print analyses suggested that the binding sites for both complexes are almost entirely overlapping. One of the complexes was eliminated by oligonucleotide competitors possessing known AP-1 binding sites. This same complex reacted strongly with anti-fos immunoglobulin suggesting a role for c-fos in governing c-myc expression. Precipitation of fos protein bound to c-myc DNA that was immobilized on beads confirmed the involvement of c-fos in a specific complex with the c-myc upstream sequence. In contrast, the other complex seen by the c-myc probe could not be competitively inhibited by AP-1 binding sites and was not affected by anti-fos antibody. Instead, this complex was effi- ciently eliminated by unlabeled oligonucleotides containing the octamer DNA motif found in immunoglob- ulin gene promoters. Purified octamer-binding pro- teins formed stable complexes with the 26-base pair c-myc sequences. These results demonstrate that degen- eracy in the consensus recognition sequences of these distinct

A negative element has previously been localized to a 57-base pair segment approximately 300 base pairs upstream of the human c-myc promoter P1. Within this element, a 26-base pair region was protected in vitro from DNase I digestion with a HeLa cell nuclear factor(s). Two specific DNA-protein complexes were identified in gel retardation assays using HeLa cell nuclear extracts and an oligonucleotide probe spanning the footprinted region. Exonuclease and chemical footprint analyses suggested that the binding sites for both complexes are almost entirely overlapping. One of the complexes was eliminated by oligonucleotide competitors possessing known AP-1 binding sites. This same complex reacted strongly with anti-fos immunoglobulin suggesting a role for c-fos in governing c-myc expression. Precipitation of fos protein bound to c-myc DNA that was immobilized on beads confirmed the involvement of c-fos in a specific complex with the cmyc upstream sequence. In contrast, the other complex seen by the c-myc probe could not be competitively inhibited by AP-1 binding sites and was not affected by anti-fos antibody. Instead, this complex was efficiently eliminated by unlabeled oligonucleotides containing the octamer DNA motif found in immunoglobulin gene promoters.
Purified octamer-binding proteins formed stable complexes with the 26-base pair cmyc sequences. These results demonstrate that degeneracy in the consensus recognition sequences of these distinct factors allows each of them to bind the c-myc negative element. The interaction of known transcriptional activators with a negative element suggests that the same factors can mediate both transcriptional activation and repression.
Nuclear proto-oncogene products play critical roles in cell growth and differentiation, mediate signal transduction in the nucleus, and participate in the regulatory processes of transcription and DNA replication (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11). The proto-oncogene cmyc has been studied extensively. Expression of the c-myc gene is closely associated with the growth state of cells; the deregulation of c-myc gene expression by various mechanisms, including chromosomal translocation, retroviral insertion, gene amplification, and point mutation, contributes to the immortalization or transformation of normal cells (12)(13)(14)(15). The expression of c-myc gene is regulated by several mechanisms, including transcriptional initiation, transcriptional * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
7l To whom correspondence should be addressed NCI, NIH, Bldg. 10, Rm. 2N105, Bethesda, MD 20892. elongation, mRNA stability, and translation (16)(17)(18)(19). The human c-myc gene is composed of three exons: the first exon is a noncoding leader sequence, and the second and third exons encode the protein product of c-myc. Transcriptional regulation of c-myc is responsive to many physiological signals, such as growth factors, and to biological agents, some of which trigger the differentiation of certain cell types (3,5). Transcription of c-myc starts from at least three initiation sites, with distinct promoters (PO, P1, and P2). P1 and P2 are the major promoters located upstream of exon I and within exon I, respectively. PO is a minor promoter, located 700 base pairs (bp)' upstream of P I , and accounts for less than 10% of the total c-myc transcription (20). The transcription of c-myc from P1 and P2 is regulated by a composite of positive and negative elements located both upstream and downstream of these promoters (17,(21)(22)(23)(24)(25). A negative transcriptional element, which acts on both P1 and P2, has been mapped between -350 to -294 bp relative to the transcriptional starting site, P1. DNase I footprint and exonuclease assays have identified a nuclear factor binding to this element (17).
In this report, we show that two biochemically distinct factors bind specifically to the 26-bp DNase I footprinted region of the c-myc negative element. One of the factors was identified as the fos/jun (AP-1) complex (40)(41)(42)(43)(44)(45)(46), and the other as the ubiquitous octamer-binding protein (27,28,47,62,63). Interestingly, the two complexes bind highly overlapping sequences. The presence of interdigitating binding sites for distinct factors may confer multiple functions on a single cis-element or allow cooperative interactions between transacting factors. fos/jun (AP-1) and the ubiquitous octamerbinding factor are known to participate in transcriptional activation and/or DNA replication (10, 26-29, 40, 45); this study suggests that repression of c-myc expression may employ components common to transcriptional activation and/ or DNA replication.

EXPERIMENTAL PROCEDURES
Cells and Media-HeLa and WEHI 231 cells were grown in Dulbecco's modified Eagle's medium containing 10% fetal calf serum.
PC12 cells were grown in medium with 7% fetal calf serum and 7% horse serum. NGF (7S-NGF, Collaborative Research, Inc., Bedford, MA) treatment of PC12 cells was at a concentration of 125 ng/ml for 90 min.
Preparation of Nuclear Extracts-Nuclear extracts were prepared by the procedure of Dignam et al. (30) with the following modifications: buffer C contained 20% glycerol and 0.45 M NaCl; buffer D contained 20% glycerol and 80 mM KCI. All buffers contained 20 mM HEPES (pH 7.5). The protein concentration of the extracts was 30 The abbreviations used are: bp, base pairs; DTT, dithiothreitol; FRA, fos-related antigen; FT, flow-through; GALV, gibbon ape leukemia virus; PAGE, polyacrylamide gel electrophoresis; HEPES, 4-(2-hydroxyethylpiperazineethanesulfonic acid); NaDodSO1, sodium dodecyl sulfate. 8992 mg/ml. Whole cell extracts were made from PC12 cells; cell pellets were extracted with the addition of 3 volumes of buffer C and incubated for 40 min on ice.
Antibodies and Peptides-A fos peptide (amino acid from 129 to 153) (33) and affinity-purified antibodies to this peptide (fos-I) were kindly donated by Dr. Michael Iadarola (National Institutes of Health, Bethesda, MD). Affinity-purified antibodies to the M peptide of mouse c-fos ( fos-M) were the generous gift of Dr. T. Curran (Roche Institute of Molecular Biology, Nutley, NJ). Rabbit antisera against a human c-fos fusion protein (fos-S) were prepared by Dr. Dennis Salomon (UCLA School of Medicine). Affinity-purified antibodies to @-galactosidase were prepared as described previously (34).
Fractionation of Nuclear Extracts by Bio-Rex 70 Chromatography-HeLa cell nuclear extracts were fractionated by cation chromatography on Bio-Rex 70 (Bio-Rad). The extracts were dialyzed against buffer D containing 150 mM KC1 and loaded onto a Bio-Rex 70 column equilibrated with the same buffer. Approximately 1 ml of resin was used for every 10-20 mg of nuclear protein. The flowthrough (FT) fraction was collected, and the column was washed with buffer D, 150 mM KCI, until the absorbance of 280 nm was less than 0.05. The column was eluted with a linear gradient from 150 to 800 mM KC1 in buffer D. Gradient fractions were collected and, along with the FT, subjected to gel retardation assay using specific, doublestranded, radiolabeled probes. The FT fraction was precipitated with 70% ammonium sulfate. The peak binding fractions resulting from gradient elution were pooled and precipitated with ammonium sulfate from 0 to 35% and from 35 to 70%. The precipitated proteins were dissolved in a minimum volume of buffer D and dialyzed against the same buffer.
Immunobeads-Monoclonal mouse anti-@-galactosidase was purchased from Promega Corp. (Madison, WI) and coupled to Immunobeads (Bio-Rad), as described by the manufacturer. The lnc repressorgalactosidase fusion protein was prepared using the procedure of Fowler and Zabin (35).
Exonuclease Assay-Exonuclease assays were performed as described (17,36). PouII/SmaI or PuuIIIScaI fragments of the human c-myc gene were cloned into the SmaI site of pUC19 or pUC18 DNA (pUCPvuII/SmaImyc or pUCPvuII/ScaImyc, respectively). The cmyc and lac operator-bearing BstNI fragments of pUCPvuII/ScaImyc or pUCPvuII/SmaImyc were purified and end-labeled with [w3*P] deoxyribonucleotides a t both 3' ends by Escherichia coli DNA polymerase (large fragment). One to three nanograms of the DNA probes were immobilized on Immunobeads as described previously (17,34,36). Binding reactions, with or without specific competitors, were performed in a total of 15 pl of buffer D, including extract and 5 pg of poly[d(I-C)] (Sigma). Following incubation for 25 min a t 20 "C, the beads were recovered by centrifugation and resuspended in 15 pl of exonuclease buffer (10 mM Tris-HCI, pH 8.0.4 mM MgCI,, 0.2 mM dithiothreitol (DTT), and 5% glycerol). Twenty units of T7 gene 6 exonuclease (United States Biochemicals, Cleveland, OH) were added; the reactions were incubated for 3 min a t 30 "C and stopped with the addition of an equal volume of 10 mM Tris-HCI, pH 7.5, 20 mM EDTA, and 0.5% NaDodS04. The reactions were then extracted sequentially with phenol and chloroform. Then the DNA was recovered by ethanol precipitation, separated by gel electrophoresis (6% acrylamide/bisacrylamide (291), 8.3% urea), and visualized with autoradiography.
Chemical Footprint Analysis-Chemical footprint analysis was performed with the procedure described by M. Kuwabara and Sigman (37). A "2P-5'-end-labeled 45-bp oligonucleotide was annealed with unlabeled complementary strand to generate probe DNA. Sequences of the 45-bp oligonucleotides are: template strand 5"GCAGCTGnontemplate strand: 5'-CCGCATCCTTGTCCTGTGAGTATAAA-TCATCGCAGGCGGAACAGC-3'. A 6-fold increased binding reaction followed by gel retardation was performed as described above. After electrophoresis, the gel was soaked in 50 mM Tris-HCI, pH 8.0, 0.4 mM 1.10-phenanthroline, 0.09 mM CuS04, and 3-mercatopro-TTCCGCCTGCGATGATTTATACTCACAGGACAAGGATG-3', pionic acid (1/2000 diluted) for 7-12 min a t room temperature. The cleavage reaction was quenched by the addition of 2,9-dimethylphenanthrone to a final concentration of 2.8 mM. The gel was rinsed with distilled water and autoradiographed for 30 min a t room temperature. The bands were excised from the gel, and the DNA was eluted in 10 mM Tris-HCI, pH 7.8, 5 mM EDTA, 0.5% NaDodSO,, 50 pg/ml proteinase K overnight a t 37 "C. After filtration, the recovered DNA was extracted with phenol/chloroform (l:l), precipitated with ethanol, analyzed by denaturing 13% polyacrylamide gel electrophoresis (PAGE), and subjected to autoradiography.
Precipitation and Imrnunoblot Analysis of the c-myc DNA-Protein Complex-Eighty microliters of HeLa cell nuclear extract (2.4 mg, total protein) was mixed with 10 pg of poly[d(I-C)], 0.8 mg of bovine serum albumin, and 0.1% Tween 20 for 4 min a t 4 "C. The mixture was spun (9000 X g, 5 min a t 4 "C), and the supernatant was incubated with 150 ng of c-myc DNA (multimer of PuuII-ScaI fragment) or FIG. 1. Nuclear factors bind to a negative element of the human c-myc gene. A, DNA map of the upstream region and of exon 1 of the human c-myc gene. A negative element, from -350 to -294 bp relative to the transcriptional initiation site P1 of the human c-myc gene, is shown. A sequence within this element defined by DNase I footprint analysis is also shown. Hd, HindIII; Cl, ClaI; P o l l , PuuII; Sc, ScaI. B, gel retardation assays identify multiple c-myc DNA-protein complexes. c-myc oligonucleotide was radiolabeled and mixed with HeLa nuclear extract and assayed by gel retardation as described under "Experimental Procedures." In each lane, 0.1 ng of the probe DNA was used. Used as competitor DNAs were 30 ng of unlabeled oligonucleotides, c-myc, and HIV-Env. The DNA-protein complexes are designated as 1-111. F indicates free probe. Oligonucleotide sequences are shown in Table I. Binding Site on Human c-myc gene Complementary 45-base oligonucleotides encompassing the c-myc negative element were separately labeled with [r-'*P]ATP by polynucleotide kinase, and each was annealed with the appropriate unlabeled complementary strands. Gel retardation and chemical footprint analysis were performed as described under "Experimental Procedures." I and I1 indicate c-myc DNA-protein complexes I and 11, respectively, and F indicates a free probe.
DNA sequences of chemical footprints are shown, and the lines indicate the bases with clear protections.
control pUCI9 DNA previously immobilized on Immunobeads. Competitor DNA (1 mg) was included in some experiments. The binding mixture was incubated for 25 min a t room temperature and washed twice with buffer D (80 mM KCI). The bound proteins were eluted with 20 pl of 0.2% NaDodSO.. The eluted proteins were separated by 10% NaDodS0,-PAGE and transferred to nitrocellulose filter (550 mAmp 4 "C for 90 min). The filter was blocked with Blotto (5% Carnation non-fat dry milk, 50 mM Tris-HCI, pH 7.5, 50 mM NaCI, 1 mM EDTA, and 1 m~ DTT for 90 min a t room temperature), incubated with affinity-purified anti-fos antibodies (1:2500 dilution in Blotto a t room temperature overnight), and rinsed with TNE-50 (10 mM Tris-HCI, pH 7.5, 50 mM NaCI, 1 mM EDTA, and 1 mM DTT). Anti-rabbit Ig, "'I-labeled F(ab')2 (Amersham), was allowed to react with the filter in Blotto for 2 h a t room temperature. After several washes with TNE-50, the filter was dried for autoradiography.
Purification of Octamer-binding Factors-The ubiquitous and lymphoid octamer-binding proteins were partially purified from a nuclear extract prepared from the mouse B-cell line, WEHI 231, by affinity chromatography on an Ig octamer affinity column essentially as described (38). The sequence of the oligonucleotides used to make the affinity column was derived from the VI, 41 Ig promoter, shown in Tahle I. The nuclear extract was passed four times over the affinity column in the presence of poly[d(I-C)], and the column was eluted as described (38). The affinity-purified protein was subjected to electrophoresis on a 12.5% NaDodS0,-polyacrylamide gel. Proteins ranging from 43 to 120 kDa were eluted from 23 contiguous gel slices and renatured as described (39). The fractions containing octamer-binding proteins were determined by gel retardation assay. T o further characterize the nuclear factor(s), gel retardation assays were performed using an HeLa cell nuclear extract and a double-stranded oligonucleotide probe spanning the protected region from DNase I digestion. Three well-separated DNA-protein complexes were observed in the assay. Formation of complexes I and I1 was significantly inhibited by the inclusion of unlabeled c-myc oligonucleotide as competitor, confirming the specificity of these complexes for c-myc sequence. Formation of complex I11 was reduced slightly. In contrast, a double-stranded oligonucleotide containing reading frame of the HIV-1 envelope gene inhibited none of the complexes (Fig. 1B). In order to explore the relation of the two specific complexes, HeLa cell nuclear extract was subjected to cation exchange chromatography with Bio-Rex 70. The binding activity producing retarded complex I was present in the FT fraction, whereas complex I1 was retained and eluted as a single peak with a salt gradient ( Fig. 2A). When an oligonucleotide for the gibbon ape leukemia virus (GALV) enhancer factor binding site (36) was used as a probe, formation of DNA-protein complexes was also observed in both the FT and peak fractions, and unexpectedly the mobilities of those complexes were indistinguishable from the cmyc DNA-protein complexes. Exonuclease binding assays using HeLa cell nuclear extract were performed as described under "Experimental Procedures," with or without the addition of competitor oligonucleotides. 10 p1 of HeLa cell nuclear extracts were mixed with a radiolabeled BstNI fragment of pUCPvuII/ScaImyc-bearing rnyc sequence in this experiment. 600 ng each of c-rnyc and GALV oligonucleotides were used as competitor DNAs. The arrow indicates the bands produced by the hydrolysis of probe DNAs. C, effects of oligonucleotides bearing consensus AP-1 binding site on c-myc DNA-protein complexes in the gel retardation assays. Gel retardation assays were performed with radiolabeled c-rnyc oligonucleotides and HeLa cell nuclear extracts as described under "Experimental Procedures." Indicated amounts of unlabeled oligonucleotides were used as competitor DNAs. The sequences of the competitor oligonucleotides are shown in Table I. Sequence-The Bio-Rex 70 FT and peak fractions were also assayed for sequence-specific DNA-binding proteins using exonuclease binding assay.

Two
Both fractions yielded barriers to exonuclease hydrolysis a t the same position as unfractionated crude extract, indicated with an arrow in Fig. 2B. A mixture of FT and peak fractions yielded coincident exonuclease-resistant bands, suggesting that both factors bound to the same, or highly overlapping, elements (Fig. 2B).
Chemical footprint analysis was performed to confirm the overlapping binding motif of the two complexes. Following nondenaturing electrophoresis, the gel was soaked in 1,lOphenanthroline and copper sulfate; complexes I and I1 as well as the free probe were then excised from the gel, and the digested DNA was recovered and analyzed as described under "Experimental Procedures." The complexes I and I1 showed indistinguishable footprints (Fig. 2C). Thus, by gel retardation, exonuclease digestion, and chemical footprint analysis, these two factors (complexes) were shown to share a common binding site.

fosljun (AP-I) Binds to the c-myc Negative Element-As
shown in Fig. 2A, the GALV probe interacted more strongly with the peak fraction than with the FT fraction, whereas the binding pattern was reversed with the c-myc probe (Fig. 2 A ) . A cross-competition test was performed to determine if the GALV and c-myc DNA-protein complexes were related. Complex I (formed with the FT fraction) was more efficiently reduced by the c-myc competitor than by the GALV oligonucleotide, whereas complex I1 (formed with the peak fraction) was more efficiently reduced by GALV than by the c-myc competitor (Fig. 3A). Together with the direct binding data shown in Fig. 2 A , the results suggest that the protein responsible for c-myc DNA-protein complex 11, which fractionates in the Bio-Rex peak fraction, was closely related to GALVenhancer binding factor (36).
Exonuclease binding assays were performed to confirm that the GALV-enhancer binding factor participated in the c-myc DNA-protein complex. The exonuclease assay was performed using HeLa cell nuclear extract and c-myc DNA probe, with the inclusion of unlabeled oligonucleotides as competitors. The GALV oligonucleotide competitor partly reduced binding to the c-myc probe, whereas the complete competition was observed with c-myc competitor (Fig. 3B). This result supports the notion that one of the two c-myc DNA-protein complexes contained GALV-enhancer binding factor. Inclusion of unlabeled GALV oligonucleotide as competitor in gel retardation assays abolished formation of c-myc DNA-protein complex I1 (data not shown).
Because the consensus binding sequence for the transcription factor AP-1 (TGAC/GTCA) is present in the GALV oligonucleotide (Table I), we examined whether the c-myc DNA-protein complex I1 also contained AP-1. Other oligonucleotides containing AP-1 binding site (Table I) were employed as competitors in gel retardation assays. Collagenase and stromelysin AP-1 sites significantly inhibited the second c-myc DNA-protein complex, whereas non-Ap-1 oligonucleotide, HIV-Env, did not. Complex I11 was also detectably reduced by the oligonucleotides of collagenase and stromelysin (Fig. 3C).
The c-fos protein has been shown to function as a transcriptional regulator (10). Recent reports have shown that the cfos protein associates with a cellular protein, p39, which is the product of the c-jun proto-oncogene, and that the fosljun complex binds to AP-1 sites (40-46). We examined the effect of affinity-purified anti-fos immunoglobulin on the c-myc DNA-protein complexes using gel retardation assays. One antibody, anti-fos-I, inhibited the second complex specifically, and a new shift was observed as indicated with an arrow in Fig. 4. In contrast, anti-P-galactosidase antibody had no effect on the c-myc DNA-protein complexes. The inhibitory effect of the fos antibody was abolished by addition of the c-fos peptide that was used to generate the rabbit anti-fos antibody. Independent preparations of anti-fos immunoglobulin, antifos-M and anti-fos-S, also inhibited the formation of complex 11, but did not produce new bands. It is noteworthy that complex 111, which was inhibited by AP-1 oligonucleotides (Fig. 3C), was not affected by the anti-fos antibodies (Fig. 4).
T o directly demonstrate the participation of the c-fos protein in a c-myc DNA-protein complex, HeLa cell nuclear extract was incubated with c-myc DNA immobilized on beads. The resulting DNA-protein complex was precipitated and washed the bound proteins were eluted with NaDodS04 and analyzed by immunoblot. Anti-fos antibody reacted with a major band of 55 kDa and minor bands of lower molecular masses. The molecular weight of the major band conforms to that reported for the c-fos protein, and the minor bands are consistent with the presence of fos-related antigens (FRAs) (46). c-myc oligonucleotide competitor completely abolished the precipitation of c-fos and FRAs, while &X DNA showed only slight inhibition. pUC DNA immobilized on beads precipitated only a trace amount of immunoreactive material (Fig. 5, left). These results demonstrate the specific binding of fos and FRA to the c-myc negative regulatory region. Unfractionated HeLa cell nuclear extract as well as PC12 The sequences of the competitor oligonucleotides are shown under "Experimental Procedures." B, octamer-binding activity is observed in the F T fraction of HeLa cell nuclear extracts. Fractionated HeLa cell nuclear extracts were probed with octamer oligonucleotides. The FT fraction contained binding activity, whereas the peak fractions did not. a and b indicate the peak fractions precipitated with 0-35% and 35-70% ammonium sulfates, respectively. 1 pl of each fraction was incubated with a radiolabeled 26-bp oligonucleotide bearing the octamer motif as shown in Table I. C, the octamer oligonucleotide competes for the complex I more efficiently than c-myc oligonucleotide.
The efficiency of competition for the first c-myc DNA-protein complex was compared using either c-myc or octamer oligonucleotide competitors. The FT fraction (1 pl) was incubated with probe oligonucleotide and either c-myc or octamer (Octa) oligonucleotides with the sequences shown in Table I. D, purified octamer-binding proteins bind to c-myc sequence. Octamer-binding proteins were purified from a B-cell line, WEHI 231, as described under "Experimental Procedures," and incubated with radiolabeled c-myc oligonucleotide. Both ubiquitous and lymphoidspecific octamer-binding proteins formed complexes with c-myc DNA. Radiolabeled octamer oligonucleotide was also used as a control probe, which showed stronger binding than the c-myc probe.
whole cell extracts, with and without NGF treatment, showed the expected pattern of fos and FRA immunoreactivity (Fig. 5,right).
An Octamer-binding Factor Recognizes the c-myc Negative Element-Gel retardation analysis in the presence of a variety of competitor sequences revealed that c-myc DNA-protein complex I was efficiently inhibited by sequences derived from the immunoglobulin K promoter, which contains an octamer motif (48). Inclusion of a 40-bp double-stranded octamer oligonucleotide as competitor abolished c-myc DNA-protein complex I. Unrelated oligonucleotides of the same length did not reduce the signal, indicative of specific binding by octamer factor (Fig. 6A). When an octamer oligonucleotide of 26 bp was radiolabeled and mixed with the FT or peak fractions, specific binding was observed to be selective with the FT fraction ( Fig. 6B): this complex comigrates with the c-myc DNA-protein complex (data not shown). Comparison of identical length c-myc and octamer oligonucleotides as competitors in gel retardation assays, with c-myc probe and the FT fraction, revealed that the octamer oligonucleotide reduced binding more efficiently than the c-myc oligonucleotide, suggesting that an octamer-binding factor constitutes the protein in c-myc DNA-protein complex I (Fig. 6C).
Two types of octamer-binding factors have been identified; one is ubiquitous (27,28,47,62,63), and the other is lymphoid specific (48,49,60,61,64). Because c-myc DNA complex I is formed with HeLa cell factors, it is likely to contain the ubiquitous factor. T o directly demonstrate the interaction of c-myc sequences with octamer binding factors, both the ubiquitous and lymphoid-specific octamer protein, purified from WEHI 231 cells, were employed in gel retardation assays using the c-myc probe. The purified octamer proteins bonded to c-myc DNA, although the interaction was less than with the octamer probe. As expected, the purified lymphoid-specific factor produced a complex with greater mobility than the ubiquitous factor (Fig. 6D). fos/jun(AP-1) and the Octamer-binding Factor Share a Common Binding Site in the c-myc Negative Element-From the results of the exonuclease assay and the chemical footprint analysis as well as from a comparison of the c-myc, GALV, and octamer DNA sequences, it is apparent that the binding sites on c-myc for complexes I and I1 overlap extensively in the central portion of the DNase I footprinted region. When comparing the c-myc and octamer sequence, one homologous segment was noted. Ten contiguous bp match each other a t the purine-pyrimidine level with 6 identical bp (Fig. 7A). Between c-myc and GALV, there are two homologous sequences within the chemically footprinted region. In one case, 8 of 9 base pairs match at the level of purine-pyrimidine; out of the 8, 7 bp are identical (Fig. 7A). In the other case, 6 consecutive bp of each match at the purine-pyrimidine level; of these 6 bp, 5 are identical (Fig. 7B). The binding site shown in Fig. 7 for fosljun (AP-1) and the octamer binding protein overlap more extensively in case of A than in the case of B. The overlapping sequence is located in the central portion of the 26-bp c-myc negative element (Fig. 7).
Because purified ubiquitous octamer-binding protein produced a complex with mobility similar to complex I from crude extract and because this complex is refractory to inhibition by anti-fos or AP-1 oligonucleotide competition, it can be concluded that fosljun is not a component of complex I.
Similarly, because complex I1 is not altered by octamer competition and because it is unlikely that addition of fosljun to complex I would increase electrophoretic mobility, it is apparent that complex I1 contains no octamer-binding protein.
No complex migrating slower than complex I is present, which suggests that a quaternary complex of fos, jun, octamerbinding protein, and probe does not form.

DISCUSSION
We have shown that two entirely distinct factor complexes bind to a c-myc negative element. These two complexes bind to highly overlapping segments of the c-myc sequence. One of the complexes is fos/jun (AP-l), and the other is the ubiquitous octamer-binding factor.
There is no obvious homology between the consensus recognition sequences for these two factors. The binding assays shown above clearly demonstrate that AP-1 and octamer consensus sequences will not cross-compete binding to their respective proteins except at very high molar excess. It was unexpected, therefore, that a single element with reduced homology to each recognition sequence would bind both the fosljun (AP-1) and the ubiquitous octamer protein. It is apparent that enough degeneracy exists within the sequences recognizable by AP-1 and the octamer-binding proteins to allow binding of both factors to a common site. Examination of the known binding sites for most eukaryotic transcription factors reveals considerable plasticity in the permissible deviations from the deduced consensus sequences. Such plastic-ity is not intrinsic to the process of sequence recognition; for example most restriction enzymes are extremely rigid with respect to sequence recognition.
By creating heterogeneity among the binding sites for a single trans-acting factor, some of those sites may be allowed to overlap with the binding sites for other regulatory proteins. A variety of antagonistic or synergistic interactions between factors may therefore expand the range of physiological responses mediated by a given cis-element. Without degeneracy in the recognition sequences the opportunities for allowing alternative binding of different factors would be reduced. For example, whereas this paper demonstrates that there exists an overlap between AP-1 and octamer-binding sites, the degeneracy of the AP-1 consensus allows overlapping binding sites for AP-1 and AP-4 in both viral and cellular genes (50).
The degeneracy of recognition sites that allows overlapping of different cis-elements also allows the assembly of compact, powerful regulatory regions such as enhancers. Viral enhancers, such as those found in SV40 or polyoma, possess a complicated and interdigitating array of positive and negative cis-elements. Without degeneracy, the constraints of rigid recognition sequences might well increase the amount of DNA needed to specify a regulatory response.
The 26-bp segment containing the fosljun (AP-1) and octamer-binding sites has been shown to contain the principal determinant of a negative cis-element for c-myc.' At this time it is not possible to say whether the fosljun (AP-1) or the octamer-binding protein is responsible for the down regulation of c-myc expression mediated by their common binding site. Each of these factors has already been demonstrated to play a role in more than one regulatory process. We cannot exclude a role for other proteins in decreasing c-myc expression from this site.
AP-1 was discovered as a transcriptional activator protein, which binds to regulatory elements of viral and cellular genes, such as SV40 and metallothionein (26). The proto-oncogene c-jun, a cellular homologue of the retroviral oncogene v-jun, has been identified as one of the proteins that binds to AP-1 sites (40, 45). Recent results indicate that the fos protein interacts with the product of c-jun, p39, and that this complex binds to 46). The role of the fos protein in transcriptional activation has been shown (10). However, it has been shown that c-fos transcription is negatively autoregulated by the fos protein itself and the fosljun (AP-1) complex also binds to a negative regulatory element of an adipocyte-related gene, aP2 (51, 52). Identification of the fos/jun (AP-1) complex binding to a c-myc negative element comforms to these observations. It is not yet known how the fos/ jun complex mediates both transcriptional activation and repression.
The octamer motif has been implicated in transcriptional activation of cellular and viral regulatory regions (for review see Ref. 64). Interestingly, this factor also binds to the origin of DNA replication on the adenovirus genome and mediates the initiation of replication (28,29). Recently, one of the octamer-binding proteins was implicated as a negative regulator of immunoglobulin heavy chain gene expression? There is precedent for cis-elements and trans-acting factors common to transcriptional activation, repression, and replication (53-56).
Extensive study will be required to determine the relative contributions of fosljun (AP-1) and octamer-binding proteins to transcriptional repression of c-myc. However, this problem is complicated. First, it is difficult to mutate the binding site for a single factor, because both binding sites overlap. A mutation in the recognition sequence for one factor will be also a mutation in the other. Second, it has been shown for both the thyroid hormone and glucocorticoid receptors that some binding sites are positive elements whereas others are negative. Alteration in protein tertiary structure or conformation, caused by subtle differences in DNA-binding sequences, may alter the nature of the physiological response (56). If the deviation of the c-myc negative element from the consensus AP-1 and octamer sequences is critical for producing the negative effect on c-myc expression, then alteration of this sequence may abolish the negative response. Therefore, the direct substitution for the c-myc sequence by consensus octamer or AP-1 sequences cannot necessarily determine the relative contribution of either factor. Third, it is possible that both factors may act synergistically to maintain a negative effect on c-myc expression. Because the ubiquitous octamerbinding factor has been reported to work selectively in the S phase of the cell cycle, and fos and jun have been shown to be induced during the Go/GI transition (27,57-59), the factors may repress c-myc transcription at different points in the cell cycle.