Products of the fos and jun proto-oncogenes bind cooperatively to the AP1 DNA recognition sequence.

The products of the proto-oncogenes c-fos and c-jun form a tight protein complex that is a major component of the transcription factor AP1. To analyze the role of fos in the binding of this complex to the AP1 DNA recognition sequence and the mechanism of interaction in further detail, we have expressed a fos protein in E. coli using an expression vector containing the temperature-inducible lambda PL promoter and a synthetic translational start codon. The fos protein encoded by this construct (termed Baf) was enriched by biochemical purification techniques and was found to form a specific complex with c-jun obtained by in vitro transcription/translation. As shown in gel retardation assays, the baf/jun complex binds to the AP1 DNA recognition sequence with high affinity, while no significant binding was observed with either of the individual protein components, indicating cooperative DNA binding of the two proteins. The fact that the bacterial baf protein does not undergo glycosylation indicates that the post-translational modification of eukaryotic c-fos with N-acetylglucosamine is not required for the formation of a stable fos/jun/DNA complex. ImagesFIGURE 2.FIGURE 3.FIGURE 4.FIGURE 5.FIGURE 6.FIGURE 7.


Introduction
The nuclear product (c-Fos) of the proto-oncogene c-fos has been implicated in a variety of biological processes including growth control, differentiation, and signal transduction in neurons (1)(2)(3). These multiple roles of c-Fos seem to be cell-type specific and may thus be due to the transregulation of defined sets of genes in specific cell types. A great number of results obtained within the last 2 years has provided overwhelming evidence that Fos indeed plays a pivotal role in transcriptional regulation (4). Based on the observation that c-Fos is present in a transcription complex bound to the promoter of the gene encoding the adipocyte protein -2 (aP2) (5), it could be demonstrated that the Fos protein complex binds to the DNA recognition sequence (TRE) of the transcription factor AP1 (6)(7)(8)(9)(10)(11). AP1 had previously been identified by virtue of its sequencespecific interaction with the SV40 early promoter region and its inducibility by the tumor promoter TPA (12)(13)(14). Interestingly, a major component of AP1 is highly related to and probably identical to the product of the proto-oncogene c-jun (15,16), pointing to a direct link between the two nuclear proto-oncogene products. In addition, it is now clear that the Fos-associated protein p39, previously detected by virtue of its co-precipitation by Fos antibodies (17)(18)(19)(20) is identical to AP1 (8,10,21). It could also be shown that Fos can transactivate APidependent transcription in transient assays where Fos expression vectors were co-transferred with a chimeric TRE-HSVtK promoter-CAT construct into 3T3, HeLa, or F9 cells (10,11,22,23). In addition, Fos can transrepress certain promoters such as those of the c-fos and HSP70 genes, but the mechanism of transrepression is unclear (9,22,24).
In the present study we have investigated the reconstitution of a Fos/Jun complex in vitro to obtain direct evidence for the association of these proteins, to analyze the mechanism of their interaction, and to investigate the role of the complex formation in their binding to the TRE. For this purpose, we expressed Fos in E. coli and used a fraction enriched in Fos protein in in vitro reconstitution assays. Our results clearly show that the bacterially expressed Fos forms a complex with Jun and that this complex-in contrast to the individual proteins-possesses a high affinity for the TRE. We also show that glycosylation, which is found with cellular Fos but not with Baf protein, is not required for the formation of the Fos/Jun/TRE complex.

Materials and Methods
Expression of Baf in E. coli E. coli RR1 containing the pbaf plasmid ( Fig. 1) were grown in a New Brunswick gyrotory shaker at 30°C in LB-medium supplemented with ampicillin (50 pRg/mL).
At early log-phase (OD600 0.1-0.3), cultures were transferred for 30 min to a water bath at 42°C followed by 1 hr incubation at 39°C on the shaker. Cells were collected by centrifugation. The resulting pellet was resuspended in 50 mM Tris, pH 8.0, 2 mM EDTA, 0.1 mM DTT, 5% glycerol, 0.1 mM PMSF (buffer A, 3 mL/g cell pellet) and treated with lysozyme (1 mg/mL) for 15 min at 25°C. The cells were disrupted on ice with a sonicator (Branson Sonifier B-12) by three 5-sec bursts using a microprobe. The lysate was centrifuged at lOOOg for 5 min at 4°C, and the resulting pellet was extracted twice in buffer A (5-mL/g starting material) containing 1 M urea. For each extraction step, cells were homogenized in extraction buffer using a Dounce homogenizer and were left for 45 min at 4°C on a rotating wheel. Insoluble material was sedimented by centrifugation at 4°C in a Beckman JA 20 rotor (14,000 rpm for 10 min). The resulting final pellet was extracted overnight with 5M urea in buffer A (5 mL/g starting material). After centrifugation, as before, the supernatant was used for further purification of Baf. The presence of Fos in the various extraction steps was determined by immunoblotting.

Ion-Exchange Chromatography
A Q-Sepharose column (1 x 5 cm, Pharmacia, Uppsala, Sweden) was equilibrated in 20 mM Tris-HCl pH 8.0, 2.5 M urea (buffer B). Five molar urea extract containing Fos protein was diluted 1:1 with buffer A and applied at a flow rate of 0.8 mL/min. After washing with buffer B, Fos protein was eluted with a linear 0 to 1 M NaCl gradient. Fractions of 1 mL were collected and assayed for Fos protein by immunoblotting.

In Vitro Reconstitution and Immunoprecipitation
Linearized T7 expression plasmid, 1 ,ug containing the c-jun cDNA (25) was incubated with 0.5 mM nucleoside triphosphates (ATP, CTP, GTP, UTP), 1 unit of RNAsin (Genofit), 0.4 OD260 units CAP nucleotide (Boehringer) and 5 units of T7 RNA polymerase for 30 min at 37°C in a total volume of 50 ,uL transcription buffer (40 mM Tris, pH 7.9, 6 mM MgCl2, 2 mM spermidine, 10 ,uM dithiothreitol, 5 ,ug bovine serum albumin). Another 5 units of enzyme were added, and the reaction mixture was incubated for 30 min. The RNA was checked for integrity on a 1% agarose gel. In vitro transcribed RNA was mixed with 60 ,uL of reticulocyte lysate (Amersham)rplus, if the protein was to be labeled, 30 ,uL (450 ,uCi) of 3 S-methionine. The reaction mixture was adjusted to 150 ,uL with H20 and incubated at 30°C for 1 hr.
Complex formation was carried out by incubating in vitro translated 35S-labeled c-Jun and unlabeled extracts (13 pL and 10 ,uL, respectively), in 30 ,uL H20 for 1 hr at 30°C. After the addition of 5 ,uL of 10 x RIPA buffer, the mixture was incubated with 4 ,uL of antiserum 455 (26,27) for 1 hr on ice followed by incubation with 40 ,uL Pansorbin (Calbiochem) for 1 hr. Immune complexes were sedimented and washed twice with RIPA buffer.
SDS sample buffer was added to the pellet, the samples heated to 95°C for 3 min, and separated on a 11.5% SDS-polyacrylamide gel. After electrophoresis, gels were processed in Amplify (Amersham), dried, and exposed to Fuji RX film.

Gel Retardation Analysis
Binding reactions were performed by preincubating 10 ,uL Baf and 5 ,iL in vitro translated c-Jun, either alone or combined, with 1 pug of poly(dI-dC) in a buffer containing 10 mM Hepes, pH 7.9, 60 mM KCI, 4% Ficoll, 1 mM EDTA, and 1 mM DTT for 60 min at 25°C. Doublestranded oligonucleotide, 32P-labeled with polynucleotide kinase, 3.5 fmole, was added and incubation was continued for 30 min at room temperature. The reaction mixtures were separated on 4% polyacrylamide gels at room temperature at 10 V/cm (28). Gels were dried and exposed to Kodak RP film.

Lectin Binding
Wheat germ agglutinin (WGA) immobilized on Agarose (Pharmacia) was washed three times with TTBS (50 mM Tris, pH 7.4, 150 mM NaCl, 0.05% TWEEN 20), resuspended in 1 volume TTBS, and incubated for 16 hr at 4°C with an equal volume of HeLa cell or bacterial.cell extract, respectively (300 jig protein). Prior to the lectin binding reaction, the bacterial cell extract was incubated with reticulocyte lysate for 2 hr at 25°C. After extensive washing in TTBS, the beads were extracted at 95°C for 5 min with SDS sample buffer. The extracts were resolved by SDS-PAGE and analyzed by Western blotting using anti-Fos antibodies (27). For competition experiments the WGA beads were suspended in 2 volumes of TTBS containing 0.5 M N-acetylglucosamine for 2 hr at 4°C on a rotating wheel. The binding reaction was carried out in the presence of 0.25 M N-acetylglucosamine.

Results
Expression of Fos Protein in E. coli and Enrichment for Baf by Ion-Exchange Chromatography As the expression vector for production of Fos in E. coli we chose pEV1 containing the heat-inducible PL promoter, a 5' located start codon and multiple cloning site (29). To generate the Fos insert, we first subcloned an intronless hybrid v-fos/c-fos gene into pUC19, released the insert with EcoRI and Hind III, and introduced this fragment into pEV1 via the corresponding restriction sites generating the pbaf construct (Fig. 1). The product encoded by pbaf consists of 8 N-terminal vector-derived amino acids fused to 359 amino acids of Fos (residues 22-380 of c-Fos) (Fig. 1). Otherwise, this protein which will be subsequently referred to as Baf, differs from c-Fos only in the 5 FBJ-MSV-specific single amino acid changes (30). The pbaf construct was introduced into E. coli. RR1 cells and individual colonies were analyzed for Baf expression. Two different clones were analyzed in Figure 2 and used for subsequent analyses, one clone expressing Baf only after heat induction and one clone expression Baf constitutively (Fig. 2). The Mr of full-length Baf is approximately 45 kd, the other bands visible on the Western blot in Figure 2 presumably represent degradation products of Baf. The absolute amount of Baf produced in these cells is relatively low, since we were not able to detect Baf reproducibly on Coomassie blue-stained polyacrylamide gels without further purification (not shown). In the bacterial extract, the majority of the Baf protein was found to be insoluble. After centrifugation of the homogenate the supernatant did not contain any detectable amount of the full-length product and only a. small fraction of the shorter Mr form (Fig. 2). However, Baf protein could be readily solubilized from the pellet by dissolving in 5M urea (Fig. 2). The extracted Baf protein was further enriched by chromatography on a Q-Sepharose ion exchange column run in the presence of 2.5 M urea and eluted with a linear gradient of 0 to 1 M NaCl (Fig. 3A). The protein eluted from this column was collected in approximately 60 fractions, which were analyzed for the presence of Baf by Western blotting. Figure 3B shows that Baf eluted within a narrow range, approximately in pools C and D, corresponding to fractions 43 to 48 in Figure  3A. These fractions were pooled and used for the subsequent binding studies.

Complex Formation between Baf and In Vitro Translated c-Jun
Baf protein was then used in an in vitro assay to reconstitute a Fos/Jun complex. For this purpose, radioactively labeled c-Jun was synthesized by in vitro transcription/translation of a c-Jun cDNA in a T7 poly- The mixture was then subjected to immunoprecipitation using a Fos-specific antibody that does not cross-react with Jun (Fig. 4). The precipitation of 35S-labeled c-Jun in the presence of Baf indicates the formation of a protein complex between Baf and Jun. No significant amount of Jun was precipitated when control preparation from the parental RR1 cells (which do not harbor the pbaf plasmid) were incubated with the labeled c-Jun or when the fos-specific antibody was omitted from the immunoprecipitation. These findings strongly suggest that the Baf protein is capable of forming a complex with c-Jun protein. Baf and c-Jun Act Synergistically to Form a Complex of High Affinity for the API Binding Site To investigate the role of complex formation between Fos and Jun, we analyzed the binding properties of Baf and Jun individually or as a complex for the AP1 DNA recognition sequence. This experiment was performed by using these proteins and a 2P-labeled synthetic oligonucleotide (designated apl) containing a palindromic AP1 binding site in a gel retardation assay. Figure 5 shows that neither c-Jun nor Fos produced any detectable band shift, while the Baf/c-Jun complex gave a very strong shift. The Baf/Jun/DNA complex formation was also performed in the presence of excess unlabeled competitor oligonucleotides. Figure 6 shows that only the specific apl oligonucleotide but not the random oligonucleotide (rd) was able to abolish the band shift. This demonstrates that the retarded band indeed represents a specific complex between Baf, c-Jun, and the AP1 binding site. In addition, a complex of very similar elec- trophoretic mobility was found when nuclear extract from serum-stimulated HeLa cells was used in the gel retardation experiment (Fig. 6).

O-Glycosylation of Fos Is Not Required for the Formation of a Stable Fos/Jun/TRE Complex
It has recently been shown that the transcription factor SP-1 purified from HeLa cells is post-translationally modified by 0-glycosylation and binds to wheat germ agglutinin (WGA) (31). The presence of the 0-linked N-acetylglucosamine residues appears to be important for transcriptional activation by SP1 (31). In the same study it was shown that other transcription factors, among them the APl/Fos protein complex, can be labeled in vitro by 3H-galactose and galactosyl transferase (an enzyme transferring galactose onto terminal GlcNAc residues), but the nature and role of this modification was not investigated further. To elucidate some of these questions, we analyzed the cellular Fos protein complex and Baf protein for their affinity for WGA. Figure 7 shows that the Fos protein complex from nuclear HeLa cell extract binds to WGA and that this binding can be blocked by N-acetylglucosamine. In contrast, Baf protein does not bind to any significant extent, even after incubation with reticulocyte extract. (This control was important to show that no glycosylation occurs during complex formation with the in vitro translated c-Jun.) These results therefore indicate that FIGURE 6. Competition of Baf/Jun/TRE complex formation by specific (apl) and nonspecific (rd) oligonucleotides. Lanes 1, 2, 3, and 4 represent 10-, 100-, 1000-and 10,000-fold excesses of competitor DNA. The sequence of apl is identical to the labeled oligonucleotide; rd has the sequence 5' GCGACTAACATCGATCG (upper strand). The right-most lane shows a band shift with nuclear extract from serum-stimulated HeLa cells.
the formation of a stable Fos/Jun/TRE complex does not require 0-glycosylation of Fos.

Discussion
It is now well established that Fos protein plays a crucial role in the regulation of APi-dependent transcription. This is mainly based on two lines of evidence. First, Fos is a constituent of transcription complexes forming over AP1 DNA recognition sequences (5)(6)(7)(8)10,32), and second, Fos can transactivate promoters containing AP1 binding sites (TREs), such as the human collagenase promoter or the HSV-tk-promoter, after insertion of 5 TRE elements (8,10,11,22). In these transcription complexes Fos tightly associates with APi/ ....: p. i I ,comp. Jun (6,8,10) which is identical with the previously identified Fos-associated protein p39 (17)(18)(19)(20). In the present study we have attempted to reconstitute a protein complex between Fos and Jun proteins. For this purpose, we have expressed and partially purified Fos protein in bacteria. This Baf protein is largely identical with c-Fos, except for the substitution of 21 amino acids at the N-terminus with 6 vector-derived amino acids and 5 internal point mutations (Fig. 1). All these alterations, however, have no noticeable effect on the biological properties of the Fos oncogene product (11 26; unpub-lished data).
Our data clearly show that Baf forms a specific complex with c-Jun protein obtained by in vitro translation (Fig. 4) and that only this complex binds with high affinity to the APi DNA recognition sequence (Fig. 5). In contrast, the individual Baf and Fos proteins do not bind to any detectable extent in this assay. It has been described that bacterially expressed Jun can bind to the TRE (15). This may be due to the high concentration of Jun present in bacterial extract, which may compensate for the low affinity of Jun protein for the TRE. In a recent study, it was indeed shown that in vitro translated Jun protein binds to an oligonucleotide containing an APi site when the concentration of operator DNA is very high (32)g Taken together these observations suGgest thapteFos p wroein stbluies tbyhea wheak bindingr of Jun protein to the TRE. We have recently been able to show by site-directed mutagenesis that specific amino acid sequences in Fos are required for the formation of a stable Fos/Jun/TRE complex (33)(34)(35), pointing to a sequence-specific DNA recognition by Fos. It thus appears that in the protein complex forming between Fos and Jun via the "leucine zipper" (36)(37)(38) both constituents bind to specific DNA sequences, probably located within the TRE.
In agreement with our results obtained with the bacterial Fos protein, it has been shown in several other studies that both in vitro Fos and Jun are required for the formation of a stable complex with the TRE in vitro (32,33,37,39,40). In one of these studies (39), it was also shown that Fos is unable to form homodimers, which probably explains why Baf does not show any specific binding to the TRE in our gel retardation assays (Fig.   5).
The Fos protein complex has recently been reported to be labeled in vitro by 3H-galactose and galactosyl transferase (31). We therefore investigated whether O-glycosylation of Fos protein is required for the formation of a Fos/Jun/TRE complex. Our findings confirm the post-translational addition of sugar residues to cellular Fos protein complex that could be shown to bind to WGA, indicating the presence of N-acetylglucosamine (Fig. 7). In contrast, no such modification is present in Baf (Fig. 7). This observation shows that 0glycosylation of Fos is not required for the sequencespecific DNA binding of the Fos/Jun complex, an observation that is in agreement with the data obtained for the transcription factor SP1 (31). It may be that, as in the case of SP1, glycosylation of Fos is required for its transregulatory properties.