Two Purified Factors Bind to the Same Sequence in the Enhancer of Mouse

Volume 17 Number 13 1989 Nucleic Acids Research

Two purified factors bind to the same sequence in the enhancer of mouse MHC class I genes: one of them is a positive regulator induced upon differentiation of teratocarcinoma cells

Alain Israel, Osamu Yano*, Fr6dIrique Logeat, Mark Kieran and Philippe Kourilsky

Unite de Biologie Moleculaire du Gene, U.277 INSERM, UAC 115 CNRS, Institut Pasteur, 25 rue du Dr Roux, 75724 Paris CUdex 15, France

Received April 4, 1989; Revised and Accepted May 24, 1989

ABSTRACT The MHC class I murine and beta-2-microglobulin genes are silent in embryonal carcinoma (EC) cells but are induced upon differentiation of these cells. We have previously shown that enhancer-like sequences located in the promoter of the H-2Kb gene are non-functional in F9 and PCC3 cells. We have previously purified a 48 kd protein (KBFI) from a mouse T cell line which binds to a palindromic sequence located in this enhancer and to a similar sequence in the promoter of the beta-2-microglobulin gene. We des- cribe here the purification of a second protein (KBF2, 58 kd) which also binds to this sequence. While both activities are present in differentiated cells, KBF1 binding activity is absent in undifferentiated EC cells, where the palindromic sequence shows no enhancer activity. Upon differentiation, KBF1 binding activity is induced and the palindromic sequence becomes active as an enhancer. Thus, the absence of KBF1 activity in undifferentiated EC cells is at least in part responsible for the lack of expression of H-2 class I and beta-2-microglobulin genes in these cells and suggests that KBF1 activity is regulated during differentiation.

INTRODUCTION The major class I transplantation antigens specified by class I genes of the major histocompatibility complex (MRC) play a key role in a number of immunological processes, particularly in the recognition of foreign antigens by cytotoxic T cells of the host (1). Their expression is developmentally regulated : class I mRNA and proteins are not expressed until midsomite sta- ge of mouse embryogenesis (2). Then, they are expressed on most somatic cells of the adult organism. Undifferentiated F9 cells exhibit a variety of molecular and cellular properties characteristic of early embryos (3) ; in particular, they do not express class I antigens (4, 5, 6). These cells can be induced to differentiate after treatment with retinoic acid and dibutyryl cyclic AMP (7). This differentiation is accompanied by expression of class I antigens (8) at the cell surface. This regulation involves activation of both beta-2-microglobulin and H-2 genes, and is likely to operate at the transcriptional level (5, 9). It has been reproduced after transfection of

EC cells by a cloned H-2 gene (10). Fusion experiments between undifferentiated EC and differentiated cells have given conflicting results as to whether positive or negative regulation is involved (11, 12). Differential DNA methylation between undifferentiated and differentiated cells has also been invoked (13). We (14, 15) and others (16, 17) have shown that a major element of the mouse MHC H-2Kb gene promoter is a palindromic sequence located approximately 160-180 bp upstream from the cap site. This sequence, TGGGGATTCCCCA is part of a larger sequence exhibiting enhancer activity in differentiated but not in undifferentiated embryonal carcinoma (EC) cells (14). A protein, called KBF1, which recognizes this palindrome, and a similar sequence located in the beta-2-microglobulin promoter, has been characterized (15, 17) and purified from a mouse T cell line (18). A similar activity, called H2TFI, has been characterized in human HeLa cells (17, 19). Mutations which abolish enhancer activity in the Kb promoter also prevent binding of KBF1 (15). In vivo competition experiments have suggested that KBF1 is a positive trans-acting factor regulating the expression of mouse MHC class I genes (15, 17). We wanted to know whether the lack of expression of class I and beta-2-microglobulin genes in undifferentiated EC cells correlates with the absence of KBFI activity. This question turned out to be more difficult to answer than we anticipated. We found that a second protein, named KBF2, binds to the same sequence as KBFI, and is present in undifferentiated EC cells. We report here the characterization and purification of KBF2. This knowledge has allowed us to unambiguously show that KBF1 binding activity is absent in undifferentiated EC cells while KBF2 activity is present. There- fore, KBF1 activity appears to be regulated during the differentiation of these cells and may in part be involved in the activation of class I gene expression observed in differentiated EC cells.

MATERIALS AND METHODS Cell lines F9 cells were obtained from Dr. K. Ozato and were grown in DMEM supplemented with 10 % fetal calf serum. They were subcultured every 2 days. The undifferentiated state was regularly checked by immunofluorescence for the presence of the SSEAI antigen. Differentiation was induced by treatment with 3x10-7 M retinoic acid and 10-3 M dibutyryl cyclic AMP. PCC3, PCC4 and 3A1D3 were obtained from Dr. J.F. Nicolas and were subcultured every 2

5246 Nucleic Acids Research days. 3T6 cells were grown in the same medium, while BW5147 cells were grown in RPMI supplemented with 10 % fetal calf serum. Transfection and CAT assays Transfection by the calcium phosphate coprecipitation technique and CAT assays were carried out as described in Kimura et al., (14) with some modi- fications : undifferentiated cells were broken by treatment with 0.05 x Tri- ton X-100 and no freeze-thawing ; cytoplasmic extracts were heated for 10 minutes at 65°C before the CAT assays. Preparation of nuclear extracts Nuclear extracts were prepared as described in IsraEl et al. (15) ex- cept that for undifferentiated cells, 0.05 % Triton X-100 was used and no mechanical disruption was required. Retardation experiments For each binding assay, 0.25 ng of 5'-end labelled oligonucleotide was incubated in a volume of 10 jil with 200 ng of poly(dI-dC) and 3-6 Ujg of nuclear extract in the buffer described in Yano et al. (18). When purified protein was used, the binding reaction contained no poly(dI-dC) but 100 pg/ml of bovine serum albumin. The reaction was incubated for 15 min. at room temperature and then loaded onto a 5 % polyacrylamide gel in 0.5 X Tris-Borate EDTA buffer, which was run for 2 hours at 12 V/cm. The gel was then fixed, dried and exposed for 2-5 hours with Kodak X-AR film. Purification of KBF2 from nuclear extracts of BW5147 cells The fractions giving rise to bands 1 and 3 eluted from the KBF1 affinity column (see Yano et al., (18) for details) at 0.2 M and KBF2 was further purified by repeated passages through an affinity column made with concatenated beta-2-microglobulin oligonucleotide which has the following sequence GATCAAGGGACTTTCCCAT TTCCCTGAAAGGGTACTAG (see Table 1 for details). UV cross-linking experiments Cross-linking was carried out as described in Cereghini et al. (30). The probe was prepared by annealing the two oligonucleotides GATGGGGATTCCCCATCTCCACAGTTTCACTTCTGCA GAGGTGTCAAAGTGAAGACGTCT followed by elongation using the Klenow enzyme with 0.1 mM BudR, dATP, dGTP and 5 iM alpha-32P-dCTP. The sequence in common between the two oligonucleo-

5247 Nucleic Acids Research

tides represents the interferon response sequence of the H-2Kb gene and does not bind KBF1 or KBF2. The probe was incubated with purified KBFI, KBF2 or BW5147 crude nuclear extract and a retardation gel was run as described above. The gel was irradiated with a 312 nm UV light lamp for 30 min. at 4°C and autoradiographed. The retarded bands were excised, incubated for 10 min. at 650C in SDS-beta-mercaptoethanol protein sample buffer, and loaded directly in the slots of a 10 % SDS-polyacrylamide gel, which was then fixed, dried and autoradiographed. Methylation protection experiments For DMS protection experiments, a double-stranded oligonucleotide including the binding site for KBF1 and KBF2 (see legend to figure 3) was labelled at the 5' end of one strand and incubated with purified KBF1 or KBF2 in a 5 times scaled up binding reaction. After 20 min. at room temperature, 1 ll of DMS was added for 5 min. After phenol extraction and ethanol precipitation, the samples were analysed on a 20 % polyacrylamide 8 M urea gel.

RESULTS 1.- Detection and purification of a new factor, KBF2, that binds to the enhancer of the Kb gene In the course of KBFI purification we performed bandshift assays using as a probe, a short 18 bp palindromic sequence derived from the H-2Kb enhancer called KBF (cf. legend to Figure 1). By using crude nuclear extracts derived from various differentiated murine cells, we could detect three specific retarded bands (bands 1, 2 and 3 in Figure IA) which were all competed by an excess of cold homologous oligonucleotide (not shown). Star- ting from a crude nuclear extract derived from the BW5147 T cell line (18), the protein(s) responsible for these three bands copurified through the first conventional columns used in the purification of KBFl (18 : hydroxyapatite, heparine agarose, gel filtration). However, bands I and 3 were obtained with the 0.2 M NaCl eluate from the DNA affinity column used as a last purification step (this column is made of concatenated KBF oligonucleotide bound to Affigel 15). In contrast, the activity responsible for band 2 was eluted between 0.6 M and 1 M NaCl and constituted the 48 kd KBFl protein (18). We decided to purify further the activity corresponding to bands 1 and 3 (Table 1). Our strategy was based on the following observation : using as a probe, an 18 bp oligonucleotide (beta-2-microglublin) (cf. Materials and

5248 Nucleic Acids Research

A 1 2 3 B

,. band 1 band 2 1 2 3 S band 3 92 M i 30 ,__ -4621

Figure 1 A) Gel retardation assay using the H-2Kb derived 18 bp palindromic sequence as a probe. The H-2Kb derived KBF oligonucleotide GATCTGGGGATTCCCCAT ACCCCTAAGGGGTACTAG was 5' end labelled and 0.25 ng was incubated with crude extracts (3-6 Vjg) or purified fractions (0.5 pl) as described in (18), but with 0.2 pg poly-dIdC as a non-specific competitor. Lane 1 : purified KBF1 (18). Lane 2 : 0.20 M NaCl eluate from the KBF1 affinity column. Lane 3 : nuclear extract derived from BW5147 mouse T cells. B) Silver stain of an SDS-polyacrylamide gel of purified KBF1 and KBF2. KBF1 was purified as described in (18) including, as a final step, chromatography on an affinity column made with concatenated KBF oligonucleotide. KBF1 eluted at about 0.6 M NaCl. The activity responsible for bands 1 and 3 as determined by the bandshift assay, was purified as described in the Materials and Methods section. 50 Ill of the most purified peak fractions giving rise to band 2 (lane 1) or bands 1 and 3 (lane 2) were TCA precipitated, resuspended in SDS sample buffer without beta-mercaptoethanol and loaded onto a 10 % SDS- polyacrylamide gel. Lane 1 : KBF1 ; lane 2 ; KBF2 ; lane 3 : MW markers.

Methods) derived from the beta-2-microglobulin sequence, which also binds KBF1 (15), we observed a much stronger band 3 than by using the H-2Kb derived probe. We thus passed the pooled fractions from the KBF affinity column (Table 1) corresponding to bands 1 and 3 through a DNA affinity column made of concatenated beta-2-microglobulin derived oligonucleotide.

5249 Nucleic Acids Research

TABLE 1 Total Totala Specificb Volume protein activity activity Yield Purificatior (ml) (mg) (ng) (ng/ug) (%) (fold)

Crude extract 25 80 800B 0.1 100 1 Hydroxyapatite 5 6.15 6460 1.05 81 1C KBF affinity 7.3 0.440 2904 6.6 36 66 column 0.2 M Beta-2 1st affinity 5.0 0.037 2400 64.9 30 649 Beta-2 2nd affinity 5.0 0.020 2020 101 25 1010 Beta-2 3rd affinity 4.0 0.005 1440 288 18 2880 a : Total activity was obtained by multiplying the specific activity by the total amount of protein (in ujg). b : Specific activity is defined as the nanogram amount of labelled DNA (KBF oligonucleotide) bound per ,;g of protein, which was determined by counting radioactivity contained in the specific retarded bands (1 + 3) excised after gel retardation. The KBF affinity column was made by concatenating and coupling to Affigel 15 the KBF oligonucleotide: GATCTGGGGATTCCCCAT ACCCCTAAGGGGTACTAG as described in Yano et al. (18). The beta-2 affinity was made by concatenating and coupling to Sepharose the beta-2 oligonucleotide: GATCAAGGGACTTTCCCAT TTCCCTGAAAGGGTACTAG. The KBF2 activity eluted around 0.25 M NaCl from this column.

The activities responsible for bands 1 and 3 coeluted around 0.25 M NaCl. After three passages, SDS gel analysis showed the presence of a 58 kd protein (Figure IB) which coeluted with the binding activity and which we called KBF2. The protein was purified about 3000 fold with an 18 % yield by this procedure. We estimate that KBF2, in the cell line from which it was purified, is present in similar amounts to that of KBF1 (18). 2.- Nature of the two retarded bands associated with KBF2 To resolve the relationship between KBF2 and the two retarded bands, we carried out titration experiments as well as UV-cross-linking experiments. Figure 2A, right panel, shows that at low concentration of purified protein, only band 3 is present, while at higher concentration, band 1 appears. Figu- re 2B shows that UV-cross-linking detects a 48 kd protein associated with band 2, corresponding to KBF1, while bands 1 and 3 are associated with the same 58 kd protein. These observations suggest that bands I and 3 correspond to the same protein (KBF2) initially detected after the affinity column. Owing to the palindromic nature of the binding site, we tested the possibility that bands I and 3 might correspond to dimer and monomer forms of KBF2.

5250 Nucleic Acids Research

A B

a b c d .2.5 1 .2.5 1

*,,@_ ~~~~~~~92 .0109~~~:*B 69 ow ~~~~46

Figure 2 A) Gel retardation assay using purified KBF2. Numbers above each lane indicate the amount of KBF2 added (in 1-l). The oligonucleotide used in the right panel (palindromic site) is KBF. The oligonucleotide used in the left panel (half site) has a double transversion which prevents KBF1 binding GATCTGGGGATTCCGGAT ACCCCTAAGGCCTACTAG B) Autoradiogram of an SDS-polyacrylamide gel showing UV-crosslinking of the KBF sequence to the proteins giving rise to bands 1, 2 and 3. Cross-linking has been carried out as described in Materials and Methods. The probe was incubated with purified KBF1, KBF2 or BW5147 crude extracts and a retardation gel was run as described in the legend to Figure 1. The gel was irradiated and autoradiographied. Bands 1, 2, 3 or 2 and 3 together were excised, incubated for 10 mmn at 65% in SDS- beta-mercaptoethanol protein sample buffer, and loaded directly in the slots of a 10 % SDS-polyacrylamide gel, which was then fixed, dried and autoradiographed. Lane a band 2 obtained with purified KBF1. Lane b band 3, lane c band 1, obtained with purified KBF2 ; lane d bands I and 2 together, as obtained with crude extract (cf. Figure lA lane 3). MW (KD) are shown on the right. The doublet appearance of KBF2 has been found reproducibly. We are cur- rently investigating the possible differences between these two species.

We synthesized an oligonucleotide with a mutation that prevents both the binding of KBF1, as well as its enhancer activity TGGGGATTCCGGA. Bandshift assay using this probe detected only band 3 (Figure 2A, left panel). A similar result was obtained with an oligonucleotide which includes only the right half of the palindrome ATTCCCCA... (not shown). These results, in addition to the respective migration rates of the two complexes (see below), suggest that KBF2 can bind either as a monomer on half the palindrome or as

5251 Nucleic Acids Research

Figure 3 DMS protection experiment using purified KBF1 or KBF2. We used the following oligonucleotide : 5' GAATGGGGAATCCCCAGCCCTGGGC 3' 3' TACCCCTTAGGGGTCGGGACCCGCT* 5' where the binding site is underlined. The lower strand was labelled with gamma-ATP at the 5' end. 5 ng of double-strand KBF oligonucleotide was incubated with 10 pl of KBF1 (lane 2) or 10 pl of KBF2 (lane 3) or with serum- albumin only (lane 1). The assay was run as described in Materials and Methods and the samples analysed on a 20 % polyacrylamide 8 M urea gel.

a dimer to the entire palindromic sequence. The fact that UV-cross-linking experiments with band 1 detect only the monomer form of KBF2 is probably due to lack of cross-linking between the 2 protein subunits after UV-treatment. If one admits that migration of a protein-DNA complex in a bandshift assay varies with the logarithm of the molecular weight of the protein part (20, 21), the assumption that the MW associated with band 1 is 116 Kd (KBF2 dimer) and with band 3 is 58 Kd (KBF2 monomer) leads to a calculated molecular weight of 95-100 Kd for band 2 (data not shown) consistent with the binding of KBF1 as a dimer. This tentative conclusion is also supported by the fact that KBF1 does not bind to the mutated oligonucleotide described above although formal proof of such dimerisation requires more thorough investiga- tion. These results also suggest that mutations which abolish KBF1 binding also prevent binding of dimeric KBF2. DMS protection experiments (figure 3) indicate that KBF1 and KBF2 contact the same residues in the binding site (the four G's on each strand : 15, 17, 19).

5252 Nucleic Acids Research

1 2 3 4 5 67 8

band 1... band 2_

band 3 -i

Figure 4 Gel retardation assay. The probe is the KBF oligonucleotide. Differentiation of F9 cells was induced by treatment with 3x10-7 M retinoic acid and 10-3 M dibutyryl cyclic AMP. Retardation assays were carried out as described in the legend to Figure 1. Similar amounts of nuclear extract were used in each lane, derived from the following cell lines : Lane 1: BW5147 mouse T cells. Lane 2 3T6 mouse fibroblasts. Lane 3 : undifferentiated F9 cells. Lane 4 F9 cells, 2 days after induction of differentiation. Lane 5 F9 cells after 4 days of differentiation. Lane 6 : PCC4 cells. Lane 7 PCC3 cells. Lane 8 3A1D3 cells (a differentiated derivative of PCC3).

3.- Absence of KBF1 in undifferentiated EC cells The lack of enhancer activity of the H-2Kb enhancer in undifferentiated EC cells (14) prompted us to investigate the presence of KBF1 and KBF2 activities in these cells. Figure 4 shows that in nuclear extracts derived from three undifferentiated EC cells (F9, PCC3 and PCC4), only bands 1 and 2 corresponding to KBF2 can be detected (lanes 3, 6, 7). However, induction of differentiation of F9 cells by treatment with retinoic acid and dibutyryl- cyclic AMP is followed by the progressive appearance of KBF1 binding activity. Similarly, the line 3A1D3 (lane 8) which is a differentiated derivative of PCC3, exhibits KBFl binding activity. Differentiated cells like BW5147 thymoma (lane 1) and 3T6 fibroblasts (lane 2) exhibit variable amounts of KBF2, suggesting that the amount of this factor does not correlate with the state of differentiation. On the other hand, all cell lines tested so far which express class I genes (Figure 4, lanes 1, 2 ; 15, 19 ; A. Israel, unpublished results) exhibit KBF1 activity. To correlate the induction of KBF1 binding activity in differentiating teratocarcinoma cells with a functional assay, we cloned either one or three copies of the KBF oligonucleoti-

5253 Nucleic Acids Research

TABLE 2 F9 RA4D 3T6

conaCAT 1 1 1 (KBF) conaCAT 1 3.2 8.5 (KBF)3 conaCAT 0.9 4.5 20

(KBFm)3conaCAT 1.1 1.0 1.1 Effect of the wild-type or mutated KBF oligonucleotide on the activity of an heterologous promoter in undifferentiated or differentiated cells. The CAT activities were determined in a transient assay, 40 hrs after transfection, as described in Kimura et al. (14). For a given cell line, variations in transfection efficiency were normalised by cotransfec- tion with the RSV-beta-gal plasmid. conaCAT is described in Kimura et al. (14). One or three copies of the KBF or KBFm (GATCTGGGGATTCCGGA ACCCCTAAGGCCTCTAG) oligonucleotides were cloned in the BamHl site located at -102 bp relative to the conalbumin transcription start. Values indicated are the average of at least 4 experiments and variations never exceeded 25 %. RA4d indicates F9 cells treated from 2 days before transfection with 3x10-7 M retinoic acid and 10-3 M dibutyryl cyclic AMP.

des or three copies of the KBFm oligonucleotide described above (TGGGGATTCC- GGA : binds only the putative KBF2 monomer) in the conaCAT expression vector previously used for characterizing H-2Kb enhancer activity (14). Table 2 shows that (KBF)conaCAT and (KBF)3conaCAT are inactive in undifferentiated F9 cells, but are active in differentiated cells. As reported for other protein binding sites, polymerisation of the oligonucleotide increases transcription efficiency. In correlation with the results obtained in figure 4, induction of differentiation of F9 cells by retinoic acid and dibutyryl cAMP results in increased activity of the (KBF)conaCAT and (KBF)3conaCAT cons- tructs. On the other hand, the (KBFm)3conaCAT construct containing a mutated oligonucleotide which does not bind KBF1 nor the putative KBF2 dimer is inactive in every cell type tested. These results establish a good correlation between the presence of KBF1 binding activity and enhancer activity of the sequence.

DISCUSSION We report here that, in nuclear extracts derived from differentiated mouse cells, two protein factors, KBF1 and KBF2, which bind to the same palindromic sequence in the enhancer region of both the H-2 class I and beta-2-microglobulin genes can be detected. KBF1 had been previously puri-

5254 Nucleic Acids Research fied and shown to have a molecular weight of approximately 48 Kda. Purified KBF2 has an approximate molecular weight of 58 Kda. Our experiments suggest that KBF2 could bind as either a monomer or a dimer depending on its concentration, while only one form of bound KBF1 is detectable. Experiments shown in figure 2 suggest that this form is dimeric but a more thorough investiga- tion is required to unambiguously establish this point. The possibility exists that band 1 could be due to association of KBF2 with a copurifying factor. However glycerol gradient analysis showed only one peak of KBF2 activity, giving rise to both complexes 1 and 3 (not shown). This is an indication that band I consists only of a DNA-KBF2 complex. This also suggests that the putative dimer cannot form in the absence of the DNA binding site. These data allowed us to interpret the unexpectedly complex situation observed in undifferentiated and differentiated (or differentiating) EC cells. In the undifferentiated state, KBF2 activity is present, while KBF1 binding activity is undetectable. When EC cells differentiate, KBF1 binding activity appears. Our major conclusion, therefore, is that the absence of KBF1 binding activity is, at least in part, responsible for the lack of enhancer activity and lack of H-2 genes expression in undifferentiated EC cells. At present, we cannot say whether the KBF1 gene is silent in undifferentiated cells and is turned on after differentiation, or whether this factor is present in an inactive form in EC cells, similar to what has recently been described for NFKB, a B cell specific factor which is present in an inactive form in the cytoplasm of pre-B and non-B cells (22). So far, all attempts to induce KBF1 activity in F9 cells by treatments which do not induce differentiation have been unsuccessful. Incidently, NFKB and KBF1 recognize the same set of sequences but with different affinities (19). The results described here together with a report describing another factor with similar specificity (EBP-1 (23)) indicate that at least four factors can bind to this sequence. The functional complexity of the H-2Kb promoter (14, 24, 25, 26) is such that the developmental regulation of KBF1 activity may not be the only reason for the non-expression of class I genes in EC cells. Tanaka et al. (13) have shown a differential methylation pattern in the 3' region of the H-2Kb gene upon differentiation of F9 cells. Miyazaki et al. (16) have observed that a 35 bp region from the H-2Kb promoter which includes the palindrome studied here can act as a repressor in F9 cells and as an activa- tor in 3T3 cells. We have not been able to demonstrate a negative effect of the palindromic sequence in F9 cells, although this may be due to the fact

5255 Nucleic Acids Research

that the 35 bp region used by Miyazaki et al. (16) contains binding sites for other factors (A. Israel et al., submitted for publication). We are cur- rently analyzing whether other proteins which bind to the H-2Kb promoter in differentiated cells also bind in EC cells. The question remains as to the function of KBF2 and its relationship, if any, to KBF1. It could act as a negative element in F9 cells superseded by KBF1 after differentiation. Its molecular cloning (in progress) will help to answer this question. KBF2 could also represent an inactive precursor form of KBF1 ; again the isolation of cDNA clones encoding KBFI and KBF2 will help to determine whether these two factors are encoded by different genes, or are two products of a single gene. The same question holds for any relationship existing beween EBP1, NFKB and KBF1 or KBF2, and again a defi- nite answer will await isolation of cDNA clones. Potential relationships between KBF1 and NFKB or KBF1 and EBP1 have been discussed already (19, 23) and KBF2 is certainly different from the mature form of NFKB which is absent from non-B cells, and when present, gives rise to a retarded complex exhibiting a mobility different from bands I or 3 seen in figure 1A (data not shown). These results show that during differentiation, the expression of cer- tain genes is, at least in part, regulated by the availability of DNA-binding regulatory proteins which, therefore, appear to be themselves regulated. This has also been shown in undifferentiated EC cells for PEAI (the mouse homologue to API) binding to the polyoma enhancer (27), for the c-fos protooncogene (28), and for AP2 and AP3 (29). These factors are known to regulate a series of different genes. The KBF1 factor described here also binds to the SV40 enhancer (17), to the promoter of the C4 complement gene (M.K., unpublished results), and to the PRDII element of the beta-interferon promoter (A.I., unpublished results). It thus appears possible that diffe rentiation of EC cells is accompanied by the induction of the binding activity of several regulatory proteins, the cloning of which should help in the understanding of the molecular mechanisms underlying embryonic differentiation.

ACKNOWLEDGEMENTS This work was supported by INSERM, Institut Pasteur and the Ligue Nationale Fran9aise contre le Cancer. M.K. was a recipient of a fellowship from the National Cancer Institute of Canada.

5256 Nucleic Acids Research

*Present address: Institute of Biological Sciences, Mitsui Pharmaceuticals Inc., Chiba-ken 297, Japan

REFERENCES 1.- Zinkernagel, R.M. & Doherty, P.C. (1980) Adv. Immunol., 27, 51-177. 2.- Ozato, K., Wan, Y.J. & Orrison, B.M. (1985) Proc. Natl. Acad. Sci., USA, 82, 2427-2431. 3.- Silver, L.M., Martin, G.R. and Strickland, S. (1983) Teratocarcinoma Stem Cells. Cold Spring Harbor Laboratory. 4.- Artz, K. and Jacob, F. (1974) Transplantation, 17, 632-634. 5.- Croce, C.M., Linnenbach, A., Huebner, K., Parnes, J.R., Margulies, D.H., Appella, E. and Seidman, J.G. (1981) Proc. Natl. Acad. Sci., USA, 78, 5754-5758. 6.- Morello, D., Gachelin, G., Dubois, P., Tonigaki, N., Pressman, D. & Jacob, F. (1978) Transplantation, 26, 119-125. 7.- Strickland, S., Smith, K.K. and Marotti, K.R. (1980) Cell, 21, 347- 355. 8.- Gachelin, G. (1978) Biochim. Biophys. Acta, 516, 27-60 (1978). 9.- Morello, D., Daniel, F., Baldacci, P., Cayre, Y., Gachelin, G. & Kourilsky, P. (1982) Nature, 296, 260-262. 10.- Rosenthal, A., Wright, S., Cedar, H., Flavell, R. and Groveld, F. (1985) Nature, 315, 579-581. 11.- GmUr, R., Solter, D. & Knowles, B.B. (1980) J. Exp. Med., 151, 1349-1357. 12.- Andrews, P.W. & Goodfellow, P.N. (1980) Somat. Cell. Genet., 6, 271-280. 13.- Tanaka, K., Appella, E. & Jay, G. (1983) Cell, 35, 457-465. 14.- Kimura, A., Israel, A., Le Bail, 0. & Kourilsky, P. (1986) Cell, 44, 261-272. 15.- Israel, A., Kimura, A., Kieran, M., Yano, O., Kanellopoulos, J., Le Bail, 0. & Kourilsky, P. (1987) Proc. Natl. Acad. Sci., USA, 84, 2653-2657. 16.- Miyazaki, J., Appella, E. & Ozato, K. (1986) Proc. Natl. Acad. Sci., USA, 83, 9537-9541. 17.- Baldwin, A.S. & Sharp, P.A. (1987) Mol. Cell. Biol., 7, 305-313. 18.- Yano, O., Kanellopoulos, J., Kieran, M., Le Bail, O., Israel, A. & Kourilsky, P. (1987) EMBO J., 6, 3317-3324. 19.- Baldwin, A.S. & Sharp, P.A. (1988) Proc. Natl. Acad. Sci., USA, 85, 723-727. 20.- Bading, H. (1988) Nucl. Acids Res., 16, 5241-5248. 21.- Schreiber, E., Mathias, P., MUller, M.M. & Schaffner, W. (1988) EMBO J., 7, 4221-4229. 22.- Baeuerle, P. & Baltimore, D. (1988) Cell, 53, 211-217. 23.- Clark, L., Pollock, R.M. & Hay, R.T. (1988) Genes Dev., 2, 991-1002 24.- Chodosh, L.A., Baldwin, A.S., Carthew, P. & Sharp, P.A. (1988) Cell, 53, 11-24. 25.- Shirayoshi, Y., Burke, P.A., Appella, E. and Ozato, K. (1987) Mol. Cell. Biol., 7, 4542-4548. 26.- Korber, B., Mermod, N., Hood, L. & Stroynowski, I. (1988) Science, 239, 1302-1306. 27.- Kryszke, M.H., Piette, J. & Yaniv, M., (1987) Nature, 328, 254-256. 28.- Ruther, U., Wagner, E.F. & Muller, R., (1985) EMBO J., 4, 1775-1781 29.- Chiu, R., Imagawa, M., Imbra, R., Bockoven, J.R. & Karin, M. (1988) Cell, 54, 541-552. 30.- Cereghini, S., Blumenfeld, M. & Yaniv, M. (1988) Genes Dev., 2, 957-974.

5257