Eur. J. Biochem. 181, 555-561 (1989) 6 FEBS 1989

Evidence that the upstream stimulatory factor and the Spl transcription factor bind in vitro to the promoter of the human-growth-hormone gene Frederic P. LEMAIGRE, Stephane J . COURTOIS, Dominique A. LAFONTAINE and Guy G. ROUSSEAU

Hormone and Metabolic Research Unit, Louvain University Medical School and International Institute of Cellular and Molecular Pathology, Brussels

(Received October 20, 1988/January 2, 1989) - EJB 88 1235

Expression of the human-growth-hormone gene is restricted to pituitary somatotrophs. Two protein-DNA complexes that are specific to the pituitary, and two that are not, had been demonstrated in vitro on the promoter of this gene. The two pituitary-specific footprints had been ascribed to a single protein called growth hormone factor 1. We have now characterized the factors responsible for the two other footprints by means of deoxyribonuclease-I protection and gel-retardation experiments.

The first footprint, located between -257 and -290 relative to the transcription initiation site, involves at least two factors present in pituitary cells. One of these factors binds between nucleotides -257 and - 267, and is indistinguishable from the upstream stimulatory factor, also called major late transcription factor or upstream element factor, initially described in HeLa cells. Earlier work by others had shown that the activator protein 2 purified from HeLa cells can bind to nucleotides - 263 and - 290. Our experiments suggest that a factor different from activator protein 2 is involved in the protection of this region against deoxyribonuclease I.

The second footprint, located between nucleotides - 116 and - 140, involves only one factor. This factor, present in pituitary cells, recognizes a GC box and is indistinguishable from transcription factor Spl, previously described in HeLa cells. The human-growth-hormone gene is therefore a candidate for regulation by these factors in vivo.

In eukaryotes, RNA polymerase I1 catalyzes transcription of DNA into mRNA. The correct positioning of the transcrip- tion initiation (cap) site and the efficiency of transcription by the polymerase depend on complex interactions between regulatory factors and defined DNA sequences, respectively referred to as trans-acting factors and cis-acting elements. Several cis-acting elements are common to many genes, whereas others are restricted to particular genes. Some trans- acting factors appear to bind to many genes, while others bind only to particular genes. The latter family of factors includes the upstream stimulatory factor (USF) also known as major late transcription factor (MLTF) or upstream element factor [l - 31, and the Spl transcription factor also known as the GC-box-binding protein [4 - 71. Furthermore, the expression of some genes is tissue-specific, which is frequently the result of interactions between tissue-specific factors and gene- specific cis-acting elements (for a review, see [8, 91).

The human-growth-hormone (hGH) gene is expressed only in pituitary somatotrophs, and the promoter of this gene

Correspondence to G. G. Rousseau, UCL-ICP, Box 7529, Avenue Hippocrate 75, B-1200 Bruxelles, Belgium

Abbreviations. hGH, human growth hormone; USF, upstream stimulatory factor; Ad-2 MLP, adenovirus 2 major late promoter; DNase I, deoxyribonuclease I; GHF2 and GHF3, growth-hormone footprints 2 and 3; AP-2, activator protein 2; GHF-1, growth hor- mone factor 1.

Enzymes. Restriction endonucleases BamHI, EcoRI, PvuII, XhoI (EC 3.1.21.4); T4 polynucleotide 5’-hydroxyLkinase (EC 2.7.1.78); DNA-directed DNA polymerase, Klenow fragment (EC 2.7.7.7); deoxyribonuclease I (EC 3.1.21.1).

contains binding sites for ubiquitous trans-acting factors and for a pituitary-specific trans-acting factor called growth hor- mone factor 1 [lo, 111. In vitro protection of the hGH-gene promoter by extracts from growth-hormone-producing cells (pituitary GC cells) against deoxyribonuclease I (DNase I) digestion shows two “footprints” due to growth hormone factor 1. Two additional footprints are seen with GC-cell extracts but also with extracts from cells (HeLa) that do not express the hGH gene [lo]. These two footprints due to ubiquitous factors are called here growth hormone footprints 2 and 3 (GHF2 and GHF3). Since the number and identity of the factors involved in GHF2 and GHF3 remained un- known, we have approached these questions by in vitro DNase-I protection and gel-retardation experiments. We provide here evidence that GHF3 is due to at least two factors, one of which is indistinguishable from USF, and that GHF2 is due to only one factor which is indistinguishable from Spl.

MATERIALS AND METHODS Materials

Enzymes were purchased from Promega Biotec (Leiden, The Netherlands) or Boehringer (Mannheim, FRG). Radio- labeled compounds were from Amersham International plc. (Little Chalfont, UK). Synthetic oligodeoxynucleotides were from Eurogentec (Likge, Belgium). Polyvinylalcohol and poly(d1-dC) were from Sigma (St. Louis, MO, USA). Poly(A) was from Pharmacia (Uppsala, Sweden). Tissue culture components were from Flow Laboratories (Brussels, Belgium) or from Gibco BRL (Ghent, Belgium).

556

Table 1. Comparison of the hGH-gene-promoter sequences protected against DNase I by GC- and HeLa-cell extracts, with the binding site of USF on the Ad-2 MLP and with the consensus for Spl binding Mismatches are underlined

Site Sequence

hGH : GHF3 (-2SO)ACCATGGCCTGCGGCCAGAGGGC&CCACGTGACC(-256) Ad-2 MLP: USF binding (-63)@CCACGTGACC(-52) hGH : GHF2 Spl consensus

(-140)TGTGTGGGAGGAGCTTCTAAATTAT(-ll6) GG GGGC T A ~ ~ C ~ T A A T

Cells and cell extracts

Human cervical carcinoma (HeLa S3) and rat pituitary tumor (GC) cells were grown in modified Eagles's medium for suspension cultures, supplemented, respectively, with 10% newborn calf serum or with 5% foetal calf serum plus 12.5% horse serum. Whole-cell extracts (10 - 20 mg protein/ml) were prepared as described by Manley [12].

Deoxyribonuclease (DNase) I jootprinting

This was performed essentially according to Briggs et al. [5]. 1-5 ng DNA, end-labeled by [ U - ~ ~ P I ~ G T P with the Klenow fragment of DNA polymerase I, was incubated with the cell extract (15- 100 pg protein) in a final volume of 50 p1 containing 10 mM Tris (pH 7.9), 50 mM KCI, 6.25 mM MgCI2,0.05 mM EDTA, 1 mM dithiothreitol, 8.5% glycerol, 2% polyvinylalcohol and 1 pg poly(d1-dC). After 15 min at O'C, followed by 2 rnin at 20°C, 50 pl of a solution of 5 mM CaClz plus 10 mM MgClZ were added. Digestion with 25- 100 mU of DNase I (Boehringer) was then allowed to proceed at 20°C for 1 min. The reaction was stopped with 100 p1 1% SDS, 200mM NaC1, 20mM EDTA. The DNA was then phenol-extracted, ethanol-precipitated with poly(A) as a car- rier, washed with 70% ethanol, vacuum-dried and loaded on an 8% polyacrylamide, 42% urea sequencing gel. For competition footprinting assays the cell extracts were preincu- bated for 15 min with 1 pg poly(d1-dC) and various amounts of double-stranded oligodeoxyribonucleotides in the same buffer as above, before the addition of end-labeled DNA fragments. Binding reactions were then allowed to proceed for 15 min before DNase I digestion.

Gel-retardation assay

This was performed essentially as described by Carthew et al. [2]. Complementary single-stranded oligodeoxy- ribonucleotides were hybridized for 5 min at 95°C and pro- gressively cooled to room temperature in 10 mM Tris (pH 7.9), 1 mM EDTA and 100 mM NaCl. The resulting double-stranded oligodeoxyribonucleotides were labeled with [Y-~~PIATP using T4 polynucleotide kinase. 0.1 -0.5 ng radiolabeled probe, without or with competing oligonucle- otide, was incubated with the cell extract (1 -10 pg protein) in a final volume of 20 pl of a solution (pH 7.9) containing 12 mM Hepes, 10 mM Tris, 60 mM KCl, 6.25 mM MgC12, 0.6 mM EDTA, 1 mM dithiothreitol, 12% glycerol and 1 pg poly(d1-dC). After 30 rnin at room temperature, the samples were loaded onto a low-ionic-strength, 5% polyacrylamide

Fig. 1. DNase Iprotection on the hGH-genepromoter. The DNA probe consists of a 330-bp-long EcoRI - BarnHI fragment of phGH494d - 256/ - 185 A - 148/ - 56, labeled at the BamHI end of the sense strand. The probe was incubated without (lane 1) or with 45 pg GC (lanes 2-4) or HeLa (lanes 5-7) cell-extract protein. A 200-fold excess (30 ng) of unlabeled oligo AP-2 and/or oligo USF was present in the incubations as indicated above the lanes

gel (acrylamide/bisacrylamide, 80 : 1). Gels were preelectro- phoresed, run and autoradiographed according to [2].

Plasmid constructions

Plasmid phGH494 contains a 496-bp EcoRI - BarnHI (-494 to + 2) insert of the hGH-1 (i.e. hGH-N) gene in pBR322. Plasmid phGH494d - 256/ - 185 A - 148/ - 56 is phGH494 with deletion of nucleotides -256 to -185 and -148 to -56. Plasmid pBRBalI-E is pBR322 containing the 2.4-kb BalI - E fragment of adenovirus 2 that includes the adenovirus 2 major late promoter.

557

Fig. 2. DNase I protection on the Ad-2 MLP. A 291-bp-long XhoI- PvuII fragment of pBRBalI - E, labeled at the PvuII end of the anti- sense strand, was incubated without (lane 1) or with 45 pg (lanes 2 and 4) or 90 pg (lane 3) of GC-cell extract protein. The incubation of lane 4 contained a 500-fold excess (100 ng) of unlabeled oligo USF

RESULTS Factors involved in growth hormone footprint 3 (GHF3)

The growth hormone footprint 3 (GHF3) produced by GC- or HeLa-cell extracts on the hGH-gene promoter is located between nucleotides -257 and -290 relative to the cap site [lo]. Because of its length we thought that GHF3 was produced by more than one factor. One of them could be USF since the sequence of the proximal moiety of this footprint resembles the binding site (Table 1) for USF detected by methidiumpropyl/Fe’ +/EDTA footprinting on the adeno- virus 2 major late promoter (Ad-2 MLP) [l -31.

To test this hypothesis we performed a DNase-I protec- tion experiment with GC-cell extracts or HeLa-cell extracts on the hGH-gene promoter, in the absence or in the presence of a competing double-stranded oligonucleotide (Fig. 1). The sequence of this oligonucleotide, 5’- GTAGGCCACGTGACCGGG-3’ (oligo USF), was chosen to correspond to the nucleotide -49 to -66 USF sequence protected against DNase I on the Ad-2 MLP [l]. In the ab- sence of oligo USF, a typical GHF3 was seen, which extended from nucleotides -257 to -290 as expected (Fig. 1, lanes 2 and 5). In the presence of oligo USF, the extent of GHF3 on the hGH-gene promoter was reduced to the sequence between nucleotides -268 and -290 (Fig. 1, lanes 3 and 6). This is 11 bp shorter than the footprint observed in the absence of oligo USF. We conclude from this experiment that the region spanning from nucleotides - 257 to - 290 binds at least two factors. One of them interacts with a sequence of the hGH- gene promoter (nucleotides -257 to -267) which is similar

Fig. 3. Detection of factor - oligo- USF complexes by gel-retardation assay. Labeled oligo USF was incubated without (lane 1) or with 3 kg HeLa(lanes2-4) orGC(lanes 5-7)celLextract protein. Theamount of competing unlabeled oligo USF or unrelated oligo GHF-1 is indi- cated above the lanes. The arrow points to the specific factor-DNA complexes

to that for USF on Ad-2 MLP. This factor is also able to recognize an oligonucleotide identical to the binding sequence of USF on the Ad-2 MLP.

The USF factor was originally described in HeLa cells. Since it is known that different factors can bind to the same DNA sequence [13 - 151, we could not exclude the fact that GC cells may contain a factor different from USF but able to recognize the USF-binding sequence. Against this possibility, we found that GC-cell extracts produced a typical USF foot- print on the Ad-2 MLP which was prevented by oligo USF (Fig. 2) . A second argument in favor of the presence of USF in GC cells stemmed from gel-retardation experiments [16]. We incubated labeled oligo USF with GC-cell extracts or with HeLa-cell extracts in parallel, and examined the complexes obtained. Only one specific oligo-USF -protein complex was seen with each extract (arrow in Fig. 3, lanes 2 and 5). A 50- fold excess of unlabeled oligo USF inhibited the formation of this protein-DNA complex (Fig. 3, lanes 3 and 6). In contrast, an unrelated oligonucleotide (oligo GHF-1 which is identical to the target sequence for growth hormone factor 1, see below) was unable to compete for the formation of this oligo-USF - protein complex (Fig. 3, lanes 4 and 7). Still, this unrelated oligonucleotide was in a form competent for producing a complex of its own with its cognate factor (data not shown). Thus, the oligo-USF - protein complex detected with both HeLa- and GC-cell extracts is specific. The other, less re- tarded, band seen in Fig. 3 was considered as nonspecific, since it was not consistently seen and was not competed for by unlabeled oligo USF. The differences in intensity of the specifically retarded bands between the HeLa- and GC-cell extracts probably reflect different relative concentrations of the protein that recognizes oligo USF in the two extracts. Treatment of the GC- and HeLa-cell extracts at 70°C for 5 min prior to incubation with labeled oligo USF eliminated

558

Fig. 4. Detection of’fuctor - oligo-AP-2 complexes by gel returdution. Labeled oligo AP-2 was incubated without (lane 1 ) or with 15 bg GC (lane 2) or HeLa (lanes 3-5) cell-extract protein. The amount of competing unlabeled oligo AP-2 or oligo USF is indicated above the lanes. The arrow points to the AP-2-DNA complex

this nonspecific band but did not affect the specific oligo- USF - protein coinplex seen by the gel-retardation assay (data not shown). This is consistent with the known thermostability of USF [l]. Finally, the extent of retardation of the oligo- USF- protein complex was identical with GC- and HeLa-cell extracts. Purified USF has been recently shown to consist of two polypeptides with respective M , of 43000 and 44000. Although the corresponding protein-DNA complexes ex- hibited different mobilities by gel-shift assay using DNA fragments longer than 200 bp, they did not with a 30-bp oligonucleotide [17], which is consistent with our data.

We conclude that GC and HeLa cells contain a factor that is able to recognize the USF-binding site on Ad-2 MLP as well as on the hGH-gene-promoter sequence. By our assays, the factor detected in GC (growth-hormone-producing) cells is therefore indistinguishable from the USF previously de- scribed in HeLa cells.

Imagawa et al. 1181 have shown that the activator protein 2 (AP-2) purified from HeLa cells can protect the remaining part of GHF3, namely nucleotides -268 to -290, against DNase I digestion. We therefore determined whether an oligonucleotide corresponding to the high-affinity AP-2-bind- ing site on the human metallothionein IIA gene

5’-ATGAACTGACCGCCCGCGGCCCGTGCATG-3‘ 3’-CTTGACTGGCGGGCGCCGGGCAC -5 ’

could prevent GHF3. This was not the case (Fig. 1, lanes 4 and 7). This oligonucleotide was functional since in gel- retardation experiments it was specifically retarded by HeLa- cell extracts, as expected from the presence of AP-2 in these cells (Fig. 4, lanes 3 - 5). No such specific retardation (arrow in Fig. 4) was seen with GC-cell extracts (Fig. 4, lane 2). These results suggest that the GC-cell extracts which produce GHF3 by footprinting do not contain AP-2 and that the factor which binds to the hGH gene at position - 268 to - 290 is not AP-2.

Fig. 5. DNuse-I protection experiment on the hGH-gene promoter. A 494-bp-long EcoRI - BumHI fragment of plasmid phCH494 was labeled at the BumHI end of the sense strand, and incubated without (lane 1) or with (lanes 2-4) 45 pg HeLa-cell-extract protein. Lanes 3 and 4 correspond to incubations performed in presence of 80 ng concatemerized and 380 ng nonconcatemerized oligo Spl, respec- tively

Factors involved in growth hormone,footprint 2 (GHF2)

By DNase-I protection, GHF2 has been located on the hGH-gene promoter between nucleotides - 116 and - 140 [lo]. GHF2 overlaps partially with the footprint (-106 to - 131) generated by the binding of growth hormone factor 1 on its distal binding site. The central part of the DNA se- quence corresponding to GHF2 is very similar to the so- called GC box, i.e. the binding-site consensus sequence of transcription factor Spl 1191 (Table 1). This similarity, and the presence in GC-cell extracts of a factor that produces typcial Spl footprints on the herpes simplex virus thymidine-kinase- gene promoter [lo], led us to propose 1201 that GHF2 or a part of it might be due to the Spl factor.

To test this hypothesis, we performed, on the hGH-gene promoter, a DNase-I footprinting experiment with HeLa-cell extracts, since Spl was initially described in these cells. The experiment was performed in the absence or in the presence of a competing double-stranded oligonucleotide (oligo Spl) whose sequence 5’-GGGGCGGGGC-3’ corresponds to the high-affinity consensus sequence reported for Spl [6]. In the absence of competitor, the expected GHF2 was observed (Fig. 5, lane 2). Still, GHF2 persisted in the presence of the competing Spl decanucleotide (lane 4). However, when the Spl decanucleotide was polymerized, the resulting concatemers were very efficient in preventing GHF2 (lane 3). The need for concatemerization of oligo Spl for observing its binding to Spl has been reported previously [19, 211.

559

Fig. 6. DNase Iprotection on the hGH-genepromoter. The same probe as in Fig. 4 was incubated without (lane 1) or with 30 pg GC-cell- cxtract protein (lanes 2-6). The amount of competing unlabeled oligo GHF-1 and of concatemerized oligo Spl present during the incubation is indicated above the lanes. The black box delineates GHF2. White boxes delineate the two binding sites of growth hormone factor 1

To see if similar results were obtained with extracts from growth-hormone-producing cells, we repeated the experiment with GC-cell extracts. A typical GHF2 could not be observed with these extracts unless binding of growth hormone factor 1 on the hGH-gene promoter was prevented (Fig. 6, lanes 3-5) by a competing oligonucleotide. The se- quence of this oligonucleotide (oligo GHF-1) was 5’- GATCCCATGCATAAATGTACACAG-3’ and corresponds to the proximal binding site (nucleotides -55 to -90) of growth hormone factor 1 on the hGH-gene promoter. There- fore, to investigate whether GHF2 detected in the presence of GC-cell extracts is due to Spl we had to perform footprinting experiments in the presence of oligo GHF-1 to permit visual- isation of GHF2. Fig. 6, lane 6 shows that the nucleotide - 116 to - 140 footprint, namely GHF2, seen with GC-cell extracts thanks to the presence of oligo GHF-1 was totally prevented by concatemerized oligo Spl .

We conclude from these experiments that GHF2 is due to a single factor which is able to recognize the Spl-binding- site consensus sequence, and that binding of this factor on the hGH-gene promoter is inhibited by growth hormone factor 1.

Gel-retardation experiments were then performed to compare the factors of HeLa- and GC-cell extracts capable

Fig. 7. Detection of ,factor - oligo-Spl complexes by gel-retardation assay. A labeled 30-bp-long concatemer containing three Spl -binding sites was incubated without (lane 1) or with 15 pg HeLa (lanes 2-4) or GC (lanes 5-7) cell-extract protein. The amount of competing unlabeled unrelated oligo GHF-1 or of concatemerized oligo Spl present during the incubation is indicated above the lanes. Arrows point to the specific factor-DNA complexes

of binding the Spl consensus sequence. A 30-bp-long con- catemer containing three oligo Spl repeats was purified to obtain a probe that had the minimal length to bind Spl in gel-shift experiments [19, 211. Samples were analyzed after incubation of GC- or HeLa-cell extracts with this 30-bp oligo Spl. Two specifically retarded bands (arrows in Fig. 7) were observed when the Spl probe was incubated with either type of extract. The additional retarded bands resulted from nonspecific interactions since they did not disappear by com- petition with an excess of unlabeled concatemerized oligo Spl (Fig. 7, lanes 3 and 6). There was no qualitative difference between the specific mobility shifts generated by the incu- bation with GC- or with HeLa-cell extracts. Thus, the factors involved in formation of the complexes were indistinguishable by these criteria. Spl reportedly occurs as two polypeptides, one of 95 kDa and one of 105 kDa [5]. This could explain our finding of two specific complexes within each incubation. Alternatively, these two types of complexes with oligo Spl might correspond to two conformations since the Spl protein could bind as a monomer or as a multimer to a probe contain- ing three binding sites. Indeed, Spl purified to near homo- geneity from HeLa cells displays multiple protein-DNA inter- actions [5]. Quantitative differences between the specifically retarded bands observed with HeLa- and GC-cell extracts probably reflect a higher relative concentration of Spl in GC- cell extracts. We conclude from these experiments that GC cells contain a factor that is identical to Spl or indistinguish- able from this protein by our biochemical and biophysical criteria, and that it binds to the hGH-gene promoter.

560

DISCUSSION

We have investigated here, by DNase-I footprinting and gel-retardation assay, the interactions of factors with defined regions of the hGH-gene promoter. We have concluded from these experiments that the footprint located between nucleotides -257 and -290 (GHF3) is due to at least two factors present in pituitary-cell extracts. One of these factors was indistinguishable from USF described previously in HeLa cells. As to the other footprint located between nucleotides -116 and -140 (GHF2), it was due to one factor present in pituitary cells. This factor was indistinguishable from Spl and its binding on the hGH-gene promoter was inhibited by the pituitary-specific growth hormone factor 1.

USF had been identified in HeLa cells. It is therefore interesting that we have found an USF-like factor in pituitary cells also. Peritz et al. [22] showed recently that a factor present in HeLa cells binds to the hGH-gene promoter between nucleotides - 257 and - 275. The binding of this factor on the hGH-gene promoter was prevented by competing amounts of a 450-bp-long Ad-2 MLP fragment that includes the USF- binding site. They claimed to have detected a similar factor in GC cells but did not investigate whether the same or different factors recognize the same DNA sequence. The known targets of USF included the Ad-2 MLP [l -31 and the promoters of the y-fibrinogen [23] and metallothionein-I [24] genes. Our data extend these observations by showing that an USF-like factor can also bind to the hGH-gene promoter. Concerning the function of USF, this factor was shown to interact cooperatively with the TATA-box-binding factor called TFII D to stimulate transcription from Ad-2 MLP [l]. Likewise, USF might trans-activate the hGH gene. However, this factor binds on the hGH-gene promoter about 200 bp further up- stream than USF does on the Ad-2 MLP. Other functions of USF-like factors were suggested by Bram and Kornberg [25] who purified a centromere-DNA-binding protein related to USF from yeast. The functional significance of the binding of the USF-like factor on the hGH-gene promoter therefore remains to be defined.

We have concluded from our results that GHF3 involves not only an USF-like factor but also at least one other factor. The latter could have been activator protein 2 (AP-2). Indeed, AP-2 purified from HeLa cells was able to bind on the hGH- gene promoter between nucleotides - 263 and - 290, namely that part of GHF3 which is not due to the USF-like factor [18]. Our results from competition footprinting and from gel- retardation experiments lead to the conclusion that a factor distinct from AP-2 and present in both GC and HeLa cells is involved in this part of GHF3. Work is in progress to characterize this factor.

The binding of Spl to GC boxes of several gene promoters has been reported (see [26] for references). We have shown here that Spl or an Spl-like molecule binds to a GC box of the hGH-gene promoter. It is noteworthy that this GC box differs by one nucleotide (Table 1) from the GC-box consensus sequence. The hGH-gene promoter contains another putative GC-box sequence [20] from nucleotides -401 to -410 (5’- TGGGTGGAGG-3‘). Still, we failed to detect any GHF2- like footprint in this region with either HeLa- or GC-cell extracts. This could be due to the two mismatches (underlined) between this hGH-gene sequence and the Spl consensus shown in Table 1. It is known that a single GC box is sufficient for activation of transcription [6]. It is therefore tempting to speculate that the Spl-like factor can stimulate hGH-gene

like factor, the length of this footprint (24 bp) exceeds the Spl decanucleotide consensus sequence. This could result from steric hindrance of DNase I activity or from the presence of Spl-associated proteins. We also found that, under our in vitro conditions, the binding of the Spl-like protein on the hGH-gene promoter is inhibited by the binding of growth hormone factor 1. An association of the latter with the Spl- like factor is unlikely since the band shift seen with oligo Spl was the same whether growth hormone factor 1 was present (GC-cell extracts) or not (HeLa-cell extracts). The relative affinities and concentrations of growth hormone factor 1 and Spl are obviously critical in determining the actual protein- DNA interactions in this region of the hGH-gene promoter. Cell extracts do not necessarily reflect the situation in the intact cell. The respective contribution of these two factors to the control of hGH-gene transcription therefore deserves further investigation.

We thank S. Durviaux for technical assistance, and T. Lambert and M. Marchand for secretarial help. We are grateful to Dr J. Martial (University of Litge) for the hGH-1 clone and to Dr D. Christophe (Free University of Brussels) for providing the oligonucleotide AP-2. F. P. L. and D. A. L. are research assistants of the National Fund for Scientific Research (Belgium). S. J. C. holds a Fellowship from the Institut pour I’Encouragement de la Recherche Scientifique duns I’lndustrie et I’Agriculfure (Belgium). This work was supported by the Fonds de la Recherche Scientzjique Medicale (Belgium), by Stimulation Action Grant ST2J00751 B from the European Economic Com- munity, and by the Belgian State - Prime Minister’s office - Science Policy Programming (incentive program in Life Sciences grant no 20).

REFERENCES 1. Sawadogo, M. & Roeder, R. G. (1985) Cell 43,165- 175. 2. Carthew, R. W., Chodosh, L. A. & Sharp, P. A. (1985) Cell 43,

3. Miyamoto, N. G., Moncollin, V., Egly, J.-M. & Chambon, P.

4. Dynan, W. S. & Tjian, R. (1985) Nature 316, 774-777. 5. Briggs, M. R., Kadonaga, J. T., Bell, S. P. & Tjian, R. (1986)

6. Kadonaga, J. T., Jones, K. A. & Tjian, R. (1986) Trends Biochern.

7. Kadonaga, J. T., Carner, K. R., Masiarz, F. R. & Tjian, R. (1987)

8. Maniatis, T., Goodburn, S. & Fischer, J. A. (1987) Science 236,

9. La Thangue, N. B. & Rigby, P. W. J. (1988) in Transcription and splicing, 1st edn (Hames, B. D. & Glover, D. M., eds) pp. 1 - 42, IRL Press, Oxford, Washington DC

10. Lefevre, C . , Imagawa, M., Dana, S., Grindlay, J., Bodner, M. & Karin, M. (1987) EMBO J . 6, 971 -981.

11. Bodner, M. &Karin, M. (1987) CeN50, 267-275. 12. Manley, J . L. (1984) in Transcription and translation. A practical

approach, 1st edn (Hames, B. D. & Higgins, S. J., eds) pp. 71 - 88, IRL Press, Oxford, Washington DC

13. Pfeiffer, K., Arcangioll, B. & Guarente, L. (1987) Cell 49, 9-18.

14. Weinberger, J., Baltimore, D. & Sharp, P. A. (1986) Nature 322,

15. McKnight, S. & Tjian, R. (1986) Cell46, 795-805. 16. Fried, M. & Crothers, D. M. (1981) Nucleic Acids Res. 9, 6505-

17. Sawadogo, M., Van Dyke, M. W., Gregor, P. D. & Roeder, R.

18. Imagawa, M., Chiu, R. & Karin, M. (1987) CeN51, 251 -260. 19. Westin, G. & Schaffner, W. (1988) Nucleic Acids Res. 16, 5711 -

439 -448.

(1985) EMBO J . 4,3563-3570.

Science 234,47 - 52.

Sci. II,20-23.

CeN.51, 1079-1090.

1237- 1245.

846-848.

6525.

G. (1988) J . Biol. Chern. 263, 11 985 - 11 993.

transcription. Although GHF2 is accounted for by an Spl- 5781.

561

20. Lafontaine, D. A,, Economidis, I. V. & Rousseau, G. G. (1987) Horm. Metab. Res. Suppl. Series 17, 5 - 10.

21. Harrington, M. A,, Jones, P. A., Imagawa, M. & Karin, M. (1988) Proc. Natl Acad. Sci. USA 85,2066-2070.

22. Peritz, L. N., Fodor, E. J. B., Silversides, D. W., Cattini, P. A,, Baxter, J . D. & Eberhardt, N. L. (1988) J . Bid. Chem. 263, 5005 - 5007.

23. Chodosh, L. A,, Carthew, R. W., Morgan, J. G., Crabtree, G.

24. Carthew, R. W., Chodosh, L. A. & Sharp, P. A. (1987) Genes

25. Bram, R. J. & Kornberg, R. D. (1987) Ma/ . Cell. Bid. 7, 403-

26. Wingender, E. (1988) Nucleic Acids Res. 16, 1879 - 1902.

R. & Sharp, P. A. (1987) Science 238,684-688.

Dev. I , 973 -978.

409.