The Human Growth Hormone Gene Contains Both Positive and Negative Control Elements*

A subset of DNA sequences in the 5“flanking DNA of the human growth hormone (hGH) gene was exam-ined by protein-DNA binding and gene transfer- expression experiments. Two adjacent cis-acting elements (I and 11) were identified between nucleotides -308/-235 of the hGH gene that modulated the expression of a linked reporter gene in transfected HeLa cells. Elements I and I1 repressed gene expression whereas alone it. whole cell extracts two factors that bind hGH DNA carrying elements I and 11. Factor I binds to single-stranded DNA, and its binding is correlated with repression of gene expression. Factor I1 binds between nucleotides -275/ -257 of the hGH gene. This region is homologous to the binding site for the adenovirus major late tran- scription factor, and factor I1 binding to hGH DNA is competed by adenovirus major late promoter DNA, indicating that the hGH and major late adenovirus promoters share a transcription regulatory element.

A subset of DNA sequences in the 5"flanking DNA of the human growth hormone (hGH) gene was examined by protein-DNA binding and gene transferexpression experiments. Two adjacent cis-acting elements (I and 11) were identified between nucleotides -308/-235 of the hGH gene that modulated the expression of a linked reporter gene in transfected HeLa cells. Elements I and I1 repressed gene expression whereas element I1 alone activated it. HeLa whole cell extracts contain two factors that bind hGH DNA carrying elements I and 11. Factor I binds to single-stranded DNA, and its binding is correlated with repression of gene expression. Factor I1 binds between nucleotides -275/ -257 of the hGH gene. This region is homologous to the binding site for the adenovirus major late transcription factor, and factor I1 binding to hGH DNA is competed by adenovirus major late promoter DNA, indicating that the hGH and major late adenovirus promoters share a transcription regulatory element.
Transcription is regulated by positive and negative control mechanisms including enhancer elements that activate expression (1) and silencers that inhibit transcription and share the orientation-and position-independence properties of enhancers (2). Several genes in yeast and higher eukaryotes are regulated by positive and negative control elements (3-6).
Our laboratory has been studying transcriptional control elements that regulate hormonal (7, 8) and tissue-specific (9) expression of the hGH'-related genes. In this communication, we describe two proteins, present in HeLa and rat anterior pituitary (GC) cell extracts, that bind to the hGH 5"flanking * This work was supported by National Institutes of Health Grant HD 17838 (to N. L. E.). The costs of publication of this article were defrayed in p&. by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. DNA. These protein-binding sequences and the proteins that bind to them regulate expression of a linked reporter gene in transfected HeLa cells. One element activates transcription and the other represses it.

MATERIALS AND METHODS
Plasmid Construction and DNA Transfer Experiments-Fragment hGH(-278/-250) with terminal BamHI linkers was synthesized by California Biotechnology Inc. Fragment hGH(-308/-235) was obtained from a Bal313'-deletion fragment (nucleotide -235 converted to a BamHI site) of the hGH 5"flanking DNA by digestion with ThaI (nucleotide -308) and BamHI (nucleotide -235). The fragment was subcloned between the SmaI and BamHI sites of pUC18 to yield plasmid pUChGH(-308/-235). The EcoRI site in the pUC polylinker was converted to a BglII site, and hGH DNA was excised by BamHI/ BglII digestion. Both fragments, hGH(-278/-250) and hGH(-308/ -235), were cloned into a BglII site at the 5'-end of the -100hGHp.cat gene (9). DNA transfer and chloramphenicol acetyltransferase assays were performed as previously described (9), except that 200-500 pg of cell extract was used in each chloramphenicol acetyltransferase assay.
Probe Preparation-The pUChGH(-308/-235) plasmid was linearized with BamHI, 32P-end-labeled on the lower strand by polynucleotide kinase, and the fragment excised with EcoRI. To label the upper strand, the plasmid was linearized with EcoRI, 3ZP-end-labeled, and the fragment excised with BamHI. The probes were purified by polyacrylamide gel electrophoresis.
Extract Preparation-HeLa cells, grown in suspension (5-8 X lo6 cells/ml), were provided by the Cell Culture Facility (University of California, San Francisco). Protein (10-20 mg) from a whole cell extract (10) was passed over a 1-ml phosphocellulose column (Whatman P-11) in Buffer A (20 mM Hepes, pH 8.0, 1 mM EDTA, 20% glycerol, 50 mM KC1). The column was washed with Buffer A and then eluted stepwise with Buffer A containing 0.1,0.3, and 0.5 M KCl. Fractions were dialyzed in Buffer A, and protein was determined by Coomassie Blue G-250 binding (Pierce Chemical Co.). The proteins were aliquoted, quick frozen, and stored at -70 "C.
DNA Biding Assay-The DNA binding assays were performed essentially according to Carthew et al. (11). Protein (5-10 pg) was incubated at 0 "C for 10 min with all assay components except the probe, and then 3ZP-labeled probe DNA (0.1-0.5 ng) was added and the incubation continued for 5 min on ice. When competitor DNA was included in the experiment, it was mixed with the probe before addition to the incubation. The samples were incubated at room temperature for 30-40 min before electrophoretic analysis (11) on a 6% polyacrylamide gel. DNase I footprinting was performed by scaling-up the DNA binding experiments (11).

RESULTS
HeLa cell proteins in step-eluted phosphocellulose column fractions were tested for binding to a 32P-labeled hGH(-308/ -235) DNA fragment using the gel electrophoresis binding assay (11). Two proteins bind the hGH(-308/-235) fragment; these proteins designated factors I and I1 are present in the 0.1 and 0.3 M KC1 eluates, respectively (Fig. 1A). We determined that factor I binds optimally to ssDNA, since it bound to boiled hGH(-308/-235) DNA but not native DNA (Fig.  lB, lanes a-c) and binds to strand-separated DNA (Fig. 1B,  lanes d and e). Factor I also binds both strands of the hGH(-308/-235) ssDNA (Fig. lB, lanes a and c ) . As discussed below, factor I1 binds native DNA. Both factors I and I1 bind specifically to hGH DNA. DNA binding experiments with synthetic oligonucleotides indicate that there are two factor I binding sites located on the upper strand between nucleotides -308/-281 and the lower strand between nucleotides -279/-250 of the hGH gene, and factor I binding to these sites is not competed by unrelated ssDNA.' In addition, factor I1 binding to 3ZP-labeled hGH(-308/-235) native DNA was competed by a 10-fold molar excess of unlabeled hGH(-308/-235) DNA but not a 500-fold excess of a 75-base pair HinfI fragment of pBFC322 (data not shown).
To map the factor I and I1 binding sites directly we employed DNase I footprinting assays. The "P-labeled hGH(-308/-235) ssDNA and native DNA fragments were mixed with column fractions containing factors I and 11, respectively, as described (11). Factor I binding to hGH(-308/ -235) ssDNA yielded a prominent and reproducible hypersensitive site at nucleotide -280; however, conditions were not found which yielded a DNase I footprint ( Fig. 2A). The inability to detect a footprint may be related to our finding of multiple factor I binding sites as discussed above. Factor I1 binding to hGH(-308/-235) native DNA produced a clear footprint at nucleotides -275/-257 (Fig. 2B). Factor I1 only binds native DNA and protects both strands of DNA between nucleotides -275/-257 (data not shown). This sequence contains a 10-base pair sequence (5'-GGTCACGTGG-3') present in the binding site for adenovirus major late transcription factor (MLTF), a protein found in uninfected HeLa cell extracts (11) that activates adenovirus transcription both in uiuo and in vitro (11,12).
To determine if factor I1 and MLTF may be identical, fragments of DNA derived from the adenovirus major late promoter and other viral and cellular promoters were used in competition experiments for factor I1 binding to hGH(-308/ -235) DNA. The data in Fig. 3 demonstrate that the adenovirus DNA fragment containing the MLTF binding site competes with the hGH-derived probe for factor I1 binding, whereas the DNA fragments from the SV40, Rous sarcoma virus, and rat insulin promoters had little effect. Some competition was observed with a large amount of hepatitis B virus DNA, but the significance of this is unknown, since we could find no obvious sequence similarity to the factor I1 binding L. N. Peritz and N. L. Eberhardt, manuscript submitted for publication. site on this fragment. The ability of adenovirus DNA to compete with hGH DNA for factor I1 binding suggests that these genes bind the same protein.
To determine whether the DNA containing elements I and I1 has transcription regulatory activity, two fragments, hGH(-279/-250) (containing element 11) and hGH(-3081 -235) (containing elements I and 11), were independently fused upstream from the -100hGHp.cat gene (9). The hybrid genes were introduced into HeLa cells, and chloramphenicol acetyltransferase activity was measured after 48 h (9) (Fig. 4). The -279/-250hGH.-lOOhGHp.~at gene expressed 3-fold more activity than the -100hGHp.cat gene. The -3081 -235hGH.-lOOhGHp.~at gene expressed only 50% of the -100hGHp.cat gene activity and &fold less activity than the -279/-250hGH.-lOOhGHp.~at gene. These results indicate that positive and negative control elements are located between nucleotides -308/-235 of the hGH gene and suggest that factor I1 activates gene expression and factor I represses it. Since the hGH(-308/-235) DNA fragment contains binding sites for both factors I and 11, repression by factor I must be dominant to activation by factor 11.

DISCUSSION
We have shown that the hGH 5'-flanking DNA contains two cis-acting elements (I and 11) that mediate negative and positive control, respectively, of hGH promoter activity and have identified two trans-acting factors (I and 11) that may mediate this activity. Based on chromatographic (Fig. lA), , and functional properties (Fig. 4), we conclude that factor I1 is related to MLTF (11, 12) or upstream stimulatory factor (13,14) that activates adenovirus major late promoter transcription and that the adenovirus and hGH genes share a common transcription element. Interestingly, a yeast centromere DNA-binding protein that also binds to many yeast promoters binds to a sequence that is homologous to the MLTF binding site (15), suggesting that MLTF-related DNA binding motifs may represent conserved protein binding sites (15). Our studies indicating that MLTFrelated proteins can affect hGH gene expression and recent evidence that MLTF activates the rat fibrinogen promoter (16) support this concept. Evidence has been presented that the interaction of upstream stimulatory factor with transcription factor IID (adenovirus major late promoter TATAAA element binding factor) is involved in promoter activation (13,14), and similar mechanisms have been proposed for yeast gene activation by the MLTF-related protein (15). Thus, factor I1 interactions with transcription factor IID-related proteins may be important for hGH gene expression as well.
The physiological significance of factors I and I1 on hGH gene expression is unknown. Our data indicate that the repressor activity that is correlated with the binding of factor I to ssDNA (Fig. 1B) is dominant over activation by factor I1 (Fig. 4). It is possible that factor I1 activity may be exhibited when factor I levels are limiting, suggesting a potential switch mechanism. Both factors I and I1 are found in HeLa cells which lack hGH gene expression and rat somatotrophic tumor (GC) cells (data not shown) that express the rat GH gene and serve as an excellent model for studying hormonal (7, 8) and tissue-specific (9, 17) regulation of the hGH gene. Our earlier work demonstrates that deletion of elements I and I1 does not affect hGH gene expression in transfected GC cells (9). This suggests that elements downstream of elements I and I1 such as the tissue-specific control elements (9, 17) may act to overcome the effects of these elements. Accordingly, it is possible that the activity associated with factor I may be involved in repressing hGH gene expression in nonpituitary cells.