Portal de Periódicos da CAPES

Cell-specific Function of cis-Acting Elements in the Regulation of Human Alcohol Dehydrogenase 5 Gene Expression and Effect of the 5′-Nontranslated Region

1995; Elsevier BV; Volume: 270; Issue: 15 Linguagem: Inglês

10.1074/jbc.270.15.9002

1083-351X

Man‐Wook Hur, Howard J. Edenberg,

Pancreatic function and diabetes

The human alcohol dehydrogenase 5 gene (ADH5) differs from all other human alcohol dehydrogenase genes in its ubiquitous expression, although there are tissue-specific differences in the level of expression. To understand the expression of ADH5, we characterized the structure and function of its 5′ region by DNase I footprinting and transient transfection assays. The region from base pair (bp) −34 to +61, flanking the major transcription start site, had strong promoter activity in three different cell lines: HeLa, H4IIE-C3, and CV-1, and could explain the ubiquitous expression. Two Sp1 sites within that region are footprinted by nuclear extracts from all tissues and cells tested. There are sites further upstream that show cell- and tissue-specific differences in both their patterns of occupancy and their effects on promoter activity. The region between bp −34 and −64 strongly increases promoter activity in H4IIE-C3 cells, weakly activates in CV-1 cells, but has no effect in HeLa cells. The region between bp −127 and −163 is a positive element in both HeLa cells and CV-1 cells, but is a negative regulatory element in H4IIE-C3 cells. These differences in part explain the levels of expression of ADH5 in various tissues. Two regions (bp −64 to −127 and bp −163 to −365) contain negative regulatory elements that reduce promoter activity in all three cells. The 5′-nontranslated region of ADH5 contains two upstream ATGs. Insertion of 12 bp within the putative coding region of these upstream ATGs led to a 1.6-2.3-fold increase in activity. This suggests that the 5′-nontranslated region has regulatory significance. The human alcohol dehydrogenase 5 gene (ADH5) differs from all other human alcohol dehydrogenase genes in its ubiquitous expression, although there are tissue-specific differences in the level of expression. To understand the expression of ADH5, we characterized the structure and function of its 5′ region by DNase I footprinting and transient transfection assays. The region from base pair (bp) −34 to +61, flanking the major transcription start site, had strong promoter activity in three different cell lines: HeLa, H4IIE-C3, and CV-1, and could explain the ubiquitous expression. Two Sp1 sites within that region are footprinted by nuclear extracts from all tissues and cells tested. There are sites further upstream that show cell- and tissue-specific differences in both their patterns of occupancy and their effects on promoter activity. The region between bp −34 and −64 strongly increases promoter activity in H4IIE-C3 cells, weakly activates in CV-1 cells, but has no effect in HeLa cells. The region between bp −127 and −163 is a positive element in both HeLa cells and CV-1 cells, but is a negative regulatory element in H4IIE-C3 cells. These differences in part explain the levels of expression of ADH5 in various tissues. Two regions (bp −64 to −127 and bp −163 to −365) contain negative regulatory elements that reduce promoter activity in all three cells. The 5′-nontranslated region of ADH5 contains two upstream ATGs. Insertion of 12 bp within the putative coding region of these upstream ATGs led to a 1.6-2.3-fold increase in activity. This suggests that the 5′-nontranslated region has regulatory significance. The human ADH5 gene encodes χ-alcohol dehydrogenase (EC 1.1.1.1), which is expressed in all tissues tested and at all stages of development. This makes χ-alcohol dehydrogenase unique among the mammalian alcohol dehydrogenases, since the others are all are expressed in tissue-specific patterns (1Adinolfi A. Adinolfi M. Hopkinson D.A. Ann. Hum. Genet. 1984; 48: 1-10Crossref PubMed Scopus (44) Google Scholar, 2Seeley T.L. Mather P.B. Holmes R.S. Comp. Biochem. Physiol. B. 1984; 78: 131-139Crossref PubMed Scopus (20) Google Scholar, 3Duley J.A. Harris O. Holmes R.S. Alcoholism (N. Y.). 1985; 9: 263-271Crossref Scopus (96) Google Scholar, 4Holmes R.S. Duley J.A. Algar E.M. Mather P.B. Rout U.K. Alcohol Alcohol. 1986; 21: 41-56PubMed Google Scholar, 5Holmes R.S. Courtney Y.R. VandeBerg J.L. Alcoholism (N. Y.). 1986; 10: 623-630Crossref Scopus (31) Google Scholar, 6Julia P. Farres J. Pares X. Eur. J. Biochem. 1987; 162: 179-189Crossref PubMed Scopus (171) Google Scholar, 7Edenberg H.J. Crabbe J.C. Harris R.A. The Genetic Basis of Alcohol and Drug Actions. Plenum Press, New York1991: 165-223Crossref Google Scholar). The level of expression of χ-alcohol dehydrogenase differs among different tissues, however. Human liver and kidney show two to three fold higher levels of χ-alcohol dehydrogenase expression than the other tissues. In rodents, baboons, and horses the difference in expression is even greater (1Adinolfi A. Adinolfi M. Hopkinson D.A. Ann. Hum. Genet. 1984; 48: 1-10Crossref PubMed Scopus (44) Google Scholar, 2Seeley T.L. Mather P.B. Holmes R.S. Comp. Biochem. Physiol. B. 1984; 78: 131-139Crossref PubMed Scopus (20) Google Scholar, 3Duley J.A. Harris O. Holmes R.S. Alcoholism (N. Y.). 1985; 9: 263-271Crossref Scopus (96) Google Scholar, 4Holmes R.S. Duley J.A. Algar E.M. Mather P.B. Rout U.K. Alcohol Alcohol. 1986; 21: 41-56PubMed Google Scholar, 5Holmes R.S. Courtney Y.R. VandeBerg J.L. Alcoholism (N. Y.). 1986; 10: 623-630Crossref Scopus (31) Google Scholar, 6Julia P. Farres J. Pares X. Eur. J. Biochem. 1987; 162: 179-189Crossref PubMed Scopus (171) Google Scholar). χ-Alcohol dehydrogenase also differs from other mammalian alcohol dehydrogenases in its substrate specificity. χ-ADH 1The abbreviations used are:ADHalcohol dehydrogenaseCATchloramphenicol acetyltransferasekbkilobase(s)bpbase pair(s)HPLChigh performance liquid chromatography. most efficiently oxidizes long chain alcohols and ω-hydroxyfatty acids (8Pares X. Vallee B.L. Biochem. Biophys. Res. Commun. 1981; 98: 122-130Crossref PubMed Scopus (109) Google Scholar, 9Wagner F.W. Pares X. Holmquist B. Vallee B.L. Biochemistry. 1984; 23: 2193-2199Crossref PubMed Scopus (104) Google Scholar, 10Kaiser R. Holmquist B. Hempel J. Vallee B.L. Jornvall H. Biochemistry. 1988; 27: 1132-1140Crossref PubMed Scopus (101) Google Scholar, 11Giri P.R. Linnoila M. O'Neill J.B. Goldman D. Brain Res. 1989; 481: 131-141Crossref PubMed Scopus (39) Google Scholar). It is also the NAD+ and glutathione-dependent formaldehyde dehydrogenase (EC 1.2.1.1), as shown by the identity in partial amino acid sequence and in structural and enzymatic properties of the homologous rat liver alcohol dehydrogenase and formaldehyde dehydrogenase (12Koivusalo M. Baumann M. Uotila L. FEBS Lett. 1989; 257: 105-109Crossref PubMed Scopus (294) Google Scholar). χ-Like alcohol dehydrogenases are highly conserved during mammalian evolution (13Sharma C.P. Fox E.A. Holmquist B. Jornvall H. Vallee B.L. Biochem. Biophys. Res. Commun. 1989; 164: 631-637Crossref PubMed Scopus (38) Google Scholar, 14Giri P.R. Krug J.F. Kozak C. Moretti T. O'Brien S.J. Seuanez H.N. Goldman D. Biochem. Biophys. Res. Commun. 1989; 164: 453-460Crossref PubMed Scopus (29) Google Scholar, 15Edenberg H.J. Brown C.J. Carr L.G. Ho W.H. Hur M.W. Adv. Exp. Med. Biol. 1991; 284: 253-262Crossref PubMed Google Scholar, 16Hur M.-W. Ho W.-H. Brown C.J. Goldman D. Edenberg H.J. DNA Seq. 1992; 3: 167-175Crossref PubMed Scopus (17) Google Scholar). Escherichia coli contains an alcohol/formaldehyde dehydrogenase with 60% amino acid sequence identity (17Gutheil W.G. Holmquist B. Vallee B.L. Biochemistry. 1992; 31: 475-481Crossref PubMed Scopus (79) Google Scholar), indicating much longer evolutionary conservation. This suggests χ-like alcohol dehydrogenases play important physiological roles. alcohol dehydrogenase chloramphenicol acetyltransferase kilobase(s) base pair(s) high performance liquid chromatography. ADH5 was recently cloned in our laboratory (18Hur M.-W. Edenberg H.J. Gene (Amst.). 1992; 121: 305-311Crossref PubMed Scopus (47) Google Scholar). The 5′ nontranslated region of ADH5 is unusual; it contains two upstream ATG codons that are in frame with each other but out of frame with the χ-ADH coding sequence (18Hur M.-W. Edenberg H.J. Gene (Amst.). 1992; 121: 305-311Crossref PubMed Scopus (47) Google Scholar). The high conservation and presumed physiological importance of χ-ADH, the quantitative differences in expression in different tissues, and the unusual 5′ nontranslated region all make ADH5 an interesting gene to study. To analyze the molecular mechanisms underlying the expression of ADH5 in different tissues, we used DNase I footprinting to detect cis-acting elements. We carried out functional assays of the promoter using transient transfections of a reporter gene into three different cell lines: H4IIE-C3 (a rat hepatoma line), HeLa (a human cervical carcinoma line), and CV-1 (a monkey kidney fibroblast line). We found that a small promoter region functions well in all three cell lines, and we detected multiple positive and negative regulatory regions further upstream. Several cis-acting elements act as positive regulatory elements in one cell line but as negative regulatory elements in another. We also showed that a minor alteration in the 5′-nontranslated region can have a significant effect on gene expression. All enzymes were purchased from Life Technologies, Inc., Promega (Madison, WI), and Boehringer Mannheim. The buffers for restriction enzymes were supplied by the manufacturers. Poly(dI˙dC) was purchased from Boehringer Mannheim and Pharmacia Biotech Inc. The exonuclease III nested deletion kit and phosphorothioate dNTP were purchased from Promega (Erase a Base System, Promega). Other nucleotides were purchased from Pharmacia. [γ-32P]ATP and [α-32P]dNTPs were purchased from DuPont. Silica gel TLC plates were purchased from Eastman Kodak Co. or J. T. Baker (Baker Flex; Phillipsburg, NJ). pCAT Control and pCAT Basic were purchased from Promega. pCAT Control contains the SV40 promoter, enhancer, and CAT coding sequence. pCAT Basic contains the CAT coding sequence but lacks a eukaryotic promoter and enhancer. pCMV-Luc (19de Wet J.R. Wood K.V. DeLuca M. Helinski D.R. Subramani S. Mol. Cell. Biol. 1987; 7: 725-737Crossref PubMed Scopus (2646) Google Scholar) contains a firefly luciferase gene driven by a Cytomegalovirus (CMV) promoter in a vector called pCDNAI (Invitrogen); it was obtained from Dr. Y.-C. Yang (Indiana University School of Medicine). pCAT-X-1342 contains the 1.4-kb FokI fragment of ADH5 subcloned into pCAT Basic at a blunted XbaI site in front of the CAT gene; it contains ADH5 sequences extending from bp −1342 to +61 relative to the transcription initiation site. The pCAT-X series of plasmids was derived from pCAT-X-1342 by 5′ nested deletions using exonuclease III or using restriction endonucleases followed by Klenow treatment to create blunt ends before subcloning. All pCAT-X plasmids extend to bp +61; their 5′ ends are at the positions designated in their names (i.e. pCAT-X-163 contains from bp −163 to +61 of the ADH5 promoter; see Fig. 2). pCAT-AA was created by cloning the AvaI restriction fragment (bp −726 to −61) into pCAT Basic at the PstI site (blunted). pCAT-H-1342 contains the same 1.4-kb FokI fragment of ADH5 as in pCAT-X-1342 (bp −1342 to +61), except that it is subcloned into pCAT Basic at a blunted HindIII site in front of the CAT gene. pCAT-H-253 was made from pCAT-H-1342 by 3′ deletion as described above; it contains ADH5 sequences from bp −1342 to −253 (Fig. 2). The deleted plasmids were sequenced from both direction using primers flanking the insertion site (HE13, CAGGAAACAGCTATGACC and HE66, CAACGGTGGTATATCCAGTG). Two additional oligonucleotides were used to sequence regions originally obtained in only one direction. Nuclear extracts were prepared from liver, kidney, spleen, and brain of C57BL/6J mice according to Gorski et al. (20Gorski K. Carneiro M. Schibler U. Cell. 1986; 47: 767-776Abstract Full Text PDF PubMed Scopus (1063) Google Scholar). Nuclear extracts from cultured cells (CV-1, H4IIE-C3) were made according to Shapiro et al. (21Shapiro D.J. Sharp P.A. Wahli W.W. Keller M.J. DNA (N. Y.). 1988; 7: 47-55Crossref PubMed Scopus (525) Google Scholar). HeLa cell nuclear extract, Sp1, AP-2, and TFIID were purchased from Promega. Antibodies to Sp1 (Sp1(PEP2)) were obtained from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). To examine the proximal region, pCAT-H-1342 was digested with NarI (at bp −128), dephosphorylated with calf intestine alkaline phosphatase, and labeled with polynucleotide kinase and [γ-32P]ATP. After removal of the free nucleotides by Sephadex G25 spin column chromatography (22Maniatis T. Fritsch E.F. Sambrook J. Molecular Cloning: A Laboratory Manual. Cold Spring Harbor Laboratory, Cold Spring Harbor, NY1982Google Scholar), the DNA was digested with XbaI (a restriction site in the polylinker). For footprinting from the opposite end of the fragment, the order of digestion was opposite, so the XbaI site was labeled. The NarI to XbaI fragment was purified by anion exchange HPLC column chromatography (Gen-Pak Fax column, Waters, Milford, MA). For footprinting upstream of the NarI site, the AvaI fragment from pCAT-X-1342 was dephosphorylated, kinased with [γ-32P]ATP, and digested with HinfI; the 331-bp fragment (bp −402 bp to −61) was isolated by HPLC as described above. For footprinting in opposite direction, the AvaI fragment was digested with HinfI and HaeIII, the resulting HinfI to AvaI fragment was separated, dephosphorylated, kinased, and further digested with EcoRII. The HinfI/ EcoRII fragment was separated by HPLC and used for the DNase I footprint assays. DNase I digestion and electrophoresis were as described previously (23Brown C.J. Baltz K.A. Edenberg H.J. Gene (Amst.). 1992; 121: 313-320Crossref PubMed Scopus (19) Google Scholar). Nuclear extract (10-40 μg), probe (40,000 cpm) and poly(dI˙dC) (1-2 μg) were mixed in binding buffer (10 m M HEPES (pH 7.9), 60 m M KCl, 1 m M EDTA, 7% glycerol, 0.2 m M dithiothreitol) and incubated at room temperature for 15 min. 5 μl of DNase I (0.2-0.4 unit) was added and incubated for 2 min. When extract was omitted, 0.04 unit of DNase I were used. For purified transcription factors and their control reactions, 60 ng of factors (Sp1, AP-2, TFIID), 0.004 unit of DNase I, and no poly(dI˙dC) were used. DNase I digestion was stopped by adding 75 μl of DNase I stop solution (20 m M Tris (pH 7.5), 20 m M EDTA, 5 m M EGTA, and 5 μg/ml yeast tRNA). The mixture was extracted once with phenol/CHCl3and precipitated by the addition of 0.5 volume of 7.5 M ammonium acetate and 2.5 volume of ethanol. The pellets were resuspended in 5 μl of sequencing loading buffer. The digested DNA (10,000 cpm) was electrophoresed in a 6% polyacrylamide/7 M urea sequencing gel (Life Technologies, Inc.) for about 50 min. An autoradiogram was obtained by exposing the dried gel to film (XAR-5, Eastman Kodak Co.) overnight at −70°C with a Quanta III image enhancing screen (DuPont). Gel mobility shift assays were carried out as described previously (23Brown C.J. Baltz K.A. Edenberg H.J. Gene (Amst.). 1992; 121: 313-320Crossref PubMed Scopus (19) Google Scholar). The site B oligonucleotide was a double-stranded oligonucleotide with the sequence (top strand) TAGGCGCTCGCCACGCCCATGCCTCCGTC. The mutant site B oligonucleotide had the sequence TAGGCGCTCGCTCTAGATATGCCTCCGTC. The site C oligonucleotide was a double-stranded oligonucleotide with the sequence (top strand) CCCCCACGCCCCGCCCCCCTCGCTAGGCGC. The mutant site C oligonucleotide had the sequence CCCCCCACGCAAGATCTCACTCGCTAGGC. One strand of each oligonucleotide was labeled using polynucleotide kinase as described above, and the complementary oligonucleotides were annealed. Each assay contained (in 20 μl) 10 m M HEPES (pH 7.9), 60 m M KCl, 1 m M dithiothreitol, 1 m M EDTA, 7% glycerol and the appropriate extract (5.2 μg of HeLa nuclear extract or 1 footprinting unit of purified Sp1. Where indicated, 0.5 or 2.0 μl of unlabeled competitor oligonucleotide (at 4 pmol/μl) or 1 μg of antibody against Sp1 was added to the incubation. Electrophoresis was at 80 V for 4 h in 4% polyacrylamide gel with 0.167 × TAE buffer (1 × TAE, 40 m M Tris, 20 m M acetic acid, 1 m M EDTA). CV-1 (African green monkey kidney cells), H4IIE-C3 (rat hepatoma cells), and HeLa (human cervical carcinoma cells) were grown on 10-cm dishes in minimal essential medium (Life Technologies, Inc.) supplemented with 5% fetal calf serum (CV-1 and HeLa) or in SWIM'S medium (Sigma) supplemented with 10% fetal calf serum (H4IIE-C3). Medium was replaced 4 h before the transfections. 15 μg of pCAT-X-1342 (or the molar equivalent of other plasmids), 2 μg of pCMV-Luc, and sufficient pUC18 to bring the total DNA to 25 μg were transfected into the cells by the CaCl2-DNA coprecipitation method (24Graham F.L. van der Eb A.J. Virology. 1983; 52: 456-457Crossref Scopus (7037) Google Scholar). The DNA was allowed to remain on the cells for 4 h (CV-1, HeLa cells) or 18 h (H4IIE-C3 cells), then the medium was removed and replaced for 2 min with medium containing either 20% (CV-1, HeLa) or 15% (H4IIE-C3) glycerol. The glycerol-containing medium was replaced with the appropriate growth medium, and incubation was continued for a total of 48 h. The plates were gently washed four times with phosphate-buffered saline, the cells were harvested, and cell pellets were resuspended in 150 μl of lysis buffer (100 m M K˙PO4(pH 7.8), 1 m M dithiothreitol) and broken by sonication. Luciferase activity was used as an internal control to normalize the plate to plate variation in transfection efficiency. Luciferase activity was assayed (19de Wet J.R. Wood K.V. DeLuca M. Helinski D.R. Subramani S. Mol. Cell. Biol. 1987; 7: 725-737Crossref PubMed Scopus (2646) Google Scholar) using 30 μl of each supernatant in 370 μl of assay buffer (25 m M Gly-Gly (free base), 15 m M MgCl2, 5 m M ATP). 100 μl of luciferin (6 mg/ml) was added automatically, and activity was measured as relative light units for 10 s in a LB9501 Luminometer (Berthold Analytical Instruments, Nashua, NH). CAT assays (25Gorman C.M. Moffat L.F. Howard B.H. Mol. Cell. Biol. 1982; 2: 1044-1051Crossref PubMed Scopus (6047) Google Scholar) were conducted by incubating cell extracts containing 20,000 relative light units of luciferase activity in 0.25 M Tris-HCl (pH 7.8), 50 m M acetyl-CoA, 75 nCi [14C]chloramphenicol, 5 m M EDTA in 200 μl at 37°C for 2.5 h. The acetylation of chloramphenicol was analyzed by silica gel thin layer chromatography (25Gorman C.M. Moffat L.F. Howard B.H. Mol. Cell. Biol. 1982; 2: 1044-1051Crossref PubMed Scopus (6047) Google Scholar) and quantitated with the AMBIS Radioanalytic Imaging System (AMBIS, San Diego, CA). CAT activity is expressed as percent chloramphenicol acetylated per 20,000 luciferase units (average of three to four independent experiments). When CAT activity was high, smaller aliquots were reassayed for shorter times to keep the reactions in the linear range, and the results were normalized to 2.5 h assays containing 20,000 relative light units of luciferase. The proximal promoter region does not contain either a TATA box or a CCAAT box (Fig. 1). It is very high in G + C content (73% in the first 200 bp upstream of the translation initiation codon and 61% in the 358 bp upstream region). This region has the characteristics of a CpG island (26Gardiner-Garden M. Frommer M. J. Mol. Biol. 1987; 196: 261-282Crossref PubMed Scopus (2726) Google Scholar, 27Bird A.P. Nature. 1986; 321: 209-213Crossref PubMed Scopus (3128) Google Scholar) : CpGs are found approximately as often as expected from the base composition, and CpGs are about equal in frequency to GpCs (18Hur M.-W. Edenberg H.J. Gene (Amst.). 1992; 121: 305-311Crossref PubMed Scopus (47) Google Scholar). These are characteristics of the promoters of housekeeping genes. ADH5 is the only human alcohol dehydrogenase gene with these characteristics (28Edenberg H.J. Brown C.J. Pharmacogenetics. 1992; 2: 185-196Crossref PubMed Scopus (27) Google Scholar). There are consensus sequences for Sp1 (29Kadonaga J.T. Jones K.A. Tijan R. Trends Biochem. Sci. 1986; 11: 20-24Abstract Full Text PDF Scopus (950) Google Scholar, 30Courey A.J. Tjian R. McKnight S.L. Yamamoto K.R. Transcriptional Regulation. Cold Spring Harbor Laboratory, Plainview, NY1992: 743-769Google Scholar) and Sp1-related factors (31Hagen G. Muller S. Beato M. Suske G. Nucleic Acids Res. 1992; 20: 5519-5525Crossref PubMed Scopus (527) Google Scholar, 32Kingsley C. Winoto A. Mol. Cell. Biol. 1992; 12: 4251-4261Crossref PubMed Scopus (491) Google Scholar, 33Imataka H. Sogawa K. Yasumoto K.-I. Kikuchi Y. Sasano K. Kobayashi A. Hayami M. Fujii-Kuriyama Y. EMBO J. 1992; 11: 3663-3671Crossref PubMed Scopus (314) Google Scholar) and for AP-2 (34Williams T. Admon A. Luscher B. Tjian R. Genes & Dev. 1988; 2: 1557-1569Crossref PubMed Scopus (487) Google Scholar) clustered around the transcription start point (Fig. 1); these factors create footprints in this region (see below). There are consensus sequences (35Locker J. Hames B.D. Higgins S.J. Gene Transcription: A Practical Approach. IRL Press, Oxford1993: 321-345Google Scholar) for C/EBP and other CCAAT-binding proteins, and for HNF5, which might be involved in the higher expression of ADH5 in liver. Consensus sequences for other nuclear transcription factors, e.g. E-boxes, AP1, SRE, and XRE, are also found. It is interesting that there are five regions that contain groups of heat shock elements. Normally three or more clustered nGAAn sites are needed; the group of sites from −290 to −258 is the most likely of these to be functional. There are as yet no data about the induction of ADH5 by heat shock, but since the protein is protective against formaldehyde and other potential toxins, this might be physiologically relevant. We prepared a series of plasmids in which various portions of the ADH5 promoter were placed in front of the cat gene in the vector pCAT Basic (Fig. 2). We carried out transient expression assays in three different cell lines (H4IIE-C3, HeLa, and CV-1) to determine the extent to which each promoter fragment could direct transcription. In H4IIE-C3 cells, pCAT-X-1342 produced high levels of CAT activity, 42% of that obtained with the SV40 promoter plus enhancer construct pCAT Control (Fig. 3 A). Deletions extending down to bp −365 had little effect on CAT activity. The next three deletions each increased CAT activity: pCAT-X-163 showed 57% more CAT activity than pCAT-X-365, pCAT-X-127 showed 65% more activity than pCAT-163, and pCAT-X-64 showed 39% more activity than pCAT-X-127. pCAT-X-127 and pCAT-X-64 were more active than the SV40 promoter/enhancer combination in pCAT Control. Further deletion to −34 bp (pCAT-X-34) reduced activity 4-fold. This smallest promoter tested, pCAT-X-34, was nearly as active as the entire 1.4-kb fragment contained in pCAT-X-1342. pCAT-H-253, a derivative of pCAT-H-1342 in which the promoter is deleted from the 3′ end to bp −253 was not active (data not shown). In HeLa cells, pCAT-X-1342 produced high levels of CAT activity, 31% of that obtained with pCAT Control (Fig. 3 B). The effects of deletions differed from those seen in H4IIE-C3 cells. Deletion of the region between bp −1342 and −838 and between bp −583 and −465 reduced CAT activity by 15 and 25%, respectively. Deletion of the region between bp −365 and −163 increased activity 77%. Deletion of the region from bp −163 to −127 decreased activity 3-fold, and there was a 27% increase in activity upon further deletion to bp −64. pCAT-H-253 was not active (Fig. 3 B and data not shown). In CV-1 cells, as in the other cells tested, pCAT-X-1342 produced high levels of CAT activity, 44% of that obtained with pCAT Control (Fig. 3 C). The effects of the deletions differed from those seen in either H4IIE-C3 or HeLa cells. Deletions extending to bp −583 did not substantially affect the activity of the ADH5 promoter, and further deletions to bp −365 increased activity slightly. In CV-1, as in the other two cell lines, deletion of the region between bp −365 and −163 increased activity by 58%. Deletion of the region from bp −163 to −127 decreased activity 2-fold. Deletion to bp −64 increased activity by 55%, and further deletion to bp −34 decreased activity by 31%. pCAT-H-253 was not active (Fig. 3 C). Both pCAT-X-1342 and pCAT-H-1342 contain the same 1.4-kb promoter fragment, cloned into different positions in the polylinker of pCAT Basic (see “Experimental Procedures”). Both contain the two upstream ATGs that are found in the ADH5 gene itself (18Hur M.-W. Edenberg H.J. Gene (Amst.). 1992; 121: 305-311Crossref PubMed Scopus (47) Google Scholar), followed by in-frame (and also out of frame) termination codons, as they are in ADH5. pCAT-H-1342 has an additional 12 nucleotides (from the polylinker) inserted within the putative coding region of these upstream ATGs (Fig. 4). Both upstream ATGs are out of frame with the ATG of the cat gene, just as they are out of frame with the ATG of ADH5 (18Hur M.-W. Edenberg H.J. Gene (Amst.). 1992; 121: 305-311Crossref PubMed Scopus (47) Google Scholar). In all three cell lines, pCAT-H-1342 directed from 1.6- to 2.3-fold more CAT activity than did pCAT-X-1342 (Fig. 4). Primer extension assays (18Hur M.-W. Edenberg H.J. Gene (Amst.). 1992; 121: 305-311Crossref PubMed Scopus (47) Google Scholar) showed two major transcription start sites (bp +1 and +3 in Fig. 1) and one minor transcription start site (at bp −82). The minor start site suggested that ADH5 might have a second promoter upstream. To investigate this, pCAT-AA was analyzed. pCAT-AA contains sequences from bp −726 to −61 bp, which includes the putative upstream transcription start site at bp −82 but not the major downstream sites at bp +1 and +3. pCAT-AA showed no promoter activity in any of the cell lines tested (Fig. 3). To identify the location of cis-acting elements in the proximal ADH5 promoter, we carried out DNase I footprinting assays (36Galas D.J. Schmitz A. Nucleic Acids Res. 1978; 5: 3157-3170Crossref PubMed Scopus (1430) Google Scholar) on fragments spanning the region from bp −401 to +61 relative to the major transcription start point of ADH5 (18Hur M.-W. Edenberg H.J. Gene (Amst.). 1992; 121: 305-311Crossref PubMed Scopus (47) Google Scholar). Footprinting experiments were carried out using nuclear extracts prepared from the cell lines used for transfection studies and from various mouse tissues; we also examined the binding of purified transcription factors Sp1, AP-2, and TFIID. Extracts from mouse liver, kidney, and spleen showed footprints B, C, and D at bp +3 to +22, bp −38 to −4 and bp −57 to −40, respectively (Fig. 5, A and B). Liver showed a cluster of hypersensitive sites from bp +24 to +27 (Fig. 5 A). The pattern in mouse brain extract is more similar to that of the other cells when 20 μg of extract was used (data not shown; it did not footprint well in Fig. 5 A, lane 7, where only 10 μg of brain extract was used). Among these tissues, only kidney showed footprint A (bp +32 to +61), and only on one strand (Fig. 5, A versus B). Further upstream, additional DNase I footprints were detected (Fig. 5 C). There were differences among the cell lines. CV-1 showed a strong footprint J (bp −384 to −343) that was not seen with other extracts (Fig. 5 C). Footprint I (bp −318 to −284) just downstream of that is seen in liver extracts and also in CV-1 and H4IIE-C3 extracts. Fig. 7 summarizes the footprints. There are differences in the boundaries of the footprints, e.g. the footprint in the region from bp −57 to −40 is slightly shorter in brain and spleen than in liver and kidney (Fig. 5 A, lanes 7-10). Footprint F (bp −151 to −112) is seen in liver extract but not seen with the other tissue extracts; a slightly smaller footprint F (−151 to −118) is detected in CV-1 and H4IIE-C3 extracts but not in HeLa cells.FIG. 7Summary of DNase I footprints and their effects on gene expression. Boxes indicate footprints (see Fig. 5); solid boxes are footprints seen in liver; stippled boxes A and J are seen in other tissues. The arrows beneath reflect the regions tested in successive deletion mutants (Fig. 3); their effects on CAT activity are represented as + for positive, 0 for no effect, and - for negative effect.View Large Image Figure ViewerDownload Hi-res image Download (PPT) Nuclear extracts prepared from H4IIE-C3, HeLa, and CV-1 produced two strong footprints, B and C, immediately flanking the transcription start point (Fig. 5, A lanes 4-6, and B, lanes 3-5). There are hypersensitive sites right around the transcription start site (+1) in all tissues except brain. There are two weaker footprints flanking B and C: D from bp −57 to −40 and A from bp +32 to +61 (Fig. 5, A and B). The HeLa extracts produced the strongest footprints B and C and looked most like the pattern seen with purified Sp1 (see below). Both H4IIE-C3 and CV-1 produced several footprints upstream of bp −60, whereas HeLa cells did not. As was seen for different tissues, the patterns of protection differ slightly among the extracts from different cells (Fig. 5). Footprint J was seen in CV-1 cells. Given the high G + C content and presence of Sp1 consensus sequences, we tested the binding of Sp1 and AP-2 to this region of the ADH5 promoter. Sp1 produced four footprints; two strong footprints, B and C, correspond to footprints seen in the cell extracts. To further test whether Sp1 was the protein responsible for the footprinting of sites B and C, gel mobility shift assays were carried out using antibodies to Sp1. Fig. 6 A shows that purified Sp1 can bind to site B (lane 8) but not to a mutated site B (lane 12). Antibody to Sp1 shifts the retarded band to lower mobility. The protein(s) in the HeLa extract also bind specifically to site B, as shown in lane 4, and most of the retarded band is shifted to lower mobility by the antibody to Sp1, indicating that Sp1 is indeed the major protein binding to the site. The small residue of unshifted band might be due to insufficient antibody or to the presence of a second protein able to bind to the oligonucleotide. The mobility of the shifted band in the HeLa extract is slightly greater than that due to purified Sp1; this might be due to post-translational modifications in the protein, since the Sp1 gene product can produce bands with mobilities approximating 95 and 105 kDa. Fig. 6 B shows that the predominant protein that binds to site C is also Sp1; again a fraction of the band does not shift with antibody, and again the mobility of the shifted band in the HeLa extract is slightly greater than that due to purified Sp1, but similar to that which binds to site B. These data demonstrate that the bulk of the protein binding to sites B and C in the HeLa extract is Sp1. The pattern of AP-2 binding differs from that seen in any of the cell lines. AP-2 protects both strands at site A; this differs from the footprinting on only one strand seen in the cell extracts. AP-2 also produces weak footprints in sites extending from bp −66 to +27, partially overlapping footprints produced by various extracts and Sp-1. TFIID did not produce detectable footprints under our conditions. The human ADH5 gene has characteristics of both a housekeeping gene and a tissue-specific gene. It is expressed in all tissues and in all stages of development, but at higher levels in liver and kidney. The data presented here demonstrate that although the proximal promoter of ADH5 has some characteristics of a housekeeping gene, particularly a G + C-rich and TATA-less promoter, it also contains several tissue-specific cis-acting elements that affect expression differently in different cellular contexts. A small region between bp −34 and +61 (contained in pCAT-X-34) has surprisingly strong promoter activity in all three cell lines tested. Its activity in these cells ranges from 18 to 46% of the activity of the SV40 promoter + enhancer combination. This small region contains two Sp1 sites flanking the major transcription start site (sites B and C) that were footprinted by all nuclear extracts (Fig. 5). These two sites are bound strongly by purified Sp1 protein. The protein in HeLa extract that binds to these sites is Sp1, as shown by antibody double-gel shifts (Fig. 6). In the absence of a TATA box, Sp1 bound to this unusual arrangement of sites might directly stimulate transcription by actively recruiting the TFIID complex to the promoter (37Pugh B.F. Tjian R. Cell. 1990; 61: 1187-1197Abstract Full Text PDF PubMed Scopus (819) Google Scholar, 38Pugh B.F. Tjian R. J. Biol. Chem. 1992; 267: 679-682Abstract Full Text PDF PubMed Google Scholar). Tazi and Bird (39Tazi J. Bird A. Cell. 1990; 60: 909-920Abstract Full Text PDF PubMed Scopus (383) Google Scholar) showed that the G + C-rich promoters of housekeeping genes are depleted of histones and actively transcribed. Croston et al. (40Croston G.E. Kerrigan L.A. Lira L.M. Marshak D.R. Kadonaga J.T. Science. 1991; 251: 643-649Crossref PubMed Scopus (298) Google Scholar) suggested that Sp1 can act as an antirepressor by actively displacing histone. Sp1 or related factors might also displace histones from the ADH5 promoter region. Sp1 is expressed in all tissues, but the mRNA level can vary as much as 100-fold (30Courey A.J. Tjian R. McKnight S.L. Yamamoto K.R. Transcriptional Regulation. Cold Spring Harbor Laboratory, Plainview, NY1992: 743-769Google Scholar, 41Saffer J.D. Jackson S.P. Annarella M.B. Mol. Cell. Biol. 1991; 11: 2189-2199Crossref PubMed Scopus (496) Google Scholar). Thus, Sp1 (30Courey A.J. Tjian R. McKnight S.L. Yamamoto K.R. Transcriptional Regulation. Cold Spring Harbor Laboratory, Plainview, NY1992: 743-769Google Scholar, 42Kadonaga J.T. Carner K.R. Masiarz F.R. Tjian R. Cell. 1987; 51: 1079-1090Abstract Full Text PDF PubMed Scopus (1335) Google Scholar) or Sp1-related proteins (31Hagen G. Muller S. Beato M. Suske G. Nucleic Acids Res. 1992; 20: 5519-5525Crossref PubMed Scopus (527) Google Scholar, 32Kingsley C. Winoto A. Mol. Cell. Biol. 1992; 12: 4251-4261Crossref PubMed Scopus (491) Google Scholar, 33Imataka H. Sogawa K. Yasumoto K.-I. Kikuchi Y. Sasano K. Kobayashi A. Hayami M. Fujii-Kuriyama Y. EMBO J. 1992; 11: 3663-3671Crossref PubMed Scopus (314) Google Scholar) may be important in the widespread expression of χ-ADH. Upstream of this minimal promoter, there are cell- and tissue-specific differences in both the binding of transcription factors and the effects of the cis-acting sequences on gene expression (Fig. 7). For example, the region containing site D had different effects in different cells. Site D was footprinted by all nuclear extracts and strongly by Sp1 (Fig. 5, A and B). Its effect on promoter function, however, differed. In H4IIE-C3 cells, it was a strong positive regulatory element, stimulating CAT expression 5-fold (cf. pCAT-X-64 and pCAT-X-34, Fig. 3 A). pCAT-X-64 was a better promoter in the hepatoma cells than the SV40 promoter and enhancer (by 1.5-fold). In CV-1 cells, this sequence was a weak positive element that increased CAT expression by 45% (Fig. 3 C). In contrast, the same sequence did not have any effect in HeLa cells (Fig. 3 B). Thus the differential expression of Sp1 (30Courey A.J. Tjian R. McKnight S.L. Yamamoto K.R. Transcriptional Regulation. Cold Spring Harbor Laboratory, Plainview, NY1992: 743-769Google Scholar, 41Saffer J.D. Jackson S.P. Annarella M.B. Mol. Cell. Biol. 1991; 11: 2189-2199Crossref PubMed Scopus (496) Google Scholar) and/or related proteins may be important in the differential expression of this promoter. The region from bp −163 to −128 is another cell-specific element; it acts as positive element in CV-1 and HeLa cells, causing the 2-2.8-fold higher CAT expression of pCAT-X-163 compared with pCAT-X-127 (Fig. 3, B and C). Surprisingly, this region down-regulated gene expression in H4IIE-C3 cells by 39% (Fig. 3 A). Thus the sequences between bp −163 and −128 bp can act as either a positive or negative regulatory element in different cellular contexts. The cell-specific function of this element suggests that it is bound by different transcription factors in different cells or that the factor(s) bound interact differently with other tissue-specific factors also bound to the promoter. We observed a footprint in the region from bp -to −130 bp in CV-1 and H4IIE-C3 extracts and a larger footprint from bp −151 to −112 in liver extracts (Fig. 5 C). This region contains a consensus sequence for C/EBP or other CAAT-binding proteins (35Locker J. Hames B.D. Higgins S.J. Gene Transcription: A Practical Approach. IRL Press, Oxford1993: 321-345Google Scholar). The region from bp −127 to −65 acted as a weak negative element in all three cell lines tested, as evidenced by the lower expression of pCAT-X-127 relative to pCAT-X-64 (Fig. 3). A reduction in CAT activity by 21, 28, and 36% was seen in HeLa, H4IIE-C3, and CV-1 cells, respectively. A very weak footprint was observed in region from bp −100 to −61 only with CV-1 nuclear extract (Fig. 5 C). The region between bp −163 and −365 reduced CAT activity to the level of pCAT-X-34 in all three cell lines tested (Fig. 3). This region contains several footprinted sequences (Figs. 5 and 7). Regions further upstream did not substantially change promoter activity in H4IIE-C3 or CV-1 cells (Fig. 3, A and C), but in HeLa cells there was a modest stimulation by sequences located between bp −464 to −583 and between bp −837 to −1342 (Fig. 3 B). The 5′ non-translated region of ADH5 is unusual in having two upstream ATGs. pCAT-X-1342 and pCAT-H-1342 have the same 1.4 kb ADH5 upstream sequence inserted at different sites in the polylinker of pCAT Basic. This introduces an extra 12 bp in front of the CAT coding sequence in pCAT-H-1342, relative to pCAT-X-1342 (Fig. 4A). pCAT-H-1342 produced a 1.6-2.3-fold higher CAT activity than pCAT-X-1342 in all cell lines tested (Fig. 4 B). The higher CAT activity in pCAT-H-1342 may be due to the slightly longer upstream coding sequences (4 amino acids longer) in pCAT-H-1342, which may affect the translocation and reinitiation of ribosomes (43Kozak M. J. Biol. Chem. 1991; 266: 19867-19870Abstract Full Text PDF PubMed Google Scholar). Alternatively, the enhancement may be the result of an alteration in the secondary structure of the 5′-nontranslated region. We thank Ronald E. Jerome for excellent technical assistance with tissue culture, Lu Zhang and Jinghua Zhao for assistance with the antibody shift experiments, and Dr. Celeste Brown, Dr. Mang Yu, and Lu Zhang for their helpful advice.