The Transcriptional Regulation of Human Aldehyde Dehydrogenase I Gene
1995; Elsevier BV; Volume: 270; Issue: 29 Linguagem: Inglês
10.1074/jbc.270.29.17521
ISSN1083-351X
AutoresYuchio Yanagawa, James C. Chen, Lily C. Hsu, Akira Yoshida,
Tópico(s)Adipose Tissue and Metabolism
ResumoHuman cytosolic aldehyde dehydrogenase 1 (ALDH1) plays a role in the biosynthesis of retinoic acid that is a modulator for gene expression and cell differentiation. Northern blot analysis showed that liver tissue, pancreas tissue, hepatoma cells, and genital skin fibroblast cells expressed high levels of ALDH1. Sequence analysis showed that the 5'-flanking region contains a number of putative regulatory elements, such as NF-IL6, HNF-5, GATA binding sites, and putative response elements for interleukin-6, phenobarbital and androgen, in addition to a noncanonical TATA box (ATAAA) and a CCAAT box. Functional characterization of the 5'-regulatory region of the human ALDH1 gene was carried out by a fusion to the chloramphenicol acetyltransferase gene. A construct containing 2.6 kilobase pairs of the 5'-flanking region was efficiently expressed in hepatoma Hep3B cells, but not in erythroleukemic K562 cells or in fibroblast LTK− cells, which do not express ALDH1. Within this region, we define a minimal promoter (−91 to +53) that contains positive regulatory elements. The study using site-directed mutagenesis demonstrated that the CCAAT box region is the major cis-acting element involved in basal ALDH1 promoter activity in Hep3B cells. Gel mobility shift assays showed that NF-Y and other octamer factors bound CCAAT box and an octamer motif sequence, but not GATA site existing in the minimal promoter region. Two additional DNA binding activities associated with the minimal promoter were found in the nuclear extract from Hep3B cells, but not from K562 cells. These results offer the possible molecular mechanism of the cell type-specific expression of ALDH1 gene. Human cytosolic aldehyde dehydrogenase 1 (ALDH1) plays a role in the biosynthesis of retinoic acid that is a modulator for gene expression and cell differentiation. Northern blot analysis showed that liver tissue, pancreas tissue, hepatoma cells, and genital skin fibroblast cells expressed high levels of ALDH1. Sequence analysis showed that the 5'-flanking region contains a number of putative regulatory elements, such as NF-IL6, HNF-5, GATA binding sites, and putative response elements for interleukin-6, phenobarbital and androgen, in addition to a noncanonical TATA box (ATAAA) and a CCAAT box. Functional characterization of the 5'-regulatory region of the human ALDH1 gene was carried out by a fusion to the chloramphenicol acetyltransferase gene. A construct containing 2.6 kilobase pairs of the 5'-flanking region was efficiently expressed in hepatoma Hep3B cells, but not in erythroleukemic K562 cells or in fibroblast LTK− cells, which do not express ALDH1. Within this region, we define a minimal promoter (−91 to +53) that contains positive regulatory elements. The study using site-directed mutagenesis demonstrated that the CCAAT box region is the major cis-acting element involved in basal ALDH1 promoter activity in Hep3B cells. Gel mobility shift assays showed that NF-Y and other octamer factors bound CCAAT box and an octamer motif sequence, but not GATA site existing in the minimal promoter region. Two additional DNA binding activities associated with the minimal promoter were found in the nuclear extract from Hep3B cells, but not from K562 cells. These results offer the possible molecular mechanism of the cell type-specific expression of ALDH1 gene. Aldehyde dehydrogenases (aldehyde:NAD+ oxidoreductase, EC 1.2.1.3, ALDH) 1The abbreviations used are: ALDHaldehyde dehydrogenasePCRpolymerase chain reactionCATchloramphenicol acetyltransferasebpbase pair(s). 1 play a role in the detoxification of alcohol-derived acetaldehyde and the metabolism of corticosteroids, biogenic amines, and lipid peroxidation (reviewed in (1Yoshida A. Hsu L.C. Yasunami M. Prog. Nucleic Acids Res. Mol. Biol. 1991; 40: 255-287Crossref PubMed Scopus (180) Google Scholar)). Many human ALDH isozymes are distinguishable on the basis of separation by physicochemical methods, tissue and subcellular distributions, and enzymatic properties(1Yoshida A. Hsu L.C. Yasunami M. Prog. Nucleic Acids Res. Mol. Biol. 1991; 40: 255-287Crossref PubMed Scopus (180) Google Scholar). Cytosolic ALDH1 is active for retinalaldehyde oxidation and is considered to play a major role in the synthesis of retinoic acid(2Yoshida A. Hsu L.C. Davé V. Enzyme (Basel). 1992; 46: 239-244Crossref PubMed Scopus (136) Google Scholar), which is a potent modulator for gene expression and cell differentiation (reviewed in (3DeLuca L.M. FASEB J. 1991; 5: 2924-2933Crossref PubMed Scopus (816) Google Scholar)). The characterization of human ALDH1 cDNA and gene has shown that this gene spans 53 kilobase pairs, and contains 13 exons, which encode 501 amino acid residues including the chain initiation Met(4Hsu L.C. Tani K. Fujiyoshi T. Kurachi K. Yoshida A. Proc. Natl. Acad. Sci. U. S. A. 1985; 82: 3771-3775Crossref PubMed Scopus (165) Google Scholar, 5Hsu L.C. Chang W.-C. Yoshida A. Genomics. 1989; 5: 857-865Crossref PubMed Scopus (50) Google Scholar). The location of human ALDH1 was assigned to chromosome 9q21(6Hsu L.C. Yoshida A. Mohandas T. Am. J. Hum. Genet. 1986; 38: 641-648PubMed Google Scholar, 7Raghunthan L. Hsu L.C. Klisak I. Sparkes R.S. Yoshida A. Genomics. 1988; 2: 267-269Crossref PubMed Scopus (40) Google Scholar). aldehyde dehydrogenase polymerase chain reaction chloramphenicol acetyltransferase base pair(s). Human ALDH1 isozyme is expressed at different levels in various tissues examined, with the highest level in the liver and the lowest or undetectable level in the heart(8Harada S. Agarwal D.P. Goedde H.W. Life Sci. 1980; 26: 1773-1780Crossref PubMed Scopus (127) Google Scholar), suggesting that the expression of ALDH1 is tissue-specific. During embryonic development, ALDH1 is detectable at an early stage(9Yoshida A. Shibuya A. Davé V. Nakayama M. Hayashi A. Experientia (Basel). 1990; 46: 747-750Crossref PubMed Scopus (4) Google Scholar). The position- and developmental stage-specific expression of ALDH1 were observed in the mouse and chick retina, suggesting that the level of ALDH1 is correlated with the biosynthesis of retinoic acid(10Godbout R. Exp. Eye Res. 1992; 54: 297-305Crossref PubMed Scopus (36) Google Scholar, 11McCaffery P. Posch K.C. Napoli J.L. Gudas L. Drager U.C. Dev. Biol. 1993; 158: 390-399Crossref PubMed Scopus (132) Google Scholar). However, little is known regarding the molecular mechanism of the tissue and developmental stage-specific expression of the ALDH1 gene. Recent studies indicated that human ALDH1 was produced in the normal genital skin fibroblast cells but not in the cells obtained from X-linked androgen receptor-negative testicular feminization patients(12Pereira F. Rosenmann E. Nylen E. Kaufman M. Pinsky L. Wrogemann K. Biochem. Biophys. Res. Commun. 1991; 175: 831-838Crossref PubMed Scopus (40) Google Scholar, 13Yoshida A. Hsu L.C. Yanagawa Y. Adv. Exp. Med. Biol. 1993; 328: 37-44Crossref PubMed Google Scholar). These findings suggest that ALDH1 is induced by androgen receptor-androgen complex in genital cells, and retinoic acid produced by ALDH1 plays an important role in testicular differentiation(13Yoshida A. Hsu L.C. Yanagawa Y. Adv. Exp. Med. Biol. 1993; 328: 37-44Crossref PubMed Google Scholar). ALDH1 expression is modulated by phenobarbital and cyclophosphamide. Phenobarbital induces ALDH1 activity in cultured human hepatoma cells (14Marselos M. Strom S.C. Michalopoulus G. Chem. Biol. Int. 1987; 62: 75-88Crossref PubMed Scopus (29) Google Scholar) and in the liver of some rat strains(15Dunn T.J. Koleske A.J. Lindahl R. Pitot H.C. J. Biol. Chem. 1989; 264: 13057-13065Abstract Full Text PDF PubMed Google Scholar). Transcriptional activation of ALDH1 was observed in cyclophosphamide-resistant human carcinoma cells (16Yoshida A. Davé V. Han H. Scalon K.J. Adv. Exp. Med. Biol. 1993; 328: 63-70Crossref PubMed Scopus (34) Google Scholar) and mouse leukemic cell line L121O(17Radin A.I. Zhoa X.-L. Woo T.H. Colvin O.M. Hilton J. Biochem. Pharmacol. 1991; 42: 1933-1939Crossref PubMed Scopus (14) Google Scholar). To elucidate the regulatory mechanisms of ALDH1 expression and to define the promoter regions that are essential for its tissue-specific and inducible expression, we have characterized the 5'-flanking region of the human ALDH1 gene. Total cellular RNAs were prepared from various cultured cell lines by the established method(18Chomczynski P. Sacchi N. Anal. Biochem. 1987; 162: 156-159Crossref PubMed Scopus (63232) Google Scholar). Twenty micrograms of total RNAs were electrophoresed, transferred onto nitrocellulose membrane, and hybridized with a human ALDH1 cDNA probe (4Hsu L.C. Tani K. Fujiyoshi T. Kurachi K. Yoshida A. Proc. Natl. Acad. Sci. U. S. A. 1985; 82: 3771-3775Crossref PubMed Scopus (165) Google Scholar) and with a human β-actin cDNA probe, which served as an internal reference. A low background promoterless CAT expression vector, PUMSVOCAT, with an unique SmaI cloning site(19Salier J.P. Hirosawa S. Kurachi K. J. Biol. Chem. 1990; 265: 7062-7068Abstract Full Text PDF PubMed Google Scholar), and the genomic clone (λDASH-19) of human ALDH1 gene containing the extended 5'-region (5Hsu L.C. Chang W.-C. Yoshida A. Genomics. 1989; 5: 857-865Crossref PubMed Scopus (50) Google Scholar) were used for preparation of the expression constructs. Using an EcoRI/EcoRI fragment −673/+350 of the clone as a template, a fragment −673/+53 with artificial HindIII site at the 5' end and XbaI site at the 3' end was created by PCR, and subcloned into pBluescript II KS+. The ligation of the HindIII/EcoRI fragment −2536/-673 with the −673/+53 fragment yielded the 5' region of ALDH1 gene covering −2536/+53. A fragment from −1736 (PvuII site) to −673 (EcoRI site) generated from the clone −2536/+53 was ligated to a fragment −673/+53 described above. Two fragments, i.e. −266/+53 and −149/+53, were obtained from the −673/+53 clone by PvuII and DraI digestion respectively. The truncated fragments from −120, −91 and −50 to +53 were obtained by PCR using the −673/+53 fragment as a template. The nucleotide sequences of these truncated clones were confirmed by sequencing. pUMSVOCAT vector was digested by SmaI, tailed with dideoxythymidine triphosphate and ligated with a PCR product (−673/+53) which has a HindIII site at the 5' and XbaI site at the 3' end. Digestion of this product by HindIII and XbaI yielded CAT-assay constructs with a series of deleted ALDH1 promoter region. The fragments −2536/+53, −1726/+53, −673/+53, −266/+53, −149/+53, −120/+53, −91/+53 and −50/+53 were subcloned into the HindIII/XbaI site of pUMSVOCAT vector. These products are designated as pCAT-2536, pCAT-1726, pCAT-673, pCAT-266, pCAT-149, pCAT-120, pCAT-91, and pCAT-50, respectively. The SV40 promoter region prepared from pCAT-promoter vector (Promega) ligated with pUMSVOCAT, and the resultant pCAT-SV40 vector was used as an internal reference for CAT-assay. The deleted vectors, i.e. pCAT-120ΔC and pCAT120ΔA containing the deletion of CCAAT box (−74/-70) and the deletion of ATAAA box (−32/-28), respectively, were generated by a two-step PCR directed mutagenesis by the following procedures. First, a fragment −673/-60 was prepared using −673/-655 as 5' primer and 5'-CTCGGATACGATGAACAAACTCAG-3' as 3' primer, and a fragment −84/+53 was prepared using 5'-GTTTGTTCATCGTATCCGAGTATG-3' as 5' primer and +53/+36 as 3' primer. Subsequently these two fragments were mixed and used for another round of PCR in the presence of −120/-103 as 5' primer and +53/+36 as 3' primer. The final PCR product was subcloned into pBluescript vector, and the internal deletion was confirmed by sequencing. pCAT-120ΔA was constructed by the same procedure except for using −673/-655 and 5'-TTGTTCCTTTCTGCACGGGCTAAA-3' as primers for amplification of −673/-12, and 5'-GCCCGTGCAGAAAGGAACAAATAAAG-3' and +53/+36 as primers for amplification of −43/+53. All cell lines were obtained from American Type Culture Collection. Human non-genital fibroblast cells (GM8447) were cultured in minimal essential medium under 5% CO2. Human genital skin fibroblast cells(9024), human hepatoma cells (HepG2 and Hep3B), human breast cancer cells (MCF-7), and mouse fibroblast cells (LTK−) were cultured in Dulbecco's modified Eagle's medium under 5% CO2. Human prostate cancer cells (LNCaP), human erythroleukemic cells (K562), and human promyelocytic leukemia cells (HL60) were grown in RPMI 1640 under 5% CO2. All media were supplemented with penicillin (100 units/ml), streptomycin (100 εg/ml), and 10% fetal bovine serum (Life Technologies, Inc.). Human hepatoma Hep3B cells and mouse fibroblast cells LTK− were transfected by the use of Lipofectin (Life Technologies, Inc.), and human erythroleukemic cells (K562) were transfected by the use of Transfectam (Promega). The cells were cotransfected with 10 εg of the test plasmid and 2 εg of pCMVβgal control plasmid. After 48 h, the treated cells were washed with phosphate-buffered saline and harvested from the plate by scraping. The cells were then pelleted by centrifugation, resuspended in 0.25 M Tris chloride (pH 8.O), and disintegrated by freezing/thawing four times. Aliquots of the centrifuge supernatant of the lysate were used for assay of β-galactosidase activity(20Maniatis T. Fritsch E.F. Sambrook J. Gene Cloning: A Laboratory Manual. 2nd Ed. Cold Spring Harbor Laboratory, Cold Spring Harbor, NY1989Google Scholar). The remainder of the supernatant was heated to 60°C for 10 min to inactivate deacetylase and stored at −70°C prior to determination of CAT activity. CAT assays were performed by the phase-extraction method (21Seed B. Sheen J.-Y. Gene (Amst.). 1988; 67: 271-277Crossref PubMed Scopus (830) Google Scholar) using [14C]chloramphenicol as substrate. The CAT activity was normalized to β-galactosidase activity and expressed as -fold increase in activity over that of the simian virus 40 (SV40) early promoter. Nuclear extracts prepared by the method of Shapiro et al.(22Shapiro D.J. Sharp P.A. Wahli W.W. Keller M.J. DNA. 1988; 7: 47-55Crossref PubMed Scopus (478) Google Scholar) were preincubated in a 20-εl reaction mixture containing 20 mM HEPES (pH 7.9), 0.5 mM dithiothreitol, 0.5 mM EDTA, 7% glycerol, 1 εg of poly(dI-dC), 25 mM KCl, and 25 mM NaCl at 25°C. After 10 min, approximately 2 × 104 cpm of a 32P-end-labeled nucleotide probe was added and the incubation continued for 20 min. The mixtures, together with 2 εl of loading buffer (50% glycerol, 1 mM EDTA, 0.25% xylene cyanole, and 0.25% bromphenol blue), were electrophoresed in a 4% polyacrylamide gel in 0.5 × TBE buffer (45 mM Tris borate, pH 8.4, 0.1 mM EDTA). In competition assays, a large excess of unlabeled double-strand oligonucleotide competitor was incubated together with the nuclear extract prior to adding the 32P-labeled probe. For supershift assay, anti-NF-YB antibody (23Mantovani R. Pessara U. Tronche F. Li X.-Y. Knapp A.-M. Pasquali J.-L. Benoist C. Mathis D. EMBO J. 1993; 11: 3315-3322Crossref Scopus (165) Google Scholar) or anti-Oct-1 or -Oct-2 antibodies (Santa Cruz Biotech, Inc.) were incubated with the probe-nuclear extract mixtures for 30 min more prior to gel electrophoresis. The oligonucleotides NF-Y (24Urano Y. Watanabe K. Sasaki M. Tamaoki T. J. Biol. Chem. 1986; 261: 3244-3251Abstract Full Text PDF PubMed Google Scholar) CTF/NF1(25Chodosh L.A. Baldwin A.S. Carthew R.W. Sharp P.A. Cell. 1988; 53: 11-24Abstract Full Text PDF PubMed Scopus (435) Google Scholar), SP1(26Briggs M.R. Kadonaga J.T. Bell S.P. Tjian R. Science. 1986; 234: 47-52Crossref PubMed Scopus (1059) Google Scholar), and OCT (27O'Neill E.A. Fletcher C. Burrow C.R. Heintz N. Roeder R.G. Kelly T.J. Science. 1988; 241: 1210-1213Crossref PubMed Scopus (117) Google Scholar) used for competition assays are: NF-Y: 5'-CGGTTGGCAGCCAATGAAATACAAAGATGA-3'; CTF/NF1: 5'-CCTTTGGCATGCTGCCAATATG-3'; Sp1: 5'-ATTCGATCGGGGCGGGGCGAGC-3'; OCT: 5' TGTCGAATGCAAATCACTAGAA-3'. Northern blot analysis demonstrated that ALDH1 mRNA is abundant in human genital skin fibroblast cells(9024) and in hepatoma cells (HepG2 and Hep3B). ALDH1 mRNA is undetectable in non-genital skin fibroblast cells (GM8447 and LTK−) and in other cancer cell lines examined (MCF-7, LNCaP, K562, and HL60) (Fig. 1). The β-actin gene is expressed at comparable levels in all cell types. In a variety of human tissues examined, ALDH1 gene is expressed highly in liver and pancreas, moderately in skeletal muscle and kidney, at low levels in brain, heart, and lung, and is undetectable in placenta (data not shown). Two overlapping clones for the 5'-region of the gene were obtained by screening human genomic libraries. The nucleotide sequence of the region (starting −2536 counting from the transcription start site numbered +1) is shown in Fig. 2. In comparison with the reported sequence from −676 to exon 1(5Hsu L.C. Chang W.-C. Yoshida A. Genomics. 1989; 5: 857-865Crossref PubMed Scopus (50) Google Scholar), the present sequence displays a T/G transversion at −396 and a C/T transition at −100. The sequence of the entire 5'-flanking region was scanned on both strands for the search of potential protein binding motifs. Based on the published libraries of such motifs(28Faisst S. Meyer S. Nucleic Acids Res. 1992; 20: 2-26Crossref Scopus (938) Google Scholar), a noncanonical TATA box exists at −32, and a potential CCAAT box exists at −74. A series of potential NF-IL6-responsive elements are scattered through the entire sequence. Several other potential liver-specific sequences also exist in the region (Fig. 2). In an attempt to delineate the DNA elements that are important for the activity of the ALDH1 promoter, the 5'-sequences of the human, marmoset (sequenced in this laboratory), mouse(29Rongnoparut P. Weaver S. Gene (Amst.). 1991; 101: 261-265Crossref PubMed Scopus (28) Google Scholar), and rat (15Dunn T.J. Koleske A.J. Lindahl R. Pitot H.C. J. Biol. Chem. 1989; 264: 13057-13065Abstract Full Text PDF PubMed Google Scholar) ALDH1 genes are compared (Fig. 3). The sequence of the human ALDH1 proximal promoter region is very similar to that of marmoset, rat, and mouse. Within the upstream 100-bp region from the transcription start site, a 10-bp deletion (−14/-5) exists in the mouse and rat genes, and a 13-bp deletion (−53/-41) exists in the rat gene. However, CCAAT and ATAAA boxes are well conserved in all species. An octamer binding site is also conserved in the promoter regions of human, marmoset, and mouse genes, suggesting the functional importance of these elements for the transcriptional regulation of the ALDH1 gene. The role of the 5' promoter region in cell type-dependent expression of the ALDH1 gene was examined by expressing CAT constructs containing progressive deletions of the 2536-bp fragments (Fig. 4). In hepatoma (Hep3B) cells, which constitutively expressed ALDH1, the CAT activity with the vector carrying −2536/+53 (pCAT-2536) was 32-fold higher than that of the cells transfected with the promoter less pUMSVOCAT vector (Fig. 5). By contrast, the expression of pCAT-2536 was similar to that of pUMSVOCAT in fibroblast LTK− cells and erythroleukemic K-562 cells, which did not express ALDH1. The cell type-specific activity of the ALDH1 promoter was further evidenced by comparison of the CAT activity yielded by the ALDH1 promoter and that by the SV40 early promoter in different cell lines. The relative activity of the ALDH1 promoter (measured using pCAT-SV40 as reference) was found to be 61.2 in Hep3B, 2.9 in LTK−, and 7.7 in K562 cells, implying that pCAT-2536 stimulates transcription 21 and 8 times more efficiently in Hep3B than LTK− and K562 cells, respectively (Fig. 5). These results indicate that the 5'-flanking region can direct the cell type-specific expression. The progressive removal of the 5'-sequences from −2536 to −673 resulted in augmented promoter activities, suggesting that the 5'-flanking region (−2536/-673) of the ALDH1 gene may contain mild negative elements. Further stepwise deletions of the sequences from −673 to −91 did not significantly affect the reporter gene expression. However, a drastic decrease of CAT activity occurred by deletion from −91 to −50. The CAT activity of the deletion construct pCAT-50 was about 8-fold higher than that of the promoterless plasmid pUMSVOCAT. These results suggest that at least two positive cis-acting regulatory elements exist in the region between −91 to +53. In Hep3B cells, the region from −93/+53, containing ATAAA and CCAAT boxes, functions as a promoter with activity similar to or even greater than the SV40 promoter (Fig. 5). The 5'-flanking region also stimulated the CAT expression in other cell lines. However, in comparison with that in Hep3B cells, the degree of stimulation is very low in these cells, which do not express the ALDH1 (Fig. 5). The construct containing the proximal promoter (pCAT-91) exhibited the CAT activity 65-fold of that of the promoterless pUMSVOCAT in Hep3B cells, but only 6-fold in LTK−, and 7-fold in K562 cells (Fig. 5). These results indicate that the promoter element (−91/+53) directs the cell type-specific expression of the ALDH1 gene. In order to identify the positive regulatory elements within the proximal promoter region (−120/+53), the basal promoter activity of the internal deletion mutants was examined. Deletion from −74 to −70 of the CCAAT box (pCAT-120ΔC) resulted in a 12.5-fold decrease of the CAT activity compared to that expressed by the undeleted pCAT-120, indicating that the binding of a nuclear factor(s) to the CCAAT box is essential for the basal promoter activity in Hep3B cells (Fig. 6). On the other hand, deletion from −32 to −28 of the ATAAA box (pCAT-120ΔA) did not significantly affect the CAT activity, suggesting that the ATAAA box is not a primary regulatory element in Hep3B cells. Gel retardation analysis showed that unlabeled NF-Y, a well characterized human albumin promoter(24Urano Y. Watanabe K. Sasaki M. Tamaoki T. J. Biol. Chem. 1986; 261: 3244-3251Abstract Full Text PDF PubMed Google Scholar), but not CTF/NFI and Sp1 sequences, competed with labeled Oligo I (Fig. 7A, lanes 3-14). Similarly, unlabeled Oligo I but not CTF/NFI and Sp1 oligonucleotides, prevented the binding of the labeled NF-Y to the Hep3B nuclear protein (data not shown). Furthermore, the antibody specific to NF-YB protein, a member of NF-Y transcription factors(23Mantovani R. Pessara U. Tronche F. Li X.-Y. Knapp A.-M. Pasquali J.-L. Benoist C. Mathis D. EMBO J. 1993; 11: 3315-3322Crossref Scopus (165) Google Scholar), disturbed the formation of Oligo I-nuclear factor complex (Fig. 8A, bandNB). These results indicate that the nuclear factor interacting with the CCAAT box region is identical to the nuclear factor NF-Y.Figure 8:Antibody recognition of NF-Y and octamer-binding proteins. A, end-labeled Oligo I was incubated with Hep3B nuclear extracts (lanes 1-3) or K562 nuclear extracts (lanes 4-6) in the presence or absence of the specific antibody and subjected to gel shift assay. Lanes1 and 4, no antibody; lanes2 and 5, with anti-NF-YB antibody; lanes3 and 6, with rabbit Ig G. The supershifted band is indicated by S, and NF-Y•DNA complex is indicated by NB. B, end-labeled Oligo II was incubated with Hep3B nuclear extracts (lanes 1-4) or K562 nuclear extracts (lanes 5-8) in the presence or absence of the specific antibody, and subjected to gel shift assay. Lanes 1 and 5, no antibody; lanes2 and 6, with anti-Oct-1; lanes3 and 7, with anti-Oct-2; lanes4 and 8, with rabbit IgG. The supershift band is indicated by S.View Large Image Figure ViewerDownload Hi-res image Download (PPT) The consensus octamer sequence, ATGCAAT, exists adjacent to the CCAAT box in the ALDH1 promoter (Fig. 2). When Oligo II (−68/-41) was incubated with the nuclear extracts from a slow moving complex, OB1 was observed in the presence of the Hep3B extract, whereas two complexes, OB1 and OB2, were detected in the presence of the K562 extract (Fig. 7B, lanes2 and 9). The unlabeled Oligo II and Oct oligonucleotides abolished the formation of the labeled OB1 and OB2 complexes (Fig. 7B, lanes 3-6 and 11-13). The Sp1 oligonucleotide did not affect the formation of OB1 and OB2 complexes (Fig. 7B, lanes 7, 8, 14, and 15). Similarly, Oligo II and Oct oligonucleotides, but not Sp1, prevented the binding of the labeled Oct oligonucleotide to the nuclear extracts from Hep3B or K562 (data not shown). The formation of OB1 complex was inhibited by the anti-Oct-1 antibody, and a supershift band was produced in the gel retardation analysis (Fig. 8B, lanes2 and 6). The formation of OB2 complex was not disturbed by the anti-Oct-1 antibody and anti-Oct-2 antibody (Fig. 8B, lanes 6-8). From these results, one can conclude that the nuclear protein producing the slow moving OB1 is Oct-1, ubiquitously expressed in various cells (30Scholer H.R. Trends Genet. 1991; 7: 323-329Abstract Full Text PDF PubMed Scopus (330) Google Scholar) and that the second complex, OB2, is produced by another octamer factor, tentatively designated as Oct-X, existing in K562 cells but not in Hep 3B cells. Cooperative binding of these factors to the ALDH1 promoter was examined by the mobility shift analysis using several nucleotide fragments shown in Fig. 9. Four complexes (A, B, C, and D) were produced by incubating Frag I (−91/-1) with the Hep3B nuclear extract (Fig. 10). Competitive binding assay using Oligos I, II, III, and IV revealed that complex B is related to both Oligo I (i.e. NF-Y binding site) and Oligo II (i.e. octamer binding site) (Fig. 10, lanes4 and 5). Oligo I strongly interfered with the formation of complex D, but not complex C, whereas Oligo II disturbed the formation of complex C, but not complex D (Fig. 10, lanes3 and 4). Oligos III and IV and Sp1 did not compete with Frag I in the formation of complexes A, B, C, and D. The results suggest that complex C results from the binding of the octamer factor to the probe, while complex D is a product of NF-Y and the probe. Complex B consists of the probe to which both NF-Y and octamer factors have bound.Figure 10:Binding of nuclear factors at the minimal promoter element (−91 to −1) with Hep3B nuclear extract. A Frag I (region −91 to −1) was labeled and used in a gel mobility shift assay. Lane 1, control without Hep3B nuclear extract; lanes 2-8, with Hep3B nuclear extract. The following lanes were obtained in the presence of unlabeled competitive oligonucleotides; lane3, 50-fold excess of Frag I; lane4, 500-fold excess of Oligo I; lane5, 500-fold excess of Oligo II; lane6, 500-fold excess of Oligo III; lane7, 500-fold excess of Oligo IV; lane8, 500-fold excess of Sp1.View Large Image Figure ViewerDownload Hi-res image Download (PPT) The formation of complex A was abolished by self-competitor, Frag I, and an excess of a truncated oligonucleotide Frag II, but not other oligonucleotides. Complex A was not detectable using K565 nuclear extracts (data not shown). These results indicate that the protein involved in the formation of complex A is cell type-specific and reacts to the region −50 to −1 of the ALDH1 promoter. In order to further elucidate the hepatocyte-specific transcription factor(s), the labeled nucleotide region −50/+53 was incubated with the nuclear extracts from Hep3B or K562 cells and subjected to the mobility shift analysis. Complex L1 was produced by the Hep3B extract but not by the K562 extract (Fig. 11A). The unlabeled Frag I, Frag II, and Frag III abolished the formation of complex L1 (Fig. 10A, lanes 3-5), suggesting that the region −50/-1 bind to the hepatocyte-specific transcription factor(s). However, Oligo III, Oligo IV, and Sp1 did not strongly interfere with the formation of L1 complex (Fig. 10A, lanes6, 7, and 9). To find out a possible involvement of GATA site in the formation of L1, Frag II with altered GATA site (−34/-29, AGATAA → ACCGAA) was used for the gel shift assay. The mutated Frag II abolished the formation of L1 complex (data not shown), suggesting that GATA binding site may not be important for the formation. Interestingly, the unlabeled Oligo V (−14/+20), reduced the formation of complex L1 (Fig. 11A, lane8), suggesting the presence of two binding factors, one for the sequence from −51 to −1, and the other for the sequence from −14 to +20. To confirm this possibility, labeled Oligo V (−14/+20) was used for the mobility shift analysis. When Oligo V was incubated with the Hep3B nuclear extract, a unique complex, L2, was produced (Fig. 11B). Complex L2 was not produced by the K562 nuclear extract. The hepatocyte-specific nuclear factors involved in formation of complexes L1 and L2 are designated as L1F and L2F, respectively. Northern blot analysis indicated that the level of ALDH1 in various types of cells is regulated at the transcriptional level (Fig. 1). Sequence analysis revealed that the extended 5'-region (−2539 to transcription start site) contains various putative regulatory elements (Fig. 2). The putative binding sites for transcription factors such as liver-specific HNF-5 and NF-IL6, muscle-specific MCBF, and hematopoietic cell-specific GATA and Est-1 may be involved in tissue- and cell type-specific expression of ALDH1. Although human ALDH1 is constitutively expressed at a high level in the liver, it is also inducible by phenobarbital in human hepatic cells and the liver of some rat strains(14Marselos M. Strom S.C. Michalopoulus G. Chem. Biol. Int. 1987; 62: 75-88Crossref PubMed Scopus (29) Google Scholar, 15Dunn T.J. Koleske A.J. Lindahl R. Pitot H.C. J. Biol. Chem. 1989; 264: 13057-13065Abstract Full Text PDF PubMed Google Scholar). A conserved 17-bp phenobarbital response element has been identified in phenobarbital-inducible rodent cytochrome P450(31He J.-S. Fulco A.J. J. Biol. Chem. 1991; 266: 7864-7869Abstract Full Text PDF PubMed Google Scholar), mouse glutathione S-transferase Ya(32Daniel V. Tichauer Y. Sharon R. Nucleic Acids Res. 1988; 16: 351Crossref PubMed Scopus (18) Google Scholar), rat ALDH(15Dunn T.J. Koleske A.J. Lindahl R. Pitot H.C. J. Biol. Chem. 1989; 264: 13057-13065Abstract Full Text PDF PubMed Google Scholar), rat epoxide hydrolase (33Falany C.N. McQuiddy P. Kasper C.B. J. Biol. Chem. 1987; 262: 5924-5930Abstract Full Text PDF PubMed Google Scholar), and rabbit cytochrome P450 11C(34Zhao J. Chan G. Govind S. Bell P. Kemper B. DNA Cell Biol. 1990; 9: 37-48Crossref PubMed Scopus (15) Google Scholar). A similar sequence (77% homology) exists in human AHLD1 (Fig. 2). Human ALDH1 is expressed in genital skin fibroblast cells from normal subjects, but not in the cells obtained from androgen receptor-negative testicular feminization patients(12Pereira F. Rosenmann E. Nylen E. Kaufman M. Pinsky L. Wrogemann K. Biochem. Biophys. Res. Commun. 1991; 175: 831-838Crossref PubMed Scopus (40) Google Scholar, 13Yoshida A. Hsu L.C. Yanagawa Y. Adv. Exp. Med. Biol. 1993; 328: 37-44Crossref PubMed Google Scholar, 35Kovacs W.J. Turney M.K. Skinnen M.K. Endocrinology. 1989; 124: 1270-1277Crossref PubMed Scopus (8) Google Scholar). Two putative androgen-responsive elements exist at positions −688/-674 and −322/-306, suggesting the possibility of androgen-dependent regulation of the ALDH1 expression in genital skin cells. It has been suggested that NF-IL6 element and HAPE-1 element are cooperatively involved in the expression of several acute phase genes (36Majello B. Arcone R. Toniatti C. Ciliberto G. EMBO J. 1990; 9: 457-465Crossref PubMed Scopus (145) Google Scholar). These two elements are closely located in the 5'-region from −2200 to −2182 of ALDH1. It is of interest to examine whether or not the two elements are involved in modulation of ALDH1 expression by interleukin-6 cytokine. The present functional studies of the promoter region revealed several regulatory elements and nuclear proteins involved in the cell type-specific expression of ALDH1. The deletion analysis demonstrated that two elements, one between −91 and −51, and the other between −50 and +53, drive the CAT expression in Hep3B cells. Deletion of CCAAT sequence (−74/-70), but not deletion of ATAAA sequence (−32/-28), decreased the promoter activity (Fig. 6). Therefore, CCAAT box-binding protein is critical for promoter activity in Hep3B cells. Previous studies showed that mutations of CCAAT sequence in the albumin and major histocompatibility class II promoters could result in decrease of promoter activity(37Herbowel P. Rollier A. Tronche F. Ott M.-O. Yaniv M. Weiss M. Mol. Cell. Biol. 1989; 9: 4750-4758Crossref PubMed Google Scholar, 38Sherman P.A. Basta P.V. Moore T.L. Brown A.M. Ting J.P.-Y. Mol. Cell. Biol. 1989; 9: 50-56Crossref PubMed Scopus (42) Google Scholar). In contrast with other mammalian species, ALDH1 is hardly expressed in the liver without an inducer in various rat strains, including Long-Evans(15Dunn T.J. Koleske A.J. Lindahl R. Pitot H.C. J. Biol. Chem. 1989; 264: 13057-13065Abstract Full Text PDF PubMed Google Scholar). A part of the octamer motif and adjacent sequences (13 bp) are deleted in rat Long-Evans strain (Fig. 3) and in other common rat strains, 2Y. Yanagawa, J. C. Chen, L. C. Hsu, and A. Yoshida, unpublished observation. 2 indicating the importance of this region for constitutive expression of ALDH1 in the liver. Although multiple factors existing in Hep3B cells, such as C/EBP, CTF/NF1, and NF-Y, could interact with CCAAT motif(39McKnight S. Tjian R. Cell. 1986; 46: 795-805Abstract Full Text PDF PubMed Scopus (377) Google Scholar, 40Landschultz W. Johnson P.F. Adashi F.Y. Graves B.J. McKnight S. Genes & Dev. 1988; 2: 786-800Crossref PubMed Scopus (630) Google Scholar, 41Gil G. Smith J.R. Goldstein J.L. Slaughter C.A. Orth K. Brown M.S. Osborne T.F. Proc. Natl. Acad. Sci. U. S. A. 1988; 85: 8963-8967Crossref PubMed Scopus (137) Google Scholar, 42Hooft van Huijsduijnen R. Li X.Y. Black D. Matthes H. Benoist C. Mathis D. EMBO J. 1990; 9: 3119-3127Crossref PubMed Scopus (211) Google Scholar), the mobility shift assay strongly supports NF-Y as the primary factor interacting with the CCAAT region (Figure 8:, Figure 9:, Figure 10:). Moreover, the octamer region adjacent to the CCAAT box binds to nuclear proteins. Interestingly, only an ubiquitous transcription factor, Oct-1, was found in Hep3B nuclear extracts, while another octamer binding factor, termed Oct-X, was detected in K562 nuclear extracts in addition to Oct-1. It was reported that neuronal Oct-2 does not activate reporter constructs containing an octamer motif, but it could interfere with the activation of such constructs by Oct-1(43Dent C.L. Lillycrop K.A. Estridge J.K. Thomas N.S.B. Latchman D.S. Mol. Cell. Biol. 1991; 11: 3925-3930Crossref PubMed Scopus (36) Google Scholar). It is conceivable that Oct-X, like Oct-2, is an inhibiting factor, competing with Oct-1 at the octamer binding site of ALDH1 in K562 cells. Both NF-Y and octamer factor(s) appeared to bind with the promoter region (−91/-1, Frag I), producing a complex B (Fig. 10). Synergy between NF-Y motif and an adjacent C/EBP site was observed in the expression of albumin gene(44Milos P.M. Zaret K.S. Genes & Dev. 1992; 6: 991-1004Crossref PubMed Scopus (109) Google Scholar), between Oct-1 and glucocorticoid receptor in activation of the mouse mammary tumor virus promoter (45Bruggemeier U. Kalff M. Franke S. Scheidereit C. Beato M. Cell. 1991; 64: 565-572Abstract Full Text PDF PubMed Scopus (161) Google Scholar) and between Oct-1 and SP-1 in activation of U2 small nuclear RNA gene (46Janson L. Pettersson U. Pro. Natl. Acad. Sci. U. S. A. 1990; 87: 4732-4736Crossref PubMed Scopus (86) Google Scholar). A similar synergism may exist between NF-1 and Oct-1 motifs in ALDH1 expression. CAT assay indicates that the ALDH1 promoter region −91/-50 is functionally active not only in Hep3B cells but also in LTK− and in K562 cells (Fig. 5). However, judging from the fact that the stimulation is much higher (2 orders of magnitude) in Hep3B cells compared with LTK− and K562 cells, the promoter region −91/-50 is essential but is not sufficient to direct the hepatocyte-specific expression of ALDH1. The mobility shift analysis demonstrated the existence of two unique nuclear proteins (L1F and L2F) producing L1 and L2 complexes, in Hep3B cells but not in K562 cells (Fig. 11). Competitive mobility shift studies indicated that L1F binding was only partially suppressed by Oligo V, which completely abolished L2F binding. The two nuclear proteins, L1F and L2F, might belong to the same family of transcription factors and play a major role in the high level expression of ALDH1 in Hep3B cells. Alternatively, the cell type-specific expression could be due to the presence of different NF-Y related proteins or octamer factors in different cell types(47Tafuri S. Wolffe A. Proc. Natl. Acad. Sci. U. S. A. 1990; 87: 9028-9032Crossref PubMed Scopus (163) Google Scholar). It is also conceivable that the concentration and ratio of ubiquitous transcription factors in particular cell types affect the gene expression(48Diamond M.I. Miner J.N. Yoshinaga S.K. Yamamoto E.R. Science. 1990; 249: 1266-1272Crossref PubMed Scopus (1070) Google Scholar, 49Liu Y.H. Yang N.Y. Teng C.T. Mol. Cell. Biol. 1993; 13: 1836-1846Crossref PubMed Scopus (97) Google Scholar, 50Mietus-Snyder M. Sladek F.M. Ginsburg G.S. Kuo C.F. Ladias J.A. Darnell J.E. Karathanasis S.K. Mol. Cell. Biol. 1992; 12: 1708-1718Crossref PubMed Scopus (231) Google Scholar). Based on the present structural and functional analysis, the following possible molecular mechanism can be proposed for the cell type-specific expression of ALDH1 gene. In ALDH1-positive Hep3B cells, the factors involved in the ALDH1 expression are NF-Y and Oct-1 binding to the promoter region −91/-50, and L1F and L2F acting on the promoter region −50/+20. In ALDH1-negative K562 cells, NF-Y, Oct-1, and Oct-X act on the promoter region −91/-50, but L1F and L2F factors are missing and thus ALDH1 cannot be strongly expressed in the cells. Oct-X might interfere with the binding of Oct −1 to the octamer sequence and suppress the promoter activity in K562 cells. The molecular mechanism of the androgen-receptor mediated expression of ALDH1 remains to be elucidated. We thank Dr. K. Kurachi for providing us with the CAT vector and Dr. D. Mathis for the anti-NF-Y antibody. We are also grateful to Vibha Davé for assistance and to Dr. S. Tamura for encouragement.
Referência(s)