Lactase-phlorizin hydrolase (LPH) synthesis is restricted to differentiated small intestinal enterocytes and is highly regulated during development. Analysis of expression of LPH promoter segments fused with luciferase transfected in Caco-2 cells, a line that uniquely expresses LPH mRNA, mapped an 18-base pair (bp) segment 100 bp upstream of the transcription start site that is required for transactivation. Remarkably, the LPH upstream element (LUE) has no stimulatory activity in both human intestinal and nonintestinal lines in which LPH mRNA is absent. Electrophoretic analysis of sequence-specific DNA-nuclear protein complexes demonstrated the presence of a Caco-2 cell-specific protein(s) (CCP), which is uniformly absent in LPH nonproducer cell lines. Mutational analysis of the LUE demonstrated that bases contained within a GATA consensus motif are critical for both CCP binding and transcription from the LPH promoter. Caco-2 cells express high levels of GATA-6 mRNA in a cell line- specific manner, suggesting that GATA-6 is a CCP that complexes with the LUE. When expressed by a plasmid, GATA-6 transactivated the LPH promoter. The stimulation was abrogated with mutations in the GATA consensus motif as well as mutations in a flanking downstream element. These studies are consistent with an important role of an intestinal GATA binding protein in cell type-specific transactivation of the LPH promoter.
- lactase-phlorizin hydrolase
- Caco-2 cells
- GATA binding proteins
the small intestinal epithelium is composed of a self-renewing monolayer of cells. As cells migrate from the crypts to the villus tips, they cease to proliferate and terminally differentiate before apoptotically dying and exfoliating. A single stem cell in each crypt gives rise to four distinct epithelial lineages, which include Paneth cells, goblet cells, enteroendocrine cells, and absorptive enterocytes (11). Each lineage of cells expresses a distinct repertoire of molecules that are required for physiological functions of the intestine.
To assimilate lactose, differentiated absorptive enterocytes express lactase-phlorizin hydrolase (LPH), an enzyme that hydrolyzes the β-1,4-galactoside bond of the disaccharide into its monosaccharide components. Similar to other glycohydrolases, whose expression is restricted to absorptive enterocytes, LPH transcription is regulated during development (29). LPH expression on the surface of apical membrane microvilli, which commences at the crypt-villus junction, is programmed in enterocytes both during maturation and development and is unaffected by the presence of substrate in the lumen (20). In the human LPH promoter, DNA sequences upstream of a single TATA box possess binding sites for transcription factors that are expressed in a variety of cell lineages, including Sp1, SRF AP-2, CTF/nuclear factor-1, cyclic adenosine monophosphate response element binding protein, and Oct1/Oct2 (4). In addition, there is a GATA motif, ∼60 bp upstream of the TATA box. This motif is conserved in the LPH promoters of all the mammalian species that have so far been analyzed (4, 5, 27).
Members of the family of GATA binding proteins all contain a segment with two conserved zinc finger DNA binding domains. These target the consensus sequence (A/T)GATA(A/G). GATA-1, GATA-2, and GATA-3 play a critical role in the regulation of hematopoiesis and are also expressed in nonhematopoietic tissues, such as Sertoli cells (GATA-1), endothelial cells (GATA-2), and nerve tissue (GATA-2 and GATA-3) (30).
Recently, the cDNAs of three new GATA binding proteins have been cloned (1, 17, 21). The mRNAs of GATA-4, GATA-5, and GATA-6 are expressed primarily in the heart and gut. Each of these transcripts encodes amino acid sequences in the zinc finger domains, which are conserved across species (1, 17, 26, 33). Both GATA-4 and GATA-6 are expressed in mammalian enterocytes. Although overlapping, the developmental and tissue-specific patterns of their expression differ significantly. In vitro, studies have indicated that GATA-4 contributes to the regulation of genes crucial for myocardial function and for extracellular matrix formation in the yolk sac (3, 13, 23). Targeted disruption of GATA-4 in mouse embryonic stem cells resulted in the loss of visceral endoderm formation (23). The rat homologue of GATA-6 has been shown to bind to and activate the promoters of both the α- and β-subunit genes of the H+-K+-adenosinetriphosphatase enzyme in the gastric parietal cells of rats and humans (26). These findings, along with the presence of conserved GATA elements in the promoters of intestinal fatty acid-binding protein and sucrase-isomaltase, suggest that GATA binding proteins may play a significant role in the regulation of some gut-specific genes. In this report, we show evidence that GATA-6 may play an important role in stimulating cell type-specific activation of the human LPH promoter.
MATERIALS AND METHODS
Cell lines, plasmids, RNA purification, and DNA sequencing.
Caco-2 and HT-29 human colon adenocarcinoma cells, SW-13 human adrenal carcinoma cells, and cervical carcinoma HeLa cells (purchased from the American Type Culture Collection) and human jejunal HIE-7 cells (24) were cultured in Dulbecco’s modified Eagle’s medium with 10% heat-inactivated fetal bovine serum. A 5.5-kb segment of the human LPH gene that contains 3.5 kb of the promoter region, exon 1 and part of intron 1, was excised from λLPH 7 (4) (generously provided by N. Mantei, Swiss Federal Institute of Technology, Zurich, Switzerland), using Sac I andSal I, and cloned into pUC 19, to produce p19/5.5.
Plasmid DNA was purified using columns from QIAGEN (Chatsworth, CA) and quantitated spectrophometrically and by electrophoretic analysis. RNA isolation was performed as described (8). Polymerase chain reaction (PCR) amplifications were performed usingTaq DNA polymerase, 25 pmol primers, and 100 ng of LPH promoter containing plasmid templates.
Oligonucleotide primers with linker sites were used to amplify increasingly long upstream segments of the LPH promoter whose 3′ end corresponded to the previously designated −1 position (relative to the LPH translation start site) (4). For this purpose, the 3′ primer LPH (3′) (5′-TAGCTAAGCTTGTCGACTTTCTAGGAACTGTTAGGAGG-3′) was paired with 5′ primers that included LPH-315 (5′-CTACAGGCGCATGCCACGATGCCTG GCTAA-3′), LPH-95 (5′GCAGGATCCTTAAATATTAAGTCTTAATTA-3′), LPH-85 (5′CAAGCTCGAGTCTTAATTATCACTTAG-3′), and LPH-46 (5′-CAAGCTCGAGTTATAAAGTAAGGGTTCC-3′). PCR segments with 5-bp-long blocks of mutations in a segment between −112 and −93 were produced with the following 5′ primers (mutated sequence shown in bold lowercase letters): LPH-mutA (5′-GCAGGATCCtcgacTAGATAACCCAGTTAAA-3′), LPH-mutB (5′-GCAGGATCCGATCAgctcgAACCCAGTTAAA-3′), LPH-mutC (5′-GCAGGATCCGATCATAGATccaaaAGTTA- AA-3′), and LPH-mutD (5′-GCAGGATCCG ATCATAGATAACCCctggcAA-3′). For electrophoretic mobility shift assays (EMSAs), the same mutations were introduced into annealed 22-bp double-stranded oligonucleotide probes, whose corresponding wild-type upper strand sequence was 5′-GATCATAGATA ACCCAGTTAAA-3′, to yield LPH upstream element (LUE)/mutA, LUE/mutB, LUE/mutC, and LUE/mutD. The PCR products listed above containing the wild-type and mutant sequences were cloned into the polylinker of pXP1 (Pharmacia Biotech) to produce LXP/−315, LKP/−95, LXP/−85, LXP/−46, LXP/mutA, LXP/mutB, LXP/mutC, and LXP/mutD, respectively. An LPH promoter restriction fragment encompassing LPH bases −3500 and −1 was also cloned into pXP1 to produce LXP/−3500. By restriction of a Bam I site in the polylinker of LXP/−315 and of unique sites in the LPH promoters at either −200 (BstX I) or −112 (Bcl I), LXP/−200 and LXP/−112 were produced. LXP/−315Δ, a plasmid with an internal deletion between bases −112 to −46, was produced by ligation of a plasmid fragment encompassing bases −315 to −112 to the linearized 5′ end of the LPH segment of LXP/−46 (site of the TATA box). DNA sequences of the LPH promoter segments cloned into pXP1 were analyzed with the Sequenase 2.0 kit (U.S. Biochem).
Transient transfections of Caco-2 cells were performed by electroporation, using a Bio-Rad gene pulser system and capacitance extender set at 250 V and 960 μF. Transfections were performed with 2 × 107 cells, 40 μg of a selected LXP reporter construct, and 2 μg of pOBCAT0 (2). Cells grown to 80% confluence were trypsinized, thoroughly suspended in 0.5 ml of cell culture media, chilled to 4°C, and aliquoted in 0.4-cm electroporation cuvettes (Bio-Rad). After electroporation, the cells were cultured for 42 h in petri dishes 100 mm in diameter. To prepare cellular extracts from transfected Caco-2 cells, the adherent and nonadherent cells were harvested together. At the time of harvest for analysis of reporter expression, the adherent cells were 60–80% confluent and growing in clusters of 10–20 cells. In some experiments, 10 μg of an RSV/chicken GATA-6 expression vector (9) (generously provided by J. B. E. Burch, Fox Chase Cancer Center, Philadelphia, PA) was cotransfected with 35 μg of an LXP construct and 2 μg of pOBCAT0 into cells. Identical amounts of DNA were also transiently transfected into Cos-1 cells, by calcium phosphate transfection. Each culture of transfected cells was lysed in 100 μl of buffer containing 1% Triton X-100 and 0.1 M potassium phosphate (pH 7.8) and stored at −70°C. To measure chloramphenicol acetyltransferase (CAT) expression, we mixed 10 μl of the cleared supernatants in scintillation vials with 200 μl CAT reaction buffer containing 10 μl [3H]acetyl-CoA (Dupont NEN) diluted 1:35 with 2.5 mM acetyl-CoA, 50 μl 5 mM chloramphenicol, and 140 μl 0.1 M tris(hydroxymethyl)aminomethane (Tris) ⋅ Cl (pH 7.7). The assay mix was gently overlaid with 2 ml of Econofluor scintillation fluid and immediately placed in a 1209 Rack Beta liquid scintillation counter. The samples were then counted at 1-h intervals. Sample readings of reaction rates were used to establish transfection efficiencies. Based on transfection efficiencies, standardized volumes of each extract (range 4–20 μl) were then analyzed for luciferase content, using a Monolight 2010 luminometer (Analytic Bioluminescence Laboratories). For each cell line tested, differences in the volumes of extracts that were tested were less than twofold.
Ribonuclease protection, reverse transcription-PCR, and Northern blot analysis.
To detect LPH mRNA from human enterocytes and Caco-2 cells32P radiolabeled riboprobes were made from a 578-bp LPH segment (bases −198 to +380) inserted into Bluescript SK and transcribed from the T3 and T7 promoters to produce sense and antisense probes, respectively. To detect reporter mRNA from Caco-2 cells transiently transfected with LXP plasmids, riboprobes were made from a Bluescript SK plasmid containing a 208-bp segment between bases −112 of LPH and +95 of the luciferase reporter. After linearization of the plasmid, the probes were transcribed from the T7 and T3 promoters to produce sense and anti-sense probes, respectively. After hybridization, the mRNA-riboprobe duplexes were treated with ribonucleases (RNases) A and T1, denatured, and separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis. The gels were then dried and autoradiographed. To determine the presence of LPH mRNA in cell lines, RNA samples were reverse transcribed with random hexamer primers and reverse transcriptase (RT). The RT products were then amplified with LPH primers to generate a 225-bp segment between +15 and +237 (5′ primer: 5′-GAGCTGTCTTGGCATGTAGT-3′; 3′ primer: 5′-GGAGACTGCTGAAGTACTCTGGC-3′), which was detected by agarose gel electrophoresis. For Northern blot analysis of GATA-6 mRNA, electrophoretically separated total RNA was transferred to a nylon membrane. GATA-4 mRNA was detected with a 44-base oligonucleotide probe (AGGCTGTGCAGGACCGGGCTGTCGAAGGGGCCGGCG- GAGGCGGC), which is complementary to both human and rodent GATA-4 forms, but not to other GATA-binding protein family members (1, 21, 26, 33), after32P end-labeling with T4 polynucleotide kinase. A human GATA-6 cDNA probe (kindly provided by T. Evans, Albert Einstein College of Medicine, Yeshiva University, Bronx, NY), consisting of a 1.5-kb EcoR I fragment not overlapping with the conserved zinc finger DNA binding domain, was 32P radiolabeled with random hexamers and the Klenow fragment ofEscherichia coli DNA polymerase, and hybridized to membrane-bound mRNA. After washing, the membranes were dried and autoradiographed.
Nuclear extract preparation and analysis of protein-DNA binding.
Nuclear extracts of cell lines were prepared according to a modified protocol of Dignam et al. (10), and extracts were stored at −70°C. Briefly, after being rinsed and scraped in cold phosphate-buffered saline (PBS), the cells were centrifuged. The cell pellet was then resuspended in an equal volume of cold 2×buffer A [10 mMN-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid (HEPES), pH 7.9, 1.5 mM MgCl2, 10 mM KCl, 0.25 mM dithiothreitol (DTT), 100 μg/ml phenylmethylsulfonyl fluoride (PMSF), 1 μg/ml leupeptin, 1 μg/ml pepstatin, and 1 μg/ml aprotinin]. After the cells were swollen on ice and disrupted by passage through a 26-gauge needle, the mixture was layered over a sucrose cushion (0.8 M sucrose, 10 mM Tris ⋅ Cl, pH 7.5, 80 mM KCl, 5 mM MgCl2, 0.25 mM DTT, and the protease inhibitors listed above) and centrifuged at 2,000 revolutions/min (rpm) for 15 min. The nuclear pellet was resuspended in an equal volume of 2×buffer C (20 mM HEPES, pH 7.9, 420 mM NaCl, 1.5 mM MgCl2, 200 mM EDTA, 0.25 mM DTT, and the protease inhibitors listed above in 20% glycerol). The mixture was then centrifuged at 37,500 rpm for 30 min, and the supernatant was stored at −70°C. Protein quantitation was determined by the Bradford protein assay (Bio-Rad) (7). EMSAs were performed using plasmid restriction endonuclease fragments or double-stranded DNA oligonucleotides that were32P end-labeled with T4 polynucleotide kinase. The radiolabeled double-stranded oligonucleotides or restriction fragment probes (0.2–0.4 ng) were added to 8 μl of EMSA binding buffer [25 mM Tris ⋅ Cl, 200 mM glycine, 1 mM EDTA, 10% glycerol, 1 mM 2-mercaptoethanol, 50 mM NaCl, 0.025% bromophenol blue, and 1 μg poly(dI-dC)] before the addition of 1 μl nuclear extract (5 μg/sample). The mixtures were incubated at room temperature for 30 min and then loaded onto 5% polyacrylamide gels. After separation of the protein-DNA complexes, the gels were dried and autoradiographed.
The LPH promoter is only activated in Caco-2 cells.
To identify a human cell line that endogenously expresses LPH mRNA, PCR amplification was performed on the RT products of total RNA isolated from intestinal and nonintestinal cell lines using primers from LPH exon 1. Of the five cell lines tested, only Caco-2 cells exhibited expression of the LPH transcript (Fig.1 A). This human colon carcinoma cell line forms a monolayer of polarized cells, after growing to confluence, and expresses a number of proteins associated with the brush borders of small intestinal enterocytes (14, 15). LPH mRNA expression is low in preconfluent Caco-2 cells and increases significantly in postconfluent culture, as the cells become more differentiated (Fig. 1 B). Although the electrophoretically separated RT-PCR product of preconfluent cell LPH mRNA was not detected by ethidium bromide staining, after Southern blotting it was detected by hybridization with a radiolabeled LPH cDNA probe (data not shown). A series of LPH promoter segments of increasing length were fused with luciferase (Fig. 2). To determine whether LPH promoter activity is restricted in a cell line-specific manner, the LXP fusion constructs were transiently transfected into Caco-2 cells as well as the other four cell lines tested above (Table1). Luciferase expression was then analyzed after standardization to CAT levels generated by a cotransfected plasmid containing the CAT gene linked to the SV40 early promoter. All the cell lines tested were efficiently transfected with this CAT reporter plasmid and exhibited high levels of CAT expression. In sharp contrast, transcription from the LPH promoter was only measurable in Caco-2 cells. In these LPH-producing cells LPH promoter activity of the LXP constructs was high, whereas the promoterless construct was transcriptionally silent (Table 1). Initial promoter mapping, using the LXP constructs with increasingly extended LPH promoter segments, indicated that the region required for transcriptional activation in Caco-2 cells is between the LPH TATA box (LXP/−46) and a site 66 bp further upstream (LXP/−112). Consistent with this finding, a construct with an internal deletion between the TATA box and −112 (LXP/−315Δ) was transcriptionally inactive. These experiments revealed the unique ability of Caco-2 cells to activate transcription from the LPH promoter.
Analysis of transcription start sites of the endogenous and transfected LPH promoters.
To determine if the 5′ end of LPH mRNA in Caco-2 cells is consistent with the transcription start site in human primary enterocytes, RNase protection analysis was performed. We hybridized 10 μg of total RNA from both human enterocytes and Caco-2 cells to a 578-bp riboprobe made from the LPH bases −198 to +380. Nuclease digestion resulted in a 399-bp protected fragment from both samples, suggesting that identical LPH transcription start sites are utilized 15 bp upstream of the translation start site (Fig. 3). This site is within a 5-bp segment that was previously found to encompass the transcription start site (4). To determine whether the appropriate LPH promoter sequence was utilized by the transfected LXP constructs, mRNA from Caco-2 cells transiently transfected with LXP/−315 was analyzed by RNase protection. For this purpose, a 207-bp riboprobe, which consisted of 112 bp of the LPH promoter segment fused with a 95-bp segment of the reporter plasmid, was used. RNase protection analysis of mRNA from transfected cells demonstrated a 114-bp protected fragment, which was absent when mRNA from nontransfected cells was analyzed (Fig. 4). The size of the protected fragment is consistent with appropriate usage of the LPH transcription start site in the transfected LPH-luciferase gene.
Identification of an 18-bp LPH upstream segment that is required for LPH transcription and that binds a Caco-2 cell line-specific nuclear protein(s).
To locate a regulatory element within the 66-bp region required for the activation of the LPH promoter in Caco-2 cells, LXP constructs were made that contained segments of the LPH promoter 39 and 49 bp upstream of the TATA box (LXP/−85 and LXP/−95, respectively). When transiently transfected into Caco-2 cells, these constructs exhibited no significantly greater activity than the TATA box construct LXP/−46 (Fig. 5). This result indicated that the crucial area for LPH transcription activation is in an 18-bp segment that is located between −95 and −112 bp upstream of the LPH translation start site. This segment has been designated the LUE.
It was of interest to determine if Caco-2 cells produce a protein(s), absent in the other tested cell lines, that binds the LUE in a sequence-specific manner. Analysis of proteins in Caco-2 cell nuclei that bind in the vicinity of the LUE was performed using a32P radiolabeled 22-bp double-stranded oligonucleotide encompassing the 18-bp LUE (Fig.6 A). Binding to this probe was also tested by EMSA with nuclear extracts from other cell lines in which the LPH promoter is not transactivated. As shown in Fig.6 B, Caco-2 cells contain a protein(s) that binds the LUE in a sequence-specific manner. In contrast, none of the other cell lines contain this protein. To determine which bases are necessary for binding by the Caco-2 cell line-specific protein(s) (CCP), excess molar amounts of 22-bp double-stranded oligonucleotides encompassing the LUE sequence with tandem blocks of 5-bp mutations (Fig. 6 A) were used as cold competitors for CCP binding to the radiolabeled LUE. Binding of the CCP to the LUE was effectively competed by the wild-type sequence as well as by sequences with base pair substitutions in segments containing bases −112 to −108 (LUE/MutA) and bases −97 to −93 (LUE/MutD) (Fig. 6 B). In contrast, mutations with base pair substitutions containing bases −107 to −103 (LUE/MutB) and −102 to −98 (LUE/MutC) reversed efficient competition of the excess oligonucleotides with the radiolabeled LUE for binding to the CCP. The loss of CCP binding indicates the crucial role of the 10-bp site within the LUE in forming a DNA sequence-specific complex with the nuclear protein. In the EMSA, a series of protein-DNA complexes that migrated more rapidly than the CCP-LUE complex were also observed. By competition with excess wild-type and mutant LUE sequences, these were not DNA sequence specific. To confirm that the LUE-containing LPH promoter binds the CCP in a sequence-specific manner, a restriction fragment containing bases −112 to −1 was added in excess molar quantities to the binding reaction. This fragment successfully competed with the 22-bp radiolabeled LUE probe for CCP binding. In contrast, a nonspecific 125-bp restriction fragment was ineffective as a competitor (Fig. 6 C), corroborating that the CCP binds to the double-stranded LPH promoter in a sequence-specific manner.
Analysis of the LUE DNA sequence indicated that the 10-bp segment that is crucial for CCP binding contains a GATA consensus site (WGATAR). As described above, the binding analysis suggested that the CCP is a GATA binding protein. Members of this family of factors are well known for their tissue-specific and developmental regulation of transcription. GATA-4 and GATA-6 are expressed in chicken as well as in human and mouse intestinal epithelial cells and may play a significant role in the tissue-specific transcriptional activation of gastrointestinal genes (1, 17, 21-23, 26).
LUE bases within and 3′ of the GATA motif are required for promoter activity.
If CCP binding to the LUE is necessary to stimulate the LPH promoter, it is predicted that mutations that abolish the protein-DNA interaction also block transactivation. To test this prediction, four LXP constructs were made that incorporated each of the four tandem blocks of LUE mutations (LXP/MutA, LXP/MutB, LXP/MutC, and LXP/MutD), previously analyzed for binding in Fig.6 B. Expression by each of these altered LPH promoter-luciferase constructs in Caco-2 cells was compared with expression by the wild-type LUE reporter construct (LXP/−112) (Fig. 7). Although none of the constructs containing mutated sequences demonstrated transcriptional activity as high as the wild-type control (LXP/−112), the construct with mutations in the segment encompassing bases −112 to −107 (LXP/MutA) reduced transcriptional activity by only 1.5-fold (Fig. 7). In contrast, mutations that block CCP binding (LXP/MutB and LXP/MutC) reduced transcriptional activity more dramatically, consistent with a critical role for binding of the CCP in LPH promoter activation. Unexpectedly, LXP/MutD, which contains mutated bases located downstream of the GATA element that do not disrupt CCP binding, was also transcriptionally silent. Interestingly, the mutated bases in LXP/MutD are part of a DNA sequence in which 11 of 13 bases are identical with the reported consensus sequence for hepatocyte nuclear factor-1 (HNF-1) (28, 31). Although contiguous, the GATA element is not overlapping with this site.
GATA-6 stimulates the LPH promoter.
GATA-4 and GATA-6 mRNA are both expressed in small intestinal epithelial cells. Compared with the rat heart mRNA, none of the cell lines tested by Northern analysis, including Caco-2 cells, expressed significant levels of GATA-4 mRNA (Fig.8 A and data not shown). In contrast, Caco-2 cells produce high levels of GATA-6 mRNA. Expression of GATA-6 transcripts at equivalent levels by the LPH nonproducer cell lines was not observed. The cell line-specific expression of high amounts of GATA-6 mRNA parallels the restricted expression of CCP detected by EMSA analysis. These data suggest that GATA-6 is a CCP that binds the LUE in vivo. Because of the difference between GATA-4 and GATA-6 mRNA levels, it is unlikely that GATA-4 is as abundant as GATA-6 in the CCP-LUE complex. This is supported by the observation that mobility of the CCP-LUE bound product, detected by EMSA analysis, is not affected by GATA-4 “supershifting” antibodies (data not shown). Unlike LPH expression, GATA-6 mRNA synthesis in Caco-2 cells is constitutive and is not influenced by confluence-induced differentiation (Fig.8 B).
To test whether GATA-6 stimulates the LPH promoter, LXP wild-type and mutant reporter constructs were cotransfected into Caco-2 cells with a GATA-6 expression plasmid (pRSV/GATA-6). Consistent with the results shown in Fig. 7, both the wild-type (LXP/−112) and LXP/MutA promoters demonstrated low levels of basal activity in the absence of cotransfected pRSV/GATA-6 (5- and 6-fold mean increases in activity, respectively, relative to the activity of a promoterless reporter;n = 3; data not shown). Cotransfection of pRSV/GATA-6 significantly stimulated LPH promoter activity of both reporter constructs (Fig.9 A). Not surprisingly, the constructs with mutations overlapping the GATA motif (LXP/MutB and LXP/MutC) did not respond to the stimulatory effects of the product of the GATA-6 expression vector. The mutated LPH promoter with base pair substitutions on the 3′ flank of the GATA motif (LXP/mutD) was also not susceptible to transcriptional activation by GATA-6.
Similarly, cotransfection of the LXP constructs with pRSV/GATA-6 into Cos-1 cells, which do not express GATA-6 mRNA endogenously, stimulated luciferase expression (Fig. 9 B). Moreover, transcription of the LXP/MutD promoter, which contains mutated bases 3′ of the GATA consensus sequence, was not stimulated by GATA-6. In contrast to their low levels of activity in Caco-2 cells, when the GATA-6 expression plasmid was absent, both the wild-type (LXP/−112) and LXP/MutA promoters were silent in Cos-1 cells (data not shown).
The coincident expression of GATA-6 and LPH mRNA in Caco-2 cells and the ability of a GATA-6 expression vector to stimulate transcription from cotransfected LXP plasmids establish a connection between the GATA binding protein and LPH promoter activation. Consistent with an activating role for GATA-6, mutations that altered the GATA binding site also abrogated promoter activation in Caco-2 cells. The GATA sequence in the LUE, AGATAA, is identical to a sequence to which GATA-6 is known to bind preferentially, as measured by selected and PCR-amplified binding (J. B. E. Burch, personal communication). When linked to a minimal promoter, this sequence has been found to effectively mediate GATA-6-induced transactivation (9). Further evidence of the activation of the LPH promoter by GATA-6 was demonstrated by transiently cotransfecting the GATA-6 expression plasmid and LXP reporter constructs into Cos-1 cells, which ordinarily do not express significant amounts of GATA-6. In association with the absence of GATA-6 expression in these cells, the transfected LPH promoter was transcriptionally silent. However, when GATA-6 was produced by a transfected expression plasmid, LPH promoters containing wild-type sequences or a block of substitution mutations 5′ to the GATA motif (LXP/MutA) in the LUE were strongly activated. In contrast, the promoters with mutations overlapping the GATA site (LXP/MutB and LXP/MutC) were not stimulated by GATA-6 to produce significant levels of luciferase.
In both Caco-2 cells and small intestinal enterocytes, the expression of GATA-6 alone is not sufficient to stimulate LPH transcription. This is supported by the observation that the level of GATA-6 mRNA expression is the same in preconfluent and confluent Caco-2 cells, whereas LPH mRNA expression is low in preconfluent cells and dramatically increases in 7- and 14-day postconfluent cells. There are also important differences in the time of onset of both GATA-6 and GATA-4 mRNA expression and LPH mRNA expression during development of the mouse. In the primitive gut, the transcripts that encode these GATA binding proteins are detectable as early as embryonicday 9.5(21). In contrast, LPH mRNA only appears after the cytodifferentiation of endodermal cells into a monolayer of polarized epithelium, which occurs several days later (29). Thus, in addition to GATA binding protein expression, another differentiation-associated event(s) is required to stimulate transcription from the genomic LPH promoter, in both preconfluent Caco-2 cells and undifferentiated small intestinal epithelial cells. This may be an alteration(s) in chromatin structure or a modulation(s) of another nuclear protein(s) that activates or suppresses LPH transcription.
The inability of GATA-6 to activate the LPH promoter with mutations downstream of the GATA binding site (LXP/mutD) contrasts with binding studies that indicate that the GATA-binding CCP is able to form a stable complex with the LUE containing these mutations (LUE/mutD). There are two possible explanations that may account for this discrepancy. First, the binding affinity of a putative GATA-6/LUE complex in vivo might be modulated by bases 3′ to the GATA motif. Previously, it has been observed that bases flanking the GATA motif affect the binding of GATA-1, GATA-2, and GATA-3 (16, 18). These proteins demonstrate base preferences that affect the binding affinities at bases +3 and +4 on the 3′ end of the GAT core. Moreover, cooperative GATA protein binding to second contiguous sites composed of a variety of sequences that deviate from the consensus sequence has been demonstrated. However, it is unlikely that the base pair substitutions in LUE/mutD cause a significant change in the affinity of the GATA binding protein-DNA interaction, since the mutated bases 3′ to the GATA motif, which disrupted LPH promoter transactivation, did not affect formation of the CCP-LUE complex in vitro. This was demonstrated by the identical efficiency of competition by titrated amounts of excess LUE/mutD and LUE/wild-type (LUE/WT) sequences for binding of the CCP with the radiolabeled LUE/WT.
A second explanation for the discrepancy between preservation of CCP binding to LUE/mutD and the loss of transactivation by LXP/MutD is that GATA-6 may require the presence of a downstream accessory factor to activate the LPH promoter. Analogously, GATA-1 has previously been shown to functionally interact with SP1 and erythroid Kruppel-like factor (12). It has recently been demonstrated that GATA-6 functionally cooperates with other transcription factors (9). In the vitellogenin II promoter, estrogen-induced transcription depends on the contribution of this GATA binding protein, which binds a site flanking the estrogen receptor element. In the case of the LUE, accessory factor binding may be a necessary step for LPH promoter transactivation. By EMSA analysis, we have found that a second Caco-2 cell protein binds a separate site that is immediately 3′ of the GATA binding site (data not shown). This site (GTTAAATATTAAG) includes 11 of 13 bases that are identical with the reported consensus motif for HNF-1 (GTTAATNATTAAC) (28, 31). In vitro binding of the second Caco-2 cell nuclear protein at the HNF-1-like site in the LPH promoter occurs independently of binding at the GATA element by CCP and is selectively disrupted by the base substitutions identical with those in LUE/mutD. Both the related homeoproteins HNF-1α and HNF-1β have been identified in primary enterocytes and Caco-2 cells (31). These transcription factors, along with CDX homeodomain proteins, were shown to regulate the small intestine- specific product of the sucrase-isomaltase gene (SI) (31). Consistent with a role in the activation of LPH expression in the small intestine during murine development, HNF-1α mRNA is expressed by embryonic day 12.5(19), which precedes LPH mRNA expression by a few days. The possibility that a GATA binding protein(s) cooperates with a member(s) of the HNF-1 family to stimulate the LPH promoter awaits further studies.
The region between the GATA motif in the LUE and the TATA box of LPH contains other elements known to bind nuclear transcription factors. A second GATA binding site located 77 to 72 bp upstream of the LPH translation start site, which has previously been designated as +1 and is ∼62 to 57 bp upstream of the LPH transcription start site (4), does not stimulate transcription from the LPH promoter in Caco-2 cells, as measured in the transfection analysis. This element, whose orientation is opposite of the GATA element in the LUE, may not be positioned to effectively interact with other critical accessory protein binding sites and/or the RNA polymerase complex. The CE-LPH1 element is located 63 to 51 bp upstream of the LPH translation start site (27). This element has been shown to bind an intestinal nuclear protein (NF-LPH1), whose expression in pig enterocytes changes in parallel to lactase expression. Although it may also play a role in LPH activation, the range of NF-LPH1 expression does not parallel the cell type-restricted pattern of LPH mRNA synthesis (6).
Whether GATA-4, which similar to GATA-6 is expressed in the small intestine, stimulates or regulates LPH transcription has yet to be studied. Overlapping expression of GATA proteins in cells may lead to additive, synergistic, or inhibitory effects on LPH transactivation. There is precedent for such regulatory interactions between GATA factors. GATA-1 and GATA-2 show signs of mutual cooperativity. In addition, GATA-1 has been shown to repress GATA-2 expression (30). Studies of the ranges of expression and functional interactions of intestinal GATA binding proteins will lead to a clearer understanding of how they modulate one another in the gut. Not surprisingly, there are GATA motifs upstream of other genes whose expression is restricted to enterocytes, including those which encode the intestinal fatty acid binding protein (25) and SI (32). It will be of interest to determine whether an intestinal GATA binding protein(s) plays a functional role in regulating the transcription of these and/or other promoters that are active in intestinal cells.
We thank Dr. N. Mantei for providing the LPH promoter-containing phage λLPH 7, Dr. J. B. E. Burch for providing a GATA-6 expression vector, and Dr. T. Evans for providing a GATA-6 cDNA probe. In addition, we acknowledge the excellent technical assistance of M. Potts, D. Meighen, the help of P. D. Avigan, and the generous sharing of reagents by Dr. I. Sunitha. Finally, we also thank Drs. S. Irving, A. Riegle, B. Jenson, and R. Glazer for helpful discussions and their critical reviews of this manuscript.
Address for reprint requests: M. I. Avigan, Depts. of Pathology and Medicine, Georgetown Univ. School of Medicine, 3900 Reservoir Rd. NW, Washington, DC 20007.
M. I. Avigan was supported by National Cancer Institute Grant CA-54818.
- Copyright © 1998 the American Physiological Society