Silencing of GATA5 gene expression as a result of promoter hypermethylation has been observed in lung, gastrointestinal and ovarian cancers. However, the regulation of GATA5 gene expression has been poorly understood. In the present study, we have demonstrated that an E (enhancer)-box in the GATA5 promoter (bp −118 to −113 in mice; bp −164 to −159 in humans) positively regulates GATA5 transcription by binding USF1 (upstream stimulatory factor 1). Using site-directed mutagenesis, EMSA (electrophoretic mobility-shift analysis) and affinity chromatography, we found that USF1 specifically binds to the E-box sequence (5′-CACGTG-3′), but not to a mutated E-box. CpG methylation of this E-box significantly diminished its binding of transcription factors. Mutation of the E-box within a GATA5 promoter fragment significantly decreased promoter activity in a luciferase reporter assay. Chromatin immunoprecipitation identified that USF1 physiologically interacts with the GATA5 promoter E-box in mouse intestinal mucosa, which has the highest GATA5 gene expression in mouse. Co-transfection with a USF1 expression plasmid significantly increased GATA5 promoter-driven luciferase transcription. Furthermore, real-time and RT (reverse transcription)–PCR analyses confirmed that overexpression of USF1 activates endogenous GATA5 gene expression in human bronchial epithelial cells. The present study provides the first evidence that USF1 activates GATA5 gene expression through the E-box motif and suggests a potential mechanism (disruption of the E-box) by which GATA5 promoter methylation reduces GATA5 expression in cancer.
GATA5 is a member of the GATA transcription factor family whose six members contain a highly conserved DNA-binding domain consisting of two zinc fingers. Whereas GATA1, GATA2 and GATA3 are expressed in the haemopoietic system, GATA4, GATA5 and GATA6 are expressed within multiple mesoderm- and endoderm-derived tissues, and are involved in tissue-specific transcriptional regulation during cell differentiation and embryogenesis. Silencing of the genes encoding GATA4 and GATA5, or of their downstream activation targets TFF (trefoil factor) 1, TFF2, TFF3 and inhibin-α, impairs proper maturation of endoderm-derived epithelial cells [1–4]. Loss of GATA5 gene expression due to promoter hypermethylation has been observed in malignancy in various organs, including lung [5–8], pancreas , ovarian , breast , oesophageal , stomach [13–15] and colon [16–18]. Despite its apparent importance in differentiation and development, very little is known about how GATA5 transcription is regulated normally, or about specifically how hypermethylation reduces its expression. In the present study, we address the transcriptional regulation of GATA5 gene expression.
In many genes, an E (enhancer)-box lies within the promoter region to provide a binding site for members of the bHLH (basic helix–loop–helix) transcription factor family and enhance transcription of the downstream gene. In the 5′-CANNTG-3′ E-box core sequence, the two central nucleotides (NN) are, in most cases, either GC or CG, and altering the spacing between the CA and TG abrogates nuclear factor binding . USF (upstream stimulatory factor) 1 is a member of the eukaryotic evolutionarily conserved bHLH–leucine zipper transcription factor family, and exhibits higher binding affinity for the 5′-CACGTG-3′ core motif than the 5′-CATGTG-3′ motif. In the present study, we identified an E-box sequence (5′-CACGTG-3′) in the promoter regions of both the human GATA5 and the murine Gata5 genes, and found that USF1 activates GATA5/Gata5 gene transcription by specially binding to the E-box motif of the GATA5/Gata5 promoter. Furthermore, experimental methylation of the E-box prevented USF1 binding, suggesting a mechanism by which GATA5 promoter hypermethylation might reduce GATA5 expression in cancers.
Cell culture and animal tissue
16HBE14o-, CMT-93, HEK (human embryonic kidney)-293, HCT116 and NIH 3T3 cells (A.T.C.C., Manassas, VA, U.S.A.) were maintained in DMEM (Dulbecco's modified essential medium)/Ham's F12 supplemented with 10% fetal bovine serum, 100 units/ml penicillin and 100 μg/ml streptomycin at 37°C in a 5% CO2 humidified incubator. Mouse tissue was used in accordance with institutional and National Institutes of Health guidelines and was approved by the University of Chicago Institutional Animal Care and Use Committee.
The pGL3-basic vector (Promega) was used to construct reporter plasmids to measure GATA5 promoter activity as described previously . A series of DNA fragments comprising bp −3053, −1706, −1289, −984, −731, −367, −317, −258, −150 or −71 to +311 [relative to the known GATA5 major TSS (transcription start site); Ensembl Gene ID ENSMUSG00000015627] of the mouse Gata5 gene were generated by PCR amplification using primers listed in Supplementary Table S1 (at http://www.BiochemJ.org/bj/446/bj4460089add.htm) and inserted into MluI/SpeI sites upstream of the firefly luciferase gene. Plasmids in which luciferase expression was driven by GATA5 bp −258 to +85 harbouring either of two double-point mutations [−108/−107 CG>TA (Mut −108/−107) or −115/−114 GT>AC (Mut −115/−114)] were generated similarly. PCR amplifications were performed with LA Taq™ polymerase and GC buffer (TaKaRa). The mammalian USF1 expression vector (psv-USF1), AUSF (a dominant-negative mutant USF1) (psv-AUSF) and their control vector (pSG424) have all been described previously  and were generously provided by Dr Nanyue Chen (University of Texas MD Anderson Cancer Center, Houston, TX, U.S.A.). All constructs were verified by DNA sequence analysis and prepared using an EndoFree Plasmid Maxi Kit (Qiagen).
Transient transfection and luciferase assay
The Dual-Luciferase Reporter Assay system (Promega) was used to determine promoter activity in HEK-293 cells. Transient transfection was performed in 12-well plates at 60000 cells/well with 1 μg of pGL3-basic or equal molar amounts of pGL3-basic-derived plasmids containing mouse Gata5 promoter constructs, and 3 ng of pKT-null vector containing the Renilla luciferase reporter gene as an internal control for transfection efficiency. pGL3-basic or pGL3-basic-derived plasmids containing the GATA5 promoter region −258 to +85 (wild-type, Mut −115/−114, Mut −108/−107) were co-transfected with 1 μg of expression vectors pGS424, psv-USF1 or psv-AUSF. All mammalian cells were transfected with Qiafect Reagent (Qiagen) according to the manufacturer's instructions. At 48 h post-transfection, cells were harvested and reporter activity was measured using the Dual-Luciferase Assay System. Promoter activity was calculated by normalizing luciferase to Renilla and was expressed as means±S.D. from at least three independent experiments, each performed in triplicate wells.
EMSA (electrophoretic mobility-shift assay)
Nuclear extracts from CMT-93 and 16HBE14o- cells were prepared using the NE-PER® Nuclear and Cytoplasmic Extraction Kit (Pierce) and used to identify protein–DNA interactions in mice and in humans respectively.
The oligonucleotide probe with mouse Gata5 promoter sequence was from bp −127 to −100 (WT −127 to −100). Modified oligonucleotide probes contained the methylated mCpG dinucleotide at a single site on each strand (Met −116) or two double-point mutations (Mut −108/−107 or Mut −115/−114). Another oligonucleotide probe corresponding to bp −173 to −146 of the human GATA5 promoter (hWT −173 to −146) was generated. Modified probes contained methyl-CpG dinucleotide (hMet −162) at a single site on each strand or a double-point mutation at −162/−161 CG>AA (hMut −162/−161) (Supplementary Table S1). All oligonucleotides were synthesized from IDT (Integrated DNA Technologies), labelled with [γ-32P]ATP using T4 polynucleotide kinase and purified by chromatography through ProbeQuant™ G-50 micro columns (GE Healthcare).
Samples of 5 μg of nuclear proteins or 2.5 μg of nuclear extracts depleted of USF1 with an anti-USF1 antibody (H-86 X, Santa Cruz Biotechnology) were incubated with [γ-32P]ATP-labelled probe (20000 c.p.m.) at room temperature (25°C) for 20 min in a total of 20 μl of a reaction mixture containing (final concentrations): 10 mM Tris/HCl (pH 7.5), 50 mM NaCl, 1 mM MgCl2, 0.5 mM EDTA, 0.5 mM DTT (dithiothreitol), 4% (v/v) glycerol and 1 μg of poly(dI-dC)·(dI-dC) as a non-specific competitor. For competition EMSAs, 100 pmol or a serial dilution of unlabelled probe was added in the reaction mixture and pre-incubated at room temperature for 10 min to verify the specificity of protein–DNA interactions. Supershift assays were performed using polyclonal antibodies directed against Sp1 (stimulating protein 1), AP2 (activating protein 2), Oct-1 (octamer transcription factor-1), USF1 or USF2 (Santa Cruz Biotechnology). The resulting protein–DNA complexes were analysed by electrophoresis on a 5% polyacrylamide gel followed by autoradiography and densitometry analysis.
Bisulfite sequencing to identify DNA methylation
Genomic DNA was extracted from wild-type C57Bl/6J mouse intestinal mucosa or from cultured HCT116 human colon carcinoma cells using the Puregene DNA Purification System (Gentra Systems) and was treated with sodium bisulfite using the EZ DNA Methylation Kit (Zymo Research) by following the manufacturer's instructions. The DNA fragment containing the GATA5 promoter E-box was amplified by PCR using the primers in Supplementary Table S1 and cloned into pCR2.1-TOPO (Invitrogen). Five clones for each sample were sequenced with M13 reverse primer via an automated sequencing system (University of Chicago Sequencing Core Facility).
Biotin-affinity purification and Western blot analysis of DNA-binding proteins
Biotin-labelled sense strand and unlabelled antisense oligonucleotides, including WT −127 to −100, Mut −115/−114, Mut −108/−107, WT −215 to −173, and WT −610 to −575 (Supplementary Table S1), were purchased from IDT. Biotinylated double-stranded oligonucleotides were prepared by annealing 2.5 nmol of each single-stranded oligonucleotide resuspended in 100 μl of annealing buffer (10 mM Tris/HCl, pH 7.5, 50 mM NaCl and 10 mM MgCl2), heated for 3 min at 80°C and cooled to room temperature.
Dynabeads MyOne Streptavidin C1 (Invitrogen) were washed three times for 10 min each in washing buffer (10 mM Tris/HCl, pH 7.5, 100 mM NaCl and 2 mM DTT, 0.1% BSA). Biotinylated double-stranded DNA (100 pmol) in 50 μl of coupling buffer (10 mM Tris/HCl, pH 8.0, 1 M NaCl, 1 mM EDTA, 2 mM DTT and 0.1% BSA) was mixed with 1 mg of streptavidin C1-coated Dynabeads and incubated at room temperature for 30 min on a rotating wheel. After incubation, the Dynabeads coated with biotinylated double-stranded DNA were washed three times in coupling buffer and resuspended in 100 μl of TNE buffer (10 mM Tris/HCl, pH 8.0, 100 mM NaCl and 1 mM EDTA) supplemented with 2 mM DTT and 20 μg/ml insulin .
DNA-coupled Dynabeads (100 μl) were washed four times with 500 μl of binding buffer [20 mM Tris/HCl, pH 8.0, 10% (v/v) glycerol, 2 mM EDTA, 0.01% Triton X-100, 150 mM NaCl, 1 mM DTT, 10 μg/ml insulin and 0.5 mM Pefabloc], resuspended in 50 μl of 2× binding buffer and incubated with an equal volume of CMT-93 nuclear extract (30 μg of protein) for 15 min at room temperature. After magnetic separation, supernatants were removed and reserved for subsequent electrophoresis analysis. The beads containing the protein–biotinylated fragment–streptavidin complex was washed three times with the binding buffer and boiled in Laemmli sample buffer. The bound proteins were resolved by SDS/PAGE (12% gels) along with the corresponding supernatant described above. Proteins were electroblotted on to nitrocellulose membranes following standard protocols. Each membrane was blocked with 2% BSA in TBS (Tris-buffered saline: 10 mM Tris/HCl, pH 8.0, and 150 mM NaCl) for 1 h at room temperature. The membrane was then incubated overnight at 4°C with rabbit polyclonal anti-USF1 or anti-Sp1 antibody at a 1/200 dilution in TBS with 2% (w/v) BSA. After five washes in TBST (TBS with 0.1% Tween 20), the membrane was incubated with horseradish-peroxidase-conjugated secondary antibody at a 1/10000 dilution in TBS with 5% (w/v) non-fat dried skimmed milk powder for 1 h at room temperature. The membrane was washed three times with TBST and the specific protein bands were visualized using the SuperSignal Chemiluminescence system (Pierce).
ChIP (chromatin immunoprecipitation) assay
A ChIP assay was performed on mouse intestinal mucosa using a ChIP Assay Kit (Upstate Biotechnology). Briefly, the cleaned intestinal mucosa from C57B/6J mice was removed by scratching with a glass slide and fixed at room temperature for 10 min in the presence of DMEM/Ham's F12 culture medium with 1% (w/v) formaldehyde following by quenching with 125 mM glycine for 5 min. After two washes with ice-cold PBS containing protease inhibitor cocktail (Roche Diagnostics), the pellet of intestinal mucosa was homogenized in SDS lysis buffer (1% SDS, 10 mM EDTA and 30 mM Tris/HCl) plus proteinase inhibitor on ice for 10 min and then sonicated to shear the DNA into 200–1000 bp fragments using a Branson Sonifier Cell Disruptor 185. Immunoprecipitation was conducted with rabbit anti-USF1 or anti-USF2 antibody (Santa Cruz Biotechnology) or without any antibody by incubation at 4°C overnight. The chromatin–protein complex was immobilized with Protein A beads and washed to remove unbound or non-specific DNA. After cross-links were reversed and the protein was digested with proteinase K (40 μg/ml), DNA was recovered and purified using a DNeasy Tissue Kit (Qiagen) and resuspended in 20 μl of elution buffer. The input control DNA and immunoprecipitated DNA were amplified and PCR products were separated by 1.5% agarose gel electrophoresis. Quantitative PCR analysis was also performed in duplicate for each of three independent ChIP experiments and the fold enrichment of the GATA5 E-box promoter region from anti-USF1 and anti-USF2 antibody-bound chromatin DNA over a non-promoter region of the GATA5 gene was calculated and normalized to a non-immune serum control, essentially as described in [23,24]. All PCR primer sequences are given in Supplementary Table S1.
Construction of standard curves and real-time PCR assay
To quantitatively measure human GATA5 gene expression activated by overexpression of USF1, real-time PCR standard curves for the GATA5 gene and for a housekeeping control gene ACTB (β-actin) were generated by 10-fold serial dilution of each modified pCR 2.1-TOPO plasmid (Invitrogen) into which was inserted the target sequence and its flanking sequence of the human GATA5 or ACTB genes.
Total RNA was isolated from 16HBE14o- cells transfected with pSG424, psv-USF1 or psv-AUSF. RNA samples were treated with DNase I (Roche) to remove any contaminating DNA, and cDNAs were synthesized from 0.5 μg of RNA using the iScript cDNA Synthesis Kit (Bio-Rad Laboratories). A fragment of human GATA5 (192 bp) and ACTB (159 bp) were amplified during real-time PCR using gene-specific primer pairs listed in Supplementary Table S1. All reactions (standard and unknown samples) were run in triplicate for each condition on the same 96-well plate. Thermal cycling conditions were 94°C for 2 min followed by 40 cycles of 94°C for 30 s, 57°C for 30 s and 72°C for 30 s in a MyiQ™ Single Color Real-Time PCR Detection System (Bio-Rad Laboratories). Results reported are averages of the triplicates. Human GATA5 gene expression level was normalized to the ACTB expression level. RT (reverse transcription)–PCR was also performed using gene-specific RT–PCR primer pairs (Supplementary Table S1). The RT–PCR amplification protocol was 94°C for 3 min, then 30 cycles of 94°C for 30 s, 60°C for 30 s and 72°C for 30 s, followed by extension at 72°C for 5 min in an Applied Biosystems Thermocycler. The PCR products of human GATA5 (336 bp) and ACTB (159 bp) were visualized after electrophoresis on 2% agarose gels.
All experiments were performed independently at least three times. Results are means±S.D. The statistical significance of differences of the means was determined by a paired Student's t test. P≤0.05 was considered statistically significant.
bp −150 to −71 of the mouse Gata5 gene contains positive regulatory elements
To determine the location of putative regulatory elements of the mouse Gata5 promoter, the promoter activity of the cloned 5′ end of the Gata5 gene was studied by transient expression assays with the luciferase gene as a reporter. A genomic DNA fragment containing 3053 bp of 5′ flanking sequence and 311 bp of Gata5 exon 1 was inserted into MluI/SpeI sites upstream of the firefly luciferase gene in pGL3-basic vector. This construct or equal molar amounts of a series of deletion constructs were transfected into HEK-293 cells, and the luciferase activity in cell extracts was analysed after 48 h. The construct containing bp −150 to 311 exhibited the greatest promoter activity. Further 5′ truncation from −150 to −71 led to a remarkable decrease in luciferase activity, indicating the presence of positive regulatory elements in this region (Figure 1). Similar results were observed in canine tracheal smooth muscle cells (not shown). These findings indicate that the upstream sequence −150 to −71 contains strongly positive regulatory element(s) that regulate mouse Gata5 gene expression.
Functional analysis of mouse Gata5 promoter in HEK-293 cells
The GATA5 promoter contains an E-box motif which is a critical binding site
To understand how these positive regulatory elements regulate GATA5 gene expression, we analysed the sequence of the 500-bp region upstream from the human GATA5 and the mouse Gata5 TSSs. We found 79% homology between genomic sequences −136 to −28 of the mouse Gata5 and −182 to −74 of the human GATA5 genes. This region contains a conserved E-box sequence (5′-CACGTG-3′, boxed in Figure 2A) located at −118 to −113 in the mouse and −164 to −159 in the human GATA5 promoters.
EMSA of E-box binding
To determine whether this E-box is capable of binding transcription factors, EMSAs were performed with wild-type oligonucleotide probe corresponding to −127 to −100 of the mouse Gata5 promoter (WT −127 to −100) and nuclear extracts from mouse CMT-93 cells. A complex of nucleic acid probe bound to protein was observed (Figure 2B, lane 1), while formation of this protein–DNA complex was specifically competed by addition of 100 pmol of unlabelled probe (Figure 2B, lane 2). Furthermore, mutation of the E-box from 5′-CACGTG-3′ to 5′-CACACG-3′ within this fragment (Mut −115/−114) abolished the formation of protein–DNA complex (lane 7). On the other hand, mutation at −108/−107 (Mut −108/−107), which left the E-box intact, had no such effect on complex formation (lane 5). Similar findings were observed in EMSAs using wild-type probe hWT −173 to −146 or mutated E-box probe hMut −162/−161 and nuclear extracts from human 16HBE14o- cells (Figure 2C). Additional competition experiments were carried out using mutated oligonucleotide probes with or without intact E-box. As can be seen in Figure 3, protein–DNA complexes were observed with labelled WT −127 to −100 probe (lane 1) as well as Mut −108/−107 probe (lane 5), but not with labelled Mut −115/−114 probe in which the E-box was disrupted (lane 9). Complex formation was completely prevented by the addition of 100 pmol of unlabelled competitor WT −127 to −100 (lanes 2 and 6) or Mut −108/−107 (lanes 3 and 7), but not by addition of unlabelled Mut −115/−114 (lanes 4 and 8).
Competition assay of E-box binding
The E-box of the GATA5 promoter contains a CpG dinucleotide that might be methylated, and which in turn might alter its ability to bind nuclear factors. We therefore examined the E-box methylation status in mouse intestinal mucosa, which has the highest Gata5 gene expression in all mouse tissues , and in HCT116 cells, a human colon cancer cell line without GATA5 gene expression, using bisulfite sequencing analysis. We found that the cytosine of the E-box CpG dinucleotide is unmethylated in mouse intestinal mucosa, but is indeed methylated in HCT116 cells. We therefore tested whether such methylation modifies the ability of this E-box to bind nuclear factors. As seen in Figure 2(B) (lane 3), methylation of the E-box in the mouse oligonucleotide probe (Met −117) significantly diminished probe binding of nuclear factors. A competition EMSA was also performed using nuclear extracts from mouse CMT-93 cells and labelled WT −127 to −100 probe and unlabelled WT competitor or unlabelled Met −117 competitor (Figure 4A). One representative experiment was quantified by densitometry and is shown in Figure 4(B). Approximately 80% of protein–DNA complex was efficiently competed by addition of 1 pmol of unlabelled WT −127 to −100 competitor, whereas 1 pmol of unlabelled methylated E-box competitor reduced formation of this complex by only 10%. Analogous results were observed using nuclear extracts from human 16HBE14o- cells and a labelled methylated E-box probe from the human GATA5 promoter (hMet −162) in EMSAs (Figure 2B, lane 3) or in additional competition experiments using a labelled probe (hWT −173 to −146) with a series of dilutions of unlabelled hWT −173 to −146 or unlabelled hMet −162 competitor (Figure 4C).
Effect of CpG methylation on mouse Gata5 promoter E-box binding
Taken together, these results indicate that the E-box within the GATA5 promoter can bind nuclear factors in vitro, and that methylation prevents such nuclear factor binding.
USF1 specifically binds to the E-box of the GATA5 promoter-positive regulatory region
TRANSFAC® analysis of the sequence around the E-box suggested possible binding sites for USF, Sp1, AP2 and Oct-1, among other factors. To determine whether any of these was included in the protein–DNA complexes described above, we performed supershift EMSAs, using labelled oligonucleotide WT −127 to −100 probe, nuclear extracts from mouse CMT-93 cells and antibodies directed against the transcription factors named above. As seen in Figure 5, the protein–DNA complex was specifically supershifted by the addition of anti-USF1 antibody, but not by anti-USF2, anti-Sp1, anti-AP2 or anti-Oct-1 antibodies (Figure 5A, left-hand panel). Similar results were found using the human DNA sequence and nuclear proteins from human bronchial epithelial cells (Figure 5A, right-hand panel). To confirm that USF1 specifically binds to the GATA5 promoter E-box, we performed EMSAs using USF1-depleted nuclear extracts. 16HBE14o- cell nuclear proteins (40 μl) at 0.5 μg/μl were incubated either without antibody or with 1 or 2 μl of anti-USF1 antibody (H-86 X) overnight at 4°C and then incubated with 20 μl of salmon sperm DNA–Protein A–agarose for 1 h at 4°C with agitation. A 5 μl sample of the USF1-depleted supernatant fraction and the labelled hWT probe −173 to −146 were used for EMSAs as described in the Experimental section. Depletion of USF1 from 16HBE14o- cell nuclear extracts using anti-USF1 antibody specifically eliminated the complex formation on the GATA5 promoter E-box (Figures 5B and 5C). In the light of our demonstration that mutation or methylation of the E-box prevents nuclear factor binding (above), the results in Figure 5 indicate that the GATA5 promoter E-box specifically binds USF1 in vitro.
Binding of USF1 to the E-box of the GATA5 promoter in both mice and humans
To provide additional evidence that the E-box binds USF1, we performed affinity-purification analysis. Double-stranded oligonucleotides corresponding to WT −127 to −100 or containing the E-box mutation (Mut −115/−114), or two other sequences corresponding to bp −215 to −173 or bp −610 to −575 in the mouse Gata5 promoter, were labelled with biotin for affinity purification of CMT-93 nuclear extracts. As seen in Figure 6, USF1 could be recovered with streptavidin-coated beads bound to biotin-labelled wild-type bp −127 to −100 after incubation with nuclear extracts. However, double-point mutation at −115/−114 to disrupt the E-box eliminated binding to USF1, so that almost all of the USF1 remained in the supernatant. Similarly, the genomic sequences −215 to −173 or −610 to −575, which lack E-boxes, also failed to bind USF1. This experiment provides further evidence that USF1 specifically binds to the E-box of the GATA5 promoter region in vitro.
Biotin-affinity purification and Western blot analysis
Next, we performed ChIP analysis to determine whether USF1 physiologically interacts with the GATA5 promoter-positive regulatory region. As shown in Figure 7, the E-box containing sequence of the Gata5 promoter was recovered after immunoprecipitation of sheared genomic DNA from mouse intestinal mucosa with anti-USF1 antibody, but not after immunoprecipitation with anti-USF2 antibody. Quantitative PCR confirmed that USF1 is present at the GATA5 E-box promoter region and the corresponding genomic DNA was enriched with anti-USF1 antibody, but not anti-USF2 antibody (P<0.05, Figure 7C). These data provide additional evidence for the occupancy by USF1, but not USF2, of the GATA5 promoter E-box in vivo.
Physiological interaction of USF1 with the E-box of Gata5 promoter in mouse intestinal mucosa
The GATA5 promoter E-box regulates GATA5 expression
To assess the functional role of the E-box in the GATA5 promoter, the wild-type GATA5 promoter fragment bp −258 to +85, or its mutated counterparts harbouring the E-box mutation at −115/−114 or the non-E-box mutation at −108/−107, were cloned upstream of the luciferase cDNA in pGL3-basic and transient transfection was performed in HEK-293 cells (Figure 8). There was significant promoter activity for the wild-type fragment and its mutated counterpart with intact E-box (compared with empty pGL3-basic luciferase reporter; P=0.0035 and 0.0043 respectively). However, promoter activity was significantly reduced when the E-box was mutated from 5′-CACGTG-3′ to 5′-CACACG-3′ (P=0.021 compared with wild-type; P=0.025 compared with Mut −108/−107). This indicates that the E-box at bp −118 to −113 is functionally crucial for full GATA5 gene promoter activity.
Mutation of the E-box reduces GATA5 promoter activity
Next, we evaluated whether overexpression of USF1 further activates transcription from the GATA5 promoter. We co-transfected pGL3-basic-derived plasmids containing the GATA5 promoter region from bp −258 to +85, or its E-box-mutated counterparts, together with expression vectors pGS424, psv-USF1 or psv-AUSF into NIH 3T3 cells, which have no endogenous GATA5 expression . As shown in Figure 9(A), USF1 significantly activated the wild-type GATA5 promoter fragment (as compared with empty pSG424 vector or dominant-negative USF vector; each P<0.05). Mutation of this promoter fragment at −108/−107 (which leaves the E-box intact) did not reduce GATA5 promoter activity during transactivation by USF1 overexpression. However, mutation of the E-box at bp −115/−114 significantly reduced promoter activity even during overexpression of USF1 (P<0.05). These results demonstrate that USF1 transactivates the GATA5 promoter by binding to its E-box.
Transactivation of GATA5 promoter activity by USF1 overexpression
Lastly, we tested whether USF1 increases endogenous GATA5 expression. Human bronchial epithelial cells were transfected with either USF1 expression vector, control vector pSG424 or plasmid expressing dominant-negative AUSF. GATA5 mRNA in transfected cells was quantified using real-time PCR. As shown in Figure 9(B), USF1 overexpression increased GATA5 gene expression in human bronchial epithelial cells, compared with cells transfected with empty pSG424 vector or dominant-negative USF vector. These quantitative results were confirmed qualitatively using RT–PCR (Figure 9C).
Together, these results indicate that USF1 transactivates GATA5 expression by binding to the E-box in its promoter.
Despite its importance in cell differentiation and tissue development [26,27], little is known about the transcriptional regulation of GATA5 expression. In the present study, we identified a positive regulatory region within bp −150 to −71 of the mouse Gata5 5′ flanking sequence that contains an E-box consensus sequence (5′-CACGTG-3′) (Figure 2A). Using EMSA, biotinylated DNA purification and ChIP, we confirmed that this E-box specifically binds USF1 in vitro and in vivo. Such E-box–USF1 binding is functionally important in activating GATA5 expression, because: (i) mutation of the E-box to prevent USF1 binding reduces GATA5 promoter activity (Figure 3); (ii) USF1 overexpression activates E-box-dependent GATA5 promoter activity (Figure 8); and (iii) USF1 overexpression increases endogenous GATA5 mRNA expression (Figure 9). Taken together, these data demonstrate conclusively that USF1 activates GATA5 transcription.
USFs are members of the bHLH transcription factor family that bind the E-box in a promoter to activate transcription of the target gene. A wide variety of genes are direct targets of USFs, including genes involved in the immune response , glucid and lipid pathways , cell cycle [30,31], cell proliferation  and carcinogenesis [33–36]. In the present study, we found that GATA5 is another target gene of USF1 and that GATA5 gene expression is activated through binding of USF1 to the canonical E-box sequence in the GATA5 promoter region, as discussed above. In other genes, it had previously been shown that the E-box modulation by methylation at the core CpG in the USF1 recognition site (5′-CACpGTG-3′) [37–39] significantly hinders transcription factor–DNA interaction and so negatively regulates gene expression. Others have reported that GATA5 gene expression occurs in association with GATA5 promoter hypermethylation in lung, oesophageal, colon and gastric cancers [7,12–14,17]. Because we have demonstrated that the E-box in the GATA5 promoter plays an important role in GATA5 expression, we hypothesized that methylation of CpG dinucleotides in the E-box  suppresses USF1 binding, and thereby down-regulates the expression of GATA5. Indeed, we have shown that experimental methylation of this E-box does inhibit its binding of USF1 (Figure 4). Furthermore, we found that the GATA5 promoter E-box is not methylated in mouse intestinal mucosa, which exhibits the greatest endogenous GATA5 mRNA expression , whereas the corresponding E-box in human HCT116 colon cancer cells (which lack GATA5 expression) is highly methylated. Finally, Ting et al.  and McGarvey et al.  reported that treatment of HCT116 cells with 5-azacytidine to demethylate genomic DNA restored GATA5 expression. Together, these results strongly imply that methylation of the GATA5 promoter E-box in certain cancers is one mechanism by which they lose GATA5 expression.
Besides E-box–USF1 binding, other transcriptional regulatory mechanisms probably also influence GATA5 expression. Using TRANSFAC® analysis, we found that the GATA5 gene promoter lacks a TATA box. The 5′-flanking sequence at bp −48 to −38 contains the recently identified X core promoter element , which might initiate GATA5 transcription by RNA polymerase II, as well as a potential binding site for Sp1, which might enhance gene transcription. Furthermore, our 5′ deletion analysis suggests the presence of a negative regulatory region within bp −3053 to −150 (Figure 1). Additional research is needed to delineate these other transcriptional regulatory mechanisms more completely.
Finally, we have reported previously that an N-terminally truncated, but still functional, isoform of mouse GATA5 is transcribed in intestinal and gastric epithelium from an alternative promoter within intron 1 . Sequence analysis fails to disclose a consensus E-box within 1000 bp of the minor TSS (which is 82 bp upstream of exon 2). As such, this alternative GATA5 promoter must be differentially regulated from the principal GATA5 promoter (upstream of exon 1) investigated in the present study.
activating protein 2
Dulbecco's modified essential medium
electrophoretic mobility-shift assay
human embryonic kidney
octamer transcription factor-1
stimulating protein 1
TBS with 0.1% Tween 20
transcription start site
upstream stimulatory factor
a dominant-negative mutant USF1
Bohao Chen conceived, designed and carried out the experiments and drafted the paper. Rona Hsu, Zhenping Li, Paul Kogut, Qingxia Du and Kelly Rouser conducted experiments. Blanca Camoretti-Mercado contributed to experiments and data interpretation. Julian Solway designed the experiments, interpreted data and revised the paper. All authors read and approved the final paper.
We thank Dr Nanyue Chen (University of Texas MD Anderson Cancer Center, Houston, TX, U.S.A.) who kindly provided psv-USF1, psv-AUSF and pSG424 plasmids.
This work was supported in part by the National Institutes of Health [grant numbers R01 HL 079398, R01 HL 097805 and K12 HL 090003] and the Cancer Research Foundation.