Transcriptional regulation of the rat NHE3 gene. Functional interactions between GATA-5 and Sp family transcription factors.

Expression of sodium-hydrogen exchanger isoform 3 (NHE3) in the intestinal and renal epithelium plays a critical role in sodium absorption and acid/base homeostasis. To decipher rat NHE3 gene regulation, its cis-acting regulatory elements and associated transcription factors were characterized by transient transfection of Caco-2, IEC-6, Qt6, and Drosophila SL2 cells. Deletion and mutational analyses demonstrated that the atypical TATA box located at bp -26/-31 was not necessary for promoter activity, and that a -20/+8-bp fragment represents a functional initiator. Within the 81-bp upstream region, three Sp transcription factor binding sites were critical because their mutation drastically reduced promoter activity. The roles of Sp1 and Sp3 were further demonstrated by electromobility shift assay and by transactivation of the NHE3 promoter in SL2 cells by forced expression of Sp1 and Sp3. Both of these transcription factors were found to act synergistically with GATA-5 bound to a GATA box in exon 1 (+20/+23 bp). These studies demonstrate that rat NHE3 promoter is initiator-driven and controlled mainly by Sp1 and Sp3, which functionally interact with GATA-5. This interaction represents a novel regulatory mechanism, which is likely to participate in a gradient of intestinal gene expression along the crypt-villus axis.

NHE3 activity is regulated at different levels: phosphorylation-mediated regulation of the transporter dependent on its linkage to the cytoskeletal protein ezrin with NHERF and E3KARP (6), regulation by trafficking of NHE3 protein on and off the apical membrane via changes in endocytosis and apical membrane recycling (7), regulation through association with lipid rafts in the brush-border membrane (8), and through transcriptionally mediated changes in NHE3 mRNA levels. An array of factors has been implicated in transcriptional regulation of NHE3 gene expression including glucocorticoid hormones (9,10), thyroid hormone (11), protein kinase C (12), and sodium butyrate (13,14). Previous studies from our laboratory also suggested that transcriptional regulation is a critical component of maturational changes in intestinal NHE3 activity and gene expression during rat postnatal development (15). Although laboratory rodents have become a common model to study physiology and pathophysiology of Na ϩ /H ϩ exchange, and the rat NHE3 gene promoter has been cloned by two independent groups of investigators (9,16), a comprehensive analysis of its function and basal regulation has been lacking to date.
NHE3 exhibits not only a temporal but also a horizontal, differentiation-specific pattern of expression along the cryptvillus axis. NHE3 expression is very low in the small intestinal epithelium of suckling animals and drastically increases after weaning (15). Additionally, NHE3 protein is not expressed in undifferentiated crypt cells of the small intestine or colon, while it is present in high abundance in differentiated absorptive enterocytes with a detectable gradient of expression directed toward the tips of the intestinal villi (17)(18)(19). Due to its physiological importance and complex expression patterns, NHE3 may be considered a marker gene suitable for studying the role of specific transcription factors and their interactions during enterocyte differentiation as well as during postnatal intestinal development, similar to the already established models of lactase-phlorizin hydrolase (LPH), sucrase-isomaltase (SI), and intestinal fatty acid-binding protein (iFABP). The body of knowledge provided by studying regulation of intestinal gene expression provided by studies of the three latter genes is already significant, but despite this fact, a complete understanding of the genetic programming of mammalian intestinal differentiation and maturation is still far from being achieved.
In recent years, a concept of combinatorial gene regulation has become a paradigm in understanding complex regulation of gene transcription in eukaryotes. The activation of eukaryotic genes in vivo often requires the coordinated binding of multiple transcription factors to promoter-enhancer regions of genes. Many of these transacting factors are not only expressed in tissue-and differentiation-specific manners, but they are also regulated by distinct signaling pathways. Furthermore, in many cases, cooperative binding of multiple transcription fac-tors to gene promoter requires a unique composition and spatial arrangement of transcription factor binding sites. These facts add to the complexity of transcriptional networks participating in regulation of gene expression.
In this article, we describe the first comprehensive analysis of NHE3 promoter function. We confirm previously disputed transcription initiation site, describe a potentially novel initiator sequence, delineate the role of Sp family transcription factors in regulation of NHE3 promoter activity, and identify a functional binding site for GATA transcription factors in exon 1. Most importantly, we also provide evidence for a novel synergistic interaction between GATA-5 and Sp1 and Sp3 in regulating NHE3 promoter function.

EXPERIMENTAL PROCEDURES
Cell Culture-Human colonic adenocarcinoma (Caco-2) cells, normal rat small intestinal epithelial (IEC-6) cells, NIH-3T3 mouse fibroblasts, quail fibrosarcoma (Qt6) cells, and Drosophila SL2 cells were obtained from American Type Culture Collection (Manassas, VA). Caco-2 and IEC-6 cells were maintained in high-glucose Dulbecco's Modified Essential Medium (DMEM) supplemented with 2 mM glutamine, 1 mM sodium pyruvate, 0.1 mM non-essential amino acids, 100 g/ml streptomycin, and 100 units/ml penicillin G (further referred to as antibiotics), and 10% (v/v) fetal bovine serum. NIH-3T3 cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% bovine calf serum and antibiotics. QT-6 cells were maintained in F12K medium with 10% tryptose phosphate broth, 5% bovine calf serum, and antibiotics. All of the aforementioned cells were grown at 37°C in a 5% CO 2 , 95% air incubator. Drosophila SL2 cells were maintained in Schneider's insect medium supplemented with 10% (v/v) fetal bovine serum and antibiotics, and were grown at 25°C without CO 2 .
Plasmid Constructs-Rat NHE3 promoter constructs were prepared in pGL-3 basic luciferase reporter vector (Promega, Madison, WI) as described earlier (14). Since overexpression of Sp or GATA transcription factors activated promoter-less pGL3-basic but not p␤Gal-basic vector (Clontech, Palo Alto, CA) in transfected cells, transactivation experiments with these factors were performed with NHE3 promoter fragments subcloned into p␤Gal-basic. Site-directed mutations were introduced by standard PCR with mutant forward primers utilizing a high fidelity DNA polymerase (Pfx, Invitrogen, Carlsbad, CA) or by two-step overlap PCR as previously described (20). A construct spanning Ϫ20/ϩ8 bp of the NHE3 gene was created by subcloning a double-stranded synthetic oligonucleotide into XmaI/XhoI sites of pGL3-basic. All constructs were confirmed by DNA sequencing.
5Ј-RLM-RACE-Since two previous reports (9, 16) described different transcription initiation sites within the rat NHE3 gene, RNA ligasedependent rapid amplification of cDNA ends (5Ј-RLM-RACE) was employed as an alternative method to identify the 5Ј-end of the NHE3 transcript utilizing components of FirstChoice RLM-RACE kit (Ambion, Austin, TX). High quality total RNA was isolated with trizol reagent (Invitrogen) followed by extraction with acid phenol. RNA from the rat small intestine and colon was used to confirm the transcription initiation site in the endogenous NHE3 gene, whereas DNase I-treated RNA isolated from Caco-2 cells transfected with NHE3 gene reporter plasmid (Ϫ118/ϩ58Luc) was used to confirm the transcription initiation site utilized in synthetic reporter gene constructs. Briefly, 10 g of RNA was dephosphorylated to remove the 5Ј-phosphate group from RNA or contaminating DNA molecules. Tobacco acid pyrophosphatase (TAP) was then used to specifically remove the cap structure from mRNA molecules. An RNA oligonucleotide was next ligated to newly de-capped mRNA using T4 RNA ligase and the resulting RNA was reverse-transcribed using SuperScript III (Invitrogen) and random primers. Outer and nested hot-start PCR reactions were performed using proofreading DNA polymerase (Platinum Taq Hi Fidelity, Invitrogen) with adapterand gene-specific primers. 1 l of the nested PCR product was subcloned into pCR4Blunt-TOPO vector (Invitrogen). 25-35 independent clones were sequenced for each 5Ј-RLM-RACE reaction.
Transfections and Reporter Gene Assays-Caco-2 cells were transfected in 24-well plates at ϳ80% confluency with 200 ng of NHE3 promoter constructs and 2 ng of promoter-less pRL-null vector containing the Renilla luciferase reporter gene as an internal control. IEC-6, NIH-3T3, and Qt6 cells were transfected in 6-well plates with 1 g of NHE3 promoter constructs and 10 ng of pRL-null. In experiments with forced expression of GATA transcription factors, 0.2 or 1 g of pCDNA3 (Invitrogen) or pCDNA3-based expression vectors with mouse GATA-4, GATA-5, or GATA-6 cDNAs (generously provided by Dr. E. Morrissey) were co-transfected along with NHE3 promoter constructs into respective cell lines. All mammalian cell lines were transfected with Lipo-fectAMINE (Invitrogen) and Qt6 cells were transfected with FuGENE 6 (Roche; Indianapolis, IN) according to manufacturer's instructions. 24 h post-transfection, cells were harvested and assayed for reporter gene activity using a dual luciferase assay (Promega; for constructs made in pGL3-basic) or with Galacto-Light Plus (Applied Biosystems, Foster City, CA) and Renilla luciferase assay system (Promega; for constructs in p␤Gal-basic). Endogenous ␤-galactosidase activity in mammalian cell lines was reduced by incubating cell lysates at 48°C for 50 min prior to performing the ␤-galactosidase assay.
SL-2 cells were seeded at 500,000 cells per well in 24-well plates 24 h prior to transfection with 200 ng of p␤GalϪ118/ϩ58 and various amounts of Sp1 or Sp3 and/or GATA-5 expression plasmids containing the Drosophila actin 5 promoter (pPacSp1 and pPacUSp3 were generously provided by Drs. R. Tjian and G. Suske, respectively; pAc-GATA-5 construct was made by subcloning a SmaI/EcoRV fragment of mouse GATA-5 into the EcoRV site of pAc5.1A plasmid (Invitrogen)). ␤-galactosidase activity was expressed as relative light units (RLU) per g of protein.
The protein concentration was determined by BCA protein assay (Pierce, Rockford, IL).
Preparation of Nuclear Extracts and Electrophoretic Mobility Shift Assays-Nuclear extracts were prepared using a modification of the method by Dignam (21) as described by Dent and Latchman (22). The obtained nuclear protein was dialyzed in 10,000 MWCO mini dialysis units (Pierce) against buffer containing 20 mM HEPES (pH 7.9), 100 mM KCl, 1 mM MgCl 2 , 20% glycerol, 0.5 mM dithiothreitol, and 0.5 mM phenylmethylsulfonyl fluoride and stored at Ϫ70°C until use.
For gel shift assays, PAGE-purified double-stranded oligonucleotides (Integrated DNA Technologies, Coralville, IA) were end-labeled with [␥-32 P]ATP by T4 polynucleotide kinase (Promega). Assays were performed by incubating 5 g of nuclear protein with 10,000 to 20,000 cpm of labeled probe for 20 min. Competing unlabeled oligonucleotides were added 20 min prior to addition of the radiolabeled probe. Sp protein gel shift assays were performed with commercially optimized reagents (Sp1/Sp3 Gelshift Kit; Geneka; Montreal, Canada) with binding reactions carried out at 10°C. Binding reactions for GATA transcription factors were carried out at room temperature in binding buffer containing 10 mM Tris-HCl (pH 7.5), 1 mM MgCl 2 , 0.5 mM EDTA, 0.5 mM dithiothreitol, 50 mM NaCl, 4% glycerol, 1 g/reaction poly(dI-dC) and 1 g/reaction polylysine.
Southwestern Blotting-All procedures were carried out at 4°C. 50 g of Caco-2 and IEC-6 cell nuclear extract was fractionated on a 4 -20% SDS-polyacrylamide gel and transferred to a nitrocellulose membrane in a buffer containing 25 mM Tris and 190 mM glycine for 30 min at 100 V. The membrane was then incubated with blocking buffer (2% nonfat dry milk, 1% bovine serum albumin, 50 mM HEPES (pH 7.9), 75 mM MgCl 2 , 40 mM KCl, 0.05 mM EDTA, 5% glycerol, 140 mM ␤-mercaptoethanol, and 16 g/ml sonicated salmon sperm DNA) for 2 h and then incubated with binding buffer (same as blocking buffer but with 0.2% nonfat dry milk) containing 32 P-labeled double-stranded probe (Ϫ20/ϩ8 bp NHE3 promoter fragment; ϳ10 6 cpm/ml) for 16 h, washed, and subjected to autoradiography. An identical binding reaction was performed with 200-fold excess of unlabeled probe as a negative control. Statistical Analysis-One-way analysis of variance or the Student's t test were employed for statistical analyses utilizing StatView software (v. 4.0; SAS Institute, Inc., Cary, NC). Post-hoc multiple comparisons were carried out with Fisher's PLSD test.

RESULTS
Transcription Initiation Site Mapping-Since previous reports utilizing primer extension (9) and S1 nuclease protection assay (16) techniques produced contradictory results, we employed 5Ј-RLM-RACE as an alternative method of mapping the 5Ј-end of the NHE3 transcript. This technique identified multiple transcription initiation sites (TIS), overlapping those detected by Kandasamy and Orlowski (16) by S1 nuclease protection assay and localized within the Ϫ3 ACCTG ϩ2 sequence ( Fig.  1). Interestingly, the major transcription initiation site reported by Kandasamy and Orlowski (16), which is used for nucleotide numbering throughout this study, was not seen in the jejunal RNA pool or in the heterologous NHE3/Luc gene, and only 4% of colonic NHE3 transcripts started at this position. Utilization of TIS in different tissues varied slightly, as summarized in Fig. 1. None of the sequenced clones matched the TIS described by Cano (9) by primer extension. Small proportion of 5Ј-RLM-RACE products (4% jejunum; 4% colon) started 20 -70 bp downstream of the 5Ј-end of NHE3 transcript identified by initial cloning experiments with the cDNA library (23). Since cDNA libraries rarely contain full-length 5Ј-untranslated regions, these results were dismissed as likely artifacts.
The same cluster of TIS was utilized to drive expression of the luciferase reporter gene from the Ϫ118/ϩ58Luc NHE3 promoter construct (Fig. 1). While 91% of detected transcripts originated within the same cluster of nucleotides ( Ϫ3 AC-CTG ϩ2 ), 4% of transcripts of the reporter gene started at nt Ϫ9 and 5% at nt ϩ21. Relevance of these sites is however, unclear.
Functional Mapping of the Rat NHE3 Promoter-A series of reporter plasmids containing various lengths of the NHE3 5Јflanking regions (from nt Ϫ1360 to nt Ϫ20) upstream of the firefly luciferase (Luc) gene was transfected into Caco-2 cells. The levels of Luc activity were normalized to Renilla luciferase activity. A short promoter fragment spanning nt Ϫ118 to ϩ58 bp conferred 82% of the activity of the longest cloned promoter fragment (p ϭ 0.015 for Ϫ1360/ϩ58 versus Ϫ118/ϩ58) (Fig. 2). Minor fluctuations in activity were observed among constructs including more than 118 bp of the promoter sequence, suggesting the presence of enhancing or repressive elements, which conceivably play a role in fine-tuning transcriptional activity. Also, further deletion to Ϫ81 bp only slightly altered the promoter activity (18% decrease; p ϭ 0.03 for Ϫ118/ϩ58 versus Ϫ81/ϩ58). Surprisingly, a short construct spanning Ϫ20/ϩ58 FIG. 2. Deletion analysis of the rat NHE3 promoter. Firefly luciferase reporter constructs containing various lengths of the NHE3 promoter were generated as described under "Experimental Procedures" and were cotransfected into Caco-2 cells along with a promoter-less pRL-null plasmid. Twenty-four hours later, the cells were harvested and firefly and Renilla luciferase activities were measured. The results are the means (ϮS.D.) of at least three independent experiments. The sequence of the NHE3 gene spanning Ϫ118/ϩ58 bp with annotated putative cis-elements is depicted below. Predicted binding sites shown to be important for basal promoter activity are depicted in bold letters. The arrow at T ϩ1 represents the major TIS identified previously (16), whereas other arrows mark the 5Ј-ends of promoter fragments studied in reporter gene assays.
FIG. 1. Verification of transcription initiation sites in the NHE3 gene by 5-RLM-RACE. Shown above is sequence of the proximal rat NHE3 promoter (plain font) and the 5Ј-end of cDNA from initial cloning experiment (Ref. 23; bold letters) with TIS mapped by Cano (9) and Kandasamy and Orlowski (16). Long vertical arrows indicate major TIS, short arrows indicate minor TIS from the two previous studies. Shown in the middle are the results of 5Ј-RLM-RACE performed with jejunal and colonic RNA and with RNA isolated from Caco-2 cells transfected with the Ϫ118/ϩ58Luc NHE3 reporter gene construct. The table depicts TIS utilization expressed as percent of all sequenced clones. Location of 5Ј-ends of NHE3 transcripts unaccounted for in the table is described in more detail under "Results." Shown below is gel analysis of the nested PCR reactions from 5Ј-RLM-RACE with jejunal and colonic RNA. TAPϩ and TAPϪ lanes represent RACE reactions with RNA processed with or without tobacco acid pyrophosphatase (TAP), respectively. bp exhibited significant activity (4.9-fold above background; p ϭ 0.008 for Ϫ20/ϩ58 versus pGL3-basic), whereas extending the sequence further upstream (Ϫ35/ϩ58 bp) to include the reported atypical TATA box (ATTAAA; Ref. 16) resulted in a decrease to levels not statistically different from the promoterless vector (Fig. 2). Since extension of the promoter past the TATA-like box had a negative rather than a positive effect on NHE3 promoter activity, we set out to address its relevance by mutating it (ATTAAA 3 ATGCAA; mutated bases in bold) in the Ϫ118/ϩ58-bp construct and testing its activity in transiently transfected Caco-2 and IEC-6 cells. Mutation of this atypical TATA box did not result in significant changes in promoter activity in either cell line (Fig. 3A). Similar results were obtained with an internal deletion construct (not shown).
Additional studies showed that 28 bp immediately surrounding the transcription start site (Ϫ20/ϩ8 bp), were able to promote low level transcription (Fig. 3B). In gel mobility shift assays with Caco-2 cell nuclear extract and a labeled Ϫ20/ϩ8 bp oligonucleotide as a probe, we were able to demonstrate a specific protein-DNA complex, which could be competed in a dose-dependent fashion with a consensus TATA box containing oligonucleotide, which was previously demonstrated to bind TATA-box binding protein (TBP; Ref. 24) (Fig. 3C). Additionally, in Southwestern blotting experiments with nuclear protein from Caco-2 and IEC-6 cells, we demonstrated weak binding of labeled Ϫ20/ϩ8 bp probe to a protein of ϳ40 kDa, which is consistent with the molecular weight of the p36 TFIID protein (Fig. 3D).
Downstream Elements-Deletion of 50 nt of exon 1 in construct Ϫ118/ϩ8 resulted in a significant reduction in NHE3 promoter activity (Fig. 4B). Computer prediction analysis (25) of putative cis elements located in this exonic region suggested binding sites for AP4 and GATA transcription factors (Fig. 2) and an overlapping binding site for AP1 and CREB. Mutation of the AP1/CREB site resulted in no significant change in promoter activity (data not shown). We therefore focused on the sequence spanning nt ϩ8 to ϩ30 and performed gel mobility shift assays with this region as labeled probe and with competing double-stranded oligonucleotides with scanning mutations in blocks of four base pairs (Fig. 4A). Only mutant 4 (M4), which included a mutated GATA box was not able to compete for nuclear protein binding, suggesting that GATA transcription factors bind to this element. Furthermore, this shifted complex could also be competed with a commercially obtained consensus GATA box oligonucleotide (Geneka) from the tal-1 gene (26). The importance of this GATA-box in the NHE3 gene was further demonstrated by introducing an internal deletion (⌬GATA) into the Ϫ118/ϩ58 construct. In transfected Caco-2 cells, the activity of this construct was significantly lower than that of the wild-type Ϫ118/ϩ58-bp plasmid, and was not significantly different from the construct with 3Ј deletion of 50 nt of exon 1 (Fig. 4B).
Proteins Binding to a ϩ20/ϩ23 NT GATA-Box-GATA-4, -5, and -6 subfamily proteins are expressed in an overlapping pattern in the developing heart and endoderm-derived organs of the gastrointestinal tract including the stomach, intestine, liver, and pancreas (27). To address binding specificity of the reported GATA box, a 20-bp GATA box-containing probe (nt ϩ10/ϩ30) was used in gel mobility shift assays. Nuclear protein from Caco-2 (Fig. 5A) and IEC-6 cells (not shown) formed a specific complex which could be competed with excess unlabeled probe (Comp. A) and a tal-1 consensus GATA box (Comp. B), but not with a probe containing a mutation in the GATA box from the NHE3 gene (Comp. C). Also, the complex could be supershifted with an antibody specific for GATA-6 (Geneka), the GATA isoform predominantly expressed in Caco-2 cells. Gel shift experiments with IEC-6 cell nuclear protein showed identical results (data not shown). Although IEC-6 cells were found by RT-PCR to express GATA-4, -5, and -6 (data not shown), we were not able to demonstrate their binding in supershift experiments due to lack of reliable antibodies for GATA-4 and GATA-5. Instead, we used nuclear protein from Qt6 cells transfected with empty expression vector or with GATA expression plasmids. Gel mobility shift assays with Qt6 cell nuclear extract confirmed that these cells lack the endogenous nuclear proteins, which bind to mammalian GATA sequences. Moreover, nuclear protein from GATA-transfected Qt6 cells formed specific complexes with the NHE3 GATA probe (ϩ10/ϩ30 nt), which upon longer separation corresponded with differences in molecular weight of respective GATA proteins (Fig. 5B). This experiment indicated that all three intestinally expressed GATA transcription factors were capable of binding to the NHE3 GATA probe.
Transactivation of NHE3 Promoter by Overexpressed GATA Transcription Factors-Forced expression of GATA-4, -5, or -6 in transiently transfected Caco-2 and IEC-6 cells resulted in increased NHE3 promoter activity (Fig. 6A), with GATA-5 being the strongest stimulator. Similar results were obtained in the non-intestinal cell lines, NIH-3T3 (data not shown) and Qt6 fibroblasts (Fig. 5C). Surprisingly, overexpression of GATA-5 did not affect the activity of a short promoter fragment (nt Ϫ20/ ϩ58), despite the fact that it contained an intact GATA box (nt ϩ20/ϩ23) (Fig. 6B), suggesting that elements upstream of nt Ϫ20 are necessary for GATA-5 stimulation of the NHE3 promoter. Extending the 5Ј-flanking region to nt Ϫ81 restored the transactivation of the promoter to levels observed with the Ϫ118/ϩ58 construct. Furthermore, mutation of three putative Sp family transcription factor binding sites, with core cis elements located at nt Ϫ71/Ϫ68 (SpA), Ϫ58/Ϫ55 (SpB; complementary strand), and Ϫ46/Ϫ43 (SpC), again eliminated the stimulatory effects of GATA-5 on NHE3 promoter activity (Fig.  6B). This suggested cooperative or synergistic effects of GATA-5 and Sp proteins interacting at one or more of the three putative Sp binding sites.
The Role of the Putative Sp Binding Sites in NHE3 Promoter Activity-Gel mobility shift assays were performed with a 42-bp DNA probe spanning all three putative Sp consensus elements and nuclear protein purified from Caco-2 and IEC-6 cells. We demonstrated that this sequence forms a specific shifted complex, which could be competed with excess unlabeled probe (Fig. 7A). We also tested the ability of individual and composite mutants of the three Sp sites to compete for binding as an indirect measure of their affinity for Sp transcription factors. Single site A and C mutants effectively competed for binding with labeled wild type probe, although not as efficiently as the unlabeled wild type probe, suggesting that these sites have very low affinity for Sp proteins. Site B mutant oligonucleotide was much less effective as a competitor, which signifies the importance of site B in interacting with Sp transcription factors. Double mutation of sites A and C did not result in decreased competition; however, double mutations of sites A and B as well as B and C reduced their ability to compete for binding. Furthermore, mutation of all three Sp binding sites resulted in complete loss of competition. Identical results were obtained with Caco-2 and IEC-6 nuclear protein (Fig. 7A). These data suggest that while site B is the predominant site of Sp/DNA interaction, sites A and C may also play a supportive role, with Sp binding to site B being a prerequisite and possibly preceding event.
The relevance of site B in nuclear protein binding is further demonstrated in Fig. 7B, where individual sites A, B, and C from the NHE3 promoter were used as competitors along with FIG. 4. The role of GATA-box (nt ؉20/؉23) in NHE3 gene promoter activity. A, gel mobility shift assays utilizing scanning mutations in competing oligonucleotides demonstrated specific binding of Caco-2 nuclear protein to a GATA box at nt ϩ20/ϩ23. In the top panel, WT is the sequence derived from NHE3 gene exon 1 and was used as a labeled probe and as a cold competitor. M1 through M6 are sequences representing progressive mutations introduced into the WT sequence in blocks of four base pairs. M4 represents a competitor with a mutated GATA box. GATA is an oligonucleotide containing a consensus GATA box derived from the tal-1 gene (see "Experimental Procedures"). All competitors were used in 50-fold excess. Below is a typical gel shift experiment. Sc and ns indicate specific and nonspecific protein/DNA complexes, respectively. B, in transiently transfected Caco-2 cells, deletion of 50 bp of exon 1 from NHE3 promoter construct (Ϫ118/ϩ58 3 Ϫ118/ϩ8) resulted in a similar decrease in reporter gene expression as compared with a construct with internal deletion of the GATA box in position ϩ20/ϩ23 (Ϫ118/ϩ58⌬GATA). Different letters next to bars indicate statistical differences (p Ͻ 0.05; data are means Ϯ S.D.; n ϭ 4 -6). a commercial Sp consensus probe and a nonspecific doublestranded oligonucleotide (TATA-box). Competitor probes with individual Sp sites were slightly modified at their 5Ј-and 3Ј-ends to avoid overlap and to provide sufficient probe length. In this series of experiments, site B was able to compete for binding along with excess wild type unlabeled probe and Sp consensus probe. Furthermore, functional data from Caco-2 cells transfected with the Ϫ81/ϩ58 bp constructs with individual or composite mutations of the putative Sp sites presented in Fig. 7C, correlated well with the results of gel mobility shift assays.
To identify proteins binding to the cluster of Sp sites in the NHE3 promoter, supershift assays were performed with the same labeled probe (see Fig. 7) and Caco-2 or IEC-6 nuclear protein. As shown in Fig. 8A, Sp1 appeared to be the predominant protein binding to this sequence in Caco-2 cells, since the majority of the specific complex was supershifted by Sp1-spe-cific antibody. A blocking antibody specific for Sp3 only weakly reduced the intensity of the complex. In IEC-6 cells, however, Sp3 was the predominant isoform found in the protein-DNA complex, as the band intensity was significantly reduced by anti-Sp3 antibody, while Sp1 antibody only very weakly supershifted the formed complex. Binding of Sp1 and Sp3 to this region of the NHE3 promoter was further demonstrated by gel mobility shift assay with nuclear protein from Drosophila SL2 cells transfected with empty expression vector or Sp1 or Sp3 expression constructs (Fig. 8B).
Synergistic Interaction between GATA-5 and Sp1 and Sp3-Results reported in Fig. 6 suggested that GATA-5 functionally interacts with Sp transcription factors to regulate NHE3 promoter activity. To determine if these transcription factors indeed act synergistically, we utilized Sp-deficient SL2 cells. Cotransfection of an NHE3 promoter construct (nt Ϫ118/ϩ58 in p␤Gal-basic) with increasing amounts (5-200 ng) of GATA-5 FIG. 6. Transactivation of NHE3 promoter by GATA transcription factors. A, activity of an NHE3 promoter construct (Ϫ118/ϩ58) in Caco-2 (black bars) or IEC-6 (open bars) cells contransfected with empty vector (pCDNA3) or with respective GATA cDNA cloned into the same vector. B, in transfected Caco-2 cells, the shortest active promoter construct (Ϫ20/ϩ58) containing an intact GATA box in exon 1 did not respond to overexpression of GATA-5 (black bars) when compared with cells contransfected with empty expression vector (gray bars). However, extension of the 5Ј-flanking region to include putative Sp transcription factor binding sites in construct Ϫ81/ϩ58 restored the response, which was again abolished by mutating all three Sp binding sites (81/ϩ58 SpMutABC). Depicted below is a scheme representation of NHE3 promoter constructs used in this experiment. ϩ1 indicates TIS (16). Data presented in panels A and B are means Ϯ S.D. (n ϭ 4 -6).
expression plasmid under control of the Drosophila actin 5 promoter, did not result in a significant increase in reporter gene activity in the absence of Sp proteins (data not shown). In order to determine the amount of transfected Sp1 or Sp3 expression plasmids that would result in submaximal stimulation of NHE3 promoter activity, we cotransfected SL2 cells with the Ϫ118/ϩ58␤Gal promoter construct along with increasing amounts of pPacSp1 or pPacUSp3 vectors. The stimulatory effect of overexpressed Sp1 saturated with a lower amount of cotransfected plasmid (Fig. 9A) as compared with Sp3, which increasingly activated reporter gene expression throughout the range of transfected plasmid amount (Fig. 9B). In subsequent experiments, the Ϫ118/ϩ58 bp NHE3 promoter construct was contransfected into SL2 cells with 20 ng of GATA-5 expression vector, 5 ng of Sp1 or 10 ng of Sp3 vector, or with a combination of GATA-5 and Sp1 or Sp3 plasmids (Fig. 9C). In agreement FIG. 7. The role of putative Sp binding sites in NHE3 gene promoter activity. A, gel mobility shift assay with Caco-2 and IEC-6 cell nuclear protein and a wild-type oligonucleotide containing a cluster of putative Sp sites (A, B, and C; nt Ϫ77/Ϫ36) in the proximal promoter of the NHE3 gene. Sequences depicted below represent the labeled probe (WT) and competing oligonucleotides used in the assay. Core Sp cis-elements are emphasized by rectangles and bold letters indicate mutations introduced to one or more of the Sp sites of competing oligonucleotides (all competitors were used in 50ϫ molar excess). B, similar gel mobility shift analysis with oligonucleotides containing individual NHE3 gene promoter Sp sites used as competitors and Caco-2 nuclear protein. SpA, B, and C sequences were slightly modified on the 5Ј-and 3Ј-ends to prevent overlap and to provide sufficient length for binding. Sp and SpM are consensus Sp1 elements from an unrelated gene and its mutated form, respectively. ns indicates a consensus TATA-box-containing oligonucleotide, which was used as a nonspecific competitor. with previously obtained data, overexpression of GATA-5 alone did not stimulate NHE3 promoter activity. However, the same amount of GATA-5 expression plasmid cotransfected with Sp1 or Sp3 resulted in synergistic activation of NHE3 promoter activity. Additive, synergistic, or antagonistic interactions among transcription factors were identified by calculating the interaction response (IR), which is a measure of comparing the effect of two expression vectors cotransfected together to the additive effect of each of the expression vectors cotransfected separately (28) as shown in Equation 1.
Values of Ϫ0.1 to 0.1 are defined as additive effects, greater than 0.1 represent synergistic effects, and less than Ϫ0.1 represent antagonistic effects. According to these criteria, both Sp1 and Sp3 activated the NHE3 promoter synergistically with GATA-5 (IR ϭ 0.44 and 0.19, respectively).

DISCUSSION
The rat NHE3 promoter was cloned in 1996 by two independent laboratories (9,16). Both reports provided genomic sequence of approximately the same region and discussed several putative, prediction analysis based, transcription factor binding sites. However, no detailed characterization of the promoter or functional mapping of cis-acting elements was presented. Moreover, different transcription initiation sites were demonstrated by both groups. Kandasamy and Orlowski (16) used S1 nuclease protection analysis and mapped it to a cluster of five nucleotides with a major site at Ϫ97 (T), and two minor sites at Ϫ100 (A) and Ϫ96 (G) nt relative to the translation start codon. Cano (29) mapped the transcription initiation site to the atypical TATA box (GGATTAAA; ϩ1 nt in bold) located at Ϫ128 nt 5Ј of the AUG initiation codon. To address this discrepancy, we utilized an alternative method to map the NHE3 gene transcriptional start site, 5Ј-RLM-RACE. Sequencing data obtained from RACE products obtained with RNA isolated from the small intestine, colon, as well as from Caco-2 cells transiently transfected with the Ϫ118/ϩ58 Luc construct, mapped the transcription initiation site to a cluster of nucleotides reported by Kandasamy and Orlowski (16). Also, functional data from transiently transfected Caco-2 and IEC-6 cells described in this paper, argue against the transcription initiation site reported by Cano (29). A reporter construct containing Ϫ20 to ϩ58 bp of the NHE3 gene lacking the TATA-like sequence (positioned at nt Ϫ30/Ϫ27) was transcriptionally active (5-fold higher then promoter-less vector). Extending the 5Јflanking region of the gene to include the only start site reported by Cano (in construct Ϫ35/ϩ58 Luc) in fact decreased the activity of the promoter to background levels. Many A/Trich sequences can convey TATA box activity, which is partly because TATA-box-binding protein (TBP) recognizes the minor DNA groove, where protein-DNA interactions are more often influenced by T/A content than by a specific sequence (30). Therefore, we also mutated or deleted this atypical TATA box in constructs Ϫ118/ϩ58mTA (Fig. 3A) and Ϫ118/ϩ58⌬TA (not shown). Both promoter constructs exhibited the same activity as the wild-type sequence in transfected Caco-2 and IEC-6 cells. Taken together, these data suggest that transcription intiation occurs independently of the presence of this TATAlike sequence, and that sequence surrounding the transcription start site as mapped by S1 nuclease protection assays and FIG. 8. Proteins binding to the cluster of Sp sites in the NHE3 promoter. Gel mobility shift assays were performed with Caco-2 and IEC-6 cell nuclear protein and WT sequence (see the legend to Fig. 7) as a labeled probe. Supershift analysis (sc, specific complex; ss, supershifted complex) indicated that Sp1 is the predominant Sp transcription factor binding this promoter region in Caco-2 cells, while Sp3 is the predominant protein in IEC-6 cells. B, similar analysis with nuclear protein from SL2 cells transfected with empty vector or Sp1 or Sp3 expression plasmid further confirmed that both Sp1 and Sp3 were capable of binding the probe. 5Ј-RLM-RACE may form the actual NHE3 gene core promoter. In support of this hypothesis, we demonstrated that the sequence spanning Ϫ20/ϩ8 bp of the NHE3 gene was able to promote low level transcription, consistent with typical low activity of isolated initiator (Inr) elements.
In mammals, the diversity of core promoters is quite significant, although precise analyses on a genome-wide scale are complicated by the lack of accurate descriptions of transcription initiation sites for the majority of genes. Promoters can be classified into those that contain a functional TATA-box, TATA-box paired with an Inr, Inr element with downstream promoter elements (DPM), and CpG island-rich promoters which apparently lack all three core elements (31). These core elements are believed to serve as recognition sites for the TFIID complex, which contains TBP and various TBP-associated factors (TAFs) (32). Although far from being conclusive, our data obtained from gel mobility shift assays as well as Southwestern blotting (Fig. 3, C and D) suggests that TBP bind to the NHE3 core promoter Inr element, despite the fact that it has a weak homology to the otherwise loose Inr consensus sequence (Py-Py-A ϩ1 -N-(T/A)-Py-Py) (33).
Downstream control elements located 3Ј of the transcription initiation site within Inr-containing promoters have been found essential for promoter activity in vivo, e.g. in the adenoviral major late (AdML) (34) and murine terminal transferase (TdT) (35) genes. Although some studies indicated that TFIID, TFII-I, cap-binding protein (CAP), or USF proteins may bind to these regions, downstream elements remain poorly understood. In the NHE3 promoter, deletion of nt ϩ9/ϩ58 resulted in a significant loss of promoter activity. Further studies unequivocally demonstrated that a GATA-box, located at position ϩ20/ ϩ23, was critical for promoter function, since deletion of it was equivalent to the gross deletion of 50 bp of exon 1. We further demonstrated that all three GATA isoforms expressed in the intestinal epithelium, GATA-4, -5, and -6 are capable of binding to this GATA sequence. Of these three transcription factors, however, GATA-5 was capable of the strongest NHE3 promoter transactivation when overexpressed in cell lines of both intestinal and non-intestinal origin. Strikingly, shorter constructs or mutated constructs that lacked putative Sp transcription factor binding sites upstream of TIS, yet had preserved GATAbox in exon 1, failed to respond to overexpression of GATA-5. This suggested that the presence of Sp transcription factors binding their upstream cis-elements is necessary for GATA-5 to exert its stimulatory effect. We further confirmed this observation by demonstrating that overexpression of GATA-5 in Spdeficient SL2 cells had no effect on NHE3 promoter activity.
To delineate the role of Sp transcription factors in regulating NHE3 promoter activity, we performed a series of gel mobility shift and functional analyses of individual or composite mutations of the three putative Sp binding sites in transiently transfected Caco-2 cells. From these experiments, we concluded that site B, with core binding sequence located at nt Ϫ58/Ϫ55, represents the highest affinity Sp protein binding site. We also observed that on a functional level this site acts in concert with the two neighboring Sp elements to control basal activity of the NHE3 promoter. Furthermore, this cluster of Sp binding sites was shown to bind both Sp1 and Sp3 proteins, as demonstrated by supershift analysis with nuclear protein from Caco-2 and IEC-6 cells and specific antibodies, as well as by gel mobility shift assays with transfected SL2 cells. Both Sp1 and Sp3 stimulated NHE3 promoter activity in cotranfected SL2 cells in a dose-dependent manner.
Functional interaction between GATA-5 and Sp1 or Sp3 was confirmed in experiments with SL2 cells, in which GATA-5 considerably potentiated the stimulatory effects of Sp1 or Sp3. Such synergy has not been described before and therefore represents a novel mechanism of transcriptional regulation. Within the intestinal epithelium, GATA-5 has recently been shown to act synergistically with hepatocyte nuclear factor 1␣ (HNF1␣; Ref. 28), and both transcription factors were shown to physically interact in regulating the activity of the human lactase-phlorizin hydrolase gene (36). It remains to be determined whether GATA-5 physically interacts with Sp1 or Sp3. The proximity of GATA sites and the HNF1␣ binding site in the lactase promoter (both core cis-elements within 23 bp) may facilitate physical interaction. In the NHE3 promoter, however, the cluster of Sp binding sites and the GATA box are fairly distant (ϳ100 nt) and divided by the putative initiator and binding of RNA polymerase II holoenzyme. If such an interaction occurred, whether direct or indirect, it would most likely result in looping the initiator sequence out, and it may facilitate recruitment or assembly of the basal transcriptional machinery. Interestingly, in cardiac myocytes, GATA-5 has been found to physically interact with p300 (37). p300 proteins are negative regulators of cell proliferation, which function as transcriptional coactivators that mediate the interaction between transcription factors binding on enhancer DNA sequences and the basal transcriptional complex formed on the core promoter (38). The acetyltransferase region of p300 can also functionally and physically interact with Sp1, stimulating its DNA binding properties (39). It is therefore conceivable that GATA-5 may indirectly interact with Sp1 or Sp3 via an adaptor protein such as p300, linking them to the basal transcriptional complex to stabilize or enhance its activity.
In conclusion, we present a functional model of basal regulation of the rat NHE3 gene promoter. In this model GATA-5, through binding to a GATA element in exon 1, enhances transcription through functional interaction with Sp1 or Sp3 binding upstream of an atypical initiator sequence. Intestinal GATA-5 is believed to function during differentiation to activate genes in fully differentiated absorptive cells of the villi (40). Therefore, the cooperative activation of an intestinal promoter through GATA-5 and Sp family transcriptional factors likely represents a novel mechanism of regulating expression of genes along the crypt-villus axis of the intestinal epithelium.