NeuroD1/E47 Regulates the E-box Element of a Novel Zinc Finger Transcription Factor, IA-1, in Developing Nervous System*

IA-1 is a novel zinc finger transcription factor with a restricted tissue distribution in the embryonic nervous system and tumors of neuroendocrine origin. The 1.7-kilobase 5′-upstream DNA sequence of the human IA-1 gene directed transgene expression predominantly in the developing nervous system including forebrain, midbrain, hindbrain, spinal cord, retina, olfactory bulb, and cerebellum, which recapitulated the expression patterns of neuroendocrine tissues and childhood brain tumors. The IA-1 promoter deletion reporter gene constructs revealed that the sequence between -426 and -65 bp containing three putative E-boxes (∼361 bp) upstream of the transcription start site was sufficient to confer tissue-specific transcriptional activity. Further mutation analysis revealed that the proximal E-box (E3) closest to the start site is critical to confer transcriptional activity. Electrophoretic mobility shift assay and transient transfection studies demonstrated that the NeuroD1 and E47 heterodimer are the key transcription factors that regulate the proximal E-box of the IA-1 promoter. Therefore, we concluded that the IA-1 gene is developmentally expressed in the nervous system and the NeuroD1/E47 transcription factors up-regulate IA-1 gene expression through the proximal E-box element of the IA-1 promoter.

The diffuse neuroendocrine system includes pancreatic islets, gastrointestinal and respiratory neuroendocrine cells, thyroid C, adrenal medulla, and pituitary cells (1). They share phenotypes with neuronal cells by expressing common neuroendocrine markers and signaling pathways.
A number of basic helix-loop-helix (bHLH) 1 transcription factors, the neurogenins, Hes-1, and NeuroD1 (also known as ␤2), have been shown to play important roles in both pancreatic endocrine and nervous system development (2)(3)(4)(5). bHLH transcription factors play a critical role in the cell type-specific expression of a variety of genes in many different tissues (6,7).
The bHLH proteins are classified into two separate groups based on their DNA binding properties and tissue distribution. In general, the class A members are ubiquitous factors including E47 and E12. The class B members are tissue-specific. However, the class B proteins can dimerize with the class A proteins as heterodimers and bind to DNA with high affinity and confer tissue-specific expression (8 -12). NeuroD1 is a bHLH transcription factor that was shown to be associated with late neuronal differentiation in Xenopus laevis (13). Gene targeting experiments revealed that deletion of the NeuroD1 gene resulted in defective pancreatic morphogenesis and abnormal enteroendocrine differentiation, which led to the early development of diabetes (14). Furthermore, NeuroD1 is required for differentiation of the granule cells in the cerebellum and hippocampus (4,5). NeuroD1 regulates several downstream target genes in the pancreatic islets, the intestine, the pituitary, and the developing neural retina (15)(16)(17)(18)(19). Previously, we have identified NeuroD1 as a potential target gene of a novel transcriptional repressor, insulinoma-associated antigen-1 (IA-1) (20). Interestingly, in this study, we show evidence that NeuroD1 also regulates IA-1 gene expression.
IA-1 encodes a novel zinc finger DNA-binding protein that was isolated from a human insulinoma subtraction library (21). In vitro induction of the AR42J amphicrine cell line into insulin-producing cells suggested that IA-1 gene expression is closely associated with the expression of islet-specific transcription factors including NeuroD1 (22). Functional studies revealed that IA-1 is a transcriptional repressor that binds to a specific DNA element and can autoregulate itself and the Neu-roD1 gene (20). The IA-1 gene has a very restricted expression pattern to tumors of neuroendocrine origin (21,23), including insulinoma, medulloblastoma, retinoblastoma, pituitary tumor, pheochromacytoma, medullary thyroid carcinoma, and small cell lung carcinoma. This pattern of expression largely overlaps the expression pattern observed for NeuroD1 (24). However, little is known about the transcriptional regulation of the IA-1 gene in neuroendocrine tissues. In this study, we have analyzed the IA-1 promoter activity in vivo by introducing an IA-1 promoter-LacZ transgene and monitoring the expression of LacZ activity in the transgenic mice. In addition to the restricted expression pattern of the IA-1 gene, we have identified both the E-box element and the E47-NeuroD1 transcription factors that contribute to the tissue restricted expression of the IA-1 gene. Since IA-1 possesses transcriptional repressor activity against the NeuroD1 gene and their expression patterns largely overlap in the neuroendocrine cells, it suggests that the NeuroD1 and the IA-1 gene may counterregulate their expression levels during nervous system development. blastoma), U87MG (human glioblastoma), and ␤TC-1, (mouse insulinoma) cells were maintained in Dulbecco's minimal essential medium with either high or low glucose supplemented with 10% heat-inactivated fetal bovine serum (Invitrogen), 100 units/ml penicillin, and 100 g/ml streptomycin. Twenty-four hours prior to the transfection, the adherent cell lines were seeded at a density of 0.25-0.3 ϫ 10 6 cells/well in a 6-well culture dish. Alternatively, the suspension cell lines were collected by centrifugation on the day of transfection and seeded at a density of 0.5 ϫ 10 6 cells/ml. A 1:1 ratio of 5 l of LipofectAMINE 2000 reagent (1 mg/ml; Invitrogen) to 5 g of DNA was used according to the manufacturer's instructions. The DNA complex was added per well in the absence of serum for 5 h at 37°C. The medium was removed and replaced with fresh fetal bovine serum containing medium for a total of 48 h at 37°C. The cells were collected, lysed in 400 l of reporter lysis buffer by repeated freeze-thaw, and assayed for CAT (Promega) and ␤-galactosidase (Promega) activity or the protein concentration was determined using the BCA protein assay kit (Pierce).
Construction of the IA-1 Promoter-LacZ Reporter Transgene and the Generation of Transgenic Mice-The 5Ј-flanking region (Ϫ1661 to ϩ40 bp) of the human IA-1 gene was subcloned from two human genomic clones, pL3S2.0 and pL3S0.8 (25). A 1.7-kb fragment was subcloned into pBluescript (Stratagene) containing the 3.7-kb Escherichia coli LacZ gene with a SV40 poly(A) ϩ tail. The transgene fragment (5.4 kb) was excised from the parental plasmid by digestion with XhoI and BamHI. The fragment was purified and contracted with the NICHD Transgenic Mouse Development Facility at the University of Alabama at Birmingham for transgenic mouse production. Founder mice were identified by PCR amplification of genomic DNA prepared from tails. Forward primer pIA-1 (5Ј-AAGCGGGAGGCGAGAACAAT-3Ј) and reverse primer ␤-gal/fusion (5Ј-TTCGCTATTACGCCAGCTGG-3Ј) yielded a PCR product of 501 bp overlapping IA-1 promoter and ␤-gal sequences using a touch-down PCR condition. The PCR cycles are as follows: in the first step, 94°C for 45 s, 68°C (each cycle decrease 0.5°C) for 45 s, and 72°C for 1 min for 19 cycles; in the subsequent step, 94°C for 45 s, 58°C for 45 s, and 72°C for 1 min for 19 cycles; last, extension for 10 min in 72°C.
Southern Blot Analysis of Copy Numbers from Transgenic Genomic DNA-To measure the copy number of transgene in three lines of transgenic mice, we performed a Southern blot analysis using known concentrations of the transgene as control. Fifteen micrograms of normal mouse or transgenic mouse tail genomic DNA were digested with BamHI and XhoI and then separated on 0.8% agarose gel electrophoresis overnight. After denaturation and neutralization, the gel was transblotted onto nitrocellulose membrane for Southern blot. The IA-1p-LacZ transgene was radiolabeled using a random prime labeling kit (Invitrogen). The hybridization procedure was followed as previously described (25). The copy number of the transgene was estimated by quantification of band intensity using a Bio-Rad image analysis system.
Whole Mount Histochemical Detection of ␤-Galactosidase Activity-For timed pregnancy and embryo staging, the morning of vaginal plug observations were considered as e0.5 embryo, and tissues were dissected at the designated ages, fixed in solution containing 0.2% glutaraldehyde, 1.5% formaldehyde, 5 mM EGTA, 2 mM MgCl 2 , and 100 mM sodium phosphate, pH 8.0, for 30 min to 2 h at room temperature depending upon the embryo size. The embryo and tissue were washed three times with 100 mM sodium phosphate buffer, pH 8.0. ␤-Galactosidase activity was developed in staining solution consisting of 1 mg/ml X-gal, 5 mM K 4 Fe(CN) 6 , 5 mM K 3 Fe(CN) 6 in sodium phosphate buffer, pH 8.0, at room temperature for up to 48 h. Whole mount stained embryo was photographed through a dissecting microscope. X-galstained embryos or tissues were further fixed in 10% buffered formalin for storage or subjected to paraffin embedding.
DNA Constructs-The various IA-1 promoter CAT constructs were created by subcloning the human IA-1 Ϫ1661/ϩ40 bp promoter fragment into the XhoI site of the pCAT3 basic vector (Promega). The Ϫ426/ϩ40 bp construct was created by NheI digestion of the Ϫ1661/ϩ40 bp IA-1p/CAT3 construct followed by religation to eliminate the Ϫ1661 to Ϫ427 bp region. The Ϫ65/ϩ40 bp construct was created by subcloning into the SacI/SmaI site in the pCAT3 basic vector. The E47 expression construct was kindly provided by Dr. R. Stein (Vanderbilt University). The pCR3.1␤2 construct was kindly provided by Dr. M. J. Tsai (Baylor College of Medicine). The E-box deletion constructs were created by using the Ϫ426/ϩ40 bp IA-1 promoter/CAT3 plasmid as a template. Mutagenesis was performed using the QuikChange site-directed mutagenesis kit (Stratagene). Mutagenic oligonucleotides were designed to bind to complementary strands of the IA-1 promoter spanning the individual E-box elements (⌬E1 Ϫ361 to Ϫ356 bp, ⌬E2 Ϫ315 to Ϫ310 bp, ⌬E3 Ϫ182 to Ϫ177 bp) to delete the 6-bp binding site. The synthetic oligonucleotides used were ⌬E1 (Ϫ361 to Ϫ356 bp; 5Ј-ATTCTCGCGC-TGATGGACGGGCCGCGGCTCCGCGCCCCCCGGAGGAGA-3Ј), ⌬E2 (Ϫ315 to Ϫ310 bp; 5Ј-GACACAAAGCCCAGGCGCCTCCCCATAGAG-3Ј), ⌬E3 (Ϫ182 to Ϫ177 bp; 5Ј-CCGCGCCCTCAGGTACCGCACCTAC-CGGGC-3Ј), and the complementary strand primers, respectively. PCR conditions were performed using 1ϫ PCR buffer (20 mM Tris-HCl pH 8.8, 10 mM KCl, 10 mM (NH 4 )SO 4, 2 mM MgSO 4 , 0.1% Triton X-100, 0.1 mg/ml bovine serum albumin), 125 ng of each primer, 20 ng of doublestranded DNA, and 2 units of Pfuturbo DNA polymerase in a 50-l reaction. PCR was performed using an initial denaturation step of 95°C for 30 s followed by 18 cycles of denaturation at 95°C for 30 s, annealing at 60°C for 30 s (E-box 2 and 3) or 63°C (E-box 1), and extension at 68°C for 10 min. The PCR products (10-l aliquot) were visualized on a 1% agarose gel, and the remainder of the reaction mixture was digested with DpnI restriction endonuclease for 1 h at 37°C. Following digestion with DpnI, 5 l of the mutagenesis reaction was transformed into XL-1 Blue competent cells. Mutants were confirmed by sequence analysis (U.S. Biochemical Corp.).
The 3ϫ E-box constructs were created using synthetic oligonucleotides containing three tandem copies of the individual E-box elements: E-box 1 oligonucleotide (5Ј-CCCATTTGGCCCATTTGGCCCATTTGGA-3Ј), E-box 2 oligonucleotide (5Ј-GGCACGTGCGGCACGTGCGGCACG-TGCGA-3Ј, and E-box 3 oligonucleotide (5Ј-TACATCTGCCTACATCT-GCCTACATCTGCCA-3Ј) and their complementary strand oligonucleotides. The complementary oligonucleotides were annealed by mixing equal molar amounts of oligonucleotides and heating at 95°C for 5 min and slow cooled at 1°C/min to room temperature. The oligonucleotides were synthesized with an A overhang to allow for cloning into the TA TOPO vector (Invitrogen). The 3ϫ E-box constructs were sequenced to determine orientation and then subcloned into the HindIII/XhoI site or HindIII/XbaI cloning site of the E1bTATACAT vector.
Electrophoretic Mobility Shift Assay (EMSA)-EMSA analysis was performed using a double strand oligonucleotide spanning the Ϫ192 to Ϫ165 bp region of the IA-1 promoter (containing E-box 3), 5Ј-CCCT-CAGGTACATCTGCCGCACCTACCG-3Ј, and the complementary strand. The double strand oligonucleotide was end-labeled using [␥-32 P]ATP (3000 Ci/mmol; PerkinElmer Life Sciences) and T 4 polynucleotide kinase (New England Biolabs). E47 and NeuroD1 proteins were synthesized using the expression plasmids pCR3.1␤2 and pcDNA3E47 and the TNT-coupled rabbit reticulocyte lysate kit (Promega). Synthesis of E47 and NeuroD1 was confirmed by Western blot analysis using an anti-E47 rabbit polyclonal antibody (Santa Cruz Biotechnology, Inc., Santa Cruz, CA) or an anti-NeuroD1 rabbit polyclonal antibody (CeMines). The EMSA binding reaction included various amounts of in vitro translated proteins in a binding reaction composed of 10 mM Tris-Cl, pH 7.5, 50 mM NaCl, 1 mM EDTA, 1 mM dithiothreitol, 5% glycerol, 200 ng of poly(dI⅐dC), and 5 ϫ 10 4 cpm of probe. For supershift assay, 1 l of a 1:5 dilution of E47 antibody (Santa Cruz Biotechnology) or a 1:2.5 dilution of NeuroD1 rabbit polyclonal antibody (CeMines) was added to the mixture following a 10-min preincubation at room temperature with in vitro translated protein and probe. The whole mixture was then incubated at room temperature for an additional 20 min. The protein-DNA complexes were resolved on a 4% PAGE (40:1) gel in 0.25ϫ TBE buffer. The gels were dried and exposed to autoradiography.
Northern Blot Analysis-Total RNA was extracted from Daoy, D283MED, U87MG, WERI-Rb1, Y79, and ␤TC-1 cell lines using Trizol reagent (Invitrogen) according to the manufacturer's instructions. Twenty micrograms of total RNA were separated on a 1% agarose/ formaldehyde gel in 1ϫ MOPS running buffer. Northern blot analysis was carried out as previously described (21). The full-length human IA-1 cDNA (2.8 kb of EcoRI fragment), a partial mouse IA-1 cDNA fragment (1.3 kb), and full-length hamster ␤2 cDNA (a kind gift from Dr. Ming Tsai; Baylor College of Medicine) probes were random primelabeled (Invitrogen) in the presence of [␣-32 P]dCTP (PerkinElmer Life Sciences). 1 ϫ 10 6 cpm/ml of labeled probe was added to the hybridization mixture. Following an overnight incubation at 55°C, the blots were washed three times for 15 min at 55°C in 0.1ϫ SSPE, 0.1% SDS and exposed to autoradiography at Ϫ70°C for 2 days. A photograph of the ethidium bromide-stained agarose gel prior to transfer is shown for a lane to lane loading comparison.

RESULTS
Generation of IA-1p-LacZ Transgenic Mice-Previous analysis of a Ϫ2090/ϩ169 bp fragment of the IA-1 5Ј-flanking sequence demonstrated that the promoter was active in AtT-20 cells and not detectable in HeLa cells, consistent with the IA-1 tissue-restricted mRNA expression pattern (26). In order to analyze the IA-1 promoter activity in vivo, we subcloned a 1.7-kb 5Ј-upstream sequence of the IA-1 gene in front of a LacZ reporter gene as a transgenic construct. We have successfully generated three lines of IA-1 promoter/LacZ transgenic mice, named Tg 2-5, 2-7, and 2-8. The estimated copy number and the LacZ expression pattern in each line of transgenic mice was determined by Southern blot analysis and ␤-gal activities. High copy numbers were found in two lines of transgenic mice, Tg 2-5 (ϳ40 -50 copies) and Tg 2-8 (ϳ80 -90 copies), whereas Tg 2-7 contains a low copy number (Ͻ5 copies).
Strong Expression of the IA-1p-LacZ Transgene in Nervous System-We performed a whole mount embryo analysis of ␤-gal activity from timed pregnant female mice. Early embryos were collected for whole mount fixation and stained for ␤-gal activity. The blue color (Fig. 1) represents ␤-gal activity driven by the IA-1 promoter sequence (Ϫ1661 to ϩ40 bp). Strong staining was observed in the nervous system of the early stage embryo (e11.5-e14.5) in the forebrain, midbrain, hind brain, spinal cord, cerebellum, olfactory bulb, and retina ( Fig. 1). No ␤-gal activity was detected in other tissues. In general, the staining patterns of Tg 2-5 and Tg 2-8 are similar. There are differential intensities between the Tg 2-5 and Tg 2-8 lines. For example, the retina staining in the Tg 2-8 was stronger than in Tg 2-5 (e11.5). Further sectioning of ␤-gal stained embryos (e12.5 day) revealed strong activities in the olfactory bulb, olfactory epithelium, retinal neuron, lens, spinal cord, and trigerminal ganglion. We also stained the postpartum brains collected from e16.5, e18.5, newborn, 1 week, and adult. Strong staining was observed in cerebellum and pineal gland at e18.5, whereas the LacZ activity gradually decreased while the brain matured to adulthood (Fig. 2). Age-matched nontransgenic control animals failed to show any detectable levels of ␤-gal staining in the brain at the time points that were chosen (data not shown). The earliest onset of ␤-gal staining was observed in e9.5 embryos with a less intense staining pattern in the midbrain, forebrain, hind brain, and spinal cord. Detection of LacZ expression at e9.5 precedes the previously reported endogenous expression of mouse IA-1 mRNA (27).
The Ϫ426/ϩ40 bp IA-1 Promoter Sequence Contains the Tissue-specific Activity-Based upon the results of the in vivo IA-1 promoter/LacZ transgenic mice, we demonstrated that the Ϫ1661/ϩ40 bp human IA-1 promoter region is sufficient to direct both the proper spatial and temporal expression of the IA-1 gene. In order to determine the minimal region(s) necessary to direct tissue-specific expression, three separate deletion constructs of the IA-1 promoter were linked to the CAT reporter gene. We analyzed the promoter activity in three different cell lines, HeLa, Y79, and ␤TC-1 (Fig. 3). Comparison of the three deletion constructs revealed that the Ϫ426/ϩ40 bp/CAT construct activated the reporter gene to the same level as the Ϫ1661/ϩ40 bp construct in both Y79 and ␤TC-1 cells. However, the overall promoter activity was significantly higher in Y79 (ϳ40-fold increase) as compared with ␤TC-1 cells (ϳ7-fold increase). As expected, promoter activity was much lower in HeLa cells (ϳ3-fold). Therefore, this result demonstrates that the Ϫ426/ϩ40 bp region contains most of the cis elements necessary to direct the proper tissue-specific expression of the IA-1 gene.
E-box 3 Is Critical for IA-1 Promoter Activity-Close inspection of the Ϫ426/ϩ40 bp IA-1 promoter region revealed the  2. The postpartum brains collected from e16.5, e18.5, newborn, 1 week, and adult were also subjected to ␤-gal staining. Strong staining was observed in cerebellum and pineal gland at e16.5 to e18.5 days, whereas the LacZ activity was decreased while the brain matured to adulthood.
presence of three putative E-box elements. In order to demonstrate the importance of the E-box elements in the IA-1 promoter, we performed site-directed mutagenesis of the E-box elements singly or in various combinations from the Ϫ426/ϩ40 bp IA-1 promoter/CAT reporter construct (designated as wild type (WT in Fig. 4)). We assessed the overall activity each of the E-box elements contributes to the Ϫ426/ϩ40 bp IA-1 promoter activity in ␤TC-1 cells. Deletion of one E-box element at a time demonstrated that deletion of E-box 3 (E3) resulted in a significant loss in promoter activity (ϳ70%) in ␤TC-1 cells (Fig. 4). Deletion of E1 or E2 alone did not result in a significant change in basal promoter activity. Double deletion of E1 and E2, as expected, did not show any significant change to overall basal promoter activity (Fig. 4). As expected, double deletion of either E1 and E3 or E2 and E3 or triple deletion of E1, E2, and E3 mimicked the single deletion of E3 alone, resulting in a ϳ75, 60, or 60% loss, respectively, in promoter activity when compared with the wild type promoter. Therefore, this result demonstrates the importance of the E3-box for tissue-specific IA-1 promoter activity.
E47-NeuroD1 Heterodimer Activates the Ϫ426/ϩ40 bp IA-1 Promoter-Deletion analysis of the IA-1 promoter in various cell lines demonstrated that the Ϫ426/ϩ40 bp region contained the majority of the critical information necessary to result in the tissue-specific expression of the IA-1 gene (Fig. 3). Further analysis of this region revealed the presence of three E-box elements in this region. Deletion of the individual E-box elements revealed that E3 was crucial for 70% of the basal promoter activity in ␤TC-1 cells. The sequence of the E3 element resembles a NeuroD1 binding site found in both the rat insulin and glucagon genes (16). Therefore, we performed transient transfections with the wild type Ϫ426/ϩ40 bp IA-1 promoter/ CAT construct to determine whether NeuroD1 could activate the IA-1 promoter/CAT construct. Using various concentrations of a NeuroD1 expression vector, a dosage-dependent increase in IA-1 promoter activity was observed in HeLa cells (Fig. 5A). NeuroD1 is a tissue-restricted bHLH protein that binds to a 6-bp E-box element as a heterodimer with a ubiquitous partner E47 (15). Therefore, we then performed transient transfections in HeLa cells using the Ϫ426/ϩ40 bp IA-1 promoter/CAT construct along with a fixed amount of E47 and NeuroD1 expression vectors to demonstrate that there was a synergistic effect of co-expression of these two molecules on the IA-1 promoter construct. As shown in Fig. 5B, transfection of an E47 expression vector alone could not stimulate the IA-1 promoter/CAT activity; however, the combination of both E47 and NeuroD1 resulted in a higher increase in CAT activity (ϳ4.5-fold) than transfection of NeuroD1 alone (ϳ2-fold) (Fig.  5B). Therefore, this result demonstrates that the IA-1 promoter is activated by the E47-NeuroD1 heterodimer and is responsible at least in part for the tissue-specific expression of the IA-1 gene.
Activation of the Individual E3-box by E47-NeuroD1 Heterodimer-Deletion analysis of the three individual E-box elements from the IA-1 promoter demonstrated that E3 is critical for tissue-specific IA-1 promoter activity. Co-expression of E47-NeuroD1 could significantly activate the Ϫ426/ϩ40 bp IA-1 promoter/CAT construct. To demonstrate whether an individual E-box or a combination of the E-boxes could indeed mediate the activating effect of E47-NeuroD1, we cloned three tandem copies of each E-box into a TATA box-containing CAT reporter (E1bTATApCAT3). The individual E-box/CAT constructs and Schematic diagram of the IA-1 promoter deletion constructs that were subcloned in a pCAT3-basic vector. The IA-1 5Ј-flanking sequence (Ϫ1661/ϩ40 bp) used for the transgenic study and the two other fragments (Ϫ426/ϩ40 and Ϫ65/ϩ40 bp) were subjected to a reporter gene assay. Both 1.7 kb and Ϫ426 bp fragments confer maximal promoter activity, which is consistent with the IA-1 expressing cell lines (Fig. 1). The CMV-␤-gal vector was used as an internal control to normalize transfection efficiency. The transfections were repeated three times.

FIG. 4. E3-box is important for the IA-1 promoter activation.
Mutational analyses of the E-box elements in the IA-1 promoter (Ϫ426/ ϩ40 bp) in ␤-TC1 cells. The 6-bp E-box was deleted from the Ϫ426/ϩ40 bp IA-1 promoter construct as shown in the hatched box. All deletion constructs were compared with the wild-type Ϫ426/ϩ40 bp IA-1 promoter activity. The deletion of the E3-box diminished by more than 60% the IA-1 promoter activity. The CMV-␤-gal vector was used as internal control to normalize transfection efficiency. The transfections were repeated three times. E47-NeuroD1 cDNA expression vectors were co-transfected into HeLa cells. Transfection of the E47 expression vector alone resulted in a ϳ3-fold increase in 3ϫ E1 or 3ϫ E2 activity and a ϳ28-fold activation of E3 (Fig. 6). However, the intact Ϫ426/ ϩ40 bp IA-1 promoter/CAT construct could not be activated by the addition of E47 cDNA alone (Fig. 5B). The reason for this observed difference between the two experiments may be due to in part the inclusion of a heterologous promoter region (E1bTATA box). Alternatively, the 3ϫ E-box constructs may lack critical flanking sequences required for the binding specificity of the E47-NeuroD1 heterodimer. Co-transfection of E47 and NeuroD1 expression vectors did not result in a significant change in the 3ϫ E1 or the 3ϫ E2-box activities; however, the 3ϫ E3 activity was stimulated by ϳ47-fold, indicating that the E47-NeuroD1 heterodimer most likely exerts its stimulatory activity through the E3 element. The activation of the 3ϫ E3 was significantly higher than with the Ϫ426/ϩ40 bp IA-1 promoter construct. One reason for this discrepancy could be the presence of three copies of the E3-box that results in higher -fold activation than with the wild type promoter. Additionally, the 3ϫ E3-box is linked to a heterologous minimal promoter that may behave differently than the intact IA-1 promoter region.
E47-NeuroD1 Heterodimers Bind to the E3-box (Ϫ192 to Ϫ165 bp Region) in an Electrophoretic Mobility Shift Assay-Transient co-transfection experiments using E47 and NeuroD1 cDNA expression vectors demonstrated a synergistic activation of the wild type (Ϫ426/ϩ40 bp) IA-1 promoter or the 3ϫ E3/ CAT promoter constructs. We further performed EMSA analysis to confirm binding of NeuroD1 protein to the E3-box in the IA-1 promoter. The cDNAs of E47 and NeuroD1 were transcribed and translated in vitro, and the proteins were incubated either alone or together with a double-stranded oligonucleotide spanning the Ϫ192 to Ϫ165 bp region in the IA-1 promoter (containing E3). The addition of E47 protein alone resulted in a single complex, different from the control rabbit reticulocyte lysate (Fig. 7A, lanes 1 and 2). Co-translation of E47 and NeuroD1 proteins generated one complex with a faster mobility than the E47 homodimer complex (Fig. 7A, lane 5). The addition of excess E47 protein results in the formation of two complexes, one slower migrating complex that co-migrates with the E47 homodimer complex and a faster migrating complex that represents an E47-NeuroD1 heterodimer (Fig. 7A,  lane 8). The identities of the complexes were confirmed by antibody supershift experiments. The addition of an E47 antibody results in the supershift of the slower migrating complex or an elimination of the slower and faster migrating complexes, demonstrating that E47 is present in both complexes (Fig. 7A,  lanes 3 and 6). NeuroD1 protein alone does not form a complex (data not shown), whereas NeuroD1 antibody does not supershift but partially blocks the E47-NeuroD1 complex (Fig. 7A,  lane 7). This result clearly demonstrates that E47-NeuroD1 heterodimers bind to the E3-box in the IA-1 promoter. Additionally, we performed competition experiments using the 3ϫ E3 oligonucleotide along with 50-and 100-fold molar excesses of cold 3ϫ E3 (self), 3ϫ E1, and 3ϫ E2 (Fig. 7B). Incubation of E47 and NeuroD1 proteins with radiolabeled 3ϫ E3 oligonucleotide in the absence of competitor resulted in the formation FIG. 6. Homodimer and heterodimer of E47 and NeuroD1 activated the individual E3-box of the IA-1 promoter. Three copies of each E-box (E1, E2, and E3) were cloned into E1bTATA-CAT vector. Each reporter vector was co-transfected with pcDNA, NeuroD1, E47, or E47-NeuroD1 into HeLa cells. The E47 homodimer and the E47-Neu-roD1 heterodimer showed a strong activation of the E3-box construct. The CMV-␤-gal vector was used as internal control to normalize transfection efficiency. The transfections were repeated three times. FIG. 7. Electrophoretic mobility shift assay. A, the NeuroD1/E47 heterodimer and the E47 homodimer can specifically bind to the E3-box. The radiolabeled IA-1 promoter (Ϫ192/Ϫ165 bp) containing the E3-box was incubated with E47 or E47-NeuroD1 with or without antibodies to E47 or NeuroD1. Both the homodimer and the heterodimer were shown in shifted bands. E47 antibody supershifted the homodimer, whereas the NeuroD1 antibody interfered with the heterodimer binding. NS, nonspecific band. B, competitive inhibition of the E47-NeuroD1 complex binding with 3ϫ E3-box. Incubation of E47 and NeuroD1 proteins with labeled 3ϫ E3 oligonucleotide in the presence or absence of competitor (50-or 100-fold molar excess) showed that E47-NeuroD1 binding is specific for the E3-box alone. of a complex that represents the E47-NeuroD1 heterodimer. The addition of a 50-fold molar excess of cold 3ϫ E3 oligonucleotide results in a ϳ50% inhibition in complex formation, whereas the addition of a 100-fold molar excess results in the complete inhibition of the E47-NeuroD1 complex formation (Fig. 7B). As expected, the addition of a 50-or 100-fold molar excess of cold 3ϫ E1 or 3ϫ E2 oligonucleotide did not compete with the E47-NeuroD1 DNA-protein complex formation, demonstrating that the interaction between the E3-box and E47-NeuroD1 is specific for the E3-box alone.
Northern Analysis of Various Neuroendocrine Cell Lines-Northern blot analysis for IA-1 mRNA has demonstrated a restricted expression pattern during early mouse embryo development (27) and in tumors of neuroendocrine origin (21,23). In order to determine whether the expression of IA-1 overlaps with NeuroD1, a potential regulatory protein for IA-1 expression, we performed Northern blot analysis on a select number of human neuroendocrine tumor cell lines. Northern blot analysis was performed using Daoy, D283MED, U87MG, WERI-Rb1, Y79, and ␤TC-1 cell lines. Shown in the left panel of (Fig.  8) is the result with a mixture of full-length human IA-1 and mouse IA-1 cDNA probes. IA-1 expression was strongest in ␤TC-1 cells, followed by D283MED, WERI-Rb1, and finally Y79 cell lines (Fig. 8A). Similarly, NeuroD1 expression is strongest in ␤TC-1 and D283MED cell lines (Fig. 8B). However, in contrast to IA-1, the level of NeuroD1 is higher in WERI-Rb1 and Y79 cell lines. Daoy and U87MG are negative for both IA-1 and NeuroD1 expression. Therefore, the overall expression pattern of IA-1 and NeuroD1 completely overlap in the cell lines tested, supporting a role for the potential regulation of IA-1 by NeuroD1. DISCUSSION Endocrine cells of the pancreas and neuronal cells, despite their different embryological origin, co-express a variety of tissue-specific transcription factors that are critical for their formation (28,29). We isolated and identified a novel zinc finger transcriptional repressor, IA-1, from a human insulinoma subtraction library, which is restricted to a limited number of tissues during early embryonic development and is also activated in tumors of neuroendocrine origin (20,21).
Based upon the tissue-restricted expression pattern observed for the IA-1 mRNA, we sought to identify regions within the IA-1 gene responsible for directing its proper tissue-specific expression. Using in vivo generation of the IA-1 promoterdriven LacZ transgenic mouse line, we convincingly demonstrated that the information required for the tissue-specific expression of the IA-1 gene is present in the 5Ј-upstream regulatory region. Staining of the early stage whole mouse embryos showed prominent staining in regions of the central nervous system such as the forebrain, midbrain, hind brain, cerebellum, spinal cord, retina, and olfactory bulb. Analysis of brains from e16.5, e18.5, newborn, 1-week-old, and adult mice established a strong expression of the IA-1 gene during early brain development that was rapidly lost as the mice matured. Our results showed the earliest time point for transgene expression was e9.5. Northern blot analysis using whole mouse embryo total RNA showed that the first detectable lA-1 message is at e10.5 (27). This discrepancy is most likely due to the increased sensitivity of detection of LacZ expression due to the high copy number of the transgene construct. Endogenous IA-1 expression was shown to peak around E11.5 and declined by E18.5 in the whole mouse embryo (27). Northern blot analysis on postpartum mouse brain RNA revealed strong expression of IA-1 in e17.5 brain that persisted until 2 weeks postpartum and was completely undetectable by 4 weeks postpartum (27). Our results show a similar temporal expression pattern for the IA-1 transgene construct as compared with the Northern blot analysis for endogenous IA-1 expression. This observation is consistent with the IA-1 promoter region directing both the appropriate spatial and temporal expression of the IA-1 gene and further supports a role of IA-1 during early embryonic brain development.
Interestingly, homology searches of other genomes have revealed that IA-1 belongs to a highly conserved group of zinc finger-containing proteins that show the highest homology in the first two zinc fingers (30). Members of this group include two novel proteins, Nerfin-1 and Nerfin-2 (for nervous finger-1 and -2) from Drosophila and a C. elegans protein Egl-46 that are expressed in the developing nervous system (30,31). Nerfin-1 expression was detected during early central nervous system development in neuroblasts prior to lineage formation and was not expressed in neurons and glia. In contrast, Nerfin-2 is expressed in only a subset of brain neurons. Egl-46 represses transcription of touch cell characteristics in FLP cells. Egl-46 mutants show a serotonin-sensitive, imipramineresistant egg-laying defect, and the mutant animals have multiple hermaphrodite-specific neuron defects, including abnormalities in cell migration, axonal outgrowth, and serotonin production (32,33). Based upon the neuronal restricted expression of these genes, it is likely that they may have conserved function. The highest degree of homology (90%) in these four proteins resides in zinc finger 2 (31). Interestingly, we have shown that zinc fingers 2 and 3 of the IA-1 protein are critical for the DNA binding activity of IA-1, and the NeuroD1 transcription factor was identified as a potential target gene for IA-1 (20). Since NeuroD1 was originally isolated based upon its ability to induce neuronal differentiation when ectopically expressed in Xenopus oocytes (13), it further supports a functional role of the IA-1 gene in neuronal differentiation.
The expression profile seen in the transgenic mouse established that the 1.7-kb region of the IA-1 promoter directs the FIG. 8. Northern blot analysis of neuroendocrine tumor cell lines. Twenty micrograms of total RNA from Daoy, D283MED, U87MG, WERI-Rb-1, Y79, and ␤TC-1 cell lines were resolved on a 1% agarose/formaldehyde gel and transferred to a nitrocellulose membrane. The samples were run as two separate sets on the same gel. The membrane was divided in half, and the left half was hybridized with a mixture of human and mouse IA-1 probes (A) and the right half was hybridized with a hamster NeuroD1/␤2 probe (B). The ethidium bromide-stained gel was shown below the blot for comparison of loading differences. The IA-1 gene expression in the neuroendocrine cell lines overlaps with the NeuroD1 message completely. appropriate spatial and temporal expression of the IA-1 gene in brain. Using transient transfection studies, we demonstrated that a region between Ϫ426 and Ϫ65 bp exhibited strong promoter activity among a series of reporter constructs made from 5Ј-end deletions of the 1.7-kb promoter in ␤-TC-1 and Y79 cells. Although the relative promoter activity was significantly higher in the Y79 cells versus the ␤-TC-1, it is likely that human IA-1 promoter is more active in human Y79 than in mouse ␤-TC-1 cells. Northern blot analysis revealed that IA-1 message is stronger in ␤-TC-1 than in Y79 cells (Fig. 8). Computer analysis of the Ϫ426/ϩ40 bp region revealed the presence of three E-box motifs. E-box elements are bound by members of the bHLH transcription factor family. Belonging to this diverse family of transcription factors, there are both tissue-restricted and ubiquitously expressed members. A growing number of neurogenic bHLH factors have been detected in neural tissues and shown to be critical in their formation. The expression of two of these factors, neurogenin 3 (Ngn3) and NeuroD1, has also been shown in the developing endocrine pancreas and nervous system (14,34). NeuroD1 expression lies downstream of Ngn3 expression due to the lack of NeuroD1 expression in Ngn3 knockout mice and the identification of two Ngn3 binding sites in the NeuroD1 promoter responsible for activation of the NeuroD1 gene (35). Close analysis of the three E-box motifs in the IA-1 promoter showed that E3 was homologous to the NeuroD1-responsive E-box motif in the insulin, glucagon, and secretin gene promoters (10,(15)(16)(17). Individual deletion of the three E-box elements revealed that E3 was critical for ϳ65% of basal IA-1 promoter activity in ␤TC-1 cells. The E3-box resembles a NeuroD1 binding site and was activated by ϳ40-fold with the addition of both E47 and NeuroD1 expression plasmids. Finally, we showed binding of in vitro translated Neu-roD1 protein to the E3-box in the IA-1 promoter using EMSA analysis. Therefore, we have demonstrated that the neuroendocrine-specific bHLH transcription factor NeuroD1 is involved in the regulation of the IA-1 gene. This has functional significance in tissues that co-express both IA-1 and NeuroD1 in the development of the neuronal tissues.
Northern blot analysis on various neuroendocrine tumor cell lines revealed the expression of IA-1 and NeuroD1 overlaps in all the cell types tested. ␤TC-1, D283MED, WERI-Rb-1, and Y79 expressed detectable levels of both IA-1 and NeuroD1 message, whereas Daoy and U87MG cells were negative for their expression. This strong correlation between the expression of IA-1 and NeuroD1 in neuroendocrine cells supports a role for NeuroD1 in the regulation of IA-1 expression.
In conclusion, we report that the IA-1 promoter directs the appropriate spatial and temporal expression of the IA-1 gene and that one of the major cis-acting regulatory elements is bound by the E47-NeuroD1 heterodimer. This establishes a dynamic relationship between the tissue-restricted bHLH transcription factor NeuroD1 and the neuroendocrine-specific transcriptional repressor protein, IA-1. The fact that IA-1 is a target gene for NeuroD1 regulation and vice versa suggests that the IA-1 gene may play a pivotal role during early neuronal development by modulating NeuroD1 gene expression.