Critical Role of the Transcriptional Repressor Neuron-restrictive Silencer Factor in the Specific Control of Connexin36 in Insulin-producing Cell Lines*

Connexin36 (Cx36) is specifically expressed in neurons and in pancreatic β-cells. Cx36 functions as a critical regulator of insulin secretion and content in β-cells. In order to identify the molecular mechanisms that control the β-cell expression of Cx36, we initiated the characterization of the human 5′ regulatory region of the CX36 gene. A 2043-bp fragment of the human CX36 promoter was identified from a human BAC library and fused to a luciferase reporter gene. This promoter region was sufficient to confer specific expression to the reporter gene in insulin-secreting cell lines. Within this 5′ regulatory region, a putative neuron-restrictive silencer element conserved between rodent and human species was recognized and binds the neuron-restrictive silencing factor (NRSF/REST). This factor is not expressed in insulin-secreting cells and neurons; it functions as a potent repressor through the recruitment of histone deacetylase to the promoter of neuronal genes. The NRSF-mediated repression of Cx36 in HeLa cells was abolished by trichostatin A, confirming the functional importance of histone deacetylase activity. Ectopic expression, by viral gene transfer, of NRSF/REST in different insulin-secreting β-cell lines induced a marked reduction in Cx36 mRNA and protein content. Moreover, mutations in the Cx36 neuron-restrictive silencer element relieved the low transcriptional activity of the human CX36 promoter observed in HeLa cells and in INS-1 cells expressing NRSF/REST. The data showed that cx36 gene expression in insulin-producing β-cell lines is strictly controlled by the transcriptional repressor NRSF/REST indicating that Cx36 participates to the neuronal phenotype of the pancreatic β-cells.

Connexin36 (Cx36) is specifically expressed in neurons and in pancreatic ␤-cells. Cx36 functions as a critical regulator of insulin secretion and content in ␤-cells. In order to identify the molecular mechanisms that control the ␤-cell expression of Cx36, we initiated the characterization of the human 5 regulatory region of the CX36 gene. A 2043-bp fragment of the human CX36 promoter was identified from a human BAC library and fused to a luciferase reporter gene. This promoter region was sufficient to confer specific expression to the reporter gene in insulinsecreting cell lines. Within this 5 regulatory region, a putative neuron-restrictive silencer element conserved between rodent and human species was recognized and binds the neuron-restrictive silencing factor (NRSF/ REST). This factor is not expressed in insulin-secreting cells and neurons; it functions as a potent repressor through the recruitment of histone deacetylase to the promoter of neuronal genes. The NRSF-mediated repression of Cx36 in HeLa cells was abolished by trichostatin A, confirming the functional importance of histone deacetylase activity. Ectopic expression, by viral gene transfer, of NRSF/REST in different insulin-secreting ␤-cell lines induced a marked reduction in Cx36 mRNA and protein content. Moreover, mutations in the Cx36 neuron-restrictive silencer element relieved the low transcriptional activity of the human CX36 promoter observed in HeLa cells and in INS-1 cells expressing NRSF/REST. The data showed that cx36 gene expression in insulin-producing ␤-cell lines is strictly controlled by the transcriptional repressor NRSF/REST indicating that Cx36 participates to the neuronal phenotype of the pancreatic ␤-cells.
Adjacent cells share ions, second messengers, small metabolites, and signaling molecules with a weight as high as 1000 Da (1) through intercellular channels that form gap junctions. This type of intercellular communication permits coordinated cellular activity and allows cells to review the functional state of their neighbors, a critical feature for the homeostasis of multicellular organisms (1)(2)(3)(4). Intercellular channels result from the association of two half-channels named connexons, which are contributed to separately by each of two adjacent cells. Each connexon is an assembly of membrane proteins named connexins, which are encoded by a gene family consisting of at least 20 members (5,6).
It has been reported that Cx36 1 is the predominantly expressed connexin in rodent neuronal cells (7)(8)(9). Immunogold labeling using freeze-fracture replicas have shown that Cx36 was found only in neurons (10) and indicate that the specific expression of connexins by the different cell types of the nervous system may define separate pathways for neuronal versus glial gap junction communication (10 -12). We recently found that Cx36 is the predominant if not the sole connexin isoform expressed in insulin-producing ␤-cells of rat and mouse pancreatic islets (13,14). Recently, Cx36 content was modified in insulin-secreting cells by using an adenoviral gene transfer approach (15) or a lentivirus-mediated transduction of cx36 cDNA (16) in order to evaluate the contribution of Cx36 in the control of insulin secretion and content. These experiments have demonstrated that tight levels of Cx36 are crucial for proper function of insulin-producing cells. Moreover, stably transfected insulin-secreting MIN6 cells with a cx36 antisense cDNA impaired the synchronization of glucose-induced Ca( 2ϩ ) oscillations and insulin secretion in response to glucose (17). These data provided evidence that proper storage of insulin as well as the regulation of insulin release required that the levels of Cx36 be maintained within native values.
The objective of the present study was to characterize the mechanisms of cell-specific expression of Cx36. The presence of several gene transcripts in pancreatic ␤-cells and neuronal cells was correlated with the absence in these cell types of a transcription factor named NRSF/REST (neuron-restrictive silencer factor/repressor element silencing transcription factor) (18). NRSF/REST binds to a 21-bp cis element named neuronal restrictive silencer element (NRSE). Through this NRSE, NRSF/REST has been identified as a negative regulatory system that silences neuronal gene expression in non-neuronal tissues, neuronal precursor cells, and certain differentiated neurons (19 -27). The repression effect induced by NRSF/REST required an interaction of NRSF/REST with the corepressor mSin3 and histone deacetylase I (28), which induces hypoacetylation of histone (29,30). We have shown recently that the expression of the JNK-interacting protein-1 (IB1/JIP-1) gene is repressed by this mechanism in cells expressing REST (31). Many other genes specifically expressed in ␤-cells and neuronal cells, such as type II sodium channel and the rat NMDA receptor, were shown to be controlled by REST through the NRSE (18,31,32).
Here we identified a conserved NRSE sequence in the promoter of human and rodent cx36 genes, and we demonstrated that NRSF/REST specifically binds to this human NRSE. Forced expression of REST in the nucleus of pancreatic ␤-cell lines decreases the cx36 gene expression. After linking up-stream to a luciferase gene reporter, the human Cx36 transcriptional activity was shown to be undetectable in RESTexpressing cells, whereas it was highly detectable in insulinsecreting cells. Ectopic expression of REST decreased by 50% the transcriptional activity of CX36 promoter in insulin-secreting cell lines. Introduction of mutated NRSEs into the human CX36 promoter both abolished the silencing effect observed in INS-1-expressing REST and relieved the low transcriptional activity detected in HeLa cells expressing REST. This transcriptional activity was relieved and the cx36 gene expression became detectable in trichostatin-treated HeLa cells expressing REST, indicating that the cx36 gene silencing by NRSF/ REST required histone deacetylase activity. Finally, these observations showed that the transcriptional repressor NRSF/ REST controls cx36 gene expression in pancreatic ␤-cell lines.  Conservation of NRSEs of human, mouse, and rat Cx36 NRSE. The sequences bordering the NRSEs are less conserved than the NRSE between the human and rodent Cx36 5Ј-untranslated region. Note that the CGGAC central region of the NRSE consensus is 100% conserved between the three species. Cx36 mut-1 or Cx36 mut-2, mutants Cx36; IB1/JIP-1, scaffold protein termed JNK-interacting protein 1 (JIP-1) or Islet-Brain 1 (IB1); GABA, ␥-aminobutyric acid; Hsynapto, human synaptophysin; Rat nicotinic, rat nicotinic acetylcholine receptor ␣7; Rat type II Na, rat type II sodium channel.
Screening of a BAC Library of Human Genomic DNA and Plasmid Construction-A human genomic library in the pBeloBAC11 vector (Genome Systems, St. Louis, MO) was screened by PCR using specific rat Cx36 primers (forward, 5Ј-AACGCCGCTACTCTACAGTCTTCC-3Ј; 5Ј-GATGCCTTCCTGCCTTCTGAGCTT-3Ј). The identity of a single positive clone was first confirmed by Southern hybridization with the 269-bp rat Cx36 DNA probe (35). Four kilobases of the 5Ј-flanking sequence were sequenced. A 2043-bp fragment was PCR-amplified with primers 5Ј-GCGGTACCTAGGGGATCACCATTCCATCTG-3Ј (sense) and 5Ј-GCAGATCTCGCTGTGCATCCGGAGGCAGCAGACAA-3Ј (antisense), and a 621-bp fragment was amplified using a specific sense primer, 5Ј-GCGGTACCCCCCTCGGGCTCCGACGC-3Ј, and subcloned in pGL3 basic (Cx36LUC-1 and Cx36LUC-2, respectively) (Promega). Plasmid constructs were verified by DNA sequencing. The fragments were subcloned at KpnI/BglII sites into the promoterless expression vector pGL3 basic (Promega). Mutations in the NRSE sequence of the human CX36 promoter were generated by PCR site-directed mutagenesis using high fidelity Pfu DNA polymerase according to the manufacturer's protocol (QuickChange, Stratagene, La Jolla, CA). In vitro mutagenesis was carried out on a human CX36 promoter fragment of 621 bp (Cx36LUC-2) from double-stranded 5.5-kb plasmid DNA using oligonucleotide primers, each complementary to opposite strands of the vector (two forward primers, 5Ј-CCAGACCCGAACCCCTTACGGCGC-CCGAGG-3Ј and 5Ј-CCAGACCCGAACCCTTTTCGGCGCCCGAGG-3Ј; and two reverse primers, 5Ј-CCTCGGGCGCCGTAAGGGGTTC-GGGTCTGG-3Ј and 5Ј-CCTCGGGCGCCGAAAAGGGTTCGGGTCT-GG-3Ј. The mutated nucleotides are underlined. Both constructs (m1Cx36LUC-2 and m2Cx36LUC-2) were verified by DNA sequencing to confirm the integrity and the presence of the desired mutations in the constructs. The sequence of the fragment of the human CX36 promoter (bases 1-2046) was assigned GenBank TM accession number AY341000.
RACE-For 5Ј-rapid amplification of cDNA ends (RACE), 3-4 g of total RNA from INS-1 or ␤TC-3 were reverse-transcribed by avian myeloblastosis virus-reverse transcriptase according to the 5Ј/3Ј-RACE kit procedure (Roche Applied Science) by using an antisense Cx36specific primer, 5Ј-AAGATGGTCCATTCCCCCAT-3Ј. The purified cDNA was then subjected to a terminal transferase tailing reaction, and the deoxyribosyladenosine-tailed cDNA was then used as a template of PCR using an oligo(dT) anchor primer provided by the 5Ј/3Ј-RACE kit and an antisense Cx36-specific primer (5Ј-ACGTTCTTGGTCCTG-GCCC-3Ј). A second nested-PCR round reaction was then performed using the PCR anchor primer provided by the race kit and a third antisense Cx36-specific primer 5Ј-ATTGCCCCCAATTAAAG-3Ј. The RACE products were subcloned into the T-Easy vector (Promega) according to the manufacturer's protocol, and individual clones were sequenced.
Transient Transfection and Luciferase Assays-Cells were transfected using the liposome-mediated Transfast Reagent (Promega) at a DNA/lipid ratio of 1:1 in RPMI1640 medium without fetal calf serum. Cells were plated in 12-well plates and grown to 60% confluence. On the day of the transfection, 1 g of DNA was mixed with 3.0 l of Transfast reagent and incubated at room temperature. After 15 min of incubation, plated cells were overlaid with DNA/Transfast mixture and incubated at 37°C for 1 h. To each well, 2 ml of RPMI1640 with serum were added. Cells were incubated for 48 h and harvested with 110 l of the passive lysis buffer (Promega). Luciferase activities (from Cx36LUC and normalization with Renilla pRLCMV [pRLCMVrenilla] vectors) were measured with 50 l of protein extract solution by using the dualluciferase reporter assay system (Promega) and a Turner TD-20/20 luminometer. All experiments were repeated three times in triplicate.
Electrophoretic Mobility Shift Assay (EMSA)-Nuclear extracts from cells were prepared according to the kit procedure (Activ Motif, CA). The protein concentration was measured using a Bio-Rad protein measurement kit. For EMSA, the double strand DNA probe corresponding to the Cx36 NRSE sequence sense (5Ј-GGATCCAGACCCGAACCCCGG-ACGGCGCCCGA-3Ј) and antisense (5Ј-CCTCGGGCGCCGTCCGGGG-TTCGGGTCTGGAT-3Ј) oligonucleotides was labeled with [␣-32 P]dCTP by the Klenow fragment of DNA polymerase I (Roche Applied Science). The labeled probe was incubated with 5 g of the nuclear protein, and 2 g of poly(dI-dC) in a total volume of 20 l containing 10% glycerol, 20 mM Hepes-NaOH, pH 6, 2.5 mM MgCl 2 , 5 mM dithiothreitol, and 50 mM KCl. After incubation at 4°C for 30 min, the reaction mixtures were resolved on a 4% non-denaturing PAGE in 0.25ϫ TBE buffer (1ϫ TBE: 89 mM Tris, 89 mM boric acid, 2 mM EDTA, pH 8.0) at 4°C for 2 h, and the products were detected by autoradiography. For the supershift experiments, affinity-purified REST polyclonal antibodies were added to the extracts, and the mixture was incubated for 10 min at room temperature prior to the addition of the labeled probe. Similar experiments were performed using nonspecific rabbit affinity-purified polyclonal antibodies. In the competition study, unlabeled specific double strand DNA competitors (30 -100-fold in excess) were added into the reaction mixtures prior to incubation. The regions comprising bp Ϫ1730 to ϩ313 (Cx36LUC-1) and bp Ϫ308 to ϩ313 (Cx36LUC-2) of the human CX36 promoter region drove high luciferase activity in the insulin-secreting INS-1 and ␤TC3 but not in unrelated HeLa and RAW cells. Luciferase activities were normalized using pRLCMVrenilla. RLU, relative light units. Columns represent mean Ϯ S.E. of three independent experiments performed in triplicate. ***, p Ͻ 0.001.

FIG. 4. Specific binding activity of NRSF/REST to the human
Cx36 NRSE. EMSA with 32 P-labeled Cx36 NRSE using nuclear extracts (NE) from HeLa cells. A slow-migrating complex (arrowhead) was detected and compared with free-probe migration. NRSF/REST shift was competed with the incubation of polyclonal antibodies against REST but not with nonspecific antibody (NS). The DNA binding activity with human Cx36 NRSE was abolished by adding a 30-or 100-fold molar excess of unlabeled wild-type NRSE.

Generation of Recombinant Adenoviruses and Cell
Infection-To express REST in insulin-secreting cells, we generated recombinant adenoviruses (36 -38) comprising the complete cDNA of human REST (31) in the sense orientation (Ad-sREST). Adenoviruses expressing the green fluorescent protein (GFP) (Ad-GFP) were used as control. The recombinant adenoviruses allowed for expression of REST and GFP were under the control of the immediate, early cytomegalovirus promoter (CMV). The cDNAs were inserted into plasmid pXC15 (39), and the adenoviruses were then generated by cotransfecting 293 cells with plasmids pJM17 and the pXC15 containing human REST cDNA, using a calcium phosphate procedure. Viruses were further plaque-purified three times on HER911 cells (IntroGene, Leiden, The Netherlands). Stocks were purified by two successive CsCl centrifugations, after which the virus band (1.5 ml) was collected and dialyzed at 4°C against three changes (at least 200 volumes each) of 10 mM Tris-HCl buffer, pH 8.0, in a Slide-A-Lyzer (0.5-3.0-ml capacity), ␥-irradiated 10K dialysis cassette (Pierce). INS-1 or ␤TC3 cells were seeded in 12-well dishes or 6-cm dishes (Falcon) and cultured for 48 h. For infection, cells were incubated for 60 min with different dilutions (multiplicity of infection (m.o.i.)) of a stock of the recombinant adenoviruses Ad-sREST or Ad-GFP and then cultured in RPMI 1640 medium for 24 h before any experiment.
Antibodies-Cx36 peptide corresponding to amino acids 289 -303 (peptide B, AKRKSVYEIRNKDLP) (13,15,17) was conjugated to keyhole limpet hemocyanin and used to immunize rabbits (Eurogentec; Parc Scientific du Sart Tilman, Herstal, Belgium). Plasmid pQE containing the His 6 tag was used to direct the synthesis by Escherichia coli of a polypeptide comprising amino acids 7-355 of the human REST. An overnight culture of E. coli that had been transformed with the recombinant plasmid was diluted 1:10 in fresh medium and grown for 1 h at 37°C before addition of 1 mM isopropyl-␤-D-thiogalactopyranoside. Four hours later, cells were pelleted and lysed on ice by sonication. The recombinant protein was incubated in the presence of nickel-nitrilotriacetic acid to purify the His 6 -tagged REST fusion protein. Site-directed antibodies were affinitypurified by passing the sera diluted in PBS through HiTrap affinity columns (HiTrap NHS-activated Sepharose; Amersham Biosciences) that had been coupled with the Cx36 COOH-terminal tail peptide or recombinant REST protein. The columns were washed sequentially with PBS, 0.5% PBS/Tween 20, and PBS. Antibodies solutions were eluted with 100 mmol/liter glycine, pH 2.5, in a 1.5 mol/liter Tris-HCl buffer, pH 8.8, to reach neutrality (13,40).
Immunofluorescence-For immunofluorescence labeling, cells were grown on glass coverslips, fixed 3 min in Ϫ20°C acetone, air-dried, rinsed in PBS, and incubated in a buffer containing 2% bovine serum albumin and 0.1% Triton. Cells were then exposed for 20 h at room temperature in the presence of the polyclonal rabbit antibodies against human REST, diluted 1:200. After repeated washing, cells were incubated a second time for 1 h at room temperature, using fluoresceinconjugated antibodies against rabbit IgGs, diluted 1:500. After additional rinsing, cells were counterstained with Evans blue, coverslipped, and viewed on a microscope (Leica, Switzerland) fitted with filters for fluorescein detection. In these experiments, negative controls included INS-1 or ␤TC-3 cells.
For reverse transcription PCR, total RNA was treated for 30 min in the presence of DNase I (DNA-free kit, Ambion, Cambridge, UK). One-g aliquots of DNase-treated RNA were reverse-transcribed using Im-Prom-II reverse transcription system (Promega, Switzerland), as described by the manufacturer. One-half of the reverse-transcribed products was used for PCR in the presence of 20 ng of sense and antisense specific primers using recombinant Taq DNA polymerase (Invitrogen). The products obtained after 30 PCR cycles performed in a PCR machine (Biometra, Switzerland) with the different primers were analyzed by agarose gel. Negative controls included amplification of INS-1 samples that had not been reverse-transcribed. The following primers were used, a sense primer 5Ј-CACAGCGATGGGGGAATGGA-3Ј and an antisense primer 5Ј-TGCCCTTTCACACATAGGCA-3Ј to detect Cx36 mRNA. These primers amplified a fragment of 980 bp. The amplified DNA was visualized after ethidium bromide staining on agarose gel.
Western Blotting-INS-1 cells were lysed by sonication in a buffer containing 5% SDS supplemented with 5 mM EDTA, and the DC protein assay reagent kit (Bio-Rad) was used to determine protein content. Aliquots of cell lysates or nuclear extracts from INS-1 cells were heated at 95°C in loading buffer and were fractionated by electrophoresis in a 12.5% polyacrylamide gel and immunoblotted onto Immobilon polyvinylidene difluoride membranes (Millipore) overnight at a constant voltage of 20 V. Membranes were incubated for 1 h at room temperature in PBS containing 5% milk and 0.1% Tween 20 (blocking buffer) and then incubated overnight at 4°C with antibodies directed against rat Cx36 in blocking buffer. Antibodies directed against ␤-tubulin (Sigma) were incubated overnight at 4°C in blocking buffer. Specific antigen-antibody complexes were detected with the horseradish peroxidase Western blot detection system (Amersham Biosciences).
Statistical Analysis-Densitometric analysis of immunolabeled proteins (Western blots) and mRNA signals (autoradiograms) was performed using an Amersham Biosciences scanner that integrates areas and corrects for background. Data were expressed as mean Ϯ S.E. Differences between means were assessed by Student's t test and/or analysis of variance t test. Statistical significance was defined at a value of p Ͻ 0.05, p Ͻ 0.01, and p Ͻ 0.001.

RESULTS
Identification of the NRSE within the CX36 Promoter Region-A 2043-bp fragment of the human CX36 promoter was PCR-amplified using specific human Cx36 primers. The nucleic acid sequence of this region is shown in Fig. 1. The transcriptional start site in insulin-secreting cells was determined using a 5Ј-RACE reaction with several primers. For this analysis, we used as template total RNA isolated from INS-1 and ␤TC3 cells, and the most prominent 5Ј nucleotide identified the start site (bp ϩ1 of the sequence shown in Fig. 1) located Ϫ313 bp from the 1st base of the translational start codon ATG in INS-1 and ␤TC3 cells. Sequence comparison between human, rat, and mouse cx36 promoter region indicates high homology in the region of the start site between the three species (data not shown). Different putative sequences for the binding of a variety of transcription factors were identified within the 2043 bp by computer analysis (bioinformatics.weizmann.ac.il/transfac/), in particular an NRSE. As shown in Fig. 2, the 21-bp NRSE identified in human (bp 42-62) is highly conserved in mouse and rat (bp 40 -60) sequences as well as with other known NRSE sequences and is homologous to the consensus NRSE sequence (23). Moreover, these Cx36 NRSE sequences are highly homologous compared with adjacent cx36 promoter sequences suggesting a functional role of these elements for REST activity.
The Promoter Region of Cx36 Drives High Transcriptional Activity in Insulin-secreting ␤-Cell Lines but Not in Unrelated Cell Lines-We evaluated the transcriptional activity of two fragments comprising bp Ϫ1730 to ϩ313 (Cx36LUC-1) and Ϫ308 to ϩ 313 (Cx36LUC-2), respectively, of the promoter region of the human CX36 gene in insulin-secreting ␤-cells and in unrelated cell lines (Fig. 3). As shown in Fig. 3, both constructs drove high transcriptional activity in INS-1 and ␤TC3 cells, whereas no significant reporter luciferase activity was detected in HeLa and Raw cells compared with the promoterless pGL3 basic vector (normalized to 1). These data indicate that important cis regulatory elements are present within the human CX36 promoter to confer the ␤-cell-specific expression of the reporter gene.
REST/NRSF Interacts with Human NRSE of CX36 Promoter-To investigate whether the NRSF/REST was able to bind to the newly identified human Cx36 NRSE, we performed an EMSA using nuclear extracts prepared from HeLa cells expressing endogenous NRSF/REST and the putative NRSE of the human CX36 promoter as the labeled probe (Fig. 4). To determine the specificity of this binding activity, competition experiments were performed using a 100-or 30-fold molar excess of unlabeled NRSE. Two DNA binding complexes were observed; the fast migrating complex reflects nonspecific binding as this complex is only slightly blocked in the presence of REST antibodies, whereas the upper complex disappeared. This DNA-binding complex was not detected using nuclear extracts obtained from ␤TC3 (not shown). These data indicate that NRSF/REST is able to bind to the human Cx36 NRSE.

Ectopic Expression of REST Decreased Cx36 Expression in
Pancreatic ␤-Cell Lines-NRSF was detected by immunofluorescence in wild-type HeLa cells using specific NRSF/REST A, Cx36LUC-2 containing a region of the CX36 promoter comprising bp Ϫ308 to ϩ313 and mutated constructs (m1Cx36LUC-2 and m2Cx36LUC-2) were transiently transfected into REST-expressing HeLa cells. The luciferase activity was relieved with the mutants constructs. Columns represent mean Ϯ S.E. of three independent experiments performed in triplicate; ***, p Ͻ 0.001. RLU, relative light units. B, the CX36 promoter containing wild-type or mutated NRSE, Cx36LUC-2 and m1Cx36LUC-2 and m2Cx36LUC-2, respectively, were transfected in the INS-1 cells infected or not with Ad-sREST. The luciferase activity of Cx36LUC was decreased by 50% in REST-expressing cells, whereas REST expression has no effect on the Cx36LUC-2 mutants activities. Columns represent mean Ϯ S.E. of three independent experiments performed in triplicate; **, p Ͻ 0.01. antibodies but not in INS-1 cells (Fig. 5, A and B). In contrast, the protein was detected in the nucleus of INS-1 or ␤TC3 cells infected with Ad-sREST (Fig. 5, C and D). Western blots of nuclear extracts, using the same REST/NRSF antibodies, con-  (Fig. 5B). Quantitative assessments for Cx36 mRNA showed that levels of these Cx36 transcripts were markedly decreased (70%) (Fig. 6C). As positive control, we confirmed the reduction of IB1/JIP-1 (31) by the heterologous expression of REST (Fig. 6C). Quantification of multiple blots showed that Cx36 mRNA levels were also markedly decreased (50 -80%) in different ␤-cell lines such as ␤TC3 and Min-6 cells (Fig. 7, A and B). These results demonstrate the functional role of REST in the control of the cell-specific expression of IB1/ JIP-1 and cx36 genes.
Transcriptional Activation of the CX36 Promoter Mutated in the NRSE in INS-1 and HeLa Cells-By using site-directed mutagenesis, the NRSE of the human CX36 promoter was modified to the mutated NRSEs (substitution of the two, respectively, four nucleotides) described in Fig. 2. In transient transfection (Fig. 8A) the reporter gene activity mediated by the mutated NRSEs in the CX36 promoter (m1Cx36LUC-2 and m2Cx36LUC-2) was relieved compared with Cx36LUC-2 activity in HeLa cells. This implies that the NRSE contributes to the repression of CX36 promoter activity in non-␤ pancreatic HeLa cells. To assess whether NRSF/REST could repress human CX36 promoter activity, the insulin-secreting INS-1 cells were infected with Ad-sREST and then transfected with the Cx36LUC-2 or mutated NRSEs constructs m1Cx36LUC-2 and m2Cx36LUC-2 (43). REST significantly decreased the Cx36LUC relative activity by 50%, whereas it has no effect on the mutated constructs. These data demonstrated that forced expression of REST in INS-1 was sufficient to repress, through the NRSE, the human CX36 promoter activity in the non-REST-expressing INS-1 cells (Fig. 8B).
Induction of Endogenous CX36 Transcription in TSA-treated Cells-We assessed whether repression of the transcription of the endogenous cx36 gene could be relieved in REST-expressing cells after trichostatin A (TSA) treatment (Fig. 9). Total RNA from HeLa, INS-1, and ␤TC3 cells treated with the vehicle dimethyl sulfoxide or with TSA were analyzed by RT-PCR for expression of the Cx36 transcript using a Cx36-specific primers set. The reaction yielded a PCR fragment detectable in insulin-secreting cells and only in NRSF/REST expressing cells treated with TSA. This observation indicates that deacetylase activity is required for the repression of endogenous cx36 gene transcription.
The present study provides new insight into the transcriptional mechanisms that control the cell-specific expression of the cx36 gene (9,10,12). Each of the gap junction proteins is characterized by a cell type-specific but overlapping expression pattern; cell-specific expression of connexin genes can take place at the transcriptional levels (47). The human Cx43 promoter contains regulatory sequences including the putative responsive element as SP1, E box, estrogen responsive element half-sites, AP-2 cAMP-responsive element-binding protein, and Ets-1 sites, which can serve as transcription factor targets (48,49). In rat Cx43, a putative regulatory element-binding thyroid hormone receptor has been described as well as elements responsive to parathyroid hormone (50). Recently, it has been shown that the transcriptional up-regulation of Cx43 by Ras-Raf-MAPK is mediated by the interaction of a novel Cx43 promoter element with a protein complex that contains both HSP90 and c-Myc (51). Overexpression of Nkx2.5, a member of the NK2 homeodomain proteins was associated with a decrease in Cx43 expression, suggesting that Nkx2.5 may function as a transcriptional repressor (52). SP1 has also been found involved in the transcriptional regulation of Cx26, in rat mammary gland and uterus, as well as in the activation of Cx40 and Cx32 proximal promoter (53)(54)(55). Moreover, it has been demonstrate that Sox10, a transcription modulator, activates Cx32 expression (56).
We have identified a fragment of the human CX36 promoter and characterized its initiation start site in insulin-secreting cells. This fragment drives high transcriptional activity in pancreatic ␤-cell lines but not in unrelated cell lines. We next identified an NRSE highly conserved between rodent and human. This cis element binds the transcription factor NRSF/ REST prepared from HeLa cells. REST was shown to silence CX36 promoter activity in non-␤-cell lines. NRSF/REST contains two transcriptional repression domains on its NH 2 and COOH termini that recruit histone deacetylases (29,(57)(58)(59). Likewise, repression mediated by NRSF was shown to be sensitive to inhibitors of histone deacetylases such as TSA. Moreover, forced expression of REST in INS-1, ␤TC3, and Min-6 was associated with a marked decrease in Cx36 mRNA expression as well as in the mRNA expression of the c-Jun NH 2 -terminal interacting protein-1 (IB1/JIP-1), known to be regulated by NRSF (31).
In pancreatic ␤-cells, the scaffold protein termed JNK-interacting protein 1 (JIP-1) or islet-brain 1 (IB1) (41, 60) is required for proper JNK activation and may play a critical role in the pro-apoptotic program observed during diabetes (38,60,61). We demonstrated recently (62) that IB1-JIP-1 content in insulin-producing ␤-cells is critical for ␤-cell function (38) and that an IB1/JIP-1 promoter variant was associated with Alzheimer's disease. We have shown recently that the absence of REST allows the cell-specific expression of the human IB1 gene (31). This study further demonstrated that endogenous IB1 mRNA expression may be efficiently inhibited in ␤-cells expressing REST, validating the use of REST gene transfer approach to inhibit genes regulated by NRSF/REST and further indicating the importance of REST to repress IB1 gene expression in non-␤ or non-neuronal cells. ␤-Cells and neuronal cells share similarities as both cell types are electrically excitable and may respond to hormonal stimuli and glucose by depolarization and exocytosis in a process similar to neurotransmitter release from synaptic vesicles (18,63). Among genes that are linked to a direct pancreatic ␤-cell function, Cx36 (13-17, 38, 41, 60, 61) and IB1/JIP-1 (38,61) are the first found to be regulated by NRSF/REST (31).
It has been shown that NRSF/REST represses a subset of neuronal genes such as SCG10, synapsin I, the brain type II voltage-dependent sodium channel, the dopamine ␤-hydroxylase, synaptophysin, and the IB1/JIP-1 genes (18,31,44,64) which are also detected in pancreatic ␤-cells. Here we confirm that NRSF/REST is a potent repressor that contributes to the preferential expression of Cx36. The endogenous Cx36 transcriptional activity was relieved in trichostatin-treated cells, and the mechanism would be that NRSF/REST silences Cx36 transcription through histone acetylation, indicating that the expression of cx36 gene is controlled by alterations of the chromatin structure. REST-induced hypoacetylation around the NRSE may change the nucleosomal structure and therefore could have a direct effect on TFIID-RNA polymerase II holocomplex access to the CX36 promoter (29,65). Indeed, the NRSE at the CX36 promoter is located 41 bp downstream from the transcription start site, and so local changes to nucleosomal structure may affect the basal transcription complex.
In conclusion, the data provide evidence that NRSF/REST is a critical transcriptional factor that contributes to the cellspecific expression of the human and rodent cx36 genes and demonstrate that Cx36 participates to the neuronal phenotype of the pancreatic ␤-cells.