The repressor element silencing transcription factor (REST)-mediated transcriptional repression requires the inhibition of Sp1.

The terminal differentiation of neuronal and pancreatic beta-cells requires the specific expression of genes that are targets of an important transcriptional repressor named RE-1 silencing transcription factor (REST). The molecular mechanism by which these REST target genes are expressed only in neuronal and beta-cells and are repressed by REST in other tissues is a central issue in differentiation program of neuronal and beta-cells. Herein, we showed that the transcriptional factor Sp1 was required for expression of most REST target genes both in insulin-secreting cells and neuronal-like cells where REST is absent. Inhibition of REST in a non-beta and a non-neuronal cell model restored the transcriptional activity of Sp1. This activity was also restored by trichostatin A indicating the requirement of histone deacetylases for the REST-mediated silencing of Sp1. Conversely, exogenous introduction of REST blocked Sp1-mediated transcriptional activity. The REST inhibitory effect was mediated through its C-terminal repressor domain, which could interact with Sp1. Taken together, these data show that the inhibition of Sp1 by REST is required for the silencing of its target genes expression in non-neuronal and in non-beta-cells. We conclude that the interplay between REST and Sp1 determines the cell-specific expression of REST target genes.

pancreas (1, 6 -8). Target genes of REST include those encoding Nav1.2 sodium channel (4), SCG10 (9), synapsin I (2, 3, 10), the GluR2 glutamate receptor subunit (11), subunits of the muscarinic acetylcholine receptor (12), islet-brain-1/c-Jun Nterminal kinase-interacting protein-1 (IB1) (6,13), complexin I, and connexin 36 (14). These genes have been shown to be mandatory for neuronal and ␤-cell functions. Re-introduction of exogenous REST suppresses expression of its target genes, preventing cells from differentiating and secreting insulin in response to glucose in neuronal-like PC12 cells and in insulinsecreting cells, respectively (15,16). In vivo, expression of REST target genes is tightly controlled and is thought to be a marker for the terminal differentiation of neuronal cells. The homozygous disruption of REST in mice leads to a precocious expression of its target genes in neuronal progenitors and induces an early embryonic lethality (7). On the other hand, suppression of REST target genes by introducing REST in differentiating neurons of chicken causes axon pathfinding errors (17).
In non-neuronal and in non-␤-cells, REST silences the expression of its target genes by its two independently acting repressor domains (18). The N-terminal repressor domain of REST has been shown to recruit some co-repressors such as mSIN3 and histone deacetylases (HDACs) into the vicinity of the promoter. Histone deacetylation leads to a more compact chromatin that prevents accessibility of transcription factors. The C-terminal repressor domain (CTRD) of REST has been shown to interact with at least one factor, the transcriptional co-repressor CoREST that may serve as a platform protein for the recruitment of molecular machinery that imposes silencing across a chromosomal interval (19,20). CoREST is able to interact with HDAC1/2, indicating that the CTRD also contributes to deacetylation of histones (15). Some studies have revealed the existence of additional repression mechanisms mediated by REST. This has been postulated because HDAC inhibitors fail to derepress the CTRD repression of some REST target genes such as GluR2 receptor and SCG10 (21,22). In this study, we hypothesized that inactivation of transcriptional activators might be an additional mechanism by which REST silences the expression of its target genes. Elucidating such a mechanism could help in understanding the mechanism by which expression of most REST target genes is positively activated in neuronal and ␤-cells Herein, we showed that the Sp1 transcriptional activator controlled expression of most REST target genes in neuronallike PC12 cells and in insulin-secreting ␤TC3 cells where REST is absent. REST prevented Sp1-dependent activation in REST endogenously expressing HeLa cells. This result was reproduced in ␤TC3 cells transiently expressing REST. Deletion constructs followed by co-immunoprecipitation and transfection assays indicated that the CTRD of REST was required for interaction with Sp1 and inhibited its activity. Altogether, our data indicated that the inhibition of Sp1 is required to silence expression of REST target genes outside neuronal and ␤-cell types. We propose that the silencing of Sp1 by REST is a required event to determine the ␤-cell and the neuronal cell expression of REST target genes.

MATERIALS AND METHODS
Plasmid Construction-Mutation in the sp1 sequence of the MAPK8IP1 promoter (mIB1luc) was generated by PCR site-directed mutagenesis using high fidelity Pfu DNA polymerase according to the manufacturer's protocol (QuikChange TM , Stratagene). In vitro mutagenesis was carried out on full-length human MAPK8IP1 promoter linked to luciferase gene reporter (IB1luc) (6) using primers as follows: sense, 5Ј-TCCGGATAAGGGGCTGCGG-3Ј and antisense, 5Ј-CCGCAGCCCCT-TATCCGGA-3Ј. The mutated nucleotides are underlined. To create the plasmid encoding the full-length cDNA of human REST under the control of a CMV promoter, the fragment encompassing the coding region (amino acids 1-1098) of REST was amplified by PCR from RIP REST (6) and was subcloned into the pCDNA3 vector. The plasmids REST⌬C and REST⌬N encode for REST mutants amino acids 1-417 and 212-1098 of REST, which retain either the N-terminal or the C-terminal repression domain of REST. These different fragments of REST were generated by PCR using primers containing BamHI and XbaI restriction sites in the sense and antisense primers, respectively. The primer set used were as follows: REST⌬C sense, 5Ј-TGGCCGAATGGATCCATGGCCACC-3Ј and antisense, 5Ј-TGAGACTCTAGATTGTTAATTAGGACA-3Ј and REST⌬N sense, 5Ј-GGGAGATGGATCCAAGGGCCCC-3Ј and antisense, 5Ј-TTCAAGTCTAGATCATTACTCCTGCC-3Ј. The PCR products were then subcloned into pCDNA3 vector. Constructs were verified by DNA sequencing. The construct encoding the Sp1 transcription factor under a CMV promoter (pCMV-Sp1) was kindly provided by Dr. Robert Tjian (University of California, Berkeley, California).
Cell Lines, Transient Transfection, and Luciferase Assays-Human carcinoma HeLa cells, the mouse insulin-secreting ␤TC3 cells, and the rat pheochromocytoma PC12 cells that are derived from neural crestderived tissue were cultured as described previously (6). All cells were transiently transfected using Effectene transfection kit (Qiagen) according to manufacturer's protocol. Cells were then incubated for 48 h, and luciferase activities were measured by using the Dual-Luciferase TM reporter assay system (Promega).
Nuclear Protein Extract Preparation and Electromobility Shift Assays-Nuclear protein extracts and electromobility shift assays were performed as previously reported (6). Primers used as labeled probes were as follows: sp1 sense, 5Ј-TCCGGGGGCGGGGCTGCGG-3Ј and antisense, 5Ј-CCGCAGCCCCGCCCCCGGA-3; MLTF sense, 5Ј-TAG-GTGTAGGCCACGTGACCGGGTGTTC-3Ј and antisense, 5Ј-GAACAC-CCGGTCACGTGGCCTACACCTA-3Ј. Complementary sense and antisense oligonucleotides were hybridized and then filled in by the Klenow fragment of DNA polymerase I (Roche Diagnostics) in the presence of [␥ 32 P]deoxycytosine triphosphate (Amersham Biosciences). For binding, 5 g of nuclear proteins were preincubated on ice, in the presence or absence of an excess of unlabeled competitor DNA, for 10 min in 20 l of a solution containing 15 mM HEPES, pH 7.8, 50 mM NaCl, 1.5 mM MgCl 2 , 1 mM EDTA, 1 mM dithiothreitol, 10% glycerol, and 2 g of poly(dI-dC)⅐poly(dI-dC). Approximately 100 fmol of double strand-labeled oligonucleotide were mixed with nuclear proteins, and the mixture was incubated for 30 min at room temperature. For supershift assays, 5-10 g of nuclear proteins were preincubated on ice with or without SP1 or REST antibodies (Santa Cruz Biotechnology) for 30 min before the addition of the labeled probe.
Western Blotting Analysis and Immunoprecipitation-Nuclear proteins (50 g) from HeLa and ␤TC3 cells were prepared and subjected to Western blot as described (23). For co-immunoprecipitation experiments, 100 g of nuclear extracts were incubated with 0.5 g of the appropriate antibodies (Sp1 and green fluorescent protein (GFP) from Santa Cruz Biotechnology) for 16 h at 4°C in 500 l of immunoprecipitation buffer (20 mM Hepes-KOH, pH 7.9, 25% glycerol, 420 mM NaCl, 1.5 mM MgCl 2 , 0.2 mM EDTA, 0.5 mM dithiothreitol, and 0.2 mM phenylmethylsulfonyl fluoride). Protein A-agarose beads (25 l) were used to precipitate the associated protein. Samples were then separated by SDS-PAGE and detected by Western blot analysis using the indicated antibodies.

FIG. 1. Mithramycin A inhibits expression of REST target
genes. ␤TC3 (A) and PC12 (B) cells were preincubated with mithramycin A (0, 100, and 400 nM) for 16 h. Total RNAs were prepared and analyzed by reverse transcription PCR (25 cycles) for REST target genes expression (IB1, connexin 36, complexin 1 (CPLX1)) using specific primer pairs as described under "Material and Methods." Expression of the ␤-actin gene that is insensitive to mithramycin A (29) was used as negative control. JIP-1, c-Jun N-terminal kinase-interacting protein-1.

RESULTS
The Transcriptional Activator Sp1 Controls the Expression of REST Target Genes-A computer search was performed to identify the regulatory elements that could be responsible for the neuronal and ␤-cell-specific expression of REST target genes. G/C boxes sequences (GGGGGCGGGGC), consensus nucleotide-binding sites for the Sp1 transcription factor, were found in most REST target genes (Table I). This observation is consistent with the fact that the promoters of these genes lack identifiable TATA elements. Independent studies have confirmed the functional implication of Sp1 in expression of some of REST target genes. These include N-methyl-D-aspartate receptors, dynamin I, synaptophysin, and glutamate receptor ionotropic kainate-5 genes (24 -27). To assess the functionality of G/C box sequences on expression of other REST target genes, ␤TC3 cells and PC12 cells that contain a detectable expression of some REST target genes (6) were treated with mithramycin A, a DNA-binding drug that binds to G/C-specific regions of DNA including G/C box sequences. This has been already shown to prevent subsequent Sp1-binding activity and, therefore, to selectively inhibit the transcription of a number of Sp1-dependent genes (28). As shown in Fig. 1, the mRNA levels of REST target genes including MAPK8IP1 encoding IB1/c-Jun N-terminal kinase-interacting protein-1, complexin I, and connexin 36 were decreased by mithramycin A in a dose-dependent manner both in ␤TC3 and in PC12 cells. In accordance with other studies, mithramycin A did not modify mRNA levels of ␤-actin (29). Electromobility shift assay analyses confirmed the loss of Sp1-binding activity in ␤TC3 cells treated with mithramycin A (data not shown). These results indicate that Sp1 is required for expression of REST target genes in cells where REST is absent.
The Transcriptional Repressor REST Represses Sp1-dependent Activation-As mentioned above, the REST target genes are not expressed outside differentiated neuronal and ␤-cells that contain the transcriptional repressor REST (1,6). Sp1 is also expressed in these cells expressing REST (data not shown). We hypothesized that REST might prevent Sp1-dependent activation of REST target promoters. To verify this hypothesis, we took advantage of the previously described model of the MAPK8IP1 promoter. This promoter region has been previously reported to contain a functional REST-binding motif (NRSE) (6) and a sp1 motif was found from positions Ϫ202 to Ϫ191 upstream of the initiation site of the transcription ( Fig. 2A). To determine the functionality of this motif, we generated IB1luc constructs containing either wild type (IB1luc) or mutated (mIB1luc) sp1 motifs. Electromobility shift FIG. 2. A GC-box within the human MAPK8IP1 promoter is required for Sp1 activation. A, mutations were introduced in a sp1 motif located at the position from Ϫ231 to Ϫ191 bp from the transcription start site of the human MAPK8IP1 promoter (6). B, the wild type (IB1luc) or the mutated (mIB1luc) constructs were transiently transfected into ␤TC3 cells in the absence or in the presence of 0.05 g (ϩ) and of 0.1 g (ϩ ϩ) of vector encoding Sp1. As expected, the wild type MAPK8IP1 promoter (IB1luc) drove a high luciferase activity, which was markedly increased by overexpression of Sp1 in a dose-dependent manner. The Sp1 activation was abolished when the sp1 motif is mutated. C, mithramycin A (nM), a drug that binds to the G/C boxes sequence, inhibited the Sp1-mediated activation of the MAPK8IP1 promoter. Experiments were performed at least six times in triplicate. Results were expressed as mean Ϯ S.E. (**, p Ͻ 0.001).
assays confirmed the inability of the cold-mutated sp1 motif to compete for the sp1-binding pattern compared with the unlabeled wild type sequence in competition experiments (data not shown). IB1luc and mIB1luc were transiently transfected in ␤TC3 cells. As expected, IB1luc drove a high luciferase activity in ␤TC3 cells. Moreover, co-introduction of the vector encoding Sp1 (Sp1) increased the luciferase activity of the MAPK8IP1 promoter in a dose-dependent manner (Fig. 2B). This Sp1-dependent activation was abolished when the sp1 motif was mutated, confirming the direct effect of Sp1 on the MAPK8IP1 promoter activity. Similar results were obtained when cells were treated with mithramycin A, which acts as a competitor for the binding G/C box sequences (Fig. 2C). In the presence of mithramycin A, the MAPK8IP1 promoter activity was reduced by 50% and the Sp1-mediated activation was abolished. These experiments were confirmed in PC12 cells (data not shown).
The inactivation of REST in HeLa cells by using a REST dominant negative has been shown to partly restore the activity of IB1luc (6). To examine the implication of Sp1 in this phenomenon, HeLa cells were transiently co-transfected with IB1luc and the Sp1-encoding vector in the presence of the dominant negative construct of REST (DNREST). DNREST corresponds to the human cDNA encoding only the DNA-binding domain of REST without the two repressor domains of the protein. The mutant has been shown to compete with the wild type protein for the binding to the NRSE (6). Overexpression of Sp1 did not significantly increase IB1luc activity in the absence of DNREST (Fig. 3). However, in the presence of DNREST, the IB1luc activity was derepressed and overexpression of Sp1 enhanced the IB1luc activity in a dose-dependent manner (Fig.  3). In contrast, mutation in the sp1 motif reduced by 50% the relieved activity of IB1luc and abolished the Sp1 activation.
To validate the negative action of REST on Sp1-induced activation, the IB1luc construct was transiently co-transfected in ␤TC3 cells in the presence of vectors encoding the human repressor REST and Sp1. As shown in Fig. 4, REST prevented the increase of IB1luc activity mediated by Sp1. This effect was abolished when NRSE was mutated (data not shown), indicating that the silencing of Sp1 requires the presence of the NRSE within the promoter. To verify that the silencing of Sp1 activity is not because of a particular context of the MAPK8IP1 promoter, another Sp1-dependent promoter construct was used. Two copies of the NRSE motif from the MAPK8IP1 promoter have been cloned upstream of the viral SV40 promoter that has been described to be controlled by Sp1 (NRSEluc) (6,30). The constructs were transiently transfected in ␤TC3 cells in the presence or in the absence of REST and Sp1. In the absence of REST, overexpression of Sp1 increased the activity of NRSEless SV40 promoter (SV40luc) and NRSEluc by 3-fold (Fig. 5). As expected, REST did not affect the Sp1-mediated activation on the SV40luc. In contrast, REST abolished the Sp1-mediated activation of the NRSEluc. Altogether these results indicate that REST represses the Sp1-mediated activation through the NRSE.
The Silencing of Sp1-mediated Activation Is Trichostatin A (TSA)-sensitive-Previous studies have shown that the RESTmediated repression of IB1luc is TSA-sensitive (6). To assess whether the silencing of Sp1 is HDAC-dependent, HeLa cells were transiently transfected with IB1luc or mIB1luc and incubated for 24h with 100 nM of TSA, a specific inhibitor of HDAC (6). In the presence of TSA, as expected the luciferase activity of the IB1luc was relieved (Fig. 6). Overexpression of Sp1 increased the IB1luc activity in a dose-dependent manner, and this effect was abolished when the sp1 motif was mutated. This result shows that the REST-mediated silencing of Sp1 activity requires histone deacetylases activity.
The CTRD Is Required for Interaction with Sp1 and Abolishes Sp1-dependent Activation-We next explored the possibility that REST might repress expression of its target genes via interaction with Sp1. To assess this hypothesis, we examined the physical interaction of endogenous Sp1 and REST by co-immunoprecipitation assays in HeLa cells. As shown in Fig.  7A, Sp1 and REST were co-immunoprecipitated with anti-Sp1 antibody using nuclear extracts from HeLa cells. The inclusion in the complex of two co-repressors of REST, mSin3A and HDAC1, was investigated because REST was previously reported to associate with these co-factors to regulate expression of its target genes (15). mSin3A and HDAC1 were found to co-precipitate with Sp1 and REST in nuclear extracts from HeLa cells. As a control, we did not detect the presence of the ubiquitous transcription factor E47 in the immunoprecipitate, although it was present in lysates.
To confirm these results described above, co-expression followed by coimmunoprecipitation was performed. A REST fused FIG. 3. Inactivation of REST allows Sp1 activation. IB1luc and mIB1luc constructs were transiently co-transfected with 0.05 g (ϩ) and 0.1 g (ϩ ϩ) of vector encoding Sp1 into REST-expressing HeLa cells in the presence or in the absence of a DNREST. In the absence of DNREST, the IB1luc activity did not function and Sp1 had no significant effect on the promoter activity. In contrast, the IB1luc activity was relieved in cells expressing DNREST, and this was further increased by overexpression of Sp1. Mutations in sp1 motif reduced the Sp1-induced promoter activity in the presence of DNREST (mIB1luc). All experiments were performed six times in triplicate. Results were expressed as mean Ϯ S.E. **, p Ͻ 0.001.

FIG. 4. REST represses Sp1 activity in insulin-secreting cells.
IB1luc construct were transiently co-transfected in the insulin-secreting cell line (␤TC3) with 0.05 g (ϩ) and 0.1 g (ϩ ϩ) of Sp1 in the presence of an expression vector that encodes the repressor REST under the control of the CMV promoter (REST) or the empty vector as a control. REST prevented the Sp1-induced activity of IB1luc. Experiments were performed at least five times in triplicate. Results were expressed as mean Ϯ S.E. **, p Ͻ 0.01.
with GFP (REST-GFP) vector was constructed and transfected in HeLa cells. The lysates from cells expressing REST-GFP and GFP (negative control) were immunoprecipitated with anti-GFP or anti-Sp1 polyclonal antibodies and subjected to SDS-PAGE followed by Western blot analysis with anti-GFP or anti-Sp1 antibodies. As shown in Fig. 7B, both REST-GFP and GFP expressed in HeLa cells were immunoprecipitated with anti-GFP antibody. The endogenous Sp1 factor and REST-GFP were co-immunoprecipitated with anti-GFP and anti-Sp1 antibodies. In contrast, Sp1 did not co-precipitate with GFP alone using anti-Sp1 and anti-GFP antibodies.
We further examined the domains of REST that were able to interact with Sp1. The REST protein displays a modular structure (18). The DNA-binding domain has been localized within the cluster of eight zinc fingers at the N terminus. Two repressor domains have been mapped at the N and the C termini of the protein (31, 32). We constructed expression vectors encod-ing two mutants of REST that lack either the C-terminal (REST⌬C) or N-terminal repression (REST⌬N) domain of REST (Fig. 8A). Both proteins retain the eight zinc finger cluster that functions as the DNA-binding domain targeting the protein to its target genes and as a signal required for nuclear localization (18,33). Thus, REST and the REST mutants should interact with the REST-binding site (NRSE) in a sequence-specific manner. Vectors encoding REST⌬N-GFP and REST⌬C-GFP were constructed and were transiently transfected in HeLa cells. GFP and REST-GFP were used as negative and positive controls, respectively. The REST-GFP and REST⌬N-GFP were co-immunoprecipitated with Sp1 using Sp1 antibodies (Fig. 8B). However, we did not find immunoprecipitation of Sp1 with GFP and REST⌬C-GFP.
We next sought to identify the domains of REST that were able to inhibit Sp1-dependent activation. Transient transfections of IB1luc in the presence of wild type or mutant REST constructs were performed in ␤TC3 cells. As expected, all REST constructs significantly reduced the luciferase activity of IB1luc in ␤TC3 cells (Fig. 9A) indicating that REST mutants were functionally active. No effect on IB1luc activity was observed when the NRSE was mutated in the MAPK8IP1 promoter, indicating that these constructs play a direct role through this element (data not shown). To assess the ability of these different mutants to prevent Sp1 activation, ␤TC3 cells were transiently co-transfected with IB1luc and mutant constructs in the presence of Sp1. The construct encoding the REST⌬N abolished the effect of Sp1, whereas REST⌬C did not inhibit the Sp1-induced activity of IB1luc (Fig. 9B). Taken together, these results indicate that the N-terminal domain of REST is required for interaction with Sp1 and prevents its transcriptional activation. REST interacts with Sp1. A, co-precipitation of endogenous REST and Sp1 proteins were performed using 100 g of nuclear extracts from REST-expressing HeLa cells. Anti-Sp1 antibodies were used to immunoprecipitate (IP) Sp1. The membrane was blotted with the indicated antibodies. B, interaction of overexpressed REST with Sp1 was evaluated by co-immunoprecipitation experiments. HeLa cells were transiently transfected with GFP-tagged REST (REST-GFP) or GFP alone as a negative control for 48 h. In the left panel, REST was immunoprecipitated with anti-GFP antibodies. The Western blot was performed using either anti-GFP or anti-Sp1 antibodies. Conversely, in the right panel, Sp1 was immunoprecipitated with anti-Sp1 antibodies, and the membrane was also blotted with either anti-GFP or anti-Sp1 antibodies.

DISCUSSION
In this study, we investigated the possibility that REST activity might include the inhibition of a transcriptional activator to control the neuronal-and ␤-cell expression of its target genes. First, we showed that the ubiquitous transcriptional factor Sp1 plays a key role in the activation of most REST target genes. Conversely, in REST-expressing cells, Sp1 was unable to exert its transcriptional activation on expression of REST target genes. Co-immunoprecipitation experiments showed that Sp1 interacted with REST. This interaction required the CTRD of REST, which abolished the Sp1-dependent transcriptional activation. This result confirms the singular role of the CTRD to play different functions including the capture of transcriptional activators. A recent study has shown that CTRD is capable of interacting with the TATA-binding protein, a core promoter factor of the basal transcriptional machinery (34). This interaction highlights another mechanism by which REST silences the expression of its target genes in addition to HDAC activity. Here, we provide evidence that inhibition of the transcriptional factor Sp1, an activator of the transcriptional machinery, is also required for the silencing of REST target genes. Sp1 plays a key role in maintaining expression of genes that lack TATA box. The TATA box is bound by a protein complex called TFIID. TFIID contains the TATA-binding protein and associated factors called TAFs such as hTAF II 130, hTAF II 55, and dTAF II 110 that are also able to interact with Sp1 (35)(36)(37)(38)(39)(40). In the absence of TATA box, Sp1 facilitates the binding of TFIID to the promoter, which in turn initiates the stepwise assembly of the preinitiation transcriptional complex for RNA polymerase II-mediated transcription (41,42). This is followed by the ordered assembly of TFIIA, TFIIB, polymerase II/TFIIF, TFIIE, and TFIIH. Sp1 is then thought to be an essential factor for efficient initiation of genes transcription. The fact that REST inhibits Sp1 and the TATAbinding protein activities suggests the hypothesis of an early silencing of the transcriptional machinery. As a possible mechanism, interaction of REST with Sp1 could block the ability of Sp1 to make the necessary contacts with the transcriptional machinery, thus preventing the formation of the initiation transcriptional complex on the core promoter. Consistent with this hypothesis, an interaction of Sp1 with the Krü ppel-like zinc finger repressor ZBP-89 has been proposed to interfere the association of Sp1 with hTAF II 130 (43), thereby leading to the repression of the vimentin gene expression. Several data showed that the recruitment of HDAC1/2 can also contribute to the inactivation of Sp1. This scenario is supported by the fact that Sp1 co-precipitated with HDAC1 and that the treatment of HeLa cells with TSA, an inhibitor of histone deacetylase, allowed Sp1 transcriptional activity to be restored. Moreover, the CTRD can recruit HDAC1/2 (15). The ability of a repressor to inhibit the transcriptional activators in a HDAC-dependent manner has been observed for the retinoblastoma protein Rb, which exerts transcriptional repression by interacting with the FIG. 8. REST interacts with Sp1 through its C-terminal repressor domain. A, the functional domains for DNA binding and transcriptional repression of REST, REST⌬C, and REST⌬N are indicated. REST⌬C and REST⌬N lack the C-and N-terminal repression domain of REST, respectively. B, the REST mutants were evaluated for their ability to interact with Sp1. The GFP-tagged REST mutants were transiently transfected in HeLa cells and co-immunoprecipitation experiments were performed using anti-Sp1 antibodies. When the membrane was blotted with anti-GFP antibodies, REST-GFP and REST⌬N-GFP were detected, whereas the REST⌬C-GFP and GFP were absent. The same membrane was blotted with anti-Sp1 as control for Sp1 immunoprecipitation (IP). M, methionine; E, glutanic acid; K, lysine; F phenylalinine.
FIG. 9. The C-terminal repressor domain of REST abolishes Sp1-activation. A, REST, REST⌬C, and REST⌬N were transiently co-transfected in ␤TC3 cells with IB1luc. All of these constructs repressed the luciferase activity driven by IB1luc. Experiments were performed at least three times in triplicate. Luciferase activities were normalized using pRLCMV Renilla, and results were expressed as mean Ϯ S.E. **, p Ͻ 0.001; *, p Ͻ 0.05. B, to assess the ability of REST mutants to repress Sp1 activation, the IB1luc construct was transiently co-transfected with 0.1 g of Sp1 into ␤TC3 cells in the presence of wild type and mutants of REST. REST and REST⌬N abolished the Sp1induced promoter activity whereas the REST⌬C construct did not inhibit the induction of the promoter by Sp1. Experiments were performed at least five times in triplicate and results were expressed as mean Ϯ S.E. **, p Ͻ 0.001. transcription factor E2F. This interaction has been shown both to inhibit activation of E2F and additionally to allow the recruitment of HDAC1 to the E2F site (44 -46). Thus, Rb has been proposed to repress the transcription of genes both by interfering with the function of transcriptional activators and by effecting changes in the structure of the chromatin that surrounds the promoter. Similarly, REST could allow the recruitment of HDAC1 to Sp1. This in turn could prevent the interaction with other transcriptional factors and/or the accessibility of other transcriptional activators into the vicinity of the promoter.
Compelling data show that REST is able to silence expression of its target genes via different mechanisms. These include the change of chromatin structure and/or the blocking of transcriptional core factor/activator. All these mechanisms are required to mediate full silencing of the target genes. The mechanism of Sp1 inactivation by REST is of importance to understand the molecular pathophysiology of certain diseases. In Huntington's disease, it has been proposed that mutant huntingtin inhibits Sp1-mediated transcription by interfering interaction of Sp1 with hTAF II 130 (47). An independent study has reported that the huntingtin mutant increases the levels of REST in the nuclei of neuronal cells leading to REST-mediated gene silencing (48). We suggest that silencing of REST target genes in Huntington's disease could also be because of a loss of the Sp1-dependent activation resulting from the abnormal presence of REST in neuronal cells. The REST target genes play important roles in controlling neuronal differentiation and glucose-induced insulin secretion (15,16,23). The proper timing of REST target-gene expression during development is required for an appropriate terminal differentiation of the central nervous system (7). The absence of REST and the presence of its target genes in differentiated ␤-cells suggest a similar function for REST during pancreas development. The spatial and temporal silencing of REST is therefore thought to be related with the appearance of ␤-cells and neuronal cell phenotypes during pancreas and neuronal development (15,16,23). Conversely, Sp1 expression is postulated to be required for the maintenance of terminally differentiated cells (49). We propose that Sp1-dependent gene activation, because of the relief of REST activity, could be a central issue for both neuronal and ␤-cell differentiation.