RBP-Jκ/SHARP Recruits CtIP/CtBP Corepressors To Silence Notch Target Genes

Plasmids.

The bait vector for two-hybrid screening, pGBT-SHARP(3291-3664), was constructed by inserting the blunted 1,355-bp NcoI fragment from pcDNA3-FLAG3-SHARP(2002-3664) into the blunted BamHI site from pGBT9 (Clontech). Expression vectors for the Gal4 fusion proteins, G4-VP16, G4-SHARP-RD-VP16, and G4-VP16-SHARP-RD, used in the transcriptional repression assay were made using PCR-assisted cloning. Details on the construction of the pFA-CMV (Stratagene)-based expression plasmids are available on request. The pGa981/6 luciferase reporter plasmid as well as the pCMV-RBP-VP16, pcDNA-3-mNotch-1ΔE, pcDNA3-FLAG2-SHARP, and pcDNA3-FLAG3-SHARP(2002-3664) expression plasmids was described previously (20, 21). The SHARP-specific expression plasmid pcDNA3-FLAG2-SHARP-RD (C-terminal repression domain only) as well as pcDNA3-FLAG2-SHARPΔC and pcDNA3-FLAG3-SHARP(2002-3411) lacking the repression domain was made using PCR-assisted cloning (construction details available on request). For the bacterial expression plasmid pGEX-2TK-SHARP-RD, the SHARP repression domain was amplified by PCR (5′-CGGGATCCGAATTCCAGCCAGCCCC-3′ and 5′-ATCCCGGGTCACACGGAGGCAATGACAATCATG-3′), digested with BamHI and XmaI, and inserted into the corresponding sites of the pGEX-2TK vector (Amersham). The vectors for the expression of glutathione S-transferase (GST)-SHARP deletion mutants (amino acids [aa] 3477 to 3628, 3477 to 3604, and 3477 to 3545) were constructed as follows. Starting from pcDNA3-FLAG2-SHARP-RD, PCR products were amplified using the upstream primer 5′-ATTAATACGACTCACTATAGGGAGACC-3′ and the following downstream primers: 5′-GCCTCGAGTTAGATCTGCAGCACGTAGGCAG-3′, 5′-TACTCGAGTTACTGCAGGTAAGTGATGAAGGC-3′, and 5′-TACTCGAGTTAGGGCCCTCCTTCAGAAAGGG-3′. The PCR products were then digested with EcoRI and XhoI and inserted into the corresponding sites of pGEX6P1 (Amersham). The eukaryotic expression vector for T7-tagged human CtBP1 and the prokaryotic expression vector for GST-CtBP1 were provided by G. Chinnadurai. The CtIP expression vector (pSP6-CtIP) was a gift from R. Baer. Myc-tagged CtIP constructs (wild type, Δ LXCXE, and ΔPLDLS) were supplied by J. R. Nevins, and CtIP constructs pcDNA3.1HA-CtIP [59 to 320] and pcDNA3.1HA-CtIP (281 to 620) were provided by J. E. Visvader. A 1.6-kb fragment was isolated after XbaI digestion of pEFrHAPGKpuropA-CtIP (aa 371 to 897), which was supplied by J. E. Visvader, and inserted into the XbaI site of pcDNA3-FLAG1, resulting in pcDNA3-FLAG1-CtIP (aa 371 to 897). After EcoRI/XbaI digestion of pGEXCtIP (aa 620 to 897), which was supplied by R. Baer, a 1.1-kbp fragment was isolated and inserted into the corresponding sites of pcDNA3-FLAG3, resulting in pcDNA3-FLAG3-CtIP (aa 620 to 897). The luciferase reporter plasmid for the Hey1 promoter (−95/+87) was a gift from M. Gessler, and the E1A expression plasmids E1A-Exon2 (pc-dl1119) and E1A-Exon2ΔCID (pc-dl1135) were supplied by C. Svensson.

Cell lines.

The HEK-293 (ATCC CRL 1573) and HeLa (ATCC CCL 2) cell lines as well as mouse embryonic fibroblasts (as described in reference 8), which were kindly provided by J. D. Hildebrand, were grown at 37°C under 5% CO₂ in Dulbecco's modified Eagle medium (Gibco) supplemented with 10% fetal calf serum.

Yeasttwo-hybridscreening.

Yeast (Saccharomyces cerevisiae) two-hybrid screening was performed using Gal4 fusion proteins as previously described (38). Briefly, Saccharomyces cerevisiae strain Y153 was transformed with the pGBT-SHARP(3291-3664) bait plasmid using the lithium acetate method and stably maintained in the absence of tryptophan. Yeasts were subsequently transformed with a pACT-based cDNA library derived from Epstein-Barr virus-transformed human peripheral lymphocytes and grown (2.8 million primary transformants) on His/Leu/Trp dropout plates containing 20 mM 3-aminotriazole. His⁺ colonies were analyzed for beta-galactosidase activity using the X-gal (5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside) filter assay as previously described (38). Prey plasmids from clones with positive results in both assays were isolated and propagated in Escherichia coli strain DH10B and sequenced using the gal843 primer (5′-GCGTTTGGAATCACTACAGGG-3′).

Preparation of cell extracts.

Whole-cell lysates were prepared as follows. Cells were washed three times in phosphate-buffered saline (PBS) and pelleted by centrifugation at 300 × g. The pellet was resuspended in 5 volumes of ice-cold CHAPS lysis buffer consisting of 10 mM 3-[(cholamidopropyl)-dimethylammonio]-1-propanesulfonate, 50 mM Tris-HCl (pH 7.9), 150 mM NaCl, 2 mM EDTA, 5 mM NaF, 10 μg/ml aprotinin, 10 μg/ml leupeptin, 1 mM dithiothreitol (DTT), and 0.5 mM phenylmethylsulfonyl fluoride and incubated on ice for 40 min. The lysate was cleared by centrifugation at 80,000 × g for 30 min. Protein concentrations were determined using the Bradford assay method (Bio-Rad). Extracts were used for immunoprecipitation, in vitro interaction assays, and Western blotting.

In vitro protein translation.

Proteins were translated in vitro in the presence of [³⁵S]methionine using the reticulocyte lysate-coupled transcription/translation system according to the manufacturer's instructions (Promega). Translation and labeling quality were monitored by SDS-PAGE.

GST pull-down assay.

GST fusion proteins were expressed in E. coli strain BL21-CodonPlus-RIL (Stratagene) and stored as whole bacterial lysates. GST protein and GST fusion proteins (approximately 1 μg) were immobilized with Sepharose beads (Amersham) and incubated at 4°C for 1 h together with in vitro-translated proteins under rotation. The reaction mixtures were washed three times with 600 μl buffer A (40 mM HEPES [pH 7.5], 5 mM MgCl2, 0.2 mM EDTA, 0.5 mM DTT, 1% Nonidet P40 [NP-40], and 100 mM KCl) and three times with buffer B (equivalent to buffer A, but containing 300 mM KCl). After these washing steps, the reaction mixtures were boiled and proteins were separated by SDS-polyacrylamide gel electrophoresis. Gels were dried and exposed to X-ray films. In addition to cell-free synthesized and radiolabeled proteins, extracts from transfected HEK-293 cells were also used in pull-down assays. In these cases, proteins interacting with GST fusion proteins were detected by Western blotting.

Cell sorting and quantitative RNA analysis.

A total of 10⁶ mouse embryonic fibroblasts was transfected either with an expression vector for enhanced green fluorescent protein (EGFP) (pEGFPC1; Clontech) or together with expression vectors for T7-CtBP1 or FLAG-CtBP1. Forty-eight hours after transfection, cells were harvested and GFP-positive cells (2 × 10⁵) were sorted on a FACS Star Plus cell sorter (Becton Dickinson). Total RNA was isolated, and mRNA expression levels were quantified using real-time PCR (TaqMan; PE Applied Biosystems). For PCR, cDNAs were reverse transcribed from 2 μg of total RNA. The PCR (denaturation 95°C for 2 min, followed by 40 cycles of 95°C for 15 s and 60°C for 1 min) was performed using SYBR green PCR core reagents (PE Applied Biosystems) and primer combinations listed below. Endogenous cyclophilin mRNA levels were used as internal controls. The following primers were used: mCyclophilin-F, 5′-ATGGTCAACCCCACCGTGT-3′; mCyclophilin-R, 5′-TTCTTGCTGTCTTTGGAACTTTGTC-3′; mHey1F, 5′-CACTGCAGGAGGGAAAGGTTATT-3′; mHey1R, 5′-GCCAGGCATTCCCGAAAC-3′; mNotch-1F, 5′-ACATCCGTGGCTCCATTGTCTA-3′; and mNotch-1R, 5′-TCTTGTAAGGAATATTGAGGCTGCC-3′.

DNA transfection and luciferase assay.

A total of 10⁶ HEK-293 cells was transfected in 90-mm culture dishes with 5 to 10 μg of expression plasmid DNA using calcium phosphate coprecipitation as previously described (21). Proteins were extracted 24 h after transfection and assayed for protein expression or used in pull-down assays. HeLa cells were transfected (2 × 10⁵) in 35-mm culture dishes with 2 μg of reporter plasmid DNA together with various amounts of expression plasmid using the FuGENE transfection reagent (Roche). Mouse embryonic fibroblasts were transfected with Lipofectamine (Invitrogen). Luciferase activity was determined from at least four independent transfections with 20 μl of cleared lysate in an LB 9501 luminometer (Berthold) using the luciferase assay system from Promega. All transfections were normalized using total cellular protein concentrations (Bradford assay; Bio-Rad).

Immunofluorescence assay.

HEK-293 cells were cultured on glass coverslips in a 25-well plate (Bibby Sterilin, Ltd.) at a density of 10⁵ cells per cm². After 16 h, cells were transfected with 500 ng of expression plasmids. Cells were rinsed with PBS 24 h after transfection and fixed and permeabilized with 0.1% Triton X-100. Nonspecific immunostaining was blocked by incubating the cells in 3% bovine serum albumin in PBS with 0.1% Tween 20. An antibody directed against the FLAG epitope (M5; Sigma) was used for detection of SHARP, and an antibody directed against the Myc epitope (9B11; Cell Signaling) was used for detection of the CtIP proteins. Staining was performed using an Alexa Fluor 488-conjugated goat anti-mouse immunoglobulin G (IgG; Molecular Probes), or a Cy3-conjugated goat anti-mouse IgG (Jackson Immuno Research) secondary antibody. After washing and mounting, the cells were analyzed using a fluorescence microscope (DMIRB; Leica).

Western blotting.

Western blotting was performed as previously described (21). For the detection of the endogenous proteins, the following antibodies were used: anti-RBP-Jκ (rat monoclonal IgG2a, T6709 [Institute of Immunology Co., Ltd.] secondary antibody; peroxidase-conjugated goat anti-rat IgG [Dianova]), anti-CtIP (rabbit polyclonal IgG, 612L, raised against residues 58 through 369 [gift from R. Baer and described in reference 40] secondary antibody; peroxidase-conjugated donkey anti-rabbit IgG [Amersham]), anti-CtBP (rabbit polyclonal IgG, H-440 [Santa Cruz]), and anti-p65/RelA (goat polyclonal IgG, C-20 [Santa Cruz] secondary antibody; peroxidase-conjugated rabbit anti-goat IgG [Dianova]). To analyze coimmunoprecipitated CtBP1 proteins, membranes were incubated with the primary antibody directed against the T7 epitope (Bethyl). Coimmunoprecipitated CtIP proteins were detected with an antibody directed against the Myc epitope (9B11; Cell Signaling). As a secondary antibody, a 1:7,000 dilution of peroxidase-conjugated sheep anti-mouse IgG (Amersham) or a 1:5,000 dilution of peroxidase-conjugated donkey anti-rabbit antibody (Amersham) was used.

Coimmunoprecipitation.

Immunoprecipitation was carried out using HEK-293 cell extracts 24 h after cotransfection with FLAG2-SHARP-RD (aa 3411 to 3663) and expression plasmids for human CtBP1 or CtIP. Cells were lysed in 900 μl CHAPS lysis buffer. Extracts were incubated with 40 μl of an agarose-conjugated anti-FLAG antibody (M2; Sigma) at 4°C overnight. Beads were washed three times with CHAPS lysis buffer containing 300 mM LiCl. After a further wash with CHAPS lysis buffer containing 150 mM LiCl, the beads were resuspended in SDS-PAGE loading buffer.

Injection of mRNA into Xenopus laevis embryos and whole-mount in situ hybridization.

Microinjections of capped mRNA of mNotch-1ΔE, SHARP, and SHARPΔC and whole-mount in situ hybridization using a digoxigenin-labeled antisense RNA probe for N-tubulin were performed as previously described (20). Stained embryos were postfixed overnight in MEMFA (0.1 M morpholinepropanesulfonic acid [MOPS], 2 mM EGTA [pH 8.0], 1 mM MgSO₄, 3.7% formaldehyde) and depigmented using 10% H₂O₂ in methanol. The uninjected side was used as a control for normal primary neurogenesis.

Purification of RBP-Jκ DNA-binding complexes.

For the preparation of the DNA affinity column, a 340-bp fragment containing 12 RBP-Jκ-binding sites was isolated by NsiI/HindIII digestion of pGa981/6. Approximately 0.15 nmol of the fragment was biotinylated and immobilized on streptavidin Sepharose (Amersham). Whole-cell extract from 10⁹ Jurkat T cells in a total volume of 10 ml was first precleared by incubating with streptavidin Sepharose alone at 4°C for 1 h. The precleared lysate was then incubated with the affinity column at 4°C for 2 h. The supernatant was collected, and, after three washing steps with 7 ml of CHAPS lysis buffer, DNA-binding complexes were each eluted with 1 ml of CHAPS lysis buffer containing increasing NaCl concentrations (elution E1, 180 mM; E2, 200 mM; E3, 300 mM; E4, 400 mM; E5, 500 mM; E6, 700 mM; and E7, 1,000 mM) and analyzed by electrophoretic mobility shift assay (EMSA) and Western blotting.

EMSA.

Cell extract, flowthrough, washing steps, and eluted fractions (2 μl) were used for electromobility gel shift assays in a binding buffer consisting of 10 mM Tris-HCl (pH 7.5), 100 mM NaCl, 0.1 mM EDTA, 0.5 mM DTT, and 4% glycerol. For binding reaction, 2 μg poly(dI-dC) (Amersham) and approximately 0.5 ng of ³²P-labeled oligonucleotides were added. The sequence of the double-stranded oligonucleotide FO233 (5′-CCTGGAACTATTTTCCCACGGTGCCCTTCCGCCCATTTTCCCACGAGTCG-3′) corresponds to the two RBP-Jκ-binding sites within the EBVTP-1 promoter. The reaction products were separated using 5% polyacrylamide gels with 1× Tris-glycine-EDTA at room temperature. Gels were dried and exposed to X-ray films.

Functional characterization of the SHARP repression domain.

Previous studies have shown that the evolutionarily conserved repression domain of SHARP (SHARP-RD), also called the SPOC (Spen paralog and ortholog C-terminal) domain, resides at the C terminus (1, 20, 27). Here, we show that the SHARP-RD is necessary and sufficient to repress transcription in transient cotransfection experiments. In addition, we demonstrate that SHARPΔC, lacking the repression domain, acts in a dominant negative fashion in a Notch-dependent neurogenesis assay using Xenopus laevis embryos. SHARP-RD was fused to the Gal4 DNA-binding domain and the VP16 transactivation domain as shown in Fig. 1A. The expression plasmids G4-VP16, G4-SHARP-RD-VP16, and G4-VP16-SHARP-RD were transiently cotransfected into HeLa cells together with the pFR-LUC reporter construct containing five Gal4-binding sites. The Gal4-VP16 control exhibited a dose-dependent increase of luciferase activity, whereas this effect was abrogated by either the Gal4-SHARP-RD-VP16 or the Gal4-VP16-SHARP-RD expression plasmid (Fig. 1B). This suggests that SHARP-RD alone is sufficient to recruit additional corepressors capable of overriding the strong effect of the VP16 transcriptional activator. Subsequently, we investigated whether full-length SHARP requires its C-terminal repression domain for RBP-Jκ-mediated transcriptional regulation. The expression plasmids for RBP-VP16, full-length SHARP, or a truncated SHARP lacking the C-terminal repression domain were transiently transfected into HeLa cells together with an RBP-Jκ-responsive reporter construct, pGa981/6 (Fig. 1C). This luciferase reporter plasmid contains six repeats of the EBNA-2-responsive element within the Epstein-Barr virus TP-1 promoter upstream to a minimal β-globin promoter. Cotransfection of the RBP-VP16 expression plasmid resulted in approximately 100-fold transcriptional activation. Transactivation mediated by RBP-VP16 was gradually reduced after cotransfection of the full-length corepressor, SHARP. In contrast, SHARP lacking its C-terminal repression domain (SHARPΔC) had a dominant negative effect, further stimulating the reporter gene instead of repressing it. Therefore, the SHARP-RD is necessary for transcriptional repression in transient transfection assays.

We have shown previously that SHARP acts as a Notch antagonist during Xenopus laevis neurogenesis (20). To test the role of the SHARP repression domain in this context, we overexpressed SHARPΔC in one half of the embryo. Whole-mount in situ hybridization for N-tubulin allowed for the identification of primary neurons in Xenopus laevis neurulae (Fig. 1D). As shown previously, embryos injected with Notch-1ΔE mRNA had no lateral and, in some cases, no intermediate primary neurons in the injected side (Fig. 1D, panel b). In contrast, overexpression of SHARP resulted in more primary neurons and a broader zone of lateral primary neurons in the injected sides (Fig. 1D, panel c). Interestingly, overexpression of SHARPΔC induced, like Notch-1ΔE, a neurogenic phenotype (Fig. 1D, panel d), resulting in a loss of primary neurons (75 out of 138 neurons [54%]). SHARPΔC showed a dominant negative effect on primary neurogenesis, which was analogous to our experiments assessing gene activation.

SHARP associates with the CtBP corepressor.

During Drosophila melanogaster Notch signaling, H can interact with the RBP-Jκ homolog SuH as well as corepressors such as dCtBP (2). In mammalian Notch signaling, SHARP interacts with both RBP-Jκ and corepressors such as SMRT (20, 27). Here, we show that endogenous RBP-Jκ also interacts strongly with GST-SHARP (Fig. 2A). So far, no sequence similarities could be detected between Drosophila H and mammalian SHARP. Assuming that SHARP may still have functional similarities to Drosophila H, the interaction of SHARP and the CtBP corepressor was tested in vivo. HEK-293 cells were transfected with an expression plasmid for FLAG-tagged SHARP-RD alone or together with T7-tagged human CtBP1. SHARP was immunoprecipitated using an anti-FLAG antibody coupled to beads. Subsequently, the coimmunoprecipitated T7-CtBP protein was detected on Western blots using an antibody against the T7 tag (Fig. 2B).

In an attempt to verify these results in an in vitro system, a GST pull-down assay was conducted using a glutathione S-transferase fusion protein for the SHARP repression domain (GST-SHARP-RD). Glutathione-Sepharose beads were coated with bacterially expressed GST or GST-SHARP-RD and used as bait for cell-free synthesized and radiolabeled CtBP1. No binding or extremely weak binding of in vitro-translated CtBP1 to GST-SHARP-RD was detected (Fig. 2C, lane 2) compared to that of our positive control, in vitro-translated CtIP (Fig. 2C, lane 3, see also Fig. 3). However, GST-SHARP-RD bound FLAG-tagged CtBP in cell lysates from FLAG-CtBP1-transfected HEK-293 cells (Fig. 2D). We concluded either that posttranslational modifications were necessary for SHARP-CtBP binding or that binding was indirect and required a bridging factor present in HEK-293 lysates.

SHARP interacts with the CtIP corepressor directly.

A yeast two-hybrid screen was performed using the SHARP repression domain fused to the Gal4 DNA-binding domain and a human B-cell library (38). We identified SMRT and NCor, which had been previously described to interact with SHARP (27). CtIP was among the newly identified interacting proteins and fit the requirements of a possible bridging factor for SHARP and CtBP1 interaction.

The interaction of SHARP and CtIP was verified in vitro using the pull-down assays with GST-SHARP-RD (Fig. 3A and B). Full-length CtIP strongly bound to GST-SHARP-RD (Fig. 3A, lane 3) as well as to CtBP (Fig. 3A, lane 2) but not to GST alone (Fig. 3A, lane 5) or GST-RBP-2N (Fig. 3A, lane 1). The CtIP protein contains a binding site for the Rb protein LXCXE in addition to the binding site for CtBP, PLDLS. Neither of these sequences are necessary for the SHARP/CtIP interaction (Fig. 3B, lane 4), but, as previously shown, the PLDLS motif is required for the CtIP/CtBP interaction (24). Pull-down assays were also performed with in vitro-translated CtIP and various GST-SHARP-RD deletion constructs. Deletion of the C-terminal 36 aa of SHARP-RD completely abolished binding to CtIP (Fig. 3C). While this 36-aa segment is necessary for CtIP interaction, it is not sufficient for binding (data not shown). Pull-down assays using GST-SHARP-RD and various in vitro-translated and radiolabeled CtIP proteins identified two regions of CtIP as binding to GST-SHARP-RD, the N terminus (aa 59 to 320) and the C terminus (aa 620 to 897) (Fig. 3D to F). The same interaction domains were previously identified for transcription factor LMO4 (32).

The SHARP/CtIP interaction was further examined in vivo using coimmunoprecipitation experiments. FLAG-tagged SHARP-RD was expressed either alone or together with Myc-tagged CtIP in HEK-293 cells. After the precipitation of SHARP-RD from cell lysates using a monoclonal anti-FLAG antibody, CtIP was detected on Western blots using an antibody against the Myc-tag. CtIP was detected in the coimmunoprecipitate only when both proteins were expressed (Fig. 4A, lane 3). Furthermore, SHARP colocalized with CtIP in the cell nucleus (Fig. 4C). In fact, we demonstrated that CtIP colocalizes with wild-type SHARP(12-3664) in the nucleus and SHARP lacking a nuclear localization signal (aa 2002 to 3664) in the cytoplasm (Fig. 4D, panels a and d and panels b and e). CtIP did not colocalize in the cytoplasm with SHARP(2002-3411) lacking C-terminal RD (Fig. 4D, panels c and f).

Importantly, we show that GST-RBP-2N, a splice variant of RBP-Jκ, FLAG-SHARP, and Myc-CtIP form a complex (Fig. 4B). FLAG-SHARP(2002-3664) or C-terminally truncated SHARP(2002-3411) lacking the RD was coexpressed with Myc-CtIP in HEK-293 cells. GST-RBP-2N specifically bound FLAG-SHARP proteins and Myc-CtIP. However, this ternary RBP-SHARP/CtIP complex did not form when cell lysate expressing the truncated FLAG-SHARP protein was used (Fig. 4B, lanes 7 and 10).

SHARP, CtIP, and CtBP act together to repress transcription.

The significance of our biochemical findings that CtIP and CtBP are novel constituents of the RBP-SHARP corepressor complex was investigated using a transrepression assay for CtIP and CtBP function on SHARP-mediated transcriptional repression. Expression plasmids for either Gal4-VP16 or Gal4-VP16-SHARP-RD in combination with expression plasmids for CtBP1 or CtIP were cotransfected into HeLa cells together with a Gal4-responsive reporter construct. Neither CtBP1 nor CtIP on its own could disrupt Gal4-VP16-mediated transactivation (Fig. 5A). However, when G4-VP16-SHARP-RD was used in the transrepression assay, the expression of either CtBP1 or CtIP increased the repressive effect of SHARP (Fig. 5A). In addition, a CtIP construct lacking the CtBP interaction domain (CtIP ΔPLDLS) failed to repress transcription (Fig. 5A). SHARP-dependent transcription requires, in part, the presence of CtBP (Fig. 5B). Mouse embryonic fibroblasts (MEFs) were transfected with the Gal4-dependent reporter construct pFR-LUC and the SHARP-RD specific expression plasmid G4-VP16-SHARP-RD. In CtBP-deficient MEFs, SHARP-mediated repression is less effective than in the heterozygous MEFs. Importantly, we could demonstrate that the transcriptional repression of CtIP, when it functions as a corepressor for SHARP, is dependent on CtBP (Fig. 5C). MEFs were transfected with the reporter construct and the SHARP-RD-specific expression plasmid G4-VP16-SHARP-RD, together with increasing amounts of a CtIP expression plasmid. A dose-dependent repressive activity for CtIP was detected only in heterozygote MEFs, not in CtBP1/CtBP2-deficient MEFs (Fig. 5C). These results strongly suggest that CtIP corepressor function with SHARP-RD requires CtBP.

SHARP-mediated repression requires the CtBP corepressor.

Hey1 was identified as a Notch target gene in mammals (15). To test the effect of CtBP- and CtIP-mediated transcriptional repression on a naturally existing Notch-responsive promoter, we performed cotransfection experiments with the human Hey1 promoter fused to the luciferase gene. Cotransfection of the Hey1 promoter construct together with the RBP-VP16 activator into HeLa cells strongly activated transcription, and increasing amounts of full-length SHARP resulted in a clear repression of RBP-VP16-mediated transactivation (Fig. 6A). In order to investigate the functional role of CtBP in transcriptional repression, the ability of the E1A adenoviral protein to derepress Hey1 transcriptional control was explored. It was previously shown that the portion of the E1A protein encoded by exon 2 strongly binds CtBP, thereby relieving CtBP-mediated transcriptional repression (33). Cotransfection of increasing amounts of E1A exon 2 lessened CtBP-mediated repression of the Hey1 promoter in HeLa cells, thereby increasing promoter activity up to sixfold (Fig. 6B). A mutant E1A exon 2 that no longer binds to CtBP, however, had no effect on transcriptional activity of the Hey1 promoter. Similar experiments conducted in CtBP-deficient mouse embryonic fibroblasts showed that transfection of increasing amounts of CtBP1 rescued transcriptional repression of the Hey1 promoter (Fig. 6C). Additionally, CtBP1-induced repression was relieved by the cotransfection of wild-type E1A exon 2 but not by mutant E1A exon 2. Neither wild-type nor mutant E1A exon 2 has an effect on Hey1 promoter activity in the absence of CtBP1. Finally, we utilized quantitative reverse transcription-PCR of either heterozygous or CtBP-deficient MEFs to demonstrate that in the absence of CtBP, the endogenous Hey1 message is strongly derepressed or upregulated (Fig. 6D). When CtBP was reintroduced into CtBP-deficient MEFs, repression at the endogenous Hey1 promoter was partially restored (Fig. 6E).

Formation of the endogenous corepressor complex.

Both biochemical and functional data suggest the presence of an endogenous RBP-Jκ/SHARP-CtIP/CtBP corepressor complex. We set out to isolate also the endogenous complex taking advantage of the high affinity of RBP-Jκ to its DNA-binding site 5′-CGTGGGAA-3′. RBP-Jκ itself binds tightly to a biotinylated multimerized RBP-Jκ-oligonucleotide column as seen in EMSA and Western blotting with anti-RBP-Jκ antibodies (peak fractions 8 to 10, elution fractions E3 to E5; Fig. 7). As monitored by silver staining and Western blotting, our DNA affinity purification approach yielded a defined spectrum of bands for the putative RBP-Jκ-containing complex(es) (data not shown); the achieved purification was from 60-fold (E2) to 800-fold (E6). Importantly, CtIP coelutes with RBP-Jκ in elution fractions E4 and E5. CtBP can still be found to bind to the RBP-Jκ oligonucleotide column. However, most of it comes off in elution E1 (180 mM NaCl).

We propose that both CtIP and CtBP are new constituents of the mammalian RBP-Jκ (Notch) corepressor complex. Because SHARP is capable of interacting with both RBP-Jκ and CtIP directly, it is conceivable the SHARP plays a functionally similar role to Drosophila Hairless.

CtIP and CtBP as novel components of RBP/SHARP-corepressor complex.

Both HDAC-independent and -dependent mechanisms have been proposed for RBP-Jκ-mediated transcriptional repression and are not necessarily mutually exclusive. It was shown that RBP-Jκ is able to recruit an HDAC/SMRT/SKIP-containing complex (9, 42). In contrast, Honjo and coworkers showed recently that the activity of the Hes1 and Hes5 promoters is not increased in RBP-Jκ-deficient T cells (34). One interpretation is that neither RBP-Jκ nor an associated RBP-Jκ corepressor complex is needed for transcriptional repression. In our case, looking at the endogenous Notch target gene Hey1 (Fig. 6D and E), the absence of corepressor CtBP does lead to strong derepression. This discrepancy might be explained by different regulation at different Notch target genes.

Our previous findings describing SHARP-mediated repression as being sensitive to trichostatin agree with findings that HDAC activity is required for optimal RBP-Jκ corepressor function (20). Here, we further elucidated the mechanism of RBP-Jκ-mediated transcriptional repression by identifying the CtIP and CtBP corepressors as novel constituents of the RBP-Jκ/SHARP corepressor complex. Transcriptional repression via the joint action of CtIP and CtBP has previously been demonstrated for Rb/p130- and Ikaros-mediated repression (10, 16). In both cases, Ikaros and Rb utilize HDAC-dependent and -independent mechanisms of transcriptional repression. CtBP is a broadly expressed corepressor protein that binds Pro-X-Asp-Leu-Ser (PXDLS) motifs present in diverse transcriptional regulators. CtBP also binds to HDACs and Polycomb group proteins (26). The ability of CtBP to form homodimers may allow it to serve as a link between these transcriptional silencing complexes and regulators bearing PXDLS motifs. Previously, Shi and colleagues described a CtBP repressor complex that exhibited a number of enzymatic activities, including HDAC and histone methyltransferases (29). Most recently, the same group showed that lysine-specific demethylase 1, another component of the CtBP complex, is capable of demethylating histone H3 at the Lys 4 position (28). Combining the data we presented here with these findings allows speculation that histone demethylation plays a role in Notch target gene repression. Alternatively, the recruitment of Polycomb-containing repressor complexes via CtIP/CtBP may be involved in maintaining Notch target gene silencing.

Is SHARP the functional analog of Drosophila Hairless?

SHARP belongs to the split ends (Spen) protein family, bearing characteristic N-terminal RRMs and a conserved SPOC domain (1). Spen-like proteins have been identified in vertebrates but also in Caenorhabditis elegans and Drosophila. However, it is unlikely that Drosophila Spen (dSpen) is the functional SHARP homolog. (i) Despite the homologies between dSpen and SHARP within the N-terminal RRMs and the C-terminal SPOC domain, the intervening region shows only poor homology and differs greatly in size. (ii) We have shown that SHARP physically interacts with RBP-Jκ via a well-defined interaction domain and antagonizes Notch-mediated transcriptional activation (20). The RBP-Jκ-interaction domain of SHARP is highly conserved from human to Xenopus but is not present in dSpen. Furthermore, no physical interaction between dSpen and SuH, the RBP-Jκ ortholog in Drosophila, has yet been reported.

SuH was originally identified on the basis of its dominant suppression of the H phenotype, and genetic evidence indicates that H negatively regulates the activity of SuH and the Notch pathway (17). H is a potent antagonist of Notch pathway activity and binds not only to SuH but also to dCtBP and Groucho corepressors (2, 17). However, to date, no mammalian homolog of H has been identified. Functionally analogous to Drosophila H, the vertebrate protein SHARP plays the bridging role between RBP-Jκ and corepressors. Since Drosophila H is able to recruit the dCtBP corepressor, we investigated whether SHARP had a similar function and recruits the mammalian CtBP protein. The initial requirement is that SHARP is capable of interacting with CtBP, and, indeed, we show here that this protein coimmunoprecipitates. However, the SHARP-CtBP interaction is not direct but requires an additional bridging factor, CtIP.

The C-terminal SPOC domain of SHARP in transcriptional repression.

We and others have previously reported SHARP to be important for transcriptional repression. The murine homolog of SHARP, MINT, is required for the repression of the homeodomain transcriptional repressor, Msx2 (18). The SHARP SPOC domain interacts with the corepressors SMRT and NCor (27). X-ray crystallography of the SHARP SPOC domain shows conserved residues that are important for the interaction with SMRT or NCor corepressors (1). SMRT and NCor, in turn, recruit larger multiprotein complexes containing HDAC activity.

Here we show that the SPOC domain alone is able to repress transcription. Furthermore, the C-terminal 36 amino acids are required to interact with the coiled-coil domains of the CtIP corepressor. Therefore, we suggest that the SHARP SPOC domain binds not only SMRT and NCor but also corepressors CtIP and CtBP. RBP/SHARP-mediated repression possibly recruits redundant corepressors that work at different Notch target genes in combination or alone and that function not only in an HDAC-dependent but also in an HDAC-independent manner.

The role of CtIP/CtBP in Notch signaling in vivo.

It is well established that Notch signaling is important for several differentiation decisions during embryonic development. It would follow that deficiency of key components such as RBP-Jκ, SHARP, CtBP, or Notch target genes should cause severe developmental defects. As expected, RBP-Jκ deficiency in mice causes gross developmental anomalies, resulting in embryonic death before day 10.5 of gestation (19). MINT-deficient mice die at embryonic day 12.5 to 13.5 and exhibit defects in cardiac development (12). Inactivation of CtIP leads to early embryonic lethality mediated by a G₁ restraint (3). Here, we show that the Notch target gene Hey1 is negatively regulated by corepressors SHARP, CtIP, and CtBP. Deficiency in Hey1/Hey2 results in death after embryonic day 9.5 (E9.5) with vascular defects. Similar defects are observed in Jagged1 and Notch1 knockout mice (7). Deficiency in the CtBP corepressor leads to embryonic death at embryonic day 10.5 due to defects in extraembryonic vascularization (8). The different knockout mouse models clearly demonstrate important roles for RBP-Jκ, SHARP, CtIP, and CtBP in early embryonic development, as expected for mice with aberrant Notch signaling.

In summary, our results indicate that corepressors CtIP and CtBP are novel components of the RBP/SHARP corepressor complex and that CtBP is required for the repression of the Notch target gene Hey1. It will be interesting to investigate the impact of CtIP and CtBP on Notch signaling and nuclear hormone signaling in vertebrate systems. Conditional knockout mice will be key in supplying many answers that describe the control mechanisms in these signaling pathways.