Assembly of a Notch Transcriptional Activation Complex Requires ...

MOLECULAR AND CELLULAR BIOLOGY, Apr. 2011, p. 1396–1408 0270-7306/11/$12.00 doi:10.1128/MCB.00360-10 Copyright © 2011, American Society for Microbiology. All Rights Reserved.

Vol. 31, No. 7

Assembly of a Notch Transcriptional Activation Complex Requires Multimerization䌤 Rodrigo Vasquez-Del Carpio,¶ Fred M. Kaplan,†¶ Kelly L. Weaver, Jeffrey D. VanWye, Marie-Clotilde Alves-Guerra,‡ David J. Robbins, and Anthony J. Capobianco* Molecular Oncology Program, Division of Surgical Oncology, Dewitt Daughtry Family of Surgery and Sylvester Comprehensive Cancer Center, Miller School of Medicine, University of Miami, 1600 NW 10th Avenue, Miami, Florida 33136 Received 26 March 2010/Returned for modification 17 May 2010/Accepted 5 January 2011

Notch transmembrane receptors direct essential cellular processes, such as proliferation and differentiation, through direct cell-to-cell interactions. Inappropriate release of the intracellular domain of Notch (NICD) from the plasma membrane results in the accumulation of deregulated nuclear NICD that has been linked to human cancers, notably T-cell acute lymphoblastic leukemia (T-ALL). Nuclear NICD forms a transcriptional activation complex by interacting with the coactivator protein Mastermind-like 1 and the DNA binding protein CSL (for CBF-1/Suppressor of Hairless/Lag-1) to regulate target gene expression. Although it is well understood that NICD forms a transcriptional activation complex, little is known about how the complex is assembled. In this study, we demonstrate that NICD multimerizes and that these multimers function as precursors for the stepwise assembly of the Notch activation complex. Importantly, we demonstrate that the assembly is mediated by NICD multimers interacting with Skip and Mastermind. These interactions form a preactivation complex that is then resolved by CSL to form the Notch transcriptional activation complex on DNA. Numerous intricate cellular processes are implemented through direct cell-to-cell interactions. Depending on the cell type, the Notch signal transduction pathway initiates a variety of cellular processes, including proliferation, differentiation, and apoptosis, through these cell-to-cell interactions (2, 9, 23, 24, 28, 32). For proper development and cellular homeostasis, tight regulation of Notch signaling is essential. Inappropriate Notch signaling is observed in neoplasms of many tissue types, indicating that deregulation of Notch signaling is involved in the initiation and/or maintenance of the neoplastic phenotype (3–6, 11, 14, 31). Therefore, understanding the mechanisms governing the tight regulation of Notch signaling is essential. In mammals, there are four known Notch genes (Notch1 to -4 genes) that encode single transmembrane-spanning cell surface receptors (2). Current models of the Notch signal transduction pathway suggest that the extracellular domain of Notch interacts with the extracellular domains of ligands found on adjacent cells. Ligands from the DSL (Delta, Serrate, and Lag-2) family of proteins interact with Notch receptor, and these interactions dictate a series of proteolytic events that release the intracellular domain of Notch (NICD) from the plasma membrane. NICD translocates into the nucleus, where it

interacts with the DNA binding protein CSL (for CBF-1/Suppressor of Hairless/Lag-1) and transcriptional coactivators of the Mastermind-like family to regulate transcription (20, 30, 37, 38). Other proteins have been postulated to be associated with this complex, one of which is the Ski-interacting protein (Skip). Skip was initially identified as a bifunctional nuclear receptor of vitamin D and as a repressor of Notch signaling in association with the protein SMRT (8). Subsequently, Skip was shown to interact with NICD and also function as a coactivator for Notch transcriptional activation, although no mechanistic detail is known (13, 39). In naturally occurring tumors and in model systems, cells transformed by Notch contain two distinct high-molecularweight NICD complexes (16). One of these complexes is localized predominantly in the nucleus, were NICD associates with Mastermind-like 1 (Maml1) and CSL to form the activation complex (termed the A complex). The second of these NICD complexes is localized predominantly in the cytoplasm and does not contain either Maml1 or CSL (termed the P complex). The relationship between these complexes is unclear; however, one intriguing possibility is that the formation of the activation complex is derived from and perhaps depends on the smaller P complex. Some of the molecular details of this trimeric complex were revealed in the crystal structure of CSL and truncated polypeptides of NICD and Maml1 (26, 36). Although these structures reveal important interactions between NICD, Maml1, and CSL, they do little to reveal the molecular events that lead to the assembly of this transcriptional activation complex. Here, we report that NICD forms multimers and that this multimerization is the initial step in transcriptional activation complex assembly. Subsequently, the NICD multimer forms a complex with Skip, which then provides a docking site to recruit Maml1 and form a preactivation complex. The interaction between the preactivation complex and CSL results in the

* Corresponding author. Mailing address: Molecular Oncology Program, Division of Surgical Oncology, Dewitt Daughtry Family of Surgery and Sylvester Comprehensive Cancer Center, Miller School of Medicine, University of Miami, 1600 NW 10th Avenue, Miami, FL 33136. Phone and fax: (305) 243-6308. E-mail: [email protected]. † Present address: Department of Cancer Biology, 233 S. 10th St., Philadelphia, PA 19107. ‡ Present address: Institut Cochin, INSERM U1016, CNRS UMR8104, Faculte´ Paris Descartes, 24 rue du Faubourg SaintJacques, 75014 Paris, France. ¶ R.V.-D.C. and F.M.K. contributed equally. 䌤 Published ahead of print on 18 January 2011. 1396

VOL. 31, 2011

Notch MULTIMERIZATION AND TRANSCRIPTIONAL COMPLEX ASSEMBLY

loading of NICD and Maml1 onto CSL to form the transcriptional activation complex on DNA. These data reveal the molecular events of a stepwise assembly that lead to the formation of the Notch transcriptional activation complex. MATERIALS AND METHODS Cell culture. H1299 and 293T cells were propagated in Dulbecco’s modified Eagle medium (Life Technologies) supplemented with 10% fetal bovine serum, 2 mM L-glutamine, 100 U of penicillin per ml, and 100 ␮g of streptomycin per ml (Life Technologies) under standard conditions. Spodoptera frugiperda IPLB-Sf21 (Sf21) cells were maintained in Sf-900 II SFM medium (GibcoBRL) supplemented with 100 U penicillin per ml and 100 ␮g streptomycin per ml (Life Technologies). Generation of baculovirus. For recombinant-baculovirus production, bacmids containing cDNA of NICD, Maml1, and CSL were generated following the Bac-to-Bac baculovirus expression system (Invitrogen). The bacmids were transfected into Sf21 cells using Cellfectin reagent (Invitrogen). Recombinant baculovirus was amplified, and the titer was determined, following the Bac-to-Bac baculovirus expression system (Invitrogen). The expression of the recombinant proteins was confirmed by Western blot analysis of cell lysates. Plasmid constructs. The NICD deletion constructs used in this study were described previously (15, 16). NICD-Flag wild-type (wt) and RAM point mutant constructs were generated by subcloning NICD into pBabePuro Flag. NICD-Flag was then subcloned into pCDNA3.1 (Invitrogen). NICD⌬2203-2216-Myc was generated using Quick change Site-Directed Mutagenesis (Stratagene), NICD-Myc in pCDNA3.1 as a template, and a combination of the following primers: 5⬘ CTC TCGCCCGTGGACCCGCCACTGCTGCCC 3⬘ and 5⬘ GGGCAGCAGTGGC GGGTCCACGGGCGAGAG 3⬘. pBabePuro Flag was generated by digesting pBabePuro with EcoRI and XhoI and ligating the following primers: 5⬘-AATT CAGAAGACTCGAGGATTATAAGGACGACGACGATAAGTAGTAA G-3⬘ and 5⬘-TGCACTTACTACTTATCGTCGTCGTCCTTATAATCCTCGA GTCTTCTG-3⬘. For baculovirus expression, CSL, NICD, Skip, and Maml1 were cloned into the baculovirus vector pFastBac1 3⬘hexaHis, pFastBac Hta (GibcoBRL), or pFastBac1 3⬘Flag. The pFastBac1 3⬘Flag vector was constructed by digesting the pFastBac1 vector (GibcoBRL) with EcoRI and XhoI and ligating the primers used to generate the pBabePuro Flag construct into pFastBac1. For bacterial expression, the RAM domain of NICD (amino acids [aa] 1759 to 1847) was subcloned into the pGEX4T-2 vector (Amersham Pharmacia Biotech) with BamHI and XhoI to generate glutathione S-transferase (GST) fusion proteins. Site-directed mutants were generated using Quick Change Site-Directed Mutagenesis (Stratagene) and combinations of the following primers: 5⬘ GACGA CCAGACAAACCACCAGCAGTGGACTCAGCAG 3⬘ and 5⬘ CTGCTGAGT CCACTGCTGGTGGTTTGTCTGGTCGTC 3⬘ for Mut A. Mut B was generated by two rounds of mutagenesis using the following primers: 5⬘ CAGA ATGAGTGGGGGCAGCAGCAGCTGGAGACCAAGAAGTTC 3⬘ and 5⬘ GA ACTTCTTGGTCTCCAGCTGCTGCTGCCCCCACTCATTCTG 3⬘ for the first mutagenesis and 5⬘ CAGCTGGAGACCCAGCAGTTCCAGTTCCAGCA GCCCGTGGTTCTG 3⬘ and 5⬘ CAGAACCACGGGCTGCTGGAACTGGAA CTGCTGGGTCTCCAGCTG 3⬘ for the second round. Nucleotide changes (underlined) were confirmed by DNA sequencing. The residues selected for mutagenesis were changed to asparagine or glutamine, depending on their chemical nature. 293T transfections. 293T cells (4 ⫻ 106) were plated on 10-cm culture dishes and transfected with a total of 8 ␮g of cDNAs of different Notch constructs, Maml1, Skip, and CSL, using Lipofectamine 2000 (Invitrogen). Expression of GST fusion proteins. GST-NICD fusion proteins were expressed in the bacterial strain BL21 (Stratagene). Cells were grown at 37°C in LB medium with 100 ␮g/ml ampicillin and induced for 3 h at 30°C with 1.0 mM IPTG (isopropyl-␤-D-thiogalactopyranoside) when the cultures reached an optical density at 600 nm (OD600) of 0.6. Bacteria were collected by centrifugation at 5,000 rpm for 10 min at 4°C. Bacterial pellets were resuspended in 30 ml of ice-cold phosphate-buffered saline (PBS) (150 mM NaCl, 1.47 mM KH2PO4, 0.8 mM Na2HPO4) containing 2 M urea and disrupted by sonication in the presence of 1% Triton X-100. The disrupted cells were centrifuged at 14,000 ⫻ g for 20 min at 4°C. To the supernatants, glutathione-agarose beads (Sigma) were added and incubated at 4°C for 4 h with agitation. The beads were washed with PBS and resuspended in an equal volume of PBS with 0.5 mM dithiothreitol (DTT) and protease inhibitors (2 mM Pefabloc [Roche], 5 ␮g/ml leupeptin [Roche], 1 ␮g/ml aprotinin [Roche]), aliquoted, and stored at ⫺80°C. Baculovirus protein expression. Recombinant Flag-tagged proteins were expressed as follows. Sf21 cells (3 ⫻ 107) in 150-mm plates were infected at a

1397

multiplicity of infection (MOI) of 2 to 5 and harvested 48 to 72 h after infection. Cells were resuspended in high-salt lysis buffer (500 mM NaCl, 40 mM Tris-HCl, pH 7.4, 1 mM EDTA, 10% glycerol, 0.02% NP-40, 1 mM DTT, and protease inhibitors) and lysed in a glass Dounce homogenizer with 30 strokes of a pestle over 30 min on ice. The lysates were centrifuged at 100,000 ⫻ g at 4°C for 30 min. An equal volume of dilution buffer (40 mM Tris-HCl, pH 7.4, 10% glycerol, 0.02% NP-40, 1 mM DTT, and protease inhibitors) was added to the supernatant. Proteins were incubated with 100 to 200 ␮l of M2 ␣-Flag agarose beads (Sigma) for 3 to 5 h at 4°C with rocking. The beads were washed in lysis buffer (150 mM NaCl, 40 mM Tris-HCl, pH 7.4, 1 mM EDTA, 10% glycerol, 0.02% NP-40, 1 mM DTT, and protease inhibitors). Proteins were eluted with elution buffer (lysis buffer supplemented with 0.4 mg/ml Flag peptide [Sigma]) and further dialyzed in storage buffer [100 mM KCl, 20 mM HEPES, pH 7.9, 20% glycerol, and 1 mM DTT), aliquoted, and stored at ⫺80°C. Coimmunoprecipitation assays and Western blots. Forty-eight hours posttransfection, total cell lysate was obtained using NP-40 lysis buffer (150 mM NaCl, 50 mM HEPES, pH 7.4, 1.5 mM EDTA, 10% glycerol, 1% NP-40, supplemented with 0.5 mM DTT, 50 mM NaF, 0.5 mM vanadate, and protease inhibitors). Cell lysates were centrifuged at 100,000 ⫻ g for 30 min at 4°C. NICD was immunoprecipitated by adding M2 beads at 4°C for 3 to 6 h. After extensive washing with the NP-40 lysis buffer, the samples were resolved by SDS-PAGE and transferred onto Immobilon-P membranes (Millipore) for Western blot analysis. Western blot analysis was performed using the indicated antibodies and an enhanced chemiluminescence (ECL) protocol (16). For the dual-affinity purification of NICD multimer complexes, 48 h posttransfection, total cell lysate was obtained using NP-40 lysis buffer. Cell lysates were centrifuged at 100,000 ⫻ g for 30 min at 4°C. NICD1767-70A-Flag was immunoprecipitated by adding M2 beads at 4°C for 3 to 6 h. After extensive washing with the NP-40 lysis buffer, NICD1767-70A-Flag complexes were eluted from the M2 resin using 1 mg/ml Flag peptide in the NP-40 lysis buffer. The eluted proteins were then dialyzed in NP-40 lysis buffer for 3 to 6 h at 4°C. Next, the NICD1767-70AFlag complexes were subjected to Ni-nitrilotriacetic acid (NTA) resin (Invitrogen) for 3 to 6 h at 4°C. NICD1771-74A⌬2444-His complexes were eluted from the Ni-NTA resin with 100 mM imidazole in the NP-40 lysis buffer. These mutants were used in the experiment due to their increased stability as multimers in cells. Samples were analyzed on a 4 to 12% NuPage Novex Bis-Tris gel, followed by colloidal blue staining following the protocol of the manufacturer (Invitrogen) or by Western blotting using the specified antibodies. For purified proteins, 100 to 400 ng (each) of NICD, Maml1, Skip, and CSL containing Flag epitope tags were mixed together and incubated in modified in vitro complex buffer (150 mM NaCl, 20 mM HEPES, pH 7.9, 20% glycerol, 0.5% NP-40, supplemented with 0.5 mM DTT and protease inhibitors). Proteins were immunoprecipitated with either anti-Mastermind antibody, anti-CSL antibody, anti-Skip (C-15; Santa Cruz Biotechnology), or a nonspecific rabbit immunoglobulin G (IgG) control antibody for 1 h at 4°C, followed by the addition of protein A-Sepharose beads (Sigma) for 60 min. After extensive washing with the modified in vitro complex buffer, the samples were resolved by SDS-PAGE and transferred onto Immobilon-P membranes (Millipore) for Western blot analysis. Western blot analysis was performed as described previously. Luciferase reporter assay. For luciferase assays, 6.0 ⫻ 104 H1299 cells were seeded in six-well plates. The cells were transfected with a total of 2 ␮g of DNA. Cotransfections consisted of 0.4 ␮g of 8⫻ CSL luciferase reporter vector as previously described (16), 0.4 ␮g of simian virus 40 (SV40) ␤-galactosidase (␤-Gal) (internal transfection control plasmid; Clontech), and 0.2 ␮g of the indicated NICD expression plasmid. The concentration of DNA was brought to 2 ␮g with either pCDNA or pBabe. For titration assays, the titrations ranged from 0.2 ␮g to 0.8 ␮g of the indicated NICD construct. Cells were transfected with 6 ␮l of Fugene 6 (Roche) in a total volume of 1 ml of Opti-MEM and 1 ml of Dulbecco’s modified Eagle medium. At 48 h posttransfection, lysates were prepared and luciferase activities were determined in a Xylux Femtomaster FB 12 luminometer according to the manufacturer’s suggested protocol (Promega). Luciferase values were corrected for transfection efficiency by normalizing them to ␤-galactosidase activity. Gel filtration and TCA precipitation. Recombinant proteins were fractionated using fast protein liquid chromatography (FPLC) (Pharmacia) on a Superose 12 HR column equilibrated in column buffer B (150 mM NaCl, 20 mM HEPES, pH 7.9, 1 mM DTT, and 10% glycerol). Samples were trichloroacetic acid (TCA) precipitated as previously described (16) and analyzed by SDS-PAGE and Western blotting using specific antibodies.

1398

VASQUEZ-DEL CARPIO ET AL.

RESULTS ICD

N forms multimers in cells. Baculovirus-expressed recombinant NICD, when analyzed by size exclusion chromatography on a Superose 6 column, displays an elution profile similar to that of the P complex observed in cells transformed with NICD (16) (Fig. 1D). Since this recombinant NICD migrates with a size much greater than its molecular mass (⬃85 kDa) and the purified fraction contains no other detectable protein, we reasoned that the P complex might be comprised of Notch multimers. To determine if NICD is a multimer in cells, coimmunoprecipitation assays were performed in 293T cells cotransfected with an NICD construct carrying a Flag epitope tag (NICDF) and an NICD⌬2444 construct carrying a Myc epitope tag (NICDM) (Fig. 1B). NICD⌬2444-Myc was utilized to distinguish between the two NICD proteins because of its migration as a smaller species on SDS-PAGE. When NICDF was immunoprecipitated with anti-Flag antibody, NICDM was coimmunoprecipitated (Fig. 1A, lane 5). In agreement with this result, when a reciprocal coimmunoprecipitation was performed using anti-Myc (9E10) antibody, NICDF was coimmunoprecipitated with NICDM (Fig. 1A, lane 6). As controls, when NICDF was not present in the lysate, NICDM did not immunoprecipitate with Flag antibody, and NICDF did not immunoprecipitate with anti-Myc antibody when NICDM was not present in the lysate (Fig. 1A, lanes 7 and 8). This result was also observed in H1299 cells (see Fig. 1 posted at http: //biomed.miami.edu/?p⫽886&s⫽154). These data indicate that the P complex is formed by NICD multimers and that these multimers occur in the absence of Maml1 and CSL. To further demonstrate that NICD forms multimers in cells, we performed two sequential affinity purifications from lysates of 293T cells cotransfected with an NICD construct carrying a Flag epitope tag (NICDF) and an NICD⌬2444 construct carrying dual Myc and His epitope tags (NICDH). Proteins were immunoprecipitated first using M2 beads and eluted with Flag peptide to end with a mix of Flag-tagged complexes (NICDFNICDF and NICDF-NICDH) (Fig. 1C, II, lane E1). The NICDFNICDH complexes were subsequently affinity purified from E1 using Ni-NTA resin, which interacts with His tag proteins, and eluted with imidazole. After the sequential affinity purifications were performed, NICDF-NICDH complexes were detected in equal ratios based on colloidal Coomassie blue staining (Fig. 1C, II, lane E2). Double affinity purification was also analyzed by Western blotting (Fig. 1C, II). Importantly, no other proteins were observed. Collectively, these data indicate that NICD forms multimers in cells. NICD multimerizes through the N-terminal domain and a region C terminal to the ankyrin repeats. To further analyze the nature of the Notch multimer complex, baculovirus-expressed recombinant NICD wt and mutant proteins were fractionated by size exclusion chromatography on a Superose 12 column. Similar to what was described above, elution volumes from gel filtration suggest a multimeric state of NICD (Fig. 1D). In contrast, NICD⌬2202, a mutant with a deletion of the C-terminal region of the protein but retaining an intact ankyrin repeat domain, showed an elution profile expected for its predicted molecular mass (⬃48 kDa) (Fig. 1D). These results show the importance of the C-terminal region for multimerization. To determine if the NICD⌬2202 mutant protein, which does

MOL. CELL. BIOL.

not form homomultimers, can form heteromultimers in the presence of NICD, 293T cells were cotransfected with NICD⌬2202 carrying either Flag or Myc epitope tags and NICD (Fig. 1E). NICD was coimmunoprecipitated when NICD⌬2202Flag was pulled down (Fig. 1E, lane 3). In contrast NICD⌬2202Myc was absent when NICD⌬2202-Flag was immunoprecipitated (Fig. 1E, lane 4), demonstrating that NICD⌬2202 can form hetero- but not homomultimers. Furthermore the fact that NICD⌬2202 can form heteromultimers but not homomultimers implies that the N-terminal region of Notch is required for multimerization. These results demonstrate that, in addition to the C-terminal region of NICD, a second site located toward the N-terminal region of the protein is involved in multimerization. To further study the interaction site, deletion mapping of NICD was performed. Since the presence of the ankyrin domain was previously seen as not sufficient for multimerization, we decided to test if a region N terminal to the ankyrin repeats, the RAM domain (aa 1759 to 1847), is the second site mediating multimerization of NICD. To assess this, GST pulldown assays were performed using the RAM domain and lysates from 293T cells transfected with NICD C-terminal truncations (Fig. 2A, left, lanes 1 to 5). The RAM domain interacted with NICD C-terminal truncations, which contained amino acids up to 2240 (Fig. 2A, lanes 2 to 5), but not with a truncation at residue 2202 (Fig. 2A, lane 6). Sequence analysis of that region shows conservation between the four Notch receptors that noticeably decreases after residue 2216 (Fig. 2A). Accordingly, when an internal deletion of the region between amino acids 2203 and 2216 was tested, the RAM domain no longer interacted with NICD (Fig. 2A, right, lane 7). These data indicate that the RAM domain interacts with the C-terminal region of NICD localized between amino acids 2203 and 2216. In order to further study the N-terminal interaction site, a deletion mutant lacking the last 28 residues of the RAM domain (⌬1819) was designed based on sequence conservation of that region. The removal of the last 28 amino acids drastically affected the binding of the RAM domain to NICD (Fig. 2B, left, lane 3). The region presents a long stretch of charged amino acids that are conserved from NICD1 to NICD3 (Fig. 2C). Site-directed mutagenesis of these residues to a polar noncharged amino acid, like glutamine, showed a profound loss of binding compared to the wild-type RAM (Fig. 2B, right, lanes 5 to 7). As a control, GST alone did not present any detectable pulldown (Fig. 2B, right, lane 8). Further analysis of this interaction by increasing the ionic strength showed that binding of the RAM domain to the C-terminal region can be mostly inhibited above 250 mM NaCl (Fig. 2D). Thus, the presence of these two multimerization sites (an Nand a C-terminal multimerization site) allows NICD to associate in an antiparallel fashion (Fig. 2E) and explains why the NICD⌬2202 mutant cannot form stable homomultimers but can form heteromultimers. Transcriptionally deficient NICD mutants can be rescued through multimerization. We reasoned that if heteromultimers could be formed, it might then be feasible to rescue the activity of an NICD mutant that cannot efficiently incorporate into a complex with Maml1 and CSL by forming a heteromultimer with a second NICD mutant that harbors a different mutation. To test this hypothesis, two mutants, NICD⌬2105 and

VOL. 31, 2011


1399

FIG. 1. NICD forms multimers in cells. (A) 293T cells were cotransfected with NICD-Flag and NICD⌬2444-Myc. The whole-cell lysate was immunoprecipitated (IP) with anti-Flag antibody to immunoprecipitate NICD, and Western blotting (WB) was performed with anti-Flag and anti-Myc (9E10) antibodies to detect NICD and NICD⌬2444, respectively. (B) Schematic representation of NICD-Flag and NICD⌬2444-Myc tagged proteins. (C) 293T cells were cotransfected with vectors expressing NICD carrying a Flag tag (NICDF) and NICD⌬2444 carrying a His tag (NICDH). (I) The lysate was subjected to two affinity purification procedures. First, the lysate was subjected to M2 resin, which interacts with Flag tag proteins. Proteins that did not interact with the M2 resin were labeled as flowthrough 1 (FT 1). NICDF complexes were eluted from the M2 resin using Flag peptide (E1). Next, the NICDF complexes were subjected to Ni-NTA resin, which interacts with His tag proteins. Proteins that did not interact with the Ni-NTA resin were labeled as flowthrough 2 (FT 2). NICDH complexes were eluted from the Ni-NTA resin (E2) with imidazole. (II) Samples were analyzed by SDS-PAGE on a 4 to 12% NuPage Novex Bis-Tris gel and stained with colloidal blue or analyzed by Western blotting. (D) Purified NICD and NICD⌬2202 were analyzed by size exclusion chromatography on a Superose 12 column. Fractions were collected, TCA precipitated, and detected by SDS-PAGE and Western blotting using anti-Flag or Myc (9E10) antibody. (E) Flag-tagged NICD⌬2202 and Myc-tagged NICD and NICD⌬2202 were cotransfected into 293T cells. Forty-eight hours posttransfection, whole-cell lysate was immunoprecipitated with anti-Flag antibody to immunoprecipitate NICD⌬2202. The proteins were analyzed by SDS-PAGE and Western blotting using anti-Myc and anti-Flag antibodies.

1400


MOL. CELL. BIOL.

FIG. 2. The N-terminal domain interacts with the C-terminal region of NICD. (A) (Left) Myc-tagged NICD C-terminal deletion constructs were transfected into 293T cells. Whole-cell lysates were subjected to GST pulldown assays with GST-RAM (amino acids 1759 to 1847). Coprecipitated samples were run on an SDS-PAGE gel and Western blotted using anti-Myc antibody (9E10). (Right, top) 293T cell lysates that were transfected with NICD wt or NICD containing an internal deletion of 14 residues (NICD⌬2203-2216) expressing a Myc tag were incubated with GST-RAM. After pulldown, samples were analyzed as described above. (Bottom) Sequence alignment of the predicted C-terminal interaction site of NICD (NICD1). The alignment shows sequence conservation in Homo sapiens (Hs) Notch sequences. The boldface letters denote similarity, and the boxes denote identity in three or more sequences. (B) (Left) 293T cell lysates that were transfected with NICD⌬2278 were incubated for pulldown with either GST-RAM or a GST-RAM deletion lacking the last 27 residues (GST-RAM⌬1819). Samples were further analyzed by SDS-PAGE and Western blotting. (Right) Similar to what was described above, 293T cell lysates were incubated with GST-RAM wt or GST-RAM constructs harboring point mutations (mut A and B). Amino acids that are marked by a dot above the sequence in panel C were mutated to glutamine or asparagine. Samples were analyzed by SDS-PAGE and Western blotting. (C) Alignment of the H. sapiens Notch sequences showing the predicted N-terminal interaction site of NICD. The boldface letters denote similarity, and the boxes denote identity in three or more sequences. (D) 293T cell lysate transfected with NICD⌬2278 were incubated with GST-RAM. After incubation, samples were split and beads were washed using different NaCl concentrations. Samples bound to the beads

VOL. 31, 2011


1401

FIG. 3. NICD multimerization rescues transcription-deficient mutants. (A) H1299 cells were transfected with 0.2 ␮g of each individual construct. For titrations, H1299 cells were transfected with 0.2 ␮g of NICD⌬2105 or NICD⌬R⌬2202 and increasing amounts of NICD⌬2105 or NICD⌬R⌬2202 DNA (0.2 to 0.8 ␮g). Forty-eight hours posttransfection, the cells were lysed, and luciferase assays were performed in triplicate. The results are shown in relative luciferase units (RLU). The error bars represent the standard deviations from each set. (B) 293T cells were transfected with NICD-Myc, NICD⌬R⌬2202-Flag, and NICD⌬2105-Myc tagged plasmid constructs. Whole-cell lysates were immunoprecipitated with anti-Flag antibody to immunoprecipitate NICD⌬R⌬2202. Samples were then analyzed by SDS-PAGE and Western blotting using anti-Myc and anti-Flag antibodies. (C) H1299 cells were transfected with 0.2 ␮g of each plasmid. For titrations, increasing amounts of NICD⌬2202⌬2105 or NICD⌬2105 (0.2 to 0.8 ␮g) were used relative to a fixed amount of plasmid (0.2 ␮g). Forty-eight hours posttransfection, the cells were lysed and luciferase assays were performed in triplicate. The results are shown in relative luciferase units. The error bars represent the standard deviations from each set.

NICD⌬R⌬2202, which are both transcriptionally inactive in reporter assays (16), were cotransfected together and assayed for reporter activity. NICD⌬R⌬2202 is inactive in reporter assays because its RAM domain, which is the high-affinity binding site for CSL, is deleted (data not shown). However, the reporter activity of NICD⌬R⌬2202 can be rescued by Maml1, indicating that this mutant is still competent to activate transcription (15, 16). The mutant weakly associates with Maml1 and CSL and transforms cells less efficiently than NICD, indicating that NICD⌬R⌬2202 retains NICD function in cells (15, 16). NICD⌬2105 has a 10-amino-acid deletion in the seventh ankyrin repeat and forms a stable but inactive complex with Maml1 and CSL (data not shown). In contrast to NICD⌬R⌬2202, the transcriptional activity of NICD⌬2105 cannot be rescued by Maml1 in reporter

assays and NICD⌬2105 fails to transform cells (15, 16). Figure 3 demonstrates that heteromultimerization can rescue activity. Specifically, when a constant level of NICD⌬R⌬2202 was cotransfected with increasing amounts of NICD⌬2105, there was a dosedependent increase in reporter activity (Fig. 3A). A similar result was obtained when a constant level of NICD⌬2105 was cotransfected with increasing amounts of NICD⌬R⌬2202. As a control, when either mutant alone was assayed for activity, the reporter activities were similar to background (Fig. 3A). These data indicate that cotransfecting NICD⌬R⌬2202 with NICD⌬2105 rescues reporter activity through heteromultimerization. To test if the rescue in activity was due to the NICD mutant proteins associating with one another, coimmunoprecipitation experiments were performed in 293T cells cotransfected with

were analyzed by SDS-PAGE and Western blotting. (E) Schematic representation of NICD showing the distinct domains described to date and multimerization sites. Below, a model for NICD multimerization is illustrated. The ankyrin repeats and the RAM domain are shown in gray and orange, respectively. ter, terminus.

1402


MOL. CELL. BIOL.

FIG. 4. NICD is monomeric in the active complex. (A) NICD-Flag was cotransfected into 293T cells with NICD⌬2444-Myc in the presence or absence of Maml1 and CSL. Forty-eight hours posttransfection, the whole-cell lysate was immunoprecipitated with anti-Flag antibody and detected by SDS-PAGE and Western blotting. Both NICD and NICD⌬2444 were detected with anti-Notch antibody (927), Maml1 was detected using anti-Myc antibody, and CSL was detected using antihemagglutinin (anti-HA) antibody. (B) (I) Purified NICD (2 pmol) and Maml1 (4 pmol) were incubated together and immunoprecipitated with anti-Maml1 antibody. (II) Purified Maml1 (4 pmol) and CSL (2 pmol) were incubated together and immunoprecipitated with anti-CSL antibody. (III) Purified NICD (2 pmol), Maml1 (4 pmol), and CSL (2 pmol) were incubated together and immunoprecipitated with anti-Maml1 antibody. All samples were analyzed by SDS-PAGE and Western blotting using an anti-Flag antibody.

NICD⌬R⌬2202 carrying a Flag tag and NICD⌬2105 carrying a Myc tag (Fig. 3B). When immunoprecipitations were performed for the Flag epitope tag, NICD⌬2105 coimmunoprecipitated with NICD⌬R⌬2202 (Fig. 3B, right, lane 3). As a control for nonspecific binding, NICD-Myc alone did not immunoprecipitate with anti-Flag antibody (Fig. 3B, right, lane 4). These results demonstrate that NICD⌬2105 and NICD⌬R⌬2202 can form heteromultimers and that this association rescues transcriptional activity. It has been suggested that expression of the RAM domain in trans can increase NICD activity. This activity has been purported to be due to binding of RAM to CSL and displacing corepressors (7). Alternatively, Friedmann et al. proposed that binding of a RAM peptide facilitates the association between the NICD ankyrin domain, Maml1, and CSL (12). To determine if the rescue observed was simply due to an increase in the RAM domain present in the cell or to heteromultimerization, rescue reporter assays were performed using NICD⌬2202⌬2105 (Fig. 3C). When NICD⌬2202, which does not self-associate and contains an intact RAM domain, was cotransfected with increasing amounts of NICD⌬2202⌬2105, there was no increase in reporter activity (Fig. 3C). However, when NICD⌬2202 was cotransfected with increasing amounts of NICD⌬2105, a dosedependent increase in activity was observed (Fig. 3C). These data indicate that the rescue observed was not due to overex-

pressing the RAM domain in trans but was in fact due to multimerization. The active complex contains NICD monomers. We demonstrated that NICD forms multimers when not in complex with Maml1 and CSL. However, in the active complex, NICD polypeptides were shown to be monomeric (26, 36). To investigate the mechanism of complex formation, coimmunoprecipitation assays were performed with NICDF and NICDM in the presence or absence of Maml1 and CSL. In the absence of Maml1 and CSL, NICDM coimmunoprecipitated with NICDF (Fig. 4A, lane 5). However, in the presence of Maml1 and CSL, NICDM did not coimmunoprecipitate with NICDF (Fig. 4A, lane 8). These data suggest that NICD might convert from a multimer to a monomer in the active complex. This observation and suggestion are consistent with a complex seen in the crystal structure, which shows one molecule of NICD is associated with Maml1 and CSL (26, 36). To further dissect the mechanism of complex formation, coimmunoprecipitation experiments were performed with NICDF and NICDM in the presence of either Maml1 or CSL. When CSL was cotransfected with NICDF and NICDM, NICD no longer self-associated (Fig. 4A, lane 7). These data indicate that the interaction between CSL and NICD plays an important role in converting NICD multimers to monomers. Interestingly,

VOL. 31, 2011


1403

FIG. 5. A functional ⌽W⌽P motif in NICD is required for efficient multimer-monomer conversion and further active-complex assembly. (A) Flag-tagged NICD and NICD RAM point mutants were cotransfected into 293T cells with NICD⌬R⌬2444 Myc tag in either the presence or absence of CSL. Forty-eight hours posttransfection, whole-cell lysates were immunoprecipitated with anti-Flag antibody and subsequently analyzed by SDS-PAGE and Western blotting. Anti-Notch antibody was used to detect NICD and NICD RAM point mutants, anti-Myc antibody to detect NICD⌬R⌬2444-Myc, and anti-HA antibody to detect CSL. (B) Schematic representation of NICD⌬R⌬2444-Myc and NICD-Flag tagged proteins. An alignment of NICD RAM point mutations used in the experiment is also shown. The boldface denotes amino acid conservation. (C) H1299 cells were transfected with 0.2 ␮g of each construct. For titrations, H1299 cells were transfected with 0.2 ␮g of NICD⌬R⌬2444 or NICD1767-70A cDNA and increasing amounts of NICD⌬2105, NICD⌬R⌬2444, or NICD1767-70A cDNA (0.2 to 0.8 ␮g). Forty-eight hours posttransfection, the cells were lysed and luciferase assays were performed in triplicate. The results are shown in relative luciferase units. The error bars represent the standard deviations from each set.

a similar result was observed when Maml1 was cotransfected with NICDF and NICDM (Fig. 4A, lane 6). However, 293T cells express CSL, but not Maml1, so the conversion from NICD multimers to monomers observed when Maml1 is introduced is likely due to the formation of a complex between NICD, Maml1, and endogenous CSL (16). To determine if Maml1 directly interacts with NICD to mediate the conversion from multimers to monomers, coimmunoprecipitation experiments were performed using purified NICD, Maml1, and CSL proteins carrying Flag epitope tags (Fig. 4B). The purified proteins were incubated together in different combinations and immunoprecipitated using the indicated antibodies. When Maml1 and NICD were incubated together, NICD did not coimmunoprecipitate with Maml1 (Fig. 4B, I, lane 3). A similar result was observed when Maml1 was incubated with CSL (Fig. 4B, II, lane 3). However, when all three proteins were incubated together, an interaction between NICD, Maml1, and CSL was detected (Fig. 4B, III, lane 2). These data indicate that Maml1 does not directly convert

Notch multimers to monomers. Instead, Maml1 either facilitates or stabilizes the interaction between NICD and CSL, and it is the combination of these interactions that converts NICD multimers to monomers in the presence of Maml1. Active complex formation requires an intact RAM ⌽W⌽P motif for the conversion of NICD multimers into monomers. Since the interaction between NICD and CSL effectively converts NICD multimers to monomers (Fig. 4A, lane 7), we reasoned that inhibiting the interaction between NICD and CSL would prevent the conversion of multimers to monomers. To test this, assays were performed using NICD RAM point mutants that are defective in binding CSL, in combination with NICD⌬R⌬2444, which has the RAM domain deleted and does not bind CSL. NICD1767-70A and NICD1771-74A are both mutants that harbor point mutations in the ⌽W⌽P motif of the RAM domain, which abrogates NICD binding to CSL (Fig. 5B) (7). When CSL was absent, all of the NICD RAM point mutant proteins coimmunoprecipitated efficiently with NICD⌬R⌬2444Myc (Fig. 5A, right, lanes 9 to 12). However, in the presence of

1404


CSL, only the two NICD mutants that cannot interact with CSL (NICD1767-70A and NICD1771-74A) remained associated with NICD⌬R⌬2444-Myc (Fig. 5A, right, lanes 14 and 15). In contrast, the RAM domain point mutant that still interacts with CSL, NICD1775-78A, did not coimmunoprecipitate with NICD⌬R⌬2444Myc in the presence of CSL (Fig. 5A, right, lane 16). These data indicate that binding of CSL to the ⌽W⌽P motif of NICD is required for the conversion of NICD multimers to monomers. Since the NICD RAM ⌽W⌽P motif point mutants remain multimers in the presence of CSL, it is possible that they could be inactive in reporter assays because they cannot be converted from multimers to monomers. To test whether these RAM mutants are active in reporter assays, luciferase assays similar to those in Fig. 3A were performed. When the NICD mutants were transfected alone, no reporter activity was detected. However, when NICD⌬2105 was cotransfected with NICD1767-70A, an increase in activity was observed, indicating that NICD⌬2105 is interacting with NICD1767-70A and providing a CSL binding site to foster complex formation (Fig. 5C). In contrast, no increase in activity was observed when a constant amount of NICD1767-70A was cotransfected with increasing amounts of NICD⌬R⌬2444, which does not contain a RAM domain and cannot complement a nonfunctional RAM point mutant (Fig. 5C). A similar result was obtained when a constant amount of NICD⌬R⌬2444 was cotransfected with increasing amounts of NICD1767-70A. These results indicate that a functional ⌽W⌽P motif is needed for the conversion of Notch multimers to monomers. In addition, these results indicate there is a separation between the CSL binding site (⌽W⌽P) and the N-terminal multimerization region in the RAM domain of NICD. Skip binds NICD multimers and recruits Maml1. Previous reports have demonstrated that Skip interacts with NICD and increases transcription of the Hes promoter when in the context of NICD and Maml1 (13, 39). However, the mechanistic details of how Skip functions with NICD have not been described. Since NICD does not stably interact with Maml1 (Fig. 4B, I, lane 3, and 6A, lane 1), we reasoned that Skip might function to recruit Maml1 to the active complex. In order to address this, different combinations of purified recombinant proteins were incubated together and coimmunoprecipitated (Fig. 6). As was previously described, an interaction between NICD and Skip is observed by coimmunoprecipitation (Fig. 6A, lane 2), but an interaction of NICD with Maml1 is detected only in the presence of Skip (Fig. 6A, lanes 3 and 4). This result shows that Skip is required for Maml1 to interact with NICD to form a preactivation complex. Furthermore, the interaction between NICD, Maml1, and Skip is reminiscent of the interaction between NICD, Maml1, and CSL in that Maml1 requires a bridging interaction to associate with Notch (Fig. 6A, lanes 5 and 6). Skip was reported to bind to the ankyrin repeat domain of NICD (39). The ankyrin repeats of NICD are also one of the two sites for interaction with CSL (26, 36). To determine if CSL can compete with Skip for the interaction with NICD and resolve the preactivation complex, recombinant proteins were added in different combinations and subjected to coimmunoprecipitation. As CSL was added to the NICD-Skip-Maml1 mixture, we observed a dose-dependent decrease in Skip binding with a concomitant increase in CSL binding (Fig. 6B, lines 4 to 8). These data indicate that CSL displaces Skip from the

MOL. CELL. BIOL.

interaction with NICD and Maml1, resulting in the transition from the preactivation complex to the active transcriptional complex. To determine if a failure of NICD to multimerize results in an inability to form the preactivation complex, coimmunoprecipitation experiments were performed using recombinant NICD, NICD⌬2202, Maml1, and Skip (Fig. 6C). NICD⌬2202 was utilized because it does not form multimers but still retains the binding region for Skip (Fig. 2) (39). Similar to what was observed with NICD, Maml1 does not associate with NICD⌬2202 in the absence of Skip (Fig. 6C, lane 1). Surprisingly, and in contrast to NICD, NICD⌬2202 failed to interact with Maml1 in the presence of Skip (Fig. 6C, lane 3), indicating that Notch multimerization is necessary to form a complex between NICD, Skip, and Maml1. Moreover, the addition of NICD to the mixture can rescue the interaction between Maml1, Skip, and NICD⌬2202 (Fig. 6C, lane 4). On the other hand, interaction of Skip with NICD⌬2202 monomers is considerably lower than the interaction of Skip with NICD (Fig. 6C, lane 2, and A, lane 2). These data indicate that the association between NICD and Skip occurs after the formation of NICD multimers and serves as a docking site for Maml1. To further demonstrate that Skip associates with NICD multimers and provides a docking site for Maml1, coimmunoprecipitation experiments were performed in 293T cells cotransfected with NICD1771-74A carrying a Flag tag, NICD⌬R⌬2444 carrying a Myc tag, Skip, and Maml1 (Fig. 6D). NICD1771-74A and NICD⌬R⌬2444 mutants were used because their heteromultimers cannot be efficiently disrupted by the presence of CSL, thus making a transient intermediate easier to detect. A SkipNICD multimer complex is detected when Skip is present (Fig. 6D, lane 7). In the presence of Skip, Maml1 interacted with the NICD1771-74A-NICD⌬R⌬2444 complex (Fig. 6D, lane 9). In the presence of only Maml1, NICD1771-74A and NICD⌬R⌬2444 coimmunoprecipitation was less efficient (Fig. 6D, lane 8). The finding that Skip recruits Maml1 to NICD multimers indicates that Skip might interact with Maml1 in cells. In addition, it indicates that Maml1 does not resolve NICD multimers and that this activity is CSL dependent. As a control to determine if Skip interacts with Maml1, coimmunoprecipitation experiments were performed in 293T cells cotransfected with Maml1 and Skip (Fig. 6E). When Maml1 was immunoprecipitated, Skip was not detected, indicating that in the absence of NICD, Skip cannot interact with Maml1 (Fig. 6E, lane 3). Taken together, these data indicate that Skip preferentially interacts with NICD multimers and that this interaction recruits Maml1 to form a preactivation complex. This preactivation complex then interacts with CSL to form a transcriptional activation complex on DNA (Fig. 7). DISCUSSION Previously, we demonstrated that NICD forms two distinct protein complexes in cells (16). One complex is predominately localized in the nucleus and is composed of NICD, Maml1, and CSL. This complex is thought to be the transcriptionally active form of Notch on DNA. In addition, a smaller Notch-containing complex is also detected in cells. This complex is primarily localized in the cytoplasm and does not contain either Maml1 or CSL. The nature and function of this complex has remained

VOL. 31, 2011


1405

FIG. 6. Skip interacts with NICD multimers and Maml1 to form a preactivation complex. (A) Purified recombinant NICD (8 pmol), Maml1 (4 pmol), Skip (4 pmol), or CSL (4 pmol) Flag-tagged proteins were incubated in different combinations on ice for 30 min. The reaction mixtures were then immunoprecipitated using anti-Maml1 or anti-Skip antibodies. Samples were subsequently analyzed by SDS-PAGE and Western blotting using the specified antibodies. (B) Reactions were performed in a similar manner to those for panel A with the difference that CSL was added in increasing quantities to the NICD-Maml1-Skip mixture prior to immunoprecipitation using anti-Maml1 or anti-Skip antibodies. Proteins were detected by SDS-PAGE and Western blotting using the specified antibodies. (C) Purified recombinant NICD⌬2202-Myc (8 pmol) and NICD (4 pmol), Maml1 (4 pmol), and Skip (4 pmol) Flag-tagged proteins were incubated in different combinations. The proteins were immunoprecipitated using anti-Maml1 or anti-Skip antibodies and further analyzed by SDS-PAGE and Western blotting. (D) NICD1771-74A-Flag was cotransfected into 293T cells with NICD⌬R⌬2444-Myc in the presence or absence of Maml1-Myc and Skip-HA. Forty-eight hours posttransfection, the whole-cell lysate was immunoprecipitated with anti-Flag, and samples were then detected by SDS-PAGE and Western blotting. NICD1771-74A was detected using anti-Notch antibody. NICD⌬R⌬2444 and Maml1 tags were detected using anti-Myc antibody, and Skip was detected using anti-HA antibody. (E) 293T cells were transfected with Maml1-Flag and Skip-HA. Forty-eight hours posttransfection, whole-cell lysates were immunoprecipitated with anti-Flag antibody and analyzed by SDS-PAGE and Western blotting. Anti-Flag antibody was used to detect Maml1 and anti-HA antibody to detect Skip.

unclear. Here, we demonstrate that prior to forming a transcriptionally active complex, NICD forms multimers, and these multimers serve as precursors to the assembly of Notch activation complexes. We further provide evidence for a stepwise

assembly of the Notch activation complex that is mediated by Skip and Maml1. It includes the formation of a preactivation complex composed of Skip, Maml1, and NICD multimers. This intermediate complex is then resolved by interaction with CSL,

1406


FIG. 7. Stepwise progression model of an active NICD signaling complex. (I) NICD monomers associate to form multimers, depicted in this model as dimers. (II) Skip binds to NICD multimers. (III) Maml1 is recruited to the NICD multimer-Skip complex, forming the preactivation complex. (IV) This intermediate complex then interacts with the CSL-corepressor complex on DNA, which loads an NICD monomer and Maml1 onto CSL, generating the Notch activation complex. The NICD monomer that is not incorporated into the complex is released, along with Skip and corepressors.

resulting in the formation of the Notch activation complex consisting of monomeric NICD, Maml1, and CSL. Stepwise assembly of the Notch activation complex. Maml1 is an essential component in Notch signaling, forming a stable complex with NICD and CSL and functioning as a “coactivator” (20, 37). A critical role of this protein in the active complex was observed when Maml1 deletion mutants that bind NICD but lack the C-terminal region inhibit Notch transactivation and can act as dominant negatives in Notch signaling (16, 35, 38). Therefore, it is thought that Maml1 functions to recruit other factors to drive Notch function (16, 34, 35, 38). Although it is clear from the crystal structure that Maml1 makes formal contacts with Notch and CSL, purified Notch and Maml1 do not interact (this report and references 26 and 36). In fact, using purified proteins, Maml1 does not interact with either NICD or CSL alone. Maml1 can do so only in the presence of all three proteins (Fig. 4B and 6A). Therefore, a question that remains to be resolved is how Maml1 is incorporated into the Notch activation complex. Skip was initially identified as a cofactor for the Ski oncoprotein (8). Subsequently, Skip has been reported to act both as a corepressor in association with the CSL corepressors SMRT/NCoR and Sharp and as an enhancer or coactivator of the Notch signaling pathway (13, 19, 29, 39). The mechanistic details of how Skip works in Notch transcriptional activation are not known. How does Skip potentiate Notch signaling? Based on our results, we propose that the role of Skip in Notch transactivation is to initiate complex assembly by binding to Notch multimers and to recruit Maml1 to form a preactivation complex (Fig. 7). We demonstrate that Skip preferentially binds to NICD multimers (Fig. 6D), and since NICD monomers and not multimers are in the Notch activation complex (Fig. 4A), we suggest that Skip is likely involved in the early events of Notch activation complex assembly. Moreover, we show that in the presence of Skip, a protein complex containing NICD multimer, Maml1 and Skip can be detected (Fig. 6A and D). Therefore, it appears that the NICD multimer-Skip complex is

MOL. CELL. BIOL.

assembled to provide a docking site for Maml1 to form a preactivation complex. Our data indicating that NICD, Maml1, and Skip assemble into a complex prior to interacting with CSL provide a mechanism for previous observations showing that both Maml1 and Skip are found at the HES-1 promoter only when NICD is present (13). Based on our data, we predict that by preventing the NICD multimer-Skip interaction, Maml1 would not be efficiently recruited to the activation complex and thus the intensity of Notch signaling would be decreased. CSL appears to be the mediator involved in the conversion of a preactivation complex to the Notch transcriptional activation complex. The interaction between the preactivation complex and CSL essentially loads NICD and Maml1 onto CSL bound to DNA and initiates transcriptional activation. How does CSL perform this conversion? It appears that CSL is involved in destabilizing the interaction between the NICD molecules (Fig. 4A and 5A). Previous studies demonstrated that CSL interacts with a 4-amino-acid motif (⌽W⌽P) found within the RAM domain of NICD (7, 17). Here, we show that the RAM domain also interacts with a region between amino acids 2203 and 2216 of NICD, which we now define as the C-terminal multimerization region (CTM) (Fig. 2D). C-terminal deletion mutants of Notch that terminate at amino acid 2240 still form multimers, but deletion mutants that terminate at amino acid 2202 are monomeric (Fig. 2). Furthermore, a deletion mutant that is monomeric is severely compromised for transcriptional activation (Fig. 3C). This is not simply due to the loss of a transactivation domain, since activity can be restored by rescue with a multimerization-competent, transactivation-deficient mutant of Notch (Fig. 3A). The interaction between the RAM domain and CTM does not require a functional ⌽W⌽P motif (Fig. 5A). Therefore, it is possible to physically separate the RAM domain of NICD into an N-terminal multimerization (NTM), amino acids 1820 to 1847, and a CSL binding region (⌽W⌽P motif). Since the RAM domain has two distinct components, we propose that the RAM domain functions as a switch between the preactivation complex and the Notch activation complex. In this model, a region of the RAM domain (NTM) C-terminal to the ⌽W⌽P motif interacts with the CTM. Thus, the NTM and CTM are the main sites of interaction for multimer formation, although other low-affinity sites might be present and contribute to the overall stability of the multimer, like the ankyrin repeats. Later, when CSL interacts with the ⌽W⌽P motif, a conformational change likely occurs in the RAM domain. This results in the RAM domain no longer interacting with the CTM, which drives the conversion from the preactivation complex to the Notch activation complex. Moreover, during the transition from preactivation complex to activation complex, it is not difficult to envision how CSL can displace Skip by steric hindrance from the complex, since both interact with the ankyrin repeat of Notch (26, 33, 36, 39). Depending on Notch presence or absence, it has been reported that Skip can be found forming part of a CSL-repressor complex or a transcriptional activation complex (13, 19, 29, 39). The data that support the model in which Skip is associated with the Notch activation complex come from chromatin immunoprecipitation (ChIP) analysis, in which Skip and other proteins can be detected sitting on the chromatin when Notch is present (12). In this model, Skip will be displaced from the repressor complex and be recruited again once the Notch ac-

VOL. 31, 2011


tivation complex is formed. In our model, Skip is displaced during the transition from the preactivation complex to the transcriptional activation complex by CSL. Considering the results provided, we cannot exclude the possibility that Skip may form part of the final Notch activation complex on DNA, either by staying in the complex upon CSL interaction or by rerecruitment after the transcriptional activation complex is formed. This issue cannot be resolved by ChIP assay, since the technique provides a snapshot in a certain time frame of proteins interacting and not the dynamics of complex formation or transcription. Biochemical and biophysical studies have been mostly focused on the ankyrin repeat domain of NICD (18, 21, 27). However, these studies did not detect multimeric forms of NICD. Why were Notch multimers not detected? In our study, we demonstrate that the C-terminal region of NICD is required for multimer formation. Deletion of the C-terminal region of NICD impairs the formation of NICD multimers (Fig. 2). Moreover experiments using only the ankyrin repeats showed that this domain is not sufficient for multimerization, although its contribution to the stability of the multimer may be important. Furthermore, the truncated forms of NICD utilized in these biophysical studies did not contain the RAM domain or the C-terminal region of NICD, which we demonstrated were necessary for the formation of NICD multimers (10, 22, 40). NICD has been shown to form dimeric activation complexes with Mastermind and CSL on DNA that contained two CSL binding sites positioned in a head-to-head arrangement (1, 25). In this crystal structure, the dimeric complexes are stabilized by the interaction of the ankyrin domain from one NICD with the ankyrin domain of another NICD. The mutation of a residue within the ankyrin domain involved in the dimer formation into alanine (R1985A) prevented the formation of this dimeric structure on DNA. Is this dimerization on DNA similar to the multimerization we observe? To determine if arginine at position 1985 of NICD plays a role in multimerization, we mutated the residue and determined if NICD still formed multimers. As previously published (25), the NICDR1985A mutant still retains reporter activity on a promoter that contains a tandem array of CSL binding sites, 8⫻ CSL promoter, but not on a promoter that contains two CSL binding sites positioned head to head, Hes-1 promoter (data not shown). These data indicate that our mutant is functioning similarly to the published mutant. To determine if NICDR1985A forms multimers, 293T cells were cotransfected with NICDR1985A carrying Flag (NICDR1985AF) and Myc (NICDR1985AM) epitope tags. When NICDR1985AF was immunoprecipitated, NICDR1985AM coimmunoprecipitated (data not shown). These data indicated that the multimerization we observe does not involve the R1985 residue found within the ankyrin domain of Notch. Furthermore the observed dimeric complexes in the crystal structure result from cooperative binding of transcriptional complexes on DNA and not from an intrinsic multimerization property of Notch that we have described for complex assembly. Why do Notch proteins multimerize with each other? We propose that multimerization has evolved to regulate Notch function by controlling the timing/duration of Notch signaling. How does multimerization regulate the timing/duration of Notch signaling? Once released from the plasma membrane, we propose that NICD forms multimers. This establishes the

1407

initial step in regulation, because multimerization is a function of free (monomeric) NICD concentration. NICD multimer formation is necessary to form a complex with Skip. This provides a second step of regulation in Notch activation, because the interaction between NICD multimers and Skip is required to recruit Maml1 to form the preactivation complex. Therefore, Skip availability is predicted to be a limiting factor in Notch signaling. Once the preactivation complex is assembled, the formation of an activation complex with DNA-bound CSL is thought to be rapid. Interaction of the preactivation complex with CSL triggers the loading of NICD and Maml1 to form the activation complex with CSL and concomitantly the release of NICD and Skip. In this step, the NICD multimer is disassociated by CSL, resulting in the retention of only one NICD molecule in the activation complex (Fig. 7). The released NICD monomer is then free to multimerize and initiate another round of activation complex assembly. Once in the activation complex, NICD is rapidly degraded following the initiation of transcription (13). Therefore, we propose that the duration of Notch signaling is a function of the rate of assembly and subsequent destruction of the Notch activation complex and that the cycling of NICD monomers and multimers may provide a mechanism for the oscillation of Notch transcriptional activity. ACKNOWLEDGMENTS We thank members of the Capobianco laboratory for support and technical assistance. We are grateful to S. Artavanis-Tsakonas and Robert Lake (Harvard University) for helpful discussions and critical reading of the manuscript. We thank Emery Bresnick (University of Wisconsin—Madison) for providing the Notch RAM point mutant constructs. This work was funded in part by grants from the National Cancer Institute (ROI CA 83736 to A.J.C.) and the Samuel Waxman Foundation for Cancer Research (to A.J.C.). REFERENCES 1. Arnett, K. L., et al. 2010. Structural and mechanistic insights into cooperative assembly of dimeric Notch transcription complexes. Nat. Struct. Mol. Biol. 17:1312–1317. 2. Artavanis-Tsakonas, S., M. D. Rand, and R. J. Lake. 1999. Notch signaling: cell fate control and signal integration in development. Science 284:770–776. 3. Aster, J. C., and W. S. Pear. 2001. Notch signaling in leukemia. Curr. Opin. Hematol. 8:237–244. 4. Aster, J. C., et al. 2000. Essential roles for ankyrin repeat and transactivation domains in induction of T-cell leukemia by notch1. Mol. Cell. Biol. 20:7505– 7515. 5. Bellavia, D., et al. 2000. Constitutive activation of NF-kappaB and T-cell leukemia/lymphoma in Notch3 transgenic mice. EMBO J. 19:3337–3348. 6. Capobianco, A. J., P. Zagouras, C. M. Blaumueller, S. Artavanis-Tsakonas, and J. M. Bishop. 1997. Neoplastic transformation by truncated alleles of human NOTCH1/TAN1 and NOTCH2. Mol. Cell. Biol. 17:6265–6273. 7. Chu, J., and E. H. Bresnick. 2004. Evidence that C promoter-binding factor 1 binding is required for Notch-1-mediated repression of activator protein-1. J. Biol. Chem. 279:12337–12345. 8. Dahl, R., B. Wani, and M. J. Hayman. 1998. The Ski oncoprotein interacts with Skip, the human homolog of Drosophila Bx42. Oncogene 16:1579–1586. 9. Demarest, R. M., F. Ratti, and A. J. Capobianco. 2008. It’s T-ALL about Notch. Oncogene 27:5082–5091. 10. Ehebauer, M. T., D. Y. Chirgadze, P. Hayward, A. Martinez Arias, and T. L. Blundell. 2005. High-resolution crystal structure of the human Notch 1 ankyrin domain. Biochem. J. 392:13–20. 11. Ellisen, L. W., et al. 1991. TAN-1, the human homolog of the Drosophila notch gene, is broken by chromosomal translocations in T lymphoblastic neoplasms. Cell 66:649–661. 12. Friedmann, D. R., J. J. Wilson, and R. A. Kovall. 2008. RAM-induced allostery facilitates assembly of a notch pathway active transcription complex. J. Biol. Chem. 283:14781–14791. 13. Fryer, C. J., J. B. White, and K. A. Jones. 2004. Mastermind recruits CycC: CDK8 to phosphorylate the Notch ICD and coordinate activation with turnover. Mol. Cell 16:509–520. 14. Gallahan, D., and R. Callahan. 1997. The mouse mammary tumor associated

1408

15. 16.

17. 18.

19.

20.

21.

22.

23. 24. 25.

26.

27.


gene INT3 is a unique member of the NOTCH gene family (NOTCH4). Oncogene 14:1883–1890. Jeffries, S., and A. J. Capobianco. 2000. Neoplastic transformation by Notch requires nuclear localization. Mol. Cell. Biol. 20:3928–3941. Jeffries, S., D. J. Robbins, and A. J. Capobianco. 2002. Characterization of a high-molecular-weight Notch complex in the nucleus of Notch(ic)-transformed RKE cells and in a human T-cell leukemia cell line. Mol. Cell. Biol. 22:3927–3941. Kovall, R. A., and W. A. Hendrickson. 2004. Crystal structure of the nuclear effector of Notch signaling, CSL, bound to DNA. EMBO J. 23:3441–3451. Kurooka, H., K. Kuroda, and T. Honjo. 1998. Roles of the ankyrin repeats and C-terminal region of the mouse Notch1 intracellular region. Nucleic Acids Res. 26:5448–5455. Leong, G. M., et al. 2004. Ski-interacting protein, a bifunctional nuclear receptor coregulator that interacts with N-CoR/SMRT and p300. Biochem. Biophys. Res. Commun. 315:1070–1076. Lin, S. E., T. Oyama, T. Nagase, K. Harigaya, and M. Kitagawa. 2002. Identification of new human mastermind proteins defines a family that consists of positive regulators for notch signaling. J. Biol. Chem. 277:50612– 50620. Lubman, O. Y., M. X. Ilagan, R. Kopan, and D. Barrick. 2007. Quantitative dissection of the Notch:CSL interaction: insights into the Notch-mediated transcriptional switch. J. Mol. Biol. 365:577–589. Lubman, O. Y., R. Kopan, G. Waksman, and S. Korolev. 2005. The crystal structure of a partial mouse Notch-1 ankyrin domain: repeats 4 through 7 preserve an ankyrin fold. Protein Sci. 14:1274–1281. Miele, L., and B. Osborne. 1999. Arbiter of differentiation and death: Notch signaling meets apoptosis. J. Cell. Physiol. 181:393–409. Milner, L. A., and A. Bigas. 1999. Notch as a mediator of cell fate determination in hematopoiesis: evidence and speculation. Blood 93:2431–2448. Nam, Y., P. Sliz, W. S. Pear, J. C. Aster, and S. C. Blacklow. 2007. Cooperative assembly of higher-order Notch complexes functions as a switch to induce transcription. Proc. Natl. Acad. Sci. U. S. A. 104:2103–2108. Nam, Y., P. Sliz, L. Song, J. C. Aster, and S. C. Blacklow. 2006. Structural basis for cooperativity in recruitment of MAML coactivators to Notch transcription complexes. Cell 124:973–983. Nam, Y., A. P. Weng, J. C. Aster, and S. C. Blacklow. 2003. Structural

MOL. CELL. BIOL.

28.

29. 30. 31.

32.

33.

34.

35. 36. 37.

38.

39.

40.

requirements for assembly of the CSL.intracellular Notch1.Mastermind-like 1 transcriptional activation complex. J. Biol. Chem. 278:21232–21239. Nye, J. S., R. Kopan, and R. Axel. 1994. An activated Notch suppresses neurogenesis and myogenesis but not gliogenesis in mammalian cells. Development 120:2421–2430. Oswald, F., et al. 2002. SHARP is a novel component of the Notch/RBPJkappa signalling pathway. EMBO J. 21:5417–5426. Petcherski, A. G., and J. Kimble. 2000. Mastermind is a putative activator for Notch. Curr. Biol. 10:R471–R473. Robbins, J., B. J. Blondel, D. Gallahan, and R. Callahan. 1992. Mouse mammary tumor gene int-3: a member of the notch gene family transforms mammary epithelial cells. J. Virol. 66:2594–2599. Ronchini, C., and A. J. Capobianco. 2001. Induction of cyclin D1 transcription and CDK2 activity by Notch(ic): implication for cell cycle disruption in transformation by Notch(ic). Mol. Cell. Biol. 21:5925–5934. Tani, S., H. Kurooka, T. Aoki, N. Hashimoto, and T. Honjo. 2001. The Nand C-terminal regions of RBP-J. interact with the ankyrin repeats of Notch1 RAMIC to activate transcription. Nucleic Acids Res. 29:1373–1380. Wallberg, A. E., K. Pedersen, U. Lendahl, and R. G. Roeder. 2002. p300 and PCAF act cooperatively to mediate transcriptional activation from chromatin templates by notch intracellular domains in vitro. Mol. Cell. Biol. 22: 7812–7819. Weng, A. P., et al. 2003. Growth suppression of pre-T acute lymphoblastic leukemia cells by inhibition of notch signaling. Mol. Cell. Biol. 23:655–664. Wilson, J. J., and R. A. Kovall. 2006. Crystal structure of the CSL-NotchMastermind ternary complex bound to DNA. Cell 124:985–996. Wu, L., et al. 2000. MAML1, a human homologue of Drosophila mastermind, is a transcriptional co-activator for NOTCH receptors. Nat. Genet. 26:484–489. Wu, L., T. Sun, K. Kobayashi, P. Gao, and J. D. Griffin. 2002. Identification of a family of mastermind-like transcriptional coactivators for mammalian Notch receptors. Mol. Cell. Biol. 22:7688–7700. Zhou, S., et al. 2000. SKIP, a CBF1-associated protein, interacts with the ankyrin repeat domain of NotchIC To facilitate NotchIC function. Mol. Cell. Biol. 20:2400–2410. Zweifel, M. E., D. J. Leahy, F. M. Hughson, and D. Barrick. 2003. Structure and stability of the ankyrin domain of the Drosophila Notch receptor. Protein Sci. 12:2622–2632.