|Home | About | Journals | Submit | Contact Us | Français|
The Notch signaling pathway regulates gene expression programs to influence the specification of cell fate in diverse tissues. In response to ligand binding, the intracellular domain of the Notch receptor is cleaved by the γ-secretase complex and then translocates to the nucleus. There, it binds the transcriptional repressor CSL, triggering its conversion to an activator of Notch target gene expression. The events that control this conversion are poorly understood. We show that the transcriptional corepressor, MTG16, interacts with both CSL and the intracellular domains of Notch receptors, suggesting a pivotal role in regulation of the Notch transcription complex. The Notch1 intracellular domain disrupts the MTG16-CSL interaction. Ex vivo fate specification in response to Notch signal activation is impaired in Mtg16−/− hematopoietic progenitors, and restored by MTG16 expression. An MTG16 derivative lacking the binding site for the intracellular domain of Notch1 fails to restore Notch-dependent cell fate. These data suggest that MTG16 interfaces with critical components of the Notch transcription complex to affect Notch-dependent lineage allocation in hematopoiesis.
The specification of alternate cell fates during development requires nuclear factors that receive and reconcile inputs from signal transduction cascades activated by multiple cell surface receptors. The importance of many of these factors to normal tissue formation and homeostasis is underscored by their roles in cancer pathogenesis. New therapeutic strategies for cancer hinge upon our understanding of these factors, their relationships to one another, and how these relationships are regulated.
Notch receptors and their nuclear partners are critical determinants of cell lineage allocation, commitment, and differentiation in diverse tissues. Dysfunction within the Notch signal transduction machinery characterizes developmental and malignant conditions, including lymphoid and myeloid leukemias (2, 4, 5, 18, 24, 36). The four Notch receptors in vertebrates are type I transmembrane proteins that undergo complex posttranslational processing en route to the plasma membrane (7, 14, 51). This includes O-linked glycosylation and proteolytic cleavage by furin protease, leaving Notch receptors noncovalently associated at the cell surface. There, they are activated by interactions with ligands of the Delta-Serrate-Lag-2 (DSL) and Jagged families expressed on adjacent cells. Ligand binding prompts serial proteolytic cleavages of the Notch receptor, releasing the Notch intracellular domain (N-ICD) (7, 14, 17, 28, 29). Upon its release, the N-ICD enters the nucleus, partners with CSL (CBF1/Su[H]/Lag-1) and Mastermind-like (MAML) proteins to form the canonical Notch transcription complex, and triggers expression of Notch target genes (30). In the absence of N-ICD, CSL is a constitutive repressor of Notch target genes, reflecting its direct and indirect interactions with components of the transcriptional repression machinery (7). Corepressors that interact directly with CSL include NCoR/SMRT (26), SHARP/MINT/SPEN (41), CIR (21), SKIP (54), HAIRLESS, CtBP, and Groucho/TLE (8, 38, 39). While the conversion of CSL from constitutive repressor to activator of Notch targets is triggered by N-ICD, the underlying mechanism is poorly understood.
MTG16 is a member of the myeloid translocation gene (MTG) family of transcriptional corepressors, which includes MTG8 and MTGR1 (11). Through interactions with DNA binding proteins, MTG family members recruit other corepressors and histone deacetylases (HDACs) to target promoters to regulate gene expression (22). The importance of MTG family members to cell growth and development is revealed by their participation in chromosomal translocations seen in acute myeloid leukemia (AML). RUNX1-MTG8, the product of t(8;21), characterizes approximately 15% of AML cases, while RUNX1-MTG16 is found in cases of secondary AML (15, 42, 52). Notably, a Notch “transcriptional signature” is associated with RUNX1-MTG8 expression in primary human AML isolates (1). More recently, mutations or altered expression at MTG loci have been described in nonhematologic malignancies, such as ductal carcinoma of the breast and colorectal cancer, reinforcing their essential contributions to cell growth and development (27, 47, 49). Mice with constitutive deletion of Mtg16 display defects in hematopoietic stem/progenitor cell functions (10). Inactivation of Mtg16 skews hematopoietic progenitors toward the granulocyte/macrophage lineage and impairs megakaryocytic-erythroid progenitor cell expansion in response to hemolytic stress.
MTG proteins are evolutionarily related to Drosophila Nervy and share with it four highly conserved Nervy homology regions (NHR1 to NHR4) and three divergent regions rich in proline, serine, and threonine (PST domains) (11). The NHR regions serve as docking sites for DNA binding proteins, corepressors, and HDACs to localize epigenetic effectors to target promoters. Functions localized to the NHR domains are necessary for the immortalizing and transforming properties of RUNX1-MTG8 (32, 50). The PST domains are likely to confer family member-specific functions, but their precise contributions to MTG proteins or their proleukemic derivatives are not known.
Binding partners for MTG family proteins have been vigorously explored. Among DNA binding proteins, interactions with GFI1 (34) and GFI1B (46), PLZF (35), TAL1/SCL (46), E2A/HEB (16), TCF4 (37), LDB1 (19), BCL6 (9), and GATA1 (20) have been described. Class I HDACs interact directly with MTG family proteins, while interactions with NCoR/SMRT, SIN3A, and SHARP/MINT/SPEN provide additional modes of HDAC and corepressor recruitment (3, 33, 41, 44). MTG family proteins also bind one another in an antiparallel fashion through their NHR2 regions to form homo- and hetero-oligomers (32, 53). As such, MTG proteins are strategically placed within the hierarchy of transcriptional repressor complex assembly.
Given the pivotal position occupied by MTG proteins in transcriptional regulatory complexes, the shared impact of MTG16 and Notch signaling on hematopoietic cell fate determination, and the capacity of RUNX1-MTG8 to both bind MTG16 and elicit a Notch gene expression signature, we explored the relationship between MTG16 and the Notch transcription complex. We show that MTG16 interacts with the core components of the Notch transcription complex, CSL and N-ICD, and map their binding domains on MTG16 to distinct regions. We also show that Notch1 intracellular domain (N1-ICD) expression interferes with the MTG16-CSL interaction. Furthermore, Notch-dependent specification of lymphoid fate ex vivo is defective in Mtg16−/− hematopoietic progenitor cells and restored by MTG16 expression, but not by an N1-ICD binding-deficient mutant. These findings suggest that MTG16 is a key component of the Notch regulatory network and reveal another aspect of the mechanism for Notch-dependent transcriptional activation.
OP9-DL1 stromal cells, a generous gift of Juan Carlos Zúñiga Pflücker, were maintained in minimal essential medium alpha (α-MEM) supplemented with 10% fetal bovine serum (FBS). NMuMG cells, generously provided by Al Reynolds, and COS7L and BOSC23 cells were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% FBS. NIH 3T3 fibroblasts were propagated in DMEM supplemented with 10% bovine calf serum (BCS). The T-cell leukemia cell line Jurkat was propagated in RPMI 1640 supplemented with 10% FBS. All cell culture media were supplemented with 2 mM l-glutamine, 50 units/ml penicillin, and 50 μg/ml streptomycin. Molecular-weight standards for sucrose gradient sedimentation, trichostatin A, Hoechst 33342, and protein G-Sepharose were obtained from Sigma-Aldrich, Inc. Monoclonal antibodies against the myc (9E10) and hemagglutinin (HA; 12CA5) epitope tags were obtained from the Vanderbilt Monoclonal Antibody Core facility. The corresponding rabbit polyclonal antibodies to myc (A-14) and HA (Y-11) epitope tags, as well as anti-MTG16 (G-20) and anti-Gal4-DBD (RK5C1), were obtained from Santa Cruz Biotechnology, Inc. The anti-green fluorescent protein (anti-GFP) antibody (7.1/13.1) was obtained from Roche Applied Science, Inc. Normal goat serum (NGS), normal mouse serum (NMS), and the anti-α-tubulin antibody (B-5-1-2) were obtained from Sigma, Inc. All other materials were of reagent grade. The anti-N1-ICD antibody bTan20 was the generous gift of Thao Dang.
To produce anti-MTG16 polyclonal antibody A/J-R, an A/J mouse was successively immunized with Freund's adjuvant and a mixture of glutathione S-transferase (GST) fusion proteins expressing the NHR regions of MTG16 with their flanking sequences, thus representing the entire MTG16 primary structure. Serum samples were collected during peak titers and evaluated for recognition of myc epitope-tagged MTG16 to confirm non-cross-reactivity with GST and to allow direct comparison to immunoblots probed with anti-myc monoclonal antibody 9E10. In parallel, an identical evaluation was performed with preimmune serum. The A/J-R antibody recognizes myc-MTG16 expressing COS7L cells and a single band from whole-cell extract of endogenous proteins prepared from Jurkat cells. No bands were detected using preimmune serum with either endogenous or transiently expressed MTG16 (see Fig. S1 in the supplemental material). Secondary antibody conjugates with horseradish peroxidase (HRP) were obtained from Sigma, Inc. (anti-mouse antibody-HRP and anti-goat antibody-HRP) and Jackson Labs, Inc. (anti-rat antibody-HRP and anti-rabbit antibody-HRP).
The pcDNA3-myc-CSL expression plasmid was the generous gift of Warren Pear. Thao Dang generously provided the pcDNA3-hNotch1 and pCMV-Tag4a-N3-ICD-Flag constructs. The pHyTc-hNOTCH4-HA plasmid was provided by Jan Kitajewski. The pBIG plasmid, which expresses GFP, was the generous gift of Jennifer Pietenpol. pCMV(M2)-Gal-MTG16, pCMV5-myc-MTG16, and expression plasmids for HA epitope-tagged MTGR1, MTG16, and RUNX1-MTG8 in pCMV5 have been previously described (3). Fragments for the MTG16 NHR regions and their flanking sequences were prepared by PCR amplification, using primers that incorporated XbaI sites at their 5′ ends. Fragments were digested, gel purified, and subcloned into an XbaI-restricted pCMV(M2)-Gal vector, yielding pCMV(M2)-Gal-MTG16NHR1, pCMV(M2)-Gal-MTG16NHR2, pCMV(M2)-Gal-MTG16NHR3, and pCMV(M2)-Gal-MTG16NHR4. pCMV-Flag-NCoR, pCMV-Flag-Sin3a, pCMV-Flag-HDAC3, MIG-myc-MTG16, Gal-TK-luciferase, and Hes1-luciferase have been previously described (3, 10, 13, 23, 25, 43). To produce GST fusion proteins of MTG16 NHR domains, each domain, with its N-terminal and C-terminal flanking sequences, was isolated from the corresponding pCMV(M2)-Gal-NHR expression plasmid as an XbaI fragment, filled in with Klenow and subcloned into SmaI-linearized pGEX4T3 to generate pGEX4T3-NHR1, pGEX4T3-NHR2, pGEX4T3-NHR3, and pGEX4T3-NHR4. Correct orientation was confirmed by restriction and GST fusion proteins were prepared essentially as described (12). Flag epitope-tagged N1- and N2-ICD constructs were produced by PCR amplification of the N1- and N2-ICDs with Pfu Turbo by using primers that incorporated BamHI and EcoRI sites in the 5′ and 3′ primers, respectively. Gel-purified amplimers were digested with BamHI and EcoRI and subcloned into BamHI/EcoRI-restricted pCMV-Tag4a, creating pCMV-Tag4a-N1-ICD-Flag and pCMV-Tag4a-N2-ICD-Flag. The pCMV5-myc vector was produced by excising the MTG16 insert from pCMV5-myc-MTG16 with XbaI and by ligating the gel-purified, linearized vector sequence. PCR was used to amplify MTG16 fragments with serial deletions from the N (ΔN) and C (ΔC) termini to create the serial deletion panels shown in Fig. Fig.8.8. For the ΔN panel, 5′ primers incorporated an in-frame XbaI site, and the common 3′ primer incorporated a STOP-XhoI-XbaI sequence. The primer sequences can be found in Table S1 in the supplemental material. PCR fragments were gel purified, digested with XbaI, and subcloned into a pCMV5-myc vector at the XbaI site, placing the insert in-frame and immediately downstream of the myc epitope tag. Fragments for the ΔC panel were also created by PCR amplification with Pfu Turbo. A common 5′ primer contained an XbaI site in-frame with the MTG16 coding sequence. The 3′ primers each contained an in-frame stop codon flanked by SpeI and XbaI sites. Again, the primer sequences used are found in Table S1 in the supplemental material. Amplimers were digested with XbaI, gel purified, and subcloned into an XbaI-linearized pCMV5-myc vector. For each deletion mutant, the correct orientation of the inserted fragment was confirmed by restriction digest. The MIG-myc-MTG16Δ2N plasmid was created by excising the MTG16Δ2N insert from pCMV5-myc-MTG16Δ2N with EcoRI and XhoI and subcloning into MSCV-Ires-GFP (MIG) linearized with EcoRI and XhoI.
Lin−/Sca-1+/c-Kit+ (LSK) hematopoietic progenitors were purified from total bone marrow cells of Mtg16+/+ and Mtg16−/− mice as described previously (10), and mRNA was purified using the 5 Prime PerfectPure kit per the manufacturer's instructions. RNA was reverse transcribed and the resulting cDNA amplified by real-time PCR in a Bio-Rad iCycler instrument, using appropriate Hes1 (Mm00468601_m1)-, Hes5 (Mm00439311_g1)-, Hey1 (Mm00468865_m1)-, Notch1 (Mm03053614_s1)-, Notch2 (Mm03053783_s1)-, and Nrarp (Mm428529_s1)-specific TaqMan probes (Applied Biosystems).
Expression plasmids in the combinations indicated were transfected into COS7L cells with Lipofectamine according to the manufacturer's instructions. Cell monolayers were harvested at 60 h posttransfection. Harvest and all subsequent steps were performed at 4°C. Monolayers were scraped into ice-cold phosphoprotecting lysis buffer (PPLB; 50 mM Tris, 150 mM NaCl, 10% glycerol, 1 mM EDTA, 1 mM dithiothreitol [DTT], 1% Triton X-100, 0.1% Na-deoxycholate, 0.1% SDS, 20 mM β-glycerophosphate, 20 mM p-nitrophenylphosphate [PNPP], 2 mM Na-pyrophosphate, 1 mM NaVO4, 5 mM NaF, 1 mM phenylmethylsulfonyl fluoride [PMSF], 10 μg/ml aprotinin [pH 8.0]) supplemented with 3% Casamino Acids, disrupted by sonication, and after 30 min of rocking at 4°C, clarified by centrifugation. As indicated in the figures, antibodies were added to clarified extracts for 90 min, and then immune complexes and coprecipitating proteins were collected with protein G-Sepharose preadsorbed in PPLB with 3% Casamino Acids. Immune complexes were washed successively with PPLB and boiled in an SDS-PAGE sample buffer supplemented with DTT.
N1-ICD and N3-ICD were expressed in vitro from pCMV-Tag4a-N1-ICD-Flag and pCMV-Tag4a-N3-ICD-Flag, respectively, using the Promega T3 coupled transcription/translation system incorporating biotin-lysyl-tRNA to label the expressed proteins. An aliquot was removed for analysis, and then expressed proteins were mixed with GST or GST-NHR1 purified from bacterial extracts and bound to glutathione agarose beads. After washing extensively with 1× phosphate-buffered saline (PBS) with 0.05% Tween 20 (PBS-T), coprecipitating proteins were separated by SDS-PAGE, transferred to nitrocellulose filters, and probed with streptavidin-conjugated HRP. Proteins were visualized by chemiluminescence detection.
Aliquots of whole-cell lysates or immune complexes were separated by SDS-PAGE as described previously (12) and transferred to nitrocellulose filters in 25 mM Tris-192 mM glycine (pH 8.35) overnight at 4°C. Nitrocellulose filters were blocked in PBS-T supplemented with 0.25% gelatin and 0.02% NaN3 (GBB) and then incubated in primary antibodies as indicated in the figures. Membranes were washed successively in PBS-T and then probed with secondary antibody-HRP conjugates appropriate to the species of primary antibody employed. Proteins were visualized by chemiluminescence detection.
NIH 3T3 fibroblasts were grown on six-well plates for reporter experiments. Forty-eight hours after transfection with expression plasmids and reporters as shown, monolayers were analyzed for firefly and Renilla luciferase expression by using the Promega dual-luciferase reporter system according to the manufacturer's instructions. Assays were performed in triplicate. Firefly luciferase data were normalized to the internal control Renilla luciferase activity and were presented as the mean ± 1 standard deviation of the results from a representative experiment.
Transfected cells were subjected to hypotonic lysis and cytoplasmic and nuclear fractions were prepared essentially as described previously (48). Briefly, cell monolayers were washed with 1× PBS and then resuspended in hypotonic lysis buffer (10 mM Tris, 140 mM NaCl, 1.5 mM MgCl2, 0.5% Triton X-100 [pH 8.4]). Samples were fractionated by centrifugation at 500 × g. Supernatants were removed as cytoplasmic fractions. Pellets were washed with hypotonic lysis buffer, resuspended in Laemmli sample buffer, and subjected to sonication. After centrifugation at 18,000 × g, nuclear extracts were removed as supernatants. Protein concentrations were determined by the method of Bradford, and equal amounts of total protein from cytoplasmic and nuclear fractions were analyzed by immunoblotting as noted above.
NIH 3T3 fibroblasts were grown on glass coverslips and transfected with pBIG and pCMV5-myc-MTG16 by using Lipofectamine. Cells were fixed in buffered 3.7% formalin, washed extensively with PBS, made permeable with 1% Triton X-100, again washed extensively with PBS, and blocked with 10% normal goat serum in PBS. MTG16 was detected in GFP-positive cells with anti-myc antibody 9E10 or NMS followed by Alexa 568-conjugated goat anti-mouse secondary antibody and fluorescence microscopy. Nuclei were visualized by Hoechst 33342 staining. Transfections with the pCMV5-myc vector were processed in parallel as negative controls.
COS7L cells were transfected with pcDNA3-myc-CSL, pCMV5-HA-MTG16WT, and pCMV-Tag4a-N1-ICD-Flag in the combinations shown in Fig. Fig.6.6. After 48 h, cells were harvested in 500 μl of lysis buffer (50 mM Tris, 150 mM NaCl, 5% glycerol, 1 mM ETDA, 1 mM DTT, 0.1 mM NaN3, 0.1% Triton X-100), disrupted by sonication, and after 30 min of rocking at 4°C, layered over gradients of sucrose from 10% to 40% (wt/wt) in HEMG buffer (25 mM HEPES, 100 mM KCl, 0.1 mM EDTA, 12.5 mM MgCl2, 10% glycerol [pH 7.9]) that had been prepared in five stepwise layers and allowed to diffuse for 18 h at 4°C. Aliquots of extracts were removed for immunoblot analysis. The samples were subjected to ultracentrifugation at 40,000 rpm in an SW40Ti rotor for 18 h. A gradient run in parallel contained sedimentation standards. Following centrifugation, 55 200-μl fractions were removed manually from each gradient and slowly drawn through nitrocellulose filters by using a slot blot apparatus and vacuum manifold. The filters were then probed with antibodies directed toward the epitope tags on the desired protein. Proteins were detected using fluorescently labeled secondary antibodies and the Li-Cor Odyssey detection system. In parallel, the sedimentation standards were identified by silver staining of SDS-PAGE gels. The peak fraction and range for each sedimentation standard are indicated.
LSK progenitors were isolated from whole bone marrow collected from the femurs and tibias of wild-type or Mtg16−/− mice as described previously (10). Retroviruses were prepared by transfecting a MIG vector, MIG-myc-MTG16WT or MIG-myc-MTG16Δ2N, into BOSC23 packaging cells. Supernatants containing the desired retrovirus were used to infect LSK progenitors for 1 h by spinoculation. Transduced LSK progenitors were cocultured over irradiated OP9-DL1 bone marrow stromal cells in the presence of interleukin-7 (IL-7) and Flt3 ligand. Cells were collected at day 21, labeled with antibodies directed toward CD4 and CD8 as shown, and subjected to analysis by flow cytometry.
Altered expression of Notch target genes HES1 and HEY1 was recently described in the RUNX1-MTG8-expressing leukemia cell line Kasumi-1 (44). Notably, we have previously described a hematopoietic defect in mice bearing a germ line deletion of Mtg16. To gain additional insights into the alterations in gene expression that correlate with this defect, we isolated mRNA from LSK progenitors of Mtg16+/+ and Mtg16−/− mice and quantified the relative expressions of Notch targets Hes1, Hes5, Hey1, Notch1, Notch2, and Nrarp by qRT-PCR. We observed 8.6-, 5.3-, and 2.5-fold elevation of Hes1, Notch1, and Nrarp expression, respectively, in mRNA from Mtg16−/− LSK cells relative to wild type (Fig. (Fig.1),1), while no significant change in Hes5, Notch2, or Hey1 was observed. These findings suggest that Mtg16 influences the steady-state expression of a subset of Notch target genes.
The repressor function of CSL on Notch-responsive promoters can be augmented by MTG8 through its interaction with the CSL corepressor, SHARP (44). Given the gene expression data shown in Fig. Fig.1,1, we extended this analysis to MTG16. Myc epitope-tagged CSL was transiently expressed either alone or with HA epitope-tagged MTG16, MTGR1, or RUNX1-MTG8 in COS7L cells and whole-cell extracts subjected to immune precipitation with anti-myc antibody. MTG family proteins were detected in immune complexes by anti-HA immunoblot analysis, and equivalent expression of each protein was determined using antibodies directed toward the appropriate epitope tags. Equivalent transfection efficiency and loading were confirmed by immunoblotting for GFP and α-tubulin, respectively. We observed MTG16, MTGR1 and RUNX1-MTG8 in anti-myc immune complexes in a CSL-dependent manner (Fig. (Fig.2a),2a), in keeping with a contribution to repressor complexes coordinated by CSL.
The N-terminal half of MTG8 is sufficient to associate with SHARP (44). To define the CSL binding motif in MTG16, we expressed myc epitope-tagged CSL with Gal fusion derivatives of MTG16 NHRs (Fig. (Fig.2b)2b) and probed for CSL in anti-Gal immune complexes. Notably, CSL coprecipitated with Gal-MTG16NHR3. However, neither Gal-MTG16NHR2 nor Gal-MTG16NHR4 copurified with CSL in parallel immune precipitations. Because the N-terminal and C-terminal flanking sequences in Gal-MTG16NHR3 are shared with Gal-MTG16NHR2 and Gal-MTG16NHR4, respectively, these data suggest that the NHR3 domain coordinates the interaction between MTG16 and CSL. Furthermore, because SHARP binding maps to the N-terminal half of MTG8, these data suggest an interaction between MTG16 and CSL that is SHARP independent.
CSL recruits NCoR/SMRT (26), SHARP/MINT/SPEN (41), CIR (21), SKIP (54), HAIRLESS, CtBP, and Groucho/TLE (8, 38, 39) to repress transcription of canonical Notch target genes. Activation of Notch signaling stimulates the release of N-ICD, which enters the nucleus to bind CSL and trigger its conversion to an activator of Notch target genes. This coincides with recruitment of the transcriptional coactivator MAML. The relationship between N-ICD and corepressors has not been extensively addressed. Using a candidate approach, we tested proteins widely represented in repressor complexes for their ability to bind N1-ICD. Flag-tagged N1-ICD was expressed alone, with myc-tagged MTG16 or myc-tagged CSL used as a positive control. A robust interaction was observed between MTG16 and N1-ICD that rivaled its binding to its transcriptional partner, CSL (Fig. (Fig.3a).3a). No evidence of binding between N1-ICD and either NCoR, SIN3A, or HDAC3 was observed (Fig. (Fig.3b3b).
To confirm the N1-ICD-MTG16 interaction at endogenous levels of protein expression, we performed bidirectional immune precipitations from extracts prepared from Jurkat T-cell leukemia cells. A mutation in the E3 ubiquitin ligase FBW7 in these cells protects N1-ICD from ubiquitin-mediated degradation (40). Jurkat cell extracts were subjected to immune precipitation with nonimmune serum (NGS or NMS), anti-MTG16 (G-20 [Fig. [Fig.3c,3c, left panel]), or anti-N1-ICD (bTan20 [Fig. [Fig.3c,3c, right panel]) antibodies. Immune complexes were analyzed for the presence of the interacting partner. The interaction between N1-ICD and MTG16 was identified in both cases. These findings intimate a binding relationship between MTG16 and N-ICDs at endogenous levels of expression.
There are four Notch receptors in mammals (Notch1 to Notch4) (4, 5, 7, 14, 17, 28). Notch1 and Notch2 are structurally similar, and deletion of either in the germ line results in embryonic lethality. In contrast, Notch3 and Notch4 receptors are structurally distinct, and mice deficient for either Notch3 or Notch4 survive with no gross phenotypic abnormalities. These findings suggest incomplete functional overlap among the Notch receptors, and broaden the scope of their importance as a family to tissue development and homeostasis. To generalize the relationship between MTG16 and Notch signaling systems, we examined its binding to the intracellular domain of each Notch receptor. Flag-tagged intracellular domains of Notch1, Notch2, Notch3, and HA-tagged Notch4-ICD were expressed alone or with myc-MTG16 in COS7L cells. Immune precipitation was performed using anti-Flag or anti-myc antibodies as shown, and the presence of coprecipitating MTG16 or N-ICD was determined by immunoblotting. Notably, a robust interaction was observed between MTG16 and each N-ICD (Fig. (Fig.4).4). Coupled with the interactions between MTG family proteins and CSL, these findings indicate a generalized relationship between MTG family proteins and core elements of the canonical Notch transcription complex.
MTG16 concentrates in the nucleus, binds multiple nuclear proteins that regulate gene expression, and can repress gene expression when tethered to promoters by DNA binding domains of transcriptional regulators (11, 22) (see Fig. S2 in the supplemental material). Similarly, N-ICDs translocate to the nucleus following release from the plasma membrane, bind transcriptional regulators, and alter expression of Notch target genes. The association between N-ICDs and MTG16 suggests a mechanism for altering the interaction between CSL and MTG16. To clarify their binding relationships, we examined the impact of N1-ICD expression on the interaction between MTG16 and CSL by using immune coprecipitation (Fig. (Fig.5).5). Myc epitope-tagged CSL and HA epitope-tagged MTG16 were expressed together in COS7L cells with an empty vector or increasing amounts of Flag epitope-tagged N1-ICD. CSL-containing immune complexes were collected from whole-cell extracts with anti-myc antibody and protein G-Sepharose and then divided into equal aliquots for anti-HA and anti-Flag immunoblots to determine MTG16 and N1-ICD coprecipitation, respectively. Expressing N1-ICD-Flag abolished the MTG16-CSL interaction at each level of its expression. Concomitantly, we observed a progressive rise in N1-ICD-Flag binding to myc-CSL. Notably, N1-ICD expression similarly destabilized the interaction between MTG16 and NCoR, yet failed to disrupt the interaction between CSL and NCoR (see Fig. S3 in the supplemental material). These data suggest that N1-ICD destabilizes interactions between MTG16 and both CSL and NCoR and that the MTG16-N1-ICD interaction may help to integrate N1-ICD functions in the nucleus following Notch signal activation.
To confirm this finding, we utilized cosedimentation in sucrose density gradients to assess the impact of N1-ICD expression on the CSL-MTG16 interaction (Fig. (Fig.6).6). We first determined the sedimentation characteristics for myc-CSL, HA-MTG16, and N1-ICD-Flag alone in 10 to 40% (wt/wt) sucrose gradients. Each gradient was then divided into 55 equal fractions, and the proteins in each fraction were bound to nitrocellulose filters in a slot blot format. Filters were probed with antibodies directed toward the epitope tags specific to each expressed protein. Molecular-weight standards were fractionated in parallel and visualized in silver-stained SDS-PAGE gels. The peak fraction and distribution for each marker are indicated below Fig. Fig.6D.6D. Additionally, an aliquot of each whole-cell extract prior to fractionation was subjected to SDS-PAGE and epitope-tag specific immunoblotting to confirm that only the transiently expressed protein was responsible for the signal observed in the slot blot.
When expressed individually, CSL and N1-ICD cosediment with low-molecular-weight standards, consistent with their monomeric molecular weights. In contrast, MTG16 was widely distributed across gradient fractions, consistent with its propensity to form higher-order oligomers. When coexpressed with MTG16, a portion of CSL shifted into higher-molecular-weight fractions that overlapped with the sedimentation of MTG16. However, when N1-ICD expression was imposed upon CSL-MTG16 cosedimentation, this higher-molecular-weight CSL population was lost, and N1-ICD cosedimented with CSL. Additionally, an N1-ICD peak emerged in the high-molecular-weight fractions of the gradient. This peak was minimally detected when N1-ICD was expressed alone but became more prominent in the context of MTG16-N1-ICD coexpression and overlapped with the distribution of MTG16 in the gradient. Collectively, these data suggest that the MTG16-CSL interaction is disrupted by N1-ICD binding and coincides with the emergence of a high-molecular-weight complex that includes N1-ICD and MTG16.
To define the binding site for N-ICDs, we created a panel of Mtg16 mutants with sequential deletions from the N and C termini (Fig. 7B and D) and used these constructs in immune precipitation experiments with N1-ICD. Each Mtg16 deletion construct carries a myc epitope at its N terminus. GFP and α-tubulin were employed as controls for equivalent transfection and gel loading, respectively. Deletion of the N-terminal 40 amino acids (MTG16Δ1N) had no effect on the N1-ICD interaction, but further deletion to residue 85 from the N terminus (MTG16Δ2N) abolished N1-ICD binding (Fig. (Fig.7A).7A). These findings indicate that MTG16 residues 40 to 85 are necessary for the interaction with N1-ICD. This observation is further supported by coprecipitation analysis using fragments of MTG16 with serial deletion from the C terminus. Each fragment in this deletion panel retained the ability to bind N1-ICD, including the amino-terminal 145 amino acids (Fig. (Fig.7C).7C). To determine if MTG16 interacts directly with N-ICDs, we utilized biotin-lysyl-tRNA in synthesizing N1- and N3-ICDs in rabbit reticulocyte lysate and then examined interactions of each biotinylated protein with bacterially expressed GST-NHR1 in vitro. GST-NHR1 contains the N-ICD binding region from MTG16 (Fig. (Fig.8).8). Both N1- and N3-ICDs coprecipitate with GST-NHR1 but not GST. Collectively, these data define residues 40 and 145 of MTG16 as the N- and C-terminal boundaries of the N1-ICD binding domain and suggest that the interaction is direct.
LSK hematopoietic progenitor cells from Mtg16−/− mice fail to make CD4/CD8-double-positive T cells when cultured with OP9-DL1 cells ex vivo (A. Hunt et al., unpublished data). OP9-DL1 is a bone marrow stromal cell line that constitutively expresses the Notch ligand Delta-like-1 (DL1) and directs T-cell development in vitro (6, 45). We employed this developmental assay to test the contribution of the MTG16 N1-ICD binding motif in primary hematopoietic progenitor cells. Mtg16−/− LSK progenitors were transduced with a MIG vector, MIG-myc-Mtg16WT or MIG-myc-Mtg16Δ2N, which lacks the N1-ICD interaction domain (Fig. (Fig.9).9). In parallel, we transduced Mtg16+/+ LSK cells with MIG vector as a technical control. CD4 and CD8 expression in MIG vector-transduced Mtg16+/+ LSK cells was comparable to that seen in nontransduced populations, suggesting no appreciable effect of retroviral transduction on lymphoid lineage allocation. Notably, Mtg16−/− LSK cells transduced with MIG vector remained CD4/CD8 negative after OP9-DL1 coculture. When Mtg16−/− LSK cells are transduced with MIG-myc-MTG16WT, expression of CD4 and CD8 was restored in GFP-positive cells. However, neither CD4 nor CD8 expression was restored by OP9-DL1 coculture in Mtg16−/− LSK cells transduced with MIG-MTG16Δ2N. Thus, this N-terminal motif in MTG16, which supports N1-ICD binding and is distinct from its highly conserved NHR domains, is essential for Mtg16-dependent lineage specification and suggests an intimate relationship between MTG family proteins and Notch-dependent phenotypes.
Notch controls binary cell fate decisions during development and adult tissue renewal to produce and maintain normal tissue architecture. To achieve this effect, Notch signals work together with an intricate network of nuclear factors that collectively coordinate and integrate incoming signals to direct complex phenotypes. Central to the effects of Notch on cell fate specification is the conversion of CSL from being a repressor to an activator of gene expression in response to Notch-activating stimuli. Before Notch activation, CSL nucleates assembly of transcriptional repressor complexes on Notch-responsive promoters. Several proteins have been implicated in CSL-mediated transcriptional repression, including NCoR/SMRT (26), SHARP/MINT/SPEN (41), CIR (21), SKIP (54), HAIRLESS (39), and CtBP and Groucho/TLE (8, 38). MTG family proteins expand the roster of corepressors that bind CSL and reinforce the notion that stringent repression of gene expression is desired to control Notch-dependent phenotypes. Recently, Salat et al. showed that MTG8 enhances repression by SHARP at RBP-Jκ responsive promoters and that knockdown of MTG8 increases Notch target gene expression without overt Notch receptor activation (44). Similarly, we found enhanced expression of Notch targets Hes1, Notch1, and Nrarp in Mtg16−/− LSK hematopoietic progenitors, with no appreciable alteration in Hes5, Hey1, or Notch2 expression. The reason for distinct effects of Mtg16 deletion on expression of Notch target genes is unknown. One plausible explanation is that distinct repression complexes are assembled on the promoters of genes regulated by Notch, and these complexes are differentially sensitive to the loss of MTG16 or permit compensation by other MTG family proteins. Alternatively, altered Notch target gene expression in Mtg16−/− progenitors may reflect more complex compensatory mechanisms established to preserve homeostasis in the context of Mtg16 nullizygosity. From this perspective, it is notable that MTG16 binds HDAC6 and HDAC8, in addition to HDAC1, HDAC2, and HDAC3, which MTG8 binds also. Likewise, while MTG8 binds both NCoR and SIN3A, MTG16 fails to bind the latter (3). These findings suggest that MTG16 contributes selectively to basal repression of canonical Notch target genes and that compensation by other MTG family members might alter transcriptional outcomes. Furthermore, binding domains for SHARP and CSL map to distinct regions in MTG proteins. Our results do not exclude an indirect interaction between CSL and MTG16, mediated by an NHR3 binding protein. However, it is attractive to speculate that CSL, SHARP, and MTG family proteins form a ternary complex that, in concert with other corepressors and HDACs, limits promiscuous expression of Notch target genes. In joining the Notch transcription complex, intracellular domains of Notch receptors must overcome this repression.
Before Notch signaling is activated, CSL and Notch receptors are segregated to the nucleus and plasma membrane, respectively. Notch-activating stimuli induce proteolytic cleavage of the Notch receptor, liberating the N-ICD to permit its interaction with CSL in the nucleus. N-ICD binding to CSL overcomes transcriptional repression and initiates assembly of the activation-competent Notch transcription complex. To trigger this conversion, we hypothesized that N-ICDs would bind components of CSL repressor complexes. Utilizing a candidate approach, we characterized interactions between the N1-ICD and repressor complex proteins to discover a robust interaction with MTG16. Indeed, MTG16 interacts with the intracellular domains of all mammalian Notch receptors, suggesting a generalized relationship between it and Notch signaling. The N1-ICD binding motif in MTG16 is localized to the PST1 region, distinct from NHR domains and flanking structures responsible for interactions with DNA binding proteins, corepressors, and HDACs. No other proteins have been identified that bind this domain in MTG16.
N1-ICD expression disrupts the MTG16-CSL and MTG16-NCoR interactions. However, it fails to bind NCoR, SIN3A, or HDAC3 and fails to disrupt the interaction between CSL and NCoR (see Fig. S3 in the supplemental material). Combined, these data suggest that N1-ICD is insufficient to overcome the CSL-NCoR interaction and instead may act through MTG16 to expel NCoR from CSL repressor complexes. Notably, Kao et al. have shown that CSL interacts with SMRT and that SMRT disrupts the CSL—N1-ICD interaction and antagonizes N1-ICD-dependent transcription via CSL (26). These data suggest that SMRT binding by CSL is favored over its interaction with N1-ICD and that N1-ICD-dependent disruption of the CSL repressor complex must be facilitated by other factors. MTG16 demonstrates characteristics consistent with this role and, in so doing, could both contribute to basal repression at CSL-regulated promoters and facilitate the transition to transcriptional activation in response to N1-ICD binding.
A hypothetical model reflecting these roles for MTG16 in the Notch transcription complex is presented in Fig. Fig.10.10. By interacting with CSL, corepressors, and HDACs, MTG16 would reinforce repression of Notch target genes in the basal state. Notch activation would liberate the N-ICD for nuclear entry, permitting its interaction with MTG16 and causing conformational changes that destabilize repressor complex interactions. Whether N-ICD-MTG16 binding is sufficient to disrupt repressor complex integrity is not yet known. It may be that interactions between N-ICD and other repressor complex components are also required. Additionally, our data do not exclude the possibility that MTG16 remains an integral component of the Notch transcription complex despite attenuated CSL binding. In support of this notion, MTG16 interacts directly with N-ICDs and they can be found together in high molecular weight complexes in which CSL is not readily identified. To clarify these issues will require an understanding of the spatiotemporal relationships between MTG16 and Notch transcription complex components. Understanding these dynamic relationships may illuminate as-yet-undefined options for regulating responses driven by physiologic and pathological Notch activation.
Because CSL and N-ICDs constitute core components of the canonical Notch transcription complex, our observations intimate an important relationship between MTG family proteins and Notch-dependent phenotypes. This notion is supported by ex vivo fate specification of hematopoietic progenitors in response to Notch-activating stimuli. Expression of T-cell lineage markers CD4 and CD8, normally promoted by Notch signal activation, is impaired in Mtg16−/− hematopoietic progenitors and restored by enforced expression of MTG16. However, an MTG16 mutant deficient in N1-ICD binding (MTG16Δ2C) fails to restore T-cell lineage marker expression in the context of Notch activation. This mutant protein retains each of the NHR regions in MTG16, preserving binding sites for DNA binding proteins, corepressors, and HDACs. These findings not only implicate MTG16 functionally in Notch-dependent hematopoietic lineage decisions but also underscore the importance of the N1-ICD binding region to this effect. Furthermore, the fact that MTG16 functions are important for a Notch-dependent outcome intimates a possible role in T-cell acute lymphoblastic leukemia, where mutations affecting the Notch signaling pathway are found in approximately 60% of patients.
Yet, it has been their contributions to AML pathogenesis that have motivated studies of the MTG family, and we now appreciate their central role in specifying cell fate. MTG family members interact with multiple DNA binding proteins implicated in control of hematopoiesis. Among these are E2A/HEB (16), ZNF652 (31), GFI1 (34) and GFI1B (46), PLZF (35), TAL1/SCL (46), TCF4 (37), LDB1 (19), BCL6 (9) and GATA1 (20). Disregulated function of one or more of these factors could contribute to the fate-specifying defect seen in Mtg16−/− progenitors. Alternatively, Notch-dependent lineage allocation could require cooperation with these or other systems regulated by MTG16. This raises the possibility that MTG proteins may provide platforms to integrate Notch signals with other transcription factors regulated by MTG family members. Because the N-ICD and CSL binding domains are retained in proleukemic fusion proteins generated from t(8;21) and t(16;21) translocations, these interactions may be important in AML pathogenesis. These speculations provide a framework for future experiments to broaden our understanding of Notch signal integration in normal and malignant hematopoiesis.
We are indebted to Thao Dang, Warren Pear, Jan Kitajewski, Jennifer Pietenpol, Juan Carlos Zúñiga Pflücker, and Al Reynolds for their gifts of reagents. We gratefully acknowledge our colleagues Christopher Williams, Utpal Davé, Joseph Amann, and Steve Brandt for their critical evaluation and thoughtful suggestions regarding the manuscript.
We also thank the Vanderbilt Monoclonal Antibody and Vanderbilt Flow Cytometry shared resource facilities for their assistance (supported by NIH grants P30 CA68485 and DK058404). Cell imaging studies were performed in part through the use of the Vanderbilt University Medical Center Cell Imaging shared resource (supported by NIH grants CA68485, DK20593, DK58404, HD15052, DK59637, and Ey008126). M.E.E. is a recipient of a young investigator award from Alex's Lemonade Stand Pediatric Research Foundation, a Hyundai Scholar award from the Hyundai Automotive Corporation, and a career development award from the St. Baldrick's Foundation for Pediatric Cancer Research. J.M. is a recipient of a medical student research fellowship from the Howard Hughes Medical Institute. The work is supported in part by National Institutes of Health grants F32 CA119859 (M.E.E.), F30 HL093993 (A.H.), and R01 HL088494 (S.W.H.).
We declare no competing financial interests.
Published ahead of print on 1 February 2010.
†Supplemental material for this article may be found at http://mcb.asm.org/.