|Home | About | Journals | Submit | Contact Us | Français|
Notch proteins are plasma membrane-spanning receptors that mediate important cell fate decisions such as differentiation, proliferation, and apoptosis. The mechanism of Notch signaling remains poorly understood. However, it is clear that the Notch signaling pathway mediates its effects through intercellular contact between neighboring cells. The prevailing model for Notch signaling suggests that ligand, presented on a neighboring cell, triggers proteolytic processing of Notch. Following proteolysis, it is thought that the intracellular portion of Notch (Nic) translocates to the nucleus, where it is involved in regulating gene expression. There is considerable debate concerning where in the cell Notch functions and what proteins serve as effectors of the Notch signal. Several Notch genes have clearly been shown to be proto-oncogenes in mammalian cells. Activation of Notch proto-oncogenes has been associated with tumorigenesis in several human and other mammalian cancers. Transforming alleles of Notch direct the expression of truncated proteins that primarily consist of Nic and are not tethered to the plasma membrane. However, the mechanism by which Notch oncoproteins (generically termed here as Nic) induce neoplastic transformation is not known. Previously we demonstrated that N1ic and N2ic could transform E1A immortalized baby rat kidney cells (RKE) in vitro. We now report direct evidence that N1ic must accumulate in the nucleus to induce transformation of RKE cells. In addition, we define the minimal domain of N1ic required to induce transformation and present evidence that transformation of RKE cells by N1ic is likely to be through a CBF1-independent pathway.
Notch proteins are cell surface transmembrane-spanning receptors which mediate critically important cellular functions through direct cell-cell contact. Interaction between Notch and its proposed ligand, Delta or Serrate/Jagged, initiates a signaling cascade that governs cell fate decisions such as differentiation, proliferation, and apoptosis in numerous tissue types (2, 9, 27, 57, 58, 69). There are four mammalian Notch genes, Notch1/TAN-1, Notch2, Notch3, and Notch4/Int-3, which are expressed in both overlapping and distinct patterns throughout mammalian tissues (11, 12, 17, 22, 45, 49, 71, 86, 90, 91, 93). Multiple genes for each ligand (6, 29, 38, 53, 60, 62, 64, 77, 87) as well as genes for numerous modulators of Notch signaling (3, 5, 13, 19, 25, 34, 35, 37, 55, 88, 94) have been identified. The growing array of elements in the Notch pathway confers extensive complexity on this signaling system.
Substantial evidence indicates three members of the mammalian Notch family are involved in the generation of neoplasia (8, 17, 50, 54, 67, 70, 74, 93). TAN-1 was identified as a chromosomal translocation, t(7;9)(q34;q34.3), involving the T-cell receptor β locus and the human Notch1 gene in T-cell acute lymphoblastic leukemias (17). This translocation resulted in constitutive expression of aberrant Notch1 transcripts that consist primarily of the intracellular domain. Mouse bone marrow infected with retroviral vectors expressing analogous truncated Notch1/TAN-1 failed to demonstrate transformation in vitro but, when reconstituted into irradiated syngeneic mice, gave rise to T-cell neoplasms with a latency of 11 to 40 weeks (3, 67).
Our lab has demonstrated that an E1A-immortalized baby rat kidney cell line (RKE) can be directly transformed by ectopic expression of constitutive alleles of Notch1 and Notch2 (N1ic and N2ic), indicating these proteins have a direct role in the transformation of cells. Furthermore, N1ic has the ability to transform primary baby rat kidney cells in cooperation with E1A (8).
Genetic evidence indicates Notch1 can cooperate with cellular proto-oncogenes to accelerate tumorigenesis in different tissue types. Retroviral insertion into the Notch1 locus has led to acceleration of tumorigenesis in mouse mammary tumor virus (MMTVD)/c-myc transgenic mice and in MMTV/neu transgenic mice (14, 26). Both insertional events led to expression of aberrant truncated transcripts encoding intracellular sequences of Notch1.
Notch2 and Notch4 have also been implicated in neoplasia. Recombinant feline leukemia virus isolated from a thymic lymphoma was found to contain sequences derived from the feline Notch2 gene (74). Evidence that Notch4 is involved in neoplasia has been derived through studies of MMTV insertional mutagenesis. The int-3 locus was initially identified as a common site of integration for MMTV (24). Proviral insertion in this locus led to expression of Notch4 intracellular sequences. The resulting aberrant Notch4 proteins have been shown to contribute to the generation of mammary carcinoma in the mouse as well as directly transform mouse mammary epithelia in vitro (22, 23, 70, 80).
In all of the described incidences of Notch genes associated with tumorigenesis, the common theme is the constitutive expression of truncated Notch proteins consisting primarily of the intracellular domain that are not tethered to the plasma membrane. Once deleted of extracellular sequences, Notch proteins are thought to be constitutively active; when expressed in both vertebrates and invertebrates, they localize primarily in the nucleus (1, 8, 18, 20, 35, 43, 52, 63, 75, 82).
A current model of Notch signal transduction suggests that ligand binding to the extracellular domain induces proteolytic cleavage that releases the intracellular domain (Nic) which then translocates to the nucleus, where it alters gene expression (2, 20, 36, 44, 75, 81, 83). The mechanism by which Notch influences gene expression is not well understood. In one model, Notch is believed to directly control transcription via interactions with CBF1 [c-promoter binding factor, Su(H) (Suppressor of Hairless), or RBP-Jκ]. CBF1 is a sequence-specific DNA binding protein that functions to repress transcription of cellular genes (15, 85). Ligand-activated Notch proteins are thought to bind to CBF1 in the nucleus, displace a complex of corepressors, and, in turn, lead to activation of target gene transcription (10, 30, 31, 33, 35, 40, 47, 48, 66, 84). However, recent evidence indicates that Notch participates in cellular functions that are independent of CBF1 activation, implying alternative effectors of the Notch signaling pathway (56, 61, 65, 78, 89).
The relevance of Notch nuclear translocation in signal transduction remains controversial (2). A recent report has demonstrated evidence of nuclear accumulation of endogenous Notch upon ligand stimulation in mouse cortical neurons. However, these authors were not able to define a strict correlation between Notch signaling and nuclear translocation (76).
Indirect evidence that Notch might function within the nucleus has been demonstrated in Drosophila, where ligand-activated Notch-GAL4-VP16 fusion proteins were found to activate transcription of an upstream activation sequence-lacZ transgene containing GAL4 DNA binding sites, presumably through proteolytic cleavage and subsequent nuclear translocation (51, 81).
Using the in vitro RKE transformation assay, we report that N1ic must accumulate within the nucleus to elicit biological function. In addition, we define the minimal domain of Nic able to induce transformation of RKE cells, and we report that transformation is likely to be independent of CBF1 activation.
Human Nic deletion mutants were generated with high-fidelity PCR using a full-length human Notch1 expression vector as the template. Two 5′ cloning primers were designed to initiate translation at either Nic residue 1759 or residue 1848. Each contains a BamHI restriction site upstream of the initiator codon and the following sequences: Nic 1759 AAAGGATCCACCATGGCACGCAAGCGCCGGCGCAGTCAT; and Nic 1848, AAAGGATCCACCATGGCACACCTGGATGCCGCTGACCTG (BamHI site is shown in italics, initiator methionine is in bold, and Nic sequences are underlined). Each of the 5′ primers was paired with the following list of 3′ cloning primers to generate C-terminal truncations. Each primer contains an XhoI restriction site (shown in italics), Nic sequences are underlined, and the numbers indicate the C-terminal residue: Nic 2556, GCGCCTCGAGCTTGAACGCCTCCGGGATGCG; Nic 2444, GCGCCTCGAGCACGTCTGCCTGGCTCGG; Nic 2358, GCGCCTCGAGGCTGGCAGCAAGGCTACT; Nic 2202, GCGCCTCGAGGTCCACGGGCGAGAGCAT; Nic 2171, GCGCCTCGAGGCTTCCACAGGCCAGGCCTTT; Nic 2120, GCGCCTCGAGCACCAGGTTGTACTCGTCCAG; Nic 2095, GCGCCTCGAGGTCCATATGATCCGTGAT; and Nic 1991, GCGCCTCGAGGGCATCCAGGTCTGTGGC. PCR products were generated with Vent polymerase (New England Biolabs) using the following cycling parameters: denature at 94°C for 1 min, anneal at 58°C for 1 min, and extend at 72°C for 2 min (25 cycles). PCR products were ligated into pBp283, a pBabe-puro derived retroviral vector containing a C-terminal Myc epitope tag (EQKLISEEDL). Sequences for all constructs were verified by automated sequencing. Inserts containing the Myc epitope tag were subsequently subcloned into pcDNA3.1 (Invitrogen).
Site-directed mutagenesis of Nic was accomplished using the QuickChange mutagenesis system (Stratagene) according to the manufacturer's protocol. The following primer combinations were used to introduce point mutations within the fourth ankyrin repeat: NicANKm1 top, CGCATGCATGATGCGACGGCGCCAGCGATCCTGGCTGCC; NicANKm1 bottom, GGCAGCCAGGATCGCTGGCGCCGTCGCATCATGCATGCG; NicANKm2 top, GACGCCACTGATCCTGGAATTCCGCCTGGCCGTGGAGG; and NicANKm2 bottom, CTCCACGGCCAGGCGGAATTCCAGGATCAGTGGCGTC. Substituted codons are in bold. A 10-amino-acid deletion of Nic residues 2105 to 2114 was engineered using the primers: NicΔ2105–2114 top (CGCGACATCGCACAGGAGGACGAGTACAACCTGGTG) and NicΔ2105–2114 bottom (CACCAGGTTGTACTCGTCCTCCTGTGCGATGTCGCG. 5′ and 3′ boundaries of the deletion are in bold.
Nic constructs containing either a nuclear export signal (NES) or a nuclear localization signal (NLS) were constructed by subcloning Nic fragments into modified pBp283 vectors. To generate pBp283NES, oligonucleotides encoding the following amino acids were annealed and ligated into pBp283: LALKLAGLDLEQKLISEEDL (NES sequence is in bold and Myc epitope is underlined ). pBp283NLS was constructed to encode the following residues: PKKKRKVEQKLISEEDL (NLS sequence is in bold, and Myc epitope is underlined ).
Generation of baby rat kidney cells immortalized with E1A (RKE cells) has been previously described (8). RKE, HeLa, Bosc23 (68), and 293T cells were incubated at 37°C in 5.5% CO2 and propagated in Dulbecco's modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum (FBS), 2 mM l-glutamine, 100 U of penicillin per ml, and 100 μg of streptomycin per ml (Life Technologies). For focus formation assays 106 RKE cells were seeded onto 100-mm-diameter dishes 12 h prior to transfection; 10 μg of pBp283 expression construct DNA was diluted in 1.5 ml of OptiMEM (Life Technologies) containing 15 μl of Lipofectamine (Life Technologies) and incubated with cells for 8 h at 37°C in 5.5% CO2 in a total volume of 6.5 ml of OptiMEM. At 48 h posttransfection, cells were trypsinized and split; half of the cells were selected in puromycin (2 μg/ml) to score transfection efficiency, while the remaining half were maintained in culture and fed fresh medium twice per week. After 2 weeks, the puromycin-selected colonies were fixed in methanol and stained with a solution of 0.5% methylene blue in 70% isopropanol, and drug-resistant colonies were counted. After 4 weeks, the focus assay plates were processed as described above and foci were counted. Focus formation efficiency was calculated as the number of foci formed divided by the number of puromycin-resistant colonies for each assay.
Clonal lines of transformed RKE cells were established by trypsin dissociation of foci isolated with cloning cylinders. Isolated cells were propagated in DMEM supplemented with the appropriate selectable marker (2 μg of puromycin per ml or 400 μg of neomycin per ml). Polyclonal cell lines for mutant constructs were established by transfection of expression vector DNA and subsequent drug selection. After 2 weeks, drug-resistant colonies were pooled and propagated under selection conditions.
Transformed RKE cells were assayed for anchorage-independent growth using the following protocol. A total of 5.0 × 103 cells from the indicated stable cell lines were suspended in 6 ml of DMEM containing 20% FBS and 0.35% low-melting-point agarose. The suspension was overlaid onto a 6-ml base of DMEM, 10% FBS, and 0.7% low-melting-point agarose in 60-mm-diameter dishes. Plates were incubated at 37°C in 5.5% CO2 for 3 weeks and then photographed with a Zeiss IM-35 inverted microscope at a magnification of ×100.
For transient expression, 3.0 × 106 Bosc23 or 293T cells were transfected with 5 μg of the indicated plasmid DNA using calcium phosphate precipitation. At 48 h posttransfection, cell lysates were prepared in standard lysis buffer (150 mM NaCl, 50 mM HEPES [pH 7.4], 1 mM EDTA, 0.5 mM dithiothreitol, 10% glycerol, 1% NP-40) supplemented with the protease inhibitors Pefabloc (2 mM), leupeptin (5 μg/ml), and aprotinin (2 μg/ml). Proteins were separated on sodium dodecyl sulfate (SDS)–8% polyacrylamide gels and immobilized on a nitrocellulose membrane (Schleicher & Schuell) by wet transfer, and equal loading was verified by staining with Ponceau S. Notch proteins were detected with the indicated antibodies using standard immunoblotting procedures. Proteins were visualized with enhanced chemiluminescence (ECL kit; Amersham) and exposure to X-ray film.
Cells expressing the indicated Nic mutant were seeded at 5.0 × 104 cells per well on four-chamber glass slides (LabTech) 12 h prior to processing. Unless noted otherwise, all steps were performed at 4°C. Cells were washed twice with ice-cold phosphate-buffered saline (PBS) and then fixed for 30 min in 3% paraformaldehyde. Cells were washed in PBS supplemented with 3% FBS, permeabilized for 5 min in 0.2% Triton X-100, and washed in PBS–3% FBS. Cells were incubated for 1 h at room temperature in a blocking solution consisting of PBS–3% FBS and 0.5% Tween 20. Where indicated, either 9E10 (1:500) or bTAN15A (1:2) diluted in PBS–3% FBS was added to each chamber and incubated with rocking at room temperature for 1 h. Slides were washed in PBS–3% FBS and were then incubated with an appropriate Cy3-conjugated secondary antibody (1:1,000 dilution) for 30 min at room temperature. Following final washes, cells were mounted in 70% glycerol containing 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (DAPI; 0.1 μg/ml), and Nic proteins were visualized and photographed on a Zeiss Axiophot fluorescent microscope with a Hamamatsu digital camera at a magnification of ×400.
A luciferase reporter plasmid (8x-CBF1-luc) containing eight copies of the CBF1 DNA binding consensus sequence (CGTGGGAA) cloned into the pGL2 reporter vector (Promega) (21) (provided by P. D. Ling, Baylor College of Medicine) was used to analyze the ability of Nic deletion constructs to interact with CBF1 within cells. For luciferase activity, 2.5 × 105 HeLa cells were seeded in six-well plates and were cotransfected with 400 ng of 8x-CBF1-luc, 400 ng of RL-TK (renilla luciferase control plasmid; Promega), and 800 ng of the indicated expression plasmid. Cells were transfected using 8 μl of Lipofectamine in a total volume of 2 ml of OptiMEM; 48 h posttransfection, cells were lysed and luciferase and renilla light units were measured in a Xylux Femptomaster FB 12 luminometer according to manufacturer's suggested protocol for the Dual Luciferase assay (Promega). Luciferase values were corrected for transfection efficiency by dividing luciferase light units by renilla light units (expressed as relative luciferase units).
Whole cell lysates were made from stable RKE cell lines expressing the indicated mutant construct. Approximately 2 × 107 cells were washed twice in ice-cold PBS, scraped into centrifuge tubes, collected by centrifugation, and lysed on ice for 20 min in 1 ml of standard lysis buffer. Cell debris was cleared by centrifugation for 30 min at 10,000 × g in a Heraus benchtop 4°C centrifuge. The resulting supernatant was divided into two 500-μl aliquots. To determine protein expression, one aliquot was immunoprecipitated with Nic polyclonal antiserum 925, a rabbit polyclonal antiserum directed against residues 1759 to 2095 of human Nic. Immune complexes were collected on protein A-Sepharose beads, washed in standard lysis buffer, and separated by SDS-polyacrylamide gel electrophoresis (PAGE) on an 8% gel, and Nic proteins were detected by Western blotting against the Myc epitope. The other 500-μl aliquot was precleared for 2 h with 10 μg of glutathione S-transferase (GST)–Sepharose beads. Beads were removed following brief centrifugation, and the lysates were then split into two equal aliquots (250 μl); one received approximately 10 μg of GST-CBF1 fusion protein-bound beads, and the other received approximately 10 μg of GST-bound beads. Complexes were formed at 4°C with constant rocking for 8 h. Following incubation, beads were washed five times in standard lysis buffer, and protein complexes were separated by SDS-PAGE (8% gel) Nic proteins were detected by Western blotting against the Myc epitope.
Previously we reported that ectopic expression of Nic can transform primary baby rat kidney cells in cooperation with E1A or a cell line immortalized by E1A (RKE). Moreover, cells transformed by Nic exhibit anchorage-independent growth and are tumorigenic in nude mice (8).
To determine the functional domains in Nic that are necessary and/or sufficient for neoplastic transformation, we generated a series of N- and C-terminal deletion mutants of Nic (see Materials and Methods for details; Fig. Fig.1A).1A). The boundaries of these deletion mutants were designed to systematically eliminate primary sequence motifs in Nic. NicΔ2444 is deleted for the entire PEST region at the C-terminal tail of Nic. NicΔ2358 is deleted of sequences that encode both the PEST and OPA motifs. NicΔ2202 is deleted of 156 additional residues N terminal to the OPA motif and terminates adjacent to a putative NLS. An additional C-terminal truncation (NicΔ2095) removes all 461 amino acids C terminal to the ankyrin repeat (ANK) domain, leaving the ANK domain sequences intact. The sole N-terminal deletion (NicΔR) removes two putative NLSs and the binding site for the mammalian Su(H) homologue CBF1 (RAM domain). Deletion of the RAM domain has previously been shown to abolish the ability of Nic to interact with CBF1 (4, 31, 75, 84). C-terminal deletions were made in the context of both the full-length Nic and the N-terminally deleted form of Nic. All deletion mutants encode a C-terminal Myc epitope tag for detection with 9E10.
To verify that these constructs expressed stable proteins of the anticipated size, expression plasmids were transiently transfected into the retroviral packaging cell line Bosc23. Western blot analysis of cellular lysates from transfected Bosc23 cells revealed that all deletion constructs expressed the predicted size protein (Fig. (Fig.11B).
To assay for focus formation, plasmids encoding Nic C-terminal deletion mutants expressed from a retroviral long terminal repeat were transfected into RKE cells. At 48 h posttransfection, cells were harvested and split into two plates. Half of the cells were plated into medium containing puromycin (2 μg/ml) to select for the linked drug resistance marker (Fig. (Fig.2A,2A, bottom row). Cells were maintained in culture for 2 weeks until drug-resistant colonies were detected. These plates were used to control for total number of transfected cells and to determine transformation efficiency for each deletion construct (Table (Table1).1). The remaining transfected cells were plated and monitored for focus formation as described in Materials and Methods (Fig. (Fig.2A,2A, top row).
Deletion of 354 C-terminal residues flanking the putative NLS (NicΔ2202) did not greatly affect ability of Nic to induce foci. Cells transfected with NicΔ2202 formed foci with similar efficiency as cells transfected with Nic (Table (Table1);1); however, NicΔ2202-induced foci were consistently smaller than Nic-induced foci (Fig. (Fig.2A).2A). We found that a C-terminal deletion that removes both the OPA and PEST domains (NicΔ2358) resulted in an increase from 46.9% to 67.1% in the efficiency of focus formation compared to full-length Nic (Table (Table1).1). Although more efficient in transformation, the foci resulting from transfection of RKE cells with NicΔ2358 are indistinguishable in appearance from Nic-induced foci (Fig. (Fig.2A,2A, top row). Removal of the PEST domain (NicΔ2444) had no effect on Nic-induced focus formation (Table (Table1;1; plate not shown). In contrast, NicΔ2095 resulted in a complete loss of transformation when transfected into RKE cells (Fig. (Fig.2A).2A). Transfection efficiencies for these constructs were essentially equal as determined by selection in puromycin (Fig. (Fig.2A,2A, bottom row).
We determined if deletion of amino-terminal NLSs and the binding site for the transcriptional regulatory protein CBF1 (RAM domain) had any consequence on the ability of Nic to transform RKE cells. Nic constructs comprised of both C-terminal and N-terminal deletions were assayed for focus formation (Fig. (Fig.2B,2B, top row). Each NicΔR mutant was approximately 70 to 85% as efficient in focus formation as their cognate Nic deletion clone which contains the RAM domain (Table (Table1).1). The growth characteristics of NicΔR-induced foci are equivalent to Nic deletion constructs pictured in Fig. Fig.22A.
To determine protein expression and the ability of the transformed cells to form colonies in semisolid media, individual foci were isolated for each transforming Nic mutant. Figure Figure2C2C shows the results of Western blot analysis of cellular lysates from transformed cell lines. All isolated foci expressed proteins of the predicted size, indicating that expression of Nic deletion constructs are responsible for the observed transformed phenotype.
Cells from transformed RKE lines were seeded into semisolid media to assay for anchorage-independent growth. Results obtained from this analysis revealed that all of the Nic and NicΔR deletion mutants that formed foci were equally capable of forming colonies in soft agar (Fig. (Fig.2D).2D). The parental RKE cells and cells stably expressing green fluorescent protein (GFP) failed to form colonies (Fig. (Fig.2D).2D). No apparent difference in colony size was observed between Nic and NicΔR mutant constructs. Furthermore, RKE cells expressing NicΔR were tumorigenic when injected into nude mice (data not shown).
The results obtained from focus formation analysis and soft agar assays indicate that the minimal transformation domain of Nic consists of 355 amino acids encompassing residues 1848 through 2202. These sequences encode the ANK domain and 107 additional C-terminal residues. The minimal transformation domain will subsequently be referred to as the TFD (transformation domain).
To define elements within the TFD that are required for focus formation, we engineered C-terminal deletion mutants in the region between NicΔ2202 and NicΔ2095. In addition, we constructed a small internal deletion and two site-directed mutations within the context of Nic (Fig. (Fig.3A).3A). NicΔ2171, which is deleted of only 31 N-terminal amino acid residues from NicΔ2202, was severely impaired for focus formation (Fig. (Fig.3B).3B). Transfection of RKE cells with this construct resulted in only 18.1% efficiency of focus formation (Table (Table1).1). However, the size of foci generated with NicΔ2171 was greatly reduced compared to that of NicΔ2202 (Fig. (Fig.3B;3B; compare to Fig. Fig.2A).2A). NicΔ2120, which expresses a mutant protein missing an additional 51 amino acids N terminal to 2171, resulted in a phenotype identical to that of NicΔ2171 (Table (Table1;1; focus plate not shown).
NicΔ2105–2114 has a 10-amino-acid deletion (2105-RMHHDIVRLL-2114) located between NicΔ2120 (a weakly transforming construct) and NicΔ2095 (a nontransforming construct). Deletion of these 10 amino acids from Nic is sufficient to eliminate transforming activity as determined by the focus assay (Fig. (Fig.3B).3B). The requirement for an intact ANK domain was determined by making two separate mutants (NicANKm1 and NicANKm2) that encode amino acid substitutions in the fourth ANK repeat (40, 42, 43, 79). NicANKm1 has a three-amino-acid substitution that changes the Nic primary sequence from 1995-DGTTPLI-2001 to 1995-DATAPAI-2001. In NicANKm2, amino acids 2003 to 2005 are changed from 2002-LAAR-2006 to 2002-LEFR-2006. Both of these ANK mutants were incapable of transforming RKE cells (Fig. (Fig.33B).
Subcellular localization of Nic constructs in RKE cells was determined by IF in polyclonal cell lines expressing Nic mutant constructs. For IF, cells were seeded onto plastic chamber slides and Nic mutant proteins were detected using 9E10 and visualized with a Cy3-conjugated secondary antibody as described in Materials and Methods (Fig. (Fig.4).4).
Previously we reported that Nic is primarily localized to the nuclei of RKE cells (8). Analysis of the C-terminal deletion mutants revealed that all mutant proteins resided primarily within the nucleus to the same extent as Nic (Fig. (Fig.4A).4A). In contrast, deletion of both N-terminal putative NLSs and the RAM domain (NicΔR) resulted in an increase of cytoplasmic staining, though each NicΔR construct retained prominent nuclear immunoreactivity.
When subcellular localization was examined for weakly transforming and nontransforming Nic mutants, we observed that all Nic mutant proteins localized to the nucleus regardless of their ability to induce focus formation in RKE cells (Fig. (Fig.44B).
Activation of CBF1-dependent transcription is considered to be an important component of the Notch signal transduction pathway. We analyzed the requirement of CBF1-mediated transcriptional activation using transient transfection assays in HeLa cells (Fig. (Fig.5A).5A). Transfection of 8x-CBF1-luc with a GFP expression plasmid or wild-type Notch1 resulted in low basal activity. In contrast, transfection of a Nic expression vector resulted in a 45-fold increase in luciferase activity over that obtained with GFP (expressed as relative luciferase units in Fig. Fig.5A).5A). Transfection of vectors expressing deletion mutants NicΔ2202 and NicΔ2120 resulted in a 30- and 6-fold increases of luciferase activity compared to the GFP control.
Deletion of the RAM domain in NicΔR mutants abolished their ability to activate CBF1-mediated transcription from the 8x-CBF1 reporter (Fig. (Fig.5A).5A). All NicΔR mutants tested in this assay exhibited luciferase activities at basal levels, demonstrating loss of functional association with CBF1.
To establish that NicΔR constructs do not physically associate with CBF1, we tested for protein-protein binding interactions using a GST pull-down assay. Whole cell lysates prepared from transformed RKE cells were incubated with either GST or GST-CBF1 adsorbed to glutathione-agarose beads and processed as described in Materials and Methods. Using a GST-CBF1 fusion protein, we could affinity precipitate only Nic deletion proteins that encode the RAM domain (Fig. (Fig.5B).5B). Deletion of the RAM domain completely abolished the ability of NicΔR proteins from transformed RKE cells to physically interact with GST-CBF1. To control for the presence of Notch proteins in this assay, an equivalent amount of each lysate was immunoprecipitated with the anti-Notch1 polyclonal antiserum 925 and detected by Western blotting against the Myc epitope (Fig. (Fig.22C).
Mutational analysis of the TFD revealed that site-directed mutations in the fourth ANK repeat (NicANKm1 and NicANKm2) and a 10-amino-acid deletion (NicΔ2105–2114) resulted in transformation-defective Nic proteins (Fig. (Fig.3).3). Since each of these constructs encodes the RAM domain, we determined if these Nic site mutants maintained molecular interactions with CBF1. Previous studies have shown that the NicANKm1 and NicANKm2 point mutations ablate Nic mediated transcriptional activation of a CBF1-responsive reporter (35, 40, 48). As predicted, cotransfection of NicANKm1 and NicANKm2 expression plasmids with 8x-CBF1-luc resulted in negligible reporter activity in HeLa cells (Fig. (Fig.6A).6A). Similarly, when assayed for CBF1 reporter activity, NicΔ2105–2114 also failed to appreciably activate transcription from the reporter gene (Fig. (Fig.6A).6A).
To determine if loss of reporter activity was due to a failure of Nic mutants to physically associate with CBF1, we assayed for protein binding using the GST-CBF1 pull-down experiment (Fig. (Fig.6B).6B). Whole cell lysates were prepared from Bosc23 cells transiently transfected with either NicANKm1, NicANKm2, or NicΔ2105–2114 and assayed for GST-CBF1 binding as described in Materials and Methods. In contrast to deletion of the RAM domain, NicANKm1, NicANKm2, and NicΔ2105–2114 mutants bound CBF1 in GST pull-down experiments as well as Nic (Fig. (Fig.6B).6B). This result indicates that binding CBF1 is not sufficient to either activate transcription from a reporter or induce focus formation.
We showed that Nic and all deletion mutants of Nic are primarily localized to the nucleus of RKE cells, regardless of their ability to transform RKE cells (Fig. (Fig.4).4). To address the significance of nuclear localization in Nic-mediated transformation, we used a mechanism to efficiently control subcellular localization of Nic proteins. We created a series of NicΔ2444 and NicΔR-2444 constructs that encode at their C termini either a cyclic AMP-dependent protein kinase inhibitor (PKI)-like NES (amino acids LALKLAGLDL ) or the simian virus40 large T NLS (amino acids PKKKRKV ) followed by the Myc epitope tag. Addition of an ectopic NES should result in active export of NicΔ2444 and NicΔR-2444 from the nucleus to the cytoplasm. The NLS should not have any effect on nuclear localization of NicΔ2444 but should rescue the loss of nuclear localization associated with the ΔRAM mutation in NicΔR-2444.
To test the ability of these constructs to induce focus formation, we transfected plasmids expressing either the NES or NLS derivatives into RKE cells and performed the focus assay. The ability of NicΔ2444 to form foci in RKE cells was significantly diminished by addition of the NES (Fig. (Fig.7A,7A, top row: +NES, Table Table1).1). In contrast, addition of the NLS had no effect on transformation by NicΔ2444 (Fig. (Fig.7A,7A, top row, +NLS; Table Table1).1). When combined with the N-terminal RAM domain deletion in NicΔR-2444, the effect of the NES was more pronounced, resulting in a dramatic loss of focus formation (Fig. (Fig.7A,7A, bottom row, +NES; Table Table1).1). Addition of an NLS to NicΔR-2444 increased transformation efficiency to that observed with NicΔ2444 (Fig. (Fig.7A,7A, bottom row, +NLS; Table Table1).1). This indicates that the reduction in transformation efficiency associated with the ΔRAM Notch mutants is a result of decreased nuclear localization and not with a loss of CBF1 binding.
Subcellular localization of +NES and +NLS constructs was determined by IF in polyclonal RKE cell lines. Addition of the NES to NicΔ2444 resulted in more diffuse total cell distribution compared to the exclusively nuclear localization of NicΔ2444; however, it retained prominent nuclear staining (Fig. (Fig.8A,8A, +NES). Addition of the NLS had no effect on subcellular localization of NicΔ2444 (Fig. (Fig.8A,8A, +NLS). Analysis of NicΔR-2444 subcellular localization revealed an increase in cytoplasmic staining compared to NicΔ2444 (Fig. (Fig.8A).8A). Addition of the NES to NicΔR-2444 resulted in a complete exclusion of the protein from the nucleus (Fig. (Fig.8A,8A, +NES). In contrast, addition of the NLS to NicΔR-2444 rescued the decrease in nuclear localization associated with the RAM domain deletion (Fig. (Fig.8A,8A, +NLS).
To authenticate that the NicΔR-2444NES construct was being exported via the nuclear export machinery, we treated RKE cell lines with the nuclear export inhibitor leptomycin B (Fig. (Fig.8B)8B) (46). Treatment of the NicΔR-2444NES expressing polyclonal cell line with leptomycin B resulted in nuclear accumulation of the protein, whereas treatment with vehicle had no effect on subcellular localization as determined by indirect IF (Fig. (Fig.8B,8B, center column, +NES). Subcellular localization of NicΔR-2444 and the NLS derivative was not affected by treatment with either leptomycin B or vehicle (Fig. (Fig.8B,8B, right and left columns). Taken together, these results indicate the NES derivatives are functioning through the NES pathway and fail to proficiently transform RKE cells due to a lack of nuclear localization.
To determine if localization of NicΔR-2444NES and NicΔ2444NES was truly reflective of the subcellular distribution of these proteins and not a result of selecting polyclonal RKE cell lines, we performed IF on transiently transfected HeLa cells. HeLa cells were transfected with the indicated expression vectors, and subcellular localization of the Nic proteins was determined as described above (Fig. (Fig.8C).8C). As we demonstrated for RKE cells, NicΔ2444NES retained prominent nuclear localization and to a lesser extent cytoplasmic localization. The subcellular distribution of NicΔ2444NES was now remarkably similar to that of NicΔR-2444, indicating that addition of an NES is equivalent to removal of the RAM-NLS (Fig. (Fig.8C).8C). Addition of an NES to NicΔR-2444 resulted in the complete redistribution of the protein to the cytoplasm, consistent with the results obtained with polyclonal RKE lines.
Nuclear accumulation is a hallmark of ectopically expressed Nic proteins in diverse cellular systems from invertebrates to vertebrates (2, 57, 58). This property of Notch proteins has confounded interpretations of where within the cell this plasma membrane-spanning receptor exerts its influence. Previously, we have shown that permanently tethering Nic to the plasma membrane as a CD8-Nic fusion protein abolishes Nic transformation activity in RKE cells (8). Drosophila expressing myristylated-Nic derivatives that predominantly localize to membranes also have a loss of Nic activity (81). These results support the hypothesis that Notch does not function at the membrane and must be released from the membrane to signal. However, neither of these approaches was able to discriminate between cytoplasmic or nuclear signaling events required for Nic activity. To resolve the issue of nuclear versus cytoplasmic signaling, we engineered Nic proteins that encode a functional NES at the C terminus. Our hypothesis was that if nuclear localization was required for transformation, addition of an NES to Nic would block nuclear accumulation and thus Nic would fail to transform RKE cells. Our data clearly demonstrate that active export of Nic molecules from the nucleus to the cytoplasm results in loss of transforming activity.
Analysis of the subcellular localization of NicΔ2444NES in a polyclonal RKE cell line revealed that this protein retained prominent nuclear localization. When transfected into RKE cells, this molecule is capable of eliciting focus formation, albeit at a lesser efficiency that NicΔ2444. However, the NicΔR-2444NES clone is completely excluded from the nucleus and is severely compromised in the transformation assay. One explanation for the incomplete penetrance of the NicΔ2444NES clone is that this clone retains a strong NLS within the RAM domain that cannot be completely overcome by the addition of a single NES. Recent reports have demonstrated that in cases where proteins are actively exported out of the nucleus, there is a balance between NLS and NES strength (16, 28). This is likely the case in our situation since removal of a strong NLS in the RAM domain in NicΔR-2444NES results in a complete loss of nuclear staining. Moreover, analysis of subcellular localization in polyclonal RKE cells might overemphasize nuclear localization. Since we know that NicΔ2444NES is capable of transforming RKE cells (at a low level), it is reasonable to suspect that polyclonal lines contain a large percentage of cells that have selected for active nuclear NicΔ2444NES proteins. In support of this explanation, transient expression of this clone in HeLa cells reveals a whole cell distribution of NicΔ2444NES more intense than what we observe in the polyclonal RKE line (Fig. (Fig.8C).8C). Taken together, these results strongly establish that neoplastic transformation of RKE cells by Nic requires nuclear localization.
Recently, we have characterized chimeras of Nic and the hormone binding domain of the estrogen receptor (Notchic-ER) that requires stimulation with 4-hydroxytamoxifen (OHT) to induce transformation of RKE cells (C. Ronchini and A. J. Capobianco, submitted for publication). In the absence of OHT, the Notchic-ER chimeras are diffusely distributed throughout the cell and are unable to induce transformation. When stimulated with OHT, Notchic-ER proteins induce transformation of RKE cells and the proteins accumulate in the nucleus. This supports our argument that nuclear accumulation of Nic is required for its biological effects.
Deletion analysis of Nic demonstrates that the C-terminal sequences encoding the PEST and OPA domains are not required for transforming activity. Deletions of these sequences from mouse Notch1 have previously been shown to have no effect of Nic inhibition of granulocytic differentiation or myogenesis (7, 43, 59). Our data also indicate that deletion of the OPA and PEST domains results in an increase in transforming activity, indicating that these domains might mediate negative regulation of Nic within RKE cells.
We have defined the TFD as amino acids 1848 to 2202 of human Notch1. This domain consists primarily of the ankyrin repeats and an additional 107 C-terminal residues. Furthermore, we have shown that the 107 amino acids juxtaposed to the ANK domain are sensitive to mutation, indicating that their integrity as a domain is necessary. Notch molecules corresponding to the TFD are endowed with intrinsic signaling ability in other systems (7, 43, 56, 59, 72, 73, 82), indicating the conserved role of the ANK domain in Nic-mediated biological effects.
C-terminal deletion and site-directed mutagenesis have implicated the region located immediately downstream of the ANK domain as a region important for transformation. C-terminal deletion of 31 residues into the TFD (N terminal to amino acid 2202) greatly attenuated focus formation, without affecting nuclear localization. Furthermore, a site-directed deletion of 10 amino acids within this region was sufficient to destroy the ability of Nic to transform RKE cells. These results indicate important functional interactions are likely to occur within this region.
Previously we reported that N2ic was considerably weaker as a transforming molecule than N1ic in the focus assay (8). It is likely that sequences within the TFD account for this difference. Sequence comparison between the TFD of Notch1 and the corresponding region in Notch2 reveals that these proteins share approximately 66% identity, considering similarity it is 74%. Interestingly, this region has been reported to mediate differences in signaling between Notch1 and Notch2 in granulocyte differentiation by granulocyte-macrophage colony-stimulating factor (GM-CSF) and granulocyte colony-stimulating factor (G-CSF), in that GM-CSF signals through N2 and G-CSF signals through N1 to inhibit differentiation (7).
CBF1-dependent versus independent signaling pathways is a topic of considerable debate within the Notch field (2, 40, 61). Based on published reports, we engineered an Nic mutant with an N-terminal deletion that removes both primary and secondary CBF1 binding sites (NicΔR) within the RAM domain (31, 32, 48, 61, 84). Consistent with these reports, we demonstrate that NicΔR mutants no longer retain the ability to interact with CBF1 from analysis of protein-protein interaction and CBF1 reporter assays. NicΔR constructs retain the ability to induce transformation when expressed in RKE cells even though they are unable to interact with CBF1. Based on this evidence, we conclude that transformation of RKE cells by Nic does not require direct activation of CBF1. Moreover, we have demonstrated that the ability of Nic to bind CBF1 is not sufficient to induce transformation. We analyzed several Notch constructs that had point mutations within the ANK domain. These ANK mutants were defective for transformation and for the ability to activate the CBF1 reporter in HeLa cells. These constructs all contained the RAM domain and were able to bind to CBF1 in a GST pull-down experiment (Fig. (Fig.6B).6B). Similar ANK mutants have been reported to abolish Nic activation of CBF1-responsive reporter assays (35, 40, 48) and fail to inhibit muscle differentiation in vitro (40, 43).
Several lines of evidence indicate that the C terminus of Nic encodes an intrinsic transcriptional activation domain. These observations were primarily made by experiments utilizing either LexA- or GAL4-Nic fusion proteins (31, 41, 48). Data from these reports define the Nic transcriptional activation domain as sequences between the OPA domain and the C-terminal putative NLS (C terminal to amino acid 2202) (48). Our CBF1 reporter data confirm that the C-terminal portion of Nic does contain a transcriptional activation domain, but with one consideration. We show that reporter activity descreases with C-terminal truncations between the OPA domain and the NLS (amino acid 2202), which is in agreement with the above reports. However, our minimal transforming domain (amino acids 1848 to 2202) does not contain the C-terminal residues that have been attributed to the intrinsic transcriptional activation potential of Nic. While not arguing against a role for Nic in transcriptional activation, our data instead imply that amino acids within the TFD (i.e., N-terminal to amino acid 2202) appear to be directly responsible for transcriptional activation.
We have concluded that CBF1 activation is not a requirement for transformation of RKE cells. However, we support the hypothesis that a transcriptional activation domain is required for Notch activity. All of our data indicate that the RAM domain is solely responsible for mediating the interaction with CBF1 and that deletion of this domain does not affect the ability of Nic to transform cells. How can we justify the apparent correlation between loss of CBF1 activity and transformation for the ANK mutants and the 10-amino-acid deletion in the TFD? If we assume that the CBF1 reporter assay in HeLa cells is simply reflective of transcriptional activation and not of a CBF1 function in RKE cells (i.e., similar to a GAL4 fusion experiment), then we can conclude that the correlation is between transcriptional activation and transformation of RKE cells. Furthermore, we suggest that the transcriptional activation required for transformation is not mediated through CBF1.
A recent report describing in vitro transformation of a mouse mammary epithelial cell line (HC11) by intracellular Notch1 concluded that transformation required the N-terminal putative NLS and the RAM domain (14). Data concerning subcellular localization of Nic constructs of functional association with CBF1 within HC11 cells were not presented. However, the authors did attempt to show activation of a potential cellular target of Nic via CBF1 in Nic-transformed HC11 cells. The ERBB2 promoter contains CBF1 binding sites and can be activated by Nic expression in reporter assays (10). The study by Dievart et al. was unable to detect elevated ERBB2 message (14), indicating Nic may not target CBF1 repressed genes within HC11 cells to induce transformation. Alternatively, the requirement for CBF1 might be cell type specific in that immortalized mammary epithelial cells but not RKE cells require CBF1 activity. One could imagine that in RKE cells, expression of E1A somehow provides a redundant function for CBF1 activity, and therefore additional CBF1 activity from interaction with Nic is not necessary.
We thank members of the Capobianco lab for support and technical assistance. We also thank David Robbins and his laboratory for insightful comments on this work. We are grateful to P. D. Ling (Baylor College of Medicine) for providing the 8x-CBF1-luc reporter and to Minoru Yoshida (University of Tokyo) for generously supplying leptomycin B. We thank W. Pear (University of Pennsylvania) for Bosc23 cells.
This work was funded in part by ACS grant LBC99465 (to A.J.C.).