γ-secretase inhibitors (GSIs) can block NOTCH receptor signaling in vitro and therefore offer an attractive targeted therapy for tumors dependent on deregulated NOTCH activity. To clarify the basis for GSI resistance in T cell acute lymphoblastic leukemia (T-ALL), we studied T-ALL cell lines with constitutive expression of the NOTCH intracellular domain (NICD), but that lacked C-terminal truncating mutations in NOTCH1. Each of the seven cell lines examined and 7 of 81 (8.6%) primary T-ALL samples harbored either a mutation or homozygous deletion of the gene FBW7, a ubiquitin ligase implicated in NICD turnover. Indeed, we show that FBW7 mutants cannot bind to the NICD and define the phosphodegron region of the NICD required for FBW7 binding. Although the mutant forms of FBW7 were still able to bind to MYC, they do not target it for degradation, suggesting that stabilization of both NICD and its principle downstream target, MYC, may contribute to transformation in leukemias with FBW7 mutations. In addition, we show that all seven leukemic cell lines with FBW7 mutations were resistant to the MRK-003 GSI. Most of these resistant lines also failed to down-regulate the mRNA levels of the NOTCH targets MYC and DELTEX1 after treatment with MRK-003, implying that residual NOTCH signaling in T-ALLs with FBW7 mutations contributes to GSI resistance.
The mammalian NOTCH proteins are heterodimeric transmembrane receptors that control cell proliferation, apoptosis, and cell fate during the development of diverse cellular lineages (1). Aberrant NOTCH signaling has been extensively linked to cancer and development. In mouse models, constitutive NOTCH signaling contributes to the genesis of breast cancer, medulloblastoma, and T cell leukemia (2–9), whereas in human cancer its role is best exemplified by T cell acute lymphoblastic leukemias (T-ALLs). Human NOTCH1 was first discovered as a gene activated at the breakpoint of the t(7;9), a very rare chromosomal translocation that fuses the intracellular form of NOTCH1 to the T cell receptor β locus in lymphoblasts of T-ALL patients (10). Recently, 50% of human T-ALL cell lines and primary patient samples were shown to harbor activating mutations in NOTCH1 that result in aberrant NOTCH signaling (11). Although mutations that directly activate NOTCH receptors have not been identified in other types of human cancers, there is abundant evidence to support the importance of deregulated NOTCH activity in the development of ovarian cancer (12), breast cancer (13), anaplastic large cell lymphoma and Hodgkin disease (14), melanoma (15), gliomas (16), lung carcinomas (17, 18), and cancers of the pancreas (18) and prostate (19). Hence, modulation of the NOTCH signaling cascade at one or more points could short-circuit this pathway in NOTCH-supported tumors, leading to clinically important antitumor effects.
Blocking the intramembranous cleavage of NOTCH is an especially attractive strategy of targeted therapy. When the NOTCH receptor is recognized by its membrane-bound ligand, a conformational change exposes the receptor to sequential rounds of protease cleavage. Binding of the ligand results in proteolytic cleavage of the receptor, first outside the cell by TNF-α–converting enzyme and then by the γ-secretase membrane protease complex, releasing the NOTCH intracellular domain (NICD), which translocates to the nucleus where it regulates the expression of its target genes, including MYC and DELTEX1 (20–23). Small molecule inhibitors of γ-secretase activity are now available that effectively inhibit NOTCH signaling in vitro. One commercial product, compound E, induces growth arrest in several different T-ALL cell lines by inhibiting the NOTCH pathway (11). Recently, we showed that treatment of T-ALL cells with the MRK-003 γ-secretase inhibitor (GSI) results in prolonged cell cycle arrest followed by apoptosis (24).
Despite the promise of GSI therapy for tumors driven by aberrant NOTCH signaling, most human T-ALL cell lines are resistant to these agents and grow normally despite GSI treatment. Thus, to establish the molecular basis of GSI resistance in tumor cells, we used T-ALL cell lines as a model system to test the ability of GSI treatment to reduce cellular levels of NICD, as well as its transcriptional targets MYC and DELTEX1. We found that mutations affecting the FBW7 ubiquitin ligase occur in GSI-resistant cell lines and are associated with sustained NOTCH signaling. Moreover, we found FBW7 mutations in primary T-ALL samples, and the mutational spectrum suggests that they produce dominant-negative FBW7 alleles. Our findings implicate FBW7 mutations in both the pathogenesis of T-ALL and leukemic cell resistance to GSIs.
MRK-003 treatment leads to Notch-dependent antiproliferative effects in a subset of T-ALL cell lines
To inhibit NOTCH-mediated signal transduction, we treated each of 20 T-ALL cell lines with the Merck GSI MRK-003 (24) at 1 μM or with DMSO (vehicle control) for 7 d. To study the effects of GSI treatment, we analyzed the cell counts, cell cycle profiles, and apoptosis at various time points after treatment. We did not observe any effects on proliferation or viability at 6 h, 24 h, or 3 d in any of the 20 lines. Five of the T-ALL cell lines (DND41, Koptk1, ALL-SIL, HPB-ALL, and TALL1) were sensitive to GSI treatment, exhibiting two- to threefold fewer live cells in the GSI-treated flasks compared with DMSO-treated flasks on day 7 of treatment (Table I). Cell cycle analysis at that time revealed a G0/G1 cell cycle arrest with decreased cells in the S phase in each of the five lines (representative histograms are shown for one line in Fig. 1 A). Annexin V staining showed a two- to fourfold increase in the percentage of cells undergoing apoptosis in the five GSI-sensitive cell lines, indicating that MRK-003 acts by inducing apoptosis as well as by blocking cell growth (Fig. 1 B). Importantly, the altered proliferation, cell cycle arrest, and increase in apoptosis could be rescued by overexpressing the NICD, demonstrating that these effects indeed result from inhibition of the NOTCH signaling pathway (24 and unpublished data). The cell counts, cell cycle profiles, and percentages of apoptotic cells in the remaining 15 T-ALL cell lines were not affected by treatment with MRK-003 (Fig. 1, C and D, and Table I).
NICD levels are high in several T-ALL cell lines without truncating mutations in
Treatment with MRK-003 led to reduced levels of NICD in each of the T-ALL cell lines in which the intracellular protein could be detected, indicating that the compound effectively inhibited NOTCH1 cleavage at the cell surface (Fig. 1 E). However, only a subset of the 20 T-ALL cell lines had a GSI-sensitive cellular phenotype, indicating that a resistant phenotype can be observed with or without evidence for constitutive NOTCH signaling (i.e., without detectable expression of NICD; Table I and Fig. 1 E). Western blot analysis also demonstrated that all of the cell lines with C-terminal truncating mutations of NOTCH1 expressed high levels of NICD (Table I and Fig. 1 E), reflecting aberrant activation of the NOTCH pathway. Interestingly, there were seven cell lines without C-terminal truncating mutations in NOTCH1 that also expressed high levels of NICD (RPMI8402, CEM, BE13, DU528, HSB2, Jurkat, and PEER), implying that NOTCH is stabilized in these cell lines through a mechanism other than mutation of NOTCH1.
FBW7 is mutated in T-ALL cell lines and primary samples
Cellular levels of NICD are determined by the net effects of the rates of production and destruction of the protein. The SCF-FBW7 ubiquitin ligase plays a critical role in NICD degradation that is dependent on an intact PEST domain of NOTCH (25–30). Other known FBW7 substrates include cyclin E, MYC, SR-EBP, and c-Jun (31). FBW7 interacts with its substrates via the β propeller formed by its WD40 repeats after the substrate becomes phosphorylated within highly conserved Cdc phosphodegrons (CPDs) (32, 33). Human tumors harbor nonsense mutations that truncate FBW7 as well as missense mutations that target key arginine residues in the WD40 repeats. These mutations disable FBW7–substrate interactions and impair substrate degradation by FBW7 (34–37). Mutations affecting CPDs in tumor-derived alleles of both MYC and JUN have been described (38–41). Thus, mutations removing or altering the substrate CPD can also lead to loss of regulation by FBW7. Because the PEST domain truncations found in human T-ALL are predicted to disrupt FBW7 binding to NICD, we hypothesized that FBW7 mutation could represent an alternative mechanism for NOTCH deregulation in human cancers.
To test this hypothesis, we sequenced the entire coding region of FBW7 to determine whether mutations affecting this ubiquitin ligase gene might account for the accumulation of NICD in T-ALL cell lines that lacked truncations of the NOTCH1 PEST degradation domain. Sequencing of the entire coding region of FBW7 in 20 T-ALL cell lines revealed mutations in 6 of the 20 lines (30%) (Table I). The GSI-resistant cell line RPMI8402 harbored a homozygous arginine-to-histidine mutation at residue 465; all of the other mutations were heterozygous missense mutations of critical WD40 arginine residues (465 and 505). One additional cell line (BE13) has a homozygous deletion of FBW7 as determined by quantitative DNA PCR. Each of the seven cell lines with increased levels of NICD but without C-terminal truncating mutations in NOTCH1 (RPMI8402, CEM, BE13, DU528, HSB2, Jurkat, and PEER) harbored either a mutation in FBW7 or homozygous deletion of the gene, thus providing an explanation for persistent NICD expression in these lines (Fig. 1 E). Conversely, none of the cell lines with FBW7 mutations had PEST domain mutations in NOTCH1, suggesting that there is no selective pressure for FBW7 mutation in cells with stabilized NOTCH1.
We also sequenced exons encoding the substrate-binding domain (exons 7, 8, 9, 10, and 11) of FBW7 in 150 primary acute myeloid leukemia (AML) samples, 60 primary myelo dysplastic syndrome (MDS) samples, and 81 primary T-ALL samples. Although the AML or MDS samples lacked mutations in FBW7, 7 of 81 (8.6%) of the T-ALL samples harbored mutations in the substrate-binding domain of FBW7. All of the mutations in the patient samples were heterozygous missense mutations that altered arginine residues critical for substrate binding (Table II). As in the T-ALL cell lines, none of the primary samples with an FBW7 mutation had C-terminal truncating mutations in NOTCH1. These results indicate that mutation of FBW7 in primary T-ALL may represent an alternative mechanism of NOTCH deregulation that contributes to the pathogen esis this disease.
The NOTCH1 T2512 region comprises an FBW7 phosphodegron
As outlined above, loss of FBW7-mediated regulation of its substrates can result from mutation of FBW7 itself or from mutation of the substrate. The PEST domain of NICD is known to regulate protein stability and is a hot spot for mutations in primary human and mouse T-ALLs (7, 11). Furthermore, FBW7 binding to NOTCH is dependent on an intact PEST domain (27), strongly suggesting that the NOTCH CPD lies within this region. Alignment of the human NOTCH1 PEST domain with other FBW7 substrates demonstrated a CPD consensus motif anchored around threonine 2512 (T2512) in human NOTCH1. This residue is highly conserved across species as well as across members of the NOTCH family (Fig. 2 A). Notably, T-ALL–associated NOTCH1 PEST mutations identified in both human and mouse tumors occur upstream, or very nearby, this putative CPD, suggesting that these mutations may disrupt its function (7, 11).
To determine if the T2512 region contains a functional NOTCH CPD, we examined its role in binding interactions between the NICD and FBW7. For these studies, we used an expression plasmid encoding the mouse NICD, where T2487 corresponds to human T2512 (human numbering is used to refer to this CPD throughout this article). We first mutated T2512 to alanine (T2512A) and cotransfected 293 cells (Fig. 2 B) or K562 erythroleukemia cells (Fig. 2 C) with FBW7 and NICD. Because the interaction of FBW7 with its substrates leads to ubiquitination and degradation by the proteasome, we uncoupled substrate binding from turnover so that we could observe stable FBW7–substrate interactions. To disrupt this pathway, we expressed two previously described SCF mutants: (a) dominant-negative cullin-1 (dnCul1), which interacts with F-box proteins, but not ubiquitin-conjugating enzymes (42), and (b) a truncated version of FBW7 (termed Fb-WD), which encodes the WD40 substrate binding domains but lacks the N-terminal domains, including the F-box motif, and thus is unable to associate with cullin-1 and with the SCF (43). In both cases we found that the T2512A NICD mutant showed markedly decreased binding to FBW7 when compared with WT-NICD (Fig. 2, B and C). In contrast, mutation of another possible NOTCH1 CPD consensus motif anchored around threonine 2133 (38) did not affect FBW7 binding (unpublished data).
Because two putative CDK8 sites in the T2512 region have been previously implicated in NOTCH degradation by FBW7 (S2514 and S2517), we also tested the role of these residues in FBW7 binding (26). We found that FBW7 binding was disrupted by the S2514A mutation and that binding was restored when S2514 was replaced by a phosphomimetic residue (S2514E) (Fig. 2 D). Although these data are consistent with a role for S2514 phosphorylation in FBW7 binding, a negative charge in the +2 position (relative to the central threonine) is not a characteristic CPD feature. We thus mutated S2514 to proline, which conforms to the CPD consensus, and found that S2514P also restored FBW7 binding. Thus, although a negative charge is tolerated in the +2 position, it is not required for FBW7 binding. We also examined the binding of the NICD to each of the three FBW7 isoforms, which reside in distinct subcellular compartments (44). The results show that NICD binds to both FBW7α (nucleoplasmic) and FBW7γ (nucleolar), but not to FBW7β (cytoplasmic), supporting the idea that the NICD interacts with FBW7 in the nucleus (Fig. 2 E).
We next examined the role of T2512 in directing FBW7-mediated endogenous NICD ubiquitination and, in accord with our binding studies, found that the T2512A mutation markedly reduces NICD ubiquitination (Fig. 2 F). The extent to which this mutation inhibited ubiquitination was similar to that seen with a C-terminal PEST deletion, termed deltaCT, which is defective in ubiquitination by FBW7 (27). The small amount of residual ubiquitination could reflect either basal, T2512A-independent interactions of the NICD with FBW7, or could be due to alternative NICD ubiquitination pathways. Finally, we used metabolic labeling and pulse chase to compare the half-life of WT-NICD to the T2512A mutant in 293 cells. As predicted, the T2512A mutant shows a prolonged half-life, reflecting impaired degradation in vivo (Fig. 2 G). Collectively, these data indicate that the T2512 region comprises a functional CPD that directs binding interactions between NOTCH and FBW7.
GSI treatment results in decreased MYC expression in a subset of T-ALL cell lines
Recent studies have demonstrated that MYC is a direct target of NOTCH in both human and mouse T-ALL (20–22). To determine if MYC protein levels are affected by treatment with MRK-003 in our panel of T-ALL cell lines, we performed Western blot analysis of lysates from a panel of 19 T-ALL cells that were either treated with vehicle (DMSO) or 1 μM MRK-003 for 3 d (Fig. 3). A reduction of MYC protein levels upon GSI treatment was apparent in 8 of the 19 lines (Koptk1, HPB-ALL, DND41, RPMI8402, Molt13, PF382, Supt7, and TALL1), including all of the GSI-sensitive lines tested (Koptk1, HPB-ALL, DND41, and TALL1). 4 of the other 10 lines (Loucy, Supt11, Supt13, and Molt16) lack activated NOTCH signaling (i.e., NICD was not detectable by Western blot analysis; Fig. 1 E and Table I), and thus we would not expect GSI treatment to affect the expression of NOTCH target genes in these lines. The remaining seven cell lines (Jurkat, Molt4, PEER, CEM, DU528, BE13, and HSB2) have activated NOTCH signaling but do not display decreased MYC expression upon GSI treatment. Of note, all of the cell lines in which MYC expression is not decreased upon GSI treatment are resistant to the drug, reinforcing the importance of this NOTCH target gene for T-ALL cell growth. Because over expression of MYC has been shown to rescue the growth-suppressive effects of GSIs (20, 21), lack of MYC down-regulation in these T-ALL cell lines may contribute to their GSI resistance.
Because MYC is also a FBW7 target, we examined MYC half-life in a subset of five T-ALL cell lines (Molt4, DND41, KOPTK1, CEM, and Jurkat) by pulse-chase analysis. Although MYC half-life was prolonged in each cell line compared with a B cell lymphoblastoid cell line (JY) (Fig. 4), it was not increased in cell lines harboring mutations in FBW7 (CEM and Jurkat) compared with those with WT FBW7. This result was not surprising because several lymphoblastic leukemia cell lines and patient samples have been shown to have aberrantly stabilized MYC (45). Thus, MYC may be stabilized in T-ALL cell lines without FBW7 mutations due to mutations in other components of the MYC degradation pathway. Furthermore, in some contexts, FBW7 loss may lead to MYC stabilization in subcellular compartments (e.g., the nucleolus) without grossly altering bulk MYC turnover (44).
We also examined if MYC mRNA levels responded to GSI treatment by performing expression arrays on 18 of the T-ALL cell lines treated with DMSO or MRK-003 (1 μM) for 3 d. We observed either a very modest effect or no effect of GSI treatment on the levels of MYC RNA in four of the resistant lines that express NICD and have mutated FBW7 (CEM, BE13, DU528, and HSB2) (Fig. 5). Hence, the lack of down-regulation of MYC by GSI treatment in these lines is due to mechanisms acting upstream of MYC transcription rather than at the protein level. In the microarray expression analysis, other NOTCH target genes such as DELTEX1 and SHQ1 (22) (Fig. 5 and not depicted) were also not affected by GSI treatment in the CEM, DU528, BE13, and HSB2 cell lines. In the PEER cell line, which also has a mutation in FBW7, we detected residual expression of NICD after 3 d of treatment with MRK-003, and quantitative RT-PCR showed that MYC and DELTEX1 RNA expression levels did not decrease in this line upon GSI treatment (Fig. 1 E and not depicted). This suggests that the FBW7 mutations (or homozygous deletion) found in these five lines result in residual signaling through NICD, even in the face of GSI treatment, thus sustaining MYC transcription and promoting continued cell proliferation.
FBW7 mutants are functionally impaired and may act as dominant-negative mutants to prevent MYC degradation
Because FBW7 is haploinsufficient for tumor suppression (46), the single allele FBW7 mutations most commonly found in T-ALL could lead to stabilization of its substrates simply by reducing FBW7 activity by 50%. However, if the selective pres sure for these mutations consisted simply of loss-of-function of one allele, we would predict that we would find nonsense mutations in the FBW7 gene (36, 47). Our finding of mutational hot spots affecting the three key arginine residues of FBW7 argues against this possibility. Instead, we propose that the restricted mutational spectrum indicates a strong selection for expression of an FBW7 protein that is defective for productive substrate interactions and/or turnover, which suggests activity as a dominant-negative allele.
To explore this possibility, we analyzed each of the three mutant FBW7 proteins identified in T-ALL (Tables I and II) for their ability to bind to both MYC and NOTCH. As expected, the mutants were defective in NOTCH binding, although FBW7R479Q showed a modest amount of residual binding (Fig. 6 A, right). In contrast, each of the mutant FBW7 proteins bound to MYC, consistent with our recent finding that in addition to its canonical interaction with the MYC threonine 58 (T58) CPD, full-length FBW7 also binds to MYC through a second interaction that is independent of the T58 CPD (Fig. 6 A, left, and unpublished data). Importantly, although these FBW7 mutants bound to MYC, they were unable to promote MYC degradation (Fig. 6 B). We also examined whether a T-ALL–derived FBW7 mutant can dominantly interfere with FBW7-driven MYC degradation by expressing either increasing amounts of FBW7-γ or a mixture of a fixed amount of FBW7-γ in combination with increasing amounts of FBW7-γ-R465C. This analysis suggests that the mutant impairs FBW7-driven MYC degradation in a dose-dependent manner (Fig. 6 C).
At least two mechanisms may contribute to the dominant-negative activities of the mutant FBW7 proteins. The first is suggested by our finding that the FBW7 mutants still interact with Myc through a binding interaction independent of the MYC CPD that cannot promote MYC degradation. Hence, in the case of MYC, they may simply compete with WT FBW7 for MYC binding. However, a second mechanism of dominant-negative function is suggested by recent findings that FBW7 (and its orthologs) form dimers (48–51), thus raising the possibility that WT and mutant FBW7 proteins form heterodimeric dominant-negative complexes that are unable to degrade MYC. However, because mutation of the FBW7 dimerization region also disrupts its noncanonical binding interaction with MYC, we cannot currently distinguish between these two possibilities (unpublished data). We thus propose that FBW7 mutations may contribute to T-ALL pathogenesis and GSI resistance by leading to the stabilization of MYC as well as NICD.
The early successes of molecularly targeted therapy, such as imatinib mesylate directed against the BCR-ABL fusion protein in chronic myelogenous leukemia, stimulated enormous interest in subverting abnormal signaling pathways in cancer cells. The NOTCH signaling cascade influences normal development by regulating differentiation, proliferation, and apoptosis (52). Activation of the NOTCH signaling pathway is firmly established in T-ALL and is likely involved in the genesis of many other tumor types (53). Thus, it is not surprising that efforts to block NOTCH signaling as a novel therapeutic strategy are under way in T-ALL as well as in solid tumors. One of the most promising approaches has been to inhibit NOTCH receptor signaling using GSIs. This strategy suppresses the generation of NICD and thus, in principle, should inhibit the downstream transcriptional events normally induced by this key signaling component after it traverses to the nucleus. Although effective against some T-ALL cell lines, GSIs do not uniformly eliminate leukemic cells with activated NOTCH signaling. Understanding the mechanisms of GSI resistance may lead to better treatments for T-ALL.
We found missense FBW7 mutations or homozygous FBW7 deletion in GSI-resistant T-ALL cell lines and in primary T-ALL samples. Moreover, we have defined the NOTCH phosphodegron and demonstrated that the mutant forms of FBW7 found in T-ALL cannot bind to the NICD. Each T-ALL cell line with constitutive NICD expression harbored either NOTCH1 PEST domain or FBW7 mutations, suggesting that these two classes of mutations provide a mutually exclusive means of prolonging the NICD half-life. We also show that the expression of NOTCH target genes including DELTEX1 and MYC are not affected by GSI treatment in five resistant T-ALL cell lines with mutations in FBW7 (CEM, BE13, PEER, DU528, and HSB2), demonstrating that the mechanism of resistance in these leukemias lies upstream of MYC and DELTEX1 transcription. It appears that in these T-ALL lines, one consequence of FBW7 mutation is stabilization of the NICD resulting in sustained NOTCH signaling, and thus promoting resistance to GSIs. However, other FBW7 substrates, such as MYC, may also account for selection for FBW7 mutations and for GSI resistance in T-ALL, as suggested by our finding that ALL-associated FBW7 mutations can dominantly inhibit MYC degradation.
An important question is why FBW7 mutation confers GSI resistance, whereas NOTCH PEST domain truncations that remove the FBW7 interaction domain do not. In each cell line with PEST mutations, only one allele is affected by these heterozygous mutations, leaving the remaining normal allele, which encodes an NICD that is stabilized in the presence of mutant FBW7. Thus, although the amount of increased NOTCH activity resulting from the single allele PEST mutations may be sufficient to underlie the primary selection for these mutations, disruption of FBW7 function may be more active in sustaining NOTCH signaling and also may prolong the half-life of MYC (as well as other substrates), thus promoting GSI drug resistance.
The association between FBW7 mutations and resistance to GSIs has implications for clinical testing of these agents in patients whose cancers show deregulation of the NOTCH pathway. Molecular analysis of the FBW7 gene, as well as genes encoding the relevant NOTCH receptors and other key components of the NOTCH signaling pathway, such as NUMB, may contribute to identification of patients likely to be most responsive to GSI therapy. Because Myc has recently been shown to be an important target of Notch in mammary tumorigenesis as well as in T-ALL (5), it might also be possible to overcome GSI resistance by combining GSIs with other drugs that block the MYC pathway to synergistically reduce MYC levels and block tumor cell growth. One attractive candidate is TMPyP4, a cationic porphyrin that binds to and stabilizes guanine quadruplexes in DNA. MYC contains a sequence in its promoter that forms a guanine quadruplex, and TMPyP4 has been shown to inhibit MYC transcription and the growth of tumor cells in vivo (54). Thus, TMPyP4 or other agents that inhibit MYC transcription may be useful in combination with GSIs to overcome resistance in patients harboring FBW7 mutations. Pharmacologic or genetic strategies that restore the normal function of FBW7 in tumor cells could also be therapeutically useful.
Materials And Methods
Bone marrow or peripheral blood samples were collected at diagnosis from T-ALL patients at the Sophia Children's Hospital/Erasmus MC and the Dutch Childhood Oncology Group, as well as from AML, MDS, and T-ALL patients at Dana-Farber Cancer Institute. Informed consent and Institutional Review Board approval was obtained to use leftover patient material for research purposes.
T-ALL cell lines were obtained from American Type Culture Collection or Deutsche Sammlung von Mikroorganismen und Zellkulturen GmbH. 293a cells were provided by S. Reed (The Scripps Research Institute, La Jolla, CA).
Treatment of T-ALL cell lines with MRK-003.
T-ALL cell lines were plated at a density of 5 × 104–1.5 × 105 cells per ml (depending on doubling time) and were treated with either DMSO or 1 μM MRK-003 for 0 h, 6 h, 24 h, 3 d, or 7 d. For each time point, triplicate flasks were analyzed by the trypan blue exclusion assay to determine cell viability. For propidium iodide staining to analyze the cell cycle profile, cells were washed with PBS, stained in a PBS buffer containing 50 μg/ml propidium iodide and 200 μg/ml RNase A, and analyzed on a Becton Dickinson FACSCalibur Instrument. For annexin V staining to determine the percentage of cells undergoing apoptosis, the annexin V–FITC (or annexin V–APC) Apoptosis detection kit I (BD Biosciences) was used according to the manufacturer's instructions.
Microarray gene expression analysis.
Total RNA isolated from cultured cells was used to make fluorescently labeled cRNA that was hybridized to DNA oligonucleotide microarrays as described previously (55, 56). Human microarrays contained oligonucleotide probes corresponding to ∼21,000 genes. All oligonucleotide probes on the microarrays were synthesized in situ with inkjet technology (Agilent Technologies) (55). After hybridization, arrays were scanned and fluorescence intensities for each probe were recorded. Ratios of transcript abundance (experimental to control) were obtained after normalization and correction of the array intensity data. Gene expression data were analyzed with the Rosetta Resolver gene expression analysis software (version 6.0; Rosetta Biosoftware). Microarray data have been deposited in the GEO database under accession number GSE8416.
For analysis of endogenous protein expression, T-ALL cell lines were treated with MRK-003 for 3 d and lysed in RIPA buffer. Blots were probed with 9E10 anti-MYC antibody (1:500; Santa Cruz Biotechnology, Inc.), the cleaved NOTCH1 (Val1744) antibody (1:1,000; Cell Signaling Technology) as described previously (24), and anti–β-actin (1:7,000; Abcam) for the loading control. Signal was detected with SuperSignal West Femto ECL reagent (Pierce Chemical Co.). For transfections, MYC-tagged proteins were detected with 9E10 hybridoma supernatant (1:4), HA-tagged proteins were detected with HA-11 (1:1,000), and FLAG-tagged proteins were detected with M2 anti-FLAG (1:4,000; Sigma-Aldrich). These same antibodies were conjugated to protein G sepharose and used for immunoprecipitations as described previously (57).
The entire coding region of FBW7 was sequenced in genomic DNA isolated from the 20 T-ALL cell lines. Exons 7, 8, 9, 10, and 11 of FBW7 were sequenced in primary patient samples. All sequencing was performed at Agencourt Bioscience Corporation. Residues are numbered according to their location in the α isoform of FBW7 (GenBank RefSeq NM_033632).
Plasmids and transient transfections.
The following expression plasmids have been described (43, 44): pFLAG-FBW7-α, -β, and -γ; pFLAG-FbWD; pFLAG-FBW7-γ-R465H (also termed R298H); and HA-MYC, HA-dncul1, and HA-ubiquitin. All mutagenesis were performed by the QuickChange method (Stratagene) and confirmed by sequencing. Point mutations in the FBW7 common region R465H, R479Q, and R505C were generated by site-directed mutagenesis using pFLAG-FBW7 vectors as template. Although numbering is different between isoforms, the α numbering is used for simplicity. CS2-mNICD-MT was provided by R. Kopan (Washington University, St. Louis, MO). It encodes the mouse Notch1 intracellular domain from Val1744 to the end of the WT protein and includes six in-frame MYC epitope tags at the C terminus. Mutagenesis of this construct was used to generate the T2512A, S2514A, S2514E, and S2514P mutants. The deltaCT mutation, generated by PCR, deletes the PEST domain but maintains the C-terminal MYC tags. 293A cells were transfected by the calcium phosphate precipitation method as described previously (57). Cells were harvested at 36–48 h after transfection using TENT buffer (50 MM Tris-Cl, pH 8.0, 2 mM EDTA, 150 mM NaCl, 1% Triton X-100) (30) with protease and phosphatase inhibitors and analyzed by immunoprecipitation and Western blotting as described previously (57).
Metabolic labeling and pulse chase.
NICD pulse-chase assays were performed as described previously (43). In brief, 293a cells were transfected with plasmids encoding MYC-tagged NICD or NICD T2512A. At 36 h after transfection, cells were preincubated with media lacking cysteine and methionine, pulsed with media containing Trans 35S label, and chased with media with cold methionine. Samples were obtained at the indicated time points and immunoprecipitated with anti-Notch serum MN1, a mouse monoclonal antibody recognizing the NICD (provided by I. Bernstein, Fred Hutchinson Cancer Research Center, Seattle, WA). MYC pulse-chase experiments were performed as described previously (45).
Cell-based ubiquitination assays were performed as described previously (58). In brief, transiently transfected 293a cells were harvested at 48 h after transfection. Cells were trypsinized, rinsed once with PBS, and lysed in 100 μl 2% SDS in TBS (10 mM Tris-HCl, pH 8.0, 150 mM NaCl) with boiling for 10 min. These samples were diluted at 1:10 with 1% triton in TBS containing protease inhibitors, 5 mM N-ethymaleimide, and 5 mM MG132, sonicated, precleared with protein G sepharose, and centrifuged to pellet-insoluble debris. Aliquots were removed for Western blot analysis, and the remaining lysates were immunoprecipitated with anti-HA antisera prebound to protein G sepharose. Samples were analyzed by SDS-PAGE and probed with 9E10 to visualize MYC-tagged ubiquitinated Notch proteins.
The authors wish to thank Theresa Zhang, Leslie Carlini, and Xudong Dai for helpful discussions regarding mRNA profiling; Hellen Kim for discussions about Western blotting methods; Keith McKenna and Chitra Raghunathan for technical assistance; and John Gilbert for editorial review.
This work was funded by a sponsored research agreement from Merck Research Laboratories (to A.T. Look), an American Society of Hematology Fellow Scholar Award and by the National Institutes of Health (1K08CA109124-01A2; to J. Grim), National Institutes of Health grants (R01CA84069 and R01CA102742) and the Burroughs Welcome Foundation (to B.E. Clurman), as well as a Harvard Medical School Hematology Training Grant (to J. O'Neil).
The authors have no conflicting financial interests.
Abbreviations used: AML, acute myeloid leukemia; CPD, Cdc phosphodegron; GSI, γ-secretase inhibitor; MDS, myelodysplastic syndrome; NICD, NOTCH intracellular domain; T-ALL, T cell acute lymphoblastic leukemia.
J. O'Neil and J. Grim, and B.E. Clurman and A.T. Look contributed equally to this work.