Notch3 Functions as a Tumor Suppressor by Controlling Cellular Senescence

Notch proteins are a family of transmembrane receptors that play fundamental roles in many biologic processes, including stem cell maintenance, lineage commitment, and cell fate determination (1). Notch signaling is initiated through ligand–receptor interaction between neighboring cells, resulting in 2 successive proteolytic cleavages that release the intracellular domain of Notch (NICD) from membrane. NICD subsequently translocates into nucleus where it associates with transcription factor CBF1/RBP-Jκ in mammals. CBF1 binds to promoters of Notch target genes in a sequence-specific manner and normally acts as a transcriptional repressor in the absence of active Notch signaling. NICD converts CBF1 into a transcriptional activator to transactivate Notch target genes by recruiting Mastermind-like (MAML) family of transcriptional activators and their coactivators (1). Notch receptors share conserved structural features, and the core Notch signaling is evolutionarily and mechanistically conserved. However, it is becoming increasingly evident that the 4 mammalian Notch receptors (Notch1–4) have distinct nonoverlapping functions, and the biologic outcome of Notch signaling is determined by cellular context (1, 2).

Notch signaling is often dysregulated during tumorigenesis (1–3). An oncogenic function of Notch signaling has been documented in several types of cancer including leukemia/lymphoma (4–7), breast cancer (8, 9), melanoma (10, 11), and brain tumor (12, 13). It is thought that Notch signaling exerts its oncogenic function by promoting proliferation, blocking differentiation, or inhibiting apoptosis (2). Interestingly, Notch signaling has also been implicated to have a tumor suppressor function in myeloid leukemia (14) and solid tumors of skin (15, 16), lung (17), liver (18–20), prostate (21), and head and neck (22, 23). The molecular mechanism underlying the tumor suppressor function of Notch signaling is not clear (3). In this report, we identified Notch3 as an important regulator of senescence. As senescence restricts proliferation of cells at risk of malignant transformation (24), our study offers a plausible mechanism underlying Notch3-mediated tumor suppression.

Human cell lines (BJ, WS1, MDA-MB-231, MDA-MB-435, MCF7, SK-BR-3, A375P, A2058, HT144, T98G, U-87 MG, J82, T24, HeLa, and 293T) were purchased from American Type Culture Collection. Telomerase-immortalized BJ-hTERT and WS1-hTERT were described previously (25). Human fibroblast LF1, telomerase-immortalized human fibroblasts LF1-hTERT, and p21^−/−-hTERT (26) were kindly provided by Dr. John Sedivy (Brown University, Providence, RI). These cells were cultured in Dulbecco's Modified Eagle Medium (Invitrogen) supplemented with 10% FBS (Hyclone). Two postselection human mammary epithelial cell (HMEC) lines, 48R and 184 (both from normal female breasts), were gifts from Dr. Martha Stampfer (Lawrence Berkeley National Laboratory, Berkeley, CA) and were maintained in mammary epithelial cell growth medium (MEGM) medium (Biowhittaker) buffered with HEPES (Sigma) to pH7.4. All cells were cultured in a humidified chamber containing 5% CO₂ at 37°C

Early passage BJ and WS1 fibroblasts at subconfluence density were irradiated (10 Gy) using a Cs¹³⁷ radiator and analyzed a week after irradiation. Similarly, these cells were treated with 1 μmol/L doxorubicin (Sigma) for 2 hours or with 2.5 μmol/L etoposide (Sigma) for 17 hours, and analyzed 24 hours after treatment. Early passage confluent fibroblasts were treated with 250 μmol/L of H₂O₂ for 2 hours and plated at subconfluent density 48 hours after treatment. After these treatments, cells entered senescence, as indicated by proliferative arrest and positive staining for SA-β-gal activity.

Fragments of HA-tagged NICD3 and its deletions were PCR amplified using a construct containing human Notch3 cDNA (kindly provided by Dr. Anne Joutel, INSERM, France) as template and inserted into pLenti-CMV-Hyg (a gift from Dr. Paul Kowalski, University of Toronto, Toronto, Canada). A NICD1 fragment was released from MSCV-NICD1-IRES-GFP provided by Dr. Nadia Carlesso (Harvard Medical School, Boston, MA) and inserted into pLenti-CMV-Hyg. A fragment of DnMAML1-GFP fusion was excised from MSCV-DnMAML1-GFP provided by Dr. Michele Kelliher (University of Massachusetts Medical School, Worcester, MA) and inserted into pLenti-CMV-IRES-BSD. A p53 expression plasmid pC53-SN3 was provided by Dr. Bert Vogelstein (The Johns Hopkins University School of Medicine, Baltimore, MD). A fragment containing human p21 promoter was released from pWWP-Luc (Addgene) and subcloned into pGL3-Basic (Promega) to create pGL3-p21-Luc. Lentiviral constructs expressing short-hairpin RNA (shRNA)-targeting Notch3 (TRCN0000020238 and TRCN0000020234) were purchased from Open Biosystems, and a scramble shRNA (provided by Dr. Stanley N. Cohen, Stanford University School of Medicine, Stanford, CA) was used as a control. All constructs were verified by sequencing.

Lentiviral packaging and infection were carried out as described previously (27). Briefly, lentiviral vectors were cotransfected into 293T cells with a plasmid (pMD2.VSV-G) encoding vesicular stomatitis virus glycoprotein (VSV-G) and a plasmid (pCMVdR8.74) encoding packaging proteins. VSV-G pseudotyped virus were collected 48 hours after transfection and used to infect target cells in the presence of 8 μg/mL polybrene (Sigma). Two days later, infected cells were selected with 1 μg/mL puromycin (Sigma), 150 μg/mL hygromycin B (Calbiochem), or 10 μg/mL blasticidin (Invitrogen).

Staining of cells with crystal violet and for SA-β-gal activity were carried out as previously described (25). To analyze cell proliferation, cells were seeded at 2 × 10⁴ per well in 6-well plates, harvested in triplicate, and counted every other day for a week using a Z1 Coulter Particle Counter (Beckman Coulter). To determine replicative life span, the cells were plated at 1 to 3 × 10⁵ cells per 10-cm dish and subcultured before they reached high cell density. Population doubling (PD) per passage was calculated as log₂ (number of cells at time of subculture/number of cells plated). Cumulative PD was plotted against total time in culture to assess replicative life span.

Total RNA was isolated using RNeasy Mini Kit (Qiagen), and reverse-transcribed using Superscript II (Invitrogen) according to manufacturer's instructions. Real-time PCR was carried out using SYBR Green PCR Kit (Bio-Rad). The following primers were used: p16 (5′-GCCCAACGCACCGAATAGT-3′ and 5′-CGATGGCCCAGCTCCTCAG-3′), p21 (5′-CCCAAGCTCTACCTTCCCAC-3′ and 5′-ACAGGTCCACATGGTCTTCC-3′), Notch3 (5′-TCTCAGACTGGTCCGAATCCAC-3′ and 5′-CCAAGATCTAAGAACTGACGAGCG-3′), and RNA polymerase II (RPII, 5′-GCACCACGTCCAATGACAT-3′ and 5′-GTGCGGCTGCTTCCATAA-3′). Relative transcript levels of p16, p21, or Notch3 in control cells were set to be 1 after normalization with RPII.

Western blot analysis was carried out using total cell lysates as described previously (28). Rabbit polyclonal Notch3 antibody was raised against a synthetic peptide (NPGTPVSPQERPPPYLA) corresponding to C-terminus of human Notch3 (Covance). Other primary antibodies used in this study were p21 (SX118), p53 (DO-1), Notch1 (mN1A), Hes1 (H-140), HA (Y-11), β-actin (C4), GAPDH (Santa Cruz Biotechnology), GFP (Roche), and α-tubulin (DM1A, Sigma).

Cells were treated with 1% formaldehyde (Sigma) and cross-linking was carried out at room temperature for 10 minutes. After neutralization with glycine (125 nmol/L), cells were lysed in SDS lysis buffer. Chromatin was sonicated to fragments of approximately 500 bp, and chromatin immunoprecipitation (ChIP) with anti-Notch3 antibody or matched IgG as a control was carried out using a ChromaFlash one step ChIP kit (Epigentek) according to manufacturer's instruction. After reverse of cross-linking at 65°C for 3 hours, amplification of the specific region in p21 promoter containing the CBF1 binding site was detected in quantitative PCR with the following primers: F: 5′-TCTGGCCTCAAGATGCTTTG-3′ and R: 5′-CACTCTGGCAGGCAAGGATT-3′. Chromatin before ChIP was used as input for comparison.

HeLa cells were plated at 5 × 10⁴ cells per well in 24-well plate and transiently transfected with 200 ng of indicated NICD or p53 expression plasmid, 100 ng of p21 promoter firefly luciferase reporter construct, and 17 ng of Renilla luciferase plasmid (Promega) using Lipofectamine 2000 (Invitrogen). Luciferase activities were determined 48 hours after transfection using a dual-luciferase reporter assay system (Promega). Results were represented as the ratio of firefly to Renilla luciferase activity and normalized to vector control.

Notch3 expression (log₂ transformed) in microarray datasets of breast cancer (GSE3165) and melanoma (GSE3189 and GSE7553) was retrieved from Gene Expression Omnibus (GEO; http://www.ncbi.nlm.nih.gov/geo), and analyzed with dot plot. One-way ANOVA was used for statistical analysis.

In other experiments, data were presented as mean ± SD. Two-tailed and unpaired Student t test, Spearman correlation, and two-way ANOVA were used for statistical analyses, with P < 0.05 considered as statistically significant.

We have previously identified a gene expression signature that is characteristic of replicative senescence in human fibroblasts (29). Notch3, a member of Notch family receptors, is among this list of genes showing elevated expression in senescent fibroblasts in our microarray analysis. Quantitative RT-PCR confirmed that Notch3 transcripts were increased in senescent fibroblasts (Fig. 1A). Furthermore, Notch3 protein, in particular the intracellular domain of Notch3 (NICD3, bottom band in Fig. 1A), was increased in senescent fibroblasts after serial passage in culture compared with early passage proliferating cells (Fig. 1A). HMECs, similar to fibroblasts (30), entered senescence in culture after experiencing progressive telomere shortening (31). We found that Notch3 expression was increased in senescent HMECs (Fig. 1B), indicating that Notch3 elevation is a common change during replicative senescence activated by telomere shortening. The closely related Notch1 did not show a similar increase in expression (Fig. 1), suggesting a specific change in Notch3 during senescence.

In addition to replicative senescence activated by telomere shortening, senescence can be induced by various stress stimuli, such as oxidative stress or DNA damage (32–34). To investigate whether elevated Notch3 expression is unique to replicative senescence or a general response during senescence, we examined Notch3 expression in cells triggered to enter senescence by different senescence-inducing stimuli. Treatment of hydrogen peroxide or DNA-damaging agents (ionizing radiation, doxorubicin, or etoposide) induced senescence in early passage fibroblasts (25, 32–34). The expression of Notch3 was increased in senescent cells induced by oxidative stress (Fig. 1C) or DNA damage (Fig. 1D), suggesting that elevation of Notch3 is a general response during senescence. Interestingly, we did not observe an increase in the cleaved NICD3 in senescent cells induced by oxidative stress or DNA damage compared with cells entering replicative senescence. The difference may reflect the duration of cells being passaged in culture. In support of this notion, the cleaved NICD3 was also found in telomerase-immortalized fibroblasts (Supplementary Fig. S1) that have been passaged extensively in culture.

Notch1 has been shown to transactivate p21 in mouse keratinocytes (16). Because there is a concomitant increase of Notch3 and p21 in senescent cells (Fig. 1), we investigated whether elevated Notch3 expression is responsible for upregulation of p21 in senescent cells. The senescent cells were infected with lentivirus to stably express a shRNA targeting Notch3. Downregulation of Notch3 by shRNA resulted in decreased p21 expression (Fig. 2A), suggesting that Notch3 is required for elevated p21 expression in senescent cells. Interestingly, the protein level of p53, the known regulator of p21 expression, remained largely unchanged upon Notch3 knockdown (Fig. 2A). Furthermore, downregulation of Notch3 in mid-passage cells resulted in a delayed onset of replicative senescence and an extended replicative lifespan (Fig. 2B and Supplementary Fig. S2), suggesting that Notch3 plays an important role in senescence activation.

Conversely, we ectopically expressed NICD3, an active form of Notch3, in early passage fibroblasts to investigate whether Notch3 is sufficient to activate p21 expression and induce senescence. As expected, expression of Hes1, a well-characterized target of Notch signaling, was increased following Notch3 upregulation (Fig. 2C). Ectopic expression of NICD3 resulted in an increase of p21 at the transcript and protein levels compared with vector controls (Fig. 2C). In contrast, NICD3 only increased the expression of p16, another important regulator of senescence, in WS1 but not BJ cells (Supplementary Fig. S3A). Furthermore, ectopic expression of NICD3 induced senescence in early passage cells, indicated by cessation of proliferation, morphologic changes, and positive staining of SA-β-gal activity (Fig. 2D and E and Supplementary Fig. S3). Collectively, these results indicate that Notch3 is an important regulator of senescence and p21 expression.

In nucleus, NICD associates with transcription factor CBF1, which binds to promoters of Notch target genes through consensus sequence of 5′-GTGGGAA-3′ (35). There is a conserved CBF1 binding site in the p21 promoter in human and mouse (Supplementary Fig. S4A). Using ChIP, we found that there was no significant binding of Notch3 on p21 promoter in early passage human fibroblasts, whereas this binding was significantly increased in senescent cells (Fig. 2F), suggesting that Notch3 directly regulates p21 expression in senescent cells. Consistently, NICD3 transactivated human p21 promoter in a luciferase reporter assay (Supplementary Fig. S4B).

To gain further insight into the mechanism of Notch3-induced p21 expression and senescence, we investigated domain(s) in Notch3 that are responsible for its ability to activate senescence and p21 expression. NICD3 or its truncations were ectopically expressed in hTERT-immortalized fibroblasts (Fig. 3A). Similarly to what was observed in early passage cells, upregulation of Notch3 was sufficient to induce p21 expression and senescence in hTERT-immortalized cells (Fig. 3A and B and Supplementary Fig. S5A and S5B). The N-terminal domain (RAM-ANK) was sufficient to induce p21 expression and senescence, whereas deletion of RAM domain (ΔRAM) completely abolished NICD3′s ability to induce p21 expression and senescence (Fig. 3A and B and Supplementary Fig. S5A and S5B), indicating that the RAM domain of Notch3 is required to activate p21 expression and senescence.

NICD associates with CBF1 and recruits Mastermind-like (MAML) family of transcriptional coactivators to transactivate Notch target genes. The RAM domain of NICD is essential for the formation of the CBF1:NICD:MAML ternary complex (36). Our finding that the RAM domain is required for NICD3 to activate p21 expression and senescence led us to hypothesize that Notch3 induces p21 expression and senescence through the canonical Notch signaling. To test this hypothesis, a dominant negative MAML1 (DnMAML1) was used to suppress Notch signaling (37). As expected, Notch3-induced Hes1 expression was attenuated by DnMAML1 (Fig. 3C and Supplementary Fig. S5C). Importantly, the ability of NICD3 to induce p21 expression and senescence was inhibited by the expression of DnMAML1 (Fig. 3C–F and Supplementary Fig. S5C–S5F), indicating that Notch3-induced p21 upregulation and senescence is dependent on a functional MAML1. Collectively, these results suggest that the canonical Notch signaling is at least partially required for Notch3-induced senescence and p21 expression.

As p21 is a critical regulator of senescence (26), our observation that Notch3 induced p21 expression and senescence prompted us to investigate whether p21 is responsible for Notch3-induced senescence. We compared the ability of Notch3 to induce senescence in telomerase-immortalized human fibroblasts LF1 and its derivative p21^−/− cells, which were generated through 2 sequential rounds of targeted homologous recombination (26). Similar to BJ and WS1 fibroblasts, ectopic expression of NICD3 in LF1 cells induced senescence and p21 expression (Fig. 4). However, the ability of Notch3 to induce senescence was significantly attenuated in p21^−/− cells, as indicated by increased cell proliferation and decreased SA-β-gal staining compared with the parental LF1 cells (Fig. 4). These results suggest that Notch3-induced senescence is partially mediated by p21.

As senescence is an important tumor suppressor (24), the senescence regulatory function of Notch3 suggests that it may act as a tumor suppressor. Exome sequencing of human tumors has identified nonsense and missense mutations in Notch3 in several types of cancer, including breast cancer (38, 39), colorectal cancer (40), melanoma (41, 42), head and neck carcinoma (22), glioblastoma (43), liver (44), lung (39, 45), ovarian (46), and prostate cancer (39, 47). 5.3% (46 of 871, as of 10/7/2012) of tumor samples reported in the COSMIC database (http://www.sanger.ac.uk/genetics/CGP/cosmic) have nonsense and missense mutations in Notch3. Some of these mutations are in the RAM domain, which is required for Notch3 to induce senescence (Fig. 3). In particular, nonsense mutations [K1473* in head and neck carcinoma (22), G1612* in lung squamous cell carcinoma (39), and Q1699* in melanoma (41)] would result in Notch3 without the intracellular domain (K1473* and G1612*) or with truncated RAM domain (Q1699*) that is unable to induce senescence. The loss-of-function mutations in Notch3 found in tumors support its role in senescence regulation and tumor suppression.

As deletion or amplification of Notch3 is not noted in the analyzed tumors, we retrieved Notch3 expression in microarray datasets containing primary human breast cancer and melanoma samples from Gene Expression Omnibus (GEO: http://www.ncbi.nlm.nih.gov/geo). Compared with normal tissues, the expression of Notch3 was significantly decreased in primary breast cancer (Fig. 5A) and melanoma samples (Fig. 5B). In contrast, no significant change in Notch1, Notch2, or Notch4 expression was observed in breast cancer (Supplementary Fig. S6A) or melanoma samples except that Notch2 showed decreased expression in one of 2 melanoma datasets (Supplementary Fig. S6B, S6C). Furthermore, a significant correlation between Notch3 and p21 expression was observed in these tumor samples (Supplementary Fig. S7).

We noticed that Notch3 was undetectable in several human tumor cell lines (Fig. 5C). When NICD3 was ectopically expressed in 3 tumor (breast cancer, melanoma, and glioblastoma) cell lines without detectable Notch3 protein, p21 expression was effectively induced (Fig. 5D). It is interesting that Notch3 induced p21 expression in MDA-MB-231 and T98G cells, because these cells express only mutant p53 protein lacking the transactivation function (48, 49). This finding, together with the observation of unchanged p53 level upon Notch3 knockdown in normal fibroblasts (Fig. 2A), suggests that Notch3 induces p21 expression independently of p53. More importantly, restoration of Notch3 expression led to inhibition of cell proliferation and senescence in these tumor cells (Fig. 5E and Supplementary Fig. S8), suggesting that loss of Notch3 expression allows the escape of senescence and thus plays a critical role in tumor development.

In this report, we found that the expression of Notch3 was elevated in senescent cells activated by various senescence-inducing stimuli, including telomere shortening, DNA damage, and oxidative stress. Downregulation of Notch3 impaired the senescence response and attenuated p21 expression, whereas ectopic expression of Notch3 was sufficient to induce senescence and p21 expression in early passage, telomerase-immortalized or tumor cells, indicating that Notch3 plays a critical role in regulating senescence. As senescence is an important tumor suppressor (24), this novel function of Notch3 suggests that it acts as a tumor suppressor. Consistent with this notion, loss-of-function mutations in Notch3 that are unable to induce senescence are identified in several types of cancer. In addition, we found that the expression of Notch3 was significantly decreased in primary human breast cancer and melanoma samples compared with normal control tissues. Moreover, we found that ectopic expression of Notch3 in tumor cell lines that normally do not express Notch3 led to growth arrest and senescence. Collectively, these results reveal a novel function of Notch3 in senescence regulation and tumor suppression.

Different Notch receptors show distinct patterns of expression during senescence. We found that Notch3 showed upregulation during senescence in fibroblasts and mammary epithelial cells. In contrast, the closely related Notch1 showed decreased expression in senescent fibroblasts and unchanged expression in senescent HMECs (Fig. 1). Notch1 has also been shown to be downregulated in senescent keratinocytes (50) but increased in senescent endothelial cells (51). Notch1 induces apoptosis in fibroblasts (52), neural progenitor cells (53), and B cells (54, 55), whereas activates senescence in endothelial cells (51) when constitutively expressed. We tried multiple times to express Notch1 in early passage fibroblasts and were unable to establish stable cell lines expressing Notch1, consistent with the previous finding that Notch1 induces apoptosis in fibroblasts (52). It is becoming increasingly evident that the 4 mammalian Notch receptors have distinct nonoverlapping functions, even though the core Notch signaling initiated by them is thought to be evolutionarily and mechanistically conserved (1, 2). The outcome of Notch signaling is highly dependent on cellular context, and Notch signaling has been found to have both oncogenic and tumor suppressor functions in cancer. For instance, an oncogenic role for Notch1 in T-cell acute lymphoblastic leukemia is well established (6, 7), whereas Notch1 is found to be a tumor suppressor in skin cancer, myeloid leukemia, and head and neck cancer (14–16, 22, 23). It is thought that Notch signaling exerts its oncogenic function by promoting proliferation, blocking differentiation, or inhibiting apoptosis (2), whereas the molecular mechanism underlying the tumor suppressor function of Notch signaling is not entirely clear (3). The senescence regulation function of Notch3 identified in this study provides a molecular mechanism underlying the tumor suppressor function of Notch3. Different Notch receptors play different, even opposing, roles in tumor development, showing the complexity of Notch signaling in cancer. As targeting Notch signaling is suggested for cancer therapeutics, our study points to unwanted dire consequence if Notch signaling is indiscriminately targeted. Understanding the specific role of different Notch receptors in cancer development offers a better option in targeting specific Notch receptor for cancer therapeutics.

No potential conflicts of interest were disclosed.

Conception and design: H. Zhang

Development of methodology: H. Cui, Y. Kong, M. Xu, H. Zhang

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): H. Cui, Y. Kong, M. Xu, H. Zhang

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): H. Cui, Y. Kong, M. Xu, H. Zhang

Writing, review, and/or revision of the manuscript: H. Cui, H. Zhang

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): H. Zhang

Study supervision: H. Zhang

The authors thank Drs. Nadia Carlesso, Stanley N. Cohen, Anne Joutel, Michele Kelliher, Paul Kowalski, John Sedivy, Martha Stampfer, and Bert Vogelstein for kindly providing reagents.

This work was supported by grants from the National Cancer Institute (R01CA131210) and The Ellison Medical Foundation (AG-NS-0347-06) to H. Zhang.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.