Cell-Cell Signaling Through NOTCH Regulates Human Embryonic Stem Cell Proliferation



Unlike pluripotent mouse embryonic stem (ES) cells, human ES cells and their malignant equivalents, embryonal carcinoma (EC) cells, require close cell-cell contact for efficient growth. Signaling through the NOTCH receptor, initiated by interaction with ligands of the DELTA/JAGGED family expressed on neighboring cells, plays a role in regulating the self-renewal of several stem cell systems. Members of the NOTCH and DELTA/JAGGED families are expressed by human EC and ES cells, and we have therefore investigated the possible role of NOTCH in the maintenance of these cells. Cleavage of both NOTCH1 and NOTCH2 to yield the intracellular domain responsible for the canonical signaling pathway of NOTCH was detected in several human EC and ES cell lines, suggesting that NOTCH signaling is active. Furthermore, the proliferation of human EC cells, as well as the expression of several downstream NOTCH target genes, was markedly reduced after small interfering RNA knockdown of NOTCH1, NOTCH2, and the canonical effector CBF-1 or after blocking NOTCH signaling with the γ-secretase inhibitor L-685,458. The inhibitor also caused a reduction in the growth of human ES cells, although without evidence of differentiation. The results indicate that cell-cell signaling through the NOTCH system provides a critical cue for the proliferation of human EC and ES cell in vitro.

Disclosure of potential conflicts of interest is found at the end of this article.


Human embryonic stem (ES) cells differ from the corresponding mouse ES cells in many ways, despite their similar derivation from the inner cell mass of the blastocyst stage of early embryos and their expression of key transcription factors characteristic of pluripotent cells, such as OCT4, NANOG, and SOX2. One notable difference is the lack of responsiveness of human ES cells to leukemia inhibitory factor [1], which can maintain the proliferation of undifferentiated mouse ES cells in the absence of feeders [2, 3]. Another is indicated by reports that bone morphogenetic proteins (BMPs) promote differentiation of human ES cells but act to maintain the undifferentiated state of mouse ES cells [4]. Many differences between the two species' ES cells were presaged by studies of embryonal carcinoma (EC) cells, the malignant counterparts of ES cells and the stem cells of teratocarcinomas [5]. Thus, some human EC cells are induced to differentiate by BMP7 [6], whereas the expression of surface marker antigens such as stage-specific embryonic antigen 3 (SSEA3) and SSEA4 by human ES cells, and their lack of SSEA1, contrasting with the SSEA1+/SSEA3−/SSEA4− phenotype of mouse ES cells [7], was first described in studies of human EC cells [8, [9], [10]–11]. Yet another difference is the ability of human EC and ES cells to undergo trophectodermal differentiation [12, [13]–14], whereas this does not typically occur with mouse ES cells unless they are genetically modified to reduce levels of OCT4 expression [15].

One particular feature of human EC cells is their tendency to differentiate when cultured at low densities, a phenomenon not influenced by addition of medium conditioned by cells grown at high density [16]. This phenomenon is mirrored in the low cloning efficiency reported for human ES cells [17, 18]. These observations suggest that the survival and proliferation of human EC and ES cells may be promoted by direct contact between the cells. A candidate for such a signaling system is that mediated by NOTCH. This pathway is an evolutionarily conserved cell-cell signaling system that regulates the survival, proliferation, and differentiation of a range of cell types throughout the development and life span of all metazoans [19]. Intercellular signaling mediated by NOTCH also plays a key role in maintaining various mammalian stem cell systems, notably hematopoietic, neural, skin, intestinal, and skeletal stem cells. In these systems, NOTCH has been shown to regulate proliferation, differentiation, and survival in a context-dependent manner [20].

Canonical NOTCH signaling is mediated by a family of cell surface transmembrane receptors. Upon interaction with cell surface ligands of the DELTA/JAGGED family, expressed on neighboring cells, the intracellular domain of the NOTCH receptor is cleaved and translocates to the nucleus, where it associates with the transcription factor CBF-1 (RBPJK, RBPSUH) [21]. Recently, we have observed the expression of NOTCH1 and NOTCH2 and their ligands in human EC and ES cells, and we have suggested the possibility that NOTCH2 in particular might play a role in the etiology of human teratocarcinomas [22, 23]. We now report functional evidence indicating that NOTCH signaling does play a crucial role in the maintenance of undifferentiated human EC and ES cells.

Materials and Methods

Cell Culture and Treatment with γ-Secretase Inhibitor

Human ES cells H1 and H7 [24] were cultured in knockout Dulbecco's modified Eagle's medium (DMEM) (Invitrogen, Carlsbad, CA, http://www.invitrogen.com), supplemented with 20% Serum Replacer (Invitrogen) and 4 ng/ml basic fibroblast growth factor (Peprotech, Rocky Hill, NJ, http://www.peprotech.com), on mouse embryonic fibroblast feeder layers, mitotically inactivated with mitomycin C (Sigma-Aldrich, St. Louis, http://www.sigmaaldrich.com) as described previously [24]. Human EC cells NTERA2 cl.D1 (NTERA2) and 2102Ep cl. 2A6 (2102Ep) [9, 25] were cultured in DMEM containing 10% heat-inactivated fetal calf serum (FCS) (Invitrogen) as described previously. For treatment with L-685,485, human ES cells were harvested by treatment with collagenase IV followed by scraping with sterile 3-mm glass beads and seeded as small colonies composed of approximately 5–20 cells at a concentration of 1.3 × 104 colonies per well of a six-well plate; human EC cells were seeded as single cells obtained via treatment with 0.25% (wt/vol) trypsin in 2 mM EDTA at a concentration of 2 × 105 cells per well of a six-well plate. Human ES and human EC cells were treated 24 hours after plating and every day thereafter with 20 μM L-685,485 (Sigma-Aldrich) diluted in prewarmed medium from a 10 mM stock prepared in dimethyl sulfoxide (DMSO).

RNA Interference

Double-stranded small interfering RNAs (siRNAs) corresponding to β2-microglobulin (B2M), Ok(a), CBF-1, NOTCH1, and NOTCH2 were selected on the basis of closeness to an ideal G/C content of 50% and sequence specificity as assessed by public sequence interrogation using BLAST and were chemically synthesized by Xeragon Inc. (now Qiagen, Valencia, CA, http://www1.qiagen.com), with the following sense and antisense sequences:

  • B2M: sense, 5′-GAUUCAGGUUUACUCACGUdTdT-3′; antisense, ACGUGAGUAAACCUGAAUCdTdT starting from nucleotide 91 of B2M sequence (GenBank accession no. AB021288);

  • OK(a): sense, 5′-GAGCAGGTTCTTCGTGAGTTC-3′; antisense, 5′-GAACCACGAATAACTGCTC-3′ starting from nucleotide 911 of human BSG (Ok blood group) sequence (GenBank accession no. NM_2001728);

  • CBF-1: sense, 5′-UCGACUACGAUCCCAGACAdTdT-3′; antisense, UGUCUGGGAUCGUAGUCGAdTdT starting from nucleotide 564 of CBF-1 sequence (GenBank accession no. XM_517135);

  • NOTCH1: sense, 5′-AACAUCAACGAGUGCUCCAGCdTdT-3′; antisense, 5′-GCUGGAGCACUCCUUGAUGUU-3′ starting from nucleotide 1,807 of NOTCH1 sequence (GenBank accession no. NM_17617);

  • NOTCH2: sense, 5′-GCAACACGGUCGAGUGCCUdTdT-3′; antisense, 5′-AGGCACUCGACCGUGUUGCdTdT-3′ starting from nucleotide 4,707 of NOTCH2 sequence (GenBank accession no. NM_024408).

For treatment with siRNA, NTERA2 and 2102Ep cells were trypsinized to produce single cells and plated into six-well plates at a density of 2 × 105 cells per well. Treatment with siRNA was carried out 24 hours after plating. To prepare siRNA for transfection, 10 μl of a 20-μM solution of siRNA was incubated with 4 μl of Oligofectamine (Invitrogen) in 190 μl of Opti-MEM (Invitrogen) for 20 minutes at room temperature. During this time, the cells were fed with 1 ml of fresh medium. Cells were incubated with 200 μl of the Oligofectamine/siRNA mixture for 24 hours at 37°C and then fed the next day with fresh medium.

Reverse Transcription-Polymerase Chain Reaction

  • β-Actin: forward, 5′-ATCTGGCACCACACCTTCTACAATGAGCTGCG-3′; reverse, 5′-CGTCATACTCCTGCTTGCTGATCCACATCTGC-3′; annealing temperature (Ta) = 60;

  • NOTCH-2: forward, 5′-TCGTGCAAGAGCCAGTTACCC-3′; reverse, 5′-AATGTCATGGCCGCTTCAGAG-3′; Ta = 60;

  • CBF-1: forward, 5′-TCCTGTGCCTGTGGTAGAGA-3′; reverse, 5′-ACTGTGGCTGTAGATGATGTGA-3′; Ta = 60;



  • TLE-4: forward, 5′-TGGCTGCTTTCTTCCCCCTTTCTC-3′; reverse, 5′-CCCGCCGGCAGCTCCTCTCG-3′; Ta = 53;

  • Id1: forward, 5′-CACCCTCAACGGCGAGATC-3′; reverse, 5′-CCACAGAGCACGTAATTCCTC-3′; Ta = 61;

  • Id3: forward, 5′-GCGGCTGCTACGAGGCGGTGT-3′; reverse, 5′-AAGTGGGCAGGGCGAAGTTGG-3′; Ta = 63;

  • Hey 2: forward, 5′-CGACGTGGGGAGCGAGAACAA-3′; reverse, 5′-GTGGCGCAAGTGCTGAGATGAGAC-3′; Ta = 57.


The following primary monoclonal antibodies were used in Western blotting and immunostaining after titration: bTAN 20 (rat IgG anti-NOTCH1) [26], Developmental Studies Hybridoma Bank, Iowa City, IA, http://www.uiowa.edu/∼dshbwww; C651.6DbHn (rat IgG anti-NOTCH2) [26], Developmental Studies Hybridoma Bank; and T6719 (rat IgG anti-CBF-1) [27], a gift from the Institute of Immunology, Tokyo. The monoclonal antibody TRA-1–60 (mouse IgM) was prepared and used as described previously [28].


For immunocytochemistry all cells were fixed in 4% paraformaldehyde for 15 minutes at room temperature. Cells were blocked and stained in 1% goat serum (Sigma-Aldrich); this was supplemented with 0.01% Triton X-100 during staining of intracellular proteins (NOTCH1 and NOTCH2). Incubations with all antibodies were for 30 minutes at room temperature while shaking. For detection of nuclear antigens, cells were stained on glass coverslips and costained with the nuclear dye Hoechst 33342 (Sigma-Aldrich) at a concentration of 5 μg/ml for 10 minutes prior to mounting in CFPVOH semipermanent mountant with AF100 antifadent (Citifluor, London, http://www.citifluor.co.uk). The staining was imaged using an Olympus IX70 inverted microscope with mercury arc illumination (Olympus UK Ltd., Watford, U.K., http://www.olympus-global.com) with a Hamamatsu ORCA ER black and white cooled charge-coupled device camera (Hamamatsu, Hamamatsu City, Japan, http://www.hamamatsu.com) controlled via Simple PCI software (Hamamatsu). Overlays were generated from single channel images for each fluorochrome used. All images were adjusted to same contrast level using the levels tool and overlaid in Photoshop (Adobe Systems Inc., San Jose, CA, http://www.adobe.com).

Western Blotting

To prepare protein extracts, cells were recovered by trypsinization (5 minutes for ES cells and 2 minutes for EC cells), counted, and resuspended in lysis buffer (1% [wt/vol] Nonidet P40, 1% [vol/vol] sodium deoxycholate, 0.1 mM phenylmethylsulfonyl fluoride in phosphate-buffered saline [PBS]) at a concentration of 2.5 × 107 cells per milliliter. Genomic DNA was sheered and removed by passing the lysate through a 21-gauge needle followed by centrifugation at 10,000g for 10 minutes at 4°C. To prepare samples for SDS-polyacrylamide gel electrophoresis, protein extracts were heated in 1 × sample loading buffer diluted 1:6 from 6 × loading buffer (0.125 M Tris-HCl, pH 6.8, 4% SDS, 3% glycerol, 0.02% β-mercaptoethanol, 0.02% bromphenol blue in water) at 60°C for 5 minutes. Samples were run on 6% or 10% gels at a concentration of 1.25 × 105 cells per lane and transferred to nitrocellulose membrane. Blocking and antibody incubations were carried out in 5% fat-free milk powder in PBS for 1 hour at room temperature followed by extensive washing. Visualization was achieved using enhanced chemiluminescence (GE Healthcare, Little Chalfont, U.K., http://www.gehealthcare.com). Antibodies and concentrations were as follows: anti-NOTCH1 antibody (bTAN 20, Developmental Studies Hybridoma Bank), tissue culture supernatant, diluted 1:5; anti-NOTCH2 antibody (C6576.DbHn, Developmental Studies Hybridoma Bank), ascites fluid, diluted 1:1,000; and anti-mCBF-1 (Rat IgG, a gift from the Institute of Immunology), diluted 1:2,500 [27]. Anti-mouse IgG horseradish peroxidase-labeled secondary antibody (Santa Cruz Biotechnology Inc., Santa Cruz, CA, http://www.scbt.com) (diluted 1:1,000) was used to detect the NOTCH2 primary; anti-rat IgG horseradish peroxidase-labeled secondary antibody (Santa Cruz Biotechnology) (diluted 1:5,000) was used to detect NOTCH1 and CBF-1.

Flow Cytometry

Cell surface antigen expression was assayed by immunofluorescence and flow cytometry as described previously using the following monoclonal antibodies: MC631, anti-SSEA3; MC813-70, anti-SSEA4; TRA-1–60; and TRA-1–81. Immunofluorescence of specific antibodies was compared with that of a negative control antibody obtained from the parent myeloma cell line P3X63Ag8 as described previously.

Apoptosis Analysis

Apoptotic analysis of human ES and human EC cells was carried out by costaining live cells with recombinant human fluorescein isothiocyanate (FITC)-conjugated Annexin V (R&D Systems Inc., Minneapolis, http://www.rndsystems.com) and propidium iodide (PI) (Sigma-Aldrich). Staining with Annexin V/PI was carried out on floating and attached cells as follows. Floating cells in the medium were collected and retained. Attached cells were then harvested via treatment with trypsin/EDTA for 2 minutes at 37°C and added to the floating cell suspension. Floating and attached cells were then pelleted via centrifugation and resuspended in Annexin buffer (10 mM HEPES, 140 mM NaCl, 2.5 mM CaCl2, pH 7.4) at a concentration of 1 × 106 cells per milliliter. A total of 1 × 105 cells (100 μl of the cell suspension) were incubated with 10 μl of Annexin V-FITC for 30 minutes at room temperature in the dark. Costaining with PI was carried out at a concentration of 50 μg/ml for 2 minutes prior to analysis by flow cytometry. During apoptotic analysis of human ES cells, costaining with the stem cell marker SSEA3 was carried out prior to staining with Annexin V/PI to facilitate the analysis of cell death in the undifferentiated stem cell compartment. Staining for SSEA3 was carried out as described above with the exception that antibody incubations were carried out for 5 minutes at room temperature.

Cell Cycle Analysis

To analyze cell cycle phase distribution by PI staining, human EC and human ES cells were harvested as single cells using trypsin/EDTA, washed three times in Ca2+, Mg2+-free PBS, and then fixed for 48 hours in 70% ethanol at 4°C. Enzymatic removal of RNA was carried out using 0.25 units of RNase per 5 × 105 cells at room temperature. Cells were then stained with PI at a final concentration of 50 μg/ml at 4°C. Flow cytometric analysis was carried out on 10,000 gated events using a CyAn ADP benchtop flow cytometer (DakoCytomation, Glostrup, Denmark, http://www.dakocytomation.com). Cells were first gated according to size and then according to the area and intensity of PI to exclude doublets. Cell cycle phase distribution was analyzed using Modfit Flow Cytometry Modeling Software, version 3 (Verity Software House, Topsham, ME, http://www.vsh.com).

Statistical Analysis

Single-time-point data were analyzed using the t test procedure using SAS, version 9.1 (SAS Institute, Cary, NC, http://www.sas.com); equality of variance was conformed. For RNA interference data, multiple regression analysis was carried out using the PROC MIXED algorithm in SAS, version 9.1. (SAS Institute).


NOTCH Receptors Undergo Proteolytic Cleavage in Human EC and ES Cell Cultures

To assess whether the NOTCH receptors are expressed in human EC and ES cells and undergo the four cleavage events to yield the S1, S2, S3, and S4 fragments required for canonical NOTCH signaling, we used Western blot analysis with two monoclonal antibodies that recognize the intracellular domain of NOTCH1 and NOTCH2 (Fig. 1). Both antibodies identified a band corresponding in size to the S1 and S3 product from both NOTCH1 and NOTCH2 in extracts from two human EC cell lines, 2102Ep and NTERA2; two karyotypically normal human ES cell lines, H1, H7.s14; and one culture-adapted human ES cell line, H7.s6. A third band, most likely representing the product released following S4 cleavage, was also visible in samples analyzed for NOTCH2, but no equivalent band was ever seen during Western blot analysis for NOTCH1. Similarly, a band corresponding to S2 was rarely, if ever, detectable for NOTCH1 but could, on occasion, be seen in samples probed for NOTCH2. The presence of the S3 and S4 bands indicates that the NOTCH receptors undergo proteolytic cleavage and suggests that NOTCH signaling through NOTCH1 and NOTCH2 may be active in all these cell cultures.

Figure Figure 1..

Cleavage of NOTCH proteins was detected in a variety of human embryonic stem and human embryonal carcinoma cell lines using monoclonal antibodies specific for the intracellular domain of NOTCH1 and NOTCH2. Both receptors appear to undergo S1 and S3 cleavage. Production of S4 cleavage products was consistently seen only in the case of NOTCH2. Neither antibody reacts significantly with protein samples isolated from Mefs. Abbreviations: Mef, mouse embryonic fibroblast; MW, molecular weight; NEXT, c-terminal intracellular fragment of the NOTCH receptor; NICD, transcriptionally active form of Notch; TIMC, functionally mature heterodimeric NOTCH receptor.

Expression and Localization of NOTCH Proteins in Undifferentiated Human EC and ES Cells

By costaining fixed cells with the antibodies to the NOTCH intracellular domain and the stem cell marker TRA-1–60, we confirmed the presence of cleaved NOTCH receptors in the pluripotent stem cell compartment. The intracellular domain of NOTCH1 was detected in most TRA-1–60-positive human EC and ES cells (Fig. 2A), Nuclear localization of NOTCH was evident in the EC cells, but it was much more diffusely spread, without clear nuclear localization, in the ES cells. By contrast, the expression pattern of NOTCH2 was almost identical in ES and EC cells (Fig. 2B), in which it was readily detected in the nuclei of all TRA-1–60-positive cells (supplemental online Fig. 2B).

Figure Figure 2..

Photomicrographs illustrating the expression of the NOTCH receptors NOTCH1 and NOTCH2 in human pluripotent stem cells. (A): TRA-1–60 and N1 expression in NTERA2 embryonal carcinoma (EC) and H1 embryonic stem (ES) cells (magnification, ×60): TRA-1–60 (Aa, Ae), N1 (Ab, Af), Cy3 secondary-only control (Ac, Ag), and Hoechst (Ad, Ah). (B): TRA-1–60 and N2 expression in NTERA2 EC and H1 ES cells (magnification, ×60): TRA-1–60 (Ba, Be), N2 (Bb, Bf), Cy3 secondary-only control (Bc, Bg), Hoechst (Bd, Bh). TRA-1–60. Abbreviations: N1, NOTCH1; N2, NOTCH2.

NOTCH Signaling Is Required for the Proliferation of Pluripotent Human Stem Cells

We next sought to investigate the role of NOTCH signaling in human stem cells using transient siRNA to knock down expression of NOTCH1, NOTCH2, and CBF-1 in the pluripotent human EC cell line NTERA2. We observed a marked reduction in levels of the corresponding proteins within 1–2 days post-transfection (Fig. 3A). At the same time, several known downstream target genes of NOTCH were also downregulated, including TLE1, TLE4, and HEY2, although HES1, a target of NOTCH signaling in other systems, was not affected. These knockdowns also resulted in downregulation of ID1 and ID3, two members of the ID family that encodes helix-loop-helix transcription factors. Notably, these have been implicated in control of mouse ES cells, but as targets of BMP rather than NOTCH signaling [4].

Figure Figure 3..

Canonical NOTCH signaling is important for the survival of human embryonal carcinoma (EC) cells. (A): Reduction in the expression of NOTCH1, NOTCH2, and CBF-1 induced by treatment of the human EC cell line NTERA2 with small interfering RNA (siRNA), shown by Western blot analysis, and the consequent downregulation of putative downstream target genes assessed by reverse transcription-polymerase chain reaction. (B): The morphology of NTERA2 cells 4 days after treatment with siRNA. Scale bars = 100 μm. (C): Growth curves of NTERA2 cells treated with siRNA to NOTCH1, NOTCH2, and CBF-1. (D): Induction of apoptosis as measured by staining for Annexin V and PI. Abbreviations: MW, molecular weight; PI, propidium iodide; RNAi, RNA interference.

Following knockdown of NOTCH2 and CBF-1, the growth of the cells was reduced (Fig. 3B, 3C) in comparison with a mock transfection and a transfection of siRNA against a gene (Ok(a)) whose removal has no consequence for pluripotent cells (p < .05 and p ≪ .001 at days 4 and 6 after siRNA treatment, respectively). NOTCH2 and CBF-1 knockdowns also resulted in an increase in cell death due to apoptosis, which was evident at day 4 in comparison with a mock transfection (p < .05 and p < .0001, respectively; n = 6) or knockdown of OK(a) (p < .05 and p < .0001, respectively; n = 6) (Fig. 3D). A reduction in growth was also apparent after knockdown of NOTCH1, but this was not statistically significant, and no obvious cell death occurred despite the reduction in growth rate (Fig. 3B–3D). Similar results for NOTCH2 and CBF-1 were also obtained with 2102Ep human EC cells (supplemental online Fig. 4).

To confirm the role of NOTCH signaling in the maintenance of EC cells, we used the γ-secretase inhibitor L-685,458 [29]. γ-Secretase is a multienzyme complex responsible for cleavage of the NOTCH intracellular domain in response to ligand binding, resulting in activation of the NOTCH signaling cascade [30]. As with inactivation of the NOTCH signaling pathway by siRNA, exposure of NTERA2 human EC cells to L-685,458 resulted in a significant reduction in cell growth (DMSO control compared with L-685,485, p < .05; n = 3) and marked death of the cells due to apoptosis (DMSO control compared with L-685,485, p < .05; n = 3) (Fig. 4). This response to L-685,458 was dose-dependent (supplemental online Fig. 5).

Figure Figure 4..

Human embryonal carcinoma (EC) cells undergo apoptotic cell death when treated with the γ-secretase inhibitor L-685,485. (A): The morphology of the human EC cell line NTERA2 following 3 days coculture with L-685,485. Note the high levels of cell death in comparison with DMSO and untreated controls. Scale bars = 100 μm. (B): Induction of apoptosis following 3 days of coculture with inhibitor L-685,458, assessed by Annexin V and PI staining and by cell counting. Abbreviations: DMSO, dimethyl sulfoxide; PI, propidium iodide.

We then tested the effect of the L-685,485 inhibitor on human ES cells. As with the EC cells, the inhibitor caused a substantial reduction in growth rate in H1 and early- and late-passage sublines of H7 human ES cells, H7.s14 and H7.s6, respectively (DMSO vs. L-685,458 treatment; H7 S6, p < .05, n = 3; H7 S14, p ≪ .01, n = 3; H1, p ≪ .01, n = 3) (Fig. 5A, 5B). In contrast to the EC cells, no substantial cell death was observed in the ES cultures (Fig. 5C) following treatment with L-685,458. However, H7.s14 were found to undergo a statistically significant increase in apoptosis in comparison with untreated cells and control cells treated with DMSO (DMSO vs. L-685,485 treatment, p = <0.0001, n = 3) when examined more closely using Annexin V to detect specifically apoptotic cells. In addition, H7 S6 cells showed a significant increase in G1/G0 and a significant decrease in S-phase for cells treated with L-685,485 compared with DMSO-treated controls (p < .05, n = 3, and p < .05, n = 3, respectively) (supplemental online Fig. 6), although, surprisingly, they did not appear to undergo significant levels of cell death (Fig. 5C). These data suggest that NOTCH signaling plays a role in regulating the proliferation of human ES cells and their malignant counterparts.

Figure Figure 5..

The response of human embryonic stem (ES) cells to inhibition of NOTCH signaling. (A): The morphology of human ES cell lines H1, H7.s14, and H7.s6 following treatment with the NOTCH inhibitor L-685,485. Note the marked reduction in colony size Scale bars = 100 μm. (D): Reduction in cell number after 5 days of coculture with L-685,485. (B): Reduction in cell number after 5 days of coculture with L-685,485. (C): Induction of apoptosis in human ES cells after 5 days of coculture with inhibitor L-685,458, assessed by Annexin V and PI staining. Abbreviations: DMSO, dimethyl sulfoxide; PI, propidium iodide.

Despite the reduction in cell proliferation that resulted from an inhibition of NOTCH signaling, the human ES cells retained an undifferentiated stem cell phenotype, as measured by morphological examination (data not shown) and as indicated by surface antigen expression (Fig. 6A). This is in contrast, for example, to retinoic acid-treated cultures of H7 S6 cells, which showed a significant reduction in surface marker expression (Fig. 6B), in accordance with our previous results [7].

Figure Figure 6..

Cell surface antigen expression of human embryonic stem (ES) cells following treatment with the GSI L-685,485. (A): Human ES cells were treated with GSI for 5 days and assayed for cells surface antigens associated with the pluripotent state. Results are displayed as the percentage of positive cells. (B): Example histograms of surface marker expression (percentage positive and mean fluorescence) of cells treated for 5 days with GSI or RA compared with untreated or DMSO-treated cultures. Antibodies used were P3X (negative control) anti-SSEA4, TRA-1–60, and TRA-1–81. The analysis gate used to estimate the percentage of positive cells is indicated by the bar marked R2. Abbreviations: DMSO, dimethyl sulfoxide; GSI, γ-secretase inhibitor; RA, retinoic acid; SSEA, stage-specific embryonic antigen.


A significant difference between mouse and human EC and ES cells is their behavior at low cell densities. Whereas mouse EC and ES cells proliferate and clone as undifferentiated cells under such conditions, human EC and ES cells tend to differentiate and/or grow poorly under such conditions. Earlier data suggested that the low density behavior of human EC cells is due to a loss of direct cell-cell contact [9], although apparently not involving cadherin-based signaling [31]. Our observations now show that the NOTCH pathway provides a cell-cell proliferation signal for human ES cells and their malignant counterparts, EC cells.

Consistent with this, it has been reported that stimulation of the NOTCH pathway increases the plating efficiency of human ES cells by activation of the serine/threonine kinase AKT and a novel phosphorylation site on STAT-3 [32]. Phosphoinositide-3 kinase/Akt signaling has also been found to play an essential role in preventing apoptosis of human ES cells [33, 34]. In another study, Noggle et al. [35] also found that the γ-secretase inhibitor N-[(3,5-Difluorophenyl)acetyl]-L-alanyl-2-phenyl]glycin e-1,1-dimethylethyl ester reduced the proliferation of two human ES cell lines, BGN01 and BGN02, although they attributed this to a reduction in spontaneous differentiation. In our experiments, however, we did not find evidence that NOTCH signaling affects the undifferentiated state of the ES cells when grown under standard conditions, but rather affects cell cycle progression and apoptosis, especially in the case of EC cells. Lowell et al. [36] did find that NOTCH signaling plays a role in regulating differentiation, particularly promoting selection of the neural lineage, but only when the cells are grown under conditions favoring differentiation.

The NOTCH pathway has been linked to the progression of a wide variety of human cancers and can function as either an oncogene or tumor suppressor depending upon the cellular context [37]. For example, hyperactive NOTCH signaling is believed to act as a survival factor during the formation of T-cell leukemia [38]. We have previously proposed that NOTCH signaling could also play a significant role in the etiology of teratocarcinomas, a form of germ cell tumor for which EC cells provide the malignant stem cells [22, 23]. Our present observations support this hypothesis. It is notable that human EC cells appear to be more sensitive to apoptosis than human ES cells in the absence of NOTCH signaling, which might reflect an exaggerated dependence of EC cells resulting from a loss of regulation and overstimulation of NOTCH signaling during tumorigenesis, or it might reflect their origin from primordial germ cells [39]. In the case of human ES cells, we found that inhibition of NOTCH signaling resulted in inhibition of cell cycle progression with accumulation of cells in G0/G1; the acquisition of an apoptotic response might reflect an acquisition of additional changes during their tumorigenic progression. It was notable, however, that upregulation of a putative NOTCH ligand, DLK1, can be a feature of human ES cell culture adaptation, which we have suggested might reflect the partial progression of a malignant phenotype [18, 40].

The transcription factor Id-1, which we have found to be regulated in response to NOTCH signaling in human EC cells, has also been identified as a downstream target of the NOTCH pathway during T-cell leukemogenesis [41]. In addition, members of the nuclear factor κB (NF-κB) family of transcription factors have emerged as important downstream mediators of the antiapoptotic NOTCH signaling during development and tumorigenesis [42, [43]–44]. NF-κB has also recently been reported to function as a survival factor and downstream component of CD30 in culture-adapted human ES cells and may also be active in karyotypically normal populations [45]. These observations raise the possibility that NF-κB and NOTCH may collaborate to regulate human ES and EC cell survival.

The role of NOTCH signaling in the growth of human EC and ES cells provides yet another difference from the corresponding mouse cells. This is not entirely unexpected given that mouse ES cells do not appear to require cell-cell contacts to the same extent as their human counterparts and may readily be passaged as single cells. Mice homozygous for a CBF-1 knockout mutation and lacking canonical NOTCH signaling survive until after gastrulation, whereas ES cells derived from these embryos are viable and able to proliferate [46]. Thus, NOTCH signaling does not appear to be required either for development of the inner cell mass and epiblast in mouse embryos or for the maintenance of mouse ES cells in culture. Nevertheless, it is striking that both ID1 and ID3, which are regulated by BMP in mouse ES cells and have been suggested to play a role in maintenance of the undifferentiated state of these cells, are apparently regulated by NOTCH in the human cells. It may then be that the same fundamental intrinsic regulatory mechanisms operate in both human and mouse ES cells but that these mechanisms are subject to different environmental controls, reflecting the different circumstances of embryonic development in these different species.

Disclosure of Potential Conflicts of Interest

P.W.A. owns stock in and has served as an officer or member of the Board for Axordia Ltd., and he has a financial interest in Wistar Institute (Philadelphia). P.J.G. owns stock in Axordia Ltd.


This work was supported by grants from the Biotechnology and Biological Science Research Council, Juvenile Diabetes Research Foundation, Medical Research Council, and the Anatomical Society of Great Britain and Ireland. We are also grateful to the Institute of Immunology, Tokyo, for the kind of gift of the CBF-1 antibody used in this study. V.F. and P.J.G. contributed equally to this work. V.F. is currently affiliated with the Center for Stem Cell and Regenerative Medicine, Keck School of Medicine, University of Southern California, Los Angeles.