Unlocking the Secrets of Cancer Stem Cells with γ-Secretase Inhibitors: A Novel Anticancer Strategy

The dysregulation of Notch signaling is associated with a wide variety of different human cancers. Notch signaling activation mostly relies on the activity of the γ-secretase enzyme that cleaves the Notch receptors and releases the active intracellular domain. It is well-documented that γ-secretase inhibitors (GSIs) block the Notch activity, mainly by inhibiting the oncogenic activity of this pathway. To date, several GSIs have been introduced clinically for the treatment of various diseases, such as Alzheimer’s disease and various cancers, and their impacts on Notch inhibition have been found to be promising. Therefore, GSIs are of great interest for cancer therapy. The objective of this review is to provide a systematic review of in vitro and in vivo studies for investigating the effect of GSIs on various cancer stem cells (CSCs), mainly by modulation of the Notch signaling pathway. Various scholarly electronic databases were searched and relevant studies published in the English language were collected up to February 2020. Herein, we conclude that GSIs can be potential candidates for CSC-targeting therapy. The outcome of our study also indicates that GSIs in combination with anticancer drugs have a greater inhibitory effect on CSCs.


Introduction
Despite the remarkable progress being made in cancer treatment, cancer is the leading cause of death worldwide. There is evidence that a rare subset of cancer cells are responsible for resistance to therapy and holding stemness functions/properties, which are known as cancer stem cells (CSCs). Therefore, these cell subpopulations induce tumor perpetuation, even after effective therapies, and result in aggression of the tumor. The CSC theory of cancer progression proposes that a tumor is a hierarchically organized tissue with CSCs at the top position in the hierarchy, which produces more differentiated cancer cells with a reduced or restricted potential of proliferation. In recent years, the CSC theory has been subjected to increasing attention and excitement, with researchers believing this theory would augment our knowledge about the molecular and cellular events within tumor progression, contributing to metastasis, recurrence, and resistance to therapy [1]. First, the Notch receptor interacts with the Notch ligand, such as DLL, and initiates proteolytic cleavage at the extracellular site, followed by cleavage at the intracellular site by γ-secretase, leading to the release of NICD. Then, NICD is translocated into the nucleus, where it interacts with CSL and MAML to form a transcription-activating complex. GSIs can inhibit these steps, including receptor/ligand binding, the release of NICD, and the interaction of NICD and downstream targets, as well as NICD protein stability, and can thus have anticancer effects. Abbreviations: CSL, CBF1, suppressor of hairless and LAG1; DLL, Delta-like ligand; EGFR, epidermal growth factor receptor; EMT, epithelial-mesenchymal transition; GSIs, γ-secretase inhibitors; HER2, human epidermal growth factor receptor 2; HIF1α, hypoxia-inducible factor 1α; MAML, mastermind-like; NF-κB, nuclear factor-kB; NICD, Notch intracellular domain; PI3K, phosphatidylinositol-3-kinase; PTEN, phosphatase and tensin homolog.

Methodology for the Literature Search and Study Selection
The current systematic review was conducted following the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines [32]. The objective of this review is to provide a systematic review of in vitro and in vivo studies for investigating the impact of GSIs on cancer stem cells. Various electronic scholarly databases, including Scopus, PubMed, Science Direct, Google Scholar, and Web of Science, were searched and relevant studies in English language were collected up to February 2020. The search syntax included "cancer stem cell" OR "tumor stem cell" OR "initiating tumor cell" OR "neoplastic stem cell" OR "colony forming unit" AND "gamma-secretase inhibitor" OR "γ-secretase inhibitor". The primary search was conducted by two researchers separately, and unrelated studies were excluded based upon their titles and abstracts. Review articles, meta-analyses, books, book chapters, conference abstracts, case reports, clinical trials, and non-English articles were also excluded. An overview of the literature search and selection process is presented in Figure 3.

Anticancer Activities of GSIs against CSCs
Among 118 eligible articles, 64 and 4 studies were performed using in vitro cancer cell lines and in vivo animal models, respectively, and in 50 studies, both in vitro and in vivo models were used. Considering the major aspects of the total included studies and based upon the location of the cancer, the results are presented in the next sections (Tables 1 and 2).

Adenoid Cystic Carcinoma
A study by Panaccione et al. [33] investigated the anticancer mechanism of GSIs in adenoid cystic carcinoma (ACC) cell lines under both in vitro and in vivo conditions. The in vitro study was performed by treating Accx11 cells with 1-10 µM DAPT with or without radiation for 24-96 h. For the in vivo assay, mice were treated with 50 mg/kg DAPT for 35 days. The treatment blocked S-phase kinase-associated protein 2 and the Notch 1 intracellular domain (N1-ICD), suppressed the growth of ACC in vivo, reduced CD133 + cells and sensitized them to radiation, and led to the induction of cyclin-dependent kinase inhibitor 1B (p27 Kip1 ). Therefore, the combination of radiation and GSI exerted a greater effect on the elimination of CD133 + cells compared with either agent alone.

Blood Cancer
In an experiment, different lymphoma cell lines were exposed to 0.1-100 µM L685458 for 24 h and subjected to a colony-forming unit (CFU) assay. L-685458 potently blocked the CFU and reduced the lymphoma stem cell population [34].
In another study, Ikram et al. (2018) used a three-dimensional (3D) cell culture to assess the antitumor effects of DAPT. The pretreatment of lymphoma cells with DAPT (5 µM) with or without NSC23766 (Rac-specific small-molecule inhibitor) for 24 h showed a significant decrease of the lymphoma stem cell population and an increased sensitivity to doxorubicin [35].
The administration of varying concentrations of DAPT (8-16 µM) for 14 days decreased the colony number in leukemic stem cells, leading to a decline in the size of large colonies by suppressing their proliferation in a concentration-dependent way [36]. Furthermore, another study showed that treatment with MRK-003 (150 mg/kg) eliminated leukemia-initiating cells in a Tal1/Lmo2 mouse model of T-cell acute lymphoblastic leukemia (T-ALL) [37].
In a T-ALL cell line with NOTCH1 mutations (DND-41) and AML cell line (NB4), GSI-XXI (compound E, 10 µM) treatment, alone or in combination with cyclopamine or quercetin for 1-7 days, suppressed the colony formation ability of these cells. The addition of cyclopamine or quercetin to compound E enhanced the inhibitory effect on DND-41 cell line growth and blocked the activation of NOTCH1 in NB4 cell lines [38].
In another study, AA and HEL erythroid leukemia cells were cultured with three GSIs (GSI-IX, 20 µM; GSI-XII, 5 µM; and GSI-XXI, 10 µM) for 1-7 days. The study claimed that treatment with GSI induced the differentiation of morphologic erythroid and enhanced the production of hemoglobin. It also showed that treatment with GSIs inhibited the colony formation ability and short-term cell growth, while GSI-XXI treatment enhanced the AA cell line growth [39].
In multiple myeloma cancer stem cells (MM-CSCs), the effect of bruceantin (a quassinoid isolated from Brucea species) was evaluated in the presence of GSI. Bruceantin effectively controlled the MM-CSCs' viability, migration, proliferation, and angiogenesis. MM-CSC pretreatment with the GSI (RO4929097, 10 µM) and increasing doses of bruceantin for 1 day inhibited the proliferation of these cells [40].

Brain Cancer
In brain cancer cell lines, it was established that the suppression of Notch signaling with DAPT inhibited hypoxia-induced GSC expansion [41]; abolished the effects of STC1 on N1-ICD production, SOX2 expression, and the sphere-forming capacity [42]; reduced the CSC of CD133 + and inhibited the proliferation of SHG-44 cells [43]; suppressed the transition from CD1331/CD1442 to double-positive (DP) [44]; inhibited cell growth and reduced the sphere formation capacity in glioblastoma neurosphere cultures [45]; and downregulated HIF-1α and hes1, reduced the number of nestin + cells, increased the number of β-III-tubulin + cells, and enhanced MKI67 and neuronal differentiation [46]. However, one study showed that DAPT treatment reduced brain CSCs, but had no survival benefit for mice injected with DAPT-treated GBM neurosphere cells [47].
DAPT treatment in combination with radiation [48], gleevec and amph1D peptide [49], D341Med with HBMEC [50], and imatinib [51], resulted in an increase of apoptosis and radio-sensitivity in ihBTC2 cells [48]; the induction of neurosphere dispersion that resulted in cell death [49]; the downregulation of Bmi-1, CDK6, c-Myc, and CCND1 expression in D341Med, and a reduction in the tumor size and volume [50]; and the effective growth inhibition of GBM cells [51].
The administration of DAPT and INCB3619 downregulated the expression of HES1 and HEY1 Notch target genes in both 0822 and 0308 cell lines. In the 0308 cell line, INCB3619 and DAPT also downregulated the expression of YKL-40/CHI3L1, while the survival was prolonged in mice [52].
In four different studies, DAPT, L685,458, BMS-708163, and RO4929097 treatment led to an increase of the ASCL1 levels in ASCL1 hi GSCs and a decrease in sphere-forming cells (SFCs) [53]; inhibited glioma tumor-initiating cell growth in a concentration-dependent manner, suppressed tumor growth, and prolonged the survival rate in vivo [54]; increased radiation-induced apoptosis and decreased the clonogenic survival of GSCs [55]; and decreased the number of CSCs by reducing proliferation and increasing cell death that was associated with decreased levels of STAT3 and Akt phosphorylation and resulted in the inhibition of tumor growth and enhancement of the survival rate [56].
Upon the usage of different concentrations of GSI-18 in vitro and in vivo, two studies reported a reduction in Hes1 protein and mRNA levels in DAOY cells, the suppression of clonogenicity, and the induction of anticancer effects mediated by suppression of the Notch signaling pathway [57], and the induction of a phenotype transformation towards non-tumorigenic cells, along with a decrease in proliferation and increase in differentiation, as well as apoptosis [58].
MRK-003, alone or combined with GSNO or chloroquine, reduced the baseline side population in primary glioma cultures and suppressed the increase of the side population induced by GSNO [59]; prevented neurosphere formation in HCMV-infected GBM cells and reduced the functionality or number of CSCs [60]; decreased the viability and sphereformation capacity and increased apoptosis through suppression of the Akt pathway [61]; and induced autophagy in glioma neurosphere lines and reduced cell proliferation, cell growth, and the colony formation ability [62].
GSI-I treatment sensitized U251 and U87 cell lines to radiation through the reduction of radio-resistant CD133 + cells, enhanced the radio-sensitivity in cancer cells, and suppressed the tumor growth [63]. GSI-I also enhanced the therapeutic effect of temozolomide and led to an increase in CD133 + glioma cytotoxicity [64].
In a study by Pietras et al. [65], MK-003 (10 µM), alone or in combination with tetradecanoyl phorbol acetate, suppressed the glioma primary cells induced by PDGF and eliminated the cancer cells expressing stem cell markers.
In GSCs, RO4929097, either alone or in combination with farnesyltransferase inhibitors, blocked the Akt pathway and inhibited the cell-cycle progression, thus enhancing the radiosensitivity and reducing the tumor growth. This combination, in addition to radiation, led to a durable response in orthotopic tumor models [66].
Treatment with MK0752 (25 µM) decreased the proliferation and self-renewal ability of GSCs and reduced the number of secondary neurospheres by differentiating GSCs into less proliferative glioma progenitor cells [67].
Another study by Dai et al. [68] investigated the change of glycosylation patterns upon treatment with GSI in GBM CSCs. For this purpose, compound E was cultured with these cancer cells, resulting in a phenotype transformation of CSCs toward a less tumorigenic form upon compound E treatment. Moreover, GSI-II treatment (0.2 µg) for 20 days effectively suppressed the CSC generation in U87 cells and significantly abrogated the proliferation and differentiation of U87 tumor-initiating cells [69].

Breast Cancer
The outcomes of seven in vitro and in vivo studies on breast cancer cells indicated that suppression of the Notch pathway with DAPT suppressed the activation of Notch by integrin-linked kinase (ILK). ILK knockdown blocked breast CSCs in vitro [70]; reduced the expression of Notch signal effectors NICD, Jagged 1, HES1, and signal peptide CUB EGFlike domain-containing protein 2 (SCUBE2); and decreased the self-renewal ability of breast CSCs [71]. It also blocked the cleavage of the CD44 intracytoplasmic domain, reduced the mammosphere-forming ability, decreased cell invasion and proliferation of triplenegative breast cancer (TNBC) cells, and suppressed tumor formation in mouse xenograft models [72,73]. These studies also showed that treatment with DAPT decreased the number of mammary progenitor cells and stem cells in p53-deficient mammary epithelium [74], reduced the percentage of CD44 hi /CD24 lo cells, lessened micro-and macro-metastases in mice, and inhibited the colony formation ability of brain metastatic MDA-MB-231 cell lines [75]. Moreover, DAPT treatment led to a significant inhibition of Notch-mediated cell survival and invasion under hypoxia through increasing E-cadherin expression and suppressing the phosphorylation of Akt in breast cancer cells [76].
Five other studies reported that treatment with various concentrations of DAPT, alone or combined with different factors (e.g., radiation, lapatinib, gefitinib, tamoxifen, and 6-shogaol), led to inhibition of the mammosphere-forming ability and an increased TIC gene expression signature, and suppressed the expansion of CD44 + CD24 low+ TRCs after radiation [77], decreased mammosphere formation and the acini size in DCIS cell lines by inhibiting Notch and ErbB1/2 [78], blocked CSC activity induced by estrogen both in vitro and in vivo, and helped gefitinib to entirely suppress the estrogen effect [79]; inhibited the estrogen receptor-α promoter activity, reduced the tamoxifen sensitivity, and enhanced the expression of markers associated with basal-like breast cancers [80]; and blocked the spheroids and breast cancer cell proliferation and suppressed the colony formation capacity and number of spheroids through inhibiting the Notch pathway [81].
In a study by Mamaeva et al. [82], the anticancer effect of glucose-functionalized nanoparticles carrying DAPT on breast CSCs was assessed. To induce Notch suppression, the mesoporous silica nanoparticles were loaded with the compounds. In vitro, breast CSCs were exposed to variable concentrations of DAPT nanoparticles (5-50 µg/mL) for 24 h, indicating that DAPT treatment reduced the CSC population. In vivo, treatment with DAPT-loaded particles or free DAPT led to a significant decrease in the size of the cancer cell population. As a result, glucose-functionalized mesoporous silica nanoparticles carrying DAPT significantly eliminated the number of CSCs.
GSI-XVII therapy (5 µM, with or without radiation) decreased the self-renewing capacity and prevented the recombinant human erythropoietin-induced enhancement in primary sphere formation [83], while preventing the radiation-induced DLL3, Notch2, Jagged1, and DLL1 gene expression and significantly decreased the number of breast CSCs [84]. It was reported that various concentrations of GSI-I, ranging from 1-10 µM for 24, significantly inhibited breast CSCs [85].
The in vitro and in vivo treatment of different concentrations of MRK-003, either alone or in combination with lapatinib, trastuzumab, and docetaxel, resulted in Notch-1 inhibition, prevented mammosphere formation, inhibited the proliferation of bulk HER2 + HCC1954 cells, and prevented tumor relapse in xenograft models [86], and suppressed Notch signaling activation and decreased the number of CSCs, while a combination of MRK-003 and docetaxel enhanced this activity [87]. MRK-003 was also shown to decrease the viability of cells derived from tumorspheres in vitro, reduce tumor-initiating cells, and block the proliferation and self-renewal ability of mammosphere-resident cells, and induced their apoptosis and differentiation [88].
Two studies using different concentrations of DAPT and compound E with or without AD-01 reported that the treatment inhibited the growth and metastasis of the cancer stemlike cells of 231BrM in the brain through suppressing the expression of HES5 in vitro and in vivo [89]. In addition, the combined treatment of DAPT and compound E with AD-01 led to a reduction in the mammosphere-forming efficiency (MSFE) and enhanced the anti-CSC effects [90].
In another study, mammosphere or monolayer breast cancer cell cultures were treated with DAPT (10 µM), and MCF-7 cells were treated with dibenzazepine (DBZ) (10 µM) for 3 days. DAPT treatment decreased the proportion of ESA + /CD44 + /CD24 low cells. Both DAPT and DBZ significantly reduced the N1-ICD and decreased the HEY2 and HES1 expression. In an in vivo study, mice were treated with 1 mg/mL DBZ for 3 days. DBZ treatment completely ablated tumor initiation and significantly decreased the size and volume of MCF-7 tumors [91].
MK-0752 (25 µM) and RO4929097(10 µM), alone or in combination with tocilizumab, suppressed tumor growth, but enhanced the CSCs in breast cancer cells expressing Notch3, while inducing IL-6. Treatment with MK-0752 led to the induction of IL-6 through Hey2-Notch3 signaling inhibition. Furthermore, hypoxia-inducible factor 1α (HIF1α) downregulated breast CSCs by reducing the IL-6 levels in breast cancer cells expressing Notch3. Using in vivo xenograft models, the concurrent use of tocilizumab and MK-0752 caused a significant reduction in breast CSCs and suppressed tumor growth [93]. Another study also indicated that MK-0752 treatment with or without docetaxel in mice bearing human tumorgrafts reduced the primary and secondary mammosphere-forming efficiency (MSFE); decreased the ALDH + and CD44 + /CD24 − subpopulations; downregulated NICD, Hes1, Hey1, Hes5, and Myc; and reduced tumor growth. Taken together, GSI treatment decreased breast CSCs and increased the efficacy of docetaxel [31].
The co-culture of CD44 + CD24 low+ and CD44 + CD24 neg cells with different GSIs (RO492-0927, 10 µM and DAPT, 5-10 µM) for 1-12 days led to a significant reduction in N1-ICD and decreased the expression of Sox2 and the sphere formation ability. Taken together, the blockade of Notch decreased the Sox2 expression and colony-and sphere-forming capacity. In in vivo nude mice, RO4920927 inhibited tumor growth in CD44 + CD24 low+ cells [94].

Colorectal Carcinoma
DAPT treatment (10-20 µmol/L) led to the blockage of Notch signaling and partially suppressed the effect of KRAS on Hes1 in colorectal carcinoma cells. The DAPT treatment also reduced the number of CSCs [96].
In another study, the inhibition of Notch by DAPT significantly reduced the Lgr5-GFP cell population in Lgr5-EGFP-CreER T2 organoids. The suppression of Notch by GSI led to a reduction in Ascl2 levels and also enhanced apoptotic cells shed into the lumen [97]. Furthermore, the suppression of Notch by DAPT decreased the colon cancer stem cell (CCSC) population and enhanced the non-CCSC population. DAPT treatment also inhibited asymmetric division and decreased symmetric CCSC-CCSC division [98].
When utilizing soluble Jagged-1-Fc protein and DAPT in colorectal cancer cells, it was observed that the Notch-1 signaling pathway activates epithelial-mesenchymal transition (EMT)/stemness-associated proteins Slug, Smad-3, and CD44 by inducing the expression of Jagged-1. Treating the parental cells with DAPT decreased the proteins, such as Jagged-1, Smad-3, and CD44, and induced a significant reduction of Slug in the ICN1 cells [99].
The administration of JLK6 and DAPT led to a significant decrease in the number of colonspheres in SW620 cell lines [100]. Moreover, another study showed that GSI-X treatment suppressed endothelial cell conditioned medium-induced Notch signaling activation and CSC enrichment in HCT116 cells [101].
Treating mice bearing CRC cells with PF-03084014 (125 mg/kg), alone or combined with irinotecan, for 28 days resulted in a significant reduction in tumor recurrence and tumor growth and also decreased the ALDH + population. This combination therapy had a greater antitumor effect when compared to PF-03084014 or irinotecan alone [102].

Gastric Cancer
Hayakawa et al. [103] presented a study examining the effect of Notch signaling inhibition on gastric CSC elimination. For this purpose, organoids were co-cultured with 25 µM DAPT for 10 days. The addition of a GSI suppressed the growth of the corpus organoid. For in vivo experiments, the Notch inhibitor dibenzazepine was injected intraperitoneally (30 µmol/kg) into mice for 2 weeks. Dibenzazepine treatment decreased the Mist1-lineage tracing expansion and proliferation in the isthmus.
In another study, the CD44 + population of gastric CSCs was targeted by DAPT, both in vitro and in vivo. In gastric CSCs, DAPT (2.5-15 µM) treatment for 24-96 h decreased the size of the CD44 + population in a time-and concentration-dependent way. This study claimed that DAPT treatment led to the inhibition of the invasion, migration, and proliferation of CD44 + CSC [104].
Epithelial-mesenchymal transition (EMT) is a process associated with tumor initiation, invasion, metastasis, and resistance to therapy. The exposure of CD44 − and CD44 + cell lines with DAPT (10 µM) for 72 h suppressed the EMT markers and Hes1 expression, inhibited the CSC properties, and blocked the CD44 + cell proliferation and invasion. In addition, in vivo GSI treatment significantly suppressed the growth of CD44 + cell xenograft tumors [105].
The co-culture of CS12 and MKN45-133 + cells with DAPT (5µM) for a day resulted in the suppression of Notch1 activation and enhancement of CD133 and stemness genes. DAPT also inhibited the sphere-forming ability and decreased the size and number of spheres [106].
DAPT (25-50 µM) alone or in combination with cisplatin reduced the gastric cancer cell viability and decreased the number of CD44 high Lgr-5 high cancer cells, representative of CSC properties. Treatment with GSI alone did not affect cell proliferation [107].

Head and Neck Cancer
In a study by Chen et al. [108], the administration of DAPT (100 µM), alone or in combination with cetuximab and cisplatin, suppressed the viability of OECM1 cells. DAPT treatment or Kruppel-like factor 4 (KLF4) knockdown led to a significant reduction in the number of KLF4 + /CD44 + cells in overexpressing Twist1 OECM1 cells. The combination of DAPT and cetuximab exerted an antitumor effect on xenograft models of head and neck cancers.
The addition of DAPT (1-10 µM), alone or in combination with 5-fluorouracil (5-FU), to esophageal adenocarcinoma OE33 cells inhibited their growth. For in vivo studies, animals were treated with 20 mg/kg DAPT for up to 10 weeks. DAPT treatment suppressed tumor growth, decreased HES1 expression and the level of NICD, and also led to the enhancement of apoptosis and reduction in cell proliferation in vivo. Taken together, Notch signaling inhibition sensitized cancer cells to chemotherapeutic agents and eliminated CSCs [109].
In 2011, Mendelson et al. [110] examined the role of Compound E in Barrett's esophageal adenocarcinoma. For in vitro studies, cell lines were treated with TGF-β and Compound E (500 nM-5 µM) for 72 h. The results showed that the treatment only suppressed cell proliferation in BE3 cell lines with high Notch signaling and TGF-β dysfunction.
Another study demonstrated that the combination of DAPT (5-10 µM and 10-20 mg/kg) with chemotherapeutic agents (docetaxel, cisplatin, and 5-FU) led to a synergistic increase in chemotherapy-enriched head-neck CSCs, both in vitro and in vivo. Taken together, the inhibition of NOTCH1 signaling reduced the tumor self-renewal capacity and number of CSCs and also decreased transcription factors of self-renewal and markers related to CSCs [111].
The results of another study indicated that NOTCH inhibition (GSI XXI, 5-10 µM) resulted in inhibition of the spheroid-forming ability and also inhibited the survival, migration, and transformation of head and neck squamous cell carcinoma cells [112].

Liver Cancer and Cholangiocarcinoma
It was demonstrated that treatment with 10 µmol/L of different GSIs (L-685,485 and DAPT) inhibited the growth of the epithelial cell adhesion molecule (EpCAM)-positive fraction in hepatocellular carcinoma cells, and in vivo Notch suppression caused significant antitumor effects in hepatomas, showing that Notch inhibition could block the stem cell properties of hepatic cancer cells [113].
In another study, PF-03084014 suppressed the self-renewal ability and proliferation of CSCs. It also decreased the tumor growth in vivo and inhibited the metastasis of hepatocellular carcinoma to the lung [114].
Cao et al. [115] showed that Notch signaling independent of CSL (CBF1, suppressor of hairless and LAG1) might have an important role in hepatic CSCs, and MRK003 treatment significantly suppressed the sphere-forming capacity and reduced the size of the human stem-like hepatocarcinoma cell population. Similarly, sorafenib and PF-03084014 suppressed the self-renewal ability and proliferation of hepatocellular carcinoma spheroids [116].
One study showed that Notch pathway inhibition by either miR-34a overexpression or treatment with DAPT suppressed the growth and colony-forming ability of human cholangiocarcinoma cells. The results also indicated that targeting miR-34a combined with DAPT is an effective treatment for cholangiocarcinoma [117]. Additionally, the proportion of CD24 + CD44 + cells, colony-forming capacity, and mice tumorigenicity were suppressed. A combination of GEM and GSI significantly decreased viable TFK-1 and RBE cells compared with GEM alone [118].

Lung Cancer
Five studies revealed that DAPT, alone or in combination with cisplatin, suppressed the Notch pathway, reduced the number of primary pulmospheres, decreased the expression of the Notch pathway target genes (Hey1 and Hes1), and reduced the self-renewal capacity of primary spheres [119]; decreased the number of CD133 + cells induced by cisplatin and enhanced the sensitivity to doxorubicin and paclitaxel [120]; decreased the number of CD44 + /CD24 − cells and inhibited the growth capacity of lung CSCs [121]; significantly decreased the ALDH + lung cancer cells, and led to cell-cycle arrest, depletion in the number of ALDH + cancer cells in a concentration-dependent manner, and a reduction in tumor cell proliferation and clonogenicity [122]; and suppressed the proliferation of CD133 − and CD133 + cells and had a small effect on the cell cycle [123].
It was demonstrated that RO4929097 (1-10 µmol/L) with or without cisplatin modulated the self-renewing activity of LCSCs via p-STAT3 and HES1 in human non-small cell lung cancer (NSCLC) cells, whereas RO4929097 treatment increased their platinum resistance. Moreover, treatment with RO4929097 inhibited the self-renewal of LCSCs and increased the platinum sensitivity, both in vitro and in vivo [124].
In another study, the exposure of MRK-003 (5-20 µM), with or without docetaxel, inhibited the sphere formation and self-renewal ability and decreased the NICD2 expression. In a mouse tumor xenograft model, MRK-003 reduced the expression of downstream effectors of Notch signaling [125].
PF-03084014 (1 µM) treatment of lung cancer cells, either alone or in combination with erlotinib, resulted in the elimination of the erlotinib-induced stem-like cell population by reducing Notch signaling activity. The inhibition of Notch3 and EGFR receptors decreased the stem-like cell population expansion [126].
Ali et al. [127] reported that in vitro and in vivo treatment of lung adenocarcinoma cells with DBZ, with or without auranofin, led to the inhibition of oncosphere growth, cell viability, soft agar growth, and significantly inhibited tumor growth. This combination therapy resulted in a significantly higher level of inhibition compared with either compound alone.

Melanoma
In 2016, a study by Kumar et al. [129] investigated the effects of targeting Notch1 on CSC-mediated melanoma progression. Treatment with DAPT and L-685,458 suppressed the expression of CD133 in CD133 + cells and enhanced the number of CD133 − cells. The results of this study demonstrated that inhibition of the Notch signaling pathway led to CD133-dependent mitogen-activated protein kinase (MAPK) signaling inhibition and eventually increased the interaction of tumor-associated endothelial cells and the migration of CD133 + cells, and blocked metastasis, melanoma growth, and angiogenesis.
It has also been shown that the Notch pathway has an important role in the differentiation of tumor pericyte precursors. DFPAA treatment led to a reduction in the number of NG2 + cells (oligodendrocyte precursor cells or polydendrocytes) [128].
The effect of the combination of GSIs with a Bcl-2 inhibitor on killing melanomainitiating cells was investigated in a study. For both in vitro and in vivo studies, GSI-I (0.83 µM) was used, either alone or in combination with ABT-737, for 24-21 days. Combination therapy decreased the cell viability and promoted the non-melanoma-initiating cell apoptosis, inhibited primary sphere formation, decreased the number of ALDH + cells, and suppressed the melanoma-initiating cells' self-renewability ability in vitro. In a mouse xenograft model, combination therapy caused a significant decline in the tumor-initiating capacity [130].
In melanoma cancer stem-like cells (MCSLCs), DAPT (10 µM) treatment increased the expression of E-cadherin and inhibited the expression of VE-cadherin and Twist1, demonstrating that DAPT plays an important role in inhibiting melanoma metastasis [131].

Osteosarcoma
Yu et al. [132] investigated the role of Notch inhibition in the elimination of osteosarcoma stem cells (OSCs). OSCs were treated with 20 µM GSI (DAPT and RO4929097) for 24 h. GSI pretreatment suppressed the spheroid-forming ability and blocked the activity of cisplatin-enriched OSCs. The administration of GSI (DAPT, 10 mg/kg/d) for 14 days in mice bearing chemoresistant xenograft tumors reduced the sarcosphere-forming ability and suppressed the recurrence of the tumor. GSI treatment also downregulated the expression of stem-like cell markers.
In another study, different osteosarcoma cell lines were exposed to GSI (DAPT, 0-50 µM), alone or in combination with cisplatin, for 8-72 h. Moreover, nude mice were treated with 8 and 10 mg/kg DAPT, intraperitoneally (i.p.), alone or combined with cisplatin (CDDP), for 5 weeks. The study suggested that GSI treatment increased the anticancer effect of CDDP in resistant OSCs through the inhibition of proliferation, reduction in motility, induction of apoptosis, and cell-cycle arrest. In addition, GSI treatment reduced the number of OSCs and enhanced the platinum sensitivity of the tumor. It was also shown that the cotreatment of GSI and CDDP suppressed phosphorylated extracellular-regulated kinase (ERK) and Akt, and thus enhanced the antitumor effects. In mice, combination therapy exhibited a greater effect on suppressing the CDDP-resistant tumor xenograft metastasis and growth, compared with the compound alone. Taken together, the GSI and CDDP combination sensitized CDDP-resistant human osteosarcoma cell lines to CDDP through Notch signaling downregulation [133].

Ovarian Cancer
Vathipadiekal et al. [134] investigated the effect of DAPT on the elimination of ovarian cancer side population cells. The SKOV3 SP and MP cells were cultured with 10 and 20 µg DAPT, for 8 days. DAPT treatment repressed the colony-formation and survival of ovarian cancer side population cells. This side population had a concentration-dependent sensitivity to DAPT treatment.
In another study, primary ovarian tumor cells were exposed to GSI-I (1-10 µM), alone or in combination with cisplatin, for 1-2 days. GSI-I decreased the CSC population and enhanced the tumor platinum-sensitivity. The knockdown of Notch3 using small interfering RNAincreased the platinum therapy response, demonstrating that tumor chemo-sensitivity modulation by GSI-I is Notch signaling specific. In addition, the study demonstrated that the combined treatment of DAPT and cisplatin had a synergistic antitumor effect in Notch-dependent cancer cells through increasing the G(2)/M cell cycle arrest, apoptosis, and DNA-damage response [135].
The treatment of ovarian cancer stem-like cells with DAPT (1-20 µM, 1-3 days) showed that the Notch blockade with DAPT significantly hampered ovarian CSC proliferation and the self-renewal ability, reduced the ovarian cancer stem cell (OCSC)-specific surface marker expression, and decreased the mRNA and protein expression of Sox2 and Oct4 in OCSCs [136].
Recently, the effect of Notch3 signaling pathway suppression was investigated in NR2F6-overexpressing epithelial ovarian cancer (EOC) in vitro and in vivo. RO4929097 (10 µM treatment) enhanced the stem cell phenotype and increased the CDDP resistance in EOC cells by inducing Notch3 activation. RO4929097 treatment also inhibited ovarian cancer cell proliferation and increased apoptosis [137].

Pancreatic Carcinoma
In pancreatic cancer cells, DAPT-induced Notch downregulation led to a reduction in the number of CD133 + , the inhibition of cell proliferation, and abrogation of the DLL4/Notch-induced chemo-resistance [138], and also resulted in apoptosis, the inhibition of EMT, and the suppression of tumorigenesis by eliminating pancreatic cancer-initiating cells (CD44 + /EpCAM+), both in vitro and vivo [139].
In four different studies, the administration of 1-80 µM DAPT, alone or in combination with leptin, gemcitabine (GEM), and 5-FU, inhibited the Notch pathway, reduced the proliferation of MiaPaCa and BxPC-3 cells, decreased the number of leptin-induced CD133 + and ALDH + cells, and inhibited tumorsphere formation [140]. In addition, the number of CM and insulinomas (INS) CSC-enriched spheres decreased, whereas the INS CSC-like cells' sensitivity to 5-FU improved and the clonogenicity and tumorigenicity of INS CSClike cells were decreased [141]. The proportion of CD24 + CD44 + cells and pAkt, Hes1, and β-catenin expression in cell lines treated with gemcitabine declined, and the invasion and migration abilities were reduced [142].
It was reported that the in vitro and in vivo treatment of cells with different GSIs (MK-0752 and RO4929097), with or without gemcitabine, led to a decrease in the number of CSCs and the inhibition of tumorsphere formation and blocked tumor growth in NOD/SCID mice. Treatment with MK-0752 or RO4929097 combined with gemcitabine displayed the highest percentage of apoptosis compared with either compound alone [144].
MRK-003 with or without gemcitabine downregulated the intracellular domain of the Notch protein (NICD), eliminated the CSCs, and inhibited the colony-forming capacity in pancreatic cancer cells. The combination of gemcitabine and MRK-003 enhanced the antitumor effects, reduced tumor cell proliferation, and led to the induction of intratumoral necrosis and apoptosis [145].
Treating mice bearing pancreatic cancer xenografts with PF-03084014 (150 mg/kg), alone or combined with gemcitabine, downregulated NICD Hey-1 and Hes-1, resulted in tumor regression, and reduced cancer stem cells. The results also showed that combination therapy with gemcitabine and PF-03084014 was effective in apoptosis induction, the suppression of angiogenesis, and tumor cell proliferation, leading to a reduction of primary tumor growth, as well as controlling metastatic dissemination, compared to treatment with gemcitabine alone [146].

Prostate Cancer
In a recently published study, Wang et al. [147] investigated the antitumor effect of PF-03084014 on prostate cancer. Prostatic cancer stem cells were exposed to 1-100 µM PF-03084014 in combination with oncolytic herpes simplex virus for 6 days. The results of this study showed a reduction in the cancer cell population.
It was shown that DAPT (1 nM-400 µM) treatment reduced the protein levels of NICD1 and blocked γ-secretase activity in prostate cancer cells [148].
In castration-resistant prostate cancer (CRPC), PF-03084014 (0.1-10 µM) treatment, with or without docetaxel, resulted in a greater decrease in the tumor growth of both docetaxel-resistant and docetaxel-sensitive CRPC compared with either compound alone. The results also showed that this treatment increased the anticancer activity and apoptosis, decreased the epithelial to mesenchymal transition, reduced survival signals (EGFR, cyclin E, NF-κB, and PI3K/Akt pathway; Bcl-xL; and Bcl-2), decreased the density of microvessels, and eliminated CSCs in the tumor [149].
Various studies, as presented in this review, have demonstrated the anti-CSC effect of GSIs. GSIs have different chemical structures, but despite these differences, all of them can inhibit Notch signaling, making them excellent candidates for potential anti-CSC agents. The anti-CSC effect of GSIs has been tested in various cancer cell lines and animal models. Various GSIs, with different concentrations, doses, and treatment durations, have been used in the reviewed studies. The elimination of CSCs using GSIs depends on the GSI type, cell line used, GSI concentration, and other anticancer agents used for combination treatment. In vitro anti-CSC activity and in vivo tumor growth inhibition accompanied by longer survival provides the proof of the efficacy of GSIs. All reviewed studies indicate consistency of their results and confirm Notch inhibition by using GSIs.
The reviewed studies showed that increased concentrations of GSIs reduce the CSC growth rate and their anti-CSC effects are time-and concentration-dependent [36,54,104,122,134]. Higher concentrations of GSIs have a higher risk of toxicity. Therefore, a balance between toxicity and enhancement of the concentration should be set. One study showed that the combination of GSI with glucose-functionalized nanoparticles increased the anti-CSC effect and, at the same time, required a lower concentration to effectively reduce the CSCs in breast cancer [82]. If GSI-loaded nanoparticles are selectively localized in tumor cells, cancerous tissue can receive a higher dose in comparison to normal tissue. Therefore, the toxicity and side effects of GSI could be reduced.

Conclusions
There is evidence that CSCs induce tumor perpetuation, even after effective therapies, and result in aggression of the tumor. Given that Notch signaling is implicated in regulation of the cell fate, the aberrant activation of this pathway can lead to tumorigenesis. Therefore, the suppression of this pathway could be a potential therapeutic target for cancer management. This is the first study that provides a systematic review of the literature on the mechanisms of action of GSIs against various CSCs, although there are many other studies on the effect of GSIs on cancer cells rather than CSCs. Various studies have revealed that GSIs have potent anti-CSC effects and can inhibit the neoplastic activities, angiogenesis, and tumor growth of cancer cells, especially when combined with chemotherapeutic or targeted therapy drugs. Since Notch signaling has a critical role in the self-renewal and maintenance of CSCs, GSIs have a greater effect on CSCs in comparison to cancer cells. Therefore, the eradication of CSCs by GSIs could lead to higher survival rates.
The potential ability of GSIs to target CSCs is mediated through the inhibition of various stemness-related signaling pathways and transcription factors. Of note, the importance of the other cellular events and signaling pathway interactions that contribute to tumor progression should be considered when it comes to designing a therapeutic plan that involves GSIs. Apart from their advantages, GSIs cause adverse severe side effects. For some GSIs, a high potency and optimal physical properties have been achieved, but the biological mechanism imposes inbuilt Notch-related side effects. Therefore, they are dose-limiting and need moderate, intermittent administration. Consequently, the effective inhibition of Notch pathway activity requires more targeted delivery approaches and the effective delivery of GSIs to CSCs to target Notch signaling in this population. One of the strategies for overcoming these challenges is using the specific properties of CSCs in the design of nanoparticles containing GSIs to improve the outcome of targeted treatment. Another strategy could be the use of small molecules, such as CB-103, which might target the Notch transcription complex to downregulate Notch signaling. In this systematic review, we conclude that inhibiting Notch signaling by GSIs is a promising strategy for achieving efficient cancer therapy.