NOTCH receptors in gastric and other gastrointestinal cancers: oncogenes or tumor suppressors?

The abnormal richness of NOTCH1-4 mRNA was found to be associated with unfavorable overall survival for 876 GC patients for 20 years [48]. In GC, activated NOTCH1 was a poor prognostic factor for patients [49]. Also, increased NICD1 was observed in tumor dedifferentiation, depth of tumor invasion, lymph node metastasis, surface morphology and Lauren classification [50]. In GC cell lines, DLL1 expression is epigenetically regulated by promoter methylation although DLL1 activates Notch1 signaling pathway. Aberrant DLL1 promoter hypermethylation has been showed in 52% primary tumors in at least one region but not in healthy controls. Therefore, epigenetic regulation of the NOTCH ligand DLL1 only partly explained the activated NOTCH1 signaling in GC [51, 52]. NOTCH1 expressed in both premalignant and cancer tissues, especially in samples of intestinal metaplasia and well-differentiated intestinal type. It may be crucial in both promoting the metaplastic transition of gastric epithelial cells and maintaining a constant proliferation of internalized epithelial cells [53, 54]. Over-activated NOTCH1 was considered to prevent gastric carcinoma BGC-823 cells from TNFα-induced apoptosis [55]. Apart from faciliating GC progression via cyclooxygenase-2 (COX-2) [56], activation of the NOTCH1 signaling was also suggested to be related with metastasis of human malignancies [57]. Moreover, over-expressed NOTCH1 enhanced interaction between nuclear STAT3 and Twist promoter and activated NOTCH1/STAT3/TWIST signaling axis, further to promote GC progression [58]. Meanwhile, NOTCH1 silencing reduced proliferation and invasion in SGC-7901 GC cells [59]. A group of scientists pointed out that NOTCH1 acted as a significant part in the maintenance of the cancer stem-like phenotype of diffuse type GC through a RBP-Jƙ binding motif in the 5′ promoter region of CD133 gene. They also suggested NOTCH1 inhibition might serve as an effective therapy against CD133-positive diffuse type GC [60]. Some reports also suggested NOTCH1 regulatory mechanisms by some tumor-suppressive miRNAs, such as miR-34 family [61], miR-124 and miR-935 [58, 62]. All these miRNAs have been proved to repress NOTCH1 expression during GC progression. Meanwhile, NOTCH1 pathway, together with miR-151-5p, interplayed with p53 to form a reciprocal regulation loop in controlling gastric carcinogenesis [63]. However, there was a paper reported anti-tumor role of NOTCH1 in GC. Zhou W et al. demonstrated that NOTCH1 was absent or minimally expressed in GC tissues but abundant in paired normal gastric mucosa. Sequentially, they highlighted a novel AKT1/NF-ƙB/NOTCH1/PTEN axis as a key mechanism of chemoresistance in GC [64]. In addition, the active intracellular domain of NOTCH2 binds with COX-2 promoter region and induces COX-2 expression [65]. These findings implied that NOTCH1 and NOTCH2 boosted GC carcinogenesis through up-regulating COX-2. High NOTCH2 expression was identified as a prognostic parameter, as it was correlated with poor survival in GC patients [66]. Recently, there were results also revealing the NOTCH2 regulation by miR-23b in GC [67]. Contradictorily, tumor suppressive role of NOTCH2 was also reported in one publication. The authors stated that NOTCH2 decreased cell invasion through the PI3K/AKT pathway in MKN45 cells [68]. NOTCH3 profusion was found in the intestinal type of GC, with a better histological differentiation, indicating its role as a favorable prognostic indicator [69]. NOTCH4 activation promoted GC growth through the overexpression of Wnt1, β-catenin and downstream target genes, c-Myc and cyclin D1 [70].

Moreover, a study suggested that restraint of NOTCH receptors by two γ-secretase inhibitors (GSIs) suppressed cell proliferation and induced cell apoptosis. DAPT, a γ-secretase inhibitor, diminished GC growth, invasion, metastasis and epithelial-mesenchymal transition (EMT) through NOTCH1 pathway [71]. Additionally, combined treatment with both GSIs and chemotherapeutic agents significantly minimized the orthotopically transplanted gastric tumors in mice [72]. As for the therapeutic strategies, Hyun-Woo Lee et al. pointed out that targeting Notch signaling by GSIs enhanced the cytotoxic effect of 5-FU in GC [73]. There was also another project indicated that the IL-6/STAT3/JAG1/NOTCH axis might be a target for improving the efficacy of trastuzumab in GC treatment [74]. To better under the Notch signaling in gastric tumorigenesis, we summarized the main reports in a schematic presentation (Fig. 1).

To further address the Notch signaling cascade in GC with more updated information, we summarized the genetic alteration rates including copy number changes (amplification and deep deletion), somatic mutations and mRNA upregulation of NOTCH1-4 in the TCGA cohort (Fig. 2) [75, 76]. From the TCGA cohort analyzed by cBioPortal, NOTCH2 and NOTCH3 have the highest mutation rate (7% respectively) and mutations include truncating mutation, missense driver mutation and missense passenger mutation. The genomic amplification was merely one of the multiple reasons for high NOTCH1-4 mRNA expression in some GC cases, so we checked the promoter methylation status of NOTCH1-4 in GC. We found the promoter methylation level of NOTCH2, but not NOTCH1, is negatively correlated with its mRNA expression with significance (P < 0.001, Additional file 1: Figure S1). As H. pylori and EBV infection are the main risk factors for GC, we then checked the expression of NOTCH1-4 with the H. pylori or EBV infection status. However, we did not achieve any positive correlation between the expression of NOTCH1-4 with H. pylori infection (Additional file 2: Figure S2) or EBV infection (Additional file 3: Figure S3).

In addition to GC, aberrant Notch pathway has been linked to liver malignancies. Notch cascade was activated in human HCC samples and promoted hepatic carcinogenesis in mice as previous research showed [77]. To address which NOTCH member took the key position during the progress of liver cancer, Huntzicker et al. used antibodies to specifically target NOTCH1-3 and JAG1 respectively in xenograft mouse model of primary HCC driven by AKT and N-RAS [78]. They found that different NOTCH receptors had drastically different functions during HCC development and inhibition of NOTCH2 represented the most significant therapeutic option in the treatment. Moreover, Wang et al. also proposed a non-proteasome mediated feedback loop between NOTCH1 and Wnt/β-catenin signaling in activating liver cancer stem cells [79]. NOTCH1 activation contributed to tumor cell growth and proliferation while NOTCH1 down-regulation inhibited the invasion and migration by inactivating the Cox-2/Snail/E-cadherin pathway or through regulation of PTEN and FAK [80‐82]. However, NICD1 was also demonstrated as a tumor suppressor gene in HCC. It induced the degradation of Snail protein by ubiquitination and inhibited Snail-induced cell invasion [83]. During tumor progression, Notch signaling exerted a tumor-suppressive role through feedback loop in response to E2F transcription factors activation in Rb-family-triple-knockout liver cells [84]. Also, Pofut1 overexpression accelerated the cell proliferation and migration in HCC through the activation of Notch pathway [85]. The profusion of NOTCH1 might predict poor survival and more aggressive behavior in patients with HCC [86, 87]. Both NOTCH1 and NOTCH4 were immunohistochemical biomarkers predicting HCC patients with short disease specific survival [88]. There was a report suggesting that NOTCH1 functions in endothelial progenitor cells to initiate tumor vasculogenesis in HCC [89]. Activated NOTCH1 expression was strongly associated with HCC metastatic through NOTCH1-Snail-E-cadherin pathway [90]. NOTCH1 and ROS-induced PI3K/AKT pathways cooperatively increased Snail expression and promoted malignancy in HCC [91]. On the other hand, downregulation of Notch signaling activity inhibited HCC invasion by inactivation of matrix metalloproteinase-2 and -9 (MMP-2, -9) and vascular endothelial growth factor (VEGF) [92]. Defective NOTCH signaling led to impaired ability of repairing liver damage [93]. The histone deacetylases inhibitor vaproic acid induced cell growth arrest in HCC via suppressing NOTCH1 and its downstream gene HES1 [94]. NOTCH1 downregulation suppressed the expression of endothelial markers and impaired tube formation. Constitutive NOTCH2 signaling activation played an oncogenic role and induced hepatic tumor formation in mice [95]. Abundant NOTCH2 expression was correlated with anaplasia in human HCC cell lines. The NOTCH2 signaling conferred aggressive behavior and immature morphology to human HCC cells [96]. NOTCH2 was activated in liver cancer stem cells (CSCs) and its activation levels were consistent with clinical severity and prognosis of HCC patients. C8orf4, which attenuated the self-renewal capacity of liver CSCs and tumor propagation, negatively regulated self-renewal of CSCs through suppression of NOTCH2 signaling [97]. Again, self-renewal deficiency and cell growth reduction after NOTCH2 depletion indicated its oncogenic role in HCC [98]. NOTCH3 expression exhibited positive correlation with more aggressive traits and shorter survival in HCC [99]. And it participated in modulating the stemness of tumor cells via inactivation of Wnt/β-catenin pathway [100]. The abundance of NICD3, a symbol of constitutively activated NOTCH signaling, was the only detectable NOTCH3 subunit in HepG2 [101]. Interestingly, despite high NOTCH3 expression, HepG2 showed low NOTCH4 expression [102]. More importantly, NOTCH3 inhibition enhanced the effect of sorafenib by overcoming drug resistance [103]. Limited data manifested that NOTCH4 overexpression might be an independent predictor of short disease specific survival in HCC [88, 102].

NOTCH1-3 were reported to be oncogenic and highly expressed in human colon adenocarcinomas [104]. Enhanced NOTCH1 was correlated with progression, tumor grade and metastasis resulting from apoptosis inhibition [105, 106]. It also positively regulated the proliferation, colony formation, cell cycling, and tumorsphere formation of human colon cancers [106]. The elevated copy number gain of NOTCH1 together with its mRNA overexpression made it an independent predictor of prognosis in CRC [107, 108]. In colorectal carcinoma cells, NOTCH1-deppendent activation of cell cycle and proliferation were mediated by repression of cyclin-dependent kinase inhibitor p27. Retroviral-transduction activated NOTCH1 resulted in increased expression of stemness related proteins [109]. Moreover, NOTCH1 downregulation significantly sensitized CRC cells to chemotherapy and ionizing radiation [110]. The subcellular localization of β-catenin converged with NICD1 to induce proliferation through the activation of cyclin D1 and HES1 [111]. However, there was a paper uncovered an unexpected suppressive role of NOTCH1 on WNT/β-catenin targeted genes in CRC. Activation of NOTCH1 converted high-grade adenoma into low-grade adenoma in an APC ^min mouse colon cancer model [112]. NOTCH2 expression was decreased in CRC and was associated with tumor differentiation [113]. Negative correlation between NOTCH1 and NOTCH2 was identified in CRC. Increased NOTCH1 expression or decreased NOTCH2 expression represent a risk factor for poor overall survival of CRC patients [114, 115]. NOTCH3 was remarkably up-regulated and promoted tumorigenesis in CRC [116]. Its nuclear expression was related with tumor recurrence and might serve as a novel predictive marker in recurrent CRC patients of stage II and III [117]. Activated NOTCH3 increased MSI-1 level, which was a well established stem cell marker in CRC cells [118]. The miR-1-NOTCH3-Asef pathway was also crucial for CRC cell migration. In this axis, NOTCH3 up-regulated Asef expression and Asef activation was required for colorectal tumorigenesis [119]. miR-206 was another miRNA that potentially regulated NOTCH3 expression in CRC. This miRNA attenuated tumor proliferation and migration through downregulation of NOTCH3 [120].

There are two major pathological subtypes of esophageal carcinoma (EC): esophageal adenocarcinomas (EACs) and esophageal squamous cell carcinomas (ESCCs). EACs is considered to arise from a clonal stem like population of cells, in which NOTCH signaling cascade was closely involved [121]. This pathway promoted cell growth and maintained stemness of EAC cells [122]. Moreover, inhibition of NOTCH activity by GSIs decreased tumor growth using patient derived xenograft models [123]. The function of NOTCH1 in ESCC was firstly identified as a tumor suppressor. Activated NOTCH1 inhibited cell proliferation and induced apoptosis in EC9706 [124]. Subsequently, the oncogenic function of NOTCH1 in ESCCs was demonstrated by different groups. NOTCH1 expression was associated with cell aggressiveness and 5-FU drug resistance in ESCC patients [125]. NOTCH1 increased invasion and metastasis of ESCC cell line EC-9706 through EMT transducted by Snail [126]. High NOTCH1 protein expression was related to poor survival in ESCC [127]. Mutually exclusive mutations in NOTCH1 and PIK3CA were identified in ESCCs from Chinese patients with genetic analysis. Mutation in NOTCH1 was related to well-differentiated, early-stage malignancy and less metastasis to regional lymph nodes [19]. NOTCH3 contributed to esophageal cell fate decisions by promoting squamous cell differentiation while preventing dedifferentiation to mesenchymal cell lineages expressing ZEBs, through which inhibition of NOTCH pathway promoted TGF-β-mediated EMT [128].

In human samples, Notch pathway components were highly expressed in pancreatic adenocarcinoma. Targeting Notch signaling pathway by natural agents eliminates pancreatic CSCs, which suggested a treatment of patients with PC [129]. Ectopic NOTCH activation induced accumulation of nestin-positive precursor cells and expansion of metaplastic ductal epithelium, which was identified as precursor lesion for PC [130]. Moreover, some reports suggested Notch mediated tumor-initiating effects by expanding undifferentiated precursor cells through TGFα. Inhibition of Notch pathway by GSIs reduced PC cell growth. In invasive PCs, NOTCH1 and downstream targets such as HES1 were up-regulated in lesions varying from tubular complexes to carcinoma [131]. Moreover, a profusion of NOTCH1 resulted in induction of EMT phenotype [132]. In PC cells, the tyrosine kinase c-Src directly mediated NOTCH1 and Furin interaction, which regulated carcinogenesis and cancer cell growth [133]. Downregulation of NOTCH1 inhibited proliferation, increased apoptosis, reduced cell migration and invasion of PC cells [134]. However, NOTCH1 exerted tumor suppressor function in a model of K-RAS-induced pancreatic ductal adenocarcinoma [135]. NOTCH2 was highly expressed in ductal cells and pancreatic intraepithelial neoplasia lesions (PanIN) using genetically engineered mice. Conditional ablation of NOTCH2 slowed down PanIN progression and delayed survival time through Myc signaling inhibition [136]. Nuclear accumulation of NOTCH3 was observed in pancreatic adenocarcinomas, which was associated with adverse clinical features and correlated with STAT3 overexpression and phosphorylated AKT [137]. Suppression of NOTCH3 inhibited cell proliferation and decreased PI3K/AKT activity in PC [138].