The Role of Notch3 Signaling in Cancer Stemness and Chemoresistance: Molecular Mechanisms and Targeting Strategies

Aberrant Notch signaling profoundly affects cancer progression. Especially the Notch3 receptor was found to be dysregulated in cancer, where its expression is correlated with worse clinicopathological features and poor prognosis. The activation of Notch3 signaling is closely related to the activation of cancer stem cells (CSCs), a small subpopulation in cancer that is responsible for cancer progression. In addition, Notch3 signaling also contributes to tumor chemoresistance against several drugs, including doxorubicin, platinum, taxane, epidermal growth factor receptor (EGFR)–tyrosine kinase inhibitors (TKIs) and gemcitabine, through complex mechanisms. In this review, we mainly focus on discussing the molecular mechanisms by which Notch3 modulates cancer stemness and chemoresistance, as well as other cancer behaviors including metastasis and angiogenesis. What’s more, we propose potential treatment strategies to block Notch3 signaling, such as non-coding RNAs, antibodies and antibody-drug conjugates, providing a comprehensive reference for research on precise targeted cancer therapy.


INTRODUCTION
Notch signaling is a highly conserved among multicellular organisms, and it is involved in cell fate decision, cell proliferation/differentiation, as well as cell lineage specification (Bray, 2006;Bray, 2016). The activation of Notch signaling is mediated by cell-to-cell interactions with a Notch ligand. In mammals, there are four Notch receptors (Notch1-4) and five ligands [Jagged (JAG)1, 2 and Delta-like ligand (DLL)1, 3 and 4]. Before it is trafficked to the cell membrane, the full-length Notch receptor undergoes initial cleavage, also called S1 cleavage, in the Golgi apparatus. When a Notch ligand (JAG or DLL) in the cell membrane of an adjacent signal-sending cell interacts with a Notch receptor in the cell membrane of the signal-receiving cell, the Notch receptor is activated and undergoes another proteolytic cleavage. This socalled S2 and S3 cleavage steps are induced by A Disintegrin And Metalloprotease domain 10 (ADAM10) and the γ-secretase complex, respectively (Bray, 2006;Kopan and Ilagan, 2009;Bray, 2016). Subsequently, the Notch intracellular domain (NICD) is released and translocated into the nucleus, where it binds to the effector DNA-binding transcription factor CSL. The latter then recruits the transcription co-activator mastermind-like protein (MAML) to induce the transcription of downstream target genes. Finally, the Notch receptor or NICD undergoes proteasomal/lysosome degradation (Bray, 2006;Kopan and Ilagan, 2009;Bray, 2016;Xiu and Liu, 2019) (Figure 1, shown on the example of Notch3 signaling).
Notch3 overexpression in cancer is mainly caused by alterations of the Notch3 gene. According to TCGA, the Notch3 gene was altered in 5% of cancer samples, mainly via amplification and mutation ( Figure 3). Notch3 has been reported to be amplified in 10-25% of ovarian carcinoma (OC) (Park et al., 2006;Etemadmoghadam et al., 2009;Cancer Genome Atlas Research Network, 2011;Hu et al., 2014). Among all cancer types in the TCGA database, OC has the highest Notch3 amplification rate (11.64%, 68 of 584 cases) (Figure 3). In addition to amplification, mutations in the negative regulatory region (NRR) and proline (P), glutamic acid (E), serine (S), threonine (T)-rich (PEST) domains of Notch3 gene can cause Notch3 activation (gain-of-function/activating mutations), as seen in human T-cell acute lymphoblastic leukemia (T-ALL) (Bernasconi-Elias et al., 2016).
Notch signaling plays complex roles in regulating cellular behaviors during cancer progression, and each Notch receptor has its specific pattern (Majumder et al., 2021). A major role of Notch3 is maintaining the stemness of cancer stem cells (CSCs). As a population of self-renewing cells with high tumorigenic potency, CSCs are found to be activated by Notch3 signaling in several kinds of cancer and contribute to cancer progression through complex mechanisms (See Notch3 and Cancer Stem Cell Properties). Another main feature of Notch3 signaling is to induce FIGURE 1 | Notch3 signaling. (A) The structure of Notch3 and its ligand. The full-length Notch3 protein consists of 34 epidermal growth factor (EGF) repeats, a negative regulatory region (NRR) [composed of three Lin-Notch repeats (LNRs) and a heterodimerization domain (HD)], a transmembrane domain, an RBPJ-association module (RAM) domain, seven ankyrin (ANK) repeats, two nuclear localization signals (NLS), a transactivation domain (TAD) and a C-terminal domain rich in proline, glutamic acid, serine, and threonine (PEST domain) (Xiu and Liu, 2019). The Notch ligand consists of EGF repeats and a Delta/Serrate/LAG-2 (DSL) domain. (B) Notch3 signal transduction process.
There are numerous published review articles on the effects of Notch signaling in cancer treatment (Giovannini et al., 2016;Bellavia et al., 2018;Giuli et al., 2019;Katoh and Katoh, 2020). By contrast, this review mainly focuses on the underlying Notch3related molecular mechanisms that regulate cancer stemness and chemoresistance. In addition, the relationships between Notch3 and other tumor biological characteristics, including metastasis and angiogenesis are also discussed. Finally, we summarize known Notch3-targeting strategies/methods for cancer therapy.
Overall, this review provides comprehensive information on the role of Notch3 signaling in cancer and its value as a therapeutic target.

NOTCH3 AND CANCER STEM CELL PROPERTIES
Tumor initiation and progression is driven by a small population of cancer cells with self-renewal and tumor-formation capacity, known as CSCs (Dawood et al., 2014). The activation of Notch3 signaling is widely found in CSCs, where it regulates their abundance and activity through several molecular mechanisms ( Figure 4). The expression of aldehyde dehydrogenase (ALDH), a recognized CSC marker, is significantly positively correlated with Notch3 expression, as seen in OC, lung carcinoma (LC), hepatocellular carcinoma (HCC) and breast carcinoma (BC) (Sullivan et al., 2010;Xiao et al., 2011;Zhang et al., 2015;Kim et al., 2017a). Suppression of Notch3 signaling in LC cells by treatment with either a γ-secretase inhibitor (GSI) or short hairpin RNA (shRNA) against Notch3 resulted in a significant decrease of ALDH + CSCs, indicating that Notch3 is critical for ALDH expression (Sullivan et al., 2010).
In OC, the zinc transporter ZIP4 was identified as a novel CSC marker that physically interacts with Notch3 and activates Notch3 signaling (Fan et al., 2020). Several studies also found that the activation of Notch3 signaling enhances CSC activity, especially in chemoresistant OC tumors (Kim et al., 2017b;Jeong et al., 2017;Fang et al., 2020), and the relevant mechanisms are discussed in Platinum and Taxane. In HCC, the activation of Notch3 signaling was found to inhibit Wnt/β-catenin signaling and increase the expression of the stemness-related protein Nanog, which promotes the maintenance of the CSC population, thereby contributing to the pathogenesis of HCC . In addition, Notch3 signaling in liver CSCs is supported by cancerassociated fibroblasts in the tumor microenvironment and maintains tumor cell self-renewal (Liu et al., 2018). Mechanistically, Notch3 signaling activates LSD1, a histonemodifying enzyme that promotes cancer stemness, by inducing its deacetylation by activating the class-III histone deacetylase (HDAC) SIRT1 (Liu et al., 2018).
In the tumor microenvironment of BC, a secreted protein named Cartilage Oligomeric Matrix Protein (COMP) physically bridges Notch3 and JAG1 on the cell membrane of CSCs, thus driving JAG1/Notch3 signaling and subsequently activating the β-catenin and Akt signaling pathways to maintain CSC status (Papadakos et al., 2019). A recent study showed that Notch3 signaling contributes to the overexpression of the T-cell inhibitory molecule PD-L1 in breast CSCs by activating mTOR signaling (Mansour et al., 2020). Specific knockdown of Notch3 can downregulate PD-L1 expression on CSCs and reduce CSC activity, providing a novel strategy for anti-PD-L1 combination therapies (Mansour et al., 2020). In addition to the mechanisms that promote CSC activity, Notch3 signaling is also found to reduce the population of breast CSCs by negatively regulating IL6 (Wang et al., 2018a). Furthermore, the activation of HIF1α in response to hypoxia is involved in Notch3-mediated IL6 inhibition in breast CSCs via direct binding to the Notch3 promoter. The combination of Notch and IL6 inhibitors significantly decreases the abundance of breast CSCs and inhibits BC growth, suggesting it might serve as a novel therapeutic strategy for treating Notch3-expressing BC (Wang et al., 2018a).

NOTCH3 AND DRUG RESISTANCE
A large number of studies have shown that Notch3 signaling is closely related to the ability of tumors to chemotherapy. Here, we mainly introduce the roles of Notch3 in the resistance of tumors to five kinds of chemotherapeutic drugs (doxorubicin, platinum, taxane, EGFR-TKIs and gemcitabine), whose mechanisms are comparatively well-understood ( Figure 5).

Doxorubicin
The resistance of tumor cells to doxorubicin, a DNA topoisomerase II inhibitor, is associated with the activation of Notch3 signaling (Giovannini et al., 2009;Michishita et al., 2011;Wang et al., 2018b). In HCC cells, Notch3 signaling contributes to doxorubicin resistance by inhibiting p53 expression, doxorubicin uptake and DNA damage, which can be reversed by Notch3 depletion (Giovannini et al., 2009). In osteosarcoma, both in vivo and in vitro experiments indicate that the lncRNA SOX2OT variant seven can activate DLL3/Notch3 signaling, maintaining the stemness and doxorubicin-resistance of tumor cells (Wang et al., 2018b). Treatment of osteosarcoma cells with epigallocatechin gallate, a polyphenol from green tea, can counteract the SOX2OT-7/DLL3/Notch3 axis, thus inhibiting cancer progression (Wang et al., 2018b).

Platinum and Taxane
Notch3 signaling plays a critical role in tumor resistance to platinum, taxane, or their combination chemotherapy regimens, especially in OC. In a clinical study of 61 OC patients, the high expression of Notch3 was correlated with shorter PFS and OS in patients with stage III and IV disease treated with a standard platinum and taxane chemotherapy regimen (Rahman et al., 2012). In vitro experiments in OC cell lines also confirmed that Notch3 was significantly overexpressed in cisplatin-resistant A2780cis cells (2.5-fold) and paclitaxel (PTX)-resistant SKpac cells (25.5-fold) compared to chemo-sensitive A2780 cells (Kim et al., 2017a).
In cisplatin-resistant OC, Notch3 signaling was found to be induced by the activation of the c-Kit/prohibitin axis and the nuclear orphan receptor NR2F6, as well as P53 mutations Xu et al., 2019;Fang et al., 2020). The activation of Notch3 signaling activates the stem cell reprogramming factor PBX1, which supports the activity of CSCs contributing to platinum chemoresistance (Fang et al., 2020). In addition, Notch3 signaling also upregulates the expression of Cyclin G1 and SUSD2, which contributes to both tumor metastasis and cisplatin resistance of OC Xu et al., 2019). As seen in OC, Epstein-Barr virus (EBV)-associated nasopharyngeal carcinoma and colorectal carcinoma (CRC), inhibition of Notch3 signaling can significantly enhance the cisplatin chemosensitivity of tumor cells, indicating the value of Notch3-targeted therapy (Man et al., 2012;McAuliffe et al., 2012;Tzeng et al., 2014;Xie et al., 2016).

EGFR-TKIs
Studies have revealed novel mechanisms by which Notch3 induces EGFR-TKI resistance in EGFR-mutated tumors. Notch3 receptor was identified as a substrate for EGFRmediated tyrosine phosphorylation, and EGFR kinase activity induces tyrosine phosphorylation of Notch3, thus inhibiting Notch3 signaling (Arasada et al., 2014). Unfortunately, EGFR-TKI therapy relieves this inhibition, resulting in Notch3 activation and subsequent CSC enrichment (Arasada et al., 2014). In response to EGFR-TKI therapy of non-small-cell lung carcinoma (NSCLC), Notch3 physically binds to β-catenin in the cytoplasm of tumor cells to activate β-catenin signaling (Arasada et al., 2018). The combination of EGFR-TKIs and a β-catenin inhibitor abrogates the Notch3-dependent activation of β-catenin, which strongly attenuates tumor onset, improving the OS and RFS of NSCLC xenograft mice (Arasada et al., 2018). Zhang et al. found that recovering the expression of miR-150 can directly downregulate Notch3 in TKI-resistant NSCLC cell lines, providing another method for reversing Notch3-mediated TKI resistance .

Gemcitabine
In a clinical study of 71 PC patients, Notch3 was identified as a novel biomarker for predicting the efficacy of gemcitabine (GEM), whereby low Notch3 expression was associated with better GEM treatment efficacy and longer OS of PC patients (Eto et al., 2013). Mechanistically, Notch3 increases the activity of PI3K/Akt signaling in PC cells in response to GEM treatment, and this effect can be reversed by Notch3-specific siRNAs (Yao and Qian, 2010). In addition, Notch3 signaling also contributes to GEM resistance in NSCLC cells. Treatment with GEM and GSI significantly enhances GEM sensitivity and leads to tumor cell apoptosis, but the underlying molecular mechanisms remain unclear (Hu et al., 2018).

NOTCH3 IN OTHER ASPECTS OF CANCER BIOLOGY Notch3 in Cancer Epithelial-Mesenchymal Transition and Metastasis
Notch3 has a close relationship with tumor metastasis (Figure 6). In clinical studies, high expression of Notch3 was found to be associated with tumor metastasis in OC, NSCLC, prostate carcinoma (PCa), HCC, PC and gallbladder carcinoma (Ye et al., 2013;Zhou et al., 2013;Liu et al., 2016a;Liu et al., 2016b;Zhou et al., 2016;Kim et al., 2017a;Lin et al., 2018;Kim and Gu, 2019). Matrix metalloproteinases (MMPs) cascade with Notch3 signaling and promote tumor metastasis. The Notch3-MMP-3 axis contributes to bone metastasis of PC by promoting the formation of osteoblastic lesions and decreasing osteoblastogenesis (Ganguly et al., 2020). In HCC and pancreatic ductal adenocarcinoma (PDAC), Notch3 signaling activates the COX-2 and ERK1/2 pathways, which subsequently enhance the migration and invasion of tumor cells by upregulating the expression of MMP-2 and MMP-9 (Zhou et al., 2013;Zhou et al., 2016). In addition, MMP-14 and MMP-28 were found to promote tumor metastasis by inducing Notch3 signaling. The MMP-14-Notch3-β1-integrin axis can be activated by interactions between lymphatic endothelial cells and melanoma cells, leading to the transformation of nonmetastatic melanoma cells into invasively sprouting melanoma cells (Pekkonen et al., 2018). The MMP-28-Notch3 axis promotes the Epithelial-Mesenchymal Transition (EMT), migration and invasion of HCC cells in vivo and in vitro (Zhou et al., 2019).
In advanced CRC, Notch3 expression is positively correlated with lymph node as well as distant metastasis, and its expression is dependent on the activation of Akt signaling (Varga et al., 2020). Additionally, DLL4/Notch3 signaling was found to upregulate the expression of the RNA-binding protein MSI-1 in metastatic CRC tumors. Active MSI-1 can inhibit the expression of NUMB, a negative regulator of Notch signaling, which maintains the activation of oncogenic Notch1 and Notch3 signaling pathways (Pastò et al., 2014).
The role of Notch3 in BC metastasis is controversial. A study of 72 BC cases reported that Notch3 expression is correlated with a lower risk of lymph node metastasis, as well as the expression of estrogen receptor α (ERα), progesterone receptor (PR) and GATA3 . N3ICD in the nucleus of BC cells can bind to the promotors of ERα and GATA3 to promote their expression. ERα and GATA3 activated by Notch3 signaling upregulate vimentin expression and repress E-cadherin expression, which then suppresses the EMT and metastasis of BC by maintaining a luminal phenotype (Dou et al., 2017;Lin et al., 2018). Another Notch3-mediated EMT-suppression mechanism in BC relies on the activation of Hippo/YAP signaling by upregulating the transcription of KIBRA, an upstream factor of Hippo signaling (Zhang et al., 2016b). The inhibitor of DNA binding 2 (Id2), a transcription factor belonging to the bHLH family, can promote the transcription of Notch3, thus attenuating the EMT in BC (Wen et al., 2018). By contrast, microRNAs 221 and 222 were found to target the 3' UTR of Notch3 and suppress its protein translation in BC cells, which reverses EMT inhibition by Notch3 signaling .
Although the inhibitory effect of Notch3 on BC metastasis has been confirmed in several studies, a pro-EMT function of Notch3 has been also identified. It was found that the activation of Notch3 signaling is linked to BC seeding and lung/brain metastasis, while abrogation of Notch3 reduces the self-renewal and invasion ability of BC cells, restoring a luminal CD44 low /CD24 high / ERα high phenotype (Leontovich et al., 2018). Mechanistically, aberrant Aurora A kinase activity activates Notch3 in breast CSCs and contributes to metastatic growth (Leontovich et al., 2018). In addition, IL6 was found to activate Notch3 signaling in CD133 high /ERα low /IL6 high breast CSCs, where it promotes endocrine resistance and metastatic progression (Sansone et al., 2016). In bone-metastatic BC, the activation of JAG1/ Notch3 signaling induced by osteoblasts and osteoblast-derived TGF-β1 contributes to aggressive osteolytic metastasis and bone destruction in vivo (Zhang et al., 2010). These findings indicate that the relationship between Notch3 and BC metastasis should be explored further.

Notch3 and Tumor Angiogenesis
The functions of Notch signaling in tumor vasculature are mainly determined by Notch ligands. JAG1-mediated Notch signaling induces neovascularization and sprouting angiogenesis, while DLL4-mediated Notch signaling inhibits tumor angiogenesis (Xiu et al., 2020b;Xiu et al., 2020a). Notch3 is also involved in the regulation of tumor angiogenesis (Figure 7). Immunohistochemistry for Notch3 expression in 105 TNBC tissues showed that its expression is positively correlated with tumor microvascular density (MVD), which suggests a potential pro-angiogenic role of Notch3 (Xue et al., 2017). MUC4, a large membrane-anchored glycoprotein, can facilitate tumor angiogenesis and increase tumor MVD in PC by activating Notch3 signaling and downstream pro-angiogenic genes, including VEGF-A and ANG-2 (Tang et al., 2016). In addition, Notch3 signaling can be activated by interactions between tumor cells and cells in the tumor microenvironment, which contribute to tumor angiogenesis in several cancers. The interactions between tumor cells and cancer-associated fibroblasts activate Notch3 signaling, thus promoting angiogenesis in oral squamous cell carcinoma (Kayamori et al., Frontiers in Molecular Biosciences | www.frontiersin.org June 2021 | Volume 8 | Article 694141 8 2016). Similarly, the interactions between CSCs and endothelial cells (ECs) activate Notch3 signaling to promote angiogenesis in OC, which can be inhibited by Notch3-targeting miRNAs, including miR-136 and miR-150 (Kim et al., 2017b;Jeong et al., 2017). Additionally, the interactions between CSCs and ECs activate Notch3 signaling to promote angiogenesis and vasculogenic mimicry in melanoma (Hsu et al., 2017).
Notch3 signaling promotes tumor angiogenesis in a canonical CSL-dependent manner, which requires cell-cell interactions and is driven by Notch ligands. However, Notch3 was also found to act as a dependent receptor in tumor ECs to negatively regulate tumor angiogenesis, circumventing CSL . Mechanistically, Notch3 receptors on the surface of tumor ECs can directly activate caspase-9, which induces the caspasedependent cell death of ECs . Interestingly, overexpression of JAG1 ligand can abrogate this effect and promote tumor neovascularization. This indicates that the effects of Notch3 on tumor angiogenesis depend on the cellular context, such as the availability and amounts of Notch ligands in tumors .

NOTCH3-TARGETING STRATEGIES FOR CANCER THERAPY
GSIs that prevent the S3 cleavage of Notch receptor are the most commonly used therapeutic option for blocking Notch signaling in cancer (Shih Ie and Wang, 2007). However, they lack specificity and may interfere with the processing of other transmembrane proteins. What's more, pre-clinical studies have shown that the use of GSIs is associated with severe side effects, such as gastrointestinal tract toxicity (Milano et al., 2004;van Es et al., 2005). Therefore, it is essential to propose other potential methods/strategies that target Notch3 signaling.

Small Interfering RNAs and Short Hairpin RNAs
siRNAs and shRNAs are powerful molecules that can directly knock down the expression of target genes (Rao et al., 2009). In pre-clinical experiments, Notch3-specific siRNAs or shRNAs were able to prevent Notch3 activation and potently inhibit tumor cell growth in vivo and in vitro (Yao and Qian, 2010;Serafin et al., 2011;Hassan et al., 2016;Kang et al., 2016;Diluvio et al., 2018). However, the low efficiency of traditional siRNA/ shRNA delivery vehicles remains a significant obstacle for their clinical application. To overcome this, researchers developed a novel aptamer-siRNA chimera-based delivery system to inhibit Notch3 expression . The chimera consisting of an aptamer and Notch3-specific siRNA was bonded with cationic Au-Fe 3 O 4 nanoparticles (NPs). This chimera exhibited a high Notch3 silencing efficiency in OC cell lines, as well as potent antitumor effects . However, there are still no reports on the in vivo use of this Au-Fe 3 O 4 NP-chimera, and its potential value also needs evaluation in clinical studies.

Antibodies
Antibodies that target Notch receptors/ligands have been confirmed to effectively modulate Notch signaling activity (Xiu et al., 2020b;Gharaibeh et al., 2020). The monoclonal antibodies (mAbs) named A4, A8, MOR20350 and MOR20358, were designed to bind the NRR domain (Lin-Notch repeat (LNR) and heterodimerization domain (HD) domain) of Notch3 protein, which prevents the exposure of the S2 cleavage site and blocks Notch3 activation (Li et al., 2008;Tiyanont et al., 2013;Bernasconi-Elias et al., 2016). In T-ALL harboring Notch3 gain-of-function mutations, anti-Notch3 NRR mAbs show potent anti-leukemic activity in T-ALL cell lines and tumor xenografts (Bernasconi-Elias et al., 2016).  In addition to anti-Notch3 NRR mAbs, another mAb against epidermal growth factor (EGF) repeats of Notch2/3 named tarextumab (also called OMP-59R5) has been used to block Notch2/3 signaling in pre-and clinical studies (O'Reilly et al., 2015;Yen et al., 2015;Hu et al., 2019;Smith et al., 2019). Tarextumab was found to significantly inhibit the growth of PC, BC, OC and small-cell lung carcinoma (SCLC) xenograft tumors, partly by reducing the abundance of CSCs. Additionally, the combination of tarextumab with GEM plus nab-paclitaxel exhibited more potent anti-tumor effects (Yen et al., 2015). In the phase 1b clinical study NCT01647828, tarextumab in combination with gemcitabine plus nab-PTX was evaluated in 38 untreated metastatic PDAC patients, and the overall response rate (ORR) was 29% (O'Reilly et al., 2015). The recommended phase 2 dose was 15 mg/kg with standard doses of the cytotoxic agents. The frequent tarextumab-related emergent adverse events (TEAEs) were diarrhea (60%) and fatigue (43%), which were mostly grade 1 or 2 (O'Reilly et al., 2015). In another phase 1 study (NCT01277146) dose escalation and expansion of tarextumab was evaluated in 42 patients with solid tumors (Smith et al., 2019). Tarextumab was well tolerated at doses of 2.5 mg/kg weekly, as well as 7.5 mg/kg every 14 or 21 days. Diarrhea (81%) was the most common TEAE, followed by fatigue (48%), nausea (45%) and decreased appetite (38%) (Smith et al., 2019). Unfortunately, the results of a phase 2 study (NCT01859741) indicated that tarextumab treatment in combination with platinumbased therapy in 145 untreated SCLC patients did not improve PFS, OS, or ORR of patients (Daniel et al., 2017). In another phase 2 study (NCT01647828) of 177 untreated metastatic PDAC patients, tarextumab treatment in combination with GEM plus nabpaclitaxel also did not improve the OS, PFS, or ORR, while PFS was even statistically worse in tarextumab-treated patients . Due to the adverse effects of tarextumab shown in phase 2 clinical trials, its clinical development was discontinued.
As mentioned in EGFR-TKIs, co-blockage of EGFR and Notch receptors is necessary in some cases. In recent studies, bispecific mAbs targeting both Notch2/3 (tarextumab) and EGFR/HER3 (panitumumab/RG7116/MEHD7945A) have been established using the "Knobs into holes" and "CrossMAb" technologies (Hu et al., 2017;Fu et al., 2019). In vivo and in vitro experiments on NSCLC and TNBC showed that EGFR/Notchbispecific mAbs exhibit potent anti-tumor effects, especially decreasing the abundance of CSCs, which limits tumor resistance to EGFR-TKIs and has potential value for clinical applications (Hu et al., 2017;Fu et al., 2019).

Antibody-Drug Conjugates
Antibody-drug conjugates (ADCs) are mAbs conjugated to small-molecule chemotherapeutic agents via a chemical linker. ADCs can selectively bind to specific targets on the surface of cancer cells and directly deliver the ultra-toxic payload, thus killing cancer cells (Chau et al., 2019). PF06650808, a novel Notch3-targeting ADC, contains a humanized anti-Notch3 IgG1 antibody, a cleavable maleimidocapronic-valinecitrulinep-aminobenzylooxycarbonyl peptide linker, and an auristatinbased cytotoxic payload (Geles et al., 2015). Pre-clinical experiments revealed that PF06650808 can effectively inhibit the growth of TNBC, OC and NSCLC xenograft tumors (Geles et al., 2015). In a recent phase 1, dose-escalation study with 40 solid tumor patients, PF-06650808 was well tolerated at doses ≤2.0 mg/kg, and the maximum tolerated dose was 2.4 mg/kg (Rosen et al., 2020). The most common TEAEs were fatigue (40.0%), decreased appetite (37.5%), nausea (35.0%) and alopecia (32.5%). The ORR and clinical benefit response in the 31 response-evaluable patients was 9.7 and 35.5%, respectively (Rosen et al., 2020). However, the study has been terminated due to a change in sponsor prioritization.

Other Drugs/Compounds
Temozolomide (TMZ) is an alkylating chemotherapeutic agent that can penetrate the blood-brain barrier and is clinically used in the treatment of glioblastoma (GBM). One study demonstrated that the inhibition of Notch3 expression contributes to TMZinduced GBM cytotoxicity . Mechanistically, TMZ enhances the expression of the ER stress protein CHAC1 by activating JNK1/c-JUN signaling. Subsequently, CHAC1 binds to Notch3 protein, which reduces the generation of N3ICD, thus preventing Notch3 signaling . However, a recent study found that TMZ can also activate DLL4/Notch3 signaling to maintain CSC properties in GBM by upregulating MMP14 expression (Ulasov et al., 2020). Thus, the effect of TMZ on Notch3 signaling needs further exploration.
In HCC chemotherapy, a well-tolerated combination of sorafenib and valproic acid was found to synergistically inhibit tumor growth by downregulating Notch3 and p-Akt . Mangiferin, a C-glucosyl xanthone (1,3,6,7-tetrahydroxyxanthone-C2-β-D-glucoside), can specifically repress Notch3 signaling, which increases apoptosis and inhibits OC tumor growth both in vitro and in vivo (Zou et al., 2017). In spite of these success stories, more drugs/compounds that potentially target Notch3 should be screened in the future.

DISCUSSION AND CONCLUSION
Notch3 signaling plays critical roles in cancer progression, and the related molecular mechanisms have been studied in some detail. Stem cell-like properties are a primary feature of Notch3-positive cancer To maintain the stemness and proliferation of tumor cells, Notch3 signaling can activate the expression of downstream genes, such as cell cycle-related genes (CCND1, C-MYC and NF-kB1), antiapoptotic genes (SURVIVIN and BCL2), as well as stemnessrelated genes (OCT-4, ALDH1, NANOG, PBX1, CD44 and CD133) (Park et al., 2010;Man et al., 2012;McAuliffe et al., 2012;Alqudah et al., 2013;Jeong et al., 2017). In addition, there are several associations and cross-talk interactions between Notch3 signaling and other signaling pathways, mainly including the Wnt/ β-catenin, Akt, IL6, EGFR, TGF-β and NF-κB signaling pathways, which affect several aspects of cancer cell behavior (Table 3). In tumor metastasis, Notch3 signaling was found to cascade the MMP, Wnt, Akt, IL6 and TGF-β signaling pathways, thereby promoting the invasion and EMT of tumor cells (See Notch3 in Cancer EMT and Metastasis). However, anti-EMT properties and mechanisms of Notch3 signaling were also found in BC, suggesting a controversial role of Notch3 in BC metastasis (Zhang et al., 2016b;Dou et al., 2017;Liang et al., 2018;Lin et al., 2018;Wen et al., 2018). Moreover, Notch3 signaling activated by cell-cell interactions between tumor cells and tumor ECs can promote tumor angiogenesis and vasculogenic mimicry (See Notch3 and Tumor Angiogenesis). These findings suggest that Notch3 has diverse, complex and wide-ranging roles in tumor cells.
To prevent the abnormal activation of Notch3 signaling in cancer, key Notch3-targeting strategies have been proposed and confirmed effective in pre-clinical studies, including the application of siRNAs/shRNAs, ncRNAs, antibodies, ADCs, and HDACi (See Notch3-Targeting Strategies for Cancer Therapy). Different mAbs against Notch3 can specially block Notch3 signaling. However, phase 2 clinical trials of the anti-Notch2/3 mAb drug tarextumab have showed poor efficacy, and the relevant clinical trials have been terminated (Daniel et al., 2017;Hu et al., 2019). Recently, bispecific mAbs targeting both Notch2/3 and EGFR/HER3 have been developed. The main advantage of these bispecific mAbs is that they can target/ block both Notch3 and EGFR signaling, which reverses the activation of Notch3 signaling in response to EGFR-TKIs (Hu et al., 2017;Fu et al., 2019). The efficacy of bispecific mAbs may be worth testing in future clinical trials.
ADCs are novel drugs that can kill tumor cells which express specific target molecules. The first Notch3-targeting ADC drug PF06650808 has been evaluated in pre-clinical experiments and a phase 1 clinical trial. Although preliminary, the results demonstrate a manageable safety profile and early signs of anti-tumor activity in cancer patients (Geles et al., 2015;Rosen et al., 2020). Other Notch3-targeting strategies, such as siRNAs/ shRNAs, non-coding RNAs and HDACi, have not been tested in clinical trials, so that their efficacy and safety in the treatment of cancer patients need to be evaluated in the future.
In order to develop Notch3-targeting methods/drugs for cancer treatment, the following potential strategies should be considered: 1) Inhibiting Notch3 gene expression using siRNAs, shRNAs, or ncRNAs. 2) Preventing the cleavage of Notch3 protein using small molecules such as using ADAM10 inhibitors (prevent S2 cleavage) and GSIs (prevent S3 cleavage). 3) Antibodies targeting Notch3 protein. 4) Killing Notch3-positive tumor cells by ADCs. 5) Promoting Notch3 degradation by HDACi. Notably, Notch3-specific inhibitors such as antibodies and ADCs are more specific than pan-Notch inhibitors such as GSIs, which may merit further pre-and clinical evaluation. In summary, Notch3 signaling affects cancer progression through complex molecular mechanisms. Future studies should investigate the relevant mechanisms and exact roles of Notch3 signaling in regulating different cancer behaviors (such as CSC properties, Epithelial-Mesenchymal Transition (EMT), metastasis, drug resistance and angiogenesis) in different tumor types. Furthermore, it is necessary to propose, establish and evaluate more potential Notch3-targeting methods/strategies for cancer treatment.

AUTHOR CONTRIBUTIONS
MX wrote the manuscript. YW, BL, XW, FX, SC, LZ, BZ, and FH directed the project. All authors read and approved the manuscript.

FUNDING
This work was supported by grants from the National Natural Science Foundation of China (81860259); Provincal Science foudation of Jiangxi (20202BABL206016); Youth Team Project of the Second Affiliated Hospital of Nanchang University (2019YNTD12003).