Regulation of signal transduction pathways in colorectal cancer: implications for therapeutic resistance

Mitogen-activated protein kinase (MAPK) pathway

The MAPK pathway is mediated by mitogen-activated protein kinases (MAPKs), which comprise a family of serine/threonine-specific protein kinases that regulate a variety of cellular processes and play crucial roles in the pathogenesis of many diseases such as cancer, infection, inflammatory and autoimmune diseases (Kim & Choi, 2010). The MAPK family is divided into several subgroups. Conventional MAPKs are (i) extracellular signal-regulated kinase 1 and 2 (ERK1/2), (ii) c-Jun N-terminal kinases (JNKs), (iii) p38 (also known as MAPK14) and (iv) ERK5, all are summarised in Fig. 1 (Morrison, 2012). Atypical MAPKs comprise ERK3/4, ERK7/8 and Nemo-like kinase (NLK) (Cargnello & Roux, 2011). In the canonical ERK/MAPK signalling pathway, extracellular signals (e.g., growth factors, stress, mitogens) bind to the receptors, most of which are receptor tyrosine kinases (RTKs) at the surface of the cell membrane, leading to auto-phosphorylation of growth factors receptors and recruitment of adaptor proteins (e.g., growth factor receptor-bound protein 2 (GRB2) and son of sevenless 1 (SOS1)) (Morrison, 2012). This in turn results in the switching of the inactivated form of Ras-family GTPase (Ras in GDP bound form, Ras-GDP) to the active form of Ras-family GTPase (Ras in GTP bound form, Ras-GTP). The external signal is then transmitted via Ras-GTP to other downstream phosphorylation targets within the cytoplasm where the signalling cascade converges at the activation of a series of MAPKs, starting from MAPK kinase kinase (MAPKKK, e.g., Raf1) followed by MAPK kinase (MAPKK, e.g., MEK1/2) and MAP kinase (MAPK, e.g., ERK1/2). Finally, the MAP kinase translocates to the nucleus to phosphorylate transcription factors (e.g., c-Jun, STAT1, c-Myc) that regulate transcription of genes for different cellular processes (Wei & Liu, 2002; Fang & Richardson, 2005; Morrison, 2012) (Fig. 1).

In multidrug-resistant CRC, MAPK pathway is often reprogrammed, usually by the overexpression of RTKs, Ras and Raf; or gain-of-function mutations of Ras and Raf, which sustain the activity of MAPK signalling pathway upon treatment with MAPK and RTK inhibitors, 5-FU and oxaliplatin (Wan et al., 2004; Kavuri et al., 2015; Martinelli et al., 2017; Ressa et al., 2018). EGFR has been a favourable target for the treatment of mCRC since the last decade, mainly because they are highly expressed in most human tumours, including CRC (Yarden & Pines, 2012). In particular, monoclonal antibody targeting EGFR, such as cetuximab and panitumumab are widely used to treat mCRC patients due to their initial benefit of improving patients’ survival (Jonker et al., 2007; Vermorken et al., 2008; Pirker et al., 2009). However, some studies have shown that anti-EGFR based therapy may not be effective in treating mCRC, indicating that a subset of CRC is resistant to the anti-EGFR treatment (Bokemeyer et al., 2009; Van Cutsem et al., 2009). Common molecular mechanisms associated with the resistance are KRAS, NRAS, BRAF and PI3KCA mutations (De Roock et al., 2010; Diaz et al., 2012). In CRC which is quadruple wild-type for KRAS, NRAS, BRAF and PI3KCA genes, HER2 gene amplification and activating mutations at the phosphorylation sites of the catalytic domain have been shown to bypass EGFR blockade by activating a compensatory signalling mechanism for cell survival (Kavuri et al., 2015; Belli et al., 2019). It has been reported that HER2 could form heterodimers with either EGFR or ERBB3 with consequent activation of ERK and Akt signalling respectively, in which the latter has been shown to promote anti-EGFR resistance (Zhang et al., 2014a; Zhang et al., 2014b). Besides, it has been found that aberrant ERBB2 activation could result in the stimulation of ERK 1/2 signalling that mediates cetuximab resistance (Yonesaka et al., 2011).

B-Raf is a MAPKKK that mediates cell growth and differentiation via the ERK/MAPK subfamily of MAPK pathway, in response to growth factors and mitogens (Morrison, 2012). B-Raf mutations, more often BRAF V600E, occur in approximately 8% of CRC and are associated with poor prognosis (Davies et al., 2002; Richman et al., 2009). BRAF V600E mutation leads to conformational changes at the catalytic domain which renders B-Raf constitutively active, independent of Ras-GTP activation and dimerization with Raf-1 (also known as C-Raf) (Durrant & Morrison, 2018). This results in prolonged phosphorylation and activation of MEK1/2 and ERK1/2 kinases which, in turn, activates downstream substrates that mediate cell growth and survival (Fang & Richardson, 2005). It has been reported that targeting BRAF V600E using mono-therapeutic agents, such as vemurafenib (a B-Raf inhibitor, also known as PLX4032) which binds to the ATP-binding site of BRAF V600E to inhibit its activity, shows limited therapeutic response in CRC (Prahallad et al., 2012). This is because targeting BRAF V600E results in feedback activation of EGFR characterised by enhanced phosphorylation of RAS and CRAF upstream of MAPK pathway and downstream activation of RAF, MEK and ERK (Corcoran et al., 2012). In order to circumvent resistance to B-Raf inhibition, B-Raf inhibitor is used in combination with EGFR inhibitor or MEK inhibitor or both which were initially shown to offer therapeutic benefit of at least 12% response rate to the drugs and improve the suppression of ERK/MAPK pathway (Bendell et al., 2014; Corcoran et al., 2014; Tabernero et al., 2014). Despite the initial success in suppressing B-Raf resistance using the multi-target approach, there is compelling evidence that BRAF V600E mutant CRC patients could also develop resistance to the new treatment (Oddo et al., 2016). Ahronian et al. (2015) has shown that the reactivation of the MAPK pathway confers cross-resistance to the combined RAF/EGFR or RAF/MEK inhibition in BRAF-mutant CRC and further demonstrated that the use of ERK inhibitor could overcome the resistance by suppressing the MAPK signalling.

On the other hand, it has been reported that treatment using ATP-competitive inhibitors produces opposing mechanisms of action that is dependent on the cellular context and genotype of the tumour. It was found that RAF inhibitors effectively block MAPK pathway in BRAF V600E cells but activate the MAPK pathway in wild-type BRAF tumours by inducing RAF dimerization and MEK/ERK phosphorylation leading to enhanced tumour growth, suggesting that other strategies to block RAF activation are needed to improve the treatment efficacy (Hatzivassiliou et al., 2010). Furthermore, RAS mutant tumours are also known to exhibit poor response to RAF inhibitors. A time-course phosphoproteomic analysis of vemurafenib-treated RAS mutant CRC cell lines has found potential cross-talk between ERK signalling with Hippo and Rho pathways and revealed novel functional targets downstream of ERK (Kubiniok et al., 2017).

Apart from post-translational regulation of proteins as a regulatory checkpoint for cellular signalling, microRNAs (miRNAs) also exhibit a functional role in the regulation of MAPK pathway in CRC drug resistance (Rasmussen et al., 2016; Angius et al., 2019). A previous miRNA profiling study has uncovered the link between high expression of miRNA-625-3p and poor clinical response towards oxaliplatin-based therapy (Arango et al., 2004; Rasmussen et al., 2013). Mechanistically, it has been demonstrated that miRNA-625-3p mediates oxaliplatin resistance by targeting MAPK kinase MAP2K6 and abrogates MAPK14 signalling, leading to increased cell cycle progression and reduced apoptosis (Rasmussen et al., 2016).

As one of the most frequently altered signalling pathways and its important roles in CRC drug resistance, MAPK pathway represents a promising target for cancer therapy. Significant progress has been made on the development of therapeutics targeting MAPK kinases with considerable clinical success (Bendell et al., 2014). Notably, the combination of BRAF and MEK inhibitors has been shown to improve response rates and may offer potential therapeutic benefit in BRAF-mutated CRC (Corcoran et al., 2014). Nevertheless, it is clear that we are still far from any complete understanding of the MAPK pathway. Moreover, the emergence of new resistance mechanisms prompts more further research to provide a deeper understanding on the complex regulation and interconnectivity of the underlying biological processes to overcome resistance and increase therapeutic efficacy.

Phosphoinositide 3-kinase (PI3K)/protein kinase B (PKB, also known as AKT) pathway

The PI3K/PKB pathway regulates cell metabolism, cell growth and cell survival. In normal condition, PI3K/PKB pathway is activated by four major sensors upstream of the pathway, namely (i) receptor tyrosine kinases (RTKs) which bind to growth factors (Hemmings, 1997), (ii) cytokine receptors (Chang et al., 2003), (iii) G protein-coupled receptors (GPCRs) that are activated by various biological molecules (Murga et al., 1998), and (iv) integrins which detect cell–cell or cell–matrix communication (Su et al., 2007). Upon ligand binding, these receptors, together with their cofactors, will activate PI3K family proteins. There are three classes of PI3K family proteins, among which only class I PI3Ks and the signalling networks they regulate are covered in this review (Fig. 2). Information about Class II and Class III PI3Ks and their roles in cellular signalling are covered in other reviews or journal articles (Falasca & Maffucci, 2012; Okkenhaug, 2013; Backer, 2016; Hawkins & Stephens, 2016). Within the class I PI3K subfamily itself, there are four catalytic isoforms (p110α, p110β, p110γ and p110δ encoded by PIK3CA, PIK3CB, PIK3CG, and PIK3CD respectively) which catalyse the phosphorylation of phosphatidylinositol-4,5-bisphosphate (PI(4,5)P₂) to phosphatidylinositol (3,4,5)-trisphosphate (PI(3,4,5)P₃) or PIP₃. p110α and p110β are expressed ubiquitously while p110γ and p110δ are expressed in immune cells. Each catalytic isoform forms a dimer with a regulatory subunit that controls the activity and subcellular localisation of the PI3K complex (Fig. 2). In response to specific external stimuli, PIP₃, which acts as a secondary messenger, will recruit cytoplasmic proteins with PIP₃ binding domains (typical examples of which are phosphoinositide-dependent kinase-1 (PDK1) and PKB) to specific cell membrane locations. Shortly after the transmission of the signal to downstream effectors, PIP₃ is then metabolised by phosphatase and tensin homolog (PTEN), which is a tumour suppressor that negatively regulates the PI3K signal by removing 3′-phosphate from PIP₃ (Danielsen et al., 2015; Fruman et al., 2017). Activation of PKB is a two-step process, whereby PDK1 phosphorylates PKB on threonine-308 to partially activate PKB (Alessi et al., 1997), followed by phosphorylation of PKB by mTORC2 on serine-473 to fully activate PKB (Sarbassov et al., 2005). The activated PKB will subsequently regulate the phosphorylation of the target substrates (the most well-known examples are glycogen synthase kinase 3 beta (GSK3β), BCL2-antagonist of death (BAD), mouse double minute homolog 2 (MDM2), forkhead box O (FOXO) and mechanistic target of rapamycin complex (mTORC1)) which mediate important cellular functions, such as glucose uptake, protein synthesis, cell survival and cell cycle progression (Manning & Cantley, 2007). The activity of PKB is also negatively regulated by protein phosphatase 2A (PP2A), PH domain leucine-rich repeat protein phosphatase 1/2 (PHLPP1/2) and carboxyl-terminal modulator protein (CTMP) (Hemmings & Restuccia, 2012) (Fig. 2). In multidrug-resistant CRC, PI3K signalling is prolonged by PIK3CA mutations, null mutation of PTEN, and RAS mutations upon treatment with standard chemotherapeutic drugs and targeted therapeutic agents (Laurent-Puig et al., 2009; De Roock et al., 2010; Kim et al., 2013; Hamada, Nowak & Ogino, 2017).

PIK3CA is regarded as one of the most frequently mutated genes in CRC, which accounts for approximately 10–30% of all CRC cases (Samuels et al., 2004; Velho et al., 2005; Hamada, Nowak & Ogino, 2017). It is reported that non-random somatic mutations occurring in the coding region, mainly exon 9 (helical domain) and exon 20 (catalytic domain), have heightened basal PI3K and PKB activities which then promote cancer progression (Kang, Bader & Vogt, 2005; Ikenoue et al., 2005). Furthermore, it has been shown that PIK3CA mutations (E545K, E542K and E545D on exon 9; H1047R and H1047L on exon 20) could mediate resistance to standard chemotherapy (FOLFOX and FOLFORI) by inducing phosphorylation of PKB and expression CRC stem cell markers (LGR5) via sustained PI3K signalling, thereby promoting cancer cell survival and proliferation (Wang et al., 2018). Interestingly, PIK3CA mutations have also been found to mediate acquired cetuximab resistance in mCRC complementary to the previously reported RAS mutations. A recent circulating tumour DNA sequencing analysis from mCRC patients has revealed five novel mutations on exon 19 of PIK3CA (K944N, V955I, F930S, V955G and K966E) that may potentially drive resistance to cetuximab via EGFR-mediated activation of the PI3K/PKB signalling pathway, suggesting that combined regimens of PI3K/mTOR inhibitors (PP242 and NVP-BEZ235) with anti-EGFR therapy may be beneficial to overcome the resistance (Xu et al., 2017).

Apart from PIK3CA mutations, there is increasing evidence of the emergence of mutations involving components of MAPK and PI3K/PKB signalling pathways when treated with targeted therapeutic agents (De Roock et al., 2010; Vitiello et al., 2019). Although anti-EGFR based therapy is commonly prescribed for KRAS wild-type CRC patients, clinical evidence has indicated that the KRAS mutation status alone is insufficient to predict therapeutic response to the therapy (Allegra et al., 2009). Retrospective cohort studies of CRC cases have identified several key players in the EGFR signalling pathway (which is a signalling network shared by both MAPK and PI3K/PKB pathways) that hinder the effectiveness of anti-EGFR monoclonal antibodies in KRAS wild-type CRC patients (Sartore-Bianchi et al., 2009; Laurent-Puig et al., 2009; De Roock et al., 2010). It has been previously reported that BRAF, NRAS, PTEN and PIK3CA mutations are associated with the efficacy and clinical outcome of EGFR-targeted therapy but their exact roles in driving the resistance are still unclear (Laurent-Puig et al., 2009; De Roock et al., 2010). Surprisingly, KRAS mutations also negatively affect the outcome of treatment against PIK3CA mutant CRC. Kim et al. (2013) have shown that KRAS and PIK3CA mutations attenuate sensitivity to treatment with a dual inhibitor of PI3K and mTOR by suppressing BIM-induced apoptosis via activation of PI3K/MTOR pathway, leading to cell survival.

More recently, miRNAs have also been reported to control CRC pathogenesis via the modulation of PI3K/PKB pathway (Soleimani et al., 2019). Based on previous studies, it has been shown that miR-29a could induce or suppress tumour progression in drug-resistant cancer cells (Zhong et al., 2013; Liu et al., 2018). Yuan et al. (2018) showed that higher level of miR-29a is expressed in CRC cell lines resistant to paclitaxel which resulted in downregulation of PTEN and upregulation of phosphorylated AKT. This suggests that miR-29a has a regulatory function to PI3K/PKB pathway via inhibition of PTEN which reduces drug sensitivity and supports cancer growth. In contrast, miR-29a could potentially reduce P-gp-mediated chemoresistance via modulation of PTEN and P-gp expression in doxorubicin-resistant CRC cell lines. This means that miR-29a exerts tumour suppressive function by inhibiting membrane transporter activity through PI3K/PKB pathway (Shi et al., 2020). These seemingly contradictory roles of miR-29a have also been observed in other malignancies. In the context of regulation of drug resistance, miR-29a has been reported to increase sensitisation to gemcitabine in pancreatic cancer cells (Kwon et al., 2016) as well as sensitization to tamoxifen in breast cancer cells (Muluhngwi et al., 2017). On the other hand, miR-29a was shown to play a role in mediating adriamycin resistance in breast cancer cells via inhibiting the PTEN/AKT/GSK3β pathway (Shen et al., 2016). It is unclear whether the opposing results are due to heterogeneity in the cancer cells or that miR-29a could exhibit multifaceted functions in a context dependent manner. Nevertheless, these paradoxical findings demand the need for further work to clarify and elucidate the function of miR-29a more comprehensively. Similar to miR-29a, miR-543 has recently been identified as the mediator of chemoresistance in CRC cells, also by suppressing the expression of PTEN which activates PI3K/PKB pathway (Liu, Zhou & Dong, 2019). Interestingly, miR-4689 which has been identified as a negative regulator of KRAS and AKT is downregulated in KRAS mutant CRC cells and confers resistance to molecular-targeted therapy, suggesting that miR-4689 could be a promising therapeutic agent to control multidrug resistance in CRC via modulation of PI3K/PKB pathway and MAPK pathway (Hiraki et al., 2015).

Taken together, current evidence clearly shows that mutations of the MAPK/PI3K/PKB signalling pathway components are frequently observed in cancer drug resistance. However, the signalling mechanisms associated with these mutations are still not well elucidated which necessitate further investigation. Importantly, distinct molecular events that regulate drug resistance such as in KRAS wild-type and mutant CRC suggests the importance of identifying relevant drug targets in CRC with different mutational status. Moreover, future research should also focus on dissecting the link of miRNAs with cancer drug resistance and their underlying molecular mechanisms.

Wnt/β-catenin pathway

The Wnt/β-catenin pathway is an evolutionarily conserved system that regulates cell development, cell differentiation, cell proliferation and cell migration. The Wnt/β-catenin pathway can be grouped into β-catenin-dependent Wnt pathway (canonical Wnt pathway) and β-catenin-independent Wnt pathway (non-canonical Wnt pathway) which are further divided into the planar cell polarity Wnt pathway and the Wnt/Ca²⁺ pathway (Komiya & Habas, 2008). The canonical Wnt/β-catenin pathway is made up of the membrane proteins, degradation complex and β-catenin protein. In the absence of Wnt ligands, the degradation complex which comprises of adenomatous polyposis coli (APC), Axin, GSK3 and CK1α is formed through phosphorylation of Axin and APC by GSK3 and casein kinase 1α (CK1α). As a result, β-catenin is ubiquitinated by E3-ligase protein βTrCP (β-transducin repeats-containing proteins) through phosphorylation and targeted for proteasomal degradation (Van Kappel & Maurice, 2017) (Fig. 3). In the presence of Wnt ligands, lipoprotein receptor-related protein 5/6 (LRP5/6) co-receptor and Frizzled (Fzd) receptor are activated which leads to phosphorylation of LRP5/6 co-receptors and binding of adaptor protein disheveled (Dvl) to the phosphorylated LRP5/6. This is followed by the recruitment of the remaining degradation complex components to the Fzd-LRP5/6 complex to inactivate the degradation complex (Janda et al., 2012) (Fig. 3). The molecular mechanism for Wnt-mediated degradation complex inactivation is still heavily disputed due to conflicting findings on the inhibition of GSK3 in the presence of Wnt signal. Several models have been proposed: (1) blockade of GSK3 catalytic site by binding to the phosphorylation motif of LRP5/6 (Wu et al., 2009). (2) Wnt-mediated dissociation of APC from GSK3 (Valvezan et al., 2012). (3) sequestration of GSK3 in endosomal vesicles through endocytosis of Fzd-LRP5/6 complex (Taelman et al., 2010). Disruption of the degradation complex integrity in the presence of Wnt signal promotes stabilisation of β-catenin, leading to accumulation of newly synthesised β-catenin in the cytoplasm and their subsequent translocation to the nucleus. Interestingly, it is also known that, in the presence of Wnt signal, the degradation complex could remain intact to target β-catenin for degradation via phosphorylation, but ubiquitination of β-catenin is impaired which inhibits its degradation by the degradation complex (Gerlach et al., 2014). Within the nucleus, β-catenin binds to T-cell factor/lymphoid enhancing factor (TCF/LEF) which are the transcription factors that activate Wnt-responsive genes required for cell growth and survival (e.g., c-Myc, Cyclin D1). In addition, β-catenin interacts with TCF/LEF to recruit transcriptional co-activators (p300/CBP and BCL9) to the transcription factors to activate gene expression (Kretzschmar & Clevers, 2017; Taciak et al., 2018) (Fig. 3). In multidrug-resistant CRC, Wnt/β-catenin pathway is reprogrammed by overexpression of Dvl protein and non-coding RNAs that interfere with the activities of downstream signalling mediators (Yue et al., 2016; Han et al., 2017; Bian et al., 2017; Zhang et al., 2017b; Jiang et al., 2019).

Dysregulation in the key Wnt/β-catenin pathway components such as upstream regulator (Dvl protein), β-catenin degradation complex and its downstream targets (β-catenin and TCF/LEF) instigate tumour progression in many types of cancer (Van Kappel & Maurice, 2017). Moreover, recent studies indicate that aberrant Wnt/β-catenin signalling could trigger anti-cancer drug resistance (Zhang et al., 2017a). Zhang et al. (2017b) reported that DVL1-3 proteins are overexpressed in CRC resistant to vincristine (a chemotherapeutic drug that interferes with microtubule synthesis leading to cell cycle arrest) which results in overexpression of ABC transporters (P-glycoprotein (P-gp), MRP2 and BCRP) and anti-apoptotic proteins (Survivin and Bcl-2). Contrary to previous findings which suggest that DVL promotes β-catenin accumulation and subsequent translocation to the nucleus, it was found that DVL1-3 translocate to the nucleus and bind to β-catenin to form a transcriptional complex, independent of β-catenin accumulation and nuclear translocation (Gao & Chen, 2010; Shang, Hua & Hu, 2017). Moreover, the study also showed that silencing DVL1-3 could re-sensitise CRC cells to vincristine, 5-FU and oxaliplatin, suggesting that DVL could be a potential therapeutic target in multidrug-resistant CRC.

It has been increasingly recognised that long non-coding RNA (lncRNA, a non-coding regulatory RNA with greater than 200 nucleotides in length) regulates Wnt/β-catenin pathway in multidrug-resistant CRC (Ma et al., 2016; Lu et al., 2017). Recent studies have shown that lncRNA cytoskeleton regulator RNA (CYTOR, also known as LINC00152) is overexpressed in CRC which confers resistance to oxaliplatin-induced apoptosis (Yue et al., 2016). In addition, elevated expression of LINC00152 is also observed in CRC that gives rise to 5-FU resistance and cancer metastasis (Bian et al., 2017). However, the regulatory mechanism of LINC00152 in the Wnt/β-catenin pathway of mCRC is still unknown (Yue et al., 2016). In a recent study, Yue et al. (2018) demonstrated that LINC00152 competitively binds to β-catenin to prevent CK1α from phosphorylating β-catenin. As a result, β-catenin accumulates and translocates to the nucleus to activate the expression of epithelial-mesenchymal transition (EMT) markers (N-cadherin and Vimentin) which are hallmarks of metastatic cancer. Reciprocally, β-catenin/TCF4 transcriptional complex promotes the expression of LINC00152 to sustain Wnt/β-catenin signalling in mCRC.

Likewise, miRNA is also known to influence cancer growth and the sensitivity of cancer cells towards anti-cancer drugs (Jansson & Lund, 2012; Piletič & Kunej, 2016). Jiang et al. (2019) have demonstrated that overexpression of miR-30-5p negatively regulates the expression of Wnt/β-catenin pathway target genes (Axin2 and c-Myc) and inhibits chemoresistance in CRC cells by targeting ubiquitin-specific peptidase 22 (USP22). Mechanistically, it has been reported that USP22 induces β-catenin nuclear localisation and upregulates FoxM1 expression to promote G1/S cell cycle transition and cell proliferation (Ning et al., 2014). In another study conducted by Chen et al. (2019), miR-103/107 has been shown to repress the activity of Axin2 leading to sustained activation of Wnt/β-catenin signalling that potentiates cancer stemness and chemotherapeutic resistance. Nevertheless, despite that non-coding RNAs such as lncRNA and miRNA are known to promote resistance to therapeutic agents in CRC, the interaction network between LncRNA and miRNA is not well defined (Gao et al., 2019). Han et al. (2017) has reported that lncRNA Colorectal Neoplasia Differentially Expressed (CRNDE) binds to miR-181a-5p to repress its expression, resulting in increased levels of its downstream targets β-catenin and transcription factor TCF4 in the Wnt/β-catenin signaling pathway. It was also demonstrated that CRNDE knockdown and miR-181a-5p overexpression inhibit Wnt/β-catenin signaling and could reduce chemoresistance and attenuate cell proliferation in CRC cells, suggesting that it could be a novel cancer therapeutic strategy.

Over the years, substantial efforts have been directed to studying the molecular mechanisms and functional effects of Wnt signalling pathway. Accumulating studies confirm the critical role Wnt signalling in drug resistance and convey important insights into its underlying mechanisms that confer resistance to different therapies. Notably, such knowledge could be potentially harnessed to facilitate the development of specific inhibitors or drug combinations to improve anticancer efficacy. Nevertheless, owing to the complexity of Wnt signalling, there are still numerous details remain be uncovered with regard to its connections to therapy resistance that warrant further investigation.

Notch pathway

Similar to other signalling pathways (Wnt/β-catenin, Hedgehog (Hh), and transforming growth factor-beta (TGF-β)/bone morphogenic protein (BMP)), the Notch pathway is highly conserved across species and is known to control cell development, apoptosis, cell differentiation and proliferation (Artavanis-Tsakonas, Rand & Lake, 1999). Notch receptors (Notch 1–4) are synthesised as precursors from mRNAs (known as pre-Notch receptor) which then undergo fucosylation (a type of glycosylation) in the endoplasmic reticulum. In the Golgi apparatus, Notch receptors are further modified by enzymes (one typical example is Fringe) and cleaved at site 1 (S1) by furin-like convertase to induce heterodimerisation of Notch receptors (Siebel & Lendahl, 2017) (Fig. 4). At the cell surface, Notch receptors bind to Notch ligands (e.g., Jagged1, Jagged2, Delta-like ligand 1 (Dll1), Delta-like ligand 3 (Dll3) and Delta-like ligand 4 (Dll4)) of neighbouring cells. This initiates subsequent cleavages of Notch receptors by ADAM10/17 metalloproteases and presenilin–γ-secretase enzyme complex at the outer side (site 2 (S2)) and inner side (site 3 (S3)) of the cell membrane (Bray, 2006) (Fig. 4). The end product of the proteolytic cleavages of Notch receptors known as Notch intracellular cleaved domain (NICD, an active form of the molecules which acts as transcriptional activators), travels to the nucleus to displace co-repressors (e.g., recombining binding protein J-kappa (RBPJκ) or CSL) and interacts with transcriptional co-activators (e.g., mastermind-like (MAML), histone acetyltransferase (HAT), p300), in order to activate the transcription of Notch target genes (e.g., Hes and Hey family proteins, cyclin D3, c-Myc). Notch signalling is terminated when the intracellular domain of Notch (ICN or NICD) is targeted for proteasomal degradation through a ubiquitin pathway (Vinson et al., 2016; Siebel & Lendahl, 2017) (Fig. 4). In multidrug-resistant CRC, cross-regulation of signalling pathways, post-transcriptional regulation and overexpression of genes in the Notch signalling pathway are among the common mechanisms underlying the development of resistance towards targeted or chemotherapeutic regimens (Rodilla et al., 2009; Majidinia et al., 2018; Negri et al., 2019).

There is growing evidence that upregulation of Notch receptors, Notch ligands and Notch target genes could lead to the maintenance of CRC stem cell populations and the acquisition of metastatic phenotype which are strongly related to poor survival of CRC patients and drug resistance (Mohamed et al., 2019; Weber et al., 2019; Shaik et al., 2020). It has been previously reported that Notch-1 signalling pathway induces EMT in non-small cell lung cancer resistant cells with acquired resistance to EGFR tyrosine kinase inhibitor, suggesting that Notch signalling may contribute to cancer drug resistance (Xie et al., 2013). In another study, Mirone et al. (2016) has reported that CRC cells that are resistant to regorafenib (a small-molecule multi kinase inhibitor) showed significant upregulation of Notch-1 and the target genes (HES1 and HEY1). The study demonstrated that knockdown of Notch-1 could partially restore the sensitivity to regorafenib and inhibit cell growth, indicating that Notch-1 may play a role in tumour resistance. In mCRC patients treated with bevacizumab-based therapy, high NICD expression was found to be associated with poorer response whereas no correlation was observed between Dll-4 expression and clinical response (Negri & Ardizzoni, 2015). In a follow-up study, Negri et al. (2019) has shown that high expression levels of NICD and CD44 are linked to cancer stemness in patients with advanced CRC treated with bevacizumab. Notably, the study demonstrated that NICDs (the functional components of Notch signalling pathway) instead of Dll-4 induce resistance to anti-angiogenic therapy in CRC via activation of Notch-induced regulation of colon cancer stem cells. Nevertheless, the functional roles of NICDs and CD44s in the CRC microenvironment during anti-angiogenic treatment are still unclear (Negri et al., 2019). HES1, which is a downstream target of Notch pathway and one of the important markers of CRC stem cells, is known to contribute to tumour relapses in CRC patients after 5-FU based chemotherapy, but the role of HES1 in chemoresistant CRC has not yet been elucidated (Gao et al., 2014). A recent study by Sun et al. (2017) has shown that HES1 modulates gene expression related to drug metabolism and EMT, notable overexpression of ABC transporters (ABCC1, ABCC2 and P-gp1) with depressed E-cadherins and elevated N-cadherins in CRC cell lines treated with 5-FU, supporting the crucial role of HES1 in promoting chemoresistance.

miRNAs are known to regulate Notch signalling pathway which results in various tumour pathology, such as metastasis, tumour relapses, cancer stemness and low survival rate (Majidinia et al., 2018; Khan et al., 2019). Recently, accumulating evidence also suggests that cross-regulation between miRNAs and Notch signalling pathway plays a critical role in cancer drug resistance. miR-139-5p is a tumour suppressor that has been found to be frequently downregulated in CRC (Shen et al., 2014). It has been reported that miR-139-5p targets Notch-1 and regulates its signal transduction to exert tumour suppressive effect in CRC (Zhang et al., 2014b). Furthermore, miR-139-5p/Notch-1 signalling has also been correlated with drug resistance in CRC. Liu et al. (2016) has shown that miR-139-5p sensitises CRC cells to 5-FU by inhibiting Notch-1 and its downstream multidrug-resistant genes (MRP-1 and BCL-2). Similarly, miR-195-5p is known to suppress cancer growth by inhibiting cell cycle progression, cell proliferation and cell migration (Luo et al., 2014). A recent study by Jin et al. (2018) has revealed that miR-195-5p could inhibit CRC cell stemness and 5-FU resistance by targeting the Notch signalling proteins Notch-2 and RBPJ, suggesting that miR-195-5p could be a potential therapeutic target in chemoresistance. On the other hand, a recent study by Xie et al. (2020b) reported that miR-34a could negatively regulate multidrug resistance protein ABCG2 via DLL1-mediated Notch signalling pathway and demonstrated that overexpression of miR-34a could overcome 5-FU resistance in CRC cells.

Overall, these findings suggest the critical role of Notch signalling pathway in cancer drug resistance. Notably, recent evidence indicates that Notch contribute to the maintenance of CRC stem cells and resistance to therapeutic agents, hence targeting Notch pathway may hold a promising prospect for cancer therapy. Thus, further studies are needed to elucidate the underlying mechanisms and the crosstalk between Notch and other signalling pathways to facilitate the design of better therapeutic approach.