1. Introduction
The first WNT gene, then known as
int-1 (mouse mammary tumor virus integration site 1), was isolated from mouse mammary tumors in 1982
[5][1].
Int-1 turned out to be highly conserved across multiple species and especially similar to the
Drosophila gene
wingless (Wg) a gene found to be involved in wing development, segmentation and formation of body axis during flight development
[6,7,8,9][2][3][4][5]. The name
WNT comes from a fusion of
wg and
int [10,11][6][7]. Most animals have several WNT genes with mice and humans encoding 19 WNT genes, seven in
Drosophila and five in
Caenorhabditis elegans (
C.elegans)
[6][2].
The WNT proteins are secreted, lipid-modified glycoproteins, usually 350–400 amino acids in length
[12][8]. WNT proteins act as ligands interacting with Frizzled (FZD) receptors on the cell surface to activate intracellular signaling pathways
[6,12,13][2][8][9]. Frizzled receptors are seven-pass transmembrane proteins (similar to G-protein coupled receptors or GPCRs) that act as primary receptors for WNT signals. In addition to the interaction between WNT ligands and FZD receptors, a variety of co-receptors, such as, the low-density lipoprotein receptor related protein (LRP) may be required to mediate WNT signaling
[14,15][10][11]. Upon activation, a signal is transduced by the pathway activating protein Disheveled (Dsh or Dvl)
[13,16][9][12]. Dsh proteins have highly conserved protein domains comprised of an amino-terminal DIX domain (named for Dsh and Axin), a central PDZ domain (named for postsynaptic density-95, discs-large and zonula occludens-1) and a carboxy-terminal DEP domain (named for Dsh, Egl-10 and pleckstrin). Dsh acts as a key switch in WNT signaling where, depending on which of the three domains is activated, the WNT signal can be branched off into multiple downstream pathways
[16][12]. The resulting pathways can be categorized into the canonical WNT pathway (β-catenin dependent pathway) and the non-canonical WNT pathways (β-catenin independent pathways) which include polarity and the WNT/Ca
2+ pathway
[15][11].
Canonical pathway: The stability of cytoplasmic β-catenin is mediated by a multimeric protein complex known as the destruction complex formed by the scaffolding proteins (Axin)
[17][13], the tumor suppressor adenomatous polyposis coli (APC)
[18,19][14][15], glycogen synthase kinase 3 (GSK-3) and casein kinase 1 (CK1). In the absence of canonical WNT ligands, β-catenin binds to the destruction complex and is phosphorylated by CK1 and GSK-3. Phosphorylated β-catenin is then ubiquitinated and subsequently degraded through the proteasome (
Figure 1)
[14,20,21,22][10][16][17][18].
Figure 1. The canonical WNT pathway in on and off states. APC, adenomatous polyposis coli; CK1, casein kinase 1; GSK3, glycogen synthase kinase 3; TCF/LEF, T-cell factor/lymphoid enhancing factor; LRP6, lipoprotein receptor related protein 6; FZD/Fz, Frizzled; Dsh, Disheveled.
In the presence of canonical WNT ligands, β-catenin phosphorylation and degradation is inhibited. Gene expression downstream of canonical WNT signaling is regulated by controlling the amount of the transcriptional co-activator β-catenin. When WNT ligands bind to FZD receptors and LRP co-receptors, a cascade of events is initiated, resulting in the disassembly of the destruction complex stabilizing β-catenin
[22][18]. The signaling cascade is initiated when Dsh is activated through the DIX and PDZ domain, and the FZD-associated Disheveled (Dsh) protein then binds to Axin; Axin then inhibits GSK-3 phosphorylation of β-catenin in the destruction complex. Axin and the destruction complex are thus recruited to the plasma membrane forming signalosomes
[23][19]. β-catenin accumulates in the cytoplasm and localizes to the nucleus where it forms a nuclear complex with DNA-bound T-cell factor/lymphoid enhancing factor (TCF/LEF) transcription factor, resulting in the activation of WNT-responsive genes
[14,16,20,24][10][12][16][20].
Non-canonical pathways: In contrast to the β-catenin-dependent canonical pathway, WNT proteins are able to activate additional signaling pathways that are independent of β-catenin. These pathways are called non-canonical pathways, which can be further categorized into two distinct branches, the planar cell polarity (PCP) pathway and the WNT/Ca2+ pathway (Figure 2).
Figure 2. The non-canonical planar cell polarity and Calcium pathways. DAAM1, Disheveled-associated activator of morphogenesis 1; Rac; ROCK, Rho-associated kinase; JNK, Jun kinase; PLC, Phospholipase C; CamKII, calcium/calmodulin-dependent kinase II.
The non-canonical PCP pathway regulates a variety of cellular behaviors including planar cell polarity, cell movements during gastrulation and cell migration of neural crest cells
[25,26,27,28][21][22][23][24]. This PCP pathway is activated through the binding of WNT ligands to the FZD receptor independently of the LRP
[16][12]. The non-canonical pathways do not need to utilize the majority of canonical pathway components including WNT itself, but most do involve Dsh and specifically the PDZ and DEP domains and the Disheveled-associated activator of morphogenesis 1 (DAAM 1). This links FZD and Dsh to the small GTPases Rho, and Rho further activates Rho-associated kinase (ROCK), thus leading to cytoskeletal reorganization
[29][25]. Dsh utilizes the DEP domain to form a complex with Rac GTPase, independent of the DAAM 1, then stimulates Jun kinase (JNK) activity and mediates profilin binding to actin
[16,28,29][12][24][25]. The polarity signaling pathway is likely best described as a collection of WNT-dependent and WNT-independent effects on cellular organization both within a cell sheet and within the cell itself, showing an intricate interplay between polarity and adhesion
[30,31,32,33,34,35][26][27][28][29][30][31].
A second branch of non-canonical WNT signaling is termed the WNT/Ca
2+ pathway. The role of the WNT/Ca
2+ pathway is to regulate the release of the intracellular Ca
2+ from the endoplasmic reticulum
[36,37][32][33] and modulate signaling for dorsal axis formation and PCP signaling for gastrulation cell movements, cell adhesion, migration and tissue separation during gastrulation
[15,38][11][34]. Similar to other WNT pathways, the activation of the WNT/Ca
2+ pathway requires the binding of the WNT ligands to the FZD receptor. The activated FZD receptor directly interacts and activates the heterotrimeric G-proteins, leading to an increase in the intracellular Ca
2+ concentration
[15][11]. The released Ca
2+ then activates calcium/calmodulin-dependent kinase II (CamKII), calcineurin or Protein Kinase C
[36][32].
4. WNT Signaling in Cancer
The WNT pathway plays a complex role in cancer development. Often, mutations of key components are associated with processes like uncontrollable cell proliferation, epithelial–mesenchymal transition (EMT) and metastasis. This section will discuss the main WNT components that have been investigated in oncology and will review the current progress of their roles in cancer development (Figure 3).
Figure 3. Canonical WNT pathway components associated with disease.
Frequent APC mutations in colorectal cancer: The APC protein is an important component of the destruction complex in canonical WNT signaling. In addition to this role, APC is also essential for the rapid transition of Axin immediately after WNT stimulation and facilitates the association of Axin and co-receptor LRP6/Arrow in
Drosophila [59][55].
As APC negatively regulates the canonical WNT pathway, it is likely that it can act as a tumor suppressor. Consistent with this idea, APC mutations, which are present in approximately ~80% of colorectal cancers, are indeed mostly loss of function and/or truncating mutations
[60,61][56][57]. These APC mutations frequently occur in the mutation cluster region (MCR) between codons 1285 and 1513 which accounts for only 10% of the entire coding region
[62][58]. To illustrate this, a study concluded that 28 out of 43 (65%) somatic mutations in colorectal cancer cells occur in the MCR
[62][58]. These mutations can inhibit β-catenin ubiquitination and cause uncontrollable transcription
[63][59]. Interestingly, an analysis of 630 human sporadic colorectal cancer tumors revealed that different APC mutations can result in different levels of canonical WNT signaling, and each region of the large intestine has its own optimal threshold of WNT signaling for tumorigenesis
[60][56]. APC mutations also occur in the germline and are present in up to 85% of patients with classical familial adenomatous polyposis (CFAP). Most germline mutations are also truncating (e.g., frameshift, nonsense)
[64][60], confirming APC’s tumor suppressor role.
The prevalence of
APC mutations in colorectal cancers suggest
APC as a powerful target for therapy. TASIN-1 (truncated APC selective inhibitor-1) was identified as a molecule that could specifically kill cells with truncated APC’s. TASIN-1′s administration into xenograft and mouse models has shown its effectiveness in tumor suppression and its specificity in killing cells with truncated APC’s while sparing normal cells in vivo
[65][61]. Considering the high frequency of mutated
APC in patients with colorectal cancer, further studies are needed to investigate APC’s molecular mechanism in β-catenin degradation, its use as a potential biomarker and other possible therapeutic targeting approaches.
AXIN1/2: Axin proteins (AXIN1 and AXIN2) are important scaffold proteins that help assemble the β-catenin destruction complex
[66][62]. AXIN1/2 act as negative regulators of β-catenin levels and tumor suppressor proteins. This is consistent with the finding that the overexpression of AXIN
1 inhibits cell growth in hepatocellular carcinoma (HCC)
[67][63]. However, the role of AXIN1/2 as a tumor suppressor through the canonical WNT pathway remains controversial, as other findings have suggested that most human
Axin1 mutated HCCs do not show a β-catenin activation program. These HCCs might have occurred independently of the WNT/β-catenin pathway and instead involved other pathways like YAP and Notch
[68][64].
AXIN
2 has been shown to behave as both a tumor suppressor and an oncogene. Its mutations are also observed in various cancers. Chapman et al. observed genetic changes (mostly deletions) of
AXIN
2 in 7% of human adrenocortical adenomas tumor samples and 17% of adrenocortical carcinomas tumor samples
[69][65]. Consistent with the idea that
AXIN
2 might act as a tumor suppressor,
AXIN
2 downregulation is associated with poorer overall survival of patients with breast cancer
[70][66]. However, in vivo findings also suggest
AXIN
2 as tumor-promoting, as it upregulates the transcriptional repressor, SNAI1, leading to increased EMT and metastatic activity in colorectal cancers in mice
[71][67]. As these contradictory studies provide only preliminary evidence of the functional significance of Axin in canonical WNT pathways, future studies aimed towards exploring Axin’s expression in different cancers and its mechanism in affecting tumor progression should be conducted. The contradictory findings in cancer are not surprising as Axin’s role in development is complex
[72][68].
LRP5/6: LRP5 and LRP6 in the low-density lipoprotein receptor (LDLR) family are single-pass transmembrane coreceptors in the WNT canonical signaling pathway. The FZD receptor may form a WNT-induced FZD–LRP6 (or LRP5) complex with these coreceptors. After WNT ligands bind to these receptors, the β-catenin signaling is initiated
[73,74][69][70].
LRP5 has contradictory roles in cancer progression as well. When osteosarcoma cells are transfected with dominant negative, soluble LRP5 (sLRP5), epithelial–mesenchymal transition (EMT) is reversed, suggesting wildtype LRP5′s role in promoting EMT and metastasis
[75][71]. However, Horne et al. observed an opposite effect in which dominant-negative
LRP5 failed to block osteosarcoma cell formation, suggesting LRP’s role as a tumor suppressor
[76][72].
Similarly, LRP6′s role in cancer progression has been controversial. A total of 45% of human hepatocellular carcinoma cells have overexpressed
LRP6 and as a result, increased β-catenin levels, suggesting LRP6 as tumor-promoting
[77][73]. Consistent with this finding, overexpression of
LRP6 is also found in triple negative breast cancers. When LRP6 is knocked-down in triple negative breast cancer cells, tumor growth is suppressed in vivo
[78][74]. LRP’s also seem to have a promoting role in breast cancer metastasis to bone. TM40D-MB breast cells which are highly metastatic to bone have higher mRNA levels of LRP5, LRP6 and β-catenin, as compared to TM40D cells, which are breast cancer cells that are non-bone metastatic
[79][75]. However, contrary to previous studies, Ren et al. reported the novel role of
LRP5/6 as suppressing metastasis. This study showed that
LRP5/6 downregulation is crucial for metastasis in mouse breast cancer models
[80][76], suggesting that the binding of LRP5/6 to FZD inhibits the FZD-regulated non-canonical pathway and its further tumor metastasis.
Therapeutic targeting of LRP5/6 shows promising results. Mesd, a specialized chaperone that binds to LRP5/6 on the cell surface, inhibits LRP5/6 ligands and suppresses downstream WNT/β-catenin signaling in prostate cancer
[81][77]. Niclosamide, another inhibitor targeting LRP6 on the cell surface, also induces cancer cell apoptosis
[81][77]. Future studies can aim to discover new therapeutic approaches targeting LRP5/6 and other members in the LDLR family. Other LDLR family members like LRP8 and LRP10 are also good potential targets to be further investigated given that previous studies have showed their involvement in breast cancers and hepatocellular carcinomas, respectively
[82,83][78][79].
WNT 5A: The WNT5A ligand binds to certain receptors (e.g., ROR2, ROR1, etc.) and activates non-canonical WNT signaling pathways
[84][80]. Mutations of WNT5A have been shown to associate with cancer development, possibly through noncanonical WNT signaling pathways
[84][80]. For example, WNT5A promotes cancer through binding its receptor ROR2 and enhancing human osteosarcoma invasiveness
[85][81].
WNT5A expression induces EMT in a PKC-dependent pathway in human melanoma cell lines
[86][82], suggesting that the increased ability of cells to undergo EMT promotes their invasiveness. Kanzawa et al. reported that the expression of
WNT5A in human gastric carcinoma-derived MKN-7 cells promoted cancer cell invasiveness by upregulating a transcription factor involved in EMT, namely SNAI1. This study also suggested that
WNT5A may play a role in creating favorable tumor microenvironments and inducing cancer stem cell properties
[87][83]. Aside from its role of promoting cell invasion,
WNT5A drives pseudo-senescence in melanoma cancer cells. When exposed to stress, pseudo-senescent cells display classical senescence markers, but are highly invasive, metastatic and resistant to therapy. In response to stressors, tumor cells with knocked down
WNT5A do not display this pseudo-senescent phenotype, whereas highly expressed
WNT5A tumor cells do, suggesting WNT5A’s role in promoting cancer cell adaptiveness under stress
[88][84]. WNT5A has also been shown to be involved with cancer cell metabolism and inflammation
[89][85]. Future studies should investigate other WNT ligands (e.g., WNT11) and their roles in cancer progression through non-canonical pathways with the hope of providing novel targets for therapies.
RNF43: Ring finger protein 43 (RNF43) is an E3 ubiquitin-protein ligase that degrades WNT receptors such as FZD and LRP5/6. It serves as a negative regulator of the WNT pathway. The mechanisms of WNT signaling suppression are different between non-canonical and canonical pathways. In non-canonical pathways, the suppression involves the interaction between RNF43′s C-terminal cytoplasmic region and Dsh’s PDZ domain. Suppression of the canonical pathway involves FZD’s extracellular protease-associated domain, cysteine-rich domain and the intracellular ring finger domain of RNF43
[90][86]. A recent study has suggested a novel way that RNF43 suppresses the pathway in the nucleus where a physical interaction between RNF43 and TCF4 translocates TCF4 to the nuclear membrane, inhibiting TCF4 transcriptional activity
[91][87].
RNF43 acts as a tumor suppressor. When 185 human colorectal tumor samples were analyzed, there were
RNF43 somatic mutations in over 18% of colorectal adenocarcinomas and endometrial carcinomas. This study also showed that most
RNF43 mutations (73%–75% of non-silent mutations) found were truncating, confirming that
RNF43 acts as a tumor suppressor
[92][88].
RNF43 mutations were also found in pancreatic cancers
[93][89] and mucinous ovarian carcinomas
[94][90]. However, few studies have been conducted to explore the role of
RNF43 in these cancers, so further studies are needed to investigate how certain mutations in
RNF43 can possibly lead to the development of these cancers.
TCF4/TCF7L2: T-cell factor/lymphoid enhancer-binding factor (TCF/LEF) transcription factors are the downstream effectors of the WNT pathway. When WNT ligands bind to upstream receptors, β-catenin is released from the cytoplasmic destruction complex and moves to the nucleus, where it associates with TCFs to regulate the transcription of target genes
[95][91].
TCF4, one of the most studied members of the TCL/LEF family, seems to have contradictory roles as both tumor-promoting and tumor-suppressing. TCF4 was originally believed to be tumor-promoting. Van De Wetering et al. demonstrated that dominant-negative TCF4, which fails to bind to β-catenin, halted cell proliferation in colorectal cells
[96][92]. However, Angus-Hill et al. reported TCF4 as tumor-suppressing as its loss of function increased proliferation in colon tumors in mice
[97][93]. This controversy may be resolved by considering that TCL/LEF undergoes alternative splicing, and the isoforms produced may differ in functional domain composition. Thus, these TCL/LEF isoforms will differ in DNA binding and target gene activation
[98][94]. To illustrate this, Tsedensodum et al. identified 14 TCF4 isoforms from four hepatocellular carcinoma cell lines. Functional analysis showed that one isoform, TCF-4K is tumor growth promotive, and another isoform, TCF-4J, is tumor growth suppressive. Strikingly, these two isoforms only differ by five amino acids
[99][95]. These data suggest that the variants caused by alternative splicing may be a reason to explain the contradictory role of TCF4
[100][96]. Further studies should elucidate the roles of TCF in cancer and development
[101][97], taking into consideration not only the downstream target genes’ roles, but also the different isoforms that result from alternative splicing. Based on this understanding, more effective therapeutic approaches can be developed to aim for certain isoforms that cause malignant phenotypes.
PTK7: Another gene linking the various types of WNT signaling to cancer is the gene
PTK7, an orphan receptor linked to a variety of cancers as a highly upregulated biomarker
[102][98]. PTK7 was originally discovered as a colon carcinoma kinase-4 upregulated in cancer tissue
[103][99]. Its function remained unclear until the ortholog was discovered as a neuronal pathfinding gene in
Drosophila (known as off-track in flies)
[104][100] and subsequently shown to be a planar polarity gene during mouse development
[105][101]. The mechanism of action remains controversial with WNT and non-canonical functions proposed
[105[101][102][103][104],
106,107,108], but the use of PTK7 in cancer treatment could well be under way
[109][105].
WNT pathway as a tumor suppressor: Although WNT is usually referred to as an oncogenic pathway, it can also function in quite the opposite direction depending on cellular context. This has primarily been studied in malignant melanoma, where activation of WNT reduces cell proliferation and can actually function as a tumor suppressor. Overall, the main lesson from these studies is that context is crucial, namely depending on what transforming factor is driving oncogenicity since WNT activation can lead to unexpected consequences
[110,111,112][106][107][108].