• Wnt signalling;
  • β-catenin;
  • tumourigenesis


  1. Top of page
  2. Abstract
  3. Introduction
  4. The Wnt signalling pathway
  5. The Wnt signalling pathway in cancer
  6. Future perspectives
  7. References

Abnormalities in the Wnt signalling pathway are found in a wide range of cancers. The diverse origin of these malignancies implies that the contribution that disrupted Wnt signalling makes to tumourigenesis is not limited to specific tissue types and thus can be regarded as a step which is ‘generic’ to the process of carcinogenesis. In recent years, rapid progress has been made in the understanding of the Wnt signalling pathway, giving an insight into how inappropriate activation of this pathway may facilitate the neoplastic conversion of a normal cell. Furthermore, elucidation of the mechanisms that regulate Wnt signalling has led to the possibility of manipulating these mechanisms in order to down-regulate Wnt signalling in established tumours. In this review, the Wnt signalling pathway is described. The role of aberrant Wnt signalling in tumour development is discussed together with its clinical implications for anti-tumour therapy. Copyright © 2005 Pathological Society of Great Britain and Ireland. Published by John Wiley & Sons, Ltd.


  1. Top of page
  2. Abstract
  3. Introduction
  4. The Wnt signalling pathway
  5. The Wnt signalling pathway in cancer
  6. Future perspectives
  7. References

Cancers arise from normal tissue due to the sequential accumulation of somatic gene mutations 1. In most instances, mutations must occur in at least four to five key genes before malignant change occurs 2. The mutations that occur and the order in which they occur represent the genetic pathway of tumour development. Although the genetic pathway will be unique for each individual tumour, there will be a great deal of overlap in cancers of similar origin. For example, in most cases of colorectal cancer, APC gene mutations will occur as the first step, whilst mutations of the TP53 gene will occur as a late event 3. Cancers arising in different tissues will have distinct genetic pathways, although there are some steps which seem to be shared amongst a wide range of tumours. Such steps, which are not limited to specific types of cancer, can be regarded as steps that are ‘generic’ to the carcinogenetic process. Thus, mutation of the TP53 gene is found in almost half of all cancers and occurs in both epithelial and mesenchymal malignancies 4, 5. Similarly, aberrant Wnt signalling occurs in cancers of diverse origin (Table 1) and as such, represents an important generic step in tumour development.

Table 1. Reported alterations in the Wnt signalling pathway in diverse tumours (with selected references). This table shows the mechanisms by which Wnt signalling may be activated and the diversity of tumours in which aberrant Wnt signalling is reported
Wnt pathway componentChangeCancer typeReference
  1. NSCLC = non-small cell lung cancer; HNSCC = head and neck squamous cell carcinoma; CLL = chronic lymphatic leukaemia; BCC = basal cell carcinoma; GOF = gain-of-function; LOF = loss-of-function.

Gastric cancer133, 134
sFRPsDown-regulationBreast139, 140
Colorectal125, 126, 143
CTNNB1GOF mutationGastric154, 155
Colorectal106, 156
Intestinal carcinoid157
Ovarian158, 159
Pulmonary adenocarcinoma160
Hepatocellular164, 165
Thyroid169, 170
Prostate171, 172
Wilms' tumour175
Juvenile nasopharyngeal angiofibromas178
Desmoid179, 180
Synovial sarcoma181
APCLOF mutationColorectal3, 109
Melanoma173, 182
Medulloblastoma167, 183
Desmoid179, 184
AxinLOF mutationHepatocellular164, 185
medulloblastoma186, 187

The Wnt signalling pathway

  1. Top of page
  2. Abstract
  3. Introduction
  4. The Wnt signalling pathway
  5. The Wnt signalling pathway in cancer
  6. Future perspectives
  7. References

The Wnt signalling pathway was first implicated in carcinogenesis over 20 years ago in viral carcinogenesis experiments conducted in mice. A common site of viral integration in mouse mammary tumour virus (MMTV)-induced mammary tumours was identified 6. The site of integration was in the promoter of a gene which was called Int-1 (for ‘integration’) and resulted in increased production of full-length Int-1 protein 7, 8. Forced expression of Int-1 protein in transgenic mice caused the development of mammary tumours, confirming a causative role for this protein in mammary tumourigenesis 9. Sequence analysis showed that Int-1 was orthologous to the Drosophila segment polarity gene Wingless (Wg) and the terms were combined to produce the name ‘Wnt’ for the mammalian Int-1 gene and its paralogs 10, 11.

Since the original identification of the Int-1 gene (now known as WNT1), the WNT gene family has grown and currently comprises at least 19 members. They show varying degrees of sequence identity but all contain 23–24 cysteine residues which show highly conserved spacing 12, 13. The Wnt proteins are small (39–46 kD) lipid-modified secreted glycoproteins which play key roles in both embryogenesis and mature tissues. During embryological development, the expression of Wnt proteins is important in patterning through control of cell proliferation and determination of stem cell fate. Alterations of Wnt signalling in animal models lead to abnormal morphogenesis 14–16 and, in humans, germline mutations of genes involved in the Wnt signalling pathway lead to congenital defects 17–19. In mature tissues, Wnts are involved in the self-renewal of stem cells and may be responsible for the maintenance of many normal tissues 20–23.

Wnt proteins act as ligands for the Frizzled family of seven-pass transmembrane receptors 24–26. There are ten known members of this family and these are characterized by the presence of a cysteine-rich domain (CRD) 27. On binding to Frizzled receptors, Wnts can activate one of three different pathways: (a) the canonical Wnt signalling pathway (which results in stabilization and increased transcriptional activity of β-catenin 28, 29); (b) the Wnt/planar cell polarity (Wnt/PCP) pathway 30; or (c) the Wnt/calcium (Wnt/Ca2+) pathway 30. Although the Wnt and Frizzled proteins can be loosely classified in accordance with the pathway that they activate, this is not rigid as there is a degree of promiscuity in the Wnt–Frizzled interactions. Each of the pathways, although distinct, appears to be transduced initially through the cytoplasmic protein Dishevelled 31–33 and the ultimate response to Wnt–Frizzled interaction will most probably depend on cellular context at that time.

(i) Canonical Wnt signalling (Figure 1)

thumbnail image

Figure 1. A schematic representation of the canonical Wnt signalling pathway and the mechanisms by which it is regulated. See main text for details

Download figure to PowerPoint

The canonical Wnt signalling pathway is the most relevant of the Wnt signalling pathways to the development of cancer. Activation of this pathway sets into motion a series of downstream events culminating in the stabilization and increase in the levels of β-catenin protein. This is normally found in the cytoplasm but, on activation of the canonical pathway, it will translocate into the nucleus, where it acts as a transcription factor to activate transcription of a number of target genes. These genes are the ultimate ‘effecters’ of neoplastic change and maintained activity of these effecter genes is the cause of tumour development. It is thus not surprising that considerable effort is invested in ensuring that canonical Wnt signalling is tightly controlled. These methods of regulation are discussed below, but for the sake of simplicity, the pathway is described as being in either an ‘Off’ or an ‘On’ state of activity. This state is maintained through molecules acting as ‘Off’ or ‘On’ switches and, as will become apparent later, tumour development occurs due to either inappropriate inactivation of ‘Off’ switches or inappropriate activation of ‘On’ switches.

(a) The ‘Off’ state
Extracellular inhibition

In order to ensure that Wnt signalling is activated only when required, inhibitory mechanisms have been developed which act at several different levels. In the extracellular milieu, secreted proteins can act to sequester Wnt ligand from its receptor. Amongst this group are secreted Frizzled related proteins (sFRPs 34), Wnt inhibitory factor-1 (WIF-1 35), Cerberus 36, and Coco 37. In humans, the sFRP family consists of five members, each containing a cysteine-rich domain (CRD) which shares 30–50% sequence homology with the CRD of Frizzled receptors 38. As well as complexing with Wnt proteins, sFRPs are thought to form function-inhibiting complexes with Frizzled receptors. The biology of sFRPs is, however, complex and in some cases, they may act as Wnt agonists 39. WIF-1 does not have any sequence homology with sFRPs, but contains a unique evolutionarily conserved WIF domain together with five epidermal growth factor (EGF)-like repeats. Cerberus and Coco are proteins related to each other which can complex with a variety of different growth factors including Wnt and bone morphogenetic protein (BMP) to inhibit signalling of the respective pathways in Xenopus. However, inhibition of Wnt signalling by the mammalian orthologs of Cerberus and Coco has not yet been shown.

Another class of extracellular Wnt inhibitor is represented by the Dickkopf (Dkk) family of secreted proteins 40, 41. Three members of the Dkk family (Dkk-1, -2, and -4) can antagonize Wnt signalling through inactivation of the surface receptors LRP5 and LRP6 (which are essential for activation of the canonical pathway 42, 43). The Dkks form a ternary complex with LRP5/6 and the single pass transmembrane receptors Kremen 1 or Kremen 2 44–46. This complex undergoes endocytosis, thereby removing LRP5/6 receptors from the cell surface. By targeting the LRP5/6 receptors specifically, Dkks can function as antagonists of canonical signalling only. In contrast, the sFRP group, by targeting Wnts, can inhibit both the canonical and the non-canonical pathway.

Phosphorylation-dependent intracellular β-catenin degradation

Within the cell, the inhibitory mechanisms act to ensure that the levels of β-catenin stay below the threshold beyond which aberrant transcriptional activity will occur. β-Catenin is constitutively produced and is present in the cytoplasm as pools of monomeric protein 29. The main mechanism for controlling cytoplasmic β-catenin levels is through direct physical destruction of the protein after recruitment to a large multi-protein complex (Figures 1 and 2). Axin1 (or its homologue Axin2) forms the central scaffold of this complex 47 and provides binding sites for β-catenin, adenomatous polyposis coli (APC), glycogen synthase kinase 3β (GSK3β), casein kinase Iα (CK Iα), and protein phosphatase 2A (PP2A) 48–52. APC also has binding sites for β-catenin and PP2A 53, 54, but GSK3β, in contrast to the other members of this complex, only has binding sites for Axin and thus cannot physically associate with either APC or β-catenin 55. Once the complex is formed, it is stabilized by the GSK3β-mediated phosphorylation of Axin and APC. The complex is further stabilized by the activity of PP2A when it is in a trimeric form which contains the B56 subunit 56.

thumbnail image

Figure 2. Activation of Wnt signalling through mutation of β-catenin or APC. The β-catenin protein is targeted for degradation by phosphorylation of a GSK3β target motif in the N-terminus (between residues 33 and 45). A large multi-protein complex is essential for the phosphorylation to occur. β-Catenin must be initially primed by phosphorylation of the serine 45 residue by CK Iα. This is then followed by sequential phosphorylation of residues threonine 41, serine 37, and serine 33. The phosphorylated protein is then recognized by βTrCP and poly-ubiquinated. Failure of any one of the four residues to be phosphorylated results in failure to be recognized by βTrCP. In the example shown, mutation of serine 45 results in failed priming of the protein, thereby preventing phosphorylation of the remaining residues. The overwhelming majority of mutations reported in CTNNB1 result in the loss of one of these residues or in-frame deletion of the GSK3β target motif. An alternative method of activating Wnt signalling is through APC mutation. Most mutations result in truncation of the protein just proximal to the Axin binding domain, which then causes destabilization of the multi-protein complex and a failure to phosphorylate β-catenin. Similarly, AXIN mutations are generally truncating mutations causing destabilization of the large complex

Download figure to PowerPoint

Within the stabilized complex, GSK3β functions to phosphorylate the N-terminus of β-catenin. There is a GSK3β recognition motif between amino acids 33 and 45 of β-catenin wherein lie four serine/threonine residues which are targeted for phosphorylation. In order for GSK3β-mediated phosphorylation to proceed, the motif must be primed by phosphorylation of the first amino acid—serine-45—by CK Iα. Thereafter the remaining target residues—threonine-41, serine-37, and serine-33—are sequentially phosphorylated by GSK3β 57. Thus phosphorylated, β-catenin is recognized by β-transducin repeat containing protein (β-TrCP) as a protein which is to be ubiquitinated 58, 59. β-TrCP is an F-box containing protein which, together with Skp1, Cullen, and Rbx-1, constitutes the enzyme ubiquitin ligase (E3). This, together with ubiquitin conjugating enzyme (E2) and ubiquitin activation enzyme (E1), causes ubiquitination of β-catenin, which is then destroyed by the proteasome system 60. All four serine/threonine residues in the GSK3β recognition motif of β-catenin must be phosphorylated in order to be recognized by β-TrCP; failure to phosphorylate even one of the residues will result in failure to ubiquitinate and degrade β-catenin.

Phosphorylation-independent intracellular β-catenin degradation

Although Axin/APC/GSK3β-mediated destruction is the main mechanism for controlling cytoplasmic levels of β-catenin, another degradation pathway has been recently identified which does not depend on N-terminal phosphorylation of β-catenin. Siah1 is induced by p53 and is able to complex with both β-catenin and the C-terminus of APC 61, 62. Siah1 can also associate with a ubiquitin E2 conjugating enzyme and a ubiquitin E3 ligase. The E3 ligase consists of the F-box protein EBI together with Skp1 and SIP. Whilst EBI associates with β-catenin, SIP associates with Siah1 and the whole complex can cause ubiquitination of β-catenin in the absence of GSK3β-mediated phosphorylation. Induction of Siah1 (or indeed p53) causes a reduction in the levels of cytoplasmic β-catenin 63. The precise role of this method of inhibition of Wnt signalling is not understood, although, from first principles, one would assume that it would involve inhibition of cell-cycle progression in the presence of DNA damage.

Sequestration of β-catenin at the cell membrane

In addition to its role as a transcription factor, β-catenin is also involved in the control of cellular adhesion 64, 65. β-Catenin can be found at the cell surface at sites of intercellular contact known as adherens junctions, where it is in complex with E-cadherin and α-catenin. Part of the cytoplasmic pool of β-catenin will thus be directed to the cell membrane, making it unavailable for translocation to the nucleus. An increase in surface E-cadherin expression will deplete cytoplasmic levels of free β-catenin and thereby inhibit Wnt signalling. Mice containing a mutant β-catenin transgene have been shown to up-regulate E-cadherin 66 expression, which presumably represents a feedback mechanism to control levels of cytoplasmic β-catenin. Events which cause breakdown of the E-cadherin–catenin complex can cause an increase in the level of free cytoplasmic β-catenin and consequent transcriptional activity 64. This shows that there is a dynamic interaction between the cytoplasmic pool and the E-cadherin associated pool, with protein able to shuttle back and forth between the two pools.

Nuclear events in the ‘Off’ state

In the ‘Off’ state of the canonical Wnt signalling pathway, β-catenin should not translocate to the nucleus and thus its target genes are not activated. β-Catenin cannot physically bind to DNA and in order to activate transcription, it must complex in the nucleus with members of the LEF/TCF family of high mobility group (HMG) proteins 67, 68. The LEF/TCF family consists of four proteins (LEF1, TCF1, TCF3, and TCF4) which complex with DNA at the heptameric consensus motif (A/T)(A/T)CAA(A/T)G 69. These proteins serve to provide a DNA binding domain for β-catenin. In the absence of nuclear β-catenin, the LEF/TCF proteins are found in complex with the transcriptional repressor Groucho 70, 71. Other transcriptional repressors, such as histone deacetylases, are also recruited to the complex, thereby ensuring that the target genes are not activated. Should β-catenin inappropriately enter the nucleus, another level of inhibition lies in factors such as the recently described Chibby 72, which can compete with β-catenin for the LEF/TCF proteins.

The role of other signalling pathways

The Wnt signalling pathway is not alone in governing embryogenesis and tissue maintenance and it is likely that a dynamic equilibrium exists between the various signalling pathways, which will interact with each other to ensure that the final outcome is correct. It has been shown that in some cases, non-canonical Wnt signalling itself can inhibit canonical Wnt signalling 73. Similarly, the bone morphogenic protein (BMP) and Hedgehog signalling have been shown to be able to inhibit Wnt signalling in certain circumstances 20, 74.

(b) The ‘On’ state
Activation through Wnt proteins

All Wnt signalling pathways are initiated by interaction of Wnt ligands with Frizzled receptors. However, the canonical Wnt signalling pathway will only be activated if this interaction takes place in the presence of the low-density lipoprotein receptor related protein (LRP) 5 or LRP 6 42, 43. The formation of the trimolecular complex (Wnt–Frizzled–LRP5/6) has two consequences. Firstly, Dishevelled is recruited (through an unknown mechanism) to the cell surface and phosphorylated by casein kinase Iε (CK Iε) 75. The phosphorylated Dishevelled protein can form a complex with Frat1 and GSK3β which serves to inhibit the activity of GSK3β 75, 76. Secondly, the Wnt–Frizzled–LRP5/6 interaction facilitates the LRP5/6-mediated degradation of Axin 42. The net effect is destabilization and functional failure of the complex responsible for phosphorylating β-catenin. In the absence of phosphorylation, β-catenin is not ubiquitinated and thereby escapes proteasomal degradation. The result of this is an increase in the level of β-catenin protein and its translocation to the nucleus.

The mechanism by which β-catenin translocates into the nucleus is not completely clear, as it does not contain a nuclear localization signal and thus may be transported by other proteins. The β-importin/karyopherin system for nuclear transportation does not appear to be essential since inhibition of this system will not prevent nuclear localization of β-catenin 77. APC protein has been shown to shuttle in and out of the nucleus and it has been suggested that this may be another method of transporting β-catenin to the nucleus 78–80. More recently, it has been shown that the proteins pygopus and Bcl9/legless can form a complex with β-catenin in the cytoplasm and, due to the inherent nuclear localization activity of pygopus, the complex can translocate into the nucleus 81.

Other mechanisms of activating canonical Wnt signalling

The amount of available β-catenin for target gene activation can also be increased by shifting protein from the cadherin-bound pool to the cytoplasmic pool. A number of receptor tyrosine kinases (RTK), on binding of their cognate ligand, are able to phosphorylate β-catenin at tyrosine residues, which causes dissociation of β-catenin from the cadherin–catenin complex. The β-catenin protein thus liberated can be recycled to the cytoplasmic pool, which is then followed by increased expression of β-catenin target genes. Thus, surface receptors, such as c-RON, epidermal growth factor receptor (EGFR) and c-ErbB2, can stimulate canonical Wnt signalling 82–84 as well as activating the signalling pathways with which they are more usually associated.

Just as other signalling pathways can inhibit Wnt signalling, they are also able to activate Wnt signalling or facilitate the effects of Wnt signalling 85. Integrin signalling, through integrin-linked kinase, can cause nuclear localization of β-catenin. The mechanisms by which the other signalling pathways achieve their effects are uncertain, although it will most likely be through the manipulation of the various ‘On’ and ‘Off’ switches. For example, GSK3β is integral to the insulin-like growth factor (IGF) signalling pathway, and IGF1 and -2 can activate Wnt signalling through inhibition of GSK3β 86–90.

Finally, certain oncogenic viruses are associated with the development of malignancy. The Epstein–Barr virus (EBV) is associated with both epithelial and lymphoid malignancies. It has been shown that EBV can activate canonical Wnt signalling through GSK3β inhibition 91, 92.

Nuclear events in the ‘On’ state

Once within the nucleus, β-catenin can compete with Groucho for binding with the LEF/TCF proteins. The mechanism by which competition occurs is not clear and is thought to be more than just direct concentration-dependent competition. The LEF/TCF proteins allow β-catenin to bind to the DNA, where it forms the basis of a large complex for activating transcription. The complex includes the essential co-factors pygopus and Bcl9/legless 93 together with a large array of other proteins [such as p300/Creb binding protein (CBP)] which allow specific target genes to be transcribed 69, 94.

There is considerable interest in the β-catenin target genes, as aberrant expression of these mediates the neoplastic conversion of normal cells. The list of β-catenin target genes is constantly growing and the reader is referred to the Wnt genes home page for a comprehensive and up-to-date list (∼rnusse/wntwindow.html). Recent studies using microarrays have identified a large number of genes which undergo altered expression after disruption of Wnt signalling 95–99. Since transcription factors (such as c-myc, c-Jun, and Sox9) are included amongst the β-catenin targets 100–102, they will in turn alter the expression of their own target genes. Some of the genes identified in the microarray studies may therefore not be true β-catenin targets; rather their altered expression may represent a secondary consequence of aberrant Wnt signalling.

The panoply of genes that are deregulated, whether as primary or secondary consequences of aberrant Wnt signalling, cause disturbance of a variety of different cellular processes. Therein probably lies the reason why aberrant Wnt signalling is germane to the development of so many different types of tumours. Each cell type is under the control of a number of different growth constraints (both intrinsic and environmental). Disruption of multiple cellular processes will allow cells to overcome many of these constraints and become neoplastic.

(ii) Non-canonical Wnt signalling

For the sake of completeness, the non-canonical Wnt signalling pathways are briefly described. The role of these pathways in cancer development is uncertain; in some instances, they can antagonize the canonical Wnt signalling pathway 73, whilst there is evidence that Wnts acting through the non-canonical pathway can promote tumour progression 103, 104. The Wnt/planar cell polarity (Wnt/PCP) pathway is thought to regulate cell polarity through altering the cytoskeleton. It is mediated through activation of the c-Jun N-terminal kinase (JNK) pathway and is thought to play an important role during gastrulation in vertebrates 30. The Wnt/calcium (Wnt/Ca2+) pathway is activated through a heterotrimeric G protein and results in an increase in intracellular calcium and activation of protein kinase C (PKC) 30. The function of this pathway is as yet unknown.

The Wnt signalling pathway in cancer

  1. Top of page
  2. Abstract
  3. Introduction
  4. The Wnt signalling pathway
  5. The Wnt signalling pathway in cancer
  6. Future perspectives
  7. References

(i) Genetic/epigenetic activation of canonical Wnt signalling (Figure 2, Table 1)

The canonical Wnt signalling pathway is regulated by several switches serving to ensure that levels of β-catenin protein are commensurate with the requirements of the cell at any particular moment in time. It would be expected that disruption of any of these switches could alter the equilibrium and result in the stabilization of and increase in β-catenin protein levels. Table 1 shows that this is indeed the case and that canonical Wnt signalling can be activated by changes occurring at any level of the pathway.

Mutations of the β-catenin gene (CTNNB1) itself occur frequently in many types of cancer. Analysis of the spectrum of CTNNB1 mutations shows that a wide variety of changes are reported including missense mutations and small in-frame deletion. Almost all reported mutations occur in exon 3 and specifically cause loss of the GSK3β recognition motif. Since phosphorylation of β-catenin by GSK3β is essential for ubiquitination and degradation of β-catenin, the effect of the mutations is to produce a protein which is transcriptionally active but which cannot be degraded by the APC–Axin complex. This in turn causes an increase in protein levels, leaving the Wnt signalling pathway in a constant ‘On’ position 105, 106. Increased expression of other ‘On’ switches, such as Frizzled and LRP5 surface receptors, together with a variety of Wnt proteins has also been reported, although mutations of these have not been hitherto described.

Molecules that normally inhibit Wnt signalling are inactivated through either loss-of-function mutations or epigenetic silencing. Mutations are reported in APC and AXIN1/2. Both of these are integral to the β-catenin phosphorylation complex and, not surprisingly, the majority of mutations that occur are protein truncating mutations which cause destabilization of the complex. Analysis of the location of APC mutations shows that there is a very well-defined 3′ boundary in the APC gene which is at the site of the Axin binding region of APC 107, 108. Almost all somatic APC mutations will therefore result in a truncated protein that has lost the Axin binding sites, whilst mutations beyond this boundary—which would produce a truncated protein that includes the Axin binding domain—are extremely rare. This shows that loss of β-catenin regulation is the major selective drive for APC mutations.

The sFRPs, which would also inhibit Wnt signalling, have been shown to be inhibited in tumours by promoter hypermethylation. Reactivation of sFRPs with demethylating agents has an inhibitory effect on tumour cells.

(ii) Wnt signalling and cancers: one size does not fit all

Although many cancers harbour CTNNB1 mutations, there are some differences between tumours with regards to the preferred mechanism for activating Wnt signalling. For example, around 80% of colorectal tumours have APC gene mutation 109, whilst approximately 12% contain CTNNB1 mutations 110. The predominance of APC mutation over CTNNB1 mutation in these tumours is puzzling, since APC is a tumour suppressor requiring at least two mutagenic events for complete loss of APC activity, whilst, in contrast, only a single mutation of CTNNB1 is required for activation of β-catenin. Furthermore, it has been shown that, in mice at least, mutation of the β-catenin gene alone is sufficient for adenoma development in the intestine 111. Part of the explanation lies in the fact that APC is a large multifunctional protein and mutation will lead to the loss of other functions in addition to that of controlling β-catenin levels 107, 108. The loss of these other APC functions obviously gives a selective advantage over and above that found from CTNNB1 gene mutation alone. The fact that APC mutations are rarely found in other tumour types implies that these advantages are restricted to the colon. Thus, the mutation of genes involved in regulating Wnt signalling may, as well as activating Wnt signalling, also have other non-Wnt signalling-related effects.

Activation of Wnt signalling (through mutation of either AXIN1/2 or CTNNB1) is common in human hepatocellular carcinomas. However, inducing mutation of the β-catenin gene in the mouse liver results in epithelial hyperplasia only, whilst, in contrast, mutation of the β-catenin gene in the mouse intestine is sufficient to cause intestinal neoplasia 111, 112. The differences in the effect of β-catenin gene mutation in the liver and intestine show that the effect of aberrant Wnt signalling will not be the same in all cell types. Some of the β-catenin target genes will also be regulated by other pathways; therefore, genes which are deregulated in one cell type may not be altered in another cell type where other mechanisms may maintain tight transcriptional inhibition. It follows from this that although aberrant Wnt signalling may make a significant contribution to different types of cancer, the precise contribution that it makes may differ between tumour types.

(iii) The mechanics of tumourigenesis: subverting normal processes to cause neoplastic change (Figure 3)

thumbnail image

Figure 3. The mechanistic basis of neoplastic change induced by aberrant Wnt signalling (adapted from Hanahan and Weinberg 113). Inappropriate elevation of β-catenin levels results in constant expression or inhibition of a large number of genes. These genes affect a large number of processes which can contribute to most of the features that characterize a malignant tumour. A small selection of deregulated genes is shown (up-regulated in red; down-regulated in green; reference in parentheses) together with an indication of some of the features that they may influence. Single target genes may influence a variety of processes. For example, matrix metalloproteinases (MMPs), through digestion of extracellular matrix, can facilitate tumour invasion. In addition, through cleavage of osteopontin, MMPs may facilitate cell migration and angiogenesis. The extensive range of effects resulting from aberrant Wnt signalling explains why canonical Wnt signalling is disrupted so frequently and in so many different types of cancer

Download figure to PowerPoint

Whichever mechanism is used by any particular tumour to activate canonical Wnt signalling, the net result is a maintained increase in the level and transcriptional activity of β-catenin protein. The induced β-catenin target genes are not mutated and therefore execute their normal functions, albeit in an inappropriate situation. The expression of the target genes is unremitting and if the usual compensatory mechanisms cannot cope with the changes, the result is derangement of the normal processes which control a cell's fate and, consequently, escape from growth constraints. This raises the question of which processes are actually deranged by Wnt signalling to facilitate cancer development. Hanahan and Weinberg identified a number of features (or ‘hallmarks’) which are acquired by most malignancies 113. Subsumed within each feature are a number of functions and biological processes that need to be modulated in order for that feature to emerge. Analysis of the genes modulated by Wnt signalling (either as bona fide β-catenin target genes or as genes showing altered expression by microarray analysis) shows that manifold processes are affected by these genes and that these can contribute to nearly all of the features identified by Hanahan and Weinberg. A modified version of the Hanahan and Weinberg model is shown in Figure 3, together with a small selection of genes showing altered expression upon induction of Wnt signalling.

‘Limitless replicative potential’ is one of the hallmarks attributed to cancer, although this has been modified in Figure 3 to ‘Inappropriate stem cell phenotype’. The latter term is preferred by the author as it assumes ‘limitless replicative potential’ but also takes cognisance of the fact that most tumours undergo differentiation towards the tissue of origin. Wnt signalling is thought to be essential for maintenance and self-renewal of stem cells 20, 21, 23. In the intestine, TCF4 is the main nuclear binding factor for β-catenin and the failure of Tcf4 knockout mice to develop stem cells in the small intestine further supports the role of canonical Wnt signalling in stem cell maintenance 114. A number of other genes stimulating the cell cycle (such as cyclin D1 115) are also induced by β-catenin and these contribute to the process of cell proliferation, which is a characteristic of the stem cell phenotype.

The effect of Wnt signalling on multiple biological processes is exemplified by the matrix metalloproteinase (MMP) genes. Three different genes—MMP7, MMP14, and MMP26—have been shown to be direct targets of β-catenin 116–118, whilst several Mmp genes have been shown to be expressed in intestinal adenomas in mice 96. The MMPs are proteolytic enzymes which can breakdown stromal collagen to allow tumour cells to acquire the feature of ‘Tissue invasion and metastasis’. The enzymatic activity also allows the release of latent growth factors in the stroma, which, together with growth factors directly secreted by tumour cells and up-regulation of growth factor receptors, will contribute to ‘Self-sufficiency of growth signals’ 119, 120. MMPs can also act on osteopontin (a secondary Wnt-induced target 96) to release fragments which together with vascular endothelial growth factor (VEGF—a direct β-catenin target) will contribute to the feature of ‘Sustained angiogenesis’ 121, 122.

There are also several genes transcriptionally altered by Wnt signalling which facilitate ‘Evasion of apoptosis’ and some of the down-regulated genes may result in ‘Insensitivity to growth inhibitors’ (Figure 3). Two additional features have been added to the model by the author. Firstly, there is ‘Failure of terminal differentiation’. Although the stem cells of most tumours still activate the genetic programmes causing differentiation towards the tissue of origin, from the earliest stages of tumour development the differentiation programme is not completed. Histologically, this is manifest as abnormal architecture and abnormal cellular morphology, and expression profiling studies have confirmed the down-regulation of genes associated with terminal differentiation. Recent evidence has shown that forced expression of one gene associated with goblet cell differentiation can reduce the tumourigenicity of a colorectal cancer cell line 123. This may be indicative of the importance of down-regulation of genes associated with terminal differentiation to tumour development.

The second additional feature is ‘Evasion of immune response’. Development of the cytotoxic immune response depends on the expression of MHC class I molecules. These are shown to be down-regulated in human cancers 124 and in response to aberrant Wnt signalling 96, and may contribute to a tumour's ability to escape the host defences.

(iv) Wnt signalling as a target for anti-cancer therapy

Although aberrant Wnt signalling alone can cause tumour development, it is unlikely that it will be sufficient for malignant change to occur. Aberrant Wnt signalling can make a contribution to most of the features necessary for malignancy in a tumour but other mutations are necessary for the features to emerge fully. Mutation of at least four to five key genes is thought to be necessary for a cancer to develop; thus, before targeting Wnt signalling for cancer treatment, it is essential to ascertain whether, in the context of the other mutations, tumour cells are still dependent on aberrant Wnt signalling for survival. In colorectal cancers, disruption of Wnt signalling is the first step to occur during tumour development, with several other mutations occurring subsequently. Some studies in colorectal cancer cell lines have shown that inhibition of Wnt signalling results in reduced tumourigenicity of the cell lines 125, 126. This shows that advanced cancers may still be dependent on aberrant Wnt signalling, even though it is the initiating event. Another study has, however, shown that somatic deletion of the β-catenin gene has little significant effect on a colorectal cancer cell line 127. Studies of lung tumourigenesis in mice have shown that tumours induced by forced overexpression of Wnts will regress on inhibition of Wnt signalling. If, however, a p53 mutation has occurred, the tumours become independent of Wnt signalling and inhibition of Wnt signalling is ineffective 128. One can conclude that targeting Wnt signalling is feasible, although, in some cases, tumours will be non-responsive despite the presence of mutations which activate Wnt signalling.

As more information emerges about the genes deregulated by aberrant Wnt signalling, these genes (many of which are known oncogenes) may also become potential targets for cancer therapy. Since a large number of genes are deregulated by aberrant Wnt signalling, it is important to determine which are the genes that are most important in the process of tumour development. For example, Neu transgenic mice are prone to developing mammary tumours. However, crossing Neu transgenic mice with cyclin D1 knockout mice abrogates mammary tumour development in the compound mutant progeny 129. This demonstrates that, in this model, cyclin D1 is absolutely essential to tumour development and as such, it is a good potential therapeutic target. However, Wnt1 transgenic mice also develop mammary tumours which, despite the fact that cyclin D1 is a well-described target of β-catenin, will develop in the absence of functional cyclin D1 129. This is also the case in intestinal tumours which develop in mice as a result of Apc mutation 130. In these tumours, therefore, cyclin D1 would be a poor therapeutic target. For effective therapeutic targeting of β-catenin-induced genes, it will be necessary to know which are the most important genes in each tumour type.

Future perspectives

  1. Top of page
  2. Abstract
  3. Introduction
  4. The Wnt signalling pathway
  5. The Wnt signalling pathway in cancer
  6. Future perspectives
  7. References

Wnt signalling is a very rapidly evolving and important field of investigation in cancer research. Expression profiling technology is allowing all the genes that are disrupted by aberrant Wnt signalling to be identified. The challenge will be to identify those genes and processes which are important to neoplastic change in each tumour type. As more is learnt about the mechanisms that regulate Wnt signalling activity, new therapeutic targets will emerge. Individual components of the Wnt signalling pathway could be targeted and antagonists of Wnt proteins are being developed. In addition, the genes that are induced by Wnt signalling may become therapeutic targets and it may become possible to manipulate other signalling pathways to inhibit Wnt signalling. As the β-catenin target genes which are important to cancer development are better delineated, it may be possible to identify function-altering polymorphisms in these genes. It may then be possible to discriminate between levels of risk of tumour development and even develop predictive screening strategies. One can be optimistic that further understanding of Wnt signalling will aid the battle against cancer.


  1. Top of page
  2. Abstract
  3. Introduction
  4. The Wnt signalling pathway
  5. The Wnt signalling pathway in cancer
  6. Future perspectives
  7. References