Wnt signaling in human and mouse breast cancer: Focusing on Wnt ligands, receptors and antagonists.

Wnt proteins, a group of secreted glycoproteins, mainly combine with receptors Frizzled (FZD) and/or low-density-lipoprotein receptor-related proteins 5/6 (LRP5/6), initiating β-catenin-dependent and -independent signaling pathways. These pathways, which can be regulated by some secreted antagonists such as secreted Frizzled-related proteins (SFRP) and dickkopf-related protein (DKK), play a critical role in embryo development and adult homeostasis. Overactivation of Wnt signaling has been implicated in some human diseases including cancer. Wnt transgenic mice provide convincing evidence that Wnt signaling is involved in breast cancer initiation and progression, which is further strengthened by observations on human clinical breast cancer patients and studies on in vitro cultured human breast cancer cells. This review focuses on the roles of Wnt ligands, receptors and antagonists in breast cancer development instead of molecules or signaling transactivating β-catenin independent on Wnt upstream components.

cancer‐associated fibroblasts

cysteine‐rich domain

cancer stem cells

dickkopf‐related protein

epithelial‐mesenchymal transition

estrogen receptor

Frizzled

glycogen synthase kinase 3 beta

human epidermal growth factor receptor 2

low‐density lipoprotein receptor‐related proteins 5/6

mammary stem cells

metastasis‐initiating cells

mouse mammary tumor virus

planar cell polarity

progesterone receptor

protein C receptor

R‐spondin

secreted Frizzled‐related proteins

α‐smooth muscle actin

T‐cell factor/lymphoid enhancer factor

transforming growth factor beta

triple‐negative breast cancer

Wnt inhibitory factor 1

INTRODUCTION

Since the discovery of the first Wnt gene, Wnt1 (initially named int‐1),1 19 human Wnt genes have been identified. These genes encode a group of highly conserved secreted glycoproteins, which are critical to embryo development and adult tissue homeostasis. FZD proteins are seven transmembrane receptors for Wnt ligands. There are 10 members in the FZD family, sharing a conserved extracellular CRD to which Wnt ligands bind. LRP5 and LRP6 are coreceptors for Wnt ligands. Both FZD and LRP5/6 are required for the Wnt/β‐catenin pathway.2

Conventionally, Wnt pathways are defined as canonical and noncanonical according to whether β‐catenin signaling is affected. Some members in the Wnt family such as Wnt1 and Wnt3a bind to FZD and LRP5/6, leading to the dissociation of the β‐catenin degrading complex in which β‐catenin is phosphorylated by GSK3β. Consequently, β‐catenin in the cytoplasm escapes from phosphorylation and subsequent degradation and translocates to the nucleus, where it combines with TCF/LEF, thereby promoting transcription of some target genes. Noncanonical Wnt pathways are independent of β‐catenin, and are usually initiated by Wnt5a and Wnt11. Until now, some secreted proteins have been found to antagonize Wnt signaling, including SFRP, DKK and WIF‐1.

TRANSGENIC MICE PROVIDE EVIDENCE FOR WNT‐DRIVEN BREAST CARCINOGENESIS

Mouse mammary tumor virus‐Wnt1 transgenic mice have provided solid evidence that Wnt signaling can initiate breast cancer. These mice were established by the insertion of MMTV‐LTR upstream of the gene Wnt1 in the opposite transcriptional orientation.3 MMTV‐Wnt1 mice show apparent ductal hyperplasia, and some of them can develop breast cancer as early as 6 months of age; histological, MMTV‐Wnt1 tumors show heterogeneous containing myoepithelial (basal‐like) cells and luminal epithelial cells.4 MMTV‐Wnt10b mice and MMTV‐LRP6 mice have a similar phenotype to MMTV‐Wnt1 mice.5, 6 Tumors from these mice show activation of the β‐catenin pathway (Figure 1). Consistently, active β‐catenin transgenic (MMTV‐βcatΔN) mice and MMTV‐c‐Myc mice also develop mammary hyperplasia and adenocarcinoma.4, 7

Figure 1

Mouse breast carcinogenesis induced by the activation of the canonical Wnt pathway. CK1, casein kinase 1; DVL, Dishevelled; GSK3β, glycogen synthase kinase 3 beta; LRP5/6, low‐density lipoprotein receptor‐related proteins 5/6; MMTV, mouse mammary tumor virus; TCF/LEF, T‐cell factor/lymphoid enhancer factor

WNT LIGANDS, RECEPTORS AND ANTAGONISTS ARE ABERRANTLY EXPRESSED IN HUMAN BREAST CANCERS

Although Wnt1 transgenic mice prove the capacity of the Wnt/β‐catenin pathway to initiate breast cancer, Wnt1 protein was hardly found overexpressed in human breast cancers.8 In contrast, Wnt10b is highly expressed in TNBC; Wnt10b activates the canonical β‐catenin pathway and contributes to increased cell proliferation and renewal.9 Wnt7b is expressed in several breast cancer cells and is overexpressed in approximately 10% of breast cancer patients.10 Moreover, Wnt receptors LRP6 and FZD7 are also overexpressed. LRP6 knockdown suppresses breast cancer cell growth, accompanied by a reduction in β‐catenin signaling activity.11 Similarly, FZD7 downregulation suppresses tumor formation in vivo as a result of inhibition of β‐catenin signaling.12 SFRP1 is expressed in normal breast epithelial cells but is frequently lost in invasive breast cancer tissues.13 Gene promoter methylation is responsible for SFRP1 expression loss and is correlated with unfavorable prognosis.14 Other SFRP such as SFRP2 and SFRP5 as well as DKK and WIF‐1 are also downregulated in breast cancer as a result of gene methylation.15, 16

WNT SIGNALING ENRICHES PROGENITOR CELLS/CANCER STEM CELLS IN BREAST CANCERS

Tumors from MMTV‐Wnt1 mice were shown containing mammary progenitor cells and CSC: also called tumor‐initiating cells, TIC), which can further differentiate into myoepithelial cells and luminal epithelial cells.4, 17, 18, 19 FZD7 knockdown in Wnt1 tumor cells reduces CSC subpopulation and tumor‐initiating capacity.20 Moreover, LRP5 deficiency delays Wnt1‐induced tumorigenesis accompanied by reduced progenitor cell accumulation.21 In contrast, a reduction in DKK1 or DKK3 promotes self‐renewal of progenitor cells/CSC by activating the β‐catenin pathway.22, 23

In human breast cancer, CD44+/highCD24−/low cells show stem‐like and high tumorigenicity.24 CD44+/highCD24−/low cells are enriched in basal‐like tumors25, 26 and show the EMT phenotype.27 Both canonical and noncanonical Wnt signalings are required for maintaining CD44+/highCD24−/low‐like cell EMT and stem phenotype.28 Moreover, DKK1 overexpression in breast cancer cells reduces CD44+/highCD24−/low subpopulation and inhibits tumorigenicity.29

Lgr5 has been identified as a stem cell marker in a series of organs.30 Subsequent studies further showed that Lgr5 is involved in the maintenance of stem cells. Binding of Lgr5 to Rspo sequesters E3 ubiquitin ligases RNF43 and ZNRF3 which ubiquitinate Wnt receptors FZD for degradation.31 Therefore, in the presence of Rspo and Wnt ligands, Lgr5 can potentiate Wnt signaling.32, 33 Actually, both Rspo and Wnt ligands are required for Lgr5+ stem cell renewal and to prevent them from differentiating.34 In this process, the Wnt/β‐catenin pathway is responsible for maintaining Lgr5 expression; Rspo/Lgr5 interaction is involved in stem cell expansion.34, 35, 36

In mice mammary gland, Lgr5 cells are chiefly within the Lin−CD24+CD49fhigh subpopulation; they can differentiate into both basal and luminal mammary epithelial cells, and regenerate functional mammary glands.37, 38 In human breast cancers, both Lgr5 and Rspo are overexpressed with activation of the Wnt/β‐catenin pathway, which contributes to increased tumor growth, metastasis and stemness.39, 40 Tenascin C (TNC) is an extracellular matrix protein abundantly expressed by mammary stem cells.41 Produced by breast cancer cells in which ~90% is CD44+CD24−, TNC maintains the expression of Lgr5 and the response of Lgr5 to Wnt ligands, and is associated with aggressive lung metastasis.42 These data demonstrate a critical role of the Rspo/Lgr5/Wnt feedback loop in maintaining CD44+CD24−Lgr5+ cells and promoting breast cancer metastasis (Figure 2).

Figure 2

Positive feedback loop between the Rspo/Lgr5 pathway and the Wnt/β‐catenin pathway. LRP5/6, low‐density lipoprotein receptor‐related proteins 5/6; Rspo, R‐spondin

Protein C receptor is a single‐pass transmembrane protein and is expressed in hematopoietic, neuronal and epithelial progenitor populations.43 As a Wnt3a‐target protein, it is also expressed in mouse Lin−CD24+CD29hi mammary stem cells. Procr+CD24+CD29hi cells are multipotent, contribute to both basal and luminal cell lineages, and are important for mammary gland development.44 In human breast cancer, Procr is a CD44+ cell‐specific gene; Procr+ cells are enriched for genes involved in cell motility, chemotaxis, and angiogenesis.43, 45 Neuropilin‐1 (Nrp1), induced by Wnt/β‐catenin signaling, functions critically for the activity of Procr+ MaSC, and Nrp1 knockdown suppresses MMTV‐Wnt1 tumor growth.46 These studies indicate that Procr is a marker of mammary stem cells and cancer stem cells. However, unlike Lgr5, functional roles of Procr in these cells remain unknown.

WNT SIGNALING IS OVERACTIVATED IN BASAL‐LIKE BREAST CANCER/TNBC

Basal‐like breast cancers and TNBC are similar, although not identical, in molecular, clinical and pathological profiles, and they overlap by a high percentage. Although MMTV‐Wnt1 tumors express ER, they share a similar gene expression signature to human basal‐like tumors.47, 48 Crossing MMTV‐Wnt1 mice with ERα knockout mice showed that ERα absence delays, but cannot prevent, the appearance of tumors.49 MMTV‐Wnt10b tumors lack ERα, PR and HER2, and express the basal‐epithelial markers CK5 and CK6, therefore resembling human TNBC.9

Wnt10b expression is absent or low in ER+, PR+ and HER2+ human tumors, but high in TNBC, and is correlated with an unfavorable prognosis.9 Moreover, LRP6 and FZD7 are also predominantly overexpressed in basal‐like breast cancer or TNBC.11, 12, 50 The above observations suggest that the canonical Wnt10b/FZD7/LRP6 pathway triggers basal‐like breast cancer/TNBC initiation.

Some other Wnt ligands and receptors including Wnt7a, Wnt5a/5b, FZD6 and ROR1 were also found overexpressed in human basal‐like breast cancer/TNBC and correlated with poor clinical outcome. Wnt7a recruits and activates fibroblasts to facilitate tumor invasion and metastasis dependent on TGF‐β instead of on the β‐catenin pathway.51 Wnt5a/5b and ROR1 are involved in breast cancer invasiveness and metastasis independent of β‐catenin signaling.52, 53, 54 FZD6 promotes TNBC cell motility and distant metastasis through the fibronectin‐actin axis.55 These studies indicate that noncanonical Wnt signaling is involved in basal‐like breast cancer/TNBC progression.

WNT SIGNALING INDUCES BREAST CANCER CELL EPITHELIAL‐MESENCHYMAL TRANSITION

Epithelial‐mesenchymal transition plays critical roles in embryogenesis. This process is also involved in tumor metastasis, providing cancer cells increased motility. Additionally, EMT confers metastatic cancer cell stemness. Ectopic expression of EMT transcription factor such as twist or snail or exposure to TGF‐β1 induced EMT in non‐tumorigenic, immortalized human mammary epithelial cells, and these cells subsequently showed CD44highCD24low phenotype.56 In contrast, mouse Procr+ stem cells and CD49fhigh/CD24med stem cells, as well as human CD44highCD24low stem cells, all showed features of EMT.44, 56

Both canonical and noncanonical Wnt pathways are involved in EMT of breast stem cells or cancer cells. Wnt ligands (Wnt10b/3) or receptors (FZD7/LRP6) mediate activation of the β‐catenin pathway responsible for EMT induction.29, 57, 58, 59, 60 In contrast, Wnt5a/5b‐induced EMT is not dependent on β‐catenin, but is involved in FZD2/STAT3 signaling.61 Ectopic expression of DKK1/SFRP1 or recombinant DKK1/SFRP1 suppresses breast epithelial cell EMT, further proving the role of Wnt ligands and receptors in this process.28, 29 Rspo3 is expressed in mouse MaSC‐enriched basal (Lin−CD24+CD29high) epithelial populations, human basal ER−PR− cell lines, and human basal‐like tumors. Treatment of non‐tumor cells with recombinant Rspo3 induces EMT, whereas knockdown of Rspo3 in breast cancer cells induces mesenchymal‐epithelial transition (MET) with a reduction in cell migration capacity.62 Mechanically, Rspo3 controls EMT through modulating FZD ubiquitination by RNF43 and ZNRF3, thereby affecting the β‐catenin pathway triggered by Wnt ligands.

Transforming growth factor‐β pathways have crosstalk with the Wnt pathway in EMT induction (Figure 3). TGF‐β induces Wnt7a/7b through Smad2/3, which inversely enhances TGF‐β‐induced EMT of mammary epithelial cells.63 Moreover, TGF‐β can activate the β‐catenin pathway to induce EMT through other molecules such as P38 in a Wnt ligand‐nondependent method.64 Both TGF‐β and β‐catenin induce EMT involving a series of EMT transcription factors including Snail, Slug and Twist.65 A combination of SFRP1 with TGF‐β inhibitor exerts greater inhibitory action on mammary epithelial cell migration and mammosphere formation than either of them, further indicating that TGF‐β signaling synergizes with Wnt signaling in maintaining cell mesenchymal phenotype.28 Actually, TGF‐β is involved in MMTV‐Wnt1 tumors. These tumors show extensive nuclear phospho‐Smad2/3 staining, and TGF‐β abrogation significantly delays MMTV‐Wnt1 tumor development.66

Figure 3

Crosstalk between the transforming growth factor‐beta (TGF‐β) pathway and the Wnt/β‐catenin pathway in the induction of breast cancer epithelial‐mesenchymal transition (EMT). LRP5/6, low‐density lipoprotein receptor‐related proteins 5/6. EMT‐TF, EMT‐transcription factor

WNT SIGNALING IS INVOLVED IN EPITHELIAL‐STROMAL CROSSTALK OF BREAST CANCER

It has been established that epithelial‐stromal crosstalk in the tumor microenvironment plays a crucial role in tumor development. In breast cancer stroma, fibroblasts are the major component. Fibroblasts in breast tumors of Wnt‐met mice show a myofibroblast/CAF phenotype characterized with expression of α‐SMA.67 Furthermore, Wnt ligand from breast cancer cells can activate fibroblasts into CAF in a TGF‐β signaling‐dependent way.51 CAF inversely secrete many molecules including Wnt ligands such as Wnt10b in exosomes to induce cancer cell EMT, metastasis and stemness.51, 57, 67 Interestingly, exosomes from fibroblasts can stimulate Wnt11 secretion by breast cancer cells, and then load Wnt11 to activate the Wnt/PCP pathway and drive cancer cell motility and metastasis.68 Together, these data indicate that Wnt signals in breast cancer environment are involved in epithelial‐stromal crosstalk and promote breast cancer progression (Figure 4).

Figure 4

Crosstalk between breast cancer cells and mesenchymal fibroblasts in the breast cancer microenvironment. CAF, cancer‐associated fibroblasts; Fb, fibroblasts; T, tumor cells; T‐EMT, tumor cells with epithelial‐mesenchymal transition

Similarly, Rspo3 is not only secreted by mammary epithelial cells, but also by α‐SMA‐positive cells in breast cancer stroma.62 It is reasonable to think that, similar to autocrine Rspo3, paracrine Rspo3 may also be involved in the activation of the Wnt/β‐catenin pathway and in the induction of cancer cell EMT and stemness. It has been shown that myofibroblast‐derived Rspo3 is required for the expression of Axin2 and Lgr5 in gastric epithelial stem cells.69 Moreover, both Wnt ligands and Rspo3 from platelet‐derived growth factor receptor alpha (PdgfRα)‐positive myofibroblasts are required to support the intestinal stem‐cell niche.70

WNT SIGNALING PLAYS A ROLE IN BREAST CANCER METASTASIS‐INITIATING CELLS

It has now been accepted now that metastasis is an early event in cancer progression. Using HER2 transgenic mice, studies have shown that breast cancer cells can disseminate in tumor premalignant stage,71, 72 and EMT induced by both canonical and noncanonical Wnt signaling contributes to early dissemination.73 These early disseminated cells show stem‐like characteristics and have tumor‐initiating capability. They are quiescent, but they can release from dormancy leading to distant metastasis. Therefore, they are also defined as MIC which may be responsible for tumor metastasis even years after removal of the primary tumors. Notably, one recent study showed that during dormancy, DKK1 suppresses Wnt signaling in MIC which was defined as latent competent cancer (LCC); downregulation of Wnt signaling induces LCC into quiescence to avoid immune surveillance.74 Together, these findings raise a possibility that after early MIC dissemination, Wnt signaling activity may decline for subsequent MIC dormancy.

WNT LIGANDS/RECEPTORS MAY BE TARGETS FOR BREAST CANCER

Porcupine (PORCN), a membrane bound O‐acyltransferase, is required for Wnt secretion and activity. PORCN inhibitor C59 suppresses the growth of MMTV‐Wnt1 tumors, and these C59‐treated tumors show a decrease in β‐catenin, CyclinD1 and c‐Myc.75 Moreover, another PORCN inhibitor LGK974 similarly induces regression of MMTV‐Wnt1 tumors, accompanied by a reduction in LRP6 phosphorylation and AXIN2 expression.76 LRP6 antagonist Mesd/Mesd peptide or LRP6 antagonistic antibodies also inactivate the Wnt/β‐catenin pathway and suppress MMTV‐Wnt1 tumor growth.11 Similar effects were observed in MMTV‐Wnt1 tumors treated with a soluble Wnt Receptor Frizzled8 CRD‐hFc.77 Moreover, FZD7 antibody was shown to block the canonical Wnt pathway and suppress breast cancer cell growth in human tumor xenograft.78 These observations further confirm the roles of Wnt ligands and receptors in breast cancer development.

CONCLUSIONS

Studies from Wnt transgenic mice clearly indicate that Wnt signaling, especially the β‐catenin pathway, is capable of initiating breast carcinogenesis. Observations on clinical human patients show overactivated Wnt signaling in a section of breast cancers in part as a result of an upregulation of Wnt ligands and/or receptors, as well as a downregulation of Wnt antagonist. CSC are enriched in both mouse Wnt transgenic breast cancers and human Wnt overactivated breast cancers. Wnt‐driven breast cancers show basal‐like/triple‐negative and EMT phenotype, and canonical and noncanonical Wnt pathways both contribute to MIC generation and cancer early dissemination. Although not sufficient, evidence still suggests that Wnt signaling inactivation may help MIC into a dormant state to escape immune attack before overt metastasis occurs.

Targeting Wnt ligands/receptors has been shown to be effective on breast tumor growth in animal models and human xenografts, but its effect on breast cancer metastasis remains unclear. Wnt signaling block may suppress tumor cell EMT and early dissemination. However, a reduction in Wnt signaling activity may induce MIC dormancy. Moreover, recent studies have shown that DKK1 level is increased in patients with bone metastasis,79 and Wnt signaling inhibition by DKK1 promotes breast cancer bone metastasis.80 These findings cause us to question the application of Wnt inhibitors in breast cancer.

CONFLICTS OF INTEREST

Authors declare no conflicts of interest for this article.