Introduction
The mammalian Dickkopf genes (
DKK) encode a class of extracellular signalling molecules that control cell fate during embryonic development and regulate tissue homeostasis in adults [
1,
2]. Four DKK gene members have been identified so far.
DKK1,
DKK2 and
DKK4 antagonise canonical Wnt/β-catenin signalling by interaction with LDL-receptor-related proteins (LRP5 and LRP6) [
3]. In contrast,
DKK3 does not sequester LRPs or Wnt ligands [
2,
4,
5]. Its function in antagonising nuclear β-catenin levels, designated as the hallmark of an activated Wnt pathway often found in human tumour tissues [
6], has received conflicting reports [
7‐
9]. Most evidence suggest
DKK3 exerts a tumour suppressive function by inhibiting a non-canonical Wnt signalling branch referred to as the planar cell polarity (PCP) pathway. The PCP pathway is characterised by the activation of c-Jun kinase (JNK) via recruitment of small GTPases of the Rho/Rac family [
10]. It results in changes in cell adhesion, motility and polarity [
11] rather than interfering with the networks of proliferation and differentiation, which is mediated by canonical Wnt/β-catenin signalling [
6].
In agreement with its putative tumour-suppressive function [
9,
12‐
14]
DKK3 is commonly downregulated in human cancers such as lung cancer [
15‐
17], renal clear cell carcinoma [
18], pancreatic cancer [
19], leukaemia [
20], prostate cancer [
7,
21], bladder cancer [
22], melanoma [
23] and gastrointestinal tumours [
24]. In many of these diseases transcriptional loss is tightly associated with methylation of the
DKK3 promoter [
15,
16,
18,
20‐
22,
24], whereas in other malignancies the cause of downregulation remains to be elucidated or is not related to 5'-cytosine methylation [
23]. A study on lung cancer revealed that the rate of
DKK3 methylation increased steadily from normal lung tissue, to low-grade and high-grade atypical adenomatous hyperplasia to invasive adenocarcinoma [
25], suggesting a potential role of
DKK3 methylation in lung cancer progression. In mouse cancer models,
DKK3 has proved a promising therapeutic agent capable of repressing tumour progression, for example, in testicular germ cell cancer [
14] and prostate cancer [
13]. More recently, a breast cancer xenotransplantation model demonstrated that a single adenoviral-mediated intra-tumoural injection of a
DKK3 expression vector efficiently discontinued tumour growth, with the induction of apoptosis in these cells [
26]. This suggests that
DKK3 may have an important tumour-suppressive function that either prevents tumour initiation or attenuates cancer progression. Interestingly, loss of
DKK3 expression was first observed in numerous immortalised tumour-derived cell lines [
27]. Immortalisation, that is escape from cellular senescence, is an early event in malignant transformation [
28], so
DKK3 could act as a tumour suppressor gene by mediating the effects of senescence stimuli. In concordance with this hypothesis,
DKK3 expression was found to be elevated in organs with predominantly growth-arrested post-mitotic cells, for example in the heart and brain [
29] and also in senescent prostate epithelial cells [
30].
However, to the authors' knowledge, a comprehensive study on DKK3 gene regulation and its implication in breast cancer has not yet been published. In our study we investigated DKK3 mRNA expression, DKK3 protein expression and DKK3 promoter methylation in breast cell lines as well as in normal and malignant primary breast tissues. Our results demonstrate for the first time that DKK3 expression is frequently downregulated in human breast cancer as a consequence of aberrant DNA methylation within the DKK3 gene promoter.
Discussion
It was previously reported that expression of the putative Wnt antagonist
DKK3 was downregulated in several tumour entities as a consequence of epigenetic DNA modification [
15,
16,
18,
20‐
22,
24]. Our study is the first to analyse
DKK3 gene regulation in human breast cancer. Malignant breast cell lines showed strong reduction of
DKK3 mRNA in association with
DKK3 promoter methylation. Consistently,
DKK3 mRNA expression was induced after promoter DNA demethylation in these cells. In primary breast carcinomas,
DKK3 mRNA expression was downregulated in 68% of invasive tumours with significant association with methylation of the
DKK3 gene promoter (p < 0.001). The total frequency of
DKK3 methylation was 61% in breast carcinomas, whereas corresponding normal breast tissues were unaffected by this epimutation. We further showed that a loss of DKK3 protein in breast carcinomas is also associated with
DKK3 promoter methylation (p = 0.001) whereas protein expression is abundant in epithelial cells of the normal breast. In summary, our data demonstrate for the first time that promoter methylation-mediated downregulation of
DKK3 expression is a frequent and tumour-related epigenetic alteration in the development of human breast cancer.
The implication of aberrant canonical Wnt/β-catenin signalling in the pathogenesis of human cancer has become widely accepted [
40]. Its oncogenic effect is mediated by uncontrolled activation of target genes that for example, enhance cell proliferation, such as c-myc and cyclin D1. In breast cancer, several genes encoding inhibitors of canonical Wnt/β-catenin signalling have been reported to be frequently hypermethylated, for example,
SFRP1 [
34,
41],
SFRP2 [
42],
SFRP5 [
43],
WIF1 [
44] and
DKK1 [
42]. We suggest that disruption of a non-canonical Wnt signalling branch, the PCP pathway, may also be implicated in human carcinogenesis by pathologically altering the networks of cellular adhesion, motility and cell polarity, because it has been shown that expression of the putative PCP pathway inhibitor
DKK3 is commonly downregulated in malignant tissues. As a consequence, loss of
DKK3 may promote hyperactivation of the PCP pathway, thereby potentially enhancing tumour aggressiveness.
Recent
in vivo experiments support a hypothesis that the loss of
DKK3 expression promotes an aggressive cancer phenotype. In a mouse model,
DKK3 proved to be a promising therapeutic agent to significantly inhibit tumour growth in testicular germ cell cancer [
14]. In an orthotopic prostate cancer model a similar treatment resulted in tumour regression, decreased metastasis and prolonged survival of the host [
13]. The most recent findings from a breast cancer study revealed that
DKK3 not only attenuates tumour growth in a xenotransplantation mouse model, it also re-sensitised multidrug-resistant MCF7/ADR cells to doxorubicin treatment in a JNK-c-Jun dependent manner [
26]. This highlights its potential utility as a gene therapeutic agent in human breast cancer. Our study adds important information to this aspect, because it so far remained unknown if methylation-mediated loss of
DKK3 expression also occurred in primary breast cancer, and, if so, how many patients were affected by this epimutation. We have shown that a large proportion (61%) of breast cancer patients have
DKK3 promoter methylation in the carcinoma tissue, leading to a functional inactivation of the tumour-protective protein. Therefore, we conclude that a potential gene therapeutic treatment with
DKK3 might be of benefit for a large target population of breast cancer patients.
In contrast to other studies,
DKK3 promoter methylation in our cohort was not associated with clinicopathological factors indicative of a progressive cancer subtype, such as tumour size, node status or histological grade. The existence of such an association has been demonstrated in prostate cancer [
7,
12], in which expression of
DKK3 was predominantly lost in high-grade prostatic tumours. Moreover, siRNA-mediated downregulation of
DKK3 expression in prostate epithelial cells disrupted acinar morphogenesis [
7], which taken with its prevalent expression in growth-arrested cells suggests a functional role of
DKK3 in post-mitotic tissue differentiation processes. Whether
DKK3 is also involved in maintaining glandular morphology in the normal mammary gland will be elucidated in a further study.
In human breast cancer, hypermethylation of Wnt antagonist genes was reported to be of clinical relevance. Both
SFRP1 and
SFRP5 methylation were shown to occur frequently and be tumour specific with a strong association to poor clinical patient outcome [
34,
43]. An impact of
DKK3 promoter methylation on cancer patient survival has been repeatedly found. It was shown to be associated with reduced disease-free survival in acute lymphoblastic leukaemia [
20], and also with shorter overall survival in kidney cancer [
45] and lung cancer [
46]. Due to its functional properties as a potential tumour suppressor in human cancers including breast cancer [
26], together with the finding that
DKK3 methylation is a significant prognostic factor in three human malignancies, we speculate that
DKK3 methylation might also bear prognostic power in breast cancer. This hypothesis is currently being approached in our laboratory in a further study.
In summary, we demonstrate for the first time that the Wnt antagonist gene DKK3 is a frequent target of epigenetic inactivation in human breast cancer, leading to downregulation of DKK3 mRNA and DKK3 protein expression in tumourous tissues. These results suggest a causative implication of DKK3 in the development of human breast cancer. Since DKK3 is believed to negatively regulate Wnt signalling, these results underline the pivotal role of a deregulated Wnt signalling pathway commonly found in this disease.
Competing interests
Edgar Dahl has declared that he has submitted a patent application on the use of DKK3 promoter methylation. The other authors have no competing interests.
Authors' contributions
JV carried out the gene expression analyses, immunohistochemical studies, methylation experiments and statistical evaluations, participated in the conception and design of the study, and wrote the manuscript. NB participated in the immunohistochemical analysis, performed data interpretation and critically revised the manuscript. AH provided clinical samples and clinicopathological data, performed data interpretation, supported in statistical analyses and critically revised the manuscript. GK provided clinical samples and clinicopathological data, performed data interpretation and critically revised the manuscript. UH provided clinical samples and clinicopathological data, participated in data interpretation and critically revised the manuscript. RK participated in the design and co-ordination of the study and critically revised the manuscript. ED planned and co-ordinated the study, and critically revised the manuscript. All authors have given final approval of the text to be published.