Tumours have an aberrant protein expression profile as a consequence of genomic and proteomic alterations. Frequently, genes specialised in embryonic development are abnormally expressed in tumours. We describe here DKK1 expression in human tumours of various origins, including breasts, lungs and kidneys. Dickkopf-1, which is involved in some aspects of embryonic development, was detected in mature human tissues, mainly in the placenta, an observation reported by Fedi et al (1999)
. Interestingly, by analysing its expression profile in breast cancer patients, DKK1 appears in tumours with a poor outcome, specifically hormone-independent cases. Also, we reported preferential tumour expression in women with familial cases of the disease. Finally, we observed substantial DKK1 protein secretion in breast cancer lines, which was further confirmed in crude extracts prepared from breast cancer specimens.
In the embryo, DKK1 functions as a secreted protein interfering with the canonical Wnt pathway (Mao et al, 2002
). In the absence of DKK1, Wnt interacts with two co-receptors, namely, LRP5/6 and Fz, which results in β
-catenin accumulation and migration to the nucleus. Consequently, interaction with the transcription factor TCF delivers positive signals for cell proliferation (reviewed in Rothbacher and Lemaire, 2002
; Brennan and Brown, 2004
). Interestingly, in 1982, Nusse and Varmus (1982)
identified the first Wnt
gene as being a mammary oncogene, and several members of the Wnt family have been linked to cancer development, especially of the breast (reviewed in Li et al, 2000
). Surprisingly, low levels of membranous β
-catenin expression have been associated with significantly worse outcomes (Dolled-Filhart et al, 2006
), which contradicts other studies (Lin et al, 2000
; Chung et al, 2004
). As such, there is still much debate about the link between tumour aggressiveness and β
-catenin expression. Interestingly, DKK1 negatively affects the Wnt pathway. At the adequate time during embryogenesis, DKK1 is secreted and binds to the LRP5/6 co-receptor (Semenov et al, 2001
), blocking interaction with secreted Wnt protein, causing β
-catenin degradation and stopping TCF-regulated gene expression in the nucleus. This mechanism of DKK1 action is important in limb and head development (Glinka et al, 1998
; Mukhopadhyay et al, 2001
). Conversely, inhibition of the Wnt pathway by DKK1 initiates cardiogenesis early in vertebrate embryos (Marvin et al, 2001
; Foley and Mercola, 2005
Dickkopf-1 has been studied in the context of colon and gastric cancers. In colon cancer, Gonzalez-Sancho et al (2005)
reported that the loss of DKK1 expression may open the door to cancer by removing the inhibitory effect on the Wnt/β
-catenin pathway. Dickkopf-1 epigenetic inactivation may be a consequence of CpG methylation (Aguilera et al, 2006
; Mikata et al, 2006
). However, hypermethylation has been observed in only 17% of colon cancer clinical specimens, which indicates that this phenomenon is real but cannot be generalised (Aguilera et al, 2006
). Also, the convincing mechanistic demonstration was performed mostly with cancer cell lines treated with the demethylating agent 5-aza-2′-deoxycytidine (DAC) and contradicts the fact that DKK1 is secreted by many highly proliferative cancer cell lines (, and ) and is detected in many breast, kidney and lung cancer specimens.
Consequently, DKK1 may have either negative or positive consequences on development, depending on time and tissue distribution during embryogenesis. In cancer, DKK1 expression does not apparently alter cell growth, especially since we noted its expression in tumours with poor prognosis. The link between DKK1 and Wnt in the context of cancer progression is plausible and currently under investigation. Previous studies have shown that artificial DKK1 expression in some tumour lines with constitutive activation of the β
-catenin pathways resulted in some decrease of cell viability but only in the presence of an oxidative stress inducer (Bafico et al, 2004
). In cancer cell lines such as MDA231 and HCT116 where β
-catenin is upregulated, the addition of inhibitors of the canonical Wnt pathway (other than DKK1) led to a marked reduction of free β
-catenin (Bafico et al, 2004
; Gregorieff and Clevers, 2005
). However, according to our findings, these two cell lines already secrete high levels of DKK1 protein, which is known to be an inhibitor of the canonical Wnt pathway. Consequently, none of these in vitro
studies correlate with the clinical observation we report here, about the presence of DKK1 protein in growing tumours from breast cancer patients. It is too early to speculate as to whether DKK1 plays a role in cancer similar to its known function in normal cells and in embryogenesis. It may be possible that DKK1 overexpression in in vitro
systems may be masked by its other features when expressed at a physiological level. Still, a high cytoplasmic β
-catenin level was found in patients with poor prognosis (Lin et al, 2000
). DKK1 has been linked to other attributes specific to cancer cells. For example, Hall et al (2005)
have recently reported that prostate cancer-derived DKK1 is involved in osteoblastic activity in bone metastases.
Dickkopf-1 could also be involved in particular phenotypes of hormone responsive tumours. We observed statistically significant preferential DKK1 expression in hormone receptor-negative (ER−
) breast tumours (). Dickkopf-1 is regulated by progesterone in normal endometrial stroma cells (Tulac et al, 2006
), but there is insufficient topical DKK1 expression in normal tissue for it to be linked to the expression profile reported here in breast cancer. Interestingly, Faivre and co-workers recently reported that the Wnt pathway can be upregulated by the progesterone receptor in breast cancer (Faivre and Lange, 2007
). However, it is too soon to establish a link between DKK1 expression and the absence of hormone receptors. In fact, we observed DKK1 in two hormone-independent prostate cancer lines (DU45 and PC3; ) but not in a hormone-dependent tumour (LNCaP). This expression profile is similar to that observed in breast cancer. Finally, we observed coexpression of DKK1 and HER-2/neu
in breast cancer cells in only one out of the 21 DKK1+
tumours (data not shown). Consequently, only one of those tumours would be eligible for treatment with HerceptinTM
, an antibody interfering with tumour progression. This further emphasises the necessity of finding additional targets for immunotherapy.
Interestingly, DKK1 could have potential applications as a secreted tumour marker for cancer diagnosis, staging and monitoring of relapse. Additional investigations are required to establish the feasibility of DKK1 protein detection in different body specimens or fluids.
In conclusion, as DKK1 is specifically expressed in common cancers, and absent from essential normal tissues, this protein is a potential TA for cancer immunotherapy. Its role as an inhibitor of the Wnt canonical pathway in normal cells aside, it may be possible to target DKK1 for a cytotoxic response through CD8+ T-cell recognition as a consequence of internal antigen processing leading to MHC class I presentation. In addition, a humoral response may be involved, as antigen-presenting cells can take up secreted tumour-derived DKK1 and elicit a CD4+ helper T-lymphocyte response. Importantly, considering that DKK1 is preferentially expressed in tumours from women with a family history, but absent from important normal tissues, the protein could be targeted in a preventive vaccine for women at risk of developing the condition. Actually, about 70–80% of women at high risk for breast cancer are predicted to develop the disease and, presently, with the exception of radical mastectomy, no effective prevention strategies are available.