Introduction
Bone metastases remain a serious long-term complication in patients suffering from breast cancer and prostate cancer. Bisphosphonates (BPs) are established antiresorptive agents for the treatment of bone metastases. More recently, direct antitumor effects of BPs have been suggested, including induction of apoptosis and inhibition of migration, invasion, and (neo) vascularization [
1]. BPs with antitumor activity are generally nitrogen-containing BPs that inhibit farnesyl pyrophosphate synthase, a key enzyme of the mevalonate pathway [
2,
3]. Statins are a second group of clinically established agents that inhibit the mevalonate pathway and have also been associated with an antitumor potential [
4]. Hence, a number of clinical trials were initiated to investigate whether these promising preclinical results would translate into clinical efficacy for women with nonmetastatic breast cancer. Although modestly favoring the adjuvant use of bisphosphonates, these trials failed to provide a clear evidence for their general efficacy in adjuvant therapy. Of note, positive results were repeatedly reported for women with deprived estrogen levels as a result of menopause or hormone ablation therapy [
5‐
8].
Dickkopf-1 (DKK-1), a soluble inhibitor of the canonical Wnt signaling pathway, has been linked to osteolytic bone disease [
9]. In multiple myeloma, increased levels of DKK-1 are associated with the presence of lytic bone lesions [
10]. DKK-1 is thought to promote osteolytic bone disease directly by inhibition of osteoblasts and concurrently promoting osteoclasts via suppression of osteoprotegerin (OPG) and enhancing receptor activator of nuclear factor-κB ligand [
11]. In murine models of myeloma, inhibition of DKK-1 reduced the extent of osteolytic lesions [
12,
13]. In contrast to myeloma bone disease, the role of DKK-1 in breast cancer and prostate cancer is less clear. In metastatic breast cancer, serum levels of DKK-1 are increased [
14], and breast cancer-derived DKK-1 has the ability to inhibit osteoblast differentiation [
15]. In women with triple-negative breast cancer, which is associated with a high risk of recurrence, expression of DKK-1 indicated poor outcome of patients [
16]. In prostate cancer, DKK-1 expression increases in early stages while decreasing during progression towards metastatic disease [
17]. These data indicate that DKK-1 may have a pathophysiological role in skeletal metastases of breast and prostate cancer. Here we identified DKK-1 as a novel target of the mevalonate pathway in estrogen receptor (ER)-negative breast cancer that can be modulated by BPs and statins via inhibiting of geranylgeranylation.
Methods
Cells
Breast cancer cells (MDA-MB-231, MCF-7 and T47D) and prostate cancer cells (PC-3 and MDA-PCa2b) were purchased from ATCC (Manassas, VA, USA), except for MDA-BONE cells (also known as MB-231-TxSA) that were obtained from the University of Texas (San Antonio, TX, USA) and MDA-MET cells that were a kind gift from Prof. L Suva (Center for Orthopaedic Research, University of Arkansas AR, USA). For osteoblast experiments, the murine myoblast cell line mC2C12 was used. C2C12 cells are a murine myoblast cell line capable of differentiation towards osteoblasts in the presence of Wnt ligands. Murine C2C12, control L-cells and WNT3A-L-Cells were a kind gift from Dr Michael Stock (University of Erlangen, Germany). Prostate cancer cell lines were kept in RPMI 1640 medium from Bio West (Renningen, Germany). All other cell lines were cultured in Dulbecco’s modified Eagle’s medium/Ham’s F-12 (PAA, Pasching, Austria), 10% fetal calf serum supreme (Lonza, Cologne, Germany) and 1% penicillin/streptomycin (PAA). Human microvascular endothelial cells-1 were cultured in endothelial cell growth medium 2 (PromoCell, Heidelberg, Germany). Primary cultures of human umbilical vein endothelial cells were isolated using collagenase II as described previously [
18]. All cell lines that were not purchased from ATCC were obtained following the ethical guidelines of our institution and the providing institution. Cell lines were sent to DSMZ (German Collection of Microorganisms and Cell Culturs), where authenticity was determined by short tandem repeat profiling and by matching with the known profiles.
Small interfering RNA, antibodies and reagents
Zoledronic acid was provided by Novartis (Basel, Switzerland), and atorvastatin, mevalonate, geranyl-geranyl-pyrophosphate (GGPP), farnesyl pyrophosphate (FPP), GGTI-298 and FTI-277 were obtained from Sigma-Aldrich (Munich, Germany). Rac1 inhibitors #1 (553502) and #2 (553511), Y-27632, rho kinase inhibitor (H-1152P) and Clostridium difficile toxin A were from Merck Chemicals (Darmstadt, Germany). Cdc42 inhibitor ML-141 was from Tocris Bioscience (Bristol, UK). Rho inhibitor #2 (BML-EI394) was obtained from Enzo (Lörrach, Germany). Rho inhibitor #1 (CT04) and Rho/Rac/Cdc42 activator I were from Cytoskeleton Inc. (Denver, CO, USA). Recombinant human DKK-1 was obtained from Peprotech (Hamburg, Germany). Antibody for RAP1A (sc-1482) was from Santa Cruz (Heidelberg, Germany), RAS (610001) antibody was from BD Biosciences (Heidelberg, Germany), the DKK-1 (MAB10962) antibody was from R&D Systems (Wiesbaden, Germany) and all other antibodies were obtained from Cell Signaling Technology (Frankfurt, Germany). DKK-1 small interfering RNA (siRNA; s22721 and s22723), Cdc42 siRNAs and nontarget siRNA were purchased from Applied Biosystems (Darmstadt, Germany). Cells were transfected using Dharmafect (Thermo Scientific, Waltham, Massachusetts (MA) USA).
RNA isolation, reverse transcription and real-time polymerase chain reaction
RNA from cell cultures was isolated using the HighPure RNA extraction kit from Roche Applied Science (Mannheim, Germany) according to the manufacturer’s protocol. RNA (500 ng) was reverse transcribed using Superscript II (Life Technologies, Darmstadt, Germany) and used for SYBR green-based real-time polymerase chain reaction (PCR) using a standard protocol (Applied Biosystems). Primer sequences were: hu glyceraldehyde 3-phosphate dehydrogenase (GAPDH) sense, AGCCACATCGCTCAGACAC; hu GAPDH antisense, GCCCAATACGACCAAATCC; hu Dkk-1 sense, AGCACCTTGGATGGGTATTC; hu DKK-1 antisense, CACACTTGACCTTCTTTCAGGAC; mu ACTB sense, GATCTGGCACCACACCTTCT; mu ACTB antisense, GGGGTGTTGAAGGTCTCAAA; mu ALP sense, CTGGTGGCATCTCGTTATCC; mu ALP antisense, CTACTTGTGTGGCGTGAAGG; and mu OPG sense, CCTTGCCCTGACCACTCTTA; mu OPG antisense, CCTTGCCCTGACCACTCTTA. PCR conditions were 50°C for 2 minutes and 95°C for 10 minutes followed by 40 cycles with 95°C for 15 seconds and 60°C for 1 minute. The melting curve as assessed in the following program: 95°C for 15 seconds, 60°C for 1 minute and 95°C for 30 seconds. The results were calculated applying the ΔΔCT method and are presented as the x-fold increase relative to the housekeeping gene (GAPDH or β-actin) or as a percentage of control.
WNT array
The human WNT signaling pathway PCR array (PAHS-0437; QIAGEN, Hilden, Germany) was used to screen for Wnt-related targets and was performed using the provided protocols. Briefly, cDNA templates of zoledronic acid-treated and control-treated MDA-231 cells were mixed with the appropriate PCR master mix. Equal volumes were pipetted to each well of the same PCR array. PCRs were performed as described above. Assessment of three individual experiments was conducted using the software provided by the manufacturer.
cDNA polymerase chain reaction array
The breast cancer cDNA Array II was purchased from Origene (Rockville, MD, USA) and was assessed for DKK-1 expression, normalized to GAPDH using the supplier’s protocol. This array contains 48 samples, covering five normal breast tissues and 43 samples of breast cancer, and provides clinical information including hormone status and pathological grade.
Dickkopf-1 and osteoprotegerin enzyme-linked immunosorbent assay and serum markers of bone turnover
Human DKK-1 and murine OPG enzyme-linked immunosorbent assays were obtained from Biomedica (Vienna, Austria) and were performed according to the manufacturer’s instructions. Cell supernatants were prediluted as determined by pretesting. Serum samples of breast cancer patients and healthy controls were obtained after informed consent and Institutional Review Board approval. Details of patient characteristics are shown in Figure S3B in Additional file
1. Patients were women with hormone-negative, nonmetastatic, breast cancer who were blinded to receive infusions of either 4 mg zoledronic acid or placebo every 3 months as adjuvant therapy. These patients received 1,000 mg calcium and 1,000 IU vitamin D per day for the duration of the study after Institutional Review Board approval (97/05 (A)/KKS 1009 and informed patient consent had been obtained. Detailed patient characteristics are shown in Figure S4A in Additional file
1. No additional concomitant drugs known to influence bone turnover were given. Control serum was taken before the first administration of zoledronic acid and before each further administration. Blood samples were immediately worked up and stored at -80°C until the analyses were performed. Markers of bone turnover were measured as routine parameters by our clinical laboratory.
Immunoblotting
Western blot analyses were performed as described previously [
19]. Briefly, after completion of the experiments, cells were lysed and quantified. In general, 20 μg protein were loaded for SDS-PAGE and transferred onto a 0.2 μm nitrocellulose membrane. After blocking for 1 hour with 5% nonfat dry milk in Tris-buffered saline with 1% Tween-20, membranes were incubated with a primary antibody overnight. After washing, the membrane was incubated for 1 hour with the horseradish peroxidase-conjugated secondary antibody. Membranes were then washed three times with Tris-buffered saline with 1% Tween-20, and proteins were visualized with Super Signal (Pierce, Bonn, Germany) enhanced chemiluminescence.
Immunohistochemistry and quantitative assessment
Primary breast cancer tissue was assessed using immunohistochemistry. Multiple tissue microarrays (TMA 8501) were purchased from Tristar (Rockville, MD, USA). Paraffin-embedded sections were dewaxed, rehydrated using an alcohol gradient, and heat-retrieved from antigens. Endogenous peroxidase activity was blocked using 0.3% H
2O
2/phosphate-buffered saline for 10 minutes at room temperature and nonspecific binding sites using the blocking buffer of the VECTASTAIN Elite ABC Kit (VECTOR Laboratories, Peterborough, UK) for 45 minutes at room temperature. Afterwards, sections were incubated with a polyclonal anti-DKK-1 antibody (ab22827; Abcam, Milton, UK) overnight at 4°C. Subsequently, slides were treated with an anti-goat secondary antibody conjugated to biotin and then developed utilizing avidin-conjugated horseradish peroxidase with diaminobenzidine as substrate (DAKO, Hamburg, Germany). Specificity was validated in cell sections where DKK-1 levels were diminished using siRNA (Figure S3A in Additional file
1). Staining intensity was assessed by two individual and experienced pathologists (MHM and GBB) and rated as either none (0), weak (1), moderate (2) or strong (3). Interobserver variability was measured using the Cohen’s κ test.
Statistical analysis
Results are presented as the mean ± standard deviation. All experiments were repeated at least three times. Statistical evaluations were performed using a one-way analysis of variance or Student’s t test. P < 0.05 was considered statistically significant.
Discussion
We show that DKK-1 is a direct target of the mevalonate pathway (Figure
7C). As an inhibitor of osteoblast differentiation and function, elevated DKK-1 levels are thought to contribute to the imbalance of osteoblastic bone formation and osteoclastic bone resorption, a predominant feature in osteolytic bone disease [
21]. DKK-1 is best investigated in myeloma bone disease [
22] and a monoclonal antibody, BHQ880, directed against DKK-1 is currently being evaluated in a combined phase I/II trial [
23].
The role of DKK-1 in breast cancerand prostate cancer is less well characterized. In breast cancer, where predominantly osteolytic lesions are found, a mechanism similar to that proposed in multiple myeloma is plausible. In this study, we identified an increased expression of DKK-1 in hormone-negative and highly osteotropic breast cancer cell lines. This subtype is particularly associated with an unfavorable outcome and a high risk of recurrence. Therefore, adjuvant treatments decreasing the risk of relapse are warranted. Assessment of primary tissue revealed an overexpression of DKK-1 in ER-negative breast cancer cases, but the expression appears independent of tumor grade or stage. Moreover, our data show that breast cancer-derived DKK-1 potently inhibits osteoblast differentiation, as seen assessed by ALP expression, as well as osteoblast-derived production of OPG, a potent inhibitor of osteoclast activity. Zoledronic acid prevented this osteoblast inhibition by negative regulation of DKK-1. This novel mechanism may help to consolidate bone turnover in the metastatic process and provides a rationale to further explore the adjuvant use of BPs in patients with high DKK-1 levels.
In contrast to breast cancer, bone metastases from prostate cancers are predominantly osteosclerotic [
24]. Here, the role of DKK-1 is less obvious. DKK-1 levels have been described to increase early in disease development, followed by a decrease with tumor progression and bone metastases, which may depict the molecular switch that transitions the osteolytic phenotype to an osteoblastic phenotype [
17]. This concept is supported by the finding that overexpression of DKK-1 in osteoblastic prostate cancer cells induces an osteolytic phenotype [
25]. In line with these findings, only the osteolytic PC3 cells expressed relevant levels of DKK-1 in this report. Lowering DKK-1 levels may have direct practical implications for patients with osteotropic tumors because osteoblast activity may be normalized directly and indirectly via OPG and receptor activator of nuclear factor-κB ligand signaling. Silencing of DKK-1 delayed the development of bone lesions in a model of prostate cancer [
26], and neutralizing DKK-1 antibodies have been successfully applied to prevent bone lesions in a preclinical model of myeloma bone disease [
13]. Besides the promising effects of DKK-1 inhibition on bone metastases, inhibiting DKK-1 also bears the potential risk of promoting tumor proliferation as a result of activated Wnt signaling and a number of studies have defined DKK-1 as a tumor suppressor [
27‐
29]. However, these studies focused on effects of overexpressing DKK-1 mainly in hormone receptor-positive breast cancer cells with low or undetectable baseline levels of DKK-1, and these results may not be transferable to other breast cancer subtypes and other entities. Indeed, DKK-1-overexpressing prostate cancer cells were recently reported to have an increased subcutaneous tumor mass upon ectopic transplantation and a higher incidence of bone metastases after intracardiac injection [
30]. Furthermore, genetic disorders of dysfunctional Wnt inhibitors, such as van Buchem’s disease and sclerosteosis, are not associated with an increased risk of malignancies [
31]. However, there are no clinical data available to assess the effects of inhibiting Wnt inhibitors (DKK-1 or sclerostin) in patients with existing malignancies, and the potential risk of aggravation should be considered.
In our study, inhibition of the mevalonate pathway resulted in a profound suppression of DKK-1
in vitro and in breast cancer patients receiving zoledronic acid. We have previously shown that similar concentrations of zoledronic acid are sufficient to induce apoptosis in treated cancer cells [
32,
33]. Furthermore, decreased vascular endothelial growth factor levels have been reported in women receiving zoledronic acid [
34]. Recent results from two large clinical trials have yielded inconclusive results regarding the general use of zoledronic acid as an adjuvant therapy for breast cancer patients [
6,
7]. While especially postmenopausal women appeared to benefit from early treatment with zoledronic acid, exact characteristics of these groups remain elusive. More recently, a meta-analysis of three placebo-controlled trials underlined better survival for zoledronic acid treatment of patients with bone metastases from solid tumors who displayed poor prognosis features such as elevated NTX levels as a surrogate for aggressive bone lesions [
35]. Here, we found an overexpression of DKK-1 in ER-negative breast cancer patients. This is in line with another study, which reported a relative increase of DKK-1 in ER/progesterone receptor-negative cancer [
15]. In our patients, baseline levels of DKK-1 largely varied. This limits the use of DKK-1 as a potential screening marker. Also, it remains unclear whether increases in DKK-1 during the course of a disease are associated with tumor progression. Elevated DKK-1 levels are described in breast cancer patients with confirmed bone lesions [
14] and a recent study defined high levels of DKK-1 as a negative prognostic marker in triple-negative breast cancer [
16]. It is therefore feasible that lowering DKK-1 levels may have positive effects. Hence, patients with high baseline DKK-1 levels may especially benefit from early therapy with zoledronic acid or potentially another mevalonate pathway inhibitor. We found DKK-1 levels to remain effectively suppressed for at least 12 months. Whereas zoledronic acid is a standard agent for the treatment of established bone metastases, statins are not used in this context.
In vitro, considerably lower concentrations of atorvastatin than zoledronic acid were needed to achieve a comparable suppression of DKK-1. However, whether the use of statins also translates into a clinically relevant suppression of DKK-1, as seen with zoledronic acid, remains unclear.
Our study has potential limitations. The number of patients included in this study is small and we have no available survival data. Larger clinical trials are therefore needed to fully define the prognostic value and pathophysiological role of DKK-1 in breast cancer. Furthermore, the tissue or cellular source of DKK-1 suppression remains unclear. Women received zoledronic acid as an adjuvant treatment and were tumor free. The drop in DKK-1 is unlikely to stem from the removal of the tumor, because there was no decrease in the placebo-treated group. Bone markers in the zoledronic acid-treated patients are suppressed as expected following antiresorptive treatment. The decrease in serum DKK-1 may therefore be explained by an impaired osteoblastic activity. Alternatively, since cells of vascular origin are also sensitive to zoledronic acid, DKK-1 suppression may also be a cumulative effect derived from multiple cell types.
Furthermore, the doses required to suppress DKK-1 in vitro probably exceed achievable serum concentrations in vivo. Serum concentrations of zoledronic acid decline rapidly after infusion and, although we have shown that a short exposure time of 2 hours is sufficient to mimic the effects of a continuous BP exposure on DKK-1, it remains unclear whether the observed effects in breast cancer cells in vitro can be translated into the clinical setting. BP uptake in vitro is likely to be mediated by fluid phase endocytosis and the cells’ ability to interfere with the mevalonate pathway in cellular targets outside the bone is likely to be limited by availability and other pharmacokinetic properties.
Currently, it remains unclear to what extent the DKK-1 serum pool in breast cancer patients affected by bone metastasis is derived from tumor cells versus reactively enhanced bone remodeling and whether the suppression of DKK-1 by zoledronic acid has a beneficial effect on osteoblast/osteoclast activity at the site of an established bone lesion or might even prevent the establishment of disseminated tumor cells in the bone.
Competing interests
The authors have received grants or honorarium for advisory boards or lectures for the individual or for the institution from Amgen (TDR, LCH, PH, FJ), AstraZeneca (PH), Eli Lilly (PH, FJ), GlaxoSmithKline (PH), Novartis (TDR, LCH, PH, FJ), Pfizer (PH), Roche (PH, FJ), Servier (LCH, FJ), Merck (LCH, TDR, FJ), and Nycomed (LCH, FJ). The remaining authors declare that they have no competing interests.