RANK, RANKL and OPG Expression in Breast Cancer – Influence on Osseous Metastasis

Abstract

In women, malignant breast tumours are among the most common malignant diseases in Europe. In advanced breast cancer, the risk of bone metastasis increases to 65–75 %. The discovery of the physiological bone metabolism parameters RANK (receptor activator of nuclear factor-κB), RANKL (receptor activator of nuclear factor-κB ligand) and OPG (osteoprotegerin) as well as their pathophysiological involvement in bone-related diseases is the subject of new therapeutic strategies. The formation of osteolytic bone metastasis requires increased osteoclast activity. Activation of osteoclasts by excessive direct RANKL or reduced OPG expression of osseous metastatic tumour cells remains to be elucidated. More than 50 % of primary breast cancer cells express OPG and RANK, while RANKL could be detected only in 14–60 %. Increased OPG concentrations in the serum of patients with bone metastases have been shown in several studies, whereas the RANKL results are described in an opposite manner. The use of OPG as a biomarker for the detection of osteolytic bone metastases is not consistent and needs to be proved in further studies. Increased RANKL activity was found in diseases characterised by excessive bone loss and formed the basis of new therapeutic options. In several studies, a human monoclonal antibody to RANKL (denosumab) was investigated for the treatment of bone diseases. Denosumab is a promising therapeutic option due to its bone-protective effects.

Introduction

With a proportion of 29 % of all tumour-related diseases in Germany, breast cancer is one of the most common malignant conditions in women. The contextual relationship behind breast cancerʼs increased tendency to form distant metastases in bone is the subject of ongoing research projects aimed at developing new therapeutic options.

This overview paper intends to discuss the direct pathophysiological involvement of breast tumour cells in bone metabolism through the expression of the bone metabolism mediators RANK (receptor activator of nuclear factor κB), RANKL (receptor activator of nuclear factor κB ligand) and OPG (osteoprotegerin), along with their therapeutic applications.

Bone Metabolism

Bony tissue is made up of 65 % hydroxyl apatite crystals and around 35 % organic matrix. The organic matrix is made up of 90 % type 1 collagen and 10 % non-collagenous proteins such as fibronectin, osteocalcin, osteopontin and bone sialoprotein 1.

Bone metabolism is a lifelong, dynamic process characterised by interspersing periods of bone resorption and bone formation. Bone cells are involved in the exchange of old bony tissue and the formation of new bony substance. These include osteoblasts, responsible for bone growth, and osteoclasts, responsible for bone resorption. Osteoblasts are formed from pluripotent mesenchymal stem cells, produce the organic base substance (osteoid) and are responsible for the mineralisation of bone. Osteoclasts are formed from haematopoietic stem cells, differentiate into multinucleated giant cells and occur in the boneʼs absorption zone. The interaction between the boneʼs stromal cells, the osteoblasts and the osteoclasts is controlled in part by the bone metabolism mediators RANK, RANKL and OPG 1, 2, 3.

Physiological Function of RANK, RANKL and OPG in Bone Metabolism

The discovery of the RANK/RANKL/OPG signal pathway has contributed significantly to the understanding of physiological bone metabolism.

RANKL belongs to the family of TNF (tumour necrosis factor) ligands and is mainly formed by bone marrow stromal cells, osteoblasts and T lymphocytes 4, 5. It is a membrane-bound peptide which can also be converted into a secreted form following post-translational processing by TACE (TNF-α-converting enzyme like protease) 6. The interaction between RANK and RANKL stimulates the differentiation and fusion of osteoclast precursor cells as well as the activation of mature osteoclasts 5 (Fig. 1). RANKL also play a role in the accumulation of osteoclasts on the surface of the bone and in the extension of their life cycle through the inhibition of apoptosis 7, 8. RANK is a homotrimeric membrane protein (616 amino acids) from the TNF receptor family and, in addition to osteoclasts, is also expressed by lymphocytes and dendritic cells 9, 10. RANK and RANKL-deficient mice exhibited reduced or non-existent osteoclast differentiation, severe osteopetrosis and also immunological defects such as missing lymph nodes 11, 12. However rats who over-expressed OPG, in which RANKL is continually inhibited, demonstrated a higher bone density without malformation of the lymphoid organs and with no impairment of the immune response 13. By contrast, over-expression of RANKL in mice led to a reduced bone density and – as a result of the increased number of osteoclasts – to osteoporosis 14. The subcutaneous injection of recombinant RANKL in vivo also led to increased bone loss, the increased production of bone absorption markers and a reduction in bone strength 15.

Fig. 1.

 Control of osteoclast differentiation by the RANK/RANKL/OPG signal pathway. RANKL is formed by osteoblasts and bone marrow stromal cells and binds to the RANK receptor on osteoclast precursor cells (CFU-M: colony forming unit-macrophage) and mature osteoclasts. This interaction promotes the differentiation and activation of osteoclasts. OPG is also formed by bone marrow stromal cells as well as osteoblasts and neutralises RANKL. OPG therefore has an inhibitory effect on osteoclast differentiation and activation. The ratio between RANKL and OPG determines whether bone-forming or bone-removing processes dominate (based on 60).

Over-expression of OPG, a member of the TNF receptors family, in vivo leads to increased bone mass and reduces the number and activity of osteoclasts 16. Independent of this, scientists in Japan have been able to confirm the osteoclast-inhibiting effect of OPG in vitro 17. Within the bone, OPG is produced by cells from the osteoblast series and secreted as a monomer or homodimer 18. As the level of differentiation of the cell increases, so does OPG production 19. OPG acts as a competitive receptor antagonist by binding and neutralising membrane-bound and soluble RANKL. This suppresses the differentiation and fusion of osteoclast precursor cells and inhibits the activation of mature osteoclasts 20, 21, (Fig. 1). Transgenic mice who over-express OPG also exhibit, similar to RANKL and RANK-deficient mice, a bone-protective effect and osteopetrosis 11, 12, 16. The switching off of OPG in vivo, however, leads to massive osteoclast activity and associated osteoporosis 22.

Interaction between Tumour and Bone Cells

In some malignant diseases (multiple myeloma, prostate cancer and breast cancer) with associated bony involvement, impaired regulation of the RANK/RANKL/OPG system plays a significant role.

An invasion of tumour cells into the bony tissue is usually associated with osteolytic lesions (breast cancer 23, multiple myeloma 24) and, more rarely, with osteoblast metastases (prostate cancer 25). Interaction between the tumour cells and the boneʼs micro-environment, based on the principle of the “soil and seed” hypothesis, is crucial for the initiation of osteolytic bone metastasis 26, 27. Tumour cells secrete soluble factors (e.g. cytokines, hormones and growth factors) which ultimately lead to osteoclast activation and thus to bone-destroying processes 28, 29. The PTHrP (parathyroid hormone-related protein) produced by tumours induces the increased expression of RANKL by the osteoblast stromal cells, thereby leading to increased osteoclast activation 30. Osteoclastic bone absorption releases growth factors such as TGF-β (transforming growth factor β), BMPs (bone morphogenic proteins) or IGF (insulin-like growth factor) from the bone matrix, which in turn contributes towards the increased proliferation of tumour cells. This results in a vicious circle involving the growing proliferation of tumour cells and increased osteolysis 31, 32, 33 (Fig. 2).

Fig. 2.

 Interaction between tumour and bone cells. Tumour cells secrete soluble factors that lead directly or indirectly, through the activation of osteoblasts, to increased osteoclast activity and associated bone destruction. Bone resorption leads to the release of growth factors and increased tumour proliferation. A vicious circle arises involving the increasing proliferation of tumour cells and increased osteoclast activation (based on 71).

To what extent breast cancer cells are themselves able to express the bone metabolism mediators RANK, RANKL and OPG and, as a result, actively intervene in the metabolism of bone will be summarised in the following section.

RANK, RANKL and OPG Expression in Breast Cancer

Expression analyses at mRNA level

Several studies have analysed the RANK, RANKL and OPG expression of breast cancer cell lines (HCC70, MCF-7, MCF-7 3.1, MCF-7 aro, MDA-MB-231, MDA-MB-435, MDA-MB-453, T47D and ZR 75-1) using RT-PCR analyses. MCF-7, MDA-MB-231 and T47D expressed RANK, and all investigated cell lines, with the exception of MDA-MB-453, expressed OPG, whereas RANKL expression was determined only in HCC70 30, 34, 35 (Table 1).

Table 1 RANK, RANKL and OPG expression analysis at mRNA level in breast cancer cell lines.

Cell line	RANK	RANKL	OPG	Reference
Note: a only analysed by Thomas et al. 30; n. d.: not determined.
HCC70	n. d.	+	+	35
MCF-7	+a	–	+	30, 35
MCF-7 3.1	n. d.	–	+	35
MCF-7 aro	n. d.	–	+	35
MDA-MB-231	+a	–	+	30, 34, 35
MDA-MB-435	n. d.	–	+	35
MDA-MB-453	n. d.	–	–	35
T47D	+a	–	+	30, 35
ZR 75-1	n. d.	–	+	35

A lack of RANKL and different degrees of RANK and OPG expression were also confirmed in primary breast cancer tissue 30, 34 (Table 2). Using real-time PCR analyses, it was possible to detect a significantly reduced RANKL expression in primary breast cancer cells compared to healthy breast gland tissue, whereas only trends were seen in terms of the expression data in distant metastases and in the OPG expression studies 36 (Table 2).

Table 2 RANK, RANKL and OPG expression analysis at mRNA level of primary/distant metastasised breast tumours and healthy control tissue.

Tissue	RANK	RANKL	OPG	Reference
Note: b qualitative evidence; c quantitative evidence (compared with healthy control tissue [n = 18]); * p < 0.05; n. d.: not determined.
Primary tumour (n = 12)	100 %b	0 %b	100 %b	30
Primary tumour (n = 30)	n. d.	0 %b	n. d.	34
Primary tumour (n = 24)	n. d.	↓*c	↑ (Trend)c	36
Liver metastasis (n = 3)	n. d.	↓ (Trend)c	↑*c	36
Bony metastasis (n = 1)	n. d.	↓ (Trend)c	↑ (Trend)c	36
Ovarian metastasis (n = 1)	n. d.	↓ (Trend)c	↓ (Trend)c	36

Expression analyses at protein level

The results of a lack of RANKL expression at mRNA level in the MDA-MB-231 30, 34 and MCF-7 cell lines 30 contrast with the results of two further studies. The transmembrane-bound and extra-cellular expression of RANKL in MCF-cells 37 was demonstrated at protein level, as was the co-expression of OPG and RANKL in MDA-MB-231 cells 38. OPG and RANKL expression in the HCC70 cell line was also confirmed at protein level using immuno-histochemical stainings 35 (Table 3).

Table 3 RANK, RANKL and OPG expression analysis at protein level in breast cancer cell lines.

Cell line	RANK	RANKL	OPG	Reference
Note: n. d.: not determined.
HCC70	n. d.	+	+	35
MCF-7	n. d.	+	n. d.	37
MDA-MB-231	n. d.	+	+	38

Several studies have determined the expression of bone metabolism mediators at protein level in human tissue. In the majority of analysed cases, RANK was demonstrated in healthy breast gland tissue (100 %) and in primary (65–100 %) and osseously metastasised tumour cells (50–100 %) 39, 40, 41. In only one study, by Trinkaus et al., RANK expression was not confirmed in healthy tissue and in the primary tumour cells, possibly due to the low case numbers (n = 4) 41 (Table 4). It has recently been shown that increased RANK expression in primary breast tumour tissue correlated positively with the development of bone metastasis and is associated with shorter disease-free skeletal survival. Increased RANK expression is also closely associated with negative prognostic parameters (tumour greater than 2 cm; G3; oestrogen receptor-negative tumours). The data was collected at mRNA and protein level 42.

Table 4 RANK, RANKL and OPG expression analysis at protein level of primary/distant metastasised breast tumours and healthy control tissue.

Tissue	RANK	RANKL	OPG	Reference
Note: d primary tumour tissue from patients without bony metastases; e primary tumour tissue from patients with bony metastases; f expression only in epithelium showing columnar alteration; n. d.: not determined; n. a.: not analysable.
Normal tissue (breast) (n = 10)	100 %	90 %	n. d.	39
Primary tumour (n = 58)	100 %	61.5 %d 31.3 %e	n. d.	39
Bony metastasis (n = 43)	100 %	2 %	n. d.	39
Normal tissue (breast) (n = 4)	0 %	0 %	n. a.	41
Primary tumour (n = 4)	0 %	0 %	n. a.	41
Bony metastasis (n = 22)	50 %	0 %	n. a.	41
Normal tissue (breast) (n = 5)	n. d.	100 %	100 %f	43
Primary tumour (n = 40)	n. d.	60 %	55 %	43
Primary tumour (n = 400)	n. d.	n. d.	40 %	46
Primary tumour (n = 400)	n. d.	14 %	n. d.	44
Primary tumour (n = 14)	65 %	n. d.	n. d.	40
Bony metastasis (n = 19)	70 %	n. d.	n. d.	40
Bony metastasis (n = 4)	n. d.	100 %	n. d.	45

RANKL is expressed heterogeneously in healthy breast gland epithelium and in primary and osseous tumour tissue. In 90–100 % of cases, healthy breast gland epithelial cells expressed RANKL, whereas this figure decreased to 14–62.5 % in primary breast tumour tissue 39, 43, 44. One of the studies observed a negative correlation between the RANKL expression of primary tumour cells and the oestrogen receptor status, as well as a positive correlation with the tumour grading 44. Accordingly, differences in the percentage distribution of RANKL-expressing primary breast cancer tumours, as in the comparison of the van Poznak and Cross study, could relate to the composition of the collective with regard to the oestrogen receptor status. Only Huang et al. were able to observe RANKL expression in all investigated cases (n = 4) in the osseously metastasised tumour cells 45, whereas in the other studies with larger case numbers, osseously metastasised tumour cells expressed RANKL in no more than 2 % of the cases 39, 41. Bhatia et al. also demonstrated a negative correlation between RANKL expression and the osseously metastasising phenotype 39 (Table 4).

OPG is not 46 expressed in lobules and glandular tissue with non-neoplastic changes, but was strongly expressed in epithelium showing columnar alteration 43. Van Poznak et al. and Holen et al. were able to demonstrate OPG expression in primary breast tumour tissue in 40 % or 55 % of cases, which also correlated positively with oestrogen receptor status 43, 46 and negatively with the ascending tumour grading 46 (Table 4).

Intervention in bone metabolism by tumour-related osteoclast activation as a result of direct RANKL expression by osseously metastasised tumour cells cannot be definitively explained due to contrasting study results. In the majority of cases, RANKL expression was lacking in osseously metastasised tumour cells 39, 41, 45. There is debate over whether or not the existing RANK expression takes priority in osseously metastasised tumour cells due to the lack of RANKL expression. RANK could serve as an anchor and contribute to any possible interaction between tumour cells and bone, or directly activate osteoblasts or stromal cells through interaction with RANKL, which can ultimately result in secondary osteoclast activation and increased osteolysis 39. The study by Trinkaus et al. also discusses a possible chemotactic stimulus of osteoblast RANKL expression for RANK-expressing, disseminated tumour cells 41. In vitro data confirms that RANKL can have a contributory effect, depending on its concentration, towards the regulation of the migration of RANK-expressing, healthy epithelial cells and cancer cells 47.

The functional significance of reduced RANKL expression (compared to healthy mammary gland) can only be surmised. OPG is not only able to bind to RANKL as a receptor antagonist but also has a binding affinity to TRAIL (TNF-related apoptosis-inducing ligand). By blockading the TRAIL receptor, TRAIL-induced apoptosis can be inhibited. The lower binding affinity between OPG and TRAIL described in earlier studies 48 has been refuted in a more recent work and is comparable with the affinity between OPG and RANKL 49. The reduced RANKL expression could lead to an increased interaction between OPG and TRAIL and consequently to a reduced rate of apoptosis 44. Further studies are needed to clarify the functional significance and the underlying mechanism of reduced RANKL expression.

The following section discusses the extent to which, compared to a healthy test subject, there is a disrupted ratio of soluble OPG and RANKL in the serum of patients with breast cancer and the resulting possible contextual relationships with regard to bone metastasis.

RANKL and OPG Serum Levels in Patients with Breast Cancer

An early study demonstrated that patients with breast cancer had no significant differences in their serum OPG level compared to the healthy control group 50 (Table 5).

Table 5 Serum RANKL and OPG level in patients with breast cancer compared with the control group.

Patients	Control	RANKL	OPG	Reference
Note: g on Anastrazole therapy; h trend towards a reduced RANKL/OPG ratio after treatment; * p < 0.05; ** p < 0.001; n. d.: not determined; n. s.: not significant; BP: bisphosphonate.
Breast cancer	Healthy	n. d.	n. s.	50
Breast cancer (bony metastasis)	Healthy	↑*	↑*	51
Breast cancer (bony metastasis)	Healthy	↓**	↓**	52
Breast cancer (bony metastasis)	Breast cancer (no bony metastasis)	n. s.	↑*	53
Breast cancerg (bony metastasis)	Breast cancerg (no bony metastasis)	n. d.	↑*	54
Breast cancer (bony metastasis) before BP therapy	Breast cancer (bony metastasis) after BP therapy	n. s.h	n. s.	56

A comparison of the serum levels of OPG and RANKL in breast cancer patients with osseous involvement and in healthy patients produced conflicting results in two studies. In the study by Mountzios et al., OPG levels and, with the exception of prostate cancer patients, RANKL levels were raised for all tumour types (breast, prostate and lung cancer). A summary of the three tumour types yielded a positive correlation between OPG and the extent of bone metastases 51. In a recent study by Mercatali et al., on the other hand, significantly lower OPG and RANKL levels were observed in patients with breast cancer and bone metastases. In view of its high specificity (87.7 %) and sensitivity (74.1 %), the use of OPG as a bio-marker for detecting bone metastases would appear plausible 52 (Table 5). One explanation for the contrasting study results could be the different testing methods (ELISA analyses 51, quantitative real-time PCR 52) used for the determination of RANKL and OPG.

Two further studies confirmed increased serum OPG levels in patients with breast cancer and bony metastases compared to patients with tumours but with no distant metastases. It was, however, not possible to use RANKL or OPG as a bio-marker for detecting bony metastases 53, 54 (Table 5). One possible explanation discussed for the raised serum OPG level in patients with bony metastases was a secondary reaction aimed at balancing the bone metabolism. Increased OPG production could restrict increasing osteoclast activity in tumour patients with bony metastases 54. Raised serum OPG levels in patients with post-menopausal osteoporosis, and therefore with associated increased osteoclast activity, were discussed in a similar manner in an earlier work 55.

In a further study, the serum concentrations of OPG and RANKL in breast cancer patients with distant bony metastasis were determined before and after treatment with zoledronate, a bisphosphonate. A trend was observed towards a reduced RANKL/OPG ratio following treatment 56 (Table 5).

Further studies are needed to definitively explore any possible clinical use of RANKL and/or OPG as bio-markers. The importance of the RANK/RANKL/OPG system in bone metabolism and the need for increased osteoclast activity for the development of bony metastases, however, open up new therapeutic possibilities.

RANK/RANKL/OPG System – the Basis for Future Treatment Options for Breast Cancer

Initial clinical studies have investigated the effectiveness of recombinant OPG constructs in healthy, post-menopausal women on the treatment of osteoporosis and in patients with malignant diseases on the treatment of tumour-related bone loss. A one-off, subcutaneous administration of Fc-OPG in healthy, post-menopausal women resulted in a suppression of bone absorption markers 57. In a phase I study, a second construct (OPG-Fc) was tested on patients with multiple myeloma as well as patients with breast cancer and osteolytic bone lesions. In a similar way to the control group treated with bisphosphonate (pamidronate), the one-off administration led to a rapid drop in the bone absorption marker 58. Risks associated with the use of OPG constructs included the possible generation of neutralising, anti-OPG antibodies and the associated neutralisation of the endogenous OPGs, as well as potential binding to TRAIL and consequently interference with natural tumour defence mechanisms 59. With the development of the monoclonal human RANKL antibody (denosumab), no further studies involving recombinant OPG constructs were carried out in view of denosumabʼs high specificity and longer half-life 60. Several studies demonstrated the bone-protecting effect of denosumab in patients with post-menopausal osteoporosis 61, 62 and in patients with rheumatoid arthritis 63, 64.

The results of a study investigating the possible use of denosumab in patients with breast cancer with bony involvement are promising. Treatment with denosumab led to a reduced excretion of the bone absorption marker uNTX (urinary N-telopeptide). The reduction in uNTX was of a similar magnitude to that seen in the control group treated with pamidronate but was effective for longer 65. Supplementary phase II studies confirmed a similar level of suppression of bone turnover in patients with breast cancer and bony involvement treated with denosumab and in those who were given intravenous bisphosphonate. The risk of skeletal complications was also reduced 66, 67. In patients who had high uNTX levels despite bisphosphonate therapy, switching to denosumab resulted in a greater drop in uNTX levels than in patients whose treatment was continued with bisphosphonates 68. In a recent clinical phase III study, patients with confirmed breast cancer and at least one bony metastasis were treated with either denosumab (120 mg s. c. and placebo i. v.) or zoledronate (4 mg i. v. and placebo s. c.) and the results compared. In the patients treated with denosumab, suppression of bone turnover was confirmed and skeletal complications were observed to develop more slowly 69.

A further clinical phase III study showed that even patients with breast cancer without bony metastases but with reduced bone density (caused by treatment with aromatase inhibitors) were able to benefit from treatment with denosumab. The treatment significantly increased bone density after 12 or 24 months compared with the control group treated with the placebo 70.

Denosumab, in view of its high specificity, long half-life and good tolerability, represents a highly promising treatment option for bone diseases.

Summary for Implementation in Practice

The discovery of the pathophysiological involvement of the bone metabolism parameters RANK, RANKL and OPG in bone-related malignant diseases has formed the subject of new treatment options. The ratio of RANKL to OPG controls bone-forming and bone-removing processes that are partly responsible for the development of bony metastases.

Contrasting or as yet unreported analyses mean that it has not yet been possible to definitively explain whether an excessive or insufficient expression of RANKL and OPG by breast tumour cells has a direct impact on the physiological ratio of RANKL/OPG. However, several studies have shown that a blockade of RANKL by monoclonal antibodies (denosumab) exhibits a bone-protective effect and represents a highly promising therapeutic option for the treatment of bone-related diseases (e.g. bony metastases, rheumatoid arthritis, post-menopausal osteoporosis).