David Nicol, University of Queensland, School of Medicine, Brisbane, Queensland, Australia. e-mail: firstname.lastname@example.org
The interplay in prostate cancer bone metastases between the ‘seed’ (the prostate cancer cells) and the ‘soil’ (the bone microenvironment) has been increasingly recognized as integral to the remarkable tropism for bone shown by prostate cancer. Increasing research into this area is elucidating the mechanisms involved in this complex ‘cross-talk’. Recent developments, including the use of bisphosphonates in metastatic disease, highlight the important role of bone cells in the development and progression of metastatic prostate cancer. We review the current reports emphasising these possible mechanisms and indicating possible factors for future treatment directions.
A feature of prostate cancer is its predilection to metastasize to bone, resulting in pathological fractures, bone pain and spinal cord compression. These complications severely affect the patients' quality of life . Most prostate cancer bone lesions are osteoblastic (increased deposition of bone), unlike other skeletal metastases which are typically osteolytic due to bone destruction. The complex interplay between prostate cancer and the bone microenvironment will be discussed in detail in this mini-review.
MOLECULAR CONTROL OF BONE STRUCTURE AND FUNCTION
Bone is a biphasic composite material composed of mineral (calcium hydroxyapatite) and organic matrix (osteoid), giving high strength and resilience to the skeleton. There are two commonly described types of bone tissue, i.e. woven bone and lamellar bone. Woven bone is structurally characterized by random orientation of its collagen fibrils and often high mineral density. It is normally seen in the fetal skeleton at the growth plates. Woven bone in an adult is always indicative of a pathological state [2,3].
Lamellar bone, which usually replaces woven bone, is deposited much more slowly and in a more orderly, layered manner, and is therefore much stronger. It requires a pre-formed solid scaffold and the newly formed lamellae run strictly parallel to the underlying surface [2,3]. The cells involved in bone function include osteoprogenitor cells, osteoblasts, osteoclasts and osteocytes (Fig. 1) [2,3].
Osteoblasts line the surface of bone and synthesise, transport and arrange the proteins of the organic matrix. Osteoblasts express various receptors for many hormones, e.g. parathyroid hormone (PTH), and vitamin D, oestrogen and androgens, cytokines and growth factors (BMPs, TGF-β, IGF, endothelin-1, FGF and Wnt) all of which control osteoblastic functions [2,4–6].
The transcription factor RUNX-2, or core-binding factor α1, drives the expression of most genes associated with osteoblast differentiation [2–4,7]. It is activated by many growth factors, e.g. FGF, platelet-derived growth factor, IGF and TGF-β. RUNX-2 also up-regulates expression of other transcription factors such as osterix, another essential factor in the control of osteogenesis .
Osteoclasts mediate bone resorption, arising from haematopoietic precursor cells of the monocyte-macrophage lineage [1–4]. Osteoclast differentiation, maturation and activation depends on cytokines and growth factors including interleukin (IL)-1, IL-3, IL-6 and IL-11, TNF, granulocyte-macrophage colony-stimulating factor (GM-CSF), M-CSF, PTH, activated vitamin D, and thyroxine [1,2,4].
Both bone stromal cells and osteoblasts produce important factors mediating bone metabolism, e.g. RANK, RANKL, and OPG. RANK is a member of the TNF family of receptors, mainly expressed on the cell surface of osteoclast precursors. RANKL is expressed on the surface of osteoblasts and bone marrow stromal cells, and is released by activated T cells [1,2,4]. The major role of RANKL is stimulation of osteoclast formation, fusion, differentiation, activation and survival. Most of the systemic osteotropic factors such as PTH, activated vitamin D and prostaglandin E2 (PGE2) induce osteoclast formation by increasing the expression of RANKL on stromal cells and osteoblasts rather than a direct effect on the osteoclast precursors .
When RANKL binds the RANK receptor on osteoclast precursors, it induces osteoclast formation. The soluble protein OPG, acting as a decoy RANK receptor, inhibits osteoclast differentiation and activation [2,4]. The ratio of RANKL to OPG regulates osteoclast activity, allowing osteoblasts and stromal cells to regulate osteoclast function, influencing bone homeostasis.
Wnt proteins are primarily involved in developmental control of body axis symmetry and branching morphogenesis in utero[8,9]. In mature tissues, Wnts are involved in self-renewal of stem cells and maintenance of many normal tissues, as well as oncogenesis . Disruption of Wnt signalling results in limb defects in the developing embryo. Adult bone remodelling is also affected by defects in Wnt antagonists .
Human and mouse studies suggest that Wnt signalling increases bone mass at least in part by stimulating osteoblastogenesis. There appears to be a temporal importance of Wnt signalling in osteoblast differentiation. Disruption of Wnt signalling by expression of the Wnt antagonists DKK-1 and -2 blocks osteoblast differentiation in immature osteoblasts, but is required to promote terminal differentiation in late-stage osteoblasts. Mesenchymal stem cells from which osteoblasts derive can differentiate into several cell types, and there is considerable evidence that Wnt signalling stimulates osteoblastogenesis and represses alternative differentiation pathways . Wnt signalling increases the expression of osteoblastogenic transcription factors such as RUNX-2 and osterix, possibly by direct binding of a β-catenin/T cell factor complex to the RUNX-2 promoter . Recent studies suggest that Wnt signalling increases the growth rate of undifferentiated and proliferating osteoblast precursor populations, and inhibits osteoblast apoptosis, increasing numbers and survival .
Wnt activation might also affect bone formation by increasing the mineralizing activity of osteoblasts. Activation of β-catenin in osteoblasts specifically increases the expression of type I collagen (a major component of the organic matrix). It also affects osteoclast function by increasing osteoblastic expression of OPG, decreasing osteoclast differentiation and bone resorption .
Skeletal remodelling involves substitution or replacement of packets of bone and occurs continuously throughout life as a means of preserving the mechanical integrity of the skeleton. During growth it contributes to bone maturation. In the adult it provides metabolically active tissue for calcium homeostasis, eliminates avascular, necrotic bone compartments and prevents fatigue by local repair of micro-cracks and fractures [2,3].
Bone remodelling is characterized by the activation-resorption-formation sequence, which has a duration of 3–4 months in humans. It is initiated with the activation of resting cell populations on or near a bone surface. Osteoclasts are the first to invade the area and resorb bone, closely followed by osteoblasts that fill the excavated site with new bone matrix. Bone formation and resorption are both temporally and quantitatively coupled in optimal conditions, such that the activity of osteoclasts and osteoblasts is tightly coordinated. The remodelling rate varies from bone to bone and with age. It is activated by various growth hormones, thyroid and PTH, and inhibited by calcitonin, cortisone and possibly calcium .
EFFECTS OF HORMONES AND AGEING ON BONE
Peak bone mass is achieved early in adulthood and is related to nutrition, level of physical activity, age and hormonal status. Men experience a gradual age-related loss of bone mineral density (BMD) of 7–12% per decade after 30 years of age. Bone turnover in males is active before 25 years of age, decreases rapidly up to 40 years of age, when it slows reaching a nadir at 55–60 years old. After 60 years old, bone resorption, as measured by bone turnover markers, increases (as does remodelling rate), whilst bone formation remains stable, therefore leading to an overall negative balance [3,11,12]. However, osteoporosis in men is rarely encountered in men aged <70 years.
Balanced bone remodelling depends on appropriate hormonal signals. As previously noted, both osteoclasts and osteoblasts have receptors for oestrogens and androgens [5,6,13]. Androgens have direct actions on osteoblasts, inducing proliferation and differentiation, and inhibiting apoptosis . Testosterone was thought to be most important in men for control of bone metabolism, but recent evidence suggests that oestrogens play the major role, especially in elderly men, in controlling bone resorption [5,6,14,15].
The age-related decline in BMD might be due to reduced oestradiol and testosterone bioavailability with age-related increases in sex hormone-binding globulin. This reduces testosterone available for conversion to oestradiol through aromatization, decreasing the protective effects of oestrogen. Thus bone resorption is increased, reducing BMD [5,14,15].
Oestrogen might act in concert with paracrine factors secreted by osteoblasts, reducing pro-resorptive cytokines such as IL-1 and IL-6, TNF-α, M-CSF and PGE2, and increasing the anti-resorptive cytokine TGF-β. Oestrogen appears to increase OPG production and down-regulates the expression of RANKL, altering the resorptive function of osteoclasts via the RANKL/OPG system [13,16].
Androgen deprivation therapy produces significant osteoporosis, reducing BMD by 3–7% per year. This exacerbates pre-existing bone loss, leading to osteopenia or osteoporosis in nearly 90% of patients after 1 year of treatment, and a five-fold increase in skeletal fractures for men treated with androgen deprivation compared to age-matched controls [5,15]. Whether or not this active, fertile ‘soil’ increases the risk of bone metastases in prostate cancer is currently unknown.
Androgen deprivation therapy has been increasingly broadened and is now prescribed for locally advanced disease and biochemical relapse after local therapies. Many men are now being treated at an early age for longer periods [15,17]. The potential and actual major public health cost due to osteoporotic secondary complications in patients with prostate cancer is increasingly recognized.
MOLECULAR BIOLOGY OF PROSTATE CANCER METASTASES: CROSS-TALK AND INTERACTION WITH THE BONE MICROENVIRONMENT IN METASTATIC DISEASE
Effects of bone on prostate cancer
The bone microenvironment has direct effects on prostate cancer cells, which may explain the tropism of prostate cancer for bone. Many of the factors discussed previously produced by bone-marrow stromal cells, osteoclasts and osteoblasts, also affect the prostate cancer cells (Fig. 2).
Various cytokines and growth factors produced by or released from the bone tissue interact with their receptors on the prostate cancer cell (shown as IGFR for the IGF receptor and the Wnt receptor ‘frizzled’ and its co-receptor LRP5 and 6 in Fig. 2). Once stimulated, these receptors act via various intracellular signalling pathways such as the SMAD, androgen receptor (AR)-associated protein 55, transducers and activators of transcription (STAT), phosphatidylinositol 3′-kinase/protein kinase B (PI3K/AKT) and mitogen-activated protein kinase signal transducers. These act on various downstream effectors but many interact with the AR itself, which once activated translocates to the nucleus and activates the androgen response elements to initiate transcription of various androgen-sensitive genes. β-catenin, which is increased when the Wnt pathway is activated, controls transcription factors of lymphoid enhancer-binding factor/T cell factor family (LEF/TCF). β-catenin is targeted for constitutive degradation by glycogen synthase kinase 3β (GSK-3β). In the absence of Wnt ligand, GSK phosphorylates the bound β-catenin, which directs it for ubiquination and subsequent proteosomal degradation, thereby maintaining low levels of cytoplasmic β-catenin. The PI3K/AKT pathway can inhibit the activity of GSK-3β, thereby preventing degradation of β-catenin and increasing the levels. E-cadherin is involved in cell-cell adhesion and is down-regulated in metastatic prostate cancer. It also has β-catenin complexed to it and therefore when down-regulated might increase the pool of available β-catenin intracellularly. Phosphatase and tensin homologue (PTEN) causes cell-cycle arrest and apoptosis, as well as inhibition of cell motility. PTEN function is often down-regulated or lost in prostate cancer, which can increase the activity of the PI3/AKT pathway.
Factors specific to bone are involved in chemotaxis and attachment of the prostate cancer cell to bone. The bone protein SPARC has been implicated as an important chemotactic factor, as has the cytokine CXCL12, also known as stromal-derived factor. These, and specific integrins (some of which are up-regulated by TGF-β) might mediate the initial attraction and attachment of prostate cancer cells to the bone tissue, and promote metastatic deposit growth [18–20].
The bone matrix has several growth factors bound within its structure, and once released upon bone resorption, might promote growth of the tumour. These include IGF, TGF-β, BMPs and FGF.
The IGFs are abundant in bone and are potent mitogens stimulating the growth of tumour cells. Once bound, IGFs are released via bone resorption, and they can enhance metastasis in two ways, i.e. by increasing cell numbers (proliferation), and by attracting the tumour cells to bone (chemotaxis) . Prostate cancer cells can also increase IGF levels through degradation of the IGF binding proteins (BPs), potentiating these effects. IGF-1 activates the transcriptional targets of the AR via the PI-3K/AKT pathway, as well as several anti-apoptotic mechanisms .
Bone is a rich source of TGF-β; studies indicate that low levels of TGF-β result in cellular proliferation of prostate cancer . High levels of TGF-β paradoxically inhibit proliferation.
EFFECTS OF PROSTATE CANCER ON BONE
Prostate cancer cells alter bone homeostasis by secreting factors directly affecting osteoblast function, or influencing bone formation indirectly by modifying the bone matrix or microenvironment. Cancer cells synthesise and deposit bone matrix proteins such as osteopontin, osteocalcin, osteonectin and bone sialoprotein within the bone. Through this osteomimetic ability they might directly contribute to bone formation [16,23].
The prostate cancer cells produce both pro-osteoblastic and pro-osteoclastic factors. Some factors can function in both manners, depending on the timing of production or concentration. Whilst radiological secondaries appear osteoblastic, evidence from histology and bone resorption and formation markers indicate that they are mixed osteoblastic and osteolytic lesions. Osteoblastic metastases form on trabecular bone at sites of previous osteoclastic resorption, and are characterized by the weak woven bone tissue predisposing the site to fracture. The increased bone production is via an overall increase in bone remodelling with induction of osteoblastic-mediated bone formation outweighing osteoclastic resorption [7,18,24]. The exact mechanisms by which this occurs are likely to be many and are still poorly understood.
Pro-osteolytic factors produced by prostate cancer cells
As noted, it appears that osteolysis is required before osteoblastic bone deposition in metastatic deposits. The cancer cells produce several pro-osteolytic factors that can enhance this initial bone resorption (Fig. 3).
RANKL stimulates osteoclast differentiation and action, and decreases apoptosis. OPG acts as a decoy receptor for RANK. The balance between OPG and RANKL is critical in controlling osteoclast activity. The prostate cancer cells produce factors that can stimulate or inhibit the activity and regulation of osteoclasts.
Prostate cancer cells produce RANKL and can directly initiate osteoclastogenesis and therefore stimulate bone resorption [16,18,23–25]. Up-regulation and expression of RANKL by prostate cancer cells and osteoblasts is controlled by several factors produced by the prostate cancer cells themselves, and therefore might act in a paracrine and/or autocrine fashion. These factors include PTH-related protein (PTHrP), IL-6 and IL-1, and PSA.
PTHrP is an endocrine hormone that evokes the same biological activity at the PTH receptor as PTH, increasing bone resorption. PTHrP is produced by prostate cancer cells and leads to the expression of RANKL on bone marrow stromal cells, inducing the formation of osteoclasts and bone resorption. This releases among other factors, TGF-β, which further increases PTHrP production by prostate cancer cells [4,24,26]. PTHrP might protect prostate cancer cells and osteoblasts from apoptosis, and act as a mitogen to promote tumour growth in addition to its osteoclastogenic properties .
Other potent osteoclastogenic factors produced by prostate cancer cells include IL-1 and IL-6 [16,18,26]. The latter is a potent stimulator of osteoclast formation and can enhance the effects of PTHrP on the formation of osteoclasts. Elevated serum levels of IL-6 correlate strongly with objective markers of prostate cancer morbidity, and suggest that it might be useful as a marker of prostate cancer activity and possibly disease progression [4,16,18,26]
Matrix metalloproteinases (MMPs) promote osteolysis and possibly metastasis by degrading bone matrix, and are secreted by prostate cancer cells (Fig. 4). MMP-2 and MMP-9 blood and urine levels are increased in patients with prostate metastases. MMPs are also active during osteoclast recruitment to sites of bone remodelling . The mechanism by which prostate cancer produces MMPs and induces bone resorption is not clear. It might involve induction of osteoclastogenesis, as inhibition of MMPs reduced the number of osteoclasts associated with prostate tumour growth in human bone implants in an experimental mouse model [18,27].
Many of the effects of prostate cancer cells on the extracellular matrix pertain to metastasis and epithelial-to-mesenchymal transformation, the latter being necessary for the successful spread of cancer cells from the prostate to the bone metastatic site.
The kallikrein-related proteases, a family of serine proteases, have a specific involvement in both normal prostate and prostate cancer. PSA (or KLK3) is one such kallikrein-related protease. PSA hydrolyses the seminal vesicle proteins, seminogelin I and II, in ejaculate, liquefying the seminal clot . PSA has recently been shown to decrease OPG mRNA expression and increase RANKL mRNA expression, suggesting that PSA might induce osteoclast formation 
Pro-osteoblastic factors produced by prostate cancer cells
Many products of prostate cancer promote the hallmarks of the osteoblastic reaction, i.e. increased osteoid surface, osteoid volume and mineralization rate (Fig. 5). Studies suggest a positive association between the presence of metastatic prostate cancer and raised OPG levels. Whilst most OPG is likely to be produced by bone marrow cells, prostate cancer cells themselves have also been shown to produce OPG .
OPG in human prostate cancer cells has also been shown to be a survival factor due to its ability to bind a TNF-related apoptosis-inducing ligand (TRAIL) suppressing apoptosis. Production of OPG might therefore be a strategy for survival by providing a decoy target for TRAIL produced in and around tumour foci by patient monocytes and other cell types [24,26,30].
RUNX-2 controls transcription and when activated increases osteocalcin, bone sialoprotein, osteopontin, alkaline phosphatase and type I collagen. Protein kinase C is another intracellular messenger transduction system. Osterix is another transcription factor similar to RUNX-2.
BMPs and TGF-β are members of the TGF-β superfamily. BMPs have many functions in bone, including apoptosis, differentiation, proliferation and morphogenesis. Target genes of the BMPs in osteoblasts include OPG and RUNX-2 [18,24]. BMPs can induce uncommitted stem cells and myoblasts to express osteoblast characteristics, such as alkaline phosphatase or osteocalcin. Their osteogenic properties appear to be specific to the differentiation stage of target cells. They do not stimulate mature osteoblasts or fibroblasts to increase expression of these proteins [18,24]. Prostate carcinoma cells produce increasing levels of BMPs as they progress to a more aggressive phenotype, suggesting that up-regulation of BMP expression by cancer cells in bone is a critical component in the development of osteoblastic lesions [7,18,24].
TGF-β up-regulates RUNX-2 and similar controllers of osteogenesis. TGF-β levels are higher in patients with prostate cancer bone metastases than in those without [7,26]. TGF-β1 might act directly on the stroma, regulating angiogenesis and tumour progression as well as inducing differentiation in bone cell populations, inducing growth/survival factors by/for tumour cells, and regulating tumour cell attachment to matrices [16,26].
Serum levels of IGF-1 correlate with the risk of prostate cancer and the IGF-1 receptor is required for neoplastic transformation . Serum levels of IGF-BPs are inversely related to the risk of developing prostate cancer. IGF-1 binds receptors on osteoblasts activating RUNX-2 [7,18]. This might provide a link between IGF-1 and the development of osteoblastic metastases.
Endothelin-1 is a potent vasoconstrictor that belongs to a family of three 21-amino-acid peptides. The effects of endothelin-1 are mediated mainly through the endothelinA receptor. Endothelin-1 has been detected in osteocytes, osteoblasts, osteoclasts and vascular endothelial cells; it stimulates mitogenesis in osteoblasts which have both endothelinA and B receptors. It also enhances the effects of other osteoblast-stimulatory factors such as BMP-7 to induce bone formation [4,18,24,32].
Prostate epithelium produces endothelin-1 and has high-affinity receptors throughout the gland. Endothelin-1 levels are increased in patients with osteoblastic metastases from prostate cancer . Tumour-produced endothelin-1 might have paracrine (on bone cells) and/or autocrine effects (on tumour growth and apoptosis). Exogenous endothelin-1 increases prostate cancer cell proliferation and augments the effects of IGF-1, platelet-derived growth factor, EGF and FGF-2. It has also been shown that endothelin-1 production is increased by prostate cancer cells in contact with bone [7,18,24,32]. Substantial data associates endothelin-1 with osteoblastic metastases in prostate cancer.
VEGF is a key regulator of physiological and pathophysiological angiogenesis; it promotes endothelial cell proliferation, survival and migration. The effects of VEGF are mediated via several receptors. The two key receptor tyrosine kinases are VEGFR-1 and VEGFR-2. VEGF has previously been shown to regulate bone formation indirectly by controlling vascularity within the developing growth plate . VEGF has a direct effect on bone formation by stimulating migration and proliferation of human osteoblasts .
Prostate cancer cells produce VEGF, facilitating tumour growth by enhancing angiogenesis and possibly migratory ability. The VEGF produced by tumour cells binds neuropilin-1 on pre-osteoblasts, inducing osteoblast differentiation and, in conjunction with other tumour-related pro-osteoblastic factors, results in osteosclerotic lesions [7,33].
PSA and KLK2 mediate cell proliferation in both the normal and malignant prostate by interactions with the IGF axis. PSA has potent mitogenic activity for osteoblasts. This might be through elevation of IGF-1 acting as an osteoblastic growth factor increasing bone deposition. PSA might achieve this by degrading IGFBP-3, thereby increasing the bioavailability of IGF-1. Another pathway might be through PSA activating the latent form of TGF-β[4,26,28,29,34]. PSA might also have a direct role in modulating genes involved in bone remodelling, including up-regulation of genes such as RUNX-2, osteopontin and TGF-β.
PSA increases bone deposition by cleaving PTHrP ; the latter has many effects on the bone and prostate cancer cell populations, but its degradation might reduce bone resorption, thereby tending toward increased deposition [4,7,28,29,34,35].
Urinary plasminogen activator is another serine protease produced by prostate cancer that acts as an osteoblast growth factor. This might be due to increasing IGF-1 levels by hydrolysing IGFBP-3 and activating latent growth factors such as TGF-β similarly to PSA [4,7,18].
The Wnt pathway has been implicated in the development of osteoblastic metastases in prostate cancer in several ways. Wnt signalling by the prostate cancer cells might promote osteomimicry. Expression by tumour cells of BMP, osteopontin, the osteopontin receptor CD44 and RUNX-2, and the ability to produce mineralized matrix has been reported. The Wnt pathway might be involved in this osteomimicry, in that both osteopontin and CD44 are Wnt-regulated genes, and the canonical Wnts stimulate osteoblast mineralization and differentiation [8,9].
Other evidence indicating the involvement of the Wnt pathway is conflicting, showing the complexity of the Wnt pathway and its role in bone metastases. Hall et al. suggested an elegant mechanism of Wnt involvement in prostate cancer osteoblastic metastases. They proposed that the involvement of Wnt agonists and antagonists is integrated with many of the previously mentioned factors to produce a phasic model; Fig. 6 provides an overview of interactions between prostate cancer cells and bone.
In this integrated phasic model of metastatic prostate cancer, initially the prostate cancer cells target bone and establish metastases and produce pro-osteolytic factors such as RANKL, IL-6, PTHrP and the Wnt antagonist and inhibitor of osteoblastic activity DKK-1. The osteolytic activity releases growth factors stored in the bone, modifying the bone microenvironment, which then alters the prostate cancer phenotype. Tumour cells then produce osteoblastic factors including BMP, PTHrP (which can act as an anabolic factor by inhibiting osteoblastic apoptosis) and factors which block osteoclastic activity, such as OPG. DKK-1 expression also decreases, activating the Wnt pathway, which increases osteogenesis. This therefore transforms an initial osteolytic phenotype to an osteoblastic one. As the deposit expands, it outgrows its initial blood supply, producing hypoxia, inducing VEGF and endothelin-1 expression to promote angiogenesis. Both the cytokines also have osteoblastic activity, enhancing bone production . This model might explain why apparently conflicting data has been reported by various research groups, due to the phasic activity of the factors involved.
As is obvious from this discussion and the figures, the complex interplay of the pathways involved is significant. As new technologies, such as microarrays, are developed and are used more extensively, this complexity is likely to continue to increase. A continued challenge for researchers is to identify and characterize the important pathways and components of the bone/cancer interplay. Only through this effort will effective therapeutic interventions become available to affect this pre-eminent issue in men's health.
CONFLICT OF INTEREST
None declared. Dr Vela is supported by a research grant from the Australasian Urological Foundation. Associate Professor Gardiner is supported by the Australian National Health and Medical Research Council, project grant no. 276415.