Metastatic PCa cells can undergo a phenotypic switch towards an OB-like phenotype by synthesizing and secreting osteocalcin, osteopontin (22), osteoprotegerin (OPG), etc. OPG is an anti-osteolytic agent that can antagonize receptor activators of nuclear factor κB ligand (RANKL, central to OC differentiation) (23). Primary PCa tissues can also express OPG, and it has been observed that their OPG expression increases with elevated PCa grade (24). PCa cells with OB phenotype can facilitate osteogenic responses at metastatic sites, known as “osteomimicry” (10, 11). Previous studies suggested that PSA could potentiate osteomimicry of PCa cells. Stable expression of endogenous active PSA in non-PSA-secreting DU-145 cells could induce a phenotypic switch towards OB phenotype (25). Also, Chiao et al. found that PSA could induce up to a 50% increase in endothelin-1 secretion by DU-145, PC-3 and JCA-1 cells in a dose-dependent manner, and ET-1 is recognized as a strong osteogenic factor (1). Another study showed that co-culture of androgen-dependent LNCaP cells and bone stromal cells could induce the formation of androgen-independent C4→C4-2→C4-2B cell sublines step by step — with their gradually increased PSA secretion (26), their expression of osteocalcin and osteopontin was simultaneously elevated (27), and their capabilities to induce osteoblastic bone metastases in immunodeficient mice were also gradually augmented (27, 28). These results substantiate that PSA can promote the phenotypic switch of PCa cells towards OB phenotype. However, the mechanisms whereby PSA induces osteomimetic properties in PCa cells remain elusive.

Overwhelming evidences substantiate that most osteolytic factors, such as 1, 25-dihydroxy vitamin D3, parathyroid hormone (PTH), IL-6, etc., cannot act directly on OCs due to the absence of corresponding receptors in OC precursors, immature and mature OCs (29–31). Instead, these factors mainly act on OBs and bone stromal cells to produce RANKL and macrophage colony-stimulating factor (M-CSF). Subsequently, RANKL and M-CSF act on their receptors expressed on surfaces of OC precursors, thereby inducing OC proliferation and differentiation (6). In other words, OC differentiation, maturation and secondary osteolysis are activated by OBs, and the biological effects of most osteolytic factors are mediated by OBs (29–31). If RANKL produced by OBs is deactivated by its antibodies, the activating effects of osteolytic factors on OCs are abolished (32).

In a study by Goya et al. where differentiation of OC precursor RAW264.7 was induced with RANKL, supplementation of RANKL alone in RAW264.7 culture media led to mature OCs on day 7. However, concurrent addition of PSA with RANKL in RAW264.7 culture media resulted in few mature OCs. Meanwhile, if PSA was added in culture media two days after supplementation of RANKL, only 60-70% of RAW264.7 cells became immature OCs on day 7; but if PSA was added five days after supplementation of RANKL, the maturation process of RAW264.7 cells towards mature OCs was unaffected. This phenomenon indicated that PSA could suppress differentiation of OC precursors, but yielded no inhibitory effects on mature OCs. The study also demonstrated that PSA could induce apoptosis of OC precursor RAW264.7 in a dose-dependent manner, and the apoptosis rate could be up to 43%. Moreover, PSA-induced apoptosis of RAW264.7 cells could be significantly reduced when PSA activity was antagonized by ACT, suggesting that the enzymatic activity of PSA promoted apoptosis of OC precursors (33). In summary, PSA can inhibit differentiation and induces apoptosis of OC precursors, therefore suppressing the osteolytic effects of OCs and enhancing PCa osteoblastic bone metastasis. However, currently little is known about the molecular or biochemical mechanisms underlying how PSA acts on OCs, and further studies are warranted.

PSA can potentiate OB proliferation. Yonou et al. reported a greater than 10-fold increase of TGFβ2 expression of human osteosarcoma SaOS-2 cells (originated from osteoblasts) after the addition of exogenous active PSA in culture media, and that SaOS-2 cell proliferation increased in a dose-dependent manner (up to 124-162%); on the contrary, deactivation of PSA using antagonist ACT or anti-PSA antibodies led to 65-75% decline of SaOS-2 cell proliferation. Additionally, they revealed that direct injection of active PSA in human bone grafts of NOD/SCID mice significantly increased the number of OBs and the volume of nascent osteoid in the bone grafts, which could also be antagonized by ACT (21). These results indicate that PSA-induced OB proliferation depends on the enzymatic activity of PSA.

PSA can activate OBs and enhance their osteogenic functions. Nadiminty et al. found that after the expression of endogenous active PSA in SaOS-2 cells, levels of multiple osteogenic factors were upregulated (34). For example, crucial osteogenesis-related transcription factor RUNX2 exhibited a 25-fold increase in expression, promoting the transcription of various downstream osteogenic factors (35). The expression of other osteogenic factors, such as osteocalcin, osteopontin, TGFβ2, BMP4 and BMP8, was also upregulated. However, interestingly, central OC differentiation factor RANKL was upregulated concomitantly, whereas OPG, the antagonist of RANKL with anti-osteolytic effects, showed a 95-fold decrease in expression. This finding suggests that after the expression of PSA in SaOS-2 cells, SaOS-2 cells demonstrate not only enhanced osteogenic functions, but also strengthened capabilities to activate OCs and secondary osteolysis — this might have a subtle connection with the phenomenon that during early PCa bone metastasis, osteogenic responses are preceded by osteolytic processes which are produced by OB-activated OCs (5, 36).

In another study, Yonou et al. stimulated MG-63 and SaOS-2 cells with exogenous active PSA, which led to a greater than 3-fold increase of OPG expression in both cell lines, while RANKL expression was downregulated. The results imply that exogenous active PSA enhances the osteogenic functions of OBs but yields no effects on their capabilities to activate OCs and secondary osteolysis (2). This is exactly the opposite of the results obtained by Nadiminty et al. as described above (34). To explain this difference, Yonou et al. proposed that exogenous active PSA stimulation was more consistent with the actual conditions of PCa metastatic sites where PSA stimulates OBs, compared to intracellular expression of endogenous PSA. Besides, a recent study found that BMP4 could induce transition of endothelial cells to OBs at PCa metastatic sites, and thus another possible mechanism whereby PSA promotes PCa osteoblastic bone metastasis could be inferred: PSA upregulates BMP4 expression and secretion of OBs, thereby facilitating endothelial-OB transition at PCa metastatic sites (37). In summary, PSA may promote PCa osteoblastic bone metastasis by potentiating OB proliferation and activation and enhancing their osteogenic functions. However, the molecular or biochemical mechanisms underlying the effects of PSA on OBs require to be clarified.

IGF-I/II are the most abundant osteogenic factors in bone matrix (38) and are components of IGF system together with 2 IGF receptors and 6 IGF-binding proteins (IGFBP) distributed in tissues and blood (39). IGFs can be synthesized by PCa cells and OBs and then secreted to bone matrix. They can augment osteogenic responses by stimulating the proliferation and activation of metastatic cancer cells and OBs (40, 41). IGFBP3 is the main binding protein of IGF-I, which can function as the transport carrier of IGF-I (42). On the other hand, IGFBP3 can also block the biological effects of IGF-I via blocking its binding with IGF receptors (43). PSA can degrade IGFBP3 and then decrease its binding with IGF-I to release more active IGF-I, thus potentiating IGF-I-induced OB proliferation and activation and PCa osteoblastic bone metastasis (44). Smith et al. analyzed IGFBP3 concentration at metastatic sites and serum PSA levels of six PCa patients, and found that IGFBP3 concentration negatively correlated with PSA level (45). In addition, Miyata et al. discovered that serum IGFBP3 level and IGFBP3/PSA ratio were significantly reduced in patients with progressive PCa (46). These studies substantiate the degradative effect of PSA on IGFBP3.

TGFβ2, synthesized by PCa cells and OBs and secreted to bone matrix, can promote proliferation and inhibit apoptosis of OBs, thereby enhancing osteogenic responses (41, 47). After being synthesized in cells, TGFβ first forms inactive “small latent” TGFβ via covalently binding with “latency-associated peptide”; then “small latent” TGFβ binds with latent TGFβ binding protein 1 (LTBP1) by a disulfide bond to form inactive “large latent” TGFβ, which is finally secreted to extracellular bone matrix and restored. To sum up, TGFβ synthesized and secreted by cells is inactive, and its activation requires the removal of LTBP1 and latency-associated peptide (48). Researchers noted that PSA could degrade the latency-associated peptide binding with TGFβ2 to accelerate the activation of TGFβ2 (49). Besides, PSA can stimulate OBs to express and secrete TGFβ2 (21, 34). Accordingly, PSA produced by metastatic PCa cells may potentiate OB proliferation and activation by facilitating the expression, secretion and activation of TGFβ2, which can further enhance PCa osteoblastic bone metastasis.

PTHrP is an important osteolytic factor secreted by cancer cells (41). Like PTH, it can act on cell membrane PTH/PTHrP receptors, activate adenylate cyclase and then induce downstream signal transduction (50). Importantly, PTHrP can act on OBs to increase their synthesis of RANKL and decrease their expression of OPG, thus promoting OC proliferation and differentiation (51). Moreover, PTHrP can bind with the receptors on OC surfaces and directly stimulate OC differentiation (52). During this process, the osteolytic effects of OCs usually outperform the osteogenic effects of secondarily activated OBs, and hence PTHrP secreted by cancer cells generally induces osteolytic bone metastases (8, 9). This can strongly explain the predominance of osteolytic bone metastases in breast cancer given that breast cancer cells secrete large amounts of PTHrP (53, 54). However, osteoblastic bone metastases are still predominantly observed in PCa, although metastatic PCa cells can also secrete fairly much PTHrP (55). We hypothesize that this discrepancy is attributed to PSA, which can cleave PTHrP at the position of its 23rd amino acid; the degraded and deactivated PTHrP fails to bind with PTH/PTHrP receptors on OB/OC surfaces, resulting in decreased osteolytic effects of PTHrP and relatively enhanced osteogenic processes at PCa metastatic sites (50, 56). Furthermore, Schluter et al. found that the N-terminal peptide of PTHrP cleaved by PSA could bind with and activate endothelin receptor, generating osteogenic effects similar to that of endothelin-1 (57). Therefore, PSA may augment osteogenic responses at PCa metastatic sites by degrading and deactivating PTHrP and suppressing its osteolytic effects.

uPA, uPA receptors and 2 uPA inhibitors constitute uPA system. uPA initially synthesized in PCa cells is an inactive single-chain proenzyme, known as a single-chain urokinase-type plasminogen activator (scuPA), which can be cleaved by proteases such as plasmin to form active uPA consisting of a light chain and a heavy chain. uPA can facilitate OB proliferation and activation by binding with uPA receptors on OB surfaces (41, 58). Besides, uPA is a serine protease that can degrade extracellular matrix and promote cancer cell spreading and metastasis (59). Intriguingly, uPA produced by metastatic PCa cells can lyse bone matrix and then accelerate PCa bone metastasis and bone remodeling (60). Moreover, uPA can activate TGFβ2 precursor to further augment osteogenic responses at PCa metastatic sites (61). Taken together, these findings reveal that uPA derived from PCa cells can potently induce PCa osteoblastic bone metastasis. Yoshida et al. found that PSA could cleave scuPA at lysine 158 to generate active uPA (62). Therefore, PSA may activate uPA to advance PCa osteoblastic bone metastasis.

Interactions prevail among these PSA-influenced factors involved in PCa osteoblastic bone metastasis. For instance, IGFBP3 is also a type V TGFβ receptor ligand that can compete with TGFβ2 to bind with its receptors. Hence when PSA secreted by metastatic PCa cells degrades IGFBP3 and decreases its concentration at metastatic sites, it can be extrapolated that the osteogenic effects of TGFβ2 would be in turn enhanced (63). Furthermore, IGFBP3 can be hydrolyzed by uPA derived from metastatic PCa cells, resulting in the release of active IGF-II and enhanced osteogenic responses (64). In addition, the ability of uPA receptors to bind with IGF-II receptors further facilitates cancer cell-OB and OB-OB interactions, which subsequently promotes the proliferation of both cancer cells and OBs (65). Such interactions may further intensify the possible effects of PSA in inducing PCa osteoblastic bone metastasis.