RARγ: The Bone of Contention for Endothelial Cells in Prostate Cancer Metastasis

Bone metastasis is associated with reduced overall survival in patients with prostate cancer. Aberrant bone formation associated with advanced prostate cancer can contribute to the progression of bone metastases. Among the most significantly overexpressed secreted factors within the metastatic tumor microenvironment that contribute to osteogenic differentiation are TGFβ and its family members, bone morphogenetic proteins (BMP; ref. 1). Elevated bone turnover by metastatic prostate cancer cells involves bone resorption by osteoclasts through paracrine TGFβ signaling and enhanced matrix deposition by osteoblasts through paracrine BMP signaling. Several osteoblast differentiation factors like alkaline phosphatase, osteopontin, and osteocalcin are found to be enriched in the prostate cancer cells themselves (2). This expression of a bone-like phenotype, referred to as osteomimicry, by osteotropic tumor cells is thought to support adaptation in the bone microenvironment. There are multiple mechanisms of elevated tissue ossification; one in particular involves endothelial–mesenchymal transdifferentiation (endoMT) that is followed by osteoblastic differentiation of the mesenchymal cells (3). In this two-step process, endoMT can be induced by inflammatory cytokines such as IL1β and TNFα for subsequent BMP-mediated osteoblastic differentiation, respectively (3). In this issue, Yu and colleagues add to the understanding of how retinoic acid receptor (RAR), specifically RARγ, mediates BMP signaling in the prostate cancer bone niche (4).

The nuclear receptor family of RARs have a long history of regulating cancer cell survival and stemness. They form dimers with the retinoid X receptors (RXR) that, when bound to a ligand, can signal through response elements to transcriptionally activate downstream genes. Differential expression of RAR/RXRs as well as disruption of retinoic acid signaling is common to the etiology of various cancers. In fact, the cognate ligand all-trans retinoic acid (ATRA) is found at concentrations around an order of magnitude lower in prostate tumors compared with that found in noncancerous prostate tissues (5). Within the bone microenvironment, the depletion of retinoic acid signaling drives the osteoblastic phenotype often associated with prostate cancer bone metastasis (6). Then, intuitively, one would expect the activation of RARγ signaling to limit osteoblastic differentiation. In line with this hypothesis, Yu and colleagues presented evidence that a RARγ agonist can inhibit BMP4-induced effects, limiting osteoblastic transdifferentiation in a variety of endothelial cell lines irrespective of their tissue origin (4). The authors tested treatment with palovarotene, an orally bioavailable RARγ–specific agonist that has been reported to inhibit BMP signaling to limit heterotopic ossification observed in fibrodysplasia ossificans progressiva (7). In the context of prostate cancer, it was similarly found to prevent chondrogenesis by vascular endothelia (4). Mechanistically, BMP signaling promoted Smad1 phosphorylation and interaction with RARγ following nuclear translocation, and palovarotene supported the recruitment of Smurf1, an E3-ubiquitin ligase, to the Smad1–RARγ complex, facilitating its proteosomal degradation (Fig. 1; ref. 4). The success of palovarotene in blocking osteoblast transdifferentiation of bone endothelia for prostate cancer metastasis could be generalized to target other instances of osteoblastic bone metastasis, such as for carcinoid, small cell lung cancer, Hodgkin's lymphoma, or medulloblastoma.

The osteoblastic response is coupled to prostate tumor expansion, and the association is more than a mere correlative observation. There are multiple bone niche–derived mediators of metastatic tumor progression. FGF2, periostin, and Notch signaling are among the paracrine/juxtacrine signaling mediators that cooperate to support prostate cancer progression. On the basis of the reduction in tumor volume observed in mice treated with palovarotene, the authors concluded that the effect was at least in part due to inhibition of osteoblastic transdifferentiation of bone endothelial cells rather than a direct inhibition of cancer cell growth (4). While there was no change in cancer cell proliferation, reduction in endothelial cell proliferation and angiogenic capacity is a plausible contributor to the observed reduction in the osteoblastic phenotype. This warrants the question of whether preventing lineage conversion of endothelial cells, which supports angiogenic potential, would serve to promote or limit the metastatic potential of tumors. However, angiogenesis within the primary tumor microenvironment and the sinusoidal vasculature of the bone supports tumor expansion differently. So blocking endothelial-to-osteoblast conversion with palovarotene could complement standard-of-care therapies following metastatic progression.

Before making that leap, one must consider other cell types involved in osteogenesis and other roles of RARγ signaling. There is the possibility of cell types other than endothelial cells differentiating into osteoblasts. Bone marrow adipocytes are a strong contender, because elevation of adiposity with age in men supports tumor progression. Incidentally, palovarotene was found to promote adipocyte transdifferentiation into osteoblasts (8). As there was a reduction in osteoblast markers with palovarotene treatment, the contribution of marrow adipocytes was considered to be minimal in mouse models (4). Further, retinoic acid signaling plays a role in the immune microenvironment, as it has been demonstrated to promote the expansion of regulatory T cells, limiting immune surveillance. The use of xenograft prostate cancer models by Yu and colleagues necessitated the use of immunocompromised mice, so the effects of palovarotene in immunocompetent models requires further investigation. ATRA inhibited BMP4-induced bone formation and prostate cancer tumor growth under castrate conditions (4). In patients with intact androgen signaling, its effects on RARγ may be muted because testosterone is a known promoter of RARγ expression (9). In addition, retinoic acid signaling in prostate cancer cells is known to block androgen synthesis via inhibition of 5α-reductase (10), suggesting a mutually inhibitory feedback loop. A concentration-dependent biphasic effect of directly promoting prostate cancer growth at lower concentrations could make RAR agonist treatment untenable as a single agent. The observation that RARγ activity suppresses prostate cancer bone metastasis raises the possibility for the use of an agonist like palovarotene for a narrower patient population in the context of standard of care. These are aspects an interventional clinical trial for patients with advanced prostate cancer with palovarotene may help clarify. Thus far, palovarotene has been tested in relatively young people without cancer.

Biomarkers of response can facilitate the clinical translation of new treatment strategies. Palovarotene was found to inhibit expression of tenascin C (4), a secreted factor identified to be expressed by endothelial cells undergoing osteoblastic differentiation. In a palovarotene cancer trial, changes in plasma levels of tenascin C could support on-target monitoring of palovarotene response. Other blood-based markers could include similar downregulation of osteopontin (SPP1), calcitonin gene–related peptide (CGRP, CALCA), endothelin 1 (EDN1), and possible elevation of plasminogen activator inhibitor 1 (PAI1, SERPINE1) by agonizing retinoic acid signaling. The marker panel will clearly be subject to the standard-of-care strategy given in conjunction with palovarotene. Androgen receptor antagonists, docetaxel, and radiopharmaceuticals are viable complements to palovarotene treatment that need to be considered. Ultimately, dose-finding studies will be critical for palovarotene efficacy in men with prostate cancer bone metastasis.