Retinoids and demethylating agents- looking for partners.
M. Sporn
DOI: https://doi.org/10.1093/JNCI/92.10.780
2000-05-17
Journal of the National Cancer Institute
Abstract:The concept that carcinogenesis may be the result of a deficiency or a failure of the action of substances that are negative regulators of control of proliferation and growth of cells is hardly new (1) and certainly antedates current interest in the mutation of tumor suppressor genes in the causation of cancer. Thus, in their classic article on the role of retinoid deficiency in control of epithelial proliferation and differentiation, Wolbach and Howe(2) noted that retinoid-deficient epithelial tissues had a premalignant phenotype that was characterized by enhanced mitotic activity and loss of differentiation. Retinoids may indeed be considered prototypes of such negative regulators of epithelial cell growth, and there is an immense literature on their mechanism of action and potential use for cancer prevention [reviewed in(3)]. However, although there have been a few major advances in the clinical use of retinoids to prevent or to treat several types of cancer(4–6),the widespread practical use of these agents in the field of oncology, either preventively or therapeutically, has yet to be achieved. Thus, while the proven efficacy of selective estrogen receptor modulators (SERMs) as agents to prevent breast cancer (7,8) justifies the present large clinical Study of Tamoxifen and Raloxifene (STAR) to compare the relative benefits of two SERMs (tamoxifen and raloxifene) without a placebo control, the practical use of retinoids to prevent breast cancer in postmenopausal women remains to be demonstrated, in spite of an impressive amount of data from animal experiments that strongly suggest that prevention should be possible. Retinoids are still a hope, rather than a practical reality, in most of clinical oncology. The report by Widschwendter et al. (9) in this issue of the Journal suggests one possible mechanism for this lack of clinical efficacy of retinoids and indicates a practical strategy to make retinoids more effective, both as preventive and as therapeutic agents. The antiproliferative and differentiative effects of retinoids are mediated through the actions of six known receptors of the steroid receptor superfamily (10).These are the three “classical” retinoic acid receptors (RARa, -b, and -g), as well as the three “orphans” (RXR-a, -b, and -g) (now “adopted,” since their ligand, 9cis-retinoic acid, is known). All six of these receptors are ligand-activated transcription factors. With respect to human carcinogenesis, there has been particular interest in the loss of function of RAR-b as a causative factor in the genesis of many common types of cancer, particularly in the lung, oropharynx, and breast, and it has been suggested that the RARb gene should be considered as a tumor suppressor gene (11–16).Although it has been shown that 13cis retinoic acid can induce increased expression of RARb in lesions of the oropharynx and can prevent recurrence of second primary tumors at this site (5,17), retinoids have not yet had a useful impact on the treatment of carcinoma of the lung or breast. In their report, Widschwendter et al. (9) have examined the role of methylation of the RARb2 gene (the gene for one of the four transcripts encoded by RARb) in breast carcinoma cells in conferring refractoriness to the ability of alltrans-retinoic acid (ATRA) to induce expression of RARb2 messenger RNA (mRNA); it is well known that the RARb2 gene has a retinoic acid response element that normally mediates such induction. In four commonly used human breast cancer cell lines, the authors have found that ATRA could induce expression of RARb2 mRNA only after these cells were treated with the demethylating agent 5-aza-2 8-deoxycytidine (Aza-CdR). Sequencing studies in s veral breast cancer cell lines, as well as in a series of primary breast cancer biopsy specimens, showed methylation of the RAR-b2 gene in most of the cell lines, as well as in a substantial number of the tumors. Expression of RARb2 mRNA was shown to be repressed in samples in which this gene was methylated; conversely, the RARb2 gene from non-neoplastic breast tissue was unmethylated and expressed. What is the molecular basis for these new findings, and what are their overall implications? Alterations in DNA methylation are now perceived as an important epigenetic mechanism that contributes significantly to the process of carcinogenesis (18– 20). Although tumor suppressor genes may, in some instances, be inactivated by mutation, it is increasingly recognized that their inactivity may also be the result of transcriptional inactivation by epigenetic mechanisms (18–21).Indeed, the discovery of numerous hypermethylated promoters of tumor suppressor g nes has moved DNA methylation from obscurity to recognition as an important mechanism in the carcinogenic process. Recent molecular studies provide important insights as to mechanism. Thus, it has been shown that MeCP2, an abundant mammalian protein that binds to methylated CpG sites in DNA, is also a transcriptional repressor (22). Furthermore, transcriptional repression by the joint action of methylated DNA and its binding protein, MeCP2, involves their recruitment of histone deacetylase to this complex, resulting in chromatin condensation and gene silencing(23–25).Several regulatory genes that are silent in cancer cells can be reactivated by combined treatment with the demethylating agent Aza-CdR together with the histone deacetylase inhibitor trichostatin A (26). Although it has been shown in both transgenic and chemically induced mouse tumor models that Aza-CdR can suppress carcinogenesis (27,28),this agent is too toxic for practical use as a chemopreventive agent. There thus is a major need for the development of less toxic demethylating agents, which most likely will require the addi-