Downloaded from genesdev.cshlp.org on September 25, 2021 - Published by Cold Spring Harbor Laboratory Press PERSPECTIVE Orphan nuclear receptors—new ligands and new possibilities Bruce Blumberg1,3,4 and Ronald M. Evans1,2,4 1Gene Expression Laboratory, 2Howard Hughes Medical Institute, The Salk Institute for Biological Studies, La Jolla, California 92037 USA Nuclear hormone receptors comprise a large superfamily drostane receptor), PXR (pregnane X receptor), and SXR of ligand-modulated transcription factors that, in part, (steroid and Xenobiotic receptor)] with brief updates on mediate response to steroids, retinoids and thyroid hor- recent progress for FXR, PPAR␥, and LXR␣. To restrict mones and play key roles in development and body the scope of the review, we will focus only on orphans physiology (for review, see Beato et al. 1995; Kastner et forming heterodimers with RXR and studies that have al. 1995; Mangelsdorf and Evans 1995; Mangelsdorf et al. been reported in the last year. We will emphasize what 1995; Enmark and Gustaffson 1996; Willy and Man- has been learned about the nature of the ligands, new gelsdorf 1998). Ten years ago the first orphan nuclear modes of regulation, insights into physiology, and poten- receptors were isolated, raising the first portents toward tial new drug targets. a new era in physiology (Giguere et al. 1988). Although structurally related to the known receptors, no physi- ological ligands or activators were known for these or- RXR heterodimers and ligand dependence phan receptors which are more numerous (19 families) than receptors with known ligands (10 families). Orphan Although inherently promising, not all orphans (and per- receptors represent a diverse and ancient component of haps only a few) can be expected to represent unidenti- the nuclear receptor superfamily, being found in nearly fied endocrine systems. Therefore, a central and critical all animal species examined. Orphan nuclear receptors problem is identifying which of the many orphan recep- provide a unique and, until recently, largely untapped tors are good candidates to be ligand responsive. An in- resource to uncover regulatory systems that impact on teresting feature of the known, nonsteroidal ligand-acti- both health and human disease. Shortly after their iso- vated nuclear receptors is that they all require het- lation, strategies were devised to identify orphan ligands erodimerization with RXR, for high-affinity DNA using the receptor as a screening target. This process, binding. We and others speculated that RXR het- referred to as ‘reverse endocrinology,’ has led previously erodimerization might be a signature feature for one to the identification of ligands for RXRs [9-cis retinoic class of ligand-responsive receptors. Indeed, with the acid receptors (RARs), a.k.a. retinoid X receptors], PPARs identification of BXR, SXR, PXR, and CAR␤, all nuclear (peroxisome proliferator-activated receptors), FXRs (far- receptors now known to require heterodimerization nesoid X receptors), and LXRs (liver X receptors). with RXR for DNA-binding are ligand-responsive (Fig. 1). In a review eight years ago, Bert O’Malley predicted Thus, RXR is a common partner in at least 11 distinct that the field of orphan receptors would provide much signaling pathways. The activation state of RXR varies excitement for the future (O’Malley 1990). The promise among these heterodimers, adding to the complexity and of orphan receptors is that they provide an opportunity potential diversity of hormonal signaling. RXR can be to identify not only novel ligands but perhaps unsus- completely silent in nonpermissive heterodimers with pected classes of ligands and potentially new principles the thyroid hormone receptor (TR) or the vitamin D re- of physiology. Because all known receptors are impor- ceptor (VDR) (Kurokawa et al. 1994; Forman et al. 1995a) tant in the treatment of human disease, their signaling although it is freely activatable in permissive het- pathways have direct implications for drug discovery. erodimers with PPAR (Kliewer et al. 1992b), LXR␣ (Willy The purpose of this review is to highlight advances made et al. 1995), and FXR/RIP14 (Forman et al. 1995b; on four new orphan receptor-mediated signaling path- Zavacki et al. 1997). In the case of the RAR heterodimer, ways [BXR (benzoate X receptor), CAR␤ (constitutive an- RXR can be activated only after the addition of an RAR ligand (Forman et al. 1995a; Schulman et al. 1997). The ability of RXR to be activated in other heterodimers has 3Present address: Department of Developmental and Cell Biology, Uni- not been reported. The activation state of RXR in a par- versity of California, Irvine, California 92697-2300 USA. 4Corresponding authors. ticular heterodimer can have important consequences E-MAIL [email protected]; [email protected]; FAX (619) 455-1349. for treating disease. For example, in animal models for GENES & DEVELOPMENT 12:3149–3155 © 1998 by Cold Spring Harbor Laboratory Press ISSN 0890-9369/98 $5.00; www.genesdev.org 3149 Downloaded from genesdev.cshlp.org on September 25, 2021 - Published by Cold Spring Harbor Laboratory Press Blumberg and Evans hormone response element is as direct repeats (DR) of AGTTCA (rather than the canonical AGGTCA) sepa- rated by four (DR-4, BXR, PXR, SXR) or five nucleotides (DR-5, CAR). PXR, SXR, and CAR are highly expressed in the liver and respond to steroidal ligands. Together, they are the first new steroid-responsive nuclear recep- tors described since the isolation of the mineralocorti- coid receptor in 1987 and the first new steroidal ligands since aldosterone in 1952. CAR␤—a ligand inactivated receptor All of the ligand-dependent nuclear receptors described to date are activated by a ligand-induced association with coactivator proteins. Recent studies with CAR␤ ap- pear to reverse this paradigm (Forman et al. 1998). When isolated, CAR␤ was found to be constitutively active on candidate DR-5 response elements. This could be the result of the production of an endogenous intracellular ligand or inherent activity in an unliganded state. Be- cause CAR␤ is also active in yeast, production of an en- dogenous ligand seemed unlikely. Thus, although lack- ing precedent, a ligand screen was established that would register an increase or a decrease in CAR␤ activity. This resulted in the identification of androstanol and andro- stenol as surprising but selective transcriptional inhibi- tors (Forman et al. 1998). These androgens completely lack activity on the androgen receptor whereas classic androgens such as testosterone fail to act on CAR␤.How the ligand reverses activation by inducing coactivator release, rather than recruitment, remains a structural co- nundrum and a solution will likely require a combina- tion of mutational studies and crystal structure in the presence and absence of ligand. Although CAR␤-specific androstanes are related to mammalian pheromones, the CAR␤ expression pattern makes it unlikely to partici- Figure 1. Schematic illustration of RXR heterodimerizing with pate in this process. Rather, this work suggests a com- an orphan receptor. RXR typically occupies the 5Ј half-site in pletely new area of androgen physiology that will require the response element. Representative natural ligands are shown knockouts and synthetic ligands to tease out its rel- next to the corresponding receptor. Synthetic activators are evance. shown where endogenous ligands are not known. BXR—an embryonic benzoate receptor noninsulin-dependent diabetes mellitus (NIDDM) the The specification of positional information and the addition of RXR-selective ligands can potentiate the ef- transfer of patterning signals are key events during early fects of known antidiabetic PPAR␥ activators and sensi- embryogenesis. One class of information transfer is me- tize animals to insulin (Mukherjee et al. 1997). This diated by diffusible substances called morphogens, some RXR-targeted therapy is now in clinical trials. of which, like retinoic acid, are small lipophilic mol- ecules. We hypothesized that additional related develop- mental signaling systems might exist and thus searched for novel embryonic orphan receptors, with the hope Four new orphan receptor signaling pathways that one or more of these might be ligand dependent Sequence analysis shows that the four newly described (Blumberg et al. 1992). One of these receptors, BXR (pre- ligand-dependent receptors (BXR, CAR␤, PXR, and SXR) viously called ONR-1; Smith et al. 1994a), was identified are closely related to each other and they appear to com- as an obligate RXR heterodimeric partner, both in vitro prise a distinct branch of the family tree. Therefore, one and in vivo, establishing it as a candidate for a ligand might expect them to share some common properties. screen (Blumberg et al. 1998a). BXR mRNA is expressed Each requires heterodimerization with RXR for high-af- during early development suggesting embryos as a likely finity DNA binding. The preferred organization of the ligand source. HPLC fractionation of embryonic ex- 3150 GENES & DEVELOPMENT Downloaded from genesdev.cshlp.org on September 25, 2021 - Published by Cold Spring Harbor Laboratory Press New orphan receptor ligands tracts, guided by the ability of fractions to stimulate BXR A paradoxical feature of catatoxic compounds is that, target genes, revealed several activity peaks. Scale-up pu- whereas many of the most potent compounds (e.g., PCN, rification followed by mass spectrometry and 1H-NMR spironolactone, cyproterone acetate) are steroid