Orphan nuclear receptor 4A1 (NR4A1) and novel ligands

NR4A1 (Nur77, TR3), NR4A2 (Nurr1) and NR4A3 (Nor1) are early immediate genes that are induced by diverse physiological and physical stimuli and play a role in adaptation to cellular stress and in pathophysiology ( 20 – 25 ). These genes exhibit similarities in their C-terminal (58–65%) (AF2) and DNA binding (94–95%) domains whereas sequence conservation in their N-terminal (AF1) domains was 26–28% and AF-1 dependent difference in transactivation by NR4As has been reported ( 26 – 28 ). There are also differences in NR4A null mutant mice; NR4A1 −/− mice are viable, NR4A2 mice die soon after birth due to deficits in the dopaminergic system. NR4A3 −/− mice exhibit inner ear deficits and one study also reported embryo lethality in NR4A3 −/− mice due to a failure to complete gastrulation ( 29 – 34 ). Endogenous ligands for NR4A1, NR4A2 and NR4A3 have not been identified and X-ray crystallographic analysis of the LBD of NR4A2 indicates that bulky amino acid side chains may preclude their interactions with ligands ( 35 ). Ongoing studies demonstrate a role for NR4A1 in multiple diseases ( 18 – 20 , 36 – 38 ) that could potentially be targeted by ligands and this is now paralleled by studies on the development and potential applications of structurally diverse NR4A1 ligands ( 18 – 20 ).

Humans and mice encode 48 and 49 nuclear receptor genes respectively which are classified based on their common structural domains which include N-terminal A/B and C-terminal E/F domains, a DNA binding domain (C) and a hinge region (D) ( 1 , 2 ). These receptors are broadly classified as ligand-activated transcription factors although there is evidence that some NRs exhibit extranuclear functions and ligand-independent activities. The NR superfamily and their receptor variants play a critical role in maintaining cellular homeostasis associated with development, reproduction and metabolism, and they also are important in multiple disease processes. The glucocorticoid receptor (GR) and estrogen receptor α (ERα) were the first NRs identified ( 3 , 4 ) and are part of the endocrine receptor subfamily which also include the progesterone, androgen and mineralocorticoid receptors. Other sub-classes of the NR family include the heterodimeric receptors, adopted orphan receptors (lipid sensors and enigmatic orphans) and the orphan receptors ( 1 ). Endogenous ligands have been identified for all but the orphan receptors. The activities of NRs are dependent on cell context where tissue/cell-specific expression (or lack thereof) of nuclear cofactors including coregulators and corepressors and chromatin structure are essential elements for NR function ( 5 – 8 ). For most NRs their endogenous ligands such as 17β-estradiol (E2) for ERα are also critical drivers of NR functions ( 9 ). Initial studies on ERα defined the high affinity interactions of E2 with ERα and for other high affinity NR ligands and their cognate receptors; however, it has been demonstrated that ERα and other NRs bind structurally diverse ligands usually with lower affinity and many of these compounds have important endogenous functions or are associated with adverse side-effects ( 10 – 12 ). For example, diethylstilbestrol binds with high affinity to ERα and use of this pharmacologic agent by pregnant women resulting in serious health problems in their male and female offspring ( 13 , 14 ). Tamoxifen is a synthetic drug that exhibits modest binding affinity for ERα and has been invaluable as an antiestrogen for treating early stage ER-positive breast cancer in women ( 15 ). In contrast, other synthetic ligands such as bisphenol-A that bind ERα with modest-low affinity are classified as endocrine disruptors and it has been hypothesized that they may be linked to multiple adverse health effects ( 16 , 17 ). Thus, the structure of receptor ligands can dramatically modulate the functions of ligand-mediated responses and this review will focus on NR4A1, a member of the NR4A sub-family of orphan nuclear receptors and the effects of ligands on receptor-mediated activity ( 18 – 20 ).

The important immune cell functions of NR4A1 and its expression in multiple inflammatory cell types make this receptor an attractive drug target for treating multiple diseases. This review outlines several structural classes of NR4A1 ligands that differentially bind and modulate NR4A1-dependent responses and genes. For example, Csn-B and structurally related TMPA both bind NR4A1, however, in in vivo models of metabolic disease Csn-B increased and TMPA decreased blood glucose levels. There are conflicting and unresolved issues in breast cancer where there is evidence that NR4A1 is both a tumor promoter and inhibitor ( 46 – 48 , 72 ). However, both Csn-B and CDIM/NR4A1 ligands inhibit mammary tumorgenesis despite their different modes of action. NR4A1 exhibits pro-oncogenic functions in solid tumors, however, NR4A1 is a tumor suppressor in most blood derived cancers and proposed receptor-derived therapies for leukemia include agents that induce NR4A1 expression (rev. in ( 20 , 90 )). Thus, development of pharmacologic agents targeting NR4A1 may require some disease-specificity. However, studies with CDIMs suggest that these compounds may be selective NR4A1 modulators since NR4A1-active CDIMs inhibited growth of solid tumor derived cancers, enhanced glucose uptake in muscle cells, decreased neuronal inflammation in models of Parkinson’s disease, inhibited endometriosis and enhanced learning and memory in mouse models ( 20 , 91 – 96 ).

Celastrol is a naturally-occuring triterpenoid ( Fig. 1A ) with anticancer activity; this compound bound NR4A1 with a K D value of 0.29 μM ( 57 , 58 ) and celastrol inhibited NR4A1-dependent transactivation. Molecular modeling studies showed that celastrol interacted with a region on the surface near the LBD of NR4A1 as previously described for TMPA ( 55 ) and like TMPA, celastrol inhibited high fat diet-induced chronic inflammation and weight gain. Celastrol induces nuclear export of NR4A1 where it interacts with mitochondrial tumor necrosis factor receptor-associated factor 2 (TRAF2) and this triggers anti-inflammatory responses ( 57 ) ( Fig. 1C ). Modeling of celastrol-NR4A1 binding showed interactions with GIn547 and Asp499, and the K D value was 0.32 μM. Structure activity studies ( 58 ) identified the two hydroxyl groups on the A ring as important binding determinants and addition of substituents (other than H) at C-6 in the B ring also resulted in the loss of activity as an NR4A1 ligand.

In solid tumors NR4A1 is overexpressed and for breast, colon, lung and ovarian tumors NR4A1 is a negative prognostic factor for patient survival or recurrence (rev. in 20 ). Several studies show that in most solid tumor-derived cell lines knockdown of NR4A1 results in decreased growth, survival, migration and invasion, and associated genes ( Fig. 2 ) demonstrating that NR4A1 is a pro-oncogenic factor. Moreover, in alveolar rhabdomyosarcoma (ARMS), a devastating pediatric cancer, NR4A1 not only regulates multiple pro-oncogenic pathways/genes as illustrated in Figure 2 but also the PAX3-FOX01 fusion gene that is the major transcriptional driver of this tumor ( 59 ). Bis-indole derived (CDIM) compounds were initially identified as PPARγ ligands ( 60 ). Structure-induction study of a series of ten 4-substituted phenyl CDIM analogs showed that the 4-trifluoromethyl, 4-methoxy and 4-H analogs were most active as NR4A1 agonists whereas the 4-trifluoromethyl, 4-bromo- and 4-tbutyl analogs were most active NR4A2 agonists in pancreatic cancer cells ( 61 , 62 ). However, subsequent structure-transactivation screening using GAL4-receptor chimeras identified a group of C-DIMs that not only bound NR4A1 but inhibited NR4A1-dependent transactivation in multiple cancer cell lines ( 20 ). Moreover, NR4A1-active CDIMs inhibited most of the pro-oncogenic pathways and associated genes illustrated in Figure 2 in colon, pancreatic, lung, breast, rhabdomyosarcoma, kidney and endometrial cancer cell lines ( 63 – 70 ). Thus, knockdown of NR4A1 or inhibition of NR4A1 by CDIMs gave comparable effects in cancer cell lines and the CDIMs were classified as NR4A1 antagonists even though array or RNAseq analysis show that both NR4A1 silencing or antagonism results in both induction and inhibition of gene expression ( 64 ). The designation of CDIMs as NR4A1 antagonists is based on their antagonism of the functional pro-oncogenic properties of NR4A1. Initial studies on NR4A1-active CDIMs focused on the 1,1-bis(3΄-indolyl)-1-(p-hydroxyphenyl)methane (DIM-C-pPhOH or CDIM8) and the corresponding p-carboxymethyl analog (DIM-C-pPhCO 2 Me) ( 70 ) ( Fig. 3 ). CDIM8 inhibits cancer cell and tumor growth (20–40 mg/kg/d) in athymic nude mouse xenograft models, however, pharmacokinetic studies show that this compound is rapidly metabolized and blood levels are low ( 71 ). Recent studies show that addition of substituents ortho to the 4-hydroxyl at C-3 and C-5 in the phenyl ring results in a buttressing effect which decreases metabolic conjugation of the hydroxyl group and enhances potency of tumor growth inhibition in athymic nude mouse xenograft (breast cancer cells) models by approximately 10-fold ( 72 ). Direct binding of CDIM8 to NR4A1 has been reported using a fluorescence quenching assay and by circular dichroism and modeling studies have identified Asp594, His556, Arg515 and Glu445 as key amino acids interacting with CDIM8 ( 70 ). Oxidized analogs of several 4-substitutedphenyl CDIMs have also been reported ( 73 , 74 ) ( Fig. 3 ) and these compounds are also potent inhibitors of cancer cell growth and appear to be more active than their parent precursors in several cancer cell lines and in vivo (prostate cancer). The oxidized mesylate derivative of DIM-C-pPhCF 3 (4-trifluoromethylphenyl) bound NR4A1 and key interactions with side chains of His372 and Tyr453 located in helices 1 and 5 respectively were key binding determinants. The mechanisms associated with CDIM modulation of gene expression involve direct interactions with nuclear NR4A1 which bind cognate NBRE and NuRE sequences as a monomers and dimers respectively as previously described ( 75 – 77 ) ( Figs. 4A and 4B ). Another mechanism of NR4A1-dependent transactivation involves NR4A1-RXR complexes ( Fig. 4C ) that interact with a DRE motif ( 78 ) and recent studies have identified that NR4A2 interactions with novel promoter sequences that may also bind NR4A1 ( 79 ). Genomic analysis of NR4A1 regulated genes shows that many of these genes such as survivin and epidermal growth factor receptor (EGFR) have previously been characterized as specificity protein 1 (Sp1)-regulated genes. NR4A1 knockdown or treatment with CDIM/NR4A1 antagonists decreased expression of survivin in pancreatic cancer cells and subsequent analysis showed that levels of survivin are dependent not only on Sp1 but also NR4A1 and p300 ( 63 ). Moreover, individual knockdown of Sp1, NR4A1 and p300 decreased survivin expression and indicating that NR4A1 is acting as a nuclear cofactor and there is extensive evidence showing that many nuclear receptors are cofactors for Sp1-regulated genes ( 80 ). Subsequent studies have shown that NR4A1/Sp1 and/or NR4A1/Sp4 regulate multiple genes through their NR4A1/Sp interactions with GC-rich promoters and these include PAX3-FOX01, PD-L1, G9a and several integrins ( 63 – 65 , 81 , 82 ) ( Fig. 4D ). TGFβ plays a key role in invasion of breast and lung cancer cells and this involves phosphorylation and subsequent export of nuclear NR4A1 which interacts with a proteasome complex that degrades inhibitory SMAD7 ( 83 , 84 ). CDIM/NR4A1 interacts with NR4A1 to inhibit nuclear export and this is accompanied by decreased degradation of SMAD-7 and TGFβ-induced invasion ( 47 , 48 ) ( Fig. 4E ). CDIM NR4A1 antagonists enhance apoptosis in RD embryonal rhabdomyosarcoma (ERMS) cells by interacting with constitutive cytosolic NR4A1 which results in apoptosis ( 83 ) ( Fig. 4F ). The oxidized CDIM + compounds induce nuclear export of NR4A1 which forms a pro-apoptotic NR4A1-bcl2 complex as described for celastrol ( Fig. 1C ) and this pathway is also activated by other pro-apoptotic agents that do not directly bind NR4A1 (rev. in ( 20 , 84 )). A survey of several lipids as NR4A1 ligands identified unsaturated fatty acids arachidonic and docosahexaenoic acids as compounds that bound NR4A1 whereas this was not observed for saturated fatty acids ( 85 ). Binding and receptor stabilization studies suggested that the unsaturated fatty acids might play a role in stabilizing NR4A1 oligomer complex formation. Prostaglandin A 2 , another lipophilic endogenous compound also binds NR4A1 and forms a covalent adduct at Cys566 of NR4A1 via Michael addition associated with the cyclopentenone ring ( 86 ). PGA 2 also induces NR4A1-dependent transactivation in normal human bronchial epithelial cells. Interestingly, PGA 2 also binds NR4A3 ( 87 ) and is thus a dual receptor ligand. Computation based modeling studies identified 2-imino-6-methoxy-2H-chromene-3-carbothioamide (IMCA) as an NR4A1 ligand and functional studies indicated that this compound acts via induction of nuclear export of NR4A1 in medullary thyroid cancer cells ( 88 ). A recent study on a series of putative NR4A ligands ( 89 ) showed that cytosporone B also bound NR4A2; many of the compounds exhibited NR4A2-independent activities and this may also be true for some of the NR4A1 ligands described in this review.

Wu and coworkers pioneered the identification and subsequent applications of NR4A1 ligands by their screening of a natural products library which identified cytosporone B (Csn-B) ( 39 ) an octaketide fungal metabolite as an NR4A1 ligand ( Fig. 1A ). Csn-B but not structurally related Csn-C induced NR4A1-dependent transactivation in human gastric BGC-823 cells and directly bound the ligand binding domain (LBD) of NR4A1 with a K D value of 7.4 × 10 −7 M. Moreover, Csn-B quenched the fluorescence of Tyr453 which is conserved in the ligand binding pocket of many nuclear receptors ( 40 – 43 ) whereas fluorescence was not quenched in the binding of Csn-B to the mutant NR4A1 (LBD)-Y453A. The results of Csn-B induced transactivation assays in BGC-823 cells implies a nuclear function for NR4A1 however, Csn-B also induces nuclear export of approximately 70% of cellular NR4A1 where it forms a pro-apoptotic bcl2-NR4A1 complex and induced cell death in cancer cell lines and inhibited tumor growth in vivo. At doses of 50 mg/kg Csn-B acts as an NR4A1 agonist and also enhances blood glucose levels and hepatic expression of gluconeogenic genes and NR4A1 in mouse models ( 39 ). Subsequent structure-activity studies identified synthetic analogs of Csn-B that were also NR4A1 ligands and these compounds enhanced expression of NR4A1, activated nuclear NR4A1 and also nuclear export of the receptor ( 44 ). Another paper also developed a synthesis of Csn-B and showed the importance of the 3-hydroxyl group for maintaining NR4A1 binding activity ( 45 ). Wu and coworkers recently reported the crystal structure of the NR4A1 (LBD) – Csn-B complex and showed that the ligand bridges the LBDs of an NR4A1 homodimer. Essential amino acids for this novel interaction include residues Asp481, GIn571 and Arg572 ( 46 ). These results were obtained in a study on the role of the NR4A1 homodimer as an inhibitor of breast cancer progression through suppression of genes involved in fatty acid uptake into cancer cells. This paper also demonstrated that loss of NR4A1 from genetic and carcinogen-induced mouse models of breast cancer resulted in enhanced tumorigenesis ( 46 ) and this was in contrast to other studies showing a pro-oncogenic role for NR4A1 in some breast cancer cell lines ( 47 , 48 ). Csn-B has been extensively used to investigate the role of NR4A1 in the presence or absence of ligand in multiple inflammatory diseases in mouse models ( 49 – 53 ). n-Pentyl 2-[3,5-dihydroxy-2-(1-nonanoyl)phenyl acetate (PDNPA) is another NR4A1 ligand that has structural similarities to Csn-B and PDNPA competes with p38 for binding to the LBD of NR4A1 ( 54 ). Modeling and binding studies show that amino acids Leu437, Ser441 and Asp549 are important for PDNPA-NR4A1 (LBD) binding and by inhibiting NR4A1-p38 interaction PDNPA (50 mg/kg) decreased LPS induced inflammation. However, the effects of PDNPA are specific for inhibiting NR4A1-p38 binding and it does not interact with the canonical binding pocket of NR4A1 ( Fig. 1B ). Interestingly, PDNPA also binds NR4A2 and NR4A3 but these interactions were not accompanied by phenotypic effects. Ethyl 2-[2,3,4-trimethoxy-6-(1-octanoyl)phenyl acetate (TMPA) is another Csn-B – like compound which inhibits NR4A1-LKB1 interactions resulting in nuclear export of LKB1 and activation (phosphorylation) of AMPK ( 55 ). This process results in the inhibition of gluconeogenesis and TMPA decreased blood glucose levels in mouse models of insulin resistant and these effects were in direct contrast to Csn-B – dependent increase in blood glucose levels ( 39 ). X-ray crystallographic analysis of TMPA-NR4A1 (LBD) interactions shows interactions of the ligand with side chains of several amino acids (Arg515, Glu445, Thr595, His372, Arg450, Tyr453, Leu492 and Val498) and the ligand primarily binds close to the surface of the LBD and not deep within the binding pocket. Mutational analysis confirmed that Thr595 is necessary for NR4A1 – LKB1 interactions whereas Cys566 is required for TMPA binding to NR4A1. Differences in the binding of PDNPA and TMPA to NR4A1 correlated with their functional differences; PDNPA but not TMPA rescued mice from a lethal LPS challenge whereas TMPA but not PDNPA decreased blood glucose levels in diabetic mice ( 54 ). Presumably this is due to antagonist (TMPA) and agonist (PDNPA) activities. Wu and coworkers also identified another Csn-B analog, 1-(3,4,5-trihydroxyphenyl)nonan-1-one (THPN) ( 56 ) which bound NR4A1 and induced autophagic cell death via activation of mitochondria in some melanoma cell lines. THPN did not induce nuclear export of NR4A1 but induce targeting of cytosolic NR4A1 (found in melanoma cells) to mitochondria. THPN bind surface residues Arg563 and Ser553 around the binding cavity and these sites were necessary for interaction of the THPN-bound NR4A1 with the mitochondrial Nix protein which plays a role in activation of autophagy. Thus, although THPN did not directly bind Nix, interaction of this ligand with NR4A1 was required for NR4A1-Nix binding and subsequent activation of autophagic cell death through specific mitochondrial interactions. Thus, Csn-B and related analogs exhibit structure-dependent interactions with different amino acids in the ligand binding AF2 domain of NR4A1 and it is possible that this variability may contribute to their diverse agonist and antagonist activities. These differences may also be related to other NR4A1 interactants (e.g.: NR4A1-Nix) where ligand-induced responses are due to both receptor binding and parallel interactions with other factors.