Natural products as modulators of retinoic acid receptor-related orphan receptors (RORs)

: 1994 to 2020 Retinoic acid receptor-related orphan receptors (RORs) belong to a subfamily of the nuclear receptor superfamily and possess prominent roles in circadian rhythm, metabolism, in ﬂ ammation, and cancer. They have been subject of research for over two decades and represent attractive but challenging drug targets. Natural products were among the ﬁ rst identi ﬁ ed ligands of RORs and continue to be of interest to this day. This review focuses on ligands and indirect modulators of RORs from natural sources and explores their roles in a therapeutic context.


Introduction
The retinoic acid receptor-related orphan receptors (ROR) a, b, and g are a subfamily of nuclear receptors, encoded by the RORA-C (or NR1F1-3) genes.In general, nuclear receptors are ligand-dependent transcription factors that translate endocrine and dietary signals into differential gene expression patterns.][3] All members of the nuclear receptor superfamily feature a signicant sequence homology and conserved structure.The ligand-independent activating function 1 (AF1) is located at the N-terminus, followed by a DNA-binding domain (DBD), hinge region, and ligand-binding domain (LBD).The most conserved region is the DBD, which contains two zinc-nger motifs that mediate binding to response elements located in the promoter region of target genes and are involved in receptor dimerization.ROR response elements (ROREs) consist of the AGGTCA consensus sequence proceeded by an A/T-rich region.Nuclear receptors can either bind DNA as monomers like RORs, homodimers, or heterodimers with a member of the retinoid X receptor subfamily as partner.The LBD consists of twelve ahelices that create a hydrophobic cavity to which ligands can bind.The AF2 domain, also referred to as helix 12 and included in the LBD, provides the structural surface for the interaction with co-activator and co-repressor proteins. 4,5n their unliganded basal state, nuclear receptors are bound to co-repressor proteins and act as transcriptional repressors.Binding of agonistic ligands leads to conformational changes, primarily stabilizing helix 12, which entails the displacement of co-repressor and recruitment of co-activator proteins, ultimately leading to the modulation of target gene expression.Interestingly, inverse ROR agonists like digoxin have been reported to destabilize helix 12, resulting in loss of co-activator interaction, however, without increased recruitment of corepressors (for more details on digoxin's mechanism of action, see chapter 3.4.1). 6Co-activator proteins are bound to the LBD via their LXXLL interaction domain by a charge clamp (reviewed in ref. 4 and 7).X-ray studies revealed that, in addition, stabilization of the active conformation of RORs is established by formation of a hydrogen bond between a histidine residue in helix 11 and a tyrosine residue in helix 12 (His-Tyr lock).Both, histidine and tyrosine further form p-p interactions with a phenylalanine residue on helix 12, which overall stabilizes the active conformation.][10] Inverse agonistic ligands are characterized by their ability to repress the transcriptional activity of its nuclear receptor below basal level via recruitment of additional co-repressors and have been shown to disrupt the active conformation of helix 12. Co-repressors contain a different interaction motif than co-activators with a similar amphipathic core (4XX44; 4 is a hydrophobic amino acid) but additional anking sequences and increased length (reviewed in ref .11).Other mechanisms include NCoR or SMRT tethering via other transcription factors 12 (reviewed in ref. 13) or "water trapping", which was proposed for two synthetic compounds by Kallen et al. in 2017. 9 This mechanism involves a water molecule becoming "trapped" in a partially hydrophobic environment when the inverse agonists are bound to the RORgt-LBD.Subsequent release of the water molecule into bulk solvent leads to destabilization of helix 12. 9 However, many more mechanisms are involved in transcriptional regulation by nuclear receptors (reviewed in ref. 14).
RORs have been reported to inuence various physiological processes such as circadian rhythm, neuronal cell development, and immune cell differentiation.At the same time, they are implicated in several pathologies like autoimmune, inammatory, and metabolic diseases.RORa is expressed in many peripheral tissues like the liver, skeletal muscle, skin, lung, and adipose tissue.RORb expression is restricted to brain, retina, bone, and pineal gland.The RORC gene encodes two isoforms via the use of alternative promoters.RORg (RORg1) differs from RORgt (RORg2) only at the rst 100 nucleotides at the Nterminus. 15RORg1 is expressed in muscle tissue, prostate, pancreas, heart, liver, and testicles, whereas RORgt is exclusively expressed in lymphatic tissues. 5,14,161][22][23] Even though natural products have been an important source for medicinal preparations, the focus on them in the pharmaceutical industry has diminished over the last decades.With high-throughput screening (HTS) and combinatorial chemistry on the rise, natural products were believed not to t the requirements of these systems. 20,21While many chemical probes have been discovered by screening, it is not a magic bullet. 21,22Natural products offer a wide range of pharmacophores and a high number of stereocenters, which provides libraries containing such compounds a higher hit rate (reviewed in ref. 22 and 23).As a guide to obtain sufficient oral bioavailability, Lipinski's rule of ve is used.As many natural products are substrates for active cellular transporters, they oen do not have to t these rules.This is a big advantage, as it makes such compounds more likely to succeed. 23This is apparent by the fact that between 1981 and 2014, approximately 50% of newly approved drugs are inspired by natural products, be it natural products, natural product analogues, or synthetic mimetics. 23Biological roles of RORs and their potential as drug targets

Biological roles of RORs
As the biological roles of RORs have already been explored in depth earlier (examples: ref. 16, 17 and 24-26), preference will be given to the aspects necessary to understand the studies covered in the present review.
The rst encounter with the at that time still unknown RORs was made, when a naturally occurring mutant strain of mice was discovered in the 1960s. 27These mice were called "staggerer" because of their staggering gait.Severe cerebellar underdevelopment with a lack of up to 90% of the Purkinje cells compared to wild type (WT) 28 and a shorter life span were noticed, among others. 27Over 30 years later, it was discovered that staggerer mice possess a deletion in the Rora gene (thus also referred to as Rora sg/sg mice) that leads to the elimination of the LBD, leaving the nuclear receptor inactive as interactions with co-activators are not possible anymore. 29The phenotype of Rora À/À mice is very similar compared to that of staggerers. 30nterestingly, Rora sg/sg mice experience a variety of metabolic benets (see chapter 2.2) but also deciencies like an impaired immune system, increased inammation, osteopenia, muscle atrophy, atherosclerosis (all reviewed in ref. 31), and irregularities in circadian rhythm. 32Since RORb is found mainly in the CNS, especially in areas connected to processing of sensory information (retina) or involved in circadian rhythm (suprachiasmatic nucleus), 33 it is not surprising that Rorb À/À mice show several related issues including retinal degeneration and blindness in adulthood as well as abnormalities in circadian rhythm and male sexual behavior. 34Lastly, RORg was proven to be key for the development of lymphatic tissues and reduces apoptosis of CD4 + CD8 + cells by up-regulating the anti-apoptotic gene Bcl-XL. 35,36nterestingly, circadian rhythm and RORs have been shown to be directly connected.RORs upregulate the expression of brain and muscle ARNT-like 1 (Bmal1), 32,37 a subunit of a key transcription factor in circadian regulation called CLOCK-BMAL1 (CLOCK ¼ circadian locomotor output cycles kaput) (reviewed in ref. 38).Of note, neuronal PAS domain protein 2 (NPAS2), a paralog of CLOCK, can substitute for it 39 and was also shown to be under the control of RORs. 40Briey, CLOCK-BMAL1 controls the expression of cryptochrome (Cry) and period (Per) as well as various other clock-regulated genes.Aer being expressed in a sufficient amount, the CRY-PER dimer can inhibit CLOCK-BMAL1 and is later subjected to ubiquitination, aer which the cycle can start anew. 38On the other hand, RORs themselves experience a rhythmic expression. 41For instance, RORg was shown to be under the control of CLOCK-BMAL1 in certain tissues (e.g. the liver) while its isoform RORgt exhibits constitutive expression. 42It was proposed that RORs act as "intermediaries" between the circadian clock and the cyclic expression of certain genes, affecting the extent of gene expression rather than rhythmicity itself. 43,44This was proven for metabolic genes like insulin induced gene 2a (Insig2a), elongation of very long chain fatty acids 3 (Elovl3), Cyp8b1, glucose-6-phosphatase (G6pc) and phosphoenolpyruvate carboxykinase (Pepck), among others. 43,44Furthermore, the expression of numerous phase I and II enzymes was shown to be controlled by RORa and RORg and thus a connection to bile acid synthesis, drug and fatty acid metabolism as well as glutathione conjugation was established, to name a few. 45everal types of cancer are linked to an increase or decrease in the activity of all three RORs (reviewed in ref. 46) as well.In short, RORa showed tumor suppressive activities that were amongst others mediated by p53 (ref.47-50) and the value of RORgt as a target in tumor therapy is currently under investigation (see chapter 2.2).
RORg and especially its isoform RORgt are the most researched RORs due to their connection to various inammatory and autoimmune diseases such as multiple sclerosis, 51 rheumatoid arthritis (RA), 52 systemic lupus erythematosus (SLE), 53 psoriasis, 54,55 asthma, 56,57 and, again, cancer 46,58,59 (also reviewed in ref. 60).This is due to the role of RORgt as a critical regulator of Th17 cell differentiation. 61The underlying mechanisms of Th17 cell differentiation are complex (Fig. 1).The expression of RORgt requires IL-6 and TGF-b. 61Upon activation, both cytokines are secreted by dendritic cells, which promotes differentiation of CD4 + cells into Th17 cells.Then, these cells up-regulate the IL-23 receptor and increase the expression of key cytokines like IL-17A/F. 61Subsequently, IL-17 is able to promote IL-6 production in various cell types (reviewed in ref. 62).While IL-23 is not necessary for Th17 cell differentiation, it is required to maintain their differentiated state. 63Via signal transducer and activator of transcription 3 (STAT3), IL-6 also increases the expression of Il21, which henceforth acts in an autocrine manner, promoting Th17 differentiation. 64Of note, both cytokines can increase RORgt protein levels STAT3dependently (reviewed in ref. 65).However, not only RORgt, but also RORa expression is necessary for Th17 differentiation through the aforementioned cytokines and STAT3. 66It is known that by upregulating forkhead box P3 (Foxp3), TGF-b promotes an immunosuppressive response via Treg differentiation, 67 while FOXP3 also inhibits RORgt function and thus Th17 differentiation. 68However, TGF-b is also required for Th17 differentiation via up-regulation of Il23r. 69It was elucidated that

Consequences of ROR (inverse) agonism and RORs as drug targets
Inverse agonists of RORa will face difficulties to succeed as drug targets, mainly due to the aforementioned tumor suppressive capabilities of this nuclear receptor (see chapter 2.1).From a metabolic perspective, however, there are some interesting implications of RORa inverse agonism that were rst discovered in Rora sg/sg mice, including drastically reduced triglyceride and apo-CIII levels, 73 enhanced breakdown of fatty acids, reduction in lipogenesis, prevention of weight gain 74 and elevated glucose uptake in skeletal muscle cells. 75Furthermore, the synthetic RORa inverse agonist SR3335 was shown to decrease the expression of two major gluconeogenic enzymes in mice, G6pc and Pepck, thus lowering blood glucose levels and potentially being useful in type 2 diabetes therapy. 76Of note, the inhibition of G6PC as a therapy strategy for type 2 diabetes has been proposed before, 77 although not in the context of RORa inverse agonism.Conversely, RORa agonists could play a role in the therapy of inammatory diseases (e.g., RORa promotes Ikba expression 78 ), atherosclerosis (e.g., RORa promotes Abca1/ Abca8/g1 and Apoa1 expression, thus increasing cholesterol efflux and HDL formation 74,79 ), cancer (e.g. the synthetic RORa agonist SR1078 increased p53 stability 49 ) or possibly even disorders linked to circadian rhythm.Furthermore, RORa was shown to promote Ibsp expression, with the according protein being involved in bone mineralization. 80While the therapeutic potential of RORb has not been explored much hitherto, in a more recent study, a connection between this nuclear receptor and bone loss was reported via RORb-dependent inhibition of RUNX2. 81Thus, RORa and RORb possibly could be targets in osteoporosis therapy.Moreover, in the last few years an aminothiazole compound has been identied as a dual inverse agonists of RORb and RORg 82 and derivatives thereof were reported to be neutral antagonists of RORb. 83These ndings could benet further research on RORb.Both, RORgt agonists and inverse agonists were shown to have the potential to be used as therapeutics.RORgt agonism using the synthetic compound SR0987 showed an increase in IL-17 and a decrease in programmed cell death protein 1 (PD-1) mRNA levels in vitro, which indicates a possible benecial combination in cancer therapy.More importantly, they found a decline in T cells expressing PD-1 on their surface following SR0987 treatment, although it is unclear how exactly RORgt and PD-1 are connected. 84Still, the mechanisms involved in an antitumor activity of RORgt agonism seem to be far more complex than that, with a wide range of co-stimulatory receptors (e.g.CD137) upregulated and co-inhibitory receptors (e.g.TIGIT) downregulated in T17 cells in response to a RORgt agonist. 85Interestingly, the synthetic RORgt agonist cintirorgon (¼ LYC-55716) was deemed safe for use in various types of cancer in a recent phase I clinical trial. 86Myeloid-derived suppressor cells (MDSC) are known suppressors of the immune systemespecially T cells -(reviewed in ref. 87) and can directly exert various protumor effects (reviewed in ref. 88).In a paper published in 2015 by Strauss et al., 89 a connection between MDSC-expansion and RORg was described.RORg acts by promoting positive (C/ EBPb) and suppressing negative (Socs3 and Bcl3) transcriptional regulators of myelopoiesis. 89When transplanting bone marrow of Rorc-decient mice into lethally irradiated WT mice, they saw a signicant decrease in tumor growth, metastasis and splenic MDSC in the recipients.Conversely, treating tumorbearing WT-mice with the RORg agonist SR1078 increased lung metastatic burden and splenic MDSC. 89RORg(t) inverse agonists are interesting due to their anti-inammatory potential.Most of the RORg(t) ligands currently in clinical development have psoriasis as their target indication (reviewed in ref. 19).This is probably due to the promising results gathered from compounds like A213. 90A213 was successfully used for oral treatment of psoriasis in two different mouse models of this disease. 90Although one of the most promising candidates in this eld, the RORgt inverse agonist VTP-43742, failed in phase II, the development of novel compounds is on the rise. 19In 2016, Wang et al. reported that RORg is overexpressed in tumors of patients suffering from metastatic castrationresistant prostate cancer (mCRPC) and able to increase the expression of the androgen receptor.Consequently, RORg inverse agonists (e.g.SR2211) were found to inhibit androgen receptor signaling and could therefore represent novel therapy options in mCRPC. 91Recently it was discovered that RORg is a pivotal regulator of cholesterol biosynthesis in triple-negative breast cancer cells and that its inhibition exhibits antitumor effects, for instance in patient-derived xenogras. 92From a metabolic perspective, Rorc À/À mice displayed a timedependent decrease in gluconeogenesis and an improvement in glucose tolerance and insulin sensitivity, suggesting a therapeutic potential for RORg inverse agonism in diabetes type II as well. 43Noteworthy, RORa oen exerts its effects in synergy with RORg.For instance, staggerer-Rorc À/À double knockout mice showed signicantly lowered blood glucose levels compared to WT littermates, though these effects could not be observed in either, staggerer or Rorc À/À mice alone. 45Regarding anti-inammatory capabilities, it was shown that Th17 differentiation is not completely abolished in the absence of RORgt alone, 61 but rather by a RORa-RORg-double deciency. 66mportantly, mice that lacked both nuclear receptors (Rora sg/sg / c À/À ) experienced complete protection against experimental autoimmune encephalomyelitis (EAE), 66 an animal model for multiple sclerosis. 93Both examples indicate that in some instances, inhibition of more than one ROR at once is desirable.

Natural ligands directly binding to RORs
An overview of natural ligands directly binding to RORs is depicted in Fig. 2 and Table 1.    C 50 ¼ 3.9 mM (promotion of NCOR2 co-repressor peptide binding) 155 When the crystal structure of the human RORa LBD was rst elucidated the authors found a ligand within its ligand binding pocket that turned out to be cholesterol. 8This was conrmed using mass spectrometry (MS). 8,96Mutations in the LBD impairing cholesterol binding as well as inhibition of cholesterol synthesis using a statin resulted in a decrease in transcriptional activity in a full-length RORa luciferase reporter assay.Depleting cells of cholesterol using hydroxypropyl-bcyclodextrin (HPCD) and a statin also led to a decrease in transcriptional activity, which could be reversed by the addition of e.g.cholesterol and, even more effectively, 7-dehydrocholesterol (¼ provitamin-D 3 ) at 10 mM. 8 Although helix 12 in the LBD is in an active conformation when bound to cholesterol, cholesterol does not directly interact with this helix. 4,8,97However, it is known that agonists can function without this ability, for instance by stabilizing the hydrogen bond of the His-Tyr lock 10 aer binding, which in turn stabilizes helix 12. 9,98,99 Hence, aer mutating this tyrosine residue to phenylalanine, effectively eliminating the His-Tyr lock, a decreasebut no obliterationin transcriptional activity was observed by the authors, 8 indicating that other interactions must be able to stabilize the active conformation as well.In later studies, cholesterol could increase RORg co-activator recruitment, while having no effect on co-repressor interaction and on RORa co-activator recruitment, indicating RORg-specic agonistic properties.The affinity of cholesterol was, however, much lower when compared to other cholesterol metabolites. 100][102][103] Interestingly, cholesterol sulfate was able to replace cholesterol in the ligand binding pocket of RORa and was predicted to bind the RORa LBD with even higher affinity by docking, which was proven experimentally via electrospray ionization (ESI)-MS and differential scanning calorimetry. 8,96,104The crystal structure of cholesterol sulfate in complex with the RORa LBD 104 was structurally very similar to that of cholesterol and showed an agonistic conformation. 8A luciferase reporter assay in cholesterol-depleted and statin-treated cells (as described before 8 ) showed that treatment with 10 mM cholesterol sulfate led to a higher transcriptional activity relative to cholesterol.On the other hand, a critical mutation (RORa Cys 288 / Gln) within the LBD selectively decreased the affinity of cholesterol sulfate while not affecting cholesterol binding and thus resulted in a reduced transcriptional activity relative to cholesterol. 104oreover, cholesterol sulfate was active in a RORg co-activator recruitment assay, RORg-Gal4 and RORa-Gal4 reporter assays in the presence of either the inverse agonist ursolic acid or an azole (CYP51 inhibitor). 2 However, another study could not observe an activity of cholesterol sulfate at 500 mM in RORg-Gal4 or RORa-Gal4 reporter assays where cells were cultivated in lipid depleted and statin and mevalonate supplemented medium, although cholesterol sulfate was active in a 25-[ 3 H] OHC competition assay. 103In a cell-based study, 40 mM cholesterol sulfate increased the mRNA expression levels of RORa and the RORa-regulated epidermal barrier precursor protein prolaggrin in a RORa-dependent manner in normal human epidermal keratinocytes. 105aken together, due to the low affinity for the LBD of RORs, cholesterol seems unlikely to be a physiological ligand.However, cholesterol sulfate has been shown to be present in Th17 cells and although functional assays have not been performed, data with desmosterol sulfate (see chapter 3.1.2)suggests similar properties for cholesterol sulfate. 2 3.1.2Cholesterol biosynthetic intermediates.In 2015, Santori et al. 1 were able to show for the rst time that sterols from the cholesterol biosynthetic pathway are a sufficient requirement for RORg transcriptional activity.RORg activity in an insect cell-based reporter system was dependent on sterol lipids, with a broad specicity for a wide range of sterols.In different mammalian cells lines, such as HEK293T, Hela, or HepG2 cells, cultivated in cholesterol-free medium, RORg is In the same year Hu et al. 2 found that endogenous sterol metabolites control Th17 differentiation via RORg agonism.Notably, the inhibition of the mevalonate-cholesterol synthetic pathway via statins reduced Th17 differentiation and IL-17A production. 106,107Desmosterol as well as zymosterol potently increased co-activator recruitment in the presence of the inverse agonists ursolic acid or digoxin and RORg-Gal4 transcriptional activity, thus suggesting that these compounds occupy the same binding site.They also both increased IL-17A production and Th17 differentiation in the presence of ursolic acid.Moreover, in Th17 cells desmosterol increased RORg target gene expression but not RORgt itself.The RORg-dependence of the IL-17A increase elicited by desmosterol was conrmed via knockdown of RORg with siRNA during differentiation and the use of T cells from RORg knockout mice.Quantication of selected sterols in Th17 cells revealed that only cholesterol and desmosterol were detectable.Furthermore, sulfated sterols, especially desmosterol sulfate, were basally or in the presence of ursolic acid more potent agonists of RORg than the corresponding 3-OH sterols.Higher production of sulfated sterols together with the fact that desmosterol sulfate as well as cholesterol sulfate could be quantied in Th17 cells, suggests that sterol sulfates might act as endogenous RORg agonists in Th17 cells.
Taken together, upregulation of cholesterol biosynthesis and uptake and simultaneous downregulation of cholesterol metabolism and efflux during Th17 differentiation leads to the accumulation of the cholesterol precursor desmosterol and its sulfate conjugates, which then act as endogenous RORg agonists in Th17 cells. 2 3.1.3Oxysterols.Several oxysterols have been investigated regarding their effect on different RORs.20a-Hydroxycholesterol (20a-OHC), 22(R)-hydroxycholesterol (22R-OHC), and 25-hydroxycholesterol (25-OHC) were all active with similar affinity in RORg co-activator recruitment assays with EC 50 values between 20 and 40 nM, while being inactive in RORg co-repressor recruitment and RORa co-activator recruitment assays. 100Notably, in another study 22R-OHC and 22S-OHC (10 mM) were both not able to elicit an effect in RORg-or RORa-Gal4 assays. 103Co-crystal structures with the RORg LBD, the co-activator peptide NCOA2-2 and the putative ligands 20a-OHC, 22R-OHC, and 25-OHC revealed very similar LBD structure for all of them with the C-terminal AF-2 in the active conformation, suggesting they act as RORg agonists.Although they take up the same position in the LBD, binding is dependent on unique pocket residues relevant for size and polarity, as could be shown with a mutagenesis approach.Transcriptional activation of RORg-Gal4 and full-length RORg conrmed the agonistic properties of these oxysterols and mutation of the unique binding residues in the RORg LBD abolished the activity of the respective oxysterols.Interestingly, sulfation of cholesterol and oxysterols carried out via SULT2B1 overexpression led to a decrease in transcriptional activity and re-supplementation of these oxysterols partly reversed this effect.It was further revealed that 25-OHC interacts with the two amino acid residues involved in the His-Tyr lock, indirectly with tyrosine on helix 12 and directly with histidine on helix 11. 9,10 Although 25-OHC was not active in a RORa co-activator recruitment assay, 100 it was shown to directly bind to the LBD of RORa in a MS approach 96 and slightly but non-signicantly to reduce activity in a RORa-Gal4 assay, 103 suggesting inverse agonism.Interestingly, the enzyme CH25H , which is responsible for 25-OHC production, is downregulated in bone marrow-derived macrophages from sg/ sg mice. 108Lipid storage is disturbed in these cells, which could be restored by treatment with physiological concentrations of 25-OHC, 109 pointing to 25-OHC as endogenous ligand.
Next to the already mentioned compounds, other oxysterols, like 22S-OHC and 27-OHC did not inuence RORa or RORg activity. 100,103iven the role of RORs in bile acid metabolism, 45 7ahydroxycholesterol produced by CYP7A1, the key enzyme in bile acid synthesis, 110 and other related 7-oxysterols (7-OS, 7bhydroxycholesterol and 7-ketocholesterol) were investigated on RORa/g. 103A competitive radioligand binding assay performed against tightly bound 25-[ 3 H]OHC indeed showed that all these 7-OS bind with high affinity to the RORa/g LBD.Furthermore, 7-OS decreased transcriptional activity in a RORa/g-Gal4 and fulllength RORa/g assay with the RORE-containing G6PC promoter 43,111 in HEK-293 cells, which suggests an inverse agonistic mechanism of action.When mutating this RORE, the effects of the sterols vanished. 103Furthermore, 7-OS inhibited the mRNA expression of G6PC in HepG2 cells and ChIP experiments revealed a 7-OS-dependent decrease in NCOA-2 recruitment to the G6PC promoter.A reChIP experiment using 7a-OHC conrmed the RORa dependence of the decrease in NCOA-2 recruitment to the promoter.Together with the data obtained by hydrogen deuterium exchange MS, the authors proposed a model of RORa/g being in a constitutively active (and thus co-activator-bound) state, where inverse agonists such as 7-OS can interfere with this process.Finally, in a more functional setting using RORa/g siRNA in murine hepatocytes, the metabolic effects of 7-aOS (decrease in G6pc/Pepck gene expression and glucose output) was obliterated, as expected from direct ligands like 7-OS.Additionally, the tested 7-OS showed no affinity towards LXR in a Gal4 luciferase assay.Possible effects on other nuclear receptors were not examined. 103In addition to these three 7-OS, GC-MS revealed that 7-dehydrocholesterol acts as a ligand of RORa. 96ater additional oxysterols were investigated. 101In RORg-Gal4, full-length RORg and RORgt reporter assays several oxysterols showed signicant agonistic activity in the presence of the inverse agonist ursolic acid, with the highest potency and efficacy observed for 27-OHC and 7b, 27-OHC.Notably, in this assay 7a, 25-OHC was inactive, while 7-keto, 27-OHC was only active in the Gal4 but not in the full-length transcriptional assays.Several oxysterols that were previously reported as RORg or RORa inverse agonists or agonists 101,103,112 including 25epoxycholesterol and 7-ketocholesterol were only weakly to moderately active.Cholestenoic acid derivatives of 27-hydroxylated sterols were only barely active, emphasizing the importance of the hydroxyl group at carbon 27 for RORgt agonism.Many of the tested oxysterols had activities on other nuclear receptors, however, 7b, 27-OHC and 7a, 27-OHC seemed to be the most selective RORgt agonists.Direct binding of oxysterols was investigated in thermal shi assays with RORa/b/g LBD and the co-activator peptide NCOA1, where 27-OHC oxysterols bound most potently to the RORg LBD (7b, 27-OHC > 7-keto, 27-OHC > 27-OHC > 7a, 27-OHC).Notably, 27-OHC binding to the RORg LBD was signicantly lower in the presence of NCOA1, suggesting that this oxysterol is no endogenous RORg agonist. 101In a functional approach 7b, 27-OHC and 7a, 27-OHC, but not 7a, 25-OHC increased the number of IL-17A producing cells from total or naïve mouse and human CD4 + T cells under Th17 differentiating conditions in the absence or presence of ursolic acid.RORgt dependency was conrmed with RORgt-decient mouse CD4 + T cells.In addition, it was conrmed that 7b, 27-OHC and 7a, 27-OHC do not activate RORa, as functional RORa is expressed in RORgt knockout cells.The production of 27-OHCs are dependent on the enzyme CYP27A1.Mouse Cyp27a1 knockout-naïve CD4 + cells showed signicantly reduced Th17 differentiation and exogenous addition of 7b, 27-OHC restored this effect, suggesting a physiological role of 27-OHC oxysterols in this process.The importance of CYP27A1 and 7b/a, 27-OHCs for Th17 differentiation was conrmed in vivo in mice.Cyp27a1 knockout mice had elevated 25-OHC levels, which led the authors to the conclusion that this oxysterol is unlikely to function as endogenous RORgt agonist. 101nother oxysterol studied in more detail is 24S-OHC (cerebrosterol), mostly found in the brain. 112,113It showed similar effects on RORa and RORg transcriptional activity in Gal4 assays as 7-oxysterols.Using ChIP-reChIP assays with the coactivator NCOA2, 24S-OHC was demonstrated to reduce recruitment of this peptide to RORa.Notably, also 24(S),25 epoxycholesterol, found in micromolar concentrations in the liver and brain, 114,115 and 24R-OHC act as specic partial inverse agonists for RORg with an IC 50 of 280 nM and 90 nM in a Gal4 assay and a K i value of 20 nM and 102 nM in a competition assay against 25-[ 3 H]OHC, respectively, with no activity on RORa. 112he high abundance of cerebrosterol in young children 113 and its RORa agonistic properties led the authors to propose a role for it in the developing brain. 112It is interesting to see that only small structural changes in the molecules can evidently lead to a specicity for a certain ROR protein, which can be explained by the differences in the LBD amongst RORs (reviewed in ref. 24).Noteworthy, all compounds tested in this study were previously identied to be agonists of LXR. 116n differentiated Th17 cells, oxysterols could not be detected 2 but in naïve T-cells, 27-OHCs were quantiable. 101Moreover, the level of 27-OHC has been reported to be 5 times lower than that of desmosterol in human plasma. 117,118Taken together, the studied secosteroids acted as inverse agonists on RORa and RORg, which might explain the multitude of effects elicited by vitamin D.
3.1.5Neoruscogenin.Neoruscogenin, a steroidal sapogenin, which can be found in Ruscus aculeatus (Asparagaceae), was identied as an agonist of RORa. 121A novel HTS method utilizing a variant of a pull-down assay, in which the liganddependent recruitment of a co-activator peptide to RORa LBD is quantied using luminescence, was able to identify RORa ligands within fractionated plant extracts.Using this method combined with subsequent isolation and chromatographic purication steps led to the identication of 25S-ruscogenin (from Dalbergia cambodiana, Fabaceae) as a potent RORa agonist.Due to its better availability, neoruscogenin, which was found to be even more potent in the HTS method than 25Sruscogenin, was subjected to further studies, which proved the agonistic properties of neoruscogenin in a RORa-Gal4 luciferase assay and by a signicant increase in gene expression of RORa target genes in HepG2 cells.The latter was later conrmed in vivo aer treating mice with the compound and harvesting their livers (e.g.Bmal1 and G6pc were up-regulated). 121Regarding selectivity, it must be noted that although being specic for RORa and not activating other RORs, neoruscogenin was found to increase the transcriptional activity of PXR in a Gal4 luciferase assay, whereas other nuclear receptors (such as LXR and FXR) were not affected. 1212 Terpenoids 3.2.1 Ursolic acid.Ursolic acid, a pentacyclic triterpenoid carboxylic acid common in most plant species, 122 was found to selectively inhibit RORgt.123 It was able to inhibit Th17 differentiation of naïve CD4 + T cells (murine, human) and downregulated IL-17 but not RORa/gt gene expression.Other cell types (e.g.Th1, Treg) were not affected as much, although a slight rise in IFN-g + cells upon ursolic acid treatment was detected.Measurements of (i) Il17a/f expression aer transduction of RORs into neutrally differentiated T cells, (ii) transcriptional activity in a luciferase assay when RORE reporter and RORa/gt were transfected in HEK293T cells and (iii) coactivator peptide binding to the LBDs (TR-FRET assay), revealed selective inhibition of RORgt by ursolic acid.123 In a mouse EAE model, ursolic acid delayed the onset of the disease by a few days, but aer seven days, 80% of the mice were sick in both, the control and the ursolic acid group.Clinical scores, CNS invading helper T cells (Th17 and Th1 cells) and splenic IL-17 production, on the other hand, were decreased signicantly in the treatment group.123 In another study 124 ursolic acid lessened the incidence and severity of collagen-induced arthritis in mice while decreasing the expression of the proinammatory cytokines TNF-a, IL-1b, IL-6, IL-21, and IL-17 and the oxidative stress markers iNOS and nitrotyrosine. Urslic acid moved the balance between Treg and Th17 cells in the spleen of these mice to the Treg side, consistent with a reduced expression of IL-17, IL-21, phosphorylated (p)-STAT3, and RORgt.Moreover, the inhibitory effect of ursolic acid on Th17 cell differentiation was conrmed in an in vitro model.However, in this model the amount of Treg cells was not inuenced, as shown via Foxp3 expression. 124Some of these results are in contrast to the ndings from Xu et al. as outlined before.123 In their study ursolic acid did not inuence STAT3 phosphorylation and RORgt mRNA levels in Th17 cells.The authors argued that these discrepancies might result from different Th17 differentiation cocktails or cell types used.123,124 Moreover, ursolic acid has been reported to inhibit STAT3 activation in many other model systems.[125][126][127][128][129][130][131] STAT3 has been previously shown to be required for Th17 differentiation in vivo and to act upstream of RORgt.3,61,132 Furthermore, ursolic acid is a known inhibitor of the NF-kB pathway.133 Inhibition of NF-kB leads to decreased expression of IL-6, which in turn acts as activator of STAT3 signaling.[134][135][136][137] This suggests that ursolic acid modulates the transcriptional activity of RORg also indirectly via NF-kB inhibition.More mechanistically, ursolic acid was able to displace a co-activator and co-repressor peptide in vitro.138,139 Direct interaction of ursolic acid with the LBD was suggested via an increase in melting temperature (T m ) in a thermal shi assay and via surface plasmon resonance (SPR).139 A co-crystal structure of ursolic acid and the RORg-LBD suggested a unique mode of action.Ursolic acid was shown to form a hydrogen bond with a histidine residue leading to a ip of helix 11, moving it closer towards helix 12, thereby causing a disordered C-terminus.They hypothesize that this ip leads to the displacement of the co-activator peptide and prevents the recruitment of a co-repressor.138 Lastly, RORgt selectivity (over RORa and RORb) was evaluated by an AlphaScreen® assay.139 Since ursolic acid has many other targets besides RORgt (for a comprehensive overview, see ref. 140) it must be carefully evaluated to what extent this could become an issue when using it for therapeutic purposes in humans.
3.2.2Betulinaldehyde.In a recent study, 139 betulinaldehyde and 3b,28-dihydroxy-lupan-29-oic acid were discovered as agonist and inverse agonist of RORgt, respectively.Betulinaldehyde was able to enhance co-activator and suppress corepressor binding to the RORgt LBD in an AlphaScreen® and HTRF® assay, respectively, while 3b,28-dihydroxy-lupan-29-oic acid inhibited both, co-activator and co-repressor binding.Thermal shi assays saw an increase in T m for both substances (3b,28-dihydroxy-lupan-29-oic was the more effective one, even at lower concentrations), suggesting a direct interaction between the substances and the RORgt LBD.K d values were determined by SPR and the selectivity of the substances for the RORgt LBD was demonstrated by yet another AlphaScreen® assay. 139Binding modes were proposed by employing molecular docking, suggesting that the His-Tyr lock was indirectly stabilized (similar to cholesterol sulfate 9 ) through hydrophobic interactions in the case of betulinaldehyde but was disrupted when 3b,28-dihydroxy-lupan-29-oic acid was docked, conrming the experimental results. 139.2.3Ganoderone A. Ganoderone A, a triterpenoid from Ganoderma pfeifferi (Ganodermataceae), was identied as a potent agonist of RORgt.9 X-ray analysis of RORgt LBD in complex with ganoderone A revealed stabilization of the receptor via the disruption of the His-Tyr lock in favor of two new direct hydrogen bonds established between ganoderone A and the two amino acid residues in helix 11 and 12. Probably due to these direct interactions, ganoderone A was found to be slightly more potent in a HTS-FRET assay compared to the indirect stabilizer of the active conformation cholesterol sulfate.9 3.2.4 Mthyl corosolate, uvaol and oleanolic acid.Three triterpenoids isolated from Eriobotrya japonica (Rosaceae), methyl corosolate, uvaol and oleanolic acid, were found to possess RORgt inhibitory effects in a Gal4 luciferase assay performed in Jurkat cells.141 Murine Th17 differentiation and Il17a/f gene expression were signicantly and dose-dependently inhibited.141 Subsequent in vivo studies were in lupus nephritis (LN) mice using only oleanolic acid but not the other two compounds due to poor extraction yields.Particularly uvaol would have been interesting to study since it possessed the lowest EC 50 value and effectively inhibited Th17 differentiation even in the nanomolar concentration range.Anti-dsDNA antibodies (markers for SLE, but their role as such have been questioned in the past 142 ) were reduced signicantly in the blood of LN mice treated with oleanolic acid.141 Lastly, kidney damage and renal IgG/IgM depositions were found to be reduced in mice treated with oleanolic acid.141 Interestingly, treatment with oleanolic acid had a greater effect compared to the positive control prednisolone acetate in vivo.However, different doses were used (50 mg kg À1 oleanolic acid vs. 15 mg kg À1 prednisolone acetate), making a direct comparison difficult.
3.2.5 Rockogenin.The plant sterol rockogenin, isolated from Agave gracilipes (Asparagaceae), inhibits IL-17 production in Th17 cells, decreases RORg transactivation, and displaces a co-activator peptide. 138,143Co-crystal structures with rockogenin, the RORg-LBD and the co-repressor peptide SMRT22 could determine that rockogenin interacts with the RORg-LBD via two direct hydrogen bonds and van der Waals contacts.The suggested mechanism involves the disruption of the His-Tyr lock.This then leads to the release of the co-activator and the recruitment of co-repressor peptides. 138.2.6Retinoids.The vitamin A metabolite and RAR agonist all-trans retinoic acid has been reported to specically bind the RORb LBD and completely replace the fortuitous pseudoligand stearate, which copuries with the RORb and acts as a ller molecule from the expression host E. coli without inuencing transcriptional activity, 144 in an ESI-MS assay.The formation of co-crystals with the RORb LBD, all-trans retinoic acid, and the co-activator peptide NCOA1 revealed that the binding pose of all-trans retinoic acid did not include interactions with helix 12.A scintillation proximity assay showed that [ 3 H] all-trans retinoic acid specically binds to the LBD of RORb.Moreover, in a Gal4 reporter assay in HT22 cells, all-trans retinoic acid inhibited RORb and RORg transactivation, while being inactive on RORa.Using the same assay, different cell types were tested and it was revealed that all-trans retinoic acid inhibited transcriptional activity in the neuronal cells HT22 and Neuro2A but not in NIH3T3, HEK293 or P19 cells, suggesting some kind of cell-type specicity.The authors suggest that all-trans retinoic acid might be important for crosstalk between RAR and ROR pathways. 102.2.7 Amethinol A. Amethinol A, a diterpene isolated from Isodon amethystoides (Lamiaceae), was shown to possess inhibitory effects on RORgt at 10 mg ml À1 in a Gal4-based luciferase assay performed in Jurkat cells. 145No further pharmacological information was provided by the authors.

Isoavones.
Isoavones are common plant constituents from the family of Fabaceae and the multitude of their benecial properties, for example in the prevention or treatment of cancer, metabolic syndrome and cardiovascular disease, have been reviewed recently. 146The isoavones biochanin A, genistein, formononetin, and daidzein (0.1-10 mM) have been reported to dose-dependently enhance RORa-and RORg-mediated transcriptional activity in a transactivation assay with a ROREresponsive reporter in CHO and the Il17a promoter in Jurkat cells.Furthermore, these four isoavones dose-dependently increased the interaction between the RORa-or RORg LBD and the co-activator LXXLL-peptide EBIP96 and the mRNA expression of IL-17A in mouse T lymphoma EL4 cells.Biochanin A, in particular, was shown to increase IL-17A mRNA levels RORa/gand STAT3-dependently and to enhance the interaction between RORgt and the co-activator NCOA1 as shown with immunoprecipitation and immunoblotting assays. 147,148Furthermore, biochanin A increased STAT3 phosphorylation in a Src kinasedependent manner. 148In another study, genistein treatment delayed the onset and reduced the severity of EAE.Interestingly, genistein-treated mice had a lower expression level of RORgt and reduced production of IL-6 in the spinal cord, however, IL-17 levels were not changed. 149The isoavones formononetin and isoformononetin, reduced IL-17a production and Th17 differentiation in a mouse model of osteoporosis. 150The isoavone puerarin, found in the herbal medicine Puerariae radix (Pueraria lobata, Fabaceae), decreased the amount of Th17 cells found in blood in a rat model of acute lung injury. 151hese contradicting results regarding IL-17 for isoavones might stem from the different model systems used.
3.3.2Nobiletin.Nobiletin is a natural polymethoxylated avone found in citrus peels.In a competitive radio-ligand binding assay for RORs using 25-[ 3 H]OHC nobiletin showed robust competitive binding to the LBDs of RORa and RORg, but with higher affinity to RORg.In addition, nobiletin was active in RORa-and RORg-Gal4 mammalian one-hybrid assays, indicating direct binding of nobiletin to the RORa and RORg LBD.Moreover, nobiletin dose-dependently and RORa/g-dependently increased RORa and RORg transactivation of the Bmal1 promoter in Hepa1-6 cells and the expression of ROR target genes such as Cyp7b1, IkBa, and Gck in mice livers with dietinduced obesity. 152Follow up studies showed a benecial effect of nobiletin on metabolic tness in naturally aged mice fed a regular diet or a high-fat diet via increased ROR-dependent mitochondrial respiration.In skeletal muscle of high-fat diet fed mice, expression of ROR target genes (Bmal1, Npas2) were increased upon nobiletin treatment and RORa and RORg protein levels were induced at zeitgeber time 18. 153 Based on these metabolic effects, 153 the impact of nobiletin on cholesterol homeostasis in metabolically challenged aged mice was investigated by the same group.Overall, the cholesterol prole of nobiletin treated high-fat diet fed aged mice improved and the reduced hepatic expression of ROR target genes involved in bile acid synthesis in these mice was abolished. 1544 Cardiac glycosides 3.4.1 Digoxin and derivatives.In 2011, digoxin, a cardenolide from Digitalis lanata (Plantaginaceae) was identied among nearly 5000 substances as a ligand of RORg using a RORg-Gal4 mammalian one hybrid assay in insect cells.155 Digoxin acted as an inverse agonist, decreasing the transcriptional activity of RORg (but not RORa or other nuclear receptors, e.g.hAR or LXR) when pretreated with 22-OHC at 10 mM by binding to its LBD.6,155 Binding to the LBD was proven by an in vitro competition assay using uorescein-conjugated 25-OHC as well as circular dichroism analysis.Importantly, the related cardenolides digitoxin and b-acetyldigoxin were also shown to possess RORg(t) inhibitory activity while the aglycone of digoxin, digoxigenin, was inactive.Interestingly, according to the authors, 155 digoxigenin does not even bind to the RORgt LBD, a notion challenged by later observations.156 Treatment of murine CD4 + cells with digoxin during polarization resulted in a selectively reduced expression of RORgt-controlled genes (e.g.Il17a/f, Il23r).Reduction of Th17 differentiation upon digoxin treatment was specic, since other cells lines (e.g.5 Moreover, only naïve murine CD4 + cells transduced with RORgt (and not RORa 66 ) experienced a decrease in Il17 expression upon digoxin treatment, as shown using ow cytometry.Gene expression proling (GEP) revealed that digoxin treatment and RORgt deciency mostly (>90%) impacted the same genes.This led to the suggestion that RORgt is the dominant target of this cardenolide.155 Moreover, RORgt gene expression itself was le unaltered by digoxin, indicating a direct inhibitory effect.Using ChIP analysis, digoxin was shown to inhibit RORgt-binding to key gene loci (Il17a/f, Il23r) of Th17 cells.Furthermore, co-activator binding to the LBD of RORgt was decreased while co-repressor binding was promoted using digoxin in vitro.6,155 Importantly, this mechanism of action diverges from the ones observed using certain sterol ligands, where only co-activator binding to ROR, but not binding of the nuclear receptor itself to the promoter was inhibited, indicating the existence of different modes of action of ROR inverse agonists.103,112 The selective inuence of digoxin on already differentiated Th17 cells was proven using in vitro, ex vivo and in vivo experiments, implying that digoxin treatment promoted dedifferentiation of Th17 cells (e.g.absence of Il17 expression) due to RORgt inhibition.155 Interestingly, digoxin was effective only in relatively high concentrations ($1 mM). 6 Morover, the co-crystal of the human RORgt LBD bound to digoxin was solved and revealed that digoxin occupies the same site within the ligand binding pocket as the agonist 25-OHC (shown in ref. 100).Furthermore, helix 12 is destabilized through hydrogen bonding between digitoxose and the histidine on helix 11 involved in the His-Tyr lock.9,10 Also, digoxin impedes the proper agonistic positioning of helix 12 by sticking out between helices 3 and 11 with its sugar moieties, thus hindering coactivator recruitment.6 Importantly and in accordance with previously collected data, 155 the lack of sugars would explain why digoxigenin, the aglycone of digoxin, did not act in an inverse agonistic fashion in contrast to the glycoside.
In contrast to the studies mentioned above, digoxin showed no effect on murine Th17 cell differentiation at 10 mM in a later study. 157er digoxin was established as an inhibitor of RORg(t), 6,155 its value for treating various, mostly inammatory diseases was examined in several preclinical studies.
3.4.1.1Preclinical studies performed with digoxin.In a study by Huh et al., digoxin but not its aglycone was able to ameliorate EAE in mice when compared to the vehicle control DMSO. 155he onset of the disease was pushed back by a few days and the clinical scores were consistently (and signicantly) lower in the treatment group.A signicantly decreased amount of Th17 cells were found in the spinal cords of mice in the treatment group compared to control while Th1 cells were le mostly unaffected, 155 which contrasts with other observations. 123Previously, it has been shown that transplant rejection is connected to Th17 cells and IL-17, 158,159 and direct antagonism of IL-17 (ref.160)  could suppress this process in rats.Accordingly, in a study investigating digoxin's effect on heart transplant rejection in mice, treatment with the cardenolide doubled the survival time compared to control.Moreover, inammation and necrosis of allogras were decreased and re-balancing of the Th17/Treg ratio in favor of Tregs was accomplished. 161Another study suggested a benet when treating abdominal aortic aneurysm (AAA) using digoxin in mice, 162 which is in line with the important role of IL-17 in AAA pathophysiology. 163The authors found a reduction of aortic diameter (key for the diagnosis and risk assessment of AAA 164 ), a reduced incidence of AAA and a rebalancing of the Th17/Treg ratio in favor of Tregs.The survival ratio did not change, indicating a rather prophylactic value of digoxin for this indication. 162Furthermore, digoxin was shown to possess prophylactic and therapeutic capabilities in vivo as it was able to suppress and ameliorate collagen-induced arthritis in mice. 165Inammation and cartilage loss were markedly reduced in the ankles of digoxin treated mice, as were arthritis scores and disease incidence in general.Lower expression levels of certain proinammatory cytokines (e.g.IL-6, -17 and -21) in arthritic joints were accompanied by a signicant decrease in Th17 and a rise in Treg cells in murine spleens. 165Another study examined the therapeutic potential of digoxin regarding atherosclerosis in ApoE À/À mice on a western-type diet. 166istologically, a reduction of atherosclerosis could be quanti-ed.Aer 12 weeks, a statistically signicant reduction in total cholesterol, triglyceride and LDL levels was found, while HDL levels were unchanged.mRNA levels of three RORg target genes involved in metabolism (Insig2a, Elovl3, Cyp8b1 (ref.44)) were examined and found to be decreased signicantly in the digoxin groups.In the spleens, owcytometric measurements saw a signicant decrease in Th17 cells and a signicant increase in Tregs following digoxin treatment and the same was true for Tregs in atherosclerotic plaques.Also, Th17 cell invasiveness of the plaques was decreased upon digoxin treatment. 166A further study used a murine model in order to examine the possible benets of digoxin treatment in inammatory bowel diseases (IBD). 167Weight loss due to colitis and colon colitis scores were reduced signicantly in the treatment group and fewer proin-ammatory CD3 + T cells were found in the colonic mucosa.Th17 cells (together with IL-17A and IL-23R mRNA levels) were decreased and Tregs (together with IL-10 mRNA levels) were increased signicantly in the colon due to digoxin treatment.When colitis was induced in mice using CD4 + cells of Il10 knockout mice, digoxin treatment led to a signicant decrease in weight loss, but histological scores were not signicantly different between the groups.Therefore, they suggested a benecial effect of digoxin treatment in disorders like Crohn's disease through direct inhibition of RORgt in a partly IL-10 dependent fashion. 167mportantly, these studies were all performed in mice.Cardiac glycosides are known to bind to the Na + /K + -ATPase a1 subunit of rodents with a signicantly decreased affinity.9][170] For this reason, the results of such studies cannot simply be applied to humans but give an impression of the clinical values of RORg(t) inhibition in general.
3.4.1.2Digoxin as RORg(t)-agonist?Surprisingly, another group reported agonism of digoxin on RORg(t) in a RORg transactivation assay. 171This contrasts with the results published in 2011 by Huh et al. 155 and Fujita-Sato et al. 6 who proposed the exact opposite.While Huh et al. and Fujita-Sato et al. worked in the micromolar concentration range using mostly insect or murine cells, 6,155 Karaś et al. employed 100 nM, since cytotoxic effects on HepG2 and human Th17 cells were observed at concentrations as low as 200 nM.Interestingly, when using 10 mM of digoxin (as in ref. 66 and  155) the authors observed cell viabilities of approx.40% and 5% for HepG2 and human Th17 cells, respectively.They also reported increased expression of the RORg-regulated genes G6PC and NPAS2 in HepG2 cells following digoxin treatment.In agreement with previous studies, a mammalian onehybrid assays with RORg-LBD:Gal4-DBD and RORa-LBD:Gal4-DBD constructs showed a RORg specic activity of digoxin. 6,155,171Interestingly, when overexpressing human and mouse RORg(t) in (RORE)6-tk-Luc transfected HepG2 reporter cells, digoxin was shown to have a greater effect on the murine compared to the human variants. 171However, this nding was not explored further.Moreover, the authors performed an electrophoretic mobility shi assay, which showed that digoxin treatment increased nuclear protein binding to a RORE-containing DNA probe.A ChIP assay showed increased RORgt and NCOA-1 (but decreased NCOA-2) binding in the promoter region of RORg-regulated genes (G6PC, NPAS2, IL17). 171It would be interesting to see whether the increase in co-activator occupancy is dependent on RORg via a sequential ChIP experiment, as conducted in previous studies. 103,112Application of 100 nM digoxin during polarization of CD4 + cells into Th17 cells yielded higher IL-17 mRNA and protein levels and a transcriptome analysis of digoxin-treated Th17 cells derived from human donors showed the induction of certain Th17-related genes like IL17A/F.Additional docking experiments with digoxin gave the best results when the active conformation of the RORg LBD was used and NCOA-2 was absent.Of note, docking of digoxin to the RORgt LBD domain in the inverse agonistic conformation resulted in the formation of a hydrogen bond to the histidine residue of the His-Tyr lock.Importantly, the hydrogen bond was not established via digitoxose (as seen in the co-crystal structure of Fujita-Sato et al. 6 ) but the carbonyl residue of the butenolide ring instead. 171Why these results drastically differed compared to the earlier studies 6,155 is not conclusively claried.The authors speculate that differences in the protein biosynthesis of the insect cells (see ref. 155)e.g. with respect to co-activatorscould be responsible.However, this does not explain why Huh et al. 155 found a decreased co-activator and an increased co-repressor binding to the RORgt LBD in vitro when applying digoxin, although different co-activators were used in both studies.The conversion of digoxin into its aglycone digoxigenin under experimental conditions could be another explanation for the different results, since the agonistic effect of digoxigenin was suggested in an earlier study. 156Assuming that no conversion took place, questions remain to be answered, e.g.why the X-ray analysis by Fujita-Sato et al. 6 suggested an inverse agonistic rather than an agonistic mechanism of action of digoxin.Further X-ray or NMR analyses of digoxin in complex with RORg(t) LBD could potentially provide clarication on this subject.Also, it would be interesting to repeat some key experiments of Huh et al. 155 and Fujita-Sato et al. 6 at lower digoxin concentrations to see if the results of this study can be reproduced.Lastly, it could indeed be possible that digoxin may function as both, agonist and inverse agonist at different concentrations, as the authors suggested. 171Natural products indirectly affecting RORs

Melatonin
The rst proposed endogenous ligand for RORs was the amino acid hormone melatonin (N-acetyl-5-methoxytryptamine). 95,173Melatonin is produced in the pineal gland and regulates circadian rhythm, sleep-wake cycles, and seasonal reproduction in mammals, among others.Although direct binding of melatonin to RORb has been shown initially, these results were not reproducible and the respective report has been retracted. 95However, a study from the same group showing a direct interaction of melatonin with RORa has not been withdrawn. 173In a more recent study, 94 melatonin was shown to decrease RORa transactivation in the human breast cancer cell line MCF-7 in the presence of 10% FCS to 34% of control at a concentration of 10 mM and to decrease the ability of RORa to bind to its response elements, as shown in a transfection assay and a gel mobility shi assay, respectively, without affecting RORa protein levels.It has been shown before that increased [Ca 2+ ] i levels and enhanced Ca 2+ /calmodulin (CaM)-dependent protein kinase IV activity stimulates RORa transcriptional activity. 174CaM kinase IV may inuence the phosphorylation status of ROR co-factors, thereby modulating its activity.Accordingly, a calmodulin antagonist, calmidazole, decreased RORa transactivation similarly than melatonin. 946][177] In MCF-7 cells, melatonin had no direct effect on [Ca 2+ ] i levels, suggesting that melatonin inuences RORa activity via CaM antagonism. 94In in vitro cultured goat spermatids, 0.1 mM melatonin increased RORa mRNA and protein levels.However, in CHO cells and human keratinocytes, melatonin was not able to inhibit RORa or RORg transactivation.In addition, relatively low docking scores in in silico modeling have been obtained for RORa or RORg, suggesting low affinity of melatonin for these receptors. 119In summary, the ability of melatonin to inuence ROR transcriptional activity seems to be cell-type specic and at least in human breast cancer cells a link to the calmodulin antagonism of melatonin is suggested.However, a causal relation could not be established.Moreover, it is likely that melatonin modulates ROR expression via its inuence on circadian rhythm. 178,179

Selected indirect modulators of RORs
A summary of studies on selected indirect ROR modulators is available in Table 2.The mechanism of action of most indirect ROR-modulators has not been elucidated, though hypotheses have been proposed.36][137]180 Similarly, anti-inammatory mechanisms might explain the effects of compound sophorae decoction 181 and ginger extract 182 on ROR.Other proposed mechanisms include the STAT5dependent modulation of IFN-g, or PPARg expression, ultimately decreasing RORgt levels (e.g.primisterin, 183 arctigenin, 184 astragalus polysaccharide 185 ) and the downregulation of the transcription factor HIF-1a via mTOR, leading to a decrease in RORC transactivation as suggested for rapamycin 186,187 (also reviewed in ref.

188).
It must be stated that the extent of RORs' inuence on the outcomes of these studies was not explicitly investigated in most cases.It can be assumed (and is oen pointed out by the authors themselves) that other mechanisms are involved as well.

Conclusion and outlook
Many studies discussed the question as to whether the transcriptional activity of ROR is ligand dependent.Interestingly, one study showed that apo-RORa (expressed in E.

Steroids and terpenoids
Dioscin, pristimerin, 3b-acetyloxyoleanolic acid, saikosaponin A Decrease of RORgt mRNA 189,190 and protein 180,183 levels Amelioration of inammatory diseases in rodents 180,183,189,190 Inhibition of RORgt transcriptional activity 190 Decrease of proinammatory cytokines (e.g.IL-17) 180,183,189,190 Decrease in Th17 cell differentiation 183,189,190 Polyketides Bavachalcone, poncirin/ponciretin, quercetin, baicalein Increase of RORa transcriptional activity 191 and Rora/RORA expression 191,192 Increased expression of RORa target genes (e.g.BMAL1, Fgf21) 191,192 Decreased expression of RORgt 193,194 Decrease of proinammatory cytokines (e.g.IL-17) 193,194 Decrease in Th17 cell differentiation 193,194 Protection from liver damage in rodents 194 Cardiac glycosides Uscharin, calcein, calotropin, digoxigenin, dihydroouabain, strophanthidine Decrease of RORgt transcriptional activity 157 Decrease in Th17 cell differentiation 157 Decrease of RORgt protein levels 157 Increased expression of RORg(t) target genes (e.g.G6PC, IL-17) 156 Increase of RORg transcriptional activity 156 Favorable scores when docked into RORg-LBD 156,157 Direct interaction with RORgt suggested by docking 156,157 Other substance classes Arctigenin (lignan), epigallocatechin-3-gallate, astragalus polysaccharide & astragaloside IV (saponin), oxymatrine (quinolizidine alkaloid), rapamycin (macrolide), amangostin (xanthone) Decrease of RORgt mRNA 184,186,[195][196][197][198] and protein 185,198 levels Amelioration of inammatory diseases 184,185,195,196,198 and emphysema 186 in rodents Decrease of RORgt transcriptional activity 199 Decrease of proinammatory cytokines (e.g.IL-17) 184,185,195,198,199 Decrease in Th17 (ref.Amelioration of inammatory diseases in rodents 181 Increase of RORa transcriptional activity in a Gal4 system, thus direct interaction suggested 200 Decrease of proinammatory cytokines (e.g.IL-17) 181 Decrease of RORgt mRNA levels 181,182 Decrease in Th17 cell differentiation 181 coli and therefore deemed ligand free) is active, 103 but in most reports, RORs were active only in the presence of sterol ligands (examples: ref. 1, 8, 104 and 155).To explain this discrepancy, the concept of "silent ligands" was brought up, 18 which, however, also excludes RORs in an apo-state.Additionally, expression of RORs in E. coli do not necessarily yield empty LBDs. 1442][3] Furthermore, it was shown that 500 nM of an endogenous ligand are sufficient to occupy 80% of available RORg.Calculations revealed that a further increase in occupancy would require very high concentration of ligands, which might explain the moderate effect of exogenous ligands in reporter assays containing serum. 1 When discussing natural products interacting with RORs, one of the most prominent and heavily investigated ligands is the cardiac glycoside digoxin.Despite its relatively high IC 50 value, 155 its benets were examined in numerous preclinical studies in rodents.The cardenolide was deemed both, inverse agonist 6,155 and agonist 171 of RORg(t) in different studies.Variations in experimental settings and the possibility of observing concentration-dependent effects were proposed as explanations, 171 but further validation is needed.As inhibitors of the Na + /K + -ATPase, cardiac glycosides like digoxin are used to treat heart failure, among other conditions, due to their positive inotropic effects.However, the cardenolide is used with caution and only when strictly indicated due to its narrow therapeutic window. 201Digoxin dose-dependently decreased cell viability in 10 human tumor cell lines with a mean IC 50 of 80 nM (ref.168) and it is recommended to aim for serum concentrations not exceeding 0.8 ng ml À1 (approx.1.0 nM) when treating patients. 202However, digoxin's inhibition of RORg occurs at the micromolar concentration range 6,155 and even its possible agonistic effect occurs at much higher concentrations. 171In a study on digoxin's use in atherosclerosis therapy in mice, the authors measured plasma levels aer the last injection and ascertained that they were "at or below the therapeutic range for humans", 166 at least in the low dose group.But even if digoxin could be used for its effects on RORg(t) in humans, pharmacokinetic issues arising for instance from kidney impairments 203 would then have to be taken into account to prevent poisoning.
In general, RORs are challenging targets, since their apparent therapeutic potential is accompanied by various difficulties that still need to be overcome.RORs were rst discovered in the mid-90s, the crystal structures were solved in the early 2000s and yet, to this day, there are no drugs on the market that have RORs as their target.As RORs are connected to several prominent biological systems, from circadian rhythm to metabolism and cancer, caution and a targeted approach is vital.The question to be answered is how natural products can be of use in this regard.Most of the compounds described in this study are either (i) toxic in the concentrations needed (e.g.digoxin), (ii) too ineffective (e.g.betulinaldehyde), (iii) not bioavailable enough (e.g.uvaol), and/or (iv) lack selectivity (e.g.ursolic acid).On the other hand, natural products have provided important mechanistical insights and may serve as templates for improved synthetic substances, as it was oen the case in the past.Such substances would ideally be selective for one ROR protein, with certain exceptions (see chapter 2.2), and active in lower nanomolar concentrations while having no to little off-target activity and good safety.A good example for such an approach is the chemical conversion of digoxin to 20,22-dihydrodigoxin-21,23-diol which did not exhibit cytotoxic effects on human cells even at 40 mM while still inhibiting RORg with an IC 50 of 12 mM (in vitro competition assay). 155In the last couple of years, progress has been made in this regard 204,205 (some further examples reviewed in ref.

206). 6 Conflicts of interest
There are no conicts of interest to declare.

Acknowledgements
This work was supported by the Vienna Anniversary Foundation for Higher Education (H-268438/2018).Special thanks to Andrea Szabo for drawing the gures.

Verena
Dirsch received her PhD from the University of Munich (1993) and then joined as postdoctoral fellow of the German Research Council the group of Koji Nakanishi at the Columbia University, New York.From 1995 to 2004 she held several academic positions in the group of Prof. Angelika Vollmar at the University of Munich.Since 2004, she is full professor at the University of Vienna and since 2006 head of the Department of Pharmacognosy.She also served as Vice-Dean at the Faculty of Life Sciences (2008-2014).Her main interest is to study the molecular mechanisms of natural products.Angela Ladurner studied Molecular Biology at the University of Vienna (Austria) and completed her doctoral studies on natural product research in the group of Verena Dirsch in 2013.She then worked as a postdoctoral scientist at the Medical University of Vienna and at the University of Vienna.Her research focused on the characterization of the mechanism of action of natural products.Patrik Schwarz studied Pharmacy at the University of Vienna (Austria) where he graduated in 2019.Since March 2020, he works as a doctoral student at the Department of Pharmacognosy, University of Vienna, under the supervision of Verena Dirsch.He is interested in nding novel (natural) ligands of the nuclear receptor RORg.
these phenomena were dependent on the TGF-b concentration and the presence or absence of certain cytokines during differentiation: high TGF-b concentrations led to an increase in Foxp3 expression and a decrease in Il23r expression, thus favoring Treg differentiation.On the other hand, low concentrations of TGF-b were found to enhance Il23r and inhibit Foxp3 expression in concert with IL-6 and IL-21, 64,70,71 thus favoring Th17 differentiation.68,72

Fig. 1
Fig. 1 (a) Differentiation of CD4 + cells into Th17 or Treg cells.Activated dendritic cells secrete a variety of cytokines like IL-6, IL-23 and TGF-b.These cytokines at varying concentrations are responsible for either Th17 or Treg differentiation.High concentrations of IL-6 and low concentrations of TGF-b favor Th17 differentiation.STAT3 downstream of IL-6 signaling is responsible for the expression of IL-21, which henceforth acts in an autocrine manner.Via STAT3, IL-6 and IL-21 promote RORa and RORgt expression, the latter being an important regulator of Th17 cell differentiation.Both RORs then drive IL-17A/F and IL-22 expression.Inverse agonists of RORgt like digoxin were shown to inhibit Th17 differentiation.IL-23 is necessary for the maintenance of the Th17 phenotype and TGF-b seems to promote the expression of its receptor.Tregs are created in the presence of higher TGF-b concentrations and in the absence of pro-inflammatory cytokines like IL-6.Via up-regulation of FOXP3, TGF-b inhibits RORgt function, thus favoring Treg differentiation.(b) Inverse RORg(t) agonists like digoxin can lead to a re-balance of the Th17/Treg ratio in favor of anti-inflammatory Tregs.

3. 1
Steroids 3.1.1Cholesterol and cholesterol sulfate.The rst hint in the search for endogenous ROR ligands was provided by the nding that cells stimulated with fetal calf serum (FCS) show an increased transactivation of RORs in different cell systems.94,95

Fig. 2
Fig. 2 Natural products as ligands of RORs.Various natural products were shown to act as ROR (inverse) agonists, thus affecting ROR target gene expression.

Table 2
Selected indirect modulators of RORs