Physiological roles of RORs involve several cellular processes including the regulation of development, circadian rhythm, metabolic pathway, and immunity [23]. The connection of melatonin and ROR physiological effects involve circadian gene regulation. Melatonin and RORs are considered the circadian oscillators regulating clock and clock-controlled gene transcription [24]. RORs bind to the ROR response elements (ROREs) on target genes [25] and activate the transcription of several circadian genes including Bmal1, NPAS2, and REV-ERBα [26,27,28]. The study in non-neuronal HeLa cells found that RORα and RORγ upregulate the transcription of the circadian transcription factors, Bmal1, up to 16 and fivefold, respectively [29]. The functional dysregulation of RORs has been reported in many diseases including multiple sclerosis, autoimmune diseases, metabolic diseases, cancer, and asthma [30].

The hypothesis of melatonin nuclear receptor was proposed since melatonin has been found to localize in the nucleus [31, 32]. Afterward, several studies demonstrated that RORs, a member of the nuclear receptor superfamily, were the nuclear receptor for melatonin. Recently, some emerging data opposing this idea has been proposed suggesting that melatonin may not a ligand for the putative nuclear RORs (for a review, see ref. [33]). In 1994, melatonin was firstly reported to bind directly with RORβ and stimulate its transcriptional activity [9]. However, in 1997, the authors requested the publisher for additions and corrections of the original paper that they could not reproduce the experimental part concerning the direct interaction of melatonin and RORβ [34]. Later, few studies reported that melatonin failed to activate RORβ transcriptional activity in Neuro2A cells and could not inhibit RORα/γ transcriptional activity in Chinese hamster ovary cells overexpressing with RORα/γ vectors [35, 36]. In addition, Slominski et al. (2016) reported that melatonin is not a natural ligand of RORα due to its lower affinity, respective to cholesterol and other fatty acid-derived compounds [36, 37].

On the contrary, several studies claiming that melatonin acts via its nuclear receptors have been proposed up until now. Multiple lines of evidence have shown that all isoforms of RORs are the mediators conveying melatonin effects, as proven by several models of studies ranging from RORα/β/γ-knock out models, specific inhibitors, or siRNA blockage [13, 14]. Melatonin has been suggested to inhibit hepatic stellate cell (HSC) activation through RORα, as melatonin membrane receptors (MT1/2) are not expressed in this cell line [12]. Furthermore, the RORα antagonist (SR1001) could also block the effect of melatonin on suppressing HSC activation [12]. The study carried out in human Jurkat T-cells showed co-localization of melatonin with RORα in the nucleus, emphasizing its role as a ligand for RORs [11]. Another study proposed that melatonin treatment enabled the restoration of RORα protein content in the nucleus and enhanced Sirt1 mRNA expression, resulting in an inhibition of inflammatory responses in septic mice [10]. Additionally, these effects of melatonin could not convey in the septic RORα functional knockout mice [10]. Furthermore, functional disruption of RORβ resulted in an alteration of animal behavior related to circadian rhythm dysregulation and the degeneration of the retina in postnatal development [38]. The expression of RORβ was tightly regulated by the biological clock showing its highest expression at night in the pineal gland and the retina, major RORβ expressing areas [15]. Moreover, the study using radioactive labeling melatonin showed that the binding of melatonin to its membrane and nuclear receptors in the thymus and spleen could be blocked by their specific inhibitors [39]. The discordance in the findings regarding the nuclear effects of melatonin led to a broad confusion among the scientists in the field of melatonin research. Nonetheless, solid evidence of direct interaction between RORs and melatonin has not yet been established. We, therefore, investigated whether melatonin can directly bind with the LBD of RORs, in particular RORβ, by exploiting computational and biophysical approaches.

In this study, we initially employed protein sequence analysis, molecular modeling, and molecular docking to study the potential binding property of melatonin on human RORs. The sequence analyses showed that the LBD of human RORα/β/γ isoforms shared 50–60% sequence identity, despite that the secondary structures of these isoforms are conserved (Supplementary Fig. 1). Further analysis showed that all human RORs contained an LBD and DNA binding motifs. Interestingly, the domains of unknown function and the PX domain were uniquely identified in the RORγ (Fig. 1A). Physiological roles of the PX domain involve membrane phosphoinositide binding anchoring the protein toward organelle membrane lipids [40], suggesting that RORγ may possess other unknown/diverse functions in cellular processes. Other motifs found in all RORs were C1- and C4-type zinc finger domains and hormonal ligand-binding domains (Fig. 1A, B). These motifs and domains suggested that RORs are likely nuclear receptors of some types which are capable of regulating gene expression. These findings are in accordance with previous reports on RORs as transcription regulators [25,26,27,28]. We further analyzed the possible binding sites for melatonin on the RORβ using rat and human RORs as the model. The docking results suggested that melatonin was potentially able to bind the LBD of all RORs with moderate predicted ΔGbinding in all ROR isoforms (− 7.14 to − 7.25 kcal/mol). Furthermore, all isoforms of these nuclear receptors shared a potential binding site and conserved residues for predicted melatonin interaction (Fig. 4A–C). However, the intermolecular forces of melatonin and RORs largely arose from van de Waals attractive forces, and no polar interaction was observed. Based on ΔGbinding and FullFitness scores, ARL showed more favorable binding to the LBD of RORβ. However, the favorable scores of Fitness function (FullFitness) concerning ligand binding modes (i.e., position, orientation, and conformation) of these two compounds were comparable. Considering the structure of ARL (C23H32O2, 57 atoms), it contains a larger molecular size compared to that of melatonin (C13H16N2O2, 33 atoms). Typically, a bigger molecule occupies more space and has more atomic elements to interact with the protein receptors by creating van der Waals attractive forces. To elucidate this precaution, we compared the estimated binding energy of melatonin-RORβ to those predicted using MT1 (PDB: 6ME5) and MT3 receptor (PDB: 2QWX) as the known target references. The predicted binding energies were commensurate with that of melatonin with its well-characterized MT1 and MT3 at − 7.52 and − 7.36 kcal/mol, respectively. Taken together, these docking experiments using rat RORβ and human RORs suggested that melatonin was potentially able to bind RORβ. We therefore opted to study the direct interaction between the LBD of RORβ and melatonin experimentally using biophysical approaches, DSF and ITC.

DSF is a medium-throughput biophysical assay based on the accessibility of the hydrophobic dye SYPRO Orange to the protein. The assay will provide insight into the thermal stability of the protein as reflected by the apparent melting temperature, Tma. A protein with high thermal stability will have high Tma. The thermal stability of a protein is affected by many factors, one of which is the ligand binding [41]. DSF analysis of the LBD of RORβ with various concentrations of melatonin showed no significant Tma shift from the control, although a slight decrease in Tma was observed (Fig. 5A, B). We were aware that the LBD of RORβ might not be in a preferred conformation as there is evidence showing that the LBD of rat RORβ adopted its active conformation upon binding with co-activator peptide SRC-1 [16]. Therefore, we further analyzed the Tma of the LBD of RORβ in the presence of various concentrations of SRC-1 co-activator peptide. As expected, SRC-1 peptide could stabilize the LBD of RORβ as a significant shift in Tma was observed (Fig. 5C, D). However, melatonin did not alter the Tma of the LBD of RORβ, albeit in the presence of SRC-1 at all concentrations (Fig. 5C, D). As the DSF result did not support the computational finding, we sought for another biophysical assay that is more robust for interaction elucidation, isothermal titration calorimetry (ITC), as it is possible that true binding may not affect thermal stability of the protein, especially for the low to moderate affinity binders [42]. ITC results agreed with the finding from DSF assay. No interaction could be detected between the LBD of RORβ and melatonin with or without SRC-1 peptide (Fig. 6A, B, E, and F), while a moderate interaction with a KD of 19.8 µM could be calculated for the LBD of RORβ and SRC-1 co-activator peptide, and the binding is thermodynamically favorable (Fig. 6C, D). Our data showed that the human RORβ isoform could not directly bind with melatonin as confirmed by both biophysical techniques even in the presence of co-activator peptide SRC-1.

Overall, we found that molecular docking, although useful for predicting ligand binding modes and affinities computationally, may not always align with experimental data, particularly due to its lack of the physiological environment. Additionally, DSF assesses changes in protein stability upon ligand binding rather than direct binding affinity. In contrast, ITC directly measures heat changes upon ligand binding, providing accurate thermodynamic parameters including KD values in the nanomolar to micromolar range [43]. Therefore, we employed a complementary approach using these techniques to investigate the direct molecular interaction of RORβ and melatonin, acknowledging the unique characteristics and limitations of each method. However, we could not entirely rule out the possibility that RORs and melatonin are somehow related as many reports have shown the association between the two. We thus investigated the effect of melatonin on RORs gene expression in SH-SY5Y neuroblastoma cell line. Our RT-ddPCR gene expression analysis showed an increased expression of neuron-enriched RORs, RORα, and RORβ, upon melatonin treatment of various concentrations for 24 h, while very stably low RORγ transcript was detected at all concentrations tested. Although the increasing trend was observed, only treatments with higher concentrations (1 and 10 µM) of melatonin significantly affected the level of RORα and RORβ (Fig. 7). Furthermore, the upregulation of RORβ mRNA levels was not completely abolished by luzindole pre-treatment. It is commonly accepted that melatonin binds and activates membrane receptors MT1 and MT2 at picomolar to nanomolar concentrations, thereby modulating downstream signaling pathways [44, 45]. At micromolar to millimolar range, melatonin intrinsically exhibits potent antioxidant and free radical scavenging properties [46, 47]. Therefore, there are several reasons to believe that the elevated expression level of RORβ is not wholly through the MT1 and/or MT2 receptors and cascading signal transduction. Firstly, mRNA level of RORβ should be affected by melatonin at nanomolar concentration. Secondly, luzindole pre-treatment should restore RORβ expression levels to the baseline. These findings suggest that the relationship and interplay between melatonin, RORs, and other target genes are much complicated than previously anticipated.

Taken together, our results indicated that melatonin was likely not the natural ligand for RORβ per se, while its effects on RORs might be mediated via its antioxidant properties or its downstream mediators, such as sirtuin 1 (Sirt1), regarding several physiological cellular roles of melatonin and RORs have in common [24, 37, 48]. In addition, melatonin may indirectly relate to RORs by regulating the expression levels of these genes. However, how melatonin is linked with the downstream effect of RORs remains the challenge, and the missing links are remained to be uncovered. Our postulation is that melatonin might bind to a yet-to-identify nuclear receptor and in turn regulates the expression of RORs, which affect the gene expression of the others (Fig. 8). We, therefore, urged the melatonin research community for the paradigm shift in the RORβ as the direct nuclear receptor for melatonin and warrant for the identification of nuclear target(s) of melatonin.

Melatonin was not the ligand for RORs. The actions of melatonin are mediated through signal transduction and its associated melatonin receptors, including the membrane-bound MT1/2 G protein-coupled receptors (GPCRs) and the cytoplasmic MT3 receptor, the latter having antioxidant and free radical scavenging characteristics. Herein, our results showed that melatonin was likely not the ligand for RORβ. Melatonin might bind to an unidentified nuclear receptor and influence the expression of RORs, which in turn affects the gene expression of the others