Structural Perspective of NR4A Nuclear Receptor Family and Their Potential Endogenous Ligands

Abstract

There are 48 nuclear receptors in the human genome, and many members of this superfamily have been implicated in human diseases. The NR4A nuclear receptor family consisting of three members, NR4A1, NR4A2, and NR4A3 (formerly annotated as Nur77, Nurr1, and NOR1, respectively), are still orphan receptors but exert pathological effects on immune-related and neurological diseases. We previously reported that prostaglandin A₁ (PGA₁) and prostaglandin A₂ (PGA₂) are potent activators of NR4A3, which bind directly to the ligand-binding domain (LBD) of the receptor. Recently, the co-crystallographic structures of NR4A2-LBD bound to PGA₁ and PGA₂ were reported, followed by reports of the neuroprotective effects of these possible endogenous ligands in mouse models of Parkinson’s disease. Based on these structures, we modeled the binding structures of the other two members (NR4A1 and NR4A3) with these potential endogenous ligands using a template-based modeling method, and reviewed the similarity and diversity of ligand-binding mechanisms in the nuclear receptor family.

1. INTRODUCTION

Nuclear receptors comprise a superfamily of hydrophobic low molecular ligand-activated transcription factors that regulate gene expression in many physiological and pathological events. Endogenous ligands have only been identified for about half of 48 nuclear receptors in the human genome.^1,2) The remaining receptors, called orphan nuclear receptors, form promising pharmaceutical targets for research and development^2–5) as well as orphan G-protein coupled receptors. Although the nuclear receptor family, nuclear receptor subfamily 4 group A (NR4A), consisting of three members,⁶⁾ NR4A1, NR4A2, and NR4A3 (old annotations are Nur77, TR3, or NGFI-B,⁷⁾ Nurr1,⁸⁾ and NOR1⁹⁾) are orphan receptors, they are promising therapeutic targets in immune diseases,^10–13) neurological diseases,^14–19) and cancer.^20–24) They have been classified as immediate-early genes whose expressions are tightly regulated by extracellular signals. In addition, as shown in Supplementary Fig. S1, NR4A1 and NR4A2, but not NR4A3, can heterodimerize with retinoid X receptor alpha (RXRα) and mediate efficient trans-activation through a direct repeat-5 (DR-5) element in response to the RXR-specific ligand, 9-cis retinoic acid.^25–28) Although the NR4A family shares features of typical steroid receptor organization comprising an N-terminal transactivation domain, a central DNA-binding domain and a C-terminal putative ligand-binding domain (LBD) as shown in Supplementary Fig. S1A, there were few reports about specific ligands that regulate NR4A-dependent transcriptional activity. Moreover, no function has yet been inclusively ascribed to the LBDs of the NR4A superfamily.

Dysfunction of NR4A nuclear receptors is associated with human diseases similar to other nuclear receptor families. NR4A3 is implicated in oncogenesis as part of the EWS fusion protein, resulting from a chromosomal translocation found in human extraskeletal myxoid chondrosarcoma tumors.²⁹⁾ In addition, NR4A3, as well as NR4A1, play a key role in the apoptosis of T lymphocytes, eosinophils, and various other cell types,^{10–13,30–32)} and thus may be implicated in disorders related to genetically- or environmentally-induced defects of activation-induced apoptosis, including autoimmune diseases and allergic diseases. Recently, NR4A3 was also reported to be closely associated with cardiovascular diseases such as atherosclerosis, pulmonary arterial hypertension, and abdominal aortic aneurysm through vascular remodeling.³³⁾ Regarding NR4A2, there have been many reports about its relationship with neurological diseases especially Parkinson’s disease, which is related to the differentiation of midbrain dopaminergic neurons.^14–19)

We examined a large number of lipid mediator metabolites, including those produced by the arachidonate and glycerolipid pathways, which might be induced in various cells following pathological stimuli such as antigen-induced apoptotic signaling. We found that only prostaglandin (PG) A₁ and PGA₂ have the ability to activate NR4A3-dependent transcription, and could bind directly to NR4A3-LBD. Furthermore, primary cultured spleen cells derived from transgenic mice overexpressing NR4A3 showed PGA₂ specific apoptosis in comparison with those derived from wild-type mice.^31,32)

Recently, Rajan et al. reported that PGA₁, PGA₂, and PGE₁ interacted directly with the LBD of NR4A2 to stimulate and NR4A2 nuclear receptor functions. They also reported the co-crystallographic structures of NR4A2-LBD bound to PGA₁³⁴⁾ and PGA₂.³⁵⁾ Furthermore, these PGs exhibited neuroprotective effects in an NR4A2-dependent manner. Based on these results, they suggested these arachidonate metabolites may function as bona fide endogenous ligands of the NR4A2 receptor and have beneficial effects in Parkinson’s disease. However, the crystallographic structure of NR4A3 have not been clarified yet. Regarding NR4A1, the crystallographic structure of the LBD has been reported,³⁶⁾ but there is no information about the transcriptional functions of these arachidonate metabolites or the co-crystallographic structure of NR4A1-LBD bound to these candidate endogenous ligands.

The three-dimensional (3D) structures of molecular complexes assessed by co-crystallographic structures reveal the atomic details of molecular interactions between compound-protein pairs, and thus provide important information to help us understand their biological functions. Computer modeling approaches are frequently performed to extend the data of these 3D complex structures. Both template-based modeling and de novo modeling are performed to model 3D complex structures. Usually, template-based modeling is performed first because it has advantages in terms of accuracy and computation costs, if proper templates are available. Template-based modeling has also been applied to the modeling of small compound-protein complexes. For this approach, a target chemical compound is superimposed onto the template compound using the 3D structural alignment programs of chemical compounds.³⁷⁾ Standard chemical compound-protein docking calculations are often improved using known compound-protein complexes as templates.^38–43) Because the accuracy of template-based modeling largely depends on the choice of the template structure, using a database containing the latest 3D structures with molecular similarity search is essential. The Homology Modeling of Complex Structure (HOMCOS) server (http://homcos.pdbj.org), constructed by Kawabata and Fukuhara in 2008,^44,45) was significantly updated for the template-based modeling of 3D complexes for proteins and small compounds in the Protein Data Bank (PDB) in 2016.⁴⁶⁾

Then, we tried to apply this in silico approach to analyze interactions between the LBDs of NR4A in the nuclear receptor family and compounds including candidate endogenous ligands.

2. THE NR4A NUCLEAR RECEPTOR FAMILY HAS MANY POSSIBLE PHARMACOLOGICAL FUNCTIONS

2.1. NR4A Family Members as Therapeutic Targets for Allergic and Neurologic Diseases

To identify novel genes related to the clinical signs of allergic diseases such as atopic dermatitis (AD), differentially expressed genes were comprehensively evaluated by comparing peripheral blood leukocytes from patients and healthy volunteers.^47,48) We found all three genes, NR4A1, NR4A2, and NR4A3, members of the NR4A nuclear family, had significantly higher levels in AD patients than in healthy volunteers.³⁰⁾ Differential expressions of NR4A1 and NR4A2 were shown in whole blood leukocytes. However, NR4A3 was selected as a peripheral blood eosinophil-specific gene. NR4A3 expression in peripheral blood eosinophils of AD patients was enhanced with exacerbation and decreased with remission of the disease. The NR4A3 gene was also highly expressed in surgically removed nasal polyps containing large eosinophil accumulations.³²⁾ We also reported that the NR4A1 and NR4A3 genes were markedly induced in correlation with eosinophil-specific apoptosis.³¹⁾ In contrast, the coincidental downregulation of NR4A1 and NR4A3 was reported in myeloid leukemogenesis.²⁴⁾ Such information suggests that these 2 receptors of the NR4A family are related to each other, with similar pathological roles.

Regarding NR4A2, there are many reports that this nuclear receptor is indispensable for the differentiation, maturation, and maintenance of midbrain dopaminergic neuron clusters.^14–19) NR4A2-deficient mice failed to generate midbrain dopaminergic neurons.⁴⁹⁾ Then, it was reported that NR4A2 expression in the brains of Parkinson’s disease patients and rodent models of Parkinson’s disease was downregulated.^17–19) To date, no mechanistic hypothesis has been put forth to explain this reduced expression. Therefore, the development of selective NR4A2 agonists may provide target-based therapeutic interventions for Parkinson’s disease and other related diseases.⁵⁰⁾

2.2. Activation of NR4A3 by PGA₁ and PGA₂

We thought that the NR4A orphan nuclear receptors may not be genuinely orphan in nature, and that some endogenous ligands may be present close to where these receptors are expressed at high levels. It is the similar strategy that chenodeoxycholic acid was proved to be an farnesoid X receptor (FXR) ligand in the liver, an organ with a high expression of the nuclear receptor subfamily 1, group H (NR1H) family member FXR.^51,52) Thus lipid mediators such as arachidonate and glycerolipid cascade metabolites in the mammalian lipid metabolic pathways, which might be induced in activated leukocytes following pathological stimuli, were screened. We need a mammalian hybrid system for this screening, in which GAL4-NR4A receptor-linked chimera plasmids and the GAL4 binding domain-firefly luciferase reporter plasmid were co-transfected into cultured cells, and luciferase activity after ligand candidate addition was assayed. PGA₁ and PGA₂ were found to have NR4A3 nuclear receptor-LBD dependent transactivation activity in this system (13,14-dihydro-15-keto-PGA₂ and PGE₁ were negative in this system.).³¹⁾

The putative LBD of the nuclear receptor was also shown to bind directly to PGA₁ and PGA₂ using high-resolution Biacore S51.³¹⁾ Evaluation of the effects of PGA₁ and PGA₂-like cyclopentenone derivatives on NR4A3 LBD-dependent transcription demonstrated subtype-specific activators and clear structure–activity relationships, as assessed by minor modification of the functional groups in these similar PG structures.

We also established NR4A3 transgenic mice and surprisingly found that PGA₂ had a strong stimulatory effect on apoptosis induction in spleen cells derived from these mice compared with wild-type mice.³¹⁾ These findings also suggest specific molecular linkage between PGA and NR4A3.

PGA₂ is the non-enzymatic breakdown product of PGE₂, and PGE₂ is one of the main products of arachidonate cascade highly produced by cyclooxygenase in inflammatory stimulated leukocytes including eosinophils. It means that our endogenous ligand screening strategy might be successful as in case of chenodeoxycholic acid and FXR.^51,52) Peroxisome proliferator-activated receptor γ (PPARγ) is the reference as a more familiar nuclear receptor in this review. Other arachidonate metabolites, PGJ₂ and 15d-PGJ₂, the non-enzymatic products of PGD₂, were proved to be endogenous ligands of PPARγ in 1995.⁵³⁾ These evidences might be correlated well with our findings. However, PGJ₂ and 15d-PGJ₂ were not active in our system.

2.3. Interaction of NR4A2 with PGs and Binding Structures between NR4A2 and Its Endogenous Ligand Candidates by X-Ray Co-crystallography

After more than 15 years of our publications,^30–32) two exciting papers were published in 2020³⁴⁾ and 2022³⁵⁾ by the same group. Rajan et al. showed that PGE₁ and its dehydrated metabolites, PGA₁ interacts directly with the LBD of NR4A2 and stimulates its transcriptional function.³⁴⁾ They additionally reported that PGA₂ has the same activities.³⁵⁾ They also proved the X-ray crystallographic structures of NR4A2 bound to PGA₁ and PGA₂. Each co-crystal structure of NR4A2-LBD bound to PGA₁ and PGA₂ revealed the covalent couplings of PGA₁ and PGA₂ with the LBD through Cys566 by forming Michael adducts. Both PGA₁ and PGA₂ bindings also induced notable conformational changes, including a 21° shift of the activation function 2 region (AF-2) in the C-terminal helix H12 away from the protein core. Both PGA₁ and PGA₂ physical bindings with NR4A2-LBD were also supported by NMR titration. PGE₁ was interacted with NR4A2-LBD, but was not covalently bound to the LBD.

The pharmacological functions of PGA₁, PGA₂, and also PGE₁ were demonstrated in the 1-methyl-4-pheny-1,2,3,6-tetrahydropyridine (MPTP)-induced lesioned mouse model³⁴⁾ or leucine-rich repeat kinase2 (LRRK2) G2019S transgenic fly model,^35,54) both models of Parkinson’s disease, in which these arachidonate metabolites rescued the locomotor deficits and neuronal degeneration. These findings by the Rajan group suggest these arachidonate metabolites may represent bona fide endogenous ligands of NR4A2. Together with our previous findings, PGA₁/PGA₂ might be the endogenous ligands of NR4A nuclear receptor superfamily members, although there have been no reported bindings of the compounds PGA₁ and PGA₂ to the protein NR4A1.

There have been few structural studies of the NR4A receptor family with endogenous ligand candidates, except NR4A2. Concerning NR4A1, there were some reports about 3D structures and synthetic low molecular agonists,^55–58) but no wet data related with arachidonate metabolites. Therefore, we modeled 3D structures between these ligand candidates and this nuclear receptor family using the template-based modeling methods described in the following chapters.

3. INTERACTIONS OF NR4A NUCLEAR RECEPTOR FAMILY MEMBERS WITH ENDOGENOUS LIGAND CANDIDATES REVEALED BY EXPERIMENTAL AND MODELED 3D STRUCTURES

3.1. 3D Structures of Ligand Binding Domains of NR4A

Several X-ray crystallographic structures of the LBD of NR4A1 and NR4A2 have been solved, although no structures are available for the LBD of NR4A3 to date. The PDB IDs of these structures are summarized in Table 1, and three of them are shown in Fig. 1. Between similar families of nuclear receptors, PPARγ was one of the earliest factors to have its 3D structure elucidated, revealing it possessed a canonical ligand-binding site. We demonstrate PPARγ as a positive control of the most similar and successful pharmaceutical target among nuclear receptor families. Because there were many references about 3D structure and the natural ligands were proved to be arachidonate cascade products, PGJ₂ and 15d-PGJ₂.⁵³⁾ Some synthetic ligand compounds are also clinically successful in improvement of insulin resistance. Figure 1C shows that the canonical binding site of PPARγ holds the compound 15d-PGJ₂ by a covalently-bonded cysteine. Many agonistic and antagonistic compounds for PPARγ were reported to bind to the canonical binding site.^59,60)

Table 1. PDB IDs of 3D Structures of the Ligand Binding Domain of the NR4A Family and PPARγ

	Apo	With synthetic liganda)	With natural liganda)
NR4A1	3v3e	3v3q[TMY](TMPA), 4rzg[ZHN](PDNPA), 4kzi[ZJ0], 4ref[3N0], 4whg[3NB],4jgv[T94](THPN), 4ree[3MZ], 4whf[3MX], 4re8[3MJ]	NA
NR4A2	1ovl	8cyo[OBJ]	5y41[RPG](PGA1), 5yd6[8SU](PGA2), 6dda[G7J](dopamine metabolite)
NR4A3	NA	NA	NA
PPARγ	1prg	2xkw[P1B](pioglitazone)	2zk1 [PTG](15d-PGJ2)

a) In square brackets [ ], the compound IDs (comp_id) of the PDB are indicated, and in parentheses ( ), the common names of the compounds are shown.

Fig. 1. 3D Structures of the NR4A2 Family and PPARγ

Molecular graphics were created using UCSF Chimera.⁸¹⁾ Using the program MATRAS,⁸²⁾ orientations of these protein structures are aligned to match the NR4A2 structure with PGA₁ (PDB ID: 5y41). (A) The holo structure of NR4A2 bound to PGA₁. The PDB ID of the structure is 5y41.³⁴⁾ The helix H12 has an open conformation. The orange region is the I-box region (K554-L562), which is important for NR4A2-RXRα heterodimerization.⁷⁷⁾ The cyan region is the AF-2 (activation function 2) region (A586-L593), which is important for a transcription activation function.⁸³⁾ (B) The holo and the apo structure of NR4A2. The apo structure is taken from PDB ID 1ovl,⁶¹⁾ and is colored dark slate blue. Upon the binding of the ligand PGA₁, the helix H12 (AF-2 region) rotates 19.3 degrees outward from the protein’s core. The angle change of Helix H12 (19.3 degrees) was measured by the bond angle between the Cα atoms at the C-terminus, with the Cα atom at the N-terminus as the center. (C) The structure of NR4A1 with nine compounds.^55–58) The protein structure is taken from PDB ID:3v3q. The helix H12 has a closed conformation, whereas the helix H10 is slightly open. Compound names are 3-letter comp_id of the Protein Data Bank. Three compounds (TMY, ZHN, and ZJ0) bound to the binding site of PGA₁ are marked by a dotted red ellipse. The other six compounds (T94, 3 MJ, 3MZ, 3MZ, 3NB and 3N0) are bound to other sites of NR4A1. Compound structures (shown in Supplementary Fig. S2) are taken from the PDB structures 3v3q for TMY, 4rzg for ZHN, 4kzi for ZJ0, 4jgv for T94, 4re8 for 3MJ, 4ree for 3MZ, 4whf for 3MX, 4whg for 3NB, and 4ref for 3N0. (D) The structure of PPARγ bound to 15d-PGJ₂. The PDB ID of the structure is 2zk1.⁶⁵⁾ The binding site of PPARγ is called “canonical binding site.” The helices H10 and H12 have closed conformations.

In 2003, the first X-ray crystal structure of NR4A2 in the apo state was reported by Wang et al. They reported that NR4A2 lacked the canonical binding pocket due to the tight packing of hydrophobic side chains⁶¹⁾ (Figs. 2A, B). Therefore, it was hypothesized that the NR4A superfamily might be ligand-independent orphan nuclear receptors. However, more than fifteen years later, Bruning et al. reported a complex structure of NR4A2 with a dopamine metabolite IQ (an oxidized DHI, as shown in Supplementary Fig. S3) (PDB ID:6dda),⁶²⁾ and Rajan et al. solved the structures of NR4A2 with PGA₁ and PGA₂ (PDB ID:5y41 and 5yd6).^34,35) These three compounds were placed in a non-canonical pocket (PGA₁: Figs. 1A, 5A; PGA₂: Fig. 5B; IQ: Fig. 5F). This pocket did not exist in the apo structure; the H12 helix was reoriented away from the apo structure by an angle of 21° to form a non-canonical binding pocket, as shown in Figs. 1B, 2C, and D. These three ligands were covalently bound to the Cys566 of NR4A2, as shown in Figs. 1A, 2E, 5A, B, and F. Figure 2E shows a 2D ligand–protein interaction diagram for PGA₁ and NR4A₂ drawn using LigPlot+.⁶³⁾ Figures 2D and E indicate that the binding site of PGA₁ is mainly hydrophobic, except for the carbonyl group (O9) on the five-membered ring. The carboxyl group (C1, O1, and O2 atoms) of PGA₁ was exposed to bulk water and did not form any bonds with specific protein atoms.

Fig. 2. Binding Sites of NR4A2 and PPARγ

(A) A cross section of the “canonical binding site” of PPARγ with 15d-PGJ₂. (PDB_ID: 2zk1). (B) A cross section of the region of NR4A2 corresponding to the “canonical binding site” (PDB_ID: 1ovl). The binding site is occupied by the tightly packed hydrophobic side chains. (C) The binding site of the apo structure of NR4A2 (PDB_ID: 1ovl).⁶¹⁾ (D) Binding site of the holo structure of NR4A2 bound to PGA₁ (PDB_ID: 5y41).³⁴⁾ Molecular surfaces are colored using the Kyte–Doolittle hydrophobic scale. Hydrophobic residues are colored in red, whereas hydrophilic residues are colored in blue. (C) 2D ligand–protein interaction diagram for PGA₁ and NR4A2 (PDB ID:5y41) drawn by LigPlot+.⁶³⁾

The binding structures of NR4A1 for nine synthetic chemical compounds have been solved, as shown in Fig. 1C. These artificial low molecular compounds were reported to be agonists and antagonists of NR4A1.^55–58) TMPA [TMY] antagonizes the NR4A1-LKB1 interaction,⁵⁵⁾ THPN [T94] induces autophagy with the participation of NR4A1,⁵⁶⁾ and PDNPA [ZHN] inhibits the interaction between NR4A1 and p38α.⁵⁷⁾ The chemical structures of these compounds are shown in Supplementary Fig. S2. These nine compounds can be classified into two groups by the locations of their binding sites. The three compounds (shown in the dotted red circle in Fig. 1B) were bound to the PGA-binding site, whereas the other six compounds (shown in dotted blue circles in Fig. 1B) were bound to a site that is far from the canonical binding site and the PGA-binding site. The chemical structures of the six compounds share a common substructure (3,4,5-trihydroxyphenyl group) and their bound 3D structures overlapped compared to the 3D structure of the three compounds. Thus, these three compounds might inhibit PGA₁, PGA₂, and IQ from binding to NR4A proteins because they bind close to the PGA-binding site.

3.2. Multiple Sequence Alignment of the Ligand Binding Domains of NR4A

The multiple alignment of the proteins NR4A1, NR4A2, NR4A3, and PPARγ was generated using the program ClustalW2,⁶⁴⁾ and shown in Fig. 3 with the binding sites and secondary structures. The binding sites of PGA1 for NR4A2 are shown in green boxes, and the canonical ligand binding sites for PPARγ are shown in magenta boxes.

Fig. 3. Multiple Sequence Alignment (MSA) of the Ligand Binding Domain (LBD) of the NR4A Family with Structural Information

MSA was calculated by ClustalW2.⁶⁴⁾ Secondary structures of NR4A2 (PDB_ID: 5y41) are shown in the upper MSA. The alignment between PPARγ and NR4A2 was obtained by 3D structural alignment (between PDB IDs 1prg and 5y41) using the protein structure comparison program MATRAS.⁸²⁾ The amino acid sequence of PPARγ (PPARG_HUMAN) is shown at the bottom of MSA with its secondary structures. Green boxes on MSA are the binding sites of PGA₁ obtained from NR4A2 (PDB_ID:5y41). Binding sites for three compounds (TMY, ZHN, and ZJ0) are indicated by red boxes on the sequence NR4A1, whereas those for six compounds (T94, 3 MJ, 3MZ, 3MZ, 3NB and 3N0) are shown by blue boxes. The bound structures of these compounds are shown in Fig. 1B. Binding sites of PPARγ for 15d-PGJ₂ (canonical binding sites) are shown in purple boxes on the sequence PPARG_HUMAN. The I-box region (K554-L562) in NR4A2 is colored orange, which was reported to be important for NR4A2-RXRα heterodimerization.⁷⁷⁾ The AF-2 (activation function 2) region (A586-L593) is colored cyan, which was reported to be important for a transcription activation function.⁸³⁾

A comparison of the two binding sites indicates the following three points. First, the canonical binding sites and the PGA₁-NR4A2 binding sites have few common sites. This is consistent with the spatial difference between the two binding sites shown in Figs. 1A and C. Second, the PGA₁/PGA₂-NR4A2 binding sites are conserved with nearly the same amino acids in the proteins NR4A1 and NR4A3, but not in PPARγ. This suggests that NR4A1 and NR4A3 are likely to bind to PGA₁ and PGA₂ molecules in a similar way. Third, the Cys566 involved in the covalent binding of PGA₁ to NR4A2 (at “TLCTQ” in NR4A2) is replaced by valine (V) in PPARγ, whereas the Cys285 (Cys313 in the UniProt PPARG_HUMAN) involved in the covalent binding of 15d-PGJ₂ to PPARγ (at “QGCQF” in PPARγ) is replaced by phenylalanine (F) in NR4A. These findings suggest that PGA1 and PGA2 cannot bind to PPARγ in the same way as NR4A2, and that 15d-PGJ₂ cannot bind to NR4A proteins in the same way as PPARγ.

3.3. Chemical Structures of Endogenous Ligand Candidates and Their Binding to Proteins

The chemical structures of the endogenous ligand candidates are shown in Fig. 4, which are related to their ability to form covalent bonds with protein cysteine residues. The complex structures of NR4A2 with PGA₁ and PGA₂ reveal that a covalent bond was formed between the C11 carbon atom of the cyclopentenone ring (red circles in Figs. 4A, B) of the ligand and the thiol group of Cys566 in NR4A2.^34,35) This suggests that PGE₁, PGJ₂, and 15d-PGJ₂ cannot form a covalent bond because the C11 carbons of these three ligands (dotted red circles in Figs. 4C–E) have additional hydroxyl or carbonyl groups.

Fig. 4. Chemical Structures of Endogenous Ligand Candidates

(A) Prostaglandin A₁ (PGA₁). (B) Prostaglandin A₂ (PGA₂). (C) Prostaglandin E₁ (PGE₁). (D) Prostaglandin J₂ (PGJ₂). (E) 15-Deoxy-Delta12,14-prostaglandin J₂ (15d-PGJ₂). (F) Dopamine metabolite 5,6-indole-quinone (IQ). Note that the molecule 5,6-dihydroxyindole (DHI) is self-oxidized into the molecule 5,6-indole-quinone (IQ), and IQ binds to NR4A2 as Cysteine IQ adduct as shown in Supplementary Fig. S3. (G) Docosahexaenoic acid (DHA). The solid red circles are carbon atoms (the C11 atoms of the rings in PGA₁ and PGA₂, the C2 atom in the indole ling of DHI), which form covalent bonds between the thiol group and Cys566 of NR4A2. The dotted red circles are the corresponding carbon atoms of PGE₁, PGJ₂, and 15d-PGJ₂. These carbons cannot form covalent bonds with proteins because –OH or = O groups are attached. The solid green circle is the unsaturated carbon atom of 15d-PGJ₂, which form a covalent bond with the thiol group of Cys285 of PPARγ.

However, in the complex structure of PPARγ with 15d-PGJ₂, a covalent bond was formed between the C13 carbon atom of the hydrophobic tail (a green circle in Fig. 4E; described as C8 in PTG of PDB ID: 2zk1) and the Cys285 in PPARγ.⁶⁵⁾ The C13 carbon of 15d-PGJ₂ has a double bond with the five-membered ring. This double bond does not exist in the other four prostaglandins. Therefore, these four molecules are unlikely to form covalent bonds with the Cys313 in PPARγ similar to 15d-PGJ₂. Additionally, the two molecules PGJ₂ and 15d-PGJ₂ lack a carbonyl group (O9) in the five-membered ring (purple circles in Figs. 4A–C), which can form a hydrogen bond with Thr567 or Arg563 (Fig. 2E). The lack of the carbonyl group may weaken the interaction between PGJ₂ and 15d-PGJ₂ for the NR4A proteins.

3.4. Models of the Bound Structures of Candidate Endogenous Ligands on NR4A

As explained previously, experimental complex structures are only available for NR4A2 with the PGA₁, PGA₂, and IQ molecules. To investigate experimentally undetermined binding structures, we conducted a computer modeling to determine the binding structures of candidate endogenous ligands and proteins from the NR4A subfamily. Using the three binding structures as templates, a template-based modeling approach can be applied to model the complex structures of similar compounds with similar proteins. We generated the 3D structural models of seven molecules (PGA₁, PGA₂, PGE₁, PGJ₂, 15d-PGJ₂, IQ, and docosahexaenoic acid [DHA]) on the three NR4A proteins (NR4A1, NR4A2, and NR4A3). The cysteine IQ adduct was employed as an oxidized binding form of IQ as shown in Supplementary Fig. S3. Protein sequences were taken from UniProt,⁶⁶⁾ chemical structures were obtained from the KEGG database,⁶⁷⁾ and their initial 3D conformations were generated by the program OpenBabel.⁶⁸⁾ We used the HOMCOS server⁴⁶⁾ to prepare the template structures, the program MODELLER⁶⁹⁾ for the homology modeling of proteins, the program fkcombu^41,70) for the flexible alignment of chemical structures, and the program AMBER⁷¹⁾ to refine conformations. The details of computational modeling procedures are described in the Supplementary materials. The model structures are available from the Biological Structure Model Archive (BSM-Arc) under BSM-ID BSM00054 (https://bsma.pdbj.org/entry/54).⁷²⁾

The compounds and proteins had sufficient similarities with the template complex structures; the five prostaglandin molecules share a common chemical structure where a five-membered ring is attached to a hydrophilic tail and hydrophobic tail (Fig. 4), and the binding sites of PGA₁ on NR4A2 are well conserved in NR4A1 and NR4A3 (Fig. 3). In the template structures of the ligands PGA₁ and PGA₂, the atom C11 forms a covalent bond with the proteins’ cysteine residue (Cys566). When superimposing the target ligands PGE₁, PGJ₂, and 15d-PGJ₂ onto the template ligand of PGA₁ or PGA₂ using the program fkcombu,^41,70) the functional group attached to C11 (–OH or = O) clashes with the side chain of the cysteine. These atomic clashes were avoided by performing energy minimization calculations using the AMBER program,⁷¹⁾ which allows for the reorientation of the compound and deformation of the side chain of the cysteine.

Modeled structures for NR4A2 and NR4A3 are shown in Figs. 5 and 6, respectively, and those for NR4A1 are shown in Supplementary Fig. S5. No atomic clashes or unnatural torsions existed in the created model structures. Due to the amino acid conservation of the binding sites, the interactions in the modeled structures NR4A1 and NR4A3 should be similar to those of the template structure NR4A2. Observation of the binding structures of the template structures PGA₁ and NR4A2 demonstrated that the interactions between the ligand and protein primarily involved hydrophobic interactions and covalent bonding with cysteine (Fig. 2E). The most significant difference in binding among the seven modeled compounds was the presence or absence of a covalent bond with cysteine. PGA₁, PGA₂, and IQ (an oxidized DHI, as shown in Supplementary Fig. S3) can form a covalent bond with cysteine, whereas PGE₁, PGJ₂, 15d-PGJ₂, and DHA cannot. The absence of covalent bonds does not necessarily mean weaker binding. For example, Rajan et al. reported that the binding affinities of the NR4A2 protein for PGA₁ and PGE₁ did not differ significantly; NMR titration experiments showed that the binding affinity of PGA₁ with NR4A2 was slightly higher than that with PGE₁, whereas competitive inhibition experiments for [³H]-PGE₁ and [³H]-PGA₁ showed that the affinity of PGA₁ was slightly lower.³⁴⁾ However, the nature of the binding for covalent-bonded compounds may differ from non-covalent-bonded compounds. The 'covalent advantage' was coined to describe the unique feature of covalent modification.⁷³⁾ Since irreversible modifications of proteins can accumulate over time and amplify a signal, even low levels of ligands may be able to initiate signaling.

Fig. 5. Experimental and Model Complex Structures of NR4A2 and Endogenous Ligand Candidates

Bound compounds are drawn as a ball-and-stick model, and Cys566, Arg563 and Thr567 are drawn as a stick model. (A) Experimental structure with PGA₁ (PDB ID:5y41).³⁴⁾ (B) Experimental structure with PGA₂ (PDB ID:5yd6).³⁵⁾ (C) Model structure with PGE₁, (D) PGJ₂, and (E) 15d-PGJ₂. (F) Experimental structure with Cysteine IQ adduct (Supplementary Fig. S3) (PDB ID:6dda).⁶²⁾ (G) Model structure with DHA.

Fig. 6. Model Complex Structures of NR4A3 and Endogenous Ligand Candidates

Bound compounds are drawn as a ball-and-stick model, Cys594, Arg591, and Thr595 are drawn as a stick model. (A) Model structure with PGA₁, (B) PGA₂. (C) PGE₁, (D) PGJ₂, (E) 15d-PGJ₂, (F) Cysteine IQ adduct (Supplementary Fig. S3), and (G) DHA.

The two molecules PGJ₂ and 15d-PGJ₂ lack a carbonyl group (O9) on the five-membered ring, which can form a hydrogen bond with Thr567 or Arg563 (Figs. 2E, 4D, and 4E). Note that the molecule PGE₁ cannot make a cysteine-covalent bond but has the O9 carbonyl group to form a hydrogen bond with the protein. Rajan et al. reported that PGJ₂ and 15d-PGJ₂ had no effect on the transcriptional activity of NR4A2-LBD, whereas PGE₁ prominently stimulated NR4A2-LBD reporter activity, and NMR spectroscopy revealed the same binding region of NR4A2 was recognized by PGE₁ and PGA₁.³⁴⁾

A metabolomic study reported interactions between DHA and the NR4A1 protein.⁷⁴⁾ Although a crystal structure of bound DHA is not available to date, de Vera et al. reported several interacting sites of NR4A2 for DHA using chemical shift perturbation by solution NMR spectroscopy.^75,76) They identified more than 20 perturbed residues by DHA-binding, covering both the canonical binding site and the PGA1-binding site of NR4A2. DHA binding also affected the methyl groups of Leu410, Ile483, and Ile486. The residue Leu410 is placed on the PGA1-binding site of NR4A2. Because our 3D structure model of DHA was built by fitting on the bound-PGA₁ molecule, it has contact with Leu410 (Fig. 5G). However, the residues, Ile483 and Ile486 are far from the position of the modeled DHA, rather than close to the canonical binding site. In that sense, our NR4A2-DHA model is hypothetical and cannot explain all the perturbed residues reported by de Vera et al.^75,76) We suggest that multiple binding sites for DHA can exist in the NR4A proteins, and our model should be one of them.

3.5. Heterodimerization and Mechanism of Transactivation

A detailed mechanism of how LBD activates DNA binding is not well understood; however, most researchers assume that the binding of an agonist to LBD leads to a conformational change of LDB, which enhances the dimerization of the LBD of NR4A proteins, and finally, the dimerization affects DNA-binding (Supplementary Fig. S1). When NR4A1 or NR4A2 form heterodimers with RXRα, their DNA-binding ability is enhanced, whereas NR4A3 does not form heterodimers.^25–28) Experimental 3D structures for the LBD heterodimer of NR4A and RXRα are not available to date, but limited structural information has been reported. Aarnisalo et al. reported that the sequence motif I-box (K554-L562; KLLGKLPEL) in NR4A2 is essential for NR4A2-RXRα interactions.⁷⁷⁾ The location of the I-box motif is shown in the structure in Fig. 1A and in the alignment in Fig. 3. A conformational change of the I-box by ligand binding might be necessary for dimerization. Interestingly, the sequences of the I-box are not entirely conserved within the NR4A family, namely, NR4A1: RLLGKLPEL, NR4A2: KLLGKLPEL, and NR4A3: KVLGALVEL, as shown in Fig. 3. The difference in these I-box sequences might explain the different heterodimerization between NR4A1, NR4A2, and NR4A3. Recently, Zhao et al. modeled the structure of the NR4A2-RXRα heterodimer with the help of cross-linking mass spectrometry and molecular dynamics.⁷⁸⁾ Their 3D model of the heterodimer is highly flexible, especially in the DBD. Estimated interacting residue pairs from their model are consistent with the I-box reported previously.

Such precise analysis between conserved amino acid sequences and conformational changes of these receptors by various ligands might be important to clarify the similarity and diversity of activation mechanisms in nuclear receptor families. For the purpose of drug discovery, relationships between ligand binding and pharmaceutical activity caused by activation of each receptor will also be important.^2,79)

4. CONCLUSION

Our previous group in the Genox Research Institute published an original paper in this Journal, Biol. Pharm. Bull., entitled as “Prostaglandin (PG) A₂ acts as a trans-activator for NOR1 (NR4A3) within the nuclear receptor superfamily.”³¹⁾ After more than 15 years, the citation rates and inquiries of this paper have increased dramatically, which might be related to the recent publication of two papers by another group.^34,35)

As a type of de novo hydrophobic ligands, steroid compounds have important roles in health and have contributed to the clinical environment, working through various nuclear receptors. Besides steroids, there have been some successful nuclear receptor–ligand relationships such as retinoic receptors vs. retinoids and PPARγ vs. thiazolidine derivatives. There are still some orphan receptor families in the 48 nuclear receptor genome, whose endogenous ligands have not been identified yet.²⁾ The NR4A superfamily, which consists of 3 receptors, is a typical example. There are no clinically successful cases of NR4A superfamily members yet, although there have been many preclinical studies on the pathological importance of this family.

This might be because of an X-ray structural report in 2003 by Wang et al. in which the Nurr1 (NR4A2) LBD contained no cavity and Nurr1 lacked a ‘classical’ binding site for coactivators.⁶¹⁾ At that time, technology related to in silico ligand analysis or ligand design approaches was limited. We previously reported specific molecular linkage between PGA₁/PGA₂ and NR4A3 and suggested that PGA₁/PGA₂ might be endogenous ligands of the NR4A nuclear receptor family,^31,32) but these ideas were not widely accepted.

After recent two reports about the co-crystallographic structures of NR4A2-LBD bound to PGA₁ and PGA₂,^34,35) we focus on the further possibilities of the NR4A family as promising pharmaceutical molecular targets for immunological or neurological diseases. It is especially because information of pathological importance related to these receptors has increased and in silico simulation technologies between receptors and ligands have progressed greatly during the last 15 years.

Recent reports on ligand-bound 3D structures of NR4A proteins have enabled us to perform template-based comparative modeling to build 3D complex structures of many similar proteins and compounds, as shown in this paper. For a broader range of compounds, molecular dynamics, and covalent ligand docking will be necessary to predict their binding structures and affinities, because there are two challenging aspects for modeling and simulating NR4A-LBD and ligands. First, the conformation of NR4A-LBD is highly dynamic and flexible. This has been demonstrated through NMR studies^75,76) and the significant structural differences between the apo and holo forms (Figs. 1B, 2C, D). The binding of unknown ligands may lead to novel binding pockets that have not been identified yet. Long-time molecular dynamics or coarse-grained simulation may be required to capture the flexibility of NR4A-LBD. Second, covalent bonding with the cysteine of the NR4A is necessary for the binding of some natural ligands (PGA₁, PGA₂, and IQ). Because classical molecular mechanics cannot represent the formation of covalent bonds, most docking programs struggle to calculate conformational potential energy and binding free energy for covalently bound ligands. Recently, several programs for covalent ligand docking have been developed, which might help model the LBD-ligand system.⁸⁰⁾

The NR4A nuclear receptor family is among the most promising pharmaceutical targets as well as other nuclear receptor families although they are still recognized as an orphan receptor family. To analyze the binding and activation mechanisms of these receptor family members with ligand candidates such as the PGs discussed here, will be very informative. The design of novel compounds that function as agonists and activate these receptors will help develop future therapeutic strategies for NR4A-related human disorders. For this purpose, using in silico simulation tools for template-based modeling will be very useful.

Acknowledgments

This research was supported by the Research Support Project for Life Science and Drug Discovery (Basis for Supporting Innovative Drug Discovery and Life Science Research (BINDS)) from AMED under Grant Number: JP23ama121019. The model structures have been submitted to the Biological Structure Model Archive (BSM-Arc) under BSM-ID BSM00054 (https://bsma.pdbj.org/entry/54).⁷²⁾

Conflict of Interest

The authors declare no conflict of interest.

Supplementary Materials

This article contains supplementary materials.

REFERENCES

• 1) Maglich JM, Sluder A, Guan X, Shi Y, McKee DD, Carrick K, Kamder K, Willson TM, Moore JT. Comparison of complete nuclear receptor sets from human, Caenorhabditis elegans and Drosophila genomes. Genome Biol., 2, 1–7 (2001).
• 2) Burris TP, de Vera IMS, Cote I, et al. International union of basic and clinical pharmacology CXIII: nuclear receptor superfamily-update 2023. Pharmacol. Rev., 75, 1233–1318 (2023).
• 3) Kliewer SA, Lehmann JM, Willson TM. Orphan nuclear receptors: shifting endocrinology into reverse. Science, 284, 757–760 (1999).
• 4) Chawla A, Repa JJ, Evans RM, Mangelsdorf DJ. Nuclear receptors and lipid physiology: opening the X-files. Science, 294, 1866–1870 (2001).
• 5) Isigkeit L, Merk D. Opportunities and challenges in targeting orphan nuclear receptors. Chem. Commun. (Camb.), 59, 4551–4561 (2023).
• 6) Maruyama K, Tsukada T, Ohkura N, Bandoh S, Hosono T, Yamaguchi K. The NGFI-B subfamily of the nuclear receptor superfamily. Int. J. Oncol., 12, 1237–1243 (1998).
• 7) Chang C, Kokontis J, Liao SS, Chang Y. Isolation and characterization of human TR3 receptor, a member of steroid receptor superfamily. J. Steroid Biochem., 34, 391–395 (1989).
• 8) Okabe T, Takayanagi R, Imasaki K, Haji M, Nawata H, Watanabe T. cDNA cloning of a NGFI-B/nur77-related transcription factor from an apoptotic human T cell line. J. Immunol., 154, 3871–3879 (1995).
• 9) Ohkura N, Hijikuro M, Yamamoto A, Miki K. Molecular cloning of a novel thyroid/steroid receptor superfaily gene from cultured rat neuronal cells. Biochem. Biophys. Res. Commun., 205, 1959–1965 (1994).
• 10) Cheng LE, Chan FK, Cado D, Winoto A. Functional redundancy of the Nur77 and Nor1 orphan steroid receptors in T-cell apoptosis. EMBO J., 16, 1865–1875 (1997).
• 11) Steff AM, Trop S, Maira M, Drouin J, Hugo P. Opposite ability of pre-TCR and alpha beta TCR to induce apoptosis. J. Immunol., 166, 5044–5050 (2001).
• 12) He YW. Orphan nuclear receptors in T lymphocyte development. J. Leukoc. Biol., 72, 440–446 (2002).
• 13) Winoto A, Littman DR. Nuclear hormone receptors in T lymphocytes. Cell, 109 (Suppl.), S57–S66 (2002).
• 14) Kim JH, Auerbach JM, Rodriguez-Gomez JA, Velasco I, Gavin D, Lumelsky N, Lee SH, Nguyen J, Sanchez-Pernaute R, Bankiewicz K, McKay R. Dopamine neurons derived from embryonic stem cells function in an animal model of Parkinson’s disease. Nature, 418, 50–56 (2002).
• 15) Zetterstrom RH, Williams R, Perlmann T, Olson L. Cellular expression of the immediate early transcription factors Nurr1 and NGFI-B suggests a gene regulatory role in several brain regions including the nigrostriatal dopamine system. Brain Res. Mol. Brain Res., 41, 111–120 (1996).
• 16) Saucedo-Cardenas O, Quintana-Hau JD, Le WD, Smidt MP, Cox JJ, De Mayo F, Burbach JP, Conneely OM. Nurr1 is essential for the induction of the dopaminergic phenotype and the survival of ventral mesencephalic late dopaminergic precursor neurons. Proc. Natl. Acad. Sci. U.S.A., 95, 4013–4018 (1998).
• 17) Decressac M, Kadkhodaei B, Mattsson B, Laguna A, Perlmann T, Bjorklund A. Alpha-synuclein-induced down-regulation of Nurr1 disrupts GDNF signaling in nigral dopamine neurons. Sci. Transl. Med., 4, 163ra156 (2012).
• 18) Le W, Conneely OM, He Y, Jankovic J, Appel SH. Nurr1 expression increases the vulnerability of mesencephalic dopamine neurons to MPTP-induced injury. J. Neurochem., 73, 2218–2221 (1999).
• 19) Chu Y, Kompoliti K, Cochran EJ, Mufson EJ, Kordower JH. Age-related decreases in Nurr1 immunoreactivity in the human substantia nigra. J. Comp. Neurol., 450, 203–214 (2002).
• 20) Maruyama K, Tsukada T, Bandoh S, Sasaki K, Ohkura N, Yamaguchi K. Retinoic acids differentially regulate NOR-1 and its closely related orphan nuclear receptor genes in breast cancer cell line MCF-7. Biochem. Biophys. Res. Commun., 231, 417–420 (1997).
• 21) Ohkubo T, Ohkura N, Maruyama K, Sasaki K, Nagasaki K, Hanzawa H, Tsukada T, Yamaguchi K. Early induction of the orphan nuclear receptor NOR-1 during cell death of the human breast cancer cell line MCF-7. Mol. Cell. Endocrinol., 162, 151–156 (2000).
• 22) Deutsch AJA, Angerer H, Fuchs TE, Neumeister P. The nuclear orphan receptors NR4A as therapeutic target in cancer therapy. Anticancer. Agents Med. Chem., 12, 1001–1014 (2012).
• 23) Garattini E, Bolis M, Gianni M, Paroni G, Fratelli M, Terao M. Lipid-sensors, enigmatic-orphan and orphan nuclear receptors as therapeutic targets in breast-cancer. Oncotarget, 7, 42661–42682 (2016).
• 24) Lin S-C, Yao C-Y, Hsu C-A, Lin C-T, Calkins MJ, Kuo Y-Y, Tang J-L, Tien H-F, Wu S-J. Functional association of NR4A3 downregulation with impaired differentiation in myeloid leukemogenesis. Ann. Hematol., 101, 2209–2218 (2022).
• 25) Perlmann T, Jansson L. A novel pathway of vitamin A signaling mediated by RXR heterodimerization with NGFI-B and NURR1. Genes Dev., 9, 769–782 (1995).
• 26) Zetterström RH, Solomin L, Mitsiadis T, Olson L, Perlmann T. Retinoid X receptor heterodimerization and development expression distinguish the orphan nuclear receptors NGFI-B, Nurr1, and Nor1. Mol. Endocrinol., 10, 1656–1666 (1996).
• 27) Vivat V, Zechel C, Wurtz JM, Bourguet W, Kagechika H, Umemiya H, Shudo K, Moras D, Gronemeyer H, Chambon P. A mutation mimicking ligand-induced conformational change yields a constitutive RXR that senses allosteric effects in heterodimers. EMBO J., 16, 5697–5709 (1997).
• 28) Maira M, Martens C, Philips A, Drouin J. Heterodimerization between members of the Nur subfamily of orphan nuclear receptors as a novel mechanism for gene activation. Mol. Cell. Biol., 19, 7549–7557 (1999).
• 29) Clark J, Benjamin H, Gill S, Sidhar S, Goodwin G, Crew J, Gusterson BA, Shipley J, Cooper CS. Fusion of the EWS gene to CHN, a member of the steroid/thyroid receptor gene superfamily, in a human myxoid chondrosarcoma. Oncogene, 12, 229–235 (1996).
• 30) Kagaya S, Hashida R, Ohkura N, Tsukada T, Sugita Y, Terakawa M, Tsujimoto G, Katsunuma T, Akasawa A, Matsumoto K, Saito H. NR4A orphan nuclear receptor family in peripheral blood eosinophils from patients with atopic dermatitis and apoptotic eosinophils in vitro. Int. Arch. Allergy Immunol., 137 (Suppl. 1), 35–44 (2005).
• 31) Kagaya S, Ohkura N, Tsukada T, Miyagawa M, Sugita Y, Tsujimoto G, Matsumoto K, Saito H, Hashida R. Prostaglandin A₂ acts as a transactivator for NOR1 (NR4A3) within the nuclear receptor superfamily. Biol. Pharm. Bull., 28, 1603–1607 (2005).
• 32) Hashida R, Ohkura N, Saito H, Tsujimoto G. The NR4A nuclear receptor family in eosinophils. J. Hum. Genet., 52, 13–20 (2007).
• 33) Martínez-González J, Cañes L, Alonso J, Ballester-Servera C, Rodriguez-Sinovas A, Corrales I, Rodríguez C. NR4A3: a key nuclear receptor in vascular biology, cardiovascular remodeling, and beyond. Int. J. Mol. Sci., 22, 11371 (2021).
• 34) Rajan S, Jang Y, Kim C-H, et al. PGE₁ and PGA₁ bind to Nurr1 and activate its transcriptional function. Nat. Chem. Biol., 16, 876–886 (2020).
• 35) Rajan S, Toh HT, Ye H, Wang Z, Basil AH, Parnaik T, Yoo JY, Lim KL, Yoon HS. Prostaglandin A₂ interacts with Nurr1 and ameliorates behavioral deficits in Parkinson’s disease fly model. Neuromolecular Med., 24, 469–478 (2022).
• 36) Flaig R, Greschik H, Peluso-Iltis C, Moras D. Structural basis for the cell-specific activities of the NGFI-B and the Nurr1 ligand-binding domain. J. Biol. Chem., 280, 19250–19258 (2005).
• 37) Lemmen C, Lengauer T. Computational methods for the structural alignment of molecules. J. Comput. Aided Mol. Des., 14, 215–232 (2000).
• 38) Wu G, Vieth M. SDOCKER: a method utilizing existing X-ray structures to improve docking accuracy. J. Med. Chem., 47, 3142–3148 (2004).
• 39) Marialke J, Tietze S, Apostolakis J. Similarity based docking. J. Chem. Inf. Model., 48, 186–196 (2008).
• 40) Fukunishi Y, Nakamura H. A new method for in silico drug screening and similarity search using molecular dynamics maximum volume overlap (MD-MVO) method. J. Mol. Graph. Model., 27, 628–636 (2009).
• 41) Kawabata T, Nakamura H. 3D flexible alignment using 2D maximum common substructure: dependence of prediction accuracy on target-reference chemical similarity. J. Chem. Inf. Model., 54, 1850–1863 (2014).
• 42) Brylinski M, Skolnick J. FINDSITE^LHM: a threading-based approach to ligand homology modeling. PLOS Comput. Biol., 5, e1000405 (2009).
• 43) Dalton JA, Jackson RM. Homology-modeling protein-ligand interactions: allowing for ligand-induced conformational change. J. Mol. Biol., 399, 645–661 (2010).
• 44) Fukuhara N, Go N, Kawabata T. Prediction of interacting proteins from homology-modeling of complex structures using sequence and structure scores. Biophysics (Nagoya-shi), 3, 13–26 (2007).
• 45) Fukuhara N, Kawabata T. HOMCOS: a server to predict interacting protein pairs and interacting sites by homology modeling of complex structures. Nucleic Acids Res., 36 (Web Server), W185–W189 (2008).
• 46) Kawabata T. HOMCOS: an updated server to search and model complex 3D structures. J. Struct. Funct. Genomics, 17, 83–99 (2016).
• 47) Ogawa K, Hashida R, Miyagawa M, Kagaya S, Sugita Y, Matsumoto K, Katsunuma T, Akasawa A, Tsujimoto G, Saito H. Analysis of gene expression in peripheral blood eosinophils from patients with atopic dermatitis and in vitro cytokine-stimulated blood eosinophils. Clin. Exp. Immunol., 131, 436–445 (2003).
• 48) Hashida R, Ogawa K, Miyagawa M, Sugita Y, Takahashi E, Nagasu T, Katsunuma T, Akasawa A, Tsujimoto G, Matsumoto K, Saito H. Analysis of gene expression in peripheral blood eosinophils from patients with atopic dermatitis by differential display. Int. Arch. Allergy Immunol., 131 (Suppl. 1), 26–33 (2003).
• 49) Zetterström RH, Solomin L, Jansson L, Hoffer BJ, Olson L, Perlmann T. Dopamine neuron agenesis in Nurr1-deficient mice. Science, 276, 248–250 (1997).
• 50) Willems S, Marschner JA, Kilu W, Faudone G, Busch R, Duensing-Kropp S, Heering J, Merk D. Nurr1 modulation mediated neuroprotective effects of statins. Adv. Sci. (Weinh.), 9, e2104640 (2022).
• 51) Makishima M, Okamoto AY, Repa JJ, Tu H, Learned RM, Luk A, Hull MV, Lustig KD, Mangelsdorf DJ, Shan B. Identification of a nuclear receptor for bile acids. Science, 284, 1362–1365 (1999).
• 52) Parks DJ, Blanchard SG, Bledsoe RK, Chandra G, Consler TG, Kliewer SA, Stimmel JB, Willson TM, Zavacki AM, Moore DD, Lehmann JM. Bile acids: natural ligands for an orphan nuclear receptor. Science, 284, 1365–1368 (1999).
• 53) Kliewer SA, Lenhard JM, Willson TM, Patel I, Morris DC, Lehmann JM. A prostaglandin J₂ metabolite binds peroxisome proliferator-activated receptor gamma and promotes adipocyte differentiation. Cell, 83, 813–819 (1995).
• 54) Ng CH, Mok SZ, Koh C, Ouyang X, Fivaz ML, Tan E-K, Dawson VL, Dawson TM, Yu F, Lim K-L. Parkin protects against LRRK2 G2019S mutant-induced dopaminergic neurodegeneration in Drosophila. J. Neurosci., 29, 11257–11262 (2009).
• 55) Zhan Y-Y, Chen Y, Zhang Q, et al. The orphan nuclear receptor Nur77 regulates LKB1 location and activates AMPK. Nat. Chem. Biol., 8, 897–904 (2012).
• 56) Wang W-J, Wang Y, Chen H-Z, et al. Orphan nuclear receptor TR3 acts in autophagic cell death via mitochondrial signaling pathway. Nat. Chem. Biol., 10, 133–140 (2014).
• 57) Li L, Liu Y, Chen H-Z, Li F-W, Wu J-F, Zhang H-K, He J-P, Xing Y-Z, Chen Y, Wang W-J, Tian X-Y, Li A-Z, Zhang Q, Huang P-Q, Han J, Lin T, Wu Q. Impeding the interaction between Nur77 and p38 reduces LPS-induced inflammation. Nat. Chem. Biol., 11, 339–346 (2015).
• 58) Wang W-J, Wang Y, Hou P-P, Li F-W, Zhou B, Chen H-Z, Bian X-L, Cai Q-X, Xing Y-Z, He J-P, Zhang H, Huang P-Q, Lin T, Wu Q. Induction of autophagic death in cancer cells by agonizing TR3 and attenuating Akt2 activity. Chem. Biol., 22, 1040–1051 (2015).
• 59) Li J, Guo C, Wu J. 15-Deoxy-∆^12,14-prostaglandin J₂ (15d-PGJ₂), an endogenous ligand of PPAR-γ: function and mechanism. PPAR Res., 2019, 7242030 (2019).
• 60) Agostini M, Schoenmakers E, Beig J, Fairall L, Szatmari I, Rajanayagam O, Muskett F, Adams C, Marais AD, O’Rahilly S, Semple RK, Nagy L, Majithia AR, Schwabe J, Blom DJ, Murphy R, Chatterjee K, Savage DB. A pharmacogenetic approach to the treatment of patients with PPARG mutations. Diabetes, 67, 1086–1092 (2018).
• 61) Wang Z, Benoit G, Liu J, Prasad S, Aarnisalo P, Liu X, Xu H, Walker NP, Perlmann T. Structure and function of Nurr1 identifies a class of ligand-independent nuclear receptors. Nature, 423, 555–560 (2003).
• 62) Bruning JM, Wang Y, Oltrabella F, Tian B, Kholodar SA, Liu H, Bhattacharya P, Guo S, Holton JM, Fletterick RJ, Jacobson MP, England PM. Covalent modification and regulation of the nuclear receptor Nurr1 by a dopamine metabolite. Cell Chem. Biol., 26, 674–685 (2019).
• 63) Laskowski RA, Swindells MB. LigPlot+: multiple ligand-protein interaction diagrams for drug discovery. J. Chem. Inf. Model., 51, 2778–2786 (2011).
• 64) Thompson JD, Higgins DG, Gibson TJ. CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice. Nucleic Acids Res., 22, 4673–4680 (1994).
• 65) Waku T, Shiraki T, Oyama T, Fujimoto Y, Maebara K, Kamiya N, Jingami H, Morikawa K. Structural insight into PPARgamma activation through covalent modification with endogenous fatty acids. J. Mol. Biol., 385, 188–199 (2009).
• 66) Bateman A, Martin M-J, Orchard S, et al. UniProt: the universal protein knowledgebase in 2023. Nucleic Acids Res., 51 (D1), D523–D531 (2023).
• 67) Kanehisa M, Furumichi M, Sato Y, Kawashima M, Ishiguro-Watanabe M. KEGG for taxonomy-based analysis of pathways and genomes. Nucleic Acids Res., 51 (D1), D587–D592 (2023).
• 68) O’Boyle NM, Banck M, James CA, Morley C, Vandermeersch T, Hutchison GR. Open Babel: an open chemical toolbox. J. Cheminform., 3, 33 (2011).
• 69) Šali A, Blundell TL. Comparative protein modelling by satisfaction of spatial restraints. J. Mol. Biol., 234, 779–815 (1993).
• 70) Kawabata T. Build-up algorithm for atomic correspondence between chemical structures. J. Chem. Inf. Model., 51, 1775–1787 (2011).
• 71) Case DA, Aktulga HM, Belfon K, et al. Amber 2021, Univ. California, San Fr. (2021).
• 72) Bekker GJ, Kawabata T, Kurisu G. The biological structure model archive (BSM-Arc): an archive for in silico models and simulations. Biophys. Rev., 12, 371–375 (2020).
• 73) Higdon A, Diers AR, Oh JY, Landar A, Darley-Usmar VM. Cell signalling by reactive lipid species: new concepts and molecular mechanisms. Biochem. J., 442, 453–464 (2012).
• 74) Vinayavekhin N, Saghatelian A. Discovery of a protein-metabolite interaction between unsaturated fatty acids and the nuclear receptor Nur77 using a metabolomics approach. J. Am. Chem. Soc., 133, 17168–17171 (2011).
• 75) de Vera IMS, Giri PK, Munoz-Tello P, Brust R, Fuhrmann J, Matta-Camacho R, Shang J, Campbell S, Wilson HD, Granados J, Gardner WJ Jr, Creamer TP, Solt LA, Kojetin DJ. Identification of a binding site for unsaturated fatty acids in the orphan nuclear receptor Nurr1. ACS Chem. Biol., 11, 1795–1799 (2016).
• 76) de Vera IMS, Munoz-Tello P, Zheng J, Dharmarajan V, Marciano DP, Matta-Camacho E, Giri PK, Shang J, Hughes TS, Rance M, Griffin PR, Kojetin DJ. Defining a canonical ligand-binding pocket in the orphan nuclear receptor Nurr1. Structure, 27, 66–77 (2019).
• 77) Aarnisalo P, Kim CH, Lee JW, Perlmann T. Defining requirements for heterodimerization between the retinoid X receptor and the orphan nuclear receptor Nurr1. J. Biol. Chem., 277, 35118–35123 (2002).
• 78) Zhao M, Wang N, Guo Y, Li J, Yin Y, Dong Y, Zhu J, Peng C, Xu T, Liu J. Integrative analysis reveals structural basis for transcription activation of Nurr1 and Nurr1-RXRα heterodimer. Proc. Natl. Acad. Sci. U.S.A., 119, e2206737119 (2022).
• 79) Jang Y, Kim W, Leblanc P, Kim C-H, Kim K-S. Potent synthetic and endogenous ligands for the adopted orphan nuclear receptor Nurr1. Exp. Mol. Med., 53, 19–29 (2021).
• 80) Sotriffer C. Docking of covalent ligands: challenges and approaches. Mol. Inform., 37, e1800062 (2018).
• 81) Pettersen EF, Goddard TD, Huang CC, Couch GS, Greenblatt DM, Meng EC, Ferrin TE. UCSF Chimera-a visualization system for exploratory research and analysis. J. Comput. Chem., 25, 1605–1612 (2004).
• 82) Kawabata T, Nishikawa K. Protein structure comparison using the Markov transition model of evolution. Proteins, 41, 108–122 (2000).
• 83) Castro DS, Arvidsson M, Bondesson Bolin M, Perlmann T. Activity of the Nurr1 carboxyl-terminal domain depends on cell type and integrity of the activation function 2. J. Biol. Chem., 274, 37483–37490 (1999).