Mapping the molecular architecture required for lipid-binding pockets using a subset of established and orphan GPCRs

Abstract

G-Protein Coupled Receptors (GPCRs) sense a wide variety of stimuli including lipids and transduce signals to the intracellular environment to exert various physiological responses. However, the structural features of GPCRs responsible for detecting and triggering responses to distinct lipid ligands have only recently begun to be revealed. 14,15-epoxyeicosatrienoic acid (14,15-EET) is one such lipid mediator that plays an essential role in the vascular system, displaying both vasodilatory and anti-inflammatory properties. We recently reported multiple low-affinity 14,15-EET-binding GPCRs, but the mechanism by which these receptors sense 14,15-EET remained unclear. Here, we have taken a combined computational and experimental approach to identify and confirm critical residues and properties within the lipid-binding pocket. Further, we generated mutants to engineer selected GPCR’s predicted binding sites to either confer or abolish 14,15-EET-induced signaling. Our structure-function analyses indicate that hydrophobic and positively charged residues of the receptor-binding pocket are prerequisites for recognizing lipid ligands like 14,15-EET and possibly other eicosanoids.

Graphical Abstract

Introduction

Bioactive lipids play important physiological roles¹ where they act in both inter- and intracellular signaling events^2–5. Consistent with the importance of these pathways, impaired lipid signaling is linked to several conditions including pulmonary fibrosis⁶, cancer⁷, inflammation⁸, neurological⁹ and cardiovascular¹⁰ diseases and apoptosis¹¹. A number of lipid molecules such as eicosanoids^12–21, lysophosphatidic acids^{22, 23} and sphingosine²⁴ signal through GPCRs²⁵ as summarized in Figure 1. The functional significance of specific receptor and lipid-ligand binding has been explored using hallmark experimental techniques^{15, 26–32} such as mutagenesis, ligand binding, and protein crystallography.

Figure 1:

Schematic representation of biosynthetic pathway of membrane derived lipid molecules and known GPCR receptors.

Arachidonic acid metabolite 14,15-epoxyeicosatrienolic acid (14,15-EET) is a cytochrome P450 epoxygenase-derived endogenous lipid molecule with potent vasodilatory^{33, 34} and anti-inflammatory³⁵ properties. Several lipid molecules are known to exert their response via GPCRs³⁶. Yang et al.³⁷ reported 14,15-EET binding to a 48kD membrane protein from coronary arterial membranes and later described receptor-like membrane binding of 14,15-EET in U937³⁸, collectively suggesting GPCR binding. Our previous β-arrestin recruitment screening effort led to the identification of several low-affinity 14,15-EET receptors including Gs-coupled PTGER₂³⁹. Here, we set out to identify 14,15-EET-senstive receptors by screening a library of 111 GPCRs by measuring ERK phosphorylation levels in response to 1 μM 14,15-EET treatment. This led to the identification of several receptors stimulated by 14,15-EET, including the receptors known to bind nucleotides, chemokines, and prostaglandins.

In the current study, we take a comprehensive approach to further understand the common binding mechanism for 14,15-EET in this diverse set of receptors. We used molecular modeling and site-directed mutagenesis to explain their shared lipid-sensing property. In the absence of experimentally determined structural information for the receptors, we took advantage of homology modeling and docking to predict 14,15-EET binding poses that allowed us to assess the residue and chemical properties underlying receptor-lipid interactions. We here define a common chemical signature that is critical to recognize the lipid molecules in various classes of GPCRs which includes a positive charge residue and extended hydrophobic volumes that are required to sense lipid ligands such as 14,15-EET.

We further tested our hypothesis using distinct GPCRs with diverse canonical ligands, the Prostaglandin E₂ receptor 4 (PTGER₄) and the C-C chemokine type 1 and type 3 receptors (CCR₁ and CCR₃). 14,15-EET binding activates both PTGER₄ and CCR₃ and we took parallel routes to confirm the critical nature of these features in each receptor. In the case of PTGER₄, we engineered mutants that eliminate 14,15-EET binding and took a gain-of-function approach for the 14,15-EET-insenstive CCR₁, leveraging the fact that its closest homolog CCR₃ responds to 14,15-EET. The sequence/structural homology between CCR₃ and CCR₁ allowed us to design activating mutants that confer 14,15-EET responsiveness to CCR₁. In each case, the effect of these mutations on receptor responses provide a mechanistic understanding of how lipid ligand recognition occurs in the binding site. The findings we present here offers valuable insights into lipid-sensing mechanisms used by a subset of GPCRs for a given ligand. We believe our work also provides a novel analytic strategy to predict cryptic lipid binding pockets within receptors that use lipids as a secondary or alternative ligands, even possibly aiding in identification of lipid ligands for orphan receptors.

2. Materials and methods

In this section we describe the experimental and computational approaches taken to predict and confirm the lipid sensing properties of selected GPCRs.

2.1. GPCR cDNA Constructs.

GPCR cDNAs were obtained from the DNASU Plasmid Repository (The Biodesign Institute, Arizona State University. Tempe, AZ) and the UMR cDNA Resource Center (University of Missouri-Rolla, Rolla, MO), and sub-cloned into pcDNA3.1+ which incorporated a hemagglutinin (HA) epitope tag. Mutant versions of PTGER₄ and CCR₁ based on modeling predictions were generated by gene synthesis including a HA epitope tag (Invitrogen) and sub-cloned into pcDNA3.1+ identically to the wild-type versions. Nucleotide and protein sequences for each mutant are included as Supplemental Figures 1 and 2.

2.2. Cell Culture

HEK-293 cells (American Type Culture Collection, Manassas, VA, USA) were grown in Dulbecco’s modified Eagle’s medium (DMEM) (Invitrogen) containing 10% fetal bovine serum (FBS) and maintained in 5% CO₂ at 37 °C. Cells were thawed from cryopreservation, grown for three passages and then used through passage 12. One day prior to transfection cells were removed by trypsinization and plated into a 12-well plate. Culture media was exchanged for Opti-MEM (Invitrogen) and transfections of each well were performed using 2.5 μg of each GPCR plasmid and 4 μL lipofectamine 2000 reagent (Invitrogen). One hour prior to agonist stimulation, the medium of the transfected cells was replaced with serum-free DMEM. For the initial 14,15-EET screen, cells were treated for 15 min with 1 μM concentration of 14,15-EET in dimethyl sulfoxide (DMSO) (Cayman Chemical Company) or DMSO alone. In the mutant analyses, serum-starved cells were treated for 15 mins with various concentrations of 14-15-EET, RANTES (R and D Systems), PGE2 (Cayman Chemical Company) or vehicle (PBS For RANTES or DMSO for 14,15-ETT and PGE₂).

2.3. Western Blot

Following stimulation, HEK-293 cells were lysed in buffer (50 mM Tris (pH 7.4), 150 mM NaCl, 1% NP-40, 0.25% Na-deoxycholate, 1 mM ethylenediaminetetraacetic acid, protease inhibitor (Roche, Nutley, NJ, USA)). Lysates were denatured in SDS sample buffer (2% SDS, 10% glycerol, 80 mM Tris, pH 6.8,0.15 M β-mercaptoethanol, 0.02% bromphenol blue) and separated using 4–12% SDS–polyacrylamide gels and transferred to polyvinylidene difluoride membranes. The membranes were blocked with 5% nonfat dry milk (NFDM) in TBST (10 mM Tris, pH 7.5,150 mM NaCl, 0.05% Tween 20) for 30 minutes at room temperature and incubated overnight at 4 °C with primary antibodies in 5% NFDM/TBST which were detected using peroxidase-linked anti-rabbit or anti-mouse secondary antibodies and enhanced chemi-luminescence method (ECL-plus, GE Amersham, Lafayette, CO, USA). Membranes were stripped and reprobed for our loading control (beta-actin) using stripping buffer (37.5 mM Tris, pH 6.8, 2% SDS, 1% β-mercaptoethanol) incubations at 50 °C for 60 min followed by washing three times with TBST followed by immunoblotting with 1:2000 mouse anti-actin monoclonal antibody (Millipore, Temecula, CA, USA). In each case, the intensity of immunoblot bands was detected and quantified using the Fluor Chem FC2 Image Analysis System (Alpha Innotech, USA). The following antibodies were used for Western blot detection: Cell Signaling Technology mouse anti-HA (1:1000), rabbit anti-ERK (1:1000), and mouse anti-pERK (1:1000). Millipore mouse anti-beta actin monoclonal antibody (1:2000).

2.4. Homology modeling

The amino acid sequences of the human proteins corresponding to PTGER₄, CCR₃, GPR17, GPR85, CXCR₄, PTGIR, GPR31, PAR₁ and GPR63 were collected from UniProt (http://www.uniprot.org/) database. The template structures used to model the target sequence are outlined in Supplemental Table 1. The 3D structures of templates were collected from Protein Data Bank (PDB) (http://www.rcsb.org/). Targets sequences were aligned to template sequences manually to match the conserved residues found in the transmembrane and extracellular loop-2 (EC-2) regions. Similarity and identity between the aligned regions of sequences were computed using Discovery-studio viewer program (http://www.accelrys.com). During the alignment chimeric portion of the crystal structures were removed, the alignments served as an input to MODELLER 9.14 program⁴⁰. For each receptor, a minimum of 100 protein models were generated, the best protein model was selected based on the DOPE score and visual inspection. Stereochemical quality of each model was determined using a Psi/Phi Ramachandran plot. In the case of CXCR₄ and PAR₁, only missing loop portions were modeled.

2.5. Induced-fit docking and SiteMap calculation.

The 3-D structure of 14,15-EET was built on the crystal structure of Cis-5,8,11,14-eicosatetraenoic acid extracted from PDB structure 1GNJ⁴¹. Similarly, Prostaglandin E2 structure was extracted from PBD (3WFH) and Prostaglandin I₂ (PGI₂) structure was downloaded from PubChem database (5282411). For carboxylic acid portions of the ligands, each were treated as negatively charged carboxylate groups. The receptor structures were prepared using Protein Preparation Wizard tool available in Maestro Suite, and OPLS-2005⁴² force field terms were used for protein preparation and docking. The water molecules and ligands were removed from the receptor if crystal structure was used for docking. The induced fit docking protocol (IFD) was used to predict the binding pose and estimate the binding energy. For each ligand, a maximum of 80 docking poses were generated using the GLIDE standard precision mode and each of these poses were subjected to prime energy minimization to produce an optimized residue/ligand pose. One final round of GLIDE docking was performed on the induced fit conformation to assess the docking score. A rigid ligand-docking protocol within IFD was adopted in cases where no reliable binding poses were predicted by flexible ligand-docking. Visual analysis and docking scoring function were used to select the best ligand pose. The binding site property and binding pose complementarity were determined using the SiteMap tool (Schrödinger, LLC, New York, NY, 2009). The binding complementarity was visually evaluated by how well 14,15-EET fit onto the contour maps. For example, a good fit means the hydrophobic carbons and polar oxygens of the ligand superimposed on to the hydrophobic and h-bond acceptor maps, respectively.

2.6. Phylogenetic tree

The phylogenetic tree was generated based on protein sequences downloaded from the Uni-Prot database (http://www.uniprot.org/). The N-J based tree was generated using ClustalX⁴³ and the dendrogram was visualized using FigTree v1.4.2. (http://tree.bio.ed.ac.uk).

3. Results and Discussion

3.1. 14,15-EET activates multiple receptors.

GPCR activation is transduced through a number of downstream pathways with nearly all leading to activation of MAP kinases including ERK_1/2⁴⁴. Given this common readout of receptor activation, we chose this pathway as an indication of receptor activation to screen a library of 111 receptors for activation by 1 μM 14,15-EET treatment (Supplemental Figure 3), identifying nine receptors demonstrating at least a 50% increase in pERK_1/2 levels (Figure 2A). This group included a number of receptors known to be lipid sensing included GPR31, PTGIR and Prostaglandin E₂ receptor type 4 (PTGER₄); the latter we and others have observed to be responsive to 14,15-EET^39,45. Three orphan receptors (GPR17, GPR63 and GPR85) and, surprisingly, two chemokine receptors CCR₃ (C-C motif) and CXCR₄ (C-X-C motif) were also found to be responsive to 14,15-EET. We hypothesized that this shared property of 14,15-EET sensing offered a unique opportunity to better understand the common structural characteristics necessary to confer lipid ligand engagement by these otherwise divergent receptors that include previously reported lipid binding receptors^{15, 46–49} (Figure 2B). Thus, we sought to develop a theoretical and empirical explanation of the chemical properties shared by receptors that recognize 14,15-EET.

Figure 2 A. Representative Western blot of ERK_1/2 phosphorylation by various GPCRs in response to 14,15-EET treatment. B. Quantification of the response for each receptor. C. Reported lipid ligands that share a similarity with 14,15-EET and their respective GPCRs.

3.2. Receptor-ligand interaction mapping

A computational approach was used to understand the structural basis of the 14,15-EET binding among this divergent set of receptors. At the time of our analysis, crystal structures existed for only two of the selected receptors (PAR₁ and CXCR₄), necessitating predictions of the lipid ligand binding sites that required generation of homology models for the remaining receptors. One hundred models were generated for each receptor and the energetically optimized model was selected based on the DOPE score.

14,15-EET was docked against each receptor using a rigorous induced-fit docking protocol where receptor’s orthosteric pocket residues were set to flex. The ligand could generate conformers during docking to sample various degrees of freedom related to conformation modeling for the flexible docking. The only exception is GPR63 where rigid docking (ligand is not allowed to generate flexible conformers) was performed to ensure binding complementarity. The docking poses were observed within the TM bundle core region and the observed docking scores are summarized in Table 1. The docking score ranges from −9.7 to 7.0 kcal/mol and, interestingly, the results for GPR63 in the rigid docking conditions produced an improved docking score (−11.4 kcal/mol).

Table 1:

Docking Score calculated for 14,15-EET against several receptors.

No.	Protein	Docking score (kcal/mol)	Method
1	PTGER4	−8.4	Flex
2	CCR3	−8.0	Flex
3	GPR17	−9.6	Flex
4	GPR85	−7.0	Flex
5	CXCR4	−7.6	Flex
6	PTGIR	−9.7	Flex
7	GPR31	−7.9	Flex
8	PAR1	−8.3	Flex
9	GPR63	−11.4	Rigid

The predicted pose complementarity was assessed by mapping SiteMap volumes onto the ligand binding-pose. The calculations of binding site property included probing the receptor surface to identify site-points that reflect chemical property map (opposite contour) of the binding-site. As a test case, we calculated SiteMap on the well-established lipid receptor, sphingosine 1-phosphate receptor 1 (SP1R₁) in complex with an antagonist sphingolipid mimic (ML056)²⁶ that possesses a charge distribution similar to that of the endogenous ligand sphingosine-1-phosphate (S₁P), amphipathic in nature with both hydrophobic and polar groups (Figure 3A). Our calculations indicated that ML056 establishes similar interactions as that of S₁P with hydrophobic and charged groups interactions with S₁PR₁ orthosteric pocket (Figure 3B). Similarly, the phosphate group interacts with two positively ionizable lysine residues (Lys-34 & Lys-120) and Glu-121 neutralizes charged amino group and several hydrophobic residues form a pocket that accommodates lipid tail binding (Figure 3C). S₁P mimicking ligands require both the charged groups and hydrophobic group found in ML056. The success of this approach suggested similar interaction analyses could provide valuable insight on the 14-15-EET binding requirements.

Figure 3:

SiteMap analysis on the lipid binding receptor SP1R₁ (PDBID:3V2Y). A. View of orthosteric pocket and SiteMap volumes. B. Sphingolipid mimic (ML056) positioned on the SiteMap volume where yellow, red and blue volumes corresponds to complementary hydrophobic, h-bond acceptor and donor, respectively. C. 2D interaction diagram of ML056 and S1PR1, where hydrophobic (green) residues coordinate lipid tail, Lys-34 and Arg-120 interact with negatively charged phosphate group and Glu-121 with positively charged amino group.

The properties of 14,15-EET indicate that the key features are the negatively charged carboxylate and the hydrophobic nature of the rest of the molecule. When probing the binding pockets of our 14,15-EET-responsive receptors, we find the expected hydrophobic carbons positioned in the hydrophobicity map (yellow) or carboxylate in the hydrogen bond acceptor map (red) (Figure 4A). The docking and SiteMap calculations indicate each of the responsive receptors has the complementary pockets as expected (Figure 4B–J). The map volumes and shapes differ according to the residue distribution in each receptor. In the case of PTGER₄, the hydrophobic volume is closer to the major pocket and EC-2 loop, and the positive charge occupies the minor binding pocket (Figure 4B), but another member of the prostanoid family, PTGIR, has increased hydrophobic and hydrophilic volume (Figure 4G). This property volume may reflect the high-affinity, endogenous ligand binding preference for PGE₂ and PGI₂ receptors.

Figure 4:

Mapping of predicted 14,15-EET binding pose on to the sitemap volumes of receptors responded in pERK_1/2 assay. A. 14-15-EET ligand chemical properties. B-J. The predicted pose of 14,15-EET overlaid on the receptor maps; only hydrophobic and H-bond acceptor properties are shows as this ligand has no H-bond donor moiety.

3.3. Role of positive charge, hydroxyl group and hydrophobic residues.

Within the model receptors, the carboxylate (COO⁻) moiety of 14,15-EET interacts with the positively charged residues of the receptors (e.g. Arg, Lys) possibly through ionic- or hydrogen bond interactions. GPR63 is the one exception where the S247 hydroxyl group donates the hydrogen bond (Figure 4) to the carboxylate oxygen, but this receptor retains the hydrophobic interactions present in the other receptors. Charged residues are known to contribute to protein folding and structural stabilization^50,51 and a growing body of literature implicates positive charges in critical roles for ligand binding^{29, 52, 53}. The lack of the critical positive charge residue suggests that GPR63 may bind to neutral lipid molecules like endocannabinoids. Moreover, the predicted ligand pose indicates that the carboxylate group interacts with at least one hydroxyl group of the residues like tyrosine or threonine (Table 2). Our observations mirror those made based on the S1PR₁²⁶ and GPR40⁵⁴ crystal structures, and several other GPCRs, where the hydroxyl group positioned closer to charged residues. From this analysis, we propose that primary and secondary contacts for the 14,15-EET carboxylate are positive charges and the hydroxyl group can act as anchor to the lipid ligand binding. The epoxy group present in the 14,15^th position can possibly act as a hydrogen bond acceptor. The oxygen in this three-member ring forms hydrogen bond in CCR₃, GPR17, GPR31 and PAR₁; however the interaction is not observed in other GPCRs.

Table 2:

Primary and secondary residues interacting with carboxylate group of 14,15-EET.

Protein	Primary contact	Secondary contact
PTGER4	R316	T76
CCR3	K202	Y114 & Y118
GPR17	R283	Y140
GPR85	R69	Y301
CXCR4	R188	Y190
PTGIR	R279 & R12	-
GPR31	R98	Y272
PAR1	K158	Y162
GPR63	-	S247

The long hydrophobic tail of the lipid ligands interacts with non-polar environment generated by hydrophobic and aromatic residues and since 14,15-EET is composed of saturated and unsaturated carbons, it can adopt several conformations in the binding pocket.

The distribution of non-polar residues in all the receptors that may form a hydrophobic enclosure around the lipid tail are indicated in the SiteMap (Figure 4) and interaction diagram (Figure 5), with the CXCR₄ receptor having smaller hydrophobic volumes relative to other receptors (Figure 4). Interestingly, the hydrophobic volume and shape vary across receptors, possibly corresponding to the endogenous ligand binding complementarity. The flexibility of 14,15-EET may allow it to adopt the conformations within the receptors that mimic the other endogenous lipid ligand interactions and complement the binding-site properties to activate the receptor. We generated our homology model prior to the release of the PTGER₄ experimentally determined structure. Cryo-EM structure demonstrates that PGE₂ interacts with EC-2 loop including a conserved threonine residue that has been reported to be required for ligand binding⁵⁵ and 14,15-EET appears to form a similar conformation also forming an interaction with the EC-2 loop. 14,15-EET modeling predicts the interaction is mainly at the minor-binding pocket, as observed for PGE₂^56,57 and the likely case for PGI₂ binding to its receptor. Whereas in the structurally distinct receptors such as CCR₃, GPR17 and GPR63, the observed orthosteric pocket hydrophobic volumes (Figure 4) may allow 14-15-EET to adopt a linear conformation.

Figure 5:

Interaction pattern of various receptors with 14,15-EET molecule. The blue and green color residues correspond to positive and hydrophobic residues, respectively.

To test the binding-site properties required to accommodate lipid ligands, we designed mutations to abolish the 14,15-EET binding of PTGER₄ and introduced residues in CCR₁ that confer lipid binding. In both cases, mutant forms of the receptors were heterologously expressed in HEK293 cells using transient transfection and tested by 14,15-EET treatment.

3.4. Eliminating the PTGER4 response to 14,15-EET

We proposed that both hydrophobic and polar properties are critical aspects for receptor engagement by 14,15-EET, and a mutant version of PTGER₄ was produced by replacing the positive charge by negative (R316E) and decreasing hydrophobicity by introducing four mutations (T76S, T79E, L99A, I315A) (Figure 6A, B and Supplemental Figure 1). The Sitemap calculation of this mutant construct indicates reduced hydrophobic volume and an increased hydrogen bond donor property. This binding site’s properties negatively complement the predicted binding requirements and is expected to abolish 14,15-EET binding, with the carboxylate interaction with E316 leading to a repulsive charge interaction. Our docking of 14,15-EET ligand against this mutant receptor failed to predict any successful binding poses due to our binding-site perturbation (Figure 6 C & D), leading us to expect that the mutant would either abolish or reduce 14,15-EET activation.

Figure 6.

Design of inactivating mutants to PTGER₄ receptor. A. Top view of predictied 14,14-EET binding pose (green) in PTGER₄ receptor. Residues in close proximity are highlighted in stick model. B. sequence alignment of ligand binding area lining residues in WT and mutant type; asterisk and arrow indicate conserved residue and the residues mutated, respectively. C. & D. PTGER₄-WT and PTGER-Mutant binding Sitemaps superimposed on the predicted 14,15-EET pose. E-F. Western blot analysis of receptor activation in the context of wild-type or mutant ligand binding pockets. Levels of phospho-ERK stimulation following exposure of either wild-type PTGER₄ receptor (E) or mutant (F) to various concentrations of vehicle, PGE2 or 14,15-EET.

As others have previously observed^{14, 58–61}, we found that the wild-type receptor displayed a robust, graduated response to both 1μM and 100 nM PGE₂ while the mutant version exhibited a clear, but diminished response at both concentrations (Figure 6E,F). These results suggest that 14,15-EET and PGE₂ use the same positive charge for receptor engagement, but that other residues are required for 14,15-EET binding of PTGER₄. Recent Cryo-EM structure provides evidence that PGE₂ establishes an interaction with minor-binding pocket residues from TM1, TM2, TM3, EC-2, TM6 and TM7 and that the conserved R-316 makes ionic interaction with carboxylate group. This predicted pose is in line with our experimental finding; 14,15-EET’s flexibility allows it to adopt an alternative binding pose as there are no structural constrains such as the cyclopentane found in PGE₂. Taken together, our ERK activation data suggest that both ligands slightly adopt different binding orientation in the minor binding pocket where R-316 interaction is intact as the mutations eliminate 14,15-EET activation and only somewhat diminish the effect of PGE₂. It is important to note that our studies are based on 14,15-EET predicted interactions with the homology model and we did not undertake docking studies exploring the influence of EC-2 orientation as was recently reported in the PGE₂ bound structure⁶² and therefore we cannot rule out the possibility that additional residues may be involved.

3.5. Conferring 14,15-EET responsivity to a CCR₁ receptor.

Our initial 14,15-EET-response screen found the chemokine receptor CCR₃ to be 14,15-responsive while its closely related paralog CCR₁ was not (Supplemental Figure 3). The overall homology between these receptors (63% identity) presented the possibility that selective substitutions of amino acids in the orthosteric pocket of CCR₁ with corresponding residues found in CCR₃ may transfer 14,15-EET ligand sensitivity to CCR₁, directly testing our predictions as to the critical residues for 14,15-EET detection. CCR₃ and CCR₁ both recognize the small-protein ligand CCL₅ Regulated upon Activation, Normal T Cell Expressed and Presumably Secreted (RANTES)^{63, 64} through binding occurring mainly on the extracellular loops. Our modeling predicts that the 14,15-EET binding site lies in the transmembrane core region. These facts suggest that it should be possible to introduce mutations that enhance 14,15-EET sensitivity without affecting overall receptor conformation or response to its cognate ligands such as CCL5. We accomplished this by introducing 14-point mutations that remodel the CCR₁ binding pocket into a CCR₃-like one with the following amino acid substitutions: P23L, G33M, L37V, K94V, L95R, F187Y, L192V, K196R, Q199H, A200T, L203M, V263S, V283M and Q284L (Figure 7A &E, Supplemental Figure 2). Here, the objective is to reduce ligand strain by moving the carboxylate neutralizing positive charge from 94^th position to 95^th observed in CCR₃. The SiteMaps indicate that these changes result in reduced polar and increased hydrophobic areas that should enhance 14,15-EET binding. Consistent with this result, the docking score for the CCR₃-WT, CCR₁-WT and CCR₁-Mut. are −8.0, −6.7 and −9.6 kcal/mol, respectively. Thus, it is clear from our modeling that CCR₁-Mut. is energetically more favorable compared to CCR₃-WT, indicating improved residue architecture in the presumptive binding pocket.

Figure 7.

Design of activating mutants to CCR₁ receptor. A. Top view of predictied 14,14-EET binding pose (green) in CCR2 receptor. Residues in close proximity are highlighted in stick model. B. Sequence alignment of ligand binding area lining residues in WT and mutant types, asterisk and arrow indicate conserved residue and the residues mutated, respectively. C-E. CCR₃, CCR₁ and CCR₁-mutant binding site maps superimposed on to the predicted 14,15-EET pose, optimized hydrophobic and H-bond acceptor properties are shown in the box. F-G. Western blot analysis of receptor activation in the context of wild-type or mutant ligand binding pockets via increased Erk phosphorylation following exposure of either wild-type or mutant-type receptor. F. Stimulation of wild-type CCR₃ and wild-type or mutant CCR₁ using multiple concentrations of RANTES. G. 14,15-EET response by CCR₃ and wild-type or mutant CCR₁.

We also used the heterologous expression system to test the ability of directed mutagenesis to generate a permissive binding pocket in CCR₁. RANTES is known ligand for both CCR₁ and CCR₃. RANTES treatment of cells expressing either the 14,15-EET-responsive CCR₃, wildtype CCR₁ or its mutant form demonstrated that both CCR₃ and wildtype CCR1 responded with similar pERK profiles while the mutant was only slightly reduced in its response (Figure 7F and 7G). When each construct was tested using 14,15-EET, CCR₃ generated a substantial response and CCR₁ failed to respond (Figure 7F and 7G), corroborating our previous observation. In contrast, mutant CCR₁ produced a response at both concentrations of 14,15-EET that was not distinguishable from the 1 μM response of CCR₃, consistent with the docking score that suggested these mutations should produce strong interactions with 14,15-EET even at lower concentration.

Taken together, these proof-of-concept results indicate that selective replacement of critical residues within the receptor binding pocket can significantly alter eicosanoid recognition. These examples validate the modeling findings that lipid ligands required hydrophobic and polar (preferably positive charge) to coordinate lipid-tail and carboxylate warhead, respectively. Similar approaches can be undertaken to explore more lipid sending receptor or to potentially deorphanize the GPCRs. Our findings indicate that in the absence of structural data, homology modeling can fuel our structural understanding of lipid ligands binding interactions.

4. Conclusions

In summary, we identified nine receptors that have a significant response to 14,15-EET. Receptor sequence/structure and predicted binding pose information allowed us to understand the chemical properties of potential binding sites required to recognize eicosanoid molecules like 14,15-EET. The induced-fit docking simulation predicted salt bridge formation between negatively charged carboxylate (COO⁻) of ligand and positively charged residues such as arginine and lysine as a primary requirement to initiate ligand-binding.

We also find that secondary interactions to the carboxylate group occur with at least one of the hydroxyl groups from tyrosine, serine, or threonine residues. SiteMap analysis reveals all these receptors also have well defined hydrophobic maps that can accommodate 14,15-EET in the orthosteric pocket. Our analyses of this modeling uncovered critical binding features that complement 14,15-EET binding. We verified these findings by designing mutant receptors that either eliminated the PTGER₄ response to 14,15-EET or that became responsive to the eicosanoid ligand in the case of the CCR₁ mutant harboring 14 mutations. Among the receptors analyzed, many have links to lipid sensing, with PTER₄ and PTGIR having cognate ligands that are long-chain fatty acid with fused ring structure ending with a negatively charged carboxylate, prostaglandins PGE₂ and PGI₂, respectively. Similarly, GPR31 was recently reported to bind 12-(S)-HETE molecule¹⁵, a 20-carbon long fatty acid that shares chemical similarity with 14,15-EET and GPR17 has been proposed to bind lipids, nucleotides⁶⁵ and leukotriene molecules such as LTE₄, LTD₄, LTC₄ and LTD₄^{48, 66}. In the case of GPR63, classes of sphingosine molecules have demonstrated efficacy as low-affinity agonists⁴⁷. Thus, our finding of activation by 14,15-EET at 1 μM concentration indicates that similar eicosanoids may be cognate ligands for this receptor. Our results also suggest that a larger population of GPCRs may be able to sense lipid-like molecules that potentially act as a primary or secondary ligand. Such receptors could be identified based on examinations of binding-site properties complementing lipid-ligand binding. The findings generated from our work advance our understanding of cross-talk between lipid ligands (off-targeting) and insights for de-orphanization and predicting cryptic lipid binding sites on known GPCRs.

Abbrevations

GPCRs

G-Protein Coupled Receptors

14,15-EET

14,15-epoxyeicosatrienoic acid

PTGER₄

Prostaglandin E₂ receptor type 4

PTGIR

Prostaglandin I2 Receptor

CXCR₄

C-X-C Motif Chemokine Receptor

CCR1

C-C chemokine type 1 Receptors

CCR3

C-C chemokine type 3 Receptors

GPR31

Orphan G-Protein Coupled Receptor 31

GPR17

Orphan G-Protein Coupled Receptor 17

GPR63

Orphan G-Protein Coupled Receptor 63

GPR85

Orphan G-Protein Coupled Receptor 85

RANTES

Regulated upon Activation, Normal T Cell Expressed and Presumably Secreted