|Home | About | Journals | Submit | Contact Us | Français|
Kappa-opioid (KOP) receptor agonists exhibit analgesic effects without activating reward pathways. In the search for non-addictive opioid therapeutics and novel chemical tools to study physiological functions regulated by the KOP receptor, we screened in silico its recently released inactive crystal structure. A selective novel KOP receptor agonist emerged as a notable result, and is proposed as a new chemotype for the study of the KOP receptor in the etiology of drug addiction, depression, and/or pain.
Opioids remain the most widely prescribed and abused class of medicines.1, 2 Addiction is not the only limiting factor for the effective use of these compounds as powerful painkillers, antitussives, antidepressants, or antipruritic agents. In addition to social and legal issues associated with their use for non-medical, recreational purposes, several adverse effects (e.g., dysphoria, constipation, respiratory depression, nausea, vomiting, etc.)3 hinder their clinical usefulness and justify the enormous effort put forth by numerous investigators over the years to discover safer opioid therapeutics and/or non-addictive medications. Notwithstanding the continued development of many compounds with opioid activity, ranging from useful agents in the clinic to important chemical tools to study the endogenous opioid system, a safe, non-addictive and effective opioid drug is yet to be discovered.3
Notable members of the superfamily of G protein-coupled receptors (GPCRs), mu-, delta-, and kappa-opioid (or MOP, DOP, and KOP) receptor subtypes4–7 are natural targets for the majority of opioid ligands. The most clinically used opioid drugs act as agonists at the MOP receptor,3 and exert addiction liability through activity at this receptor.8, 9 Thus, it has been proposed that high-affinity selective ligands of the DOP and KOP receptors would provide more effective routes to discovering non-addictive analgesics.10, 11 In particular, KOP receptor agonists have been shown to be unable to activate the reward pathway12 while still acting as effective pain suppressors on the central nervous system (CNS) and/or the periphery,11 most likely through the Gi/o protein-mediated inhibition of cAMP production,13 the blockade of calcium channels,14 and/or the activation of the inward rectifier potassium channels.15 Unfortunately, the KOP receptor agonists developed to date are not ideal drugs as they exert side effects such as dysphoria.16 However, KOP receptor-mediated dysphoric effects have recently been attributed to the activation of the p38 MAPK pathway following arrestin recruitment to the activated KOP receptor.16–18 Therefore, KOP receptor-selective G-protein biased agonists, which do not recruit arrestin, have been proposed to be more effective analgesics, without the adverse effects triggered by the arrestin pathway.18 We have recently reported on such a functionally selective G protein-biased KOP receptor ligand:19 6’-guanidinonaltrindole (6’-GNTI). Although this morphine-derivative ligand is a promising lead compound for non-addictive analgesics acting at the KOP receptor with reduced liability for dysphoria, its effective use as a drug is severely limited by its physicochemical properties and its inability to cross the blood brain barrier.
The lack of a detailed molecular-level understanding of the interactions between opioids and their receptors has hindered successful receptor-based drug design. By revealing how opioid ligands bind to their receptors, recent high-resolution crystal structures of all 4 opioid receptor subtypes, i.e., the MOP,20 DOP,21 KOP,22 and nociceptin/orphanin FQ23 receptors, offer an unprecedented opportunity to discover novel chemotypes targeting these proteins that might eventually be developed into more efficacious therapeutics.24, 25
In the search for these non-addictive therapeutics targeting opioid-receptors, we screened in silico over 4.5 million commercially available, ‘lead-like’ small molecules accessible in ready-to-dock three-dimensional format in the ZINC database,26 based on complementarity with the crystallographic binding mode of JDTic into the KOP receptor crystal structure.22 To the best of our knowledge, this is the first virtual screening study at the KOP receptor using its recently released crystal structure.22 The study led to the identification of 4 novel small-molecule chemotypes, out of twenty-two tested molecules, acting at the KOP receptor. The S-stereoisomer of one of these compounds was further characterized as a novel selective KOP receptor ligand with agonistic activity at this receptor, and as such it represents a promising candidate for structure-based drug design.
We used the ‘lead-like’ subset version of the ZINC database26 that was accessible online on February, 2nd, 2012 when the molecular docking study was performed. This subset version contained ~4.5 million commercially available small-molecules selected using the filtering criteria specified on the ZINC database website. Molecular docking at chain A of the recently released inactive KOP receptor structure (PDB ID: 4DJH22), following removal of all non-protein atoms, was performed with DOCK3.6.27–30 The atom positions of the JDTic crystallographic ligand within the KOP receptor binding pocket were replaced by forty-five spheres that had been labeled for chemical matching based on the local protein environment. Default parameters, i.e., a bin size of 0.2 Å, a bin size overlap of 0.1 Å, and a distance tolerance of 1.2 Å for both the binding site matching spheres and each docked small-molecule from the ‘lead-like’ subset, were used for ligand conformational sampling. Partial charges from the united atom AMBER force field31 were used for all receptor residues with the exception of Asp138 in transmembrane helix 3. The dipole moment of this residue was increased by 0.4 per polar atom to favor identification of small molecules that would form ionic interactions with this residue.32 The KOP receptor was kept rigid while each small molecule was docked into the binding pocket in an average of 3,073 orientations relative to the receptor, and an average of 2,132 conformations for each orientation. A score corresponding to the sum of the receptor-ligand electrostatic and van der Waals interaction energies, corrected for ligand desolvation, was assigned to each docked molecule and configuration within the KOP receptor binding pocket. The specific energy estimates were obtained as we recently described for an analogous study.33 The best scoring conformation of each docked molecule was further subjected to 100 steps of energy minimization with the protein residues kept rigid. Twenty-two compounds, termed here MCKK-1–22, and listed in Table S1, were selected from visual inspection of the 500 top-scoring docked compounds (Table S2) based on criteria discussed in Results and Discussion. Similarity between these molecules and the 9,934 opioid receptor ligands that are annotated in the ChEMBL database [https://www.ebi.ac.uk/chembl/] (Table S1) was quantified using an in-house script in R language that calculates Tanimoto coefficients (Tc) to the nearest neighbors based on extended connectivity fingerprint maximum distances 4 (ECFP4) and the protocol we recently reported.33 Tc values range from 0 to 1, with the 0 value indicating maximally dissimilar compounds and 1 indicating maximally similar ones.34 As suggested in the literature,35 molecules are considered reasonably similar if their Tc value is above 0.40. Molecules for testing were purchased from commercial vendors. Specifically, compounds MCKK-1, MCKK-4, MCKK-8, MCKK-15, MCKK-18 and MCKK-21–22 were obtained from ChemBridge, MCKK-2 from the National Cancer Institute of the National Institutes of Health, MCKK-3 from Labotest, MCKK-5–7, MCKK-9–13, MCKK-16–17 and MCKK-19 from Enamine, MCKK-14 from Florida Heterocyclic Compounds, and MCKK-20 from Molecular Diversity Preservation International.
The cDNAs for human KOP (hKOP) receptor and the G protein GαoB were obtained from the Missouri S&T cDNA Resource Center. For arrestin recruitment experiments, full-length Renilla luciferase 8 (RLuc8, provided by S. Gambhir) was fused in-frame to the C terminus of the hKOP receptor in the pcDNA3.1 vector. The following human G protein constructs were provided by C. Gales:36, 37 GαoB with RLuc8 inserted at position 91 (GαoB-RLuc8); untagged Gβ1 (β1); untagged Gγ2 (γ2). The human γ2 subunit was fused to full-length mVenus at its N terminus (mVenus-γ2), and we used the fusion construct human arrestin3-mVenus previously described.38 All constructs were confirmed by sequencing analysis. A total of 20 µg of plasmid cDNA (e.g. 0.2 µg of hKOR-RLuc8, 15 µg of arrestin3-mVenus, and 4.8 µg of pcDNA3.1) was transfected into HEK-293T cells using polyethylenimine (Polysciences Inc.) in a 1:3 ratio in 10-cm dishes. Cells were maintained in culture with DMEM supplemented with 10% FBS. The transfected ratio among receptor, Gα, β1, and γ2, or arrestin was optimized by testing various ratios of plasmids encoding the different sensors. Experiments were performed 48 h after transfection.
Two days after transfection with human KOP receptor and GαoB, HEK293T cells were lysed and membranes were prepared in HEPES buffer (NaCl 140 mM, KCl 5.4 mM, HEPES 25 mM, EDTA 1 mM, MgCl2 2 mM, BSA 0.006%, pH 7.4) using a Polytron homogenizer. Membranes were incubated with 3H-diprenorphine (0.3 nM) (PerkinElmer) at room temperature for 1 h in a final volume of 1 ml, in the absence or presence of various concentrations of each small-molecule selected from the virtual screening. Membranes were then harvested using a Brandel cell harvester through a Whatman FPD-24 934AH glass-fiber filter and washed three times with ice-cold wash Buffer (Tris-HCl 10 mM, NaCl 120 mM, pH 7.4). Non-specific binding was determined using 400 nM of NorBNI.
BRET was performed as described.39 Briefly, two days after transfection, cells were harvested, washed, and re-suspended in a phosphate-buffered saline (PBS) solution. Approximately 200,000 cells/well were distributed in 96-well plates, and 5 µM coelenterazine H (luciferase substrate) was added to each well. Five minutes after the addition of coelenterazine H, ligands were added to each well. After 2 min for G protein activation or 5 min for arrestin recruitment, the BRET signal was determined by quantifying and calculating the ratio of the light emitted by mVenus, the energy acceptor (510–540 nm), over that emitted by RLuc8, the energy donor (485 nm). The drug-induced BRET signal was normalized, taking the Emax of the ethylketocyclazocine (EKC)-induced response as 100%. To measure cAMP accumulation, we used a BRET-based cAMP in a previously described YFP-Epac-RLuc (CAMYEL) assay.40 GαoB, β1, and γ2 were co-expressed to enhance the signal-to-noise ratio, and the cells were treated for 5 min with 100 µM forskolin prior to stimulation.40 The data were normalized and represented as the percentage of forskolin-stimulated cAMP accumulation with 0 defined as the maximal inhibition elicited by EKC.
All reagents purchased from chemical suppliers were used without further purification and reactions monitored using Thin-layer chromatography (TLC) on 0.25 mm Analtech GHLF silica gel plates using EtOAc/n-hexanes and visualized at 254 nm. Column chromatography was performed on silica gel (40 – 63 µm particle size, 230–400 mesh) from Sorbent Technologies (Atlanta, GA). NMR spectra were recorded on either a Bruker DRX-400 with a H/C/P/F QNP gradient probe or a Bruker Avance AV-III 500 with a dual carbon/proton cryoprobe using δ values in ppm as standardized from tetramethylsilane (TMS) and J (Hz) assignments for 1H resonance coupling and 13C fluorine coupling. High resolution mass spectrometry data were collected on a LCT Premier (Waters Corp.) time-of-flight mass spectrometer. Analytical HPLC was performed on an Agilent 1100 Series Capillary HPLC system with diode array detection at 254.8 nm on a CRIRALCEL OD-H column (4.6 × 150 mm), Daicel Chemical Industries, Ltd. using isocratic elution in 97% hexanes and 3% 2-Propanol at a flowrate of 1.25 mL/min. General procedures for the synthesis of tertbutyl 2-(thiazol-2-ylcarbamoyl)pyrrolidine-1-carboxylate (2S and 2R), as well as 1-(2-(3-fluorophenylamino)-2-oxoethyl)-N-(thiazol-2-yl)pyrrolidine-2-carboxamide (MCKK-17S and MCKK-17R) are provided in Supporting Information.
We screened in silico 4,554,059 commercially-available, “lead-like” compounds from the ZINC database26 based on complementarity with the crystallographic binding mode of JDTic into the KOP receptor binding pocket. The 500 top-scoring docking hits (Table S2; 0.01% of the docked library) were visually inspected and prioritized based on features that an automatic molecular docking screen does not take into account. Specifically, molecules were selected based on the following criteria: a) chemotype diversity; b) the presence of polar interactions between the ligand and the Asp138 residue; c) interactions with KOP receptor residues in the binding pocket that are different in DOP and MOP receptors, d) limited flexibility, e) different binding modes from classical alkaloids as revealed by DOP21 and MOP20 receptor crystal structures, and f) purchasability, i.e., molecules were readily available for purchase. Based on these criteria, twenty-two small molecules were purchased from the set of 500 highest-scored compounds. These molecules, labeled MCKK-1–22 in Table S1, corresponded to the DOCK scoring ranks 1, 80, 87, 97, 111, 127, 137, 210, 253, 269, 276, 346, 347, 360, 379, 402, 403, 404, 411, 427, 452, and 472, respectively. As shown in Table S1, these compounds were found to be significantly different from annotated opioid receptor ligands in the ChEMBL database, as indicated by small ECFP4-based Tc values.35 These data confirm the chemotype novelty of all selected agents.
The primary experimental testing of MCKK-1–22 consisted of performing a competitive inhibition binding assay at the hKOP receptor. Membranes of HEK293T cells transfected with the hKOP receptor and GαoB were prepared and incubated with 3H-diprenorphine (0.3 nM) in the absence or presence of 10 or 100 µM of each small-molecule from the virtual screening. Four molecules, MCKK-4, MCKK-5, MCKK-13 and MCKK-17 partially but significantly inhibited 3H-diprenorphine binding (Figure S1) at 100 µM and their properties were therefore further investigated.
To assess whether any of these molecules had agonistic activity, we used a BRET-based G protein activation assay where the hKOP receptor was co-expressed in HEK293T cells with GαoB-RLuc8, β1, and mVenus-γ2, as discussed in Methods. The drug-induced BRET signal is interpreted as a dissociation of and/or conformational change within the Gαβγ complex, and thus, as the activation of the co-expressed G protein. Among the selected molecules, only the racemic mixture MCKK-17R/S activated GαoB with a potency of 8.3 ± 4.0 µM (Figure 1A). Thus, we proceeded to the chemical synthesis of the R and S stereoisomers of MCKK-17 using commercially available N-(tert-butoxycarbonyl)-L-proline (1R) or N-(tert-butoxycarbonyl)-D-proline (1S) (Figure S2) to identify the active molecule. The appropriate proline was coupled to 2-aminothiazole using 1,1’-carbonyldiimidazole (CDI) in CH2Cl2 under anhydrous conditions41, 42 to afford the corresponding Boc-protected thiazoles (2R and 2S). Removal of the Boc group under acidic conditions followed by alkylation with 2-chloro-N-(3-fluorophenyl)acetamide43, 44 under basic conditions in DMF overnight at 80 °C gave stereoisomers MCKK-17S and MCKK-17R. The purity of MCKK-17S and MCKK-17R was determined to be at least 99% by integration of the UV trace from chiral HPLC (data not shown). MCKK-17S resulted in the most active stereoisomer at the hKOP receptor. Indeed, MCKK-17S displayed full agonism relative to EKC at the hKOP receptor with an EC50 of 7.2 ± 3.8 µM, whereas MCKK-17R displayed a potency of only 120 ± 9 µM. None of the other molecules, MCKK-4, MCKK-5 and MCKK-13, significantly activated GoB (Figure 1A), suggesting that, in contrast to MCKK-17R/S, those molecules are antagonists at the hKOP receptor.
To assess the degree of selectivity of MCKK-4, MCKK-5, MCKK-13, MCKK-17R or MCKK-17S for the hKOP over the DOP and MOP receptors, we performed competitive inhibition of 3H-diprenorphine (0.3 nM) binding in the absence or presence of various concentrations of each molecule, and determined their Ki values for each receptor (Table 1). EKC was used as a reference at all three receptors with recorded Ki values of 20 ± 6 nM at hKOP receptor, 250 ± 31 nM at the DOP receptor, and 170 ± 100 nM at the MOP receptor. Consistent with the first experimental testing (Figure S1), the aforementioned five molecules displayed a relatively weak affinity (Ki values) at the hKOP receptor, between 100 and 500 µM (Table 1). Among them, only MCKK-17S exhibited selectivity for the hKOP receptor with a measured affinity of 120 ± 38 µM at this receptor, and no detectable affinity at DOP and MOP receptors (>1000 µM).
To further assess the selectivity of MCKK-17S for the hKOP receptor (Table 1), we also investigated whether the racemic mixture MCKK-17R/S, as well as the two stereoisomers, displayed agonistic activity at the DOP and MOP receptors (Figure 1B). Neither the racemic nor the MCKK-17 stereoisomers displayed significant activity at the MOP receptor, and they only weakly activated the DOP receptor (> 1000 µM). Notably, MCKK-17R, the less active steroisomer at the hKOP receptor, was the most active enantiomer at the DOP receptor. In contrast, MCKK-17S did not significantly activate the DOP receptor, confirming its selectivity at the hKOP receptor. These results were confirmed using a BRET-based cAMP accumulation inhibition assay (Figure 1C) to monitor the agonistic activity of the selected molecules at the hKOP receptor (Figure 1C), as well as at the DOP and MOP receptors (Figure S3).
Finally, we investigated whether MCKK-17 and the corresponding stereoisomers could recruit arrestin. MCKK-17R/S and MCKK-17S recruited arrestin3, with potencies of 160 ± 38 µM and 120 ± 47 µM, respectively, whereas MCKK-17R only weakly recruited arrestin3 at the highest concentration (potency > 1000 µM) (Figure 1D), consistent with its weaker potency at the hKOP receptor for G protein activation.
Although several non-peptidic selective molecules targeting the KOP receptor have been developed, we are still far from a therapeutically effective KOP receptor drug. Our structure-based virtual screening and compound selection criteria yielded the discovery of a novel small-molecule chemotype that acts as a selective, full agonist at the KOP receptor. To the best of our knowledge, this is the first time that a virtual screening based on an antagonist-bound GPCR crystal structure has identified an agonist. We were pleased to note that the chemical scaffold of the identified hit refers to a selective KOP receptor compound that has very little similarity with all opioid receptor agonists or antagonists annotated in ChEMBL, and that it has never been reported to be an opioid receptor ligand. In summary, MCKK-17S is a promising new lead compound for structure-based ligand optimization aimed at discovering potent non-addictive analgesics. Although the parent compound MCKK-17S is not biased toward G protein activation over arrestin, the chemical scaffold is well suited to structure-guided modifications, raising the prospects of maintaining selectivity while increasing potency and building G protein bias.
We thank Dr. Raymond C. Stevens for sharing with us the atomic coordinates of the KOP receptor crystal structure as well as Drs. Celine Gales and Sam Gambhir for providing cDNA constructs, and Dr. James Woods for providing EKC. This work was supported, in whole or in part, by National Institutes of Health grants DA026434 and DA034049 (to M.F.), DA022413 and MH054137 (to J.A.J.), and DA018151 (to T.E.P.). MF is also grateful to CTSA grant UL1TR000067 awarded to the Icahn School of Medicine at Mount Sinai. Computations were run on resources available through the Scientific Computing Facility of Mount Sinai School of Medicine.
The general procedures for the synthesis of MCKK-17S and MCKK-17R are provided. Table S1 provides a list of the twenty-two tested compounds. Table S2 reports details of the 500 top-scoring docked compounds from virtual screening at the KOP receptor. Figure S1 shows the plot of competitive inhibition of 3H-diprenorphine binding at the KOP receptor. Figure S2 shows the synthetic scheme used to obtain MCKK-17R and MCKK-17S stereoisomers. Figure S3 shows the cAMP accumulation inhibition curves at DOP and MOP receptors. This information is available free of charge via the Internet at http://pubs.acs.org.