Potent and subtype-selective dopamine D2 receptor biased partial agonists discovered via an Ugi-based approach

Potent and Subtype-Selective Dopamine D2Receptor Biased Partial Agonists Discovered via an Ugi-Based Approach Ana Mallo-Abreu, Irene Reyes-Resina, Jhonny Azuaje, Rafael Franco, Aitor García-Rey, Maria Majellaro, Darío Miranda-Pastoriza, Xerardo García-Mera, Willem Jespers, Hugo Gutiérrez-de-Terán, Gemma Navarro,*and Eddy Sotelo* Cite This: J. Med. Chem. 2021, 64, 8710−8726 Read Online ACCESS Metrics & More Article Recommendations * sıSupporting Information ABSTRACT: Using a previously unexplored, efficient, and versatile multicomponent method, we herein report the rapid generation of novel potent and subtype-selective DRD2biased partial agonists. This strategy exemplifies the search for diverse and previously unexplored moieties for the secondary/allosteric pharmacophore of the common phenyl-piperazine scaffold. The pharmacological characterization of the new compound series led to the identification of several ligands with excellent DRD2affinity and subtype selectivity and remarkable functional selectivity for either the cAMP (22a and 24d) or the β-arrestin (27a and 29c) signaling pathways. These results were further interpreted on the basis of molecular models of these ligands in complex with the recent DRD2 crystal structures, highlighting the critical role of the secondary/allosteric pharmacophore in modulating the functional selectivity profile. ■INTRODUCTION The superfamily of seven transmembrane receptors (7TMR), commonly referred to as G protein-coupled receptors (GPCRs), is the largest target class in the druggable genome. 1 The receptors in this superfamily regulate virtually every aspect of human physiology, and they are sensors of a wide array of extracellular stimuli. 1 As a consequence, GPCRs are the target of more than 30% of all prescription drugs. 2 The synergistic use of innovative experimental and computational approaches in the last decade led to the increasing appreciation of the key role of conformational plasticity on GPCR signaling events (e.g., constitutive activity, inverse agonism, or biased agonism). 3 For a number of GPCRs, the propensity to activate distinct G proteins can elicit diverse responses depending on the cellular environment. 4 However, an even more subtle but intriguing mode of signaling has been attributed to the ability of a receptor to activate signaling pathways independent of G- protein activation. This occurs through the scaffolding of signaling complexes by β-arrestin, a component of the GPCR desensitization and internalization machinery. 5,6 The process by which ligands differentially modulate G-protein-dependent and/or G-protein-independent (β-arrestin) pathways to mediate specific downstream signal transduction routes is a phenomenon known as functional selectivity or biased agonism. 3 The concept of biased agonism has progressively reshaped our understanding of GPCR signaling and shifted the paradigm of GPCR drug discovery. 7,8 However, the molecular mechanisms behind biased signaling remain elusive since the study of the functional contributions of G-protein and βarrestin mediated signaling pathways of endogenous/exogenous ligands still constitute a challenge. 7,9 Biased GPCR ligands can trigger a specific pathway that is responsible for a Received: April 19, 2021 Published: June 10, 2021 Articlepubs.acs.org/jmc © 2021 American Chemical Society 8710 https://doi.org/10.1021/acs.jmedchem.1c00704 J. Med. Chem. 2021, 64, 8710−8726 Downloaded via UNIV DE SANTIAGO DE COMPOSTELA on July 8, 2022 at 11:10:22 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles. given therapeutic effect, while not activating other signaling events eventually implicated in side effects. Such ligands are extremely useful to elucidate the key structural contributors to signal transduction pathways, besides their significant potential to develop therapeutic agents with fewer side effects. 10−12 A paradigmatic case is Oliceridine, a G-protein-biased μ-opioid (MOR) agonist that has shown encouraging results in clinical studies, combining a potent analgesic effect with reduced Figure 1. Selected examples from the literature of DRD2biased ligands. 12,25−38 Journal of Medicinal Chemistry pubs.acs.org/jmc Article https://doi.org/10.1021/acs.jmedchem.1c00704 J. Med. Chem. 2021, 64, 8710−8726 8711 incidence of β-arrestin-mediated adverse effects (e.g., respiratory depression and constipation). 13−15 The dopamine D2receptor (DRD2) is a prototypical GPCR for which exploration of the biased agonism concept is becoming the new paradigm to provide better drugs. 12,16,17 DRD2is the primary target of antipsychotics and antiparkinsonian agents and is also implicated in the mode of action of several drugs associated with abuse and addiction. 18,19 Schizophrenia is characterized by positive, negative, and cognitive symptoms. 20 Classical antipsychotics are effective at targeting the positive symptoms, but they do have adverse extrapyramidal symptoms (EPS). 21 Atypical antipsychotics have overcome some of the problems associated with typical APDs in the clinic, and they are better at targeting the positive symptoms of schizophrenia without inducing EPS. However, these compounds have their own distinct side-effect profile, which includes weight gain, agranulocytosis, and hypotension. 21 Aripiprazole and cariprazine (Figure 1) are prototypes of a new generation of atypical antipsychotics, 22 and they were approved by the FDA for the treatment of schizophrenia, bipolar I manic/mixed episodes, and depressive disorder. 23 From a structural point of view, they are considered bitopic ligands, bearing a canonical primary pharmacophore (arylpiperazine) and a secondary (or allosteric) pharmacophore linked through an spacer group 24 (Figure 1). These drugs changed the view of antipsychotic action on dopamine signaling and introduced for the first time in the treatment of psychosis a clinically relevant mechanism based on DRD2 occupancy without DRD2blockade. They are thought to act as antagonists in the striatum, where excessive dopamine activity is believed to cause positive symptoms, but they do show agonist activity in the mesocortical pathway, where reduced dopamine activity is associated with negative symptoms and cognitive impairment. 25 Inspired by the unique antipsychotic profile of aripiprazole, a novel series of DRD2biased agonists have been developed over the past decade (Figure 1). Inspection of the pharmacological data available for these ligands enabled to identify DRD2 partial agonists eliciting either β-arrestin mediated recruitment and G-protein biased DRD2ligands (Figure 1). A structural analysis reveals their chemical analogy, particularly with the atypical antipsychotics that inspired their design (Figure 1). While effective in retaining the desired biased profiles, the limited structural diversity inherent to their design hindered the exploration of alternative allosteric regions of the receptor (e.g., secondary pocket) adjacent to the primary (canonical) binding site. In other words, this conservative strategy could not deliver structurally novel ligands that are able to stabilize alternative conformational states of receptors. 39 Furthermore, most DRD2biased agonists developed to date do not show remarkable subtype selectivity toward the other receptors of the dopamine D2family (D3and D4,Figure 1). 40 Evidence from clinical practice indicates that most effective antipsychotics exhibit a rather promiscuous receptor profile, with important affinity toward several GPCRs (usually defined as selectively nonselective drugs). 41,42 However, from a chemical biology perspective, the development of molecular probes that simultaneously elicit subtype selectivity and signaling bias profiles is key to determine the molecular and physiological determinants that underpin DRD2biased signaling. Such pharmacological tools would contribute to our understanding of the molecular basis of DRD2signaling not only in transfected cells but also in complex and physiologically relevant environments. This information enables the elucidation of the real contribution of β-arrestin and G-protein signaling in dopaminergic receptors and the development of safer and more effective medications for schizophrenia and Parkinson’s disease. To overcome the limitations of previous approaches, we herein report the design, synthesis, and pharmacological characterization of several series of DRD2partial agonists that exhibit either G-protein or β-arrestin biased signaling profiles, uniquely combined with exquisite subtype selectivity. The new families of compounds were designed and assembled by using a highly versatile multicomponent approach. The experimental data provided structure−activity relationship (SAR) and structure−functional selectivity relationship (SFSR) trends that were consistent with the proposed binding modes, as defined in a receptor-driven docking model. The overall results of the study represent a successful proof-of- concept of an unexplored strategy for the rapid identification of novel structurally diverse and functionally selective DRD2 ligands. ■RESULTS AND DISCUSSIONS Design and Synthesis. From a structural point of view, most studied DRD2biased ligands fit in the pharmacophoric model presented in Figure 2. These compounds are bitopic Figure 2. (A) Representative structure of most DRD2biased ligands. (B) General structure of herein described ligands. Blue: common (primary) pharmacophore in the series. Black: structure of the scaffolds explored in the secondary pharmacophore. Journal of Medicinal Chemistry pubs.acs.org/jmc Article https://doi.org/10.1021/acs.jmedchem.1c00704 J. Med. Chem. 2021, 64, 8710−8726 8712 ligands, containing three well-defined regions: 43 (1) the primary pharmacophore (PP) [commonly referred to as the left-hand side (LHS) or head group], consisting of a mono- or disubstituted phenyl-piperazine scaffold, (2) the central linker, that is usually variable in length and nature (e.g., acyclic or cyclic), and (3) the secondary (or allosteric) pharmacophore (SP) [commonly referred to as the right-hand side (RHS) or tail group], generally consisting of a heterocyclic core. Although the aromatic piperazines of the primary pharmacophore (PP) dictate the efficacy profile and is sufficient to allow binding to the primary (orthosteric) binding site of DRD2 (and to that of the closely related DRD3subtype), enlargement of the chemical structures by addition of a flexible linker and a second, mostly lipophilic system (SP) has been found to promote enhanced affinity and subtype selectivity. 44,45 In the present study, it was decided to maintain the primary pharmacophore (Figure 2), with the 1-(2,3-dichlorophenyl)- piperazine moiety selected (which is present in aripiprazole), and a shorter than usual (four atoms) linear linker, which is present in UNC9995 12,33 (Figure 1). Six previously unexplored secondary pharmacophoric (SP) groups (Figure 2) were proposed to examine the effect of these structural modifications on subtype selectivity (DRD2, DRD3, and DRD4) and also their effect on the DRD2functional selectivity profile of the novel ligands. The selected SP frameworks provide novel topologies, physicochemical features, and alternative binding modes that should enable the capture of diverse conformational states within the receptor. In addition to the heterocyclic and functional diversity introduced, some of the proposed SP fragments bear a stereogenic center within the heterocyclic framework (Scheme 1,compounds27 and 29), thus introducing stereochemical diversity that would enable the future investigation of scarcely explored stereoselective interactions within the SP region. The appropriate decoration of targeted structures (Figure 2), according to the specific requirements of the target receptor, would allow effective interaction with subpockets in the secondary (allosteric) binding site, ultimately resulting in optimized bioactivity levels. Moreover, the achievement of such decorations by synthetically feasible approaches is an emerging criterion for library design. The implementation of concise and efficient synthetic methodologies that reconcile molecular complexity with experimental simplicity, thus allowing rapid access to privileged molecular frameworks, constitutes a highly desirable goal within the competitive environment of drug discovery. A valuable addition to the compendium of preparative methods to generate drug candidate libraries is provided by multicomponent reactions Scheme 1. Ugi-Based Assembly of the Target Compounds Journal of Medicinal Chemistry pubs.acs.org/jmc Article https://doi.org/10.1021/acs.jmedchem.1c00704 J. Med. Chem. 2021, 64, 8710−8726 8713 (MCRs), 46,47 which have emerged as a tailored synthetic paradigm in the context of medicinal chemistry and chemical biology programs. MCRs combine three major principles in organic synthesis: convergence and atom and step economies. In addition, such reactions are highly flexible and their extraordinary exploratory power allows maximum structural complexity to be generated from simple starting materials in just a single step. 48 The main goal of the study reported here was to identify novel DRD2biased agonists by introducing unexplored structural elements in the secondary pharmacophore (SP) region. The selection criteria for the proposed SP groups were guided by the principles of synthetic feasibility and structural diversity (Figure 2). Thus, we envisioned different divergent, highly exploratory, and experimentally simple MCR-assisted pathways (Scheme 1). The selected synthetic approaches, which are based on the Ugi fourcomponent reaction (U-4CR), exploit the potential of this transformation for structural diversification. Thus, starting from the readily available carboxylic acid 16, which contains the primary pharmacophoric moiety and the linker, we envisioned a set of reactions (Scheme 1) in which 16 would be combined with diversely functionalized amine inputs (17), carbonyl partners (18), and three representative isocyanides (19). The simplest set of ligands (20a−c) was obtained by the Ugi reaction of 16 with methylamine (17a), formaldehyde (18a), and isocyanides 19a−cin methanol at room temperature for 48 h. 47 The assembly of the other five subsets (22,24,26,27, and 29) involved the use of polyfunctional reactive substrates and/or the versatility of the Ugi-Deprotect-Cyclize (UDC) strategy. 47,49,50 As shown in Scheme 1, the feasibility of the selected pathways relies heavily on the latent reactivity of the different functionalized Ugi adducts (21,23,25, and 28), which, upon direct cyclization (22,26, and 27) or removal of the protecting group, undergo an intramolecular cyclization (24 and 29) to furnish the target structures in an efficient transformation that takes place in one pot. In this way, compounds 22a−cwere obtained by reaction of 16 with methylamine (17a), isocyanides 19a−c, and phenylglyoxal (18b) as the key precursor (Scheme 1). The superior reactivity of the formyl group in 18b ensured the chemoselectivity of the reaction to produce an Ugi adduct (21a−c) that contains an enolizable ketone group, and this was transformed in situ to give 22a−cby treatment with ammonium acetate at 100 °C. 51 Treatment of the carboxylic acid 16 with formaldehyde (18a), isocyanides 19a−c, and the mono-BOC protected phenylenediamines 17b−cafforded the Ugi adducts 23, which, upon acidic BOC cleavage and thermal treatment, afforded 24a−f Table 1. Structures and Pharmacological Data at the D2Receptor Family for the Ligands a , b a pIC50 and Emax values are the average of five experiments, each performed in duplicate with ±SEM values that are three times lower than the average. Emax relative to the effect of the reference agonist quinpirole. b Tested using the experimental protocols described in the Experimental Section. Journal of Medicinal Chemistry pubs.acs.org/jmc Article https://doi.org/10.1021/acs.jmedchem.1c00704 J. Med. Chem. 2021, 64, 8710−8726 8714 (Scheme 1). 52 The Ugi reaction of 16, formaldehyde (18a), isocyanides 19a−c, and aminoacetaldehyde dimethyl acetal (17d) generated the adducts 25, which were transformed in situ to 26a−cby an acid-mediated transformation that involved an intramolecular cyclization and subsequent elimination (Scheme 1). 53 The four-component reaction of 16, isocyanides 19a−c, and the bifunctional precursors 18c (chloroacetaldehyde) and 17e (N-methylethylenediamine) under basic conditions (NaHCO3) directly afforded the piperazine derivatives 27a− c(Scheme 1). 54 The sequence involves the formation of an Ugi adduct, which, under basic conditions, undergoes an intramolecular nucleophilic substitution reaction. Finally, the assembly of piperazin-2-ones 29a−cwas accomplished by a similar pathway (Scheme 1), starting from carboxylic acid 16 and isocyanides 19a−c, but using two alternative bifunctional precursors [i.e., ethyl glyoxylate (18d) and N-BOC-ethyl- enediamine (17f)]. Acid-mediated cleavage of the BOC group in the Ugi adducts 28 provided the target ligands (29). 55 Ligands of series 27 and 29, which contain a stereocenter within the SP heterocyclic fragment, were isolated and evaluated as racemic mixtures. A detailed description of the synthetic methods and the complete structural, spectroscopic, and analytical data for all compounds are provided in the Experimental Section. The five heterocyclic cores explored as secondary pharmacophoric groups (22,24,26,27, and 29) can be considered as conformationally restricted analogs of the early acyclic series 20, with differences in structure, topology, physicochemical descriptors, and complexity. Thus, the Ugibased diversification strategy enables the rapid differentiation of the structural elements of the acyl-aminoamide scaffold into highly diverse molecular frameworks. Biological Evaluation. The newly synthesized ligands were all initially tested in cAMP inhibition assays with three dopamine receptor subtypes (DRD2, DRD3, and DRD4), i.e., the DRD2-like receptors, to evaluate their functional behavior and selectivity profile (Table 1). All experiments were performed in vitro on transfected HEK-293T cells, with the evaluation of the efficacy (Emax) and half maximum inhibitory concentration (IC50) for the cAMP assays, using previously described experimental protocols. 56 Quinpirole was used as a control and reference drug during these studies. Compounds 27a−cand 29a−cwere tested as racemic mixtures. On the basis of its DRD2potency (pIC50 > 8) and subtype selectivity criteria, seven ligands (20a,20b,22a,27a,27b,29a, and 29c) were selected for further investigation of the DRD2- mediated potency (EC50) and efficacy (Emax) for β-arrestin-2 recruitment (Table 2). As a consequence of its excellent DRD2 potency (pIC50 = 8.66), albeit without selectivity toward DRD4, ligand 24d was also included in the set of compounds selected for bias characterization. The β-arrestin-2 recruitment study involved BRET experiments performed on transfected HEK-293T cells using previously described experimental protocols. 57 Aripiprazole and quinpirole were employed as controls during these studies. Table 2. Ligands and Structure−Selectivity Relationship (SSR) Data for Selected Ligands a , b a EC50,IC 50, and Emax values are the average of five experiments, each performed in duplicate with ±SEM values that are 3-fold less than the average. Emax relative to the effect of the reference agonist quinpirole. b Tested using the experimental protocols described in the Experimental Section. c Bias factors were quantified by the operational model using quinpirole as a positive control (see the Experimental Section). Ligand bias values >0 indicate preference for the cAMP pathway, and values < 0 indicate preference for the β-arrestin signaling pathway. Values above 0.5 are considered significant and are highlighted. *Emax is not shown due to low affinity of the ligand. Journal of Medicinal Chemistry pubs.acs.org/jmc Article https://doi.org/10.1021/acs.jmedchem.1c00704 J. Med. Chem. 2021, 64, 8710−8726 8715 With the aim of better exploring the pharmacological profile of the novel series herein documented it was decided to evaluate a set of ligands (Table 2,20a,20b,22a,24d,27a, 27b,29a, and 29c) in antagonist mode. In this case, cells were pretreated with the selected compounds before treatment with the agonist quinpirole (see Methods). As can be observed in Figure 3A, there were no significant variations in the efficacy between each of the compounds tested with quinpirole compared to the quinpirole tested alone. These results enable to discard a potential antagonistic behavior for the studied compounds. Figure 3B shows a comparative profile of the cAMP dose response curves obtained for ligands 22a and 24d and quinpirole at DRD2, DRD3, and DRD4. Structure−Activity and Structure−Selectivity Relationships. The cAMP functional data for the novel compounds (Table 1) reveal that some ligands behave as DRD2selective partial agonists. Inspection of reported data enables to identify eight novel and highly potent (pIC50 >8) DRD2ligands (e.g., 20a,20b,22a,24d,27a,27b,29a, and 29c), six of which elicit remarkable selectivity (>1000-fold) toward DRD3and DRD4. Furthermore, some potent and selective DRD3(i.e., 24b and 26c, pIC50 = 7.94 and 8.47, respectively) or DRD4(i.e., 22c and 24c, pIC50 = 8.24 and 8.10, respectively) ligands were identified. These data emphasize the potential of herein disclosed MCR-based diversification of the secondary pharmacophore region has in modulating the interaction with DRD2. Additionally, our Figure 3. (A) Emax values for 100 nM quinpirole in cAMP assays performed in HEK-293T cells expressing DRD2, pretreated or not (reference black column) with 100 nM of the selected compounds. Data are normalized (right) to the effect of quinpirole alone (100%). (B) HEK-293T cells expressing human DRD2, DRD3, or DRD4were treated with the indicated compounds. The effect of the compounds on the decrease of 500 nM- induced cAMP levels was determined as described in the Experimental Section. Data are given relative to the value of forskolin alone and then normalized to the effect of quinpirole. Journal of Medicinal Chemistry pubs.acs.org/jmc Article https://doi.org/10.1021/acs.jmedchem.1c00704 J. Med. Chem. 2021, 64, 8710−8726 8716 results exemplify how subtle structural modifications on the secondary pocket can provoke important differentiation in the biological profile of the synthesized ligands. For a more immediate and efficient analysis of the variation of both affinity and selectivity, the pIC50 values at DRD2(X axis) versus DRD3(Yaxis, top panel) and DRD4(Yaxis, bottom panel) are provided as independent scatter plots using the same scale and range for both axes (square plot). Each subset was represented in a different color and shape in order to facilitate a more comprehensive analysis of both potency and selectivity within a series. In both plots, the DRD2selective compounds appear below the diagonal (right bottom zone), with the distance from the diagonal being proportional to the degree of selectivity, confirming that the identified DRD2 partial agonists also show a high degree of selectivity versus DRD3/DRD4. This subset was selected for further pharmacological characterization (see Table 2). The functional data presented in Table 1 highlight the relevance of the amide group in the secondary pharmacophore for effective interaction within DRD2. The only subset that did not provide potent DRD2ligands (26) has this amide group embedded within the heterocyclic core, which means that they lacked the polar hydrogen and had a conformational restraint, while the rest of the series provided at least one ligand with significant DRD2potency. In contrast to the low affinity on the DRD2, series 26 provided compound 26c, a highly potent (pIC50 = 8.47) and selective (>300-fold) novel DRD3partial agonist. Series 20,27, and 29 generally yielded potent and subtypeselective DRD2partial agonists, and these included the most attractive ligands identified in this study (Table 1 and Figure 4). In these series, compounds bearing a benzyl group on the amide moiety (20a,27a, and 29a) systematically exhibited a low nanomolar potency (pIC50 = 8.34, 8.96, and 8.94 respectively). In contrast, the cyclohexyl group seems to be well tolerated only in acyl-aminoamides (20b) and the N- methylpiperazines (27b). Conversely, those compounds that contained a tert-butyl residue generated ligands (20c and 27c) that systematically exhibited micromolar potency, apart from 29c, thus suggesting that this group could not facilitate optimal complementarity within DRD2. Although most ligands with imidazole- or benzimidazole-based SP groups (Table 1, ligands 22 and 24) have low potency at DRD2, the pIC50 values determined for ligands 22a and 24d (pIC50 = 8.30 and 8.66, respectively) reveal that these scaffolds, when appropriately decorated on the exocyclic amide group (i.e., with a benzyl group), can provide potent and selective DRD2partial agonists. As previously discussed, 1-acyl-N-methylpiperazine-2-car- boxamides 27 and 1-acyl-N-methyl-3-oxopiperazine-2-carbox- amides 29 can be considered as conformationally restricted analogs of the acyl-aminoamides 20. Thus, their similar biological profile (potency and selectivity) could be a consequence of the close structural similarity of these three series. Despite the structural analogy, the cyclic constrained analogs (piperazine-2-carboxamides 27 and 3-oxopiperazine-2- carboxamides 29) exhibited slightly superior potency (Table 1) when compared to the acyclic series (20). This trend suggests that the cyclic derivatives are more similar to the bioactive conformation. As observed in the early series, and with the exception of 29c (pIC50 = 8.42), ligands bearing the tert-butyl group in the exocyclic amide afforded the weakest potency (pIC50 = 5.76−6.79). Another interesting structural feature of the conformationally restricted series 27 and 29 is the presence of a stereogenic center at position 2 of the heterocyclic core. Although these compounds were tested as racemates, it is reasonable to expect diverse pharmacological profiles for the different enantiomers. Accordingly, the potential influence of the absolute configuration of the stereogenic center in these series will be explored in future work. DRD2-mediated signaling events are initiated either by G protein-dependent (G-protein-coupled) and/or G proteinindependent pathways (β-arrestin recruitment). The ability of a (partial) agonist to selectively activate one of these specific signaling pathways is a pharmacological phenomenon known as functional bias (or functional selectivity). A key goal of this study was to explore the relationship between the biased selectivity and the structural features of the ligand, which ultimately lead to the establishment of specific interactions with DRD2. Thus, we selected the seven derivatives (20a,20b, 22a,27a,27b,29a, and 29c) that exhibited a high cAMP potency (pIC50 > 8) and optimal DRD2selectivity (Table 1)to perform a β-arrestin-2 recruitment BRET assay in transfected HEK-293T cells, which determines the potency and efficacy Figure 4. Potency−selectivity DRD3−DRD2and DRD4−DRD2plots. Journal of Medicinal Chemistry pubs.acs.org/jmc Article https://doi.org/10.1021/acs.jmedchem.1c00704 J. Med. Chem. 2021, 64, 8710−8726 8717 for β-arrestin-2 recruitment. Although it was not selective (DRD4, pIC50 = 8.01), the benzimidazole derivative 24d was included in this study due to its excellent DRD2potency (pIC50 = 8.66). Aripiprazole, a known biased ligand, was used as a reference ligand and quinpirole 58 (a full agonist of DRD2) was used as a positive control in both cAMP and β-arrestin-2 recruitment BRET assays. The comparative cAMP and βarrestin-2 data are presented in Table 2. In order to identify functional bias rapidly, a bias factor was calculated using the Black and Leffoperational model 59 with respect to quinpirole (see Table 2). Most of the evaluated ligands exhibited excellent efficacy in the β-arrestin recruitment pathway (Emax over quinpirole in the range 68−142%, see Table 2), thus behaving as full agonists for this pathway. The most salient data emerging from β-arrestin recruitment assays evidenced two pairs of ligands that elicit opposite signaling profiles. Thus, while ligands 27a and 29c exhibit a very attractive subnanomolar profile in the β-arrestin recruitment test (pIC50 = 9.67 and 9.43, respectively), derivatives 22a and 24d showed only weak potency (micromolar range). The availability of ligands bearing different groups on the exocyclic amide in series 20,27, and 29 provided evidence of the key role of the alkyl group (benzyl, cyclohexyl, or tert-butyl) on the β-arrestin recruitment potency. Interestingly, the compounds that elicited the poorest β-arrestin recruitment potency (22a and 24d) contain an aromatic heterocyclic core with an N-benzyl group within the secondary pharmacophore. Six of the ligands (20b,22a,24d,27a,29a, and 29c) showed a clear functional selectivity profile (biased agonism) according to the bias factor parameter (Table 2), where a positive value indicates a preference for the cAMP pathway and a negative value denotes that β-arrestin recruitment is dominant. As one would expect, the weak potency in the βarrestin recruitment assay and excellent cAMP data mean that ligands 22a and 24d show a significant bias toward cAMP [ΔΔlog(τ/KA) = 2.223 and ΔΔlog(τ/KA) = 2.768, respectively]. These values represent 167-fold and 586-fold bias, respectively, toward the cAMP pathway. Furthermore, compound 29a also shows a moderate [ΔΔlog(τ/KA)= 0.502] 3-fold bias toward cAMP inhibition. In contrast, ligands 27a and 29c, due to their sub-nanomolar effect and excellent efficacy in the β-arrestin pathway (pEC50 = 9.67 and 9.43, respectively) and its potency and moderate efficacy in the cAMP pathway, showed 10-fold and 11-fold β-arrestin biased agonism. Compound 24d, besides being one of the most potent binders at DRD2and indeed the partial agonist with strongest bias toward the cAMP pathway (Table 2), lacks the required D2/D4selectivity profile (Table 1) to warrant further characterization of this particular compound. In any case, compound 24d was used as a tool to understand the molecular basis of its biased profile. In order to investigate the structural basis for the different biased signaling profiles, a complex of each of the molecules listed in Table 2 with DRD2was generated by different docking approaches, initially using the crystal structure of DRD2in complex with risperidone. 60 Despite the fact that this is an inactive conformation of the receptor, the chemical similarity of the general scaffold of our compounds with the cocrystallized antagonist (risperidone) supported the use of this crystal structure. Moreover, the orthosteric binding site of the aminergic receptor is not expected to change substantially upon complexation with partial agonists. 61 As derived from the binding mode obtained (Figure 5), the 2,3-dichlorophenyl ring on the piperazine scaffold (primary pharmacophore, commonly referred to as LHS) is analogous to the benzisoxazole moiety of risperidone, which uniquely extends into a deep binding pocket defined by the side chains of residues in TM3 (Cys1183.36 and Ile1223.40), TM5 (Ser1975.46 and Phe1985.47), and TM6 (Phe3826.44, Phe3906.52, and Trp3866.48) as opposed to other dopaminergic ligands crystallized to date. 60,62,63 The common anchoring point throughout the series is the salt bridge interaction between the charged nitrogen in the piperazine and the sidechain of Asp1143.32. The position of the secondary pharmacophore (SP) is, as expected, more variable. Interestingly, there is a correlation between the pharmacological activity and the structural features introduced, thus providing an initial proposal for the structural interpretation of ligand bias on this receptor. Thus, the two strongest G-protein biased ligands (22a and 24d, magenta in Figure 5) place the benzyl tail toward the extracellular region, thus making distinct contacts with the tip of TM7 (Pro4057.32 and Tyr4087.35). This arrangement is in contrast to the other benzyl-containing SP, which were moderately selective for the β-arrestin signaling pathway (20a and 27a) and bend the SP toward TM2 (Val912.61 and Leu942.64, ligands in cyan in Figure 5). A similar orientation was found for the cyclohexyl substituent in ligands 20b (β-arrestin-biased) and 27b (nonbiased) or even for the benzyl-containing 29a, which shows a less pronounced bias toward the G-protein pathway. According to this model, imidazole- or benzimidazole-based SP groups, specifically decorated with the exocyclic amide benzyl substituted, occupy a distinct subpocket that might be related to their G-protein biased profile, while there is no clear specific subpocket for β-arresting biased ligands. Additionally, this Figure 5. Binding mode of compounds in Table 2 on the inactive DRD2, PDB code 6CM4 (A). Compounds are color coded according to their pharmacological activity as G-protein biased (magenta), β-arrestin biased (cyan), or no bias (green). The detailed binding mode for each compound class is shown, for compound 24d (B) and compound 27a (C), depicting the residues involved in interactions with the ligand in each case, and the variable region occupied by the SP designated with a blue circle. Journal of Medicinal Chemistry pubs.acs.org/jmc Article https://doi.org/10.1021/acs.jmedchem.1c00704 J. Med. Chem. 2021, 64, 8710−8726 8718 J. D2Dopamine Receptor G Protein-Biased Partial Agonists Based on Cariprazine. J. Med. Chem. 2019,62, 4755−4771. (34) Free, R. B.; Chun, L. S.; Moritz, A. E.; Miller, B. N.; Doyle, T. B.; Conroy, J. L.; Padron, A.; Meade, J. A.; Xiao, J.; Hu, X.; Dulcey, A. E.; Han, Y.; Duan, L.; Titus, S.; Bryant-Genevier, M.; Barnaeva, E.; Ferrer, M.; Javitch, J. A.; Beuming, T.; Shi, L.; Southall, N. T.; Marugan, J. J.; Sibley, D. R. Discovery and Characterization of a G Protein-Biased Agonist That Inhibits β-Arrestin Recruitment to the D2Dopamine Receptor. Mol. Pharmacol. 2014,86,96−105. (35) Chun, L. S.; Vekariya, R. H.; Free, R. B.; Li, Y.; Lin, D. T.; Su, P.; Liu, F.; Namkung, Y.; Laporte, S. A.; Moritz, A. E.; Aubé, J.; Frankowski, K. J.; Sibley, D. R. Structure-Activity Investigation of a G Protein-Biased Agonist Reveals Molecular Determinants for Biased Signaling of the D2Dopamine Receptor. Front. Synaptic Neurosci. 2018,10,2. (36) Männel, B.; Dengler, D.; Shonberg, J.; Hubner, H.; Möller, D.; Gmeiner, P. Hydroxy-Substituted Heteroarylpiperazines: Novel Scaffolds for β-Arrestin-Biased D2R Agonists. J. Med. Chem. 2017, 60, 4693−4713. (37) Jacobson, M.; Childers, W.; Abou-Gharbia, M. Dopamine D 2 Partial Agonists −Discovery, Evolution, and Therapeutic Potential. In Successful Drug Discovery; Wiley, 2019; Chap. 4, 83−130. (38) Männel, B.; Hubner, H.; Möller, D.; Gmeiner, P. β-Arrestin Biased Dopamine D2 Receptor Partial Agonists: Synthesis and Pharmacological Evaluation. Bioorganic Med. Chem. 2017,25, 5613− 5628. (39) Trincavelli, M. L.; Daniele, S.; Orlandini, E.; Navarro, G.; Casadó, V.; Giacomelli, C.; Nencetti, S.; Nuti, E.; Macchia, M.; Huebner, H.; Gmeiner, P.; Rossello, A.; Lluís, C.; Martini, C. A New D2Dopamine Receptor Agonist Allosterically Modulates A2A Adenosine Receptor Signalling by Interacting with the A2A/D2 Receptor Heteromer. Cell. Signal. 2012,24, 951−960. (40) Moritz, A. E.; Benjamin Free, R.; Sibley, D. R. Advances and Challenges in the Search for D2and D3Dopamine Receptor-Selective Compounds. Cell. Signal. 2018,41,75−81. (41) Roth, B. L.; Sheffer, D. J.; Kroeze, W. K. Magic Shotguns versus Magic Bullets: Selectively Non-Selective Drugs for Mood Disorders and Schizophrenia. Nat. Rev. Drug Discov. 2004,3, 353−359. (42) Sexton, P. M.; Christopoulos, A. To Bind or Not to Bind: Unravelling GPCR Polypharmacology. Cell 2018,172, 636−638. (43) Kopinathan, A.; Draper-Joyce, C.; Szabo, M.; Christopoulos, A.; Scammells, P. J.; Lane, J. R.; Capuano, B. Subtle Modifications to the Indole-2-Carboxamide Motif of the Negative Allosteric Modulator N-((trans)-4-(2-(7-cyano-3,4-dihydroisoquinolin-2(1H)-yl)ethyl)- cyclohexyl)-1H-indole-2-carboxamide (SB269652) Yield Dramatic Changes in Pharmacological Activity at the Dopamine D2Receptor. J. Med. Chem. 2019,62, 371−377. (44) Löber, S.; Hubner, H.; Tschammer, N.; Gmeiner, P. Recent Advances in the Search for D3- and D4- Selective Drugs: Probes. Models Candidates. Trends Pharmacol. Sci. 2011,32, 148−157. (45) Klein Herenbrink, C.; Verma, R.; Lim, H. D.; Kopinathan, A.; Keen, A.; Shonberg, J.; Draper-Joyce, C. J.; Scammells, P. J.; Christopoulos, A.; Javitch, J. A.; Capuano, B.; Shi, L.; Lane, R. Molecular Determinants of the Intrinsic Efficacy of the Antipsychotic Aripiprazole. ACS Chem. Biol. 2019,14, 1780−1792. (46) Dömling, A.; Ugi, I. Multicomponent Reactions with Isocyanides. Angew. Chemie 2000,39, 3168−3210. (47) Dömling, A.; Wang, W.; Wang, K. Chemistry and Biology Of Multicomponent Reactions. Chem. Rev. 2012,112, 3083−3135. (48) Cioc, R. C.; Ruijter, E.; Orru, R. V. A. Multicomponent Reactions: Advanced Tools for Sustainable Organic Synthesis. Green Chem. 2014,16, 2958−2975. (49) Abdelraheem, E. M. M.; Khaksar, S.; Kurpiewska, K.; Kalinowska-Tuscik, J.; Shaabani, S.; Dömling, A. Two-Step Macrocycle Synthesis by Classical Ugi Reaction. J. Org. Chem. 2018,83, 1441−1447. (50) Kurhade, S.; Diekstra, E.; Sutanto, F.; Kurpiewska, K.; Kalinowska-Tłuscik, J.; Dömling, A. Multicomponent Reaction Based Synthesis of 1-tetrazolylimidazo[1,5- a]pyridines. Org. Lett. 2018,20, 3871−3874. (51) Madhavachary, R.; Zarganes-Tzitzikas, T.; Patil, P.; Kurpiewska, K.; Kalinowska-Tłuscik, J.; Dömling, A. Synthesis of Highly Substituted Imidazole Uracil Containing Molecules via Ugi- 4CR and Passerini-3CR. ACS Comb. Sci. 2018,20, 192−196. (52) Hulme, C.; Chappeta, S.; Griffith, C.; Lee, Y. S.; Dietrich, J. An Efficient Solution Phase Synthesis of Triazadibenzoazulenones: ’designer Isonitrile Free’Methodology Enabled by Microwaves. Tetrahedron Lett. 2009,50, 1939−1942. (53) La Venia, A.; Lemrová, B.; Krchnák, V. Regioselective Incorporation of Backbone Constraints Compatible with Traditional Solid-Phase Peptide Synthesis. ACS Comb. Sci. 2013,15,59−72. (54) Rossen, K.; Sager, J.; Dimichele, L. M. An Efficient and Versatile Synthesis of Piperazine-2-Carboxamides. Tetrahedron Lett. 1997,38, 3183−3186. (55) Hulme, C.; Cherrier, M. P. Novel Applications of Ethyl Glyoxalate with the Ugi MCR. Tetrahedron Lett. 1999,40, 5295− 5299. (56) Reyes-Resina, I.; Samadi, A.; Navarro, G.; Saadeh, H. A.; Khasawneh, M. A.; Mestres, J.; Marco-Contelles, J.; Franco, R. Identification of a Tool Compound to Study the Mechanisms of Functional Selectivity between D2and D3Dopamine Receptors. ACS Omega 2018,3, 17368−17375. (57) Navarro, G.; Cordomí, A.; Brugarolas, M.; Moreno, E.; Aguinaga, D.; Pérez-Benito, L.; Ferre, S.; Cortés, A.; Casadó, V.; Mallol, J.; Canela, E. I.; Lluís, C.; Pardo, L.; McCormick, P. J.; Franco, R. Cross-Communication between Gi and Gs in a G-Protein-Coupled Receptor Heterotetramer Guided by a Receptor C-Terminal Domain. BMC Biol. 2018,16, 24. (58) Tsuruta, K.; Frey, E. A.; Grewe, C. W.; Cote, T. E.; Eskay, R. L.; Kebabian, J. W. Evidence That LY-141865 Specifically Stimulates the D-2 Dopamine Receptor. Nature 1981,292, 463−465. (59) Van Der Westhuizen, E. T.; Breton, B.; Christopoulos, A.; Bouvier, M. Quantification of Ligand Bias for Clinically Relevant β2- Adrenergic Receptor Ligands: Implications for Drug Taxonomy S. Mol. Pharmacol. 2014,85, 492−509. (60) Wang, S.; Che, T.; Levit, a.; Shoichet, B. K.; Wacker, D.; Roth, B. L. Structure of the D2Dopamine Receptor Bound to the Atypical Antipsychotic Drug Risperidone. Nat. Publ. Gr. 2018, 269−273. (61) Wacker, D.; Fenalti, G.; Brown, M. A.; Katritch, V.; Abagyan, R.; Cherezov, V.; Stevens, R. C. Conserved Binding Mode of Human β2Adrenergic Receptor Inverse Agonists and Antagonist Revealed by X-Ray Crystallography. J. Am. Chem. Soc. 2010,132, 11443−11445. (62) Zhou, Y.; Cao, C.; He, L.; Wang, X.; Zhang, X. C. Crystal Structure of Dopamine Receptor D4Bound to the Subtype Selective Ligand, L745870. eLife 2019,8, e48822. (63) Chien, E. Y. T.; Liu, W.; Zhao, Q.; Katritch, V.; Han, G. W.; Hanson, M. A.; Shi, L.; Newman, A. H.; Javitch, J. A.; Cherezov, V.; Stevens, R. C. Structure of the Human Dopamine D3Receptor in Complex with a D2/D3Selective Antagonist. Science 2010,330, 1091−1095. (64) Fan, L.; Tan, L.; Chen, Z.; Qi, J.; Nie, F.; Luo, Z.; Cheng, J.; Wang, S. Haloperidol Bound D2Dopamine Receptor Structure Inspired the Discovery of Subtype Selective Ligands. Nat. Commun. 2020,11 (). (65) Yin, J.; Chen, K. Y. M.; Clark, M. J.; Hijazi, M.; Kumari, P.; Bai, X.; chen; Sunahara, R. K.; Barth, P.; Rosenbaum, D. M. Structure of a D2Dopamine Receptor−G-Protein Complex in a Lipid Membrane. Nature 2020,584, 125−129. (66) Sherman, W.; Day, T.; Jacobson, M. P.; Friesner, R. A.; Farid, R. Novel Procedure for Modeling Ligand/Receptor Induced Fit Effects. J. Med. Chem. 2006,49, 534−553. (67) Navarro, G.; Hradsky, J.; Lluís, C.; Casadó, V.; McCormick, P. J.; Kreutz, M. R.; Mikhaylova, M. NCS-1 Associates with Adenosine A2A Receptors and Modulates Receptor Function. Front. Mol. Neurosci. 2012,5, 53. (68) Schrodinger. Schrodinger. LLC New York 2009. Journal of Medicinal Chemistry pubs.acs.org/jmc Article https://doi.org/10.1021/acs.jmedchem.1c00704 J. Med. Chem. 2021, 64, 8710−8726 8725 (69) Halgren, T. A.; Murphy, R. B.; Friesner, R. A.; Beard, H. S.; Frye, L. L.; Thomas Pollard, W.; Banks, J. L. Glide: A New Approach for Rapid, Accurate Docking and Scoring. 2. Enrichment Factors in Database Screening. J. Med. Chem. 2004,47, 1750−1759. Journal of Medicinal Chemistry pubs.acs.org/jmc Article https://doi.org/10.1021/acs.jmedchem.1c00704 J. Med. Chem. 2021, 64, 8710−8726 8726