Pharmacological and Physicochemical Properties Optimization for Dual-Target Dopamine D3 (D3R) and μ‑Opioid (MOR) Receptor Ligands as Potentially Safer Analgesics

Pharmacological and Physicochemical Properties Optimization for Dual-Target Dopamine D3 (D3R) and μ-Opioid (MOR) Receptor Ligands as Potentially Safer Analgesics
Alessandro Bonifazi, Elizabeth Saab, Julie Sanchez, Antonina L. Nazarova, Saheem A. Zaidi, Khorshada Jahan, Vsevolod Katritch, Meritxell Canals, J. Robert Lane, and Amy Hauck Newman
Journal of Medicinal Chemistry 2023 66 (15), 10304-10341
DOI: 10.1021/acs.jmedchem.3c00417

ABSTRACT: A new generation of dual-target μ opioid receptor (MOR) agonist/dopamine D3 receptor (D3R) antagonist/partial agonists with optimized physicochemical properties was designed and synthesized. Combining in vitro cell-based on-target/off-target affinity screening, in silico computer-aided drug design, and BRET functional assays, we identified new structural scaffolds that achieved high affinity and agonist/antagonist potencies for MOR and D3R, respectively, improving the dopamine receptor subtype selectivity (e.g., D3R over D2R) and significantly enhancing central nervous system multiparameter optimization scores for predicted blood−brain barrier permeability. We identified the substituted trans-(2S,4R)-pyrrolidine and trans-phenylcyclopropyl amine as key dopaminergic moieties and tethered these to different opioid scaffolds, derived from the MOR agonists TRV130 (3) or loperamide (6). The lead compounds 46, 84, 114, and 121 have the potential of producing analgesic effects through MOR partial agonism with reduced opioid-misuse liability via D3R antagonism. Moreover, the peripherally limited derivatives could have therapeutic indications for inflammation and neuropathic pain.

INTRODUCTION
The on-going opioid epidemic remains a serious public health crisis affecting social and economic welfare globally. The COVID-19 pandemic has sadly increased opioid-related overdose with no end in sight.1−3 Although there are medications to treat both opioid use disorders (OUD) (e.g., buprenorphine and methadone) and overdose (e.g., naloxone), these have proven inadequate to reverse the surge of opioid-related deaths in the past decade. Moreover, prevention of dependence on opioids prescribed for pain management, such as oxycodone, has not been addressed pharmacologically. Indeed, dependence on prescription opioids remains a major contributor to the current opioid crisis in the United States,1−6 although recently, the easy and cheap access to synthetic opioids such as fentanyl has become the major driver in opioid overdose.7 Alongside social and public health programs, the National Institutes of Health (NIH) has supported improvement in access to treatment and recovery, campaigns to promote safe opioid prescription practices and overdose prevention, and movements to combat the stigma associated with OUD. The development of innovative medications for OUD and safe, effective, non-addictive strategies to manage pain, while minimizing risk of relapse, is the focus of a highly translational NIH scientific effort, culminating with the launch of the HEAL initiative (helping to end addiction long-term). In 2019, the National Institute on Drug Abuse (NIDA) published a list of the most wanted medication development and therapeutic priorities, targeting pharmacological mechanisms to prevent and treat OUD and opioid overdose.8 One of the main priorities was the development of dopamine D3 receptors (D3R) antagonists and partial agonists. The D3R belongs to the D2-like receptor family and is activated by the endogenous neurotransmitter dopamine (DA). This DA receptor subtype is predominantly expressed in the mesolimbic DA region of the brain, which controls behaviors associated with drug-related cues, reinforcement, motivation, and reward.9 Numerous labs have pioneered drug design and pre-clinical development of highly selective D3R partial agonists and antagonists as pharmacotherapeutics for the treatment of psychostimulant use disorders and OUD.10 For example, recent drug candidates (±)-1 and (R)-2, two highly selective D3R antagonists, both decrease oxycodone drug-seeking and self-administration without decreasing anti-nociception, and importantly, without affecting peripheral biometric cardiovascular parameters when administered alone or in the presence of oxycodone or cocaine. Also, (±)-1 decreases dose escalation of opioid selfadministration in both male and female rats and reduces the acquisition of drug seeking behavior.These data have supported the preclinical development of these and other related compounds for not only the treatment of OUD, but potential prevention of opioid dependence, if administered with a prescription opioid for pain management.These observations have supported drug development campaigns toward combination treatments with opioid analgesics, or potentially with methadone or buprenorphine, to improve their efficacy in preventing relapse and minimizing the possibility of cardiovascular or other opioid-driven side effects. Furthermore, achieving D3R subtype selectivity over D2R may limit extrapyramidal side effects, locomotor impairments, and metabolic disorders, associated with D2R antagonism,improving their tolerability and compliance in this patient population. Despite significant efforts toward developing analgesics that are non-opioids, the primary target for pain medications continues to be the μ-opioid receptors (MOR); the most effective opioid painkillers act as agonists at MOR. MOR, within the central nervous system (CNS), are expressed in brain regions associated with reward and aversion and in areas controlling pain sensation and respiratory processes that have high concentrations of GABA neurons. MOR activation efficiently reduces severe pain, especially acute or perioperative pain, but the concomitant reward, tolerance, and respiratory depression effects pose a significant threat and risk for OUD and overdose. Recent efforts directed toward generating safer opioid analgesics21 have focused on developing biased agonists that preferentially activate the G-protein-dependent signaling cascade associated with MOR, and consequent analgesia, while limiting β-arrestin recruitment. Observations made in βarrestin2 knockout mice suggested that β-arrestin2 MOR signaling may mediate side effects such as constipation and respiratory depression.22 However, recent conflicting evidence shows that β-arrestin2 does not mediate opioid-induced respiratory depression,23−27 and some such biased agonists still appear to induce self-administration in experimental animals.21,28 Recent studies have suggested that the low intrinsic efficacy (weak partial agonism) of these compounds may cause the improved separation between anti-nociceptive and respiratory depressive doses and their safety profile.24,29−33 It is evident that parsing out mechanisms underlying pharmacologically desired versus unwanted effects and differentiating these at a cellular signaling level is still difficult when exclusively targeting the MOR. TRV130 (3) and PZM21 (4)34−36 are among some of the most recently studied MOR agonists. The former was initially reported to exclusively activate the Gi/o signaling pathway, without β-arrestin recruitment, with a rapid analgesic profile and limited side effects. However, despite ongoing investigations into its application for post-surgery pain, preclinical studies report abuse potential and constipation. Similar findings have been observed for the biased agonist 4,37 which was recently shown to induce respiratory depression comparable to morphine.38 TRV734, a more recent structural analog of 3 with a significantly improved oral bioavailability,39 is being evaluated in translational studies for its safety, pharmacodynamic profile, and pharmacokinetic parameters.40 Opioid-based pain medications can swiftly induce tolerance, dose escalation, and opioid-induced hyperalgesia (OIH),41 a paradoxical effect of increased peripheral nerve hypersensitivity. Proposed biased agonists, 3 and 4, have also been found to induce OIH and hyperalgesic priming in an animal model for transition to chronic pain.42 In an effort to control pain at a peripheral level and limit addictive liability, the development of peripherally limited MOR agonists that induce antinociception by targeting the peripheral opioid receptors located in sensory neurons has been pursued. In particular, peripheral opioid agonists can attenuate inflammation-induced excitability of primary afferent neurons and reduce the release of proinflammatory neuropeptides from peripheral terminals. In injured tissues, this can lead to analgesia and anti-inflammatory effects.43 Loperamide; an over-the-counter anti-diarrheal medication) is a peripherally restricted MOR agonist, due to its high affinity as a P-glycoprotein (P-gp) transporter substrate, which limits its ability to be retained in the CNS. Loperamide (6) was the first commercially available peripherally limited MOR agonist to show anti-hyperalgesic properties on its own, particularly in reducing heat and mechanical hyperalgesia in nerve injured rats,21,44 mediated at the peripheral terminals of the afferent fibers.45,46 It has also been found to produce a synergistic peripherally mediated analgesia in a mouse inflammatory pain model, when administered in combination with the δ-opioid receptor agonist oxymorphindole.44 Based on our interest in both opioid and DA D2-like receptors systems, we recently developed an innovative drug design approach,10−12,48 merging the analgesic properties of MOR agonists with D3R antagonism/partial agonism predicting reduced addictive liability.47 We designed compounds using a bivalent drug design49 and scaffold hybridization50 strategy, linking together two pharmacophores: (i) MOR agonist PP, targeting, and activating MOR in its OBS and (ii) D3R antagonist/partial agonist PP, binding within the D3R OBS. We also demonstrated that carefully chosen MOR PP can be accommodated in the D3R secondary binding pocket (SBP)47 and, similarly, well-designed D3R PP can also interact with the MOR SBP, increasing the affinity of these bivalent ligands for both targets. With this approach, we wanted to provide an alternative drug discovery strategy which could potentially speed up the preclinical development process by simplifying pharmacological and toxicological studies, that until now have been conducted with two different drugs in combination, by obtaining a single molecule with merged pharmacological properties. Considering the need for both centrally as well as peripherally limited analgesics, we took a two-pronged approach to (1) discover safer CNS active analgesics, with reduced addictive liability, as well as (2) generate new families of primarily peripherally acting ligands, for their therapeutic potential in peripheral pain and inflammation. Using the 6- scaffold, given its structural features resembling classical D3R ligands such as haloperidol (7; Figure 1), allowed for an initial structure−activity relationships (SAR) campaign,47 where we identified and characterized the most potent, first in class, dualtarget MOR agonists-D3R antagonists (Figure 1). In our first reported series,47 inspired and designed around 6 as the MOR PP, most of the lead molecules (8, 9, and 10; Figure 1 and Table 1) presented CNS multiparameter optimization scores [CNS-MPO; a value indicating the likelihood of a drug to cross the blood−brain barrier (BBB)]51 <2.5, suggesting a more peripherally restricted profile. In this study, we implemented a combination of in silico methods to predict the ability of new drugs to target the CNS, and we sought to expand the SAR by specifically adjusting the physicochemical properties necessary to obtain higher CNS drug-like parameters, while still retaining high affinity agonist and antagonist potencies at both MOR and D3R targets of interest, respectively. Since the beginning of this new campaign, the SAR design was assisted by CNS-MPO calculations and predictions, as detailed in depth in the discussion below, and by parallel in vitro binding screening to assess the effect of every structural modification. We directed the chemistry and drug design around two MOR agonist pharmacophores: (1) 6, to continue improving the previous generation of ligands with the aim of increasing potency and CNS profile, allowing for the comparison of centrally and peripherally active ligands; and (2) 3, the well-known potent and centrally active MOR agonist, thus allowing us to specifically improve its pharmacology to reduce addictive liability with the dual target approach. Due to extensive SAR studies already published,36 the most favorable positions to link the D3R pharmacophores were revealed. Moreover, we were particularly focused on the possibility of using the 3-scaffold for modulating different maximal efficacy at MOR, creating libraries of full and partial agonists to further establish the correlation between intrinsic efficacy and therapeutic window.30 Herein, we aimed to synthesize dual target compounds with maximized D3R affinity and subtype preferential selectivity. In addition, we aimed to obtain compounds with moderate MOR affinity and moderate MOR efficacy, still sufficient to elicit significant in vivo analgesia, but with reduced receptor desensitization, in contrast to the extremely potent and efficacious addictive synthetic opioids. One of the most important challenges was to find the right balance between MOR and D3R affinity, which can be successfully translated in vivo occupancy. Ultimately, to have an effect on limiting the opioid pharmacophore rewarding profile, a significant level of D3R occupancy52,53 needs to be achieved at the dose used for effective anti-nociception. This would likely require a higher affinity D3R antagonist/partial agonist; meanwhile, a more moderate MOR agonist/partial agonist affinity would be able to elicit relevant analgesia,54,55 due to amplification of the agonist activated cellular signaling cascade and/or MOR receptor reserve. For the D3R PP, in addition to the classical high affinity/ selectivity-inducing scaffolds, like the 2,3-dichlorophenyl piperazine [inspired by PG648 (11)]56 and the 1-(3-chloro5-ethyl-2-methoxyphenyl)piperazine [inspired by eticlopride (12) and 1] 57 (Figure 1), we also decided to investigate new chemical space and tether less explored scaffolds, such as 1-(6- (trifluoromethyl)pyridin-2-yl)piperazine (predicted to enhance the D3R subtype selectivity), the 3-(piperazin-1-yl)benzo[d]- isothiazole [inspired by the atypical antipsychotic perospirone58−60 (13)], and variously substituted phenyl cyclopropylamines [inspired by tranylcypromine (14) and its analogues]61 (Figures 1 and 2). Finally, to evaluate the role of key physicochemical properties, such as pKa and clog P, a significant effort was directed in designing complex linkers, such as substituted pyrrolidine rings (inspired by 12 and its bitopic analogues, i.e., 15; Figure 1) focusing on the basicity of the harboring amines, effect of electron withdrawing groups (EWG; inspired by 1,16 and 2,17 pairs, Figure 1) 57,62 or donating substituents, presence of hydrogen bond (H-bond) donor/acceptor groups, and stereochemistry, which we previously reported to be a fundamental factor in bitopic and bivalent drug design, when targeting two sites within the same receptors, exploiting unique ligand poses49,63,64 (Figure 2).

■ RESULTS AND DISCUSSION Drug Design and Synthesis. This generation of bivalent dual target ligands can be sub-divided into two main categories based on the MOR agonist PP: (A) the N,N-dimethyl-2,2- diphenylacetamides, derived from 6, and (B) the 2- (tetrahydro-2H-pyran-4-yl)pyridines, inspired by 336 and its analogues. For both classes of PPs, variations in the substitution patterns around the MOR agonist scaffold, the length, stereochemistry, and structural composition of the linker, and the connected D3R pharmacophores with antagonists or partial agonists functionality were explored to create a large SAR library. It was important to not only understand the biological implications of each of the Journal of Medicinal Chemistry pubs.acs.org/jmc Article https://doi.org/10.1021/acs.jmedchem.3c00417 J. Med. Chem. 2023, 66, 10304−10341 10307 MOR and D3R substituted pharmacophores attached in a bivalent fashion, but also to investigate the role of the linker in its geometry and chemical space. All of the compounds were designed and synthesized to retain or improve the desired pharmacological profile and receptor subtype selectivity without neglecting the importance of balanced physicochemical properties essential for CNS activity. The CNS-MPO scores predicting BBB permeability and CNS activity are based on six factors (Table S1): molecular weight (MW), clog P, clog D, pKa of the most basic group, topological polar surface area (TPSA), and number of hydrogen bond donor groups. Molecular weight >360, clog P >3, and clog D (at pH 7.4) >2 negatively impact the CNSMPO scores, with 0 scoring values (T0) for these parameters at 500 for molecular weight, 4 for clog D, and 5 for clog P. 51 To fulfill essential bivalent requirements of dual-target D3RMOR pharmacophores, we generally end up with unfavorable high molecular weight compounds, along with higher clog P/ clog D values, and very basic pKa due to presence of multiple secondary/tertiary basic amine groups, fundamental for the respective target binding. Hence, we implemented the following designs to improve each of the parameters involved in CNS-MPO calculations: (i) decreasing MW, clog P, and clog D; CNS-MPO score tends to add significant penalties on MW and clog P; however, there are very good reasons to keep these values low. Larger clog P values can improve the rate of penetration across the membranes but, at the same time, large lipophilic molecules also show non-specific protein binding, thus limiting unbound “active” form of the drug (both in plasma and brain); (ii) decreasing pKa of the basic nitrogen and incorporating EWGs close to basic amine, which might increase concentration of non-ionized diffusible form and reduce P-gp recognition;65 (iii) reducing H-bond donors by replacement of H-bond donor atoms or incorporating H-bond acceptors capable of engaging H-bond donors in intramolecular H-bonds, which reduce availability to water solvent and P-gp and also reduce flexibility of the molecule; ultimately, (iv) while TPSA was not specifically used as a factor in the design, it is important to note that removal of heteroatoms will reduce TPSA, while including H-bond acceptors will increase the value. CNS-MPO66−68 scores have been calculated for each new candidate and are reported in Tables 1, 2, and S1. The first series of compounds, based on 6, was designed using previous SAR47 that highlighted the importance of linker rigidity and the presence of basic tertiary amine groups that help to achieve optimal pKa values in comparison to previously reported secondary amines. This led to the use of welltolerated pyrrolidine (L-proline inspired) linkers. The versatility and importance of using pyrrolidine moieties with specific stereochemistry and substitution patterns when designing D3R antagonists has been the focus of previous SAR campaigns inspired by 12 (Figure 1).63 It was expected that the presence of a hydroxyl group with the well-established optimal trans-(2S,4R) 63 stereochemistry around the pyrrolidine (Figure 2) ring would reduce the lipophilicity of these highly functionalized bivalent compounds. The hydroxy group of commercially available starting material (2S,4R)-1-(tert-butoxycarbonyl)-4-hydroxypyrrolidine-2-carboxylic acid (18; Scheme 1) was first protected as a tert-butyl dimethyl silyl ether (19), prior to reducing the carboxylic acid group to the corresponding primary alcohol 20. Further oxidation, assisted with Dess−Martin periodinane (DMP), provided the aldehyde intermediate for reductive amination. As detailed in the Experimental Methods section, due to the poor solubility of the intermediate iminium salt, after work-up, the reductive amination proceeded to the formation of the enamine product 22, as a mixture of diastereoisomers, instead of the fully reduced tertiary amine. Nevertheless, the unsaturated scaffold was considered an interesting additional derivative. Thus, the deprotection of the silyl ether with TBAF and the Boc-group with TFA allowed for the deprotection of the key secondary amine 24 to react with N,N-dimethyl-4-oxo-2,2-diphenylbutanamide (25) 47 in a reductive amination to prepare 26.

■ CONCLUSIONS It is evident how challenging it is to combine drug design features that satisfy multiple-receptor affinity, subtype selectivity, and exploration of innovative chemical space resulting in retention of pharmacologically desired efficacy at both receptor targets, while attempting to also optimize peripheral versus CNS activity. Often chemical modifications that improve CNS penetrability are not favorable for dual receptor engagement, and on the contrary, highly decorated scaffolds that can push the limit of drug design to extremely high affinity and potency, exploiting the nuances of structurebased drug design, tend to be detrimental for brain/plasma distribution and CNS activity. Ultimately, the key resides in being able to find the right balance. Herein, we reported an extensive SAR study conducted in parallel between in vitro cell-based assays screening and in silico prediction models that ultimately identified promising leads based on 6 (i.e., 46, 55, 56, 58, and 84) and 3 (102, 114, and 121), which presented high to moderate affinities and selectivity for both MOR and D3R, while still retaining CNS-MPO scores between 2.8 and 3.7. We not only introduced an innovative approach for dualtarget pharmacology, but with highly methodical synthetic work, supported by CADD, we explored untapped chemical space, probing the respective receptor’s binding sites and thus identifying scaffold compositions for maximized pharmacological profiles. The new leads offer potential for in vivo analgesia with reduced addictive liability and possibly present a new direction to develop safer pain medications. In addition, some of the most potent and active compounds, demonstrating the desired dual−pharmacology profile, have been predicted to be more peripherally limited (CNS-MPO <2.5: compounds 8, 9, and 10 from the previous generation;47 compounds 34, 68, 78, and 80 from this new series), yet may be valuable therapeutics for inflammation and pain mediated by peripheral MOR mechanisms. From a medicinal chemistry perspective, these new leads offer additional opportunities for future optimization campaigns. Moreover, they underscore how starting from selective ligands for specific receptor families, it is possible toidentify shared structural moieties to design and synthesize unique dual target pharmacotherapeutics.