Synthesis and pharmacological characterization of ethylenediamine synthetic opioids in human μ‐opiate receptor 1 (OPRM1) expressing cells

Abstract Opioids are powerful analgesics acting via the human μ‐opiate receptor (hMOR). Opioid use is associated with adverse effects such as tolerance, addiction, respiratory depression, and constipation. Two synthetic opioids, AH‐7921 and U‐47700 that were developed in the 1970s but never marketed, have recently appeared on the illegal drug market and in forensic toxicology reports. These agents were initially characterized for their analgesic activity in rodents; however, their pharmacology at hMOR has not been delineated. Thus, we synthesized over 50 chemical analogs based on core AH‐7921 and U‐47700 structures to assess for their ability to couple to Gαi signaling and induce hMOR internalization. For both the AH‐7921 and U‐47700 analogs, the 3,4‐dichlorobenzoyl substituents were the most potent with comparable EC50 values for inhibition of cAMP accumulation; 26.49 ± 11.2 nmol L−1 and 8.8 ± 4.9 nmol L−1, respectively. Despite similar potencies for Gαi coupling, these two compounds had strikingly different hMOR internalization efficacies: U‐47700 (10 μmol L−1) induced ~25% hMOR internalization similar to DAMGO while AH‐7921 (10 μmol L−1) induced ~5% hMOR internalization similar to morphine. In addition, the R, R enantiomer of U‐47700 is significantly more potent than the S, S enantiomer at hMOR. In conclusion, these data suggest that U‐47700 and AH‐7921 analogs have high analgesic potential in humans, but with divergent receptor internalization profiles, suggesting that they may exhibit differences in clinical utility or abuse potential.


| INTRODUC TI ON
The ability of opiates to suppress painful stimuli is undisputed.
Opiates are a class of small molecules or peptides regardless of structure that can bind and activate opiate receptors. Morphine is an opioid extracted from the opium poppy that has been used as early as the third century BC to treat dysentery, pain, and suffering. 1 It relieves pain by binding to and activating any of the three gene products encoding for cell surface G protein-coupled receptors (GPCRs).
These receptors are the μ-, δ-, or κ-opiate receptors and they are encoded by the OPRM1, OPRD1, and OPRK1 genes, respectively. 2,3 Opioid analgesic properties stem from opioid receptor gene expression in sensory neurons of the brain and peripheral CNS and their coupling to intracellular heterotrimeric G proteins. Opiate binding induces a conformational change in opiate receptors and signals to rapidly suppress neuronal excitability by G protein-dependent modulation of Ca 2+ , K + , and Na + currents resulting in a profound reduced perception of pain. Of the three main opiate receptor subtypes, only compounds with relatively high selectivity for the μ-opiate receptor (MOR) have achieved widespread clinical utility due, in part, to increased adverse effects such as dysphoria, convulsions, or poor selectivity of agents that have been developed to selectively target the δ-or κ-opiate receptors. 1,2 Agonist binding-induced conformational changes of the MOR, in addition to activating inhibitory Gα i proteins, cause the phosphorylation of intracellular residues such as Ser 375 by a number of kinases (eg G protein-coupled receptor kinases [GRKs], PKC). 4 The MOR phosphorylation sites and the efficacy of phosphorylation can differ based on the agonist structure. MOR phosphorylation leads to β-arrestin recruitment, receptor desensitization, and internalization which are all regulatory processes central to the development of opiate tolerance. 5,6 While many synthetic and naturally occurring MOR agonists have high potency and efficacy for coupling to Gα i , they can diverge significantly in their ability to promote β-arrestin recruitment and receptor internalization. [7][8][9] For example, despite morphine's high potency and efficacy for Gα i coupling and widespread clinical use, it has very low efficacy for β-arrestin recruitment and causes very little MOR internalization. 7 Conversely, endogenously produced opioids, such as the enkephalins and β-endorphins, are within a 10-fold range of morphine in terms of Gα i coupling potency; however, they are far more efficacious than morphine for recruiting β-arrestin and inducing receptor internalization. 9 The concept that agonists could be designed to preferentially couple to Gα i vs β-arrestin recruitment, referred to as "biased signaling," was a driving force behind the development of new opiates such as oliceridine. 10 This idea was bolstered by reports showing that β-arrestin-2 knockout mice display increased analgesia, decreased tolerance, and have less respiratory depression after morphine administration. [11][12][13] Hence new opiates, such as oliceridine, were designed and selected for the ability to couple strongly to the Gα i pathway but with low efficacies for β-arrestin recruitment and internalization, similar to morphine. However, recent events, such as the failure of FDA approval for oliceridine in 2018 and new studies using mouse models with targeted mutations in the OPRM1 gene that prevent MOR internalization, suggest that this may not be the best approach for developing safer opiates with fewer side effects.
Mutations of carboxyl tail serine and threonine residues, including Ser 375 , that are phosphorylated by GRKs reveal that respiratory depression and constipation are significantly exacerbated when MOR receptors fail to internalize after agonist binding 14 while analgesic tolerance is reduced. Thus, perhaps the pursuit of new opioids that better mimic the endogenous system might result in novel more "balanced" agents with improved clinical utility over morphine-like derivatives.
In light of this, we sought to synthesize and pharmacologically characterize a series of structural analogs based on the ethylenediamine structural analogs AH-7921 and U-47700 and compare their pharmacology to morphine and the endogenous opioid mimetic DAMGO. These two synthetic opioids, first synthesized and patented in the early 1970s, have naloxone (NLX)-reversible analgesic potential in rodent models 15,16 in the potency range of morphine.
However, their pharmacological properties, including their ability to cause internalization of human μ-opioid receptors (hMORs), are unknown. Thus, they represent a potentially useful series of core structures that are relatively easy to synthesize from where new "balanced" or "biased" opiates could be designed.
The design of analogs was to assess both compounds described in the patents, as well as related novel analogs, for their pharmacological selectivity and efficacy for causing hMOR internalization.
Additionally, while the prior literature on the U-series compounds indicated that stereoisomers differ in their selectivity for κ-vs μopioid receptors, we sought to more clearly define the impacts of these differences on hMOR pharmacology by synthesizing and testing single stereoisomers of the U-series compounds. We present herein findings related to structure activity relationships of these two compounds and over 50 structural analogous to provide new insights on how chemical structures affect potency for Gα i signaling and efficacy for hMOR internalization.

| Acylation of starting amines
The respective acid chloride (1.05 eq) was added to a solution of the appropriate precursor amine (1.0 eq) and triethylamine (1.0 eq) in 5 mL of dry diethyl ether and stirred at room temperature for 16 hour. The reaction mixture was extracted with ethyl acetate (3×), washed with brine, dried over Na 2 SO 4 , and concentrated in vacuo.
The crude product was recrystallized from dichloromethane or precipitated with ethyl acetate upon sonication to give desired product as a solid. Reported hydrochloride salts of final compounds were made using 2.0 N HCl in diethyl ether solution.

| Validation of synthesized compound purity and structure
Nuclear magnetic resonance (NMR) spectra were recorded on a

| DNA constructs
Human µ-opioid receptor, OPRM1, with three sequential hemagglutinin antigen (3xHA) tags at the N-terminus (cDNA Resource Center OPRM10TN00) was subcloned into pENTR/D-TOPO vector (Thermo Fisher K240020). The subcloned plasmid underwent a three-way Gateway LR recombination along with pENTR-EF1α vector containing EF1α promoter and the lentiviral 2k7bsd destination vector with blasticidin resistance coding sequence, as described previously. 18 The resulting lentiviral plasmid vector is referred to herein as EF1α-3xHA-OPRM1-2k7bsd.

| Chemical syntheses, stereochemical considerations, and structural analog design
AH-7921 ( Figure 1) is an achiral compound made from cyclohexanone via a Strecker reaction as described in the original patents and in (Figure 2A). 17,19,20 The

| Potency of AH-7921 and structural analogs at hMOR
Seventeen analogs based on the AH-7921 core structure were synthesized and assessed for hMOR pharmacological activity (Table   S1). In addition to A01 (AH-7921), two analogs were found to result in significant suppression of FSK-induced cAMP accumulation that was reversed by NLX coadministration ( Figure 5). A02 significantly decreased FSK-induced cAMP levels at (0.1 µmol L −1 ) and showed further increasing dose-dependent decrease of cAMP levels. Thus, of the AH-series analogs we synthesized and tested, A01 ( Figure 4C; EC 50 26.9 ± 11.2 nmol L −1 ) and A02 ( Figure S6; EC 50 59.3 ± 2.0 nmol L −1 ) were classified as high potency hMOR agonists followed only by A04, which we classified as a low potency agonist ( Figure 5). The remainder of the AH-7921 compound series demonstrated either no activity or activity that was not reversible by NLX treatment (Table   S1, Figure S2A).
Pharmacological profiling also revealed U02, U05, and U08 as low potency agonists. Analogous to findings with AH-7921 analogs, were inactive based on comparison to FSK-only control ( Figure S3A).
U03 and U09 showed significant decreases in cAMP levels in the range of concentrations tested, but their regulation of cAMP was not NLX reversible ( Figure S3B). The summary of the U-47700 series of opioid compound analogs is shown in Table S2.

| Potency of Udes-and US-series of analogs at hMOR
To study the effect of removing N-methyl group on the amide in (R,R)-U-series, Udes-series (compounds Udes01-09) was synthesized and screened for hMOR signaling ( Figure S4 and Table S3).
Initial screening of Udes01 resulted in a significant decrease of FSKinduced cAMP levels at 10 nmol L −1 , and specificity for hMOR was confirmed by NLX reversibility (Figure 7). The EC 50 value of Udes01 was subsequently determined to be 3.0 ± 0.3 nmol L −1 ( Figure S6) characterizing it as the most potent analog that we discovered. The summary of the pharmacological activity of Udes-series of compound analogs at the hMOR is shown in Table S3. The remaining Udes-series (Udes02-09) did not significantly alter FSK-induced cAMP accumulation at any of the tested concentrations ( Figure S4).
The US-series was synthesized to assess for pharmacological activity of the S,S enantiomer of the U-47700 analogs. The nine USseries analogs were synthesized and screened for hMOR agonism (Table S4, Figure S5). Of these nine, only US01 (3,4-dichlorobenzoyl substituent) demonstrated significant decrease in cAMP levels at F I G U R E 6 Drug potency of active U-47700 series compounds. HT1080 hMOR cells are treated with 100 μmol L −1 FSK only, or FSK with 0.01 μmol L −1 (low), 0.1 μmol L −1 (mid), 1 μmol L −1 (high) of the indicated compound, and high concentration dose of the compound with 10 μmol L −1 NLX. Each treatment dose was performed in triplicate (n = 2; data from representative experiment shown) and data were analyzed using one-way ANOVA with Dunnett multiple comparison test and FSK only as standard. *P < 0.05 the highest screening dose of 1 μmol L −1 but not at the lower (0.1 or 0.01 µmol L −1 ) concentration ranges (Figure 7, Figure S5) indicating that the S,S enantiomer of U-47700 was significantly less potent than the R,R enantiomer.

| Efficacy of hMOR internalization by high potency AH-7921, U-47700 analogs
Agonist-induced receptor endocytosis remains a hallmark feature of GPCR activation and regulation and provides insights into the level of receptor desensitization that occurs after agonist stimulation. 8  U04 and Udes01 demonstrated hMOR internalization similar to morphine and the two AH-series compounds (Figure 8).

| D ISCUSS I ON
We present a synthesis and screening strategy that enabled us to identify the pharmacological activity of over 50 ethylenediamine-containing compounds at the hMOR. Of the high potency hMOR agonists that we found, the rank order of potencies for functional Gαi coupling and cAMP inhibition is as follows: showing U01 to be the most potent U-series analog followed by U04, then U02, U05, and U08; however, our data indicate that Udes01 was ~3X more potent than U01, which is much more potent than indicated in the aforementioned studies. This could be explained by differences in human vs rodent MOR protein-coding sequences or if the Udes01 tested contained a racemic mixture, which is quite likely and could explain its much lower potency.
In light of this, our U-series analog synthesis approach benefitted from the availability of the single stereoisomers of the advanced amine intermediates. This allowed for the synthesis of single stereoisomer analogs rather than the racemic versions of the analogs. Our findings, the (1R,2R) stereoisomer of U-47700 (U01), were substantially more potent than the (1S,2S) stereoisomer (US01). The influence of stereochemistry in the U-series compounds on μ-vs κ-opioid receptor selectivity has been reported 25,27 indicating that the (1S,2S) isomer has high κ-opioid receptor selectivity, while the (1R,2R) isomer has high μ-opioid receptor selectivity.
Differentiating the effects of single isomers vs racemic U-series compounds may help delineate in vivo effects in humans mediated by μ-and κ-opioid receptors. One report of a seized sample of U-49900, a diethyl amine version of U-47700 which is a dimethyl amine, demonstrated that it was a racemic mixture, as determined by circular dichroism. 28  It has been proposed that hMOR endocytosis is closely associated with the development of drug tolerance in the opiate user 8,30 and in mouse models. 14 Our data reveal that A01 and A02 demonstrated low levels (~15%) of hMOR internalization with similar drug potency for Gα i coupling. However, U01 and U04/Udes01 significantly (P < 0.05) diverged in their desensitization capacities: U01 induced high levels of internalization (~26%) while U04 and Udes01 induced between 8% and 12% of cell surface hMORs to internalize. Thus, the levels of hMOR internalization for A01, A02, U04, and Udes01 are similar to the reported internalization levels of well-characterized opioids, such as morphine and fentanyl, while U01 is much more similar to the endogenous opioids, the endorphins and the enkephalins. 7,9,31 Thus, if the goal was to design new opiates with safer and improved pharmacodynamic properties that mimic endogenously produced "balanced/ natural" opiate responses at the hMOR, further assessment of the (1R,2R) single isomer U01 is likely to produce internalization and potentially receptor desensitization effects more aligned with endogenous opiates although further studies are needed.
In summary, our results demonstrate that this in vitro functional assay for hMOR pharmacology provides a foundation from which effects of these agents in humans can be anticipated and they provide a scaffold for the rational design of potentially superior analgesics that better mimic the endogenous opioid system. 14,32 Through our combinatorial approach, we discovered some novel moderately potent hMOR agonists. Whether these novel compounds have different tolerance profiles or addictive potential needs to be further investigated in vivo.