Attenuated G protein signaling and minimal receptor phosphorylation as a biochemical signature of low side-effect opioid analgesics

Multi-receptor targeting has been proposed as a promising strategy for the development of opioid analgesics with fewer side effects. Cebranopadol and AT-121 are prototypical bifunctional ligands targeting the nociceptin/orphanin FQ peptide receptor (NOP) and µ-opioid receptor (MOP) that elicit potent analgesia in humans and nonhuman primates, respectively. Cebranopadol was reported to produce typical MOP-related side effects such as respiratory depression and reward, whereas AT-121 appeared to be devoid of these liabilities. However, the molecular basis underlying different side effect profiles in opioid analgesics remains unknown. Here, we examine agonist-induced receptor phosphorylation and G protein signaling profiles of a series of chemically diverse mixed MOP/NOP agonists, including cebranopadol and AT-121. We found that these compounds produce strikingly different MOP phosphorylation profiles. Cebranopadol, AT-034 and AT-324 stimulated extensive MOP phosphorylation, whereas AT-201 induced selective phosphorylation at S375 only. AT-121, on the other hand, did not promote any detectable MOP phosphorylation. Conversely, none of these compounds was able to elicit strong NOP phosphorylation and low NOP receptor phosphorylation correlated with partial agonism in a GIRK-channel assay. Our results suggest a close correlation between MOP receptor phosphorylation and side effect profile. Thus, bifunctional MOP/NOP opioid ligands combining low efficacy G protein signaling at both NOP and MOP with no detectable receptor phosphorylation appear to be devoid of side-effects such as respiratory depression, abuse liability or tolerance development, as with AT-121.


Western Blot analysis.
Stably transfected HEK293 cells were seeded onto poly-l-lysine-coated 60 mm dishes and grown to 80% confluency. After compound-treatment, cells were lysed in detergent buffer in the presence of protease (Complete Mini) and phosphatase (PhosSTOP) inhibitors (Sigma-Aldrich, Steinheim, Germany). Glycosylated MOP was enriched using wheat germ lectin-agarose (WGA) beads and NOP receptor was enriched using HA-beads (ThermoFisher Scientific, Schwerte, Germany) as described in detail 19

Results
Chemical structures of the novel AT compounds are depicted in Fig. 1. In vitro binding affinities of the compounds at NOP and MOP were determined using competitive radioligand displacement 35 . As shown in Table 1 2). At NOP, only AT-324 displayed full intrinsic activity (Table 2). Cebranopadol and AT-034 showed high intrinsic activity although with lower potency than AT-324. AT-201 and AT-121 displayed partial agonist efficacy at NOP. Conversely, at MOP, cebranopadol and AT-034 exhibited full intrinsic activity, whereas AT-324 and AT-201 had partial agonist efficacy. AT-121 displayed lower intrinsic activity at MOP compared to AT-201 but has similar agonist potency. We further investigated functional efficacy of the compounds using a GIRK channel-based fluorescent screening assay in AtT-20 cells expressing NOP or MOP, as an additional read-out for G protein signaling. All compounds produced a concentration-dependent change in the membrane potential which was recorded as a change in fluorescent signal of the FMP dye. E max was determined by normalizing the maximal change in fluorescent signal of each compound to that of standard full agonists at the respective receptors (Table 2). Both N/OFQ and DAMGO yielded EC 50 values in the nanomolar range at NOP and MOP, respectively (Fig. 3). Corresponding to results obtained in GTPɣS binding assay, AT-324 displayed full intrinsic activity at NOP whereas AT-201, AT-121,  www.nature.com/scientificreports/ www.nature.com/scientificreports/ cebranopadol and AT-034 showed partial agonist efficacy at NOP. At MOP, cebranopadol showed superagonist activity in our hands, with an E max of 127% compared to DAMGO. AT-034, AT-201 and AT-324 displayed high agonist efficacy while AT-121 showed the lowest E max at MOP in the GIRK assay compared to the other bifunctional ligands. Next, we studied agonist-induced receptor phosphorylation at NOP and MOP in stably transfected HEK293 cells (Fig. 4). At MOP, cebranopadol produced robust phosphorylation at T370, S375, T376 and moderate phosphorylation at T379 in a concentration-dependent manner. On the contrary, cebranopadol failed to elicit phosphorylation at NOP, except for very weak phosphorylation at the primary site S346. AT-034 induced robust phosphorylation at T370, S375, T376 as well as T379 in MOP, whereas no phosphorylation was seen at NOP. AT-324 elicited strong phosphorylation at T370 and S375 at MOP but was barely able to stimulate detectable phosphorylation at NOP. Conversely, AT-201 elicited phosphorylation only at S375 at MOP; however, no phosphorylation was detected at NOP. AT-121 was the only compound that failed to induce any phosphorylation at MOP or NOP even at higher concentrations. In summary, cebranopadol and AT-121 show major differences in their in vitro profile with respect to G protein signaling and receptor phosphorylation (Fig. 5).

Discussion
Various strategies have emerged in medicinal chemistry over the past few years to address the adverse sideeffects associated with the use of MOP-targeted opioids. These include biased MOP agonists, multi-or bi-functional ligands, positive allosteric modulators, or partial agonists; yet, a consensus strategy remains a significant challenge 7 . Biased agonism seemed to be a promising concept initially, however, clinical trials with the presumed G protein biased agonist TRV130 (oliceridine) failed to produce therapeutic results devoid of side-effects such as constipation 36 or respiratory depression, although respiratory effects appeared lower than in the morphine control group 37 . Another approach that shows substantial potential is co-activation of NOP and MOP receptors. Several studies indicate that activation of NOP can attenuate opioid-mediated adverse effects such as reinforcing and rewarding behaviors [38][39][40] . Moreover, co-activation of NOP and MOP has been shown to produce synergistic enhancement of MOP-mediated anti-nociception 10 . www.nature.com/scientificreports/ www.nature.com/scientificreports/ The compounds used in this study, AT-034, AT-121, AT-201, and AT-324 are bifunctional NOP-MOP ligands belonging to different chemical scaffolds 41 (Fig. 1), designed using medicinal chemistry strategies to have varying levels of NOP and MOP binding affinities and intrinsic efficacies 13,35,42,43 . AT-034 and AT-201 are from a series of piperidinyl-1,3-dihydroindolone-based NOP ligands whose chemical modifications resulted in varying affinity at the MOP receptor and differential pharmacological effects. AT-201 (previously called SR16435), which shows high binding affinity at NOP and MOP and has partial agonist efficacy at both receptors, shows naloxone-reversible antinociceptive activity in mice but also showed naloxone-reversible rewarding effects in the conditioned place preference model in mice 44,45 . However, compared with morphine, AT-201 showed lower tolerance development to its antinociceptive effects after chronic treatment, attributed to its NOP efficacy and bifunctional MOP/NOP profile 44,45 . The piperidinyl-indolinone AT-034, on the other hand, has higher binding affinity at MOP than at NOP (Table 1), and higher agonist efficacy at MOP than at NOP (Table 2). AT-034 was shown to significantly reduce cocaine self-administration in rats, similar to buprenorphine, an effect that required co-activation of both MOP and NOP receptors 46 .
AT-121 belongs to a novel class of spiro-isoquinolinone NOP ligands which were optimized using structurebased drug design to produce a MOP/NOP bifunctional ligand with partial agonist efficacy at both receptors 13 . In nonhuman primates, AT-121 produced potent anti-nociceptive and anti-hyperalgesic effects devoid of reinforcing effects or physical dependence. In addition, AT-121 did not compromise respiratory or cardiovascular functions functions at supra-analgesic doses and showed significantly lower tolerance development compared to morphine after chronic treatment in nonhuman primates 13 . AT-324 belongs to yet a different chemical scaffold, the triazaspirodecanone scaffold, which has yielded highly selective NOP ligands such as the selective NOP receptor agonist too compound Ro64-6198 4 . While AT-324 has similarly high binding affinity at NOP as Ro64-6198, medicinal chemistry-driven structural modifications of this scaffold afforded high MOP binding affinity and a bifunctional MOP/NOP agonist efficacy profile in vitro 41 , shown in Tables 1 and 2. AT-324 also shows potent antinociceptive activity in nonhuman primates (unpublished results). As a comparator, we used cebranopadol, which is now well-characterized for its pharmacological effects in various animal models as well www.nature.com/scientificreports/ as in human clinical studies 47 . Cebranopadol belongs to a class of spiro[cyclohexane-dihydropyrano [3,4-b] indol]-amines and is a potent NOP and MOP agonist 15 , chemically distinct from the AT-series of compounds. It produced potent analgesic effects in various rat models of acute and chronic pain 15,16 . Although it did not elicit respiratory or motor deficits, or itching, it produced reinforcing effects in nonhuman primates 48,49 . Recent clinical studies showed that cebranopadol is effective in treatment of post-operative pain, cancer-related pain, as well as lower back pain 47,[50][51][52] .
In this study, we assessed the in vitro pharmacological effects of these chemically diverse novel mixed MOP/ NOP agonists on agonist-induced G protein signaling and receptor phosphorylation. In addition to the standard ligand binding and GTP Ɣ S assays, we used two novel pharmacological measures, i.e., GIRK channel-based fluorescent assay and phosphosite-specific antibodies, to study different stages of agonist-induced receptor activation in vitro. The GIRK channel-based membrane potential assay is a reliable method that uses a fluorescent membrane potential (FMP) dye to study G protein signaling 24 . The role of GIRK channels as a key effector protein in signaling of receptors involved in pain pathways has been demonstrated in various animal models [53][54][55] . Receptor phosphorylation and its functional importance in GPCR regulation and desensitization have been extensively demonstrated 56,57 . Phosphosite-specific antibodies have proven to be reliable tools to study temporal and spatial aspects of receptor regulation. We have previously characterized phosphosite-specific antibodies that effectively detect phosphorylation at the carboxyl terminus of their respective G protein-coupled receptors. These include pT370, pS375, pT376 and p T379 at MOP and pS346, pS351 and pT362/S363 at NOP receptors [19][20][21] .
AT-034, which could be described as a 'MOP-dominant' MOP/NOP bifunctional agonist (i.e. intrinsic activity at MOP > intrinsic activity at NOP) from our assays (Fig. 5), elicited strong phosphorylation at pT370, S375, T376 and T379 residues at the C-terminal of MOP but showed no phosphorylation at NOP (Fig. 4). Although AT-324 Figure 5. Radar graphs illustrating the balance between agonist-mediated effects of the investigated compounds at NOP (orange) and MOP (blue). All axes represent values normalized to maximal concentration of N/OFQ or DAMGO set at 100%. Phosphorylation values were calculated for the primary phosphorylation site S346 at NOP and S375 at MOP. Scale ranging from − 10 to 150% with an interval of 20%. www.nature.com/scientificreports/ was more 'NOP-dominant' in G protein signaling assays, it induced robust phosphorylation of MOP, especially at S375, but no phosphorylation of NOP at any of the tested sites. In contrast, AT-201 is a 'balanced' MOP/NOP agonist (MOP intrinsic activity ≅ NOP intrinsic activity) that induced only S375 phosphorylation at MOP, but none at NOP. Along with its analgesic effects, it also displayed rewarding properties in mice 44 . Interestingly, clinically used buprenorphine which shows partial agonism at both MOP and NOP, has a phosphorylation profile similar to AT-201 wherein it stimulates only S375 phosphorylation at MOP but none at NOP 20,21 Cebranopadol showed a more 'MOP-dominant' profile and partial to high efficacy at NOP in G protein signaling assays. Despite its clinical efficacy in pain, cebranopadol has been reported to produce reinforcing effects in rats, nonhuman primates and produced drug liking in human clinical trials 48,58,59 . Correspondingly, it elicited strong multi-site phosphorylation at MOP and weak NOP phosphorylation at S346. Interestingly, AT-121 which is a partial agonist at both MOP and NOP did not induce any detectable phosphorylation signal at either of the receptors and only AT-121 was reported to be devoid of any opioid-related side-effects in a nonhuman primates 13 unlike AT-201 and cebranopadol 44,60 . Our current studies expand and reinforce the hypothesis that bifunctionalMOP/NOP compounds with lower intrinsic activity at both targets may have promising profiles with wider therapeutic windows and reduced adverse effects 49 . Recent findings that suggest low-efficacy partial agonism at MOP as an approach to design safer analgesics further support this hypothesis 61,62 . Altogether, the two parameters of (1) partial agonism in G protein signaling at MOP and NOP, combined with (2) low or absent agonist-induced receptor phosphorylation, appear to have predictive validity for favorable therapeutic effects avoiding abuse liability. However, the optimal balance of NOP-or MOP-related effects for development of novel and safe opioids remains to be examined in more detail. www.nature.com/scientificreports/