Activation of the μ‐opioid receptor by alicyclic fentanyls: Changes from high potency full agonists to low potency partial agonists with increasing alicyclic substructure

Abstract Fentanyl analogs represent an important group of new psychoactive substances and knowing their efficacy and potency might assist in interpreting observed concentrations. The potency of fentanyl analogs can be estimated from in vitro studies and can be used to establish structure–activity relationships. In this study, recombinant CHO‐K1 cells (AequoScreen) expressing the human μ‐opioid receptor were used to establish dose–response curves via luminescent analysis for cyclopropyl‐, cyclobutyl‐, cyclopentyl‐, cyclohexyl‐, and 2,2,3,3‐tetramethylcyclopropylfentanyl (TMCPF), on three separate occasions, using eight different concentrations in an eight‐fold serial dilution in triplicates starting at ~60 μM. Fentanyl was used as a full agonist reference while morphine and buprenorphine were included for comparison. Cyclopropylfentanyl (EC50 = 4.3 nM), cyclobutylfentanyl (EC50 = 6.2 nM), and cyclopentylfentanyl (EC50 = 13 nM) were full agonists slightly less potent than fentanyl (EC50 = 1.7 nM). Cyclohexylfentanyl (EC50 = 3.1 μM, efficacy 48%) and TMCPF (EC50 = 1.5 μM, efficacy 65%) were partial agonists less potent than morphine (EC50 = 430 nM). Based on the results, cyclopropyl‐, cyclobutyl‐, and cyclopentylfentanyl would be expected to induce intoxication or cause fatal poisonings at similar concentrations to fentanyl, while the toxic or fatal concentrations of cyclohexylfentanyl and TMCPF would be expected to be much higher.


| INTRODUCTION
Fentanyl analogs, belonging to the opioids, represent an important group of new psychoactive substances (NPS), especially when looking at drug overdoses and the number of deaths. 1 Opioids exert their effects by activating the three major types of opioid receptors (μ, δ, and κ). 2 However, the primary receptor involved in opioid addiction and the often lethal respiratory depression caused by opioid overdose is the μ-opioid receptor. 2,3 In cases of suspected intoxication, knowing the efficacy and potency of the NPS might assist toxicologists and medical examiners in their interpretation. Similarly, predictions of potency based on the structure could be of value in scheduling decisions on new uncharacterized NPS. Potency can be studied in several ways, including functional studies in laboratory animals 4 or receptor-based studies in vitro. The latter can be divided into receptor binding assays and receptor activity studies. Receptor binding studies measure the affinity of a ligand to the opioid receptor 5,6 and are suitable for high throughput. However, binding studies cannot differentiate between full and partial agonists, and even an antagonist such as naloxone shows binding affinity. 6 Receptor activation assays instead measure proximal or downstream activation of the receptor, allowing the potency and efficacy values for a certain agonist to be determined, making them more informative than binding assays. Receptor activation can be measured via different assays based on, for example, aequorin, 7 GTPγS binding, 8,9 cAMP, 10,11 and β-arrestin recruitment. 12 The interpretation of the data is complicated by the fact that potency is substantially affected by the assay conditions such as the reagent concentration, which reporting system was used, the cell type, and if a human or murine receptor was used, making it difficult to compare the results from different assays.
Alicyclic fentanyls share their overall chemical structure with fentanyl, except for the substitution of the propionamide with amides with increasingly large cyclic structures containing either a cyclopropyl-, cyclobutyl-, cyclopentyl-, cyclohexyl-, or 2,2,3,3-tetramethylcyclopropyl ring (see Figure 1). In 2017, cyclopropylfentanyl was reported in 59 deaths in Sweden. 13 Cyclopentylfentanyl 14 and 2,2,3,3-tetramethylcyclopropylfentanyl (TMCPF) have also been encountered as NPS in Sweden. 15 Similarly, cyclopropylfentanyl findings have been reported in the USA 16-18 and Switzerland. 19 Information about the pharmacokinetics of alicyclic fentanyls is limited but we know that they are metabolized differently in biological systems. 15 The structure-activity relationships for fentanyl analogs have been investigated and reviewed. 12,[20][21][22][23] The binding affinity of cyclopropylfentanyl has been reported as 0.088, 1.2 0.77, and 2.4 nM in different assays. 5,23,24 In addition, Hassanien et al 23  Previous literature indicates that both the potency and the efficacy differ between the different alicyclic fentanyl analogs but it is difficult to ascertain how the ring size impact these parameters.
Therefore, the aim of this study was to determine the μ-opioid receptor activity induced by a complete series of alicyclic fentanyls and to investigate the structure-activity relationships regarding potency and efficacy.

j Cell lines and cultivation
The receptor activation assay was carried out on AequoScreen recombinant CHO-K1 cell lines purchased from Perkin Elmer (Groningen, the Netherlands) expressing the human μ-opioid receptor (ES-542-AV) and subunit Gα16 coupling receptor activation to an increase in intracellular Ca 2+ concentration. 25 The cells also expressed apoaeqourin which, when combined with externally added coelenterazine, forms the photoprotein aequorin. When aequorin is exposed to Ca 2+ , coelenterazine is oxidized with the emission of light. 26 The flash luminescence can easily be read by a plate reader.
When combined, the μ-opioid receptor coupled to Gα16 and aequorin provide a convenient model system for measuring receptor activation.
The cell lines were cultured at 37 C in a humidified air atmosphere containing 5% CO 2 , in Ham's F12 medium supplemented with 10% FBS and passaged every 3-4 days. The cells were not cultured beyond 30 passages.

j Dose-response assay
Prior to the dose-response assays, the cells were cultured to a confluency of 70-90% and then trypsinized, centrifuged (150 × g for 5 min at room temperature), and resuspended in pre-warmed assay medium (DMEM/Ham's F12 without phenol red supplemented with 15 mM HEPES, L-glutamine, and 0.1% protease-free BSA) at a con- The experiments for each substance were repeated on three different days comprising 6 days of analysis. Fentanyl was included as a reference every day and the data set for fentanyl contains seven experiments, including two from the same day.

j Data analysis
Luminescence data from each well were summarized over the total reading time and blank measurements were subtracted. The response signals were normalized to the digitonin signal for each plate and then normalized to the plateau signal of fentanyl (average of top two concentrations analyzed in the same experiment), denoted as 100% activity.
The EC 50 values and efficacy with 95% confidence intervals (profile likelihood) and curve fittings (non-linear fit, three parameters, bottom constrained to 0%) were calculated using all data points analogs was compared with that of the full agonist fentanyl (5 comparisons, n = 3; fentanyl n = 7) using a one-way ANOVA with Dunnett correction for multiple tests. Differences in potency, as Log (EC 50 ) values from the regression, were compared between fentanyl and all fentanyl analogs (15 comparisons, n = 3; fentanyl n = 7) using a one-way ANOVA with Dunnett correction for multiple tests.

j RESULTS
Full dose-response curves were obtained for cyclopropyl-, cyclobutyl-, cyclopentyl-, and cyclohexylfentanyl, as well as TMCPF, fentanyl, morphine, and buprenorphine, see Figure 1. All the potency and efficacy data can be found in Table 1.

j DISCUSSION
In the present study, all alicyclic fentanyls tested showed activity at the μ-opioid receptor indicating the potential for both abuse and fatal intoxication caused by respiratory depression.
As discussed by others, the measured K i values from receptor binding studies are highly dependent on the assay conditions, 6 and most likely this is also true for the EC 50 values obtained in this study.
To increase the relevance, the efficacies and potencies of the alicyclic fentanyls were compared with those of the well-characterized agonists fentanyl, morphine, and buprenorphine.
To the best of our knowledge fentanyl, morphine, and buprenorphine have never been compared using an assay based on the aequorin system, but a few studies using human receptors have been conducted using other reporting systems. Lipinski et al 5  fentanyl to be 180× more potent than morphine, while the Drug Enforcement Agency reported morphine to be more potent than fentanyl in their assay. This can be compared with the present study where fentanyl was 250× more potent than morphine. Three studies 9-11 have evaluated both morphine and buprenorphine and all three studies reported buprenorphine to be around >250× more potent than morphine, while we found them to be equipotent. For reference, buprenorphine has previously been reported as 25-100 times more potent than morphine in vivo. 27 The difference is surprising and the reason for this discrepancy is unknown. It might be related to the T A B L E 1 Efficacy and potency of alicyclic fentanyls. Adjusted P values are given for the difference in efficacy compared with fentanyl and values < 0.05 were considered significant. For potency, all differences are significant except for between cyclopropyl-and cyclobutylfentanyl. TMCPF, 2,2,3,3-tetramethylcyclopropylfentanyl slow receptor dissociation reported for buprenorphine. 27 In a flash assay, such as the one used in this study, with read times of a few seconds it is possible that the slow dissociation does not impact the assay in the same way as in cAMP and GTPγS binding assays with incubation times of an hour or more.
The five alicyclic fentanyls can be divided into two distinct groups. The three smaller analogs, cyclopropyl-, cyclobutyl-, and cyclopentylfentanyl were all full agonists with potencies similar to fentanyl, while cyclohexylfentanyl and TMCPF behaved like partial agonists of similar efficacy to buprenorphine but with lower potency. It was remarkable how suddenly the agonist behavior changed when going from a cyclopentyl ring to a cyclohexyl ring, and it is postulated that this might be due to steric hindrance at the binding site.
Except for cyclohexylfentanyl, the alicyclic fentanyl analogs have been studied before, although not together. 5,12,23,24 As there is considerable variability between the different assays the results were compared as fold changes from the EC 50 value and the efficacy of fentanyl. In our study, cyclopropylfentanyl was a full agonist 2.5-fold less potent than fentanyl. In other studies the EC 50 varied from 2.9-fold more potent to 1.7-fold less potent than fentanyl with efficacies between 84% and 107% compared with fentanyl. 5,12,23,24 Cyclobutylfentanyl was a full agonist 3.1-fold less potent than fentanyl in this study, while in the study by Hassanien et al 23  Earlier studies from our group have shown that the main routes of metabolism for alicyclic fentanyls include dealkylation to form normetabolites as well as oxidation of the alicyclic rings. With increasing ring size fewer normetabolites are formed, in favor of oxidation. 15 If the oxidized metabolites maintain their activity while the normetabolites, analogous to norfentanyl, are inactive this could contribute to the observed potency and/or duration of effects of fentanyls with larger cyclic structures in vivo, but further experiments are needed to verify that this is the case.

j CONCLUSIONS
In our study, using the AequoScreen assay, all alicyclic fentanyl analogs exhibited activity at the μ-opioid receptor. Cyclopropyl-, cyclobutyl-, and cyclopentylfentanyl were all full agonists with a similar potency to fentanyl. On the contrary cyclohexylfentanyl and TMCPF were partial analogs of similar efficacy to buprenorphine but with lower potency.