**1. Introduction**

The δ-opioid receptor (δOR) has great potential as a therapeutic target to treat a myriad diseases and disorders. Preclinical use of δOR agonists suggest their utility to reduce anxiety, depression, alcohol use, migraine, neuropathic and inflammatory pain [1,2]. Yet, to this day roughly 30 years since the δOR was cloned [3,4] no δOR selective molecule has been FDA approved for clinical use. Between 2008–2010 a small set of δOR agonists entered phase II clinical trials (NCT00993863, NCT01058642, NCT00759395 and NCT00979953) for acute and chronic pain conditions as well as to treat depressive disorders [5]. However, none of these trials progressed to phase 3 clinical trials. A common shared feature of the phase 2 drug candidates, ADL5859, ADL5747 and AZD2327 was that their structure was based on that of previously developed potent and highly selective δOR agonists SNC80 and BW373U86 (SNC86), (Figure 1, [6–10]). A major concern with the original

**Citation:** Meqbil, Y.J.; Su, H.; Cassell, R.J.; Mores, K.L.; Gutridge, A.M.; Cummins, B.R.; Chen, L.; van Rijn, R.M. Identification of a Novel Delta Opioid Receptor Agonist Chemotype with Potential Negative Allosteric Modulator Capabilities. *Molecules* **2021**, *26*, 7236. https://doi.org/ 10.3390/molecules26237236

Academic Editor: Jay McLaughlin

Received: 5 November 2021 Accepted: 27 November 2021 Published: 29 November 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

SNC compounds was their propensity to induce severe seizures in rodents [11]. AZD2327 exhibited proconvulsant effects [8], whereas Adolor was able to modify the SNC structure enough to not observe tonic-clonic seizures [6,7]. However, recent studies have led to a better understanding of the mechanism by which SNC80 can cause seizures, implicating β-arrestin as a critical factor [12,13]. The SNC compounds are super-recruiters of β-arrestin, and it appears that ADL5859 and AZD2327, recruit β-arrestin on par with the endogenous agonist Leu-enkephalin, if not stronger [14–16]. In 2020, Conibear et al., developed a novel δOR agonist PN6047 (Figure 1), based on the SNC80 scaffold, which was not proconvulsant, and which recruited β-arrestin with efficacy slightly less than ARM390 (which in our hands has an Emax on par with Leu-enkephalin) [14,17]. Thus, PN6047 shared similarity with the failed Adolor and Astra Zeneca compounds, looking promising in terms of preclinical in vivo effects, but retaining high risk for a failure once moved into human clinical trials. Thus, in order for the δOR field to progess and produce a clinically viable candidate it is important to divert from the SNC80 scaffold. A handful of δOR selective small molecules have been produced that suggest this is possible: TAN-67 and KNT-127 (Figure 1) have distinct scaffolds and under-recruit β-arrestin, respectively with Emax for β-arrestin 2 recruitment of 30%, 70% and do not induce convulsions [14,18,19]. Similarly, kratom alkaloids, while displaying pan-opioid activity, are highly G-protein biased in that they do not show detectable β-arrestin 2 recruitment [20]. Our goal for this study was to identify novel δOR agonist scaffold(s) that under-recruit β-arrestin (relative to SNC80). In this study, we screened over 5000 chemical compounds from CNS-focused drug libraries. We were able to identify a molecule (compound **1**) with a novel chemical scaffold that was selective for δOR over the µ- and κ-opioid receptors (µOR and κOR) with micromolar affinity and potency. Computational modelling of compound **1** into the δOR crystal structure (PDB: 6PT3) suggests it is able to partially occupy the known orthosteric binding pocket as well as an allosteric binding pocket in the presence of Leu-enkephalin. Further in vitro analysis showed that compound **1** potentially negatively modulates the potency of Leu-enkephalin in an allosteric manner. in rodents [11]. AZD2327 exhibited proconvulsant effects [8], whereas Adolor was able to modify the SNC structure enough to not observe tonic-clonic seizures [6,7]. However, recent studies have led to a better understanding of the mechanism by which SNC80 can cause seizures, implicating β-arrestin as a critical factor [12,13]. The SNC compounds are super-recruiters of β-arrestin, and it appears that ADL5859 and AZD2327, recruit βarrestin on par with the endogenous agonist Leu-enkephalin, if not stronger [14–16]. In 2020, Conibear et al., developed a novel δOR agonist PN6047 (Figure 1), based on the SNC80 scaffold, which was not proconvulsant, and which recruited β-arrestin with efficacy slightly less than ARM390 (which in our hands has an Emax on par with Leuenkephalin) [14,17]. Thus, PN6047 shared similarity with the failed Adolor and Astra Zeneca compounds, looking promising in terms of preclinical in vivo effects, but retaining high risk for a failure once moved into human clinical trials. Thus, in order for the δOR field to progess and produce a clinically viable candidate it is important to divert from the SNC80 scaffold. A handful of δOR selective small molecules have been produced that suggest this is possible: TAN-67 and KNT-127 (Figure 1) have distinct scaffolds and under-recruit β-arrestin, respectively with Emax for β-arrestin 2 recruitment of 30%, 70% and do not induce convulsions [14,18,19]. Similarly, kratom alkaloids, while displaying pan-opioid activity, are highly G-protein biased in that they do not show detectable βarrestin 2 recruitment [20]. Our goal for this study was to identify novel δOR agonist scaffold(s) that under-recruit β-arrestin (relative to SNC80). In this study, we screened over 5,000 chemical compounds from CNS-focused drug libraries. We were able to identify a molecule (compound **1**) with a novel chemical scaffold that was selective for δOR over the µ- and κ-opioid receptors (µOR and κOR) with micromolar affinity and potency. Computational modelling of compound **1** into the δOR crystal structure (PDB: 6PT3) suggests it is able to partially occupy the known orthosteric binding pocket as well as an allosteric binding pocket in the presence of Leu-enkephalin. Further in vitro analysis showed that compound **1** potentially negatively modulates the potency of Leu-enkephalin in an allosteric manner.

highly selective δOR agonists SNC80 and BW373U86 (SNC86), (Figure 1, [6–10]). A major concern with the original SNC compounds was their propensity to induce severe seizures

*Molecules* **2021**, *26*, x FOR PEER REVIEW 2 of 15

**Figure 1.** Chemical structure of δOR agonists. BW373U86 (SNC86), SNC80, AZD2327, PN6047, ADL5859, ADL5747, (- )TAN-67 and KNT-127. **Figure 1.** Chemical structure of δOR agonists. BW373U86 (SNC86), SNC80, AZD2327, PN6047, ADL5859, ADL5747, (-)TAN-67 and KNT-127.

#### **2. Results 2. Results**

#### *2.1. Identification of a Novel δOR Agonist with Sub-Maximal β-Arrestin Recruitment Efficacy 2.1. Identification of a Novel δOR Agonist with Sub-Maximal β-Arrestin Recruitment Efficacy*

We have previously reported that SNC80 super-recruits β-arrestin 2 relative to Leuenkephalin but has equal β-arrestin 1 recruitment efficacy [14,15]. Thus, for ease of setting a cut-off threshold we decided to perform a high-throughput screen with the β-arrestin 1 We have previously reported that SNC80 super-recruits β-arrestin 2 relative to Leuenkephalin but has equal β-arrestin 1 recruitment efficacy [14,15]. Thus, for ease of setting a cut-off threshold we decided to perform a high-throughput screen with the β-arrestin 1 cells. We tested ~5100 compounds and identified a single positive hit, that, at a 10 µM concentration, displayed ~50% β-arrestin 1 recruitment relative to SNC80 (Figure 2).

*Molecules* **2021**, *26*, x FOR PEER REVIEW 3 of 15

**Figure 2.** Screening of a CNS-targeted compound library for β-arrestin 1 recruitment at δOR**.** 5200 compounds from sixteen 384-well plates from diverse CNS-targeted drug libraries were tested at 10µM for β-arrestin 1 recruitment at δORs in a PathHunter assay. The red dot represents the hit compound (**1**). 10 µM SNC80 was utilized for normalization. **Figure 2.** Screening of a CNS-targeted compound library for β-arrestin 1 recruitment at δOR. 5200 compounds from sixteen 384-well plates from diverse CNS-targeted drug libraries were tested at 10 µM for β-arrestin 1 recruitment at δORs in a PathHunter assay. The red dot represents the hit compound (**1**). 10 µM SNC80 was utilized for normalization. **Figure 2.** Screening of a CNS-targeted compound library for β-arrestin 1 recruitment at δOR**.** 5200 compounds from sixteen 384-well plates from diverse CNS-targeted drug libraries were tested at 10µM for β-arrestin 1 recruitment at δORs in a PathHunter assay. The red dot represents the hit compound (**1**). 10 µM SNC80 was utilized for normalization.

#### *2.2. Compound 1 Displays 10-Fold Selectivity over µOR and κOR 2.2. Compound* **1** *Displays 10-Fold Selectivity over µOR and κOR 2.2. Compound 1 Displays 10-Fold Selectivity over µOR and κOR*  Pharmacological characterization of compound **1** revealed that it had a micromolar

Pharmacological characterization of compound **1** revealed that it had a micromolar affinity (Figure 3A) and potency (Figure 3B) at the δOR, which was roughly 10-fold stronger than for the µOR and κOR (Table 1). Within the testable dose range (<100 µM) we were unable to detect any β**-**arrestin 2 recruitment for compound **1** at the µOR and κOR (Table 1, Figure 3C). At the highest concentration we were able to detect β**-**arrestin 1 and 2 recruitment at the δOR (Figure 3C), but we were unable to generate pEC50 or alpha values in these assays as we had not reached the maximum effect yet. Pharmacological characterization of compound **1** revealed that it had a micromolar affinity (Figure 3A) and potency (Figure 3B) at the δOR, which was roughly 10-fold stronger than for the µOR and κOR (Table 1). Within the testable dose range (<100 µM) we were unable to detect any β-arrestin 2 recruitment for compound **1** at the µOR and κOR (Table 1, Figure 3C). At the highest concentration we were able to detect β-arrestin 1 and 2 recruitment at the δOR (Figure 3C), but we were unable to generate pEC<sup>50</sup> or alpha values in these assays as we had not reached the maximum effect yet. affinity (Figure 3A) and potency (Figure 3B) at the δOR, which was roughly 10-fold stronger than for the µOR and κOR (Table 1). Within the testable dose range (<100 µM) we were unable to detect any β**-**arrestin 2 recruitment for compound **1** at the µOR and κOR (Table 1, Figure 3C). At the highest concentration we were able to detect β**-**arrestin 1 and 2 recruitment at the δOR (Figure 3C), but we were unable to generate pEC50 or alpha values in these assays as we had not reached the maximum effect yet.

cells. We tested ~5100 compounds and identified a single positive hit, that, at a 10 µM concentration, displayed ~50% β-arrestin 1 recruitment relative to SNC80 (Figure 2).

cells. We tested ~5100 compounds and identified a single positive hit, that, at a 10 µM concentration, displayed ~50% β-arrestin 1 recruitment relative to SNC80 (Figure 2).

**Figure 3.** Pharmacological characterization of compound **1**. **A.** Binding affinity for compound **1**. At δOR, µOR and κOR. **Figure 3.** Pharmacological characterization of compound **1**. **A.** Binding affinity for compound **1**. At δOR, µOR and κOR. **B.** Inhibition of forskolin induced cAMP by compound **1** in cells expressing δOR, µOR and κOR. **C.** β-arrestin recruitment **Figure 3.** Pharmacological characterization of compound **1**. (**A**). Binding affinity for compound **1**. At δOR, µOR and κOR. (**B**). Inhibition of forskolin induced cAMP by compound **1** in cells expressing δOR, µOR and κOR. (**C**). β-arrestin recruitment for compound **1** following stimulation of δOR, µOR and κOR.

for compound **1** following stimulation of δOR, µOR and κOR.

**B.** Inhibition of forskolin induced cAMP by compound **1** in cells expressing δOR, µOR and κOR. **C.** β-arrestin recruitment

for compound **1** following stimulation of δOR, µOR and κOR. **Table 1.** Pharmacological characterization of compound **1**. All assays were run in three or more **Table 1.** Pharmacological characterization of compound **1**. All assays were run in three or more independent trials. ND = not detected.


### *2.3. Compound* **1** *Derivatives Exhibit Lower δOR Potency*

The hit compound (**1**), *N*'-(2-hydroxy-3-methoxybenzylidene)-3-(2-thienyl)-1*H*-pyrazole-5-carbohydrazide, had a novel chemotype and in contrast to well-established δOR agonists Leu<sup>5</sup> -enkephalin, SNC80 and ADL5859 appears to lack a basic nitrogen. Next, we performed a structure activity relationship (SAR) by catalog using 14 analogs of compound **1** (Figure 4, Table 2) to investigate how compound **1** may bind to δOR and to possibly identify compounds with improved pharmacology. In our experience, potency for δOR agonism in the PathHunter β-arrestin assay is generally lower than for the cAMP assay [21]. Therefore, to assess if analogs of compound **1** displayed improved δOR potency we first characterized the compounds in the cAMP assay. We found that none of the purchased analogs had stronger potency for δOR activation than compound **1** (Table 2). *Molecules* **2021**, *26*, x FOR PEER REVIEW 5 of 15

**Figure 4.** Chemical structures of compound **1** and **14** analogs of compound **1**. **Figure 4.** Chemical structures of compound **1** and **14** analogs of compound **1**.

*2.4. Compound 1 Engages Amino Acid Residues That Form the Orthosteric Binding Pocket* 

Given the novelty of compound **1**′s scaffold, we wanted to model possible interactions of compound **1** at the δOR. We utilized the active-like crystal structure of δOR (PDB: 6PT3 [22]) to perform docking and molecular dynamics (MD) simulations in Schrödinger 2021-1. The crystal structure (6PT3) contains nine thermostabilizing mutations, three of which are near at the sodium binding pocket (N 902.45, D 952.50, N 1313.35) and near ECL2 in transmembrane helix 2 (TM2) (Q1052.60 and K1082.62). Subsequently, we reverted all nine mutations to the wild-type (WT) residues (see methods and suppl.). Our initial docking suggested that the thiophene ring of compound **1** occupies a hydrophobic pocket near the orthosteric site formed by W114ECL1, V1243.28, L1253.29, C198ECL2 where it forms ionic bonds with K1082.63 and hydrophobic interactions with W114ECL1 and C198ECL2 (Figure 5A). Additionally, compound 1 appeared to extend further into the orthosteric site where it was in proximity to and interacted with D1283.32, Y1293.33 and Y3087.42 (Figure 5B). To confirm

phobic pocket (Supplementary Figure S1). We then decided to further model the interactions of compound **1** at δOR using dynamic structures where we performed three independent all-atom MD simulations which showed a relatively stable pose for compound **1** where it interacts with residues in TM2, ECL1, TM3 and ECL2 (L200ECL2) and occasionally with residues in TM5 (K2145.40) and TM7 (Figure 5C, D, Supplementary Figures S2 and

S3).


**Table 2.** Potency (pIC50) and standard error (SEM) of analogs of compound **1** to inhibit cAMP signaling at δOR. The sigma catalog number for each compound is provided. All compounds were tested in three or more independent trials.

#### *2.4. Compound* **1** *Engages Amino Acid Residues That Form the Orthosteric Binding Pocket*

Given the novelty of compound **1** 0 s scaffold, we wanted to model possible interactions of compound **1** at the δOR. We utilized the active-like crystal structure of δOR (PDB: 6PT3 [22]) to perform docking and molecular dynamics (MD) simulations in Schrödinger 2021-1. The crystal structure (6PT3) contains nine thermostabilizing mutations, three of which are near at the sodium binding pocket (N 902.45, D 952.50, N 1313.35) and near ECL2 in transmembrane helix 2 (TM2) (Q1052.60 and K1082.62). Subsequently, we reverted all nine mutations to the wild-type (WT) residues (see methods and Supplementary Material). Our initial docking suggested that the thiophene ring of compound **1** occupies a hydrophobic pocket near the orthosteric site formed by W114ECL1, V1243.28, L1253.29, C198ECL2 where it forms ionic bonds with K1082.63 and hydrophobic interactions with W114ECL1 and C198ECL2 (Figure 5A). Additionally, compound **1** appeared to extend further into the orthosteric site where it was in proximity to and interacted with D1283.32, Y1293.33 and Y3087.42 (Figure 5B). To confirm the initial docked poses, we docked compound **1** into multiple potential binding sites generated using SiteMap and confirmed similar interactions with residues within the hydrophobic pocket (Supplementary Figure S1). We then decided to further model the interactions of compound **1** at δOR using dynamic structures where we performed three independent all-atom MD simulations which showed a relatively stable pose for compound **1** where it interacts with residues in TM2, ECL1, TM3 and ECL2 (L200ECL2) and occasionally with residues in TM5 (K2145.40) and TM7 (Figure 5C,D, Supplementary Figures S2 and S3).

**Figure 5.** Molecular Dynamic simulation of Compound **1** binding to the δOR**.** (**A**)**.** Compound **1** bound at δOR where its positioned within the hydrophobic pocket, a predicted allosteric site. (**B**)**.** Compound **1** interacts with residues forming the hydrophobic pocket as well as with residues deeper into the orthosteric site K1082.63, W114ECL1, L1253.29, D1283.32, Y1293.33, C198ECL2, L200ECL2 and K2145.40. (**C**)**.** A rolling average of 3 ns of the RMSD of compound **1** in a 300 ns MD simulation showing a relatively stable binding pose for compound **1**. (**D**)**.** Interaction fractions between compound **1** and the δOR in 3 different MD simulations. **Figure 5.** Molecular Dynamic simulation of Compound **1** binding to the δOR. (**A**). Compound **1** bound at δOR where its positioned within the hydrophobic pocket, a predicted allosteric site. (**B**). Compound **1** interacts with residues forming the hydrophobic pocket as well as with residues deeper into the orthosteric site K1082.63, W114ECL1, L1253.29, D1283.32, Y1293.33 , C198ECL2, L200ECL2 and K2145.40. (**C**). A rolling average of 3 ns of the RMSD of compound **1** in a 300 ns MD simulation showing a relatively stable binding pose for compound **1**. (**D**). Interaction fractions between compound **1** and the δOR in 3 different MD simulations.

#### *2.5. Compound* **1** *Can Occupy an Allosteric Space Alongside Leu-Enkephalin 2.5. Compound* **1** *Can Occupy an Allosteric Space alongside Leu-Enkephalin*

Our modeling suggests that compound **1** interacts with residues in TM2 and TM7, which have been previously reported to interact, potentially, with the positive allosteric modulator BMS 986187 [23]. At the orthosteric site, compound **1** forms water-mediated interactions, hydrogen bonds and hydrophobic interactions with D1283.32, Y1293.33 and H2786.62 residues which were reported to be involved in δOR activation [22]. Additionally, compound **1** interacts with W114ECL1 (π-π stacking), V1243.28, L1253.29, C198ECL2 where its thiophene moiety occupies a partially hydrophobic pocket that is adjacent to the orthosteric site (Figure 5A). These unique interactions which include amino acid residues in the orthosteric and the potential allosteric binding sites prompted us to model compound **1** in the presence of Leu-enkephalin using molecular dynamics (MD) simulations (Figure 6A, Supplementary Figure S4). Intriguingly, compound **1** appears to maintain a relatively stable orientation as shown by the relatively stable RMSD in three independent MD simulations whereas Leu-enkephalin undergoes more dramatic confirmational changes in the presence of compound **1** (Figure 6B–D). Specifically, the presence of compound **1** appears to disrupt the π-π interaction between Leu-enkephalin with W2846.58 where the phenyl group of Phe4 rotates away from W2846.58 (Figure 6C). We also observed an inward shift in ICL2 as well as conformational changes at the intracellular side of δOR in ICL2, TM5 Our modeling suggests that compound **1** interacts with residues in TM2 and TM7, which have been previously reported to interact, potentially, with the positive allosteric modulator BMS 986187 [23]. At the orthosteric site, compound **1** forms water-mediated interactions, hydrogen bonds and hydrophobic interactions with D1283.32, Y1293.33 and H2786.62 residues which were reported to be involved in δOR activation [22]. Additionally, compound **1** interacts with W114ECL1 (π-π stacking), V1243.28, L1253.29, C198ECL2 where its thiophene moiety occupies a partially hydrophobic pocket that is adjacent to the orthosteric site (Figure 5A). These unique interactions which include amino acid residues in the orthosteric and the potential allosteric binding sites prompted us to model compound **1** in the presence of Leu-enkephalin using molecular dynamics (MD) simulations (Figure 6A, Supplementary Figure S4). Intriguingly, compound **1** appears to maintain a relatively stable orientation as shown by the relatively stable RMSD in three independent MD simulations whereas Leu-enkephalin undergoes more dramatic confirmational changes in the presence of compound **1** (Figure 6B–D). Specifically, the presence of compound **1** appears to disrupt the π-π interaction between Leu-enkephalin with W2846.58 where the phenyl group of Phe<sup>4</sup> rotates away from W2846.58 (Figure 6C). We also observed an inward shift in ICL2 as well as conformational changes at the intracellular side of δOR in ICL2, TM5 and TM6 when compared to the thermostabilized crystal structure (Supplementary Figure S5).

S5).

and TM6 when compared to the thermostabilized crystal structure (Supplementary Figure

**Figure 6.** Molecular dynamic simulation of Compound **1** bound to the δOR in the presence of Leu-enkephalin. (**A**)**.** a representative binding pose for compound **1** in the presence of Leu-enkephalin obtained from a 300 ns MD simulation where compound **1** stably occupies the partially hydrophobic pocket. (**B**)**.** Leu-enkephalin forms H-bonds and water mediated interactions with K1082.63, D1283.32, R192ECL2, C198ECL2, H3017.35, C3037.37 and hydrophobic interactions with Y3087.42 whereas compound **1** mostly interacts with W114ECL1, L1253.29, C198ECL2 and L200ECL2 and K2145.40. (**C**)**.** Poses of Leu-enkephalin and compound **1** showing the first frame of a 300 ns MD simulation (Leu-enkephalin: light green, compound **1**: light pink, W284: cyan) aligned on the clustered poses of Leu-enkephalin and compound **1** (Leu-enkephalin: dark green, compound **1**: red, W2846.58: light grey). (**D**)**.** A rolling average of 3 ns of the RMSD of compound **1** in the presence of Leuenkephalin obtained from a 300 ns MD simulation showing a relatively stable pose for compound **1** whereas the disruption of Leu-enkephalin's interaction with W2846.58 causes a relatively large change in its RMSD. **Figure 6.** Molecular dynamic simulation of Compound **1** bound to the δOR in the presence of Leu-enkephalin. (**A**). a representative binding pose for compound **1** in the presence of Leu-enkephalin obtained from a 300 ns MD simulation where compound **1** stably occupies the partially hydrophobic pocket. (**B**). Leu-enkephalin forms H-bonds and water mediated interactions with K1082.63, D1283.32, R192ECL2, C198ECL2, H3017.35, C3037.37 and hydrophobic interactions with Y3087.42 whereas compound **1** mostly interacts with W114ECL1, L1253.29, C198ECL2 and L200ECL2 and K2145.40. (**C**). Poses of Leu-enkephalin and compound **1** showing the first frame of a 300 ns MD simulation (Leu-enkephalin: light green, compound **1**: light pink, W284: cyan) aligned on the clustered poses of Leu-enkephalin and compound **1** (Leu-enkephalin: dark green, compound **1**: red, W2846.58: light grey). (**D**). A rolling average of 3 ns of the RMSD of compound **1** in the presence of Leu-enkephalin obtained from a 300 ns MD simulation showing a relatively stable pose for compound **1** whereas the disruption of Leu-enkephalin's interaction with W2846.58 causes a relatively large change in its RMSD.

#### *2.6. Compound* **1** *Potentially Negatively Modulates Potency of Leu-Enkephalin through an Allosteric Mechanism 2.6. Compound* **1** *Potentially Negatively Modulates Potency of Leu-Enkephalin through an Allosteric Mechanism*

Given that our modelling efforts suggested binding poses in a slightly allosteric binding pocket, we next decided to measure to what degree compound **1** modulated the activity profile of leu-enkephalin in the cAMP glosensor assay. We noted an increase in baseline (or τβ) when Leu-enkephalin was co-incubated with increasing concentrations of compound **1** (Figure 7A–B), without observing a chance in Emax (β = 1). We observed a left-shift in Leu-enkephalin potency suggestive of a negative allosteric modulation that is affinity (or α) driven (Figure 7A–B). As such, compound **1** appears to act as a negative allosteric modulator (NAM)-agonist [24] in the cAMP glosensor assay. It is well known that, for example, irreversible antagonists by lowering the receptor reserve will right-shift the potency of an agonist [25]. Thus, the potency shift could also be driven by the decrease in Given that our modelling efforts suggested binding poses in a slightly allosteric binding pocket, we next decided to measure to what degree compound **1** modulated the activity profile of leu-enkephalin in the cAMP glosensor assay. We noted an increase in baseline (or τβ) when Leu-enkephalin was co-incubated with increasing concentrations of compound **1** (Figure 7A,B), without observing a chance in Emax (β = 1). We observed a left-shift in Leu-enkephalin potency suggestive of a negative allosteric modulation that is affinity (or α) driven (Figure 7A,B). As such, compound **1** appears to act as a negative allosteric modulator (NAM)-agonist [24] in the cAMP glosensor assay. It is well known that, for example, irreversible antagonists by lowering the receptor reserve will right-shift the potency of an agonist [25]. Thus, the potency shift could also be driven by the decrease in receptors available for Leu-enkephalin to bind to since radioligand binding indicates that compound **1** can bind and displace agonists (Figure 3) from the binding pocket.

*Molecules* **2021**, *26*, x FOR PEER REVIEW 8 of 15

**Figure 7.** Compound **1** acts as a negative allosteric modulator for leu-enkephalin potency in the cAMP glosensor assay. (**A**)**.** Dose-dependent inhibition of forskolin-mediated cAMP production by Leu-enkephalin (Leu-Enk) in the absence or presence of increasing concentrations of compound **1**. (**B**)**.** The decrease in Leu-enkephalin pIC50 is correlated with in-**Figure 7.** Compound **1** acts as a negative allosteric modulator for leu-enkephalin potency in the cAMP glosensor assay. (**A**). Dose-dependent inhibition of forskolin-mediated cAMP production by Leu-enkephalin (Leu-Enk) in the absence or presence of increasing concentrations of compound **1**. (**B**). The decrease in Leu-enkephalin pIC50 is correlated with increasing concentration of compound **1**. (**A**)**.** Dose-dependent inhibition of forskolin-mediated cAMP production by Leu-enkephalin (Leu-Enk) in the absence or presence of increasing concentrations of compound **1**. (**B**)**.** The decrease in Leu-enkephalin pIC50 is correlated with increasing concentration of compound **1**. **3. Discussion** 

#### **3. Discussion** Here we report on a novel δOR-selective agonist chemotype that was identified from

creasing concentration of compound **1**.

**3. Discussion**  Here we report on a novel δOR-selective agonist chemotype that was identified from a 5120-compound high-throughput screen of CNS-targeted chemical libraries. The scaffold lacks a basic protonated amine, which is generally considered a hallmark feature of opioid ligands, needed to form a stable salt-bridge with aspartate D3.32.[22] Using MolgpKa [26], the predicted pKa of the basic nitrogen in the pyrazole ring of compound **1** is 1.4, in sharp contrast with the pKa for protonated basic amines that is closer to physiological pH. A second interesting feature of compound **1** is the apparent negative allosteric modulation of the endogenous agonist Leu-enkephalin. Positive allosteric modulators (PAMs) have been identified for the opioid receptors, including the G-protein-biased δOR 'PAM-agonist' BMS 986187 (Figure 8) [24,27–29]. Cannabidiol and tetrahydrocannabinol have been proposed to be allosteric modulators of the δOR, specifically accelerating nal-Here we report on a novel δOR-selective agonist chemotype that was identified from a 5120-compound high-throughput screen of CNS-targeted chemical libraries. The scaffold lacks a basic protonated amine, which is generally considered a hallmark feature of opioid ligands, needed to form a stable salt-bridge with aspartate D3.32 [22]. Using MolgpKa [26], the predicted pKa of the basic nitrogen in the pyrazole ring of compound **1** is 1.4, in sharp contrast with the pKa for protonated basic amines that is closer to physiological pH. A second interesting feature of compound **1** is the apparent negative allosteric modulation of the endogenous agonist Leu-enkephalin. Positive allosteric modulators (PAMs) have been identified for the opioid receptors, including the G-protein-biased δOR 'PAM-agonist' BMS 986187 (Figure 8) [24,27–29]. Cannabidiol and tetrahydrocannabinol have been proposed to be allosteric modulators of the δOR, specifically accelerating naltrindole dissociation rate [30], however to our knowledge no NAM-agonist has previously been reported. a 5120-compound high-throughput screen of CNS-targeted chemical libraries. The scaffold lacks a basic protonated amine, which is generally considered a hallmark feature of opioid ligands, needed to form a stable salt-bridge with aspartate D3.32.[22] Using MolgpKa [26], the predicted pKa of the basic nitrogen in the pyrazole ring of compound **1** is 1.4, in sharp contrast with the pKa for protonated basic amines that is closer to physiological pH. A second interesting feature of compound **1** is the apparent negative allosteric modulation of the endogenous agonist Leu-enkephalin. Positive allosteric modulators (PAMs) have been identified for the opioid receptors, including the G-protein-biased δOR 'PAM-agonist' BMS 986187 (Figure 8) [24,27–29]. Cannabidiol and tetrahydrocannabinol have been proposed to be allosteric modulators of the δOR, specifically accelerating naltrindole dissociation rate [30], however to our knowledge no NAM-agonist has previously been reported.

receptors available for Leu-enkephalin to bind to since radioligand binding indicates that

receptors available for Leu-enkephalin to bind to since radioligand binding indicates that

compound **1** can bind and displace agonists (Figure 3) from the binding pocket.

compound **1** can bind and displace agonists (Figure 3) from the binding pocket.

**Figure 8.** Chemical structures of the allosteric G-protein-biased δOR modulator BMS 986187, the G-protein-biased µOR agonist PZM21 and the G-protein-biased κOR agonist compound 81. **Figure 8.** Chemical structures of the allosteric G-protein-biased δOR modulator BMS 986187, the G-protein-biased µOR agonist PZM21 and the G-protein-biased κOR agonist compound 81.

**Figure 8.** Chemical structures of the allosteric G-protein-biased δOR modulator BMS 986187, the G-protein-biased µOR agonist PZM21 and the G-protein-biased κOR agonist compound 81. The PAM-agonist BMS 986187 does not possess an ionizable group and thus resembles our compound **1**, which also lacks the protonated amine commonly present in opioids. However, comparisons between the suggested mode of binding of BMS 986187 and compound **1** at δOR show distinct interactions that could account for the differences The PAM-agonist BMS 986187 does not possess an ionizable group and thus resembles our compound **1**, which also lacks the protonated amine commonly present in opioids. However, comparisons between the suggested mode of binding of BMS 986187 and compound **1** at δOR show distinct interactions that could account for the differences The PAM-agonist BMS 986187 does not possess an ionizable group and thus resembles our compound **<sup>1</sup>**, which also lacks the protonated amine commonly present in opioids.However, comparisons between the suggested mode of binding of BMS 986187 and compound **1** at δOR show distinct interactions that could account for the differences in their mode of action. Notably, in the presence of the endogeneous peptide Leu-enkephalin, compound **1** appears to occupy a partially hydrophobic pocket adjacent to the orthosteric site which allows compound **1** to interact with residues in ECL1 (W114ECL1), ECL2 (C198, L200) and TM7, whereas BMS 986187 is reported to interact with residues in TM2 and TM7 in its lowest relative free-energy state in the presence of SNC80 [23]. Moreover, most of the residues reported to interact with BMS 986187 were shown to interact

with residues in the active-like structures of δOR that constitute the orthosteric binding site [22,23]. These differences in the interactions could account for the distinct pharmacology of compound **1** and BMS 986187. Intriguingly, in the presence or absence of Leu-enkephalin, compound **1** maintains a relatively stable orientation that enables it to retain its hydrophobic and water-mediated interactions at the thiophene and pyrazole rings, respectively (Figure 6A,D). The presence of Leu-enkephalin, however, appears to disrupt the water-mediated interactions between compound **1** and orthosteric residues D1283.32 and Y1293.33 (Figure 6B) and changes the number of hydrogen donors or acceptors in compound **1** (Supplementary Figure S5). On the other hand, the presence of compound **1** disrupts the hydrophobic interaction between Phe<sup>4</sup> and δOR by causing the phenyl group of Leu-enkephalin to rotate away from the side chain of W2846.58 (Figure 6C). Additionally, H-bond and water-mediated interactions between Leu-enkephalin and R192ECL2 appear to move ECL2 toward Leu-enkephalin which could open a cryptic binding site similar to a previously reported allosteric binding site in the angiotensin II (AngII) type 1 receptor [31] (Figure 6B, Supplementary Figure S6). As such, we predict that compound **1** may induce NAM activity by either destabilizing Leu-enkephalin or by playing an analogous role to BMS 986187 where it stabilizes the Na<sup>+</sup> binding at δOR which increases the likelihood of receptor deactivation. It should be noted that comparisons between the binding modes of compound **1** and BMS 986187 at the δOR are limited due to the differences in the crystal structures used for modeling (agonist-bound vs antagonist-bound, respectively), chemotype differences between compound **1** and BMS 986187, the modeling method utilized, and the co-simulated ligand. Hence, future studies should examine the binding of compound **1** at the δOR in the presence of small molecule agonists and the implementation of enhanced sampling methods to model its interactions in the presence or absence of δOR agonists.

After identifying compound **1** in our screen, we had hoped to find analogs with higher potency, through a SAR by catalog. However, none of the purchased analogs displayed improved potency for the δOR. Our choice of catalog analogs was driven primarily by price and availability and much less guided by intelligent design. As a result of this strategy, we were only able to explore minor derivatization at the thiopene moiety and the 2-hydroxy-3 methoxybenzene moiety. Therefore, it is possible that compound **1** may still be improved on, for example, by altering or substituting on the pyrazole group, or by adding hydrogen bond-forming and/or accepting groups on the thiophene moiety.

Another feature we set out to find in our screen was a δOR agonist that underrecruited β-arrestin. Much effort has been devoted to identify opioids that display a preference to recruit and activate G-proteins relative to β-arrestin recruitment [17,21,32–34]. Our screen was designed with the purpose of finding molecules that underrecruit β-arrestin, but that are not G-protein-selective i.e., that entirely avoid β-arrestin recruitment and as such, compound **1** does still recruit β-arrestin. Surprisingly, we noted an unusual steep increase in β-arrestin recruitment at the δOR when stimulated with 100 µM compound **1**, such that we were unable to accurately predict an Emax. The sharp rise in β-arrestin recruitment at 100 µM did not appear to be a pan- interference assay effect, as we did not observe a similar response in our µOR and κOR PathHunter cell lines (Figure 3C). The mechanism or implication of compound **1** 0 s β-arrestin recruitment at 100 µM will require further investigation.

With increased availability of apo-state, antagonist-bound and agonist bound opioid structures, drug screening has moved away from screening physical libraries to screening virtual libraries. A computational model created using the crystal structure of an antagonist bound κOR [35] supported a virtual chemical library screen of 5 million molecules at κOR resulting in the identification of compound 81 (Figure 8), which is a G-protein-biased agonist with an 0.16 µM affinity and 0.53 µM potency at the κOR [36]. A virtual screen of 3 million molecules docked at a computational model of the µOR based on the antagonistbound µOR crystal structure [37] resulted in the identification of a hit with 2.5 µM affinity at the µOR, which through an analog screen was improved to a lead compound with a 42 nM affinity and G protein bias. Further structure guided optimization of the lead compound

resulted in the design of PZM21 (Figure 8), a G-protein-biased µOR-selective agonist with 1 nM affinity and unique chemotype [32]. Recent advances now allow for virtual screening of libraries containing more than a billion compounds [38,39]. While it is undeniable that large virtual screens can identify completely novel chemical matter, the ability to discover molecules with novel pharmacology may be more limited or biased by the type of structure (e.g., an orthosteric agonist-bound structure stabilized by a heterotrimeric G-protein or nanobody-mimic in a single active conformation) used for docking. Thus, in conclusion, our results highlight a current persisting value of chemical library screens in identifying molecules with unique binding modes and pharmacology.

#### **4. Materials and Methods**

#### *4.1. Chemicals*

Leu<sup>5</sup> -enkephalin, compounds **1**–**15** and forskolin were purchased from Sigma-Aldrich (St. Louis, MO USA). [D-Ala<sup>2</sup> , N-MePhe<sup>4</sup> , Gly-ol] enkephalin (DAMGO), SNC80 and U50,488 were purchased from Tocris Bioscience (Minneapolis, MN, USA). Radiolabels were from Perkin Elmer (Waltham, MA, USA).

#### *4.2. Library Screen*

In consultation with the Chemical Genomics Facility within the Purdue Institute for Drug Discovery, we screened sixteen 384-well plates that were part of CNS-targeted drug libraries. Specifically, we screened eleven plates part of a CNS-Chemdiv library, three plates part of a Chembridge ion channel library, and two plates part of a CNS-TimTec library. Each plate contained 320 compounds and four spare columns that were utilized to run positive (10 µM SNC80, 32 wells) and negative controls (0.02% DMSO, 32 wells), which were used to calculate Z-factors (average: Z'= 0.53, hit plate: Z'= 0.58) and normalize the data across plates. Using an Echo 525 acoustic liquid handler (Labcyte, San Jose, CA, USA), depending on the stock concentration (1, 10 or 20 mM) of the library plate 5, 10 or 100 nL of each compound was transferred from the library plate to the assay plate, the final concentration of each library compound was 10 µM.

#### *4.3. Radioligand Binding Assay*

Radioligand binding was performed as previously described [40,41]. For the binding assay 50 µL of a dilution series of peptide was added to 50 µL of 3.3 nM [3H]DPDPE (*K*<sup>d</sup> = 3.87 nM) or 2.35 nM of [3H]DAMGO (*K*<sup>d</sup> = 1.07 nM) or 0.8 nM of [3H]U69,593 (*K*<sup>d</sup> = 1.2 nM) in a clear 96 well plate. Next, 100 µL of membrane suspension containing 7 µg protein was added to the agonist wells and incubated for 90 min at room temperature. The reaction mixture was then filtered over a GF-B filter plate (Perkin Elmer) followed by four quick washes with ice-cold 50 mM Tris HCl. The plate was dried overnight, after which 50 µL scintillation fluid (Ultimagold uLLT) was added and radioactivity was counted on a Packard TopCount NXT scintillation counter. All working solutions were prepared in a radioligand assay buffer containing 50 mM Tris HCl, 10 mM MgCl2, and 1 mM ethylenediaminetetraacetic acid at pH 7.4.

#### *4.4. Cellular Signaling Assays*

cAMP inhibition and β-arrestin 1 and 2 recruitment assays were performed as previously described [18]. In brief, for cAMP inhibition assays HEK 293 (Life Technologies, Grand Island, NY, USA) cells were transiently transfected in a 1:3 ratio with FLAG-mouse δOR, or HA-mouse µOR and pGloSensor22F-cAMP plasmids (Promega, Madison, WI, USA) using Xtremegene9 (Sigma). Two days post-transfection cells (20,000 cells/well, 7.5 µL) were seeded in low volume Greiner 384-well plates (#82051-458, VWR, Batavia, IL, USA) and were incubated with Glosensor reagent (Promega, 7.5 µL, 2% final concentration) for 90 min at room temperature. Cells were stimulated with 5 µL drug solution for 20 min at room temperature prior to stimulation with 5 µL forskolin (final concentration 30 µM), for an additional 15 min at room temperature. For β-arrestin recruitment assays, CHO-

human µOR PathHunter β-arrestin 2 cells, CHO-human δOR PathHunter β-arrestin 2 cells, U2OS κOR PathHunter β-arrestin 2 cells or U2OS PathHunter β-arrestin 1 cells (DiscoverX, Fremont, CA, USA) were plated (2500 cells/well, 10 µL) one day prior to stimulation with 2.5 µL or 5–100 nL (in the screen) drug solution for 90 min at 37 ◦C/5% CO2, after which cells were incubated with 6 µL cell PathHunter assay buffer (DiscoverX) for 60 min at room temperature as per the manufacturer's protocol. Luminescence for each of these assays was measured using a FlexStation3 plate reader (Molecular Devices, Sunnyvale, CA, USA). As positive control we utilized Leu<sup>5</sup> -enkephalin or SNC80 (in the screen) for δOR, [D-Ala<sup>2</sup> , N-MePhe<sup>4</sup> , Gly-ol] enkephalin (DAMGO) for µOR and U50,488 for κOR.

#### *4.5. Assessment of Allosteric Modulation*

We ran log-step concentration response curves for Leu-enkephalin (10 µM–1 pM) in the presence of 0, 0.1, 0.3, 1, 3, or10 µM compound **1** in the δOR glosensor cAMP assay.

#### *4.6. Data and Statistical Analysis*

All data are presented as means ± standard error of the mean, and analysis was performed using GraphPad Prism 8 software (GraphPad Software, La Jolla, CA, USA). For in vitro assays, nonlinear regression was conducted to determine pIC<sup>50</sup> (cAMP) or pEC<sup>50</sup> (β-arrestin recruitment). Technical replicates were used to ensure the reliability of single values, specifically each data point for binding and β-arrestin recruitment was run in duplicate, and for the cAMP assay in triplicate. The averages of each independent run were counted as a single experiment and combined to provide a composite curve in favor of providing a 'representative' curve.

#### *4.7. Receptor and Ligand Preparation for Molecular Modeling*

The crystal structure of the active-like δOR (PDB: 6PT3) bound to small molecule agonist, DPI-287, was obtained from the Protein Data Bank (PDB) [22]. Molecular modeling was performed via Maestro (Schrödinger suite 2021-1, Schrödinger, Inc., New York, NY, USA). The *Protein Preparation Wizard* was used to prepare the structures before docking. The crystal structure was preprocessed to cap the N-terminus, remove the BRIL tag, membrane lipids and other crystal waters or ions not involved in mediating receptor-ligand interaction. Preliminary modeling and energy minimization of the thermostabilized receptor [22] and the WT-reverted receptor (data not shown) showed the feasibility of performing MD simulations using a truncated version of the WT receptor (residues 41–289) where all 9-thermostabilizing mutations were reverted to the WT (Supplementary Figure S5). Missing loops and side chains in the crystal structure were modeled using *Prime* within Schrödinger [42–44]. H-bond were assigned using the PROPKA algorithm [45,46]. All-atom MD simulations were performed on the modeled receptor using Desmond (Schrödinger, Inc.) implementing the OPLS4 force field. Compound **1** was prepared using LigPrep where the ionization states were assigned using Epik at pH 7.0 ± 2.0 [47,48]. Docking grids were generated for a representative structure from the MD simulations using *Receptor Grid Generation* in Schrödinger Release 2021-1 (Schrödinger, Inc.) using default parameters.

#### *4.8. Ligand Docking Using Glide*

Compound **1** and a set of known δOR ligands (Table 1) were docked into a model WT δOR using Glide (Table 2) [49–51]. Further structural optimization was needed to improve the docking accuracy of the model WT δOR (Supplemental Table S3). Additionally, given the novelty of the compound **1** 0 s chemotype, δOR ligands were docked into several models with predicted binding sites that were generated using SiteMap [52,53]. The best model was selected for further production MD simulations. Standard precision (SP) scoring function in Schrodinger 2021-1 was used for the initial docking of the molecules. The extra precision (XP) scoring function was then to further refine the docked poses. Postdocking energy minimization was performed for the top 50 poses of each small molecule, after which top 10 poses were visually inspected. The top 50 docked poses were also

scored using Prime MM-GBSA scoring [54]. The best pose (based on docking, visual inspection and MM-GBSA score) was selected for subsequent production MD simulations (Supplemental Tables S4–S6).

### *4.9. Molecular Dynamics Simulations of Compound* **1** *at δOR*

Production molecular dynamics simulations (MD) were performed in Desmond as reported previously [55]. Ligand-receptor complexes were embedded in a POPC membrane contained in a SPC-solvated orthorhombic box while maintaining a 10 Å distance from box boundaries. Na<sup>+</sup> and Cl<sup>−</sup> ions at a concentration of 0.15 M were added to mimic biological conditions using System Builder in Schrodinger 2021-1. The default membrane relaxation protocol in Desmond was used for membrane relaxation. Then a constant pressure and temperature (NPT) equilibration run was performed for 100 ns. The RESPA integrator with a 2 fs integration step for bonded interactions and a 6 fs step for non-bonded interactions. The Nosé-Hoover thermostat (and Martyna-Tobias-Klein barostat with semi-isotropic coupling to maintain temperature at 300 K and pressure at 1 bar. For the production MD simulations, three independent 200 ns NPT simulations were carried out for compound **1** in complex with modeled δOR or compound **1** and Leu-enkephalin in complex with modeled δOR. Each trajectory was assembled into 10 clusters using the trajectory clustering protocol implemented in Desmond. The top five clusters with the most interacting members were further assessed using Prime MM-GBSA (Supplemental Tables S7 and S8). The top poses were further inspected and used for analyses and figures presented here.

**Supplementary Materials:** The following are available, Figure S1: Binding sites within the δOR structure generated using SiteMap, Figure S2: *Cα* RMSD of δOR and compound **1** obtained from 3 independent MD simulations with varying trajectory time lengths and starting points, Figure S3: Receptor and ligand RMSD across several MD simulations, Figure S4: Summary of key δOR amino acid interactions with compound **1** and Leu-Enkephalin in the presence of compound **1**, Figure S5: Pharmacophore mapping analysis using the receptor-ligand complex. Figure S6: Comparison of the thermostabilized and simulated wild-type agonist-bound δOR structures. Table S1: Smiles of δOR agonists and antagonists used to validate the initial docking models, Table S2: Docking and glide scores for known δOR agonists and antagonists used to validate the initial docking model before structural optimization of the model δOR, Table S3: Docking and glide scores for known δOR agonists and antagonists used to validate the initial docking model after structural optimization, Table S4: Compound **1** docking scores using the SP scoring function. Top 10 poses were rescored XP scoring function, Table S5: Top 15 Leu-enkephalin poses docked into model δOR in the presence of compound **1**, Table S6: Rescoring of top 50 poses of Leu-enkephalin docked into model δOR using Prime MM-GBSA, Table S7: MM-GBSA scoring of top 5 clusters from a 300 ns MD simulation for Leu-enkephalin and compound **1**, Table S8: MM-GBSA scoring of top 5 clusters from a 300 ns MD simulation for compound **1**.

**Author Contributions:** Conceptualization, R.M.v.R. and Y.J.M.; methodology, L.C., Y.J.M.; validation, Y.J.M. and H.S.; formal analysis, R.M.v.R., Y.J.M., H.S. and R.J.C.; investigation, R.M.v.R., Y.J.M., H.S., R.J.C., K.L.M., A.M.G. and B.R.C.; writing—original draft preparation, R.M.v.R. and Y.J.M.; writing—review and editing, R.M.v.R. and Y.J.M.; visualization, R.M.v.R. and Y.J.M.; supervision, R.M.v.R.; project administration, R.M.v.R.; funding acquisition, R.M.v.R. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by is supported by funds provided by the National Institute on Alcohol Abuse and Alcoholism (AA025368, AA026949, AA026675) and Drug Abuse (DA045897) of the National Institutes of Health, the Purdue Institute for Drug Discovery and the Department of Medicinal Chemistry and Molecular Pharmacology.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Not applicable.

**Conflicts of Interest:** The authors declare no conflict of interest.

**Sample Availability:** The compounds are available from commercial vendors e.g., Sigma.

### **References**


#### *Article* **Synthesis, Pharmacological Evaluation, and Computational Studies of Cyclic Opioid Peptidomimetics Containing** β **3 -Lysine**

**Karol Wtorek <sup>1</sup> , Piotr F. J. Lipi ´nski <sup>2</sup> , Anna Adamska-Bartłomiejczyk <sup>1</sup> , Justyna Piekielna-Ciesielska <sup>1</sup> , Jarosław Sukiennik <sup>3</sup> , Alicja Kluczyk <sup>4</sup> and Anna Janecka 1,\***


**Abstract:** Our formerly described pentapeptide opioid analog Tyr-c[D-Lys-Phe-Phe-Asp]NH<sup>2</sup> (designated **RP-170**), showing high affinity for the mu (MOR) and kappa (KOR) opioid receptors, was much more stable than endomorphine-2 (EM-2) in the rat brain homogenate and displayed remarkable antinociceptive activity after central (intracerebroventricular) and peripheral (intravenous) administration. In this report, we describe the further modification of this analog, which includes the incorporation of a β 3 -amino acid, (*R*)- and (*S*)-β 3 -Lys, instead of D-Lys in position 2. The influence of such replacement on the biological properties of the obtained analogs, Tyr-c[(*R*)-β 3 -Lys-Phe-Phe-Asp]NH<sup>2</sup> (**RP-171**) and Tyr-c[(*S*)-β 3 -Lys-Phe-Phe-Asp]NH<sup>2</sup> , (**RP-172**), was investigated in vitro. Receptor radiolabeled displacement and functional calcium mobilization assays were performed to measure binding affinity and receptor activation of the new analogs. The obtained data revealed that only one of the diastereoisomeric peptides, **RP-171**, was able to selectively bind and activate MOR. Molecular modeling (docking and molecular dynamics (MD) simulations) suggests that both compounds should be accommodated in the MOR binding site. However, in the case of the inactive isomer **RP-172**, fewer hydrogen bonds, as well as instability of the canonical ionic interaction to Asp147, could explain its very low MOR affinity.

**Keywords:** opioid receptors; β-amino acids; peptide synthesis; receptor binding studies; functional assay

## **1. Introduction**

Among the three opioid receptors, mu (MOR), delta (DOR), and kappa (KOR), MOR plays the most important role in the modulation of pain signals and, therefore, is an important target in medicinal chemistry and drug development [1]. The two endogenous compounds activating MOR are endomorphin-1 (EM-1, Tyr-Pro-Trp-Phe-NH2) and endomorphin-2 (EM-2, Tyr-Pro-Phe-Phe-NH2) [2]. Over the years, numerous chemical modifications of these ligands have been reported in order to provide specific information on their structure–activity relationship and to find drug candidates with improved therapeutic properties [3–5]. Among various modifications of opioid peptides, cyclization of their linear structures was used to restrict flexibility and to obtain better-defined conformations, allowing for the identification of receptor binding sites [6–9].

Endomorphins are very short peptides lacking reactive side chain groups, which makes their cyclization difficult. One of the structural elements considered essential for their binding to MOR is the free cationic amino group of Tyr<sup>1</sup> [10–12], and this feature does not encourage head-to-tail cyclization. In order to obtain cyclic analogs based on the structure of

**Citation:** Wtorek, K.; Lipi ´nski, P.F.J.; Adamska-Bartłomiejczyk, A.; Piekielna-Ciesielska, J.; Sukiennik, J.; Kluczyk, A.; Janecka, A. Synthesis, Pharmacological Evaluation, and Computational Studies of Cyclic Opioid Peptidomimetics Containing β<sup>3</sup> -Lysine. *Molecules* **2022**, *27*, 151. https://doi.org/10.3390/ molecules27010151

Academic Editors: Mariana Spetea and Richard M. van Rijn

Received: 13 November 2021 Accepted: 25 December 2021 Published: 28 December 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

EM-2 but still to preserve the free N-terminal amino group, we introduced into the sequence of EM-2 additional amino acids with functionalized side chains. A pentapeptide analog Tyrc[D-Lys-Phe-Phe-Asp]NH<sup>2</sup> (designated **RP-170**), in which cyclization was achieved through the amide bond between D-Lys and Asp side chains, displayed high affinity for MOR, was much more stable than EM-2 in the rat brain homogenate and showed remarkable antinociceptive activity after central (i.c.v.) and peripheral (i.v.) administration [13]. The presence of a D-amino acid in position 2 (as in opioid peptides isolated from amphibian skin) was shown to enforce a different conformation of a peptide, greatly improving MOR binding as compared with Tyr-c[Lys-Phe-Phe-Asp]NH<sup>2</sup> [14]. Molecular docking studies of **RP-170** revealed that the amino group of Tyr<sup>1</sup> provided ionic interactions with Asp<sup>147</sup> residue in the transmembrane helice TM III of the receptor, while Asp amide effectively interacted with Asp<sup>216</sup> and Cys<sup>217</sup> belonging to the extracellular loop EL II. The presence of a Lys residue allowed for the formation of another strong interaction between Asp<sup>147</sup> and Lys-NH [15]. sequence of EM-2 additional amino acids with functionalized side chains. A pentapeptide analog Tyr-c[D-Lys-Phe-Phe-Asp]NH2 (designated **RP-170**), in which cyclization was achieved through the amide bond between D-Lys and Asp side chains, displayed high affinity for MOR, was much more stable than EM-2 in the rat brain homogenate and showed remarkable antinociceptive activity after central (i.c.v.) and peripheral (i.v.) administration [13]. The presence of a D-amino acid in position 2 (as in opioid peptides isolated from amphibian skin) was shown to enforce a different conformation of a peptide, greatly improving MOR binding as compared with Tyr-c[Lys-Phe-Phe-Asp]NH2 [14]. Molecular docking studies of **RP-170** revealed that the amino group of Tyr1 provided ionic interactions with Asp147 residue in the transmembrane helice TM III of the receptor, while Asp amide effectively interacted with Asp216 and Cys217 belonging to the extracellular loop EL II. The presence of a Lys residue allowed for the formation of another strong interaction between Asp147 and Lys-NH [15]. Further modifications of **RP-170** produced analogs with different opioid receptor

Endomorphins are very short peptides lacking reactive side chain groups, which makes their cyclization difficult. One of the structural elements considered essential for their binding to MOR is the free cationic amino group of Tyr1 [10–12], and this feature does not encourage head-to-tail cyclization. In order to obtain cyclic analogs based on the structure of EM-2 but still to preserve the free N-terminal amino group, we introduced into the

*Molecules* **2022**, *27*, x FOR PEER REVIEW 2 of 15

Further modifications of **RP-170** produced analogs with different opioid receptor preferences. Introduction of Dmt instead of Tyr<sup>1</sup> increased cyclopeptide affinity to MOR [16]. The reduction in the ring size increased MOR selectivity [17]. Substitution of the Phe residues by amino acids fluorinated in the aromatic ring (4-F-Phe, 2,4-diF-Phe, 4-CF3Phe) produced either high-affinity MOR/KOR agonists, non-selective MOR/DOR/KOR agonists, or selective KOR agonists [18], indicating that even small modifications in the side chains can completely change their orientation in the receptor cavity. preferences. Introduction of Dmt instead of Tyr1 increased cyclopeptide affinity to MOR [16]. The reduction in the ring size increased MOR selectivity [17]. Substitution of the Phe residues by amino acids fluorinated in the aromatic ring (4-F-Phe, 2,4-diF-Phe, 4-CF3Phe) produced either high-affinity MOR/KOR agonists, non-selective MOR/DOR/KOR agonists, or selective KOR agonists [18], indicating that even small modifications in the side chains can completely change their orientation in the receptor cavity. In the present study, we investigated the influence of a β-amino acid on the biological

In the present study, we investigated the influence of a β-amino acid on the biological properties of **RP-170**. D-Lys was replaced by (*R*)- or (*S*)-β 3 -Lys, obtained by homologation of D- or L-ornitine (Orn). This modification produced compounds isomeric to **RP-170** with the same size of the macrocyclic ring (17-membered), as in the parent peptide. Opioid receptor binding and activation were studied, and the obtained results were rationalized by molecular docking and molecular dynamics (MD) simulations. properties of **RP-170**. D-Lys was replaced by (*R*)- or (*S*)-β3-Lys, obtained by homologation of D- or L-ornitine (Orn). This modification produced compounds isomeric to **RP-170** with the same size of the macrocyclic ring (17-membered), as in the parent peptide. Opioid receptor binding and activation were studied, and the obtained results were rationalized by molecular docking and molecular dynamics (MD) simulations.

#### **2. Results 2. Results**

#### *2.1. Synthesis of Protected (R)- and (S)-β 3 -Lys 2.1. Synthesis of Protected (R)- and (S)-β3-Lys*

(R)- and (S)-Fmoc-β 3 -Lys (Mtt), which are not available commercially, were obtained by homologation of D- and L-Orn, respectively, according to the general procedure [19]. The synthetic protocol is outlined in Scheme 1. (R)- and (S)-Fmoc-β3-Lys (Mtt), which are not available commercially, were obtained by homologation of D- and L-Orn, respectively, according to the general procedure [19]. The synthetic protocol is outlined in Scheme 1.

**Scheme 1.** Synthesis of Fmoc-(R)- and (S)-β3-Lys(Mtt)*.*  **Scheme 1.** Synthesis of Fmoc-(R)- and (S)-β 3 -Lys(Mtt).

#### *2.2. Synthesis of Cyclopeptides 2.2. Synthesis of Cyclopeptides*

Cyclopeptides containing a β-amino acid, Tyr-c[(R)-β 3 -Lys-Phe-Phe-Asp]NH<sup>2</sup> (**RP-171**) and Tyr-c[(S)-β 3 -Lys-Phe-Phe-Asp]NH<sup>2</sup> (**RP-172**) (Figure 1) were synthesized on the solid support, using Fmoc/t-Bu strategy, with the hyper-acid labile groups (Mtt and O-2- PhiPr) for the selective protection of amine/carboxyl side chains of (R)- and (S)-β 3 -Lys and Asp, engaged in cyclization. After deprotection of the functionalized side chains, the linear

sequences were cyclized through amide bond formation. Final products were obtained with a purity greater than 95%, as assessed by semi-preparative RP-HPLC. The detailed analytical data of the synthesized peptides are provided in the Supplementary Materials (Table S1, Figures S2 and S3). ear sequences were cyclized through amide bond formation. Final products were obtained with a purity greater than 95%, as assessed by semi-preparative RP-HPLC. The detailed analytical data of the synthesized peptides are provided in the Supplementary Materials (Table S1, Figures S2 and S3).

Cyclopeptides containing a β-amino acid, Tyr-c[(R)-β3-Lys-Phe-Phe-Asp]NH2 (**RP-171**) and Tyr-c[(S)-β3-Lys-Phe-Phe-Asp]NH2 (**RP-172**) (Figure 1) were synthesized on the solid support, using Fmoc/t-Bu strategy, with the hyper-acid labile groups (Mtt and O-2- PhiPr) for the selective protection of amine/carboxyl side chains of (R)- and (S)-β3-Lys and Asp, engaged in cyclization. After deprotection of the functionalized side chains, the lin-

*Molecules* **2022**, *27*, x FOR PEER REVIEW 3 of 15

**Figure 1***.* Chemical structure of **RP-171** and **RP-172**. **Figure 1.** Chemical structure of **RP-171** and **RP-172**.

#### *2.3. LC-MS, LC-MSn, and Quantum Chemical Calculation Studies 2.3. LC-MS, LC-MS<sup>n</sup> , and Quantum Chemical Calculation Studies*

During the routine LC-MS analysis of analogs **RP-171** and **RP-172,** we noticed a distinct difference in retention times and MSn patterns for these diastereoisomeric peptides. To confirm our observation, we subjected a mixture of these peptides to LC-MS and MSn experiments. The HPLC analysis in reversed-phase mode revealed that the isomeric peptides separate easily, using both C18 column (Aeris Peptide) and biphenyl column (Kinetex Biphenyl), known for additional π-π interactions [20], with a nearly 0.5 min retention time difference in both cases in a 10 min gradient run from 5 to 80% acetonitrile in water (Figures 2 and S4–S6). Such a difference in retention time suggests altered interactions with the stationary phase, probably due to the shape of the molecules. It is interesting that the elution order from the biphenyl column was the same as from the C18 column. To assign the order of isomeric peptides in the LC-MS experiment on the **RP-171** and During the routine LC-MS analysis of analogs **RP-171** and **RP-172**, we noticed a distinct difference in retention times and MS<sup>n</sup> patterns for these diastereoisomeric peptides. To confirm our observation, we subjected a mixture of these peptides to LC-MS and MS<sup>n</sup> experiments. The HPLC analysis in reversed-phase mode revealed that the isomeric peptides separate easily, using both C<sup>18</sup> column (Aeris Peptide) and biphenyl column (Kinetex Biphenyl), known for additional π-π interactions [20], with a nearly 0.5 min retention time difference in both cases in a 10 min gradient run from 5 to 80% acetonitrile in water (Figure 2 and Figures S4–S6). Such a difference in retention time suggests altered interactions with the stationary phase, probably due to the shape of the molecules. It is interesting that the elution order from the biphenyl column was the same as from the C<sup>18</sup> column.

**RP-172** peptide mixture (Figure 2), we used retention times obtained during analysis of pure peptides, supported by MSn spectra. As expected, the MS spectra of peptides **RP-171** and **RP-172** were identical (panels **RP-171** MS and **RP-172** MS), and the difference in their To assign the order of isomeric peptides in the LC-MS experiment on the **RP-171** and **RP-172** peptide mixture (Figure 2), we used retention times obtained during analysis of pure peptides, supported by MS<sup>n</sup> spectra. As expected, the MS spectra of peptides **RP-171** and **RP-172** were identical (panels **RP-171** MS and **RP-172** MS), and the difference in their collision-induced dissociation (MS<sup>2</sup> , panels **RP-171** MS/MS, and **RP-172** MS/MS) was related to an intensity of 586 *m*/*z* fragment ion. To find a more reliable distinction, the MS<sup>3</sup> spectra were obtained for the precursors 586 from MS<sup>2</sup> spectra (panels **RP-171** MS/MS/MS and **RP-172** MS/MS/MS, MS<sup>3</sup> discussion in the Supplementary Materials).

**Figure 2.** LC-MSn analysis of peptides **RP-171** and **RP-172**. Extracted ion chromatogram (XIC) showing two peaks for the mixture of peptides (top panel): MS panels, all spectra recorded for the retention time of indicated peaks; MS spectra, MS/MS spectra for the precursor ions *m*/*z* 700.34, and MS/MS/MS spectra for the precursor ions *m*/*z* 586.30. **Figure 2.** LC-MS<sup>n</sup> analysis of peptides **RP-171** and **RP-172**. Extracted ion chromatogram (XIC) showing two peaks for the mixture of peptides (top panel): MS panels, all spectra recorded for the retention time of indicated peaks; MS spectra, MS/MS spectra for the precursor ions *m*/*z* 700.34, and MS/MS/MS spectra for the precursor ions *m*/*z* 586.30.

collision-induced dissociation (MS2, panels **RP-171** MS/MS, and **RP-172** MS/MS) was related to an intensity of 586 *m*/*z* fragment ion. To find a more reliable distinction, the MS3 spectra were obtained for the precursors 586 from MS2 spectra (panels **RP-171** MS/MS/MS

and **RP-172** MS/MS/MS, MS3 discussion in the Supplementary Materials).

The fragment ions observed in the MS/MS spectra were typical for peptide amides (consecutive loss of ammonia, 683 *m*/*z*, and carbon monoxide 655 *m*/*z*), whereas the 586 The fragment ions observed in the MS/MS spectra were typical for peptide amides (consecutive loss of ammonia, 683 *m*/*z*, and carbon monoxide 655 *m*/*z*), whereas the 586 ion resulted from ring-opening and removal of the Asp residue. The difference in intensity of the 586 *m*/*z* ions in panels **RP-171** MS/MS and **RP-172** MS/MS suggests that the fragmentation of peptide **RP-171** occurs easier than in the case of **RP-172**, suggesting that peptide **RP-172** containing (S)-β 3 -Lys is more stable.

This observation corresponds to the results of quantum chemical calculations (performed with Gaussian09 [21], Table S2) for both isomers. The lowest-lying (at the B3LYP/6-31G(d,p) level) gas-phase conformer of the [(S)-β 3 -Lys<sup>2</sup> ]- analog is more stable by 3.5 kcal/mol (∆G298) than the lowest-lying conformer of the [(R)-β 3 - Lys]- analog. The structures differ with respect to the intramolecular hydrogen bonds present (Figure 3). In the [(R)-β 3 -Lys]- analog, the Tyr<sup>1</sup> amino group interacts with the backbone carbonyl oxygens of Phe<sup>4</sup> and Asp<sup>5</sup> . This arrangement might facilitate internal cyclization upon the Asp residue loss. On the other hand, in **RP-172**, the Tyr<sup>1</sup> amino group interacts with the carbonyl oxygens of (S)-β 3 -Lys<sup>2</sup> and of the exocyclic CONH2. by 3.5 kcal/mol (ΔG298) than the lowest-lying conformer of the [(R)-β3- Lys]- analog. The structures differ with respect to the intramolecular hydrogen bonds present (Figure 3). In the [(R)-β3-Lys]- analog, the Tyr1 amino group interacts with the backbone carbonyl oxygens of Phe4 and Asp5. This arrangement might facilitate internal cyclization upon the Asp residue loss. On the other hand, in **RP-172**, the Tyr1 amino group interacts with the carbonyl oxygens of (S)-β3-Lys2 and of the exocyclic CONH2.

This observation corresponds to the results of quantum chemical calculations (performed with Gaussian09 [21], Table S2) for both isomers. The lowest-lying (at the B3LYP/6-31G(d,p) level) gas-phase conformer of the [(S)-β3-Lys2]- analog is more stable

ion resulted from ring-opening and removal of the Asp residue. The difference in intensity of the 586 *m*/*z* ions in panels **RP-171** MS/MS and **RP-172** MS/MS suggests that the fragmentation of peptide **RP-171** occurs easier than in the case of **RP-172**, suggesting that pep-

*Molecules* **2022**, *27*, x FOR PEER REVIEW 5 of 15

tide **RP-172** containing (S)-β3-Lys is more stable.

**Figure 3.** The lowest-lying conformers of **RP-171** and **RP-172** (at the B3LYP/6-31G(d,p) in the gas phase. Green dots show intramolecular hydrogen bonding.

**Figure 3.** The lowest-lying conformers of **RP-171** and **RP-172** (at the B3LYP/6-31G(d,p) in the gas

#### phase. Green dots show intramolecular hydrogen bonding. *2.4. Receptor Binding and Functional Activity*

*2.4. Receptor Binding and Functional Activity*  The binding affinities of cyclopeptides **RP-171** and **RP-172** toward MOR, DOR, and KOR were determined by competitive binding against [3H]DAMGO, [3H][D-Ala2]del-The binding affinities of cyclopeptides **RP-171** and **RP-172** toward MOR, DOR, and KOR were determined by competitive binding against [3H]DAMGO, [3H][D-Ala<sup>2</sup> ]deltorphin-2, and U-69593, respectively, using membranes of CHO cells transfected with opioid receptors and are summarized in Table 1.


torphin-2, and U-69593, respectively, using membranes of CHO cells transfected with opioid receptors and are summarized in Table 1. **Table 1.** Receptor binding affinities (K<sup>i</sup> ) of novel cyclic analogs at MOR, DOR, KOR.

**RP-171** Tyr-c[(*R*)-β3-Lys-Phe-Phe-Asp]NH2 29 ± 4.32 >1000 420 ± 23 **RP-172** Tyr-c[(*S*)-β3-Lys-Phe-Phe-Asp]NH2 950 ± 45 >1000 >1000 <sup>a</sup> Binding affinities were determined by competitive displacement of the selective radioligands, [3H]DAMGO (MOR), [3H]deltorphin-2 (DOR), and [3H]U-69593 (KOR) using commercial membranes of CHO cells transfected with human opioid receptors. Values are expressed as mean ± SEM, n = 3.

The parent compound **RP-170** displayed subnanomolar affinity to MOR, nanomolar to KOR, and did not show substantial DOR affinity. Replacement of D-Lys with (R)-β 3 - Lys generated **RP-171**, which showed about 50-fold lower affinity for MOR but did not bind to the other two opioid receptors, which made this analog much more selective. The diastereoisomeric **RP-172**, incorporating (S)-β 3 -Lys, did not bind to any of the three opioid receptors, showing that affinity of these analogs depended on the configuration of the β-amino acid.

The functional activities of the cyclopeptides in vitro were assessed at all three opioid receptors in calcium mobilization assay in which CHO cells co-expressing human recombinant opioid receptors and chimeric G proteins were used to monitor changes of intracellular calcium levels, reflecting activation of the G protein-coupled receptors (GPCR) [22,23].

The obtained results are summarized in Table 2. Agonist potencies of peptides are given as the negative logarithm of the molar concentration of an agonist that produces 50% of the maximal possible effect (pEC50). Ligand efficacy was expressed as intrinsic activity (α). Dermorphin, DPDPE, and dynorphin A were used as standard agonists for calculating efficacy at MOR, DOR, and KOR, respectively. In CHO-MOR cells, the parent analog **RP-170** induced a significant concentration-dependent release of Ca2+ ions (pEC<sup>50</sup> = 8.93, α = 1.00), with efficacy and potency even higher than those of dermorphin (pEC<sup>50</sup> = 8.57, α = 1.00). For peptides **RP-171** and **RP-172**, the calculated pEC<sup>50</sup> values were 6.87 and 5.45, respectively (for concentration-response curves, see Figure S7). In CHO-DOR cells, DPDPE elicited a strong concentration-dependent Ca2+ release, showing high potency and maximal effect (pEC<sup>50</sup> = 7.23, α = 1.00), while all three cyclopeptides were inactive. In CHO-KOR cells, dynorphin A induced a significant concentration-dependent Ca2+ release (pEC<sup>50</sup> = 9.04, α = 1.00). The potency of **RP-170** was only slightly lower, showing high potency and maximal effect (pEC<sup>50</sup> = 8.60, α = 1.00), **RP-171** displayed significantly lower potency but high efficacy (pEC<sup>50</sup> = 5.99, α = 0.82), and **RP-172** was inactive. Summing up, in this assay, **RP-171** had similar receptor preferences as the parent **RP-170,** while **RP-172** was completely inactive, which points to the importance of the R-chirality at position 2 of these cyclopeptides.

**Table 2.** Effect of new analogs at human recombinant opioid receptors coupled with calcium signaling via chimeric G proteins.


Dermorphin, DPDPE, and dynorphin A were used as reference agonists for calculating intrinsic activity at MOR, DOR, and KOR, respectively: pEC50, aAgonist potency values; α, befficacy values; n ≥ 3.

#### *2.5. Molecular Modeling*

In order to obtain insight into the structural basis for the observed affinities, the analogs **RP-171** and **RP-172** were docked into the structure of the activated MOR (PDB accession code: 6DDF [24]) using AutoDock 4.2.6 [25]. The best scored poses were then subjected to molecular dynamics (MD) simulations (100 ns production, see Figure S8 for RMSD plots).

A general view of the binding pose of **RP-171**, as found in the MD simulations at t = 100.0 ns, is shown in Figure 4A. The interaction scheme is presented in Figure 4B. The compound is anchored in the MOR binding site first and foremost by the canonical interaction of the protonated amino group of Tyr<sup>1</sup> with Asp147. Additionally, the amide hydrogen of the peptide bond joining Tyr<sup>1</sup> and β 3 -Lys<sup>2</sup> interacts with Asp147. These two interactions are stable throughout the simulation (Figure 4C,D). Other polar contacts stabilizing the complex are hydrogen bonds between the exocyclic carbonyl oxygen and Gln<sup>124</sup> or Asn127, but these interactions fluctuate in the simulation time (Figure 4E,F). The remaining contacts are of apolar character. The aromatic ring of Tyr<sup>1</sup> is involved with π-π stacking with Tyr<sup>148</sup> and π-alkyl interactions with Ala<sup>240</sup> and Val<sup>236</sup> side chains. Other residues in the close vicinity of this aromatic ring are Met<sup>151</sup> and His297. The aromatic ring of the Phe<sup>3</sup> residue approaches Trp318, Lys303, and Ala304, while the Phe<sup>4</sup> aromatic ring is

exposed to the solvent close to the extracellular outlet of the binding site. For other residues participating in van der Waals contacts, refer to Figure 4B. the solvent close to the extracellular outlet of the binding site. For other residues participating in van der Waals contacts, refer to Figure 4B.

In order to obtain insight into the structural basis for the observed affinities, the analogs **RP-171** and **RP-172** were docked into the structure of the activated MOR (PDB accession code: 6DDF [24]) using AutoDock 4.2.6 [25]. The best scored poses were then subjected to molecular dynamics (MD) simulations (100 ns production, see Figure S8 for

A general view of the binding pose of **RP-171**, as found in the MD simulations at t = 100.0 ns, is shown in Figure 4A. The interaction scheme is presented in Figure 4B. The compound is anchored in the MOR binding site first and foremost by the canonical interaction of the protonated amino group of Tyr1 with Asp147. Additionally, the amide hydrogen of the peptide bond joining Tyr1 and β3-Lys2 interacts with Asp147. These two interactions are stable throughout the simulation (Figure 4C,D). Other polar contacts stabilizing the complex are hydrogen bonds between the exocyclic carbonyl oxygen and Gln124 or Asn127, but these interactions fluctuate in the simulation time (Figure 4E,F). The remaining contacts are of apolar character. The aromatic ring of Tyr*1* is involved with π-π stacking with Tyr148 and π-alkyl interactions with Ala240 and Val236 side chains. Other residues in the close vicinity of this aromatic ring are Met151 and His297. The aromatic ring of the Phe3 residue approaches Trp318, Lys303, and Ala304, while the Phe4 aromatic ring is exposed to

*Molecules* **2022**, *27*, x FOR PEER REVIEW 7 of 15

RMSD plots).

**Figure 4.** (**A**) Binding mode of **RP-171** (white sticks) in the MOR binding site, as found at t = 100.0 ns of the MD production run. The receptor is shown in a simplified manner, with only selected helices (ribbons) and side chains (thin sticks) shown. Display of nonpolar hydrogens is suppressed for clarity. (**B**) Diagram showing interactions between **RP-171** and the MOR binding site (interaction types colored according to the legend). (**C**–**F**) Time evolutions of selected distances associated with protein‒ligand interactions during the MD simulations**. Figure 4.** (**A**) Binding mode of **RP-171** (white sticks) in the MOR binding site, as found at t = 100.0 ns of the MD production run. The receptor is shown in a simplified manner, with only selected helices (ribbons) and side chains (thin sticks) shown. Display of nonpolar hydrogens is suppressed for clarity. (**B**) Diagram showing interactions between **RP-171** and the MOR binding site (interaction types colored according to the legend). (**C**–**F**) Time evolutions of selected distances associated with protein–ligand interactions during the MD simulations.

A general view of the binding pose of **RP-172**, as found in the MD simulations at t = 100.0 ns, is shown in Figure 5A. The interaction scheme is presented in Figure 5B. The only polar contact that is consistently present throughout the whole MD production is the Hbond interaction of amide hydrogen of the peptide bond joining Tyr1 and β3-Lys2 interacts A general view of the binding pose of **RP-172**, as found in the MD simulations at t = 100.0 ns, is shown in Figure 5A. The interaction scheme is presented in Figure 5B. The only polar contact that is consistently present throughout the whole MD production is the H-bond interaction of amide hydrogen of the peptide bond joining Tyr<sup>1</sup> and β 3 - Lys<sup>2</sup> interacts with Asp<sup>147</sup> (Figure 5D). Contrary to what was found for **RP-171**, and contrary to what would be expected for strong MOR agonists, the interaction between the protonated amino group of Tyr<sup>1</sup> and Asp<sup>147</sup> is unstable (Figure 5C). This H-bond, while present in the binding pose found by docking, is broken after the 65 ns of the MD simulations. Another polar contact that is broken during the MD run involves the interaction of exocyclic carbonyl oxygen with the Arg<sup>211</sup> side chain guanidine group. By the end of the simulation, the carbonyl oxygen of Phe<sup>3</sup> starts with backbone amide hydrogen of Leu<sup>218</sup> and hydroxyl hydrogen of Thr<sup>219</sup> (Figure 5E,F). With respect to apolar contacts (found in the final snapshots of the simulation), the Tyr<sup>1</sup> aromatic ring interacts with the Met<sup>151</sup> side chain (π-alkyl interaction). The Phe<sup>3</sup> aromatic ring approaches Trp<sup>318</sup> and participates in π-alkyl interactions with the side chains of Leu219, Lys233, and Val236. Other receptor residues interacting with the peptide are shown in Figure 5B.

with the peptide are shown in Figure 5B.

**Figure 5.** (**A**) Binding mode of **RP-172** (white sticks) in the MOR binding site, as found at t = 100.0 ns of the MD production run. The receptor is shown in a simplified manner, with only selected helices (ribbons) and side chains (thin sticks) shown. Display of nonpolar hydrogens is suppressed for clarity. (**B**) Diagram showing interactions between **RP-172** and the MOR binding site (interaction types colored according to the legend). (**C**–**F**) Time evolutions of selected distances associated with protein‒ligand interactions during the MD simulations. **Figure 5.** (**A**) Binding mode of **RP-172** (white sticks) in the MOR binding site, as found at t = 100.0 ns of the MD production run. The receptor is shown in a simplified manner, with only selected helices (ribbons) and side chains (thin sticks) shown. Display of nonpolar hydrogens is suppressed for clarity. (**B**) Diagram showing interactions between **RP-172** and the MOR binding site (interaction types colored according to the legend). (**C**–**F**) Time evolutions of selected distances associated with protein–ligand interactions during the MD simulations.

with Asp147 (Figure 5D). Contrary to what was found for **RP-171**, and contrary to what would be expected for strong MOR agonists, the interaction between the protonated amino group of Tyr1 and Asp147 is unstable (Figure 5C). This H-bond, while present in the binding pose found by docking, is broken after the 65 ns of the MD simulations. Another polar contact that is broken during the MD run involves the interaction of exocyclic carbonyl oxygen with the Arg211 side chain guanidine group. By the end of the simulation, the carbonyl oxygen of Phe3 starts with backbone amide hydrogen of Leu218 and hydroxyl hydrogen of Thr219 (Figure 5E,F). With respect to apolar contacts (found in the final snapshots of the simulation), the Tyr1 aromatic ring interacts with the Met151 side chain (π-alkyl interaction). The Phe3 aromatic ring approaches Trp318 and participates in π-alkyl interactions with the side chains of Leu219, Lys233, and Val236. Other receptor residues interacting

#### **3. Discussion**

**3. Discussion**  β-Amino acids, although much less abundant than their α-analogs, are also present in nature and exhibit interesting pharmacological properties. The difference between αand β-amino acids is in the number of carbon atoms (one or two, respectively) that separate an amino and a carboxy termini. β-Amino acids with side chains other than H can exist as *R* or *S* isomers at either the α (C2) carbon or the β (C3) carbon, producing β2- or β-Amino acids, although much less abundant than their α-analogs, are also present in nature and exhibit interesting pharmacological properties. The difference between α- and β-amino acids is in the number of carbon atoms (one or two, respectively) that separate an amino and a carboxy termini. β-Amino acids with side chains other than H can exist as *R* or *S* isomers at either the α (C2) carbon or the β (C3) carbon, producing β 2 - or β 3 -amino acids, respectively.

β3-amino acids, respectively. The most common naturally occurring β-amino acid is β-alanine, which is a component of pantothenic acid (vitamin B5), which, in turn, is a component of coenzyme A. Another example of a natural β-amino acid is (1*R*,2*S*)-2-aminocyclopentanecarboxylic acid The most common naturally occurring β-amino acid is β-alanine, which is a component of pantothenic acid (vitamin B5), which, in turn, is a component of coenzyme A. Another example of a natural β-amino acid is (1*R*,2*S*)-2-aminocyclopentanecarboxylic acid (cispentacin), an antifungal antibiotic isolated from *Bacillus cereus* [26].

(cispentacin), an antifungal antibiotic isolated from *Bacillus cereus* [26]. β-Peptides (made of only β-amino acids) in general do not appear in nature, though among the opioid peptides, there are examples of such synthetic analogs [27]. More often mixed α/β-peptides, in which one or more β-residues are incorporated instead of some α-amino acids, were constructed [28–34].

An important advantage of peptide analogs incorporating β-amino acids over natural peptides is their stability against proteolytic degradation [35,36], which makes β-amino acids desirable building blocks in the preparation of peptide-based drugs [37].

In this report, we used (*R*)- and (*S*)-β 3 -Lys to assess the influence of a β-amino acid on the conformation of the macrocycle of Tyr-c[D-Lys-Phe-Phe-Asp]NH<sup>2</sup> (**RP-170**), which has nanomolar MOR and KOR affinity. The obtained diastereoisomeric analogs **RP-171** and **RP-172** are also isomers of the parent compound **RP-170**, with which they share the same number of atoms in the whole structure and in the macrocycle. The difference between **RP-171**/**RP-172** and the parent **RP-170** is the point at which the exocyclic Tyr<sup>1</sup> is attached to the ring (one carbon atom shift as compared with **RP-170**). The experimental evaluation of the binding affinity and functional activity of **RP-171** showed that such minor structural change had a significant effect on the biological properties causing 53- and 276-fold loss of affinity for MOR and KOR, respectively, as compared with the parent. The inversion of the configuration of β 3 -Lys in **RP-172** induced an almost complete loss of affinity of this peptide for the opioid receptors.

The diastereomers **RP-171** and **RP-172** exhibit slightly different lipophilicity, as could be seen from their chromatographic behavior in the reversed-phase liquid chromatography on a C<sup>18</sup> column. No additional effects were observed when the biphenyl stationary phase was used, which may suggest that the arrangement of aromatic rings in the isomers was not suitable for interactions with a biphenyl motif.

Further differences between isomers were revealed after a thorough analysis of fragmentation patterns in the MS<sup>n</sup> experiments. Peptide **RP-172** containing (*S*)-β 3 -Lys turned out to be more stable, and this observation corresponds with the results of conformational analysis and quantum chemical calculations. They show a significant difference in the structure and the energetics of the lowest-lying conformers of both diastereoisomers.

In our former work [33], we devised an interaction model for **RP-170** and its analogs, in which the peptides were anchored in the MOR binding pocket by interactions at the three key binding subsites. According to that model, in the S1 subsite, the protonable amino group of Tyr<sup>1</sup> interacts with Asp<sup>147</sup> (a typical contact for high-affinity MOR agonists of both peptide [24] and non-peptide character [38]). In the S2 and S3 subsites reside the aromatic rings of the Phe residues. Our analyses suggested that the ability to place Tyr<sup>1</sup> , Phe<sup>3</sup> , and Phe<sup>4</sup> in these subsites is important for high MOR affinity.

The present results seem to corroborate this model. Replacement of D-Lys by (R) or (S)-β3-Lys produced a topographical shift of Phe<sup>3</sup> and Phe<sup>4</sup> in regard to Tyr<sup>1</sup> . As a conse-quence (according to the molecular docking and molecular dynamics), neither **RP-171** nor **RP-172** could accommodate their Phe<sup>3</sup> and Phe<sup>4</sup> aromatic rings in the way the parent compound did. This explains the lower MOR affinity of **RP-171**. This compound exhibits, however, the canonical interaction between the Tyr<sup>1</sup> amino group and Asp147. On the contrary, for **RP-172**, such interaction (while present in the docked pose) is unstable in the MD simulations. This could be correlated to a much-diminished MOR affinity found experimentally for this analog. The obtained experimental and theoretical data form the basis for further work on **RP-170** analogs, an important element of which will be ADME/T evaluation.

#### **4. Materials and Methods**

#### *4.1. Materials*

All protected α-amino acids were purchased from Bachem A (Bubendorf, Switzerland). Opioid radioligands, [3H]DAMGO, [3H]deltorphin-2, and [3H]U-69593, and human recombinant opioid receptors were purchased from PerkinElmer (Krakow, Poland). GF/B glass fiber strips were obtained from Whatman (Brentford, UK). Purity of peptides was determined by RP-HPLC and exact mass. Analytical and semi-preparative RP-HPLC was performed using Waters Breeze instrument (Milford, MA, USA) with dual absorbance detector (Waters 2487, Milford, MA, USA). All ESI-MS experiments were performed on a Shimadzu IT-TOF mass spectrometer (Shimadzu, Japan) equipped with ESI source connected to Nexera HPLC system (Shimadzu, Japan). The instrument was operated in the positive-ion mode. Peptide solutions (1 µL) were introduced in a 0.2 mL/min flow of mobile phase. For LC-MS experiments, Aeris Peptide C<sup>18</sup> and Kinetex Biphenyl (Phenomenex, Torrance, CA, USA) were used, in a gradient reversed-phase mode, from 5 to 80% acetonitrile in water (both containing 0.1% HCOOH). <sup>1</sup>H NMR spectra were recorded on a 500 MHz Brucker instrument in DMSO-d6, using residual DMSO as a resonance reference at 2.5 ppm.

#### *4.2. Synthesis of Fmoc-Protected (R)- and (S)-β 3 -Lys(Mtt)*

To the 500 mL three-necked, round bottom flask with Liebig's condenser equipped with thermometer, magnetic stirrer and protected from moisture with a tube with anhydrous calcium chloride, a solution of Fmoc-D-Orn-(Boc)-OH (**1**) (3 g, 6.6 mmol, 1 eq) in 50 mL of tetrahydrofuran (THF) was added, stirred and cooled to −30 ◦C. Then, Nmethylmorpholine (1.52 mL, 13.9 mmol, 2.1 eq) was added, followed by methyl chloroformate (0.56 mL, 7.3 mmol, 1.1 eq) added dropwise, and stirring was continued for 30 min at −30 ◦C. Next, the diazomethane obtained, using standard procedure, from Diazald®(8.48 g, 13.9 mmol, 6 eq.) was distilled along with diethyl ether directly to the flask. The temperature in the flask was maintained below −10 ◦C, and after 1 h, the cooling bath was removed. The reaction was completed in 2 h (LC-MS analysis). Acetic acid (5 mL) was added to decompose the excess diazomethane, and stirring was continued for 30 min. Then, 100 mL of diethyl ether was added, and the solution was washed with water (2 × 100 mL), 5% NaHCO<sup>3</sup> (2 × 50 mL), and brine. The organic fraction was dried over MgSO<sup>4</sup> to obtain, after evaporation, 3 g (95%) of diazoketone (**2**), which was used in the next step without further purification.

Diazoketone (**2**) (3 g, 6.3 mmol, 1 eq) was dissolved in the mixture of THF and water (55 mL; 10:1) in a 250 mL round bottom flask. Triethylamine (1.78 mL, 17.5 mmol, 2.8 eq) and silver trifluoroacetate (0.15 g, 0.7 mmol, 0.11 eq) were added, and stirring was continued for 30 min. The solution was diluted with diethyl ether (200 mL), followed by 5% NaHCO<sup>3</sup> (200 mL). The white precipitate was filtrated off and combined with the aqueous phase. Its pH was adjusted to 2 with 2 M HCl, and the product was extracted with ethyl acetate (3 × 150 mL). The organic solution was washed with brine (100 mL) and dried over MgSO4. Evaporation of the solvent gave a white product, which, after purification by flash chromatography (hexane:ethyl acetate = 1:1; Rf = 0.3), yielded 1.5 g (51%) of **3**.

A total of 0.9 g of **3** was dissolved in dioxane (5 mL); 4N HCl/dioxane (10 mL) was added, and the mixture was stirred until the reaction was completed (LC-MS). The solid residue obtained after evaporation was suspended in propylene oxide (10 mL) and refluxed for 2 h until all chloride ions reacted with the silver nitrate solution. Then, diethyl ether was added, and the white precipitate was filtered and dried. The obtained zwitterionic product **4** (0.71 g, yield ~100%) was used in the next step without further purification.

In total, 0.71 g (1.92 mmol, 1 eq) of **4** was suspended in DCM (20 mL). N,O-bis(trimethylsilyl) acetamide (0.535 g, 2.5 mmol, 1.3 eq) was added and stirred for 30 min. Next, 4-methyltrityl chloride (0.562 g, 1.92 mmol, 1 eq) was added along with DIPEA (2.5 mmol, 0.44 mL, 1.3 eq). The reaction was kept overnight at r.t. and controlled with LC-MS. When completed, the solvent was evaporated, and the residue was dissolved in ethyl acetate (200 mL) and washed with 5% NaHCO<sup>3</sup> (2 × 100 mL) and brine (100 mL). The organic layer was dried over MgSO<sup>4</sup> and evaporated. The product was purified by flash chromatography (Rf = 0.2 in DCM:MeOH = 10:1), giving 0.67 g of the final product **5** with a 50% yield. <sup>1</sup>H NMR spectrum (Figure S1) confirmed the structure.

#### *4.3. Peptide Synthesis*

Synthesis of linear precursors of cyclopeptides was performed by the standard manual solid-phase procedure on MBHA Rink-Amide resin (100–200 mesh, 0.8 mmol/g), using 9-fluorenylmethoxycarbonyl (Fmoc) protection for the α-amino groups of amino acids. N<sup>ε</sup> -amino group of (R)- and (S)-β 3 -Lys was protected by the 4-methyltrityl (Mtt), β-carboxy group of Asp by 2-phenyl-isopropyl ester (O-2 PhiPr) and hydroxy group of Tyr by tbutyl (t-Bu). Piperidine in DMF (20%) was used for the deprotection of Fmoc groups, and 2-(1H-benzotriazol-1-yl)-1,1,3,3-tetramethyluronium tetrafluoroborate (TBTU) was employed as a coupling agent and diisopropylethylamine (DIEA) as a neutralizing base. Fully assembled Fmoc-protected peptides were treated with 1% trifluoroacetic acid (TFA) in dichloromethane (DCM) to remove the side chain Mtt and O-2PhiPr protecting groups, followed by on-resin cyclization (TBTU). Cleavage from the resin was accomplished by treatment with TFA/triisopropylsilane (TIS)/water (95:2.5:2.5) for 3 h at room temperature.

Crude peptides were purified by preparative reversed-phase HPLC on a Vydac C<sup>18</sup> column (10 µm, 22 × 250 mm), flow rate 2 mL/min, 20 min linear gradient from water/0.1% (*v*/*v*) TFA to 80% acetonitrile/20% water/0.1% (*v*/*v*) TFA. The purity of the final peptides was verified by analytical HPLC employing a Vydac C<sup>18</sup> column (5 µm, 4.6 × 250 mm), flow rate 1 mL/min, and the same solvent system over 50 min. The purity of the obtained peptides was >95%. Calculated values for protonated molecular ions were in agreement with those determined by high-resolution mass spectroscopy with electrospray ionization (ESI-MS) (Table S1).

#### *4.4. Opioid Receptor Binding Assays*

The opioid receptor binding assays were performed according to the described method [39], using commercial membranes of Chinese Hamster Ovary (CHO) cells transfected with human opioid receptors. The binding affinities for MOR, DOR, and KOR were determined by radioligand competition analysis using [3H]DAMGO, [3H]deltorphin-2, and [ <sup>3</sup>H]U-69593, respectively, as specific radioligands, respectively. Membrane preparations were incubated at 25 ◦C for 120 min with appropriate concentrations of a tested peptide in the presence of 0.5 nM radioligand in a total volume of 0.5 ml of 50 mM Tris/HCl (pH 7.4) containing bovine serum albumin (BSA) (1 mg/mL), bacitracin (50 µg/mL), bestatin (30 µM), and captopril (10 µM). Non-specific binding was determined in the presence of 1 µM naloxone. Incubations were terminated by the rapid filtration through the GF/B Whatman (Brentford, UK) glass fiber strips (pre-soaked for 2 h in 0.5% (*v*/*v*) polyethylamine) using Millipore Sampling Manifold (Billerica, MA, USA). The filters were washed three times with 4 ml of ice-cold Tris buffer solution. The bound radioactivity was measured in a Packard Tri-Carb 2100 TR liquid scintillation counter (Ramsey, MN, USA) after overnight extraction of the filters in 4 mL of a Perkin Elmer Ultima Gold scintillation fluid (Wellesley, MA, USA). Three independent experiments for each assay were carried out in duplicate. The data were analyzed by a nonlinear least square regression analysis computer program Graph Pad PRISM 6.0 (Graph Pad Software Inc., San Diego, CA, USA). The IC<sup>50</sup> values were determined from the logarithmic concentration–displacement curves, and the values of the inhibitory constants (K<sup>i</sup> ) were calculated according to the equation of Cheng and Prusoff [40].

#### *4.5. Calcium Mobilization Assay*

Calcium mobilization assay was performed, as reported in detail elsewhere [41], using CHO cells stably co-expressing human recombinant MOR or KOR and the C-terminally modified Gαqi5 and CHO cells co-expressing human recombinant DOR and the GαqG66Di5 chimeric protein (a generous gift from Prof. Girolamo Calo, University of Padova, Italy). Cells were cultured in a culture medium consisting of Dulbecco's MEM/HAMS F12 (1:1) supplemented with 10% fetal bovine serum, penicillin (100 IU/mL), streptomycin (100 µg/mL), L-glutammine (2 mM), fungizone (1 µg/mL), geneticin (G418; 200 µg/mL) and hygromycin B (100 µg/mL). Cell cultures were kept at 37 ◦C in 5% CO2/humidified air. Cells were seeded at a density of 50,000 cells/well into 96-well black, clear-bottom plates. After 24 h incubation, the cells were loaded with a medium supplemented with probenecid (2.5 mM), calcium-sensitive fluorescent dye Fluo-4 AM (3 µM), pluronic acid (0.01%), and HEPES (20 mM) and kept for 30 min at 37 ◦C. Then, the loading solution was aspirated, and 100 µL/well of assay buffer (HBSS supplemented with 20 mM HEPES, 2.5 mM probenecid, and 500 µM Brilliant Black) was added. After placing both plates (cell culture and compound plate) into the FlexStation II (Molecular Device, Union City, CA, USA), the on-line additions were carried out in a volume of 50 µL/well and the fluorescence changes were measured. Ligand efficacies, expressed as the intrinsic activity (α), were calculated as the E*max* ratio of the tested compound and the standard agonist. At least three independent experiments for each assay were carried out in duplicate.

Curve fittings were performed using Graph Pad PRISM 5.0 (GraphPad Software Inc., San Diego, CA, USA). Data have been statistically analyzed with one-way ANOVA followed by the Dunnett's test for multiple comparisons; *p* values < 0.05 were considered significant.

#### *4.6. Quantum Chemical Calculations*

One hundred conformers for compounds **RP-171** and **RP-172** were generated by an in-house Python script using the improved ETKDG method [42]. The compounds were protonated at the N-terminal nitrogen atom. The geometries were optimized in Gaussian09 [21] at the B3LYP/6-31G level in a gas phase or in water using the PCM solvent model. The resulting geometries were then reoptimized at the B3LYP/6-31G(d,p) level. Further attempts to increase the theory level were unsuccessful for the lack of convergence. Top conformers were subject to harmonic frequency calculations at the B3LYP/6-31G(d,p) level in order to ascertain that the geometries are minima (no imaginary frequencies) and to calculate thermochemical values.

#### *4.7. Molecular Docking*

One hundred conformers of **RP-171** and **RP-172** (obtained as described in Section 4.6) were docked into the activated structure of the MOR (PDB accession code: 6DDF [24], a complex of mu opioid receptor with Gi protein, with DAMGO peptide in the orthosteric binding site) using AutoDock 4.2.6 [25]. The ligands and the protein were processed in AutoDock Tools 4 [25]. The ligands' side chains were allowed to rotate, and the receptor structure was kept rigid. The docking box was set around the position of the DAMGO molecule in the 6DDF structure [24]. The grids (82 × 78 × 104 points, with 0.375 Å spacing) were calculated with AutoGrid, and the docking was performed using Lamarckian Genetic Algorithm local searches according to the pseudo-Solis and Wets algorithm. Each docking consisted of 100 runs. The results were clustered, and the top scored solutions were visually inspected to examine their conformity to the known literature data on ligand MOR interactions [43]. Molecular graphics were prepared in Biovia Discovery Studio Visualizer [44].

#### *4.8. Molecular Dynamics*

The complexes of MOR with **RP-171** and **RP-172** (obtained by molecular docking, described in Section 4.7) were subject to molecular dynamics simulations in GROMACS 5.1.2 [45]. The complexes were embedded in a lipid bilayer of POPC molecules (128 molecules) solvated with water molecules (TIP3P type, 13,000 molecules) and supplied with ions (Na<sup>+</sup> and Cl−, 0.154 M). These steps were performed with the CHARMM-GUI service [46]. CHARMM 36 force field was used for modeling the proteins, lipids, water, and ions. The ligands were modeled using CHARMM CGenFF [47].

The complexes were minimized and equilibrated, whereafter 100 ns production was performed (NPT ensemble, temperature = 303.15 K, integration step = 2 fs, cut-off scheme Verlet, Nose-Hoover thermostat, Parrinello–Rahman barostat, LINCS H-bonds constraints).

**Supplementary Materials:** The following are available online. Figure S1: H<sup>1</sup> NMR spectrum of Fmoc- (*R*)- β 3 -Lys(Mtt); Figures S2 and S3: high-resolution mass spectra of analogs **RP-171** and **RP-172**; Figures S4–S6: LC-MS and MS<sup>n</sup> analysis of analogs **RP-171** and **RP-172**; Figure S7: concentration– response curves of analogs in the functional assay; Figure S8: root mean square deviations of protein and ligand in the MD simulations; Table S1: physicochemical characterization of analogs 2–9; Table S2: total energies of top 15 conformers for **RP-171** and **RP-172**.

**Author Contributions:** Conceptualization, K.W. and A.J.; investigation, K.W., A.A.-B., J.P.-C., P.F.J.L. and A.K.; methodology, K.W., P.F.J.L., A.A.-B., J.S. and A.K.; supervision, A.J.; writing—original draft, K.W., A.J., P.F.J.L. and A.K.; writing—review and editing, A.J., P.F.J.L. and A.K. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by a grant from the Medical University of Lodz No. 503/1-156- 02/503-11-001-19-00.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Not applicable.

**Acknowledgments:** The calculations were performed at Swierk Computing Centre, National Centre ´ for Nuclear Research, Swierk, Poland. The authors would like to express their gratitude to Girolamo ´ Calo from the University of Padova for the kind gift of CHO cells expressing opioid receptors and chimeric G proteins and to Andrzej Reszka (Shim-Pol, Poland) for providing access to the Shimadzu IT-TOF instrument.

**Conflicts of Interest:** The authors have no conflict of interest to declare.

#### **References**

