*3.3. [3H]NOP Receptor Binding Assay*

Competitive binding assays at the human NOP receptor stably transfected into CHO cells were performed according to the published procedure [62]. Cell membranes (15 µg) were incubated in 50 mM Tris-HCl buffer (pH 7.4) with [3H]nociceptin (0.1 nM) and various concentrations of test compounds in a final volume of 1 mL, for 60 min at 25 ◦C. Nonspecific binding was determined using 10 µM of unlabeled nociceptin. After incubation, reactions were terminated by rapid filtration through 0.5% PEI-soaked Whatman GF/C glass fiber filters. Filters were washed three times with 5 mL of ice-cold 50 mM Tris-HCl buffer (pH 7.4) using a Brandel M24R cell harvester (Brandel, Gaithersburg, MD, USA). Radioactivity retained on the filters was counted by liquid scintillation counting using a Beckman Coulter LS6500 (Beckman Coulter Inc., Fullerton, CA, USA). Inhibition constant (K<sup>i</sup> , nM) values were determined by the method of Cheng and Prusoff [64] from concentration-response curves by nonlinear regression analysis using the GraphPad Prism 5.0 Software (GraphPad Prism Software Inc., San Diego, CA, USA). All experiments were performed in duplicate and repeated three times with independently prepared samples. Data are presented as means ± SEM.

#### *3.4. Protein Preparation and Modeling of the KOR Active Conformation*

For classical opioid receptors, X-ray crystal structures of the active state proteins are published and provided in the protein data bank (PDB [65]). The respective structures with PDB-IDs 5C1M for the MOR [42], 6PT2 for the DOR [43] and 6B73 for the KOR [41] were prepared using MOE v2020.0901 [66]. The X-ray crystal structure of the inactive state NOP receptor (PDB-ID: 5DGH) was prepared analog. Only the chain with the best resolution was processed. Fusion proteins (antibody fragment in MOR, thermostabilized cytochrome b562 (BRIL) in the DOR, nanobody in the KOR) and the unresolved parts of the N-terminus, as well as of the C-terminus of the opioid receptors were deleted. Thermostabilizing mutations in the DOR and KOR were subsequently reverted to the human wild-type sequence obtained from the UniProt-Databank [67] (human DOR: P41143, human KOR: P41145). The MOR structure (PDB-ID: 5C1M) is of a murine receptor. Hence, the sequence was manually mutated to obtain the human wild-type MOR model (UniProt-ID: P35372). The NOP receptor structure already contained the human sequence. Missing side chain atoms were automatically generated using the protein builder integrated in MOE. The unresolved parts of ECL2, ECL3 and ICL3 of the KOR and ICL2 of the NOP receptor were

modeled using the loop modeler panel within MOE. To obtain high quality structures, Ramachandran outliers [68] and atom clashes were resolved using energy minimization with the OPLS-AA force field [69].

Homology modeling of the active state NOP receptor was performed using MOE v2020.0901 with default settings in a similar as described in [70]. The chain with the best resolution (3.10 Å) of the active KOR structure (PDB-ID: 6B73, sequence identity of 59% and sequence similarity of 73%) with the NOP receptor (Figure S1) served as a template. The protein target sequence (human NOP receptor) was obtained from the UniProt-Database (human NOP receptor P41146). Both Ramachandran outliers as shown in Figure 3 are located in flexible loops far away from the binding site (T206 of extracellular loop 2, ECL2, and S251 of intracellular loop 3, ICL3). Hence, we assume that these Ramachandran outliers are unlikely to influence ligand binding. Visual inspection revealed that the side chain orientations of the residues forming the orthosteric binding pocket, including D3.32 (number denote Ballesteros–Weinstein numbering [44]), responsible for the crucial ionic interaction between opioids and their receptors, show a similar orientation in the generated model as in the template.

'Interaction potential maps' as implemented in MOE v2020.0901 were used to determine putatively relevant water molecules inside the binding site of the KOR (resolution too low to determine co-crystallized waters) and the NOP receptor (homology model without water coordinates; too low resolution in the crystal structure). The interaction potential is an energy-based function that probes water molecules within the protein and calculates the interaction energy between water molecule and protein [66]. For this calculation the KOR binding site was defined as all residues within 4.5 Å around the crystalized ligand MP1104 in the KOR structure (PDB-ID: 6B73). Since the KOR and NOP receptor share a high sequence identity (59%) the same resides were used to define the NOP binding site in the active state homology model. For the NOP receptor crystal structure again, all residues within 4.5 Å around the crystalized ligand C-35 were used.

### *3.5. Protein-Ligand Docking*

The starting conformation of HS-731 (IUPAC name: 2-[(4,5α-epoxy-3-hydroxy-14β-metho xy-17-methylmorphinan-6β-yl)amino]acetic acid) was generated using Corina v3.00 [71,72]. All five opioid receptor structures were protonated at a pH of 7.0 using the protonate 3D function [73] included in MOE (v2020.0901). GOLD v5.2 [74] was used for docking HS-731 into the receptors. The binding site was defined as a 20 Å sphere around the side chain carboxylate C (γC)-atom of D3.32 and restricted to the solvent-accessible surface. Pyramidal nitrogen atoms in the ligand were allowed to flip during the docking process. A total of 30 genetic algorithm runs per receptor structure were performed, generating diverse solutions (the root mean square deviation between docking poses was more than 1.5 Å). The generated binding hypotheses were scored using the GoldScore docking function [75,76]. The search efficiency was held at 100%. A constraint maintaining a maximum distance of 5.5 Å between the nitrogen in the morphinan scaffold and the γC-atom of D3.32 was set to ensure a crucial ionic interaction [41,56,57,77].

The obtained binding poses were energy-minimized in the protein environment using the MMFF94 force field [78] implemented in LigandScout v4.4.3 [79,80]. The binding poses of HS-731 in complex with the MOR, DOR and KOR were visually inspected and filtered according to the reported binding mode of the morphinan scaffold of opioid agonist BU72 co-crystallized with the MOR (PDB-ID: 5C1M [42]) and the morphinan scaffold of the opioid agonist MP1104 co-crystallized with the KOR (PDB-ID: 6B73 [41]). Additionally, MP1104-KOR interactions were used to score the DOR docking results as MP1104 also exhibits full agonism at the DOR. The relevant interactions are summarized in Table 3. Rescoring of the MOR and KOR clearly identified one docking result as most plausible that was chosen for further evaluation. At the DOR however, several docking results were scored equal. Thus, the pose with the lowest distance between the positively charged

nitrogen in the morphinan scaffold and the carboxylate of D3.32 out of the best scored docking results was chosen at the DOR.


**Table 3.** Ligand–receptor interactions used for rescoring of docking results.

PI, positive ionizable interaction; HY, hydrophobic interaction; HBA, hydrogen bond acceptor; HBD, hydrogen bond donator; HOH refers to water molecules.

None of the crystallized opioid ligands exhibit agonist activity to the NOP receptor, but due to high identity and similarity to the classical opioid receptors (Figure S1) a similar binding mode of HS-731 in all active state opioid receptors was assumed. MP1104 shares the morphinan scaffold of HS-731 and an alignment and superposition of the KOR crystal structure and the NOP receptor homology model revealed the same orientation of the residues that interact with MP1104 in the MP1104-KOR-complex and their NOP receptor equivalent, with the exception of Y1313.33. Therefore, the binding poses were evaluated according to the geometry of the other interactions detected in the MP1104-KOR complex (Table 3). For the inactive state NOP receptor (PDB-ID: 5DGH) the orientation and interaction pattern of the cocrystallized ligand C-35 was used to evaluate the docking poses. C-35 only exhibit the crucial ionic interaction towards D1303.32 as well as several hydrophobic interactions (to I1273.29, I1293.31, Y1313.33, M1343.36, V2796.51, V2836.55).

#### *3.6. Molecular Dynamics Simulations and Analysis to Evaluate Docking Poses*

Five replicates of molecular dynamics (MD) simulations of 100 ns were performed for each receptor-ligand complex. The systems were set up using Maestro v2020-4 [81] and parametrized using the OPLS 2005 force field [82,83]. The MD simulations were performed using Desmond v2020-4 [84]. The protein was placed in a cubic box with 10 Å padding either side to the protein surface filled with TIP4P water molecules [85] and ions (0.15 M NaCl), to ensure isotonic conditions, and was embedded in a 1-palmitoyl-2-oleoylphosphatidylcholine (POPC) bilayer. The membrane placement was carried out according the OPM database (PDB-ID: 5C1M for the MOR, 6PT2 for the DOR, 6B73 for the KOR). The simulation was performed under periodic boundary conditions as an NPγT ensemble, i.e., a constant number of particles, pressure (1.01325 bar), lateral surface tension (0 N/m) and temperature (300 K) throughout the simulation. Each simulation resulted in 1000 system conformations, according to a 100 ps recording interval. VMD v1.9.3 [86] was used to center the protein and to align the trajectory onto the backbone heavy atoms of the starting protein conformation.

For MD simulation analysis, dynamic pharmacophores, so called dynophores [55,87], were calculated. Dynamic pharmacophores encompass pharmacophoric information derived from an ensemble of protein conformations obtained from MD simulations. Interactions are grouped into feature groups according to their interaction type (e.g., lipophilic interaction, hydrogen bond acceptor, hydrogen bond donator). The interaction occurrence over the trajectory of each interaction group is statistically determined and reported as percentages. The dynophore algorithm is implemented the ilib framework, on which also LigandScout [79,80] is based upon. To assess the quality of interactions occurring during the MD simulations distances between HS-731-COO-(C-atom)-KOR-K2275.39 (Nz), HS-731-COO-(C-atom)-DOR-K2145.39 (Nz), HS-731-COO-(C-atom)-MOR-K2355.39 (Nz) and HS-731-COO-(C-atom)-MOR-K3056.58 (Nz), HS-731-COO-(C-atom)–DOR-R291 (Cz), HS-731-COO-(C-atom)-KOR-K200, and HS-731-secundary amine-KOR-E209 (CD) were measured using VMD. The violin plots (Figure 7) representing the distribution of measured distances were generated using the python v3.8.5 [88] packages seaborn v0.11.2 [89] and matplotlib v3.4.3 [90].

#### **4. Conclusions**

In this study, we assessed the difference in binding affinity and activity values of the peripheral opioid antinociceptive, HS-731, at the opioid receptors, and generated a binding hypothesis at each opioid receptor subtype. HS-731 shows extensive ionic interactions with the classical opioid receptors, MOR, DOR and KOR, and the differences in the frequency and quality of those interactions mediate differences in the affinity and activity of HS-731 to these receptors. At the MOR, HS-731 forms four ionic interactions over the majority of the MD simulations. At the DOR and KOR, there were only two noteworthy ionic interactions present. A closer examination of the interaction quality facilitated by an interaction distance assessment revealed by far the strongest ionic interactions at the MOR followed by the DOR. The quality at the KOR was much weaker than at the DOR. A salt bridge between K2275.39 and E2976.58 was observed in about 50% in the case of the KOR. This interaction is likely to cause the KOR to adopt an intermediate-state conformation as supported by the decreased distance between the bottom of TM6 and TM4 as a surrogate parameter for the TM6 translocation and GPCR activation, and therefore could explain the partial agonism of HS-731 to the KOR. The MOR and DOR that did not exhibit TM5-TM6 ionic interactions, and thus were not forced to adopt an intermediate state conformation are able to be fully activated by the agonist HS-731.The present results highlight the importance of ionic interactions for the binding of the 6β-glycine substituted agonist HS-731 to the opioid receptors, and accentuate the non-conserved residue 6.58 and the N-terminus, as important selectivity determinants for the classical opioid receptors. We experimentally demonstrate that HS-731 displayed no substantial binding to the NOP receptor. We surmise that Y1313.33 is responsible for this observation, in that it points further into the active state binding pocket than in the classical opioid receptors and prevents HS-731 binding within the orthosteric binding pocket. Furthermore, the hydroxyl group of HS-731 is likely to abolish ligand binding to the NOP receptor in that it mimics the tyrosine within the message address of endogenous peptides for the classical opioid receptors instead of the phenylalanine within the message address of the NOP receptor agonist nociceptin.

In conclusion, our findings offer significant structural insights into HS-731 interactions with the opioid receptors that are important for understanding the pharmacology of this peripheral opioid analgesic.

**Supplementary Materials:** The following are available online, Figure S1: Sequence identity and similarity among the opioid receptors, Figure S2: Docking pose of HS-731 to the inactive NOP receptor, Figure S3: Root mean square deviation of HS-731 in complex with the MOR over the simulation time, Figure S4: Root mean square deviation of the MOR backbone atoms in complex with HS-731 over the simulation time, Figure S5: Root mean square deviation of HS-731 in complex with the DOR over the simulation time., Figure S6: Root mean square deviation of the DOR backbone atoms in complex with HS-731 over the simulation time, Figure S7: Root mean square deviation of HS-731 in complex with the KOR over the simulation time, Figure S8: Root mean square deviation of the KOR backbone atoms in complex with HS-731 over the simulation time, Figure S9: Comparison of the binding modes of HS-731 at the MOR derived by docking and after MD simulations, Figure S10: Comparison of the binding modes of HS-731 at the DOR derived by docking and after MD simulations. Figure S11: Comparison of the binding modes of HS-731 at the KOR derived by docking and after MD simulations. Figure S12: Ionic interaction distances. Figure S13: Comparison between the active state KOR and the intermediate state KOR.

**Author Contributions:** Conceptualization, M.S. and G.W.; methodology, K.P., H.S. and M.S.; formal analysis, K.P., H.S., G.W. and M.S.; investigation, K.P. and M.S.; resources, H.S., G.W. and M.S; writing—original draft preparation, K.P. and M.S.; writing—review and editing, all authors.; visualization, K.P.; supervision, M.S. and G.W.; funding acquisition, G.W. and M.S. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by Deutsche Forschungsgemeinschaft (DFG: 435233773) and the Austrian Science Fund (FWF: P15481 and I4697).

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

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Data is available from the authors upon reasonable request.

**Acknowledgments:** Administrative support was provided by Szymon Pach and Theresa Noonan from the field of computational drug design at Freie Universität Berlin. We gratefully acknowledge the High-Performance Computing Facilities (Curta) provided by the Zedat at Freie Universität Berlin.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

**Sample Availability:** Sample of compound is available from the authors.

#### **References**

