Molecular Dynamics Simulations to Investigate How PZM21 Affects The

International Journal of Molecular Sciences

Article Molecular Dynamics Simulations to Investigate How PZM21 Aﬀects the Conformational State of the µ-Opioid Receptor Upon Activation

Zhennan Zhao, Tingting Huang and Jiazhong Li *

School of Pharmacy, Lanzhou University, Lanzhou 730000, China; [email protected] (Z.Z.); [email protected] (T.H.) * Correspondence: [email protected]; Tel.: +86-931-8915-587

Received: 6 May 2020; Accepted: 29 June 2020; Published: 1 July 2020

Abstract: Opioid analgesics such as morphine have indispensable roles in analgesia. However, morphine use can elicit side effects such as respiratory depression and constipation. It has been reported that G protein-biased agonists as substitutes for classic opioid agonists can alleviate (or even eliminate) these side effects. The compounds PZM21 and TRV130 could be such alternatives. Nevertheless, there are controversies regarding the efficacy and G protein-biased ability of PZM21. To demonstrate a rationale for the reduced biasing agonism of PZM21 compared with that of TRV130 at the molecular level, we undertook a long-term molecular dynamics simulation of the µ-opioid receptor (MOR) upon the binding of three ligands: morphine, TRV130, and PZM21. We found that the delayed movement of the W2936.48 (Ballesteros–Weinstein numbering) side chain was a factor determining the dose-dependent agonism of PZM21. Differences in conformational changes of W3187.35, Y3267.43, and Y3367.53 in PZM21 and TRV130 explained the observed differences in bias between these ligands. The extent of water movements across the receptor channel was correlated with analgesic effects. Taken together, these data suggest that the observed differences in conformational changes of the studied MOR–ligand complexes point to the low-potency and lower bias effects of PZM21 compared with the other two ligands, and they lay the foundation for the development of G protein-biased agonists.

Keywords: G protein-biased agonists; µ-opioid receptor (MOR); morphine; PZM21; molecular dynamics; simulation

1. Introduction Pain is a highly unpleasant physical sensation caused by illness or injury. It is a common symptom associated with many diseases. With increases in people’s wishes for having a good quality of life, the expectation of treating pain (especially cancer-related pain) is increasing. Morphine [1] is used worldwide because of its strong analgesic activity. However, morphine can produce a series of toxic reactions and side eﬀects: tolerance, constipation, respiratory depression, and addiction. The US Centers for Disease Control and Prevention (Atlanta, GA, USA) has announced that opioid abuse has become one of the most serious public health crises in the USA in the past decade. The number of poisoning incidents caused by opioid abuse is numerous. In 2015, more than 33,000 people in the USA died from an overdose of opioids [2]. In 2017, approximately 50,000 deaths were caused by opioid abuse. From 2001 to 2016, the total number of opioid-related deaths was >335,000 [3]. The World Health Organization claims that about 275 million people worldwide (5.6 per cent of the global population aged 15–64 years) used drugs at least once during 2016. Among them, there were about 34 million people who used opioids and about 19 million who used

Int. J. Mol. Sci. 2020, 21, 4699; doi:10.3390/ijms21134699 www.mdpi.com/journal/ijms Int. J. Mol. Sci. 2020, 21, 4699 2 of 16 opiates [4]. However, >125 million US citizens are suffering from acute or chronic pain, and this number is even greater in the world. Int. J. Mol. Sci. 2020, 21, x FOR PEER REVIEW 2 of 17 Morphine targets the µ-opioid receptor (MOR), which is a member of the G protein-coupled receptorused (GPCR)opiates [4]. family. However, Originally, >125 million it was US citizens believed are thatsuffering GPCRs from could acute or activate chronic only pain, downstreamand this G proteins.number is However, even greater researchers in the world. have found that GPCRs not only activate G proteins, but also activate otherMorphine signal targets transducers, the μ-opioid such receptor as β-arrestin (MOR), and which sca ffisold a member proteins of [ 5the]. MiceG protein-coupled lacking β-arrestin havereceptor been shown (GPCR) to havefamily. lower Originally, resistance it was to believed morphine, that anGPCRs increased could analgesicactivate only eff ect,downstream and prolonged G drugproteins. action compared However, with researchers those from have wild-type found that mice, GPCRs and not the desensitizationonly activate G eproteins,ffect is reducedbut also [6–8]. Theseactivate results other indicate signal indirectly transducers, that such side as β e-arrestinffects are and associated scaffold proteins with the [5].β Mice-arrestin lacking pathway. β-arrestin It has have been shown to have lower resistance to morphine, an increased analgesic effect, and prolonged not been demonstrated directly that the β-arrestin pathway mediates side effects such as respiratory drug action compared with those from wild-type mice, and the desensitization effect is reduced [6– depression8]. These in results mice. indicate However, indirectly opioid th agonistsat side effects are now are associated recognized with widely the β-arrestin as activating pathway. the It G-protein has pathwaysnot been to producedemonstrated an analgesic directly that effect, the andβ-arrestin that some pathway side mediates effects may side beeffects associated such as respiratory with activation of thedepressionβ-arrestin in pathway. mice. However, opioid agonists are now recognized widely as activating the G-protein Developingpathways to opioidproduce substitutes an analgesic that effect, are eandfficacious that some with side few effects side emayffects be is associated the best long-termwith solutionactivation for controlling of the β-arrestin the pathway. side effects mentioned above. Following this hypothesis, several G-protein-biasedDeveloping agonists opioid substitutes have been that developed are efficacious in the with 21st few century.side effects Foris the example, best long-term using the naturalsolution product for mitragyninecontrolling the (an side indole-based effects mentioned alkaloid above. from MitragynaFollowing speciosethis hypothesis,) and its several analogs [9], G-protein-biased agonists have been developed in the 21st century. For example, using the natural Schmid et al. [10] postulated that SR17018, SR15099, and SR-15098 could be potential G protein-biased product mitragynine (an indole-based alkaloid from Mitragyna speciose) and its analogs [9], Schmid agonists. TRV130 (Oliceridine) [11–13] (Trevena, Chesterbrook, PA, USA) not only has good analgesic et al. [10] postulated that SR17018, SR15099, and SR-15098 could be potential G protein-biased activityagonists. but also TRV130 has G protein-biased(Oliceridine) [11–13] properties, (Trevena so it, hasChesterbrook, been granted PA, the USA) designation not only of “breakthroughhas good therapy”analgesic by the activity US Food but also and has Drug G protein-biased Administration prop (FDA)erties, andso it hashas been completed granted phase the designation III clinical of trials. In November“breakthrough 2017, therapy” Trevena by submitted the US Food a new and drug Drug application Administration registration (FDA) and to has the completed FDA. phase AIII Gclinical protein-biased trials. In November agonist reported 2017, Trevena recently submitted is the compound a new drug PZM21. application The latter registration has been to screenedthe usingFDA. molecular docking and modified subsequently [14]. Manglik and colleagues showed that PZM21 A G protein-biased agonist reported recently is the compound PZM21. The latter has been has a strong analgesic effect (EC50 = 4.6 nM) and no morphine-like side effects. It has garnered considerablescreened using attention molecular because docking of its and potential modified medicinalsubsequently value, [14]. Manglik and researchers and colleagues have showed conducted that PZM21 has a strong analgesic effect (EC50 = 4.6 nM) and no morphine-like side effects. It has in-depth research into PZM21. However, Hill et al. [15] demonstrated that PZM21 does not have a garnered considerable attention because of its potential medicinal value, and researchers have close association with a G protein-based pathway (EC = 110 nM) or β-arrestin pathway, and it is a conducted in-depth research into PZM21. However, Hill50 et al. [15] demonstrated that PZM21 does dose-dependent analgesic and G protein-biased agonist. not have a close association with a G protein-based pathway (EC50 = 110 nM) or β-arrestin pathway, Weand wishedit is a dose-dependent to clarify the eanalgesicffect of PZM21 and G protein-biased in a theoretical agonist. fashion. We used an all-atomic molecular dynamicsWe (MD) wished simulation to clarify to comparethe effect the of movementsPZM21 in a oftheoretical the non-biased fashion. agonist We used morphine an all-atomic and biased agonistsmolecular TRV130 dynamics and compound (MD) simulation PZM21. Into thiscomp way,are wethe investigatedmovements theof the conformational non-biased agonist changes of the MORmorphine influenced and biased by the agonists three ligands,TRV130 and respectively. compound The PZM21. chemical In this structure way, we of investigated the three ligands the is shownconformational in Figure1. changes of the MOR influenced by the three ligands, respectively. The chemical structure of the three ligands is shown in Figure 1.

Figure 1. Chemical structures of the studied ligands: (a) TRV130, (b) PZM21, and (c) morphine. Figure 1. Chemical structures of the studied ligands: (a) TRV130, (b) PZM21, and (c) morphine. 2. Results and Discussion

2.1. Molecular Docking Results of Each System The two-dimensional diagram of the interactions between a ligand and various protein residues are shown in Figure2. The tertiary amine cations of morphine, TRV130, and PZM21 had a hydrogen bond with D1473.32 (Ballesteros–Weinstein numbering [16]), data that are consistent with results from previous studies [14,17]. An interaction between morphine and Y1483.33 was noted, and strong π–π

Int. J. Mol. Sci. 2020, 21, x FOR PEER REVIEW 3 of 17

2. Results and Discussion

2.1. Molecular Docking Results of Each System

Int. J. Mol. Sci.The2020 two-dimensional, 21, 4699 diagram of the interactions between a ligand and various protein residues 3 of 16 are shown in Figure 2. The tertiary amine cations of morphine, TRV130, and PZM21 had a hydrogen bond with D1473.32 (Ballesteros–Weinstein numbering [16]), data that are consistent with results from stackingprevious was studies present [14,17]. between An interaction the pyridine between ring morphine of TRV130 and and Y148 W3183.33 was7.35 noted,. PZM21 and strong contains π–π three nitrogenstacking atoms, was so present there werebetween two the interactions pyridine ring in thisof TRV130 system: and (1) W318 a hydrogen7.35. PZM21 bond contains between three PZM21 and Y326nitrogen7.43 ,atoms, and (2) soπ there–π stacking were two between interactions the in benzene this system: ring (1) and a H297hydrogen6.52. bond We used between these PZM21 molecular and Y3267.43, and (2) π–π stacking between the benzene ring and H2976.52. We used these molecular docking results as the initial structures for the next MD simulation. docking results as the initial structures for the next MD simulation.

Figure 2. Molecular docking results for each system. (a) Morphine_docked, (b) TRV130_docked, Figure 2. Molecular docking results for each system. (a) Morphine_docked, (b) TRV130_docked, (c) (c) PZM21_docked. PZM21_docked. 2.2. Stability of Each MD Simulation System 2.2. Stability of Each MD Simulation System We wished to monitor the stability of the system. Hence, after three 500 ns unbiased MD We wished to monitor the stability of the system. Hence, after three 500 ns unbiased MD simulations,simulations, we calculatedwe calculated the RMSDthe RMSD of the of proteinthe protein backbone backbone atoms atoms and and ligands ligands (Figure (Figure3). 3). The The protein volatilityprotein in volatility TRV130_sys in TRV130_sys was relatively was relatively large larg ande and eventually eventually stabilized stabilized atat approximately 2.7 2.7 Å, followedÅ, followed by Morphine_sys by Morphine_sys at approximately at approximately 2.6 Å 2.6 and Å PZM21_sysand PZM21_sys at approximately at approximately 2.3 2.3 Å. Å. The The results suggestedresults thatsuggested TRV130_sys that TRV130_sys had a large had fluctuation a large fluctuation in amino in amino acid changes acid changes compared compared with with those of Morphine_systhose of Morphine_sys and PZM21_sys, and PZM21_sys, which aff whichected affect the conformationaled the conformational changes chan ofges the of the protein protein to to a large extent.a large However, extent. forHowever, the ligands, for the the ligands, fluctuations the fluctuations of the three of the systems three systems were ranked were asranked PZM21_sys as (2.0 Å), TRV130_sys (1.3 Å), and Morphine_sys (0.3 Å). These fluctuations occurred because the molecular structure of PZM21 has a longer side chain, which is more flexible and prone to fluctuation. The molecular structure of morphine contains five rings of a morphinan core that is relatively rigid and does not undergo fluctuation readily. Overall, the three systems eventually reached a relatively stable state after 400 ns, so we used the last 100 ns of the MD simulation for subsequent analyses. Int. J. Mol. Sci. 2020, 21, x FOR PEER REVIEW 4 of 17

PZM21_sys (2.0 Å), TRV130_sys (1.3 Å), and Morphine_sys (0.3 Å). These fluctuations occurred because the molecular structure of PZM21 has a longer side chain, which is more flexible and prone to fluctuation. The molecular structure of morphine contains five rings of a morphinan core that is relatively rigid and does not undergo fluctuation readily. Overall, the three systems eventually reached a relatively stable state after 400 ns, so we used the last 100 ns of the MD simulation for Int. J. Mol. Sci. 2020, 21, 4699 4 of 16 subsequent analyses.

Figure 3. (a) Root-mean-square deviation (RMSD) of the protein backbone and (b) RMSD of the ligand. Figure 3. (a) Root-mean-square deviation (RMSD) of the protein backbone and (b) RMSD of the 2.3. Residue D1473.32 and H2976.52 in the Ligand-Bindingligand. Pocket An interaction between the morphinan tertiary amine cation of the ligand and the polar residue 3.32 6.52 2.3. D147Residue3.32 D147was found and in H297 the crystal in the structure Ligand-Binding of the activated Pocket MOR bound to the agonist BU72 and theAn crystal interaction structure between of the inactivethe morphinan MOR [18 tertiary] bound am to theine antagonistcation of theβ-funaltrexamine. ligand and the Basedpolar onresidue 6.52 D147this3.32 interaction,was found compoundin the crystal PZM21 structure was discovered of the activated by means MOR of molecular bound to docking the agonist [14]. BU72 H297 and the crystal(also structure located in of the the active inactive pocket) MOR formed [18] a hydrogenbound to bond the antagonist with the phenolic β-funaltrexamine. hydroxyl group Based of BU72 on this or generated a water-mediated network of hydrogen bonds. In the activated pocket, this hydrogen interaction, compound PZM21 was discovered by means of molecular docking [14]. H2976.52 (also bond network extended to Y1483.33 [19]. To estimate the aﬃnity between the ligands morphine, locatedPZM21, in the TRV130, activeand pocket) protein, formed we calculated a hydrogen the bond proportion with ofthe these phenolic hydrogen hydroxyl bonds group in the entireof BU72 or generatedMD simulation, a water-mediated including D147 network3.32 with of ligandshydrogen and bo waternds. (Table In the1). activated At least one pocket, hydrogen this bond hydrogen bondwas network formed betweenextended D147 to 3.32Y148and3. 33 PZM21 [19]. To/TRV130, estimate so both the ofaffinity them could between bind closelythe ligands to the MOR.morphine, PZM21,The hydrogen TRV130, bond and betweenprotein, D147 we3.32 calculatedand morphine the pr accountedoportion for of only these 6.58%, hydrogen and the bonds hydrogen in bondthe entire MDbroke simulation, rapidly including during the D147 MD simulation.3.32 with ligands To conﬁrm and water this result (Table intuitively, 1). At least we one aligned hydrogen the docked bond was formedcomplex between and simulated D1473.32 complexand PZM21/TRV130, (Figure4). In the so docked both of result, them the could distance bind between closely morphine to the MOR. and The 3.32 hydrogenD147 bondwas 1.5between Å, and D147 after3.32 a period and morphine of dynamic accounted simulation, for the only distance 6.58%, was and increased the hydrogen to 5.9 Å. bond brokeThis rapidly result was during due tothe the MD fact simulation. that morphine To penetrates confirm deepthis result into the intuitively, pocket and formswe aligned a more the stable docked 6.52 complexhydrogen and bondsimulated with H297 complex. Conversely, (Figure 4). PZM21 In the docked and TRV130 result, did the not distance move deep between inside themorphine pocket and during the simulation. The distance between D1473.32 and PZM21/TRV130 was kept at a short range, D1473.32 was 1.5 Å, and after a period of dynamic simulation, the distance was increased to 5.9 Å. and the hydrogen bond could continue to be formed. Moreover, PZM21 also formed a stable hydrogen This result was due to the fact that morphine penetrates deep into the pocket and forms a more bond with H2976.52. These results indicated that these ligands were targeted to the MOR. stable hydrogen bond with H2976.52. Conversely, PZM21 and TRV130 did not move deep inside the pocket during the simulation. The distance between D1473.32 and PZM21/TRV130 was kept at a short

Int. J. Mol. Sci. 2020, 21, 4699 5 of 16

Int. J. Mol. Sci. 2020, 21, x FOR PEER REVIEW 5 of 17 Table 1. Occupancy of hydrogen bonds between D3.32 and water, and D3.32 and ligands, in each system. range,Data wereand the calculated hydrogen by bond Groningen could Machinecontinue Chemicalto be formed. Simulations Moreover, (GROMACS). PZM21 also Determinationformed a stable of hydrogen bond with H2976.52. These results indicated that these ligands were targeted to the MOR. hydrogen bonding is based on the conditions of α 30 and r 0.35 nm. ≤ ◦ ≤ Table 1. Occupancy of hydrogen bonds between D3.32 and water, and D3.32 and ligands, in each H-bond D3.32_H2O D3.32_Ligand system. Data were calculated by Groningen Machine Chemical Simulations (GROMACS). Number Morphine (%) PZM21 (%) TRV130 (%) Morphine (%) PZM21 (%) TRV130 (%) Determination of hydrogen bonding is based on the conditions of α ≤ 30° and r ≤ 0.35 nm. 0 a 0.04% 0.23% 93.42% 17.43 − − H-bond1 0.03D3.32_H 3.402O 1.18 5.62 D3.32_Ligand 1.18 65.44 Number2 Morphine 0.35 (%) PZM21 51.47(%) TRV130 1.85(%) Morphine 0.96 (%) PZM21 41.68(%) TRV130 17.14(%) a 03 − 3.78 0.04% 35.85 0.23% 3.67 93.42% − 40.03 17.43 − − 14 0.03 31.79 3.40 7.49 1.18 20.90 5.62 1.1816.09 65.44 − − 25 0.35 47.21 51.47 1.58 1.85 52.27 0.96 41.681.02 17.14 3 3.78 35.85 3.67 − − 40.03 − − 6 14.19 0.16 15.36 4 31.79 7.49 20.90 − − 16.09 − − − 7 2.43 0.01 4.00 5 47.21 1.58 52.27 − − 1.02 − − − 8 0.20 0.52 6 14.19 0.16 − 15.36 − − − − − − 9 0.02 0.02 7 2.43 0.01 − 4.00 − − − − − − 8 0.20a The symbol “ − ” denotes no related 0.52 interaction with − a hydrogen −bond. − 9 0.02 − 0.02 − − − a The symbol “−” denotes no related interaction with a hydrogen bond.

Figure 4. Cont.

Int. J. Mol. Sci. 2020, 21, 4699 6 of 16 Int. J. Mol. Sci. 2020, 21, x FOR PEER REVIEW 6 of 17

Figure 4. Superposition of the docked structure and simulated structure of the three systems and an Figure 4. Superposition of the docked structure and simulated structure of the three systems and an image of the ligands and partial residues in the binding pocket. (a) Docked complex/blue vs. image of the ligands and partial residues in the binding pocket. (a) Docked complex/blue vs. simulated simulated complex/gray in Morphine_sys. (b) Docked complex/blue vs. simulated complex/orange complex/gray in Morphine_sys. (b) Docked complex/blue vs. simulated complex/orange in PZM21_sys. in PZM21_sys. (c) Docked complex/blue vs. simulated complex/green in TRV130_sys. (c) Docked complex/blue vs. simulated complex/green in TRV130_sys.

2.4. Residues W2936.48, Y3267.43, W3187.35, and Y3367.53 Aﬀect Receptor Function Studies have shown that W6.48 is an “activation switch” [20] that activates “water channels”; W6.48 has an important eﬀect on activation of opioid receptors. Conformational changes of W6.48 in the δ-opioid receptor regulate the conformation of transmembrane 5 (TM5) and TM6, and they transmit signals to the intracellular loop region to upregulate the selectivity of the G protein [21]. Int. J. Mol. Sci. 2020, 21, x FOR PEER REVIEW 7 of 17

Studies have shown6.48 that7.43 W6.487.35 is an “activation7.53 switch” [20] that activates “water channels”; Int. J. Mol.2.4. Residues Sci. 2020, W29321, 4699 , Y326 , W318 , and Y336 Affect Receptor Function 7 of 16 W6.48 has an important effect on activation of opioid receptors. Conformational changes of W6.48 in the δ-opioid receptor regulate the conformation of transmembrane 5 (TM5) and TM6, and they Hence,transmit we focused signals onto studythe intracellular of the key loop residue region W6.48 to upregulate and calculated the selectivity the change of the in G the protein torsion [21]. angle of Hence, we focused on study of the key residue W6.48 and calculated the change in the torsion angle the W2936.48 side chain. of the W2936.48 side chain. The W2936.48 side chains in PZM21_sys and Morphine_sys were finally reversed (Figure5a). The W2936.48 side chains in PZM21_sys and Morphine_sys were finally reversed (Figure 5a). ThatThat reversal reversal occurred occurred in in Morphine_sys Morphine_sys at at 5050 ns, but it it occurred occurred in inPZM21_sys PZM21_sys at 300 at 300ns. A ns. statistical A statistical 6.48 histogramhistogram for thefor the W293 W2936.48frequencyfrequency (Figure(Figure 66)) shows that that the the reversed reversed angle angle in Morphine_sys in Morphine_sys was was stablestable at approximately at approximately 76 76°,◦, whereas whereas in in PZM21_sysPZM21_sys it it was was stable stable at atapproximately approximately 62°. 62We◦ .deduced We deduced thatthat the ditheff erencedifference in variationin variation of of the the W293 W2936.486.48 sideside chain chain between between Morp Morphine_syshine_sys and andPZM21_sys PZM21_sys mightmight be anbe importantan important factor factor leading leading to to a higher dose dose of of PZM21 PZM21 than than morphine morphine to achieve to achieve similar similar agonistagonist ability. ability. This This conclusion conclusion isis consistent with with the the study study of Hill of Hillet al. et[15]. al. The [15 ].torsion The torsionangle of the angle of 6.48 the W293W2936.48sideside chain chain of of TRV130_sys TRV130_sys barely barely changed changed throughout throughout the thesimulation. simulation. Overlap Overlap of the of the docking structure and simulated structure revealed that a translation to the binding pocket occurred docking structure and simulated structure revealed that a translation to the binding pocket occurred in in W2936.48, so TRV130 may differ from other ligands in terms of activating the MOR. W2936.48, so TRV130 may differ from other ligands in terms of activating the MOR.

Figure 5. Change in torsion angle of residues (a) W2936.48 and (b) Y3267.43 in three systems.

Figure 5. Change in torsion angle of residues (a) W2936.48 and (b) Y3267.43 in three systems. Int. J. Mol. Sci. 2020, 21, 4699 8 of 16 Int. J. Mol. Sci. 2020, 21, x FOR PEER REVIEW 8 of 17

Morphine 18 PZM21 16

6 Probability(%) 4

0 30 60 90 120 6.48 Torsion angle_W293 6.48 FigureFigure 6. Distribution 6. Distribution of theof the torsion torsion angle angle of of residue residue W2936.48 inin Morphine_sys Morphine_sys and and PZM21_sys. PZM21_sys.

7.43 7.35 Moreover,Moreover, Y326 Y326and7.43 and W318 W3187.35also also hadhad a certain effect effect on on the the bi biasas of ofthe the agonist. agonist. In the In Y7.43F the Y7.43F mutantmutant system, system, [D-Ala2,N-MePhe4,Gly-ol]-enkephalin [D-Ala2,N-MePhe4,Gly-ol]-enkephalin (DAMGO) (DAMGO) completelycompletely loses loses the the recruitment recruitment of β-arrestin-2,of β-arrestin-2, but in but the in W7.35A the W7.35A mutant mutant system, system, it it increases increases the relative relative activity activity of β of-arrestin-2β-arrestin-2 [22]. [22]. We calculatedWe calculated the changethe change in torsionin torsion angle angle of of the the Y326 Y3267.437.43 sideside chain chain and and root-mean-square root-mean-square fluctuation fluctuation 7.43 (RMSF)(RMSF) of residues. of residues. Y326 Y3267.43 in Morphine_sys Morphine_sys had had almost almost no turning, no turning, and the and deflection the deflection angle was angle stable at about −75° (Figure 5b). The Y3267.437.43 side chain in PZM21_sys and TRV130_sys was was stable at about 75◦ (Figure5b). The Y326 side chain in PZM21_sys and TRV130_sys was deflected, and the− deflection angle was stable at approximately 125°. The Y3267.43 side chain of deflected, and the deflection angle was stable at approximately 125 . The Y3267.43 side chain of TRV130_sys fluctuated from −100° to 150°, and in PZM21_sys, it fluctuated◦ from 100° to 175° (Figure TRV130_sys7). The Y326 fluctuated7.43 sidefrom chain of100 TRV130_sys◦ to 150◦, andwas inmore PZM21_sys, active than it that fluctuated in PZM21_sys. from 100 Figure◦ to 175 8 exhibits◦ (Figure 7). 7.43 − The Y326the RMSFside of chainthe MOR of TRV130_sys backbone in was three more systems, active and than it that shows in PZM21_sys. that W3187.35 Figure in TRV130_sys8 exhibits the RMSFfluctuated of the MOR the most, backbone but it inmaintained three systems, similar and stability it shows in Morphine_sys that W3187.35 andin PZM21_sys. TRV130_sys The fluctuated MD the most,simulation but it maintainedof Cheng et al. similar [23] showed stability that in the Morphine_sys stability of Y326 and7.43 PZM21_sys. and W3187.35The was MDrelated simulation to the of Chengbias et al.of the [23 ]agonist.showed Hence, that the the stability changes of of Y326 these7.43 residuesand W318 in our7.35 simulationwas related results to the confirmed bias of the that agonist. Hence,PZM21 the changes and TRV130 of these had a residues certain degree in our of simulationbias. However, results the bias confirmed of PZM21 that was PZM21 not as strong and TRV130 as had athat certain of TRV130, degree and ofbias. these However,data are in the accordance bias ofPZM21 with the was experimental not as strong results as of that Hill of et TRV130, al. [15]. and Furthermore, Y3367.53 in TRV130_sys and PZM21_sys had similar fluctuations of the side-chain these data are in accordance with the experimental results of Hill et al. [15]. Furthermore, Y3367.53 in torsion angle ranging from approximately 75° to 125°, but TRV130_sys was more stable than TRV130_sys and PZM21_sys had similar fluctuations of the side-chain torsion angle ranging from PZM21_sys. Fluctuations of the side-chain torsion angle in Morphine_sys ranged from −80° to 175°, approximatelyand these fluctuations 75◦ to 125 ◦were, but in TRV130_sys a wide range. was Y336 more7.53 was stable located than in PZM21_sys. the G protein-binding Fluctuations region, of the side-chain torsion angle in Morphine_sys ranged from 80 to 175 , and these fluctuations were in a and the distribution of the side-chain torsion angle indicated− ◦ that the◦ mechanism of the three ligands wide range.affecting Y336 the 7.53downstreamwas located pathway in the of G MOR protein-binding activation was region, different. and the distribution of the side-chain torsion angle indicated that the mechanism of the three ligands affecting the downstream pathway of MOR activation was different.

2.5. The Water Channel Associated with Activation Studies have shown that a ligand located in the orthosteric site can mediate the entry of water molecules into the G protein-coupled receptor to cause receptor activation [20,24]. An image of the equilibrated conformations of the three systems (Figure9) revealed that the water molecules entering the transmembrane (TM) region were the most in Morphine_sys, followed by TRV130_sys, and finally PZM21_sys. These data were validated when the statistics of the hydrogen-bond ratio were calculated (Table1). Compared with PZM21_sys, TRV130_sys and Morphine_sys had more hydrogen bonds between the residue of the binding pocket and water. The solvent-accessible surface area (SASA) of the MOR TM region can reflect the activation effect of the ligand on the receptor. Thus, we calculated the SASA of the MOR TM region in the three systems (Figure 10). The SASA had an upward trend in the three systems. Eventually, Morphine_sys reached 14.800 Å2, TRV130_sys reached 14.600 Å2, Int. J. Mol. Sci. 2020, 21, 4699 9 of 16 and PZM21_sys reached 14.400 Å2. These results were in accordance with the data shown in Figure9, andInt. indicatedInt. J. J.Mol. Mol. Sci. Sci.that 2020 2020, the21, 21, x, abilityxFOR FOR PEER PEER to REVIEW activateREVIEW the MOR followed the order of morphine, TRV130 and9 9of of 17PZM21. 17

7.43 7.53 FigureFigureFigure 7. Distribution 7. 7. Distribution Distribution of of theof the the torsion torsion torsion angleangle angle of of residues residues residues( (a(a)a ))Y326 Y3267Y3267.43 .43and andand ( b(b) )Y336 ( bY336) Y3367.53 in in7.53 three threein systems. three systems. systems.

Figure 8. The root-mean-square ﬂuctuation (RMSF) of the MOR backbone in three systems. FigureFigure 8. 8. The The root-mean-square root-mean-square fluctuation fluctuation (RMSF) (RMSF) of of the the MOR MOR backbone backbone in in three three systems. systems.

Int.Int. J. J. Mol. Mol. Sci. Sci. 2020 2020, ,21 21, ,x x FOR FOR PEER PEER REVIEW REVIEW 10 10 of of 17 17

2.5.2.5. TheThe WaterWater ChannelChannel AssociatedAssociated withwith ActivationActivation StudiesStudies havehave shownshown thatthat aa ligandligand locatedlocated inin thethe orthostericorthosteric sitesite cancan mediatemediate thethe entryentry ofof waterwater moleculesmolecules intointo thethe GG protein-coupledprotein-coupled receptorreceptor toto cacauseuse receptorreceptor activationactivation [2[20,24].0,24]. AnAn imageimage ofof thethe equilibratedequilibrated conformationsconformations ofof ththee threethree systemssystems (Figure(Figure 9)9) revealedrevealed thatthat thethe waterwater moleculesmolecules enteringentering thethe transmembranetransmembrane (TM)(TM) regionregion werewere ththee mostmost inin Morphine_sys,Morphine_sys, followedfollowed byby TRV130_sys,TRV130_sys, andand finallyfinally PZM21_sys.PZM21_sys. TheseThese datadata werewere validatedvalidated whenwhen thethe statisticsstatistics ofof thethe hydrogen-bondhydrogen-bond ratioratio werewere calculatedcalculated (Table(Table 1).1). ComparedCompared withwith PZM21_sys,PZM21_sys, TRV130_sysTRV130_sys andand Morphine_sysMorphine_sys hadhad moremore hydrogenhydrogen bondsbonds betweenbetween thethe residueresidue ofof thethe bindbindinging pocketpocket andand water.water. TheThe solvent-accessiblesolvent-accessible surfacesurface areaarea (SASA)(SASA) ofof thethe MORMOR TMTM regionregion cancan reflreflectect thethe activationactivation effecteffect ofof thethe ligandligand onon thethe receptor.receptor. Thus,Thus, wewe calculatedcalculated thethe SASASASA ofof thethe MORMOR TMTM regionregion inin thethe threethree systemssystems (Figure(Figure 10).10). TheThe 2 SASASASA hadhad anan upwardupward trendtrend inin thethe threethree systsystems.ems. Eventually,Eventually, Morphine_sysMorphine_sys reachedreached 14800Å14800Å2,, 2 2 TRV130_sysTRV130_sys reachedreached 14.60014.600 ÅÅ2,, andand PZM21_sysPZM21_sys reachedreached 14.40014.400 ÅÅ2.. TheseThese resultsresults werewere inin accordanceaccordance withwith thethe datadata shownshown inin FigureFigure 9,9, andand indicatedindicated thatthat thethe abilityability toto activateactivate thethe MORMOR followedfollowed thethe

order of morphine, TRV130 and PZM21. order of morphine, TRV130 and PZM21. Int. J. Mol. Sci. 2020 21 , , 4699 10 of 16

FigureFigureFigure 9. Image 9.9. ImageImage showing showingshowing the thethe equilibratedequilibratedequilibrated structstruct structureureure ofof the ofthe the“water“water “water channel”channel” channel” inin ((aa)) Morphine_sys,Morphine_sys, in (a) Morphine_sys, ((bb)) (b) PZM21_sys,PZM21_sys,PZM21_sys, andand and ((ccc))) TRV130_sys. TRV130_sys. Each Each Each red red red dot dot dot represents represents represents a a water water a water molecule. molecule. molecule.

1520015200 PZM21 PZM21 Morphine 1500015000 Morphine TRV130 TRV130 1480014800

1460014600 ) 2 )

2 14400

Å 14400 Å

1420014200 SASA ( SASA SASA ( SASA 1400014000

1380013800

1360013600

1340013400 00 50 50 100 100 150 150 200 200 250 250 300 300 350 350 400 400 450 450 500 500 Time (ns) Time (ns) Figure 10.FigureFigureSolvent-accessible 10. 10. Solvent-accessible Solvent-accessible surface surface surface area (SASA)areaarea (SASA) (SASA) of the of of transmembranethethe transmembrane transmembrane region region region of of of the the the MOR MOR MORin in in di ﬀerent systems with time. differentdifferent systems systems with with time. time.

2.6. Flexibility and Conformational Change in the Loop Area The RMSF of the three systems is shown in Figure8, and the overall fluctuation was similar. The less flexible parts were the relatively rigid TM regions, and the seven regions with lower RMSF values can be seen in Figure8. Seven-segment TM helices were represented, and the residue numbers were 65–96 (TM1); 101–130 (TM2); 136–171 (TM3); 180–106 (TM4); 225–262 (TM5); 268–305 (TM6); and 311–339 (TM7). The flexible regions with large RMSF values corresponded to the intracellular and extracellular loop regions of the MOR: 97–100 (Intracellular 1, I1), 131–135 (Extracellular 1, E1), 172–179 (Intracellular 2, I2), 207–224 (Extracellular 2, E2), 263–267 (Intracellular 3, I3) and 306–310 (Extracellular 3, E3), respectively. 52Gly and 347Phe, which locate the N-terminus and C-terminus of the receptor, respectively, had greater flexibility and higher RMSF values compared with other regions. In terms of overall similarity, there were subtle differences in the different systems due to the different ligands. Morphine_sys and PZM21_sys had similar RMSF distributions and similar dynamic characteristics, but the RMSF of TRV130_sys was larger than that of the other two systems, especially in the E1, E2, I3, and E3 regions. These dynamic differences can elicit different biologic effects by modulating the interaction between the receptor and a downstream signal-transduction protein such as a G protein or β-arrestin. Studies have shown that E2 and E3 have important roles in G-protein selectivity [5], and that I3 has a hydrophobic interaction with Gi or a polar interaction with the sixth β-sheet of Gα in the MOR. Therefore, our simulation results showed that TRV130, Int. J. Mol. Sci. 2020, 21, 4699 11 of 16 morphine, and PZM21 had different effects on the MOR, whereas PZM21 exhibited similar movements to that of morphine in some respects. In order to better understand the effects of morphine, PZM21, and TRV130 on the conformational state of MOR, we calculated the dynamic cross-correlation matrix (DCCM) of the Cα atom of the receptor throughout the entire simulated trajectory (Figure 11). The red and yellow parts were called the “positive regions”, and they represented positive correlations between the residues (i.e., the dynamic movements of the residues were in the same direction). The blue portion was called the “negative region” and represented strong negative correlations between the residues (i.e., the movements of the residues were opposite). We noted few negative regions in Morphine_sys containing the non-biased agonist, and many negative regions in the system in which the bias agonist TRV130 was present. A negative-correlation motion of PZM21_sys was between Morphine_sys and TRV130_sys. These results suggested that (1) morphine, PZM21, and TRV130 had different effects on the local dynamics and conformation of the receptor protein; (2) the biased agonist could cause more negative correlationInt. J. Mol. movement Sci. 2020, 21 of, x theFOR receptorPEER REVIEW protein compared with unbiased agonist morphine. 12 of 17

FigureFigure 11. Dynamic 11. Dynamic cross-correlation cross-correlation matrix matrix diagram diagram forfor each system. system.

3. Materials3. Materials and Methods and Methods

3.1. Preparation3.1. Preparatio andn Molecular and Molecular Docking Docking The crystalThe crystal structure structure of the of activated the activated MOR MOR (PDB: (PDB 5C1M: 5C1M [19]) [19]) was was downloaded downloaded from from the the Protein Protein Data Bank (www.rcsb.orgData Bank (www.rcsb.org).). The crystal The structure crystal structure bound withbound the with agonist the agonist BU72 containsBU72 contains a nanobody a nanobody (Nb39) to stabilize(Nb39) the to MOR stabilize structure. the MOR To carrystructure. out MDTo carry simulations out MD on simulations the wild-type on the MOR, wild-type we removed MOR, we Nb39 and mutatedremoved the Nb39 non-standard and mutated amino the non-standard acid YCM back amino to cysteine.acid YCM We back left to onlycysteine. the ligandWe left BU72 only the at the ligand BU72 at the orthosteric site as a binding site for the subsequent molecular-docking study. Then, we complemented missing amino acids and loop fragments in the crystal. The three-dimensional structures of morphine, TRV130, and PZM21 were constructed in Discovery Studio 2.5 [25]. Ligands were prepared by the “Prepare Ligands” module to obtain the lowest energy and more stable ligand structure. First, we used Maestro v10.1 (Schrödinger, New York, NY, USA) for docking studies. Initially, “Protein Preparation Wizard” was used to examine and optimize the protein. The “PROPKA” tool was employed to adjust the protein to a physiologic pH = 7 environment. Hydrogen atoms were added to the structure, and a Monte Carlo multiple minimum conformation search was done under the OPLS_2005 force field. The “Ligprep” tool is used to neutralize ligands and determine chiralities from 3D structure. Second, the ligand BU72 in the protein structure was set to the center of the binding-site box, the radius of which was 15 Å. Finally, the three ligands were docked into the Int. J. Mol. Sci. 2020, 21, 4699 12 of 16 orthosteric site as a binding site for the subsequent molecular-docking study. Then, we complemented missing amino acids and loop fragments in the crystal. The three-dimensional structures of morphine, TRV130, and PZM21 were constructed in Discovery Studio 2.5 [25]. Ligands were prepared by the “Prepare Ligands” module to obtain the lowest energy and more stable ligand structure. First, we used Maestro v10.1 (Schrödinger, New York, NY, USA) for docking studies. Initially, “Protein Preparation Wizard” was used to examine and optimize the protein. The “PROPKA” tool was employed to adjust the protein to a physiologic pH = 7 environment. Hydrogen atoms were added to the structure, and a Monte Carlo multiple minimum conformation search was done under the OPLS_2005 force ﬁeld. The “Ligprep” tool is used to neutralize ligands and determine chiralitiesInt. J. Mol. from Sci. 3D2020 structure., 21, x FOR PEER Second, REVIEW the ligand BU72 in the protein structure was set to the 13 centerof 17 of the binding-site box, the radius of which was 15 Å. Finally, the three ligands were docked into the receptorreceptor using using Glide Glide 5.8 5.8 XP XP [26 [26].]. The The conformation conformation results were were arranged arranged in inorder order of scoring, of scoring, and a and a reasonablereasonable (i.e., (i.e., the the docking docking score score was was high high and and thethe key interaction interaction between between the the ligand ligand and andreceptor receptor was present)was present) complex complex conformation conformation was was selected selected as as the the initial initial structure. structure. ToTo ensureensure the reliability of of our our docking results, we used the same method to dock BU72 in the crystal structure to the receptor docking results, we used the same method to dock BU72 in the crystal structure to the receptor with a with a root-mean-square deviation (RMSD) of 0.496 Å. Therefore, our docking results were credible. root-mean-square deviation (RMSD) of 0.496 Å. Therefore, our docking results were credible. 3.2. Construction of the Simulation System 3.2. Construction of the Simulation System The simulation system consisted of 1-palmitoyl-2-oleoylsn-glycero-3-phosphocholine (POPC), Thethe MOR, simulation and ligands system (Figure consisted 12). ofThe 1-palmitoyl-2-oleoylsn-glycero-3-phosphocholine membrane was composed of two layers of POPC lipid (POPC), the MOR,molecules, and ligands each of (Figure which had12). the The same membrane initial conformation was composed (the of hydrophobic two layers of tail POPC of the lipid molecule molecules, each ofwas which straight). had the same initial conformation (the hydrophobic tail of the molecule was straight).

Figure 12. Sketch map of the simulation system including the µ-opioid receptor (MOR), lipid membranes, Figure 12. Sketch map of the simulation system including the μ-opioid receptor (MOR), lipid and water molecules. For clearer display, some water molecules and the ligand are hidden. membranes, and water molecules. For clearer display, some water molecules and the ligand are hidden. First, the complexes of the MOR and ligand were inserted into a lipid membrane composed of 89 POPCFirst, lipid the molecules,complexes of and the theMOR major and ligand axis of were the inserted MOR was into made a lipid substantially membrane composed perpendicular of to the89 membrane POPC lipid surface.molecules, Second, and the major some axis of theof the amino MORacids was made in the substantially protein were perpendicular charged, to so we utilizedthe NaClmembrane solution surface. (0.15 Second, M) to some maintain of the amino the neutrality acids in the of protein the system. were charged, Finally, so the we systemutilized was solvatedNaCl by solution adding (0.15 a certain M) to maintain number the of waterneutrality molecules of the system. of the Finally, TIP3P modelthe system throughout was solvated the system.by Eachadding simulation a certain system number had a of size water of about molecules 8 8 of10 the nm TIP3P3 and model contained throughout approximately the system. 40.000 Each atoms. × × 3 Constructionsimulation of system the entire had a simulationsize of about system 8 × 8 × was10 nm completed and contained using approximately the CHARMM-GUI 40.000 atoms. Internet serverConstruction [27–29](www.charmm-gui.org of the entire simulation), and system the simulationwas completed details using are the shown CHARMM-GUI in Table2. Internet server [27–29] (www.charmm-gui.org), and the simulation details are shown in Table 2.

Table 2. Simulations systems used in this study.

Number Simulation Name of System POPC TIP3P Na+ Cl− Ligand Receptor Time/ns Morphine_MOR 89 7764 31 45 Morphine MOR 500 PZM21_ MOR 89 7765 31 45 PZM21 MOR 500 TRV130_ MOR 89 7003 18 32 TRV130 MOR 500

Int. J. Mol. Sci. 2020, 21, 4699 13 of 16

Table 2. Simulations systems used in this study.

Number Name of System Simulation Time/ns + POPC TIP3P Na Cl− Ligand Receptor Morphine_MOR 89 7764 31 45 Morphine MOR 500 PZM21_MOR 89 7765 31 45 PZM21 MOR 500 TRV130_MOR 89 7003 18 32 TRV130 MOR 500

3.3. MD Simulations The conduction of MD simulations were proceeded with software GROMACS 5.1.4. (manual. gromacs.org/). The CHARMM36 force field [30,31] was described for the protein and membrane. The Charmm General Force Field (CGenFF) via the Paramchem Internet site was described for parametrized ligands. van der Waals interactions were calculated using a double cutoff radius of 10–12 Å. The long-range electrostatic force was calculated using the Particle–Mesh–Ewald method [32]. Using periodic boundary conditions, hydrogen atoms were constrained by the LINear Constraint Solver (LINCS) algorithm [33]. The time step was 2 fs. First, 5000-step energy optimization was undertaken by the steepest-descent method to avoid the local irrationality of the system. Second, due to the particularity of the membrane system, a special equilibrium scheme was adopted. The Bernendson method [34] is used to control the temperature at 310K while maintaining a constant pressure of 1 atm for a conical ensemble (NVT) and isobaric–isothermal ensemble (NPT) ensemble. In this process, the backbone and side chains of the protein and ligands are given a different constraint potential in each step. Finally, all the constraints are removed, and the system is released completely for 1 ns pre-equalization. The equilibration time and the constraint force parameters are shown in Table3. The temperature and pressure of this process were controlled by the Nosé–Hoover method [35] and Parrinello–Rahman method [36], respectively. Finally, each system was subjected to a MD simulation of 500 ns.

Table 3. Equilibration time and constraint force parameters of each step. NPT: isobaric–isothermal ensemble, NVT: conical ensemble.

2 Time Steps Equilibration Force Constant/KJ/mol nm− Step Ensemble · /fs Time/ps Protein Backbone Protein Side Chain Ligand 1 NVT 1 25 4000 2000 4000 2 NVT 1 25 2000 1000 2000 3 NPT 1 25 1000 500 1000 4 NPT 2 100 500 200 500 5 NPT 2 100 200 50 200 6 NPT 2 100 50 0 50 7 Pre-Production 2 1000 0 0 0

4. Conclusions The development of safe and efficacious analgesic agents with few side effects is an urgent need. The common feature of such analgesics is that they are biased toward a G-protein pathway. The biased properties of compound PZM21 and its analgesic activity are controversial. We investigated the effects of the biased agonist TRV130, unbiased agonist morphine, and compound PZM21 on the MOR by means of MD simulations. Our results enabled five main conclusions to be drawn. First, morphine, PZM21, and TRV130 formed stable hydrogen bonds with residues D1473.32 and H2976.52 to aid ligand binding to the MOR. All ligands targeted the MOR. Second, we analyzed some key residues associated with receptor selectivity and noted two main trends. The first trend we noticed was that the side chain of W2936.48 was reversed in PZM21_sys and Morphine_sys, but it reversed slowly in PZM21_sys. This may be an important factor why PZM21 is a Int. J. Mol. Sci. 2020, 21, 4699 14 of 16 dose-dependent agonist. The second trend we identified was that the Y3267.43 side chain rotated to a certain degree in PZM21_sys and TRV130_sys, but that it had a larger fluctuation range and was more active in TRV130_sys. Y3367.53 in PZM21_sys was more active and W3187.35 in TRV130_sys was more active compared with PZM21_sys. Rotation of the side chains in Y3267.43 and Y3367.53 was similar but slightly different in TRV130_sys and PZM21_sys. These data may imply that they were biased, but PZM21 was less biased than TRV130. Third, the extent of openness of the water channel affected the analgesic activity of the ligand. The SASA of Morphine_sys was the largest, followed by TRV130, and finally PZM21. Therefore, the analgesic activity of PZM21 merits additional study. Fourth, by analyzing the flexibility of the loop region, the fluctuations of Morphine_sys and PZM21_sys in the E1, E2, I3, and E3 regions were significantly smaller than those of TRV130_sys. In addition, PZM21 exhibited similar dynamic properties to those of morphine in these regions. Finally, DCCM analyses revealed that negative correlation movements between residues in the biased agonist system were more common compared with the unbiased agonist system, whereas the negative correlation between the residues of PZM21_sys was between Morphine_sys and TRV130_sys. In summary, our simulation work indicated that PZM21 did not appear to have an obvious advantage such as that of TRV130 in terms of analgesia or fewer side effects. From its dynamic behavior, PZM21 was a low-potency analgesic and a low-potential biased agonist. These data are in good agreement with the experimental results of Hill et al. [15]. We also summarized some movements related specifically to biased agonists, which lay the foundation for the design of morphine substitutes with fewer side effects (i.e., G protein-biased agonists).

Author Contributions: Conceptualization and supervision, J.L.; MD simulations, Z.Z.; data analyses, Z.Z. and T.H. All authors analyzed the results and approved the final version of the manuscript for submission. Funding: This research received no external funding. Acknowledgments: We acknowledge support for computing resources by Gansu Computing Center and the High-Performance Computing Center within Lanzhou University. Conflicts of Interest: The authors declare no conflict of interest.

References

1. Stefano, G.B.; Pilonis, N.; Ptacek, R.; Kream, R.M. Reciprocal Evolution of Opiate Science from Medical and Cultural Perspectives. Med. Sci. Monit. Int. Med. J. Exp. Clin. Res. 2017, 23, 2890–2896. [CrossRef] 2. Skolnick, P. The Opioid Epidemic: Crisis and Solutions. Annu. Rev. Pharmacol. Toxicol. 2018, 58, 143–159. [CrossRef] 3. Park, K.; Otte, A. Prevention of Opioid Abuse and Treatment of Opioid Addiction: Current Status and Future Possibilities. Annu. Rev. Biomed. Eng. 2019, 21, 61–84. [CrossRef] 4. World Health Organization. Available online: https://www.who.int/substance_abuse/information-sheet/en/ (accessed on 19 November 2018). 5. Seyedabadi, M.; Ghahremani, M.H.; Albert, P.R. Biased signaling of G protein coupled receptors (GPCRs): Molecular determinants of GPCR/transducer selectivity and therapeutic potential. Pharmacol. Ther. 2019, 200, 148–178. [CrossRef] 6. Bohn, L.M.; Gainetdinov, R.R.; Lin, F.T.; Lefkowitz, R.J.; Caron, M.G. Mu-opioid receptor desensitization by beta-arrestin-2 determines morphine tolerance but not dependence. Nature 2000, 408, 720–723. [CrossRef] [PubMed] 7. Bohn, L.M.; Lefkowitz, R.J.; Gainetdinov, R.R.; Peppel, K.; Caron, M.G.; Lin, F.T. Enhanced Morphine Analgesia in Mice Lacking β-Arrestin 2. Science 1999, 286, 2495–2498. [CrossRef] 8. Raehal, K.M.; Walker, J.K.; Bohn, L.M. Morphine Side Eﬀects in β-Arrestin 2 Knockout Mice. J. Pharmacol. Exp. Ther. 2005, 314, 1195–1201. [CrossRef] Int. J. Mol. Sci. 2020, 21, 4699 15 of 16

9. Váradi, A.; Marrone, G.F.; Palmer, T.C.; Narayan, A.; Szabó, M.R.; Le Rouzic, V.; Grinnell, S.G.; Subrath, J.J.; Warner, E.; Kalra, S.; et al. Mitragynine/Corynantheidine Pseudoindoxyls As Opioid Analgesics with Mu Agonism and Delta Antagonism, Which Do Not Recruit β-Arrestin-2. J. Med. Chem. 2016, 59, 8381–8397. [CrossRef][PubMed] 10. Schmid, C.L.; Kennedy, N.M.; Ross, N.C.; Lovell, K.M.; Yue, Z.; Morgenweck, J.; Cameron, M.D.; Bannister, T.D.; Bohn, L.M. Bias Factor and Therapeutic Window Correlate to Predict Safer Opioid Analgesics. Cell 2017, 171, 1165–1175. [CrossRef][PubMed] 11. Chen, X.T.; Pitis, P.; Liu, G.; Yuan, C.; Gotchev, D.; Cowan, C.L.; Rominger, D.H.; Koblish, M.; Dewire, S.M.; Crombie, A.L.; et al. Structure-activity relationships and discovery of a G protein biased mu opioid receptor ligand, [(3-methoxythiophen-2-yl)methyl]({2-[(9R)-9-(pyridin-2-yl)-6-oxaspiro-[4.5]decan-9-yl]ethyl})amine (TRV130), for the treatment of acute severe pain. J. Med. Chem. 2013, 56, 8019–8031. [CrossRef][PubMed] 12. DeWire, S.M.; Yamashita, D.S.; Rominger, D.H.; Liu, G.; Cowan, C.L.; Graczyk, T.M.; Chen, X.T.; Pitis, P.M.; Gotchev, D.; Yuan, C.; et al. A G protein-biased ligand at the mu-opioid receptor is potently analgesic with reduced gastrointestinal and respiratory dysfunction compared with morphine. J. Pharmacol. Exp. Ther. 2013, 344, 708–717. [CrossRef][PubMed] 13. Ok, H.G.; Kim, S.Y.; Lee, S.J.; Kim, T.K.; Huh, B.K.; Kim, K.H. Can oliceridine (TRV130), an ideal novel micro receptor G protein pathway selective (micro-GPS) modulator, provide analgesia without opioid-related adverse reactions? Korean J. Pain 2018, 31, 73–79. [CrossRef][PubMed] 14. Manglik, A.; Lin, H.; Aryal, D.K.; McCorvy, J.D.; Dengler, D.; Corder, G.; Levit, A.; Kling, R.C.; Bernat, V.; Hubner, H.; et al. Structure-based discovery of opioid analgesics with reduced side effects. Nature 2016, 537, 185–190. [CrossRef][PubMed] 15. Hill, R.; Disney, A.; Conibear, A.; Sutcliffe, K.; Dewey, W.; Husbands, S.; Bailey, C.; Kelly, E.; Henderson, G. The novel mu-opioid receptor agonist PZM21 depresses respiration and induces tolerance to antinociception. Br. J. Pharmacol. 2018, 175, 2653–2661. [CrossRef] 16. Ballesteros, J.A.; Weinstein, H. Integrated methods for the construction of three-dimensional models and computational probing of structure-function relations in G protein-coupled receptors. Methods Neurosci. 1995, 25, 366–428. 17. Sader, S.; Anant, K.; Wu, C. To probe interaction of morphine and IBNtxA with 7TM and 6TM variants of the human µ-opioid receptor using all-atom molecular dynamics simulations with an explicit membrane. Phys. Chem. Chem. Phys. PCCP 2018, 20, 1724–1741. [CrossRef] 18. Manglik, A.; Kruse, A.C.; Kobilka, T.S.; Thian, F.S.; Mathiesen, J.M.; Sunahara, R.K.; Pardo, L.; Weis, W.I.; Kobilka, B.K.; Granier, S. Crystal structure of the mu-opioid receptor bound to a morphinan antagonist. Nature 2012, 485, 321–326. [CrossRef] 19. Huang, W.; Manglik, A.; Venkatakrishnan, A.J.; Laeremans, T.; Feinberg, E.N.; Sanborn, A.L.; Kato, H.E.; Livingston, K.E.; Thorsen, T.S.; Kling, R.C.; et al. Structural insights into µ-opioid receptor activation. Nature 2015, 524, 315–321. [CrossRef] 20. Yuan, S.; Palczewski, K.; Peng, Q.; Kolinski, M.; Vogel, H.; Filipek, S. The mechanism of ligand-induced activation or inhibition of mu- and kappa-opioid receptors. Angew. Chem. 2015, 54, 7560–7563. [CrossRef] 21. Sun, X.; Laroche, G.; Wang, X.; Agren, H.; Bowman, G.R.; Giguere, P.M.; Tu, Y. Propagation of the Allosteric Modulation Induced by Sodium in the delta-Opioid Receptor. Chemistry 2017, 23, 4615–4624. [CrossRef] 22. Hothersall, J.D.; Torella, R.; Humphreys, S.; Hooley, M.; Brown, A.; McMurray, G.; Nickolls, S.A. Residues W320 and Y328 within the binding site of the mu-opioid receptor influence opiate ligand bias. Neuropharmacology 2017, 118, 46–58. [CrossRef][PubMed] 23. Cheng, J.X.; Cheng, T.; Li, W.H.; Liu, G.X.; Zhu, W.L.; Tang, Y.Computational insights into the G-protein-biased activation and inactivation mechanisms of the µ opioid receptor. Acta Pharmacol. Sin. 2017, 39, 154–164. [CrossRef][PubMed] 24. Yuan, S.; Filipek, S.; Palczewski, K.; Vogel, H. Activation of G-protein-coupled receptors correlates with the formation of a continuous internal water pathway. Nat. Commun. 2014, 5, 1–10. [CrossRef][PubMed] 25. Accelrys Discovery Studio; Version 2.5; Accelrys: San Diego, CA, USA, 2010. 26. Friesner, R.A.; Banks, J.L.; Murphy, R.B.; Halgren, T.A.; Klicic, J.J.; Mainz, D.T.; Repasky, M.P.; Knoll, E.H.; Shelley, M.; Perry, J.K.; et al. Glide: A New Approach for Rapid, Accurate Docking and Scoring. 1. Method and Assessment of Docking Accuracy. J. Med. Chem. 2004, 47, 1739–1749. [CrossRef][PubMed] Int. J. Mol. Sci. 2020, 21, 4699 16 of 16

27. Jo, S.; Kim, T.; Im, W. Automated builder and database of protein/membrane complexes for molecular dynamics simulations. PLoS ONE 2007, 2, e880. [CrossRef][PubMed] 28. Jo, S.; Kim, T.; Iyer, V.G.; Im, W. CHARMM-GUI: A web-based graphical user interface for CHARMM. J. Comput. Chem. 2008, 29, 1859–1865. [CrossRef] 29. Lee, J.; Cheng, X.; Swails, J.M.; Yeom, M.S.; Eastman, P.K.; Lemkul, J.A.; Wei, S.; Buckner, J.; Jeong, J.C.; Qi, Y.; et al. CHARMM-GUI Input Generator for NAMD, GROMACS, AMBER, OpenMM, and CHARMM/OpenMM Simulations Using the CHARMM36 Additive Force Field. J. Chem. Theory Comput. 2016, 12, 405–413. [CrossRef] 30. Klauda, J.B.; Venable, R.M.; Freites, J.A.; O’Connor, J.W.; Tobias, D.J.; Mondragon-Ramirez, C.; Vorobyov, I.; MacKerell, A.D.; Pastor, R.W. Update of the CHARMM All-Atom Additive Force Field for Lipids: Validation on Six Lipid Types. J. Phys. Chem. B 2010, 114, 7830–7843. [CrossRef] 31. Vanommeslaeghe, K.; Hatcher, E.; Acharya, C.; Kundu, S.; Zhong, S.; Shim, J.; Darian, E.; Guvench, O.; Lopes, P.; Vorobyov, I.; et al. CHARMM general force field: A force field for drug-like molecules compatible with the CHARMM all-atom additive biological force fields. J. Comput. Chem. 2010, 31, 671–690. [CrossRef] 32. Darden, T.; York, D.; Pedersen, L. Particle mesh Ewald: An N log(N) method for Ewald sums in large systems. · J. Chem. Phys. 1993, 98, 10089–10092. [CrossRef] 33. Hess, B.; Bekker, H.; Berendsen, H.J.; Fraaije, J.G. LINCS: A linear constraint solver for molecular simulations. J. Comput. Chem. 1997, 18, 1463–1472. [CrossRef] 34. Berendsen, H.J.C.; Postma, J.P.M.; Gunsteren, W.F.; DiNola, A.; Haak, J.R. Molecular dynamics with coupling to an external bath. J. Chem. Phys. 1984, 81, 3684–3690. [CrossRef] 35. Nosé, S. A molecular dynamics method for simulations in the canonical ensemble. Mol. Phys. 1984, 52, 255–268. [CrossRef] 36. Parrinello, M.; Rahman, A. Polymorphic transitions in single crystals: A new molecular dynamics method. J. Appl. Phys. 1981, 52, 7182–7190. [CrossRef]

© 2020 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 (http://creativecommons.org/licenses/by/4.0/).