Any opinions, findings, conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of UKDiss.com.
Computational insight into molecular mechanism of differentiated allosteric modulations at the mu opioid receptor by structurally similar opioid ligands
ABSTRACT: Activating or blocking the mu opioid receptor (MOR) through orthosteric ligands are current methods to treat opioid abuse and addiction. Unfortunately, most MOR orthosteric ligands suffer from significant side effects mainly due to their low selectivity to the kappa and delta opioid receptors. In contrast, some G protein-coupled receptor (GPCR) allosteric ligands have been reported to exhibit high subtype selectivity and can modulate effectively the potency and efficacy of orthosteric ligands. Recently, we identified structurally similar NAQ and its analog NCQ as novel MOR ligands. Interestingly, NAQ acted mainly as a MOR antagonist but NCQ as a MOR partial agonist. In the present work, molecular modeling methods were applied to explore their molecular mechanism on their different efficacy profiles on the MOR. When NAQ and NCQ bound with the inactive state MOR, the ‘address’ moieties of NAQ and NCQ interacted with the allosteric binding site but showed no allosteric modulation to the binding of the ‘message’ moieties of NAQ and NCQ at the orthosteric binding site, similarly to the silent allosteric modulator. In contrary, in the active state MOR, the ‘address’ moieties of NAQ and NCQ showed allosteric modulation function of a positive allosteric modulator at different levels. The “address” moieties of both ligands seemed to be able to positively modulate the efficacy of the ‘message’ moieties of the two ligands at the orthosteric binding site. Owing to the relatively larger ‘address’ moiety of NCQ, its positive allosteric modulation seemed to be much stronger than that of NAQ. These computational results combined with the experimental data helped us understand the allosteric modulation mechanism of NAQ and NCQ, and provided valuable information to develop our next generation of MOR selective ligands with high affinity and subtype selectivity to potentially treat opioid abuse and addiction.
KEYWORDS:NAQ, NCQ, mu opioid receptor, molecular modeling, allosteric binding site, orthosteric binding site
America is under an opioid crisis. According to the National Institute on Drug Abuse, every day more than 115 people in the United States died from opioid overdose in 2017.1Opioid abuse and addiction are the main cause of opioid overdose. Currently, there are three medications approved by the Food and Drug Administration (FDA) to treat opioid abuse and addiction, including opioid agonist methadone, partial agonist buprenorphine, and antagonist naltrexone.2 However, due to their side effects, such as abuse liability and high relapse rate concerns associated with methadone and buprenorphine maintenance, patient compliance issues associated with naltrexone treatment, the utilization rates of these pharmacotherapies are low.3Thus, new and effective opioid abuse and addiction treatments still need to be developed.
Mu opioid receptor (MOR) plays a central role in the treatment of opioid abuse and addiction. The abuse and addiction liability of clinically available opioids is mainly caused by their interactions with the MOR.4, 5Hence, activation of the MOR through MOR agonists or partial agonists or keeping the MOR at its inactive state via MOR antagonists may provide potential therapies for opioid abuse and addiction.6-8 Many MOR agonists, partial agonists, and antagonists also can be called as orthosteric ligands, which target the primary binding site of the MOR, termed as orthosteric binding site.9, 10Among three subtypes of the opioid receptors, the MOR, kappa opioid receptor (KOR), and delta opioid receptor (DOR),their orthosteric binding sites are highly conserved and the residues acting critical roles in the binding with the orthosteric ligands are almost identical.11Therefore, the development of subtype-specific MOR orthosteric ligands is challenging.On the other hand, different from the orthosteric binding site, the allosteric binding site of G-protein coupled receptors (GPCRs) have been reported to locate at the non-conserved regions and can accommodate allosteric ligands, also termed as allosteric modulators.12-14Hence, allosteric modulators binding with the allosteric binding sites may achieve high subtype selectivity. Moreover, allosteric modulators may have the potentialities to promote or reduce the binding affinity of orthosteric ligands, or modulate the functional response of GPCRs to the binding of orthosteric ligands via inducing long-range conformational changes of the receptors, including the opioid receptors.15-18
Depending on whether the functional response of the receptor is enhanced, inhibited, or had no effect on the affinity or the efficacy of orthosteric ligands, allosteric modulators can act as a positive allosteric modulator (PAM), a negative allosteric modulator (NAM), or a silent allosteric modulator (SAM), respectively.17, 19 On the other hand, these potentialities of allosteric modulators only act when orthosteric ligands bind to the receptor. Recently, several ligands targeting both the orthosteric and allosteric binding sites of GPCRs simultaneously have been identified with enhanced receptor affinity and subtype specificity, e.g. McN-A-343 for the M1-muscarinic acetylcholine receptor,20R-22 for the Dopamine D3-like receptor,21SPM-242 and SPM-354 for the sphingosine-1-phosphate receptor.22, 23 These ligands have been termed as bitopic ligands due to the fact that they carry the characters of both orthosteric and allosteric ligands with unique pharmacological profiles.24-26Thus, design and development of bitopic ligands selectively targeting the MOR may be a plausible route to treat opioid abuse and addiction.
Figure 1. The chemical structures of NAQ (a) and NCQ (b). The chemical structures with atom notation were derived from the complexes after molecular dynamics (MD) simulations.
As one example, our lab recently reported that 17-cyclopropylmethyl-3,14β-dihydroxy-4,5α-epoxy-6β-[(2’-indolyl)acetamido]morphinan (INTA) was a MOR agonist and 17-cyclopropylmethyl-3,14β-dihydroxy-4,5α-epoxy-6α-(indole-7-carboxamido)morphinan (NAN) was a MOR antagonist.27, 28Molecular modeling studies indicated that both two ligands may be termed as bitopic ligands. Their epoxymorphinan moieties bound to the orthosteric binding site of MOR and their indole rings interacted with different domains of the MOR allosteric binding site. Hence, the indole ring of INTA acted as a PAM to the epoxymorphinan moiety but the indole ring of NAN acted as a NAM.28Recently, based on the ‘message-address’ concept, our lab developed a C(6)-isoquinoline substituted naltrexone derivative NAQ (Figure 1a).29According to the docking study, we found that the ‘message’ moiety of NAQ bound with the same binding sites of the epoxymorphinan moieties of INTA and NAN while its ‘address’ moiety may interact with the allosteric binding site of MOR.30NAQ displayed a high binding affinity for the MOR at Ki = 0.55 nM with about 48-fold selectivity for the MOR over the KOR and about 241-fold selectivity over the DOR.29It acted as a low-efficacy MOR partial agonist compared with DAMGO in the 35S-GTP[γS]-binding assay with MOR expressing CHO cell lines.31-33In the comparative opioid withdrawal precipitation study, NAQ exhibited less withdrawal effect compared with the well-known opioid antagonists, naloxone and naltrexone.34These properties associated with NAQ may make it serve as a potential lead to develop more potent and selective MOR ligands to treat opioid abuse and addiction.
In this regard, a series of NAQ analogs were synthesized and pharmacologically characterized to understand their structure-activity relationship (SAR) on the MOR. Among them, NCQ (Figure 1b) was the most efficacious MOR ligand. The binding affinity of NCQ to MOR was 0.55 nM with about 40-fold selectivity for the MOR over the KOR and approximately 62-fold selectivity over the DOR.35Notably, Figure 1 showed that the ‘message’ moieties (epoxymorphinan moiety) of NAQ and NCQ were identical. For the ‘address’ moieties (isoquinoline ring) of NAQ and NCQ, there were two substitutions on the 1’- and 4’-position of the isoquinoline ring of NCQ. Based on the MD simulations on another NAQ analog NNQ,35we anticipated that the ‘message’ moiety of NCQ may bind with the same orthosteric binding site of NAQ in MOR. However, compared to NAQ, NCQ acted as a high-efficacy MOR partial agonist in its [35S]GTPγS binding assay (Table 1).35The SAR studies of NAQ analogs on the MOR indicated that the substitutions on the 1’- and/or 4’-position of the isoquinoline ring (the address moiety) seemed to be critical to the efficacies of the MOR ligands.36
Therefore, in the present work, in order to comprehend the molecular mechanism of how the substitutions of the ‘address’ moieties of NAQ and NCQ induced different efficacy profiles on the MOR, NAQ and NCQ were docked into the crystal structures of antagonist-bound and agonist-bound MORs followed by MD simulations. These computational results together with the pharmacological characterization may help us understand the allosteric characteristics of the MOR binding site and the putative mechanism of action for the ‘address’ moieties of both NAQ and NCQ interacting with the MOR. Eventually, such knowledge will provide critical clues to modify the ‘address’ moiety and further design and develop more potent and selective MOR modulators for opioid abuse and addiction treatment.
|Compound||Ki (nM)||Selectivity ratio||MOR [35S]GTPγS binding|
|MOR||KOR||DOR||MOR vs KOR||MOR vs DOR||EC50 (nM)||% Emax of DAMGO|
- RESULTS AND DISCUSSION
2.1. Docking studies of NAQ and NCQ in the inactive state and active state MORs. In order to gain insights into the binding modes of NAQ and NCQ in theinactive state and active state MORs, NAQ and NCQ were docked into the crystal structures of antagonist-bound (PDB ID: 4DKL)37and agonist-bound (PDB ID: 5C1M)38MORs using GOLD 5.4.39, 40 The orientation of the ‘address’ moieties of NAQ and NCQ in the allosteric binding site of the MOR was mainly decided by favorable interactions between the ligands and the protein, that is, the hydrophobic portion of the ligands with the hydrophobic transmembrane helices of the protein and the hydrophilic portion of the ligands with the polar extracellular loop region of the protein,29, 31 which can be reflected by different scoring systems. In our operation, the highest CHEM-PLP scores and the reasonable orientations of NAQ and NCQ in the inactive state and active state MORs were considered together to choose the optimal docking poses, and the results were displayed in Figures 2 (inactive state MOR) and 3 (active state MOR).
Similar to other MOR ligands that were designed based on the “message-address” concept,30, 41 the epoxymorphinan moieties (‘message’ moiety) of NAQ and NCQ occupied a similar ‘message’ domain of MOR, which also can be termed as the orthosteric binding site: the epoxymorphinan moiety formed hydrophobic interactions with M151, W293, and H297, the dihydrofuran oxygen of the epoxymorphinan moiety formed hydrogen bonding interaction with Y148, and the piperidine quaternary ammonium nitrogen atom formed ionic interactions with D147. Moreover, owing to the amide linker between the ‘message’ moiety and the ‘address’ moiety, the ‘address’ moieties of NAQ and NCQ can reach to the allosteric binding site of the MOR and may occupy different domains of the allosteric binding site in different binding poses. According to the docking results, there were mainly three different domains in the allosteric binding site of the MOR interacting with the ‘address’ moieties of NAQ and NCQ, termed as allosteric binding domain 1 (ABD1); ABD2, and ABD3. As shown in Figures 2 and 3, ABD1 formed between transmembrane helices (TMs) 6 and 7, ABD2 formed between TM 5 and ECL2, and ABD3 formed between TMs 2, 3, and ECL 1. The residues V3006.55 and W3187.35 at ABD1, L219ECL2 and L2325.38 at ABD2, Q1242.60, W133ECL1, and I1443.29 at ABD3 could directly interact with the isoquinoline rings of NAQ and NCQ in different binding poses. In the inactive state MOR, the ‘address’ moieties of NAQ and NCQ may occupy either ABD1 or ABD2 (Figure 2). In the active state MOR, the ‘address’ moiety of NAQ seemed to interact with ABD1 or ABD3, and the ‘address’ moiety of NCQ may interact with ABD2 or ABD3 (Figure 3), as summarized in Table 2.
Figure 2. The conformations of NAQ (a) and NCQ (b) in the inactive state MOR with the highest CHEM-PLP scores from docking studies. The protein showed as cartoon model in gray; NAQ and NCQ showed as stick model. Carbon atoms: NAQ in NAQ_MORinactive_ABD1 complex (cyan); NAQ in NAQ_MORinactive_ABD2 complex (orange); NCQ in NCQ_MORinactive_ABD1 complex (magentas); NCQ in NCQ_MORinactive_ABD2 complex (light-blue); key amino acid residues showed as stick and ball model in green. The surface models in yellow, light-orange, and cyan represented the ABD1, ABD2, and ABD3, respectively.
Figure 3. The conformations of NAQ (a) and NCQ (b) in the active state MOR with the highest CHEM-PLP scores from docking studies. The protein showed as cartoon model in light-green; NAQ and NCQ showed as stick model. Carbon atoms: NAQ in NAQ_MORinactive_ABD1 complex (cyan); NAQ in NAQ_MORinactive_ABD3 complex (orange); NCQ in NCQ_MORinactive_ABD2 complex (magentas); NCQ in NCQ_MORinactive_ABD3 complex (light-blue); key amino acid residues showed as stick and ball model in green. The surface models in yellow, light-orange, and cyan represented the ABD1, ABD2, and ABD3, respectively.
Table 2. The binding domains of the ‘address’ moieties of NAQ and NCQ in the allosteric binding site of inactive and active state MORs, and CHEM-PLP scores for the optimal binding poses in the inactive and active state MORs from docking studies.
|Ligand||Inactive MOR||CHEM-PLP||Complex||Active MOR||CHEM-PLP||Complex|
2.2. Molecular dynamics (MD) simulations of NAQ and NCQ in the inactive state and active state MORs. Previous studies had indicated that molecular dynamics (MD) simulation was a feasible method to evaluate the reliability of the molecular docking results.35, 41Thus, to determine which binding pose may be favored for NAQ and NCQ in the inactive state and active state MORs and understand their binding mechanisms in the MOR, MD simulations were performed on the eight optimal docking poses obtained from docking studies. In order to obtain stable systems, we conducted 10 ns MD simulations on the eight systems within a POPC lipid bilayer membrane system solvated by TIP3 water layers. As described in previous studies, the root-mean-square deviation (rmsd) values of a system smaller than 3.0 Å can be used as a criterion to evaluate the dynamics equilibrium of the system.42, 43Therefore, the rmsd values of all the protein backbone atoms based on the respective starting structures were calculated and the results were displayed in Figure 4. It seemed apparent that the rmsd values of all the eight systems showed minimum variance after 5 ns of MD simulations. The average rmsd values of NAQ_MORinactive_ABD1, NAQ_MORinactive_ABD2, NCQ_MORinactive_ABD1, NCQ_MORinactive_ABD2, NAQ_MORactive_ABD1, NAQ_MORactive_ABD3, NCQ_MORactive_ABD2, and NCQ_MORactive_ABD3 systems were 1.49, 1.46, 1.31, 1.12, 1.34, 1.28, 1.36, and 1.36 Å, respectively. Based on the above analysis, we concluded that all eight systems achieved stability after 5 ns MD simulations. Therefore, it was reasonable and reliable to conduct the energy and distance analyses based on the snapshots extracted from 7-10 ns MD simulations.
Figure 4. The root-mean-square deviation (rmsd) of the protein backbone atoms of the eight systems relative to the respective starting structures.
The average non-bonded interaction energies of the last 4 ns MD simulations between ligands (NAQ and NCQ) and their surrounding environment (including protein and water molecules) within different cutoff distances of the ligands in the eight systems were calculated and the results were summarized in Table 3. As shown in Table 3, the total interaction energies within different cutoff distances followed the trend: 10 Å < 8 Å < 5Å, which meant the cutoff distance of the ligands had a significant effect on the total interaction energies and comparing the total interaction energies between different binding poses had to be within the same cutoff distances. For the inactive state MOR, comparing the total interaction energies of NAQ_MORinactive_ABD1 and NCQ_MORinactive_ABD1 systems with that of NAQ_MORinactive_ABD2 and NCQ_MORinactive_ABD2 systems, respectively, we found that the total interaction energies for the binding of NAQ and NCQ with ABD1 and ABD2 of the allosteric binding site at the same cutoff distance followed the trend: ABD2 < ABD1. While in the active state MOR, the total interaction energies of the two different binding poses of NAQ and NCQ at the same cutoff distance showed no significant difference from each other.
Table 3. The interaction energies (kcal/mol) between ligands (NAQ and NCQ) and their surrounding environment (including protein and water molecules) within different cutoff distances of the ligands
|Complex||Cutoff distancea (Å)||Eleb||Vdwc||Totald|
|a Distance from the ligands NAQ and NCQ.
b Ele: electrostatic interaction.
c VDW: Van der Waals’ interaction.
d The data in Italic and bold represent the total interaction energies.
Additionally, as described above, residues V3006.55 and W3187.35 at ABD1, L219ECL2 and L2325.38 at ABD2, Q1242.60, W133ECL1, and I1443.29 at ABD3 could directly form interactions with the isoquinoline rings of NAQ and NCQ. Thus, the shortest distances between the atom at the isoquinoline ring of the two ligands and the atom at these key amino acid residues of the eight protein-ligand systems during the last 4 ns MD simulations were measured and the results were summarized in Table 4. These distances may provide important information to determine which domain of the allosteric binding site may favorably interact with the ‘address’ moieties of NAQ and NCQ (described in the following section).
Table 4. The shortest distances between the atom of the isoquinoline ring of both NAQ and NCQ and the atom of the key residues in ABD1, ABD2, and ABD3 of the six ligand-protein systems after MD simulationsa
|Complexes||Atom of ligand||Atom of residue||distance (Å)|
|a The distances in Italic and bold represent the possible domain of the allosteric binding site of MOR interacted with the isoquinoline ring of the NAQ and NCQ.|
2.3. The allosteric modulation of the ‘address’ moiety of NAQ and NCQ in the inactive state MOR. According to the non-bonded interaction energy analysis (Table 3) and the distance analysis between the atoms at the isoquinoline ring of the two ligands (NAQ and NCQ) and the atoms on the key amino acid residues in ABD1 and ABD2 of the four inactive protein-ligand systems (Table 4), the NAQ_MORinactive_ABD2 and NCQ_MORinactive_ABD2 systems were chosen as the representative binding poses of NAQ and NCQ in the inactive state MOR.
For the NAQ_MORinactive_ABD2 complex, the binding domain of the isoquinoline ring of NAQ shifted from the ABD2 toward the ABD1 after 10 ns MD simulations (Figures 2a, 5a, and Table 4). As a result, the binding mode of the NAQ_MORinactive_ABD2 complex was similar to that of the NAQ_MORinactive_ABD1 complex after 10 ns MD simulations (Figure 5a). That is, the ‘message’ moiety of NAQ interacted with the orthosteric binding site and the ‘address’ moiety bounded with the ABD1 of the allosteric binding site. In the NAQ_MORinactive_ABD2 complex, the cyclohexyl ring that the isoquinoline ring of NAQ attached to adopted a twisted boat conformation to accommodate the α-configuration of the isoquinoline ring to the ABD1 of the MOR (Figure 5a). Consequently, the aromatic residue W318 locating at ABD1 formed - stacking interaction with the ‘address’ moiety of the NAQ, which was consistent with our previous observation.30In the NCQ_MORinactive_ABD2 complex, the chloride and methoxyl substitutions on the isoquinoline ring yielded a relatively larger ‘address’ moiety than that of NAQ, which would make the ABD1 of the inactive state MOR not spacious enough to accommodate the ‘address’ moiety of NCQ. In addition, as shown in Figure 5b and Table 4, the ‘address’ moiety of NCQ could form additional hydrophobic interactions with residues L219and L232 that located at ABD2. Thus, the cyclohexyl ring of NCQ was in a chair conformation to make the ‘address’ moiety of NCQ interact favorably with the ABD2 of the inactive state MOR after 10 ns MD simulations.
Figure 5. Binding modes of NAQ_MORinactive_ABD1, NAQ_MORinactive_ABD2 (a) and NCQ_MORinactive_ABD2 (b) systems with key amino acid residues in the binding site after 10 ns MD simulations. The protein showed as cartoon model in gray; NAQ and NCQ showed as stick model. Carbon atoms: NAQ in NAQ_MORinactive_ABD1 complex (cyan); NAQ in NAQ_MORinactive_ABD2 complex (orange); NCQ in NCQ_MORinactive_ABD2 complex (light-blue); key amino acid residues showed as stick and ball model in green. The surface models in yellow, light-orange, and cyan represented the ABD1, ABD2, and ABD3, respectively.
As described above, M151, W293, and H297 located at the orthosteric binding site of the MOR, which could directly form hydrophobic interactions with the ‘message’ moieties of the NAQ and NCQ. Notably, the hydrophobic interaction between W293 and the ‘message’ moiety of the NAQ was mainly formed with the six-membered benzene ring of the indole ring of W293. The ‘address’ moiety of NAQ binding to ABD1 (ABD2) resulted in the ‘message’ moiety of NAQ moving deeply into the orthosteric binding site of the inactive state MOR (Figure 5a and Table 5), which may induce the side chain of W293 to exhibit a significant rotation (comparing Figure 2a with Figure 5a). In fact, the benzene ring of the indole ring of W293 seemed to move away from the ‘message’ moiety of NAQ while the pyrrole ring of the indole ring of W293 moved close to the ‘message’ moiety of NAQ and formed hydrophobic interaction with the ‘message’ moiety of NAQ, which may compensate the weaken hydrophobic interaction between the benzene ring of the indole ring of W293 and the ‘message’ moiety of NAQ (Table 5). Thus, the binding of the ‘address’ moiety of NAQ with ABD1 (ABD2) of the allosteric binding site didn’t affect the binding of the ‘message’ moiety of the NAQ with the orthosteric binding site. In the NCQ_MORinactive_ABD2 complex, the distance analyses between the ‘message’ moiety of NCQ and the key residues M151, W293, and H297 located at the orthosteric binding site showed that the ‘address’ moiety of NCQ binding to ABD2 had no significant effect on the binding of the ‘message’ moiety of the NCQ with the orthosteric binding site (Table 5).
Therefore, in the inactive state MOR, the ‘address’ moieties of NAQ and NCQ in the different allosteric binding domains of the MOR showed the same modulation to the ‘message’ moieties of NAQ and NCQ binding with the orthosteric binding site. That is, the ‘address’ moieties of NAQ and NCQ bound with the allosteric binding site of the inactive state MOR seemed to have no significant effects on the binding of the ‘message’ moieties of NAQ and NCQ with the orthosteric binding site, which was agreement with the function of SAM.17, 19
Table 5. The average distances in NAQ_MORinactive_ABD2 and NCQ_MORinactive_ABD2 complexes from docking studies and 10 ns MD simulationsa
|Complexes||Atom of ligand||Atom of residue||Docking distance (Å)||MD distance (Å)|
a The distances in Italic represented the distances showed significant changes after 10 ns MD simulations.
2.4. The allosteric modulation of the ‘address’ moiety of NAQ and NCQ in the active state MOR. Fromthe non-bonded interaction energy analyses of NAQ and NCQ in the active state MOR (Table 3) and the distance analyses between the atom on the ‘address’ moiety of the two ligands (NAQ and NCQ) and the atom on the key amino acid residues at ABD1, ABD2, and ABD3 of the active state MOR (Table 4), the ‘address’ moiety of NAQ seemed to interact with both ABD1 and ABD3 stably and both ABD2 and ABD3 can accommodate the ‘address’ moiety of NCQ. The previous study on the crystal structures of the active MOR and KOR (PDB ID: 5C1M and 6B73)38, 44 indicated that the MOR agonist BU72 and the KOR agonist MP1104 could interact with both the orthosteric and allosteric binding sites and further induce the MOR and KOR into their active conformations. Moreover, the ‘message’ moieties of NAQ and NCQ (Figure 1) were almost identical to those of BU72 and MP1104, which could interact with the orthosteric binding site of the receptor. Most importantly, analyzing the binding modes of BU72 with the active MOR and MP1104 with the active KOR, the binding domains for both BU72 and MP1104 in the allosteric binding sites of the MOR and KOR were coincident with our defined binding domain, ABD3. Therefore, in our study, the NAQ_MORactive_ABD3 and NCQ_MORactive_ABD3 systems were chosen as the favorable binding poses of NAQ and NCQ in the active state MOR.
Figure 6. Binding modes of NAQ_MORactive_ABD3 (a) and NCQ_MORactive_ABD3 (b) complexes with key amino acid residues in the binding site after 10 ns MD simulations. The protein showed as cartoon model in light-green; NAQ and NCQ showed as stick model. Carbon atoms: NAQ in NAQ_MORinactive_ABD3 complex (orange); NCQ in NCQ_MORinactive_ABD3 complex (light-blue); key amino acid residues showed as stick and ball model in green. The surface models in yellow, light-orange, and cyan represented the ABD1, ABD2, and ABD3, respectively. The red dash lines represented the distances between T120 (TM2) and D147 (TM3).
As described above, the ABD3 had three important residues Q124, W133, and I144, which can directly form interactions with the ‘address’ moiety of NAQ and NCQ. From Table 4, the hydrophobic interaction between the ‘address’ moiety of NAQ and W133 in NAQ_MORactive_ABD3 complex was stronger than that between the ‘address’ moiety of NCQ and W133in NCQ_MORactive_ABD3 complex, while the hydrophobic interactions between the ‘address’ moiety of NAQ and residues I144 and Q124 in NAQ_MORactive_ABD3 complex were weaker than those between the ‘address’ moiety of NCQ and residues I144 and Q124in NCQ_MORactive_ABD3 complex. These three different interactions of the ‘address’ moiety of NAQ and NCQ were due to the ‘address’ moiety of NCQ larger than that of NAQ. Owing to the steric hindrance caused by the chloride and methoxyl groups at the isoquinoline ring of NCQ, the TM3 could keep close enough to the TM2 to maintain the MOR in its active conformation (Figure 6b, 4.5 Å), which was almost similar to that in the crystal structure of agonist-bound MOR (4.7 Å). While in the NAQ_MORactive_ABD3 complex, owing to the smaller ‘address’ moiety of NAQ, the steric hindrance caused by the isoquinoline ring of NAQ was not strong enough to keep the conformations of the TM2 and TM3 stay at its active state (Figure 6a, 5.6 Å). For comparison purpose, the average distance between TM2 and TM3 in the crystal structure of antagonist-bound MOR was 6.8 Å.
Taken together, the ‘address’ moieties of NAQ and NCQ interacted with the same domain (ABD3) of the allosteric binding site of the active state MOR. Owing to the chloride and methoxyl groups at the isoquinoline ring of NCQ, the ‘address’ moiety of NCQ can positively modulate the functional response of the orthosteric binding site of MOR and keep the MOR at its active conformation, which was similar to the function of PAM. For NAQ, as the ‘address’ moiety of NAQ was relatively smaller than that of NCQ, the positive modulation of the ‘address’ moiety of NAQ to the binding of the ‘message’ moiety of NAQ with the orthosteric binding site was much weaker than that of the ‘address’ moiety of NCQ. Therefore, NCQ showed higher efficacy than NAQ when bound with the active state MOR.
The mu opioid receptor (MOR) is a critical target for the treatment of drug abuse and addiction. NAQ and its analog NCQ showed high binding affinity to the MOR and exhibited high selectivity for the MOR over the KOR and DOR. Interestingly, NAQ acted as a MOR antagonist or low efficacy partial agonist but its analog NCQ acted as a MOR potent partial agonist. In the present study, the molecular mechanism of different substitutions as the ‘address’ moieties led to the different allosteric modulations between NAQ and its analog NCQ to the MOR was investigated by using molecular docking and molecular dynamics (MD) simulations.
Docking studies revealed that there were three different binding domains in the allosteric binding site of MOR: ABD1, ABD2, and ABD3. From the docking studies, we obtained eight optimal binding poses for NAQ and NCQ in the inactive and active state MOR. Based on the results of the MD simulations and the following energy and distance analyses, NAQ_MORinactive_ABD2 and NCQ_MORinactive_ABD2, NAQ_MORactive_ABD3, and NCQ_MORactive_ABD3 systems were chosen as the representative binding poses. In the inactive state MOR, similar to SAM, the ‘address’ moieties of NAQ and NCQ bound with ABD1 and ABD2, respectively, which may occupy the allosteric binding site of the inactive state MOR but showed no significant effect on the binding of the ‘message’ moieties of NAQ and NCQ with the orthosteric binding site. In the active state MOR, the ‘address’ moieties of NAQ and NCQ showed positive allosteric modulation to the ‘message’ moieties of NAQ and NCQ binding with the orthosteric binding site. As the chloride and methoxyl substituents on the isoquinoline ring of NCQ caused the ‘address’ moiety of NCQ larger than that of NAQ, the positive allosteric modulation of the ‘address’ moiety of NCQ was more significant than that of ‘address’ moiety of NAQ. Therefore, NCQ exhibit higher efficacy than NAQ.
In conclusion, the application of the molecular modeling methods, combined with the energy and distance analyses, helped us exploring the molecular mechanisms of the allosteric modulation of NAQ and NCQ. The finding will provide useful information to the design and development of novel and highly selective MOR ligands that can link orthosteric and allosteric binding sites together to treat substance abuse and addiction.
- Materials and methods
4.1. Preparation of structures. The X-ray crystal structures for antagonist-bound (PDB ID: 4DKL)37and agonist-bound (PDB ID: 5C1M)38MOR were obtained from the Protein Data Bank at http://www.rcsb.org and adopted in the present work. Hydrogen atoms were added to each structure (without ligand). After that, each structure was optimized with 10,000 energy minimization iterations while all heavy atoms were held a fixed aggregate Gasteiger-Hückel charge under the Tripos Forcefield (TAFF) in Sybyl-X 2.1. NAQ and NCQ were sketched in Sybyl-X 2.1, then assigned Gasteiger-Hückel charges and energy minimized (10,000 iterations) to a gradient of 0.05.45
4.2. Docking studies. GOLD 5.4 was employed to conduct the docking studies with standard default settings.39, 40 As widely accepted for other epoxymorphinan skeletons, D147locating in the orthosteric binding site of MOR and the protonated nitrogen atom at the 17-amino group of NAQ and NCQ may form an ionic interaction. Thus, the atoms within 10 Å of the γ-carbon atom of D147in MOR were applied to define the binding site.29During the docking process, the distance between the protonated nitrogen atom in the 17-amino group of the ligands and the oxygen atom at the side chain of D147 was limited to 4.0 Å to ensure the formation of an ionic interaction between them. The distance between the ligands’ dihydrofuran oxygen atom and the phenolic oxygen atom of Y148was limited to 3.5 Å to ensure the formation of a hydrogen bonding interaction. Default scoring function of GOLD 5.4, CHEM-PLP score,46together with the binding orientations of NAQ and NCQ within the binding site of MOR were used as criteria to choose the optimal docking pose.
4.3. Construction of lipid membrane system. To our knowledge, the GPCRs comprise the largest group of membrane receptors in eukaryotes.In order to simulate the MOR in a membrane-aqueous system, the 1-palmitoyl-2-oleoylphosphatidylcholine (POPC) was selected to construct a 100 Å 100 Å POPC lipid membrane with the MEMBRANE plugin in the Visual Molecular Dynamics (VMD) program.47The orientation of the protein in the membrane (OPM) server was employed to orient the receptor-ligand complex within the POPC lipid membrane.48Next, each system was TIP3 solvated with a 50 Å water layer on both sides of the membrane at the vertical axis by using the SOLVATE plugin in VMD.After that, water molecules or POPC lipid molecules at a distance of 0.75 Å or less from the receptor-ligand complex were removed. Finally, sodium and chloride ions were added to the solvated systems to make the ion concentration approximately 0.1 nM by the AUTOIONIZE plugin in VMD. The ions were distributed above and below the POPC lipid membrane. A receptor-ligand complex within the POPC lipid membrane solvated by TIP3 water layers was displayed in Figure 7.
Figure 7. Receptor-ligand complex within a POPC lipid bilayer membrane system solvated by TIP3 water bilayers. The protein showed as cartoon model in cyan; ligand showed as sphere model in magentas; POPC lipid membrane showed as stick model in green; water molecules showed as line model in red.
4.4. Molecular dynamics simulation. Molecular dynamics (MD) simulations were performed using the NAMD V2.8 program.49MD simulations were conducted in four steps in the present work. First, the lipid tails were energy minimized (1,000 iterations) in the NPT ensemble for a period of 0.5 ns to make the fluid-like POPC lipid bilayer to reach equilibrium. Second, an NPT equilibration of the system was conducted for a period of 1.0 ns by restraining the protein and ligand (5 kcal/(mol-Å)). Third, the whole system was energy minimized without any restraint in the NVT ensemble for a period of 1.0 ns. Final, the production run was conducted with the same parameters as the third step for a period of 10.0 ns.
Simulations were conducted using the CHARMM force field with CHARMM36 force field parameters for proteins50, 51 and lipids.52 The parameters of NAQ and NCQ were obtained from CHARMM General Force Field (CGenFF) program.53-55In the process of MD simulations, periodic boundary conditions were employed. The long-range electrostatic interactions were calculated using Particle Mesh Ewald (PME) method.56The non-bonded van der Waals interactions were computed with a smooth cut off between 10 and 12 Å. All simulations were carried out at constant pressure (p = 105 Pa) and temperature (T = 310 K) with an integration time step of 2 fs. The constant temperature was maintained via Langevin dynamics with a damping coefficient of 5 ps-1. The hybrid Nose-Hoover Langevin piston method, with a damping time of 200 fs, was used to keep the pressure constant. Based on MD simulations, energy landscape analysis was conducted by using the ENERGY plugin in NAMD with a dielectric constant of 6.5.57
1. Opioid Overdose Crisis. https://www.drugabuse.gov/drugs-abuse/opioids/opioid-overdose-crisis
2. Alderks, C.-E. (2017) Trends in the use of methadone, buprenorphine, and extended-release naltrexone at substance abuse treatment facilities: 2003-2015 (update). In The CBHSQ report: August 22, 2017 Center for Behavioral Health Statistics and Quality, Substance Abuse and Mental Health Services Administration, Rockville, MD.
3. Gray, A. (2007) Systematic review of the safety of buprenorphine methadone and naltrexone, Department of Therapeutics and Medicines Center for the AIDS Programme of Research in South Africa Congella, South Africa.
4. Matthes, H.-W., Maldonado, R., Simonin, F., Valverde, O., Slowe, S., Kitchen, I., Befort, K., Dierich, A., Le Meur, M., and Dollé, P. (1996) Loss of morphine-induced analgesia, reward effect and withdrawal symptoms in mice lacking the µ-opioid-receptor gene. Nature 383, 819.
5. Law, P.-Y., Wong, Y.-H., and Loh, H.-H. (2000) Molecular mechanisms and regulation of opioid receptor signaling. Annu. Rev. Pharmacol. Toxicol. 40, 389-430.
6. Biggio, G., Casu, M., Corda, M. G., Di Bello, C., and Gessa, G. (1978) Stimulation of dopamine synthesis in caudate nucleus by intrastriatal enkephalins and antagonism by naloxone. Science 200, 552-554.
7. Wood, P. L., Stotland, M., Richard, J., and Rackham, A. (1980) Actions of mu, kappa, sigma, delta and agonist/antagonist opiates on striatal dopaminergic function. J. Pharmacol. Exp. Ther. 215, 697-703.
8. Spanagel, R., Herz, A., and Shippenberg, T.-S. (1992) Opposing tonically active endogenous opioid systems modulate the mesolimbic dopaminergic pathway. Proceedings of the National Academy of Sciences 89, 2046-2050.
9. Haddou, T.-B., Béni, S., Hosztafi, S., Malfacini, D., Calo, G., Schmidhammer, H., and Spetea, M. (2014) Pharmacological investigations of N-substituent variation in morphine and oxymorphone: Opioid receptor binding, signaling and antinociceptive activity. PLoS One 9, e99231.
10. Huang, P., Kehner, G.-B., Cowan, A., and Liu-Chen, L.-Y. (2001) Comparison of Pharmacological Activities of Buprenorphine and Norbuprenorphine: Norbuprenorphine Is a Potent Opioid Agonist. J. Pharmacol. Exp. Ther. 297, 688-695.
11. Katritch, V., Cherezov, V., and Stevens, R. C. (2013) Structure-function of the G protein-coupled receptor superfamily. Annu. Rev. Pharmacol. Toxicol. 53, 531-556.
12. Melancon, B.-J., Hopkins, C.-R., Wood, M.-R., Emmitte, K.-A., Niswender, C.-M., Christopoulos, A., Conn, P.-J., and Lindsley, C.-W. (2012) Allosteric modulation of seven transmembrane spanning receptors: theory, practice, and opportunities for central nervous system drug discovery. J. Med. Chem. 55, 1445-1464.
13. Raddatz, R., Schaffhauser, H., and Marino, M.-J. (2007) Allosteric approaches to the targeting of G-protein-coupled receptors for novel drug discovery: a critical assessment. Biochem. Pharmacol. 74, 383-391.
14. Yanamala, N., and Klein-Seetharaman, J. (2010) Allosteric Modulation of G Protein-Coupled Receptors by Cytoplasmic, Transmembrane and Extracellular Ligands. Pharmaceuticals (Basel) 3, 3324-3342.
15. Flor, P.-J., and Acher, F.-C. (2012) Orthosteric versus allosteric GPCR activation: the great challenge of group-III mGluRs. Biochem. Pharmacol. 84, 414-424.
16. Dalton, J.-A., Gomez-Santacana, X., Llebaria, A., and Giraldo, J. (2014) Computational analysis of negative and positive allosteric modulator binding and function in metabotropic glutamate receptor 5 (in)activation. J. Chem. Inf. Model. 54, 1476-1487.
17. Burford, N.-T., Watson, J., Bertekap, R., and Alt, A. (2011) Strategies for the identification of allosteric modulators of G-protein-coupled receptors. Biochem. Pharmacol. 81, 691-702.
18. Conn, P.-J., Christopoulos, A., and Lindsley, C.-W. (2009) Allosteric modulators of GPCRs: a novel approach for the treatment of CNS disorders. Nat. Rev. Drug Discov. 8, 41-54.
19. Burford, N.-T., Clark, M.-J., Wehrman, T.-S., Gerritz, S.-W., Banks, M., O’Connell, J., Traynor, J.-R., and Alt, A. (2013) Discovery of positive allosteric modulators and silent allosteric modulators of the mu-opioid receptor. Proc. Natl. Acad. Sci. U. S. A. 110, 10830-10835.
20. Chen, X., Klockner, J., Holze, J., Zimmermann, C., Seemann, W.-K., Schrage, R., Bock, A., Mohr, K., Trankle, C., Holzgrabe, U., and Decker, M. (2015) Rational design of partial agonists for the muscarinic m1 acetylcholine receptor. J. Med. Chem. 58, 560-576.
21. Newman, A.-H., Grundt, P., Cyriac, G., Deschamps, J.-R., Taylor, M., Kumar, R., Ho, D., and Luedtke, R.-R. (2009) N-(4-(4-(2,3-dichloro- or 2-methoxyphenyl)piperazin-1-yl)butyl)heterobiarylcarboxamides with functionalized linking chains as high affinity and enantioselective D3 receptor antagonists. J. Med. Chem. 52, 2559-2570.
22. Sanna, M.-G., Vincent, K.-P., Repetto, E., Nguyen, N., Brown, S.-J., Abgaryan, L., Riley, S.-W., Leaf, N.-B., Cahalan, S.-M., Kiosses, W.-B., Kohno, Y., Brown, J.-H., McCulloch, A.-D., Rosen, H., and Gonzalez-Cabrera, P.-J. (2016) Bitopic Sphingosine 1-Phosphate Receptor 3 (S1P3) Antagonist Rescue from Complete Heart Block: Pharmacological and Genetic Evidence for Direct S1P3 Regulation of Mouse Cardiac Conduction. Mol. Pharmacol. 89, 176-186.
23. Jo, E., Bhhatarai, B., Repetto, E., Guerrero, M., Riley, S., Brown, S.-J., Kohno, Y., Roberts, E., Schurer, S.-C., and Rosen, H. (2012) Novel selective allosteric and bitopic ligands for the S1P(3) receptor. ACS Chem. Biol. 7, 1975-1983.
24. Kamal, M., and Jockers, R. (2009) Bitopic ligands: all-in-one orthosteric and allosteric. F1000 Biol. Rep. 1, 77.
25. Fronik, P., Gaiser, B.-I., and Sejer Pedersen, D. (2017) Bitopic Ligands and Metastable Binding Sites: Opportunities for G Protein-Coupled Receptor (GPCR) Medicinal Chemistry. 60, 4126-4134.
26. Lane, J.-R., Sexton, P.-M., and Christopoulos, A. (2013) Bridging the gap: bitopic ligands of G-protein-coupled receptors. Trends Pharmacol. Sci. 34, 59-66.
27. Le Naour, M., Lunzer, M.-M., Powers, M.-D., Kalyuzhny, A.-E., Benneyworth, M.-A., Thomas, M.-J., and Portoghese, P.-S. (2014) Putative kappa opioid heteromers as targets for developing analgesics free of adverse effects. J. Med. Chem. 57, 6383-6392.
28. Obeng, S., Wang, H., Jali, A., Stevens, D.-L., Akbarali, H.-I., Dewey, W.-L., Selley, D.-E., and Zhang, Y. (2018) Structure-Activity Relationship Studies of 6alpha- and 6beta-Indolylacetamidonaltrexamine Derivatives as Bitopic Mu Opioid Receptor Modulators and Elaboration of the “Message-Address Concept” To Comprehend Their Functional Conversion. ACS Chem. Neurosci published online Sep 7, 2018. DOI: 10.1021/acschemneuro.8b00349.
29. Li, G., Aschenbach, L.-C., Chen, J., Cassidy, M.-P., Stevens, D.-L., Gabra, B.-H., Selley, D.-E., Dewey, W.-L., Westkaemper, R.-B., and Zhang, Y. (2009) Design, synthesis, and biological evaluation of 6alpha- and 6beta-N-heterocyclic substituted naltrexamine derivatives as mu opioid receptor selective antagonists. J. Med. Chem. 52, 1416-1427.
30. Zaidi, S.-A., Arnatt, C.-K., He, H., Selley, D.-E., Mosier, P.-D., Kellogg, G.-E., and Zhang, Y. (2013) Binding mode characterization of 6alpha- and 6beta-N-heterocyclic substituted naltrexamine derivatives via docking in opioid receptor crystal structures and site-directed mutagenesis studies: application of the ‘message-address’ concept in development of mu opioid receptor selective antagonists. Bioorg. Med. Chem. 21, 6405-6413.
31. Mitra, P., Venitz, J., Yuan, Y., Zhang, Y., and Gerk, P.-M. (2011) Preclinical disposition (in vitro) of novel mu-opioid receptor selective antagonists. Drug Metab. Dispos. 39, 1589-1596.
32. Altarifi, A.-A., Yuan, Y., Zhang, Y., Selley, D.-E., and Negus, S.-S. (2015) Effects of the novel, selective and low-efficacy mu opioid receptor ligand NAQ on intracranial self-stimulation in rats. Psychopharmacology (Berl.) 232, 815-824.
33. Obeng, S., Yuan, Y., Jali, A., Selley, D.-E., and Zhang, Y. (2018) In vitro and in vivo functional profile characterization of 17-cyclopropylmethyl-3,14beta-dihydroxy-4,5alpha-epoxy-6alpha-(isoquinoline-3-car boxamido)morphinan (NAQ) as a low efficacy mu opioid receptor modulator. Eur. J. Pharmacol. 827, 32-40.
34. Yuan, Y., Li, G., He, H., Stevens, D.-L., Kozak, P., Scoggins, K.-L., Mitra, P., Gerk, P.-M., Selley, D.-E., Dewey, W.-L., and Zhang, Y. (2011) Characterization of 6alpha- and 6beta-N-heterocyclic substituted naltrexamine derivatives as novel leads to development of mu opioid receptor selective antagonists. ACS Chem. Neurosci. 2, 346-351.
35. Yuan, Y., Zaidi, S.-A., Stevens, D.-L., Scoggins, K.-L., Mosier, P.-D., Kellogg, G.-E., Dewey, W.-L., Selley, D.-E., and Zhang, Y. (2015) Design, syntheses, and pharmacological characterization of 17-cyclopropylmethyl-3,14beta-dihydroxy-4,5alpha-epoxy-6alpha-(isoquinoline-3′-ca rboxamido)morphinan analogues as opioid receptor ligands. Bioorg. Med. Chem. 23, 1701-1715.
36. Yuan, Y., Elbegdorj, O., Beletskaya, I.-O., Selley, D.-E., and Zhang,-Y. (2013) Structure activity relationship studies of 17-cyclopropylmethyl-3,14beta-dihydroxy-4,5alpha-epoxy-6alpha-(isoquinoline-3′-ca rboxamido)morphinan (NAQ) analogues as potent opioid receptor ligands: preliminary results on the role of electronic characteristics for affinity and function. Bioorg. Med. Chem. Lett. 23, 5045-5048.
37. 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., and Granier, S. (2012) Crystal structure of the micro-opioid receptor bound to a morphinan antagonist. Nature 485, 321-326.
38. 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., Granier, S., Gmeiner, P., Husbands, S.-M., Traynor, J.-R., Weis, W.-I., Steyaert, J., Dror, R.-O., and Kobilka, B.-K. (2015) Structural insights into micro-opioid receptor activation. Nature 524, 315-321.
39. Jones, G., Willett, P., and Glen, R.-C. (1995) Molecular recognition of receptor sites using a genetic algorithm with a description of desolvation. J. Mol. Biol. 245, 43-53.
40. Jones, G., Willett, P., Glen, R.-C., Leach, A.-R., and Taylor, R. (1997) Development and validation of a genetic algorithm for flexible docking. J. Mol. Biol. 267, 727-748.
41. Wang, H., Zaidi, S.-A., and Zhang, Y. (2017) Binding mode analyses of NAP derivatives as mu opioid receptor selective ligands through docking studies and molecular dynamics simulation. Bioorg. Med. Chem. 25, 2463-2471.
42. Gohlke, H., Kiel, C., and Case, D.-A. (2003) Insights into protein-protein binding by binding free energy calculation and free energy decomposition for the Ras-Raf and Ras-RalGDS complexes. J. Mol. Biol. 330, 891-913.
43. Hou, T., McLaughlin, W., Lu, B., Chen, K., and Wang, W. (2006) Prediction of binding affinities between the human amphiphysin-1 SH3 domain and its peptide ligands using homology modeling, molecular dynamics and molecular field analysis. J. Proteome Res. 5, 32-43.
44. Che, T., Majumdar, S., Zaidi, S.-A., Ondachi, P., McCorvy, J.-D., Wang, S., Mosier, P.-D., Uprety, R., Vardy, E., Krumm, B.-E., Han, G.-W., Lee, M.-Y., Pardon, E., Steyaert, J., Huang, X.-P., Strachan, R.-T., Tribo, A.-R., Pasternak, G.-W., Carroll, F.-I., Stevens, R.-C., Cherezov, V., Katritch, V., Wacker, D., and Roth, B.-L. (2018) Structure of the Nanobody-Stabilized Active State of the Kappa Opioid Receptor. Cell 172, 55-67.
45. Clark, M., Cramer, R.-D., and Van Opdenbosch, N. (1989) Validation of the general purpose tripos 5.2 force field. J. Comput. Chem. 10, 982-1012.
46. Korb, O., Stutzle, T., and Exner, T.-E. (2009) Empirical scoring functions for advanced protein-ligand docking with PLANTS. J. Chem. Inf. Model. 49, 84-96.
47. Eisen, D., Planatscher, H., Hardie, D.-B., Kraushaar, U., Pynn, C.-J., Stoll, D., Borchers, C., Joos, T.-O., and Poetz, O. (2013) G protein-coupled receptor quantification using peptide group-specific enrichment combined with internal peptide standard reporter calibration. J. Proteomics 90, 85-95.
48. Lomize, M.-A., Lomize, A.-L., Pogozheva, I.-D., and Mosberg, H.-I. (2006) OPM: orientations of proteins in membranes database. Bioinformatics 22, 623-625.
49. Phillips, J.-C., Braun, R., Wang, W., Gumbart, J., Tajkhorshid, E., Villa, E., Chipot, C., Skeel, R.-D., Kale, L., and Schulten, K. (2005) Scalable molecular dynamics with NAMD. J. Comput. Chem. 26, 1781-1802.
50. MacKerell, A.-D., Jr., Feig, M., and Brooks, C.-L., 3rd. (2004) Improved treatment of the protein backbone in empirical force fields. J. Am. Chem. Soc. 126, 698-699.
51. MacKerell, A.-D., Bashford, D., Bellott, M., Dunbrack, R.-L., Evanseck, J.-D., Field, M.-J., Fischer, S., Gao, J., Guo, H., Ha, S., Joseph-McCarthy, D., Kuchnir, L., Kuczera, K., Lau, F.-T., Mattos, C., Michnick, S., Ngo, T., Nguyen, D.-T., Prodhom, B., Reiher, W.-E., Roux, B., Schlenkrich, M., Smith, J.-C., Stote, R., Straub, J., Watanabe, M., Wiorkiewicz-Kuczera, J., Yin, D., and Karplus, M. (1998) All-atom empirical potential for molecular modeling and dynamics studies of proteins. J. Phys. Chem. B 102, 3586-3616.
52. Klauda, J.-B., Venable, R.-M., Freites, J.-A., O’Connor, J.-W., Tobias, D.-J., Mondragon-Ramirez, C., Vorobyov, I., MacKerell, A.-D., Jr., and Pastor, R.-W. (2010) Update of the CHARMM all-atom additive force field for lipids: validation on six lipid types. J. Phys. Chem. B 114, 7830-7843.
53. Vanommeslaeghe, K., and MacKerell, A.-D., Jr. (2012) Automation of the CHARMM General Force Field (CGenFF) I: bond perception and atom typing. J. Chem. Inf. Model. 52, 3144-3154.
54. Vanommeslaeghe, K., Raman, E.-P., and MacKerell, A.-D., Jr. (2012) Automation of the CHARMM General Force Field (CGenFF) II: assignment of bonded parameters and partial atomic charges. J. Chem. Inf. Model. 52, 3155-3168.
55. Vanommeslaeghe, K., Hatcher, E., Acharya, C., Kundu, S., Zhong, S., Shim, J., Darian, E., Guvench, O., Lopes, P., Vorobyov, I., and Mackerell, A.-D., Jr. (2010) CHARMM general force field: A force field for drug-like molecules compatible with the CHARMM all-atom additive biological force fields. J. Comput. Chem. 31, 671-690.
56. Darden, T., York, D., and Pedersen, L. (1993) Particle mesh Ewald: An N⋅log(N) method for Ewald sums in large systems. The Journal of Chemical Physics 98, 10089-10092.
57. Kukic, P., Farrell, D., McIntosh, L.-P., Garcia-Moreno, E.-B., Jensen, K.-S., Toleikis, Z., Teilum, K., and Nielsen, J.-E. (2013) Protein dielectric constants determined from NMR chemical shift perturbations. J. Am. Chem. Soc. 135, 16968-16976.