ORIGINAL RESEARCH published: 06 March 2020 doi: 10.3389/fmolb.2020.00038 Exploring the Activation Mechanism of the mGlu5 Transmembrane Domain Isaias Lans 1,2 , Óscar Díaz 1,3,4 , James A. R. Dalton 1,3,4 and Jesús Giraldo 1,3,4* 1 Laboratory of Molecular Neuropharmacology and Bioinformatics, Unitat de Bioestadística and Institut de Neurociències, Universitat Autònoma de Barcelona, Bellaterra, Spain, 2 Biophysics of Tropical Diseases, Max Planck Tandem Group, University of Antioquia, Medellín, Colombia, 3 Unitat de Neurociència Traslacional, Parc Taulí Hospital Universitari, Institut d’Investigació i Innovació Parc Taulí (I3PT), Institut de Neurociències, Universitat Autònoma de Barcelona, Barcelona, Spain, 4 Instituto de Salud Carlos III, Centro de Investigación Biomédica en Red de Salud Mental, CIBERSAM, Madrid, Spain As a class C GPCR and regulator of synaptic activity, mGlu5 is an attractive drug target, potentially offering treatment for several neurologic and psychiatric disorders. As little is known about the activation mechanism of mGlu5 at a structural level, potential of mean force calculations linked to molecular dynamics simulations were performed on the mGlu5 transmembrane domain crystal structure to explore various internal mechanisms responsible for its activation. Our results suggest that the hydrophilic interactions between intracellular loop 1 and the intracellular side of TM6 have to be disrupted to Edited by: reach a theoretically active-like conformation. In addition, interactions between residues Huiyong Sun, that are key for mGlu5 activation (Tyr6593.44 and Ile7515.51 ) and mGlu5 inactivation China Pharmaceutical University, China (Tyr6593.44 and Ser8097.39 ) have been identified. Inasmuch as mGlu5 receptor signaling Reviewed by: is poorly understood, potentially showing a complex network of micro-switches and Tong Zhu, subtle structure-activity relationships, the present study represents a step forward in the East China Normal University, China Jagdish Suresh Patel, understanding of mGlu5 transmembrane domain activation. University of Idaho, United States Keywords: G protein-coupled receptors, mGlu5 receptor, molecular dynamics computer simulation, potential of *Correspondence: mean force calculations, free energies, mGlu, class C GPCR Jesús Giraldo Jesus.Giraldo@uab.es INTRODUCTION Specialty section: This article was submitted to Metabotropic glutamate (mGlu) receptors belong to Class C G protein-coupled receptors (GPCRs). Molecular Recognition, mGlu receptors are composed of 8 subtypes assembled into 3 (I, II, and III) groups (Pin et al., 2003). a section of the journal The focus of the present study, mGlu receptor subtype 5 (mGlu5), belongs to Group I (Pin and Frontiers in Molecular Biosciences Bettler, 2016). mGlu5 is involved in several neurologic and psychiatric disorders (Foster and Conn, Received: 20 November 2019 2017). Inhibition of mGlu5 might potentially alleviate depression, Parkinson’s disease, Fragile X Accepted: 18 February 2020 syndrome and chronic pain symptoms whereas positive modulation of mGlu5 may provide a new Published: 06 March 2020 treatment for schizophrenia (Nicoletti et al., 2015). Citation: From a structural point of view, mGlu receptors are characterized by a dimeric arrangement of Lans I, Díaz Ó, Dalton JAR and three protein domains: the extracellular Venus flytrap (VFT) domain, the transmembrane (TM) Giraldo J (2020) Exploring the Activation Mechanism of the domain, and the cysteine rich domain (CRD). The VFT domain includes the orthosteric site where mGlu5 Transmembrane Domain. glutamate and synthetic agonists and antagonists bind. The TM domain includes the seven helices Front. Mol. Biosci. 7:38. typical of all GPCRs and the intracellular regions responsible for G protein recognition. The CRD doi: 10.3389/fmolb.2020.00038 connects the VFT and TM domains (Pin and Bettler, 2016) mGlu receptors are obligate dimers, Frontiers in Molecular Biosciences | www.frontiersin.org 1 March 2020 | Volume 7 | Article 38 Lans et al. mGlu5 TM Activation Mechanism which adds both versatility and complexity to signal transduction structure: while a slight increase or decrease in co-operativity was (Rovira et al., 2008; Zhou and Giraldo, 2018; Pin et al., 2019). observed for some PAM and NAM scaffolds, respectively, a NAM In contrast to Class A GPCRs, the TM domains of mGlu to PAM switch was detected for two different ligand scaffolds receptors bear allosteric binding sites exclusively (Bennett et al., (Gregory et al., 2014). Also, it was found that Thr7816.46 Ala 2014). Ligand binding to the TM domain allows for positive, and Ser8097.39 Ala mutations switched the pharmacology of some negative and silent allosteric modulation of glutamate activity alkyne type PAMs (Gregory et al., 2013). Thus, the receptor (PAM, NAM, and SAM, respectively) (Kenakin, 2005; Conn et al., machine performs differently depending on the molecular 2009). NAM can be either full or partial depending on whether program the allosteric modulator selects. full occupancy at the allosteric site provides total or partial target The recent determination of the crystal structures of the inhibition, respectively. Allosteric modulation may affect both TM domain of the mGlu5 receptor, in an inactive state and the affinity and the efficacy of agonists. This can be done either bound to NAMs (Dore et al., 2014; Christopher et al., 2015) in a convergent or a divergent way, as it has been recently has definitively established the allosteric modulator binding shown (Kenakin and Strachan, 2018). As an example of the latter region. This structural knowledge has confirmed the role of behavior, PAM-antagonists were defined as allosteric modulators these and other residues previously identified by mutagenesis that increase the affinity but decrease the efficacy of agonists studies (Pagano et al., 2000; Malherbe et al., 2003, 2006; Gregory (Kenakin and Strachan, 2018). Moreover, mGlu PAMs may exert et al., 2013). More recently, a study was presented in which the a functional effect by themselves because their binding to the inactive and active conformations of full-length mGlu5 dimer TM domain may directly activate the receptors, in particular were elucidated by a combination of X-ray crystallography, cryo- mGlu5 receptor (Noetzel et al., 2012). This property divides electron microscopy and signaling studies (Koehl et al., 2019). mGlu PAMs into two groups: pure and ago-PAMs. The former As part of the mechanism of receptor activation, this work group comprises those compounds that lack intrinsic efficacy determined the intersubunit TM conformational change leading and are not able to activate the receptor in the absence of to a TM6-TM6 interface, as already proposed in a previous study glutamate or any other orthosteric agonist. Their mechanism of (Hlavackova et al., 2012). In addition, the relevance of the ECL2 action is a pure allosteric enhancement of agonist activity. The in the propagation of structural changes from the VFT to the TM latter group comprises those compounds that have both agonist through the CRD domain was pointed out (Koehl et al., 2019). and allosteric function (Foster and Conn, 2017). This different However, the intrasubunit rearrangement which, accordingly to mechanism of action of mGlu PAMs may have therapeutic REFs (Hlavackova et al., 2012; Grushevskyi et al., 2019), should consequences. It has been shown that, in contrast to pure mGlu5 sequentially occur after the intersubunit conformational change PAMs, mGlu5 ago-PAMs may present severe side effects such as was not identified, probably because of the absence of the G induction of seizures and behavioral convulsions (Rook et al., protein in the agonist-receptor complex. Nevertheless, though 2013). This indicates that mGlu5 ligand recognition is complex the structural characteristics of each of the protomers in the active and interpreted by at least two (agonist- and allosteric-) chemical TM domain remain unsolved, some features, which potentially languages that share many grammatical features. could lead to TM activation, were found. In particular, an upward Significant efforts have been made to decipher the mGlu5 movement of TM3, a slight outward movement of TM5 and a receptor structure-activity translation machinery (Kaae et al., destabilization of the ionic lock (Koehl et al., 2019). 2012; Dalton et al., 2014, 2017; Gregory et al., 2014; Gomez- Because of the complexity of mGlu molecular functioning, Santacana et al., 2017; Cong et al., 2019; Jojart et al., 2019; a complete solution to the problem can only be obtained by Llinas Del Torrent et al., 2019). In general, a delicate molecular- collecting different pieces of knowledge from complementary gear receptor system whose mechanism of action is not yet well techniques. While analyses of crystallographic structures provide understood has become apparent. In this regard, it is remarkable precise comparisons between static snapshots (Dalton et al., that very small changes in ligand structure can have large effects 2015; Lans et al., 2015a) molecular dynamics (MD) simulations in ligand function, transforming a PAM into a NAM or vice versa supply the time dimension to the problem allowing the receptor (Wood et al., 2011; Gomez-Santacana et al., 2014). Apparently, to make use of its flexibility, thereby making visible dynamic multiple triggers for agonism and allosterism coexist in the interactions between particular receptor residues and revealing mGlu5 TM domain and work in a concerted and variable fashion. conformational effects (Latorraca et al., 2017). This finding seems not to be rare because as reviewed in Dosa To take a step forward in previous MD simulations of the and Amin (2016), functional switches appears to be a broad mGlu5 receptor (Dalton et al., 2016, 2017) we provide herein phenomenon reflecting the dynamic nature of GPCRs. potential of mean force (PMF) calculations (using umbrella Of note, six key residues have been identified as part of the sampling) linked to MD simulations of the TM domain of affinity receptor network in a rat mGlu5 receptor (Gregory et al., this receptor in its apo form. The present computational study, 2014). These residues can be translated to a human receptor although conceived under a reductionist approach because it sequence corresponding to Pro6553.40 , Tyr6593.44 , Thr7816.46 , includes only one TM domain of a dimeric 3-domain receptor, Trp7856.50 , Ser8097.39 , and Ala8107.40 [superscript numbering has allowed the identification of some conformational features according to Pin et al in class C GPCRs (Pin et al., 2003), that can help to understand the intricacies of mGlu activation adapted from the Ballesteros-Weinstein scheme in Class A mechanism. The approach is consistent with experimental data GPCRs (Ballesteros and Weinstein, 1995)]. Interestingly, the that showed that the TM domain of a truncated mGlu5 receptor Trp7856.50 Ala mutation has different effects depending on ligand displays the same agonist-independent constitutive activity as Frontiers in Molecular Biosciences | www.frontiersin.org 2 March 2020 | Volume 7 | Article 38 Lans et al. mGlu5 TM Activation Mechanism the wild-type receptor (Goudet et al., 2004). Thus, the present that the side chain nitrogen of Trp7856.50 , the hydroxyl group of study aims to structurally explore the mGlu5 TM domain Tyr6593.44 , the hydroxyl group of Ser8097.39 and the Thr7816.46 constitutive activity. side chain form direct hydrogen bonds between them rather than ones mediated by a water molecule. However, the hydroxyl group of Ser8097.39 is able to form a new hydrogen bond with a RESULTS AND DISCUSSION water molecule (Figure 1, right and Supplementary Figure S3). This collection of hydrogen bonds has, as a consequence, the MD Simulations of Apo mGlu5 TM narrowing of the allosteric site. Domain The present work focuses on revealing key structural details of the activation mechanism of apo mGlu5 TM domain at the atomic mGlu5 TM Domain Activation Through level. To this end, the reported crystallographic structure of the Umbrella Sampling Simulations mGlu5 TM domain in complex with the NAM mavoglurant (PDB Starting from the last receptor structure of the MD simulation ID: 4OO9) (Dore et al., 2014) was used as a reference state. Thus, described above, we calculated the PMF, using umbrella sampling the receptor displays the typical structural features of an inactive simulation, for the activation of the TM domain of apo mGlu5. state. In particular, the ionic lock between Lys6653.50 on TM The PMF provides us with a free energy profile as a function of a helix 3 (TM3) and Glu7706.35 (TM6), which is characteristic of collective variable, which can describe a specific process. Three the inactive state of Class C GPCRs, (Pin and Bettler, 2016) is collective variables were tested for modeling mGlu5 receptor closed. Additionally, Asn767 on intracellular loop 3 (ICL3) makes activation: (1) the center of mass distance between the Cα atoms hydrogen bond interactions with Ser612 on intracellular loop 1 of residues Ser612 and Ser614 (both on ICL1) and the Cα (ICL1), which extends from residues Tyr604 to Ser614. atoms of residues Asn767 (ICL3) and Glu7706.35 (intracellular Because the crystallographic structure of the mavoglurant- end of TM6). This distance represents the disruption/formation mGlu5 complex is partially incomplete and it also includes of an interaction at the intracellular side between TM6/ICL3 some modifications of the receptor amino acid sequence, it was and ICL1. (2) The distance between the Lys6653.50 -Nξ and necessary to work with a complete model of the wt TM domain Glu7706.35 -Cδ atoms (K665-Nξ and E770-Cδ). This distance structure in the apo form. To this end, a model developed defines the ionic lock between TM3 and TM6 at the intracellular previously by our group was used (see section “Methods”). Then, side. (3) The center of mass distance between ICL1 and atom to simulate the activation of the mGlu5 TM domain, the receptor Lys6653.50 -Nξ (ICL1 and K665-Nξ ). This distance represents the model was equilibrated and, subsequently, an MD simulation disruption/formation of an interaction between TM3 and ICL1 of 380 ns was carried out. The RMSD of the backbone atoms (specifically between K665 and S613) at the intracellular side through the MD simulation showed a stabilization of the TM of the receptor (Figure 2) as observed in the crystal structure domain (Supplementary Figures S1, S2). Briefly, the TM domain (Dore et al., 2014). maintains its inactive state, which means that the Lys6653.50 - The PMFs were computed at different simulation times Glu7706.35 ionic lock and the Asn767-Ser612 interactions are (500 ps and 1, 3, and 4 ns). It was considered that the PMFs conserved along the course of the MD simulation. Because of converged when the maximum change between the last PMF this, we did not observe any cavity formation at the intracellular and the previous one was lower than 2 kcal/mol (Supplementary side of the TM domain, which is presumably necessary for Gq Figure S5). It can be seen that, in general, the histograms for the coupling, based on currently known Class A GPCR-G protein collective variables show a strong overlapping between adjacent crystal structures (Weis and Kobilka, 2018). windows. Such overlapping is a key point in umbrella sampling However, despite this, in our MD simulation of the mGlu5 TM simulations. Two neighboring windows need to overlap with domain, a number of structural changes were identified, with the each other in order that a continuous PMF can be obtained main variations affecting the allosteric binding site (Figure 1). from the calculations. Moreover, strong overlapping is required In particular, we observed a movement of the extracellular by the weighted histogram analysis method (WHAM) for PMF part of TM7 through which TM7 approaches TM3 (Figure 1, construction (Kästner, 2011). However, the parameter choice for left). It is worth noting that our model was constructed from umbrella sampling simulations (i.e., the spring constant and the a crystal structure where the NAM mavoglurant is bound to number of umbrella sampling windows) is critical because greater the receptor in the allosteric binding site (Dore et al., 2014). overlapping requires more computational resources. Thus, a However, in our apo model the allosteric binding site is empty. balance between overlapping and computation time is needed Thus, the space of the allosteric site was partially filled by TM7 to reach a feasible PMF. In this regard, the ionic lock collective during the MD simulation. As expected, the movement of TM7 variable contained a considerable overlap in the present study. affects some interactions in the allosteric site. In the mGlu5 Consequently, in future studies, a better separation between crystal structure (Figure 1, center) there is a hydrogen bond adjacent windows would be necessary to ensure more efficient network, mediated by a water molecule, involving Tyr6593.44 and PMF computations. Thr7816.46 side chains and the backbone carbonyl of Ser8097.39 . Comparison between the energetic profiles of the three In our simulation, Trp7856.50 , Tyr6593.44 and Ser8097.39 side PMFs used shows that the PMF calculated using the collective chains rearrange themselves in order to fill empty spaces inside variable ICL1-TM6/ICL3 yielded the lowest energy values when the allosteric pocket. These movements are made in such a way attempting to activate the TM domain of mGlu5 (Figure 3). Frontiers in Molecular Biosciences | www.frontiersin.org 3 March 2020 | Volume 7 | Article 38 Lans et al. mGlu5 TM Activation Mechanism FIGURE 1 | (A) The inactive TM domain of mGlu5. The initial structure is shown in brown and a snapshot after 380 ns of the MD simulation is shown in green. During the MD simulation, we observed an inward movement of TM7, with this helix approaching TM3. (B) Crystallographic structure of the allosteric site of the TM domain of the mGlu5-NAM complex (PDB ID: 4OO9) (Dore et al., 2014). (C) TM7 movement during the MD simulation promotes conformational changes of several side chains in the allosteric pocket. Tyr6593.44 , Thr7816.46 , Trp7856.50 , and Ser8097.39 side chains rearrange to fill empty spaces inside the allosteric pocket yielding a narrowing of the allosteric site, which is maintained via a new hydrogen bonding network. did not show any clear minimum along its trajectory. This is potentially in agreement with previous studies that showed that the breaking of the ionic lock is not a sufficient condition for GPCR activation (Dror et al., 2009; Lans et al., 2015b) because it is possible to find some inactive Class A GPCR structures with a broken ionic lock [see for instance A2AR PDB ID: 3EML (Jaakola et al., 2008) and β2AR PDB ID: 2RH1 (Cherezov et al., 2007)]. Furthermore, in MD simulations revealing ligand- mediated mGlu4 conformational change, it was found that receptor “activation” resulted in transient disruption of the ionic lock but not a clear break (Dalton et al., 2017). We admit that the critical point found within the ICL1- TM6/ICL3 PMF is not a pronounced minimum and the structural and functional properties characteristic of a true FIGURE 2 | The three collective variables tested on the intracellular side of minimum may not be given in this case. The reason for that may mGlu5 TM domain. (1) The center of mass distance between the Cα atoms of lay in the limitations of our approach, which has reduced the residues Ser612 and Ser614 (ICL1) and the Cα atoms of residues Asn767 structural complexity of a three domain (VFT, CRD, and TM) (ICL3) and Glu7706.35 (TM6) (orange line). (2) The distance between the mGlu dimer to a single TM domain. Moreover, the G protein, Lys6653.50 -Nξ and Glu7706.35 -Cδ (K665-Nξ and E770-Cδ) atoms (red line). which would allow the intracellular stabilization of the active (3) The center of mass distance between ICL1 and Lys6653.50 -Nξ atom (ICL1 and K665-Nξ ) (blue line). state, has not been included either. However, despite its structural simplicity, our approach may be useful enough for the structural exploration of constitutive mGlu5 activity. As such, we tentatively call the TM conformations included within the blue circle in Interestingly, in contrast to the others, which display nearly linear Figure 3, “active-like conformations.” relationships, the ICL1-TM6/ICL3 PMF shows an energetic Due to the absence of G protein-bound crystal structures of profile that could reflect the exploration of structural features class C GPCRs, the activation mechanism of mGlu receptors involved in the triggering of receptor activation. In particular, has been proposed as homologous to class A GPCRs. However, a region was detected (blue circle, top of Figure 3) which it should be noticed that class A and class C GPCRs share includes a critical point that appears to identify a higher-energy low sequence similarity (Pin et al., 2003) and, more noticeably, “pseudo-stable” receptor state. It could be hypothesized that the micro-switches that have been linked to class A GPCR this region corresponds to an active-like receptor conformation activation are not conserved in class C GPCRs. This includes on the basis that separation of intracellular loops is predicted sequence motifs in TM3 (DRY), TM6 (CWxP except for to be a necessary feature of mGlu receptor activation (Dalton conserved Trp6.48a6.50c ), TM7 (NPxxY) and the Ile3.40a , Pro5.50a , et al., 2017). Interestingly, the PMF whose collective variable and Phe6.44a residues of the “transmission switch” of class A involves the explicit breaking of the Lys665-Glu770 ionic lock GPCRs (Tehan et al., 2014) (superscripts numbering followed Frontiers in Molecular Biosciences | www.frontiersin.org 4 March 2020 | Volume 7 | Article 38 Lans et al. mGlu5 TM Activation Mechanism residue function is still uncertain, as no class C GPCR has been crystallized into a G protein-bound state up to date. The present study has identified some structural features that may distinguish the mGlu5 TM activation from that of class A GPCRs. Our simulations suggest that the most relevant interactions for the stabilization of the inactive state are those between TM6 and ICL1. The path from inactive to “active-like conformation” includes the reorientation of the flexible ICL3 (Figure 4 and Supplementary Figure S4) (an outward movement) and the breaking of the first H-bond of TM6: O(Glu770)-N(Ile774), which could be associated to the ICL3 outward movement. These changes were made without relevant movements or changes in TM6. Nevertheless, we have not found evidence indicating that the O(Glu770)-N(Ile774) H-bond needs to break for the receptor activation. Moreover, our results show that the cavity formation for the G-protein coupling (see section"Docking”) only requires slight movements of TM6. Thus, our results suggest a different activation mechanism of the TM domain of mGlu5 with respect to Class A GPCRs. Activation of Class A GPCRs requires the breaking of the ionic lock between TM3 and TM6 at their intracellular side and a large outward movement of TM6. However, it appears that the activation of the Class C GPCR mGlu5 requires, in addition to disruption of the ionic lock (Supplementary Figure S6), the breaking of interactions between the intracellular end of TM6 and ICL3 and ICL1. These changes can be made without affecting the general structure and stability of the receptor (Supplementary Figure S7 and Supplementary Movie S1). Furthermore, the present study suggests that the activation of the TM domain of mGlu5 does not involve an explicit outward movement or lengthening of TM6. Instead, TM5 and ICL3 may display a slight outward movement (Figure 4), which is a consequence of the ICL3-ICL1 separation and the very FIGURE 3 | Calculated PMFs for the activation of the mGlu5 TM domain using short ICL3 at mGlu5. This is further supported by the remarkably three different collective variables. (A) The center of mass distance between shorter length of the intracellular ends of TM5 and TM6 in ICL1 (Cα atoms of residues Ser612 and Ser614) and ICL3/TM6 (Cα atoms of mGlu receptors compared to class A GPCRs (i.e., TM5 and TM6 residues Asn767 and Glu7706.35 ). (B) Distance between the ionic lock atoms of β2 adrenergic receptor extend for 18 and 9 residues more Lys6653.50 -Nξ and Glu7706.35 -Cδ (K665-Nξ -E770-Cδ). (C) Distance between than mGlu5 into the intracellular region, respectively, see section the center of mass of ICL1 and Lys6653.50 -Nξ atom (ICL1 and K665-Nξ ). “Docking”). This data is in agreement with the importance of the conformation of intracellular loops for mGlu receptor activation and G protein binding found both experimentally (Gomeza et al., by “a” according to Ballesteros-Weinstein scheme (Ballesteros 1996; Francesconi and Duvoisin, 1998) and in MD simulations and Weinstein, 1995)] for class A GPCRs or followed by “c” (Dalton et al., 2017). according to Pin et al. (2003) scheme for class C GPCRs). Despite The hypothesis proposed in this study for the mGlu5 TM previous studies have proposed functional similarities between domain activation is compatible with experimental studies. The class A GPCR micro-switches and certain non-homologous class TM domains linked to the active and inactive VFT states of the C GPCR residues (Perez-Benito et al., 2017; Llinas Del Torrent full-length mGlu5 obtained from cryo-EM did not significantly et al., 2019) their respective conformational changes have been differ between them (Koehl et al., 2019). Importantly, the outward observed only in MD simulations of mGlu receptor homology movement of TM6, observed in the activation of class A GPCRs, models, but not in mGlu receptor crystal structures (Dore was not observed in active mGlu5 (Koehl et al., 2019). In addition, et al., 2014; Christopher et al., 2015; Koehl et al., 2019) or in both active and inactive structures show a dimeric arrangement, MD simulations starting from those (Feng et al., 2015; Dalton with the two TM domains completely separated in the inactive et al., 2017; Perez-Benito et al., 2017; Llinas Del Torrent et al., VFT state. However, in the active VFT state the TM domains 2019). Furthermore, while the outward movement of TM6 is a show a strong interaction between their TM6s, which appears hallmark of class A GPCR activation, the extent of this movement to be a hallmark of the mGlu activation process (Koehl et al., is noticeably lower in mGlu5 even when enhanced sampling 2019). Thus, it is reasonable to hypothesize that if the TM6-TM6 methods are employed (Cong et al., 2019). Thus, the analogy mutual stabilization is strong enough, the intrasubunit outward between class A and class C GPCR conformational change and movement of TM6 is less likely. Alternatively, it has also been Frontiers in Molecular Biosciences | www.frontiersin.org 5 March 2020 | Volume 7 | Article 38 Lans et al. mGlu5 TM Activation Mechanism FIGURE 4 | Molecular superposition between the inactive crystallographic (brown) and active-like (from-umbrella-sampling/MD, green) TM domains of mGlu5. (A) The arrows indicate the separation between ICL1-TM6/ICL3 after umbrella sampling simulation. (B) Lateral view of inactive and post-PMF/MD TM domains of mGlu5. Note the outward movement undergone by TM5 to reach an active-like conformation. (C) Bottom view of the inactive and post-PMF/MD TM domains of mGlu5. The dotted line shows the separation between Tyr6593.44 and Ser8097.39 . suggested that there is enough space for the intracellular part of the TM6 of at least one of the subunits to move outward (Pin et al., 2019). Finally, it is worth mentioning that a slight outward movement of TM5 can be observed in the VFT-active cryo-EM structure (Koehl et al., 2019) which if it could be associated to TM activation it would be consistent with the proposal presented in the present study. In the active-like conformations identified in the present study by umbrella sampling simulation and through the unbiased trajectory of the MD simulation of the initially inactive mGlu5, we observed some residues in the allosteric binding site that undergo re-packing because of the deletion of the co-crystallized FIGURE 5 | The intracellular active-like conformation of the TM domain of NAM (Figure 1). In order to detect conformational changes in mGlu5. The structure was obtained by umbrella sampling/molecular dynamics the allosteric site that could be involved in the activation of simulation for the collective variable ICL1-TM6/ICL3. The interaction the TM domain of mGlu5, we sought to identify similarities ASN767_Glu7706.35 -ICL1 is disrupted. Because of this, there is a cavity formation in the intracellular side of the TM domain, which is likely to be and differences between the inactive crystal state and active-like necessary for Gq protein coupling. conformations of the receptor obtained from PMF application. We observed that Tyr6593.44 undergoes a conformational change in the active-like conformation with respect to the inactive one. As a consequence of this change, the aromatic ring of Tyr6593.44 Protein-Protein Docking of Gq With the is showed more displaced toward Ile7515.51 and the hydrogen Active-Like Conformation of the TM bond between Tyr6593.44 and Ser8097.39 side chains is disrupted, Domain of mGlu5 showing a longer distance (Figure 4 and Supplementary Figure Most of what is known of GPCR activation at structural level S3). All these structural features could be a consequence of the has come from the determination of active and inactive Class A outward movement of TM5 through the activation-like process crystal structures (Weis and Kobilka, 2018). Structural analyses (Figure 4). However, this chain of events could occur in the have shown that agonist binding at the upper part of the receptor opposite way. Thus, we speculate that having a bound ago-PAM, is not sufficient to activate the receptor and that concomitant G this could modulate Tyr6593.44 conformation on TM3 in a way protein binding at the cytoplasmic side of the transmembrane that Tyr6593.44 is moved toward Ile7515.51 on TM5, thus forcing domain of receptors is necessary (Gregorio et al., 2017). the outward movement of TM5. This movement could induce a To evaluate to what extent the “active-like” conformation of conformational change of Asn767 located on intracellular loop 3 the TM domain of mGlu5 is active in the classical sense, we (ICL3) and Glu7706.35 (intracellular end of TM6). Thus, Asn767 performed a protein-protein docking to examine the capability and Glu7706.35 could break their interactions with ICL1 allowing of the receptor to accommodate a bound GTPase and the helical the formation of a cavity in the intracellular side, necessary for domain of Gq protein (PDB ID: 3AH8) at its intracellular side the receptor activation (Figure 5). Thus, as it happens in class A (Nishimura et al., 2010). If the resulting docking complex would GPCRs, one of the molecular triggers for the activation of the TM resemble the crystallographic β2 adrenergic receptor-Gs protein domain of mGlu5 could be a structural change initiated at TM3 complex (PDB ID: 3SN6) (Rasmussen et al., 2011) it would be (Dalton et al., 2015; Lans et al., 2015a). reasonable to consider the mGlu5 conformation as active. Frontiers in Molecular Biosciences | www.frontiersin.org 6 March 2020 | Volume 7 | Article 38 Lans et al. mGlu5 TM Activation Mechanism We used Cluspro webserver (Kozakov et al., 2013) and ID: 3SN6) (Rasmussen et al., 2011; Figure 6). Such as in the HADDOCK2.2 webserver (van Zundert et al., 2016) to carry crystal complex, the C terminal of the α5 helix within the Gα out the protein-protein docking. To validate the usefulness of subunit of the Gq protein fits between the intracellular ends of Cluspro and Haddock to predict the correct interaction complex TM3, TM5 and TM6 (as well as in between ICL1, ICL2, and in this kind of system we carried out a docking of the Gs protein ICL3). Additionally, it is observed from Cluspro conformation, to the β2 adrenergic receptor. The closest pose of Gs, predicted that Lys6653.50 (the positive counterpart of the ionic lock) forms from the docking to the crystal complex (3SN6) (Rasmussen et al., a cation-π interaction with Tyr352 of the Gα subunit of Gq, 2011) using Cluspro, belongs to the second populated cluster showing a distance of 4.0 Å between the Cγ_Y351 and Nζ_K98 and has a RMSD of 7.2 Å with respect to the crystal structure atoms (in the crystal Gs-β2 adrenergic receptor complex, the (Supplementary Figure S8 and Supplementary Tables S1A,B). equivalent distance Cγ_Y391-Cζ_R131 is of 4.7 Å). Additionally, On the other hand, the best pose from Haddock has a RMSD of a docking between the mGlu5 TM inactive conformation and the 2.41 Å with respect to the crystal structure. Thus, both servers Gq protein was carried out. The results indicate that the “active- are able to predict acceptable and medium quality poses of the like” conformation accommodates the Gq protein deeper in the complex Gs-β2 adrenergic receptor according to CAPRI (Critical intracellular part than the inactive conformation of the receptor Assessment of Prediction of Interactions) criterion (Lensink et al., does (Supplementary Figure S9). This is an indication that 2007). Then, we performed a protein-protein docking between the receptor conformation obtained in the present study might Gq and the “active-like” conformation of the TM domain of represent an active conformation of the TM domain of mGlu5. mGlu5 using Cluspro (Kozakov et al., 2013) and Haddock (van Zundert et al., 2016) web servers. The results show that our active-like receptor conformation is able to accommodate the CONCLUSION binding of Gq at its intracellular side (Figure 6). In particular, two conformation of the Gq-mGlu5 complex, belonging to clusters 19 The present study shows a plausible pathway for the activation and 3, predicted using the balanced Cluspro scoring function and of the TM domain of mGlu5. Our results indicate that polar the Haddock score, respectively (Supplementary Tables S2A,B) interactions between ICL1 and the intracellular side of TM6 resemble the crystal Gs-β2 adrenergic receptor-complex (PDB have to be disrupted to allow the cavity formation at the FIGURE 6 | The conformations of the complex between Gq and the TM domain of mGlu5 obtained by using protein-protein docking using Cluspro (B,E) and Haddock (C,F) are similar to the crystallographic β2 adrenergic receptor-G protein complex (PDB ID: 3SN6) (Rasmussen et al., 2011) (A,D). The C terminal of α5 helix within Gα subunit of Gq is located between the intracellular ends of TM3, TM5 and TM6. Frontiers in Molecular Biosciences | www.frontiersin.org 7 March 2020 | Volume 7 | Article 38 Lans et al. mGlu5 TM Activation Mechanism intracellular side of the transmembrane domain, which is likely to (Berger et al., 1997) (obtained from http://moose.bio.ucalgary. be necessary for Gq protein coupling. This finding is in agreement ca/). The embedding protocol was achieved using the inflateGRO with a previous study (Dore et al., 2014) which reports that methodology (Kandt et al., 2007). The water model used was Ser613Ala mutation, in ICL1, and Glu770Ala mutation, in TM6, the single point charge (SPC) model (Berendsen et al., 1981). To which presumably disrupt polar interactions between ICL1 and remove the positive net charge of the system, 14 Cl− ions were TM6, result in constitutive receptor activity. In this regard, the added as counterions. The resulting systems were minimized up importance of ICL1 for recognizing G proteins in a class A to the maximum force <1000.0 kJ.mol/nm using the steepest GPCR has been recently shown by NMR experiments (Sounier descent algorithm. During the minimization the positions of the et al., 2015). In addition, we have noted that the interaction protein and membrane atoms were restrained by using a force between Tyr6593.44 and Ile7515.51 is maintained in the “active- constant of 1000 kJ.mol−1 .nm−2 . like” mGlu5 conformation, indicating that this interaction might contribute to activation or positive modulation of mGlu5. Equilibration and MD Production Moreover, the interaction between Tyr6593.44 and Ser8097.39 , The equilibration process was carried out in three steps. The mediated by a water molecule in the NAM-mGlu5 crystal first step consisted of 100 ps of NVT simulation, in which the structure or directly observed in our unbiased MD and missing in z-coordinate of the POPC phosphorous atoms in the membrane the active-like conformation, could serve to stabilize the inactive as well as the heavy atoms of the mGlu5 TM domain were conformation. Indeed, previous studies (Gregory et al., 2013) restrained. The reference temperature was set at 310 K, using showed that Tyr659Val, Thr781Ala, and Ser809Ala mutations the modified Berendsen thermostat coupling method with a time induce changes in the pharmacology of several modulator constant of 0.1 ps. The second step consisted of 3 substeps. The compounds, changing these from PAMs to NAMs or SAMs. first substep consisted of 3 ns of NPT simulation (restraining Finally, we have shown that the active-like conformation of both the z-coordinate of the POPC phosphorous atoms in the TM domain of mGlu5 obtained in the present study is able to the membrane and the heavy atoms of the TM domain); a accommodate the Gq protein. It is worth noting that although we force constant of 10,000 kJ.mol−1 .nm−2 for these atoms was have not identified a clear minimum corresponding to the active- used. In the second substep (2 ns length) only the movements like conformation, the obtained conformation could represent a of the heavy atoms of the TM domain were restrained by useful model of an active-like state. In this regard, it is likely that using a force constant of 10,000 kJ.mol−1 .nm−2 . In the third to obtain a clear minimum of the TM active state it would be substep (3 ns length) only the movements of the heavy atoms necessary to simulate the activation process in presence of the of the TM domain were restrained by using a force constant extracellular VFT domain and as a dimeric receptor: a receptor of 1000 kJ.mol−1 .nm−2 . Finally, the third step consisted of model system beyond the scope of the present study. 12 ns without any restraint. Then, an unbiased MD simulation production of 380 ns was carried out without imposing any condition. In each step, the reference temperature was set at EXPERIMENTAL PROCEDURES 310 K, the Nosé-Hoover thermostat coupling method with a time constant of 1.0 ps was used, and the pressure was set at 1 atm Classical MD simulations were performed to study the activation by using a semi-isotropic method with a time constant of 5.0 ps. of the TM domain of mGlu5 in the apo form. The GROMOS53A6 The particle mesh Ewald method (Essmann et al., 1995) was used force field (Oostenbrink et al., 2004; Hess et al., 2008) was used. to estimate the long range Coulomb interaction and periodic The initial structure used was a complete mGlu5 TM model, built boundary conditions were used in all the steps. in a previous study (Dalton et al., 2016) from the mavoglurant- mGlu5 crystal structure (PDB ID: 4OO9) (Dore et al., 2014) in Potential of Mean Forces Calculation which the NAM mavoglurant and all the co-crystallized water The PMF was calculated to estimate the free energy profile of molecules were removed. In particular, the water molecule at the activation of apo mGlu5 TM domain. The starting structure the bottom of the allosteric binding pocket was removed in our was the last snapshot of the previous MD production phase. mGlu5 apo model because in the crystal structure it displays Three different PMFs were determined using the umbrella molecular interactions with the receptor that are dependent on sampling technique (Kumar et al., 1992; Boczko and Brooks, the NAM presence (Dore et al., 2014). Thus, a destabilization 1993; Rajamani et al., 2003) along three different collective of the water molecule is expected when the NAM is absent. In variables (Figure 2): (1) The center of mass distance between agreement with this assumption, it was shown in a previous ICL1 (Cαs of residues Ser612 and Ser614) and Cαs of residues study (Dalton et al., 2017) that this water molecule is not stable Asn767 (ICL3) and Glu7706.35 (TM6) (orange line); (2) Distance in the MD simulations performed on the free mGlu5 receptor. between the K6653.50 -Nξ and E7706.35 -Cδ (K665-Nξ and E770- We performed all simulations on a standard workstation using Cδ) atoms (red line); (3) the center of mass distance between ICL1 GROMACS v4.6.1 (Hess et al., 2008). and Lys6653.50 -Nξ atom (ICL1 and K665-Nξ ) (blue line). Minimization Umbrella Sampling Simulation The ligand-free mGlu5 model was inserted into a pre-equilibrated To choose the starting structures of the windows, we carried and fully hydrated POPC lipid bilayer of 123 molecules out 40 sequential MD simulations of 500 ps each, starting from (Supplementary Figure S1). Berger lipids parameters were used the last snapshot of the unbiased apo MD simulation. The Frontiers in Molecular Biosciences | www.frontiersin.org 8 March 2020 | Volume 7 | Article 38 Lans et al. mGlu5 TM Activation Mechanism velocities and coordinates of the last configuration generated in Gq-mGlu5 Docking each of the MD simulations were used as starting points for The GTPase and helical domains of Gq protein [PDB ID: 3AH8 the following one and successively in this way to complete the (Nishimura et al., 2010) Gly1-Val355] were used as ligand while 40 MD simulations. The sampling of each MD simulation was our “active-like” conformation of the TM domain of mGlu5 centered at a specific value of the collective variable (0.05 nm (obtained from the umbrella sampling simulation) was used as larger than the previous simulation) by using a spring constant of receptor. To mimic the membrane effect, in Cluspro web server, 4000 kJ/mol.nm2 . Finally, from the set of 40 MD simulations, we the residues Ser568-Val599, Leu622-Ser660, Val695-Phe756 and localized 24 configurations whose values in the collective variable Tyr779-Glu815 (at the receptor) and Gly1-Met20 (at the ligand) periodically increased 0.05 nm between them and, thus, let the were set as repulsive residues. On the contrary, Lys665 residue receptor going from the inactive configuration to the active- (positive counterpart in the ionic lock at mGlu5) and Leu349- like one. Each configuration was stabilized for at least 50 ps Glu351 residues were set as attractive residues in the receptor in each MD simulation. In this way, we started each umbrella and ligand, respectively. Finally, in Haddock web Server, Lys665 sampling window from a stabilized structure. In the case of residue (positive counterpart in the ionic lock at mGlu5) and the ionic lock collective variable, it was necessary to localize Leu349-Glu351 residues were set as active residues. 70 starting structures whose values in the collective variable periodically increased 0.017 nm between them and, thus, let the receptor going from the inactive configuration to the active- DATA AVAILABILITY STATEMENT like one. This procedure allowed us to obtain a good sampling of the entire range of the collective variable. Subsequently, we The datasets generated for this study are available on request to carried out 4 ns MD per window for the ICL1-TM6/ICL3 and the corresponding author. ICL1-K665 collective variables and 1 ns for the ionic lock. During the umbrella sampling simulations the trajectories were saved every 2 ps. AUTHOR CONTRIBUTIONS Finally, the PMFs were calculated using WHAM (Hub et al., 2010) implemented in GROMACS 4.6.1. Statistical errors IL performed the MD simulations and PMF studies, analyzed the were estimated using Bayesian bootstrap analysis (N = 50) results, and wrote the manuscript. ÓD participated in structural (Hub et al., 2010). analysis. JD revised the manuscript. JG supervised the study and wrote the manuscript. Protein-Protein Docking The protein-protein docking was carried out using Cluspro FUNDING webserver (Kozakov et al., 2013) with the balanced scoring function and Haddock 2.2 (van Zundert et al., 2016). To validate This study was supported in part by the Ministerio de Economía the usefulness of Cluspro and Haddock programs to predict the y Competitividad (SAF2014-58396-R) and Ministerio de Ciencia, correct interaction complex in the GPCR-G protein system, we Innovación y Universidades (SAF2017-87199-R). attempted to reproduce the Gs-β2 adrenergic X-RAY complex (PDB ID: 3SN6) (Rasmussen et al., 2011). Thus, we carried out a docking of the Gs protein to the β2 adrenergic receptor using SUPPLEMENTARY MATERIAL the Gs and β2 adrenergic chains from the 3SN6 PDB structure (Rasmussen et al., 2011). The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmolb.2020. Cluspro Docking 00038/full#supplementary-material The A chain (Thr9-Leu394) was used as ligand while the R chain MOVIE S1 | mGlu5 TM receptor model inserted in a POPC membrane model. (Glu30-Cys341) was used as receptor. To mimic the membrane Backbone RMSD of the mGlu5 TM domain relative to the starting structure of the or avoid poses of Gs protein interacting with the receptor part in unbiased molecular dynamics simulation. Time-window average (using a 10 ns contact with the membrane, we set some repulsive residues in the window) of selected distances in the allosteric site along the unbiased molecular dynamics simulation. Backbone RMSD of the ICL3, relative to the starting receptor (Glu30-Val48; Cys77-Glu122; Ile159-Tyr219 and Ile278- structure of the unbiased molecular dynamics simulation. The PMF for each Gly320) and also in the ligand (Thr9-Gln19). Additionally, the collective variable calculated at different simulation times. Average distance of the Arg131 residue (in the receptor) and Arg389-Glu392 residues in Ionic lock along the ICL1-TM6/ICL3 collective variable. RMSD of the receptor the ligand were considered as attracting residues. backbone with respect to the inactive structures as a function of the value of the ICL1-TM6/ICL3 collective variable. The conformation of the complex between Gs and the TM domain of β2 adrenergic receptor, obtained by protein-protein docking Haddock Docking using Cluspro (blue-cyan complex) and Haddock2.2 (blue-yellow complex), are The A chain (Thr9-Leu394) was used as ligand while the R chain similar to the crystallographic β2 adrenergic receptor-G protein complex (PDB ID: 3SN6, blue-green complex) with a RMSD of 7.2 Å and 2.4 Å, respectively. (Glu30-Cys341) was used as receptor. The Arg131 residue (in Comparison of the inactive mGlu5 TM-Gq pose with the “active-like” mGlu5 the receptor) and Arg389-Glu392 in the ligand were defined as TM-Gq pose. Cluspro docking rank between the Gs protein and the β2 adrenergic active residues. receptors. Haddock docking rank between the Gs protein and the β2 adrenergic Frontiers in Molecular Biosciences | www.frontiersin.org 9 March 2020 | Volume 7 | Article 38 Lans et al. mGlu5 TM Activation Mechanism receptors. Cluspro docking rank between the “active-like” conformation of the TM Cluspro docking rank between the inactive conformation of the TM domain of domain of mGlu5 and the GTPase and helical domain of Gq protein (3AH8 PDB). mGlu5 and the GTPase and helical domain of Gq protein (PDB ID: 3AH8). Haddock docking rank between the “active-like” conformation of the TM domain Haddock docking rank between the inactive conformation of the TM domain of of mGlu5 and the GTPase and helical domain of Gq protein (PDB ID: 3AH8). mGlu5 and the GTPase and helical domain of Gq protein (PDB ID: 3AH8). REFERENCES Essmann, U., Perera, L., Berkowitz, M. L., Darden, T., Lee, H., and Pedersen, L. G. (1995). A smooth particle mesh Ewald method. J. Chem. Phys. 103, 8577–8593. Ballesteros, J. A., and Weinstein, H. (1995). Integrated methods for the doi: 10.1063/1.470117 construction of three-dimensional models and computational probing of Feng, Z., Ma, S., Hu, G., and Xie, X. Q. (2015). Allosteric binding site and structure-function relations in G protein-coupled receptors. Methods Neurosci. activation mechanism of class C G-protein coupled receptors: metabotropic 25, 366–428. doi: 10.1016/s1043-9471(05)80049-7 glutamate receptor family. AAPS J. 17, 737–753. doi: 10.1208/s12248-015- Bennett, K. A., Dore, A. S., Christopher, J. A., Weiss, D. R., and Marshall, F. H. 9742-8 (2014). Structures of mGluRs shed light on the challenges of drug development Foster, D. J., and Conn, P. J. (2017). Allosteric modulation of GPCRs: new insights of allosteric modulators. Curr. Opin. Pharmacol. 20, 1–7. doi: 10.1016/j.coph. and potential utility for treatment of schizophrenia and other CNS disorders. 2014.09.022 Neuron 94, 431–446. doi: 10.1016/j.neuron.2017.03.016 Berendsen, H. J., Postma, J. P. M., van Gusteren, W. F., and Hermans, J. (1981). Francesconi, A., and Duvoisin, R. M. (1998). Role of the second and third “Interaction models for water in relation to protein hydration,” in Proceedings of intracellular loops of metabotropic glutamate receptors in mediating dual signal the Fourteenth Jerusalem Symposium on Quantum Chemistry and Biochemistry transduction activation. J. Biol. Chem. 273, 5615–5624. doi: 10.1074/jbc.273.10. Held in Jerusalem, Israel: Intermolecular forces, ed. B. Pullman (Dordrecht: 5615 Reidel Publishing Company), 331–342. doi: 10.1007/978-94-015-7658-1_21 Gomeza, J., Joly, C., Kuhn, R., Knopfel, T., Bockaert, J., and Pin, J. P. (1996). The Berger, O., Edholm, O., and Jahnig, F. (1997). Molecular dynamics simulations second intracellular loop of metabotropic glutamate receptor 1 cooperates with of a fluid bilayer of dipalmitoylphosphatidylcholine at full hydration, constant the other intracellular domains to control coupling to G-proteins. J. Biol. Chem. pressure, and constant temperature. Biophys. J. 72, 2002–2013. doi: 10.1016/ 271, 2199–2205. doi: 10.1074/jbc.271.4.2199 s0006-3495(97)78845-3 Gomez-Santacana, X., Dalton, J. A., Rovira, X., Pin, J. P., Goudet, C., and Gorostiza, Boczko, E. M., and Brooks, C. L. III. (1993). Constant-temperature free energy P. (2017). Positional isomers of bispyridine benzene derivatives induce efficacy surfaces for physical and chemical processes. J. Phys. Chem. 97, 4509–4513. changes on mGlu5 negative allosteric modulation. Eur. J. Med. Chem. 127, doi: 10.1021/j100119a043 567–576. doi: 10.1016/j.ejmech.2017.01.013 Cherezov, V., Rosenbaum, D. M., Hanson, M. A., Rasmussen, S. G., Thian, F. S., Gomez-Santacana, X., Rovira, X., Dalton, J. A., Goudet, C., Pin, J. P., and and Kobilka, T. S. (2007). High-resolution crystal structure of an engineered Gorostiza, P. (2014). A double effect molecular switch leads to a novel potent human beta2-adrenergic G protein-coupled receptor. Science 318, 1258–1265. negative allosteric modulator of metabotropic glutamate receptor 5. Med. Christopher, J., Aves, S. J., Bennett, K. A., Dore, A. S., Errey, J. C., and Jazayeri, Chem. Commun. 5, 1548–1554. doi: 10.1039/c4md00208c A. (2015). Fragment and structure-based drug discovery for a class C GPCR: Goudet, C., Gaven, F., Kniazeff, J., Vol, C., Liu, J., Cohen-Gonsaud, M., et al. discovery of the mGlu5 negative allosteric modulator HTL14242 (3-chloro-5- (2004). Heptahelical domain of metabotropic glutamate receptor 5 behaves [6-(5-fluoropyridin-2-yl)pyrimidin-4-yl]benzonitrile). J. Med. Chem. 58, 6653– like rhodopsin-like receptors. Proc. Natl. Acad. Sci. U.S.A. 101, 378–383. doi: 6664. doi: 10.1021/acs.jmedchem.5b00892 10.1073/pnas.0304699101 Cong, X., Cheron, J. B., Golebiowski, J., Antonczak, S., and Fiorucci, S. (2019). Gregorio, G. G., Masureel, M., Hilger, D., Terry, D. S., Juette, M., and Zhao, Allosteric modulation mechanism of the mGluR5 transmembrane domain. H. (2017). Single-molecule analysis of ligand efficacy in beta2AR-G-protein J. Chem. Inform. Model. 59, 2871–2878. doi: 10.1021/acs.jcim.9b00045 activation. Nature 547, 68–73. doi: 10.1038/nature22354 Conn, P. J., Christopoulos, A., and Lindsley, C. W. (2009). Allosteric modulators Gregory, K. J., Nguyen, E. D., Malosh, C., Mendenhall, J. L., Zic, J. Z., and Bates, of GPCRs: a novel approach for the treatment of CNS disorders. Nat. Rev. Drug B. S. (2014). Identification of specific ligand-receptor interactions that govern Discov. 8, 41–54. doi: 10.1038/nrd2760 binding and cooperativity of diverse modulators to a common metabotropic Dalton, J. A., Gomez-Santacana, X., Llebaria, A., and Giraldo, J. (2014). glutamate receptor 5 allosteric site. ACS Chem. Neurosci. 5, 282–295. doi: 10. Computational analysis of negative and positive allosteric modulator binding 1021/cn400225x and function in metabotropic glutamate receptor 5 (in)activation. J. Chem. Inf. Gregory, K. J., Nguyen, E. D., Reiff, S. D., Squire, E. F., Stauffer, S. R., and Model. 54, 1476–1487. doi: 10.1021/ci500127c Lindsley, C. W. (2013). Probing the metabotropic glutamate receptor 5 Dalton, J. A., Lans, I., and Giraldo, J. (2015). Quantifying conformational changes (mGlu(5)) positive allosteric modulator (PAM) binding pocket: discovery of in GPCRs: glimpse of a common functional mechanism. BMC. Bioinformatics point mutations that engender a “molecular switch” in PAM pharmacology. 16:124. doi: 10.1186/s12859-015-0567-3 Mol. Pharmacol. 83, 991–1006. doi: 10.1124/mol.112.083949 Dalton, J. A., Lans, I., Rovira, X., Malhaire, F., Gomez-Santacana, X., Pittolo, Grushevskyi, E. O., Kukaj, T., Schmauder, R., Bock, A., Zabel, U., and Schwabe, T. S., et al. (2016). Shining light on an mGlu5 photoswitchabe NAM: a (2019). Stepwise activation of a class C GPCR begins with millisecond dimer theoretical perspective. Curr. Neuropharmacol. 14, 441–454. doi: 10.2174/ rearrangement. Proc. Natl. Acad. Sci. U.S.A. 116:201900261. doi: 10.1073/pnas. 1570159x13666150407231417 1900261116 Dalton, J. A. R., Pin, J. P., and Giraldo, J. (2017). Analysis of positive and negative Hess, B., Kutzner, C., van der Spoel, D., and Lindahl, E. (2008). GROMACS allosteric modulation in metabotropic glutamate receptors 4 and 5 with a dual 4: algorithms for highly efficient, load-balanced, and scalable molecular ligand. Sci. Rep. 7:4944. doi: 10.1038/s41598-017-05095-5 simulation. J. Chem. Theor. Comput. 4, 435–447. doi: 10.1021/ct700301q Dore, A. S., Okrasa, K., Patel, J. C., Serrano-Vega, M., Bennett, K., and Cooke, Hlavackova, V., Zabel, U., Frankova, D., Batz, J., Hoffmann, C., and Prezeau, L. R. M. (2014). Structure of class C GPCR metabotropic glutamate receptor 5 (2012). Sequential inter- and intrasubunit rearrangements during activation of transmembrane domain. Nature 511, 557–562. doi: 10.1038/nature13396 dimeric metabotropic glutamate receptor 1. Sci. Signal. 5:ra59. doi: 10.1126/ Dosa, P. I., and Amin, E. A. (2016). Tactical approaches to interconverting scisignal.2002720 GPCR agonists and antagonists. J. Med. Chem. 59, 810–840. doi: 10.1021/acs. Hub, J. S., de Groot, B. L., and van der Spoel, D. (2010). g_wham-A Free weighted jmedchem.5b00982 histogram analysis implementation including robust error and autocorrelation Dror, R. O., Arlow, D. H., Borhani, D. W., Jensen, M. O., Piana, S., and Shaw, estimates. J. Chem. Theory Comput. 6, 3713–3720. doi: 10.1021/ct100494z D. E. (2009). Identification of two distinct inactive conformations of the beta2- Jaakola, V. P., Griffith, M. T., Hanson, M. A., Cherezov, V., and Chien, E. Y. (2008). adrenergic receptor reconciles structural and biochemical observations. Proc. The 2.6 angstrom crystal structure of a human A2A adenosine receptor bound Natl. Acad. Sci. U.S.A. 106, 4689–4694. doi: 10.1073/pnas.0811065106 to an antagonist. Science 322, 1211–1217. doi: 10.1126/science.1164772 Frontiers in Molecular Biosciences | www.frontiersin.org 10 March 2020 | Volume 7 | Article 38 Lans et al. mGlu5 TM Activation Mechanism Jojart, B., Orgovan, Z., Marki, A., Pandy-Szekeres, G., Ferenczy, G. G., and Keseru, the GROMOS force-field parameter sets 53A5 and 53A6. J. Comput. Chem. 25, G. M. (2019). Allosteric activation of metabotropic glutamate receptor 5. 1656–1676. doi: 10.1002/jcc.20090 J. Biomol. Struct. Dyn. 1–9. doi: 10.1080/07391102.2019.1638302 Pagano, A., Ruegg, D., Litschig, S., Stoehr, N., Stierlin, C., and Heinrich, M. (2000). Kaae, B. H., Harpsoe, K., Kvist, T., Mathiesen, J. M., Molck, C., and Gloriam, The non-competitive antagonists 2-methyl-6-(phenylethynyl)pyridine and 7- D. (2012). Structure-activity relationships for negative allosteric mGluR5 hydroxyiminocyclopropan[b]chromen-1a-carboxylic acid ethyl ester interact modulators. ChemMedChem 7, 440–451. doi: 10.1002/cmdc.201100578 with overlapping binding pockets in the transmembrane region of group I Kandt, C., Ash, W. L., and Tieleman, D. P. (2007). Setting up and running metabotropic glutamate receptors. J. Biol. Chem. 275, 33750–33758. doi: 10. molecular dynamics simulations of membrane proteins. Methods 41, 475–488. 1074/jbc.m006230200 doi: 10.1016/j.ymeth.2006.08.006 Perez-Benito, L., Doornbos, M. L. J., Cordomi, A., Peeters, L., Lavreysen, H., Pardo, Kästner, J. (2011). Umbrella sampling. WIREs Comput. Mol. Sci. 1, 932–942. doi: L., et al. (2017). Molecular switches of allosteric modulation of the metabotropic 10.1002/wcms.66 glutamate 2 receptor. Structure 25, 1153.e4–1162.e4. doi: 10.1016/j.str.2017. Kenakin, T. (2005). New concepts in drug discovery: collateral efficacy and 05.021 permissive antagonism. Nat. Rev. Drug Discov. 4, 919–927. doi: 10.1038/ Pin, J. P., and Bettler, B. (2016). Organization and functions of mGlu and GABAB nrd1875 receptor complexes. Nature 540, 60–68. doi: 10.1038/nature20566 Kenakin, T., and Strachan, R. T. (2018). PAM-antagonists: a better way to block Pin, J. P., Galvez, T., and Prezeau, L. (2003). Evolution, structure, and activation pathological receptor signaling? Trends Pharmacol. Sci. 39, 748–765. doi: 10. mechanism of family 3/C G-protein-coupled receptors. Pharmacol. Ther. 98, 1016/j.tips.2018.05.001 325–354. doi: 10.1016/s0163-7258(03)00038-x Koehl, A., Hu, H., Feng, D., Sun, B., Zhang, Y., and Robertson, M. J. (2019). Pin, J. P., Kniazeff, J., Prezeau, L., Liu, J. F., and Rondard, P. (2019). GPCR Structural insights into the activation of metabotropic glutamate receptors. interaction as a possible way for allosteric control between receptors. Mol. Cell. Nature 566, 79–84. doi: 10.1038/s41586-019-0881-4 Endocrinol. 486, 89–95. doi: 10.1016/j.mce.2019.02.019 Kozakov, D., Beglov, D., Bohnuud, T., Mottarella, S. E., Xia, B., Hall, D. R., et al. Rajamani, R., Naidoo, K. J., and Gao, J. (2003). Implementation of an adaptive (2013). How good is automated protein docking? Proteins 81, 2159–2166. doi: umbrella sampling method for the calculation of multidimensional potential of 10.1002/prot.24403 mean force of chemical reactions in solution. J. Comput. Chem. 24, 1775–1781. Kumar, S., Rosenberg, J. M., Bouzida, D., Swendsen, R. H., and Kollman, P. A. doi: 10.1002/jcc.10315 (1992). The weighted histogram analysis method for free energy calculations on Rasmussen, S. G., DeVree, B. T., Zou, Y., Kruse, A. C., Chung, K. Y., and Kobilka, biomolecules. I. The method. J. Comp. Chem. 13, 1011–1021. doi: 10.1002/jcc. T. S. (2011). Crystal structure of the beta(2) adrenergic receptor-Gs protein 540130812 complex. Nature 477, 549–557. doi: 10.1038/nature10361 Lans, I., Dalton, J. A., and Giraldo, J. (2015a). Helix 3 acts as a conformational Rook, J. M., Noetzel, M. J., Pouliot, W. A., Bridges, T. M., Vinson, P. N., and Cho, hinge in class a GPCR activation: an analysis of interhelical interaction energies H. P. (2013). Unique signaling profiles of positive allosteric modulators in crystal structures. J. Struct. Biol. 192, 545–553. doi: 10.1016/j.jsb.2015.10.019 of metabotropic glutamate receptor subtype 5 determine differences Lans, I., Dalton, J. A., and Giraldo, J. (2015b). Selective protonation of acidic in vivo activity. Biol. Psychiatry 73, 501–509. doi: 10.1016/j.biopsych.2012. residues triggers opsin activation. J. Phys. Chem. B 119, 9510–9519. doi: 10. 09.012 1021/acs.jpcb.5b01908 Rovira, X., Roche, D., Serra, J., Kniazeff, J., Pin, J. P., and Giraldo, J. Latorraca, N. R., Venkatakrishnan, A. J., and Dror, R. O. (2017). GPCR dynamics: (2008). Modeling the binding and function of metabotropic glutamate structures in motion. Chem. Rev. 117, 139–155. doi: 10.1021/acs.chemrev. receptors. J. Pharmacol. Exp. Ther. 325, 443–456. doi: 10.1124/jpet.107. 6b00177 133967 Lensink, M. F., Mendez, R., and Wodak, S. J. (2007). Docking and scoring protein Sounier, R., Mas, C., Steyaert, J., Laeremans, T., Manglik, A., Huang, W., et al. complexes: CAPRI 3rd Edition. Proteins 69, 704–718. doi: 10.1002/prot.21804 (2015). Propagation of conformational changes during µ-opioid receptor Llinas Del Torrent, C., Casajuana-Martin, N., Pardo, L., Tresadern, G., and Perez- activation. Nature 524, 375–378. doi: 10.1038/nature14680 Benito, L. (2019). Mechanisms underlying allosteric molecular switches of Tehan, B. G., Bortolato, A., Blaney, F. E., Weir, M. P., and Mason, J. S. (2014). metabotropic glutamate receptor 5. J. Chem. Inform. Model. 59, 2456–2466. Unifying family A GPCR theories of activation. Pharmacol. Ther. 143, 51–60. doi: 10.1021/acs.jcim.8b00924 doi: 10.1016/j.pharmthera.2014.02.004 Malherbe, P., Kratochwil, N., Muhlemann, A., Zenner, M. T., Fischer, C., and Stahl, van Zundert, G. C. P., Rodrigues, J., Trellet, M., Schmitz, C., Kastritis, P. L., and M. (2006). Comparison of the binding pockets of two chemically unrelated Karaca, E. (2016). The HADDOCK2.2 web server: user-friendly integrative allosteric antagonists of the mGlu5 receptor and identification of crucial modeling of biomolecular complexes. J. Mol. Biol. 428, 720–725. doi: 10.1016/j. residues involved in the inverse agonism of MPEP. J. Neurochem. 98, 601–615. jmb.2015.09.014 doi: 10.1111/j.1471-4159.2006.03886.x Weis, W. I., and Kobilka, B. K. (2018). The molecular basis of G protein-coupled Malherbe, P., Kratochwil, N., Zenner, M. T., Piussi, J., Diener, C., Kratzeisen, receptor activation. Annu. Rev. Biochem. 87, 897–919. doi: 10.1146/annurev- C., et al. (2003). Mutational analysis and molecular modeling of the binding biochem-060614-033910 pocket of the metabotropic glutamate 5 receptor negative modulator 2-methyl- Wood, M. R., Hopkins, C. R., Brogan, J. T., Conn, P. J., and Lindsley, C. W. 6-(phenylethynyl)-pyridine. Mol. Pharmacol. 64, 823–832. doi: 10.1124/mol. (2011). “Molecular switches” on mGluR allosteric ligands that modulate modes 64.4.823 of pharmacology. Biochemistry 50, 2403–2410. doi: 10.1021/bi200129s Nicoletti, F., Bruno, V., Ngomba, R. T., Gradini, R., and Battaglia, G. (2015). Zhou, B., and Giraldo, J. (2018). An operational model for GPCR homodimers Metabotropic glutamate receptors as drug targets: what’s new? Curr. Opin. and its application in the analysis of biased signaling. Drug Discov. Today 23, Pharmacol. 20, 89–94. doi: 10.1016/j.coph.2014.12.002 1591–1595. doi: 10.1016/j.drudis.2018.04.004 Nishimura, A., Kitano, K., Takasaki, J., Taniguchi, M., Mizuno, N., and Tago, K. (2010). Structural basis for the specific inhibition of heterotrimeric Gq Conflict of Interest: The authors declare that the research was conducted in the protein by a small molecule. Proc. Natl. Acad. Sci. U.S.A. 107, 13666–13671. absence of any commercial or financial relationships that could be construed as a doi: 10.1073/pnas.1003553107 potential conflict of interest. Noetzel, M. J., Rook, J. M., Vinson, P. N., Cho, H. P., Days, E., and Zhou, Y. (2012). Functional impact of allosteric agonist activity of selective positive allosteric Copyright © 2020 Lans, Díaz, Dalton and Giraldo. This is an open-access article modulators of metabotropic glutamate receptor subtype 5 in regulating central distributed under the terms of the Creative Commons Attribution License (CC BY). nervous system function. Mol. Pharmacol. 81, 120–133. doi: 10.1124/mol.111. The use, distribution or reproduction in other forums is permitted, provided the 075184 original author(s) and the copyright owner(s) are credited and that the original Oostenbrink, C., Villa, A., Mark, A. E., and van Gunsteren, W. F. (2004). A publication in this journal is cited, in accordance with accepted academic practice. No biomolecular force field based on the free enthalpy of hydration and solvation: use, distribution or reproduction is permitted which does not comply with these terms. Frontiers in Molecular Biosciences | www.frontiersin.org 11 March 2020 | Volume 7 | Article 38