Effectiveness of abdominal radiographs in visualizing chewable iron supplements following overdose

Theory

Molecular Switches of Allosteric Modulation of the Metabotropic Glutamate 2 Receptor Graphical Abstract

Authors Laura Pe´rez-Benito, Maarten L.J. Doornbos, Arnau Cordomı´, Luc Peeters, Hilde Lavreysen, Leonardo Pardo, Gary Tresadern

Correspondence [email protected] (L.P.), [email protected] (G.T.)

In Brief Allosteric modulators of class C metabotropic glutamate GPCRs modulate functional activity by interaction with similar amino acids as orthosteric ligands for class A GPCRs. Pere´z-Benito et al. use combined experimental and computational approaches to define the functional mechanism of action of negative and positive allosteric modulators of mGlu receptors.

Highlights d

Experimental and computational study of mGlu receptor allosteric modulation

d

Elucidated the binding mode of mGlu2 receptor allosteric modulators

d

Allosteric functional effects act via trigger and transmission switch amino acids

d

These amino acids are analogous to those important for class A GPCR activation

Pe´rez-Benito et al., 2017, Structure 25, 1–10 July 5, 2017 ª 2017 Elsevier Ltd. http://dx.doi.org/10.1016/j.str.2017.05.021

Please cite this article in press as: Pe´rez-Benito et al., Molecular Switches of Allosteric Modulation of the Metabotropic Glutamate 2 Receptor, Structure (2017), http://dx.doi.org/10.1016/j.str.2017.05.021

Structure

Theory Molecular Switches of Allosteric Modulation of the Metabotropic Glutamate 2 Receptor Laura Pe´rez-Benito,1,2 Maarten L.J. Doornbos,3,4 Arnau Cordomı´,1 Luc Peeters,3 Hilde Lavreysen,3 Leonardo Pardo,1,* and Gary Tresadern2,3,5,* 1Laboratori de Medicina Computacional, Unitat de Bioestadistica, Facultat de Medicina, Universitat Autonoma de Barcelona, 08193 Bellaterra, Spain 2Computational Chemistry, Janssen Research and Development, Calle Jarama 75A, 45007 Toledo, Spain 3Janssen Research and Development, Turnhoutseweg 30, 2340 Beerse, Belgium 4Division of Medicinal Chemistry, Leiden Academic Centre for Drug Research (LACDR), Leiden University, P.O. Box 9502, 2300 RA Leiden, the Netherlands 5Lead Contact *Correspondence: [email protected] (L.P.), [email protected] (G.T.) http://dx.doi.org/10.1016/j.str.2017.05.021

SUMMARY

Metabotropic glutamate (mGlu) receptors are class C G protein-coupled receptors (GPCRs) crucial for CNS function and important drug discovery targets. Glutamate triggers receptor activation from an extracellular domain binding site while allosteric modulators bind in the seven-transmembrane domain. Little is known about how allosteric modulators produce their functional effects at the molecular level. Here we address this topic with combined experimental and computational approaches and reveal that mGlu receptor allosteric modulators interact with the homologous ‘‘trigger switch’’ and ‘‘transmission switch’’ amino acids as seen in class A GPCRs, in short, the characteristic hallmarks of class A agonist activation translate to the mGlu allosteric modulator. The proposed ‘‘trigger switch’’ for the mGlu2 involves the side chains of F6433.36a.40c, N7355.47a.47c, and W7736.48a.50c, whereas the ‘‘transmission switch’’ involves the Y6473.40a.44c, L7385.50a.50c, and T7696.44a.46c amino acids. The work has wide impact on understanding mGlu GPCR function and for future allosteric modulator drugs.

INTRODUCTION G protein-coupled receptors (GPCRs) are crucial for cell signaling and are the target for a large proportion of drugs that act primarily as agonists or inverse agonists at the orthosteric site. However, allosteric modulation of GPCRs using small molecules that bind to distinct sites is of high interest, particularly in the field of metabotropic glutamate (mGlu) receptors. The mGlu family consists of eight class C GPCRs (Niswender and Conn, 2010) and the mGlu2 receptor is of interest as it negatively modulates glutamate release into synapses. Hence, mGlu2 receptor activation can dampen glutamate hyperfunction in disorders such as anxiety and schizophrenia (Ghose et al., 2009; Gu et al., 2008), whereas

blocking of mGlu2 receptors can be beneficial for glutamate hypofunction in depression (Celanire et al., 2015). Activation of mGlu2 receptors with glutamate analogs (Fell et al., 2012) presents challenges for selectivity, brain penetration, and chemical novelty. These difficulties resulted in a shift toward allosteric modulators that bind at less-conserved allosteric sites (Conn et al., 2009), increasing (positive allosteric modulator [PAM]) or decreasing (negative allosteric modulation [NAM]) the affinity and/or efficacy of glutamate. Many PAMs are supported with preclinical animal testing (Cid et al., 2012; Fell et al., 2011; Johnson et al., 2013; Lavreysen et al., 2015; Trabanco et al., 2011; Tresadern et al., 2010) while two have advanced into clinical trials, AZD8529 (ClinicalTrials.gov, 2009) and JNJ-40411813 (Cid et al., 2014) (also known as ADX71149). Reference molecules in the field include BINA (1) (Galici et al., 2006), JNJ40068782 (2) (Lavreysen et al., 2013), and JNJ-46281222 (3) (Doornbos et al., 2016). Inhibition of mGlu2 receptors also began with orthosteric ligands such as LY351495 from Eli Lilly (Nicholls et al., 2006), and shifted to allosteric antagonists. Reported NAM reference molecules include Ro-676221 (4) (Kew and Kemp, 2005), Ro-4491533 (5) (Campo et al., 2011), and Ro-4995819/Decoglurant (6) characterized in vivo (Goeldner et al., 2013). Molecule 6 has advanced into clinical trials (ClinicalTrials.gov, 2011) (Figure 1). Mutagenesis has demonstrated that mGlu2 allosteric modulators bind in an evolutionary conserved site in the seven-transmembrane (7-TM) domain similar to orthosteric ligands in class A (Farinha et al., 2015; Lundstro¨m et al., 2011). The recent class C crystal structures of mGlu1 and mGlu5 receptors have confirmed the 7-TM allosteric binding site (Dore et al., 2014; Wu et al., 2014). Despite vast interest in the field, the interplay between the allosteric modulator and receptor activation observed experimentally remains unclear (Pin and Bettler, 2016). Given the attraction of allosteric GPCR modulation and especially mGlu receptors, we have addressed this question with an extensive experimental and computational study. Experimental functional activity, binding data, selectivity, and site-directed mutagenesis have been combined with computational structure-activity relationships (SARs), docking and molecular dynamics (MD) simulations. We define how mGlu2 allosteric modulators bind and exert Structure 25, 1–10, July 5, 2017 ª 2017 Elsevier Ltd. 1

Please cite this article in press as: Pe´rez-Benito et al., Molecular Switches of Allosteric Modulation of the Metabotropic Glutamate 2 Receptor, Structure (2017), http://dx.doi.org/10.1016/j.str.2017.05.021

Figure 1. Selected mGlu2 Receptor PAMs 1–3 and NAMs 4–6

their function. We find remarkable similarities between class A and mGlu class C GPCRs. This work provides crucial insight for research and drug discovery in this area as well as furthering our understanding of GPCR allosteric modulation. RESULTS AND DISCUSSION Molecules 1 to 6 Are Functional Allosteric Modulators with an Overlapping Binding Site Molecules were tested in functional assays to assess PAM or NAM activity. In brief, a variable concentration of test compound and 10 mg membrane protein were pre-incubated with buffer (to detect agonist effects) or an EC20 or EC80 equivalent concentration of glutamate (to detect PAM or NAM effects respectively). After 30 min of incubation at 30 C, 0.1 nM [35S]GTPgS was added and after another 30 min incubation, rapid filtration stopped the reaction and filter-bound radioactivity levels of [35S]GTPgS were measured to infer functional activity. In addition, a binding displacement assay was performed with a tritiated mGlu2 receptor PAM, [3H]JNJ-46281222. Molecules 1, 2, and 3 showed PAM activity with pEC50 values of 7.03, 6.90, and 8.09, respectively (Table 1) and had no NAM effect up to 10 or 30 mM in the NAM assay. Correspondingly, molecules 4, 5, and 6 were mGlu2 receptor NAMs with pIC50’s of 8.29, 8.57, and 8.60, respectively, with no PAM effect in the PAM assay. All PAMs and NAMs displaced the tritiated PAM [3H]JNJ-46281222, displaying pKi values ranging from 7.22 to 8.33 (Hill slopes of 1, suggesting a single population of binding sites). Hence, 1–6 augment or inhibit glutamate response at mGlu2 receptors by binding to a common allosteric site. Allosteric binding sites of PAMs and NAMs were located with mutagenesis experiments. Details confirming comparable expression and activity for wild-type (WT) and mutants receptors are provided in Table S1. All mutated receptors showed proper receptor expression in immunoblotting. Mutants also showed comparable [35S]GTPgS functional activity compared with WT. Furthermore, radioligand binding studies using the orthosteric mGlu2/3 receptor antagonist [3H]LY341495 showed similar specific binding and Bmax determinations for WT and mutant receptors confirming the orthosteric site was intact. The mutagenesis experiments revealed that R6353.28a.32cA, L6393.32a.36cA, 2 Structure 25, 1–10, July 5, 2017

F6433.36a.40cA, N7355.47a.47cD, and 6.48a.50c A decreased the potency of W773 PAMs 1–3 (Table 2; Figure S1). The H7235.34a.35cV mutation affected activity of 2 while F7766.51a.53cA affected 1 and 2. Meanwhile, F6433.36a.40cA, L7325.43a.44cA, W7736.48a.50cA, F7766.51a.53cA, and F7806.55a.57cA mutations reduced the activity of NAMs 4, 5, and 6 (Table 3; Figure S2). Mutating L6393.32a.36cA only reduced the activity of 4 and 6, whereas D7255.36a.37cA, and I6934.56a.46cM, 7.42a.36c A only affected 6. Overall, V798 L6393.32a.36c, F6433.36a.40c, W7736.48a.50c, and F7766.51a.53c were important for the action of both PAMs and NAMs, suggesting common interactions in the shared allosteric site. Importantly, N7355.47a.47c affects only PAMs and F7806.55a.57cA only NAMs. PAM and NAM SARs Pharmacophore models that discriminate active from inactive mGlu2 PAMs and NAMs were computed (see STAR Methods). The PAM pharmacophore had a clear 3D overlap with key shared features (see Figure 2A). (1) A central aromatic feature (AR1) orients the molecules via the p electrons through aromatic-aromatic or aromatic-hydrophobic interactions. (2) A key hydrogen bond acceptor (HBA) feature (the carbonyl group of the indanone or pyridone in 1 and 2, or the sp2 nitrogen in the triazolopyridine of 3). (3) A hydrophobic feature (HYD1) at one end of the molecule, and two hydrophobic features (HYD2-HYD3) located at the other end permitting various substituents. (4) Active compounds contain hydrophobic groups (methyl, cyano, trifluoromethyl, or chloro) on the AR scaffold. Thus, an additional occupancy feature (OCC) was added to the pharmacophore model. (5) In contrast, increasing the size of the HYD1 feature was detrimental for activity, thus, an excluded feature (EX) was added in this region. This pharmacophore is consistent with previous hypotheses and has been used to identify new mGlu2 PAMs (Tresadern et al., 2010, 2014). The NAM pharmacophore contained benzodiazapinones (4 and 5) and pyrazolopyrimidines (6) that could be overlaid to share common pharmacophoric features (Figure 2B). (1) An aromatic feature (AR1). (2) A HBA feature (HBA1 and HBA2) matched by the carbonyl of the benzodiazapinone scaffold or by the amide carbonyl. While this HBA is satisfied by all benzodiazapinones, the pyrazolopyrimidines do not always contain an amide carbonyl in this region (acetylenic spacers can also be active). (3) Two hydrophobic features (HYD1–HYD2) located at one end of the molecule. In Supplemental Information the same SAR and pharmacophore analysis is shown to recapitulate the interactions for binding of mGlu1 NAMs in the known crystal structure. Binding Mode of PAMs and NAMs at mGlu2 Receptors To understand how PAMs and NAMs interact with the mGlu2 receptor, 1–3 were docked in the active-like mGlu2 receptor model

Please cite this article in press as: Pe´rez-Benito et al., Molecular Switches of Allosteric Modulation of the Metabotropic Glutamate 2 Receptor, Structure (2017), http://dx.doi.org/10.1016/j.str.2017.05.021

Table 1. mGlu2 Receptor Activity and Affinity of Positive and Negative Allosteric Modulators Determined by [35S]GTPgS and [3H]JNJ-46281222 Binding Experiments Using Stably Expressing hmGlu2-CHO Cells Compound

PAM pEC50a

NAM pIC50b

Binding pKic

Table 2. Effect of mGlu2 Receptor Mutations on Activity of PAMs as Determined by [35S]GTPgS Binding Assay in the Presence of an EC20 Glutamate Concentration, 4 mM PAM Compound 1, BINA

2, JNJ-40068782

3, JNJ-46281222

1, BINA

PAM

7.03 ± 0.14

–

7.22 ± 0.26

Mutant

pEC50

pEC50

pEC50

2, JNJ-40068782

PAM

6.90 ± 0.10

–

7.58 ± 0.16

Transient WT

7.11 ± 0.30

7.08 ± 0.13

8.22 ± 0.23

3, JNJ-46281222

PAM

8.09 ± 0.23

–

8.33 ± 0.34

Stable WT

7.03 ± 0.14

6.88 ± 0.13

8.09 ± 0.23

4, Ro-676221

NAM

–

8.29 ± 0.21

7.96 ± 0.13

R6353.28a.32cA

6.21a***

6.42 ± 0.03***

n.c.

5, Ro-4491533

NAM

–

8.57 ± 0.22

8.09 ± 0.16

R6363.29a.33cA

n.c.

n.c.

n.c.

6, Ro-4995819

NAM

–

8.60 ± 0.07

7.56 ± 0.07

L6393.32a.36cA

5.85; 5.87b***

6.27 ± 0.16***

n.c.

5.91; 5.79b***

5.64 ± 0.18***

6.50; 6.14b***

a

Functional activity of mGlu2 receptor PAMs determined by the enhancement of glutamate (EC20) induced [35S]GTPgS binding. b Functional activity of mGlu2 NAMs determined by the reduction of glutamate (EC80) induced [35S]GTPgS binding. c Affinity for the allosteric binding pocket of the mGlu2 receptor as determined by [3H]JNJ-46281222 binding experiments. Data are shown as mean ± SD of at least three individual experiments performed in duplicate. – indicates the PAMs did not show NAM activity in the NAM assay, and NAMs did not show PAM activity in the PAM assay.

and NAMs 4–6 into the ‘‘inactive’’ model and subjected to three independent unrestrained 1 ms MD simulations (see STAR Methods). PAMs 1–3 showed a robust mode of interaction (Figure 3 and Figure S5), consistent with the SAR analysis. The carbonyl group of 1 and 2 or the nitrogen atom of the triazole ring of 3 (HBA in Figure 2A) forms a hydrogen bond with N7355.47a.47c, the indanone ring of 1 or the pyridone ring of 2 or the triazolopyridine ring of 3 (AR1) forms an aromatic interaction with F6433.36a.40c, the cyclopentyl moiety of 1 or cyclopropyl moieties of 2 and 3 (HYD1) expands toward the intracellular side to interact with W7736.48a.50c (also with F7766.51a.53c in 2) without reaching the EX feature, the methyl groups of 1 or cyano group of 2 or the trifluoromethyl of 3 (OCC) enter into a small hydrophobic cavity between TM3 and TM5 delimited by G6403.33a.37c (CAV1) and the biphenyl group of 1 or the 4-phenylpiperidine group of 2 and 3 (HYD2-3) expands toward the extracellular side interacting with L6393.32a.36c (except 3). These binding modes are compatible with the mutagenesis experiments (Figures 3A–3C and Table 2). Moreover, the carboxylic acid of 1 forms an ionic interaction with R6353.38a.32c (Figure 3A) and the phenyl ring of 2 interacts with the H7235.34a.35cR6353.38a.32c pair (Figure 3B). Accordingly, R6353.38a.32cA has a significant effect on 1 and 2, whereas H7235.34a.35cV only in 2 (Table 2) Figure 2A (bottom) summarizes the amino acids in the 7-TM domain of the mGlu2 receptor interacting with the ligands and their pharmacophore features. The binding mode of NAMs 4–6 is shown in Figures 3D–3F, S5, and S9. All three NAMs bind in front of TM6 with the central scaffold interacting with the aromatic W7736.48a.50c, F7766.51a.53c, and F7806.55a.57c amino acids, as confirmed by side-directed mutagenesis (Table 3). In addition, the carbonyl group of 4 and 5 or the nitrogen atom of the pyrazole ring of 6 (HBA1 in Figure 2B) interacts with S7977.41a.35c, mutated in the mGlu5 receptor (Gregory et al., 2014), and the benzodiazapinone group of 4 and 5 or the pyrazolopyrimidine scaffold of 6 (HYD2) forms aromatic interactions with F6433.36a.40c, W7736.48a.50c, and F7766.51a.53c. These binding modes are compatible with

F643

3.36a.40c

A

3.37a.41c

A

6.99 ± 0.15

6.7 ± 0.22

n.c.

S6884.51a.41cL

6.93 ± 0.33

6.90 ± 0.44

7.69

G6894.52a.42cV

6.20 ± 0.37*

6.27 ± 0.20***

7.21 ± 0.30*

I6934.56a.46cM

n.c.

n.c.

n.c.

V7004.63a.53cL

n.c.

n.c.

n.c.

H7235.34a.35cV

n.c.

6.26 ± 0.06***

n.c.

D7255.36a.37cA

n.c.

n.c.

n.c.

S644

M7285.39a.40cA

n.c.

n.c.

n.c.

S7315.42a.43cA

n.c.

n.c.

n.c.

L7325.43a.44cA

7.80 ± 0.03

7.03 ± 0.06

7.31 ± 0.33

N7355.47a.47cD

5.13a***

5.52a***

6.51; 6.83b***

V7365.48a.48cA

n.c.

n.c.

n.c.

W7736.48a.50cA

<5.0***

<5.0***

7.11a,b***

F776

6.51a.53c

6.70 ± 0.06

6.38 ± 0.23***

n.c.

F7806.55a.57cA

A

n.c.

n.c.

n.c.

V7987.42a.36cA

n.t.

n.t.

n.t.

*p < 0.05, **p < 0.01, ***p < 0.001 significantly different from value obtained for transiently transfected WT mGlu2 receptor. Determined using one-way ANOVA with Dunnett’s post-test. pEC50 values were averaged (mean ± SD) for three or more individual experiments performed in triplicate; for n = 2, individual values are provided. a For one or two experiments, pEC50 was <5; b n = 2. n.c., refers to no change in activity compared with WT, performed by testing compound at 2 concentrations (concentrations chosen are equivalent to concentration producing half-maximal or maximal response as determined for these compounds on the WT receptor); n.t., not tested. See also Figure S1.

the mutagenesis experiments (Figures 3D–3F; Table 3). The trifluoromethyl group of 5 (HYD1) enters into a small hydrophobic cavity between TM6 and TM7 (CAV2) (Figure 3E). However, the larger phenylacetylene group of 4 (HYD1) cannot fit into CAV2, entering deeper into the receptor (Figure 3D) as the phenylacetylene group of Mavoglurant in the crystal structure of mGlu5 receptor (Dore et al., 2014). NAM 6 contains two trifluoromethyl groups, which could bind in CAV1 (TM3 and TM5) and CAV2 (TM6 and TM7) in a similar manner as 3 and 5, respectively (see above). Between the two possible orientations, only the mode of binding in which the phenyl-CF3 group binds in CAV1 and the pyrazole-CF3 group (HYD1) in CAV2 (Figure 3F), was unchanged during MD simulations and fulfills the results of the mutagenesis experiments. The phenyl ring of 4 and 5 and the aminopyridine group of 6 (AR1) forms an aromatic-aromatic interaction with F7806.55a.57c. We also proposed that the Structure 25, 1–10, July 5, 2017 3

Please cite this article in press as: Pe´rez-Benito et al., Molecular Switches of Allosteric Modulation of the Metabotropic Glutamate 2 Receptor, Structure (2017), http://dx.doi.org/10.1016/j.str.2017.05.021

Table 3. Effect of mGlu2 Mutations on Activity of NAMs as Determined by [35S]GTPgS Binding Assay in the Presence of an EC80 Glutamate Concentration of 60 mM NAM Compound 4, Ro-676221

5, Ro-4491533

6, Ro-4995819

Mutant

pIC50

pIC50

pIC50

Transient WT

8.18 ± 0.28

8.69 ± 0.18

9.00 ± 0.21

Stable WT

8.29 ± 0.21

8.57 ± 0.22

8.60 ± 0.07

R6353.28a.32cA

8.14; 8.01a

9.26; 8.55a

n.t.

R6363.29a.33cA

8.29 ± 0.24

8.75 ± 0.28

8.68 ± 0.11

L6393.32a.36cA

7.79 ± 0.28**

8.40 ± 0.25

8.41 ± 0.09***

F6433.36a.40cA

7.39 ± 0.21***

7.50 ± 0.20***

7.17 ± 0.07***

S6443.37a.41cA

8.32 ± 0.35

8.77 ± 0.37

n.t.

S6884.51a.41cL

8.17 ± 0.12

8.43 ± 0.35

n.t.

I6934.56a.46cM

8.06 ± 0.09

8.67 ± 0.03

8.47 ± 0.11***

V7004.63a.53cL

8.33; 8.19a

8.65 ± 0.23

n.t.

H7235.34a.35cV

8.30 ± 0.29

8.61 ± 0.30

8.62 ± 0.08

D7255.36a.37cA

8.15 ± 0.16

8.38 ± 0.10

8.07 ± 0.14***

M7285.39a.40cA

8.55 ± 0.07

8.49 ± 0.07

9.23 ± 0.04 9.14 ± 0.08

S731

5.42a.43c

8.55 ± 0.16

8.71 ± 0.23

L7325.43a.44cA

A

7.64 ± 0.28***

7.21 ± 0.15***

8.33 ± 0.09***

N7355.47a.47cD

8.59; 7.83a

9.01; 8.26a

8.70; 8.63a

V7365.48a.48cA

7.98; 8.01a

9.28; 8.95a

n.t.

<5

6.73 ± 0.09***

6.75 ± 0.06***

F7766.51a.53cA

7.46 ± 0.64***

7.35 ± 0.14***

6.47 ± 0.09***

F7806.55a.57cA

7.58 ± 0.21***

7.77 ± 0.12***

7.54 ± 0.02***

V7987.42a.36cA

7.98 ± 0.20

8.45 ± 0.57

8.57 ± 0.26**

6.48a.50c

W773

A

*p < 0.05, **p < 0.01, ***p < 0.001 significantly different from value obtained for transiently transfected WT mGlu2 receptor. Determined using one-way ANOVA with Dunnett’s post-test. pEC50 values were averaged (mean ± SD) for three or more individual experiments performed in triplicate; for n = 2, individual values are provided. a n = 2. n.t., not tested. See also Figures S2 and S3.

2-aminopyridine group of 6 interacts with D7255.36a.37c, which is confirmed experimentally (Table 3). PAMs and NAMs 1–6 were screened in mGlu receptor selectivity assays (Table S2). PAMs 1–3 only showed activity at mGlu2, and no activity versus other mGlu receptors. In contrast, NAMs 4–6 lack selectivity, showing antagonistic activity at both mGlu2 and mGlu3 receptors. This curious lack of selectivity of mGlu2 NAMs versus mGlu3 receptors has been seen previously (Celanire et al., 2015) although without providing a structural explanation. Because the mGlu2 receptor contains N7355.47a.47c, which is Asp at the homologous position of the mGlu3 receptor, and the N7355.47a.47cD mutation affects the activity of PAMs 1–3 but not NAMs 4–6 (Tables 2 and 3), suggests that the proposed hydrogen bond with PAMs (HBA, Figure 2A) is responsible for the observed selectivity versus mGlu3. Mechanisms of Receptor (In)Activation by PAMs and NAMs Families A and C of GPCRs bind the same G proteins, maintain the spatial conservation of the TM helices, and share the 7-TM binding site. Monomeric mGlu2 receptors couple to G proteins 4 Structure 25, 1–10, July 5, 2017

upon activation by a PAM alone (El Moustaine et al., 2012) while PAM binding in only one allosteric site in the mGlu homodimer can achieve maximal potentiation (Goudet et al., 2005). Therefore we hypothesize similarities in receptor (in)activation for classes A and C, as previously suggested for family B (Cordomı´ et al., 2015; Spyridaki et al., 2014). Thus, (in)activation of mGlu2 receptors involves rearrangement of an analogous ‘‘transmission switch’’ in TM3, TM5, and TM6, involving positions 3.40a, 5.50a, and 6.44a, as described for class A (inward movement of TM5 at the highly conserved P5.50a, steric competition between a bulky hydrophobic side chain at position 3.40a, counterclockwise rotation of TM3 when viewed from the extracellular side, reposition of F6.44a, and outward movement of TM6 for receptor activation) (Sansuk et al., 2011; Venkatakrishnan et al., 2013). Moreover, it is known that the initial agonist-induced structural changes of the receptor (‘‘trigger switch’’), responsible for rearrangement of the ‘‘transmission switch’’, are a hydrogen bond interaction between agonists and S5.46a (Rasmussen et al., 2011; Warne et al., 2011), the movement of W6.48a (Xu et al., 2011), or a conformational toggle switch of the side chain of the amino acid at position 3.36a (Pellissier et al., 2009). Thus, these ‘‘switches’’ were studied during the MD simulations of active-like and inactive mGlu2 receptor (Figures 4 and 5). Figure 4A illustrates the position of the side chains forming the ‘‘trigger switch’’ amino acids (F6433.36a.40c, N7355.47a.47c, and W7736.48a.50c) for NAMs. No significant differences in TM5 are observed because NAMs do not interact with N7355.47a.47c, while small differences are observed in TM3 because NAMs form aromatic interactions with F6433.36a.40c (Figure 4A). However, the most significant change is found for W7736.48a.50c in TM6. Notably, the extra methyl group of 5 relative to 6, keeps the conformation of W7736.48a.50c outside the 7-TM bundle, as observed in the crystal structures of the Mavoglurant-mGlu5 receptor (Dore et al., 2014) and FITM-mGlu1 receptor (Wu et al., 2014) complexes. In contrast, 4 and 6 permit a conformational switch of W7736.48a.50c toward the inside of the 7-TM bundle, as also observed in mGlu5 crystal structures (Christopher et al., 2015). Accordingly, 4 and 6 favor the gauche+ conformation (inside the bundle) while 6 favors the trans conformation (outside) of W7736.48a.50c (Figure 4C). Thus, the conformation of W7736.48a.50c depends on the chemical structure of the NAM. MD simulations on mGlu1 and mGlu5 receptor crystal structures confirmed that NAMs do not induce conformational changes in ‘‘trigger’’ or ‘‘transmission’’ switch amino acids (Figures S7 and S8). PAMs form a crucial hydrogen bond with N7355.47a.47c in TM5 that is crucial for allosteric activity, as the N7355.47a.47cD mutation results in a 100-fold loss in potency (Table 2) and lower Emax (Figure S1). This hydrogen bond remains constant throughout the MD simulations (Figure S6) and the side chain of N7355.47a.47c stays in a similar position for PAMs 1–3 (Figure 4B). Moreover, in all MD simulations of PAMs, W7736.48a.50c, which was initially modeled pointing to the membrane, rotates inside the receptor (Figure 4B) thereby reducing the volume of the bottom of the receptor cavity (EX pharmacophoric feature in Figure 2A). The HYD1 pharmacophoric feature of 1–3 interacts with the side chain of W7736.48a.50c and assists its movement (gauche+ conformation, Figure 4D). Figure 4E illustrates the position of the ‘‘transmission switch’’ side chains (Y6473.40a.44c, L7385.50a.50c, I7395.51a.51c,

Please cite this article in press as: Pe´rez-Benito et al., Molecular Switches of Allosteric Modulation of the Metabotropic Glutamate 2 Receptor, Structure (2017), http://dx.doi.org/10.1016/j.str.2017.05.021

Figure 2. Key Interactions for PAMs and NAMs Pharmacophore models of PAMs (A) and NAMs (B) of the mGlu2 receptor. The structural features of the pharmacophore models are shown on the top panels, whereas the predicted amino acids in the mGlu2 receptor, determined by a combination of docking/MD simulations (Figure 3) and mutagenesis experiments (Tables 2 and 3) are shown on the bottom panels. See also Figure S3 for application to mGlu1 receptor.

and T7696.44a.46c) during the MD simulations for NAMs 4–6. Clearly, all NAMs stabilized similar conformations of these amino acids. Most obviously T7696.44a.46c moves so little throughout the simulations that the dihedral angle distributions in Figure 4G are indiscernible. Conversely, for PAMs it flips for each ligand throughout the simulations (Figure 4H). To compare the mechanism of negative and positive allosteric modulation at the mGlu2 receptor, we plot the movement of the

amino acids in the first activation ‘‘trigger switch’’ and the ‘‘transmission switch’’ during the MD simulations in the presence of PAM 1 and NAM 6 (Figure 5). The hydrogen bond between the carbonyl group of 1 and N7355.47a.47c, absent for NAMs, moves N7355.47a.47c toward TM3, whereas F6433.36a.40c that interacts with both PAMs and NAMs, moves toward TM7 in the PAM-bound simulation. This hydrogen bond interaction between PAMs and N7355.47a.47c resembles the interaction between agonists and TM5 in class A, but how does it affect the ‘‘transmission switch?’’ Relative to the inactive simulation, in the active simulation there is an inward movement of TM5 (at L7385.50a.50c) toward TM3, relocation of Y6473.40a.44c toward TM6, and finally reposition of the side chain of T7696.44a.46c. This includes a conformational change of T7696.44a.46c from the gauche+ to the gauche or

Figure 3. The Binding Mode of mGlu2 Allosteric Modulators Detailed view of the binding mode of PAMs 1 (A) green, 2 (B), blue, and 3 (C) yellow and NAMs 4 (D) magenta, 5 (E) orange, and 6 (F) red to the mGlu2 receptor. Amino acids involved in the binding of ligands, as determined by site-directed mutagenesis experiments reported in Tables 2 and 3 are shown in white, whereas S7977.41a.35c that was reported for the binding of NAMs to mGlu5 receptor (Gregory et al., 2014) is shown in green. See also Figure S9.

Structure 25, 1–10, July 5, 2017 5

Please cite this article in press as: Pe´rez-Benito et al., Molecular Switches of Allosteric Modulation of the Metabotropic Glutamate 2 Receptor, Structure (2017), http://dx.doi.org/10.1016/j.str.2017.05.021

Figure 4. Receptor Side-Chain Movements in Response to NAMs and PAMs Distribution of the positions of representative atoms (CZ of F6433.36a.40c, ND2 of N7355.47a.47c, and NE1 of W7736.48a.50c) (A and B) and (OH of Y6473.40a.44c, CD2 of L7385.50a.50c, CG2 of I7395.51a.51c, and OG1 of T7696.44a.46c) (E and F) of the amino acids of mGlu2 receptor at the homologous positions of the ‘‘trigger switch’’ and ‘‘transmission switch’’ of class A, respectively, during MD simulations of the active-like model of mGlu2 receptor in complex with Gi and PAMs 1–3 (B and F) and of the ‘‘inactive’’ model in complex with NAMs 4–6 (A and E). Evenly spaced snapshots extracted from the 1 ms of unbiased MD simulations are depicted. PAMs 1–3 are shown in green, blue and yellow and NAMs 4–6 are shown in magenta, orange, and red, respectively. Timeevolution of the c1 rotamer of W7736.48a.50c and T7696.44a.46c for NAMs 4–6 (C and G) and PAMs 1–3 (D and H) during the MD simulations (line color matches with the bound ligand). See also Figures S7 and S8 for application to mGlu1 and mGlu5 receptors.

trans conformer (Figures 4G and 4H), which was not observed for NAMs (see above). A conformational change of this type has been shown to cause bend in a helices (Ballesteros et al., 2000), which in this case would facilitate the outward movement of TM6 for receptor activation. Experimental Validation of the Mechanisms of Receptor Activation by PAMs To further validate our hypothesis on the role of ‘‘trigger’’ and ‘‘transmission’’ switch amino acids we performed further site-directed mutagenesis. Clearly, the ‘‘trigger switch’’ (F6433.36a.40cA, N7355.47a.47cD, and W7736.48a.50cA) mutations impede receptor activation by PAMs 1–3 (Table 2 and Figure S1). Moreover, these mutations affect the activity of a larger set of PAMs for the mGlu2 receptor (Farinha et al., 2015; Lundstro¨m et al., 2016). Consistent with our results, it was shown by others that the N7355.47a.47cD mutation had no effect on NAM activity but did affect PAM activity (Hemstapat et al., 2007). Further agreement is seen that F6433.36a.40c and W7736.48a.50c mutations 6 Structure 25, 1–10, July 5, 2017

affected the activity of NAMs but N7355.47a.47c mutation did not (Lundstro¨m et al., 2011). However, two reports show no loss of activity for two structurally different PAMs upon N7355.47a.47cD mutation (Schaffhauser et al., 2003; Rowe et al., 2008), possibly due to alternative binding modes and mechanism of action. In contrast to our mGlu2 receptor NAMs, mutation of N7475.47a.47c affects mGlu5 NAM activity (Mølck et al., 2012; Gregory et al., 2014), whereas mutation of W7986.48a.50c does not affect mGlu1 NAM activity Fukuda et al., 2009. However, neither mutation of these three amino acids forming the ‘‘trigger switch’’ nor any of the 40 mutations reported in Table S1 (many involved in PAM activity) influenced glutamate receptor activity. In contrast, as predicted by the simulations, mutations of Y6473.40a.44cA and T7696.44a.46cA in the ‘‘transmission switch’’ prevent glutamate-induced receptor activation (Figure 6). This indicates that the ‘‘transmission switch’’ is the link between the extracellular (where glutamate binds) and intracellular (where Gi binds) environments. Interestingly, T7696.44a.46cS marginally impairs receptor activation suggesting the methyl group of the b-branched side chain of Thr plays a role in receptor activation. T7696.44a.46c is four amino acids below W7736.48a.50c, which are conserved among mGlu receptors. The Y6473.40a.44cV and T7696.44a.46cS/V mutations in the mGlu2 receptor, performed by others, show similar reduction in orthosteric activity (Lundstro¨m et al., 2011, 2016), while mutation of positions 3.44c and 6.46c in the mGlu5 receptor dramatically reduce (Malherbe et al., 2006; Gregory et al., 2014) and also invert (Turlington et al., 2013) allosteric ligand activity.

Please cite this article in press as: Pe´rez-Benito et al., Molecular Switches of Allosteric Modulation of the Metabotropic Glutamate 2 Receptor, Structure (2017), http://dx.doi.org/10.1016/j.str.2017.05.021

Figure 5. Comparison of the Mechanism of Positive and Negative Allosteric Modulation at the mGlu2 Receptor Relative position of the ‘‘trigger switch’’ amino acids involved in the initial agonist-induced structural changes on the receptor responsible for the rearrangement of the ‘‘transmission switch’’ amino acids that finally lead to receptor activation during MD simulations of PAM 1 (in green) and NAM 6 (in red) (see legend of Figure 4 for details of the MD simulations and the atoms depicted).

Figure 5 also shows a water molecule involved in a network of interactions with T7696.44a.46c and Y6473.40a.44c in the ‘‘transmission switch.’’ This water molecule enters from bulk solvent in all the simulations despite not being included in the initial system. Via hydrogen bonds with the side chains of T7696.44a.46c and Y6473.40a.44c it helps to stabilize a specific orientation of these amino acids. This water-mediated hydrogen bond is important as the Y6473.40a.44cF mutant showed lower activation by glutamate. The crystal structure of the mGlu5 receptor contains a water molecule in a similar position that was proposed to be a key element (Dore et al., 2014). Conclusions We have compared the binding of PAMs and NAMs and their mechanism of negative and positive allosteric modulation at

Figure 6. Mutation of Transmission Switch Amino Acids and Effect on Functional Glutamate-Induced [35S]GTPgS Binding Curves show the glutamate concentration response Y6473.40a.44cA/F and T7696.44a.46cA/S. Error bars show SDs.

for

mutants

the mGlu2 receptor via a combination of experiment and computational methods. We observed differences between residues important for PAMs and NAMs, such as R6363.29a.33cA, I6934.56a.46cM, F7806.55a.57cA, and V7987.42a.36cA, which affected the activity of NAMs but had no effect on PAMs. On the other hand, mutations L6393.32a.36cA, S6443.37a.41cA, S6884.51a.41cL/ G6894.52a.42cV, and N7355.47a.47cD, did not affect the NAMs but had a pronounced effect on the PAMs. Thus, our results show that analogous to class A, in which agonist and antagonist binding overlaps in the orthosteric site but form different sets of interactions (Rasmussen et al., 2011), the allosteric binding site of NAMs and PAMs at mGlu2 receptors also overlaps but with significant differences (Figure S9). While the binding site of PAMs is close to TM3–5, NAMs bind in front of, and parallel to, TM6. In addition, NAMs expand deeper toward the intracellular side than PAMs. As recently discussed (Harpsøe et al., 2016) based on analysis of all published class C GPCR mutants in the TM domain, this allosteric binding site is most probably conserved in the class C family. The mechanisms of receptor inactivation by NAMs and receptor activation by PAMs were studied by unbiased MD simulations in the microsecond timescale to explore conformational changes at the receptor. Our study proposes that either positive (PAMs) or negative (NAMs) allosteric modulation involves rearrangement of homologous ‘‘switches’’ as (in)activation of class A by either orthosteric antagonists or agonists. We have identified an activation ‘‘trigger switch’’ that is rearranged by PAM binding and a ‘‘transmission switch’’ that is not directly involved in ligand interactions but links the binding site with the outward movement of TM6 for receptor activation and G protein binding. The combined experimental and computational results strongly support that despite the low degree of sequence similarity between classes A and C of GPCRs, the two families likely share conserved elements in their mechanisms of receptor activation within the 7-TM domain. Overall, our hypothesis agrees with the large body of experimental mutagenesis data that is available for mGlu receptors. Our work goes beyond previous reports of mutagenesis to provide a dynamic picture of how allosteric modulators elicit their effects. While we may expect subtle differences in mGlu receptor allosteric mechanism of action as reflected in the different binding modes in the mGlu1 and mGlu5 crystal structures, the allosteric modulators act upon an activation pathway passing through the trigger and transmission switch amino acids that is consistent with class A receptors. Structure 25, 1–10, July 5, 2017 7

Please cite this article in press as: Pe´rez-Benito et al., Molecular Switches of Allosteric Modulation of the Metabotropic Glutamate 2 Receptor, Structure (2017), http://dx.doi.org/10.1016/j.str.2017.05.021

STAR+METHODS Detailed methods are provided in the online version of this paper and include the following: d d d

d

d

KEY RESOURCES TABLES CONTACT FOR REAGENT AND RESOURCE SHARING EXPERIMENTAL MODEL AND SUBJECT DETAILS B Datasets for Pharmacophore Building B PAM and NAM Pharmacophore Models B Pharmacophore for mGlu1 B Computational Models of the mGlu2 Receptor in Complex with PAMs 1-3 and NAMs 4-6 B MD Simulations Performed on mGlu1 and mGlu5 Crystal Structures METHOD DETAILS B Plasmids, Cell Transfection and Cell Culture B Membrane Preparation 3 B [ H]LY341495 Binding Assay 35 B [ S]GTPgS Binding Assay 3 B [ H]JNJ-46281222 Binding Assay B Selectivity Assays B Data Analysis QUANTIFICATION AND STATISTICAL ANALYSIS

SUPPLEMENTAL INFORMATION Supplemental Information includes nine figures and two tables and can be found with this article online at http://dx.doi.org/10.1016/j.str.2017.05.021. AUTHOR CONTRIBUTIONS Conceptualization Ideas, L.P.B., H.L., L. Pardo, and G.T.; Methodology, L.P.B., M.L.J.D., L. Peeters, H.L., and G.T.; Validation, L.P.B., M.L.J.D., L. Peeters, H.L., and G.T.; Formal Analysis, L.P.B., M.L.J.D., and L. Peeters; Investigation, L.P.B., M.L.J.D., L. Peeters, and G.T.; Writing – Original Draft, L.P.B., L. Pardo, and G.T.; Writing – Review & Editing, L.P.B., M.L.J.D., A.C., H.L., L. Pardo, and G.T.; Supervision, A.C., H.L., L. Pardo, and G.T. ACKNOWLEDGMENTS We thank members of the wider Janssen mGlu2 receptor PAM and NAM drug discovery teams. This study was supported by a grant from MINECO (SAF2016-77830-R). The authors thankfully acknowledge the computer resources of RES (BCV-2016-1-0010). L. Peeters, H.L., and G.T. are current employees, and L.P.B. and M.L.J.D. are former employees of Janssen Pharmaceutical Companies of Johnson & Johnson. This work was supported in part by a grant from Janssen.

Bento, A.P., Gaulton, A., Hersey, A., Bellis, L.J., Chambers, J., Davies, M., €ger, F.A., Light, Y., Mak, L., McGlinchey, S., et al. (2014). The ChEMBL Kru bioactivity database: an update. Nucleic Acids Res. 42, D1083–D1090. €hnig, F. (1997). Molecular dynamics simulations Berger, O., Edholm, O., and Ja of a fluid bilayer of dipalmitoylphosphatidylcholine at full hydration, constant pressure, and constant temperature. Biophys. J. 72, 2002–2013. Campo, B., Kalinichev, M., Lambeng, N., Yacoubi, M.E., Royer-Urios, I., Schneider, M., Legrand, C., Parron, D., Girard, F., Bessif, A., et al. (2011). Characterization of an mGluR2/3 negative allosteric modulator in rodent models of depression. J. Neurogenet. 25, 152–166. Celanire, S., Sebhat, I., Wichmann, J., Mayer, S., Schann, S., and Gatti, S. (2015). Novel metabotropic glutamate receptor 2/3 antagonists and their therapeutic applications: a patent review (2005 – present). Expert Opin. Ther. Pat. 25, 69–90. Chemical Computing Group. (2014). MOE v2014.9 (Chemical Computing Group). Christopher, J.A., Aves, S.J., Bennett, K.A., Dore, A.S., Errey, J.C., Jazayeri, A., Marshall, F.H., Okrasa, K., Serrano-Vega, M.J., Tehan, B.G., et al. (2015). Fragment and structure-based drug discovery for a class C GPCR: discovery of the mGlu5 negative allosteric modulator HTL14242 (3-chloro-5-[6-(5-fluoropyridin-2-yl)pyrimidin-4-yl]benzonitrile). J. Med. Chem. 58, 6653–6664. Cid, J.M., Tresadern, G., Vega, J.A., de Lucas, A.I., Matesanz, E., Iturrino, L., Linares, M.L., Garcia, A., Andre´s, J.I., Macdonald, G.J., et al. (2012). Discovery of 3-cyclopropylmethyl-7-(4-phenylpiperidin-1-yl)-8-trifluoromethyl[1,2,4]triazolo[4,3-a]pyridine (JNJ-42153605): a positive allosteric modulator of the metabotropic glutamate 2 receptor. J. Med. Chem. 55, 8770–8789. €tjens, R., Finn, T., Rocher, J.-P., Poli, S., Cid, J.M., Tresadern, G., Duvey, G., Lu Vega, J.A., de Lucas, A.I., Matesanz, E., et al. (2014). Discovery of 1-butyl-3chloro-4-(4-phenyl-1-piperidinyl)-(1H)-pyridone (JNJ-40411813): a novel positive allosteric modulator of the metabotropic glutamate 2 receptor. J. Med. Chem. 57, 6495–6512. ClinicalTrials.gov. (2009). Study to Assess the Efficacy, Safety, and Tolerability of AZD8529 in Adult Schizophrenia Patients (ClinicalTrials.gov). ClinicalTrials.gov. (2011). A Study of RO4995819 in Patients with Major Depressive Disorder and Inadequate Response to Ongoing Antidepressant Treatment (ClinicalTrials.gov). 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. Cordomı´, A., Caltabiano, G., and Pardo, L. (2012). Membrane protein simulations using AMBER force field and Berger lipid parameters. J. Chem. Theor. Comput. 8, 948–958. Cordomı´, A., Ismail, S., Matsoukas, M.T., Escrieut, C., Gherardi, M.J., Pardo, L., and Fourmy, D. (2015). Functional elements of the gastric inhibitory polypeptide receptor: comparison between secretin- and rhodopsin-like G protein-coupled receptors. Biochem. Pharmacol. 96, 237–246. Doornbos, M.L.J., Perez-Benito, L., Tresadern, G., Mulder-Krieger, T., Biesmans, I., Trabanco, A.A., Cid, J.M., Lavreysen, H., IJzerman, A.P., and Heitman, L.H. (2016). Molecular mechanism of positive allosteric modulation of the metabotropic glutamate receptor 2 by JNJ-46281222. Br. J. Pharmacol. 173, 588–600.

Received: February 28, 2017 Revised: April 21, 2017 Accepted: May 24, 2017 Published: June 22, 2017

Dore, A.S., Okrasa, K., Patel, J.C., Serrano-Vega, M., Bennett, K., Cooke, R.M., Errey, J.C., Jazayeri, A., Khan, S., Tehan, B., et al. (2014). Structure of class C GPCR metabotropic glutamate receptor 5 transmembrane domain. Nature 511, 557–562.

REFERENCES

El Moustaine, D., Granier, S., Doumazane, E., Scholler, P., Rahmeh, R., Bron, P., Mouillac, B., Bane`res, J.L., Rondard, P., and Pin, J.P. (2012). Distinct roles of metabotropic glutamate receptor dimerization in agonist activation and G-protein coupling. Proc. Natl. Acad. Sci. USA 109, 16342–16347.

Abraham, M.J., Murtola, T., Schulz, R., Pa´ll, S., Smith, J.C., Hess, B., and Lindahl, E. (2015). GROMACS: high performance molecular simulations through multi-level parallelism from laptops to supercomputers. SoftwareX 1–2, 19–25. Ballesteros, J.A., Deupi, X., Olivella, M., Haaksma, E.E.J., and Pardo, L. (2000). Serine and threonine residues bend a-helices in the c1 = g conformation. Biophys. J. 79, 2754–2760.

8 Structure 25, 1–10, July 5, 2017

Farinha, A., Lavreysen, H., Peeters, L., Russo, B., Masure, S., Trabanco, A.A., Cid, J., and Tresadern, G. (2015). Molecular determinants of positive allosteric modulation of the human metabotropic glutamate receptor 2. Br. J. Pharmacol. 172, 2383–2396. Fell, M.J., Witkin, J.M., Falcone, J.F., Katner, J.S., Perry, K.W., Hart, J., Rorick-Kehn, L., Overshiner, C.D., Rasmussen, K., Chaney, S.F., et al.

Please cite this article in press as: Pe´rez-Benito et al., Molecular Switches of Allosteric Modulation of the Metabotropic Glutamate 2 Receptor, Structure (2017), http://dx.doi.org/10.1016/j.str.2017.05.021

(2011). N-(4-((2-(trifluoromethyl)-3-hydroxy-4-(isobutyryl)phenoxy)methyl) benzyl)-1-methyl-1H-imidazole-4-carboxamide (THIIC), a novel metabotropic glutamate 2 potentiator with potential anxiolytic/antidepressant properties: in vivo profiling suggests a link between behavioral and central nervous system neurochemical changes. J. Pharmacol. Exp. Ther. 336, 165–177. Fell, M.J., McKinzie, D.L., Monn, J.A., and Svensson, K.A. (2012). Group II metabotropic glutamate receptor agonists and positive allosteric modulators as novel treatments for schizophrenia. Neuropharmacology 62, 1473–1483. Fukuda, J., Suzuki, G., Kimura, T., Nagatomi, Y., Ito, S., Kawamoto, H., Ozaki, S., and Ohta, H. (2009). Identification of a novel transmembrane domain involved in the negative modulation of mGluR1 using a newly discovered allosteric mGluR1 antagonist, 3-cyclohexyl-5-fluoro-6-methyl-7-(2-morpholin-4ylethoxy)-4H-chromen-4-one. Neuropharmacology 57, 438–445. Galici, R., Jones, C.K., Hemstapat, K., Nong, Y., Echemendia, N.G., Williams, L.C., de Paulis, T., and Conn, P.J. (2006). Biphenyl-indanone A, a positive allosteric modulator of the metabotropic glutamate receptor subtype 2, has antipsychotic- and anxiolytic-like effects in mice. J. Pharmacol. Exp. Ther. 318, 173–185. Ghose, S., Gleason, K.A., Potts, B.W., Lewis-Amezcua, K., and Tamminga, C.A. (2009). Differential expression of metabotropic glutamate receptor 2 and 3 in schizophrenia: a mechanism for antipsychotic drug action? Am. J. Psychiatry 166, 812–820. Goeldner, C., Ballard, T.M., Knoflach, F., Wichmann, J., Gatti, S., and Umbricht, D. (2013). Cognitive impairment in major depression and the mGlu2 receptor as a therapeutic target. Neuropharmacology 64, 337–346. Goudet, C., Kniazeff, J., Hlavackova, V., Malhaire, F., Maurel, D., Acher, F., Blahos, J., Pre´zeau, L., and Pin, J.P. (2005). Asymmetric functioning of dimeric metabotropic glutamate receptors disclosed by positive allosteric modulators. J. Biol. Chem. 280, 24380–24385. Gregory, K.J., Nguyen, E.D., Malosh, C., Mendenhall, J.L., Zic, J.Z., Bates, B.S., Noetzel, M.J., Squire, E.F., Turner, E.M., Rook, J.M., et al. (2014). Identification of specific ligand–receptor interactions that govern binding and cooperativity of diverse modulators to a common metabotropic glutamate receptor 5 allosteric site. ACS Chem. Neurosci. 5, 282–295. Gu, G., Lorrain, D.S., Wei, H., Cole, R.L., Zhang, X., Daggett, L.P., Schaffhauser, H.J., Bristow, L.J., and Lechner, S.M. (2008). Distribution of metabotropic glutamate 2 and 3 receptors in the rat forebrain: implication in emotional responses and central disinhibition. Brain Res. 1197, 47–62. Harpsøe, K., Isberg, V., Tehan, B.G., Weiss, D., Arsova, A., Marshall, F.H., €uner-Osborne, H., and Gloriam, D.E. (2015). Selective negative allosteric Bra modulation of metabotropic glutamate receptors – a structural perspective of ligands and mutants. Sci. Rep. 5, 13869. €uner-Osborne, H., and Gloriam, Harpsøe, K., Boesgaard, M.W., Munk, C., Bra D.E. (2016). Structural insight to mutation effects uncover a common allosteric site in class C GPCRs. Bioinformatics 33, 1116–1120. Hemstapat, K., Da Costa, H., Nong, Y., Brady, A.E., Luo, Q., Niswender, C.M., Tamagnan, G.D., and Conn, P.J. (2007). A novel family of potent negative allosteric modulators of group II metabotropic glutamate receptors. J. Pharmacol. Exp. Ther. 322, 254–264. Johnson, P.L., Fitz, S.D., Engleman, E.A., Svensson, K.A., Schkeryantz, J.M., and Shekhar, A. (2013). Group II metabotropic glutamate receptor type 2 allosteric potentiators prevent sodium lactate-induced panic-like response in panic-vulnerable rats. J. Psychopharmacol. 27, 152–161. Kew, J.N.C., and Kemp, J.A. (2005). Ionotropic and metabotropic glutamate receptor structure and pharmacology. Psychopharmacology 179, 4–29. Lavreysen, H., Langlois, X., Ahnaou, A., Drinkenburg, W., te Riele, P., Biesmans, I., Van der Linden, I., Peeters, L., Megens, A., Wintmolders, C., et al. (2013). Pharmacological characterization of JNJ-40068782, a new potent, selective, and systemically active positive allosteric modulator of the mGlu2 receptor and its radioligand [3H]JNJ-40068782. J. Pharmacol. Exp. Ther. 346, 514–527. €tjens, R., Lavreysen, H., Langlois, X., Donck, L.V., Nun˜ez, J.M.C., Pype, S., Lu and Megens, A. (2015). Preclinical evaluation of the antipsychotic potential of

the mGlu2-positive allosteric modulator JNJ-40411813. Pharmacol. Res. Perspect. 3, e00097. Lundstro¨m, L., Bissantz, C., Beck, J., Wettstein, J.G., Woltering, T.J., Wichmann, J., and Gatti, S. (2011). Structural determinants of allosteric antagonism at metabotropic glutamate receptor 2: mechanistic studies with new potent negative allosteric modulators. Br. J. Pharmacol. 164, 521–537. Lundstro¨m, L., Bissantz, C., Beck, J., Dellenbach, M., Woltering, T.J., Wichmann, J., and Gatti, S. (2016). Pharmacological and molecular characterization of the positive allosteric modulators of metabotropic glutamate receptor 2. Neuropharmacology 111, 253–265. €hlemann, A., Zenner, M.T., Fischer, C., Stahl, Malherbe, P., Kratochwil, N., Mu M., Gerber, P.R., Jaeschke, G., and Porter, R.H. (2006). Comparison of the binding pockets of two chemically unrelated allosteric antagonists of the mGlu5 receptor and identification of crucial residues involved in the inverse agonism of MPEP. J. Neurochem. 98, 601–615. Mølck, C., Harpsøe, K., Gloriam, D.E., Clausen, R.P., Madsen, U., Pedersen, €uner-Osborne, L.Ø., Jimenez, H.N., Nielsen, S.M., Mathiesen, J.M., and Bra H. (2012). Pharmacological characterization and modeling of the binding sites of novel 1, 3-bis (pyridinylethynyl) benzenes as metabotropic glutamate receptor 5-selective negative allosteric modulators. Mol. Pharmacol. 82, 929–937. Navarro, G., Cordomı´, A., Zelman-Femiak, M., Brugarolas, M., Moreno, E., Aguinaga, D., Perez-Benito, L., Corte´s, A., Casado´, V., Mallol, J., et al. (2016). Quaternary structure of a G-protein-coupled receptor heterotetramer in complex with Gi and Gs. BMC Biol. 14, 1–12. Nicholls, R.E., Zhang, X.-l., Bailey, C.P., Conklin, B.R., Kandel, E.R., and Stanton, P.K. (2006). mGluR2 acts through inhibitory Ga subunits to regulate transmission and long-term plasticity at hippocampal mossy fiber-CA3 synapses. Proc. Natl. Acad. Sci. USA 103, 6380–6385. Niswender, C.M., and Conn, P.J. (2010). Metabotropic glutamate receptors: physiology, pharmacology, and disease. Annu. Rev. Pharmacol. Toxicol. 50, 295–322. Pellissier, L., Sallander, J., Campillo, M., Gaven, F., Queffeulou, E., Pillot, M., Dumuis, A., Claeysen, S., Bockaert, J., and Pardo, L. (2009). Conformational toggle switches implicated in basal constitutive and agonist-induced activated states of 5-HT4 receptors. Mol. Pharmacol. 75, 982–990. Pin, J.P., and Bettler, B. (2016). Organization and functions of mGlu and GABAB receptor complexes. Nature 540, 60–68. Rasmussen, S.G., Choi, H.J., Fung, J.J., Pardon, E., Casarosa, P., Chae, P.S., Devree, B.T., Rosenbaum, D.M., Thian, F.S., Kobilka, T.S., et al. (2011). Structure of a nanobody-stabilized active state of the beta(2) adrenoceptor. Nature 469, 175–180. Rowe, B.A., Schaffhauser, H., Morales, S., Lubbers, L.S., Bonnefous, C., Kamenecka, T.M., McQuiston, J., and Daggett, L.P. (2008). Transposition of three amino acids transforms the human metabotropic glutamate receptor (mGluR)-3-positive allosteric modulation site to mGluR2, and additional characterization of the mGluR2-positive allosteric modulation site. J. Pharmacol. Exp. Ther. 326, 240–251. Sansuk, K., Deupi, X., Torrecillas, I.R., Jongejan, A., Nijmeijer, S., Bakker, R.A., Pardo, L., and Leurs, R. (2011). A structural insight into the reorientation of transmembrane domains 3 and 5 during family A G protein-coupled receptor activation. Mol. Pharmacol. 79, 262–269. Schaffhauser, H., Rowe, B.A., Morales, S., Chavez-Noriega, L.E., Yin, R., Jachec, C., Rao, S.P., Bain, G., Pinkerton, A.B., Vernier, J.M., and Vernier, J.M. (2003). Pharmacological characterization and identification of amino acids involved in the positive modulation of metabotropic glutamate receptor subtype 2. Mol. Pharmacol. 64, 798–810. Sharma, S., Rodriguez, A.L., Conn, P.J., and Lindsley, C.W. (2008). Synthesis and SAR of an mGluR5 allosteric partial antagonist lead: unexpected modulation of pharmacology with slight structural modifications to a 5-(phenylethynyl) pyrimidine scaffold. Bioorg. Med. Chem. Lett. 18, 4098–4101. Schro¨dinger. (2015). Maestro v2015.4 (Schro¨dinger, LLC). Spyridaki, K., Matsoukas, M.T., Cordomı´, A., Gkountelias, K., Papadokostaki, M., Mavromoustakos, T., Logothetis, D.E., Margioris, A.N., Pardo, L., and Liapakis, G. (2014). Structural-functional analysis of the third transmembrane

Structure 25, 1–10, July 5, 2017 9

Please cite this article in press as: Pe´rez-Benito et al., Molecular Switches of Allosteric Modulation of the Metabotropic Glutamate 2 Receptor, Structure (2017), http://dx.doi.org/10.1016/j.str.2017.05.021

domain of the corticotropin-releasing factor type 1 receptor: role in activation and allosteric antagonism. J. Biol. Chem. 289, 18966–18977. Tesmer, J.J.G., Berman, D.M., Gilman, A.G., and Sprang, S.R. (1997). Structure of RGS4 bound to AlF4- activated Gia1: stabilization of the transition state for GTP hydrolysis. Cell 89, 251–261. Trabanco, A.A., Cid, J.M., Lavreysen, H., Macdonald, G.J., and Tresadern, G. (2011). Progress in the development of positive allosteric modulators of the metabotropic glutamate receptor 2. Curr. Med. Chem. 18, 47–68. Tresadern, G., Cid, J.M., Macdonald, G.J., Vega, J.A., de Lucas, A.I., Garcı´a, A., Matesanz, E., Linares, M.L., Oehlrich, D., Lavreysen, H., et al. (2010). Scaffold hopping from pyridones to imidazo[1,2-a]pyridines. New positive allosteric modulators of metabotropic glutamate 2 receptor. Bioorg. Med. Chem. Lett. 20, 175–179. Tresadern, G., Cid, J.-M., and Trabanco, A.A. (2014). QSAR design of triazolopyridine mGlu2 receptor positive allosteric modulators. J. Mol. Graph Model. 53, 82–91. Turlington, M., Noetzel, M.J., Chun, A., Zhou, Y., Gogliotti, R.D., Nguyen, E.D., Gregory, K.J., Vinson, P.N., Rook, J.M., Gogi, K.K., et al. (2013). Exploration of

10 Structure 25, 1–10, July 5, 2017

allosteric agonism structure-activity relationships within an acetylene series of metabotropic glutamate receptor 5 (mGlu5) positive allosteric modulators (PAMs): discovery of 5-((3-fluorophenyl) ethynyl)-N-(3-methyloxetan-3-yl) picolinamide (ML254). J. Med. Chem. 56, 7976–7996. Venkatakrishnan, A.J., Deupi, X., Lebon, G., Tate, C.G., Schertler, G.F., and Babu, M.M. (2013). Molecular signatures of G-protein-coupled receptors. Nature 494, 185–194. Warne, T., Moukhametzianov, R., Baker, J.G., Nehme, R., Edwards, P.C., Leslie, A.G., Schertler, G.F., and Tate, C.G. (2011). The structural basis for agonist and partial agonist action on a beta(1)-adrenergic receptor. Nature 469, 241–244. Wu, H., Wang, C., Gregory, K.J., Han, G.W., Cho, H.P., Xia, Y., Niswender, C.M., Katritch, V., Meiler, J., Cherezov, V., et al. (2014). Structure of a class C GPCR metabotropic glutamate receptor 1 bound to an allosteric modulator. Science 344, 58–64. Xu, F., Wu, H., Katritch, V., Han, G.W., Jacobson, K.A., Gao, Z.G., Cherezov, V., and Stevens, R.C. (2011). Structure of an agonist-bound human A2A adenosine receptor. Science 332, 322–327.

Please cite this article in press as: Pe´rez-Benito et al., Molecular Switches of Allosteric Modulation of the Metabotropic Glutamate 2 Receptor, Structure (2017), http://dx.doi.org/10.1016/j.str.2017.05.021

STAR+METHODS KEY RESOURCES TABLES

REAGENT or RESOURCE

SOURCE

IDENTIFIER

Guanosine 5’-diphosphoste sodium salt

Sigma

G7127

L-Glutamic acid, monosodium salt monohydrate

Sigma

G2834

Saponin

Cabiochem

558255

[35S]GTPgS

PerkinElmer

NEG030X001MC

[3H]LY341495

ARC

ART 1439

[3H]JNJ-46281222

Janssen

Chemicals, Peptides, and Recombinant Proteins

JNJ-42341806

Janssen

BINA

Janssen

JNJ-40068782

Janssen

JNJ-46281222

Janssen

Ro-676221

Janssen

Ro-4491533

Janssen

Ro-4495819

Janssen

Critical Commercial Assays Lipofectamine LTX transfection reagent

Life Technologies

15338

CHO-K1 WT cells

ATCC

CCL-61

CHO-K1_hmGlu2 stable cell line

Janssen

Experimental Models: Cell Lines

Recombinant DNA mGlu2 WT and mutated hmGlu2

Life Technologies

GeneArt

Graphpad Prism v4.02

Graphpad Software Inc.

https://www.graphpad.com/

Molecular operating environment (MOE) pharmacophore elucidator

Chemical Computing Group, 2014

https://www.chemcomp.com/

Maestro protein and ligand preparation, GLIDE XP docking

Schro¨dinger Software LLC (2015)

https://www.schrodinger.com/

Software and Algorithms

Gaussian vG09 for calculation of RESP derived charges

Gaussian Inc

http://gaussian.com/

GROMACS v5.0.6 molecular dynamics

Abraham et al. (2015) SoftwareX 1-2 19-25.

http://www.gromacs.org/

CONTACT FOR REAGENT AND RESOURCE SHARING Requests for further information or reagents may be directed to the Lead Contact, Gary Tresadern ([email protected]) EXPERIMENTAL MODEL AND SUBJECT DETAILS Datasets for Pharmacophore Building Known mGlu2 PAMs and NAMs were retrieved from ChEMBL (Bento et al., 2014) using keywords ‘GRM2’ and ‘mGlu2’. Molecules with concentration response bioactivity <1 mM were kept. Orthosteric ligands containing amino acid substructures were removed. Thus 296 active PAMs but only 17 active NAMs were identified. Therefore, the NAMs dataset was augmented by retrieving examples from 12 patents (WO2007110337, WO2005040171, WO2008128889, WO2008119689, WO2006099972, WO2006084634, WO2005123738, WO2003066623, WO2002023665, WO2002083652, WO2001129012, US20070072879) delivering a final NAM dataset of 289 unique molecules. Known inactive molecules are often absent in public bioactivity databases. Hence these were taken from Janssen in-house mGlu2 receptor PAM and NAM high throughput screening (HTS) data. Inactives were retained in a similar MW range to the PAMs (233 to 515 Dalton) and NAMs (303 to 658 Dalton). A subset was selected to initially derive the pharmacophore in an automated manner. Hence, in the case of PAMs 86 molecules were selected whereas for NAMs 82 were chosen. These were combined with two randomly selected sets of 235 inactive molecules from either the Janssen PAM or NAM HTS.

Structure 25, 1–10.e1–e4, July 5, 2017 e1

Please cite this article in press as: Pe´rez-Benito et al., Molecular Switches of Allosteric Modulation of the Metabotropic Glutamate 2 Receptor, Structure (2017), http://dx.doi.org/10.1016/j.str.2017.05.021

PAM and NAM Pharmacophore Models Datasets for PAMs (296 molecules) and NAMs (289 molecules) were created from ChEMBL and from patents (see Supplemental Information). Inactive molecules were taken from Janssen mGlu2 receptor PAM and NAM high throughput screening. PAM pharmacophore elucidation was performed on a training set (86 active similar to PAMs 1-3 and 235 inactive) using MOE which delivered a satisfactory alignment and placed four features. The best statistical pharmacophore (99% actives and only 46% inactives hit) was augmented with additional features (see main text). Further statistical testing on the full dataset (296 actives and 2791 inactives) identified 121 actives (41%) and only 204 inactives (7%). This is expected because a single pharmacophore is unlikely to hit diverse chemical series in the full dataset. A NAM pharmacophore was also developed using a training set of 82 active and 235 inactive. The automated elucidation delivered a satisfactory alignment of actives and placed five features. The pharmacophore hit 85% of actives and only 17% of the inactive molecules. Again, the initial pharmacophore was augmented with additional features based on knowledge of the SAR (see text) and tested further on the entire dataset (289 actives and 2800 inactives). Hitting all features was too restrictive; hence we permitted molecules to match only 9 of the 11. This identified 171 of the actives (59%) and 28 inactives (1%). Hence both PAM and NAM pharmacophores discriminate actives and inactives and captured the SAR. Pharmacophore for mGlu1 An analogous approach was used here for mGlu1 as described for mGlu2. The aim was to investigate if a pharmacophore model can be informative for defining the binding mode of mGlu allosteric modulators. This was done by generating a ligand based pharmacophore for mGlu1 receptor NAMs and comparing to the corresponding crystal structure (PDB 4OR2(Wu et al., 2014)). Known mGlu1 NAMs were retrieved from ChEMBL using keywords ‘GRM1’ and ‘mGlu1’. Molecules with concentration response less than 1 mM were retained, orthosteric amino acid substructures were removed, and NAMs were identified with activity type matching ‘‘antagonist’’, ‘‘inhibitory’’ or ‘‘negative allosteric’’. This retrieved a total of 321 sub-micromolar mGlu1 NAMs. To enable subsequent comparison with the crystal structure, only molecules containing an aminothiazole scaffold were retained, resulting in 19 NAMs. Known inactives were extracted from Janssen in-house mGlu1 receptor NAM high throughput screening (HTS) data. A set of 423 inactives were chosen containing the aminothiazole substructure and being in a similar MW range to the actives (300 to 450 Dalton). The mGlu1 active and inactive NAMs was used as input for the ph4elucidate tool in MOE with all settings as described in the manuscript. The automated elucidation delivered a five feature pharmacophore with excellent statistical behavior to identify mGlu1 NAMs, hitting all 19 of the actives, and only 2 of the inactives. The initial best pharmacophore was relatively simple, containing three aromatic centers and an acceptor, Figure S3A. Comparing with the two inactives that were hit by the pharmacophore shows that one example extends beyond the region of the lower (as shown) aromatic ring, Figure S3B. Hence, placing an excluded volume prevented these inactives being hit and improved the performance of the pharmacophore, with all 19 active NAMs being retrieved but no inactives. In addition, the pharmacophore was augmented with additional pi aromatic features to specify the orientation of the two distal aromatic rings, Figures S3C and S3D. The pharmacophore was then used to search against a large set of 65136 inactives from the previous mGlu1 NAM HTS. The pharmacophore only hit 83 molecules. Interestingly, although built to be very specific towards aminothiazole mGlu1 NAMs, the pharmacophore hit 85 of the 321 ChEMBL actives, a pleasing result suggesting similarities between active NAM chemical series. Overall, the pharmacophore shows excellent statistical performance. The pharmacophore specifies that active molecules contain three aromatic centers and a central acceptor group. Meanwhile the distal phenyl (often halogen substituted) does not permit large substituents as captured by the excluded volume. It is known that mGlu allosteric modulators bind deeper than conventional class A ligands and in a parallel orientation to the alpha helices. Meanwhile, they are often lipophilic and contain multiple aromatic rings connected linearly. Hence, ligands may adopt either of two, ‘up’ or ‘down’, binding modes. In this case, the excluded volume suggests the space for substitution is limited in this region and hence this part of the ligand likely enters deepest into the receptor. Therefore, the orientation of the ligands provided by the pharmacophore alignment and SAR analysis which is shown in Figures S3A–S3D is strikingly similar to the binding mode in the crystal structure, Figure S3E. Indeed the overlay is good enough to directly superpose the pharmacophore onto the protein ligand complex structure, Figure S3F. The important features from the pharmacophore are well matched when compared with the crystal structure of mGlu1 7-TM, Figure S3E. The pharmacophore predicts the necessity of the lower aromatic ring which in turn is satisfied by interaction with amino acids Trp798, Phe801, Ala818 and Val819 in the crystal structure. The excluded volume corresponds to the bottom of the pocket explaining why molecules cannot extend further in that region. The hydrogen bond acceptor feature complements with a hydrogen bond donor from Asn760 (this interaction has been previously predicted, (Harpsøe et al., 2015) and is consistent with our molecular dynamics simulations, discussed in more detail below). Meanwhile the thiazole and ‘upper’ aromatic rings correspond to the aromatic pharmacophore features that interact with amino acids such as Leu648, Val664, Val753, Leu757, Tyr805 and Thr815. In summary, we have applied an analogous pharmacophore and SAR analysis approach to that used in the manuscript for mGlu2 ligands. The results agree with the interactions that are seen in the mGlu1 7-TM NAM crystal structure, demonstrating the usefulness of the approach to discriminate plausible interaction motifs and hence provide insight when defining binding modes. Taken along with the wide array of experimental data (for instance, mutagenesis and ligand binding displacement) it can be seen how the pharmacophore can also contribute towards defining the binding mode. Whilst the pharmacophore is different to the mGlu2 cases some consistencies are seen such as the prevalence of lipophilic and scarcity of polar interactions. Computational Models of the mGlu2 Receptor in Complex with PAMs 1-3 and NAMs 4-6 Structural models (inactive and active-like states) of the mGlu2 receptor (Uniprot code Q14416) 7-TM were built. The crystal structure of the highly similar inactive mGlu5 receptor (PDB 4OO9, 51% sequence identity in 7-TM, Figure S4) was used for the inactive state. e2 Structure 25, 1–10.e1–e4, July 5, 2017

Please cite this article in press as: Pe´rez-Benito et al., Molecular Switches of Allosteric Modulation of the Metabotropic Glutamate 2 Receptor, Structure (2017), http://dx.doi.org/10.1016/j.str.2017.05.021

The ECL2 is not resolved in the mGlu5 receptor, hence it was modeled using the mGlu1 receptor structure (PDB 4OR2). The active-like state was modeled by changing the conformation of the intracellular TM6 of the mGlu2 receptor (N7576.32a.34c-M7666.41a.43) for the active conformation of b2- 6.32-6.41 in the b2AR-Gs complex (PDB 3SN6). The Gi protein was included in this active-like model with its C-terminal a5 helix bound in the intracellular cavity. The extracellular part, including receptor side chain conformation, was comparable in the active-like and inactive models. The a-helical domain of Gia was modeled in the ‘‘closed’’ conformation, using the crystal structure of [AlF4-]-activated Gi (PDB 1AGR) (Tesmer et al., 1997). A similar approach has recently been used (Doornbos et al., 2016; Navarro et al., 2016). Models were constructed with MOE and Maestro (2015). Ligand conformers were docked into the models using Glide XP. Docking poses were further studied in explicit membrane MD simulations with GROMACS v5.0.6 (Abraham et al., 2015). The complexes of Mavoglurant-mGlu5 receptor (Dore et al., 2014), FITM-mGlu1 receptor (Wu et al., 2014), NAMs 4-6 with the inactive mGlu2 receptor, and PAMs 1-3 with the active mGlu2 receptor in complex with Gi were embedded in a pre-equilibrated box (9x9x9 for inactive or 10x10x19 nm for active) containing a lipid bilayer (205 or 297 POPC molecules) with explicit solvent (14000 or 47000 waters) and 0.15 M concentration of Na+ and Cl- (140 or 490 ions). Each system was energy minimized and subjected to a 5 step MD equilibration (10+5+2+2+2 ns) in which constraints in hydrogen atoms, protein loops, and protein and ligand atoms were subsequently relaxed. Three independent replicas of unrestrained MD trajectories were produced for 1 ms using a 2 fs time step and constant temperature of 300K. The AMBER99SD-ILDN force field was used for the protein, the parameters described by Berger et al. (Berger et al., 1997) for lipids, and the general Amber force field and HF/6-31G*-derived RESP atomic charges for the ligand. This combination of protein and lipid parameters has recently been validated (Cordomı´ et al., 2012). MD Simulations Performed on mGlu1 and mGlu5 Crystal Structures The aim was to use Molecular Dynamics (MD) simulations on the mGlu1 and mGlu5 receptor 7-TM crystal structures to check for consistency in our findings when using X-ray coordinates as input for modelling compared with models of mGlu2. The method was the same as that described for mGlu2. The receptor was prepared in the same way. The ligand and receptor starting coordinates were taken from the crystal structures. The receptor was placed in a pre-equilibrated membrane and solvated. The system was subjected to the same multistep equilibration procedure (21 ns in total). Production dynamics were performed for 1 ms. The amide group within the mGlu1 ligand was observed to rotate and adopt an alternative conformation which still fits the electron density, as commented in the above discussion on the mGlu1 NAM pharmacophore. This effect has been reported before, (Harpsøe et al., 2015) and does not change the overall conformation of the molecule, but instead rotates the carbonyl and methyl amine to a more optimal conformation featuring a hydrogen bond to amino acid Asn7605.47. Regarding the mGlu5 receptor, the extracellular loop 2 (ECL2) is not resolved in the X-ray structure, and was therefore modelled to provide an initial input geometry prior to equilibration. Also, the ligand that was used in the simulation was a close analogue, compound 12a from Sharma et al, (Sharma et al., 2008) a full NAM containing a similar substructure and phenylacetylene scaffold to Mavoglurant. This is unlikely to impact the results and the simulations show that the ligand is extremely stable and adopts an optimal orientation. The results confirmed the observations seen for the mGlu2 NAMs in the manuscript. In short, NAMs do not induce conformational changes in the amino acids forming part of the ‘trigger switch’ or ‘transmission switch’. This can be seen clearly from the close overlap of the amino acids throughout the simulations, and their lack of movement, in Figures S7A and S7B along with Figures S8A and S8B. The dihedral angles in panels C and D for both Figures also show there is no flipping or change throughout the simulations. Finally, panel E in both Figures S7 and S8 show the protein and ligand to be highly stable and well behaved throughout the simulations. Hence, MD simulations performed with an analogous approach, but on crystal structures in place of a homology model, confirm our observations for mGlu2 NAMs, thereby adding further support to our hypothesis of the role of these amino acids in functional activity of allosteric modulators. METHOD DETAILS Plasmids, Cell Transfection and Cell Culture cDNA constructs encoding human non-mutated and mutated mGlu2 receptors were synthesized by GeneArt (Life Technologies), subcloned to the mammalian expression vector pcDNA3.1(+) (Life Technologies) and amplified through E. coli transformation. CHO-K1 cells were used for transient transfection. 24 hours prior to transfection, cells were seeded at high density (20,000 cells/cm2) into 14 cm Ø plates. Transfections were performed using lipofectamine LTX reagent (Life technologies). CHO-K1 cells expressing mutated and non-mutated mGlu2 receptors were maintained in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% heat-inactivated FBS, penicillin, streptomycin, pyruvic acid and L-Glutamine. Cells were kept in a humidified atmosphere at 37 C and 5% CO2 and were subcultured twice weekly. Membrane Preparation Cells were detached from the plates by scraping into 50 mM Tris-HCl buffer, pH 7.4 and subsequently centrifuged for 10 min at 16,000 rpm in a Sorvall 5C Plus SS34 centrifuge at 4 C. Pellets were resuspended in ice-cold 5 mM hypotonic Tris-HCl, pH 7.4 and homogenized using an Ultra Turrax homogenizer (IKA-Werke GmbH & Co.KG, Staufen, Germany) at 24,000 rpm. Homogenates were centrifuged at 18,000 rpm for 20 min at 4 C. Remaining pellets were suspended in 50 mM Tris-HCl pH 7.4 and the homogenization step was repeated. Aliquots were stored at -80 C. Protein concentrations were determined using the Bradford method (Bio-Rad, Hercules, CA, USA) using BSA as a standard.

Structure 25, 1–10.e1–e4, July 5, 2017 e3

Please cite this article in press as: Pe´rez-Benito et al., Molecular Switches of Allosteric Modulation of the Metabotropic Glutamate 2 Receptor, Structure (2017), http://dx.doi.org/10.1016/j.str.2017.05.021

[3H]LY341495 Binding Assay Membranes were allowed to thaw and subsequently homogenization was performed using an Ultra Turrax homogenizer at 24,000 rpm. After determination of the membrane protein concentration using the Bradford method (Bio-Rad), samples were diluted in ice-cold assay buffer (50 mM Tris-HCl pH 7.4, 10 mM MgCl2 and 2 mM CaCl2). Binding assays were performed in a total volume of 500 ml, containing increasing concentrations of glutamate (10 nM to 1 mM), 10 mg membrane protein and 3 nM [3H]-LY341495. Nonspecific binding was determined using 1 mM glutamate. After incubation for 1 hour at room temperature, membranes were rapidly filtered through a 96-well GF/B filterplate (PerkinElmer) on a PerkinElmer filtermate harvester and washed three times with ice-cold wash buffer (50 mM Tris-HCl pH 7.4). Plates were allowed to dry overnight. Filter-bound radioactivity was determined by scintillation spectrometry using a Topcount NXT Microplate Scintillation and Luminescence Counter (PerkinElmer). [35S]GTPgS Binding Assay Membranes were thawed and homogenized using an Ultra Turrax homogenizer at 24,000 rpm. Samples were diluted in ice-cold assay buffer (20 mM HEPES pH 7.4, 100 mM NaCl, 3 mM MgCl2, and 10 mM GDP and 14.3 mg/mL saponin). DMSO concentrations were %1%. Assay mixtures containing a variable concentration of test compound and 10 mg membrane protein were pre-incubated with buffer (to detect agonist effects) or an EC20 or EC80-equivalent concentration of glutamate (to detect PAM or NAM effects respectively). After 30 minutes of incubation at 30  C, 0.1 nM [35S]GTPgS was added. The reaction was stopped after another 30 minute incubation at 30  C by rapid filtration through a 96-well GF/B filterplate (PerkinElmer) on a PerkinElmer filtermate harvester. Plates were washed three times with ice-cold wash buffer (10 mM NaH2PO4/10 mM Na2HPO4, pH 7.4) and dried overnight. Filterbound radioactivity was counted in a Topcount microplate scintillation and luminescence counter (PerkinElmer). [3H]JNJ-46281222 Binding Assay Membranes were allowed to thaw and subsequently homogenization was performed using an Ultra Turrax homogenizer at 24,000 rpm. Samples were diluted in ice-cold assay buffer (50 mM Tris-HCl pH 7.4, 10 mM MgCl2 and 2 mM CaCl2). Binding assays were performed in glass tubes in a total volume of 500 ml, containing variable concentrations of test compound, 75 mg membrane protein for the stably expressed hmGlu2 or 150 mg for the mutagenesis constructs and 4 nM [3H]JNJ-46281222. Nonspecific binding was determined using 10 mM JNJ-42341806. DMSO concentrations were %1%. The reaction was stopped after incubation for 1 hour at room temperature by rapid filtration over pre-coated (PEI 0.1%, Sigma-Aldrich) GF/C filters through a Brandel harvester 96 (Brandel, Gaithersburg, MD, USA). Filters were washed three times with ice-cold wash buffer (50 mM Tris-HCl pH 7.4). Filter-bound radioactivity was determined using liquid scintillation spectrometry on a Tri-Carb 2810TR counter (PerkinElmer). Selectivity Assays mGlu receptor panel selectivity assays: Ca2+ assays with human mGlu1, 3, 5, 7, or 8 receptor expressing HEK 293 cells were performed as reported in Lavreysen et al. (2013), except for a slight change in the procedure for mGlu5: cells expressing the human mGlu5 receptor were seeded at 40,000 cells/well in MW384. Twenty-four hours after seeding, cells were incubated for 90 min in Ca2+ assay kit (Molecular Devices) dissolved in saline PBS supplemented with 5 mmol/L probenecid, pH 7.4 (f.c. 2.5 mmol/L probenecid as loading buffer was added on the cell layer without removal of medium) before measurements. Measurement of [35S]GTPgS binding to membranes from CHO cells expressing the rat mGlu6 receptor and membranes from L929sA cells expressing the human mGlu4 receptor were conducted also as described in Lavreysen et al. (2013). Molecules 1 to 6 were tested in a functional mGlu receptor assay panel (see Table S2). All PAMs did not activate any of the other human mGlu receptor subtypes or the rat mGlu6 receptor up to 10 mM concentration limit. However, they did show weak mGlu2 agonistic activity as described previously for 1 (Galici et al., 2006), 2 (Lavreysen et al., 2013) and 3 (Doornbos et al., 2016) which may be due to residual levels of endogenous glutamate. PAMs did not inhibit glutamate-induced signaling at any of the receptors up to 10 mM concentration. Considering NAMs 4, 5 and 6, no activation of any mGlu receptors was seen up to the same concentration limit. Importantly however, NAMs showed antagonistic inhibitory effects at mGlu3 receptors in a comparable range to their activity in the analogous assay at mGlu2 receptors. Hence, NAMs were not selective versus mGlu3 receptors. Data Analysis Data analyses were performed using Prism 4.02 (GraphPad software, La Jolla, CA, USA). For the data sets of [3H]JNJ-46281222 binding experiments, pIC50 values were obtained using non-linear regression curve fitting into a sigmoidal concentration-response curve using the equation: Y = Bottom + (Top - Bottom) / {1 + 10^(X - LogIC50)}. pKi values were obtained from pIC50 values using the Cheng-Prusoff equation. Concentration-response curves obtained in [35S]GTPgS binding experiments were fitted using non-linear regression curve fitting into a sigmoidal concentration-response curve using the equation: Y = Bottom + (Top - Bottom) / {1 + 10^ [(LogEC50 - X) 3 Hill Slope]}. Statistical analysis was performed if indicated, using one-way ANOVA with Dunnett’s post-test. QUANTIFICATION AND STATISTICAL ANALYSIS Methods and parameters used for statistical analyses are reported in the Method Details section.

e4 Structure 25, 1–10.e1–e4, July 5, 2017