Structural insights into the activation initiation of full-length mGlu1

Glutamate is the main excitatory neurotransmitter in the human brain, and it exerts diverse responses through ionotropic glutamate receptors (iGluRs) and metabotropic glutamate receptors (mGluRs) (Nakanishi and Masu, 1994). mGluRs are members of the C family of GPCRs, and are divided into three groups based on G protein coupling, sequence homology, and ligand selectivity (Stansley and Conn, 2019). mGlu1 and mGlu5 belong to group I and predominantly couple to Gq/11. They are postsynaptic glutamate receptors that respond much slower than the typical postsynaptic AMPA or NMDA-type iGluRs (millisecond timescale) (Scheefhals and MacGillavry, 2018). mGlu1 is involved in multiple physiological processes in the central nervous system (CNS) and is a promising therapeutic target for treating CNS associated disorders. Notably, positive allosteric modulators (PAMs) of mGlu1 show the selectivity in the striatum dopamine signaling inhibition and hold the potential for treating schizophrenia with fewer side effects (Stansley and Conn, 2019). Constitutive homoor hetero-dimerization is the defining feature of class C GPCRs, and mGluRs form homodimers mediated by the intermolecular disulfide bond within the N-terminal extracellular domain (ECD) (Wu et al., 2014). The large ECD of mGluRs can be divided into two parts: the venus flytrap (VFT) domain and the cysteine-rich domain (CRD) (Muto et al., 2007). In contrast to GPCRs of other classes, ligands bind to the orthosteric site in the VFT domain or the allosteric site in the 7TM domain (Pin and Bettler, 2016). So far, crystal structures have been determined for the orthosteric ligand bound (agonist or antagonist) and ligand free form of VFT domain of mGlu1 (Kunishima et al., 2000; Tsuchiya et al., 2002), and the negative modulator (NAM) bound 7TM domain of mGlu1 (Wu et al., 2014). These structures provide clues about the dimerization of the VFT domain and high-resolution architecture of the mGlu1 transmembrane region. However, the full-length structure of mGlu1 is still unknown and the agonist-induced conformational transition of mGlu1 remains elusive due to the missing apo and active full-length mGlu1 structures. Here, we present two full-length mGlu1 structures in the apo and intermediate active states at 3.96 Å and 3.65 Å, respectively, using cryo-EM single particle analysis. This study captures a new intermediate state of mGluRs and provides additional insights into the dynamic activation process of mGluRs. To enhance the expression and stability of mGlu1, the wild-type mGlu1 was modified by Nand C-terminal truncations and the optimized construct was expressed in the insect cell system (See Supplementary Materials). To seek the mGlu1 active conformation, the orthosteric agonist L-Quisqualic acid together with PAM Ro 67-4853 were added during protein purification. L-Quisqualic acid is a group I mGluRs preferred agonist that shows higher potency than glutamate (Schoepp et al., 1999), and Ro 67-4853 is a selective PAM of mGlu1 (Knoflach et al., 2001). Eventually, we obtained the cryo-EM structure of the full-length mGlu1 homodimer at a global resolution of 3.96 Å (Fig. S1, Table S1, Supplementary Materials). Similar to the full-length mGlu5, the CRD, VFT and 7TM domains constitute the whole homodimer conformation of mGlu1 with central symmetry (Fig. 1A). Generally, the VFT domain contains the orthosteric binding site and shows two major conformational states: open or closed state. An open state is when the VFT domain is in inactive state in presence or absence of ligand, while a close state is induced by agonists, which may lead to receptor activation (Kunishima et al., 2000; Tsuchiya et al., 2002). However, the activation of mGlu receptors from a resting state (apo or inactive state) to an active state requires additional large conformation changes of both receptors in homodimer. In the resting state, the orientations of lobe 2s in the two VFTs are distant, while in the active state, lobe 2s rotate and become closer (Kunishima et al., 2000). Based on the solved crystal structures, there exist several different conformation combinations of VFTs in mGluR dimers: “Roo” (rest open-open), “Rcc” (rest close-close), “Aoo” (active open-open), “Acc” (active close-close), and “Aco” (active close-open) (Doumazane et al., 2013). Although an excess amount of agonist and PAM were added during mGlu1 purification, in the cryo-EM structure, there is no electron density either for L-Quisqualic acid in the ligand binding pocket of VFTs, or for PAM Ro 67-4853 in 7TM domains, indicating that mGlu1 is in the ligand free form. Compared to the crystal structure of the ligand free


Constructs, expression and purification of Nb43
Nb43 was synthesized and sub-cloned into a modified pET-28a (+) vector with a 6× His tag before the N-terminus. The construct was transformed into Escherichia coli BL21 for protein expression. Cells were grown in Luria Broth in the presence of kanamycin at 37 °C until OD 600 reached 0.6 -0.8. Protein expression was induced by adding 0.5 mM IPTG, and the cells were harvested after incubation at 25 °C for 16 -20 h. Cells were harvested by centrifugation and then resuspended in buffer A, containing 300 mM NaCl, and 40 mM Tris (pH 8.0). The supernatant was isolated by ultracentrifugation and filtration after sonication. For Nb43 purification, protocols were as follows. In brief, Ni-NTA resin was loaded to supernatant to incubate at 4 °C for 1 -2 h with constant stirring. The resin was washed with 10 CVs of buffer B containing 200 mM NaCl, 50 mM HEPES (pH 7.5), and 50 mM imidazole. Nb43 was then eluted by 3 CVs of buffer C, containing 200 mM NaCl, 50 mM HEPES (pH 7.5), and 300 mM imidazole. Nb43 was concentrated and frozen at −80 °C for future use.
For apo mGlu1, cryo-EM datasets were collected on a Titan Krios electron microscope (Thermo Fisher Scientific) equipped with a Gatan K3 summit direct electron camera (Gatan, Inc.) in super resolution mode operating at 300 kV accelerating voltage. Movies were taken under the EFTEM nanoprobe mode with a 70 μm C2 aperture and calibrated magnification of 22,500×, corresponding to a pixel size of 1.06 Å. Each movie comprises 45 frames with a total dose of 80 electrons per Å 2 and 3 s exposure time with a dose rate of 30 e -/pix/s. Serial EM software (Mastronarde, 2005) was used to collect data with a defocus range of -1.0 to -2.0 μm.
For agonist-bound mGlu1, cryo-EM datasets were collected on a Titan Krios electron 4 microscope equipped with a Gatan K2 summit direct electron camera and a Gatan Quantum energy filter with a slit width of 20 eV in super resolution mode. Movies were taken under the nanoprobe mode, with a 50 μm C2 aperture and calibrated magnification of 130,000×, corresponding to a pixel size of 1.04 Å. Each movie comprises 45 frames with a total dose of 60 electrons per Å 2 and 8.1 s exposure time with a dose rate of 8 e -/pix/s. Serial EM software (Mastronarde, 2005) was used to collect data with a defocus range of -1.0 to -2.0 μm.
6,604 movies were collected and patch motion correction was performed. Patch CTF estimation was used to determine contrast transfer function (CTF) parameters for each non-dose micrograph. 7,322,716 particles were yielded by template based autopicking, and a subset of seven million particles was used to do three cycles of 2D classification. Four classes of initial models were generated by selecting 1,715,013 particle projections. After several rounds of 3D classification, the best model was generated and then used as the reference model for the final 3D classification.
The final dataset of 134,512 particle projections was used for the final homogenous and nonuniform refinement in cryoSPARC, and a density map with a nominal resolution of 3.96 Å was determined by gold standard Fourier shell correlation (FSC) using the 0.143 criterion.
Estimation of local resolution was determined with the local resolution refinement in cryoSPARC.
18,832 movies were collected and motion correction was performed by MotionCor2 (Zheng et al., 2017). CTFFIND 4.1 (Rohou and Grigorieff, 2015) was used to determine CTF parameters for each non-dose micrograph. 2,909,206 particles were yielded by Gautomatch v 0.56 (http://www.mrc-lmb.cam.ac.uk/kzhang/Gautomatch), and a subset of one million particles was used for 2D classification. 217,068 particles were extracted for generating the initial model. The best model was made from 58,627 particles after several rounds of 3D classification and was then subjected to the 3D refinement with C1 symmetry and Bayesian polishing. A final 3.65 Å map was obtained based on the FSC using the 0.143 criterion. Local resolution estimation was also performed with the Bsoft package (Heymann, 2001).

Model building and refinement for mGlu1 structures
The apo form of mGlu5 (Koehl et al., 2019) (PDB code 6N52), the ligand-free form of mGlu1 VFT (Kunishima et al., 2000) (PDB code 1EWT), and NAM-bound mGlu1 7TM (Wu et al., 2014) (PDB code 4OR2) were used as the starting models for model building and refinement against the electron density map of apo mGlu1. L-Quisqualic acid, Nb43-bound mGlu5 (Koehl et al., 2019) (PDB code 6N51), and glutamate-bound mGlu1 VFT (Kunishima et al., 2000) (PDB code 1EWK) were used as the initial models for agonist-bound mGlu1. Models were docked into the EM density map using Chimera (Pettersen et al., 2004). Iterative manual adjustments and rebuilding were performed in COOT (Emsley et al., 2010) and phenix.real_space_refine in Phenix (Adams et al., 2010). The model statistics were validated using MolProbity (Chen et al., 2010). Structural figures were prepared in Chimera and PyMOL (http://www.pymol.org). The final refinement statistics are provided in Supplementary Table 1. The extent to which any model was overfitted during refinement was measured by refining the final model against one of the half-maps and by comparing the resulting map versus model FSC curves with the two half-maps and full model.

Intracellular calcium mobilization assay
CHO-K1 cells were seeded at a density of 10 6 cells in 10 cm dishes overnight at 37 ℃ in F-6 12 supplemented with 10% FBS. On the day of transfection, 3 μg DNA encoding the mGlu1 or its mutants was transfected to the cells by TransIT2020 (Mirus Bio). After 24 h, cells were trypsinized and seeded in black-sided, clear-bottomed 384-well plates (Greiner Bio-one) at a density of 15,000 cells per well.
On the day of measurement, growth medium was removed, and cells were loaded with 20 μL/well of 1x Fluo-4 Direct Calcium dye (Invitrogen) (prepared in HBSS buffer) and incubated at 37 °C in the dark. The FLIPR Tetra High Throughput Cellular Screening System (Molecular Devices) was programmed to initially take 10 readings (1 read per second) as a baseline before the addition of 10 μl of 3x drug solutions (prepared in HBSS buffer with 0.1% BSA). The fluorescence intensity was recorded for 2 min after drug addition to detect agonist activity. To measure PAM activity, cells were incubated at room temperature for 10 min followed by a second addition of 10 μL L-Quisqualic acid at a final concentration of 145 nM. Data were analyzed by nonlinear regression using GraphPad Prism 8.0.