Gq/11 family proteins.

The CCK1R is preferentially coupled to Gq/11 proteins. Few other structures of GPCRs (5HT_2AR, OX₂R, and M1 AChR) bound to Gq mimetic proteins have been determined [23,24,27], where Gq or G11 is also the primary transducer. In each of these receptors, TM6 remains in an extended helical conformation that is translated out at the intracellular base relative to inactive structures [23,24]; the G protein αH5 binds into the narrow cavity formed by this translational movement (S5 and S7 Figs). Overlay of the CCK1R/mGsqi complex structure with each of these other receptor–G protein complexes revealed good correlation in the location and orientation of the G protein and in the conformation of the far carboxyl-terminal residues of the αH5 (S7A–S7C Fig). The largest difference was observed between the CCK1R/mGqsi and the M1 AChR/G11/i complexes, where there both the α5 and αN helices were translated relative to each other (S7C Fig); nonetheless, the conformation of the G protein itself is very similar. The M1 AChR/G11/i is the only GPCR-Gq/11 complex solved where the Gα subunit is primarily the Gq/11 sequence, and this could contribute to the observed differences. However, it is important to note that in the 3DVA of the CCK1R/mGsqi cryo-EM data, all the conformations presented in the available consensus structures of the active M1 AChR, OX₂R, and 5HT_2AR are sampled (S2 and S4 Videos). As such, sample preparation and vitrification conditions could also contribute to the observed differences, in addition to the potential for real distinctions in the most stable conformations for each complex.

Gs proteins.

The Gs protein heterotrimer was the first transducer to be solved in complex with an activated GPCR [39], and multiple complexes with Gs have now been determined for class A and class B GPCRs where Gs is the primary transducer [7,10–12,40–44]. These structures have revealed both common and diverse features that govern Gs binding. In the inactive GDP-bound Gαs protein, the αH5 interacts with the αN helix and β1 sheet, with the far carboxyl terminus assuming the bulky “hook” conformation that is prototypical of the active G protein conformation (Fig 4E and 4F). In complex with activated GPCRs, the αH5 is rotated anti-clockwise and translated away from the nucleotide binding site (S5 Video). In most class A GPCR complexes, and all class B GPCR complexes, the bulky “hook” conformation is maintained with binding into the receptor intracellular binding cavity, primarily enabled by a large outward movement at the base of TM6 following receptor activation [7,10] (Fig 4). In prototypical receptors, such as the β-ARs or the adenosine A_2AR, this is facilitated by kinking of TM6 (Fig 4A, 4G and 4I), and this is even more marked in class B GPCRs [40] (Fig 4K and 4L). This was initially thought to be a required feature for Gs coupling, at least for receptors where Gs is the primary transducer [7,9,10]. Recent structures of class A GPCRs from different evolutionary subclades have demonstrated that these receptors can bind and activate Gs through alternative mechanisms. In both the bile acid receptor, GPBAR [12], and the prostanoid receptor, EP4R [11], TM6 is an unkinked, extended α-helix, similar to the TM6 conformation in structures of the active Gq/11 coupled receptors (Fig 4H and 4J, S7 Fig), with a correspondingly narrow pocket for binding of the carboxyl-terminal αH5. In the GPBAR–Gs complex, the αH5 is in the prototypic “hook” conformation, and this is accommodated by a shallower engagement with the receptor core that is nonetheless stabilised by more extensive interactions of the Ras domain with ICL3 and TM5 that has a markedly extended, stable, helix [12]; a similar elongated TM5 helix and extended interactions with the Ras domain is observed in the GPR52–Gs structure [41]. Nonetheless, the bulky “hook” conformation is maintained in these structures, suggesting that it is energetically favoured for Gs proteins. In contrast, in the CCK1R–Gs structure, the far carboxyl-terminal residues are “unwound” leading to a less bulky conformation, and the αH5 binds to the same depth as both the mGsqi protein in the CCK1R structure and the Gs protein in prototypical structures, with the carboxyl-terminal hydrophobic Leu residues oriented towards the lipid/detergent interface (Fig 4A, 4B and 4D). Intriguingly, in the EP4R that preferentially couples to Gs, unwinding of the “hook” also occurs (S5 Video), enabling the αH5 to bind to similar depth in the narrower receptor cavity, although here the carboxyl-terminal residues orient between the base of TM7 and TM1 [11] (Fig 4J). These data demonstrate that, while there is a common mode for Gs to engage with GPCRs, functional engagement can occur by multiple divergent mechanisms in a receptor-dependent manner for both canonically coupled GPCRs and those, like CCK1R, that preferentially engage other G protein subfamilies; these diverse modes of binding contribute to the difficulties in using bioinformatic and computational approaches to predict G protein selectivity. Further investigation will be required to understand GPCR–Gs interactions that can trigger destabilisation of the Gs αH5 “hook” conformation that enables novel modes of engagement with receptors; this will be necessary for prediction of functional Gs binding to GPCRs where limited outward movement of TM6 is an intrinsic structural feature.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Gαs adopts a carboxyl terminally extended conformation of the αH5 to enable binding to CCK1R.

(A–D). The far carboxyl terminus of the αH5 of Gαs forms a bulky “hook” conformation for binding to the prototypical Gs-coupled β2-AR (A–C) and occupies a large intracellular binding cavity relative to Gs-bound to the CCK1R that has a narrower binding cavity with binding facilitated by formation of an extended conformation of the Gαs carboxyl terminus (A, B, D). In the inactive Gαs-GDP, the Gs carboxyl terminus exhibits the “hook” conformation indicating that this is the principal ground state conformation (E, F). (G–L) Comparison of the conformation and orientation of Gαs when bound to CCK1R relative to class A, A2AR (G, PDB:6GDG), GPBAR (H, PDB:7CFM), β1-AR (I, PDB:7JJO), and EP4R [J, PDB:7D7M], or class B, GLP-1R (K, PDB:6X18) and SCTR (L, PDB:6WZG) that couple predominantly to Gs. Overall, the receptors engage Gs via wide outward movement of TM6 to accommodate the carboxyl-terminal hook motif. However, the GPBAR interacts with Gs in a narrower cavity through shallower engagement (H), while the EP4R also has a smaller intracellular pocket where Gs engagement is facilitated by unwinding of the far carboxyl terminus of the αH5. However, unlike CCK1R, the side chains are directed into the junction of the base of TM7 and TM1. αH5, α5 helix; CCK, cholecystokinin; CCK1R, cholecystokinin type 1 receptor; SCTR, secretin receptor.

https://doi.org/10.1371/journal.pbio.3001295.g004

Currently, only 1 other GPCR has been solved with 2 different G protein subfamily transducers: the GCGR bound to Gs or Gi [14]. For this receptor, the backbone conformation of the receptor was equivalent, regardless of the bound G protein. Here, the default conformation was the prototypical open pocket that facilitates Gs binding with the less bulky Gi1 protein binding into the same pocket with fewer interactions [14]. In the current study, the CCK1R also exhibited the same backbone conformation, regardless of the bound G protein, but instead this was enabled by conformational change within the Gs protein that allowed it to bind into the smaller CCK1R intracellular binding pocket. While much more work is required, the current data support the contention that the intrinsic conformational flexibility/stability of the receptor [9,45] is a critical feature in modes of G protein coupling for both primary and secondary transducers. The current work also provides supporting evidence for conformational flexibility of the G protein αH5 as a mechanism for promiscuous coupling.

Materials.

Peptide ligands, natural CCK-26-33 (CCK-8), and an analog for radiolabeling, D-Tyr-Gly-[(Nle^28,31) CCK-26-33], were synthesised in-house as previously described [47]. LANCE cAMP kit and sodium ¹²⁵Iodine were from Perkin Elmer Health Sciences (Shelton, Connecticut, United States of America). Fluo-8-AM was from AAT Bioquest (Sunnyvale, California, USA). Fetal Clone II was from Hyclone Laboratories (Logan, Utah, USA), and Dulbecco’s Modified Eagle Medium, glutamine, zeocin, and soybean trypsin inhibitor were from Thermo Fisher Scientific (Waltham, Massachusetts, USA). Methyl-β cyclodextrin, 3-isobutyl-1-methyl xanthine and poly-lysine were from Sigma-Aldrich (St Louis, Missouri, USA). All other reagents were of analytical grade.

Cell lines.

The HEK-293(S) parental cell line and its G protein (Gq, Gs) knockout lines (GsKO, GqKO) generated using the CRISPR/Cas-9 approach [48,49] were kindly provided by Dr. Asuka Inoue (Tohoku University, Miyagi, Japan). Each cell line was transfected with human type 1 CCK receptor using PEI [50], with clonal cell lines expressing similar levels of receptor selected using CCK-like radioligand binding [25]. Cells were grown in Dulbecco’s Modified Eagle Medium supplemented with 10% Fetal Clone II in a humidified atmosphere containing 5% CO₂ at 37°C and were passaged approximately twice a week.

Cholesterol modifications.

Cells had their cholesterol composition enhanced using MβCD–cholesterol inclusion complex [51].

Receptor binding assays.

CCK-like radioligand binding assays were performed with intact cells grown in poly-lysine coated 24-well tissue culture plates, as described [52]. Nonspecific binding was determined in the presence of 1 μM unlabeled CCK. All assays were performed in duplicate and repeated at least 3 times in independent experiments. Competition binding curves were analysed and plotted using the nonlinear regression analysis programme in the PRISM software suite v8.02 (GraphPad, San Diego, California, USA).

Intracellular calcium response assays.

CCK-stimulated intracellular calcium responses were quantified in intact cells using Fluo-8-AM using a FlexStation (Molecular Devices, Sunnyvale, California, USA), as described [52]. All assays were performed in duplicate and repeated at least 3 times in independent experiments. Calcium response curves were analysed and plotted as percentages of the maximal stimulation by 100 μM ATP using nonlinear regression analysis in the PRISM software suite v8.02.

Intracellular cAMP response assays.

CCK-stimulated intracellular cAMP responses were measured in intact cells in 384-well white Optiplates using LANCE cAMP kits, as described [53]. Assays were performed in duplicate and repeated at least 3 times in independent experiments. The cAMP concentration–response curves were analysed and plotted using the nonlinear regression analysis in PRISM v8.02.

Statistical comparisons.

Comparisons of experimental groups were assessed using analysis of variance (ANOVA) followed by Dunnett posttest, or using the Mann–Whitney test, as provided in PRISM 8.0 (GraphPad). The threshold for statistical significance was set at p < 0.05.

Constructs.

Human WT CCK1R was modified to include an N-terminal hemagglutinin (HA) signal sequence, FLAG-tag, 22 amino acids of the N-terminus of the M₄ muscarinic receptor to improve expression [28,29], a 3C-protease recognition sequence to remove the N-terminal modifications, and a carboxyl-terminal 8xhistadine (8xHis) tag (S1A Fig). This construct was cloned in to the pFastBac vector for insect cell expression and pcDNA3.1 for mammalian expression. A construct without modifications was also cloned into pcDNA3.1 for construct validation. Previously described constructs were used to express the heteromeric G protein in insect cells, including a dominant negative form of Gα_S (DNGα_S) and a dual expressing vector containing Gγ₂ and 8xHis tagged Gβ₁ [8,54]. 8xHis tagged Nb35 in pET20b was obtained from B. Kobilka [39]. For the CCK1R/mGsqi complex, the CCK1R was modified to include the HA signal sequence, N-terminal FLAG-tag, and carboxyl-terminal 3C-protease recognition site, 8xHis-tag and mGsqi (S1B Fig) [55].

Insect cell expression.

CCK1R, DNGα_S, and Gβ₁γ₂ were co-expressed in the Tni insect cell line (Expression Systems, Davis, CA, USA) using the Bac-to-bac baculovirus system (Invitrogen, Carlsbad, CA, USA). The Tni insect cells were grown to a density of 4 million cells/ml in ESF 921 serum free media (Expression Systems) before infection with a 4:2:1 ratio of CCK1R:DNGα_S:Gβ₁γ₂ baculoviruses. Insect cells were harvested by centrifugation at approximately 48 hours post infection, and cell pellets were stored at −80°C. For the CCK1R/miniGsqi similar methods were used, except a 4:1 ratio of CCK1R–mGsqi:Gβ₁γ₂ baculoviruses was used.

Nb35 and ScFv16 expression and purification.

Nb35 was expressed in the periplasm of BL21(DE3) Rosetta Escherichia coli cell line using an autoinduction method [56]. Transformed cells were grown at 37°C in a modified ZY media (50 mM Phosphate buffer pH 7.2, 2% tryptone, 0.5% yeast extract, 0.5% NaCl, 0.6% glycerol, 0.05% glucose and 0.2% lactose), in the presence of 100 μg/ml carbenicillin and 35 μg/ml chloramphenicol. At an OD₆₀₀ of 0.7, the temperature was changed to 20°C for approximately 16 hours before harvesting by centrifugation and storage at −80°C. NB35 was purified as described previously by extracting the periplasm supernatant and Ni-affinity chromatography [57].

ScFv16 was expressed in the Tni insect cell line as an excreted product using baculovirus. Media was harvested by centrifugation 72 hours post infection and chelating agents quenched by the addition of 5 mM CaCl₂. The media was separated by precipitation by centrifugation and batch bound to EDTA-resistant Ni-Sepharose resin. The column was washed with a high salt buffer (20 mM HEPES pH 7.2, 500 mM NaCl, and 20 mM imidazole), followed by a low salt buffer (20 mM HEPES pH 7.2, 500 mM NaCl, and 20 mM imidazole), before elution in 20 mM Hepes pH 7.5, 100 mM NaCl, and 250 mM imidazole. The eluted protein was dialysed to 20 mM Hepes pH 7.5 and 100 mM NaCl before storage at −80°C.

CCK1R/CCK-.8/DNGα_S complex purification.

Insect cell pellet from 1 L of culture was thawed and solubilised in 20 mM HEPES pH 7.4, 100 mM NaCl, 5 mM MgCl₂, 5 mM CaCl₂, 0.5% LMNG (Anatrace, Maumee, OH, USA), 0.03% CHS (Anatrace), and a Roche cOmplete Protease Inhibitor Cocktail tablet (Sigma-Aldrich, North Ryde, New South Wales, Australia). The resuspended pellet was homogenised in a Dounce homogeniser and complex formation initiated by the addition of 10 μM CCK-8, 5 μg/ml NB35 and 25 mU/ml apyrase (NEB, Notting Hill, Victoria, Australia). The solubilised complex was incubated with stirring at 4°C for 2 hours before insoluble material was removed by centrifugation at 30,000 g for 30 minutes. The solubilised complex was batch bound to equilibrated M1 anti-Flag affinity resin, rotating at 4°C for 1 hour. The resin was packed into a glass column and washed with 20 column volumes of 20 mM HEPES pH 7.4, 100 mM NaCl, 5 mM MgCl₂, 0.01% LMNG, 0.0006% CHS and 1 μM CCK-8 in the presence of 5 mM CaCl₂, before elution in the same buffer with 5 mM EGTA and 0.1 mg/ml Flag peptide. Eluted complex was concentrated to less than 500 μL in an Amicon Ultra-15 100 kDa molecular mass cut-off centrifugal filter unit (Millipore, Burlington, MA, USA) and further purified by SEC on a Superdex 200 Increase 10/300 GL (Cytiva, Marlborough, MA, USA) with 20 mM HEPES pH 7.4, 100 mM NaCl, 5 mM MgCl₂, 0.01% LMNG, 0.0006% CHS, and 1 μM CCK-8. Eluted fractions were containing complex were pooled and concentrated to 4.1 mg/ml before being flash frozen in liquid nitrogen and stored at −80°C.

CCK1R/CCK-8/mGsqi complex purification.

Insect cell pellet from 1.5 L of culture was thawed and resuspended in a hypotonic solution of 20 mM Hepes pH 7.4, 5 mM MgCl₂, and 1 μM CCK-8. The cells were stirred at room temperature for 15 minutes to allow lysis to occur. Cell pellet was collected by centrifugation and solubilised in 20 mM HEPES pH 7.4, 100 mM NaCl, 5 mM MgCl₂, 5 mM CaCl₂, 0.5% LMNG (Anatrace), and 0.03% CHS (Anatrace) and a Roche cOmplete Protease Inhibitor Cocktail tablet (Sigma-Aldrich). The resuspended pellet was homogenised in a Dounce homogeniser and complex formation initiated by the addition of 10 μM CCK-8, 5 μg/ml ScFv16 and 25 mU/ml apyrase (NEB). The solubilised complex was incubated with stirring at 4°C for 2 hours before insoluble material was removed by centrifugation at 30,000 g for 30 minutes. The solubilised complex was batch bound to equilibrated M1 anti-Flag affinity resin, rotating at 4°C for 1 hour. The resin was packed into a glass column and washed with 20 column volumes of 20 mM HEPES pH 7.4, 100 mM NaCl, 5 mM MgCl₂, 0.01% LMNG, 0.0006% CHS, and 1 μM CCK-8 in the presence of 5 mM CaCl₂, before elution in the same buffer with 5 mM EGTA and 0.1 mg/ml Flag peptide. Eluted complex was concentrated to less than 500 μL in an Amicon Ultra-15 100 kDa molecular mass cut-off centrifugal filter unit (Millipore) and further purified by SEC on a Superdex 200 Increase 10/300 GL (Cytiva) with 20 mM HEPES pH 7.4, 100 mM NaCl, 5 mM MgCl₂, 0.01% LMNG, 0.0006% CHS, and 1 μM CCK-8. The sample was then cleaved with 3C-protease at room temperature for 1 hour and reapplied to SEC. Eluted samples were pooled and concentrated to 4 mg/ml before being flash frozen and stored at −80°C.

SDS-PAGE and western blot.

Purified samples were analysed by SDS-PAGE and western blot. Samples were applied to precast TGX gels (Bio-Rad, Hercules, CA, USA) before staining with InstantBlue Coomassie stain (Sigma-Aldrich) or immediately transferred to PVDF membrane (Bio-Rad) for western blot analysis. Western blots were stained with primary rabbit anti-GNAS (Novus Biologicals, Littleton, CO, USA NBP1-31730), primary mouse anti-His antibody (Qiagen, Hilden, Germany, 34660), secondary goat anti-rabbit 800CW antibody (LI-COR Biosciences, Lincoln, NE, USA, 926–32211), secondary anti-mouse 680CW antibody (LI-COR Biosciences, 926–68070) and an in-house made Alexa Fluor 488 conjugated mouse anti-flag antibody. Western blots were imaged on a Typhoon 5 imaging system (Amersham, Little Chalfont, UK).

Vitrified sample preparation and data collection.

Samples (3 μL) were applied to a glow-discharged UltrAufoil R1.2/1.3 300 mesh holey grid (Quantifoil GmbH, Großlöbichau, Germany) and were flash frozen in liquid ethane using the Vitrobot mark IV (Thermo Fisher Scientific) set at 100% humidity and 4°C for the prep chamber with a blot time of 10 seconds. Data were collected on Titan Krios microscope (Thermo Fisher Scientific) operated at an accelerating voltage of 300 kV with a 50 μm C2 aperture with no objective aperture inserted and at an indicated magnification of 130kX in nanoprobe TEM mode. A Gatan K3 direct electron detector positioned post a Gatan BioQuantum energy filter (Gatan, Pleasanton, California, USA), operated in a zero-energy-loss mode with a slit width of 15 eV, was used to acquire dose fractionated images of the samples. Movies were recorded in hardware-binned mode (previously called counted mode on the K2 camera) with the experimental parameters listed in S1 Table using 18-position beam-image shift acquisition pattern by custom scripts in SerialEM [58].

CCK1R/DNGs/CCK-8.

A total of 7,146 micrographs were motion corrected using UCSF MotionCor2 [59] and dose weighted averages had their CTF parameters estimated using CTFFIND 4.1.8 [60]. Particles were picked using the crYOLO software package [61] on a pretrained set of weights for GPCRs yielding 6.4 M particle positions. These particles were extracted from the micrographs and subjected to 2D classification and ab initio 3D and 3D refinement in the cryoSPARC (v3.1) software package [62], which resulted in a homogeneous well-centred particle stack containing 1 M particles. These were then fed into the RELION (v 3.1) software package for further rounds of 2D and 3D classification, which led to 800 k particles for initial 3D refinement in RELION. These particles where then polished in RELION and underwent a further round of 3D classification and CTF envelope fitting. A final consensus 3D refinement using 643 k particles was performed in RELION. The maximisation step was carried out using the SIDESPLITTER algorithm [63] to yield a final resolution of 1.95 Å (FSC = 0.143, gold standard) for the consensus map. Further, receptor-focused refinements were performed using a mask generated from an initial PDB model and searching a local 1.8 degree Euler angle space.

CCK1R/mGsqi/CCK-8.

A total of 7,182 micrographs were motion corrected using UCSF Motioncor2 and dose weighted averages had their CTF parameters estimated using CTFFIND 4.1.8. Particles were picked using the automated template picking routine in RELION 3.1. Moreover, 2.3 M particles were extracted and cryoSPARC was employed to perform 2D classification, ab initio 3D model generation and initial 3D refinement. The resulting particle stack contained 441 k particles, which were then polished in RELION 3.1. The polished particle stack was then fed back into the cryoSPARC software package for a non-uniform 3D consensus refinement and CTF envelope fitting which yielded a 2.44 Å resolution map (FSC = 0.143, gold standard). Due to a large amount of conformational flexibility between the receptor and G proteins, further local refinements in cryoSPARC were used to calculate high quality maps of either the CCK1 receptor (2.57 Å) or the mGsqi G protein complex (2.43 Å) which were used to generate a PDB model.

CCK1R/CCK-8/DNGs model building.

An initial homology model of CCK1R was created using SWISS-MODEL [64] using the active structure of the serotonin 5-HT1B receptor as a template (PDB ID: 6G79). The CCK1R model was placed into the receptor-focused cryo-EM map by the MDFF routine in namd2 [65]. The CCK-8 peptide was manually built using COOT [66], the sulfotyrosine residue was imported from the monomer library (TYS), and geometry restrains were generated with eLBOW within PHENIX [67]. The model was refined with repeated rounds of manual model building in COOT and real-space refinement within PHENIX [68]. The G protein (DNGα_S/Gβ₁/Gγ₂) and NB35 from the structure of GLP-1 receptor-Gs complex (PDB ID: 6B3J) was rigid body placed into the consensus cryo-EM map using the map fitting tool of UCSF ChimeraX [69]. The G protein was further refined to the consensus cryo-EM map with repeated rounds of manual model building in COOT and real-space refinement within PHENIX. Lastly the CCK1R/CCK-8 model was combined with the G protein and real-space refined to the consensus cryo-EM map using PHENIX. The model quality was assessed using MolProbity [70] before PDB deposition.

CCK1R/CCK-8/mGsqi model building.

The higher resolution model of CCK1R/CCK-8 obtained from the Gα_S structure was rigid-body placed into a receptor-focused cryo-EM map using ChimeraX. The model was refined with repeated rounds of manual model building in COOT and real-space refinement within PHENIX. The β₁ and γ₂ were obtained from the higher resolution Gα_S structure, the miniGsqi was modified from the OX₂R structure (PDB ID: 7L1U), and the subunits were rigid-body placed in the G protein–focused map. The model was refined with repeated rounds of manual model building in COOT and real-space refinement within PHENIX [68]. Lastly, the CCK1R/CCK-8 model was combined with the G protein and real-space refined to the consensus cryo-EM map using PHENIX. The model quality was assessed using MolProbity [70] before PDB deposition. The scFv16 was not modelled due to poor side-chain density.

Model interaction analysis.

Interactions and hydrogen bonds were analysed using UCSF chimeraX package, with relaxed distance and angle criteria (0.4 Å and 20° tolerance, respectively). The interfaces were further analysed using PDBePISA [71]. Figures were generated using UCSF chimeraX and PyMOL Molecular Graphics System, Version 2.3 (Schrödinger, New York, NY, USA).

3DVA.

3DVA was performed using the cryoSPARC software package, based on the consensus refinements of the complexes. For the CCK1R–mGsqi data, 6 principal components (Comp0 to Comp5) were used, and for the CCK1R–Gs data, 5 principal components (Comp0 to Comp4) were used to analyse the principal motions in the cryoEM data, each separated into 20 frames (frames 0 to 19). Cryosparc 3DVA outputs were used for dynamic analyses of the CCK complexes and visualised using the command vseries as implemented in ChimeraX [69]. The backbones of the consensus refinement models were flexibly fitted into the frames 0 and 19 density maps of the 3DVA principal components using Isolde [72], implemented in ChimeraX [69]. Morphs of models (aligned using the matchmaker command, or aligned by densities) were created in Chimera and ChimeraX [69,73]. Movie editing was performed using Adobe Premiere Pro 2020.