Introduction

G-protein-coupled receptors (GPCRs) are the largest superfamily of membrane proteins and are involved in virtually all physiological phenomena1,2. GPCRs regulate intracellular pathways by binding extracellular ligands and transmitting signals across the plasma membrane to intracellular effector molecules via heterotrimeric G proteins composed of Gα, Gβ, and Gγ subunits3. The 16 Gα subtypes in human cells are classified into four Gα subfamilies (Gαs, Gαi, Gαq, and Gα12) that transmit unique signals (e.g., cAMP, Ca2+, inositol phosphate, and Rho activation) and Gα subtypes belonging to each of these subfamilies have been thought to exert similar functions1. However, recent studies have revealed that subtypes belonging to the same subfamily have different patterns of coupling to GPCRs and exert distinct functions, requiring further analysis focusing on each subtype4,5,6.

Gα12 and Gα13, which belong to the G12 subfamily, activate the small GTPase Rho through Rho-specific guanine nucleotide exchange factors7. Some reports indicate that Gα12 and Gα13 are functionally different, interacting with different effector target proteins leading to different outcomes. For example, Gα13-knockout mice are embryonic lethal due to defective angiogenesis, whereas Gα12 knockout mice develop normally despite the expression of both subtypes in the vascular endothelial cells8,9,10. This phenotypic difference has been attributed to myocyte-specific enhancer factor-2 (MEF2), which is specifically activated downstream of Gα1311. Several other effector proteins, such as Hsp90 and PYK2, have been identified to bind specifically to Gα12 or Gα13, respectively12,13.

Some GPCRs selectively couple to either Gα12 or Gα13, indicating that they are used effectively in various physiological conditions. In our previous study, we generated a dataset quantifying the coupling of 11 chimeric Gα proteins including Gα12 and Gα13 to 148 GPCRs by using TGFα shedding assay5. A meta-analysis integrating this dataset with the Guide to Pharmacology and the Bouvier datasets showed that 25 GPCRs were found to be coupled to both Gα12 and Gα13, 5 GPCRs specifically to Gα12, and 8 GPCRs specifically to Gα135,14,15,16. However, it remains uncertain how GPCRs interact with Gα12 or Gα13 and what determines their coupling selectivity.

The C-terminal helix (α5 helix) of the Gα subunits is an important determinant for coupling selectivity. The α5 helix consists of 26 amino acids and interacts with the transmembrane core of GPCRs, occupying > 70% of the GPCR–G protein interface. Structural studies of the GPCR–Gα protein complex and cell-based assays confirmed the role of the α5 helix in determining Gα coupling selectivity17,18,19. In GPR35, out of the eight amino acids of the α5 helix that differ between Gα12 and Gα13, G.H5.23, the fourth residue from the C-terminus of the Gα subunit, contributes to its Gα13 preference over Gα1220. Thus, focusing on the interaction between a receptor and the α5 helix would help determine the key elements regulating the coupling selectivity between Gα12 and Gα13.

We have previously generated a Gα12-coupled Designer Receptor Exclusively Activated by Designer Drugs (DREADD) through a series of modifications. The first generation of Gα12-coupled DREADD was designed by modifying Gαq-coupled DREADD (M3D) based on the TGFα shedding assay database5,21. Subsequently, an F931.57V mutation was introduced into the first-generation DREADD to create a second-generation Gα12-DREADD (G12D), which couples strongly to Gα12 but weakly to Gα1322. In the present study, we used the second generation G12D as an experimental model for analyzing the mechanism and determinants of Gα12-vs-Gα13-coupling selectivity.

In this study, by utilizing chimeric Gα proteins, we identified the specific residues in the α5 helix of Gα12 responsible for Gα12 selectivity. Through molecular dynamics analysis, we pinpointed residues in G12D that involve interactions with the selectivity determinants in the α5 helix. Finally, by mutating G12D, we successfully developed a Gα12/13-DREADD that increased both Gα12/13 coupling to similar extents.

Results

The Gα12 selectivity of G12D requires specific amino acid residues in the α5 helix including isoleucine G.H5.23

To determine which amino acid residues in the α5 helix are critical for preferential coupling for Gα12 over Gα13 by G12D, we constructed chimeric Gα proteins, in which the α5 helix (G.H5.01–G.H5.26) of Gαq was replaced by the α5 helix of Gα12 or Gα13 (Gαq-12C or Gαq-13C, respectively). By following a previous strategy5 where we introduced a chimeric Gα subunit in the Gαq/Gα11/Gα12/Gα13 quadruple-deficient HEK293 cells (∆Gq/∆G12 cells), we herein co-expressed Gαq-12C or Gαq-13C along with G12D and a reporter gene encoding alkaline phosphatase-tagged TGFα (AP-TGFα) in ∆Gq/∆G12 cells. Cells were stimulated with increasing concentrations of clozapine-N-oxide (CNO) and the relative coupling of Gαq-12C and Gαq-13C to G12D was evaluated by measuring the activity of cleaved AP-TGFα (Fig. 1A,B). The relative intrinsic activities (RAi) of Gαq-13C to Gαq-12C was calculated from the concentration–response curves in the TGFα shedding assay5,23. This parameter analysis showed that the RAi was 4.6-fold higher in cells expressing Gαq-12C than in cells expressing Gαq-13C, confirming a previous report that G12D preferentially couples with Gαq-12C5 (Fig. 1C). The amino acid sequences of the α5 helices of Gα12 and Gα13 differ at the following positions: G.H5.04, G.H5.06, G.H5.09, G.H5.10, G.H5.17, G.H5.18, G.H5.22, and G.H5.23 (Fig. 1D). To determine which of these amino acid differences confer coupling selectivity, we constructed eight pairs of chimeric Gα protein mutants. Specifically, the amino acid at one of the eight variable positions in the α5 helix was switched from its Gα12 residue to its Gα13 residue or vice versa, while the rest of the Gαq-12C and Gαq-13C protein sequences remained unchanged (Fig. 1E, Supplementary Fig. 1). We found that coupling selectivity was reversed when the amino acid at the position G.H5.23 was reversed; thus, Gαq-13C carrying a leucine to isoleucine mutation (i.e., Gαq-13C [LG.H5.23I]) coupled preferentially to G12D, while the reverse mutants, Gαq-12C (IG.H5.23L), did not (Fig. 1F,G, Supplementary Fig. 2). In addition, Gαq-13C (RG.H5.09H) enhanced coupling with G12D, while Gαq-12C (HG.H5.09R) tended to decrease it (Fig. 1F,G, Supplementary Fig. 2). These results indicate that Gα12 and Gα13 coupling specificity is attributable to the specific residues in the α5 helix, most notably at the position G.H5.23 (isoleucine and leucine, respectively).

Figure 1
figure 1

Identification of residues in the α5 helix involved in the Gα12-vs-Gα13-coupling selectivity by the TGFα shedding assay. (A) Schematic representation of the TGFα shedding assay for evaluating coupling between G12D and Gαq-12C or Gαq-13C. G12D, chimeric Gα (Gαq-12C or Gαq-13C), and AP-TGFα are transiently expressed in ∆Gq/∆G12 cells. Upon stimulation with CNO, G12D is coupled with the chimeric Gα, of which activation leads to ectodomain shedding of the AP-TGFα reporter. Cleaved AP-TGFα is quantified by measuring AP activity in the conditioned media based on the production of para-nitrophenol (p-NP) from para-nitrophenyl phosphate (p-NPP). See the method section for detail. (B) Concentration–response curve for the TGFα shedding responses induced by G12D activation upon CNO stimulation. The vehicle-treated condition is set as the baseline. The magenta and cyan circles represent the response under conditions expressing Gαq-12C and Gαq-13C, respectively. (C) Logarithmic values of relative intrinsic activity (RAi) calculated from (B) using the formula shown in the figure. The values were normalized by the Gαq-12C expressing condition. Bars and error bars represent the mean and SEM, respectively, for three independent experiments with each dot representing an individual experiment. ** represents P < 0.01 with the two-tailed t-test. (D) Amino acid sequence of the α5 helices of Gα12 and Gα13. Positions where amino acids differ between Gα12 and Gα13 are highlighted in magenta and cyan, respectively. (E) Design of swapped mutants in the Gαq-12C and the Gαq-13C backbones. Mutants in positions G.H5.09, G.H5.10, G.H5.17, G.H5.18, and G.H5.22 are omitted. (F) Concentration–response curve for the TGFα shedding responses of the swapped mutants. The dashed lines represent the responses of Gαq-12C (magenta) and Gαq-13C (cyan) shown in (B). In all panels, the symbols and error bars represent the mean and SEM, respectively, for three independent experiments performed in triplicate. For many data points, the error bars are smaller than the symbols and, thus, are not visible. (G) Logarithmic values of RAi of the swapped mutants calculated from (F). Bars and error bars represent the mean and SEM, respectively, for three independent experiments with each dot representing an individual experiment. ns, *, ** and *** represent P > 0.05, < 0.05, < 0.01 and < 0.001, respectively, with two-way ANOVA, followed by the Dunnett’s multiple comparisons test (Gαq-12 mutants with wild-type Gαq-12C; Gαq-13 mutants with wild-type Gαq-13C).

Full size image

MD simulations reveal interactions involving Gα12 selectivity of G12D

To gain insight into the Gα12 selectivity of G12D, we performed molecular dynamics (MD) simulations of the G12D–Gα12 and the G12D–Gα13 complexes. The two complexes were modeled with SWISS-MODEL and I-TASSER structure prediction software using M1R–Gα11 complex structure (PDB: 6OIJ) as a template and embedded into lipid bilayers using CHARMM-GUI24,25,26,27 (Fig. 2A). Both models were simulated for 500 ns using a conventional MD protocol to estimate the Poisson-Boltzmann-based energy of the Gα protein to the receptor as described previously28. Because calculations of the G12D–Gα12 total energy indicated that low energy states were not conformationally sampled, 500-ns Gaussian Accelerated Molecular Dynamics (GaMD) simulations were performed29. The GaMD snapshots of the G12D–Gα12 complex clustered into two distinct conformations and the snapshots of the G12D–Gα13 complex clustered into five distinct conformations, indicating that the latter complex was more dynamic. The average structure that best represented each cluster was then chosen as the starting point for a final 500-ns conventional MD allowing for unbiased free energy calculations and receptor–G protein interaction analyses. For the G12D–Gα12 complex, most structural variation between the two clusters was in the internal dynamics of the Gα12 protein and TM6 of the receptor. For the G12D–Gα13 complex, most structural variation among the five clusters was in the internal dynamics of the Gα13 protein, the positioning of its α5 helix, and the intracellular facing regions of TM5/TM6 domains of the receptor.

Figure 2
figure 2

Comparison of the stability between the G12D–Gα12 and the G12D–Gα13 complexes through MD simulations. (A) Schematic representation of the molecular dynamics simulations. On the right side, the occupancy (%) of each cluster and the ∆G (kcal/mol) calculated from the cMD results are displayed. (B) Overlap of snapshots every 20 ns from 500-ns cMD simulations starting from each cluster 1 (C1). For clarity, G12D is shown with the average structure of each simulation and residues other than the α5 helix are hidden. (C) RMSD for the α5 helix calculated from 500-ns cMD simulations starting from C1.

Full size image

The results of the MD simulations revealed significant differences in the stability of Gα12 and Gα13 interaction with G12D. From the MD trajectories, we calculated the total energy (∆G) of Gα12 and Gα13 through MMPBSA analysis30. We found that ∆G was weaker (less negative) in the Gα13 simulation than in the Gα12 simulation, indicating that the interaction between G12D and Gα13 is less stable than that between G12D and Gα12 (Fig. 2A). We also observed that the position in the α5 helix of Gα13 in the transmembrane core of G12D was variable, while its position in Gα12 was more stable and essentially stayed in the transmembrane core of G12D (Fig. 2B). Indeed, the root-mean-square deviation (RMSD) for the α5 helix was higher in the Gα13 simulation than in the Gα12 simulation for the most populated cluster (C1; Fig. 2C). RMSD analysis for the α5 helix for all clusters showed that, compared with the C1 cluster of the G12D–Gα12 complex, all of the other clusters including those of the G12D–Gα13 complex were sampled more frequently in positions with RMSD scores beyond 4 Å (Supplementary Fig. 3A). These observations are consistent with the results of cell-based assays (Fig. 1F,G).

The most notable differences in the simulations of the G12D–Gα12 and the G12D–Gα13 complexes were in ICL2. More specifically, when comparing the average structures of the two complexes, the position of ICL2 was tilted 3.5 Å more toward the α5 helix in the G12D–Gα12 complex than in the G12D–Gα13 complex (Fig. 3A). Among residues in the ICL2, Y17634.53 showed the most noticeable difference between the simulations (Fig. 3B). In the simulation of the G12D–Gα12 complex, Y17634.53 was positioned between N1032.39 and S1693.53 and oriented toward Gα12 residue IG.H5.23, such that Y17634.53, N1032.39, and S1693.53 form a pocket for the methyl group of IG.H5.23 (Fig. 3B,C, Supplementary Fig. 3B). These observations show that the transmembrane core of G12D contains a pocket that accommodates IG.H5.23, and this likely stabilizes the G12D–Gα12 complex in a low energy state. On the other hand, in the α5 helix of Gα13, LG.H5.23 appeared disoriented and tilted toward TM5 and TM6 (Fig. 3D), while G12D Y17634.53 was oriented upward and failed to form a pocket with N1032.39 and S1693.53, likely disfavoring an interaction between these residues and Gα13 LG.H5.23 (Fig. 3D, Supplementary Fig. 3B). Together, these observations suggest that the pocket formed by G12D residues N1032.39, S1693.53 and Y17634.53 plays a significant role in the preferential coupling between G12D and Gα12.

Figure 3
figure 3

Structural comparison of the G12D–Gα12 and the G12D–Gα13 complexes through MD simulations. (A) A comparison of ICL2 structures between the G12D–Gα12 and the G12D–Gα13 complexes. Shown are representative structures during the 500-ns cMD simulations of the G12D complexes, each started from the most populated clusters (C1) observed in the GaMD simulations. These structures were aligned with G12D transmembrane bundles. In the G12D–Gα12 complex, G12D and Gα12 are shown in orange and magenta, respectively. In the G12D–Gα13 complex, G12D and Gα12 are shown in yellow and cyan, respectively. (B) A comparison of the positions of Y17634.53 between the G12D–Gα12 and the G12D–Gα13 complexes. (C,D) Detailed positions of Gα12 IleG.H5.23 (C) and Gα13 LeuG.H5.23 (D) in the transmembrane core of G12D. (E) A comparison of ICL3 structures between the G12D–Gα12 and the G12D–Gα13 complexes. (F,G) Detailed orientations of G.H5.09 residue.

Full size image

In the G12D structures, we found another difference in the region composed of the C-terminal portion of the TM5, ICL3 and the N-terminal portion of TM6 (TM5–ICL3–TM6). In the simulation of the G12D–Gα12 complex, the TM5–ICL3–TM6 region was located farther from the transmembrane core side than in the Gα13 simulation (Fig. 3E). In general, the active form of GPCR displays an outward shift of TM5–ICL3–TM6, which facilitates the accommodation of α5 helix17. Therefore, the TM5–ICL3–TM6 opening observed in the G12D–Gα12 complex likely reflects a favorable form for coupling. In the simulations, TM5–ICL3–TM6 was in the proximity of the G.H5.09 residue position out of the eight different residues in the α5 helices of Gα12 and Gα13. Consistent with the observations, in the mutant experiment in Fig. 1E,F, the mutants in which these positions of Gα13 were swapped for Gα12 residues showed enhanced coupling to G12D, indicating their involvement in the selectivity. In the simulations, the G.H5.09 residue appeared to be closest to Q2645.71 in G12D (Fig. 3F,G). Therefore, Q2645.71, in addition to the previously mentioned N1032.39, S1693.53 and Y17634.53, are potential residues involved in the selectivity.

N1032.39, S1693.53, Y17634.53 and Q2645.71 in G12D collectively contribute to the Gα12-vs-Gα13 selectivity

To examine the importance N1032.39, S1693.53, Y17634.53 and Q2645.71 in the Gα12-vs-Gα13-coupling selectivity as well as to generate a G12D mutant with enhanced Gα13 coupling, we next performed mutation studies in G12D. We engineered 19 amino acid substitution mutations at these positions and examined the coupling of Gα12 and Gα13 for a total of 76 mutants. Each G12D mutant carried an N-terminal FLAG-epitope tag to allow cell surface expression to be quantified by flow cytometry. Most mutants carrying substitutions of N1032.39 and Y17634.53 were expressed at a lower level than the wild-type (WT) control, while expression of mutants carrying substitutions of S1693.53 and Q2645.71 showed relatively smaller changes in expression level (Supplementary Fig. 4). Next, the Gα12-vs-Gα13 selectivity of these mutants was assessed using the TGFα shedding assay. The RAi values were calculated using the equation shown in Fig. 4A, offering indexes into the extent to which the mutants enhanced (or reduced) coupling to Gαq-12C and Gαq-13C relative to the wild-type G12D (Fig. 4B,C, Supplementary Fig. 5–9). G12D mutants — with substitutions at N1032.39 with R, K or P; Y17634.53 with H, S, A or W; or Q2645.71 with R or P — showed higher RAi and stronger coupling to Gαq-12C and Gαq-13C than WT. In particular, Y17634.53H and Y17634.53W greatly enhanced coupling to Gαq-13C, and this enhancement effect was stronger than that for Gαq-12C. None of the S1693.53 mutations enhanced coupling to Gαq-13C, whereas S1693.53R and S1693.53K selectively enhanced coupling to Gαq-12C. Although these mutants of G12D display varying degrees of selective coupling to Gα12 or Gα13, the results of these experiments confirm the importance of N1032.39, S1693.53, Y17634.53, Q2645.71 as determinants of G12D coupling.

Figure 4
figure 4

Determining G12D residues involved in the Gα12-vs-Gα13 selectivity. (A) Calculation method for RAi of the G12D mutants. (B) Logarithmic values of RAi of the G12D mutants. ∆Gq/∆G12 cells transiently expressing the indicated G12D mutants along with AP-TGFα and Gαq-12C (or Gαq-13C) (Gαq-12C-expressing condition and Gαq-13C-expressing condition are shown as magenta and cyan, respectively). The cell-surface expression levels are shown in gray with different intensities to allow comparison with the wild-type G12D. WT (1:2) to (1:20) denote twofold to 20-fold less volumes, respectively, of transfected plasmids than those of the mutant plasmids. Bars and error bars represent the mean and SEM, respectively, for 3–15 independent experiments with each dot representing an individual experiment. The data for WT and its dilutions are reused in all panels. (C) Concentration–response curve for the TGFα shedding responses of the representative G12D mutants. The dashed lines represent the responses of the wild-type G12D. In all panels, the symbols and error bars represent the mean and SEM, respectively, for three independent experiments. For many data points, the error bars are smaller than the symbols and, thus, are not visible.

Full size image

To further enhance coupling between Gα13 and G12D, we introduced additional mutations on top of the Y17634.53H mutant, which demonstrated the highest Gα13-coupling capability in both RAi and Emax parameters (Fig. 4B,C, Supplementary Fig. 5–9). We chose N1032.39P and Q2645.71R because they enhance coupling to Gαq-13C (Fig. 4B,C, Supplementary Fig. 5–9). Compared to Y17634.53H, both N1032.39R/Y17634.51H and Y17634.53H/Q2645.71R showed increased RAi values under both Gαq-12C and Gαq-13C expressing conditions. Furthermore, Y17634.53H/Q2645.71R also exhibited an increasing effect on Emax (Fig. 5A,B, Supplementary Fig. 10). Based on these results, we designated Y17634.53H/Q2645.71R double mutant, which exhibits strong coupling to the α5 helices of Gα12 and Gα13, as Gα12/13-DREADD (Fig. 5C).

Figure 5
figure 5

Development of Gα12/13-DREADD. (A) Logarithmic values of RAi of the G12D mutants (Gαq-12C expressing condition = magenta, Gαq-13C expressing condition = cyan). The cell-surface expression levels of the mutants are shown in gray with different intensities to allow comparison with the wild-type G12D. WT (1:2) to (1:20) denote twofold to 20-fold less volumes of transfected plasmids, respectively, than those of the mutant plasmids. Bars and error bars represent the mean and SEM, respectively, for three independent experiments performed in triplicate. The data for WT and its dilutions are reused from Fig. 4. (B) Concentration–response curve for the TGFα shedding responses of the representative G12D mutants. The dashed lines and solid lines represent the responses of the wild-type G12D and Y17634.53H, respectively. In all panels, the symbols and error bars represent the mean and SEM, respectively, for three independent experiments performed in triplicate. For many data points, the error bars are smaller than the symbols and, thus, are not visible. (C) A schematic model of the Y17634.53H/Q2645.71R mutant showing potent coupling to both Gα12 and Gα13 via recognition of the three residues in the α5 helix.

Full size image

Discussion

In this study, we showed that IG.H5.23 at position G.H5.23 in the α5 helix of Gα12 is a major determinant of selective coupling between Gα12 and G12D. Conversely, LG.H5.23 in Gα13 has been reported to contribute to the Gα13 selectivity of GPR3520. These findings support the notion that the interaction between G.H5.23 in the α5 helix and target GPCRs is critical for the Gα12-vs-Gα13 coupling selectivity. Structural analyses of S1PR2–Gα13 and GPR35–Gα13 complexes have revealed that L G.H5.23 is oriented toward TM2 and TM3 and is accommodated by a pocket formed by residues 2.39, 3.53, and 34.5331,32. In particular, in the S1PR2–Gα13 complex, ICL2 lies inside the transmembrane core, highlighting the importance of 34.5331. These studies support the accuracy of our MD-simulated structures of G12D–Gα12 and G12D–Gα13 complexes. Notably, the molecular properties of isoleucine and leucine are very similar except for the different positions of the methyl groups in the amino acid side chains, and yet GPCRs distinguish one from the other. This is likely achieved by how well the pocket in the GPCR transmembrane core accommodates the isoleucine versus the leucine side chain. In the MD simulations described in the present study, the methyl group of Gα13 LG.H5.23 is predicted to encounter steric hindrance in the G12D pocket formed by TM2, TM3, and ICL2, which would create instability in the G12D–Gα13 complex. This idea is consistent with the fact that amino acid substitutions that altered the shape of the transmembrane core of G12D enhanced coupling to Gα13, presumably by creating a pocket that is optimized to accommodate LG.H5.23.

One caveat of this study is that it only identifies determinants of GPCR coupling selectivity in the 26-amino-acid α5 helix of Gα12 and Gα13. The rationale for this limitation is that prior studies using chimeric G proteins revealed a crucial role of the α5 helix, especially G.H5.23 and G.H5.24 positions, for determining coupling selectivity18,19. By focusing here on the α5 helix, we avoid difficulties arising from differences in the efficiency of the TGFα shedding assay as a surrogate measure of Gα12- and Gα13-mediated signaling. However, additional determinants of G protein coupling specificity exist in other protein regions and structures such as the αN–β1 hinge and the β2–β3 loop33,34,35. To fully understand the mechanisms underlying Gα12-vs-Gα13-coupling selectivity, a broad examination of determinants of coupling selectivity linked to these elements will be needed in the future. Another caveat is that the present study is based on structural models instead of experimentally verified protein structural data. That said, the full Gα subunit was used in the simulations to ensure that the α5 helix had the proper local environment for positioning relative to the receptor. In addition, to ensure the accuracy of the modeled structures, we performed GaMD followed by clustering before the production runs to ensure that we captured stable conformations of the receptor and the Gα proteins in the modeled complexes. The resulting Gα12 and Gα13 simulations showed significant differences in stability. In the future, once the structures of GPCR–Gα12 or GPCR–Gα13 complexes are resolved, they can be used to validate and/or improve upon the simulations presented here.

DREADDs that couple specifically to one or more Gα proteins were designed for and have been useful in drug discovery research36. Gαq-DREADD and Gαi-DREADD have been universally used to regulate neuronal activation37,38. DREADDs have also been used to analyze G protein signaling in a particular cell population and to predict the effects of agonists that activate specific G-protein-signaling pathways. Administration of CNO to mice genetically modified to express DREADD in a cell type-specific fashion allows for spatio-temporal regulation of specific G-protein signaling. For instance, analyses of DREADD-transgenic mice have revealed the enhancing effect of hepatic glucose production by Gαs, Gαi, and Gαq signaling in hepatocytes, and our G12D discovered the enhancing effect of browning by Gα12 signaling in adipocytes22,39,40,41. In the present study, an engineered mutant of G12D created the novel Gα12/13-DREADD, which is useful for research on Gα12/13 signaling. In addition, mutations that enhance coupling to Gα12, including N1032.39R, N1032.39K, S1693.53R, and S1693.53K, were discovered in the present study. These mutations could be used to design next-generation G12D that are more potent for Gα12. Gα12 and Gα13 signaling have been reported to have important functions in the immune system. Gα12/13-coupled GPCRs such as S1PR2 activate Gα12/13 signaling and regulate immune cell activities and functions in a temporally regulated manner42,43. Immune cell species-specific expression of Gα12/13-DREADD and administration of CNO at various stages of the immune response will aid in a comprehensive understanding of the tightly regulated Gα12/13 signaling function. However, for functional analysis of Gα13 signaling alone, future modification of Gα12/13-DREADD will be necessary to reduce its coupling to Gα12.

Methods

Reagents and plasmids

CNO dihydrochloride was purchased from Tocris Bioscience or synthesized at the International Institute for Integrative Sleep Medicine (WPIIIIS), University of Tsukuba. The detailed synthetic method was described previously22. A plasmid encoding alkaline phosphatase (AP)-TGFα was described previously5. G12D (M3D-GPR183/ICL3 F931.57V) and G12D mutants carrying the N-terminal FLAG-epitope tag (DYKDDDDK) were cloned into the pcDNA3.1 expression vector5. The amino-acid sequence of FLAG-epitope-tagged G12D is denoted below; the ICL3 swap, the F931.57V mutation, the FLAG-epitope tag and the substitutions of Y3.33C and A5.46G mutations from the human M3 receptor are underlined and highlighted in bold:

MDYKDDDDKGSTLHNNSTTSPLFPNISSSWIHSPSDAGLPPGTVTHFGSYNVSRAAGNFSSPDGTTDDPLGGHTVWQVVFIAFLTGILALVTIIGNILVIVSVKVNKQLKTVNNYFLLSLACADLIIGVISMNLFTTYIIMNRWALGNLACDLWLAIDCVASNASVMNLLVISFDRYFSITRPLTYRAKRTTKRAGVMIGLAWVISFVLWAPAILFWQYFVGKRTVPPGECFIQFLSEPTITFGTAIAGFYMPVTIMTILYWRIYKETERTAKQNPLTEKSGVEKKAAQTLSAILLAFIITWTPYNIMVLVNTFCDSCIPKTFWNLGYWLCYINSTVNPVCYALCNKTFRTTFKMLLLCQCDKKKRRKQQYQQRQSVIFHKRAPEQAL.

Gαq-12C, Gαq-13C, and their mutants were inserted into the pcDNA3.1 expression vector. The amino acid sequence of Gαq-12C and Gαq-13C are as follows (underlining denotes the α5 helices of Gα12 and Gα13).

Gαq-12C:

MTLESIMACCLSEEAKEARRINDEIERQLRRDKRDARRELKLLLLGTGESGKSTFIKQMRIIHGSGYSDEDKRGFTKLVYQNIFTAMQAMIRAMDTLKIPYKYEHNKAHAQLVREVDVEKVSAFENPYVDAIKSLWNDPGIQECYDRRREYQLSDSTKYYLNDLDRVADPAYLPTQQDVLRVRVPTTGIIEYPFDLQSVIFRMVDVGGQRSERRKWIHCFENVTSIMFLVALSEYDQVLVESDNENRMEESKALFRTIITYPWFQNSSVILFLNKKDLLEEKIMYSHLVDYFPEYDGPQRDAQAAREFILKMFVDLNPDSDKIIYSHFTCATDTENVRFVFHAVKDTILQENLKDIMLQ;

Gαq-13C:

MTLESIMACCLSEEAKEARRINDEIERQLRRDKRDARRELKLLLLGTGESGKSTFIKQMRIIHGSGYSDEDKRGFTKLVYQNIFTAMQAMIRAMDTLKIPYKYEHNKAHAQLVREVDVEKVSAFENPYVDAIKSLWNDPGIQECYDRRREYQLSDSTKYYLNDLDRVADPAYLPTQQDVLRVRVPTTGIIEYPFDLQSVIFRMVDVGGQRSERRKWIHCFENVTSIMFLVALSEYDQVLVESDNENRMEESKALFRTIITYPWFQNSSVILFLNKKDLLEEKIMYSHLVDYFPEYDGPQRDAQAAREFILKMFVDLNPDSDKIIYSHFTCATDTENIRLVFRDVKDTILHDNLKQLMLQ.

Cell culture and transfection

Parent and Gαq/Gα11/Gα12/Gα13 quadruple-deficient HEK293 cells5 were maintained in Dulbecco’s modified Eagle medium (DMEM, Nissui Pharmaceutical) supplemented with 10% fetal bovine serum (Gibco), 100 U/mL penicillin (Sigma-Aldrich), and 100 µg/mL streptomycin (Gibco) (complete DMEM) in a 37 °C incubator with 5% CO2. Transfection of plasmid DNAs was performed using lipofection reagent polyethylenimine solution (PEI Max, Polysciences). Typically, cells were seeded in each well of a 12-well culture plate at a cell density of 2 × 105 to 3 × 105 cells/mL in 1 mL complete DMEM and cultured for 1 day in a 37 °C incubator with 5% CO2. For transfection, plasmid solution was diluted in 50 µL Opti-MEM (Gibco) and mixed with 2.5 µL of 1 mg/mL PEI solution in 50 µL Opti-MEM. Cells were incubated for 1 day after transfection before use in an experiment.

TGFα shedding assay

The TGFα shedding assay was performed as described previously5,23. Plasmid transfection was performed in a 12-well plate using a mixture containing 250 ng of plasmid DNA encoding AP-TGFα, 100 ng of plasmid encoding FLAG-G12D, and 50 ng of plasmid encoding Gαq-12C or Gαq-13C. After incubation for 1 day, transfected cells were harvested by trypsinization, pelleted by centrifugation at 190 × g for 5 min, and suspended in 3.5 mL Hank’s Balanced Salt Solution (HBSS) containing 5 mM HEPES (pH 7.4). After incubation for 15 min at room temperature, during which spontaneous AP-TGFα shedding settled down, the cells were centrifuged at 190 × g for 5 min, and cell pellets were resuspended in 3.5 mL HBSS. The resuspended cells were plated in a 96-well plate at 90 µl per well (typically 24 total wells; triplicate measurement) and placed in a 37 °C incubator with 5% CO2 for 30 min. After incubation, 10 µl of 10× CNO serial dilution in HBSS containing 5 mM HEPES (pH 7.4) and 0.01% (v/w) bovine serum albumin (BSA) was added to each well and incubated for 1 h at 37 °C in 5% CO2. The 96-well cell plates were centrifuged at 190 × g for 2 min. After centrifugation, 80 µL supernatant from each well was transferred to another 96-well plate (conditioned media plate or CM plate), leaving attached cells and 20 µL supernatant in the original well of the cell plate. Then, 80 µL of para-nitrophenyl phosphate (p-NPP) solution (10 mM p-NPP; 40 mM Tris-HCl, pH 9.5; 40 mM NaCl; 10 mM MgCl2) was added to each well of each plate. Absorbance at 405 nm (Abs405) was measured for both plates before (background) and after a 1-h or 2-h incubation at room temperature using a microplate reader (SpectraMax 340 PC384, Molecular Devices). The percent of AP-TGFα release was calculated using the following equation:

$${\text{AP}} \mbox{-} {\text{TGF}}\upalpha \, (\% {\text{ CM}}) = \, [\Delta {\text{Abs}}_{{{4}0{\text{5 CM}}}} /\left( {\Delta {\text{Abs}}_{{{4}0{\text{5 CM}}}} + \, \Delta {\text{Abs}}_{{{4}0{\text{5 cell}}}} } \right)\; \times \;{1}.{25}]\; \times \;{1}00$$

where CM denotes conditioned media.

Vehicle-treated AP-TGFα (%CM) was subtracted from CNO-stimulated AP-TGFα (%CM) and the resulting value (%AP-TGFα release) was used to represent a G-protein response.

Flow cytometry

∆Gq/∆G12 cells were seeded in a 12-well culture plate at a concentration of 3 × 105 cells per mL (1 mL per well) 1 day before transfection. Transfection was performed using 100 ng of plasmid encoding FLAG-G12D (or mutant FLAG-G12D), 250 ng plasmid encoding AP-TGFα and 50 ng plasmid encoding Gαq-12C (or Gαq-13C). After incubation for 1 day, cells were collected by adding 100 μL 0.53 mM EDTA-containing D-PBS, followed by 100 μL of 5 mM HEPES (pH 7.4)-containing HBSS. The cell suspension was transferred to a 96-well V-bottom plate and fluorescently labeled using anti-FLAG-epitope tag monoclonal antibody (Clone 1E6, FujiFilm Wako Pure Chemical; 10 μg per mL) diluted in 2% goat serum, 2 mM EDTA in D-PBS (blocking buffer), followed by goat anti-mouse IgG secondary antibody conjugated with Alexa Fluor 488 (Thermo Fisher Scientific; 10 μg per mL). After washing with D-PBS, the cells were resuspended in 200 μL of 2 mM EDTA in D-PBS and filtered through a 40-μm filter. Single-cell fluorescence was quantified using an EC800 flow cytometer with a 488-nm laser (Sony). Live cells were gated with a forward scatter (‘FS-Peak-Lin’) cutoff of 390 and a gain of 1.7. Mean fluorescence intensity of all recorded events (approximately 20,000 cells per sample) was analyzed with FlowJo software (FlowJo) and used for statistical analysis.

Western blot

∆Gαq/∆G12 cells were lysed in SDS-PAGE sample buffer (62.5 mM Tris–HCl [pH 6.8], 50 mM dithiothreitol, 2% SDS, 10% glycerol, and 4 M urea) containing 1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride and 2 mM sodium orthovanadate. Lysates of an equal number of cells were separated using 8% or 12.5% SDS-polyacrylamide gel electrophoresis. Subsequently, the proteins were transferred to a nitrocellulose membrane (GE Healthcare). The membrane was blocked in 5% skim milk in blotting buffer (10 mM Tris–HCl (pH 7.4), 190 mM NaCl, and 0.05% Tween 20), and proteins were labeled and visualized by immunoblotting with appropriate primary and secondary antibodies. The primary antibody used in this study was an anti-Gq goat polyclonal antibody (Abcam, ab128060, D-6, lot GR108939-6, 1:2000 dilution). The secondary antibody was horseradish peroxidase (HRP)-conjugated anti-goat IgG (American Qualex, A201PS lot 7A0327H). The membrane was soaked in luminol reagent (100 mM Tris–HCl (pH 8.5), 50 mg per mL of Luminol Sodium Salt HG (FujiFilm Wako Pure Chemical), 0.2 mM p-Coumaric acid and 0.03% (v/v) of H2O2). Chemiluminescence imaging was performed and band intensities were measured using Amersham Imager 680 (Cytiva).

Modeling the G12D–Gα12 and the G12D–Gα13 complexes

To build computational models of the G12D–Gα12 and the G12D–Gα13 complexes, we used SWISS-MODEL and I-TASSER structure prediction software packages with the muscarinic acetylcholine receptor 1 (M1R)–G11 protein complex (PDB: 6OIJ) as a template 24,25,26 (Fig. 2A). These predicted structures were embedded in a phosphatidylcholine lipid bilayer using the Bilayer Builder tool in CHARMM-GUI27.

Molecular dynamics simulation and clustering

Each complex was relaxed and equilibrated through 500-ns NPT ensemble conventional molecular dynamics (cMD) simulations using a 2-fs time step, an AMBER force field, and AMBER’s pmemd.cuda program44. For the duration of all simulations, pressure and temperature were kept at 1 atm and 310.15 K, respectively. Before full-length cMD simulations, both complexes underwent six relaxation steps to minimize energy and equilibrate the systems as described previously28. G12D–Gα12 and G12D–Gα13 complexes were then subjected to 50-ns GaMD equilibration followed by a 500-ns GaMD production run using default parameters, which includes harmonic dual-boost potential to reduce energy barriers to enhance conformational sampling29. To find distinct conformations in the GaMD snapshots, the DBScan clustering algorithm was used with the “minpoints” parameter set to 4 conformations and the distance cutoff “epsilon” parameter of 2.25 Å45. The full receptor–Gα complex was aligned during clustering. The DBScan clustering algorithm used pairwise-RMSD of this complex between any two trajectory snapshots as the distance metric for clustering. Any two conformation with an RMSD greater than 2.25 Å is put in different clusters and all final reported clusters should have at least 4 conformations in each cluster based on the parameters used. The advantage of this algorithm over other popular algorithms like K-means clustering and hierarchical clustering is that in DBScan clustering one does not need to specify the target number of clusters, because the number of clusters should be determined by the underlying dynamics of the molecular system. The conformation that best represented each DBScan cluster as its centroid was chosen as a starting point for an additional 500-ns cMD simulation to allow for unbiased MMPBSA energy calculations and analysis of receptor–G-protein interactions. Residue interaction analysis was performed on the average structure of the most populated cluster for each complex.

Receptor–Gα binding free energy analysis

MMPBSA-based total energy (∆G) was determined by calculating the difference between the complex free energy and the sum of the G protein and receptor free energies as described previously28.

Data analysis

Concentration–response curves were fitted using Nonlinear Regression Analysis: Variable slope (four parameters) function in the GraphPad Prism 9 software (GraphPad) was used with absolute Hill Slope values less than 2. Sigmoid maximum effect (Emax, also referred to as Span) and the negative log of the half-maximal excitation concentration (pEC50) were used to evaluate G-protein-coupling activity. Protocols for normalizing and for generating experimental replicates are described in the legends of each Figure, as appropriate.