Wen X , Shang P , Chen H , Guo L , Rong N , Jiang X , Li X , Liu J , Yang G , Zhang J , Zhu K , Meng Q , He X , Wang Z , Liu Z , Cheng H , Zheng Y , Zhang B , Pang J , Liu Z , Xiao P , Chen Y , Liu L , Luo F , Yu X , Yi F , Zhang P , Yang F , Deng C , Sun JP .

	Graphical abstract
	Figure S1. Evolution analysis and potential mechanism of GPR4 for pH sensing in different species, related to Figure 1 (A) Genomic location of GPR4 in vertebrates. Synteny for chromosomal regions containing the GPR4. The GPR4 and the neighboring genes in vertebrate genome fragments are shown. The genomic location of GPR4 is flanked by Outer Mitochondrial Membrane Lipid Metabolism Regulator OPA3 (OPA3) and Gastric Inhibitory Polypeptide Receptor (GIPR) and is conserved in vertebrates, but GPR4 is absent at the corresponding positions in the bird genome. The complete names of these species were shown in Table S1. (B and C) Effects of different pH buffers on 5 μM forskolin-induced cAMP accumulation in cells overexpressing GloSensor (B) and activation of Gs-coupled receptor TGR5 by a synthetic agonist P395 as controls (C). The results suggested pH changes do not significantly affect intracellular cAMP accumulation measurement within the extracellular pH range from 7.8 to 5.8 in HEK293 cells. The values are presented as the means ± SEMs of 3 independent experiments (n = 3). ns, no significant difference. (D) pH-dependent curves of GPR4 across different species in the cAMP accumulation assay. (E) Table of blood pH values for each species. The blood pH values of M. unicolor and G. seraphini have not been reported yet, so the blood pH values of Typhlonectes compressicauda85 and Boulengerula taitanus,86 which are closely related to M. unicolor and G. seraphini, are cited as references. The blood pH values of B. bufo have not been reported yet, so the blood pH values of Bufo gargarizans, which is closely related to B. bufo, are cited as references. Other references cited in our tables are listed below: C. harengus,87 B. taurus,88 and H. sapiens.89
	Figure 1. Optimal pH for GPR4 across different species and adaptive evolution of Xenopus GPR4 (A) Summary of blood pH and optimal pH of GPR4 across different species. Orange solid circle indicates blood pH reported in the literature. Green solid circle indicates that blood pH was experimentally determined in current manuscript. Blue solid circle indicates that optimal pH was experimentally determined in current manuscript. (B) Summary of optimal pH of GPR4 from different species. (C) Partial sequence alignment of the positively selected sites (red) for the GPR4 in Xenopus. The construction of phylogenetic tree based on TimeTree (http://www.timetree.org/home 28). Xenopus was shown with the red line. (D) A linear correlation between the blood pH and optimal pH of GPR4 among different species was identified. The blood pH of M. musculus (n = 10), B. bufo (n = 10), R. catesbeiana, (n = 12), X. tropicalis (n = 16), and X. laevis (n = 18) was experimentally determined in the current manuscript, whereas all other blood pH values were obtained from the existing literature. All experimental animals are adult male individuals. Species include M. unicolor, G. seraphini, C. harengus, B. taurus, M. musculus, H. sapiens, B. bufo, R. catesbiana, X. tropicalis, and X. laevis. (E) Adult male X. tropicalis and R. catesbeiana were submerged for different periods of time (0 and 90 min; 12 and 18 h), and immediately after anesthesia, arterial blood was collected for blood gas analysis. (F) Measurement of arterial blood pH in adult male X. tropicalis and R. catesbeiana after submerged for different time periods (0 and 90 min). X. tropicalis was submerged for 0 and 90 min (n = 10), R. catesbeiana for 0 min (n = 13), and R. catesbeiana for 90 min (n = 8). ∗∗∗p < 0.001, the submerged groups compared with the normal 0 min groups; the values are represented as the means ± SEMs. All the data were statistically analyzed using Student’s t test. (G) Measurement of arterial blood pH in adult male X. tropicalis and X. laevis after being submerged for 0, 12, and 18 h. n = 10, ∗∗∗p < 0.001, the submerged groups compared with the normal 0 h groups; the values are represented as the means ± SEMs. All the data were statistically analyzed using Student’s t test. (H) pH-dependent curves of xtGPR4 (left) and mmGPR4 (right) in the cAMP accumulation assay. Middle, simplified evolutionary tree of vertebrate species (including fishes, amphibians, reptiles, and mammals) expressing GPR4. In left and right, the ascending pH range (Ar) of cAMP accumulation was shown with green background. The optimal pH range (Op) was shown with blue background. The values are represented as means ± SEMs of three independent experiments (n = 3). The abbreviation and complete names of species in this figure were shown in Table S1. The cartoon images in (A), (D), (E), and (H) were generated using BioRender (https://biorender.com). See also Figure S1.
	Figure 2. Cryo-EM structures of xtGPR4-Gs and mmGPR4-Gs complexes (A) Cryo-EM maps and ribbon representation of the receptor portion of GPR4apo-anti BRIL Fab complex. The xtGPR4apo structure is shown in gray. (B) Cryo-EM maps and ribbon representation of xtGPR4-Gs complexes at different pH values. xtGPR4 in the xtGPR4-Gs complexes at pH 7.2, pH 6.7, and pH 6.2 were shown in magenta, gray blue, and soft purple, respectively. Gαs: orange, Gβ1: soft blue, Gγ2: very soft blue, and scFv16: light gray. (C) A dendrogram clustering representation of GPR4 evolution from aquatic animals to mammals. The animal images were generated using BioRender (https://biorender.com). (D) Cryo-EM maps and ribbon representation of mmGPR4-Gs complexes at different pH values. mmGPR4 in the mmGPR4-Gs complexes at pH 7.6, pH 7.2, and pH 6.2 were shown in sea blue, grass green, and violet, respectively. Gαs: orange, Gβ1: soft blue, Gγ2: very soft blue, and scFv16: light gray. (E and F) Left: comparison between the mmGPR4AF2/xtGPR4apo generated by AlphaFold2 (gray) and the s-mmGPR4apo (green)/s-xtGPR4apo (blue) model generated by molecular dynamics simulation based on mmGPR4AF2 model or xtGPR4apo model. Right: the average RMSD values of the simulated xtGPR4/mmGPR4 apo structure during triplicate 200 ns molecular dynamic (MD) simulations. See also Figure S2.
	Figure S2. Representative size-exclusion chromatography elution profiles of mmGPR4-Gs, xtGPR4-Gs, and apo-xtGPR4 using Superose 6 Increase 10/300 GL, and cryo-EM images and single-particle reconstruction of the EM map of the mmGPR4-Gs, xtGPR4-Gs, and apo-xtGPR4 complexes, related to Figure 2 (A) To increase the expression levels of the two GPR4 receptors, we introduced thermostable cytochrome b562RIL (BRIL) at the N terminus of full-length xtGPR4 and mmGPR4. We co-expressed BRIL-mmGPR4/BRIL-xtGPR4, modified Gαs91,92,93 Gβ1, Gγ2, and scFv16 in Spodoptera frugiperda (Sf9) insect cells and assembled GPR4-Gs complexes in vitro. The structures of the mmGPR4-Gs complexes at pH 7.6, pH 7.2, and pH 6.2, and the xtGPR4-Gs complexes at pH 7.2, pH 6.7, and pH 6.2, were resolved. The cryo-EM structure of xtGPR4apo at pH 8.0 was solved with an overall resolution of 3.25 Å, using the previously described antibody-BRIL-fusion strategy.94 Representative elution profiles of in vitro-reconstituted mmGPR4pH6.2-Gs complex, mmtGPR4pH7.2-Gs complex, mmGPR4pH7.6-Gs complex, xtGPR4pH6.2-Gs complex, xtGPR4pH6.7-Gs complex, xtGPR4pH7.2-Gs complex, and xtGPR4apo on Superose 6 Increase 10/300 column. The size-exclusion chromatography peaks are marked by triangles. (B) Representative cryo-EM micrograph and two-dimensional (2D) class averages of xtGPR4pH6.2-Gs particles (scale bar: 100 nm). (C) Flow chart for three-dimensional (3D) classification of xtGPR4pH6.2-Gs particles. About 3,491,325 particles were used for step-by-step 2D classification as well as ab-initio reconstruction and hetero refinement. 3D volume of classes and refinements are shown and red box indicated selected class. We have performed mask for receptor-alone for xtGPR4pH6.2-Gs complex structures. The Fourier shell correlation curves indicated that mask of receptor region in xtGPR4pH6.2-Gs complexes improved the map resolution of specific region of xtGPR4pH6.2. At the FSC 0.143 cut-off, the overall and the local resolutions for the map were 2.60 and 2.35 Å, respectively. (D) Representative cryo-EM micrograph and two-dimensional (2D) class averages of xtGPR4pH6.7-Gs particles (scale bar: 100 nm). (E) Flow chart for three-dimensional (3D) classification of xtGPR4pH6.7-Gs particles. About 3,099,005 particles were used for step-by-step 2D classification as well as ab-initio reconstruction and hetero refinement. 3D volume of classes and refinements are shown and red box indicated selected class. We have performed mask for receptor-alone for xtGPR4pH6.7-Gs complex structures. The Fourier shell correlation curves indicated that mask of receptor region in xtGPR4pH6.7-Gs complexes improved the map resolution of specific region of xtGPR4pH6.7. At the FSC 0.143 cut-off, the overall and the local resolutions for the map were 2.87 and 2.38 Å, respectively. 3D density map colored according to local resolution (Å) of the xtGPR4pH6.7-Gs complex were shown. (F) Representative cryo-EM micrograph and two-dimensional (2D) class averages of xtGPR4pH7.2-Gs particles (scale bar: 100 nm). (G) Flow chart for three-dimensional (3D) classification of xtGPR4pH7.2-Gs particles. About 3,910,308 particles were used for step-by-step 2D classification as well as ab-initio reconstruction and hetero refinement. 3D volume of classes and refinements are shown and red box indicated selected class. We have performed mask for receptor-alone for xtGPR4pH7.2-Gs complex structures. The Fourier shell correlation curves indicated that mask of receptor region in xtGPR4pH7.2-Gs complexes improved the map resolution of specific region of xtGPR4pH7.2. At the FSC 0.143 cut-off, the overall and the local resolutions for the map were 2.65 and 2.45 Å, respectively. (H) Representative cryo-EM micrograph and two-dimensional (2D) class averages of xtGPR4apo-Gs particles (scale bar: 100 nm). (I) Flow chart for three-dimensional (3D) classification of xtGPR4apo particles. About 1,939,133 particles were used for step-by-step 2D classification and ab-initio reconstruction. 3D volume of classes and refinements are shown and red box indicated selected class. We have performed mask for receptor-alone for xtGPR4apo structure. The Fourier shell correlation curves indicated that mask of receptor region in xtGPR4apo structure improved the map resolution of specific region of xtGPR4apo. At the FSC 0.143 cut-off, the local resolutions for the map was 2.36 Å, respectively. 3D density map colored according to local resolution (Å) of the xtGPR4apo structure were shown. (J) Representative cryo-EM micrograph and two-dimensional (2D) class averages of mmGPR4pH6.2-Gs particles (scale bar: 100 nm). (K) Flow chart for three-dimensional (3D) classification of mmGPR4pH6.2-Gs particles. About 4,703,895 particles were used for step-by-step 2D classification, Topaz Train as well as ab-initio reconstruction and hetero refinement. 3D volume of classes and refinements are shown and red box indicated selected class. We have performed mask for receptor-alone for mmGPR4pH6.2-Gs complex structures. The Fourier shell correlation curves indicated that mask of receptor region in mmGPR4pH6.2-Gs complexes improved the map resolution of specific region of mmGPR4pH6.2. At the FSC 0.143 cut-off, the overall and local resolutions for the map were 2.76 and 2.72 Å, respectively. 3D density map colored according to local resolution (Å) of the mmGPR4pH6.2-Gs complex were shown. (L) Representative cryo-EM micrograph and two-dimensional (2D) class averages of mmGPR4pH7.2-Gs particles (scale bar: 100 nm). (M) Flow chart for three-dimensional (3D) classification of mmGPR4pH7.2-Gs particles. About 4,123,782 particles were used for step-by-step 2D classification, Topaz Train as well as ab-initio reconstruction and hetero refinement. 3D volume of classes and refinements are shown and red box indicated selected class. We have performed mask for receptor-alone for mmGPR4pH7.2-Gs complex structures. The Fourier shell correlation curves indicated that mask of receptor region in mmGPR4pH7.2-Gs complexes improved the map resolution of specific region of mmGPR4pH7.2. At the FSC 0.143 cut-off, the overall and local resolutions for the map were 2.52 and 2.44 Å, respectively. 3D density map colored according to local resolution (Å) of the mmGPR4pH7.2-Gs complex were shown. (N) Representative cryo-EM micrograph and two-dimensional (2D) class averages of mmGPR4pH7.6-Gs particles (scale bar: 100 nm). (O) Flow chart for three-dimensional (3D) classification of mmGPR4pH7.6-Gs particles. About 4,260,515 particles were used for step-by-step 2D classification, Topaz Train as well as ab-initio reconstruction and hetero refinement. 3D volume of classes and refinements are shown and red box indicated selected class. We have performed mask for receptor-alone for mmGPR4pH7.6-Gs complex structures. The Fourier shell correlation curves indicated that mask of receptor region in mmGPR4pH7.6-Gs complexes improved the map resolution of specific region of mmGPR4pH7.6. At the FSC 0.143 cut-off, the overall and local resolutions for the map were 2.62 and 2.37 Å, respectively. 3D density map colored according to local resolution (Å) of the mmGPR4pH7.6-Gs complex were shown. (P–V) Cryo-EM density maps and models are shown for TM1–TM7 of xtGPR4, mmGPR4, and α5 helix of Gαs. (P) In the xtGPR4pH6.2-Gs complex using the global map of xtGPR4pH6.2-Gs complex with map threshold at 0.15. (Q) In the xtGPR4pH6.7-Gs complex using the global map of xtGPR4pH6.7-Gs complex with map threshold at 0.15. (R) In the xtGPR4pH7.2-Gs complex using the global map of xtGPR4pH7.2-Gs complex with map threshold at 0.15. (S) In the xtGPR4apo using the global map of xtGPR4apo with map threshold at 0.15. (T) In the mmGPR4pH6.2-Gs complex using the global map of mmGPR4pH6.2-Gs complex with map threshold at 0.15. (U) In the mmGPR4pH7.2-Gs complex using the global map of mmGPR4pH7.2-Gs complex with map threshold at 0.15. (V) In the mmGPR4pH7.6-Gs complex using the global map of mmGPR4pH7.6-Gs complex with map threshold at 0.15. When using “Color Zone” tool in chimera, we set the color radius at 2 Å. Several key proton-sensing residues of xtGPR4apo only have poor EM density for their side chains; therefore, we used molecular dynamics simulations to simulate the side chains of those residues, including D161ECL2-45.43, E170ECL2-45.52, and H159ECL2-45.41. Although the cryo-EM data of the mmGPR4-Gs complexes and xtGPR4-Gs complexes at ascending pH range can be classified into two major classes, resulting in the resolution of two structures (mmGPR4pH7.6 and mmGPR4pH7.6b, xtGPR4pH7.2 and xtGPR4pH7.2b, xtGPR4pH6.7 and xtGPR4pH6.7b), the overall configurations of these two structures at each pH condition are in general similar. Due to insufficient EM density of mmGPR4pH7.6b, xtGPR4pH7.2b, xtGPR4pH6.7b, and mmGPR4pH6.2b, we were unable to accurately assign the side chains of the key H involved in proton sensing. As a result, we did not include the details of the structure of mmGPR4pH6.2b, mmGPR4pH7.6b, xtGPR4pH7.2b, and xtGPR4pH6.7b. In further analysis, only the structures of xtGPR4apo, mmGPR4pH6.2, mmGPR4pH7.2, mmGPR4pH7.6, xtGPR4pH6.7, xtGPR4pH6.2, and xtGPR4pH7.2 are used. However, the classifications indicated that multiple conformational states of key proton-sensing residue H may exist under specific pH condition, in particularly the ascending range. Further studies involving the collection of more cryo-EM data may provide a more accurate explanation of the conformational changes in extracellular His in response to pH changes. A receptor region mask was carried out for all the xtGPR4/mmGPR4-Gs complexes and xtGPR4apo to improve the EM density of the extracellular residues (Figures S2B–S2O; Table S5). Through the EM densities of these structures, models were built for most residues of xtGPR4/mmGPR4 and the G protein trimer (Figures S2P–S2V; Table S6). To obtain further knowledge on protonation-induced GPR4 activation by comparing structural differences between active and inactive states, the mmGPR4 receptor at apo state (s-mmGPR4apo) was modeled according to the AlphaFold 2 prediction, followed by molecular dynamics simulation (Figures 2E and 2F).
	Figure 3. Structural basis of proton sensing by mmGPR4 (A) Frontal and top views of the three-dimensional (3D) structural representation of the extracellular surface histidines of mmGPR4pH7.2. (B) HIP (histidine with hydrogens on both nitrogens) proportions of extracellular surface histidines at apo state (designated the inactive state, colored pink), pH7.6 (designated the less active state, colored orange), and pH 7.2 (designated the full active state, colored green). (C) Effects of the H191.32A, H812.66A, H822.67A, H87ECL1-23.52A, H1574.63A, H167ECL2-45.47A, and H2717.36A mutations on mmGPR4-induced cAMP accumulation under Op conditions (left) or Ar conditions (right). Mutants resulting in a significant difference were highlighted with a red background. The values were presented as the means ± SEMs of 4 independent experiments (n = 4). ∗∗∗p < 0.001; ns, not significant. All the data were analyzed by one-way ANOVA with Tukey’s test (compared with wild-type [WT] mmGPR4). (D) The 3D structure representation detailing the interactions between H1574.63, H167ECL2-45.47, H2717.36, and surrounding residues in the apo structures of mmGPR4. (E) Effects of the mutations on mmGPR4-induced cAMP accumulation under Op conditions (left) or Ar conditions (right). Mutants resulting in a significant difference were highlighted with a red background. The values were presented as the means ± SEMs of 4 independent experiments (n = 4). ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001, ns, not significant, compared with WT GPR4 via one-way ANOVA with Tukey’s test. #p < 0.05; ##p < 0.01; ###p < 0.001, compared with the indicated group using unpaired t test. See also Figure S3.
	Figure S3. Structural basis of proton sensing by mmGPR4, and proportion of protonated Asp (D)/Glu (E) residues in the extracellular region of mmGPR4 and xtGPR4, and comparison of TM6 movement in mmGPR4 and xtGPR4 after activation with classical GPCR, related to Figure 3 (A) The EM density and 3D structural representation of extracellular histidines of mmGPR4pH7.2. Through the EM densities of mmGPR4pH7.6 and mmGPR4pH7.2, we unambiguously modeled 7 histidine residues in the extracellular region of mmGPR4: H191.32, H812.66, H822.67, H87ECL1-23.52, H1574.63, H167ECL2-45.47, and H2717.36. (B) The protonation states and ratios of the 8 histidines on the extracellular surface of mmGPR4 at pH 7.8 (apo state), pH 7.6 (less active state state), and pH7.2 (active state). (HID, histidine with hydrogen on the delta nitrogen; HIE, histidine with hydrogen on the epsilon nitrogen; HIP, histidine with hydrogens on both nitrogen.) Cartoon displays histidine protonation states for those with more than 50% protonation. (C) The pH-dependent curves of cAMP accumulation assay in cells overexpressing WT mmGPR4 and H191.32A, H812.66A, H822.67A, H87ECL1-23.52A, H1574.63A, H167ECL2-45.47A, or H2717.36A mutants of mmGPR4. Values are represented as means ± SEMs of 3 independent experiments (n = 4) (left). ELISA data summarizing the expression levels of WT and mutants of mmGPR4. Values are represented as means ± SEMs of 3 independent experiments (n = 3). n.s., no significant difference. All the data were analyzed by one-way ANOVA with Tukey’s test (right). (D) The 3D structural representation of H822.67 and H87ECL1-23.52 of mmGPR4pH7.2 and mmGPR4apo. (E) The 3D structural representation of H191.32, H1574.63, H167ECL2-45.47, and H2717.36 of mmGPR4pH7.2 and mmGPR4apo. (F) The 3D structural representation of H1574.63, F1564.62, and W1795.34 of mmGPR4pH7.6 and mmGPR4apo. (G and H) Cryo-EM density of H1574.63 and surrounding residues, including F1564.62, W1795.34, and A1534.59, in the structures of mmGPR4pH7.6-Gs and mmGPR4pH7.2-Gs. (I) Cryo-EM density of H167ECL2-45.47 and surrounding residue D163 ECL2-45.43 in the structures of mmGPR4pH7.6-Gs. (J) The 3D structural representation of H167ECL2-45.47, F169ECL2-45.49, and F2677.32 of mmGPR4pH7.6 and mmGPR4apo. The imidazole ring of H167ECL2-45.47 in mmGPR4pH7.6 traversed by approximately 4.6 Å and formed hydrophobic packing with F169 ECL2-45.49 and F2677.32. (K) Cryo-EM density of H167ECL2-45.47 and surrounding residues, including F169ECL2-45.49, D163ECL2-45.43, D83ECL1-23.48, and F2677.32, in the structures of mmGPR4pH7.2-Gs. (L and M) Cryo-EM density of H2717.36 and surrounding residues, including E172ECL2-45.52, in the structures of mmGPR4pH7.6-Gs and mmGPR4pH7.2-Gs. (N) The pH-dependent curves of cAMP accumulation assay in cells overexpressing WT mmGPR4 and D772.62A, D8323.48A, D8323.48N, F973.30A, D16345.43A, D16345.43N (left), F16945.49A, E17245.52A, E17245.52Q, or F2677.32A (middle) mutants of mmGPR4. The dashed curves for mmGPR4-WT shown in (P) are reprinted from (O) for direct comparisons. Values are represented as means ± SEMs of 3 independent experiments (n = 4). ELISA data summarizing the expression levels of WT and mutants of mmGPR4. Values are represented as means ± SEMs of 3 independent experiments (n = 3). n.s., no significant difference. All the data were analyzed by one-way ANOVA with Tukey’s test (right). (O) Proportions of protonated Asp (D)/Glu (E) residues at the extracellular surface in mmGPR4 at apo and pH 7.2. We next calculated the pKa values of all D/E residues in the extracellular region of both mmGPR4apo and mmGPR4pH7.2 (optimal pH). In the mmGPR4pH7.2 structure, all D/E residues in the extracellular region exhibited less than 10% protonation, showing no significant increase in protonation states. This result suggests that extracellular H residues, rather than D/E residues, are most likely the protonation sensors responsible for inducing mmGPR4-Gs signaling activation. (P) The examined Asp (D)/Glu (E) residues in the extracellular part of mmGPR4 at pH 7.2. (Q) Cryo-EM density of H1574.63, H167ECL2-45.47, H2717.36, and surrounding residues, including A1534.59, E159ECL2-45.39, F169ECL2-45.49, D163ECL2-45.43, D83ECL1-23.48, F2677.32, and E172ECL2-45.52, in the structures of mmGPR4pH6.2-Gs. The protonation of H167ECL2-45.47 enabled a hydrogen bond with D163ECL2-45.43, while losing the hydrogen bonding with D83ECL1-23.48 compared with the mmGPR4pH7.2 structure. (R and S) Structural comparisons between the TMs of s-xtGPR4apo and β2ARinactive, and between the ECL1 and ECL2 of s-xtGPR4apo and xtGPR4AF2. The overall architecture of xtGPR4apo structure was very similar to inactive β2 adrenoceptor (β2ARinactive) structure, showing that an overall Cα root-mean-square deviation (RMSD) value is approximately 2.9 Å, except for inward displacement of the upper portion of TM1 by approximately 8–10 Å (R). Notably, the experimentally resolved xtGPR4apo structure showed significant differences when comparing with the xtGPR4 model predicted by AlphaFold2 (xtGPR4AF2) for the configurations of extracellular loops (ECL1 and ECL2). In xtGPR4, ECL1 moved upward by approximately 3–4 Å compared with xtGPR4AF2, accompanied with downward movement of ECL2 by approximately 10 Å, resulting in a juxtaposition between ECL1 and ECL2 in xtGPR4apo (S). (T) Structural superposition of inactive and active states of mmGPR4pH7.2, xtGPR4pH6.2, β2AR (PDB: 7BZ2), and 5-HT6R (PDB: 7YS6) viewed from the side. The directional movements of TM6 in these receptors relative to the inactive state are depicted with red arrows, along with the deviation distances of TM6. mmGPR4pH7.2, xtGPR4pH6.2, β2AR, and 5-HT6R were colored in soft green, very soft blue, sky blue, and grayish magenta, respectively.
	Figure 4. Structural basis of proton sensing by xtGPR4 (A) Frontal and top views of the 3D structural representation of the extracellular surface histidines of xtGPR4pH6.2. (B) HIP proportions of extracellular surface histidines at apo state (designated the inactive state, colored gray) and pH 7.2 (designated the full active state, colored pink). (C) The 3D structure depicting detailed interactions among H159ECL2-45.41, H165ECL2-45.47, H2767.36, and the surrounding residues of xtGPR4pH6.2 (purple), xtGPR4pH6.7 (blue), and xtGPR4pH7.2 (pink) and the apo structures (gray). Hydrogen bonds are represented by red dashed lines. Only the partial EM density of the side chains of E156ECL2-45.38 and S171ECL2-45.53 in the xtGPR4pH6.2 structure, that of D161ECL2-45.43 in the xtGPR4pH6.7 structure, can be traced. We utilized MD simulation to model the side chains of these four residues, which are shown with 40% opacity in the figure and are labeled in orange font. (D) Effects of the mutations on xtGPR4-induced cAMP accumulation at the optimal pH (Op) and ascending range pH (Ar). Mutants with significant differences were highlighted with a purple background. The values were presented as the means ± SEMs of 3 independent experiments (n = 3). ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001, ns, not significant, compared with WT GPR4 via one-way ANOVA with Tukey’s test. #p < 0.05; ##p < 0.01; ###p < 0.001, compared with the indicated group using unpaired t test. (E) Effects of the mutations on xtGPR4-induced cAMP accumulation at the optimal pH (Op) and ascending range pH (Ar) values. Mutants with significant differences were highlighted with a purple background. The values are presented as the means ± SEMs of 3 independent experiments (n = 3). ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001, ns, not significant, compared with WT GPR4 via one-way ANOVA with Tukey’s test. #p < 0.05; ##p < 0.01; ###p < 0.001, compared with the indicated group using unpaired t test. (F) Heatmap of Op (100% WT) and Ar (100% WT) of the indicated GPR4 mutant of different species compared with WT GPR4 through GloSensor cAMP assay. The red represents Op and blue represents Ar. The pH-dependent curves were shown in Figure S4. The values were from three independent experiments (n = 3). The complete names of these species were shown in Table S1. (G) Sequence alignment of key residues involved in the proton sensing of GPR4 among different species across 500 million years evolution. The construction of phylogenetic tree and the species divergence time based on TimeTree (http://www.timetree.org/home). The animal images were generated using BioRender (https://biorender.com). See also Figures S4 and S5.
	Figure S4. Positive selection sites critical for acidic shift of xtGPR4 and the evolution of key residues involved in proton sensing by GPR4 across different species, related to Figure 4 (A) The protonation states and ratios of the 4 histidines on the extracellular surface of xtGPR4 at pH 8.0 (apo state), pH7.2 (less active state), and pH6.2 (active state). (HID, histidine with hydrogen on the delta nitrogen; HIE, histidine with hydrogen on the epsilon nitrogen; HIP, histidine with hydrogens on both nitrogen.) Cartoon displays histidine protonation states for those with more than 50% protonation. (B and C) Cryo-EM density of H165ECL2-45.47 and surrounding residues, including D81ECL1-23.48 and S131.28, in the structures of xtGPR4pH7.2 (B) and xtGPR4pH6.7 (C). (D) The average RMSD values of D161ECL2-45.43 in the structure of xtGPR4pH6.7 during triplicate 200 ns MD simulations. (E) Cryo-EM density of H165ECL2-45.47 and surrounding residues, including D161ECL2-45.43, D81ECL1-23.48, and S131.28, in the structures of xtGPR4pH6.2-Gs. (F) pH-dependent curves of cAMP accumulation assay in cells overexpressing WT xtGPR4, and S131.28A, D81ECL1-23.48A, D81ECL1-23.48N, D81ECL1-23.48K, H159ECL2-45.41A, D161ECL2-45.43A, D161ECL2-45.43N, D161ECL2-45.43K, H165ECL2-45.47A, D161.31K, D161.31A, Y772.64F, E156ECL2-45.38A, E170ECL2-45.52A, E170ECL2-45.52Q, E170ECL2-45.52K, S171ECL2-45.53A, and H2767.36A. The dashed curves for xtGPR4-WT shown in right are reprinted from left for direct comparisons. Values are represented as means ± SEMs of 3 independent experiments (n = 3). Mutations in D81ECL1-23.48A/K or D161ECL2-45.43A/K significantly decreased xtGPR4 activity. For comparison, the D161ECL2-45.43N and D81ECL1-23.48N mutant had less effect for protonation-induced activation of xtGPR4. In contrast, eliminating the potential H-bond between S131.28 and H165ECL2-45.47 by mutations did not significantly affect the activity of xtGPR4 in the ascending pH range or the optimal pH range. (G) Elisa experiments to determine the expression levels of the xtGPR4S131.28A, D161.31A, D161.31K, Y772.64F, D81ECL1-23.48A, D81ECL1-23.48N, D81ECL1-23.48K, E156ECL2-45.38A, H159ECL2-45.41A, D161ECL2-45.43A, D161ECL2-45.43N, D161ECL2-45.43K, H165ECL2-45.47A, E170ECL2-45.52A, E170ECL2-45.52Q, E170ECL2-45.52K, S171ECL2-45.53A, and H2767.36A. The values are presented as means ± SEMs from three independent experiments (n = 3). n.s., no significant difference. All the data were analyzed by one-way ANOVA with Tukey’s test. (H–J) Cryo-EM density of H2767.36 and surrounding residues, including Y772.64, Y172ECL2-45.54, D161.31, and E170ECL2-45.52, in the structures of xtGPR4pH7.2 (H), xtGPR4pH6.7 (I), and xtGPR4pH6.2-Gs (J). (K–T) Curves showing pH-dependent cAMP accumulation in cells overexpressing GPR4 D81ECL1-23.48A, D161ECL2-45.43A, H165ECL2-45.47A, E170ECL2-45.52A, and H2767.36A of different species. (K) M. terrapin, (L) C. mydas, (M) B. bufo, (N) R. temporaria, (O) G. seraphini, (P) L. chalumnae, (Q) D. rerio, (R) L. bergylta, (S) C. variegatus, (T) C. milii. The values are represented as means ± SEMs of 3 independent experiments (n = 3). The amino acid number of each species corresponds to the amino acid of X. tropicalis. The complete names of these species were shown in Table S1. (U and V) Elisa experiments to determine the expression levels of the GPR4 D81ECL1-23.48A, D161ECL2-45.43A, H165ECL2-45.47A, E170ECL2-45.52A, and H2767.36A of different species. (U) M. terrapin, C. mydas, B. bufo, R. temporaria, G. seraphini, (V) L. chalumnae, D. rerio, L. bergylta, C. variegatus, C. milii. The values are presented as means ± SEMs from three independent experiments (n = 3). n.s., no significant difference. All the data were analyzed by one-way ANOVA with Tukey’s test. The amino acid number of each species corresponds to the amino acid of X. tropicalis. The complete names of these species were shown in Table S1.
	Figure S5. Three key evolved sites critical for the acidic shift of xtGPR4, related to Figure 4 (A–C) Cryo-EM density of H159 and surrounding residues, including E156ECL2-45.38, in the structures of xtGPR4pH7.2-Gs, xtGPR4pH6.7-Gs, and xtGPR4pH6.2-Gs. (D and E) The average RMSD values of S171ECL2-45.53 and E156ECL2-45.38 in the structure of xtGPR4pH6.2 during triplicate 200 ns MD simulations. Whereas the side chain of H159ECL2-45.41 was easily assigned to the EM density in the xtGPR4pH6.2 structure, only partial EM densities of the side chains of E156ECL2-45.38 and S171ECL2-45.53 could be traced, and we used MD simulations to model the side chains of these two residues. (F) Curves showing pH-dependent cAMP accumulation in cells overexpressing xtGPR4 S171.32G and E156ECL2-45.38D. The values are represented as means ± SEMs of 3 independent experiments (n = 3). (G) Elisa experiments to determine the expression levels of the xtGPR4-WT, S171.32G, and E156ECL2-45.38D. The values are means ± SEMs from three independent experiments (n = 3). n.s., no significant difference. All the data were analyzed by one-way ANOVA with Tukey’s test. (H–J) WT R. catesbeiana and those injected with AAV sh-rcGPR4 together with AAV xtGPR4 (rcGPR4 xtGPR4), xtGPR4 mutants E156D (rcGPR4xtGPR4-E156D), or H159F (rcGPR4 xtGPR4-H159F) were submerged for 1.5 h (n = 10). Total RNA of the indicated tissues from the aforementioned Rana was isolated and reverse transcribed. Quantitative PCR was performed to examine the mRNA levels of rcGPR4 (H) and xtGPR4 (I). The expression of rcGPR4 and xtGPR4 was normalized to the housekeeping gene β-actin. The values were calculated as 2−ΔCt values and are presented as the means ± SEMs. ∗∗∗p < 0.001, the AAV groups compared with the WT group. All the data were statistically analyzed using Student's t test. WT R. catesbeiana and those injected with AAV sh-rcGPR4 together with AAV xtGPR4 (rcGPR4 xtGPR4), xtGPR4 mutants E156D (rcGPR4xtGPR4-E156D), or H159F (rcGPR4 xtGPR4-H159F) were submerged for 1.5 h (n = 10). Total RNA of the indicated tissues from the aforementioned frogs was isolated and reverse transcribed. Quantitative PCR was performed to examine the mRNA levels of VCAM1. The expression of VCAM1 was normalized to the housekeeping gene β-actin. The values were calculated as 2−ΔΔCt values and are presented as the means ± SEMs. ∗∗∗p < 0.001, ∗∗p < 0.01, #p < 0.05, the AAV groups compared with the WT groups (∗); the rcGPR4xtGPR4-E156D and rcGPR4 xtGPR4-H159F groups compared with the rcGPR4 xtGPR4 group (#). All the data were statistically analyzed using Student's t test. These frogs were subsequently subjected to submersion experiments to determine the evolutionary importance of the key residues responsible for the acid shift property of xtGPR4. The results revealed that the overexpression of WT xtGPR4 in R. catesbeiana (with endogenous rcGPR4 knockdown) (rcGPR4xtGPR4 group) extended the duration of submersion (3 h for AAV xtGPR4 WT). Notably, R. catesbeiana with xtGPR4 mutant (E156D and H159F) overexpression (rcGPR4xtGPR4-E156D group and rcGPR4 xtGPR4-H159F group) presented a shorter submersion duration than those with WT xtGPR4 overexpression (2.4 h for xtGPR4 mutants), although both durations were longer than those observed in WT R. catesbeiana expressing endogenous rcGPR4. Consistent with these results, the gene expression of VCAM1 was greater in the xtGPR4 mutant (E156D and H159F) group of R. catesbeiana than in the xtGPR4 WT group, whereas the expression of VCAM1 in the xtGPR4 WT group and its related two mutant groups was lower than that in WT R. catesbeiana.
	Figure 5. Mechanism of xtGPR4 and mmGPR4 activation after proton sensation (A) Schematic illustration delineating the pathways involved in the mechanism of xtGPR4 activation in response to proton sensing. (B–G) Compared with the apo state at pH 7.8 (colored gray), the protonation of H165ECL2-45.47 and formation of a polar network between H165ECL2-45.47, D81ECL1-23.48, and D161ECL2-45.43 in the xtGPR4pH6.2 structure (colored purple) result in a structural rearrangement of key residues along the propagation pathway, which is consistent with the activation hub—the traditional toggle switch Y2386.48. (H) Sequence alignment of key residues involved in the activation pathway of GPR4 from different species revealing their conservation across various species over 300 million years. The construction of phylogenetic tree and the species divergence time based on TimeTree (http://www.timetree.org/home). (I and J) 3D structural representation of the interaction between the D3.49R3.50Y3.51 motif of mmGPR4 and xtGPR4 and surrounding residues. Hydrogen bonds are depicted as red dashed lines. (K) 3D structural representation of the conformational changes of Y7.53 in the D7.49P7.50xxY7.53 motif of mmGPR4 and xtGPR4 upon proton sensing. See also Figure S6.
	Figure S6. Common mechanism shared by activation of xtGPR4 and mmGPR4, related to Figure 5 (A–F) Cryo-EM density of key residues in the activation pathway of xtGPR4pH6.2-Gs. (G) Curves showing pH-dependent cAMP accumulation in cells overexpressing xtGPR4-WT, F973.32A, Y983.33A, L1013.36A, Y172ECL2-45.54A, Y2386.48A, Y2416.51A (left), H2426.52A, R2486.58A, F2727.32A, Y2757.35A, and L2797.39A (middle). The dashed curves for xtGPR4-WT shown in (H) are reprinted from (G) for direct comparisons. The values are represented as the means ± SEMs of 3 independent experiments (n = 3). Elisa experiments to determine the expression levels of the xtGPR4-WT, F973.32A, Y983.33A, L1013.36A, Y2386.48A, Y2416.51A, H2426.52A, R2486.58A, F2727.32A, Y2757.35A, and L2797.39A. Values are mean ± SEM from three independent experiments (n = 3). n.s., no significance. All data were analyzed by one-way ANOVA with Tukey’s test (right). (H) A schematic illustrating the pathways involved in the mechanism of mmGPR4 activation in response to proton sensing. (I–M) Compared with the apo state at pH 7.8 (colored gray), in the mmGPR4pH7.2 structure (colored soft green), the protonation of H167ECL2-45.47 and the formation of a polar network between H167ECL2-45.47, D83ECL1-23.48, and D163ECL2-45.43 result in the structural rearrangement of key residues along the propagation pathway, which was consistent with the activation hub—the traditional toggle switch F2386.48. Hydrogen bonds were depicted as red dashed lines. (N–R) Cryo-EM density of key residues in the activation pathway of mmGPR4pH7.2-Gs. (S) Curves showing pH-dependent cAMP accumulation in cells overexpressing mmGPR4-WT, F993.32A, Y1003.33A, I1033.36A, H1233.56A, F17445.54A (left), Y2426.51A, H2436.52A, R2496.58A, Y2707.35A, and L2747.39A (middle). The dashed curves for mmGPR4-WT shown in (O) are reprinted from (N) for direct comparisons. The values are represented as the means ± SEMs of 3 independent experiments (n = 3). Elisa experiments to determine the expression levels of the mmGPR4-WT, F993.32A, Y1003.33A, I1033.36A, H1233.56A, F17445.54A, R2496.58A, Y2707.35A, and L2747.39A. Values are mean ± SEM from three independent experiments (n = 3). n.s., no significance. All data were analyzed by one-way ANOVA with Tukey’s test (right). (T–W) Cryo-EM density of the D3.49R3.50Y3.51 motif and surrounding residues in the structures of mmGPR4 and xtGPR4. (X) The average RMSD values of R1153.50 and H1213.56 in the structure of xtGPR4apo during triplicate 200 ns MD simulations. (Y and Z) Cryo-EM density of the D7.49P7.50xxY7.53 motif in the structures of mmGPR4 and xtGPR4.
	Figure 6. Connection of protonation of H159ECL2-45.41 to xtGPR4 and H157ECL2-45.41 to mmGPR4 activation (A–F) Structural rearrangements of key residues along the propagation path connecting the protonation of H159ECL2-45.41 site to the toggle switch Y2386.48 position, determined by comparison of the s-xtGPR4apo structure (gray) and the xtGPR4pH6.2 (blue-purple) structures. Hydrogen bonds are depicted as red dashed lines. (G) Effects of key residue mutations along propagation pathway of xtGPR4 connecting H159ECL2-45.41 site to the toggle switch Y2386.48 position on cAMP accumulation at optimal pH (Op) and ascending range pH (Ar). The values are represented as means ± SEMs of 3 independent experiments (n = 3). ∗p < 0.05; ∗∗∗p < 0.001; ns, not significant. All the data were analyzed by one-way ANOVA with Tukey’s test (compared with WT xtGPR4). (H–J) Structural rearrangements of key residues along the propagation path connecting the histidine sites responsible for pH sensing to the toggle switch F2396.48 position, determined by comparison of the s-mmGPR4apo structure (gray) and the mmGPR4pH7.2 (soft green) structures. (K) Effects of key residue mutations in the propagation pathway of mmGPR4 activation on cAMP accumulation under Op conditions (left) or Ar conditions (right). The values are presented as the means ± SEMs of 3 independent experiments (n = 3). ∗∗∗p < 0.001; ns, not significant. All the data were analyzed by one-way ANOVA with Tukey’s test (compared with the WT mmGPR4). See also Figure S7.
	Figure S7. Histidine protonation and the activation mechanism and propagation pathway of GPR4 in different species, related to Figure 6 (A–F) Cryo-EM density of key residues in the activation pathway of xtGPR4pH6.2-Gs. (G and H) Curves showing pH-dependent cAMP accumulation in cells overexpressing xtGPR4-WT, Y772.64F, F973.32A, Y983.33A, L1013.36A, Y169ECL2-45.51A (G), Y2416.51A, H2426.52A, Y2757.35A, and L2797.39A (H). The dashed curves for xtGPR4-WT shown in (H) are reprinted from (G) for direct comparisons. The values are represented as the means ± SEMs of 3 independent experiments (n = 3). (I) Elisa experiments to determine the expression levels of the xtGPR4-WT, Y772.64F, F973.32A, Y983.33A, L1013.36A, Y169ECL2-45.51A, Y2416.51A, H2426.52A, Y2757.35A, and L2797.39A. The values are presented as the means ± SEMs from three independent experiments (n = 3). n.s., no significant difference. All the data were analyzed by one-way ANOVA with Tukey’s test. (J) Residues of the conserved propagation path are preserved across 200 million years between different species. The residue numbers are annotated according to xtGPR4. Notably, these four residues were differentially clustered according to their optimal pH range of special GPR4 subtypes (from different species): L1013.36, Y169ECL2, S171ECL2, and Y172ECL2 are presented in GPR4 subtype with a more acidic optimal pH range (such as xtGPR4), whereas I1013.36, F169ECL2, K171ECL2, and F172ECL2 occurred in the GPR4 subtype with a more alkaline optimal pH range (such as mmGPR4). We therefore speculated that they might contribute to different optimal pH ranges among different species along the propagating path connecting extracellular H159ECL2-45.41 of xtGPR4 to the toggle switch position. (K) Structural rearrangements of key residues along the propagation path connecting the histidine sites responsible for pH sensing to the toggle switch position, determined by comparison of the active structures of mmGPR4-pH 7.2 (green) at the optimal pH and xtGPR4-pH 6.2 (purple). Conformational changes in the residues are highlighted by red arrows. (L) pH-dependent curves of the cAMP accumulation assay in cells overexpressing WT xtGPR4, H159ECL2-45.41F, Y169ECL2-45.51F, S171ECL2-45.53K, and Y172ECL2-45.54F of xtGPR4. The values are presented as the means ± SEMs of 3 independent experiments (n = 3). (M) ELISA experiments to determine the expression levels of xtGPR4-WT, H159ECL2-45.41F, Y169ECL2-45.51F, S171ECL2-45.53K, and Y172ECL2-45.54F. The values are presented as the means ± SEMs from three independent experiments (n = 3). n.s., no significant difference. All the data were analyzed via one-way ANOVA with Tukey’s test. (N and P) Cryo-EM density of key residues in the activation pathway of mmGPR4pH7.2-Gs. (Q and R) Curves showing pH-dependent cAMP accumulation in cells overexpressing mmGPR4-WT, F993.32A, Y1003.33A, I1033.36A, Y1043.37A, P1544.60A (Q), 15945.39A, W1795.34A, M1835.38A, and R1875.42A (R). The dashed curves for mmGPR4-WT shown in (E) are reprinted from (D) for direct comparisons. The values are represented as the means ± SEMs of 3 independent experiments (n = 3). Consistent with these structural comparisons and analyses, the F993.32A, Y1003.33A, I1033.36A, Y1043.37A, and M1835.38A mutations significantly impaired mmGPR4 activation in response to pH changes. (S) Elisa experiments to determine the expression levels of the mmGPR4-WT, F993.32A, Y1003.33A, I1033.36A, Y1043.37A, P1544.60A, E15945.39A, W1795.34A, M1835.38A, and R1875.42A. The values are presented as the means ± SEMs from three independent experiments (n = 3). ns, no significant difference. All the data were analyzed by one-way ANOVA with Tukey’s test.
	Figure 7. Coupling of xtGPR4/mmGPR4 to Gs (A) Structural depiction of the interfaces between mmGPR4 and Gαs. The interface between the receptor and Gαs consists of TM1–3, TM5–7, intracellular loops 2–3 (ICLs 2–3), and helix 8 of mmGPR4 and the αN- and C-terminal α5 helices, α4-β6 loop, and β2-β3 loop of Gαs. (B) Structural depiction of the interfaces between xtGPR4 and Gαs. The interface between the receptor and Gαs consists of TM1–3, TM5–7, ICLs 2–3, and helix 8 of xtGPR4 and the αN- and C-terminal α5 helices, α4-β6 loop, and β2-β3 loop of Gαs. (C) The residues of mmGPR4, xtGPR4, and β2AR that contribute to G protein binding in the mmGPR4pH7.2-Gs, xtGPR4pH6.2-Gs, and β2AR-Gs complexes were illustrated as follows. The filled soft blue circles indicated mmGPR4 or xtGPR4 residues that interact with Gs. The filled soft green circles indicated β2AR residues that interact with Gs. Blank circles represented receptor residues that do not interact with Gs in specific complex structures. Additionally, circles surrounded by red lines indicated common residues interacting with Gs. (D) The 3D structure representation of the interactions of mmGPR4pH7.2 and Gαs. (E) The 3D structure representation of the detailed interactions of xtGPR4pH6.2 and Gαs. (F) Sequence alignment of key residues involved in the Gs interface of GPR4 from different species revealing their conservation across various species over 400 million years. Green font: amino acid sequence number in xtGPR4. The construction of phylogenetic tree and the species divergence time based on TimeTree (http://www.timetree.org/home). (G and H) Effects of key residue mutations along the interface between the mmGPR4 (G), xtGPR4 (H), and Gs on GPR4-induced cAMP accumulation under Op conditions (left) or Ar conditions (right). Mutants resulting in a significant difference were colored soft green (G) or purple (H). The values are presented as the means ± SEMs of 3 independent experiments (n = 3). ∗∗∗p < 0.001; ns, not significant. All the data were analyzed by one-way ANOVA with Tukey’s test (compared with WT GPR4). (I) Effects of key residue mutations in cmGPR4 coupled to Gs on cmGPR4-induced cAMP accumulation under Op conditions (left) or Ar conditions (right). Mutants resulting in a significant difference were colored orange. The values are presented as the means ± SEMs of 3 independent experiments (n = 3). ∗∗∗p < 0.001; ns, not significant. All the data were analyzed by one-way ANOVA with Tukey’s test (compared with WT cmGPR4). See also Figure S8.
	Figure S8. Gs-coupling interfaces of xtGPR4 and mmGPR4, related to Figure 7 (A) The demonstration of the narrower cavities between TM3, TM6, and TM7 in xtGPR4pH6.2 and mmGPR4pH7.2 compared with β2AR using the tube helices model. In contrast to other classical class A GPCRs (such as the β2AR-Gs complex), mmGPR4 and xtGPR4 contain significantly narrower cytoplasmic cavities, enabling Gs to directly contact the cytoplasmic side residues of TM1 and TM2 of GPR4. (B and C) The EM density and 3D structural representation of key residues about the detailed interactions of E532.38, L542.39, A1203.53, V1213.54, L12534.51, Y2035.58, I2065.61, V2105.65, and I2246.33 in mmGPR4, H41G.S1.02, V227G.s3h2.03, F376 G.H3.09, I383G.H5.15, L388G.H5.20, L393G.H5.25, and L394G.H5.26 in Gαs. (D and E) The EM density and 3D structural representation of key residues about the detailed interactions of E512.38, D1143.49, A1183.53, V1193.54, L12334.51, L2055.61, L2095.65, E2196.29, V2236.33, and L2266.36 in xtGPR4, H41 G.S1.02,V227G.s3h2.03, F376G.H3.09, R380G.H5.12, I383G.H5.15, L388G.H5.20, L393G.H5.25, and L394G.H5.26 in Gαs. (F–H) Curves showing pH-dependent cAMP accumulation in cells overexpressing mmGPR4-WT, E532.38A, V1213.54A, P12434.50A, L12534.51A, S213A, E2176.26A (F), xtGPR4-WT, E512.38A, V1193.54A, P12234.50A, L12334.51A, S214A, E2196.29A (G), cmGPR4-WT, E512.38A, V1193.54A, P12234.50A, L12334.51A, S214A, and E2186.26A (H). The values are represented as the means ± SEMs of 3 independent experiments (n = 3). (I–K) Elisa experiments to determine the expression levels of the mmGPR4-WT, E532.38A, V1213.54A, P12434.50A, L12534.51A, S213A, E2176.26A (I), xtGPR4-WT, E512.38A, V1193.54A, P12234.50A, L12334.51A, S214A, E2196.29A (J), cmGPR4-WT, E512.38A, V1193.54A, P12234.50A, L12334.51A, S214A, and E2186.26A (K). The values are presented as the means ± SEMs from three independent experiments (n = 3). ns, no significant difference. All the data were analyzed by one-way ANOVA with Tukey’s test.