Development, Benchmarking and Validation of Gcoupler Computational Framework.

(a) Schematic workflow depicting different modules of the Gcoupler package. Of note, Gcoupler possesses four major modules, i.e., Synthesizer, Authenticator, Generator and BioRanker. (b) AUC-ROC curves of the finally selected model for each of the indicated GPCRs. Note: Experimentally validated active ligands and decoys were used in the testing dataset. (c) Bar graphs depicting the sensitivities and specificities of the indicated GPCRs with experimentally validated active ligands and reported decoys. (d) The AUC-ROC curve indicating the model performance in the indicated conditions. (e) Bar graphs indicating the prediction probabilities for each indicated experimentally validated ligand. (f) Schematic workflow illustrates the steps in measuring and comparing the structural conservation of the GPCR-Gα-protein interfaces across human GPCRs. (g) Snakeplot depicting the standard human GPCR two-dimensional sequence level information. Conserved motifs of the GPCR-Gα-protein interfaces are depicted as WebLogo. Asterisks represent residues of conserved motifs present in the GPCRs-Gα-protein interfaces. Of note, the location of the motifs indicated in the exemplary GPCR snake plot is approximated. (h) Schematic workflow illustrates the steps in measuring and comparing the structural conservation of the GPCR-Gα-protein interfaces across human GPCRs. (i) Representative structures of the proteins depicting highly conserved (low RMSD) and highly divergent (high RMSD) GPCR-Gα-protein interfaces. PDB accession numbers are indicated at the bottom. (j) Heatmap depicting the RMSD values obtained by comparing all the GPCR-Gα-protein interfaces of the available human GPCRs from the protein databank. Of note, the RMSD of the Gα-protein cavity was normalized with the RMSDs of the respective whole proteins across all pairwise comparisons. (k) Heatmap depicting the pairwise cosine similarities between the in silico synthesized ligands of the GPCR-Gα-protein interfaces of the available human GPCRs using Gcoupler. (l) Schematic diagram depicting the hypothesis that the intracellular metabolites could allosterically modulate the GPCR-Gα interaction.

Identification of Endogenous, Intracellular Allosteric Modulators of Ste2p Using Gcoupler

(a) Schematic diagram depicting the topology of all the cavities identified using the Synthesizer module of the Gcoupler python package. Of note, the cavity nomenclature includes the cavity location, i.e., EC (extracellular), IC (intracellular), and TM (transmembrane), succeeded by numerical number. (b) Diagram depicting the three-dimensional view of the Ste2 protein, with highlighted Gα-protein binding site (Gpa1) and the Gcoupler intracellular cavities (IC4 and IC5). The venn diagram at the bottom depicts the percentage overlap at the amino acid levels between the Gα-binding site and predicted IC4 and IC5. (c) Schematic representation of the overall workflow used to predict the endogenous intracellular allosteric modulators of Ste2 receptor using Gcoupler and molecular docking technique. Of note, Yeast Metabolome DataBase (YMDB) metabolites were used as query compounds. (d) Overlapping density plots depicting and comparing the distributions of synthetic compounds predicted to target the IC4 and IC5 of the Ste2 receptor using the Gcoupler package. Of note, the Authenticator module of Gcoupler segregated the synthesized compound for each cavity (IC4 or IC5) into High-Affinity Binders (HAB) and Low-Affinity Binders (LAB). (e) AUC (Area under the curve) plots representing the performance of the indicated models. Notably, the models were trained using the cavity-specific synthetic compounds generated using the Gcoupler package. (f) Scatter plots depicting the relationship (correlation) between the binding prediction probabilities using Gcoupler and binding free energies computed using molecular docking (AutoDock). (g) Scatterplot depicting the Pathway Over Representation Analysis (ORA) results of the endogenous metabolites that were predicted to bind to the GPCR-Gα-protein (Ste2p-Gpa1p) interface using both Gcoupler and molecular docking. (h) Alluvial plot showing five level sub-activity spaces screening of the selected metabolites for IC4. (i) Schematic diagram depicting the workflow opted to narrow down on the single metabolic gene mutants. (j) Schematic diagram depicting the experimental design used to screen single metabolic gene mutants for α-factor-induced PCD. Cell viability was assessed using Propidium iodide-based fluorometric assay. (k) Scatter plot depicting the impact of α-factor stimuli on cellular viability, assessed using Propidium iodide-based fluorometric assay. The y-axis represents -log10(p-value) of the one-sample student’s t-test between the normalized PI fluorescence of untreated and treated conditions. The x-axis represents the percentage inhibition or increase in cellular viability, estimated using a Propidium Iodide-based assay. The mutants reported to be involved in mating, PCD, or both are indicated in orange, green, and blue, respectively. The statistically non-significant mutants are indicated below the dashed line in black. (l) Heatmap depicting the relative enrichment/de-enrichment of differentially enriched metabolites in the indicated conditions. Of note, four biological replicates per condition were used in the untargeted metabolomics. (m) Venn diagram depicting the overlap between the predicted endogenous intracellular allosteric modulators of Ste2p and differentially enriched metabolites (DEMs) identified using untargeted metabolomics. (n) Mean-whisker plot depicting the relative abundance of ubiquinone 6 (CoQ6) and zymosterol (ZST) in the indicated conditions. Student’s t-test was used to compute statistical significance. Asterisks indicate statistical significance, whereas ns represents non-significance.

Elevated Endogenous Metabolite Levels Stabilize Ste2p-Gpa1p Interactions and Selectively Inhibit GPCR Signaling

(a) Scheme representing the key steps opted for preparing Ste2p structure for downstream computational analysis. (b) Barplots depicting the binding energies obtained by the docking of Ste2p and indicated metabolites across IC4 and IC5. (c) Line plots depicting the Root Mean Square Deviation (RMSD) changes over simulation timeframes from the three independent replicates of the indicated conditions in the indicated conditions. The spread of the data is indicated as Standard Deviation (SD). Notably, RMSD is provided in Angstroms (Å), whereas the simulation time is in nanoseconds (ns). (d) Workflow depicting the steps involved in Ste2p-miniG-protein docking using HADDOCK and PRODIGY web servers. (e) Barplots depicting the fold change of the dissociation constant (Kd) in the indicated conditions. Notably, fold change was computed with respect to the wild-type condition (Ste2p-miniG-protein). Inlets represent molecular representations of Ste2p-miniG-protein and the highlighted interface residues. (f) The schematic diagram depicting the experimental workflow used to quantify α-factor-induced PCD using a propidium iodide-based cell viability fluorometric assay. Box plot on the right depicting the rescue from the α-factor-induced PCD in the indicated conditions as inferred using propidium iodide-based cell viability fluorometric assay (n=9 or 10 biological replicates; Heatkilled = 2). The y-axis represents the fold change of the propidium iodide fluorescence values with respect to their respective controls. Mann Whitney U test was used to calculate statistical significance. Asterisks indicate statistical significance, whereas ns represents non-significance. (g) Schematic representation (left) of the experimental approach used to measure cell vitality and viability using microscopy-based FUNTM1 staining. Representative micrographs (right) depicting the FUNTM1 staining results in the indicated conditions, Scale 10 µm. Mean-whisker plot depicting the relative proportion of the vital and viable yeast cells observed using FUNTM1 staining in the indicated conditions (n = 3 biological replicates). A Student’s t-test was used to compute statistical significance. Asterisks indicate statistical significance, whereas ns represents non-significance. Error bars represent the standard error of the mean (SEM). (h) Schematic representation (left) of the experimental design for the mating assay (n = 3 biological replicates, each with three technical replicates). MATa yeast cells were preloaded with the metabolites and then mated with MATα cells to evaluate the mating efficiency. Representative micrographs in the middle qualitatively depict the mating efficiency in the indicated conditions. The bar plots on the right depict the mating efficiency (mean ± SEM) in the indicated conditions. Student’s t-test was used to compute statistical significance. Asterisks indicate statistical significance, whereas ns represents non-significance. (i) Schematic representation depicting the experimental design of phospho-MAPK activity-based Western blot. Barplots depicting the p-Fus3 levels (mean ± SEM) in the indicated conditions. Error bars representing standard error of mean (SEM). A Student’s t-test was used to compute statistical significance. Asterisks indicate statistical significance, whereas ns represents non-significance. (j) Schematic representation (left) of the experimental approach used to measure the fluorescence in PFUS1-eGFP-CYC1 yeast cells. Representative micrographs (right) depicting the eGFP expression in the yeast cells in the indicated conditions. Scale 20 µm. Barplot depicting the Corrected Total Cell Fluorescence (CTCF) value (mean ± SEM) in the indicated conditions. A Student’s t-test was used to compute statistical significance. Asterisks indicate statistical significance, whereas ns represents non-significance.

Site-Directed Ste2p Mutants Disrupt Metabolite-Mediated Rescue

(a) Workflow depicting the steps involved in Ste2p-miniG-protein docking of the wild-type and site-directed Ste2p mutants. Notably, docking was performed using HADDOCK and PRODIGY web servers. (b) Barplots depicting the dissociation constant (Kd) fold change in Ste2p site-directed mutants and wild-type. Notably, fold change was computed with respect to the metabolite influenced wild-type condition (Ste2p-miniG-protein). (c) The schematic diagram depicting the experimental workflow used to quantify α-factor-induced programmed cell death in generated site-directed missense mutants (T155D, L289K, S75A), alongside reconstituted wild-type STE2 (rtWT), using a propidium iodide-based cell viability fluorometric assay. The box plot (left) depicts the increase in the relative proportion of dead cells upon α-factor exposure. Box plot (right) depicting the loss of rescue phenotype from the α-factor-induced programmed cell death in the indicated conditions when pre-loaded with metabolites as inferred using propidium iodide-based cell viability fluorometric assay. The y-axis represents the fold change of the propidium iodide fluorescence values with respect to their respective controls. Mann Whitney U test was used to calculate statistical significance. Asterisks indicate statistical significance, whereas ns represents non-significance. (d) Schematic representation (top) of the experimental approach used to measure cell vitality and viability using microscopy-based FUNTM1 staining. Representative micrographs (below) depicting the FUNTM1 staining results in the indicated conditions, Scale 10 µm. Mean-whisker plot depicting the relative proportion of the vital and viable yeast cells observed using FUNTM1 staining in the indicated conditions (n = 4 biological replicates). A Student’s t-test was used to compute statistical significance. Asterisks indicate statistical significance, whereas ns represents non-significance. Error bars represent the standard error of the mean (SEM). (e) Schematic representation (up) depicting the experimental design of phospho-MAPK activity-based Western blot. Barplots (down) depicting the p-Fus3 levels (mean ± SEM) in the indicated conditions. The y-axis represents the p-Fus3/Fus3 ratio for the stimulated condition normalized by its corresponding unstimulated sample. A Student’s t-test was used to compute statistical significance. Asterisks indicate statistical significance, whereas ns represents non-significance. (f) Schematic representation depicting the experimental design of RNA sequencing, featuring treatment duration and the sequencing parameters. (g) Heatmap depicting the expression of differentially expressed genes obtained from RNA sequencing in the indicated conditions. Notably, Control and LST represent yeast cells unloaded and pre-loaded with lanosterol, respectively. α-factor is represented as α, where plus and minus signs represent its presence and absence, respectively. (h) Schematic representation of the experimental workflow followed to deduce the impact of indicated metabolites treatment on isoproterenol (ISO)-induced, GPCR-mediated hypertrophy response in human (AC16) and neonatal rat cardiomyocytes. Notably, in the case of AC16 cells, Wheat germ agglutinin (WGA) was used to stain the cardiomyocytes, whereas, for neonatal cardiomyocytes, alpha-sarcomeric actinin staining was used. (i) Micrographs depicting the human (above; green colored) and neonatal rat (below; red colored) cardiomyocytes in the indicated conditions. Scale 50 µm. (j) Box plots depicting the surface area of human (AC16) and neonatal rat cardiomyocytes in the indicated conditions. Statistical significance of indicated metabolites with untreated control and isoproterenol-treated conditions are indicated in green and grey text, respectively. Mann Whitney U test with Bonferroni corrected p-values was used to compute statistical significance.