Cell‐permeable high‐affinity tracers for Gq proteins provide structural insights, reveal distinct binding kinetics and identify small molecule inhibitors

Background and Purpose G proteins are intracellular switches that transduce and amplify extracellular signals from GPCRs. The Gq protein subtypes, which are coupled to PLC activation, can act as oncogenes, and their expression was reported to be up‐regulated in cancer and inflammatory diseases. Gq inhibition may be an efficient therapeutic strategy constituting a new level of intervention. However, diagnostic tools and therapeutic drugs for Gq proteins are lacking. Experimental Approach We have now developed Gq‐specific, cell‐permeable 3H‐labelled high‐affinity probes based on the macrocyclic depsipeptides FR900359 (FR) and YM‐254890 (YM). The tracers served to specifically label and quantify Gq proteins in their native conformation in cells and tissues with high accuracy. Key Results FR and YM displayed low nanomolar affinity for Gαq, Gα11 and Gα14 expressed in CRISPR/Cas9 Gαq‐knockout cells, but not for Gα15. The two structurally very similar tracers showed strikingly different dissociation kinetics, which is predicted to result in divergent biological effects. Computational studies suggested a “dowel” effect of the pseudoirreversibly binding FR. A high‐throughput binding assay led to the discovery of novel Gq inhibitors, which inhibited Gq signalling in recombinant cells and primary murine brown adipocytes, resulting in enhanced differentiation. Conclusions and Implications The Gq protein inhibitors YM and FR are pharmacologically different despite similar structures. The new versatile tools and powerful assays will contribute to the advancement of the rising field of G protein research.

Upon activation by a GPCR, the α-subunit releases GDP, binds GTP instead and dissociates from the βγ subunits. There are four different G protein subfamilies, G s , G i , G q and G 12/13 , that interact with different second messenger systems. Gα q proteins activate PLC β which leads to the release of inositol 1,4,5-trisphosphate (IP 3 ) and subsequent calcium mobilization resulting in a number of intracellular effects. Four different Gα q proteins exist: Gα q , Gα 11 , Gα 14 and Gα 15/16 (Kamato et al., 2017;Mizuno & Itoh, 2009).
YM is produced by Chromobacterium sp. (Taniguchi et al., 2003), while FR was isolated from the plant Ardisia crenata SIMS and is produced by the bacterial endophyte Candidatus Burkholderia crenata that is present as a symbiont in the leaves of the plant Fujioka, Koda, & Morimoto, 1988). A few analogues of FR have also been isolated, however, in tiny amounts Reher et al., 2018). Recently, the total syntheses of 1 and 2 and some analogues were described, but they represent labour-intensive procedures providing only small amounts of the products; all of the synthesized analogues showed moderate potency or were inactive (Xiong et al., 2019;Zhang et al., 2017). In functional studies, FR and YM were found to be similarly potent and selective Gα q/11 protein inhibitors. Both are exceedingly useful for studying G q protein signalling and for dissecting signalling pathways (Inamdar, Patel, Manne, Dangelmaier, & Kunapuli, 2015;Roszko et al., 2017;Schrage et al., 2015). However, more readily available inhibitors would be highly desirable. Moreover, such compounds may have potential as drugs, for example, for the treatment of chronic pulmonary disease (Matthey et al., 2017) and certain types of cancer (Feng et al., 2014).
Currently, no tools or methods are available to directly label G q proteins in their native conformation, which would enable a number of biological and clinical applications. In the present study, we developed G q -specific high-affinity chemical probes for detection and quantification with high accuracy and demonstrate their power for labelling of G q proteins in a variety of organs, cells and tissues. We show that the tracers derived from the structurally related G q inhibitors FR and YM are in fact strikingly different possessing extremely divergent binding kinetics, which is anticipated to translate into disparity regarding their pharmacological actions. Moreover, we established a high-throughput binding assay that has led to the discovery of small molecule G q protein inhibitors, one of which-Ebselen-was further characterized and found to inhibit G q signalling in recombinant cells and native brown adipocytes leading to enhanced adipocyte differentiation.

| Synthesis of radiotracers
The radioligands were prepared by Quotient Bioresearch, now Pharmaron (Cardiff, UK), by catalytic hydrogenation of YM and FR with tritium gas (custom synthesis). YM was used for preparing

What is already known
• G q proteins can act as oncogenes and their expression is up-regulated in cancer and inflammation.

What this study adds
• Tritium-labelled radiotracers for Gq proteins were developed leading to the discovery of novel Gq inhibitors.
• Hydrogenated FR and YM are structurally similar Gq inhibitors differing dramatically in their residence time.
What is the clinical significance • The developed Gq tracers may be useful for diagnostic purposes, for example in cancer.
• A small molecule Gq inhibitor resulted in enhanced differentiation of brown adipocytes.

| Preparation of intact human platelet suspensions
Apheresis-purified human thrombocyte-plasma concentrates (PRP) were obtained from the blood bank, University of Bonn, in accordance with ethical guidelines, and directly used. For radioligand binding assays, they were diluted with 50-mM Tris-HCl buffer, pH 7.4, to obtain a concentration of 7.5 × 10 6 platelets/50 μl (corresponding to a 1:10 dilution).

| Platelet membrane preparation
Apheresis-purified human thrombocyte-plasma concentrates were centrifuged (1,000 g at 4 C for 10 min), and the supernatant was decanted and centrifuged again (at 48,400 g, 4 C, for 60 min). The supernatant was kept and used as platelet-poor plasma for control experiments; the formed pellet was resuspended in buffer 1 (50-mM Tris-HCl, 5-mM EDTA, and 150-mM NaCl) and centrifuged at 48,400 g, 4 C for 60 min. The supernatant was discarded again and the pellet was resuspended in buffer 2 (5-mM Tris-HCl and 5-mM EDTA). The cell suspension was homogenized using an Ultra-Turrax ® (IKA Laborechnik, Staufen, Germany) for 1 min at a speed level of 4, subsequently transferred into cryovials, and stored at −80 C before use (the preparation is stable for several months). For radioligand binding assays, 50 μg of protein per vial was used, determined by the method of Bradford (Bio-Rad Protein Assay).
2.5 | Retroviral transfection of Gα q -knockout HEK293 cells with Gα 11 , Gα 14 , Gα 15 and Gα q proteins cDNAs encoding for the α-subunits of the human G 11 , G 14 , G 15 and mouse G q guanine nucleotide-binding proteins in a pcDNA3.1(+) vector were obtained from E. Kostenis. CRISPR/Cas9-Gα q -knockout of the human and mouse Gα q proteins differ only by one amino acid in position 171 (alanine in human and serine in mouse). The mouse Gα q protein contained an internal hemagglutinin (HA) tag introduced through the following modifications: E125D, N126V, Y128D, V129Y and D130A. Retroviral transfection was performed as previously described (Hillmann et al., 2009). Briefly, the coding sequence of the human Gα 11 , Gα 14 , Gα 15 and mouse Gα q proteins was cloned into the pQCXIN retroviral expression vector, amplified, purified and sequenced prior to the transfection of GP + envAM-12 packaging cells together with vesicular stomatitis virus G protein DNA using lipofectamine 2000. After 16 hr, 3 ml of DMEM containing 10% FBS, 1% of a penicillin/streptomycin solution (final concentrations: penicillin = 100 UÁml −1 , streptomycin = 0.1 mgÁml −1 ), and sodium butyrate (5 mM) was added to the packaging cells. They were kept at 32 C and 5% CO 2 for 48 hr, during which the viral vectors containing the receptor sequence were produced and released into the surrounding medium. These were harvested, filtered (45-μm filter pore diameter), and added to HEK293 cells that had previously been modified via CRISPR/Cas9 genome editing to lack Gα q expression. Polybrene solution (6 μl, 4 mgÁml −1 in H 2 O, filtered) was added. After 2.5 hr, the virus-containing medium was discarded, and DMEM supplemented with 10% FBS and 1% of a penicillin/streptomycin solution (final concentrations: penicillin = 100 UÁml −1 , streptomycin = 0.1 mgÁml −1 ) was given to the cells. These were incubated for 2 days, followed by selection of successfully transfected cells with geneticin resistance by adding G418 (200 μgÁml −1 ) to the medium. The cells were cultured in these media at 37 C and 10% CO 2 until membrane preparations were generated.

| Membrane preparations of recombinant HEK cells, rodent and human cancer cell lines
Recombinant HEK cells were cultured to 70% of confluence. The culture medium was discarded, and cells were washed once with approximately 5 ml of PBS. Dishes were stored at −20 C until further use.
After defrosting, cells were detached with a rubber scraper after adding 5-mM Tris-HCl, 2-mM EDTA, pH 7.4. The collected cell suspension was homogenized using an Ultra-Turrax ® for 1 min at a speed level of 4. The cell suspension was transferred into tubes and centrifuged for 10 min at 1,000 g, 4 C. The pellet (P1) was discarded, and the supernatant was centrifuged for 45 min at 48,400 g, 4 C. The resulting supernatant was discarded, and the pellet (P2) was resuspended in 50-mM Tris-HCl, pH 7.4, and centrifuged again for 45 min at 48,400 g, 4 C. This step was repeated once again, and the suspension was subsequently homogenized using an Ultra-Turrax ® for 1 min at a speed level of 4 (IKA Laborechnik), transferred into cryovials, and stored at −80 C (for up to several months). Protein determination was performed by the method of Bradford (Bio-Rad Protein Assay). Membrane preparations of rodent and human cancer cell lines were obtained according to the same method as described for HEK cells.

| Rat brain membrane preparations
Rat brains obtained from Pel-Freez ® (Rogers, Arkansas, USA) were defrosted in sucrose (0.32 M) and dissected to obtain cortex and striatum, respectively. Membrane preparations were prepared, P1 from striatum, P2 from cortex, according to the methods described above for HEK cells.

| Mouse tissue membrane preparations
Female CD1 mice (Janvier Labs, Le Genest-Saint-Isle, France) were housed under normal light-dark cycles with ad libitum supply of chow and water. At the age of 10-12 weeks, healthy animals were killed by cervical dislocation, and the respective organs (kidney, liver, lung, heart and brain) were dissected, snap-frozen in liquid nitrogen, and stored at −80 C until use. Each independent experiment was conducted with tissue from an individual animal. The tissue samples were weighed and subsequently homogenized in 50-mM Tris-HCl buffer, pH 7.4, using a bead mill (TissueLyser LT, Qiagen, Venlo, NL) for 15 min 50 strokes per second, and protein determination was performed by the method of Bradford (Bio-Rad Protein Assay). The lysate was transferred into cryovials and stored at −80 C.

| Protein determination
Determination of proteins in solutions was carried out by the Bradford method. The stock solution (Bio-Rad Protein Assay) was diluted 1:5 in H 2 O, at a ratio of 50:1, added to the protein solution, and incubated for 5 min at room temperature. The protein concentrations were determined photometrically at 595 nm (using a DU ® 530 spectrophotometer, Beckmann-Coulter, Krefeld, Germany). As a reference, a standard dilution of BSA in 50-mM Tris-HCl, pH 7.4, buffer was prepared and measured, and a regression curve for the quantification of protein concentration was calculated and applied.

| Competition binding studies
The experiments were performed in 50-mM Tris-HCl buffer, pH 7.4.
The final assay volume of 200 μl contained 5 μl of test compound in DMSO, 50 μl of [ 3 H]PSB-15900 or [ 3 H]PSB-16254, respectively, in buffer, 5 nM of final concentration, and 100 μl of human platelet membrane preparation (50 μg of protein) or intact human platelets (7.5 × 10 6 cells/vial). Non-specific binding was determined in the presence of 5-μM FR. The incubation was started by the addition of membrane preparation and was performed at 37 C for 1 hr with gentle shaking. It was terminated by rapid vacuum filtration through GF/C glass-fibre filters using a Brandel 24-well harvester. The filters were rinsed three times with approximately 3 ml each of ice-cold Tris-HCl buffer, 50 mM, pH 7.4 containing 0.1% Tween 20 and 0.1% BSA, to separate bound from free radioligand. The filters were punched out and transferred to scintillation vials. Luma Safe ® scintillation cocktail (2.5 ml) was added, and after 6 hr of incubation, the samples were counted for 1 min each, using a liquid scintillation counter (counting efficiency 53%). Experiments in the presence of GTP were additionally performed after preincubation of the membrane preparation with GTP for 1 hr. pIC 50 and pK i values were calculated using equations for one-site competition as implemented in GraphPad Prism 6.01 (GraphPad Inc., La Jolla, CA, USA).

| Kinetic binding experiments
Association experiments were performed by incubation of platelet membrane preparations (50 μg of protein/vial) or intact platelets (7.5 × 10 6 cells/vial) with [ 3 H]PSB-15900 or [ 3 H]PSB-16254, respectively, 10 nM of final concentration, at 0 C, room temperature (21 C), or 37 C, respectively, for different periods of time ranging from 1 to 420 min depending on the incubation temperature. Incubation was performed for 3 hr at 0 C, for 1.5 hr at 21 C, and for 1 hr at 37 C to reach binding equilibrium. Dissociation was subsequently initiated by the addition of FR (5 μM). All other conditions and the remaining procedure were as described above for competition experiments. The presented association half-lives were not corrected for radioligand dissociation. For data evaluation, raw data (in cpm) from kinetic binding experiments were taken, nonspecific binding was subtracted, and data were normalized to percentage of specific binding at time 0 (= 100%), while 0 cpm corresponded to 0%. The "one phase association" equation (Y = Y max * [1 − exp(−k obs * X)]) implemented in GraphPad Prism was employed to determine t 1/2 from k obs . For dissociation experiments, the value at time 0 was defined as 100%, and 0 cpm were defined as 0%. The "one phase exponential decay" equation Prism was used to determine t 1/2 and k off .

Different concentrations of [ 3 H]PSB-15900 or [ 3 H]PSB-16254,
respectively, ranging from 1 to 60 nM were incubated with platelet membrane preparations or intact platelets for 3 hr at 0 C, 1.5 hr at 21 C, or 1 hr at 37 C. All other conditions and procedures were as described above for competition experiments. Specific binding (cpm) was calculated by subtracting non-specific binding from total binding. Total and specific binding were fitted in GraphPad Prism 6.01 (GraphPad Inc.) with the equations "One site -Total" (Y = (B max * X)/(K D +X)+NS * X) and "One site -Specific" (Y = (B max * X)/(K D +X)), respectively, while non-specific binding was fitted as a linear function. The pK D values shown in Table 1 represent the means ± SD of all individual experiments, obtained from the "One site -Specific" fit. All cpm values were converted into B max values considering the specific activity of the radioligands and the amount of protein in the assay tubes.

| High-throughput screening assay
Competition binding experiments were performed as described above (for incubation conditions, washing steps, and buffers) but on 96-well plates using a Brandel 96-well plate harvester with GF/C Unifilters (Perkin-Elmer). After harvesting, 50 μl per cavity of Ultima Gold scintillation cocktail (Perkin-Elmer) was added, and counting was performed as described above. FR (1 μM) was used as a positive control showing complete inhibition.

| Calcium mobilization assays
The functional activity of YM, FR, and Ebselen as G q protein inhibitors was determined by assessing their ability to antagonize ADP-induced T A B L E 1 Affinities of tracers and expression levels of G q family (Gα q/11/14 ) proteins determined by saturation binding expressed as pK D ± SD

[ 3 H]PSB-15900
Tissue and temperature pK D ± SD (nM) B max ± SD (pmolÁmg −1 of protein)  Figure S4. calcium mobilization in 1321N1 astrocytoma cells stably transfected with the human P2Y 1 receptor, which is a G q -coupled receptor. In addition, the compounds' inhibitory effects on Carbachol-induced calcium mobilization were studied in the same cell line, which endogenously express the G q -coupled muscarinic M 3 ACh receptor. The assays were performed according to a previously described procedure (Rafehi et al., 2017) using a NOVOstar microplate reader (BMG Labtech, Offenburg, Germany).

| Animal studies
Animal studies are reported in compliance with the ARRIVE guidelines (Kilkenny, Browne, Cuthill, Emerson, & Altman, 2010;McGrath & Lilley, 2015) and with the recommendations made by the British Journal of Pharmacology.

| Analysis of brown adipocytes
Ten-to 12-week-old pregnant wild-type C57Bl6/J mice were purchased from Charles River and housed in pathogen-free environment under a 12-hr light/12-hr dark cycle having free access to standard rodent diet and water. Brown adipose tissue (BAT)-derived mesenchymal stem cells were isolated from interscapular BAT of newborn wildtype mice and further cultured as previously described (Klepac et al., 2016). Specifically, 170,000 cells per well were seeded in growth medium (GM; DMEM supplemented with 10% FBS and 1% penicillin-

| Docking studies
The crystal structure of the heterotrimeric G protein (Gα i/q βγ) with the G q -selective inhibitor YM (3AH8.pdb, resolution 2.9 Å) was obtained from the Protein Data Bank (Berman et al., 2000).  et al., 2009). As an initial step, the co-crystallized water molecules and the ligand were removed from the X-ray structure. The atomic partial charges were added using AutoDockTools (Morris et al., 2009;Sanner, 1999). Three-dimensional energy scoring grids of 60 × 60 × 60 points with a spacing of 0.375 Å were computed, and the grids were centred based on the cocrystallized ligand, YM-254890. During the docking simulations, the ligands were fully flexible while the residues of the receptor were treated as rigid. Fifty independent docking calculations using the varCPSO-ls algorithm from PSO@Autodock implemented in AutoDock4.2 were performed and terminated after 500,000 evaluation steps (Namasivayam & Gunther, 2007 (Huang et al., 2017).
Ligand parameters were generated using the ParamChem service (https://cgenff.paramchem.org) implemented in CHARMM-GUI. The prepared complexes were equilibrated using NAMD (Phillips et al., 2005) over 5 ns using the input files generated from CHARMM-GUI.
During the simulations, non-bonded interactions were gradually switched off at 10 Å, and the long-range electrostatic interactions were calculated using the Particle-mesh Ewald method (Darden, Your, & Pedersen, 1993). The temperature was maintained at 303.15 K using the Langevin thermostat, and the pressure was

| Statistical analysis and randomization
The data and statistical analysis comply with the recommendations of the 3 | RESULTS

| Preparation of FR-and YM-derived tracers
A prominent structural feature of both YM (1) and FR (2; Figure 1a) is an exocyclic double bond conjugated to a carbonyl group (see shown by X-ray crystallography (Nishimura et al., 2010); the double bond is merely involved in hydrophobic interactions. When analysing the structure of FR, we figured that the double bond could undergo catalytic hydrogenation and that this reaction might be utilized for the labelling of the compound.
In fact, hydrogenation of FR and YM was reported to lead to derivatives that were still capable of inhibiting G q signalling, although with somewhat lower potency (Schrage et al., 2015;Taniguchi et al., 2004). Thus, we decided to perform catalytic hydrogenation of the double bond of FR and YM with tritium gas (Figure 1b), and the resulting mixtures of tritium-labelled stereoisomers (3/5 and 4/6), Rand S-configurated at the new stereocentre, were separated by HPLC coupled to radio and diode array detectors and analysed by MS ( Figures S1 and S2) showed high-affinity binding, while much lower specific binding was seen with the other stereoisomer (see Figure 1c). The eutomers were assumed to be the R-configurated epimers based on the study by Taniguchi et al. (2004).

| Development of binding assays
Native cell membrane preparations from rat brain cortex and human blood platelets, which were expected to express high G q protein levels (Inamdar et al., 2015;Kawasaki et al., 2005;Kamato et al., 2017;Offermanns, 2000;Taniguchi et al., 2003) Table 1). The pK D value determined for [ 3 H]PSB-15900 at human platelet membranes was almost identical to that measured at rat brain cortical membranes, which can be explained by the remarkably high conservation of the Gα q protein sequences in different species (Mizuno & Itoh, 2009 Figure 2a,b). These assays allowed the accurate determination of expression levels of G q proteins with high accuracy (see Table 1). In human platelet membranes, an expression level of approximately 5 (3.86-7.09) pmolÁmg −1 of total protein amount was determined. The expression level of Gα q proteins in rat brain cortical membranes was determined to be about fourfold higher (B max 19.7 pmolÁmg −1 protein). Saturation assays at intact human platelets using [ 3 H]PSB-15900 resulted in a comparable pK D value (8.00), and 10,600 ± 2,000 binding sites (Gα q molecules) per platelet were detected (see Figure S4).  their interactions within the binding pocket were compared. These docking studies allowed us to speculate that, due to its additional lipophilic "handles," FR is anchored in the binding pocket like a dowel forming a latch, while YM lacks those anchor points and can therefore be more readily released (see Figure 4).

|
The docking studies also indicated that the large differences in binding affinities between the hydrogenated epimers (3/5, 4/6, see

| Molecular dynamics simulation
To test our hypotheses, we performed MD simulations of the G q protein complexes bound to either YM or FR as a next step ( Figure S7a,b).
The root mean square deviation (RMSD) values of the C α atoms of the heterotrimeric complex rapidly reached equilibrium with approximately 1 Å deviation from the first frame of 200 ns of MD simulation ( Figure S7a). The interaction pattern that emerged from the trajectory visualizations (see selected runs in Videos S1 and S2) shows that both ligands were anchored in the binding site cleft by their phenyl ring establishing a strong hydrophobic interaction with the residues Ile56, Val184, and Ile190 (for sequence alignment of G q proteins and amino acid residues involved in YM and FR binding, see Figure S8)

| Competition binding assays
Both tracers were subsequently employed for competition binding studies on human platelet membranes (at 37 C). Selected competition curves are displayed in Figure 5a, Table 1).
Pseudohomologous competition binding will be useful to simplify the procedure for estimating pK D and B max values in different cells and tissues.
Binding studies were also performed in living cells. PRP showed high specific binding of [ 3 H]PSB-15900, while plasma depleted from platelets did not display any specific binding (see Figure 5c). Non-  Figure S9a,b), which were well in agreement with previously determined pK D values.

| Effects of potential G q protein modulators on [ 3 H]PSB-15900 binding
Next, we studied whether binding of the G q tracer [ 3 H]PSB-15900 would be modulated by metal cations, nucleotides, phospholipids, agonists of G q protein-coupled receptors or the known G protein inhibitors BIM-46187 (BIM dimer) and BIM-46174 (BIM monomer; Ayoub et al., 2009;Schmitz et al., 2014). Effects of sodium, lithium, magnesium and calcium chloride (10 and 100 mM) on the binding of  Figure S12).
Agonists of G q -coupled GPCRs are signalling by inducing a conformational change of G q proteins and thus it might be possible that they affect the binding of the G q inhibitor [ 3 H]PSB-15900. Therefore, we investigated agonists of G q -coupled receptors that are known to be highly expressed in platelets (Koupenova & Ravid, 2018;Offermanns, 2000), including PAR-1, TXA 2 , 5-HT 2A , adenosine A 2B , and P2Y 1 receptor agonists, using membrane preparations (see Figure S13). No modulation was observed by any of the employed GPCR agonists, except for a minor effect of the PAR-1-selective peptide agonist TFLLR-NH 2 (16% inhibition at 100 μM).
The G protein inhibitor BIM-46187 (BIM dimer), a symmetrical disulfide, and its corresponding monomeric thiol (BIM-46174; for structures, see Figure S14) were previously shown to directly inhibit G q proteins (Schmitz et al., 2014); however, their binding site is still

unknown. Binding studies on human platelet membranes versus [ 3 H]
PSB-15900 (preincubation with BIM molecules for 3 hr) did not show any inhibition or modulation of radioligand binding at concentrations ranging up to 100 μM (see Figure S14).

Our results show that binding of the FR-derived tracer [ 3 H]PSB-
15900 is not allosterically modulated by compounds that are known to bind directly to G q (e.g. GTP and BIM) or that affect the conformation of G q proteins (e.g., GPCR agonists).

| G q subtype selectivity
Subsequently, we studied the G q protein selectivity of both [ 3 H]PSB-15900 and [ 3 H]PSB-16254 by utilizing HEK (HEK293) cells whose Gα q proteins had been knocked out by CRISPR/Cas9 gene editing.
Using retroviral transfection (Hillmann et al., 2009), the different Gα q protein subtypes, Gα q , Gα 11 , Gα 14 and Gα 15 , were stably expressed in those HEK293-G q -KO cells. [ 3 H]PSB-15900 showed no specific binding to membrane preparations of HEK293-Gα q -KO cells (see Figure 5d,e), which demonstrates its selectivity for Gα q proteins. In contrast, it displayed high specific binding to cell membranes recombinantly expressing Gα q , Gα 11 or Gα 14 protein, but not to those that expressed Gα 15 (see Figure 5d,e). The same results were observed with [ 3 H]PSB-16254 (Figure 5f,g). The affinities of FR and YM for the three Gα q proteins were very similar (pIC 50 and pseudo-B max values are collected in Table S2).

| Quantification of G q proteins in different cells and tissues
In a next step, we utilized the new probes for the determination of Gα q protein expression levels in different human and rodent cells, tissues, and organs (Figure 5h). Since the sequences of rat, mouse, and human Gq proteins are highly conserved, the data are directly comparable. High expression was detected in mouse liver, lung and brain, while lower levels were observed in kidney and especially in the heart.
Human platelets showed a comparably high expression level. Some cancer cell lines, for example, Jurkat T cells, a human T-cell leukaemia cell line, and human 1321N1 astrocytoma cells, were found to exhibit particularly high G q expression levels (Figure 5h).

| High-throughput screening
Since the natural products YM and FR are complex molecules that are not easily accessible in large quantities, we established a highthroughput screening assay for the screening of compound libraries with the aim to identify starting points for novel, more readily accessible G q protein inhibitors. Thus, a binding assay with [ 3 H]PSB-15900 was established in a 96-well format (for details, see Section 2). The assay was validated by determining the pK i value of FR, which was identical to the previously determined pK i value ( Figure S15). A Z 0 value of 0.69 was calculated for the assay indicating its suitability for high-throughput screening (a value of 0.5 or greater is required; Zhang, Chung, & Oldenburg, 1999). We subsequently screened part of our in-house compound collection, initially 2,400 compounds, which resulted in two hits (see Figure S16). One of the hit molecules, Ebselen (Noguchi, 2016; for structure, see Figure 6a)

| Effects of the discovered Gq inhibitor Ebselen on brown adipocytes
Since the G q signalling pathway has been described to be involved in the differentiation of BAT (Klepac et al., 2016), we used primary murine brown adipocytes to study Ebselen's G q inhibitory effect. G q signalling in brown adipocytes can be quantified by measuring IP 1 production (Klepac et al., 2016). To specifically activate G q signalling, we used brown adipocytes transduced with a G q -coupled designer GPCR, a modified M 3 muscarinic receptor (Dq) that is activated by otherwise pharmacologically inactive CNO (Conklin et al., 2008). Ebselen significantly reduced the CNO-induced increase in IP 1 levels demonstrating its ability to inhibit G q signalling also in intact cells, albeit not as strongly as FR (Figure 6e). In addition, we studied the effect of Ebselen on endogenous G q activators using ET-1, which we previously identified as an autocrine regulator of G q signalling in brown adipocytes (Klepac et al., 2016). ET-1 increased IP 1 production, which was also attenuated by Ebselen pretreatment (Figure 6f)

| Selective tracers and assays for Gα q proteins
The structurally related macrocyclic depsipeptides YM and FR, which are produced by bacteria, were previously found to be potent, selective inhibitors of Gα q proteins (Gao & Jacobson, 2016;Inamdar et al., 2015;Kamato et al., 2017;Kukkonen, 2016 ;Schrage et al., 2015). We have now prepared tritium-labelled hydrogenated derivatives of FR and YM and developed assays to address important biological questions. Tracers with high specific activity (37-77 CiÁmmol −1 ), indicating the average introduction of 1.3-2.6 tritium atoms per molecule were obtained. The more potent stereoisomers or eutomers of both tracers (presumably Rconfigurated) were isolated in high purity and in contrast to previous studies (Schrage et al., 2015;Taniguchi et al., 2004), they were found to be equipotent to the parent compounds YM and FR. This may be explained by a moderate purity of previously studied nonradiolabelled hydrogenation products.
With these new tool compounds, the direct interaction of compounds with G q proteins was measured in competition assays using cell membrane preparations. They provide an important complement to functional studies, which are indirect measurements that can be influenced by many factors, such as the ability to penetrate cells or various off-target effects. The tracers bind with high affinity and selectivity to their targets, Gα q , Gα 11 and Gα 14 , but not to Gα 15 , which unambiguously answers the question of the Gα q subtype selectivity of FR and YM (Kukkonen, 2016;Schrage et al., 2015). This is due to the high homology of Gα q , Gα 11 , and Gα 14 whose binding site for YM and FR is virtually identical, while that of the Gα 15 protein shows lower homology (for sequence alignment, see Figure S8).
The established assays are anticipated to be highly useful for structural studies on G q proteins and their complexes in order to assess and confirm the correct folding of purified G q proteins used, for example, for X-ray crystallography, NMR, or electron cryomicroscopy studies. Importantly, the tracers readily penetrate cell membranes as demonstrated for human platelets, and binding kinetics were similar in membranes as compared to intact cells (Figure 3e,f, Table S1, and Figure S9 Figure S4). Human blood platelets, for example, were determined to express 10,600 G q protein molecules per cell or approximately 4-7 pmolÁmg −1 of protein (Table 1). In rat brain cortical membranes, the Gq expression level was about fourfold higher (20 pmolÁmg −1 protein). This is significantly higher than the typical expression levels of GPCRs in peripheral or brain cells. We observed that some cancer cells, for example, Jurkat T cells, express particularly high Gα q levels and it is known that Gα q and Gα 11 proteins can act as oncogenes (Chiariello, Marinissen, & Gutkind, 2015). Future comprehensive studies on different cancer cells and tissues, as well as immune cells, isolated from cancer patients, will be highly useful to elucidate the role of G q proteins in cancer progression. Moreover, the tracers and the established assays may be used for diagnostic purposes and to test whether G q proteins could serve as biomarkers for precision medicine. Tumours with high G q levels may profit from treatment with G q protein-inhibiting drugs. Gα q expression levels may also be altered in autoimmune diseases, such as rheumatoid arthritis (Zhang & Shi, 2016), and the expression level of Gα q proteins appears to be related to inflammatory diseases, for example in the lung (John et al., 2016), to pathological conditions of the skin (Doçi et al., 2017) and even to heart diseases (Frey et al., 2017).  (Rensing, Uppal, Blumer, & Moeller, 2015;Xiong et al., 2019;Zhang et al., 2017), and by biotechnological approaches Taniguchi et al., 2003). The new tracers will help to design and develop G q inhibitors with specific, desired kinetic profiles. This resulted in only two hits (0.08% hit rate), one of which, Ebselen, has been further characterized.

| Ebselen blocks Gq functions in recombinant cells and native adipocytes
Our results show that the established assay is suitable for identifying novel G q protein inhibitors. Ebselen is a thiol-reactive multi-target drug (Noguchi, 2016) that has not been previously reported to interact with heterotrimeric G proteins. It may bind to a free cysteine residue present in the Gα q protein near the binding pocket of FR and thereby block the binding of the tracer. Moreover, it also inhibits G q protein function as shown in calcium mobilization assays, in which G q was activated via two unrelated GPCRs, the M 3 and the P2Y 1 receptor. Also in intact primary brown adipocytes, Ebselen decreased G q DREADD-induced as well as ET-1-induced IP 1 production. Like FR, Ebselen induced an increase in the differentiation of brown adipocytes.
Although Ebselen did not affect basal IP 1 levels, it increased differentiation of brown adipocytes. This difference might be explained by the different time scales of the two experiments: IP 1 assays were performed in washed preadipocytes over a short period of about 2 hr (acute treatment). The extracellular environment is likely devoid of any G q stimulus in this setting, and, therefore, Ebselen would not affect basal IP 1 production. However, differentiation experiments were performed over a period of 11 days (chronic treatment). Brown adipocytes are known to secrete a variety of factors during differentiation, which signal via autocrine/paracrine loops (Ali Khan et al., 2018). We previously showed that ET-1 is one factor that is secreted by cultured brown adipocytes, which strongly inhibits differentiation in an autocrine manner via the activation of the Gq-coupled endothelin-A receptor (Klepac et al., 2016).
One would therefore anticipate an inhibition of adipogenic differentiation by the application of an antioxidant molecule. In fact, other antioxidants such as N-acetyl-L-cysteine and Tempol blocked adipogenic differentiation in 3T3-L1 cells (Samuni et al., 2010;Vigilanza, Aquilano, Baldelli, Rotilio, & Ciriolo, 2011). As Ebselen rather increased the adipogenic differentiation in our experiments, this effect should not be caused by its antioxidant properties but could be due to G q protein inhibition.
In conclusion, we have developed probes and assays which have provided new insights and will further advance the field of GPCR and G protein research in general, and that of G q proteins in particular.
Future studies will be directed towards investigating the molecular mechanism of G q inhibition by Ebselen and explore its suitability as a lead structure for developing potent and selective inhibitors for G q proteins and eventually for other G protein subtypes.
to compound screening, and we thank Jessica Nagel for her support within the framework of her master thesis. Finally, we would like to like to thank E. Kostenis for cDNAs encoding for the α-subunits of the human G 11 , G 14 , G 15 , and mouse G q guanine nucleotide-binding proteins.