A Dynamic, Split-Luciferase-Based Mini-G Protein Sensor to Functionally Characterize Ligands at All Four Histamine Receptor Subtypes

In drug discovery, assays with proximal readout are of great importance to study target-specific effects of potential drug candidates. In the field of G protein-coupled receptors (GPCRs), the determination of GPCR-G protein interactions and G protein activation by means of radiolabeled GTP analogs ([35S]GTPγS, [γ-32P]GTP) has widely been used for this purpose. Since we were repeatedly faced with insufficient quality of radiolabeled nucleotides, there was a requirement to implement a novel proximal functional assay for the routine characterization of putative histamine receptor ligands. We applied the split-NanoLuc to the four histamine receptor subtypes (H1R, H2R, H3R, H4R) and recently engineered minimal G (mini-G) proteins. Using this method, the functional response upon receptor activation was monitored in real-time and the four mini-G sensors were evaluated by investigating selected standard (inverse) agonists and antagonists. All potencies and efficacies of the studied ligands were in concordance with literature data. Further, we demonstrated a significant positive correlation of the signal amplitude and the mini-G protein expression level in the case of the H2R, but not for the H1R or the H3R. The pEC50 values of histamine obtained under different mini-G expression levels were consistent. Moreover, we obtained excellent dynamic ranges (Z’ factor) and the signal spans were improved for all receptor subtypes in comparison to the previously performed [35S]GTPγS binding assay.


Introduction
G protein-coupled receptors (GPCRs) transduce external stimuli to intracellular events by the activation of heterotrimeric G proteins. Upon receptor activation, the heterotrimeric G protein binds to the receptor, which is followed by a GDP-GTP nucleotide exchange at the Gα subunit. The resulting conformational change of Gα promotes the uncoupling of the G protein from the receptor and the dissociation of the heterotrimer into a Gα monomer and a Gβγ dimer [1,2]. Both are then capable to modulate effector proteins inside the cell. Canonical GPCR-mediated signaling is determined by Gα, the subtypes of which target different membrane-bound effectors, such as phospholipase C [3,4] (PLC) and adenylyl cyclase [5,6] (AC). In drug discovery, GPCRs are the most studied drug targets and are addressed by more than 30% of approved drugs [7]. Fundamental criteria for successful drugs are a high binding affinity and potency at the target receptor, as well as a distinct pharmacological action ((full,

Principle and Characteristics of the Mini-G Protein Recruitment Assay
To study the G protein signaling of histamine receptor ligands, we applied the split-NanoLuc technology [29] to the human histamine receptor subtypes H 1 , H 2 , H 3 or H 4 (NlucC) and the corresponding mini-G proteins mGsq, mGs and mGsi (NlucN) (Supplementary Figure S1). Upon receptor activation, the mini-G protein was recruited by the receptor leading to the formation of a functional NanoLuc ( Figure 1A). Thus, agonist concentration-dependent luminescence signals were obtained in the presence of the substrate furimazine ( Figure 1B). To investigate antagonists, the response of the reference agonist histamine at EC 80 concentration (H 1 R: 10 µM, H 2-4 R: 1 µM) was measured after a pre-incubation period of the respective antagonists. In order to verify the histamine receptor expression, radioligand saturation binding experiments were performed, and adequate binding of [ 3 H]mepyramine to the H 1 R co-expressed with mGsq, [ 3 H]UR-DE257 to the H 2 R co-expressed with mGs and [ 3 H]UR-PI294 to the H 3 R and H 4 R each co-expressed with mGsi were observed (Supplementary Figure S2 and Table S1). functional NanoLuc ( Figure 1A). Thus, agonist concentration-dependent luminescence signals were obtained in the presence of the substrate furimazine ( Figure 1B). To investigate antagonists, the response of the reference agonist histamine at EC80 concentration (H1R: 10 µM, H2-4R: 1 µM) was measured after a pre-incubation period of the respective antagonists. In order to verify the histamine receptor expression, radioligand saturation binding experiments were performed, and adequate binding of [ 3 H]mepyramine to the H1R co-expressed with mGsq, [ 3 H]UR-DE257 to the H2R coexpressed with mGs and [ 3 H]UR-PI294 to the H3R and H4R each co-expressed with mGsi were observed (Supplementary Figure S2, Supplementary Table S1). Scheme of the mini-G protein recruitment assay. The split-NanoLuc technology was applied to the H1-4R (C-terminus) and the mini-G proteins (mG; N-terminus). Upon receptor activation, the mini-G protein is recruited to the GPCR and the split-NanoLuc fragments form a functional enzyme leading to the oxidation of the substrate and thus luminescence signals in an agonist concentration-dependent manner. (B) Representative luminescence traces of the mini-G protein recruitment of mGsq to H1R, mGs to H2R and mGsi to H3R and H4R. Baseline and inter-well corrected luminescence traces of histamine at various concentrations and the assay medium Leibovitz's L-15 (L-15) as negative control are plotted. (C) Plotted signal-to-background ratios (S/Bs) were calculated from 100% and 0% values of the respective assays, representing top and bottom values of the concentration response curves. For the mini-G recruitment assay (mG), peak or plateau values of the response to 100 µM histamine (100%) and L-15 (0%) are displayed, whereas for the [ 35 S]GTPγS binding assay (GTPγS) responses to 1 mM histamine for H1,2R or to 10 µM histamine for H3,4R (100%) and H2O (0%) were taken. Presented data are the means ± SEM of at least five independent experiments (n ≥ 5), each performed in triplicate.

Kinetics and Dynamic Ranges of Mini G Protein Recruitment
The dynamic split-NanoLuc approach allows for monitoring the G protein response to a ligand in real-time, demonstrating the differences in kinetics for each receptor and mini-G protein combination upon histamine stimulation ( Figure 1B). The mGsq recruitment to the H1R is comparatively slow, leading to a plateau, whereas the luminescence signals of the mGs and mGsi recruitment to the H2R, H3R and H4R reach very sharp maxima and then flatten gradually. As the deletion of the membrane anchor and the Gβγ binding site were key modifications in the development of the utilized mini-G proteins, we assume that these observed kinetics will differ to the behavior of endogenous heterotrimeric G proteins. Further, other properties of the test system, such as the split-luciferase complementation reaction and the protein expression levels of the receptors and mini-G proteins, could influence the kinetics. Nevertheless, tracing the mini-G protein recruitment upon receptor activation in real-time could unveil differences in receptor regulation (e.g., receptor desensitization and internalization) [30,31] and may also serve as a useful tool to supplement studies of ligand binding kinetics, such as association and dissociation rate constants (kon/off) and residence time [32,33].
Using the mini-G sensors, the signal amplitudes of the assay were improved for all four receptor subtypes compared to the [ 35 S]GTPγS binding assay ( Figure 1C). For uniform comparison of the (A) Scheme of the mini-G protein recruitment assay. The split-NanoLuc technology was applied to the H 1-4 R (C-terminus) and the mini-G proteins (mG; N-terminus). Upon receptor activation, the mini-G protein is recruited to the GPCR and the split-NanoLuc fragments form a functional enzyme leading to the oxidation of the substrate and thus luminescence signals in an agonist concentration-dependent manner. (B) Representative luminescence traces of the mini-G protein recruitment of mGsq to H 1 R, mGs to H 2 R and mGsi to H 3 R and H 4 R. Baseline and inter-well corrected luminescence traces of histamine at various concentrations and the assay medium Leibovitz's L-15 (L-15) as negative control are plotted. (C) Plotted signal-to-background ratios (S/Bs) were calculated from 100% and 0% values of the respective assays, representing top and bottom values of the concentration response curves. For the mini-G recruitment assay (mG), peak or plateau values of the response to 100 µM histamine (100%) and L-15 (0%) are displayed, whereas for the [ 35 S]GTPγS binding assay (GTPγS) responses to 1 mM histamine for H 1,2 R or to 10 µM histamine for H 3,4 R (100%) and H 2 O (0%) were taken. Presented data are the means ± SEM of at least five independent experiments (n ≥ 5), each performed in triplicate.

Kinetics and Dynamic Ranges of Mini G Protein Recruitment
The dynamic split-NanoLuc approach allows for monitoring the G protein response to a ligand in real-time, demonstrating the differences in kinetics for each receptor and mini-G protein combination upon histamine stimulation ( Figure 1B). The mGsq recruitment to the H 1 R is comparatively slow, leading to a plateau, whereas the luminescence signals of the mGs and mGsi recruitment to the H 2 R, H 3 R and H 4 R reach very sharp maxima and then flatten gradually. As the deletion of the membrane anchor and the Gβγ binding site were key modifications in the development of the utilized mini-G proteins, we assume that these observed kinetics will differ to the behavior of endogenous heterotrimeric G proteins. Further, other properties of the test system, such as the split-luciferase complementation reaction and the protein expression levels of the receptors and mini-G proteins, could influence the kinetics. Nevertheless, tracing the mini-G protein recruitment upon receptor activation in real-time could unveil differences in receptor regulation (e.g., receptor desensitization and internalization) [30,31] and may also serve as a useful tool to supplement studies of ligand binding kinetics, such as association and dissociation rate constants (k on/off ) and residence time [32,33].
Using the mini-G sensors, the signal amplitudes of the assay were improved for all four receptor subtypes compared to the [ 35 S]GTPγS binding assay ( Figure 1C). For uniform comparison of the signal-to-background (S/B) ratios, we also implemented the [ 35 S]GTPγS binding assay for the H 1 R (Table S1, Supplementary Methods). Remarkably, in the case of the H 1 R, the S/B ratio was up to 29-fold higher in the mini-G protein recruitment assay than in the [ 35 S]GTPγS binding assay ( Figure 1C). Such favorable S/Bs are beneficial for the determination of agonist efficacies and will allow for a reduction of the agonist concentration when exploring antagonists. To evaluate the overall assay quality, we calculated the Z' factor, a dimensionless figure of statistical effect size. Classically, the Z' factor has been used in the validation process of HTS methods, as it numerically evaluates the dynamic range of an assay and its ability to identify biologically active molecules [34]. For all four receptor subtypes, we obtained a Z' factor that was between 0.5 and 1.0 (H 1 R: 0.79 ± 0.07, H 2 R: 0.85 ± 0.03, H 3 R: 0.80 ± 0.04, H 4 R: 0.68 ± 0.05; Supplementary Figure S3) indicating a sufficient separation of maximal effect and baseline values. Consequently, the presented mini-G protein recruitment assays can be classified as excellent screening methods [34].

Mini-G Protein Recruitment-Based Investigation of Histamine Receptor Ligands with Diverse Pharmacological Profiles
To demonstrate the applicability of these novel assays for future drug research, we tested a set of standard ligands (Supplementary Figure S4), which are described as (inverse) agonists or antagonists. We experienced a broad range of potencies and efficacies for ligands at all four receptor subtypes ( Figure 2) and the order of potencies of all studied agonists was in good agreement with literature data (Figure 2A, Tables 1-4). However, as general observation, agonists probed at the H 3 R and the H 4 R displayed lower potencies (up to one magnitude) than in published [ 35 S]GTPγS binding and steady-state GTPase activity assays (cf . Tables 3 and 4). Likewise, this phenomenon was observed for agonists studied in NanoBRET binding assays using intact cells expressing either the H 3 R or the H 4 R, as well as for agonists investigated with a H 3 R conformational sensor [35][36][37]. This finding was proposed as a consequence of an altered GPCR-G protein-guanine nucleotide composition, and therefore a more transient formation of the ternary complex compared to cell membrane preparations or cell homogenates [17]. By testing a large set of agonists, we validated the mini-G protein recruitment approach to report on a multifaceted spectrum of pharmacological actions. Efficacies ranged from weak partial agonism, discovered for histaprodifen at the H 1 R (E max = 33% ± 2.0) and UR-PI294 at the H 1 R (E max = 29% ± 1.4) and H 3 R (E max = 11% ± 1.1), to full agonism, demonstrated by e.g., N α -methylhistamine at the H 1 R (E max = 99% ± 2.0), dimaprit at the H 2 R (E max = 94% ± 2.6) and histamine (by definition: 100%) at all four receptor subtypes (Tables 1-4). Strikingly, the efficacies of UR-KUM530 at the H 1 R (E max = 112 ± 1.0) and N α -methylhistamine at the H 3 R (E max = 111 ± 1.6) were significantly higher (α < 0.05) as those of the endogenous ligand histamine (Tables 1 and 3), which is hypothesized as "superagonism" [38]. Similar results were previously observed for UR-KUM530 and were suggested to originate from a differing orientation in the binding pocket of the H 1 R compared to histamine [39,40]. In contrast, N α -methylhistamine has always been reported as a full agonist at the H 3 R [8].
Additionally, we extended the application of the mini-G sensor to the characterization of antagonists. The cells expressing the histamine receptors in combination with the respective mini-G proteins were pre-incubated with the antagonists and the response to the subsequently added agonist histamine was assessed. In this setting, standard antagonists exhibited expected pK b values at all receptor subtypes ( Figure 2B, Tables 1-4). Only in the cases of the tricyclic H 1 R antagonists maprotiline (pK b = 10.58 ± 0.11) and cyproheptadine (pK b = 10.19 ± 0.10), we determined up to two magnitudes higher pK b values than reported (Table 1). In the past, histamine receptors were reported to be constitutively active [41] in recombinant systems [42][43][44][45]. Investigation of the inverse agonistic potential of antagonists revealed that nearly all antagonists reduced the basal activity of the histamine receptors in the mini-G protein recruitment assay in a concentration-dependent manner ( Figure 2C, Tables 1-4). However, in our system the maximal inverse efficacies were small (H 1 R: −4%, H 2 R: −8%, H 3 R: −3%, H 4 R: −8% normalized to 100 µM histamine). Contrary to the literature, the constitutive activity of the Gα i -coupled receptors H 3 R and H 4 R was less pronounced [10,46]. However, as thioperamide demonstrated inverse agonism at the H 4 R, we confirmed JNJ7777120 and A943931 as neutral H 4 R antagonists.

Influence of Mini-G Protein Co-Expression on Potencies and Dynamic Ranges
Mini-G proteins functionally mimic active Gα subunits and thus a mutual cooperativity between mini-G protein and agonist binding to GPCRs has been proposed [23]. We probed histamine at the H 1-3 receptors co-expressed with increasing mini-G protein levels, but not at the H 4 R due to its weak transient expression. HEK293T cells were transiently transfected with constant receptor DNA amounts (1 µg) and increasing mini-G DNA amounts (0.125, 0.25, 0.5 and 1.0 µg) and were tested in the mini-G protein recruitment assay with histamine. In all three setups, the transfection of increasing mini-G gene doses were correlated with mini-G protein expression levels, which was demonstrated by a Western blot analysis (Supplementary Figure S5). In the mini-G protein recruitment assay, pEC 50 values of histamine were not significantly shifted (α = 0.05) by increasing mini-G expression levels at the three receptor subtypes (Figure 3A,B) in contrast to suggestions of Wan et al. (2018) [23]. However, the signal amplitudes were affected differently for the three receptor/mini-G pairs. In the case of the H 1 R, the signal span was not altered by different mGsq expression levels ( Figure 3A,C). On the contrary, the mGsi expression level determined by the highest gene dose of 1 µg significantly decreased the dynamic range at the H 3 R and, even more striking, all applied mGs gene doses in rising order led to significantly lowered dynamic ranges at the H 2 R (α = 0.05; Figure 3A,C). Similar to the collision coupling model of GPCR-G protein interaction [19,63], a possible explanation for the decreased signal amplitudes is that the basal activity of the histamine receptors increase due to higher mini-G expression levels and, thus, a more likely collision of constitutively active receptors and the respective mini-G protein. However, one has to be careful judging the extent of the signal span reduction observed for the H 1 R/mGsq, H 2 R/mGs and H 3 R/mGsi pairs, as e.g., same gene doses of mGsq led to considerably lower expression levels compared to mGs in the Western blot analysis (Supplementary Figure S5) [23].  Significance levels (C) were calculated using one-way ANOVA followed by Tukey's multiple comparison test calculated as **p < 0.01, ***p < 0.005, ****p < 0.0001.

Stabilization of the Active H2R Conformation by the Minimal Gαs Protein
As the signal amplitude at the H2R could be correlated to the mGs expression (Figure 3), we further explored binding properties of the endogenous agonist histamine and the antagonist famotidine by displacement of [ 3 H]UR-DE257 at HEK293T cells stably expressing the NlucN-mGs and H2R-NlucC fusion proteins ( Figure 4A, Supplementary Table S2). Whereas the radioligand displacement by famotidine followed a monophasic curve supporting a one-site binding model (pKi = 7.68 ± 0.01), notably a two-sites binding model was preferred for the agonist histamine (pKi,low = 3.87  50 values and (C) AUCs of histamine obtained in the mini-G recruitment assay using HEK293T cells transiently transfected with indicated DNA amounts (in µg) of the H 1-3 R-NlucC and NlucN-mGsq/mGs/mGsi constructs 72 h prior to the experiments. Presented data are from five independent experiments (n = 5), each performed in triplicate. Whiskers (B) represent 95% confidential intervals. Significance levels (C) were calculated using one-way ANOVA followed by Tukey's multiple comparison test calculated as ** p < 0.01, *** p < 0.005, **** p < 0.0001.

Stabilization of the Active H 2 R Conformation by the Minimal Gα s Protein
As the signal amplitude at the H 2 R could be correlated to the mGs expression (Figure 3), we further explored binding properties of the endogenous agonist histamine and the antagonist famotidine by displacement of [ 3 H]UR-DE257 at HEK293T cells stably expressing the NlucN-mGs and H 2 R-NlucC fusion proteins ( Figure 4A, Supplementary Table S2). Whereas the radioligand displacement by famotidine followed a monophasic curve supporting a one-site binding model (pK i = 7.68 ± 0.01), notably a two-sites binding model was preferred for the agonist histamine (pK i,low = 3.87 ± 0.13; pK i,high = 6.94 ± 0.14). Thus, we assumed there was a high affinity binding site at the H 2 R as previously described for the ternary H 2 R-G protein complex [53]. To correlate the observation to the amount of co-expressed mGs, we probed the binding of histamine at the H 2 R by transient transfections of increasing mGs gene doses (from 0 µg to 1 µg of mGs DNA) and a constant gene dose of H 2 R (1 µg; Figure 4B, Supplementary Table S2). The expression of the H 2 R alone (0 µg of mGs DNA) led to a rightward shifted, but monophasic concentration response curve of histamine. In contrast, by increasing mGs gene doses, we recorded an extended formation of the high affinity binding site ( Figure 4B, Supplementary Table S2). Therefore, we deduce that mGs stabilizes the active conformation of the H 2 R in a concentration-dependent manner. Although endogenously expressed G proteins are also intended to stabilize active receptor conformations, we did not detect a high affinity binding site using HEK293T cells that were transiently transfected with the H 2 R alone. On the one hand, this could be traced back to the lower native expression levels of G proteins compared to the overexpressed mGs. On the other hand, mGs constitutes the active GTPase domain of Gα s and therefore is immediately accessible for binding to the H 2 R in active state, whereas endogenous G proteins presumably exist in diverse conformations [53].

Discussion
Our study focused on the development of a novel live cell assay that reports on functional properties of histamine receptor ligands at an early stage of signal transduction. We achieved this by applying the split-NanoLuc to the four histamine receptors and minimal (chimeric) G proteins. We observed excellent signal amplitudes at all four receptor subtypes, which was of particular importance for the weakly expressed recombinant H4R. Moreover, we are the first to provide timeresolved courses of agonist-mediated functional responses using mini-G sensors with split-NanoLuc complementation for an entire receptor family, the subtypes of which couple to three different types of mini-G proteins (mGs, mGsi and mGsq). As the presented biosensor is becoming available for an increasing number of GPCRs [23][24][25][26][27], it will be appealing to extend the application in prospective studies to analyse time-resolved differences of distinct GPCRs coupling to the same minimal G protein. The methodology could also be used to investigate one GPCR coupling to different minimal G proteins (coupling specificity), and to supplement ligand binding studies with kinetic input, for example association and dissociation rate constants (kon/off) and residence time [32,33]. This might

Discussion
Our study focused on the development of a novel live cell assay that reports on functional properties of histamine receptor ligands at an early stage of signal transduction. We achieved this by applying the split-NanoLuc to the four histamine receptors and minimal (chimeric) G proteins. We observed excellent signal amplitudes at all four receptor subtypes, which was of particular importance for the weakly expressed recombinant H 4 R. Moreover, we are the first to provide time-resolved courses of agonist-mediated functional responses using mini-G sensors with split-NanoLuc complementation for an entire receptor family, the subtypes of which couple to three different types of mini-G proteins (mGs, mGsi and mGsq). As the presented biosensor is becoming available for an increasing number of GPCRs [23][24][25][26][27], it will be appealing to extend the application in prospective studies to analyse time-resolved differences of distinct GPCRs coupling to the same minimal G protein. The methodology could also be used to investigate one GPCR coupling to different minimal G proteins (coupling specificity), and to supplement ligand binding studies with kinetic input, for example association and dissociation rate constants (k on/off ) and residence time [32,33]. This might contribute to an even better pharmacological understanding of receptor regulation, as well as signal formation and transduction [64,65].
By investigating a large set of standard ligands, we demonstrated the usefulness of the mini-G sensor to reliably characterize agonists and antagonists. In our system, all four histamine receptor subtypes were constitutively active, although to a lesser extent than reported in other recombinant systems with the H 4 R [46,60]. The occurrence of such constitutively active receptors depends on the expression levels and the stoichiometry of the GPCRs and the G proteins according to the extended ternary complex (ETC) model of GPCR function [66]. Thus, the applied test system limits the detectability of the constitutive activity and the extent of the inverse efficacy of a ligand [67]. In future routine characterization of histamine receptor ligands, it would be convenient to introduce a reference ligand that produces inverse deflection of bioluminescence in the mini-G protein recruitment assay, such as diphenhydramine (H 1 R), famotidine (H 2 R) and thioperamide (H 3 R, H 4 R).
In the literature, two models of the GPCR-G protein interaction are discussed: a collision coupling, and a pre-coupled model [19,20,63]. In the case of the H 2 R, the lower the gene dose of the mGs, the higher the dynamic range. Further, we observed a high affinity binding site for the agonist histamine subject to the mGs expression level in radioligand competition binding experiments. Both results agreed with the collision coupling model of GPCR-G protein interaction, which supports an increased constitutive activity of GPCRs highly expressed in recombinant systems [19,63]. Therefore, it was not surprising that we did not detect such correlation for the H 1 R and H 3 R. Both receptors were expressed to a lesser extent compared to the H 2 R (Supplementary Table S1) and also the expression level of mGsq was considerably lower compared to mGs (Supplementary Figure S5).
Concisely, this study describes the establishment and usefulness of a mini-G sensor for prospective drug discovery at the histamine receptors. Due to the homogenous nature and the non-radioactive readout with an, per definition, excellent dynamic range (Z' factor), this assay will be automatable, and should be compatible with HTS.

Molecular Cloning
The human codon-optimized cDNA fragments encoding the mini-G proteins mGs, mGsi and mGsq (corresponding to mini-Gs393, mini-Gs/i43 and mini-Gs/q71 published by Nehmé [22], Supplementary Figure S1), were synthesized by Eurofins Genomics (Eurofins Genomics LLC, Ebersberg, Germany). Plasmids containing the split-NanoLuc fragments (NlucN, 159 amino acids; NlucC, 11 amino acids) were from Promega and cDNAs encoding the histamine receptors were purchased from the Missouri cDNA research center (Rolla, MO, USA). All cDNAs were amplified by PCR and subcloned into vector backbones by standard molecular cloning techniques. For this purpose, a set of pIRESpuro3 vectors was generated encoding the respective mini-G protein, which was N-terminally fused to the large split-luciferase fragment (NlucN) separated by a flexible glycine-serine-linker (encoding -GSSGGGGSGGGGSS-). The sequence encoding the H 1 R-NlucC described by Littmann et al. (2019) was subcloned into pcDNA3.1 using the restriction enzymes HindIII and SacII, and the receptor sequence was then replaced by either the H 2 R, H 3 R or H 4 R gene using HindIII and XbaI [71]. The optimal arrangement of a split-luciferase system to study the interaction of GPCRs and intracellular proteins of interest (GPCR-NlucC and NlucN-protein) was reported previously [23,71]. Plasmid DNA was quantified by UV-Vis absorbance using a NanoDrop spectrophotometer (ThermoFisher, Braunschweig, Germany). All sequences were verified by sequencing performed by Eurofins Genomics.

Cell Culture
HEK293T cells were a kind gift from Prof. Dr. Wulf Schneider (Institute for Medical Microbiology and Hygiene, Regensburg, Germany) and cultured in DMEM supplemented with 10% FBS at 37 • C in a water-saturated atmosphere containing 5% CO 2 . Cells were periodically inspected for mycoplasma contamination by means of the Venor GeM Mycoplasma Detection Kit (Minerva Biolabs, Berlin, Germany) and proven negative.

Generation of Stable Transfectants
In order to generate stable cell lines, wildtype HEK293T cells were stepwise transfected with a pIRESpuro3 vector encoding either the NlucN-mGs, -mGsi or -mGsq protein, and with the respective pcDNA3.1 plasmid encoding the histamine H 1-4 receptor-NlucC fusion protein according to the XtremeGene HP transfection protocol (Merck). The cells were then cultured in DMEM supplemented with 10% FBS, 1 µg/mL puromycin and 600 µg/mL G418 for sustained selection pressure.

Generation of Transient Transfectants
Adjusted to a cell density of 0.3 × 10 6 cells/mL, HEK293T cells were seeded into a 6-well cell culture plate (Sarstedt, Nürnbrecht, Germany) and allowed to attach overnight. The next day, the cells were transfected using linear polyethyleneimine (PEI, 1 mg/mL in PBS; 1:5 ratio (2 µg DNA: 10 µL PEI)) and incubated for another 48 h to allow for adequate protein expression. For mini-G protein recruitment assays and radioligand competition binding experiments, we applied a constant amount of 2 µg of total DNA per 6-well (total volume of 2 mL) comprising 1 µg of pcDNA3.1 H 1/2/3 R-NlucC and increasing amounts of the pIRESpuro3 NlucN-m/Gsq/mGs/mGsi DNA (0.125, 0.25, 0.5, or 1.0 µg). To ensure a uniform transfection efficiency, the empty pIRESpuro3 vector was co-transfected as mock DNA (0.875, 0.75, 0.5 µg or none). For Western blot analysis of the mini-G protein expression, the cells were transfected with a total amount of 2 µg DNA comprising 0.125, 0.25, 0.5 or 1.0 µg of the pIRESpuro3 NlucN-m/Gsq/mGs/mGsi and 1.875, 1.750, 1.5 and 1.0 µg, respectively, of the empty pIRESpuro3 vector as mock DNA.

Western Blot Analysis
Cells were lysed using a RIPA lysis buffer (50 mM Tris, 0.1% sodium dodecyl sulfate, 0.5% sodium deoxycholate, 1% Triton X-100, 150 mM NaCl) supplemented with SIGMAFAST protease inhibitor cocktail tablets according to the manufacturer's protocol (Sigma-Aldrich). Lysates (15 µg protein) and 10 µL of the Precision Plus Protein TM Dual Color Standard (Bio-Rad, Feldkirchen, Germany) were loaded to an 8-16% Novex Tris-glycine polyacrylamide gel (Thermo Scientific) and SDS-page was performed at 225 V for 1 h. Thereafter, the proteins were blotted on a nitrocellulose membrane (0.2 A, 1 h). By incubation with 5% skim milk powder in phosphate-buffered saline supplemented with 0.05% Tween 20 (PBS-T) for 1 h at RT, nonspecific binding sites of the membrane were blocked. After three washing steps with PBS-T, blots were incubated overnight at 4 • C with the primary antibodies α-Nluc (1:5000; in PBS-T; polyclonal, produced in rabbit, kindly provided by Promega) and α-vinculin (1:500; in PBS-T; monoclonal; MAB6896, produced in mouse, R&D Systems Inc., MN, USA). After additional three washing steps on the next day, the membranes were incubated with the HRP-conjugated secondary antibodies (raised against IgG, respectively) α-rabbit (1:10,000 in PBS-T; sc-2313, produced in donkey, Santa Cruz, TX, USA) and α-mouse (1:100,000 in PBS-T; A0168, produced in goat; Sigma-Aldrich) for 3 h at RT. The blots were washed three times with PBS-T and developed using the Clarity Western ECL substrate (Bio-Rad, Feldkirchen, Germany). Subsequently, the colorimetric and luminescent images of the stained blots were captured using a ChemiDoc MP imager (Bio-Rad).

Mini-G Protein Recruitment Assay
The day before the experiment, cells were detached by trypsinization (0.05% trypsin, 0.02% EDTA in PBS) and centrifuged (700 g, 5 min). Subsequently, the cells were resuspended in L-15 supplemented with 10 mM HEPES (Serva, Heidelberg, Germany) and 5% FBS. Thereafter, 100.000 cells per well were seeded onto a white flat-bottom 96-well microtiter plate (Cat. No. 781965, Brand GmbH + CoKG, Wertheim, Germany) and incubated at 37 • C in a water-saturated atmosphere without additional CO 2 overnight. Shortly before the experiment, the substrate furimazine was diluted in L-15 and 10 µL were added to the cells (final dilution 1:1000). Then, the plate was transferred to a pre-heated (37 • C) EnSpire plate reader (Perkin Elmer Inc., Rodgau, Germany). After recording the basal luminescence for 15 min, 10 µL of the agonist serial dilutions were added to the cells (final volume: 100 µL) and luminescence traces were recorded for 45 min (agonist mode). When investigating antagonists, the antagonist dilutions were added before the reference agonist histamine (EC 80 concentration; H 1 R: 10 µM, H 2-4 R: 1 µM) and the cells were incubated for 15 min (antagonist mode). Luminescence was captured with an integration time of 0.1 s per well. Data were analyzed using GraphPad Prism8 software (San Diego, CA, USA). The relative luminescence units (RLU) were corrected for (slight) inter-well variation caused by differences in cell density and substrate concentration, as well as for baseline drift, by dividing all data by the mean luminescence intensity of the respective L-15 control. AUCs of the luminescence traces for each concentration were calculated and normalized to the maximum response of 100 µM histamine (100% control) and L-15 (0% control). The logarithmic ligand concentrations were fitted against the normalized intensities with variable slope (log(c) vs. response-variable slope (four parameters)). The fit yielded pEC 50 and E max values in the case of agonists, and pIC 50 values in the case of antagonists, which were used to calculate pK b values according to the Cheng-Prusoff-equation [72]. In order to assess Z' factors, the baseline-corrected relative luminescence units (RLU) of 100 µM histamine and L-15 were inter-well corrected and AUCs were used for the calculation of means and standard deviations [34].
Significant differences in the efficacies obtained in the mini-G protein recruitment assay were assessed using a one-sample t-test (n = 5; α = 0.05). When investigating the influence of the mini-G protein expression level, significant differences between AUCs and pEC 50 values were calculated using one-way ANOVA followed by Tukey's multiple comparison test (n = 5, α = 0.05).

Radioligand Binding Experiments
Radioligand saturation binding experiments were performed using intact HEK293T cells co-expressing either NlucN-mGsq/H 1 R-NlucC, NlucN-mGs/H 2 R-NlucC, NlucN-mGsi/H 3 R-NlucC or NlucN-mGsi/H 4 R-NlucC. The following radioligands were used to verify the receptor expressions: For both, radioligand saturation and competition binding experiments, all (radio)ligand dilutions were prepared 10-fold concentrated in L-15 and 10 µL/well were transferred to a round bottom polypropylene 96-well microtiter plate (Greiner Bio-One, Frickenhausen, Germany). The cells were detached by trypsinization (0.05% trypsin + 0.02% EDTA), harvested by centrifugation (700 g, 5 min) and resuspended in L-15. The cells were adjust-ed to a density of 1.0 × 10 6 cells/mL and 80 µL of the cell suspension were added to each well (final assay volume of 100 µL). Then, the cells were incubated at room temperature under shaking for 60-120 min, and the cells were collected by filtration and washed with ice-cold PBS using a 96-well harvester (Brandel Inc., Unterföhring, Germany). The cell-associated radioactivity was measured by liquid scintillation counting, as previously described [73].
All data were analyzed using GraphPad Prism8 software. In the case of saturation binding experiments, all data were best fitted to a one-site saturation binding model (one site-total and nonspecific binding; one site-specific binding) yielding K d values. For competition binding experiments, data of the agonist histamine were best fitted to a two-sites competition binding model (two sites-fit logIC50) yielding pIC 50,high and pIC 50,low . Except, competition binding data of histamine using cells transiently transfected with the H 2 R alone and data of the antagonist famotidine obtained at cells stably co-expressing the H 2 R and mGs were fitted to the one-site three parameter logistic fit (one-site-fit logIC50) to determine pIC 50 values. Obtained pIC 50 values (pIC 50 , pIC 50,high , pIC 50,low ) were then used to calculate pK b values according to the Cheng-Prusoff-equation [72].