ThermoBRET: A Ligand‐Engagement Nanoscale Thermostability Assay Applied to GPCRs **

Measurements of membrane protein thermostability reflect ligand binding. Current thermostability assays often require protein purification or rely on pre‐existing radiolabelled or fluorescent ligands, limiting their application to established targets. Alternative methods, such as fluorescence‐detection size exclusion chromatography thermal shift, detect protein aggregation but are not amenable to high‐throughput screening. Here, we present a ThermoBRET method to quantify the relative thermostability of G protein coupled receptors (GPCRs), using cannabinoid receptors (CB1 and CB2) and the β2‐adrenoceptor (β2AR) as model systems. ThermoBRET reports receptor unfolding, does not need labelled ligands and can be used with non‐purified proteins. It uses Bioluminescence Resonance Energy Transfer (BRET) between Nanoluciferase (Nluc) and a thiol‐reactive fluorescent dye that binds cysteines exposed by unfolding. We demonstrate that the melting point (Tm) of Nluc‐fused GPCRs can be determined in non‐purified detergent solubilised membrane preparations or solubilised whole cells, revealing differences in thermostability for different solubilising conditions and in the presence of stabilising ligands. We extended the range of the assay by developing the thermostable tsNLuc by incorporating mutations from the fragments of split‐Nluc (Tm of 87 °C versus 59 °C). ThermoBRET allows the determination of GPCR thermostability, which is useful for protein purification optimisation and drug discovery screening.


Introduction
G protein coupled receptors (GPCRs) are a large family of membrane proteins that are important drug discovery targets. [1]tructural and biophysical studies of GPCRs have significant importance in modern drug discovery [2] but one major hurdle is their successful solubilisation from their native membrane environment and subsequent purification.Optimisation of receptor stability during this process is a key component to success. [3]Additionally, the ability of a bound ligand to stabilise the receptor structure is a property which can be exploited in screening efforts to find novel drug candidates. [4]xisting GPCR protein stability assays rely on the availability of a high-affinity radioligand to act as a tracer for receptor functionality. [5]In the absence of the radioactive tracer, temperature-induced aggregation-based techniques, such as technology developed by Heptares Therapeutics (now Sosei Heptares) [6] or say, which utilises a thiol-reactive fluorescent fluorochrome.This dye reacts with exposed cysteines, acting as a sensor of protein stability in the temperature-dependent unfolding process. [8]ther thiol-reactive dyes such as BODIPY-FL-Cystine (BLC) or 4-(aminosulfonyl)-7-fluoro-2,1,3-benzoxadiazole (ABD) are also available for stability measurements. [9]However, both these techniques currently require purified protein in microgram quantities which is a considerable drawback.
Low abundance of GPCRs even in over-expressing systems and their inherently low stability in detergents [10] calls for sensitive protein stability assays that can be used without protein purification or pre-existing tracer compounds.
Here, we present the ThermoBRET assay based on bioluminescence resonance energy transfer between the bright Nanoluciferase (Nluc), [11] acting as a donor of light and a thiol reactive Sulfo-Cyanine3 maleimide (SCM) dye, the acceptor, allowing us to quantify the relative thermostability of nonpurified GPCRs solubilised into detergent micelles.As a test case we focus on two GPCRs, the cannabinoid receptor 2 (CB 2 ) as a therapeutically promising [12] but unstable drug target [13] and the previously well characterised β 2 adrenergic (β 2 AR) receptor.This assay detects picomolar concentration, corresponding to nanogram amounts, of target protein.Due to the nature of the homogeneous assay format, negating the need to separate bound and unbound ligand, it can be used to detect binding of low-affinity stabilising ligands and fluorescent tracers.Since we employ NanoBRET detection, these assays are safer than radiometric alternatives and can be readily performed in 96-and 384well assay format.

ThermoBRET provides reliable measurements of GPCR stability
We fused a Nluc [11] to the receptor N-terminus, preceded by a cleaved signal peptide to ensure its' successful expression and plasma membrane trafficking (Supporting Information 1).Detergent solubilised receptor samples containing a thiol reactive Sulfo-Cyanine3 maleimide (SCM) acceptor are incubated at varying temperatures using a gradient forming PCR thermocycler.As the receptor unfolds on heating the SCM covalently binds to exposed cysteine residues (Figure 1).
We chose SCM because of its suitability as a BRET acceptor for Nluc, water solubility, and relatively low cost compared to other thiol-reactive fluorophores.In principle, any maleimide or other thiol-reactive conjugated fluorescent dye with overlapping donor-acceptor emission-absorption spectra can be used.The unfolded state of the receptor due to thermal denaturation is measured as NanoBRET between the Nluc tag and the SCM acceptor and is quantified as a ratio of the donor and acceptor light emissions, termed the NanoBRET ratio.The relative thermostability of a receptor in different solubilised non-purified membrane preparations can be easily determined, first by thermal denaturation across a temperature gradient on a thermocycler block, rapid cooling to 4 °C, and then following the addition of the Nluc substrate furimazine and measurement of the NanoBRET ratio in a 384-well luminescence plate reader at room temperature (Figure 1).The midpoint of the transition curve is found by fitting the data to a Boltzmann sigmoidal equation to obtain a T m .

Detergents affect stability of receptors
When solubilising GPCRs into detergent micelles, the choice of detergent and other additions is an important factor which influences thermostability of the GPCR.Two commonly used detergents are dodecyl-β-D-maltopyranoside (DDM) and laurylmaltose-neopentyl glycol (LMNG), with the latter often imparting greater stability. [14]A general observation in the field of GPCRs is that longer-chain detergents provide milder and more stabilising environment for GPCRs.Another detergent commonly added to solubilitisation buffers is 3-[(3-cholamidopropyl)dimethylammonio]-2-hydroxy-1-propanesulfonate (CHAPSO).CHAPSO has proven to be another useful detergent, its addition helping to further preserve the functional properties of certain GPCRs.Many GPCRs happen to bind cholesterol in the native environment of the cell.Addition of the cholesterol derivative cholesteryl hemisucciniate (CHS), which is much easier to handle due to its improved solubility in water.CHS has also been shown to impart greater thermostability to GPCRs through specific interactions with the protein as well as modulation of micelle architecture. [15]gure 1.Principle of ThermoBRET assay measured in 384-well plate format.Detergent solubilised non-purified membrane preparations expressing GPCRs fused at the N-terminus with Nluc (or thermostabilised tsNluc) are heated using a PCR thermocycler in the presence of sulfo-Cy3 maleimide (SCM).As the protein unfolds due to thermal denaturation, SCM reacts with newly exposed cysteine residues putting the sulfo-Cy3 acceptor fluorophore in proximity with the Nluc donor.At higher temperatures, protein aggregation leads to a decrease in the NanoBRET signal and these points are truncated before fitting to a Boltzmann sigmoidal equation to obtain a melting point (T m ).
We therefore solubilised Nluc-CB 2 in a selection of different solubilisation buffers that are routinely used in our lab.
When solubilised in DDM detergent, CB 2 had a T m of around 33 °C (Figure 2A) and was marginally more stable in LMNG (T m = 35 °C).Addition of CHAPSO and CHS in the detergent micelles provided the highest thermostability for CB 2 (T m = 43 °C in LMNG/ CHAPSO/CHS).13A] As maltoside detergents (DDM and DDM/CHS mixture) are by far the most popular choice for structural studies of membrane proteins, [16] and CB 2 was sufficiently stable in DDM/CHAPSO/CHS as reported previously, [13A] and produced sigmoidal temperatureresponse curve, we selected this condition for subsequent experiments.
Differences in the detergent stability of the adrenergic β 2receptor were also found (Supporting Information Figure S1).Hence, this assay can be readily used to screen for the best detergent solubilising conditions before attempting a large-scale purification.

Ligands stabilise CB 2
We also tested a selection of endogenous and synthetic cannabinoid receptor ligands for their ability to increase the thermostability of CB 2 (Figure 2B).The lipophilicity of its ligands has made the CB 2 receptor a particularly challenging target for ligand binding experiments due to their high non-specific binding.These ligands were all tested at a concentration of 20 μM, well above their dissociation constant (K D ) at room temperature, in order to ensure full occupancy of the solubilised receptors.Interestingly, the endogenous cannabinoid 2-arachidonoylglycerol (2AG) increased the T m of CB 2 by around 6 °C, whereas the other endogenous cannabinoid anandamide (AEA) only increased the T m by around 2 °C.The initial expectation would be that the degree of stabilisation at a constant ligand concentration reflects binding affinity.2-AG has been reported to bind CB 2 ca.2.5 times stronger compared to AEA. [17] Subsequently, a meta-analysis of reported affinity values suggested that both have ca.0.5-1 μM affinity for CB 2 , [18] with a large degree of variation of values reported by different groups.From the thermodynamic perspective, the degree of protein stabilisation by ligands depends on ligand affinity and concentration, the ability of the ligand to bind to the unfolded state of the protein and on the temperature dependence of the affinity that is determined by the entropy contribution to the binding [19] and may differ to different ligands.It needs to be noted that 2-AG and AEA structurally are very close so that last argument is unlikely to be important in this case.Both ligands are prone to degradation and have specific lipases that degrade them, present in cells and cell lysates.A more systematic study with ligands covering a larger range of affinities would really help to establish a link between ligand affinity and degree of stabilisation.
Other synthetic cannabinoid ligands HU308 and SR144528 also produced appreciable increases in thermostability, and the pattern of ligand stabilisation appeared different for the related CB 1 receptor (Supporting Information Figure S2), suggesting the observed stabilisation reflects ligand affinity.

tsNluc extends the range of the ThermoBRET assay
One problematic aspect of the ThermoBRET is the thermostability of the Nluc donor itself, which has been reported to unfold at around 55-60 °C. [11]This limits the thermal range for this assay and prevents accurate T m determination in conditions where the receptor itself is particularly thermostable, for example when CB 2 is bound to the high affinity non-selective cannabinoid agonist HU210 (Figure 2C).We therefore combined Nluc mutations which had been developed by Promega as part of their efforts to create a stable split-luciferase system [20] and found that these mutations improved thermostability of the full length luciferase by about 30 °C (Figure 2D).In line with previous reports [11] we found that purified Nluc had a T m of 59 °C, and that purified thermostable Nluc (tsNluc) had a T m of 87 °C (Figure 2D), making it preferable for thermostability measurements across a wide temperature range.Importantly, tsNluc contains no cysteine residues (Supporting Information 2) and thus is unaffected by maleimide/thiol chemistry.Further characterisation showed tsNluc to have a similar luminescence emission profile as Nluc with furimazine as a substrate, although with a lower luminescence output (Supporting Information Figure S3).Applying this novel tsNluc fusion improved the working temperature range of the Thermo-BRET assay and allowed successful T m determination for CB 2 in the presence of HU210 (Figure 2E).Strikingly, HU210 was able to stabilise CB 2 by around 12 °C, the highest level achieved of any of the CB 2 ligands tested.
To further assess the ability of this improved assay format to determine the stability of GPCRs in detergent we created a tsNluc-β 2 AR expression construct.Previous work employing the CPM assay indicates that this receptor can be stabilised by high affinity antagonists, [21] solubilised in DDM/CHS.Our initial experiments suggested that tsNluc-β 2 AR was also sufficiently stable in DDM only, a solubilisation condition we have used for this study.Due to the higher throughput achievable with our BRET-based system we were able to assess the receptor stabilising effects of both β-adrenergic agonists and antagonists (Figure 3 A and B).Both high affinity antagonist and agonist were able to stabilise the receptor to a degree which was dependent on the affinity of the ligands for the receptor (Figure 3C).In contrast, k off and k on were by themselves more poorly correlated with receptor stabilisation (Figure 3D and E).
The Tm values obtained by ThermoBRET were consistent with those reported previously [4A, 21] for ICI118551 (51 °C vs 51.6 °C), alprenolol (54 °C vs 53.6 °C) and carazolol (59 °C vs 65.6 °C).There were differences in construct design as this was a construct used for structure determination and it contained T4 lysozyme replacing the ICL3 of the receptor.There were differences in the assay format -they used purified protein in DDM/ CHS and continuous ramping of temperature. [8]Unfortunately, the concentration of the ligands used in previous experiments was not reported.Nevertheless, it supports the notion that ThermoBRET provides results that are consistent with the wellestablished CPM assay.

Lipid and total protein concentration in lipid-detergent micelles affects stability of the receptor
Having established a robust assay to measure receptor stability, we examined the more subtle effects of lipid-detergent ratio, as well as total protein concentration, on the stability of solubilised tsNluc-CB 2 .
Firstly, we examined if the stability of CB 2 is affected by the receptor concentration, while keeping the membrane fraction/ detergent ratio the same by supplementing the fraction of CB 2 membranes with "empty" non-transfected HEK293 membranes to the same total amount (Supporting Information Figure S4A).The reported value was the same, suggesting that CB 2 stability is not affected by its concentration, at least within the tested range.
Secondly, we examined if an increase in membrane/ detergent ratio could affect CB 2 stability, by supplementing a fixed amount of CB 2 membranes with an increasing amount of "empty" membranes (Supporting Information Figure S4B).Surprisingly, the receptor stability was decreased by 2-3 °C at higher membrane/detergent ratio, although it plateaued at concentrations above 20 ng/μL, as generally increasing concentration of lipids may lead to stabilisation of receptors. [23]A possible explanation for this is that by adding more "empty" membrane we also increase total protein concentration (HEK293 membranes contain a significant amount of different proteins).If some of the proteins have lower or comparable thermostability to CB 2 , this may accelerate aggregation process and reduce observed Tm, irrespective of protein identity.This observation suggests that for each new target this relationship needs to be explored, and concentration of membrane and detergent should be kept constant and an appropriate point on the plateau region should be chosen.

Stability of the receptor measured by ThermoBRET reflects a loss of ligand binding
Receptor denaturation is a complex process, progressing through a loss of tertiary structure through potential intermediates that may resemble the overall organisation of the correctly folded receptor (and protecting cysteines from modification) to complete loss of tertiary structure and aggregation.It is important to understand what process is sensed by the ThermoBRET assay.We hypothesised that ligand binding (as one of the functional properties of the receptor) should be sensitive to the smallest conformational changes.In this case unfolding of a target protein in response to an increase in temperature will result in a loss in specific binding.We measured binding of a fluorescent ligand to a detergent solubilized receptor using NanoBRET detection. [24]igure 4A shows the saturation binding curve for (S)-propranololgreen binding to the tsNluc-β 2 AR receptor solubilised in DDM.The K d of (S)-propranolol-green for the β 2 AR was determined to be 5.7 � 1.1 nM. Figure 4B shows the loss in (S)-propranolol-green specific binding signal as the protein unfolds in response to an  increase in temperature, following preincubation of a saturating concentration of fluorescent ligand (1 μM) with the receptor.
The observed Tm value reported by the loss of fluorescent ligand binding is the same as reported by ThermoBRET.These data strongly support the notion that ThermoBRET reports the loss of ligand binding function as the receptor unfolds.

Isothermal ThermoBRET
Finally, we assessed the dependency of agonist and antagonist ligand concentration on tsNluc-β 2 AR thermostability at a constant temperature of 35 °C, just above the T m of the apo form of the receptor.Receptor stability measurements expressed as a function of ligand concentration are shown in Figure 5A.
ThermoBRET IC 50 values were derived for this smaller test set and correlated with the thermal shift values obtained at fixed concentrations of individual agonists and antagonists.A linear relationship was observed between these two measures (Figure 5B).Finally, the ThermoBRET IC 50 values were correlated with radioligand binding affinity values obtained for the different ligands (Figure 5C).Again, an excellent correlation was observed between the two data sets apart from the very slowly dissociating antagonist cyanopindolol.There are two plausible explanations for this: either the very slow off-rate of this ligand means that for this ligand equilibrium is not achieved prior to the determination of ThermoBRET IC 50 values, samples being kept on ice prior to melting, or we are observing the phenomenon of ligand depletion due to nM concentrations of receptor present in the reaction mixture.

Assay sensitivity
Solubilised receptor concentration was estimated by extrapolation from a standard curve of the luminescence emission from purified tsNluc protein of known concentrations.ThermoBRET assay sensitivity was then determined by dilution of this solubilized β 2 AR receptor sample with receptor stabilisation assessed by monitoring the change in BRET ratio as a function of increasing temperature, see Supporting Information Figure S5.This confirms that ThermoBRET is a nanoscale system with sensitivity in line with that of a traditional ligand binding assay.
Additional parameter that could influence the assay window is the concentration of the cysteine-reactive dye.There should be sufficient, at least stoichiometric, amount or reactive dye to react to every available cysteine in the sample.On the other hand, too high concentration of the fluorophore in the sample may lead to increased background through bystander BRET as well as reduction in the signal through the inner filter effect.In further Radioligand binding data values were taken from. [22]xperiments we tested the effect of SCM dye concentration on the thermal unfolding of the β 2 AR (see Supporting Information Figure S6), confirming that dye concentrations in the region of 1-3 μM are sufficient to monitor the unfolding of GPCRs and without significant quenching of the luminescent signal.

Receptor solubilisation directly from cells
In addition, we assessed the stability of the β 2 AR solubilised directly from whole cells and the ability of orthosteric ligands to stabilise the receptor.Figure 6 shows the measured stability of the apo β 2 AR solubilised from whole cells in the absence and presence of a selection of antagonist and agonist ligands and the resulting correlation of T m values obtained with the same ligands from membrane solubilised receptors.The measurement of receptor thermostability directly from whole cells further reduces the number of steps required the assay (i.e. no need to prepare membranes) and allows target engagement to be assayed when receptors are still in a native membrane environment if it is added prior to solubilisation.

Discussion
Here, we aimed to establish a sensitive technique to measure receptor stability in crude non-purified detergent-solubilised membrane preparations, without any need for a unique tracer compound.ThermoBRET proved to be extremely sensitive and selective technique for reporting the stability of receptors directly solubilised from crude membrane preparations and even those directly solubilised from intact cells.It utilises universal cysteinereactive fluorescent dyes for readout and can be used to detect the binding of specific non-fluorescent ligands.
The processes of protein unfolding and protein aggregation are related, yet very separate phenomena.As mentioned in the introduction, a decrease in GFP fluorescence [7B, 7C] or luciferase signal [6] has been used as a measure of temperature-induced aggregation of GPCRs.Nluc has also been successfully used in similar applications which monitor protein aggregation of soluble proteins.These applications fuse either full length Nluc [25] or split Nluc [26] to the protein of interest and the decrease in luminescence activity is monitored as a measure of aggregation following thermal denaturation.In contrast, the ThermoBRET assay described here captures the initial conformational unfolding events which expose maleimide reactive cysteine residues in the protein of interest.In addition, ThermoBRET measurement is buffered from changes in the concentration of the luciferase fused target protein because a ratiometric method is used to calculate resonance energy transfer, contrasting with assays which measure luminescence intensity only.This means the measurement remains robust even when using low concentrations of the target protein where the magnitude of the measured signal is at the lower end of detection capabilities.We note that under the test conditions used, the luminescence activity of low pM purified Nluc and tsNluc was measurable in a 384-well plate (Supporting Information Figure S3A), making the assay extremely sensitive.This makes the assay particularly amenable to situations where there are limitations on the amounts of target protein available, for example, protein targets which are poorly expressed in vitro and/ or in vivo.This assay principle could even be applied in more physiologically relevant in vivo cellular models whereby (or the 11 amino acid HiBiT tag) is fused to an endogenously expressed protein via CRISPR-mediated insertion. [27]he novel thermostable tsNluc we describe has clear advantages compared to Nluc.Firstly, due to its improved thermostability it is less likely to unfold before the protein of interest and cause sample aggregation and other possible artefacts.The T m of tsNluc was measured at 87 °C which puts an upper temperature limit on the ThermoBRET method.However, such a high thermostability situation would be very unexpected for an integral membrane protein solubilised in detergents.Secondly, whilst previous reports [11] and our own data (Figure 2D) showed the T m of Nluc to be 59 °C, more recent use of Nluc to monitor protein aggregation showed clear luminescence activity after protein samples had been heated to temperatures > 60 °C before cooling. [25]We speculate that Nluc has propensity to spontaneously refold after thermal denaturation, and that the presence of detergents in the buffers of the latter report either aided Nluc refolding or delayed irreversible protein aggregation.In our ThermoBRET assays, the cysteine exposed upon Nluc unfolding could potentially react with the SCM and prevent its refolding, whereas tsNluc avoids these pitfalls.The cysteine-less sequence of tsNluc also allows easy in vitro chemical tagging of tsNluc-fusions.There is interest in producing conjugates of Nluc fused to other biomolecules, though usually this involves incorporation of sequence-specific ligation motifs onto Nluc followed by enzyme-mediated ligation to the molecule of interest. [28]By introducing a cysteine at the C-terminus of tsNluc, any molecule could be conjugated to tsNluc by thiol-reactive chemistry and mild reaction conditions, enhancing its potential for protein engineering and enabling a wide scope for future applications.
In comparison to our previous ThermoFRET application using a terbium cryptate labelled receptor as a FRET donor, [29] the ThermoBRET approach offers potential advantages.ThermoFRET requires cell surface labelling of the receptor-fused SNAP tag with the terbium cryptate donor molecule, adding to assay cost, but perhaps more importantly creating an extra labelling step that can be problematic if the tag is not readily exposed at the plasma membrane.In contrast, the use of a genetically encoded bioluminescent donor (ie.tsNluc or Nluc) omits this labelling step.This means that fused proteins which are poorly trafficked to the plasma membrane, are now amenable as they do not require labelling at the cell surface.Additionally, ThermoFRET requires more sophisticated detection by plate readers with time-resolved fluorescence detection capabilities, whereas BRET only requires a luminometer with filtered light detection, more readily available in many labs.A comparison of biophysical techniques used in drug screening cascades is shown in Figure 7 along with the relative protein requirements and assay throughput potential of each technique.
The ability of the ThermoBRET assay to quantify ligandinduced changes in the receptor T m makes it an ideal tool for studying ligand binding to GPCRs.4A] In principle, this assay can detect compounds which bind the target at any site, assuming this interaction influences the thermodynamic conformational landscape of the protein.It can be used to screen potential ligands for orphan GPCRs as it does not depend on the availability of tool compounds or known binders to develop a competition assay.Moreover, it could be used to detect the combined stabilisation of several ligands to discover positive and negative allosteric modulators of GPCRs.
Despite a number of advantages ThermoBRET offers, and the positive results obtained for β 2 AR, CB 2 and CB 1 receptors, this technique is not immune to the general limitations of thermal shift assays.The protein needs to be in a native state once solubilised, and some assay condition optimisation may be required for less stable receptors.Correspondingly, it may not be successful for every target receptor tried as there is a requirement that the receptor contains buried cysteines which become exposed upon thermal denaturation This limitation is also inherent for the CPM assay, and in such situations where no free thiol exists, then cysteines could be rationally introduced into the receptor sequence to generate a ThermoBRET signal.
Potential applications of ThermoBRET are not limited to GPCRs.Stability of any membrane protein that has a folded transmembrane domain, such as transporters and ion channels, could potentially be investigated.The only requirement is the presence of buried cysteines and an opportunity to engineer a tsNluc at the N or C-terminus, or alternatively a loop proximal to the luciferase donor.The principles of resonance energy transfer dictate that the acceptor dye covalently attached to an exposed buried cysteine should be within 1-10 nm depending on the donor and acceptor combination to achieve a useful BRET signal. [30]In principle, this method could be applied to any integral membrane protein provided that this distance requirement is met and that Nluc fusion does not disrupt the normal functioning of the protein.
It is also important to consider a possibility that fusing an additional stable domain such as tsNluc to the protein of interest may affect its stability.The potential mechanisms involved may include removal of existing interactions, if the N-terminal region needs to be shortened, or a chance introduction of new interactions that may stabilise the protein.It is important to remember, as it is with many other biophysical techniques, that the conditions employed in assay design are likely to deviate significantly from physiological conditions.As such our assays are unlikely to accurately reflect the stability of membrane proteins in the cell due to differences in environment (detergent micelle versus membrane), and therefore any interpretation of the results must take this into account.Nonetheless, ThermoBRET is excellent at measuring differences in target protein stability upon addition of interacting ligands and can be reliably used to draw conclusions about whenever a ligand binds the target, irrespective of the potential effects of tsNluc fusion.
To establish its real-life performance, larger compound libraries, which include potential PAINS (Pan-Assay INterference compoundS), will need to be screened to further understand any limitations of this technique.Overall, ThermoBRET is an excellent and highly sensitive tool for the optimisation of target protein solubilisation conditions and for biophysical screening of GPCR compound libraries to support structural biology, and to aid drug discovery efforts.

Drug compounds and reagents
Sulfo-Cy3 maleimide (SCM) (Lumiprobe GmbH, Germany) was obtained in powder form, dissolved in DMSO at a concentration of 10 mM and stored in the dark at À 20 °C.Furimazine, the substrate for Nluc, was obtained from the Nano-Glo Luciferase Assay System kit (Promega, UK) provided at a concentration of 5 mM.Cannabinoid ligands (anandamide [AEA], 2-arachydonylglycerol [2AG], SR144528, HU210, HU308, cannabinol) were obtained from Tocris Bioscience and dissolved in DMSO to a storage concentration of 10 mM, except AEA and 2AG which were dissolved in EtOH.Rimonabant was obtained from Roche Pharmaceuticals GmbH (Germany).(S)-propranolol-green (CA200693) was from CellAura, UK, and supplied by Hello Bio (Bristol, U.K).

Plasmid construction
For mammalian cell expression, receptor constructs were cloned into pcDNA4/TO using Gibson assembly. [31]All GPCR constructs contained an N-terminal signal peptide (which is cleaved by signal peptidases during protein maturation and trafficking) to improve expression, followed by a TwinStrep affinity tag, then Nluc (or tsNluc) followed by the receptor sequence.The TwinStrep affinity tag was not required for these studies but was present to facilitate receptor purification and antibody-based detection if required.The synthesized cDNA for tsNluc was obtained from GeneArt Gene synthesis (Invitrogen).For bacterial cell expression of Nluc and tsNluc, cDNA sequences were cloned into the pJ411 expression plasmid with an N-terminal 10X histidine affinity tag and TEV cleavage site encoded upstream of the protein of interest.Amino acid sequences of the constructs used are provided in Supporting Information 1.The correct sequence within the expression cassette of all plasmid constructs was verified by Sanger sequencing (Genewiz, UK).

Mammalian cell culture
The T-Rex TM -293 cell line (HEK293TR; ThermoFisher Scientific) was used to make stable expressing cell lines for receptors cloned into pcDNA4/TO.HEK293TR cells were cultured in growth medium (DMEM, 10 % FCS, 5 μg/mL blasticidin) in a 37 °C humidified incubator with 5 % CO 2 .Stable cell lines were generated by PEI transfection of pcDNA4/TO plasmids into HEK293TR cells.24-48 h after transfection, 20 μg/mL zeocin was incorporated into the growth medium until stable expressing, zeocin-resistant cell populations remained (2-4 weeks).To produce cells for membrane preparations, 1X T175 culture flask of confluent stable cells was treated with 1 μg/mL tetracycline for 48 h to induce receptor expression.Following this, cells were lifted by trituration and centrifuged at 500 g for 10 min.Cell pellets were then frozen at À 80 °C until membranes were prepared.

Membrane preparations
HEK293TR cell pellets were resuspended in 20 mL of ice-cold buffer (10 mM HEPES (4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid) pH 7.4, 10 mM EDTA (Ethylenediaminetetraacetic acid)) and homogenised using a Ultra Turrax (Ika Work GmbH, Germany).The homogenised cell suspension was then centri-fuged at 4 °C for 5 min at 500 g to remove whole cells and large debris, and the remaining supernatant was then centrifuged twice at 4 °C and 48,000 g for 30 min before the membrane pellet was resuspended in buffer (10 mM HEPES pH 7.4, 0.1 mM EDTA).Protein concentration of resuspended membranes was determined using Pierce BCA Protein assay kit (ThermoFisher Scientific) and was adjusted to 3-10 mg/mL before being aliquoted and stored at À 80 °C.

ThermoBRET experiments
The CORE buffer for thermostability experiments contained 20 mM HEPES pH 7.5, 150 mM NaCl, 10 % w/v glycerol, 0.5 % w/v BSA.Cell membranes were diluted in CORE buffer to approximately 0.1-0.5 mg/mL total protein and were then centrifuged at 16,000 g for 60 min at 4 °C to remove residual EDTA from the membrane preparation buffers.Membrane pellets were then resuspended in CORE buffer containing detergent, and samples were incubated at 4 °C with gentle shaking for 1 h to solubilise membranes.Detergent/CHS concentrations used were either 1 % DDM, 1 % DDM / 0.5 % CHAPSO / 0.3 % CHS, 0.5 % LMNG, or 0.5 % LMNG / 0.5 % CHAPSO / 0.3 % CHS.Samples were then centrifuged again at 16,000 g for 60 min at 4 °C to remove unsolubilised material, and the resulting supernatant containing detergent micelles was transferred to a fresh tube.These supernatants were then kept on for up to 48 h during testing.For thermostability testing, solubilised receptors were diluted 10fold in CORE buffer with the addition of 1 μM SCM and 20 μM of ligand (if used).This was incubated on ice for 15 min before being aliquoted across 96-well PCR plates and placed in the precooled (4 °C) PCRmax Alpha Cycler 2 Thermal Cycler (Cole-Palmer Ltd, St. Neots, UK).Samples were then incubated at different temperatures for 30 min via a temperature gradient across the plate.Following rapid cooling of the samples to 4 °C, samples were then transferred to white 384-well proxiplates (Perkin Elmer) containing furimazine at a final concentration of 10 μM, a commonly used concentration of the substrate.The Michaelis-Menten response of the NanoLuc has been reported [11] and also shown in the Supporting Information Figure S3.It is important to mention that the concentration of 10 μM is below the K m of furimazine for Nluc (ca 20 μM) and is not in sufficient excess to achieve pseudo zero order reaction kinetics.However, the efficiency of resonance energy transfer is a ratio between the emission intensities of the acceptor and of the donor, changes in brightness of the Nluc donor are cancelled out.The plate was then read using a PHERAstar FSX plate reader (BMG) at room temperature and the 450BP80/550LP filter module.Measurements were performed in singlet for each temperature point.Whole cell experiments (10 million cells/mL of 1 % DDM) were performed essentially as described above but receptor solubilisation was performed in the presence of protease inhibitor cocktail (cOmplete TM mini EDTA-free Protease Inhibitor cocktail (Roche)).

Nluc and tsNluc expression and purification
NiCo21(DE3) chemically competent E. coli were transformed with pJ411 bacterial expression plasmids and plated onto LB/agar plates containing 2 % w/v glucose and 50 μg/mL kanamycin.After incubation at 37 °C for 16-24 h, a single colony was picked to inoculate 20 mL of terrific broth containing 0.2 % w/v glucose and 50 μg/mL kanamycin.After 16-24 h in a shaking incubator set at 37 °C, 15 mL of overnight culture was added to 3 L of terrific broth containing 0.2 % w/v glucose and 50 μg/mL kanamycin, grown in a shaking incubator at 37 °C until OD 600 of 0.7-1, when 500 μM of isopropyl-β-D-thiogalactopyranoside (IPTG; VWR Chemicals) was added to induce protein expression.Cells were then grown overnight (16-20 h) at 25 °C in a shaking incubator before being harvested by centrifugation and frozen at À 80 °C.Cell pellets were then thawed on ice, and resuspended in 100 mL lysis buffer (100 mM Tris (tris(hydroxymethyl)aminomethane) pH 7.5, 300 mM NaCl, 0.25 mg/mL chicken lysozyme, 1 μg/mL bovine DNAse I, 4 mM MgCl 2 , and 3 cOmplete TM mini EDTA-free Protease Inhibitor cocktail tablets (Roche)).After 1 h on ice in lysis buffer, cells were then lysed further by French press.Cell lysates were then clarified by centrifugation at 25,000 rcf for 30 min and then by passing through a 0.45 μm syringe filter.The His-tagged proteins from the resulting lysate were then purified using a 5 mL HiTrap TALON Crude column on an ÄKTA start protein purification system (Cytiva Life Sciences) and eluted with 150 mM imidazole.Elution fractions were analysed by SDS-PAGE and fractions which contained no visible contaminants proteins were pooled together.Protein concentration was determined by A 280 measurement on a Denovix DS-11 FX series spectrophotometer assuming the calculated molar extinction coefficient (ɛ 280 ) of 26,930 mol À 1 cm À 1 for both proteins.

Luminescence activity thermostability experiments
Purified Nluc and tsNluc proteins were serially diluted from around 200 μM down to 100 pM in CORE buffer.Proteins were then aliquoted across 96-well PCR plates (100 μL per well) and placed in the pre-cooled (4 °C) PCRmax Alpha Cycler 2 Thermal Cycler (Cole-Palmer Ltd, St. Neots, UK).Samples were then incubated at different temperatures for 30 min via a temperature gradient across the plate.Following rapid cooling to 4 °C, 85 μL of samples were then transferred to white 96 well plates (Perkin Elmer) containing 15 μL of diluted furimazine to a final concentration of 10 μM.
After 30 seconds of gentle shaking, the luminescence intensity was measured in a PHERAstar FSX plate reader at room temperature.Measurements were performed in triplicate for each temperature point, and three independent experiments were performed.

Curve fitting and data analysis
All curve fitting and data manipulation were performed using GraphPad Prism 8.For ThermoBRET measurements, NanoBRET ratio was defined as the 550LP emission divided by the 450BP80 emission.In situations in which the NanoBRET ratio decreased at high temperatures (presumably due to protein aggregation and loss of signal), the data was manually truncated after the highest point.Data was then normalised to the upper (100 %) and lower (0 %) data points and fitted using a Boltzmann sigmoidal equation constrained to upper and lower values of 0 % and 100 %.For luminescence thermostability measurements, unfiltered luminescence was normalised to the top point of the dataset and fitted using a Boltzmann sigmoidal equation with no constraints.

Figure 2 .
Figure 2. ThermoBRET measurements in different detergent conditions and with stabilising ligands, demonstrating superior performance of tsNluc over Nluc for high thermostability situations.ThermoBRET thermostability curves and pooled T m measurements for (A) Nluc-CB 2 solubilised in the indicated detergent conditions.(B) in DDM/CHAPSO/CHS, in the presence (20 μM)/absence of ligands.(C) ThermoBRET curve for Nluc-CB 2 solubilised in DDM/CHAPSO/CHS, showing that the curve for the receptor bound to HU210 cannot be fitted as it is stable beyond the point of Nluc stability.(D) Luminescence thermostability curves of purified Nluc and tsNluc.(E) ThermoBRET using tsNluc-CB 2 in the presence/absence of HU210, showing a full fit for both curves.(A) and (B) show pooled normalised data showing mean � standard deviation for the number of experimental replicates evident in the far-right graph (n�2).(C) and (E) are raw fitted data from a single experiment performed 3 times.(D) is pooled normalised data from 3-independent experiments.

Figure 3 .
Figure 3. tsNluc β 2 AR thermoBRET measurements in 0.1 % DDM with stabilising ligands and high thermostability situation.ThermoBRET thermostability curves and pooled T m measurements for tsNluc-β 2 AR solubilised in DDM in the presence (20 μM)/absence of (A) agonist ligands and (B) antagonist ligands.The magnitude of shifts is shown on panels on the right, with ligands as indicated on the left panel.Correlations of ligand thermostability (ΔT m ) with (C) ligand affinity, (D) ligand k off and (E) ligand k on .(A) and (B) show pooled normalised data showing mean � standard deviation for the number of experimental replicates evident in the far-right graph (n�3), pooled normalised data from 3 or more independent experiments.Radioligand binding data values were taken from.[22]

Figure 5 .
Figure 5. tsNluc β 2 AR thermoBRET measurements in 0.1 % DDM with stabilising ligands expressed as a function of concentration.(A) ThermoBRET IC 50 curves obtained at a fixed temperature of 35 °C following a 30 min incubation with either agonists or antagonist ligands.Normalised data is plotted as the mean � standard deviation for 3 replicates.ThermoBRET IC 50 values were derived for this smaller test set and correlated with (B) the change in thermal shift obtained at fixed concentrations of individual agonist or antagonist and (C) radioligand binding derived ligand affinity values.All correlations are derived from mean thermostability measurements consisting of at least 3 replicates.Radioligand binding data values were taken from.[22]

Figure 6 .
Figure 6.tsNluc β 2 AR thermoBRET measurements in 0.1 % DDM following solubilisation from whole cells.ThermoBRET thermostability curves and pooled T m measurements for tsNluc-β 2 AR solubilised in DDM in the presence (20 μM)/absence of (A) agonist ligands and antagonist ligands.The magnitude of shifts shown on the panel on the right (B) shows the correlation between T m 's determined following solubilisation of the β 2 AR from HEK293TR membranes or whole cells.(A) and (B) show pooled normalised data showing mean � standard deviation for the number of experimental replicates evident in the far-right graph (n �3), pooled normalised data from 3 or more independent experiments.

Figure 7 .
Figure 7.A comparison of biophysical techniques used in drug screening cascades.Outlined are the protein requirements of each technique and their estimated daily throughput screening potential, along with the advantages of ThermoBRET and some potential uses.