Multiple GPCR Functional Assays Based on Resonance Energy Transfer Sensors

G protein-coupled receptors (GPCRs) represent one of the largest membrane protein families that participate in various physiological and pathological activities. Accumulating structural evidences have revealed how GPCR activation induces conformational changes to accommodate the downstream G protein or β-arrestin. Multiple GPCR functional assays have been developed based on Förster resonance energy transfer (FRET) and bioluminescence resonance energy transfer (BRET) sensors to monitor the conformational changes in GPCRs, GPCR/G proteins, or GPCR/β-arrestin, especially over the past two decades. Here, we will summarize how these sensors have been optimized to increase the sensitivity and compatibility for application in different GPCR classes using various labeling strategies, meanwhile provide multiple solutions in functional assays for high-throughput drug screening.


INTRODUCTION
G protein-coupled receptors (GPCRs) represent one of the largest membrane receptor superfamily, which is encoded by approximately 3% of human genes and over 800 members (Klabunde and Hessler, 2002;Fredriksson et al., 2003;Pin et al., 2019). They are widely expressed in all cells and organs from brain tissue to blood vessels, and are responsible for sensing a variety of external stimuli, ranging from light and temperature to neurotransmitters, peptides, and lipids (Lagerstrom and Schioth, 2008). GPCRs are involved in diverse physiological activities and play critical roles in pathogenesis, making them important drug targets (Sriram and Insel, 2018).
Members of the GPCR superfamily share a common seven-transmembrane (7TM) topology, and are classified into classes A, B, C, and F according to sequence similarity (Fredriksson et al., 2003;Fredriksson and Schioth, 2005). Generally, class A GPCRs possess a short extracellular domain (ECD), while class C GPCRs have a large ECD called Venus Flytrap domain (VFT). Ligand binding induces conformational changes from the extracellular ligand-binding site to the intracellular side of the receptor. The outward movement of the cytoplasmic end of transmembrane domain (TM) 6 in class A GPCRs opens up an intracellular cavity to accommodate the Gα subunit and activate G protein; in class B GPCRs, TM6 shows a disruption of the helical fold and the formation of a sharp kink to bind Gα subunits (Hilger et al., 2020). In contrast to Class A and B GPCRs, which are reported to function as monomers, class C GPCRs are reported as obligatory dimers . Ligand binding to class C GPCR leads to the closure of VFT, triggering the conformational change in the cysteine-rich domain or stalk domain, further rearranging the TMs from inactive interface to TM6/TM6 active interface, which is similar in class C GPCR homodimers such as metabotropic glutamate receptor type 2 (mGlu2) (Xue et al., 2015), mGlu5 (Koehl et al., 2019), and calcium sensing receptor (CaSR) , or heterodimer, like metabotropic γ-aminobutyric acid receptors (GABA B receptor) (Xue et al., 2019;Mao et al., 2020;Papasergi-Scott et al., 2020;Park et al., 2020;Shaye et al., 2020). Hence, monitoring the conformational changes of GPCRs provide a structural basis to evaluate GPCR activation.
The classical functional assays used to measure the activity of GPCRs are mainly based on downstream messengers, such as Ca 2+ release, 1, 4, 5-inositol phosphate (IP3)/IP1 accumulation, cyclic adenosine monophosphate (cAMP) production, or reporter gene expression (Thomsen et al., 2005). Most of them have been successfully developed into high-throughput screening (HTS) and robust assays, and widely applied in the pharmaceutical industry and academic research ( Figure 1A). Furthermore, with the discovery of G protein-independent β-arrestin signaling, functional assays have been developed by detecting β-arrestin recruitment to GPCRs or β-arrestin-induced GPCR internalization (Zhang and Xie, 2012; Figure 1B), while the first GPCR biased drug have been approved by FAD recently (Mullard, 2020). In addition to these classical assays, multiple functional assays based on resonance energy transfer (RET), which is a technology to detect the protein-protein interaction, have been developed in recent years for directly monitoring conformational changes in GPCRs, G proteins, and β-arrestins ( Figure 1C). Using these sensors, the GPCR signaling profiles and GPCR activation process have been investigated at multiple scales.
RET sensors have been well-reviewed to illustrate GPCR activation and signaling previously (Lohse et al., 2012;Kauk and Hoffmann, 2018;El et al., 2019;Haider et al., 2019;Quast and Margeat, 2019). In this review, we have summarized the FRET and BRET sensors, which contributed to G protein and βarrestin signaling, intra-GPCR rearrangement, and inter-GPCR movement investigations, especially in recent years. Meanwhile, we will focus on how these sensors are optimized to better investigate GPCR signaling and adapted to HTS in functional assays, as well as what new mechanism have been identified based on these sensors.

THE PRINCIPLES OF RET
RET is a photo physical process, in which the energy of a fluorescent donor is transferred to a suitable fluorescent energy acceptor (Förster, 1948; Figure 2). The efficiency of RET depends on three parameters: (1) the emission spectrum of the donor overlaps with the excitation spectrum of the acceptor; (2) the distance between the fluorophores is within 100 Å; (3) the relative orientation of their dipole moments toward each other (the parallel dipole orientation gets highest RET) (Stryer, 1978). According to the fluorescent labels, RET sensors can be normally classified into FRET and BRET. The excitation of FRET donor fluorophores needs an extra excitation laser, while BRET is based on the use of lightemitting enzymes-luciferase and other different variations. RET is a good approach to measure the GPCR activation and signaling in a living system (Cottet et al., 2012;Lohse et al., 2012;Kauk and Hoffmann, 2018).

GPCR/G PROTEIN SIGNALING FUNCTIONAL ASSAYS
Sensors for GPCR/G Protein Interaction G protein heterotrimer activity is initiated by exchange of GDP with GTP, when ligand binding to GPCR triggers the G protein coupling to the receptor (Neubig, 1994). Though some RETbased sensors have used to detect the dynamic interaction of G proteins to GPCRs upon stimulation (Table 1; Azpiazu and Gautam, 2004;Gales et al., 2005;Hein et al., 2005;Nobles et al., 2005;Philip et al., 2007;Audet et al., 2008), agonistinduced G protein recruitment represents high diversity and specificity among different G protein subtypes and GPCRs (Du et al., 2019) and dynamic GPCRs and G proteins interactions cannot always easily be detected. MiniG proteins are used to improve the stability of the GPCR-G protein complex (Nehme et al., 2017). They are modified by deleting membrane anchor domains and the Gβγ binding surface in wild-type Gα proteins, and mutated in many positions to increase GPCR/G protein complex stabilization (Nehme et al., 2017;Wan et al., 2018). MiniG BRET sensors used energy pair of Renilla luciferase (Rluc) and Venus fused in GPCR and miniG protein, respectively ( Figure 3A), which can effectively recognize different families of GPCRs, even class F GPCRs (Wright et al., 2019). MiniG BRET sensors can be used to investigate the dynamic interactions of GPCRs with almost all G protein families, including Gα i/o , Gαs, Gα 12/13 , and Gα q/11 (Wan et al., 2018). And then, the miniG q sensor is successfully applied in HTS compare of 5-HT 2A serotonin receptor hallucinogen agonists . Further optimization of miniG BRET sensors by replacing the BRET donor Rluc with NanoLuc (Nluc), which has stronger brightness and smaller size (Hall et al., 2012), or replacing the BRET donor and acceptor with NanoLuc Binary Technology (NanoBiT) pairs, which is based on the Nluc complementation system consisting of LgBiT (18 kDa) and SmBiT (1.3 kDa) proteins (Dixon et al., 2016), can further increase the sensitivity. The optimized miniG sensors can be used to detect GPCR FIGURE 1 | GPCR functional assays. (A) Classical GPCR functional assays based on the measurement of GTPγS, and Ca 2+ , IP1, IP3, cAMP, and reporter gene. (B) Biased GPCR functional assays. Assays were used to identify the biased signals between G proteins and β-arrestins, through detecting β-arrestin recruitment and GPCR internalization. (C) Multiple GPCR pharmacology profiling. GPCR sensors based on the conformational changes of different G protein subtypes, β-arrestins, and GPCRs. activation in intracellular compartments, such as the Golgi apparatus (Wan et al., 2018).

Sensors for G Protein Heterotrimer
As G protein can pre-associate to some GPCRs in the absence of ligand (Gales et al., 2005;Nobles et al., 2005;Philip et al., 2007;Audet et al., 2008), detecting G protein heterotrimer activity provides a more general way to evaluate GPCR activation. Measuring the proximity change between Gα and Gβγ subunits through BRET assay (Figure 3B), can reflect the G protein heterotrimer states and activation of GPCRs (Galés et al., 2006). Three flexible regions in Gα i around amino acid numbers 91, 122, and 60 were reported to be accessible for BRET sensors insertion, according to Gα subunit structure. Interestingly, the BRET ratio between Gα i1 -91Rluc or Gα i1 -122Rluc and green fluorescent protein (GFP) 10-Gγ decreased significantly, whereas Gα i1 -60Rluc and GFP10-Gγ exhibited an increase of BRET ratio during α 2A AR activation (Galés et al., 2006). This indicates that small rearrangements have occurred in the Gαβγ heterotrimer, which can be detected using BRET assay. Rluc can be replaced by Rluc8 and other enhanced luciferases (Rluc II) to increased brightness and quantum yield in BRET (Loening et al., 2006;Sauliere et al., 2012). Based on optimized BRET sensors, activation of different G protein subtypes can be measured, such as Gα i , Gα o , Gα s , Gα q , Gα 12 , and Gα 13 . GPCR ligand such as S II, which was previous known as β-arrestin-biased agonist of angiotensin II receptor type 1A (AT 1 R) using classical functional assays, is considered as partial agonist, as it can induce ∼20% BRET ratio change in G i/o , G q , and G 13 compared with the full agonist Ang II (Sauliere et al., 2012). It indicated the advantage and the necessity to combine BRET sensor in GPCR functional assays and proved the HTS application of BRET sensors in G protein signaling.
The G protein BRET sensors can be useful tools to evaluate the activity of individual G protein subtypes and many sensors have been reported as summarized in Table 1. However, not all the subtypes can be detected with good sensitivity. The new BRET Gαβγ biosensors TRUPATH have systematically optimized the insertion positions of the donor in Gα and the best combination of Gβ and Gγ subtypes. 14 optimized sensors have been developed, including the first Gα 15 and Gα Gustducin sensors . TRUPATH biosensors extremely increase the sensitivity of G protein BRET functional assays, contributing to the development of a powerful platform to investigate most G proteins activation in an array of GPCRs agonists, antagonists, inverse agonists and biased ligands. Meanwhile, NanoBiT system M2R/Gα o -Gβ. Azpiazu and Gautam, 2004 Rluc-GFP BRET sensor measuredthe direct and kinetic interaction of Gα s β 1 γ 2 proteins and β 2 AR in living cells, and supported the precoupling of Gα s β 1 γ 2 -β 2 AR by basal BRET signal.
Gβγ-masGRK3ct Masuho et al., 2015 Gprotein-specific unimolecules Nluc-YFP Membrane-anchored unimolecular BRET sensor specifically binds to GTP-Gαprotein and produce the BRET signal to indicate the activity of endogenous GPCRs/G proteins without any modifications.
Rluc8-GFP10 With high sensitivity, the BRET sensors re-defined the SII as the partial agonist of AT 1 R rather than β-arrestins biased agonist by detecting multiple G protein heterotrimersactivities.
AT 1 R/Gαβγ Sauliere et al., 2012 Rluc8-GFP The study generated through exhaustive protein engineering and empirical testing, building the TRUPATH suite of Gαβγ biosensors includes the first Gα 15 and Gα Gustducin probes. CFP-YFP Using FRET-based assay developed the direct sensors in mammalian to measure multiple G protein subtypes heterotrimer changes, it is indicated that Gα i and Gα z undergo rearrangement rather than dissociation, whereas Gα o dissociate or rearrange in distinct manner after α 2A AR activation.
YFP-mTurquoise The improved G q FRET sensor (with the best CFP variant) firstly allowed the detection of K on of G q and the FRET sensor indicated the dissociation of G q from G protein heterotrimer after stimulation of H1R.
TRHRs/β-arrestins Kroeger et al., 2001;Hanyaloglu et al., 2002 Rluc-YFP Using OTR/β-arrestin BRET sensor obtained the result that the BERT signal started at 10s and achieved the peak at 35s, indicated the delay and slow course of β-arrestin recruitment may be limited by receptors phosphorylation via GRK.
AT 1 R/β-arrestin2 Namkung et al., 2016 CFP-YFP The FRET sensor was used to detect the dynamic GPCR/β-arrestins interaction and provided the evidence oftime delay compared to the activation of PTHR.
PTHR/β-arrestin2 Vilardaga et al., 2002Vilardaga et al., , 2003 CFP-GFP/YFP The study used FRET sensors to classify the P2Y 2 receptor as a class A receptor when stimulated with ATP or as a class B receptor when stimulated with UTP, according to the interaction manners with β-arrestins.
P2Y 2 R/β-arrestins Hoffmann et al., 2008 FlAsH/ReAsH Double site-specific and orthogonal labeled FRET sensor was proposed in PTHR/β-arrestin2interaction investigation with minimal disturbance of their function proved the method may be widely applied.
β 2 AR/β-arrestin2 AT 1 R/β-arrestin2 PTH1R/β-arrestin2 Shukla et al., 2008 Nluc-CyOFP1 The optimized BRET sensorsdetected β-arrestin2 partial active state under AT 1 R stimulations with increased brightness and wider spectral separation, and the sensors can be applied with a wide panel of class A and B receptors even orphan receptors.

CFP-FlAsH
The FlAsH-FRET sensorshave advantages of the small size and robust fluorescence signal. The multiple sites inserted FlAsH-FRET sensors detected different conformational changes in β-arrestin2 under β 2 AR and M 2 AChR after agonist stimulation, and supportedreceptor-specific patterns of conformational changes in β-arrestin2.
Nuber et al., 2016 Rluc-FlAsH A series of intramolecular FlAsH-BRET sensors were designed todetect different conformational changes in β-arrestins under multiple GPCRs, such as PTH 1 R, β 2 AR and AT 1 R, supporting that specific ligand-receptors could invoke different conformational changes in β-arrestins and provide the insight into mechanism of GPCRs generating diverse functions.  Luciferase and YFP-tagged Gα-Gβγ constitute the G protein intermolecular BRET sensor for detecting the rearrangements of G protein heterotrimers (upper). G protein heterotrimer NanoBiT sensor based on Nluc complementation system was applied to monitor G protein heterotrimer rearrangements (lower). (C) Gβγ-GRK BRET sensor. The BRET signal was produced between the Gβγ and GRK after GPCR activation.
(D) Unimolecular BRET sensor, BERKY biosensor (BRET biosensor with ER/K linker and YFP). After G protein activation, the detector module YFP-KB-1753 binds to active G-protein Gα i -GTP on membranes, and the BRET signal produced between N-terminus Nluc and YFP.
has been used to modify the G protein sensors ( Figure 3B). The advantages of the smaller size and strong signal can avoid possible steric hindrance induced by large proteins such as Rluc, GFP or Venus, and enable the stable detection in hours (Dixon et al., 2016;Inoue et al., 2019).
To further compare the activation efficacy between different G protein subtypes, Gβγ-GRK NanoBRET sensor has been developed based on the Gβγ subunit and lipid-modified reporter peptide GRK3ct (masGRK3ct) (Hollins et al., 2009;Masuho et al., 2015; Figure 3C). The NanoBRET strategy was achieved by fusing Nluc to the GRK3ct and Venus to Gβγ (Masuho et al., 2015). It can determine both the kinetics and extent of G protein activation to independently analyze the catalytic activities of GPCRs and their signaling efficacy toward various targeted Gα protein subtypes (Masuho et al., 2015;Hauser et al., 2018).
Up to now, tools that are suitable for primary cells or native tissues of endogenous GPCRs remain limited. Very recently, Maziarz et al. (2020) developed a type of BRET biosensor with ER/K linker and YFP, called BERKY biosensor to capture the GTP form of the Gα protein ( Figure 3D). The membrane-anchoring sequence-fused Nluc was used as the BRET donor, and the YFPfused synthetic peptide KB-1753 served as the acceptor (Johnston et al., 2006). This unimolecular biosensor can specifically and sensitively bind to Gα i -GTP, and causes an increased BRET signal. It allows endogenous Gα-GTP and free Gβγ to be detected in primary living cells, and record the activation of G proteins in native, physiological environments. BERKY biosensors have been developed for endogenous Gα q -GTP, Gα 13 -GTP, free Gβγ, and Rho-GTP in cells via a similar strategy (Maziarz et al., 2020).

GPCR/β-ARRESTIN SIGNALING FUNCTIONAL ASSAYS
β-arrestins are considered to be prominent mediators of GPCR internalization, facilitating GPCR desensitization and the negative regulating G proteins (Lefkowitz, 1998). β-arrestins also act as key modulators of GPCRs to initiate G proteinindependent signaling pathways (Lefkowitz and Shenoy, 2005). Based on the interaction with β-arrestins, GPCRs can be classified into class A and B. Class A GPCRs interact with β-arrestins rapidly and transiently, whereas class B GPCRs stably associate with β-arrestins with a higher affinity (Oakley et al., 2000;Hasbi et al., 2004). β-arrestins can adopt different conformational changes while interacting with phosphorylated GPCRs (Shukla et al., 2008). The dynamic of β-arrestin conformational rearrangement can be longer than GPCR/β-arrestins interaction (Nuber et al., 2016). Accordingly, there are two main types of biosensors for studying the kinetics of GPCR/β-arrestin signal based on BRET and FRET: intermolecular sensors used for monitoring GPCR/β-arrestin dynamic interactions and intramolecular sensors used for measuring β-arrestin conformational rearrangement (Table 1; Bertrand et al., 2002;Charest and Bouvier, 2003;Krasel et al., 2005).

Sensors for GPCR/β-Arrestin Interaction
BRET sensors detecting recruitment of β-arrestins to active GPCRs in living cells was firstly reported by Angers et al. (2000). They used Rluc and GFP as BRET donor and acceptor, which were fused to the C terminus of β2 Adrenergic receptor (β 2 AR) and β-arrestin2, respectively ( Figure 4A). The results showed a large increase in BRET ratio following β 2 AR stimulation, and represented an agonist concentration-dependent manner. Subsequently, many similar studies investigated the recruitment of β-arrestins to GPCRs (Table 1). Furthermore, this kind of BRET sensors were successfully applied in HTS for GPCRs antagonists, which show compatibility and sensitivity as a functional assay (Hamdan et al., 2005).
While the conventional BRET sensors use the non-natural combination of donor and acceptor from different species such as Rluc from Renilla reniformis and GFP from Aequorea Victoria to limit non-specific signals from random interaction, enhanced bystander BRET (EbBRET) sensor is composed with BRET energy pairs both from Renilla reniformis. It possesses moderate affinity, optimally transferred energy, and improved signal, which shows the advantage in investigation of compartmental and ligand-biased β-arrestin trafficking in the plasma membrane and endosomal membrane in real time (Molinari et al., 2008;Namkung et al., 2016). EbBRET sensors can work in real time image recording up to 20 min, greatly facilitating the studies of β-arrestin and GPCR trafficking at high spatial and temporal resolutions (Lan et al., 2012;Namkung et al., 2016;Cao et al., 2019).
Compared to BRET sensors, FRET sensors are more suitable for visualizing the kinetics of GPCRs and β-arrestins with high temporal resolution in cells (Lohse et al., 2012; Figure 4B). Using the FRET sensor, which consists of the CFP-labeled parathyroid hormone 1 receptor (PTH1R) and YFP-labeled β-arrestin2, has shown a time delay of β-arrestin2 recruitment to PTHR after receptor activation (Vilardaga et al., 2002(Vilardaga et al., , 2003. Through comparing the recruiting rate of β-arrestin2 to β 2 AR, repeated short-term stimulation promoted β 2 ARs phosphorylation and showed very rapid increase of FRET with t 0.5 at 2.1 s, while the first stimulation caused fairly slow FRET change (t 0.5 of 19.6 s), which indicate the GPCR phosphorylation is the rate-limiting step for β-arrestin recruitment (Krasel et al., 2005). Furthermore, β-arrestin subtypes: β-arrestin1 and β-arrestin2 can exist different recruitment manner to the same GPCR when stimulated with different ligand, such as P2Y 2 R (a Purine receptor subtype) agonist UTP induced both β-arrestin1 and β-arrestin2 interacted with P2Y 2 R, whereas ATP caused stronger interaction of β-arrestin1 than β-arrestin2 with P2Y 2 R (Hoffmann et al., 2008).
To reduce the size of FRET acceptor, a smaller fluorescent probe fluorescein arsenical hairpin binder (FlAsH) was introduced as FlAsH-FRET sensors ( Figure 4B). A short peptide of 6 amino acids containing tetracysteine was inserted into the targeted protein, which can specifically bind to FlAsH dyes that produce FRET signal between proximate fluorophore pairs. The CFP/FlAsH FRET sensors showed almost five times the signal amplitude compared to the CFP/GFP FRET sensors . And then, the double site-specific and orthogonal labeling FRET sensor such as FlAsH/ReAsH (a red arsenical hairpin binder) can also be introduced to investigate GPCR/β-arrestin dynamic interaction (Zurn et al., 2010).

β-Arrestin Intramolecular Sensors
β-arrestin conformational change occurs following recruitment to the receptor (Charest et al., 2005;Nuber et al., 2016). βarrestin intramolecular BRET biosensor is based on the proximity change between the N-and C-terminus of β-arrestin ( Figure 4C). Intramolecular BRET sensors (Luc-β-arrestin-YFP) indicated that β-arrestin can adopt multiple active conformations with different ligands treatment (Shukla et al., 2008). It can be optimized by using Nluc and red-shifted fluorescent protein (CyOFP1) to increase brightness and wider spectral separation (Oishi et al., 2020). This sensor can monitor the early conformational changes of β-arrestin 2 in complex with GPCRs, with a wide panel of different class A and class B GPCRs upon agonist activation, and with orphan GPCRs known to spontaneously recruit β-arrestin2. After the R170E mutation was introduced, the sensor was able to detect the partial active state of β-arrestin2. It permits the monitoring of β-arrestin change in different stage during the GPCR activation (Oishi et al., 2020). Additionally, intramolecular FlAsH-BRET sensors using Rluc and FlAsH pair or intramolecular FlAsH-FRET sensors using CFP and FlAsH pair have also been developed ( Figure 4C). They confirmed distinct conformational changes in β-arrestins induced by different ligands and GPCRs (Lee et al., 2016;Nuber et al., 2016;Strungs et al., 2019).

INTRAMOLECULAR CONFORMATIONAL GPCR SENSORS
Though GPCR intramolecular RET sensors have already been developed in 1995 in purified GPCRs (Gether et al., 1995), Vilardaga et al. firstly reported a FRET sensor to detect GPCR conformational changes in living cells, which inserts CFP at the ICL3 and YFP at the C-terminus ( Figure 5A) in PTHR and α 2A AR (Vilardaga et al., 2003(Vilardaga et al., , 2005. Using this sensor, the authors presented different FRET signals induced by full agonists (a strong decrease) and partial agonists (a weak decrease) or inverse agonists (a significant increase) of α 2A AR, indicating the dynamic activation process and distinct receptor conformation rearrangements specific to different ligands. Then, many similar intramolecular conformational GPCR sensors have been developed, such as β 1 AR (Rochais et al., 2007), β 2 AR , A 2A -adenosine receptor (A 2 AR) , and B 2 -Bradykinin receptor (B 2 R) (Chachisvilis et al., 2006) as shown in Table 2. FlAsH labeling provided an alternative choice to replace CFP or YFP ( Figure 5A). The FlAsH labeling (∼1 kDa) can be induced into the ICL3 of GPCRs as the energy acceptor, while inserting CFP as donor at the C-terminus. This FlAsH-FRET sensor confirmed the similar fast kinetics of GPCR activation and also showed a fivefold improvement in signal-tonoise ratio .
Similar intramolecular BRET sensors of GPCRs were also used to monitor the conformational rearrangement promoted by diverse ligands (Sleno et al., 2016Bourque et al., 2017;Devost et al., 2017). Intramolecular FlAsH-BRET sensors are modified the C-tail with Rluc II, and introduced FlAsH labeling in the ICL3 (Figure 5B). In the case of the F prostanoid receptor, AT1R, and β 2 AR, although three FlAsH-BRET sensors were constructed in a similar way, the BRET signals exhibited receptorspecific behavior, indicating that different GPCRs have unique conformational profile . Furthermore, distinct patterns of conformational changes can be observed by biased ligand. Multiple positioned FlAsH-BRET sensors of AT 1 R showed rapid, sustained and robust BRET signals to allow the comparison of the full agonist and biased ligands stimulation . Moreover, intramolecular BRET sensors can be used to investigate the allosteric interactions between two receptors, such as the heterodimeric AT1R and prostaglandin F2α receptors .
Alternatively, BRET assay with Nluc and YFP or GFP10 can be used to detect the GPCR conformational rearrangements (Figure 5B), which can be induced by various compounds or mutation in the receptor as well as the impact of interacting proteins (Picard et al., 2018). Schihada et al. (2018) screened the efficacy of different fluorescent proteins or dyes as acceptors, combined with Nluc as a BRET-based α 2A AR biosensor, NanoBRET 618 (Nluc and Halo Tag dye pair) showed the highest amplitude upon agonist stimulation. It is a powerful approach to distinguish slight differences induced by partial agonist in the BRET signal instead of the full agonist effect identified by classical cAMP assay. These BRET sensors are also adaptable for micro-liter plate assays with HTS formats.
A novel assay for detecting conformational changes in GPCRs, based on nanobodies recognizing specific conformations, has been reported in recent years. Several conformation-specific nanobodies for GPCRs have been developed, including κ opioid receptor (KOR) (Che et al., 2018), µ opioid receptor (Huang et al., 2015), M2-muscarinic receptors (Kruse et al., 2013), β 2 AR (Rasmussen et al., 2011), AT 1 R (Szalai et al., 2012), and mGlu2 (Scholler et al., 2017b). For example, two conformation-sensitive nanobodies of KOR, Nb39 and Nb6, recognize the active and inactive states, respectively (Che et al., 2018). Combined with the BRET approach, it can be used to detect KOR activity. For the Nb6 sensor, a strong BRET ratio decreased upon KOR activation induced by agonist, in which Nb6 dissociated from activated receptors and recovered after the antagonist addition, while the Nb39 sensor had the opposite effect ( Figure 5C). Moreover, the conservative binding of Nb6 in the region provides a compatible tool for ligand-induced active conformational changes of other class A GPCRs, when replace their ICL3 by KOR ICL3 . Nanobody-based GPCR conformational sensors also have the advantage of investigating the transactivation induced by other receptors. For example, Nb80, an active β2AR sensitive nanobody, was used to analyze the effect of vascular endothelial growth factor receptor 2 on β 2 AR activation (Kilpatrick et al., 2019). Nanobody application reduced the modification in GPCRs. However, as there are only a few nanobodies available for GPCRs and the intellectual property protection, the application of nanobodies in GPCR functional assays remains limited.
Regarding to the application in native cells or animals in vivo model, genetically encoded sensors based on GPCR conformational changes have incorporated circularly permuted fluorescent protein (cpGFP) to optical visualize the neurotransmitter release in brain (Sun et al., 2018;Peng et al., 2020). The cpGFP is modified from original GFP, in which the amino and carboxyl portions have been interchanged and reconnected by a short spacer between the original terminus. It is more flexible and accessible than original protein, and the fluorescence intensity of cpGFP is related to its conformation (Baird et al., 1999). Thus, cpGFP offered a suitable strategy for

RET-sensors applied in purified protein
Polarity sensitive fluorescent dyes IANBD β 2 AR The first direct evidence for ligand specific conformational changes occurring in a GPCR. Gether et al., 1995 Fluorescence quenching Fluorescein maleimide (FM) and oxyl-N-hydroxysuccinimide as quencher β 2 AR Showed the relative movements of the two labeled amino acid and 20 s activation kinetics through the fluorescent change after ligand addition. Ghanouni et al., 2001a,b Intramolecular RET-sensors in cells FRET CFP/YFP PTHR; α 2A AR; β1AR; β 2 AR; A 2 AR; B 2 R The first generation of FRET sensor to detect GPCR conformational changes in living cells, and showed a fast activation kinetic induced by various ligands in single cell assay using microscopy. Vilardaga et al., 2003Vilardaga et al., , 2005Hoffmann et al., 2005;Chachisvilis et al., 2006;Rochais et al., 2007;Reiner et al., 2010 FRET CFP/FlAsH A 2 AR; α 2A AR The smaller FlAsH tag avoided to disturb G protein signaling and provided a better signal-to-noise ratio compared with CFP/YFP FRET sensor. nanoBRET Nluc/NB80-GFP β 2 AR NanoBRET sensor based on NB80-GFP confirmed the activation of β 2 AR in VEGFR2-β 2 AR complexes was not influenced by VEGFR. conformation-sensitive sensors. As similar movements between TM5 and TM6 occurred during class A GPCRs activation (Weis and Kobilka, 2018;Hilger et al., 2020), fusing cpGFP at the ICL3 would allow significant fluorescence signal changes of cpGFP following GPCR activation upon ligand binding (Doi and Yanagawa, 1999;Sun et al., 2018). Mutation to abolish the G protein coupling in these sensors is required to not change the physiological GPCR function when expressed in the animals. The advantage of genetically encoded sensors is the rapid and high resolution in two-photon imaging systems for spatial neurotransmitters detecting in living animal (Sun et al., 2018;Peng et al., 2020). However, to obtain sensors with high sensitivity, large number of screening has been done for the cpEGFP insertion and linker residues (Sun et al., 2018). The experience in developing the sensor of one neurotransmitter is not always well adapted to another. Neuromodulator sensors are available for dopamine, serotonin, norepinephrine, acetylcholine, endocannabinoid, adenosine and gastrin-releasing peptide, but for glutamate and γ-aminobutyric acid remains difficult (Labouesse and Patriarchi, 2021).

INTERMOLECULAR CONFORMATIONAL GPCR DIMERIZATION SENSORS
Intermolecular FRET sensors are considered as good approaches for investigating the dimerization/oligomerization of GPCRs, especially in class C GPCRs (Milligan and Bouvier, 2005;Kniazeff et al., 2011). The classical CFP/YFP FRET sensor can detect the inter-subunit conformational change. When inserting the fluorescent protein in ICL2 of mGlu1 receptor, it showed an increased FRET signal indicating the relative movement of two mGlu1 subunits (Tateyama and Kubo, 2006). Subsequently, through measuring the FRET between two mGlu1 subunits in real-time; a fast increased intersubunit FRET signal between protomers was detected within 10 ms after glutamate application (Marcaggi et al., 2009). However, difficulties remain in these classical CFP/YFP FRET sensor applications due to low sensitivity, photobleaching and limitations of inserting position. Time-resolved FRET (TR-FRET) use long-fluorescence lifetime fluorophore, such as lanthanide cryptate instead of fluorescent protein. The fluorescence lifetimes of these molecules are very long, ranging from 100 to 1,000 µs, which leads the efficiency of FRET is not affected by the orientation of dipole moments between donor and acceptor, and becomes truly dependent on their distance (Mathis, 1995;Selvin, 2002). Taking a fixed delay time before acquiring the signal allows the removal of most of the fluorescent background provided by biological media and instrument, which largely improves the signal-tonoise ratio, compared with classical FRET sensors (Maurel et al., 2008;Scholler et al., 2017a). TR-FRET sensors can be adapted in multi-well plates format from 96 well to 384 well for drug HTS (Scholler et al., 2017a;Liu et al., 2020).
Antibodies labeled with long-lifetime lanthanide-based cryptate fluorophores were used in the first-generation TR-FRET sensors, which target small tags fused in GPCRs, such as HA, Flag, or c-Myc ( Figure 6A). It was used to prove the protein interactions and indicate the interface (Kniazeff et al., 2004;Liu et al., 2004). However, it failed to monitor the dynamic changes between the GPCR subunits, might because of the large size of the antibodies.
The SNAP tag is two-thirds the size of GFP, derived from the O 6 -guanine nucleotide alkyltransferase that covalently reacts with benzyl-guanines (BG). The SNAP tag can be specifically and covalently labeled with any fluorophore carried by the BG benzyl group. By using non-permeant BG derivatives compatible with TR-FRET measurements, it allows to identify the dimer of GPCRs ( Figure 6B). Using SNAP and CLIP labeling TR-FRET sensors, the mGluRs, which are considered strict homodimers, have been found to be heterodimers between different sub-groups (Doumazane et al., 2011). Meanwhile, SNAP labeling sensors can be used to identify oligomers, such as the GABA B receptor, which was previously considered to be heterodimers (Maurel et al., 2008;Comps-Agrar et al., 2011).
Furthermore, a N-terminal SNAP tag strategy combined with TR-FRET technology was developed to monitor the dynamic changes between two subunits in class C GPCR dimer. The mGlu receptors are dimeric entities with a large ECD, and during activation, a large conformational change results from the rotation of almost 70 • of one ECD relative to the other (Huang et al., 2011). According to the available structures of mGlu receptor ECDs (Kunishima et al., 2000;Tsuchiya et al., 2002;Muto et al., 2007;Koehl et al., 2019), the distance between the N terminus varies from 2.8 nm in the inactive conformation to 3.3 nm in the active conformation. Consistently, in the TR-FRET measurement, inactive mGlu2 receptors were found to have a high FRET signal, while a low FRET signal was obtained in the active state ( Figure 6B). The agonist-induced decrease in intersubunit FRET efficiency was caused by a large change in the distance between the separated fluorophores (Doumazane et al., 2013). Such a movement is closely related to receptor activation, as agonist potencies presented by TR-FRET are perfectly correlated with those determined in cell-based functional assays (including IP1, cAMP, and Ca 2+ release), indicating the efficiency and accuracy of the conformational change-sensitive TR-FRET sensor. The N-terminal SNAP strategy is feasible for most class C GPCRs, including all mGlu family members (mGlu1-8) and CaSR (Scholler et al., 2017a;Liu et al., 2020; Figure 6B). SNAPtagged TR-FRET sensor of the mGlu5 receptor based on ECD conformation showed enhanced basal activation in the presence of D1R (Sebastianutto et al., 2020).
To increase the compatibility of SNAP-tagged TR-FRET sensor to different GPCRs, several modifications are required to obtain a large signal-to-noise ratio, including the insertion position, labeling strategies, and methods to quantify signals. For example, the N-terminal SNAP strategy does not work for the GABA B receptor because GABA B receptors do not undergo a strong conformational change similar to that of the mGlu receptors during activation (Geng et al., 2012;Lecat-Guillet et al., 2017). To detect the conformational change of this receptor, researchers kept the SNAP tag on the N-terminal of the GABA B1 subunit and introduced a short version of the acyl carrier protein (ACP)-tag (ACP 17 ) within extracellular lobe 2 of the same subunit based on the knowledge of the activation of the GABA B receptor (Figure 6B). A high TR-FRET signal was largely decreased upon GABA B receptor activation and was suppressed by the competitive antagonist CGP54626 (Lecat-Guillet et al., 2017;Scholler et al., 2017a). Different sizes of tags, such as ACP (8 kDa) and ACP 17 (2 kDa) (George et al., 2004;Yin et al., 2005) or the O 6 -alkylguanine-DNA alkyl transferase derivatives, SNAP (Keppler et al., 2003), CLIP (23 kDa) (Gautier et al., 2008), and Halo Tag (33 kDa) (Encell et al., 2012), combined with compatible fluorophore pairs build a toolbox for TR-FRET sensors optimization (Figures 6D,E). Another representative application of intra-subunit sensor is the luteinizing hormone (LH) receptor from the class A family, which has a large extracellular domain. After the addition of agonist, TR-FRET signals increased while the ACP 17 and SNAP tags were constructed at the loop and N terminus, respectively, revealing an important conformational change within the extracellular domain of the LH receptor. The strategy of the extracellular intrasubunit sensor is feasible for these kinds of monomeric GPCRs, in which the extracellular domain undergoes a large conformational change during activation (Scholler et al., 2017a).
In addition to class C GPCR, some classes A and B GPCRs were found to form dimer or oligomer (Carrillo et al., 2003;Berthouze et al., 2005;Harding et al., 2009;Kasai et al., 2018). These heterodimers showed distinct functions and related to pathogenesis. For example, AT 1 R and B 2 R heteromerization was found to occur in human placental biopsies from pregnancies complicated by preeclampsia, and the aberrant heteromerization of AT 1 R-B 2 R was found to result in exaggerated calcium signaling and high vascular smooth muscle mechanosensitivity (Quitterer et al., 2019). TR-FRET sensors are also adaptable to other GPCR dimers, such as class A relaxin family peptide receptor 3 (RXFP3) and the LH receptor, class B PTHR, corticotropin-releasing factor receptor 1 (CRF1R), and pituitary-activating cAMP polypeptide (PACAP) receptor 1 (PAC1). The potency of a series of agonists obtained from measurements of the TR-FRET assay correlated with those obtained in functional assays (Scholler et al., 2017a). On the other side, BRET approaches using Rluc and YFP pair can be used to identify the formations of GPCR dimers (Ayoub and Pfleger, 2010;Johnstone and Pfleger, 2012;El et al., 2019). But few intermolecular BRET conformational change sensors have been reported. Though ligand-induced BRET changes have been presented in MT1/MT2 melatonin receptor heterodimers (Ayoub et al., 2002(Ayoub et al., , 2004, conformational changes in other dimers such as F prostanoid receptor, were not robust enough for investigation (Sleno et al., 2016).
The unnatural amino acid (UAA) site-directed modification strategy is considered to be a potential way to build flexible RET sensors that minimize the labeling tag size into one residue. The UAA technology can be used to investigate the interaction sites between two proteins, such as β-arrestin binding to AT 1 R (Gagnon et al., 2019). Then, with an optimization in UAA incorporation strategy, it can measure FRET signal between two specific labeling sites of GPCR at the living cell level and the single-molecule level (Liauw et al., 2021). This UAA incorporation strategy in living cells provide a good protocol to apply UAA instead of other tags in GPCR sensors, which may have higher sensitivity to monitor more differential conformational change.

CONCLUSION AND PERSPECTIVES
In this review, we summarized four types of conformational sensors for GPCR signaling and activation based on FRET and BRET. These sensors have identified new mechanisms of GPCRs activation process and also lead to significant breakthroughs in high-throughput drug screening toolboxes. Generally, most FRET sensors show strong intensity and microscopy compatibility, which possess better spatial and temporal resolution for imaging purposes. The TR-FRET sensor can also be applied in HTS. BRET sensors have more sustainable signals, higher signal-to-noise ratios and HTS applications. These sensors have been optimized using various labeling strategies to increase the sensitivity and compatibility, from heterogeneous systems to endogenous conditions. Using these assays, GPCR signaling and activation have been investigated on a large scale and at multiple levels. However, the introduction of BRET and FRET sensors without breaking normal expression and function remains challenging. The UAA site-directed modification strategy for FRET sensors may be a promising approach (Liauw et al., 2021). BERKY biosensors for endogenous G proteins will be a good choice to investigate endogenous GPCR activation (Maziarz et al., 2020).
Meanwhile, although smaller and smaller tags can be preferred to minimize the extra influence, a few FRET sensors based on traditional antibodies (Liu et al., 2004;Comps-Agrar et al., 2011) or labeled small molecule ligands (Albizu et al., 2010) have shown advantages in the detection of GPCR in native samples. However, due to the excessive molecular weight of antibodies or insufficient specificity of some antibodies and small molecule ligands, it is difficult to detect GPCR complexes in native tissue. Nanobodies, which have a smaller size, higher affinity, and conformation specificity, may provide breakthroughs in native GPCR functional assays and signaling (Scholler et al., 2017b; Figure 6C). As the hetero-complexes of GPCRs have received increasing attention for their connection with diseases (Prinster et al., 2005;Gomes et al., 2016), nanobody-based assays will provide useful tools for investigating roles of GPCR heteromers in physiological and pathological processes in the future.

AUTHOR CONTRIBUTIONS
All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.