WO2022258437A1 - Quantification of protein-protein interaction of membrane proteins using high-mass mass spectrometry - Google Patents

Abstract

G-protein-coupled receptors (GPCRs) are important pharmaceutical targets for the treatment of a broad spectrum of diseases. Although there are structures of GPCRs in their active conformation with bound ligands and G-proteins, the detailed molecular interplay between the receptors and their signaling partners remains challenging to decipher. To address this, a high-throughput quantification method to quantify protein-protein interaction of membrane proteins is disclosed, comprising the steps of: a) providing a solution comprising a membrane protein, such as a GPCR, in the absence or presence of a chemical compound / compounds or ligands that can bind to or modulate the membrane protein;; b) adding a partner protein to the solution and after a predetermined time interval adding additionally a crosslinker that reacts with proteins' surface amino acid, such as lysines or other amino acids depending on the specific crosslinker that is used, in order to form chemical linkages that stabilize compound complexes of biomolecules; c) detecting and quantifying stabilized native and transient complexes of the biomolecules and the non-interacting biomolecule counterparts by mass spectrometry using a reference peak, i.e. in terms of a normalization strategy, to investigate the binding ability of the partner protein to the membrane protein. Thus, this high-sensitivity, high-throughput mass spectrometry method interrogates the first stage of signal transduction. The membrane protein and partner protein complex formation is detected as a proxy for the effect of ligands on membrane protein conformation and on coupling selectivity. The method requires only very little probe amounts,

Description

Quantification of protein-protein interaction of membrane proteins using high-mass mass spectrometry

The present invention relates to a high-throughput mass spectro etric method for the detection and quantification of protein-protein interaction of membrane proteins and partner proteins.

G-protein-coupled receptors (GPCRs) are the largest family of membrane receptors in humans and play essential roles in physiology and disease. Their physiological and cellular signaling effects, modulated by chemically diverse ligands, are exerted through coupling to and activating heterotrimeric G- protein complexes

In humans, there are 16 Ga subunits that are classified into four families and

Goii2_/i3)· Each Ga subunit is involved in a specific signal transduction pathway. Although the understanding of GPCR signaling has been greatly enhanced by the remarkable progress in GPCR structural biology, much remains to be discovered to fully understand the molecular mechanisms of allostery and ligand-induced coupling selectivity (or functional selectivity) between GPCRs and their cytoplasmic transducers (G-proteins, but also kinases and arrestins) that lead to precise signal transduction cascades and biased signaling.

Investigation of the interplay between GPCRs, ligands, and intracellular binding partners is challenging due to the complexity of their interactions. The functional outcome of GPCR activity depends on a still poorly understood network of protein interactions. To date, there are no high-throughput methods to study every G-protein and its ability to couple to a given receptor under a standard set of conditions. Many GPCR assays use radio-/fluorescent-labelled ligand binding or measurement of second messenger molecules. More recent methods involve cell- based biosensors, including dynamic mass redistribution (DMR) and cellular dielectric spectroscopy (CDS), that display an overall cellular response and translate GPCR signaling into distinct optical or impedance readouts respectively. However, these assays do not provide a direct readout of G-protein coupling to GPCRs. Current biophysical methods that measure such protein interactions directly to provide information on selectivity and affinity - such as surface plasmon resonance (SPR), fluorescence resonance energy transfer (FRET), isothermal titration calorimetry (ITC) and analytical ultracentrifugation (AUC) - only provide limited information on dynamic protein interactions and either are not suited for high-throughput screening or lack information on all interacting components. Bioluminescence resonance energy transfer (BRET) has been extensively used over the last two decades to study GPCR-protein interactions; however, BRET requires labeling of the proteins and, because their level of expression can vary considerably, quantification can be difficult. Native electrospray ionization mass spectrometry (nESI-MS) has been successfully applied to study G-protein complexes and membrane proteins. However, it is difficult to find buffer conditions that are compatible with both ESI-MS and functional membrane proteins.

The majority of current methods are focused on the interaction between the ligand/modulator with the GPCR but lack a robust method to quantitatively measure all interacting components, including the ligand/modulator, the GPCR, and the transducer protein(s). The quantification of all interacting components, comprising the first stage of the GPCR signaling event, can provide an improved and more accurate method for drug discovery. Therefore, it is the objective of the present invention to provide a high-throughput mass spectrometric method that unravels ligand-mediated GPCR-protein complex interplay in a more detailed and quantitative way. This objective is achieved according to the present invention by a high-throughput quantification method to quantify protein- protein interaction of membrane proteins, comprising the steps of: a) providing a solution comprising a membrane protein, such as a GPCR, in the absence or presence of a chemical compound / compounds or ligands that can bind to or modulate the membrane protein;; b) adding a partner protein to the solution and after a predetermined time interval adding additionally a crosslinker that reacts with proteins' surface amino acid, such as lysines or other amino acids depending on the specific crosslinker that is used, in order to form chemical linkages that stabilize compound complexes of biomolecules; c) detecting and quantifying stabilized native and transient complexes of the biomolecules and the non-interacting biomolecule counterparts by mass spectrometry using a reference peak, i.e. in terms of a normalization strategy, to investigate the binding ability of the partner protein to the membrane protein.

Thus, this high-sensitivity, high-throughput mass spectrometry method interrogates the first stage of signal transduction. The membrane protein and partner protein complex formation is detected as a proxy for the effect of ligands on membrane protein conformation and on coupling selectivity. The method requires only very little probe amounts, such as little as 1.25 pmol protein per sample. The normalization step allows to quantitatively measure the binding affinities of membrane proteins with partner proteins. It is anticipated that this methodology will find broad use in screening and characterization of GPCR-targeting drugs. The crosslinking proteins provide a way that the GPCR, membrane or soluble protein (target protein) interaction with its partner protein are stably held together wherein the crosslinker reacts with the surface residues of the membrane protein and partner protein.

The crosslinking molecule comprises two or more reactive ends, thus being capable of chemically attaching to specific functional groups, such as primary amines or sulfhydryls.

The crosslinked 'target protein' with its interacted partner protein is then detected by high-mass MALDI mass spectrometry as a complexed peak. The reference protein for the normalization can be any soluble protein, other than beta-galactosidase, that is stable and monomeric in solution, and is detected as a stable single peak on high-mass MALDI mass spectrometer.

In a preferred embodiment of the present invention, BS(PEG)g, a bifunctional amine reactive reagent with a spacer arm length of 38.5 A, can be used to crosslink the partner protein via lysine residues wherein the lysine residues are present at the G- protein interacting interfaces of the membrane protein. In general, the crosslinker in general requires reactive terminals for specific functional groups, such as primary amines and sulfhydryls, present on the membrane protein and the partner protein.

Preferably, for the mass spectrometry an optimized MALDI sandwich spotting method can be used, preferably comprising a third layer of saturated sinapinic acid thereby considerably improving the signal level of the membrane proteins by MALDI detection and thus improving sensitivity of the overall method.

Further preferably, as reference peak in a normalization strategy b-galactosidase (b-gal) can be used.

In a further preferred embodiment of the present invention, the partner protein, such as nanobody, GTPase domain of Ga subunit (mGa), Ga subunit, a G-protein, or an arrestin, can be mutated and/or truncated in order to address a certain section and/or binding ability and/or functionality of the membrane protein, such as the GPCR.

Preferred embodiments of the present invention are hereinafter described in more detail with reference to the attached drawings which depict in:

Figure 1 schematically a workflow for the analysis of the selective coupling between GPCRs and partner proteins via high-mass MALDI-MS;

Figure 2 Selectivity in complex formation of apo- and ligand- bound GPCRs with partner proteins assayed by high-mass MALDI-MS;

Figure 3 Binding affinities between GPCRs and partner proteins;

Figure 4 schematically the role of the C-terminus of mGo and mGi on binding to GPCRs;

Figure 5 Quantitative competition between partner proteins and between ligands for binding to GPCR; and

Figure 6 Quantitation of the ligand-biased binding between βlAR and Gi/arrestin proteins.

The present invention realizes a quantitative high-mass matrix- assisted laser desorption/ionization mass spectrometry (MALDI- MS) strategy that combines chemical crosslinking and quantification based on an internal standard to assay the interplay between receptors, ligands, and interacting proteins. This versatile method enables to:

(i) elucidate the selectivity profile of G-proteins to GPCR;

(ii) dissect the molecular details of complex formation and probe the conformational regulation of GPCRs in an unprecedented way; and

(iii) determine the binding constant values and characterize ligand-ligand and protein-protein competitions.

Thus, this method has a much higher tolerance to buffer, salts, detergents, or lipids than ESI-MS. Moreover, it does not require any immobilization or chemical labelling of the purified proteins that might alter their bioactivity and integrity of the complexes during detection. The present high-throughput method (384 sample spots per MALDI plate, but other sample plates, like 1536 sample spots per MALDI plate, and the use of robotics may further increase the throughput) is sensitive (the required amount per sample is only 1.25 pmol), rapid (one spectrum can be recorded within 8 seconds), and quantitative. More than 70 ligand-GPCR-partner combinations have been studied during the creation of the present invention.

RESULTS

Optimization of Crosslinking Reaction and Spotting Method. The combination of crosslinking and mass spectrometry is a rapidly emerging approach to provide information on the structure and interaction networks of proteins. The GPCR-G protein interaction is transient and the complex is considered to be intrinsically unstable. Thus, capturing this interaction requires the use of certain stratagems such as stabilization of the complexes with nanobodies or antibodies, or recombinant technology to prevent their dissociation.

Lysine residues are present at the G-protein interacting interfaces of GPCRs. Based on this, BS(PEG)g, a bifunctional amine reactive reagent with a spacer arm length of 38.5 A, is used to crosslink interacting proteins via lysine residues.

After reaction, samples will contain intramolecular crosslinks, monolinks, and, most importantly, intermolecular crosslinks as shown in Fig. 1A. that stabilize and capture the protein-protein complexes in their equilibrium state, preventing them from dissociating during the MALDI process. The experimental conditions and crosslinking times were optimized using the prototypical photoreceptor rhodopsin (Rho), which couples effectively to mGo (a truncated form of Ga₀ subunit). It was found that even short (£1 min) pre-incubation with BS(PEG)g prevents the association between Rho and mGo, probably due to quick reaction of the crosslinker with lysine residues near the binding interfaces of Rho and mGo, precluding assembly of the complex. Using an optimized experimental procedure, it was found that in all of the Ga proteins or their truncated versions tested, 6-9 lysine residues react with BS(PEG)g, resulting in the formation of ~2 intermolecular crosslinks in each complex.

GPCRs are extremely challenging integral membrane proteins to work with as they are unstable in detergent solution and require the use of an appropriate condition for their extraction from the membranes. Since they are available in low quantity only, a sensitive detection method will therefore help reduce protein sample consumption. Thus, the MALDI sandwich spotting method was optimized by trial and error by testing various chemicals and the number of layers in the sandwiching method, and it was found in one preferred embodiment of the present invention that an addition of a third layer of saturated sinapinic acid considerably improved the signal level of GPCR proteins by MALDI detection and thus improved sensitivity. The method thus enabled to even detect picomole quantities of the GPCR-partner protein compound.

Ligand-Mediated GPCR Selective Coupling. Using this optimized crosslinking protocol, the present method is first showcased by examining the coupling ability of three class A GPCRs to a panel of mini-Ga proteins (hereafter abbreviated as mGa: mGs, mGo, mGi, mGq) and nanobody 80 (Nb80), in the presence or absence of various ligands (Fig. 2). The GPCRs studied were a constitutively active mutant of bovine Rho, thermostabilized turkey βlAR, and the F117W mutant of mouse angiotensin II type 1 receptor (AT1R) (protein sequences.

The detection and analysis of multi-component proteins complexes (such as GPCRs with their heterotrimeric G proteins) by any biophysical method is challenging. One therefore established our method by using mGa proteins, which are simplified versions of their full-length counterpart (Ga) containing the GTPase domain but lacking the a-helical domain, and are widely used in biochemical, biophysical, cellular and structural biology studies for studying GPCR-G-protein interactions and GPCR activation mechanisms. Swapping the c-tail (a5 helix) of the G protein is commonly performed to switch selectivity between G- protein subtypes. The present mGo and mGs are thermostabilised version of their truncated wild-type G-protein, and mGq and mGi are engineered from mGs by introducing nine and seven mutations on the a5 helix that correspond to residues of Gq and Gi, respectively. Mixing and incubation of the binding partners is followed by treatment with BS(PEG)g, and the resulting complexes and remaining unbound partners in the sample are detected by high-mass MALDI-MS by monitoring the peak intensities of each species. Examples of measured spectra are shown in Fig. IB, the results are summarized in Fig. 2. The present method allows to indirectly detect conformational changes and ensembles of the receptor by following receptor-complex formation, which can be read out directly from the mass sepctra.

GPCR orthosteric ligands fall into three categories: activating (agonists), inactivating (inverse agonists) and neutral (antagonists). The present assay largely displays the expected GPCR-G-protein recognition patterns. The constitutively active Rho mutant couples to the two members of the Gai_/0 family, mGo and mGi, both in the apo (apo-Rho) and agonist-bound (atr-Rho) forms (Fig. 2). This was expected, as constitutively active Rho has been shown to strongly recruit Gi and Go. The iso-βlAR was found to bind to Nb80 (a Gs mimetic nanobody), proving that the present βlAR construct can achieve a fully active conformation and that Nb80 binding is conformation specific. It has been shown that this receptor can couple to Ga_s, Gcg and Ga_q families and, indeed, it is observed that agonist-bound βlAR (iso-βlAR) can couple to some extent to all mGa subtypes (Fig. 2). Apo-βlAR can specifically couple to mGo, which showed similar selectivity profiles with known antagonists (propranolol, nadolol, and carvedilol) and s32212. Based on these profiles, it is possible to classify s32212 as an antagonist for βlAR. Finally, it is observed that the agonist-bound AT1R (angll-ATIR) couples to both mGq and mGo, but not mGi (Fig. 2). This could be because the mGi construct lacks some key residues required for receptor binding. As mGi is engineered from mGs and contains only the Gi fragment on the a5 helix, this suggests the a5 helix of Gi is not the main determinant for its coupling to AT1R and instead the globular part of Gi could be more important. This may also explain why a weak interactions of mGi to iso-βlAR and potentially weak interactions also to car-βlAR and angll-ATIR was observed (Fig. 2). Azilsartan, a potent inverse agonist can compete off many AT1R blockers. It is expected that this ligand stabilizes the receptor in an inactive conformation with severely impaired mGa coupling. Indeed, this ligand abolished coupling of all mGa proteins to the AT1R, including mGo (Fig.

2). These data illustrate how the apo, agonist-bound, antagonist-bound and inverse agonist-bound forms of receptors exist in different conformational ensembles with different profiles of G-protein recognition.

From the perspective of the mGa proteins, mGo is found to be the most promiscuous G-protein, as it binds to all agonist/antagonist-bound receptors and, remarkably, to all apo receptors (Fig. 2). Native Go protein is highly expressed in the central and peripheral nervous systems, endocrine cells, and cardiomyocytes, being the most abundant G-protein subtype in neurons. There is considerable evidence for the existence of functional complexes of apo-GPCRs with G-protein and the Go subtype seems particularly predisposed to such pre-coupling.

Thus, it is conjectured that the promiscuity of mGo observed in the present assay represents its ability to recognize apo (through pre-coupling), agonist-bound and antagonist-bound receptors.

A normalization strategy to determine binding Affinity of

GPCR»partner Complexes. Since ionisation efficiencies of proteins are highly variable in MALDI and could change upon crosslinking, there is no direct correlation between peak intensity and protein concentration. To be able to quantify individual protein components in the spectra, a normalization strategy was developed using b-galactosidase (b-gal) as a reference protein (an example of calibration and standard curve for Rho is shown in Fig. 3A and B), which is stable in its monomeric form and does not interfere with the analytes of the sample. This allows to calculate the concentrations of each species at equilibrium and the corresponding dissociation constants of the complexes between GPCRs and their partner proteins as shown in Fig. 3C.

The measured dissociation constants between GPCRs and interacting proteins (K_d) are in the high nanomolar to low micromolar range (summarized in Fig. 3). Literature K_d values are scarce because such measurements are challenging. A comparison of the MALDI-based K_d data with literature and a microscale thermophoresis measurement showed good agreement. It was observed that mGo generally had a higher affinity to the GPCRs compared to other partner proteins (Fig. 3). For blAR, the dissociation constant of mGo (0.25 mM) was hardly influenced by the ligands (Fig. 2) and was considerably lower than that of mGs (0.35 mM), mGq (1.24 mM), and mGi (1.62 mM). Among the receptors, bIAϊ! generally has higher affinities to the test partner proteins. For AT1R, binding to mGo is twice as strong than to mGq (Fig. 3). One quantitatively elucidated the interaction strength between the protein-protein complexes.

These interactions are the key determinant of information transmission within a signaling network.

Effect of the G-protein C-terminus on the Interaction with GPCRs. Many aspects of the formation of signaling complexes between GPCRs and G-proteins are still unclear, such as the molecular determinants of coupling selectivity or the role of pre-coupling of G-proteins to inactive receptors. Recent structural and biophysical studies have confirmed the C-terminus of the Ga subunit as one of the primary determinants of the interaction with GPCRs. The binding characteristics of the present mGa constructs show indeed that a few amino acid substitutions in the C-terminus of mGs, mGi, and mGq can alter their selective coupling to AT1R and Rho and impact the binding affinity to βlAR (Fig. 3).

To further assess the role of the mGa C-terminus, the last five residues from mGo and mGi (mGo_Δ5 and mGi_Δ5) were truncated and their binding affinity has been assessed to the present panel of receptors. The data show that mGi truncation abolished coupling to both apo and agonist-bound receptors (Fig. 4). However, truncation of mGo affected coupling to Rho and AT1R, but not to βlAR, which still binds mGo_Δ5 with similar affinities to mGo in both the apo (0.28 mM) and agonist-bound (0.23 mM) states. This indicates that the last five residues of G-protein are not always the main determinant for receptor recognition and other regions can mediate high-affinity binding. Based on the observation that ligands did not affect the affinity between βlAR and mGo, but had a significant effect on the binding of Rho and AT1R to mGo, it is speculated that ligand-induced GPCR conformational changes have a greater influence on the C- terminal contribution of the binding to the G-protein, and that GPCR and mGo interactions are receptor-dependent.

Ligand-Mediated Competition between Partner Proteins. To explore the interplay between affinity and selectivity in GPCR binding partners, the formation of βlAR complexes with mGa proteins (mGs, mGo, and mGq) is measured in the presence of the competitor Nb80 at equimolar amounts (Fig. 5A). In the absence of ligand, βlAR binds only to mGo due to its pre-coupling ability (Kd of 0.25mM) (Fig. 3), indicating that the ligand-free receptor ensemble is conformationally specific for mGo only.

Isoprenaline-bound βlAR selectively coupled with Nb80 in the presence of mGs or mGq, but couple with both mGo and Nb80. This is due to the tighter binding of Nb80 for isoprenaline-bound βlAR (0.21mM) compared to mGs (0.35 mM) and mGq (1.24 mM), while mGo binds with similar affinity to Nb80 (0.25 mM) (Fig. 3).

To measure the inhibition ability of Nb80 to mGo, the formation of βlAR^»mGo complexes was measured at increasing concentrations of Nb80 (Fig. 5D), and the inhibitory constant (Kj_.) of Nb80 to mGo (1.5710.24 mM) was calculated. One also measured the effects of isoprenaline on the competition between mGo and Nb80, and as expected, the competitiveness of Nb80 increases with rising isoprenaline concentration. These results show that when multiple partner proteins coexist, while GPCRs prefer to couple with partners of higher affinity, changes in ligand and partner concentrations can alter this coupling selectivity. It is substantiated that the promiscuous binding of mGo is specific for the two following reasons: first, one was able to displace mGo binding to AT1R in the present of the inverse agonist azilsartan, showing that mGo binding can be allosterically modulated by ligands (Fig. 2B). Second, Nb80 can also displace mGo binding to βlAR in a competitive manner (Fig. 5D). These results strongly suggest that mGo binds to the 'canonical' recognition site in the cytoplasmic side of the activated receptor.

Allosteric Influence of Ligands on GPCRs. One also investigated the allosteric conformational regulation of GPCR^»G-protein complexes by several ligands (Fig. 5B, C). All antagonists tested had the same effect on the coupling ability of bIAϊI, which binds only to mGo in their presence (Fig. 2). To further characterize these antagonists, one measured their ability to compete with the agonist and affect formation of the receptor*mGa complexes by incubating 2.5 mM apo-βlAR with equimolar amounts (50 mM) of antagonist (s32212, propranolol, carvedilol or nadolol) and agonist (isoprenaline) (Fig. 5B and C). At these concentrations, isoprenaline cannot compete off propranolol or carvedilol, and propranolol/carvedilol-bound βlAR still only recruits mGo, but it can compete off s32212 and recovers coupling to mGs, Nb80, and, partially, to mGq. Interestingly, in nadolol-bound βlAR, isoprenaline only partially recovers its recruiting ability with Nb80, but not with mGs and mGq.

One next explored in more detail the inhibitory ability of these antagonists on the formation of GPCR complexes. For that, one measured the formation of the βlAR^»mGs and βlAR^»Nb80 complexes in the presence of 1 or 25 mM of antagonists at increasing concentrations of isoprenaline (Fig. 5E and F). S32212 behaves as a surmountable competitive antagonist, as raising the isoprenaline concentration recovers near-maximal formation of the βlAR^»mGs complex (80%); the Ki of s32212 was determined to be 3.5610.26 mM.

On the contrary, propranolol behaves as an insurmountable competitive antagonist, as isoprenaline (at any concentration) cannot recover maximal βlAR^»mGs complex formation. Nadolol shows dual behaviour in different complex systems: it is insurmountable in βlAR^»mGs but surmountable in βlAR^»Nb80 (Fig.

5F), likely due to the higher affinity of Nb80 with isoprenaline-bound βlAR compared to mGs, and the allosteric effect of Nb80, which assists displacement of nadolol to isoprenaline. The positive cooperative effect of Nb80 on isoprenaline binding observed here is consistent with a previous report and demonstrates the allosteric mechanistic property of GPCRs. The present data agree with the concept that ligands induce (or stabilize) specific receptor conformations and the sensitivity of the present method reveals in detail the complexity of their interactions. It is shown that nadolol is more surmountable than propranolol, in agreement with their reported pKi values (-8.2 and -7.2, respectively). Furthermore, one shows for the first time that S32212 is a weaker antagonist for βlAR than nadolol, as shown by its less prominent inhibitory effect.

Ligand-Biased Assembly of the piAR^»G Protein/Arrestin Complexes.

Next, the present method is expanded by using full-length wild- type protein partners - GcgpY and b-arrestin-l (Fig. 6). One first incubated apo-, isoprenaline-, or carvedilol-bound βlAR with Gog, Gai^Gβ^Gy or b-arrestin-l at equimolar concentration and tested the formation of blAR·protein complexes. Artefacts were excluded by measuring mixtures of proteins that were pretreated with the crosslinker, which could not form protein complexes (Fig. 6B). It is found that isoprenaline-bound bIAA and ligand-free bIAA exhibited similar binding affinity to Gog and arrestin (~60% and 32% complex formation, respectively), while carvedilol-bound bIAA showed a higher affinity to Gog and arrestin (~92% and 88% complex formation, respectively).

One also tested the complex formation in an equimolar mixture of bIAA, Gai, and arrestin. It is found that both the b1AA·6aί and b1AA·arrestin complexes were present, but that the former formed much more readily than the latter (four times higher intensity with apo- or iso-βlAR and three times higher intensity with car- βlAR). This also illustrates that Gai possesses a higher binding affinity with βlAR than arrestin.

One then studied the interaction between ligand-bound βlAR and Gai^»Gβ^»GY. One incubated Gcq with Gβ^»Gy at equimolar concentration, and, as expected, peaks for the crosslinked complexes Gβ^»Gy (47,600 Da) and Gai^Gβ^Gy (91,500 Da) were detected (Fig. 6C). Additionally, a peak m/z at 53,200 Da is observed corresponding to a cross-linked complex of Gcq with Gy (Fig. 6C). Following addition of βlAR, one observed the simultaneous presence of the cross-linked complexes Ga^Gy,

Gcq ·Gβ · Gy, piAR'Gcy (82,800 Da) and piAR'Gcy * GP * Gy (130,900 Da) (Fig. 6C). The presence of isoprenaline hardly altered the relative intensity of these protein peaks compared to the absence of ligand, while carvedilol increased the formation of βlAR^Gcq^»Gβ ^» Gy resulting in a complete disappearance of the βlAR, Gβ^»GY, and Gcii ^» Gy peaks.

As car-βlAR does not bind mGi (Fig. 2), these data show that mGi did not inherit all the bioactivity from Gi, indicating that other regions of the Ga core domain make a large contribution to its receptor binding specificity. The present receptors were not treated with kinases or phosphorylation enzymes; in addition, the present βlAR construct is truncated at the C-terminus and intracellular loop 3, meaning that the majority of the phosphorylation sites is absent. The absence of phosphorylation, which precludes PKA-dependent Gs/Gi switching in the βlAR, is the probable cause of the lack of Gai^»Gβ^»GY recruitment observed for iso-βlAR (i.e., same response than the apo receptor; Fig.

2C).

Moreover, the data suggest that carvedilol-mediated arrestin coupling to βlAR is phosphorylation-independent. Importantly, the present method allows the quantification of Gi- and arrestin-complex formation induced by carvedilol, which quantitatively shows how ligands modulate the extent of the recruitment of G-proteins and arrestin.

DISCUSSION

Several recent technological advances have enhanced the understanding of various aspects of GPCR activation mechanisms and signaling. For example, structural biology studies by NMR, X-ray crystallography and cryo-EM have provided high-resolution structural insights, enabling the molecular characterization of different protein complexes. In addition, functional studies using biophysical and signaling assays have allowed the characterization of ligand properties and ligand-mediated cellular response. However, the characterization of the network of GPCR-protein interactions following receptor activation remains difficult to tackle.

While the traditional view of GPCR signaling involves a more or less sequential course of events, it is now clear that receptors can adopt multiple active states and engage multiple intracellular binding partners in a complex interaction network. To better understand the network of ligand-mediated GPCR^»G- protein interactions, the present invention discloses a method to address this by directly monitoring the GPCR-protein complex formation. One demonstrated the use of the present method by screening three class A GPCRs against a panel of engineered Ga proteins and generated a selectivity profile for each ligand tested (Fig. 2B). In agreement with a previous study, a G_±/0~ coupled receptor (Rho in this case) is more selective and couples only to G_± and G₀. Our G_s- and G_q-coupled receptors (βlAR and AT1R) are more promiscuous and always couple to some extent to the Gi_/o family as well (Fig. 2B). In order to fully understand the promiscuity of agonist-bound receptors, probably high- resolution structures of the same receptor bound to different transducers would be required to provide the molecular details and insights into this aspect.

The selectivity profiles of the three GPCRs indicate that each ligand-free or ligand-bound receptor has its unique coupling profile (Fig. 2B). Concurring with previous studies, it is also shown that agonist-bound GPCRs exist in multiple conformations (Fig. 2). This explains the complexity of the GPCR signaling mechanism, which is not governed simply by 'active' and 'inactive' states, or a ternary model. The method presented here allows to investigate GPCR interactions in an unprecedented way. The proportion of different ligands (agonist and antagonist) can further fine-tune the receptor conformational ensembles. Thus, the present data enables to observe the allosteric conformational regulation of GPCRs, which helps to explicate the plasticity of GPCR signal transduction.

The development and application of efficient GPCR binding assays are critical in the early stages of drug development. Current high-throughput technologies for assaying the function of GPCRs mainly depend on the measurement of second messenger output, such as inositol phosphate, calcium and cAMP. These readouts are distant from the actual information of the GPCR-effector complex, and rely on cellular responses that can be modulated by several separate or even cross-talking signaling pathways.

Therefore, the second messenger output does not directly indicate the 'recruiting' activity of a ligand and does not provide an accurate way to profile ligands according to this measure. Unraveling the relationships between ligand, receptor, and the coupling complexes (with G proteins and arrestins) that mediate downstream signaling events is the key to unscramble allosterism and biased signaling. It is showed here that the presented method can effectively be used to study the coupling of both G protein and arrestin (Fig. 6) and thus could potentially be used in drug discovery for ligand profiling.

Investigating the pentameric complex system

(ligand·βlAR·Gcg·Gβ·Gy) (Fig. 6C) was more complicated than the three-component systems (ligand»GPCR»mGa/G-protein/arrestin) and posed a challenge to obtain the binding affinity values for all components. However, the present data provide a unique profile for such pentameric system at equilibrium (Fig. 6C). Further expansion of the present method to study other members of the G protein, arrestin and G-protein kinase families may be of great relevance to future GPCR deorphanization approaches, or to dissect partially overlapping signaling pathways occurring in some of the G protein families, such as the Gi/o/z.

GPCRs are allosterically dynamic proteins. Multiple biophysical techniques are currently being used to fully understand how different ligands produce different signaling patterns. Complementary to previous techniques, the present strategy represents the first mass spectroscopic method that allows characterization of the direct ligand-induced receptor-protein complex formation in detail. One developed a powerful all-in-one method, unraveling the G-protein coupling selectivity to GPCRs and receptor conformational regulation, to provide information regarding the protein/analyte concentrations, their competition, affinity constants, molecular size and structure. It is therefore anticipated that the present method will emerge as a valuable strategy for high-throughput screening and for unravelling the molecular details of ligand-GPCR-protein interaction. Figure captions of the Figures 1 to 6

Figure 1 shows schematically a workflow for the analysis of the selective coupling between GPCRs and partner proteins via high- mass MALDI-MS. (A) Schematic of the crosslinking procedure resulting in stabilised GPCR^»G-protein complex plus unbound partners "decorated" with monolinks. (B) For assessing the ligand-mediated selectivity of a GPCR to a partner protein, the GPCR is first incubated with a mGa, nanobody 80 (Nb80), or G- protein, in the presence or absence of ligand. The GPCR^»partner complexes formed are then stabilised by chemical crosslinking, followed by detection of the protein components by high-mass MALDI-MS.

Figure 2 depicts the selectivity in complex formation of apo- and ligand-bound GPCRs with partner proteins assayed by high- mass MALDI-MS. (A) Three-dimensional structural models of mGa proteins and Nb80. The amino acid sequences of the C-terminal tail (helix 5, box) of the Ga subunit, accounting for ~70% of the interacting surface between GPCRs and G proteins, are shown for all mGa proteins (homology models of mGi, and mGq were built using SWISS-MODEL with mGs, PDB - 3SN6, as template); the last five key amino acids in mGa involved in selectivity determinant are underlined. (B) Complex formation propensity of three GPCRs - Rho, βlAR, and AT1R - in the presence or absence of agonists, antagonists, or inverse agonists with their partner proteins mGs, mGo, mGi, mGq and Nb80 is measured by comparing the relative peak intensity of the GPCR^»partner protein complex with that of the non-complexed GPCR. The ligands used were atr = all trans-retinal, iso = isoprenaline, pro = propranolol, nad = nadolol, car = carvedilol, angll = angiotensin II, azi = azilsartan; apo designates the ligand-free forms. Error bars represent standard deviations determined from three independent replicates. Figure 3 shows the binding affinities between GPCRs and partner proteins. (A) Calibration of different concentrations of Rho normalized to 2 mM of β-galactosidase. (B) Peak intensity ratio of Rho to β-galactosidase vs. Rho concentration in the sample.

(C) Evaluation of the affinities (dissociation constants K_d, measured in mM) for different GPCR with various partner proteins (mGs - orange, mGo - green, mGi - beige, mGo - turquoise, and Nb80 - magenta), using both apo (top panels) and ligand-bound (bottom panels) forms of the GPCRs. The data were obtained by titrating the G-protein against the GPCR in 20 mM Hepes buffer, pH 7.5, 40 mM NaCl, 0.01% lauryl maltose neopentyl glycol (LMNG). Error bars represent standard deviations from three independent replicates. N.D. = not determined.

Figure 4 elucidates the role of the C-terminus of mGo and mGi on binding to GPCRs. (A) Mass spectra showing the coupling between ligand-bound GPCRs (from left to right: apo-Rho, atr-Rho, apo- βlAR, iso-bIA, apo-ATIR, angll-ATIR) and truncated mGo (mGo_Δ5, first row) and mGi (mGi_Δ5, second row) proteins. (B) K_d values of apo-βlAR^»mGo_Δ5, (solid light green empty squares), iso- βlAR»mGo_Δ5 (dark green solid circle), apo-βlAR»mGi_Δ5 (light brown empty square), and iso-βlAR»mGi_Δ5 (dark brown solid square) (right panel). Error bars represent standard deviations from three independent repeats.

Figure 5 illustrates the competition between partner proteins and between ligands for binding to GPCR. (A) Schematic of the competition between Nb80 and other mGa proteins (mGs, mGo, and mGq) for binding to βlAR (in the presence or absence of ligand) and the different assembly possibilities. (B) Schematic of GPCR conformational ensembles induced by the competition between antagonist and agonist ligands. The GPCRs are stabilized in a suitable conformation under the combined effect of both ligands and partner proteins. (C) Schematic of the competition between nadolol and isoprenaline and the formation of the βlAR^»Nb80 complex, modulated by the presence of a partner protein. (D) Conversion of βlAR^»mGo (solid green circles) to βlAR^»Nb80 (solid magenta diamonds) using 2.5 mM βlAR, 3.0 mM mGo, and increasing concentrations of Nb80, and conversion to βlAR^»Nb80 in the absence of mGo (empty magenta diamonds). (E) βlAR^»mGs complex formation modulated by different ligands at different concentrations of isoprenaline. (F) Comparison of the βlAR^»mGs and βlAR^»Nb80 complex formation as revealed by titration with isoprenaline. Error bars represent standard deviations from three independent repeats.

Figure 6 shows ligand-biased binding between βlAR and Gi/arrestin proteins. (A) Structural models of the pentameric complex βlAR^»Gai^»Gβ^»Gy with bound isoprenaline (left; assembled using molecular graphics software (PyMOL) and the templates 3SN6, 2Y03, and 1GP2), and βlAR·b-arrestin-1 complex (right; PDB code 6TKO) with lysine residues highlighted in red. (B) Control experiment showing the absence of complex formation if the interaction partners are first treated with crosslinker (top panel), and complex formation between βlAR and y/arrestin

+arrestin in ligand-free and isoprenaline- and carvedilol-bound receptor. Complex formation in percentages was calculated by normalisation with b-Gal as a standard. (C) Formation of diverse complexes of Gog, b, Gy, and bIAB following incubation and treatment with BS(PEG)g, in the absence and presence of isoprenaline or carvedilol. Grey dashed traces are spectra recorded without applying crosslinker, blue dashed traces are spectra recorded after pre-treating mixture components with crosslinker before incubation. Percentage complex formation are calculated from three independent repeats.

Claims

Patent Claims

1. A high-throughput quantification method to quantify protein- protein interaction of membrane proteins, comprising the steps of: a) providing a solution comprising a membrane protein, such as a GPCR, in the absence or presence of a chemical compound / compounds or ligands that can bind to or modulate the membrane protein; b) adding a partner protein to the solution and after a predetermined time interval adding additionally a crosslinker that reacts with proteins' surface amino acid, such as lysines or other amino acids depending on the specific crosslinker that is used, in order to form chemical linkages that stabilize compound complexes of biomolecules; c) detecting and quantifying stabilized native and transient complexes of the biomolecules and the non-interacting biomolecule counterparts by mass spectrometry using a reference peak, i.e. in terms of a normalization strategy, to investigate the binding ability of the partner protein to the membrane protein.

2. The method according to claim 1, wherein BS(PEG)g, a bifunctional amine reactive reagent with a spacer arm length of 38.5 A, is used to crosslink the partner protein via lysine residues wherein the lysine residues are present at the G- protein interacting interfaces of the membrane protein.

3. The method according to claim 1 or 2, wherein, for the mass spectrometry an optimized MALDI sandwich spotting method has been used, preferably comprising a third layer of saturated sinapinic acid thereby considerably improving the signal level of the membrane proteins by MALDI detection and thus improving sensitivity of the overall method.

4. The method according to any of the preceding claims wherein, as reference peak in a normalization strategy b-galactosidase (b-gal) is used.

5. The method according to any of the preceding claims, wherein the mGa, nanobody 80 (Nb80), or a G-protein are mutated and/or truncated in order to address a certain section and/or binding ability and/or functionality of the membrane protein, such as the GPCR.