Allosteric modulation of adenosine A1 and cannabinoid 1 receptor signaling by G‐peptides

Abstract While allosteric modulation of GPCR signaling has gained prominence to address the need for receptor specificity, efforts have mainly focused on allosteric sites adjacent to the orthosteric ligand‐binding pocket and lipophilic molecules that target transmembrane helices. In this study, we examined the allosteric influence of native peptides derived from the C‐terminus of the Gα subunit (G‐peptides) on signaling from two Gi‐coupled receptors, adenosine A1 receptor (A1R) and cannabinoid receptor 1 (CB1). We expressed A1R and CB1 fusions with G‐peptides derived from Gαs, Gαi, and Gαq in HEK 293 cells using systematic protein affinity strength modulation (SPASM) and monitored the impact on downstream signaling in the cell compared to a construct lacking G‐peptides. We used agonists N6‐Cyclopentyladenosine (CPA) and 5’‐N‐Ethylcarboxamidoadenosine (NECA) for A1R and 2‐Arachidonoylglycerol (2‐AG) and WIN 55,212‐2 mesylate (WN) for CB1. G‐peptides derived from Gαi and Gαq enhance agonist‐dependent cAMP inhibition, demonstrating their effect as positive allosteric modulators of Gi‐coupled signaling. In contrast, both G‐peptides suppress agonist‐dependent IP1 levels suggesting that they differentially function as negative allosteric modulators of Gq‐coupled signaling. Taken together with our previous studies on Gs‐coupled receptors, this study provides an extended model for the allosteric effects of G‐peptides on GPCR signaling, and highlights their potential as probe molecules to enhance receptor specificity.


| INTRODUC TI ON
G protein-coupled receptors (GPCRs) have been the most successful class of drug targets in clinical medicine, due in part to their widespread distribution and important roles in physiology. 1 The pharmacological success of GPCRs derives from their selective coupling to specific heterotrimeric G proteins, triggering the corresponding physiological response. Recent drug discovery efforts have focused on the development of allosteric modulators for GPCRs. 2 Allosteric modulators have the potential to increase receptor specificity by targeting sequence motifs unique to receptor family subtypes and isoforms. Furthermore, allosteric modulators require the presence of an orthosteric ligand, providing physiological context-dependent control of GPCR signaling. 3 Therefore, compared to orthosteric ligands, large doses of allosteric modulators can be administered with a lower risk of target-based toxicity. 2 An emerging target site for allosteric modulators is the GPCR-G protein-binding interface.
The GPCR-G protein-binding interface contains sequence divergent structural elements including three intracellular loops and the GPCR C-tail. 4 However, the intrinsically disordered nature of the loop and C-tail, combined with the potential for binders in these regions to disrupt GPCR-G protein coupling has limited efforts to rationally design allosteric modulators that target the GPCR-G protein interface. 5 In this study, we examine the potential for the G protein α subunit C-terminus (G-peptide) to serve as an allosteric modulator of GPCR signaling. The G-peptide is a well-established determinant of GPCR-G protein coupling selectivity. 6,7 The G-peptide interacts at the cytosolic GPCR-G protein interface, which is distinct from the orthosteric ligand-binding pocket. The GPCR interaction with a cognate G-peptide triggers nucleotide exchange in the Gα subunit (GDP to GTP) resulting in G protein activation and downstream signaling.
While interactions with noncognate G-peptides do not precipitate G protein activation, we have recently shown that noncognate interactions alter receptor conformation resulting in enhanced ligand efficacy. 8,9 Previous studies show that while the noncognate G-peptide interactions are transient, the GPCR conformational state persists following dissociation resulting in the allokairic modulation (AKM) of downstream signaling. 8,9 Allokairic modulators bind asynchronously with the ligand and rely on the temporal persistence of GPCR conformation to exert their influence on orthosteric ligand efficacy. 9 Our previous studies focused on the Gs-coupled β2-adrenergic (β2-AR) and dopamine (D1R) receptors, which show enhanced cyclic AMP generation in the presence of a noncognate Gq protein. Likewise, the Gq-coupled V1 vasopressin receptor (V1R) shows enhanced IP 1 levels in the presence of the noncognate Gs protein. 8 In this study, we examine the potential for allokairic modulation of two canonical Gi-coupled receptors, adenosine type 1 (A1R) and cannabinoid type 1 (CB1) using G-peptides derived from Gs, Gi, and Gq subtypes.
While β2-AR and D 1 R principally signal through Gs, and A 1 R and CB 1 primarily signal through Gi. However, A 1 R and CB 1 display signaling through multiple G proteins with A 1 R signaling through Gi and Gq, and CB 1 signaling through Gi, Gq, and Gs. 10-13 CB 1 , the most widely expressed GPCR in the central nervous system, primarily signals through Gi producing euphoria and analgesia upon binding tetrahydrocannabinol (THC) in the brain. 14,15 CB 1 has also been shown to signal through Gq in human embryonic kidney (HEK) 293 cells after treatment with WIN55,212-2 (WN) 12 and through Gs in rat globus pallidus, HEK 293, COS-7, CHO, and 3T3 cells after treatment with WN. 10,13,16 However, the physiological effects of CB 1 signaling through Gs, Gq, and non-G protein-mediated pathways is less clear since there have not been biased ligands identified that specifically target these pathways. A 1 R is another example of a promiscuous receptor that can activate different signal transduction pathways in an agonist-dependent manner. A 1 R is ubiquitously expressed and most well known for being antagonized by caffeine, producing stimulant effects. 17 While A 1 R canonically signals through Gi, there is evidence that A 1 R has a diverse G protein-activating profile where A 1 R can adopt agonist-specific conformations, arising from small changes in ligand structure, which lead to the differential activation of G proteins including Gi and Gq. 11 This promiscuity of coupling in these canonical Gi receptors allows us to examine the allosteric effects of the G-peptide on multiple G protein signaling pathways.
The goal of this focused study is to examine the allosteric effects of G-peptides derived from three distinct Gα C-termini peptides (Gαs, Gαi, and Gαq) on signaling from two promiscuous Gi-coupled receptors (A 1 R and CB 1 ). The C-termini of three G proteins, Gαs, Gαi, and Gαq, will be referred to as s-pep, i-pep, and q-pep (or collectively as G-peptides) throughout this manuscript. We expressed A 1 R and CB 1 fusions with the s-, i-, or q-pep in HEK 293 cells using systematic protein affinity strength modulation (SPASM) and monitored the impact on downstream signaling in the cell compared to a construct lacking this G-peptide, referred to henceforth as no-pep.
We have extensively reported on this SPASM technique, which allows systematic control of the intramolecular interaction between a GPCR and a G-peptide. 6,8,18,19 This technology allows us to directly compare the influence of different G-peptides on the cognate G protein signaling pathways in cells. While this is a tethered system, we have shown that these engineered GPCR constructs yield similar results to reconstituted systems of GPCR membranes and recombinant G proteins with regards to allokairic modulation of G protein activation. 8,19 Hence, despite the synthetic nature of our approach, it provides insight into the impact of receptor interactions with G-peptides on downstream signaling.
To investigate the allosteric effects of G-peptides on Gicoupled receptors, we used N 6 -Cyclopentyladenosine (CPA) and 5'-N-Ethylcarboxamidoadenosine (NECA) for A 1 R and 2-Arachidonoylglycerol (2-AG) and WIN 55,212-2 mesylate (WN) for CB 1 . Our current study confirms what we previously found in Gs-coupled receptors β 2 -AR and D 1 -R, where s-pep and q-pep positively modulate canonical Gs signaling. 8 cAMP response at high concentrations of 2-AG and WN is enhanced by q-pep (~30% and 95% increase in cAMP, respectively). Likewise, cAMP stimulation by WN at CB 1 is enhanced by s-pep (~40% increase). In contrast, i-pep diminishes cAMP response from CB 1 for both 2-AG and WN (30 and 50% decreases, respectively). At low concentrations of 2-AG, WN, and CPA (nmol/L) we observed inhibition of cAMP, associated with signaling through Gi. We found that the presence of q-pep or i-pep enhanced canonical Gi signaling in A 1 R after activation by CPA (~35% increase), and in CB 1 after activation by WN (~700% increase) and 2-AG (~125% increase), respectively.
These findings extend our previously reported allosteric effects of G-peptides to Gi-coupled signaling. 8,9 At high concentrations of 2-AG, WN, CPA, or NECA (μmol/L), stimulation of inositol phosphate (IP 1 ) is observed, associated with signaling through Gq. We found that the presence of different G-peptides universally inhibits IP 1 signaling through Gq (decreases ranging from 30% to 65%), with the exception of s-pep (~50% increase) on CB 1 following activation by WN. Taken together, our data provide an extended model for the allosteric effects of distinct G-peptides on signaling through Gs, Gi, and Gq pathways and highlight the ability of G-peptides to differentially impact signaling in a receptor and ligand-dependent manner.

linker was inserted
between all protein domains as part of the primer sequence to allow for free rotation between domains. All sensors also contained either an N-terminal HA-tag or a His-tag. All constructs were confirmed by sequencing.

| IP 1 assays
HEK293T cells expressing the indicated sensor were harvested 28-32 h posttransfection (X-tremeGENE HP) to assess IP 1 levels using the IP-One HTRF assay kit (Cisbio). Cells were gently suspended in their original media, counted using a hemocytometer, and spun down

| Statistical analysis
Data are represented as mean values ± SEM. All experiments were repeated for at least three independent trials, with three to six technical repeats per condition (N > 3). Statistical analysis was performed using GraphPad Prism 7.0c (Graphpad Software, Inc). To assess significance across experimental repeats, pooled or un-pooled data underwent subsequent pairwise ANOVA analysis. Tukey's post hoc test was performed to assess significance when evaluating comparisons between multiple conditions with P-values *P ≤ .05; **P ≤ .01; ***P ≤ .001; ****P ≤ .0001; *****P ≤ .00001.

| Nomenclature of targets and ligands
Key protein targets and ligands in this article are hyperlinked to corresponding entries in http://www.guide topha rmaco logy.
org, the common portal for data from the IUPHAR/BPS Guide to PHARMACOLOGY, 20 and are permanently archived in the Concise Guide to PHARMACOLOGY 2019/20: G protein-coupled receptors. 21
From N-to C-terminus, each SPASM sensor contains a GPCR, mCitrine (to monitor sensor integrity), 10 nm ER/K linker, mCerulean (for matching receptor expression), and a 27-amino acid peptide derived from the α5-helix at the C-terminus of the Gα subunit (s-pep, i-pep, q-pep, or no-pep). We chose the 10 nm linker based on previous work, where we found that a shorter linker corresponded to a higher effective concentration of the protein interaction ( Figure S1, left). 22 We had previously shown that a peptide derived from Gαs (s-pep) could enhance Gs signaling through β2-AR, and we confirmed this in Figure S1 with β2-AR producing a significant increase in cAMP when tethered to the s-pep (Sp) by either a 10 or 20 nm linker. 8 However, we observed no significant increase in cAMP production by β2-AR when tethered to s-pep by a 30 nm linker ( Figure S1). We therefore used a 10 nm linker to tether peptides to GPCRs for subsequent experiments, since it appeared that the effective concentrations enforced by either a 10 or 20 nm linker were required to modulate signaling. The Gα C-terminal peptides have been shown to be essential for activation by the GPCR but do not themselves trigger downstream effectors. 6,8,[23][24][25][26][27] In previous studies we have shown the ability of SPASM sensors to be expressed and localized primarily to the plasma membrane in HEK 293 cells. 28 Our SPASM sensors are therefore designed to modulate the interaction between the attached receptor (A 1 R or CB 1 ) and endogenous G proteins in cells, allowing one to study the impact of the tethered Gα peptides on canonical GPCR signaling. 19 SPASM A 1 R and CB 1 constructs lacking a C-terminal peptide (no-pep) were used to measure background cAMP and IP 1 levels and for characterization of ligand dose-response.

| Impact of Gα C-terminal peptides on cAMP response in the Cannabinoid (CB 1 ) Receptor
Cells expressing the CB 1 sensor display potentiation of forskolin-   Figure 1B and C, red dashed lines) to characterize the impact of peptides on cAMP stimulation and Gs signaling ( Figure 1D and E) and at low concentration (50 or 300 nmol/L, Figure 1B Figure 1D) or WN ( Figure 1E) to stimulate cAMP production through the Gs pathway ( Figure 1B and C). The q-pep sensor was found to increase signaling through Gs in CB 1 , as evidenced by a significant increase in cAMP levels ( Figure 1D and E, blue bars). This finding in a Gi-coupled receptor extends our previous results where q-pep exhibited enhanced signaling in the Gs pathway in Gs-coupled receptors. 8 S-pep sensors also increased signaling through Gs in CB 1 after stimulation by WN ( Figure 1E, red bar). In contrast, the presence of i-pep inhibited Gs signaling in CB 1 after stimulation by 2-AG or WN, decreasing cAMP levels ( Figure 1D and E, green bars). These findings are also summarized in the schematic ( Figure 1F Figure 1I).  Figure S2). We therefore could not characterize the impact of Gα peptides on Gs signaling in A 1 R.

| C-terminal Gα Peptides Inhibit Gq Signaling from Promiscuous Receptors
Previous work from our lab suggests that the effect of noncanonical G proteins on IP 1 signaling are more receptor specific. 8 We found that Gs enhances IP 1 production and signaling through Gq in the vasopressin receptor (V 1A -R) but not the α1 adrenergic receptor (α1-AR). 8 In the current study we examined the impact of Gα peptides on

| D ISCUSS I ON
In this study we demonstrate the allosteric modulation of two Gicoupled receptors, A 1 R and CB 1 , using peptides derived from the C-terminus of the Gα subunit (G-peptides). G-peptides derived from Gαi and Gαq (i-pep and q-pep) enhance agonist-dependent cAMP inhibition, demonstrating their function as positive allosteric modulators of Gi-coupled signaling. In contrast, i-pep and q-pep suppress agonist-dependent IP 1 levels suggesting that they function as negative allosteric modulators of Gq-coupled signaling.
Taken together with our previous studies focused on Gs-coupled receptors, our findings reinforce the potential of G-peptides to allosterically modulate signaling from class A GPCRs. 8,9 While allosteric modulation of GPCR signaling has gained prominence to address the need for receptor specificity, efforts have mainly focused on allosteric sites adjacent to the orthosteric ligand-binding pocket and lipophilic molecules that target transmembrane helices. 2 In contrast, here we use as G-peptides as probe molecules to demonstrate allosteric modulation through the GPCR-G protein binding interface.
The two Gi-coupled receptors (CB 1 and A 1 R) examined in this study have also been reported to signal to varying degrees through other G proteins. [10][11][12][13]16 Figure S5). In accordance with a previous studies, we confirmed CB 1 did indeed signal through Gs, as no significant potentiation of forskolin-induced cAMP production was observed in untransfected HEK 293 cells stimulated by the CB 1 agonists 2-AG or WN ( Figure S2). 10,13,16 We therefore used CB 1 to examine the impact of G-peptides on Gs signaling, with findings consistent with our previous report for the Gs selective β2-AR receptor ( Figure 1D-F). 8 Our data contrast with previous studies that report inhibition of GPCR signaling by native cognate G-peptides. [32][33][34] In these studies, minigene vectors were used to overexpress cognate G-peptides in cells at arbitrarily high concentrations, in order to identify and selectively inhibit cognate G protein engagement with the receptor. Accordingly, we have previously shown that high concentrations of cognate G-peptides (100 μmol/L s-pep) can competitively inhibit signaling from Gs-coupled receptors. 8,9 In contrast, we find that noncognate G-peptides can bind weakly to the receptor and serve as positive allosteric modulators. 8,9 While no significant positive allosteric effects were noted in studies with minigene vectors encoding noncognate G-peptides, these could be attributed to the variation and/or lack of control in expression since saturating levels would result in inhibition. [32][33][34] To alleviate the confounding effects of G-peptide concentration, we used the SPASM constructs to provide equivalent effective concentrations of distinct G-peptides across different receptor-ligand-pathway combinations. Furthermore, the ER/K linker in the SPASM sensors provides an effective concentration of approximately 10 μmol/L, 18 which is significantly lower than our previously reported threshold for competitive inhibition by cognate G-peptides. Using this technology, we observe differential effects of G-peptides on distinct pathways emerging from the same receptor. Specifically, while both i-pep and q-pep augment Gi-mediated cAMP inhibition, they suppress IP 1 accumulation downstream of Gq activation. Given that sensor expression levels were matched between cAMP and IP 1 assays and the ER/K linked G-peptides (i-pep and q-pep) are presented at equal effective concentrations, it is unlikely that inhibition of Gq signaling stems from a simple competitive inhibition mechanism. Instead, the differential effects of G-peptides likely stem from the dynamic conformational landscape of GPCRs. 35,36 We propose a model wherein transient interactions with G-peptides alter receptor conformation. The receptor does not form a stable ternary complex with the G-peptide and therefore at low concentrations (10 μmol/L) does not interfere with the kinetics of the receptor-cognate G protein interaction. 9 However, the altered receptor conformation triggered by G-peptide binding impacts ligand efficacy for cognate G protein activation, resulting in positive or negative allosteric modulation of downstream responses. The inability of the G-peptides, especially those derived from noncognate G proteins, to form stable interactions with the receptor has been previously observed in A 1 R-Gi fusions. 37 The lack of stable ternary complex formation with noncognate G proteins has been suggested as a kinetic proofreading mechanism to prevent noncognate GPCR-G protein coupling. 37 Nonetheless, we have previously shown that both cognate and noncognate G-peptide interactions influence receptor conformation. 9 Transient interactions of the G-peptide at the cognate G protein binding site on the receptor stabilize a distinct receptor conformational state. This conformational state persists following G-peptide dissociation enabling increased efficacy of subsequent cognate G protein coupling and enhanced downstream signaling. 9 Given that the G-peptide and cognate G protein share the same binding site, albeit staggered in time, we propose that the G-peptides function as allokairic modulators (AKMs) of cognate GPCR signaling. Allokairy is an established concept in enzymatic reactions, wherein increased substrate concentrations can increase maximal reaction rates, especially if the substrate stabilizes a distinct active enzyme conformation. 38 AKMs can bind asynchronously with the orthosteric ligand and rely on temporally persistent conformational states of the enzyme to exert their effects. 9 G-peptides as AKMs provide access to the entire GPCR-G protein interaction interface for allosteric modulation, without necessarily competing with cognate G protein coupling. Targeting the GPCR-G protein interface offers the potential to enhance receptor specificity, especially given the three intrinsically disordered loop regions with considerable isoform specific sequence homogeneity.

E TH I C S S TATEM ENT
No animals, human tissue, human volunteers, or patients were used in this study. Touma and Sivaramakrishnan wrote manuscript.

DATA AVA I L A B I L I T Y S TAT E M E N T
Additional information and requests for data and/or reagents should be directed to the corresponding author, Dr Sivaraj Sivaramakrishnan.