GAIN domain–mediated cleavage is required for activation of G protein–coupled receptor 56 (GPR56) by its natural ligands and a small-molecule agonist

Adhesion G protein–coupled receptors (aGPCRs) represent a distinct family of GPCRs that regulate several developmental and physiological processes. Most aGPCRs undergo GPCR autoproteolysis-inducing domain–mediated protein cleavage, which produces a cryptic tethered agonist (termed Stachel (stinger)), and cleavage-dependent and -independent aGPCR signaling mechanisms have been described. aGPCR G1 (ADGRG1 or G protein–coupled receptor 56 (GPR56)) has pleiotropic functions in the development of multiple organ systems, which has broad implications for human diseases. To date, two natural GPR56 ligands, collagen III and tissue transglutaminase (TG2), and one small-molecule agonist, 3-α-acetoxydihydrodeoxygedunin (3-α-DOG), have been identified, in addition to a synthetic peptide, P19, that contains seven amino acids of the native Stachel sequence. However, the mechanisms by which these natural and small-molecule agonists signal through GPR56 remain unknown. Here we engineered a noncleavable receptor variant that retains signaling competence via the P19 peptide. We demonstrate that both natural and small-molecule agonists can activate only cleaved GPR56. Interestingly, TG2 required both receptor cleavage and the presence of a matrix protein, laminin, to activate GPR56, whereas collagen III and 3-α-DOG signaled without any cofactors. On the other hand, both TG2/laminin and collagen III activate the receptor by dissociating the N-terminal fragment from its C-terminal fragment, enabling activation by the Stachel sequence, whereas P19 and 3-α-DOG initiate downstream signaling without disengaging the N-terminal fragment from its C-terminal fragment. These findings deepen our understanding of how GPR56 signals via natural ligands, and a small-molecule agonist may be broadly applicable to other aGPCR family members.

Most aGPCRs undergo GAIN domain-mediated autoproteolysis during protein maturation to generate an N-terminal fragment (NTF) containing adhesion domains and most of the GAIN domain and a C-terminal fragment (CTF) containing a cryptic tethered peptide agonist, called Stachel, that emanates extracellularly from the seven-transmembrane-spanning (7TM) domain (6,7). The NTF and CTF remain noncovalently associated on the plasma membrane (1)(2)(3)(4)(5)8). Currently, there are three models to describe aGPCR activity modulation: ligand binding dissociates the NTF from its CTF, exposing the tethered agonist to activate downstream signaling via heterotrimeric G proteins (6,7,9,10); direct and transient interaction between the extracellular and 7TM domains alters downstream signaling in a Stachel-independent manner (11,12); or ligand binding induces a conformation change that exposes the Stachel sequence and leads to receptor activation without cleavage and/or removal of the NTF (13). These models are unlikely to be mutually exclusive and could, in principle, occur in the same receptor. ADGRG1 (GPR56) has critical functions in the development of several organ systems with broad implications for human diseases and their treatment (3, 14 -31). Mutations in GPR56 cause a severe human brain malformation called bilateral frontoparietal polymicrogyria, characterized by cortical lamination defects, cerebellar hypoplasia, and central nervous system (CNS) hypomyelination (18,32). In the CNS, GPR56 regulates embryonic brain development and postnatal CNS myelination (22,26,27,(33)(34)(35)(36)(37). Myelin is a multilayered glial membrane surrounding axons in the vertebrate nervous system. In the CNS, myelin is formed by specialized glial cells called oligodendrocytes (38). GPR56 promotes oligodendrocyte development and CNS myelination by coupling to G␣ 12/13 and leading to activation of the RhoA pathway (26,27). In the peripheral nervous system, GPR56 regulates myelin formation and maintenance (23).
As a member of the aGPCR family, GPR56 undergoes GAIN domain-mediated cleavage to generate an NTF and a CTF, which remain noncovalently associated on the plasma membrane (8,39). Current literature supports cleavage-dependent (7,40) and cleavage-independent signaling mechanisms for GPR56 (11,12). To date, there are two natural ligands, collagen III and tissue transglutaminase 2 (TG2), and one small molecule agonist, 3-␣-acetoxydihydrodeoxygedunin (3-␣-DOG), of GPR56. Collagen III is the ligand of GPR56 in the developing cerebral cortex and skeletal muscles (28,35,36,41) and TG2 in melanoma cells (39) and oligodendrocytes (22). Additionally, a 19-amino-acid synthetic peptide (P19) that mimics the GPR56 Stachel sequence functions as a natural tethered agonist for GPR56 (42). However, it remains elusive whether these confirmed natural and small-molecule agonists of GPR56 require GAIN domain-mediated cleavage to activate G-protein signaling downstream of the receptor.
In this study, we show that P19 can equally activate both cleaved and noncleaved receptors, whereas the other three agonists require GAIN domain-mediated receptor autoproteolysis. Furthermore, we demonstrate that TG2 ligand action demands the presence of a matrix protein, laminin, thus mediating a tripartite form of signaling among GPR56, TG2, and matrix protein. Taken together, our results provide support for cleavage-dependent GPR56 activation by its natural and smallmolecule agonists.

GAIN domain-mediated cleavage is required for GPR56 activation by natural and small-molecule agonists
To investigate the requirement of GAIN domain-mediated autoproteolysis for GPR56 signaling, we first generated GPR56 mutants that abolish receptor cleavage but retain cell surface expression. Based on the published GAIN domain crystal structures (5), we made two missense mutants, H381S and L382A (Fig. 1, A-C). Immunostaining with an anti-mouse GPR56 mAb, H11 (35), showed that both mutant receptors had cell surface abundances that are comparable with the WT receptor (Fig. 1, D and E). Western blot analysis revealed that GPR56 H381S completely lacks the cleaved NTF, whereas ϳ 56% of the GPR56 L382A is cleaved (Fig. 1, F and G), showing that the GPR56 H381S receptor is defective in undergoing GAIN Cleavage-dependent ADGRG1/GPR56 signaling domain-mediated cleavage. Therefore, we chose GPR56 H381S for the remainder of the study. We next examined the basal activity of this mutant receptor compared with WT GPR56 using a serum response element (SRE)-luciferase assay and found that GPR56 H381S mutant has significantly reduced basal activity (Fig. 1H). The lower basal activity associated with the GPR56 H381S mutant is not the result of poor overall health of the transfected cells because there were comparable cell numbers over a 48-h period after transfection between cells transfected with WT and mutant receptors (Fig. S1).
We next investigated whether GAIN domain-mediated receptor cleavage is required for GPR56 activation by the known agonists collagen III (35), 3-␣-DOG (42), and P19 (42), none of which require cofactors for receptor activation. Collagen III, P19, and 3-␣-DOG all equally activated WT receptormediated downstream G protein-dependent signaling by SRE-luciferase assay ( Fig. 2A). However, only P19 could elicit signaling from the GPR56 H381S mutant receptor (Fig. 2). To further confirm this observation, we performed GTP-RhoA pulldown assays to measure levels of activated RhoA upon ligand stimulation. Treatment with 3-␣-DOG, P19, and collagen III resulted in robust RhoA activation (Fig. 2, B and C). This result indicated that the GPR56 H381S mutant receptor retained signaling competence but did not respond to natural or smallmolecule agonist stimulation.

TG2-mediated GPR56 activation requires both receptor cleavage and the presence of laminin
TG2 was first identified as a binding partner of GPR56 in melanoma cells, although the signaling consequences of this interaction were not revealed (39,43). Very recently, we discovered that TG2 is also the ligand of oligodendrocyte-expressed GPR56 during CNS myelination by promoting OPC proliferation (22). We performed in vitro experiments using Gpr56 ϩ/ϩ and Gpr56 Ϫ/Ϫ oligodendrocyte precursor cells (OPCs) and recombinant TG2 and found that TG2 can only augment the proliferation of OPCs in the presence of laminin (Fig. 3A). This effect was GPR56-dependent and highly specific because TG2 plus laminin had no effect on Gpr56 Ϫ/Ϫ OPCs and fibronectin, another extracellular matrix protein that is known to interact with TG2, had no effect in this assay (Fig. 3A).
To extend our characterization of TG2 as the ligand for GPR56, we investigated its ability to activate GPR56 downstream signaling in vitro on WT GPR56 and cleavage-deficient GPR56 H381S using SRE-luciferase assays. We found that TG2 could activate WT GPR56 in the presence of laminin, whereas TG2 or laminin alone were not effective (Fig. 3B). Similarly, we only observed a robust elevation of GTP-RhoA upon dual stimulation of TG2 and laminin (Fig. 3, C and D). In contrast, the GPR56 H381S receptor was not responsive to stimulation with TG2 and laminin together (Fig. 3B), indicating that GPR56 cleavage is required for TG2/laminin-mediated GPR56 activation.

Ligand-induced shedding of GPR56 NTF is required for downstream signaling
A potential explanation for the inability of TG2 binding alone to activate GPR56 is that this interaction is not sufficient to dissociate the NTF from its CTF. As a corollary, we hypothesized that GAIN domain-mediated cleavage was also required to generate a receptor permissive for dissociation of the NTF by TG2 plus laminin. To test this, we treated serum-starved, GPR56-transfected HEK293T cells with TG2 alone or TG2 plus laminin for 10 min. The cell pellets were used to measure GPR56-stimulated GTP-RhoA levels by RhoA pulldown assay (Fig. 3, C and D), and the supernatants were concentrated and subjected to Western blot analysis with the anti-GPR56 NTF antibody H11. Strikingly, we found that TG2 in the presence of laminin causes GPR56 NTF dissociation and shedding in conjunction with increased GTP-RhoA levels in the cell pellet, whereas TG2 alone was incapable of dissociating the GPR56 NTF and did not lead to increased GTP-RhoA levels (Figs. 4, A and C, and 3, C and D). We demonstrate that GPR56 mediates a tripartite interaction of receptor, ligand, and matrix protein and that ligand-induced shedding of GPR56 NTF is required for downstream signaling.
We next asked whether dissociation of the GPR56 NTF is a general mechanism for all of its known ligands. To test this, we performed similar NTF shedding experiments for 3-␣-DOG, P19, and collagen III. Only collagen III treatment, but not 3-␣-DOG and P19, resulted in increased levels of NTF in the supernatant (Fig. 4, B and D).

3-␣-DOG is a partial orthosteric agonist of GPR56
3-␣-DOG is a small molecule that was first identified in a high-throughput screening assay (42). It activates truncated GPR56 7TM with an EC 50 of ϳ5 M (42). Small-molecule aggregations or nonspecific inhibiting or activating proteins in vitro could be a major source of false positive results in drug discovery (44 -46). Therefore, before exploring the 3-␣-DOG Cleavage-dependent ADGRG1/GPR56 signaling activation mechanism, we examined whether 3-␣-DOG is an aggregator using an AmpC ␤-lactamase enzymatic assay. AmpC is one of the most sensitive enzymes and has been studied extensively for aggregate-based inhibition (47). 2 nM AmpC ␤-lactamase was incubated with varying concentrations of 3-␣-DOG (from 1.56 M to 100 M) for 5 min at room temperature. 3-␣-DOG reduced AmpC activity to ϳ65% only at the highest concentration of 100 M (Fig. 5A). To investigate whether 3-␣-DOG forms colloids, a dynamic light scattering assay was used (48). Different concentrations of 3-␣-DOG were measured at a wavelength of 826.6 nm. 3-␣-DOG scattered as an aggregate only at 50 and 100 nM (Fig. 5B). Taken together, these results suggest that 3-␣-DOG is not an aggregator at any relevant concentration near its EC 50 .

Cleavage-dependent ADGRG1/GPR56 signaling
To investigate whether 3-␣-DOG functions as either an orthosteric activator or an allosteric modulator, we performed a competition assay by examining the efficacy of P19 in the presence of 3-␣-DOG. To our surprise, 3-␣-DOG inhibits P19 activity at concentrations as low as 1 M (Fig. 5C). It is worth noting that a previously published result demonstrated that 3-␣-DOG requires the presence of three of the seven amino acids (VLM) in the Stachel sequence to activate the receptor (Fig. 5D) (42). Taken together, our findings support that 3-␣-DOG is a partial orthosteric agonist that requires the presence of a partial Stachel sequence to activate the receptor.

Discussion
The evidence to date indicates aGPCR signaling modes that are either GAIN domain-mediated receptor cleavage-dependent or cleavage-independent (1,3,6,7,9,(11)(12)(13). However, most of the studies that were performed to arrive at these models did not test the influence of natural ligands or agonists. In this study, we investigated the signaling requirements of natural GPR56 ligands, a synthetic peptide agonist, and a smallmolecule agonist. We focused on G␣ 12/13 and the RhoA pathway because this GPR56 signaling pathway is the only one to have been documented in vivo (22,26,27,35) and in vitro (7,41,42). To investigate the requirement of receptor cleavage for its downstream signaling, we generated a cleavage-deficient receptor, GPR56 H381S . We demonstrated that both natural ligands (collagen III and TG2) require receptor cleavage and activate GPR56 via a Stachel-dependent mechanism. Our previous work showed that collagen III facilitates RhoA activation by removing the GPR56 NTF from its CTF (40). Here we further demonstrate that collagen III requires GAIN domainmediated receptor cleavage to initiate downstream signaling (Fig. 6A).
Our results are consistent with published results. Using engineered truncated GPR56 CTFs that either possess or lack the Stachel sequence, Hall and co-workers (12) demonstrated that the Stachel sequence is required in a pathway-dependent manner; activation of SRE luciferase required the presence of the Stachel sequence, whereas transforming growth factor ␣ shedding and activation of nuclear factor of activated T cells (NFAT) luciferase were Stachel-independent.
TG2 has been shown previously to be the GPR56 binding partner in melanoma cells (39). Recently, we showed, through a combined genetic and unbiased proteomic approach, that TG2 is the ligand of oligodendrocyte GPR56 (22). However, the downstream signaling of this binding remained largely unknown. This study elucidates a novel tripartite signaling model for GPR56, TG2, and laminin. TG2 has also been shown to interact directly with laminin (49). Here we show that TG2,  (42). The seven-amino-acid Stachel is depicted as seven solid green circles (each represents one amino acid).

Cleavage-dependent ADGRG1/GPR56 signaling
in the presence of laminin, binds to the GPR56 NTF and dissociates the NTF from the CTF, allowing the endogenous GPR56tethered peptide agonist to initiate G-protein signaling (Fig. 6, B and C). It is not possible to directly test the need for the Stachel sequence for GPR56 signaling because this will render a noncleavable receptor. Nevertheless, our data are very compelling in support of the Stachel-dependent signaling mechanism by demonstrating that shedding of the NTF leads to receptor activation (Figs. 3 and 4).
As P19 is a synthetic mimetic of the authentic tethered peptide agonist, it presumably engages the receptor orthosteric site, which is unaffected in both the cleaved and cleavage-deficient receptor (Fig. 6D). On the contrary, 3-␣-DOG can only activate the cleaved receptor, unlike the synthetic peptide agonist P19 (Fig. 2). Combining the previously published results using a series of truncated GPR56 7TM constructs that contain various Stachel sequence lengths (Fig. 5D), our results support the notion that 3-␣-DOG functions as a partial orthosteric agonist of GPR56 that requires the presence of part of the Stachel sequence (Fig. 6E). The crystal structure of GPR56 NTF confirms partial exposure of the Stachel sequence in cleaved GPR56 (50). We suspect that the noncleavable GPR56 H381S receptor does not possess such exposure and therefore cannot be activated by 3-␣-DOG (Fig. 6E).

Cell surface immunofluorescence imaging and Western blotting
HEK293T cells were used for immunoblotting and immunocytochemistry experiments. For cell surface staining of GPR56, HEK293T cells were plated on poly-D-lysine coated coverglass and transfected with WT, H381S, or L382A mutant Gpr56 cDNA using Xfect transfection reagent (Clontech) and cultured for 48 h. The transfected cells were fixed in 2% (PFA) paraformaldehyde for 10 min without permeabilization and then incubated with a mouse anti-GPR56 (H11) antibody visualized by mouse Alexa 546. Images were captured by a confocal LSM 510 NLO system.
For Western blotting, transfected HEK293T cells were harvested 2 days after transfection and solubilized for 1 h at 4°C in RIPA buffer (Boston BioProducts). The cell lysates were used for Western blotting with a standard Western blotting protocol.

TG2 activation assay
HEK 293T cells were transiently transfected with mouse Gpr56 cDNA. Twenty-four hours after transfection, the cells were subjected to serum starvation for 36 h, followed by addition of recombinant TG2 (200 nM) with or without recombinant laminin (2 g/ml) or acetic acid as a control for 10 min. The conditioned media were harvested, filtered, and concentrated as described previously (8). Equal volumes of the concentrated media were used for Western blotting (40), whereas the cell pellets were used for the Rho-GTP pulldown assay.

Rho-GTP Rhotekin pulldown assay
The Rho-GTP pulldown assay was performed as described previously (35), using HEK293T cells transfected with mouse Gpr56 cDNAs. The transfected cells were stimulated with recombinant TG2 (200 nM) with or without recombinant laminin (2 g/ml). The treated cells were lysed in 300 l of ice-cold RIPA buffer containing protease inhibitor mixture (Boston BioProducts). Equal amounts of total lysate protein were incubated with 60 g of GST-RBD beads at 4°C for 90 min. The beads were washed twice with RIPA buffer and once with Tris-buffered saline. Bound Rho proteins were eluted by Laemmli sample buffer and detected by Western blotting using mouse monoclonal anti-RhoA antibody.

Luciferase reporter assays
HEK293T cells were seeded at 1 ϫ 10 4 cells in 96-well plates in DMEM and 10% FBS 24 h prior to transfection. For each well, 20 ng of dual SRE-luciferase reporter plasmid and 2 ng of GPR56 (WT) or GPR56 H381S (mutant) were transfected using Lipofectamine 2000 (Invitrogen). Six hours after transfection, cells underwent 16 h of serum starvation prior to treatment with various agonists. Cells were lysed and analyzed using the DualGlo luciferase assay system (E2920, Promega, Madison, WI) according to the manufacturer's protocol. Luminescence was read using a Synergy-2 microplate reader (BioTek, Winooski, VT). All firefly luciferase data were normalized to the Renilla luciferase signal and expressed as -fold increase over the signal obtained from cells transfected with WT GPR56.

Dynamic light scattering assay
3-␣-DOG dilutions were delivered from concentrated DMSO stocks into filtered 50 mM potassium P i buffer (pH 7.0), at room temperature to reach a final concentration of 1% DMSO. Measurements were made in triplicate for each concentration using a DynaPro Plate Reader II (Wyatt Technology) with a 75-milliwatt laser at a wavelength of 826.6 nm. Normalized scattercounts per second were obtained to measure critical aggregation concentration.

Statistical analysis
All data were analyzed using GraphPad Prism8 (San Diego, CA). Results are expressed as means Ϯ S.D. An unpaired, twosided Student's t test was used to compare two groups. p values of less than 0.05 were considered to indicate statistical significance. One-way ANOVA followed by Tukey's post hoc test was used to analyze multiple treatment condition experiments. Sample size was not predetermined by statistical methods but based on similar studies in the field.