Optimization of a peptide ligand for the adhesion GPCR ADGRG2 provides a potent tool to explore receptor biology

The adhesion GPCR ADGRG2, also known as GPR64, is a critical regulator of male fertility that maintains ion/pH homeostasis and CFTR coupling. The molecular basis of ADGRG2 function is poorly understood, in part because no endogenous ligands for ADGRG2 have been reported, thus limiting the tools available to interrogate ADGRG2 activity. It has been shown that ADGRG2 can be activated by a peptide, termed p15, derived from its own N-terminal region known as the Stachel sequence. However, the low affinity of p15 limits its utility for ADGRG2 characterization. In the current study, we used alanine scanning mutagenesis to examine the critical residues responsible for p15-induced ADGRG2 activity. We next designed systematic strategies to optimize the peptide agonist of ADGRG2, using natural and unnatural amino acid substitutions. We obtained an optimized ADGRG2 Stachel peptide T1V/F3Phe(4-Me) (VPM-p15) that activated ADGRG2 with significantly improved (>2 orders of magnitude) affinity. We then characterized the residues in ADGRG2 that were important for ADGRG2 activation in response to VPM-p15 engagement, finding that the toggle switch W6.53 and residues of the ECL2 region of ADGRG2 are key determinants for VPM-p15 interactions and VPM-p15-induced Gs or arrestin signaling. Our study not only provides a useful tool to investigate the function of ADGRG2 but also offers new insights to guide further optimization of Stachel peptides to activate adhesion GPCR members.

The adhesion GPCR ADGRG2, also known as GPR64, is a critical regulator of male fertility that maintains ion/pH homeostasis and CFTR coupling. The molecular basis of ADGRG2 function is poorly understood, in part because no endogenous ligands for ADGRG2 have been reported, thus limiting the tools available to interrogate ADGRG2 activity. It has been shown that ADGRG2 can be activated by a peptide, termed p15, derived from its own N-terminal region known as the Stachel sequence. However, the low affinity of p15 limits its utility for ADGRG2 characterization. In the current study, we used alanine scanning mutagenesis to examine the critical residues responsible for p15-induced ADGRG2 activity. We next designed systematic strategies to optimize the peptide agonist of ADGRG2, using natural and unnatural amino acid substitutions. We obtained an optimized ADGRG2 Stachel peptide T1V/F3Phe(4-Me) (VPM-p15) that activated ADGRG2 with significantly improved (>2 orders of magnitude) affinity. We then characterized the residues in ADGRG2 that were important for ADGRG2 activation in response to VPM-p15 engagement, finding that the toggle switch W 6.53 and residues of the ECL2 region of ADGRG2 are key determinants for VPM-p15 interactions and VPM-p15-induced Gs or arrestin signaling. Our study not only provides a useful tool to investigate the function of ADGRG2 but also offers new insights to guide further optimization of Stachel peptides to activate adhesion GPCR members.
There are several proposed activation mechanisms of aGPCRs. One of these mechanisms is the "tethered agonist" model (13,21). In this model, the GAIN domain autoproteolysis exposes a tethered cryptic agonist sequence, designated as Stachel (German for stinger), within the N-terminal region of an aGPCR member, which induces receptor activation through its interaction with the 7TM domain. The tethered Stachel sequence located at the CTF plays an important role in aGPCR activation in response to different physiological stimulations, such as ligand binding, mechanostress, or removing of the NTF (13,(18)(19)(20). Furthermore, synthetic peptides mimicking the Stachel sequence are called Stachel peptides for each aGPCR receptor and have been shown to activate their corresponding aGPCRs (13,21).
Adhesion G protein-coupled receptor G2(ADGRG2), also known as G protein-coupled receptor 64 (GPR64) or human epididymal gene product 6 (HE6), has attracted substantial attention for its specific expression and essential functions in the male reproductive system (22). Recent studies have found that its deficiency results in the dysfunction of fluid reabsorption and male infertility (5). Like most aGPCRs, ADGRG2 is still regarded as an orphan receptor with no identified endogenous ligands (11). In vitro, the overexpression of This article contains supporting information. ‡ These authors contributed equally. * For correspondence: Jin-Peng Sun, sunjinpeng@sdu.edu.cn; Xiao Yu, yuxiao@sdu.edu.cn.
ADGRG2 caused constitutive Gs-and Gq-coupling activity, and a stronger effect was observed with the CTF truncation mutant ADGRG2 (ADGRG2-β) (5). In our recent studies, we have found that ADGRG2 is specifically coupled to the cystic fibrosis transmembrane conductance regulator (CFTR) through a Gq-and β-arrestin1-mediated mechanism, regulating the ion/water homeostasis in efferent ductules, which is essential for male fertility. It has been shown that ADGRG2 could be activated by the Stachel peptide p15 (sequence: TSFGILLDLSRTSLP) derived from the Stachel sequence (23), to elicit downstream Gs, Gq, G12/13 signaling (Fig. 1A) (23)(24)(25). p15 does not activate or control other aGPCRs (GPR110, GPR133), thereby displaying apparent specificities for the receptor it originated from (23). In contrast to other peptidergic GPCRs, Stachel sequencederived synthetic peptides have a significantly low affinity to the 7TM (13,26). For example, the Stachel peptide p15 has a significantly low affinity toward ADGRG2 (23). The low affinity of most Stachel peptides limits their in vivo studies. Chemical modification of the Stachel peptides may help to improve their affinity. Meanwhile, it remains to be revealed how the key residues in the 7TM domain sense the Stachel peptide binding and activate the receptor. Gaining more understanding into this question may greatly facilitate the design of peptidomics-based agonists for individual aGPCR members, which remains a great challenge in the field. In the current study, we identified important residues that contributed to the Stachel ADGRG2-peptide p15-induced ADGRG2 activation. According to our findings, we designed a systematic strategy to optimize the peptide-derived ADGRG2 agonist. We then obtained an optimized ADGRG2 Stachel peptide T1V/F3Phe(4-Me) (VPM-p15) that activated ADGRG2 with significantly improved affinity. Furthermore, we characterized the residues in ADGRG2 that are important for ADGRG2 activation in response to VPM-p15 engagement.

Results
The aGPCR ADGRG2 is activated by the Stachel peptide p15 First, it is necessary to confirm the activation of Stachel peptide p15 on ADGRG2. We generated an ADGRG2 mutant in which the entire ECD (including the entire GPS motif) was removed (ADGRG2-ΔGPS-β). Then, we measured the concentration-dependent p15 activities toward ADGRG2-ΔGPS-β and full-length ADGRG2(ADGRG2-FL) receptor. Increasing the amount of p15 led to a significant increase in cAMP levels, which was determined by the GloSensor assay (Fig. S1A). p15 also elicited a rise in the intracellular calcium level, which was determined using a CalfluxVTN Ca 2+ assay. This change could be inhibited by the Gq/11 selective inhibitor YM-254890, suggesting that p15 induced the calcium influx via a Gq-dependent pathway (Fig. S1B). Of interest, we found through the bioluminescence resonance energy transfer (BRET) assay that p15 promoted the recruitment of β-arrestin1 and β-arrestin2 to ADGRG2, respectively (Fig. S1, C-D). It is notable that p15 induced significantly more recruitment of β-arrestin1 or β-arrestin2 through the truncated mutant compared with the wildtype ADGRG2. Moreover, the ADGRG2-β (CTF fragment) shows a significantly higher constitutive activity for arrestin recruitment compared with that of the wildtype ADGRG2 (Fig. S1, E-F). Because the Stachel sequence of ADGRG2-β could have more chance to access to the 7TM domain compared with that of the Stachel sequence in the ADGRG2 full length, these results indicate that the arrestin recruitment to ADGRG2 might be strengthened by the activation of ADGRG2 via Stachel sequence interaction. Taken together, these data demonstrate that the Stachel peptide p15 activates multiple signaling pathways downstream of ADGRG2.
Alanine scanning within the p15 peptide identified critical amino acids for activity Among the three Stachel peptides of ADGRG2 (7, 13, 15 amino acids in length), p15 induced the strongest ADGRG2 activation (Fig. 1B). To find out which amino acid (aa) residues of p15 are required for receptor activation, we performed a systematic alanine scanning. The results showed that mutations of T1A and T12A enhanced p15 activity. Conversely, mutations of any residues from the middle of LLDL at positions ranging from +6 to +9, F3A, or R11A all greatly impaired p15 activity (Fig. 1C, Fig. S2 and Table S1).

Optimization of Stachel peptide p15
We next optimized p15 by amino acid substitution, according to the following strategies: (1) diversifying the noncritical residues identified in the alanine scanning to amino acids with different properties; (2) changing critical amino acids into natural or unnatural amino acids with different structures but similar properties (hydrophobicity, polarity, or charge); (3) enhancing the hydrophobicity of the C terminus of the Stachel peptide ( Fig. 2A).

Expression of ADGRG2 and its mutants on the cell surface
To provide a foundation for further optimization of the ligand that activates ADGRG2, we then investigated the molecular mechanism of the engagement of VPM-p15 with ADGRG2. Through sequence alignment, we found that ADGRG2 Stachel sequences of the region p15 are highly conserved among different species (Fig. S4A). Because adhesion GPCRs are most homologous to class B GPCRs, we therefore used the PTH1R (a class B GPCR) and a Swiss-model software to generate an ADGRG2 three-dimensional coordinate model (Fig. 4A). According to this model, we then selected ADGRG2 mutations in extracellular loops 1, 2, 3, and transmembrane helices and then mutated them to alanine by QuikChange (Fig. 4, A-B and Fig. S4A). We transfected HEK293 cells with ADGRG2-ΔGPS-β WT or its corresponding mutants and detected the expression of receptors on the cell surface by cell-surface ELISA (Fig. S4B). We adjusted the amount of transfected plasmid and used the mutants without significant alterations of receptor-surface expression for further delineation of the ADGRG2-VPM-p15 engagement.

Identification of important ADGRG2 residues mediating VPM-p15 engagement
The binding capacities of ADGRG2 WT or its mutants for VPM-p15 were determined by BRET assay, which has been used in previous studies (27)(28)(29). We fused the N terminus of ADGRG2 with a Rluc fragment and coupled FITC to the C terminus of VPM-p15 to facilitate BRET measurement. We stimulated the HEK293 cells transfected with Rluc-ADGRG2-ΔGPS-β wildtype (WT) or mutants by increasing the concentration of FITC-conjugated VPM-p15, and their binding abilities were determined by fitting the data with the saturation curves.
The concentration-response curves of the binding abilities of the receptors to VPM-p15 are shown in the supplemental materials ( Fig. S5 and Table S3). Of importance, mutations of two residues in ECL2, including W757 ECL2 and I758 ECL2 , as well as the mutations of W824 6.53 in TM6 (superscripts are based on Wooten's numbering system for class B family GPCRs (30)), almost abolished the binding of VPM-p15 to ADGRG2 (Fig. 4C and Table S3). The binding results indicate that the TM6 and ECL2 regions are important for mediating the binding of VPM-p15 to ADGRG2.
Both p15 and VPM-p15 were able to induce β-arrestin recruitment via ADGRG2. We therefore randomly picked several mutants that did not affect ligand binding and examined their effects on arrestin recruitment. Interestingly, two mutations, F764A and Y765A, showed improved activities toward both G protein and arrestin. In contrast, the mutations of V768A and F772A impaired both G protein and arrestin activities ( Fig. 4D and Fig. S6, A-B). These results indicate that G protein and arrestin activity are mediated by at least several common residues located in the ADGRG2 7TM region.
The obstacle in working with Stachel peptides is that they require large amounts to elicit significant activation levels in vitro with EC 50 values ranging from 80 to 400 μM in general (26). The reason for this low-affinity interaction can be explained by the physiologic 1:1 stoichiometry of the Stachel agonist and the 7TM interface, which makes high-affinity agonists dispensable. At present, the mechanism underlying the interaction between aGPCR and Stachel ligands remains to be solved, and a systematic way to acquire a high-affinity Stachel ligand for aGPCR member is in great need to facilitate further aGPCR studies. Here, by performing the alanine scanning of p15, which has been known to activate ADGRG2, we identified the key residues, including F 3 , L 6 LDL 9 , and R 11 , that are important for p15-induced ADGRG2 activity. According to this achievement, we then performed systematic modifications to improve peptide affinity. Among the three strategies that we have exploited, we found two that significantly increased peptide activity. These strategies include (1) diversifying noncritical residues to natural or unnatural amino acids with different properties and (2) changing critical amino acids into natural or unnatural amino acids with similar properties with different structures (Fig. 5). These findings may provide valuable guidance for future peptide agonists optimization toward other aGPCR members.
Currently, no atomic-resolution structures of any aGPCRs have been determined, which brings great challenges to agonist development toward individual aGPCR members. We therefore performed alanine scanning of a group of residues within the upper part of the 7TM domain of ADGRG2 to examine its engagement with VPM-p15. The results suggest that the extracellular loop 2 and the conserved toggle switch W 6.53 are essential for VPM-p15 binding (Figs. 4 and 5). These clues may also help further peptide agonist design for ADGRG2.
Future drug development for these aGPCRs will benefit greatly from further understanding of the mechanism of aGPCR activation in response to Stachel peptides, by using structural, biochemistry, and biophysics methods. Considering the number of human diseases associated with aGPCR mutations and interesting phenotypes observed after aGPCR gene deletion (1,37), there are compelling reasons to argue that the optimization of Stachel peptides for aGPCR activation will contribute to the development of new therapies for human diseases.

Peptide synthesis and solubilization
ADGRG2 Stachel peptides were synthesized using standard Fmoc-chemistry on an automated peptide synthesizer Multi-Pep (Intavis AG). Final side chain deprotection and cleavage Optimization of ADGRG2 peptide agonist Figure 4. Key residues of ADGRG2 for VPM-p15 binding and signal transduction. A, a cartoon presentation of ADGRG2-ΔGPS-β highlighting the existence of the possible interactions between the VPM-p15 ligand and the binding site. The ADGRG2 structure was modeled by using PTH1R (Protein Data Bank: 6NBI) as a template. The extracellular loops are colored yellow and the ligand-binding residues are shown as side chain types and colored pink. B, a schematic serpentine representation of the ADGRG2 7TM domain residues highlighting its mutation sites. Extracellular and intracellular loops (ECL and ICL) are indicated (B). C, binding capacities of ADGRG2-ΔGPS-β WT or its mutants for VPM-p15 monitored by BRET experiments. HEK293 cells were transfected with Rluc-ADGRG2-ΔGPS-β or mutants. Transfected cells were stimulated by increasing the concentration of FITC-VPM-p15. Binding capacities were from the solid support was achieved by using TFA, water, and thioanisole (95:2.5:2.5vol%). Peptides were subsequently purified to >95% purity by preparative RP-HPLC (Shimadzu LC-8). Peptides were suspended in dimethyl sulfoxide and diluted to ≤4% dimethyl sulfoxide (vol/vol) in experiments.

GloSensor cAMP assay
The Glosensor cAMP assay was performed as described (6,39,40). HEK293 cells were transfected with the GloSensor plasmid and the desired expression plasmids in 24-well dishes. Twenty-four hours after transfection, the cells were plated on 96-well plates at a cell density of 20,000 cells/well. The cells were maintained in Dulbecco's modified Eagle's medium for another 24 h, washed with PBS, and then incubated with 100 μl of solution containing 10% fetal bovine serum, 2% (v/v) Glo-Sensor cAMP reagent, and 88% CO 2 -independent medium in each well for 2 h. The cAMP signal was examined using a luminescence counter (Mithras LB 940).

CalfluxVTN Ca 2+ assay
For the Ca 2+ assays, HEK293 cells in 24-well dishes were cotransfected with plasmids encoding ADGRG2 or its mutants, CalfluxVTN constructs were used at a 6:1 ratio. Empty vector pcDNA3.1(+) was used to normalize the total amount of transfected plasmid DNA. The transfection and CalfluxVTN Ca 2+ assays were performed as previously reported (28,29,41) with minor modifications. In brief, 24 h post transfection, cells were harvested and distributed in 96-well flat-bottomed white microplates. After another 24 h, the cells were processed accordingly and then were added with the Nluc substrate furimazine (5 μM). The BRET signal was calculated by measuring the ratio of the light emitted by the Venus reporter (535 nm) relative to light emitted by the Nluc reporter (475 nm). The average baseline value (basal BRET ratio) recorded prior to agonist stimulation was subtracted from the experimental BRET signal values to obtain the ΔBRET ratio.

Bioluminescence resonance energy transfer assay
The ADGRG2-YFP and β-arrestin1-Rluc or β-arrestin2-Rluc plasmids were transiently transfected into HEK293 cells. Forty-eight hours after transfection, the cells were washed with PBS, detached with PBS and 5 mM EDTA, and resuspended in PBS with 0.1% (w/v) glucose at room temperature. The cells were then distributed (80 μg of protein per well) in a 96-well microplate (Corning Inc, Corning, NY, USA) and incubated in the presence or absence of ADGRG2 determined by BRET. Kd values of ADGRG2-ΔGPS-β WT and its mutants for binding VPM-p15 were calculated by GraphPad. n.s. p > 0.05, *p < 0.05, **p < 0.01, Binding capacities of ADGRG2-ΔGPS-β mutants were compared with ADGRG2-ΔGPS-β WT. D, effects of ADGRG2-ΔGPS-β mutants on VPM-p15-induced cAMP accumulation. HEK293 cells transfected with ADGRG2-ΔGPS-β or its mutants were stimulated by 100 μM VPM-p15. cAMP levels were detected by the Glosensor assay. Data were normalized by paralleling experiments with ADGRG2-ΔGPS-β WT. n.s. p > 0.05; *p < 0.05; ***p < 0.001; ADGRG2-ΔGPS-β mutants were compared with ADGRG2-ΔGPS-β WT. C-D, each experiment was repeated six times. Figure 5. Schematic diagram depicting the optimization strategies of Stachel peptides and the activation of ADGRG2. ADGRG2 could be activated by the Stachel peptide p15 derived from the Stachel sequence. However, the Stachel peptide p15 has a significantly low affinity toward ADGRG2. We then designed systematic strategies to optimize the peptide agonist of ADGRG2, using natural or unnatural amino acid substitutions. Subsequently, we obtained an optimized ADGRG2 Stachel peptide VPM-p15 that activates ADGRG2 more potently compared with the wildtype p15 peptide. Then, the binding results indicate that TM6 and ECL2 regions are important positions to constitute ADGRG2 ligand binding, which recognizes VPM-p15.

Optimization of ADGRG2 peptide agonist
Stachel peptides for 1 min. BRET between Rluc and YFP was measured after the addition of the Rluc substrate coelenterazine 400a (5 μM, Interchim) under a Thermo plate reader. The BRET signal was calculated as the ratio of emission of YFP (527 nm) to Rluc (370-480 nm).
In the detection of binding capacities of ADGRG2 WT or its mutants for VPM-p15, HEK293 cells were transfected with Rluc-ADGRG2-ΔGPS-β or mutants. Transfected cells were stimulated by increasing the concentration of FITCconjugated VPM-p15. Binding capacities were determined by BRET, EC 50 values of ADGRG2 WT and its mutants for binding VPM-p15 were calculated by GraphPad, and data were normalized by EC 50 of WT.

Cell-surface ELISA assay
After transfection for 24 h, HEK293 cells were cultured at 37 C in 24-well plates pretreated with polylysine. After another 24-h incubation, the cells were washed with Tris Buffer Saline Tween20 (TBST) once and fixed with 4% paraformaldehyde at room temperature for 10 min. Then the cells were washed three times with TBST and sealed with 5% bovine serum albumin solution for 1 h at room temperature. After three washes with TBST, the cells were incubated with primary antibody overnight at 4 C. After another three washes with TBST, secondary antibody was added, and the cells were incubated for 1 h at room temperature. Tetramethylbenzidine was added after washing three times with TBST. After 5 to 20 min, the reaction was stopped with 0.25 M HCl solution until the color of the solution changed appropriately to blue. Then the reaction solution was transferred to a 96-well plate, and the absorbance at OD450 was measured with a microplate reader.

Statistics
All the data are presented as the mean ± SEM from at least three independent experiments. Statistical comparisons were performed using an ANOVA with GraphPad Prism 6.0 (GraphPad Software, San Diego, CA, USA). Significant differences were accepted at p < 0.05. The sequence alignments were performed using T-coffee.

Data availability
The structure presented in this paper has been deposited in the Protein Data Bank with the following code: 6NBI. All remaining data are contained within the article.