Generating thermostabilized agonist-bound GPR40/FFAR1 using virus-like particles and a label-free binding assay

Abstract

Elucidating the detailed mechanism of activation of membrane protein receptors and their ligand binding is essential for structure-based drug design. Membrane protein crystal structure analysis successfully aids in understanding these fundamental molecular interactions. However, protein crystal structure analysis of the G-protein-coupled receptor (GPCR) remains challenging, even for the class of GPCRs which have been included in the majority of structure analysis reports among membrane proteins, due to the substantial instability of these receptors when extracted from lipid bilayer membranes. It is known that increased thermostability tends to decrease conformational flexibility, which contributes to the generation of diffraction quality crystals. However, this is still not straightforward, and significant effort is required to identify thermostabilized mutants that are optimal for crystallography. To address this issue, a versatile screening platform based on a label-free ligand binding assay combined with transient overexpression in virus-like particles was developed. This platform was used to generate thermostabilized GPR40 [also known as free fatty acid receptor 1 (FFAR1)] for fasiglifam (TAK-875). This demonstrated that the thermostabilized mutant GPR40 (L42A/F88A/G103A/Y202F) was successfully used for crystal structure analysis.

• GPR40/FFAR1
• size exclusion chromatography/liquid chromatography mass spectroscopy-based binding assay
• thermostabilized mutant
• virus-like particle

Introduction

Membrane proteins are important drug targets due to their wide variety of biogenic roles. One of the largest membrane protein superfamilies is the G-protein-coupled receptors (GPCRs), which are the targets of approximately 25% of all known drugs in clinical use (Lagerstrom & Schioth, 2008). In recent years, structural insights into GPCRs that have revealed the mechanism of receptor activation and its ligand binding have successfully been used for structure-based drug discovery (Congreve et al., 2012; Kolb et al., 2009; Venkatakrishnan et al., 2013). GPCRs generally become unstable when extracted from the lipid bilayer membrane by detergent. In addition, maintaining the activity of the purified proteins is difficult because of their flexibility and poor stability when solubilized from the cell membrane. One of the main characteristics of GPCR proteins is the equilibrium between the active and inactive states (Ulrik et al., 1997). These characteristics of GPCRs make X-ray crystallographic studies challenging compared with soluble proteins. Nevertheless, determination of the crystal structure of membrane proteins such as drug-targeted GPCRs has become feasible through the use of recent protein engineering techniques (Venkatakrishnan et al., 2013). Increasing their thermostability by introducing point mutations is one such technique, which has substantially contributed to the elucidation of several GPCR protein structures (Lebon et al., 2011; Magnani et al., 2008; Miller & Tate, 2011; Serrano-Vega et al., 2008; Shibata et al., 2009; Tate, 2012). It is known that increased thermostability of GPCRs tends to decrease their flexibility (Bertheleme et al., 2013; Ulrik et al., 1997). Nevertheless, protein crystal structure analysis using thermostabilized mutant GPCRs has improved our understanding of structural aspects of mechanisms of GPCR activation and ligand binding modes (Congreve et al., 2011a, b, Dore et al., 2011; Robertson et al., 2011). As the methods for selecting thermostabilized mutants for crystallographic studies reported so far require solubilized membrane protein samples, universal application of this technique to GPCRs is often hindered by a low expression level and protein instability of the wild-type. Existing methods generally require a radiolabeled ligand binding assay, which is a major hurdle for many GPCRs due to the unavailability of high-affinity radiolabeled ligands for the receptor. These limitations can be resolved by combining a size-exclusion chromatography/liquid chromatography mass spectroscopy (SEC/LC-MS)-based binding assay with transient overexpression in mammalian virus-like particles (VLP) (Muckenschnabel et al., 2004; Zehender et al., 2004; zur Megede et al., 2000). A major advantage of this approach is that the VLP system does not require solubilization of the membrane fraction. Furthermore, the expression of the target membrane protein is higher than that of proteins prepared from the membrane fraction. The SEC/LC-MS-based binding assay is a well-established technique that reveals the interaction between the ligand and the protein of interest. This technique has potential for broader use mainly because it does not require a radiolabeled ligand. Here, we report that this approach can produce a thermostabilized GPCR that is suitable for crystal structure analysis.

As a case study, we examined G-protein-coupled receptor 40 [GPR40; also known as free fatty acid receptor 1 (FFAR1)]. GPR40 is activated by medium and long chain free fatty acids and mediates free fatty acid (FFA)-induced glucose-dependent insulin secretion from pancreatic β cells (Itoh et al., 2003). A GPR40-selective agonist, [(3S)-6-({2′,6′-dimethyl-4′-[3-(methylsulfonyl)propoxy]biphenyl-3-ylethoxy)-2,3-dihydro-1-benzofuran-3-yl] acetic acid hemihydrate (fasiglifam, development code: TAK-875), was demonstrated to have robust glucose-lowering effects in patients with type 2 diabetes mellitus (Araki et al., 2012; Burant et al., 2012; Leifke et al., 2012; Naik et al., 2012; Tsujihata et al., 2011; Yashiro et al., 2012). Recent research has indicated that fasiglifam acts as an ago-allosteric modulator of GPR40 (Yabuki et al., 2013). However, the binding mode of fasiglifam to GPR40 is not fully understood. For better understanding of its binding mode by crystal structure analysis, thermostabilized mutant GPR40 constructs were generated using the SEC/LC-MS-based binding assay combined with a mammalian VLP system. We also determined agonistic activities and affinities of GPR40 ligands compared with selected thermostabilized mutants to explore the relationship between ligand binding affinity and receptor function.

Methods

Materials

Fasiglifam was synthesized at Takeda Pharmaceutical Company Limited (Kanagawa, Japan). Docosahexaenoic acid (DHA) was purchased from Tocris Bioscience (Bristol, UK).

Mutagenesis

GPR40 was amplified from genomic DNA using a standard polymerase chain reaction (PCR) method and cloned into pcDNA3.1. The human GPR40 expression vector pcDNA3.1/hGPR40 was used as a template for in vitro site-directed mutagenesis. Point mutant vectors were created using an In-Fusion® Advantage PCR Cloning kit (Clontech Laboratories) (Zhu et al., 2007). First, pcDNA3.1 was linearized with the restriction enzymes Nhe1 (Takara Bio Inc.) and Pme1 (BioLabs Inc.). Second, to generate the mutation, DNA fragments were generated by PCR using pcDNA3.1/GPR40 as the template. Mutagenic sense and antisense primers were designed as 5′-6mer-XXX-20mer-3′, where XXX represents the amino acid codon of the desired mutation. The primer pairs overlapped by 15 bp at their 5′ ends. A sense primer overlapping with the Nhe1 restriction site was designed as 5′-ACCCAAGCTGGCTAGATGGACCTGCCCCCGCAGCT-3′, and an antisense primer overlapping with the Pme1 restriction site was designed as 5′-CTGATCAGCGGGTTTTTACTTCTGGGACTTGCCCCCTTGCGT-3′. Pairs of DNA fragments – a DNA fragment from the sense primer with mutagenic antisense primer, and a DNA fragment from the antisense primer with mutagenic sense primer – were generated by PCR. Finally, the generated DNA fragments pairs were joined with linearized pcDNA3.1 by In-Fusion. In-Fusion samples were transformed into ECOS™ competent E. coli JM109 (Nippon Gene), and full-ORF sequences of individual clones were verified using a Genetic Analyzer (Life Technologies). Residues predicted as N-terminal, C-terminal, and intracellular loop regions were excluded from the mutagenesis study for the below reasons. First, these regions were used to fuse proteins, such as T-4 lysozyme and BRIL and to increase protein hydrophilicity and stability for crystal structure analysis (Cherezov et al., 2007; Chun et al., 2012). Second, based on previous literature, the mutation sites effective in increasing GPCR thermostability may be difficult to find in these regions (Tate, 2012). A total of 265 mutated GPR40 expression vectors were prepared by changing amino acid residues to alanine; ordinal alanine was changed to valine. In addition, tyrosine was changed to alanine and phenylalanine based on in-house mutagenesis data from the compound binding assay.

Protein expression

FreeStyle293 (Life Technologies) cells were used to express wild-type and mutant GPR40 constructs in VLPs. FreeStyle293 cells were maintained in culture in FreeStyle293 expression medium (Life Technologies) with 1 mgml⁻¹ of gentamycin in an 8% CO₂ incubator with shaking at 37 °C and were passaged twice weekly. The GPR40 expression vector and HIV-1 GAG protein expression vector were cotransfected into FreeStyle293 cells using NeoFectin reagent (Astec) according to the manufacturer’s instructions. The cells were incubated in an 8% CO₂ incubator with shaking at 37 °C; 48 h after transfection, cells were centrifuged at 500 g for 5 min, and supernatants were pooled. The supernatants containing GPR40-expressing VLPs were centrifuged at 32 000 g for 60 min. The supernatant was discarded, and the pellet was resuspended in dilution buffer [50 mM Tris–HCl (pH 7.5), 5 mM EDTA, and Complete Protease Inhibitor Cocktail (Roche Diagnostics)]. Diluted samples were stored at −80 °C. Using this protocol, we obtained about 9 mg of GPR40-expressing VLPs from 2.5 × 10⁸ cells. To prepare GPR40 expressed from the cell membrane, the GPR40 expression vector was transfected into FreeStyle293 cells and cultured under the same conditions as those used for VLP preparation. Cultured cells were centrifuged at 500 g for 5 min, and the supernatant was discarded. The cell pellet was resuspended in distilled water with Complete Protease Inhibitor Cocktail and centrifuged at 32 000 g for 60 min to remove cytoplasmic proteins. After the supernatant was discarded, the pellet was resuspended in dilution buffer and homogenized on ice. The homogenized sample was centrifuged at 6000 g for 30 min, and a supernatant sample was collected as the cytoplasmic membrane sample.

Saturation binding experiments using the SEC/LC-MS-based binding assay

The binding affinities of fasiglifam for GPR40 expressed by VLPs and cytoplasmic membrane samples were measured by the SEC/LC-MS-based binding assay. VLP samples were diluted to 3 μg ml⁻¹, and membrane samples were diluted to 50 μg ml⁻¹ with the binding assay buffer [50 mM Tris–HCl (pH 7.5), 5 mM MgCl₂, 5 mM EDTA, and 0.005% Tween 20]. Diluted samples were dispensed into a 96-well plate. Fasiglifam in DMSO was added to the wells at final concentrations of 250, 100, 40, 16, 6.4, 2.5, 1.0, and 0.4 nM, and the plate was incubated at 4 °C. Unbound compound in 25 μl of the reaction sample was separated by SEC based on a 96-well plate (Zehender et al., 2004). Subsequently, 50 μl of 70% acetonitrile/0.2% formic acid was added to each well to dissociate the bound compound from the receptor protein. The dissociated ligand was applied to a C18, 3.0 mm i.d. × 20 mm column (Intakt) by ultrafast liquid chromatography (Shimadzu), followed by identification and quantification with an API5000 electrospray-ionization mass spectrometer (AB SCIEX). The apparent K_d () value and B_max of fasiglifam were calculated using the one site-specific binding curve equation of Prism 5.03 software (GraphPad Software). Data points are from at least quadruplicate measurements.

Thermostabilized mutant screening

Wild-type and mutant GPR40-expressing VLPs were diluted with assay buffer [10 mM Tris–HCl (pH 7.5), 5 mM EDTA, 5 mM MgCl₂, 0.005% Tween 20] to 20 μg ml⁻¹ in the presence of 100 nM fasiglifam (assay format with fasiglifam, hereafter called format A) or in the absence of fasiglifam (assay format without fasiglifam, hereafter called format B). After incubation at 4 °C for 30 min, samples were again incubated at three different temperatures for 30 min. Format A samples were incubated at 4, 41, and 55 °C and format B samples were incubated at 4, 38, and 55 °C. After the heating process, samples were immediately placed on ice for 5 min. Fasiglifam (100 nM) was added to the format B samples, which were further incubated at 4 °C for 30 min. Receptor activity remaining after heat treatment was assessed by measuring the amount of receptor-bound fasiglifam using the SEC/LC-MS-based binding assay. Before thermostabilized mutant screening, we determined the apparent T_m () value of wild-type GPR40 in formats A and B (details of the method are given in the next section). Then, the screening temperature was set based on the each apparent T_m value.

Apparent T_m value measurements

The value (°C) of the receptor was determined by measuring the amount of receptor-bound fasiglifam using the SEC/LC-MS-based binding assay with eight-point heating temperature under format A and B conditions. The value was calculated by nonlinear regression analysis using the dose–response curve equation in Prism 5.03 software (Magnani et al., 2008).

Ca²⁺ flux assay (FLIPR)

Human Embryonic Kidney 293T (HEK293T) cells were seeded in a 25 cm² flask at 1.5 × 10⁶ cells/flask and incubated overnight in 5% CO₂ at 37 °C. The next day, GPR40 expression vector was transfected into HEK293T cells using Fugene® HD Reagent (Promega) according to the manufacturer’s instructions. The cells were seeded 1 day after transfection and assayed the next day. Cells were seeded onto poly-D-lysine-coated 384-well plates (BD) at 15 000 cells/well and incubated overnight in 5% CO₂ at 37 °C. Cells were incubated in loading buffer [Hanks’ buffered saline solution supplemented with 20 mM HEPES (pH 7.5), 2.5 μg ml⁻¹ fluorescent calcium indicator Fluo 4-AM (Dojindo Laboratories), 2.5 mM probenecid (Dojindo Laboratories) and 0.01% fatty acid-free BSA (Sigma Aldrich)] for 60 min at room temperature. Various concentrations of fasiglifam or DHA were added to cells, and the increase in intracellular Ca²⁺ concentration was monitored using the FLIPR Tetra system (Molecular Devices) for 90 s. Agonist-stimulated responses in wild-type GPR40, thermostabilized mutant GPR40, and pcDNA3.1 vector-transfected cells as a negative control were determined as the maximum value minus the minimum value after subtracting the baseline response. The dose-response relationship was analyzed by fitting to sigmoid dose-response curves. The agonist EC₅₀ values were calculated using Prism 5.03 software. Data points are from quadruplicate measurements.

Fluorescence-activated cell sorting (FACS) analysis

HEK293T cells were seeded in six-well plates at 6.0 × 10⁵ cells/well and incubated overnight in 5% CO₂ at 37 °C. Next day, GPR40 expression vector was transfected into the cultured cells using Fugene® HD (Promega) reagent according to the manufacturer’s instructions. After 48 h of incubation, cells were harvested with Cell Dissociation Buffer, enzyme-free, PBS-based (Life Technologies). After the PBS wash, cells were diluted to 1.0 × 10⁷ cells/well in 50 μl of FACS assay buffer [TBS (pH 7.4) with 2% FBS]. Allophycocyanin (APC)-conjugated anti-FLAG Surelight antibody (Perkin Elmer) at 10 μg ml⁻¹ was added on ice for 30 min, followed by washing with 1 ml FACS assay buffer and centrifuging at 500 g for 5 min. The supernatant was discarded, and samples were resuspended in 500 μl of FACS assay buffer and filtered prior to analysis of the mean values of APC area signal using FACSAria II (BD). Data points are from triplicate measurements.

Results

Screening of thermostabilized mutant

A total of 265 single-point mutants of GPR40 were expressed in VLPs. The thermostabilities of the mutants were determined under format A and B conditions (Figure 1). In the mutants that exhibited significantly lower ligand binding at 4 °C compared with the wild-type, the reduced binding resulted from a lower expression level or reduced ligand binding affinity. The mutants in which the signal level was >50% lower compared with the wild-type were excluded from the subsequent thermostability analysis because they were not suitable for crystal structure analysis. Thermostabilized mutants were selected based on the percentage of remaining binding signal, which was calculated as follows:

Figure 1. Snake plot of the amino acid sequence of human GPR40 showing the thermostabilized sites. Blue, increase (%) in activity in the fasiglifam-binding condition (G29, T31, L38, C127, A128, A129, W131, A132, L133, V134, E145, G149 and V225); yellow, increase (%) in activity in the compound-free condition (V64, A109, Y122, L135, A199, F200, G204, V237, G265, T286 and V287); red, increase (%) in activity in both conditions (L42, F88, G103, R104, Y202 and K285); gray, not tested or fasiglifam binding signal for the mutant <25% of the signal for the wild-type; *binding signal of fasiglifam not determined for both tyrosine-to-alanine and tyrosine-to-phenylalanine mutants; **binding signal of fasiglifam not determined for tyrosine-to-alanine mutant but determined for tyrosine-to-phenylalanine mutant; ***increase (%) in activity for tyrosine-to-phenylalanine mutant; ****increase (%) in activity for both tyrosine-to-alanine and tyrosine-to-phenylalanine mutants.

The percentage of remaining binding signal in the wild-type was 58% under the format A condition and 53% under the format B condition. These results were obtained by averaging six independent experiments. The percentages of remaining binding signal in the prepared mutants were determined, and thermostabilized single-point mutants were selected using the following criteria: The percentage of activity was greater than +2 SD of the percentage of activity in the wild-type (84% under format A and 74% under format B).

determination for thermostabilized mutants

To confirm the thermostability of selected single-point mutants, values were measured for the selected mutants, and some single-point mutants showed increased thermostability (Table 1). Assuming that combining most stable single-point mutants could further increase thermostability, we conducted multiple rounds of mutation combinations that generated a four-point mutant GPR40 (L42A/F88A/G103A/Y202F). The value of the four-point mutant was increased by approximately 17 °C when compared with that of the wild-type (Figure 2 and Table 2).

Figure 2. Apparent T_m (, °C) of the wild-type and the thermostabilized four-point mutant of GPR40. Binding signals at each temperature were normalized as percentages of the binding signal at 4 °C. This experiment was performed in sextuplet. Data are expressed in means ± SEM and were fitted using the sigmoidal dose-response curve equation of Prism 5.03 software. Closed circles, the wild-type under format A condition; open circles the wild-type under format B condition; closed triangles, thermostabilized four-point mutant under format A condition; open triangles, thermostabilized four-point mutant under format B condition.

Table 1. Apparent T_m (, °C) values of wild-type and single-point thermostabilized mutants of GPR40.

	(Format A)	(Format A)	(Format B)	(Format B)
WT	37.2	–	35.5	–
T31A	38.3	1.1	38.3	2.8
L42A	42.1	4.9	39.8	4.3
V64A	39.6	2.4	37.7	2.3
F88A	44.0	6.8	43.9	8.3
G103A	42.7	5.5	41.2	5.7
R104A	40.9	3.7	39.2	3.7
Y122F	41.1	3.8	39.2	3.7
E145A	42.8	5.6	39.6	4.1
A199V	39.8	2.6	37.6	2.1
Y202A	40.5	3.3	38.9	3.4
Y202F	43.7	6.5	42.7	7.2
V225A	40.4	3.2	38.2	2.7
G265A	41.1	3.9	38.9	3.3

Table 2. values (°C) of wild-type and the thermostabilized four-point mutant of GPR40.

	(°C)
	Format A	Format B
Wild-type	38.3 ± 0.3	36.5 ± 0.3
L42A/F88A/G103A/Y202F	55.7 ± 0.3	55.7 ± 0.4

Each data point was assayed in sextuplet means ± SEM

Binding affinity of fasiglifam for wild-type GPR40 and thermostabilized mutants

The binding affinities of fasiglifam for the wild-type, thermostabilized four-point mutant, FLAG-tagged wild-type, and FLAG-tagged thermostabilized four-point mutant of GPR40 expressed in VLPs and cytoplasmic membrane were determined using the SEC/LC-MS-based binding assay. Total compound binding in the presence of GPR40 was measured, while non-GPR40-expressing VLPs were used to estimate non-specific binding. Specific binding was calculated by subtracting nonspecific binding from total binding. Wild-type GPR40 and the thermostabilized four-point mutant GPR40 had values of 11.5 and 10.9 nM, respectively (Figure 3), which confirmed that insertion of these point mutations into the GPR40 receptor retains the ligand-binding properties of the wild-type receptor. values of fasiglifam and B_max of the GPR40 expressed by VLPs and cytoplasmic membrane are shown in Table 3.

Figure 3. Fasiglifam binding affinity for the wild-type and the thermostabilized four-point mutant of GPR40. Saturation binding curves of fasiglifam for the wild-type (a) and the thermostabilized four-point mutant (b) were determined by the SEC/LC-MS-based binding assay. The specific binding signal for the receptor (open triangles) was calculated as the difference between the total binding signal (solid circles) and the non-specific binding signal (solid squares). This experiment was performed in sextuplet Data are expressed in means ± SEM and were fitted using the one site-specific binding curve equation of Prism 5.03 software.

Table 3. values of fasiglifam and B_max for wild-type and the thermostabilized four-point mutant of GPR40.

Sample form	Construct	(nM)	Bmax (fmol)	Total protein per assay (ng)
VLP	Wild-type*	11.9 ± 1.4	3.6 ± 0.1	75
	L42A/F88A/G103A/Y202F*	11.4 ± 3.1	2.8 ± 0.2	75
	FLAG/wild-type†	8.0 ± 0.8	3.6 ± 0.1	75
	FLAG-L42A/F88A/G103A/Y202F†	11.8 ± 1.2	1.7 ± 0.04	75
Cytoplasmic membrane	Wild-type†	7.7 ± 0.7	2.7 ± 0.1	1250
	L42A/F88A/G103A/Y202F†	8.3 ± 0.6	3.6 ± 0.1	1250

*Each data point was assayed in sextuplet means ± SEM

†Each data point was assayed in quadruplicate means ± SEM

Agonist activity of fasiglifam for wild-type and thermostabilized four-point mutant GPR40 and cell surface receptor expression level

Three (L42A, G103A, and Y202F) of four thermostabilized point mutations were close to the intracellular face of the transmembrane helices, and it is an intriguing question whether these mutations hinder the downstream signaling of GPR40 receptor. The effect of mutations on Ca²⁺ signaling was studied in the presence of both fasiglifam and DHA for FLAG-tagged wild-type and the four-point thermostabilized mutant GPR40 (Figure 4 and Table 4). The thermostabilized mutants appeared to decrease Ca²⁺ signaling while maintaining tight ligand binding. FACS analysis revealed that the thermostabilized four-point mutant had a higher expression level than the wild-type (Figure 5).

Figure 4. Agonistic activity of fasiglifam (a) and DHA (b) in wild-type (circles) and the thermostabilized four-point mutant of GPR40 (squares). The experiment was performed in quadruplicate. Data are expressed in means ± SEM and were fitted using the dose-response equation of Prism 5.03 software.

Figure 5. Transient expression level of wild-type and the thermostabilized four-point mutant GPR40 on the HEK-293T cell surface detected by FACS. The experiment was performed in triplicate. Data are expressed in means ± SEM.

Table 4. Agonistic activity (EC₅₀) of fasiglifam and DHA for the FLAG-tagged wild-type and FLAG-tagged thermostabilized four-point mutant of GPR40.

	log EC50 (M)
	Fasiglifam	DHA
FLAG/wild-type	−7.89 ± 0.11	−6.32 ± 0.08
FLAG/L42A/F88A/G103A/Y202F	−4.69 ± 0.05	−4.70 ± 0.02

Each data point was assayed in quadruplicate means ± SEM

Discussion

We developed a versatile assay platform to obtain thermostabilized GPCRs. We selected a total of 13 single-point mutants showing an increase of >1 °C in thermostability (Table 1). It is noteworthy that several single-point mutants gave a shift of >5 °C. Combining these mutations resulted in a further increase in thermostability. Our protocol provides a high-throughput parallel preparation of thermostabilized mutants because of the miniaturized format, which takes advantage of VLP expression and label-free ligand. GPCR expression in VLPs is superior to expression in the cytoplasmic membrane, particularly in terms of throughput and receptor density per unit of total protein. The use of VLPs transiently expressing the target GPCR streamlines the process by eliminating the membrane solubilization step. Regarding receptor density per unit of total protein, a higher density of VLP expressing GPR40 can be obtained when compared with FreeStyle293 cell membranes (Table 3), resulting in lower non-specific binding signals. Moreover, cell surface receptor expression levels of thermostabilized mutants also increase when compared with those of the wild-type. As large amounts of the receptor protein are required for crystal structure analysis, the VLP expression system may be a good choice for binding assay-based screening of thermostabilized mutants.

The SEC/LC-MS-based binding assay technique also offers the advantage of using label-free compounds to evaluate binding affinity and to identify single-point thermostabilized mutations that can be combined, resulting in a thermostabilized receptor suitable for crystallization. It has been suggested that a unique set of thermostabilized mutants are required for each ligand of interest due to potential conformational changes in the receptor on ligand binding (Magnani et al., 2008). This was confirmed by measurement of thermostability under two conditions and the agonistic activity of the two ligands fasiglifam and DHA.

We determined the thermostability of GPR40 under two conditions: Format A and format B. The thermostability of the active-state receptor was determined under the format A condition. The active state is generally unstable compared with the inactive state (Kobilka & Deupi, 2007). We anticipated different thermostabilized mutants between formats A and B because of the difference in receptor stability between the screening conditions. However, thermostabilized receptors that shared the same mutation sites were selected by screening under conditions A and B (Table 1). The of the wild-type under format A was 3 °C higher than that under format B (Table 2), implying the binding of fasiglifam, known to be an ago-allosteric modulator of GPR40, to the receptor in the intermediate active state conformation of GPR40 and only a small contribution to the stability of GPR40. The FLIPR assay revealed that the Ca²⁺ signaling of fasiglifam decreased >1000-fold for the four-point thermostabilized mutant, with little impact on binding affinity (Tables 3 and 4). On the other hand, Ca²⁺ signaling of DHA, a natural full agonist, decreased only approximately 40-fold. These results indicate that fasiglifam does not alter the conformation of the four-point mutant GPR40 to the active state. Therefore, the four-point thermostabilized mutant GPR40 is suitable for crystal structure analysis with fasiglifam. In our recent research, this four-point mutant in combination with a T4 lysozyme fusion strategy was successfully used to determine the crystal structure of GPR40 bound to fasiglifam (in preparation). However, for crystal structure analysis, it may be necessary to decrease receptor flexibility in the full agonist-bound state. In conclusion, it was demonstrated that this platform can be used to identify thermostabilized mutant GPCRs optimized for particular ligands, and further trials with the other membrane protein targets are under way.

Conclusions

A versatile assay platform that can be used to obtain a thermostabilized mutant GPCR for the purpose of crystal structure analysis was developed. Several thermostabilized single amino acid mutants of GPR40 were identified using this platform. Combining these single-point mutations resulted in a further increase in thermostability in the four-point GPR40 mutant L42A/F88A/G103A/Y202F. The four-point mutant GPR40 had lower Ca²⁺ signaling for fasiglifam, whereas the mutant GPR40 did not show altered binding affinity for fasiglifam. These results indicate that the four-point thermostabilized mutant was suitable for crystal structure analysis of the GPR40/fasiglifam complex.

Declaration of interest

The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.

Acknowledgements

The authors are grateful to Drs G. Kefala, A. Srivastava, J. Yano, and F. Gruswitz for valuable discussions and careful proofreading of the manuscript.