Chemogenetic Approach Using Ni(II) Complex-Agonist Conjugates Allows Selective Activation of Class A G-Protein-Coupled Receptors.

Investigating individual G-protein-coupled receptors (GPCRs) involved in various signaling cascades can unlock a myriad of invaluable physiological findings. One of the promising strategies for addressing the activity of each subtype of receptor is to design chemical turn-on switches on the target receptors. However, valid methods to selectively control class A GPCRs, the largest receptor family encoded in the human genome, remain limited. Here, we describe a novel approach to chemogenetically manipulate activity of engineered class A GPCRs carrying a His4 tag, using metal complex-agonist conjugates (MACs). This manipulation is termed coordination tethering. With the assistance of coordination bonds, MACs showed 10-100-fold lower EC50 values in the engineered receptors, compared with wild-type receptors. Such coordination tethering enabled selective activation of β2-adrenoceptors and muscarinic acetylcholine receptors, without loss of natural receptor responses, in living mammalian cells, including primary cultured astrocytes. Our generalized, modular chemogenetic approach should facilitate more precise control and deeper understanding of individual GPCR signaling pathways in living systems.

Introduction

G-protein-coupled receptors (GPCRs) trigger a myriad of cellular functions, including regulation of nervous and immune systems, by activation of intracellular signaling proteins, such as heterotrimeric G-proteins (Gs, Gi/o, Gq proteins).[1] GPCRs constitute the largest superfamily of membrane receptors and have been among the most valuable drug targets because of their close association with many diseases.[2] Recent biological studies unveiled the higher complexity of GPCR signaling, that is, that each subtype induces different signaling pathways and physiological responses.[3−5] These involve downstream signaling and/or GPCR-interacting proteins, influenced by the characteristic localizations and expression patterns of various GPCRs.[6−8] Also, the huge diversity and structural similarities of GPCR subtypes and their complicated overlapping expression patterns have hampered the deeper understanding of specific GPCR signaling that would be critical for drug design. While conventional pharmacology regulates receptor functions by small organic molecules, insufficient receptor selectivity and cell type specificity of these compounds often prevent interrogation of the biological functions of each GPCR subtype. Illuminating characteristics of individual GPCR subtypes in complex biological environments remains challenging.

A set of toolkits, such as optogenetics[9] and chemogenetics,[10−12] for artificial switching of membrane receptor functions now contribute greatly to analyses of many physiological events, including those involved in animal behaviors. Optogenetics, relying on photoactivatable membrane receptors, is valuable for interrogating neural circuits in vivo.[9,13] For example, OptoXR can activate specific G-protein pathways by artificial opsin-receptor chimeras.[14] The most widely used chemogenetic approach, designer receptors exclusively activated by designer drugs (DREADD), enables regulation of cell signaling through G-protein pathway-selective activation by a bioorthogonal pair of a pharmacologically inert agonist and an artificial GPCR with a genetically engineered ligand binding site.[11,15−19] Though powerful for unraveling biological phenomena at the cellular level of resolution, these methods cannot determine in detail the roles of individual subtypes of receptor proteins in their corresponding cell signaling and/or physiological events.

The design of chemical switches and handles on the target proteins is promising to investigate the protein functions, such as caged proteins, chemical rescues, and allosteric regulation with small molecules.[20−29] Among them, a pioneering strategy for addressing the functions of membrane receptor proteins is optochemical genetics, developed by Trauner et al., which constructs photoresponsive membrane receptors through covalent modification of engineered receptors possessing a genetically introduced Cys with maleimide–azobenzene–ligand conjugates.[30−32] This approach enables control of ligand-gated ion channels (e.g., ionotropic glutamate receptors) and class C GPCRs (e.g., metabotropic glutamate receptors) with large extracellular clamshell-like ligand binding domains. However, its application to class A GPCRs has been limited.[33] Thus, even though a few chemogenetic tools have emerged, these are not yet adequate for controlling cell signaling associated with class A GPCRs, the largest family of GPCRs encoded in the human genome. Tadross and co-workers recently developed drugs acutely restricted by tethering (DART) pharmacology, a chemogenetic method applicable to in vivo study, enabling cell type-specific inactivation of endogenous membrane receptors (AMPA-type glutamate receptor and a muscarinic acetylcholine receptor [mAChR]) through covalent immobilization of antagonists/inverse agonists on a HaloTag protein expressed on the cell surface.[34] Furthermore, Hodson and co-workers reported a novel method, termed RECON, for prolonged, reversible activation of the large number of SNAP-tagged class A and B GPCRs with benzylguanine-linked peptide agonists.[35] These findings are promising, but potential concerns remain regarding subtype selectivity and design flexibility. For an investigation of the complex and ubiquitous class A GPCR signaling, the simple and generalized design of the chemical switches and handles reflecting receptor specificity would be highly desirable.

We herein describe a new strategy for selective activation of an engineered GPCR. We modified a class A GPCR, adding a His tag as a tether by mutagenesis. We also prepared a metal complex–agonist conjugate (MAC) for the corresponding GPCR by organic synthesis. The high affinity of the MAC, with assistance from the coordination bond, allowed selective activation of the target GPCR in living cells, a process termed coordination tethering (Figure ). We demonstrated that two class A GPCRs, the β2-adrenoceptor (β2AR) and mAChR (M1R), were chemogenetically activated in living cells because of the 10–100-fold greater affinity of MAC for the mutated GPCR, relative to that for the wild-type receptor, without loss of natural receptor responses. We also showed that coordination tethering enabled selective activation of the His-tag-fused M1R in primary cultured astrocytes which endogenously express a variety of mAChRs. This method is modular and would be generally applicable to class A GPCRs, thus providing a powerful approach for deciphering characteristics of these receptors, in molecular detail.

Figure 1

Schematic illustration of coordination tethering. A His4 tag was introduced into the N-terminal domain of a class A GPCR as a metal coordination site by genetic engineering. The coordination bond between the His4 tag and a Ni–NTA tethered on a metal complex–agonist conjugate (MAC) enables selective activation of the target GPCR by increasing potency of the MAC.

Results and discussion

Design of His Tag-Fused β2-Adrenoceptors and MACs for Coordination Tethering

We recently reported chemogenetic activation of ionotropic and metabotropic glutamate receptors (ligand-gated ion channel and class C GPCR, respectively), two examples of excitatory neurotransmitter receptors in the central nervous system.[36] Our strategy relied on stabilization of their active conformations by complexation of two genetically incorporated His residues with Pd(bpy) (bpy: 2,2′-bipyridine) in a bidentate coordination bond. This method, though useful, was limited to receptors undergoing a large conformational change of their ligand binding domains during activation. It cannot, however, be applied to class A GPCRs that lack such large domains. The goal of this study was to develop a new approach for artificial modulation of the activities of class A GPCRs, on the basis of a distinct strategy of coordination chemistry. Coordination tethering employs a pair of multi-His sequences (known as a His tag) and a Ni2+–nitrilotriacetic acid (Ni–NTA) complex, a widely used tool for protein purification, to increase agonist affinity.[37,38] The dissociation constant of Ni–NTA and His6 tag was reported to be 14 μM.[39] We envisioned that the coordination bond between the His tag incorporated into a target GPCR and the Ni–NTA tethered in MAC would assist agonist binding. This would evoke selective activation of the engineered GPCR by increasing potency of the modified agonist.

β2AR was the first model protein chosen for proof-of-principle, because its structure and activation mechanisms have been most intensely investigated, by single-crystal and cryoEM structural analyses, NMR, and site-directed mutagenesis.[40,41] These previous findings allowed us to select the N-terminal domain as the insertion site for the His4 tag. This was because other extracellular loops, ECL1, ECL2, and ECL3, were regarded as essential for ligand binding kinetics and specificity of β2AR, as in other class A GPCRs (Figure a).[42] Nevertheless, a lack of structural information for the N-terminal region in β2AR prevented rational design of the engineered β2AR. To obtain an appropriate pair of receptor and MAC, we screened a focused small library consisting of the engineered β2ARs with the His4 tag placed at various positions, along with a few possible MACs. Expression plasmids for the engineered β2ARs were constructed with the His4 tag inserted at four different positions, between G1.1/Q1.2, F1.9/L1.10, P1.19/D1.20, and R1.27/D1.28. The resulting products were β2AR(1.1H4), β2AR(1.9H4), β2AR(1.19H4), and β2AR(1.27H4), respectively, as designated by Ballesteros–Weinstein nomenclature[43] (Figure a, see the Supporting Information for plasmid construction). Additionally, we chemically synthesized three MACs (MAC(βAR, O0), MAC(βAR, O4), and MAC(βAR, O8)) (Figure b,c, see the Supporting Information for details). These had both phenylephrine (PE), an agonist for ARs, and Ni–NTA connected by ethylene glycol linkers of varied lengths (O0, 20 Å; O4, 39 Å; O8, 53 Å; with distances calculated between secondary amine of PE and ternary amine of NTA) (Figure b). As shown in Figure c, this modular molecular design would allow for efficient syntheses of a library of MACs for β2AR or other GPCRs.

Figure 2

Design of His4 tag-fused β2ARs and MACs. (a) Design of His4 tag-fused β2ARs. Four different β2AR mutants with the His4 tag were constructed by genetic engineering. The insertion sites were indicated by red triangles. The full sequence of the N-terminal domain of each mutant is shown in Table S4. (b) Chemical structures of MACs for β2AR. (c) Synthetic scheme of MAC(βAR, O4). MAC(βAR, O4) was prepared by complexation of control-1 and 1 equiv of NiSO4 at room temperature at 30 min, followed by being used for cellular assays without isolation. Since the affinity between the Ni2+ ion and NTA is strong at neutral pH (pKa: 11.54),[44] the free Ni2+ ion is negligible under our experimental conditions. Syntheses of other MACs and their control compounds are shown in the Supporting Information.

Screening Experiment

With the focused library of His4 tag-fused β2ARs and MACs in hand, we set up an efficient screening system based on fluorescence Ca2+ imaging with the indicator Fura-2. To assess β2AR activity by changes in intracellular Ca2+ concentration ([Ca2+]i), we transiently transfected HEK293T cells with plasmids for both β2AR and Gα15 proteins[45] (Figure a). For use in screening experiments, six different types of HEK293T cells were prepared harboring a vector plasmid, plasmids of wild-type (WT) β2AR, or four β2AR mutants. Using a fluorescence microscope, we then monitored [Ca2+]i changes upon addition of each MAC or control (0.1 μM) in HEPES-buffered saline (HBS). We defined Δratio as the difference between the initial and maximum ratio values and quantitatively evaluated the activity as shown in Figure b.

Figure 3

Chemogenetic activation of His4 tag-fused β2AR by coordination tethering. (a) Schematic illustration of fluorescence Ca2+ imaging. HEK293T cells were transiently transfected with both β2AR and Gα15 proteins. PLC, phospholipase C; PIP2, phosphatidylinositol 4,5-biphosphate; IP3, inositol 1,4,5-triphosphate; ER, endoplasmic reticulum. (b) Ca2+ responses induced by 0.1 μM (red) MAC(βAR, O4) or (black) control-1 in HEK293T cells expressing β2AR(1.19H4). Δratio was defined as the difference between the maximum and initial ratio values (n = 29, 52). (c) The color map shows the averaged Δratio evoked by 10 min of treatment with 0.1 μM MACs or controls for the control vector, WT β2AR, and β2AR mutants (n = 10–52). O0, MAC(βAR, O0); O4, MAC(βAR, O4); O8, MAC(βAR, O8). (d, e) Concentration–response curves of (d) β2AR(1.19H4) and (e) WT β2AR for (red) MAC(βAR, O4) and (black) control-1 (n = 15–60). (f) Reversible Ca2+ responses of β2AR(1.19H4) evoked by 0.1 μM MAC(βAR, O4) (n = 28). The data represent the mean ± SEM. Reproducibility of all experiments was confirmed at least three times.

The Δratio values are summarized as a color map in Figure c, and these data were used to identify two hit pairs, β2AR(1.19H4) or β2AR(1.27H4) mutants, each paired with MAC(βAR, O4). HEK293T cells expressing β2AR(1.19H4), for instance, showed a significant change in [Ca2+]i upon addition of 0.1 μM MAC(βAR, O4) (Δratio = 0.48 ± 0.12), while this effect was not observed with control-1 (Δratio = 0.037 ± 0.007) (Figure b,c, Figure S1). The other hit pair, β2AR(1.27H4) and MAC(βAR, O4), showed a similar increase in Δratio (0.5 ± 0.2 in MAC(βAR, O4) and 0.030 ± 0.006 in control-1). In contrast, with WT β2AR, there was no change in [Ca2+]i on treatment with MAC(βAR, O4) or control-1 (Figure c, second row). These data suggested that the coordination bond between the His4 tag and Ni–NTA was essential for activation of β2AR(1.19H4) and β2AR(1.27H4).

The screening data also showed that the linker length of MAC was crucial for coordination tethering. For the hit β2AR(1.19H4) (Figure c, second row from the bottom), a MAC bearing the shortest linker (MAC(βAR, O0)) failed to activate the receptor, presumably because the Ni–NTA moiety could not reach the His4 tag on β2AR. The MAC with the longest linker (MAC(βAR, O8)), on the other hand, did activate the β2AR(1.19H4) but with much lower potency than that of the hit MAC(βAR, O4). This was related to entropic loss attributable to an excessively long and flexible ethylene glycol linker. More interestingly, the incorporation site of the His4 tag was also critical. When the His4 tag was inserted at sites more distant to the ligand binding site (that is in 1.1H4 and 1.9H4 mutants), neither the hit MAC(βAR, O4) nor the longer MAC(βAR, O8) could activate the mutant receptors. With β2AR(1.27H4), on the other hand, insertion of the His4 tag too close to the transmembrane domain may have impaired the natural function of β2AR. Therefore, we selected β2AR(1.19H4) as the optimal mutant for more detailed analyses.

Detailed Responses of the Hit Pair β2AR(1.19H4) and MAC(β

For the optimal pair, β2AR(1.19H4) and MAC(βAR, O4), the concentration dependence of MAC(βAR, O4) on β2AR activity was evaluated. In the case of control-1, both WT β2AR and β2AR(1.19H4) had nearly identical concentration-dependent saturation curves, with EC50 values of 1.3 and 1.3 μM, respectively (Figure d,e, black lines). This indicated that introduction of the His4 tag did not impair receptor activity. As expected, the EC50 value of MAC(βAR, O4) for β2AR(1.19H4) was increased by 93-fold (0.014 μM) (Figure d, red line). In sharp contrast, the EC50 value of MAC(βAR, O4) for WT β2AR was barely shifted (0.68 μM) (Figure e, red line). We also confirmed that MAC(βAR, O4) activated the engineered β2AR(1.19H4) repeatedly, at least three consecutive times (Figure f).

As a control experiment, we examined the concentration dependence of agonist-induced activation of the β2AR(1.19H4) mutant by PE, in the absence and presence of Ni–NTA or Ni2+ ion (Figure S2a,b). The dose dependence was not affected by Ni–NTA or Ni2+ ion (EC50 values: 0.32 μM for without Ni–NTA or Ni2+ ion, 0.38 μM for with Ni–NTA, and 0.36 μM for with Ni2+ ion) (Figure S2c). Also, the enhanced response of MAC(βAR, O4) was completely inhibited by ICI-118551, an antagonist for β2AR (Figure S3). These results revealed that two points of interaction between β2AR and MAC (using the orthosteric (PE) binding sites and the His tag (Ni–NTA) coordination site) were essential for β2AR activation by coordination tethering. In addition, we confirmed that no response was detected in the presence of U73122, a PLC inhibitor (Figure S4). This indicated that the observed response induced by MAC(βAR, O4) was evoked by activation of the Gα15/PLC pathway.

We also confirmed that coordination tethering could activate the endogenous Gs pathway of β2AR by monitoring the concentration change of cAMP. In the case of β2AR(1.19H4), the EC50 value of MAC(βAR, O4) for cAMP production was lower than that of control-1 by 24-fold (0.011 and 0.26 μM, respectively) (Figure S5a). In contrast, WT β2AR showed no prominent difference in the EC50 values between MAC(βAR, O4) and control-1 (0.23 and 0.16 μM, respectively) (Figure S5b).

We additionally investigated whether the number of His residues in the His tag sequence affected β2AR activation, using a mutant (β2AR(1.19H8)) bearing eight His residues at the same insertion position as that of the hit His4 mutant (β2AR(1.19H4), see Figure S6a for the mutant design). The EC50 value of MAC(βAR, O4) for β2AR(1.19H8) was significantly lower than that of control-1 (0.017 and 1.0 μM, respectively). This indicated that coordination tethering worked well for this receptor, similarly to results with β2AR(1.19H4), and that the longer His tag did not impair ligand–receptor binding (Figure S6b). Meanwhile, the EC50 was not considerably improved by increasing the number of His residues, suggesting that four His residues were sufficient for MAC(βAR, O4)-induced receptor activation.

Selective Activation of Muscarinic Acetylcholine Receptors

To validate the general applicability of this strategy, we next examined selective activation of a second class A GPCR. Muscarinic acetylcholine receptors (mAChRs) are widely expressed in the body, such as in the central and peripheral nervous systems, and are mainly involved in cognitive and behavioral processes.[46,47] Among five subtypes of mAChRs, we selected Gq-coupled M1R as the target and incorporated a His4 tag sequence into the M1R at nearly the same site (M1R(1.18H4)) as in the hit β2AR(1.19H4) mutant (using Ballesteros–Weinstein nomenclature, see the Supporting Information for plasmid construction) (Figure a). Concurrently, a MAC was designed for mAChRs (MAC(M1R) in Figure b,c) that included iperoxo, the selective agonist for mAChRs, and Ni–NTA connected by a tetra-ethylene glycol linker. On the basis of the crystal structures of M2R and iperoxo,[48] the linker was connected at the terminal ammonium moiety.

Figure 4

Chemogenetic activation of mAChR, M1R, by coordination tethering. (a) Design of the His4 tag-fused M1R, M1R(1.18H4). Red and green letters represent a His4 tag and a HA tag, respectively. (b) Chemical structure of MAC(M1R). (c) Synthetic scheme of MAC(M1R). MAC(M1R) was prepared by complexation of control-2 and 1 equiv of NiSO4 at room temperature at 30 min, followed by being used for cellular assays without isolation. (d) Ca2+ responses evoked by 0.1 μM of (red) MAC(M1R) and (black) control-2 in CHO cells expressing M1R(1.18H4) (n = 18). Concentration–response relationships of (e) M1R(1.18H4) and (f) WT M1R on (red) MAC(M1R) and (black) control-2 (n = 17–27). The data represent the mean ± SEM. Reproducibility of all experiments was confirmed at least two times.

The activities of WT M1R and M1R(1.18H4) transiently expressed in CHO cells were evaluated with fluorescence Ca2+ imaging, using a protocol similar to that used for β2AR screening. The concentration dependence for activating M1R(1.18H4) showed that EC50 of MAC(M1R), compared with control-2, was decreased by approximately 16-fold (0.14 μM in MAC(M1R) and 2.3 μM in control-2, Figure d,e). In contrast, WT M1R showed no substantial difference in EC50 values for MAC(M1R) and control-2 (3.7 μM in MAC(M1R) and 2.4 μM in control-2, Figure f). This clearly implied that selective M1R(1.18H4) activation may be achieved with less than 100 nM MAC(M1R), even in the presence of WT M1R. We also found that the EC50 value of M1R(1.18H4) for control-2 was comparable to that of WT M1R. This suggested that the M1R(1.18H4) mutant retained its natural responsiveness. Additionally, the concentration–response of M1R(1.18H4) for iperoxo was unaffected by addition of 1 μM Ni–NTA (Figure S7). Similar to that of the β2AR mutant, activation of M1R(1.18H4) by MAC(M1R) was completely inhibited by pretreatment with pirenzepine (an M1R-selective antagonist), YM254890 (a Gq GDP/GTP exchange inhibitor), or U73122 (a PLC inhibitor) (Figures S8–S10, respectively). This demonstrated that coordination tethering indeed relied on interactions at two sites, enabling activation of the Gq/PLC pathway. It is noteworthy that our design strategy for chemogenetic activation was successfully applied to not only β2AR, but also mAChR, even though these two receptors have the lowest sequence homology among the aminergic GPCRs.[49] Furthermore, we confirmed the orthogonality of the His-tagged receptor–MAC pairs. As shown in Figure S11, MAC(βAR, O4) could not activate M1R(1.18H4) at all and vice versa.

Application of Coordination Tethering in Primary Cultured Astrocytes

Finally, we applied the pair MAC(M1R)/M1R(1.18H4) to a primary cultured cell system (Figure a). In cultured cortical astrocytes, there are several endogenous Gq-coupled mAChR signaling pathways involved in DNA syntheses, cell proliferation, and regulation of neuronal and vascular functions through local release of gliotransmitters.[50,51] We prepared cultured astrocytes transiently transfected, using nucleofection, with M1R(1.18H4), WT M1R, or vector plasmids. Transfection was confirmed by Western blotting (WB) analysis and immunostaining using a human influenza hemagglutinin (HA) tag inserted at the N-termini of the M1Rs (Figure a). The WB data indicated that both mutant M1R(1.18H4) and WT M1R were successfully expressed in cultured astrocytes (Figure S12). Immunostaining with an anti-HA tag antibody showed strong fluorescence signals on the membrane surface of astrocytes transfected with M1R(1.18H4) or WT M1R plasmids (Figure b and Figure S13a). This indicated that these receptors were localized to the plasma membrane of the astrocytes. In contrast, almost no fluorescence was detected on the cell surface of astrocytes transfected with the vector plasmid (Figure S13b).

Figure 5

Selective activation of the engineered M1R in primary rat cortical astrocytes. (a) Schematic illustration of selective activation of the His4 tag-fused M1R over endogenous mAChRs by coordination tethering. (b) Confocal microscopic images of astrocytes co-transfected with AcGFP (green) and M1R(1.18H4) immunostained by an anti-HA tag antibody (magenta). Nuclei were stained with Hoechst33258 (blue). Scale bar: 20 μm. Ca2+ responses evoked by 30 nM (c) MAC(M1R) and (d) control-2 in astrocytes harboring the plasmids of (red) M1R(1.18H4), (blue) WT M1R, or (black) the control vector (n = 18). Fura-2 ratiometric images of astrocytes expressing M1R(1.18H4) (e) before and (f) during perfusion of 30 nM MAC(M1R). Red and blue represent high and low Ca2+ concentration, respectively. Scale bar: 100 μm. (g) Average Δratio plot evoked by 5 min of treatment of 30 nM MAC(M1R) and control-2 (n = 11–23). *** denote significant differences from the group of M1R(1.18H4)/MAC(M1R)/inhibitor(−) (***: P < 0.001, one way analysis of variance with Dunnett’s post hoc test). F(6/117) = 23.05. Concentration–response curves of (h) M1R(1.18H4) and (i) WT M1R on (red) MAC(M1R) and (black) control-2 (n = 20–25). The data represent the mean ± SEM. Reproducibility of all experiments was confirmed at least two times.

As shown in Figure c–g, Ca2+ imaging experiments clearly revealed that astrocytes expressing M1R(1.18H4) were activated by MAC(M1R) (30 nM), but not by control-2. MAC(M1R)-induced receptor activation was inhibited by the M1R-selective antagonist, pirenzepine (Figure g). In contrast, cells expressing WT M1R or vector did not show Ca2+ signals in response to MAC(M1R) (30 nM) (Figure c,d,g). Detailed MAC(M1R) concentration dependences indicated that its EC50 value for M1R(1.18H4) was shifted downward by 4-fold when compared with control-2 (39 nM in MAC and 170 nM in control) (Figure h). In contrast, its EC50 values of MAC(M1R) and control-2 for WT M1R showed no prominent difference (390 nM in MAC(M1R) and 340 nM in control-2) (Figure i). Therefore, we concluded that selective receptor activation in astrocytes transfected with mutant M1R(1.18H4) was attributed to a coordination bond-assisted affinity shift of MAC(M1R), selective for the engineered M1R(1.18H4). More importantly, our findings indicated that our coordination tethering strategy is applicable to primary cultured astrocytes containing a variety of endogenous mAChRs.

Conclusion

In summary, we have developed coordination tethering, enabling selective activation of His tag-fused class A GPCRs by MAC, a modified ligand, in living mammalian cells, including primary cultured astrocytes. This approach was validated with two different subfamilies of class A GPCRs, using almost identical design principles in engineering the receptors and chemically modifying the corresponding MACs. We confirmed that both MACs showed no toxicity toward mammalian cells (Figure S14). Also of note, coordination tethering can be applied to receptors with native amino acid residues at extracellular coordination sites accessible to Ni–NTA. In our study, while there are 6 Glu, 5 Asp, 6 His, and 2 Met residues distributed on the N-terminal domain and extracellular loops of β2AR (Figure S15), MAC exhibited high selectivity toward the inserted His4 tag without interference by these native residues. These results suggested that a coordination tethering approach can be extended, with flexibility, to a wide range of class A GPCRs. Of further importance, His4 tag incorporation into the target receptors did not have a detrimental impact on their responses to ligands, a key advantage for investigating natural receptor functions under equivalent conditions. A critical limitation, on the other hand, is that our coordination tethering method is not applicable to orphan receptors with unknown agonists. Recently intensified efforts to deorphanize GPCRs, however, should gradually resolve this shortcoming.[52] Another possible limitation is the lack of switching systems when we apply this coordination tethering to more complex environments where the EC50 value of MAC on His-tagged receptors is more close to that on WT receptors. We will attempt to improve coordination tethering by incorporating a switching mechanism such as a binary ON/OFF step like RECON.[35] In addition, it is conceivable that our short His4 peptide insertion may be compatible with rapidly developing in vivo genome editing techniques (such as the CRISPR-Cas9 system), enabling incorporation of peptide tags (e.g., HA tags) into endogenous proteins in a cell-specific manner.[53,54] We anticipate that the rational combination of our method with these genetic or virus infection techniques would modulate activity of endogenous GPCRs. Thus, we envision that coordination tethering will facilitate deeper understanding and more precise regulation of cell-specific receptor functions in living tissues and/or animals.

Experimental Procedures

Detailed procedures for organic syntheses, the site-directed mutagenesis, Western blot analyses, and cell viability assay are provided in the Supporting Information.

Culture and Transfection of HEK293T, HEK293, and CHO Cells

HEK293T, HEK293, and CHO cells were maintained in Dulbecco’s modified Eagle’s medium (DMEM, Sigma-Aldrich) for HEK293T and HEK293 cells or DMEM-F12 (Sigma-Aldrich) for CHO cells supplemented with 10% fetal bovine serum (FBS) (Gibco), 100 unit/mL penicillin, 100 μg/mL streptomycin, and 0.25 μg/mL amphotericin B (Gibco) at 37 °C in a humidified atmosphere of 95% air and 5% CO2. For β2ARs, HEK293T and HEK293 cells were transiently transfected with plasmids (WT β2ARs, the β2AR mutants, or the control vector) using Lipofectamine2000 (Invitrogen) and Superfect transfection reagent (Qiagen), respectively, in DMEM supplemented with 10% FBS according to the manufacture’s instruction. For M1R, CHO cells were transiently transfected with plasmids (WT M1R, the M1R mutants, or the control vector) using Lipofectamine2000 transfection reagent in DMEM-F12 supplemented with 10% FBS according to the manufacture’s instruction. The cells were co-transfected with pEGFP-F (Clontech), pmCherry-F,[36] or pDsRed monomer-F (Clontech) as a transfection marker. For Ca2+ imaging, the cells were grown for 24–36 h, seeded on glass coverslips (Matsunami) coated with poly-l-lysine solution (Sigma-Aldrich), and subjected to Ca2+ imaging 4–10 h after seeding.

Culture and Transfection of Primary Cultured Rat Cortical Astrocytes

Primary rat cortical astrocytes were obtained from P2 neonatal SD rat pups (both male and female were used) (Japan SLC) according to the protocol reported by Guaza et al.[55] The cortexes were dissected and digested by trypsin (Nacalai) at 37 °C for 30 min under humidified atmosphere (95% air and 5% O2). After centrifugation, the cells were resuspended in DMEM–GlutaMAX (Gibco) containing 10% FBS and 1% penicillin–streptomycin, gently dissociated by pipetting, and filtered by a cell strainer (100 μm). After cell counting, the cells were seeded into poly-d-lysine-coated (Sigma-Aldrich) T75 cultured flasks and grown in humidified atmosphere (95% air and 5% O2). At 5 DIV, the culture flasks were shaken at 250 rpm for 16 h at 37 °C. After removal of nonadherent cells, the adherent cells were collected by treatment of trypsin–EDTA. The collected astrocytes were transiently transfected with the plasmids of WT M1R, M1R(1.18H4), or the vector plasmid by Nucleofector (Lonza) according to the manufacture’s instruction. The cells were co-transfected with AcGFP as a transfection marker. The transfected cells were seeded on poly-d-lysine-coated glass coverslips (Matsunami), cultured in humidified atmosphere (95% air and 5% O2), and subjected to Ca2+ imaging 24–48 h after seeding. The purity of astrocytes (>80%) was verified by immunofluorescent imaging with an anti-GFAP antibody (Figure S16). All the experimental procedures were performed in accordance with the National Institute of Health Guide for the Care and Use of Laboratory Animals and approved by the Institutional Animal Use Committees of Kyoto University.

Fluorescence Ca2+ Imaging

The HEK293T, CHO cells, and astrocytes were loaded with 5 μM Fura-2 AM (Dojindo) for 15–30 min in growth medium, respectively. Fura-2 fluorescence was measured in HBS (107 mM NaCl, 6 mM KCl, 1.2 mM MgSO4, 11.5 mM glucose, 0.2 mM CaCl2, and 20 mM HEPES at pH 7.4), respectively. Control compounds for β2AR and M1R were dissolved in a mixture of HEPES buffer (10 mM, pH 7.4) and DMSO, and D2O, respectively. Precursor control compounds (control-1, control-2, control-3, and control-4) and 1 equiv of NiSO4 were mixed 20–30 min before fluorescence Ca2+ imaging. U73122 (Cayman Chemical), YM254890 (Wako), ICI-118551 (Sigma-Aldrich), and pirenzepine (Sigma-Aldrich) were dissolved in HBS from 1000× DMSO stocks. (R)-Phenylephrine (TCI) and iperoxo were dissolved in HBS from 1000× H2O stocks. Fluorescence images were obtained using a fluorescence microscopy (IX71, Olympus) instrument equipped with a CMOS camera (ORCA-flash 4.0, Hamamatsu Photonics) under xenon lamp illumination, and analyzed with a video imaging system (AQUACOSMOS, Hamamatsu Photonics) according to the manufacture’s protocol. The ratio of 340:380 nm fluorescence was determined from the images, on a pixel-by-pixel basis. For facilitation of the screening assay in HEK293T cells (Figure c and Figure S1), two different cell lines expressing one of the constructs were cocultured on a glass coverslip. Each mutant can be distinguished by co-transfected fluorescent proteins having distinct colors as a marker, and the agonist responses of two different mutants were simultaneously assayed.

cAMP Assay

A cAMP assay was conducted by using the cyclic AMP select ELISA kit (Cayman Chemical). The transfected HEK293 cells were seeded on cultured dishes and incubated for 2 days in growth medium at 37 °C in a humidified atmosphere of 95% air and 5% CO2. The cultured medium was replaced into serum-free DMEM supplemented with 10% fetal bovine serum (FBS) (Gibco), 100 unit/mL penicillin, 100 μg/mL streptomycin, and 0.25 μg/mL amphotericin B, and the cells were incubated at 37 °C for 60 min (serum-starvation). After removal of DMEM, HBS solutions containing agonists and a phosphodiesterase inhibitor (200 μM IBMX (Nacalai tesque)) were added to the cells followed by incubation at 37 °C for 15 min. After HBS solutions were removed, 0.1 M HCl solution was added to the cells, and incubated at room temperature for 20 min. The resultant cells were collected by cell scrapers and homogenized by pipetting. The homogenized solutions were analyzed by the cyclic AMP select ELISA kit (Cayman Chemical) according to the manufacture’s instruction.

Immunostaining of Cultured Astrocytes

Astrocytes cultured on glass coverslips (5 DIV) were fixed with 4% paraformaldehyde at room temperature (rt) for 30 min and washed with PBS buffer. For immunostaining by an anti-GFAP antibody, this was followed by permeabilization with PBS containing 0.1% triton X-100 at rt for 15 min. The cells were blocked with PBS containing 10% normal goat serum for 1 h at rt. After blocking, primary antibodies in PBS buffer containing 1% BSA were added and incubated overnight at 4 °C. Secondary antibodies in PBS buffer containing 1% normal goat serum were added and incubated at rt for 1 h. Used primary antibodies were as follows: mouse anti-GFAP antibody (CST, 3670, ×300) and rabbit anti-HA tag antibody (abcam, ab9110, ×300). Used secondary antibodies were as follows: goat anti-mouse-IgG-Alexa633 antibody (Invitrogen, A21050, ×1000) and goat anti-rabbit-IgG-Alexa633 antibody (Invitrogen, A21070, ×1000). The cells were also stained with Hoechst33258 (Dojindo, 1 mg/mL, ×1000). Cell imaging was performed with confocal microscopy (LSM800, Axio Observer.Z1, ZEISS) equipped with a 63×, numerical aperture (NA) 1.40 oil objective. Fluorescence images were acquired using 405, 488, and 640 nm lines of semiconductor lasers for excitation of Hoechst33258, AcGFP, and Alexa633, respectively.

Statistical Analysis

Comparison of two groups was made by two-sided unpaired Welch’s t-test. Post hoc multiple comparisons were conducted by using one-way ANOVA and Dunnett’s test. Difference with P < 0.05 was considered significant. All statistical analyses were conducted with Kaleidagraph 4.5 software (Synergy Software). All the measurements were taken from distinct samples. The exact n values used to calculate the statistics are summarized in the Supporting Information.