Activation of Multiple G Protein Pathways to Characterize the Five Dopamine Receptor Subtypes Using Bioluminescence Technology

Abstract

G protein-coupled receptors show preference for G protein subtypes but can recruit multiple G proteins with various downstream signaling cascades. This functional selection can guide drug design. Dopamine receptors are both stimulatory (D₁-like) and inhibitory (D₂-like) with diffuse expression across the central nervous system. Functional selectivity of G protein subunits may help with dopamine receptor targeting and their downstream effects. Three bioluminescence-based assays were used to characterize G protein coupling and function with the five dopamine receptors. Most proximal to ligand binding was the miniG protein assay with split luciferase technology used to measure recruitment. For endogenous and selective ligands, the G-CASE bioluminescence resonance energy transfer (BRET) assay measured G protein activation and receptor selectivity. Downstream, the BRET-based CAMYEN assay quantified cyclic adenosine monophosphate (cAMP) changes. Several dopamine receptor agonists and antagonists were characterized for their G protein recruitment and cAMP effects. G protein selectivity with dopamine revealed potential G_q coupling at all five receptors, as well as the ability to activate subtypes with the “opposite” effects to canonical signaling. D₁-like receptor agonist (+)-SKF-81297 and D₂-like receptor agonist pramipexole showed selectivity at all receptors toward G_s or G_i/o/z activation, respectively. The five dopamine receptors show a wide range of potentials for G protein coupling and activation, reflected in their downstream cAMP signaling. Targeting these interactions can be achieved through drug design. This opens the door to pharmacological treatment with more selectivity options for inducing the correct physiological events.

G protein-coupled receptors (GPCRs) are the most studied targets in drug discovery and are the most common transmembrane receptors in the mammalian genome.¹⁻⁴ GPCRs are connected to heterotrimeric proteins containing α, β, and γ subunits and can be classified according to their primary coupling behavior to the Gα subunit, which is divided into α_s-, α_i/o-, α_q/11-, and α_12/13 families.^4,5 The activation of a GPCR subsequently regulates intracellular signaling by engaging heterotrimeric G proteins or β-arrestins, which transfer the signal to further downstream effector proteins or lead to internalization and desensitization of the receptor.⁶⁻⁸

The neurotransmitter dopamine acts as an endogenous ligand through five dopamine receptors belonging to the superfamily of class A GPCRs that can be divided into two subtype families. The D₁-like family comprises the D₁ receptor (D₁R) and D₅ receptor (D₅R), which activate the second messenger pathway by stimulating adenylyl cyclase, leading to cyclic adenosine monophosphate (cAMP) production. The D₂ receptor (D₂R), D₃ receptor (D₃R), and D₄ receptor (D₄R) form the D₂-like family and, in contrast, block the cAMP-initiated second messenger cascade.^2,3,9,10 Dopamine receptors are highly expressed in ventral tegmental areas and substantia nigra pars compacta in the CNS and are of great importance in learning processes, motivation, and motorial output.¹⁰⁻¹² Dysregulation of the catecholamine dopamine plays diverse roles in pathogenesis and therapy for different diseases such as schizophrenia, Parkinson’s disease (PD), drug addiction, and bipolar disorders.¹³⁻¹⁵

In the preceding few decades, several test systems have been developed to characterize binding properties or functionality of ligands concerning either the D₁R or D₂R.^7,16,17 Unfortunately, the other subtypes have not been as much of a focus. Selectivity studies between all five dopamine receptors, coupling studies contrary to primary coupling of dopamine receptors, and the monitoring of their functional response all remain under-researched.¹⁴ The three less-targeted receptors (D₃R, D₄R, D₅R) are nevertheless important as they are associated with diseases as potential drug targets independent of the D₁R and D₂R. For example, both the D₄R and D₅R have been implicated to play a role in attention deficit hyperactivity disorder (ADHD), whereas the D₃R receptor is associated with autism spectrum disorder (ASD) and substance abuse disorder.¹⁸⁻²³ Additionally, the coupling profile of all five dopamine receptors is often referred to as “predominantly coupling” and has been previously researched in other studies to detect the receptors’ G protein preferences (the canonical predominant G protein families being G_s for D₁R and D₅R,²⁴ G_i for D_2lR, and G_o for D₃R and D₄R²⁵). Although predominant, these do not represent exclusive G protein coupling at the dopamine receptors. For example, the D₁R prefers to couple to the Gα_s subunit but can also recruit the Gα_i1, Gα_i2, Gα_o, and Gα_q subunits.²⁶ There are hints that the D₂R and D₃R can recruit β-arrestins with a ligand-dependent bias that creates new opportunities for dopamine receptor drug discovery.^16,27,28 For example, improving β-arrestin2 recruitment of the D₂R is thought to help with the therapeutic effect of L-DOPA in PD, whereas decreasing β-arrestin2 recruitment at the D₃R and using a G protein biased ligand may help treat dyskinesia.^29,30 Such selectivity profiles may also be possible, targeting the different G proteins and their coupling.

To address this, we have applied three different bioluminescence-based assay procedures for functional characterization of all five dopamine receptors in live cells (Figure 1). Through these assays, we aim to thoroughly describe and compare the G protein activation profile of the five dopamine receptors with their endogenous and selective ligands.

Figure 1.

Overview of all three test systems. (A) Scheme of the miniG recruitment assay with split NanoLuc technology,³¹ where the large fragment (LgBit) is fused N-terminally to the miniG protein and the small fragment (SmBit) C-terminally to the human dopamine receptors. After ligand binding, an induced conformational change leads to the recruitment of the respective miniG protein and complementation of the two NanoLuc fragments. Bioluminescence intensity is measured in the presence of a substrate. (B) Scheme of the G-CASE assay, where the NanoLuc is cloned into the Gα subunit of a G protein trimer, and the cpVenus protein is fused to the N-terminus of the corresponding G_γ subunit. After receptor activation by a ligand, the Gα-NanoLuc donor and cpVenus-G_γ acceptor proteins are spatially displaced, decreasing the BRET ratio and enabling measurement of the specific activation of a chosen Gα subunit. (C) Principle of a CAMYEN BRET-based biosensor. The NanoLuc is fused to the C-terminus of an Epac cAMP binding domain, and mCitrine to the N-terminus. Conformational change of the CAMYEN protein decreases the BRET ratio between the NanoLuc donor and mCitrine fluorescent acceptor when bound to cAMP. Created with BioRender.com.

Materials and Methods

Materials

The cDNAs for human (h)D_2longR (D_2lR; NM_000795.4) and hD₃R (D₃R; NM_000796.5) were kindly provided by Harald Hübner (Friedrich-Alexander-University, Erlangen, Germany), and cDNAs of hD₁R (D₁R; NM_000794.5), hD₅R (D₅R; NM_000798.5), and hD_4.4R (D₄R; NM_000797.4) were purchased from the cDNA Resource Center (Rolla, MO). Molecular biology enzymes and reagents were from New England Biolabs (NEB; Frankfurt am Main, Germany), unless otherwise described. G-CASE plasmids for G_i1, G_i2, G_i3, G_o1, G_s, G_q, and G₁₃ sensors were a kind gift from Gunnar Schulte (available on Addgene³²), and G_z-CASE was made and verified previously in-house.³³

Dulbecco’s modified Eagle’s medium high glucose (DMEM) and HEPES (1 M in distilled (d)H₂O, pH = 7.4, sterilized and stored at 4 °C) were from Sigma (Taufkirchen, Germany). Leibovitz’s L-15 medium without phenol red (L-15) was from Gibco (Taufkirchen, Germany). Fetal calf serum (FCS), trypsin (0.05% trypsin, 0.02% EDTA in PBS), and Geneticin (G418) were from Merck (Darmstadt, Germany). Puromycin and zeocin were obtained from Invivogen (Toulouse, France). The NanoLuciferase substrate furimazine (Nano-Glo) was from Promega (Walldorf, Germany), and coelenterazine H (CZH; 5 mM in methanol, stored at −80 °C) was from BioSynth s.r.o (Bratislava, Slovakia). HEK293T cells (RRID:CVCL_0063) were a kind gift from Wulf Schneider (Institute for Medical Microbiology and Hygiene, Regensburg, Germany).

Depending on their physicochemical properties, when possible, ligands were dissolved in dH₂O; otherwise, DMSO (Merck) was used as a solvent. Dopamine dihydrochloride (Dopa), (+)-butaclamol hydrochloride (Buta), pramipexole dihydrochloride (Prami), and (+)-SCH-23390 hydrochloride (SCH) were purchased from Sigma (Taufkirchen, Germany). Haloperidol (Halo) was from TCI Deutschland GmbH (Eschborn, Germany). R-(−)-Apomorphine (Apo), (+)-SKF-81297 hydrobromide (SKF), spiperone hydrochloride (Spip), nemonapride (Nemo), and (−)-quinpirole hydrochloride (Quin) were obtained from TOCRIS (Bristol, U.K.). Radioligands [³H]N-methyl-spiperone (77 Ci/mmol) and [³H]SCH-23390 (81 Ci/mmol) were purchased from Novandi Chemistry AB (Södertälje, Sweden).

Molecular Cloning

Plasmids for the split NanoLuc (also known as NanoBiT³¹) miniG protein assay were generated by standard PCR amplification and restriction enzyme techniques. The cDNAs of miniG_s (miniG_s_393; mG_s), miniG_s/i (mG_s/i_43; mG_i1), miniG_s/q (mG_s/q_71; mG_q), and miniG_o1 (miniG_o1_12; mG_o1) were a customized gene synthesis from Eurofins Genomics, according to the published protein sequence of Nehmé et al.³⁴ As described by Höring et al.³⁵ for miniG_s, miniG_i1, and miniG_q, the miniG_o1 protein was cloned within a pIREspuro3 vector backbone that encodes the large fragment of split NanoLuc (NlucN); MVFTLEDFVGDWEQTAAYNLDQVLEQGGVSSLLQNLAVSVTPIQRIVRSGENALKIDIHVIIPYEGLSADQMAQIEEVFKVVYPVDDHHFKVILPYGTLVIDGVTPNMLNYFGRP YEGIAVFDGKKITVTGTLWNGNKIIDERLITPDGSMLFRVTINS; also known as LgBiT³¹ C-terminally fused via a flexible glycine-serine linker (–GSSGGGGSGGGGSS–) downstream of the inserted cDNA sequence. The fusion of the human dopamine receptors D_2lR, D₃R, and D₄R with the small fragment of the split NanoLuc (NlucC; VTGYRLFEEIL; also known as SmBiT³¹) was facilitated by PCR, restriction enzyme digest, and ligation. The receptors were subcloned into a vector backbone of pcDNA3.1_neo-NlucC between restriction enzymes HindIII and XbaI, allowing for the fusion of NlucC on the receptor C terminus. For the D₁R and D₅R, Gibson assembly reactions were performed to subclone the gene of interest into the pcDNA3.1 vector backbone. Therefore, overlaps of 25 base pairs matching the vector backbone were attached to the receptor cDNA by PCR using the following primers, where receptor complementary bases are underlined:

D₁R fw: CAAGCTGGCTAGTTAagcttccaccATGAGGACTCTGAACACCTC

D₁R rv: cgccacctccaTCTAGACtcgagccGGTTGGGTGCTGACCGTTT

D₅R fw: CAAGCTGGCTAGTTAagcttccaccATGCTGCCGCCAGGCAG

D₅R rv: cgccacctccaTCTAGACtcgagccATGGAATCCATTCGGGGTGA

For use with the CAMYEN and G-CASE assays, whose biosensor expressions were both neomycin resistant, dopamine receptor cDNA was cloned into a pcDNA3.1_zeo-5HT_3A-FLAG-SacB vector backbone, made in-house. This vector contained zeocin antibiotic resistance, with the murine 5HT_3A signal peptide (5HT_3A; MRLCIPQVLLALFLSMLTGPGEGSR; to improve plasma membrane targeting and expression)³⁶ and FLAG tag (DYKDDDDK; to detect receptor expression) for N-terminal fusion to the receptor. The SacB gene was used as a counter-selection gene to the receptor cDNA.³⁷ For D₁R, D_2lR, D₄R, and D₅R, the receptor cDNA was cloned into the vector using standard PCR, restriction digest, and ligation techniques between BamHI and XhoI restriction sites. For the D₃R, Gibson assembly was used with an NEBuilder reaction for 1 h at 50 °C using the following primers for the PCR:

D₃R fw: acgatgacgacaagggatccaagcttGCATCTCTGAGCCAGCTG

D₃R rv: cctctagaggtaccctcgagTCAGCAAGACAGGATCTTG

A HindIII site was added upstream of the receptor sequence and two silent mutations from the original sequence were found (T873C, G897A).

Positive bacterial colonies were first extracted in a mini-prep from overnight cultures (Miniprep Kit, Nippon Genetics, Düren, Germany) and then maxi-prepped for use in mammalian cells (Maxiprep Kit, Qiagen, Hilden, Germany). All plasmids were quantified by ultraviolet–visible (UV–vis) absorbance using a NanoDrop spectrophotometer (Thermo Fisher, Braunschweig, Germany) and sequences were confirmed by custom DNA sequencing from Eurofins Genomics (Ebersberg, Germany).

The well-characterized cAMP sensor called “CAMYEL” (cAMP sensor using YFP-Epac-RLuc) was improved by replacing Renilla luciferase (Rluc) with the much brighter Nanoluciferase (Nluc).³⁸ This was achieved by PCR amplification of the DNA encoding Nluc and insertion into the CAMYEL plasmid (a kind gift from Alastair Keen) via Gibson Assembly as described above. We named the resulting construct “CAMYEN” (cAMP sensor using YFP-Epac-Nluc).

Cell Culture

HEK293T cells (passage 10–35) were cultured in DMEM supplemented with 10% FCS and 2 mM l-glutamine at 37 °C, 5% CO₂ in a H₂O-saturated atmosphere. Cells were periodically inspected for mycoplasma contamination by a Venor GeM Mycoplasma Detection Kit (Minerva Biolabs, Berlin, Germany).

Generation of Stable HEK293T Cell Lines

For biosensors requiring stable expression, HEK293T cells (passage 10–15) were seeded on a sterile 6-well dish at a cell density of 300,000 cells/mL in DMEM supplemented with 10% FCS and 2 mM l-glutamine. The next day, cells were transfected with 2 μg of cDNA of pIRES_puro-NlucN-miniG_o1, pcDNA3l-His-CAMYEN, or G_s-CASE using the transfection reagent XtremeGene HP (Merk) according to the supplier’s protocol (1:3 cDNA (μg)/XtremeGene (μL) ratio). After an incubation period of 48 h, the cells were detached by using trypsin and seeded in a 25 or 175 cm³ cell culture flask with 5 or 25 mL of DMEM supplemented with 10% FCS, respectively. Cells were allowed to attach and thereafter treated with the antibiotic puromycin (miniG_o1; 3 μg/mL) or G418 (CAMYEN and G_s-CASE; 1000 μg/mL) to achieve stable expression. The media was refreshed every 3 days, and puromycin/G418 levels were dropped to 1/600 μg/mL for continued selection pressure in later passages.

HEK293T cells stably expressing the miniG_s, miniG_i1, or miniG_o1 protein with the N-terminal NlucN were then stably transfected in the same manner with the dopamine receptor-NlucC plasmids to generate the stable cell lines miniG_s_D₁R, miniG_s_ D₅R, miniG_o1_D₃R, miniG_o1_D₄R, and miniG_si1_D_2lR. These cells then underwent continued antibiotic selection pressure through passages with 1 μg/mL puromycin and 600 μg/mL G418. HEK293T-CAMYEN cells were also stably transfected with XtremeGene HP using the same protocol to insert the pcDNA3.1_zeo-5HT_3A-FLAG-D₁R, D_2lR, D₃R, D₄R, or D₅R plasmid. All five cell lines generated were selected using 300 μg/mL zeocin and thereafter cultured using DMEM containing 100 μg/mL zeocin and 600 μg/mL G418.

MiniG Protein Recruitment Assay

HEK293T cells stably expressing the miniG protein of interest and corresponding dopamine receptor tagged by the split NanoLuc fragments were seeded in a 75 cm³ flask in DMEM supplemented with 10% FCS and 2 mM l-glutamine and allowed to grow until 80% confluence was reached. After trypsinization, the cells were suspended in Leibovitz’ L-15 media (with 5% FCS and 10 mM HEPES) and centrifuged at 700 rpm for 5 min. After discarding the supernatant, the cells were resuspended in L-15 media supplemented with 5% FCS and 10 mM HEPES and the cell density was adjusted at 1.25 × 10⁶ cells/mL. 80 μL of the cell suspension was transferred to each well of a white 96-well plate (BRANDplates cellGrade 781965, VWR) and incubated at 37 °C overnight in a humidified atmosphere.

The miniG protein recruitment assay was performed as described before by Höring et al.³⁵ at 37 °C with an EnSpire plate reader (PerkinElmer, Baesweiler, Germany) or CLARIOstar Plus plate reader (BMG LABTECH, Ortenberg, Germany) to characterize agonists (agonist mode). The dilution of the substrate and samples were prepared in Leibovitz’s L-15 media supplemented with 10 mM HEPES prior to the experiment. The basal luminescence was recorded immediately after adding 10 μL of the substrate to each well for 33 plate repeats (CLARIOstar Plus, 13 plate repeats), with an integration time of 100 ms per well (CLARIOstar Plus, 0.5 s). Ten μL of every concentration of the ligand dilution series was added in triplicate, and the final luminescence measurement was performed for a further 100 plate repeats (CLARIOstar Plus, 39 plate repeats).

A similar procedure was performed for the characterization of antagonists (antagonist mode). Here, a further 33 plate repeats were recorded (CLARIOstar Plus, 13 plate repeats) after the addition of the antagonist ligand dilutions to evaluate possible inverse agonist effects, followed by the final measurement of 100 cycles (CLARIOstar Plus, 39 plate repeats) after adding 10 μL of the endogenous agonist dopamine at an EC₈₀ concentration of 100 nM for D₁R and D₅R, 100 nM dopamine, 1 μM pramipexole in the case of D₂R and D₃R or 1 μM (−)-quinpirole for the D₄R. For normalization of the data, the negative control (solvent) and positive control (maximum level of 10 μM dopamine for D₁R and D₅R, 10/100 μM pramipexole for D_2lR/D₃R, and 10 μM (−)-quinpirole for the D₄R) were included on every plate. The resulting pK_b values were determined according to the Cheng–Prusoff equation.³⁹

In the case of miniG_o1 with the D₄R, HEK293T cells were seeded on a sterile 6-well dish at a cell density of 300,000 cells/mL in DMEM supplemented with 10% FCS and 2 mM l-glutamine (2 mL/well). The next day, cells were transfected with 1 μg of cDNA of pIRES_puro-NlucN-miniG_o1 and pcDNA3.1_neo-D₄R-NlucC using the Transporter 5 PEI transfection reagent (Polyscience, Inc., Warrington) according to the supplier’s protocol (1:5 cDNA (μg)/PEI (μL) ratio) in 200 μL of unsupplemented DMEM, after 20 min incubation at room temperature. After an incubation period of 48 h, cells were trypsinated and 80 μL of the adjusted cell suspension of 1.25 × 10⁶ cells/ml was transferred to each well of a white 96-well plate (BRANDplates cellGrade 781965, VWR, Darmstadt, Germany) and incubated at 37 °C overnight in a humidified atmosphere. For luminescence measurements, a Tecan Infinite Lumi (Tecan, Männedorf, Switzerland) plate reader was used with 33 plate repeats for baseline recording and 66 plate repeats after the addition of the ligands, with an integration time of 200 ms.

The same procedure was used for transient experiments probing the selectivity of miniG protein coupling to dopamine receptors, except HEK293T cells were seeded on a sterile 24-well dish at a cell density of 70,000 cells/ml in DMEM supplemented with 10% FCS and 2 mM l-Glu (500 μL/well) and transfected with 250 ng of respective plasmid DNA. After 48 h incubation period, 40 μL/well of the cell suspension was seeded in a 384-well plate (LUMITRAC medium binding, Greiner Bio-One, 781075, UV-sterilized prior to experiment). For luminescence measurements, a Tecan Infinite Lumi plate reader was used with 6 plate repeats for baseline recording and 19 plate repeats after the addition of the ligands, with an integration time of 200 ms.

G-CASE Assay

To measure the specificity of dopamine receptor G protein activation, the BRET-based G-CASE sensors were used. After trypsinization and centrifugation (1000 rpm for 5 min), HEK293T or HEK-G_s-CASE cell lines were seeded on a sterile 24-well plate at a cell density of 300,000 cells/mL in DMEM supplemented with 10% FCS (500 μL/well). The next day, cells in each well were transfected with 500 ng of pcDNA (negative control) or dopamine receptor plasmid (pcDNA3.1_zeo-5HT_3A-FLAG-D₁R, D_2lR, D₃R, D₄R or D₅R) plus 500 ng of the G-CASE plasmid (G_i1, G_i2, G_i3, G_o1, G_z, G_q, G₁₃, or pcDNA 3.1 for the G_s-CASE cell line) using linear polyethylenimine (PEI 1 mg/L, 5 μL) in 100 μL of unsupplemented DMEM, after 15 min incubation at room temperature. All 46 variants were left for 48 h, then trypsinized and centrifuged. Cells were then plated onto white, opaque 384-well plates (LUMITRAC medium binding or BRANDplates 781981; VWR, Darmstadt, Germany) in L-15 media supplemented with 5% FCS and 10 mM HEPES at 40 μL/well (24 wells of the 384-well plate from a single 24 well) and left overnight at 37 °C in a humidified atmosphere. On the day of the experiment, 5 μL of 50 μM CZH in L-15 with 10 mM HEPES (final concentration 5 μM) was added to each well and the baseline was read for three cycles (35 min) at 200 ms integration time for both “Blue” (<470 nm) and “Green” (520–580 nm) filter wavelengths on a Tecan Infinite Lumi. Prediluted ligands to 10× final concentration or negative solvent control in L-15 + 10 mM HEPES were then added in triplicate at 5 μL per well and the plate was read with the same parameters for six cycles (75 min). The BRET ratio for each time point was calculated by dividing the Green filtered light emission by the Blue filtered light emission.

CAMYEN cAMP Assay

HEK-CAMYEN_D₁R, D_2lR, D₃R, D₄R, or D₅R cells were trypsinized and centrifuged (1000 rpm for 5 min) and resuspended in L-15 media supplemented with 5% FCS and 10 mM HEPES. After adjusting the cell count to 600,000 cells/mL, the cell suspension was added to a sterile, white 96-well plate (BRANDplates cellGrade 781965) at 80 μL per well. Plates were incubated overnight at 37 °C in a humidified environment. On the day of the experiment, 10 μL of furimazine 200× diluted in L-15 with 10 mM HEPES with either 100 μM forskolin (D₂-like receptors) or 1 mM 3-isobutyl-1-methylxanthine (IBMX; D₁-like receptors) was added to each well (final assay concentrations: furimazine 2000× diluted with 100 μM IBMX or 10 μM forskolin), and the baseline luminescence was taken for both “Blue” (<470 nm) and “Green” (520–580 nm) filter wavelengths on a Tecan Infinite Lumi (Tecan, Männedorf, Switzerland) at an integration time of 100 ms for 11 cycles (30 min). Ligands diluted to 10× final concentration in L-15 media with 10 mM HEPES were added in triplicate at 10 μL/well. The luminescence response was then measured for a further 23 cycles (1 h) and the raw BRET ratio was calculated for each time point by dividing the Green filtered light emission by the Blue filtered light emission.

Calcium Mobilization Assay

HEK293A cells (passage 20–30, kindly gifted from Asuka Inoue) were plated at 300,000 cells/well in 6-well plates in DMEM with 5% fetal bovine serum (FBS) and transiently transfected the following day using PEI Max (MW 40,000; Polysciences Asia Pacific, Taipei, Taiwan) at a 1:5 DNA (μg)/PEI Max (μL) ratio. Variables tested used the following cDNA amounts per well: 1 μg of dopamine receptor plasmid plus 1 μg of empty vector; 2 μg of empty vector; 1 μg of wildtype Gα_q (#GNA0Q00000, cDNA Resource Centre) with 1 μg of dopamine receptor plasmid, or 1 μg of wildtype Gα_q with 1 μg of empty vector. Cells were detached with versene (PBS with 0.5 mM EDTA, pH = 7.4) and then plated onto black-walled, clear bottom 96-well plates (PerkinElmer) in 100 μL/well DMEM with 5% FBS the day before the experiment. On the day of the experiment, the plates were washed once with calcium imaging buffer (150 mM NaCl, 2.6 mM KCl, 1.18 mM MgCl₂, 10 mM d-glucose, 10 mM HEPES, 2.2 mM CaCl₂, 0.5% (w/v) bovine serum albumin, 4 mM probenecid, pH = 7.4) and media was aspirated before addition of 90 μL/well calcium imaging buffer with 1 μM Fluo-8-AM dye (1 mM stock made up in DMSO; AAT Bioquest, Pleasanton, CA) and 1 μM propranolol ((S)-(−)-propranolol hydrochloride, 10 mM in dH₂O, Sigma). Plates were then incubated at 37 °C for 45 min −1 h before reading on the FDSS/μCELL kinetic plate imager (#C13299, Hamamatsu Photonics, Hamamatsu, Shizuoka, Japan). Prior to read, dopamine serial dilutions and ionomycin control (1 μM final concentration) were prepared at 10× final concentration in a v-bottom 96-well plate using calcium buffer. Calcium mobilization via Fluo-8-AM fluorescence intensity (ex/em = 490:520 nm, K_d = 389 nM) was simultaneously measured in all wells every second at 37 °C for 30 s baseline. The drug dilutions were then automatically pipetted from the 96-well compound plate, and the fluorescence was read for a further 4.5 min after ligand addition.

Radioligand Binding Assay

In order to measure the receptor density for the stably expressed dopamine receptor cell lines (HEK-miniG_x_D_yR and HEK-CAMYEN_D_yR), radioligand binding assays were performed. The cell density was adjusted to 80,000 cells/well after counting in a “Neubauer” hemocytometer, followed by seeding 80 μL of the cell suspension in a binding buffer (50 mM Tris-HCl, 1 mM EDTA, 5 mM MgCl₂·6 H₂O, and 100 μg/mL bacitracin, pH = 7.4) in each well of a 96-well plate. The radioligand for D₁-like receptors was [³H]SCH-23390 (81 Ci/mmol, Novandi Chemistry AB, Södertälje, Sweden; K_d = 0.2 nM D₁R/0.3 nM D₅R^16,40,41), and [³H]N-methyl-spiperone (77 Ci/mmol, Novandi Chemistry AB, Södertälje, Sweden; K_d = 0.014 nM D₂R/0.026 nM D₃R/0.078 nM D₄R¹⁶) was used for D₂-like receptors, both with increasing concentrations in the range of approximately 1/10K_d–10K_d. Total binding was determined in the absence of any competitor, and nonspecific binding was measured by incubating the cell suspension in the presence of radioligand and (+)-butaclamol in a 2000-fold excess with a total volume of 100 μL per well. Incubation periods of 1 h for D₂-like receptors or 1.5 h for D₁-like receptors were terminated by separating bound and free radioligand with an automated cell harvester (Brandel, Gaithersburg) utilizing rapid filtration through Whatman GF/C filters precoated with 0.3% polyethylenimine. Filters were transferred to flexible 96-well sample plates (PerkinElmer, Rodgau, Germany) and incubated with scintillation cocktail for at least 5 h before radioactivity was measured using a MicroBeta² 1450 scintillation counter (PerkinElmer, Rodgau, Germany). Data were analyzed using Prism 9 (GraphPad, La Jolla, CA), and after subtracting nonspecific binding, K_i values were determined from IC₅₀ values according to the Cheng–Prusoff equation.³⁹

ELISA and Immunofluorescence

ELISA and immunofluorescence were used to assess relative amounts and localization of dopamine receptors, miniG and G-CASE components in their transient expression. Clear 96-well plates (Cellstar, 655180, Greiner Bio-one) were coated with 0.5% gelatin, cross-linked using 2.5% glutaraldehyde, and washed 10× with distilled H₂O. Cells were transiently transfected with PEI using the same amounts and ratios as the G-CASE and miniG coupling selectivity assays. After passaging, cells were diluted to a density of 600,000 cells/mL, then 500 μL of cells were added to 100 μL of cDNA and PEI, and 100 μL of this mix was seeded per well. Following overnight incubation at 37 °C, 5% CO₂, all steps were conducted at room temperature. The next day, the cells were washed 1× with PBS and fixed with 4% paraformaldehyde in PBS (Alfa Aesar) for 10 min, washed 3× with PBS, and then permeabilized using 0.1% Triton X-100 for 10 min. After washing 3-times with PBS, 1 h incubation with 0.5% bovine serum albumin (BSA) in PBS was used to block unspecific sites. Primary antibodies for either the FLAG protein (1:500; DYKDDDDK monoclonal antibody from mouse; #MA1-91878, LOT #SLCD3524, Invitrogen), fused to the N terminus of the dopamine receptors, or NanoLuc (2 μg/mL; monoclonal antibody from mouse; #MAB10026, LOT CLUG0221101, RnD systems), internally expressed within the G-CASE Gα subunit or targeting the NlucN-miniG, was dissolved in 1% BSA in PBS and incubated with cells for 1 h. Cells were then washed 3× with 0.5% BSA in PBS and either horseradish peroxidase-conjugated (ELISA, 1:3000, #31430, LOT #077M4820 V, Thermo Fisher Scientific) or Cy3-conjugated (immunofluorescence, 1:1000, #AP124C, Merk) goat antimouse polyclonal antibodies were diluted using 1% BSA in PBS and incubated with cells for 1 h. The cells were then washed 3× with PBS. For the ELISA, cells were incubated in 50 μL/well 3,3′,5,5′-tetramethylbenzidine (Sigma) for 20 min, and then 50 μL of 2 M HCl was added. The 450 nm absorbance was measured on an EnSpire plate reader (PerkinElmer, Rodgau, Germany) at 37 °C. For immunofluorescence, DAPI (1:50 in PBS) was used to stain cell nuclei after 10 min of incubation and then washed 3× with PBS. Images were taken on a widefield Eclipse Ts2-FL inverted microscope (Nikon Europe, Amstelveen, The Netherlands) using a 10×/0.25 Ph1 ADL WD 6.2 objective. For each variable, three fluorescence images were taken; DAPI (385 nm filter cube) for nuclei, with 30 ms exposure, Cy3 (525 nm filter cube) for antibody-tagged proteins, with 1 s exposure, and cpVenus (470 nm filter cube) for tagged Gγ subunits, with 400 ms exposure.

Data Analysis

The luminescence or raw BRET ratio traces were recorded for each ligand addition in a time-resolved manner, averaged between the well repeats, and corrected to the baseline measurements. The area under the curve (AUC) was calculated and used in concentration–response curves for all assays (time period of 1 h for CAMYEN and G-CASE assays, 45 min for the miniG protein recruitment assay, and 4.5 min for the calcium assay). Different time periods were used to capture the peak or sustained plateau of each assay, and the AUC was used as it provided a more holistic representation of ligand action than a single time point, measuring the total effect over a given time period and negating any small variations in timing between experiments while including important kinetic differences in ligand action into the final measurement. Finally, the CAMYEN assay data were normalized to 1 μM dopamine response for each receptor. Any ligand that produced a cAMP accumulation/inhibition that was not 100% (including the 95% confidence interval) of the 1 μM dopamine response was classed as a partial agonist. For the miniG recruitment assay, the data were normalized to the highest concentration of either the endogenous agonist dopamine (10 μM D₁R and D₅R), pramipexole (10 μM D_2lR and 100 μM D₃R), or (−)-quinpirole (10 μM D₄R), and for the calcium assay, data were normalized to 1 μM ionomycin response. G-CASE assay data for efficacy (E_max) were normalized to the dopamine response as a reference ligand for each G protein subunit. All calculations were conducted using Prism 9 or 10 (GraphPad, La Jolla, CA).

For comparison of assay responses where receptor and sensor levels varied, Δlog(E_max/EC₅₀) was calculated using EC₅₀ values and E_max percentage responses from the concentration–response curves best fit in Prism. Only drug responses with both an E_max ≥ 5% and pEC₅₀ > 4.50 (miniG) or >6.00 (G-CASE) were considered true responses. A three-parameter fit was used for all assays. These values were then referenced to the sensor with the highest response with dopamine as a reference ligand (G_s-CASE for D_1,5Rs, G_z-CASE for D_2l,3,4Rs).

For statistical analysis, data were first tested for normal Gaussian distribution using the Shapiro–Wilk test and, when confirmed, one-way analysis of variation (ANOVA) with Šídák’s multiple comparisons test was used (calcium response data). Statistical analysis was only performed on data where n = 5 from different experiments on different days (achieved with all primary experiments). A p value < 0.05 was classed as significant.

Data are expressed as mean ± SEM, unless otherwise specified.

Results and Discussion

Characterization of Canonical MiniG Coupling by Dopamine Receptors

The miniG protein recruitment assay combined with split NanoLuc technology is an excellent tool for the functional characterization of ligands and investigation of different G protein coupling behaviors of GPCRs in live cells.¹ Upon activation by a ligand, a conformational change of the receptor is induced and recruitment of the miniG protein occurs with concomitant complementation of the split luciferase fragments. The resulting bioluminescence is measured under the presence of the NanoLuc substrate, and recruitment of the specific miniG protein upon receptor activation can be monitored in a concentration-dependent manner of the ligand in real time. An additional advantage is the ability of the miniG protein to bind to the receptor in the absence of a ligand for monitoring of inverse agonism.

In accordance with the canonical signaling pathways of dopamine receptors, the miniG_s was used with D_1,5Rs, miniG_i1 with the D_2lR, and miniG_o1 with D_3,4Rs in our assays. To examine the influence of receptor modification (C-terminus NlucC fusion) and receptor expression in all stable NlucN-miniG_x_D_yR-NlucC cell lines, radioligand saturation binding experiments were performed with the radiolabeled antagonist [³H]SCH-23390 for the D₁-like and [³H]N-methylspiperone ([³H]NMSP) for the D₂-like family constructs (Figure S1 and Table S1). Although the D_1,2l,3,5R-NlucCs expressed at a similar level, the D₄R-NlucC expressed poorly (roughly 5-fold lower). We performed an additional experiment to investigate the impact of varying the plasmid cDNA, therefore changing receptor expression in our system (Figure S30). The D_2lR-NlucC was chosen as a representative receptor due to its varied expression levels between the miniG and CAMYEN assays (Table S1). The D_2lR-NlucC was transfected in differing amounts (0.125–1 μg) using an equivalent amount of the NlucN-mG_i1 sensor (1 μg) and made up to 2 μg with empty vector cDNA when required. Upon (−)-quinpirole stimulation, no difference in potency was observed for differing D_2lR amounts.

Sensor expression was determined using an ELISA with an anti-NanoLuc antibody against the NlucN (Figure S10). The mG_s_D₁R expressed the highest, followed by a modest reduction in sensor expression for mG_i1_D_2lR and mG_s_D₅R. The mG_o sensor in conjunction with the D_3,4Rs expressed the lowest in our system (about 50% of mG_s_D₁R).

We tested dopamine as the endogenous ligand, (+)-SKF-81297 and (+)-SCH-23390 as D₁R-like ligands, and pramipexole and (−)-quinpirole as D₂-like ligands at all five dopamine receptors. Additionally, we tested (−)-apomorphine, (+)-butaclamol, haloperidol, spiperone, and nemonapride for selected dopamine receptors. For an accurate differentiation between the selected standard ligands, we used an agonist mode, where ligands are added after reaching baseline, and antagonist mode, where an extra incubation step with an antagonist was added before the agonist.

Canonical MiniG Coupling with Dopamine Receptor Agonists

We have obtained potencies and efficacies of a broad range, which were comparable with literature data (Table S2). Dopamine was characterized as a full agonist at all receptors except the D₄R_mG_o1 (Figures 2 and 3). At the D₃R_mG_o1, the potency of dopamine was lower than expected (pEC₅₀ = 5.78 ± 0.35) compared to a published radioligand binding assay (pK_i = 7.74⁴²). (+)-SKF-81297, which is described as either D₁-like partial agonist^43,44 or full agonist,^45,46 was confirmed to be a full agonist for D₁-like receptors in this assay (E_max: D₁R_mG_s = 105 ± 3%, D₅R_mG_s = 104 ± 2% dopamine response) and showed high potency at the D₁R_mG_s (pEC₅₀ = 7.89 ± 0.12) and D₅R_mG_s (pEC₅₀ = 7.33 ± 0.10). The selective D₂-like ligand pramipexole was a full, potent agonist at the D₂-like receptors D_2lR_mG_i1 (pEC₅₀ = 7.69 ± 0.06) and D₃R_mG_o1 (pEC₅₀ = 8.08 ± 0.07), but not the D₄R_mG_o1, where no activation of the receptor was observed.

Figure 2.

Concentration–response curves of standard ligands at the D₁-like receptors obtained in the miniG_s protein recruitment assay using agonist mode and antagonist mode. (A, C) Agonist mode for the D₁R and D₅R, respectively. (B, D) Antagonist mode was performed in the presence of the agonist dopamine (100 nM) for the D₁R and D₅R, respectively. All experiments were performed using whole cells stably expressing NlucN-miniG_s and the respective dopamine receptor-NlucC. Dashed lines represent incomplete or flat curve fits. Data are expressed as mean ± SEM of n = 5 independent experiments, each performed in triplicate.

Figure 3.

Concentration–response curves of standard ligands at the D₂-like receptors obtained in the miniG protein recruitment assay using agonist mode and antagonist mode. (A) Agonist mode of standard ligands for the D_2lR_mG_i1. (B) Antagonist mode was performed in the presence of the agonist pramipexole (1 μM) for the D_2lR_mG_i1. (C) Agonist mode for the D₃R_mG_o1. (D) Antagonist mode was performed in the presence of the agonist pramipexole (1 μM) for the D₃R_mG_o1. (E) Agonist mode for the D₄R_mG_o1. (F) Antagonist mode was performed in the presence of 1 μM (−)-quinpirole for the D₄R_mG_o1. All experiments were performed on whole cells stably expressing NlucN-miniG_s/i1/o1 and the respective dopamine receptor-NlucC, except in the case of the D₄R. HEK293T cells were transiently transfected with 1 μg of plasmid DNA of NlucN-miniG_o1 and D₄R-NlucC. Dashed lines represent incomplete or flat curve fits. Data are expressed as mean ± SEM of n = 5 independent experiments, each performed in triplicate.

Canonical MiniG Coupling with Dopamine Receptor Antagonists

The reported antagonists (+)-SCH-23390, (+)-butaclamol, and spiperone were successfully characterized in the miniG protein recruitment assay as competitive antagonists. By taking advantage of the constitutive activity in the miniG assay, the three ligands were also tested in the agonist mode and showed an additional partial (inverse) signal. Inverse or partial agonistic effects were also observed in the antagonist mode, lowering the signal under the baseline in the highest concentrations ((+)-butaclamol at D_1,3,5R and spiperone at D₃R; Figures 2 and 3), or by an incomplete displacement of dopamine by (+)-SCH-23390 to an efficacy of approximately 40% (D₁R_mG_s) and 20% (D₅R_mG_s; Figure 2).

CAMYEN-Measured cAMP Response at the Five Dopamine Receptor Subtypes

To compare miniG coupling with downstream effects, cAMP responses of the five dopamine receptors were investigated. Relative accumulation of intracellular cAMP was measured in real time using the CAMYEN BRET-based biosensor with stably expressed dopamine receptors. As with the miniG protein recruitment assay, receptor expression was evaluated using radioligand binding, where the D_1,3,5Rs expressed at a similar level, the D_2lR about 10-fold lower, and the D₄R 6-fold higher (Figure S2 and Table S1). Compared to the NlucC-tagged receptors, receptor densities were comparable for the D_1,3,5Rs. D_2lR expressed about 10-fold lower than the D_2lR-NlucC, and the D₄R had a 25-fold increase in receptor density than the D₄R-NlucC.

Concentration–response curves using area under the curve of the BRET ratio after 1 h were generated for the five dopamine receptors with seven different ligands (dopamine, (−)-apomorphine, (+)-butaclamol, (+)-SCH-23390, (+)-SKF-81297, pramipexole, and (−)-quinpirole), normalized to 1 μM dopamine (100%) and either 10 μM forskolin (D₂-like, 0%) or buffer (D₁-like, 0%; Figure 4). Kinetic traces for each ligand at each receptor and the CAMYEN response to ligands without transfected receptors are provided in Figures S3–S9. The kinetic traces show that all ligand/receptor combinations producing a response form a plateau at peak cAMP inhibition/accumulation over the 1 h time period. There is little discrepancy between the times taken to reach the peak plateau response for any receptor/ligand combinations using this assay, which is within approximately 6 min after ligand addition. Without dopamine receptor transfection, the wildtype HEK293T cells produced a small, transient increase in cAMP production upon 1–10 μM (+)-SKF-81297 addition. No other ligand produced any response in either cAMP inhibition or stimulation.

Figure 4.

Ligand-induced changes of cAMP in HEK293T cells stably expressing the CAMYEN biosensor with the five dopamine receptors. Concentration–response curves were generated using area under the curve of the BRET ratio/buffer control trace over 1 h. D₁-like receptor cAMP production in (A) D₁R and (B) D₅R was measured using 100 μM IBMX (30 min preincubation) and normalized to buffer (0%) and 1 μM dopamine (DA; 100%) responses. D₂-like receptor cAMP inhibition in (C) D_2lR, (D) D₃R, and (E) D₄R was taken with 10 μM forskolin (30 min preincubation) and normalized to 10 μM forskolin (0%) and 1 μM dopamine (DA; 100%) responses. For all graphs, the purple bar represents either 10 μM forskolin cAMP response (D₁-like) or the baseline CAMYEN read in buffer (D₂-like) to show the dynamic range of the sensor in each stable cell line. Ligands include agonists dopamine and (−)-apomorphine, antagonists (+)-butaclamol and (+)-SCH-23390, D₁-like selective agonist (+)-SKF-81297, and D₂-like selective agonists quinpirole and pramipexole. Dashed lines represent incomplete or flat curve fits. Data are expressed as mean ± SEM of n = 5 independent experiments, each performed in triplicate.

Dopamine Receptor Agonist cAMP Responses

In general, pEC₅₀ values were increased in the CAMYEN cAMP assay compared with the miniG recruitment assay, typically by 1–1.5 log units (Table S2). Dopamine produced similar potency responses at the five receptor subtypes, with EC₅₀ values ranging from 0.8 to 6.0 nM (pEC₅₀ = 9.09–8.22). Comparably, (−)-apomorphine was also a potent full agonist at all receptors with EC₅₀ values between 0.8 and 17.4 nM (pEC₅₀ = 9.09–7.76). The D₁-like selective compound (+)-SKF-81297 had subnanomolar potency at D_1,5Rs and, unlike miniG recruitment, was only a partial agonist for cAMP production at the D₁R (84% dopamine response). (+)-SKF-81297 had a low-potency inverse agonist effect at D₃R, yet had a partial agonist cAMP response via the D₄R (pEC₅₀ = 6.19, E_max = 61%) and D_2lR (pEC₅₀ = 6.87, E_max = 22%). Both D₂-like selective agonists, (−)-quinpirole and pramipexole, were unable to produce full concentration–response curves for D_1,5Rs up to 10 μM (Figure 4). D₂-like receptors however had comparable cAMP potencies for (−)-quinpirole and pramipexole, ranging from 0.3 to 1.4 nM (pEC₅₀ = 9.49–8.86) and were always full agonists.

Dopamine Receptor Antagonist cAMP Response

In agreement with the miniG protein recruitment, the antagonist (+)-butaclamol produced an inverse agonist effect on cAMP production (E_max: D₁R = −43%, D_2lR = −44%, D₃R = −32%, D₄R = −43%, and D₅R = −37% dopamine response), and the other antagonist (+)-SCH-23982 acted as a partial agonist at all subtypes except D₃R (E_max: D₁R = 56%, D_2lR = 58%, D₄R = 83%, and D₅R = 54% dopamine response).

Selectivity of MiniG Protein Coupling by Dopamine Receptors

To investigate the selectivity of miniG protein coupling by dopamine receptors, HEK293T cells were transiently transfected with 1 μg of dopamine receptor plasmid cDNA (D_1–5R-NlucC) and one of the four miniG proteins (NLucN-mG_s/i1/q/o1). For characterization, we have used dopamine, (+)-SKF-81297 as a D₁-like selective ligand, and pramipexole as a D₂-like selective ligand. For evaluation of the data, Δlog(E_max/EC₅₀) was calculated and normalized to canonical coupling of the respective dopamine receptor activated by dopamine. The canonical signaling was set to “0.00”, whereby a negative value shows a decreased selectivity and a positive value an increased selectivity.

D₁-Like Receptor MiniG Protein Selectivity

Within the D₁-like family, both receptors showed preference in their selectivity of miniG protein coupling and also exhibit the ability to couple with other subtypes (Figure 5). Additional to the canonical coupling with mG_s, the D₁R-NlucC recruited mG_i1 through dopamine and (+)-SKF-81297 activation (Δlog(E_max/EC₅₀): dopamine = −0.13/(+)-SKF-81297 = −0.15) with a high efficacy, but lower potency compared to mG_s. The D₅R-NlucC was also able to recruit mG_i1 (Δlog(E_max/EC₅₀): dopamine = −0.11/(+)-SKF-81297 = −0.14); however, both ligands had 10-fold lower efficacy compared to mG_s coupling. The D₁R-NlucC recruited mG_q upon binding any of the three ligands, where, surprisingly, the activation profile was pramipexole > dopamine > (+)-SKF-81297 (Δlog(E_max/EC₅₀): −0.008 > −0.29 > −0.48). However, in the case of efficacy, the selectivity changed to (+)-SKF-81297 > dopamine > pramipexole. The D₅R-NlucC failed to recruit mG_q, and neither receptor recruited mG_o1 with the compounds tested.

Figure 5.

Selectivity profile of miniG protein recruitment at D₁-like receptors. (A) Selectivity for the recruitment of four different miniG proteins (miniG_s/i1/q/o1) at the D₁R using Δlog(E_max/EC₅₀) normalized to the dopamine/mG_s response. (B) pEC₅₀ and (C) E_max are shown separately as heatmaps, where 100% E_max is taken as the mG_s response to dopamine. (D) Selectivity for the recruitment of four different miniG proteins (miniG_s/i1/q/o1) at the D₅R using Δlog(E_max/EC₅₀) normalized to the dopamine/mG_s response. (E) pEC₅₀ and (F) E_max are shown separately as heatmaps, where 100% E_max is taken as the mG_s response to dopamine. Data are expressed as the means of n = 5 independent experiments, each performed in triplicate. DA = dopamine, PR = pramipexole, SKF = (+)-SKF-81297.

D₂-Like Receptor MiniG Protein Selectivity

In the D₂-like family, the D₃R-NlucC and D₄R-NlucC were both selective for mG_i1/o1 (Figure 6). Only the D₃R-NlucC coupled to mG_o1 with all three ligands (Δlog(E_max/EC₅₀): dopamine = 0.00/pramipexole = 0.29/(+)-SKF-81297 = −3) but with low potency for dopamine and pramipexole. The mG_o1 protein was recruited to the D₄R-NlucC after being activated by dopamine (Δlog(E_max/EC₅₀): −0.06) and (−)-quinpirole, with low potency and efficacy for dopamine and surprisingly low potency for (−)-quinpirole. In the case of the D_2lR-NlucC, dopamine coupled the receptor to mG_q additional to its canonical coupling (mG_i1 and mG_o1), with very low efficacy. Interestingly, (+)-SKF-81297 and pramipexole were not able to couple D_2lR-NlucC to mG_q but recruited both mG_i1 and mG_o1 (Δlog(E_max/EC₅₀): mG_i1 dopamine = 0.00/(+)-SKF-81297 = 0.22/pramipexole = 0.15; mG_o1 dopamine = −0.05/(+)-SKF-81297 = −0.29/pramipexole = −0.06). Concentration–responses for all ligand/receptor pairs are shown in Figures S21 and S22.

Figure 6.

Selectivity profile of miniG protein recruitment at D₂-like receptors. (A) Selectivity for the recruitment of four different miniG proteins (miniG_s/i1/q/o1) at the D₂R using Δlog(E_max/EC₅₀) normalized to the dopamine/mG_i1 response. (B) pEC₅₀ and (C) E_max are shown separately as heatmaps, where 100% E_max is taken as the mG_i1 response to dopamine. (D) Selectivity for the recruitment of four different miniG proteins (miniG_s/i1/q/o1) at the D₃R using Δlog(E_max/EC₅₀) normalized to the pramipexole/mG_o1 response. (E) pEC₅₀ and (F) E_max are shown separately as heatmaps, where 100% E_max is taken as the mG_o1 response to pramipexole. (G) Selectivity for the recruitment of four different miniG proteins (miniG_s/i1/q/o1) at the D₄R using Δlog(E_max/EC₅₀) normalized to the (−)-quinpirole/mG_o1 response. (H) pEC₅₀ and (I) E_max are shown separately as heatmaps, where 100% E_max is taken as the mG_o1 response to (−)-quinpirole. Data are expressed as the means of n = 5 independent experiments, each performed in triplicate. DA = dopamine, PR = pramipexole, SKF = (+)-SKF-81297; QUN = (−)-quinpirole.

Activation of G Proteins by the Five Dopamine Receptor Subtypes Using G-CASE Sensors

In order to expand on the selectivity observations seen with the miniG protein recruitment, the five dopamine receptors were transiently transfected with eight different G-CASE sensors (G_i1, G_i2, G_i3, G_o1, G_z, G_q, and G₁₃; G_s-CASE was used as a stable cell line) into HEK293T cells to measure ligand-dependent G protein activation. Receptor expressions were consistent between the G-CASE sensors, with D₃R and D₅R usually expressing the lowest, although showing some variation between experiments, and dopamine receptor co-expression had no effect on sensor expression (ELISA and immunofluorescence; Figures S11–S20). As with the miniG assay, ligands were selected for checking subtype and endogenous activity; endogenous ligand dopamine, D₁-like selective (+)-SKF-81297, and D₂-like selective pramipexole.

The endogenous ligand dopamine was used as the reference ligand for sensor selectivity measurements. G₁₃ was not activated by any receptor/ligand combination. Concentration–responses for all ligand/receptor/sensor combinations are shown in the Supporting Information, along with negative control ligand responses in wildtype HEK293T cells expressing only the G-CASE sensor (Figures S23–S28).

D₁-Like Receptor Noncanonical G Protein Activation

For the D₁-like receptors, G_s-CASE was used as the canonical reference sensor. The D₁R was the most selective for G protein activation, limited to G_s,z,o1,q-CASE (Figure 7). The selectivity profile for dopamine was G_s > G_q > G_o1 > G_z (Δlog(E_max/EC₅₀): 0.00, −1.18, −1.29, −1.86, respectively). (+)-SKF-81297 produced a 20-fold greater G_s-CASE response at the D₁R with a selectivity profile for G_s > G_o1 > G_q > G_z (Δlog(E_max/EC₅₀): 1.47, 0.92, −0.11, −0.85, respectively), whereas pramipexole could only produce weak G_i2,q,z-CASE activation. E_max and therefore Δlog(E_max/EC₅₀) could not be calculated for the G_i2 sensor as the reference ligand, dopamine, did not activate G_i2-CASE up to the 10 μM tested.

Figure 7.

G protein selectivity profile of D₁-like receptors determined by the G-CASE assay. The endogenous ligand dopamine, D₂-like selective pramipexole, and D₁-like selective (+)-SKF-8129 are assessed for ligand-induced G protein selectivity. (A) Web of selectivity at the D₁R for the eight different G-CASE constructs (G_i1, G_i2, G_i3, G_o1, G_z, G_s, G_q, G₁₃) using Δlog(E_max/EC₅₀) referenced to the dopamine/G_s response. (B) pEC₅₀ and (C) E_max are shown separately as heatmaps, where 100% E_max is taken as the response to dopamine for each sensor. (D) Web of selectivity at D₅R for the eight different G-CASE constructs (G_i1, G_i2, G_i3, G_o1, G_z, G_s, G_q, G₁₃) using Δlog(E_max/EC₅₀) referenced to the dopamine/G_s response. (E) pEC₅₀ and (F) E_max are shown separately as heatmaps, where 100% E_max is taken as the response to dopamine for each sensor. Data are expressed as the means of n = 5 independent experiments, each performed in triplicate. DA = dopamine, PR = pramipexole, SKF = (+)-SKF-81297. Note the logarithmic scale of the web axes. White spaces without numbers in the heatmaps indicate no measurable response up to 10 μM ligand, apart from *no E_max could be calculated for the G_i2 sensor with pramipexole at D₁R due to no response to dopamine up to 10 μM.

When bound to dopamine, the D₅R could activate all G_s,i/o/z,q subtypes, with a selectivity for G_s > G_i1/z > G_i2/o1 > G_q/i3 (G_s 4–40-fold over G_z/i1-G_q/i3, respectively; Figure 7). Pramipexole concentrations up to 10 μM could not activate G_q-CASE, G_i2-CASE, or G_s-CASE at the D₅R; however, similar potencies and efficacies to dopamine were seen for G_i1,i3,o1,z-CASE. Similarly, (+)-SKF-81297 could not produce G_i1,q-CASE responses through D₅R; however, the D₁-selective compound had a 5-fold greater G_s response than dopamine, and a 40-fold decrease in G_z activation (Δlog(E_max/EC₅₀) G_s = 0.48; G_z = −1.64).

D₂-Like Receptor Noncanonical G Protein Activation

G_z-CASE was used as the reference sensor for D_2l,3,4Rs as this sensor produced the highest potency response to dopamine. Dopamine could activate all G_i/o/z-CASE subtypes at D₂-like receptors, with a preference for G_z, followed by G_o1 (Figure 8). For the G_i subtypes, D₃R and D₄R both showed selectivity for G_i3 > G_i2 > G_i1 (Δlog(E_max/EC₅₀) D₃R = 0.11, −0.60, −0.96, D₄R = −0.77, −0.79, −1.19, respectively), whereas the D_2lR had a greater response for G_i1 > G_i2 > G_i3 (Δlog(E_max/EC₅₀) D_2lR = −0.47, −0.57, −1.15, respectively). Dopamine could activate G_q-CASE at all three receptors. G_s-CASE activation by dopamine was weak at D_2lR and D₄R, and not detectable at the D₃R up to 10 μM (Δlog(E_max/EC₅₀): −2.08, −2.14, respectively). The selective ligand pramipexole could not activate G_s at any of the D₂-like receptors, but produced an increase in Δlog(E_max/EC₅₀) compared to dopamine for most G_i/o/z subtypes (around 5-fold increase, excluding G_o1,i1 for D₃R and G_z,i3 for D₄R; Figure 9). (+)-SKF-81297 could activate G_s-CASE, G_o1-CASE, and G_z-CASE for all D₂-like receptors, although no Δlog(E_max/EC₅₀) value could be calculated for the G_s-CASE with D₃R as the reference ligand dopamine did not produce a response up to 10 μM. At the D_2lR, there was activation of G_i1,i2,q-CASE using (+)-SKF-81297 that was unique between the D₂-like receptors.

Figure 8.

G protein selectivity profile of D₂-like receptors determined by the G-CASE assay. The endogenous ligand dopamine, D₂-like selective pramipexole, and D₁-like selective (+)-SKF-8129 are assessed for ligand-induced G protein selectivity. (A) Web of selectivity at D_2lR for the eight different G-CASE constructs (G_i1, G_i2, G_i3, G_o1, G_z, G_s, G_q, G₁₃) using Δlog(E_max/EC₅₀) relative to the dopamine/G_z response. (B) pEC₅₀ and (C) E_max are shown separately as heatmaps, where 100% E_max is taken as the response to dopamine for each sensor. (D) Web of selectivity at the D₃R for the eight different G-CASE constructs (G_i1, G_i2, G_i3, G_o1, G_z, G_s, G_q, G₁₃) using Δlog(E_max/EC₅₀) relative to the dopamine/G_z response. (E) pEC₅₀ and (F) E_max are shown separately as heatmaps, where 100% E_max is taken as the response to dopamine. (G) Web of selectivity at D₄R for the eight different G-CASE constructs (G_i1, G_i2, G_i3, G_o1, G_z, G_s, G_q, G₁₃) using Δlog(E_max/EC₅₀) relative to the dopamine/G_z response. (H) pEC₅₀ and (I) E_max are shown separately as heatmaps, where 100% E_max is taken as the response to dopamine for each sensor. Data are expressed as the means of n = 5 independent experiments, each performed in triplicate. DA = dopamine, PR = pramipexole, SKF = (+)-SKF-81297. Note the logarithmic scale of the web axes. White spaces without numbers in the heatmaps indicate no measurable response up to 10 μM ligand, apart from *no E_max could be calculated for the G_s sensor with (+)-SKF-8129 at the D₃R due to no response to dopamine up to 10 μM.

Figure 9.

Calcium mobilization of the five dopamine receptors in HEK293A cells. HEK293A cells were transiently transfected with dopamine receptor or empty plasmid vector with or without Gα_q. (A) Concentration–response curves of calcium ion influx for D₁-like and D₂-like receptors with wildtype or overexpressed Gα_q. The bar on the left represents the 1 μM ionomycin response (100%) used to normalize the data. (B) E_max and pEC₅₀ values from each individual experiment used to create the response curves in panel (A), with differential Gα_q expression paired by the experimental day/transfection. Repeated measures one-way ANOVA was used to determine if E_max (F(11, 44) = 8.42, p < 0.0001) or pEC₅₀ values (F(11, 44) = 1.63, p = 0.12) significantly differed between wildtype and overexpressed Gα_q. Šídák’s multiple comparisons test concluded that overexpression of Gα_q increased the E_max of the calcium response for D₁R (*p = 0.0005) and D_2lR (**p = 0.0009). n = 5 individual experiments from different transfections on different days, carried out in triplicate. DA = dopamine, wt = wildtype, oe = overexpressed.

Calcium Mobilization of Dopamine Receptors

The coupling of mG_q to the D₁R-NlucC and D_2lR-NlucC and dissociation of G_q-CASE with all subtypes prompted the investigation of the downstream calcium ion mobilization of the five dopamine receptors with the endogenous agonist dopamine. HEK293A cells were transiently transfected with the wildtype dopamine receptor constructs used in the CAMYEN cAMP assay, with or without the overexpression of Gα_q, and loaded with 1 μM Fluo-8 AM calcium dye, which increases in fluorescence intensity when bound to calcium ions. To selectively block native adrenoceptors and serotonin receptors that should be present in the cells and may bind dopamine, 1 μM propranolol was added at the same time as the dye and remained throughout the experiment.

Concentration-dependent calcium responses to dopamine were detected with D₁R and D₅R, both with and without overexpression of the Gα_q subunit (Figure 9). The D_2lR could only produce an influx of calcium in response to dopamine activation when the Gα_q subunit was also overexpressed in the HEK cells. For the D₃R and D₄R, the efficacy of calcium mobilization was similar to wild-type cells without dopamine receptor expression. Coupling and activation of G_s can cause a downstream calcium response; therefore, the statistical significance of the addition of overexpressed Gα_q was tested for both the E_max and pEC₅₀ generated in each calcium experiment to determine the impact of Gα_q expression in the system (Figure 9). Both the D₁R and D_2lR caused a significant increase in the efficacy of the calcium response detected in the cells when Gα_q was overexpressed, suggesting that coupling to the G_q is at least partly responsible for the calcium mobilization with dopamine for these receptors in HEK293A cells. No change in pEC₅₀ was discovered for any receptor. Kinetic response curves are provided in Figure S29.

Discussion

Canonical G Protein Signaling

The three bioluminescence-based assays used in this study have highlighted important pharmacological responses of the five dopamine receptor subtypes. Functional characterization of the dopamine receptors through their canonical pathways have been performed using the CAMYEN cAMP assay and miniG protein recruitment. All agonists (dopamine, pramipexole, (+)-SKF-81297, (−)-apomorphine, and (−)-quinpirole) have been successfully characterized for both acute receptor activation with the canonical miniG proteins and downstream cAMP response. The assays are suitable for detecting ligand potency, efficacy, and selectivity. Either methodology can be used for basic functional characterization of dopamine ligands.

Antagonist Canonical Response

As expected, haloperidol, (+)-SCH-23390, and (+)-butaclamol acted as antagonists at all dopamine receptors in the miniG recruitment assay. (+)-Butaclamol acted as an inverse agonist in agonist mode and for cAMP production. Confirming our results, (+)-butaclamol has previously been reported to reverse the D_1,5Rs confirmation from the active to inactive form,^47,48 and has shown to be an efficacious inverse agonist at D₂-like receptors.^48,49 (+)-SCH-23390 is known as a selective D₁-like receptor antagonist. Intriguingly, (+)-SCH-23390 could act as a partial agonist by itself at the D_1,2l,4,5Rs and was unable to completely block miniG_s coupling in antagonist mode, noted before in other assays.^47,48 Affinity of (+)-SCH-23390 for 5HT_2A receptors could cause G_q responses,⁵⁰ reflecting the cAMP increase by D₁-like receptors; however, this would not account for the agonist effect of (+)-SCH-23390 at the D_2l,₄Rs in our cAMP assay. We therefore hypothesize that (+)-SCH-23390 acts as a partial agonist at the D_1,2l,4,5Rs, but not the D₃R, creating an interesting prospect for selective drug design.

Comparison between cAMP Generation and Canonical G Protein Recruitment

Between the assays, higher potencies were always observed in the CAMYEN assay than with miniG protein recruitment due to the more proximal position of G protein recruitment to ligand binding than the cAMP response further downstream in the signaling cascade.^1,35 Of interest, while (+)-SKF-81297 produced a partial response at the D₁R in the cAMP and G_s-CASE assays, there was a full agonist response with the miniG_s. This partial response could be due to a potent β-arrestin recruitment by (+)-SKF-81297, measured in previous studies, causing internalization and reducing G protein activation at the membrane and when the receptor is recycled.^45,51 The miniG_s recruitment was perhaps not affected because of the higher affinity of miniG_s, lowering the turnover of the sensor when tightly bound to the receptor and obstructing arrestin interaction.¹ Our theory represents only one postulation, and indeed other factors such as G protein subunit promiscuity and off target effects may also be involved in the difference between the assay results.

Compared to the CAMYEN assay and literature-reported values, reduced dopamine responses were seen with the D_3,4Rs in the miniG recruitment assay. With the D₃R, this was unexpected as the D_2lR_miniG with dopamine, D₃R_mG_o1 with all other agonists, and the cAMP responses with dopamine at the D₃R produced values comparable to literature data. We theorize that the NlucC fusion on the C-terminus of the D₃R is impacting the receptor activation in a way that is ligand- and receptor-dependent. Currently, there is no structure of the D₃R with dopamine bound, making it difficult to hypothesize the exact structural significance of these data. For the D₄R, where all ligands except quinpirole failed to produce full agonist responses, we propose the low dopamine response is because the D₄R-NlucC had a 25-fold lower expression than the nonmodified receptors (radioligand saturation, Figures S1, S2, and Table S1). Speculatively, the NlucC fragment on the C-terminus could interfere with D₄R trafficking and cell membrane expression, but not the similar D_2lR. Palmitoylation of both the D₃R and D₄R C terminus regions has been shown to be more extensive and crucial than the D_2lR.⁵²⁻⁵⁴ The D₃R has also been shown to interact with the GPCR-regulating protein GIPC via the C-terminus regions,⁵⁵ protecting the receptor from degradation. There may also be other factors involved in the differing dopamine activity with C-terminus modification of the D₃R-NlucC and D₄R-NlucC, such as a decreased/enhanced ability for miniG binding that could be ligand-dependent. Lower responses for both these receptors could also be compounded by the reduced presence of the mG_o1 sensor (ELISA; Figure S10). To the best of our knowledge, this is the first report of C-terminus modifications for an assay affecting the dopamine response at the D_3,4Rs, and the importance of the C-terminus in their signaling and expression is of interest for future studies.

Comparison of Individual Dopamine Receptor G Protein Subunit Coupling

Aside from canonical coupling, patterns of G protein recruitment are important even within subfamilies of Gα subunits where the downstream messenger is the same, as subtle differences in the effectors and kinetics of the response cascade are present.⁵⁶⁻⁵⁸ Targeting G protein selectivity in drug development can therefore lead to useful physiological responses.⁵⁹⁻⁶² To investigate dopamine receptor selectivity of G protein subunits, the miniG recruitment assay was used in combination with the G-CASE G protein activation assay. G-CASE α and γ subunits showed no difference in expression level; however, direct comparison between the assay sensors is not possible due to inherent differences in signal windows for the different constructs due to small spatial changes in the position of the Nluc tag within the α subunit and differences in the γ subunits used within each sensor (Figures S10–S19).

G Protein Selectivity of the Endogenous Ligand Dopamine

G_z and Dopamine

Dopamine coupled to various G protein subtypes in our miniG and G-CASE assays. Of the inhibitory G protein subunits, G_z-CASE had both the highest potency and efficacy when activated by dopamine binding of all five receptors. G_z couples to both D₂-like⁶³⁻⁶⁷ and D₁-like^62,66,68 families, making it an interesting drug target. Abundantly expressed in the brain, but elsewhere limited to platelets and pancreatic islet cells,⁶⁹ G_z protein activation is thought to be involved with the response to psychoactive drugs, circadian rhythm, brain development, and reward systems, although the subunit remains under-researched.⁷⁰⁻⁷² G_z activation has been described before in G protein selectivity with the D_2lR and D₃R, though these previous experiments have reported a higher activity of the receptors with G_o proteins.^67,73 This discrepancy is likely due to the difference in kinetics between the G_z and G_o proteins, as G_z is much slower in GTPase catalytic activity than other inhibitory G protein subunits.⁷⁴

G_s and G_i/o Coupling by Noncanonical Dopamine Innervation

Along with G_z, dopamine could recruit and activate inhibitory subunits at D₁-like receptors. D₅R has been previously shown to recruit the G_z protein²⁷ and D₁R with the G_o proteins.⁷⁵ The inhibitory G protein coupling may therefore have physiological relevance, such as in renal proximal tubules.⁷⁶ Conversely, for D₂-like receptors, while there was no indication of coupling between G_s and D₃R for miniG_s or G_s-CASE up to 10 μM, the D_2l,4Rs showed a low-potency activation of the G_s-CASE. To the best of our knowledge, there is no known cell type or physiological relevance known for the impact of this coupling. Recently published bias studies by Hauser and Inoue have shown no coupling between G_s and D₂-like receptors.^26,66

Possible G_q Activation and Calcium Mobilization by Dopamine

G_q coupling (mG_q) and/or dissociation (G_q-CASE) were also detected at all dopamine receptors. Our calcium mobilization assay successfully linked overexpressed Gα_q with increased dopamine receptor calcium influx efficacy for D₁R and D_2lR. In agreement with our findings, D₁-like receptors have been known to couple to G_q subunits.^{26,27,77−79} This coupling appears to be important in both the cerebral cortex and striatum, although the exact impact of this activation remains unclear. For the D_2lR, this is in contrast to results using a different G protein BRET-based assay where no G_q activation was reported, potentially due to a change in βγ subunit expression compared to our G-CASE assay.^67,73,80 Physiologically, studies looking at heterodimers of D₁R and D₂R or astrocytic D₂Rs have also seen calcium influx and potential G_q coupling.^81,82 Although not proved in this study, there is still the potential that the other dopamine receptors may produce a downstream response through G_q as calcium responses were detected if not significantly altered by Gα_q overexpression or different to wild-type HEK293A cells. Indeed, D₃R has also been previously reported to show coupling with G_q in some cell types.^79,83,84 Further assays with knockout ΔGα cells, G protein inhibitors, initisol-1-phosphate detection, and potentially other cell types would be required to confirm this. Investigation into the potential for ligand-based selectivity of G_q would also be interesting as pramipexole appears to decrease G_q activation in D₁-like receptors and increase in D₂-like receptors in relation to dopamine in our G-CASE assay.

Native G Protein and Dopamine Receptor Expression

To understand the impact of the G protein coupling profiles with the endogenous ligand dopamine, it is important to consider the expression and abundance of the subunits and receptors in different cell populations. Of the G protein subunits used in our assays, the Gα_s, Gα_i3, Gα_q, and Gα₁₃ are thought to be ubiquitous, with the Gα_i1 and Gα_i2 also widely spread.⁸⁵ Both Gα_z and Gα_o1 are found primarily in neuronal tissue, and Gα_z can also be found in platelets and pancreatic islet cells.⁸⁶ Dopamine receptor expression of all five subtypes is highest in the brain, particularly in the basal ganglia.⁸⁷ The basal ganglia similarly expresses all G protein subunits investigated in this paper. However, the expression level of these G protein subunits will vary when compared to another area of the brain, such as the hypothalamus, or body, such as the stomach (where the D₅R is expressed at a high level), or cell type, where Gα_q has higher expression in astrocytes than neurons, for example (Table S3). Specialized studies into the abundance of receptor and G protein at a single cell level would be of great value to help understand the impact of promiscuous G protein coupling of the dopamine receptors.

G Protein Coupling of Selective Ligands

The D₁-like receptor agonist (+)-SKF-81297 produced some interesting new insights into G protein specificity. For D₁-like receptors, (+)-SKF-81297 activated/recruited the same subunits as dopamine, except for G_q,i1-CASE at the D₅R. With the D₂-like receptors, (+)-SKF-81297 at the D_2l,3Rs activated G_s-CASE at a higher potency than dopamine (shown previously possible for D₃R with G_s in CHO and COS-7 cell lines^63,88), although they both failed to recruit the mG_s. Along with an increase in G_s activation, the D_2lR also retains some potency for the G_i/o/z-CASE and mG_i1/o1 with (+)-SKF-81297, hence creating a low-efficacy agonistic cAMP response in the CAMYEN assay. Alternatively, (+)-SKF-81297 does not activate G_i subunits at the D₃R nor recruit the mG_i1/o1, and only weakly activated the G_z/o1-CASE. Therefore, the result in the cAMP assay where no agonist response was detected could be due to a more dominant G_s recruitment. Future studies could assess the impact of this through knockout cells and G protein inhibitors to tease out the G_s response. D₄R did not couple to G_i in any assay in response to (+)-SKF-81297 activation and displayed a partial, low-potency agonistic cAMP decrease, likely via the recruitment of G_z/o1, as detected with G-CASE.

While (+)-SKF-81297 increased G_s-CASE activation at all receptors (except D₄R), the D₂-like receptor agonist pramipexole recruited and activated G_i/o/z proteins. These data suggest that the selective dopamine ligands show an inherent bias in G protein recruitment, although the exact structural mechanism is speculative. Structural insights into the D₁R and D₂R were investigated by Zhuang et al.⁵¹ The steric clash between the D₂R transmembrane helix 6 (TM6) and α5 helix in the G_s protein impairs G_s protein recruitment. However, interacting residues with (+)-SKF-81297 in the TM5 are conserved between the receptors (Figure 10). It is therefore feasible that (+)-SKF-81297 could induce enough conformational change in the TM domains to overcome the steric hindrance of TM6 and accommodate G_s binding. The position of interacting residue His^6.55 in the D₄R is also altered compared to the D_2,3Rs due to the change of Tyr^7.35 to Val^7.35, perhaps explaining why there is reduced G_s activity at the D₄R with (+)-SKF-81297 compared to the D_2,3Rs.⁸⁹ For pramipexole, G_i protein binding was facilitated through the gap between TM3 and TM5/6 in the D₃R;⁸⁹ therefore, binding of pramipexole to TM3 in D_1,5Rs could help with this interaction, causing the increase in G_i/o/z subunit recruitment and activation (Figure 10). In summary, structural studies into these phenomena would be of interest to understand the apparent ligand-induced selectivity of G protein binding at the dopamine receptors.

Figure 10.

Interacting residues between the dopamine receptors and (+)-SKF-81297 or pramipexole. Important transmembrane helix residues are outlined in bold. For (+)-SKF-81297, interacting residues in the TM5 are conserved between the receptors (Ser^5.42 and Ser^5.46), Trp^7.43 in TM7 of the D_1,5Rs is changed for a similar bulky residue, Tyr^7.43, in the D_2,3,4Rs, and the TM6 interacting residue Asn^6.55 in the D₁R is replaced with His^6.55 in the D_2,3,4Rs, still allowing for hydrogen bonding with the ligand. Tyr^7.35 becomes Val^7.35 in the D₄R, changing availability for His^6.55. Pramipexole also interacts with His^6.55 and the three interacting residues in TM3, Asp^3.32, Cys/Ser^3.36, and Thr^3.37, are all conserved between receptor subtypes. Sequence alignment was performed using GPCRdb.^96,97.

G-CASE vs MiniG Recruitment Results

There were several differences between the miniG recruitment and G-CASE activation. One hypothesis is that this is due to the loss of the GTPase region in miniG proteins. Recently, Jang et al.⁹⁰ described a model whereby the GTPase activity of the G protein is a critical determinant of its coupling selectivity, perhaps more than the structural compatibility between the GPCR and the G protein. However, it has also been shown that some apparent subunit “dissociation” in G₁₂ BRET assays similar to the G-CASE shown here are in fact not effective to downstream signaling and may simply be subtle movements of Gα relative to Gγ.⁹¹ Tagging positions of the Gα subunits are also different between the sensors, which may change the true sensor dissociation range; for example, when the Gα-Gβγ interface opens under ligand stimulation of the receptor but only rotates the Gα subunit, not causing full dissociation.⁹²

We therefore suggest that subunit dissociation, G protein recruitment, and secondary messengers are important to assess the receptor–ligand response as no method alone produces a full profile of G protein responses.

Study Limitations

Although we have been thorough in our characterization of the five dopamine receptors with these assays, some Gα subtypes and common receptor isoforms were not included in this study due to time constraints. Responses to G_olaf, G_oB, other G_q family subunits, G_11/14/15, and the D_2shortR or known D₄R isoforms would be of great interest. Along with investigating G protein activation of ligands specific for the two different subclasses (SKF-81297 for D₁-like and pramipexole for D₂-like), it would also be advisable to use selective compounds for the individual receptors, although it is particularly difficult to find subtype-selective ligands for the D₂-like receptors (D₂R, D₃R, and D₄R). Nevertheless, it would certainly be interesting to investigate compounds such as benperidol, 7-OH-DPAT, and aripriprazole.⁹³⁻⁹⁵ Future studies could also investigate the impacts of natively expressed G proteins in the recombinant cells by using CRISPR/Cas9 knockout cell lines and G protein inhibitors with even more downstream assays. For our cell systems, receptor expression was comparable but not identical, which may alter the potencies and signal windows of our bioluminescence assays. As discussed, the C-terminus modification of a fused NlucC protein used in the miniG recruitment assay could also have an impact on the dopamine receptor expression, binding and signaling in subtle ways that will require further examination. It is also necessary to confirm findings first discovered in overexpressed recombinant systems, such as the HEK293 cells used here, with endogenous protein expression, primary cells, and animal models to aid with translational efforts.

Conclusions

In conclusion, this study has successfully characterized the five dopamine receptors with both canonical and selective G protein activation. We have demonstrated that it is important to undertake full ligand pharmacological characterization to understand the acute and downstream effects. G protein selectivity by dopamine receptors is a complex and interesting target for drug discovery, for which our assay systems represent valuable and important tools.

Glossary

Abbreviations

ADHD

attention hyperactivity deficit disorder

Apo

R-(−)-apomorphine

ASD

autism spectrum disorder

BRET

bioluminescence resonance technology

BSA

bovine serum albumin

Buta

(+)-butaclamol hydrochloride

cAMP

cyclic adenosine monophosphate

CAMYEN

cAMP sensor using YFP-Epac-Nluc

cDNA

complementary DNA

CNS

central nervous system

cpVenus

circular permutation Venus

CZH

coelenterazine H

dH₂O

distilled water

DMEM

Dulbecco’s modified Eagle’s medium

Dopa

dopamine

(h)DxR

(human) dopamine Dx receptor

ELISA

enzyme-linked immunosorbent assay

Epac

exchange protein activated by cAMP

FCS

fetal calf serum

G-CASE

G protein-based, tricistronic activity sensor

GPCR

G protein-coupled receptor

Halo

haloperidol

HEPES

4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid

L-15

Leibovitz’s L-15 medium without phenol red

miniG or mG

minimal G protein

NanoBiT

nanoluciferase binary technology

NanoLuc or Nluc

nanoluciferase

Nemo

nemonapride

NlucC or SmBiT

small C terminus part of the split nanoluciferase

NlucN or LgBiT

large N terminus part of the split nanoluciferase

PBS

phosphate buffered saline

PD

Parkinson’s disease

PEI

linear polyethylenamine

Prami

pramipexole dihydrochloride

Quin

(−)-quinpirole hydrochloride

SCH

(+)-SCH-23390 hydrochloride

SKF

(+)-SKF-81297 hydrobromide

Spip

spiperone hydrochloride

Data Availability Statement

All data used in the present study will be available by the corresponding authors upon reasonable request.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsptsci.3c00339.

• Radioligand binding experiments (Figures S1, S2 and Table S1); kinetic responses of the CAMYEN cAMP assay (Figures S3–S9); ligand pEC₅₀/pK_b of miniG recruitment and cAMP stimulation/inhibition (Table S2); NlucN-miniG sensor expression by ELISA (Figure S10); evaluation of receptor/G-CASE expression by immunofluorescence and ELISA (Figures S11–S20); concentration–response curves for G protein selectivity experiments (Figures S21–S28); calcium mobilization kinetics (Figure S29); miniG_i1 experiments varying the D_2lR-NlucC expression level (Figure S30); and gene expression levels of dopamine receptors and G protein subunits in basal ganglia, hypothalamus, and stomach (Table S3) (PDF)

Author Contributions

^§ D.M. and L.J.H. contributed equally to this work. D.M., L.J.H., B.L.H., and S.P.: conceptualization; D.M., L.J.H., C.H., and L.F.: data curation; D.M., L.J.H., C.H., and L.F.: formal analysis; S.P.: funding acquisition; D.M., L.J.H., C.H., and L.F.: investigation; D.M., L.J.H., and B.L.H.: methodology; L.J.H. and S.P.: project administration; S.P.: resources; L.J.H. and S.P.: supervision; D.M., L.J.H., and S.P.: visualization; D.M. and L.J.H.: writing—original draft; D.M., L.J.H., C.H., B.L.H., L.F., and S.P.: writing—review and editing.

L.J.H. and C.H. were funded by the Deutsche Forschungsgemeinschaft (DFG, Research Training Group GRK 1910). S.P. was supported by the Fonds der Chemischen Industrie (No. 661688).

The authors declare no competing financial interest.

Notes

The graphical abstract and Figure 1 were created using BioRender.com.