Dopamine Negatively Regulates Insulin Secretion Through Activation of D1-D2 Receptor Heteromer

Pancreatic β-cells secrete insulin, which plays a central role in glucose homeostasis in the body. Insulin secretory response and glucose metabolism are tightly coupled and regulated by a complex mechanism. In β-cells, insulin granule exocytosis is initiated by the transport of glucose into β-cells. Glucose-stimulated insulin secretion (GSIS) comprises a transient first phase and sustained second phase (1). Glucose metabolism leads to ATP generation, promoting the closure of K_ATP channels on the plasma membrane, followed by cell depolarization and opening of the voltage-dependent calcium (Ca²⁺) channels. Subsequently, Ca²⁺ influx triggers the rapid first phase of insulin granule exocytosis (2,3). The second phase involves amplifying the triggering pathway in which metabolic signals other than ATP are involved (4,5). Hormones and neural transmitters modulate GSIS. A G-protein–coupled receptor (GPCR) activates the downstream signaling events, which involve cAMP, diacylglycerol, and inositol 1,4,5-triphosphate (6). cAMP potentiates insulin granule exocytosis through protein kinase A– and Epac2A-dependent mechanisms (7–10).

Dopamine (DA) is a monoamine neurotransmitter identified in neuronal cells and nonneural cells. DA receptors are members of the GPCR superfamily and are divided into two major classes: D1 and D2. The D1 class receptors couple to the G_sα proteins and stimulate adenylyl cyclase, whereas the D2 class receptors couple to G-_i/oα proteins and inhibit adenylyl cyclase (11). Evidence shows that DA-D2 receptor signaling functions as a negative regulator of GSIS in β-cells (12–18). A mouse model of global disruption of D2 receptor exhibited impaired glucose tolerance and diminished β-cell mass (19). β-Cell–selective D2 mutant mice exhibit postprandial hyperinsulinemia (20). Pancreatic β-cells possess the machinery to synthesize, convert, store, and catabolize monoamines, including DA (15). β-Cells express tyrosine hydroxylase (TH), convert l-tyrosine to dihydroxyphenylalanine (DOPA), express aromatic l-amino acid decarboxylase (AADC) that converts l-DOPA to DA, and express vesicular monoamine transporter 2 (VMAT2) that uptakes monoamines and prevents their degradation by monoamine oxidase (15,21). The administration of DA or l-DOPA increases islet DA content and inhibits GSIS in vitro and in vivo (16,22). Treatment of mouse islets with a VMAT2 inhibitor, tetrabenazine (TBZ), led to a depletion of total pancreatic DA, which not only enhanced GSIS (21) but also potentiated mouse pancreatic β-cell differentiation (23). β-Cell–specific VMAT2 mutant mice showed reduced DA content and enhanced GSIS in vitro (24). DA-D2 receptor signaling potentiates dedifferentiation in primary β-cells through heteromer formation with adenosine A2 receptor (25).

GPCR heteromer formation modulates receptor function (26). D2 acts as a hub receptor and interacts with other GPCRs (27). Receptor heteromer formation leads to novel receptor dynamics during which the receptor changes its target recognition and downstream signaling (28). The activation of the D1-D2 heteromer by its selective compound leads to the transduction of a novel signaling pathway with unique functional properties (29–31).

Total internal reflection fluorescence microscopy (TIRFM) is a tool used to analyze the spatial-temporal dynamics of insulin granule exocytosis. Using TIRFM, Ca²⁺ imaging, and cAMP imaging, we investigated the molecular mechanism of DA regulation of GSIS in β-cells. DA is reported to colocalize with insulin in insulin vesicular granules, cosecreted with insulin at GSIS, and mediates negative regulation of GSIS (15,17). Here, we found that D1 antagonists rescued the DA-mediated inhibitory effect on glucose-stimulated Ca²⁺ influx. We investigated the underlying mechanism and found that the D1 receptor is involved in a DA-mediated inhibitory effect on GSIS through the formation of a D1-D2 heteromer. DA facilitates D1-D2 heteromer complex formation and transiently inhibits insulin secretion. Through our findings, we sought to determine the role played by D1 through interaction with D2 to modulate the robust inhibitory and toxic signaling mediated by D2.

All studies involving animals were performed according to local guidelines and regulations and were approved by the institutional committee for animal research at the Tokyo Institute of Technology.

TBZ, DA, domperidone, and haloperidol were purchased from Tocris Bioscience; α-methyl-dl-tyrosine methyl ester (AMPT) and GBR12909 from Sigma-Aldrich; SCH23390 from Abcam; SKF38393 and SKF83566 from R&D Systems; SKF83959 from Cayman Chemical; and tolbutamide and forskolin from Fujifilm Wako Pure Chemical Co. The compounds were dissolved in DMSO. Pertussis toxin (PTX) from List Biological Laboratories was dissolved in water.

The following antibodies were used: mouse antiglucagon (1/1,000, G2654; Sigma-Aldrich); guinea pig anti-insulin (1/10, A0564; Dako); rabbit anti-TH (1/1,000, P40101-0; Pel-Freez); goat anti-AADC (1/1,000, AF3564; R&D Systems); rabbit anti-DA (1/100, IS1005; ImmuSmol); rabbit anti-D2 (1/100, 376203; Synaptic Systems GmbH); goat anti-D1 (1/100, sc-31479; Santa Cruz); mouse antisynaptophysin (1/100, ab8049; Abcam); Alexa Fluor 568 donkey anti-rabbit IgG (1/1,000, A10037) or donkey anti-goat IgG (1/1,000, A11057; Invitrogen); Alexa Fluor 488 donkey anti-guinea pig IgG (1/1,000, 706-546-158), Alexa Fluor 647 donkey anti-rabbit IgG (1/1,000, 711-606-152), or donkey anti-mouse IgG (1/1,000, 715-606-150; Jackson ImmunoResearch Laboratories); and DAPI (Roche Diagnostics). Images were acquired with an Axio Observer Z1 and LSM 780 (Zeiss) or an ImageXpress microscanning system and MetaXpress image analysis software (Molecular Devices).

C57BL/6 mice were purchased from Charles River Laboratories Japan and bred in a 12-h light-dark cycle. Islets from 10-week-old male mice were isolated (Sakano et al., 2016). Isolated islets were dissociated with Accumax (Innovative Cell Technologies, Inc.) at 37°C for 5 min and cultured in DMEM (high glucose 4.5 g/L [25 mmol/L]) supplemented with 10% FBS, 100 μmol/L nonessential amino acids, 2 mmol/L l-glutamine, 50 units/mL penicillin, 50 μg/mL streptomycin, and 100 μmol/L 2-mercaptoethanol at 37°C in 5% CO₂.

The mouse D1 and D2 receptor fragments were cloned from mouse islet cDNA and fused with 3× FLAG tag or 6× His tag at the N terminus, respectively. Human TH (hTH) was cloned from human islet cDNA and fused with T2A-mCherry cDNA. These gene cassettes and cAMP indicator Pink Flamindo (32) were inserted into the multiple cloning site of the pENTR1A vector (A10462; Invitrogen), respectively. Then these cassettes were transferred to pAd/CMV/V5-DEST vector (493-20; Invitrogen) using LR recombination reaction by the Gateway LR Clonase II Enzyme Mix (Invitrogen). pENTR-preproinsulin-Venus (33) was also transferred to the pAd/CMV/V5-DEST vector.

MIN6 infected with adenovirus carrying hTH-T2A-mCherry or mCherry were lysed with sample buffer (0.1 mol/L Tris-HCl [pH 6.8], 4% SDS, 20% glycerol) with 12% 2-mercaptoethanol and boiled at 95°C for 5 min. One microgram of samples were applied to 8% polyacrylamide gel for SDS-PAGE, then blotted onto a polyvinylidene difluoride membrane (Millipore) and probed overnight at 4°C with rabbit anti-TH antibodies (1/1,000). The dissociated mouse primary islets were seeded at 1.0 × 10⁶ cells per well on six-well plates and infected with adenovirus carrying 3× FLAG-mD1 and/or 3× FLAG-mD2 or 6× His-mD2. Islet cells were lysed with 1% Triton X-100 in 0.1 mol/L PBS, and then lysates were mixed with a sample buffer with 12% 2-mercaptoethanol and incubated at 37°C for 10 min. Samples of 100 ng or 25 ng were loaded with 7.5% polyacrylamide gel for SDS-PAGE. The primary antibodies used were mouse anti-D1 antibodies (1/1,000, ab78021; Abcam) or rabbit anti-D2 antibodies (1/5,000). The secondary antibodies used were horseradish peroxidase–conjugated goat anti-rabbit IgG (H+L) (1/10,000, 12-348, Merck Millipore) and goat anti-mouse IgG (H+L) (1/10,000, AP308P; Merck Millipore). Membranes were reacted with Immobilon Forte Western HRP Substrate (Merck Millipore) and scanned using FUSION Solo 4S WL (M&S Instruments).

Islet cells were treated with chemicals indicated in the figures. Cells were lysed with lysis buffer containing 3% Triton X-100 (Nacalai Tesque) in 0.1 mol/L PBS. Lysates were assayed using the DA ELISA Kit (ImmuSmol) according to the manufacturer’s instructions. Data were normalized, with protein contents measured using a Micro BCA Protein Assay Kit (Thermo Fisher Scientific).

The dissociated islets cells were seeded at 1.0 × 10⁵ cells per dish on 35-mm dish with a 5mm-hole in the center, glass-bottom of 0.16–0.19 mm thickness, refractive index n=1.52 (Matsunami) coated with Easy iMatrix-511 (Nippi) at 0.25 μg/cm². The cells were infected with adenovirus carrying insulin-Venus. In the experiment of TH overexpression, to express TH, an aliquot of the dissociated cells was infected by adenovirus carrying hTH-T2A-mCherry for 1 h at 37°C, then thoroughly washed out. The TH-infected islet cells were mixed with uninfected islet cells and seeded onto the dish. Forty-eight hours after infection, the cells were preincubated in low-glucose (2.5 mmol/L) Krebs-Ringer bicarbonate HEPES buffer (KRBH) (133.4 mmol/L NaCl, 4.7 mmol/L KCl, 1.2 mol/L KH₂PO₄, 1.2 mmol/L MgSO₄, 2.5 mmol/L CaCl₂, 5.0 mmol/L NaHCO₃, 10 mmol/L HEPES [pH 7.4], and 0.2% BSA) at 37°C in a Stage Top Incubation System (Tokai Hit). After preincubation, glucose and various agents were added to the dish at the indicated final concentrations. Observation of exocytosis of β-cells was performed by using a total internal reflection system with an inverted microscope (IX83; Olympus). Dishes were set on the stage and observed with an objective lens (UAPON100×HOTIRF numerical aperture = 1.49; Olympus) by type F immersion oil (Olympus). The focus was maintained during image acquisition with IX3 Z-Drift Compensator (Olympus). Cells were excited by a 488-nm argon laser at an incident angle of 73.5°, a calculated penetration depth of 65.1 nm. Images were collected by Zyla sCMOS camera (Andor) and acquired every 500 ms, which was operated with MetaMorph image analysis software (Molecular Devices).

The dissociated islet cells were seeded on a 384-well plate. The cells were infected by adenovirus carrying Pink Flamindo (32). After a 48-h culture, the cells were incubated in low-glucose KRBH for 5 min. The transition of cytosolic cAMP induced by forskolin and indicated chemicals were then monitored with inverted microscopy as mentioned above. Images were acquired every 3 s using MetaMorph.

The dissociated islet cells were seeded on a 384-well plate coated with Easy iMatrix-511 at a density of 7,500 cells per well. The cells were infected with adenovirus carrying various genes, as indicated in the figures. After culturing for 48 h, the cells were preincubated with 4.6 μmol/mL Fluo-4 AM, 0.25 mmol/L Probenecid, and 0.08% Pluronic F-127 (Dojindo Molecular Technologies) in KRBH for 1 h at 37°C. After preincubation, cells were washed with KRBH, and cytosolic Ca²⁺ was monitored by using an inverted microscope as mentioned above. Images were acquired every 1 s using MetaMorph.

Direct protein-protein interactions were examined by Duolink In Situ Detection Reagents Red (Sigma-Aldrich). After incubation of dissociated islet cells infected by adenovirus for 48 h, cells were fixed with 4% paraformaldehyde, washed, and then permeated. After blocking with 5× diluted Blocking One (Nacalai Tesque), primary antibodies rabbit anti-D2 antibody (376203; Synaptic Systems GmbH) and goat anti-D1 antibody (sc-31479; Santa Cruz) were added for 16 h at 4°C. Primary antibody washout, Duolink Anti-Rabbit PLUS and Anti-Goat MINUS probes (Sigma-Aldrich), ligation reaction, and amplification steps were performed according to the manufacturer’s instruction. Cells were counterstained with anti-insulin antibodies and DAPI. Images were acquired and analyzed with the ImageXpress microscanning system and MetaXpress.

The TUNEL assay was performed using the In Situ Cell Death Detection Kit, Fluorescein (Roche Applied Science).

Acquired image sequences by TIRFM and Ca²⁺ imaging experiments were analyzed with MATLAB (MathWorks). For the analysis of TIRFM images, cells were manually captured as regions of interest in the sum of image stacks, and the regions that had higher values than the threshold of the subtracted intensity (i.e., the change of intensity) were filtered by the area to detect flashes following fusion events. The number of fusion events was collected every 30 s in each cell.

An embryonic stem cell line bearing a targeted mutation at the D1 receptor (Mouse Genome Informatics identifier 99578) (Drd1^{tm1a(KOMP)Wtsi}; #EPD0507_1_H11) (C57BL/6), produced by the Trans-NIH Knockout Mouse Project was used. A DNA cassette composed of flippase recognition target sites, lacZ sequence, and loxP sites was inserted at position 54206198 of chromosome 13 upstream of exon 2. A third loxP site was inserted at position 54209082 downstream of the target gene (Mouse Genome Informatics identifier 4451251). LoxP sites thus flank exon 2. The targeted embryonic stem cells were injected into the perivitelline space of one-cell stage C57BL/6 mouse embryos to produce heterozygous D1 receptor knockout (Drd1^tm1a/WT) mice.

Real-time PCR analysis was done using THUNDERBIRD SYBR qPCR Mix (Toyobo), and reactions were run on the StepOnePlus Real-Time PCR System (Applied Biosystems). The sequences of primers for Drd1 were forward CTGCCCTACAACGAATAATG and reverse CATAGTCCAATATGACCGATAAG. β-actin was used as an internal control (forward GTGATGGTGGGAATGGGTCA, reverse TTTGATGTCACGCACGATTTCC).

Data are presented as mean ± SD, with individual data or medians shown in box-and-whisker plots, except where indicated. Data were analyzed by one-way ANOVA and Dunnett multiple comparisons test, except where indicated. Significant differences were considered as P < 0.05. All data were obtained from more than three independent experiments, except where indicated.

Representative underlying imaging data sets generated and analyzed during the current study are available in the supplementary material that accompanies this article. Other data sets generated and/or analyzed during the current study are available from the corresponding authors upon reasonable request.

To investigate the role of dopamine signals in β-cells, we re-examined the distribution of TH that converts tyrosine into l-DOPA. As previously reported, only a subpopulation of β-cells expresses TH (Fig. 1A, top). In contrast, AADC, which converts l-DOPA into DA, is observed in almost all insulin-expressing β-cells (Fig. 1A, bottom). The immunostaining results of TH and AADC are in agreement with those of previous reports (34,35). l-DOPA is incorporated intracellularly into β-cells by DA transporter (DAT) or L-type amino acid transporters and converted to DA by AADC (16,20). We confirmed that the exogenous addition of DA to isolated islets increased intracellular DA content. The addition of GBR12909, an inhibitor of the DAT (36), abolished the increase in DA levels (Fig. 1B). Therefore, the incorporation of DA also seems to contribute to the redistribution of DA to the neighboring β-cells.

We then examined the subcellular localization of DA in β-cells. DA-positive staining was observed in the cytoplasm of nearly all insulin-expressing β-cells. DA staining overlapped with insulin staining in a dotted pattern (Fig. 1C). DA staining was distinct from that of synaptophysin (Fig. 1D), a marker for synaptic-like microvesicles (37) where nonpeptide neurotransmitters, such as γ-aminobutyric acid, are stored. The results agree with those previously reported that DA localizes in insulin vesicular granules and is cosecreted with insulin in a glucose-stimulated manner.

To investigate the role of DA in the regulation of insulin secretion, we then used TIRFM to visualize the dynamics of insulin granule exocytosis in mouse islet primary β-cells. We transfected a previously reported construct for visualizing insulin granule exocytosis (33). In exocytosis, insulin granules are docked and fused to the membrane. Flash events of Venus revealed the release of insulin from the insulin granules, upon which fluorescence intensity increased as a result of a change in pH. We constructed an analytical program to automatically detect the fusion events of insulin granules using MATLAB. In brief, we calculated the differences between each frame and the extracted flash events by thresholding areas to distinguish docked granules (Supplementary Fig. 1A and B). Glucose challenge (16.7 mmol/L) induced biphasic insulin granule exocytosis (33). Using TIRFM, we detected an immediate-early phase (first phase) within 5 min of glucose challenge and sustained insulin exocytosis (second phase) within 5–20 min of glucose challenge (Supplementary Fig. 1C, left). We then examined the effect of TBZ, a VMAT2 inhibitor, on glucose-stimulated insulin granule exocytosis. As a result, an overnight TBZ treatment did not affect the first phase but increased the second phase of insulin granule exocytosis (Supplementary Fig. 1C and D). Although other monoamines are also substrates for VMAT2, DA is the most abundant in mouse β-cells (23). We then focused on DA to examine its role in β-cells. We attempted to increase DA synthesis in β-cells. Treatment with AMPT, an inhibitor of TH, alone did not affect GSIS (F.U., unpublished observations). We hypothesized that this was due to the limited population of β-cells expressing TH (Fig. 1A) and examined whether TH-expressing cells could exert opposing effects on the GSIS of the neighboring β-cells. We then tried to increase the population of TH-expressing β-cells using an adenoviral vector to overexpress TH fused with T2A-mCherry (TH OE). We first infected MIN6, a β-cell line, with the adenovirus. We detected a 55.6-kDa band that originated from TH OE by Western blot analysis (Supplementary Fig. 1E) and an increase in DA content (Supplementary Fig. 1F) in TH OE MIN6. We then mixed the adenovirus-infected TH-expressing β-cells (coexpressing mCherry) with untransfected β-cells. When mixing with ∼20% of β-cells overexpressing TH, we observed that nearly all β-cell clusters contained at least one TH-expressing cell (Fig. 1E). Using TIRFM, we observed a reduction in the second phase of insulin exocytosis in β-cell clusters containing TH-expressing cells compared with that in the control uninfected β-cells. In contrast, the first phase of insulin exocytosis was not affected (Fig. 1F). Administration of AMPT to the mixture of TH OE and β-cells reversed the decrease in the second phase of insulin exocytosis (Fig. 1F). A quantitative analysis of the fusion events further confirmed these results (Fig. 1G). Our results suggest that without l-DOPA or DA supplies from the blood stream, the low level of endogenous β-cell–derived DA exerts an inhibitory effect on the second phase of insulin granule exocytosis. Exogenous DA addition at 5 min after the glucose challenge confirmed the inhibition of the second phase of insulin exocytosis (Supplementary Fig. 2A and B). We also confirmed that exogenous DA addition before the high-glucose challenge inhibited both the first and second phases of insulin exocytosis (Supplementary Fig. 2C and D). In the absence of an exogenous source of DA or its precursor l-DOPA, islets do not accumulate significant intracellular DA stores (20). We interpreted the result to mean that the amount of endogenously derived DA in the islets determines the extent of its inhibitory effect. Although DA has the capacity to inhibit both the first and second phases of insulin exocytosis, at conditions of low endogenous DA content without supplies from the blood stream, DA seems to exert negative effects only on the second phase of insulin exocytosis.

Using the red fluorescent protein-based cAMP indicator Pink Flamindo, we performed imaging of cAMP dynamics (32). DA inhibited intracellular cAMP accumulation induced by forskolin (an adenylyl cyclase activator) at high glucose. PTX, a G-_i/oα-protein blocker, reversed the DA-mediated cAMP inhibition (Fig. 2A).

We then performed Ca²⁺ imaging using Fluo-4 am, a cell-permeable fluorescent Ca²⁺ indicator, to monitor cytosolic Ca²⁺ accumulation. Images were acquired every second. Glucose challenge triggered increased Ca²⁺ flux in controls (DMSO), which was inhibited dose dependently by DA (Fig. 2B and C). PTX treatment partially reversed the inhibitory effect of DA on Ca²⁺ flux (Fig. 2D and E). The result suggests that DA-mediated inhibition of the Ca²⁺ flux functions partially through a G-_i/oα-coupled DA receptor that possibly involves G_βγ activation.

We also confirmed DA-mediated inhibition of Ca²⁺ flux in the presence of tolbutamide, an inhibitor of the K_ATP channel (38) (Fig. 2F). The Ca²⁺ flux was confirmed to be inhibited by nifedipine, a voltage-gated Ca²⁺ channel blocker (Fig. 2G). The results suggest that DA-mediated inhibition of Ca²⁺ flux acts downstream of the K_ATP channel and partly through G-_i/oα to reduce the cAMP level and/or possibly through a G_βγ-mediated pathway.

We further tested whether DA receptor antagonists could counteract the inhibitory effects of DA on insulin exocytosis in β-cells. Pancreatic β-cells express D2 receptors that transduce negative regulatory signals for GSIS, and D2 receptor disruption causes glucose intolerance (19). Unexpectedly, the addition of D1 antagonist SCH23390 (39), but not D2 antagonist domperidone, counteracted the inhibitory effect of DA on glucose-stimulated Ca²⁺ accumulation (Fig. 3A). Dose-dependent studies revealed that SCH23390 (1 × 10⁻⁶ mol/L) significantly increased Ca²⁺, even in the absence of DA (Fig. 3B). The result suggests that DA signaling through the D1 receptor plays a crucial role in the DA-mediated negative regulation of glucose-stimulated Ca²⁺ accumulation. We then confirmed the effect of the D1 antagonist on GSIS using TIRFM, with another D1 antagonist, SKF83566 (40), or a D1 agonist, SKF38393 (41). Indeed, SKF83566, but not SKF38393, increased the second phase of insulin granule exocytosis (Fig. 3C and D).

D1 antagonists enhanced the glucose-stimulated Ca²⁺ flux, and insulin exocytosis suggested that DA exists in the culture environment. DA has been shown to be released into the culture in response to high-glucose stimulation (20). In the synapses of the central nervous system, the local concentration of DA is estimated to be ∼0.5–100 μmol/L within a 5-μm radius from the release site (17,42). We interpreted the result to mean that a D1 receptor might be involved in the signaling complex of the negative regulation of GSIS. D1 antagonists competed with DA to bind to its receptor in the signaling complex and rescued the inhibitory effect of DA on glucose-stimulated Ca²⁺ influx, which led to increased insulin granule exocytosis.

To gain insight into the possible role of D1 and D2 receptors in the negative regulation of GSIS, we overexpressed D1 or D2 receptors in primary islet cells using the adenoviral vector system. The intensity shown in Fig. 4A and B was adjusted to make endogenously expressed D1 and D2 visible. Supplementary Fig. 3 shows lower contrasts of the overexpression of D1 (D1OE) or D2 (D2OE) β-cells. In control, uninfected β-cells, most insulin-expressing β-cells were costained weakly with anti-D1 (Fig. 4A, top) or anti-D2 antibody (Fig. 4B, bottom). The result suggests that β-cells express the D1 and D2 receptors. D1OE or D2OE in islet cells increased the intensity of the D1 or D2 immunoreactivity, respectively (Fig. 4A and B). Interestingly, D1OE islet cells exhibited extended morphologies, showing higher attachment to the plate.

We then examined the effects of D1OE or D2OE on glucose-stimulated Ca²⁺ flux. In control, uninfected β-cells, DA inhibited glucose-stimulated Ca²⁺ flux. However, in D1OE cells, DA potentiated the increase in glucose-stimulated Ca²⁺ flux (Fig. 4C and D). By contrast, D2OE in islet β-cells abolished glucose-stimulated Ca²⁺ flux, even without exogenous DA (Fig. 4E).

We then assayed the effects of D1OE or D2OE on GSIS using TIRFM (Fig. 4F). In D1OE β-cells, DA addition significantly increased the basal insulin granule exocytosis at low glucose. DA-mediated inhibition of the second phase of insulin granule exocytosis was partially rescued in D1OE β-cells compared with control (Fig. 4F, left and middle, and G). By contrast, D2OE exhibited a robust inhibitory effect on insulin granule exocytosis. The first phase of insulin granule exocytosis was also abolished, even before exogenous DA addition, along with the inhibition of the second phase (Fig. 4F, left and right, and G).

We interpreted that the D2OE β-cell lack of response to glucose challenge is due to a hypersensitization of the D2 receptor. The β-cells are continuously exposed to the cosecreted DA with insulin in the vicinity (17,42). We interpreted the results to mean that D2OE cells become desensitized and unresponsive to high glucose. We tested the response of the D1OE or D2OE β-cells to forskolin and found that D2OE, but not D1OE, β-cells failed to exhibit forskolin-induced intracellular cAMP accumulation at high glucose (Supplementary Fig. 3B). The D2OE β-cells also lost their response to tolbutamide after culture at high (25 mmol/L) glucose conditions (Supplementary Fig. 3C).

These results suggest that either D1OE or D2OE alone could not confer the transient inhibitory effect of DA. The results also raise the possibility of D1-D2 receptor involvement in the DA-mediated inhibitory signal.

We next investigated the involvement of the D1-D2 heteromer, using an agonist against the D1-D2 heteromer, SKF83959 (31), in place of DA. Under the glucose challenge, Ca²⁺ influx occurred. SKF83959 at 10 μmol/L (at 301 s) triggered a rapid, but transient decrease in Ca²⁺ flux, which rapidly returned to a normal level before the agonist addition within 120 s (Fig. 5A and B). Compared with the prolonged inhibition of Ca²⁺ flux by DA (Fig. 2B and D), the inhibition by SKF83959 was transient. D1OE and D2OE receptors (D1D2OE) sensitized β-cells such that they responded to a concentration of SKF83959 as low as 1 nmol/L to inhibit Ca²⁺ flux (Fig. 5C and D).

We then examined the effect of SKF83959 on insulin secretion using TIRFM (Fig. 5E). SKF83959 or DA was added 5 min after the glucose challenge. We found that SKF83959 triggered a transient decrease in insulin exocytosis in contrast to the prolonged decline by DA (Fig. 5E). Quantitative analysis confirmed that the inhibition of insulin granule exocytosis by SKF83959 occurred transiently and lasted for up to 2.5 min after the addition (range 0–2.5 min). Subsequently, insulin exocytosis was recovered in SKF83959-treated, but not DA-treated, cells. Note that the average fusion events in SKF83959-treated cells were no different from those of the controls after 2.5 min. This transient inhibitory feature of SKF83959 differed from that of DA (Fig. 5F).

We then asked whether the activation of the D1-D2 heteromer affects cAMP accumulation. Unlike DA, SKF83959 treatment did not inhibit the forskolin-dependent increase of cAMP under glucose challenge (Fig. 5G). D1-D2 heteromer activation does not affect cAMP accumulation, which might explain its transient inhibitory effect on insulin secretion.

We performed a Duolink Proximity Ligation Assay (PLA) (Fig. 6) and noticed that a certain level of D1-D2 heteromer exists in insulin-expressing cells in control islets. DA or SKF83959 significantly increased the intensity, suggesting that agonist activation increased D1-D2 heteromer formation (Fig. 6A and B). D2OE significantly increased D1-D2 heteromer formation compared with control (scramble) or D1OE alone. D1D2OE further increased heteromer formation (Fig. 6C and D). Our results suggest that D2 is a limiting factor for D1-D2 heteromer formation. The coexpression of D1 with D2 significantly increases D1-D2 heteromer formation.

We then performed Western blot analysis to confirm D1-D2 receptor heteromer formation in pancreatic β-cells. Intermolecular interactions between transmembranes are important in the case of D2 receptor dimerization. D2 homodimer and higher order oligomers were reported to be sodium dodecyl sulfate resistant, and disruption of disulfide linkages did not significantly dissociate the dimer formation (43). Antibodies against the D1 receptor recognized a band of ∼48 kDa in uninfected islet lysates (Fig. 6E), whereas anti-D2 antibody did not detect any band (F.U., unpublished observations). In D1OE islet lysate, anti-D1 antibody detected a band of ∼100 kDa, which seems to correspond to D1 homodimer (Fig. 6F). On the one hand, in D2OE and D1D2OE islet lysates, bands corresponding to D1-D2 heteromer (∼86 kDa) and D1-D2 oligomer were detected by both anti-D1 antibody and anti-D2 antibody (Fig. 6F and G). On the other hand, anti-D2 antibody detected D2 monomer (∼37 kDa) and D2 homodimer (∼74 kDa) in D2OE and D1D2OE (Fig. 6G). The results suggest that D1-D2 heteromer formation was increased in D2OE and D1D2OE, agreeing with the PLA results (Fig. 6C and D).

We then investigated the functionality of the D1-D2 heteromer by Ca²⁺ imaging. In D1OE primary β-cells, the D1-D2 heteromer agonist SKF83959 exerted no effects. However, in D2OE and D1D2OE β-cells, transient decreases in Ca²⁺ influx were observed at doses of 10 or 100 nmol/L SKF83959 (Fig. 6H). The transient inhibition of Ca²⁺ influx by SKF83959 agrees with the PLA and Western blot analysis results, demonstrating that D2OE and D1D2OE increased the formation of D1-D2 heteromer, which mediates the inhibitory effect on Ca²⁺ influx.

We then confirmed the ability of D1 and D2 antagonists to compete with the activation of D1-D2 heteromer by SKF83959 in D1D2OE β-cells (Supplementary Fig. 4A). SKF83959 binds to the D1 and D2 receptor sites in the D1-D2 heteromer (31). SKF83959 does not bind to the D2 receptor itself. The SKF83959 binding site on the D2 receptor appears in the D1-D2 heteromer, and SKF83959 competes with the D2 antagonist on the D2 binding site (31). The D1 antagonist SCH23390 or D2 antagonist haloperidol competed with SKF83959 and reverted the transient decrease in Ca²⁺ flux triggered by SKF83959. The application of both SCH23390 and haloperidol simultaneously almost entirely canceled the inhibitory effect of SKF83959 on Ca²⁺ flux. SCH23390 counteracted the inhibitory effect of SKF83959 and triggered an increase in Ca²⁺ accumulation similar to that observed with DA (Fig. 3A). The results further support our hypothesis that D1 receptors participate in the DA-mediated inhibitory signal. We tested the addition of SKF83959 in the presence of various concentrations of DA. The transient inhibitory effect of SKF83959 was not affected by DA (Supplementary Fig. 4B, top). Of note is that in control β-cells, DA decreased glucose-stimulated Ca²⁺ flux in a concentration-dependent manner. On the other hand, in D1D2OE cells, low doses (1 and 10 nmol/L) of DA activated PTX-sensitive high-affinity D2 receptor, whereas high doses (100 nmol/L, 1 μmol/L, or 10 μmol/L) of DA activated PTX-resistant low-affinity D1 receptor (Supplementary Fig. 4B, top right and bottom). The results agree with the Western blot results (Fig. 6F) that D1 and D2 receptor homodimers both exist in D1D2OE cells.

We previously reported that DA promotes β-cell dedifferentiation and decreases β-cell mass and β-cell survival through the D2 receptor. D2 receptor acted through the formation of D2-A2A heteromer, counteracting adenosine-mediated β-cell proliferating signal (25). To investigate the possible involvement of the D1-D2 heteromer in regulating β-cell mass, we overexpressed D1 (D1OE), D2 (D2OE), or both (D1D2OE) in primary β-cells; cultured them for up to 5 days; and then assayed them for total cell numbers, β-cell numbers, and the proportion of cells that underwent apoptosis.

Under culture, the number of control untransfected β-cells gradually decreased. D2OE cells showed a more significant decrease in total cell number or β-cell number than the control untransfected cells. DA addition accelerated the reduction of total cell numbers or β-cell numbers. D1OE did not show significant effects, whereas D1D2OE exerted beneficial protective effects on β-cell number or total cell number, particularly in the presence of DA (Fig. 7B and C). TUNEL assay further demonstrated that cell death increased upon DA activation in D2OE. However, D1D2OE rescued cell death (Fig. 7D). Since D2OE also showed an increase in D1-D2 heteromer (Fig. 6G), we further compared adding DA or SKF83959 to D2OE cells and found that unlike DA, SKF83959 addition to D2OE cells did not decrease β-cell numbers (Fig. 7E and F).

Our results reveal a protective role of the D1 receptor to prevent β-cell failure. Our finding that the D1 antagonist SCH23390 increased glucose-stimulated Ca²⁺ flux and GSIS led to our present discovery of the unexpected role of the D1 receptor in the formation of a receptor complex in the DA-mediated negative regulation of GSIS. The D1 antagonist SCH23390 shows a high D1 affinity (K_i ≤0.12 nmol/L) and a much lower D2 (K_i ∼1,100 nmol/L) or D3 (K_i ∼800 nmol/L) affinity (39,44). To further confirm the specificity of SCH23390 against the D1 receptor, we used a Drd1^tm1a/WT mouse model (Supplementary Fig. 5 and Supplementary Materials and Methods). The Drd1 expression level in the Drd1^tm1a/WT islets is about one-half of the wild-type littermate islets (Supplementary Fig. 5B). SCH23390 increased the glucose-stimulated Ca²⁺ flux from 100 nmol/L. The effect of SCH23390 was smaller in the Drd1^tm1a/WT mutant than in the wild type at SCH23390 50, 100, and 500 nmol/L (Supplementary Fig. 5C). The result demonstrates that the effect of SCH23390 is mediated through the D1 receptor and further confirms our findings that the D1 receptor participates in the negative regulation of the glucose-stimulated Ca²⁺ flux and GSIS in β-cells. In conclusion, the D1 receptor functions through D1-D2 heteromer formation to sequester D2 from forming a homodimer (Fig. 8).

We performed cAMP imaging, Ca²⁺ imaging, and TIRFM to study the effects of DA through its receptors. DA is colocalized in the insulin vesicular granule, but not in synaptic-like microvesicles, in pancreatic β-cells. Upon GSIS, DA is cosecreted with insulin to regulate the insulin granule exocytosis negatively. Although DA can inhibit the first phase of insulin exocytosis, endogenous DA cosecreted with insulin upon GSIS seems to inhibit the second phase of insulin exocytosis. We found that the D1 receptor plays an essential role in transducing the DA-mediated negative signal for GSIS using antagonists. From overexpression studies, we found that D2OE exerted a robust inhibitory effect on Ca²⁺ flux and almost eliminated GSIS, which is toxic for β-cell survival. D1 sequesters D2 to enable the formation of the D1-D2 heteromer, avoiding excess toxic signaling through the D2 homodimer in β-cells. A D1-D2 heteromer–specific agonist, SKF83959, triggered a transient inhibition of GSIS without eliciting harmful effects on β-cells.

Upon insulin secretion, DA is cosecreted into the environment. β-Cells express DAT, which reuptakes DA, or l-DOPA in the circulation (16). The local DA concentration might be high in β-cells upon GSIS, similar to that estimated in the synapse (17,42). D2OE receptor increased the inhibitory D2 homodimer and could elicit autoactivation of the receptor. We previously reported that D2OE increased β-cell dedifferentiation and apoptosis (25). We interpret this to mean that healthy β-cells show a limited expression level of D2 receptors so that signaling through the D1-D2 heteromer is dominant. An increase in D2 expression level might increase the formation of the D2 homodimer, lead to a toxic effect on β-cell survival, and trigger β-cell dysfunction.

It is of interest that only a limited population of β-cells express TH. We found that islet cells containing 20% TH OE cells exhibited inhibition in the second phase of insulin secretion. In rats, TH activity was reported to be subject to changes by diet manipulation or the microenvironment (34,45). Studies comparing eight mouse strains described that high TH expression was associated with increased DA, decreased insulin secretion, and lean, resistant to high-fat/high-sucrose diet–induced metabolic phenotypes. Preincubation of l-DOPA in mice with low TH expression mimicked the increase in islet DA levels and blunted GSIS. Notably, the mouse strain with high TH expression was the most insulin-sensitive strain compared with those with low TH expression (46). This suggests the importance of DA-mediated signals in maintaining β-cell and glucose homeostasis.

D3 receptors belong to the D2 class, are expressed in the mouse islets, and act in combination with D2 to inhibit GSIS (16,20). The expression of D2 was reported in β-cells in mice, rats, and humans (14). D1 expression was also reported in rat islet β-cells (14,47). D1 and D2 expression levels might associate with GSIS responsiveness, which remains to be investigated. To our knowledge, there are no reports on the differential expression of D1 or D2 receptors.

SKF83959 was initially identified as a D1 agonist. SKF83959 has a high D1 affinity (K_i ≤1.10 nmol/L) and a low D2 affinity (K_i ≤920 nmol/L) (44,48). The D1 antagonist SCH23390 showed an even higher affinity to D1 (K_i ≤0.12 nmol/L) but not D2 (K_i ∼1,100 nmol/L) or D3 (K_i ∼800 nmol/L) (39,44). SKF83959 binds to the D1 receptor site of the D1-D2 heteromer. The SKF83959 binding site in the D2 receptor appears upon D1-D2 heteromer formation (31). Our overexpression and competition results demonstrate the selectivity of SKF83959 and SCH23390 on the D1 receptor. Our interpretation of the result is that D1 antagonist SCH23390 competes with DA/SKF83959 in binding to the D1-D2 heteromer, thereby rescuing the inhibitory effect.

Activation of G_βγ modulates the exocytosis of insulin granules through interaction with voltage-dependent Ca²⁺ channels, G-protein–activated inward rectifying potassium channels, and soluble N-ethylmaleimide–sensitive factor attachment protein (SNARE) receptor (49). We found that DA inhibited Ca²⁺ influx partially through G-_i/oα. G_βγ might also be involved in D1-D2 heteromer activation, which remains to be investigated.

The D1-D2 heteromer was reported in the striatum in a unique subset of neurons. In the striatum, D1-D2 heteromer coupled to G-α_q/11 increased the Ca²⁺/calmodulin-dependent protein kinase IIa (CaMKIIα) in the nucleus accumbens, suggesting a potential role in synaptic plasticity. There is evidence for functional precoupled complexes of receptor heteromers and adenylyl cyclase (50). The signaling pathway downstream of the D1-D2 heteromer in β-cells is unknown and awaits further investigation.

The expression of the D1 receptor that forms a heteromer with D2 to fine-tune the D2 inhibitory signal is a key to keeping β-cells healthy. Our results suggest that signaling through the D1-D2 heteromer in β-cells enables fine-tuning of the Ca²⁺ signal. Signaling through the D1-D2 heteromer triggers a transient and rapid decrease in Ca²⁺ influx to inhibit GSIS temporarily. However, when the D2 homodimer becomes dominant, signaling through the D2 receptor exhibits a robust inhibitory effect on Ca²⁺ influx and GSIS. Understanding the regulation of DA signaling may improve methods for preventing and treating diabetes.

Acknowledgments. The authors thank Dr. Nobuaki Shiraki for the discussion. The authors thank the members of the Animal Centers and Biomaterials Analysis Division, Open Facility Center, and Open Research Facilities for Life Science and Technology at the Tokyo Institute of Technology, for technical assistance.

Funding. This work was supported by grants-in-aid from the Ministry of Education, Culture, Sports, Science and Technology, Japan (20K08325 to D.S. and 21H02978 to S.K.). This work was also supported in part by the Takeda Science Foundation and Japan Insulin Dependent Diabetes Mellitus (IDDM) Network.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. F.U. performed the experiments and collected, analyzed, and discussed the data. T.A. provided technical advice for the analysis of the imaging data. T.K. provided technical advice for cAMP imaging. H.T. and S.S. provided technical advice for TIRFM and discussed the data. D.S. discussed the data and provided conceptual input and technical advice. S.K. provided conceptual input, discussion, writing and revision of the manuscript, and approval of the final version of the manuscript and obtained funding. D.S. and S.K. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.