GluDs are ionotropic dopamine receptors tuned by G-proteins

Dopamine is a neurotransmitter essential for cognition, and its dysregulation is associated with neurological diseases^1,2. Historically, dopamine has been understood to signal exclusively through metabotropic receptors³. Delta-type ionotropic glutamate receptors (GluDs), which have recently been established as ligand-gated ion channels^4,5, are fundamental for synaptic maintenance, are implicated in neurological disorders, and co-localize with dopaminergic machinery. Here, we report that dopamine is a direct agonist of GluDs, eliciting ionotropic activity, as visualized by cryo-electron microscopy (cryo-EM), bilayer recordings, mutagenesis, and patch clamp recordings. Dopamine binds to the GluD ligand binding domain, inducing clamshell closure and channel activation through a distinct molecular interface. GluD channel activity is tightly regulated by G-proteins, which act as molecular switches to tune GluD activity: free Gβγ inhibits ligand-gating, while Gα or inactive G-protein heterotrimers enable dopamine-induced GluD currents. This tuning of GluD activity by G-proteins is uncoupled in a point mutation associated with neurodegeneration. These findings expand mechanisms of neuronal dopaminergic signaling, uncover how G-proteins tune GluD channel activity, and provide a framework for targeting GluDs in neurological diseases.

Dopamine was discovered over a century ago and has since been appreciated as a neurotransmitter that is essential for learning, motor control, and motivation^1,6. The dysregulation of dopaminergic signaling is linked to Parkinson’s disease, schizophrenia, addiction, depression, attention deficit hyperactivity disorder, and bipolar disorder². Unlike many other neurotransmitters, dopamine has not been found to directly activate ligand-gated ion channels in the nervous system³. Rather, dopamine has long been thought to act via volume transmission by activating G-protein coupled receptors (GPCRs), which then stimulate second-messengers to modulate neuronal excitability and plasticity^6–9. Despite this view, dopamine is concentrated into vesicles similarly to ionotropic neurotransmitters and is released high concentrations, suggesting the presence of synaptic-like signaling^3,10,11.

Intriguingly, dopamine release machinery co-localizes with delta-type ionotropic glutamate receptors (GluDs; GluD1 & GluD2) in the molecular layer of the cerebellum (GluD2) and the dopaminergic midbrain (GluD1)^12–26. The GluDs have recently been established as bona fide ligand-gated ion channels and are unique amongst ionotropic glutamate receptors (iGluRs): rather than binding glutamate, they bind distinct neurotransmitters like D-serine and GABA. GluDs have fundamental roles in regulating both excitatory and inhibitory synapses and are implicated in diseases such as schizophrenia, cerebellar ataxia, intellectual disability, dementia, and glioblastoma^27–29. Notably, at Purkinje cell dendrites—where GluD2 is highly localized—dopamine release induces a depolarizing current, hinting at a direct functional connection¹³.

However, cellular currents from GluDs have not been readily recorded in cells. Most cellular recordings have relied on GluDs containing a missense mutation (the “Lurcher” mutation; Lc), which are capable of conducting current, and do so in the absence of ligand^30,31. Recent evidence has suggested that currents from GluDs are heavily regulated in cells by G-protein coupled receptor (GPCR) signaling networks, supported by data showing that GPCR activation enables currents to be recorded from GluDs^{22–25,32,33}. Thus, GluDs appear to be ligand-gated ion channels that are heavily regulated by GPCR signaling cascades in cellular contexts.

GPCR activation leads to splitting of the inactive G-protein heterotrimer Gαβγ into Gα and Gβγ, which are released from the GPCR to stimulate signaling cascades^34–36. We hypothesize that G-proteins directly regulate the channel activity of GluDs. This would not be without precedent; inward rectifying potassium channels, voltage-gated calcium channels, and transient receptor potential channels are directly regulated by G proteins^37–46. However, G-proteins have not been shown to directly regulate iGluRs.

Our ideas are twofold: (1) that dopamine is a ligand for GluDs, and (2) that G-proteins regulate GluDs, directly through modulating their ionotropic capabilities. To test this, we purified human GluD1 and GluD2. Dopamine elicits ligand-gated currents from both GluDs in brain lipid bilayers in a concentration-dependent manner. Through cryo-EM of GluD2 in the presence of dopamine, we confirm that dopamine directly binds to the GluD2 ligand binding domain (LBD), through a binding mechanism unique from the gliotransmitter D-serine. G-protein binding to GluD2 directly regulates the ligand-gating activity (of both dopamine and D-serine), and we observe through patch clamp recording that ligands activate GluD2 in cells following stimulation by Gα. In total, our data expand the signaling pathways of dopamine in the nervous system, redefine our understanding of how GluD channels are regulated, expand the understanding of how G-proteins regulate ion channels, and provide avenues for therapeutic intervention of GluD signaling networks.

Results

Dopamine gating

To test if dopamine elicits ligand-gated responses from the GluDs, we first purified human GluD2 (Extended Data Fig. 1) and reconstituted the receptors into brain lipid bilayers to test for ligand-gated activity. In the presence of 0.1 mM dopamine, we observe canonical ligand-gated currents reflected by oscillations between the closed (C) state and open states (represented by conductance states O1-O4; Fig 1a). The current histogram is best fit with three populations of sub-conductance levels (± standard error of the mean, SEM; Fig. 1b): O1 (22 ± 0.9 pS), O2 (45 ± 0.3 pS), and O3 (70 ± 0.1 pS).

Fig. 1 |. Ligand-gating of GluDs by dopamine.

a, Example of a GluD2 current trace in the presence of 0.1 mM (top), 1 mM (middle), and 10 mM dopamine (bottom) without or with (100 μM NASPM, purple) at 37°C. b, Current histogram of GluD2 in response to dopamine, fit with two to four Gaussian components at 37 °C. 0.1 mM: N = 7 traces; R² for O1 = 0.99 (6886 events), O2 = 0.98 (1041 events), O3 = 0.99 (41 events). 1 mM: N = 7 traces; R² for O1 = 0.99 (2454 events), O2 = 0.99 (512 events), O3 = 0.98 (176 events), O4 = 0.82 (37 events). 10 mM: N = 13 traces; R² for O1 = 0.92 (30543 events), O2 = 0.99 (8501 events), O3 = 0.99 (3755 events), O4 = 0.94 (1009 events). 10 mM + 100 μM NASPM: N = 5 traces; R² for O1 = 0.99 (2075 events), O2 = 0.99 (52 events). c, Example of a GluD1 current trace in the presence of 0.1 mM (top), 1 mM (middle), and 10 mM dopamine (bottom) at 37°C. d, Current histogram of GluD1 in response to dopamine, fit with two to four Gaussian components at 37 °C. 0.1 mM: N = 11 traces; R² for O1 = 0.99 (6612 events), O2 = 0.98 (179 events). 1 mM: N = 7 traces; R² for O1 = 1.0 (1224 events), O2 = 0.86 (94 events). 10 mM: N = 13 traces; R² for O1 = 0.99 (10150 events), O2 = 0.97 (991 events), O3 = 0.99 (139 events), O4 = 0.97 (22 events). All data are shown as mean ± SEM.

If these currents are dopamine-dependent, we expect to observe concentration-dependent shifts in the distribution of open-state occupancies, like the glutamate concentration-dependence of the conductance levels in α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA)-subtype iGluRs (AMPARs)^47–59. Accordingly, as we increase the dopamine concentration from 0.1 mM to 1 mM and 10 mM, we not only observe a qualitative increase in the dopamine responses from both GluD1 and GluD2 in bilayers, but also a leftward shift in the current histogram populations, reflecting the concentration-dependent occupancy of sub-conductance levels (Extended Data Fig. 1). In the presence of 1 mM dopamine, the GluD2 current histogram fits with four sub-conductance levels (± SEM; Fig. 1b): 20 ± 6 pS (O1), 50 ± 0.4 pS (O2), 74 ± 0.4 pS (O3), and 97 ± 0.6 pS (O4). The relative frequency of sub-conductance occupation increases as dopamine increases to 10 mM, reflected by an increase in O4 occupancy. This ligand-dependence of the GluD2 dopamine current histograms is reminiscent of the four subconductance levels observed in AMPARs that are glutamate concentration-dependent^47–59.

To verify that the dopamine currents are being mediated by GluD2, we tested whether the currents are blocked by 1-naphthyl acetyl spermine (NASPM), a polyamine-based toxin that blocks GluD currents^4,30,60,61. Accordingly, we observe attenuation of currents from 10 mM dopamine in the presence of 100 μM NASPM, reflected by a rightward shift in the current histogram (Fig. 1b).

Next, we tested whether GluD1 ligand-gating is also sensitive to dopamine (Extended Data Fig. 1). Like GluD2, we observe dopamine concentration-dependent responses (0.1, 1, 10 mM) from GluD1. In the presence of 0.1 mM dopamine, we observe activation of GluD1 to O1 and O2 (18 ± 0.3 pS and 47 ± 0.5 pS, respectively; Fig. 1d). As we increase to 1 mM dopamine, we observe O1 and O2 (18 ± 0.5 pS and 51 ± 1.1 pS), and in 10 mM we observe O1-O4 (19 ± 0.2 pS, 48 ± 0.4 pS, 74 ± 0.4 pS, 98 ± 0.5 pS), along with a concentration-dependent change in the occupancies. However, compared to GluD2, we observed lower values for mean conductance and open probability (P_O) in GluD1, reflecting a difference in open state occupancy and single-channel kinetics between GluD subtypes (Extended Data Fig. 1). Collectively, dopamine appears to elicit ligand-gated ion channel activity from both isolated GluD1 and GluD2.

To-date, the canonical method of assaying GluD cellular function has been through recording spontaneous cellular currents from GluD1_Lc or GluD2_Lc. Ligands such as D-serine reduce the spontaneous currents, presumably through a desensitization-like mechanism. Therefore, we tested whether dopamine produces a similar response on GluD2_Lc through cellular recordings from GluD2_Lc. Indeed, we see a dopamine-dependent reduction of GluD2_Lc current, consistent with binding to the ligand binding domain (LBD) and with D-serine reduction of current (Extended Data Fig. 2).

Dopamine binding to GluD2

To identify the dopamine binding site, we used cryo-EM to reconstruct the three-dimensional architecture of GluD2 in the presence of dopamine (Fig. 2a; Extended Data Figs. 3,4). The overall architecture of GluD2 is reminiscent of GluD2 bound to D-serine: the receptor has an overall italic “Y” shape in a non-domain-swapped arrangement, where local dimers in the amino terminal domain (ATD) form between the A and B subunit positions, as well as the C and D subunit positions, and dimer pairings do not swap at the level of the LBD. Below the LBD is the transmembrane domain (TMD), which houses a central cation channel. The dopamine-bound GluD2 structure is similar to the D-serine-bound structure⁴ with an overall root mean squared deviation (RMSD) of 1.2 Å, with each domain layer having an RMSD of 1.0 Å or less (Extended Data Fig. 5a).

Fig. 2 |. Cryo-EM of dopamine-bound GluD2.

a, Composite cryo-EM map of dopamine-bound GluD2. b, Local LBD map and slice through the dopamine-bound LBD. c, Structure of the LBD subunit, with inset highlighting the dopamine binding pocket, showing the D742 residue. d, i, Example of GluD2 (D742A) current trace in the presence of 10 mM (n = 4), 50 mM (n = 3) and 100 mM (n = 4) dopamine (top); or 10 mM (n = 4), 50 mM (n = 5) and 100 mM (n = 3) D-serine (bottom); ii, example of GluD2 (R530K) current trace in the presence of 10 mM dopamine (n = 4) or 10 mM D-serine (n = 10).

All LBDs have closed clamshells (closed ~16 Å compared to GluD2 in the absence of ligand), which is a general indicator of ligand binding in iGluRs (Extended Data Fig. 5b)⁶². However, due to intermediate resolution, whether dopamine was bound to the LBDs was not immediately apparent. To elucidate the binding site, we performed signal subtraction and local refinements on each individual LBD (Extended Data Figs. 3,4). The D subunit position LBD refined to sufficient resolution to enable definitive assignment for a dopamine feature in the center of the LBD clamshell (Fig. 2b; Extended Data Fig. 5b). The map feature for dopamine is flat and elongated like the chemical structure of dopamine and the map is shaped distinctly from smaller ligands such as D-serine (Extended Data Fig. 5c).

Dopamine sits between the upper (D1) and lower (D2) lobes of the LBD and appears to be coordinated by a hydrophobic pocket with D742 coordinating the amine moiety (Fig. 2c, inset i), though the map resolution occludes complete visualization. However, the binding site is unique from the D-serine binding site in GluD2, where while D742 is essential for D-serine binding, R530 also plays a major role in D-serine binding through coordination of the α-carboxyl group on D-serine^4,63, however, R530 appeared to be absent from the dopamine binding pocket (Extended Data Fig. 5d,e). Thus, we hypothesized that mutagenizing D742 would reduce the effects of dopamine, while mutagenesis at R530 would not.

To test this, we first mutagenized D742 to D742A, which would negate the amine coordination. Indeed, we do not observe ligand-gating from GluD2(D742A) until extreme dopamine concentrations (e.g., 100 mM; Fig. 2d, inset i). While GluD2(R530K) has a similar effect on D-serine ligand-gating^4,63, it does not reduce dopamine gating (Fig. 2d, inset ii), further supporting the absence of R530 from dopamine coordination in the LBD (Fig. 2c, inset i).

G-protein binding and regulation

There is a wealth of data suggesting that activation of GPCRs contributes to the stimulation of currents from GluDs^22–25,33. However, the molecular mechanisms of how cellular GluD currents are stimulated are unknown. We hypothesized that activated Gα subunits may regulate this, considering the GTP-dependence of GluD currents³³, and the potential of direct Gα binding to iGluRs^64,65. Toward this end, we first tested whether G-proteins can directly bind GluDs, starting with the most predominant family of G-proteins in the brain, the G_i/o inhibitory family⁶⁶.

We directly tested if GluD2 can form complexes with G-proteins by using purified recombinant GluD2 immobilized on affinity resin as bait to assay GluD2 binding to Gα, Gβγ, and Gαβγ (Fig. 3a, inset i; Extended Data Fig. 6). We observe that Gα, Gβγ, and Gαβγ are independently pulled down in a GluD2-specific manner (Fig. 3a, inset ii). We then tested if Gα can bind GluD2 independently of Gβγ by assaying if GluD2-Gβγ can pull down Gα (pre-incubated with GTPγS to prevent heterotrimer formation). We observe binding (Fig. 3a, inset ii), indicating that Gα and Gβγ can simultaneously and independently bind GluD2 in GluD2-Gβγ+Gα complexes.

Fig. 3 |. Reconstitution of GluD2 G-protein complexes.

a, Schematic of G-protein affinity pulldown by GluD2-eGFP-Streptag II (i) and example gel of the elutions (ii; n = 3). b, Example current traces in the presence of 10 mM dopamine (yellow) or 10 mM D-serine (blue) of GluD2 bound to Gα (i; n = 5 and 6 traces for dopamine and D-serine, respectively), Gβγ (ii, n = 5 and 14 traces), Gαβγ (iii, n = 8 and 23 traces), and Gα + Gβγ (iv, n = 6 and 6). c, Mean P_O ± SEM of Apo GluD2 currents, dopamine: 4.9 ± 0.4% (n = 302 10s event detections), D-serine: 5.8 ± 0.8% (n = 69); GluD2-Gα, dopamine: 5.9 ± 0.4% (n = 248), D-serine: 4.0 ± 0.6% (n = 147); GluD2-Gαβγ, dopamine: 13.1 ± 1.4% (n = 158), D-serine: 15.6 ± 2.5% (n = 75). Dopamine: Apo versus Gα, P value 0.77, 95% CI of difference −2.6 to 0.67, degrees of freedom (df) = 500.4; apo versus Gαβγ, P value <0.0001, 95% CI of difference −12.2 to −4.1, df = 179.4. D-serine: Apo versus Gα, P value 0.66, 95% CI of difference −1.1 to 4.7, df = 161.3; apo versus Gαβγ, P value 0.0039, 95% CI of difference −17.3 to −2.3, df = 87.8. d, Overall mean conductance ± SEM of Apo GluD2 currents, dopamine: 36.71 ± 0.10 pS (43795 events), D-serine: 38.25 ± 0.41 pS (3003 events); GluD2-Gα, dopamine: 30.71 ± 0.14 pS (7608 events), D-serine: 24.56 ± 0.31 pS (1555 events); GluD2-Gαβγ, dopamine: 35.05 ± 0.16 pS (13549 events), D-serine: 40.93 ± 0.31 pS (4503 events). Dopamine: Apo versus Gα, P value <0.0001, 95% CI of difference 5.50 to 6.49, d.f. = 16444; apo versus Gαβγ, P value <0.0001, 95% CI of difference 1.13 to 2.20, df = 25630. D-serine: Apo versus Gα, P value <0.0001, 95% CI of difference 12.24 to 15.13, df = 4548; apo versus Gαβγ, P value <0.0001, 95% CI of difference −4.13 to −1.23, df = 6114. e, Mean open dwell time duration ± SEM of apo GluD2currents, dopamine: 8.8 ± 0.04 ms (42295 events), D-serine: 9.6 ± 0.18 ms (3013 events); GluD2-Gα, dopamine: 8.7 ± 0.07 ms (5520 events), D-serine: 9.7 ± 0.39 ms (1550 events); GluD2-Gαβγ, dopamine: 8.7 ± 0.06 ms (9989 events), D-serine: 10.1 ± 0.13 ms (8535 events). Dopamine: Apo versus Gα, P value 0.65, 95% CI of difference −0.11 to 0.37, df = 9624; apo versus Gαβγ, P value 0.36, CI of difference −0.067 to 0.37, df = 20320. D-serine: Apo versus Gα, P value 1.0, 95% CI of difference −1.35 to 1.10, df = 2193; apo versus Gαβγ, P value 0.22, 95% CI of difference - 1.11 to 0.14, df = 6526. f, Mean closed dwell time duration ± SEM of Apo GluD2 currents, dopamine: 209 ± 7.7 ms (24775 events), D-serine: 359 ± 21 ms (2038 events); GluD2-Gα, dopamine: 198 ± 23 ms (5421 events), D-serine: 386 ± 27 ms (1287 events); GluD2-Gαβγ, dopamine: 113 ± 5 ms (7143 events), D-serine: 187 ± 8 ms (3942 events). Dopamine: apo versus Gα, P value 1.0, 95% CI of difference −58.96 to 81.76, df = 6648; apo versus Gαβγ, P value <0.0001, 95% CI of difference 69.95 to 122.8, df = 30781. D-serine: Apo versus Gα, P value 0.97, 95% CI of difference −125.8 to 71.15, df = 2660; apo versus Gαβγ, P value <0.0001, 95% CI of difference 108.5 to 236.4, df = 2621. **P<0.01; ****P< 0.0001; ns, not significant.

How does GluD2 function in the context of G-proteins? We reconstituted the GluD2 and GluD2 G-protein complexes into brain lipid bilayers for single-channel bilayer recordings to assay how G-proteins alter GluD2 ligand-gating. In the absence of G-proteins, GluD2 exhibits canonical ligand-gating behavior in the presence of D-serine and GABA⁴, as well as dopamine (Fig. 1). We tested conditions in the presence of D-serine, which is understood to be a canonical ligand for GluD2, as well as dopamine.

We observe ionotropic behavior from GluD2-Gα (Fig. 3b, inset i) and find that apo GluD2 has a P_O of 5.8 ± 0.8% in the presence of D-serine and 4.9 ± 0.4% in the presence of dopamine. With GluD2-Gα, these values changed modestly: 4.0 ± 0.8% and 5.8 ± 0.4% in the presence of D-serine and dopamine, respectively (Fig. 3c). However, the mean conductances significantly decreased to 24 ± 0.3 pS (D-serine) and 31 ± 0.1 pS (dopamine) for GluD2-Gα (Fig. 3d). Unexpectedly, Gβγ silenced the GluD2 ionotropic response regardless of ligand (Fig. 3b, inset ii). However, we observe increased P_O from GluD2-Gαβγ for D-serine and dopamine, to 15.6 ± 2.4% and 13.1 ±1.4% (Fig. 3c), and modest changes in the mean conductance from 43 ± 0.3 pS and 35 ± 0.2 pS (dopamine) (D-serine, Fig. 3d). The increase in GluD2-Gαβγ P_O appears to be specifically due to a decrease in the mean closed duration, shifting from 359 ± 21 ms and 209 ± 8 ms to 187 ± 8 ms and 113 ± 5 ms in D-serine and dopamine, respectively (Fig. 3e,f). GluD2-Gαβγ did not affect the mean open duration compared to apo GluD2.

We next tested whether Gα can rescue the silencing of GluD2-Gβγ complexes because Gα can bind to GluD2 along with Gβγ (Fig. 3a). Indeed, both D-serine and dopamine can elicit ionotropic behavior from GluD2-Gβγ+Gα, suggesting that Gα can relieve inhibition by Gβγ (Fig. 3b, inset iv). Collectively, these results from in vitro reconstitution support a dynamic, modulatory role of G-proteins on GluD2 function, dependent on the specific G-protein assembly.

G-protein tuning of cellular GluD currents

Our in vitro data suggest that GPCR regulation of GluD currents is from direct G-protein tuning of GluDs. We tested this directly in situ by expressing human GluD2 in heterologous cells and performing inside-out patch electrophysiology to allow manipulating the cytosolic face of the channel.

As expected, cellular GluD2 does not basally have ligand-gated responses to D-serine (Fig. 4a, inset i) nor dopamine (Fig. 4b, inset i). Next, on the same patch, we stimulated GPCR activation with the Gβγ binding peptide (mSIRK) to stimulate Gα release from Gαβγ⁶⁷. Correspondingly, we observe cationic currents after mSIRK treatment in the presence of D-serine or dopamine (Fig. 4a,b, inset ii). Next, we washed in NASPM, which attenuates the currents (Fig. 4a,b, inset iii), as expected based on our single-channel recordings (Fig. 1). In sum, we observe GluD2 is inactive in the absence of mSIRK with a P_O of 0.2 ± 0.2% and 0.5 ± 0.3% in the presence of D-serine and dopamine, respectively, which increases to 9.2 ± 0.8% and 7.7 ± 1.3% during mSIRK treatment, and is then attenuated to 0.2 ± 0.1% and 1.0 ± 0.4% from NASPM (Fig. 4c,d).

Fig 4. |. Cellular modulation of GluD2 by G-proteins.

a,b, representative unitary current traces recorded from inside-out patches with D-serine (a) or dopamine (b) without (i), or with intracellular mSIRK (ii) or mSIRK + NASPM (iii). c, Mean Po ± SD of baseline D-serine currents, baseline: 0.2 ± 0.2; intracellular mSIRK: 9.2 ± 0.8; intracellular mSIRK + NASPM: 0.2 ± 0.1, n = 7. Baseline vs mSIRK P value <0.0001, 95% CI of difference −11.4 to −6.6, df = 6.6; Baseline vs mSIRK + NASPM P value <0.0001, 95% CI of difference 6.5 to 11.5, df = 6.2. d, Mean Po ± SD of baseline dopamine currents, baseline: 0.5 ± 0.3; intracellular mSIRK: 7.7 ± 1.3; intracellular mSIRK + NASPM: 1.0 ± 0.4, n = 5. Baseline vs mSIRK P value 0.016, 95% CI of difference - 12.3 to −2.1, df = 4; Baseline vs mSIRK + NASPM P value 0.013, 95% CI of difference 2.2 to 11.3, df=4. e,f, representative unitary current traces recorded from inside-out patches with D-serine (e) or dopamine (f) without (i), or with intracellular Gα (ii) or Gα + NASPM (iii). g, Mean Po ± SD of baseline D-serine currents, baseline: 0.4 ± 0.1; intracellular Gα: 12.3 ± 1.7; intracellular Gα + NASPM: 0.5 ± 0.1, n = 8. Baseline vs Gα P value 0.0006, 95% CI of difference −17.0 to −6.9, df = 7; Baseline vs Gα + NASPM P value <0.0001, 95% CI of difference 6.9 to 16.6, df = 7. h, Mean Po ± SD of baseline dopamine currents, baseline: 0.07 ± 0.03; intracellular Gα: 12.9 ± 3.1; intracellular Gα + NASPM: 0.9 ± 0.7, n = 5. Baseline vs Gα P value 0.030, 95% CI of difference −23.6 to −1.9, df = 4; Baseline vs Gα + NASPM P value 0.027, 95% CI of difference 2.0 to 21.8, df = 4. *P<0.05; **P<0.01; ***P<0.001; ****P<0.0001.

To further verify that our observed cellular currents are due to Gα binding to GluD2, we repeated these inside-out patch experiments in the presence of our purified Gα, which binds to GluD2 in vitro and relieves Gβγ inhibition of ligand-gating (Fig. 3). Akin to the mSIRK treatment, Gα enables GluD2 ionotropic behavior within cells in the presence of D-serine (Fig. 4e, insets i-ii) and dopamine (Fig. 4f, insets i-ii), and these currents are attenuated by NASPM (Fig. 4e,f, inset iii). GluD2 P_O increases to 12.3 ± 1.7% from 0.4 ± 0.1% and to 12.9 ± 3.1% from 0.07 ± 0.03% in the presence of D-serine and dopamine, respectively, and NASPM reduces the P_O to 0.5 ± 0.1% and 0.9 ± 0.7%. Finally, we confirmed that our cellular currents from GluD2 (with mSIRK or Gα internally) are dependent on D-serine or dopamine, and do not occur in the presence of glutamate, which does not activate GluDs (Extended Data Fig. 7).

These data, in combination with our in vitro reconstitutions, suggest that the GPCR-dependence of GluD currents^22–25,33 is from direct regulation of GluDs via G-proteins.

Mechanism of the lurcher mutation

For decades, Lc GluDs have been the standard for interrogating cellular GluD function. GluD2_Lc (GluD2 A654T) has a missense mutation at the top of the ion channel helix at an essential point for channel gating⁴. This mutation decouples the LBD from the TMD, enabling ion channel opening in the absence of ligand³⁰. We hypothesized that GluD2_Lc has spontaneous channel activity because it is uncoupled from channel silencing by Gβγ.

One potential idea for such a mechanism is that the GluD2_Lc does not interact with Gβγ. To test this, we purified human GluD2_Lc and tested whether it interacts with Gβγ. Surprisingly, GluD2_Lc pulls down Gβγ like normal GluD2, and even inactive Gαβγ (Fig. 5a). Next, we tested whether GluD2_Lc is inhibited by reconstituting GluD2_Lc-Gβγ into brain lipid bilayers for single-channel recordings. Apo GluD2_Lc behaves as expected in bilayers, where we observe leak currents in the absence of ligand (Fig. 5b). However, we also observe similar leak currents with GluD2_Lc-Gβγ, suggesting that while GluD2_Lc can interact with Gβγ, it is not inhibited by it like wild-type GluD2.

Fig 5. |. Reconstitution of G-proteins with GluD2_Lc.

a, SDS-PAGE gel of G-protein affinity pulldown by GluD2_Lc-eGFP-StreptagII. b, Example of a GluD2_Lc current trace in the absence of ligand at 37°C, n = 4 traces. c, Example of a GluD2-Gβγ current trace for WT GluD2 with 10 mM D-serine (blue, n = 14 traces) and GluD2_Lc in the absence of ligand (red, n = 3 traces).

Model of G-protein channel tuning

GluDs are ligand-gated ion channels that exhibit canonical ligand-gating in isolation (Fig. 6a). Neurotransmitters such as dopamine, D-serine, and GABA can bind to the LBDs, which in turn open the M3 ion channel helices. Gα binds to GluD2, does not inhibit activation, but may attenuate the channel conductance. Gβγ inhibits the receptor from ligand-gating, which can be rescued by Gα binding. Surprisingly, GluD2 forms a complex with inactive heterotrimer Gαβγ, which positively augments the channel activation and conductance. Most GluDs in cells are silenced, which we expect to be from Gβγ (Fig. 6b). However, Gα can bind and prime GluD2 for ligand-gating in GluD2-Gβγ-Gα complexes.

Fig 6. |. Proposed mechanism for G-protein modulation of GluDs.

a, In isolated preparations, GluD receptors are not bound to G-proteins and exhibit traditional ligand-gated ionotropic properties by neurotransmitters (e.g., D-serine, GABA, and dopamine). b, Cellular GluDs are silenced by Gβγ, which directly binds to the receptor, preventing ligand-gated currents. This silencing is alleviated by GluD associating with Gα, recovering ionotropic properties. c, Due to decoupling of the TMD and LBD, GluD_Lc is not affected by Gβγ silencing despite its binding, resulting in constitutive currents from both isolated and cellular preparations.

Based on our findings with GluD2_Lc, we hypothesize that silenced GluD2 receptors are allosterically decoupled from ligand-gating when just Gβγ is present. GluD2_Lc is unique from wildtype receptors in that, because of its mutation, it has a TMD that is functionally decoupled from ligand-gating; its ion channel helices can open in the absence of ligand (Fig. 6c). Thus, even when Gβγ is bound, it is not inhibited because the allosteric network is disrupted.

Discussion

Dopamine has long been understood to not directly activate vertebrate ligand-gated ion channels^1–3. However, here we show that dopamine directly elicits cationic currents via GluDs in vitro and in cells. Thus, there is now evidence to support an expanded molecular community for dopamine function, and a mechanism for direct excitation by dopamine.

However, despite the colocalization, the observations of GluD activation by dopamine have been overlooked. The reasons for this are likely two-fold: GluDs have only recently been defined as bona fide ligand-gated ion channels, and as we demonstrate here, their ligand-gating function is highly regulated by G-proteins.

G-proteins have multimodal effects on GluD signaling. A key future direction will be how the diverse G-protein community, including other alpha subunits such as the stimulatory (G_s), G_q/11, and G_12/13 families, tunes GluD signaling across different cellular contexts, neurotransmitter responses, and disease mutations.

Of interest is that GluD2 forms a complex with inactive Gαβγ. It has been suggested that GluDs may signal through metabotropic means by acting as trans-synaptic signal transduction devices⁶⁸. One possibility is that trans-synaptic proteins such as the cerebellins and neurexins, which are tightly correlated with GluDs throughout the brain^27,29,69,70, modulate the G-protein regulation of GluD channel activity. A second possibility is that the trans-synaptic proteins stimulate activation of Gαβγ in a GluD-Gαβγ complex, effectively making GluDs an ionotropic-metabotropic hybrid receptor.

There is now evidence that GluDs, once considered orphan receptors, do not bind glutamate but do bind neurotransmitters such as dopamine and GABA, as well as the gliotransmitter D-serine. Thus, GluDs likely have a central role in the brain, which has long been hinted at by studies pointing to their role in regulating glutamatergic, GABAergic, and dopaminergic signaling. However, the key to unlocking the molecular function is their G-protein regulators, which render GluDs as conditional ionotropic receptors. Understanding the depth of GluD regulation by G-proteins throughout brain tissue will be a key step to reclassifying the roles of GluDs in the future.

Methods

Construct design.

The DNA coding sequence of full-length human GluD2 (UniProt ID O43424–1) was fused to a Thr-Gly-Gly linker, a thrombin cleavage site (LVPRGS), an enhanced GFP (eGFP) sequence, a Ser-Gly-Leu-Arg-Ser linker, a Strep-Tag II (WSHPQFEK) sequence, and a stop codon at its C-terminus. The construct was introduced into a pEG-BacMam vector for baculovirus-driven protein expression in HEK293 cells⁷¹. The GluD2 Lurcher (A654T) mutant was generated by site-directed mutagenesis. The human Gα_i1 sequence (Uniprot ID P63096) with an amino-terminal 6× histidine tag, a HRV 3C protease cleavage site (LEVLFQGP), and a flexible glycine/serine linker were inserted into a pET21a plasmid (gifted by Dr. Daniel Hilger)⁷². Human Gβ_1γ2 sequences (Uniprot ID P62873 for Gβ₁ and P59768 for Gγ₂) were used to express and purify the G-protein. The Gβ1 sequence was preceded by an N-terminal 6x histidine-tag, an HRV 3C protease cleavage site, and a Gly-Ser-Ser-Gly glycine/serine linker⁷².

Expression and purification of human GluD2.

Bacmids and baculoviruses of the full-length GluD2 and GluD2 Lurcher mutant were made using standard methods⁷³. In short, each cloned pEG-BacMam plasmid carrying a specific hGluD2 construct was transformed into DH10Bac (Gibco, cat# 10301012) to generate bacmids. Those bacmids were then transfected into Expisf9 cells (Gibco, A35243) at 2.5 million cells per mL concentration using ExpiFectamine Sf transfection reagent (Gibco, A38915). The transfected cells were cultured at 27 °C with constant shaking. After 7 days, supernatant of the cultured cells containing P1 baculoviruses was harvested and stored at 4 °C until further usage. To increase the virus titer, P2 viruses were further gathered: 1 mL of the P1 viruses was added to 200 mL of fresh Expisf9 cells at a concentration of 1 million cells per mL. The infected cells were incubated at 27 °C with constant shaking. The P2 viruses were isolated after 7 days via ultracentrifugation of the supernatant (41,300 xg for 60 mins at 4 °C) and resuspended with 40 mL of Expi293 expression media (ThermoFisher, A1435102). The concentrated P2 viruses were stored at 4 °C until further usage.

For protein expression in mammalian cells, 20 mL of the concentrated P2 virus was added to Expi293F cells (Gibco, A14527) at ~ 3 to 4 million cells per mL concentration. The infected Expi293F cells were incubated at 37 °C with 5% CO₂ and constant shaking. At ~16 hours post-infection, the cells were treated with 10 mM sodium butyrate (Sigma, 303410) and transferred to a 30 °C, 5 % CO₂ incubator. At 72 hours post-infection, the cells were harvested by centrifugation (500 xg for 20 mins at 4 °C), washed with 1x PBS, and spun down again. The cell pellets were snap-frozen in liquid nitrogen and stored at −80 °C until further usage.

For purification, the frozen cell pellets were resuspended at 4 °C in lysis buffer (20 mM Tris-HCl, pH 8.0, and 150 mM NaCl) with protease inhibitors (0.8 μM aprotinin, 2 μg/mL leupeptin, 2 μM pepstatin A, and 1 mM phenylmethylsulfonyl fluoride). The resuspended cells were then lysed on ice using a Misonix sonicator for four cycles (Fisher Scientific, 1s on and 1s off for 1 min per cycle, power level 7). The supernatant was collected by centrifugation (2,500 xg for 20 mins at 4 °C), and cell membranes were further isolated by ultracentrifugation (125,000 xg for 45 mins at 4 °C). The pelleted membranes were then resuspended in lysis buffer (20 mM Tris-HCl, pH 8.0 and 150 mM NaCl) and mechanically homogenized before solubilized in solubilization buffer (20 mM Tris-HCl, pH 8.0 and 150 mM NaCl, 1% n-dodecyl-β-d-maltopyranoside (Anatrace, D310) and 0.2% brain extract total (Avanti Research, 131101P) as the final concentration) for 2 hours at 4 °C with constant stirring. Insoluble fractions were further removed by ultracentrifugation (125,000 xg for 45 mins at 4 °C). The supernatant containing the solubilized membrane proteins was incubated with 2 mL of Strep-Tactin XT 4Flow resin (IBA, 2–5010) overnight at 4 °C with constant stirring. The following day, the resin was collected via a gravity column and washed with 10 column volumes (CVs) of glyco-diosgenin (GDN) buffer (150 mM NaCl, 20 mM Tris pH 8.0, and 0.01% GDN [Anatrace, GDN101]).

Bound proteins were eluted with GDN buffer containing 50 mM biotin (Thermo Scientific, PI29129) and concentrated down to 500 μl at 4 °C using a 100 kDa molecular weight cut-off concentrator (Millipore Sigma, UFC9100). The concentrated sample was then digested with thrombin (1:200 mass ratio of thrombin to eluted protein) for 1 hour at room temperature to remove the eGFP and StrepTagII. For the pulldown experiments, the thrombin digestion was skipped to preserve the tag. The protein samples (cleaved or uncleaved) were further separated via size-exclusion chromatography on a Superose 6 Increase 10/300 GL column (Cytiva, 29091596) equilibrated with the GDN buffer. The peak fractions corresponding to GluD2 tetramers were pooled and concentrated up to ~ 4 mg/mL (Millipore Sigma, UFC5100) for cryo-EM specimen preparation. The samples were further ultracentrifuged to remove insoluble material after concentration. For the pulldown experiments, the protein sample was concentrated up to 8–10 mg/mL and snap-frozen in liquid nitrogen, stored in the −80 °C freezer until further usage.

Expression and purification of human G-proteins.

Human Gα_i subunit was expressed and purified as previously described⁷². The pET21a plasmid carrying the Gα_i sequence was transformed into BL21(DE3) cells (NEB, C2527H). The cells were grown at 37 °C in Terrific Broth (Supelco, T0918) to an O.D.600 of 0.6, and protein expression was induced with 0.5 mM IPTG (Invitrogen, 15529019). The induced cells were incubated at 25 °C overnight with constant shaking (210 rpm) before being harvested and resuspended in lysis buffer (50 mM HEPES pH 7.5, 100 mM NaCl, 1 mM MgCl₂, 50 μM GDP, 5 mM β-mercaptoethanol (βME), 5 mM imidazole, and protease inhibitors) the next day. The resuspended cells were lysed on ice by sonication using 50% power and 50% duty cycle for four times 60s. The supernatant was collected by centrifugation (2,500 xg for 20 mins at 4 °C) and incubated with 5 mL of Talon resins (TaKaRa, 635503) for an hour. The bound resins were washed with 10 CV with the lysis buffer supplemented with 20 mM imidazole in a gravity column before the protein was eluted with the lysis buffer containing 250 mM imidazole. The eluted protein solution was dialyzed overnight in dialysis buffer (20 mM HEPES pH 7.5, 100 mM NaCl, 1 mM MgCl₂, 20 μM GDP, 5 mM βME, and 5 mM imidazole). The histidine tag was cleaved by 3C protease (w/w 1:100), and 10 μL calf intestinal alkaline phosphatase (NEB, M0525S) was added to remove phosphorylation during the dialysis. The next day, uncleaved protein and 3C protease were removed by incubation with Talon resin for 1 hour at 4 °C. The flowthrough was concentrated down to 500 μL and run on a Superdex 200 Increase 10/300 GL column (Cytiva, 28990944) with SEC buffer (20 mM HEPES pH 7.5, 100 mM NaCl, 1 mM MgCl₂, 20 μM GDP, and 100 μM TCEP). The corresponding fractions were pooled, concentrated, and stored at −80 °C until further usage.

The human Gβ₁γ₂ heterodimer was expressed in Trichoplusia ni (High Five) insect cells (Gibco B85502) using a single baculovirus encoding the Gβ₁ and Gγ₂ sequences generated by the BestBac method (Expression Systems)^72,74. The cells were infected at a density of 3–4 × 10⁶ cells per mL and incubated at 130 rpm and 27 °C for ~ 60 hours. The cells were harvested by centrifugation (550 xg for 15 mins at 4 °C), washed with 1x PBS, and spun down again. The cell pellets were snap-frozen in liquid nitrogen and stored at −80 °C until further usage. The harvested cell pellets were resuspended in lysis buffer (10 mM Tris, pH 7.5, 5 mM βME, and protease inhibitors) for 1.5 hours. The membrane fraction was collected by ultracentrifugation (21,000 xg for 45 mins at 4 °C) and homogenized manually using a glass dounce homogenizer (Bellco Glass, 1984–10100) with solubilization buffer (20 mM HEPES pH 7.5, 100 mM NaCl, 1.0% sodium cholate, 0.05% DDM, 5 mM βME, and protease inhibitors). The solubilization reaction was further incubated at 4 °C for 1 hour with constant stirring. Insoluble debris was removed by ultracentrifugation (21,000 xg for 45 mins at 4 °C). The supernatant was incubated with 4 mL talon resins for 1 hour at 4 °C before washing with 10 CV of the solubilization buffer supplemented with 20 mM imidazole. The bound protein was eluted with elution buffer (20 mM HEPES pH 7.5, 100 mM NaCl, 0.1% DDM, 5 mM βME, 250 mM imidazole). The eluted protein was treated with 3C protease (w/w 1:10) and 10 μL calf intestinal alkaline phosphatase (NEB, M0525S) and dialyzed overnight in 20 mM HEPES pH 7.5, 100 mM NaCl, 0.1% DDM, 5 mM βME, and 5 mM imidazole. The uncleaved protein and 3C protease were removed by incubating the overnight solution with 2 mL Talon resins. The flowthrough solution was collected and concentrated and run on a Superdex 200 Increase 10/300 GL column (Cytiva, 28990944) with SEC buffer (20 mM HEPES pH 7.5, 100 mM NaCl, 1 mM MgCl₂, 20 μM GDP, and 100 μM TCEP). The corresponding fractions were pooled, concentrated and stored in −80 °C freezer until further usage.

Human Gα_iβ₁γ₂ heterotrimer was expressed and purified as previously described⁷². In brief, the heterotrimer was expressed and purified similarly to the Gβ₁γ₂ heterodimer, with the following differences. Two baculoviruses, one encoding the wild-type Gα_i1 sequence and the other encoding the His-tagged Gα₁ and Gγ₂ sequences, were generated by the BestBac method (Expression Systems). These viruses were used to co-transduce High Five insect cells (Gibco B85502) at a density of 3–4 × 10⁶ cells per mL and incubated at 130 rpm and 27 °C for ~ 48–60 hours. During protein expression, all buffers were supplemented with GDP (10 μM) and MgCl₂ (100 μM for lysis, 5 mM for solubilization, and 1 mM for dialysis and anion exchange buffers). Following reverse Ni affinity chromatography to remove uncleaved, His-tagged protein and 3C protease, the flowthrough solution, supplemented with 1 mM manganese chloride, was collected and treated with CIP (NEB, M0525S), Antarctic phosphatase (NEB, M0289S), and λ phosphatase (NEB, P0753S) for 1 hour at 4 °C to dephosphorylate the protein. After adjusting the NaCl concentration to 50 mM by adding dilution buffer composed of 20 mM HEPES pH 7.4, 1 mM MgCl₂, 10 μM GDP, 100 μM TCEP, and 0.05% DDM, the protein was loaded onto a Capto HiRes Q 10/100 (Cytiva) anion exchange column equilibrated in buffer A composed of 20 mM HEPES pH 7.4, 50 mM NaCl, 1 mM MgCl₂, 10 μM GDP, 100 μM TCEP, and 0.05% DDM. The column was washed with buffer A, and bound protein was eluted over a linear gradient from 0 to 50% buffer B, which is near-identical to buffer A, but contains 1 M NaCl. Eluted lipidated G protein heterodimer was concentrated using a 30-kDa MWCO Amicon spin concentrator (MilliporeSigma), flash frozen, and stored at −80°C until future use.

Pulldown of G-proteins.

7 μM GluD2 (GluD2 or GluD2_Lc) was incubated with 70 μM G protein components (Gα_i, Gβ₁γ₂, or Gα_iβ₁γ₂, 1:10 molar ration) in a 50 μL reaction with buffer (20 mM HEPES pH 7.5, 100 mM NaCl, 1 mM MgCl₂, 20 μM GDP, and 100 μM TCEP, 0.01% GDN) at 37 °C for 1 hour. Each reaction was then diluted to 200 μL in a 1 mL Eppendorf tube and incubated with 100 μL of strep-tactin resin at room temperature for 1 hour with constant mixing. The bound resins were washed three times with 200 μL buffer. For elution, the resins were incubated with 200 μL of reaction buffer supplemented with 50 mM biotin for 30 mins twice at room temperature. A total of 400 μL elution for each reaction was then pooled and concentrated using a 30K concentrator (Thermo Scientific, 88502) before running on a 4–20% Tris-glycine denaturing gel (ThermoFisher, WXP42012BOX). 1 mM GTPγS was added in the specific samples as indicated throughout the entire pulldown process (incubation/wash/elution).

Cryo-EM sample preparation and data collection.

Homemade grids were used to prepare the cryo-EM samples in this study. In brief, C-flat holey carbon grids (Electron Microscopy Sciences, CF213–50-Au, Electron Microscopy Sciences) were first coated with 50 nm gold by a sputter coater (Leica EM ACE600) and then plasma cleaned using a Tergeo Plasma Cleaner (Pie Scientific) with Ar/O₂ to make the final 1.2/1.3 gold grids with gold mesh, based on previously published methods^75,76. The grids were glow-discharged for 120 s with 25 mA current and 10 s hold time in a Pelco Easiglow (Ted Pella, 91000) before sample application.

20 μl of the concentrated GluD2 sample was first pre-incubated in a thermocycler set at 37 °C for 10 mins before use. Then, a 3-μl protein sample was spiked with 0.5 μl of the 70 mM dopamine (Sigma-Aldrich, H8502; final dopamine concentration, 10 mM), and mixed vigorously before being applied to freshly-discharged grids in a FEI Vitrobot Mark IV (Thermo Fisher Scientific) chamber set at 37 °C and 90 % humidity (wait time 30 s; blot force 5 N; blot time 4 s). The grids were then immediately plunge-frozen in liquid ethane.

A total of 7,244 micrographs were collected on a 200-kV Glacios microscope (ThermoFisher Scientific) equipped with a Falcon 4i camera. The micrographs were collected with a total dose of 40.00 e⁻/Ų, a dose rate of 9.93 e⁻/pixel/s with a pixel size of 1.2 Å/pixel, and the defocus range was set from −0.5 μm to −2.9 μm.

Image processing.

CryoSPARC⁷⁷ (version 4.6.2) was used for processing sample images while final particle picking was performed with TOPAZ⁷⁸. Details can be found in Extended Data Figures 3&4.

Model building, refinement, and structural analyses.

ChimeraX⁷⁹, Coot⁸⁰, ISOLDE⁸¹ and PHENIX⁸² compiled by the SBgrid Consortium⁸³ were used in combination to perform model building, refinement, and structural analysis. All visualizations and measurements were performed in ChimeraX. Model quality was assessed with MolProbity⁸⁴.

Bilayer recording and analyses.

Unitary currents were recorded from purified GluD receptors using an Orbit mini with a temperature control unit (Nanion Technologies), following previously established methods^4,5. Before all recordings, the Orbit mini was calibrated with a standard test cell chip (Nanion Technologies) to normalize baseline currents to 0 pA. A MECA recording chip (4× 100 μm cavities, Nanion Technologies, 132002) was then plugged into the Orbit mini and filled with recording solution (150 mM KCl, 20 mM HEPES, pH 7.4). A lipid bilayer with a capacitance of 10–25 pF was then formed over each recording cavity using a 10 μL pipette tip dipped in 100 mg/mL brain extract total (Avanti, 131101C) dissolved in decane (Sigma-Aldrich, D901–100ML). 1 μL of purified apo protein diluted to ~ 4 μg/mL was added on top of each painted recording cavity. D-serine or dopamine was then diluted in the recording solution to the noted concentrations (0.1, 1, 10, and 100 mM). To measure pharmacological block of dopaminergic currents, 100 μM NASPM (Tocris, 2766) was co-applied with dopamine. All recordings were conducted with a holding potential (V_hold) of −80 mV and a maximum applied current of 200 pA at 37 °C. Currents were low-pass filtered at 0.20 kHz, sampled at 20 kHz, and written into digital files using the Elements Data Reader 4 (EDR4, Elements) software v.1.7.1.

Currents were analyzed using Clampfit software version 11.3 (Molecular Devices). Currents were filtered with an 8-pole Bessel filter with a −3 dB cut-off frequency of 100 Hz and adjusted for baseline before analysis. These data were then idealized using the half-amplitude algorithm built into the Single-Channel Search function in Clampfit. Five detection levels (−2, −4, −6, −8, and −10 pA) were predefined, and an automatic adjustment for baseline and simultaneous update for all levels (with a 10% level contribution) function was used to correct for baseline drift. We omitted any event lasting less than 0.5 ms. The corresponding current amplitude values (i) and dwell times for each event were then used to calculate unitary point conductance (γ), mean open dwell time (MOT), mean closed dwell time (MCT), and open probability (P_O). P_O values were calculated using the dwell times for closed and open events across files for each condition. These values were estimated by calculating P_O within a 10 s sliding window in RStudio 2026.01.1, and mean P_O values were compared across conditions. γ was calculated by the following equation:

$$\gamma =\frac{i}{{\text{V}}_{\text{hold}}}$$

All the following analyses were done using GraphPad Prism 11. Open conductance histograms were created from frequency distributions of open event current amplitudes, and Gaussian distributions were fit to individual open-levels (O1-O4) with a nonlinear regression of the following equation:

$$y={\text{y}}_{\text{max}}{e}^{-0.5{(\frac{x-\mathit{\text{mean}}}{\mathit{\text{SD}}})}^2}$$

Where SD is the standard deviation, and mean is the mean current amplitude.

The Robust regression and Outlier (ROUT) function in Prism 11, with a default setting of coefficient Q = 1%, was used to identify the outliers in each data set. Significance of difference was set to α = 0.05 and was determined using Brown-Forsythe and Welch’s one-way ANOVA test or by Welch’s t-test, as noted, assuming variances among the experimental groups were unequal. P-values were corrected for multiple comparisons using the Game–Howell post-hoc test.

Inside-out recordings and whole-cell recordings.

HEK293T cells were maintained in Dulbecco’s modified Eagle’s medium (GenDEPOT) supplemented with 10% fetal bovine serum (GenDEPOT) and penicillin–streptomycin (Invitrogen). Cells were transiently transfected with GluD2 (human GluD2-eGFP-StrepTag-II) using iMFectin Poly DNA Transfection Reagent (GenDEPOT) according to the manufacturer’s instructions.

Recordings were performed in the inside-out patch-clamp configuration. Patch pipettes (8–12 MΩ) were filled with an external solution containing 150 mM NaCl, 4 mM KCl, 2 mM CaCl₂, 10 mM HEPES, and either 10 mM D-serine or 10 mM dopamine (pH 7.4). The internal (bath) solution contained 135 mM CsF, 33 mM CsOH, 2 mM MgCl₂, 1 mM CaCl₂, 11 mM EGTA, and 10 mM HEPES. Recordings were performed at a holding potential of −100 mV. 25 μM mSIRK (Sigma-Aldrich, 371818), 70 μM Gi, and 300 μM NASPM were locally perfused around the patch. Currents were acquired at 50 kHz and low-pass filtered at 5–10 kHz using an Axopatch 200B amplifier and Digidata 1550A digitizer (Molecular Devices). For analysis, traces were further filtered at 1 kHz⁸⁵.

Single-channel events were detected and analyzed in Clampfit (Molecular Devices) using the single-channel search function. Open and closed states were identified by amplitude threshold crossing, and traces were idealized accordingly. P_O was calculated as the total open time divided by the total recording duration.

For the GluD2_Lc cellular recordings, mouse GluD2 (A654T) was transfected as described above into HEK293T Cells with iMFectin at a 1:0.2 ratio per 35-mm dish. Patch pipettes (3–5 MΩ) were filled with an internal solution containing 135 mM CsF, 33 mM CsOH, 2 mM MgCl₂, 1 mM CaCl₂, 11 mM EGTA, and 10 mM HEPES. The extracellular solution contained 150 mM NaCl, 4 mM KCl, 2 mM CaCl₂, 10 mM HEPES, and dopamine (pH 7.4). External solutions were locally applied to lifted cells using a SF-77B perfusion fast-step (Warner Instruments). Whole-cell patch-clamp recordings were performed 24 h post-transfection at a holding potential of −60 mV. Currents were sampled at 10 kHz and low-pass filtered at 5 kHz using a 16-bit A/D converter (Digidata 1550A, Molecular Devices) controlled by pCLAMP 10 software (Molecular Devices).

The statistical analyses were performed using GraphPad Prism 11. Significance of difference was set to α = 0.05 and was determined using a paired One-way ANOVA with the Game–Howell correction, assuming variabilities among the experimental groups were unequal. Tukey’s multiple comparison post-hoc test was performed to correct P-values for multiple comparisons.

Supplementary Material

Supplemental Information The online version contains supplemental information.

Data Availability

The cryoEM reconstructions are deposited into the Electron Microscopy Data Bank (EMDB). The whole maps are the primary cryoEM maps in each deposition, and each local map, as applicable, and half maps are supplied as supplemental files in each deposition. All structural coordinates are deposited in the Protein Data Bank (PDB). The accession codes for each state (EMDB, PDB) will become available after publication.