Disrupting D1-NMDA or D2-NMDA receptor heteromerization prevents cocaine's rewarding effects but preserves natural reward processing

Abstract

Addictive drugs increase dopamine in the nucleus accumbens (NAc), where it persistently shapes excitatory glutamate transmission and hijacks natural reward processing. Here, we provide evidence, from mice to humans, that an underlying mechanism relies on drug-evoked heteromerization of glutamate N-methyl-d-aspartate receptors (NMDAR) with dopamine receptor 1 (D1R) or 2 (D2R). Using temporally controlled inhibition of D1R-NMDAR heteromerization, we unraveled their selective implication in early phases of cocaine-mediated synaptic, morphological, and behavioral responses. In contrast, preventing D2R-NMDAR heteromerization blocked the persistence of these adaptations. Interfering with these heteromers spared natural reward processing. Notably, we established that D2R-NMDAR complexes exist in human samples and showed that, despite a decreased D2R protein expression in the NAc, individuals with psychostimulant use disorder display a higher proportion of D2R forming heteromers with NMDAR. These findings contribute to a better understanding of molecular mechanisms underlying addiction and uncover D2R-NMDAR heteromers as targets with potential therapeutic value.

Fig. 1.. Detection and temporally controlled disruption of striatal D1R-NMDAR and D2R-NMDAR heteromerization in vivo.

(A) Example images of D1R-NMDAR heteromer PLA detection and related negative (Neg. cont) showing the absence of signal when one primary antibody is omitted. (B) Top: Viral-based strategy for expression of interfering peptide to disrupt D1R-NMDAR interaction [Tet-On-GluN1C1 (C1)] and control [Tet-On-GluN1C1Δ (C1Δ)] in the NAc. Bottom: Example image of dox (+dox)–mediated RFP expression [anterior commissure (ac)]. TRE, tetracycline response element; P2A, 2A peptide; PGK, phosphoglycerate kinase 1 promoter; rtTA, reverse tetracycline-controlled transactivator. (C) Example image of D1R-NMDAR heteromer detection by PLA in C1Δ- and C1-transduced neurons. (D) Quantifications of the fluorescent PLA signal in C1Δ- and C1-transduced neurons. Two-sided Student’s t test, t = 4.078, df = 13, **P = 0.0013, n = 7 to 8 cells from four mice per group. (E) Neuronal survival of C1Δ- and C1-transduced neurons. Two-sided Student’s t test, t = 0.354 df = 6, P = 0.735, n = 4 mice per group. (F) Same as in (A), for D2R-NMDAR heteromer PLA detection. (G) Viral strategy for expression of interfering peptide to prevent D2R-NMDAR heteromerization [Tet-On-D2R-IL3 (IL3)] and control [Tet-On-D2R-IL3-scr (IL3-scr)]. (H) Representative image D2R-NMDAR heteromer detection by PLA in IL3-scr– and IL3-transduced neurons. (I) Quantifications of the fluorescent PLA signal in IL3-scr– and IL-3 infected neurons. Two-sided Student’s t test, t = 2.393, df = 25, *P = 0.0246, n = 11 to 16 cells from 4 mice per group. (J) Neuronal survival of IL3-scr– and IL3-transduced neurons. Two-sided Student’s t test, t = 0.2767 df = 6, P = 0.7913, n = 4 mice per group. Scale bars, 10 μm. Error bars denote SEMs.

Fig. 2.. Behavioral sensitization to cocaine is associated with transient D1R-NMDAR heteromerization and prolonged D2R-NMDAR heteromerization in the NAc.

(A) Experimental time frame [NAc core and NAc shell and measurements of locomotor activity before and during 5 days of saline or cocaine (15 mg/kg) injections]. Two-way analysis of variance (ANOVA): treatment effect, F_1,24 = 67.79, ***P < 0.0001 saline versus cocaine on day 5, n = 13 mice per group. DM, dorso-medial striatum; DL, dorso-lateral striatum. (B) Detection and quantifications of D1R-NMDAR heteromerization in saline- and cocaine-treated groups. PLA signal is represented as fold increase normalized to the saline group. Two-sided Student’s t test: *P < 0.05, **P < 0.01, and ***P < 0.001 (saline versus cocaine); n = 28 to 84 fields of view per structure (NAc core, 4; NAc shell, 12; DM, 6; DL, 6; fields of view per mice) from seven mice per group. (C) Same as for (B) for D2R-NMDAR heteromerization. (D) Experimental time frame, measurements of locomotor activity, and quantifications of D1R-NMDAR and D2R-NMDAR heteromerization. PLA signal is represented as fold increase normalized to the saline group. One-way ANOVA: *P < 0.05, **P < 0.01, and ***P < 0.001 (saline-saline versus cocaine-saline/cocaine-cocaine); #P < 0.05 and ##P < 0.01 (cocaine-saline versus cocaine-cocaine); n = 28 to 84 fields of view per structure from seven mice per group. (E) Same as for (D) except that mice received either a vehicle solution (Veh) or the D1R antagonist SCH23390 (SCH) or the D2R antagonist eticlopride (Etic) before a cocaine challenge. One-way ANOVA: **P < 0.01 and ***P < 0.001 (vehicle-saline versus vehicle-cocaine); #P < 0.05 and ##P < 0.01 (vehicle-cocaine versus SCH-cocaine or cocaine versus Etic-cocaine); n.s., not significant; n = 28 to 84 fields of view per structure from seven mice per group. Scale bars, 10 μm (B and C). Error bars denote SEMs.

Fig. 3.. D1R-NMDAR heteromerization controls cocaine-evoked potentiation of glutamate transmission onto D1R-MSN and the development of behavioral sensitization.

(A) Experimental design and example trace of AMPA+NMDA (black) and NMDA (gray) currents in neurons (asterisk) expressing GFP [i.e., D1R-MSN (see fig. S3A)], and C1 or C1Δ (RFP⁺). (B) A/N ratios. Two-way ANOVA: virus effect, F_1,30 = 2.511, *P = 0.033 and ##P = 0.0061, n = 3 to 4 mice per group and n = 6 to 13 cells per group. (C) Comparison of representative recordings of pharmacologically isolated NMDAR EPSCs, normalized to the peak amplitude (in percentage). (D) Deactivation kinetics of NMDAR EPSCs. Two-way ANOVA: virus effect, F_1,30 = 0.205, P > 0.999, n = 3 to 4 mice per group and n = 8 to 9 cells per group. (E) Top: Experimental time frame and basal locomotor activity. Bottom left: Two-sided Student’s t test, t = 0.332 df = 30, P = 0.742, n = 16 mice per group. Bottom middle: Inhibition of D1R-NMDAR heteromerization during the development of locomotor sensitization. Three-way ANOVA: virus effect, F_1,256 = 13.72, ***P = 0.0003; n.s., P > 0.9999; n = 7 to 11 mice per group. Bottom right: Sensitization scores of cocaine-treated mice injected with Tet-On-GluN1C1Δ (black) or Tet-On-GluN1C1 (blue). One-sample Student’s t test: t = 5.289, df = 10, ###P = 0.0004 (cocaine C1Δ versus 0); t = 2.197; df = 10; n.s., P > 0.05 (cocaine C1 versus 0). (F) Top: Experimental time frame. Bottom left: Inhibition of D1R-NMDAR heteromerization during 7 days of abstinence, followed by a challenge cocaine injection. Three-way ANOVA: virus effect, F_1,243 = 0.6160, P > 0.9999, n = 7 to 8 mice per group. Bottom right: Sensitization scores of cocaine-treated mice injected with Tet-On-GluN1C1Δ (black) or Tet-On-GluN1C1 (blue). One-sample Student’s t test, t = 5.289, df = 7, #P = 0.0325 (cocaine C1Δ versus 0); t = 2.599, df = 6, ΔP = 0.0407 (cocaine C1 versus 0). Error bars denote SEMs.

Fig. 4.. D2R-NMDAR heteromerization is involved in the development and maintenance of cocaine’s sensitizing effects.

(A) Experimental time frame and representative traces of AMPA + NMDA (black) and NMDA (gray) currents in neurons expressing GFP [i.e., D2R-MSN (see fig. S3B)] and IL3 or IL3-scr (RFP⁺) (B) A/N ratios. Two-sided Student’s t test: t = 0.3397, df = 21, P = 0.7375, n = 4 to 6 mice per group and n = 11 to 12 cells per group. (C) Comparison of representative recordings of pharmacologically isolated NMDAR EPSCs, normalized to the peak amplitude (in percentage). (D) Deactivation kinetics of NMDA EPSCs. Two-sided Student’s t test: t = 0.5129, df = 21, P = 0.6134, n = 11 to 12 cells to group. (E) Tet-On-D2R-IL3 spares quinpirole-induced inhibition of forskolin-induced accumulation of cAMP. LogIC₅₀ is −9.78 for D2R-IL3 and −9.65 for D2R-IL3-Scr. n = 3 independent experiments per condition. (F) Top: Experimental time frame. Bottom left: Basal locomotor activity. Two-sided Student’s t test, t = 0.994, df = 30, P = 0.3282, n = 16 mice per group. Bottom middle: Inhibition of D2R-NMDAR heteromerization blunts the development of locomotor sensitization. Three-way ANOVA: virus effect, F_1,192 = 1.984; n.s., P > 0.9999; n = 6 to 8 mice per group. Bottom right: Sensitization scores of cocaine-treated mice injected with Tet-On-D2R-IL3-scr (black) or Tet-On-D2R-IL3 (green). One-sample Student’s t test, t = 3.258, df = 7, #P = 0.0139 (cocaine IL3-Scr versus 0); t = 1.782, df = 7, P > 0.05 (cocaine IL3 versus 0). (G) Top: Experimental time frame. Bottom left: Impact of D2R-NMDAR heteromer inhibition on the maintenance of cocaine-evoked locomotor sensitization. Three-way ANOVA: virus effect, F_1,198 = 7.278, *P = 0.0330, n = 6 mice per group. Bottom right: Sensitization scores of cocaine-treated mice injected with Tet-On-D2R-IL3-scr (black) or Tet-On-D2R-IL3 (green). One-sample Student’s t test: t = 4.170, df = 5, ##P = 0.0087 (cocaine IL3-Scr versus 0); t = 0.00, df = 5, P > 0.05 (cocaine IL3 versus 0). Error bars denote SEMs.

Fig. 5.. Differential roles of D1R-NMDAR and D2R-NMDAR heteromerization in controlling the rewarding effects of cocaine.

(A) Experimental time frame and CPP score upon inhibition of D1R-NMDAR heteromerization. Two-way ANOVA: virus effect, F_1,59 = 5.281, ***P < 0.0001 and ##P = 0.0040, n = 15 to 16 mice per group. (B) Top: Low magnification images of D1R-MSN (GFP⁺; see fig. S3A) infected (RFP) shown by the asterisks. Scale bar, 10 μm. Bottom: High magnification of dendritic segments. Scale bar, 5 μm. Spine density analysis. Two-way ANOVA: virus effect, F_1,162 = 11.14, *P = 0.0446 and ##P = 0.0043, n = 27 to 69 dendrites from six mice per group. (C) Experimental time frame to study the impact of D1R-NMDAR uncoupling on the extinction and cocaine-induced reinstatement (Reinstat.) of CPP. Two-way ANOVA: virus effect, F_1,24 = 0.004, P > 0.999 (cocaine C1 versus cocaine C1Δ, CPP score on relapse day), n = 10 to 16 mice per group. (D) Same as for (A) upon inhibition of D2R-NMDAR heteromerization. Two-way ANOVA: virus effect, F_1,57 = 2.424, ***P < 0.0001 and #P = 0.0396, n = 14 to 16 mice per group. (E) Top: Low-magnification images of D2R-MSN (GFP; see fig. S3B) infected (RFP) shown by the asterisk. Scale bar, 10 μm. Bottom: High-magnification of dendritic segments. Scale bar, 5 μm. Spine density analysis. Two-way ANOVA: virus effect, F_1,166 = 0.1268; n.s., P > 0.999 (saline IL3-scr versus cocaine IL3-scr); n = 30 to 53 dendrites from six mice per group. (F) CPP score upon inhibition of D2R-NMDAR during CPP during extinction and cocaine-induced reinstatement. Two-way ANOVA: virus effect, F_1,31 = 0.899, **P = 0.0018 (cocaine IL3 versus cocaine IL3-scr CPP score on reinstatement day), n = 16 to 17 mice per group. Error bars denote SEMs.

Fig. 6.. Inhibiting D1R-NMDAR or D2R-NMDAR heteromerization does not alter CPP for food.

(A) Experimental time frame to study the consequences of inhibition of heteromerization on the development of food-induced CPP. (B) Impact of inhibiting D1R-NMDAR heteromerization on the CPP score. Two-way ANOVA: virus effect, F_1,27 = 2.756, ***P = 0.0002 (control C1∆ versus food C1∆); ###P = 0.0002 (control C1 versus food C1); n.s., P > 0.999 (food C1 versus food C1∆); n = 7 to 8 mice per group. (C) Effect of inhibiting D2R-NMDAR heteromerization on the CPP score. Two-way ANOVA: virus effect F_1,26 = 0.007, **P < 0.0098 (control IL3-scr versus food IL3-scr); #P = 0.0366 (control IL3 versus food IL3); n.s., P > 0.999 (food IL3 versus food IL3-scr); n = 7 to 8 mice per group. Error bars denote SEMs.

Fig. 7.. D2R-NMDAR heteromerization is increased in postmortem brain samples from individuals with psychostimulant use disorder despite decreased D2R expression.

(A) Representative images of D2R single recognition by PLA in human control subjects (Human: Control) or individuals having suffered from psychostimulant use disorder (Human: Psychostimulants) and negative control, in which the primary antibody is omitted (left; Neg. cont; see fig. S6). (B) Quantifications of D2R single PLA signal represented as fold decrease compared to control subjects. Two-sided Student’s t test: t = 3.331 df = 24, **P = 0.0028, n = 13 subjects per group. (C) Example images of GluN2B subunit of NMDAR single detection and Neg. cont. (D) Quantifications of Glu2NB PLA signal. Two-sided Student’s t test, t = 0.224 df = 24, P = 0.8243, n = 13 subjects per group. (E) Illustrative images of D2R-NMDAR heteromer detection by PLA and Neg. cont (GluN2B antibody omitted). (F) Quantifications of D2R-NMDAR PLA signal. Two-sided Student’s t test: t = 1.868 df = 24, P = 0.074, n = 13 subjects per group. (G) Pearson correlation between D2R expression levels and D2R-NMDAR heteromerization for each sample from all individuals having suffered from psychostimulant use disorder. R² = 0.3219, P = 0.0432. (H) Same as in (G) for the Pearson correlation between GluN2B expression levels and D2R-NMDAR. R² = 0.0087, P = 0.761. (I) Quantifications of D2R-NMDAR PLA signal normalized to D2R expression levels for each subject. Two-sided Student’s t test, t = 2.882 df = 24, **P = 0.0082, n = 13 subjects per group. Scale bars, 25 μm (A, C, and E). Error bars denote SEMs.