Trans-synaptic modulation of cholinergic circuits tunes opioid reinforcement

Abstract

Opioids trigger structural and functional neural adaptations of the reward circuit that lead to dependence. Synaptic cell adhesion molecules (CAMs) play a pivotal role in circuit organization and present prime candidates for orchestrating remodeling of neural connections in response to drug exposure. However, the contribution of CAMs to opioid-induced rewiring of the reward circuit has not been explored. Here, we used unbiased molecular profiling to identify CAMs in the nucleus accumbens (NAc) modulated by morphine administration. We found that opioid exposure induces the expression of ELFN1, a CAM selectively expressed in cholinergic interneurons in the NAc. We determined that ELFN1 acts trans-synaptically to modulate the strength and plasticity of the glutamatergic inputs onto cholinergic neurons via the recruitment of presynaptic metabotropic glutamate receptor 4 (mGlu4). Disruption of Elfn1 diminished morphine reward and intake in self-administering mice. Together, our findings identify a key molecular factor responsible for adjusting the strength of opioid effects by modulating the configuration of striatal circuitry in an experience-dependent fashion and unveil potential therapeutic target for combating opioid abuse.

Keywords: GPCR; cell adhesion molecules; mGlu receptors; opioids.

Fig. 1.

Repeated exposure to opioids drives remodeling of CAMs. (A) Schematic of drug administration in mice and experimental workflow for tissue preparation for RNA sequencing. Saline or morphine was subcutaneously (s.c.) delivered with a single daily injection for five consecutive days. Microdissected tissue from NAc was collected 24 h after the last injection and processed for RNA sequencing. (B) Volcano plot of differentially expressed genes in the NAc following repeated morphine exposure is shown in blue (up-regulated) and red (down-regulated) based on log2 ratio ≥ 0.05 and P-value ≤ 0.05. CAM ELFN1 and other known genes induced by morphine are highlighted. (C) Venn diagram showing categorical classification of protein localization. (D) In situ hybridization showing distribution of ELFN1 mRNA (Elfn1, green), MOR mRNA (Oprm1, red), and DAPI (blue) in the NAc. (Scale bar, 10 µm.) (E) Representative immunoblot for ELFN1, synaptophysin (SYP), postsynaptic density protein 95 (PSD-95), and GAPDH for samples isolated via microdissection of 2 mm × 2 mm brain punches from wild-type (WT) mice after repeated exposure to saline or morphine as shown in panel A. On the right densitometric quantification of protein expression levels as a percentage of saline-treated mice. Elfn1 P = 0.0168 unpaired t test; SYP P = 0.3186 Mann–Whitney test; PSD-95 P = 0.1434 unpaired t test (n = 12 to 13 mice). (F) Representative immunoblot for ELFN1 and GAPDH for samples isolated via microdissection of 2 mm × 2 mm brain punches from WT mice after repeated exposure to saline or naloxone and morphine as shown in panel A. Densitometric quantification of protein expression levels as a percentage of saline-treated mice. Elfn1 P = 0.1343 unpaired t test (n = 12 to 14 mice).

Fig. 2.

E1fn1 expression in CINs regulates glutamatergic signaling. (A) Schematic representation of Elfn1 KO mouse generation. The coding exon is replaced with LacZ-neo cassette for β-Galactosidase reporter expression. (B) Representative image showing an acute slice containing a SPiDER-β-Gal-labeled neuron in the NAc. (Scale bar, 20 µm.) (C) Electrophysiological recordings obtained from the same neuron showing representative traces of spontaneous firing activity (Top) and firing response to intrasomatic current injections (Bottom). (Scale bar, 10 mV, 1 s.) (D) Double staining of striatal sections containing the NAc obtained from Elfn1 KO mice using anti-β-galactosidase antibody (green) and anti-ChAT (magenta). (Scale bar, 100 µm; ac: anterior commissure.) (E) Representative traces of EPSCs upon paired-pulse stimulation (40 ms) of local glutamatergic afferents. (Scale bar, 40 pA, 20 ms.) (F) Pooled data show significant increases in the amplitude of the 2nd EPSC in both genotypes. Two-way ANOVA showing significant effect of EPSC number across groups F (1, 18) = 22.67; P = 0.0002. (G) The amplitude ratio of EPSC2/EPSC1 is significantly smaller in Elfn1 KO mice (n = 10 neurons/6 mice) compared to WT mice (n = 10 neurons/6 mice). Nonparametric t test; Mann–Whitney test, P = 0.0089. (H) Representative traces of short-term facilitation of excitatory synaptic transmission in CINs induced by five trains of 10 pulses at 20 Hz. (Scale bar, 40 pA, 100 ms.) (I) Normalized time course of EPSC amplitude showing EPSC facilitation in WT mice and EPSC depression in Elfn1 KO mice over the course of the train. (J) Quantification of the EPSC amplitude change of last (EPSC10) vs. first (EPSC1). Nonparametric t test; Mann–Whitney test, P = 0.0030. (K) The normalized ratio of last (EPSC10) and first (EPSC1) is significantly decreased in Elfn1 KO mice (n = 8 neurons/5 mice) compared to WT mice (n = 7 neurons/4 mice). Nonparametric t test; Mann–Whitney test, P = 0.0030.

Fig. 3.

Circuit mechanism for the control of converging glutamatergic and opioidergic signals in CINs. (A) Schematic for conditional elimination of Elfn1 from CINs. Mice containing a conditional KO allele (Elfn1^flx/flx) were crossed with a ChAT^Cre strain to generate Elfn1 CIN cKO. (B) Striatal sections were immunostained with anti-ChAT, and endogenous GFP fluorescence was specifically observed in ChAT⁺ CIN in the NAc only in Elfn1-cKO-CIN mice but not their Cre⁻ control littermates. (Scale bar, 10 µm.) (C) Representative traces of spontaneous firing under physiological conditions from WT and Elfn1 cKO mice. (Scale bar, 1 mV, 2 s.) (D) Time course of spontaneous firing in WT and Elfn1 cKO mice. (E) Average firing frequency from WT (n = 8 neurons/5 mice) and Elfn1 cKO mice (n = 8 neurons/8 mice) showing no difference under basal conditions. (F) Interspike interval (ISI) showing similar duration in WT and Elfn1 cKO mice. (G) Representative traces of EPSCs upon paired-pulse stimulation (40 ms) of local glutamatergic afferents. (Scale bar, 50 pA, 20 ms.) (H) Amplitude ratio of EPSC2/EPSC1 is significantly smaller in Elfn1 cKO mice (n = 6 neurons/4 mice) compared to WT mice (n = 6 neurons/5 mice). Nonparametric t test; Mann–Whitney test, P = 0.0022. (I) Pooled data show no difference in the amplitude of first EPSC between genotypes. (J) Representative traces of short-term facilitation of excitatory synaptic transmission in CINs induced by five trains of 10 pulses at 20 Hz. (Scale bar, 40 pA, 40 ms.) (K) Normalized time course of EPSC amplitude showing EPSCs facilitation in WT mice and EPSCs depression in Elfn1 cKO mice over the course of the train. (L) The normalized ratio of last (EPSC10) and first (EPSC1) is significantly decreased in Elfn1 cKO mice (n = 5 neurons/5 mice) compared to WT mice (n = 5 neurons/5 mice). Nonparametric t test; Mann–Whitney test, P = 0.0079.

Fig. 4.

Elfn1 recruits and stabilizes mGlu₄ at presynaptic excitatory terminals in the NAc. (A) Western blotting and (B) densitometric quantification of protein expression for various synaptic components compared between WT and Elfn1 cKO mice in the NAc. mGlu₄ P = 0.022; unpaired t tests, n = 14. (C) Electron micrographs of NAc region showing immunoparticles for mGlu₄, detected via a pre-embedding immunogold method. Dendritic spines (s) and axon terminals (at) are marked; arrows indicate locations of mGlu₄ immunoparticles at the synaptic terminal. (D) Quantitative analysis showed that mGlu₄ is significantly reduced at synaptic terminals of Elfn1 cKO mice, compared to WT mice (n = 3 mice/genotype). (Scale bar, 200 nm.) Unpaired t tests, Mann–Whitney test, P < 0.0001. (E) Schematic of electrophysiological experiments assessing the contribution of mGlu₄ at excitatory synapses impinging on CINs. Activation of mGlu₄ was achieved using the selective group III mGluRs agonist L-AP4 (50 µm). (F) Representative EPSC traces obtained upon different pharmacological treatments. (Scale bar, 100 pA, 40 ms.) (G) Normalized time course showing changes in EPSCs amplitude in response to pharmacological treatments with L-AP4 (50 µM, black) or L-AP4 and CPPG (100 µM, gray) in WT mice. (H) Normalized time course showing changes in EPSCs amplitude in response to pharmacological treatments with L-AP4 in Elfn1 KO mice (red) and Elfn1 cKO mice (green). (I) Quantification of EPSCs amplitude inhibition produced by L-AP4 in CIN from Elfn1 KO mice (n = 7 neurons/5 mice) and Elfn1 cKO mice (n = 6 neurons/5 mice) compared to WT mice (n = 6 neurons/4 mice). One-way ANOVA, Kruskal–Wallis test; P = 0.0018, followed by Dunn’s multiple comparison test: WT vs. Elfn1 KO mice P = 0.0242; WT vs. Elfn1 cKO mice P = 0.0049.

Fig. 5.

Elfn1 regulates morphine reward in the NAc. (A) Schematic of the experimental design for self-administration. (B) Schematic for the self-administration apparatus. An active lever (A) is associated with intravenous infusion of morphine; I indicates the inactive lever. Mice begin self-administration of morphine 1 wk after catheter implantation and then are tested with varying concentrations of morphine under a fixed ratio five schedule. (C) Raster plot from representative WT and Elfn1 cKO mice, showing infusion distribution over time. The raster plot shows last 3 d and first 3 d of descending limb of morphine doses (0.3 mg/kg → 0.1 mg/kg) and last 3 d and first 3 d of ascending limb of morphine doses (0.1 mg/kg → 0.3 mg/kg). (D) Cumulative intake of morphine self-administration (0.1 to 0.3 mg/kg) in WT mice (n = 9 mice) and Elfn1 cKO mice (n = 10 mice). Two-way ANOVA showing significant interaction between doses and genotypes: F (1,17) = 12.23, P = 0.0028, followed by Šídák's multiple comparisons test: WT (01 mg/kg vs. 0.3 mg/kg) P < 0.0001; WT vs. Elfn1 cKO (0.3 mg/kg) P < 0.0001. (E) Number of infusions earned during first hour and second hour of morphine self-administration between WT and Elfn1 cKO mice. Two-way ANOVA showing that both genotypes earned significantly more rewards during the first hour of the session F (1,17) = 18.11, P = 0.0005. (F) Interinfusion intervals during session 25, between WT and Elfn1 cKO mice. Nonparametric t test; Mann–Whitney test, P = 0.0006. (G) Interinfusion intervals distribution during session 25 and relative pie chart distribution in WT. (H) Interinfusion intervals distribution during session 25 and relative pie chart distribution Elfn1 cKO mice. Intervals are expressed in seconds. (I) PR paradigm, showing exponential response requirement relative to morphine infusion number. (J) Morphine infusions under PR schedule in WT and Elfn1 cKO mice. Nonparametric t test; Mann–Whitney test, P < 0.0001. (K) Maximal number of lever presses for morphine infusion in WT and Elfn1 cKO mice. Nonparametric t test; Mann–Whitney test, P < 0.0001. (L) Time distribution for breakpoint values in WT and Elfn1 cKO mice. Simple linear regression, R-squared = 0.5848 (WT) 0.4814 (Elfn1 cKO). F = 6.925, P = 0.0094.

Fig. 6.

Opioids drive rewiring of the cholinergic microcircuit in the NAc to adjust drug seeking. Activation of MOR by morphine acts as a signaling modifier for reward encoding through modulation of CIN activity in the NAc. Prolonged MOR activation upon morphine exposure prompts CIN reorganization via the expression of a cell-adhesion molecule—Elfn1—which trans-synaptically engages mGlu₄ located at glutamatergic terminals projecting onto CINs in the NAc. The increased mGlu₄ activity allows temporal control of sustained excitation of CINs by glutamatergic inputs which integrate with the inhibitory actions of MOR in CIN providing neuromodulatory tone for the activity of dMSN and iMSN to promote encoding of accurate opioid reward responses in the microcircuit of the NAc.