Shared and divergent transcriptomic regulation in nucleus accumbens D1 and D2 medium spiny neurons by cocaine and morphine

Abstract

Substance use disorders (SUDs) induce widespread molecular dysregulation in the nucleus accumbens (NAc), a brain region pivotal for coordinating motivation and reward. These molecular changes are thought to support lasting neural and behavioral disturbances that promote drug-seeking in addiction. However, different drug classes exert unique influences on neural circuits, cell types, physiology, and gene expression despite the overlapping symptomatology of SUDs. To better understand common and divergent molecular mechanisms governing SUD pathology, our goal was to survey cell-type-specific restructuring of the NAc transcriptional landscape in after psychostimulant or opioid exposure. We combined fluorescence-activated nuclei sorting and RNA sequencing to profile NAc D1 and D2 medium spiny neurons (MSNs) across cocaine and morphine exposure paradigms, including initial exposure, prolonged withdrawal after repeated exposure, and re-exposure post-withdrawal. Our analyses reveal that D1 MSNs display many convergent transcriptional responses across drug classes during exposure, whereas D2 MSNs manifest mostly divergent responses between cocaine and morphine, with morphine causing more adaptations in this cell type. Utilizing multiscale embedded gene co-expression network analysis (MEGENA), we discerned transcriptional regulatory networks subserving biological functions shared between cocaine and morphine. We observed largely integrative engagement of overlapping gene networks across drug classes in D1 MSNs, but opposite regulation of key D2 networks, highlighting potential therapeutic gene network targets within MSNs. These studies establish a landmark, cell-type-specific atlas of transcriptional regulation induced by cocaine and by morphine that can serve as a foundation for future studies towards mechanistic understanding of SUDs. Our findings, and future work leveraging this dataset, will pave the way for the development of targeted therapeutic interventions, addressing the urgent need for more effective treatments for cocaine use disorder and enhancing the existing strategies for opioid use disorder.

Fig 1.. Patterns of transcriptional regulation in NAc D1 and D2 MSNs induced by cocaine or morphine.

A, Experimental paradigm outlining cell-type-specific RNAseq after cocaine or morphine exposure conditions. Nuclei in D1 or D2 MSNs were GFP-tagged for purification by crossing mice with a floxed EGFP-L10a allele, which inserts a GFP tag into the ribosomal L10a locus, with D1-Cre or D2-Cre mice. Double-transgenic mice were treated with saline, cocaine (20 mg/kg), or morphine (10 mg/kg) for 10 days (cocaine cohort) or 15 days (morphine cohort), followed by a 30-day homecage withdrawal period. Subsequently, mice were challenged with an injection of either saline, cocaine, or morphine and were sacrificed 1h later from the homecage. NAc tissue was extracted, GFP-positive nuclei were purified via FANS, and RNAseq was performed on these D1 and D2 MSN nuclei. Note, cocaine and morphine cohorts were sequenced at different times. B, Rank-rank hypergeometric overlap plots illustrating threshold-free transcriptomic similarities between cocaine and morphine across first-ever exposure (SC vs. SM), withdrawal (CS vs. MS), or re-exposure after withdrawal (CC vs MM) separately for D1 or D2 MSNs. Heat indicates strength of overlap, and quadrants represent direction of gene expression (white arrows; lower-left quadrant, genes up in both; upper-right quadrant, genes down in both; upper left quadrant, genes down in cocaine but up in morphine; lower right quadrant: genes up in cocaine but down in morphine) C, Union heatmaps seeded to the CC condition showing log2FoldChange of genes identified as significantly (>20% fold change, p<0.05) upregulated (yellow) or downregulated (blue) in all experimental conditions.

Fig 2.. Gene network architecture of cocaine- and morphine-induced transcriptional regulation.

MEGENA network for D1 (top) and D2 (bottom) MSNs represented as sunburst plots with segments on each rung representing distinct networks. MEGENA enables progressive network refinement, as reflected in sunburst plots by central rungs being broken into smaller segments moving outwards from central rungs. Note that D1 and D2 MEGENA networks are represented by a single sunburst plot, which is presented repeatedly to highlight condition-specific network enrichment of DEGs. Coloration of sunburst plot segments depicts significant enrichment of upregulated (yellow) or downregulated (blue) genes in particular networks. Modules highlighted with red boxes indicate those emphasized in subsequent figures.

Fig 3.. D1 MSN gene network that integrates canonical IEGs is primed for activation by chronic cocaine.

A, Network structure of module D1 m160 showing defined hubs as diamonds and non-hub nodes as circles, both sized by degree of connections. Immediate early genes are highlighted with bold text. Differentially expressed genes (>20% FC, p<0.05) from the CC condition are overlaid onto this network (upregulated, yellow shading; no change, gray shading), demonstrating near complete activation of this network. B, IEG subnetworks were defined by a nearest neighbor approach wherein all nodes 2 steps downstream of IEGs were integrated into a subnetwork module. C, IEG subnetworks, which included two influential m160 hub genes, Aplp2 and Scn2b, with overlaid differential gene expression for each experimental condition to show gene network engagement.

Fig 4.. Hub gene analysis reveals key D2 networks that differentiate cocaine- vs. morphine-induced changes to biological functions.

A, MEGENA network structure for D1 and D2 MSNs. Note, D1 MSNs are more densely interconnected compared to D2 MSNs (mean difference of correlation coefficients p<2.2e-16, see Results). B, Hub gene analysis and clustering. Top, scheme describing the approach of extracting top-ranked gene network hubs for D1 or D2 MSNs based on exhibiting more than 20 co-expression partners. Heatmaps show expression patterns in log2FoldChange (up, yellow; down, blue) for each top-ranked hub gene across experimental conditions after hierarchical clustering. Bottom chart shows results of gene ontology analysis of biological processes enriched in D2 cluster 2 which exhibits opposite regulation between cocaine and morphine exposure conditions. C, Gene ontology analysis of two D2 MEGENA modules with most represented in D2 cluster 2, showing enrichment of biological processes associated with energetic utilization (m11) and protein trafficking (m19). D, Gene network structure of m11 and m19 presenting hub genes as diamonds and non-hub nodes as circles. Differentially expressed genes (p<0.05) from CC (left) and MM (right) conditions are overlaid showing upregulated (yellow) or downregulated (blue) genes, throughout the network.