Morphine induces gene expression changes in specific VTA cell populations.

a, Illustration of behavioral and snRNA-seq workflow. b, Treatment groups and sample sizes for snRNA-seq. c, Behavioral effects of CFA-induced inflammatory pain measured with the Von Frey test. CFA induced mechanical hyperalgesia on day 2 and day 7 after administration (two way ANOVA, significant interaction between test day and CFA treatment, F(6, 108)=14.61; p<0.0001, with Tukey post-hoc tests revealing group differences on D2 and D7). d, Integrated VTA snRNA-seq UMAP from 28 independent samples, totalling 38,468 nuclei after quality control and filtering. Clustering identified 16 distinct cell types. e, Total number of differentially expressed genes (DEGs) identified using cluster-specific pseudobulked analysis with DESeq2.

Morphine induces immediate early genes in dopamine neurons.

a, UMAP representation of neuronal cluster IDs. b, Marker genes for neurons (Syt1) dopamine neurons (Th, Slc6a3), glutamate neurons (Slc17a6), and GABA neurons (Slc32a1, Gad1, Gad2), as well as the µ opioid receptor gene Oprm1, plotted with UMAP coordinates (from a). c, Violin plot of expression for Oprm1 across neuronal cell types reveals highest expression in GABA.2 neurons. d, Expression of immediate early genes in DA.1 neuronal cluster, split by saline or morphine treatment. e, Expression of IEG module genes across VTA cluster IDs reveals selective upregulation in dopamine neurons. f, Volcano plot of DA.1 morphine DEGs identified from pseudobulk analysis with DESeq2. g, Heatmap of DA.1 morphine DEGs across all experimental groups reveals no systematic interaction between pain state and morphine transcriptional response. h, Illustration of circuit model for dopamine neuron activation by morphine.

Despite low or absent expression of opioid receptors, morphine strongly alters gene expression in VTA non-neuronal cell populations.

a, Expression heatmap for opioid receptor genes across VTA cluster IDs. b, Feature plot of Fkbp5, split by experimental group (saline and morphine). c, Volcano plots for differentially expressed genes in astrocytes, oligodendrocytes, and microglia. d, Pie charts showing the intersection of upregulated and downregulated DEG lists across three glial populations. e, Expression heatmap for steroid hormone receptor genes Nr3c1 and Nr3c2 across VTA cluster IDs. f, Gene track for shared DEG Fkbp5 identifies glucocorticoid response element (GRE) in both the rat (top) and human (bottom) genome.

Corticosterone (but not µOR activation) induces Fkbp5 in rat glial cells.

a, Treatment with corticosterone, an endogenous glucocorticoid in rats, increases Fkbp5 expression as measured by RT-qPCR in rat C6 glioma cells (One-way ANOVA, n = 8 per group). b, Pretreatment with the glucocorticoid receptor antagonist mifepristone 0.5 h prior to application of corticosterone attenuated the induction of Fkbp5 (One-way ANOVA, n = 5-6 per group). c, Treatment with the µOR agonist DAMGO for 5 h failed to induce Fkbp5 (One-way ANOVA, n = 6 per group). Data expressed as mean ± s.e.m. Multiple comparisons, ****p < 0.0001.

Cortisol (but not µOR activation) induces FKBP5 in a human-derived astrocyte model via glucocorticoid receptor signaling.

a, Illustration for generation of human-derived astrocytes from immortalized neural precursor cells (NPCs). b, Immunocytochemistry images showing acquisition of astrocyte markers GFAP and S100B across the differentiation time course, as well as depletion of NPC markers. c, Treatment with cortisol increases FKBP5 expression as measured by RT-qPCR in human-derived astrocytes in a dose-dependent manner (One-way ANOVA, n = 12 per group). d, Treatment with the glucocorticoid receptor agonist dexamethasone induces FKBP5 expression as measured by RT-qPCR in human-derived astrocytes (Unpaired t-test, n = 13 per group). e, Treatment with the µOR agonist DAMGO for 5 h failed to induce FKBP5 (One-way ANOVA, n = 15 per group). f, Pretreatment with the glucocorticoid receptor antagonist mifepristone 0.5 h prior to application of cortisol blocked the induction of FKBP5 (One-way ANOVA, n = 12 per group). h, Mifepristone blocked dexamethasone-induced increases in FKBP5 in human astrocytes. Data expressed as mean ± s.e.m. Multiple comparisons, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Cortisol induction of FKBP5 in a human-derived astrocyte model requires NR3C1.

a, Illustration for CRISPRi targeting the NR3C1 promoter. Lentiviral vectors express a dCas9-KRAB-MeCP2 fusion from the astrocyte-selective GfaABC1D promoter, and CRISPR sgRNAs for lacZ (a non-targeting control) and NR3C1 are delivered via separate viral vectors. b, Immunocytochemistry images showing expression of CRISPRi machinery in human-derived astrocytes. c, NR3C1 sgRNA paired with CRISPRi machinery successfuly reduces NR3C1 mRNA as compared to lacZ control sgRNA (Mann-Whitney nonparametric test,, n = 6 per group). d, NR3C1 CRISPRi blocked the induction of FKBP5 by cortisol in human-derived astrocytes (Two-way ANOVA, n = 6 per group). Data expressed as mean ± s.e.m. Multiple comparisons, *p < 0.05, **p < 0.01.

Quality control, integration, and marker gene analysis for snRNA-seq dataset.

a, Violin plot showing number of genes detected in each cluster ID. b, Violin plot of the number of unique molecular identifiers (UMIs) in each cluster ID. c, Bar graph of the proportion of nuclei contributed from each experimental group. d, Local Simpson’s Inverse Index (LISI) scores for each annotated cluster. e, Dot plot of expression level and percentage of known marker genes in each cluster.

Comparison of annotated cluster IDs with the Allen Brain Cell (ABC) mouse brain atlas.

a, Alluvial plot showing relationship between cluster ID assignments in this project and ABC atlas class ID. b, Integrated VTA UMAP showing predicted Allen class ID for each nucleus. c, Bootstrapping probability for predictions for each ABC class ID indicates lower confidence in predictions for glutamatergic and GABAergic neurons.

Dopamine neuron subtype marker comparison.

a, Feature plot of dopamine neuron populations with cluster IDs. b, expression of dopamine marker genes (top row) and glutamate and GABA marker genes (bottom row). c, Pie graphs showing percent of DA.1 or DA.2 neurons expressing DA and other marker genes. DA.1 neurons express high levels of Gch1 and little Slc26a7, whereas DA.2 neurons express high levels of Slc26a7 and little Gch1.

Overlap of GABAergic interneuron markers with Oprm1 expression in the VTA GABA.2 cluster.

a, UMAP of VTA neuronal populations. b, Expression feature plots of Oprm1 and known interneuron markers within the GABA.2 cluster. c, Percentage of GABA.2 neuronal population expressing common interneuron markers. d, Percentage of neurons with interneuron marker expression that also express Oprm1.

Heatmaps of DEGs in Astrocyte, Olig.1, and Microglia clusters with individual sample level data shown and split by sex.

Each row represents a single DEG, and columns represent individual samples. For each gene, pseudobulked log-normalized counts were centered and scaled across samples to calculate the Z-score using the formula Z = (x - μ) / σ, where μ is mean and σ is standard deviation.

Pain and morphine interact to alter gene expression in VTA astrocytes.

a, Heatmap of Astrocyte cluster CFA x morphine interaction DEGs. Each column represents a single DEG, and rows represent group average values. For each gene, pseudobulked log-normalized counts were centered and scaled across groups to calculate the Z-score using the formula Z = (x - μ) / σ, where μ is mean and σ is standard deviation. b, Mean and representative interaction DEGs belonging to Group A and Group B. c, Network diagram showing enriched gene ontology terms in Group B interaction DEGs.