bioRxiv preprint doi: https://doi.org/10.1101/781872; this version posted September 25, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. PREPRINT A dopamine-induced gene expression signature regulates neuronal function and cocaine response Katherine E. Savell1†‡, Morgan E. Zipperly1†§, Jennifer J. Tuscher1†§, Corey G. Duke1†§, Robert A. Phillips III1†§, Allison J. Bauman1, Saakshi Thukral1, Faraz A. Sultan1, Nicholas A. Goska1, Lara Ianov1, & Jeremy J. Day1,2* Drug addiction is a worldwide health problem, with overdose rates of both psychostimulants and opioids currently on the rise in many developed countries. Drugs of abuse elevate dopamine levels in the nucleus accumbens (NAc) and alter transcriptional programs believed to promote long-lasting synaptic and behavioral adaptations. However, even with well-studied drugs such as cocaine, drug- induced transcriptional responses remain poorly understood due to the cellular heterogeneity of the NAc and complex drug actions via multiple neurotransmitter systems. Here, we leveraged high- throughput single-nucleus RNA-sequencing to create a comprehensive molecular atlas of cell subtypes in the NAc, defining both sex-specific and cell type-specific responses to acute cocaine experience in a rat model system. Using this transcriptional map, we identified specific neuronal subpopulations that are activated by cocaine, and defined an immediate early gene expression program that is upregulated following cocaine experience in vivo and dopamine (DA) receptor activation in vitro. To characterize the neuronal response to this DA-mediated gene expression signature, we engineered a large-scale CRISPR/dCas9 activation strategy to recreate this program. Multiplexed induction of this gene program initiated a secondary synapse-centric transcriptional profile, altered striatal physiology in vitro, and enhanced cocaine sensitization in vivo. Taken together, these results define the genome-wide transcriptional response to cocaine with cellular precision, and demonstrate that drug-responsive gene programs are sufficient to initiate both physiological and behavioral adaptations to drugs of abuse. Correspondence to Jeremy Day ([email protected] | day-lab.org | @DayLabUAB) NEARLY 5 MILLION Americans reported cocaine use in contributions of DA-dependent transcriptional programs 2017, and recent increases in cocaine-related drug overdoses to neuronal physiology and behavior is not clear. Further, present significant public health challenges 1( ). A hallmark although drug experience leads to large scale transcriptional trait of drugs of abuse is the acute elevation of dopamine changes in the NAc, previous studies have focused on (DA) in the nucleus accumbens (NAc), a central integrator of individual drug-responsive genes instead of coordinated the reward circuit (2-4). Abused drugs produce increases in gene programs. Here, we sought to define DA-driven DA that are greater in both concentration and duration than gene expression signatures with single-cell precision and natural rewards (2, 5, 6), and this signaling is hypothesized understand the molecular and physiological consequences to underlie maladaptive reinforcement after repeated drug of this gene program. use (7). Exposure to drugs of abuse results in significant To map the transcriptional landscape and cocaine transcriptional and epigenetic reorganization in the NAc (8- response of the NAc with single-cell resolution, we 13), initiating synaptic and behavioral plasticity associated performed single-nucleus RNA-sequencing (snRNA-seq) on with the transition to drug addiction (7, 14, 15). However, 15,631 NAc cells from both male and female rats after acute even with well-studied drugs such as cocaine, drug-induced exposure to cocaine (Fig. 1; Supplementary Fig. 1), using transcriptional responses remain poorly understood. This a cocaine dose previously reported to result in enduring is in part due to the cellular heterogeneity of the NAc, synaptic changes in the NAc (15). From a merged dataset which is a diverse structure containing multiple neuronal containing transcriptomes from all cells, unsupervised and non-neuronal subpopulations and complex neuronal dimensionality reduction approaches identified discrete NAc circuitry. Additionally, many drugs of abuse engage multiple neuronal clusters harboring known markers of Drd1 DA neurotransmitter systems in the NAc (16-22), and the specific receptor positive and Drd2 DA receptor positive medium

1Department of Neurobiology, University of Alabama at Birmingham, Birmingham, AL 35294, USA 2Civitan International Research Center, University of Alabama at Birmingham, Birmingham, AL 35294, USA †Equal contributions ‡Present address: Behavioral Neuroscience Research Branch, Intramural Research Program, National Institute on Drug Abuse, NIH/ DHHS, Baltimore, MD 21224, USA §Position of these authors was determined by random assignment bioRxiv preprint doi: https://doi.org/10.1101/781872; this version posted September 25, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

a Single nucleus RNA-seq workflow

10X Genomics Saline FACS Cell 1 Cell 2 Cell 3

Cell 4 Cell 5 Cell 6 Cocaine Waste & Single nucleus debris suspension Nuclei Oil Library construction Locomotor Dissect NAc Harvest Dissociate Pellet Purify Capture & RNA sequencing testing tissue cells & filter nuclei nuclei barcoding Demultiplexing & analysis

b c Global single-nucleus RNA clustering (15,631 nuclei) Polydend. Microglia Drd3-MSN Pvalb Int. Sst Int. Drd2-MSN-2 Mural Olig-3 Glutamatergic Olig-1 Drd1-MSN Drd2-MSN-1 Olig-2 AstrocyteGrm8-MSNGABA Polydendrocyte Mbp Ebf1 Pvalb interneuron Olig-3 Drd2-MSN-2 Drd2

GABA Drd2-MSN-1 Sst-Interneuron undefined Slc1a3

Olig-1 Glutamatergic Grm8

Drd3-MSN Pdgfra Grm8-MSN

Olig-2 Kit Sst Mural Drd1-MSN

Microglia Astrocyte

UMAP_2 Slc17a7 UMAP_1 d Cocaine DEGs e Enhanced experimental signal f Drd2-MSN g Drd1-MSN subclustering top 10 UMAP_2 EES Grm8 1 Poly EES Nr4a3 1 Drd3 Micro 4 score Fosb 0 UMAP_1 Olig-2 1.0 Nr4a1 Fosl2 0.5 AC134224.3 3 2 Drd2-1 0.0 Drd2-MSN-1 Homer1 Egr4 Homer1 Fosb 5 1 −0.5 Penk Junb 0 0 Arc Drd1

GABA Drd1-MSN top 10 Homer1 Marker genes Nr4a3 h 1 Nr4a1 Olig-1 Homer1 Neuron-Neuron Fosl2 Glia-Glia Junb Neuron-Glia Drd1-MSN Tac1 Nr4a3 Arc Fosb Sik2

Sik2 (subcluster 1 Ntrk2 UMAP_2 cocaine response)

Ntrk2 Responder genes

10 genes ROC predictive power Dlg1 Astro UMAP_1 0 Mbnl2 0.4 0.8 0 1 Pearson’s r ROC predictive power (subcluster 1 grouping) Figure 1. Single-nucleus RNA-seq reveals cell-specific transcriptional response to cocaine. a, Single-nucleus RNA-seq workflow. Male and female adult rats (n = 4/sex) received saline or 20mg/kg cocaine (intraperitoneal injection) and underwent locomotor testing prior to tissue harvesting, nucleus accumbens dissection, nuclei purification, and single nuclei sequencing on the 10X Genomics platform. b, Global clustering across experimental treatment and sex for 15,631 individual NAc nuclei identifies all major cell classes of the rat NAc, including MSNs expressing Drd1 and Drd2 mRNA. c, Heatmap of cell-specific marker genes across all clusters (see Supplementary Figure 2 for additional details). d, Circos plot of cocaine DEGs by cluster (outer rim, ordered clockwise by adjusted p value within each cluster) and coherent changes between clusters (internal arcs). Neuron-neuron, glia-glia, and neuron-glia coherent DEGs are reflected in teal, purple, and gray, respectively. e, Unbiased detection of cocaine-activated cell clusters with MELD pipeline identifies enhanced experimental signal (EES) in distinct Drd1-MSN and Drd2-MSN neuronal clusters. f, Top 10 gene-EES correlations for Drd1-MSN and Drd2-MSN clusters. g, Subclustering of Drd1-MSNs highlights specific IEG induction and EES signal in a single subpopulation (subcluster 1). UMAPs show only Drd1-MSNs identified in panel B. h, Receiver operating characteristic (ROC) analysis on subcluster 1 identifies “marker” genes that predict membership in that cluster as compared to all other Drd1-MSN subclusters (x-axis), as well as “responder” genes that predict cocaine treatment classification (i.e., are altered by cocaine). spiny neurons (Drd1-MSNs and Drd2-MSNs, respectively), oligodendrocytes (Fig. 1b-c and Supplementary Fig. 2). as well as previously identified (23, 24) transcriptionally- Notably, this clustering also revealed potentially novel MSN defined cell classes including somatostatin (Sst) positive classes marked by expression of Drd3 (a D2 family DA interneurons, microglia, astrocytes, polydendrocytes, and receptor) and Grm8 (a metabotropic glutamate receptor). 2 bioRxiv preprint doi: https://doi.org/10.1101/781872; this version posted September 25, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. While Drd3-MSN and Grm8-MSN clusters were relatively the presence of distinct, non-overlapping, and sex-specific depleted in the classic MSN marker Ppp1r1b (the gene astrocytic transcriptional responses to cocaine. encoding DARPP-32 protein) (25), each cluster exhibited To further investigate novel upstream interactions high expression of Foxp2 and Bcl11b, genes strongly linked that may contribute to cocaine response likelihood within to MSN differentiation and function (Supplementary Fig. a cell type, we restricted our analysis to Drd1-MSNs and 2) (26-28). Moreover, whereas cell fractions in nearly all repeated dimensionality reduction mapping approaches (Fig. clusters were evenly distributed between treatment group 1g). This analysis identified 4 unique subclusters of Drd1- (saline and cocaine) and sex, 91% of Drd3-MSNs were found MSNs, only one of which (subcluster 1) exhibited robust in male animals, indicative of a significant sex bias in this cocaine responsivity. Using this subcluster map, we employed subpopulation (Supplementary Fig. 3). receiver operator characteristic (ROC) analysis to estimate To identify cocaine-activated cell clusters, we the success of every individual gene in binary classification pursued two parallel strategies. First, we performed cluster- of a cell in 2 steps. First, we identified genes that marked specific identification of differentially expressed genes cells found in the cocaine-activated subcluster (subcluster (DEGs) in saline versus cocaine conditions (collapsing 1) as compared to cells in all other Drd1-MSN subclusters across sex; Fig. 1d; Supplementary Table 1). This analysis (termed “marker” genes). Next, we identified genes that revealed robust transcriptional response in Drd1-MSNs, distinguished cocaine-treated cells from saline-treated cells which contained more DEGs (232) than any other cluster. in this cocaine-activated subcluster (termed “responder” Drd1-MSN DEGs were enriched in CREB binding motifs genes). Intriguingly, while genes such as the synaptic protein and genes involved in MAP kinase pathways, regulation of Homer1 served as both marker and responder genes, other synaptic signaling, behavior, and cognition (Supplementary genes such as Dlg1 exhibited high predictive power in Fig. 4, Supplementary Table 2). Drd1-MSN DEGs also marking the cocaine-activated subcluster without exhibiting exhibited overlap with DEGs arising from Drd2-MSNs (Fig. an altered response to cocaine (Supplementary Table 5). 1d), suggesting the induction of common transcriptional In contrast, the Drd1 receptor itself exhibited almost no pathways in these clusters. We next used an unbiased predictive power as a marker gene (ROC power = 0.138). graphical signal processing approach to stratify cellular Together, these results highlight a potential contribution clusters based on condition-specific gene signatures 29( ), of multiple genes in determining which Drd1-MSNs are termed the “enhanced experimental signal”, or EES (Fig. capable of being activated by cocaine. 1e; Supplementary Table 3). This method identified two The activation of specific neuronal ensembles by unique cocaine-responsive subclusters - one from the Drd1- cocaine may result from circuit-based mechanisms (e.g., MSN parent cluster and another from Drd2-MSN-1 parent differential projections from input structures), or could cluster. Both high-EES cell clusters exhibited expression of potentially arise from cell-autonomous mechanisms in key immediate early genes (IEGs; e.g., Fosb, Junb, and Nr4a1) response to DA signaling (7, 17, 32-34). However, drugs of in the cocaine condition, but little to no expression of these abuse target many distinct receptor classes in a variety of genes in the saline condition (Supplementary Fig. 4). Given neuronal and non-neuronal cell types (20, 21), and these that these genes are commonly used markers for neuronal complex drug actions make identification of DA-induced gene activity (18, 30), these results suggest that cocaine activates expression programs difficult usingin vivo models. Therefore, select ensembles but not the majority of Drd1- or Drd2-MSNs. we took advantage of a well-studied and controllable rat However, this analysis also revealed correlations between primary striatal neuron culture system (35, 36) to enable EES and expression of several genes central to neuropeptide comprehensive and specific identification of DA-dependent signaling in MSNs (31), including the substance P precursor gene expression programs in MSNs. Using this system, we Tac1 (in Drd1-MSNs) and the enkephalin precursor Penk (in first identified a core signature of the transcriptional response Drd2-MSNs; Fig. 1f). to DA receptor activation by performing deep RNA-seq on To determine whether genetic sex contributed to bulk striatal neuronal cultures treated with 1µM DA for 1hr, cocaine responses in each cluster, we recalculated EES-gene a treatment that closely models concentrations and temporal correlations independently for each sex (Supplementary Fig. duration of DA increases found in vivo after acute cocaine 5, Supplementary Table 4). Male and female transcriptional exposure (2, 3, 6). This approach identified 103 DEGs (Fig. responses were positively correlated in both Drd1-MSNs 2a-b, Supplemental Table 6) following DA treatment, and Drd2-MSNs, and in both cases key IEGs associated with with the majority of DEGs (100) being upregulated versus high EES scores in our sex-combined analysis also drove EES vehicle control samples. Gene ontology analysis revealed that scores when sex was considered separately. In contrast, this genes upregulated by DA receptor stimulation are primarily analysis revealed surprising divergence in astrocytes. For both associated with transcriptional mechanisms and processes sexes, numerous astrocyte-responsive genes exhibited strong important for neuronal function such as regulation of synaptic correlations with EES. However, male and female responses plasticity (Fig. 2c; Supplementary Table 7). Transcription were negatively correlated for this cluster, demonstrating factor motifs significantly enriched in DA-responsive genes

3 bioRxiv preprint doi: https://doi.org/10.1101/781872; this version posted September 25, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

a DIV 11 Primary Rat b RNA-seq c Gene ontology/Network analysis Striatal Neurons Acute dopamine Acute dopamine: upregulated genes 103 DEGs (100▲, 3▼) 40 DNA-binding transcription activator activity skeletal muscle cell differentiation 30 Fosb Junb regulation of neuronal synaptic plasticity Biological Fos Nr4a2 Nr4a1 central nervous system neuron differentiation process negative regulation of canonical Wnt signaling pathway Vehicle Dopamine 20 (1µM) CREB ATF Enriched transcription 1hr 10 E4F1 ATF6 factor motifs SRF 0 RNA-seq Wald statistic (abs. val.) -20 -10 0 (n = 3 per group) 0 2 4 10 10 10 log2(Fold change) Adjusted Dopamine/Vehicle p-value

d Global single-nucleus f Dopamine SKF-38393 KCl RNA clustering (3,361 nuclei) Cluster ID Grm8-MSN Drd1-MSN Drd2-MSN Grm8-MSN GABA GABAergic Poly Drd2-MSN undefined Bottom Top Bottom Top Bottom Top DA DEG EES correlation (rank) Drd1-MSN

g Cocaine h UMAP_2 Cluster ID GapdhFos Fosb Junb Nr4a1 Nr4a2 UMAP_1 Polydendrocyte Drd1-MSN

Drd2-MSN-1 for all IEGs Vehicle p < 0.05 Drd2-MSN-2 Drd3-MSN DA Grm8-MSN DA + SCH23390 Fold change GABA (vs control) e Pvalb 3.0 DA SKF KCl DMSO Sst for all IEGs 2.5 100 p < 0.05 Poly. Drd1 Poly. 300 Drd1 Glut DA 2.0 Olig-1 30 100 1.5 Olig-2 DA + U0124 10 30 1.0 Olig-3 DA + U0126 Astrocyte 0.5

Poly for all IEGs

DMSO p < 0.05 Grm8 Drd2 Grm8 Drd2 Microglia Mural DA

Bottom Top DA + 666-15 GABA GABA DA DEG EES correlation (rank)

Figure 2. Identification of a cell-type specific dopamine-mediated transcriptional program. a, Illustration of the experimental setup for modeling dopamine receptor activation in medium spiny neuron cultures. b, Volcano plot showing signature of differentially expressed genes (DEGs) detected following 1 hr dopamine treatment. Dotted line on y-axis shows adjusted p value < 0.05. DESeq2 identified a total of 3 downregulated genes (light gray) and 100 upregulated genes (red), including immediate early genes (IEGs) Fos, Fosb, Junb, Nr4a1, and Nr4a2. c, Top 5 significant biological processes (gene ontology) and transcription factor binding site enrichment (motif analysis) for 100 upregulated genes in dopamine treated neurons. A maximum value was assigned to CREB (adjusted p value = 0). d, Single-nucleus RNA-seq from 3,361 nuclei identified 5 major cell classes in primary striatal neuron cultures. Neurons were treated with dopamine (50μM), the DRD1 agonist SKF-38393 (1μM), or depolarized with 25mM KCl. e, Differentially expressed genes by cell cluster. Dopamine and SKF-38393 induce a transcriptional response primarily in Drd1-MSNs, whereas KCl depolarization results in broad transcriptional alterations across neuronal classes. f, Unbiased signal identification (EES) identifies high correlation between DA-induced gene set (100 DEGs from bulk RNA-seq) and transcriptional response to dopamine or DRD1 agonist SKF specifically in Drd1 MSNs. Ridgeline plots show density of EES-gene correlation ranks for core DA signature genes (100 upregulated genes from 2b). This activation signature is present in Drd2 MSNs following depolarization, but not in response to DA. g, DA signature gene induction in multple MSN clusters following acute cocaine experience in vivo (clusters taken from Fig. 1b). h, RT-qPCR for representative DA transcriptional response genes following DRD1 receptor antagonist SCH-23390 (1μM), MEK inhibition (U0126, 1μM), or CREB inhibition (666-15, 1μM). All p values < 0.05 for DA IEGs. included cyclic AMP response elements, which is consistent the adult NAc, primary striatal neuron cultures were devoid with previous reports demonstrating a key role for CREB in of oligodendrocyte, astrocyte, and microglia markers (due drug-induced transcriptional changes (10, 37, 38). to the embryonic stage of tissue harvested for culture), but To determine whether this core transcriptional did harbor polydendrocytes positive for canonical marker response to DA occurs broadly across MSN subtypes or is genes Pdgfra and Olig1 (Supplementary Fig. 6). Similar to limited to specific populations, we performed snRNA-seq cocaine experience in vivo, treatment with DA or SKF-38393 from 3,361 cultured striatal neurons (mixed from male resulted in selective transcriptional activation of Drd1-MSNs and female E18 rat brains) in 4 distinct treatment groups (Fig. 2e). This activation included core DA signature genes (Vehicle, 50µM DA, 1µM of the DRD1 receptor agonist identified from bulk RNA-seq (Fos, Fosb, Junb, Nr4a1, and SKF-38393, or 25mM potassium chloride (KCl) to induce Nr4a2), demonstrating robust induction of this gene program neuronal depolarization). Cultured neurons were again in Drd1-MSNs (Supplementary Fig. 7; Supplementary largely divided into defined Drd1-MSNs, Drd2-MSNs, and Table 8). Moreover, expression of nearly all of these genes Grm8-MSNs (Fig. 2d). Unlike scRNA-seq datasets from was positively correlated with unbiased EES scores for DA 4 bioRxiv preprint doi: https://doi.org/10.1101/781872; this version posted September 25, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. and SKF-38393 treatments only in Drd1-MSNs (Fig. 2f; receptor, interference with MAP kinase signaling cascades Supplementary Table 9), highlighting the key contribution linked to CREB activation, or inhibition of CREB itself. of this program to overall transcriptional changes in these To test these hypotheses, we performed RT-qPCR for key neurons. In contrast, both Drd1-MSN and Drd2-MSN clusters DA-response genes 1hr following DA stimulation (1µM) in responded with massive transcriptional alterations following the presence of either a DRD1 receptor antagonist (SCH- neuronal depolarization with KCl, which was reflected in the 23390, 1µM), a Mitogen-activated protein kinase kinase total number of DEGs, the fraction of cells activated, and the (MEK) inhibitor (U0126, 1µM), or a CREB inhibitor (666- percentage of reads aligning to DA-responsive genes (Fig. 2e 15, 1µM). In agreement with our prediction, we found that and Supplementary Fig. 7). Together, these results suggest induction of each DA-responsive gene was inhibited by each that both Drd1- and Drd2-MSNs are capable of producing a of these treatments (Fig. 2h). These results confirm that robust transcriptional response involving the same core gene DA-responsive gene programs are DRD1-dependent in this set, but that only Drd1-MSNs exhibit robust transcriptional culture model system and demonstrate that these effects responses to elevated DA neurotransmission in this model. require MAP kinase signaling and classical CREB-mediated Intriguingly, expression of this core gene set was positively transcriptional mechanisms. correlated with EES score in the adult snRNA-seq dataset in Although individual candidate genes in this DA- multiple MSN clusters following cocaine experience (Fig. responsive transcriptional program have been shown to 2g), potentially reflective of multimodal (e.g., DA-dependent play key roles in drug-induced adaptations (39, 40), it and DA-independent) contributions to this gene expression is hypothesized that gene expression programs work in profile after cocaine exposure. concert to exert downstream functional effects (30, 41). Given that DA-responsive genes were preferentially However, defining the significance of key gene expression activated in Drd1-MSNs and that these genes are enriched programs has remained challenging due to the lack of tools for CREB binding motifs, we hypothesized that responsivity capable of interrogating large-scale polygenic changes. of these genes could be prevented via inhibition of the DRD1 Recent developments in CRISPR-based technology have a b c CRISPR activation mCherry Experimental sgRNA VPR timeline Lenti hSyn dCas9 VPR hSyn-dCas9-VPR dCas9 FLAG NLS DIV 0 Neurons seeded Target PAM Lenti sgRNA LacZ multiplex sgRNA U6 EF1α mCherry multiplex array 3-4 Viral Gene A Gene D transduction 8x repeat

Gene B Gene E Control Dopaplex I Dopaplex II Tent5b Nptx1 Fosb Sstr2 8x LacZ -or- Junb Gsx1 + Nr4a1 Gadd45b Gene C Gene F Prox1 Gadd45g Egr4 Egr3 11 RT-qPCR/ Nr4a2 Btg2 Fos Egr2 RNA-seq Dopaplex Pt. I & II g d e f Dopaplex induction RNA-seq Dopaplex gene ontology enrichment Tent5b Dopaplex induction (Biological process; all DEGs) Junb DOWN UP 670 DEGs (339▲, 331▼) DEGs DEGs Prox1 50 Nr4a2 neuron projection morphogenesis Nptx1 Gsx1 40 cation channel activity Gadd45g voltage-gated ion channel activity Btg2 30 regulation of synaptic plasticity Fosb potassium ion transmembrane transporter activity Nr4a1

Gene ID Egr4 negative regulation of actin filament depolymerization Fos 20 forebrain cell migration Sstr2 positive regulation of neuron projection development Gadd45b 10 Egr3 negative regulation of neurogenesis all p < 0.05 negative regulation of Ras protein signal transduction Egr2 0 Wald statistic (abs. val.) 0 2 4 6 8 -4 -2 0 2 4 6 8 10-4 10-2 100 50 0 50 100 log2(Fold change) log2(Fold change) Adjusted % genes Target/LacZ Dopaplex/LacZ p-value in term Figure 3. Multiplexed CRISPRa engineering to mimic dopamine-induced gene expression changes. a, Illustration of the CRISPRa multiplex vector approach expressing the dCas9-VPR activator fusion and multiplex single guide RNAs targeting either the bacterial LacZ gene (non-targeting control) or 16 of the top dopamine-induced genes (Dopaplex). b, Live cell imaging reveals successful transduction of the CRISPRa lentiviruses. Scale bar = 50µm. c, Experimental timeline for in vitro Dopaplex induction. Primary striatal neurons were generated and transduced with multiplexed CRISPRa constructs at DIV3-4. On DIV11, RNA was extracted and subjected to both RT-qPCR and RNA-sequencing to examine gene expression. d, dCas9-VPR increases gene expression of Dopaplex genes compared to the multiplexed LacZ control (n = 8, unpaired t-test; Tent5b t13.4 = 49.6, p < 0.0001; Junb t13.62 = 29.97, p < 0.0001; Prox1 t12.05 = 11.64, p < 0.0001; Nr4a2 t7.94 = 16.82, p < 0.0001; Nptx1 t13.75 = 7.722, p < 0.0001; Gsx1 t13.71 = 9.226, p < 0.0001; Gadd45g t13.98 = 9.435, p < 0.0001; Btg2 t9.586 = 29.71, p < 0.0001; Fosb t7.423 = 28.64, p < 0.0001; Nr4a1 t10.35= 29.52, p < 0.0001; Egr4 t8.382= 2.388, p = 0.0427; Fos t13.96 = 6.982, p < 0.0001; Sstr2 t9.132 = 20.70, p < 0.0001; Gadd45b t8.104 = 29.48, p < 0.0001; Egr3 t11.87 = 19.00, p < 0.0001; Egr2 t7.828 = 27.37, p < 0.0001;). All data are expressed as mean ± s.e.m. e, RNA-seq volcano plot showing DEGs detected by DESeq2 in Dopaplex verses LacZ multiplex targeting conditions. Upregulated genes (salmon, 339 genes) and downregulated genes (yellow, 331 genes) are indicated, and a standard Wald statistic cutoff corresponding to adjusted p < 0.05 is represented by the horizontal dotted line. f, Top 10 significant molecular function (GO) for all DEGs (excluding 16 Dopaplex genes) after Dopaplex induction. 5 bioRxiv preprint doi: https://doi.org/10.1101/781872; this version posted September 25, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. enabled rapid and multiplexable transcriptional control activate multiple genes in this DA-induced gene expression in the mammalian central nervous system, providing an profile, we harnessed a neuron-optimized CRISPR activation avenue to investigate how gene programs regulate normal (CRISPRa) system (Fig. 3) in which a catalytically dead Cas9 and maladaptive brain states (35, 42-44). To simultaneously protein (dCas9) is fused to the strong hybrid transcriptional

a MEA experimental timeline c Active d Spike e Spike DIV units frequency frequency Neurons Live cell imaging 15 3 100 0 ns seeded ns 10 2 80 LacZ Viral 3-4 60 transduction 5 1 Units/well 40 LacZ Mean frequency (Hz) 729 824 Frequency (% neurons) Dopaplex 0 0 20 min. 0 5 10 15

11 Dopaplex recording LacZ LacZ Mean frequency (Hz) Dopaplex Dopaplex b Representative traces f g h Burst Burst Burst frequency LacZ Dopaplex frequency frequency (single gene targeting) 1.5 100 * ns 2 1.0 90

1 0.5 80 Bursts/min Bursts/min 10 µV LacZ 1s Dopaplex 0.0 70 0 Frequency (% neurons) 0 5 10 15 LacZ LacZ Fosb Bursts/min Tent5b Dopaplex i j Experimental timeline - CRISPRa + Dopaplex targeting mCherry DAPI lv 2 groups: LacZ multiplex sgRNAs + dCas9-VPR Dopaplex Pt. I/II sgRNAs + dCas9-VPR NAc core Cocaine locomotor sensitization

10mg/kg Cocaine ac Saline Cocaine challenge

Viral 18d 14d infusion

k l m Cocaine Cocaine Cocaine challenge locomotion locomotion locomotion 20000 10 40 1500 LacZ LacZ 10 mg/kg * Dopaplex 8 *Dopaplex 30 15000 * cocaine, I.P. 1000 6 10 10000 4 5 500

5000 (norm. to C1) (norm. to C1) 2 LacZ Distance traveled Distance traveled Dopaplex

0 Distance traveled (cm) 0 Distance traveled (cm) 0 0 Sal Coc 1 2 3 4 5 0 20 40 60 Cocaine session LacZ Time (min) Dopaplex

Figure 4. Dopaplex gene expression program increases bursting activity of MSNs and enhances cocaine locomotor sensitization. a, (left) Experimental timeline for viral transduction and MEA recordings. (right) Live cell imaging reveals successful transduction of the CRISPRa lentiviruses. b, Representative traces from 2 units each from the LacZ and Dopaplex targeting groups. The number of active units per well (c) and action potential frequency (d-e) did not change between LacZ and Dopaplex-targeted conditions (active units: n = 72 wells, Student’s t-test, p = 0.275; frequency: n = 729-824 neurons per group, Mann-Whitney U test, U = 296405, p = 0.6548). f, Burst frequency is increased in Dopaplex-targeted neurons (n = 729-824 neurons per group, Mann-Whitney U test, U = 282388, p = 0.0265). g, Cumulative distribution of burst frequency as a percent of all neurons. h, Burst frequency was not altered when CRISPRa was targeted to single genes within the Dopaplex gene expression program (n = 356-372 neurons per group, one-way ANOVA with

Dunnett’s test for multiple comparisons, F2, 1087 = 2.295, p = 0.1012). i, Experimental timeline for viral transduction and cocaine locomotor sensitization. j, Representative image of sgRNA viral transduction (mCherry reporter) in the NAc. k, Absence of baseline locomotion difference between LacZ and Dopaplex targeted animals following saline injection or the first cocaine injection (n = 12-14 rats per group, Mann-Whitney U test, U = 65, p = 0.3474). l, Over 5 cocaine administration sessions, Dopaplex-targeted animals exhibited increased locomotion (n = 12-14 rats per group, two-way ANOVA with main effect of sgRNA, F1,24 = 4.412, p = 0.0464). m, Induction of sensitization with a cocaine challenge injection 14 days after the last cocaine pairing enhanced locomotion in Dopaplex-targeted rats (n = 12-14 rats per group, Mann-Whitney U test, U = 41, p = 0.0270). All data are expressed as mean ± s.e.m. Individual comparisons, *p < 0.05.

6 bioRxiv preprint doi: https://doi.org/10.1101/781872; this version posted September 25, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. activator VPR (a concatemer of the herpes simplex overexpressing either Fosb or Tent5b did not differ from LacZ viral protein VP16, the p65 subunit of NF-kB, and the sgRNA neurons in any physiological measure (active units, gammaherpesvirus transactivator Rta) (35, 45). This system firing rate, or burst frequency; Fig. 4h), indicating that the allows multiplexed gene programming through the design increase in burst firing following induction of the Dopaplex of single guide RNAs (sgRNAs) that target the promoter program was not due to the induction of either of these genes region of selected genes. Using this system, we designed alone. sgRNAs to target 16 of the genes most robustly altered by Our results suggest that DA signaling results in DA receptor activation (including Fos, Fosb, and Nr4a1; Fig. an altered transcriptional profile that supports increased 3a). Following validation of each sgRNA using individual burst firing of MSNs, which is consistent with reports gene targeting in striatal cultures (Supplementary Fig. 8), that prior drug exposure alters the intrinsic properties of we engineered custom multiplex sgRNA lentiviruses that MSNs and enhances excitability (17). Therefore, we sought contained eight sgRNAs each for a total of 16 unique sgRNAs to understand how this gene expression program might (Fig. 3a-d). Expression of these vectors in striatal neurons in regulate behavioral changes observed after exposure to drugs tandem with neuron-specific expression of the dCas9-VPR of abuse. One such behavior is locomotor sensitization, fusion protein enabled simultaneous induction of all genes in which DA transmission in the NAc is potentiated as (which we termed “Dopaplex”) as compared to a control sensitization develops (46, 47). To examine the function multiplex virus expressing an sgRNA array targeting the of this Dopaplex gene expression program on cocaine bacterial gene LacZ (Figure 3d). locomotion sensitization, we stereotaxically infused CRISPRa Since many of the top DA-responsive genes lentiviral vectors targeting Dopaplex genes (or LacZ control) were transcription factors, we next sought to define the bilaterally into the NAc core (Fig. 4i-j, Supplemental Figure transcriptional consequences of activating this program. 9), a region critical for the development and maintenance RNA-seq comparisons following CRISPRa targeting LacZ or of sensitization (48, 49). We previously validated robust Dopaplex identified 670 DEGs, with 339 upregulated genes and neuron-selective expression of this CRISPRa system and 331 downregulated genes (Figure 3e, Supplemental in vivo within 2 weeks of viral delivery into the NAc, and Table 10). After removal of the 16 targeted Dopaplex genes demonstrated that this tool results in elevated protein levels from the gene list, gene ontology analysis revealed that of target genes (35). Eighteen days after viral infusion, critical neuronal processes were altered in the Dopaplex- rats underwent locomotor sensitization testing in a novel targeted group, including genes involved in the regulation environment using a dose of cocaine (10mg/kg) that does of ion channels, projection morphogenesis, and synaptic not reliably produce robust sensitization (50, 51). Animals plasticity (Figure 3f; Supplemental Table 11). These results receiving the CRISPRa constructs targeting Dopaplex demonstrate that activation of a limited set of DA-sensitive genes did not differ fromLacZ multiplex controls in genes can result in large-scale transcriptional consequences, baseline locomotion after an injection of saline, nor in acute including gene targets that might promote neurophysiological increases in locomotion following the first dose of cocaine alterations. (Fig. 4k). However, we observed a significant increase in Exposure to drugs of abuse or alterations in CREB the development of locomotor sensitization after 5 cocaine signaling influence the intrinsic physiological properties injections in Dopaplex-targeted rats (Fig. 4l). Likewise, this of MSNs (17, 19, 34). Therefore, we sought to understand enhancement in locomotor activity was maintained 14 days whether Dopaplex induction altered the physiological after the last cocaine pairing by administration of a cocaine activity of MSNs using a high-throughput multielectrode challenge dose (Fig. 4m). These results indicate that DA- array system in which striatal neurons are seeded directly mediated gene expression changes are sufficient to drive on a 768 electrode array divided across 48 culture wells cocaine-related changes in neural physiology and behavior. (Fig. 4a-b). Dopaplex-expressing neurons did not differ Following chronic drug administration, drug- from non-targeting LacZ multiplexed control neurons in induced dopaminergic signaling in the NAc is potentiated both the number of spontaneously active neurons and the (46, 47), providing multiple opportunities for induction of mean action potential frequency (Figure 4c-e). However, long-lasting changes in gene expression that are observed action potential burst frequency was increased in Dopaplex- after repeated drug use 8,( 9, 52). While a candidate gene targeted neurons (Fig. 4f-g), demonstrating that this gene approach has been critical in understanding some of the expression program alters MSN firing patterns. To determine fundamental transcriptional changes following exposure to if this phenotype could be mimicked by targeting individual drugs of abuse, activity-dependent transcriptional events genes in the Dopaplex program, we targeted CRISPRa produce coordinated changes that may be distinct from machinery to either the transcription factor Fosb (which has isolated changes in expression of a single gene. Our results been frequently implicated in cocaine action (10, 39, 40)), shed new light on this process by identifying specific neuronal the nucleotidyltransferase Tent5b (the gene with the highest populations altered by cocaine experience, and further fold change after DA stimulation), or LacZ control. Neurons defining unique DA-responsive gene programs that control

7 bioRxiv preprint doi: https://doi.org/10.1101/781872; this version posted September 25, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. neuronal function and behavioral responses to cocaine. Together, these results allow comprehensive exploration Neuronal Cell Cultures. Primary rat neuronal cultures were of NAc transcriptional programs in an addiction-relevant generated from E18 rat striatal tissue as described previously model species, and identify novel target cells and genes for (35, 36). Briefly, cell culture plates (Denville Scientific Inc.) were perturbation in the context of dopaminergic signaling. Given coated overnight with poly-L-lysine (Sigma-Aldrich; 50 µg/ that addiction is a polygenic disorder that involves a complex ml), supplemented with 7.5 µg/mL laminin (Sigma-Aldrich), interplay between genetic background and experience- and rinsed with diH2O. Microelectrode arrays (MEAs; Axion dependent cellular responses, use of multiplexed gene Biosystems) were coated with polyethyleneimine (Sigma- regulation and cell-specific manipulations may enable a more Aldrich). Dissected striatal tissue was incubated with papain complete understanding of how initial drug reinforcement (Worthington LK003178) for 25 min at 37°C. After rinsing transitions to long-term substance abuse disorders. in complete Neurobasal media (supplemented with B27 and L-glutamine, Invitrogen), a single cell suspension was prepared by sequential trituration through large to small fire- Materials and Methods polished Pasteur pipettes and filtered through a 100µm cell strainer (Fisher Scientific). Cells were pelleted, re-suspended Animals. All experiments were performed in accordance in fresh media, counted, and seeded to a density of 125,000 with the University of Alabama at Birmingham Institutional cells per well on 24-well culture plates (65,000 cells/cm2) or Animal Care and Use Committee. Sprague-Dawley timed 30,000 cells per well on 48-well MEA plates. Cells were grown pregnant dams and 90-120-day-old male or female rats were in complete Neurobasal media for 11 days in vitro (DIV 11) purchased from Charles River Laboratories. Dams were in a humidified CO2 (5%) incubator at 37°C with half media individually housed until embryonic day 18 (E18) for cell changes at DIV 1, 4-5, and 8-9. MEAs received a one-half culture harvest. Male or female adult rats were co-housed in media change to BrainPhys (Stemcell Technologies Inc.) pairs in plastic filtered cages with nesting enrichment in an with SM1, L-glutamine supplements starting on DIV 4-5 and AAALAC-approved animal care facility maintained between continued every 3-4 days. All MEA media was supplemented 23-24°C on a 12hr light/dark cycle with ad libitum food (Lab with penicillin-streptomycin. Diet Irradiated rat chow) and water. Bedding and enrichment were changed weekly by animal resources program staff. RNA extraction and RT-qPCR. Total RNA was extracted Animals were randomly assigned to experimental groups. (RNAeasy kit, Qiagen) and reverse-transcribed (iScript cDNA Synthesis Kit, Bio-Rad). cDNA was subject to RT- Drugs. Cocaine hydrochloride was dissolved in sterile qPCR for genes of interest, as described previously (36). A 0.9% sodium chloride. Drugs were purchased from Sigma- list of PCR primer sequences is provided in Supplemental Aldrich (St. Louis, MO) and injected intraperitoneally (i.p.) Table 12. at a dose of 20mg/kg or 10mg/kg for locomotor testing and locomotor sensitization, respectively. Cocaine solution Tissue collection from adult NAc. One hour after i.p. was made fresh immediately before behavioral testing and injection of either saline vehicle or 20mg/kg cocaine, rats protected from light. For in vitro experiments, drugs were were euthanized by live decapitation, and the brain was diluted in Neurobasal medium (Invitrogen) immediately rapidly removed and blocked into coronal sections in prior to treating cell culture plates. Dopamine hydrochloride ice-cold Hibernate A media (Thermo Fisher, A1247501) (Sigma-Aldrich, H8502-5G) was dissolved in Neurobasal supplemented with B27 and GlutaMax (Life Technologies, medium, and cells were treated at a dose of 1µM for RT-qPCR 35050-061). Tissue punches were then collected from coronal experiments or 50µM for snRNA-sequencing experiments. sections containing the NAc core and shell (~ +1.68 AP from R(+)-SCH-23390 hydrochloride (Sigma-Aldrich, D054- bregma; n=4 rats/treatment/sex) and rapidly frozen on dry 5MG) and R(+)-SKF-38393 hydrochloride (Sigma-Aldrich, ice. Tissue was stored at -80°C until the day of sequencing. S101-5MG) were dissolved in sterile Milli-Q water, and cells were treated at a dose of 1µM. For MEK inhibitor Single Nuclei Dissociation. Frozen tissue from adult NAc. experiments, U0124 (1µM; Millipore, 662006-1MG) and Frozen tissue punches were thawed slowly on ice prior U0126 (1µM; Millipore, 662005-1MG) were dissolved in to being chopped by scalpel 100 times in 2 orthogonal DMSO (Invitrogen, D12345). The CREB inhibitor 666-15 directions. Tissue from 4 animals in the same treatment (1µM; Tocris, 5661), also called 3-(3-Aminopropoxy)-N- condition was then combined and transferred to 5mL of [2-[[3-[[(4-chloro-2-hydroxyphenyl)amino]carbonyl]- chilled (4°C) Lysis Buffer (10mM Tris-HCl, 10mM NaCl,

2-naphthalenyl]oxy]ethyl]-2-naphthalenecarboxamine 3mM MgCl2 (Thermo Fisher, BP214-500), 0.1% Igepal in hydrochloride, was dissolved in DMSO. Potassium chloride Nuclease-Free Water (Sigma, 18896-50ML)) for 15 min, (Fisher Scientific, P330-500) was dissolved in Neurobasal mixing the tissue by inversion every 2 min. Lysis was media, and cells were treated with a 25mM dose. quenched after 15 min with 5mL of complete Hibernate-A

8 bioRxiv preprint doi: https://doi.org/10.1101/781872; this version posted September 25, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. (Thermo Fisher, A1247501) supplemented with B27 and reads mapping to the mitochondrial genome were removed Glutamax (Life Technologies, 35050-061). Tissue was then from further analysis. Molecular count data from each GEM triturated by fire-polished Pasteur pipette (2 pipettes of (Gelbeads in Emulsion) well was then log-normalized with a decreasing diameter; 8-10 passes per pipette) for 35 min scaling factor of 10,000. Following normalization, and before and passed through a 40mm pre-wet filter. The samples dimensionality reduction, all GEM wells were integrated were then pelleted at 500rcf for 10 min at 4C followed by a with FindIntegrationAnchors() and IntegrateData(), using wash in 10mL of Nuclei Wash & Resuspension Buffer (1X 17 principal components (PCs) for the adult dataset and PBS, 1% BSA, 0.2U/ul NxGen RNase inhibitor (Lucigen, 10 PCs for the culture dataset. Dimensionality reduction is 30281-2)). Supernatant was then removed and the pellet dependent on two main parameters, number of principal was gently resuspended in a total volume of 800µl prior to components and the resolution value. To determine the 7-AAD (Thermo Fisher, 00-6993-50) staining and FACS to appropriate values, UMAPs were generated for every further purify the nuclei for sequencing. Immediately after combination of 15 different PCs (2-30) and 20 different FACS, nuclei were washed a final time at 200rcf in 10mL resolution values (0.1-2.0), using the standard Louvain supplemented Hibernate-A containing 1% BSA for 10 min at algorithm. To produce UMAPs that capture previously 4°C to remove any remaining fine debris and nascent RNA. defined cell types, we used 17 principal components and Nuclei were then brought to a concentration of 1,700 nuclei/ resolution values of 0.2 for the adult dataset, and 10 principal µL. Finally, an average total of 2,400 nuclei pooled from 4 components with a 0.1 resolution value for the culture rats per sex and treatment group were loaded into a single dataset. To validate that the clusters are specific and consist well of the Chromium Single Cell B Chip (10x Genomics of known cell types, the log-normalized expression value of catalog #1000074), utilizing 4 of the 8 available wells. the top 2-6 genes from each striatal cell type, as identified Striatal cell culture dissociation. On DIV11, striatal primary by publicly available single-cell sequencing databases (e.g., cells cultured on 12-well plates (250K/well) were treated with DropViz.org (23) and mousebrain.org (24)), found within Neurobasal vehicle, 50µM DA, 1µM SKF, or 25mM KCL, and each dataset was overlaid on the UMAP using FeaturePlot(). placed back into the incubator for 1hr. Cells were then washed To identify stimulation-dependent DEGs within each cell with 1ml PBS per well, followed by a 5 min incubation period type, a Wilcoxon Ranked Sum test was performed on the in 600µL Lysis Buffer. Cells were then gently pipette mixed log-normalized corrected unique molecular identifier (UMI) 5 times with a P1000 before being transferred to a 1.5mL counts using custom R scripts based on Seurat source code. tube and centrifuged at 500rcf for 5 min at 4°C. Supernatant P-values were then adjusted using a Bonferroni correction was then removed and the pellet resuspended in 100µL based on the total number of genes identified within the Nuclei Wash Buffer. Tubes for each treatment group were dataset. Biological process gene ontology and transcription then combined, gently pipette mixed 5 times, and passed factor motif enrichment analysis for DEGs from specific cell through a 40µM filter prior to FACS to further purify nuclei clusters was performed in WEB-based Gene Set Analaysis for sequencing. After FACS, 2,400 nuclei per treatment Toolkit (WebGestalt) (55), using all protein coding genes as a condition were loaded into individual wells of the Chromium reference set. Correction for multiple testing was performed Single Cell B Chip (10x Genomics, #1000074), utilizing 4 of using Benjamani-Hochberg adjustment, with statistical the 8 available wells. Chromium single nuclei capture and significance assessed at FDR < 0.05. barcoding were completed on the Chromium Single Cell Enhanced Experimental Signal (EES) was calculated Controller, according to manufacturer’s instructions. using the Manifold Enhancement of Latent Dimensions (MELD, version 0.0) (29) package on a cell-state graph Single Nuclei RNA Sequencing and Analysis. Libraries were generated using 100 PCA dimensions calculated with edge constructed according to manufacturer’s instructions using weights between cells generated through the Graphtools the Chromium Single Cell 3’ library construction kit (10X library with a knn=9 and a decay=10 using Python v3.7.3 Genomics, #1000092), which utilizes version 3 chemistry after removing lowly expressed genes (genes not expressed for gene expression. 15,676 nuclei from adult rat NAc and in at least 10 cells (adult data set) or 5 cells (culture dataset)). 3,601 nuclei from cultured primary rat striatal neurons were The saline and cocaine experimental labels in the cell state sequenced on the Illumina NextSeq500 at the UAB Heflin graph were set with a smoothing parameter of Beta=1. The Genomics Core to a depth of ~38,000 reads per nuclei and Markov Affinity-based Graph Imputation of Cells (MAGIC, ~147,201 reads per nuclei, respectively. A Cell Ranger (v3.0.2) v1.5.5)(56) package was utilized to impute gene expression reference package was generated from the Ensembl Rn6 using default settings on the cell-state graph. To determine rat genome with a modified GTF file (version 95) to create genes which vary with the EES, pairwise Pearson correlation a custom pre-mRNA package which ensured alignment to r values were calculated between each gene and the EES by unspliced pre-mRNAs and mature mRNAs. Cell Ranger cluster using SCPREP (v0.12.1). To generate the ridgeline filtered outputs were analyzed with Seurat v3.0.2 using R plots, pairwise EES-gene Pearson correlation r values were v3.6.0(53, 54). Nuclei containing <200 features and >5% of calculated by cluster and ranked. Ridgeline plots reflect the

9 bioRxiv preprint doi: https://doi.org/10.1101/781872; this version posted September 25, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. ranks of 100 upregulated DEGs identified in the DA bulk DA-induced genes were inserted by PCR amplification and RNA-seq experiment. The circos plot was generated using Golden Gate assembly into a custom-made destination CIRCOS version 0.69 in Perl v5.18.4. The size of the outer plasmid co-expressing mCherry with unique BsmBI sites. rings represents the number of DEGs within each cluster. All lentiviruses were produced as described previously (35), Genes were positioned by adjusted p values, and arcs were and titers for in vitro experiments were > 1x1011 GC/mL and then drawn between genes that changed in the same direction > 1x1012 for in vivo experiments. Titers were determined in pairwise clusters and color coded to indicate neuron- with the Lenti-X RT-qPCR titer kit (Clonetech). neuon, glia-glia, and neuron-glia matches. ROC analysis was performed with a subset of snRNA-seq data from the Lentivirus production. Viruses were produced in a sterile parent Drd1-MSN cluster in Seurat v3.0.2. Dimensionality environment subject to BSL-2 safety by transfecting HEK- reduction was performed with 17 PCs and resolution of 0.2. 293T cells (ATCC CRL-3216) with the specified CRISPR For ROC evaluation, a binary classifier was built with each plasmid, the psPAX2 packaging plasmid, and the pCMV- gene using FindMarkers(test.use = “roc”, assay = “RNA”, slot VSV-G envelope plasmid (Addgene 12260 & 8454) with = “data”, logfc.threshold = 0, min.pct = 0.1). FuGene HD (Promega) for 40-48 hrs in supplemented Ultraculture media (L-glutamine, sodium pyruvate, and CRISPR-dCas9 construct design. CRISPRa experiments sodium bicarbonate) in a T225 culture flask. Supernatant was used lentivirus compatible plasmid constructs we previously passed through a 0.45µm filter and centrifuged at 106,883rcf optimized for robust neuronal expression (35). Gene-specific for 1 hr 45 min at 4°C. The viral pellet was resuspended in sgRNA targets were designed using online tools provided by 1/100th (in vitro) or 1/1000th (in vivo) supernatant volume the Zhang Lab at MIT (crispr.mit.edu) and CHOPCHOP of sterile PBS and stored at -80°C. Physical viral titer was (http://chopchop.cbu.uib.no/). To ensure specificity, all determined using Lenti-X qRT-PCR Titration Kit (Takara), CRISPR RNA (crRNA) sequences were analyzed with and only viruses greater than 1x1011 GC/ml were used. Viruses National Center for Biotechnology Information’s (NCBI) were stored in sterile PBS at -80°C in single-use aliquots. For Basic Local Alignment Search Tool (BLAST). A list of the smaller scale virus preparation, each sgRNA plasmid was target sequences is provided in Supplemental Table 12. transfected in a 12-well culture plate as described above. crRNAs were annealed and ligated into the sgRNA scaffold After 40-48 hrs, lentiviruses were concentrated with Lenti-X using the BbsI cut sites. Plasmids were sequence-verified concentrator (Takara), resuspended in sterile PBS, and used with Sanger sequencing. immediately. Selected CRISPRa target genes were the genes with the highest confidence changes in gene expression following Multi Electrode Array Recordings. Single neuron

DA treatment. We applied a log2 fold change cutoff of > 1.2 on electrophysiological activity was recorded using an Axion the bulk DA RNA-seq dataset, resulting in 23 potential target Maestro recording system (Axion Biosystems). E18 rat genes (Supplemental Table 6). We first designed sgRNAs primary striatal neurons were seeded in 48-well MEAs at to target each gene individually, excluding 3 genes due to 30,000 cells/well, as described above. Each MEA well contains significant predicted off-target localization of CRISPRa 16 extracellular recording electrodes and a ground electrode. sgRNAs (Supplemental Figure 2a). At DIV3-4, primary Neurons were transduced with CRISPRa constructs on DIV striatal neurons were transduced with lentiviruses expressing 3-4 and MEA recordings were performed at DIV 11, while sgRNAs targeting either a negative control sequence (the connected to a temperature- and CO2-controlled headstage bacterial LacZ gene) or one of the high confidence DA- (monitored at 37°C and 5% CO2). Electrical activity was induced genes (Supplemental Figure 8a). Compared to the measured by an interface board at 12.5 kHz, digitized, and non-targeting control, we measured significant induction of transmitted to an external computer for data acquisition and 17 of the high confidence genes. To determine if these genes analysis in Axion Navigator software (Axion Biosystems). All could be induced simultaneously, we pooled all 17 sgRNA data were filtered using dual 0.01 Hz (high pass) and 5,000 Hz viruses (Supplemental Figure 8b) and observed that all but (low-pass) Butterworth filters. Action potential thresholds one gene (Fosl2) was induced by this approach. Therefore, were set automatically using an adaptive threshold for each we excluded Fosl2 from the multiplex CRISPRa targeting electrode (> 6 standard deviations from the electrode’s mean experiments. To deliver sgRNAs for each of these targets in signal). Neuronal waveforms collected in Axion Navigator a more efficient way, we generated new multiplex sgRNA were exported to Offline Sorter (Plexon) for sorting of lentiviruses that contained eight sgRNA sequences each distinct waveforms corresponding to multiple units on one (driven by distinct U6 promoters) for a total of 16 unique electrode channel, and confirmation of waveform isolation sgRNAs between 2 viruses (Figure 3a). The bacterial LacZ using principal component analysis, inter-spike intervals, gene target was again used as a sgRNA non-targeting control and auto- or cross-correlograms. Further analysis of burst by constructing an 8x LacZ sgRNA array. For multiplexing activity and firing rate was performed in NeuroExplorer sgRNAs, individual sgRNAs targeting either LacZ or the (v.5.0).

10 bioRxiv preprint doi: https://doi.org/10.1101/781872; this version posted September 25, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. apparatus (Kopf Instruments). During surgical procedures, Bulk RNA-Sequencing. Bulk RNA-Sequencing (RNA- an anaesthetic plane was maintained with 1–2.5% isoflurane. Seq) was carried out at Hudson Alpha Genome Services Under aseptic conditions, guide holes were drilled using Laboratory or the Heflin Center for Genomic Science Paxinos and Watson(61) stereotaxic coordinates (all Genomics Core Laboratories at the University of Alabama coordinates in respect to bregma: AP: +1.6 mm, ML: ±1.4 at Birmingham. RNA was extracted, purified (RNeasy, mm, DV: -7.0) to target the NAc core. All infusions were Qiagen), and DNase-treated for three biological replicates made using a gastight 30-gauge stainless steel injection per experimental condition. 1µg of total RNA underwent needle (Hamilton Syringes) that extended into the infusion quality control (Bioanalyzer) and was prepared for site. Bilateral lentivirus microinfusions of (1.5µl total volume directional RNA sequencing using NEBNext reagents (New per hemisphere) were made using a syringe pump (Harvard England Biolabs) or SureSelect Strand Specific RNA Library Apparatus) at a rate of 0.25µl/min. Injection needles remained Prep Kit (Agilent Technologies) according to manufacturer’s in place for 10 min following infusion to allow for diffusion. recommendations. PolyA+ RNA libraries underwent Rats were infused bilaterally with either 1.5 µl of total sequencing (50-75 bp paired-end directional reads; ~25 lentivirus mix comprised of 0.75µl sgRNA and 0.75µl dCas9- M reads/sample) on an Illumina sequencing platform VPR viruses in sterile PBS. After infusions, guide holes were (HiSeq2000 or NextSeq2000). covered with sterile bone wax and surgical incision sites were closed with nylon sutures. Animals received buprenorphine Bulk RNA-Seq Data Analysis. Paired-end FASTQ files were and carprofen for pain management and topical bacitracin to uploaded to the University of Alabama at Birmingham’s High prevent infection at the incision site. Performance Computer cluster for custom bioinformatics analysis using a pipeline built with snakemake (57) (v5.1.4). Acute cocaine locomotion testing. For basic locomotor Read quality, length, and composition were assessed using testing, naïve male (n=8) and female (n=8) animals were given FastQC prior to trimming low quality bases (Phred < 20) a single i.p. injection of either cocaine (20mg/kg) or saline and Illumina adapters (Trim_Galore! v04.5). Splice-aware immediately before being placed in the activity chamber for alignment to the Rn6 Ensembl genome assembly (v90) was 30 min of testing. Testing took place over two consecutive performed with STAR (58) v2.6.0c. An average of 88.4% days between 2:00 PM and 5:30 PM, and sessions were of reads were uniquely mapped. Binary alignment map counterbalanced across time by sex and treatment condition. (BAM) files were merged and indexed with Samtools (v1.6). The same female experimenter conducted locomotor testing Gene-level counts were generated using the featureCounts and was present in the room during testing. Behavior (59) function in the Rsubread package (v1.26.1) in R chambers consisted of 43cm x 43cm Plexiglass locomotor (v3.4.1), with custom options (isGTFAnnotationFile = activity chambers (Medical-Associates, St. Albans, VT) with TRUE, useMetaFeatures = TRUE, isPairedEnd = TRUE, opaque white wall covering and an open top. Each chamber requireBothEndsMapped = TRUE, strandSpecific = 2, and included a 48-channel X-Y infrared array (Medical- autosort = TRUE). DESeq2 (60) (v 1.16.1) in R was used to Associates, St. Albans, VT) that was used to measure distance perform count normalization and differential gene expression traveled in conjunction with Activity Monitor software analysis with the application of Benjamini-Hochberg false (Medical-Associates, St. Albans, VT). Test chambers were discovery rate (FDR) for adjusted p-values. Differentially cleaned with 0.0156% chlorhexidine and 70% ethanol at the expressed genes (DEGs) were designated if they passed a p beginning and end of each testing day, and cleaned with 70% < 0.05 adjusted p-value cutoff and contained basemean > 39. ethanol in between trials. Animals were transported to the Gene ontology (GO) analysis was conducted on DA behavioral testing core 30 min prior to testing. Only the two (compared to vehicle control), and CRISPRa-induced animals currently undergoing locomotor testing were kept Dopaplex gene expression program (compared to LacZ in the testing room; other animals were kept in an adjacent sgRNA control) using the ClueGO application in Cytoscape. room. Behavior sessions were conducted during the light Overrepresentation enrichment analysis was performed cycle, and the overhead lights and white noise generator in using non-redundant terms in biological process, using the the behavior room remained on. protein-coding rat genome as a reference set. Enrichment analysis applied Benjamini-Hochberg correction for Cocaine locomotor sensitization. After 2 weeks to allow for multiple comparisons and required a minimum of 3 genes recovery and viral expression, animals were habituated to per enriched GO term category. Transcription factor binding handling for a minimum of 4 days before locomotion testing prediction used the WEB-based Gene Set Analaysis Toolkit began. Locomotor sensitization experiment sessions were 1 (WebGestalt) (55). hr, with a 15 min habituation time before an i.p. injection of either saline (day 1) or 10mg/kg cocaine (days 3, 5, 7, 9, 11, Stereotaxic Surgery. Naïve adult Sprague-Dawley rats were and 25). anaesthetized with 4% isoflurane and secured in a stereotaxic

11 bioRxiv preprint doi: https://doi.org/10.1101/781872; this version posted September 25, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. Immunohistochemistry. Immunohistochemistry (IHC) was This work was supported by NIH grants DA039650, performed as described previously(35). Adult male rats were DA034681, and MH114990 (J.J.D.), DA042514 (K.E.S.), transcardially perfused with formalin (1:10 dilution in PBS, DA041778 (F.A.S.). L.I. is supported by the Civitan Fisher) 24-48 hr after the cocaine challenge. Brains were International Research Center at UAB. Additional assistance removed and post-fixed for 48 hr in formalin, then sliced at to J.J.D. was provided by the UAB Pittman Scholars Program 50µm using a vibratome. Tissue slices were permeabilized and the UAB Flow Cytometry Core. We thank Samantha with 0.25% Triton X-100 in PBS, then blocked for 1 hr at room Black for assistance with stereotaxic surgery and all current temperature with blocking buffer (1x PBS with 10% Thermo and former Day Lab members for assistance and support. Blocker BSA and 1% goat serum). Expression of mCherry We specifically thank Mikael Guzman Karlsson and Svitlana was examined by incubation with an anti-mCherry primary Bach for their contributions to establishment of single-cell antibody (1:500 in PBS with 10% Thermo Blocker BSA and RNA sequencing platforms in the lab. We thank the UAB 1% goat serum, Abcam catalog #ab183628) overnight at Flow Cytometry Core for assistance with 10X snRNA-seq 4°C. Slices were washed three times with PBS and incubated and FACS. for 1 hr at room temperature with a fluorescent secondary antibody (Alexa Fluor 546 goat anti-rabbit, Thermo Fisher Author contributions Scientific catalog #A-11010, 1:500). Slices were washed three J.J.D., J.J.T., C.G.D., M.E.Z., and R.A.P. conceived of single times with PBS and mounted onto microscope slides with nuclei RNA-seq experiments. M.E.Z. and J.J.T. performed Prolong Gold anti-fade medium (Invitrogen) containing behavioral assays and generated tissue for single-nucleus 4,6-diamidino-2-phenylindole (DAPI) stain as a marker RNA-seq, with assistance from R.A.P. and C.G.D. C.G.D. for cell nuclei. 4X images of the viral infusion site were and J.J.T developed the single-nuclei dissociation protocol taken on a Nikon TiS inverted fluorescent microscope and with assistance from R.A.P., M.E.Z., and the UAB Flow viral placement was mapped onto to coronal diagrams Cytometry Core. R.A.P., C.G.D., & J.J.D. performed statistical adapted from Paxinos and Watson (61) (Supplementary and graphical analysis from single-cell RNA-seq datasets Fig. 9). Animals with strong viral expression in at least one with assistance from L.I. and J.J.T.. M.E.Z. performed hemisphere were included in behavioral studies, and one DRD1 antagonist, MEK inhibitor, and CREB inhibitor animal was excluded due to targeting. experiments. K.E.S. and J.J.D. conceived of the CRISPR/ dCas9 multiplexing experiments. K.E.S., L.I., R.A.P., and Statistical Analysis. Sample sizes were calculated using a J.J.D. performed statistical and graphical analysis of bulk freely available calculator (Lenth, R. V. (2006-9). Java Applets RNA-seq datasets. K.E.S. designed, cloned, and validated for Power and Sample Size [Computer software]. Available at custom sgRNA arrays and performed electrophysiological http://www.stat.uiowa.edu/~rlenth/Power). Correspondence and behavioral assays for CRISPR/dCas9 multiplexing, with between the DA and SKF datasets was determined with assistance from A.J.B., S.T., F.A.S., and N.A.G. All projects a linear regression. Transcriptional differences from RT- were supervised by J.J.D.. K.E.S. and J.J.D. wrote the main qPCR experiments were compared with an unpaired t-test text of the manuscript. K.E.S., C.G.D., J.J.T., R.A.P., M.E.Z., with Welch’s correction or one-way ANOVA with Dunnett’s and J.J.D. wrote the methods section of the manuscript. All or Tukey’s post-hoc tests where appropriate. MEA data authors have approved the final version of the manuscript. was compared with Mann-Whitney U-tests or a one-way ANOVA. Sensitization data was compared with either an Competing Interests unpaired t-test with Welch’s correction or two-way ANOVA. The authors declare no competing financial interests. Statistical significance was designated at a = 0.05 for all analyses. Statistical and graphical analyses were performed References with Prism software (GraphPad). Statistical assumptions (e.g., normality and homogeneity for parametric tests) were 1. C. f. D. C. a. Prevention., 2018 Annual Surveillance Report of Drug-Related Risks and Outcomes — formally tested and examined via boxplots. United States., (2018). 2. B. J. Aragona et al., Preferential enhancement Data Availability. Sequencing data that support the findings of dopamine transmission within the nucleus of this study are available in Gene Expression Omnibus accumbens shell by cocaine is attributable to a (GSE137568, GSE137658, GSE137759, and GSE137763). direct increase in phasic dopamine release events. J Neurosci 28, 8821-8831 (2008). All relevant data that support the findings of this study are 3. G. Di Chiara, A. Imperato, Drugs abused by available by request from the corresponding author (J.J.D.). humans preferentially increase synaptic dopamine With the exception of sgRNA plasmids, CRISPRa constructs concentrations in the mesolimbic system of freely are available in the Addgene plasmid repository. moving rats. Proc Natl Acad Sci U S A 85, 5274-5278 (1988). 4. P. E. Phillips, G. D. Stuber, M. L. Heien, R. M. Acknowledgements Wightman, R. M. Carelli, Subsecond dopamine 12 bioRxiv preprint doi: https://doi.org/10.1101/781872; this version posted September 25, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. release promotes cocaine seeking. Nature 422, 614- other monoamine uptake blockers. J Neurochem 66, 618 (2003). 559-568 (1996). 5. J. J. Day, M. F. Roitman, R. M. Wightman, R. M. 22. C. Bocklisch et al., Cocaine disinhibits dopamine Carelli, Associative learning mediates dynamic shifts neurons by potentiation of GABA transmission in in dopamine signaling in the nucleus accumbens. the ventral tegmental area. Science 341, 1521-1525 Nat Neurosci 10, 1020-1028 (2007). (2013). 6. G. D. Stuber, M. F. Roitman, P. E. Phillips, R. M. 23. A. Saunders et al., Molecular Diversity and Carelli, R. M. Wightman, Rapid dopamine signaling Specializations among the Cells of the Adult Mouse in the nucleus accumbens during contingent Brain. Cell 174, 1015-1030 e1016 (2018). and noncontingent cocaine administration. 24. A. Zeisel et al., Molecular Architecture of the Mouse Neuropsychopharmacology 30, 853-863 (2005). Nervous System. Cell 174, 999-1014 e1022 (2018). 7. E. J. Nestler, C. Luscher, The Molecular Basis of 25. S. I. Walaas, P. Greengard, DARPP-32, a dopamine- Drug Addiction: Linking Epigenetic to Synaptic and and adenosine 3’:5’-monophosphate-regulated Circuit Mechanisms. Neuron 102, 48-59 (2019). phosphoprotein enriched in dopamine-innervated 8. J. Feng et al., Chronic cocaine-regulated epigenomic brain regions. I. Regional and cellular distribution in changes in mouse nucleus accumbens. Genome Biol the rat brain. J Neurosci 4, 84-98 (1984). 15, R65 (2014). 26. L. J. Hachigian et al., Control of Huntington’s 9. I. Maze et al., Essential role of the histone Disease-Associated Phenotypes by the Striatum- methyltransferase G9a in cocaine-induced plasticity. Enriched Transcription Factor Foxp2. Cell Rep 21, Science 327, 213-216 (2010). 2688-2695 (2017). 10. C. A. McClung, E. J. Nestler, Regulation of gene 27. A. Delli Carri et al., Developmentally coordinated expression and cocaine reward by CREB and extrinsic signals drive human pluripotent stem DeltaFosB. Nat Neurosci 6, 1208-1215 (2003). cell differentiation toward authentic DARPP-32+ 11. W. Renthal et al., Genome-wide analysis of medium-sized spiny neurons. Development 140, chromatin regulation by cocaine reveals a role for 301-312 (2013). sirtuins. Neuron 62, 335-348 (2009). 28. P. Arlotta, B. J. Molyneaux, D. Jabaudon, Y. Yoshida, 12. A. J. Robison, E. J. Nestler, Transcriptional and J. D. Macklis, Ctip2 controls the differentiation of epigenetic mechanisms of addiction. Nat Rev medium spiny neurons and the establishment of the Neurosci 12, 623-637 (2011). cellular architecture of the striatum. J Neurosci 28, 13. A. Stipanovich et al., A phosphatase cascade by 622-632 (2008). which rewarding stimuli control nucleosomal 29. D. B. Burkhardt et al., Enhancing experimental response. Nature 453, 879-884 (2008). signals in single-cell RNA-sequencing data using 14. V. Pascoli, J. Terrier, A. Hiver, C. Luscher, Sufficiency graph signal processing. bioRxiv, 532846 (2019). of Mesolimbic Dopamine Neuron Stimulation for 30. K. M. Tyssowski et al., Different Neuronal Activity the Progression to Addiction. Neuron 88, 1054-1066 Patterns Induce Different Gene Expression (2015). Programs. Neuron 98, 530-546 e511 (2018). 15. V. Pascoli, M. Turiault, C. Luscher, Reversal of 31. D. C. Castro, M. R. Bruchas, A Motivational and cocaine-evoked synaptic potentiation resets drug- Neuropeptidergic Hub: Anatomical and Functional induced adaptive behaviour. Nature 481, 71-75 Diversity within the Nucleus Accumbens Shell. (2011). Neuron 102, 529-552 (2019). 16. D. W. Self, W. J. Barnhart, D. A. Lehman, E. J. Nestler, 32. H. L. Hsiang et al., Manipulating a “cocaine engram” Opposite modulation of cocaine-seeking behavior in mice. J Neurosci 34, 14115-14127 (2014). by D1- and D2-like dopamine receptor agonists. 33. S. Trouche et al., Recoding a cocaine-place memory Science 271, 1586-1589 (1996). engram to a neutral engram in the hippocampus. 17. A. F. MacAskill, J. M. Cassel, A. G. Carter, Cocaine Nat Neurosci 19, 564-567 (2016). exposure reorganizes cell type- and input-specific 34. Y. Dong et al., CREB modulates excitability of connectivity in the nucleus accumbens. Nat Neurosci nucleus accumbens neurons. Nat Neurosci 9, 475- 17, 1198-1207 (2014). 477 (2006). 18. D. Guez-Barber et al., FACS identifies unique 35. K. E. Savell et al., A Neuron-Optimized CRISPR/ cocaine-induced gene regulation in selectively dCas9 Activation System for Robust and Specific activated adult striatal neurons. J Neurosci 31, 4251- Gene Regulation. eNeuro 6, (2019). 4259 (2011). 36. K. E. Savell et al., Extra-coding RNAs regulate 19. A. Dobi, G. K. Seabold, C. H. Christensen, R. Bock, neuronal DNA methylation dynamics. Nat Commun V. A. Alvarez, Cocaine-induced plasticity in the 7, 12091 (2016). nucleus accumbens is cell specific and develops 37. W. A. Carlezon, Jr. et al., Regulation of cocaine without prolonged withdrawal. J Neurosci 31, 1895- reward by CREB. Science 282, 2272-2275 (1998). 1904 (2011). 38. A. A. Levine et al., CREB-binding protein controls 20. A. Matsui, V. A. Alvarez, Cocaine Inhibition of response to cocaine by acetylating histones at the Synaptic Transmission in the Ventral Pallidum Is fosB promoter in the mouse striatum. Proc Natl Pathway-Specific and Mediated by Serotonin. Cell Acad Sci U S A 102, 19186-19191 (2005). Rep 23, 3852-3863 (2018). 39. N. Hiroi et al., FosB mutant mice: loss of chronic 21. M. Y. Li, Q. S. Yan, L. L. Coffey, M. E. Reith, cocaine induction of Fos-related proteins and Extracellular dopamine, norepinephrine, and heightened sensitivity to cocaine’s psychomotor serotonin in the nucleus accumbens of freely moving and rewarding effects.Proc Natl Acad Sci U S A 94, rats during intracerebral dialysis with cocaine and 10397-10402 (1997).

13 bioRxiv preprint doi: https://doi.org/10.1101/781872; this version posted September 25, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. 40. W. Renthal et al., Delta FosB mediates epigenetic 56. D. van Dijk et al., Recovering Gene Interactions desensitization of the c-fos gene after chronic from Single-Cell Data Using Data Diffusion. Cell amphetamine exposure. J Neurosci 28, 7344-7349 174, 716-729 e727 (2018). (2008). 57. J. Koster, S. Rahmann, Snakemake-a scalable 41. S. D. Pope, R. Medzhitov, Emerging Principles of bioinformatics workflow engine.Bioinformatics 34, Gene Expression Programs and Their Regulation. 3600 (2018). Mol Cell 71, 389-397 (2018). 58. A. Dobin et al., STAR: ultrafast universal RNA-seq 42. K. E. Savell, J. J. Day, Applications of CRISPR/Cas9 aligner. Bioinformatics 29, 15-21 (2013). in the Mammalian Central Nervous System. Yale J 59. Y. Liao, G. K. Smyth, W. Shi, featureCounts: an Biol Med 90, 567-581 (2017). efficient general purpose program for assigning 43. Y. Zheng et al., CRISPR interference-based specific sequence reads to genomic features. Bioinformatics and efficient gene inactivation in the brain. Nat 30, 923-930 (2014). Neurosci 21, 447-454 (2018). 60. M. I. Love, W. Huber, S. Anders, Moderated 44. H. Zhou et al., In vivo simultaneous transcriptional estimation of fold change and dispersion for RNA- activation of multiple genes in the brain using seq data with DESeq2. Genome Biol 15, 550 (2014). CRISPR-dCas9-activator transgenic mice. Nat 61. G. Paxinos, C. Watson, The rat brain in stereotaxic Neurosci 21, 440-446 (2018). coordinates. (El Sevier, New York, ed. Fifth, 2009). 45. A. Chavez et al., Highly efficient Cas9-mediated transcriptional programming. Nat Methods 12, 326- 328 (2015). 46. N. A. Addy, D. P. Daberkow, J. N. Ford, P. A. Garris, R. M. Wightman, Sensitization of rapid dopamine signaling in the nucleus accumbens core and shell after repeated cocaine in rats. J Neurophysiol 104, 922-931 (2010). 47. P. W. Kalivas, P. Duffy, Time course of extracellular dopamine and behavioral sensitization to cocaine. I. Dopamine axon terminals. J Neurosci 13, 266-275 (1993). 48. C. Cadoni, M. Solinas, G. Di Chiara, Psychostimulant sensitization: differential changes in accumbal shell and core dopamine. Eur J Pharmacol 388, 69-76 (2000). 49. Y. Li, M. J. Acerbo, T. E. Robinson, The induction of behavioural sensitization is associated with cocaine- induced structural plasticity in the core (but not shell) of the nucleus accumbens. Eur J Neurosci 20, 1647-1654 (2004). 50. J. M. Gulley, B. R. Hoover, G. A. Larson, N. R. Zahniser, Individual differences in cocaine-induced locomotor activity in rats: behavioral characteristics, cocaine pharmacokinetics, and the dopamine transporter. Neuropsychopharmacology 28, 2089- 2101 (2003). 51. J. Sabeti, G. A. Gerhardt, N. R. Zahniser, Individual differences in cocaine-induced locomotor sensitization in low and high cocaine locomotor- responding rats are associated with differential inhibition of dopamine clearance in nucleus accumbens. J Pharmacol Exp Ther 305, 180-190 (2003). 52. D. M. Walker et al., Cocaine Self-administration Alters Transcriptome-wide Responses in the Brain’s Reward Circuitry. Biol Psychiatry 84, 867-880 (2018). 53. A. Butler, P. Hoffman, P. Smibert, E. Papalexi, R. Satija, Integrating single-cell transcriptomic data across different conditions, technologies, and species. Nat Biotechnol 36, 411-420 (2018). 54. R. Satija, J. A. Farrell, D. Gennert, A. F. Schier, A. Regev, Spatial reconstruction of single-cell gene expression data. Nat Biotechnol 33, 495-502 (2015). 55. J. Wang, S. Vasaikar, Z. Shi, M. Greer, B. Zhang, WebGestalt 2017: a more comprehensive, powerful, flexible and interactive gene set enrichment analysis toolkit. Nucleic Acids Res 45, W130-W137 (2017). 14 bioRxiv preprint doi: https://doi.org/10.1101/781872; this version posted September 25, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. a Saline Cocaine Y Y

X X b c 50 1.5 **** 40 1.0 30 raveled (m ) raveled (m ) T T 20 0.5 10 Distance Distance 0 0 0 10 20 30 Saline Time (min) Cocaine d 50 ns e *** *** 1.5 40

30 1.0 raveled (m ) T

20 raveled (m ) T 0.5 10 Distance

0 Distance 0 0 10 20 30

Saline MSaline F Time (min) Cocaine MCocaine F Supplementary Figure 1. Cocaine experience signifi- cantly increases locomotor activity in both male and female rats. a, Representative activity trace of saline (n = 8) and cocaine (n = 8) animals over the 30 minute test session. b, Mean total distance traveled is significantly greater in rats who received a single i.p. injection of cocaine (20 mg/kg, n=8) compared to saline controls (n = 2 8; unpaired t test, t(14) = 8.45, p < 0.0001, R = 0.84). c, Locomotor activity is increased in the cocaine group (20 mg/kg, n = 8) across all timepoints, compared to saline controls (n = 8). d, While there was a siginificant effect of cocaine on locomotor activity (20 mg/kg; one-way ANOVA 2 F(3,12) = 20.55, p < 0.0001, R = 0.84), Tukey’s post hoc analysis revealed no sex differences in the mean distance traveled in response to saline (n = 4/sex; p = 0.9988) or acute cocaine (n = 4/sex; p = 0.9940). e, Within each treatment group, there are no sex differences in locomotor activity at any timepoint (n = 4/sex/treatment group). All data are expressed as mean �±��� s.e.m. ***p < 0.001, ****p < 0.0001. M = male (light grey, blue), F = female (charcoal, pink).

15 bioRxiv preprint doi: https://doi.org/10.1101/781872; this version posted September 25, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. a b Drd1−MSN Average Drd2−MSN−1 expression Drd2−MSN−2 Drd3−MSN 2 Grm8−MSN 1 GABA undefined 0 Pvalb−Interneuron −1 Sst−Interneuron Glutamatergic Percent Identity Olig−1 expressed Olig−2 0 Olig−3 25 Astrocyte 50 Polydendrocyte 75 Microglia 100 Mural 0 2000 4000 6000 Kit Sst Mbp Syt1 PdynEbf1Drd2PenkDrd3Grm8 Gja1 Rgs5 Gad1 Genes expressed Elavl2 Hapln2 Pdgfra Foxp2Bcl11b Slc17a7 Ppp1r1b Arhgap15 Features c Drd1 expression d Drd2 expression

0.8 0.6 3 0.4 2 0.2 1 UMAP_2 UMAP_2 UMAP_1 0.0 UMAP_1 0 Expression Expression e f 0.25 UMAP by treatment group

Drd1-MSN 0.20 Olig-1

0.15 Drd2-MSN-1

Cocaine 0.10 Olig-2 Astrocyte Grm8-MSN

0.05 Microglia Polydendrocyte GABA undefined

Drd3-MSN Pvalb-Interneuron slope = 0.99, R2 = 0.87 Saline 0.00 UMAP_2 Cocaine 0.00 0.05 0.10 0.15 0.20 0.25 UMAP_1 Saline

Supplementary Figure 2. Identification of cell types within the rat nucleus accumbens. a, Violin plots indicating the distribution of number of total genes expressed by cell class. b, Dot plot indicating the average expression and percent of cells expressing marker genes of each identified celltype. c-d, Enrichment of Drd1 and Drd2 transcripts within identified clusters. e, Fraction of cocaine and saline treated neurons within each cluster. f, UMAP showing distribution of cells from saline and cocaine-treated rats.

16 bioRxiv preprint doi: https://doi.org/10.1101/781872; this version posted September 25, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. a b c Fraction of cells by sex Male Female UMAP by sex 0.25 Olig-1 Drd1-MSN Drd2-MSN-1 Olig-1 0.20 Drd1-MSN Olig-2 Astrocyte Grm8-MSN 0.15 Drd2-MSN-1 GABA undefined Polydendrocyte Microglia

Female Drd3-MSN 0.10 Olig-2 Cluster ID Pvalb-Interneuron Grm8-MSN Astrocyte Sst-Interneuron Microglia GABA undefined Drd2-MSN-2 0.05 Polydendrocyte Mural Male Olig-3 Pvalb-Interneuron 2

UMAP_2 slope = 1.12, R = 0.95 Female Drd3-MSN Glutamatergic 0.00 UMAP_1 0.00 0.05 0.10 0.15 0.20 0.25 0 50 100 Male Percentage of cells

Supplementary Figure 3. Sex differences in NAc cell distribution. a, Global UMAP showing male and female cells following Seurat integration. b, Linear regression of male and female cell fractions reveals highly similar cell distribution in nearly all cell clusters, with the exception of Drd3-MSNs. c, Percentage of each cluster accounted for by male and female cells. Over 90% of Drd3-MSNs were obtained from male animals.

17 bioRxiv preprint doi: https://doi.org/10.1101/781872; this version posted September 25, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

18 bioRxiv preprint doi: https://doi.org/10.1101/781872; this version posted September 25, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. a b Female EES Male EES

EES EES score score 1.0 0.5 0.5 0.0 0.0 −0.5 −0.5

UMAP_2 UMAP_1 UMAP_2 UMAP_1 c Drd1-MSN Drd2-MSN-1 Astrocytes r = 0.9222 r = 0.4483 r = -0.2219 1.0 p = 1.0235e-78 1.0 p = 5.0534e-07 1.0 p = 2.9661e-06

Homer1

Egr4 Nr4a1 Nr4a1 Egr4 0.5 0.5 Fosb Junb 0.5 Fos Junb Egr2 Gadd45g Tiparp Fos Tiparp Cnr1 Fosb Gadd45g Btg2 Hs6st3 Btg2 Egr2 0.0 0.0 0.0 (r value)

-0.5 -0.5 -0.5 Male EES Correlation

2 13 86 19 3 97 -1.0 66 -1.0 299 -1.0 124 62 314 -1.0 -0.5 0.0 0.5 1.0 -1.0 -0.5 0.0 0.5 1.0 -1.0 -0.5 0.0 0.5 1.0 Female EES Correlation Female EES Correlation Female EES Correlation (r value) (r value) (r value)

Supplementary Figure 5. Contribution of sex to cocaine-mediated transcriptional responses in specific cell types in adult rat NAc. a-b, UMAPs showing enhanced experimental signal (EES) generated independently for each sex. b, EES-gene correlation correspondence plots reveal highly similar cocaine-induced transcriptional responses in Drd1- and Drd2-MSN clusters. In contrast, cocaine-mediated transcriptional responses in astrocytes were divergent in males and females.

19 bioRxiv preprint doi: https://doi.org/10.1101/781872; this version posted September 25, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. a b c Bcl11b Gad1 4 4 Drd1−MSN 3 3 Average 2 Expression 2 Drd2−MSN 1 1 1 Expression Level 0 Expression Level −1 0 0 GABAergic Percent Olig1 Pdgfra Identity 3 Expressed 4 25 Grm8−MSN 50 2 3 75 2 1 Polydendrocyte 1 Expression Level Expression Level 0 0

0 Isl1 Syt1 Ebf1 Drd2 Olig1 Gad1 Grm8 Elavl2 1000 2000 3000 4000 5000 Foxp1 Foxp2 Pdgfra Bcl11b

Adora2a Drd1−MSNDrd2−MSNGABAergicGrm8−MSN Drd1−MSNDrd2−MSNGABAergicGrm8−MSN Polydendrocyte Polydendrocyte Genes Expressed Features Identity Identity d e UMAP by treatment group Dopamine SKF-38393 KCl UMAP_2

UMAP_1 0.30 0.30 0.30 2 R2=0.9550 R2=0.9941 R =0.6124

0.25 0.25 0.25 0.20

0.20 0.20 0.15 cells per cluster cells per cluster cells per cluster 0.10

Drd1-MSN Fraction of KCl treated Fraction of SKF treated 0.15 0.15 Drd2-MSN Fraction of dopamine treated GABAergic 0.05 Grm8-MSN Veh Polydendrocyte DA 0.10 0.00 SKF-38393 0.10 KCl 0.10 0.15 0.20 0.25 0.30 0.10 0.15 0.20 0.25 0.30 0.10 0.15 0.20 0.25 0.30 Fraction of Vehicle treated cells per cluster Supplementary Figure 6. Identification of cell types within rat embryonic striatal neurons. a, Violin plots indicating the distribution of the number of genes expressed in each cluster b, Dot plot indicating the average expression and percent of cells expressing marker genes of each identified celltype. c,Violin plots indicating the distribution of neuronal (Bcl11b,Gad1) and polydendrocyte (Olig1, Pdgfra) marker genes. d, Distribution of dopamine, KCl, or SKF treated cells e, Fraction of dopamine, KCl, or SKF treated cells treated neurons within each cluster.

20 bioRxiv preprint doi: https://doi.org/10.1101/781872; this version posted September 25, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

21 bioRxiv preprint doi: https://doi.org/10.1101/781872; this version posted September 25, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

a b Single gene CRISPRa Dopaplex v1.0 validation Tent5b Tent5b-1 Tent5b-2 * Junb Junb-1 Junb-2 * Prox1 Prox1-1 Nr4a2 Prox1-2 Prox1-3 * Nptx1 Nr4a2-1 * Gsx1 Nr4a2-2 Gadd45g Nptx1-1 * Nptx1-2 Btg2 Gsx-1 Gsx-2 * Egr4 Gadd45g-1 * Sstr2 Gadd45g-2 Gadd45b Klf4-1 Klf4-2 Egr3 Klf4-3 Klf4-4 Egr2 Klf4-5 Btg2-1 * Fosl2 Btg2-2 Fos Tiparp-1 Tiparp-2 Nr4a1 Tiparp-3 Fosb Arc-1 Arc-2 -2 0 2 4 6 8 Arc-3 Arc-4 Log2 (fold change) CRISPRa target Egr4-1 vs. LacZ Egr4-2 * Egr4-3 Sstr2-1 * Sstr2-2 Gadd45b-1 * Gadd45b-2 Egr3-1 * Egr3-2 Egr2-1 Egr2-2 * Fosl2-1 * Fosl2-2 Fos-1 * Fos-2 Fos-3 Sik1-1 Sik1-2 Sik1-3 Nr4a1-1 * Nr4a1-2 Fosb-1 Fosb-2 * -2 0 2 4 6 8 10

Log2 (Fold change) vs.LacZ

Supplementary Figure 8. CRISPRa validation at single gene targets. a, Single guide RNAs targeted with dCas9-VPR to DA-induced genes results in validated induction in 16 out of 23 genes, as compared to a LacZ control sgRNA (n = 2 per group). Asterisks indicate sgRNAs selected for multiplexed targeting. (b) Lentiviral vectors expressing sgRNAs validated individually to upregulate target genes were pooled to target 17 DA-induced genes simultaneously. Simultaneous targeting resulted in induction of all but one gene (n = 5 per group). All data are expressed as mean ± s.e.m.

22 bioRxiv preprint doi: https://doi.org/10.1101/781872; this version posted September 25, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

a CRISPRa target: LacZ b CRISPRa target: Dopaplex

LV LV LV LV CPu CPu CPu CPu

NAcCAcbC NAcCAcbC NAcC NAcC aca NAcSh NAcSh aca aca NAcSh NAcSh aca

Supplementary Figure 9. Viral placement validation for animals infused with CRISPRa constructs. CRISPRa lentivirus targeted to either the bacterial LacZ (a) or Dopaplex (b) were infused into the nucleus accumbens core (NAcC). Schematics of target regions are adapted from Paxinos and Watson. NAcC: nucleus accumbens core; NAcSh: nucleus accumbens shell; CPu: caudate putamen; LV: lateral ventricle.

23