Gene-targeted, CREB-mediated induction of ΔFosB controls distinct downstream transcriptional patterns within D1 and D2 medium spiny neurons

Abstract

Background:

The onset and persistence of addiction phenotypes are, in part, mediated by transcriptional mechanisms in the brain that affect gene expression and subsequently neural circuitry. ΔFosB is a transcription factor that accumulates in the nucleus accumbens (NAc) – a brain region responsible for coordinating reward and motivation – after exposure to virtually every known rewarding substance, including cocaine and opioids. ΔFosB has also been shown to directly control gene transcription and behavior downstream of both cocaine and opioid exposure, but with potentially different roles in D1 and D2 medium spiny neurons (MSNs) in NAc.

Methods:

To clarify MSN subtype-specific roles for ΔFosB, and investigate how these coordinate the actions of distinct classes of addictive drugs in NAc, we developed a CRISPR/Cas9-based epigenome editing tool to induce endogenous ΔFosB expression in vivo in the absence of drug exposure. After inducing ΔFosB in D1 or D2 MSNs, or both, we performed RNA-sequencing on bulk male and female NAc tissue (N = 6–8/group).

Results:

We find that ΔFosB induction elicits distinct transcriptional profiles in NAc by MSN subtype and by sex, establishing for the first time that ΔFosB mediates different transcriptional effects in males vs females. We also demonstrate that changes in D1 MSNs, but not in D2 MSNs or both, significantly recapitulate changes in gene expression induced by cocaine self-administration.

Conclusions:

Together, these findings demonstrate the efficacy of a novel molecular tool for studying cell-type-specific transcriptional mechanisms, and shed new light on the activity of ΔFosB, a critical transcriptional regulator of drug addiction.

Introduction

The onset and persistence of drug addiction is thought to be caused, in part, by cell-type-specific regulation of gene transcription in the brain’s reward circuitry. Several transcription factors are implicated in mediating the effects of drugs of abuse on gene expression (1,2), however, how these factors act in specific cell types remains mostly unknown.

ΔFosB is among the best-characterized transcription factors in the context of addiction. A product of the Fosb gene which also encodes full-length FosB, ΔFosB is uniquely stable compared to FosB and other Fos family transcription factors, and accumulates in response to chronic exposure to all classes of drugs of abuse (3,4). This phenomenon is best established within the nucleus accumbens (NAc), part of the ventral striatum that plays a key role in reward and addiction.

ΔFosB induction in NAc shows striking cell-type-specific patterns. All drugs of abuse induce ΔFosB selectively in the class of NAc medium spiny projection neurons (MSNs) that expresses predominantly D1 dopamine receptors, with the sole exception of opioids, which induce ΔFosB equally in D1 and D2 MSNs (5). Of note, natural rewards (e.g., sucrose drinking, enriched environment) also induce ΔFosB in both cell types, whereas chronic administration of antipsychotic drugs (e.g., haloperidol) induces ΔFosB selectively in D2 MSNs (5). These cell-type-specific patterns of ΔFosB expression are important given the very different, and in some cases, opposite effects of D1 vs D2 NAc MSNs on a range of behavioral endpoints (6–10).

ΔFosB expression in D1 MSNs vs D2 MSNs exerts distinct effects on cocaine-related behavior. Its overexpression selectively in D1 MSNs, achieved either by inducible bitransgenic mice (11,12) or viral-mediated gene transfer (13), promotes behavioral responses to cocaine, effects not seen upon its overexpression in D2 MSNs. ΔFosB overexpression also differentially impacts synaptic plasticity at glutamatergic synapses in D1 vs D2 NAc MSNs (13). Much less is known, however, about the cell-type-specific transcriptional effects of ΔFosB in this brain region, which could explain its effects on synaptic plasticity and behavior. An older microarray study found that ΔFosB overexpression in D1 MSNs of bitransgenic mice mimics the effects of chronic investigator-administered cocaine (14), but to date there has not been an examination of ΔFosB’s transcriptional effects with more advanced RNA-sequencing (RNA-seq) approaches.

Moreover, these earlier studies relied on overexpression systems, which induce supraphysiological levels of ΔFosB expression and could therefore be confounded by non-physiological effects of such high levels of the protein. To overcome this challenge, we turned in the present study to the newly developed tools of locus-specific epigenome editing, which make it possible to guide a protein of interest to a single locus within the genome (15,16). Several transcription factors have been shown to mediate activation of Fosb transcription and ΔFosB induction in NAc by cocaine; these include CREB, E2F3a, SRF, and EGR3 (17–19). Our approach here was to take the best-established of these factors, CREB, and target it to Fosb by use of CRISPR/Cas9 epigenome editing tools. We demonstrate that fusion of a constitutively active isoform of CREB to dead Cas9, dCas9-CREB^S133D, can be used both in vitro and in vivo in mouse brain to activate ΔFosB expression in the absence of drug exposure. We then use this epigenome editing tool in concert with Cre-driver lines of mice to induce endogenous ΔFosB selectively in D1 MSNs or in D2 MSNs, or in both cell types, and to determine the downstream transcriptional effects by RNA-seq. We show that ΔFosB induction in D1 MSNs is uniquely associated with the transcriptional regulation seen in NAc in response to cocaine self-administration.

Methods and Materials

See supplemental information for detailed methods.

Neuro2a cells transfection and RNA isolation

Neuro2a cells were cultured and transfected with 100 ng of gRNAs and dCas9 fusion constructs. After 48 hr, RNA was isolated using the RNeasy Mini Kit (Qiagen) according to manufacturer instructions.

Animals

For bulk NAc tissue experiments, C57BL/6J male mice were acquired from The Jackson Laboratory. For NAc MSN-specific experiments, D1-Cre and D2-Cre bacterial artificial chromosome (BAC) transgenic mice (http://www.gensat.org/cre.jsp) were bred in-house. All animal procedures were performed in accordance with guidelines of the Institutional Animal Care and Use Committee at the Icahn School of Medicine at Mount Sinai.

Transcription factor engineering and viral reagents

The CRISPR/Cas9 system designed to target the Fosb promoter was synthesized as described previously (20). Candidate gRNA sequences targeting Fosb from the transcription start site (TSS) to −1250 bp were determined in silica using Deskgen.

All candidate gRNAs were de novo synthesized and contained a U6 promoter, variable target sequence, gRNA scaffold, and a termination signal. These were then subcloned into bicistronic p1005 variant plasmids for packaging into Herpes simplex virus (HSV) vectors. One HSV vector was made to express dCas9-CREB^S133D and mCherry, and another was made to express a gRNA and GFP. An HSV vector was also made to express dCas9-CREB^S133A. dCas9-^CREBS133D was cloned into both Cre-dependent and Cre-independent forms of our HSV vector. HSVs were delivered in NAc by stereotactic surgery.

RNAscope® in situ hybridization

Coronal brain sections (40 μm) were fixed in 4% PFA and prepared for in situ hybridization using RNAscope® to hybridize the following probes to RNA transcripts and fluorophores: Mm-Drd1 (cat #461901), Mm-Drd2-C3 (cat #406501-C3), and Mm-Fosb-O1-C2 (cat #584751-C2).

RNA isolation, RT-qPCR, RNA-seq library preparation and sequencing

NAc was dissected rapidly and frozen on dry ice. RNA extraction, library preparation, and RNA-seq were performed according to previously described methods (21,22) using a TruSeq Stranded Total RNA – Gold (Illumina). Libraries were sequenced by GENEWIZ using an Illumina HiSeq System with 150 bp paired-end reads.

For qPCR validation of CRISPR construct efficacy, RNA was converted to cDNA with iScript (Bio-Rad). qPCR samples were analyzed in triplicate using the standard ΔΔCT method.

Statistics and bioinformatics

All bioinformatic analyses were accomplished using published approaches, with modifications (20–24). Transcriptomic analyses were carried out primarily with DESeq2 in R (version 1.27) for differential expression analysis and rank-rank hypergeometric overlap (RRHO) for threshold-free comparisons between datasets (24,25). Differential expression was calculated as fold-change in transcript expression following delivery of gRNA-Fosb over that of gRNA-NT.

All statistical analyses were performed using Prism software (GraphPad). Student’s t tests were used for any pairwise comparisons and two-way ANOVAs were used for all multiple comparisons followed by post hoc tests where appropriate. A Type I error significance level of 0.05 was used to determine statistical significance for all calculations.

Results

In vitro induction of Fosb and ΔFosb expression using dCas9-VP64

To determine whether ΔFosB expression can be induced using targeted CRISPR/Cas9 epigenome editing, we first assessed activating Fosb transcription in vitro in Neuro2a cells. We began by using a dCas9 fused to VP64, a chimera protein used for general transcriptional activation. Nine candidate gRNAs were selected from a list of in silico target sites on the Fosb promoter from the TSS to −1250 bp (Fig. 1A). We assessed the efficacy of transcriptional induction by targeting each site with dCas9 alone, or dCas9-VP64, and measured both Fosb and ΔFosb mRNA expression. For every site targeted, resulting expression was compared to that elicited by a control gRNA, which has a scrambled, non-targeted sequence (gRNA-NT) and used here as a non-functional control as in earlier work (20). Targeting each gRNA site with dCas9 alone was minimally activating, but targeting Fosb with 5 of the 9 candidate gRNAs (4, 5, 9, 11, or 15) paired with dCas9-VP64 significantly increased both Fosb and ΔFosb expression (Fig. 1B,C).

Figure 1. Induction of Fosb and ΔFosb in vitro after targeting with dCas9, dCas9-VP64, or dCas9-CREB^S133D.

(A) A schematic of the Fosb gene locus depicting sites of 9 selected gRNAs for in vitro testing. The endogenous CREB binding site (CRE motif) is depicted approximately −500 bp upstream of TSS. (B) Fosb mRNA expression was increased in Neuro2a (N2a) cells after targeting with dCas9-VP64 paired with several candidate gRNAs, but not in response to dCas9 alone (main effects of construct, F_1,37 = 23.37, p<0.0001, and gRNA, F_9,37 = 2.497, p=0.0242 by two-way ANOVA; gRNA-NT vs gRNA-4: t₃₇ = 3.605, **p=0.0082; gRNA-NT vs gRNA-5: t₃₇ = 4.988, ***p=0.0001; gRNA-NT vs gRNA-9: t₃₇ = 3.12, *p=0.032; gRNA-NT vs gRNA-11: t₃₇ = 5.079, ****p<0.0001; gRNA-NT vs gRNA-15: t₃₇ = 4.52, ***p=0.0006 by post-hoc analysis). (C) ΔFosb mRNA expression was increased in N2a cells after targeting with dCas9-VP64 paired with several candidate gRNAs, but not in response to dCas9 alone (main effect of construct, F_1,37 = 32.65, p<0.0001, by two-way ANOVA; gRNA-NT vs gRNA-4: t₃₇ = 3.946, **p=0.0031; gRNA-NT vs gRNA-5: t₃₇ = 5.295, ****p<0.0001; gRNA-NT vs gRNA-9: t₃₇ = 3.245, *p=0.0224; gRNA-NT vs gRNA-11: t₃₇ = 4.122, **p<0.0018; gRNA-NT vs gRNA-15: t₃₇ = 3.954, **p=0.003 by Bonferroni post-hoc analysis).

In vivo induction of Fosb and ΔFosb expression using dCas9-CREB^S133D

Based on Fosb induction using dCas9-VP64 in vitro, gRNA-11 (gRNA-Fosb) was selected—based on the consistency of its effects (Fig. 1B,C)—for targeting Fosb in vivo using a more specific tool to activate transcription. VP64 is not expressed in mammalian brain and therefore is not part of the organismal drug response. VP64 is also a viral promoter that induces gene transcription outside of normal physiological ranges. Thus, we designed a dCas9 effector partner that would mimic the active form of an endogenous regulator of the Fosb locus: dCas9-CREB^S133D. CREB^S133D is a constitutively active form of CREB that mimics its active, phosphorylated isoform by expressing an aspartate residue at S133. To assess the efficacy of transcriptional induction using this tool, gRNA-11 was paired with dCas9-CREB^S133D in vitro and in vivo, as well as being paired with dCas9 alone or with an inactive dCas9-CREB fusion, CREB^S133A. CREB^S133A is a dominant negative form of CREB that cannot be activated through phosphorylation at S133.

In vitro, targeting Fosb with dCas9 alone or dCas9-CREB^S133A was not effective at inducing transcription of either Fosb or ΔFosb; only dCas9-CREB^S133D paired with Fosb-targeting gRNA-11 effectively induced expression of both transcripts (Fig. 2A,B). Notably, the 2–3-fold induction seen with dCas9-CREB^S133D is of far lesser magnitude than the 10–30-fold induction seen with dCas9-VP64 (see Fig. 1).

Figure 2. Induction of Fosb and ΔFosb in NAc after targeting with dCas9-CREB^S133D.

(A-B) Fosb in N2a cells is only effectively induced by targeting with dCas9-CREB^S133D paired with gRNA-Fosb, but not when paired with gRNA-NT, dCas9 alone with either gRNA, or dCas9-CREB^S133A with either gRNA. (A) Fosb expression: F_{2, 17} = 5.488, p=0.0145; dCas9 + gRNA-Fosb vs dCas9-CREB^S133A + gRNA-Fosb: p=0.9815, dCas9 + gRNA-Fosb vs dCas9-CREB^S133D + gRNA-Fosb: p=*0.0259, dCas9-CREB^S133A + gRNA-Fosb vs dCas9-CREB^S133D + gRNA-Fosb: *p=0.0385 by 1-way ANOVA and Tukey’s HSD. (B) ΔFosb expression: F_{2, 13} = 7.702, p=0.0062; dCas9 + gRNA-Fosb vs dCas9-CREB^S133A + gRNA-Fosb: p=0.9565, dCas9 + gRNA-Fosb vs dCas9-CREB^S133D + gRNA-Fosb: **p=0.0080, dCas9-CREB^S133A + g gRNA-Fosb vs dCas9-CREB^S133D + gRNA-Fosb: *p=0.0129 by 1-way ANOVA and Tukey’s HSD. (C-D) Expression of Fosb isoforms in mouse NA is induced by dCas9- CREB^S133D paired with gRNA-Fosb over gRNA-NT and other dCas9 isoforms. (C) Fosb expression: F_{4, 57} = 4.995, **p=0.0016; dCas9-CREB^S133D + gRNA-Fosb vs GFP: **p=0.0060, dCas9-CREB^S133D + gRNA-Fosb vs dCas9^S133D + gRNA-NT: *p=0.0156, dCas9-CREB^S133D + gRNA-Fosb vs dCas9 + gRNA-Fosb: *p=0.0137, dCas9-CREB^S133D + gRNA-Fosb vs dCas9-CREB^S133A + gRNA-Fosb: p=0.0853 by 1-way ANOVA and post hoc analysis using Dunnett’s multiple comparisons test. (D) ΔFosb expression: F_{4, 48} = 5.497, p=0.0010; dCas9-CREB^S133D + gRNA-Fosb vs GFP: *p=0.0108, dCas9-CREB^S133D + gRNA-Fosb vs dCas9^S133D + gRNA-NT: **p=0.0049, dCas9-CREB^S133D + gRNA-Fosb vs dCas9 + gRNA-Fosb: **p=0.0082, dCas9-CREB^S133D + gRNA-Fosb vs dCas9-CREB^S133A + gRNA-Fosb: **p=0.0024 by 1-way ANOVA and post hoc analysis using Dunnett’s multiple comparisons test. (E) ΔFosB but not FosB protein was increased in NAc when dCas9-CREB^S133D was targeted to Fosb with gRNA-Fosb, relative to gRNA-NT (t₁₈ = 2.417, *p=0.0265). (F) In situ hybridization for Drd1, Drd2, and Fosb in the NAc of a D1-Cre mouse shows an increase in D1 MSN-specific ΔFosb expression after infection with gRNA-Fosb vs infection with gRNA-NT. Arrows denote Drd1+/Fosb+ cells.

We next assessed the efficacy of inducing Fosb expression in vivo in mouse NAc. Each of the various complexes, dCas9 alone, dCas9-CREB^S133A, or dCas9-CREB^S133D, as well as gRNA-NT or gRNA-Fosb, were packaged into HSVs and delivered bilaterally into NAc. Activation of Fosb and ΔFosb mRNA expression was compared to that elicited by infection of NAc tissue with HSV-GFP. Both Fosb and ΔFosb mRNA expression were significantly increased only after targeting the active form of CREB to Fosb (Fig. 2C,D). Additionally, ΔFosB protein was significantly induced in NAc when dCas9-CREB^S133D was targeted to Fosb with gRNA-Fosb, relative to gRNA-NT (Fig. 2E), demonstrating the viability of using a natural transcriptional effector protein in vivo to induce Fosb transcription and subsequent ΔFosB protein expression. Notably, full-length FosB protein is not appreciably expressed in NAc after either viral condition (Fig. 2E), presumably due to the very short half-life of FosB protein vs ΔFosB protein (3,4).

After successfully inducing endogenous ΔFosB expression in vivo in NAc using dCas9-CREB^S133D, we packaged dCas9-CREB^S133D into a Cre-inducible HSV vector and bilaterally infused it into NAc of Cre-expressing BAC transgenic mouse lines in order to achieve D1 or D2 MSN-specific induction of ΔFosB. RNAscope images confirm this cell-specific expression (Fig. 2F, Supplementary Fig. 1,2): in the images shown, HSV-dCas9-CREB^133D induces ΔFosb only in D1 MSNs of D1-Cre mice, and only in D2 MSNs of D2-Cre mice, and only in the presence of gRNA-Fosb.

ΔFosB elicits distinct transcriptional patterns in D1 vs D2 NAc MSNs

After inducing ΔFosB expression in NAc D1 and D2 MSNs, we performed RNA-seq on bulk NAc tissue of male and female mice in order to assess the whole NAc transcriptomic effects of ΔFosB induction in one or both cell types. D1 MSN-specific induction was intended to mimic the cocaine-elicited transcriptome with respect to ΔFosB expression, while induction in D2 MSNs or both MSN subtypes would mimic ΔFosB induction patterns after other stimuli (see Introduction).

ΔFosB induction in each cell type of male mice elicited distinct patterns of differential gene expression (Fig. 3A). Transcripts up- or downregulated in male NAc downstream of ΔFosB induction in D1 MSNs showed modest overlap with transcripts up- or downregulated in D2 MSNs. By contrast, when ΔFosB was induced in both MSN subtypes, little overlap was seen with transcripts affected in either cell type alone, suggesting that transcriptional regulation resulting from ΔFosB activity in male D1 MSNs and in D2 MSNs might partly counteract one another.

Figure 3. ΔFosB induction in D1 vs D2 MSNs results in MSN-specific and sex-specific differential gene expression.

Union heatmaps displaying differential gene expression between induction in D1 MSNs alone, D1 and D2 MSNs simultaneously, or D2 MSNs alone in NAc. All differential expression was analyzed between induction in the appropriate MSN subtype with dCas9-CREB^S133D targeted to Fosb with gRNA-Fosb, relative to gRNA-NT. All differential expression reported in heatmaps was ± 30% fold change in expression from gRNA-NT, and p<0.05. (A) Differential expression patterns in male NAc. (B) Differential expression patterns in female NAc. (C-E) Venn diagrams displaying minimal overlap in significantly differentially expressed genes between male and female NAc downstream of D1 MSN-specific induction (C), D1/D2 MSN-specific induction (D), and D2 MSN-specific induction (E). All transcripts represented were ±30% fold change in expression from gRNA-NT, and p<0.05. (F) Heatmap indicating off-target differential expression events elicited by gRNA-NT and gRNA-Fosb. Transcripts that matched with each gRNA sequence from 0 to 3 mismatches in base pairs (top panel) were assessed for significant differential expression. Hits were organized by their proximity to the nearest gene either within 300bp or farther away in the genome (bottom panel). N=6–8/group.

In female NAc, transcriptional patterns downstream of ΔFosB induction in each MSN condition were more distinct from one another than in male NAc. In fact, a subset of transcripts up- or downregulated in female NAc upon ΔFosB induction in D1 MSNs showed opposite regulation compared to ΔFosB induction in D2 MSNs (Fig. 3B). In comparison, ΔFosB induction in both MSN subtypes resulted in a transcriptional pattern mostly distinct from that observed after ΔFosB induction in either female-specific cell population alone.

When the transcriptional profiles downstream of ΔFosB induction in each MSN subtype were directly compared between male and female NAc, very little overlap of regulated genes was observed between males and females for each MSN subtype (Fig. 3C–E). Notably, a greater number of genes were differentially expressed when ΔFosB was manipulated in a single MSN subtype (Fig. 3C,E), while in the D1/D2 group, more genes were differentially expressed in female NAc than in male (Fig. 3D). We did not observe any differences in ΔFosb induction between male and female NAc (Supplementary Fig. 3).

Next, we leveraged this differential expression data to determine the specificity of targeting dCas9-CREB^S133D to Fosb by calculating off-target differential expression. Mismatches to gRNA-Fosb or gRNA-NT from 0 to 3 nucleotides were identified. gRNA-NT only had mismatches of >3 bp genome-wide, with 4 genomic regions showing 3 mismatches. The number of mismatches for gRNA-Fosb are represented in Table 1. The percent of genes differentially expressed in any of the mismatch groups upon HSV-dCas9-CREB^S133D/gRNA-Fosb injection was just 3.67% (Fig. 3F), confirming that gRNA-Fosb is specific for its intended target.

Table 1.

The Number of Mismatches for gRNA-Fosb

# mismatches	# of genomic loci
1	2
2	20
3	339

Sex-specific transcriptional effects of ΔFosB induction in D1 vs D2 NAc MSNs

To further analyze the striking sex-specific and MSN subtype-specific patterns we observed via differential expression analysis, we performed threshold-free RRHO analysis to examine global co-regulation patterns. In doing so, RRHO analysis assesses the transcriptome-wide degree of overlap between two conditions of interest, regardless of the magnitude of change in expression and in the absence of any arbitrary statistical thresholds. In this way, it complements differential expression analysis.

Directly comparing male and female NAc, we observed strong overlap downstream of ΔFosB induction in D1 MSNs, where a subset of transcripts were commonly downregulated (Fig. 4A). Much weaker overlap was evident for ΔFosB induction in D2 MSNs and no prominent overlap was seen for ΔFosB induction in both D1 and D2 MSN cell types (Fig. 4A). Together, the results from differential expression and RRHO analyses both indicate strikingly distinct sex-specific effects downstream of ΔFosB activity in D1 and D2 MSNs.

Figure 4. Rank-rank hypergeometric overlap analysis of ΔFosB induction in D1 vs D2 MSNs in male and female NAc.

RRHO analyses displaying threshold-free overlap in gene expression between male and female NAc downstream of ΔFosB induction in D1 MSNs alone, D1 and D2 MSNs simultaneously, or D2 MSNs alone. (A) Comparisons between male and female NAc within each MSN subtype condition. (B) Comparisons between MSN subtypes within male NAc. (C) Comparisons between MSN subtypes within female NAc. N=6–8/group.

When MSN subtype conditions were compared within the male and female datasets, a strong convergence of regulated genes was observed within male NAc groups (Fig. 4B), but not for female NAc (Fig. 4C). Induction in D1 MSNs vs D1/D2 MSNs, and induction in D1 vs D2 MSNs, in male NAc showed an extremely high degree of overlap in both up- and downregulated transcripts (Fig. 4B). Noticeably, no overlap was observed for the same conditions in females (Fig. 4C). However, moderate overlap was observed between D1/D2 and D2-specific induction in both male and female NAc (Fig. 4B,C). These patterns reinforce that the impact of ΔFosB induction in each MSN subpopulation is dependent upon sex-specific factors.

Molecular pathway analysis was used to identify functional networks of genes influenced by ΔFosB induction in D1 or D2 MSNs or both. Results of these analyses, shown in Supplementary Figures 4 and 5, support conclusions from RRHO comparisons that largely distinct sets of genes are affected in each cell type in a sex-specific manner.

D1 and D2-specific MSN induction of ΔFosB recapitulates transcriptional patterns elicited by cocaine self-administration

To assess the contribution of ΔFosB to the transcriptional patterns elicited by cocaine exposure, we compared the differential gene expression patterns observed downstream of ΔFosB’s action in D1 and/or D2 MSNs with previously published RNA-seq data on bulk NAc tissue from a comprehensive cocaine self-administration study (22). In this study, male mice were first food trained to acquire lever-pressing behavior, and then put through 10–15 days of cocaine or saline self-administration. Some of the mice were euthanized 24 hr after the last self-administration session, while others underwent 30 days of withdrawal (WD) at which time they were placed back into the self-administration chambers and given an intraperitoneal injection of cocaine or saline and euthanized 1 hr later. We compared ΔFosB-elicited transcriptional patterns in male NAc to that seen under these various conditions (Fig. 5). All comparisons of expression were carried out using threshold-free RRHO.

Figure 5. Threshold-free genome-wide expression overlap between ΔFosB-transcriptome and cocaine self-administration transcriptome in NAc.

(A) Experimental design. HSV-dCas9-CREB^S133D and HSV-gRNA (either gRNA-Fosb or gRNA-NT) were bilaterally co-infused into mouse NAc, and tissue was collected 6 days later. MSN-specific expression was achieved by using a Cre-inducible HSV-dCas9-CREB^S133D in D1-Cre or D2-Cre BAC transgenic mice. For induction in D1 and D2 MSNs simultaneously, a D1-Cre mouse was used but HSV-dCas9-CREB^S133D was Cre-independent. (B-E) RRHO comparisons between ΔFosB induction in D1, D1 and D2, and D2 MSNs and acute cocaine exposure (B), 10d cocaine self-administration (C), plus 30 d withdrawal + a saline challenge (D), or 30 d withdrawal + a cocaine challenge (E). N=6–8/group.

Gene expression patterns in bulk NAc tissue downstream of ΔFosB induction in D1 MSNs displayed significant overlap with each of the four cocaine conditions. Acute cocaine (saline self-administration followed by 30 d WD + cocaine injection vs saline injection) revealed a weak but significant positively correlated overlap in expression patterns as compared to ΔFosB induction in D1 MSNs (Fig. 5B). This positive overlap was far stronger for mice after chronic cocaine self-administration examined 24 hr later (Fig. 5C). Surprisingly, this pattern of overlap reversed in directionality after 30 d WD both for animals given a challenge dose of saline or of cocaine (Fig. 5D,E). The degree of this opposing overlap is more significant after administration of a cocaine challenge dose compared to saline.

Significant overlap patterns were not observed between ΔFosB-elicited transcriptomes downstream of induction in D2 MSNs alone or in both D1 and D2 MSNs. The only exception is significant overlap between transcripts regulated as a result of induction in D2 MSNs and those regulated 24 hr after cocaine self-administration, where a weak but significant overlap was apparent (Fig. 5C).

Discussion

Molecular, cellular, and circuit level changes take place to coordinate each stage of drug use and its associated behaviors (1,26). As these mechanisms can converge and diverge between different classes of substances, studies probing transcriptional dynamics offer unique and powerful insight into understanding these similarities and differences. In this study, we assessed how ΔFosB, a transcription factor hypothesized to be a “molecular switch” for the transition from volitional to compulsive drug use (27), differentially coordinates gene expression in the NAc’s two predominant cell types.

In order to characterize its control over gene transcription, we leveraged CRISPR/Cas9-based locus-specific epigenomic editing to selectively induce ΔFosB in D1 or D2 MSNs or both, and used this cell-type-specific induction to assess the transcriptomic consequences of ΔFosB activity in each cell type in the entire NAc. Our results revealed strikingly different patterns by both cell type and sex: ΔFosB induction in each MSN subtype controls distinct transcripts overall and when compared between male vs female NAc. In male NAc, differential expression analysis suggested a potential counteractive effect when ΔFosB is induced in both cell types simultaneously (Fig. 3A), whereas in female NAc, we observed a reversal in directionality: a portion of transcripts downregulated as a result of ΔFosB induction in D1 MSNs were upregulated downstream of activity in D1/D2 and D2 MSNs (Fig. 3B). Direct comparison of the transcripts affected in male and female NAc showed that the transcripts comprising these patterns essentially did not overlap (Fig. 3C–E; 4A–C). This lack of overlap between sexes is consistent with prior male-female RNA-seq comparisons conducted by our group (36,37,38) and, indeed, numerous physiological and behavioral differences in every stage of substance use have been identified between the sexes. Some prominent departures observed in human drug users include greater escalation of both amount of drug and frequency of use in women, differences in the severity of withdrawal symptoms by drug class and by sex, and that men stabilize as compulsive users at lower doses and may have longer periods of abstinence (40). These distinctions are relevant for the study of ΔFosB’s sex-specific roles because the protein is thought to be critical for coordinating the transition from casual to compulsive drug use, and our results support a hypothesis that this coordination may differ between males and females. However, given the breadth and depth of known sex differences in addiction neurobiology, extensive follow-up investigations are necessary in order to parse the possibilities for ΔFosB’s contributions.

In addition to clarifying the sex-specific and D1 and D2 MSN-specific consequences of ΔFosB induction, our transcriptomic dataset clarifies ΔFosB’s transcriptional activity in the context of different drugs. Because ΔFosB accumulates in D1 MSNs in response to repeated cocaine exposure, we predicted that manipulating ΔFosB in this cell type would recapitulate part of a cocaine self-administration transcriptome generated by our laboratory (22). This recapitulation is exactly what we observed in the highly significant overlap between transcription resulting from ΔFosB induction in D1 MSNs and that elicited by 10 d of cocaine self-administration, a result that strongly supports a long-held hypothesis for ΔFosB’s transcriptional role in D1 MSN plasticity following repeated cocaine exposure (1,11,13,14). By comparison, only modest overlap is seen with acute cocaine exposure, a time point when ΔFosB induction is at very low levels. Interestingly, when compared to the transcriptome in NAc as a result of 30 d withdrawal from cocaine, and re-exposure after withdrawal, we observed significant but opposite overlap (Fig. 5D,E), indicating that ΔFosB’s regulation is upstream of a different network of genes in NAc during stages of withdrawal and relapse compared to initial and recurring drug exposures. Indeed, while the effect of ablating or overexpressing ΔFosB during acute and repeated cocaine exposure has been studied previously (12,14,41,42), the effect of manipulating ΔFosB during other stages of drug exposure has not yet been explored. These findings indicate the importance of following up on these observations experimentally.

That ΔFosB could have a unique role in different stages of drug use is consistent with its known functions in several other cell types throughout the brain. These include its role in medial prefrontal cortex pyramidal neurons in the context of stress susceptibility (43), the hippocampus in the context of both stress and cocaine exposure (44,45), and in mediating spatial memory deficits in a mouse model of Alzheimer’s disease (46). It would be interesting to use our gene-editing tools to investigate the likely distinct transcriptional profiles downstream of ΔFosB in each of these additional cell types. Moreover, recent advances in single-cell RNA-seq has revealed considerable heterogeneity in gene expression within MSNs traditionally classified as D1 and D2 subtypes (47,48), which raises the question for future investigation of how ΔFosB influences transcription in these various subtypes of D1 and D2 MSNs.

It is important to note that the present study of ΔFosB’s effects on gene transcription in NAc was carried out in the absence of drug exposure and all attendant behavioral experiences (i.e., cue associations, withdrawal, extinction, relapse, etc.). It would therefore be interesting in future work to understand how the transcriptomic actions of ΔFosB might change in the context of these other factors which would be expected to activate many additional cellular signaling pathways. A consideration in this line of thinking is that only relatively small percentages of neurons in brain reward regions like the NAc appear to be activated in response to drug exposure and associated cues. Such “neuronal ensembles” are thought to encode certain aspects of the drug experience by controlling neural circuit activity in a coordinated manner (49). Studies of cocaine-activated neuronal ensembles suggest that ΔFosB expression correlates with initial expression of c-Fos (50), but the extent to which ΔFosB is involved in encoding the cells of a neuronal ensemble, or the extent to which it could be involved in “remembering” different aspects of a drug experience, are questions for future investigations.

One caveat of our study is that we examined the transcriptomic effects of ΔFosB acting in a single cell type by use of RNA-seq of bulk NAc extracts. We utilized this experimental design based on the knowledge that transcriptional changes that ΔFosB induces in one cell type (e.g., D1 MSNs) would be expected to influence the functioning (and hence transcriptional regulation) of other neuronal and even non-neuronal cell types within this brain region. Nevertheless, an essential next step in this line of investigation would be to perform RNA-seq on sorted populations of specific cell types to understand with greater resolution the cell autonomous effects of ΔFosB as well as its effects on other cell types.

Together, the results presented here demonstrate a novel tool for in vivo locus-specific epigenomic gene editing. We show that dCas9-CREB^S133D can be used to mimic ΔFosB expression in degree and in cell-specificity as if it were induced by drug exposure. We used this tool to assess whether or not ΔFosB expression in D1 MSNs, in D2 MSNs, or in both cell types recapitulates the cocaine-elicited transcriptomes in NAc, and how the consequences of its expression in these cell types compare between males and females. We conclude that ΔFosB expression elicits mostly distinct sex-specific transcriptional patterns downstream of its induction in each MSN subtype, and that its expression alone is capable of recapitulating some of the transcriptional changes observed in response to cocaine self-administration after short withdrawal periods.

KEY RESOURCES TABLE

Resource Type	Specific Reagent or Resource	Source or Reference	Identifiers	Additional Information
Add additional rows as needed for each resource type	Include species and sex when applicable.	Include name of manufacturer, company, repository, individual, or research lab. Include PMID or DOI for references; use “this paper” if new.	Include catalog numbers, stock numbers, database IDs or accession numbers, and/or RRIDs. RRIDs are highly encouraged; search for RRIDs at https://scicrunch.org/resources.	Include any additional information or notes if necessary.
Antibody
Bacterial or Viral Strain
Biological Sample
Cell Line	Neuro2a cells	CCL-131, ATCC	RRID:CVCL_0470
Chemical Compound or Drug
Commercial Assay Or Kit	RNeasy Mini and Micro Kits	Qiagen
Commercial Assay Or Kit	Effectene reagent	Qiagen
Commercial Assay Or Kit	Mm-Drd1	ACD Bio RNAscope assay	cat #461901
Commercial Assay Or Kit	Mm-Drd2-C3	ACD Bio RNAscope assay	cat #406501-C3
Commercial Assay Or Kit	Mm-Fosb-O1-C2	ACD Bio RNAscope assay	cat #584751-C2
Commercial Assay Or Kit	RNA Nano kit	Agilent
Commercial Assay Or Kit	iScript cDNA kit	Biorad
Commercial Assay Or Kit	TruSeq Stranded Total RNA - Gold	Illumina
Deposited Data; Public Database	Accession number pending	NCBI GEO
Genetic Reagent
Organism/Strain	C57BL/6J, male	Jackson Laboratory
Organism/Strain	D1-Cre BAC transgenic mice, male/female	bred in-house at ISMMS		www.gensat.org
Organism/Strain	D2-Cre BAC transgenic mice, male/female	bred in-house at ISMMS		www.gensat.org
Recombinant DNA	pcDNA-dCas9	Addgene	plasmid #47106	gift from Charles Gersbach
Recombinant DNA	pcDNA-dCas9-VP64	Addgene	plasmid #47107	gift from Charles Gersbach
Sequence-Based Reagent	Candidate gRNAs	Integrated DNA Technologies
Software; Algorithm	Deskgen	deskgen.com/landing/cloud.html#/
Software; Algorithm	GeCKO v.2
Software; Algorithm	NGS-Data-Charmer	https://github.com/shenlab-sinai/NGS-Data-Charmer
Software; Algorithm	SAMtools
Software; Algorithm	featureCounts	http://subread.sourceforge.net/
Software; Algorithm	DESeq2		version 1.27
Software; Algorithm	RRHO2	https://github.com/mestill7/RRHO2
Software; Algorithm	Prism	GraphPad