Transcriptional regulation of ventral hippocampus-nucleus accumbens circuit excitability drives cocaine seeking

Abstract

Ventral hippocampus (vHPC) CA1 pyramidal neurons send glutamatergic projections to nucleus accumbens (NAc), and this vHPC-NAc circuit mediates cocaine seeking and reward, but it is unclear whether vHPC-NAc neuron properties are modulated by cocaine exposure to drive subsequent behavior. The immediate early gene transcription factor FosB/ΔFosB is induced throughout the brain by cocaine and is critical for cocaine seeking, but its function in vHPC-NAc neurons is not understood. We now show that circuit-specific knockout of FosB/ΔFosB in vHPC-NAc neurons impaired cocaine reward expression and forced abstinence–induced seeking. We also found that vHPC-NAc excitability was decreased by experimenter-administered repeated cocaine and cocaine self-administration, and this cocaine-induced excitability decrease was mediated by ΔFosB expression. To uncover the mechanism of this change in circuit function, we used circuit-specific translating ribosome affinity purification to assess cocaine-induced, FosB/ΔFosB-dependent changes in gene expression in vHPC-NAc. We found that cocaine causes a FosB/ΔFosB-dependent increase in the expression of calreticulin, an endoplasmic reticulum–resident calcium-buffering protein. Calreticulin expression mediated vHPC-NAc excitability and was necessary for cocaine reward. These findings uncover a noncanonical mechanism by which cocaine increases calreticulin in vHPC leading to decreased vHPC-NAc excitability and drives cocaine seeking and reward.

ΔFosB induces calreticulin to decrease excitability of ventral hippocampus-accumbens projections and drive cocaine seeking.

INTRODUCTION

Addiction is a crucial social and health problem that can be defined as the unnatural drive to seek and take drugs despite negative consequences. Cocaine overdose deaths tripled between 2015 and 2020 (1), yet there are now no US Food and Drug Administration–approved treatments for cocaine substance use disorders. To develop novel therapeutics for cocaine substance use disroder (SUD), we require a better understanding of how drugs remodel neuronal circuitry that may underlie addictive behavior. Addiction is characterized by reward dysfunction and aberrant seeking behavior, e.g., relapse, leading to changes in neuroplasticity, gene expression, and neurocircuitry (2–5). The ventral hippocampus (vHPC) is important for emotional memory and is functionally distinct from dorsal HPC (6), and recent evidence suggests a role for vHPC in encoding external stimuli with internal drives, including the motivation to seek reward (7). vHPC is necessary for the relapse to drug seeking (8–10), and acute activation of vHPC induces drug seeking (11, 12). Glutamatergic afferents that innervate nucleus accumbens (NAc) medium spiny neurons (MSNs) are important in drug-induced plasticity, reward, and seeking (13, 14). vHPC sends glutamatergic projections to NAc (vHPC-NAc), and this circuit is critical for cocaine reward and seeking (7, 15, 16). Through excitatory synaptic plasticity, cocaine promotes indelible changes that underlie addictive behavior, like altered reward and drug seeking. Cocaine reshapes the plasticity of vHPC inputs onto NAc MSNs (16–18), and reversal of this plasticity impairs cocaine seeking (17). However, mechanisms by which cocaine remodels the function of vHPC neurons that project onto NAc MSNs remain unknown.

Psychostimulants, like cocaine and amphetamine, alter the intrinsic excitability of neurons in many brain regions. Of note, amphetamine decreases the intrinsic excitability of neurons in ventral subiculum (19), a region close in proximity to the vHPC that also sends projections to the NAc. Collectively, these findings suggest that drugs of abuse not only produce changes in synaptic plasticity in accumbal synapses but also produce changes in brain regions projecting to NAc that regulate addictive behavior, e.g., vHPC. However, it is now unknown whether cocaine directly affects the intrinsic extricability of vHPC-NAc cells.

Cocaine may be remodeling vHPC neurons via transcriptional changes. ΔFosB, a stable immediate early gene transcription factor (2, 20) that mediates hippocampal neuron function (21–23), is induced in vHPC by cocaine (24, 25), and its expression in vHPC is necessary for cocaine reward (24). Thus, we hypothesized that cocaine, via FosB/ΔFosB, may be remodeling vHPC neurons through altered gene expression and cell excitability to drive cocaine reward and seeking behavior. The current study demonstrates that cocaine induces FosB/ΔFosB, which drives expression of the target gene calreticulin to decrease vHPC-NAc excitability and drive cocaine reward and seeking.

RESULTS

Cocaine increases ΔFosB expression in vHPC-NAc neurons

We have previously established that repeated, experimenter-administered cocaine induces ΔFosB mRNA and protein expression in vHPC neurons (24). There is also semiquantitative evidence suggesting that ΔFosB is induced by cocaine self-administration in dorsal HPC (25). We sought to determine the extent to which cocaine self-administration induces ΔFosB in vHPC neurons. Separate cohorts of male mice were trained to self-administer intravenous infusions of cocaine (0.5 mg/kg per infusion) or saline for 14 days (Fig. 1, A to C, and fig. S1). One day after the last session, vHPC tissue was taken for assessment of ΔFosB mRNA and protein (Fig. 1D). Cocaine self-administration increased ΔFosB mRNA in vHPC (Fig. 1E) compared to saline self-administering mice.

Fig. 1. Chronic cocaine induces ΔFosB in vHPC neurons that project to NAc.

(A) Timeline of cocaine or saline self-administration in male mice. Arrow indicates the time of tissue collection. h, hours. (B) Cocaine self-administration performance. Nosepoke responses per min across cocaine (Coc) or saline (Sal) self-administration. Cocaine self-administering mice responded on the active nosepoke more than saline self-administering mice across sessions [two-way repeated measures (RM) analysis of variance (ANOVA): ME of group: F_1,11 = 19.54, P = 0.0010; ChamberXGroup: F_13,143 = 2.169, P = 0.0137], with significantly increased active nosepoke responding on days 6 and 11 to 14 (Holm-Šidák: *P < 0.05). No differences were observed between groups in inactive nosepoke responses. ME, mean error. (C) Cocaine or saline infusions per minute across 14 daily 2-hour sessions where responses on an active nosepoke resulted in an intravenous infusion of cocaine (Coc; 0.5 mg/kg per infusion; n = 8) or saline (Sal; n = 5). Cocaine self-administering mice received higher number of infusions across days compared to saline self-administering mice (two-way RM ANOVA: DayXGroup: F_13,143 = 2.911, P = 0.009), with significantly more infusions on days 4 to 14 (Holm-Šidák tests: *P < 0.05 for days 4 to 14). (D) Illustration of tissue biopsy punch (red circle) in vCA1. (E) Cocaine self-administration enhances ΔFosB in vHPC. vHPC tissue was taken 1 day after the last self-administration session for detection of ΔFosB mRNA or protein. Cocaine self-administration significantly increased ΔFosB mRNA compared to saline (independent samples t test: t₁₁ = 2.997, *P = 0.0121). F.C., fold change. (F) Circuit-tagging procedure. Male mice with crossed Floxed FosB and L10-GFP (KO; FosB^fl/fl:Rosa26^eGFP-L10a) or L10-GFP only (WT; Rosa26^eGFP-L10a) received retrograde HSV vector expressing Cre (HSVrg-hEf1ɑ-Cre) into NAc to drive GFP and FosB KO in projections to accumbens, including vHPC-NAc projections. All mice were treated with cocaine (Coc; 20 mg/kg, ip) or saline (Sal) for 10 days and 1 day later vHPC coronal brain slices were taken for FosB/ΔFosB immunohistochemistry. (G) Representative 20× coronal images of GFP (green, top), FosB/ΔFosB (red, bottom), and a merged image (bottom) in WT-Sal (n = 5), WT-Coc (n = 4), and KO-Coc (n = 3) mice. White arrows denote GFP⁺ neurons coexpressing FosB/ΔFosB. (H) Cocaine increases FosB/ΔFosB in vHPC-NAc neurons. Quantification of colabeled GFP⁺ and FosB/ΔFosB⁺ neurons in all three groups. WT-Coc had significantly greater colabeled neurons compared to WT-Sal and KO-Coc (one-way ANOVA: F_2,9 = 18.01, P = 0.0007; Holm-Šidák: **P < 0.01, ***P < 0.001).

The vHPC contains heterogeneous cell types, but NAc-projecting vHPC neurons (vHPC-NAc), which are primarily in vCA1 and ventral subiculum [vSub; (22)], are critical for cocaine seeking (17, 26). We assessed whether repeated cocaine increases FosB/ΔFosB expression in vHPC-NAc neurons using a circuit-tagging approach (Fig. 1F) to drive the expression of green fluorescent protein (GFP; fig. S2) and knock out the FosB gene in projections to NAc. Briefly, Cre-dependent GFP reporter mice (Rosa26^eGFP-L10a) were crossed with floxed FosB mice (FosB^fl/fl:Rosa26^eGFP-L10a). Both wild-type (WT; Rosa26^eGFP-L10a) and conditional knockout mice (FosB KO; FosB^fl/fl:Rosa26^eGFP-L10a) received intracranial infusions of a retrograde herpes simplex virus (HSV) vector expressing Cre recombinase (HSVrg-hEf1α-Cre) into NAc. Three to 6 weeks later, circuit-tagged WT and FosB KO mice received 10 days of repeated, experimenter-administered cocaine [20 mg/kg, intraperitoneally (ip)] or saline and were assessed for FosB/ΔFosB immunohistochemistry in GFP-tagged vHPC-NAc neurons. Similar to our previous findings in whole vHPC (24), cocaine increased FosB/ΔFosB expression in vHPC-NAc neurons, as well as non–NAc-projecting vHPC neurons (Fig. 1, G and H, and fig. S3). FosB/ΔFosB expression was completely blocked in vHPC-NAc neurons of the FosB KO mice, without affecting non–NAc-projecting vHPC neurons. These data show that experimenter- or self-administered cocaine increase ΔFosB in vHPC and specifically show that cocaine also induces FosB/ΔFosB in vHPC-NAc neurons that are important for drug seeking.

FosB/ΔFosB in vHPC-NAc is necessary for cocaine reward and seeking

On the basis of our finding that cocaine induces FosB/ΔFosB in vHPC-NAc neurons, we surmised that FosB/ΔFosB may be playing a role in vHPC-NAc drive of cocaine seeking and reward. We have previously demonstrated that hippocampal ΔFosB expression regulates learning and memory (21) and cocaine conditioned place preference (CPP) (24), but its role in vHPC-NAc neurons in the context of drug seeking remains unknown. Therefore, we next sought to determine whether FosB/ΔFosB expression in vHPC-NAc is necessary for cocaine reward using our previously characterized circuit-specific CRISPR approach (Fig. 2A) that knocks down FosB/ΔFosB specifically in vHPC-NAc neurons (22). CRISPR-mediated FosB KO (FosB crKO) significantly reduced the preference for a context conditionally paired with cocaine (Fig. 2, B to D), demonstrating that FosB/ΔFosB expression in vHPC-NAc neurons is necessary for the expression of cocaine reward. However, place preference does not distinguish between deficits in the processing of reward or deficits in ability to form context-reward association memories. To determine the specific role of vHPC-NAc FosB/ΔFosB expression, we turned to an operant cocaine self-administration model.

Fig. 2. Selective FosB knockout in vHPC neurons that project to NAc impairs cocaine reward and cocaine seeking.

(A) Illustration of dual virus CRISPR method to selectively knockout FosB/ΔFosB in vHPC-NAc neurons. (B) Cocaine CPP procedure and timeline. Male controls (Con; n = 15) or FosB crKO (crKO; n = 16) mice underwent cocaine CPP in a three-chamber box for twice-daily conditioning sessions of saline and cocaine (5 mg/kg, ip) followed by a drug-free posttest. (C) Saline-paired (Sal) and cocaine-paired (Coc) chamber time during the posttest for Con and crKO mice. Con mice spent significantly more time in the Coc chamber than in the Sal chamber; however, crKO had no differences in time spent between chambers (two-way RM ANOVA: ChamberXGroup: F_1,29 = 4.738, P = 0.0378; Holm-Šidák: **P < 0.01). (D) Cocaine-paired and cocaine-unpaired (i.e., saline) chamber difference in Con and crKO mice. crKO had significantly reduced paired-unpaired time compared to Con (independent samples t test: t₂₉ = 2.177, *P = 0.0378). (E) Forced abstinence–induced seeking procedure. h, hours. (F) Con (n = 6) and crKO (n = 12) mice did not differ in the number of infusions across cocaine self-administration sessions. (G) After a 7-day forced abstinence in their homecage, Con increased the number of mock infusions per minute during a seeking test compared to crKO (independent samples t test: t₁₅ = 2.377, *P = 0.0312). Created in BioRender. Eagle, A. (2025) https://BioRender.com/tfyofsj.

Forced abstinence from cocaine self-administration produces long-term changes in plasticity (27), including at vHPC synapses onto NAc MSNs (17), and reversing these changes attenuates cocaine seeking (17, 28). Therefore, we next asked whether FosB/ΔFosB expression in vHPC-NAc is necessary for acquisition and maintenance of cocaine self-administration and/or for forced abstinence–induced cocaine seeking. Mice were trained to respond on an active nosepoke for cocaine for 14 days (0.5 mg/kg per infusion, intravenously) on a fixed ratio 1 (FR1) schedule (Fig. 2E). FosB crKO had no effect on the acquisition or maintenance of cocaine self-administration, indicated by number of infusions (Fig. 2F), nosepoke responses, and accuracy (fig. S4). It is important to note that overall responding in these groups was low, likely due to the effects of prior stereotaxic surgery and viral vector expression. Mice were then left in their homecages for 1 week before testing for abstinence-induced seeking back in the cocaine context. In the seeking test, nosepoke responses only resulted in mock infusions. We found that vHPC-NAc FosB crKO impaired abstinence-induced cocaine seeking, without affecting cocaine self-administration (Fig. 2G and fig. S4, B and D), reducing the number of mock infusions (Fig. 2G) and active nosepoke responses (fig. S4), without affecting accuracy. Collectively, these findings indicate that FosB/ΔFosB expression in vHPC-NAc is required for context-induced cocaine reward and seeking.

FosB/ΔFosB mediates cocaine-induced decreases in vHPC-NAc excitability

Psychostimulants, like cocaine and amphetamine, alter the intrinsic excitability of neurons in ventral subiculum (19), cortex (29, 30), lateral habenula (31), and NAc (32–36). Of particular importance, amphetamine decreases the intrinsic excitability of neurons in ventral subiculum (19), a region close in proximity to the vHPC that also sends projections to the NAc (37). Collectively, these findings suggest that, beyond changes in synaptic plasticity at accumbal synapses, psychostimulants produce intrinsic excitability changes in brain regions that regulate drug responses and project to NAc. However, it is unknown whether cocaine directly affects the intrinsic extricability of vHPC neurons. To address this, male C57Bl/6J mice were trained to self-administer cocaine (0.5 mg/kg per infusion) or saline for 14 days on an FR1 active nosepoke schedule of reinforcement (fig. S5). One day after the last cocaine (or saline) self-administration session, vHPC was sliced for ex vivo electrophysiology, and we found that cocaine self-administration reduced vHPC CA1 pyramidal neuron excitability (Fig. 3, A to D, and table S1). This was indicated by a decrease in depolarization-induced spike frequency (Fig. 3B), a decrease in total spikes (Fig. 3C), and no changes in rheobase current (Fig. 3D). We also used circuit-tagging in cocaine and saline self-administering WT male mice (Fig. 3E) to assess excitability of ventral CA1 neurons that project to NAc (vCA1-NAc). vHPC tissue was taken 1 day after the last self-administration session, and we found that cocaine self-administration also reduced vCA1-NAc excitability (spike frequency and total spikes) (Fig. 3, F and G, and table S2). We also observed an increase in rheobase in vCA1-NAc neurons from cocaine self-administering mice (Fig. 3H). We previously established that ΔFosB expression decreases dorsal and ventral CA1 pyramidal neuron excitability (22, 23), and ΔFosB knockout enhances basal (i.e., no drug or other manipulation) vCA1-NAc excitability in male mice (22). On the basis of our finding that cocaine induces FosB/ΔFosB in vHPC-NAc neurons, we hypothesized that FosB/ΔFosB expression mediates the cocaine-dependent reduction in vCA1-NAc excitability. We used our circuit tagging approach (Fig. 3I) to express GFP and knock out FosB/ΔFosB in vCA1-NAc neurons in Cre-dependent GFP (WT) and floxed FosB (FosB KO) male mice treated with 10 days of experimenter-administered cocaine (20 mg/kg, ip) or saline. We initially did not include a FosB KO group that would only receive saline, as we have previously shown that FosB/ΔFosB knockout/knockdown increases basal vCA1 and vCA1-NAc excitability (22), as well as dorsal HPC CA1 and dentate gyrus neuron excitability (23). vHPC slices were taken 1 day after the last injection for whole-cell electrophysiology. We found that experimenter-administered cocaine in WT mice reduced vCA1-NAc excitability, as indicated by a decrease in depolarization-induced spike frequency (Fig. 3, J and K), with no change in rheobase (Fig. 3L), or other intrinsic cellular properties (table S3). Critically, FosB/ΔFosB knockout in vCA1-Nac neurons (FosB KO mice) blocked this cocaine-induced decrease in excitability. Furthermore, no differences were observed between cocaine- and saline-treated FosB KO mice, as expected (fig. S6). This further supports the conclusion that FosB/ΔFosB is required for cocaine-induced decreases in excitability. These findings collectively suggest that cocaine reduces the excitability of vHPC-NAc neurons via induction of FosB/ΔFosB.

Fig. 3. Cocaine, through ΔFosB, decreases vHPC projection neuron excitability.

(A) Illustration of ex vivo patch-clamp approach (top), representative spike traces (middle), and timeline (bottom) for patched vCA1 neurons from cocaine (Coc; n = 4 mice; n = 17 cells) and saline (Sal; n = 3; n = 13 cells) self-administering male mice. Scale bar: 50 pA, 200 ms. (B) Cocaine self-administration reduces vCA1 neuron spike frequency (two-way RM ANOVA: ME of group: F_1,28 = 7.320, P = 0.0115; CurrentXGroup: F_11,308 = 1.993, P = 0.0286; Holm-Šidák: *P < 0.05, #P < 0.10). (C) Cocaine self-administration reduces the total number of spikes (independent samples t test: t₂₈ = 2.705, *P = 0.0115). (D) There were no significant differences between groups on rheobase. (E) Circuit-tagging procedure to target vCA1-NAc neurons (top) and self-administration timeline (bottom) for cocaine (WT-Coc; n = 6 mice; n = 28 cells) or saline mice (WT-Sal; n = 5 mice; n = 17 cells) mice. (F) Cocaine self-administration significantly reduced spike frequency in vCA1-NAc neurons (two-way RM ANOVA: CurrentXGroup: F_12,516 = 4.674, P < 0.0001; Holm-Šidák: *P < 0.05, #P < 010). (G) Cocaine self-administration significantly reduced total spikes (independent samples t test: t₄₃ = 2.230, *P = 0.0310). (H) There was a trend toward a significant increase in rheobase in vCA1-NAc neurons of WT-Coc mice (independent samples t test: t₄₃ = 1.719, #P = 0.0928). (I) Circuit-tagging procedure to target vCA1-NAc neurons (top) and experimenter-administered injection timeline (bottom). WT mice received saline (WT-Sal; n = 8 mice; n = 23 cells) or cocaine (WT-Coc; n = 6 mice; n = 17 cells). FosB KO mice received cocaine (KO-Coc; n = 4 mice; n = 13 cells). (J) Cocaine significantly reduced spike frequency in WT mice (two-way RM ANOVA: CurrentXGroup: F_22,550 = 3.545, P < 0.0001; Holm-Šidák: WT-Sal versus WT-Coc, *P < 0.05). KO-Coc reduced spike frequency compared to WT-Coc (Holm-Šidák: WT-Coc versus KO-Coc, *P < 0.05 for 225 to 300 pA). (K) There was a trend toward a significant difference in the total number of spikes in WT-Coc compared to those in WT-Sal and KO-Coc (one-way ANOVA: F_2,50 = 2.728, P = 0.0751; Holm-Šidák: #P < 010). (L) There were no differences between groups on rheobase. Created in BioRender. Eagle, A. (2025) https://BioRender.com/s4xwqz2.

Chronic excitability mediates cocaine reward

Cocaine-induced changes in plasticity underlie many behavioral responses to the drug (17, 27, 28, 34). Therefore, we hypothesized that cocaine-induced reduction in vHPC-NAc excitability underlies cocaine reward and seeking behaviors. Acute activation of vHPC, including vHPC-NAc neurons, enhances or induces cocaine seeking (11), whereas acute inhibition can decrease cocaine seeking (26). However, it is unclear how a chronic decrease in membrane excitability underlies cocaine behavioral responses. Reversal of drug-induced changes in prefrontal cortex and NAc neuronal excitability ameliorates cocaine seeking (30, 34, 38, 39). In line with this evidence, we assessed whether chronic chemogenetic increases in excitability of vHPC-NAc neurons affect cocaine reward (Fig. 4, A and B). To express circuit-specific designer receptors exclusively activated by designer drugs (DREADD) G_q receptors specifically in vHPC-NAc neurons, we used an intersectional virus strategy (Fig. 4A). Male C57Bl/6J mice received infusions of retrograde HSV expressing Cre into NAc (HSVrg-hEf1α-Cre) and either AAV2-CMV-GFP or AAV2-hSyn-DIO-hM3Dq-mCherry in vHPC. We also included a separate control cohort to assess chronic decreases in vHPC-NAc excitability in mice receiving retrograde HSV-Cre in NAc and AAV2-hSyn-DIO-hM4Di-mCherry in vHPC. To chronically activate or inhibit vHPC-NAc neurons, mice received 14 days of twice-daily intraperitoneal injections of the DREADD ligand, clozapine-N-oxide (CNO; 0.3 mg/kg). During the last 4 days, mice underwent cocaine CPP (Fig. 4B). Chronic vHPC-NAc activation impaired place preference for cocaine (Fig. 4, C and D). Specifically, GFP control mice spent significantly more time in a chamber paired with cocaine than in a chamber paired with saline (Fig. 4C). However, G_q-expressing mice spent a similar amount of time in both the saline- and cocaine-paired chambers. Preference (cocaine-paired minus cocaine-unpaired chamber) was also significantly reduced in G_q mice compared to that in GFP mice (Fig. 4D). A shorter subchronic activation (4 days of CNO treatment during the CPP) did not impair the expression of cocaine reward (fig. S7). Chronic inhibition of the vHPC-NAc circuit did not affect cocaine CPP at the 5 mg/kg dose (Fig. 4, E and F). These findings demonstrate that cocaine-induced decreases in vHPC-NAc excitability mediate cocaine place preference.

Fig. 4. Chronic chemogenetic activation of vHPC neurons that project to NAc impairs the CPP for cocaine.

(A) Viral approach to express DREAD excitatory G_q and inhibitory G_i receptors in vHPC-NAc neurons of male mice. A retrograde HSV vector expressing Cre (HSVrg-hEf1ɑ-Cre) was injected into NAc and either a control (AAV2-CMV-GFP), DREADD G_q (AAV2-hSyn-DIO-hM3Dq-mCherry), or DREAD G_i viral vector (AAV2-hSyn-DIO-hM4Di-mCherry) was injected into vHPC. h, hour. (B) Cocaine CPP procedure and timeline. CNO was administered twice-daily for 14 days that ended with cocaine (5 mg/kg, ip) CPP test. (C) Saline-paired (Sal) and cocaine-paired (Coc) chamber time during the posttest for GFP (n = 8) and G_q (n = 8) mice. GFP mice spent significantly more time in the Coc chamber than in the Sal chamber; however, G_q did not differ in time spent between chambers (two-way RM ANOVA: ChamberXGroup: F_1,14 = 4.737, P = 0.0471; Holm-Šidák: **P = 0.0063). (D) Cocaine-paired and cocaine-unpaired (i.e., saline) chamber difference in GFP and G_q mice. G_q had significantly reduced paired-unpaired time compared to GFP (independent samples t test: t₁₄ = 2.176, *P = 0.0471). (E) Saline-paired (Sal) and cocaine-paired (Coc) chamber time during the posttest for GFP (n = 14) and G_i (n = 16) mice. Both GFP and G_i mice spent significantly more time in the Coc chamber than in the Sal chamber (two-way RM ANOVA: ME of chamber: F_1,28 = 16.81, P = 0.0003; Holm-Šidák: *P = 0.0240). (F) Cocaine-paired and cocaine-unpaired (i.e., saline) chamber difference in GFP and G_i mice. There were no significant differences between groups (independent samples t test: P > 0.05). Created in BioRender. Eagle, A. (2025) https://BioRender.com/0hdp9o1.

Cocaine, via FosB/ΔFosB, alters gene expression in vHPC-NAc neurons

ΔFosB regulates gene transcription in a variety of brain regions, including the hippocampus (2, 20, 22); therefore, we next sought to identify a transcriptional mechanism of cocaine-induced changes in membrane excitability and cocaine behavior. We used a transcriptomic approach in vHPC-NAc using TRAP-seq (translating ribosome affinity purification followed by sequencing) (40) to assess actively translating mRNAs in vHPC-NAc neurons (Fig. 5A). Circuit-tagged WT and FosB KO male mice were experimenter-administered 10 days of cocaine (20 mg/kg, ip) or saline. One day after the last injection, fresh-frozen vHPC tissue punches containing GFP⁺ cells were collected, and samples were pooled (n = 4 to 5 per sample) and processed for TRAP-seq. We first compared WT mice that received cocaine or saline. We plotted all 14,495 genes across confidence (P value) and fold change (Fig. 5B). We identified 587 total significant cocaine-regulated genes (P < 0.05), 270 genes that were up-regulated by cocaine and 317 that were down-regulated by cocaine. We also compared mice treated with cocaine with intact FosB (WT) and FosB KO in vHPC-NAc neurons (Fig. 5C). We identified 774 total significant cocaine-regulated genes (P < 0.05), 385 genes that were up-regulated by cocaine and 390 that were down-regulated by cocaine. Of these significantly regulated genes, we examined cocaine-regulated FosB/ΔFosB-dependent genes. We defined these genes as those that were reversibly regulated; either genes up-regulated by cocaine in WT mice and down-regulated by FosB KO in cocaine-treated mice or vice versa. We identified a total of 152 reversibly regulated genes. Of these, 108 were down-regulated by cocaine and up-regulated when FosB was knocked out in cocaine-treated mice, and 45 were up-regulated by cocaine and down-regulated when FosB was knocked out in cocaine-treated mice. These data revealed cocaine-induced, FosB/ΔFosB-dependent changes in actively translating gene expression.

Fig. 5. Cocaine-induced gene expression in vHPC neurons that project to NAc.

(A) Circuit-tagging and TRAP-seq approach [modified from (22)]. (B) Differentially regulated genes in vHPC-NAc neurons between WT-Coc versus WT-Sal mice. Gray indicates nonsignificant (N.S.) genes. (C) Differentially regulated genes in vHPC-NAc neurons between FosB KO-Coc versus WT-Coc mice. (D) vHPC tissue punches from cocaine (Coc; n = 8) and saline (Sal; n = 5) self-administering male mice were processed for Calr qPCR. Cocaine self-administration increases Calr mRNA in vHPC (independent samples t test: t₁₁ = 2.424, *P = 0.0338). (E) vHPC tissue punches from cocaine (Coc, C; n = 6) and saline (Sal, S; n = 6) self-administering male mice were processed for Western blotting for calreticulin. Cocaine increases calreticulin in vHPC (independent samples t test: t₁₀ = 2.428, *P = 0.0356). (F) Pooled vHPC tissue samples (n = 3 mice per sample; n = 4 samples per group) from experimenter-administered cocaine (10 days; Coc; 20 mg/kg, ip; n = 12) and saline male mice (Sal; n = 12) were processed for FosB/ΔFosB ChIP-PCR for Calr. Cocaine increased FosB/ΔFosB binding (% input) to the Calr promoter (two-way ANOVA: InputXGroup: F_1,12 = 21.64, P = 0.0471; Holm-Šidák: ****P < 0.0001). (G) Circuit-tagged male WT and FosB KO mice received 10 days of cocaine (Coc; 20 mg/kg, ip; n = 6) or saline (Sal; n = 5). One or 14 days after the last injection, perfused vHPC slices were collected for immunohistochemistry. Representative 40× images of calreticulin (red) and GFP (green; vHPC-NAc neurons) from vHPC coronal slices. White arrows denote GFP⁺ vHPC-NAc neurons. (H) Cocaine increased calreticulin intensity in GFP⁺ neurons from WT mice at 1 day after the last injection (one-way ANOVA: F_3,1773 = 134.5; Holm-Šidák: ****P < 0.0001 for all comparisons to WT-Sal, FosB KO-Sal, and FosB KO-Coc). (I) Cocaine increased calreticulin intensity in GFP⁺ neurons 14 days after the last injection both per cell (independent samples t test: t₁₁₃₀ = 14.32, ****P < 0.0001) and averaged per mouse (independent samples t test: t₈ = 2.678, *P = 0.0280).

Of particular interest in these potential FosB/ΔFosB target genes was the endoplasmic reticulum (ER)–resident chaperone protein, calreticulin. Calreticulin is a high-capacity, low-affinity Ca²⁺-binding chaperone protein that, along with calnexin, regulates ER Ca²⁺ storage (41). Calreticulin was up-regulated in vHPC-NAc neurons by cocaine but was down-regulated in cocaine-treated FosB KO mice compared to that in WT. Conversely, when we knockout FosB in these same neurons, cocaine no longer increased calreticulin expression. This suggests that calreticulin expression may be increased by cocaine-induced FosB/ΔFosB. To verify our sequencing results, we first assessed whether cocaine self-administration would increase calreticulin mRNA and protein. Male mice were trained to self-administer cocaine for 14 days, and vHPC tissue punches were taken 1 day after the last administration (as in Fig. 1A). Cocaine self-administration significantly up-regulated Calr mRNA in vHPC with a 2.73-fold change increase in cocaine compared to that in saline self-administering mice (Fig. 5D). From a separate cohort of cocaine and saline self-administering male mice, vHPC tissue was processed for total calreticulin protein (Fig. 5E and fig. S8). Cocaine significantly increased calreticulin protein in vHPC. We next sought to identify whether cocaine enhances FosB/ΔFosB binding to the Calr gene. We used the same experimenter-administration regimen (as Fig. 1F) in naïve male mice, and tissue was processed for chromatin immunoprecipitation (ChIP)–polymerase chain reaction (PCR). We found that cocaine treatment robustly enhanced ΔFosB binding to the Calr promoter region (Fig. 5F). We did not see substantial FosB/ΔFosB binding to the Calr promoter in saline-treated mice when compared to that in immunoglobulin G (IgG) controls, suggesting that this mechanism is driven by cocaine induction of FosB/ΔFosB.

We next sought to determine whether calreticulin is induced by cocaine in vHPC-NAc neurons. To address this, we used our circuit-tagging approach and treated mice with experimenter-administered cocaine (20 mg/kg, ip) for 10 days (Fig. 1F). vHPC tissue was taken 1 or 14 days after the last injection for immunohistochemistry staining of calreticulin. Calreticulin is highly conserved and has abundant expression across tissue and cell types (42, 43). Therefore, we measured calreticulin staining intensity for each GFP⁺ vHPC-NAc neuron as a semiquantitative measure of its expression in this circuit. We found that cocaine increases calreticulin intensity in vHPC-NAc neurons both at 1 day after chronic cocaine (Fig. 5, G and H) and 14 days after chronic cocaine (Fig. 5I). Specifically, we observed a significant increase in calreticulin intensity across total cells at both time points and calreticulin intensity averaged per mouse at 14 days. We also observed that FosB KO blocked the cocaine induction of calreticulin in vHPC-NAc neurons at 1 day (Fig. 5H). Collectively, these findings validate our TRAP-seq results and demonstrate that cocaine induction of ΔFosB drives calreticulin expression in vHPC-NAc neurons and, furthermore, that this increase is relatively long-lasting.

Calreticulin mediates cocaine changes in vHPC-NAc excitability and behavior

The role of calreticulin in neurons is poorly understood, although there is evidence suggesting that it plays a role in memory (44). Calreticulin’s role in neuronal excitability has not been investigated, although there is interesting evidence that calreticulin and neuronal excitability may be loosely linked to its role in ER Ca²⁺ homeostasis and chaperone functions (45). We therefore sought to determine whether calreticulin may play a role in vHPC-NAc excitability. Our first step was to determine whether calreticulin up-regulation was sufficient to decrease vHPC excitability. To address this question, we used viral overexpression of calreticulin in vHPC (AAV2-CMV-mCalr-GFP; or AAV2-CMV-GFP as control) in male C57Bl/6J mice. Approximately 4 to 6 weeks later, we assessed excitability using ex vivo slice electrophysiology in GFP-labeled calreticulin-overexpressing vCA1 neurons (Fig. 6A). Calreticulin overexpression reduced the excitability of vHPC neurons compared to that of control vHPC neurons (Fig. 6, C and D). Specifically, we observed a calreticulin overexpression-induced decrease in spike frequency (Fig. 6B) and the total number of spikes (Fig. 6C) and a calreticulin overexpression-induced increase in rheobase (Fig. 6D). This suggests that calreticulin is sufficient to decrease vHPC excitability in the absence of cocaine or ΔFosB.

Fig. 6. Calreticulin mediates ventral hippocampal excitability and behavior.

(A) Illustration of approach for calreticulin overexpression on vHPC excitability and representative spike traces (300 pA; scale bar: 50 pA, 200 ms) from vCA1 neurons (n = 11 cells per group; n = 3 mice per group). (B) mCalr overexpression reduced vCA1 spike frequency (two-way RM ANOVA: ME of group: F_1,20 = 7.600, P = 0.0122; CurrentXGroup: F_11,220 = 1.935, P = 0.0363; Holm-Šidák: *P < 0.05, #P < 0.10). (C) mCalr reduced total spikes (independent samples t test: t₂₁ = 2.298, *P = 0.0319) and (D) produced a trend toward significance in rheobase (independent samples t test: t₂₁ = 2.041, #P = 0.0540). (E) Circuit-tagging approach to drive GFP and KO calreticulin in vHPC-NAc neurons. (F) Representative 20× images in WT (top) and Calr KO (bottom) mice. Calreticulin (red) and GFP (green) neurons. White arrows denote GFP⁺ neurons with no calreticulin expression. (G) Representative 200-pA spike traces in GFP⁺ vCA1-NAc neurons in WT (GFP-WT; n = 6 mice; n = 40 cells) and Calr KO (GFP-Calr KO; n = 4 mice; n = 16 cells). Scale bar: 50 pA, 200 ms. (H) GFP-Calr KO significantly reduced spike frequency (two-way RM ANOVA: ME of group: F_1,54 = 28.99, P < 0.0001; CurrentXGroup: F_11,594 = 2.100, P = 0.0187; Holm-Šidák: *P < 0.05). (I) GFP-Calr KO significantly increased total spikes (independent samples t test: t₅₄ = 5.409, ****P < 0.0001) (J) with no effect on rheobase. (K) Intersectional virus strategy to selectively knockout Calr in vCA1-NAc neurons. (L) Cocaine (5 mg/kg, ip) CPP procedure in male controls (WT; n = 15) or Calr KO (Calr KO; n = 10). (M) WT mice spent significantly more time in the Coc chamber; however, Calr KO had no differences in time spent between chambers (two-way RM ANOVA: ChamberXGroup: F_1,23 = 6.311, P = 0.0195; Holm-Šidák: ***P = 0.0002). (N) Calr KO significantly reduced paired-unpaired time (independent samples t test: t₂₃ = 2.512, *P = 0.0195). Created in BioRender. Eagle, A. (2025) https://BioRender.com/fjfuo50.

We next sought to determine whether calreticulin knockout in vHPC-NAc would alter the excitability of vCA1-NAc neurons. We received a donation of floxed calreticulin mice (Calr^fl/fl) and crossed these mice with our Rosa26^eGFP-L10a strain to produce a crossed line (Calr^fl/fl: Rosa26^eGFP-L10a). We used our circuit-tagging approach (Fig. 6E) to drive both GFP expression and knockout calreticulin (Calr KO) in all of the NAc-projecting neurons, including vHPC-NAc neurons. We performed immunohistochemistry for GFP and calreticulin in vHPC tissue slices from WT (Rosa26^eGFP-L10a) and Calr KO male mice. We confirmed that Calr KO in vHPC-NAc neurons decreased calreticulin staining intensity (Fig. 6, F and G). In a separate cohort of WT and Calr KO male mice, we assessed the effects of Calr KO on vCA1-NAc excitability. We used our circuit-tagging approach (Fig. 6E) and measured excitability in GFP⁺ vCA1-NAc neurons from WT (GFP-WT) or Calr KO mice (GFP-Calr KO). We observed that knockout of calreticulin increased the excitability of vCA1-NAc neurons (Fig. 6, F and G), with no change in rheobase (Fig. 6H). These findings suggest that cocaine induction of calreticulin is a primary driver for cocaine-dependent changes in vHPC-NAc excitability.

Calreticulin changes in vHPC-NAc excitability may be a potential mediator of cocaine behavior; however, few studies have identified the role of calreticulin in neurons and its role in behavior. We sought to determine the role of calreticulin in cocaine reward on the basis of our observation that FosB knockout impairs cocaine reward (Fig. 2). We used an intersectional viral strategy in crossed Calr^fl/fl: Rosa26^eGFP-L10a mice to KO calreticulin in only vHPC-NAc neurons (Fig. 6K). Briefly, male WT (Rosa26^eGFP-L10a) and floxed calreticulin mice (Calr^fl/fl: Rosa26^eGFP-L10a) received viral infusions of AAVrg-Ef1a-FlpO into NAc and AAV8-EF1a-fDIO-Cre into vHPC. WT and Calr KO were then tested for cocaine CPP (Fig. 6L). Circuit-specific Calr KO in vHPC-NAc neurons reduced the preference for a context paired with cocaine (Fig. 6, M and N). WT mice spent more time in a cocaine-paired chamber than in a saline-paired chamber; however, Calr KO mice spent a similar amount of time in both chambers. These data suggest that vHPC-NAc calreticulin is necessary for the expression of cocaine reward and support our overall model that cocaine induces FosB/ΔFosB to increase expression of calreticulin, reducing membrane excitability of vHPC-NAc neurons and driving cocaine reward and seeking.

DISCUSSION

It has been well established that cocaine produces plasticity at vHPC synapses onto NAc neurons (17, 26, 37), yet these changes are largely postsynaptic in NAc MSNs. While there is no specific evidence that cocaine remodels presynaptic vHPC-NAc plasticity, this has been characterized in PFC synapses onto NAc MSNs (46). Compelling evidence also suggests that cocaine conditioning and self-administration reshape the function of vHPC neurons, producing changes in excitatory transmission and dendritic spine morphology (47–50). Here, we now demonstrate that cocaine remodels the intrinsic membrane excitability of vHPC-NAc neurons through induction of FosB/ΔFosB. The FosB/ΔFosB-mediated decrease in vHPC-NAc neuronal excitability mirrors prior findings that show that other psychostimulants reduce excitability in vSub (19) and our own data that ΔFosB regulates dorsal CA1 (23) and vCA1 excitability (22). The mechanisms by which FosB/ΔFosB reduces excitability were previously unclear; however, our analyses of gene expression in vHPC-NAc suggested both direct effects, e.g., ion channel expression, and indirect effects via signaling or downstream transcription factors. This lead to our in-depth investigation of the ER chaperone protein, calreticulin. In the future, it will be important to determine the time course by which FosB/ΔFosB reshapes the function of hippocampal neuron excitability and how this affects cellular and network function.

ΔFosB has a well-established role in NAc and is regulated there by stress, natural rewards, and drug exposure (51–53), but its role in HPC is less clear. Previous studies showed that ΔFosB expression is induced in HPC (21, 24, 25, 54), and we demonstrated that cocaine induces ΔFosB in vHPC neurons (24). We show here that FosB/ΔFosB is induced in vHPC-NAc neurons; however, these cocaine-induced changes are not specific to vHPC-NAc neurons. We have observed that cocaine induces FosB/ΔFosB in whole HPC (dorsal HPC, ventral DG, vCA1-NAc projections, as well as nonprojecting vCA1 neurons). ΔFosB is induced in hippocampus by many other stimuli, including learning (21), stress (22, 55), other drugs of abuse (25), chronic electroconvulsive shock (54, 56), seizures and seizure-inducing agents (57, 58), exercise (59), and antidepressants (55). It can be inferred from the aggregate of these many findings that chronic stimuli induce FosB/ΔFosB in hippocampus, with possible variations that are dependent on region and treatment. The accumulation of FosB/ΔFosB in these areas may have a number of effects; in our case, changes in cocaine effects on vHPC neurons that project to NAc. We show that increases in FosB/ΔFosB expression are quite robust with a near doubling of mRNA expression and the number of FosB/ΔFosB protein positive cells, mirroring what we and others have observed (21, 24, 25, 54). ΔFosB is a chronic activity-dependent transcription factor; therefore, its induction regulates gene expression that likely leads to indelible changes in plasticity, with many of these effects observed in NAc (25, 51–53, 56, 60). We observed a cocaine-induced change in vHPC-NAc gene expression that was, in part, FosB/ΔFosB-dependent. Chronic cocaine induces FosB/ΔFosB expression across many brain regions (25, 52), and this pattern is not limited to only cocaine. However, differences in the temporal pattern of induction, if any, between vHPC and NAc are unknown, and future studies could interrogate the pattern of induction across many different brain regions and different time points.

We also observed that vHPC-NAc FosB/ΔFosB is necessary for cocaine reward and seeking. This is in line with our prior findings that cocaine exposure epigenetically modifies the FosB gene to enable ΔFosB expression in vHPC and drive cocaine reward (24) and that ΔFosB in dorsal hippocampus (dHPC) is necessary for hippocampal-dependent learning and memory (21). ΔFosB in NAc is similarly regulated by cocaine and drives cocaine responses (60, 61). However, its role in hippocampal-dependent regulation of cocaine behavior is not as well understood. It has been hypothesized that vHPC is necessary for linking external stimuli, e.g., cues and contexts, with internal drives, such as the drive to seek and consume cocaine (7). This may require associative memory formation of cues or contexts associated with cocaine and other drugs and the underlying plasticity required for these phenomena. vHPC projections to NAc are necessary and sufficient for drug seeking (8, 9, 11, 26, 50), and the present work suggests that ΔFosB may be a key player regulating excitability of this circuit underlying behavioral responses to cocaine.

It is unclear how a FosB/ΔFosB expression-dependent decrease in vHPC-NAc neuron excitability affects NAc MSNs. It is not entirely clear whether FosB/ΔFosB-induced decreases in excitability directly causally affect cocaine behavior or whether the decreased excitability is due to some as-yet undiscovered indirect mechanism. The activity of glutamatergic afferent synapses onto NAc MSNs regulates synaptic plasticity (62), which is particularly affected by drugs of abuse including cocaine, ultimately leading to increased relapse-like behaviors in rodents (3, 17). The cocaine-induced, FosB/ΔFosB-dependent changes in vHPC-NAc neurons, therefore, may lead to downstream changes in synaptic connections between vHPC and NAc, including strengthening or weakening of these connections. This is a probable underlying mechanism for changes in the expression of cocaine reward and cocaine seeking, but future experiments will be needed to uncover any synaptic effects.

Cocaine can also produce homeostatic plasticity changes in NAc MSNs (32, 34, 63). It may be that a FosB/ΔFosB-dependent decrease in vHPC-NAc excitability indirectly leads to changes in feedforward inhibition in NAc MSNs (64). These changes could also be dependent on whether vHPC synapses onto D1 or D2 MSNs. Relapse to cocaine seeking produces increased synaptic plasticity at vHPC synapses onto D1 MSNs (3), and cocaine produces feedforward inhibition largely at D1 MSNs (64); however, recent evidence shows that cocaine biases plasticity toward vSub neuron synapses onto D2 MSNs (37). We have also observed that FosB/ΔFosB expression in vHPC-NAc neurons mediates stress susceptibility and vHPC-NAc excitability (22). Stress and drugs may produce alterations in vHPC that lead to synergistic or opposing downstream changes in the NAc, an idea supported by previous work showing that both cocaine and stress remodel vHPC synapses onto NAc (16). These feedforward changes may have widespread impacts on mood and reward function, such as stress susceptibility and susceptibility to drug seeking.

The mechanism for calreticulin changes in vHPC excitability and cocaine behavior remains to be explored. Calreticulin is a high-capacity, low-affinity Ca²⁺-binding chaperone protein that, along with calnexin, regulates ER Ca²⁺ storage and protein folding (41). Calreticulin’s low affinity for Ca²⁺ is important in maintaining large concentrations of free Ca²⁺ in the ER available for rapid release, with calreticulin accounting for ~50% of the ER-free Ca²⁺ capacity (41, 42). While cocaine effects on ER Ca²⁺ release have not been investigated, cocaine reduces ER store-operated Ca²⁺ entry (65). It is quite possible that calreticulin is a mechanism for cocaine’s effects on ER regulation of intrinsic neuronal excitability, and the ER is a vital organelle in neurons for maintaining calcium (Ca²⁺) homeostasis. The ER can enhance intracellular Ca²⁺-mediated signaling, synaptic plasticity, and vesicle release (66). ER Ca²⁺ stores are replenished primarily through sarco-endoplasmic reticulum Ca²⁺ adenosine triphosphatase and Ca²⁺ release–activated channel, whereas Ca²⁺ release is primarily conducted through ryanodine receptors and inositol triphosphate receptors. Release of Ca²⁺ from ER stores can be initiated by a variety of signaling events, including changes in cytosolic Ca²⁺ and metabotropic receptor signaling. However, the role of ER Ca²⁺ release in regulating neuronal physiology (i.e., excitability) is not well understood, although there is a compelling argument from the ER’s role in synaptic plasticity (66) that the ER may also alter long-term excitability. Therefore, cocaine induction of calreticulin through ΔFosB may produce changes in ER that lead to decreased vHPC excitability and subsequent cocaine seeking.

We use both experimenter-administration and self-administration of cocaine in this study, and we feel this is a strength of the experimental design. Intravenous self-administration (IVSA) studies provide both translational relevance and a more subtle and varied array of interpretations for behavioral results, and we used them here to show that ΔFosB is induced in vHPC-NAc neurons by cocaine IVSA and is necessary in these cells for context-induced seeking of cocaine. For dissecting the specific role of FosB/ΔFosB and its downstream targets in vHPC-NAc function and behavioral outputs, we switched to higher-throughput experimenter-administration methods. We believe that the two methods of cocaine administration are likely to have similar downstream effects on hippocampal gene expression, as multiple studies have shown that both methods induce ΔFosB in the hippocampus (2, 24, 25) and NAc (2, 25, 60), and both methods alter excitatory inputs onto NAc MSNs (67, 68).

One potential drawback to the generalizability of these findings is that we only included male mice in our studies. This was primarily due to our own observations that vHPC-NAc neuron excitability in females (in the absence of cocaine or any other manipulation) is quite different than males (69). Specifically, female mice have greater basal excitability in vHPC-NAc neurons compared to males. This difference is driven by adult testosterone and reduces stress-driven anhedonia in male mice compared to that in females. There have been multiple findings showing that female mice are more likely to escalate drug self-administration and to reinstate at greater rates and are more motivated for drugs (70, 71). We have previously observed that ΔFosB knockout increases vHPC-NAc excitability (22), an effect that we have not observed in females, which is likely due to a ceiling effect. As we show here, knockout of calreticulin, a gene target of FosB/ΔFosB, also increases vHPC-NAc excitability in males. Therefore, it is likely that, if we were to repeat this study to include females, then we would likely not observe these same knockout/knockdown effects. However, increases in FosB/ΔFosB expression in females could be a potential mediator for cocaine changes in gene expression and vHPC-NAc excitability, and that will certainly be a focus of future studies in our group and others.

MATERIALS AND METHODS

Animals

All experiments were approved by the Institutional Animal Care and Use Committee at Michigan State University (Animal Use Form no. 202500408) and in accordance with AAALAC. Male C57Bl/6J mice (three to five per cage, 7 to 8 weeks old upon arrival from the Jackson Laboratory) were allowed at least 5 days to acclimate to the facility before any experimental procedures. The floxed FosB mouse strain (FosB^fl/fl) was a gift from the laboratory of E. Nestler at the Icahn School of Medicine at Mount Sinai. The Rosa26^eGFP-L10a mice were a gift from the laboratory of G. Leinninger at Michigan State University. The floxed calreticulin mouse strain (Calr^fl/fl) was originally created by M. Ikawa at Osaka University and was a gift from the laboratory of M. Michalak at the University of Alberta. Unless otherwise stated, all mice were group housed in a 12-hour:12-hour light/dark cycle with ad libitum food and water. Temperature (22°C) and humidity (50 to 55%) were held constant in animal housing and behavioral testing rooms.

Stereotaxic surgery and viral vectors

Stereotaxic surgery was conducted as previously described (22). For circuit-tagging experiments (Fig. 1F), retrograde Cre vector (HSV-hEf1α-Cre; 0.5 μl; Gene Delivery Core, Massachusetts General Hospital) was infused into NAc [+1.6 anterior/posterior (AP), ±1.5ML, −4.4 dorsal/ventral (DV)] relative to bregma, 10° angle). For CRISPR experiments, retrograde Cas9 vector (HSV-hEf1α-LS1L-myc-Cas9; 0.5 μl; Gene Delivery Core, Massachusetts General Hospital) was infused into NAc. After 3 weeks, control vector (HSV-IE4/5-TB-eYFP-CMV-IRES-Cre; Gene Delivery Core, Massachusetts General Hospital) or FosB gRNA vector (HSV-IE4/5-TB-FosB gRNA-CMV-eYFP-IRES-CRE; Gene Delivery Core, Massachusetts General Hospital) was infused into the ventral CA1 region of vHPC [vCA1; −3.4AP, ±3.2 medial/lateral (ML), −4.8DV relative to bregma, 3° angle; 0.5 μl]. For DREADD experiments, retrograde Cre vector was infused into NAc and control vector (AAV2-CMV-GFP; 0.5 μl; UNC Vector Core), DREADD G_q vector [AAV2-hSyn-DIO-hM3D(G_q)-mCherry; 0.5 μl; Addgene], or DREADD G_i vector [AAV2-hSyn-DIO-hM4D(G_i)-mCherry; 0.5 μl; Addgene] was infused into the vCA1. For calreticulin overexpression experiments, mCalr vector (AAV2-CMV-mCalr-SV40p-eGFP; 0.5 μl; Vector Biolabs) was infused into vCA1. For circuit-specific calreticulin knockout experiments, retrograde FlpO vector (AAVrg-Ef1a-FlpO; 0.5 μl; Addgene) was infused into NAc, and a Flp-dependent Cre virus (AAV8-EF1a-fDIO-Cre; 0.5 μl; Addgene) was infused into vCA1. All behavioral and electrophysiological procedures commenced 3 to 6 weeks following surgeries.

Behavioral testing

Behavioral data were collected using an infrared–charge-coupled device camera (Panasonic) and analyzed using automated video tracking software (CleverSys). Animals were transported to the behavioral testing rooms in their homecages and allowed 30 min to habituate to the room.

Experimenter-administered cocaine

Cocaine hydrochloride (dissolved in 0.9% sterile saline) was injected intraperitoneally into mice daily in their homecages for 10 days.

Cocaine self-administration

Cocaine self-administration was conducted as previously described (72). Briefly, mice were trained and tested for seeking in a mouse operant chamber equipped for self-administration (Med-Associates Inc.), including an automated syringe pump, commutator and infusion tubing, magazine, house light, and left and right nosepokes with cue lights within the nosepoke hole. Mice received jugular catheter surgeries and, 4 days following surgery, were trained de novo to self-administer intravenous infusions of cocaine or saline (0.5 mg/kg per infusion) across 14 daily 2-hour sessions. House and cue lights remained on until a successful FR1 ratio nosepoke response in the active nosepoke hole (randomly assigned left or right). This was immediately followed by an infusion (including the sound of the syringe pump) and a 20-s timeout period of the house and cue lights. Nosepoke responses on the inactive nosepoke had no consequences. Responses were limited to 75 responses (total cocaine, 37.5 mg/kg) in each session. For the abstinence-induced seeking test, animals were left in their homecages for 6 days after the last administration session and then were tested again in the same operant chambers. The seeking test was the same as administration session except: (i) nosepoke responses on the active nosepoke resulted in no infusion but the sound of the pump (e.g., “mock infusion”); and (ii) session duration was 1 hour.

Cocaine CPP

CPP was conducted as previously described (24). Briefly, mice were tested for cocaine CPP in a three-chamber CPP box (San Diego Instruments). On day 1, mice received a pretest (no cocaine or saline) during which they were allowed to explore the entire box for 15 to 20 min. On days 2 to 3, mice received injections of saline paired with one chamber in the morning for 30 min and injections of cocaine (5 mg/kg, ip) paired with the opposite chamber for 30 min in the afternoon (chamber counterbalanced by group). On day 4, mice were again tested in a posttest (no cocaine or saline) and allowed to freely explore the entire box.

Chemogenetic experiments

G_q, G_i, and control mice were treated daily with CNO (0.3 mg/kg) dissolved in 5% dimethyl sulfoxide. Mice received twice-daily injections of CNO (in the morning and afternoon). Before CPP, CNO was injected and mice were left in their homecages. During CPP, CNO was injected 20 to 30 min before testing and conditioning.

Immunohistochemistry

Immunofluorescent analysis was conducted as previously described (22). Mice were transcardially perfused with cold phosphate-buffered saline (PBS), followed by 10% formalin. Brains from all immunostaining experiments were postfixed 24 hours in 10% formalin, cryopreserved in 30% sucrose, and sliced frozen on an SM2010R microtome (Leica) into 35-μm sections. Immunohistochemistry was performed using primary antibodies against FosB (1:1000; ab11959, Abcam), calreticulin (1:2000; ab92516, Abcam), and GFP (1:1000 to 1:5000; ab5450, Abcam); and secondary antibodies (1:200; Jackson ImmunoResearch) were conjugated to fluorescent markers (Alexa Fluor 488; Cy3; Cy5). Fluorescent images were visualized on an Olympus FluoView 1000 filter–based laser scanning confocal microscope or a Nikon Eclipse Ni-U Upright Fluorescent Microscope. Cell counts and intensity of signal in individual cells was quantified using ImageJ [National Institutes of Health (NIH)] software by an experimenter blinded to conditions.

Reverse transcription PCR

RNA extraction and reverse transcription PCR were conducted, as previously described (22). Mice were euthanized, and fresh brains were immediately dissected into 1-mm coronal sections. vHPC tissue was collected using 14-gauge biopsy punches guided by a fluorescent dissecting microscope (Leica) and stored at −80°C until processing. RNA was isolated from vHPC tissue using TRIzol (Invitrogen) homogenization and chloroform layer separation. The clear RNA layer was then processed (RNeasy Micro Kit; QIAGEN, no. 74004) and analyzed with NanoDrop. A volume of 10 μl of RNA was reverse transcribed to cDNA (High Capacity cDNA Reverse Transcription Kits; Applied BioSystems, no. 4368814). Before quantitative PCR (qPCR), cDNA was diluted to 200 μl. The reaction mixture consisted of 10 μl of Power SYBR Green PCR Master Mix (Applied Biosystems, no. 436759), 2 μl each of forward and reverse primers and water, and 4 μl of cDNA template. Samples were then heated to 95°C for 10 min (step 1) followed by 40 cycles of 95°C for 15 s, 60°C for 15 s, and 72°C for 15 s (step 2) and 95°C for 15 s, 60°C for 15 s, 65°C for 5 s, and 95°C for 5 s (step 3). Analyses were carried out using the ΔΔC(t) method (73). Samples were normalized to Gapdh.

Calr

Forward: 5′ GAA TAC AAG GGC GAG TGG AA 3′

Reverse: 5′ GGG GGA GTA TTC AGG GTT GT 3′

Western blotting

Western blotting analysis was conducted, as previously described (24). Mice were euthanized, and fresh brains were immediately dissected into 1-mm coronal sections. vHPC tissue was collected using 14-gauge biopsy punches guided by a fluorescent dissecting microscope (Leica) and stored at −80°C until processing. vHPC tissue punch samples were processed for SDS–polyacrylamide gel electrophoresis and transferred to polyvinylidene difluoride membranes for Western blotting with chemiluminescence. Blots were probed with calreticulin (1:1000; ab92516, Abcam) and then stripped and reprobed for Gapdh (14C10; 1:20,000; 2118, Cell Signaling Technology). Protein was quantified relative to Gapdh using ImageJ (NIH) software.

TRAP and cDNA library preparation

TRAP was conducted, as previously described (22). Three weeks following injection of retrograde HSV-Cre into NAc, Cre-dependent L10-GFP–expressing mice (Rosa26^eGFP/L10a) were euthanized, and fresh brains were immediately dissected into 1-mm coronal sections. vHPC tissue was collected using 14-gauge biopsy punches guided by a fluorescent dissecting microscope (Leica) and stored at −80°C until processing (n = 19 to 20 punches per group, three to four mice pooled per group). Polyribosome-associated RNA was affinity purified by homogenizing vHPC tissue in ice-cold tissue-lysis buffer [20 mM Hepes (pH 7.4), 150 mM KCl, 10 mM MgCl₂, 0.5 mM dithiothreitol, cycloheximide (100 μg/ml), protease inhibitors, and recombinant RNase inhibitors] using a motor-driven Teflon glass homogenizer. Homogenates were centrifuged for 10 min at 2000g (4°C), and supernatant was supplemented with 1% NP-40 (AG Scientific, no. P1505) and 30 mM dHPC (Avanti Polar Lipids, no. 850306P) and centrifuged again for 10 min at 20,000g (4°C). Supernatant was collected and incubated with Streptavidin MyOne T1 Dynabeads (Invitrogen, no. 65601) that were coated with anti-GFP antibodies (Memorial Sloan-Kettering Monoclonal Antibody Facility; clone names, Htz-GFP-19F7 and Htz-GFP-19C8; 50 μg per antibody per sample) using recombinant biotinylated Protein L (Thermo Fisher Scientific, no. 29997) for 16 to 18 hours on a rotator (4°C) in low-salt buffer [20 mM Hepes (pH 7.4), 350 mM KCl, 1% NP-40, 0.5 mM dithiothreitol, and cycloheximide (100 μg/ml)]. Beads were isolated and washed with high-salt buffer [20 mM Hepes (pH 7.4), 350 mM KCl, 1% NP-40, 0.5 mM dithiothreitol, and cycloheximide (100 μg/ml)], and RNA was purified using the RNeasy Micro Kit (QIAGEN, no. 74004). To increase yield, each RNA sample was initially passed through the QIAGEN MinElute column three times. Following purification, RNA was quantified using a Qubit fluorometer (Invitrogen), and RNA quality was analyzed using a 4200 Agilent TapeStation (Agilent Technologies). cDNA libraries from 5 ng of total RNA were prepared using the SMARTer Stranded Total RNA-Seq Kit (Takara Bio, USA, no. 635005), according to manufacturer’s instructions. cDNA libraries were pooled following Qubit measurement and TapeStation analysis, with a final concentration of 4 nM.

Sequencing

Sequencing was performed at the Icahn School of Medicine at Mount Sinai Genomics Core Facility (https://icahn.mssm.edu/research/genomics/core-facility). Raw sequencing reads from mice were mapped to mm10 using HISAT2. Counts of reads mapping to genes were obtained using featureCounts software of Subread package against Ensembl v90 annotation. Differential expression was conducted using the DESeq2 package (74). Sequencing data have been deposited into the GEO database (accession number GSE281142).

Chromatin immunoprecipitation–PCR

Brain tissues were suspended in PBS and disrupted by passing them through 25-gauge needles. The disrupted tissues were then fixed with 2 mM ethylene glycol bis(succinimidyl succinate) (Thermo Fisher Scientific) for 1 hour, followed by treatment with 1% formaldehyde for 15 min and 0.125 M glycine for 5 min to quench the reaction. The cells were lysed in a solution containing 1% SDS, 10 mM EDTA, and 50 mM tris-HCl (pH 8.0). The DNA was fragmented into ~200– to 400–base pair lengths using sonication (Branson Sonifier 450). Immunoprecipitation was carried out overnight at 4°C using 5 μl of rabbit polyclonal FosB antibody (lot 3; 2251S, Cell Signaling Technology) or 2 μg of rabbit polyclonal IgG isotype (3900S, Cell Signaling Technology). Antibody-bound chromatin fragments were isolated with protein G plus/protein A agarose beads (MilliporeSigma), followed by washing and elution. After the immunoprecipitated and input chromatin fragments were reverse cross-linked, the DNA was extracted using phenol/chloroform and subsequently precipitated.

FosB occupancy at the Calr promoter was analyzed using quantitative real-time PCR on the C384 Real-Time System (Bio-Rad). The primer sequences used were GACCTACAGCTGTCCCTTTC (forward) and CCCATAGTGCGACCAATAGAA (reverse). Binding enrichment was quantified as the percentage of antibody (IgG isotype or FosB) immunoprecipitated DNA fragments relative to the input.

Electrophysiology

Whole-cell, ex vivo slice electrophysiology was conducted, as previously described (22, 23). All solutions were bubbled with 95% O₂–5% CO₂ throughout the procedure. Mice were anesthetized with isoflurane anesthesia and transcardially perfused with sucrose artificial cerebrospinal fluid (aCSF; 234 mM sucrose, 2.5 mM KCl, 1.25 mM NaH₂PO₄, 10 mM MgSO₄, 0.5 mM CaCl, 26 mM NaHCO₃, and 11 mM glucose). Brains were rapidly removed, blocked, and placed in cold sucrose aCSF. Coronal sections (250 μM) containing vHPC were cut on a vibratome (Leica) and transferred to an incubation chamber containing aCSF (126 mM NaCl, 2.5 mM KCl, 1.25 mM NaH₂PO₄, 2 mM MgCl, 2 mM CaCl, 26 mM NaHCO₃, and 10 mM glucose) held at 34°C for 30 min before moving to aCSF at room temperature until used for recordings. Recordings were made from a submersion chamber perfused with aCSF (2 ml/min) held at 32°C. Borosilicate glass electrodes (3 to 6 MΩ) were filled with K-gluconate internal solution (115 mM potassium gluconate, 20 mM KCl, 1.5 mM MgCl, 10 mM phosphocreatine-tris, 2 mM MgATP, and 0.5 mM Na₃GTP; pH 7.2 to 7.4; 280 to 285 mosmol). GFP-positive cells in the ventral CA1 region of HPC were visualized using an Olympus BX51WI microscope using differential interference contrast infrared and epifluorescent illumination. Whole-cell patch-clamp recordings were made from transfected cells using a MultiClamp 700B amplifier and Digidata 1440A digitizer (Molecular Devices), and whole-cell junction potential was not corrected. Traces were sampled (10 kHz), filtered (10 kHz), and digitally stored. Cells with membrane potential more positive than −50 mV or series resistance of >20 MΩ were omitted from analysis. Rheobase was measured by giving brief (250 ms) depolarizing (0 to 300 pA, Δ5-pA steps) steps with 10-s step intervals. Elicited spike number was measured by issuing increasing depolarizing steps (0 to 300 pA, Δ25-pA steps, 500 ms) with 30-s step intervals. All electrophysiology recordings were made at ~30° to 32°C by warming the aCSF line with a single inline heater (Warner Instruments).

Statistical analyses

We used one-way and two-way analyses of variance (ANOVAs; including repeated measures or mixed effects as appropriate) followed by Holm-Šidák corrected post hoc comparisons in the case of significant omnibus effects and by independent samples t tests. Alpha criterion was set to 0.05. To ensure reproducibility, separate cohorts were included in behavioral testing.

Supplementary Materials

This PDF file includes:

Tables S1 to S5

Figs. S1 to S8