Abstract
The initiation of abstinence after chronic drug self-administration is stressful. Cocaine-seeking behavior on the first day of the absence of the expected drug (Extinction Day 1, ED1) is reduced by blocking 5-HT signaling in dorsal hippocampal cornu ammonis 1 (CA1) in both male and female rats. We hypothesized that the experience of ED1 can substantially influence later relapse behavior and that dorsal raphe (DR) serotonin (5-HT) input to CA1 may be involved. We inhibited 5-HT1A/1B receptors (WAY-100635 plus GR-127935), or DR input (chemogenetics), in CA1 on ED1 to test the role of this pathway on cocaine-seeking persistence 2 weeks later. We also inhibited 5-HT1A or 5-HT1B receptors in CA1 during conditioned place preference (CPP) for cocaine, to examine mechanisms involved in the persistent effects of ED1 manipulations. Inhibition of DR inputs, or 5-HT1A/1B signaling, in CA1 decreased drug seeking on ED1 and decreased cocaine seeking 2 weeks later revealing that 5-HT signaling in CA1 during ED1 contributes to persistent drug seeking during abstinence. In addition, 5-HT1B antagonism alone transiently decreased drug-associated memory performance when given prior to a CPP test, whereas similar antagonism of 5-HT1A alone had no such effect but blocked CPP retrieval on a test 24 h later. These CPP findings are consistent with prior work showing that DR inputs to CA1 augment recall of the drug-associated context and drug seeking via 5-HT1B receptors and prevent consolidation of the updated nondrug context via 5-HT1A receptors. Thus, treatments that modulate 5-HT–dependent memory mechanisms in CA1 during initial abstinence may facilitate later maintenance of abstinence.
Significance Statement
One of the most complex issues associated with substance use disorder treatment is the development of strategies that assist in the initiation and maintenance of abstinence from drug taking. Herein, we show that the persistent propensity for cocaine-seeking during abstinence can be attenuated by 5-HT1A/5-HT1B receptor antagonists or by inhibition of dorsal raphe signaling to the dorsal hippocampus. Our results provide insight into potential clinical applications for the treatment of addiction in human populations.
Introduction
Most preclinical research examining sex differences in cocaine addiction has focused on ongoing cocaine self-administration behaviors or reinstatement [see Lynch et al. (2002) for review]. We and others have shown that females exhibit greater drug-seeking behavior than males during early cocaine withdrawal (Fuchs et al., 2005; Kippin et al., 2005; Feltenstein et al., 2011; Kohtz and Aston-Jones, 2017), which is independent of estrous cycle phase (Feltenstein and See, 2007; Kerstetter et al., 2008) and is exacerbated by the anxiogenic drug yohimbine (Kupferschmidt et al., 2009). Cocaine withdrawal-related anxiety is highest during the first 2 d of drug absence (Coffey et al., 2000), and craving during this initial abstinence period predicts later relapse propensity (Hall et al., 1991). Therefore, initial drug abstinence may be a critical time point for understanding the factors that contribute to later relapse.
Prior reports implicate corticotropin-releasing factor 1, norepinephrine, and serotonin signaling (5-HT) in driving cocaine-seeking behavior during initial abstinence as seen on Extinction Day 1 (ED1; Cason et al., 2016; Kohtz and Aston-Jones, 2017). These studies also show that cocaine seeking on ED1 is marked by increased Fos expression in locus ceruleus, dorsal raphe (DR), median raphe, and dorsal hippocampal cornu ammonis (CA) 1, CA3, and dentate gyrus. In addition, the antagonism of β-adrenergic receptors in CA1 decreased ED1 cocaine seeking in females only, whereas blocking 5-HT1 receptors in CA1 substantially decreased ED1 drug seeking in both sexes (Kohtz and Aston-Jones, 2017). However, the effects of manipulating these circuits during ED1 on later cocaine seeking remain unknown.
Our study employed a novel approach that focused not only on cocaine seeking during ED1 but also on the persistence of drug seeking following 2 weeks of abstinence. This seeking-persistence paradigm can be used to study either transient (during ED1) or persistent effects (longer abstinence) of manipulating ED1 on cocaine-seeking behaviors. We focused on serotonin signaling in the present study, given prior reports of its role in both sexes to modulate cocaine seeking.
Materials and Methods
All experiments were conducted in accordance with procedures established by the Rutgers University Institutional Animal Care and Use Committee and the Guide for the Care and Use of Laboratory Animals (National Research Council).
Jugular catheter surgeries
Adult male and female Sprague Dawley rats were anesthetized with ketamine/xylazine (56.5/8.7 mg/kg) and given Rimadyl (1.0 mg/kg) as an analgesic. Chronic intravenous catheters were constructed in-house and inserted as described previously (Smith et al., 2009). Rats were given cefazolin (10 mg, i.v.) and heparin (10 U, i.v.) starting 3 d following surgery and daily following self-administration sessions. Rats recovered from surgery for at least 1 week before self-administration training.
Dorsal hippocampal cannulae and virus injections
Immediately following the jugular catheter surgery, rats were placed in a stereotaxic apparatus and bilaterally infused with 1.0 µl per side of AAV8-hSyn-hM4Di-mCherry targeted to DR via a glass micropipette (20 µm tip; males, AP, −7.8; ML, ±3.1; DV, −7.5; 30°; females, AP, −7.6; ML, ±3.1; DV, −7.2; 20°), and bilateral guide cannulae were implanted into CA1 (AP, −3.0; ML, ±2.0; DV, −2.5; no angle). Rats in nondesigner receptors exclusively activated by designer drug (DREADD) experiments received jugular catheters and bilateral CA1 cannulae alone. Dorsal control cannulae were implanted into the retrosplenial cortex (intra-RSC; AP, −3.0; ML, ±2.0; DV, −1.5), 2 mm dorsal to CA1 target sites. Cannula placements and DREADD expression were confirmed for each rat following the completion of behavioral analysis (details below).
Drugs
Cocaine hydrochloride (National Institute on Drug Abuse) was dissolved in 0.9% sterile saline. WAY-100635 (Tocris) and GR-127935 (Tocris) were dissolved in 0.9% sterile saline and administered intrahippocampally to CA1 (intra-CA1, bilateral; 1 nmol/1.0 µl/side) or to dorsal control site regions (intra-RSC, bilateral; 1 nmol/1.0 µl/side). Anisomycin (Tocris) was dissolved in 1 N HCl and brought to pH 7.2 with 1 N NaOH and to a final concentration of 125 mg/ml with sterile aCSF (Alberini, 2008; Qi and Gold, 2009). The DREADD agonist clozapine N-oxide (CNO; RTI International) was dissolved in sterile aCSF, and 1.0 mM was administered in a volume of 1 µl per side intra-CA1 to stimulate DREADDs on terminals transported from DR. All intra-CA1 infusions were at a rate of 0.25 µl/min, 10 min before testing (Kohtz and Aston-Jones, 2017).
Cocaine self-administration
Self-administration sessions were conducted in standard operant chambers housed in sound-attenuating cubicles and controlled via MED-PC IV software (Med Associates) as described previously (Smith and Aston-Jones, 2011). Rats were trained in daily 2 h sessions to press an active lever for intravenous cocaine for at least 10 d on a fixed ratio 1 (FR1) schedule of reinforcement to reach a criterion of >10 cocaine infusions/day (Kohtz and Aston-Jones, 2017). We used different cocaine doses in males (0.20 mg cocaine/infusion) and females (0.16 mg cocaine/infusion) to account for differences in body weight (∼200 g/female, ∼450 g/male) and obtain a similar number of cocaine–cue pairings (Fig. 1A; infusions), as in our previous publication (Kohtz and Aston-Jones, 2017). Each cocaine infusion was followed by a 20 s time-out in which lever pressing produced neither cocaine nor cues. An inactive lever was also present; presses on it were tabulated but had no consequence.
Sex differences in cocaine self-administration. A, Cocaine self-administration by male (0.20 mg/infusion) and female (0.16 mg/infusion) rats. Different doses were used for males and females to produce a similar number of cocaine–cue pairings (infusions) during conditioning (Kohtz and Aston-Jones, 2017). The asterisk indicates the day in which female lever pressing was significantly greater than that of males, p < 0.05. B, Daily cocaine intake by milligram/kilogram body weight of male and female rats. The asterisks indicate the days in which female overall intake is greater than that of males, p < 0.05.
Cocaine seeking during ED1
Twenty-four hours after the final self-administration session, rats were tested in an initial extinction session (ED1). In this session, rats were exposed to the self-administration chamber for 90 min, during which time presses on either lever had no consequence. Active-lever pressing in the absence of cocaine reward or cues on ED1 was taken as a measure of cocaine seeking during early abstinence, as previously reported (Feltenstein et al., 2011; Cason et al., 2016; Kohtz and Aston-Jones, 2017; Kohtz et al., 2018).
Cocaine-seeking persistence
Following the ED1 test, rats were returned to the homecage for 14 d of abstinence. Then, rats were returned to the context as in ED1 procedures, for a second extinction session (ED2) on Withdrawal Day 15 (WD15). Active-lever pressing during this second session was taken as a measure of cocaine-seeking persistence following abstinence.
Conditioned place preference
We used a standard two-chamber, balanced design for conditioned place preference (CPP). Rats were habituated and assessed for a side preference prior to conditioning acquisition. The cocaine-paired chamber was counterbalanced between rats. Two conditioning sessions (one each for saline and cocaine) occurred each day for 3 d consecutively prior to testing; injections of saline or cocaine alternated between morning and afternoon sessions during the CPP acquisition phase. CPP acquisition and testing were identical to our previously published reports (Harris and Aston-Jones, 2003a, b).
Novel object recognition task
We used a clear acrylic chamber (∼40 × 40 × 30 cm) to test novel object recognition (nOR) performance. Rats were habituated to the box for 30 min 1 d prior to training. On the training day, rats were exposed to two identical objects and allowed to ad libitum explore for 5 min. Time spent with each object was recorded to determine if rats favored one side of the box. Twenty-four hours later, one object was replaced with a novel object, and rats were returned to the box. The percentage of time spent interacting with the novel object indicated object memory performance.
Immunohistochemistry
Immunohistochemistry was used to confirm the localization of the AAV8-synapsin-hM4Di-mCherry receptor expression in cell bodies of the DR and in processes in CA1. We performed staining for mCherry and 5-HT on sequential sections. We identified the DR region with 5-HT antibody staining and used this to generate a region of interest for analyzing mCherry-stained neurons. The DR contains many fewer neurons that project to the dHPC CA1 compared with other 5-HT–containing nuclei, such as the median raphe (Vertes et al., 1999; Luchetti et al., 2020). However, unlike the median raphe wherein ∼25% of all projections to dHPC CA1 are 5-HT negative, <5% of dHPC projecting DR neurons are 5-HT negative (Köhler and Steinbusch, 1982). Therefore, combined with the use of CNO microinjection into CA1, this method effectively targets DR 5-HT projections to dHPC CA1. Rats with >30% of DREADD expression outside DR, as determined by 5-HT staining on alternate sections, were excluded from analyses (n = 4). Six rats died within 1 week of surgery, which we attributed to the puncture of the cerebral aqueduct. hM4Di expression was robust in DR (Fig. 4B) in the remaining rats (n = 28). Immunohistochemistry for mCherry (Mahler et al., 2014) and 5-HT (Kohtz and Aston-Jones, 2017) were performed per previously reported methods.
Experimental design and statistical analysis
Independent t tests, Pearson's R correlations, one- or two-way analysis of variance (ANOVA), or three-way multiple ANOVAs were used to compare differences between groups in responding. Three-way repeated-measure ANOVAs were used to compare treatment (vehicle vs WAY plus GR in pharmacology experiments or vehicle vs CNO in DREADD experiments) and sex (male, female) across test days (ED1 on WD1 vs ED2 on WD15) for cocaine-seeking behaviors. Two-way ANOVAs were used to determine the effects of treatment on cued-reinstatement tests, similarly comparing the effects of WAY plus GR treatment between males and females. One-way ANOVAs were used to determine the effects of WAY and/or GR treatments on drug memory acquisition or expression in the CPP paradigm, in males only. Significant ANOVA effects were followed by post hoc pairwise comparisons using Student's t test. Respective p values were Bonferroni’s corrected for multiple comparisons across statistical tests.
Results
Cocaine self-administration in male and female rats
Consistent with our prior report (Kohtz and Aston-Jones, 2017), there was no overall difference between the numbers of cocaine infusions for males versus females during FR1 (although females had significantly greater lever pressing behavior on Day 5; post hoc, df = 18; t = 3.710; p = 0.0017; Fig. 1A). Cocaine infusions escalated over 10 d of self-administration (F(9,198) = 5.484; p < 0.0001) independent of sex. As in prior reports, females had more overall cocaine consumption by body weight during FR1 (mg/kg/day; main effect of sex, F(1,22) = 10.29; p = 0.0041); this was only significant on Days 4–7 (t = 3.051; t = 3.978; t = 3.563; t = 3.210; p < 0.05; Fig. 1B). Importantly, these data confirm that the doses used in males and females during self-administration produced similar numbers of cocaine–cue pairings (Fig. 1A; infusions), a critical factor for comparing lever pressing during extinction tests.
5-HT1 receptor signaling in the dorsal hippocampus on ED1 facilitates persistent cocaine seeking
Intra-CA1 WAY plus GR attenuates persistent cocaine seeking
Following 10 d of >10 cocaine self-administration infusions per day, rats (n = 6–9/group) were returned to the operant chamber without cues or cocaine reward for ED1 testing. Rats were given an intrahippocampal infusion of saline or a cocktail of WAY-100635 plus GR-127935 (5-HT1A and 5-HT1B antagonists, respectively, WAY plus GR; 1.0 nmol each/1.0 µl/side) 10 min prior to entering the operant chamber. As previously reported, there was a sex difference in the magnitude of ED1 cocaine seeking, wherein saline-treated females had greater ED1 responding than those of saline-treated males (F(1,24) = 6.492; p = 0.0177; Fig. 2B; Kohtz and Aston-Jones, 2017). As in our prior report (Kohtz and Aston-Jones, 2017), we found that WAY plus GR significantly decreased cocaine-seeking behavior on ED1 in males and, to a greater extent, in females (F(1,24) = 28.09; p < 0.0001; Fig. 2B). That report also found that WAY plus GR infusions reduced active-lever pressing without inducing simple locomotor effects.
5-HT1A/1B receptor antagonism in dorsal hippocampal CA1 persistently reduced cocaine seeking in both sexes. A, Timeline of experimental conditions for each rat tested in panels B and C. B, Administration of WAY plus GR to dorsal CA1 10 min prior to ED1 testing decreased cocaine seeking in both sexes on ED1 and persistently decreased seeking when tested 2 weeks later on WD15 (ED2, with no additional injections). The asterisks indicate the significant differences from saline-administered controls. All CA1 injections were made prior to ED1 only. C, Survival curves for extinction behavior following ED2 testing depicted in panel B. Although females had significantly less ED2 cocaine seeking compared with seeking on ED1 and males do not differ in their seeking between ED1 and ED2 (as shown in panel B), it took longer for saline-treated females to meet extinction criteria of <25 lever presses for 3 consecutive days compared with males. Intervention with WAY plus GR on ED1 significantly reduced extinction resistance in females but not in males. D, Four groups of rats received different injections into CA1 either before or immediately after ED1 testing (see Results for description of groups and timelines of injections into CA1). All rats underwent ED1 testing and 2 weeks of abstinence as in prior experiments. The efficacy of interventions was analyzed at ED2. Administration of WAY plus GR post-ED1 had similar efficacy as administration pre-ED1 on reducing cocaine-seeking behavior tested 2 weeks later. To test if WAY plus GR attenuation of cocaine seeking was dependent on protein synthesis during the consolidation window following ED1 testing, we administered WAY plus GR pre-ED1 and anisomycin (protein synthesis inhibitor) post-ED1. The WAY plus GR–induced reduction in cocaine-seeking persistence was blocked by a post-ED1 injection of anisomycin. Thus, protein synthesis during the post-ED1 consolidation window was required for WAY plus GR's effects on cocaine-seeking persistence. †† indicate significant differences from saline controls, p < 0.05. * indicates the significant difference between WAY plus GR pre-ED1 and WAY plus GR pre-ED1 plus anisomycin post-ED1.
In the same group of rats, we then investigated if a single dose of the WAY plus GR cocktail prior to context retrieval had long-term effects on cocaine-seeking behavior. Rats were returned to their homecages for 2 weeks of forced abstinence. On WD15, rats were returned to the operant chamber for a second time without drugs or cues to test cocaine-seeking persistence (ED2), with no further hippocampal or other injections. Across multiple cohorts, rats given saline on ED1 exhibited cocaine-seeking behavior that strongly predicted persistent cocaine seeking tested 2 weeks later on ED2 [females, R2 = 0.42; p = 0.0124; males, R2 = 0.53; p = 0.0022; Fig. 3A includes unpublished findings from rats in Kohtz and Aston-Jones (2017)]. We used a three-way repeated-measure ANOVA to compare sex and treatment (WAY plus GR vs vehicle) over the ED1 and ED2 test days. Females had greater ED1 seeking, driving a main effect of sex (F(1,48) = 5.800; p = 0.0199). Intervention with intra-CA1 WAY plus GR on ED1 decreased cocaine seeking on both ED1 and ED2, because there was a main effect of treatment (F(1,48) = 48.650; p < 0.0001) but not of test day, showing that attenuation of 5-HT signaling on ED1 persistently decreased drug seeking (including at 2 weeks later at ED2) in both sexes (Fig. 2B).
5-HT1A/B receptor antagonism is ineffectual when administered to the retrosplenial cortex. A, Active-lever presses (cocaine seeking) on ED1 predicts active-lever pressing behavior on ED2 (after 2 weeks of abstinence) in both sexes. This figure incorporates rats across multiple cohorts including animals from this study (n = 14) and our prior study (Kohtz and Aston-Jones, 2017; n = 16) and indicates that this effect is highly consistent. B, Rats underwent FR1 training and were administered WAY plus GR on ED1 as shown in Figure 1A targeting 2 mm dorsal to hippocampal CA1 (into the retrosplenial cortex, intra-RSC). Administration of WAY plus GR intra-RSC had no effect on ED1 cocaine seeking or on subsequent cocaine seeking tested 2 weeks later. Females had significantly greater cocaine-seeking behaviors on ED1 compared with those in males (p < 0.05). C, Plots of the mean (+SEM) daily active-lever presses during extinction training following WAY plus GR administration on ED1. Rats underwent extinction training for at least 7 d and until they met 3+ days of <25 lever presses. There was a main effect of WAY plus GR on ED1 to decrease average lever pressing during extinction in females compared with vehicle controls (post hoc t, Day 3, t = 2.526; Day 4, t = 2.124 NS; Day 5, t = 3.563; Day 6, t = 3.226; Day 7, t = 3.767; df = 11; p = 0.0030–0.0328). D, WAY plus GR microinfusion into dorsal CA1 on ED1 had no effect on cued-reinstatement behavior following extinction. However, there was a significant main effect of sex wherein females had greater cued-reinstatement behavior than males. The asterisk indicates the significant difference from males, p < 0.05. E, Rats underwent FR1 training as in other experiments and were administered WAY plus GR on WD1 in the homecage (without ED1 testing in the self-administration chamber). Cocaine-seeking behavior was tested at WD15. Administration of WAY plus GR to dorsal CA1 on WD1 in the homecage had no effect on subsequent cocaine seeking tested 2 weeks later (F(1,19) = 0.0604, NS).
No effect of dorsal control WAY plus GR on cocaine seeking
In a separate group, rats (n = 12) were bilaterally implanted with cannulae targeted to the retrosplenial cortex, 2 mm dorsal to CA1, and along the same cannula trajectory as for CA1 injections described above. In these dorsal controls, administration of WAY plus GR intra-RSC had no effect on ED1 cocaine seeking or on persistent cocaine seeking tested 2 weeks later (Fig. 3B), although females had greater cocaine-seeking behavior than did males (F(3,20) = 5.316; p = 0.0328).
Intra-CA1 WAY plus GR decreased resistance to extinction in females
Females are normally more resistant to extinction than males (Lynch and Carroll, 2000; Becker and Hu, 2008). Thus, we next investigated whether the single infusion of WAY plus GR on ED1 altered sex differences in extinction resistance. Responding on ED1 was greater in females compared with males, as expected; however, active-lever responding on ED2 was similar for males and females given saline on ED1 (t = 0.2829; p = 0.2016; Fig. 2B). Following ED2 testing, all rats were extinguished in the operant box for at least seven daily, 2 h sessions (no cues or cocaine) until the criterion of <25 active plus inactive lever presses was met for 3 consecutive days. Saline-treated females took longer to meet the extinction criterion than did saline-treated males (X2 = 9.513; p = 0.0020). Furthermore, intra-CA1 WAY plus GR on ED1 decreased extinction resistance in females but not in males [Fig. 2C; (X2 = 11.45, p = 0.0095)]. Across Days 3–7 of extinction, there was a main effect of WAY plus GR to decrease average lever pressing during extinction in females (Fig. 3C; F(3,24) = 6.055; p = 0.0032).
No effect of intra-CA1 WAY plus GR on cued reinstatement of cocaine seeking
We then investigated if the effects on context-induced cocaine seeking (i.e., seeking during the above ED1 and ED2 extinction tests) are generalized to reinstatement of cocaine seeking induced by cocaine-associated cues. First, rats that were extinguished in the above experiment were administered intra-CA1 saline 10 min prior to a 2 h cued-reinstatement session, wherein active-lever presses yielded cocaine-associated light plus tone cues but no cocaine. There were no significant effects of the prior ED1 treatment (saline vs WAY plus GR) on cued-reinstatement responding in either sex (Fig. 3D). Rats were then subjected to additional extinction training (criterion of <25 lever presses for 3 consecutive days) and tested for cued reinstatement responding 10 min following intra-CA1 WAY plus GR infusions. A two-way repeated-measure ANOVA comparing the effects of sex and acute treatment (saline vs WAY plus GR) found an overall effect of sex, wherein females had greater cued-reinstatement behavior than males (F(1,26) = 4.852, p = 0.0367), but no effect of WAY plus GR treatment on reinstatement test day on reinstatement behavior (Fig. 3D). These data indicate that 5-HT signaling in CA1 during ED1 selectively affects persistent seeking induced by the context during extinction (ED2) but not reinstatement in response light plus tone cues.
Decrease of persistent seeking by intra-CA1 WAY plus GR requires exposure to context
We next determined if exposure to the context, e.g., ED1, was required for attenuation of cocaine-seeking behavior observed 2 weeks following the WAY plus GR injection. Separate groups of rats (n = 11–12/sex) were trained to self-administer >10 cocaine infusions/day for 10 d, as described above. Twenty-four hours following the final session, rats were infused intra-CA1 with WAY plus GR and returned to their homecages without re-exposure to the test cage. Rats underwent 14 d of forced abstinence in their homecages and were tested in a delayed ED1 (on WD15) with no further manipulations. There was no effect of sex or WAY plus GR treatment on lever pressing in these rats (i.e., in the absence of context exposure; Fig. 3E), indicating that exposure to the context is necessary for WAY plus GR treatment to reduce persistent cocaine seeking.
Protein synthesis in CA1 is required for the attenuating effects of WAY plus GR on cocaine-seeking persistence
We next tested if the effects of WAY plus GR treatment to persistently decrease cocaine-seeking behavior were dependent on protein synthesis. Four new groups of male rats (n = 6–7/group) were trained to self-administer cocaine as described above. These rats were administered into CA1 either (1) saline prior to ED1 testing, (2) WAY plus GR prior to ED1 testing, (3) WAY pls GR immediately following ED1 testing, or (4) a combination of WAY plus GR prior to ED1 testing and anisomycin (a protein synthesis inhibitor) immediately following ED1 testing. The results (shown in Fig. 2D) indicated that the effect of WAY plus GR to reduce cocaine seeking persistently was dependent on hippocampal protein synthesis: using a one-way ANOVA (F(3,30) = 9.068; p = 0.0003), we compared the effects of intra-CA1 saline versus pre-WAY plus GR (administered prior to ED1 testing) versus post-WAY plus GR (administered post-ED1 testing) versus pre-WAY plus GR plus post-anisomycin. Cocaine seeking on ED1 was significantly decreased compared with the vehicle in rats administered pre-WAY plus GR only (F(3,30) = 4.541; p = 0.0106). Post hoc t test analyses showed that WAY plus GR administered into CA1 post-ED1 testing decreased cocaine seeking tested on ED2 compared with saline controls (df = 13; t = 2.681; p = 0.0189) indicating that post-ED1 injections have similar efficacy to pre-ED1 injections on cocaine-seeking persistence tested 2 weeks later. Furthermore, the effect of pre-WAY plus GR to decrease cocaine seeking on ED2 versus saline controls (df = 13; t = 2.512; p = 0.0260) was blocked by post-ED1 infusion of anisomycin (pre-WAY plus GR vs pre-WAY plus GR plus post-anisomycin; df = 12; t = 3.239; p = 0.0071), indicating that WAY plus GR's effects to persistently attenuate cocaine seeking may be dependent on protein synthesis.
DR input to the dorsal hippocampus is needed for persistent cocaine seeking
We previously showed that Fos induction in DR, but not in the median raphe, significantly correlated to ED1 drug-seeking behaviors in male and female rats (Kohtz and Aston-Jones, 2017). Thus, we next determined if projections from DR impacted persistent cocaine-seeking behaviors. Twenty-two rats received AAV8-hSyn-hM4D(Gi)-mCherry into DR and were implanted with bilateral cannulae in CA1 (see Materials and Methods, Dorsal hippocampal cannulae and virus injections). Rats began self-administration training 4–5 weeks later, to allow time for DREADD expression and transport. Twenty-four hours after the final self-administration session, rats (n = 6–8/group) were returned to the operant chamber for ED1 testing. Ten minutes prior to ED1 testing, rats were given an intrahippocampal infusion of aCSF (1.0 µl/side) or CNO (1.0 mM, 1.0 µl/side) to test the effects of inhibiting DR inputs to CA1. DREADD inhibition of DR inputs to CA1 replicated antagonist studies described above: aCSF-treated females had greater ED1 cocaine seeking than males (F(1,24) = 12.88; p = 0.0015). Intra-CA1 CNO in DREADD rats decreased cocaine-seeking behavior on ED1 in both sexes (F(1,24) = 26.28; p < 0.0001; Fig. 4E) and was more effective in females than in males (F(1,24) = 9.420; p = 0.0053). There was no effect of intra-CA1 CNO on cocaine seeking on ED1 in rats that did not receive the DREADD vector, indicating that the results were not due to nonspecific effects of CNO (Fig. 4F).
Chemogenetic inhibition of DR signaling in CA1 persistently reduced cocaine seeking in both sexes. A, A midsagittal view of the rat brain illustrating the pAAV8-hSyn-hM4Di(Gi)-mCherry infusion target region (DR) and CNO infusion location (dorsal hippocampal CA1, intra-CA1). B, Typical DREADD tag expression (mCherry, red) in DR of a rat that received an hSyn-hM4Di(Gi)-mCherry infusion 6 weeks before. Note that DREADD expression is largely contained within DR borders. C, Terminal expression of hM4Di in the hippocampus by detection of DREADD tag (mCherry, red). hM4Di fibers were present in the dentate gyrus, CA3 and CA1. Cell nuclei are counterstained with DAPI (blue). The panel at the right is the magnification of the area indicated in low power at the left. D, Behavioral timeline for testing performed in panels E and F. E, Microinjection of CNO (1.0 mM/1.0 µl/side) in dorsal CA1 to target hM4Di terminals from DR on Extinction Test 1 decreased responding on that test and also produced persistently decreased cocaine seeking (seen in Extinction Test 2, 14 d later) in both sexes. The asterisks indicate the significant differences from aCSF controls. F, Local CNO had no effect on ED1 or ED2 behavior in non-DREADD female controls. As no effects of CNO were observed in non-DREADD females (even though females had larger effects of CNO in subjects expressing DREADDs), this control manipulation was not repeated in non-DREADD male rats to reduce animal use. G, There was a main effect of sex wherein females had greater cued reinstatement than did males; however, there was no effect of intra-CA1 CNO on cued reinstatement in either sex. The asterisks indicate the significant differences from males, p < 0.05.
Cocaine-seeking persistence was tested with a second drug-seeking session (ED2) 2 weeks later (see Materials and Methods, Cocaine-seeking persistence). As above, we used a three-way repeated-measure ANOVA to compare sex and treatment with the ED1 and ED2 test days. Intervention with CNO on ED1 decreased cocaine-seeking behavior on both ED1 and ED2 because there was a main effect of treatment (F(1,48) = 53.716; p = 0.0001) but not of test day, showing that attenuation of DR signaling in CA1 on ED1 persistently decreased drug seeking in both sexes (Fig. 3E). There was no effect of intra-CA1 CNO on cocaine-seeking persistence in rats that did not receive the DREADD vector, indicating that results were not due to nonspecific effects of CNO (Fig. 4F).
Rats were then subjected to additional extinction training over several days until criteria were met and then tested for cued reinstatement of cocaine seeking. Females had greater cued reinstatement than males (F(1,18) = 14.00; p = 0.0367; Fig. 4G). Similar to our pharmacology manipulations, we found no effect of CNO on ED1 to influence cued-reinstatement behavior in either sex, further supporting the notion that attenuation of DR (5-HT) signaling in CA1 on ED1 reduces context-, but not cue-, induced cocaine seeking persistently.
Receptor signaling in the dorsal hippocampus is involved in drug context memory: CPP experiments
We speculated that the reduction in cocaine seeking on ED1 by blockade of DR inputs to CA1 reflected reduced recall of the drug-associated context. We also hypothesized that these treatments facilitated the consolidation of new information on ED1 (e.g., that the context was no longer associated with the drug) because there was a persistent decrease in drug seeking 2 weeks later. We investigated these hypotheses by testing the effects of intrahippocampal WAY and GR given together, or separately, in a CPP paradigm.
Naive groups of rats (7–8/group) were used to test the effects of 5-HT1A and 5-HT1B receptor antagonism in CA1 on CPP for cocaine. The WAY plus GR cocktail or saline control was microinfused intra-CA1 either 10 min prior to each of the three cocaine conditioning sessions in the CPP acquisition phase or in different rats 10 min prior to the (cocaine-free) CPP test session. Intra-CA1 WAY plus GR administered during CPP acquisition increased cocaine preference on the CPP test day in both sexes (main effect of treatment, F(1,25) = 13.84; p = 0.0010; Fig. 5A). In contrast, WAY plus GR given prior to the CPP test session produced a main effect to reduce preference similarly in both sexes (F(1,25) = 27.39; p < 0.0001; Fig. 5B). These data show that WAY plus GR enhanced the acquisition of cocaine CPP when given during acquisition but decreased expression of cocaine CPP when administered on test day.
Combined inhibition of 5-HT1A/1B receptors in CA1 modulated acquisition and recall of cocaine CPP. A, Timeline for testing the effects of WAY-100635 plus GR-127935 (5-HT1A and 5-HT1B antagonists, respectively, WAY plus GR) microinjection into CA1 prior to training (acquisition group) or prior to testing (retrieval group) on CPP. B, Cocaine preference when WAY plus GR was given into CA1 prior to CPP acquisition sessions versus prior to the CPP expression/retrieval test. There were no significant differences between saline-treated rats during acquisition or retrieval; therefore, these data were combined (n = 8/sex per control group). WAY plus GR given during acquisition increased CPP preference (n = 6–7/sex), and WAY plus GR given prior to the test session decreased CPP preference (n = 6–7/sex). The asterisks indicate the post hoc tests revealing significant differences between identified groups, p < 0.05.
Next, we investigated the actions of blocking 5-HT1A or 5-HT1B receptors separately by giving the individual compounds in two new groups of rats (n = 6–8/group). As the effects of the cocktail were similar in both sexes, we investigated the effects of the individual compounds during acquisition versus expression in naive males only to reduce animal usage. WAY (selective 5-HT1A antagonist) had significant effects on preference when given prior to CPP acquisition but not when given prior to the test session, compared with saline controls (overall ANOVA, F(2,19) = 10.52; p = 0.0011; Fig. 6B). Post hoc analyses showed that WAY administered prior to CPP acquisition significantly increased preference on the test day (t = 3.746; df = 17; p = 0.0036), whereas WAY administered on the test day had no effect compared with saline. When retested 24 h later (second CPP test), rats that had WAY administered on the preceding CPP test day showed significantly decreased preference compared with the prior day (paired t = 4.449; df = 5; p = 0.0067). These results indicate that 5-HT1A receptor signaling in CA1 interferes with learning an association between context and cocaine reward.
5-HT1A and 5-HT1B receptors in CA1 modulated the acquisition and recall of cocaine CPP, respectively. A, Timeline for testing the effects of WAY-100635 (WAY; 5-HT1A antagonist) or GR-127935 (5-HT1B antagonist) microinjection into CA1 prior to training days (acquisition group) or prior to testing day (retrieval group) on CPP. B, CPP performance after WAY microinjection into CA1. WAY given during CPP acquisition enhanced cocaine preference. WAY given 10 min prior to the CPP test/retrieval session produced no effects on cocaine preference on that test but strongly decreased preference recall when retested 24 h later. †† indicates the preference when WAY administered on acquisition was significantly different from saline control or WAY administered on retrieval. The asterisk indicates the significant differences between the retrieval test and retest 24 h later, p < 0.05. C, CPP performance after GR microinjection into CA1. GR given during CPP acquisition had no effect on cocaine preference. GR administration 10 min prior to the test/retrieval session significantly decreased preference but had no effect when these rats were retested 24 h later, indicating recovery of CPP memory. †† indicates the significant differences from GR administered on retrieval. The asterisk indicates the significant difference from the retrieval test, p < 0.05.
Next, we tested GR (5-HT1B antagonist) using the same paradigm in two additional groups of naive rats (n = 6–7/group). GR microinfused into CA1 prior to CPP acquisition had no effect on preference in the later test session, whereas GR given on the test day decreased cocaine preference (F(2,20) = 19.17; p < 0.0001; GR test day vs saline, t = 5.618; df = 18; p < 0.0001; GR test day vs GR acquisition, t = 5.049; df = 18; p = 0.0007; Fig. 6C). In contrast to the effects of WAY, these effects of GR were transient, as CPP on the 24 h retest was significantly greater than that on the prior day when animals received GR (paired t = 5.908; df = 6; p = 0.0010). This indicates that inhibition of 5-HT1B signaling attenuates recall, but not consolidation, of the cocaine-associated context.
Effect of 5-HT1A plus 5-HT1B receptor antagonist cocktail on nOR performance
We then sought to determine if the effects of WAY plus GR on drug-associated memories generalize to other memory types in naive rats (n = 10/group). For this, we tested the effects of giving the WAY plus GR cocktail into dorsal CA1 on nOR performance. Administering WAY plus GR 10 min prior to training in nOR resulted in increased nOR performance on the subsequent test (F(1,33) = 5.133; p = 0.0037), whereas instead administering WAY plus GR 10 min prior to nOR testing resulted in decreased nOR performance (F(1,35) = 25.10; p < 0.0001). Post hoc tests indicated that these effects were significant in females (acquisition vs saline, t = 2.798; df = 19; p = 0.0107; retrieval vs saline, t = 4.889; df = 19; p < 0.0002) and in males (retrieval vs saline, t = 2.555; df = 18; p = 0.0199; Fig. 7B). The raw time spent interacting with the objects (novel, familiar) did not significantly differ between experimental conditions (Fig. 7C). These data indicate that inhibition of 5-HT1A plus 5-HT1B signaling drives consolidation, and inhibits recall, of episodic memories (e.g., object discrimination), similar to effects observed with combined inhibition in CPP.
5-HT1 receptor signaling modulated acquisition and recall of nOR memories. A, Timeline for testing the effects of WAY-100635 plus GR-127935 (5-HT1A and 5-HT1B antagonists, respectively, WAY plus GR) microinjection into CA1 prior to training (acquisition group) or prior to testing (retrieval group) on nOR performance. B, nOR performance. There were no significant differences between saline-treated rats during acquisition or retrieval; therefore, these data were combined (n = 10/sex per control group). WAY plus GR given into dorsal CA1 during nOR acquisition (n = 10/sex) enhanced nOR performance in females only, whereas WAY plus GR given before the nOR retrieval test (n = 10/sex) reduced nOR performance in both males and females. †† indicates all the groups that are significantly different, p < 0.05. The asterisk indicates the significant difference between indicated groups, p < 0.05. C, Total time interacting with novel (colored bars) and familiar (gray bars) objects in the nOR task. Individual data points are plotted for novel object times only.
Discussion
Our study identified the role of serotonin signaling in dorsal hippocampal CA1 in the persistence of cocaine seeking and in drug-associated memories. We confirmed our prior finding that ED1 cocaine seeking is attenuated by 5-HT1A plus 5-HT1B antagonists in CA1 (Kohtz and Aston-Jones, 2017). We extended these findings to show that attenuating DR 5-HT signaling in CA1 on ED1 not only attenuates cocaine seeking in that session but also has persistent effects to reduce cocaine seeking 2 weeks later. We found that 5-HT antagonists must be given in the cocaine self-administration context to decrease persistent seeking. Furthermore, we demonstrated that the effects of 5-HT antagonist–dependent amelioration of cocaine-seeking persistence required protein synthesis. We also found specific effects of 5-HT1A versus 5-HT1B receptors on cocaine-associated contextual memory consolidation versus recall, respectively. Together, these results indicate that 5-HT signaling in CA1 contributes to memory processes that drive cocaine seeking evoked by drug-associated contexts and support prior reports defining a role of the dorsal hippocampus in context-, but not cue-, induced drug seeking (Fuchs et al., 2005; Crombag et al., 2008; Xie et al., 2010; Raybuck and Lattal, 2014; McGlinchey and Aston-Jones, 2018).
Sex differences in 5-HT signaling
Prior reports indicate that there are sex differences in hippocampal 5-HT signaling that may contribute to our finding that attenuation of 5-HT signaling in CA1 was more effective in females compared with males. Those studies reported sex differences in 5-HT fiber innervation and release, receptor expression, and 5-HT's roles in neurogenesis, depression, stress, and mood disorders (Zhang et al., 1999; Martinowich and Lu, 2008; Franceschelli et al., 2015). In particular, sex differences in hippocampal somatodendritic 5-HT1A receptors play a substantial role in corresponding stress responses (Haleem et al., 1990, 2007; Haleem and Parveen, 1994; Haleem, 2011). Indeed, systemic administration of a 5-HT1A agonist produces a twofold decrease of 5-HT release in the hippocampus of females compared with that in males (Haleem et al., 1990), indicating that there is greater 5-HT1A receptor signaling in females. Acute stressors increase 5-HT1A receptor sensitivity, thereby decreasing 5-HT availability and producing coping deficits (Haleem et al., 2007). Conversely, repeated predictable stress decreases 5-HT1A receptor sensitivity, increasing 5-HT availability and producing stress adaptation (Lopez et al., 1998). Thus, raphe–hippocampal 5-HT signaling contributes to stress adaptation in a sex-dependent manner, wherein females may be more sensitive to acute stressors, such as stress that occurs at the onset of drug abstinence (e.g., ED1; Kohtz and Aston-Jones, 2017).
Serotonin has diverse effects on memory processes
The dorsal hippocampus receives serotonergic inputs from both the median and DR nuclei (Vertes, 1991; Vertes et al., 1999). Mnemonic- and reward-mediating effects of 5-HT signaling in the dorsal hippocampus have been primarily attributed to innervation by median raphe 5-HT neurons (Luchetti et al., 2020). Although the dorsal hippocampus primarily receives inputs from median raphe to drive memory processes, there are also significant contributions of the smaller 5-HT projection from DR to impact reward processing, cognitive flexibility, spatial memory, and decision-making (Liu et al., 2014; Miyazaki et al., 2014; Matias et al., 2017; Teixeira et al., 2018). In our prior report, we showed that Fos induction in DR, but not in the median raphe, 5-HT neurons significantly correlated to cocaine-seeking behavior on ED1, and our current results confirm that the DR projection to CA1 drives persistent cocaine-seeking behavior. Although DR contains a diverse population of projection neurons, including dopamine, glutamate, and GABA, these data together with our pharmacology results using 5-HT antagonists suggest that the primary signal relevant for driving cocaine-seeking behavior is serotonergic.
Roles of hippocampal 5-HT1A and 5-HT1B receptors in drug memory
The role of hippocampal 5-HT receptors in learning and memory is controversial, such that the release of 5-HT into the hippocampus can exert both positive and negative effects on multiple memory types, due at least in part to the large number of 5-HT receptors and diverse downstream signaling mechanisms (Hoyer and Martin, 1997). One such controversial result is that both increased 5-HT activity and inhibition of brain 5-HT synthesis can impair learning performance; this puzzling set of results may indicate that 5-HT has U-shaped effects on memory (Ogren et al., 1982; Ogren, 1985). The specific roles of 5-HT1A and 5-HT1B receptors in CA1 pyramidal and interneurons on memory acquisition and performance are of particular interest to the interpretation of results found in the current manuscript (Meneses, 2001; Ogren et al., 2008).
Previous studies indicate that CA1 5-HT1A signaling impairs spatial memory acquisition, and this effect can be reversed by the administration of 5-HT1A–specific antagonists such as WAY-100635 (Carli et al., 1999; Eriksson et al., 2007). Further support for an inhibitory role of 5-HT1A receptors in memory is reviewed in Ogren et al. (2008). Consistent with this view, we show that the 5-HT1A receptor antagonist WAY-100635 administered during cocaine CPP acquisition (but not immediately prior to testing) enhanced cocaine-associated contextual memory. A 24 h retest further supported this interpretation, as 5-HT1A antagonism on test day decreased preference when tested 24 h later, perhaps due to increased recall of the lack of drug in the prior CPP test experience.
The role of 5-HT1B receptors in memory is more complicated and appears to depend on the type of memory task and cognitive demand. Mice lacking 5-HT1B receptors show enhanced acquisition of the Morris water maze, a hippocampus-dependent memory task, but impaired working memory in the radial arm maze (Buhot et al., 2003). However, 5-HT1B receptor antagonists in CA1 promote memory formation (Cai et al., 2013) and facilitate avoidance learning (Eriksson et al., 2008). Inverse agonists of 5-HT1B receptors, or the 5-HT1B antagonist GR-127935, administered post-training can also facilitate learning in associative autoshaping tasks (Meneses, 2001). Our CPP results lead us to hypothesize that in our paradigm, CA1 5-HT1B signaling enhances drug “memory” recall (and therefore cocaine seeking on ED1) and blockade of these receptors (e.g., by GR-127935) decreases recall, as shown by decreased cocaine seeking on ED1.
We interpret our results to indicate that activation of 5-HT1A and 5-HT1B receptors in CA1 on initial abstinence (ED1) both decrease consolidation of the absence of drug and enhance contextual recall, respectively, causing increased cocaine-seeking behavior and subsequently maintaining seeking behavior 2 weeks later (Fig. 8). We propose that if these signals are blocked, either by antagonizing 5-HT1 receptors in the CA1 or by chemogenetically inhibiting raphe 5-HT input to CA1, recall of the drug-associated context is decreased thereby reducing cocaine-seeking behaviors on ED1 and consolidation of the novel information about the context (e.g., the absence of cocaine) is increased, thereby attenuating persistent seeking following 2 weeks of abstinence. Our data and interpretation are consistent with those from other studies finding that 5-HT1A and 5-HT1B receptors inhibit consolidation and facilitate recall of memories, respectively (Buhot et al., 2003; Ogren et al., 2008).
Hippocampal CA1 serotonin signaling drives cocaine-seeking persistence. A, Illustration of DR 5-HT synapses on 5-HT1A and 5-HT1B receptor–containing neurons in dorsal hippocampal CA1. B, Illustration of how 5-HT1B and 5-HT1A receptors in dorsal hippocampal CA1 pyramidal neurons engage distinct memory processes, based upon current results. 5-HT1B receptor signaling enhances contextual recall of the prior drug-associated context and thus drives increased cocaine-seeking behavior on ED1 (extinction “burst”). 5-HT1A receptor activation decreases consolidation of the absence of the drug, functionally promoting cocaine-seeking behavior 2 weeks later (persistence). 5-HT; serotonin, DR; dorsal raphe.
Conclusions
The present findings indicate that initial abstinence may be a critical time point for targeting subsequent drug-seeking behavior. Although these behaviors differ substantially between sexes, the serotonin system may be uniquely involved in both. Thus, targeting drug-seeking behavior during initial abstinence via therapeutic interventions that manipulate contextual memory or serotonin signaling has potential treatment implications for both sexes.
Footnotes
This research was supported by Public Health Service Grant DA006214 (G.A.-J.) and National Institute of Environmental Health Sciences T32ES007148 and K99DA045758 (A.S.K.). We thank the members of the Aston-Jones lab for their thoughtful comments on the project. We also thank Belle Lin and David Goldmeier for their technical assistance in completing this work.
The authors declare no competing financial interests.
- Correspondence should be addressed to Amy S. Kohtz at akohtz{at}umc.edu.
