Inhibition of estradiol signaling in the basolateral amygdala impairs extinction memory recall for heroin-conditioned cues in a sex-specific manner

Abstract

Introduction:

Relapse is a major treatment barrier for opioid use disorder (OUD). Environmental cues become associated with the rewarding effects of opioids and can precipitate relapse, even after numerous unreinforced cue presentations, due to deficits in extinction memory recall (EMR). Estradiol (E2) modulates EMR of fear-related cues, but it is unknown whether E2 impacts EMR of reward cues and what brain region(s) are responsible for E2’s effects. Here, we hypothesize that inhibition of E2 signaling in the basolateral amygdala (BLA) will impair EMR of a heroin-associated cue in both male and female rats.

Methods:

We pharmacologically manipulated E2 signaling to characterize the role of E2 in the BLA on heroin cue EMR. Following heroin self-administration, during which a light/tone cue was co-presented with each heroin infusion, rats underwent cued extinction to extinguish the conditioned association between the light/tone and heroin. During extinction, E2 signaling in the BLA was blocked by an aromatase inhibitor or specific estrogen receptor (ER) antagonists. The next day, subjects underwent a cued test to assess heroin cue EMR.

Results:

In both experiments, females took more heroin than males (mg/kg) and had higher operant responding during cued extinction. Inhibition of E2 synthesis in the BLA impaired heroin cue EMR in both sexes. Notably, E2’s actions are mediated by different ER mechanisms, ERα in males but ERβ in females.

Conclusions:

This study is the first to demonstrate a behavioral role for centrally-produced E2 in the BLA and that E2 also impacts EMR of reward-associated stimuli in both sexes.

Introduction

With over 1.5 million individuals having an opioid use disorder (OUD), the opioid epidemic rages on; contributing to over 75,000 overdose deaths in 2022 and a $1.02 trillion annual economic burden [1–3]. During substance use, environmental sensory cues are conditioned with the rewarding nature of the drug. A key feature of OUD is the ability of these drug-associated cues to evoke drug-seeking behaviors, precipitating relapse during periods of abstinence. This effect is persistent even after numerous unreinforced cue presentations (extinction resistant) and presents a major barrier to long-term abstinence for those with OUD [4, 5, reviewed in: 6]. These cues are a particularly potent relapse “trigger” in females, as they are more responsive to drug-conditioned stimuli [7, 8]. Even though OUD is more common in males, females have more rapid acquisition of opioid self-administration, greater susceptibility to the addictive properties of opioids, higher rates of prescription opioid misuse, and poorer initial responses to pharmacotherapies for OUD [ reviewed in: 9, 10, 11]. Exclusion of females from foundational studies on OUD has hindered our ability to understand, and thereby address, these disparities.

Those with OUD may have a pathologic deficit in extinction memory recall (EMR), the process of expressing an extinction memory that competes with the original conditioned memory [4, 5]. Opioids form strong conditioned associations that may disrupt EMR, conferring pro-relapse, maladaptive behaviors [4, 5]. Cue exposure therapy (CET) is one therapeutic intervention suggested for treatment of substance use disorders (SUDs). During CET, drug-related cues are presented with the goal of decreasing cue reactivity and promoting extinction memory formation, thereby preventing cue-induced relapse. CET has had mixed results for treatment of SUDs [reviewed in: 12, 13, 14]. Concerningly, despite decreasing physiologic and self-reported cue reactivity, one study found that those who underwent CET had higher dropout and relapse rates compared to placebo psychotherapy [15]. There is a more recent interest in using cognitive-enhancing pharmacologic agents in conjunction with CET to treat anxiety and substance use disorders, but none have advanced to clinical utility for SUD treatment [16].

One candidate pathway to target for this use is estrogen signaling. Estradiol (E2) is the predominant form of estrogen, a gonadal sex hormone and neurosteroid, that impacts learning and memory [reviewed in: 17] among other functions. E2 can access the brain from the peripheral circulation but is also synthesized by the enzyme aromatase within the brain, the major source for E2 involved in central signaling [18, 19]. Notably, this synthesis of E2 occurs within the brain of both males and females [20, 21]. For this reason, it is important to consider the functions of central E2 signaling in both sexes. E2 exerts its effects by binding to three types of estrogen receptors (ERs; ERα, ERβ, and GPER), which are widely expressed throughout the male and female brain [22]. ERα and ERβ are steroid receptors while GPER is a G protein coupled receptor. In addition to their canonical signaling as transcription factors, ERα and ERβ can also mediate rapid signaling from the membrane like GPER [reviewed in: 21, 23, 24]. Most evidence for the molecular mechanisms of E2 in memory processes comes from the hippocampus, where E2 facilitates induction of long-term potentiation (LTP) and stimulates dendritic spine formation and remodeling, however other regions may be similarly impacted [24]. E2’s impacts on memory have also been determined behaviorally [reviewed in: 19, 25]. Relevant to this study, decreased E2 signaling impaired EMR for fear-associated cues in males and females [26, 27]. Notably, systemic ERβ agonism, but not ERα agonism, improved fear cue EMR in females, but no agonists have been tested in males [28].

One brain region implicated in this associative reward learning is the basolateral amygdala complex (BLA), composed of the lateral (LA), basal (BA) and basomedial (BM) subnuclei [reviewed in: 29, 30]. The BLA is involved in formation and recall of both the conditioned and extinction memories, including for opioids [31]. Importantly, the BLA is subject to regulation by E2: it expresses all three types of ERs and inhibition of E2 synthesis has sex-specific effects on dendritic spine density and excitability [32, 33]. Furthermore, brain activation studies have suggested the amygdala may be an important site of action for E2 signaling during extinction [34, 35]. In the BLA, males and females express aromatase at similar levels, which yields similar levels of E2 locally, and both sexes express similar levels of ERα [32]. ERβ expression has been repeatedly detected in the BLA, but quantification and direct comparisons between sexes are complicated by antibody validation challenges that limit reliability of prior studies [36] and by conflicting results both within and between mRNA and protein analyses [37–43].

Given prior studies showing that E2 modulates fear EMR, in the present study, we tested the role of E2 on EMR of a drug reward-related cue. While fear and reward are oppositely valenced, they may share similar mechanisms of memory extinction [29, 44]. To date, there are no studies examining E2 and cue EMR for a rewarding stimulus in males or females. Based on prior studies demonstrating that central E2 signaling is important for fear EMR in both males and females [26, 27], we hypothesized that E2 signaling in the BLA is important for heroin cue EMR in males and females, therefore, disruption of BLA E2 signaling would lead to a deficit in EMR. We tested this hypothesis by incorporating established behavioral techniques of heroin self-administration and cued drug seeking into a novel experimental protocol, designed to mimic prior fear EMR studies (Fig. 1). In the present study, we examined how inhibition of E2 synthesis (Experiment 1) or specific ER antagonism (Experiment 2) in the BLA impacted heroin cue EMR in male and female rats.

Fig. 1: Behavioral protocol for assessing heroin-cue extinction memory recall.

(Left) On Days 1–8 (Conditioning), rats self-administer heroin via nose pokes. Nose pokes result in presentation of a light and tone stimulus (Neutral Stimulus) and a concomitant infusion of heroin (Unconditioned Stimulus, US). Over subsequent pairings, the light/tone becomes associated with the rewarding effects of the heroin (light/tone becomes a Conditioned Stimulus, CS). (Middle) On Day 9 (Cued Extinction), active nose pokes result in presentation of the CS without the US (light/tone without heroin), causing decreased drug-seeking behavior (nose pokes) over time as the extinction memory is formed. E2 manipulations occur on this day, prior to the session. (Right) Finally, on Day 10 (Extinction Memory Recall, EMR), nose pokes again result in presentation of the CS without the US. Rats with good EMR will seek less drug (lower nose pokes), as they recall that this behavior and the light/tone no longer predict a reward. Rats with elevated drug seeking (higher nose pokes) are said to have poor EMR, as their behavior indicates dominance of the original conditioned memory.

Materials and Methods

Subjects & Surgeries

A total of 110 male and female Wistar rats (Table S1, 8–9 weeks old on arrival, Envigo) were used, with 4 subjects excluded from final analyses due to catheter failure or missed cannula placement. All procedures were approved by the Institutional Animal Care and Use Committee of the Medical University of South Carolina in accordance with the “Guide for the Care and Use of Laboratory Rats” of the Institute of Laboratory Animal Resources on Life Sciences, National Research Council. After acclimation, rats had bilateral cannulas stereotactically implanted for pharmacologic manipulation of ERs in the BLA (relative to bregma: anterior-posterior: −2.5mm, medial-lateral: ±4.8mm, dorsal-ventral: −8.5mm). Subjects also had a catheter implanted into the right jugular vein for heroin self-administration, as detailed in prior publications [45, 46] and the Supplement. In a preliminary experiment, we tested the effects of estrous cycle and ovariectomy on heroin cue EMR (Fig. S1). We did not detect differences in EMR across groups (Figs. S1I and S1J), so we used gonadally intact males and females for the Experiments discussed herein. Further elaboration on this decision is included in the Discussion and Supplement.

Drugs & Delivery

Heroin (NIDA Drug Supply Program, Rockville, MD) was dissolved in 0.9% saline to 200mg/250mL. An aromatase inhibitor (Fadrozole, FAD, MedChemExpress, Monmouth Junction, NJ), ERα antagonist (1,3-Bis(4-hydroxyphenyl)-4-methyl-5-[4-(2-piperidinylethoxy)phenol]-1H-pyrazole dihydrochloride, MPP, Tocris Biosciences, Minneapolis, MN), and ERβ antagonist (4-(2-phenyl-5,7-bis(trifluoromethyl)pyrazolo[1,5-a]pyrimidin-3-yl)phenol; PHTPP, MedChemExpress) were dissolved in DMSO and diluted with 0.9% saline, as detailed in the Supplement. Doses for FAD, MPP, and PHTPP used herein were chosen from previously published studies [47–56] and checked against selectivity parameters for each drug (Table 1). Given prior studies demonstrating in vitro and in vivo selectivity and specificity, we did not test other doses in the present study. Intracranial infusions (1 μL per hemisphere, 2.5% v/v DMSO/normal saline) were administered at a rate of 0.2 μl/min. Infusion localization and spread/diffusion was confirmed in a separate cohort of four subjects using Fast Green FCF (ThermoFisher, Waltham, MA) dissolved in the same vehicle solution and infused at the same rate and volume (0.2 μL/min, 1 μL; Fig. 2). Brains were collected 6 hours after the infusion to determine spread in alignment with the experimental methods.

Table 1:

Drug doses and administration

Exp.	Drug	Mechanism	Dose	Volume	Rate	Citations
1	Fadrozole	Aromatase Inhibitor	1.20 μg/hemisphere	1 μL	0.2 μL/min	[47–51]
2	MPP	ERα Antagonist	2.78 ng/hemisphere	1 μL	0.2 μL/min	[52–55]
	PHTPP	ERβ Antagonist	2.12 ng/hemisphere	1 μL	0.2 μL/min	[52–54,56]

Fig. 2: Intracranial infusion spread in the basolateral amygdala.

(Left) Schematic demonstrating spread of Fast Green FCF infused through cannulas targeting the BLA of 2 males and 2 females. Opacity of the oval indicates intensity of the straining. (Right) Representative images of dye spread in a female (top) and male (bottom) with region delineations (dashed lines) and brain borders (solid lines). n = 4; BLA, basolateral amygdala; CeA, central amygdala.

Self-Administration, Cued Extinction, & Extinction Memory Recall

Detailed behavioral procedures (summarized in Fig. 1) are included in the Supplement. All behavior occurred within the same operant chamber/context. Briefly, rats self-administered heroin via nose pokes on an ascending fixed ratio (FR) for 6 hr/day for 8 days. Heroin infusions (40 μg) were paired with a light and tone stimulus (heroin-conditioned cue). On day 9, all rats underwent a 6-hour cued extinction session, during which active nose pokes (ANPs) led to presentation of the heroin-conditioned light+tone cue without heroin delivery. Prior to cued extinction, subjects received an intracranial infusion of vehicle (2.5% v/v DMSO/normal saline), fadrozole (Experiment 1; Fig. 3), MPP, or PHTPP (Experiment 2; Fig. 4) into the BLA. Subjects were randomly assigned to vehicle or treatment groups. The following day, rats were tested for extinction memory recall (EMR) during a 1-hour session under the same conditions as cued extinction, except there were no infusions prior to EMR test. In a small subset of subjects in Experiment 1, we administered fadrozole prior to both cued extinction and EMR test to rule out potential effects of state-dependent learning (FAD-FAD group).

Fig. 3: Aromatase inhibition in the basolateral amygdala during cued extinction impairs heroin-cue extinction memory recall in both males and females.

A. Experiment 1 timeline. Group n’s are as follows: 23 males (10 Veh, 10 FAD, 3 FAD-FAD), 21 females (9 Veh, 9 FAD, 3 FAD-FAD). B. Cannula placements in the BLA. C. Daily operant responding (ANPs and INPs) during heroin self-administration. There were no differences in ANPs between males and females, but females had greater INP responding. D. Heroin intake (mg/kg) during self-administration. Females took more heroin than males. E. Total ANPs during the 6hr cued extinction session. Females had greater ANPs than males. Aromatase inhibition (FAD) had no effect on ANPs during extinction in either sex. F. Hourly operant responding (ANPs and INPs) during the cued extinction session. Veh and FAD treated groups were collapsed since FAD did not impact responding.

The increased ANPs in females relative to males were driven by responding during the first hour of extinction. G. Total ANPs during the 1hr EMR test. FAD, administered into the BLA on the day prior, increased ANPs in both males and females relative to vehicle controls. H. Difference scores comparing ANPs during extinction versus ANPs during EMR test. FAD increased difference scores in both sexes, indicative of impaired EMR. I. Total ANPs during the 1hr EMR test (Veh and FAD groups collapsed across sex and reproduced from panel G above). FAD administration into the BLA just prior to cued extinction (FAD) and before cued extinction and EMR test (FAD-FAD), increased ANPs in both males and females relative to vehicle controls. J. Difference scores comparing ANPs during extinction versus ANPs during EMR test (Veh and FAD groups collapsed across sex and reproduced from panel H above). FAD and FAD-FAD groups had increased difference scores relative to Veh, indicative of impaired EMR. All data are shown as mean ±SEM with n’s indicated in each panel. **p < 0.01, *** p < 0.001, ****p < 0.0001, $p < 0.01 versus Veh. ANPs, active nose pokes; BLA, basolateral amygdala; EMR, extinction memory recall; FAD, Fadrozole; FR, Fixed Ratio; INPs, inactive nose pokes; Veh, vehicle.

Fig. 4. Antagonism of estrogen receptors in the basolateral amygdala during cued extinction has sex-specific effects on heroin-cue extinction memory recall.

A. Experiment 2 timeline. Group n’s are as follows: 19 males (5 Veh, 7 MPP, 7 PHTPP), 20 females (6 Veh, 7 MPP, 7 PHTPP). B. Cannula placements in the BLA. C. Daily operant responding (ANPs and INPs) during heroin self-administration. There were no differences in ANPs between males and females, but females had greater INP responding. D. Heroin intake (mg/kg) during self-administration. Females took more heroin than males. E. Total ANPs during the 6hr cued extinction session. Females had greater ANPs than males. MPP and PHTPP had no effect on ANPs during extinction in either sex. F. Hourly operant responding (ANPs and INPs) during the cued extinction session. Veh, MPP, and PHTPP treated groups were collapsed since drug had no effect. The increased ANPs in females relative to males were driven by responding during the first hour of extinction. G. Total ANPs during the 1hr EMR test. MPP and PHTPP, administered into the BLA on the day prior, resulted in a sex-specific pattern of responding on EMR test. Males who received MPP had greater ANPs relative to male vehicle controls. Females who received PHTPP had greater ANPs relative to all other groups. H. Difference scores comparing ANPs during extinction versus ANPs during EMR test. MPP increased difference scores in males and PHTPP increased difference scores in females. All data are shown as mean ±SEM with n’s indicated in each panel. ****p < 0.0001, $p < 0.05 versus Veh, #p < 0.05 versus all other groups. ANPs, active nose pokes; BLA, basolateral amygdala; EMR, extinction memory recall; FR, Fixed Ratio; INPs, inactive nose pokes; ME, main effect; MPP, 1,3-Bis(4-hydroxyphenyl)-4-methyl-5-[4-(2-piperidinylethoxy)phenol]-1H-pyrazole dihydrochloride (ERα antagonist); PHTPP, 4-(2-phenyl-5,7-bis(trifluoromethyl)pyrazolo [1,5-a]pyrimidin-3-yl)phenol (ERβ antagonist); Veh, vehicle.

Data & Statistical Analysis

To account for individual differences in operant responding, we calculated a difference score (Total ANPs_{EMR Test} – Afterage/Hour ANPs_Extinction) as one approach to assess EMR. Negative values are indicative of better EMR, as subjects had lower drug seeking on test than cued extinction. A score near zero is interpreted as poor EMR, as rats did not adjust their drug seeking behavior after extinction. Increased seeking on EMR test yields a positive difference score, suggesting very poor EMR. Additional details about the derivation of the difference score are included in the Supplement.

Prior to analysis, the ROUT method was applied to detect outliers in EMR test responding or difference scores (the primary behavioral measurements). At both Q = 1% (preferred threshold, more permissive to variance) and Q = 5% (more likely to identify an outlier), no outliers were detected. All subjects were therefore included in the subsequent analyses. Analyses of variance (ANOVAs; 2 or 3way ± repeated measures [RM] with Geisser-Greenhouse corrections) were used to analyze the dependent measures (operant responding [ANPs, INPs, difference scores] and intake). For all ANOVAs, the between subjects independent variables were sex (male/female) and treatment during extinction (Veh/FAD or Veh/MPP/PHTPP). Holm-Šídák’s post-hoc comparisons between both treatment group and sex were used when appropriate and are reported throughout. Complete ANOVA tables are shown in Tables S2 and S3. The significance was set at α=0.05 throughout. All analyses and graphs were produced using GraphPad Prism software (version 9.5).

Results

Inhibition of E2 synthesis in the BLA during cued extinction disrupts heroin cue EMR in both sexes

A timeline for Experiment 1 is shown in Fig. 3A with cannula placements in Fig. 3B (expanded in Fig. S2). The 3way ANOVA comparing nose poke x time x sex revealed that both males and females increased ANPs over time, consistent with the increase in FR (Fig. 3C, day x nose poke interaction F[5.14,215.7]=19.06, p < 0.0001). Despite similar operant responding, females took more heroin (mg/kg) than males (Fig. 3D, 2way RM ANOVA, main effect of sex F[1,42]=67.26, p < 0.0001). Both sexes increased intake over the eight days of self-administration (main effect of day F[4.02,168.6]=32.87, p < 0.0001). Intake did not differ between rats randomly assigned to Veh or FAD groups (Fig. S3B). During cued extinction, females had greater ANPs than males (Fig. 3E, 2way ANOVA, main effect of sex, F[1,40]=14.09, p = 0.0006) but fadrozole did not impact responding, so FAD and vehicle groups were collapsed. In the subsequent analysis of responding across cued extinction, there was a three way interaction between time, nose poke, and sex (Fig. 3F, 3way RM ANOVA, F[5,210]=6.36, p < 0.0001). Given this interaction, we separated data by nose pokes and analyzed with 2way RM ANOVAs. Sex differences in ANPs during extinction resulted from greater responding by females during the first hour of the session (2way RM ANOVA, hour x sex interaction F[5,210]=8.98, p < 0.0001). INPs decreased across the extinction session, but there were no sex differences (2way RM ANOVA, main effect of hour, F[1.66,69.61]=6.08, p = 0.006). During the EMR test, subjects who received FAD in the BLA the day prior (FAD) had greater ANPs (Fig. 3G, 2way ANOVA, main effect of FAD F[1,34]=31.66, p < 0.0001) and elevated difference scores (Fig. 3H, 2way ANOVA, main effect of FAD F[1,34]=38.54, p < 0.0001) relative to vehicle. After collapsing across sex, the group that received fadrozole prior to extinction and EMR test (FAD-FAD) had impaired EMR relative to vehicle, but did not differ from the group that received a fadrozole infusion prior to extinction only (FAD; Figs. 3I & 3J, ANPs: 1way ANOVA, F[2,41]=17.60, p < 0.0001; Difference Score: 1way ANOVA, F[2,41]=22.28, p < 0.0001). Thus, inhibition of E2 synthesis in the BLA negatively impacts heroin EMR in both males and females.

ERα and ERβ antagonism in the BLA during cued extinction has sex-specific effects on heroin cue EMR

A timeline for Experiment 2 is shown in Fig. 4A with cannula placements in Fig. 4B (expanded in Fig. S2). As in Experiment 1, the 3way ANOVA comparing nose poke x time x sex revealed that both males and females increased ANPs over time (day x nose poke interaction F[5.05,186.9]=10.78, p < 0.0001). Differently, there was also a main effect of sex (F[1,37]=16.02, p = 0.0003; Fig. 4C). Follow-up analyses determined that this sex effect was driven by differences in INPs (2way RM ANOVA, main effect of sex F[1,37]=9.69, p = 0.0036). Patterns of heroin intake were also similar to Experiment 1, with both sexes increasing intake over time and females taking more than males (2way RM ANOVA, day x sex interaction F[7,259]=2.69, p = 0.011; main effect of sex F[1,37]=20.53, p < .0001]). Additionally, intake did not differ between rats randomly assigned to Veh, MPP, or PHTPP groups (Fig. S4B). During cued extinction, females had greater ANPs than males (Fig. 4E, 2way ANOVA, main effect of sex F[1,33]=28.35, p < 0.0001), but ANPs were not impacted by ER antagonism (MPP/PHTPP). Therefore, groups were collapsed by sex and analyzed across time. The resultant ANOVA revealed a three way interaction between time, nose poke, and sex (Fig. 4F, F[5,185]=15.53, p < 0.0001), so data were separated by nose poke and analyzed with 2way RM ANOVAs. As in Experiment 1, this analysis showed that elevated first hour responding by females drove the sex difference in ANPs (2way RM ANOVA, hour x sex interaction F[5,185]=19.26, p < 0.0001). INPs decreased across the extinction session regardless of sex (2way RM ANOVA, main effect of hour, F[1.784,66.01]=10.33, p = 0.0002). EMR, as assessed by ANPs during test (Fig. 4G) and difference scores (Fig. 4H), was impacted by ER antagonism in a sex-specific pattern (2way ANOVAs, sex x antagonist interactions; ANPs: F[2,33]=15.89, p < 0.0001; Difference Scores: F[2,33]=16.92, p < 0.0001). Males who received MPP (ERα antagonist) in the BLA had impaired EMR relative to vehicle controls, but MPP did not impact EMR in females. Conversely, females who received PHTPP (ERβ antagonist) exhibited profoundly impaired EMR, but PHTPP did not significantly alter male EMR. Together, these data indicate that E2 signaling in the BLA impacts EMR via different ER subtypes between males (ERα) and females (ERβ).

Discussion/Conclusion

Decreased E2 signaling has been shown to impair cued fear EMR [26, 27], but its impact on drug reward EMR was unknown. Furthermore, region-specific actions of E2 had not been explored for cued EMR of either valence. In the present study, we tested the hypothesis that inhibition of E2 signaling in the BLA during cued extinction, by blocking E2 synthesis (Experiment 1) or specific ER antagonism (Experiment 2), would impair heroin cue EMR in male and female rats.

Sex-specific responding during self-administration and cued extinction

Across both Experiments, several sex-specific patterns emerged during heroin self-administration and cued extinction. First, despite having similar operant responding (Figs. 3C and 4C), females took more heroin than males when adjusted for differences in body weight (Figs. 3D and 4D). This finding is in agreement with previous publications from our lab [45, 46] and others [57–63] showing that females take more opioids than males. While only a single dose of heroin was used in the present study, there was no set infusion criterion and subjects were able to alter their intake by modifying their responding.

Next, regardless of E2/ER manipulation, females had greater ANPs than males during cued extinction (Figs. 3E and 4E), driven by increased responding in the first hour of the session (Figs. 3F and 4F). Prior studies from our lab [45] and others [63–66] have shown that females display increased drug seeking behavior during context (non-cued) extinction sessions, particularly early into extinction training. The present study, however, used a cued extinction model. Preclinically, females have greater cue-induced opioid seeking [60, 63] and report greater cue-induced opioid craving in clinical studies [8]. Several reports of conflicting findings to those described above, so debate regarding sex differences in cue reactivity and drug craving/seeking persist in the field [reviewed in: 67]. Presently, our findings support the perspective that females have greater drug-associated cue reactivity than males, driving increased drug seeking behaviors. Despite having greater ANPs in the first hour of cued extinction, females and males exhibit similar responses for the remainder of the session. This could reflect within-session facilitation of extinction in females, as they exhibit a greater absolute reduction in ANPs relative to males, perhaps due to having more unreinforced cue presentations than males. This increase in cue presentations did not transfer to increased EMR, since male and females exhibited similar difference scores. Additional experiments controlling the number of unreinforced cue presentations would determine how learning the change in contingency reflects memory for the extinguished association.

Estradiol synthesized in the basolateral amygdala contributes to heroin cue extinction memory recall

The primary focus of this study was to examine the role of E2 signaling in the BLA on heroin cue EMR. In Experiment 1 (Fig. 3), we aimed to test whether inhibition of E2 synthesis in the BLA with FAD would impair heroin cue EMR. We show that both males and females have poor EMR for a heroin-conditioned cue at baseline, because in both Experiments vehicle-treated subjects have difference scores close to zero, indicating similar drug-seeking responses between EMR test and cued extinction (Figs. 3H and 4H). This finding supports interpretations that a pathological deficit in EMR may contribute to relapse in OUD [4, 5].

However, despite the baseline EMR deficit, aromatase inhibition during cued extinction further disrupted EMR in both sexes (positive difference scores; Figs. 3G and 3H). This agrees with the published role of E2 in fear EMR [26, 27]. In females, the non-selective ER antagonist fulvestrant impaired fear EMR [27] as did the aromatase inhibitor, fadrozole, in males [26]. As in a prior study [26], this effect on heroin cue EMR was not due to state-dependent learning (Figs. 3I & 3J). Our results expand on these findings, showing that E2 signaling also plays a role in EMR of drug-related cues, and we have identified the BLA as an important region for E2 signaling to impact EMR. Furthermore, this is the first study showing that central synthesis of E2 contributes to EMR. Prior fear EMR studies have used systemic administration, altering both peripheral and central E2 synthesis, or directly modulated ER signaling via agonism or antagonism. Our approach, infusing the aromatase inhibitor directly into the BLA, verifies that E2 synthesis within the brain, specifically in the BLA, modulates EMR. As multiple cell types in the brain can produce and/or respond to E2 (e.g., astrocytes, principal neurons, interneurons) [18, 19], it is unclear what cell(s) in the BLA produce the E2 and what cell(s) are the targets of E2 signaling that underly these impacts.

Different estrogen receptor subtypes in the basolateral amygdala contribute to impaired extinction memory recall in males and females

In Experiment 2 (Fig. 4A), we probe the role(s) for specific ERs in the BLA that may mediate E2’s impacts on EMR. Administration of an ERα antagonist, MPP, into the BLA during cued extinction impaired heroin cue EMR in males only (Figs. 4G and 4H). The opposite, impaired EMR in females only, was found after administration of an ERβ antagonist, PHTPP. Our results demonstrate that disruption of heroin cue EMR is dependent on sex-specific ER mechanisms in the BLA. Previous studies have reported enhancing effects of systemic ERβ agonism in females during cued [28] and contextual [34] fear EMR, which are in agreement with the current study. Notably, no receptor mechanism for E2’s impacts on fear EMR has been previously identified in males, but our findings in heroin cue EMR indicate that ERα signaling may contribute.

Interestingly, different ER types have been previously shown to underly similar outcomes in males and females. For example, in the hippocampus E2 enhances synaptic potentiation via both pre- and post-synaptic mechanisms [19, 68]. Pre-synaptically, these enhancements rely on ERα in males and ERβ in females, while post-synaptic potentiation is reliant on ERβ in males and GPER in females [69]. Here, we describe a convergent behavioral outcome mediated by sex-divergent ER types (ERα in males and ERβ in females), however we did not test the contributions of the third ER type, GPER, in this study (discussed below). E2 has been shown to impact BLA structure and function, in both sex convergent and divergent manners [reviewed in: 33]; but, to our knowledge, this is the first time that different ER mechanisms in the BLA of males and females have been shown to elicit a convergent behavioral effect. The molecular features underlying this phenomenon are still being described in the hippocampus [70] but remain unexplored in the BLA. The third ER type, GPER, was not examined in this study, but is robustly expressed in the BLA of both sexes, with greater expression in females during diestrus than females during estrus or males [71]. Depending on the brain region, GPER signals via rapid modulation of several cell signaling cascades in neurons and/or astrocytes [72–74]. To our knowledge, no studies have characterized GPER function in the BLA, so it is unclear how GPER may impact neuronal/astrocytic signaling cascades or calcium levels therein. Interestingly, specific activation of GPERs may exert similar effects as E2 by different mechanisms (e.g., both enhance object recognition memory in the hippocampus but via different signaling cascades; [75]) and even exhibit estrogen-independent activity (e.g., GPER inhibits, while E2 enhances, neuron proliferation in the dentate gyrus; [76]). However, sex differential GPER expression in the BLA may lead to sex-specific outcomes. Future studies focused on the contributions of GPER to EMR and BLA function will address this question. Though speculative and beyond the scope of this project, we predict that antagonism of GPER in the BLA would impair heroin-cue EMR in both sexes, perhaps through different signaling mechanisms.

In addition to unexplored molecular mechanisms, cell-type specific expression of ERα and ERβ in the BLA has also not been fully described. One study in guinea pigs documented expression of ERα on calretinin-positive GABA neurons, while ERβ was expressed on parvalbumin-positive GABA neurons, but this did not account for all neurons in the amygdala expressing either ER subtype [77]. The remaining ERs could be expressed on other neurons, like glutamatergic principal neurons, or various glial cell types. Additional studies of the BLA show E2 impacts both glutamatergic and GABAergic neuron function [78]. Further classification of ERα and ERβ expression in the BLA could help better understand the possible receptor mechanisms by which E2 signaling impacts EMR in each sex.

Possible mechanisms for estradiol’s impacts on extinction memory recall

Fear EMR studies have described the basic mechanisms by which E2 impacts EMR. First, they determined that inhibition of E2 disrupts EMR by impairing consolidation of the extinction memory. Inhibition of E2 signaling during cued extinction or just after the session caused an EMR deficit, however manipulation after the consolidation window closed (around 4 hr) did not impact EMR [26]. Additional studies have shown that E2 signaling likely impacts fear EMR through an NMDAR-dependent mechanism [79], in agreement with E2’s molecular effects in other regions [reviewed in: 17, 80]. Here, we expand this finding by demonstrating that this effect is also present for EMR of a heroin cue and that E2 signaling in the BLA is at least partially responsible for the change in EMR.

One prior study of E2 in the BLA found that blocking E2 synthesis with an aromatase inhibitor reduced spine density in adult mice (in vivo) and neonatal rat organotypic slice cultures (in vitro), but only in females [32]. Similarly, aromatase inhibition blocked LTP in the BLA of female, but not male, juvenile rats [32]. Our results in females, that aromatase inhibition impairs heroin-cue EMR, suggesting that aromatase inhibition decreases BLA plasticity, matches the spine and LTP data from Bender and colleagues. In contrast, here fadrozole also induced an EMR deficit in males, but no changes in spines or LTP were reported previously. The most likely contributor to this discrepancy is the use of behaving subjects engaging complex intra- and extra-amygdalar regulation via numerous other transmitters and signals, which may modulate E2’s contribution to the behavioral changes. These complex interactions are not likely recapitulated using in vitro or non-behaving in vivo methods. Additionally, E2 could act via a different mechanism in the male BLA not captured by these molecular methods, such as altering inhibitory interneuron or pre-synaptic neuron function. Analysis of spine morphology and molecular markers associated with LTP induction and other molecular pathways in the brains of behaving subjects could help resolve this discrepancy. Additional contributors include differences in subject age and species (adult rats here versus adult mice, neonatal rats [spines], and juvenile rats [LTP]) and the inhibitors (fadrozole here versus letrozole).

Methodological considerations:

Extinction session duration and infusion timepoint

Modeling prior fear EMR studies, the 6-hour cued extinction session duration was selected to combat the heroin-cue association and match the self-administration session length. Across the session, our data indicate that most of the responding occurs within the first hour, therefore, by the end of the 6-hour session rats are no longer responding for the heroin cue. Our finding that control subjects had poor EMR for heroin cues after a single extinction session is not surprising, given the strong associations that are formed by drugs of abuse.

Infusions were delivered prior to the cued extinction session to test impacts of E2 signaling on EMR the following day. Administration of the test compounds prior to the extinction session alters both acquisition and consolidation, while infusion immediately after impacts memory consolidation only. In fear EMR studies, impairment of E2 signaling during or shortly after cued fear extinction led to an EMR deficit, but not after the consolidation window closed [26]. While parsing the mechanistic difference by which E2 impacts EMR (between acquisition or consolidation of the extinction memory) was not the focus of this study, we did rule out state-dependent learning by infusing fadrozole both pre-extinction and pre-EMR test to a subset of subjects (Figs. 3I & 3J). Using this approach, we found that E2 signaling in the BLA did not impact acquisition of cued extinction. Additionally, a post-extinction infusion would have occurred outside of the consolidation window (4 hr) for part of the session.

Drug duration of action and target-specificity

Clearance of the E2/ER drugs prior to the end of the cued extinction session was a concern in choosing the 6-hour session duration. Fadrozole has a half-life of approximately 10.5 hours when dosed orally [81], however, the half-lives for MPP and PHTPP are unknown [82, 83]. Based on the results herein, we interpret that the duration of action for all three drugs during the cued extinction session is sufficient, as they each result in a behavioral change during EMR test. Without a sufficient duration of action, we would have expected a false negative result (i.e., no change in EMR relative to vehicle groups).

Aromatase inhibitors alter production of other steroid hormones, including minor forms of estrogen, progesterone [84], corticosterone [85], and aldosterone [86]. We selected the aromatase inhibitor fadrozole for these experiments to directly compare to a prior fear EMR study that also used fadrozole, and because it is potent and selective for inhibiting synthesis of estrogens (IC₅₀ = 4.5–30 nM depending on the tissue) over progesterone (IC₅₀ = 160 μM), glucocorticoids (corticosterone; IC₅₀ = 100 μM), and mineralocorticoids (aldosterone; IC₅₀ = 1.0 μM) [87]. Fadrozole, as administered here, would inhibit aldosterone synthesis in the BLA, although impacts of aldosterone on opioid seeking or memory have not been determined. In contrast, progesterone and corticosterone synthesis would remain largely undisturbed at this dose. Potential impacts of the minor forms of estrogen (estrone, estriol) cannot be determined here, but are likely minimal relative to E2.

To verify our interpretation of Experiment 1, that fadrozole-induced deficits in heroin-cue EMR are mediated by decreased E2 signaling, we directly blocked E2 signaling in the BLA using ER antagonists, MPP (ERα) and PHTPP (ERβ). As discussed in the Methods, prior studies have determined in vitro and in vivo selectivity and specificity for MPP and PHTPP [47–56]. We used similar doses as these prior studies, ensuring maintenance of this selectivity (see Table 1 for pharmacokinetic parameters). MPP, while a strong ERα antagonist in vitro, was later determined to function more like a selective estrogen receptor modulator (SERM) with mixed agonist and antagonist activity at ERα under certain conditions in vivo [82, 88]. However, even after this metabolic conversion, MPP’s antagonist activity is much stronger than its agonist activity [88] and has been shown to induce anti-ERα effects in vivo [83, 89, 90]. Here, based on E2’s impacts on plasticity and proposed mechanism of action to impact EMR, MPP demonstrates an anti-estrogen effect on EMR in males. This effect is not significant in females, but additional studies using a different ERα antagonist may provide additional insights about the role of ERα in the BLA for both sexes. Notably, PHTPP does not exhibit this SERM-like activity.

Estrous cycle and peripheral hormones

In an initial preliminary experiment, we tracked female estrous cycle via vaginal cytology (Fig. S1). Contrary to findings from fear experiments [91, 92], we did not detect differences in heroin cue EMR when extinguished across different estrous cycle phases. While speculative, we posit differences in the strength of the conditioned associations (footshocks over a single day of conditioning versus heroin over eight days of conditioning) underly these effects. Specifically, in our paradigm, naturally circulating E2 may not be sufficient to enhance EMR to override a heroin-related cue, whereas it is adequate for more weakly-associated cues. This interpretation is strengthened by our finding that inhibition of central E2 synthesis impaired heroin cue EMR (Experiment 1), suggesting central E2 production is more important for heroin cue EMR than peripheral E2. Alternatively, E2 may have a ceiling effect for heroin cue EMR, such that removal of E2 can induce a deficit but further E2 signaling cannot enhance beyond the basal level. While estrous cycle does impact neuronal physiology and estrogen responsivity [78, 93, 94], our primary endpoint (EMR behavior) was not sensitive to cycle (Fig. S1) and therefore did not necessitate cycle tracking at this initial stage. In support of this, in Experiments 1 and 2, variability in female responses during cued extinction and EMR test are comparable to levels of variation in males. While estrous cycle was not tracked in these females, they are expected to be in different cycle phases (between and within independent cohorts). This consistency with male variability and responding between cohorts reinforces the finding that estrous cycle phase does not demonstrably impact heroin cue EMR. Furthermore, even though estrous cycle phase altered fear EMR [27, 92], it did not change the effects of E2 manipulations on the behavioral outcome (i.e., inhibition of E2 signaling impaired fear EMR regardless of cycle).

These interpretations are also strengthened by preliminary experiments testing heroin cue EMR in ovariectomized females, where no difference in EMR was detected relative to males or gonadally intact females (Fig. S1). Our goal here was to evaluate, and possibly uncover differences in, heroin cue EMR in males and females, so we elected to use gonadally intact females for these experiments. Ovariectomy can be a useful tool for investigating sex differences, but it dysregulates expression of ERs in the brain, which were important variables for the present study [39, 41]. We acknowledge that there are potential confounds with this approach, like central effects of circulating estradiol and progesterone, but we believe that the benefit of maintaining normal hormonal physiology and more direct translational applicability outweigh this limitation. Future studies including ovariectomized females may be used to isolate the roles of organization versus activation in the sex-specific roles of ERα and ERβ signaling in the BLA.

Basolateral amygdala and other brain regions

In the present study, we used implanted cannulas to infuse drugs directly into the BLA. This central manipulation defined the BLA as an important locus for E2 signaling in heroin cue EMR and demonstrated that central E2 synthesis is involved. Using this methodology, we cannot exclude contributions of peripheral E2 or other hormones to heroin cue EMR, but our results from estrous cycle and ovariectomized experiments indicate that central E2 production is more important than peripheral E2 for heroin cue EMR in our paradigm (Fig. S1). Furthermore, we are unable to determine if other regions (prefrontal cortex, hippocampus, etc.) are modulated by E2 signaling and contribute to heroin cue EMR.

Several studies describing the role of the BLA in extinction and EMR narrowed our focus on the BLA. For example, inactivation of the BLA after cued extinction training disrupted cocaine cue EMR in male rats [95], whereas BLA inactivation prior to a test suppressed heroin-cued reinstatement [96]. Additionally, while not direct evidence, brain activation patterns following fear EMR also suggest an important role for the amygdala. In one study, c-Fos was decreased in the amygdala of females who received E2 immediately after extinction [28]. Another study found several differentially activated amygdala subregions between E2 and vehicle treated females after cued extinction and EMR test, but no changes in prefrontal or hippocampal activity [35]. These activity patterns support the notion that E2 signaling in the BLA could similarly alter fear EMR as we have shown for heroin cue EMR.

In one prior study examining regional effects, the authors identified the hippocampus as a key locus for ERβ signaling to enhance contextual fear EMR in females [34]. To our knowledge, this is the only other region-specific manipulation of E2 signaling and EMR that has been published. The hippocampus is vital for contextual memory processes but plays a lesser role relative to the BLA in cued memory [97]. Here, our primary goal was to assess cued EMR, but contextual memories also contribute to drug seeking behaviors. Further studies under a contextual extinction protocol and/or direct hippocampus manipulations would be required to determine if E2 signaling impacts contextual heroin EMR as with fear.

Operant and classical conditioning

Throughout, we have compared our methodology and results to those from fear conditioning studies. We have already noted the different valence of the unconditioned stimulus (aversive versus rewarding) as one factor that limits direct comparison, but an additional consideration is the conditioning method used. Fear studies use classical (Pavlovian) conditioning, whereby the cue and shock are presented without any action from the subject. The conditioned response, freezing, is measured to assess the strength of the stimulus-stimulus association. Contrastingly, our paradigm is based on operant conditioning principles, whereby the heroin infusion and coincident presentation of the cues are contingent on operant responding from subjects (nose pokes), forming a response-outcome association. A Pavlovian association is still formed between the cue and the drug, but operant responding (i.e. number of nose pokes) measures the strength of the association. For both valences of unconditioned stimuli (aversive and rewarding) and conditioning approaches (classical and operant), the BLA is functionally implicated upstream of the behavioral response [98]. In other words, the association and extinction of the conditioned stimulus and the unconditioned stimulus occur in the BLA for both classical fear conditioning and operant self-administration. The BLA then signals to other regions to initiate the appropriate behavioral response, after which the regions involved differ between each methodology.

Non-drug rewards and unreinforced cues

Given the BLA’s role in forming and extinguishing conditioned associations for positive and negative stimuli, additional investigation into E2’s impacts on EMR of cues associated with non-drug rewards (food, sucrose) or aversive stimuli (high-dose cocaine, quinine solution) could better explain both the BLA’s function and how E2 modulates EMR for different stimuli. Additionally, though the BLA is thought to be recruited only by motivationally relevant stimuli, exploration of E2’s impacts on EMR of cues associated with unreinforced ‘neutral’ stimuli (saline) is also warranted.

Conclusions

The results from these experiments demonstrate that E2’s impacts on EMR are not limited to fear, but also extend to heroin cued EMR. Importantly, we identified that centrally-produced (i.e., synthesized within the brain) E2 underlies this effect, that the BLA is an important locus for E2 signaling during extinction in both sexes, and that sex-specific ER mechanisms are responsible for these effects. Future studies should determine if E2 signaling in the BLA impacts contextual reward EMR, what impacts E2 has on non-drug rewards, whether enhanced E2 signaling can enhance EMR, and the molecular differences in the BLA that mediate sex-specific ER contributions to heroin cue EMR. A better understanding of the mechanisms underlying EMR and E2’s impacts could inform better treatments to prevent relapse in OUD, like pharmacotherapy-assisted CET.

Data Availability Statement

Data is available upon request, subject to legal and ethical guidelines. Data inquiries can be directed to the corresponding author.