Depleting retinoic acid synthesis in the nucleus accumbens shell produces a protective phenotype for emotional reactivity and drug-taking in rats

Abstract

Rationale:

In previous work, a convergent transcriptomic approach strongly suggested a role for retinoic acid (RA) in controlling the emotion- and reward-related functions of the nucleus accumbens shell (NAcSh).

Objective:

Here, we causally assess the role of NAcSh RA in controlling anxiety-, emotion-, and reward-related behavior in rats and explore cellular mechanisms that may underlie this phenotype.

Methods:

Rats underwent bilateral knockdown of the retinoic acid synthesis enzyme Aldh1a1 in the NAcSh. Anxiety-related behavior was assessed using open-field exploration, elevated plus maze, and sucrose neophobia tests. Emotion-related behavior was assessed via sucrose preference, post-isolation social contact, and forced swim tests. Animals were subsequently allowed to self-administer fentanyl to assess reward- and frustration-related behavior. In parallel, electrophysiological testing of medium spiny neurons (MSNs) in the NAcSh was used to explore the role of RA in NAcSh cellular function.

Results:

We observed an anxiety-vulnerable, depression-resilient phenotype in knockdown animals compared to controls. During operant tasks, knockdown animals took fewer fentanyl infusions during FR5 maintenance and showed decreased demand intensity in behavioral economics sessions. Finally, electrophysiological assessment of NAcSh MSNs revealed attenuated excitability following Aldh1a1 knockdown. Altogether, our findings reveal a key role for NAcSh RA signaling in determining emotional resilience and drug-taking, likely via decreased MSN excitability.

Conclusions:

Our results posit the RA synthesis enzyme Aldh1a1 as a promising therapeutic target for depression-, frustration-, and addiction-associated disorders. This is the first report linking RA to frustrative nonreward, the NAcSh to operant frustration, and RA to fentanyl drug-taking behavior.

1. Introduction

Retinoic Acid (RA) is a vitamin A metabolite that exerts both immediate and transcriptional regulation of neurobiological functions (Vilhais-Neto and Pourquié 2008). A potential role for retinoids in increasing emotionality was first suggested due to associations between retinoid intake and depression-related side effects in humans (Barak et al. 2005; Zhang et al. 2022). However, administration of these in animals has led to inconsistent findings (Ferguson et al. 2005; O’Reilly et al. 2006). Whether, or how, RA can increase emotionality has drastic implications for existing retinoid prescription and usage, and conversely may also represent novel targets for anti-depressant pharmacotherapeutic development.

Previously, a convergent transcriptomics approach led us to identify RA signaling in the nucleus accumbens shell (NAcSh) as a promising target. Firstly, we found RA signaling components to be specifically and densely expressed in the NAcSh, suggesting a role for RA in controlling functions of this region (Zhang et al. 2016a). Moreover, RA signaling in the NAcSh was regulated by protection- and risk-associated conditions for depression-related behavior and drug-taking (Zhang et al. 2016b). Direct manipulation of RA signaling in the NAcSh by knocking down the degradation enzyme Cyp26b1 produced a robust phenotype of increased depression-related behavior and drug-taking in rats (Table 1; Zhang et al. 2016b, 2019). As with other genetic manipulations in the NAcSh, anxiety-related behavior was inversely regulated with respect to emotionality and drug-taking (Table 1; Zhang et al. 2019). These findings highlight the NAcSh as a potential site of action for retinoid-mediated emotionality, and may explain inconsistent findings of emotion-related phenotypes in studies of oral retinoid administration.

Table 1.

Overview of current results in context of previous work (* = trend; ^# = frustration-related behavior; NC = no change).

Manipulation	Drug Self-administration	Depression-related behavior	Anxiety-related behavior	MSN excitability	Citation
Resilient Depression/Drug-Taking with Susceptible Anxiety Phenotype
Environmental enrichment	↓ ↓ ↓ ↓	↓ ↓ ↓	↑ ↑	↓	Green et al., 2010
CREB inhibition	↓ ↓	↓ ↓ ↓	↑ ↑ ↑	↓	Dong et al., 2006; Green et al., 2010; Larson et al., 2011
ΔFosB overexpression	↓ ↓ ↓	↓	NC		Zhang et al., 2014; Wallace et al., 2008
ATF2 overexpression		↓ ↓	↑		Green et al., 2008
Aldh1a1 shRNA	↓↓	↓↓↓#↑	↑↑↑	↓	Current manuscript
Susceptible Depression/Drug-Taking with Resilient Anxiety Phenotype
GSK3 shRNA	↑↑	↑↑	↓ ↓	↑	Crofton et al., 2017
ATF3 overexpression		↑	↓ ↓		Green et al., 2008
ATF4 overexpression		↑↑	↓ ↓		Green et al., 2008
Cyp26b1 shRNA	↑↑↑↑	↑	↓		Zhang et al., 2016 Zhang et al., 2019
Inconsistent or No Change in Phenotype
Crabp2 shRNA	NC	↑	↑*		Zhang et al., 2019
Fabp5 shRNA	↓↓↓	NC	NC	↑↓	Zhang et al., 2019 Crofton et al., 2021

Arrows represent the direction of the effect, the number of arrows corresponds to the number of assays where an effect was detected. All experiments except the current evaluated cocaine self-administration. Previous manipulations associated with retinoic acid are highlighted in light blue, the current manipulation is highlighted in yellow.

Overall, prior evidence positions the RA synthesis enzyme Aldha1a1, the subject of the current manuscript, as a promising target for pharmacotherapeutic development. Here, we examine the bidirectionality of RA control of emotion and drug-taking by virally knocking down the Aldh1a1 enzyme in the NAcSh of adult rats. We expected decreased emotionality and drug-taking, which we assessed using a classic battery of tests for depression-related behavior followed by operant fentanyl self-administration. Additionally, we leveraged duration of bar pressing during operant tasks of extinction and progressive ratio for the evaluation of frustration-related behavior, a sensitive and validated proxy of frustration-based emotional reactivity (Vasquez et al. 2021a, c; Marmol Contreras et al. 2024). The integration of novel tasks of reactivity with motivation-related operant behavior and classic behavioral tasks allows us to demonstrate a role of RA signaling in the NAcSh in determining depression-related behavior, drug-taking, and for the first time ever, frustration-based reactivity in rats. Finally, we examine the effect of Aldh1a1 knockdown on MSN excitability by using patch clamp electrophysiology to elucidate cellular mechanisms that may mediate the behavioral phenotypes observed.

The term “emotionality” refers to the extent to which one experiences emotions (positive or negative) and expresses these in measurable behavior. Though the “experience” of and “reactivity” to emotions represent distinct dimensions of emotionality (Izard 2010). Animal research is limited to the “reactivity” component, as there is no way to assess an animal’s internal experience. Therefore, within this manuscript “emotionality” is used interchangeably with “emotional reactivity” to describe the degree to which animals alter their behavior in assays designed to invoke negative emotions.

In addition, within this manuscript we discuss emotionality separately from anxiety-related behavior. Though we acknowledge that many think of anxiety as an emotion, and that anxiety can arise from one or more emotions (Arroll and Kendrick 2018), we hold that there is an important distinction between emotionality and anxiety. Both are determined by psychological, cognitive, and physiological components (Izard 2010; Arroll and Kendrick 2018; Papini et al. 2024). However, anxiety is an adaptive response to a potential threat (Arroll and Kendrick 2018), not an emotion, and it is understood to be regulated by distinct circuitry from depression-related behavior (Hyman & Nestler, 1993; Janicak, 2002). This manuscript presents further evidence that these systems are not always co-regulated in circuitry, furthering the argument that grouping anxiety under emotionality may undermine progress within the field by misinforming experimental design, interpretation of findings, and the long-term development of neuropsychiatric therapeutics.

2. Methods

2.1. Animals

Two cohorts male Sprague-Dawley rats (Harlan, Houston, TX) were obtained at 225–250g. Rats were randomly assigned to a control or knockdown condition (n = 8–12 per group). Rats had ad-libitum access to food and water except during surgery and behavioral experiments. Rats were maintained in a controlled environment (22 °C; 50% relative humidity; 12 h light/dark cycle, lights on 0600 h) in an Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC) approved colony. Rats were pair-housed in standard polycarbonate cages except during specified behavioral tests. All procedures conformed to the NIH Guide for the Care and Use of Laboratory Animals and approved by The University of Texas Medical Branch Institutional Animal Care and Use Committee. The sample size for cohort 1 was n = 8 control animals and n = 10 knockdown animals. The sample size for cohort 2 was n = 12 animals per group.

2.2. In vitro knockdown of Aldh1a1 by adeno-associated viral vector

Like previous studies from our laboratory (Zhang et al. 2019), an AAV vector was designed and constructed using RNA interference to knock down expression of our target gene, Aldh1a1 in vivo. Five shRNA sequences were designed to target the rat Aldh1a1 coding region and each co-transfected into HEK293 cells with a plasmid overexpressing rat Aldh1a1 in triplicate. Cells were harvested after 24 hours and processed for mRNA using the Qiagen RNeasy kit. RNA was reverse transcribed using the Stratagene First-Strand cDNA kit. cDNA was quantified using SYBR green qPCR (Applied Biosystems) using relative quantitation from the ΔΔ_CT method using the following primers: forward: 5’ AAGAAGGGGACAAGGCAGAT 3’, reverse: 5’ CAAGTACGCATTGGCAAAGA 3’. Primers for Gapdh were used to normalize transcript levels across samples: forward: 5’ GACATCAAGAGGTGGTGAA 3’, reverse: 5’ TGTCATACCAGGAAATGAGC 3’.

2.3. In vivo knockdown of Aldh1a1 by adeno-associated viral vector

To knockdown Aldh1a1 in vivo (Wunsch et al. 2024), rats were anesthetized with isoflurane (VetEquip, Pleasanton, CA) and injected bilaterally into the NAcSh (A/P = 1.7, L = 2.2, D/V = −6.7, 10 degree lateral angle) with 1 μL of AAV2-shAldh1a1, per side, over a period of 10 minutes. Control rats received bilateral injections of equal volumes of AAV2-shCTRL expressing a non-targeted hairpin (Benzon et al. 2014). After bilateral injection, needles remained in place for an additional 10 minutes to allow for vector to disburse. At the conclusion of the study, all brains were extracted, and accurate placements were confirmed by visualizing the native enhanced green fluorescent protein (eGFP) with a Dual Fluorescent Protein Flashlight and VG2 barrier filter glasses (Nightsea, Bedford, MA) (Zhang et al. 2019). No animals were excluded based on vector placement. For AAVs, maximal expression is achieved by three weeks and has stable expression up to a year, which encompasses the duration of the studies (Green and Nestler 2006).

2.4. Immunofluorescence and microscopy

Animals were deeply anesthetized using isoflurane and euthanized using a guillotine. Brains were extracted and immediately placed in 4% paraformaldehyde overnight at 4°C. Next, brains were immersed in a 40% glycerol solution until they sank. Coronal sections (35 μm) of the NAcSh were sliced at −20°C and collected directly into a 1% PBS solution. For staining, intact slices were mounted and blocked in 3% normal donkey serum (in a 0.3% Triton PBS solution) for 1 hour at room temperature. Slices were then incubated with a primary antibody (1:2000, anti-GFP Antibody, Chicken IgY, Avēs Labs Inc.) overnight at room temperature. Next, slices were washed with 0.1% PBS 3 times and incubated with the secondary antibody (1:2000, Alexa Fluor 488 conjugated Donkey anti-Chicken IgY, Invitrogen), for 2 hours at room temperature. Finally, slices were washed with 0.1% PBS 4 times and cover slipped using a mounting medium (ProLong Gold Antifade Mountant, Invitrogen). Images were acquired using the Leica True Confocal Scanner SPE and Leica Application Suite Advanced software (Leica Microsystems, Wetzlar, Germany). Images were collected at a 20X magnification. For the final image, the over-exposure dial was set to 137 to enhance the visibility of landmarks.

2.5. Behavioral assays following vector injections

Following a 3-week recovery from vector injections, rats were subjected to a battery of behavioral tests conducted in order of increasing perceived stress (Fig. 1) such that previous tests had less probability of affecting the results of the next test. Animals in cohort 1 underwent open field exploration followed by the elevated plus maze (EPM) test. Animals in cohort 2 underwent the open-field exploration, sucrose neophobia and preference, post-isolation social contact, and forced swim tests, followed by operant self-administration tasks described below. Where applicable, apparatuses were cleaned with cavicide between animals.

Fig. 1.

Sequence of experimental procedures for behavioral battery and fentanyl self-administration. Cohort 1 completed the open-field exploration and elevated plus maze tasks. Cohort 2 completed the open-field exploration, sucrose neophobia, sucrose preference, post-isolation social contact, and forced swim tests before proceeding to fentanyl self-administration tasks.

2.6. Assessment of anxiety-related behavior

2.6.1. Open field

The open field test was used to assess differences in locomotor activity and anxiety-related behavior. The test was conducted under normal light conditions by placing the rats in a novel open field arena (clear plastic container, 45 × 40 × 45 cm) with a thin layer of bedding, where they were allowed to move around freely for 30 minutes. The movement of the rats for this period was monitored from above using a video tracking system (Anymaze from Stoelting Inc., Wood Dale, IL, USA). Chambers were cleaned with Cavicide between animals and bedding was replaced. The total distance traveled was used as a measure of locomotor activity. Generally, animals will spend more time on the edges of an arena in high anxiety situations (Belzung and Le Pape 1994). Therefore, the middle of each box was designated as the center zone (20 × 25 cm) in Anymaze prior to running the assay, and time spent in this zone was interpreted as an inverse measure of anxiety-related behavior.

For the control group, the initial sample size for cohort 1 was n = 8 animals and n = 12 animals for cohort 2. For the knockdown group, the initial sample size for cohort 1 was n = 10 and cohort 2 was n = 12 animals. For cohort 1, data for two animals in each group was not recorded due to tracking issues. One additional animal in the shAldh1a1 group was an outlier and therefore excluded from analysis. For cohort 2, two outliers in the shAldh1a1 group were identified and excluded from analysis. The final combined sample size for analysis was n = 18 for controls and n = 17 for shAldh1a1 animals. Since experimental conditions were identical for Cohort 1 and Cohort 2, data for this test were combined for analysis.

2.6.2. Sucrose neophobia

The sucrose neophobia assay also served as a test of anxiety-related behavior. Rats were exposed to a 1% sucrose water solution for the first time and allowed to consume freely for 30 minutes. Rats were separated into individual cages and their normal water was removed at 1600 h. A 1% sucrose water solution was prepared and used to fill standard rat water bottles. These were carefully weighed and placed in each cage at lights out (1800 h). After 30 minutes, the bottles were carefully withdrawn to ensure minimal spillage and re-weighed to calculate by subtraction the total amount of the 1% sucrose solution consumed by each rat in grams. Researchers thoroughly tested every bottle used in this experiment prior to determining the pre-assay weight to ensure minimal leakage and remove defective equipment. In addition, researchers practiced placing and removing bottles from cages prior to the experiment to ensure minimal spillage. The final sample size for this test was n = 12 per group.

2.6.3. Elevated plus maze

The elevated plus maze (EPM) assessed anxiety-related behavior in cohort 1 animals. The apparatus (Noldus Information Technology, Leesburg, VA, USA) consisted of two open arms and two closed arms (12 × 50 cm, 75 cm above the floor). Under low light conditions during the dark phase of the light cycle, animals were placed in the center facing a closed arm and allowed to explore the maze freely for 5 minutes. The time spent in open, closed, and center zones was monitored based on the center of the rat’s body using AnyMaze. The final sample size for this test was n = 8 animals in the control group and n = 10 animals in the knockdown group.

2.7. Assessment of depression-related behavior

2.7.1. Sucrose preference

Following the sucrose neophobia test, bottles containing the 1% sucrose solution were provided for the next 48 hours instead of normal water to enable habituation to the sweet solution. This was followed by 48 hours of access to normal water. Again, pair-housed rats were separated at 1600 h with access to food but not water. After 2 hours, two pre-weighed bottles were placed in each cage, one containing water in the usual water position and the other containing 1% sucrose water a few inches away. Rats were allowed to consume either solution freely for 16 hours, and bottles were re-weighed to determine consumption by subtraction of standard water and 1% sucrose solution. This assay models anhedonia, defined low positive affect for pleasurable activities (Der-Avakian and Markou 2012; De Fruyt et al. 2020). A decreased preference for sucrose relative to water (consumption ratio) is interpreted as decreased positive affect since, generally, healthy animals will develop a preference for a slightly sweet solution compared to plain water (Primo et al. 2023). As with the sucrose neophobia test, researchers thoroughly tested every bottle used in this experiment prior to determining the pre-assay weight to ensure minimal leakage and remove defective equipment. In addition, researchers practiced placing and removing bottles from cages prior to the experiment to ensure minimal spillage. The final sample size for this test was n = 12 per group.

2.7.2. Post-isolation social contact

This task models the social withdrawal aspect of depression. Rats were separated from their cage mates for 24 hours then reunited in an open-field (plastic container, 45 × 40 × 45 cm) arena and allowed to interact freely with the environment and each other for 30 minutes. Videos were manually scored for allo-grooming by an investigator blinded to the experimental condition. Periods where rats mutually groomed (i.e. not self-grooming) were considered allo-grooming. This test was scored in pairs to account for variability in allo-grooming related to dominance dynamics. The final sample size for this test was n = 6 per group because animals are scored in pairs (i.e. n = 12 total per group). Decreased allo-grooming in this test is interpreted as increased depression-related behavior.

2.7.3. Forced swim test

Rats were placed into a Plexiglas cylinder (Diameter: 10”, H:18”) filled with approximately 14 L of room temperature water (24 ± 0.5 °C). After each session, the rats were patted dry and returned to their home cages. Two sessions, 10 minutes and 5 minutes in length, were conducted 24 hours apart. Latency to first immobility and total time immobile was scored manually by two blinded researchers. Inter-rater reliability was assessed using Pearson correlations. Increased time immobile in this task is interpreted as increased depression-related behavior. The final sample size for this test was n = 12 per group.

2.8. Operant tasks of drug self-administration

Upon completion of the behavioral battery assessing anxiety- and depression-related behavior, Cohort 2 animals were implanted with intrajugular catheters and trained to self-administer cocaine. Two animals in the control group and one animal in the knockdown group were lost during surgery. During training, 4 animals in the control group and two animals in the knockdown group lost patency. The final sample size for all self-administration tasks was n = 6 control animals and n = 9 shAldh1a1 animals.

2.8.1. Intrajugular catheter implantation

Rats were anesthetized with ketamine (100 mg/kg IP) and xylazine (10 mg/kg, IP) and implanted with indwelling intrajugular silastic catheters as described previously (Zhang et al. 2016b; Crofton et al. 2017). To maintain catheter patency, catheters were flushed daily with 0.1 ml of heparinized (10 U/ml) saline with ticarcillin (0.067g/ml). Following a 7-day recovery period from catheter surgery, animals proceeded to self-administer fentanyl HCl (32 μg/kg/infusion; NIDA Drug Supply Program).

2.8.2. Operant schedules

Acquisition: Rats self-administered fentanyl for 3 hours under a fixed ratio 1 (FR1) schedule (32 μg/kg/infusion), where each infusion required one bar press by the rat. After the first 10 sessions, FR1 sessions were shortened to 2 hours for the next 5 sessions.

Maintenance responding: Following acquisition sessions, rats were allowed to self-administer fentanyl on an FR5 contingency for 2 hours each day until rats were responding stably. Maintenance sessions were also used to maintain responding prior to extinction and progressive ratio sessions. Rats were considered stable when there was less than 20% variability across 3 sessions, a 2:1 ratio of active to inactive lever presses, and at least 10 infusions/session. For extinction, progressive ratio, and FR5 sessions between extinction and progressive ratio sessions, stability was not required, but infusions needed to be higher than 10 per session.

Behavioral economics (BE): Every 20 minutes, doses were decreased to assess fentanyl demand intensity and elasticity (progression rounded to the hundredth place: 3.20, 3.20, 1.80, 1.01, 0.57, 0.32, 0.18, 0.10, 0.06, 0.03, 0.02, and 0.01 μg/kg/inf). The first dose was not analyzed, as always, due to loading effects (Marmol Contreras et al. 2024). The last three doses were not analyzed due to early extinction in a substantial portion of the animals.

Between-session extinction (EXT): Following stability in maintenance sessions, rats were allowed to attempt self-administration under a 3-hour extinction program. A normal contingency of FR5 was used to deliver reinforcement-associated cues every 5 presses but infusions were completely omitted. Rats underwent a total of 6 EXT sessions.

Progressive ratio (PR): Each successive reinforcement required an increasing number of bar presses according to a semi-logarithmic progression until rats spent 1 hour without achieving a reinforcement (Richardson and Roberts 1996). As with previous work, rats underwent a total of 3 PR sessions, with the first day of PR is excluded from analysis since the individual rat’s learning curve is a significant confound (Marmol Contreras et al. 2024). The intermediate FR5 sessions preceding PR were also not included in analysis.

2.8.3. Frustration-related operant behavior:

To assess emotional reactivity, duration of bar pressing was documented during frustrating programs (EXT, PR). Standard MED-PC operant chambers were reconfigured to level mode, enabling us to monitor the duration of each recorded bar press to the hundredth of a second (Vasquez et al. 2021b; Marmol Contreras et al. 2024). Increased duration of pressing during frustrating operant tasks has been validated as a proxy for frustration-related behavior in rats self-administering natural and drug rewards, allowing us to monitor motivation- and frustration-related behavior simultaneously (Vasquez et al. 2021a; Marmol Contreras et al. 2024).

2.9. Electrophysiology following Aldh1a1 knockdown

After at least 3 weeks following stereotaxic surgery, 300 μM coronal brain slices containing the NAc were prepared as previously described (Dvorak et al. 2023). After slices were prepared, they were stored in a recovery chamber filled with warm (37 ℃) and continuously oxygenated (95%/5% O₂/CO₂) standard artificial cerebrospinal fluid (aCSF) comprised of the following salts: 123.9 mM NaCl, 3.1 mM KCl, 10 mM glucose, 1 mM MgCl₂, 2 mM CaCl₂, 24 mM NaHCO₃, and 1.16 mM NaH₂PO₄ (Osmolarity = 300–310 mOsm; pH = 7.4). After at least 30 min of recovery, slices were transferred to a recording chamber perfused with heated and continuously oxygenated aCSF. Whole-cell patch-clamp recordings were performed in GFP positive MSNs of the NAcSh. Recording pipettes (3–5 MΩ) were filled with an internal solution comprised of the following salts: 45 mM K-gluconate, 2 mM MgCl₂, 0.1 mM EGTA, 2.5 mM Na₂ATP, 0.25 mM Na₂GTP, 5 mM phosphocreatine, and 10 mM HEPES (Osmolarity = 290 mOsm; pH = 7.2). After GΩ seal formation and entry into the whole-cell configuration, capacitive transients and series resistances were compensated for by 70–80%. Then, MSNs were held in I = 0 mode for approximately 2 min to determine resting membrane potential (RMP). The amplifier was then switched to current-clamp mode, and intrinsic neuronal excitability was assessed by measuring evoked action potentials (AP) in response to current injections ranging from – 20 pA to + 200 pA with 800 ms pulses and a change in injected current of 10 pA between sweeps. Recordings were performed using an Axopatch 200b amplifier (Molecular Devices, Sunnyvale, CA, USA). Prior to digitization and storage, data were acquired at 20 kHz and filtered at 5 kHz. Clampex 9.2 software (Molecular Devices), interfaced to electrophysiological equipment using a Digidata 1200 analog-digital interface (Molecular Devices), was used to control all experimental parameters. The final sample size for electrophysiology was n = 10 cells per group, sourced from n = 3 animals per group.

2.10. Statistical analysis

For all behavioral data containing more than two observations, repeated measures one-way ANOVA or a repeated measures two-way ANOVA were conducted as appropriate. Post-hoc analyses used Tukey’s multiple comparisons tests. Data containing two observations were evaluated using independent or paired student’s T-tests, as appropriate. Welch’s corrections were used for T-tests that violated the sphericity assumption. Outliers more extreme than 1.5 interquartile ranges beyond upper and lower bounds were excluded from analysis (discussed in results when applicable). Normality was determined by Shapiro-Wilk tests. When the normality assumption was violated, non-parametric tests were used for analysis. All statistics were performed using GraphPad 9 software. Data for the electrophysiological studies were analyzed using pCLAMP 10 and GraphPad Prism 9 software (La Jolla, CA, USA) as previously described (Dvorak et al. 2023).

To obtain demand intensity and demand elasticity from the behavioral economics data, we first obtained groupwise row averages and standard errors for each dose. Then, the following equation was fit to the row data with the k constant set to 0.9: $Y=Q_0*{10}^{k{e}^{(-\alpha Q_{0}c)}-1}$ (Koffarnus et al. 2015). The value for the k constant was chosen because it provided the best fit of the model.

3. Results

3.1. In vitro and in vivo vector validation

Fig. 2A shows relative knockdown efficiency of the five hairpins tested in HEK cells. Hairpin H2 (5’ TCAACCCAGGTTGGCAAATTAATC 3’) was chosen for behavioral experimentation and was packaged into an AAV2 capsid (UNC Vector Core) with hairpin expression driven by a mouse U6 promoter and co-expressing eGFP from a CMV promoter (shAldh1a1) (Zhang et al. 2019). The chosen hairpin, H2, significantly decreased Aldh1a1 mRNA expression (F(6, 14) = 11.69, p < 0.001).Fig. 2B shows a representative placement and titer of the viral shAldh1a1 vector (eGFP immunofluorescence) in the NAcSh.

Fig. 2.

Knocking down Aldh1a1 in the nucleus accumbens shell produces anxiety-vulnerable phenotype. (A) In-vitro assessment of Aldh1a1 mRNA knockdown efficacy of the five knockdown hairpins designed (H1-H5). Positive control (PC) samples contained the overexpression hairpin without any knockdown hairpin. Negative control (NC) samples did not contain any overexpression hairpin. All other samples contained the over-expression hairpin and the indicated knockdown hairpin. The H2 hairpin was chosen for packaging. (B) Representative placement and titer of viral shAldh1a1 vector (eGFP immunofluorescence) in the NAcSh (NAcSh and midline outlined, AC=anterior commissure). (C) No differences across groups were detected in total distance travelled during the open field exploration test. (D) However, shAldh1a1 rats spent less time in the center zone in this task. (E) shAldh1a1 animals also spent less time the open arms during the elevated plus maze test compared to shCTRL rats. (F) Finally, shAldh1a1 animals consumed less 1% sucrose solution upon first exposure (neophobia). Data shown are the mean ± SEM. For A, triplicates were averaged and statistical significance was assessed using a one-way ANOVA. For C-F, statistical significance between groups was assessed using independent t-tests. Asterisks (*) are used to indicate pairwise significance: *p <0.05, **p <0.01, ***p < 0.001

3.2. Knocking down Aldh1a1 expression in the NAcSh did not alter locomotor activity

For cohort 1, data for two animals in each group was not recorded due to tracking issues. One additional animal in the shAldh1a1 group was an outlier and therefore excluded from analysis. For cohort 2, two outliers in the shAldh1a1 group were identified and excluded from analysis. The final group sizes were n = 17–18 per group. Rats did not differ significantly in their mean distance traveled during the open field test across groups (t(33) = 0.746, p = 0.461; Fig. 2C). Thus, differences in subsequent spontaneous behavior assays are unlikely to be related to changes in general activity.

3.3. Effects of knocking down Aldh1a1 in the NAcSh on anxiety-like behavior

For the open-field exploration test, data from two behavioral cohorts were combined because the experimental conditions, other than the date that the test was conducted, are identical for these groups. Analysis revealed that Aldh1a1 knockdown rats spent less time in the center zone than shCTRL rats (t(33) = 2.247, p = 0.031; Fig. 2D). Individually, each cohort shows a trend in the same direction (Cohort 1: t(11) = 2.027, p = 0.068; Cohort 2: t(17.3) = 1.781, p = 0.092). Knockdown rats also spent significantly less time in the open arms during EPM compared to control rats (t(16) = 2.448, p = 0.026, Fig. 2E). Finally, knockdown animals consumed less sucrose during the neophobia task compared to controls (t(21) = 2.859, p = 0.009; Fig. 2F). All three of these results suggest increased anxiety-related behavior in experimental animals.

3.4. Effects of knocking down Aldh1a1 in the NAcSh on emotional reactivity

During the sucrose preference test, a main effect of group was not detected for amount of water and sucrose consumed (F(1,44) = 0.763, p = 0.386; Fig. 3A). In the post-isolation allo-grooming task, shAldh1a1 animals spent less time allo-grooming (t(10) = 2.346, p = 0.041; Fig. 3B), classically interpreted as increased depression-related behavior. For the forced swim test, a Shapiro-Wilk test of normality revealed that latency was not normally distributed in either group (shCTRL: W = 0.838, p < 0.001; shALDH1a1: W = 0.842, p < 0.001). Similarly, total immobility was not normally distributed in either group (shCTRL: W = 0.813, p < 0.001; shALDH1a1: W = 0.849, p < 0.001). Therefore, non-parametric tests were used to analyze data for the forced swim test. A Mann-Whitney U test revealed increased latency to first immobility (U = 32, p = 0.021; Fig. 3C) and decreased total immobility (U = 38, p = 0.049; Fig. 3D), in shAldh1a1 rats compared to shCTRL rats on the second day of the FST, suggesting decreased depression-related behavior in Aldh1a1 knockdown rats. Pearson correlations were used to verify inter-rater reliability of FST scoring (Latency: R² = 0.432, p = 0.035; Immobility: R² = 0.527, p = 0.008).

Fig. 3.

Aldh1a1 knockdown in the nucleus accumbens shell affects depression-related behavior. (A) Both groups consumed significantly greater amounts the 1% sucrose solution than normal water, with no differences detected across groups in the amount of water or 1% sucrose solution consumed. (B) During the post isolation allo-grooming task, shAldh1a1 rats spent less time grooming compared to shCTRL rats. In the forced swim test, shAdh1a1 rats exhibited an anti-depressive phenotype (C-D). (C) shAldh1a1 rats had increased latency to first immobility, as well as (D) decreased total time immobile than shCTRL rats. Data shown are the mean ± SEM. Statistical significance between groups was assessed using independent t-tests. Asterisks (*) are used to indicate pairwise significance: *p <0.05, **p <0.01, ***p < 0.001.

The operant frustration effect refers to the increase in duration of bar presses exhibited by over 90% of animals undergoing operant frustration (Marmol Contreras et al. 2024). Evaluation of this effect in shAldh1a1 animals using a 2-way repeated measures ANOVA revealed decreased duration of pressing compared to control animals during EXT (Main effect: F(1,13) = 5.51, p = 0.036; Group*Day interaction: F(6,78) = 4.054, p = 0.001; Fig. 4B) and PR (Main effect: F(1,13) = 1.959, p = 0.185, Group*Day interaction: F(1,13) = 5.775, p = 0.032; Fig. 4D) sessions, agreeing with the above evidence of emotional resilience during FST. Importantly, duration of bar pressing did not differ across groups in the low frustration maintenance FR5 sessions (F(1,13) = 0.219, p = 0.648; Fig. 5B), increasing our confidence that the effect noted in Fig. 4 is related to frustration.

Fig. 4.

Knocking down Aldh1a1 in the nucleus accumbens shell blunts operant emotional reactivity. (A, C, E) There were no differences in drug-seeking during frustration across groups. Rats achieved a similar number of infusions for EXT (A) and PR (C) and had a similar PR breakpoint (E). (B, D) Knockdown animals exhibited attenuated emotional reactivity during frustrating operant tasks. (B) For EXT, shAldh1a1 animals had significantly lower bar press durations than shCTRL rats. (D) During PR, a significant day*group interaction revealed decreased bar press durations in shAldh1a1 animals for day 3 of PR compared to shCTRL rats. Data from FR5 and PR Day 1 are shown, but as discussed are excluded from PR analysis. Data shown are the mean ± SEM. Asterisks (*) are used to indicate pairwise significance: *p <0.05, **p <0.01, ***p < 0.001. For ANOVA results, # indicates p <0.05 in a main effect and ¥ indicates a significant interaction (¥ is p <0.05, ¥¥ is p <0.01).

Fig. 5.

Knocking down Aldh1a1 synthesis in the nucleus accumbens shell decreases drug-taking during low-effort conditions. (A) During FR5 sessions, shAldh1a1 trended toward fewer number of infusions. (B) There were no differences in average bar press duration during FR5. (C) Fentanyl demand curve fitted across 7 fentanyl price points (number of responses for 1 μg/kg of fentanyl) for each groups. (D) Groupwise fentanyl demand intensity and (E) groupwise demand elasticity (i.e., slope) derived from (C). Data are mean ± SEM. Asterisks (*) are used to indicate pairwise significance, *p <0.05.

3.5. Effects of knocking down Aldh1a1 in the NAcSh on drug-taking

During maintenance FR5 sessions, shAldh1a1 animals trended toward a main effect of decreased number of infusions compared to shCTRLs (F(1,13) = 4.409, p = 0.055; Fig. 5A). In a behavioral economics assessment (Fig. 5C–E), shAldh1a1 animals showed decreased fentanyl demand intensity (Q₀, consumption at cost “zero”; t(13) = 2.238, p = 0.043; Fig. 3D), despite unaltered demand elasticity (t(13) = 0.087, p = 0.932; Fig. 5E). During EXT (F(1,13) = 1.610, p = 0.227; Fig. 4A) and PR (F(1,13) = 1.674, p = 0.218; Fig. 4C) sessions, no differences in number of infusions were detected. There were also no differences in PR breakpoint (F(1,13) = 2.282, p = 0.155; Fig. 4E) across groups. These results suggest decreased drug-taking at baseline and low-effort conditions despite unaltered drug-seeking.

3.6. Knockdown of Aldh1a1 in the NAcSh perturbs excitability of MSNs

To shed light on electrophysiological changes that could underlie the behavioral consequences of knocking down Aldh1a1 in the NAcSh, patch-clamp recordings were performed in MSNs in slices from rats previously treated with shCTRL or shAldh1a1. Importantly, MSNs represent ~95% of the total cell volume of the NAcSh and provide the sole output of the brain structure (Castro and Bruchas 2019). MSNs expressing shAldh1a1 displayed depolarization block (Fig. 6A), characterized by a lessened ability to fire repetitive APs at current steps greater than 80 pA relative to shCTRL-expressing MSNs (Fig. 6B; refer to Supplementary Table 1 for detailed statistical information), not accompanied by a change in AP kinetics (max rise: t(18) = 0.828, p = 0.418; max decay: t(18) = 0.828, p = 0.735; Fig. 6C–D) or the voltage threshold for AP initiation (t(18) = 1.748, p = 0.098; Fig. 6E). Related to passive electrical properties, MSNs expressing shAldh1a1 displayed an increased input resistance compared to MSNs expressing shCTRL (t(18) = 2.830, p = 0.011; Fig. 6G), whereas the RMP was unaffected (t(18) = 0.740, p = 0.469; Fig. 6F).

Fig. 6.

Knockdown of Aldh1a1 in the NAcSh perturbs excitability of MSNs. (A) Representative traces of evoked APs in MSNs expressing either a control hairpin (shCTRL; black) or an Aldh1a1 knockdown hairpin (shAldh1a1; blue) in response to increasing current injections (inset). (B) Number of action potentials plotted as a function of injected current for MSNs expressing either shCTRL (black) or shAldh1a1 (blue). (C, D) Comparison of the kinetics of the upstroke (C) and downstroke (D) of the AP for the indicated experimental groups. (E) Comparison of the voltage threshold for action potential initiation between the indicated experimental groups. (F) Comparison of resting membrane potential between the indicated experimental groups. (G) Comparison of input resistance between the indicated experimental groups. Data shown are the mean ± SEM (n = 10 cells/group). Statistical significance between groups was assessed using a Student’s t-test. In (B), * denotes a p-value of at least <0.05. In (G), * denotes a p-value < 0.05.

4. Discussion

Overview

Based on previous findings (Table 1), knocking down Aldh1a1 was hypothesized to decrease depression-related behavior and drug-taking, making it a promising target for pharmacotherapeutic development. In Click or tap here to enter text.Click or tap here to enter text.the current study, we show that Aldh1a1 knockdown in the NAcSh increases emotional resilience, indicated by decreased depression- and frustration-related behavior compared to controls. This manipulation also resulted in decreased drug-taking during low- but not high-effort conditions compared to controls. Additionally, we found that Aldh1a1 knockdown increased anxiety-related behavior in the open-field exploration, elevated plus maze, and sucrose neophobia tests. Finally, electrophysiological assays revealed decreased excitability of medium spiny neurons (MSNs) expressing the shAldh1a1 hairpin, suggesting that a decreased output from the NAcSh may underlie the observed phenotypes. These results, barring frustration-related behavior which we only recently gained the ability to measure, were expected based on the findings produced by knocking down the protein opposing Aldh1a1 function–– Cyp26b1 (Table 1), an RA metabolizer. Our results highlight RA signaling in the NAcSh as a potential overlapping mechanism for emotion- and addiction-associated disorders. Table 1 summarizes the results of this study in the context of previous investigations.

Knocking down retinoic acid synthesis in the NAcSh increases anxiety-related behavior and enhances emotional resilience

It is well-established that manipulations of the NAcSh typically produce similar effects in depression- and addiction-related behavior, with concomitant MSN excitability and contra-regulation of anxiety-related behavior (Table 1). Whether this can be attributed to a species- or region-specific effect, to our knowledge, has not yet been determined. However, integrating our findings with existing literature points to a possible region-specific mechanism. Specifically, it was reported that continuous intracerebroventricular administration of RA increased both anxiety- and depression-related behavior in rats (Huang and Chen 2020). Therefore, manipulating whole-brain RA levels may approximate clinical endotypes by allowing wide distribution of RA, whereas AAV manipulations can inform on how RA is used within the specific region of interest. Overall, our findings suggest that general hyperactivity of RA signaling may inform on comorbid anxiety and depression presentations, whereas non comorbid presentations may be mediated by dysregulations specific to the NAcSh, striatum, or more broadly the limbic system. Still, follow-up is necessary to fully grasp how our consistent but puzzling findings inform clinical endophenotypes.

Nevertheless, shAldh1a1 animals exhibited increased anxiety-related behavior compared to controls in the open-field exploration, sucrose neophobia, and elevated plus maze tests (Fig. 2C–F). Moreover, we observed an anti-depressant phenotype in a forced swim test, with knockdown animals showing increased latency to immobility and decreased total time immobile compared to controls (Fig. 3C–D). Robust assays of frustration-related operant behavior in fentanyl self-administration corroborated findings of emotional resilience. We previously showed that more than 90% of animals (across multiple cohorts, regardless of sex) will increase their average duration of bar pressing during frustrating operant tasks (Vasquez et al. 2021a; Marmol Contreras et al. 2024). However, this effect was significantly blunted in shAldh1a1 animals relative to controls for both extinction and progressive ratio sessions (Fig. 4B, 4D), indicating blunted emotional reactivity. No differences were detected in the sucrose preference task, which does not measure reactivity. Together, the combined use of classic tasks for emotional behavior and novel tasks of emotional reactivity point to a pivotal role of RA synthesis in controlling emotional regulation. Our findings expand upon existing literature linking global RA abundance to hyper-emotionality (Bremner and McCaffery 2008). In our case, decreasing RA production in the NAcSh alone attenuated emotional reactivity, suggesting that this region at least partially contributes to the findings reported in pre-clinical and clinical settings.Click or tap here to enter text.

The post-isolation allo-grooming task modeling social withdrawal produced a result inconsistent with the other behavior and electrophysiology results. Rats with a protective addiction and depression phenotype typically spend more time mutually grooming following isolation (Zhang et al. 2019). Therefore, an anti-depressive phenotype in the forced swim task would be expected to co-occur with increased allo-grooming in Aldh1a1 knockdown animals compared to controls; however, the opposite was observed (Fig. 3B). One possibility is that for animals in the knockdown group, separation from their cage mates did not induce the strong negative emotions presumed to increase allo-grooming upon reunion. Moreover, a pre-isolation measurement of grooming was not obtained, and it is therefore impossible to discern whether the vector produced an asocial phenotype. Importantly, it was shown that administering RA to fmr1 knockout mice rescued deficits in social behavior, and the NAcSh has been shown to control several aspects of sociability (Yang et al. 2022). Therefore, RA regulation of this NAcSh function is an intriguing possibility meriting follow-up.

Knocking down retinoic acid synthesis in the NAcSh decreased drug-taking

In subsequent fentanyl self-administration, Aldh1a1 knockdown animals showed decreased drug-taking at baseline conditions. Across FR5 maintenance sessions, experimental animals obtained fewer fentanyl infusions compared to control animals (Fig. 5A). During behavioral economics sessions (Fig. 5C), experimental animals similarly exhibited decreased demand intensity (Q₀; Fig. 5D), indicated by fewer fentanyl infusions consumed by experimental animals at the “zero cost” dose (Strickland and Lacy 2020). Notably, there were no differences detected in number of infusions obtained during EXT and PR tasks (Fig. 4A, C), PR breakpoint (Fig. 4E), nor fentanyl demand elasticity during behavioral economics assessment (Fig. 5E). These findings indicate preserved motivation-related behavior during obstruction of drug-taking despite decreased drug-taking at low effort, non-frustrating conditions (i.e. decreased intrinsic drug valuation).

Our findings of attenuated drug-taking in Aldh1a1 knockdown animals expand upon existing literature assessing the role of RA in determining drug-taking. Firstly, the phenotype uncovered in this study opposes the phenotype produced by knocking down the degradation enzyme Cyp26b1 in the NAcSh yielding increased depression-related behavior and cocaine-taking, in addition to decreased anxiety-related behavior (Zhang et al. 2016b, 2019). This suggests that RA signaling within cells of the NAcSh contributes to functions associated with emotionality, anxiety, and drug-taking. Previous studies by our lab found that knockdown of the RA trafficking enzyme Fabp5 in the NAcSh decreased operant motivation for cocaine, with no effect on emotionality (Zhang et al. 2019; Crofton et al. 2021). Therefore, it is possible that the effect of decreased fentanyl-taking in Aldh1a1 knockdown animals is at least in part related to decreased RA trafficking by Fabp5. We also previously found that knockdown of trafficking enzyme Crabp2 decreased operant motivation for sucrose (Zhang et al. 2019), but we do not have information on the effect of Aldh1a1 knockdown on operant sucrose making it hard to draw conclusions.

Integrating our behavioral findings with existing literature points to RA-mediated plasticity as an underlying mechanism for attenuated drug-taking in Aldh1a1 knockdown animals. For instance, retinoic acid signaling in the striatum was shown to mediate network expression changes associated with protection against cocaine incubation of craving (Powell et al. 2020). Similarly, knocking down Cyp26b1 and Aldh1a1 in the nucleus accumbens core was shown to bidirectionally alter glutamatergic plasticity related incubation of cocaine craving (Wunsch et al. 2024). Given our own findings of attenuated MSN excitability within the NAcSh, it is likely that knocking down retinoic acid synthesis at least partially prevents drug-induced plasticity that facilitates drug-taking. Importantly, the findings in this manuscript suggest RA-mediated plasticity in the NAc as potential mechanism for drug-related functions generally, including but not limited to cocaine and stimulants.

The role of retinoic acid synthesis in medium spiny neuron functionality and implications

Silencing of Aldh1a1 markedly lessened the ability of MSNs to fire repetitive action potentials at current injections greater than 80 pA (Fig. 6B). Implications at the circuit level are likely to explain the behavioral phenotypes observed, though the precise mechanism is likely multifaceted. Firstly, RA is somewhat of a “master” regulator as it potently modulates a wide number of other key regulators. For instance, RA causes rapid phosphorylation of cyclic AMP response element binding protein (CREB) (Cañón et al. 2004), inhibition of which also produced a protective phenotype (Table 1). Similarly, RA decreases activating transcription factor 2 (ATF-2) phosphorylation in melanoma cells (Huang et al. 2008), and overexpression of ATF-2 in the NAcSh also produced a protective phenotype (Table 1). Therefore, it is possible that the protective phenotype of Aldh1a1 knockdown is partially via control of other transcription factors. Nevertheless, perturbed MSN excitability is prevalent across all protective phenotypes (Table 1) (Dong et al. 2006; Green et al. 2010; Larson et al. 2011; Scala et al. 2018), suggesting that NAcSh disengagement from its outputs may be important. Moreover, increased input resistance of MSNs (Fig. 6G) is consistent with a reduction in voltage-gated K+ currents and a reduction in voltage-gated K+ channels open at the resting membrane potential (Pablo and Pitt 2017), which could suggest voltage-gated K+ channels being regulated by RA.

Finally, reduced intracellular Ca2+ levels are known to induce RA synthesis (Wang et al. 2011) and subsequently increase cell surface expression of calcium-permeable ligand-gated ion channels (Aoto et al. 2008). It is thereby possible that Aldh1a1 knockdown decreases MSN excitability via prevented upregulation of calcium-permeable ion channels. Importantly, however, the molecular relationship between RA and calcium-permeable ion channels is not straightforward, with RA also directly altering the biophysical properties of voltage-gated Ca²⁺ channels and reduce their activity (de Hoog et al. 2018).

Limitation: Sex as a biological Variable

There are known sex differences in retinoic acid function (Youness et al. 2022), emotionality and anxiety (Allen and Haccoun 1976; Donner and Lowry 2013), frustration/aggression (Archer 2004; Im et al. 2018), and substance use tendencies (Becker 2016). Our study did not incorporate female rats, limiting our understanding of sex as a biological variable on the observed phenotypes.

Conclusions

The evidence enclosed in this manuscript solidifies retinoic acid signaling as a key regulator of emotion- and addiction-related behaviors, specifically in determining emotional resilience and drug-taking. By design, the NAcSh is implicated as the site of action, further expanding upon existing literature demonstrating an involvement of this region in mediating emotionality and drug-taking. Attenuation of MSN excitability in this region is thought to underlie the protective phenotypes observed, possibly by preventing maladaptive changes in plasticity induced by emotional stressors and drug consumption. Overall, our results posit RA signaling in the NAcSh as a key determinant of emotionality and drug-taking, and the RA synthesis enzyme Aldh1a1 is a promising target for anti-depressant and anti-addiction therapeutic development. This is the first report linking RA to fentanyl drug-taking as well as frustration, and the first to link the NAcSh to operant frustration.

Data Availability Statement

Data is available upon reasonable request.