Opioid use disorder is characterized by compulsive drug seeking and heightened relapse vulnerability following abstinence, a phenomenon known as incubation of craving. Although preclinical data suggest similar behavioral expression of opioid use between sexes, conclusive evidence on sex differences in craving and relapse across abstinence periods remains lacking. Here, we investigated the effects of abstinence from oxycodone self-administration on neurotransmission in the paraventricular thalamus (PVT) to nucleus accumbens shell (NAcSh) pathway in male and female rats. Using optogenetics and ex vivo electrophysiology, we assessed synaptic strength, glutamate release probability, and intrinsic excitability of NAcSh medium spiny neurons (MSNs) following 1 (acute) or 14 (prolonged) days of forced abstinence. No sex differences were observed in oxycodone self-administration or somatic withdrawal. However, females exhibited greater cue-induced relapse after prolonged but not acute abstinence. Prolonged abstinence produced comparable increases in PVT-NAcSh synaptic strength and presynaptic glutamate release probability in both sexes, while inhibitory transmission and MSN excitability were largely unaltered. The dissociation between comparable circuit-level plasticity and sex-specific relapse vulnerability suggests that PVT-NAcSh strengthening represents a shared neuroadaptation to oxycodone abstinence, while mechanisms driving heightened relapse in females likely involve additional circuit elements that remain to be identified.

Subjects

Adult male (250-275g; Total N = 42) and female (200-225g; Total N = 45) Sprague Dawley rats (Charles River Laboratory, Wilmington, MA) were used in this study. Upon arrival, rats were group housed (4 rats per cage) and were habituated for 1 week to the animal colony kept on a 12-h light/dark cycle (lights on 7:00AM) with food and water ad libitum. Following surgeries, rats were singly-housed for the rest of the experiment. All animal procedures were conducted in accordance with the guidelines of the National Institutes of Health and were approved by the Institutional Animal Care and Use Committee (IACUC) at McLean Hospital/Harvard Medical School (Protocol #2017N0000277, #2018N000090 & #2017N000125). 

Surgeries

Stereotaxic surgery and viral injections

All surgeries were performed according to AAALAC guidelines. Rats were first anesthetized with ketamine and xylazine (80 mg/kg and 8 mg/kg, respectively, I.P.). A craniotomy was made to target the PVT using the following stereotaxic coordinates, based on Paxinos and Watson 6th edition rat brain atlas (Paxinos & Watson, 2006): AP: -2.6mm from Bregma, ML: +2.0mm from Bregma (20° angle ML), and DV: -6.3mm from skull surface at injection site.  A total of 1μL of the AAV5 vector carrying CaMKIIα-ChR2(H143R)-eYFP was injected unilaterally into PVT at a rate of 125 nl/min using a 10μl Hamilton syringe with a 29-gauge needle under the control of a micro-syringe pump (Harvard Instruments). Viral vectors (titers, ~10.0 x 10¹² particles/ml) were purchased from the University of North Carolina viral vector facility.

Intravenous catheter implantation surgery

After recovery from stereotaxic surgery (~7 days), rats were implanted with indwelling silastic intravenous jugular catheters (SAI infusions; RSB-SA-7.5CF and RSB-SA-7.5CM), as described in Mavrikaki, Pravetoni, et al. (2017), Mavrikaki, Lintz, et al. (2021), and Thomsen (2025). Rats were anesthetized with ketamine and xylazine (80 mg/kg and 8 mg/kg, respectively, I.P.), and catheters were implanted into the right jugular vein, secured to the vein with non-absorbable suture thread and passed subcutaneously through the rat’s back. All rats received an injection of ketofen (5mg/kg; S.C.) and gentamicin (0.1ml; 10mg/ml, I.V.) during catheter implantation. Catheters were flushed daily with 0.2ml of heparinized saline (30 units/ml; I.V.) and once a week with 0.2ml of gentamicin (10 mg/ml I.V.). Catheter patency was checked once per week using methoexital (Brevital; 0.1 ml females; 0.2 males of 10 mg/ml I.V.).

Behavioral methods

Oxycodone self-administration

Med Associates operant conditioning chambers (30.5 (l) × 24.1 (w) × 29.2 (h) cm), kept within soundproofed outer chambers with ventilation fans, were equipped with two retractable levers, each with a cue light above them, a house light, a counterbalance swivel and tether, and an infusion pump. Rats (males: n = 15 saline, 27 oxycodone; females: n= 22 saline, 23 oxycodone) underwent 8 days of short access (ShA) oxycodone self-administration training (0.06mg/kg/infusion; 1h/day) followed by 14 days of long access (LgA) regimen (0.06 mg/kg/infusion; 6h/day), similar to Mavrikaki et al. (2019). Self-administration sessions were run 7 days/week, at approximately 9:00 am each day. All self-administration was conducted during the light phase of a 12:12 light/dark cycle (lights on at 7:00am; lights off at 7:00pm). A fixed-ratio 1 (FR1) schedule of reinforcement was used such that a press on the active lever resulted in a 4-s oxycodone infusion (100 µl) followed by a 6-s time out period where a press on the active lever produced no consequences.

Assessing oxycodone dependence

To demonstrate that our oxycodone self-administration protocol induced dependence in both male and female rats, we measured spontaneous somatic withdrawal signs 24-h after the last oxycodone self-administration session. After removal from self-administration chambers and catheter flushing, rats were placed back in their home cages and brought to a quiet, temperature-maintained (20°C) room and allowed to habituate for ~15-min. Rats were then individually placed into clear, 65-cm-high by 25-cm-diameter Plexiglas cylinders that contained a small amount of bedding. Rats were allowed to habituate to the cylinders for ~15 min. At this point, a digital video system (Swann Communications, Sante Fe, CA) was used to record the rats in the cylinders for 20 minutes. Upon completion of recording, somatic withdrawal behaviors were scored for the first 15 min of the recording by a researcher who was unaware of the treatments. Every 15 seconds, the following behaviors were marked as either present or absent: diarrhea, ptosis, jumping, walking, rearing, digging, flat posture, “wet dog shakes,” grooming and teeth chattering (Mavrikaki et al., 2021; Chartoff et al., 2006). The number of occurrences of each behavior was summed. In addition, a Total Withdrawal Score was calculated by summing weighted frequencies of those behaviors most commonly and specifically observed in opioid withdrawal: Total Withdrawal = Grooming (x1.0) + Wet Dog Shakes (x1.5) + Ptosis (x1.2) (Chartoff, Mague, et al., 2006; Chartoff, Barhight, et al., 2009). Wet Dog Shakes and Ptosis were multiplied by previously determined weighting factors to account for their high importance, but low prevalence, to withdrawal signs.  

Forced abstinence and cue-induced oxycodone-seeking

a)  1-day abstinence (acute abstinence): From the total rats above, male (Saline, N = 9; Oxycodone, N = 18) and female (Saline, N = 13; Oxycodone, N = 15) rats underwent 1d of forced abstinence (rats returned to the vivarium in their home cages for 24-h) from oxycodone self-administration. A subset of 1d abstinence male (Saline, N=5; Oxycodone N=10) and female (Saline, N=7; Oxycodone N=7) rats were used to measure somatic withdrawal signs, and another subset of males (Saline, N=4; Oxycodone, N=8) and females (Saline, N=6; Oxycodone, N=8) was used to measure cue-induced oxycodone-seeking after the 1d abstinence period. After 1d of abstinence, rats were reintroduced to the operant chamber for a 2-h relapse test. The cues associated with oxycodone were presented, but no drug was delivered.

b) 14-day abstinence (prolonged abstinence): From the total rats above, male (Saline, N=6; Oxycodone N=8) and female (Saline, N=8; Oxycodone, N=7) rats underwent 14d of abstinence from oxycodone self-administration. After 14d of abstinence, rats were reintroduced to the operant chambers for a 2-h relapse test as described above.

The number of active and inactive lever presses were recorded during incubation/ cue-induced oxycodone-seeking testing following acute and prolonged abstinence periods. Active lever presses were compared between saline and oxycodone groups for cue-induced oxycodone-seeking. Rat brains were extracted thirty minutes to one hour following the cue-induced oxycodone-seeking test, such that electrophysiological recordings reflect synaptic properties associated with relapse after acute or prolonged abstinence.

Ex-vivo electrophysiology and optogenetic stimulation

Coronal slices (300 µm in thickness) containing the NAc were obtained using a vibratome in cold cutting solution containing the following in mM: 252.0 sucrose, 1.0 CaCl₂, 5.0 MgCl₂, 2.5 KCl, 1.25 NaH₂PO₄, 26.0 NaHCO₃ and 10.0 glucose and equilibrated with 95% O₂ and 5% CO₂. Slices were then incubated in artificial cerebrospinal fluid (ACSF) containing the following in mM: 125 NaCl, 2.5 KCl, 2.5 CaCl₂, 1.0 MgSO₄, 1.25 NaH₂PO₄, 26.0 NaHCO₃, and 10.0 glucose at room temperature for at least 1 hr before recordings started. Whole-cell recordings were obtained from the NAcSh neurons with patch electrodes (3-5 MΩ resistance) containing the following in mM: 135.0 Cs-methane-sulfonate, 5.0 NaCl, 1.0 MgCl₂, 10.0 BAPTA, 10.0 HEPES, 2.0 ATP and 0.20 GTP adjusted to pH 7.2 with CsOH. Neurobiotin (0.2%; Vector Laboratories) was also added to the internal solution before the recordings to allow subsequent histochemical localization of the recorded neurons in the NAcSh.

Synaptic responses were induced by photostimulation of ChR2-expressing PVT projecting terminals in the NAcSh with a LED light source (excitation wavelength: 470 nm, 5 ms in duration, Thorlabs).  Whole-cell recordings were accepted if the access resistance was ≤20 MΩ and remained stable throughout the recording period (defined as <20% change from baseline). Recordings that exceeded these thresholds were excluded from analysis. All recordings were performed at 30-32°C. After recordings, slices were placed in PBS containing 4% paraformaldehyde and kept in the refrigerator until histological processing.

The pharmacological reagents used in electrophysiological experiments included NBQX disodium salt, D-AP5, NBQX, and (-)-bicuculline methobromide, which were prepared as stock solutions in water at 1000- to 5000-fold concentrations and stored at -20°C.

Morphological analysis

Histology for Neurobiotin-filled cells

Brain slices containing Neurobiotin-filled neurons in NAcSh were washed in PBS for 20 min x 3 times and incubated with Streptavidin Alexa 568 conjugate (10-20 µg/ml, catalog number: S11226, Molecular Probes) in PBS containing 0.2% Triton X-100 at room temperature for 24 hours. The slices were then washed with PBS for 20 min x 3 times and mounted on gelatinized slides. The anti-fading mounting media with DAPI (Vectashield, Vector Laboratories) was applied to slices.

Analysis of dendritic morphology of NAcSh-MSNs

Acquisition of imaging data of Neurobiotin-stained neurons was performed using a Leica SP8 TCS confocal microscope under a 40X/1.30 NA oil-immersion objective lens. The image resolution (1024 X 1024 pixels) and z step (0.5 μm) of optical planes were kept constant throughout the study. To image entire dendritic trees of NAcSh neurons, the zoom was adjusted in a range of 0.75 – 1.0, which corresponds to the voxel size between 0.284 X 0.284 X 0.5 μm and 0.379 X 0.379 X 0.5 μm. Three-dimensional (3D) reconstruction of dendritic trees was conducted in stacked confocal images in the program NeuronStudio (version 0.9.92, Icahn School of Medicine at Mount Sinai, New York, NY; Rodriguez et al., 2003).

We used the manual tracing tool to reconstruct dendrites starting from the soma. Specifically, we started the manual tracing from the beginning of each primary dendrite and moved one node at a time to form an entire path along a branch through the views at XY, ZY and XZ orientations. The program provides numerical measurements of the reconstructed dendritic trees, including dendritic length, surface area, volume, number of branch points, and Sholl analysis. Soma size was estimated by measuring the maximal projection area of a neuron soma in a single optical coronal section (1.038 µm thick) using ImageJ. The spatial organization of the dendritic field was characterized by the ratio of the long axis to the short axis of the dendritic field at the coronal plane. The long axis of the dendritic field was defined as the distance from the soma to the most distal dendritic process, and the short axis as the distance from the soma to the most distal dendritic point located 90° from the long axis (O'Donnell & Grace, 1993). The measurements were performed on stacked images of MSNs using the Leica Application Suite X (LAS X, Leica Microsystems CMS GmbH, version 3.5.5). The Neurobiotin-stained neurons were primarily obtained from the medial division of NAcSh in two adjacent slices of caudal nucleus accumbens, corresponding to the bregma levels between 10.08 and 10.56 mm (Paxinos & Watson, 2006).

Statistical analysis

Male and female rats were randomly assigned to either saline or oxycodone groups. In electrophysiological experiments, ~2-3 neurons were recorded per animal. The numbers of rats and recorded neurons for the analysis of the different experiments are indicated in the results section of the associated article. Data are reported as mean +/- SEM. All electrophysiology data were collected using Patch Master (Heka systems). We used Prism 9 (GraphPad) for statistical analysis using Two-tailed t-tests, Mixed effects model (Restricted Maximum Likelihood/REML), and Two- or Three-way ANOVAs with Sidak’s multiple comparisons, as appropriate.

For analyses of intrinsic excitability based on spike count–current relationships, spike output (number of action potentials per current step) was analyzed using nested hierarchical Poisson generalized linear mixed-effects models (GLMMs), fitted separately for each sex and abstinence duration. Each model included injected current (mean-centered), drug condition (Saline vs. Oxycodone), and their interaction as fixed effects. Random effects included random intercepts and slopes for injected current at the animal level, and random intercepts for cells nested within animals, to account for the non-independence of repeated current steps within cells and of multiple cells recorded from the same animal. Unique identifiers were assigned to each animal and cell by concatenating condition labels with animal and cell IDs, ensuring that animals across treatment conditions were treated as independent subjects. Models were fitted using maximum likelihood estimation with Laplace approximation. Overdispersion was assessed using the Pearson chi-squared statistic divided by the residual degrees of freedom; when this ratio exceeded 1.5, a dispersion correction was applied within the Poisson framework. Statistical significance of fixed effects was assessed using marginal F-tests with residual degrees of freedom. Sex comparisons among oxycodone-treated animals were conducted as separate analyses using identical model structures, with sex as the grouping condition. For visualization, marginal model-predicted values were overlaid on each plot as dashed lines, representing the population-level fixed-effects predictions from the fitted GLMM with random effects set to zero, back-transformed from the log scale to spike counts. To assess statistical power, a posthoc power analysis was conducted for each comparison using a two-sample t-test approximation at the rat level, with rat-level mean spike output as the unit of analysis and the pooled between-rat standard deviation as the denominator for effect size estimation (Cohen's d). This approach was used to estimate the number of animals per group required to achieve 80% and 90% statistical power, given the observed effect sizes. All analyses were performed in MATLAB (R2025b).

IPSC amplitudes were analyzed using linear mixed-effects models (LMEs) to account for the hierarchical structure of the data, with repeated measurements across light intensities nested within animals and multiple cells recorded per animal. Unique identifiers were assigned to each rat and cell by concatenating condition labels with animal and cell IDs, ensuring that animals across treatment conditions were treated as independent subjects. Models included fixed effects of light intensity (mean-centered across the six stimulation levels: 0.5, 1.3, 2.0, 4.7, 7.6, and 10.4 mW), treatment condition (Saline vs. Oxycodone), and their interaction. Random effects included random intercepts and light intensity slopes for individual animals, and random intercepts for cells nested within animals, capturing both between-animal variability in baseline responses and sensitivity to light, as well as within-animal variability across cells. Model parameters were estimated using restricted maximum likelihood (REML). Statistical significance of fixed effects was assessed using marginal F-tests with residual degrees of freedom, applied consistently across male and female datasets. Data are presented as mean ± SEM, computed from rat-level averages, in which responses from multiple cells within the same animal were first averaged to yield one value per animal per light intensity, consistent with the animal-level inference of the LME. Male and female datasets were analyzed separately using identical model structures to assess sex-specific effects. All analyses were performed in MATLAB (R2025b).

The number of animals and cells used per experiment, as well as the results of all statistical analyses, are reported in the text and figure legends.

References

Chartoff, E. H., Barhight, M. F., Mague, S. D., Sawyer, A. M., & Carlezon Jr, W. A. (2009). Anatomically dissociable effects of dopamine D1 receptor agonists on reward and relief of withdrawal in morphine-dependent rats. Psychopharmacology, 204(2), 227-239.

Chartoff, E. H., Mague, S. D., Barhight, M. F., Smith, A. M., & Carlezon Jr, W. A. (2006). Behavioral and molecular effects of dopamine D1 receptor stimulation during naloxone-precipitated morphine withdrawal. The Journal of neuroscience, 26(24), 6450-6457.

Mavrikaki, M., Anastasiadou, E., Ozdemir, R. A., Potter, D., Helmholz, C., Slack, F. J., & Chartoff, E. H. (2019). Overexpression of miR-9 in the nucleus accumbens increases oxycodone self-administration. International Journal of Neuropsychopharmacology, 22(6), 383-393.

Mavrikaki, M., Lintz, T., Constantino, N., Page, S., & Chartoff, E. (2021). Chronic opioid exposure differentially modulates oxycodone self‐administration in male and female rats. Addiction biology, 26(3), e12973.

Mavrikaki, M., Pravetoni, M., Page, S., Potter, D., & Chartoff, E. (2017). Oxycodone self-administration in male and female rats. Psychopharmacology, 234(6), 977-987.

O'Donnell, P., & Grace, A. A. (1993). Physiological and morphological properties of accumbens core and shell neurons recorded in vitro. Synapse, 13(2), 135-160.

Paxinos, G., & Watson, C. (2006). The rat brain in stereotaxic coordinates: hard cover edition. Elsevier.

Rodriguez, A., Ehlenberger, D., Kelliher, K., Einstein, M., Henderson, S. C., Morrison, J. H., ... & Wearne, S. L. (2003). Automated reconstruction of three-dimensional neuronal morphology from laser scanning microscopy images. Methods, 30(1), 94-105.

Thomsen, M., & Caine, S. B. (2005). Chronic intravenous drug self‐administration in rats and mice. Current Protocols in Neuroscience, 32(1), 9-20.