Abstract
We previously reported that ventral subiculum (vSub) activity is critical to incubation of oxycodone seeking after abstinence induced by adverse consequences of drug seeking. Here, we studied the role of claustrum, a key vSub input, in this incubation. We trained male and female rats to self-administer oxycodone for 2 weeks and then induced abstinence by exposing them to an electric barrier for 2 weeks. We used retrograde tracing (cholera toxin B subunit) plus the activity marker Fos to identify projections to vSub cactivated during “incubated” relapse (Abstinence Day 15). We then used pharmacological reversible inactivation to determine the causal role of claustrum in incubation and the behavioral and anatomical specificity of this role. We also used an anatomical disconnection procedure to determine the causal role of claustrum–vSub connections in incubation. Finally, we analyzed an existing functional MRI dataset to determine if functional connectivity changes in claustrum-related circuits predict incubation of oxycodone seeking. Claustrum neurons projecting to vSub were activated during relapse tests after electric barrier-induced abstinence. Inactivation of claustrum but not areas dorsolateral to claustrum decreased incubation of oxycodone seeking after electric barrier-induced abstinence; claustrum inactivation had no effect on incubation after food choice-induced abstinence. Both ipsilateral and contralateral inactivation of claustrum–vSub projections decreased incubation after electric barrier-induced abstinence. Functional connectivity changes in claustrum–cortical circuits during electric barrier-induced abstinence predicted incubated oxycodone relapse. Our study identified a novel role of claustrum in relapse to opioid drugs after abstinence induced by adverse consequences of drug seeking.
- electric barrier
- fMRI
- incubation
- opioids
- oxycodone
- self-administration
- self-administration
- voluntary abstinence
Significance Statement
We recently reported that ventral subiculum (vSub) is critical to incubation of oxycodone craving after abstinence induced by adverse consequences of drug seeking. Here, we first showed that claustrum projections to vSub are active during tests for incubation of oxycodone seeking after electric barrier-induced abstinence. Next, we used pharmacological inactivation to test the causal roles of claustrum and its anatomical connections with vSub in incubation. Using an existing functional MRI dataset, we also tested if functional connectivity changes in claustrum-related circuits predicted “incubated” oxycodone relapse after electric barrier-induced abstinence. Our data suggest that claustrum- and claustrum-related circuits contribute to relapse after voluntary abstinence induced by adverse consequences to drug seeking but not abstinence induced by availability of alternative nondrug rewards.
Introduction
Relapse during abstinence is a defining feature of opioid addiction (Hunt et al., 1971; Sinha, 2011). In laboratory rats, opioid seeking progressively increases or incubates during abstinence from heroin (Shalev et al., 2001), fentanyl (Gyawali et al., 2020), and oxycodone (Blackwood et al., 2019) self-administration. However, in these and many other incubation-related studies, drug seeking has been assessed after homecage-forced abstinence (Venniro et al., 2016; Chow et al., 2025). This contrasts with the human condition where abstinence is often voluntary because people who use drugs choose to avoid adverse consequences of drug seeking and taking (Epstein and Preston, 2003). To mimic this human condition, we developed rat models of drug relapse and craving after voluntary abstinence (Fredriksson et al., 2021a). In one model, abstinence is achieved by introducing negative consequences to inhibit drug seeking (an electric barrier in front of the drug-paired lever; Negishi et al., 2024).
In an initial study, we reported reliable incubation of oxycodone seeking after 15 and 30 d of electric barrier-induced abstinence; in both sexes, this incubation effect was stronger than the classical incubation after homecage-forced abstinence (Fredriksson et al., 2020). Next, we reported that ventral subiculum (vSub) neuronal activity is critical to incubation of oxycodone seeking after electric barrier-induced abstinence but not homecage-forced abstinence (Fredriksson et al., 2023). These results indicate that vSub activity is specifically involved in incubation after voluntary abstinence induced by adverse consequences of opioid seeking.
The goal of our current study was to identify additional brain regions contributing to incubation of oxycodone seeking after electric barrier-induced abstinence. We first used retrograde tracing (cholera toxin B subunit; CTb) with immunohistochemical detection of the activity marker Fos (Marchant et al., 2009, 2016; Claypool et al., 2023) to determine if incubation of oxycodone seeking after electric barrier-induced abstinence is associated with activation of brain regions projecting to vSub. Incubated relapse to opioid seeking after electric barrier-induced abstinence was associated with a large increase in Fos expression in claustrum neurons projecting to vSub (see Results). Based on these correlational Fos data, we selected the claustrum for mechanistic investigation.
We also examined claustrum's role in incubation of opioid seeking after electric barrier-induced abstinence, because claustrum and its cortical projections (White et al., 2017) contribute to various brain functions (Brown et al., 2017; Narikiyo et al., 2020; Madden et al., 2022), including behavioral and physiological effects of addictive drugs (Liu et al., 2019; Brynildsen et al., 2020; Terem et al., 2020, 2023; Peretz-Rivlin et al., 2024; Zhao et al., 2024; see Discussion). Additionally, studies using Fos have reported increased claustrum activity after acute or chronic opioid exposure (Brynildsen et al., 2020; Terem et al., 2023). We also recently found that “incubated” food seeking after prolonged abstinence (60 d) is associated with increased Fos expression in the claustrum (Madangopal et al., 2022).
In follow-up mechanistic experiments, we used muscimol + baclofen inactivation (McFarland and Kalivas, 2001) to determine the causal role of claustrum in incubation of oxycodone seeking after electric barrier-induced abstinence. To assess behavioral selectivity, we determined the effect of claustrum inactivation on incubation of oxycodone seeking after food choice-induced abstinence (Caprioli et al., 2017; Reiner et al., 2020). To assess anatomical specificity, we determined the effect of inactivation of regions located dorsolateral to claustrum (insular and somatosensory cortex) on incubation. We also used an anatomical disconnection procedure (Setlow et al., 2002; Bossert et al., 2012; Claypool et al., 2023) to determine the role of anatomical connections between claustrum and vSub in incubation.
Finally, we used an existing functional MRI (fMRI) dataset to explore claustrum-related circuits potentially involved in incubation of oxycodone seeking after electric barrier-induced abstinence (Fredriksson et al., 2021b). We determined if longitudinal resting-state functional connectivity changes of claustrum with different regions predict incubation. The resting-state functional connectivity method is based on our previous studies (Lu et al., 2012; Fredriksson et al., 2021b).
Materials and Methods
Subjects
We used male (n = 102) and female (n = 103) Sprague–Dawley rats (Charles River Laboratories) weighing 300–360 and 210–240 g, respectively, prior to surgery. We maintained the rats under a reverse 12:12 h light/dark cycle (lights off at 8:00 A.M.) with food and water available ad libitum. We housed two rats/cage prior to surgery and individually after surgery. We performed the experiment in accordance with the NIH Guide for the Care and Use of Laboratory Animals (eighth edition), under a protocol approved by the local Animal Care and Use Committee. We excluded 49 of the 205 rats used in the study due to failure to acquire oxycodone self-administration (n = 1), poor health (n = 21), no CTb expression (n = 1), complications in recovery (n = 1), loss of intravenous catheter patency (n = 3), or cannula misplacement (n = 22).
Drugs
We received oxycodone hydrochloride (HCl) from NIDA pharmacy and dissolved it in sterile saline. We chose a unit dose of 0.1 mg/kg for self-administration training based on our previous studies (Fredriksson et al., 2020, 2023). In Experiments (Exp.) 2–6, we dissolved muscimol + baclofen (Tocris Bioscience) in sterile saline and injected it intracranially at a dose of 50 + 50 ng in 0.5 µl/side (Venniro et al., 2017; Reiner et al., 2020; Fredriksson et al., 2023) 30 min before the relapse test sessions.
Intravenous surgery
We anesthetized the rats with isoflurane (5% induction, 2–3% maintenance, Covetrus). We attached silastic catheters to a modified 22 gauge cannula cemented to polypropylene mesh (Amazon or Industrial Netting), inserted the catheter into the jugular vein, and fixed the mesh to the midscapular region of the rat (Caprioli et al., 2017, 2018; Venniro et al., 2018). We injected the rats with ketoprofen (2.5 mg/kg, s.c., Covetrus) after surgery and the following day to relieve pain and decrease inflammation. We allowed the rats to recover for 6–8 d before food or oxycodone self-administration training. During recovery and all experimental phases, we flushed the catheters every 24–48 h with gentamicin (4.25–5.0 mg/ml, Fresenius Kabi or Covetrus) dissolved in sterile saline. If we suspected catheter failure during training, we tested patency with Diprivan (propofol, NIDA pharmacy, 10 mg/ml, 0.1–0.2 ml injection volume, i.v.), and if not patent, we recatheterized the other jugular vein and continued training the next day or eliminated the rat from the study.
Intracranial surgery
We performed the intracranial surgery at the same time as the intravenous surgery. We anesthetized the rats and, using a stereotaxic instrument (Kopf Instruments), implanted bilateral guide cannulas (23 gauge; Plastics One) 1 mm above the claustrum (Exp. 2, 3, 5, and 6), dorsolateral to claustrum (Exp. 4), and vSub (Exp. 5 and 6). We set the nose bar at −3.3 mm and used the following coordinates from the bregma for claustrum, anteroposterior (AP), +1.0 mm; mediolateral (ML), ±5.2 mm (6° angle); and dorsoventral (DV), −5.5 mm for males and −5.3 mm for females); dorsolateral to claustrum, AP, +1.0 mm; ML, ±5.2 mm (0–2° angle); and DV, −6.0 mm for males and –5.7 mm for females; and vSub, AP, –6.0 mm; ML, ±5.3 mm (4° angle); and DV, −7.5 to −8.0 mm for males and −7.2 to −7.5 mm for females. We anchored the cannulas to the skull with jeweler's screws and dental cement. We used these coordinates based on the results of Exp. 1 below, pilot experiments, and our previous study (Fredriksson et al., 2023).
Intracranial injections
Four days before the intracranial injections, we habituated the rats to the injection procedure. Habituation consisted of three phases. We first exposed the rats to the injection cage (an empty cage containing bedding). The following day, we gently removed the cannula blockers before exposing the rats to the injection cage. On the last habituation day, we gently lowered down the injectors and placed the rats in the injection cage. On the test days, we connected the syringe pump (Harvard Apparatus) to 10 µl Hamilton syringes and attached the Hamilton syringes to the 30 gauge injectors via polyethylene-50 tubing; the injectors were extended 1 mm below the tips of the guide cannulas. We injected vehicle (saline) or muscimol + baclofen (50 + 50 ng in 0.5 µl/side) at a rate of 0.5 µl/min and left the injector in place for an additional minute to allow diffusion. After the final testing, we deeply anesthetized the rats with isoflurane and removed their brains and stored them in 10% formalin. We sectioned brains at 50 µm using a Leica Microsystems cryostat and stained sections with cresyl violet to verify the placement of the cannulas.
CTb injections into vSub
After the last day of oxycodone self-administration training, we injected CTb (List Biological Laboratories) into vSub. We injected 50 nl of 1% CTb bilaterally into vSub over 5 min, with the needle left in place for an additional 5 min for diffusion (Mahler and Aston-Jones, 2012; Venniro et al., 2017; Reiner et al., 2020; Claypool et al., 2023). We used a 1.0 µl, 32 gauge Neuros Syringe (Hamilton) attached to UltraMicroPump (UMP3) with SYS-Micro4 Controller (World Precision Instruments). The coordinates for vSub were as follows: AP, −5.8 mm; ML, ±5.4 mm (4° angle); and DV, −8.3 mm for males and −8.0 mm for females. These coordinates are based on our previous studies (Bossert et al., 2016; Marchant et al., 2016; Fredriksson et al., 2023).
Tissue processing for immunohistochemistry
The procedure is based on our previous reports (Reiner et al., 2020; Claypool et al., 2023). Ninety minutes after the relapse test in Exp. 1, we deeply anesthetized the rats with isoflurane saturated in air in an enclosed glass desiccator for 80 s and perfused them transcardially with 100 ml of 1× PBS, pH 7.4, followed by ∼400 ml of 4% paraformaldehyde in PBS. We removed and postfixed the brains in 4% paraformaldehyde for 2 h before transferring them to 30% sucrose in PBS for 48 h at 4°C. We subsequently froze the brains in powdered dry ice and stored them at −80°C until sectioning. We cut coronal sections (40 µm) using a cryostat (Leica Microsystems). We divided the sections into three series and stored them in PBS containing 0.1% sodium azide at 4°C.
Immunohistochemistry for parvalbumin and Fos and CTb double labeling (claustrum)
We used parvalbumin as an anatomical marker of the claustrum (Dillingham et al., 2019; Grimstvedt et al., 2023; Fig. 2). We based our assay parameters on dilution curve pilot assays. We processed one-in-three series for triple-label immunohistochemistry to detect parvalbumin, Fos, and CTb. We first rinsed tissues three times in PBS over 30 min and incubated them in a blocking solution containing 2% normal donkey serum and 0.3% Triton X-100 in PBS. After 2 h in blocking solution, we incubated the tissues in primary antiserum with antibodies raised against parvalbumin (1:1,000; mouse anti-parvalbumin, catalog #P3088, RRID: AB_477329), Fos (1:4,000; rabbit anti-Phospho-c-Fos, catalog #5348S, RRID: AB_10557109, Cell Signaling Technology), and CTb (1:10,000; goat anti-CTb, catalog #703, RRID: AB_10013220, List Biological Laboratories) over 48 h at 4°C. Next, we rinsed the sections three times in PBS and then transferred them into secondary antiserum containing donkey anti-mouse Alexa Fluor 647 (1:500, catalog #715-605-151, RRID: AB_2340863, Jackson ImmunoResearch Laboratories), donkey anti-rabbit Alexa Fluor 555 (1:500, catalog #711-565-152, RRID: AB_3095471, Jackson ImmunoResearch Laboratories), and donkey anti-goat Alexa Fluor 488 (1:500, catalog #711-545-152, RRID: AB_2313584, Jackson ImmunoResearch Laboratories) for 5 h at room temperature. After three rinses in PBS, we mounted free-floating sections onto glass slides and coverslipped them with Dapi-Fluoromount-G (catalog #17984-24, Electron Microscopy Sciences).
Immunofluorescence histochemistry for Fos and CTb double labeling (other brain regions)
We processed a one-in-three series for immunohistochemical detection of Fos and CTb. We rinsed the sections three times with PBS and incubated them for 2 h in a blocking buffer containing 4% bovine serum albumin (BSA) in PBS with 0.3% Triton X-100 (PBS-TX). We then incubated all sections for at least 24 h at 4°C in rabbit anti-Fos primary antibody (1:2,000; Phospho-c-Fos, catalog #5348S, RRID: AB_10557109, Cell Signaling Technology), and goat anti-CTb antibody (1:2,000, catalog #703, RRID: AB_10013220, List Biological Laboratories) in 4% BSA in 0.3% PBS-TX.
Next, we rinsed the sections three times with PBS and incubated them with donkey anti-goat Alexa Fluor 488 (1:500, catalog #705-545-147, RRID: AB_2340430, Jackson ImmunoResearch Laboratories) and donkey anti-rabbit Alexa Fluor 594 (1:500, catalog #711-585-152, RRID: AB_2340621, Jackson ImmunoResearch Laboratories) for 2 h, followed by three rinses in PBS. We then mounted the sections on chromalum–gelatin-coated slides and coverslipped them with Dapi-Fluoromount-G (catalog #17984-24, Electron Microscopy Sciences).
Analysis and quantification of Fos and CTb
We divided the claustrum, basolateral amygdala (BLA), medial septum/diagonal band (MS/DB), paraventricular nucleus of the thalamus (PVT), and nucleus reuniens (RE) into anterior and posterior locations (areas) and captured images and quantified cells at two different bregma coordinates for these areas. For each rat, we quantified cells using the following approximate coordinates from the bregma: claustrum, anterior (+2.0 to +2.4) and posterior (+0.8 to +1.2); BLA, anterior (−3.0 to −3.24) and posterior (−3.4 to −3.6); MS/DB, anterior (+1.2) and posterior (+0.96); PVT, anterior (−1.5 to −2.0) and posterior (−3.0 to −3.5); and RE, anterior (−1.5 to −2.0) and posterior (−3.0 to −3.5). Before we started to capture the images, we determined CTb injection sites in vSub for all rats (see Fig. 2C for placement of the CTb injections).
For the claustrum, we used an Olympus VS200 Slide Scanner to capture images of parvalbumin, Fos, and CTb. We used parvalbumin as an anatomical marker (see above) and only counted Fos and CTb within parvalbumin-defined claustrum (Fig. 2). For the other brain regions, we used epifluorescence microscope to capture immunofluorescence images (10× objective) for Fos and CTb immunoreactive (IR) cells using an ORCA Flash 4.0LT (Hamamatsu Photonics) attached to a Zeiss Axio Scope Imager M2 using MicroManager (V1.4).
We only included rats in which the CTb injection site was restricted to vSub. Additionally, when CTb expression was similar in both hemispheres, we counted Fos and CTb from both hemispheres. If there were hemispheric differences in CTb signal, we counted Fos and CTb from the hemisphere with stronger CTb intensity. (Note: In a pilot experiment, we found that the projection from claustrum to vSub is unilateral with no CTb labeling in the opposite hemisphere.) For each rat, we identified Fos-positive and CTb-positive neurons based on their respective nuclear and cytoplasmic localizations and through imaging conditions that produced no observable channel bleed-through. We semiautomatically quantified cells in four sections and calculated the mean of these counts per area (i.e., each rat provides an individual data point for data analyses with data averaged from four images for each rat). We performed image-based quantification of Fos-positive, CTb-positive, and Fos + CTb cells in ImageJ in a blind manner (inter-rater reliability between IF and AB: Pearson's r = 0.96–0.99; AB and SC, r = 0.95–0.97; AB and DJR, r = 0.96–0.99; and KN and AP, r = 0.90–0.97).
Self-administration apparatus
For oxycodone self-administration, we used Med Associates chambers containing two levers located 7.5–8 cm above the grid floor on opposing walls. Responding on the active retractable lever activated the infusion pump, while lever presses on the inactive, nonretractable lever had no consequences. For food self-administration, we used Med Associates chambers containing two levers located 7.5–8 cm above the grid floor on the same wall. Responding on the active retractable lever activated the food dispenser, while the lever presses on the inactive lever had no consequences. We equipped each chamber with a stainless-steel grid floor connected to a shocker (Med Associates ENV-410B).
General behavior procedure
The experiments included the following phases: food self-administration (6 d; Exp. 3 only), oxycodone self-administration training (14 d), early tests for oxycodone seeking (Abstinence Day 1), electric barrier-induced abstinence (10–11 d; Exp. 1–2 and 4–6), or food choice-induced abstinence (11 d; Exp. 3) over 2 weeks and late tests for oxycodone seeking (Abstinence Day 15 or 16). We provide details of the different phases for each experiment below.
Food self-administration training
Before the first self-administration training session, we gave rats a 1 h magazine training session which began with the presentation of the white house light, followed by the noncontingent delivery of one pellet every 3 min. We used 45 mg preferred or palatable food pellets (12.7% fat, 66.7% carbohydrate, and 20.6% protein; catalog #1811155, TestDiet; Calu et al., 2014; Caprioli et al., 2015). We then trained the rats to lever press for food during six 1 h sessions that were separated by 10 min for 6 consecutive days. The sessions began with the presentation of the white house light, followed 10 s later by the insertion of the food-paired active lever (right panel). The white house light remained on for the duration of the session and served as a discriminative cue for the palatable food.
We trained rats under a fixed-ratio 1 (FR1)–20 s timeout reinforcement schedule (or fixed interval 20 s schedule), where one lever press resulted in the delivery of five 45 mg food pellets and the presentation of a 20 s discrete tone cue, during which additional lever presses were not reinforced. At the end of each 1 h session, the white house light was turned off, and the active lever was retracted. To match the number of discrete cue presentations to that of oxycodone (see below), we limited the number of food-reinforced deliveries to 15 per hour.
Oxycodone self-administration training
We trained the rats to self-administer oxycodone–HCl for 6 h/d (six 1 h sessions separated by 10 min) for 14 d under the FR1–20 s timeout reinforcement schedule. Oxycodone was infused at a volume of 100 µl over 3.5 s at a unit dose of 0.1 mg/kg/infusion. Each session began with illumination of a red house light that remained on for the entire session, followed 10 s later by the insertion of the active lever. Active lever presses led to oxycodone infusions that were paired with a 20 s tone–light cue. At the end of each 1 h session, the house light turned off, and the active lever retracted. We limited oxycodone intake to 15 infusions per hour.
Electric barrier-induced abstinence
During this phase, oxycodone was available for 2 h per day for 10–11 d. We used the same parameters (oxycodone dose, reinforcement schedule, tone–light cues, etc.) that we used during the training phase. To achieve abstinence, we introduced an electric barrier near the active lever (“shock zone”; Cooper et al., 2007; Fredriksson et al., 2020). We separated the “shock zone” (two-third of the chamber) from a “safe zone” (remaining one-third of the chamber) with a plastic demarcation (McMaster-Carr, catalog #9852K61). If the rats approached the active lever, they received a continuous mild footshock (0.1–0.4 mA). On the first day, we set the current at 0.0 mA and then gradually increased the intensity to 0.3 mA (0.1 mA increments per day). If a rat did not suppress oxycodone self-administration (<3 infusions per day), we increased the intensity to 0.4 mA the next day. Prior to the electric barrier phase, we tested the rats' sensitivity to footshock (operationally defined as the minimal shock level that causes the withdrawal of the front paw). There were no group or sex differences in shock sensitivity in any of the experiments, assessed with 0.05 mA increments, starting at 0.05 mA, and the values for individual rats ranged from 0.1 to 0.2 mA.
Food choice-induced abstinence
The procedure is based on our previous studies (Caprioli et al., 2015; Claypool et al., 2023). We conducted 11 discrete choice sessions using the same parameters (dose of oxycodone, number of palatable food pellets per reinforced delivery, stimuli associated with the two retractable levers) used during the training phases. We divided each choice session into 20 discrete trials with each trial being 9 min in duration. Individual trials consisted of 6 min during which the rat was able to respond. Each trial began with the presentation of both discriminative cues previously associated with palatable food or oxycodone, followed 10 s later by the insertion of both the palatable food-paired and oxycodone-paired levers. Rats could then select one of the two levers. If the rats responded within 6 min, the reinforcer associated with the selected lever was delivered. Each reinforced delivery was signaled by the oxycodone-associated or food-associated cue (white cue light or tone, respectively), retraction of both levers, and shutdown of the food and oxycodone discriminative cues. Thus, on a given trial, the rat could earn either oxycodone infusion or food pellets but not both. If a rat failed to respond on either active lever within 6 min, both levers were retracted, their related discriminative cues were turned off with no reinforcer delivery until onset of the next trial, and the trial was counted as an omission.
Relapse (incubation) tests
We tested the rats for oxycodone seeking for 30 or 90 min under extinction conditions during early (Day 1) or late (Day 15 or 16) abstinence or both. During testing after electric barrier-induced abstinence, we turned off the electric barrier and removed the plastic demarcation. We gave all rats a 30 min habituation period in the self-administration chamber before the start of the test session to allow them to realize that the barrier is not electrified. During testing after food choice-induced abstinence, we also gave the rats a 30 min habituation period in the self-administration chamber before the start of the test session. Lever presses resulted in the delivery of the oxycodone-paired tone–light cue and activation of the infusion pump, but no drug infusions.
Exp. 1: effect of incubated oxycodone seeking on Fos expression in afferent projections to vSub
The goal of Exp. 1 was to determine if incubation of oxycodone seeking after electric barrier-induced abstinence is associated with selective activation of one or more afferent projections to the vSub. The experiment consisted of three phases (Fig. 2A): oxycodone self-administration training (14 d), electric barrier-induced abstinence (10 d), and late tests for oxycodone seeking (Abstinence Day 15 or 16). We used two groups of male and female rats (n = 11 per group) in an experimental design that included the between-subjects factor of test condition (no-test, test). We matched the groups for oxycodone intake during the training and electric barrier phases. On Abstinence Day 15 (no-test group) or 16 (test group), the rats were given a 90 min relapse test during which response on both levers were not reinforced. On Abstinence Day 16, rats from the test group were perfused immediately after the Day 16 relapse test, and rats from the no-test group were perfused directly from their homecage. (Note: the no-test group was tested one day before the test group so that all rats were perfused on the same day.)
Exp. 2: effect of bilateral claustrum inactivation on incubation after electric barrier-induced abstinence
In Exp. 1, we found that incubation of oxycodone seeking after electric barrier-induced abstinence was associated with increased Fos expression in claustrum neurons projecting to vSub. The goal of Exp. 2 was to determine if the claustrum plays a causal role in incubation of oxycodone seeking after electric barrier-induced abstinence. For this purpose, we used the classical muscimol + baclofen inactivation procedure (McFarland and Kalivas, 2001) to inactivate the more posterior claustrum area (AP, +1.0 mm from the bregma). The experiment consisted of four phases (Fig. 4A): oxycodone self-administration training (14 d), early tests for oxycodone seeking (Abstinence Day 1), electric barrier-induced abstinence (11 d), and late tests for oxycodone seeking (Abstinence Day 15). We used two groups of male and female rats (n = 9–12 per group) in a mixed experimental design that included the between-subjects factor of muscimol + baclofen dose (0, 50 + 50 ng/side) and the within-subjects factor of abstinence day (1 or 15). We matched the different groups for total oxycodone infusions during the training phase. We compared the number of lever presses on Day 1 and Day 15 tests in rats injected with either saline or muscimol + baclofen into the claustrum before the relapse tests.
Exp. 3: effect of bilateral claustrum inactivation on incubation after food choice-induced abstinence
In Exp. 2, we found that muscimol + baclofen inactivation of the posterior claustrum decreased incubation of oxycodone seeking after electric barrier-induced abstinence. In Exp. 3, we determined if this finding generalizes to incubation of oxycodone seeking when abstinence was achieved by giving rats access to an alternative nondrug food reward (Caprioli et al., 2017; Fredriksson et al., 2021a). The experiment consisted of five phases (Fig. 5A): food self-administration (6 d), oxycodone self-administration training (14 d), early tests for oxycodone seeking (Abstinence Day 1), food choice-induced abstinence (11 d), and late tests for oxycodone seeking (Abstinence Day 15). We used two groups of male and female rats (n = 11–12 per group) in a mixed experimental design that included the between-subjects factor of muscimol + baclofen dose (0, 50 + 50 ng/side) and the within-subjects factor of abstinence day (1 or 15). We matched the different groups for total oxycodone infusions during the training phase.
Exp. 4: effect of bilateral inactivation dorsolateral to claustrum on incubation after electric barrier-induced abstinence
The claustrum is a small region, and after injections of muscimol + baclofen into this region, the mixture likely diffused away from the claustrum to nearby regions (insular and somatosensory cortex). The goal of Exp. 4 was to determine the anatomic specificity of the inhibitory effect of claustrum inactivation by injecting muscimol + baclofen dorsolateral to the claustrum injection site (Fig. 6B).
The experiment consisted of four phases (Fig. 6A): oxycodone self-administration training (14 d), early tests for oxycodone seeking (Abstinence Day 1), electric barrier-induced abstinence (11 d), and late tests for oxycodone seeking (Abstinence Day 15). We used two groups of male and female rats (n = 13 per group) in a mixed experimental design that included the between-subjects factor of muscimol + baclofen dose (0, 50 + 50 ng/side) and the within-subjects factor of abstinence day (1 or 15). We matched the different groups for total oxycodone infusions during the training phase.
Exp. 5: effect of anatomical disconnection of claustrum–vSub connections on incubation after electric barrier-induced abstinence
The goal of Exp. 5 was to determine if the projections from claustrum to vSub are necessary for relapse after electric barrier-induced abstinence. For this purpose, we used the muscimol + baclofen inactivation procedure (McFarland and Kalivas, 2001) for anatomical disconnection. The experiment consisted of four phases (Fig. 7A): oxycodone self-administration training (14 d), early tests for oxycodone seeking (Abstinence Day 1), electric barrier-induced abstinence (11 d), and late tests for oxycodone seeking (Abstinence Day 15). We used three groups of male and female rats (n = 10–14 per group) in a mixed experimental design that included the between-subjects factor of the group (vehicle, ipsilateral muscimol + baclofen in claustrum and vSub, contralateral muscimol + baclofen in claustrum and vSub) and the within-subjects factor of abstinence day (1 or 15). The vehicle group consisted of rats injected with vehicle either ipsilateral or contralateral in claustrum and vSub because there were no significant differences between these two conditions.
Exp. 6: effect of unilateral claustrum or vSub inactivation on incubation after electric barrier-induced abstinence
In Exp. 5, we found that both ipsilateral and contralateral inactivation of the projections between claustrum and vSub decreased incubation of oxycodone seeking after electric barrier-induced abstinence. The goal of Exp. 6 was to rule out the possibility that the similar effects of ipsilateral and contralateral inactivation were due to unilateral inactivation of either the claustrum or vSub alone. We used three groups of male and female rats (n = 8–16 per group) in a mixed-factorial design that included the between-subjects factor of the group (vehicle, unilateral muscimol + baclofen into the claustrum, or unilateral muscimol + baclofen into the vSub) and the within-subjects factor of abstinence day. The vehicle group included rats injected bilaterally with vehicle into either the claustrum or vSub, as no significant differences were observed between these two brain regions. The unilateral muscimol + baclofen claustrum group included rats injected with saline in one hemisphere and muscimol + baclofen in the other hemisphere. The unilateral muscimol + baclofen vSub group included rats injected with saline in one hemisphere and muscimol + baclofen in the other hemisphere. Within each brain region, we counterbalanced the hemisphere receiving the muscimol + baclofen injection.
Statistical analyses
Behavioral and immunohistochemistry data
We analyzed the behavioral and immunohistochemistry data with repeated-measure (RM) ANOVAs or ANOVAs using SPSS (Version 24, GLM procedure). For the behavioral data, we followed significant main effects and interactions (p < 0.05) with univariate ANOVAs and Bonferroni's post hoc tests. For the immunohistochemistry data, we performed multiple t tests on three measures (Fos, CTb, and Fos + CTb) across the anterior and posterior levels of five brain areas. To account for these multiple comparisons, we also applied a Bonferroni's post hoc correction to the resulting p values. We describe the different between- and within-subject factors for the different analyses in the Results section. Because the multifactorial ANCOVAs yielded multiple main and interaction effects, we only report significant effects that are critical for data interpretation. For clarity, we indicate the results of post hoc analyses with asterisks in the figures, but do not describe them in the Results section. For a complete reporting of the statistical analyses of the data presented in Figures 1⇓⇓⇓⇓⇓⇓–8, see Table 1. We did not observe statistically significant sex differences in the data described in the figures. Therefore, we combined data from males and females within each experimental condition for all statistical analyses.
Behavioral data prior to the relapse testing in Exp. 1–6. A, B, The number of infusions, active and inactive lever presses during the oxycodone self-administration training and electric barrier phases in Exp. 1 (total n = 22, 11 males, 11 females) and Exp. 2 (total n = 21, 8 males, 13 females). C, Food choice-induced abstinence. The number of rewards, active and inactive lever presses during food self-administration (left), oxycodone self-administration (middle), and food choice-induced abstinence (right) in Exp. 3 (total n = 21, 11 males, 10 females). D–F, Electric barrier-induced abstinence. The number of infusions, active and inactive lever presses during the oxycodone self-administration training and electric barrier phases in Exp. 4 (total n = 26, 13 males, 13 females), Exp.5 (total n = 36, 15 males, 21 females), and Exp. 6 (n = 35, 16 males, 19 females). Data are mean ± SEM.
Fos and CTb expression during incubated relapse to oxycodone seeking after electric barrier-induced abstinence (Exp. 1, claustrum). A, Timeline of Exp. 1. B, Relapse (incubation) tests after electric barrier-induced abstinence: The number of active lever presses during the 90 min Day 15 test session. C, Locations of CTb injection sites in vSub and representative image. D, Representative CTb and Fos expression in parvalbumin-defined claustrum for test and no-test groups. Arrowheads indicate Fos–CTb double-labeled neurons. E, CTb and Fos-IR quantification. The number of Fos-positive cells per square millimeter, the number of CTb-positive cells per square millimeter, and the number of Fos + CTb double-labeled cells per square millimeter in the anterior and posterior claustrum. *Different from the no-test group, Bonferroni-corrected p < 0.05. Data are mean ± SEM. No-test, n = 11 (6 males, 5 females); test, n = 11 (5 males, 6 females). Scale bars: C, 1 mm; D, 1 mm (left), 500 µm (middle), 100 µm (right).
Fos and CTb expression during incubated relapse to oxycodone seeking after electric barrier-induced abstinence (Exp. 1, other brain regions). A, ROI locations for Fos and CTb quantification from Exp.1. B, Representative Fos (red) and CTb (green) immunoreactivity in ROIs of test and no-test groups. Arrowheads indicate Fos–CTb double-labeled neurons. C, CTb and Fos-IR quantification. Top to bottom, The number of Fos-positive cells per square millimeter, the number of CTb-positive cells per square millimeter, and the number of Fos + CTb double-labeled cells per square millimeter in different brain regions. *Different from the no-test group, Bonferroni-corrected p < 0.05. Data are mean ± SEM. No-test, n = 11 (6 males, 5 females); test, n = 11 (5 males, 6 females). Scale bar, 100 µm.
Effect of claustrum inactivation on incubation of oxycodone seeking after electric barrier-induced abstinence (Exp. 2). A, Timeline of Exp. 2. B, Representative image and injector tip placements in the claustrum. Scale bar, 1 mm. C, Relapse (incubation) tests Days 1 and 15: the number of active lever presses during the 30 min test session on Abstinence Day 1 and the first 30 min of the Day 15 test session after muscimol + baclofen injections into the posterior claustrum. *Different from vehicle on Day 15 for active lever presses, p = 0.004. D, Relapse (incubation) test Day 15: the number of active lever presses during the 90 min Day 15 test session. *Different from vehicle on Day 15 for active lever presses, p = 0.002. n = 9–12 rats per group (3–5 males/group, 6–7 females/group). Data are mean ± SEM.
Effect of claustrum inactivation on incubation after food choice-induced abstinence (Exp. 3). A, Timeline of Exp. 3. B, Representative images of injector tip placements in posterior claustrum. Scale bar, 1 mm. C, Relapse (incubation) tests Days 1 and 15: the number of active lever presses during the 30 min Day 1 test session and the first 30 min of the Day 15 test session after injections of muscimol + baclofen into the posterior claustrum. D, Relapse (incubation) test Day 15: the number of active lever presses during the 90 min Day 15 test session. *Different from Day 1, p < 0.05. n = 11–12 rats per group (5–6 males/group, 6 females/group). Data are mean ± SEM.
Effect of inactivation outside the claustrum on incubation after electric barrier-induced abstinence (Exp. 4). A, Timeline of Exp. 4. B, Representative images of injector tip placements in the insular and sensory cortex. Scale bar, 1 mm. C, Relapse (incubation) tests Days 1 and 15: the number of active lever presses during the 30 min Day 1 test session and the first 30 min of the Day 15 test session after injections of muscimol + baclofen into the insular and sensory cortex. D, Relapse (incubation) test Day 15: the number of active lever presses during the 90 min Day 15 test session. *Different from Day 1, p < 0.05. n = 13 rats per group (6–7 males/group, 6–7 females/group). Data are mean ± SEM.
Effect of claustrum–vSub disconnection on incubation after electric barrier-induced abstinence (Exp. 5). A, Timeline of Exp. 5. B, Representative injector tip placement images and locations in claustrum and vSub for ipsilateral and contralateral injections. C, Relapse (incubation) tests Days 1 and 15: the number of active lever presses during the 30 min Day 1 test session and the first 30 min of the Day 15 test session after injections of muscimol + baclofen into claustrum and vSub. *Main effect of group condition for active lever presses; p = 0.045 and 0.073 using Bonferroni-corrected post hoc differences between vehicle and ipsilateral and contralateral groups, respectively. D, Relapse (incubation) test Day 15: the number of active lever presses during the 90 min Day 15 test session. *Main effect of group condition for active lever presses; p = 0.025 and 0.019 using Bonferroni-corrected post hoc differences between vehicle and ipsilateral and contralateral groups, respectively. n = 10–14 rats per group (4–6 males/group, 5–8 females/group). Data are mean ± SEM. vSub, ventral subiculum; M + B, muscimol + baclofen (50 + 50 ng).
Effect of unilateral claustrum or vSub inactivation on incubation after electric barrier-induced abstinence (Exp. 6). A, Timeline of Exp. 6. B, Representative image and injector tip placements in the claustrum and vSub. C, Relapse (incubation) tests Days 1 and 15: the number of active lever presses during the 30 min test session on abstinence Day 1 and the first 30 min of the Day 15 test session after vehicle or unilateral muscimol + baclofen injections into the claustrum and vSub. D, Relapse (incubation) test Day 15: the number of active lever presses during the 90 min Day 15 test session. *Different from Day 1, p < 0.001. n = 8–16 rats per group (4–9 males/group, 4–8 females/group). Data are mean ± SEM. vSub, ventral subiculum; M + B, muscimol + baclofen (50 + 50 ng). The vehicle group (n = 16) includes rats injected with vehicle into either the claustrum or vSub.
Statistical analyses for experiments 1–6 (SPSS GLM RM module)
fMRI data
We used an existing fMRI dataset (Fredriksson et al., 2021b) to determine if functional connectivity changes of claustrum-related circuits predict incubation after electric barrier-induced abstinence. The flowchart of the fMRI analyses is shown in Fig. 9, and the experimental timeline is shown in Figure 10A. We preprocessed fMRI data with a standard pipeline, including skull stripping, motion correction, coregistration to a template, noise component removal, temporal filtering, spatial smoothing (in plane) with a 0.6 mm full-width at half-maximum Gaussian kernel, nuisance covariance regression for respiration, head motion, and body temperature. We excluded 7 of the 44 rats from the analyses because their head motion exceeded a standard threshold (Liu et al., 2020). Next, we delineated the bilateral claustrum seed with 10 imaging voxels (0.547 × 0.547 × 1 mm per voxel; Fig. 10B) based on a rat atlas (Paxinos and Watson, 2007).
Flowchart of the fMRI analyses.
Correlation between changes of functional connectivity in claustrum-related circuits and incubation score after electric barrier-induced abstinence. A, Experimental timeline. B, Voxelwise analysis of functional connectivity changes of the claustrum-related circuits: abstinence phase, oxycodone group. Functional connectivity changes of the claustrum with the olfactory bulb, orbitofrontal cortex, insula, sensory cortex, cingulate cortex, retrosplenial cortex, and visual cortex were significantly correlated with the incubation score in the oxycodone group. Imaging results were corrected for whole-brain multiple comparisons at pcorr < 0.05. C, Correlations between the changes of functional connectivity in the above circuits and the incubation score. The underlay brain images in B are from the mean of structural images that were registered to a rat brain atlas.
To account for the potential influence of signals from adjacent regions, we used a Small Region Confound Correction protocol for analyses using claustrum seed (Krimmel et al., 2019b,a). Specifically, we dilated the claustrum by two voxels and determined the overlaps between the dilated claustrum and the insula, striatum, and endopiriform seeds that were separated by at least 1 voxel (0.547 mm) from the original claustrum. These regions were called “flanking” regions, and we regressed claustrum time series from the flanking regions. Next, we conduced voxelwise Pearson's correlation coefficients between the time courses of the corrected claustrum seed and all brain voxels and converted them to z-scores (achieving an approximate normal distribution) to evaluate functional connectivity between brain regions. To verify the specificity of the claustrum seed analysis, we also put seeds in the adjacent insula (6 voxels; Fig. 11A) and striatum (10 voxels; Fig. 11B) as controls and conducted the similar analyses to that of the claustrum.
Correlation between changes of functional connectivity in control seeds (insula and striatum adjacent to claustrum) and incubation score after electric barrier-induced abstinence. A, Voxelwise analysis of functional connectivity changes of insula-related circuits: abstinence phase, oxycodone group. Functional connectivity change of insula with dorsal striatum was negatively correlated with the incubation score in the oxycodone group. B, Voxelwise analysis of functional connectivity changes of striatum-related circuits: abstinence phase, oxycodone group. No significance was found with the striatum seed. The underlay brain images are from the mean of structural images that were registered to a rat brain atlas. Imaging results were corrected for whole-brain multiple comparisons at pcorr < 0.05.
To assess if functional connectivity changes in claustrum or control regions-related circuits are associated with incubation of oxycodone craving after electric barrier-induced abstinence, we conducted a voxelwise correlation analysis between functional connectivity changes in the voluntary abstinence phase (late abstinence minus early abstinence) and the incubation score (active lever presses during the relapse tests on Abstinence Day 15 minus lever presses on Day 1 in the oxycodone group; n = 18). The incubation score was higher on Day 15 than on Day 1 for all rats, and the data are reported in Fredriksson et al. (2021b). We also conducted a similar correlation analysis between functional connectivity changes in the self-administration phase (early abstinence minus pretraining) and the incubation score in the oxycodone group. We corrected for multiple comparisons and used corrected p < 0.05 (with uncorrected p < 0.05 and cluster size >27 based on Monte Carlo simulation in Analysis of Functional NeuroImages) as a threshold to determine voxels that are significant in the analyses.
To assess the specificity of these associations, we extracted functional connectivity signals in the food group (n = 19) from brain regions showing significance in the oxycodone group. We then conducted a region-of-interest (ROI)-based correlation analysis between functional connectivity changes in the voluntary abstinence phase and incubation (or abatement) score of food.
In addition, to explore whether functional connectivity changes in claustrum–vSub circuits are associated with the incubation score, we calculated functional connectivity between claustrum (left, right, or bilateral) and vSub (left, right, or bilateral) using the similar area as in our previous study (Fredriksson et al., 2023). Finally, we examined the correlation between functional connectivity changes in these claustrum–vSub circuits during the voluntary abstinence phase and the oxycodone incubation score.
Results
In the experiments described below, we used a rat model of incubation of oxycodone seeking after food choice-induced or electric barrier-induced voluntary abstinence (Fredriksson et al., 2020). The experiments included some or all the following phases: food self-administration training (6 d), oxycodone self-administration training (14 d), early tests for oxycodone seeking (Abstinence Day 1), electric barrier-induced abstinence (10–11 d) or food choice-induced abstinence (11 d), and late tests for oxycodone seeking (Abstinence Day 15–16). We used both male and female rats in Exp. 1–6 but did not use sex as a factor in the statistical analyses described below and in Table 1 because in our previous studies, which were statistically powered to detect sex differences, we did not observe sex differences in food self-administration, oxycodone self-administration, electric barrier-induced abstinence, food choice-induced abstinence, or incubation of oxycodone seeking (Fredriksson et al., 2020).
Behavioral data for self-administration and voluntary abstinence phases
The male and female rats demonstrated reliable food and oxycodone self-administration (Fig. 1A–E), as indicated by a significant increase in the number of food rewards, oxycodone infusions, and active lever presses over the training sessions. The complete analyses for the number of rewards, infusions, and active and inactive lever presses during training are described in Table 1.
The rats in Exp. 1–2 and 4–6 voluntarily abstained from drug self-administration when we introduced an electric barrier of increasing shock intensity near the active lever, as indicated by a significant decrease in the number of infusions and active lever presses during the abstinence phase (Fig. 1A,B,D–F). In Exp. 3, the rats voluntarily abstained from drug self-administration when we introduced the discrete choice between food and oxycodone as indicated by a significant decrease in the number of infusions during the abstinence phase (Fig. 1C). The mean number of infusions for the last 3 d of electric barrier-induced abstinence or food choice-induced abstinence was <2 per session. The statistical analyses of the electric barrier-induced abstinence and food choice-induced abstinence phases are described in Table 1.
Exp. 1: effect of incubated oxycodone seeking on Fos expression in afferent projections to vSub
The goal of Exp. 1 was to determine if incubation of oxycodone seeking after electric barrier-induced abstinence is associated with selective activation of one or more afferent projections to vSub. For this purpose, we used Fos expression and CTb labeling in male and female rats tested for oxycodone seeking under extinction conditions after electric barrier-induced abstinence (see Fig. 2A for the experimental timeline). Representative locations of CTb spread of vSub injection sites are shown in Figure 2C.
Relapse (incubation) test
The rats tested after 15–16 abstinence days showed higher active lever presses than inactive lever presses (Fig. 2B). The mixed-factorial ANOVA for the number of active lever presses that included the between-subject factor of lever (inactive, active) and the within-subject factor of session time (30, 60, or 90 min) showed significant lever × session time interaction (F(2,20) = 97.4; p < 0.001).
Fos and CTb immunohistochemistry
We measured Fos, CTb, and Fos + CTb expression after the Day 15–16 relapse (incubation) test in claustrum (Fig. 2) and other brain regions (Fig. 3). The one-way ANOVA for Fos-positive, CTb-positive, and Fos + CTb-positive cells per square millimeter included the between-subject factor of test condition (test, no-test).
Fos
Relapse to oxycodone seeking during testing was associated with a significant increase (p < 0.05) in Fos expression in anterior and posterior claustrum (F(1,20) = 58.6; p < 0.001; F(1,20) = 80.3; p < 0.001), anterior and posterior RE (F(1,20) = 35.1; p < 0.001; F(1,20) = 39.9; p < 0.001), anterior BLA (F(1,20) = 14.1; p = 0.038), and posterior MS/DB (F(1,20) = 16.3; p = 0.019; Figs. 2E, 3C, top).
CTb
There were no significant group (no-test, test) differences (Bonferroni-corrected p < 0.05) in the different brain regions (Figs. 2E, 3C, middle).
Fos ± CTb
Relapse to oxycodone seeking during testing was associated with a significant increase (Bonferroni-corrected p < 0.05) in Fos + CTb expression only in anterior and posterior claustrum (F(1,20) = 20.2; p = 0.007; F(1,20) = 42.2; p < 0.001; Figs. 2E, 3C, bottom). Note: The CTb values in the claustrum of the test rats were somewhat higher than those of the no-test rats (Fig. 2E). To account for this difference, we reanalyzed the Fos + CTb data using CTb values as a covariate. The ANCOVA analysis showed significant differences (F(1,19) = 11.5; p = 0.003; F(1,19) = 28.3; p < 0.001; for anterior and posterior claustrum; Table 1), confirming that relapse after electric barrier-induced abstinence was associated with activation of the claustrum-to-vSub projection. Representative pictures of Fos and CTb labeling in each ROI of the no-test and test rats are shown in Figures 2 and 3.
Taken together, the results of Exp. 1 showed that incubation of oxycodone seeking after electric barrier-induced voluntary abstinence is associated with increased neuronal activity in several regions with neurons projecting to the vSub. Of those regions, we followed up on the Fos + CTb data of the claustrum, because of the robust double labeling in this region compared with the double labeling in the other brain regions.
Exp. 2: effect of bilateral claustrum inactivation on incubation after electric barrier-induced abstinence
In Exp. 1 we found that incubation of oxycodone seeking after electric barrier-induced abstinence was associated with increased Fos expression in claustrum neurons projecting to vSub. In Exp. 2, we determined a causal role of the claustrum in this form of incubation, using the classical muscimol + baclofen inactivation procedure (McFarland and Kalivas, 2001). We injected muscimol + baclofen to the more posterior claustrum region shown in Figure 2. We tested male and female rats for the effect of saline or muscimol + baclofen injections on oxycodone seeking 1 d after oxycodone self-administration training (Day 1) and after electric barrier-induced abstinence (Day 15). The rats were tested for 30 min on Day 1 and 90 min on Day 15. We tested the rats for only 30 min on Day 1 to minimize potential carryover effects of extinction learning that can confound data interpretation on Day 15. Representative pictures of the claustrum injector tip placements are shown in Figure 4B.
Relapse test. Active lever pressing was higher after electric barrier-induced abstinence than 1 d after self-administration training (incubation of oxycodone seeking; Fig. 4C, vehicle condition), and this effect was decreased by muscimol + baclofen injections into the claustrum.
Analysis of the first 30 min data of the tests on Days 1 and 15 showed that muscimol + baclofen injections decreased incubated oxycodone seeking on Day 15 but had no significant effect on Day 1. The mixed-factorial ANCOVA (inactive lever as a covariate) for the number of active lever presses, which included the between-subject factors of muscimol + baclofen dose (0, 50 + 50 ng/side) and the within-subject factors of abstinence day (1 and 15) and session time (10, 20, or 30 min), showed significant abstinence day × muscimol + baclofen dose interaction (F(1,17) = 20.7; p < 0.001).
The mixed-factorial ANCOVA (inactive lever as a covariate) for the number of active lever presses during the 90 min test on Day 15, which included the between-subject factors of muscimol + baclofen dose and the within-subject factor of session time (30, 60, or 90 min), showed significant effects of muscimol + baclofen dose × session time (F(2,36) = 6.3; p = 0.005; Fig. 4D).
Together, the results of Exp. 2 showed that inhibition of claustrum neuronal activity decreased incubated oxycodone seeking but had no significant effect on “nonincubated” oxycodone seeking 1 d after drug self-administration training.
Exp. 3: effect of bilateral claustrum inactivation on incubation after food choice-induced abstinence
In Exp. 2, we found that claustrum inactivation decreased incubation of oxycodone seeking after electric barrier-induced abstinence. In Exp. 3, we determined if the claustrum also contributes to incubation of oxycodone seeking after voluntary abstinence is achieved by giving rats access to an alternative nondrug food reward (Caprioli et al., 2017; Fredriksson et al., 2021a). We tested different groups of male and female rats for the effect of saline or muscimol + baclofen injections on oxycodone seeking 1 d after oxycodone self-administration training and after 2 weeks of food choice-induced abstinence. Representative images of claustrum injector tip placements are shown in Figure 5B.
Relapse test. Active lever pressing was higher after food choice-induced abstinence than after 1 d after self-administration training (incubation of oxycodone seeking; Fig. 5C), and this effect was not affected by claustrum inactivation (Fig. 5C,D).
The mixed-factorial ANCOVA (inactive lever as a covariate) for the number of active lever presses during the 30 min test of Day 1 and the first 30 min of the test of Day 15 included the between-subject factors of muscimol + baclofen dose and the within-subject factors of abstinence day and session time (10, 20, 30 min). The analysis showed significant effects of abstinence day (F(1,17) = 13.2; p = 0.002) and session time (F(2,34) = 23.7; p < 0.001), but no significant effects of muscimol + baclofen dose or interactions with this factor. The mixed-factorial ANCOVA for number of active lever presses during the 90 min test on Day 15, which included the between-subjects factors of muscimol + baclofen dose and the within-subject factor of session time (30, 60, 90 min), showed a significant effect of session time (F(2,36) = 17.5; p < 0.001), but no significant effects of muscimol + baclofen dose or interactions with this factor.
Together, the results of Exp. 3 showed that inhibition of claustrum activity had no effect on incubation of oxycodone seeking after food choice-induced abstinence.
Exp. 4: effect of bilateral inactivation dorsolateral to claustrum on incubation after electric barrier-induced abstinence
The goal of Exp. 4 was to determine the anatomic specificity of the inhibitory effect of claustrum inactivation by injecting muscimol + baclofen dorsolateral to the claustrum injection site (Fig. 6B). We tested different groups of male and female rats for the effect of saline or muscimol + baclofen injections on oxycodone seeking 1 d after oxycodone self-administration training and after electric barrier-induced abstinence (Day 15). Representative images of the injector tip placements are shown in Figure 6B.
Relapse test. Active lever pressing was higher after electric barrier-induced abstinence than after 1 d after self-administration training (incubation of oxycodone seeking; Fig. 6C), and this effect was not decreased by muscimol + baclofen injections (Fig. 6C,D).
The mixed-factorial ANCOVA (inactive lever as a covariate) for the number of active lever presses during the 30 min test of Day 1 and the first 30 min of the test of Day 15 included the between-subject factor of muscimol + baclofen dose and the within-subject factors of abstinence day and session time (10, 20, 30 min). The analysis showed significant effects of abstinence day (F(1,22) = 16.2; p < 0.001) and session time (F(2,44) = 34.8; p < 0.001), but no significant effects of muscimol + baclofen dose or interactions with this factor.
The mixed-factorial ANCOVA for the number of active lever presses during the 90 min test on Day 15, which included the between-subject factor of muscimol + baclofen dose and the within-subject factor of session time (30, 60, 90 min), showed a significant effect of session time (F(2,46) = 41.5; p < 0.001), but no significant effects of muscimol + baclofen dose or interactions with this factor.
Together, the results of Exp. 4 showed that inhibition of the insula/somatosensory cortex area near the claustrum had no effect on incubation oxycodone seeking after electric barrier-induced abstinence.
Exp. 5: effect of anatomical disconnection of claustrum–vSub connections on incubation after electric barrier-induced abstinence
The goal of Exp. 5 was to determine if the projection from claustrum to vSub is critical to incubation of oxycodone seeking after electric barrier-induced abstinence. For this purpose, we used an anatomical disconnection procedure. In this method, the role of a neuronal pathway in behavior is inferred from the observation that a lesion/inactivation (permanent or reversible) or receptor blockade of one brain site in one hemisphere, together with lesion/receptor blockade of a second brain site in the contralateral hemisphere, disrupts the behavior of interest. The assumptions of “disconnection” studies are that the target behavior is at least partially intact following ipsilateral lesion/inactivation of the two brain sites in the same hemisphere and that neuronal projections are exclusively or primarily ipsilateral (Gold, 1966; Everitt et al., 1991; Gaffan et al., 1993; Setlow et al., 2002).
We tested different groups of male and female rats for the effect of ipsilateral or contralateral claustrum and vSub saline or muscimol + baclofen injections on oxycodone seeking 1 d after oxycodone self-administration training and after electric barrier-induced abstinence. Representative pictures of the claustrum and vSub injector tip placements are shown in Figure 7B. For the statistical analysis, we combined the rats injected with vehicle ipsilaterally or contralaterally into a single vehicle group, as there were no statistically significant differences in lever presses during the relapse tests between these two conditions.
Relapse test. Active lever pressing was higher after electric barrier-induced abstinence than 1 d after self-administration training (incubation of oxycodone seeking; Fig. 7C vehicle condition). Contralateral or ipsilateral inactivation of claustrum and vSub significantly decreased incubated oxycodone seeking on Day 15 but not nonincubated oxycodone seeking on Day 1.
The mixed-factorial ANCOVA (inactive lever as a covariate) for the number of active lever presses, which included between-subject factor of the group (vehicle, ipsilateral muscimol + baclofen in claustrum and vSub, contralateral muscimol + baclofen in claustrum and vSub), and the within-subject factors of abstinence day and session time (10, 20, or 30 min), showed significant effects of abstinence day (F(1,31) = 19.3; p < 0.001), group (F(2,31) = 3.9; p = 0.030), session time (F(2,62) = 55.8; p < 0.001), abstinence day × group (F(2,31) = 3.3; p = 0.050), and abstinence day × session time (F(2,62) = 18.2; p < 0.001). Post hoc one-way ANOVAs of active lever presses showed a significant group effect on Day 15 (F(2,33) = 4.2; p = 0.023) but not Day 1 (F(2,33) = 2.3; p = 0.12).
The mixed-factorial ANCOVA for the number of active lever presses during the 90 min test on Day 15, which included the between-subject factor of the group and the within-subject factor of session time (30, 60, 90 min), showed significant effects of the group (F(2,32) = 5.5; p = 0.009) and session time (F(2,64) = 33.3; p < 0.001; Fig. 7D).
The results of Exp. 5 showed that inhibition of both ipsilateral and contralateral claustrum projections to vSub decreased incubated oxycodone seeking after electric barrier-induced abstinence.
Exp. 6: effect of unilateral claustrum and vSub inactivation on incubation after electric barrier-induced abstinence
In Exp. 5, we found that ipsilateral or contralateral inactivation of projections between the claustrum and vSub decreased incubation of oxycodone seeking after electric barrier-induced abstinence. This finding suggests that both intra- and interhemispheric claustrum–vSub connections contribute to this incubation. However, an alternative explanation is that inactivating either the claustrum or vSub in one hemisphere alone is sufficient to decrease incubation. To rule out this possibility, we tested male and female rats implanted with bilateral cannulas targeting either the claustrum or vSub. We examined the effects of bilateral vehicle injections versus muscimol + baclofen injections in one hemisphere and saline injections in the other hemisphere on incubation of oxycodone seeking. We tested the rats 1 d after oxycodone self-administration training and again after 2 weeks of electric barrier-induced abstinence. Representative images of injector tip placements in the claustrum and vSub are shown in Figure 8B. For statistical analysis, we combined rats injected with vehicle into the claustrum or vSub into a single vehicle group, as there were no statistically significant differences in lever presses during the relapse tests between these two conditions.
Relapse test. Active lever pressing was higher after electric barrier-induced abstinence than 1 d after self-administration training (incubation of oxycodone seeking; Fig. 8C, vehicle condition), and this effect was not decreased by unilateral muscimol + baclofen injections into either the claustrum or vSub.
The mixed-factorial ANCOVA (inactive lever as a covariate) for the number of active lever presses, which included between-subject factor of the group (vehicle, unilateral muscimol + baclofen in claustrum, unilateral muscimol + baclofen in vSub) and the within-subject factors of abstinence day and session time (10, 20, or 30 min), showed significant effects of abstinence day (F(1,30) = 28.6; p < 0.001), session time (F(2,60) = 54.5; p < 0.001; Fig. 8C), and group × session time (F(4,60) = 3.6; p = 0.011; Fig. 8C). The interaction is due to higher responding at the 10 min timepoint and lower responding at the 20 and 30 min timepoint in the muscimol + baclofen claustrum group than in the vehicle and muscimol + baclofen vSub groups.
The mixed-factorial ANCOVA for the number of active lever presses during the 90 min test on Day 15, which included the between-subject factor of the group and the within-subject factor of session time (30, 60, 90 min), showed a significant effect showed session time (F(2,62) = 37.0; p < 0.001) but not group or interaction between the two factors (p > 0.05; Fig. 8D).
The results of Exp. 6 showed that unilateral inactivation of claustrum or vSub had no significant effect on incubated oxycodone seeking after electric barrier-induced abstinence.
Longitudinal functional connectivity changes in claustrum-related circuits predict incubated oxycodone seeking after electric barrier-induced abstinence
In Exp. 2, we found that claustrum inactivation decreased incubated oxycodone seeking after electric barrier-induced abstinence. To determine potential claustrum-related circuits associated with incubated oxycodone seeking, we conducted whole-brain voxelwise functional connectivity analyses with the claustrum as a seed (Fig. 9B). We determined whether longitudinal claustrum-related functional connectivity changes induced by voluntary abstinence (from early abstinence to late abstinence; Fig. 10A) or oxycodone self-administration (from pretraining to early abstinence) were correlated with incubated oxycodone seeking assessed by the incubation score (active lever presses on Abstinence Day 15 minus active lever presses on Day 1).
Voluntary abstinence phase. Voxelwise correlation analyses showed that functional connectivity changes of the claustrum with the olfactory bulb, orbitofrontal cortex, insula, sensory cortex, cingulate cortex, retrosplenial cortex, and visual cortex (Fig. 10B) were significantly correlated with the incubation score in the oxycodone group. These claustrum–cortical circuits showed negative correlations with the incubation score of oxycodone (Fig. 10C). To verify the specificity of the claustrum seed results, we also conducted similar analyses using adjacent insula and striatum as control seeds (Table 2). The results showed that functional connectivity of the insula with the dorsal striatum (but not the other regions above) was significantly correlated with incubation score (Pearson's r = −0.92; p < 0.0001; Fig. 11A). No significant correlations were found for the adjacent striatum seed (Fig. 11B). The different results between the control seeds and the claustrum seed indicate that partial volume effects are minimal in our claustrum analysis.
Correlations between functional connectivity changes in control seeds (insula and striatum adjacent to the claustrum) and incubation score during electric barrier-induced abstinence in the oxycodone group
To determine whether the associations between these circuits and behavior are specific to the oxycodone group, we correlated functional connectivity changes of the same circuits in the food group in the abstinence phase with incubation (or abatement) score of food (active lever presses for the food-trained rats were lower on Day 15 than on Day 1; Fredriksson et al., 2021b). Among the seven claustrum circuits, only claustrum–cingulate cortex connectivity was associated with the incubation score in the food group (Table 3).
Correlations between functional connectivity changes in claustrum-related circuits during the electric barrier-induced abstinence phase and the incubation (or abatement) score in the food group
We also analyzed the functional connectivity changes of ipsilateral, contralateral, and bilateral claustrum–vSub circuits with incubation score (Table 4). Only the functional connectivity changes between left claustrum and left vSub during voluntary abstinence was marginally correlated with the incubation score in the oxycodone group (r = 0.46; p = 0.056).
Correlations between functional connectivity changes in the claustrum–vSub circuits during the electric barrier-induced abstinence phase and the incubation score in the oxycodone group
Self-administration phase. Voxelwise correlation analyses showed no significant correlation between functional connectivity changes of claustrum-related circuits and the incubation score in the oxycodone group.
Together, the results of the fMRI analyses showed that functional connectivity changes of claustrum–cortical circuits during the electric barrier-induced abstinence phase predicted the incubation score in the oxycodone group. In contrast, functional connectivity changes during the self-administration phase did not predict the incubation score.
Discussion
We studied the role of claustrum in incubation of opioid seeking after electric barrier-induced abstinence and report four main findings. First, incubation was associated with activation of claustrum projections to vSub. Moderate or low activation was observed in MS/DB and reuniens projections, but not in BLA or paraventricular thalamus projections. Second, claustrum inactivation decreased incubated oxycodone seeking after electric barrier-induced abstinence, with no effect on nonincubated seeking on Abstinence Day 1 or incubation after food choice-induced abstinence. Additionally, inactivation of regions dorsolateral to the claustrum had no effect. Third, contralateral and ipsilateral coinactivation of claustrum and vSub decreased incubated oxycodone seeking, an effect not observed with single-hemisphere inactivation of either region. Fourth, functional connectivity changes in claustrocortical circuits during abstinence predicted incubated oxycodone relapse.
Results indicate that the claustrum is critical to incubation of oxycodone seeking after electric barrier-induced abstinence and that claustrum-related functional connectivity changes predict this incubation. Our data also suggest that claustrum projections to vSub contribute to incubation, but these results should be interpreted with caution (see below).
The role of claustrum in behavioral effects of addictive drugs
The claustrum and its cortical projections contribute to diverse functions (Brown et al., 2017; Narikiyo et al., 2020; Madden et al., 2022), including pain responses (Stewart et al., 2024) and response to psychedelic drugs (Doss et al., 2022). Recent studies explored the claustrum role in behavioral and physiological effects of psychostimulant and opioid drugs. Regarding psychostimulants, Terem et al. (2020) reported that inhibition of claustrum or Drd1-expressing claustrum neurons projecting to the frontal cortex decrease acquisition but not expression of cocaine place preference (CPP). Zhao et al. (2024) reported that claustrum activity contributes to the effect of adolescent cocaine exposure on cocaine CPP and anxiety-like behaviors. Liu et al. (2019) reported that inhibition of the claustrum–prefrontal cortex pathway decreases methamphetamine-induced impulsivity.
Regarding opioids, Brynildsen et al. (2020) reported that withdrawal from chronic morphine exposure is associated with increased Fos in the claustrum. Terem et al. (2023) used a group-housed oral fentanyl self-administration procedure in mice (Peretz-Rivlin et al., 2024). They reported decreased activity of claustrum neurons projecting to the frontal cortex during fentanyl self-administration and that activation of these neurons decreases fentanyl intake. The authors concluded that activity of claustrum projections to the cortex inhibits fentanyl intake.
In contrast, our Fos and inactivation data indicate that claustrum activity promotes opioid relapse. What might account for the different conclusions from our study and Terem et al. (2023) study? We propose several possibilities beyond species differences. The first is the experimental manipulations. Terem et al. inhibited claustrum projections to the frontal cortex, while we reversibly inactivated all claustrum neurons. Another consideration is the “addiction phase”: ongoing opioid self-administration versus relapse during abstinence. Indeed, many studies reported that mechanisms of opioid relapse are distinct from those controlling opioid self-administration (Shalev et al., 2002). For example, mesolimbic dopamine transmission is critical to relapse/reinstatement of heroin seeking but not heroin self-administration (Badiani et al., 2011). Other possible reasons are differences in route of administration (oral vs intravenous) and the opioid drug (fentanyl vs oxycodone). Regarding the latter, there is evidence for different behavioral and physiological effects of opioid drugs (Milella et al., 2023).
Finally, we previously reported that resting-state functional connectivity changes in the orbitofrontal cortex with the dorsal striatum and related circuits during oxycodone self-administration and/or electric barrier-induced abstinence predict incubation of oxycodone seeking after electric barrier-induced abstinence (Fredriksson et al., 2021b). We also reported that abstinence-related functional connectivity changes of vSub with the retrosplenial cortex and early-abstinence functional connectivity of vSub with the dorsal hippocampus predict incubation (Fredriksson et al., 2023). Here, we used brain images from the original study (Fredriksson et al., 2021b) to determine if resting-state connectivity changes in claustrum-related circuits predict incubation. We found that functional connectivity changes of the claustrum with the orbitofrontal cortex, insula, sensory cortex, cingulate cortex, retrosplenial cortex, and visual cortex during the abstinence phase were associated with incubation of oxycodone seeking after electric barrier-induced abstinence. Future studies on the causal role of these claustrum-related circuits in relapse to opioid seeking are warranted. One area of interest is the retrosplenial cortex whose functional connectivity with both vSub (Fredriksson et al., 2023) and claustrum predicts incubation. Other questions for future research include identifying the cell types of Fos-positive neurons in the claustrum that contribute to incubation and the role in incubation of their projections to regions other than vSub.
Methodological considerations
One consideration is that claustrum inactivation caused nonspecific behavioral deficits. This possibility is unlikely because claustrum inactivation had no effect on incubation after food choice-induced abstinence or nonincubated oxycodone seeking on Day 1. Another consideration is anatomical specificity. After intracranial injections, drugs diffuse away from the injection site and can target nearby regions (Wise and Hoffman, 1992). To address this concern, we injected muscimol + baclofen dorsolateral to the claustrum and found that these injections had no effect, suggesting that the effect of claustrum inactivation is primarily due to claustrum inhibition. However, we cannot rule out that diffusion to other nearby regions (e.g., dorsal endopiriform nucleus), not targeted by our anatomical control injections, contributes to the effect of muscimol + baclofen on incubation.
Another consideration is the interpretation of the data from the disconnection experiment within the context of the role of claustrum-to-vSub projections (identified in Exp. 1) in incubation. One consideration is that the behavioral effects of the disconnection manipulation may be due to interference with activity of vSub-to-claustrum projections (Canteras and Swanson, 1992). Another consideration is the similar effect of the ipsilateral and contralateral manipulations. There are two potential reasons for these similar effects. The first is that incubation after electric barrier-induced abstinence requires intact claustrum and vSub connectivity in both hemispheres. The second is that both ipsilateral and contralateral disconnection disrupts activity in brain regions connected to the claustrum and vSub that contribute to incubation. It will be important to assess the role of cross-hemispheric communication in brain regions associated with both claustrum and vSub, particularly in cortical areas that predominantly project to contralateral claustrum (Smith and Alloway, 2010, 2014). We also found that unilateral inactivation of claustrum or vSub had no effect on incubation after electric barrier-induced abstinence, supporting the possibility that direct or indirect projections between claustrum and vSub contribute to incubation.
Another issue to consider is the interpretation of the correlational Fos + CTb double-labeling data within the context of the role of claustrum-to-vSub projection in incubation. Specifically, during the relapse test, this projection can be activated by other factors that may or may not be related to incubation of opioid seeking like locomotor activity, the operant task, exposure to the drug-paired context and cues, and more. Additionally, higher Fos–CTb double labeling can be due to higher overall Fos in the test versus the no-test groups. Because of these interpretation issues, we and others use Fos as an initial screen to guide the future causal role of mechanistic experiments, as we have done in the present study.
Two other issues to consider are that we did not assess all CTb-labeled projections to vSub for Fos coexpression. We observed CTb labeling after tracer injections to vSub in other regions like entorhinal cortex, field CA1 of the hippocampus, and midline thalamic structures that we did not quantify. The role of these projections to vSub in incubation of opioid seeking after electric barrier-induced abstinence is a subject for future research.
A methodological challenge with the fMRI analysis is that the claustrum is a small region surrounded by the insula, striatum, and endopiriform. Thus, partial volume effect of signal from adjacent structures is a potential confound. To address this concern, we regressed claustrum time series from signals of surrounding structures (Krimmel et al., 2019a,b). Additionally, to verify specificity of our claustrum seed analysis, we put seeds in the adjacent insula and striatum. These control analyses indicate that partial volume effect is minimal in our claustrum analysis.
Finally, functional connectivity changes between claustrum and vSub were not significantly correlated with the incubation score. This result appears contradictory to our anatomical disconnection results, which suggest a causal role of projections between the two regions and incubation. These inconsistent findings suggest that task-independent correlational resting-state functional connectivity fMRI data should be interpreted with caution as they reflect system-level synchrony between brain regions, not causal roles of the regions in behavior (Barttfeld et al., 2015; Smith et al., 2017).
Conclusions
Using a rat model where incubation of opioid seeking is potentiated after electric barrier-induced abstinence (Negishi et al., 2024), we identified a role of claustrum in this incubation but not incubation after food choice-induced abstinence. Our results extend findings from previous studies showing that brain regions and circuits that control the relapse-related behaviors in animal models depend on the method used to achieve abstinence (Pelloux et al., 2018; Marchant et al., 2019; Fredriksson et al., 2021a; Chow et al., 2025). We also used an existing fMRI dataset and identified several cortical regions whose functional connectivity with claustrum predicted incubation of oxycodone seeking.
Data Availability
The individual data of the behavioral and neurochemical data are available upon request (Ida Fredriksson and Yavin Shaham). The imaging data are available upon request (Yihong Yang).
Footnotes
This research was supported by funds from the Intramural Research Program of the NIDA–NIH (Y.S. and Y.Y.), the Swedish Research Council Starting Grant (2022-01538, I.F.), the Swedish Society of Medical Research Starting Grant (SG-22-0118, I.F.), and a fellowship from the NIH Center for Compulsive Behavior (K.N.).
↵*Y.Y., Y.S., and I.F. are co-senior authors.
The authors declare no competing financial interests.
- Correspondence should be addressed to Ida Fredriksson at ida.fredriksson{at}liu.se, Yihong Yang at yihongyang{at}mail.nih.gov, or Yavin Shaham at yavin.shaham{at}nih.gov.
