Effects of Withdrawal from Cocaine Self-Administration on Rat Orbitofrontal Cortex Parvalbumin Neurons Expressing Cre recombinase: Sex-Dependent Changes in Neuronal Function and Unaltered Serotonin Signaling

Abstract The orbitofrontal cortex (OFC) is a brain region involved in higher-order decision-making. Rodent studies show that cocaine self-administration (CSA) reduces OFC contribution to goal-directed behavior and behavioral strategies to avoid drug intake. This change in OFC function persists for many weeks after cocaine withdrawal, suggesting involvement in the process of addiction. The mechanisms underlying impaired OFC function by cocaine are not well-understood. However, studies implicate altered OFC serotonin (5-HT) function in disrupted cognitive processes during addiction and other psychiatric disorders. Thus, it is hypothesized that cocaine impairment of OFC function involves changes in 5-HT signaling, and previous work shows that 5-HT1A and 5-HT2A receptor-mediated effects on OFC pyramidal neurons (PyNs) are impaired weeks after cocaine withdrawal. However, 5-HT effects on other contributors to OFC circuit function have not been fully investigated, including the parvalbumin-containing, fast-spiking interneurons (OFCPV), whose function is essential to normal OFC-mediated behavior. Here, 5-HT function in naive rats and those withdrawn from CSA were evaluated using a novel rat transgenic line in which the rat parvalbumin promoter drives Cre-recombinase expression to permit identification of OFCPV cells by fluorescent reporter protein expression. We find that whereas CSA altered basal synaptic and membrane properties of the OFCPV neurons in a sex-dependent manner, the effects of 5-HT on these cells were unchanged by CSA. These data suggest that the behavioral effects of dysregulated OFC 5-HT function caused by cocaine experience are primarily mediated by changes in 5-HT signaling at PyNs, and not at OFCPV neurons.


Introduction
The orbitofrontal cortex (OFC), a prefrontal cortical (PFC) area, is a brain region involved in executive cognitive function in domains such as associative learning, reversal learning, outcome prediction, and estimation of reward value (Jones and Mishkin, 1972;Schoenbaum et al., 1998Schoenbaum et al., , 2016Gallagher et al., 1999;Tremblay and Schultz, 1999;Izquierdo and Murray, 2004;Hardung et al., 2017). A hallmark of OFC function is its contribution to flexible behavior as defined by the ability to shift behavioral strategies in the face of changing contingencies and expected outcomes (Ostlund and Balleine, 2007;Schoenbaum and Esber, 2010;Gremel and Costa, 2013;Izquierdo, 2017).
A growing body of evidence shows that 5-HT is critical for OFC-dependent behavior and that dysregulation of 5-HT function is implicated in psychiatric disorders where changes in OFC-dependent executive function are observed (Roberts, 2011). Additionally, pharmacologically targeting the 5-HT system is associated with positive treatment outcomes and increased OFC metabolism in psychiatric disorders (Saxena et al., 1999;New et al., 2004). In animal studies, OFC-dependent behavioral tasks that require shifts in behavioral strategies to obtain reward are impaired by the depletion of 5-HT (Clarke et al., 2004;Roberts, 2011;Izquierdo et al., 2012;West et al., 2013). Moreover, selective ligand activation of 5-HT receptor subtypes within the OFC alters reversal learning, a form of behavioral flexibility (Boulougouris and Robbins, 2010;Furr et al., 2012), and other OFC-dependent behaviors such as reinforcer devaluation and response inhibition are blocked by lesions of OFC 5-HT axons (West et al., 2013;Man et al., 2010).
The inability to change behavior in the absence of reward or in the presence of aversive consequences is endemic to cocaine use disorder (CUD) in humans, and cocaine exposure is associated with the dysregulation of OFC function in humans and in animal models (Volkow and Fowler, 2000;Bechara, 2005;Schoenbaum and Shaham, 2008;Lucantonio et al., 2012Lucantonio et al., , 2014Simmler et al., 2017). Thus, CUD is associated with a reduction in OFC metabolic activity that persists after cocaine withdrawal (Volkow et al., 1993;Volkow and Fowler, 2000), and enduring impairments in performance of behavioral tasks relying on intact OFC function are observed in animals following cocaine withdrawal. This includes reductions in the ability to change strategies to obtain reward (e.g., reversal learning; Jentsch et al., 2002;Schoenbaum et al., 2004;Calu et al., 2007), and in performance of non-operant behaviors such as reinforcer devaluation (Schoenbaum and Setlow, 2005) and overexpectation learning (Takahashi et al., 2013;Lucantonio et al., 2014). These findings point toward longlasting changes in OFC function contributing to disordered behavior following cocaine exposure in humans and in animals (Lucantonio et al., 2014). However, the neural mechanisms through which this occurs are poorly understood.
The changes in OFC-dependent behavior observed after cocaine experience are hypothesized to result from altered serotonergic control of OFC neuron activity as large reductions in 5-HT-mediated electrophysiological effects and altered expression of 5-HT 2A receptor transcripts are observed in OFC pyramidal neurons (PyNs) weeks after withdrawal from cocaine self-administration (CSA) or yoked cocaine administration (Wright et al., 2017). However, the output of OFC, like other cortical circuits, relies on the interaction of glutamatergic PyNs (principal cells) with less-numerous local circuit GABAergic interneurons. One class of these interneurons, known as fast-spiking parvalbumin neurons (OFC PV ), regulates PyN excitability via inhibitory synapses and plays an essential role in the generation of g oscillations that promote frontocortical synchrony necessary for OFC-dependent behavior (Bissonette et al., 2015). As cortical parvalbumin neurons also express 5-HT receptors (Santana et al., 2004;Athilingam et al., 2017), and promote cortical synchrony (Puig et al., 2010), we hypothesize that withdrawal from CSA will also alter 5-HT responses in these cells. To evaluate this, a novel transgenic rat (Pvalb-iCre) was developed in which Cre-recombinase expression was driven by the parvalbumin promoter, and a Cre-dependent viral targeting strategy was used to examine 5-HT function in these cells following withdrawal from CSA. We report that although sex-dependent differences in OFC PV neuron properties were observed after CSA, and effects of 5-HT on these cells differed between male and female cocaine-naive rats, no changes in response to 5-HT were observed after cocaine withdrawal. Therefore, we conclude that the enduring effects of cocaine on OFC 5-HT function and behavior likely occur primarily through actions at PyNs and not OFC PV cells.

Materials and Methods
Generation of a bacterial artificial chromosome (BAC) for Pvalb-iCre A BAC carrying a 70-kb fragment of rat genomic DNA containing the Pvalb gene locus was obtained from the Children's Hospital Oakland Research Institute (CH230-499N20). The BAC was transformed into the recombineering bacterial strain (GS1783) using electroporation and edited using "en passant mutagenesis" to replace the start codon of the Pvalb gene (Tischer et al., 2010;Shimshek et al., 2002). Specifically, a cassette containing an I-SceI restriction site and a KanS selection marker was amplified from pEP-KanS (a gift from Nikolaus Osterrieder; Addgene plasmid #41017; http://n2t.net/ addgene:41017; RRID:Addgene_41017) and inserted into a plasmid containing the coding region for iCre (pBS mFos TetO iCre, pOTTC161). A 50-bp portion of the iCre coding region was duplicated on both sides of the insertion to facilitate intramolecular recombination and marker removal in subsequent steps. The final plasmid, pBS mFos TetO iCre(KanS) (pOTTC4094) was used as a template to PCR amplify the entire iCre-(I-SceI-KanS) coding region with primers appended with 59 linkers corresponding to 50 bp of homology with the sequences adjacent to the Pvalb start codon. This PCR product was electroporated into the GS1783 cells containing the Pvalb BAC after heat shock, and the transformants were selected on LB agar plates containing chloramphenicol and kanamycin. Successful transformants were screened by PCR across the 59 and 39 junctions and sequence verified for error-free insertion. The kanamycin resistance marker was removed, and the iCre coding region restored by the simultaneous treatment with arabinose and heat shock. Cultures were diluted and streaked onto plates containing LB agar plates containing chloramphenicol. Successful recombinants were screened by PCR and verified by sequencing and pulse-field agarose gel electrophoresis. The final Pvalb-iCre BAC (pOTTC465) was prepped using Macherey-Nagel NucleoBond Xtra BAC columns and then used to create the transgenic rat line. See schematic of final transgene (Fig. 1, transgene and characterization).

Transgenic rat husbandry
All animal procedures were performed in accordance with National Institutes of Health Animal Care Guidelines and approved by local animal care and use committees. Procedures were initially based on those for optimizing the production of transgenic rats (Filipiak and Saunders, 2006) with modifications described here. Female Long-Evans rats purchased from Charles River Laboratories and maintained in a 6 A.M. to 6 P.M. light cycle with food and water ad lib were superovulated and mated as described (Filipiak and Saunders, 2006). Briefly, the ovulation of females, four to five weeks of age, was synchronized using 40-mg LHRH (Luteinizing Hormone Releasing Hormone, Sigma Chemicals) injected intraperitoneally on day 1. Superovulation was achieved with 20 IU pregnant mare serum gonadotropin (PMSG; National Hormone and Peptide Program) 48 h later at 10 A.M. Another 48 h later, 20 IU human chorionic gonadotropin (hCG; Novarel, Ferring) was injected intraperitoneally. Directly after these injections the females were added to cages with singly housed Long-Evans stud males. The next morning at 9 A.M. copulation plugs were identified, and fertilized oocytes collected from the females in M2 medium and then treated with M2/hyaluronidase to remove cumulus cells. They were then transferred to KSOM medium (all medium from Sigma) and incubated at 37°with 5% CO 2 in a humidified incubator. By 10 A.M. when pronuclear membranes were visible, they were injected with a solution (2-3 ng/ml) of the BAC DNA described above. Later in the day zygotes were surgically transferred to the oviducts of female rats. These recipients and vasectomized males were housed in a 3 A.M. to 3 P.M. light cycle. The females' ovulation synchronized as described above with 80 mg LHRH on day 1 at 10 A.M. and mated with vasectomized SD males on day 5. These recipient females were monitored daily after surgery. Pups were weaned at p21 and ear punches were used to identify and provide biopsies for genotyping as described below. Animals found to be positive for the transgene were then identified with ear tags. Animals were bred as Cre-positive carriers by wild-type Long-Evans from Charles River Laboratories.

Genotyping and copy number analysis
Genomic DNA was isolated from tissue biopsies or primary cells using a Macherey-Nagel Tissue Spin kit. Two genotyping protocols were used to identify animals carrying the Pvalb-iCre transgene. The 59 junction was amplified with Pvalb F16505 (59-CAGAACTCTCCAGTGTCT GCTGG-39) and iCre R173 (59-CTTCCAGGTGTGTT CAGAGAAG-39) to produce a 476-bp amplicon. The 39 junction was amplified with iCre F738 (59-GTTCTGCCG GGTCAGAAAGAATGGT-39) and Pvalb R16998 (59-CCAA CCCCGAGGATAAGGGAATG-39) to produce a 713-bp amplicon. All PCRs used OneTaq (New England Biolabs), and 40 cycles of 94°C for 30 s and 68°C for 1 min. Amplification products were analyzed by agarose gel electrophoresis. For copy number determination, the genomic DNA was isolated as described above served as template for droplet digital PCR using ddPCR Supermix for Probes (No dUTP; Bio-Rad #1863024) and FAM-labeled or HEX-labeled probe assays for reference gene Ggt1 (Witten et al., 2011) and iCre (Table 1). Droplets were generated and analyzed using a QX200 AutoDG Droplet Digital PCR System (Bio-Rad). Amplification reactions were run in a Bio-Rad T100 thermocycler.

Histologic analysis of brains
Cre-dependent AAV reporters for initial phenotyping of putative lines Parvalbumin-iCre rats used in the present experiments were injected with Cre-dependent fluorescent reporters  to identify Cre recombination present in the OFC and other brain regions. For example, an AAV vector expressing a nuclear-localized, color-changing reporter, AAV Nuc-flox-(mCherry)-eGFP (Bäck et al., 2019) was injected into OFC and four weeks later the brains were processed for fluorescent protein expression. To fix brain tissue, animals were anesthetized using isoflurane, then transcardially perfused with heparinized PBS, followed by 4% paraformaldehyde (PFA), and then postfixed for 2 h in 4% PFA. Extracted brains were immersed in succeeding gradients of 18% to 30% sucrose for cryopreservation before being flash frozen. Brains were cryosectioned at 30 mm and washed for 30 min in PBS, followed by a 1-h incubation in a 4% goat serum 0.3% Triton X-100 blocking solution. After 30 min of washing, sections were incubated in primary antibody in blocking solution (mouse anti-parvalbumin, 1:1000, Sigma, P3088) overnight, shaking at 4°C. Sections were then washed in 3Â in PBS, incubated for 1 h in a secondary antibody blocking solution (goat antimouse, 1:500, Alexa Fluor 680, Invitrogen, A21058) and washed again in PBS. Sections were then stained with DAPI and imaged at 20Â magnification using laser confocal microscopy (Nikon Eclipse E800 microscope body, Nikon C2 Confocal microscope system, NIS-Elements imaging software version 4.40; Nikon Instruments Inc.). Cell counting and quantification of parvalbumin and redto-green signal colocalization was performed in six rats (three males, three females). Six representative brain sections from each animal were used for quantification (ImageJ 1.53C). For single channel quantification, automatic cell counting of particles larger than 3 mm in diameter was applied after adjusting for the background. For colocalization, manual counts were performed. Data are presented as mean % (6SEM) of red or green cells colocalized with parvalbumin staining.

Subjects
All experimental procedures were approved by the local animal care and use committee and housed in facilities certified by the American Association for the Accreditation of Laboratory Animal Care (AAALAC). Male and Female Pvalb-iCre rats (generated as described above), 140-300 d of age at the time of electrophysiological recording, were housed two to three per cage for at least one week before experimental procedures. They were maintained in a temperature-controlled and humidity-controlled environment under a reverse 12/12 h light/dark cycle (lights on at 9 P.M., lights off at 9 A.M.) with food and water available ad libitum. All behavioral experiments were run during the dark phase of the light cycle and begun each day between 10 and 12 A.M. For animals in CSA groups, one week following intravenous catheterization surgery food was restricted to 15 g of standard rodent diet per day and made available after the daily CSA sessions. Animals were food restricted until the end of self-administration training, after which ad libitum feeding resumed. All behavioral experiments were conducted during the dark cycle.

Intracranial surgery
Rats were anesthetized with isoflurane (3% concentration at 1.5 l of 100% O 2 /min for induction, 2.5% at 1.5 l/ min for maintenance) which was applied with a Parkland Scientific V300PK Anesthesia Machine and maintained using a Somni EPS-3 Exposure Prevention System (Somni Scientific). Small holes were drilled into the skull and a stainless-steel cannula was positioned using the coordinates; 14.0 mm anterior to bregma, 2.0 mm lateral to the midline skull suture, and ventral À3.3 mm from the brain surface using a stereotaxic apparatus (Stoelting). Bilateral Injections of the ChR2 viral construct (AAV5-EF1a-DIO-hChR2(H134R)-EYFP, titer 1.8 Â 10 13 , Addgene), at a volume of 700 nl, were performed using a Hamilton 1-ml syringe at a rate of 150 nl/min. Following surgery, rats were given one injection of 4-hydroxy-2-methyl-N-(5-methyl-2-thiazolyl)-2H-1,2-benzothiazine-3-216 carboxamide-1,1dioxide (Meloxicam, 5mg/mL, Boehringer/Ingelheim, Ingelheim, FRG) for post-surgical analgesia, and then single housed with ad libitum access to food. The animals were then returned to their home cages and at least two to three weeks were permitted for full expression of the viral construct.

Intravenous self-administration (IVSA) surgery
Rats were anesthetized with equithesin (1% pentobarbital, 2% magnesium sulfate, 4% chloral hydrate, 42% propylene glycol, 11% ethanol, 3 ml/kg, i.p.), and a SILASTIC catheter (inner diameter, 0.020 inch; outer diameter, 0.037 inch; Dow Corning) was advanced 3.5 cm into the right jugular vein and secured with silk suture (5-0). Anesthetic depth was assessed continuously throughout the procedure. The catheter terminated in a modified L-shaped 22-G guide cannula (Plastics One) mounted on the skull with cranioplastic cement and secured with three stainless steel screws. Animals were given one injection of meloxicam; for postsurgical analgesia, and were single housed for the remainder of the experiment. Catheters were flushed daily with 0.1 ml of heparinized saline, to maintain patency, and with cefazolin to forestall infection, until the end of operant training. Animals recovered for at least one week after the surgery, and before food restriction commenced. Postsurgery, all animals were monitored for 3 d for signs of adynamic ileus; no such signs were observed in any subjects. Operant training started 2-3 d after the beginning of food restriction.

CSA training
Self-administration training took place in standard rat operant chambers equipped with two retractable levers (Med Associates). Each chamber was located in a sound attenuating enclosure, equipped with a ventilation fan that provided background noise. Catheters were connected to a length of metal coil-protected plastic tubing (Plastics One) via a liquid swivel (Instech) to a syringe mounted in an infusion pump (Med Associates) located outside the sound attenuating enclosure. Animals were trained to press one lever (active lever) for intravenous cocaine infusions under a fixed ratio 1 (FR1) schedule. At the beginning of the session, the two levers were inserted into the chamber and a house light located on the wall opposite the levers was illuminated. Each lever press resulted in infusion of 0.75 mg/kg cocaine (2.25 mg/ml in sterile saline) accompanied by retraction of both levers, extinction of the house light and illumination of a cue light above the active lever for 30 s. Responses on the inactive lever had no consequence. To facilitate acquisition of the response, animals were allowed to self-administer during 6-h sessions until they received 40 infusions within a session. Most of the animals acquired self-administration within one to two sessions. After that, sessions were limited to 3 h or 75 infusions, whichever occurred first. CSA under the FR1 schedule continued for a total of 12 daily sessions. Following behavioral training, rats in the CSA group remained single housed in their home cages, with ad libitum food and water available, until electrophysiology experiments.

Brain slice preparation
Animals were anesthetized with isoflurane and decapitated using a guillotine. The brains were then extracted and transferred to ice-cold NMDG cutting solution (93 mM NMDG, 2.5 mM KCl, 1.2 mM NaH 2 PO 4 , 30 mM NaHCO 3 , 20 mM HEPES, 25 mM glucose, 5 mM ascorbic acid, 3 mM sodium pyruvate, 10 mM MgCl 2 , and 0.5 mM CaCl 2 ). The tissue was then cut perpendicular to its longitudinal axis at ;1.4 mm posterior to bregma using a razor blade, and then the sectioned surface was glued to the stage of a vibrating tissue slicer (Leica VT1200S, Leica Biosystems). Six to seven coronal sections (250 mm) were collected from each brain beginning at the appearance of the dorsolateral orbital cortex and the rhinal fissure (4.7 mm anterior to bregma) and ending at ;3.2 mm anterior to bregma. For each animal, recordings were restricted to brain slices corresponding to these landmarks. The slices were then hemisected and transferred to heated (34°C) NMDG for 5 min. The brain slices were then transferred to an oxygenated (95% O 2 /5% CO 2 ) holding chamber filled with HEPES-containing artificial CSF (aCSF; 109 mM NaCl, 4.5 mM KCl, 1.2 mM NaH 2 PO 4 , 35 mM NaHCO 3 , 20 mM HEPES, 11 mM glucose, 0.4 mM ascorbic acid, 1 mM MgCl 2 , and 2.5 mM CaCl 2 ) at room temperature (23°C) for at least 30 min, and up to 7 h.

In vitro electrophysiology
Hemisectioned slices were transferred to a chamber (RC-26; Warner Instruments) mounted to a fixed stage on a vibration isolation table (TMC Vibration Control). Slices were continuously perfused (3 ml/min) with oxygenated aCSF (126 mM NaCl, 3 mM KCl, 1.2 mM NaH 2 PO 4 , 26 mM NaHCO 3 , 11 mM glucose, 1.5 mM MgCl 2 , and 2.4 mM CaCl 2 using a peristaltic pump; Cole-Parmer), and warmed to 30-32°C using an in-line solution heater (Warner Instruments). The identification of OFC PV neurons in layer V of the ventrolateral OFC was achieved through excitation of the virally expressed enhanced yellow fluorescent protein (eYFP), using light from a mercury lamp (X-Cite 120Q, Excelitas) directed through the objective of an upright microscope (BX51WI, Olympus) mounted on a motorized moveable stage. The microscope was also equipped with differential interference contrast (DIC) optics, and a 900-nm infrared light source to identify living neurons in brain slices containing the OFC.
Whole-cell recording electrodes were fabricated using borosilicate pipette glass (Sutter Instruments, 1.5 mm O. D. Â 0.86 mm I.D.) and a horizontal puller (P-97; Sutter Instruments). They were filled with a potassium-based internal solution (140 mM K-gluconate, 5 mM KCl, 10 mM HEPES, 0.2 mM EGTA, 2 mM MgCl 2 , 4 mM Mg-ATP, 0.3 mM Na 2 -GTP, and 10 mM Na 2 -phosphocreatine), neutralized to a pH of 7.2 using potassium hydroxide. Electrode resistances were 3-7MV when filled with this solution. Wholecell patch clamp recordings were performed using an Molecular Devices 700B Multi-Clamp amplifier (Molecular Devices). Depending on the experiment, cells were voltage clamped between À50 and À65 mV. Spontaneous EPSCs (sEPSCs) were recorded in the presence of picrotoxin (50 mM) to block GABA A receptor currents and sampled at 10 kHz using WinLTP software (WinLTP Ltd) and an A/D board (National Instruments, PCI-6251) housed in a personal computer. Analyses of sEPSCs were performed offline and automated using the MiniAnalysis program (v 6.0.7, Synaptosoft). The sEPSC detection parameters were as follows: amplitude threshold, 5 pA; period to search for peak, 10,000 ms; necessary baseline before peak, 5000 ms; period for decay, 20,000 ms; fraction of peak to find decay, 0.37; period of averaged baseline, 1000 ms; area threshold trigger, 20 pA/ms. Peak detection valence was 'negative', and all detected events were visually inspected to verify analysis. An examination data from randomly sampled sEPSC recordings indicated that RMS noise ranged from 1.86 to 2.88 pA, with a mean (6SEM) of 2.49 6 0.88 pA (n = 9). This indicates adequate sensitivity to resolve sEPSCs above the 5-pA threshold. This same sample of cells showed a sEPSC mean 50% rise time of 0.33 6 0.12 ms, and a 37% decay time of 0.97 6 0.34 ms. Hyperpolarizing voltage steps (À10 mV) were delivered via the recording electrode every 30s to monitor whole-cell access and series resistance. Cells with an access resistance change of .30% were excluded from analysis. Current clamp experiments were performed with depolarizing currents ranging from À30 to 600 pA. ChR2 stimulation was achieved with blue laser light (473 nm; 50 mW laser, OEM) collimated through the objective using an IS-OGP adapter (Siskyou Corp.). Serotonin (5-HT, 1 mM; Tocris), was prepared fresh each day at low pH (;2) to prevent oxidation and diluted 1:50 in aCSF flowing into the slice chamber, to achieve a final bath concentration of 20 mM using a syringe pump (Model A-99, Razel Scientific Instruments). All other drugs were dissolved at their final concentration in aCSF and delivered using the peristaltic pump. Light-evoked (optical) IPSCs (oIPSCs) were recorded in the presence of 6,7-dinitroquinoxaline-2,3-dione (DNQX; 20 mM) to block glutamatergic synaptic currents.

Statistical analyses
Male and female Pvalb-iCre rats were randomly assigned to naive or CSA groups after intra-OFC injection of AAV5-EF1a-DIO-hChR2(H134R)-EYFP and recovery from this surgery. Formal experimenter blinding procedures were not used. However, complete analysis of the data and statistical comparisons from naive and CSA groups were not performed until data were collected from all subjects. Two rats were excluded from the studies, one because of the loss of intravenous catheter patency, the other died from an unknown cause during training. All experiments were designed using estimates of effect size and dispersion obtained from experience with similar OFC brain slice preparations. These estimates were then entered into power analysis calculations using G-Power (version 3.1.9.7, University of Dusseldorf, Germany) to ensure sample sizes were within reasonable estimates. In instances where experimental variability was large, additional experiments were performed. As stated above, recording from neurons exhibiting a change in access resistance (monitored continuously throughout the experiment) .30% were ceased and the data discarded. The proportion of cells discarded for this reason is estimated to be not .15%. Graphing and statistical analyses were performed using GraphPad Prism (v9.2). Data are generally reported as mean 6 SEM or mean 6 95% confidence interval (CI), unless otherwise stated. Table  2 summarizes all statistical data, including tests of normality, significance levels, mean effect sizes, 95% CIs, and post hoc results (Tukey's post hoc multiple comparisons test) for each experiment. Statistical tests included ANOVA [one-way, two-way, with repeated measures (RM) where appropriate], and the Student's t test. The Fisher's exact test was used to determine proportional effect changes among neurons from naive and CSA rats. A minimum level of p , 0.05 was considered to be significant in all statistical tests. Throughout, the number of rats from which data were acquired is indicated by N, whereas the number of cells is indicated by n.

Generation of transgenic rats and their characterization
A BAC DNA containing the rat Pvalb coding region was successfully recombineered to contain iCre at the start codon of the Pvalb coding sequence (Fig. 1A). The resulting BAC was injected into rat embryos and resulted in three founder lines. One founder line did not propagate. Phenotyping animals from the other two lines was conducted by Cre-mediated recombination of an AAV vector expressing a nuclear-localized, color-changing reporter, AAV-Nuc-flox-(mCherry)-eGFP (Bäck et al., 2019). Line 2 exhibited Cre-recombination in the OFC (Fig. 1B-D). The number of mCherry-positive cells (no recombination) was clearly higher than the sparse number of eGFPpositive cells (Cre recombination). Quantification revealed the percentage of eGFP-positive cells coexpressing parvalbumin was variable but much higher than cells coexpressing mCherry and parvalbumin (Fig.  1D). The lack of parvalbumin immunoreactivity in some eGFP expressing cells may be because of fluctuations in parvalbumin expression, and therefore Cre, relative to when the injection occurred. The ability to colocalize parvalbumin signal with nuclear fluorescent signal may also be a limiting factor. Importantly, cells that showed no signs of recombination (mCherry positive cells) consistently showed minimal overlap with parvalbumin expression (Fig. 1B,D). Using AAV with Cre-dependent reporters in this manner, we followed several potential sublines over 3-4 generations showing Cre-recombination in areas where Pvalb is known to be expressed, such as prefrontal and somatosensory cortices, with a focus on the OFC. We also observed a subset of neurons in both somatosensory cortex and non-dopaminergic neurons in the midbrain that exhibited Cre activity. Moreover, another laboratory characterizing this same line of rats found that 98% of cells expressing a Cre-dependent eYFP in the amygdala also expressed parvalbumin immunoreactivity (unpublished observation; Gavan McNally, personal communication). Using another Cre-dependent, AAV vector expressing a translational fusion of the rhodopsin ion channel to eYFP (AAV-DIO-hChR2-eYFP), we also examined colocalization with endogenous Pvalb expression in the OFC (Fig. 1E-J). Cells with somal expression of ChR2-eYFP colocalized with Pvalb immunofluorescence. The cellular pattern of Pvalb-immunofluorescence was similar between ChR2-eYFP-positive cells at the injection sight and ChR2-eYFP-negative cells distal to the site of AAV-DIO-hChR2-eYFP injection indicating the transgene expression did not alter Pvalb expression. As with our injection of the AAV Nuc flox (mCherry) eGFP, we noted cells expressing hChR2-eYFP but not Pvalb, suggesting the presence of low level non-specific Cre expression or that the hChR2-eYFP expressing cells were no longer producing Pvalb at the time of tissue collection. Other brain regions were not extensively characterized in this line of transgenic rats, and any Cre-dependent transgene of interest should be empirically evaluated for a given brain area and testing  paradigm. Based on our phenotypic data and finding that only a single copy of the Cre transgene was present in line 2, we designated the line as LE-Tg(Pvalb-iCre)2Ottc [rat genome database (RGD) ID #10412329, RRRC line #773] and refer to this line as Pvalb-iCre for the remainder of the article.

Properties of transfected OFC PV neurons
Following delivery of AAV expressing Cre-dependent ChR2-eYFP, we confirmed eYFP fluorescence in OFC and used DIC microscopy to locate cells in spatial register with the fluorescent signal (Fig. 2). Patch clamp recordings in eYFP-positive neurons revealed a significantly  . Also shown is a biocytin-filled cell that also expresses eYFP (blue arrow) in a Pvalb-iCre rat, three weeks after receiving an OFC injection with the ChR2-eYFP viral construct. C, DIC video microscopic image showing an OFC neuron (yellow circle) in a living rat brain slice from a Pvalb-iCre rat. A patch electrode is shown to the right (v-shaped structure). D, The same neuron (white circle) as in C, under fluorescence illumination, and through the same objective in a Pvalb-iCre rat, three weeks after injection with the ChR2-eYFP virus (scale bar for C, D).
higher firing frequency in response to square-wave current injection, compared with eYFP-negative cells (significant interaction between eYFP expression and current injection, two-way RM-ANOVA; F (11,231) = 30.23, p , 0.0001; and significant main effects of firing frequency vs injected current (F (11,231) = 51.63, p , 0.0001) and firing rate as a function of YFP expression (F (1,21) = 17.61, p = 0.0004; Fig. 3A). The eYFP-positive neurons also exhibited more depolarized resting membrane potentials (RMPs) than eYFP-negative neurons (Fig. 3B), and injection of subthreshold depolarizing current revealed current oscillations in eYFP-positive cells (Fig. 3Ci), that were not seen in eYFP-negative cells (Fig. 3Cii). In addition, the current-voltage (I-V) relationships in eYFP-positive cells were linear (Fig. 3B, deviation from linearity, p = 0.26), unlike the profile seen in other interneuron subtypes. As the electrophysiological characteristics observed in eYFPpositive cells matched that of fast-spiking parvalbumin interneurons (Koós and Tepper, 1999;Weber and Andrade, 2010), we refer to these eYFP-positive cells in our Pvalb-iCre rats as OFC PV neurons.

ChR2-induced activity in OFC PV cells
Recorded OFC PV cells were stimulated using blue (473 nm) laser light across a range of power settings (200 ms in duration). Laser stimulation generated stable inward currents without decrement throughout the light pulse, at all power settings (Fig. 3C). In contrast, currents were not observed in eYFP-negative cells in response to laser stimulation (Fig. 3D, t (13) = 6.71, p , 0.0001). In a group of recordings from PyNs (eYFP-negative cells), we found that single 5 ms laser pulses reliably elicited IPSCs (oIPSCs; Fig. 4A). These oIPSCs had a short (2.78 6 1.05 ms) latency to 50% peak with a small SD (0.20 6 0.16 ms, mean 6 SD) or "jitter"; both of which suggest that these currents were monosynaptic (Fig. 4A). Pairedpulse laser stimulation at a 50-ms interpulse intervals revealed inhibition of the second oIPSC, relative to the first oIPSC (Fig. 4A), and these responses were completely blocked by the selective GABA A receptor antagonist, SR-95531 (gabazine, 10 mM; Fig. 4B, paired-Student's t test, t (13) = 6.49, p , 0.0001). Thus, ChR2 activation of OFC PV Figure 3. Properties of eYFP-positive (eYFP1) and eYFP-negative (eYFP-) neurons in the OFC of Pvalb-iCre rats, three weeks after injection of viral construct expressing eYFP and ChR2. A, left, Mean effect of depolarizing currents (80-300 pA, 500 ms), passed through whole-cell recording electrodes, on action potential firing frequency in eYFP1 and eYFP-cells. eYFP1 cells showed a significantly higher frequency of action potential discharge, compared with eYFP-cells (significant interaction between eYFP and current injection; F (11,231) = 30.23, two-way RM ANOVA). Inset, Representative examples from single cells responding to 1300-pA current injection. B, Mean I-V relationships in response to subthreshold current injections (-30 to 150 pA, 500 ms) in eYFP1 and eYFP-cells. Membrane potential measurements were made during the final 20 ms of the current step. i-iii show membrane voltage responses to current injection in representative eYFP1 and eYFP-neurons, as well as the current steps to evoke these responses (iii). C, Peak inward currents elicited by ChR2 activation with 473-nm laser light pulses of varying power in a single eYFP1 OFC neuron. The inset shows membrane currents in response to different intensities of 473-nm light (blue bar). The later current response is used to measure R in . The cell was voltage clamped at À65 mV. D, Mean 6 95% CI maximum inward currents indicating that lightevoked responses were observed in eYFP1 cells, and not in eYFP-cells (N, n = 4, 15; unpaired t test, t (22) = 5.804, p , 0.0001). In these experiments, GABAergic currents were blocked by picrotoxin (Ptx; 50 mM) to eliminate IPSCs arising from activation of nearby eYFP1 cells.
5-HT Effects in OFC PV cells. Bath application of 5-HT (20 mM) elicited inward currents at a holding potential of -65 mV in the majority of OFC PV cells examined (22 of 28 cells, 11 rats, 78.6%; Fig. 5A,B), with no change in input resistance (R in ; data not shown; one-sample t (21) = 1.25, p = 0.225). Moreover, the mean largest amplitudes of inward currents in response to 5-HT in OFC PV neurons from males and female rats were similar ( Fig. 5A-C, unpaired t test, t (80) = 0.38, p = 0.97). A small number of cells (4/28, 14.3%) from only male rats exhibited significant outward currents in response to 20 mM 5-HT (Fig. 5C, one-sample t test, t (40) = 7.94, p , 0.0001), and this was associated with a small decrease in R in (data not shown). Given the small proportion of cells showing outward currents, we did not further characterize this response. However, it has previously been shown that similar currents in OFC PyNs in male wild-type rats were mediated by 5HT 1A receptors (Wright et al., 2017).

5-HT does not alter glutamatergic transmission onto OFC PV cells
In addition to its postsynaptic effects, 5-HT increases glutamatergic transmission onto OFC PyNs, likely via depolarization of these cells and release of glutamate from PyN to PyN collateral synapses. We therefore examined properties of glutamate synaptic transmission onto OFC PV cells, and whether this was modulated by 5-HT, by recording sEPSCs in these neurons. An interaction between sex and treatment (5-HT effect) on sEPSC frequency was not significant using a two-way ANOVA (F (1,30) = 0.040, p = 0.842; Fig. 6A-D), nor was there a significant main effect of treatment with 5-HT on sEPSC frequency (F (1,30) = 0.200, p = 0.658). However, there was a significant main effect of sex on the frequency of sEPSCs (F (1,30) = 20.52, p , 0.0001). Post hoc analysis revealed that the frequency of sEPSCs were significantly higher in OFC PV cells from males in the baseline condition (p = 0.0113, Tukey's post hoc test) and that 5-HT had no effect of sEPSC frequency in these cells from males or females (p = 0.970, p = 0.997, respectively). In contrast to this baseline difference in sEPSC frequency between males and females, mean sEPSC amplitudes were not significantly different (Fig. 6E), nor did 5-HT alter this measure (Fig. 6E, two-way ANOVA, F (1,30) = 0.00164, p = 0.97). Thus, basal synaptic glutamate input is stronger in male rat OFC PV neurons, and this is not altered by 5-HT in males or females.

5-HT increases GABAergic transmission onto OFC PYRs
As shown above, 5-HT evoked inward depolarizing currents in the majority of GABAergic OFC PV neurons, and these cells are synaptically connected to OFC PyNs (Figs. 4,5). Therefore, we hypothesized that 5-HT would increase ChR2-evoked GABAergic oIPSCs recorded in OFC PyNs. Since parvalbumin interneurons can sustain high rates of firing, we activated OFC PV cells with 473-nm light stimulus trains (10 pulses at 10 or 20 Hz; laser power set to evoke oIPSCs at 75% of maximum) to activate oIPSCs in PyNs. 5-HT (20 mM) significantly increased the amplitude of oIPSCs between ;5% and 16% for the 10-Hz trains and 5% and 17% for the 20-Hz trains (Fig. 7A  application. There were no differences in the magnitude of the 5-HT-induced increase in oIPSC amplitude in OFC PV cells between male and female rats (data not shown), nor in the proportion of cells showing the 5-HT increase in oIPSCs. Together, these data suggest that 5-HT increases OFC PV neuron inhibition of OFC PyNs and that this does not differ between male and female rats.
Effects of CSA withdrawal on OFC PV neurons. Although small differences in the rates of lever pressing were noted between male and female Pvalb-iCre rats during earlier sessions of acquisition of CSA, this became similar and stable in both sexes by the end of training (Fig. 8A). Moreover, there was no difference in the total amount of cocaine intake between male and female rats at the end of this training (Fig. 8B, t (15) = 1.02, p . 0.05, unpaired t test).
OFC PV neuron membrane properties are altered after withdrawal from CSA Following 14 d of CSA training, rats were returned to their home cages for 55.7 6 5.7 d (mean 6 SEM) until electrophysiology experiments were conducted. Male and female rats were withdrawn from CSA for a similar number of days (45.8 6 4.6 and 64.6 6 9.4 d, respectively, t (17) = 1.733, p = 0.101, unpaired t test). RMP, holding current (I hold ; obtained at À65 mV) and R in , were measured in OFC PV cells from naive rats and those withdrawn from CSA. Cells from male, but not female animals, demonstrated significantly less negative (depolarized) RMPs after withdrawal from CSA (Fig. 8C, two-way ANOVA, treatment Â sex interaction, F (1,35) = 13.26, p = 0.0009; treatment main effect F (1,35) = 4.459, p = 0.0419; sex main effect, F (1,35) = 2.279, p = 0.1401) post hoc analysis (Tukey's) showed that there was no difference in OFC PV RMPs in between naive females and males (p = 0.520), that there was a significant effect of CSA on RMP in only males (p = 0.0015) demonstrated by a significant difference between females and males after CSA (p = 0.0014). Consistent with the change in OFC PV RMP values from male rats, the amount of current necessary to voltage clamp these cells at À65 mV (I hold ) was larger and more negative in OFC PV neurons from male rats after withdrawal from CSA (Fig. 8D, two-way ANOVA, treatment Â sex interaction, F (1,68) = 9.115, p = 0.0036; treatment main Post hoc analysis showed that the effect of CSA on I hold in males was significant (p , 0.0001). In contrast to the effects of CSA on RMP and I hold in OFC PV neurons from male rats, no changes in the relationship between injected current and firing frequency was observed in any of the groups after CSA (Fig. 8E,F, RM-ANOVA, F (7,21) = 1.288, p = 0.304). These data suggest that whereas OFC PV neurons from only male rats are more depolarized following CSA, this has little effect on neuronal output in response to depolarizing current injection.
Effects of 5-HT on OFC PV neurons after withdrawal from CSA. Withdrawal from cocaine did not affect the magnitude of mean change in I hold caused by 5-HT (20 mM) in OFC PV neurons from male or female Pvalb-iCre rats after withdrawal from CSA (Fig. 8G,H, unpaired t test, male t (38) = 0.623, p = 0.5373; female t (38) = 0.0742, p = 0.941). Furthermore, the proportion of cells showing inward currents with 5-HT was not changed after withdrawal from CSA (Fisher's exact test, p = 0.702).

Discussion
A growing body of evidence shows that 5-HT innervation of cortical areas is important for the expression of cognitive flexibility, as defined in tasks requiring a change in behavioral strategy to obtain reward. In marmosets, OFC 5-HT depletion impairs visual discrimination reversal (Clarke et al., 2004), and studies tapping similar cognitive domains in rodents show congruent results (Roberts, 2011;Izquierdo et al., 2012;West et al., 2013). More specifically, rodent studies show that 5-HT depletion produces deficits in behavioral tasks that require intact OFC 5-HT function to permit cognitive flexibility (Izquierdo et al., 2012;West et al., 2013;Roberts, 2011). Psychostimulants such as methamphetamine and cocaine also negatively affect OFC-dependent information processing, and in many cases, this is seen several weeks after cessation of drug exposure. Thus, long-lasting learning impairments in contingency reversal (Jentsch et al., 2002;Schoenbaum et al., 2004;Calu et al., 2007), reinforcer devaluation (Schoenbaum and Setlow, 2005), and in Pavlovian overexpectation tasks (Takahashi et al., 2013;Lucantonio et al., 2014) are reported. Moreover, the enduring nature of these deficits after withdrawal from selfadministration of psychostimulants in animal models suggests that they may be involved in the dysregulated behavior that is observed in addiction to these drugs.
Links between cocaine and the 5-HT system have been documented at several levels of analysis. Cocaine is a potent inhibitor of the serotonin transporter (SERT; Torres et al., 2003), where it might act to alter 5-HT receptor function to indirectly alter OFC-dependent behavior. In support of this, the deleterious effects of CSA on OFCdependent reversal learning are reduced in mice lacking the SERT (Nonkes et al., 2013), and some cocaine-induced behavioral changes are absent in mutant mice expressing SERT with reduced affinity for this drug (Simmler et., 2017). Based on these findings, it has been hypothesized that 5-HT signaling is involved in the contribution the OFC to cognitive flexibility, and that psychostimulant exposure may disrupt this 5-HT function in this capacity (Roberts, 2011;Wright et al., 2017;Hankosky et al., 2018). In this regard, our recent data also show profound changes to 5-HT signaling in OFC PyNs in rats withdrawn from CSA or yoked cocaine administration, and that this 5-HT hypofunction could be observed many weeks after withdrawal from cocaine (Wright et al., 2017). The changes included a decrease in 5-HT-induced inhibition of OFC PyNs via 5-HT 1A receptors, and a reduction in the ability of 5-HT 2A receptors to increase synaptic glutamate release. Although these data show 5-HT hypofunction in OFC PyNs following cocaine exposure, other potential sites of 5-HT interaction with the local OFC circuitry have not been studied. Thus, 5-HT 2 receptors are also highly expressed on cortical GABAergic interneurons (Liu et al., 2007), and analogous fast-spiking parvalbumin cells in the medial PFC are depolarized by 5-HT 2A receptors (Weber and Andrade, 2010;Athilingam et al., 2017). To determine whether CSA alters 5-HT signaling in OFC PV neurons, we developed a novel Pvalb-iCre transgenic rat line in which Cre recombinase expression is driven by the parvalbumin promoter. Cre-dependent expression of eYFP in the OFC of Pvalb-iCre rats was then used to study 5-HT signaling in OFC PV cells in both naive rats and in those withdrawn from CSA. Additionally, as clinical studies show clear sex-dependent differences in vulnerability to CUD, with females more vulnerable to all phases of drug use, and similar results observed in animal models of addiction (Carroll and Anker, 2010), we powered our study to distinguishing potential differences between males and female Pvalb-iCre rats with regard to effects of CSA and altered 5-HT signaling.  . OFC PV interneuron membrane properties and sensitivity to 5-HT following CSA in male and female Pvalb-iCre rats. A, Mean number of lever presses for cocaine in male and female Pvalb-iCre rats. B, Mean cocaine intake for all Pvalb-iCre rats across all self-administration trials. Male and female rats self-administered a similar amount of cocaine (unpaired t test, t (22) = 1.46, p = 0.158). Legend applies to C, D. C, Mean (695% CI) RMP of OFC PV neurons from naive male and female rats and those withdrawn from CSA. Cells from male rats were significantly depolarized (less negative RMP) following CSA withdrawal (two-way ANOVA, sex Â treatment interaction, F (1,35) = 13.26, p = 0.0009); male naive versus male CSA (**Tukey's post hoc test, p = 0.0015), whereas those from females were unchanged (ns, p = 0.70, Tukey's post hoc test). **p , 0.01, Tukey's post-hoc test. D, Mean (695% CI) I hold necessary to voltage clamp OFC PV cells at À65 mV. Consistent with the changes in RMP, cells from male rats required more I hold after CSA (two-way ANOVA, sex Â treatment interaction, F (1,68) = 9.115, p = 0.0036; male naive vs male CSA, p ,0.0001, Tukey's post hoc test), and cocaine exposure did not change I hold in OFC PV neurons from female rats (female naive vs female CSA, p = 0.9962, Tukey's post hoc test). **p , 0.01, Tukey's post-hoc test. E, F, Withdrawal from CSA did not alter the action potential discharge frequency of OFC PV cells caused by depolarizing current injection in either males or females (two-way ANOVA, F (7,21) = 1.288, p = 0.3047). G, H, Mean time course of the effects of 5-HT on I hold in OFC PV neurons from male and female naive and CSA rats. No changes in the sensitivity of OFC PV neurons to 5-HT were observed in cells from males (unpaired t test, t (38) = 0.6226, p = 0.5373) or females (t (38) = 0.0742, p = 0.9412) following withdrawal from CSA. ns, not significant. Consistent with prior studies of PFC parvalbumin interneurons (Weber and Andrade, 2010;Athilingam et al., 2017), we found that 5-HT application caused depolarizing inward currents in the majority of OFC PV cells in male and female Pvalb-iCre rats. However, this was blocked by the 5-HT 2 receptor antagonist, ketanserin, in only females. The reason for this sex difference in ketanserin sensitivity in OFC PV neurons is unknown. However, significantly greater binding of [ 3 H]-ketanserin to 5-HT 2 Rs in all layers of the female rat hippocampus has been reported (Zhang et al., 1999), and this may be relevant in the female OFC. Alternatively, outward currents were observed in a minority of male OFC PV neurons during 5-HT application in the present study (Fig. 5C) that were similar to outward currents mediated by 5HT 1A receptors in male rat PyNs (Wright et al., 2017). As this outward current may compete with the 5-HT-activated inward current, a partial blockade of inward current by ketanserin may be obscured.
Another possibility may arise from the observations that ketanserin has higher affinity for 5-HT 2A versus 5-HT 2C receptors, and a low affinity for 5-HT 2B receptors (Barnes and Sharp, 1999). Therefore, sex-linked differences in the expression of these 5-HT 2 R subtypes could result in insensitivity to ketanserin in the male OFC PV cells. Finally, differences in 5-HT 2A versus 5-HT 2C receptor expression on OFC PV neurons between male and female rats have also been reported after methamphetamine self-administration (Hankosky et al., 2018), suggesting some sexlinked lability in the expression of 5-HT 2 R subtypes in these neurons. In addition to these differences in cellular distribution of 5-HT receptor subtypes, sexually dimorphic effects of 5-HT 2C R and 5-HT 1A R activation on CSA have been reported in rhesus monkeys (Collins and France, 2018). Therefore, data from the present and prior studies support the idea that 5-HT 2 R receptor subtype distribution differs between male and female mammals. Figure 9. Properties of sEPSCs in OFC PV interneurons following CSA and lack of effects of 5-HT in Pvalb-iCre rats. A, Representative traces showing sEPSCs from male (blue traces) and female (orange traces) OFC PV cells from naive rats and those withdrawn from CSA. B, Mean (695% CI) sEPSC frequency in OFC PV cells from male and female naive rats and those withdrawn from CSA. There was a significant reduction in sEPSC frequency in male OFC PV cells following CSA (two-way ANOVA, sex Â treatment interaction, F (1,39) = 12.05, p , 0.0013, main effect sex, F (1,39) = 4.434, p = 0.0417, treatment main effect, F (1,39) = 2.057, p = 0.160; Tukey's post hoc test male naive vs male CSA, p = 0.006), and naive female rats exhibited a significantly lower frequency of sEPSCs compared with naive males (p = 0.005, Tukey's). **p , 0.01, Tukey's post-hoc test. Legend applies to C, D. C, Mean (695% CI) sEPSC amplitude in OFC PV cells from male and female naive rats and those withdrawn from CSA. No changes in sEPSC amplitude were observed (two-way ANOVA, sex Â treatment interaction, F (1,38) = 0.064, p = 0.802). D, Mean (695% CI) effect of 5-HT (20 mM) on sEPSC frequency in OFC PV cells from male and female naive Pvalb-iCre rats and those withdrawn from CSA. There was no effect of 5-HT on OFC PV cells from naive or CSA-withdrawn rats (two-way ANOVA, sex Â treatment interaction, F (1,38) = 0.582, p = 0.450). ns, not significant.
Sex-related differences were also noted in membrane and synaptic properties of OFC PV neurons from naive rats and from those withdrawn from CSA. Thus, OFC PV cells in naive female rats exhibited lower sEPSC baseline frequencies than males, suggesting that there is a lower probability of glutamate release onto these neurons in females. However, in contrast to the prominent increase in glutamate release by 5-HT reported in OFC PyNs (Wright et al., 2017), this was unaffected by 5-HT in OFC PV cells in naive rats and those withdrawn from CSA. Despite this, OFC PV cells in male Pvalb-iCre rats exhibited smaller holding currents, depolarized membrane potentials (Fig.  8C,D), and a reduction in sEPSCs following CSA (Fig. 9B). However, the inward currents caused by 5-HT were unchanged by CSA in OFC PV cells from males and females (Fig. 8G,H). Although the effects of 5-HT on OFC PV cells did not change after CSA withdrawal, the changes in OFC PV neuron properties that were observed in males are likely to have profound effects on cortical circuit function and its participation in behavior. These changes in male rats could be compensatory and lead to a greater resistance to cocaine addiction compared with females (Carroll and Anker, 2010), or perhaps to an increased vulnerability to relapse to cocaine seeking triggered by conditioned cues under certain conditions, as reported in male rats (Fuchs et al., 2005). Collectively, our data show that male OFC PV neurons are more sensitive to the effects of CSA itself, that 5-HT receptor subtypes may be differentially expressed on OFC PV neurons of males and females, and, unlike that reported in OFC PyNs (Wright et al., 2017), the effects of 5-HT on OFC PV cells are unaltered after CSA withdrawal. Therefore, these results suggest that withdrawal from cocaine exposure disrupts 5-HT signaling primarily at OFC PyNs, and that this may account for the dysregulated OFC-dependent behavioral and cognitive processes observed in CUD in humans and in animal models of cocaine addiction.