Optogenetic stimulation of corticostriatal circuits improves behavioral flexibility in mice with prenatal alcohol exposure

Abstract

Fetal alcohol spectrum disorder (FASD) is the most common preventable form of developmental and neurobehavioral disability. Animal models have demonstrated that even low to moderate prenatal alcohol exposure (PAE) is sufficient to impair behavioral flexibility in multiple domains. Previously, utilizing a moderate limited access drinking in the dark paradigm, we have shown that PAE 1) impairs touchscreen pairwise visual reversal in male adult offspring 2) leads to small but significant decreases in orbitofrontal (OFC) firing rates 3) significantly increases dorsal striatum (dS) activity and 4) aberrantly sustains OFC-dS synchrony across early reversal. In the current study, we examined whether optogenetic stimulation of OFC-dS projection neurons would be sufficient to rescue the behavioral inflexibility induced by PAE in male C57BL/6J mice. Following discrimination learning, we targeted OFC-dS projections using a retrograde adeno-associated virus (AAV) delivered to the dS which expressed channel rhodopsin (ChR2). During the first four sessions of reversal learning, we delivered high frequency optogenetic stimulation to the OFC via optic fibers immediately following correct choice responses. Our results show that optogenetic stimulation significantly reduced the number of sessions, incorrect responses, and correction errors required to move past the early perseverative phase for both PAE and control mice. In addition, OFC-dS stimulation during early reversal learning reduced the increased sessions, correct and incorrect responding seen in PAE mice during the later learning phase of reversal but did not significantly alter later performance in control ChR2 mice. Taken together these results suggest that stimulation of OFC-dS projections can improve early reversal learning in PAE and control mice, and these improvements can persist even into later stages of the task days later. These studies provide an important foundation for future clinical approaches to improve executive control in those with FASD.

Introduction

Fetal Alcohol Spectrum Disorder (FASD) represents a global health issue due to its persistently high prevalence. A recent estimation based on global epidemiological data reports that ~1 in 13 women consume alcohol during pregnancy with an estimated 630.000 infants born annually that may be diagnosed with FASD (Lange et al., 2017, Popova et al., 2023). FASD is associated with executive function difficulties, including problems with planning, cognitive flexibility, and cognitive control behavior (Mattson et al., 1999, Kodituwakku, 2009). In particular, behavioral inflexibility has been commonly described in individuals with FASD and has been seen in multiple rodent models of prenatal alcohol exposure (PAE) (Schneider et al., 2011, Chudasama, 2011). Across species, behavioral flexibility has commonly been measured via reversal learning tasks which require a subject to learn which of two response choices leads to a reward to a high-performance level, and then re-learn the associations when the reward contingencies are reversed (Izquierdo et al., 2017, Uddin, 2021).

The corticostriatal circuitry underlying initial discrimination learning and reversal is one of the most widely studied and highly consistent across species. Studies in rodents, non-human primates, and humans support the critical role of the orbitofrontal cortex (OFC) in updating outcome expectations when contingencies change (for review see: (Izquierdo et al., 2017)). In particular, rodent studies have consistently shown that loss of OFC function impairs the ability to change to a previously non-rewarded choice when the initial choice is still present (for review see: (Hamilton and Brigman, 2015)). In contrast, the dorsolateral striatum (dS) is involved in the formation of stimulus-outcome associations both in discrimination learning and during later reversal (Brigman et al., 2010, Graybeal et al., 2011).

Utilizing a touchscreen visual discrimination reversal paradigm, we have previously shown that moderate PAE (80-90 mg/dL) throughout gestation spares discrimination learning but significantly increases maladaptive perseveration, or repeated responding to a previously rewarded choice, during reversal learning in male and female mice (Marquardt et al., 2014). We later utilized dual-region in-vivo electrophysiology recordings of OFC and dS in male PAE mice to examine how these circuits were altered during behavior. While previous studies in untreated C57BL/6J males found that OFC pyramidal neurons tracked an initially rewarded choice through discrimination and into early reversal (Marquardt et al., 2019), the pattern was strikingly different in male PAE mice. During the period where PAE mice perseverated (repeated responses to a previously rewarded stimulus), we saw subtle decreases in OFC firing rates and a significant increase in the number of responsive neurons during choice behaviors. In contrast, dS medium spiny neurons in PAE mice had robustly increased firing rates, as well as increased responsive units across discrimination and reversal (Marquardt et al., 2020). PAE mice also displayed significantly decreased and delayed phase alignment to correct responses during early reversal in the OFC, and aberrantly prolonged synchrony between the OFC and dS (Marquardt et al., 2020). These data suggest that while in control animals, coordinated phasic activity of the OFC is essential to exert control over the dS in order to disengage a well-learned response, PAE decreases and delays this signaling in male mice, leading to increased perseveration. Studies investigating putative mechanisms of these alterations have found that PAE alters the expression and function of OFC GABAergic interneurons (Kenton et al., 2020), and decreased NMDA-evoked current density in OFC pyramidal neurons of male mice (Licheri et al., 2021). Together, these findings support the hypothesis that PAE alterations in excitatory/inhibitory balance may lead to altered timing of OFC single unit activity in male offspring. This, in turn disrupts the coordinated signaling to downstream striatal regions that control automatized behavior, and that restoration of this signaling has the potential to restore OFC-dS communication and restore impaired behavioral flexibility.

In the current study, we tested whether optogenetic stimulation of OFC-dS projection neurons immediately following correct choices in early reversal learning would be sufficient restore the impaired behavioral flexibility in adult PAE male mice. Utilizing a stimulation approach previously shown to restore OFC-dS signaling and reduce compulsive behaviors (Burguiere et al., 2013), we targeted OFC-dS projections using a retrograde adeno-associated virus (AAV) delivered to the dS, and then we delivered high-frequency optogenetic stimulation to OFC-dS projecting pyramidal neurons via optic fibers following correct choices on the first four sessions of the reversal task (Fig. 1A). Our results suggest that direct stimulation of OFC-dS projection neurons is sufficient to reduce perseveration both in controls and mice prenatally exposed to alcohol expressing ChR2 during stimulation days, and throughout the early phase of reversal. In addition, early stimulation significantly reduced PAE deficits seen during the late stage of the reversal task.

Figure 1. Touch-screen visual discrimination reversal learning paradigm.

(A) Experimental design showing PAE and SAC mice trained from Early (% correct = 50%) through Late Discrimination (criterion of > 85% correct responses). Mice then underwent surgery for injecting the retrograde AVV in dS (bregma: 0.75 mm) and implanting the optical fibers in OFC (bregma: +2.6 mm). The virus was allowed to express for 4 weeks prior to reminder sessions to test retention of discrimination learning. Finally, mice were tested on reversal learning while receiving high frequency stimulation for the first 4 sessions. Performance was analyzed by well-established stages: Early reversal characterized by perseveration where performance was <33% correct, Chance Reversal where performance was between 33-66% correct, and finally Late Reversal, where mice learned to consistently respond to the new stimulus above 85% correct. (B) Schematic diagram showing the flow of each testing trial and the delivery of optogenetic stimulation where applicable. Mice initiate via lever press (1) and then touch one of two stimuli (2) via nose poke. An incorrect choice (3a) leads to a 10-second house-light on timeout, followed by correction trials until a correct choice is made. A correct response (3b) leads to a reward, secondary reinforcers, and concomitant optical stimulation during the first 4 days of reversal.

Methods

All experimental procedures were performed in accordance with the National Institutes of Health Guide for Care and Use of Laboratory Animals and were approved by the University of New Mexico Health Sciences Center Institutional Animal Care and Use Committee.

In-vitro validation using optogenetic slice electrophysiology.

To validate the optogenetic tools, naïve male C57BL/6J mice (n=3, age ~ 3 months old) were injected with pAVV-CaMKIIa-hChR2 (H134R)-mCherry into the dS (AP: 0.75, ML: ±2.04, DV: −3.2). 4 weeks following injection, mice were anesthetized with ketamine (250 mg/kg intraperitonally) and perfused intracardially with cold cutting solution containing in (mM): 92 N-methyl-D-glucamine (NMDG), 2.5 KCl, 10 MgSO₄, 0.5 CaCl₂, 1.25 NaH₂PO₄, 30 NaHCO₃, 20 Hepes, 25 Glucose, 2 Thiourea, 3 Na-Pyruvate, 5 Na-ascorbate (pH to 7.3–7.4 with HCl and bubbled with 95% O₂ and 5% CO₂, adjusted osmolarity to 290-310 mOsm) (Ting et al., 2018). Brains were rapidly removed and transferred into an ice-cold NMDG-solution for at least 2 min. Coronal slices containing the OFC (AP: 2.6, ML: ±1.35) were prepared using a Leica VT1000 plus vibratome (Leica Microsystems, Bannockburn, IL) in NMDG-solution at 4 °C, then transferred to an incubation chamber held at 34 °C containing NMDG solution. The NaCl concentration of the NMDG solution was gradually increased to 52 mM by adding 2 M NaCl in a stepwise fashion (250 μl at 0 min, 250 μl at 5 min, 500 μl at 10 min, 1 ml at 15 min, and 2 ml at 20 min). Following, the slices were moved in a holding chamber (model BSC-PC, Warner Instruments, Hamden, CT) with artificial cerebral spinal fluid (aCSF) containing in (mM): 92 NaCl, 2.5 KCl, 1.25 NaH2PO4, 30 NaHCO3, 25 glucose, 20 HEPES, 2 Thiourea, 3 Na-Pyruvate, 5 Na-ascorbate, 2 CaCl₂, and 1 MgSO4 (pH to 7.3-7.4 with a few drops of 1 M NaOH and bubbled with 95% O2 and 5% CO2, (290-310 mOsm).

Slices were kept at room temperature in the holding solution for recovery for 1 hour prior to electrophysiological recordings. Slices were transferred to a recording chamber with slice support (Cat # RC-27L; Warner Instruments) to allow flow of solution above and below the slice. Whole-cell recordings were performed in presence of artificial cerebral spinal fluid (aCSF) containing in (mM): 125 NaCl, 2 KCl, 1.3 NaH₂PO₄, 26 NaHCO₃, 10 glucose, 2 CaCl₂, 1 MgSO₄, 5 Na-ascorbate (osmolarity 290-310 mOsm, PH 7.3-7.4 bubbled with 95% O₂/ 5% CO₂) delivered to the chamber at a flow rate of 2 ml/min using a peristaltic pump (Mater Flex, model 7518-10, Cole Parmer, Vernon Hills, IL). Slices were maintained at 34 °C using a dual automatic temperature controller (Model TC-344B) connected to an in-line solution heater (Model SH-27B) (Warner Instruments). Recording pipettes were filled with an internal solution composed of (mM): 142 K-methanesulfonate, 0.5 EGTA, 15 HEPES, 4 KCl, 2 MgATP, 0.3 NaGTP, 10 Phosphocreatine disodium salt, adjusted to 305 mOsm, pH 7.25 (adjusted with KOH). Optically evoked action potentials were generated in OFC pyramidal neurons expressing-ChR2 using a 473 nm laser (Ike-473-100-OP) connected to a power supply (IKE PS-300) (IkeCool Corporation, Anaheim, CA). Laser light was delivered using an IS-OGP optogenetics laser positioner (Siskyou, Grants Pass, OR) through 40x objective lens. The action potentials were generated using a laser pulse for 500 ms with 5-mW laser power. Recordings were conducted using an Axopatch 700B amplifier (Molecular Devices, Sunnyvale, CA), and digitized and filtered at 10 kHz and 2 kHz, respectively. Data were acquired and analysed using Clampex and Clampfit software (version 11; Molecular Devices, Sunnyvale, CA), respectively.

Prenatal Alcohol Exposure Paradigm

Control and PAE mice used in the present study were provided by the New Mexico Alcohol Research Center (NMARC) using a limited access drinking paradigm previously described (Brady et al., 2012, Licheri et al., 2021). Female C57BL/6J mice (Jackson Laboratory, Bar Harbor, ME) at postnatal day (PND) 60 were exposed to 0.066% (w/v) saccharin (SAC) (Sigma, product #S6047) or to Ethanol (EtOH) (KOPTEC, product #V1101) solution (5% w/v for 4 days, then 10% w/v) sweetened with 0.066% (w/v) saccharin for 4 hours per day (from 10:00 to 14:00 hours) during the dark cycle. After one week of acclimation to SAC or 10% EtOH solutions, individual females, and singly housed males (Jackson Laboratory, Bar Harbor, ME) were bred together for 2 hours immediately following the drinking period (from 14:00 to 16:00 hours) for 5 consecutive days. Females went on to alcohol exposure during the 5-day mating period and were weighed every 3 to 4 days to determine the pregnancy. SAC and EtOH concentrations were halved every 2 days starting the first day after birth. At ~PND 23 SAC and PAE offspring were weaned and housed in groups of 2 per cage at constant temperature and controlled humidity under a reverse 12 hr light/dark cycle (light off 08:00 hr). To control for litter effects, only 1-2 mouse per litter was used for the current experiments (n=5-6 per treatment/stimulation group). Considering previous data showing an altered OFC-dS signaling was performed in male mice, this study was conducted using male offspring {Marquardt, 2020 #91}. In the current investigation, the average alcohol consumption measured during the entire exposure was 5.37 ± 0.18 g/kg/4 hours and the blood ethanol concentration was 30.08 ± 3.19 mg/dl measured at the end of 4 hours using Analox Analysers (Analox Instruments, England).

Operant Apparatus

Operant tasks performed for assessing behavioral flexibility were conducted in custom made acrylic chambers (sizes 21.6 x 17.8 x 12.7 cm). Each chamber has a solid acrylic floor to improve ambulation and was housed within a sound- and light-attenuating box (Med Associates, St. Albans, VT). A pellet dispenser connected to a reward magazine, house light, tone generator, and ultra-sensitive lever are located at the end of the chamber. While, at the opposite end there is a touch-sensitive screen (Conclusive Solutions, Sawbridgeworth, U.K.) shielded with a black acrylic aperture plate allowing two 2 x 5 cm touch areas spaced by 0.5 cm and located at a height of 6.5 cm from the floor of the chamber. The presentation of stimuli in the response windows and touches is controlled and recorded by K-limbic Software Package (Conclusive Solutions, Sawbridgeworth, U.K.).

Pre-training

At 6-8 weeks of age SAC and PAE male mice were gradually weight reduced via food restriction to reach and maintain 85% of their free-feeding body weights. Mice were given access to the pellet reward (14 mg dustless pellets; BioServ, Frenchtown, NJ) for 3 consecutive days before moving on to pre-training. During the first stage of pre-training, mice were habituated to the operant chamber and retrieved 10 pellets within 30 minutes (min). Next, mice were trained to obtain a reward by pressing an ultra-sensitive lever, and criteria for moving to the next stage was complaining 30 reward trials in under 30 min. In the next stage, a press of the lever led to presentation of a white stimulus (variously shaped) in one of the 2 response windows, and a touch to the stimulus led to a reward delivery, cued by secondary reinforcers (1-sec pure tone and illumination of the magazine). Mice that initiated, touched, and collected 30 pellets within 30 min were moved to the final stage of pre-training. In this stage, touch of a blank window during the stimulus presentation led to 10-second (sec) houselight-on timeout. Following an error, mice were given a correction trial during which the same stimulus and left/right position were presented until a correct response was performed. Criteria for moving on to Discrimination was 70% first presentation correct responses within 30 minutes.

Discrimination Learning

Mice were trained on a pairwise visual discrimination where an initiation led to the presentation of two novel equi-luminescent visual stimuli, one per window. The rewarded stimulus (S+) (Fan or Marbles) was counterbalanced across experimental groups. Each testing session consisted of 30 first-presentation trials. Left versus right spatial presentation of the correct and incorrect stimuli was pseudorandomized for first-presentation trials within sessions, with a given configuration occurring less than four times consecutively to prevent side-bias. For each first-presentation trial, a touch of S+ was counted as a correct response and led to delivery of reward and secondary reinforcers. The next trial initiation led to a new first-presentation trial. A touch on the incorrect stimulus (S−) was counted as an incorrect response and resulted in a 10 sec houselights-on timeout. The next initiation led to a correction trial (errors) where the stimuli were presented in the same spatial orientation. A touch on S+ led to reward and the animal was able to initiate the next first-presentation trial while a touch on S− led to another correction trial (Fig. 1B). For each session, the following dependent variables were collected: first presentation correct trials, first presentation error trials, correction trials (excluding first presentation error), stimulus reaction time (time from initiation to screen touch) and reward latency (latency to magazine entry following a correct touch. When mice reached discrimination criterion (30 first presentation trials completed with an average of 85% correct responses excluding correction trials) within 30 minutes over two consecutive days, they underwent viral delivery and fiber placement.

Construction of optical fibers

Optical fibers were assembled inserting a 10 mm piece of standard hard cladding multimode fiber (high OH, ⌀ 800 μm Core, 0.39 NA, TS 2077040, Thorlabs; previously cut and stripped) into a multimode ceramic zirconia ferrule (ID 230 μm, Thorlabs). Fibers were epoxied into ferrules using 353 ND Epoxy 2 component Mix Ratio 10:1 (Part A and Part B, PFP Precision Fiber Products), then the convex end of the ferrules were polished on each grade of polishing papers (Thorlabs): diamond lapping (30 μm), silicon carbide (5 μm), oxide lapping (3 μm), aluminum oxide (1 μm) and calcined alumina (0.3 μm). Afterwards, the fibers were scored with diamond knife to 3.5 mm length. Prior to implantation, the light output of each fiber was verified using an optical power meter (PM100D, Thorlabs). To deliver an equal amount of light to each hemisphere, fibers were matched to each other so that each fiber output equal amounts of light.

Stereotaxic Adeno-associated Virus-Injection and optical fiber implantation

After completing Discrimination, mice were given access to 3 grams of food for at least 3 consecutive days before surgery. Mice were anesthetized with isoflurane and preoperative analgesic injection (0.05 mg/kg buprenorphine) was administered intraperitoneally. Mice were placed in a stereotaxic alignment system (Kopf Instruments, Tujunga, CA, US) and eyes were covered with ophthalmic ointment. Viral injections were performed using a 2 μl Hamilton Syringe (# 88400, 7002 KH 25/2.75”/3) delivering the virus at a rate of 0.1 μl/min using a syringe pump (GenieTouch-Kent Scientifica Corporation). 0.5 μl of pAVV-CaMKIIa-hChR2 (H134R)-mCherry (AVV Retrograde) or pAVV-CaMKIIa-MCherry (AVV retrograde; Addgene) was injected into both hemispheres targeting the dS (AP: 0.75, ML: ±2.04, DV: −3.2, Paxinos & Franklin, 2^nd edition, Fig. 2B lower). After the injection was completed, the injector was left in place for 5 minutes, then slowly retracted by 1.6 mm for 5 additional minutes to allow the diffusion of the virus in the tissue, and finally slowly retracted fully. Next, 230 μm diameter fiber optic with ceramic zirconia ferrule (Thorslabs) implants were bilateral implanted into the OFC (AP: 2.6, ML: ±1.35, DV: −2.3, Fig 2B upper) and fixed in place using dental acrylic.

Figure 2. Implantation of optical fibers and ex-vivo validation for optogenetic stimulation.

Schematic representation of the optical fiber position in each mouse used for the current investigation. The optical fibers were implanted bilaterally in the OFC during the surgery. Each color indicated the optical fibers bilaterally implanted in the same animal in each different experimental group (A). Representative coronal section of OFC stained with DAPI and mCherry showing the bilateral implantation of optical fibers. Scale bar 400 μm. Top picture Representative coronal section of dS stained with DAPI and mCherry showing the bilateral injection of retrograde AVV for targeting the OFC-dS projections. Scale bar 800 μm. Bottom picture (B). Representative trace of laser stimulation on OFC-pyramidal neuron expressing ChR2 showing the depolarization and a train of action potentials (473 nm for 500 ms). The blue line indicates the duration of laser stimulation (C). Representative trace showing the action potential train induced by 5 mW laser stimulation for 5 msec every 100 msec. (D).

Reversal learning task

After four weeks of viral expression, mice were given post-surgery discrimination reminder sessions to ensure retention of the previously learned response. The day after reminder criterion (85% correct responses) was attained, mice began tethered reminder sessions with optical patch-cables attached to acclimate to tethered behavioural testing. Reversal learning began on the session following tethered reminder criterion was reached. Optical fibers were connected bilaterally through optical patch cables (200/230 μm core diameter, 0.5 numerical aperture, length 0.5 m) (Plexon, Dallas, TX, USA) to a PlexBright Dual LED Commutator (Plexon, Dallas, TX, USA). Stimulation was generated using PlexBright 4 Channel Optogenetic Controller via the Radiant Software (Plexon, Dallas, TX, USA) connected to the LED commutator through a PlexBright Series-Y BNC Cable (Plexon, Dallas, TX, USA). During the first four sessions of Reversal learning, light stimulation (laser pulse 465 nm, laser power 10.65 mW, frequency 10 Hz, duration 5 ms.) was automatically delivered to all mice upon touch of the correct stimulus (Fig. 1B). Following the 4 stimulation sessions, mice were allowed to perform the Reversal task untethered until criterion (30 trials completed with an average of 85% correct responses, excluding correction trials, over two consecutive days within 30 min) was reached.

Perfusion, brain extraction and slices mounting

Following testing, mice were heavily anesthetized with ketamine (250 mg/kg intraperitonally) and perfused intracardially with with 4% w/v paraformaldehyde (PFA (J.T. Baker Chemical Co.), in phosphate buffered saline (PBS) (Thermo-Fischer Scientific, USA), pH 7.4. Brains were extracted and stored in PFA for 48 hours at 4 °C and then transferred to 1X PBS. Coronal slice containing OFC and dS were sliced at 50 μm tick using a vibratome (Pelco easiSlicer, Ted Pella, INC. Redding, CA, USA) and stored in 24-well plates containing 1X PBS. Following, the coronal sections were washed 3 times with 1X PBS and incubated 20 minutes with 4’,6-diamidino-2-phenylindole (DAPI, dilution 1:500, Thermo-Fischer Scientific, USA). After rinsing with 1X PBS 4 times, sections were mounted on Superfrost Plus microscope slide (VWR, Radnor, PA) using Fluoromount-G mounting media (Thermo Fisher Scientific, USA), covered with glass coverslips (VWR, Radnor, PA) and stored at 4 °C until imaging acquisition.

Imaging acquisition

Slides containing the coronal slices corresponding to the dS injection site and OFC fiber placement were scanned by an Axio Scan.Z1 slide scanner (Zeiss, Germany) using an objective Plan-Apochromat 20X/0.8 MZ and the following filter settings: DAPI-LED 385nm, exposure time 8 m; mCherry-LED 555 nm, exposure time 10.01 ms. Tracks of optical fibers were detected using QuPath’s cell Detection Algorithm (Version 0.4.3), and their respective locations for each animal were mapped onto appropriate schematic templates (Fig. 2A).

Statistical analysis

The following behavioral measurements were taken during reversal learning sessions: total number of sessions, number of first presentation correct responses made, first presentation errors (incorrect responses) number of correction errors (excluding first-presentation errors), choice reaction time and time to retrieve reward. Performance during the reminder sessions and during the first 4 days of stimulation were analysed separately. A stage-wise analysis was utilized that categorized each session based on individual animal’s performance. Reversal learning was analysed separately for three stages including early perseverative (performance <33% correct), chance (between 33 and 66% correct) and later learning (>66% correct) (Brigman et al., 2013, Hafez et al., 2022, Marquardt et al., 2020). Performance was statistically assessed via three-way and two-way analysis of variance (ANOVA) in Prism Version 10.0.3 (GraphPad Software, San Diego, CA). Interaction affects were further analysed by uncorrected Fisher’s LSD post hoc test. Following histological analysis 1 PAE Tag and 1 PAE ChR2 were excluded due to incorrect fiber placement.

Outlier Analysis

Given the inherent variability of individual animals across multiple phases of testing, outlier testing was performed via Rout Testing (1%) for correct, incorrect, and correction error responses for each performance stage, and outliers excluded from all measures for that stage. 1 SAC ChR2 outlier which was excluded from the early reversal stage, and 1 SAC Tag and 1 PAE ChR2 mouse were excluded from late reversal stage analysis.

Results

Ex-vivo laser stimulation induces depolarization in OFC pyramidal neurons labeled with AAV-Chr2

To validate the stimulation protocol, we performed a set of qualitative ex-vivo recordings in coronal slices (n=3) stimulating the OFC-dS projecting neurons with 5 mW laser stimulation for 500 msec (Fig. 2C). Furthermore, we mimicked the “in-vivo protocol” in ex-vivo whole-cell recordings, applying 5 mW laser stimulation for 5 msec every 100 msec (10 Hz). Both approaches revealed robust responses from OFC pyramidal neurons expressing ChR2. (Fig. 2D).

Bilateral viral injection does not affect discrimination retention

No significant differences were seen in discrimination learning between PAE and SAC mice. Both experimental groups required similar numbers of sessions to achieve the criterion (t₍₂₁₎=0.472, p=0.641), and made similar numbers of correct responses (t₍₂₁₎=1.104, p=0.281), incorrect responses (t₍₂₁₎=1.178, p=0.251) correction errors (t₍₂₁₎=1.802,p=0.085) and had similar stimulus reaction time (t₍₂₁₎=0.854,p=0.402) and reward latencies (t₍₂₁₎=0.444,p=0.661) (data not shown). Following surgical infusion of the virus (Tag and ChR2), PAE and SAC mice did not show significant differences in the number of retention sessions to re-attain discrimination criterion. The two-way ANOVA analysis found no main effects of injection and treatment, as well no injection x treatment interaction (injection effect: F_1,19=0.378, p=0.545; treatment effect: F_1,19=1.307, p=0.267; injection x treatment interaction: F_1,19=0.626, p=0.438) (Fig. 3A). Analysis of correct responses made across the reminder sessions did not show any significant differences (injection effect: F_1,19=0.413, p=0.527; treatment effect: F_1,19=1.421, p=0.248; injection x treatment interaction: F_1,19=0.785, p=0.386) (Fig 3B). Similarly, no main effect of injection, treatment or interaction were observed on the number of incorrect responses or on the number of correction errors respectively (incorrect responses, injection effect: F_1,19=0.015, p=0.901, treatment effect: F_1,19=0.137, p= 0.714, injection x treatment interaction: F_1,19=0.001, p 0.9701) (Fig. 3C). Correction errors: (injection effect: F_1,19=0.064, p=0.802; treatment effect: F_1,19=0.515, p=0.481; injection x treatment interaction: F_1,19=0.093, p=0.762) (Fig. 3D). AD)ysis revealed that the bilateral viral injections did not significantly alter reaction time to visual stimuli (injection effect: F_1,19=0.008, p=0.929; treatment effect: F_1,19=0.120, p= 0.732; injection x treatment effect: F_1,19 =0.0003, p=0.986) (Fig. 3E). In addition, we did not find any significant differences in the reward time (injection effect: F_1,19 =0.022, p=0.881; treatment effect: F_1,19 = 3.686, p=0.070; injection x treatment effect: F_1,19= 0.152, p=0.700) (Fig. 3F).

Figure 3. Reminder session of discrimination learning one month after bilateral virus injection.

Surgical delivery of virus and fiber implant did not significantly affect reminder performance as measured by number of sessions (A) or correct and incorrect responses made during reminder (B-C). There was no significant difference in the number of correction errors (D) in all experimental groups. The viral injection did not alter the reaction and reward time regardless of treatment and virus (E-F). Data are expressed as mean ± SEM. Black circles = Tag mice, White circles = ChR2 mice.

Acute optical stimulation increases correct responding in SAC and PAE mice

Analysis of stimulation sessions during the first 4 days of reversal revealed a significant main effect of stimulation (F_1,19=18.85, p=0.0004), day (F_3,57= 25.52, p<0.0001) and a day x stimulation interaction (F_3,57= 4.141, p=0.010) on the number of correct responses made. However, there was no significant main effect of treatment (F_1,19= 0.010, p=0.919), day x treatment, (F_3,57 =1.159, p=0.335), treatment x stimulation (F_{1, (F}2.475, p=0.132) or day x treatment x stimulation interaction (F_3,57=0.077, p=0.9720) (Fig. 4A). Statistical analysis of the incorrect responses made by mice across the 4 days of stimulation showed only a significant main effect of day (F_3,57=3.792, p=0.0304) with no significant main effects of treatment (F_1,19= 1.504, p=0.235) or stimulation (F_1,19=1.504, p=0.235) and no interactions (day x treatment: F_3,57=0.7522, p=0.525; day x stimulation : F_3,57 0.4077, p=0.748; treatment x stimulation: F_1,19=0.9351, p=0.345; day x treatment x stimulation: F_3,57=0.093, p=0.963) (Fig 4B).

Figure 4. Optogenetic stimulation increased correct responding in ChR2 mice regardless of treatment.

Both PAE and SAC mice expressing ChR2 significantly increased the number of correct responses across days compared to Tag control animals (A). All mice reduced incorrect responding across sessions but there was no effect of treatment or stimulation (B). Both PAE ChR2 and Tag initially made more correction errors compared to SAC groups, and all mice reduced perseveration across all 4 sessions of optogenetic stimulation (C). Stimulation decreased reaction time across sessions in ChR2, while Tag mice remained stable across sessions (D). Stimulation did not modulate the reward time in all experimental groups (E). **** p< 0.001; *** p< 0.01; * p< 0.05. Main effect of treatment during the day 1 of stimulation: & p=0.05. Data are expressed as mean ± SEM.

PAE mice regardless of stimulation group made significantly more correction trials the first day of reversal (main effects of treatment: F_1,19=6.004, p=0.024; Fig. 4C). However, stimulation led to a significant reduction in correction errors across days as shown by a significant effect of day (F_3,57=48.82, p<0.0001) and stimulation (F_1,19=0.020, p=0.887) with no interactions (day x treatment: F_3,57=2.731, p=0.0521; day x stimulation: F_3,57=0.205, p=0.892; treatment x stimulation: F_1,19=0.206, p=0.654; day x treatment x stimulation: F_3,57= 0.461, p= 0.710; Fig. 4C).

Analysis of reaction time (time from lever press initiation to screen touch) showed a significant day x stimulation interaction (F_3,57=3.274, p=0.027) as SAC ChR2 mice initially showed slower reaction times but significantly reduced them across days. No effect of day (F_1.554,29.53=2.214, p=0.136), treatment (F_1,19=0.072, p=0.790), stimulation (F_1,19=0.726, p=0.404) or other interactions (day x treatment: F_3,57=1.112, p=0.351; treatment x stimulation: F_1,19=0.978, p=0.335; day x treatment x stimulation: F_3,57=0.942, p=0.426; Fig. 4D). Reward time (time to collect reward) was also generally unaffected during stimulation sessions, with no main effect of day (F_1.845,35.05=0.405, p=0.653), treatment (F_1,19=0.741, p=0.400) or stimulation (F_1,19=1.689, p=0.209). We found no significant day x stimulation interaction (F_3,57=1.384, p=0.257), and no other significant interactions (day x treatment: F_3,57=0.691, p=0.560; treatment x stimulation interaction: F_1,19=0.576 p=0.456; day x treatment x stimulation: F_3,57=1.801, p=0.157; Fig. 4E).

Optical stimulation reduced time spent in early perseverative reversal learning phase

Analysis of the number of early, perseverative reversal sessions where performance was <33% correct, revealed a significant main effect of stimulation (F_1,18=15.03, p=0.0011), but no main effect of treatment (F_1,18=0.0182, p=0.893) and no treatment x stimulation interaction (F_1,18=0.520, p=0.479) (Fig. 5A). Similarly, analysis of correct responses showed a significant main effect of stimulation (F_1,18=6.406, p=0.020) regardless of treatment (treatment effect: F_1,18=0.058, p=0.811) with no interaction (treatment x stimulation: F_1,18=0.097, p=0.758; Fig. 5B). Consistent with improved performance, SAC and PAE mice expressing ChR2 made less incorrect responses in this specific phase of reversal learning (treatment effect: F_1,18=1.302, p=0.268; stimulation effect: F_1,18=8.336, p=0.009; treatment x stimulation interaction: F_1,18=0.0358, p=0.851) (Fig. 5C). The statistical analysis of the correction errors also found a significant main effect of stimulation (F_1,18=9.987, p=0.005) regardless of treatment (F_1,18=2.388, p=0.139), and no treatment x stimulation interaction (F_1,18=2.293, p=0.147) (Fig. 5D). Optical stimulation (F_1,18= 0.849, p=0.368) nor treatment (F_1,18=0.408, p=0.530) significantly altered reaction time to stimulus choice during early reversal learning with no interaction (F_1,18= 1.091, p=0.310; Fig. 5E). There was also no main effect of treatment (F_1,18=1.515, p=0.234), stimulation (F_1,18=0.241, p=0.629) and no treatment x stimulation interaction (F_1,19= 1.017, p=0.326) on reward time measured in this stage of reversal (Fig. 5F).

Figure 5. Optogenetic stimulation improves the early phase of reversal learning in SAC and PAE mice.

Analysis of the early reversal learning (performance less than 33%) found a significant decrease in the number of sessions performed in SAC and PAE ChR2 vs. Tag (A). There was a significant effect of stimulation in the number of correct and incorrect responses made during this phase of reversal (B-C), as well as in the number of correction errors (D). Reaction and reward time are not affected by optogenetic stimulation (E-F). Data are expressed as mean ± SEM. ** p< 0.01; *p< 0.05. Black circles = Tag mice, White circles = ChR2 mice.

Optical stimulation does not alter chance reversal learning

Analysis of the mid-point of reversal where performance was from 33% to 66% correct, revealed that neither treatment nor stimulation had an effect on performance with no interactions. This was the case for number of sessions (treatment: F_1,19=0.527, p=0.476; stimulation: F_1,19=0.527, p=0.476; treatment x stimulation: F_1,19= 0.366, p=0.552; Fig. 6A), number of correct responses (treatment: F_1,19=0.152, p=0.700; stimulation: F_1,19=1.198, p=0.287; treatment x stimulation: F_1,19=0.211, p=0.651; Fig. 6B), incorrect responses (treatment: F_1,19=0.603, p=0.446; stimulation: F_1,19=1.506, p=0.234; treatment x stimulation: F_1,19=0.203, p=0.656; Fig. 6C) and correction errors (treatment: F_1,19=0.644, p=0.431; stimulation: F_1,19= 2.328, p=0.143; treatment x stimulation: F_1,19=0.003; p=0.956) (Fig. 6D). In addition, no main effects of treatment, stimulation or interaction were seen for reaction time to choice (treatment: F_1,19=0.524, p=0.477; stimulation: F₉=2.584, p=0.124; treatment x stimulation: F_1,19=1.255, p=0.276), or reward retrieval time (treatment: F_1,19=0.036, p=0.851; stimulation: F_1,19=2.928e-006, p=0.998; treatment x stimulation: F_1,19=0.197, p=0.661) (Fig. 6E-F).

Figure 6. Chance reversal learning was not affected by previous optogenetic stimulation.

SAC and PAE mice stimulated vs non-stimulated during the performance between 33-66% did not show any significant differences in the number of sessions (A), in the number of correct and incorrect responses (B-C), and in the number of correction errors (D). Reaction and reward time show similar values in all experimental groups (E-F). Data are expressed as mean ± SEM. Black circles = Tag mice, White circles = ChR2 mice.

Optical stimulation improves the late phase of reversal in PAE mice

Analysis of the late learning stage of reversal where performance was above 66% correct revealed a significant treatment x stimulation interaction in the number of sessions performed (treatment: F_1,17=1.484, p=0.239; stimulation: F_1,17=0.540, p= 0.472; treatment x stimulation: F_1,17=4.969, p=0.039; Fig. 7A). Post-hoc analysis found that PAE Tag mice performed more session than SAC Tag mice (Fisher’s LSD: p=0.03). Furthermore, PAE ChR2 mice required less sessions than PAE Tag mice (Fisher’s LSD: p=0.04). Similarly, there was a significant interaction effect on the number of correct responses made at this stage (treatment: F_1,17=1.523, p=0.234; stimulation: F_1,17=0.694, p=0.416; treatment x stimulation: F_1,17=5.035, p=0.0.38; Fig. 7B). Post-hoc analysis revealed significant difference between SAC Tag vs PAE Tag mice (p=0.02), and that PAE ChR2 mice required less correct responses than PAE Tag mice to reach criterion (p=0.03). However, analysis of incorrect responses found that while there was no main effect of treatment (F_1,17=2.207, p=0.155) or stimulation (F_1,17= 0.209, p=0.652), there was a significant treatment x stimulation interaction (F_1,17=6.928, p=0.017; Fig. 7C). Post-hoc analysis revealed that PAE Tag mice made significantly more incorrect responses than SAC Tag mice (Fisher’s LSD: p=0.01), Additionally, PAE ChR2 made significantly less incorrect responses than PAE Tag mice (Fisher’s LSD: p=0.03) suggesting PAE impaired late-stage reversal which was rescued by early stimulation. A similar pattern was seen for correction errors with non-significant treatment (F_1,17=1.377, p=0.256) and stimulation effect (F_1,17=0.013, p=0.908) but a significant interaction (F_1,17=7.146, p=0.016). Post-hoc analysis revealed that PAE Tag mice made significantly more correction errors than SAC Tag mice (Fisher’s LSD: p=0.01), while PAE ChR2 had a non-significant reduction in correction errors (Fisher’s LSD: p=0.07 Fig. 7D). Analysis of reaction time found no main effects of treatment, stimulation, and no interaction treatment effect: (treatment: F_1,17=0.691, p=0.417; stimulation: F_1,17=1.033, p=0.323; treatment x stimulation: F_1,17=0.268, p=0.611; Fig. 7E). Two-way ANOVA also did not find any significant main effect or interaction for time to retrieve reward (treatment: F_1,17=0.324, p=0.576; stimulation: F_1,17=0.209, p=0.652; treatment x stimulation: F_1,17=2.157, p=0.160; Fig. 7F).

Figure 7. Early optogenetic stimulation improved the performance of late reversal learning in PAE, but not SAC mice.

There was no significant main effect of stimulation or treatment but a significant treatment x stimulation interaction in the number of sessions, correct responses, incorrect responses and correction errors made during late reversal. Post Hoc analysis showed PAE Tag required more sessions (A), correct responses (B) and made more incorrect responses (C) than SAC Tag. PAE ChR2 reduced sessions, correct responses and incorrect responses to criterion versus PAE Tag mice. PAE Tag mice made significantly more correction errors than SAC Tag mice (D) but early stimulation did not reduce this significantly in PAE ChR2 mice. Reaction and reward time did not show any significant differences between all experimental groups (E-F). Data are expressed as mean ± SEM. * p< 0.05. Fisher’s Post hoc: # p<0.05 (SAC Tag vs PAE Tag); & p<0.05 (PAE Tag vs PAE ChR2). Black circles = Tag mice, White circles = ChR2 mice.

Discussion

It is well established that low to moderate PAE is sufficient to impair pairwise visual reversal in male adult offspring (Marquardt et al., 2014, Marquardt et al., 2020). In-vivo electrophysiology recordings show that impaired behavioral flexibility in PAE mice is accompanied by small but significant decreases in OFC firing rate following a correct choice and concomitant increases in dS activity, as measured by increased firing rates and total choice-responsive cells. In addition, OFC-dS synchrony in PAE mice is aberrantly sustained into reversal compared to the early disengagement and later reestablishment seen in control animals (Marquardt et al., 2020). Taken together, these findings suggest that maladaptive perseveration following PAE may be driven by loss of OFC coordinated activity and a subsequent decrease in signaling to downstream dS regions necessary to disengage the previously learned behavior (Marquardt et al., 2020). In the current study we expressed ChR2 in OFC-dS projection neurons and then tested whether light stimulation in the OFC following correct choices during early reversal would rescue the behavioral inflexibility induced by low to moderate PAE. Our results demonstrate that direct stimulation of OFC-dS projection neurons is sufficient to significantly increase correct responding in PAE and control ChR2 mice during acute stimulation sessions, and significantly reduce sessions, trials and errors required during the early phase of reversal. In addition, stimulation significantly reduced PAE increase in sessions, trials and incorrect responses seen during late reversal.

Anterograde tract tracings studies have demonstrated that the rodent OFC has direct connections both to ventral and dorsal aspects of the striatum (Schilman et al., 2008) and support the framework that OFC exerts top-down control to disengage striatal habit-systems when expectancy is violated and reward values need to be updated (Rudebeck and Murray, 2014, Rudebeck et al., 2013, Schilman et al., 2008, Schoenbaum and Esber, 2010, Schoenbaum et al., 2009, Schoenbaum et al., 2011a, Schoenbaum et al., 2011b). Previous findings in PAE mice are consistent with this framework, as developmental alcohol exposure is associated with both altered OFC function and aberrantly increased dS activity. Studies of other perseverative behaviors, i.e. compulsive grooming, have shown those behaviors are also associated with increases in striatal firing subsequent to loss of cortical inhibition (Burguiere et al., 2013). Those studies further showed that optogenetic stimulation of OFC pyramidal neurons was sufficient to reintroduce cortical coherence and reduce firing in striatal sub-regions involved both in individual medium spiny neurons (MSN) and at the population level. Importantly, direct stimulation of cortical firing and the subsequent decrease in striatal activity were sufficient to reduce compulsive behavior (Burguiere et al., 2013) and motivated the current approach.

Our current data show that stimulation of OFC-dS projection significantly increased the number of correct responses during the stimulation sessions in both PAE and SAC ChR2 mice. In line with our previous recording studies, both PAE ChR2 and PAE Tag showed early perseveration, making significantly more correction trials during the first day of reversal. As reported in those previous studies, tethering mice significantly slows choice response time in all mice as choice reaction time. During the early stage of reversal, reaction time in tethered animals is triple that seen in freely moving mice, and this increased response time is associated with an overall decrease in perseveration in tethered versus freely moving PAE (Chandrasekaran et al., 2023, Marquardt et al., 2020, Marquardt et al., 2014). PAE Tag mice also made significantly more sessions, correct responses, incorrect responses, and correction errors versus SAC Tag mice during the late stage of reversal (>66%) during which the new associations were being solidified. Importantly, both PAE ChR2 and SAC ChR2 mice showed a significant increase in the number of correct responses made during the four stimulation sessions compared to Tag mice. Despite the small number of correct responses and subsequent stimulations made during the first two stimulation sessions, both PAE and SAC ChR2 mice rapidly increased their correct responding across stimulation sessions and approached the chance learning stage significantly faster than Tag mice.

Importantly, stimulation during the first four sessions was sufficient to significantly reduce the duration of early perseverative sessions in both PAE ChR2 and SAC ChR2 mice. Stimulation during days 1-4 reduced the number of sessions, errors and correction errors mice required for mice to exit early perseverative reversal and re-attain chance performance in all ChR2 mice. We also found that ChR2 and Tag delivery alone did not differentially impact learning retention, as no significant differences were seen in re-attaining discrimination criterion following viral delivery and fiber implantation between groups. Additionally, optogenetic stimulation did not alter other task behaviors including reaction time to choose or retrieve reward following a correct choice, suggesting it is not simply trial slowing or alterations in motivation that drive the effects in ChR2 groups.

While optogenetic stimulation significantly improved early reversal performance in both PAE and SAC ChR2 mice, no significant differences were seen between treatments or stimulation groups during the chance phase (33–66%). These results are consistent with previous studies in PAE mice that exposed mice were specifically impaired in the early perseverative phase and did not differ on later learning. In contrast, OFC-dS stimulation did have significant effect on later performance within the learning stage (>66%). During this stage of reversal, PAE Tag mice showed significant increases in the number of sessions, correct and incorrect responses, and correction errors vs. SAC Tag controls. Early stimulation significantly reduced this increase in sessions, trials, and incorrect responding, and led to a non-significant reduction in correction errors, suggesting that early stimulation has long-lasting impacts on behavioral flexibility in PAE mice.

While the behavioral impact of PAE on behavioral flexibility and altered OFC-dS firing and oscillatory activity is clear, the mechanism by which developmental alcohol alters these circuits is not fully understood. It is well established that both hippocampal and cortical local field potential depend on GABAergic inputs (Stark et al., 2013, Varga et al., 2014), and that GABAergic interneurons play a role in coordinating OFC function specifically during reversal tasks (Bissonette et al., 2015). In studies focused on the effects of PAE on synaptic transmission, we have shown that the moderate PAE model used here increased the amplitude and area of GABA-A receptor-mediated sIPSCs in the OFC of male mice (Kenton et al., 2020). This model of PAE has also been shown to alter OFC GABAergic interneuron numbers in male offspring, with PAE leading to a significant alteration in the number of calretinin-expressing interneurons (Kenton et al., 2020). In addition, examination of excitatory signaling in the OFC found that while PAE females had significantly larger NMDA-eEPSC amplitudes than SAC controls, our PAE model led to a significant decrease in NMDA-eEPSC current density of pyramidal neurons (Licheri et al., 2021). Together these results suggest that dysregulation of GABAergic expression and function, at least in male offspring, leads to decreases in NMDAR function, which may in turn impede the ability to generate appropriate oscillatory signaling required for the OFC to exert top-down control over the striatum, leading to increased perseveration.

Regardless of how PAE is altering excitatory/inhibitory balance in the OFC, our results suggest that increased excitation via optogenetic stimulation is sufficient to improve reversal learning in PAE male mice (Marquardt et al., 2020). Although the relationship between individual unit firing and LFP is still not completely understood (Tiesinga and Sejnowski, 2010), studies in other systems have shown that direct excitation, particularly outside the expected firing intensity, can lead to shifts in phase and entrain local oscillations (Calderone et al., 2014, Haider et al., 2016, Lepage et al., 2011). While not directly tested in this study, given the direct OFC-dS targeting via retrograde AAV here, it is possible that restoring of OFC tone through the modulation of firing rate exhibits via dS modulation. While the decreased OFC activity and coherence in PAE mice suggest a mechanism by which stimulation improves early reversal, improvements in SAC ChR2 suggest that even in optimally functioning animals, increased OFC activity is sufficient to speed disengagement of dS habit systems and update reward outcomes. The role of the dS is in the action outcome learning and habit formation (Balleine and O'Doherty, 2010, Graybiel, 2008, Yin and Knowlton, 2006) is well established, and photosilencing of dorsolateral striatum neurons during the early reversal phase speeds new association learning, by allowing more medial dS to form new action-outcome associations (Bergstrom et al., 2020). As our approach primarily targets OFC-dlS projections, this suggests that early OFC stimulation is sufficient to disengage the dS and allow for new associations to be learned.

The changes in OFC-dS circuit dynamics from early optogenetic stimulation appear to persist beyond early reversal, given the improvement in PAE ChR2 performance in late reversal. While SAC mice benefit during early reversal with no change during chance or late reversal, the reduced number of sessions, trials, and incorrect responding in PAE ChR2 suggest that optogenetic stimulation across the first four sessions of reversal is sufficient to induce long-term plasticity in this circuit. Considering the fairly long-term changes in behavioral performance, it is possible that optogenetic stimulation across the first four sessions of reversal is sufficient to induce long-term potentiation (LTP). A previous report in rats has shown that repetitive, high-frequency stimulation induced long-term changes in synaptic strength and reduced alcohol seeking (Cheng et al., 2021). While not directly tested in the current study, it is possible that excitation of OFC-dS projections during the correct choice led to modulations in neuronal function through synaptic homeostasis as our stimulation approach led to changes up to a week post-delivery.

This hypothesis finds support from our previous whole-cell recording study performed in OFC showing a significant reduction in the strength of NMDA-receptor-mediated currents in PAE male mice (Licheri et al., 2021). The optogenetic activation of OFC pyramidal neurons during the first four sessions could induce pre- and postsynaptic changes, modulating the number and strength of synapses within and between brain areas with long-lasting effects, rescuing the excitatory and inhibitory balance. However, our paradigm, low frequency stimulation for 5 ms during the ~10 correct responses per session in early phase of reversal learning was much less robust than the high-frequency optical stimulation paired with optical postsynaptic depolarization repeated three times every 5 minutes used in rats (Cheng et al., 2021). Future studies are needed to investigate if our trial-based optogenetic stimulation paradigm significantly decreases the high dS firing rate observed in PAE mice and to test whether the behavioral rescue reports is accompanied by alterations in OFC glutamate and GABAergic transmission.

Conclusion

To our knowledge the present study is the first study to date showing that optogenetic stimulation of OFC-dS projections can improve early reversal learning in both control mice and PAE animals and improve late reversal learning in PAE mice. The current data add weight to the literature confirming the crucial role played by OFC in behavioral flexibility performance. These findings also provide a foundation for the development of novel and potential therapeutic strategy for FASD patients, suggesting that a behavioral therapy combined with transcranial stimulation might improve the deficits in executive control functions such as behavioral inflexibility.

Highlights.

• Prenatal alcohol exposure increases maladaptive perseverative behavior during the first day of reversal learning and increases incorrect responding during late-stage reversal.

• Optogenetic stimulation of OFC-dS projection neurons following correct responses during the early phase reversal learning improves behavioral performance in both control and prenatally alcohol exposed mice.

• Stimulation during early-reversal reduces the increased sessions, correct and incorrect responding seen during the late phase of reversal learning in prenatally alcohol exposed mice.

• These studies demonstrate that optogenetic stimulation of OFC-dS circuit increases behavioral flexibility in all mice and can reduce impairments in later reversal learning seen following prenatal alcohol exposure.