Introduction

Autism spectrum disorder (ASD) is a neurodevelopmental disorder characterized by alterations in social communication and interaction, as well as the presentation of restricted, repetitive behaviors (APA, 2022). Given that ASD has high heritability (Sandin et al., 2017), much work has been done in the last 30 years to identify genes that confer risk of developing ASD (De Rubeis et al., 2014; Iossifov et al., 2014; Sanders et al., 2015). Through this, hundreds of high-confidence risk genes have been identified, varying greatly in the proteins for which they code (Satterstrom et al., 2020). These include transcriptional and translational regulators, ion channels, receptors, cell adhesion molecules, and others (De Rubeis et al., 2014; Delorme et al., 2013; Ebert & Greenberg, 2013). Recent work has focused on identifying brain regions and circuits that may be commonly affected by ASD-related mutations. The basal ganglia, in particular the striatum, represents one such commonly altered brain region and prior studies have demonstrated changes in striatal circuit function and striatum-associated behaviors in mice with mutations in ASD risk genes (Benthall et al., 2021; Fuccillo, 2016; Peca et al., 2011; Peixoto et al., 2016; Platt et al., 2017; Rothwell et al., 2014; Wang et al., 2016). However, whether basal ganglia circuits are convergently changed in ASD mouse models is an open question. Here we investigated whether loss of function of the syndromic ASD risk gene Cntnap2 alters striatal physiology and basal ganglia-dependent behaviors in mice.

Cntnap2 codes for a neurexin-like cell adhesion molecule called Contactin-associated protein-like 2 (Caspr2) (Poliak et al., 1999; Poliak et al., 2003). In mice, Caspr2 is expressed in several cortical and subcortical regions, including the striatum, from embryonic day 14 (E14) onward into adulthood (Penagarikano et al., 2011). Caspr2 is primarily localized at the juxtaparanodes of myelinated axons and is involved in the clustering of potassium channels (Poliak et al., 2003; Scott et al., 2019). In mice, in vitro work suggests that Caspr2 may also play a role in AMPAR trafficking and cell morphology (Anderson et al., 2012; Gdalyahu et al., 2015; Varea et al., 2015), and ex vivo studies indicate that it can control cell excitability and circuit synchronicity (Martin-de-Saavedra et al., 2022). Caspr2 is important during neurodevelopment and has been implicated in neuronal migration (Penagarikano et al., 2011), the maturation and function of parvalbumin-positive GABAergic interneurons (Penagarikano et al., 2011; Scott et al., 2019; Vogt et al., 2018), and the timing of myelination (Scott et al., 2019). CNTNAP2 mutations in people lead to a neurodevelopmental syndrome that can include language disorders, epilepsy, obsessive compulsive disorder, and ASD (Penagarikano & Geschwind, 2012; Rodenas-Cuadrado et al., 2014). A mouse model of this syndrome, Cntnap2^-/-, has been shown to exhibit good face validity for ASD-relevant social and motor behavior alterations (Brunner et al., 2015; Dawes et al., 2018; Penagarikano et al., 2011; Scott et al., 2019).

However, the impact of Cntnap2 loss on striatal synaptic physiology and excitability and corticostriatal-dependent behaviors has not been comprehensively assessed.

The striatum is primarily composed of GABAergic striatal projection neurons (SPNs), which make up two functionally distinct output pathways: the D1-receptor expressing cells of the direct pathway (dSPNs), which project to substantia nigra pars reticulata (SNr), and the D2- receptor expressing cells of the indirect pathway (iSPNs), which project to external globus pallidus (GPe) (Calabresi et al., 2014; Gerfen & Surmeier, 2011; Kravitz et al., 2010; Tai et al., 2012). The two types of SPNs are intermixed throughout the striatum and receive excitatory glutamatergic inputs from cortex and thalamus, as well as dopaminergic input from the midbrain (Ding et al., 2008; Doig et al., 2010; Gerfen & Surmeier, 2011). Coordinated activity between the populations of SPNs in response to these inputs mediates action selection, motor learning and habit formation (Hawes et al., 2015; Santos et al., 2015; Yin et al., 2005, 2006). Although SPNs comprise upwards of 95% of the cells in striatum, there are a wide range of distinct interneurons, primarily GABAergic, that contribute significantly to the inhibitory circuitry of the striatum. Parvalbumin (PV) interneurons, which make up ∼2% of the cells in striatum, provide the largest feedforward inhibition onto SPNs (Burke et al., 2017). Changes in the number and/or function of PV interneurons have been identified in several ASD mouse models, including Cntnap2^-/-mice, implicating PV circuitry as a potential common alteration across ASD mouse models (Filice et al., 2020; Juarez & Martinez Cerdeno, 2022).

To determine how loss of Cntnap2 affects striatal function, we assessed the physiology of SPNs and striatal PV-interneurons in the dorsolateral striatum and utilized a range of assays to assess striatum-associated behaviors in Cntnap2^-/- mice. We find that SPNs of both the direct and indirect pathways exhibit increased corticostriatal drive, despite unchanged excitatory cortical input onto these cells. Although decreased inhibitory function has been identified in other brain regions in Cntnap2^-/-mice, we find no deficit in broad or PV-specific inhibitory input onto SPNs in the case of Cntnap2 loss. Instead, we identify a significant increase in the intrinsic excitability of SPNs in Cntnap2^-/- mice, in particular in dSPNs, which likely underlies the increased corticostriatal drive of these cells. Behaviorally, we find that Cntnap2^-/- mice exhibit RRB-like behaviors including increased grooming, marble burying and nose poking in the holeboard assay. Cntnap2^-/- mice also exhibit increased motor routine learning in the accelerating rotarod, and cognitive inflexibility in an odor-based reversal learning task. Taken together, these findings suggest that enhanced direct pathway drive may play a role in the spontaneous and learned repetitive behaviors exhibited by these mice.

Results

Cntnap2^-/- SPNs exhibit increased cortical drive

Emerging evidence suggests that corticostriatal synapses are a common site of alteration in mouse models of ASD (Li & Pozzo-Miller, 2020). To test whether mice with loss of Cntnap2 exhibit changes in corticostriatal connectivity, we crossed Cntnap2^-/- mice to Thy1- ChR2-YFP mice, which express channelrhodopsin in a subset of layer V pyramidal neurons (Fig. 1A) (Arenkiel et al., 2007; Poliak et al., 1999). These mice were crossed to a D1-tdTomato reporter line to visually identify dSPNs (Ade et al., 2011). We recorded from SPNs in the dorsolateral striatum (DLS), as this sensorimotor striatal subregion is implicated in the acquisition of habitual and procedural behaviors (Packard & Knowlton, 2002). Changes in physiological function in this area may be connected to the acquisition of repetitive motor behaviors in ASD (APA, 2022; Fuccillo, 2016). To simulate a train of cortical inputs, we applied ten pulses of blue light over the recording site in DLS and measured the number of action potentials (APs) fired by SPNs in the absence of synaptic blockers (Fig. 1A). We altered the light intensity to vary the probability of eliciting subthreshold depolarizations or AP firing. dSPNs were identified using tdTomato fluorescence, and tdTomato negative neurons were designated putative iSPNs.

Figure 1.

We quantified the number of evoked APs at different light intensities and found that both dSPNs and iSPNs in young adult Cntnap2^-/- mice exhibited significantly increased spike probability compared to wild-type (WT) SPNs (Fig. 1B-E). This suggests increased corticostriatal drive, consistent with findings in another mouse model with loss of function of the ASD-risk gene Tsc1 (Benthall et al., 2021). To test whether the enhanced spiking probability was due to a change in the synaptic excitability of Cntnap2^-/-SPNs, we applied blue light of varying intensity over the recording site in DLS to evoke AMPAR-driven excitatory postsynaptic currents (EPSCs). We found that the average optically-evoked EPSC amplitude was not significantly different across a range of light intensities in Cntnap2^-/-dSPNs or iSPNs compared to WT (Fig. 1F-I).

To further assess synaptic inputs, we measured the number of dendritic spines in Cntnap2^-/- and WT SPNs, which are typically the sites of cortical synaptic innervation (Bouyer et al., 1984; Xu et al., 1989). To visualize spines, we injected neonate Cntnap2;Drd1a-tdTomato mice with AAV5-Syn1-GFP virus to sparsely label dSPNs and iSPNs in the DLS (Fig. S1) (Keaveney et al., 2018). We found that Cntnap2^-/- SPNs in adult mice had similar spine density as WT (Fig. S1), suggesting no overall change in synapse number. Together, these results show that dSPNs and iSPNs in Cntnap2^-/- mice exhibit enhanced cortically-driven spiking.

However, this is not due to a change in corticostriatal synaptic strength or overall synapse density.

Increased corticostriatal drive in Cntnap2^-/- mice is not due to reduced inhibition Previous work has indicated a reduction in the number and/or function of fast-spiking parvalbumin-expressing (PV) interneurons across multiple brain regions in Cntnap2^-/-mice (Ahmed et al., 2023; Antoine et al., 2019; Jurgensen & Castillo, 2015; Paterno et al., 2021; Penagarikano et al., 2011; Vogt et al., 2018). While inhibitory deficits have been identified in the cortex and hippocampus (Antoine et al., 2019; Jurgensen & Castillo, 2015), and the number of PV interneurons has been reported to be decreased in striatum (Penagarikano et al., 2011), a comprehensive assessment of inhibitory synaptic function has yet to be completed in the striatum of Cntnap2^-/- mice. We first determined whether there were global deficits in inhibition onto SPNs in Cntnap2^-/- mice using intrastriatal electrical stimulation to evoke inhibitory postsynaptic currents (IPSCs) (Fig. 2A). In Cntnap2^-/- dSPNs, the average electrically-evoked IPSCs across a range of stimulation intensities were not different from WT dSPNs (Fig. 2B,C). However, in Cntnap2^-/- iSPNs, IPSCs were, on average, slightly larger than those in WT iSPNs, although we note that the responses were variable across cells (Fig. 2D,E).

Figure 2.

There are many sources of inhibition in the striatum (Burke et al., 2017), which can all be activated with electrical stimulation. To assess whether inhibition from PV interneurons specifically is altered in Cntnap2^-/- mice, we crossed Cntnap2^-/-;D1-tdTomato mice to Pvalb-Cre; RCL-ChR2-H134R-EYFP (Ai32) mice to express channelrhodopsin in PV interneurons (Fig. 2F) (Hippenmeyer et al., 2005; Madisen et al., 2010). We applied a blue light pulse of varying intensity over the recording site to evoke PV interneuron-specific IPSCs in SPNs, in the presence of excitatory synaptic blockers (Fig. 2F). We found that the average optically-driven IPSCs did not differ significantly in amplitude in Cntnap2^-/- dSPNs or iSPNs compared to WT controls (Fig. 2G-J).

To directly measure PV neuron function, we assessed the intrinsic excitability of PV interneurons in Cntnap2^-/- mice. To visualize PV interneurons for recordings, we crossed Cntnap2^-/- mice to Pvalb-Cre;RCL-tdT (Ai9) mice (Fig. S2A). Plotting the number of APs fired as a function of current step size indicated that there were no significant differences in the intrinsic excitability of PV interneurons in Cntnap2^-/- mice compared to controls (Fig. S2B,C). There were also no changes in intrinsic cell properties such as rheobase, membrane resistance, capacitance, resting membrane potential, or AP shape in PV interneurons in Cntnap2^-/- mice (Fig. S2D-K).

Given prior reports of altered PV cell number in Cntnap2^-/-mice (Paterno et al., 2021; Penagarikano et al., 2011; Vogt et al., 2018), we counted PV-expressing cells in the striatum, using immunohistochemistry and fluorescent in situ hybridization. We found no significant difference in the number of PV-positive cells in the dorsal striatum of Cntnap2^-/- mice compared to WT (Fig. S3A-F and K-N). We also found no changes in the average single cell or total level of PV protein expression in the dorsal striatum (Fig. 3G-J). Overall, we did not observe significant changes in either PV interneuron number, PV expression, or PV interneuron- mediated inhibition in the adult Cntnap2^-/-striatum compared to WT controls.

Figure 3.

SPN intrinsic excitability is increased in Cntnap2^-/- mice

Given that the increased cortical drive onto Cntnap2^-/-SPNs could not be explained by changes in excitatory or inhibitory synaptic function, we tested whether it could be due to a change in the intrinsic excitability of SPNs. To measure this, we recorded from dSPNs and iSPNs in Cntnap2^-/-;D1-tdTomato mice and injected current steps of increasing amplitude. We found that Cntnap2^-/- SPNs exhibited significantly increased intrinsic excitability compared to WT SPNs, which was most pronounced for dSPNs (Fig. 3A-L). Cntnap2^-/- dSPNs, but not iSPNs,

also had reduced rheobase current (Fig. 3C), the minimum current required to evoke an AP, as well as increased membrane resistance (Fig. 3D). Membrane capacitance (Fig. 3E,K), resting membrane potential (Fig. 3F,L), and AP shape (Fig. S4) were not significantly changed in Cntnap2-/- SPNs. Given the lack of synaptic changes observed in Cntnap2^-/- SPNs, the increase in SPN intrinsic excitability likely underlies their enhanced corticostriatal drive (see Fig. 1).

Cntnap2^-/- mice exhibit increased repetitive behaviors

RRBs comprise one of the primary symptom domains of ASD (APA, 2022). Alterations in striatal circuits are thought to be involved in the manifestation of RRBs, given the striatum’s role in action selection and motor control (Estes et al., 2011; Fuccillo, 2016; Hollander et al., 2005; Langen et al., 2014). To determine whether changes in motor behavior accompanied the altered striatal physiology in Cntnap2^-/- mice, we assessed locomotor activity and spontaneous repetitive behaviors using the open field, marble burying, and holeboard assays (Fig. 4A-D). In the open field, we found no significant difference in the total distance traveled, average speed, or number of rears in Cntnap2^-/- mice compared to WT controls (Fig. 4E-G). We did find that Cntnap2^-/- mice made significantly more entries into the center of the open field arena than WT mice, which may reflect a reduction in avoidance behavior in these mice (Fig. 4H). Manually scored grooming behavior revealed that Cntnap2^-/- mice initiated more grooming bouts than WT controls in the open field (Fig. 4I).

Figure 4.

To further assess motor behaviors in Cntnap2^-/- mice, we utilized the marble burying assay (Fig. 4C), which takes advantage of a mouse’s natural tendency to dig or bury. The number of marbles buried is used as a measure of persistent or repetitive behavior (Angoa- Perez et al., 2013). We found that Cntnap2^-/- mice buried significantly more marbles on average than WT controls (Fig. 4J). Another measure of repetitive behavior, which is based on the natural exploratory behavior of mice is the holeboard assay (Fig. 4D). In this task, the number of nose pokes made into unbaited holes is recorded. We found that Cntnap2^-/-mice made significantly more nose pokes within a 10-minute period than WT mice (Fig. 4K). This was largely due to increased poking during the last 5 minutes of the test (Fig. 4L-M), indicating persistent poking behavior in Cntnap2^-/- mice. Together with the increased grooming identified in the open field, increased marble burying and nose poking indicate an increase in RRBs in Cntnap2^-/-mice. A summary of behavior test results by genotype and sex is shown in Table S1.

To gain further insight into the spontaneous behavior profile of Cntnap2^-/-mice, we utilized a combination of DeepLabCut and Keypoint-MoSeq to perform unbiased, machine learning-based assessment of general locomotion and behavior in an additional cohort of Cntnap2^-/- mice (Fig. 4N-P, S5) (Mathis et al., 2018; Wiltschko et al., 2020). Again, we found that Cntnap2^-/- mice did not exhibit major changes in general locomotor activity compared to WT littermates (Fig. S5A). Analysis of movement “syllables” using Keypoint-MoSeq revealed that across the 25 most frequently performed syllables, two syllables associated with grooming were performed with significantly increased frequency in Cntnap2^-/- mice (Fig. 5N). Cntnap2^-/-mice also had an increase in the total number of grooming bouts (Fig. S5B), replicating the findings in the manually scored cohort (see Fig. 4I). While syllable usage was generally similar between WT and Cntnap2^-/- mice, transitions between syllables differed between the groups (Fig. S5C).

Figure 5.

A measure of the entropy of syllable transitions revealed that Cntnap2^-/-mice exhibited less entropy, suggesting less variability in the transition from one movement syllable to the next (Fig. 4O). This rigidity in motor sequence may be indicative of more restricted motor behavior overall. Finally, we tested whether a trained decoder could accurately distinguish WT and Cntnap2^-/-mice using information about movement, syllable usage, or syllable transitions. The decoding models performed significantly better than chance at identifying WT and Cntnap2^-/- mice based on their syllable usage and transitions, but not general locomotor activity (Fig. 4P). Together, this analysis demonstrates that while overall locomotor activity is not strongly affected in Cnantp2^-/- mice, the behavior patterns of these mice are distinct from WT, reflecting enhanced presence of RRBs.

Cntnap2^-/- mice exhibit enhanced motor learning

The accelerating rotarod is a striatal-dependent measure of motor coordination and learning that has been used across a range of ASD mouse models (Cording & Bateup, 2023). Changes in corticostriatal circuits have been identified in mouse models of ASD with altered performance in the task (Cording & Bateup, 2023). Given the altered corticostriatal drive in Cntnap2^-/- mice, we tested whether motor coordination and learning were affected in these mice. In the rotarod test, mice learn to walk and then run to stay on a rotating rod as it increases in speed over the course of five minutes. Mice perform three trials a day for four days. In trials one through six, the rod increases in speed from five to 40 revolutions per minute (RPM), while in trials seven through 12 the rod increases from 10 to 80 RPM (Fig. 5A). Learning occurs over trials within a day, as well as across days, as the mouse develops and hones a stereotyped motor pattern to stay on the rod for increasing amounts of time (Rothwell et al., 2014; Yin et al., 2009). We found that Cntnap2^-/- mice performed significantly better than WT mice in this task, in particular, in the later trials when the rod rotates at the faster 10 to 80 RPM speed (Fig. 5B).

Initial performance (terminal velocity on trial one) was not significantly different between WT and Cntnap2^-/- mice (Fig. 5C), but the learning rate from trial one to trial 12 was increased in Cntnap2^-/- mice (Fig. 5D). These findings expand upon previous work indicating increased performance on both steady-state and accelerating rotarod tasks utilizing slower speeds in Cntnap2^-/-mice (Dawes et al., 2018; Penagarikano et al., 2011). These results also align with the increased rotarod performance seen in other ASD mouse models exhibiting enhanced corticostriatal drive (Benthall et al., 2021; Cording & Bateup, 2023).

Cntnap2^-/- mice exhibit cognitive inflexibility

RRBs include not just stereotyped movements, but also insistence on sameness and perseverative interests (APA, 2022). Cognitive inflexibility, a deficit in the ability to flexibly adapt to changes in the environment and update behavior, is a manifestation of ASD and some other psychiatric disorders, which are associated with striatal dysfunction (Fuccillo, 2016). Indeed, in individuals with ASD, the severity of RRBs is associated with measures of cognitive inflexibility, and evidence from imaging studies suggests that altered corticostriatal connectivity may be present in the case of both repetitive behaviors and cognitive inflexibility (Uddin, 2021). To assess cognitive flexibility in Cntnap2^-/-mice, we utilized a four-choice odor-based reversal learning assay (Johnson et al., 2016; Lin et al., 2022). Briefly, mice are trained to dig for a reward in one of four pots containing scented wood shavings (Fig. 6A). On the first day of the task (acquisition), the rewarded pot is scented with odor one (O1). Mice reach criterion when they have chosen O1 for at least eight of 10 consecutive trials. On day two, mice are given a recall test in which the rewarded pot is again scented with O1. After reaching criterion, the reversal trials begin, and the rewarded pot is scented with the previously unrewarded odor two (O2). To reach criterion, mice must learn the new association of O2 and reward and choose O2 for eight of 10 consecutive trials.

Figure 6.

During acquisition, Cntnap2^-/- mice performed similarly to WT controls, not differing in the average number of trials needed to reach criterion, the number of quadrant entries made before making a choice, or the latency to choose a pot (Fig. 6B-D). On day two, Cntnap2^-/- mice performed similarly to controls during recall, demonstrating successful consolidation of the odor- reward pairing (Fig. 6E). However, we found that Cntnap2^-/- mice exhibited a deficit in reversal learning, requiring significantly more trials on average than WTs to reach criterion once the odor-reward pairing was changed (Fig. 6F). Interestingly, during reversal, Cntnap2^-/- mice made fewer quadrant entries before making a digging choice and exhibited significantly decreased latency to make a choice compared to controls (Fig. 6G,H). Even after the first correct choice of O2 during reversal, Cntnap2^-/- mice took more trials to reach criterion than WTs (Fig. 6I). In terms of errors, Cntnap2^-/- mice made more reversal errors than WT mice (Fig. 6J), in particular perseverative (continuing to choose O1) and regressive (choosing O1 after correctly choosing O2 once) errors (Fig. 6K). Cntnap2^-/-mice did not differ from WT controls in choices of the novel (newly introduced during reversal) or irrelevant (never rewarded) odors, or in the number of omitted trials (timing out without making a choice) (Fig. 6K). Instead, the persistence in choosing O1, even after at least one correct choice of O2, drove the cognitive inflexibility in these mice (Fig. 6L). This persistence in choice may be reflective of the broader scope of RRBs in Cntnap2^-/- mice.

Discussion

In this study we tested whether loss of the ASD risk gene Cntnap2 results in altered striatal physiology or striatal-dependent behavior. We found that both direct and indirect pathway SPNs exhibited significantly increased cortical drive in Cntnap2^-/- mice. This change was likely not due to a change in excitatory or inhibitory synapses, as excitatory cortical input onto SPNs was unchanged and there were no significant deficits in inhibition onto SPNs in these mice. Instead, loss of Cntnap2 resulted in a significant increase in the intrinsic excitability of SPNs, in particular dSPNs. At the behavioral level, Cntnap2^-/- mice exhibited repetitive behaviors including increased grooming, nose poking, and marble burying. These mice also had enhanced motor learning, performing significantly better than controls in the accelerating rotarod task. Finally, Cntnap2^-/- mice exhibited cognitive inflexibility in the four-choice reversal learning task.

Cellular phenotypes of Cntnap2 loss

The loss of Caspr2 has a variable impact on intrinsic excitability across brain regions and cell types. The increased intrinsic excitability that we identified in SPNs has also been observed in Purkinje cells of the cerebellum (Fernandez et al., 2021) and pyramidal cells of the cortex (Antoine et al., 2019; Cifuentes-Diaz et al., 2023). However, we note hypoactivity (Brumback et al., 2018) or unchanged excitability (Lazaro et al., 2019) of pyramidal cells in some cortical regions in Cntnap2^-/- mice. Caspr2 is involved in the clustering of voltage-gated potassium channels, in particular at the juxtaparanodes of myelinated axons (Poliak et al., 1999; Poliak et al., 2003) and the axon initial segment (Inda et al., 2006). Indeed, there are profound deficits in the clustering of Kv1-family channels in Cntnap2^-/- mice, particularly Kv1.2 channels (Scott et al., 2019). These channels play an important role in regulating the intrinsic excitability of SPNs (Nisenbaum et al., 1994), and when blocked, result in decreased rheobase and increased firing (Shen et al., 2004). The loss of Caspr2 in Cntnap2^-/- mice may result in improper localization of these channels, which could contribute to the decreased rheobase and increased SPN excitability we observed. While Caspr2 has most often been discussed for this interaction with Kv1-family channels, recent work has shown that it can also play a role in calcium signaling. The extracellular domain of Caspr2, after undergoing cleavage through proteolytic processing, can interact with the calcium extrusion pump PMCA2, promoting calcium export and likely decreasing cell excitability (Martin-de-Saavedra et al., 2022). This process, which increases with neuronal activity, may be disrupted in Cntnap2^-/- mice, which could also contribute to the increased intrinsic excitability that we found in SPNs.

Prior studies of Cntnap2^-/- mice have identified changes in the number of PV-expressing interneurons in the cortex (Penagarikano et al., 2011; Vogt et al., 2018), hippocampus (Paterno et al., 2021; Penagarikano et al., 2011) and striatum (Penagarikano et al., 2011). However, this finding is inconsistent across studies, as others have reported no change in the number of PV interneurons in these regions (Ahmed et al., 2023; Lauber et al., 2018; Scott et al., 2019). One possible explanation for this disparity is altered PV protein expression in Cntnap2^-/- mice such that immunoreactivity varies in cell counting assessments. This is supported by the finding that the number of Vicia Villosa Agglutinin-positive (VVA+) perineuronal nets that preferentially surround PV cells is unchanged in Cntnap2^-/- mice, even when PV immunoreactivity varies (Hartig et al., 1992; Haunso et al., 2000; Lauber et al., 2018). Parvalbumin, a Ca²⁺ buffer, plays an important role in the intrinsic fast-spiking properties of PV interneurons, such that a reduction in PV protein expression is known to change PV intrinsic function (Orduz et al., 2013). However, altered intrinsic properties of PV interneurons has also been variably reported across brain regions and studies of Cntnap2^-/- mice, with subtle changes in PV firing properties reported in the developing striatum (Ahmed et al., 2023) and adult cortex (Vogt et al., 2018), but unchanged in the hippocampus (Paterno et al., 2021) and medial prefrontal cortex (Lazaro et al., 2019). In this study, we find no significant change in the number of PV interneurons or the striatal expression of PV protein in Cntnap2^-/-mice. Following from this, we find no significant deficit in inhibition onto SPNs nor do we find any differences in PV interneuron intrinsic excitability in these mice. Interestingly, we find that broad inhibition measured through intrastriatal stimulation is enhanced specifically onto iSPNs in Cntnap2^-/- mice. This change is likely driven by altered function or connectivity of another type of striatal interneuron, or other SPNs, which provide lateral inhibition onto iSPNs (Burke et al., 2017; Koos et al., 2004).

Loss of Cntnap2 alters striatal-dependent behaviors

The striatum can be separated into functionally distinct subregions. We focused on the dorsal striatum in this study because of its role in controlling motor and cognitive functions (Voorn et al., 2004), which are relevant to ASD (Fuccillo, 2016; Subramanian et al., 2017). The dorsal striatum can be further subdivided into the dorsomedial striatum (DMS) and the dorsolateral striatum (DLS), with the former considered an associative region involved in goal- directed action-outcome learning and the latter implicated in the acquisition of habitual or procedural behaviors (Packard & Knowlton, 2002). We focused on cell function in DLS as stereotyped, perseverative or persistent behaviors likely recruit DLS circuitry (Evans et al., 2024; Fuccillo, 2016). In the accelerating rotarod assay, learning and performance in the task has been associated with changes in the DLS. Positive modulation of the firing rate of DLS neurons occurs during rotarod training, in particular in later trials of the task, and synaptic potentiation of DLS SPNs in late training is necessary for intact performance (Yin et al., 2009). In line with this, lesions of the DLS impair both early and late rotarod learning (Yin et al., 2009). We found that Cntnap2^-/- mice had increased rotarod performance, most notably at the later stages when DLS function is strongly implicated. Functionally, we also found increased cortical drive of DLS SPNs in these mice, in particular of dSPNs, a change that was associated with increased rotarod performance in another mouse model with disruption of an ASD-risk gene (Benthall et al., 2021). Together, this supports a connection between the functional change observed in DLS SPN physiology and the increased motor routine learning in Cntnap2^-/- mice.

In terms of restricted, repetitive behaviors, we replicated prior studies showing increased spontaneous grooming in Cntnap2^-/- mice (Penagarikano et al., 2011). Early evidence implicates the striatum in the control of the syntax or sequence of movements in a rodent grooming bout, such that very small lesions of DLS are capable of disrupting grooming (Cromwell & Berridge, 1996). However, recent work has also outlined roles for cellular modulation in DMS and ventral striatal Islands of Calleja in the control of grooming behavior (Ramirez-Armenta et al., 2022; Zhang et al., 2021). Cntnap2^-/- mice also exhibited increased marble burying and nose poking. The precise neurobiological substrates of these behaviors are yet unclear, but evidence linking increases in these behaviors to changes in cortico-striatal and amygdala-striatal function supports the notion that these behaviors may fit into a broader basal ganglia-associated RRB- like domain (Albelda & Joel, 2012; Lee et al., 2024).

In the four-choice reversal learning task, Cntnap2^-/- mice showed no differences during the acquisition phase, suggesting that there were no broad deficits in learning. However, in the reversal stage of the task, Cntnap2^-/- mice took significantly more trials to learn a new odor- reward pairing, owing primarily to continued choice of the previously rewarded odor. The DMS and ventral striatum (nucleus accumbens) have been shown to play an important role in reversal learning (Izquierdo et al., 2017), and in the four choice task specifically (Delevich et al., 2022). Additionally, decreased dopamine release in the DLS is associated with deficits in reversal learning in this task (Kosillo et al., 2019; Lin et al., 2022). Together, the learning phenotypes seen in Cntnap2^-/- mice in the accelerating rotarod and reversal learning assay share an underlying rigidity in behavioral choice. In both cases, changes in striatal circuits likely underlie the repetitive, stereotyped behaviors.

In summary, our results fit into a model whereby divergent cellular changes in the striatum driven by a functionally diverse set of ASD risk genes similarly enhance corticostriatal drive, in particular, of the direct pathway. This in turn may facilitate striatal-dependent motor routine learning and behavioral perseveration. We speculate that a shared gain-of-function in striatal circuits may play a role in the formation of perseverative or repetitive behaviors in a sub- set of ASDs more broadly.

Experimental procedures

Mice

All animal procedures were conducted in accordance with protocols approved by the University of California, Berkeley Institutional Animal Care and Use Committee (IACUC) and Office of Laboratory Animal Care (OLAC) (AUP-2016-04-8684-2). Cntnap2^-/- mice and littermate Cntnap2^+/+controls with the following alleles were utilized for each experiment.

Table 1.

Mice were group housed on a 12 h light/dark cycle (dark cycle 9:00 AM – 9:00 PM) and given ad libitum access to standard rodent chow and water. Both male and female animals were used for experimentation. The ages, sexes, and numbers of mice used for each experiment are indicated in the respective method details and figure legends. All mice used for experiments were heterozygous or hemizygous for the Drd1a-tdTomato, Thy1-ChR2-YFP, PV-Cre, Ai32, or Ai9 transgenes to avoid potential physiological or behavioral alterations.

Electrophysiology

Mice (P50-60) were briefly anesthetized with isoflurane and perfused transcardially with ice-cold ACSF (pH = 7.4) containing (in mM): 127 NaCl, 25 NaHCO3, 1.25 NaH2PO4, 2.5 KCl, 1 MgCl2, 2 CaCl2, and 25 glucose, bubbled continuously with carbogen (95% O2 and 5% CO2). Brains were rapidly removed and coronal slices (275 μm) were cut on a VT1000S vibratome (Leica) in oxygenated ice-cold choline-based external solution (pH = 7.8) containing (in mM): 110 choline chloride, 25 NaHCO3, 1.25 NaHPO4, 2.5 KCl, 7 MgCl2, 0.5 CaCl2, 25 glucose, 11.6 sodium ascorbate, and 3.1 sodium pyruvate. Slices were recovered in ACSF at 36°C for 15 min and then kept at room temperature (RT) before recording. Recordings were made with a MultiClamp 700B amplifier (Molecular Devices) at RT using 3-5 MOhm glass patch electrodes (Sutter, #BF150-86-7.5). Data were acquired using ScanImage software, written and maintained by Dr. Bernardo Sabatini (https://github.com/bernardosabatini/SabalabAcq). Traces were analyzed in Igor Pro (Wavemetrics). Recordings with a series resistance > 25 MOhms or holding current above 200 pA were rejected.

Current-clamp recordings

Current clamp recordings were made using a potassium-based internal solution (pH = 7.4) containing (in mM): 135 KMeSO4, 5 KCl, 5 HEPES, 4 Mg-ATP, 0.3 Na-GTP, 10 phosphocreatine, and 1 EGTA. For corticostriatal excitability experiments, optogenetic stimulation consisted of a full-field pulse of blue light (470 nm, 0.5 ms pulse width, CoolLED) through a 63x objective (Olympus, LUMPLFLN60XW). Light power was linear over the range of intensities tested. No synaptic blockers were included. For intrinsic excitability experiments (both SPN and PV interneuron), NBQX (10 μM, Tocris, #1044), CPP (10 μM, Tocris, #0247) and picrotoxin (50 μM, Abcam, #120315) were added to the external solution to block synaptic transmission. One second depolarizing current steps were applied to induce action potentials.

No holding current was applied to the membrane.

Voltage-clamp recordings

Voltage-clamp recordings were made using a cesium-based internal solution (pH = 7.4) containing (in mM): 120 CsMeSO4, 15 CsCl, 10 TEA-Cl, 8 NaCl, 10 HEPES, 1 EGTA, 5 QX-314, 4 Mg-ATP, and 0.3 Na-GTP. Recordings were acquired with the amplifier Bessel filter set at 3 kHz. Corticostriatal synaptic stimulation experiments to measure evoked AMPA-mediated EPSCs were performed in picrotoxin (50 μM), and optogenetic stimulation consisted of a full- field pulse of blue light (470 nm, 0.15 ms pulse width) through a 63x objective. Synaptic stimulation experiments to measure evoked IPSCs were performed in NBQX (10 μM) and CPP (10 μM). For electrically evoked IPSCs, a concentric bipolar stimulating electrode (FHC, #30214) was placed in dorsal striatum, roughly 200 μm medial to the recording site in dorsolateral striatum, and a 0.15 ms stimulus was applied. For PV-interneuron optically evoked IPSCs, a full-field pulse of blue light (470 nm, 0.15 ms pulse width) was applied through a 63x objective at the recording site.

Dendritic imaging and spine analysis

Neonatal (P1-3) Cntnap2^-/-;D1-tdT and Cntnap2^+/+;D1-tdT mice were cryoanesthetized and injected bilaterally with 200 nL AAV1.hSyn.eGFP.WPRE.bGH (Penn Vector Core, #p1696 (Keaveney et al., 2018)), diluted 1:75 in saline to achieve sparse transduction. Injections were targeted to the dorsal striatum, with coordinates approximately 1.3 mm lateral to midline, 2.0 mm posterior to bregma, and 1.5 mm ventral to the head surface. At P50-60, mice were anesthetized with isoflurane and transcardial perfusion was performed with 10 mL of 1x PBS followed by 10 mL of ice cold 4% PFA (EMS, #15710-S) in 1x PBS. Brains were post-fixed in 4% PFA in 1x PBS overnight at 4° C. 80 μm coronal sections were made using a freezing microtome (American Optical, AO 860) and stored in 1x PBS at 4° C. Sections were blocked for 1 hour at RT in BlockAid (ThermoFisher, #B10710) and incubated for 48 hours with gentle shaking at 4° C with antibodies against GFP (1:2500, Abcam, #13970) and RFP (1:1000, Rockland (VWR, #600-401-379) diluted in PBS-Tx (1x PBS with 0.25% Triton X-100 (Sigma, #T8787). Sections were washed 3 x 10 min in PBS-Tx and incubated with gentle shaking for 1 hour at RT with Alexa Fluor 488 and 546 secondary antibodies (1:500, Invitrogen, #A11039, #A11035). Sections were washed 3 x 10 min in 1x PBS and mounted onto SuperFrost slides (VWR, #48311- 703) using VECTASHIELD HardSet Antifade Mounting Medium (Vector Laboratories, #H-1400-10). Z stack images of individual dendrites were taken on a confocal microscope (Olympus FLUOVIEW FV3000) with a 60x oil immersion objective (Olympus #1- U2B832) at 2.5x zoom with a step size of 0.4 μm and deconvoluted using Olympus CellSens software. To quantify spine density, dendrites and spines were reconstructed using the FilamentTracer module in Imaris software (Oxford Instruments). The spine density of each dendrite was calculated using Imaris. Dendrites analyzed varied in total length, but excluded the most proximal and distal portions of the dendrite.

Brain sectioning and immunohistochemistry

Adult mice were perfused as above and brains were post-fixed with 4% paraformaldehyde overnight, then sectioned coronally at 30 μm. For immunohistochemistry,

individual wells of sections were washed for 3 x 5 min with 1x PBS, then blocked for 1 hour at RT with BlockAid blocking solution. Primary antibodies diluted in PBS-Tx were added and tissue was incubated for 48 hours with gentle shaking at 4° C. Sections were then washed 3 x 10 min with PBS-Tx. Secondary antibodies diluted 1:500 in PBS-Tx were added and incubated with gentle shaking for 1 hour at RT. Sections were washed 3 x 10 min in 1x PBS. Sections were mounted onto SuperFrost slides (VWR, #48311- 703) and coverslipped with VECTASHIELD HardSet with DAPI (Vector Laboratories, #H-1500-10) or VECTASHIELD HardSet Antifade Mounting Medium (Vector Laboratories, #H-1400-10). The following antibodies were used: mouse anti-PV (1:1000, Sigma, #P3088), rabbit anti-PV (1:1000, Abcam, #11427), anti-RFP (1:500, Rockland, #600-401-379), Alexa Fluor 405, 488 and 546 conjugated secondary antibodies (1:500, Invitrogen, #A-31553, #A-11001, #A-11003, and #A-11035).

PV cell counting

To count PV+ interneurons, Z-stack images of immunostained striatal sections were taken on a confocal microscope (Olympus FLUOVIEW FV3000) with a 10x or 20x objective (Olympus # 1-U2B824 or Olympus # 1-U2B825) and step size of 2 μm. For quantification, image stacks were Z-projected to maximum intensity using Fiji (ImageJ) and cropped to a 400 μm x 400 μm image in anatomically matched sections of the DLS. All PV-expressing cells within this region were counted using the ROI manager tool in ImageJ. Designation of a cell as PV positive was determined by the experimenter and consistently maintained across animals. Experimenter was blind to genotype and ROIs were made on the DAPI channel to avoid selecting regions based on PV expression. To quantify bulk PV fluorescence, ROIs were manually defined in ImageJ using the Freehand tool to cover as much of the DLS as possible, and mean fluorescence intensity was measured. To quantify individual cell PV fluorescence, ROIs were manually defined around every PV positive cell in the previosuly drawn DLS ROI using the Freehand tool, and mean fluorescence intensity was measured.

Western Blot

Adult mice (P48-55) were deeply anesthetized with isoflurane and decapitated. Brains were rapidly dissected and 1.5 mm dorsal striatum punches (Biopunch, Ted Pella, #15111-15) were collected from both hemispheres, flash-frozen in liquid nitrogen, and stored at −80° C. On the day of analysis, frozen samples were sonicated (QSonica Q55) until homogenized in 200 μl lysis buffer containing 1% SDS in 1x PBS with Halt phosphatase inhibitor cocktail (Thermo Fisher Scientific, #PI78420) and Complete mini EDTA-free protease inhibitor cocktail (Roche, #4693159001). Sample homogenates were then boiled on a heat block at 95° C for 5 min and allowed to cool to RT. Total protein content was determined using a BCA assay (Thermo Fisher Scientific, #23227). Following the BCA assay, protein homogenates were mixed with 4x Laemmli sample buffer (BioRad, #161-0747). 12.5μg of total protein per sample were then loaded onto 12% Criterion TGX gels (BioRad, #5671044) and run at 65 V. Proteins were transferred to a PVDF membrane (BioRad, #1620177) at 11 V for 14 hours at 4° C using the BioRad Criterion Blotter (BioRad, #1704070). Membranes (BioRad, #1620177) were briefly reactivated in methanol and rinsed in water 3x. After rinsing, membranes were blocked in 5% milk in 1x TBS with 1% Tween (TBS-Tween) for 1 hour at RT before being incubated with primary antibodies diluted in 5% milk in TBS-Tween overnight at 4° C. The following day, after 3 x 10 min washes with TBS-Tween, membranes were incubated with secondary antibodies for 1 hour at RT. Following 6 × 10 min washes, membranes were incubated with chemiluminescence substrate (PerkinElmer #NEL105001EA) for 1 min and exposed to Amersham Hyperfilm ECL (VWR, #95017-661).

Bands were quantified by densitometry using ImageJ software. GAPDH was used to normalize protein content and data is expressed as a percentage of control within a given experiment. The following antibodies were used: anti-Caspr2 (1:5000, Abcam, #153856), anti- PV (1:2500, Abcam, #11427), anti-GAPDH (1:5000, Cell Signaling, #51745S), and anti-rabbit goat HRP conjugate (1:5000, BioRad, #1705046).

In situ hybridization

Fluorescent in situ hybridization was performed to quantify Pvalb mRNA expression in the striatum of Cntnap2^+/+ and Cntnap2^-/-mice. Mice were briefly anesthetized with isoflurane and brains were harvested, flash-frozen in OCT mounting medium (Thermo Fisher Scientific, #23-730-571) on dry ice and stored at -80° C for up to 6 months. 16 µm sections were collected using a cryostat (Thermo Fisher Scientific, Microm HM 550), mounted directly onto Superfrost Plus glass slides (VWR, #48311-703) and stored at -80° C for up to 6 months. In situ hybridization was performed according to the protocols provided with the RNAscope Multiplex Fluorescent Reagent Kit (ACD, #323100). Drd1a mRNA was visualized with a probe in channel 2 (ACD, #406491-C2) and Pvalb mRNA in channel 3 (ACD, #421931-C3). After incubation, sections were secured on slides using ProLong Gold Antifade Mountant with DAPI (Invitrogen, P36935) and 60 x 24 mm rectangular glass coverslips (VWR, #16004-096). Sections were imaged on an Olympus FluoView 3000 confocal microscope using a 10x objective with 1.5x zoom and a step size of 2 µm. Pvalb-expressing cells were quantified across the entire striatum using the ROI manager tool in ImageJ. A cell was considered Pvalb positive if over 50% of the cell contained fluorescent puncta when compared to the DAPI channel. Experimenter was blind to genotype.

Behavioral analysis

All behavior studies were carried out in the dark phase of the light cycle under red lights (open field) or white lights (marble burying, holeboard, rotarod, and four choice reversal learning). Mice were habituated to the behavior testing room for at least 30 min prior to testing. Mice were given at least one day between different tests. All behavior equipment was cleaned between each trial and mouse with 70% ethanol and rinsed in diluted soap followed by water at the end of the day. If male and female mice were to be tested on the same day, male mice were run first then returned to the housing room, after which all equipment was thoroughly cleaned prior to bringing in female mice for habituation. Behavioral tests were performed with young adult male and female mice (7-11 weeks old). The experimenter was blind to genotype throughout the testing and scoring procedures.

Open field assay

Exploratory behavior in a novel environment and general locomotor activity were assessed by a 60 min session in an open field chamber (40 cm L x 40 cm W x 34 cm H) made of transparent plexiglass. Horizontal infrared photobeams (Stoelting, 60001-02A) were positioned to detect rearing. The mouse was placed in the bottom right-hand corner of the arena and behavior was recorded using an overhead camera and analyzed using ANY-maze software (Stoelting). An observer manually scored self-grooming behavior during the first 20 minutes of the test. A grooming bout was defined as an unbroken series of grooming movements, including licking of body, paws, or tail, as well as licking of forepaws followed by rubbing of face with paws.

Open field assay with DeepLabCut Keypoint-MoSeq analysis

Mice were placed in the open field arena and video recorded with a monochrome camera (FLIR Grasshopper 3, GS3-U3-41C6NIR-C) and a 16 mm wide angle lens (Kowa, LM16HC) placed above the arena from a height of 50 cm. To extract the body part (keypoint) coordinates from the video recordings, DeepLabCut (DLC) 2.3.4 (Mathis, et al. 2018; Nath, et al. 2019) was used. Fourteen body parts including nose, head, left ear, right ear, left forelimb, right forelimb, spine 1, spine 2, spine 3, left hindlimb, right hindlimb, tail 1, tail 2, and tail 3 were manually labeled on a small subset of the video frames. A DLC model was then trained using the annotated frames to label those 14 body parts for all videos recorded. The total distance traveled, and number of center entries were calculated using the coordinate of bodypart tail 1.

Discrete behavior syllables were extracted using Keypoint-MoSeq 0.4.4 (Weinreb, et al. 2023). Syllable usage and transition data were obtained using built-in functions of the Keypoint-MoSeq package. Decoding and entropy analysis were performed using customized Python 3.9 script. Code available upon request in Github. Entropy was calculated using the following equation, where u_i denotes the frequency of the syllable i and p_i,j denotes the transition probability from syllable i to syllable j.:

Marble burying assay

The marble burying assay was used to test for repetitive behavior. 20 black marbles were arranged in an orderly 4 x 5 grid on top of 5 cm of clean corn cob bedding in a standard mouse cage. Overhead room lights were on and white noise was played to induce mild stress. Mice were placed in the cage with the marbles for 30 minutes. The number of unburied marbles (>50% exposed) was recorded after the session.

Holeboard assay

The holeboard assay was used to measure exploratory and repetitive behavior. The holeboard apparatus consisted of a smooth, flat, opaque gray plastic platform, suspended 10 cm from the base by four plastic pegs in each corner. The board contained 16 evenly spaced 2 cm diameter holes and was surrounded by a 30 cm high clear plastic square encasing. During testing, mice were placed into the center of the holeboard. Mice explored the board for 10 minutes while video was recorded from both an above and side-view camera. Videos were used post-hoc to manually count and map the number of nose pokes made during the task. Nose pokes were defined as the mouse’s nose passing through the board barrier when viewed through the side-view camera.

Accelerating rotarod assay

The accelerating rotarod test was used to examine motor coordination and learning.

Mice were trained on a rotarod apparatus (Ugo Basile, #47650) for four consecutive days. Three trials were completed per day with a 5 min break between trials. The rotarod was accelerated from 5-40 revolutions per minute (rpm) over 300 s for trials 1-6 (days 1 and 2), and from 10-80 rpm over 300 s for trials 7-12 (days 3 and 4). On the first testing day, mice were first acclimated to the apparatus by being placed on the rotarod rotating at a constant 5 rpm for 60 s and returned to their home cage for 5 min prior to starting trial 1. Latency to fall, or to rotate off the top of the rotarod barrel, was measured by the rotarod stop-trigger timer.

Four choice odor-based reversal learning test

The four-choice odor-based reversal learning test was used to assess learning and cognitive flexibility. Animals were food restricted for 6 days in total, with unrestricted access to drinking water, and maintained at 90-95% of ad lib feeding body weight. Food was given at the end of the day once testing was completed. Food restriction and introduction to Froot Loop cereal piece (Kellogg’s, Battle Creek, MI) began 48 hours before pre-training. The four-choice test was performed in a custom-made square box (30.5 cm L × 30.5 cm W × 23 cm H) constructed of clear acrylic. Four internal walls 7.6 cm wide partially divided the arena into four quadrants. A 15.2 cm diameter removable cylinder fit in the center of the maze and was lowered between trials (after a digging response) to isolate the mouse from the rest of the maze. Odor stimuli were presented mixed with wood shavings in white ceramic pots measuring 7.3 cm in diameter and 4.5 cm deep. All pots were sham baited with a piece of Froot Loop cereal secured underneath a mesh screen at the bottom of the pot. This was to prevent mice from using the odor of the Froot Loop to guide their choice. The apparatus was cleaned with 2.5% acetic acid followed by water and the pots were cleaned with 70% ethanol followed by water between mice. The apparatus was cleaned with diluted soap and water at the end of each testing day.

On the first habituation day of pre-training (day 1), animals were allowed to freely explore the testing arena for 30 min and consume small pieces of Froot Loops placed inside empty pots positioned in each of the four corners. On the second shaping day of pre-training (day 2), mice learned to dig to find cereal pieces buried in unscented coarse pine wood shavings (Harts Mountain Corporation, Secaucus, NJ). A single pot was used and increasing amounts of unscented wood shavings were used to cover each subsequent cereal reward. The quadrant containing the pot was alternated on each trial and all quadrants were rewarded equally. Trials were untimed and consisted of (in order): two trials with no shavings, two trials with a dusting of shavings, two trials with the pot a quarter full, two trials with the pot half full, and four trials with the cereal piece completely buried by shavings. The mouse was manually returned to the center cylinder between trials.

On the days for odor discrimination (day 3, acquisition) and reversal (day 4), wood shavings were freshly scented on the day of testing. Anise extract (McCormick, Hunt Valley, MD) was used undiluted at 0.02 ml/g of shavings. Clove, litsea, and eucalyptus oils (San Francisco Massage Supply Co., San Francisco, CA) were diluted 1:10 in mineral oil and mixed at 0.02 ml/g of shavings. Thymol (thyme; Alfa Aesar, A14563) was diluted 1:20 in 50% ethanol and mixed at 0.01 ml/g of shavings. During the discrimination phase (day 3), mice had to discriminate between four pots with four different odors and learn which one contained a buried food reward. Each trial began with the mouse confined to the start cylinder. Once the cylinder was lifted, timing began, and the mouse could freely explore the arena until it chose to dig in a pot. Digging was defined as purposefully moving the shavings with both front paws. A trial was terminated if no choice was made within 3 min and recorded as omission. Criterion was met when the animal completed eight out of ten consecutive trials correctly. The spatial location of the odors was shuffled on each trial. The rewarded odor during acquisition was anise.

The first four odor choices made during acquisition were analyzed to determine innate odor preference by the percentage of choices for a given odor: Cntnap2^+/+ mice: 60% thyme, 25% anise, 12.5% clove, and 2.5% litsea. Cntnap2^-/-mice: 47.5% thyme, 45% anise, 7.5% clove, 0% litsea. We note that both Cntnap2^+/+ and Cntnap2^-/-mice exhibited the strongest innate preference for thyme, an unrewarded odor. There were no significant differences in innate odor preference.

The reversal phase of the task was carried out on day 4. Mice first performed the task with the same rewarded odor as the discrimination day to ensure they learned and remembered the task. After reaching criterion on recall (eight out of ten consecutive trials correct), the rewarded odor was switched, and mice underwent a reversal learning test in which a previously unrewarded odor (clove) was rewarded. A novel odor (eucalyptus) was also introduced, which replaced thyme. Perseverative errors were choices to dig in the previously rewarded odor that was no longer rewarded. Regressive errors were choosing the previously rewarded odor after the first correct choice of the newly rewarded odor. Novel errors were choices to dig in the pot with the newly introduced odor (eucalyptus). Irrelevant errors were choices to dig in the pot that had never been rewarded (litsea). Omissions were trials in which the mouse failed to make a digging choice within 3 min from the start of the trial. Total errors were the sum of perseverative, regressive, irrelevant, novel, and omission errors. Criterion was met when the mouse completed eight out of ten consecutive trials correctly. The spatial location of the odors was shuffled on each trial.

Quantification and statistical analysis

Experiments were designed to compare the main effect of genotype. The sample sizes were based on prior studies and are indicated in the figure legend for each experiment.

Whenever possible, quantification and analyses were performed blind to genotype. GraphPad Prism version 10 was used to perform statistical analyses. The statistical tests and outcomes for each experiment are indicated in the respective figure legend. Two-tailed unpaired t tests were used for comparisons between two groups. For data that did not pass the D’Agostino & Pearson normality test, a Mann-Whitney test was used. Two-way ANOVAs were used to compare differences between groups for experiments with two independent variables. Statistical significance was defined in the figure panels as follows: *p < 0.05, **p < 0.01, ***p < 0.001.

Acknowledgements

This work was supported by SFARI research grant #514428 to H.S.B. and NIH fellowship #F31NS124499 to K.R.C.

Conflicts of interest

The authors declare no competing interests.

Supplementary Figures and Table

Supplementary Figure 1.

Supplementary Figure 2.

Supplementary Figure 3.

Supplementary Figure 4.

Supplementary Figure 5.

Supplementary Table 1.