Abstract
Social deficits are a hallmark of schizophrenia, often characterized by impairments in processing and integrating socially transmitted information. However, translational models that accurately capture these deficits remain scarce. The Disrupted-in-Schizophrenia 1 gene (DISC1), a key susceptibility factor implicated in the etiology of psychiatric disorders, has been shown to cause DISC1 protein aggregation and dysfunctional signaling when modestly overexpressed, ultimately resulting in aberrant dopamine homeostasis. In this study, we employed a transgenic rat model overexpressing human DISC1 (tgDISC1 rats) to investigate social reward learning and microstructural integrity in the brain. Using a modified Social Transmission of Food Preference (STFP) task, we report that male tgDISC1 rats failed to update reward preferences based on social information, despite intact nonsocial reward learning—suggesting a specific deficit in social reward learning. Diffusion tensor imaging (DTI) in a behaviorally naive cohort revealed reduced fractional anisotropy (FA) in key subcortical regions, including the nucleus accumbens, amygdala, and substantia nigra, as well as areas mediating cortical-subcortical communication as the thalamus. Structural alterations in corresponding neuroanatomical areas have also been described in DTI of schizophrenia patients. Our findings link aberrant DISC1 signaling with impaired connectivity in parts of the mesolimbic system, critical for integrating social information into decision-making. This model recapitulates both behavioral and structural endophenotypes of schizophrenia and suggests that social impairments may stem from a fine-grained circuit-selective dysfunction rather than a generalized reward processing deficit. The tgDISC1 rat thus offers a translational platform for probing the neural substrates of social dysfunction in psychiatric disorders.
Significance Statement
Disrupted-in-Schizophrenia-1 (DISC1) is a scaffold protein with broad regulatory functions implicated in psychiatric disorders. Overexpression of DISC1 leads to protein aggregation and signaling abnormalities, disrupting dopamine pathways and behavior. We tested whether DISC1 overexpression alters the impact of social information on reward valuation using a modified Social Transmission of Food Preference paradigm. Wild-type rats shifted their preference after social interaction, adopting the demonstrator’s initially nonpreferred flavor. In contrast, tgDISC1 rats failed to update reward preference based on social cues. Diffusion tensor imaging in behaviorally naive tgDISC1 rats revealed structural alterations in limbic circuits, potentially underlying this deficit. These results underscore the role of DISC1 signaling in integrating social information, informing mechanisms of impaired social reward learning in psychiatric conditions.
Introduction
Schizophrenia is a complex psychiatric disorder of heterogenous biological origin marked by a spectrum of symptoms, including pronounced deficits in social behavior. While social withdrawal and anhedonia are well-documented features, patients also show impairments in social learning—specifically, the ability to integrate social cues to guide behavior and decision-making (Bellack et al., 1990; Morrison et al., 2017; Catalano et al., 2020). For instance, individuals with schizophrenia exhibit reduced sensitivity to social rewards (Fett et al., 2019), likely impeding their ability to adapt to social environments and contributing to the broader social dysfunction characteristic of the disorder (van’t Wout et al., 2009). Yet, translational models mimicking these deficits remain limited. To address this gap, we employed a modified version of the Social Transmission of Food Preference (STFP; Galef et al., 1984) paradigm, which has recently been adapted as a model for social reward learning (Noguer-Calabús et al., 2022). In this task, observer rats first express a preference between two differently flavored rewards. A demonstrator rat is then fed the observers nonpreferred flavor, after which both rats briefly interact. Following this interaction, the observer's preference is reassessed. Typically, observers shift their preference toward the previously nonpreferred reward—reflecting integration of social information into their own value representation. This ethologically relevant paradigm thus enables investigation of social reward learning mechanisms and their potential disruption in psychiatric conditions.
Dopaminergic signaling plays a central role in social learning—especially in the encoding of reward value and guiding social decision-making in both humans and animals (Bayer and Glimcher, 2005; Kalenscher and Pennartz, 2008; Soutschek et al., 2017; Castrellon et al., 2019; Terenzi et al., 2022). Impaired sensitivity to social rewards in schizophrenia has been linked to alterations in dopaminergic brain regions (Butler et al., 2020), supporting the relevance of dopamine pathways in the disorder's social deficits (Kapur et al., 2005).
The Disrupted-in-Schizophrenia 1 (DISC1) protein is of particular interest in psychiatric research, given its strong link to dopamine signaling, brain development, and cellular processes, such as synaptic plasticity and spine formation (Brandon and Sawa, 2011; Dahoun et al., 2017).
There is substantial evidence supporting the pathogenic relevance of DISC1 protein dysfunction for schizophrenia and other mental illnesses, as insoluble DISC1 protein aggregates have been found in postmortem brain tissue and CSF in a subset of patients (Leliveld et al., 2008; Pils et al., 2023). DISC1 aggregation has also been reported in human cell lines and mouse brains, following proteostasis disruption by influenza A infection (Marreiros et al., 2020). This suggests that aberrant DISC1 protein handling and aggregation are a disease-relevant mechanism, even in the absence of overt gene mutation or widespread overexpression in the general schizophrenia population.
To model the DISC1-related pathology, a transgenic (tgDISC1) rat line was developed in which the human, non-mutant DISC1 protein is modestly overexpressed. Due to the protein's intrinsic propensity to misfold and aggregate (Leliveld et al., 2008; Cukkemane et al., 2025), the model is suitable to mimic cases of DISC1 multimerization observed in patients and investigate mechanistic effects. Supporting this, tgDISC1 rats exhibits dysregulated dopamine homeostasis—evidenced by increased dopamine transporter availability and elevated high-affinity dopamine D2 receptor expression—alongside behavioral phenotypes resembling psychiatric symptoms (Trossbach et al., 2016; Uzuneser et al., 2019; Nani et al., 2020).
Based on this, we hypothesized that DISC1 protein aggregation would disrupt the function of dopaminergic circuits involved in social reward learning, leading to impaired performance of tgDISC1 rats in the STFP paradigm. To further characterize the tgDISC1 model on a neurostructural level, we conducted in vivo diffusion tensor imaging (DTI) in a test-naive rat cohort. This analysis aimed to identify microstructural alterations associated with disrupted DISC1 signaling. Specifically, we assessed structural integrity across key neuroanatomical regions relevant to the STFP paradigm. This two-sided approach—investigating behavioral assessment and structural imaging—provides a comprehensive characterization of the tgDISC1 rat, including features potentially relevant to schizophrenia-related social dysfunction and/or associated hallmarks of microstructural brain pathophysiology.
Materials and Methods
Animals and housing
A total of 96 male Sprague Dawley rats, aged 3–6 months, were tested. The animals included transgenic homozygous DISC1 rats, as well as their wild-type littermates matched as siblings of the same parents (hereafter referred to as wild-type). Before testing, wild-type rats were assigned to either serve as controls or social demonstrators. The animals were bred at the ZETT (Zentrale Einrichtung für Tierforschung und wissenschaftliche Tierschutzaufgaben, Düsseldorf, Germany) of the Heinrich Heine University in Düsseldorf, Germany. The animals were maintained in Type 4 cages on wood chipped bedding (LASvendi) at ∼22 ± 2°C and 55 ± 5% humidity. Rats were paired-housed unless described differently (see below). For enrichment, a wooden block and a red PVC tube were added. The animals were kept in a reverse 12 h light/dark rhythm. Rats were supplied with laboratory rodent chow (Ssniff) and water ad libitum, unless stated otherwise in the behavioral task descriptions. All animal procedures were conducted in accordance with the German Tierschutzgesetz (Animal Welfare Act) and were approved by the local authority LANUV (Landesamt für Natur-, Umwelt- und Verbraucherschutz North Rhine-Westphalia, Germany).
Genotyping
The generation and genotyping of the tgDISC1 model has been described to great detail previously (Trossbach et al., 2016). Briefly, for genotyping, small parts of the tail tips were lysed in a buffer (100 mM Tris, pH 8, 5 mM EDTA, 0.2% SDS, 200 mM NaCl, and 100 mg/ml proteinase K). After overnight incubation in 500 µl of lysis (50°C, 800 rpm), genomic DNA was precipitated (100% isopropanol; 70% ethanol), centrifuged for 30 min with 14.000 rpm at 4°C and subsequently solubilized in distilled water. For qPCR detection, primers targeting PrP promoter region (forward: 5′-CTGATCTCCAGAAGCCCAAA 3′; reverse: 5′-CAGGCCTATTCCTTGACAGC-3′) and rat β-actin (forward: 5′ GCAACGCGCAGCCACTGTCG-3′; reverse: 5 ′-CCACGCTCCACCCCTCTAC-3′) as a reference gene were utilized. For polymerase chain reaction (PCR; 10 min at 95°C, followed by 40 cycles of 10 s at 95°C, 20 s at 60°C, and 30 s at 72°C), DNA samples, either PrP or β-actin primers, SYBR Green I SuperMix (Roche), FactorQ, and distilled water were mixed. PCR products were analyzed via gel electrophoresis and LightCycler 480 software (Roche), with β-actin used for normalization.
Procedure
Rats were tested in a modified variant of the STFP as well as in a number of control tasks. The tasks were conducted in a fixed, chronological order: all observer rats first completed the 3-Chamber task, followed by the Odor Discrimination task, and then the STFP task (see below for task descriptions). The Reward Magnitude Discrimination and Reversal task were performed by a separate batch of rats. In general, all rats were handled for 5 min on 3 consecutive days before starting the first experiments. Before the start of the experiments, the animals were brought into the experimental room for habituation. Between trials, the experimental set-ups were cleaned with 70% ethanol. Experiments were conducted during the animals’ dark phase.
Social Transmission of Food Preference Task
The experimental design and analysis were performed as described by Noguer-Calabús et al. (2022) (Fig. 1). Three days prior to the STFP, rats were single housed in Type 4 cages with enrichment (wood block and tube) and acclimatized to the experimental setup. During this period, they were provided with custom-made metal hanging feeders containing two bowls with either 10 grape-flavored or 10 banana-flavored pellets (TestDiet).
Study design. Illustration of the experimental adaption of the Social Transmission of Food Preferences-Paradigm by Galef et al. (1984) to assess the social influence on revaluation of food rewards (Noguer-Calabús et al., 2022). In the example, the observer rat prefers flavor A over B. Habituation and Pre-Interaction Preference Testing: During the Habituation and Pre-Interaction Testing phase, observer rats were presented with two weighed cups on 3 consecutive days, each containing a different type of flavored pellets (A or B). The testing lasted for 6 h each day. After completing testing on Day 3, individual preferences of the observer were quantified by weighing the amount consumed of each respective reward. On Day 3, demonstrators were single housed and provided with a hanging feeder overnight containing the pellets that were not preferred by their assigned observers. Demonstrators were age- and sex-matched, unfamiliar wild-type rats of the same strain (for visualization purposes, colors differ). Social Interaction: On Day 4, each matched pair of demonstrators and observers were allowed to interact freely for 20 min. Post-Interaction Preference Testing: Immediately after the Social Interaction, observer rats were returned to their individual cages and provided with two cups, each containing one of the two pellet types. As in the Pre-Interaction Testing, the cups were removed after 6 h to assess observers’ preferences. This procedure was repeated the next day. Habit., Habituation. Illustrations were sourced from SciDraw and adapted by the author (https://doi.org/10.5281/zenodo.3926077; https://doi.org/10.5281/zenodo.3926414).
The STFP protocol was implemented in several stages: Habituation to the experimental procedure (Day 1), Pre-Interaction Preference Testing (Days 2 and 3), Social Interaction (Day 4), and Post-Interaction Preference Testing (Days 4 and 5).
During the entire course of the STFP, rats were subjected to mild food restriction, maintaining them at 85% of their free-feeding body weight. Standard laboratory chow was removed each morning on test days and returned after the 6 h preference measurements were completed. Water was available ad libitum.
Pre-Interaction Preference Testing
Each day during the Pre-Interaction Preference Testing, the observer rats were offered two food cups containing two different reward flavors (grape and banana). The rats had unrestricted access to the food cups for 6 h. Afterward, the cups were removed. Cups were weighed before and after testing.
Consumption was quantified individually for each observer rat by calculating the difference in cup weight before and after the 6 h testing period. On Day 3, following the 6 h measurement, individual preferences were determined based on the quantity of each pellet type consumed. The flavor consumed in greater quantity during Pre-Interaction Preference Testing was designated as the individual's preferred flavor (see below, Exclusion criteria).
After observers’ preferences were quantified on Day 3, demonstrator rats were single housed (with enrichment) and provided with a hanging feeder containing the pellets that were not preferred by their assigned observers. Demonstrators had access to the feeder all night preceding the Social Interaction on Day 4 (see below). Demonstrators were age- and sex-matched, unfamiliar wild-type rats of the same strain.
Social interaction
On Day 4, it was confirmed that demonstrators had consumed the flavored pellets before interaction by inspecting if any pellets were left in the cups. To enhance the odor that was to be transmitted, crushed pellets were gently rubbed onto the demonstrators’ back, snout, and genital area. Subsequently, the demonstrator was marked with a black pen on the back for identification purposes. A matched pair of demonstrators and observers were then allowed to interact freely in an open field (50 cm × 50 cm × 45 cm, PVC, illuminated to 5–15lux) for 20 min. The interaction was recorded (Conrad Electronic SE), and an experimenter blinded to the genotypes scored the social initiation behavior of the observer using Solomon Coder (Solomon Coder beta 19.08.02 © András Péter).
Post-Interaction Preference Testing
Immediately after Social Interaction on Day 4, the observer rats were placed back into their individual cages. As in the Pre-Interaction Preference Testing, they were given two separate cups, each filled with one of the two pellet flavors. The cups were removed and weighed after a 6 h interval. This preference test was repeated the next day. After completion, all animals were returned to their previous group housing conditions.
Analysis
A daily Preferences Index (PI) was calculated as described below (Noguer-Calabús et al., 2022):
Exclusion criteria
Observer rats were excluded from analysis if they did not consume any of the flavored pellets during the Habituation and Pre-Interaction Preference Testing. Further, rats with inconsistent Pre-Interaction preferences were also excluded from the analysis. This criterion applied if the reward type preferred on Day 3 was different from the preferred reward on Day 2. The reason for being strict about the consistency of the observer's food preference across Pre-Interaction days was to make sure that demonstrators were fed with the food that truly contrasted with the observer's original preference. Six rats of each genotype group were excluded from further analysis due to inconsistent preferences, or not consuming any flavored pellets at all.
Control tasks
We employed a number of control tasks described below. All observer rats were tested in all control tasks, except the Reward Magnitude Discrimination and Reversal task.
3-Chamber task
The 3-Chamber task is a well-established method for examining sociability and general interest in an unfamiliar conspecific (Ku et al., 2016). The apparatus consisted of an open field (60 cm × 60 cm × 39 cm, PVC, 5–10 lux) with two 20 cm × 20 cm compartments located in the left and right corners, separated by bars that allowed snout to snout contact. The task was performed in two trials, both trials were video-recorded (Conrad Electronic SE). The first trial was a habituation trial during which the observer rat was placed in the starting chamber and was allowed to freely explore the box for 5 min before being returned to its home cage. After a 10 min intertrial interval (ITI), a second, unfamiliar rat was placed in one of the compartments, and the observer rat was again placed in the starting chamber. This social trial lasted for 5 min. We pseudorandomly counterbalanced across observers into which compartment the second rat was placed. The recorded trials were analyzed using EthoVision XT 9 (Noldus), which measured locomotion, the time the observer spent in the interaction zone with the second rat, and the time spent exploring the empty compartment.
Odor discrimination
To control for rats’ ability to identify and discriminate between specific odors, an Odor Discrimination task was implemented prior to testing the modified STFP. The protocol was adapted from Noguer-Calabús et al. (2022). All trials were video-recorded (Conrad Electronic SE). Initially, the rats were placed into an open field (50 cm × 50 cm × 45 cm, PVC, illuminated to 5–15 lux) for 10 min to allow habituation to the arena. On the following day, during the sample trial, two customized 3D-printed bowls were placed at randomized corners of the open field. These bowls contained crushed pellets, either grape or banana (TestDiet), mixed with water in a 1:3 ratio to a volume of 10 ml. During the sample trial, both bowls contained the same odor; the choice of odor was pseudorandomized across rats. A 3D-printed grid cover on the bowls prevented the animals from consuming the liquid, allowing only olfactory exploration for 5 min before the rats were returned to their home cage. Following a 15 min ITI, the test trial was conducted. In this trial, two bowls were placed in the same positions as in the sample trial. One bowl contained the familiar odor from the sample trial, while the other contained a novel odor. The location and type of odor were pseudorandomized across rats. Recorded trials were analyzed using EthoVision XT 11.5 (Noldus). During the habituation phase, exploratory drive was measured by the number of visits to the center zone. In the sample and test trials, olfactory exploration of the bowls was evaluated by the number of nose touches and by comparing the total exploration duration between genotypes.
Reward Magnitude Discrimination and Reversal
The Reward Magnitude Discrimination and Reversal task were used to assess potential differences between genotypes in goal-directed motivation, reward magnitude discrimination, and reversal learning. Rats were food restricted to 85% of their free-feeding body weight but remained pair-housed throughout the experiment. Because this task involved the consumption of sucrose pellets, which might interact with performance in the STFP, we used rats from a different batch that were not tested in the STFP.
A customized T-maze was used (Ugo Basile S.R.L.). The T-maze featured automatic sliding doors that separated each arm, which were equipped with a pellet dispenser (Noldus) and a light signal at the far end. One arm of the T-maze was designated as the “start arm” (consistent for each animal), while the remaining two arms served as choice arms. The doors were independently operable, allowing access to one arm at a time during forced trials (see below). The entire apparatus was controlled by EthoVision XT 11.5 (Noldus).
Habituation and training
The training procedure followed the protocol previously described and consisted of three stages: Habituation, Shaping 1, and Shaping 2 (Zech et al., 2022).
Habituation: During Habituation, rats were allowed to explore the T-maze for 10 min. When a rat entered a choice arm, the associated automatic sliding door closed, and a food reward (sucrose pellet, 20 mg dustless precision pellets; Bio-Serv) was delivered by the pellet dispenser, accompanied by a 1 s light flash. After a short delay, the door reopened, allowing the rat to return to the start box to begin a new trial. There was no restriction on the number of pellets delivered during Habituation.
Shaping 1: Shaping 1 was conducted over 4 d, with a maximum session duration of 40 min per day. Each session allowed a maximum of 16 free trials per rat, where both arms were accessible and provided the same reward (one sucrose pellet). A session ended either when all free trials were completed or when the maximum session duration was reached. The number of completed free trials were compared each day between genotypes.
Shaping 2: In the Shaping 2 stage, rats first completed six forced trials where only one arm (in pseudorandom order; three trials allocated to each choice arm) was accessible, followed by 16 free trials where both arms were accessible. The maximum session duration remained 40 min. The reward in both arms continued to be one pellet. The number of completed free trials was compared each day between genotypes. Although Shaping 2 was planned to last for 4 d, technical issues on the final day resulted in the doors not responding adequately. Consequently, data from Day 4 of Shaping 2 were excluded from the analysis. To control for side bias, the number of free trials completed in each choice arm was compared across genotypes, with data from Day 3 being used for this comparison.
Reward Magnitude Discrimination Task (RMDT)
Upon completing the training phase, rats were subjected to the Reward Magnitude Discrimination Task (RMDT) over the course of 3 d. In this task, the arms of the T-maze were associated with either a small reward (one pellet) or a large reward (eight pellets). The assignment of reward amounts to arms was pseudorandomized between individuals and remained consistent throughout the RMDT. Each session consisted of six forced trials followed by 16 free trials, with a maximum session duration of 40 min. The percentage of large reward choices was recorded and compared between genotypes daily.
Reversal learning
Following the completion of the RMDT, rats underwent a reversal learning session. In this session, the location of the arm delivering the large reward was swapped for each individual. The number of forced trials, free trials, and the maximum session duration remained unchanged from the previous tasks. The reversal learning was conducted over a period of 3 d. The percentage of large reward selections was compared between genotypes on a daily basis.
Exclusion criteria
Rats were excluded from the tasks if they failed to complete a minimum of 10 free trials on 2 consecutive days. One animal had to be excluded from the reversal learning.
Magnetic resonance imaging and DTI
A total of 34 rats underwent magnetic resonance imaging (MRI) examination. All MRI measurements were performed in a horizontal 11.7 T 16 cm bore magnet (Bruker BioSpin, Bruker) using a transmit–receive 40 mm volume coil. The animals were anesthetized by 1.5–2% isoflurane in pure O2 delivered via a nose cone during the whole imaging session.
First, a high-resolution T2-weighted rapid acquisition with relaxation enhancement (RARE) anatomical scan was performed with the following parameters: repetition time (TR)/echo time (TE), 2,800/30 ms; field of view (FOV), 22 × 15 mm2; in-plane resolution, 156 µm × 156 µm; slice thickness, 1 mm; number of signal averages, 4. The images were used as a quality control to check for any volumetric deviations or structural abnormalities.
DTI data acquisition
After performing magnetic field shimming, DTI images were acquired with a spin-echo echoplanar imaging sequence with the following parameters: TR/TE, 3,200/21 ms; 30 diffusion gradient directions; b = 1,000 s/mm2; FOV, 22 × 15 mm2; in-plane resolution, 156 µm × 156 µm; slice thickness, 1 mm; 4 averages in order to increase the signal-to-noise ratio. Five additional DTI images with b = 0 s/mm2 were also obtained.
Volumetric segmentation
As slight changes in ventricle size has been an established feature of the tgDISC1 rat (Trossbach et al., 2016), we used volumetric analysis to ensure comparability between genotypes in this regard. First, manual segmentation of the right and left lateral ventricles was performed on MR images using ITK-SNAP (Yushkevich et al., 2006). The ventricles were segmented from coronal slices along a 5.88 mm section, from interaural 11.28 mm/bregma 2.28 mm to interaural 5.40 mm/bregma −3.60 mm (Paxinos and Watson, 2013). The start and end points of the segmentation were best marked by the appearance of the corpus callosum and inferior lateral ventricles respectively. Prior to segmentation, MR images were thresholded and binarized to show only the top 1% of voxel intensities, which should correspond to CSF on a T2-weighted MR image. All MRI analysis was carried out by a single operator but showed excellent intra-rater consistency with a Dice coefficient 0.99 ± 0.002 across two independent segmentations using all images available in the dataset.
DTI data processing
DTI data were postprocessed using DSI-Studio (http://dsi-studio.labsolver.org). Raw Bruker data were imported and corrected for distortions. Next, the quality of the data (number of bad slices indicated by DSI-Studio) was checked. Masks were specified in order to eliminate background signals. Further, image orientation was corrected in order to align the measured volume with the rat brain atlas implemented in DSI-Studio (Johnson et al., 2021). Generalized Q-Sampling Imaging (GQI) was used in order to reconstruct fractional anisotropy (FA), axial diffusivity (AD), and radial diffusivity (RD) maps, as measures of microstructural integrity and structural properties of the measured tissue. FA, AD, and RD maps were registered to the implemented atlas and values of FA, AD, and RD were computed in the regions of interest (ROIs; Fig. 2), selected based on their potential implication in social transmission of food preference (Table S2): nucleus accumbens (NAc), basolateral amygdala (BLA), cortical amygdala (CoA), hippocampus (HPC), infralimbic cortex (IL), lateral septum (LS), diagonal band of broca (DB; as representative of a major cholinergic structure of the basal forebrain, Berger-Sweeney et al., 2000), olfactory areas (OLF), orbital cortex (ORB), piriform cortex (PIR), prelimbic cortex (PL), substantia nigra pars compacta (SNc), and the thalamus (TH).
Representative weighted color maps of Fractional Anisotropy (FA). A, FA weighted color maps from slices displaying regions of interest (ROIs). The ROIs were bilaterally selected from the atlas implemented in DSI-Studio (based on Johnson et al., 2021). NAc, nucleus accumbens; BLA, basolateral amygdala; CoA, cortical amygdala; HPC, hippocampus; IL, infralimbic cortex; LS, lateral septum; DB, diagonal band of Broca; OLF, olfactory areas; ORB, orbital cortex; PIR, piriform cortex; PL, prelimbic cortex; SNc, substantia nigra pars compacta; TH, thalamus. B, Exemplary 3D illustration of ROIs across the brain. Note that olfactory areas and piriform cortex are excluded for simplified visualization, as they cover the majority of the ventral areas. C, Individual streamlines of a weighted color map. The color represents the direction of the structural organization in which red is medial-lateral, green is anterior-posterior, and blue is dorsal-ventral direction.
Exclusion criteria
One rat died while under anesthesia for MRI scan and three rats with visible abnormalities (ventricular enlargements) in the anatomical images were excluded from further DTI measurements. Subsequent volumetric segmentation measurement uncovered substantial ventricular enlargement, in another two animals, which were also excluded from further analysis (Fig. S1). Importantly, the ventricular enlargement was not restricted to one genotype but found in both wild-types and tgDISC1 rats (wild-type n = 3; tgDISC1 n = 2), suggesting it is likely attributable to spontaneous variation within the background strain, as reported by other researchers, too (Mulla et al., 2012; Tu et al., 2014, 2017). Five more rats were excluded due to inability to finish DTI acquisition because of movement and one rat was excluded due to poor quality of the scan, indicated by DSI-Studio. Finally, 22 rats (n = 11/genotype) were included into the DTI data analysis.
Data analysis
For all parameters, data are presented as mean ± standard errors of the mean (SEM). Calculations were performed with RStudio (version 2023.06.2) and GraphPad Prism 9.5.0 (GraphPad Software). Graphs were generated with GraphPad Prism 9.5.0 (GraphPad Software). For statistical analysis, differences in the dependent variables between genotypes were compared with unpaired t tests or by nonparametric Mann–Whitney U tests, dependent on the (lack of) normal distribution, as assessed with Kolmogorov–Smirnov tests. Mixed-effects models with Restricted Maximum Likelihood estimation were used to compare the influence of multiple fixed factors. If applicable, post hoc analyses were calculated, corrected with Benjamini–Hochberg False Discovery rate (FDR) for multiple comparisons. The statistical significance level was set to α = 0.05.
Results
Behavior
tgDISC1 rats fail to update reward values after social contact with a demonstrator
We calculated an individual Preference Index before and after Social Interaction, indicating the strength of preference for one reward flavor over the other in each respective task phase (Noguer-Calabús et al., 2022). We computed a mixed-effects model of the PI, with contact (Pre-Interaction vs Post-Interaction) as within-subject factor and genotype (wild-type vs tgDISC1) as between-subject factor. The analysis revealed a main effect of contact as well as an interaction effect of contact and genotype (contact: F(1,26) = 15.62, p = 0.0005; genotype: F(1,26) = 1.93, p = 0.1769; contact × genotype: F(1,26) = 8.89, p = 0.0062; Fig. 3). Post hoc tests demonstrated a difference in PI for wild-type rats depending on social contact (p < 0.0001), indicating that these rats shifted their food preferences following Social Interaction. This result supports the occurrence of social transmission of food preferences in wild-type rats. Conversely, tgDISC1 rats showed no difference in PI before and after interaction (p = 0.4967), suggesting that tgDISC1 rats failed to show socially transmitted food revaluation.
tgDISC1 rats show no revaluation of reward values after social contact with a demonstrator. A, Preference index (PI). The PI indicating the strength of preference for one reward over the other, for both genotypes and before (pre) versus after (post) Social Interaction. While wild-type rats decreased their preference for the originally preferred reward, thus showing social transmission of food preference, the PI did not change in tgDISC1 rats. B, Consumption of originally preferred food: wild-type, but not tgDISC1 rats decreased their mean consumption of the originally preferred food type (in grams) from pre- to postinteraction. C, Consumption of the originally nonpreferred food: wild-type, but not tgDISC1, rats increased the consumption of the originally nonpreferred food type (in grams) from pre- to postinteraction. D, Total pellet consumption over the course of the STFP. No change in the total consumption of flavored pellets, regardless of the flavor (in grams) from pre- to postinteraction for both genotypes. Data are mean ± SEM. *p < 0.05 and **p < 0.01 for contact × genotype interaction.
We additionally assessed total pellet consumption, i.e., the sum of both pellet types, to test for potential differences in hunger or satiation between genotypes. We found that total consumption of flavored pellets did not differ before or after the Social Interaction or between genotypes, as revealed by a mixed-effects model (contact: F(1,26) = 2.63, p = 0.1168; genotype: F(1,26) = 0.53, p = 0.37; contact × genotype interaction: F(1,26) = 0.43, p = 0.5181). This suggests that the genotype-dependent differences in the propensity to show social transmission cannot be explained by differences in general appetite, satiation, or food consumption (Fig. 3).
We further analyzed the average amount consumed of the originally preferred and nonpreferred flavors separately, using mixed-effects models with contact as the within-subject factor and genotype as the between-subject factor. For the originally preferred reward flavor, there was a main effect of contact (F(1,26) = 7.41, p = 0.0114) and a contact × genotype interaction (F(1,26) = 4.65, p = 0.0405). Post hoc comparisons showed a decrease in preferred reward consumption in wild-type rats after interaction (p = 0.0039), but no change in tgDISC1 rats (p = 0.6924).
For the nonpreferred rewards, the mixed-effects model also revealed a main effect of contact (F(1,26) = 19.25, p = 0.0002) and a contact × genotype interaction (F(1,26) = 7.48, p = 0.0111). Wild-type rats increased their consumption of the nonpreferred reward after Social Interaction (p < 0.0001), whereas tgDISC1 rats showed no change (p = 0.2530). Thus, following Social Interaction, wild-type rats increased their intake of the previously nonpreferred reward and reduced their consumption of the initially preferred one. In contrast, tgDISC1 rats showed no change in their reward preferences.
An overview of the conducted test battery of the control tasks and control variables and their respective parameters is listed in Supplementary Table S1. None of the comparisons revealed any significant differences in any of the variables between genotypes.
tgDISC1 rats show intact social interest and motivation compared with wild-type controls
To investigate potential social behavior impairments in tgDISC1 rats, we manually scored several aspects of social behavior initiated by the observer during the STFP task, including snout to snout contact, partner exploration, allogrooming, social play, following, and genital exploration. Comparisons between tgDISC1 and wild-type rats revealed no significant differences in any of these behaviors (Fig. 4; statistics for the group comparisons are summarized in Table S1). Stereotypical behaviors, such as repetitive self-grooming and rearing, were also compared as indicators of arousal. Again, no significant differences were observed between genotypes (Fig. 4, Table S1).
Behavior during the Social Interaction in the STFP. tgDISC1 rats do not significantly differ from wild-type rats in any of the measured social (A–F) or stereotypical behaviors (grooming and rearing; G, H) during STFP interaction. All data for social initiation by the observer in seconds. A, Duration of snout to snout sniffing. B, Duration of unspecific partner exploration by sniffing. C, Duration of allogrooming. D, Duration of social play. E, Duration of following. F, Duration of genital exploration. G, Duration of self-grooming. H, Duration of rearing. All data are mean ± standard error of the mean.
Prior to the STFP task, the 3-Chamber task was conducted to assess whether general social interest in an unfamiliar conspecific differed between genotypes. During the habituation phase, we recorded the observers’ locomotion, measured by total distance traveled, and their average velocity, with no significant differences between genotypes (Table S1). In the social interest phase, a mixed-effects model with genotype as the between-subject factor and chamber (conspecific vs empty) as the within-subject factor showed a main effect of chamber on exploration duration (F(1,76) = 58.46, p < 0.0001), but no main effect of genotype (F(1,76) < 0.01, p = 0.9934) nor a interaction effect (F(1,76) = 2.95, p = 0.0898). These results suggest that we have no evidence to assume that tgDISC1 rats have impaired social interest in the 3-Chamber task.
Olfactory capacity and odor discrimination are not altered in tgDISC1 rats
To control for potential olfactory impairments that could influence the transmission of reward preference, we used an Odor Discrimination task to assess the olfactory capacity of tgDISC1 rats. During habituation in the open field, the number of visits to the center zone and general locomotion were measured, with no significant differences between genotypes (Table S1).
In the sample trial, two bowls containing the same odor were presented, and a mixed-effects model with genotype as the between-subject factor and bowl as the within-subject factor was used to compare the number of nose touches. No differences were found between genotypes (genotype: F(1,38) = 0.95, p = 0.34; bowl: F(1,38) = 0.44, p = 0.51; bowl × genotype interaction: F(1,38) = 1.29, p = 0.26). We also found no significant difference in total exploration time between genotypes (Table S1).
In the test trial, which introduced a novel odor alongside a familiar one, a mixed-effects model with genotype as the between-subject factor and novelty as the within-subject factor revealed a main effect of novelty on number of nose touches (F(1,38) = 6.99, p = 0.0118), but no main effect of genotype (F(1,38) = 0.54, p = 0.465) or an interaction effect (F(1,38) = 0.52, p = 0.474). These results indicate that tgDISC1 rats show functional olfactory behavior and are capable of odor discrimination, suggesting that olfactory processing is not impaired in these animals (Fig. 5).
No impaired social interest, odor discrimination, motivation, or cognitive flexibility in tgDISC1 rats. A, Social interest to approach an unknown conspecific in the 3-Chamber task. The duration of exploration in the apparatus shows a significant main effect of chamber (empty vs conspecific), but there was no significant difference in exploration of the conspecific between genotypes. B, Odor discrimination between a novel and a familiar odor. We found a significant effect of novelty on the number of nose touches at both odors, but no difference between genotypes. C, The ability to distinguish large and small food rewards and the motivation to obtain them was tested in the Reward Magnitude Discrimination Task (RMDT). Percentages to choose the large food rewards in 16 free trials did not significantly differ between genotypes (compare Table S1). Red line indicates chance level of 50%. D, Flexible adaptation to changed reward contingencies during a reversal phase. Percentages to choose the large food rewards in 16 free trials did not significantly differ between genotypes (compare Table S1). Red line indicates chance level of 50%. *p < 0.05, ****p < 0.0001.
Goal-directed motivation and reward magnitude discrimination are not different between tgDISC1 and wild-type rats
We assessed goal-directed motivation and reward magnitude discrimination in tgDISC1 rats using the Reward Magnitude Discrimination Task (RMDT) in a T-maze setup. We found no significant difference between genotypes in the number of free trials performed during the two Shaping phases (Table S1), suggesting no evidence for a difference in learning speed and ability.
On Day 3 of the second Shaping phase, a Wilcoxon matched-pairs signed rank test was conducted for each genotype to determine if there were potential side biases toward one arm of the T-maze. No side biases were found in neither wild-type (p = 0.500) nor tgDISC1 rats (p = 0.187).
In the RMDT, we analyzed the daily percentages of choices of the arm containing the larger reward. No significant differences in large reward choices were observed between genotypes on any of the 3 d of testing (Fig. 5, Table S1).
No difference between tgDISC1 and wild-type rats in reversal learning
Following the RMDT, we evaluated cognitive flexibility with a reversal learning task. Over the course of 3 d, we compared, between genotypes, the daily percentage of large reward choices after reversing the magnitude-to-arm contingencies. No significant differences between genotypes were detected on any of the 3 d, indicating that tgDISC1 rats exhibit similar learning strategies and cognitive flexibility as wild-type rats (Fig. 5, Table S1).
Magnetic resonance imaging
tgDISC1 rats display impaired organization of tissue in multiple limbic areas
To investigate differences in brain (micro)structures between wild-type and tgDISC1 rats, we performed in vivo DTI in a behaviorally naive cohort of rats. FA, AD, and RD, reflecting tissue structural properties, were compared on group-level between tgDISC1 and wild-type rats. tgDISC1 rats had significantly lower FA values in regions of (or closely related to) the limbic system, including the NAc, amygdala (BLA and CoA), LS, DB, OLF, SNc, and TH (Fig. 6, Table 1).
Changes in fractional anisotropy (FA) observed in tgDISC1 rats compared with wild types. Abbreviations: NAc, nucleus accumbens; BLA, basolateral amygdala; CoA, cortical amygdala; HPC, hippocampus; IL, infralimbic cortex; LS, lateral septum; BF, basal forebrain, as representative for diagonal band of Broca, DB; OLF, olfactory areas; ORB, orbital cortex; PIR, piriform cortex; PL, prelimbic cortex; SNc, substantia nigra pars compacta; TH, thalamus. Data are mean ± standard error of the mean; ns, nonsignificant; *p < 0.05.
Summary of results of group-level comparisons in fractional anisotropy (FA), axial diffusivity (AD), and radial diffusivity (RD) values of brain regions implicated for Social Transmission of Food Preference (STFP) performance
There were no significant differences in AD and RD after correcting for multiple comparisons (Table 1). The observed reductions in FA, in the absence of changes in AD or RD, suggest subtle disruptions in microstructural organization rather than gross abnormalities in tissue diffusion. These results indicate that tgIDSC1 rats exhibited compromised tissue organization in key regions, potentially associated with social reward, rather than widespread structural impairments.
Discussion
This study aimed to characterize behavioral and structural consequences of DISC1 overexpression, leading to DISC1 protein aggregation, in a rat model of schizophrenia, with a particular focus on social reward learning and brain microstructure integrity. Using a modified version of the Social Transmission of Food Preference (STFP) paradigm, we examined whether tgDISC1 rats could update subjective reward values based on social information. In line with our hypothesis, tgDISC1 rats failed to change their initial reward preference after interacting with a demonstrator rat fed with the observers’ nonpreferred food, indicating a selective impairment in socially mediated reward value updating. Notably, performance in a nonsocial reward learning task was unimpaired, suggesting that the deficit was specific to the social domain rather than reflecting general learning or motivational dysfunction. This behavioral phenotype mirrors findings in schizophrenia patients, who show reduced sensitivity to social, but not nonsocial, rewards (Catalano et al., 2018; Lee et al., 2019), and difficulties integrating cognitive and affective information during decision-making (Heerey et al., 2008).
To further characterize the tgDISC1 model on a neural level, we conducted in vivo DTI in a separate cohort of behaviorally naive animals. This design choice allowed us to assess baseline alterations caused by disrupted DISC1 signaling without the confounding effects of behavioral testing, which has been shown to rapidly modulate DTI readouts within 1 h past testing (Blumenfeld-Katzir et al., 2011; Ding et al., 2013). ROIs were selected based on their previously established involvement in social learning in an STFP task (Table S2). Our analysis revealed significant structural changes in tgDISC1 rats, evident by decreases in FA values in key limbic structures, including the NAc, amygdala (BLA; CoA), LS, and SNc. We observed additional alterations in nodes controlling cortical-subcortical communication, such as the TH and the basal forebrain (Gielow and Zaborszky, 2017; Cruz et al., 2022). These changes in FA values conform with the known biological roles of DISC1, which is critically involved in synaptic trafficking, dendritic arborization, and neurodevelopment (Brandon and Sawa, 2011; Lipina and Roder, 2014), and support the interpretation that DISC1 protein aggregation may lead to altered microstructural organization in these ROIs. Importantly, the current results extend and replicate prior neuropathological findings in tgDISC1 rats, including divergent dopaminergic cell counts in the substantia nigra (Hamburg et al., 2016). Further, the absence of significant differences in other DTI measures suggests that the observed FA reductions are unlikely to stem from myelin or axonal deficits but rather reflect alterations in neurite organization within these subcortical regions. Although speculative, this aligns with the idea that FA in gray matter is more sensitive to dendritic complexity and synaptic architecture rather than traditional white matter integrity markers (Müller et al., 2019). Together, these data suggest that tgDISC1 rats display a network-level disruption in subcortical nodes which likely is the result from a subtle impairment of brain development (Kamiya et al., 2005; Brandon and Sawa, 2011).
Social information processing critically depends on the coordinated interaction of distributed neural circuits (O’Connell and Hofmann, 2012; Huang et al., 2020; Menon et al., 2021; Wang et al., 2021; Kietzman et al., 2022; Poggi et al., 2024; Kalenscher et al., 2025). A well-characterized BLA–PL–NAc circuit, for instance, is critical for integrating socially acquired information into reward-based decisions in mice (Kietzman et al., 2022), a process closely mirrored in our STFP paradigm. The critical involvement of BLA–PL–NAc is in line with work demonstrating that the NAc is essential for STFP, as male rats with exocytotic lesions of the NAc Shell were unable to revalue the originally nonpreferred reward (Noguer-Calabús et al., 2022). Importantly, neither the disruption of the BLA–PL–NAc circuit nor the NAc Shell lesion affected odor discrimination or general social interest, paralleling the presented phenotype of tgDISC1 rats. Similarly, previous studies stressed the importance of the amygdala and its subregions for the social transmission, as lesioning the BLA prevented associating social with olfactory cues (Wang et al., 2006). In addition, the CoA plays a unique role in integrating social and olfactory information, as it was engaged during STFP but not when social or olfactory stimuli were presented in isolation (Liu et al., 2024). In our model, FA reductions in these same regions—many of which are downstream targets of mesolimbic dopamine—may reflect the underlying vulnerability of this circuitry to disturbed DISC1 signaling. Indeed, aberrant DISC1 signaling via modest DISC1 protein aggregation has been shown to impair dopamine homeostasis (Trossbach et al., 2016) and monoamine levels in postmortem NAc and amygdala tissue in tgDISC1 rats (Wang et al., 2017; Uzuneser et al., 2019). In summary, since both, animal and human studies, have demonstrated a central role for dopamine in social reward processing, revaluation, and decision-making (Burke et al., 2017; Dang et al., 2018; Castrellon et al., 2019), it is plausible that the impaired social reward learning observed in tgDISC1 rats results from dysregulated mesolimbic dopaminergic signaling.
In studies involving human patients, several of the regions showing reduced FA in our rat model have also been implicated in the pathophysiology of schizophrenia across multiple MRI modalities. DTI studies previously reported reduced FA values in the amygdala, TH, and NAc (Kalus et al., 2005; Hashimoto et al., 2009; Spoletini et al., 2011; Cuesta et al., 2021). Importantly, these findings have been interpreted as impaired modulation of dopaminergic relay stations (Kalus et al., 2005) and disturbed information flow between limbic and cortical systems (Spoletini et al., 2011). Further, reduced FA along the NAc-TH pathway, regions also found altered in our model, has been observed in individuals at high risk for psychosis, suggesting an early involvement of these tracts in schizophrenia pathophysiology (Chen et al., 2025). Beyond structural abnormalities, consistent evidence from patient fMRI studies revealed altered functional connectivity in the same regions, including the amygdala (Kim et al., 2020), the TH (Avram et al., 2018), the SNc, and ventral striatum (White et al., 2015). Further, Martino et al. (2018) identified decreased connectivity within subcortical circuits overlapping with areas included in our results, strongly supporting the idea of broader functional network disruptions in schizophrenia (Metzner et al., 2024). Importantly, the aforementioned study also emphasizes the importance of dopamine dysregulation in regard to changed connectivity (Martino et al., 2018). Supporting this, a combined PET-fMRI study revealed a causal link between dopaminergic receptor binding potential, along with abnormal dopamine release, and altered connectivity patterns in subcortical-cortical regions in schizophrenia (Horga et al., 2016). Hence, empirical evidence from patient data suggests strong parallels with the altered brain regions and aberrant dopamine signaling observed in our tgDISC1 rats, even across a methodological variety, which further supports the clinical relevance of our findings.
Previous studies in healthy individuals have uncovered DISC1-dependent differences in the volume of brain regions closely related to our findings (Mühle et al., 2017), corroborating our conclusion that disruptions in DISC1 signaling influence regional brain structure and function. Interestingly, these changes were found to be sex dependent. Other research has proposed that estradiol may exert a beneficial effect on altered DISC1 function (Erli et al., 2020). Thus, it would be worthwhile to investigate this further in female tgDISC1 rats. While current literature reports no sex differences in social or memory behavior in this model (Uzuneser et al., 2019), it has not yet examined more nuanced interactions, such as those involving our adapted STFP paradigm or microstructural integrity.
Finally, the specificity of the social deficit in tgDISC1 rats’ contributes to an ongoing discussion about whether social impairments in schizophrenia reflect a general reward processing deficit or a more selective dysfunction in integrating social information (Fett et al., 2019; Lee et al., 2019; Butler et al., 2020; Hanssen et al., 2020). Our data support the latter view, suggesting that while basic reward processing may remain intact, the ability to flexibly integrate social cues in decision-making is compromised (Lee et al., 2019; Catalano et al., 2020)—a challenge that may arise from impaired communication within circuits to integrating emotional, cognitive, and social information to guide behavior (Heerey et al., 2008).
In conclusion, this study shows that DISC1 overexpression leading to DISC1 protein aggregation and altered DISC1 signaling impairs social, but not general, reward learning and is linked to specific microstructural alterations in a fine-grained network of subcortical regions critical for social reward learning. Future research should try to establish a mechanistic link with a focus on region- and circuit-specific interventions, as well as cellular analyses, to test (regional) causal links between DISC1 protein dysfunction and impaired social value computation. Given that current clinical assessments of social deficits in patients rely heavily on subjective self-report, animal models may provide an outset for quantifiable behavioral markers that link hallmarks of subcortical pathology to functionally relevant endophenotypes. The tgDISC1 model thus provides translational value for understanding circuit-level vulnerabilities that may underlie social dysfunction in psychiatric disorders like schizophrenia.
Footnotes
The work was supported by a grant from the German Research Foundation (Deutsche Forschungsgemeinschaft, DFG; grant no KO 1679/14-1 to C.K. and grant no. KA 2675/5-3 to T.K.). The AI-based language tools DeepL and ChatGPT 3.5 have been used for grammar correction and improving readability. Following their use, the authors thoroughly reviewed and revised the material as necessary, taking full responsibility for the final content of the publication.
↵*J.D. and Y.K. contributed equally to this work.
M.R. received fees consulting, lecturing, or serving on advisory boards from Astra Zeneca, Boehringer Ingelheim, Echosens, Eli Lilly, Merck-MSD, Madrigal, Novo Nordisk, Synergy, and Target RWE and has performed investigator-initiated research with support from Boehringer Ingelheim and Novo Nordisk to the German Diabetes Center (DDZ). No conflicts of interest, financial or otherwise, are declared by the other authors.
This paper contains supplemental material available at: https://doi.org/10.1523/JNEUROSCI.1067-25.2025
- Correspondence should be addressed to José Dören at doeren{at}hhu.de.
