Highly demarcated structural alterations in the brain and impaired social incentive learning in *Tbx1* heterozygous mice

A prerequisite for precision psychiatry is identifying genetic risk factors, brain alterations, and their behavioral dimensions. Ultra-rare single-gene variants and rare CNVs are strong risk factors for psychiatric disorders, yet how they produce symptomatic elements remains unclear. Brain regional volume changes are reported across ASD, schizophrenia, ADHD, and in CNV-linked illnesses, with 22q11.2 CNV particularly well studied. Individuals with 22q11.2 deletions show altered cortical thickness and surface area and subcortical volume changes, including decreased hippocampus, thalamus, putamen, and amygdala volumes, and increased caudate and nucleus accumbens. Mouse models of 22q11.2 deletion also show amygdala and striatal subregion alterations. Recent sequencing implicates ultra-rare variants in 22q11.2-encoded genes (e.g., TBX1, CLDN5, HIRA) in idiopathic schizophrenia and ASD, and TBX1 variants have been reported in cases with ASD/intellectual disability. However, such human genetic studies are limited by rarity and co-occurring variants, and mouse models of whole 22q11.2 deletion cannot isolate single-gene effects. Single-gene manipulations and smaller CNVs in mice provide alternative approaches. Altered Tbx1 dosage impairs social interaction/communication, memory, and cognitive flexibility in mice and affects fimbria myelination and cognitive speed, paralleling findings in 22q11.2 carriers. Based on this evidence, the study explored volumetric alterations and related behavioral dimensions in congenic constitutive Tbx1 heterozygous mice using whole-brain volumetric MRI, focusing on how Tbx1 heterozygosity affects regional brain volumes and amygdala-dependent social incentive learning.

Prior work shows widespread structural brain alterations in 22q11.2 deletion carriers (cortical thickness and surface area changes; decreased volumes in amygdala, hippocampus, thalamus, putamen; increased caudate and nucleus accumbens). Mouse models of 22q11.2 deletion demonstrate focal amygdala reductions in voxel-wise analyses and mixed cortical alterations. Large-scale human sequencing links ultra-rare variants in TBX1 and other 22q11.2 genes to schizophrenia/ASD risk, and TBX1 variants have been found in cases with ASD/intellectual disability/developmental delay. Tbx1 dosage alterations in mice impact social behavior, neonatal vocal communication, spatial/working memory, cognitive flexibility, and white matter (fimbria) myelination, aligning with cognitive slowing and white matter changes in 22q11.2 deletion syndrome. These studies motivate examining TBX1’s contribution to specific regional brain volumes and related behaviors.

Ethics: All animal procedures were approved by IACUCs at Albert Einstein College of Medicine (A3312-01), UT Health San Antonio (3345-01; 20190084AR), and Azabu University (#180316-6), following NIH guidelines. Animals and genotyping: Constitutive Tbx1 heterozygous (HT, ±) and wild-type (WT, +/+) mice on a congenic C57BL/6J background were used; Tbx1 −/− is perinatally lethal. Original non-congenic Tbx1 HT mice were backcrossed >10 generations to C57BL/6J. Genotypes were determined by PCR with specific primers for WT and HT alleles. Both sexes were used across experiments; specific sample sizes per experiment are provided below. Mice were randomly assigned; experimenters were blinded to genotype. Volumetric MRI: Ex vivo MRI was used for higher accuracy/reproducibility. Whole-brain and regional volumes were quantified. Primary atlas-based analysis included 19 regions based on Ma et al. (2005). Voxel-based analyses within the amygdala assessed focal differences with permutation-based nonparametric testing (500 permutations), small-volume correction within amygdala ROI (p < 0.05). Cortical segmentation followed Ullmann et al. (2013) to analyze 72 cortical/related regions relative to whole-brain volume, including hemisphere-specific analyses with Benjamini–Hochberg FDR thresholds (5–25%). Female cohort for MRI: WT n = 11 from 6 litters; HT n = 7 from 5 litters. Immunohistochemistry: To validate MRI findings in the amygdalopiriform transition area (APir), sections were immunostained for calretinin (marker enriched in subsets of GABAergic neuropil). The calretinin-positive neuropil area within APir (−2.46 to −3.64 mm from bregma) was quantified. Female WT n = 5 (4 litters); HT n = 4 (4 litters). Behavior: Social incentive conditioning (place conditioning). Male mice were pair-housed with age-matched male littermates (same or different genotype) from weaning to testing (5–7 weeks). Apparatus had two compartments with distinct proximal cues (cob vs paper bedding). Preconditioning test (Day 1, 30 min); conditioning Day 2 (24 h, with home-cage partner in cob bedding compartment); conditioning Day 3 (24 h, no partner in paper bedding compartment); post-conditioning test Day 4 (30 min, no partner). Preference shifts were assessed within subjects (paired t-tests) and across groups with ΔT difference scores. Control experiments counterbalanced conditioning to paper bedding and included no-partner conditioning to isolate effects of social vs non-social cues. Group labels: WT/WT/WT, WT/HT/HT, HT/HT/HT, HT/WT/WT. Olfactory social cue test: Sniffing time and habituation to urine odor from age-matched C57BL/6J non-littermates were measured to assess detection of social olfactory cues. Sensorimotor gating: Acoustic startle and prepulse inhibition (PPI) assessed with acoustic pulses (75–120 dB) and prepulses at 4–24 dB above 65 dB background. Non-acoustic PPI measured with visual (light) prepulse and tactile (air puff) pulse. Startle magnitude, PPI percentage, and habituation across blocks were analyzed. Acoustic cohort: male WT n = 20 (15 litters), HT n = 28 (17 litters). Non-acoustic cohort: male WT n = 29 (21 litters), HT n = 42 (25 litters). Statistics: Analyses performed in SPSS v28 and SigmaPlot 12.5. ANOVA for >2 groups; Student’s unpaired/paired t-tests for two-group comparisons. Assumptions tested via Shapiro–Wilk (normality) and Levene’s (homogeneity); if violated, Mann–Whitney (unpaired) or Wilcoxon (paired) nonparametric tests used. Multiple comparisons controlled with Benjamini–Hochberg FDR (typically 5%). Effect sizes reported as Cohen’s D. Detailed statistics are in Supplementary Table S1; MRI region values in Supplementary Tables S2–S5.

- Whole brain: No difference in total brain volume between HT and WT. - Atlas-based subcortical analysis (19 regions): Amygdala showed the largest reduction (Cohen’s D = −1.185) as % of whole brain (U = 64, p = 0.02) but did not survive correction when all regions tested together. Increases observed in fimbria (D = 0.952), corpus callosum (D = 0.882), globus pallidus (D = 0.535), internal capsule (D = 0.489), striatum (D = 0.409), thalamus (D = 0.343), superior colliculus (D = 0.240); decreases in inferior colliculus (D = −0.450), anterior commissure (D = −0.346), basal forebrain/septum (D = −0.291). - Amygdala voxel-wise: Significant focal volume reductions in anterior and posterior right amygdala after permutation correction within amygdala ROI (p < 0.05); 409 voxels (0.21 mm³) significant. Hemisphere-specific atlas measures did not survive FDR correction (left p = 0.082, D = −0.896; right p = 0.014, D = −1.333 before correction). - Cortical segmentation (72 regions): As a set, 10 regions differed significantly after BH correction (FDR 5%). Increases: temporal association area (TeA), ventral secondary auditory cortex (AuV), primary auditory cortex (Au1). Decreases: ventral intermediate entorhinal cortex (VIEnt), amygdalopiriform transition area (APir), secondary motor cortex (M2), primary somatosensory jaw region (S1J), medial entorhinal cortex (MEnt), posteromedial cortical amygdaloid area (PMCo), cortex–amygdala transition zone (CxA). Volume reductions clustered around/within amygdala boundaries. Hemisphere-specific cortical effects only emerged at lenient FDR 25% (e.g., right VIEnt/APir/MEnt/CxA decreases; left Au1 and binocular visual cortex increases). - Immunohistochemistry (APir): Calretinin-positive neuropil area was reduced in HT vs WT (t(7) = 3.586, p = 0.009), validating MRI-detected reduction in posterior amygdala region. - Social incentive conditioning: In cob-conditioned protocol, only WT mice conditioned with WT partners (WT/WT/WT) showed a significant post-conditioning preference shift for the cob compartment (paired t(7) = −4.143, p = 0.004; survived BH FDR 5%). ΔT difference scores: WT/WT/WT differed from HT/HT/HT (t(16) = 3.070, p = 0.007) and from HT/WT/WT (t(13) = 2.756, p = 0.016), both significant after BH correction. WT mice failed to show preference when conditioned with HT partners. HT mice showed no preference shift in cob-conditioned protocol regardless of partner genotype. Paper-conditioned controls and no-partner controls indicated that apparent HT preference shifts were driven by non-social environmental cues (paper vs cob), not genuine social conditioning; WT mice were sensitive to partner genotype. - Olfactory social cue: WT and HT mice did not differ in sniffing time or habituation to urine odor (no impairment in detecting social olfactory cues). - Acoustic startle/PPI: HT mice had reduced acoustic startle at high intensities (110 dB p = 0.001; 115 dB p < 0.001; 120 dB p = 0.004; Mann–Whitney; all significant after BH FDR 5%). Acoustic PPI was reduced in HT (Genotype F(1,46) = 11.19, p = 0.002; Genotype×Prepulse F(5,230) = 3.556, p = 0.004). Post hoc: HT lower than WT at prepulses 8 dB (p < 0.001), 12 dB (p = 0.018), 16 dB (p = 0.032), 20 dB (p = 0.003), 24 dB (p < 0.001). Habituation to 120 dB startle did not differ (Mann–Whitney U = 285.5, p = 0.908). - Non-acoustic startle/PPI: No genotype differences for air-puff startle (U = 525, p = 0.326), light-induced PPI (t(69) = 0.954, p = 0.343), or habituation (t(69) = −0.609, p = 0.544).

The study isolates TBX1’s contribution to brain structure and behavior within the 22q11.2 CNV context. Constitutive Tbx1 heterozygosity led to highly focal reductions in anterior/posterior amygdala and surrounding cortical regions and to impaired amygdala-dependent social incentive learning. Concurrently, HT mice exhibited increased primary/secondary auditory cortical volumes and selective deficits in acoustic (but not non-acoustic) startle and PPI. These results align with structural/functional amygdala abnormalities and sensorimotor gating deficits reported in 22q11.2 deletion carriers and mouse models, implicating TBX1 as a driver for particular phenotypic dimensions. Not all human 22q11.2-associated volumetric changes (e.g., hippocampus, thalamus, cerebellum) appeared in Tbx1 HT mice, indicating other 22q11.2 genes likely underlie those alterations. The findings also highlight limitations of atlas-based region analysis for detecting focal gene effects; voxel-wise and segmental analyses captured demarcated changes that cross conventional anatomical boundaries. Potential peripheral contributions (e.g., TBX1-associated hearing loss) may relate to auditory cortical volume and acoustic PPI phenotypes. Overall, Tbx1 deficiency affects specific circuits (amygdala-related social learning; auditory network-related acoustic gating) rather than causing global brain alterations.

This work identifies focal amygdala and peri-amygdalar cortical volume reductions and increased auditory cortical volumes as key neuroanatomical correlates of Tbx1 heterozygosity, linking them to impaired social incentive learning and selective acoustic sensorimotor gating deficits. The results support TBX1 as a driver gene for specific structural and behavioral phenotypes observed in 22q11.2 CNV carriers and underscore the need for analyses that detect focal, cross-boundary anatomical changes. Future research should: (1) delineate causal developmental mechanisms (embryonic neurogenesis vs post-embryonic synaptogenesis/myelination) underlying these volumetric changes; (2) parse contributions of peripheral (hearing) versus central changes to acoustic PPI deficits; (3) examine interactions with other 22q11.2 genes that may influence additional brain regions; and (4) test timing-specific interventions during relevant developmental windows to ameliorate social incentive learning deficits in TBX1 deficiency.

- Sex effects: While no clear sex differences are reported for ASD prevalence, social cognition, or key brain volumes in 22q11.2 carriers, underlying molecular/cellular mechanisms may differ by sex (latent sex differences). The current study did not dissect sex-specific brain mechanisms. - Statistical power and multiple comparisons: Atlas-based regional analyses may miss focal effects; many regions showed medium-to-small effect sizes without surviving correction. Larger samples could reveal additional significant alterations. - Non-acoustic PPI parametrization: Only single light prepulse and single air-puff intensities were used; potential genotype differences at other parametric points could have been missed, although acoustic PPI deficits with comparable PPI levels in non-acoustic tests argue against a general gating deficit. - Peripheral contributions: TBX1-associated hearing impairments in mice/humans could contribute to acoustic PPI and auditory cortex volume findings; causality between peripheral and central changes was not resolved.