Tbx1 heterozygosity in the oligodendrocyte lineage shifts myelinated axon composition in the mouse fimbria without behavioral impairments Authors: Anne Marie Wells, Takaki Tanifuji, Takeshi Takano, Arumu Endo, Gina Kang, Marisa Esparza, Qian Shi, Manzoor A. Bhat, Noboru Hiroi Published: 2026 ABSTRACT Constitutive heterozygosity of Tbx1, a T-box transcription factor gene located within the 22q11.2 deletion region, results in behavioral deficits and altered composition of myelinated axons in the fimbria, together with reduced levels of an oligodendrocyte precursor cell (OPC) marker, in mice. However, the cellular origins of these effects and the extent to which axonal changes causally contribute to behavioral impairments remain unclear. We hypothesized that Tbx1 deficiency specifically within the oligodendrocyte lineage contributes to myelin and behavioral phenotypes. To test this hypothesis, we first demonstrated through in vitro siRNA knockdown that Tbx1 regulates both OPCs and mature oligodendrocytes. Subsequently, we assessed the impact of Tbx1 heterozygosity initiated in OPCs on behavioral and myelin phenotypes in male conditional PdgfrαCre;Tbx1+/flox mice. These mice exhibited Cre-mediated recombination in Pdgfrα-expressing brain regions and in the OPC progeny within the fimbria. At one month of age, the mutants displayed a higher rate of spontaneous alternation at the longest inter-trial interval in the T-maze compared to their wild-type littermates—an effect that was dissipated at two months. No significant phenotypic abnormalities were observed in conditional PdgfrαCre;Tbx1+/flox mice regarding neonatal ultrasonic vocalizations, social interaction, novel object approach, anxiety-like behavior (elevated plus maze), or open-field locomotion and thigmotaxis. Electron microscopic analysis revealed a compositional shift in myelinated axons within the fimbria of adult male mutants, characterized by an increased number of myelinated axons in the 300–800 nm diameter range and a decreased number in the ~ 1,200 nm and ~ 1,400 nm ranges, with myelin thickness remaining unchanged across diameters. These findings indicate that Tbx1 heterozygosity in the oligodendrocyte lineage leads to a selective shift towards smaller myelinated axons in the fimbria and a transiently higher level of capacity for working memory and cognitive flexibility. However, it does not replicate the full spectrum of myelination abnormalities or the broader cognitive and social deficits observed in constitutive Tbx1 heterozygotes, suggesting that Tbx1 deficiency in non-oligodendrocyte lineage cells may lead to altered myelination and neurodevelopmental behavioral impairments.Supplementary InformationThe online version contains supplementary material available at 10.1186/s13041-026-01311-4. Introduction Cognitive and social impairments are predictors of mental illness [1, 2]. Infants later diagnosed with autism spectrum disorder (ASD) exhibit delays and deviations in various aspects of motor, social, and language development [3–11]. Similarly, children who subsequently develop full-scale symptoms of schizophrenia display delayed development of social cognition, working memory, attention, and processing speed [12–16], with these deficits remaining integral to the disorder following its onset [17, 18]. These dimensional deficits in mental illness have genetic underpinnings. Copy number variations (CNVs)–which consist of deletions or duplications of up to several million base pairs at specific chromosomal loci encompassing numerous protein-coding genes—are associated with high odds ratios and penetrance rates for cognitive and social deficits [19, 20] as well as mental illnesses [21–23]. Individuals with hemizygous deletions of human chromosome 22q11.2, ranging from 1.5 Mb to 3.0 Mb, exhibit cognitive, social, and emotional impairments from childhood [24–26] and are diagnosed at elevated rates with anxiety disorders, attention-deficit hyperactivity disorder (ADHD), ASD, intellectual disability (ID), and schizophrenia [27, 28]. Duplication of 22q11.2 is linked to epilepsy, ID, ADHD, and ASD at rates higher than those observed in non-carriers [29–45]. However, the mechanisms through which over 40 protein-coding genes in 22q11.2 contribute to dimensional deficits and clinical diagnoses remain poorly understood in humans. Rare inherited cases of variants in the human TBX1 gene, a T-box transcription factor located within the 22q11.2 CNV, are associated with ASD, schizophrenia, and ID in the absence of 22q11.2 CNVs [46–51]. However, due to their rarity, the statistical reliability of the association between TBX1 variants and mental illness is limited. Moreover, these patients often carry variants in other single genes throughout the genome [49], complicating the establishment of causality from these associations. The deletion or overexpression of small murine chromosomal regions orthologous to human chromosomes or single genes provides a complementary approach [1, 52–56]. Constitutive overexpression of several hundred kilobase pair segments of the mouse ortholog of 22q11.2 recapitulates distinct social and cognitive deficits associated with 22q11.2 duplication [52, 57]. Larger hemizygous deletions that encompass one of these segments result in deficits in prepulse inhibition [50]. These studies in mouse models suggest the presence of driver genes for specific behavioral phenotypes within these segments. The Tbx1 gene is located within one of these small segments. The heterozygous deletion of Tbx1 alone results in impairments in neonatal social communication [58–60], as well as post-pubertal and adult deficits in social interaction [58], social incentive learning [61], acoustic, but not non-acoustic, prepulse inhibition [50, 61], spatial memory acquisition in the Morris water maze, and the speed to complete simple discrimination and reversal phases of attentional set-shifting, without impairing non-specific motor speed [62]. The fimbria of post-pubertal Tbx1 heterozygous mice displays a lack of large myelinated axons, enhanced myelination of medium-sized axons, a compositional shift of myelinated axons toward smaller sizes, and a gene expression profile indicative of defective oligodendrocyte precursor cells [62]. Furthermore, the volume of the secondary motor cortex and amygdala, regions involved in the vocal network [63], is reduced in post-pubertal Tbx1 heterozygous mice [61]. Additionally, virally induced overexpression of Tbx1 in the hippocampus negatively impacts the developmental maturation of working memory capacity [64]. The cellular and developmental origins of the effects of Tbx1 deficiency on behavioral and myelin phenotypes remain insufficiently elucidated; nonetheless, cells expressing Tbx1 represent potential candidate substrates. Recent single-cell transcriptomic analyses have demonstrated that Tbx1 is detectable in a broader range of cell types in the mouse than previously indicated by in situ hybridization and immunocytochemical studies. Specifically, Tbx1 is detectable in the neural tube, as well as in endothelial cells, radial glia (or neural progenitor cells), oligodendrocyte progenitor cells, oligodendrocytes, and various types of immature and mature neurons [65–67]. These findings imply that Tbx1 is expressed in radial glial cells and neural progenitor cells, along with their progeny, which encompass oligodendrocyte and neuronal lineages. During the embryonic period, oligodendrocyte precursor cells originate from embryonic neural progenitor cells in the medial ganglionic eminence of the ventricular zones around embryonic day 12.5 (E12.5), followed by a secondary wave from the lateral ganglionic eminence at E15.5. Oligodendrocyte precursor cells continue to be generated from radial glia and neural progenitor cells in the subventricular zone of the brain during the neonatal period [68–72]. Within the embryonic and neonatal mouse brain, Tbx1 is present in cells exhibiting gene expression profiles characteristic of endothelial cells, radial glia, and both immature and mature neurons [66, 67]. Our previous work demonstrated that Tbx1 heterozygosity initiated in nestin-positive cells, including radial glia and neural progenitor cells, during the neonatal period recapitulates some behavioral deficits associated with constitutive Tbx1 heterozygosity [73]. As neural progenitor cells during the neonatal period give rise to oligodendrocyte precursor cells, oligodendrocytes, neural progenitor cells, and neurons [68–72], we sought to determine whether the effects of Tbx1 heterozygosity within the oligodendrocyte lineage induces behavioral and anatomical phenotypes. Our findings indicate that Tbx1 heterozygosity in the oligodendrocyte lineage induces a compositional shift in myelinated axons in the fimbria, while not altering myelin integrity or large myelinated axons, or impairing behavior. Methods Mice All protocols for animal handling and use were approved by the Institutional Animal Care and Use Committee (IACUC) of the University of Texas Health Science Center at San Antonio (UTHSCSA) in accordance with National Institutes of Health (NIH) guidelines. Mice were housed in a vivarium with a standard light phase from 7 am to 9 pm). C57BL/6J mice: We used 1- and 2-month-old male C57BL/6J mice (Jax #000664, Jackson Laboratory, Bar Harbor, ME) as stimulus mice for social interaction. PdgfraCre;Tbx1+/flox mice: To conditionally induce Tbx1 heterozygosity in oligodendrocyte precursor cells, we chose a PdgfraCre line that initiates Cre synthesis earlier than a Cspg4Cre line to better induce recombination in the oligodendrocyte lineage [74]. The Pdgfra- promoter drives Cre-based recombination most robustly in oligodendrocyte precursor cells [75], but also in astrocytes, neurons, fibroblast-like cells, endothelial cells, vascular and leptomeningeal cells, and pericytes in the mouse brain [75–78]. We crossed hemizygous PdgfraCre mice with Tbx1+/flox mice to generate PdgfraCre;Tbx1+/flox mice. PdgfrαCre breeder mice (C57BL/6-Tg(Pdgfra-Cre)1Clc/J; Jax #013148, Jackson Laboratory, Bar Harbor, ME) have a mixed C57BL/6J;C57BL/6N background. We backcrossed non-congenic Tbx1+/flox mice [79] to C57BL/6J for more than 10 generations and confirmed that there was Cre-dependent recombination in a congenic Tbx1+/flox mice [73]. The genetic background of PdgfraCre and Tbx1+/flox lines were mixed C57BL/6J;C57BL/6N and C57BL/6J, respectively. These two C57BL/6 substrains have many behavioral differences [80]. To avoid the confounding effects of their unequal genetic backgrounds [81], we used only an F1 generation derived from crosses of the PdgfraCre and Tbx1+/flox lines. While both male and female mice were born with all genotypes, only half as many females as males survived to complete analyses up to 2 months of age. Because the sample sizes of females tested were not sufficient for statistical reliability, our analyses were based on male mice only. No sex differences are observed in social or cognitive dimensions among carriers of 22q11.2 deletion in humans [24]. PdgfraCre;ROSA-tdTomato mice: We crossed hemizygous PdgfraCre mice (C57BL/6-Tg(PdgfraCre)1Clc/J; Jax #013148, Jackson Laboratory, Bar Harbor, ME) with homozygous congenic ROSA-CAG-tdTomato mice (B6.Cg-Gt(ROSA)26Sortm14(CAG−tdTomato)Hze/J, Jax # 007914, Jackson Lab, Bar Harbor, ME) to generate PdgfraCre;ROSA-CAG-tdTomato mice. This ROSA line expresses tdTomato in the hippocampus without the action of Cre ( http://connectivity.brain-map.org/transgenic/experiment/81560256). We determined the genotypes of the mice using our previously published protocol with primers shown in Table S1. The sex of the mice was determined by inspection of their external genitalia. In vitro cell culture of oligodendrocytes Lateral ventricle tissues, including the subventricular zone (SVZ), were obtained from postnatal day 1–2 (P1-2) C57BL/6J pups. Cells were isolated in culture, with each culture derived from a single mouse. The cells were cultured in a medium (DMEM/F12, HEPES [11320–032, Gibco, Grand Island, NY, USA]) supplemented with N2 (17,502,048, Gibco, Grand Island, NY, USA), B27 (17,504,044, Gibco, Grand Island, NY, USA), epidermal growth factor (EGF) at a concentration of 20 ng/ml (AF100-15, Peprotech, Cranbury, NJ, USA). After two to four passages, the cells were dissociated from the spheres using StemPro Accutase Cell Dissociation Reagent (A1110501, Gibco, Grand Island, NY, USA). The cells were then divided into equal portions and plated on each Poly-L-ornithine (P4957, Sigma, St. Louis, MO, USA) and Bovine Fibronectin (1030-FN, R&D Systems, Minneapolis, MN, USA)-coated 6well plate (Corning, 351,146, Corning, NY, USA,). The cells were cultured for 144 h in medium with 5% fetal bovine serum after withdrawing EGF from the medium to induce and maintain differentiation. At the time of EGF withdrawal, we applied Tbx1 siRNA (10 nM, cat#4,390,771, s74767, Invitrogen; Thermo Fisher Scientific, Waltham, MA, USA) and control siRNA-A (10 nM, cat#4,390,843, Invitrogen; Thermo Fisher Scientific, Waltham, MA, USA) together with siLentFect™ Lipid Reagent for RNAi, 0.5 ml (BIO-RAD, cat# 1,703,360, Hercules, CA, USA). Quantitative reverse transcription polymerase chain reaction (qRT-PCR) In accordance with our published procedure [62], total RNA was extracted using an RNeasy Plus Mini Kit (Cat#74,134, Qiagen, Germantown, USA). Complementary DNA (cDNA) was synthesized from total RNA using SuperScript IV VILO Master Mix (Cat#11,766,050, Invitrogen, Carlsbad, USA). qRT-PCR reactions were performed in triplicate on a QuantStudio 6 Flex Real-Time PCR System (Cat#4,485,694, Applied Biosystems, Waltham, USA) using the TaqMan Fast Advanced Master Mix (Cat#4,444,963, Applied Biosystems, Waltham, USA). We used TaqMan® Gene Expression Assays (cat# 4,331,182, Thermo Fisher, Waltham, MA, USA). The Taqman probes are listed in the Supplementary Material (Table S2). Data were analyzed using the ΔΔCt method and normalized to the reference gene Pgk1; we had an identical pattern of results with another reference gene 18S. Whole brain tissue clearing and index matching We used the SHIELD and SmartBatch + (LifeCanvas Sciences, Cambridge, MA) standard pipelines for preserving and electrophoretic clearing [82] to visualize tdTomato in the brains of PdgfraCre;ROSA-CAG-tdTomato mice. We performed transcardial perfusion of 1-month old mice with ice-cold 0.9% NaCl in dH2O, followed by 4% paraformaldehyde (PFA) in phosphate-buffered saline (PBS). Brains were preserved using SHIELD as follows: perfused brains were 1) removed from the cranium and fixed in 4% PFA in PBS overnight at 4 °C with gentle shaking; 2) incubated in SHIELD OFF solution (dH2O + SHIELD Buffer [cat#SH-Bf] + SHIELD Epoxy Buffer [cat#SH-Ex]) solution at 4 °C with gentle shaking for 3 days; 3) incubated in pre-warmed SHIELD ON Buffer (cat#SH-ON) at 37 °C for 1 day with gentle shaking; 4) incubated in the delipidation buffer (cat#DB) for 3 days at 45 °C with gentle shaking. Final clearing by electrophoresis was performed with the SmartBatch + device, with the delipidation buffer inside the clearing cup containing the samples and conduction buffer (cat#CB) for 30 h. Cleared brains were index-matched using a solution of 50% EasyIndex (RI = 1.52, cat#EI-500–1.52) + 50% dH2O with shaking at 37 °C for 1 day, then 100% EasyIndex for one additional day until brains were transparent [82]. Light sheet microscopy We used a Zeiss Lightsheet 7 Microscope housed in the UTHSCSA Optical Imaging Core to perform volumetric light sheet imaging of cleared PdgfraCre;ROSA-CAG-tdTomato brains using a pair of 5 × /0.1 focusing illumination objectives and a Fluar 2.5 × /0.12 detection objective at identical settings. Image tiles were stitched together using Zeiss Zen Blue software (ver 3.4). 3D volume analysis was performed with Imaris (Oxford Instruments, ver 10.02). Peak intensity of the tdTomato signal was segmented, and volumetric measurements (mm3) and peak intensities of the total volume (arbitrary units) were calculated for each volume in each mouse. Immunofluorescence PdgfrαCre;ROSA-tdTomato mice were sacrificed at 1 month of age. Mice were deeply anesthetized using 4–5% isoflurane and perfused transcardially with 0.9% saline followed by 4% PFA. Brains were extracted and post-fixed in 4% PFA overnight at 4 °C, followed by cryoprotection in glycerol overnight at 4 °C. Coronal 60-μm thick sections were cut with a freezing microtome. For immunofluorescence labeling, sections were washed in 0.1 M PBS, blocked with 5% normal donkey serum in PBS, and incubated overnight at room temperature, as we described previously [62, 73]. We used a well-validated primary rabbit antibody against myelin basic protein (1:200, MBP, ab40390, Abcam) and Donkey anti-rabbit IgG-Alexa Fluor 488 (1:200, A21206, Invitrogen), together with a nuclear marker (DAPI,1:25,000, D3571, Invitrogen). MBP is a marker of mature oligodendrocytes [83, 84]. We captured images of sections that were mounted and cover-slipped with Vectashield® anti-fade mounting medium (H-1000) using a Keyence microscope (BZ-X800) microscope with BZX Hardware Module and Zeiss LSM 710 confocal microscope at the UT Health Optical Imaging Core. Double staining of tdTomato with markers of oligodendrocytes We captured a series of coronal brain sections from PdgfrαCre;ROSA-tdTomato mice using a Keyence BZ-X800 microscope with BZX Hardware Module and a Zeiss LSM 710 confocal microscope at the UT Health Optical Imaging Core. We used the DAPI signal to identify the anatomical boundaries of landmark structures in each image, cross-matched with equivalent levels in the Allen Brain Atlas (Allen Reference Atlas – Mouse Brain. Available from atlas.brain-map.org). We then used a color scale to represent the level of tdTomato intensity within the delineated anatomical boundaries. Behavioral analysis Male neonatal mice were tested for vocalization resulting from maternal separation at P8 and P12, followed by additional behavioral tests at 1 and 2 months of age, corresponding to peripubertal and post-pubertal, adolescent periods [85], respectively; early signs of puberty begin around 1 month of age [86]. Because body weight can alter behavior in mice [87] and Pdgfrα regulates progenitor differentiation into adipocytes [88, 89], we measured body weight at the beginning of behavioral tests. Our behavioral battery included reciprocal social interaction, novel object approach, spontaneous alternation in a T-maze, elevated plus maze, and locomotor activity and thigmotaxis in an inescapable open field [52, 57–59, 64, 90–92]. The order of the behavioral assays was based on the stress level; behaviors that occur in a home cage-like setting was given first (i.e., social interaction and novel object approach). T-maze and elevated plus maze tests permit choices and are thus considered less stressful than an inescapable open field. At least a one-day interval was provided between behavioral tests to reduce the likelihood of carryover effects [61, 62, 93], except for the T-maze, in which three different delays were given on three consecutive days. Mice were assigned randomly to experimental groups and tested during the light phase. Experimenters were blinded to genotypes. Social interaction, novel object approach, and behaviors in the elevated plus maze were recorded using a Basler GigE camera, with video images stored in Ethovision v18 software (Noldus Information Technology, Leesburg, VA). Video images of social interaction and novel object approach were manually scored according to our established criteria. Various parameters of the elevated plus maze were automatically analyzed by Ethovision. The time point of each arm entry was recorded for spontaneous alternation, and the data were subsequently tallied manually. Locomotor activity and thigmotaxis in an inescapable open field were analyzed using Med Associates Activity Monitor 7 Software (Fairfax, VT). Ultrasonic vocalization Pups were tested for vocalization induced by maternal separation at postnatal days 8 and 12. A cage containing the mother and a litter was transferred to the test room 30 min before testing. Ultrasonic vocalization was recorded for 5 min by UltraSoundGate (Avisoft, Germany) connected to a computer equipped with Avisoft-RECORDER software (Avisoft, Germany) in a test chamber (22 cm long x 16.5 cm wide x 16.5 cm high). The sampling rate and lower cutoff frequency were set at 250 kHz (format, 16-bit) and 10 kHz, respectively. A frequency window from 15–150 kHz was used for analysis. Call detection was provided by an automatic threshold-based algorithm and a hold-time mechanism (hold time = 10 ms). Sonograms were inspected, and mechanical noises were eliminated from the analysis. We used the VocalMat software, which has the lowest false positive and false negative rates [94], to determine call types. This software typically detects only the most salient components of the harmonic call type and classifies this call type differently. As the harmonic call type is often affected in genetic mouse models of neuropsychiatric disorders [58, 59, 87, 95], we manually inspected all call types and re-classified such cases as harmonic calls. Reciprocal Social Interaction Test A test mouse and an age-matched unfamiliar C57BL/6J mouse as a stimulus subject were simultaneously placed in a cage (28.5 cm long × 17.5 cm wide × 12.5 cm high). Mouse behavior was recorded during two 5-min sessions with a 30-min interval between sessions. Video images were scored manually for affiliative and aggressive social interaction by a rater blinded to genotype. The following reciprocal social behaviors were scored by the duration of interaction (minimum of 1 s): aggressive (tail rattle, bite/kicks, sideway offense, boxing/wrestling), affiliative, non-aggressive (mount, pursuit, olfactory investigation, allogrooming), or passive (side-by-side, submissive, escape/leap). We analyzed the sum of time engaged in affiliative, active, and passive social interactions, but passive behavior was rarely seen in our experimental set-up [73, 87, 92, 95]. Novel Object Approach A test mouse was placed in a cage (28.5 cm long × 17.5 cm wide × 12.5 cm high) with an empty, non-secured 50-mL Falcon tube (3 cm diameter × 8.5 cm long) lying on its side. Videos of mouse interactions with the tube were used to manually score approach behavior by two independent analysts blinded to genotype with an inter-rater reliability of > 99%. Approach behavior was scored for olfactory or non-olfactory investigations with a minimum duration of 1 s. Time spent near the tube was also analyzed. Spontaneous Alternation in a T-Maze A test mouse was placed in the start site of the long arm of a black plexiglass T-maze (33 cm long × 10.5 cm wide × 20.5 cm high) and allowed to enter the maze and explore either the left or right short arms (31 cm long × 10.25 cm wide × 20.5 cm high) of the cage. We imposed 0 s, 15 s, or 30 s interval delays to maze reentry for 10 trials, and the three interval trials were given on three consecutive days. We manually scored the percentages of alternation and latency to reach the alternated and non-alternated arms. Elevated Plus Maze Test Mice were placed in the center stage (5 × 5 cm) of the maze apparatus with four arms (30 × 5 cm) and permitted to explore two closed arms and two open arms extending from the center platform of the maze for 5 min. The maze was positioned 53 cm above the floor. The time spent in visits and the frequency of visits to the two open and two closed arms were analyzed. Open Field Mice were placed in an open-field cage (27.3 cm × 27.3 cm × 20.2 cm; Med Associates, Fairfax, VT) and permitted to explore the cage for 30 min. We measured the distance and velocity of motor activity and duration of time spent in the center (19.05 cm × 19.05 cm central square) versus the margin of an open field, using the activity monitor software (Med Associates, Fairfax, VT). Electron Microscopy (EM) We prepared tissue for EM to characterize myelination in the fimbria of PdgfrαCre;Tbx1+/flox mice as we described previously [62]. After behavioral assays were completed at 1 and 2 months of age, 4–5-month-old male PdgfrαCre;Tbx1+/flox (N = 3) and five control mice, including PdgfrαCre;Tbx1+/+ (N = 2), wild-type (WT);Tbx1+/flox (N = 2), and WT;Tbx1+/+ (N = 1), were anesthetized with 4.5% isoflurane in a chamber, and anesthesia was maintained by a nose cone with 2.0% isoflurane. The animals were transcardially perfused with 120 mL of 0.9% saline followed by fixation with 120 mL of 0.1 M buffer (pH 7.4; cat#11,653 Electron Microscopy Science, Hatfield, PA) with 2.5% sodium glutaraldehyde (cat#16,310, Electron Microscopy Sciences, Hatfield, PA) and 2.5% paraformaldehyde (cat#19,202, Electron Microscopy Sciences, Hatfield, PA). Brains were extracted and post-fixed in the same solution at 4 °C for 7 weeks. The fimbria from the two hemispheres were obtained separately using a vibratome and placed in 0.1 M sodium cacodylate buffer overnight. Tissues were rinsed with 0.1 M sodium cacodylate buffer to remove aldehydes and placed in 2% osmium tetroxide solution (OsO4; cat#19,150, Electron Microscopy Sciences, Hatfield, PA) in 0.1 M sodium cacodylate buffer for 1 h. Tissues were dehydrated in a series of ethanol solutions and embedded in molds containing Polybed resin (Poly/Bed® 812 Embedding media, cat#08791–500 Polysciences, Inc., Warrington, PA). We cut 100-nm sections and collected them on square 150-mesh copper grids (cat#7551C, Polysciences, Inc., Warrington, PA) and stained them with uranyl acetate (7 g in 100 mL deionized water)/Reynold’s lead citrate (1.33 g lead nitrate,1.76 g sodium citrate in 30 mL triple-distilled water, 8 mL 1 N NaOH). Uranyl acetate stains membranous structures and structures containing nucleic acid; lead citrate binds to RNA-containing structures and the hydroxyl groups of carbohydrates. Each fimbria section was viewed in the EM grid squares. Images were screened at 1,000 × magnification. For each section, we used all grid images that contained fimbria tissue with round axons without wrinkles, folds, or tears in the sections. One image was captured at the center of each grid field at 20,000 × magnification. Images were analyzed by MyelTracer [96]. Only axons whose myelin was fully within the image were analyzed quantitatively. We obtained multiple images for each animal from a section of the right and left hemispheres. The positions of sections sliced from the fimbria varied between the two hemispheres and did not necessarily match the two sides; in one case, only one hemisphere was available because of the accidental loss of a tissue slice. The number of images varied across tissue slices, resulting in unequal image numbers. Many axons from each grid square and many grid squares from each hemisphere were analyzed for each animal. As the number of images and positions of each image varied between the two hemispheres and did not match between the two sides, the hemisphere could not be used as a replicate. Thus, we pooled data for each animal as a technical replicate; data points within and across images were technical replicates. As there was considerable variability in the data, we could not average the values for each animal. The averages of the total axon diameter per genotype were also not appropriate for this data set, as axon numbers differed between genotypes at specific axon diameters (≥ 300 nm to < 400 nm; ≥ 1,200 nm to < 1300 nm) (genotype × axon diameter range, F(27,5880) = 6.336, P < 1.0 × 10−4). Thus, linear mixed models were used, as reported previously [62]. Data were assessed for the number of myelinated axons, the thickness of myelination, and the g-ratio along a 100 nm unit axon diameter. Data from different images and hemispheres were embedded in random models. Statistical analysis We used SPSS (v29.0.2.0 (20), IBM Corporation) to perform all statistical analyses. Among-group and between-group comparisons of the data were performed using analysis of variance (ANOVA) and Student’s two-tailed t-test (α = 0.05). We determined normality and homogeneity of variance using the Shapiro–Wilk test and Levene’s homogeneity of variance test, respectively If either assumption was violated, we analyzed the data using a linear mixed model, Kruskal Wallis tests, or Mann–Whitney U tests. The Greenhouse–Geisser correction was applied if sphericity was violated and the estimated epsilon was less than 0.75. If multiple tests were applied to a data set, the significance level was adjusted using the Benjamini–Hochberg correction, with a false discovery rate of 5%. We used GraphPad Prism software (v9; GraphPad Software, San Diego, CA) to generate all graphs. All statistical analyses are presented in Table S3. Results Effects of Tbx1 knockdown on markers of oligodendrocyte lineage in vitro We aimed to identify the earliest stage in the oligodendrocyte lineage affected by Tbx1 deficiency. To achieve this aim, we developed an in vitro screening assay utilizing neonatal neural progenitor cells derived from the subventricular zone of C57BL/6J pups sacrificed on postnatal day 2, which are capable of generating both neuronal precursor cells and oligodendrocyte precursor cells [72]. We assessed the expression of Cspg4, a marker for oligodendrocyte precursor cells, as well as Mag, Mbp, Mog, and Plp1, which are indicative of maturing and mature myelinating oligodendrocytes [68], at 144 h after EGF was withdrawn (i.e., differentiation) and Tbx1 siRNA was applied. The application of Tbx1 siRNA resulted in a significant reduction in the mRNA levels of Tbx1, Cspg4, and all markers associated with mature oligodendrocytes, with Pgk1 (Fig. 1) and 18S (Table S3–Fig. 1) as reference genes. These in vitro findings align with our in vivo data, which demonstrate that Cspg4 (also referred to as Ng2) was diminished in the fimbria of Tbx1 heterozygous mice [62].Fig. 1The relative quantification (RQ) (mean ± SEM) of expression levels at each stage of oligodendrogenesis, as determined by quantitative reverse transcription polymerase chain reaction (qRT-PCR) using Pgk1 as a reference gene. Three cell clones were derived from three different C57BL/6J pups at postnatal days 2 or 3, with data obtained from three culture wells for each clone, except for one Tbx1 siRNA-treated clone where Tbx1 mRNA was undetectable and thus excluded from the analysis. A linear mixed model showed that Tbx1 siRNA reduced expression of all genes tested equally (Treatment, F(1,89) = 14.665, P = 0.0002386; Gene, F(5,89) = 0.8440, P = 0.5222; Treatment x Gene, F(5,89) = 0.8440, P = 0.5222). Mann–Whitney tests confirmed that Tbx1 siRNA reduced expression of each gene. Statistically significant differences between control siRNA and Tbx1 siRNA groups are indicated by *, **, and *** for P < 0.05, P < 0.01, and P < 0.005, respectively; they remained significant following Benjamini–Hochberg correction at a 5% false discovery rate (FDR) The relative quantification (RQ) (mean ± SEM) of expression levels at each stage of oligodendrogenesis, as determined by quantitative reverse transcription polymerase chain reaction (qRT-PCR) using Pgk1 as a reference gene. Three cell clones were derived from three different C57BL/6J pups at postnatal days 2 or 3, with data obtained from three culture wells for each clone, except for one Tbx1 siRNA-treated clone where Tbx1 mRNA was undetectable and thus excluded from the analysis. A linear mixed model showed that Tbx1 siRNA reduced expression of all genes tested equally (Treatment, F(1,89) = 14.665, P = 0.0002386; Gene, F(5,89) = 0.8440, P = 0.5222; Treatment x Gene, F(5,89) = 0.8440, P = 0.5222). Mann–Whitney tests confirmed that Tbx1 siRNA reduced expression of each gene. Statistically significant differences between control siRNA and Tbx1 siRNA groups are indicated by *, **, and *** for P < 0.05, P < 0.01, and P < 0.005, respectively; they remained significant following Benjamini–Hochberg correction at a 5% false discovery rate (FDR) Given that our culture includes both neuronal [58, 73] and oligodendrocyte (see Fig. 1) lineages, as well as residual neural progenitor cells, the mechanism by which Tbx1 knockdown alters the expression of myelin marker genes remains unclear. It is possible that Tbx1 knockdown in oligodendrocyte precursor cells leads to a cell-autonomous reduction in myelin markers; alternatively, reduced Tbx1 levels in neonatal neural progenitor cells or their neuronal progeny may indirectly influence the expression of these oligodendrocyte markers. Localization of Recombination in Pdgfrα-Cre;ROSA-tdTomato Mice Our data indicate that Tbx1 deficiency begins to impact the stage of oligodendrocyte precursor cells (see Fig. 1). Consequently, we planned to initiate Tbx1 heterozygosity in oligodendrocyte precursor cells using a conditional heterozygous mouse model. This mouse employed Cre-based initiation of Tbx1 heterozygosity, guided by the promoter of Pdgfra, a gene expressed in oligodendrocyte precursor cells of both embryonic and post-embryonic origins. To identify the brain regions and cell types where Tbx1 recombination is anticipated, we utilized PdgfrαCre;ROSA-tdTomato mice as the best available proxy. Alternative methodologies are not applicable to Tbx1. Recent single-cell RNA sequencing analyses reveal that Tbx1 is present in proliferating cells, including neural progenitor cells, lung AT2 cells, spermatogonia, spermatids, and embryonic mesenchymal cells [65, 66]. However, although Tbx1 mRNA and protein products are expressed in actively proliferating neural progenitor cells, their levels significantly decrease in differentiated cells [73]. As a result, the levels of Tbx1 mRNA and protein products in our conditional Tbx1 heterozygous mice are expected to reflect reductions due to cell-cycle changes, recombination-induced heterozygosity, or a combination of both factors. Detecting TBX1 protein in the mouse brain presents challenges, with notable exceptions in endothelial cells [97] and neonatal neural progenitor cells [58, 73] partly due to epitope masking caused by fixation [58]. Additionally, approaches to localize TBX1 using reporters under the endogenous Tbx1 gene or its promoter have not yielded detectable signals in cells other than endothelial cells within the mouse brain [50, 97, 98], making it difficult to assess its colocalization with Pdgfrα protein. In theory, Pdgfra-positive and MBP-positive cells could be isolated via fluorescence-activated cell sorting (FACS). However, while Tbx1 is present in oligodendrocyte precursor cells and immature oligodendrocytes, its expression is notably low in mice [65, 66]. Evaluating further reduced Tbx1 levels is technically challenging due to the floor effect in heterozygous mice. We first assessed the extent of PdgfrαCre-mediated recombination throughout the brain, employing a tissue clearing technique (SmartBatch +) and utilizing lightsheet microscopy to visualize the 3D volume of tdTomato signals in the brains of PdgfrαCre;ROSA-tdTomato mice (Fig. 2A). The highest levels of tdTomato expression were observed in a pair of arch-shaped structures located along the lateral and third ventricles. Additionally, clusters of tdTomato signals were detected on the ventral surface of the posterior brain, while lower levels of signals were observed throughout other regions.Fig. 2A. A representative lightsheet image of the brain from a 1-month-old male PdgfrαCre;ROSA-tdTomato mouse. Following SmartBatch + tissue delipification, tdTomato signals were visualized using a Zeiss LSM 7 Lightsheet microscope with a 2.5 × objective (N = 3). B-D. tdTomato expression in 1-month-old PdgfrαCre;ROSA-tdTomato mice is depicted together with blue DAPI signals at the levels of the anterior striatum (B), middle striatum (C), posterior striatum/fimbria (D), anterior lateral ventricle (E), middle lateral ventricle (F), and fimbria (F). Abbreviations: ChPx, choroid plexus; LV, lateral ventricle; Str, striatum; Sept, septum; CC, corpus callosum; CTX, cortex; fmb, fimbria; SVZ, subventricular zone N = 3 A. A representative lightsheet image of the brain from a 1-month-old male PdgfrαCre;ROSA-tdTomato mouse. Following SmartBatch + tissue delipification, tdTomato signals were visualized using a Zeiss LSM 7 Lightsheet microscope with a 2.5 × objective (N = 3). B-D. tdTomato expression in 1-month-old PdgfrαCre;ROSA-tdTomato mice is depicted together with blue DAPI signals at the levels of the anterior striatum (B), middle striatum (C), posterior striatum/fimbria (D), anterior lateral ventricle (E), middle lateral ventricle (F), and fimbria (F). Abbreviations: ChPx, choroid plexus; LV, lateral ventricle; Str, striatum; Sept, septum; CC, corpus callosum; CTX, cortex; fmb, fimbria; SVZ, subventricular zone N = 3 To identify brain regions exhibiting tdTomato signals in PdgfrαCre;ROSA-tdTomato mice, we analyzed a series of coronal sections from 1-month-old PdgfrαCre;ROSA-tdTomato mice. High levels of tdTomato signals were present in the striatum, superficial layers of the neocortex (Fig. 2BC), choroid plexus within the lateral ventricle (Fig. 2DE) and the fimbria (Fig. 2F). This expression pattern, driven by the Pdgfrα promoter, aligns with previously reported distributions of Pdgfrα mRNA and protein, as well as recombination activity associated with the Pdgfrα promoter in the mouse brain [99–101]. Moreover, the intense staining of the choroid plexus is consistent with the prominently tdTomato + structures resembling the shapes of the lateral and third ventricles (see Fig. 2A). Embryonic Pdgfra-positive cells give rise to distinct pre-oligodendrocyte precursor cells by the perinatal period; however, they exhibit similar profiles during the postnatal and adult periods [68, 102]. PdgfraCre is expected to induce recombination in oligodendrocyte precursor cells [75]. Once tdTomato is expressed through recombination in the PdgfraCre;ROSA-tdTomato mouse line, it remains expressed in their progeny, including mature oligodendrocytes and myelinated fibers. Given that myelination of the fimbria is selectively affected in constitutive Tbx1 heterozygous mice [62], we examined whether tdTomato colocalized with MBP, a marker of mature oligodendrocytes and myelinated fibers, in the fimbria [103]. MBP-positive fibers were present throughout the fimbria (Fig. 3AB). TdTomato signals were more prominent laterally than medially (Fig. 3C) and were colocalized with MBP-positive fibers (Fig. 3D) in the fimbria. These findings confirm that tdTomato, activated by PdgfraCre in PdgfraCre;ROSA-tdTomato mice, is present in myelinated fibers in the fimbria.Fig. 3A Representative Case of Confocal Images of the Fimbria of a 1-Month-Old Male PdgfrαCre;ROSA-tdTomato Mouse. Sections were stained for DAPI (A, blue) and MBP (B, green). C. Red tdTomato signals originated from the PdgfrαCre;ROSA-tdTomato genotype. D. A composite image of the three colors is presented; white arrows indicate an area where fibers are positive for both MBP and tdTomato. N = 4. ChPx, choroid plexus; fmb, fimbria A Representative Case of Confocal Images of the Fimbria of a 1-Month-Old Male PdgfrαCre;ROSA-tdTomato Mouse. Sections were stained for DAPI (A, blue) and MBP (B, green). C. Red tdTomato signals originated from the PdgfrαCre;ROSA-tdTomato genotype. D. A composite image of the three colors is presented; white arrows indicate an area where fibers are positive for both MBP and tdTomato. N = 4. ChPx, choroid plexus; fmb, fimbria Intense tdTomato signals were evident in the choroid plexus located laterally to the lateral surface of the fimbria (Fig. 3D). The cell types in the mouse choroid plexus that express Pdgfra include fibroblasts, mural cells, and endothelial cells during embryonic development, as well as fibroblasts and macrophages during the post-embryonic period in the mouse brain. Although gene profiles indicative of oligodendrocyte precursor-like cells are detectable during the embryonic period, their precise location within the choroid plexus remains unclear [66, 104–106]. The pronounced tdTomato signal identified in the choroid plexus at one month of age likely arises from one or more of these cell types at any developmental time point. tdTomato signals detected at this age represent recombination events that may have occurred at any developmental time point prior one month of age. Effects of Tbx1 Heterozygosity in Oligodendrocyte Precursor Cells on Behavior The growth rate may influence behavior; PdgfrαCre;Tbx1+/flox, referred to as oTbx1+/− mice, may exhibit developmental delays, with their phenotypes potentially reflecting a lag in development compared to their wild-type littermates. However, neonatal oTbx1+/+ and oTbx1+/− pups demonstrated comparable increases in body weight (Fig. S1). To investigate the impact of Tbx1 heterozygosity in oligodendrocyte precursor cells and their progeny on behavior, mice were tested for social communication during the neonatal period and subjected to a series of tests evaluating social, cognitive, anxiety-related, and motor behaviors at one and two months of age. This timing aligns with the peak of myelination, which occurs between the fourth and fifth postnatal weeks in rodents [107, 108] and the onset of puberty around 1 month of age and adolescence around 2 months of age [85]. Constitutive Tbx1 heterozygous mice exhibit impairments in 1) neonatal social communication as early as P7, 2) social behaviors, 3) working memory/cognitive flexibility in a T-maze and attentional set shifting, and 4) affect-related behaviors in an inescapable open field, although not in an elevated plus maze during adolescence [58–62]. We first assessed differences among the three control groups: PdgfraCre;Tbx1+/+, WT;Tbx1+/flox, and WT;Tbx1+/+. No significant differences were observed among the controls for any behavioral measure, except at four time points regarding total margin time in the open field at 2 months of age (see Table S3-Fig. 8D). When analyzing all four genotypes together, a significant genotype effect was found without an interaction effect with time (Genotype, P = 0.0256; Genotype x Time, P = 0.665). However, post-hoc comparisons did not reveal significant differences between WT;Tbx1+/flox and WT;Tbx1+/+ mice except for two time points at 5 min and 15 min (see Table S3, Fig. 8D). Given the sample size for these two control groups was n = 6 each, we lacked confidence in these statistical values. Consequently, we combined all the control genotypes into a single control group, designated as oTbx1+/+, for comparison with oTbx1+/− [66]. Constitutive Tbx1+/− mice display various neonatal, peri-adolescent, and postnatal behavioral phenotypes [58, 59]. Thus, we evaluated mice for neonatal ultrasonic vocalization on P8 and P12–P13 and conducted additional tests with peripubertal (1 month) and adolescent (2 months) mice, allowing 1–2 day intervals between tasks to minimize potential carryover effects from prior testing [93]. We found no differences between oTbx1+/+ and oTbx1+/− pups regarding the number, percentage, and duration of various neonatal vocal call types on P8 and P12 (Fig. 4A–C). At 1 and 2 months of age, mice were sequentially assessed for social interaction, novel object approach, spontaneous alternation in a T-maze, anxiety-related behavior in an elevated plus maze, and locomotor activity and thigmotaxis in an inescapable open field. oTbx1+/− and oTbx1+/+ mice were indistinguishable in social interaction (Fig. 5A) and novel object approach (Fig. 5B). In a T-maze, oTbx1+/− mice at 1 month of age exhibited higher spontaneous alternation rates at the longest delay (Fig. 6A); otherwise, oTbx1+/+ and oTbx1+/− mice were indistinguishable in rates of spontaneous alternation at 2 months of age and in latencies to correct choices at both ages (Fig. 6A, B).Fig. 4The number (A), proportion (B), and duration (C) (mean ± standard error of the mean [SEM]) of 12 call types emitted on postnatal days (P) P8 and P12. The two genotypes exhibited differences in certain interaction effects: A. Number: Genotype, F(1,912) = 0.1966, P = 0.6576; Genotype x Postnatal day, F(1, 912) = 4.835, P = 0.0281; Genotype x Call type, F(11, 912) = 0.7077, P = 0.7319; Genotype x Postnatal day x Call types, F(11, 912) = 0.3686, P = 0.9679. B. Proportion: Genotype, F(1,912) = 0.0107, P = 0.9176; Genotype x Postnatal day, F(1, 912) = 0.0107, P = 0.9176; Genotype x Call type, F(11, 912) = 1.3657, P = 0.1837; Genotype x Postnatal day x Call types, F(11, 912) = 0.7966, P = 0.6437. C. Duration: Genotype, F(1,560) = 0.0435, P = 0.8349; Genotype x Postnatal day, F(1, 560) = 0.10519, P = 0.7458; Genotype x Call type, F(11, 560) = 3.361, P = 0.00016; Genotype x Postnatal day x Call types, F(11, 560) = 1.0389, P = 0.4101. However, Mann–Whitney U tests did not reveal any significant genotype effects at any postnatal day for any call type after applying Benjamini–Hochberg corrections at 5% false discovery rate (FDR). Inset in A: The total number of ultrasonic vocalizations (mean ± SEM) emitted at P8 and P12. The two genotypes did not differ on the two postnatal days (Genotype, F(1,76) = 0.1559, P = 0.694; Genotype x Postnatal day, F(1, 76) = 1.103, P = 0.2969). Inset in C: Duration: Genotype, F(1,32) = 0.0002, P = 0.9878; Genotype x Postnatal day, F(1, 32) = 0.3617, P = 0.5518). Ha, harmonic; Sd, step-down; Sh, short; Df, down frequency modulation; Uf, up frequency modulation; Ts, two steps; Fl, flat; Ch, chevron; Co, complex; Su, step-up; Ms, multiple steps; Rc, reverse chevron. oTbx1+/+ mice, N = 27; oTbx1+/− mice, N = 11 at P8; oTbx1+/+ mice, N = 30; oTbx1+/− mice, N = 12 at P12Fig. 5Interaction with Social and Non-Social Stimuli. A) Social Interaction. The oTbx1+/− (PdgfrαCre;Tbx1+/flox) mice and oTbx1 +/+ mice did not differ in the amount of time (mean ± SEM) spent in affiliative social interaction (1 month: genotype, F(1, 72) = 0.1118, P = 0.739; genotype × session, F(1, 72) = 0.262, P = 0.6102; 2 months: genotype, F(1, 58) = 0.00547, P = 0.9413; genotype x session, F(1, 58) = 1.2175, P = 0.2744). The three genotypes of oTbx1 +/+ (1 month, N = 8, PdgfrαCre;Tbx1+/+, N = 10, Wild-type;Tbx1+/+, N = 9, Wild-type;Tbx1+/flox; 2 months, N = 8, PdgfrαCre;Tbx1+/+, N = 6, Wild-type;Tbx1+/+, N = 7, Wild-type;Tbx1+/flox) did not show significant differences (1 month: genotype, F(2, 48) = 1.686, P = 0.1960; genotype × session, F(2, 48) = 0.843, P = 0.4366; 2 months: genotype, F(2, 36) = 0.0392, P = 0.9616; genotype x session, F(2, 36) = 0.2382, P = 0.7893), and were thus combined as oTbx1 +/+ . oTbx1+/+ mice, N = 27; oTbx1+/− mice, N = 11 at 1 month; oTbx1+/+ mice, N = 21; oTbx1+/− mice, N = 10 at 2 months. B) Approach to a novel, non-social object. The two genotypes did not differ in the amount of time spent in affiliative social interaction (1 month: genotype, F(1, 70) = 2.2507E-05, P = 0.9962; genotype × session, F(1, 70) = 1.0305, P = 0.31354; 2 months: genotype, F(1, 58) = 0.11634, P = 0.7342; genotype × session, F(1, 58) = 0.0821, P = 0.77545). The three genotypes of oTbx1+/+ (1 month: N = 8, PdgfrαCre;Tbx1+/+, N = 9, Wild-type;Tbx1+/+, N = 9, Wild-type;Tbx1+/flox; 2 months: N = 8, PdgfrαCre;Tbx1+/+, N = 6, Wild-type;Tbx1+/+, N = 7, Wild-type;Tbx1+/flox) did not differ (1 month: genotype, F(2, 46) = 0.1772, P = 0.8382; genotype × session, F(2, 46) = 0.9318, P = 0.4012; 2 months: genotype, F(2, 36) = 0.0580, P = 0.9437; genotype × session, F(2, 36) = 0.6581, P = 0.5239) and were therefore combined as oTbx1+/+. oTbx1+/+ mice, N = 26; oTbx1+/− mice, N = 11 at 1 month; oTbx1+/+ mice, N = 21; oTbx1+/− mice, N = 10 at 2 monthsFig. 6Spontaneous alternation in a T-maze. Compared to male oTbx1+/+ (N = 20) mice, male oTbx1+/− (N = 10) mice performed at higher percentages of correct spontaneous alternations (mean ± SEM) at the longest delay (30 s) in a T-maze at 1 month of age, but not at 2 months. Specifically, at 1 month: % alternation, genotype, F(1, 96) = 4.66, P = 0.0333; genotype × delays, F(2,96) = 3.1509, P = 0.0473; at 2 months: genotype, F(1, 81) = 2.047, P = 0.1564; genotype × delays, F(2, 81) = 0.2439, P = 0.7841. The three genotypes of oTbx1+/+ (1 month: N = 8, PdgfrαCre;Tbx1+/+, N = 7, Wild-type;Tbx1+/+, N = 9, Wild-type;Tbx1+/flox; 2 months: N = 8, PdgfrαCre;Tbx1+/+, N = 6, Wild-type;Tbx1+/+, N = 6, Wild-type;Tbx1+/flox) did not differ (1 month: genotype, F(2, 63) = 0.1046, P = 0.9008; genotype × delays, F(4, 63) = 0.2648, P = 0.8995; 2 months: genotype, F(2, 51) = 0.1197, P = 0.8874; genotype × delays, F(4, 51) = 0.6401, P = 0.6363) and consequently were combined as oTbx1+/+. oTbx1+/− mice exhibited higher rates of correct alternations at the 30-s delay compared to oTbx1 +/+ mice, as determined by the Mann–Whitney test (P = 0.0027). B) The two groups did not exhibit significant differences in the latency to correct choices in the T-maze at 1 month (genotype, F(1, 96) = 2.181, P = 0.1430; genotype × delays, F(2, 96) = 2.058, P = 0.1332) or at 2 months (genotype, F(1, 81) = 1.1135, P = 0.2945; genotype × delays, F(2, 81) = 2.5825, P = 0.0818). The three genotypes of oTbx1+/+ (1 month, N = 8), PdgfrαCre;Tbx1+/+ (N = 7), Wild-type;Tbx1+/+ (N = 9), and Wild-type;Tbx1+/flox (2 months, N = 8), PdgfrαCre;Tbx1+/+ (N = 6), Wild-type;Tbx1+/+ (N = 6), and Wild-type;Tbx1+/flox, demonstrated a significant interaction effect at 1 month of age (genotype, F(2, 63) = 1.5787, P = 0.2143; genotype × session, F(4, 63) = 4.5625, P = 0.00267), but no significant interaction at 2 months of age (genotype, F(2, 51) = 0.1631, P = 0.8500; genotype × session, F(4, 51) = 1.0611, P = 0.3853). However, Mann–Whitney U tests revealed that none of the three genotypes significantly differed from one another at 1 month of age after applying the Benjamini–Hochberg correction (P > 0.05 for all pairs). The three genotypes were combined as oTbx1+/+. The sample sizes were as follows: oTbx1+/+ mice, N = 24; oTbx1+/− mice, N = 10 at 1 month; oTbx1+/+ mice, N = 20; oTbx1+/− mice, N = 9 at 2 months The number (A), proportion (B), and duration (C) (mean ± standard error of the mean [SEM]) of 12 call types emitted on postnatal days (P) P8 and P12. The two genotypes exhibited differences in certain interaction effects: A. Number: Genotype, F(1,912) = 0.1966, P = 0.6576; Genotype x Postnatal day, F(1, 912) = 4.835, P = 0.0281; Genotype x Call type, F(11, 912) = 0.7077, P = 0.7319; Genotype x Postnatal day x Call types, F(11, 912) = 0.3686, P = 0.9679. B. Proportion: Genotype, F(1,912) = 0.0107, P = 0.9176; Genotype x Postnatal day, F(1, 912) = 0.0107, P = 0.9176; Genotype x Call type, F(11, 912) = 1.3657, P = 0.1837; Genotype x Postnatal day x Call types, F(11, 912) = 0.7966, P = 0.6437. C. Duration: Genotype, F(1,560) = 0.0435, P = 0.8349; Genotype x Postnatal day, F(1, 560) = 0.10519, P = 0.7458; Genotype x Call type, F(11, 560) = 3.361, P = 0.00016; Genotype x Postnatal day x Call types, F(11, 560) = 1.0389, P = 0.4101. However, Mann–Whitney U tests did not reveal any significant genotype effects at any postnatal day for any call type after applying Benjamini–Hochberg corrections at 5% false discovery rate (FDR). Inset in A: The total number of ultrasonic vocalizations (mean ± SEM) emitted at P8 and P12. The two genotypes did not differ on the two postnatal days (Genotype, F(1,76) = 0.1559, P = 0.694; Genotype x Postnatal day, F(1, 76) = 1.103, P = 0.2969). Inset in C: Duration: Genotype, F(1,32) = 0.0002, P = 0.9878; Genotype x Postnatal day, F(1, 32) = 0.3617, P = 0.5518). Ha, harmonic; Sd, step-down; Sh, short; Df, down frequency modulation; Uf, up frequency modulation; Ts, two steps; Fl, flat; Ch, chevron; Co, complex; Su, step-up; Ms, multiple steps; Rc, reverse chevron. oTbx1+/+ mice, N = 27; oTbx1+/− mice, N = 11 at P8; oTbx1+/+ mice, N = 30; oTbx1+/− mice, N = 12 at P12 Interaction with Social and Non-Social Stimuli. A) Social Interaction. The oTbx1+/− (PdgfrαCre;Tbx1+/flox) mice and oTbx1 +/+ mice did not differ in the amount of time (mean ± SEM) spent in affiliative social interaction (1 month: genotype, F(1, 72) = 0.1118, P = 0.739; genotype × session, F(1, 72) = 0.262, P = 0.6102; 2 months: genotype, F(1, 58) = 0.00547, P = 0.9413; genotype x session, F(1, 58) = 1.2175, P = 0.2744). The three genotypes of oTbx1 +/+ (1 month, N = 8, PdgfrαCre;Tbx1+/+, N = 10, Wild-type;Tbx1+/+, N = 9, Wild-type;Tbx1+/flox; 2 months, N = 8, PdgfrαCre;Tbx1+/+, N = 6, Wild-type;Tbx1+/+, N = 7, Wild-type;Tbx1+/flox) did not show significant differences (1 month: genotype, F(2, 48) = 1.686, P = 0.1960; genotype × session, F(2, 48) = 0.843, P = 0.4366; 2 months: genotype, F(2, 36) = 0.0392, P = 0.9616; genotype x session, F(2, 36) = 0.2382, P = 0.7893), and were thus combined as oTbx1 +/+ . oTbx1+/+ mice, N = 27; oTbx1+/− mice, N = 11 at 1 month; oTbx1+/+ mice, N = 21; oTbx1+/− mice, N = 10 at 2 months. B) Approach to a novel, non-social object. The two genotypes did not differ in the amount of time spent in affiliative social interaction (1 month: genotype, F(1, 70) = 2.2507E-05, P = 0.9962; genotype × session, F(1, 70) = 1.0305, P = 0.31354; 2 months: genotype, F(1, 58) = 0.11634, P = 0.7342; genotype × session, F(1, 58) = 0.0821, P = 0.77545). The three genotypes of oTbx1+/+ (1 month: N = 8, PdgfrαCre;Tbx1+/+, N = 9, Wild-type;Tbx1+/+, N = 9, Wild-type;Tbx1+/flox; 2 months: N = 8, PdgfrαCre;Tbx1+/+, N = 6, Wild-type;Tbx1+/+, N = 7, Wild-type;Tbx1+/flox) did not differ (1 month: genotype, F(2, 46) = 0.1772, P = 0.8382; genotype × session, F(2, 46) = 0.9318, P = 0.4012; 2 months: genotype, F(2, 36) = 0.0580, P = 0.9437; genotype × session, F(2, 36) = 0.6581, P = 0.5239) and were therefore combined as oTbx1+/+. oTbx1+/+ mice, N = 26; oTbx1+/− mice, N = 11 at 1 month; oTbx1+/+ mice, N = 21; oTbx1+/− mice, N = 10 at 2 months Spontaneous alternation in a T-maze. Compared to male oTbx1+/+ (N = 20) mice, male oTbx1+/− (N = 10) mice performed at higher percentages of correct spontaneous alternations (mean ± SEM) at the longest delay (30 s) in a T-maze at 1 month of age, but not at 2 months. Specifically, at 1 month: % alternation, genotype, F(1, 96) = 4.66, P = 0.0333; genotype × delays, F(2,96) = 3.1509, P = 0.0473; at 2 months: genotype, F(1, 81) = 2.047, P = 0.1564; genotype × delays, F(2, 81) = 0.2439, P = 0.7841. The three genotypes of oTbx1+/+ (1 month: N = 8, PdgfrαCre;Tbx1+/+, N = 7, Wild-type;Tbx1+/+, N = 9, Wild-type;Tbx1+/flox; 2 months: N = 8, PdgfrαCre;Tbx1+/+, N = 6, Wild-type;Tbx1+/+, N = 6, Wild-type;Tbx1+/flox) did not differ (1 month: genotype, F(2, 63) = 0.1046, P = 0.9008; genotype × delays, F(4, 63) = 0.2648, P = 0.8995; 2 months: genotype, F(2, 51) = 0.1197, P = 0.8874; genotype × delays, F(4, 51) = 0.6401, P = 0.6363) and consequently were combined as oTbx1+/+. oTbx1+/− mice exhibited higher rates of correct alternations at the 30-s delay compared to oTbx1 +/+ mice, as determined by the Mann–Whitney test (P = 0.0027). B) The two groups did not exhibit significant differences in the latency to correct choices in the T-maze at 1 month (genotype, F(1, 96) = 2.181, P = 0.1430; genotype × delays, F(2, 96) = 2.058, P = 0.1332) or at 2 months (genotype, F(1, 81) = 1.1135, P = 0.2945; genotype × delays, F(2, 81) = 2.5825, P = 0.0818). The three genotypes of oTbx1+/+ (1 month, N = 8), PdgfrαCre;Tbx1+/+ (N = 7), Wild-type;Tbx1+/+ (N = 9), and Wild-type;Tbx1+/flox (2 months, N = 8), PdgfrαCre;Tbx1+/+ (N = 6), Wild-type;Tbx1+/+ (N = 6), and Wild-type;Tbx1+/flox, demonstrated a significant interaction effect at 1 month of age (genotype, F(2, 63) = 1.5787, P = 0.2143; genotype × session, F(4, 63) = 4.5625, P = 0.00267), but no significant interaction at 2 months of age (genotype, F(2, 51) = 0.1631, P = 0.8500; genotype × session, F(4, 51) = 1.0611, P = 0.3853). However, Mann–Whitney U tests revealed that none of the three genotypes significantly differed from one another at 1 month of age after applying the Benjamini–Hochberg correction (P > 0.05 for all pairs). The three genotypes were combined as oTbx1+/+. The sample sizes were as follows: oTbx1+/+ mice, N = 24; oTbx1+/− mice, N = 10 at 1 month; oTbx1+/+ mice, N = 20; oTbx1+/− mice, N = 9 at 2 months The two genotypes were indistinguishable in the percentage of time spent in open arms (Fig. 7A) and visits to open arms (Fig. 7B) of the elevated plus maze.Fig. 7Anxiety-related behavior in an elevated T-maze.The oTbx1+/− and oTbx1+/+ mice did not differ significantly in the relative amount of time spent (mean + SEM) in the open arms of an elevated plus maze (A,1 month,U = 100, P = 0.3440; 2 months, U = 98, P = 0.9483) or the frequency of visits to the open arms of the elevated plus maze (B, 1 month,t(32) = 1.7412, P = 0.0912; 2 months, t(28) = 0.5925, P = 0.5583). The three genotypes of oTbx1 +/+ (1 month, N = 7, PdgfrαCre;Tbx1+/+, N = 7, Wild-type;Tbx1+/+, N = 6, Wild-type;Tbx1 +/flox; 2 months, N = 8, PdgfrαCre;Tbx1+/+, N = 6, Wild-type;Tbx1+/+, N = 6, Wild-type;Tbx1+/flox) did not differ in the percentage of time spent in open arms (1 month,genotype, F(2, 20) = 0.0745, P = 0.9285; 2 months,genotype, F(2, 17) = 2.126, P = 0.1500) or the percentage of visits to open arms (1 month,genotype, F(2, 20) = 0.1837, P = 0.8336; 2 months,genotype, F(2, 17) = 1.4821, P = 0.2551), and thus were combined as oTbx1+/+. The sample sizes were: oTbx1+/+ mice, N = 23; oTbx1+/− mice, N = 11 at 1 month; oTbx1+/+ mice, N = 20; oTbx1+/− mice, N = 10 at 2 months Anxiety-related behavior in an elevated T-maze.The oTbx1+/− and oTbx1+/+ mice did not differ significantly in the relative amount of time spent (mean + SEM) in the open arms of an elevated plus maze (A,1 month,U = 100, P = 0.3440; 2 months, U = 98, P = 0.9483) or the frequency of visits to the open arms of the elevated plus maze (B, 1 month,t(32) = 1.7412, P = 0.0912; 2 months, t(28) = 0.5925, P = 0.5583). The three genotypes of oTbx1 +/+ (1 month, N = 7, PdgfrαCre;Tbx1+/+, N = 7, Wild-type;Tbx1+/+, N = 6, Wild-type;Tbx1 +/flox; 2 months, N = 8, PdgfrαCre;Tbx1+/+, N = 6, Wild-type;Tbx1+/+, N = 6, Wild-type;Tbx1+/flox) did not differ in the percentage of time spent in open arms (1 month,genotype, F(2, 20) = 0.0745, P = 0.9285; 2 months,genotype, F(2, 17) = 2.126, P = 0.1500) or the percentage of visits to open arms (1 month,genotype, F(2, 20) = 0.1837, P = 0.8336; 2 months,genotype, F(2, 17) = 1.4821, P = 0.2551), and thus were combined as oTbx1+/+. The sample sizes were: oTbx1+/+ mice, N = 23; oTbx1+/− mice, N = 11 at 1 month; oTbx1+/+ mice, N = 20; oTbx1+/− mice, N = 10 at 2 months We assessed motor activity and anxiety-related behavior in a high-stress open-field task [58, 62] where there is no option to escape to a closed space. In this task, oTbx1+/− and oTbx1+/+ mice showed no differences in distance traveled at both 1 month and 2 months of age (Fig. 8A,B) or in time spent in the margin zone (Fig. 8C,D).Fig. 8Motor Activity and Thigmotaxis. oTbx1+/+ and oTbx1+/− mice exhibited indistinguishable distances traveled at both 1 month and 2 months of age (A, 1 month, genotype, F(1, 198) = 0.6314, P = 0.4278; genotype × time, F(5, 198) = 0.9551, P = 0.4466; B, 2 months, genotype, F(1, 28) = 0.0242, P = 0.87739; genotype × time, F(3.66, 102.485) = 0.5389, P = 0.69188). The Greenhouse–Geisser correction was applied due to a violation of sphericity, with the estimated epsilon being less than 0.75.The three genotypes of oTbx1+/+ (1 month, N = 8, PdgfrαCre;Tbx1+/+, N = 7, Wild-type;Tbx1+/+, N = 9, Wild-type;Tbx1+/flox; 2 months, N = 8, PdgfrαCre;Tbx1+/+, N = 6, Wild-type;Tbx1+/+, N = 6, Wild-type;Tbx1+/flox) demonstrated differences at both ages (genotype, F(2, 21) = 5.475, P = 0.01220; genotype × time, F(7.10, 74.58) = 1.118, P = 0.3611; at 2 months,genotype, F(2, 17) = 9.533, P = 0.00167; genotype × time, F(10, 85) = 0.3955, P = 0.9453). Comparisons of all pairs using Student’s t-tests indicated that no pair reached significance after the Benjamini–Hochberg correction (P > 0.05 for all pairs), resulting in their combination as oTbx1+/+. oTbx1+/+ and oTbx1+/− mice spent equal amounts of time in the margin zone (i.e., thigmotaxis) of the inescapable open field at both 1 and 2 months of age (C, 1 month, genotype, F(1, 198) = 0.2009, P = 0.6545; genotype × time, F(5,198) = 2.1517, P = 0.0609; D, 2 months, genotype, F(1, 168) = 0.08912, P = 0.7657; genotype × time, F(5, 168) = 0.9880, P = 0.4267). The three genotypes of oTbx1+/+ did not differ at 1 month of age (genotype, F(2, 21) = 0.3686, P = 0.6961; genotype × time, F(5.485, 57.59) = 0.9884, P = 0.4375), but differences were observed at 2 months of age (genotype, F(2, 17) = 4.880, P = 0.0211; genotype × time, F(5.049, 42.917) = 0.7180, P = 0.6147). Comparisons of all pairs using Student’s t-tests revealed that 2-month-old Wild-type;Tbx1+/flox and Wild-type;Tbx1+/+ differed at the first four time points, even after the Benjamini-Hochberg correction (P < 0.05). However, these groups had relatively small sample sizes (N = 6 for both), and PdgfrαCre;Tbx1+/+and Wild-type;Tbx1+/flox or PdgfrαCre;Tbx1+/+ and Wild-type;Tbx1+/flox did not differ at any time point (P > 0.05), thus they were combined as oTbx1+/+. Sample sizes included oTbx1+/+ mice, N = 24; oTbx1+/− mice, N = 11 at 1 month; oTbx1+/+ mice, N = 20; oTbx1+/− mice, N = 10 at 2 months Motor Activity and Thigmotaxis. oTbx1+/+ and oTbx1+/− mice exhibited indistinguishable distances traveled at both 1 month and 2 months of age (A, 1 month, genotype, F(1, 198) = 0.6314, P = 0.4278; genotype × time, F(5, 198) = 0.9551, P = 0.4466; B, 2 months, genotype, F(1, 28) = 0.0242, P = 0.87739; genotype × time, F(3.66, 102.485) = 0.5389, P = 0.69188). The Greenhouse–Geisser correction was applied due to a violation of sphericity, with the estimated epsilon being less than 0.75.The three genotypes of oTbx1+/+ (1 month, N = 8, PdgfrαCre;Tbx1+/+, N = 7, Wild-type;Tbx1+/+, N = 9, Wild-type;Tbx1+/flox; 2 months, N = 8, PdgfrαCre;Tbx1+/+, N = 6, Wild-type;Tbx1+/+, N = 6, Wild-type;Tbx1+/flox) demonstrated differences at both ages (genotype, F(2, 21) = 5.475, P = 0.01220; genotype × time, F(7.10, 74.58) = 1.118, P = 0.3611; at 2 months,genotype, F(2, 17) = 9.533, P = 0.00167; genotype × time, F(10, 85) = 0.3955, P = 0.9453). Comparisons of all pairs using Student’s t-tests indicated that no pair reached significance after the Benjamini–Hochberg correction (P > 0.05 for all pairs), resulting in their combination as oTbx1+/+. oTbx1+/+ and oTbx1+/− mice spent equal amounts of time in the margin zone (i.e., thigmotaxis) of the inescapable open field at both 1 and 2 months of age (C, 1 month, genotype, F(1, 198) = 0.2009, P = 0.6545; genotype × time, F(5,198) = 2.1517, P = 0.0609; D, 2 months, genotype, F(1, 168) = 0.08912, P = 0.7657; genotype × time, F(5, 168) = 0.9880, P = 0.4267). The three genotypes of oTbx1+/+ did not differ at 1 month of age (genotype, F(2, 21) = 0.3686, P = 0.6961; genotype × time, F(5.485, 57.59) = 0.9884, P = 0.4375), but differences were observed at 2 months of age (genotype, F(2, 17) = 4.880, P = 0.0211; genotype × time, F(5.049, 42.917) = 0.7180, P = 0.6147). Comparisons of all pairs using Student’s t-tests revealed that 2-month-old Wild-type;Tbx1+/flox and Wild-type;Tbx1+/+ differed at the first four time points, even after the Benjamini-Hochberg correction (P < 0.05). However, these groups had relatively small sample sizes (N = 6 for both), and PdgfrαCre;Tbx1+/+and Wild-type;Tbx1+/flox or PdgfrαCre;Tbx1+/+ and Wild-type;Tbx1+/flox did not differ at any time point (P > 0.05), thus they were combined as oTbx1+/+. Sample sizes included oTbx1+/+ mice, N = 24; oTbx1+/− mice, N = 11 at 1 month; oTbx1+/+ mice, N = 20; oTbx1+/− mice, N = 10 at 2 months While conditional Tbx1 heterozygosity in the oligodendrocyte cell lineage resulted in higher spontaneous alternation scores at the longest inter-trial delay at one month of age, this phenotype contrasts with the impaired spontaneous alternation observed in constitutive Tbx1 heterozygous mice [58, 62]. It is noteworthy that oTbx1+/+ mice exhibited consistent decline in alternation rates to chance levels at a 30 s delay, whereas oTbx1+/- mice did not show this decline from 15 to 30 s delay (see Fig. 6A, 1 month), indicating a delayed decay of working memory or cognitive flexibility. Furthermore, this conditional Tbx1 heterozygosity did not replicate the altered neonatal vocalizations, peri-adolescent or postnatal social interaction deficits, or heightened responses to novel, non-social objects or thigmotaxis observed in constitutive Tbx1 heterozygous mice [58]. Effects of Tbx1 Heterozygosity in Oligodendrocyte Precursor Cells on Myelinated Axons in the Fimbria Following behavioral testing, we examined myelinated axons in the fimbria utilizing electron microscopy. Constitutive Tbx1 heterozygosity distinctly results in an increased proportion of myelinated axons with diameters ranging from 200 to 600 nm diameter, while concurrently reducing the proportion of axons measuring ≥ 700 nm and < 1,200 nm. Additionally, axons within the 700–1,500 nm range demonstrated a thicker myelin sheath, with no larger myelinated axons identified in constitutive Tbx1 heterozygous mice [62]. We evaluated whether the ultrastructural alterations observed in the fimbria of constitutive Tbx1+/− mice were mirrored in oTbx1+/− mice. Densely packed axons were identified in the fimbria (Fig. 9A). To evaluate relative myelin thickness, we analyzed the ratio of inner axon diameter to outer fiber diameter (i.e., the g-ratio) (Fig. 9B). Similar to the findings in constitutive Tbx1 mice, the g-ratio plateaued at approximately 0.8, which is the optimal ratio for signal conduction in the brain [88]. However, no significant differences in g-ratios were noted between oTbx1+/− and oTbx1+/+ mice across the full range of axon diameters.Fig. 9Electron Microscopy Analysis of Myelinated Axons. A. Representative electron microscopy images of axons in the fimbria at 20,000 × magnification. Genotypes: +/+ , oTbx1+/+; +/− , oTbx1+/−. Scale bar = 1,600 nm. B. G-ratios plotted against axon diameters for 4,423 and 2,317 axons from 5 oTbx1+/+ mice and 3 oTbx1+/− mice, respectively. All axons with closed myelin sheaths were analyzed using MyelTracer. The reciprocal function best fits the data distributions. G-ratios increased as a function of axon diameter (oTbx1 +/+, R = 0.6268, P < 0.001; oTbx1 +/− , R = 0.6491, P < 0.001). There was no significant difference in the axon diameter range for G-ratios between oTbx1 +/+ and oTbx1 +/− mice (genotype, P = 0.8131; genotype × axon diameter, P = 0.221). C. Average number (± SEM) of axons per image field versus axon diameter. Significant differences were observed between oTbx1+/+ and oTbx1+/− mice (χ2 (27) = 109.7085, P = 6.22 × 10−12), primarily due to a higher number of axons with diameters ≥ 300 nm and < 800 nm and a lower number of axons with diameters ≥ 1,200 nm and < 1,500 nm in oTbx1+/− compared to oTbx1+/+ mice, as determined by Mann–Whitney U tests applied to this diameter range (*, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001). The shift in axon sizes toward smaller to medium diameters in oTbx1+/− compared to oTbx1+/+ mice resulted in a smaller average axon size for oTbx1+/− mice (Mann–Whitney non-parametric tests, P = 7.48 × 10−19). D. Differences in the percentage of myelinated axons for oTbx1+/− versus oTbx1 +/+ mice across axon diameters (Kolmogorov–Smirnov, P = 3.59 × 10−16) Electron Microscopy Analysis of Myelinated Axons. A. Representative electron microscopy images of axons in the fimbria at 20,000 × magnification. Genotypes: +/+ , oTbx1+/+; +/− , oTbx1+/−. Scale bar = 1,600 nm. B. G-ratios plotted against axon diameters for 4,423 and 2,317 axons from 5 oTbx1+/+ mice and 3 oTbx1+/− mice, respectively. All axons with closed myelin sheaths were analyzed using MyelTracer. The reciprocal function best fits the data distributions. G-ratios increased as a function of axon diameter (oTbx1 +/+, R = 0.6268, P < 0.001; oTbx1 +/− , R = 0.6491, P < 0.001). There was no significant difference in the axon diameter range for G-ratios between oTbx1 +/+ and oTbx1 +/− mice (genotype, P = 0.8131; genotype × axon diameter, P = 0.221). C. Average number (± SEM) of axons per image field versus axon diameter. Significant differences were observed between oTbx1+/+ and oTbx1+/− mice (χ2 (27) = 109.7085, P = 6.22 × 10−12), primarily due to a higher number of axons with diameters ≥ 300 nm and < 800 nm and a lower number of axons with diameters ≥ 1,200 nm and < 1,500 nm in oTbx1+/− compared to oTbx1+/+ mice, as determined by Mann–Whitney U tests applied to this diameter range (*, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001). The shift in axon sizes toward smaller to medium diameters in oTbx1+/− compared to oTbx1+/+ mice resulted in a smaller average axon size for oTbx1+/− mice (Mann–Whitney non-parametric tests, P = 7.48 × 10−19). D. Differences in the percentage of myelinated axons for oTbx1+/− versus oTbx1 +/+ mice across axon diameters (Kolmogorov–Smirnov, P = 3.59 × 10−16) We also quantified the number and proportion of axons across different diameters. The fimbria of oTbx1+/− mice exhibited an increased number of axons measuring ≥ 300 nm and < 800 nm, alongside decreased number of axons ≥ 1,200 nm and < 1,500 nm diameters in comparison to oTbx1+/+ mice (Fig. 9C). Consequently, the curve representing relative axon proportions (y-axis) versus axon diameter (x-axis) shifted to the left in oTbx1+/− mice relative to oTbx1+/+ mice (Fig. 9D). Consistent with g-ratio findings, myelin thickness showed no significant differences between oTbx1+/− and oTbx1+/+ mice across all axon diameters (Fig. S2). Thus, conditional Tbx1 heterozygosity initiated in oligodendrocyte precursor cells selectively shifted the relative proportions of small to medium myelinated axons in the fimbria without impacting myelin thickness. In summary, conditional Tbx1 heterozygosity within the oligodendrocyte lineage resulted in a shift toward smaller-sized myelinated axons while leaving myelin thickness unchanged in the fimbria. Discussion The objective of this study was to delineate the cellular origin of the diverse behavioral and myelin phenotypes observed in constitutive Tbx1 heterozygous mice [58–62]. Given that social and cognitive phenotypes are recapitulated by initiating Tbx1 heterozygosity in neonatal progenitor cells [73], and that certain neonatal neural progenitor cells in the mouse brain are responsible for generating neurons and oligodendrocytes during the neonatal period [71, 72, 109], it is possible that Tbx1 deficiency in the neuronal or oligodendrocyte lineage contributes to the observed behavioral phenotypes. This study aimed to investigate the impact of Tbx1 heterozygosity in the oligodendrocyte lineage on the behavioral and ultrastructural phenotypes. To achieve this aim, we crossed PdgfrαCre mice with congenic Tbx1+/flox mice to generate PdgfrαCre;Tbx1+/flox (oTbx1+/-) mice. We evaluated a range of social, cognitive, anxiety-related, and motor behaviors, as well as the ultrastructural composition of myelin and axons in the fimbria. The oTbx1+/- mice selectively replicated the altered proportion of axon sizes in the fimbria observed in constitutive Tbx1 heterozygous mice. In contrast, oTbx1+/- mice exhibited higher levels of spontaneous alternation following a 30-s delay at one month of age—indicative of a delayed decay of working memory and cognitive flexibility– contrary to the impaired spontaneous alternation observed in constitutive Tbx1 heterozygous mice. This observation supports the notion that the final phenotype arises from the cumulative effects of various contributory and opposing factors [54, 81]. Other phenotypes associated with constitutive Tbx1 heterozygosity were absent in oTbx1+/- mice, including neonatal social communication, responses to social or non-social stimuli, and thigmotaxis. These findings indicate that Tbx1 in the oligodendrocyte lineage does not significantly influence these phenotypes. Combined with the observation that Tbx1 heterozygosity initiated in neonatal (P1–P5) neural progenitor cells recapitulates some social and cognitive deficits observed in constitutive Tbx1 heterozygous mice [73], the current negative data suggest that Tbx1 in neonatal neural progenitor cells or their neuronal lineage may play a principal role in determining the final phenotype [73]. We cannot dismiss the possibility that conditional Tbx1 heterozygosity was incomplete. The Pdgfrα promoter is expected to induce nearly complete recombination in oligodendrocyte precursor cells, with labels persisting in their mature progeny, oligodendrocytes [110, 111]; however, variability in recombination may occur in certain cells [112]. The apparent absence of behavioral impairments may be attributed to incomplete recombination in all or some individual oligodendrocyte precursor cells. Nonetheless, oTbx1+/- mice showed an axonal phenotype at a presumed gene dosage reduction that did not significantly affect behavior, clearly indicating that behavioral phenotypes exhibit lower sensitivity to Tbx1 heterozygosity in oligodendrocyte precursor cells compared to the axonal phenotype. Constitutive Tbx1 heterozygous mice exhibit increased myelin thickness of medium-sized axons and a lack of large (≥ 1,700 nm) myelinated axons in the fimbria [62]; however, these characteristics were absent in oTbx1+/- mice. On the other hand, the fimbria of oTbx1+/- mice exhibited a greater number of axons with diameters ranging from ≥ 300 nm to < 800 nm and a reduced number of axons with diameters from ≥ 1,200 nm to < 1,500 nm, compared to oTbx1+/+ mice. This pattern partially aligns with the increase in axons measuring ≥ 200 nm and < 400 nm, as well as the decrease in axons measuring ≥ 700 nm and < 1,700 nm in constitutive Tbx1 heterozygous mice. The slight differences in the ranges of diameters of altered axons might be due to manual measurements in our previous study [62] and an automated measurements by MyelTracer in the current study. This shift in axonal composition may result from oligodendrocyte lineage effects in a non-cell-autonomous manner, or from a non-specific, cell-autonomous effect of the Pdgfrα promoter in neurons. While PdgfrαCre primarily induces recombination in oligodendrocyte precursor cells, it also initiates recombination in neurons, astrocytes, pericytes, ependyma, perivascular mesenchymal cells, and other cell types [75–78, 100, 111, 113]. The recombination initiated by PdgfraCre could have influenced the behavioral phenotype in spontaneous alternation through interactions with other cell types and their progeny. In the mouse brain, Tbx1 is undetectable in astrocytes or microglia, yet it is expressed in endothelial cells and neural progenitor cells. Furthermore, Tbx1 is detected in immature neurons as well as both excitatory and inhibitory neurons at various developmental stages, from embryonic development to adulthood [65–67, 114]. Additional research is required to clarify the impacts of Tbx1 deficiency across different cell types on phenotypic outcomes. The general absence of phenotypic abnormalities may be attributed to the high mortality rate observed in oTbx1+/- litters, potentially skewing our sample toward mice that were less affected by, or more resilient to, the mutation’s effects. In our breeder colony, only 31% and 24% of male pups survived to postnatal day (P7) and 2 months of age, respectively. Nevertheless, two lines of evidence do not support this explanation. First, a greater number of mice were alive during the neonatal period compared to 1 and 2 months of age, yet neonatal vocalizations on P8 and P12 were normal. Second, our nestinCreERT;Tbx1+/flox mice also exhibited a high mortality rate but exhibited more significant behavioral deficits [73]. Large-scale human brain imaging studies have identified alterations in white matter among carriers of CNV [115, 116], as well as in idiopathic cases of ASD and schizophrenia [117, 118]. Moreover, individuals with hemizygous deletions of 22q11.2 exhibit changes in white matter within the fornix/fimbria [119]. Our present findings do not support the hypothesis that Tbx1 deficiency in the oligodendrocyte lineage contributes to the structural or behavioral abnormalities related to 22q11.2 hemizygosity and protein-truncating variants of TBX1. Further research is necessary to identify the specific cell types responsible for the behavioral and structural phenotypes linked to Tbx1 heterozygosity, 22q11.2 hemizygosity, and other CNVs, in order to establish a generalizable mechanistic basis for dimensions associated with mental illness. Supplementary Information Supplementary Material 1 Supplementary Material 1 Supplemental Table S3 Supplemental Table S3