Bazedoxifene reverses sexually dimorphic autistic-like abnormalities in biallelic MDGA1-mutant mice.

Bazedoxifene reverses sexually dimorphic autistic-like abnormalities in biallelic MDGA1-mutant mice - PMC Skip to main content An official website of the United States government Here's how you know Here's how you know Official websites use .gov A .gov website belongs to an official government organization in the United States. Secure .gov websites use HTTPS A lock ( Lock Locked padlock icon ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites. Search Log in Dashboard Publications Account settings Log out Search… Search NCBI Primary site navigation Search Logged in as: Dashboard Publications Account settings Log in Search PMC Full-Text Archive Search in PMC Journal List User Guide PERMALINK Copy As a library, NLM provides access to scientific literature. Inclusion in an NLM database does not imply endorsement of, or agreement with, the contents by NLM or the National Institutes of Health. Learn more: PMC Disclaimer | PMC Copyright Notice EMBO Mol Med . 2026 Mar 20;18(4):1358–1398. doi: 10.1038/s44321-026-00402-y Search in PMC Search in PubMed View in NLM Catalog Add to search Bazedoxifene reverses sexually dimorphic autistic-like abnormalities in biallelic MDGA1-mutant mice Seungjoon Kim Seungjoon Kim 1 Department of Brain Sciences, Daegu Gyeongbuk Institute of Science and Technology (DGIST), Daegu, Korea 2 Center for Synapse Diversity and Specificity, DGIST, Daegu, Korea Find articles by Seungjoon Kim 1, 2, # , Hyeonho Kim Hyeonho Kim 1 Department of Brain Sciences, Daegu Gyeongbuk Institute of Science and Technology (DGIST), Daegu, Korea 2 Center for Synapse Diversity and Specificity, DGIST, Daegu, Korea Find articles by Hyeonho Kim 1, 2, # , Javier Porta Pelayo Javier Porta Pelayo 3 Genomics, Genologica, Málaga, Spain Find articles by Javier Porta Pelayo 3, # , Sara Alvarez Sara Alvarez 4 Genomics, NimGenetics, Madrid, Spain Find articles by Sara Alvarez 4, # , Gyubin Jang Gyubin Jang 1 Department of Brain Sciences, Daegu Gyeongbuk Institute of Science and Technology (DGIST), Daegu, Korea 2 Center for Synapse Diversity and Specificity, DGIST, Daegu, Korea Find articles by Gyubin Jang 1, 2 , Jinhu Kim Jinhu Kim 1 Department of Brain Sciences, Daegu Gyeongbuk Institute of Science and Technology (DGIST), Daegu, Korea 2 Center for Synapse Diversity and Specificity, DGIST, Daegu, Korea Find articles by Jinhu Kim 1, 2 , Byeongchan Kim Byeongchan Kim 1 Department of Brain Sciences, Daegu Gyeongbuk Institute of Science and Technology (DGIST), Daegu, Korea Find articles by Byeongchan Kim 1 , Victoria M Hoelscher Victoria M Hoelscher 5 Department of Pharmacology, University of Colorado School of Medicine, Anschutz Medical Campus, Aurora, CO USA Find articles by Victoria M Hoelscher 5 , Beatriz Calleja-Pérez Beatriz Calleja-Pérez 6 Pediatric Primary Care, C.S. Doctor Cirajas, Madrid, Spain Find articles by Beatriz Calleja-Pérez 6 , Hyunsu Jung Hyunsu Jung 7 Brain Science Institute, Korea Institute of Science and Technology (KIST), Seoul, Korea Find articles by Hyunsu Jung 7 , Yeji Yang Yeji Yang 8 Digital Omics Research Center, Korea Basic Science Institute (KBSI), Ochang, Korea 9 Department of Biological Sciences, Korea Advanced Institute of Science and Technology (KAIST), Daejeon, Korea Find articles by Yeji Yang 8, 9 , Hea Ji Lee Hea Ji Lee 8 Digital Omics Research Center, Korea Basic Science Institute (KBSI), Ochang, Korea Find articles by Hea Ji Lee 8 , Jihae Lee Jihae Lee 10 School of Biosystem and Biomedical Science, College of Health Science, Korea University, Seoul, Korea Find articles by Jihae Lee 10 , Seoyeon Kim Seoyeon Kim 11 Department of Integrated Biomedical and Life Science, Korea University, Seoul, Korea 12 BK21 Four R&E Center for Learning Health Systems, Korea University, Seoul, Korea 13 L-HOPE Program for Community-Based Total Learning Health Systems, Korea University, Seoul, Korea Find articles by Seoyeon Kim 11, 12, 13 , Mar Jiménez de la Peña Mar Jiménez de la Peña 14 Neuroimaging, Hospital Universitario Quirónsalud, Madrid, Spain 15 Universidad Europea de Madrid, Madrid, Spain Find articles by Mar Jiménez de la Peña 14, 15 , Yelin Lee Yelin Lee 1 Department of Brain Sciences, Daegu Gyeongbuk Institute of Science and Technology (DGIST), Daegu, Korea Find articles by Yelin Lee 1 , Sohye Kim Sohye Kim 1 Department of Brain Sciences, Daegu Gyeongbuk Institute of Science and Technology (DGIST), Daegu, Korea Find articles by Sohye Kim 1 , Ah-reum Han Ah-reum Han 16 Center for Biomolecular & Cellular Structure, Institute for Basic Science (IBS), Daejeon, Korea Find articles by Ah-reum Han 16 , Dong Sun Lee Dong Sun Lee 16 Center for Biomolecular & Cellular Structure, Institute for Basic Science (IBS), Daejeon, Korea Find articles by Dong Sun Lee 16 , Sangho Ji Sangho Ji 1 Department of Brain Sciences, Daegu Gyeongbuk Institute of Science and Technology (DGIST), Daegu, Korea Find articles by Sangho Ji 1 , Wookyung Yu Wookyung Yu 1 Department of Brain Sciences, Daegu Gyeongbuk Institute of Science and Technology (DGIST), Daegu, Korea 2 Center for Synapse Diversity and Specificity, DGIST, Daegu, Korea Find articles by Wookyung Yu 1, 2 , Ho Min Kim Ho Min Kim 9 Department of Biological Sciences, Korea Advanced Institute of Science and Technology (KAIST), Daejeon, Korea 16 Center for Biomolecular & Cellular Structure, Institute for Basic Science (IBS), Daejeon, Korea Find articles by Ho Min Kim 9, 16 , Joon-Yong An Joon-Yong An 10 School of Biosystem and Biomedical Science, College of Health Science, Korea University, Seoul, Korea 11 Department of Integrated Biomedical and Life Science, Korea University, Seoul, Korea 12 BK21 Four R&E Center for Learning Health Systems, Korea University, Seoul, Korea 13 L-HOPE Program for Community-Based Total Learning Health Systems, Korea University, Seoul, Korea Find articles by Joon-Yong An 10, 11, 12, 13 , Won Chan Oh Won Chan Oh 5 Department of Pharmacology, University of Colorado School of Medicine, Anschutz Medical Campus, Aurora, CO USA Find articles by Won Chan Oh 5 , Seok-Kyu Kwon Seok-Kyu Kwon 7 Brain Science Institute, Korea Institute of Science and Technology (KIST), Seoul, Korea 17 Division of Bio-Medical Science & Technology, KIST School, Korea University of Science and Technology (UST), Daejeon, Korea Find articles by Seok-Kyu Kwon 7, 17 , Jin Young Kim Jin Young Kim 8 Digital Omics Research Center, Korea Basic Science Institute (KBSI), Ochang, Korea 18 College of Pharmacy, Chung-Ang University, Seoul, Korea Find articles by Jin Young Kim 8, 18 , Ji Won Um Ji Won Um 1 Department of Brain Sciences, Daegu Gyeongbuk Institute of Science and Technology (DGIST), Daegu, Korea 2 Center for Synapse Diversity and Specificity, DGIST, Daegu, Korea Find articles by Ji Won Um 1, 2 , Alberto Fernández-Jaén Alberto Fernández-Jaén 18 College of Pharmacy, Chung-Ang University, Seoul, Korea 19 Department of Pediatric Neurology, Hospital Universitario Quirónsalud, Madrid, Spain Find articles by Alberto Fernández-Jaén 18, 19, ✉ , Jaewon Ko Jaewon Ko 1 Department of Brain Sciences, Daegu Gyeongbuk Institute of Science and Technology (DGIST), Daegu, Korea 2 Center for Synapse Diversity and Specificity, DGIST, Daegu, Korea Find articles by Jaewon Ko 1, 2, ✉ Author information Article notes Copyright and License information 1 Department of Brain Sciences, Daegu Gyeongbuk Institute of Science and Technology (DGIST), Daegu, Korea 2 Center for Synapse Diversity and Specificity, DGIST, Daegu, Korea 3 Genomics, Genologica, Málaga, Spain 4 Genomics, NimGenetics, Madrid, Spain 5 Department of Pharmacology, University of Colorado School of Medicine, Anschutz Medical Campus, Aurora, CO USA 6 Pediatric Primary Care, C.S. Doctor Cirajas, Madrid, Spain 7 Brain Science Institute, Korea Institute of Science and Technology (KIST), Seoul, Korea 8 Digital Omics Research Center, Korea Basic Science Institute (KBSI), Ochang, Korea 9 Department of Biological Sciences, Korea Advanced Institute of Science and Technology (KAIST), Daejeon, Korea 10 School of Biosystem and Biomedical Science, College of Health Science, Korea University, Seoul, Korea 11 Department of Integrated Biomedical and Life Science, Korea University, Seoul, Korea 12 BK21 Four R&E Center for Learning Health Systems, Korea University, Seoul, Korea 13 L-HOPE Program for Community-Based Total Learning Health Systems, Korea University, Seoul, Korea 14 Neuroimaging, Hospital Universitario Quirónsalud, Madrid, Spain 15 Universidad Europea de Madrid, Madrid, Spain 16 Center for Biomolecular & Cellular Structure, Institute for Basic Science (IBS), Daejeon, Korea 17 Division of Bio-Medical Science & Technology, KIST School, Korea University of Science and Technology (UST), Daejeon, Korea 18 College of Pharmacy, Chung-Ang University, Seoul, Korea 19 Department of Pediatric Neurology, Hospital Universitario Quirónsalud, Madrid, Spain ✉ Corresponding author. # Contributed equally. Received 2025 Jun 9; Revised 2026 Feb 12; Accepted 2026 Feb 24; Collection date 2026 Apr. © The Author(s) 2026 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ . Creative Commons Public Domain Dedication waiver http://creativecommons.org/publicdomain/zero/1.0/ applies to the data associated with this article, unless otherwise stated in a credit line to the data, but does not extend to the graphical or creative elements of illustrations, charts, or figures. This waiver removes legal barriers to the re-use and mining of research data. According to standard scholarly practice, it is recommended to provide appropriate citation and attribution whenever technically possible. PMC Copyright notice PMCID: PMC13084050  PMID: 41862769 Abstract MDGA1 reportedly suppresses GABAergic synaptic inhibition and may be associated with schizophrenia. However, it has been unclear whether and how MDGA1 dysfunction causes neurodevelopmental disorders. Here, we describe two patients with autism spectrum disorder (ASD) carrying missense mutations in MDGA1 : p.Val116Met/p.Ala688Val and p.Tyr635Cys/p.Glu756Gln. Murine in utero overexpression of MDGA1 p.Val116Met/p.Ala688Val alters normal cortical neuron migration and impairs ultrasonic vocalizations (USVs). The p.Tyr635Cys/p.Glu756Gln substitution disrupts the triangular extracellular structure of MDGA1 and renders it unable to impact GABAergic synapses in hippocampal CA1 neurons. Male Mdga1 knock-in (KI) mouse pups and adults harboring the p.Tyr636Cys/p.Glu751Gln mutation exhibit impaired USVs and sensorimotor gating, similar to male Mdga1 conditional knockout (cKO) mice. No behavioral deficits were seen in female counterparts. Bazedoxifene (a selective estrogen receptor modulator) treatment of male Mdga1 Y636C/E751Q KI mice rescues the changes in the expression and phosphorylation of a subset of GABAergic synaptic proteins, as well as behavioral performance and GABAergic synaptic strength. Thus, different MDGA1 mutations manifest as distinct MDGA1 dysfunctions and are likely to cause ASD via sexually dimorphic loss-of-function and/or gain-of-function mechanisms. Keywords: MDGA1, Autism Spectrum Disorder, Inhibitory Synapse, Communication Deficits, Social Deficits Subject terms: Neuroscience Synopsis The current study identifies MDGA1, a GABAergic synapse-specific suppressor, as a novel neurodevelopmental disorder biomarker, as supported by human genetic and clinical data as well as mouse transgenic models. Distinct MDGA1 mutations lead to varied cellular and behavioral deficits. MDGA1 dysfunction causes sexually dimorphic behavioral impairments. Bazedoxifene reverses MDGA1-mediated ASD-like phenotypes. The current study identifies MDGA1, a GABAergic synapse-specific suppressor, as a novel neurodevelopmental disorder biomarker, a conclusion supported by human genetic and clinical data as well as mouse transgenic models. The paper explained. Problem The prevailing hypothesis is that an excitatory/inhibitory (E/I) imbalance underlies autism spectrum disorder (ASD). However, prior research has focused heavily on glutamatergic synapses, leaving the specific molecular mechanisms of GABAergic synapses in ASD pathology largely underexplored. This limits our comprehensive understanding of how E/I imbalance arises. Results The current study identifies MDGA1, a GABAergic synapse-specific suppressor, as a novel putative neurodevelopmental disorder biomarker, prompted by clinical observations in two ASD patients with distinct MDGA1 missense mutations and further supported by our finding of decreased MDGA1 levels in human ASD postmortem brains. We show that these mutations uniquely affect cortical neuron migration and GABAergic synapse suppression, with transmission electron microscopy results highlighting the requirement for an intact MDGA1 extracellular domain. Using mouse models, we demonstrated that ASD-associated MDGA1 mutations impair specific social behaviors (e.g., altered USV and acoustic startle responses) and developmental stage-specific GABAergic synapse functions in mice, most notably in males. Proteomics also revealed that these mutations are associated with developmental stage-dependent alterations in post-translational modifications of synaptic proteins, particularly presynaptic synapsin II. Crucially, bazedoxifene, an FDA-approved drug, reversed these behavioral, electrophysiological, and biochemical alterations in male Mdga1 -KI mice. Impact This study significantly advances the understanding of ASD etiology by highlighting the critical, underexplored role of GABAergic dysfunction and demonstrating how different ASD-linked MDGA1 substitutions uniquely affect developmental and synaptic processes. Our findings suggest a novel therapeutic strategy for ASD by demonstrating that bazedoxifene, an FDA-approved drug, can rescue communication and startle response impairments in ASD animal models. This work lays essential groundwork for developing targeted therapies for ASD and related neurodevelopmental disorders. Introduction Mounting evidence from human genetic studies implicates synaptic proteins in the pathogenesis of neuropsychiatric disorders, such as autism spectrum disorders (ASDs), intellectual disability, and schizophrenia (Lee et al, 2017 ; Lima Caldeira et al, 2019 ; Torres et al, 2017 ). Thus, these disorders are considered synaptopathies, which researchers have attempted to model in various animal species (Brose et al, 2010 ; Lepeta et al, 2016 ). However, limitations and challenges in producing appropriate animal models of neuropsychiatric disorders still hinder our understanding of the precise mechanisms of the pathogenesis and behavior (Nestler and Hyman, 2010 ; Südhof, 2018 ). Moreover, although a series of converging mechanisms have accounted for ASD-relevant dysfunctions at glutamatergic synapses (Fernandes and Carvalho, 2016 ; Jang et al, 2017 ; Kettenmann et al, 2013 ; Lima Caldeira et al, 2019 ), the molecular and cellular mechanisms underlying GABAergic synaptic dysfunctions remain largely enigmatic. Meprin, A-5 protein, and receptor protein tyrosine phosphatase mu domain-containing glycosylphosphatidylinositol anchor protein 1 (MDGA1) is a cell surface glycoprotein attached to the cell membrane by a glycosylphosphatidylinositol motif (Litwack et al, 2004 ; Ko, 2025 ). It functions throughout the central nervous system (CNS) development (Connor et al, 2019 ; Um and Ko, 2017 ); particularly, its role in regulating cortical neuron migration in the superficial layers has been extensively investigated (Ishikawa et al, 2011 ; Perez-Garcia and O’Leary, 2016 ; Takeuchi et al, 2007 ; Takeuchi and O’Leary, 2006 ). However, MDGA1 continues to be expressed at the adult stage of CNS development (Lee et al, 2013 ), suggesting its putative role in the later phase of CNS development. In keeping with this observation, MDGA1 negatively regulates GABAergic synapse maintenance by binding to a subset of synapse organizers, such as neuroligin-2 (Nlgn2) or amyloid precursor protein (APP) (Kim et al, 2022a ; Lee et al, 2013 ; Pettem et al, 2013 ). Intriguingly, it appears that MDGA1 targets APP to tune specific GABAergic neural circuit properties in the hippocampus (Kim et al, 2022a ). Although the contribution of the Nlgn2-binding activity of MDGA1 to hippocampal synapse organization cannot be ruled out, it is plausible that MDGA1 might deploy distinct synaptic pathways to modulate GABAergic synapse development. MDGA1 has been associated with schizophrenia and bipolar disorder (Kahler et al, 2008 ; Li et al, 2011 ). Moreover, MDGA1 knockout (KO) mice exhibit impaired prepulse inhibition of the startle response with altered dopamine and serotonin metabolism (Hossain et al, 2020 ). In the present study, we identified two missense mutations in MDGA1 in two patients with autistic features. Extensive gain-of-function (GoF) and loss-of-function (LoF) analyses in cultured neurons and hippocampal CA1 neurons revealed that the two MDGA1 variants nullify the ability of MDGA1 wild-type (WT) to regulate neuronal migration and synaptic function. Moreover, the genetic introduction of one of these mutations (p.Tyr636Cys/p.Glu751Gln; Y636C/E751Q; human MDGA1 Glu756 corresponds to mouse MDGA1 Glu751) into mice recapitulated Mdga1 KO phenotypes, including abnormalities in USV. Our study provides novel insights into a potential synaptopathy mechanism contributing to ASD-related phenotypes, highlighting the pathophysiological significance of a previously underappreciated aspect of the structural features of MDGA1. Results Clinical manifestations of the two patients with MDGA1 variants and genetic analyses Two patients with autistic features were referred to our clinical institute. The first patient (P1) is a 5-year-old girl diagnosed with moderate psychomotor impairment with autistic features, according to clinical and neuropsychological evaluations (IQ of 63; ADOS score of 13). On examination, she showed some mild dysmorphic features: frontal bossing, prominent eyes, and a thin tented upper lip. The second patient (P2) is a 5-year-old boy with low intellectual functioning and autistic features, also based on clinical and cognitive assessments (IQ of 71; ADOS score of 19). He displayed a prominent forehead, blepharophimosis, and a thin vermilion of the upper lip. The family histories of both patients were not relevant; their initial neurodevelopmental milestones were significantly delayed. Both patients showed severe delays in verbal and non-verbal communication, as confirmed in their respective neuropsychological evaluations. Routine laboratory screening, neurophysiological tests (electroencephalography and auditory evoked potentials), and conventional genetic studies (karyotype and aCGH arrays) revealed no significant abnormalities. Whole-exome trio analysis revealed compound heterozygous MDGA1 variants in both cases (Table EV1 ). A paternally inherited missense variant (c.2266 G > C; p. Glu756Gln) and a maternally inherited missense variant (c.1904 A>G; p. Tyr635Cys) of MDGA1 were identified in patient P1 (Fig. 1A–D ). Both variants were classified as variants of uncertain significance (VUS). An extremely low frequency was observed for both variants in the Genome Aggregation Database (gnomAD), with a maximal frequency of 0.07% and 0.001%, respectively, and no healthy homozygotes. The gnomAD missense Z-Score of 2.57 (greater than 0.647) and the CADD (Combined Annotation-Dependent Depletion) scores of 32 and 27.2 predicted a deleterious effect. In the proband of patient P2, a paternally inherited missense variant (c.346 G > A; p. Val116Met) and a maternally inherited missense variant (c.2063 C > T; p. Ala688Val) of MDGA1 were observed (Fig. 1A–D ). These variants were classified as variants of uncertain significance and likely benign, respectively. An extremely low frequency was observed for both variants in the gnomAD population database (with a maximal frequency of < 0.001% and 0.07%, respectively, and no healthy homozygotes). Their CADD scores of 23.6 and 21.3 suggested a deleterious effect. These mutations were also observed in a compound heterozygous state in two of his three brothers (one with mild psychomotor impairment and one who is a very young infant). While direct measurement of MDGA1 protein levels in patient-derived biospecimens would be valuable, practical limitations, including challenges in obtaining patient consent, precluded their collection for the current study. Figure 1. Genetic analysis of human MDGA1 missense mutations found in individuals with autism spectrum disorders and the biochemical and cellular properties of these mutations. Open in a new tab ( A ) Pedigree of a family with MDGA1 (c.346 G > A and c.2063 C > T) mutations. Squares represent males, circles represent females, and arrows indicate affected individuals. Individuals with the c.346 G > A (p.Val116Met) mutation are denoted by red symbols, while those with the c.2063 C > T (p.Ala688Val) mutation are denoted by green symbols. ( B ) Sequencing chromatograms showing the MDGA1 (c.346 G > A and c.2063 C > T) mutations. The chromatograms display the nucleotide sequences surrounding the mutation sites. Genomic DNA was isolated from peripheral blood samples of probands and parents, and mutations were verified by Sanger sequencing. ( C ) Pedigree of a family with MDGA1 (c.1904 A>G and c.2266 G > C) mutations. Symbols and color codes are as described in ( A ). Individuals with the c.1904 A>G (p.Glu756Gln) mutation are denoted by red symbols, and those with the c.2266 G > C (p.Tyr635Cys) mutation are denoted by green symbols. ( D ) Sequencing chromatograms showing the MDGA1 (c.1904 A>G and c.2266 G > C) mutations. The chromatograms display nucleotide sequences surrounding the mutation sites. ( E ) Structural model of the hetero-tetrameric human MDGA1/neuroligin complex (PDB: 5OJ6; PDB: 5XEQ) (Elegheert et al, 2017 ; Kim et al, 2017 ) and patient mutations in human MDGA1. For clarity, only the crystal structure of Nlgn2 and the aligned structural model of human MDGA1 are presented. Domains of human MDGA1 (six Ig domains and one FnIII domain) are colored with a rainbow gradient from Ig1 (red) to FnIII (violet). Dimeric Nlgn2 is colored with light cyan. Close-up views of mutated and neighboring residues in human MDGA1 (left). Residues corresponding to patients’ mutations described in the current study are presented as sticks and labeled. ( F ) Diagram illustrating the ASD-associated MDGA1 variants used in the current study. ( G , H ) Immunoblot analyses probing the expression levels of MDGA1 WT and the indicated variants in HEK293T cells ( G ) or cultured hippocampal neurons ( H ). ( I , J ) Analyses of the surface transport activity of HA-tagged MDGA1 WT and its variants in HEK293T cells. ( I ) Representative images of transfected HEK293T cells, illustrating the expression and surface transport activity of MDGA1 variants. Transfected cells were fixed (not permeabilized) and incubated with anti-HA antibody to detect the extracellular region of MDGA1 (red). Intracellular levels of MDGA1 constructs were visualized by incubating cells with an anti-HA antibody after permeabilization (green). Scale bar, 10 μm (applies to all images). ( J ) Quantification of the surface transport activity of MDGA1 constructs. All MDGA1 variants exhibited normal surface expression comparable to WT, indicating that the ASD-associated mutations did not impair surface transport. Data are means ± SEMs (nonparametric Kruskal–Wallis test with Dunn’s post hoc test; n = 10 images/group). ( K , L ) Cell-surface-binding assays. ( K ) Representative images showing HEK293T cells expressing N-terminally HA-tagged MDGA1 WT and its variants incubated with purified Ig-fused neuroligin-2 (Ig-Nlgn2), Ig-fused APP (Ig-APP), or IgC alone (control), as analyzed by immunofluorescence imaging for Ig-fusion proteins (red) and HA (blue). Scale bar, 10 μm (applies to all images). ( L ) Quantification of cell-surface binding. All MDGA1 variants displayed comparable binding to Nlgn2 and APP. Data are means ± SEMs ( n = 12–18 cells/group). Source data are available online for this figure . Segregation analysis confirmed a compound heterozygous state, as both were identified in a heterozygous state in both the mother and father, consistent with an autosomal recessive inheritance; these mutations were confirmed by Sanger sequencing in both cases. In the WES-trio analysis, no other missense variants with a CADD score over 15 or LoF variants with a clear phenotypic association or a compatible segregation pattern were identified. In both cases, brain 3-T MRI did not reveal any significant structural malformations (Appendix Fig. S1 ). However, in patient P1, diffusion tensor imaging with 3D-tractography reconstruction showed a marked asymmetry of the superior longitudinal fasciculus (Appendix Fig. S1 ). Cellular and temporal expression of MDGA1 in human brain development and autism spectrum disorder To investigate MDGA1 expression throughout brain development, we utilized a single-cell atlas of the neurotypical human post-mortem brain (Data ref: Kim et al, 2024b ). The atlas collated eight single-nucleus and single-cell transcriptomic datasets from prior studies, covering the developmental stages across the human lifespan (from gestational week 7–90 years) (Fig. EV1A–C ). Among the 41 clusters comprising the atlas (Fig. EV1C ), MDGA1 showed the highest expression in cluster 12 (C12). This cluster comprises excitatory neurons of early cortical development, mainly fetal second-trimester cortical excitatory neurons (Fig. EV1D ). MDGA1 appears to be highly expressed in immature cortical excitatory neurons, particularly from the mid-fetal period to infancy (gestational week 15 to postnatal week 70) (Fig. EV1E ). We confirmed the excitatory neuron-specific expression of MDGA1 using a previously published single-nucleus atlas (Appendix Fig. S2A,B ) (Data ref: Wang et al, 2025 ), which demonstrated a consistently high level of MDGA1 in excitatory neurons (Appendix Fig. S2C–F ). Specifically, MDGA1 exhibited high-level expression in immature intratelencephalic excitatory neurons (EN-IT-Immature) and layer II/III intratelencephalic excitatory neurons (EN-L2 3-IT) (Appendix Fig. S2G ). Thus, MDGA1 expression exhibits cellular and temporal specificity in excitatory neurons during the early development of the human cortex. Figure EV1. MDGA1 expression in the human brain single-cell atlas. Open in a new tab ( A–D ) Uniform Manifold Approximation and Projection (UMAP) of the human brain single-cell atlas, colored according to primary cell types ( A ), developmental stages ( B ), Leiden clusters ( C ), and z-score normalized expression of MDGA1 ( D ). ( E ) Temporal expression of MDGA1 across developmental stages. Data represent sample-wise mean log-normalized expression computed using a pseudobulk method. Analysis was restricted to clusters with ≥ 4600 cells (C0–C22). ( F ) Gene set enrichment analysis between neurological disorder genes and differentially expressed genes in each cluster. Statistical significance was determined using a Fisher’s exact test. Significant enrichment was defined as a False Discovery Rate (FDR) < 0.05 and a log 2 (odds ratio) > 1.5. Since risk genes for ASD and neurodevelopmental disorders (NDDs) are enriched during early cortical neurodevelopment (Choi and An, 2021 ; Data ref: Satterstrom et al, 2020 ), we performed gene set enrichment tests for neurological genes among the cluster-specific genes. We found that the C12-specific genes ( n = 1632) were significantly enriched for ASD and developmental delay risk genes (False Discovery Rate [FDR] < 0.05; Fisher’s exact test). Additionally, C12 was robustly associated with bipolar risk genes and epilepsy risk genes (Log 2 (OR) > 1, P < 0.05). These observations collectively suggest that MDGA1 may play a role across a spectrum of neuropsychiatric conditions with a shared genetic underpinning (Fig. EV1F ; Dataset EV1 ). Notably, many of the C12-specific genes were previously reported by Satterstrom et al (Data ref: Satterstrom et al, 2020 ) as being neuronal communication genes involved in ASD risk (11/24), suggesting that MDGA1 expressed in immature neurons is relevant to early synaptogenesis. We compared the cell-type-specific MDGA1 expression between ASD patients and control individuals using a single-nucleus atlas composed of 15 ASD patients and 16 controls (Data ref: Velmeshev et al, 2019 ) (Appendix Fig. S3A–C ). We found that layer II/III excitatory neurons (L2/3) and protoplasmic astrocytes (AST-PP) from ASD individuals showed significantly decreased MDGA1 expression levels (FDR < 0.01) (Appendix Fig. S3D ; Dataset EV2 ) compared to those from control subjects. These findings suggest that MDGA1 should be examined in a cell-type-specific manner and that dysregulated MDGA1 expression in a specific cell population may contribute to the pathophysiology of ASD. Normal expression level, surface transport, ligand-binding, and folding of MDGA1 in the MDGA1 mutations of P1 and P2 To investigate the structural effect of the patient mutations on human MDGA1, we predicted the structure of hMDGA1 using AlphaFold2 (Jumper et al, 2021 ; Tunyasuvunakool et al, 2021 ). The p.Val116Met mutation is in the Ig1 domain and two mutations (p.Tyr635Cys and p.Ala688Val) are in the fibronectin type III (FNIII) domain of MDGA1 (Fig. 1E ). Human MDGA1 and neuroligin-2 (Nlgn2) form a hetero-tetrameric complex with 2:2 stoichiometry, and long-range electrostatic interactions and hydrogen bonds enable the strong binding of MDGA1 to the Nlgn2 dimer (Kim et al, 2017 ). The substitution of the small, nonpolar valine with the large methionine (p.Val116Met) in MDGA1, located at the binding interface, is likely to disrupt the interactions between MDGA1 immunoglobulin domain 1 (Ig1) and Nlgn2 by causing steric hindrance with neighboring polar residues (R243 and E247); notably, this mutation was predicted to have little effect on the 3D structure of MDGA1 (Fig. 1E ). We were unable to predict the structural alterations in the p.Glu756Gln mutation located in the MDGA1 MAM domain, the structure of which was not determined by X-ray crystallography (Elegheert et al, 2017 ). The crystal structure of chicken MDGA1 (Ig 1-6 -FNIII) showed that MDGA1 has a triangular shape and the sharp-angled Ig 2 -Ig 3 , Ig 3 -Ig 5 and Ig 6 -FNIII linkers are stabilized by multiple inter-domain contacts (Elegheert et al, 2017 ). hMDGA1 p.Tyr635 corresponding to chicken MDGA1 p.Tyr629 is located at the interface between Ig6 and FNIII (Appendix Fig. S4 ). Therefore, we hypothesized that the hMDGA1 p.Tyr635Cys mutation could structurally weaken these inter-domain interactions, which is also supported by free energy calculations (Table EV2 ), leading to the disruption of the triangular structure of hMDGA1. We performed negative-stain electron microscopy and confirmed the previous structural data (Elegheert et al, 2017 ) that hMDGA1 WT has a closed triangular shape (boxed in blue color), while the majority of particles for thehMDGA1 Y635C/E756Q mutant had a linear or irregular structure (boxed in red color) (Appendix Fig. S5A ). It is likely that the hMDGA1 p.Ala688Val mutation in the FNIII domain might mildly affect the protein stability of the FNIII domain. We next examined whether the ASD-linked MDGA1 mutants exhibit altered secondary structures, and a circular dichroism (CD) spectrum was measured. A strong negative peak near 210 nm of the CD spectra for MDGA1 wild-type (WT) suggests that the main secondary structure of MDGA1 is composed of β-sheets. The CD spectra of the MDGA1 V116M/A688V and MDGA1 Y635C/E756Q variants are similar to that of MDGA1 WT (Appendix Fig. S5B ), indicating that the ASD-linked MDGA1 mutations do not alter the secondary structure of MDGA1. To further probe the impact of amino acid substitutions on the structure and/or synaptic function of MDGA1, we generated mammalian expression vectors encoding the indicated MDGA1 variants (V116M/A688V and Y635C/E756Q), both of which are considered potentially damaging by various silico prediction tools (Table EV1 ). Notably, Y635 and E756 residues in human MDGA1 are conserved in human MDGA2 at equivalent positions. The aforementioned MDGA1 residues (except for A688) are mostly conserved among various vertebrate species, implying their possible functional significance (Appendix Fig. S4 ). To compare the expression levels and intracellular transport properties of the MDGA1 variants versus MDGA1 WT, we expressed HA-tagged variants and WT MDGA1 in human embryonic kidney 293T (HEK293T) cells (Fig. 1F ). Immunoblot analyses of the HEK293T cell lysates (Fig. 1G ) or the cultured hippocampal neuron lysates (Fig. 1H ) expressing HA-tagged MDGA1 WT or the respective variants showed that total protein expression levels of MDGA1 V116M/A688V or MDGA1 Y635C/E756Q were comparable to those of MDGA1 WT (Fig. 1F–H ). We next examined the surface and intracellular protein levels of MDGA1 WT and the mutants in HEK293T cells. MDGA1 V116M/A688V or MDGA1 Y635C/E756Q exhibited similar surface levels as MDGA1 WT (Fig. 1I,J ). To determine whether the indicated MDGA1 mutations affected the interactions with known extracellular ligands (Nlgn2 or APP), we assayed cell-surface binding of recombinant Ig-fusion proteins of MDGA1 (IgC-MDGA1) or IgC alone (negative control) with HEK293T cells expressing HA-tagged Nlgn2 or APP (Fig. 1K,L ). IgC-MDGA1 proteins robustly bound to HEK293T cells expressing HA-Nlgn2 or HA-APP (Fig. 1K,L ). IgC did not exhibit any noticeable binding activity to HEK293T cells (Fig. 1K,L ). We were unable to analyze whether the MDGA1 mutations could alter the synaptic localization of MDGA1 because overexpression of MDGA1 WT and variants at mature-stage cultured hippocampal neurons showed a diffuse distribution along the dendrites of the transfected neurons, in line with prior observations (Kim et al, 2024a ; Toledo et al, 2022 ). Overexpression of MDGA1 Y635C/E756Q abolished the MDGA1 activity in negatively regulating GABAergic synapse maintenance in cultured hippocampal neurons Given previous reports that MDGA1 overexpression specifically reduces the number of GABAergic synapses in cultured hippocampal neurons and hippocampal CA1 pyramidal neurons (Kim et al, 2022a ; Lee et al, 2013 ; Pettem et al, 2013 ), we explored whether the ASD-associated MDGA1 mutations could impair the ability of MDGA1 to negatively regulate the GABAergic synapse maintenance. To this end, we transfected the indicated MDGA1 variants at 7 days in vitro (DIV7) in cultured hippocampal neurons and measured the density of puncta positive for both gephyrin (a marker for GABAergic postsynaptic specialization) and vesicular GABA transporter (VGAT; a marker for presynaptic GABAergic nerve terminals) in the transfected neurons at DIV14 (Fig. 2A,B ). In line with previous reports (Lee et al, 2013 ; Pettem et al, 2013 ), overexpression of MDGA1 WT significantly decreased the density and area of gephyrin + VGAT + puncta. Interestingly, overexpressed MDGA1 V116M/A688V had similar effects, whereas overexpressed MDGA1 Y635C/E756Q failed to alter GABAergic synapse puncta density or area (Fig. 2A,B ). However, both ASD-associated MDGA1 variants effectively suppressed Nlgn2-induced presynaptic assembly in heterologous synapse formation assays, in a manner similar to MDGA1 WT (Appendix Fig. S6 ). Further experiments showed that the E756Q substitution, but not the Y635C substitution, was responsible for abrogating the activity of MDGA1 WT in negatively regulating GABAergic synapse number in cultured neurons (Appendix Fig. S7 ). Since MDGA2 overexpression also increases excitatory synapses (Kim et al, 2024a ), we questioned whether overexpression of MDGA2 Y700C/E820Q would also fail to suppress excitatory synapse density. However, overexpressed MDGA2 Y700C/E820Q displayed similar abilities to suppress excitatory synapses (Appendix Fig. S8A,B ). Figure 2. Analysis of GABAergic synaptic properties in hippocampal CA1 pyramidal neurons expressing ASD-associated MDGA1 variants. Open in a new tab ( A , B ) Representative images ( A ) and summary graphs ( B ) showing the density of synaptic puncta in cultured hippocampal neurons transfected at DIV7 with the indicated full-length MDGA1 expression constructs and immunostained at DIV14 with antibodies to VGAT, gephyrin and EGFP. Data are presented as means ± SEMs ( n = 14–15 neurons/group; * P = 0.0135 (vs. WT), * p = 0.0451 (vs. V116M/A688V); Kruskal–Wallis test with Dunn’s post hoc test). Scale bar, 10 μm (applies to all images). ( C – G ) Representative mIPSC traces ( C ) and quantification of frequency ( D ), amplitude ( E ), rise time ( F ), and decay time ( G ) of mIPSCs from cultured neurons transfected with MDGA1 WT or its variants. Data are presented as means ± SEMs (control, n = 19; WT, n = 22; V116M/A688V, n = 25; Y635C/E756Q, n = 24; * P = 0.0226 (vs. WT), *** P = 0.0001 (vs. V116M/A688V); Kruskal–Wallis test with Dunn’s post hoc test). ( H–J ) Representative traces ( H ) of electrophysiological recordings in CA1 pyramidal neurons expressing MDGA1 WT and its variants recorded in 7- to 8-week-old mice, 3 weeks after AAV-mediated expression of the indicated MDGA1 variants in CA1 neurons. Quantification of mIPSC frequency ( I ) and amplitude ( J ) recorded in the hippocampal CA1 pyramidal neurons. Data are presented as means ± SEMs ( n = 16–37 cells/group; * P =  0.0033 (vs. WT), * P =  0.0280 (vs. V116M/A688V); Kruskal–Wallis test with Dunn’s post hoc test). ( K – N ) Experimental setup for recording eIPSCs in hippocampal CA1 pyramidal neurons by placing stimulating electrodes in the stratum lacunosum-moleculare (SLM) layer. Representative traces ( L ) of eIPSCs obtained from dendritic recordings in CA1 pyramidal neurons. Average eIPSC I-O curve ( M ), and average eIPSC I-O slope ( N ) from hippocampal CA1 pyramidal neurons expressing the indicated MDGA1 variants. Data are presented as means ± SEMs ( n = 12–27 cells/group; ( M ) *** P = 0.0004 (10 μA), ** P = 0.0134 (20 μA), ** P = 0.0091 (30 μA), ** P = 0.0065 (40 μA), ** P = 0.0101 (50 μA), ** P = 0.0086 (60 μA) vs. WT; *** P = 0.0002 (10 μA), *** P = 0.0003 (20 μA), *** P = 0.0003 (30 μA), *** P = 0.0010 (40 μA), *** P = 0.0009 (50 μA), *** P = 0.0007 (60 μA) vs. VM/AV; ( N ) * P = 0.0445 (vs. WT), ** P = 0.0052 (vs. VM/AV); Kruskal–Wallis test with Dunn’s post hoc test). ( O , P ) Representative somatic eIPSC paired-pulse ratio (PPR) traces ( O ) and average PPR ( P ) from hippocampal CA1 pyramidal neurons expressing the indicated MDGA1 variants. Data are presented as means ± SEMs ( n = 12–27/group; ( P ) ** P = 0.0016 (vs. WT), ** P = 0.0001 (vs. VM/AV); Kruskal–Wallis test with Dunn’s post hoc test). ( Q ) Schematic showing biolistic transfection of organotypic hippocampal CA1 slices with MDGA1 mutants at EP4–5 and imaging at EP18–24. ( R ) Two-photon images of dendritic segments from CA1 pyramidal neurons co-transfected with tdTomato, gephyrin intrabody-GFP, and MDGA1 control, MDGA1 WT, MDGA1 V116M/A688V, or MDGA1 Y635C/E756Q vector. ( S , T ) Quantitative analysis of gephyrin puncta density and size ( S ) (puncta/10-μm dendrite) and dendritic spine density and size ( T ) (spines/10-μm dendrite) in CA1 pyramidal neurons expressing MDGA1 WT or its variants. Data are presented as means ± SEMs ( n = 12–17 cells/group; ( S ) ** P = 0.0088 (puncta density; vs. WT), ** P = 0.0127 (puncta density; vs. VM/AV), ** P = 0.0019 (puncta size; vs. YC/EQ); ( T ) *** P = 0.0005 (spine density; vs. VM/AV), * P = 0.0289 (spine size; vs. VM/AV); Kruskal–Wallis test with Dunn’s post hoc test). Source data are available online for this figure . To test whether the ASD-associated MDGA1 mutations could also affect the MDGA1-mediated negative regulation of GABAergic synaptic transmission, we measured miniature inhibitory postsynaptic currents (mIPSCs) using whole-cell voltage-clamp recordings. Overexpression of MDGA1 WT significantly decreased the mIPSC frequency without altering the mIPSC amplitude, rise time or decay time of mIPSCs, compared to those from neurons expressing the control plasmid (Fig. 2C–G ). Overexpressed MDGA1 Y635C/E756Q had a similar effect, whereas overexpressed MDGA1 V116M/A688V did not alter GABAergic synaptic transmission (Fig. 2C–G ). However, overexpression of MDGA2 Y700C/E820Q significantly decreased the frequency of miniature excitatory postsynaptic currents (mEPSCs), in keeping with the two-photon imaging results (Appendix Fig. S8C–G ). Taken together, these data suggest that the MDGA1 Y635C/E756Q substitutions manifested as MDGA1 LoF mutations in the regulation of GABAergic synaptic properties. Overexpressed MDGA1 Y635C/E756Q fails to negatively regulate GABAergic synaptic properties in hippocampal CA1 pyramidal neurons We recently showed that overexpression of MDGA1 in the adult mouse hippocampal CA1 suppresses GABAergic synaptic transmission and stabilization (Kim et al, 2022a ). To determine whether our ASD-associated MDGA1 mutations could affect the ability of MDGA1 to negatively regulate the GABAergic synaptic properties, we injected adeno-associated viruses (AAVs) expressing MDGA1 WT or the indicated MDGA1 variants into the hippocampal CA1 of adult mice and performed whole-cell patch-clamp recordings to measure the GABAergic synaptic transmission and strength. Consistent with our previous report (Kim et al, 2022a ), overexpression of MDGA1 WT significantly decreased the frequency (but not amplitude) of mIPSCs in the hippocampal CA1 pyramidal neurons (Fig. 2H–J ). Overexpression of MDGA1 V116M/A688V had a similar effect, whereas that of MDGA1 Y635C/E756Q failed to decrease the mIPSC frequency (Fig. 2H–J ). To examine whether MDGA1 has a similar impact on GABAergic synaptic transmission in the medial prefrontal cortex (mPFC), we measured mIPSCs in layer II/III of mPFC pyramidal neurons. Surprisingly, overexpression of MDGA1 WT and its ASD-associated variants did not alter the frequency or amplitude of mIPSCs nor the densities of gephyrin puncta or dendritic spines (Appendix Fig. S9 ). We next examined whether the ASD-associated MDGA1 mutations impacted the ability of MDGA1 to suppress evoked inhibitory postsynaptic currents (eIPSCs) in the distal dendritic compartment of hippocampal CA1 pyramidal neurons (Kim et al, 2022a ). We found that overexpression of MDGA1 WT or MDGA1 V116M/A688V significantly decreased the amplitude of distal dendritic eIPSCs and increased the paired-pulse ratio (PPR), whereas MDGA1 Y635C/E756Q failed to do so (Fig. 2K–P ). In contrast, overexpression of MDGA1 WT, V116M/A688V, or Y635C/E756Q had no effect on the frequency or amplitude of asynchronous IPSCs (aIPSCs) (Appendix Fig. S10 ). To corroborate our electrophysiological observations, we used two-photon laser-scanning microscopy to examine gephyrin puncta in hippocampal CA1 slice cultures (Fig. 2Q ). Our results revealed that overexpression of MDGA1 WT or MDGA1 V116M/A688V significantly decreased the gephyrin puncta density, whereas overexpression of MDGA1 Y635C/E756Q did not; in fact, overexpression of the latter variant tended to increase the gephyrin puncta size (Fig. 2R,S ). Unexpectedly, overexpression of MDGA1 V116M/A688V, but not MDGA1 WT or MDGA1 Y635C/E756Q, significantly reduced the density of dendritic spines while concomitantly increasing their size (Fig. 2R,T ). These results were further reinforced by data obtained from cultured hippocampal neurons, wherein overexpression of MDGA1 V116M/A688V, but not MDGA1 WT or MDGA1 Y635C/E756Q, specifically decreased the excitatory synapse density and mEPSC frequency (Appendix Fig. S11 ). Our findings collectively indicate that the Y635C/E756Q substitution renders MDGA1 unable to control GABAergic synapse organization in the hippocampal CA1, whereas the MDGA1 V116M/A688V substitution affects excitatory synaptic structures. In vivo overexpression of MDGA1 V116M/A688V, but not MDGA1 Y635C/E756Q, alters cortical neuron migration and ultrasonic vocalizations in mice In patients with ASD, the marked neuropathological characteristics include defective neuronal migration (Pan et al, 2019 ). Intriguingly, Mdga1 knockdown in migrating layer II/III cortical neurons can disrupt their normal migration (Takeuchi and O’Leary, 2006 ). Moreover, transient glutamatergic synapse formation between subplate neurons and immature migratory neurons instructs neocortical neuronal migration (Ohtaka-Maruyama et al, 2018 ), reminiscent of the excitatory synaptic phenotype of MDGA1 V116M/A688V (see Fig. 2 ; Appendix Fig. S11 ). To investigate whether our ASD-associated MDGA1 variants could alter cortical neuron migration, we performed in utero electroporation experiments. Expression plasmids encoding HA-tagged MDGA1 (WT, V116M/A688V or Y635C/E756Q) or a control plasmid (pCIG2 harboring an internal ribosome entry site-driven enhanced green fluorescent protein [EGFP] sequence) were electroporated into mice at E15.5 to mainly target progenitors of layer II/III cortical pyramidal neurons, but not GABAergic interneurons (Fig. EV2A ), as previously described (Meyer-Dilhet and Courchet, 2020 ). At P21, neuronal positions in the electroporated mice were examined by EGFP fluorescence (Fig. EV2A ). Neurons expressing MDGA1 WT or MDGA1 Y635C/E756Q exhibited patterns of migration across cortical layers similar to those of neurons expressing the control plasmid (> 95% of neurons reached layers II/III) (Fig. EV2B,C ). Strikingly, neurons expressing MDGA1 V116M/A688V displayed abnormally altered migration: only ~55% of the transfected neurons were seen in layer II/III, and a large portion was detected in layer V/VI/white matter (~44.5%) (Fig. EV2B,C ). These results suggest that the V116M/A688V substitution might have a dominant-negative inhibitory effect on MDGA1 function, leading to abnormal cortical neuron migration. Figure EV2. Analysis of cortical neuronal migration and ultrasonic vocalization in mice expressing ASD-associated MDGA1 variants. Open in a new tab ( A ) Schematic of the in utero electroporation. The indicated plasmids (control or HA-tagged MDGA1 variants [HA-MDGA1-IRES-EGFP]) were electroporated in utero at E15.5 and neuronal migration was analyzed by immunohistochemistry using confocal microscopy at P21. ( B , C ) Analysis of the migration trajectory of cortical neurons expressing the indicated MDGA1 variants. Expression of the V116M/A688V variant impaired neuronal migration compared to that seen in control neurons or those expressing MDGA1 WT. Data are presented as means ± SEMs (‘ n ’ denotes number of images/mice; control, n = 18/9; WT, n = 10/5; V116M/A688V, n = 6/3; and Y635C/E756Q, n = 5/3; **** P < 0.0001 (II-IV), **** P < 0.0001 (V/VI/WM); Mann–Whitney U test). Scale bar, 500 µm. ( D ) Schematic of the experimental setup for pup isolation and USV recording. USVs were measured at P3, P6, P9, and P12 from WT pups electroporated in utero with control or MDGA1 variant plasmids. ( E ) Number of USV calls recorded from pups expressing the control or the indicated MDGA1 variant at P3, P6, P9 and P12. Data are presented as means ± SEMs ( n = 7–9 pups/group; asterisks (*) denote significant differences between control and the VM/AV group; * P = 0.0134 (P3), * P = 0.0338 (P6), * P = 0.0385 (P9); nonparametric Kruskal–Wallis test followed by Dunn’s post hoc test). ( F ) Duration of USV calls recorded at the same time points. No significant differences were observed among groups. Data are presented as means ± SEMs ( n = 7–9 pups/group). In view of the observation that P1 and P2 patients harboring the MDGA1 missense mutations commonly exhibit communication abnormalities, we evaluated maternal isolation-induced USVs, which are distress calls emitted by pups when separated from their mother (representing infant-mother vocal communicative behavior relevant to ASD) (Kazdoba et al, 2016 ; Premoli et al, 2021 ). In utero electroporation was performed for control, MDGA1 WT, MDGA1 V116M/A688V, or MDGA1 Y635C/E756Q vectors, and maternal isolation-induced USVs of pups were recorded on postnatal day 3 (P3), P6, P9, and P12 (Fig. EV2D,E ). Call numbers in all pups increased from P3 to P6, decreased slightly by P9, and reached zero by P12 (Fig. EV2E ). Intriguingly, MDGA1 V116M/A688V-expressing mice emitted significantly fewer USVs compared to age-matched pups expressing control, MDGA1 WT, or MDGA1 Y635C/E756Q at all ages (up to P12) (Fig. EV2E ). Mdga1 -cKO mice exhibit various autistic-like abnormal behaviors Given that the overexpression of ASD-associated MDGA1 variants revealed functionally divergent outcomes depending on the mutation, we next investigated the effects of the complete loss of MDGA1 in forebrain excitatory neurons. To this end, we utilized Emx1 -Cre:: Mdga1 f/f mice (also denoted as Mdga1 -cKO mice) as a conditional knockout model. Since MDGA1 is highly enriched in excitatory neurons of the cortex and hippocampus in both mice and humans (Lee et al, 2013 ; see Appendix Fig. S2 ), this excitatory neuron-specific deletion model was selected for downstream behavioral and molecular phenotyping, as employed in our previous study (Kim et al, 2022b ; Han et al, 2024 ). Notably, previous studies reported that constitutive Mdga1 -KO mice exhibit impaired hippocampus-dependent spatial learning and memory, along with a reduced startle response—phenotypes often regarded as schizophrenia-related endophenotypes (Connor et al, 2017 ; Hossain et al, 2020 ; Turetsky et al, 2007 ). Regarding other autistic-like behavioral phenotypes, adult male (P60) Mdga1 -cKO mice exhibited levels of anxiety/exploration-related behavior, locomotor activity, working memory, and repetitive/compulsive-like behavior within the expected ranges (Fig. 3A ; Appendix Fig. S12 ). However, adult male Mdga1 -cKO mice exhibited deficits in social behavior, as evident from their reduced time spent exploring animated mice over an object (sociability) and reduced time spent exploring unfamiliar mice (social novelty recognition) in the three-chamber test (Fig. 3D–F ). These mice also emitted reduced USVs upon encountering a female mouse, as well as showing a shift in the distribution of syllable types in the USVs (Fig. 3G–J ). Notably, even at the early postnatal stage, Mdga1 -cKO pups exhibited decreased maternal isolation-induced USVs, indicating impairments in early communicative behavior (Fig. 3B,C ). Impaired sensorimotor gating was observed in adult male Mdga1 -cKO mice, as characterized by decreases in the acoustic startle response and prepulse inhibition (Fig. 3K–M ), as well as decreased novel object-recognition memory (Fig. 3N–P ). The home-cage activities of these mice were unaltered (Fig. 3Q–V ). Figure 3. Analysis of mouse behavior in male adult Mdga1 -cKO mice. Open in a new tab ( A ) Breeding scheme followed to generate Mdga1 -cKO mice. Emx1 -Cre driver mice were crossed with Mdga1 -floxed mice. The timeline of behavioral experiments performed on male Mdga1 -cKO mice, including Y-maze, open field test (OFT), novel object recognition (NOR), three-chamber, light-dark transition (LDT), elevated plus maze (EPM), tail suspension test (TST), forced swim test (FST), social isolation-induced ultrasonic vocalization (USV), and prepulse inhibition (PPI) tests. ( B ) Experimental setup for recording USVs from isolated pups. Pups were separated from their dams and recorded for 5 min on clean bedding with a heat pad (37 °C) using a condenser microphone. ( C ) Quantification of USV call numbers from control and Mdga1 -cKO pups at PND3, PND6, PND9 and PND12. Data are presented as means ± SEMs ( n = 7–9 pups/group from ≥ 3 litters; * P = 0.0110 (PND3), * P = 0.0278 (PND6), * P = 0.0187 (PND12); Mann–Whitney U test). ( D – F ) Three-chamber social interaction test design ( D ), and quantification of sociability ( E ) and social recognition ( F ). In the sociability phase ( E ), Mdga1 -cKO mice exhibited a significantly reduced discrimination index compared to controls, indicating impairment of social preference. In the recognition phase ( F ), Mdga1 -cKO mice failed to prefer a novel stranger over a familiar one, suggesting deficits in social recognition. Data are means ± SEMs ( n = 13–16 mice/group; ( E ) *** P = 0.0002 (Ctrl/sociability), **** P < 0.0001 ( Mdga1 -cKO/sociability), * P = 0.0151 (Discrimination index); ( F ) * P = 0.0398 (Ctrl/social recognition); Wilcoxon matched-pairs signed rank test). ( G – J ) Representative USV traces ( G ), quantification of USV call number ( H ) and duration ( I ), and distribution of call types ( J ; complex trill, downward ramp, flat, inverted-U, short, split, step down, step up, trill, upward ramp) in adult male mice recorded under social isolation conditions. Mdga1 -cKO mice emitted fewer and shorter calls, with the altered proportions of call types including reductions in downward ramp calls. Data are means ± SEMs ( n = 12–15 mice/group; ( H ) * P = 0.0414; ( I ) * P = 0.0155; ( J ) * P = 0.0355 ; Mann–Whitney U test). ( K – M ) Acoustic startle response and prepulse inhibition (PPI) test. In the PPI paradigm ( K ), mice were presented with a prepulse (73–82 dB) followed by a 120-dB startle pulse. Mdga1 -cKO mice displayed reduced startle response ( L ) and reduced PPI at multiple prepulse intensities ( M ), suggesting the presence of sensorimotor gating deficits. Data are means ± SEMs ( n = 12–18 mice/group; ( L ) * P = 0.0275 (95 dB), ** P = 0.0091 (105 dB), ** P = 0.0013 (110 dB), *** P = 0.0007 (115 dB), *** P = 0.0009 (120 dB); ( M ) ** P = 0.0070 (79 dB), * P = 0.0442 (82 dB); Mann–Whitney U test). ( N–P ) NOR test schematic ( N ) showing the object configurations used during the training and test phases. Mdga1 -cKO mice showed reductions in the discrimination index and exploration of the novel object ( O , P ), consistent with an impairment of recognition memory. Data are means ± SEMs ( n = 12–18 mice/group; ( P ) *** P = 0.0003 (Exploration time), * P = 0.0246 (Discrimination index); Wilcoxon matched-pairs signed rank test). ( Q – V ) Quantification of spontaneous behaviors in home cages using the LABORAS system. Measurements of food and water intake ( Q ), locomotion ( R ), climbing ( S ), rearing ( T ), grooming ( U ), and undefined behaviors ( V ) were recorded over 24 h in 2-h bins. The light and dark cycles of the day are represented by sun and moon icons, respectively. Data are means ± SEMs ( n = 6 mice/group). Source data are available online for this figure . Given that there is a strong male bias in human ASDs (Bolte et al, 2023 ) and many ASD animal models bear sexually dimorphic behavioral abnormalities (Bruce et al, 2021 ; Werling, 2016 ), we further tested whether adult female Mdga1 -cKO mice displayed similar autistic-like behavioral phenotypes (Fig. EV3 ). Indeed, adult female Mdga1 -cKO mice exhibited normal sociability, USV numbers, USV syllable compositions, sensorimotor gating, and novel object-recognition memory (Fig. EV3 ). Also, unlike their male counterparts, adult female Mdga1 -cKO mice showed reduced anxiety-like behavior (Fig. EV3D,E ). These results suggest that MDGA1 could be a potential sex-specific biomarker for ASDs. Figure EV3. Analysis of behaviors of adult female Mdga1 -cKO mice. Open in a new tab ( A – C ) Three-chamber test results showing a schematic of the test ( A ) and the exploration time and discrimination index during the sociability ( B ; stranger 1 vs. empty) and social recognition ( C ; stranger 1 vs. stranger 2) phases for female control and Mdga1 -cKO mice. Mdga1 -cKO females showed normal sociability and intact social recognition, as indicated by comparable exploration times and discrimination indices relative to controls. Data are presented as means ± SEMs ( n = 12 mice/group; ( B ) *** P = 0.0010 (Ctrl), *** P = 0.0005 (M1-cKO); ( C ) *** P = 0.0015 (Ctrl), *** P = 0.0005 (M1-cKO); Mann–Whitney U test). ( D , E ) Open-field test results showing total distance traveled, velocity, and time spent in the center for female control and Mdga1 -cKO mice. Mdga1 -cKO mice exhibited increased time spent in the center zone, while total distance traveled and velocity were comparable to those of controls. Data are presented as means ± SEMs ( n = 15–20 mice/group; ( E ) * P = 0.0105; Mann–Whitney U test). ( F ) Schematic of the prepulse inhibition (PPI) test. Mice were exposed to a prepulse sound (73–82 dB) followed by a startle pulse (120 dB), and the inhibition of the startle response was measured. ( G ) Acoustic startle response in female control and Mdga1 -cKO mice. Data are presented as means ± SEMs ( n = 17–18 mice/group). ( H ) PPI of the acoustic startle response in female control and Mdga1 -cKO mice. Data are presented as means ± SEMs ( n = 14–17 mice/group). ( I – K ) Novel object recognition test results showing a schematic of the test ( I ) and the exploration time and discrimination index during the training ( J ) and testing phases ( K ) for female control and Mdga1 -cKO mice. Quantification of exploration times during training ( J ) and testing ( K ) revealed no significant differences in the discrimination index between groups, indicating that recognition memory was intact in Mdga1 -cKO mice. Data are presented as means ± SEMs ( n = 13–18 mice/group; ( K ) **** P < 0.0001 (Ctrl), ** P = 0.0034 (M1-cKO); Wilcoxon matched-pairs signed rank test). ( L ) Marble burying test results showing the number of marbles buried by female control and Mdga1 -cKO mice. The number of buried marbles was not significantly different between Mdga1 -cKO and control mice, suggesting that there was no change in repetitive behavior. Data are presented as means ± SEMs ( n = 11–14 mice/group). ( M – Q ) USV results showing a representative sonogram ( M ) and the number of calls ( N ) and call length ( O ) for female control and Mdga1 -cKO mice. Representative waveforms ( P ) serve as standardized reference samples for each USV category and are utilized across figures (e.g., Fig. EV4P ) as a consistent visual index for classification criteria. Distribution of different call types ( Q ) is shown. No significant differences in call-type proportions were observed, suggesting that vocal communication was intact. Data are presented as means ± SEMs ( n = 18–20 mice/group; Mann–Whitney U test). ( R – W ) Quantification of spontaneous behaviors using the LABORAS system over a 24-h period. No significant differences were observed in food and water intake ( R ), locomotion ( S ), climbing ( T ), rearing ( U ), grooming ( V ), or undefined behaviors ( W ) between Mdga1 -cKO and control mice. Data are presented as means ± SEMs ( n = 6 mice/group). Adult male Mdga1 Y636C/E751Q knockin mice exhibit autistic-like behavioral abnormalities that are distinct from those of adult male Mdga1 -cKO mice Our previous functional analyses revealed that the two ASD-associated MDGA1 substitutions induce abnormalities in distinct facets of neuronal and synaptic developmental processes. However, in utero overexpression of MDGA1 Y635C/E756Q did not significantly impair pup isolation-induced USVs or affect neuronal migration (Fig. EV2 ). We further investigated whether this substitution could influence behaviors in adult mice (Fig. 4A ). To this end, we generated Mdga1 knock-in (KI) mice carrying Y636C and E751Q (Tyr636 and Glu751 in mouse Mdga1 correspond to Tyr635 and Glu756 in human MDGA1) using the clustered regularly interspaced short palindromic repeats (CRISPR)/Cas9 system. Our results showed that Mdga1 Y636C/E751Q pups exhibited reduced numbers of USV calls throughout development (Fig. 4B,C ), similar to Mdga1 -cKO pups. In addition, homozygous Mdga1 Y636C/E751Q pups showed more severe impairment than their heterozygous counterparts in USV (Fig. 4B,C ). This discrepancy may reflect a dosage-dependent effect of the mutant protein, as behavioral impairment was more pronounced in homozygous KI mice compared to heterozygotes, whereas in utero electroporation likely resulted in partial overexpression against a background of endogenous wild-type MDGA1, potentially masking the phenotype (Fig. EV2 ). Figure 4. Analysis of mouse behavior in adult male Mdga1 Y636C/E751Q KI mice. Open in a new tab ( A ) Breeding scheme used to generate Mdga1 Y636C/E751Q KI mice and timeline of behavioral testing. Behavioral assays, including the Y-maze, open field test (OFT), novel object recognition (NOR), three-chamber social test, light-dark transition (LDT), elevated plus maze (EPM), tail suspension test (TST), forced swim test (FST), social isolation-induced USV, and prepulse inhibition (PPI), were performed in adult male mice starting at 8 weeks of age. ( B ) Schematic of the setup used to record USVs in pups. The pups were separated from the dam and placed on clean bedding with a heating pad (37 °C). USVs were recorded for 5 min using a condenser microphone. ( C ) Number of USV calls recorded from Mdga1 Y636C/E751Q KI mice and control pups at PND3, PND6, PND9, and PND12. KI pups emitted significantly fewer USVs at PND6 and PND9, suggesting the presence of early-life communication deficits. Data are means ± SEMs ( n = 9–13 pups/group; * P = 0.0278 (PND6), ** P = 0.0023 (PND9); Mann–Whitney U test). ( D – F ) Schematic of the three-chamber social interaction assay ( D ), and quantification of sociability ( E ) and social recognition ( F ). Data are means ± SEMs ( n = 9–10 mice/group; ( E ) ** P = 0.0039 (Ctrl/sociability), ** P = 0.0020 (M1-KI/sociability); ( F ) * P = 0.0391 (Ctrl/social recognition), * P = 0.0371 (M1-KI/social recognition); Wilcoxon matched-pairs signed rank test). ( G – J ) Representative USV traces ( G ), quantification of USV call number ( H ) and duration ( I ), and distribution of call types ( J ; complex trill, downward ramp, flat, inverted-U, short, split, step down, step up, trill, upward ramp) in adult male mice recorded under social isolation conditions. Mdga1 Y636C/E751Q KI mice emitted a significantly greater number of USV calls compared to controls, but with altered call-type composition—specifically, a decrease in complex trill calls and an increase in step-down calls—suggesting that there is a shift in the qualitative nature of vocal communication. Data are means ± SEMs ( n = 12–15 mice/group; ( H ) * P = 0.0220; ( J ) * P = 0.0327 (CT), * P = 0.0377 (Step Down) ; Mann–Whitney U test). ( K – M ) Schematic of the PPI paradigm ( K ). Mice were presented with a prepulse (73–82 dB) followed by a 120-dB startle pulse. Mdga1 Y636C/E751Q KI mice exhibited reduced startle responses ( L ) and significantly enhanced PPI at higher prepulse intensities ( M ), indicating the presence of sensorimotor gating deficits. Data are means ± SEMs ( n = 12–13 mice/group; ( L ) * P = 0.0156 (115 dB), * P = 0.0156 (120 dB); ( M ) * P = 0.0206 (79 dB), * P = 0.0300 (82 dB); Mann–Whitney U test). ( N – P ) Schematic of the NOR test ( N ), showing object placements during the training and testing phases. Testing occurred 24 h after the training session. Mdga1 Y636C/E751Q KI mice exhibited exploration times and discrimination indices comparable to those of control mice ( O, P ), indicating that they had intact recognition memory. Data are means ± SEMs ( n = 7–10 mice/group; ( P ) ** P = 0.0020 (Ctrl), * P = 0.0156 (M1-KI); Wilcoxon matched-pairs signed rank test). ( Q – V ) Quantification of spontaneous behaviors in home cages using the LABORAS system. Food and water intake ( Q ), locomotion ( R ), climbing ( S ), rearing ( T ), grooming ( U ), and undefined behaviors ( V ) were recorded over 24 h in 2-h bins. The light and dark cycles of the day are represented by sun and moon icons, respectively. Data are means ± SEMs ( n = 6 mice/group). Source data are available online for this figure . Next, we subjected adult male Mdga1 Y636C/E751Q KI mice to a battery of behavioral tests similar to those used for Mdga1 -cKO mice. Both adult male Mdga1 Y636C/E751Q KI and Mdga1 -cKO mice were viable and fertile, exhibited no obvious abnormalities, morbidity, or premature mortality, and were comparable to control mice in brain weight and size (Appendix Fig. S13 ). Male Mdga1 Y636C/E751Q KI mice showed normal anxiety/exploration-related behavior, locomotor activity, working memory, and repetitive/compulsive-like behavior (Appendix Fig. S14 ). In addition, male Mdga1 Y636C/E751Q KI mice exhibited normal sociability and object recognition memory (Fig. 4D–F,N–P ); however, similar to their male cKO counterparts, male Mdga1 Y636C/E751Q KI mice encountering female mice exhibited an altered syllable composition in their USV calls with an increased USV call number (Fig. 4G–J ). Intriguingly, male Mdga1 Y636C/E751Q KI mice exhibited a decreased acoustic startle response (ASR) but increased prepulse inhibition (Fig. 4K–M ) and thereby differed from the male Mdga1 -cKO mice (Fig. 3 ). Overall, male Mdga1 Y636C/E751Q KI mice exhibited a narrower range of behavioral impairments than male Mdga1 -cKO mice, particularly in sensorimotor gating and sociability, suggesting that Mdga1 Y636C /E751Q KI might not compromise all the MDGA1-mediated functions. Again, the female Mdga1 Y636C/E751Q KI mice did not exhibit any of the behavioral deficits seen in their male KI counterparts, further highlighting the sexually dimorphic manifestation of MDGA1 dysfunctions (Fig. EV4 ). Figure EV4. Analysis of behaviors of adult female Mdga1 Y636C/E751Q KI mice. Open in a new tab ( A – C ) Three-chamber test results showing a schematic of the test ( A ) and the exploration time and discrimination index during the sociability ( B ; stranger 1 vs. empty) and social recognition ( C ; stranger 1 vs. stranger 2) phases for female control and Mdga1 Y636C/E751Q KI mice. Mdga1 Y636C/E751Q KI females showed normal sociability and intact social recognition, as indicated by comparable exploration times and discrimination indices relative to those of controls. Data are means ± SEMs ( n = 11–13 mice/group; ( B ) *** P = 0.0010 (Ctrl), *** P = 0.0002 (M1-KI); ( C ) ** P = 0.0068 (Ctrl), *** P = 0.0005 (M1-KI); Wilcoxon matched-pairs signed rank test). ( D , E ) Open-field test results showing total distance traveled, velocity, and time spent in the center for female control and Mdga1 Y636C/E751Q KI mice. Mdga1 Y636C/E751Q KI mice exhibited decreased time spent in the center zone, while the total distance traveled and velocity were comparable to those of controls. Data are presented as means ± SEMs ( n = 11–13 mice/group; ( E ) * P = 0.0105, Mann–Whitney U test). ( F–H ) Schematic ( F ) of the prepulse inhibition (PPI) test. Mice were exposed to a prepulse sound (73–82 dB) followed by a startle pulse (120 dB), and the inhibition of the startle response was measured. Acoustic startle response ( G ) in female control and female Mdga1 Y636C/E751Q KI mice. Data are presented as means ± SEMs ( n = 11–14 mice/group). PPI of the acoustic startle response ( H ) in female control and Mdga1 Y636C/ E751Q KI mice. Data are presented as means ± SEMs ( n = 11–14 mice/group). ( I – K ) Novel object recognition test results showing a schematic of the test ( I ) and the exploration time and discrimination index during the training ( J ) and testing phases ( K ) for female control and Mdga1 Y636C/E751Q KI mice. Quantification of exploration times during training ( J ) and testing ( K ) revealed no significant difference in the discrimination index between groups, indicating that recognition memory was intact in Mdga1 Y636C/E751Q KI mice. Data are presented as means ± SEMs ( n = 7–9 mice/group; ( K ) * P = 0.0156 (Ctrl), ** P = 0.0078 (M1-KI); Wilcoxon matched-pairs signed rank test). ( L ) Marble burying test results showing the number of marbles buried by female control and female Mdga1 Y636C/E751Q KI mice. The number of buried marbles was not significantly different between Mdga1 Y636C/E751Q KI and control mice, suggesting that there was no change in repetitive behavior. Data are presented as means ± SEMs ( n = 8 mice/group). ( M – Q ) USVs recorded from adult female mice under social isolation. Representative sonograms ( M ), quantification of USV call number ( N ) and duration ( O ), and representative waveforms ( P ) are shown. Note that the waveforms in ( P ) serve as standardized reference samples for each USV category and are utilized across figures (e.g., Fig. EV3P ) as a consistent visual index for classification criteria. Distribution of call types ( Q ) was assessed; no significant differences in call-type proportions were observed, suggesting that vocal communication was intact. Data are presented as means ± SEMs ( n = 16–18 mice/group). ( R – W ) Quantification of spontaneous behaviors using the LABORAS system over a 24-h period. No significant differences were observed in food and water intake ( R ), locomotion ( S ), climbing ( T ), rearing ( U ), grooming ( V ), or undefined behaviors ( W ) between Mdga1 Y636C/E751Q KI and control mice. Data are presented as means ± SEMs ( n = 8 mice/group). Adult male Mdga1 Y636C/E751Q KI mice phenocopy the electrophysiological abnormalities of Mdga1 -cKO mice We additionally analyzed the levels of MDGA1 and other synaptic proteins in Mdga1 Y636C/E751Q KI and cKO mice. Quantitative RT-PCR analyses revealed that forebrain regions (hippocampus and cortex), but not the cerebellum, exhibited marked decreases in the mRNA levels of Mdga1 with no change in those of Mdga2 (Appendix Fig. S15A ). Semi-quantitative immunoblotting experiments showed that Mdga1 -cKO caused a significant loss of MDGA1 whereas the Y636C/E751Q substitution did not affect the protein level of MDGA1 (Appendix Fig. S15B,C ). We also observed that the expression of GABA A Rγ2 and the vesicular GABA transporter (VGAT) was increased in cKO and KI mice, gephyrin levels were substantially increased in cKO mice, and Nlgn2 levels were increased in KI mice (Appendix Fig. S15B,C ). We did not observe any significant change in the levels of the other examined proteins, including those characteristic of excitatory synapses. These data suggest that Mdga1 Y636C/E751Q KI and cKO do not cause global changes in the molecular composition of the brain, except for the small increases seen in inhibitory synaptic markers in both KI and cKO mice. We further performed immunohistochemical analyses using antibodies against VGAT and gephyrin (to label inhibitory synapses) or VGLUT1 and PSD-95 (to label excitatory synapses) in the hippocampus and mPFC of adult cKO and KI mice. An increased density of VGAT + puncta was specifically observed in the stratum lacunosum moleculare layer of the hippocampus of male cKO and KI mice but not their female counterparts (Appendix Figs. S16 – S23 ). No change was observed in either inhibitory or excitatory synapses in mPFC layer II/III of adult cKO and KI mice. Next, we examined whether the Mdga1 Y636C/E751Q variant affects the ability of endogenous MDGA1 to regulate synaptic transmission in hippocampal CA1 pyramidal neurons (Fig. 5A–C ). Male KI mice were subjected to electrophysiological recordings at various stages for measurement of spontaneous synaptic transmission, evoked synaptic strength, and neurotransmitter release probability. Since discrepancies in electrophysiological phenotypes were previously reported for Mdga1 -KO mice at different stages of growth (Connor et al, 2017 ; Kim et al, 2022a ), we included both juvenile (further subdivided into P10–12 or P18–20) and adult (P60) male Mdga1 Y636C/E751Q mice for the study. Whole-cell patch-clamp recordings revealed no marked alterations in the frequency/amplitude of mIPSCs or mEPSCs in hippocampal CA1 neurons from both juvenile and adult male Mdga1 Y636C/E751Q mice, except for P10–12 Mdga1 Y636C/E751Q neurons, which exhibited increased frequencies of both mIPSCs and mEPSCs (Fig. 5D–O ). Parallel recordings using male Mdga1 -cKO mice showed similar phenotypes aligned with those from the same-staged male Mdga1 Y636C/E751Q mice; specifically, an increased frequency of mIPSCs and mEPSCs in P10–12 CA1 pyramidal neurons with no changes in spontaneous synaptic transmission in P18–20 or P60 CA1 pyramidal neurons (Fig. EV5 ). Figure 5. Analysis of GABAergic synaptic properties in the hippocampal CA1 pyramidal neurons of Mdga1 Y636C/E751Q KI mice. Open in a new tab ( A ) Schematic illustrating the genetic targeting strategy used to generate Mdga1 Y636C/E751Q KI mice. The point mutations p.Tyr636Cys (TAC > TGC) and p.Glu751Gln (GAG > CAG) are shown on the targeting allele and WT allele. ( B ) Breeding scheme followed to generate Mdga1 Y636C/E751Q KI mice. ( C ) Flowchart of the electrophysiology experiments conducted on Mdga1 Y636C/E751Q KI mice at different developmental stages (P10–12, P18–20 and P60). ( D , E ) Representative traces ( D ) and quantification ( E ) of mEPSCs in CA1 pyramidal neurons from control and Mdga1 Y636C/E751Q KI mice at P10–12. Mdga1 Y636C/E751Q KI mice exhibited a significant increase in mEPSC frequency. Data are presented as means ± SEMs ( n = 19 cells/group; * P = 0.0497; Mann–Whitney U test). ( F , G ) Representative traces ( F ) and quantification ( G ) of mIPSCs in CA1 pyramidal neurons from control and Mdga1 Y636C/E751Q KI mice at P10–12. Mdga1 Y636C/E751Q KI mice showed a significant increase in mIPSC frequency. Data are presented as means ± SEMs ( n = 18–19 cells/group; * P = 0.0335; Mann–Whitney U test). ( H , I ) Representative traces ( H ) and quantification ( I ) of mEPSCs in CA1 pyramidal neurons from control and Mdga1 Y636C/E751Q KI mice at P18–20. Data are presented as means ± SEMs ( n = 15 cells/group). ( J , K ) Representative traces ( J ) and quantification ( K ) of mIPSCs in CA1 pyramidal neurons from control and Mdga1 Y636C/E751Q KI mice at P18–20. Data are presented as means ± SEMs ( n = 14 cells/group). ( L , M ) Representative traces ( L ) and quantification ( M ) of mEPSCs in CA1 pyramidal neurons from control and Mdga1 Y636C/E751Q KI mice at P60. Data are presented as means ± SEMs ( n = 19 cells/group). ( N , O ) Representative traces ( N ) and quantification ( O ) of mIPSCs in CA1 pyramidal neurons from control and Mdga1 Y636C/E751Q KI mice at P60. Data are presented as means ± SEMs ( n = 19 cells/group). ( P–T ) Representative traces ( P , S ) and quantification of dendritic eIPSCs in CA1 pyramidal neurons from control and Mdga1 Y636C/E751Q KI mice at P18–20. Mdga1 Y636C/E751Q KI mice exhibited increased eIPSC amplitude ( Q , R ) and decreased PPR ( T ). Data are presented as means ± SEMs ( n = 11–12 cells/group; ( Q ) * P = 0.0358 (50 μA), * P = 0.0066 (60 μA); ( R ) * P = 0.0492; ( T ) * P = 0.0225; Mann–Whitney U test). ( U – Y ) Representative traces ( U , X ) and quantification of dendritic eIPSCs in CA1 pyramidal neurons from control and Mdga1 Y636C/E751Q KI mice at P60. Mdga1 Y636C/E751Q KI mice showed increased eIPSC amplitude ( V , W ) and decreased PPR ( Y ). Data are presented as means ± SEMs ( n = 19 cells/group; ( V ) * P = 0.0247 (50 μA), * P = 0.0178 (60 μA); ( W ) * P = 0.0219; ( Y ) * P = 0.0004; Mann–Whitney U test). Source data are available online for this figure . Figure EV5. Analysis of electrophysiological properties in the hippocampal CA1 pyramidal neurons of various aged male Mdga1 -cKO mice. Open in a new tab ( A , B ) Representative traces ( A ) and quantification ( B ) of mEPSC frequency and amplitude recorded from CA1 pyramidal neurons of P10–12 male mice. Mdga1 -cKO neurons showed significantly increased mEPSC frequency, while amplitude remained unchanged. Data are means ± SEMs (control, n = 15/6; Mdga1 -cKO, n = 14/6; * P = 0.0459, Mann–Whitney U test ) . ( C , D ) Representative traces ( C ) and quantification ( D ) of mIPSCs from P10–12 CA1 pyramidal neurons. Mdga1 -cKO mice exhibited increased mIPSC frequency and amplitude. Data are means ± SEMs (control, n = 19/6; Mdga1 -cKO, n = 19/6; (frequency) * P = 0.0233, (amplitude) * P = 0.0215; Mann–Whitney U test ) . ( E , F ) Representative traces ( E ) and quantification ( F ) of mEPSCs recorded from CA1 pyramidal neurons of P18–20 male mice. No significant differences were observed between genotypes. Data are means ± SEMs (control, n = 12/4; Mdga1 -cKO, n = 12/4). ( G , H ) Representative traces ( G ) and quantification ( H ) of mIPSCs recorded from CA1 pyramidal neurons of P18–20 male mice. No significant differences were observed between genotypes. Data are means ± SEMs (control, n = 11/4; Mdga1-cKO , n = 10/4). ( I , J ) Representative traces ( I ) and quantification ( J ) of mEPSCs recorded from CA1 pyramidal neurons of P60 male mice. No significant differences were observed between genotypes. Data are means ± SEMs ( n = 10/4 per group). ( K , L ) Representative traces ( K ) and quantification ( L ) of mIPSCs recorded from CA1 pyramidal neurons of P60 male mice. No significant differences were detected between genotypes. Data are means ± SEMs ( n = 10/4 per group). ( M – O ) Representative traces ( M ) and averages of dendritic eIPSCs ( N , O ) from CA1 pyramidal neurons of male control and Mdga1 -cKO mice at P18–20. Mdga1 -cKO mice exhibited a significant increase in eIPSC amplitude. Data are presented as means ± SEMs (control, n = 12/4; Mdga1 -cKO, n = 12/4; ( N ) (40 μA) * P = 0.0402, (50 μA) * p = 0.0449; ( O ) * p = 0.0242; Mann–Whitney U test). ( P , Q ) Representative traces ( P ) and averages of paired-pulse ratios (PPRs) of dendritic IPSCs ( Q ) from CA1 pyramidal neurons of male control and Mdga1 -cKO mice at P18–20. Mdga1 -cKO mice showed a significant reduction in PPRs. Data are presented as means ± SEMs (control, n = 13/4; Mdga1 -cKO, n = 13/4; * P = 0.0387 ; Mann–Whitney U test). ( R – T ) Representative traces ( R ) and averages of dendritic eIPSCs ( S , T ) from CA1 pyramidal neurons of male control and Mdga1 -cKO mice at P60. eIPSC amplitude was significantly increased in Mdga1 -cKO mice. Data are presented as means ± SEMs (control, n = 13/5; Mdga1 -cKO, n = 16/5; ( S ) (40 μA) * P = 0.0435, (60 μA) ** P = 0.0066; ( T ) * P = 0.0499; Mann–Whitney U test). ( U , V ) Representative traces ( U ) and averages of dendritic eIPSC-PPRs ( V ) from CA1 pyramidal neurons of male control and Mdga1 -cKO mice at P60. Mdga1 -cKO mice showed significantly decreased dendritic eIPSC-PPRs. Data are presented as means ± SEMs (control, n = 14/5; Mdga1 -cKO, n = 16/5; * P = 0.0267; Mann–Whitney U test). Hippocampal CA1-specific Mdga1 -cKO mice exhibit increased dendritic IPSC amplitudes with increased paired-pulse ratios (PPR) specifically when the distal dendrites are stimulated (Kim et al, 2022a ). Accordingly, we examined whether these electrophysiological features could be replicated in male Mdga1 Y636C/E751Q CA1 pyramidal neurons. Both P18–20 and P60 CA1 pyramidal neurons from male Mdga1 Y636C/E751Q KI mice exhibited a marked increase in eIPSC amplitudes and IPSC-PPRs (Fig. 5P–Y ). Parallel measurements from male Mdga1 -cKO mice showed identical phenotypes (Fig. EV5 ), reinforcing the LoF effects of the Y636C/E751Q substitution. Unexpectedly, we observed a significant increase in dendritic aIPSC amplitudes in CA1 pyramidal neurons from male Mdga1 Y636C/E751Q KI and Mdga1 -cKO mice (Appendix Fig. S24 ). However, measurement of somatic aIPSCs revealed a specific increase in male Mdga1 Y636C/E751Q KI mice, but not in male Mdga1 -cKO mice (Appendix Fig. S24E–H ). Strikingly, CA1 pyramidal neurons from both adult female Mdga1 Y636C/E751Q KI and Mdga1 -cKO mice displayed no alterations in eIPSC amplitudes or IPSC-PPRs compared to control neurons (Fig. EV6A–J ). In contrast, CA1 neurons from juvenile female Mdga1 Y636C/E751Q KI and Mdga1 -cKO mice showed increased eIPSC amplitudes and IPSC-PPRs (Fig. EV6K–T ), suggesting that sex-specific changes in GABAergic synaptic strength reflect developmental effects. These results are in keeping with the lack of detectable impairment of certain social behaviors in adult female Mdga1 Y636C/E751Q KI and Mdga1 -cKO mice (see Figs. EV3 and EV4 ). Figure EV6. Analysis of synchronous GABAergic evoked synaptic transmission in the hippocampal CA1 pyramidal neurons of adult female Mdga1 -cKO and Mdga1 Y636C/E751Q KI mice. Open in a new tab ( A – E ) Representative traces ( A , D ), input-output (I–O) curves ( B , C ), and paired-pulse ratios (PPRs) ( E ) of dendritic eIPSCs recorded from CA1 pyramidal neurons in adult female control and Mdga1 Y636C/E751Q KI mice. No significant differences were observed. Data are presented as means ± SEMs ( n = 12 cells/group). ( F – J ) Representative traces ( F , I ), I–O curves ( G , H ), and PPRs ( J ) of dendritic eIPSCs recorded from CA1 pyramidal neurons in adult female control and Mdga1 -cKO mice. No significant differences were observed. Data are presented as means ± SEMs ( n = 12 cells/group). ( K–O ) Representative traces ( K , N ), I–O curves ( L , M ), and PPRs ( O ) of dendritic eIPSCs recorded from CA1 pyramidal neurons in P18 female control and Mdga1 Y636C/E751Q KI mice. Mdga1 Y636C/E751Q KI mice showed significantly increased eIPSC amplitude ( M ) and decreased PPRs ( O ). Data are presented as means ± SEMs ( n = 9 cells/group; ( L ) (30 μA) * P = 0.0333, (50 μA) * P = 0.0333, (60 μA) * P = 0.0151; ( M ) * P = 0.0142; ( O ) * P = 0.0188; Mann–Whitney U test). ( P – T ) Representative traces ( P , S ), I–O curves ( Q , R ), and PPRs ( T ) of dendritic eIPSCs recorded from CA1 pyramidal neurons in P18 female control and Mdga1 -cKO mice. Mdga1 -cKO mice exhibited significantly increased eIPSC amplitude ( Q ) and decreased PPRs ( T ). Data are presented as means ± SEMs ( n = 11 cells/group; ( Q ) (40 μA) * P = 0.0199, (50 μA) * P = 0.0158, (60 μA) * P = 0.0135; ( R ) * P = 0.0192; ( T ) * P = 0.0225; Mann–Whitney U test). Numerous neuropathological studies have revealed lower interneuron density in postmortem tissues from individuals with ASDs and hypoactive parvalbumin (PV) + interneurons in various ASD animal models (Contractor et al, 2021 ; Filice et al, 2020 ). Based on these findings, we examined the density of PV + interneurons and neuronal activity across hippocampal subfields in our animal models (Appendix Fig. S25 ). No changes in interneuron density or neuronal activity (assessed via c-Fos immunostaining) were evident in either Mdga1 Y636C/E751Q KI or Mdga1 -cKO mice. Moreover, no changes in the density of neurons, astrocytes or microglia were observed in male adult Mdga1 Y636C/E751Q KI mice (Appendix Fig. S26 ). Given the previous report that MDGA1 is robustly expressed in the upper cortical layer neurons (Takeuchi and O’Leary, 2006 ), we examined whether Mdga1 Y636C/E751Q KI mice exhibited altered targeting of cortical layer II/III neurons, as seen in Mdga1 -cKO mice (see Fig. EV2 ). Immunohistochemical analyses revealed that the density of CB + and Brn2 + neurons in somatosensory cortex layer II/III was reduced in Mdga1 -cKO mice (Appendix Fig. S27 ), whereas no such changes were observed in adult male Mdga1 Y636C/E751Q KI mice when assessed using the cortical layer II/III-specific markers, Wfs1, calbindin (CB), and Brn2 (Appendix Fig. S28 ). Our data collectively demonstrate that the Y636C/E751Q substitution induces a LoF change in MDGA1 in both electrophysiological and behavioral phenotypes with sexual dimorphism, similar to the results obtained for Mdga1 -cKO neurons, but without affecting the maturation of the cortical laminar structure. Proteomic analyses identify hypophosphorylation of synapsin II in adult male Mdga1 -cKO and Mdga1 Y636C/E751Q KI mice To understand the mechanisms underlying MDGA1-LoF-induced ASD-like phenotypes, we attempted proteomic and phosphoproteomic analyses of hippocampal tissues from embryonic mice (E18), adult (P60) male Mdga1 -cKO mice, male Mdga1 Y636C/E751Q KI mice, and littermate control embryos or mice. The proteomic analysis showed that the MDGA1 protein level was significantly reduced in Mdga1 -cKO mice but not in Mdga1 Y636C/E751Q KI mice (Datasets EV3 and EV4 ; see also Appendix Fig. S15 ). Phosphopeptides were enriched in tryptic peptides using immobilized metal ion affinity chromatography (IMAC) and subjected to LC-MS/MS (Fig. 6A ; Datasets EV5 and EV6 ). Compared to control mice, our analysis of adult male Mdga1 -cKO mice identified 28 differentially expressed phosphopeptides (DEPPs, P < 0.05; Dataset EV6 ) with significant changes (Log 2 FC > 0.4 or Log 2 FC < -0.4) belonging to 27 proteins (DEPs). In contrast, adult male Mdga1 Y636C/E751Q KI mice showed 83 DEPPs belonging to 75 DEPs (Fig. 6B,C ; Dataset EV7 ). Gene Ontology (GO) analyses were performed on the DEPs using Metascape, which enriched for GO terms related to synapse-related functions in both the Mdga1 -cKO and Mdga1 Y636C/E751Q KI mouse analysis (Fig. 6D ). A substantial fraction of the DEPs from adult male Mdga1 -cKO and adult male Mdga1 Y636C/E751Q KI, mice (33.3% and 3%, respectively) belonged to the SynGO protein category (Fig. 6E–G ). Among the DEPs, synapsin II (SynII) and OTU domain-containing protein 5 (Otud5) were commonly downregulated in adult male Mdga1 -cKO and Mdga1 Y636C/E751Q KI mice (Fig. 6H,I ). Figure 6. Phosphoproteomic analysis of hippocampi from adult Mdga1 -cKO and Mdga1 Y636C/E751Q KI mice. Open in a new tab ( A ) Schematic diagram of phosphoproteomic workflow using hippocampal lysates from adult (8-week-old) male Mdga1 -cKO, Mdga1 Y636C/E751Q KI and littermate control mice ( n = 4 per group). ( B ) Volcano plot of 4338 phosphopeptides identified from hippocampal lysates of adult male Mdga1 -cKO mice. Differentially expressed phosphopeptides (DEPPs) with significant increases or decreases (Log 2 FC < -0.4 or > 0.4; P < 0.05; Welch’s T test) are shown in red (12 phosphopeptides) and blue (16 phosphopeptides), respectively. ( C ) Volcano plot of 5124 phosphopeptides identified from hippocampal lysates of adult male Mdga1 Y636C/E751Q KI mice. DEPPs with significant increases or decreases (Log 2 FC < -0.4 or > 0.4; P < 0.05; Welch’s T-test) are shown in red (35 phosphopeptides) and blue (48 phosphopeptides), respectively. ( D ) Gene ontology (GO) analysis was performed on differentially expressed phosphoproteins (DEPs) identified from the phosphoproteomic analyses of adult male Mdga1 -cKO (left) and Mdga1 Y636C/E751Q KI (right) mice. The bar charts depict the significantly enriched GO terms of the Molecular Function (green), Cellular Component (yellow), and Biological Process (red) categories, represented as -log 10 (FDR) values. ( E ) Venn diagram visualization of SynGO category proteins. ( F ) Localization annotations of SynGO category proteins from adult male Mdga1 -cKO mice. ( G ) Localization annotations of SynGO category proteins from adult male Mdga1 Y636C/E751Q KI mice. ( H ) Venn diagram showing the overlap of DEPs between Mdga1 -cKO and Mdga1 Y636C/E751Q KI mice. Two proteins, SynII and Otud5, are shared between condition; 25 DEPs are unique to Mdga1 -cKO mice and 73 are unique to Mdga1 Y636C/E751Q KI mice. ( I ) Summary table of DEPs after removal of redundancy. The table shows the total number DEPs with a breakdown of upregulated and downregulated proteins in Mdga1 -cKO and Mdga1 Y636C/E751Q KI mice, together with the number of unique peptides (# pep). ( J ) STRING analysis of downregulated DEPs from Mdga1 -cKO mice ( P < 0.05 and Log 2 FC < -0.4; Welch’s T test). ( K ) STRING analysis of downregulated DEPs from Mdga1 Y636C/E751Q KI mice ( P < 0.05 and Log 2 FC < -0.4; Welch’s T test). A STRING analysis of protein-protein interactions (PPIs) was performed using downregulated DEPs (Fig. 6J,K ). Intriguingly, clear downregulation of phosphorylation levels was seen for SynII in adult male Mdga1 -cKO and adult male Mdga1 Y636C/E751Q KI mice, respectively, compared to the corresponding control mice. Synaptic vesicle clustering and maintenance of synapse structure were identified as the most relevant biological processes; the relevant DEPs included SynI, SynII, Piccolo, Bassoon, and RIMS1 (Appendix Fig. S30 ). Notably, the phosphorylation levels of these proteins were not significantly altered in our phosphoproteomic analysis of E18 embryos from Mdga1 Y636C/E751Q KI mice (Appendix Fig. S31 ; Dataset EV8 ). These extensive proteomic analyses revealed that there are marked developmental-stage-related differences in the profiles of DEPPs, DEPs, and phosphorylation level changes in a subset of presynaptic proteins, including Syn proteins, that might manifest as sexually dimorphic abnormalities in MDGA1 functions (see below). Bazedoxifene administration rescues reduced SynII phosphorylation, impaired communicative behavior and GABAergic synaptic transmission in adult male Mdga1 Y636C/E751Q KI mice Given that both male and female Mdga1 -mutant mice exhibited comparable increases in inhibitory synaptic transmission during the juvenile period, whereas only males (and not females) retained these abnormalities into adulthood (see Figs. 5 and EV6 ), we considered the possibility that puberty-associated, sex-dependent factors contribute to this divergence (Ferri et al, 2018 ). We therefore examined whether estrogen-sensitive pathways—which are known to influence inhibitory synapse maturation and circuit refinement—modulate the persistent adult male phenotype. Proteomic analyses revealed no significant changes in the levels of estrogen receptor-associated proteins (Ddx5 and Phb2), but marginal (yet significant) changes in the levels of the NDD risk factors, Cnot1 and Parp1 (Dong et al, 2023 ; Humphries et al, 2023 ; Vissers et al, 2020 ) (Appendix Fig. S32 ), in male Mdga1 -cKO and Mdga1 Y636C/E751Q KI mice, hinting that there is a link between the MDGA1 and estrogen receptor signaling pathways. We also performed enzyme-linked immunosorbent assay (ELISA)-based measurement of estradiol levels in male juvenile and adult Mdga1 Y636C/E751Q KI mice and found no significant differences in estradiol levels between WT and KI mice (Appendix Fig. S33 ). These findings indicate that baseline estrogen signaling is not substantially altered in Mdga1 -mutants, but raise the possibility that enhancing estrogen-responsive pathways might still modulate the persistent adult male phenotypes. Thus, we utilized bazedoxifene (BZD), which is a third-generation selective estrogen receptor modulator that is blood-brain barrier (BBB)-penetrant and Food and Drug Administration (FDA)-approved for treating menopausal symptoms (Hill et al, 2020 ; Peng et al, 2017 ; Zafar et al, 2022 ). Our decision to employ acute BZD treatment was based on a previous report that BZD shows rapid brain penetrance and swift neuromodulatory potential (Hill et al, 2020 ). We acutely injected adult male Mdga1 Y636C/E751Q KI mice with a single dose of BZD (4.1 mg/kg) and, 6 h later, performed semi-quantitative immunoprecipitation analyses using phosphoserine antibodies (recognizing phosphorylated Syn residues) to measure changes in Syn protein phosphorylation, and applied a subset of behavioral tests to measure ASRs, PPI and USVs (Fig. 7A ). As hypothesized, administration of BZD rescued the reduced protein phosphorylation of SynII isoforms (but not SynI), the downregulation of GABAergic synaptic proteins (Nlgn2, GABA A Rγ2 and VGAT; Fig. 7B–E ), and several behavioral abnormalities, all of which returned to the levels seen in littermate control or female counterpart mice (Fig. 7F–K ). Notably, a subset of altered USV syllable types (including complex trill) and ASRs, but not PPI, was rescued by BZD treatment of adult male Mdga1 Y636C/E751Q KI mice (Fig. 7H–K ). In addition, the enhancement of GABAergic synaptic strength in the distal dendrites of the CA1 pyramidal neurons from adult male Mdga1 Y636C/E751Q KI mice was restored to control levels upon BZD administration (Fig. 7L–P ). The normalized eIPSC amplitudes and PPRs of eIPSCs in the CA1 pyramidal neurons from adult male Mdga1 Y636C/E751Q KI mice were partially reversed by acute treatment with the estrogen receptor antagonist, Fulvestrant (5 μM) (Robertson and Harrison, 2004 ) (Appendix Fig. S34 ). Although this partial reversal does not establish a direct molecular mechanism, it is in line with the idea that acute modulation of estrogen-sensitive signaling can influence the persistent inhibitory abnormalities observed in adult males. Viewed together, our results highlight different combinations of synaptic, behavioral and hormonal mechanisms that can produce ASD-like manifestations and suggest that MDGA1 could be a reliable biomarker for studying a putative biological basis for the male preponderance of ASDs. Figure 7. Bazedoxifene rescues sexually dimorphic alterations in the hippocampal CA1 pyramidal neurons of Mdga1 Y636C/E751Q KI mice. Open in a new tab ( A ) Schematic illustrating the experimental paradigm. Adult male control and Mdga1 Y636C/E751Q KI mice were administered bazedoxifene (BZD; 4.1 mg/kg, intraperitoneally) or vehicle (sham). Six hours post-treatment, the mice were subjected to biochemical, electrophysiological, and behavioral analyses. ( B , C ) Semi-quantitative immunoblotting of hippocampal synaptic proteins, including Nlgn2, GABA A receptor subunit, VGAT, and gephyrin from BZD- or sham-treated mice. β-actin was used as a loading control for normalization. Quantification of normalized protein levels ( C ) revealed that BZD rescued the enhanced expression of synaptic proteins observed in Mdga1 Y636C/E751Q KI mice. Data are means ± SEMs ( n = 6 per group; (Nlgn2) * P = 0.0220, # P = 0.0313; (GABA A Rγ2) ** P = 0.0021, # P = 0.0495; (VGAT) ** P = 0.0059, # P = 0.0343; ‘#’ indicates statistical comparisons with their counterparts; two-way ANOVA with Tukey’s post hoc test). ( D , E ) Representative immunoblots ( D ) and quantification ( E ) of phosphorylated SynI, SynIIa, and SynIIb proteins obtained via immunoprecipitation with anti-phospho-serine antibody and normalized to total Syn levels. BZD significantly restored phospho-SynII levels in KI hippocampi. Data are means ± SEMs ( n = 4 independent experiments; (SynIIa) **P = 0.0010, ## P = 0.0019; (SynIIb) # P = 0.0441; ‘#’ indicates statistical comparisons with their counterparts; two-way ANOVA with Tukey’s post hoc test). ( F , G ) ASRs ( F ) and PPI of ASR ( G ) in adult male control and Mdga1 Y636C/E751Q mice given BZD or sham treatment. BZD treatment reversed the reduced ASR in Mdga1 Y636C/E751Q KI mice, but failed to rescue the deficits in PPI. Data are presented as means ± SEMs ( n = 20–21 mice/group; ( F ) * P = 0.0444 (115 dB), * P = 0.0281 (120 dB), # P = 0.0344 (120 dB); ( G ) * P = 0.0195 (73 dB), * P = 0.0205 (76 dB), * P = 0.0429 (79 dB); ‘#’ indicates statistical comparisons with their counterparts; two-way ANOVA followed by Tukey’s post hoc test). ( H – K ) USVs recorded from socially isolated adult male mice. Representative traces are shown in ( H ). BZD treatment normalized the increased number of USV calls ( I ) observed in Mdga1 Y636C/E751Q KI mice. Call length ( J ) was not significantly altered across groups. The altered distribution of call types in Mdga1 Y636C/E751Q KI mice was also partially rescued by BZD, including restoration of the reduced proportion of complex trill calls ( K ). Data are means ± SEMs ( n = 9–10 mice/group; ( I ) * P = 0.0323, # P = 0.0456; ( K ) ** P = 0.0089, ## P = 0.0045; # indicates statistical comparisons with their counterparts; two-way ANOVA followed by Tukey’s post hoc test). ( L – P ) Electrophysiological recordings of dendritic eIPSCs in CA1 pyramidal neurons at P56. Representative traces ( L , O ) and quantification of I–O responses ( M , N ) and paired-pulse ratio (PPR; P ) revealed that BZD rescued the increased inhibitory input strength and the presynaptic release probability abnormalities observed in Mdga1 Y636C/E751Q KI mice. Data are means ± SEMs ( n = 14–21 cells/group; ( M ) (30 μA) ** P = 0.0065 (vs. MI-KI Sham ), ## P = 0.0064 (vs. MI-KI BZD ); (40 μA) ### P = 0.0003 (vs. Ctrl BZD ), *** P = 0.0001 (vs. MI-KI Sham ), ### P = 0.0004 (vs. MI-KI BZD ); (50 μA) #### P < 0.0001 (vs. Ctrl BZD ), **** P < 0.0001 (vs. MI-KI Sham ), #### P < 0.0001 (vs. MI-KI BZD ); (60 μA) #### P < 0.0001 (vs. Ctrl BZD ), **** P < 0.0001 (vs. MI-KI Sham ), #### P < 0.0001 (vs. MI-KI BZD ); ( N ) #### P < 0.0001 (vs. Ctrl BZD ), **** P < 0.0001 (vs. M1-KI sham ); ( P ) ### P = 0.0004 (vs. Ctrl BZD ), ** P = 0.0010 (vs. M1-KI sham ), ## P = 0.0072 (vs. MI-KI BZD ); ‘#’ indicates statistical comparisons with their counterparts; two-way ANOVA followed by Tukey’s post hoc test). Source data are available online for this figure . Discussion Here we propose that MDGA1, previously characterized as a GABAergic synapse-specific suppressor, may serve as a potential translational biomarker for MDGA1 -associated neurodevelopmental and/or neuropsychiatric disorders, including ASDs. Given that various components of MDGA1-containing synaptic complexes have consistently been linked to ASDs (Kim et al, 2021 ), we examined the possibility that MDGA1 might also be implicated in ASDs. The current study provides new insights into the pathways by which two ASD-associated MDGA1 variants induce previously recognized pathogenesis, including abnormal neuronal migration and imbalanced excitation/inhibition (E/I) ratios (Contractor et al, 2021 ; de la Torre-Ubieta et al, 2016 ; Delorme et al, 2013 ; Nisar et al, 2022 ; Pan et al, 2019 ; Sohal and Rubenstein, 2019 ; Willsey et al, 2022 ). Intriguingly, MDGA1 variants manifested distinct LoF and GoF phenotypes during various stages of postnatal development, highlighting MDGA1 as a multi-functional protein that impinges on a spectrum of neurodevelopmental processes in a spatiotemporally specific manner. Specifically, we found that the V116M/A688V mutant impacts neuronal migration in the cortex and glutamatergic transmission in the hippocampus, whereas the Y635C/E756Q mutant selectively modulates hippocampal GABAergic synapses. These results differentiate MDGA1 from other proteins encoded by known ASD risk genes (Heavner and Smith, 2020 ) and provide a framework for understanding the divergent behavioral outcomes between genotypes. In particular, the GABAergic dysfunction observed in the Y635C/E756Q mutant contrasts with classic pathophysiological mechanisms. While ASD-related alterations at glutamatergic synapses have been comprehensively investigated (Galineau et al, 2023 ; Moretto et al, 2018 ; Nisar et al, 2022 ; Ramaswami and Geschwind, 2018 ), GABAergic pathologies are typically attributed to global deficits, such as decreased GABA levels and loss of PV + interneurons that increase the E/I ratio and network hyperactivity (Coghlan et al, 2012 ; Zhao et al, 2021 ). Furthermore, a number of synaptic genes affect both glutamatergic and GABAergic systems (Zhao et al, 2021 ), complicating the evaluation of specific contributions from GABAergic synapse dysfunctions in ASD pathophysiology. However, the MDGA1 variants analyzed here did not directly alter the numbers of PV + interneurons, or intrinsic neuronal excitability, nor did they affect the binding of MDGA1 to its ligands, Nlgn2 and APP, both of which are also encoded by ASD risk genes (Ali et al, 2020 ; Sokol et al, 2006 ). Instead, the Y635C/E756Q mutant induced a circuit-specific alteration of inhibitory synapses, providing a novel mechanistic basis for the malfunction of GABAergic synapse components in ASDs distinct from global network hyperactivity. Despite providing significant mechanistic insights into the role of MDGA1 in neurodevelopmental processes, our study has several limitations. While the Mdga1 Y636C/E751Q knock-in (KI) mice meticulously recapitulate the specific patient-identified genetic variants, we acknowledge that no single animal model can fully mirror the complex and multifactorial nature of human neurodevelopmental and neuropsychiatric disorders. The full spectrum of complex human clinical features, such as those seen in ASD, is indeed only partially recapitulated in any rodent model. Our model primarily serves as a valuable tool for dissecting the causal molecular and cellular mechanisms underlying MDGA1 dysfunction and specific, quantifiable behavioral deficits, rather than offering high face validity for the entire human condition. Further research is necessary to bridge these gaps between preclinical models and human pathologies. Both MDGA1 variants led to altered USVs in pups, a phenotype reminiscent of communicative deficits in autism. In utero overexpression of MDGA1 V116M/A688V impaired both USVs and neuronal migration, whereas MDGA1 Y635C/E756Q did not elicit such effects in the same context. Intriguingly, pups from Mdga1 Y636C/E751Q KI and Mdga1 -cKO models exhibited reduced isolation-induced USVs, suggesting that endogenous MDGA1 may buffer mutant-induced defects in transient expression systems. These phenotypic discrepancies likely reflect differences in genetic context and developmental timing across models. Notably, USV impairments in Mdga1 Y636C/E751Q KI mice occurred despite intact cortical lamination, unlike the disrupted positioning seen with MDGA1 V116M/A688V overexpression. This divergence suggests that MDGA1 may regulate early communicative behaviors via region- and temporally specific mechanisms beyond its ability to modulate cortical lamination. Supporting this, MDGA1 expression is enriched in cortical excitatory neurons during mid-gestation, but its overexpression in the adult cortex has no electrophysiological effect. Together, these findings underscore the spatiotemporal specificity of MDGA1 function and highlight the need for refined models to dissect its discrete roles in ASD pathogenesis. The Y636C/E751Q substitution did not alter the hippocampal level of MDGA1, despite our in silico prediction that these changes would exert a subtle destabilizing effect. Remarkably, adult male Mdga1 Y636C/E751Q KI mice exhibited increased USV calls with abnormal USV syllable composition; this behavior contrasts with the decreased USV calls observed in adult male Mdga1 -cKO mice, which is typical of other ASD model mice. Rather than a direct GoF effect, the increased USV calls in adult male Mdga1 Y636C/E751Q KI mice likely represent a paradoxical behavioral consequence of network dysregulation driven by a LoF mechanism. While this interpretation aligns with the observed synaptic deficits, we do not exclude the potential for subtle, context-dependent GoF effects contributed by the mutation. Further studies are warranted to unravel the mechanisms by which the Y635C/E756Q substitution leads to these functional deficits despite preserved protein stability, and to determine whether the V116M/A688V substitution causes a similar change in MDGA1 protein levels. The two MDGA1 variants exerted distinct effects on synaptic properties. MDGA1 WT and MDGA1 V116M/A688V suppressed GABAergic synapses, whereas MDGA1 Y635C/E756Q did not exhibit this activity. Strikingly, MDGA1 V116M/A688V impaired the dendritic spine structure and excitatory synaptic transmission. Whereas forebrain-specific Mdga1 -cKO mice unambiguously demonstrated effects only in GABAergic synapses (Kim et al, 2024a ), MDGA1 V116M/A688V had unexpected effects in excitatory synapses. Previous studies established that glutamate plays an important role in neuronal migration, and neuronal migration defects are linked to disruptions in glutamatergic neurotransmission (Gressens, 2000 ; Luhmann et al, 2015 ; Moretto et al, 2018 ). MDGA1 forms complexes with the gap junction subunit, connexin 43 (Perez-Garcia and O’Leary, 2016 ), to provide a neurogenic niche within subventricular zones during cortical neuron migration, and contributes to sustaining glutamatergic synaptic activity (Chever et al, 2014 ). Thus, it is plausible that during the early phase of CNS development, MDGA1-mediated biological processes might be driven by glutamate. Alternatively, a small population of endogenous MDGA1 expressed at glutamatergic synapses could possibly contribute to this process. Further investigation is warranted to unravel the underlying mechanisms. The MDGA1 Y635C/E756Q mutant appears to exhibit a destabilized triangular extracellular structure of MDGA1 (as suggested by in silico prediction and confirmed by TEM analyses). Our findings thus support the recent claim that maintenance of the compact three-dimensional structure of MDGA1 is physiologically significant and required for its function at the synaptic cleft in mature neurons (Lee et al, 2023 ). Indeed, we tested the biological effects of MDGA1 mutations that could modulate compactness of its extracellular triangular structure and found that optimal integrity of MDGA1 structures is essential for its GABAergic synapse-suppressing activities (Appendix Fig. S35 ). Our investigation into the role of MDGA1 in influencing synaptic protein localization and function, particularly the observed suppression of GABAergic synapses, involved overexpression experiments. While the obtained results provide valuable mechanistic insights into the intrinsic ability of MDGA1 to modulate synaptic architecture and its interaction with key synaptic partners, it is crucial to interpret these findings with caution regarding their direct translation to physiological endogenous conditions. Overexpression systems can sometimes influence native protein trafficking, induce non-physiological interactions, or mask subtle effects that occur at endogenous expression levels. Thus, results from overexpression approaches should primarily serve as a complementary tool for exploring underlying molecular mechanisms, as partially supported by our parallel LoF experiments in Mdga1 Y636C/E751Q KI mice. Furthermore, our approach of incorporating two MDGA1 missense variants into a single expression vector to study their combined effect does not precisely replicate the compound heterozygous genetic context observed in patients, where each mutation occurs on a separate allele. While this design allowed us to efficiently investigate the synergistic or additive functional consequences of these co-occurring mutations on the properties of MDGA1, it hinders us from fully dissecting the individual contributions or potential epistatic interactions of each variant in vivo. Future studies employing separate allelic models or more sophisticated genetic engineering approaches would be invaluable for elucidating the precise role of each individual mutation within a true compound heterozygous setting. Our proteomic analyses indicated that Mdga1 -cKO and Mdga1 Y636C/E751Q KI neurons display reduced phosphorylation of presynaptic SynII, which likely desynchronizes neurotransmitter release at GABAergic synapses. This is consistent with the observation that the action of MDGA1 is transsynaptically linked to presynaptic neurons (Kim et al, 2022a ). We speculate that MDGA1-mediated negative transsynaptic regulation occurs via precisely tethered APP protein, which is further coupled with the activity of SynII proteins to orchestrate GABA release at the synaptic cleft that is designed to maximize synaptic strength via the precise alignment of pre- and postsynaptic compartments (Biederer et al, 2017 ; Savtchenko and Rusakov, 2007 ). It is strategically advantageous for MDGA1 to adopt a compact triangular conformation that facilitates mechanical stabilization and efficient transsynaptic signaling (Savtchenko and Rusakov, 2007 ; Zuber et al, 2005 ). It would be interesting to explore whether the circuit-specific introduction of extracellularly stabilized MDGA1 protein or drugs that selectively boost specific SynII phosphorylation site(s) could have beneficial effects on ASDs in the context of various neurodevelopmental trajectories. Adult male Mdga1 -cKO mice exhibited a range of autistic-like behavioral abnormalities. These changes included mild deficits in sociability, reduced USV call numbers and USV duration, and abnormal sensorimotor gating, supporting a genetic association of MDGA1 with ASDs. The presence of deficits in both sociability and USVs indicates that these mice display abnormalities in multiple ASD-relevant domains (i.e., two of the three core diagnostic domains of ASD). Importantly, USV deficits have been consistently reported in well-established ASD models (Giua et al, 2024 ; Premoli et al, 2021 ; Scattoni et al, 2008 ), aligning well with our findings and reinforcing USVs as a crucial ASD-related endophenotype. Furthermore, Mdga1 -cKO mice showed abnormal sensorimotor gating, as evidenced by deficits in PPI. While PPI has been traditionally used in schizophrenia research, it should be noted that it is also widely employed in ASD studies. This assay is particularly valuable in capturing deficits in sensory processing and sensorimotor gating, which are increasingly recognized as part of the ASD phenotype. PPI deficits have been consistently reported in multiple genetic ASD models (Doornaert et al, 2025 ; Ellison et al, 1989 ; Kokash et al, 2019 ). Human studies have similarly shown reduced PPI in individuals with autism and Asperger’s syndrome compared to neurotypical controls (McAlonan et al, 2002 ; Perry et al, 2007 ), indicating its robust translational value. The behavioral profiles of adult male Mdga1 -cKO mice differed from those recently reported for Mdga2 -KO mice (Wang et al, 2024 ), which is consistent with the report that MDGA1 and MDGA2 play distinct roles in negatively regulating GABAergic and glutamatergic synapses, respectively (Kim et al, 2024a ). However, given that MDGA2 forms complexes with Nlgn1 (or Nlgn3) and Nrxns, and Nlgn3 is closely linked to ASDs, it is highly likely that MDGA2 is also associated with ASDs. Intriguingly, the total level of MDGA paralogs appears to be critical for maintaining synaptic homeostasis in neurons (Kim et al, 2024a ). Thus, the mechanisms by which the unique action of each MDGA paralog contributes to synaptic homeostasis in neurons and whether its impairment triggers ASDs require further elucidation. Although the behavioral phenotype in the Mdga1 -cKO mice is broad compared to the selective deficits in Mdga1 Y636C/E751Q KI mice, we emphasize that their molecular and electrophysiological profiles closely mirror those observed in the Mdga1 Y636C/E751Q KI mice. This supports the notion that there is a shared synaptopathic mechanism across genotypes, reinforcing the behavioral findings. The behavioral impairments we herein documented in forebrain excitatory neuron-specific Mdga1 -cKO mice support the pathogenicity of MDGA1 deficiency. However, it is challenging to directly translate these animal model-derived results to human patients, as the same genetic mutation can often result in different clinical presentations. The functional consequences of MDGA1 mutations may depend on variable factors, such as protein stability, interacting protein network and/or neuronal functions. Given the critical role of MDGA1 in suppressing GABAergic synaptic properties, it is possible that even slight differences in MDGA1 dysfunction could contribute to variations in disease severity. While our findings established a strong connection between biallelic MDGA1 mutations and neurodevelopmental processes, further study is required to fully understand how different variant combinations influence clinical outcomes. In this regard, patient-derived neuronal models will be crucial for unraveling the genotype-phenotype relationship in MDGA1-associated NDDs. In view of this, the immediate goals of future studies should be to introduce idiopathic ASD-associated MDGA1 mutation(s) into induced neuronal cells and determine whether this can replicate major cellular disease phenotypes in human neurons. Although further explorations are warranted, our study provides important clues into how proper control of GABAergic synaptic inhibition is mechanistically linked to ASD pathogenesis. Intriguingly, adult female Mdga1- cKO mice did not show any of the prominent behavioral deficits seen in their male counterparts. Together with our extensive analyses of male and female Mdga1 Y636C/E751Q knock-in mice, these results suggest that our MDGA1 mouse models could be used in the future to elucidate the sexually dimorphic nature of ASDs. Our study raises several questions/issues that remain to be addressed. Whereas adult female Mdga1 -cKO and Mdga1 Y636C/E751Q KI mice displayed no abnormal behavioral or electrophysiological phenotype, juvenile female Mdga1 -cKO and Mdga1 Y636C/E751Q KI mice exhibited electrophysiological phenotypes similar to those of their male counterparts. Although dimorphic phenotypes have been reported in numerous ASD models (Jeon et al, 2018 ), definitive explanations are lacking. The female protective effect theory of ASDs (Dougherty et al, 2022 ; Tsai et al, 1981 ) is compelling, since a sudden peak in gonadal hormone levels during the prepubertal period of female mouse development appears to contribute to an anti-inflammatory state that further regulates hippocampal CA1 GABAergic synapse organization (Mukherjee et al, 2017 ; Tabatadze et al, 2015 ). We herein demonstrate that BZD may be useful for treating certain aspects of ASDs, and our findings further imply that estradiol signaling may act on synaptic disinhibition to normalize SynII activities (Ohtani-Kaneko et al, 2010 ). Although studies on the acute neuromodulatory effects of BZD treatment are still limited, other SERMs, such as tamoxifen and raloxifene, have demonstrated similar rapid effects on neuronal plasticity behavior (Finney et al, 2021 ; Velazquez-Zamora et al, 2012 ), supporting the broader concept that estrogen receptor modulation can exert biologically significant acute effects on the brain within hours of administration. Consistent with our model, acute application of the estrogen receptor degrader, Fulvestrant, partially suppressed the BZD-induced normalization of inhibitory synaptic function in Mdga1 -KI slices. Because Fulvestrant requires ≥6–12 h to induce ubiquitin–proteasome–mediated degradation of nuclear ERα (Hammes and Levin, 2007 ; Long and Nephew, 2006 ), the brief 20-min exposure used here is expected to selectively interfere with rapid membrane-initiated estrogen signaling (Callige and Richard-Foy, 2006 ; Vasudevan and Pfaff, 2007 ). This selective and mechanism-specific partial blockade provides direct functional evidence that the BZD-induced rescue requires intact estrogen receptor signaling, supporting the involvement of an estrogen-sensitive pathway in the MDGA1-dependent synaptic phenotype. Although the details of the relevant molecular interactions remain to be fully elucidated, we note that estrogen signaling is known to rapidly reduce inhibitory synaptic strength in the hippocampus (Huang and Woolley, 2012 ; Mukherjee et al, 2017 ). These fast-acting suppressive effects on GABAergic function may act synergistically with the previously established role for MDGA1 as an inhibitory synapse suppressor. Thus, we speculate that BZD may transiently compensate for the loss of MDGA1 function in male Mdga1 -KI mice through partially overlapping pathways that remain to be identified in future studies. Moreover, GABAergic synaptic inhibition is essential for controlling the window of the critical period of plasticity (Andrade-Talavera et al, 2023 ) and its precocious closure is linked to ASDs (Berger et al, 2013 ; LeBlanc and Fagiolini, 2011 ), giving rise to the speculation that MDGA1 is a key regulatory factor in the development of critical-period plasticity. The effects of BZD administration should be systematically examined in other ASD model animals that exhibit sexually dimorphic abnormalities. We also need to better understand the synaptic and circuit-level actions of BZD and how long its rescue effects last during development. Such studies would help inform the optimization of BZD dosing and administration frequency, with the goal of avoiding potentially adverse side effects (Raina et al, 2024 ). Given the significant expression of MDGA1 in astrocytes, it would also be interesting to establish whether both neuronal and astrocytic MDGA1 are engaged in shaping the critical periods that determine sexual dimorphism during development across diverse brain areas (Ackerman et al, 2021 ; McCarthy and Wright, 2017 ; Ribot et al, 2021 ). Another future challenge will be to determine the cell types, circuits, and/or regions that are primarily involved in eliciting various behavioral manifestations of Mdga1 -cKO and Mdga1 Y636C/E751Q KI mice. Notably, overexpression or deletion of MDGA1 in mPFC neurons failed to exert significant effects on synaptic properties and mouse behavior, partly in line with the low-level expression of MDGA1 in the mPFC. Considering the contributions of the hippocampus to the social and cognitive deficits in ASDs (Banker et al, 2021 ), our results obtained from the hippocampal CA1 GABAergic circuits are compellingly relevant to ASD pathophysiology. To gain a better understanding of ASD biology, the investigation of MDGA1 functions should be extended beyond the hippocampus and mPFC to other regions related to regulating ASD-like behaviors, such as the thalamus (Hwang et al, 2017 ; Roy et al, 2022 ), with the goals of addressing the canonical/non-canonical roles of MDGA1 and establishing its utility as a genuine biomarker of ASD. Importantly, in addition to homozygous Mdga1 Y636C/E751Q KI mice, heterozygous Mdga1 Y636C/E751Q KI mice should be studied to elucidate the detailed abnormalities relevant to ASD patients. Finally, while our study provides compelling preclinical evidence for the pathogenic role of MDGA1 in GABAergic synaptic dysfunction and neurodevelopmental phenotypes, the direct medical relevance and translational biomarker potential of MDGA1 require further rigorous validation. This potential has been suggested, but large-scale clinical and functional studies are needed for further validation (Table EV3 ). The number of reported human cases with MDGA1 mutations currently remains limited, and MDGA1 is not yet widely recognized as a major ASD risk gene in comprehensive databases such as SFARI. Our findings establish MDGA1 as a novel candidate gene and provide critical mechanistic insights, but extensive future validation in larger human cohorts, including functional studies of patient-derived cells and comprehensive genetic screenings, will be essential for firmly establishing its utility as a reliable diagnostic, prognostic, and/or therapeutic biomarker. These ongoing efforts will be crucial for bridging the gap between basic mechanistic discoveries and clinical applications. Methods Reagents and tools table Reagent/resource Reference or source Identifier or catalog number Experimental models Mouse: ICR (E15.5 embryos) Daehan Biolink N/A Mouse: Mdga1 f/f PMID: 35074912 N/A Mouse: Mdga1 Y636C/E751Q KI Biocytogen N/A Recombinant DNA pAAV-MDGA1 WT PMID: 35074912 N/A pAAV-MDGA1 V116M/A688V This paper N/A pAAV-MDGA1 Y635C/E756Q This paper N/A pCAGG-His-HA-MDGA1 WT This paper N/A pCAGG-His-HA-MDGA1 V116M/A688V This paper N/A pCAGG-His-HA-MDGA1 Y635C/E756Q This paper N/A pCIG2-HA-MDGA1 WT This paper N/A pCIG2-HA-MDGA1 V116M/A688V This paper N/A pCIG2-HA-MDGA1 Y635C/E756Q This paper N/A pCMV-IgC APP PMID: 35074912 N/A pCMV-IgC Nlgn2 PMID: 23248271 N/A pDisplay-MDGA1 WT PMID: 23248271 N/A pDisplay-MDGA1 V116M/A688V This paper N/A pDisplay-MDGA1 Y635C/E756Q This paper N/A pDisplay-MDGA1 T135C/N313C This paper N/A pDisplay-MDGA1 V314D This paper N/A pDisplay-MDGA1 T333C/Y519C This paper N/A pDisplay-MDGA1 I336D This paper N/A L313-MDGA1 WT PMID: 23248271 N/A L313-MDGA1 V116M/A688V This paper N/A L313-MDGA1 Y635C/E756Q This paper N/A pDisplay-MDGA2 WT PMID: 23248271 N/A pDisplay-MDGA2 Y631C/E751Q This paper N/A Antibodies Alexa Fluor 488-conjugated antibodies Jackson ImmunoResearch Cat# 715-545-150; RRID: AB_2341099 Alexa Fluor 647-conjugated antibodies Jackson ImmunoResearch Cat# 715-605-150; RRID: AB_2340905 Alexa Fluor 647-conjugated secondary antibodies Jackson ImmunoResearch Cat# 715-605-150; RRID: AB_2340905 Chicken polyclonal anti-EGFP Aves Labs Cat# GFP-1020; RRID: AB_10000240 Chicken polyclonal anti-Tbr1 Millipore Cat# AB2261; RRID: AB_10615497 Cy3-conjugated anti-mouse IgG Jackson ImmunoResearch Cat# 715-165-150; RRID: AB_2340813 Cy3-donkey anti-human IgG Jackson ImmunoResearch Cat# 709-165-149; RRID: AB_2340535 FITC-conjugated anti-rabbit IgG Jackson ImmunoResearch Cat# 711-095-152; RRID: AB_2340612 Goat polyclonal anti-EGFP Rockland Cat# 600-101-215; RRID: AB_218182 Guinea pig polyclonal anti-VGLUT1 Millipore Cat# ab5905; RRID: AB_2301751 Mouse anti-Synapsin II Sigma-Aldrich Cat# S2312; RRID: AB_477499 Mouse monoclonal anti-APP Millipore Cat# MAB348; RRID: AB_94882 Mouse monoclonal anti-Calbindin Swant Cat# 300; RRID: AB_10000347 Mouse monoclonal anti-GAD67 Millipore Cat# MAB5406; RRID: AB_2278725 Mouse monoclonal anti-HA BioLegend Cat# 901501; RRID: AB_2565006 Mouse monoclonal anti-NMDAR1 Millipore Cat# MAB363; RRID: AB_94946 Mouse monoclonal anti-Nlgn1 Synaptic Systems Cat# 129 111; RRID: AB_887747 Mouse monoclonal anti-PV Swant Cat# 235; RRID: AB_10000343 Mouse monoclonal anti-Synapsin I Synaptic Systems Cat# 106 011; RRID: AB_2619722 Mouse monoclonal anti-Synapsin II Synaptic Systems Cat# 106 211; RRID: AB_2619774 Mouse monoclonal anti-gephyrin Synaptic Systems Cat# 147 011; RRID: AB_887719 Mouse monoclonal anti-β-actin Sigma-Aldrich Cat# sc-47778; RRID: AB_476743 Rabbit anti-Gephyrin Synaptic Systems Cat# 147002; RRID: AB_887702 Rabbit anti-VGAT Synaptic Systems Cat# 131003; RRID: AB_887871 Rabbit monoclonal anti-HA Cell Signaling Technology Cat# 3724; RRID: AB_1549585 Rabbit monoclonal anti-c-Fos Cell Signaling Technology Cat# 14609; RRID: AB_2798537 Rabbit polyclonal anti-Brn2 Abcam Cat# ab94977; RRID: AB_10615497 Rabbit polyclonal anti-Calbindin Swant Cat# 300; RRID: AB_10000347 Rabbit polyclonal anti-GABA A receptor gamma2 Synaptic Systems Cat# 224 003; RRID: AB_2263066 Rabbit polyclonal anti-Nlgn2 Synaptic Systems Cat# 129 202; RRID: AB_993011 Rabbit polyclonal anti-PSD-95 Jaewon Ko’s Laboratory (JK016) RRID: AB_2722693 Rabbit polyclonal anti-VGAT Synaptic Systems Cat# 131 003; RRID: AB_887869 Rabbit polyclonal anti-VGLUT1 Millipore Cat# ab5905; RRID: AB_2301751 Rabbit polyclonal anti-Wfs1 Proteintech Cat# 11558-1-AP; RRID: AB_2216046 Rabbit polyclonal anti-pan-SHANK Eunjoon Kim’s Laboratory (1172) RRID: AB_2810261 Rabbit polyclonal anti-phospho-Synapsin II (Ser425) Thermo Fisher Scientific Cat# PA5-64855; RRID: AB_2663626 Rabbit polyclonal anti-phosphoserine Millipore Cat# AB1603; RRID: AB_390205 Rat monoclonal anti-SST Millipore Cat# MAB354; RRID: AB_2255365 Rabbit polyclonal anti-Iba-1 Fujifilm Wako Cat# 016–20001; RRID: AB_839506 Mouse monoclonal anti-GFAP Cell Signaling Technology Cat# 3670; RRID: AB_561049 Mouse monoclonal anti-NeuN Millipore Cat# MAB377; RRID: AB_2298772 Chemicals, enzymes and other reagents 18-plex TMT reagents Thermo Fisher Scientific Cat# A52048 Bovine serum albumin (BSA) Sigma-Aldrich Cat# A9418 Dulbecco’s Modified Eagle Medium (DMEM) Welgene Fetal bovine serum (FBS) Tissue Culture Biologicals HEPES Sigma-Aldrich Cat# H3375 Neurobasal Medium Thermo Fisher Scientific Cat# 21103049 Protein A-Sepharose beads GE Healthcare Cat# 17-0780-01 2,2,2-tribromoethyl alcohol Sigma-Aldrich Cat# T48402 B-27 Supplement Thermo Fisher Scientific Cat# 17504044 Bazedoxifene acetate MedchemExpress Cat# HY-A0036 Fulvestrant MedchemExpress Cat# HY-13636 CNQX Tocris Cat# 0190 CalPhos Transfection Kit Takara Cat# 631312 D-AP5 Tocris Cat# 0106 Triton X-100 Sigma-Aldrich Cat# X100 Trypsin Thermo Fisher Scientific Fast Green FCF Sigma-Aldrich Cat# F7252-5G GlutaMax Thermo Fisher Scientific Cat# 35050061 Glycine Sigma-Aldrich Cat# G8898 1.0 µm Gold Microcarriers Bio-Rad Cat# 1652263 Ni-NTA magnetic agarose beads Qiagen Cat# 36113 Ethylene-Diamine-Tetraacetic Acid Invitrogen Cat# AM9261 Iron(Ⅲ) chloride Sigma-Aldrich Cat# 701122 Paraformaldehyde Sigma-Aldrich Cat# 158127 Penicillin-streptomycin Thermo Fisher Scientific Cat# 15140122 Poly-L-lysine Sigma-Aldrich Cat# P4832 Polyethylene glycol Sigma-Aldrich Cat# 81260 Protease K Sigma-Aldrich Cat# P2308 Sodium pyruvate Thermo Fisher Scientific Cat# 11360070 Picrotoxin Tocris Cat# 1128 Tetrodotoxin (TTX) Cayman Chemical Cat# 14964 Mouse Estradiol (E2) ELISA Kit AssayGenie Cat# MOEB2488 Oligonucleotides and other sequence-based reagents ASD gene sets PMID : 35982160; 31981491 N/A Alzheimer’s gene sets PMID: 24162737 N/A Epilepsy gene sets PMID: 29942082 N/A Schizophrenia gene sets PMID: 35396579 N/A Software Clampfit 10.8 Molecular Devices https://www.moleculardevices.com IGV (Integrative Genomics Viewer) Broad Institute, Cambridge https://igv.org/ Functool 3D Fiber Tracking GE Medical Systems GraphPad Prism GraphPad Software RRID: SCR_002798 ImageJ National Institutes of Health (NIH) FATHMM BioCompute https://fathmm.biocompute.org.uk DeepSqueak GitHub DeepSqueak MetaMorph Molecular Devices https://www.moleculardevices.com Perseus Max Planck Institute of Biochemistry Version 1.6.15 3D Tractography Software GE Medical Systems EthoVision XT 15.5 Noldus Information Technology GenoSystem Genologica scvi-tools scvi-tools.org https://scvi-tools.org pCLAMP11 Molecular Devices https://www.moleculardevices.com SIFT A*STAR Bioinformatics Institute https://sift.bii.a-star.edu.sg Scrublet Github https://github.com/swolock/scrublet Instruments Electroporator BTX Model ECM 830 Vibratome Leica Biosystems Model VT1200S Confocal microscope Carl Zeiss Model LSM800 Transmission electron microscope FEI (Tecnai) Model T12 Bio-TWIN Liquid chromatography system Thermo Fisher Scientific UltiMate 3000 RSLCnano MRI System (3T) GE Medical Systems Milwaukee, Wisconsin Proteomics system Thermo Fisher Scientific Orbitrap Fusion Lumos Mass Spectrometer Two-photon microscope Bruker Nano, Inc. Patch pipette puller Sutter Instruments Model P-97 Circular dichroism spectrophotometer JASCO Model J-1500 Sound-attenuating chamber SD Instruments SR-LAB-Startle Response System Ti:Sapphire laser Spectra Physics Model MaiTai HP Deposited Data Single-cell brain atlas data PMID: 39363111 N/A Open in a new tab Identification of cases with MDGA1 variants In total, 505 cases of trio exome sequencing performed in our department since 2014 were reviewed in search of MDGA1 mutations in homozygous or compound heterozygous states, according to the haploinsufficiency of this gene. All studies were performed on patients with neurodevelopmental disorders of probable genetic origin. Two unrelated cases with compound heterozygous MDGA1 mutations were identified. Genetic analysis Whole exome sequencing was performed using genomic DNA isolated from whole blood from the proband and parents. Genomic DNA extraction was carried out from blood using the MagNa Pure 24 equipment (Roche Diagnostics). The quantity of extracted gDNA was measured with a fluorimeter (Quibit 3.0). The absorbance ratios at 260/280 and 260/230 were also studied to determine the quality of the DNA obtained, using a NanoDrop ND-2000 equipment. In addition, the integrity of the genomic DNA was analyzed by electrophoresis in 0.8% agarose gels. Libraries were prepared using the KAPA Hyper plus Kit (Roche Diagnostics) following the manufacturer’s specifications and a capture enrichment protocol with specific probes (KAPA HyperExome; Roche Diagnostics). Then, we performed subsequent massive parallel sequencing in a NextSeq 550 equipment (Illumina). Signal processing, base calling, alignment, and variant calling were performed with Genologica variant analysis software (GenoSystem). This software developed by Genologica contains an optimized algorithm that includes (among other steps), the following: (a) Initial quality control of the sequences; (b) Filtering the sequences by eliminating indeterminacies, adapters and low-quality areas; (c) Second quality control of the sequences; (d) Mapping on the hg19 reference genome; (e) Obtaining variants and CNVs; (f) Mapping coverage study; and (g) Annotating variants. Finally, variant prioritization was based on stringent assessments at both the gene and variant levels and taking into consideration the patient’s phenotype and the associated inheritance pattern. Candidate variants were visualized using IGV (Integrative Genomics Viewer). Candidate variants were evaluated based on stringent assessments at both the gene and variant levels, taking into consideration both the patient’s phenotype and the inheritance pattern. Variants were classified following the guidelines of the American College of Medical Genetics and Genomics (ACMG). A board of molecular clinical geneticists evaluated each variant classified as pathogenic, likely pathogenic, or a variant of uncertain significance, and decided which, if any, had to be reported. In every case, causal variants were discussed with the referring physician and/or clinical geneticist. Effects of MDGA1 missense variants were predicted using various in silico tools, including SIFT ( S orting I ntolerant f rom T olerant; https://sift.bii.a-star.edu.sg ); FATHMM ( F unctional A nalysis t hrough H idden M arkov M odels; https://fathmm.biocompute.org.uk ); Eigen ( https://www.columbia.edu/eigen ); LRT ( L ikelihood R atio T est) (Chun and Fay, 2009 ); MutationTaster ( https://www.mutationtaster.org ); and MVP ( M issense V ariant P athogenicity prediction) (Qi et al, 2021 ). Neuroimaging The brain MRIs entailed different sequences, including DTI and 3D-SPGRT1, regardless of the reason for the study. DTI images were obtained with a 3 T system (GE Medical System, Milwaukee, Wisconsin) by using a SS-SE echoplanar diffusion-weighted imaging (DWI) sequence (TR:12,000; FOV: 240 mm; section thickness: 3 mm, 0 spacing; matrix 128 × 128; bandwidth: 250; 1 NEX; diffusion encoding in 45 directions) with maximum b = 1000 sec/mm 2 . 3D-tractography was performed in an off-line workstation by using commercially available processing software as provided by the manufacturer (Functool 3D Fiber Tracking, GE, France) based on the fiber assignment by contiguous tracking (FACT) method, achieved by connecting voxel to voxel. The threshold values were 0.3 for FA and 45° for the trajectory angles, between the regions of interest (ROIs). DTI tracts were also co-registered to the 3D-T1 weighted dataset. MRI evaluation and tractography analysis were conducted by a radiologist specializing in these techniques who was unaware of the patient´s genetic diagnosis. Ethical procedure The study was carried out in accordance with the Declaration of Helsinki of the World Medical Association and approved by the local ethics committees. Informed consent was obtained from each family after a full explanation of the procedures. Identification of MDGA1 in a single-cell human brain atlas We utilized the single-cell brain atlas from our previous study (Data ref: Kim et al, 2024b ) to examine MDGA1 expression. The atlas collated eight single-cell and single-nucleus datasets representing neurotypical samples from 11 developmental stages (from gestational week 7–90 years), with the data classified into 41 clusters. Among the collected data, cells and cell clusters annotated as being of poor quality in the original studies and those lacking sample information were excluded. Metadata, including sample ID, batch, platform, demographics, brain region, and postmortem interval were standardized with age information categorized into developmental stages. Gene names were standardized with respect to the NCBI and HGNC databases, and doublets detected by Scrublet version 1.8.2 ( https://github.com/swolock/scrublet ) were excluded. Normalization, log transformation, and selection of highly variable genes were executed using Scanpy version 1.8.2 ( https://scanpy.readthedocs.io ). Integration with batch effect correction was conducted using scvi-tools version 1.0.3 ( https://scvi-tools.org ), and clustering was conducted with the Leiden algorithm (resolution = 0.6). Primary cell types were annotated based on the expression of canonical cell type marker genes defined by the Allen Brain Institute (Hodge et al, 2019 ). Genes that were differentially expressed in each cluster were identified using the Wilcoxon rank-sum test and cluster-specific genes were defined as those with an FDR < 0.05 and a log 2 -fold change >0.2. Analysis of MDGA1 expression in single-nucleus transcriptomic data from developing human neocortex The single-nucleus RNA-sequencing data from the developing human neocortex single-nucleus multiome atlas (Data ref: Wang et al, 2025 ) were utilized to examine the cell-type specificity of MDGA1 expression. This dataset comprises 38 human neocortical samples collected from the prefrontal cortex and primary visual cortex across 5 developmental stages: the first trimester, second trimester, third trimester, infancy, and adolescence. The processed and annotated h5ad data were obtained from the CELLxGENE database ( https://cellxgene.cziscience.com/collections/ad2149fc-19c5-41de-8cfe-44710fbada73 ). All data processing and visualization were performed using Scanpy version 1.8.2 (Wolf et al, 2018 ). Gene set enrichment test Gene set enrichment tests were performed for risk genes associated with various neurological disorders. ASD risk genes selected from multiple comparisons were retrieved from Fu et al ( n = 184) (Data ref: Fu et al, 2022 ) and Satterstrom et al ( n = 96) (Data ref: Satterstrom et al, 2020 ) and categorized as being involved with gene expression regulation ( n = 54) and neuronal communication ( n = 24) as previously defined (Data ref: Satterstrom et al, 2020 ). Alzheimer’s disease risk genes were obtained from Lambert et al ( n = 32) (Data ref: Lambert et al, 2013 ). Epilepsy risk genes were identified from Heyne et al ( n = 50) (Data ref: Heyne et al, 2018 ). The schizophrenia gene set was retrieved from Singh et al ( n = 32) (Data ref: Singh et al, 2022 ). Genes significantly enriched for PTVs or damaging missense variants ( P value < 0.01) in cases and listed as bipolar disorder risk genes in the BipEx dataset ( https://bipex.broadinstitute.org ; n = 58) were selected. The developmental disorder gene set ( n = 372) was sourced from Fu et al (Data ref: Fu et al, 2022 ). Intellectual disability risk genes ( n = 28) were collected from Willsey et al (Data ref: Willsey et al, 2018 ). ADHD risk genes ( n = 74) were gathered from Demontis et al (Data ref: Demontis et al, 2023 ). Gene identifiers for these gene sets were matched between gene sets by converting each gene alias into the corresponding HGNC gene symbol. For the enrichment test, a Fisher’s exact test (one-sided) with multiple comparisons by the Bonferroni procedure was applied. Comparison of MDGA1 expression between ASD and control cells The single-nucleus atlas representing cells retrieved from 15 ASD patients and 16 controls (Data ref: Velmeshev et al, 2019 ) was utilized. Data processing and UMAP visualization were performed using Scanpy version 1.8.2 (Wolf et al, 2018 ). Statistical comparisons of MDGA1 expression were performed using the two-sided Wilcoxon rank-sum test for each cell-type cluster. To account for multiple testing across cell types, FDR correction was applied using the Benjamini-Hochberg method. An overall comparison combining all cells across clusters was conducted separately for descriptive purposes and was not included in the multiple testing correction. Summary statistics, including means, medians, standard errors of the mean (SEM), standard deviations, fold-changes, and log 2 -fold changes were calculated for each comparison. Effect sizes were computed using Cohen’s d formula with pooled standard deviation. Sample sizes for each group and cell type were recorded, and statistical significance was determined using FDR-corrected p -values (FDR < 0.05) for individual cell-type comparisons (Dataset EV2 ). All calculations and visualizations were performed using Python SciPy and statsmodels. Expression vectors pDisplay-MDGA1 V116M/A688V, pDisplay-MDGA1 Y635C/E756Q, pDisplay-MDGA1 V116M, pDisplay-MDGA1 A688V, pDisplay-MDGA1 Y635C, pDisplay-MDGA1 E756Q, pDisplay-MDGA1 T135C/N313C, pDisplay-MDGA1 V314D, pDisplay-MDGA1 T333C/Y519C, and pDisplay-MDGA1 I336D were generated via site-directed mutagenesis (Stratagene) using the pDisplay-MDGA1 WT plasmid (Lee et al, 2013 ) as a template. L313-MDGA1 V116M/A688V and L313-MDGA1 Y635C/E756Q were similarly generated using the L313-MDGA1 WT construct (Lee et al, 2013 ) as a template. pDisplay-MDGA2 Y70C/E820Q was generated via site-directed mutagenesis using pDisplay-MDGA2 WT plasmid (Lee et al, 2013 ) as a template. pAAV-MDGA1 V116M/A688V and pAAV-MDGA1 Y635C/E756Q were similarly generated using pAAV-MDGA1 WT plasmid (Kim et al, 2022a ) as a template. pCAGGS-His-HA-MDGA1 variants (WT, V116M/A688V, and Y635C/E756Q) were constructed by amplification of the indicated full-length sequences via polymerase chain reaction (PCR), followed by digestion with EcoR I and Not I and subcloning into the pCAGGS-His-HA vector (Addgene). pCIG2-HA-MDGA1 variants (WT, V116M/A688V and Y635C/E756Q) were constructed by amplification of the indicated full-length sequences via PCR, followed by digestion with EcoR I and Xma I, and subcloning into the pCIG2 vector (Addgene). The pCMV-IgC APP (Kim et al, 2022a ) and pCMV-IgC Nlgn2 (Lee et al, 2013 ) constructs have been described previously. Antibodies The following commercially available antibodies were used: mouse monoclonal anti-APP (clone 22C11; Millipore; Cat# MAB348; RRID: AB_94882 ); mouse monoclonal anti-β-actin (clone AC-74; Sigma-Aldrich; Cat# A2228; RRID: AB_476743); mouse monoclonal anti-Nlgn1 (Synaptic Systems; Cat# 129 111; RRID: AB_887747); rabbit polyclonal anti-Nlgn2 (Synaptic Systems; Cat# 129 202; RRID: AB_993011); guinea pig polyclonal anti-VGLUT1 (Millipore; Cat# AB5905; RRID: AB_2301751); mouse monoclonal anti-gephyrin (Synaptic Systems; Cat# 147 011; RRID: AB_887719); rabbit polyclonal anti-VGAT (Synaptic Systems; Cat# 131 003; RRID: AB_887869); rabbit polyclonal anti-GABA A receptor γ2 (Synaptic Systems; Cat# 224 003; RRID: AB_2263066); mouse monoclonal anti-GAD67 (clone 1G10.2; Millipore; Cat# MAB5406; RRID: AB_2278725); mouse monoclonal anti-NMDAR1 (clone 54.1; Millipore; Cat# MAB363; RRID: AB_94946 ); goat polyclonal anti-EGFP (Rockland; Cat# 600-101-215; RRID: AB_218182); chicken polyclonal anti-EGFP (Aves Labs; Cat# GFP-1020; RRID: AB_10000240); mouse monoclonal anti-HA (clone 16B12; BioLegend; Cat# 901501; RRID: AB_2565006); rabbit monoclonal anti-HA (clone C29F4; Cell Signaling Technology; Cat# 3724; RRID: AB_1549585); rat monoclonal anti-SST (clone YC7; Millipore; Cat# MAB354; RRID: AB_2255365); mouse monoclonal anti-PV (clone PV235; Swant; Cat# 235; RRID: AB_10000343); rabbit polyclonal anti-Iba-1 (Fujifilm Wako; Cat# 016–20001; RRID: AB_839506); mouse monoclonal anti-GFAP (Cell Signaling Technology; Cat# 3670; RRID: AB_561049); mouse monoclonal anti-NeuN (Millipore; Cat# MAB377; RRID: AB_2298772); rabbit monoclonal anti-c-Fos (clone 9F6; Cell Signaling Technology; Cat# 14609; RRID: AB_2798537); rabbit polyclonal anti-phosphoserine (Millipore; Cat# AB1603; RRID: AB_390205); mouse monoclonal anti-Synapsin I (Synaptic Systems; Cat# 106 011; RRID: AB_2619722); mouse monoclonal anti-Synapsin II (Synaptic Systems; Cat# 106 211; RRID: AB_2619774); rabbit polyclonal anti-phospho-Synapsin II (Ser425) (Thermo Fisher Scientific; Cat# PA5-64855; RRID: AB_2663626); mouse monoclonal anti-Calbindin (Swant; Cat# 300; RRID: AB_10000347); rabbit polyclonal anti-Wfs1 (Proteintech; Cat# 11558-1-AP; RRID: AB_2216046); chicken polyclonal anti-Tbr1 (Millipore; Cat# AB2261; RRID: AB_10615497); rabbit polyclonal anti-Brn2 (Abcam; Cat# ab94977; RRID: AB_10615497); and Cy3-donkey anti-human IgG antibodies (Jackson ImmunoResearch; Cat# 709-165-149; RRID: AB_2340535). The following antibodies were previously described: rabbit polyclonal anti-pan-SHANK (1172; RRID: AB_2810261) and rabbit polyclonal anti-PSD-95 (JK016; RRID: AB_2722693) (Han et al, 2020 ). Drugs The following drugs were commercially purchased: Bazedoxifene acetate (Cat# HY-A0036) and Fulvestrant (Cat# HY-13636) from MedChemExpress (MCE); CNQX (Cat# 0190), D-AP5 (Cat# 0106) and picrotoxin (Cat# 1128) from Tocris Bioscience; Tetrodotoxin (TTX; Cat# 14964) from Cayman Chemical; and 2,2,2-tribromoethyl alcohol (Cat# T48402 ) from Sigma-Aldrich. Cell culture HEK293T cells were cultured in Dulbecco’s Modified Eagle’s Medium (DMEM; Welgene) supplemented with 10% fetal bovine serum (FBS; Tissue Culture Biologicals) and 1% penicillin-streptomycin (Thermo Fisher) at 37 °C in a humidified 5% CO 2 atmosphere. All procedures were performed according to the guidelines and protocols for rodent experimentation approved by the Institutional Animal Care and Use Committee of the Daegu Gyeongbuk Institute of Science and Technology (DGIST). Cell-surface binding assays Ig-fusion proteins of Nlgn2, APP, and IgC alone (control) were produced in HEK293T cells. Soluble Ig-fused proteins were purified using protein A-Sepharose beads (GE Healthcare) as previously described. Bound proteins were eluted with 0.1 M glycine (pH 2.5) and immediately neutralized with 1 M Tris-HCl (pH 8.0). Transfected HEK293T cells expressing the indicated plasmids were incubated with 10 μg/ml Ig-fused proteins for 2 h at 37 °C with gentle agitation. Images were acquired using a confocal microscope (LSM800; Carl Zeiss). Staining for surface/intracellular protein levels HEK293T cells were transfected with expression vectors for HA-MDGA1 WT or the indicated variants. After 48 h, cells were washed twice with PBS, fixed with 3.7% formaldehyde for 10 min at 4 °C, and blocked with 3% horse serum/0.1% bovine serum albumin (BSA; crystalline grade) in PBS for 15 min at room temperature. Surface-expressed protein was detected by staining with mouse anti-HA antibody at room temperature. After 90 min, cells were washed twice with PBS and incubated with Cy3-conjugated anti-mouse antibodies for 1 h at room temperature. Next, cells were permeabilized with PBS containing 0.2% Triton X-100 for 10 min at 4 °C and incubated with rabbit anti-HA antibody for 90 min at room temperature to label intracellularly expressed proteins, followed by FITC-conjugated anti-rabbit secondary antibodies. Images were acquired using a confocal microscope (LSM800; Carl Zeiss). Production of recombinant lentiviruses Lentiviruses were produced by transfecting HEK293T cells with three plasmids–lentivirus vectors, psPAX2, and pMD2.G–at a 2:2:1 ratio. After 72 h, lentiviruses were harvested by collecting the media from the HEK293T cells and briefly centrifuging at 1000 × g to remove cellular debris. Filtered media containing 5% sucrose were centrifuged at 117,969 × g for 2 h; supernatants were then removed, and the virus pellet was washed with ice-cold PBS and resuspended in 80 μl PBS, as previously described (Lee et al, 2013 ). Heterologous synapse formation analyses HEK293T cells were transfected with EGFP (control) or co-transfected with pDisplay-HA-Nlgn2, alone or together with the indicated MDGA1 variants using PEI. After 48 h, the transfected HEK293T cells were trypsinized, seeded onto day in vitro 12 (DIV12) hippocampal neurons, co-cultured for an additional 24 h, and double-immunostained on DIV13 with antibodies against EGFP, HA, and synapsin. All images were acquired using a confocal microscope (LSM800; Carl Zeiss). For quantification, the contours of transfected HEK293T cells were chosen as the region of interest (ROI). Fluorescence intensities of synaptic markers in each ROI were quantified for both red and green channels using MetaMorph software (Molecular Devices). Normalized synapse density on transfected HEK293T cells was expressed as the ratio of red to green fluorescence. Transfection, immunocytochemistry, confocal microscopy imaging, and analyses Hippocampal cultured neurons were prepared from E18 rat brains cultured on coverslips coated with poly-L-lysine, and grown in Neurobasal medium supplemented with B-27 (Thermo Fisher Scientific), 0.5% FBS, 0.5 mM GlutaMax (Thermo Fisher Scientific), and sodium pyruvate (Thermo Fisher Scientific), as previously described (Um et al, 2020 ). Cultured neurons were co-transfected with EGFP alone or with the indicated MDGA1 variant expressing plasmids using a CalPhos Transfection Kit (Takara) at DIV5 and immunostained at DIV14. For immunocytochemistry, cultured neurons were fixed with 4% paraformaldehyde/4% sucrose, permeabilized with 0.2% Triton X-100 in phosphate-buffered saline (PBS), immunostained with the indicated primary antibodies, and detected with the indicated Cy3-, fluorescein isothiocyanate (FITC)-, or Alexa Fluor 647-conjugated secondary antibodies (Jackson ImmunoResearch). Transfected neurons were chosen randomly, and images were acquired using a confocal microscope (LSM800; Carl Zeiss) with a ×63 objective lens; all image settings were kept constant. Z-stack images were converted to maximum intensity projection and analyzed to obtain the size, intensity, and density of puncta immunoreactivities derived from marker proteins. Quantification was performed in a blinded manner using MetaMorph software. Animals All mice were maintained and handled in accordance with protocols (DGIST-IACUC-23112809-0003) approved by the Institutional Animal Care and Use Committee of DGIST under standard, temperature-controlled laboratory conditions. Mice were kept on a 12:12-h light/dark cycle (lights on at 7:00 am) and received water and food ad libitum. All experimental procedures were performed on male mice, using littermate controls without Cre expression. Conditional Mdga1 -KO mice were previously described (Kim et al, 2022a ; Perez-Garcia and O’Leary, 2016 ). Pregnant rats (Daehan Biolink) were used to prepare in vitro cultures of dissociated hippocampal neurons. Mdga1 Y636C/E751Q -KI mice were generated by Biocytogen Pharmaceuticals (Beijing, China) based on CRISPR/Cas9 approach. To generate Mdga1 Mut/+ mice, the candidate sgRNAs, located in the intron 9 and intron 13 of Mdga1 , were searched by the CRISPR design tool ( http://www.sanger.ac.uk/htgt/wge/ ) and then were screened for on-target activity using a Universal CRISPR Activity Assay (UCA TM , Biocytogen Pharmaceuticals (Beijing Co., Ltd). A gene-targeting vector containing a 5’ homology arm, donor fragment (containing the mutations of p.Tyr636Cys; TAC > TGC and p.Glu751Gln; GAG > CAG), and 3’ homology arm was used as a template to repair the double-stranded breaks (DSBs) generated by Cas9/sgRNA. The T7 promoter sequence was added to the Cas9 or sgRNA template by PCR amplification in vitro. The Cas9 mRNA, targeting vector, and sgRNAs were co-injected into the cytoplasm of one-cell stage fertilized C57BL/6 N eggs. The injected zygotes were transferred into the oviducts of Kunming pseudopregnant females to generate F0 mice. F0 mice with the expected genotype confirmed by tail genomic DNA PCR and sequencing were mated with C57BL/6 N mice to establish germline-transmitted F1 heterozygous mice. F1 heterozygous mice were genotyped by tail genomic PCR, Southern blot, and DNA sequencing. Immunohistochemistry, confocal microscopy imaging, and analyses P60 mice were anesthetized and immediately perfused, first with PBS for 3 min, and then with 4% paraformaldehyde for 5 min. Brains were dissected out, fixed in 4% paraformaldehyde overnight, and sliced into 30-µm-thick coronal sections using a vibratome (Model VT1200S; Leica Biosystems). Sections were permeabilized by incubating with 2% Triton X-100 in PBS containing 5% bovine serum albumen and 5% horse serum albumin for 30 min. For immunostaining, sections were incubated for 8–12 h at 4 °C with primary antibodies diluted in 0.1% Triton X-100 in PBS containing bovine and horse serum. The following primary antibodies were used: anti-VGLUT1 (1:300), anti-PSD-95 (1:300), anti-VGAT (1:300), anti-gephyrin (1:500), anti-PV (1:300), anti-SST (1:50), anti-c-Fos (1:400), anti-Brn2 (1:100), anti-Wfs1 (1:100), anti-Calb (1:100), anti-Iba1 (1:1000), anti-GFAP (1:300) and anti-NeuN (1:500). Sections were washed three times in PBS and incubated with appropriate Cy3- or FITC-conjugated secondary antibodies (Jackson ImmunoResearch) for 2 h at room temperature. After three washes with PBS, sections were mounted onto glass slides (Superfrost Plus; Fisher Scientific) with VECTASHIELD mounting medium (H-1200; Vector Laboratories). Images were acquired by confocal microscopy (LSM800; Zeiss). Synaptic puncta were quantified using MetaMorph software, and their density and average area were measured. For immunohistochemical analysis of electroporated mice (P21), the mice were heart-perfused with 4% paraformaldehyde. Brains were collected, embedded in 3% low melting-temperature agarose, and sectioned at 100 µm thickness with a vibratome (VT1200S; Leica). The sections were incubated with primary antibodies (anti-EGFP, 1:2000; anti-HA, 1:1000) in PBS containing 0.1% Triton X-100, 0.1% BSA, and 2.5% goat serum, and then with fluorescent secondary antibodies (Alexa Fluor 488 and Alexa Fluor 647). Slices were mounted in Antifade mounting medium with DAPI (VECTASHIELD) and images were acquired by confocal microscopy (A1R; Nikon). Preparation of adeno-associated viruses and titration HEK293T cells were co-transfected with the indicated AAV vectors and pHelper and pAAV1.0 (serotype 2/9) vectors. 72 h later, transfected HEK293T cells were collected, lysed, and mixed with 40% polyethylene glycol and 2.5 M NaCl, and centrifuged at 2000 × g for 30 min. The cell pellets were resuspended in HEPES buffer (20 mM HEPES; 115 mM NaCl, 1.2 mM CaCl 2 , 1.2 mM MgCl 2 , 2.4 mM KH 2 PO 4 ) and an equal volume of chloroform was added. The mixture was centrifuged at 400 × g for 5 min, and concentrated three times with a Centriprep centrifugal filter (Millipore) at 1220 × g for 5 min each and with an Amicon Ultra centrifugal filter (Millipore) at 16,000 × g for 10 min. Before the titration of AAVs, contaminating plasmid DNA was eliminated by treating 1 μl of concentrated, sterile-filtered AAVs with 1 μl of DNase I (Sigma-Aldrich) for 30 min at 37 °C. After treatment with 1 μl of stop solution (50 mM ethylenediaminetetraacetic acid) for 10 min at 65 °C, 10 μg of proteinase K (Sigma-Aldrich) was added and AAVs were incubated for 1 h at 50 °C. Reactions were inactivated by incubating samples for 20 min at 95 °C. The final virus titer was quantified by qRT-PCR detection of EGFP sequences and subsequent reference to a standard curve generated using the pAAV-T2A-EGFP plasmid. All plasmids were purified using a Plasmid Maxi Kit (Qiagen GmbH). Preparation and transfection of organotypic hippocampal and prefrontal cortical slice cultures Organotypic slice cultures from mouse hippocampus or prefrontal cortex were prepared from P2–3 mice, as described previously (Kim et al, 2022a ; Ogelman et al, 2024 ), in accordance with the guidelines of the Institutional Animal Care and Use Committee at the University of Colorado Anschutz Medical Campus (protocol number: 721; IBC protocol number: 1305). Mice were acquired from the Jackson Laboratory (C57BL/6 N wild-type). Slices were transfected with pDisplay MDGA1 WT, pDisplay MDGA1 V116M/A688V or pDisplay MDGA1 Y635C/E756Q 13–20 days prior to two-photon imaging using biolistic gene transfer (180 psi). A total of 5 μg tdTomato and 5 μg GFP-gephyrin-intrabody or 20 μg of the indicated pDisplay MDGA1 vector were coated onto 6–7 mg gold particles. The age of the culture was expressed as the equivalent postnatal (EP) day; postnatal day at slice culturing + days in vitro. Two-photon imaging Imaging was performed at EP18–24 on transfected hippocampal CA1 pyramidal neurons or mPFC layer II/III pyramidal neurons within 40 μm of the slice surface at 30 °C in recirculating artificial cerebrospinal fluid (ACSF) (127 mM NaCl, 25 mM NaHCO 3 , 1.25 mM NaH 2 PO 4 , 2.5 mM KCl, 25 mM D-glucose) aerated with 95% O 2 /5% CO 2 . For each neuron, image stacks (512 × 512 pixels; 0.047 µm/pixel) with 1 μm Z-steps were acquired from one segment of secondary or tertiary apical and/or basal dendrites using a two-photon microscope (Bruker Nano, Inc) with a pulsed Ti::sapphire laser (MaiTai HP; Spectra Physics) tuned to 920 nm (4–5 mW at the sample). All images are maximum projections of three-dimensional image stacks after applying a median filter (2 × 2) to the raw image data. Quantification of dendritic spines All distinct protrusions emanating from the dendritic shaft, regardless of shape, were counted and measured on red fluorescent images using ImageJ (NIH). In all neurons included for analysis, overall spine density did not significantly change over the imaging session. This ensured that the density of synapses under examination was not influenced by cell health (which was always monitored), as widespread spine/filopodia-like structures can form in cases where cell health is compromised. Estimated spine volume was measured from bleed-through–corrected and background-subtracted red fluorescence intensities using the integrated pixel intensity of a boxed region of interest (ROI) surrounding the spine head, as described previously (Kim et al, 2022a ). Quantification of gephyrin puncta Gephyrin fluorescence intensity was calculated from bleed-through–corrected and background-subtracted green (GFP-gephyrin intrabody) fluorescence intensities using the integrated pixel intensity of a boxed region surrounding a GFP-gephyrin punctum as described previously (Kim et al, 2022a ). Gephyrin enrichment in dendritic shafts was calculated by normalizing GFP-gephyrin fluorescence intensities (as described above) for each punctum to the mean GFP fluorescence intensity determined from four background ROIs on the same dendritic shaft. GFP-gephyrin enrichment was defined as a gephyrin punctum in cases where the ratio of green fluorescence from a punctum to green fluorescence from the dendritic background (G p /G d ) was > 1. Among these GFP puncta, those with green fluorescence intensities more than two standard deviations (2 SDs) greater than the local background green fluorescence levels measured from two ROIs near the puncta were classified as true gephyrin puncta and counted in images from the green channel using ImageJ (NIH). A summary of the criteria is as follows: gephyrin puncta, G p /G d > 1 and expression level > 2 SDs of background levels; no puncta, G p /G d < 1 or expression level < 2 SDs of the background level. Stereotactic surgery and virus injections P30 mice were anesthetized by intraperitoneal injection of a saline-based 2% Avertin solution (2,2,2-tribromoethyl alcohol dissolved in tert-amyl alcohol [Sigma-Aldrich]), and their heads were fixed in a stereotactic apparatus. Recombinant AAV virus was injected into the hippocampal CA1 region (coordinates: AP −2.1 mm, ML ± 1.3 mm, and DV 1.8 mm) with a Hamilton syringe at a flow rate of 100 nl/min (injected volume, 300 nl) using a Nanoliter 2010 Injector (World Precision Instruments). Each injected mouse was restored to its home cage for 2–4 weeks and used subsequently for immunohistochemical analyses, electrophysiological recordings, or behavioral analyses. Electrophysiology Cultured hippocampal neurons Hippocampal cultured neurons from wild-type rat embryos were transfected with MDGA1 and MDGA2 patient mutant variants, followed by analysis at DIV13–16, as previously described (El Chehadeh et al, 2022 ). Using a Model P-97 pipette puller (Sutter Instruments), pipettes were pulled from borosilicate glass (0.86 mm I.D., 1.5 mm O.D.). The resistance of patch pipettes containing an external solution ranged from 3 to 6 MΩ. The internal solution included 145 mM CsCl, 0.3 mM Na-GTP, 10 mM EGTA, 10 mM HEPES, 5 mM NaCl, and 4 mM Mg-ATP, with the pH adjusted to 7.2–7.4 with CsOH and an osmolarity of 290–295 mOsm/L for mEPSCs and mIPSCs. The external solution included 130 mM NaCl, 1 mM MgCl 2 , 2 mM CaCl 2 , 4 mM KCl, 10 mM D-glucose, and 10 mM HEPES with the pH calibrated to 7.2–7.4 with NaOH and an osmolarity of 300–305 mOsm/L. The whole-cell recording was performed at room temperature using M-TSC manipulators (SENSAPEX). Electrophysiological data were obtained using an Axon MultiClamp 700B amplifier and pCLAMP11 software and then digitized using an Axon DigiData 1550B data capture board (Axon Instruments); mEPSCs and mIPSCs were recorded at a holding potential of −70 mV. Synaptic currents were analyzed offline using Clampfit 10.8 software (Molecular Devices). For mEPSC recordings, the external solution contained 1 μM TTX and 50 μM picrotoxin to block Na + currents and GABA A receptors, respectively. For mIPSC recordings, 1 µM TTX, 10 µM CNQX, and 50 µM D-AP5 were included in the extracellular solution to block Na + currents, AMPA receptors, and NMDA receptors, respectively. Acute hippocampal CA1 slices Hippocampal slices (300 μm) were prepared from mice aged P10–12 days, P18–20 days, or 8–10 weeks. Following anesthesia with isoflurane, the mice were euthanized, and their brains were swiftly extracted and placed in a chilled solution with low calcium and high magnesium levels, which was oxygenated (95% O 2 and 5% CO 2 ). This solution consisted of 3.3 mM KCl, 1.3 mM NaH 2 PO 4 , 26 mM NaHCO 3 , 11 mM D-glucose, 0.5 mM CaCl 2 , 10 mM MgCl 2 , and 211 mM sucrose. Hippocampal slices were prepared using a vibratome (VT1000S; Leica) and transferred to a storage chamber filled with oxygenated artificial cerebrospinal fluid (aCSF) containing 124 mM NaCl, 3.3 mM KCl, 1.3 mM NaH 2 PO 4 , 26 mM NaHCO 3 , 11 mM D-glucose, 2 mM CaCl 2 , and 1 mM MgCl 2 . Slices were incubated at 30 °C for at least 60 min prior to experimentation. Subsequently, slices were transferred into the recording chamber and subjected to constant perfusion with standard aCSF that was oxygenated with a mixture of 95% O 2 and 5% CO 2 . All experiments were performed at 32 °C, and slices were used within 4 h. Only cells with an access resistance (R a ) smaller than 30 MΩ were analyzed. All the recordings were conducted using a Multiclamp 700B amplifier and DigiData 1550B Digitizer. To assess mIPSCs, whole cell recordings were performed from hippocampal CA1 pyramidal neurons using glass pipettes (3–5 MΩ) filled with a solution containing 145 mM CsCl, 5 mM NaCl, 10 mM HEPES, 10 mM EGTA, 4 mM Mg-ATP, and 0.3 mM Na-GTP, adjusted to pH 7.2–7.3 with CsOH. The membrane potential was clamped at −70 mV. mIPSCs were isolated by the external application of 50 μM D-AP5, 10 μM CNQX, and 1 μM tetrodotoxin to inhibit NMDARs, AMPARs, and Na + channels, respectively. For the measurement of evoked inhibitory postsynaptic currents (eIPSCs) and delayed asynchronous IPSCs (aIPSCs), patch pipettes were filled with an internal solution consisting of 130 mM Cs-methanesulfonate, 5 mM TEA-Cl, 8 mM NaCl, 0.5 mM EGTA, 10 mM HEPES, 4 mM Mg-ATP, 0.4 mM Na-GTP, 1 mM QX-314, and 10 mM disodium phosphocreatine, adjusted to pH 7.2–7.3 with CsOH. aIPSC recordings were carried out in aCSF where Ca 2+ was replaced by Sr 2+ (124 mM NaCl, 3.3 mM KCl, 1.3 mM NaH 2 PO 4 , 26 mM NaHCO 3 , 11 mM D-glucose, 8 mM SrCl 2 , and 1 mM MgCl 2 ). Cells were voltage-clamped at 0 mV. Electrical stimulation was applied using a concentric bipolar electrode (FHC), placed on the stratum radiatum (SR) or stratum lacunosum moleculare (SLM) of the hippocampal CA1 to record somatic inhibition and dendritic inhibition, respectively. eIPSCs and aIPSCs were isolated by inhibiting NMDARs and AMPARs using 50 μM D-AP5, and 10 μM CNQX. To record eIPSCs input-output (I-O) curves, electrical stimulation was applied ranging from 10 to 60 μA with 10 μA increments. Average eIPSC I-O values were measured from three consecutive sweeps. For assessing paired-pulse ratios (PPRs), pairs of stimuli were administered at a 10-Hz frequency. PPRs were calculated as the ratio of the 2nd eIPSC amplitude to the 1st eIPSC amplitude. Average PPRs were analyzed from three consecutive sweeps. To assess delayed aIPSCs, a 40-Hz train stimulation was applied. Synaptic events during the 1-s period after train stimulation were quantified and normalized to the 1-s period before the train stimulation. Circular dichroism spectroscopy Far-UV circular dichroism (CD) from 200 nm to 260 nm was performed on a spectrophotometer (J-1500; JASCO) equipped with a temperature controller and a 1-mm path-length quartz cuvette. Samples of wild-type and mutant MDGA1 (0.2 mg/ml) in PBS were prepared for CD measurements. The bandwidth was 5 nm and the scanning speed was 20 nm/min with a 4-s digital integration time. All spectra were obtained by averaging five scans with buffer subtraction, and the temperature was maintained at 20 °C. Millidegrees were converted to molar residue ellipticity. In utero electroporation E15.5 timed-pregnant ICR mice (Daehan Biolink) were anesthetized with isoflurane, and the uterine horns were exposed by laparotomy. Endotoxin-free plasmids (pCIG2, pCIG2-HA-MDGA1WT, pCIG2-HA-MDGA1 V116M/A688V, or pCIG2-HA-MDGA1 Y635C/E756Q; 2 μg/μl) and 0.5% Fast Green were injected into a lateral ventricle of E15.5 embryos. Electroporation was performed by placing the anode on the side of DNA injection and the cathode on the other side of the head to target cortical progenitors. Four pulses of 40 V (50 ms duration, 500 ms interval) were applied using an electroporator (ECM 830; BTX). After electroporation, the uterine horns were returned into the maternal abdomen. Measurement of estradiol levels by enzyme-linked immunosorbent assay Estradiol levels were quantified using a Mouse Estradiol (E2) ELISA Kit (AssayGenie; Cat# MOEB2488) following the manufacturer’s instructions with minor modifications. Brains from Mdga1 Y636C/E751Q KI mice were rapidly dissected, rinsed in ice-cold 1× PBS to remove excess blood, and homogenized in 20 mL of 1× PBS. Tissue homogenates were stored overnight at ≤ −20 °C and subjected to two freeze–thaw cycles to ensure complete cell lysis. The samples were centrifuged at 5000 × g for 5 min at 4 °C, and the resulting supernatants were collected for ELISA measurements. For the assay, 50 µL of each sample or standard was added to the designated wells, followed by the addition of 50 µL of Detection Reagent A working solution. The plates were sealed and incubated for 1 h at 37 °C, then washed three times with Wash Buffer. After washing, 100 µL of Detection Reagent B working solution was added to each well, and the plates were incubated for 45 min at 37 °C. A second wash step was performed five times. Subsequently, 90 µL of Substrate Solution was added to each well, and the plates were incubated for 15 min at 37 °C protected from light. The reaction was terminated by adding 50 µL of Stop Solution. Absorbance was immediately measured at 450 nm using a preheated microplate reader. Mouse behavioral tests Male and female mice aged 7–9 weeks were used for all behavioral tests. Tests were conducted in a soundproofed room under dim lighting (< 5 lux). All behavioral examinations were conducted in a controlled environment, and mice were acclimated to the experimental room prior to testing. The order of testing was randomized, and experiments were conducted at consistent timeframes to minimize potential circadian influences. Open-field test Mice were placed in a white acrylic open-field box (40 × 40 × 40 cm) and allowed to freely explore the environment for 10 min in dim light (< 5 lux). The distance traveled and time spent in the center zone were recorded using a top-view infrared camera and analyzed using EthoVision XT 15.5 software (Noldus). Y-maze test A Y-shaped white acrylic maze with three 35-cm-long arms at a 120° angle from each other was used. Mice were introduced into the maze and allowed to explore freely for 8 min. An entry was counted when all four limbs of the mouse were within the arm. The movement of mice was recorded with a top-view infrared camera and analyzed using EthoVision XT 15.5 software. Novel object-recognition test An open-field chamber was used. Mice were habituated to the chamber for 10 min. For training sessions, two identical objects were placed in the center of the chamber at regular intervals, and mice were allowed to explore the objects for 10 min. After the training session, mice were returned to their home cages for 24 h. For test sessions, one of the two objects was replaced with a novel object, which was placed in the same position in the chamber. Mice were returned to the chamber and allowed to explore freely for 10 min. Movement was recorded using an infrared camera, and the number and duration of contacts were analyzed using EthoVision XT 15.5. The discrimination index was calculated as the difference between the time spent exploring novel and familiar objects during the test phase. Elevated plus-maze test A white acrylic maze with two open arms (30 × 5 × 0.5 cm) and two closed arms (30 × 5 × 30 cm) was elevated by 75 cm above the floor. Mice were individually placed at the center of the elevated plus-maze and allowed to freely move for 5 min. The time spent in each arm and the number of arm entries were analyzed using EthoVision XT 15.5. Light and dark box transition test The apparatus consisted of a roofless box divided into a closed dark chamber and a brightly illuminated chamber (350 lux). A small entrance allowed free travel between the two chambers. The time spent in each chamber and the number of transitions were analyzed. Three-chamber test The testing apparatus consisted of a white acrylic box divided into three chambers (20 × 40 × 22 cm each) with small openings on the dividing walls. Wire cups were placed in the corners of both side chambers. Mice were placed in the central chamber for a 10-min habituation period. Following this time, an age-matched social conspecific was placed in the wire cup on the left side-chamber and sociability was assessed by measuring subject exploration times for the enclosed conspecific and empty cup during a 10-min session. In the final 10-min session, a new social conspecific was placed into the previously empty wire cup, and social recognition was assessed. Prepulse inhibition test Mice were individually acclimated to the experimental room for a minimum of 30 min under standard housing conditions with reduced extraneous noise. Each mouse was placed in a sound-attenuating chamber equipped with a startle response measurement apparatus (SR-LAB-Startle Response System; SD Ins.), ensuring minimal movement during testing. On day 1, after a 5-min habituation period in the absence of stimuli, mice were exposed to a series of startle-eliciting stimuli (from 0 to 120 dB) to familiarize them with the startle response. On day 2, for the PPI trials, prepulse stimuli (73, 76, 79, or 82 dB) were presented before the startle-eliciting stimuli (120 dB). The order and inter-trial intervals of prepulse and startle stimuli were randomized, including control trials with no prepulse. Startle responses were recorded for each trial using PPI test software. Data analysis involved calculating the percentage PPI for each prepulse intensity using the formula: PPI (%) = ( R pulse − R pre − pulse )/ R pulse × 100 ( R pulse is the average response magnitude in the absence of a prepulse and R pre − pulse the average response magnitude in the presence of a prepulse). Ultrasonic vocalization test Following a 3-day period of single housing, each adult mouse was individually placed in a sound-attenuating chamber equipped with a microphone for USV recording. To elicit vocalizations, an anesthetized conspecific of the opposite sex was introduced. Interaction and vocalizations were recorded for 10 min. For pup experiments, sessions were conducted on P3, P6, P9, and P12. Each pup was temporarily separated from the dam and placed on a heat pad maintained at 370 C. USV recordings were analyzed using DeepSqueak software for the detection and categorization of ultrasonic vocalizations. Forced swim test Mice were individually placed in a glass cylinder (15 × 30 cm) containing water (24 ± 1 °C; depth, 15 cm). Mice were forced to swim for 6 min, and the duration of immobility was measured during the final 4 min. The latency to immobility from the start of the test (delay between the start of the test and appearance of the first bout of immobility, defined as a period of at least 1 s without any active escape behavior) and duration of immobility (defined as the time not spent actively exploring the cylinder or trying to escape from it) were measured. Immobility time was defined as the time the mouse spent floating in the water without struggling, making only minor movements that were strictly necessary to maintain its head above water. Tail-suspension test Mice were individually subjected to the tail suspension test, involving suspension by adhesive tape affixed approximately 1–2 cm from the tail tip. The testing apparatus consisted of a standardized setup, and mice were allowed to hang freely. The total test duration was set at 5 min during which mice were examined for despair-related behaviors. The immobility time was recorded for the final 4 min of the test. Immobility was operationally defined as the absence of limb or body movement, and the recorded duration represented the time spent suspended without exhibiting active escape behaviors or exploratory movements. Additionally, latency to immobility was measured as the period during which the mouse did not engage in any active escape behavior for at least 1 s. Marble burying test Mice were individually acclimated to the experimental room for at least 30 min prior to testing. Standard mouse bedding was placed in home cages to provide a familiar environment. For the marble burying test, a 5-cm layer of fresh bedding was uniformly spread in the Plexiglas cage. Next, 15 glass marbles were evenly distributed on the bedding surface. Each mouse was introduced into the cage and allowed free exploration for a duration of 30 min. During the marble burying phase, the number of marbles buried (defined as being at least two-thirds covered by bedding) was counted and quantified. Negative-stain electron microscopy Human MDGA1 full ectodomain (16 μg/ml) and the Y635C/E756Q mutant (19 μg/ml) were applied to a glow-discharged carbon-coated 300 mesh copper grids (Electron Microscopy Science) and stained with a 0.075% uranyl formate solution. Images were acquired using a Tecnai T12 Bio-TWIN transmission electron microscope equipped with an FEI Eagle 4 K charge-coupled device camera operating at 120 kV. Proteomics Peptide preparation and phosphopeptide enrichment Hippocampi from male control, Mdga1 -cKO, or Mdga1 Y636C/E751Q KI mice (4 replicates) of adult (P60) and E18 mouse embryos were lysed with 1× SDS buffer (5% SDS and 50 mM triethylammonium bicarbonate, pH 8.5). The lysates were digested using an S-trap method with trypsin, according to the provided protocol. The digested peptides were labeled with 18-plex TMT isotopes (Thermo Fisher Scientific), dried by SpeedVac, and desalted with Pierce peptide desalting spin columns (Thermo Fisher Scientific). The combined samples were fractionated into 20 fractions by basic reverse phase liquid chromatography. 5% of the sample was used for total proteome analysis and then the remaining 95% was reserved for phosphoproteomic analysis. The remaining latter portion was combined into 10 fractions for phosphopeptide enrichment. First, Ni-NTA magnetic agarose beads were washed three times with DW and then incubated with 100 mM EDTA (pH 8.0) for 30 min on a rotator. Next, 100 mM FeCl 3 solution was added, and the mixture was incubated with rotation for 30 min. The prepared Fe 3+ -NTA beads were washed with DW and incubated overnight with each sample in 80% ACN with 0.1% TFA, at 4 °C on a rotator. The bead-bound phosphopeptides were eluted with elution buffer (50% ACN in 1% ammonium hydroxide), promptly acidified to pH 3.5–4.0 using 10% TFA, and vacuum dried. LC-MS/MS and data analysis LC-MS/MS analysis was performed using an UltiMate 3000 RSLCnano system (Thermo Fisher Scientific) coupled to a Orbitrap Fusion Lumos mass spectrometer (Thermo Fisher Scientific). Mobile phases A and B were composed of 0 and 95.0% acetonitrile containing 0.1% formic acid, respectively. The LC gradient was applied at a flow rate of 250 nL/min during 120 min for peptide separation. The Orbitrap Fusion Lumos was operated in data-dependent mode, and the MS2 scans were performed with HCD fragmentation (37.5% collision energy). MS/MS spectra were identified and quantified using the Integrated Proteomics Pipeline software with the UniProt mouse database and the following search parameters: precursor mass tolerance, 20 ppm; fragment ion mass tolerance, 200 ppm; two or more peptide assignments for protein identification at a false positive rate < 0.01; and TMT reporter ion mass tolerance, 20 ppm. For phosphopeptide identification, phosphorylation of serine, threonine, and tyrosine was set as the differential modifications, with a maximum of three additional modifications permitted. Statistical analysis was conducted using the Perseus software (version 1.6.15), and expression levels were compared using Welch’s t-test ( P < 0.05). Statistical analysis No statistical method was used to pre-determine sample sizes. Rather, the sample sizes were selected based on previous studies published in the field (see Life Science Reporting Summary for references). Animals in the same litter were randomly assigned to the different treatment groups in the various experiments. All statistical analyses were performed using GraphPad Prism 7 software (RRID: SCR_002798 ). The normality of distributed data was determined using the Shapiro–Wilk normality test. Normally distributed data were compared using Student’s t test or the analysis of variance (ANOVA); non-normally distributed data were compared using the Mann–Whitney U test, Kruskal–Wallis test followed by Dunn’s post hoc test, or Tukey’s multiple comparison test. Outliers were identified using Grubb’s test ( P < 0.05). Dataset EV9 presents detailed statistics and numerical values. The statistical analyses are described in more detail in the Reporting Summary linked to this article. Supplementary information Table EV1 (75.5KB, doc) Table EV2 (37KB, doc) Table EV3 (56KB, doc) Appendix (5MB, pdf) Peer Review File (1.7MB, pdf) Dataset EV1 (65.9KB, xlsx) Dataset EV2 (15.3KB, xlsx) Dataset EV3 (3.8MB, xlsx) Dataset EV4 (3.1MB, xlsx) Dataset EV5 (2.1MB, xlsx) Dataset EV6 (2.4MB, xlsx) Dataset EV7 (200.2KB, xlsx) Dataset EV8 (312.8KB, xlsx) Dataset EV9 (226.2KB, xlsx) Source data Fig. 1 (62.9MB, zip) Source data Fig. 2 (83.2MB, zip) Source data Fig. 3 (7.2MB, zip) Source data Fig. 4 (6.6MB, zip) Source data Fig. 5 (74.1MB, zip) Source data Fig. 7 (14MB, zip) Expanded View Figures (7.4MB, pdf) Acknowledgements The authors thank Jinha Kim (DGIST, Korea) for technical assistance. This study was supported by the National Creative Research Initiative Program of the Ministry of Science and ICT (RS-2022-NR070708 to Jaewon K), the National Research Foundation of Korea (NRF) funded by the Korea Government (2022R1C1C200003499 to Seungjoon K; RS-2024-00339642 to HK; NR-2020R1C1C1003426 to J-YA; RS-2023-00264980 to S-KK; RS-2024-00399031 to JYK; and RS-2023-NR076948 to JWU), the Korea Basic Science Institute (C523400 to JYK), Institute for Basic Science (IBS-R030-C1 to HMK), and National Institutes of Health (R01MH124778, R21MH126073 and R21NS133681 to WCO). Author contributions Seungjoon Kim : Formal analysis; Funding acquisition; Investigation; Methodology; Writing—review and editing. Hyeonho Kim : Formal analysis; Funding acquisition; Investigation; Methodology; Writing—review and editing. Javier Porta Pelayo : Formal analysis; Investigation; Methodology. Sara Alvarez : Formal analysis; Investigation; Methodology. Gyubin Jang : Formal analysis; Investigation; Methodology. Jinhu Kim : Formal analysis; Investigation; Methodology. Byeongchan Kim : Formal analysis; Investigation; Methodology. Victoria M Hoelscher : Formal analysis; Methodology. Beatriz Calleja-Pérez : Formal analysis. Hyunsu Jung : Formal analysis; Investigation; Methodology. Yeji Yang : Data curation; Formal analysis; Investigation; Methodology. Hea Ji Lee : Data curation; Formal analysis; Investigation; Methodology. Jihae Lee : Data curation; Formal analysis; Methodology. Seoyeon Kim : Data curation; Formal analysis; Methodology. Mar Jim énez de la Peña : Formal analysis. Yelin Lee : Methodology. Sohye Kim : Methodology. Ah-reum Han : Methodology. Dong Sun Lee : Formal analysis. Sangho Ji : Formal analysis. Wookyung Yu : Formal analysis; Supervision. Ho Min Kim : Formal analysis; Supervision; Funding acquisition. Joon-Yong An : Formal analysis; Supervision; Funding acquisition. Won Chan Oh : Formal analysis; Supervision; Funding acquisition. Seok-Kyu Kwon : Formal analysis; Supervision; Funding acquisition. Jin Young Kim : Formal analysis; Supervision; Funding acquisition. Ji Won Um : Supervision; Funding acquisition; Project administration. Alberto Fernández-Jaén : Conceptualization; Formal analysis; Supervision; Writing—review and editing. Jaewon Ko : Conceptualization; Formal analysis; Supervision; Funding acquisition; Investigation; Writing—original draft; Project administration; Writing—review and editing. Source data underlying figure panels in this paper may have individual authorship assigned. Where available, figure panel/source data authorship is listed in the following database record: biostudies:S-SCDT-10_1038-S44321-026-00402-y . Data availability The raw MS data files for total proteomics have been deposited to the MassIVE repository with identifier PXD057134 ( ftp://massive-ftp.ucsd.edu/v07/MSV000096187/ ) for data from Mdga1 -cKO mice and identifier PXD057136 ( ftp://massive-ftp.ucsd.edu/v07/MSV000096188/ ) for data from Mdga1 Y636C/E751Q KI mice. The source data of this paper are collected in the following database record: biostudies:S-SCDT-10_1038-S44321-026-00402-y . Disclosure and competing interests statement The authors declare no competing interests. Footnotes These authors contributed equally: Seungjoon Kim, Hyeonho Kim, Javier Porta Pelayo, Sara Alvarez. Contributor Information Alberto Fernández-Jaén, Email: [email protected]. Jaewon Ko, Email: [email protected]. Supplementary information Expanded view data, supplementary information, appendices are available for this paper at 10.1038/s44321-026-00402-y. References Ackerman SD, Perez-Catalan NA, Freeman MR, Doe CQ (2021) Astrocytes close a motor circuit critical period. Nature 592:414–420 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Ali H, Marth L, Krueger-Burg D (2020) Neuroligin-2 as a central organizer of inhibitory synapses in health and disease. Sci Signal 13:eabd8379 [ DOI ] [ PubMed ] [ Google Scholar ] Andrade-Talavera Y, Perez-Rodriguez M, Prius-Mengual J, Rodriguez-Moreno A (2023) Neuronal and astrocyte determinants of critical periods of plasticity. Trends Neurosci 46:566–580 [ DOI ] [ PubMed ] [ Google Scholar ] Banker SM, Gu X, Schiller D, Foss-Feig JH (2021) Hippocampal contributions to social and cognitive deficits in autism spectrum disorder. Trends Neurosci 44:793–807 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Berger JM, Rohn TT, Oxford JT (2013) Autism as the early closure of a neuroplastic critical period normally seen in adolescence. Biol Syst Open Access. 10.4172/2329-6577.1000118 [ DOI ] [ PMC free article ] [ PubMed ] Biederer T, Kaeser PS, Blanpied TA (2017) Transcellular nanoalignment of synaptic function. Neuron 96:680–696 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Bolte S, Neufeld J, Marschik PB, Williams ZJ, Gallagher L, Lai MC (2023) Sex and gender in neurodevelopmental conditions. Nat Rev Neurol 19:136–159 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Brose N, O’Connor V, Skehel P (2010) Synaptopathy: dysfunction of synaptic function?. Biochem Soc Trans 38:443–444 [ DOI ] [ PubMed ] [ Google Scholar ] Bruce MR, Jones KL, Vernon AC, Silverman JL, Crawley JN, Ellegood J, Lerch JP, Van de Water J (2021) Sexually dimorphic neuroanatomical differences relate to ASD-relevant behavioral outcomes in a maternal autoantibody mouse model. Mol Psychiatry 26:7530–7537 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Callige M, Richard-Foy H (2006) Ligand-induced estrogen receptor alpha degradation by the proteasome: new actors?. Nucl Recept Signal 4:e004 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Chever O, Pannasch U, Ezan P, Rouach N (2014) Astroglial connexin 43 sustains glutamatergic synaptic efficacy. Philos Trans R Soc Lond B Biol Sci 369:20130596 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Choi L, An JY (2021) Genetic architecture of autism spectrum disorder: Lessons from large-scale genomic studies. Neurosci Biobehav Rev 128:244–257 [ DOI ] [ PubMed ] [ Google Scholar ] Chun S, Fay JC (2009) Identification of deleterious mutations within three human genomes. Genome Res 19:1553–1561 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Coghlan S, Horder J, Inkster B, Mendez MA, Murphy DG, Nutt DJ (2012) GABA system dysfunction in autism and related disorders: from synapse to symptoms. Neurosci Biobehav Rev 36:2044–2055 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Connor SA, Ammendrup-Johnsen I, Kishimoto Y, Karimi Tari P, Cvetkovska V, Harada T, Ojima D, Yamamoto T, Wang YT, Craig AM (2017) Loss of synapse repressor MDGA1 enhances perisomatic inhibition, confers resistance to network excitation, and impairs cognitive function. Cell Rep 21:3637–3645 [ DOI ] [ PubMed ] [ Google Scholar ] Connor SA, Elegheert J, Xie Y, Craig AM (2019) Pumping the brakes: suppression of synapse development by MDGA-neuroligin interactions. Curr Opin Neurobiol 57:71–80 [ DOI ] [ PubMed ] [ Google Scholar ] Contractor A, Ethell IM, Portera-Cailliau C (2021) Cortical interneurons in autism. Nat Neurosci 24:1648–1659 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] de la Torre-Ubieta L, Won H, Stein JL, Geschwind DH (2016) Advancing the understanding of autism disease mechanisms through genetics. Nat Med 22:345–361 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Delorme R, Ey E, Toro R, Leboyer M, Gillberg C, Bourgeron T (2013) Progress toward treatments for synaptic defects in autism. Nat Med 19:685–694 [ DOI ] [ PubMed ] [ Google Scholar ] Demontis D, Walters GB, Athanasiadis G, Walters R, Therrien K, Nielsen TT, Farajzadeh L, Voloudakis G, Bendl J, Zeng B et al (2023) Genome-wide analyses of ADHD identify 27 risk loci, refine the genetic architecture and implicate several cognitive domains. Nat Genet 55:198–208. [DATASET] [Supplementary Table 7] [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Dong Y, Li W, Meng J, Wang P, Sun M, Zhou F, Li D, Shu J, Cai C (2023) Pathogenicity analysis and splicing rescue of a classical splice site variant (c.1343+1G>T) of CNOT1 gene associated with neurodevelopmental disorders. Am J Med Genet A 191:2775–2782 [ DOI ] [ PubMed ] [ Google Scholar ] Doornaert EE, Mohamad AE, Johal G, Allman BL, Mohrle D, Schmid S (2025) Postnatal environment affects auditory development and sensorimotor gating in a rat model for autism spectrum disorder. Front Neurosci 19:1565919 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Dougherty JD, Marrus N, Maloney SE, Yip B, Sandin S, Turner TN, Selmanovic D, Kroll KL, Gutmann DH, Constantino JN et al (2022) Can the “female protective effect” liability threshold model explain sex differences in autism spectrum disorder?. Neuron 110:3243–3262 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] El Chehadeh S, Han KA, Kim D, Jang G, Bakhtiari S, Lim D, Kim HY, Kim J, Kim H, Wynn J et al (2022) SLITRK2 variants associated with neurodevelopmental disorders impair excitatory synaptic function and cognition in mice. Nat Commun 13:4112 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Elegheert J, Cvetkovska V, Clayton AJ, Heroven C, Vennekens KM, Smukowski SN, Regan MC, Jia W, Smith AC, Furukawa H et al (2017) Structural mechanism for modulation of synaptic neuroligin-neurexin signaling by MDGA proteins. Neuron 95:896–913 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Ellison GW, Myers LW, Mickey MR, Graves MC, Tourtellotte WW, Syndulko K, Holevoet-Howson MI, Lerner CD, Frane MV, Pettler-Jennings P (1989) A placebo-controlled, randomized, double-masked, variable dosage, clinical trial of azathioprine with and without methylprednisolone in multiple sclerosis. Neurology 39:1018–1026 [ DOI ] [ PubMed ] [ Google Scholar ] Fernandes D, Carvalho AL (2016) Mechanisms of homeostatic plasticity in the excitatory synapse. J Neurochem 139:973–996 [ DOI ] [ PubMed ] [ Google Scholar ] Ferri SL, Abel T, Brodkin ES (2018) Sex differences in autism spectrum disorder: a review. Curr Psychiatry Rep 20:9 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Filice F, Janickova L, Henzi T, Bilella A, Schwaller B (2020) The parvalbumin hypothesis of autism spectrum disorder. Front Cell Neurosci 14:577525 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Finney CA, Shvetcov A, Westbrook RF, Morris MJ, Jones NM (2021) The selective estrogen receptor modulator tamoxifen protects against subtle cognitive decline and early markers of injury 24 h after hippocampal silent infarct in male Sprague-Dawley rats. Horm Behav 134:105016 [ DOI ] [ PubMed ] [ Google Scholar ] Fu JM, Satterstrom FK, Peng M, Brand H, Collins RL, Dong S, Wamsley B, Klei L, Wang L, Hao SP et al (2022) Rare coding variation provides insight into the genetic architecture and phenotypic context of autism. Nat Genet 54:1320–1331. [DATASET] [Supplementary Table 11] [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Galineau L, Arlicot N, Dupont AC, Briend F, Houy-Durand E, Tauber C, Gomot M, Gissot V, Barantin L, Lefevre A et al (2023) Glutamatergic synapse in autism: a complex story for a complex disorder. Mol Psychiatry 28:801–809 [ DOI ] [ PubMed ] [ Google Scholar ] Giua G, Iezzi D, Caceres-Rodriguez A, Strauss B, Chavis P, Manzoni OJ (2024) Sex-specific modulation of early life vocalization and cognition by Fmr1 gene dosage in a mouse model of Fragile X Syndrome. Biol Sex Differ 15:18 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Gressens P (2000) Mechanisms and disturbances of neuronal migration. Pediatr Res 48:725–730 [ DOI ] [ PubMed ] [ Google Scholar ] Hammes SR, Levin ER (2007) Extranuclear steroid receptors: nature and actions. Endocr Rev 28:726–741 [ DOI ] [ PubMed ] [ Google Scholar ] Han KA, Kim YJ, Yoon TH, Kim H, Bae S, Um JW, Choi SY, Ko J (2020) LAR-RPTPs directly interact with neurexins to coordinate bidirectional assembly of molecular machineries. J Neurosci 40:8438–8462 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Han KA, Yoon TH, Kim J, Lee J, Lee JY, Jang G, Um JW, Kim JK, Ko J (2024) Specification of neural circuit architecture shaped by context-dependent patterned LAR-RPTP microexons. Nat Commun 15:1624 [ DOI ] [ PMC free article ] [ PubMed ] Heavner WE, Smith SEP (2020) Resolving the synaptic versus developmental dichotomy of autism risk genes. Trends Neurosci 43:227–241 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Heyne HO, Singh T, Stamberger H, Abou Jamra R, Caglayan H, Craiu D, De Jonghe P, Guerrini R, Helbig KL, Koeleman BPC et al (2018) De novo variants in neurodevelopmental disorders with epilepsy. Nat Genet 50:1048–1053. [DATASET] [Supplementary Table 3] [ DOI ] [ PubMed ] [ Google Scholar ] Hill RA, Kouremenos K, Tull D, Maggi A, Schroeder A, Gibbons A, Kulkarni J, Sundram S, Du X (2020) Bazedoxifene - a promising brain active SERM that crosses the blood brain barrier and enhances spatial memory. Psychoneuroendocrinology 121:104830 [ DOI ] [ PubMed ] [ Google Scholar ] Hodge RD, Bakken TE, Miller JA, Smith KA, Barkan ER, Graybuck LT, Close JL, Long B, Johansen N, Penn O et al (2019) Conserved cell types with divergent features in human versus mouse cortex. Nature 573:61–68 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Hossain MR, Jamal M, Tanoue Y, Ojima D, Takahashi H, Kubota T, Ansary TM, Ito A, Tanaka N, Kinoshita H et al (2020) MDGA1-deficiency attenuates prepulse inhibition with alterations of dopamine and serotonin metabolism: An ex vivo HPLC-ECD analysis. Neurosci Lett 716:134677 [ DOI ] [ PubMed ] [ Google Scholar ] Huang GZ, Woolley CS (2012) Estradiol acutely suppresses inhibition in the hippocampus through a sex-specific endocannabinoid and mGluR-dependent mechanism. Neuron 74:801–808 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Humphries EM, Ahn K, Kember RL, Lopes FL, Mocci E, Peralta JM, Blangero J, Glahn DC, Goes FS, Zandi PP et al (2023) Genome-wide significant risk loci for mood disorders in the Old Order Amish founder population. Mol Psychiatry 28:5262–5271 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Hwang BJ, Mohamed MA, Brasic JR (2017) Molecular imaging of autism spectrum disorder. Int Rev Psychiatry 29:530–554 [ DOI ] [ PubMed ] [ Google Scholar ] Ishikawa T, Gotoh N, Murayama C, Abe T, Iwashita M, Matsuzaki F, Suzuki T, Yamamoto T (2011) IgSF molecule MDGA1 is involved in radial migration and positioning of a subset of cortical upper-layer neurons. Dev Dyn 240:96–107 [ DOI ] [ PubMed ] [ Google Scholar ] Jang S, Lee H, Kim E (2017) Synaptic adhesion molecules and excitatory synaptic transmission. Curr Opin Neurobiol 45:45–50 [ DOI ] [ PubMed ] [ Google Scholar ] Jeon SJ, Gonzales EL, Mabunga DFN, Valencia ST, Kim DG, Kim Y, Adil KJL, Shin D, Park D, Shin CY (2018) Sex-specific behavioral features of rodent models of autism spectrum disorder. Exp Neurobiol 27:321–343 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Jumper J, Evans R, Pritzel A, Green T, Figurnov M, Ronneberger O, Tunyasuvunakool K, Bates R, Zidek A, Potapenko A et al (2021) Highly accurate protein structure prediction with AlphaFold. Nature 596:583–589 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Kahler AK, Djurovic S, Kulle B, Jonsson EG, Agartz I, Hall H, Opjordsmoen S, Jakobsen KD, Hansen T, Melle I et al (2008) Association analysis of schizophrenia on 18 genes involved in neuronal migration: MDGA1 as a new susceptibility gene. Am J Med Genet B Neuropsychiatr Genet 147B:1089–1100 [ DOI ] [ PubMed ] [ Google Scholar ] Kazdoba TM, Leach PT, Crawley JN (2016) Behavioral phenotypes of genetic mouse models of autism. Genes Brain Behav 15:7–26 [ DOI ] [ PMC free article ] [ PubMed ] Kettenmann H, Kirchhoff F, Verkhratsky A (2013) Microglia: new roles for the synaptic stripper. Neuron 77:10–18 [ DOI ] [ PubMed ] [ Google Scholar ] Kim D, Jung H, Shirai Y, Kim H, Kim J, Lim D, Mori T, Lee H, Park D, Kim HY et al (2022b) IQSEC3 Deletion Impairs Fear Memory Through Upregulation of Ribosomal S6K1 Signaling in the Hippocampus. Biol Psychiatry 91:821–831 [ DOI ] [ PubMed ] Kim HY, Um JW, Ko J (2021) Proper synaptic adhesion signaling in the control of neural circuit architecture and brain function. Prog Neurobiol 200:101983 [ DOI ] [ PubMed ] [ Google Scholar ] Kim J, Kim S, Kim H, Hwang IW, Bae S, Karki S, Kim D, Ogelman R, Bang G, Kim JY et al (2022a) MDGA1 negatively regulates amyloid precursor protein-mediated synapse inhibition in the hippocampus. Proc Natl Acad Sci USA 119:e2115326119 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Kim JA, Kim D, Won SY, Han KA, Park D, Cho E, Yun N, An HJ, Um JW, Kim E et al (2017) Structural insights into modulation of neurexin-neuroligin trans-synaptic adhesion by MDGA1/neuroligin-2 complex. Neuron 94:1121–1131 [ DOI ] [ PubMed ] [ Google Scholar ] Kim S, Jang G, Kim H, Lim D, Han KA, Um JW, Ko J (2024a) MDGAs perform activity-dependent synapse type-specific suppression via distinct extracellular mechanisms. Proc Natl Acad Sci USA 121:e2322978121 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Kim S, Lee J, Koh IG, Ji J, Kim HJ, Kim E, Park J, Park JE, An JY (2024b) An integrative single-cell atlas for exploring the cellular and temporal specificity of genes related to neurological disorders during human brain development. Exp Mol Med 56:2271–2282. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Ko J (2025) MDGAs as synaptic suppressors with implications in neurodevelopmental disorders. Curr Opin Neurobiol 95:103130 [ DOI ] [ PubMed ] Kokash J, Alderson EM, Reinhard SM, Crawford CA, Binder DK, Ethell IM, Razak KA (2019) Genetic reduction of MMP-9 in the Fmr1 KO mouse partially rescues prepulse inhibition of acoustic startle response. Brain Res 1719:24–29 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Lambert JC, Ibrahim-Verbaas CA, Harold D, Naj AC, Sims R, Bellenguez C, DeStafano AL, Bis JC, Beecham GW, Grenier-Boley B et al (2013) Meta-analysis of 74,046 individuals identifies 11 new susceptibility loci for Alzheimer’s disease. Nat Genet 45:1452–1458. [DATASET] [Table 2] [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] LeBlanc JJ, Fagiolini M (2011) Autism: a “critical period” disorder?. Neural Plast 2011:921680 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Lee E, Lee J, Kim E (2017) Excitation/inhibition imbalance in animal models of autism spectrum disorders. Biol Psychiatry 81:838–847 [ DOI ] [ PubMed ] [ Google Scholar ] Lee H, Chofflet N, Liu J, Fan S, Lu Z, Resua Rojas M, Penndorf P, Bailey AO, Russell WK, Machius M et al (2023) Designer molecules of the synaptic organizer MDGA1 reveal 3D conformational control of biological function. J Biol Chem 299:104586 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Lee K, Kim Y, Lee SJ, Qiang Y, Lee D, Lee HW, Kim H, Je HS, Südhof TC, Ko J (2013) MDGAs interact selectively with neuroligin-2 but not other neuroligins to regulate inhibitory synapse development. Proc Natl Acad Sci USA 110:336–341 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Lepeta K, Lourenco MV, Schweitzer BC, Martino Adami PV, Banerjee P, Catuara-Solarz S, de La Fuente Revenga M, Guillem AM, Haidar M, Ijomone OM et al (2016) Synaptopathies: synaptic dysfunction in neurological disorders—a review from students to students. J Neurochem 138:785–805 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Li J, Liu J, Feng G, Li T, Zhao Q, Li Y, Hu Z, Zheng L, Zeng Z, He L et al (2011) The MDGA1 gene confers risk to schizophrenia and bipolar disorder. Schizophr Res 125:194–200 [ DOI ] [ PubMed ] [ Google Scholar ] Lima Caldeira G, Peca J, Carvalho AL (2019) New insights on synaptic dysfunction in neuropsychiatric disorders. Curr Opin Neurobiol 57:62–70 [ DOI ] [ PubMed ] [ Google Scholar ] Litwack ED, Babey R, Buser R, Gesemann M, O’Leary DD (2004) Identification and characterization of two novel brain-derived immunoglobulin superfamily members with a unique structural organization. Mol Cell Neurosci 25:263–274 [ DOI ] [ PubMed ] [ Google Scholar ] Long X, Nephew KP (2006) Fulvestrant (ICI 182,780)-dependent interacting proteins mediate immobilization and degradation of estrogen receptor-alpha. J Biol Chem 281:9607–9615 [ DOI ] [ PubMed ] [ Google Scholar ] Luhmann HJ, Fukuda A, Kilb W (2015) Control of cortical neuronal migration by glutamate and GABA. Front Cell Neurosci 9:4 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] McAlonan GM, Daly E, Kumari V, Critchley HD, van Amelsvoort T, Suckling J, Simmons A, Sigmundsson T, Greenwood K, Russell A et al (2002) Brain anatomy and sensorimotor gating in Asperger’s syndrome. Brain 125:1594–1606 [ DOI ] [ PubMed ] [ Google Scholar ] McCarthy MM, Wright CL (2017) Convergence of sex differences and the neuroimmune system in autism spectrum disorder. Biol Psychiatry 81:402–410 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Meyer-Dilhet G, Courchet J (2020) In utero cortical electroporation of plasmids in the mouse embryo. STAR Protoc 1:100027 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Moretto E, Murru L, Martano G, Sassone J, Passafaro M (2018) Glutamatergic synapses in neurodevelopmental disorders. Prog Neuropsychopharmacol Biol Psychiatry 84:328–342 [ DOI ] [ PubMed ] [ Google Scholar ] Mukherjee J, Cardarelli RA, Cantaut-Belarif Y, Deeb TZ, Srivastava DP, Tyagarajan SK, Pangalos MN, Triller A, Maguire J, Brandon NJ et al (2017) Estradiol modulates the efficacy of synaptic inhibition by decreasing the dwell time of GABA(A) receptors at inhibitory synapses. Proc Natl Acad Sci USA 114:11763–11768 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Nestler EJ, Hyman SE (2010) Animal models of neuropsychiatric disorders. Nat Neurosci 13:1161–1169 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Nisar S, Bhat AA, Masoodi T, Hashem S, Akhtar S, Ali TA, Amjad S, Chawla S, Bagga P, Frenneaux MP et al (2022) Genetics of glutamate and its receptors in autism spectrum disorder. Mol Psychiatry 27:2380–2392 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Ogelman R, Gomez Wulschner LE, Hoelscher VM, Hwang IW, Chang VN, Oh WC (2024) Serotonin modulates excitatory synapse maturation in the developing prefrontal cortex. Nat Commun 15:1368 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Ohtaka-Maruyama C, Okamoto M, Endo K, Oshima M, Kaneko N, Yura K, Okado H, Miyata T, Maeda N (2018) Synaptic transmission from subplate neurons controls radial migration of neocortical neurons. Science 360:313–317 [ DOI ] [ PubMed ] [ Google Scholar ] Ohtani-Kaneko R, Iwafuchi M, Iwakura T, Muraoka D, Yokosuka M, Shiga T, Watanabe C (2010) Effects of estrogen on synapsin I distribution in developing hypothalamic neurons. Neurosci Res 66:180–188 [ DOI ] [ PubMed ] [ Google Scholar ] Pan YH, Wu N, Yuan XB (2019) Toward a better understanding of neuronal migration deficits in autism spectrum disorders. Front Cell Dev Biol 7:205 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Peng L, Luo Q, Lu H (2017) Efficacy and safety of bazedoxifene in postmenopausal women with osteoporosis: a systematic review and meta-analysis. Medicine 96:e8659 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Perez-Garcia CG, O’Leary DDM (2016) Formation of the cortical subventricular zone requires MDGA1-mediated aggregation of basal progenitors. Cell Rep 14:560–571 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Perry W, Minassian A, Lopez B, Maron L, Lincoln A (2007) Sensorimotor gating deficits in adults with autism. Biol Psychiatry 61:482–486 [ DOI ] [ PubMed ] [ Google Scholar ] Pettem KL, Yokomaku D, Takahashi H, Ge Y, Craig AM (2013) Interaction between autism-linked MDGAs and neuroligins suppresses inhibitory synapse development. J Cell Biol 200:321–336 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Premoli M, Memo M, Bonini SA (2021) Ultrasonic vocalizations in mice: relevance for ethologic and neurodevelopmental disorders studies. Neural Regen Res 16:1158–1167 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Qi H, Zhang H, Zhao Y, Chen C, Long JJ, Chung WK, Guan Y, Shen Y (2021) MVP predicts the pathogenicity of missense variants by deep learning. Nat Commun 12:510 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Raina PM, Patel P, Parmar M (2024) Bazedoxifene. In: StatPearls [Internet]. Treasure Island (FL): StatPearls Publishing Ramaswami G, Geschwind DH (2018) Genetics of autism spectrum disorder. Handb Clin Neurol 147:321–329 [ DOI ] [ PubMed ] [ Google Scholar ] Ribot J, Breton R, Calvo CF, Moulard J, Ezan P, Zapata J, Samama K, Moreau M, Bemelmans AP, Sabatet V et al (2021) Astrocytes close the mouse critical period for visual plasticity. Science 373:77–81 [ DOI ] [ PubMed ] [ Google Scholar ] Robertson JF, Harrison M (2004) Fulvestrant: pharmacokinetics and pharmacology. Br J Cancer 90:S7–10 [ DOI ] [ PMC free article ] [ PubMed ] Roy DS, Zhang Y, Halassa MM, Feng G (2022) Thalamic subnetworks as units of function. Nat Neurosci 25:140–153 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Satterstrom FK, Kosmicki JA, Wang J, Breen MS, De Rubeis S, An JY, Peng M, Collins R, Grove J, Klei L et al (2020) Large-scale exome sequencing study implicates both developmental and functional changes in the neurobiology of autism. Cell 180:568–584. [DATASET] [Supplementary Table 2] [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Savtchenko LP, Rusakov DA (2007) The optimal height of the synaptic cleft. Proc Natl Acad Sci USA 104:1823–1828 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Scattoni ML, Gandhy SU, Ricceri L, Crawley JN (2008) Unusual repertoire of vocalizations in the BTBR T+tf/J mouse model of autism. PLoS ONE 3:e3067 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Singh T, Poterba T, Curtis D, Akil H, Al Eissa M, Barchas JD, Bass N, Bigdeli TB, Breen G, Bromet EJ et al (2022) Rare coding variants in ten genes confer substantial risk for schizophrenia. Nature 604:509–516. [DATASET] [Supplementary Table 13] [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Sohal VS, Rubenstein JLR (2019) Excitation-inhibition balance as a framework for investigating mechanisms in neuropsychiatric disorders. Mol Psychiatry 24:1248–1257 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Sokol DK, Chen D, Farlow MR, Dunn DW, Maloney B, Zimmer JA, Lahiri DK (2006) High levels of Alzheimer beta-amyloid precursor protein (APP) in children with severely autistic behavior and aggression. J Child Neurol 21:444–449 [ DOI ] [ PubMed ] [ Google Scholar ] Südhof TC (2018) Towards an understanding of synapse formation. Neuron 100:276–293 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Tabatadze N, Huang G, May RM, Jain A, Woolley CS (2015) Sex differences in molecular signaling at inhibitory synapses in the hippocampus. J Neurosci 35:11252–11265 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Takeuchi A, Hamasaki T, Litwack ED, O’Leary DD (2007) Novel IgCAM, MDGA1, expressed in unique cortical area- and layer-specific patterns and transiently by distinct forebrain populations of Cajal-Retzius neurons. Cereb Cortex 17:1531–1541 [ DOI ] [ PubMed ] [ Google Scholar ] Takeuchi A, O’Leary DD (2006) Radial migration of superficial layer cortical neurons controlled by novel Ig cell adhesion molecule MDGA1. J Neurosci 26:4460–4464 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Toledo A, Letellier M, Bimbi G, Tessier B, Daburon S, Favereaux A, Chamma I, Vennekens K, Vanderlinden J, Sainlos M et al (2022) MDGAs are fast-diffusing molecules that delay excitatory synapse development by altering neuroligin behavior. eLife 11:e75233 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Torres VI, Vallejo D, Inestrosa NC (2017) Emerging synaptic molecules as candidates in the etiology of neurological disorders. Neural Plast 2017:8081758 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Tsai L, Stewart MA, August G (1981) Implication of sex differences in the familial transmission of infantile autism. J Autism Dev Disord 11:165–173 [ DOI ] [ PubMed ] [ Google Scholar ] Tunyasuvunakool K, Adler J, Wu Z, Green T, Zielinski M, Zidek A, Bridgland A, Cowie A, Meyer C, Laydon A et al (2021) Highly accurate protein structure prediction for the human proteome. Nature 596:590–596 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Turetsky BI, Calkins ME, Light GA, Olincy A, Radant AD, Swerdlow NR (2007) Neurophysiological endophenotypes of schizophrenia: the viability of selected candidate measures. Schizophr Bull 33:69–94 [ DOI ] [ PMC free article ] [ PubMed ] Um JW, Han KA, Choi SY, Ko J (2020) Protocol for quantitative analysis of synaptic vesicle clustering in axons of cultured neurons. STAR Protoc 1:100095 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Um JW, Ko J (2017) Neural glycosylphosphatidylinositol-anchored proteins in synaptic specification. Trends Cell Biol 27:931–945 [ DOI ] [ PubMed ] [ Google Scholar ] Vasudevan N, Pfaff DW (2007) Membrane-initiated actions of estrogens in neuroendocrinology: emerging principles. Endocr Rev 28:1–19 [ DOI ] [ PubMed ] [ Google Scholar ] Velazquez-Zamora DA, Garcia-Segura LM, Gonzalez-Burgos I (2012) Effects of selective estrogen receptor modulators on allocentric working memory performance and on dendritic spines in medial prefrontal cortex pyramidal neurons of ovariectomized rats. Horm Behav 61:512–517 [ DOI ] [ PubMed ] [ Google Scholar ] Velmeshev D, Schirmer L, Jung D, Haeussler M, Perez Y, Mayer S, Bhaduri A, Goyal N, Rowitch DH, Kriegstein AR (2019) Single-cell genomics identifies cell type-specific molecular changes in autism. Science 364:685–689. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Vissers L, Kalvakuri S, de Boer E, Geuer S, Oud M, van Outersterp I, Kwint M, Witmond M, Kersten S, Polla DL et al (2020) De novo variants in CNOT1, a central component of the CCR4-NOT complex involved in gene expression and RNA and protein stability, cause neurodevelopmental delay. Am J Hum Genet 107:164–172 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Wang L, Wang C, Moriano JA, Chen S, Zuo G, Cebrian-Silla A, Zhang S, Mukhtar T, Wang S, Song M et al (2025) Molecular and cellular dynamics of the developing human neocortex. Nature 647:160–178 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Wang X, Lin D, Jiang J, Liu Y, Dong X, Fan J, Gong L, Shen W, Zeng L, Xu T et al (2024) MDGA2 constrains glutamatergic inputs selectively onto CA1 pyramidal neurons to optimize neural circuits for plasticity, memory, and social behavior. Neurosci Bull 40:887–904 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Werling DM (2016) The role of sex-differential biology in risk for autism spectrum disorder. Biol Sex Differ 7:58 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Willsey AJ, Morris MT, Wang S, Willsey HR, Sun N, Teerikorpi N, Baum TB, Cagney G, Bender KJ, Desai TA et al (2018) The psychiatric cell map initiative: a convergent systems biological approach to illuminating key molecular pathways in neuropsychiatric disorders. Cell 174:505–520. [DATASET] [Supplementary Table 1] [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Willsey HR, Willsey AJ, Wang B, State MW (2022) Genomics, convergent neuroscience and progress in understanding autism spectrum disorder. Nat Rev Neurosci 23:323–341 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Wolf FA, Angerer P, Theis FJ (2018) SCANPY: large-scale single-cell gene expression data analysis. Genome Biol 19:15 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Zafar E, Maqbool MF, Iqbal A, Maryam A, Shakir HA, Irfan M, Khan M, Li Y, Ma T (2022) A comprehensive review on anticancer mechanism of bazedoxifene. Biotechnol Appl Biochem 69:767–782 [ DOI ] [ PubMed ] [ Google Scholar ] Zhao H, Mao X, Zhu C, Zou X, Peng F, Yang W, Li B, Li G, Ge T, Cui R (2021) GABAergic system dysfunction in autism spectrum disorders. Front Cell Dev Biol 9:781327 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Zuber B, Nikonenko I, Klauser P, Muller D, Dubochet J (2005) The mammalian central nervous synaptic cleft contains a high density of periodically organized complexes. Proc Natl Acad Sci USA 102:19192–19197 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Associated Data This section collects any data citations, data availability statements, or supplementary materials included in this article. Supplementary Materials Table EV1 (75.5KB, doc) Table EV2 (37KB, doc) Table EV3 (56KB, doc) Appendix (5MB, pdf) Peer Review File (1.7MB, pdf) Dataset EV1 (65.9KB, xlsx) Dataset EV2 (15.3KB, xlsx) Dataset EV3 (3.8MB, xlsx) Dataset EV4 (3.1MB, xlsx) Dataset EV5 (2.1MB, xlsx) Dataset EV6 (2.4MB, xlsx) Dataset EV7 (200.2KB, xlsx) Dataset EV8 (312.8KB, xlsx) Dataset EV9 (226.2KB, xlsx) Source data Fig. 1 (62.9MB, zip) Source data Fig. 2 (83.2MB, zip) Source data Fig. 3 (7.2MB, zip) Source data Fig. 4 (6.6MB, zip) Source data Fig. 5 (74.1MB, zip) Source data Fig. 7 (14MB, zip) Expanded View Figures (7.4MB, pdf) Data Availability Statement The raw MS data files for total proteomics have been deposited to the MassIVE repository with identifier PXD057134 ( ftp://massive-ftp.ucsd.edu/v07/MSV000096187/ ) for data from Mdga1 -cKO mice and identifier PXD057136 ( ftp://massive-ftp.ucsd.edu/v07/MSV000096188/ ) for data from Mdga1 Y636C/E751Q KI mice. The source data of this paper are collected in the following database record: biostudies:S-SCDT-10_1038-S44321-026-00402-y . Articles from EMBO Molecular Medicine are provided here courtesy of Nature Publishing Group ACTIONS View on publisher site PDF (16.1 MB) Cite Collections Permalink PERMALINK Copy RESOURCES Similar articles Cited by other articles Links to NCBI Databases Cite Copy Download .nbib .nbib Format: AMA APA MLA NLM Add to Collections Create a new collection Add to an existing collection Name your collection * Choose a collection Unable to load your collection due to an error Please try again Add Cancel Follow NCBI NCBI on X (formerly known as Twitter) NCBI on Facebook NCBI on LinkedIn NCBI on GitHub NCBI RSS feed Connect with NLM NLM on X (formerly known as Twitter) NLM on Facebook NLM on YouTube National Library of Medicine 8600 Rockville Pike Bethesda, MD 20894 Web Policies FOIA HHS Vulnerability Disclosure Help Accessibility Careers NLM NIH HHS USA.gov Back to Top