Abstract
Disturbances in GABAergic neurodevelopment are thought to underlie cortical network dysfunction in neurodevelopmental disorders (NDDs) such as autism and schizophrenia. The diversity of GABAergic neurons is shaped by cortical cues during prolonged postmitotic differentiation, yet how pathological environments associated with NDDs influence this process remains poorly understood. Oxysterols, oxidized cholesterol metabolites or its precursors, modulate key developmental signaling pathways. Here, using an iPSC-based model of human forebrain GABAergic neuron differentiation combined with single-cell transcriptomics, we show that 24S,25-epoxycholesterol, an oxysterol enriched in the fetal brain and dysregulated in NDDs, enhances neurogenesis while altering the distribution of GABAergic neuronal subtypes. Pharmacological and genetic perturbations further identify the liver X receptor as a critical mediator of these effects. Together, these findings uncover a link between cholesterol metabolism and GABAergic fate specification, suggesting that aberrant oxysterol signaling contributes to the pathogenesis of NDDs.
Subject terms: Stem cells, Molecular neuroscience
Introduction
Gamma-aminobutyric acid (GABA)-releasing inhibitory neurons are a diverse population essential for maintaining neural circuit balance and regulating complex behaviors [1, 2]. Dysfunction of these neurons produces excitatory-inhibitory imbalance, a hallmark of neurodevelopmental disorders (NDDs) including schizophrenia, autism spectrum disorders, and epilepsy [1, 3–5], and is also implicated in Alzheimer’s disease–associated network and cognitive deficits [6]. Genomic studies show that NDD risk variants are enriched in GABAergic neuron transcripts and regulatory regions, with effects detectable as early as fetal development [7–11], pointing to a prenatal origin of disease vulnerability.
24S,25-epoxycholesterol (24S,25-EC) is the most prevalent oxysterol in the fetal brain and plays a critical role in midbrain dopamine neurogenesis [12–17]. We previously showed that 24S,25-EC production is altered in autism iPSC-derived cortical cells, resulting in either delayed or precocious cortical neurogenesis depending on the direction of the oxysterol change [18]. Forebrain GABAergic interneurons arise from the medial ganglionic eminence (MGE) and diversify into distinct subtypes in cortical and striatal environments, a process shaped by local cues. Moreover, 24S,25-EC can also modulate hedgehog signaling via binding to its signal transducer Smoothened [19]. These observations raise the possibility that 24S,25-EC regulates MGE induction and progenitor differentiation, and that dysregulated oxysterol metabolism may disrupt GABAergic subtype specification in NDDs.
Here, we used single-cell RNA sequencing to map iPSC-derived MGE neurons against human MGE populations, confirming high-fidelity generation of major GABAergic subclasses. We find that 24S,25-EC promotes neurogenesis via LXRβ signaling and alters the balance of GABAergic subtypes. Our results reveal a cholesterol-derived developmental cue that shapes inhibitory neuron diversity and suggest a metabolic mechanism contributing to NDD pathogenesis.
Methods
Human pluripotent stem cell (PSC) culture and neuronal differentiation
The PSC lines used in this study were iCas9 hESC line, iCas9-derived LXRβ knockout line [18], corrected KOLF2 (KOLF2.1 J) hiPSC line [20], and KOLF2-derived LHX6-emerald interneuron reporter line [21]. All hPSCs were cultured on Cultrex (R&D systems) coated plastics in E8 Flex media (ThermoFisher) in a humidified 37 °C incubator with 5% CO2. Media was changed every other day and cells passaged mechanically with 0.02% EDTA at 80% confluence. MGE differentiation followed procedures described previously [21, 22]. Briefly, cells from two 80% confluent wells of a 6-well plate were plated onto a 12-well plate previously coated with Cultrex in E8 media (Day 0) and changed to N2B27 the next day. Cells were induced to neuroectoderm fate by LDN-193189 (100 nM, Tocris), SB-431542 (10 µM, Tocris) and XAV-939 (2 µM, Stratech) from day 1–10 followed by ventralisation with a combination of recombinant Sonic Hedgehog (SHH, 200 ng/ml, Peprotech) and a small molecule SHH agonist purmorphamine (PM, 1 µM, Merck) from day 11–18. To promote terminal differentiation and cell survival, cultures were treated with BDNF (10 ng/ml, Peprotech) from day 25 until harvest. For oxysterol experiments, 24S,25-epoxycholesterol (0.5–1 μM, Abcam) was added to N2B27 media from day 15–25. For testing MGE induction effect of 24S,25-EC (d10-18 treatment), neural differentiation follows the same procedure without SHH and PM.
For lanosterol synthase inhibition experiment, MM0299 (9.1–36.4 nM, Target Mol Chemicals) or DMSO control was added to MGE progenitor cultures for three days from day 20. Cultures were terminated at day 25 for immunocytochemistry. For pharmacological interrogation of LXR signaling, ethanol control or small molecule LXR agonist GW3965 (0.5, 1 µM, Stratech) and/or antagonist GSK2033 (0.5, 1 µM, stratech) were added during day 15–25.
Immunocytochemistry
Cultured cells were fixed in 3.7% PFA for 15 min at 4 °C. For nuclear antigen detection, an additional fixation with methanol gradient was performed, which includes 5 mins each in 33 and 66% methanol at room temperature followed by 100% methanol for 20 min at −20 °C. Cells were returned to PBS containing 0.3% Triton-X-100 (PBS-T) via inverse gradient. Cells were then permeabilized with three 10 min washes in PBS-T and then blocked in PBS-T containing 3% BSA and 5% donkey serum. Cells were incubated with primary antibodies in blocking solution overnight at 4 °C. Following three PBS-T washes, Alexa-Fluor secondary antibodies were added at 1:1000 in PBS-T for 1 h at ambient temperature in the dark. Three PBS washes were then performed that included once with DAPI at 1:3000. Images were taken on a PerkinElmer OperaPhenix or a Leica DMI6000B inverted microscope. Quantification was performed using CellProfiler (cellprofiler.org) or manually in ImageJ (imagej.net). Five to ten randomly selected fields of view were analysed from each of three independent differentiations (biological replicates), with each differentiation performed in 2–3 technical replicates. Each counting field is presented as a dot in relevant bargraphs. The following antibodies were used for the immunofluorescence studies: rabbit anti-FOXG1 (Abcam AB18259, 1:250), NKX2.1 (Abcam AB76013, 1:1000), OLIG2 (R&D systems AF2418, 1:200), COUPTF2 (R&D systems PP-H7147-00, 1:100), anti-Ki67 (ACK02, 1:1000), secondary antibodies (Alexa Fluor 488, 594 and 647, 1:1000).
EdU Labelling and detection
EdU labelling and detection was performed using a Click-iT EdU Assay kit (Thermo Fisher). Cells were incubated with EdU for 2 h at specified day of differentiation, followed by fixation in 3.7% PFA for 15 min at 4 °C. EdU detection was carried out as per manufacturer’s protocol before antibody staining. Cell cycle exit index of NKX2.1+ MGE progenitors was calculated as the ratio of NKX2.1+EdU+Ki67− cells to the total EdU+ population. Image capture and quantification were performed as described in ‘Immunocytochemistry’.
Single cell RNA sequencing and data processing
For scRNA-seq, differentiating MGE cells were collected and dissociated into single cells at d25 and d40. Briefly, cells were treated with Accutase (ThermoFisher) for 5–7 min at 37 °C. The suspension was centrifuged at 200 g, 4 °C for 5 min to remove supernatant. Cells were then re-suspended in cold PBS with 0.5% BSA and passed through a 40μm Flowmi cell strainer. Three biological replicates of the same stage were pooled together. For quality control, cell viability was assessed by trypan blue staining and confirmed to be above 80%. Cells were loaded onto the 10X Chromium Single Cell Platform and libraries were constructed using the Single Cell 3’ Library v3.1 kit (10X Genomics) as per manufacturer’s protocol. The libraries were sequenced on the Illumina Novaseq 6000 instrument using Novaseq S2 100 cycle kit aiming for a depth of approximately 150,000 reads/cell. Sequencing was run as paired-end, 100 cycles in a standard workflow.
Raw reads were processed using Cell Ranger 7.1.0 (10x Genomics) and aligned to Homo sapiens GRCh38-2020-A. All downstream analysis was performed using the R package Seurat (v 5.2) [23]. For quality control, a Median Absolute Deviation (MAD) method was applied using the isOutlier function from the scuttle package (1.8.4) [24]. Cells with nFeature 2 MADs lower than the median, median nCounts above or lower than 2 MADs, and the fraction of mitochondrial genes above 2 MADs were discarded. The normalisation and variance stabilisation of molecular count data were performed using the SCTransform function in Seurat [25]. This function was used to also remove confounding sources of variation such as mitochondrial mapping percentage. The batch effect was removed using harmony (v1.1.0) [26]. Uniform manifold approximation and projection (UMAP) and unsupervised Louvain clustering were performed using RunUMAP, FindNeighbors and FindClusters function based on the first 30 principal components. Cell cycle phase scores were calculated using the CellCycleScoring function based on the expression of canonical G2/M and S phase markers [27]. For annotation, cluster-specific gene markers were found using the FindConservedMarkers function with min.pct=0.25, logfc.threshold=0.25, statistically significant markers p.adjust<0.05 were tested (hypergeometric test) for gene enrichment in specific makers of MGE branches previously reported in Mayer, et al. [28].
To compare the transcriptome similarity with published human ganglionic eminence scRNAseq data [29], we performed anchor-based label transfer implemented in Seurat. The anchors between the query and reference datasets were identified and filtered using the FindTransferAnchors() function with the top 50 dimensions. Then, the MapQuery() wrapper function was used to project our data onto human brain data to visualise cell type similarities between datasets. Differential gene expression analyses were performed using the FindAllMarkers and FindMarkers function with Wilcoxon signed-rank test and Bonferroni correction for total number of genes included in the analysis. Proportion change analysis was performed using the scProportionTest 0.0.0.9 package [30] with 1000 permutations followed by FDR correction. Gene ontology (GO) enrichment was performed using the enrichGO function in the clusterProfiler 4.12.6 package [31] based on Homo sapiens GO term database retrieved by the org.Hs.eg.db 3.19.1 package. Synapse-focused GO enrichment analysis was performed using non-redundant terms of the SynGO database [32].
Gene set over-representation tests were performed using the enricher function in the clusterProfiler package. LXRα and LXRβ targets were obtained from TFLink gateway (https://tflink.net), including only TF-target interactions from Chromatin immunoprecipitation and/or electrophoretic mobility shift assay experiments. SCENIC analysis was performed using pySCENIC 0.12.1 [33, 34]. cisTarget databases used for SCENIC analysis were the Homo sapiens - hg38 – refseq_r80_v10- SCENIC+ databases downloaded from https://resources.aertslab.org/cistarget/. SCENIC regulon enrichment score (AUC) was used for differential testing using the Wilcoxon signed-rank test with Bonferroni correction for the total number of regulons detected. Gene networks were visualised and analysed with Analyze Network in Cytoscape 3.10.2 [35].
Gene enrichment analysis
Public summary statistics associated to the latest schizophrenia and autism spectrum disorder GWAS were downloaded from https://www.med.unc.edu/pgc/download-results/scz/ and OpenGWAS (dataset ieu-a-1185), respectively [36, 37]. Summary statistics were QC and cleaned using R package MungeSumstats [38]. SNPs were m apped to genes with annotation windows of 35 kb and 10 kb upstream and downstream, respectively. Dupplicated rsIDs for the same SNP were excluded from analysis. The Seurat object containing the expression dataset was converted into a CellTypeDataset object using the expression weighted cell type enrichment (EWCE) R package, as reported by Skene et al. [9]. Main cell type association analysis was carried out using the MAGMA.Celltyping running MAGMA v1. 10 [9, 36] using a combination of linear and the top 10% most specific genes for each cell type. DEGs between treated and untreated samples were used for MAGMA enrichment analysis to determine the association neuropsychiatric traits. Conditional MAGMA analysis were carried out to account for enrichment driven by brain-expressed genes using a covariate gene set obtained from https://www.syngoportal.org.
Statistical analysis
RNA sequencing data were analyzed separately as described above. The SPSS Statistics 27 software (IBM) was used for all other statistical analyses. All quantified data were plotted in Prism 10 (GraphPad Software) and are reported as the mean ± SEM from three independent experiments, each with 2–3 technical replicates. Data were tested for normality using the Shapiro-Wilk test and for equal variance with Levene’s test before performing statistical analyses by one-way ANOVA or two-way ANOVA (depending on the number of dependent variables) with post-hoc Bonferroni for equal variances or Dunnett’s T3 test for multiple comparisons where equal variances criteria were not met. The Kruskal-Wallis test with post-hoc Dunn’s test for pairwise comparison was used when the parametric test was not suitable. The results were considered statistically significant at p < 0.05.
Results
24S,25-EC promotes LHX6+ cell production from hiPSCs without affecting ventral patterning
Given its prevalence in the fetal brain and its high-affinity binding to smoothened in vitro [19], we first tested the ability of 24S,25-EC to ventralize forebrain neural progenitors. KOLF2 iPSCs were first induced into forebrain neural progenitors without patterning cues [39], followed by exposure to 1μM 24S,25-EC between days 10–18 (d10-18, Figure S1A). A published MGE induction protocol was used as the positive control [21, 22], where cultures were treated with sonic hedgehog (SHH) and smoothened agonist, purmorphamine (PM), in the same time window. Efficient forebrain neural induction was achieved in all conditions, as demonstrated by the highly enriched presence of FOXG1 on d20 (Figure S1B-C). As expected, most cells in SHH-treated positive control cultures expressed the MGE progenitor marker NKX2.1 and OLIG2 (Figure S1D-E), and a small population of cells expressed a CGE marker COUPTF2 (Figure S1F), while few were found in the basal control cultures. However, 24S,25-EC treatment did not lead to an increase in NKX2.1+ cells over the basal condition, suggesting a lack of smoothened agonist activity.
We next investigated whether 24S,25-EC could affect neuronal production from SHH patterned MGE progenitors. To this end, 24S,25-EC was added at a later stage between d15-25 of differentiation (Fig. 1A). For ease of analysis, this experiment utilized an LHX6-mEmerald reporter human iPSC line that faithfully express the genetic knock-in mEmerald in LHX6-expressing neurons [21, 22]. LHX6 marks migrating MGE-derived cortical interneurons and some GABAergic projection neurons in vivo. We observed a gradual increase of mEmerald+ cells from d25 till d50 in both the basal control cultures and those treated with 0.5μM and 1μM 24S,25-EC (Fig. 1B, C). However, there were significantly more mEmerald+ cells in 24S,25-EC treated cultures than in the controls at d25, d30 and d40, although the control cultures caught up by d45. In contrast to the increase in mEmerald+ cells, protein markers expressed by dividing MGE progenitors, such as NKX2.1, OLIG2 and FOXG1, decreased in cultures exposed to 1μM 24S,25-EC compared to the controls at d25 (Fig. 1D, E). We observed no differences in the expression of COUPTF2, a marker of dorsal CGE and LGE.
Fig. 1. 24S,25-EC promotes neuronal production from NKX2.1+ progenitors.

A Outline of the neural differentiation paradigm and 24S,25-EC treatment regime. B Endogenous mEmerald signal from the LHX6-mEmerald cultures treated with vehicle or EC at d25, d30 and d45. C Quantification of mEmerald+ cells from d25-d55. D Top and middle rows, triple immunostaining for NKX2.1(green), OLIG2 (red), and COUPTF2 (magenta) and single staining for FOXG1 (green) at d25 of differentiation. E Quantification of NKX2.1+, FOXG1+, OLIG2+ and COUPTF2+ cells shown in D. F Double immunostaining for NKX2.1 (green) and Ki67 (light blue) combined with EdU labelling (red). Arrow heads indicate NKX2.1+EdU+Ki67- cells. G Quantification of cycling NKX2.1+ Ki67+ progenitors. H Quantification of cell cycle exit index. Data presented are mean ± s.e.m. of three independent differentiation runs each with 2–3 technical replicates. Each dot in the graphs represents an image counted. The numbers of technical replicates and images counted in C, E, G, H are provided in table S1. EC treated vs vehicle control groups were compared by two-way ANOVA with post-hoc Bonferroni in C. One-way ANOVA with post-hoc Dunnett T3 or non-parametric Kruskal-Wallis with post-hoc Dunn in E, G and H (*p < 0.05; **p < 0.01; ***p < 0.001). Nuclei were counterstained with DAPI (blue) or Hoechst for live imaging in B. Scale bars: 100 µm.
To investigate the mechanism of 24S,25-EC action, we exposed d25 cultures to a 2 h EdU pulse followed by triple immunostaining for EdU, Ki67, and NKX2.1 the next day (Fig. 1F–H). This approach enabled quantification of cycling progenitors (NKX2.1⁺Ki67⁺; Fig. 1G) and calculation of the cell cycle exit index, defined as the ratio of EdU-labeled progenitors that had exited the cell cycle (NKX2.1⁺EdU⁺Ki67⁻) to the total EdU⁺ population. Although no significant difference was observed in the number of cycling NKX2.1⁺ MGE progenitors between conditions (Fig. 1G), the cell cycle exit index was significantly higher in 24S,25-EC-treated cultures compared to controls (Fig. 1H). These findings indicate that 24S,25-EC promotes cell cycle exit in MGE neural progenitors.
To investigate whether the neurogenesis-promoting effect of the exogenous 24S,25-EC reflects its physiological function, we exposed d20 MGE differentiation to MM0299, a lanosterol synthase inhibitor that stimulates endogenous 24S,25-EC production [40]. Following three days of MM0299 treatment, immunostaining for two pan-neuronal markers, βTubulin III and MAP2, revealed a dose-dependent increase in neuronal number compared to vehicle controls (Figure S2). These results indicate that the pro-neurogenic effect of 24S,25-EC is not an artifact of exogenous application, but rather reflects an endogenous regulatory mechanism governing MGE neuron differentiation, as it has been demonstrated in cortical cells [18].
iPSC-derived neurons exhibit high MGE specificity and reproduce the three principal developmental branches of the human fetal MGE
To further characterize the iPSC-derived neurons and assess the impact of 24S,25-EC, we performed scRNAseq at two developmental stages (d25 and d40) on vehicle-treated controls and 24S,25-EC-treated cultures, following the previously described treatment protocol. After quality control and batch effect correction using Harmony (Figure S3A-B), we retained 114,787 high-quality cells, encompassing 18,579 protein-coding genes (58,066 control and 56,721 oxysterol-treated; Fig. 2A). Unsupervised Louvain clustering identified 11 cell clusters, each comprising varying proportions of d25 and d40 samples (Fig. 2B, C, Figure S3C-D). In both control and 24S,25-EC-treated cultures, d25 samples contained a higher proportion of S- or G2/M- phase cells than G1-phase cells, whereas d40 samples were predominantly G1-phase cells, as predicted by the CellCycleScoring function in Seurat (Fig. 2D, E; Figure S3E–F), consistent with the temporal progression of neuronal differentiation observed in the LHX6 reporter assay.
Fig. 2. Single cell RNAseq confirms highly specific MGE fate induction and recapitulation of major MGE neuronal classes.

A UMAP embedding of scRNA-seq data of 58,066 control and 56,721 EC-treated iPSC-derived MGE cells at d25 and d40. B UMAP of control cells only at d25 and d40. C Stack plot depicting quantification of d25 and d40 in each cell cluster of the control samples. D UMAP plot of control cells color-coded by predicted cell cycle phases. E The proportion of control cells in each cell cycle phase at d25 and d40. F Feature plots showing the expression levels of brain-region-specific genes. G Heatmap showing classic progenitor- and neuron-expressed genes in each cluster group. H Annotation of cell types visualized by UMAP. I Violin plot showing differential expression of MGE neuron genes in each neuronal cluster group. J Sankey diagram showing alignment of iPSC-derived MGE neurons to human whole ganglionic eminence cells at 9–18 GW [29].
In agreement with the high proportions of FOXG1+ and NKX2.1+ cells in d25 cultures (Fig. 1D, E), FOXG1 and NKX2.1 transcripts were detected in all cell clusters, with higher levels concentrated in cycling cells (Fig. 2D, F). Abundant expression of other known MGE-specific genes was also evident, including SOX6, LHX8, MEF2C, NNAT, CALM1, and the pan-GABAergic marker GAD1 (Fig. 2F). NNAT was preferentially detected in S or G2/M cells, while the others were enriched in G1 dominated cell clusters, consistent with MGE neuronal gene markers (Fig. 2D, F). In contrast, transcripts restricted to or preferentially expressed in the dorsal telencephalon (cortex, DMRTA1, EMX1, PAX6, and TBR1) or LGE (FOXP1, FOXP2, ZNF503, and SALL3) were detected in a few cells and/or at low levels (Fig. 2F). Moreover, there was no evident expression of markers indicative of other brain regions, such as the diencephalon (IRX1, EBF3, SOX14, and SLC18A2), midbrain, and hindbrain (FOXA2, PITX3, EN2, and GBX2). These findings confirm the strong MGE specificity of the iPSC-derived neural populations.
Based on cell cycle scoring and the expression of canonical progenitor and neuronal marker genes, clusters 7, 2, and 6 were designated as progenitors, late progenitors, and nascent neurons, respectively (Fig. 2G). In contrast, clusters 0, 1, 3–5, and 8–10 were classified as neurons due to their preferential expression of neuronal genes (Fig. 2G).
We further characterized the neuronal clusters using a hypergeometric test to assess the enrichment for marker genes of the three major classes of human fetal MGE progenies defined by Mayer et al. [28] in differentially expressed genes of each cluster (Figure S3G). Cluster 1 showed strong similarity to branch 1 fetal MGE cells, which display transcriptional features of cortical interneurons expressing PLS3, DLX2, ARX, EPHA5, NXPH1, DCLK2, and SST (Fig. 2H, I; Table S1). Accordingly, this cluster was designated as B1 SST-CCK cortical interneurons. Clusters 4, 5, and 9 corresponded to branch 2 GABAergic projection neurons, differentially expressed RALYL, PCDH7, NRP1, ZFHX3, LRP1B, RELN, and TLE4; notably, cluster 4 (branch 2 group 1) also exhibited a high differential expression of EBF1 (Fig. 2H, I; Table S1). Clusters 0, 3, and 8 aligned with branch 3 neurons, characterized by enriched expression of SHISA9, CNTNAP2, ADARB2, ENC1, and LHX8 relative to branches 1 and 2. Descendants of this branch include most cholinergic projection neurons, cholinergic striatal interneurons, and GABAergic globus pallidus projection neurons [41, 42]. Finally, cluster 10 expressed genes from all branches along with elevated levels of NKX2.1 and SFTA3, and was therefore classified as unspecified neurons (Fig. 2G, I; Table S1).
To confirm the transcriptomic fidelity of iPSC-derived neurons relative to human MGE populations, we mapped our data to a published single-cell RNA-seq dataset from human fetal whole ganglionic eminence using Seurat’s FindTransferAnchors followed by MapQuery functions [29, 43]. This analysis showed that the majority of iPSC-derived clusters (0, 1, 3–5, and 8–10) mapped to the fetal MGE neuronal component of the human dataset (Fig. 2J, Figure S3I). Only a small fraction of cells in cluster 5 (1.27%) and cluster 9 (0.85%), together accounting for 0.088% of all cells, were predicted as thalamic neurons (Fig. 2J), which arise primarily from the diencephalon during development [44].
Clusters 2 and 7 mapped predominantly to human fetal neural progenitors, with a smaller fraction aligning to fetal MGE neurons, whereas cluster 6 mapped mainly to fetal MGE neurons with a minor contribution from fetal progenitors. Importantly, very few cells were assigned to other neural cell types present in the reference dataset, including LGE neurons (0.01% of total cells), oligodendrocyte progenitor cells (OPC, 0.02%), and cortical excitatory neurons (0.0008%) (Fig. 2J, Figure S3I).
These results demonstrate that iPSC-derived cell populations are transcriptionally well aligned with human fetal MGE lineages. Collectively, these observations support the utility of this iPSC system for modeling neuronal fate specification of human MGE neurons.
Schizophrenia disease genes are enriched across iPSC-derived MGE neuron populations and 24S,25-EC regulated gene sets
Schizophrenia risk alleles are over-represented in human MGE-derived neurons [7–10]. To assess the disease relevance of our iPSC-derived MGE model, we used MAGMA.Celltyping to test enrichment of schizophrenia-associated variants among transcripts expressed in iPSC-derived cell populations, using the latest schizophrenia GWAS dataset [44]. All neuronal populations showed significant enrichment (FDR padj < 0.05; Fig. 3A), whereas neural progenitors did not, consistent with findings in the human brain. To confirm cell-type specificity, we applied EWCE to the top 10% of specifically expressed genes in each group; all neuronal populations except B3 projection/interneuron group 2 remained significantly enriched (Table S2).
Fig. 3. Cell-type-specific enrichment of schizophrenia GWAS signals in MGE-derived neuronal populations.

A Bar plot depicting the -log₁₀(FDR-adjusted p-values) for MAGMA cell-type enrichment tests across the annotated cell types. Two modes of enrichment analysis were tested: linear (red) using all expressed genes, and top 10% (blue) using only the top 10% most specific genes expressed in each cell type. B Bar plot depicting the -log₁₀(FDR-adjusted p-values) for MAGMA gene test for enrichment of ASD (red) and schizophrenic (blue) risk genes in 24S,25-EC-dependent DEGs across all cell types. Tests were performed with or without brain-expressed genes as covariate. Vertical dashed line marks the significance threshold (FDR = 0.05).
We next asked whether NDD risk genes were preferentially altered by 24S,25-EC treatment. MAGMA gene-set analysis of oxysterol-dependent differentially expressed genes (DEGs; FDR padj < 0.05), combining both up- and down-regulated at day 25 and day 40 within each cluster, revealed significant enrichment of schizophrenia risk variants in most neuronal populations (Fig. 3B), with enrichment of autism spectrum disorder risk genes observed in nascent neurons and B2 group 2 neurons. Notably, progenitors, late progenitors, and nascent neurons, cell types that did not show enrichment in their global transcriptomes, displayed significant enrichment among oxysterol-responsive DEGs.
Conditional MAGMA analyses controlling for broadly brain-expressed genes yielded similar results, albeit with modestly increased p-values. Together, these findings indicate that 24S,25-EC-responsive gene programs disproportionately overlap with schizophrenia risk architecture and support the use of iPSC-derived MGE neurons as a relevant platform for modelling neurodevelopmental disorders.
24S,25-EC exhibits a differential effect on neuronal subclasses
To gain insight into the effects of 24S,25-EC on different cell populations, we compared the proportion of each cell cluster between the control and oxysterol-treated conditions. Using the whole dataset containing both d25 and d40 cells, we observed a significant reduction in cell number in the 24S,25-EC treated progenitor cluster (−0.58>Log2 fold change>0.58, FDR padj < 0.05, Fig. 4A), providing independent support for 24S,25-EC-induced cell cycle exit as measured by EdU labelling (Fig. 1F–H). Intriguingly, we detected a significant reduction in unspecified neurons and an increase in two branch 3 neuronal clusters (B3-projection/interneurons 1 and 2, Fig. 4A). By performing the comparison by date separately, we found that the reduction in unspecified neurons was already present at d25, whereas an increase in branch 3 neurons occurred later at d40 (Figure S3H). We also observed a reduction in the number of B1-cortical interneurons and an increase in the number of nascent neurons at d40. No significant changes in B2 neuronal clusters were detected by any means of comparison, suggesting a preferential effect of 24S,25-EC on Branches 1 and 3 neuronal clusters.
Fig. 4. Preferential effects of 24S,25-EC on neuronal groups.

A 24S,25-EC-dependent fold changes in the proportion of each cluster group. Significance threshold: −0.58>Log2 fold change>0.58, FDR < 0.05. B Heatmap and bar graph showing the number of DEGs of different categories by each cell cluster. C Pie chart showing GO terms of d25 DEGs of progenitor and later progenitor populations. D Heatmap showing enrichment of LXRβ target genes in DEGs of different categories and each cell cluster.
To explore the potential mechanisms by which 24S,25-EC alter cellular composition, we analyzed oxysterol-dependent DEGs within each cell cluster considering the direction of changes at d25 and d40 in tandem. Most genes were up- or downregulated in 24S,25-EC-treated samples at only one time point. These DEGs were referred to as d25-High/d25-Low and d40-High/d40-Low, respectively (Fig. 4B, Table S3). A small proportion of DEGs showed the same or opposite direction of changes at d25 and d40. These DEGs are referred to as Continuous High/Low and d25-High (or low) reverse, respectively (Table S3). Considering that oxysterol treatment ends at d25, we hypothesized that the d25 DEGs are likely to be enriched with direct targets and downstream pathways of oxysterol signaling, whereas the d40 DEGs may reflect secondary or long-term changes in the associated cell populations.
At d25, the progenitor and late progenitor populations responded to 24S,25-EC with a greater number of DEGs compared to most of the neuronal clusters, as both ‘High’ and ‘Low’ (Fig. 4B). B3 projection/interneuron 3 and B1 SST-CCK cortical interneurons also showed many d25 DEGs (Fig. 4B). In contrast, d40 DEGs were mostly found in nascent neurons, late progenitors, and neuronal populations (Fig. 4B). These observations support the notion that 24S,25-EC primarily acts on progenitor cells and disproportionately affects B1 SST-CCK cortical interneurons and B3 projections/interneurons among neuronal types.
However, d25-High DEGs displayed distinct GO enrichment profiles between progenitors and neurons. In progenitor and late progenitor clusters, these genes were predominantly enriched for processes related to cell cycle regulation, transcriptional control, and TGF-β and WNT signaling (Fig. 4C). In contrast, neuronal d25-High DEGs were enriched for pathways associated with neuronal maturation and function, including axonogenesis, synapse assembly, and intracellular transport (Table S4).
To examine the impact of oxysterol treatment on synaptic biology, we performed a synapse-focused gene ontology (SynGO) enrichment analysis within neuronal clusters (Figure S4A; Table S4 and Table S5) [32]. This analysis revealed significant upregulation at d40 of gene sets associated with active-zone architecture and presynaptic actin cytoskeleton organization in B1 SST-CCK cortical interneurons. Conversely, genes involved in synaptic translation were downregulated at day 25 in B3 projection/interneuron group 2 neurons and at day 40 in B1 SST-CCK cortical interneurons and B2 GABAergic projection neuron group 1 neurons.
Together, these results suggest that 24S,25-EC promotes presynaptic stabilization and functional maturation of neuronal populations.
24S,25-EC promotes MGE progenitor neurogenesis via LXR
Interestingly, we observed a significant over-representation of LXRβ target genes among d25-High DEGs in progenitors, nascent neurons, B3 projection/interneurons 3, and B1 SST-CCK cortical interneurons (Fig. 4D, Table S3), although B2 GABAergic neuron 2 also showed enrichment for LXRα target genes (Figure S4B, https://tflink.net/)45. These findings suggest a role for LXRβ in mediating oxysterol function and support the notion that LXRβ is the dominant LXR isoform expressed in the brain [46]. Furthermore, consistent with the EC-induced increase in neural progenitor cell cycle exit (Fig. 1H), the LXRβ targets identified in progenitor d25-High DEGs were primarily associated with cell cycle and transcription and regulation (Figure S4C).
Moreover, SCENIC gene regulatory networks analysis identified d25-High transcription factors as having the highest betweenness centrality measures within the cell type-specific transcriptomic networks (Figure S4D), suggesting that they function as key “hub” genes in oxysterol-dependent regulatory mechanisms in the responding progenitor cells. Notably, many of these transcription factors, such as E2F3, HES1, TCF3/12, SMAD3, REST and YY1, are well-established regulators of cell cycle and neurogenesis, and are also known direct targets of LXRβ (E2F3, HES1, TCF3/12, SMAD3) or its co-receptor RXR (REST, YY1) [45,47–53].
The above analysis highlights LXR as the likely signaling receptor acting in MGE progenitors that mediates the 24S,25-EC-regulation of GABAergic neurogenesis. To provide experimental support, we pharmacologically activated LXR using the selective LXR agonist, GW3965. When applied to iPSC-MGE progenitors during d15-d25, GW3965 mimicked the effects of 24S,25-EC, as demonstrated by an increase in the number of LHX6-mEmerald+ neurons (Fig. 5A–D). Moreover, the EdU incorporation assay showed that GW3965 treatment resulted in a reduction in cycling NKX2.1+ MGE progenitors (NKX2.1+Ki67+) and, similar to 24S,25-EC treatment, an increase in cell cycle exit index (Fig. 5E). In contrast, the addition of the LXR antagonist GSK2033 abolished the neurogenic promoting effect of 24S,25-EC by blocking cell cycle exit (Figure S5).
Fig. 5. 24S,25-EC promote MGE progenitor neurogenesis via LXR.

A Outline of the neural differentiation paradigm. B Bar graph showing % of NKX2.1+ Ki67+ cells in d25 cultures treated with vehicle and LXR agonist GW3965 (GW). C Endogenous mEmerald signal from the LHX6-mEmerald cultures treated with vehicle or GW at d25, d30 and d35. D Line plot showing the % of mEmerald+ neurons at d25, d30 and d35. E Dot plot showing cell cycle exit index of control and GW-treated cultures. F Double immunostaining for NKX2.1 (green) and Ki67 (light blue) combined with EdU labelling (red.) Arrow heads indicate NKX2.1+EdU+Ki67- cells. G–H Dot plot showing the portion of NKX2.1+EdU+ and NKX2.1+Ki67+ progenitors, respectively, in isogenic iCas9 control and LXRβ knockout cultures in responding to EC treatment. I Dot plot showing the quantification of cell cycle exit index in wildtype and LXRβ knockout cultures in response to EC treatment. Data presented are mean ± s.e.m. of three independent differentiation runs, each with 2–3 technical replicates. Each dot in the graphs represents an image counted. The numbers of technical replicates and images counted in B, D, E, G–I are provided in table S4. EC treated vs vehicle control groups were compared by Kruskal-Wallis with post-hoc Dunn in B. Two-way ANOVA with post-hoc Bonferroni in D. One-way ANOVA with post-hoc Bonferroni (E,I) or Dunnett T3 when equality of variances criteria was not met in G, H. (*p < 0.05; **p < 0.01; ***p < 0.001). Nuclei were counterstained with DAPI (blue) or Hoechst for live imaging in C. Scale bars: 100 µm.
LXRβ is the predominant LXR receptor in the brain [46]. To exclude potential non-specific effects of the LXR agonist/antagonist, we re-examined 24S,25-EC on neural progenitor behavior using a CRISPR-engineered hESC line deficient in LXRβ and its isogenic wild-type control line [18]. Using the same experimental design as shown in Fig. 1A, we found that while 24S,25-EC treatment resulted in a decrease in S phase (NKX2.1+EdU+) and cycling (NKX2.1+Ki67+) MGE progenitors in the isogenic wild-type control culture, the same treatment had no effect on LXRβ-knockout MGE progenitors (Fig. 5F–I, Figure S6). Together, our pharmacological and genetic approaches demonstrate that 24S,25-EC promotes neural progenitor cell cycle exit via LXRβ receptor.
Discussion
Our study reveals a previously unrecognized role for oxysterol-LXR signaling in shaping cortical GABAergic neurogenesis, providing mechanistic insight into how cholesterol metabolism regulates inhibitory neuron heterogeneity during development and how its dysregulation may alter their subtype composition in NDDs.
Oxysterols, as by-products of cholesterol metabolism, are well established as modulators of signaling pathways critical for neural development [19, 54]. Altered levels of brain-derived oxysterol in the plasma or cerebrospinal fluid have been reported in individuals with neurodegenerative diseases [55–58] and neurodevelopmental disorders [55–61], and we previously identified elevated 24S,25-EC in autism iPSC-derived cortical cells [18]. However, the functional significance of these alterations has remained unclear. By demonstrating selective effects of 24S,25-EC on late-born GABAergic subtypes, our findings suggest that oxysterol imbalance can bias progenitor fate decisions and disrupt the developmental diversification of inhibitory neurons, a process essential for establishing balanced cortical circuitry.
GABAergic neurons constitute an exceptionally diverse population, with specialized subtypes distributed across cortical layers, the striatum, and the hypothalamus [62]. Neuron birthdate is a key determinant of GABAergic subtype identity [63–65], and interneurons generated at distinct developmental stages establish different patterns of connectivity with cortical projection neurons [28, 29, 66]. Despite this, the extrinsic signals linking progenitor timing to subtype specification and circuit integration have remained poorly defined. Our data provide proof-of-principle that environmental changes in the developing cortex, such as altered 24S,25-EC levels, can influence interneuron subtype trajectories and thereby reshape the excitatory-inhibitory architecture implicated in NDDs. Supporting this notion, and consistent with our findings of altered cortical 24S, 25-EC in 15q11.2 CNV iPSC-derived cortical cells, mouse models carrying loss of function of CYFIP1, the putative causal gene within the 15q11.2 loci, exhibit cortical synaptic dysfunction and excitation/inhibition imbalance [67].
Impaired synaptic function is a hallmark of many neurodevelopmental disorders, and our SynGO enrichment analyses point to a potential role for 24S,25-EC in synaptic maturation and plasticity. The presynaptic active zone is essential for vesicle docking, priming, and neurotransmitter release, while local protein synthesis at synapses supports rapid structural and functional remodeling. Accordingly, the 24S,25-EC-induced upregulation of gene sets associated with active zone architecture and presynaptic organization is consistent with enhanced presynaptic maturation. Conversely, the downregulation of genes involved in synaptic protein synthesis may reflect a shift toward structural consolidation and a more stable synaptic state.
Unlike in progenitor populations, LXR target genes were not enriched among synaptic DEGs, suggesting that the synaptic effects of 24S,25-EC are unlikely to arise through direct LXR-dependent transcriptional regulation. Instead, these changes may result from altered lipid metabolism or trafficking or emerge indirectly downstream of earlier LXR-mediated developmental processes. Supporting this interpretation, synaptic gene changes, most prominent at day 40, represented only a small fraction of total oxysterol-responsive neuronal DEGs. In contrast, the majority of DEGs were identified in progenitor populations at day 25 and were primarily associated with cell cycle regulation and neurogenesis. This temporal progression suggests that later transcriptional changes related to neuronal function are secondary to earlier epoxycholesterol-mediated effects on progenitor behavior and lineage specification.
Notably, our regulatory network analysis in the progenitor populations identified HES1, SMAD3, and TCF3/12 as hub transcription factors downstream of LXRβ, all of which integrate major developmental signalling pathways. These findings suggest that oxysterol-LXR signaling may intersect with WNT, TGFβ, and NOTCH networks to fine-tune the differentiation timing and interneuron subtype specification. Future studies will be required to experimentally dissect this crosstalk and determine the context-specific mechanisms through which LXR signaling shapes interneuron diversity.
Finally, our scRNA-seq analysis demonstrated strong transcriptomic fidelity between iPSC-derived and human MGE-derived neurons, validating iPSC-based GABAergic differentiation as a robust platform for modeling both neurodevelopmental and psychiatric disease mechanisms. Moreover, MGE-derived neurons from rodent or stem cell sources have been shown to functionally integrate into host circuits [68–71], highlighting the therapeutic potential of engineered interneurons. Our findings further suggest that manipulating extrinsic metabolic cues can direct progenitors toward specific GABAergic subtypes, providing a potential strategy to tailor neuronal diversity for both mechanistic studies and regenerative applications.
Conclusion
Our findings identify 24S,25-epoxycholesterol as a key extrinsic regulator of human MGE-derived GABAergic neurogenesis, acting through LXRβ to control progenitor cell-cycle exit and neuronal subtype specification. By linking cholesterol metabolism to inhibitory neuron diversification and neurodevelopmental disease-associated gene networks, this study provides a mechanistic framework connecting metabolic signaling with cortical circuit development. Beyond advancing our understanding of human interneuron biology, these results establish oxysterol-LXR signaling as a promising target for manipulating interneuron fate and offer a foundation for developing stem cell-based models and therapeutic strategies for neurodevelopmental and psychiatric disorders.
Supplementary information
List of supplemental materials and supplemental figure legend
Acknowledgements
We thank Drs. Francesco Bedogni, Kathryn Peall, and members of the ML laboratory for helpful discussions during this study. RNA sequencing analysis was performed using the computational facilities of the Advanced Research Computing@Cardiff (ARCCA) Division, Cardiff University.
Author contributions
MC-S and ML conceived of the study and designed the experiments. MC-S and EK carried out and analyzed the hESC and hiPSC experiments. MC-S, SD and NNV captured the scRNA-seq experiment samples, while SD made the libraries. MC-S, ZL and EK performed data analysis. DCF performed gene enrichment analysis. MF provided guidance on scRNA-seq data analysis to MC-S and EK. ML wrote the paper. All authors edited and approved the manuscript.
Funding
This research was funded by the U.K. Medical Research Council (grant number MR/L020807) and EU Horizon 2020 Program ‘NSC-reconstruct’ (grant number No. 874758) to ML EK is funded by a PhD Award from The Hodge Centre for Translational Neuroscience at Cardiff University.
Data availability
All singel-cell RNA sequencing data from this project were deposited at NCBI Gene Expression Omnibus (GEO) with the accession number GSE298622.
Competing interests
The author declares no competing interests.
Ethics
This study used established human pluripotent stem cell (hPSC) lines obtained from approved repositories. Under UK regulations, research using established human stem cell lines does not require Ethics Committee approval. All cell lines were used in accordance with institutional policies and applicable regulatory guidelines.
Supplementary information
The online version contains supplementary material available at https://doi.org/10.1038/s41398-026-04396-6.
Footnotes
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
These authors contributed equally: Ethan Kidd, Zongze Li.
References
- 1.Marín O. Interneuron dysfunction in psychiatric disorders. Nat Rev Neurosci. 2012;13:107–20. [DOI] [PubMed] [Google Scholar]
- 2.Le Magueresse C, Monyer H. GABAergic interneurons shape the functional maturation of the cortex. Neuron. 2013;77:388–405. [DOI] [PubMed] [Google Scholar]
- 3.Lunden JW, Durens M, Phillips AW, Nestor MW. Cortical interneuron function in autism spectrum condition. Pediatr Res. 2019;85:146–54. [DOI] [PubMed] [Google Scholar]
- 4.Rapanelli M, Frick LR, Pittenger C. The role of interneurons in autism and tourette syndrome. Trends Neurosci. 2017;40:397–407. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Filice F, Janickova L, Henzi T, Bilella A, Schwaller B. The parvalbumin hypothesis of autism spectrum disorder. Front Cell Neurosci. 2020;14:577525. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Verret L, Mann EO, Hang GB, Barth AMI, Cobos I, Ho K, et al. Inhibitory interneuron deficit links altered network activity and cognitive dysfunction in Alzheimer model. Cell. 2012;149:708–21. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Cameron D, Mi D, Vinh N-N, Webber C, Li M, Marín O, et al. Single-Nuclei RNA sequencing of 5 regions of the human prenatal brain implicates developing neuron populations in genetic risk for schizophrenia. Biol Psychiatry. 2023;93:157–66. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Cameron D, Vinh N-N, Prapaiwongs P, Perry EA, Walters JTR, Li M, et al. Genetic implication of prenatal gabaergic and cholinergic neuron development in susceptibility to schizophrenia. Schizophr Bull. 2024;50:1171–84. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Skene NG, Bryois J, Bakken TE, Breen G, Crowley JJ, Gaspar HA, et al. Genetic identification of brain cell types underlying schizophrenia. Nat Genet. 2018;50:825–33. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Anderson KM, Collins MA, Chin R, Ge T, Rosenberg MD, Holmes AJ. Transcriptional and imaging-genetic association of cortical interneurons, brain function, and schizophrenia risk. Nat Commun. 2020;11:2889. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Rummel CK, Gagliardi M, Ahmad R, Herholt A, Jimenez-Barron L, Murek V, et al. Massively parallel functional dissection of schizophrenia-associated noncoding genetic variants. Cell. 2023;186:5165–82.e33. [DOI] [PubMed] [Google Scholar]
- 12.Messedi M, Makni-Ayadi F. 24S-hydroxycholesterol in neuropsychiatric diseases: schizophrenia, autism spectrum disorder, and bipolar disorder. Adv Exp Med Biol. 2024;1440:293–304. [DOI] [PubMed] [Google Scholar]
- 13.Vigne S, Pot C. Implication of oxysterols and phytosterols in aging and human diseases. Adv Exp Med Biol. 2024;1440:231–60. [DOI] [PubMed] [Google Scholar]
- 14.Broccoli V, Caiazzo M. Nuclear receptors: oxysterols detour to neurodevelopment. Nat Chem Biol. 2013;9:70–1. [DOI] [PubMed] [Google Scholar]
- 15.Theofilopoulos S, Abreu de Oliveira WA, Yang S, Yutuc E, Saeed A, Abdel-Khalik J, et al. 24(S),25-Epoxycholesterol and cholesterol 24S-hydroxylase (CYP46A1) overexpression promote midbrain dopaminergic neurogenesis in vivo. J Biol Chem. 2019;294:4169–76. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Sacchetti P, Sousa KM, Hall AC, Liste I, Steffensen KR, Theofilopoulos S, et al. Liver X receptors and oxysterols promote ventral midbrain neurogenesis in vivo and in human embryonic stem cells. Cell Stem Cell. 2009;5:409–19. [DOI] [PubMed] [Google Scholar]
- 17.Wang Y, Sousa KM, Bodin K, Theofilopoulos S, Sacchetti P, Hornshaw M, et al. Targeted lipidomic analysis of oxysterols in the embryonic central nervous system. Mol Biosyst. 2009;5:529–41. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.De La Fuente DC, Tamburini C, Stonelake E, Andrews R, Hall J, Owen MJ, et al. Impaired oxysterol-liver X receptor signaling underlies aberrant cortical neurogenesis in a stem cell model of neurodevelopmental disorder. Cell Rep. 2024;43:113946. [DOI] [PubMed] [Google Scholar]
- 19.Raleigh DR, Sever N, Choksi PK, Sigg MA, Hines KM, Thompson BM, et al. Cilia-associated oxysterols activate smoothened. Mol Cell. 2018;72:316–27.e5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Pantazis CB, Yang A, Lara E, McDonough JA, Blauwendraat C, Peng L, et al. A reference human induced pluripotent stem cell line for large-scale collaborative studies. Cell Stem Cell. 2022;29:1685–702.e22. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Cruz-Santos M, Cardo LF, Li M. A Novel LHX6 reporter cell line for tracking human ipsc-derived cortical interneurons. Cells. 2022;11:853. 10.3390/cells11050853. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Cruz MS, Li M. Identification of TGFβ signaling as a regulator of interneuron neurogenesis in a human pluripotent stem cell model. Neuronal Signal. 2021;5:NS20210020. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Hao Y, Stuart T, Kowalski MH, Choudhary S, Hoffman P, Hartman A, et al. Dictionary learning for integrative, multimodal and scalable single-cell analysis. Nat Biotechnol. 2024;42:293–304. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.McCarthy DJ, Campbell KR, Lun ATL, Wills QF. Scater: pre-processing, quality control, normalization and visualization of single-cell RNA-seq data in R. Bioinformatics. 2017;33:1179–86. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Choudhary S, Satija R. Comparison and evaluation of statistical error models for scRNA-seq. Genome Biol. 2022;23:27. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Korsunsky I, Millard N, Fan J, Slowikowski K, Zhang F, Wei K, et al. Fast, sensitive and accurate integration of single-cell data with Harmony. Nat Methods. 2019;16:1289–96. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Nestorowa S, Hamey FK, Pijuan Sala B, Diamanti E, Shepherd M, Laurenti E, et al. A single-cell resolution map of mouse hematopoietic stem and progenitor cell differentiation. Blood. 2016;128:e20–31. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Mayer C, Hafemeister C, Bandler RC, Machold R, Batista Brito R, Jaglin X, et al. Developmental diversification of cortical inhibitory interneurons. Nature. 2018;555:457–62. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Shi Y, Wang M, Mi D, Lu T, Wang B, Dong H, et al. Mouse and human share conserved transcriptional programs for interneuron development. Science. 2021;374:eabj6641. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Hinshaw DC, Shevde LA. The tumor microenvironment innately modulates cancer progression. Cancer Res. 2019;79:4557–66. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Yu G, Wang L-G, Han Y, He Q-Y. clusterProfiler: an R package for comparing biological themes among gene clusters. OMICS. 2012;16:284–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Koopmans F, van Nierop P, Andres-Alonso M, Byrnes A, Cijsouw T, Coba MP, et al. SynGO: an evidence-based, expert-curated knowledge base for the synapse. Neuron. 2019;103:217–34.e4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Aibar S, González-Blas CB, Moerman T, Huynh-Thu VA, Imrichova H, Hulselmans G, et al. SCENIC: single-cell regulatory network inference and clustering. Nat Methods. 2017;14:1083–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Van de Sande B, Flerin C, Davie K, De Waegeneer M, Hulselmans G, Aibar S, et al. A scalable SCENIC workflow for single-cell gene regulatory network analysis. Nat Protoc. 2020;15:2247–76. [DOI] [PubMed] [Google Scholar]
- 35.Shannon P, Markiel A, Ozier O, Baliga NS, Wang JT, Ramage D, et al. Cytoscape: a software environment for integrated models of biomolecular interaction networks. Genome Res. 2003;13:2498–504. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.de Leeuw CA, Mooij JM, Heskes T, Posthuma D. MAGMA: generalized gene-set analysis of GWAS data. PLoS Comput Biol. 2015;11:e1004219. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Trubetskoy V, Pardiñas AF, Qi T, Panagiotaropoulou G, Awasthi S, Bigdeli TB, et al. Mapping genomic loci implicates genes and synaptic biology in schizophrenia. Nature. 2022;604:502–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Murphy AE, Schilder BM, Skene NG. MungeSumstats: a bioconductor package for the standardization and quality control of many GWAS summary statistics. Bioinformatics. 2021;37:4593–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Cardo LF, de la Fuente DC, Li M. Impaired neurogenesis and neural progenitor fate choice in a human stem cell model of SETBP1 disorder. Mol Autism. 2023;14:8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Nguyen TP, Wang W, Sternisha AC, Corley CD, Wang H-YL, Wang X, et al. Selective and brain-penetrant lanosterol synthase inhibitors target glioma stem-like cells by inducing 24(S),25-epoxycholesterol production. Cell Chem Biol. 2023;30:214–29.e18. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Nóbrega-Pereira S, Gelman D, Bartolini G, Pla R, Pierani A, Marín O. Origin and molecular specification of globus pallidus neurons. J Neurosci. 2010;30:2824–34. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Zhao Y, Marín O, Hermesz E, Powell A, Flames N, Palkovits M, et al. The LIM-homeobox gene Lhx8 is required for the development of many cholinergic neurons in the mouse forebrain. Proc Natl Acad Sci USA 2003;100:9005–10. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Bocchi VD, Conforti P, Vezzoli E, Besusso D, Cappadona C, Lischetti T, et al. The coding and long noncoding single-cell atlas of the developing human fetal striatum. Science. 2021;372:eabf5759. 10.1126/science.abf5759. [DOI] [PubMed] [Google Scholar]
- 44.Puelles L, Rubenstein JLR. Forebrain gene expression domains and the evolving prosomeric model. Trends Neurosci. 2003;26:469–76. [DOI] [PubMed] [Google Scholar]
- 45.Savory JGA, Edey C, Hess B, Mears AJ, Lohnes D. Identification of novel retinoic acid target genes. Dev Biol. 2014;395:199–208. [DOI] [PubMed] [Google Scholar]
- 46.Willy PJ, Umesono K, Ong ES, Evans RM, Heyman RA, Mangelsdorf DJ. LXR, a nuclear receptor that defines a distinct retinoid response pathway. Genes Dev. 1995;9:1033–45. [DOI] [PubMed] [Google Scholar]
- 47.Julian LM, Vandenbosch R, Pakenham CA, Andrusiak MG, Nguyen AP, McClellan KA, et al. Opposing regulation of Sox2 by cell-cycle effectors E2f3a and E2f3b in neural stem cells. Cell Stem Cell. 2013;12:440–52. [DOI] [PubMed] [Google Scholar]
- 48.Dhanesh SB, Subashini C, James J. Hes1: the maestro in neurogenesis. Cell Mol Life Sci. 2016;73:4019–42. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 49.Singh A, Mahesh A, Noack F, Cardoso de Toledo B, Calegari F, Tiwari VK. Tcf12 and NeuroD1 cooperatively drive neuronal migration during cortical development. Development. 2022;149:dev200250. 10.1242/dev.200250. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 50.Kuwahara A, Sakai H, Xu Y, Itoh Y, Hirabayashi Y, Gotoh Y. Tcf3 represses Wnt-β-catenin signaling and maintains neural stem cell population during neocortical. development PLoS ONE. 2014;9:e94408. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 51.Hiew L-F, Poon C-H, You H-Z, Lim L-W. TGF-β/Smad signalling in neurogenesis: implications for neuropsychiatric diseases. Cells. 2021;10:1382. 10.3390/cells10061382. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 52.Ballas N, Grunseich C, Lu DD, Speh JC, Mandel G. REST and its corepressors mediate plasticity of neuronal gene chromatin throughout neurogenesis. Cell. 2005;121:645–57. [DOI] [PubMed] [Google Scholar]
- 53.Zurkirchen L, Varum S, Giger S, Klug A, Häusel J, Bossart R, et al. Yin Yang 1 sustains biosynthetic demands during brain development in a stage-specific manner. Nat Commun. 2019;10:2192. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 54.Lehmann JM, Kliewer SA, Moore LB, Smith-Oliver TA, Oliver BB, Su JL, et al. Activation of the nuclear receptor LXR by oxysterols defines a new hormone response pathway. J Biol Chem. 1997;272:3137–40. [DOI] [PubMed] [Google Scholar]
- 55.Leoni V, Caccia C. Oxysterols as biomarkers in neurodegenerative diseases. Chem Phys Lipids. 2011;164:515–24. [DOI] [PubMed] [Google Scholar]
- 56.Gamba P, Testa G, Sottero B, Gargiulo S, Poli G, Leonarduzzi G. The link between altered cholesterol metabolism and Alzheimer’s disease. Ann N Y Acad Sci. 2012;1259:54–64. [DOI] [PubMed] [Google Scholar]
- 57.Jin U, Park SJ, Park SM. Cholesterol metabolism in the brain and its association with parkinson’s disease. Exp Neurobiol. 2019;28:554–67. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 58.Karasinska JM, Hayden MR. Cholesterol metabolism in huntington disease. Nat Rev Neurol. 2011;7:561–72. [DOI] [PubMed] [Google Scholar]
- 59.Sun Z, Zhao L, Bo Q, Mao Z, He Y, Jiang T, et al. Brain-Specific oxysterols and risk of schizophrenia in clinical high-risk subjects and patients with schizophrenia. Front Psychiatry. 2021;12:711734. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 60.Grayaa S, Zerbinati C, Messedi M, HadjKacem I, Chtourou M, Ben Touhemi D, et al. Plasma oxysterol profiling in children reveals 24-hydroxycholesterol as a potential marker for autism spectrum disorders. Biochimie. 2018;153:80–5. [DOI] [PubMed] [Google Scholar]
- 61.Guidara W, Messedi M, Maalej M, Naifar M, Khrouf W, Grayaa S, et al. Plasma oxysterols: Altered level of plasma 24-hydroxycholesterol in patients with bipolar disorder. J Steroid Biochem Mol Biol. 2021;211:105902. [DOI] [PubMed] [Google Scholar]
- 62.Kessaris N, Magno L, Rubin AN, Oliveira MG. Genetic programs controlling cortical interneuron fate. Curr Opin Neurobiol. 2014;26:79–87. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 63.Inan M, Welagen J, Anderson SA. Spatial and temporal bias in the mitotic origins of somatostatin- and parvalbumin-expressing interneuron subgroups and the chandelier subtype in the medial ganglionic eminence. Cereb Cortex. 2012;22:820–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 64.Miyoshi G, Hjerling-Leffler J, Karayannis T, Sousa VH, Butt SJB, Battiste J, et al. Genetic fate mapping reveals that the caudal ganglionic eminence produces a large and diverse population of superficial cortical interneurons. J Neurosci. 2010;30:1582–94. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 65.Miyoshi G, Butt SJB, Takebayashi H, Fishell G. Physiologically distinct temporal cohorts of cortical interneurons arise from telencephalic Olig2-expressing precursors. J Neurosci. 2007;27:7786–98. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 66.Mi D, Li Z, Lim L, Li M, Moissidis M, Yang Y, et al. Early emergence of cortical interneuron diversity in the mouse embryo. Science. 2018;360:81–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 67.Davenport EC, Szulc BR, Drew J, Taylor J, Morgan T, Higgs NF, et al. Autism and Schizophrenia-Associated CYFIP1 Regulates the Balance of Synaptic Excitation and Inhibition. Cell Rep. 2019;26:2037–51.e6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 68.Tyson JA, Anderson SA. GABAergic interneuron transplants to study development and treat disease. Trends Neurosci. 2014;37:169–77. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 69.Southwell DG, Froemke RC, Alvarez-Buylla A, Stryker MP, Gandhi SP. Cortical plasticity induced by inhibitory neuron transplantation. Science. 2010;327:1145–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 70.Maisano X, Litvina E, Tagliatela S, Aaron GB, Grabel LB, Naegele JR. Differentiation and functional incorporation of embryonic stem cell-derived GABAergic interneurons in the dentate gyrus of mice with temporal lobe epilepsy. J Neurosci. 2012;32:46–61. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 71.Liu Y, Weick JP, Liu H, Krencik R, Zhang X, Ma L, et al. Medial ganglionic eminence-like cells derived from human embryonic stem cells correct learning and memory deficits. Nat Biotechnol. 2013;31:440–7. [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
List of supplemental materials and supplemental figure legend
Data Availability Statement
All singel-cell RNA sequencing data from this project were deposited at NCBI Gene Expression Omnibus (GEO) with the accession number GSE298622.
