Cortical development dynamics across autism spectrum disorder mouse models

Main

Several studies have uncovered the heterogeneous genetic landscape of ASD, identifying numerous monogenic forms 4 , 5 . Although these discoveries are crucial for understanding ASD biology, fundamental questions remain: how mutations that are scattered across many genes lead to similar disorders; and how similar these genetically defined ASDs are to one another.

Here we used single-nucleus multi-omics sequencing to study 251 samples from mouse models of high-risk genes linked to ASD, spanning developmental stages, sexes and two brain regions (Fig. 1a ). These data comprehensively represent cell states across ASD-associated mutants, uncovering converging as well as diverging phenotypes.

Fig. 1: Large-scale multi-omics profiling of high-risk genes associated with ASD. The alternative text for this image may have been generated using AI.

Full size image

a , Simultaneous transcriptome and chromatin accessibility profiling of brain samples from 11 monogenic mouse lines, each carrying a mutation in a high-risk gene linked to ASD. An overview of the experimental design is shown, indicating total samples and cells analysed per time point. Illustration of the multi-omics approach created by J. Kirchner. b – d , Uniform manifold approximation and projection (UMAP) visualization of integrated snRNA-seq and snATAC-seq data from E15.4 ( b ), P4 ( c ) and P14 ( d ). Cell clusters are coloured and numbered by cell type. ET, extratelencephalic-projecting neurons; NP, near-projecting; PN, projection neuron; SVZ, subventricular zone; Vend, vascular endothelial cells; VLMC, vascular leptomeningeal cell. e – g , Changes in cell-type abundances in individual mutants linked to ASD from E14.5 ( e ), P4 ( f ) and P14 ( g ). The arrowheads indicate relative percentage differences compared with WT. Statistical significance was assessed in edgeR using a two-tailed quasi-likelihood F -test. Nominal P values are shown. h , Pseudotime analysis of the RG lineage at E14.5. The solid line indicates the trajectory inferred from gene expression embedding ( Methods ). The dashed arrow indicates the trajectory from the RG (root) towards cortical plate cells (left). Two RG subpopulations were distinguished: RG1 (G1/S phase) and RG2 (G2 phase) (centre and right). i , Pseudotime analysis of RG cells indicating that mutants linked to ASD tend to remain in the proliferating RG1 subpopulation. The vertical line marks the boundary between two modes of the pseudotime distribution. j , Median pseudotime analysis of RG cells indicates a consistent developmental delay in mutants linked with ASD compared with WT. The centre line represents the median, the box limits show the 25–75% interquartile range (IQR), the whiskers extend to the furthest data points within 1.5× the IQR from the box edges, and the data points outside this range are represented as dots. The dashed line indicates the median of the WT per-sample averages. Animals used for analyses: n (E14.5) = 7 ( Cul3 +/ − , Kdm6b +/ − and Trip12 +/ − ), 6 ( Ash1l +/ − , Bckdk − / − , HnrnpU +/ − , Kmt5b +/ − , Setd5 +/ − , Usp7 +/ − , Wac +/ − and colony-matched controls) and 3 ( Ptchd1 −/ y ); n (P4) = 6 ( Ash1l +/ − , Bckdk − / − , Cul3 +/ − , HnrnpU +/ − , Kdm6b +/ − , Setd5 +/ − , Trip12 +/ − , Usp7 +/ − , Wac +/ − and colony-matched controls), 4 ( Kmt5b +/ − ) and 3 ( Ptchd1 −/ y ); n (P14) = 8 ( Ash1l +/ − and Cul3 +/ − ), 7 (colony-matched controls), 6 (Bckdk − / − , HnrnpU +/ − , Kdm6b +/ − , Kmt5b +/ − , Setd5 +/ − , Trip12 +/ − , Usp7 +/ − and Wac +/ − ) and 3 ( Ptchd1 −/ y ).

Single-nucleus multi-omics profiling

Droplet-based single-cell multi-omic technologies enable matched gene expression and chromatin accessibility profiling in thousands of single cells in one experiment. However, existing methods remain costly and limited in scalability. Here we optimized a multiplexing strategy for combined single-nucleus RNA sequencing (snRNA-seq) and single-nucleus assay for transposase-accessible chromatin using sequencing (snATAC-seq). We utilized a multiplexing approach using cholesterol-modified oligos (CMOs) 6 as sample-specific barcodes (Fig. 1a and Supplementary Table 1 ), enabling pooled processing. In this way, we profiled 45 conditions comprising 251 samples (Supplementary Table 1 ), including mutants and controls, allowing us to assess transcriptional and chromatin states at single-cell resolution across a broad range of genetic ASD mouse models (Fig. 1a ; Methods ).

In total, we collected the molecular profiles of 200,787 nuclei (66,969 at embryonic day 14.5 (E14.5), 56,260 at postnatal day 4 (P4) and 59,974 at P14 for the forebrain; 17,584 at P4 for the cerebellum). We detected an average of approximately 3,900 transcripts, about 2,000 genes and approximately 4,200 fragments per cell at a median sequencing depth of about 25,000 reads per cell (Supplementary Figs. 1d and 3a ), aligning with standard recommendations for these types of analyses.

We identified 12 (E14.5), 22 (P4), 26…