Chromatin accessibility during human first-trimester neurodevelopment

The human brain's development involves a tightly orchestrated series of events driven by transcription factor expression and chromatin accessibility changes. While single-cell RNA sequencing (scRNA-seq) atlases of gene expression exist, comprehensive atlases of chromatin accessibility, especially beyond the forebrain, are limited. This study aimed to create a comprehensive, single-cell resolution atlas of chromatin accessibility and paired gene expression across the entire developing human brain during the first trimester (6–13 weeks post-conception). This period is crucial for brain patterning and the establishment of neural cell type identities. Understanding the regulatory landscape during this stage is essential for interpreting genome-wide association studies (GWAS) of neurodevelopmental disorders, as many disease-associated loci reside in non-coding regions and exhibit context-specific activity. Previous studies have focused on later stages of development or specific brain regions, leaving a significant gap in our understanding of the first trimester. This study directly addresses this gap by providing a holistic view of chromatin accessibility and gene expression across the entire developing brain during this critical developmental window.

Previous research has mapped the regulatory landscapes of the developing human brain, but these studies primarily focused on the second-trimester developing cortex, whole embryos, organoids, and induced pluripotent stem cell-derived models. These studies provided valuable insights, but lacked the comprehensive, single-cell resolution across the entire brain during the first trimester. Studies using scRNA-seq have provided valuable information about cell types and states in the developing brain, revealing regional differences and subtle variations between closely related cell types. However, understanding the regulatory mechanisms behind these observed variations required the integration of chromatin accessibility data with gene expression data. Therefore, this study builds upon this previous work by integrating both scRNA-seq and scATAC-seq data to provide a more complete understanding of the regulatory landscape.

The researchers analyzed chromatin accessibility in 18 specimens using scATAC-seq, and combined scATAC-seq and scRNA-seq (multiome) in 3 specimens. Samples were dissected into major brain regions (telencephalon, diencephalon, mesencephalon, metencephalon, and cerebellum). After quality control, they analyzed chromatin profiles from 526,094 nuclei and gene expression profiles from 166,785 nuclei. They used stratified peak calling, latent semantic indexing, and Louvain clustering to identify 135 cell clusters. A modified version of Cicero was used to identify candidate *cis*-regulatory elements (cCREs). The researchers applied a convolutional neural network (CNN) to predict cell types based on sequence composition. They also employed DELAY, another CNN method, to estimate gene regulatory networks and BoolODE for simulating gene expression dynamics. Finally, they performed stratified linkage disequilibrium score regression to identify cell types vulnerable to psychiatric disorder-associated SNPs, validated in a second cohort.

The number of accessible regions increased with both age and neuronal differentiation. The CNN accurately predicted cell types (average ROC AUC of 0.92). Analysis of *ESRRB* regulation in Purkinje cells revealed a two-step activation process involving TFAP2B and LHX5. GWAS analysis revealed an association between major depressive disorder (MDD) and midbrain-derived GABAergic neurons. While many MDD-associated genes weren't cell-type specific, the associated accessible regions showed enriched TF motifs consistent with a midbrain GABAergic identity. This suggests that these genes contribute to MDD when dysregulated specifically in these midbrain cell types during early development. The study identified over 100,000 cell-type- and region-specific developmental accessible chromatin regions and predicted their regulatory syntax using CNN modeling.

This study provides a detailed, multiomic atlas of the developing human brain during the first trimester, offering significant advancements in our understanding of early neurodevelopment. The use of a CNN to predict cell-type identity from sequence composition is a novel approach that allows for a more in-depth understanding of the regulatory mechanisms driving cell fate decisions. The findings highlight the context-dependent nature of disease-associated alleles, emphasizing that genetic variations may only contribute to disease when they affect specific cell types during critical developmental windows. The strong association of MDD with midbrain GABAergic neurons warrants further investigation into the specific mechanisms underlying this vulnerability. Further research is needed to investigate the role of other cell types in MDD, especially during later stages of development.

This study offers a high-resolution, multiomic resource for understanding early human brain development. The identification of numerous cell-type-specific accessible chromatin regions, along with the elucidation of *ESRRB* regulation and the association of MDD with midbrain GABAergic neurons, significantly contributes to the field. Future research should expand on these findings by investigating the broader implications of these findings for other neurodevelopmental and psychiatric disorders and exploring the regulatory mechanisms in other developmental stages.

The study used clinical samples, leading to variability in developmental timing and incomplete data collection for some regions. The sample size was limited by the availability of early developmental human brain samples. The analysis focused on an early developmental period, potentially missing later-stage contributions to complex diseases.