Research Report: The Connection Between Cortical Gene Expression and Neurodevelopmental Disorders

Research Background

The anatomical and functional organization of the human brain is the result of the coordinated expression of numerous genes. It has been found that the first principal component (c1) of cortical gene expression plays a significant role in hierarchical expression from sensory-motor areas to association areas. However, whether there are other key components of gene expression has always been a concern of the scientific community. Over the past decade, transcriptional maps of the whole brain and genome (such as Allen Human Brain Atlas, AHBA) have shown that healthy brain tissue may depend on the “transcriptional program” of the coordinated expression of a large number of genes during development. This research is conducted in this context, aiming to reveal more gene expression components and their roles in brain development and neurodevelopmental disorders.

Authors and Publication Information

This paper was jointly completed by the team of Richard Dear, Konrad Wagstyl, Jakob Seidlitz, Ross D. Markello, Aurina Arnatkevičiūtė, Kevin M. Anderson, Richard A. I. Bethlehem, Armin Raznahan, Edward T. Bullmore, and Petra E. Vértes. The team assembled the strengths of multiple institutions including the University of Cambridge, Wellcome Centre for Human Neuroimaging in London, Children’s Hospital of Philadelphia, and the Lifespan Brain Chart Alliance. The paper was published in the June 2024 issue of “Nature Neuroscience”, and the article link is https://doi.org/10.1038/s41593-024-01624-4.

Research Process

a) Research Workflow

Data Processing and Analysis

The researchers first optimized the AHBA dataset, using data from the Allen Human Brain Atlas, performed principal component analysis (PCA) on the microarray measurement of relative mRNA levels in six adult brains, and identified three components (c1, c2, c3). To verify the universality of these components, they used other datasets such as PsychENCODE, Allen Cell Atlas, and BrainSpan. The researchers found that through optimization and dimensionality reduction methods, these components are coordinately expressed during the fetal period and postnatal development process.

Experimental Process and New Methods

In processing the AHBA data, the research team for the first time adopted the diffusion map embedding (DME) of nonlinear dimensionality reduction techniques, which is more noise-resistant than linear PCA and is more biologically plausible. Therefore, DME identified the same components in the filtered gene expression matrix, but its universality was significantly improved.

Data Filtering and Gene Expression Analysis

The researchers filtered the genes and brain regions of the AHBA dataset, selected 137 brain regions with data in at least three human brains, and preserved the most stable 50% of genes. Through the DME method, the matrix composed of the first 50% most stable gene expression and 137 brain regions was applied to the analysis, significantly enhancing the universality of the first three components. At the same time, the researchers found that the brain region scores derived from the filtered data using DME are smoother than the scores obtained from unfiltered data using PCA, indicating that higher universality means less likelihood of contamination by spatial random noise.

b) Main Research Findings

Component and its Biological Enrichment Analysis

The study revealed three universal cortical gene expression components (c1, c2, c3), which are closely related to biological processes, cell types, and brain structural features. Specifically:

• c1 is mainly correlated with the enriched genes of neurons, inhibitory interneurons, and glutamatergic neurons.
• c2 is related to enriched genes in metabolic processes and epigenetic processes.
• c3 is related to synaptic plasticity, learning, and immune processes.

These components show different axial alignments in different brain anatomical structures, and even after filtering out the highest variance component, the co-expression network of cortical areas still shows a significant anatomical structure.

Neuroimaging Comparison of Components

The study shows that the three transcription components are specifically co-localized with various neuroimaging or other macro brain phenotypes. For example:

• c1 is strongly correlated with the weighted node degree of MRI networks, but unrelated to other components.
• c2 is significantly related to theta wave (4–7 Hz) oscillations in MEG data.
• c3 expression significantly increases during adolescence, which is consistent with previous studies on cortical myelination in adolescence.

Relationship between Transcription Components and Developmental Process

The study further used the BrainSpan dataset to explore the developmental trajectory of transcription components and found:

• c1 and c2 are close to adult expression patterns in the fetal period and childhood.
• c3 starts strong expression significantly in adolescence, indicating that this component is related to extra myelination and tiling axonal pruning during adolescence.

c) Summary and Research Significance

The study shows that there is a logical relationship between the three components of cortical gene expression, and each component plays a key role in specific stages of brain development and biological processes. The important findings of the research include:

1. c1 is closely related to Autism Spectrum Disorder (ASD).
2. c2 is related to cognitive metabolic processes and is also related to ASD.
3. c3 is related to adolescent brain development and atypical upper cortical connections in individuals at high genetic risk for schizophrenia.

These findings not only extend the understanding of cortical gene expression but also provide a new perspective for exploring the pathogenesis of neurodevelopmental disorders.

d) Research Highlights

The highlight of this research is the use of unique optimization processes and dimensionality reduction methods to reveal new gene expression components and to prove the biological and clinical relevance of these components through cross-verification of multiple datasets. This transcriptional program is important for understanding the pathology of normal brain development and neurodevelopmental disorders.

e) Other Valuable Information

The study also points out that the development of new, high-granularity transcriptional data may reveal more gene expression components, providing a direction for future research. At the same time, the openness of research data and codes also opens up the possibility of further research in related fields.

Research Significance and Value

This research not only scientifically reveals the key gene expression patterns during brain development but also provides a new method in the clinical research of neurodevelopmental disorders, which is of great significance for advancing brain science and psychiatric research. These research methods and results will help develop new diagnostic and treatment strategies to address complex neurodevelopmental disorders such as autism and schizophrenia.