Cortical remodelling in childhood is associated with genes enriched for neurodevelopmental disorders

Abstract

<jats:title>Abstract</jats:title><jats:p>Cortical development during childhood and adolescence has been characterised in recent years using metrics derived from Magnetic Resonance Imaging (MRI). Changes in cortical thickness are greatest in the first two decades of life and recapitulate the genetic organisation of the cortex, highlighting the potential early impact of gene expression on differences in cortical architecture over the lifespan. It is important to further our understanding of the possible neurobiological mechanisms that underlie these changes as differences in cortical thickness may act as a potential phenotypic marker of several common neurodevelopmental and psychiatric disorders.</jats:p><jats:p>In this study, we combine MRI acquired from a large typically-developing childhood population (n=768) with comprehensive human gene expression databases to test the hypothesis that disrupted mechanisms common to neurodevelopmental disorders are encoded by genes expressed early in development and nested within those associated with typical cortical remodelling in childhood.</jats:p><jats:p>We find that differential rates of thinning across the developing cortex are associated with spatially-varying gradients of gene expression. Genes that are expressed highly in regions of accelerated thinning are expressed predominantly in cortical neurons, involved in synaptic remodeling, and associated with common cognitive and neurodevelopmental disorders. Further, we identify subsets of genes that are highly expressed in the prenatal period and jointly associated with both developmental cortical morphology and neurodevelopmental disorders.</jats:p>

Mutations associated with neuropsychiatric conditions delineate functional brain connectivity dimensions contributing to autism and schizophrenia

“…PLSR method was used to investigate the association between spatial patterns of gene expression (of the 37 and 24 genes encompassed in the 22q11.2 and 16p11.2 genomic loci) and the 16p11.2 and 22q11.2 FC signatures. PLSR is a multivariate approach, which has previously been applied to investigate the relationship between neuroimaging phenotypes and spatial patterns of gene expression [73][74][75][76] . PLSR was performed separately for 16p11.2 and 22q11.2 genes.…”

Early or late gestational exposure to maternal immune activation alters neurodevelopmental trajectories in mice: an integrated neuroimaging, behavioural, and transcriptional study

“…Indeed, these strategies disregard the ever-increasingly acknowledged network-like architecture of the brain and the inherent relationships to behaviour (98,99). In contrast, multivariate strategies, such as PLS, move beyond simplistic associations to better understand brain-behaviour relationships (100)(101)(102). In the context of small animal imaging, PLS provides a novel, streamlined method to assess cross-sectional brain-behaviour patterns from large cohorts of deeply phenotyped mice.…”

Individual variation underlying brain age estimates in typical development

“…Structural T1 volumes were processed as described previously [Ball et al, 2017;Ball et al, 2020]. Briefly, vertex-wise maps of cortical thickness and area were constructed with FreeSurfer 5.3 (http://surfer.nmr.mgh.harvard.edu).…”

“…For all subjects, cortical thickness and area maps were parcellated into n = 179 labels per hemisphere using a parcellation scheme based on anatomical and functional cortical boundaries (excluding the hippocampus) [Glasser et al, 2016]. We repeated our analysis using an alternative parcellation comprising n = 250 regions per hemisphere with approximately equal area [Arnatkeviciūtė et al, 2019;Ball et al, 2020]. Regional measures of mean cortical thickness and surface area were calculated using the 95% trimmed mean and logtransformed sum of all vertex values contained within the assigned region, respectively.…”

“…or define any inflection points. GPR has previously been used successfully to model relationships between MRI-derived metrics and age [Ball et al, 2020;Cole et al, 2015;Cole et al, 2018]. We review GPR below; for a technical overview we refer interested readers to [Rasmussen and Williams, 2006].…”