Canonical Gene Transcription Associated With Human Cerebral Cortical Volume

Objective:As one commonly used brain structural imaging measure,the cerebral cortical volume(CCV)has been associated with varieties of human behavioral performance and neuropsychiatric disorders.In our study,we decide to identify the conserved molecular programs accounting for regional CCV differences.Afterwards we annotate the CCV-related modules by module enrichment,cell type-and temporal-specific expression.Such annotation can further characterize these modules and provide possible mechanism about how gene modules affect regional CCV differences.Subjects and Methods:The study was divided into three parts,the discovery experiment was performed on the six postmortem brains from Allen Human Brain Atlas(AHBA)with both transcriptional and imaging data.Genes with highly consistent transcription patterning across cortical regions and individuals were screened out according to the index of differential stability(DS),reflecting the tendency for a gene to exhibit reproducible differential expression across cortical regions.Since the high DS genes were more specific to brain tissue and more important to brain function,we reduced the 20736 genes by half to only maintain genes with high DS value.Then the high DS genes were clustered into consensus modules by the weighted gene co-expression network analysis(WGCNA).Afterwards,Pearson correlation analyses between the expression of each module eigengene(ME)of each module and the CCV across cortical samples were performed to identify modules with consistent and significant correlations in most donated brains.The reproducibility and generalization of these associations were validated in four independent healthy populations with different age and ethnicity,which included 200(100 females,100 males)healthy young subjects from Human Connectome Project(HCP);226(108 females,118 males)healthy elder subjects from Alzheimer Disease Neuroimaging Initiative(ADNI);1102(595 females,507 males)healthy young Chinese(CHI-Y);and 21(14 females,7 males)healthy elder Chinese(CHI-E).Then the CCV-related modules were annotated by moduleenrichment,cell type-and temporal-specific expression.Results:1.Using weighted gene co-expression network analysis,10370 high DS genes were clustered into 43 consensus modules.2.Two modules were spatially correlated with CCV in five of the six donated brains from Allen Human Brain Atlas.These correlations were validated in four independent imaging datasets with different age and ethnicity.3.The CCV-related modules were enriched in neurons and synapses,and showed significant cell type-specific expression in cortical neurons.The module positively correlated with CCV showed high expression in the critical period of development.However,the module negatively correlated with CCV demonstrated greatly different temporal-specific expression patterns.Conclusion:1.Conserved gene transcription in cortical neurons rather than other components of the cerebral cortex is critical for regional CCV variances in human brain.2.The module positively correlated with CCV is possibly associated with the early development of the cerebral cortex.However,the module negatively correlated with CCV may be involved in both developmental and aging processes of the human cerebral cortex.