From transcription to function: mapping brain networks

The brain is a series of connected functional networks which coordinate the control of processes like vision and language. Mapping the molecular networks that underlie this connectivity would provide an unprecedented understanding of brain function in health and disease. However, significant technical challenges hinder our ability to study functional and molecular circuits simultaneously—one approach requires an intact brain while the other requires sampling throughout all brain regions.  To circumvent these major limitations Richiardi et al. approached the challenge by correlating and mapping two fundamentally distinct circuits in the cortex: gene transcription networks and functional networks.

Using resting state functional magnetic resonance imaging (fMRI) in subjects aged 18 to 29, the authors began by confirming four canonical functional cortical networks: sensorimotor, visuospatial, dorsal default-mode, and salience. They then mapped cortical gene transcripts from the Allen Institute for Brain Science microarray data set onto these four functional circuits. A linear correlation analysis was then applied b­­etween all mapped brain samples. This resulted in clusters of gene transcripts with similar expression levels within specific functional networks—the high strength fraction—and different or lower expression levels across unrelated networks—the low strength fraction. Finally, the authors identified the gene transcripts that drove the four functional networks and obtained a consensus list of 136 top-ranked gene transcripts.

Their findings were validated in two ways: First, the researchers used in vivo fMRI data from 259 individuals each paired to a single nucleotide polymorphism (SNP) data set. They confirmed that specific polymorphisms within the genes in the consensus list enriched the strength of functional networks. Second, the authors reported that many of the consensus list transcripts corresponded to ion channels highly expressed along axonal paths within the network. Noteworthy, a proportion of transcripts were gene candidates for neurodegeneration. These results agree with the initial functional networks found in adolescents. In addition, these findings validate the well-known involvement of ion channels as the molecular units driving transmission of information across neuronal networks and underscore the known involvement of ion channel-enriched networks in health and disease.  

Overall, the molecular-functional networks herein proposed by Richiardi et al. emerge from a data-driven—rather than hypothesis-driven—approach. This work provides a useful framework to continue investigating brain function from a multi-dimensional approach. Some questions worth exploring in the near future include: Is it enough to define a “molecular network” based solely on the linear correlation between the genetic transcripts within anatomically defined samples? How reliably can molecular networks proposed from postmortem tissue or blood samples recapitulate in vivo networks? And finally, will the proposed connectivity/functional networks differ depending on the level of brain activity or engagement?

Publication

1. Richiardi J et al. (2015) Correlated gene expression supports synchronous activity in brain networks. Science 348(6240):1241-1244. doi: 10.1126/science.1255905