Contribution of energy metabolism to structure and function of the human brain

2595 

Abstract Submission 

Moohebat Pourmajidian¹, Bratislav Misic¹, Alain Dagher²

¹McGill University, Montreal, Quebec, ²Montreal Neurological Institute and Hospital, McGill University, Montreal, QC

Moohebat Pourmajidian  
McGill University
Montreal, Quebec

Bratislav Misic  
McGill University
Montreal, Quebec

Alain Dagher  
Montreal Neurological Institute and Hospital, McGill University
Montreal, QC

The brain relies on energy from glucose metabolism to maintain its structure and carry out its functional repertoire. Perturbations of energy distribution in the brain can therefore lead to physiological and cognitive dysfunction. Abnormal neurovascular and neurometabolic coupling, oxidative metabolism, and mitochondrial function have been observed in the aging brain and several neurodevelopmental and neurodegenerative disorders, including Parkinson's disease (PD), Alzheimer's disease (AD). Metabolic dysfunctions appear early in disease progression (Watts et al., 2018) and mitochondrial dysfunction has been associated with protein aggregation in both AD and PD (Cho et al., 2010). Here, by leveraging gene expression data, we investigate how different energy metabolism pathways are distributed in the human cortex.

We curated gene sets of known pathways of glucose and lactate metabolism using Gene Ontology and Reactome databases. Microarray data from the Allen Human Brain Atlas (AHBA) was preprocessed using the abagen package (Markello et al., 2021) and parcellated into 400 cortical regions in the Schaefer parcellation (Schaefer et al., 2018). Average gene expression maps of glucose-specific metabolic pathways were produced and used as an indicator of metabolic activity across the human cortex. Distribution of metabolic expression maps across functional and cytoarchitectural classes of the human cortex were explored (Thomas Yeo et al., 2011; Triarhou, 2007)

Metabolic PET maps including glucose uptake, oxygen consumption, cerebral blood flow and glycolytic index, as well as MEG maps from the Human Connectome Project (HCP) S900 release were acquired and parcellated using the Neuromaps package (Markello et al., 2022). Structural and functional connectivity matrices were obtained from the HCP S900 release (Vos De Wael et al., 2018). Cell type gene expression maps were produced using curated data from five single-cell RNA sequencing studies (Seidlitz et al., 2020). The relationship between metabolic pathway maps and PET maps, cell type maps, structural and functional connectivity measures were explored using Pearson correlation and partial least squares (PLS) analysis. Spatial null models and bootstrap resampling were employed to test the reliability and statistical significance of the results.

Curated gene sets were used to generate mean expression maps of glycolysis, pentose phosphate pathway (PPP), tricarboxylic acid cycle (TCA), oxidative phosphorylation (OXPHOS) and lactate metabolism. These metabolic maps generally show higher average expression in the somatosensory and motor cortices and lower expression in the limbic and visual regions (Fig.1a, b). The PPP map is different from the five other metabolic pathway maps, which show a very similar distribution to each other.

Correlation analysis showed significant positive correlations between the PPP and lactate maps and metabolic PET maps. The glycolysis and OXPHOS maps only show significant positive correlation with the glycolytic index PET map (Fig.1c).

PLS analysis of metabolic PET maps with the multi-modal brain maps, shows a general dichotomy: lactate and PPP pathways, neuronal and endothelial cells, lower MEG frequencies and functional connectivity measures spatially covary in the same direction as the PET metabolic maps of glucose, oxygen and cerebral blood flow. On the other end, the negative loadings seem to be dominated by glial cells, higher MEG frequencies and measures of structural connectivity (Fig.2).

Here, using gene expression as a proxy, we produced maps of glucose metabolism pathways and investigated their distribution across the human cortex. We show a dichotomy between PPP and maps related to glycolysis. As well, we explored the spatial alignment between glucose metabolic pathway maps and metabolic PET maps, cell type distribution, neurophysiological oscillations, and structural/functional network topology.

Connectivity (eg. functional, effective, structural)

EEG/MEG Modeling and Analysis

Multivariate Approaches ²

PET Modeling and Analysis

Cerebral Metabolism and Hemodynamics ¹

Astrocyte

Data analysis

FUNCTIONAL MRI

Multivariate

Neuron

Positron Emission Tomography (PET)

STRUCTURAL MRI

Other - Energy Metabolism, Neuroimaging Transcriptomics