Neurotransmitter and cellular translation of neuroimaging phenotypes in healthy and abnormal brain

548 

Abstract Submission 

Amir Ebneabbasi¹, Mortaza Afshani², Arman Seyed-Ahmadi³, Varun Warrier⁴, Richard Bethlehem⁵, Timothy Rittman¹

¹Department of Clinical Neurosciences, University of Cambridge, Cambridge, Cambridgeshire, ²Shahid Beheshti University, Tehran, Tehran, ³Department of Statistics, University of British Columbia, Vancouver, Canada, Vancouver, British Columbia, ⁴Autism Research Centre, Department of Psychiatry, University of Cambridge, Cambridge, Cambridgeshire, ⁵Autism Research Centre, Department of Psychiatry, University of Cambridge, Cambridge, United Kingdom

Amir Ebneabbasi  
Department of Clinical Neurosciences, University of Cambridge
Cambridge, Cambridgeshire

Mortaza Afshani  
Shahid Beheshti University
Tehran, Tehran

Arman Seyed-Ahmadi  
Department of Statistics, University of British Columbia, Vancouver, Canada
Vancouver, British Columbia

Varun Warrier  
Autism Research Centre, Department of Psychiatry, University of Cambridge
Cambridge, Cambridgeshire

Richard Bethlehem  
Autism Research Centre, Department of Psychiatry, University of Cambridge
Cambridge, United Kingdom

Timothy Rittman  
Department of Clinical Neurosciences, University of Cambridge
Cambridge, Cambridgeshire

Technological advances allow for high-resolution neuroimaging. Magnetoencephalography (MEG) and magnetic resonance imaging (MRI) measure neural oscillations and morphology that can be considered proxies of cognitive and emotional status in health and disease (Rittman, 2020; Vidaurre et al., 2018). Nonetheless, how those system-level measures relate to underlying neurophysiological processes is not completely understood (Larivière et al., 2021; Seidlitz et al., 2020). Previous work has begun associating image-derived phenotypes with neurotransmitter systems using positron emission tomography (PET) radiotracers (Hansen et al., 2022; Park et al., 2022). Here, we tested the hypothesis that functional and morphological imaging markers in healthy individuals and psychiatric disorders are associated with specific molecular and cellular processes assessed by gene expression data.

We used openly-available large-scale datasets [Human Connectome Project (HCP-MEG), Enhancing NeuroImaging Genetics through Meta-Analysis (ENIGMA), Allen Human Brain Atlas (AHBA)] and unsupervised learning to specify whether the spatial patterns of healthy neural function (frequency-specific power maps: delta, theta, alpha, beta, low- and high-gamma) and abnormal cortical structure (disorder-specific atrophy maps: deletion syndrome, autism spectrum, attention-deficit hyperactivity, high-risk psychosis, schizophrenia, bipolar, major depressive and obsessive-compulsive disorder), are co-located with topographic distributions of gene expression. Specifically, partial least square (PLS) with brain region bootstrapping and spatial permutation was leveraged to rank the predictor weights of brain genes. Then, we considered fast gene set enrichment analysis (FGSEA) (Subramanian et al., 2005) to explore whether neurotransmitter-specific gene sets (GABA, glutamate, dopamine, serotonin, histamine, aspartate, glycine, epinephrine, norepinephrine, acetylcholine) and cell-specific gene sets (inhibitory/excitatory neurons, intermediate progenitors, microglia, oligodendrocyte, radial glia, astrocytes and vascular cells) (Bhaduri et al., 2021) were within the extreme bounds of the PLS-ranked genes. We leveraged agglomerative hierarchical clustering to calculate the similarity of oscillation and atrophy maps based on the underlying molecular and cellular underpinnings. Finally, we used different gene pre-processing thresholds (differential stability: r ≥ 0.1, r ≥ 0.2, r ≥ 0.4) and multimarker analysis of genomic annotation (MAGMA) for robustness and sensitivity analysis, respectively.

Our findings showed that frequency-specific power maps mostly derive from inhibitory/excitatory neurotransmitters (GABA and glutamate) and their related neurons. In addition, disorder-specific atrophy maps are mainly co-located with serotoninergic/dopaminergic pathways. Neurodevelopmental diseases spatially intersected with immune cell expression and psychiatric disorders with genes expressed in interneuron cells. Analyses were replicated using different gene pre-processing thresholds, with Kendall's coefficients of concordance being satisfactory. We also observed that serotonin/dopamine are highlighted in the genome-wide association studies (GWAS) of psychiatric disorders.

The present study indicates that functional and structural image-derived phenotypes reflect underlying transcriptional processes. Understanding the molecular and cellular underpinnings of neuroimaging changes in this way will provide critical insights into neuropsychiatric disease for developing novel therapeutic targets and biomarkers.

Psychiatric (eg. Depression, Anxiety, Schizophrenia) ¹

Transcriptomics

Microcircuitry and Modules

Brain Atlases

Neurophysiology of Imaging Signals ²

Affective Disorders

Astrocyte

Cellular

DISORDERS

GABA

Glutamate

Histamine

MEG

MRI