Modeling Neuronal Dysfunction in Schizophrenia to Predict NeuroimagingBiomarkers

Yale University, New Haven CT

Investigators

Linked publications & trials

Abstract

? DESCRIPTION (provided by applicant): Schizophrenia (SCZ) is a severe mental illness associated with devastating symptoms that are difficult to treat due to incomplete understanding of SCZ pathophysiology. Emerging genetic and post-mortem SCZ studies have revealed altered neurochemistry and implicated N-methyl-D-aspartate receptor (NMDAR) dysfunction. However, it remains unknown how NMDAR dysfunction differentially impacts excitatory (E) vs. inhibitory (I) neurons, which both express NMDARs and may produce neural deficits in SCZ through contrasting mechanisms. In parallel, non-invasive neuroimaging studies have repeatedly identified severe neural deficits during cognitive performance and during task-free resting state. However, it remains unknown if the same underlying synaptic mechanisms can link neuroimaging findings across cognitive states and task-free states. Providing this unification is vital to bridge mechanistic knowledge gaps regarding neural states in SCZ. Findings from our recently published resting-state neuroimaging study identified increased global brain signal variability in SCZ, which we reproduced in silico, using a biologically-plausible computational model of large- scale neural systems, by perturbing E/I balance - an effect expected to occur as a consequence of NMDAR dysfunction. These observations established a parsimonious link between a computationally modeled synaptic mechanism and network-level neuroimaging findings in SCZ. Based on this effect, we hypothesized that cellular-level dysfunction in SCZ may map onto key dependent measures that are quantifiable via state-of-the- art neuroimaging across cognitive states in SCZ. This research proposal is designed to test this hypothesis via two specific aims. The first aim will extend our published model in three major ways: (i) we will explicitly model E and I neural population effects separately, to dissociate E vs. I dysfunction; (i) we will expand the model to include simulations of a cognitive task-state, allowing for a unified framework capturing cellular-level dysfunction across baseline and cognitively-active states; (iii) we will implement functional differentiation across model nodes. Thus, the model will explicitly translate biophysically plausible large-scale neuronal dynamics to simulated neuroimaging data, allowing explicit neuroimaging features to be linked to focused synaptic perturbations. The objective of the second aim is to utilize new data from an ongoing task-based and resting-state study of SCZ with matched healthy subjects, allowing for empirical neuroimaging validation of modeling predictions across mental states. We will explicitly quantify dependent neuroimaging measures predicted by the model (Aim 1) to reflect distinct cellular-level dysfunction in SCZ. Critically, the proposed dependent measures (signal variance and covariance) will be assessed in parallel across the model and empirical data, allowing for concurrent progress across both aims. This project, if successful, can help elucidate cell-specific mechanisms of synaptic deficits in SCZ, directly informing neuroimaging marker development and ultimately guiding targeted treatment toward underlying sources of SCZ neuropathology.