Understanding dynamics of brain network in TDP-43 related neurodegeneration

Mislocalization of TDP-43 is a common pathological feature of several neurodegenerative diseases, including
Alzheimer’s disease, amyotrophic lateral sclerosis, and frontotemporal dementia. Many studies support the
loss-of-function mechanism caused by TDP-43 nuclear clearance as a major pathogenesis pathway toward
neurodegeneration. It is of great importance to understand the dynamics of disease progression in TDP-43
knockout mouse models.
Functional microcircuits are at the heart of the information processing capability of the brain. Calcium imaging
is a powerful tool to study functional microcircuits. Calcium imaging can generate high-dimensional longitudinal
datasets, which are collected at multiple time points over a temporal process (each observation time point is a
wave). Longitudinal analysis models the evolving temporal process. However, existing analysis methods are
primarily designed to process cross-sectional data, which provides a static view of the brain network. As a
result, existing methods have limited capability to model complex dynamic network patterns for high-
dimensional data. Lack of advanced longitudinal analysis methods is a bottleneck for using calcium imaging to
study the dynamics of brain networks in TDP-43 knockout mouse models.
This project seeks to develop a Bayesian computational system to model longitudinal functional microcircuits
and use it to examine microcircuit changes in a TDP-43 knockout mouse model. The developed system is
referred to as Bayesian Longitudinal Microcircuit Analysis (BLMA). The Specific Aims are: Aim 1. Develop a
computational system for longitudinal microcircuit modeling. The proposed system, BLMA, will include these
components: preprocessing, microcircuit construction, feature extraction, and Bayesian multivariate mixed
modeling. Aim 2. Understand microcircuit changes in a TDP-43 knockout mouse model. We will apply BLMA to
an existing longitudinal calcium imaging dataset of pyramidal neurons of the prefrontal cortex (PFC) from
awake behaving Tdp-43F/F mice. We will compare the control and knockout groups to determine whether the
knockout group exhibits abnormal PFC microcircuit trajectories.
This project will develop BLMA to advance the state-of-the-art in data analysis and modeling for longitudinal
calcium imaging. It leverages Bayesian machine learning to address critical challenges in longitudinal calcium
imaging data analysis: shared information across waves and high dimensionality. This project is innovative
because it will develop a novel Bayesian system to model microcircuit changes based on calcium imaging data
and delineate a unique brain network mechanism leading to TDP-43 related neurodegeneration. At the
completion of this project, we will have delineated a unique brain network mechanism in TDP-43 related
neurodegeneration.

Grant Number: 5R21AG085492-02
NIH Institute/Center: NIH
Principal Investigator: RONG CHEN