Determining the Dynamics and Function of the Cardiac Thin Filament

Abstract

The cardiac thin filament (CTF) is known as the regulatory unit of cardiac muscle. Point mutations to various proteins of the CTF complex have been shown to change both their structure and dynamics and often lead to cardiomyopathies and possibly death. While the link between these point mutations and cardiomyopathies has been established, the underlying mechanism of how the disease is induced is still unknown. Additionally, there is still a gap in our understanding of the mechanism for healthy cardiac muscle function. The work presented in this dissertation uses computational methods with a complete atomistic model of the cardiac thin filament to address several of these shortcomings in an effort to better understand cardiac muscle and cardiomyopathies. The first project presents a new method for predicting the pathogenicity of variants of unknown significance (VUS) on cardiac troponin T (cTnT) and tropomyosin (Tm). This method compares structural changes of the VUS to well-defined pathogenic mutations. Computational results of a subset of mutations were verified via differential scanning calorimetry to develop an experimental correlation. Potential reclassifications for nine VUS were suggested based on similarities in computational observables with pathogenic mutations. The second project studies two different transitions in the cardiac thin filament during the contraction cycle, the opening of the cTnC N-lobe and the dissociation of the cTnI C-terminal domain from actin. The free-energy surfaces of both transitions were calculated using the enhanced sampling technique metadynamics. Important intra- and interprotein interactions that contribute to free-energy barriers were determined from the simulations. The third project studies the potential use of an engineered variant, L48Q on cTnC, to reverse the mutational effects of a dilated cardiomyopathy (DCM) linked genetic mutation, D230N on Tm. Changes in the structure of the troponin core were compared for D230NTm alone, and the D230NTm and L48QcTnC double variant. The fourth project is an ongoing study to create a machine learning algorithm to predict the pathogenicity of VUS and classify them as either benign, hypertrophic cardiomyopathy (HCM), or DCM. The current algorithm being utilized is a convolutional neural network. This project aims to create a fully automated process for predicting pathogenicity and resulting phenotypes specific to the cardiac thin filament. The work presented in this dissertation provides methods that can also be extended to study other sarcomeric proteins or biological systems.