Date of Award
Doctor of Philosophy (PhD)
Biomedical Engineering (ENAS)
Mutations in proteins of the cardiac sarcomere can alter muscle function, leading to a hypercontractile or a hypocontractile state of the heart. These mutational insults can also lead to disease states over time, such as hypertrophic cardiomyopathy (HCM) or dilated cardiomyopathy (DCM). One such gene, TPM1, which encodes the sarcomeric regulatory protein α-tropomyosin, has been linked to cases of both familial HCM and DCM. However, it remains unclear how different mutations to the same gene can cause different pathogenic phenotypes. Without clear predictive genotype-phenotype relationships, the value of clinical genetic testing in screening and treating families is inherently limited. This work focuses on the development of two tools to improve understanding of the connection between genotype and phenotype for sarcomeric genes: 1) a computational model of the cardiac thin filament and 2) an engineered tissue model capable of expressing arbitrary mutations through use of adenoviral transduction. By pairing these tools with other complementary methodologies (molecular dynamics and in vitro motility assays [IVMA]) we seek to demonstrate that they can form the basis of accurate classification of TPM1 variants of unknown significance (VUS) into HCM or DCM phenotypes. We first designed and implemented a Markov chain-Monte Carlo model for simulating thin filament activation. We wanted to produce a detailed model that was capable of predicting both steady-state and dynamic force production while incorporating detailed mechanisms of regulation. To do this, we investigated the regulatory mechanism of the sarcomeric protein troponin I (TnI). It was long thought the inhibitory peptide domain (IP) of TnI acted as the sole actin-binding region that holds tropomyosin in the myosin-blocking position. More recently, evidence has arisen that the C-terminal mobile domain (MD) of TnI also binds actin and may also contribute to this inhibition. To properly incorporate these findings, we created both a 16-state model with TnI-IP as the sole regulatory domain and a 24-state TnI-IP+MD version. Comparison of these models showed that assumption of a second actin-binding site allows the individual domains to have a lower affinity for actin than with IP acting alone. We also tested the 24-state model’s ability to represent steady-state experimental data in the case of disruption of either the IP or MD and we were able to capture qualitative changes in several properties as seen in experimental data. Overall, our analyses support a paradigm in which two domains of TnI bind with moderate affinity to actin, working in tandem to regulation the thin filament. To begin the characterization of mutations to TPM1, molecular dynamics simulations were used to predict important structural and mechanical changes. We applied this to two mutants: the DCM-linked M8R and the HCM-linked S215L. M8R increased flexibility of the tropomyosin chain and enhanced affinity for the blocked or inactive state of tropomyosin on actin. S215L also increased flexibility of the tropomyosin chain while enhancing affinity for the closed state of tropomyosin on actin in which myosin binding sites are revealed. Applying these molecular effects to the 24-state Markov model reproduced the shifts in calcium sensitivity, maximum force, and cooperativity that were also observed in IVMA experiments. The model was then used to simulate the impact of M8R or S215L expression on twitch behavior. These dynamic simulations predicted that M8R would reduce peak force and duration of contraction in a dose-dependent manner. To evaluate this prediction, TPM1 M8R was expressed via adenovirus in engineered heart tissues and isometric twitch force was observed. The mutant tissues showed depressed contractility and twitch duration that agreed in detail with model predictions. For S215L, simulations predicted a hypercontractile twitch phenotype. Mechanical testing of genetically engineered tissues homozygous for mutant S215L TPM1 also showed an increase in peak force and slowed relaxation when compared to isogenic WT tissues. In the final study, we characterized four TPM1 VUS using a combination of molecular modeling, IVMA, and engineered tissue. First, 20 candidate VUS were analyzed computationally using molecular dynamics and energy minimization calculations to predict each variant’s effects on TPM1 structure and association with thin filament proteins. From this analysis, four variants representing a spectrum from most to least predicted pathogenicity (A102D, D258E, K233N, and A239T) were selected for further study. Predictions were tested for each variant via engineered heart tissues. Mechanical testing of the tissues revealed an HCM phenotype for A102D and D258E, but a DCM phenotype for A239T and K233N. The pathogenic phenotypes of these selected variants reveals robust progress toward our long-term goal of computational prediction of disease risk for novel TPM1 variants.
Creso, Jenette, "Multiscale Modeling of Familial Cardiomyopathy-linked Tropomyosin Mutations" (2022). Yale Graduate School of Arts and Sciences Dissertations. 578.