Thermodynamics and Kinetics of Chemo-mechanical Energy Conversion
by Yixin Chen
Date of Examination:2025-09-11
Date of issue:2025-10-08
Advisor:Prof. Dr. Helmut Grubmüller
Referee:Prof. Dr. Helmut Grubmüller
Referee:Prof. Dr. Stefan Klumpp
Persistent Address:
http://dx.doi.org/10.53846/goediss-11542
Files in this item
Name:PhDThesis-YixinChen.pdf
Size:47.4Mb
Format:PDF
Description:Thesis
The following license files are associated with this item:
Abstract
English
Understanding the structure-dynamics-function linkage in biomolecular systems at a mechanistic and quantitative level is a fundamental challenge in contemporary biophysics. This thesis employs molecular dynamics (MD) simulations, Bayesian inference, and Markov modeling to address this challenge for two enzymatic systems of high biological and biomedical relevance: CRISPR/Cas13a and F1-ATPase. In the first part of this thesis (Chapter 2), I investigated the conformational dynamics of CRISPR/Cas13a along its catalytic pathway. Cas13a is an RNA-guided endonuclease that has shown great potential in medical diagnostics and genetic therapy. Despite growing interest in its applications, a mechanistic understanding of its activation and catalysis remains elusive. Here, I combined all-atom molecular dynamics (MD) simulations with DEER/PELDOR spectroscopy data (provided by our collaborators) to characterize the structure and dynamics of Cas13a from the bacterium Leptotrichia buccalis in the four known binding states, including the apo state whose structure has not been experimentally resolved. Large-scale inter-domain rearrangements and increased structural flexibility of the apo state compared to the RNA-bound states were observed in the MD simulations. These predictions are qualitatively supported by the inter-residue distance distributions extracted from the DEER/PELDOR spectra, and suggest an RNA-induced ordering mechanism for Cas13a activation. However, quantitative discrepancies between the distance distributions predicted by the MD simulations and those extracted from the DEER/PELDOR spectra are present, to which a few factors might have contributed and require further investigation. To accurately interpret the DEER/PELDOR spectra used in the Cas13a study, I implemented in the second part of this thesis (Chapter 3) a Bayesian approach for extracting distance distributions from those data. DEER/PELDOR spectroscopy measures an oscillatory time-domain signal that encodes the distribution of distances between a pair of spin centers introduced into the biomacromolecule. The state-of-the-art method for extracting the encoded distance distribution from a given DEER/PELDOR spectrum, based on Tikhonov regularization (TR), suffers from overfitting and lacks reliable uncertainty estimation. To mitigate these issues, I implemented and tested a Bayesian approach. In benchmark tests against synthetic datasets and applications to experimental spectra of Cas13a, my Bayesian approach demonstrates general applicability, high robustness, and the capability to fully quantify uncertainties, thus complementing TR to enhance the reliability of structural interpretations based on DEER/PELDOR spectra. The third part of this thesis (Chapter 4) addresses the catalytic mechanism of F1-ATPase. As the catalytic domain of ATP synthase, F1-ATPase is pivotal for mechano-chemical energy conversion in mitochondria. Despite wide acceptance of the rotary catalysis concept, a quantitative description of the thermodynamics and kinetics within the catalytic pathway of F1-ATPase remains elusive. Here we developed a minimal yet thermodynamically consistent Markov model for F1-ATPase that involves the relevant conformational and chemical degrees of freedom and reproduces all relevant experimental data. Within the generic construction of a Markov model considering discrete orientations of the γ-subunit and several conformations and nucleotide binding states of each individual β-subunit, we systematically evaluated over 14,000 model variants via exhaustive Bayesian search in the large parameter space of transition rates, cross-validation, and analysis of physical and chemical restraints. Unexpectedly, we find that a fully functional minimal model requires four distinct β-subunit conformations. Further, our model reconciles the decade-long bi-site vs. tri-site controversy, clarifying that both pathways contribute depending on ATP concentration. Additionally, our model suggests a Brownian-ratchet-like mechanism that explains the observation that one ATP hydrolysis event can trigger larger than 120◦ rotations, thereby explaining seemingly over 100% chemo-mechanical coupling efficiency. The projects presented in this thesis provide not only new insights into specific biological systems, but also generally applicable tools and frameworks. Particularly, I have extended the application of Bayesian inference to the analysis of DEER/PELDOR spectroscopy data, and to the parametrization of Markov models. Further, the methodology developed in Chapter 2 combining MD simulations and DEER/PELDOR spectroscopy could be useful for studying protein conformational dynamics with structural details at atomic resolution. Finally, the theoretical framework established in Chapter 4 for Markov modeling and hypothesis testing could enable one to study many other molecular machines whose enzymatic mechanisms involve close coupling between conformational motions, substrate binding, and chemical reactions.
Keywords: molecular dynamics simulations; Bayesian inference; Markov models; CRISPR/Cas13a; F1-ATPase; DEER/PELDOR spectroscopy
