Engineering Dynamic Behavior into Nucleic Acids Guided by Single Molecule Fluorescence Microscopy

Single Molecule Fluorescence Microscopy;DNA Nanotechnology;Autonomous DNA Walker;Kinetic Fingerprinting;Chemistry;Science;Chemistry

Li, JiemingKennedy, Robert T ;

University of Michigan

关键词: Single Molecule Fluorescence Microscopy;    DNA Nanotechnology;    Autonomous DNA Walker;    Kinetic Fingerprinting;    Chemistry;    Science;    Chemistry;   

Others  :  https://deepblue.lib.umich.edu/bitstream/handle/2027.42/149970/jmli_1.pdf?sequence=1&isAllowed=y

瑞士|英语

来源: The Illinois Digital Environment for Access to Learning and Scholarship

【 摘 要 】

Single-molecule fluorescence microscopy is a powerful technique that has been used for investigating the structural dynamics of biomolecules, and is particularly useful when ensemble averaging might obscure detailed information of the system under investigation. One application of single molecule measurement is to optimize the design of DNA nano-devices. Dynamic DNA nanotechnology has yielded nontrivial autonomous behaviours such as stimulus-guided locomotion, computation, and programmable molecular assembly. Despite these successes, DNA-based nanomachines suffer from slow kinetics, requiring several minutes or more to carry out a handful of operations. In this thesis, I have pursued the speed limit of an important class of reactions in DNA nanotechnology—toehold exchange—through the single-molecule optimization of a novel class of DNA walker that undergoes cartwheeling movements over a field of complementary oligonucleotides. I identified the walking mechanism by single-molecule fluorescence resonance energy transfer (smFRET) measurement, with the stepping rate constant approaching 1 s-1, which is 10- to 100-fold faster than prior DNA walkers. I also used single-particle tracking to demonstrate movement of the walker over hundreds of nanometers within 10 min, in quantitative agreement with predictions from the stepping kinetics. These results suggest that substantial improvements in the operating rates of broad classes of DNA nanomachines utilizing strand displacement are possible.Another application of single molecule measurements is kinetic fingerprinting detection. Conventional methods for detecting small quantities of nucleic acids require amplification by the polymerase chain reaction (PCR), which necessitates prior purification and introduces copying errors. While amplification-free methods do not have these shortcomings, they are generally orders of magnitude less sensitive and specific than PCR-based methods. In this thesis, I review important experimental tips and data analysis details to provide a practical guide to a novel amplification-free method, single-molecule recognition through equilibrium Poisson sampling (SiMREPS), that provides both single-molecule sensitivity and single-base selectivity by monitoring the repetitive interactions of fluorescent probes with immobilized targets. In addition to demonstrating how this kinetic fingerprinting filters out background arising from the inevitable nonspecific binding of probes, yielding virtually zero background signal, I also investigated the detection of epigenetic mutations such as CpG methylation using SiMREPS.The analysis of single-molecule microscopy data can be very time-consuming because there is no sufficiently robust automatic method for selection of qualified single-molecule fluorescence trajectories from the generally noisy and heterogeneous raw data, necessitating manual trace selection that can take hundreds of hours for large datasets. In this thesis, I discuss the innovative use of the popular convolutional neural network AlexNet and the recurrent neural network Long Short-Term Memory (LSTM) to develop an automatic selector for single-molecule fluorescence resonance energy transfer (smFRET) traces. The average prediction accuracy is above 90% when tested on datasets from different users and experimental systems. To boost the selection accuracy and increase the diversity of training datasets, simulation data were included into the training data set and tested for selection accuracy. I expect that this new method will not only greatly expedite analysis of smFRET data and increase analysis reliability of SiMREPS data, but also introduce and validate machine learning as an effective tool for analysis of single-molecule microscopy data more generally.Together, these results provide new insights into how single molecule microscopy can be used to engineer dynamic behaviors of nucleic acids.