Quantifying & Modeling Mitochondrial Dynamics
Agent-Based Mitochondrial Dynamics Simulation (MiDyS)
Our agent-based mitochondrial dynamics simulation (MiDyS) is currently in development. We are utilizing the Netlogo simulation environment for development of both a 2D and 3D model of intracellular mitochondrial dynamics. Model development and optimization is being informed by live cell recordings of primary cells from MitoTimer reporter mice at basal states and under pharmacological and genetic manipulations.
Given that mitochondria perform numerous dynamic behaviors to preserve the overall system, we believe that agent-based modeling is an effective approach to capture the complex processes of mitochondrial dynamics. We are in the process of developing and optimizing a predictive computational model in which mitochondrial units act as agents with the ability to undergo fusion, fission, growth, and degradation.
Above: Representative live recordings of MiDyS in a 2D environment. Each circular object represents one mitochondrial unit. Blue objects are "healthy" and "functional" units, while red units are "dysfunctional" based on extreme intrinsic properties (e.g. mitochondrial membrane potential). Links between units represent the networking behavior of mitochondria via fusion.
Right: Representative live recording MiDyS in a 3D environment. Behaviors (e.g. fission, fusion, degradation) of individual mitochondrial units and networks can be observed throughout the simulation
Machine Learning-Based Mitochondrial Morphological Classification
Mitochondrial morphology is a representative snapshot of the underlying processes of mitochondrial dynamics and quality control. Morphological analyses of mitochondria have traditionally been performed using qualitative approaches or simple quantitative measurements (e.g. mitochondrial length).
Utilizing free and opensource programs and plug-ins, we have developed a semi-automated machine learning-based analysis pipeline for quantifying mitochondrial morphology. We have demonstrated our pipeline's utility in cell culture (2D) and tissue histology (3D) using immunofluorescent labeling, as well as ultrastructural 3D imaging via serial block-face scanning electron microscopy (SBF-SEM).
R functions created for machine learning model training and utilization, as well as sample images, are available in our GitHub repository: https://github.com/sanderson-lab/mitomorphology
Representative image of mouse primary cortical neurons with mitochondria immuno-labeled for ATP Synthase (green) and TOM20 (red). Insert: segmented mitochondrial objects color coded by morphological classification.
Representative 3D reconstructions of individual mitochondrial objects from each morphological phenotype (mouse hippocampus imaged via IF confocal microscopy)
3D reconstruction of mitochondrial objects auto-segmented from 100 70nm SBF-SEM slices (rat hippocampus)
3D Reconstruction (magenta) of mitochondrial objects from serial block-face scanning electron microscopy (SBF-SEM). Video moves through 70 nm SEM slices, overlays mitochondrial objects segmented via trainable Weka segmentation plug-in, and then reconstructs back through the SEM block.
Relevant Publications:
Fogo GM, Anzell AR, Maheras KJ, Raghunayakula S, Wider JM, Emaus KJ, Bryson TD, Bukowski MJ, Neumar RW, Przyklenk K, Sanderson TH. (2021). Machine learning-based classification of mitochondrial morphology in primary neurons and brain. Sci Rep. 11(1): 5133.
For a complete listing of publications: click here