Fig 1.
Mitochondrial network topology is preserved in high-framerate 4D fluorescence microscopy data.
a, Representative 4D (3D+time) lattice light-sheet microscopy data of a hiPSC colony labeled with MitoTracker (mitochondria, green) and expressing CAAX-RFP (plasma membrane, red). b, Individual cells in the colony are segmented based on the plasma membrane marker. c, Mitochondria fluorescence signal in a single cell is segmented using MitoGraph. d, Mitochondrial network skeleton dynamics over 5 min every 6.4s (time red to purple). e, Mitochondrial fluorescence density and segmented network skeleton are overlaid and shown for frame numbers 0, 1, 2, 20 at frame interval 3.2s. f, Scale-invariant feature transform (SIFT) maps image features for two frames separated by 3.2s (top), and 64s (bottom). The image size for the mapped region is 1034px by 642px. g, Pixel deviation between SIFT-mapped feature locations at different time intervals.
Fig 2.
Algorithm design and in-silico validation of 4D mitochondrial network tracking.
a, Discretized nodes along the segmented mitochondria skeleton serve as the basis for network tracking. Terminal, branching, and bulk nodes are treated equally and termed skeleton nodes. Cloud: fluorescence density; sphere: skeleton node. b-c, Cost terms used for the linear assignment problem (LAP) formulation of node tracking. Spatial proximity is measured as distances between nodes within two consecutive frames. Topology cost is computed using a graph comparison that assigns low cost for similar local topology. d, LAP formulation of node tracking for the mitochondrial network. From left to right: 1) pairwise distance matrix for nodes at frames T and T+1; 2) thresholds to eliminate nodes too far to be tracked; 3) spatial separation and network topology constraints; 4) the solution to the LAP yields the tracking results as linked node pairs, along with terminated and initiated nodes. e, Two consecutive frames of a reaction-diffusion mitochondrial network simulation with representative fusion (cyan) and fission (magenta) positions pointed out by the arrows. f, Temporal network tracking for the simulated mitochondria for two consecutive timepoints (black: correct arrow, red: incorrect arrows, blue: incorrect arrows at the fusion/fission sites). g, Magnification of the example in-silico fusion (cyan) and fission (magenta) events in e). h, The y-axis denotes tracking precision which is the percentage of nodes that are correctly tracked (based on simulation ground-truth). The x-axis denotes average node mean squared displacement (MSD) per frame, which is linearly proportional to the frame interval according to MSD = 6Dτ. MSD is computed by converting simulation units to real world units so that it can be comparable to experimental data. Tracking is then performed for three scenarios: 1) red: tracking on simulations without fusion/fission using distance cost only (similar curve for simulation with fusion/fission); 2) green: tracking on simulation without fusion/fission using distance and topology costs; 3) blue: tracking on simulation with fusion/fission using distance and topology costs.
Fig 3.
In-vitro validation and evaluation of 4D mitochondrial network tracking.
a, LLSM volumetric snapshot of a segmented cell. Green: mitochondrial network. Red: plasma membrane. b-c, Fluorescence signal and segmented network skeleton are overlaid for two consecutive frames (blue: 0s, red: 3.2s). d, Left: tracking of the skeleton nodes for the two frames in b) and c) are visualized by black arrows. Right: zoom-in to a representative region tracked over 12.8s. The skeletons are colored in blue, red, green, purple and orange in the order of time. e-g, Tracking of three representative mitochondrial network fragments for 32 seconds (time blue to red). e) A small fragment displays twisting motion. f) A medium-size fragment displays inward motion. g) A large fragment displays complex motion patterns.
Fig 4.
Mitochondrial network motility analysis.
a-c, Top: Mitochondrial nodes are colored by diffusivity at node, segment, or fragment levels from high (red) to low (blue) diffusivity. Bottom left: distribution of diffusivity values, bottom right: linking vectors compared to a fixed reference vector. d-e, Spatial structure of somatic mitochondrial network colored by mitochondria segment diffusivity in control and oligomycin-treated cells. f, Kernel-smoothed distribution of segment diffusivity for 2552 segments in control cells (blue), and 2376 segments in oligomycin-treated cells (red). T-test is performed with p-value = 2.8e-298. g, Spatial tracking vector correlations between neighboring nodes. Left: illustration. Right: heatmap of vector correlation for segment nodes (columns) at different timepoints (rows). h, Temporal tracking vector correlation for the same node at consecutive frames. Left: illustration. Right: heatmap of correlation values for segment nodes (columns) at different timepoints (rows). i, Violin plot of the spatial and temporal correlation values between control (blue) and oligomycin-treated (red) cells.
Fig 5.
Mitochondrial network remodeling analysis.
a, Representative snapshots of a tracked fusion event (left), and fission event (right). b, Node diffusivity values are significantly lower for randomly selected nodes (green) as compared to nodes undergoing fusion (cyan) and fission (magenta). Student’s t-test are applied, and p-values are 6.565e-25 and 1.237e-24 for random vs. fusion nodes and random vs. fission nodes, respectively. c, Representative tracking of mitochondrial fragments in hiPSCs over three timepoints where each fragment is uniquely colored. d, Analysis of fission/fusion preferences with respect to fragment size for control (blue) and nocodazole-treated condition (green). Normalized network size difference measures the differences in # nodes for the fragments undergoing remodeling events, normalized by # nodes of the larger fragment. Mann-Whitney test is used to test whether the underlying distribution is the same, with p-values = 1.2e-32.
Fig 6.
Temporal characteristics of mitochondrial network remodeling, flux, and damage resilience.
a, Temporal networks display node and edge dynamics that have an influence on network transport and resilience (newly added or removed nodes/edges highlighted in red). b, Mean degree difference between control, oligomycin, and nocodazole. c, Temporal intersection between control, oligomycin, and nocodazole. d, Global network reachability in a representative somatic mitochondrial network depicted as a color gradient (dark: high reachability, light: low reachability). e, Global network reachability where the top 5% highest betweenness-centrality nodes were removed. f, Mean normalized global reachability in different drug conditions. The three violin plots for each condition correspond to no nodes removed (left), 5% random nodes removed (middle), and 5% most connected nodes removed (right). g,h, Correlation of network reachability with node MSD and temporal intersection.