Correlative imaging integrates electrophysiology with three-dimensional murine heart reconstruction to reveal electrical coupling between cell types

Figures

• 

  Fig. 1. Functional characterization of AP kinetics and conduction. a, A CTRL and a DSG2 mouse heart during optical mapping. Scale bars, 2 mm. Hearts were electrically paced at the apex (yellow lightning bolt). Fluorescence signals (∆F/F) were extracted from the LV (black region of interest (ROI), CTRL; red ROI, DSG2) during a burst of 15 stimuli at LF (5 Hz) and HF (≥10 Hz). Freq, frequency. b, Average AP upstroke kinetics (TTP) and AP duration at 50%, 70% and 90% repolarization (APD50, APD70 and APD90, respectively) in the LV free wall of CTRL (black) and DSG2 (red) hearts. c, LV activation maps of a CTRL and a DSG2 heart during pacing at LF and HF, showing deformation of the advancing wavefront at HF pacing in the DSG2 heart. Activation time is color coded, and the black lines represent isochrones with 1-ms resolution. d, Average conduction velocity (CV, m s−1) and wavefront deformation (as angular dispersion, in degrees) in CTRL and DSG2 LV free walls. Data were collected from nine CTRL and nine DSG2 mouse hearts and are reported as means ± s.e.m. Two-way repeated-measures analysis of variance (ANOVA) was performed with Tukey post hoc comparison. Source data

• 

  Fig. 2. 3D reconstruction of myocardium and collagen in cleared hearts. a, Optical clearing of entire mouse hearts by a cardiac-optimized SHIELD protocol. CTRL and DSG2 hearts are shown before (in 4% paraformaldehyde (PFA)) and after clearing (in EasyIndex solution). H&E histological inspection of cleared tissue is shown for both samples. b, Detection scheme of the mesoSPIM microscope, based on a custom-designed telecentric lens. Emission light from a FOV of 10.4 mm × 10.4 mm is collected by a ×2 telecentric lens and imaged on the camera sensor, working in a rolling-shutter modality that is synchronized with the position of the light-sheet focal beam as it advances through the tissue. c, LV optical section of a cleared DSG2 heart: myocardium fluorescence (in gray) and scattering signal (in red) were imaged with a voxel size of 3.25 × 3.25 × 3.1 µm3 in XYZ. Inset 1: the cellular organization of the myocardium. Inset 2: the extracellular matrix distribution inside a fibrotic area of the LV wall. These features were consistently observed in all analyzed samples. d,e, 3D reconstructions of entire cleared mouse ventricles (d, CTRL; e, DSG2) acquired with the mesoSPIM setup by detecting the…

• 

  Fig. 3. Organ-scale reconstruction of myocardium, fibrosis and 3D cardiomyocyte organization. a, 3D reconstruction and segmentation of a DSG2 heart: myocardium (MYO, gray) was segmented by fluorescence signal, CF (black) and NCF (green) from scattered light. A transversal plane is shown, with the outer profile of CF and NCF highlighted in red and black, respectively. b, Percentage of tissue classified as CF and NCF with respect to the total tissue volume in CTRL and DSG2 hearts. Data were collected from nine CTRL and nine DSG2 mouse hearts and reported as means ± s.e.m. Two-way repeated-measures ANOVA performed with Tukey post hoc comparison was applied. c, Schematic of the image analysis workflow developed to map the 3D cellular orientation (isotropic spatial resolution of 96 μm) from muscle fluorescence signal (isotropic voxel size of 6 μm) in whole-ventricle reconstructions. Cellular architecture of a DSG2 mouse heart is shown. Inset: a virtual slice (thickness, 100 μm), where helix angle distribution of locally prevailing cardiomyocyte orientation is color coded. LV ENDO, left ventricular endocardial surface; EPI, epicardial surface. d, Example of a reconstruction of a DSG2 heart…

• 

  Fig. 4. Morpho-functional correlation between conduction and fibrosis distribution. a, Registration pipeline between functional and structural data: the ventricular shape (in yellow) is extracted from optical mapping images. 3D mesoSPIM-based tomographs are automatically oriented to match this shape (dashed gray outline). The 2D functional data (activation maps) are then mapped onto the appropriately rotated 3D mesh. b, Ventricular activation maps of Fig. 1c correlated with CF (first row, in black) and with NCF (second row, in green) surrounding CF. Isochrone resolution, 1 ms. Only the fibrosis found in the frontal free wall is shown, and its depth is indicated by color intensity. c, Transmural fibrotic area in the transversal section of the DSG2 heart shown in b, reconstructed with two-photon microscopy (black, collagen; magenta, Cx43). Cx43 punctae are visible in myocardial tissue surrounding fibrosis (1) and in strands of cells inside CF (2) (observed in two DSG2 hearts).

• 

  Fig. 5. Impact of fibrosis on conduction and arrhythmogenesis during extra-beat electrical stimulation. a, Frames of optical mapping (ΔF/F0) of one sinus beat (number 1), two stimulated extra beats (number 2 and number 3, 112 ms and 33 ms after the preceding sinus beat, respectively) applied to the right ventricular outflow tract (yellow and red lightning bolts) of a DSG2 heart during an arrhythmia induction protocol. The red panel shows a reentrant arrhythmia induced by trigger number 3. Activated tissue (during AP) is indicated in red and resting tissue (excitable) in green. Structural reconstruction of CF (white profile outlines) is superimposed. b, Fluorescence signal (extracted from the ROI shown in a, red square), showing transmembrane potential dynamics during the extra-beat stimulations in a. Colored rectangles indicate time windows selected for representative frames in a. Below: black arrows indicate the time interval between last sinus beat and the extra trigger. Extra-beat number 3 gave rise to a non-sustained ventricular tachycardia that revolves around the fibrotic region. c, Activation maps (isochrones resolution of 1 ms) of the same DSG2 heart shown in a upon apical pacing…

• 

  Fig. 6. Integrated computational modeling of AP propagation. a, Activation maps measured via optical mapping during apical pacing at LF (5 Hz) and HF (15 Hz) in 1 CTRL and 1 DSG2 heart were projected onto the corresponding 3D structural data. b, Electrical propagation during apical pacing at LF and HF, simulated using a functional model of electrical activity in the same hearts as in a. For each sample, structural information was integrated into a biventricular geometrical model (spatial resolution, 100 μm) at 3 progressive levels: a theoretical cardiomyocyte distribution (rule-based), actual cardiomyocyte organization (image-based) and with the addition of NCF distribution. 3D activation maps of the last beat (after ten pulses at LF and HF) are shown. The ventricular cavity orifice is displayed in blue for clarity.

• 

  Fig. 7. Morpho-functional registration between fibrosis and electrophysiological remodeling. a,c,e, Differential maps indicating variations in APD90 (a), local conduction time (LCT; c) and TTP (e) upon pacing at the apex at LF and HF (shown as HF minus LF), for one CTRL heart and one DSG2 heart. Functional maps are mapped onto the 3D mesh, and contours of CF and NCF areas are highlighted in black and green, respectively, for better visualization. b,d,f, Spatial heterogeneity (in ms) calculated from APD90 (b), LCT (d) and TTP (f) differential maps, measured from two angles to ensure mapping of the entire LV of each heart (two points per heart are reported). g, Quantification of Cx43 punctate area in remote (inset 1 from Fig. 5c) and fibrotic regions (inset 2) of DSG2, and compared with CTRL. Data are reported as means ± s.e.m. Data in b, d and f were collected from nine CTRL and nine DSG2 mouse hearts. A two-way repeated-measures ANOVA was performed with a Tukey post hoc comparison. Data in g were collected from two CTRL (ten randomly selected areas) and two DSG2 mouse hearts (ten randomly selected areas from remote and ten from fibrotic areas each). Data were analyzed using the…

• 

  Fig. 8. Simulation of local electrophysiological alterations in NCF. Simulations of apical pacing at LF and HF were performed on a structural model of a DSG2 heart, incorporating image-based fiber orientation and percolation-derived NCF (as in Fig. 6a, third row). Electrophysiological alterations were introduced within the fibrotic region (NCF). a, Reduced tissue conductivity by 75%, preserving normal anisotropy. b, Further reduced longitudinal conductivity achieving isotropy. c, Mimicking decreased sodium channel availability by reducing peak sodium conductance by 75%. d, Homogeneous addition of electrically coupled fibroblasts (1 myocyte connected to 12 fibroblasts). e, Combined condition: isotropic reduced conductivity, sodium current reduction and fibroblast coupling. f, Modeling of a 1D chain of electrotonically coupled myocytes and fibroblasts. Examples for successful (left; pacing rate of 8.5 Hz) and failing (right; pacing rate of 11.5 Hz) AP propagation between ventricular cardiomyocytes (red APs) via an insert of electrically non-excitable fibroblasts (green; here n = 5 cells). Signals are shifted for visualization; thus, axis labels do not indicate absolute values. g, Maximum…

• 

  Extended Data Fig. 1. Left-ventricular maps of all included control (CTRL) hearts. Hearts were paced at LF (5 Hz; left panel) and HF ( ≥ 10 Hz; right panel). Activation time (Act) is color-coded, with isochrones of 1 ms. Time-to-peak (TTP), AP duration at 50%, 70%, and 90% of repolarization (APD50, APD70, APD90, respectively) are color-coded with the scale indicated above. Average data obtained from these maps are included in Fig. 1.

• 

  Extended Data Fig. 2. Left-ventricular maps of all included DSG2 hearts. Hearts were paced at LF (5 Hz; left panel) and HF ( ≥ 10 Hz; right panel). Activation time (Act) is color-coded, with isochrones of 1 ms. Time-to-peak (TTP), AP duration at 50%, 70%, and 90% of repolarization (APD50, APD70, APD90, respectively) are color-coded with the scale indicated above. Average data obtained from these maps are included in Fig. 1.

• 

  Extended Data Fig. 3. Resolving cellular orientation in the entire heart. High-resolution reconstructions by fluorescence signal of cardiac muscle of a cleared healthy heart, collected by mesoSPIM. (a, b) The long axis (LAX) plane and Short Axis (SAX) plane sections are shown. Main cellular orientation is well resolved in both LAX and SAX views, enabling detection of cellular alignment following the helical angle distribution in the myocardial wall (highlighted in c, e, g, and h panels). Cells can be resolved also in the trabeculae inside the cardiac chamber (panels d, f). Pixel size: 3.25 µm × 3.25 µm. Scale bar: 1 mm (left panels), 200 µm (C-H inserts).

• 

  Extended Data Fig. 4. MesoSPIM tomography of a cleared mouse heart. A representative whole-heart reconstruction of a healthy mouse heart performed following the optimized SHIELD clearing protocol and the MesoSPIM-based imaging protocol. For each heart, two identical reconstructions were acquired, both with and without the long-pass filter (λ > 650 nm) to collect tissue fluorescence (here represented with a grey signal) and elastic scattering (red signal), respectively. Original images voxel size: 3.25 µm × 3.25 µm × 3.1 µm; two-channels whole-heart tomography voxel size: 6 µm × 6 µm × 6 µm.