Indirect correlative light and electron microscopy an integrated rule-based imaging approach with robust quantitative capabilities.

Abstract

Although powerful, the utility of correlative light and electron microscopy (CLEM) is severely restricted by lack of experimental throughput and quantitative capability. We present indirect CLEM (iCLEM), a rule-based imaging (RUBI) pipeline for low-cost, high throughput assessment of multiscale structure. iCLEM correlates multiscale morphometric measurements independently obtained from each technique by exploiting structural landmarks identifiable via both light and electron microscopy, sidestepping CLEM's rate-limiting steps. Exploiting landmarks robust to perturbation enables comparison between health and disease. As proof of principle, we have used iCLEM to compile a cohesive and comprehensive quantitative description of cell-cell contact structures in cardiac muscle known as intercalated discs (ID). For this, we exploited ID components, such as gap junctions, adherens junctions and desmosomes, as structural landmarks. Confocal microscopy enabled assessments on organ through cellular scales, such as cell geometry, the sizes of IDs in different parts of the heart, and the distribution of ion transport proteins between IDs and other parts of heart muscle cells. Paired sets of transmission electron micrographs at different magnifications enable quantification of ultrastructure from micro through nanoscales ranging from the size and tortuosity of the whole ID to intermembrane spacing within its different nanodomains. Next, we assessed the spatial distribution of ion transport proteins within the ID, relative to structural landmarks and each other, using super-resolution imaging (STORM single molecule localization microscopy). Molecular organization was assessed by spatial analysis of fluorescence images and machine learning-based cluster analysis of single molecule localizations. These data enabled construction of 3D finite element models of IDs populated with ion transport proteins mimicking experimental observations. Integration of these models into physiological models is revealing previously unappreciated structure-function relationships underlying health and disease.