Applying super-resolution microscopy to detect mitochondria-endoplasmic reticulum contact sites

Abstract

Membrane contact sites (MCS) are regions in the cell where the membranes of two organelles come into close proximity without fusing. Contact sites between mitochondria and the endoplasmic reticulum (MERCs) are arguably the most well-studied MCS. The study of MERCs is commonly conducted using fluorescence colocalization analysis of mitochondria and ER in diffraction limited confocal images as well as using split fluorescent probes such as the split-GFP-based contact site sensor (SPLICS). However, the inter-organelle distances of MERCs are between 10-80 nm, which is below the 200-250 nm diffraction limited resolution of traditional confocal microscopy. 3D stimulated emission-depletion (STED) super-resolution microscopy provides a roughly two-fold resolution improvement (~120 nm XY, 250 nm Z), yet is still unable to resolve MERCs. MCS-DETECT, a membrane contact site detection algorithm faithfully detects elongated ribosome-studded riboMERCs when applied to 3D STED super-resolution image volumes. Here, we expressed the SPLICSL reporter in HeLa cells co-transfected with the ER reporter RFP-KDEL and label fixed cells with antibodies to RFP and the mitochondrial protein TOM20. Using 3D STED image volumes, MCS-DETECT analysis of MERCs was compared to contacts identified via co-occurrence-based colocalization analysis of mitochondria and ER or the SPLICSL probe. MCS-DETECT found significantly smaller percent mitochondria coverage by contacts compared to either colocalization or SPLICSL and matched more closely with metrics from 3D electron microscopy (EM) analysis. Additionally, a subset of SPLICSL probe was also found to localize to mitochondria, with some puncta completely enveloped by mitochondria. Altogether, MCS-DETECT was found to identify MERCs with higher specificity than either colocalization or SPLICSL probe. We further developed and validated MCS-DETECT by applying it to analysis of various imaging modalities including 3D STED, 3D confocal, 2D STED, SoRa spinning disk, and structured illumination microscopy (SIM). In 3D STED, MCS-DETECT accurately produced MERC structures and identified larger MERCs in HT-1080 compared to COS-7 cells. However, analysis of image volumes from the other modalities produced smaller contact puncta compared to 3D STED, and often failed to capture the difference between cell lines. Altogether, this suggests that MCS-DETECT is highly specialized for the analysis of 3D STED.