Three-dimensional organization of the endoplasmic reticulum membrane around the mitochondrial constriction site in mammalian cells revealed by using focused-ion beam tomography

The endoplasmic reticulum (ER) and mitochondria associate at multiple contact sites to form specific domains known as mitochondria-ER associated membranes (MAMs) that play a role in the regulation of various cellular processes such as Ca2+ transfer, autophagy, and inflammation. Recently, it has been suggested that MAMs are also involved in mitochondrial dynamics, especially fission events. Cytological analysis showed that ER tubules were frequently located close to each other in mitochondrial fission sites that accumulate fission-related proteins. Three-dimensional (3D) imaging of ER-mitochondrial contacts in yeast mitochondria by using cryo-electron tomography also showed that ER tubules were attached near the constriction site, which is considered to be a fission site1). MAMs have been suggested to play a role in the initiation of mitochondrial fission, although the molecular relationships between MAMs and the mitochondrial fission process have not been established. Although an ER-mitochondrial membrane association has also been observed at the fission site in mammalian mitochondria, the detailed organization of MAMs around mammalian mitochondria remains to be established. To visualize the 3D distribution of the ER-mitochondrial contacts around the mitochondria, especially around the constriction site in mammalian cells, we attempted 3D structural analysis of the mammalian cytoplasm using high-resolution focused ion-beam scanning electron microscopy (FIB-SEM) tomography, and observed the distribution pattern of ER contacts around the mammalian mitochondrial constriction site. Rat hepatocytes and HeLa cells were used. Liver tissue was obtained from male rats (Wistar, 6W) fixed by transcardial perfusion of 2% paraformaldehyde and 2.5% glutaraldehyde in 0.1 M cacodylate buffer (pH 7.4) under deep anesthesia. HeLa cells were fixed with the same fixative. The specimens were then stained en bloc to enhance membrane contrast and embedded in epoxy resin2). The surface of the specimens was freshly exposed using an ultramicrotome and examined by FIB/SEM (Quanta 3D FEG, FEI, USA). Ion-beam milling and image acquisition cycles were performed under the following conditions. The milling was performed with a gallium ion beam at 30 kV with a current of 100 pA, with a milling pitch of 10 nm/step. Material contrast images using backscattered electrons (BSE) were acquired at a landing energy of 2 keV with a bias voltage of 1.5–2.5 kV using a vCD detector. The remaining acquisition parameters were as follows: beam current = 11 pA, dwell time = 6–30 µs/pixel, image size = 1024 × 883 pixel (5.9 × 5.1 µm), pixel size = 5.8 nm/pixel. The resultant image stack was processed using Avizo 6.3 and Amira 5.4(FEI, USA). Reconstructed volume showed the existence of several constriction sites on mitochondria in both chemically fixed normal hepatocytes and HeLa cells. Each material contrast image of specimen surfaces showed two types of membrane associations between the ER and mitochondria. The first was an osmiophilic bridge-like structure; these bridges were approximately 50 nm in length, and they connected the ER membrane and the mitochondrial outer membrane (OMM). The second was a close apposition (< 20 nm) of the ER membrane and the OMM. Membrane segmentation revealed the 3D distribution of the membrane contacts; 10 to 20% of the mitochondrial surface was occupied by ER contacts. No fundamental difference was observed between hepatocytes and HeLa cells in the distribution pattern of the contacts. Although ER-contacts and bridge-like structures were occasionally found to accumulate around the mitochondrial constriction area, we did not observe any ring-like ER tubules around the mammalian mitochondrial constriction site, as in yeast. These results suggest that the role of ER-membrane associations in the mitochondrial fission process may differ between mammals and yeast.