Recent advances in electron microscopy have revealed that intracellular organelles do not exist independently, but rather form inter-organelle contact sites where they approach each other within 5–30 nm. Among these, mitochondria–ER contact sites (MERCS) are highly conserved from yeast to mammals and serve as critical hubs for cellular processes such as lipid biosynthesis, Ca²⁺ signaling, and the transport of various metabolites. In mammalian cells, the ER-resident protein PDZD8 has been shown to play an essential role in MERCS formation (Hirabayashi et al., Science, 2017). However, the mitochondrial protein that interacts with PDZD8 remained unknown, and the complete molecular mechanism underlying MERCS formation had not been elucidated. In this study, Dr. Aoyama and colleagues identified FKBP8 as a candidate mitochondrial protein that directly binds to PDZD8 and mediates MERCS formation.

Furthermore, electron microscopy analysis demonstrated that PDZD8 and FKBP8 act within the same pathway to fulfill an indispensable role in MERCS formation (Figure 1).

In this experiment, they hypothesized that if FKBP8 directly binds to PDZD8, then overexpressing FKBP8 on the mitochondrial outer membrane would cause PDZD8 to accumulate near mitochondria. To test this possibility, they employed 3D CLEM.

In PDZD8 knock-in HeLa cells, they overexpressed a mutated form of FKBP8 locked to be localized at the outer mitochondrial membrane and fluorescently labeled PDZD8. After cell fixation, fluorescence imaging was performed using AX with NSPARC, followed by sectioning of the same region for electron microscopy. Serial EM images were acquired using a field- emission scanning electron microscope (JSM-IT800, JEOL) (3D CLEM; details described later).

Segmentation and 3D reconstruction of the mitochondrial outer membrane and MERCS (defined as ER membranes within 25 nm of the mitochondrial outer membrane) from EM images (Figure 2, top center) revealed that the reconstructed mitochondria from eight consecutive 50-nm sections (total thickness: 400 nm) closely matched the mitochondrial signal (Venus-FKBP8) obtained by AX with NSPARC (Figure 2, yellow and green). This confirmed that the fluorescently observed region could be re-identified within the EM images.

Fluorescence microscopy allows observation of fluorescently-labeled protein localization, but the resolution is insufficient for direct visualization of nanoscale structures such as MERCS. Therefore, using CLEM to combine the advantages of light and electron microscopy is a highly effective approach for clarifying protein localization at MERCS. In this study, PDZD8 was first observed by fluorescence microscopy, and then the same field was examined by electron microscopy to identify the position of MERCS. Specifically, AX with NSPARC was used to acquire high-resolution fluorescence images, and the corresponding regions were imaged by three-dimensional electron microscopy and correlated using 3D CLEM (Figure 5). When performing 3D CLEM, if the resolution gap between electron microscopy and fluorescence microscopy is too large, it becomes difficult to cover all optical sections of the fluorescence image in the EM dataset and to accurately align the two image sets. The super- resolution capability of AX with NSPARC produces higher resolution fluorescence images compared to conventional confocal microscopes, enabling overlay with EM images using a minimal number of serial sections.

This approach allowed three-dimensional visualization of PDZD8-derived fluorescence signals overlapping with the fine structure of MERCS.