Typically, quantitative immunogold EM requires the decoration of sections with antibodies, resulting in relatively few gold particles per decorated section. To determine the suborganellar distribution of a specific protein with this approach, numerous individual gold localizations are recorded on many images and an average protein localization is determined [4 and 5]. Hence immunogold EM is usually not selleck compound suited to study protein distribution in individual mitochondria. Fluorescence microscopy is arguably the most suitable approach to study the distribution of proteins in single mitochondria [6]. However, studies using conventional fluorescence microscopy to investigate

protein localizations in these organelles ultimately face the challenge that mitochondria are small; the width of mitochondrial tubules is typically between 250 and 500 nm [7, 8 and 9]. In conventional (confocal) microscopes diffraction limits the achievable resolution to ≥200 nm in the lateral plane and to ≥500 nm in the axial direction [10]. Hence the size of most mitochondria is just at the resolution limit of optical microscopy making the analysis of submitochondrial protein distributions always challenging and often entirely impossible using diffraction limited optical microscopes [11, 12, 13, 14 and 15]. Over the last decade several super-resolution microscopy (nanoscopy) concepts have C646 in vivo been devised that allow diffraction-unlimited optical resolution.

All concepts that fundamentally overcome the diffraction limit exploit a transition between two fluorophore states, usually RG7420 in vivo a fluorescent (on-) and a non-fluorescent (off-) state in order to discriminate adjacent features. Depending on how the transition is implemented, the current super-resolution methods may be assigned to one of two classes, namely coordinate-targeted (prominent approaches: STED [16 and 17], SPEM/SSIM [18 and 19] and RESOLFT [20, 21 and 22]) and coordinate-stochastic approaches (PALM [23], STORM [24], FPALM [25], GSDIM [26], dSTORM [27], and others). The various methods routinely provide

optical resolution well below 50 nm (i.e. they fundamentally overcome the diffraction barrier), have been implemented with more than one color, and 3D versions are available. The underlying physical concepts as well as the practical differences between the approaches have been expertly reviewed elsewhere [28•, 29• and 30]. To evaluate what can be expected when imaging mitochondria with conventional diffraction-limited microscopy or diffraction-unlimited nanoscopy, we simulated three simplified models that should reflect differently labeled mitochondria (Figure 1): a mitochondrion with regularly stacked cristae (crista to crista separation is 100 nm), as often seen in EM images [31••] where only the cristae are labeled (Figure 1b). A helical structure circumventing the matrix, which might resemble a postulated mitoskeletal element [15] (Figure 1c). Randomly distributed proteins in the outer membrane (Figure 1d).