Correlative in-resin super-resolution and electron microscopy using standard fluorescent proteins

Abstract

We introduce a method for correlative in-resin super-resolution fluorescence and electron microscopy (EM) of biological structures in mammalian culture cells. Cryo-fixed resin embedded samples offer superior structural preservation, performing in-resin super-resolution, however, remains a challenge. We identified key aspects of the sample preparation procedure of high pressure freezing, freeze substitution and resin embedding that are critical for preserving fluorescence and photo-switching of standard fluorescent proteins, such as mGFP, mVenus and mRuby2. This enabled us to combine single molecule localization microscopy with transmission electron microscopy imaging of standard fluorescent proteins in cryo-fixed resin embedded cells. We achieved a structural resolution of 40-50 nm (~17 nm average single molecule localization accuracy) in the fluorescence images without the use of chemical fixation or special fluorophores. Using this approach enabled the correlation of fluorescently labeled structures to the ultrastructure in the same cell at the nanometer level and superior structural preservation.

Figure 1. Overview of sample preparation steps for correlative in-resin super-resolution and TEM imaging.

UA: uranyl acetate, TA: tannic acid.

Figure 2. The quality of SMLM imaging is dependent on addition of tannic acid to the freeze substitution medium.

(a) Comparison of average single molecule localization accuracy, nearest neighbor distances and structural resolution of EphA2-mVenus localized in the plasma membrane and the endoplasmic reticulum (ER) of HEK293T cells in resin sections for samples with different concentrations of tannic acid in the freeze substitution medium. Nearest neighbor distances have been determined considering the 20 nearest molecule positions around each position. Error bars represent standard deviations. (b–d) Example SMLM images of cells, all acquired under the same conditions, depicting typical results for different concentrations of tannic acid. Scale bars are 1 μm. (e) Detected single molecule signals per μm² and image frame during SMLM data acquisition for different concentrations of tannic acid in the freeze substitution medium. Only areas of fluorescently labeled structures were included. Each curve represents the average values of the SMLM measurements (total 57) for each concentration of tannic acid. (f) Color code for local density of detected molecules and Nyquist limited resolution in SMLM images (b–d). (g–i) Representative TEM images showing the typical quality of the ultrastructure (ER and mitochondria (MIT)) for freeze substitution with different concentrations of tannic acid. Scale bars are 0.5 μm. Note that the TEM images in (g–i) depict different cells than those shown in the SMLM images (b–d).

Figure 3. Correlative in-resin super-resolution fluorescence and EM imaging of HEK293T cells transfected with EphA2/A4 receptor proteins fused to mGFP, mVenus or mRuby2.

The first column shows conventional wide-field fluorescence images, the second column the corresponding SMLM images with color coded local densities of detected molecules and the corresponding Nyquist limited resolution (the same color code as in Fig. 2f and Fig. 4f applies). TEM images of the same cells are depicted in the third column (plasma membrane (PM), ER and the nucleus (NUC) are indicated in the individual images), followed by an overlay of the TEM images with the conventional wide-field fluorescence images, and an overlay of TEM and SMLM images. Scale bars are 1 μm. The freeze substitution for these cells was performed with 0.1% TA.

Figure 4. Comparison of resolution achieved with in-resin super-resolution CLEM.

(a) Overlay of SMLM image on the TEM image of EphA2-mVenus in a resin embedded HEK293T cell. Plasma membrane (PM), ER and the nucleus (NUC) of the cell are indicated in the image. The rectangle marks the region shown with a higher magnification in the panels on the right: (b) Conventional wide-field fluorescence microscopy, (c) Super-resolution SMLM and (d) TEM. The line profiles in (e) show the different levels of details of membrane structures resolvable with each technique. The structural resolution of ~50 nm achieved with SMLM in resin sections using standard fluorescence proteins (here mVenus) allowed a superior correlation of fluorescent signals and EM ultrastructure (a, g). Scale bar is 1 μm in (a) and 0.5 μm in (b, c, d, g). Arrow heads point to sites of endocytosis of EphA2-mVenus from the plasma membrane into vesicles. Note that the line profile for the TEM image was generated from inverted pixel values for better comparison. (f) Color code for local density of detected molecules and Nyquist limited resolution of SMLM image in (c). The freeze substitution for this cell was performed with 0.01% TA.

Figure 5. In-resin single molecule super-resolution imaging of histones.

(a) Conventional wide-field fluorescence image of H2B-GFP in a HEK293T cell. Blue dashed lines indicate the plasma membrane (PM). (b) Corresponding SMLM image showing a very detailed distribution of H2B-GFP molecules in the nucleus of the cell. (c) Magnified image of region marked with green rectangle in (b). (d) Distribution of detected single molecules corresponding to image (c). Their positions are marked by crosses in a contour map of the local molecule densities. Average single molecule localization accuracy was ~15 nm, local densities of detected molecules reach up to ~40,000 per μm² (~2,000 molecule positions per diffraction limited volume). Scale bars in (a) and (b) are 1 μm, in (c) and (d) 20 nm. (e) Color code for local molecule densities and corresponding Nyquist limited resolution.