Preparation protocol for multimodal imaging of biological tissue samples by light and electron microscopy. A simple preparation protocol for multimodal imaging of biological tissue samples by light and electron microscopy. 

Routine approaches to correlative investigation of tissue samples normally use samples differently prepared for light (LM) and electron microscopy (EM). To overcome the handicap of data derived from differently prepared samples and to introduce the superior structural preservation of HPF into LM investigations, we developed a new freeze-substitution protocol. Such prepared samples turned out to be optimal for both, investigating complex tissue structures with confocal light microscopy as well as describing ultrastructure and constituents of the same identical cells with electron microscopy.

Structural investigation of tissue biopsies requires the coupling of optimal structural preservation with detailed information collected at the light and electron microscopic level. Unfortunately, although cryo-immobilisation by high-pressure freezing provides the best structural preservation, it is used routinely only for EM investigations, while for light microscopy chemical fixation protocols have been established. These chemically invasive fixation protocols have the drawback of introducing unpredictable fixation artefacts. Therefore, comparative histological (i.e. light microscopic) and ultrastructural (i.e. electron microscopic) results are usually obtained from parallel samples that have not been prepared identically and never by examining exactly the same features. Finally, finding an area of interest for EM investigation within a complex tissue is like searching for a needle in a haystack.
To overcome these handicaps, we modified the well-established freeze-substitution technique (FS), allowing the investigation of resin-embedded cryo-immobilised tissue by confocal laser scanning microscopy (CLSM) prior to EM examination. Thus, (1) selected cells can be depicted throughout the whole tissue block by CLSM and (2) subsequently selectively prepared by targeted sectioning for follow- up investigation of the identical structure by TEM.

Sample Preparation
Samples of native human skin were taken using standard biopsy punches (2 mm) and cut into small pieces about 500–2,000 μm in diameter and 200 μm thick.

The samples were then placed in aluminium planchettes filled with 1- hexadecene and high-pressure frozen (HPM 010, Bal-Tec). The subsequent freeze-substitution was performed for 64 h (40 h at -90°C, 12 h at -70°C, 12 h at - 50°C) in acetone saturated with uranyl acetate. In addition, fluorescent dyes were added to the substitution medium prior to cooling. After freeze-substitution, the samples were washed with acetone and twice with ethanol (1h at -50°C each). For embedding, the samples were infiltrated with HM 20 in ethanol (33% for 2 h at -50°C, 67% for 2 h at -50°C, and two times 100%, overnight and for 2 h at -50°C). UV-polymerization was in fresh pure HM 20 for at least 4 days at -50°C.

Sample Preparation and Imaging
Excess surrounding resin was removed from the sample and a plain sample surface (blockface) was produced using a diamond trimming knife. To facilitate the sub-sequent correlation of CLSM and TEM images, the resin block was marked with a 31 gauge-needle (Fig. 1). The blockface is examined by confocal laser scanning microscopy using a Leica TCS SP CLSM equipped with an Ar/Kr laser. For optimal image quality oil immersion objectives were chosen, since the refractive index of immersion oil is closer to the refractive properties of resin. 3D structural reconstruction of CLSM image stacks was performed using Imaris (Bitplane Zurich, Switzerland).
The evaluation of confocal 2D and 3D image data revealed areas of interest (AOI) within the sample for further ultrastructural investigation. The depth of these AOI within the sample was estimated using the measurement tool of the CLSM software, and ultrathin sections were taken from the sample at the respective depth in a ultra-microtome with a diamond knife (UCT, Leica). The sections were stained with uranyl acetate and lead citrate and examined in the transmission electron microscope (Tecnai12, FEI).

Immunlabelling of Thin Sections
For immunofluorescence, semi-thin sections (200 nm) were placed on masked glass cover slips. The sections then were treated with 10% normal donkey serum for 2 h, washed twice and incubated with mouse anti-milk fat globule anti-serum (1:50) for 1 h. Afterwards, the sections were washed twice and incubated with a Cy5-conjugated donkey anti-mouse IgG secondary antibody (1:200) for 1 h.

Correlating Light Microscopy with Electron Microscopy
In Fig. 1, the basic steps of the correlative multimodal microscopic investigation are shown for a skin tissue sample. First, the high-pressure frozen sample is freezesubstituted in the presence of fluorescent dyes and embedded in resin (a). Then a plain blockface is prepared, marked with a fine needle, and 3D light microscopic data are collected from the blockface with a confocal microscope (b). Due to the intense fluorescent labelling of such prepared tissue, the high complexity of tissue and cellular organisation can also be selectively visualized by 3D reconstruction, e.g. epidermal nuclei (c). After evaluation of the CLSM data, an area of interest is chosen in a single 2D image and its depth in the resin block is determined using the CLSM software. Ultrathin sections are taken at the corresponding depth, and these sections are examined in the TEM (d). Using this procedure, the tissue sample can be depicted from the macroscopic level to the electron microscopic level, and the microscopy mode used for imaging can be chosen according to the desired resolution. To visualize for example the localisation of nuclei or cell borders in tissue samples, CLSM imaging is absolutely sufficient. However, to obtain more information on the structure of cell-cell contact sites (e) or collagen bundles (f), TEM imaging is possible on the identical sample and cell.