VIS2FIX: A High-Speed Fixation Method for Immuno-Electron Microscopy


  • Matthia A. Karreman,

    Corresponding author
    1. Molecular Biophysics, Department of Physics and Astronomy, Utrecht University, Princetonplein 1, NL-3508 TA Utrecht, The Netherlands
    2. Biomolecular Imaging, Department of Biology, Utrecht University, Padualaan 8, NL-3584 CH Utrecht, The Netherlands
      Matthia A. Karreman,
    Search for more papers by this author
    • These authors contributed equally to the work presented in this paper.

  • Elly G. van Donselaar,

    1. Biomolecular Imaging, Department of Biology, Utrecht University, Padualaan 8, NL-3584 CH Utrecht, The Netherlands
    Search for more papers by this author
    • These authors contributed equally to the work presented in this paper.

  • Hans C. Gerritsen,

    1. Molecular Biophysics, Department of Physics and Astronomy, Utrecht University, Princetonplein 1, NL-3508 TA Utrecht, The Netherlands
    Search for more papers by this author
  • C. Theo Verrips,

    1. Biomolecular Imaging, Department of Biology, Utrecht University, Padualaan 8, NL-3584 CH Utrecht, The Netherlands
    Search for more papers by this author
  • Arie J. Verkleij

    1. Biomolecular Imaging, Department of Biology, Utrecht University, Padualaan 8, NL-3584 CH Utrecht, The Netherlands
    Search for more papers by this author

Matthia A. Karreman,


Immuno-transmission electron microscopy (TEM) is the technique of choice for high-resolution localization of proteins in fixed specimen. Here we introduce 2 novel methods for the fixation of sections from cryo-immobilized samples that result in excellent ultrastructural preservation. These high-speed fixation techniques, both called VIS2FIX, allow for a reduction in sample preparation time from at least 1 week to only 8 h. The methods were validated in immuno-TEM experiments on THP-1 monocytes, human umbilical vein endothelial cells (HUVECs) and Madin–Darby canine kidney (MDCK-II) cells. The fixation and retention of neutral lipids is demonstrated, offering unique prospects for the application of immuno-TEM in the lipidomics field. Furthermore, the VIS2FIX methods were successfully employed in correlative fluorescence and electron microscopy.

Immuno-transmission electron microscopy (TEM) allows for high-resolution localization of proteins by labeling with antibodies and immuno-gold on thin sections of cells or tissues. The fixation method of choice for ultrastructural analysis of immuno-gold labeled specimens is cryo-immobilization. In contrast to chemical fixation, this process is nonselective, less prone to artifacts and extremely fast; it provides the ability to stop processes as they take place. High-pressure freezing (HPF) allows for quick freezing of relatively large volumes (300 µm3) of cells or tissue without the formation of ice crystals. The vitrified samples need to be brought to room temperature in order to make immuno-labeling possible. Formation of ice crystals while increasing the temperature of the sample can be prevented by performing a freeze substitution (FS). In FS, vitrified water in the sample is replaced with an organic solvent at low temperature. Next, the temperature is raised and, simultaneously, the sample is fixed by chemicals in the solvent. The cells can then either be embedded in resin (1), or rehydrated and further processed according to the Tokuyasu method (2,3). A major disadvantage of these methods is that the processing is done on comparatively large specimen blocks. Consequently, infiltration and fixation rates are slow and the total procedure can easily take a week or more to accomplish.

Here, a novel approach to the preparation of cryo-immobilized material for immuno-TEM is presented. First, vitreous sections (VIS) of the cryo-immobilized sample are cut and adhered to the grid by using an antistatic device (see Materials and Methods). Next, the sections on the grids are fixed, brought to room temperature, and immuno-labeled. Two different methods to fixate (FIX) the VIS are developed; VIS2FIXFS and VIS2FIXH (for a schematic overview, see Figure 1A). Owing to the small volume of the sections, both methods are much faster than existing methods. Importantly, this allows users to efficiently test different fixations to optimize the protocols for their particular usage. As every ribbon of sections can be fixed independently, many fixatives can be tested on one single sample. In particular, for unique specimen (e.g. biopsies) this can be very advantageous.

Figure 1.

The VIS2FIX methods of section fixation. A) A flowchart highlighting the steps involved in preparing biological material for immuno-TEM using the two VIS2FIX methods. General steps, here indicated in blue, are described in detail in the Material and Methods section. In green (for VIS2FIXFS) and yellow (for VIS2FIXH) the two most important steps in the section fixation methods are shown, and the approximate duration of each step (in minutes) is indicated. B) A top view of the two fixation set-ups inside the AFS (cooled down to −90°C). The set-up for VIS2FIXFS is shown in the lower half of the image. Here, 5 grids with adhered vitreous sections are placed in the flow-through ring filled with fixative solution (in acetone). The left most grid is still floating on the solution, whereas the four other grids have already sunk to the bottom of the flow-through ring. Above, four grids are placed on top of the frozen fixative solution (in buffer). During the fixation procedure, the set-ups would normally be covered with an aclar disk to protect the grids and fixative from the humid air. For visualization purposes, the covers were removed.

Results and Discussion

The first method, VIS2FIXFS, is based on FS and subsequent rehydration of the sections. The grids with the adhered sections were transferred to the FS chamber (−90°C) of an automatic freeze substitution system (AFS; Figure 1B, lower part). A high-speed FS procedure was employed, which is completed in only 85 min (Table 1). To achieve this, the grids were placed on dry acetone at −90°C containing 0.1% (wt/vol) uranyl acetate, 0.2% (vol/vol) glutaraldehyde (GA) and 0.2% (wt/vol) osmium tetroxide. In the AFS, the temperature was increased from −90°C to 0°C in three steps. First the specimen was kept at −90°C for 15 min to replace the ice with acetone. Next, the sections were fixed for 15 min at −60°C and −20°C. It is known that osmium tetroxide already starts reacting at −70°C (4) and GA at −50°C (1), but choosing −60°C and −20°C yielded better results in our experiments. After leaving the sections for 15 min at −60°C, the osmium tetroxide may be removed from the FS medium to prevent any negative effect on antigenicity (5,6). At 0°C, the sections were stepwise rehydrated in the presence of 0.2% (vol/vol) GA to prepare them for immuno-labeling.

Table 1.  FS schedule for VIS2FIXFS
Temperature in °C (period in minutes)Slope (degrees per hour)
−90°C (15 min)0°C
Increase to −60°C (13 min)Maximum slope (138°C)
−60°C (15 min)0°C
Increase to −20°C (18 min)Maximum slope (133°C)
−20°C (15 min)0°C
Increase to 0°C (9 min)Maximum slope (133°C)

The second method, VIS2FIXH (‘H’ for hydrated), is easier and quicker. Like VIS2FIXFS, the grids with sections were transferred to the AFS at −90°C (Figure 1B, upper part). In this cold and dry environment, the grids were placed on water based frozen fixative. The fixative was melted on a hotplate until the surface became liquid. Next, it was placed on ice and the sections were fixed for another 10 min. To optimize the method, several combinations of fixatives were tested. The optimal fixative for fixing THP-1 human monocytes was composed of 0.05% (wt/vol) osmium tetroxide, 0.2% (vol/vol) GA and 0.2% (wt/vol) uranyl acetate in buffer. In a mixture of 0.01–0.2% osmium tetroxide and 0.2% uranylacetate, the addition of either 0.2–2% formaldehyde, 0.01–0.2% GA or 0.1–1% acrolein gave similar results. Satisfactory results were also obtained when osmium tetroxide was omitted from the fixative mixture. We note, however, that the use of osmium tetroxide aids in the stabilization of, e.g., heterochromatin. The composition of the fixative can be altered depending on the antibody to be employed for immuno-labeling or the cell type used. This freedom makes this method applicable for experiments involving antigens sensitive to osmium tetroxide or GA.

The usage of osmium tetroxide in the fixative mixture has a great advantage as it allows for the fixing of lipids. Lipid droplets (Figure 2A,B indicated with ‘LD’) are organelles composed of a core of neutral lipids surrounded by a phospholipid monolayer. The fixation and retention of neutral lipids (Figure 2A) is a unique feature of the VIS2FIXH method. In existing immuno-TEM preparation methods, neutral lipids are usually either extracted during FS or not fixed, like in Tokuyasu samples. Also using the VIS2FIXFS method the lipid core is lost from the droplets, whereas the shape of the droplet remains (Figure 2B). The preservation of the lipid droplets using VIS2FIXH offers novel prospects for immuno-TEM in the lipidomics field.

Figure 2.

Ultrastructural aspects of VIS2FIX methods and immuno-gold labeling of PDI. A) After VIS2FIXH fixation the neutral lipids in the core of a lipid droplet (LD) are preserved. This is visible from the comparatively high gray value (electron density) inside the droplet. B) This density is lost after VIS2FIXFS. C) Immuno-gold (10 nm) labeling of ER resident protein PDI on a VIS2FIXH section. D) Immuno-gold (15 nm) labeling of PDI on a VIS2FIXFS section. Note the full cytoplasm and well-preserved heterochromatin in the nuclei (N). ER, endoplasmatic reticulum; G, Golgi; Ly, lysosome; M, mitochondria; Mv, multivesicular body. Scale bars in (A) and (B) represent 300 nm, and 500 nm in (C) and (D).

When the temperature of a cryo-immobilized specimen is raised from −150°C to 0°C, like with VIS2FIXH, both small cubic ice and larger hexagonal ice can be formed (7). Small cubic ice, which is formed while warming the sample to temperatures above −135°C, has little influence on the ultrastructure of the sample and localization of macromolecules (8). When hexagonal ice is formed during vitrification, ultrastructural damage can be observed in the cell, but is most easily detectable in the nucleus and cytoplasm (8–10). Following VIS2FIXH, no ice-crystal damage was observed in the nucleus, cytoplasm or organelles of the cells (Figure 2A,C).

The effect of a temperature rise on the formation of hexagonal ice, produced at temperatures higher than −90°C, was examined. The level of damage was assessed by the integrity of the heterochromatin, the cell's organelles and the cytoplasm. Sections of THP-1 cells were cut and statically adhered to the grids, as described previously. These grids were equally divided over two sapphire disk (SD) holders. One SD holder with grids was transferred to the AFS at −90°C (see Materials and Methods). Next, these grids were heated to −5°C, and cooled down again to −90°C in the dry environment of the AFS. As a control, the grids in the second SD holder were kept in the ultramicrotome at a constant temperature of −150°C during the experiment. At the end of the thermal cycle, the control grids were transferred to the AFS at −90°C. Together with the thermally cycled sections, the controls were freeze substituted and rehydrated according to VIS2FIXFS.

During FS, the sections were fixed at low temperature. Importantly, if any ice-crystal-induced damage would be present in the samples, it would be fixed and visible in the TEM. A morphological comparison was made between the directly freeze substituted (Figure S1A) and thermally cycled sections (Figure S1B). No difference could be observed between the ultrastructure of the differently prepared sections, not even in the structure of the heterochromatin. We note that heterochromatin is known to be very sensitive to hexagonal ice-crystal damage. The observed preservation of the structural integrity of the samples implies that no ultrastructural damage is introduced by the formation of ice crystals during thermal cycling from −90°C to −5°C and cooling back down to −90°C.

At present, we do not have an explanation for these observations. According to the work of Jacques Dubochet (8,11), ice formation during devitrification of a sample has little or no influence on the ultrastructure of biological material and localization of macromolecules. It is suggested that cubic ice crystals are formed from immobile water and large hexagonal ice can be created by connecting small cubic ice crystals without displacing them (11). This process has no influence on the location of macromolecules as the ice does not move.

Another explanation for the absence of hexagonal ice crystals could be sublimation of water from the thin sections at −90°C. The removal of water from the specimen would reduce the possibility to form ice crystals. The AFS room is filled with dry nitrogen gas to prevent the inflow of air and water vapor. This cold and dry environment allows for (partial) freeze drying of the thin sections. This could explain the lack of ice-crystal damage after re-cooling the specimen from −5°C to −90°C. Further work, using for instance cryo-electron microscopy (cryo-EM), is required to come to a conclusive explanation.

To validate the VIS2FIX methods with immuno-TEM, a selection of antibodies was tested. Efficient labeling was found when membrane bound proteins like CD63, LAMP-2 (both located at the lysosome) and AP-1 (located at clathrin coated vesicles close to the Golgi) were targeted (data not shown). Furthermore, both fixation techniques allowed abundant labeling of endoplasmic reticulum (ER) lumen located PDI (Figure 2C,D). In both methods, we observed excellent membrane contrast of the ER, the nuclear envelope, the Golgi, cristae of the mitochondria and lysosomes (Figure 2). With the VIS2FIXH method the membrane contrast is somewhat less pronounced (Figure 2C), possibly because more of the cytoplasmic content is preserved that contributes to the overall contrast. A striking quality of both methods is the outstanding preservation of vesicles, in particular, in the Golgi area and organelles. This result could be due to the choice of fixatives in combination with the nature of section fixation; we hypothesize that the open structure of the sections allows for high accessibility for the fixatives.

The potential of the VIS2FIX methods for immuno-labeling was further assessed by testing the labeling efficiency of an antibody directed against Caveolin on VIS2FIX sections of human umbilical vein endothelial cells (HUVECs). This antibody no longer recognizes its epitope when the cells are fixed in a fixative mixture containing 0.2% GA and subsequently prepared following conventional immuno-TEM methods (data not shown). Following VIS2FIXH and VIS2FIXFS fixation, however, the immuno-labeling of caveolin is still possible, and very effective, when 0.2% GA is present in the fixative mixture (Figure 3A,B, respectively). Surprisingly, the negative influence of GA on the labeling of caveolin observed in other preparation methods for immuno-TEM is not found with both VIS2FIX methods. This result suggests that the VIS2FIX methods offer new possibilities for immuno-labeling with antibodies hitherto not applicable in immuno-TEM. Possibly, the shielding of epitopes caused by cross-linking of larger volumes is prevented due to the fact that only a thin section is fixed using the VIS2FIX methods.

Figure 3.

Immuno-gold labeling of Caveolin on VIS2FIX sections of HUVECs. Sections of high-pressure frozen HUVECs were fixed according to VIS2FIXH (A) and VIS2FIXFS (B) and subsequently labeled with immuno-gold (10 nm) for caveolin located to caveolae. Caveolae are invaginations of the plasma membrane and also occur in the cell resembeling small vesicles (50–100 nm). For VIS2FIXH (A) a fixative mixture of 0.2% GA + 0.2% UA in buffer was employed. For VIS2FIXFS (B), the sections were fixed with 0.1% OsO4 + 0.1% UA in Acetone between −90°C to −60°C. At −60°C the fixatives were changed to 0.2% GA + 0.1% UA in Acetone. Following FS, the sections were stepwise rehydrated in the presence of 0.2% GA (see also Materials and Methods for more details and subsequent steps). Scalebars represent 200 nm.

The similarity between the Tokuyasu method and the VIS2FIX methods is the use of non-resin embedded biological material. The lack of a resin matrix is generally considered to be an advantage of the Tokuyasu method as it improves the accessibility of the epitopes. However, the absence of a resin matrix also results in the loss of material. To investigate the retention of material following VIS2FIX, the Forssman glycolipid was immuno-labeled. Previously, the immuno-labeling of the Forssman glycolipid was performed on both Tokuyasu and Lowicryl sections of MDCK-II cells (12). Although abundant labeling of Forssman glycolipid could be achieved on Lowicryl sections, the epitope seemed to be extracted and relocated during the thawing and labeling of Tokuyasu cryo-sections. To test the retention of the Forssman glycolipid following the novel fixation methods, sections of MDCK-II cells were fixed using VIS2FIXFS and VIS2FIXH, and subsequently immuno-labeled. With both methods, excellent immuno-labeling of Forssman glycolipid was achieved (Figure 4A,B). Labeling was found on the plasma membrane, lysosomes, endosomes and Golgi, which is in agreement with previous findings (12). In conclusion, using VIS2FIX the omission of resin embedding does not lead to a loss of the epitope. The increased retention of material combined with the high accessibility of the epitopes due to lack of a resin matrix make the VIS2FIX methods excellent for immuno-labeling.

Figure 4.

Immuno-gold labeling of Forssman glycolipid on VIS2FIX sections of MDCK-II cells. Sections of high-pressure frozen MDCK-II cells were fixed according to VIS2FIXFS (A) and VIS2FIXH (B) and subsequently labeled with immuno-gold (10 nm) for Forssman glycolipid. Positive labeling of Forssman glycolipid can be found on the plasma membrane, endosomes, lysosomes, and the Golgi. Sections fixed according to VIS2FIXFS (A) were treated as described for the HUVECs (see Figure 3 legend). For VIS2FIXH (B) the cells were fixed with 0.2% OsO4 + 0.2%GA + 0.2% UA in buffer. Scalebars represent 500 nm.

To further investigate the potential of both methods, VIS2FIX was applied in integrated laser and electron microscopy (iLEM) (13). This new form of correlative microscopy combines a custom built laser scanning fluorescence microscope and a TEM within one set-up. Here, regions of interest (ROI) are identified based on the fluorescence signal of the specimen on the EM grid. The positions of the ROIs are retrieved with ∼1 µm precision in the TEM and can be investigated at high resolution. Since both the fluorescence microscope and the TEM employ the same sample and sample holder, the correlation is quick and reliable. On VIS2FIXH (data not shown) and VIS2FIXFS fixed sections, LAMP-2 was double labeled with gold and fluorescent markers (Figure 5). Both techniques yielded high fluorescence signals and rich immuno-gold labeling. It is challenging to prepare specimens that are compatible with both fluorescence microscopy and TEM; the heavy metals used for obtaining TEM contrast quench the fluorescence signal (14). Therefore, less heavy metal stain is allowed to post-contrast the iLEM samples. This specimen yielded a somewhat lower, but still satisfactory, TEM contrast than conventionally prepared samples (Figure 5B,C) and showed little quenching of the fluorescence signal. These results indicate that the VIS2FIX methods are well suited for use in both conventional immuno-TEM and (integrated) correlative microscopy. The potential of the VIS2FIX methods is further illustrated by promising first results in electron tomography on VIS2FIX prepared sections (data not shown).

Figure 5.

iLEM imaging of VIS2FIXFSsections labeled for LAMP-2. A) Fluorescence image of a 50 µm2 area on a section labeled for LAMP-2 with immuno-gold (15 nm) and immuno-fluorescence (Alexa 488 Fluor, see Online Methods). B) Overlay of the fluorescence signal of the area indicated in A (white square), and a TEM image of the same region. C) A TEM image of the area indicated in B (black square). Here it becomes clear that the lysosomes are abundantly labeled. Scale bars represent 10 µm in A, 1 µm in B, and 500 nm in C.

Here, we introduce two novel section fixation techniques for immuno-labeling of high-pressure frozen material. Section fixation was first introduced by Slot and Geuze (15), but good results could only be obtained in small areas of the section (3,15). In contrast, the VIS2FIX methods exhibit excellent conservation of ultrastructure throughout the entire section. Importantly, the methods are fast, reproducible, and results are of high quality. Now it is possible to perform cryo-immobilization, sectioning, subsequent fixation and immuno-labeling within a period of 8 h. The methods offer prospects for conventional immuno-TEM but also have potential to meet the challenges in areas such as lipidomics, correlative microscopy and tomography.

Materials and Methods

Cell culture and HPF

THP-1 human monocytes, a leukemic cell line, were purchased from ATCC (LGC Standards GmbH) and cultured according to supplier's instructions. MDCK-II cells were cultured in DMEM (Dulbecco's Modified Eagle Medium) with low glucose (PAA Laboratories, Inc.), enriched with 5% Fetal Calf Serum Gold (PAA Laboratories, Inc.). HUVECs were cultured to a cobblestone state as described by Jiménez et al. (16), and kept cobblestone in culture for 6 days until HPF. Preceding HPF, the monolayers of the adhering cells were dissociated by placing a small volume of 0.05% Trypsin + 0.02% EDTA (PAA Laboratories, Inc) in the petri dishes for 10 min (MDCK-II) or <1 min (HUVEC). The trypsinization process was stopped by adding culture medium to the cells. THP-1 cells, HUVECs and MDCK-II cells were spun down (5 min, 153 × g) at 37°C, and the pellet was resuspended 1:1 in 30% (wt/vol) dextran (Fluka) in culture medium. Copper specimen tubes were filled with the suspension of cells in dextran and frozen in high pressure with either the Leica EM HPF or the Leica EM PACT (17) at a pressure of ∼2000 bar. Prior to sectioning, the frozen tubes were stored in liquid nitrogen.

Cryo-sectioning and grid transfer to the AFS

At a temperature of −150°C, the tubes were trimmed and sectioned in the Leica EM UC6/FC6 ultramicrotome with the Leica CRION antistatic device set to discharge. For trimming the specimen, a cryotrim 45 diamond knife was employed (element 6). Ultrathin sectioning was performed with a 45° cryo-dry diamond knife (element 6). Before the sectioning, a TEM grid with formvar film, carbon coated and glow discharged, was mounted in the cross tweezers in the Leica micromanipulator. Ultrathin sectioning was facilitated with the discharge mode of the Leica CRION antistatic device. Sectioning for VIS2FIX is very similar to sectioning for cryo-EM of vitreous sections (CEMOVIS), as both involve cryo-immobilized material. Ultrathin sectioning can be challenging. For trouble shooting and helpful suggestions on cryo-sectioning please refer to the excellent work of A. Al-Amoudi (18) and J. Pierson (19). Furthermore, here some suggestions are given that yielded excellent results in our laboratory.

Thoroughly cleaning the diamond knife helped the sectioning; the cleaning appears to slightly charge the knife edge, preventing compression of the sections. Following HPF with the Leica EMpact, the copper tubes need to be cut out of their holders with a tube cutter (Leica). Tubes frozen with the Leica EM HPF were also cut with the Leica tube cutter to ensure that the tubes frozen in different HPFs had equal lengths. The cutting of the tubes can locally influence the quality of the ice in the tubes and mechanically damage the sample. Therefore, before first use, 400 µm was trimmed from the front of the tubes (feed 200 nm, speed 100 mm/s). The front of the tubes was trimmed so that a black, shiny and flat surface is obtained. Subsequently, 100 µm was trimmed from the edges of the tube. To achieve a square block face employing a 45° trimming knife, the onset point of the trimming was exactly on the interface between the sample and the tube.

Before ultrathin sectioning, a TEM grid was mounted in cross tweezers in the Leica micromanipulator (see also the Leica EM FC7 and UC7 brochure, page 13, top three images. The brochure is freely available at The tips of these tweezers (Dumont inox crossover type N5) were bended, resembling tips of cover-slip tweezers. This modification allowed for the grid, when mounted into the tweezers, to be parallel to the cutting edge of the knife. This position facilitated the mounting and adhering of the ribbons following sectioning. Furthermore, the cross tweezers were adjusted so that the tips were not pressed against each other too tightly. This allowed for placing and manipulating the grid into the tweezers while the tweezers were closed.

To mount the grid, the tweezers in the micromanipulator were rotated so that both legs of the tweezers were visible for the user. The grid was held in between the legs, and next carefully moved into the tip of the tweezers. Subsequently, the tweezers were rotated back to the original position, and the grids' surface was in line with the knife edge. The grid was positioned so that the grid bars were parallel to the ribbon produced while sectioning. Upon adhering of the ribbons, this orientation allows the majority of the sections to be in the meshes of the grid and not hidden from view on the grid bars.

While sectioning, the tip of the tweezers with the grid was rotated away so that the presence of the grid did not interfere with the sectioning. In this position the grid was still slightly charged by the ionizer, facilitating the adherence of the ribbon in a later state. Ultrathin sectioning (60–80 nm) was facilitated with the discharge mode of the Leica Antistatic device. The ribbon was guided with a hair of a guinea pig mounted on a wooden stick. Hairs of guinea pigs are slightly angled, particularly strong and often the ribbons stuck to the tip of the hairs, which helped holding onto the ribbon. When a sufficiently long ribbon was achieved, the discharge mode of the Antistatic Device was switched off, and the grid was turned back toward the edge of the knife. To avoid limited visibility of the ribbon due to the reflection of the grid's film, the grid was tilted slightly toward the knife's cutting edge. The ribbons were statically adhered to the grids with the CRION set to charge. See also supplementary movie SM2 from the work of Pierson et al. (19). Here, it was of great importance to check if the adherence was successful by trying to lift the ribbon from the grid, which should not be possible.

Following adhering of sections, the grid was removed from the micromanipulator tweezers. Remaining inside the cold and dry environment of the microtome chamber, the grid was then placed into a Leica SD holder (part of the SD FS unit), which was sitting in a petri dish. The SD holder can hold 24 sapphire disks or, for this application, grids. After collecting and storing the required number of grids, the knife and the sample were removed from the microtome. Subsequently, the SD holder with grids was transferred from the microtome chamber to the FS chamber of the AFS2 (Leica AFS unit). During this process, the grids in the SD holder must be kept cold and shielded from humid air to prevent the formation of ice crystals on the section. Therefore, the SD holder with the grids should be placed in a protective surrounding before the transfer. Hereto, a precooled tin (Leica Universal Container) was present in the microtome chamber with one cold ring (Leica bottom plate) on the bottom. The SD holder with the grids was placed in the tin, and covered with a donut-shaped aclar foil, cut from a 200-µm-thick aclar embedding film (Electron Microscopy Sciences). The aclar foil's outer diameter was 3.5 cm and the hole in the center had a diameter of 9 mm. Two more cold rings were placed on top and covered by a disk-shaped aclar foil (diameter 3.5 cm). Quickly but gently, the tin was then transferred to the AFS2 (set to −90°C), which was positioned very close to the ultramicrotome. Before cooling down the microtome for cryo-sectioning, the SD holder (kept in a closed petri dish), the tin with the cold rings and the aclar foils were placed inside the microtome chamber. A small piece of partially folded tape was stuck to the face of the aclar foils and the lids of the petri dishes, functioning as a handle to facilitate lifting and moving. The items employed for grid storage and transfer were cooled down together with the microtome to a temperature of −150°C.

VIS2FIXFS method

A Leica flow-through ring (trimmed to a height of ±6 mm) was placed in a Leica reagent bath (cut-off to a height of ±1 cm). The reagent bath with the flow-through ring was positioned on top of three cold rings (Leica bottom plates) in a tin (Leica Universal Container) at −90°C in the AFS2. At this temperature, 3 mL of precooled (−90°C) dried acetone (MERCK) with fixatives was pipetted into the flow-through ring, and covered with an aclar foil (diameter 3.5 cm). Acetone was chosen as the retention of phospholipids in the sample following FS is higher with this solvent than with methanol (20). The fixatives used were 0.1% (wt/vol) uranyl acetate (UA) (MERCK), 0.1–0.5% (vol/vol) GA (from stock solution 10% in acetone, Electron Microscopy Sciences) and 0.2–0.5% (wt/vol) osmium tetroxide (Electron Microscopy Sciences) in acetone. Following sectioning, the grids were transferred in the SD holder to the FS chamber as described above. Using a precooled, fine tip tweezers (Dumont inox crossover type N5, the tips bended to resemble cover-slip tweezers), each grid was carefully floated with the section side down on the fixative within one individual compartment of the flow-through ring. One flow-through ring holds a maximum of 10 grids, but up to three flow-through rings with fixative can be placed in the AFS2 as described above. Due to the low-surface tension of the acetone, the grids eventually sink to the bottom of the flow-through compartments. As soon as the last grid was placed on the fixative, the FS program was started.

If desired, the fixative composition can be altered at any point during the FS. To achieve this, the majority of the fixative was removed, pipetting it from the center of the flow-through ring. Here, it was important to not let the grids dry out. This can be followed by some washing steps. Subsequently, the new fixative was added to the flow-through ring with the grids. We achieved good morphology when the fixative employed from −90°C to −60°C was composed of 0.1% (wt/vol) UA and 0.2% (wt/vol) osmium tetroxide in acetone. This mixture was subsequently replaced by 0.1% (wt/vol) UA and 0.2% (vol/vol) GA in acetone. At the end of the FS program, when the temperature reached 0°C, the grids were washed five or more times with 0.2% (vol/vol) GA in acetone (approximately 3 mL per washing step) as described above. The following rehydration steps were performed in the AFS2 at 0°C, in seven sequential steps of 1–2 min: 0.2% (vol/vol) GA in 95% acetone, 0.2% GA in 90% acetone, 0.2% GA in 80% acetone, 0.2% GA in 70% acetone, 0.2% GA in 50% acetone, 0.2% GA in 30% acetone and finally 0.2% GA in 10% acetone in filtered water (Millipore 0.22 µm membrane filter). To prepare the solutions for steps 1–5 we employed a stock solution of 10% GA in acetone (Electron Microscopy Sciences), and for subsequent steps 8% GA in filtered water (EM grade; Polysciences, Inc.)

After rehydration, the grids in the flow-through ring were washed three times with filtered water. The grids were removed from the flow-through ring with fine tip tweezers, and the back of the grid was dried with slightly moist filter paper. Finally, the grids were washed seven times for 1 min by floating them on drops of filtered water placed on a strip of parafilm. At this point, the grids were ready for immuno-labeling or storage for later use. The grids can be stored similar to Tokuyasu sections (21). To this end, the grids were briefly incubated by floating them on a drop of 1:1 mixture of 1.8% (wt/vol) methylcellulose (Sigma-Aldrich) and 2.3 m sucrose in 0.1 m PHEM buffer [composed of 60 mm PIPES (MERCK), 25 mm HEPES (MERCK), 10 mm EGTA (Sigma-Aldrich) and 2 mm MgCl (MERCK), pH adjusted to 6.9] on ice. The grids were gently pulled off the viscous drop and placed on a parafilm covered glass slide with the sections and the drop facing downward. The glass slide with the grids was placed in a glass petri dish, sealed with parafilm, and stored at 4°C. To release the grids after storage, a drop of 0.1 m PHEM can be pipetted between the grid and the parafilm, dissolving the partly dried methylcellulose.

VIS2FIXH method

A 2 mL fixative was prepared containing 0.01–0.5% (wt/vol) osmium tetroxide (Electron Microscopy Sciences), 0.2% (wt/vol) UA (MERCK) and 0.01–0.2 % (vol/vol) GA (8% in filtered water, Polysciences, Inc.) in 0.1 m PHEM buffer, on ice. In the reagent bath (cut-off to a height of 6 mm), we placed a final volume of 800 µL fixative covering the bottom of the bath. The reagent bath was then transferred to the AFS2 (set to −90°C) and placed on top of three cold rings in a tin, covered with an aclar foil or petri-dish lid (diameter 3.5 cm). The fixative immediately froze. The grids with adhered sections were transferred to the AFS2 as described above. With the precooled bended tip tweezers the grids were placed gently, with the sections facing down, on the frozen fixative. The reagent bath with the grids was then placed in a cooled petri dish to protect it from the humid air, and transferred from the AFS2 onto a 40°C hot plate. After 4–5 min the surface of the fixative became partially liquid. As soon as all the grids were floating on the fixative, the petri dish with the grids and the fixative was quickly placed on ice. The fixative, which was still partly frozen, and the grids were protected from light to limit oxidization of the osmium tetroxide. The sections were allowed to fix for another 10 min. Finally, the grids were removed from the fixative and washed 10 times for 1 min on drops of 0.1 m PHEM buffer or filtered water; then they were ready for immuno-labeling or storage, respectively.

Immuno-labeling for TEM and iLEM

For immuno-TEM, the free aldehyde groups in the VIS2FIX fixed sections were quenched by washing the grids five times for 2 min on drops of 0.1 m PHEM buffer containing 0.02 m Glycine (MERCK). To block nonspecific binding of the antibody, the sections were incubated for 15 min with blocking buffer, containing 1% (wt/vol) BSA (Sigma-Aldrich) in 0.1 m PHEM buffer, followed by 1 h incubation with the specific primary antibody. Antibodies used here were directed against PDI (mouse monoclonal, 1:100 in blocking buffer, Stressgen Biotechnologies Corp.), Caveolin (rabbit polyclonal, 1:100 in blocking buffer, BD Biosciences) and Forssman glycolipid (rat monoclonal, diluted 1:10 in blocking buffer) (12,22). After five washing steps with 0.1% (wt/vol) BSA in 0.1 m PHEM buffer, the sections labeled for PDI were incubated for 20 min with a bridging antibody; rabbit anti-mouse Ig (1:300 in blocking buffer, Dako).

The sections were washed five times for 2 min in 0.1% (wt/vol) BSA in 0.1 m PHEM buffer, and labeled for 20 min with protein A gold [1:80 (this dilution is batch-dependent) in blocking buffer; Cell Microscopy Centre, University Medical Centre Utrecht]. After washing the sections three times briefly, and seven times 2 min in 0.1 m PHEM buffer, the labeling was stabilized by 5 min incubation with 1% (vol/vol) GA (Polysciences, Inc.) in 0.1 m PHEM buffer. To remove the buffer and the GA, the sections were washed 10 times for 1 min on drops of filtered water and stained for 5 min with 2% (wt/vol) uranyl oxalate in filtered water pH 7 (prepared from oxalic acid and uranyl acetate). Thereafter, we briefly washed the sections twice on filtered water. Finally, the sections were embedded in 0.4% (wt/vol) uranyl acetate (pH 4, MERCK) in 1.8% (wt/vol) methyl cellulose on ice.

A similar protocol was employed for correlative labeling for the iLEM. As a primary antibody, a mouse monoclonal antibody to LAMP-2 (1:150, BD Biosciences Pharmingen) was used. Following incubation with rabbit anti mouse and protein A gold, the grids were washed five times 2 min with 0.1% (wt/vol) BSA in 0.1 m PHEM buffer followed by a 45 min incubation with Alexa 488 fluor conjugated goat anti rabbit antibody (1:200; Invitrogen). The sections were washed five times for 2 min with 0.1 m PHEM buffer and fixed for 15 min with 4% (wt/vol) formaldehyde (from paraformaldehyde; Sigma-Aldrich). We note that the usage of GA as a fixative may lead to an increase in auto fluorescence of the section. This was followed by 10 times 1 min washing on filtered water and staining for 5 min with 2% (wt/vol) uranyl oxalate and, next, stained with 2% (wt/vol) uranyl acetate for another 5 min. Between the staining steps, we washed the sections twice briefly on filtered water. The sections were washed, two times briefly on filtered water, and embedded in 1.8% (wt/vol) methyl cellulose on ice.


The sections were imaged in our iLEM(13). The iLEM consists of a Tecnai 12 120 kV transmission electron microscope (FEI Company) equipped with a custom designed laser scanning fluorescence microscope mounted on one of its side ports. The TEM images were recorded at 80 kV with a bottom mount TEMCam-F214 (Tietz Video and Image processing systems) charge-coupled device (CCD) camera. The laser scanning microscope is equipped with a 488-nm laser (Bluephoton, Omicron Laserage Laserprodukte GmbH) and an avalanche photo diode (APD) detector. The fluorescence microscope of the iLEM is operated using custom software written in LabView 8.0. For image processing Photoshop CS3 was employed. To overlay the fluorescence image and the TEM image (Figure 5B) the γ-value of the fluorescence image was adjusted. In all figures, the TEM and FM images are scaled for optimal contrast by employing linear adjustments of the levels of the entire image.


We thank N. Jiménez, W. F. Voorhout, and R. J. Mesman for critically reading the manuscript and useful discussions. We thank C. T. W. M. Schneijdenberg for sharing his insights on the behavior of water in cryo-conditions. J. W. Slot is greatly acknowledged for sharing his thoughts and ideas on the project. We thank C. E. M. Vocking for giving creative input on naming the methods, R. Snyder for critically reading the manuscript, and R. Scriwanek for helping with the preparation of the figures. We thank G. Van Meer for providing us with the antibody to Forssman glycolipid, which he in turn received from A. Sonnenberg. We are grateful to A. V. Agronskaia for her assistance with the iLEM. M. A. K. is funded by Technology Foundation Stichting Technologische Wetenschappen. E. G. D. is supported by Innovatiegerichte Onderzoekprogramma's Genomics (IOP Genomics, SenterNovem), and the Dutch FSHD Foundation.