A new approach for cryofixation by high-pressure freezing

Authors


D. Studer. Fax: + 41 31 631 38 07; e-mail: studer@ana. unibe.ch

Abstract

A newly designed high-pressure freezing machine for cryofixation was established and tested (Leica EMPACT), based on ideas originally proposed by Moor & Riehle in 1968. The new machine, essentially an improved version of our prototype, pressurizes the sample to 2000 bar in a small container (using methylcyclohexane as hydraulic fluid) and at the same time cools the outer surface of the container with a jet of liquid nitrogen. The advantage of this approach is that the machine uses little liquid nitrogen and can be built small and light. The machine is able to vitrify and freeze well a variety of specimens, for example, plant leaves, yeast cells, liver or nerve tissue (more samples are shown at: http://www.ana.unibe.ch/empact). Cooling efficiency is the same as in the traditional machines that use liquid nitrogen to pressurize and simultaneously cool the sample.

Introduction

High-pressure freezing is at present the only way to vitrify or freeze well a large variety of bulk biological samples thicker than 50 µm. The basic idea is to immobilize instantly and in situ all components constituting a biological sample. This prevents osmotic effects (Studer et al., 1992), and shifting and loss of ions (Somlyo et al., 1985; Zierold, 1991) as they occur with chemical fixation by aldehydes in different buffers.

Based on well-known specimens, e.g. defined salt and sugar solutions and well-characterized tissues (rat liver, bovine cartilage and Zea mays root), the limits and possibilities of high-pressure freezing have been documented in several papers (Riehle, 1968; Moor, 1987; Sartori et al., 1993; Studer et al., 1995; Shimoni & Müller, 1998).

Although the method has been commercially available for almost 20 years, its impact on biomedical research has so far been comparatively small. About 100 papers have dealt with the application of high-pressure freezing as a first fixation step for the ultrastructural description of biological structures. Between 1968 and 1992 about 20 papers appeared; since 1993 about 10 articles per year have been published, although about 50 machines are in use world-wide.

In general, conventional high-pressure freezing machines work adequately. Using liquid nitrogen as both pressurizing and cooling agent, they build up 2000 bar in a few milliseconds and cool the samples efficiently as soon as the pressure is established. These machines pressurize 100–160 mL of liquid nitrogen. With the help of a large hydraulic pump and a heavy pressure accumulator, a very solid high-pressure piston drives the liquid nitrogen. The specimen, fixed in a specimen holder, is inserted into a pressure chamber through which the pressurized liquid nitrogen is jetted. To coordinate pressure rise and cooling, the pressure chamber has to be filled with a solvent prior to the freezing cycle (Moor, 1987). This design has the following consequences. Because a relatively large volume of liquid nitrogen is pressurized, the machine has to be built very solidly as mentioned previously. To cool the high-pressure piston before use, large quantities of liquid nitrogen are necessary. The use of a solvent in the conventional high-pressure freezing machines (to coordinate pressure rise and cooling) limits their cooling efficiency (Studer et al., 1995).

To build a new smaller machine and to improve the cooling efficiency of high-pressure freezing cryofixation, at least theoretically, a new design was established based on initial ideas by Moor & Riehle (1968). Pressure generating hydraulic fluid and coolant are applied to the sample by means of two separate systems. Because the pressure system can be minimized and the coolant can be recycled, the machine consumes little liquid nitrogen. There are two sample holder systems (platelets and tubes) that work well. Instead of making an impressive big bang during freezing, as in conventional machines, the new design is relatively quiet.

Subsequently we describe in detail the mode of action of the new high-pressure freezing machine (Leica EMPACT; Studer, 1999) its performance and possibilities as well as a variety of applications.

Materials and methods

The new high-pressure freezing machine

The new high-pressure freezer (Fig. 1, EMPACT, Leica Vienna) functions according to the scheme shown in Fig. 2(a) (details are described in the figure legend). In this machine, application of pressure and cold to the sample are completely separated. For building up pressure, a pneumatic piston with a diameter of c. 80 mm generates about 150 kp on a high-pressure piston, which has a diameter of 3 mm, thus generating about 2000 bar in a hydraulic system which is coupled directly to the sample. The pressurized volume is about 0.5 mL and filled with methylcyclohexane. This solvent is almost insoluble in water; it melts at −126 °C and boils at + 101 °C. It is as flammable as toluene but less toxic (Merck, product information). The pressure is applied only to the inside of the sample holder, which therefore has to withstand 2000 bar. The cold is applied to the outer surface of the sample holder; it is a jet of liquid nitrogen driven by a pressure of 10 bar. A compressed air-driven piston synchronizes pressure rise and cooling. When freezing is accomplished the sample is automatically brought into a liquid nitrogen container. The machine is controlled through a touch screen monitor. All manipulations are indicated, and every freezing cycle is registered (Fig. 2b). Twenty cycles can be stored. The pressure within the sample chamber can be easily adjusted between 1 and 2500 bar just by changing the pressure of compressed air applied to the piston generating high pressure. The weight of the machine is about 80 kg and its dimensions are 650 × 900 × 500 mm.

Figure 1.

 Overview of the Leica (EMPACT) high-pressure freezing machine. The following parts are indicated: (A) the touchscreen controlling the whole process of high-pressure freezing. (B) reservoir containing methylcyclohexane, the liquid used to apply pressure to the sample. (C) stage where the sample is mounted for high-pressure freezing and the liquid nitrogen bath into which the sample is transferred after cryofixation. (D) Dewar filled with liquid nitrogen.

Figure 2.

 (a) Schematic drawing showing the functioning of the new high-pressure freezing machine: Sample B is introduced into a copper tube, which is surrounded by a perforated steel envelope A (detail see Fig. 3B1), or into a platelet system (detail see Figs 4A and B). Samples are transferred to the machine using a mounting rod (detail see Figs 3 and 4C). The sample holder is then fixed into the machine by pressing it between the pressure system and piston 1. The two nipples on either side of the sample holder are tight at 2000 bar (detail in Fig. 2a). The pressure system has a volume of about 300 µL, is filled with methylcyclohexane, and is connected through a one-way valve to reservoir D. Once the sample is fixed, piston 2 is activated, but stopped by shutter C. Then piston 3 starts to push liquid nitrogen through the tubes at a pressure of about 10 bar for precooling the system. The liquid nitrogen jet, however, is deviated by mechanism E before hitting the sample. Finally piston 4 is activated: shutter C is released, the pneumatic piston 2 applies about 150 kp to the high-pressure piston with a diameter of 3 mm, thus generating 2000 bar in less than 20 ms and mechanism E directs the liquid nitrogen jet simultaneously onto the specimen. When high-pressure freezing is completed, piston 1 releases the specimen into the liquid nitrogen bath F. (b) The pressure and temperature course as established in the EMPACT. The pressure slope (plateau at 2000 bar) shows that the pressure is achieved in about 5 ms and the cooling starts about 25 ms after pressure rise (plateau at −196 °C). The cooling rate (dT/dt) is in the order of 7000 K s−1 between 0 and 100 °C.

Sample holders and sample preparation

Two sample holders are available as described in detail in Figs 3 and 4. Type I consists of a copper tube (Fig. 3A) 16.3 mm long, with an inner diameter of about 300 µm and an outer diameter of 650 µm, which is surrounded by a stainless steel envelope (16 mm long, Fig. 3B1) for stability and to allow easy manipulation of the copper tube. The copper tube is held in the envelope by pressing two stainless steel cones simultaneously into its open ends, deforming them to a funnel-like structure. As an assembled unit, this sample holder is easy to use for suspensions of cells, bacteria and viruses or to introduce cellulose capillaries according to Hohenberg et al. (1994). To fill the copper tube with a suspension such as yeast cells (Saccharomyces cerevisiae) in our application, a wire with a diameter of about 250 µm is used as a piston. By immersing one end of the tube into a drop of the yeast suspension and by drawing back the wire, the tube is filled. Cellulose tubes are loaded into the copper tubes in either buffer, medium or 1-hexadecene. The full copper tube fixed in its envelope is attached to a manipulator (Fig. 3C) by a circumferential spring-loaded clamp (working like a mechanical pencil) and introduced into the high-pressure-freezing machine. In other experiments pieces of rat peripheral nerve about 12 mm long were introduced into such tubes. A copper wire 80 µm in diameter was pulled twice through the tube forming a loop at one end (Fig. 3B2). The peripheral nerve was placed into the loop, and was introduced into the tube by pulling the wires at the opposite end. Type II sample holder (Fig. 4A) enables one to freeze discs of freshly excised tissue 1.2 mm in diameter and with a thickness of 0.2 mm. Here the sample is prepared as described earlier (Studer et al., 1989, 1995) for conventional high-pressure freezing machines. Vibratomes, razor blades and punches are used to trim samples. In the experiments presented here, discs of ivy leaves (Hedera helix) were punched to the right size (diameter 1.2 mm), the air of the intercellular spaces was replaces with 1-hexadecene and the specimens were introduced into the cavity of the platelets (Studer et al., 1989). Thin sections of rat liver were cut with razor blades, punched and introduced into the cavity of the type II holder. Once the holder is filled with the sample, it is tightened in its envelope (Fig. 4B) by means of a torque wrench applying a force of 20 Ncm. The thread of the screw protruding from the envelope is fixed onto a manipulator (Fig. 4C) and the sample transferred and fixed in the EMPACT. During this procedure some of the type II holder platelets are deformed and can be used only for one freezing cycle. Others stay in shape and are used several times.

Figure 3.

 Type I sample holder consists of a copper tube (A; outer diameter 0.65 mm, inner diameter 0.3 mm), which is very suitable for suspensions; a wire of about 250 µm diameter is used as a piston to fill the tube. For stabilization and better handling, the tube is surrounded by a stainless steel envelope (B1). For filling the tube with small blood vessels and muscle tissue or nerves, a 80-µm thick wire forming a loop on one end of the tube is used to pull the tissue into it (B2). Envelope and tube are fixed to a rod (C) for convenient manipulation.

Figure 4.

 Type II sample holder consists of platelets (A), with a cavity 200 or 500 µm in depth and with a diameter of 1.2 mm (the outer diameter is 2 mm). The filled platelet is tightened with a screw against an abutment in envelope B. This set-up withstands pressure and is small enough to allow good freezing. It is connected to rod C for convenient high-pressure freezing.

Freeze-substitution and ultramicrotomy

After high-pressure freezing, the copper tubes with frozen samples are punched out from the envelope under liquid nitrogen (Fig. 5A). The punched central 5 mm of the tube are cut lengthways with a cutter (Fig. 5B). This cutting is essential to enable good diffusion of the substitution medium into the sample. The samples frozen in type II holders usually stick in their cavity. Cut tubes and type II holders containing the frozen samples are transferred in liquid nitrogen to the freeze-substitution apparatus (AFS, Leica, Vienna) where Eppendorf tubes filled with medium are precooled. As a medium we use acetone (dried over calcium chloride) containing 2% osmium tetroxide (Van Harreveld & Crowell, 1964). The AFS is programmed as follows: 27 h at − 90 °C, heating at a rate of 2 °C h−1 to − 60 °C, 8 h at −60 °C, heating at a rate of 2 °C h−1 to − 30 °C, 8 h at − 30 °C, transfer of the samples to ice (0 °C). After 1 h the specimens are washed three times in anhydrous acetone. The platelets of the type II holder are removed. The sample discs and the copper tubes containing the samples are embedded stepwise in Epon 812 (30%, 70%, 100% resin) at 4 °C. The infiltration times chosen are 3 h for the first two embedding steps and 1 day for the final resin concentration. Polymerization is carried out with fresh resin at 60 °C for 3 days. The polymerized samples are sectioned with diamond knives (Diatome, Biel) on an ultramicrotome (UCT, Leica, Vienna). The copper tubes are sectioned with the sample. The ultrathin sections are collected on 150 mesh and slot grids coated with carbon coated Formvar.

Figure 5.

 Tool A punches the copper tubes under liquid nitrogen containing the frozen samples. A piece of about 5 mm is recovered and for freeze-substitution, the tube is cut up lengthwise still in liquid nitrogen, with tool B. In this tool the 5 mm copper tubes are inserted into a slide, which is pushed through two cutting wheels, opening the tubes lengthways.

Cryosectioning

Cryosectioning of the copper tube sample holder is easy to perform. Once the sample is frozen, the central part of the copper tube (5 mm) is punched out (Fig. 5A) in liquid nitrogen. The remaining little tube containing the frozen sample is transferred in liquid nitrogen to the precooled cryochamber of the cryomicrotome (Reichert FCS, Leica, Vienna). At a temperature of about −170 °C it is fixed there into a specimen holder insert (Fig. 6A) that fits the specimen holder of the cryomicrotome (Fig. 6B). With a trimming diamond (Diatome) the copper is trimmed away and the remaining specimen pyramid is cryosectioned. In the case of the platelet sample holder, the manipulations for cryosectioning are the same as for the traditional machines. By pushing a sharp needle between platelet rim and sample, one can liberate pieces, or the whole sample. These sample pieces are glued (cryoglue according to Richter, 1994) to pins, trimmed and sectioned (Sartori et al., 1993).

Figure 6.

 Tool B is the specimen holder for cryosectioning used in the FCS cryochamber (Leica). Tool A is a chuck into which the punched tubes (Fig. 5A) fit when the chuck is placed into tool B. C represents the tip of a cryofixed copper tube.

Electron microscopy

The cryosection presented here was examined at an acceleration voltage of 120 kV in a Philips CM12 transmission electron microscope on a Gatan cryoholder (Gatan Inc., Warrendale, PA, U.S.A.) at a temperature of approximately − 170 °C. Micrographs were taken at a primary magnification of 10 000× using the low dose device of the electron microscope. Estimated electron dose was 1500–3000 e nm−2 (applied to the region of interest before and during exposure). Focusing was done using the wobbler (no visible movement), setting a defocus of 2–3 µm. Electron diffraction was performed with a camera length of 0.4 m. Image recording was done on Kodak electron microscope film 4489. Brightness was adjusted in order to achieve exposure times of 2 s. The plates were developed for 5 min in Kodak D19 at 20 °C (Sartori et al., 1993).

Ultrathin Epon-sections poststained with uranyl acetate and lead citrate were investigated on a Philips 400 electron microscope. Image recording and developing were performed as described above.

Comparative temperature measurements

A thermocouple device was constructed that is able to measure the temperature drop in the conventional high-pressure freezing machines as well as in the EMPACT. A thermocouple Thermocoax typeZ AB25NN (Philips, Switzerland) 0.5 mm in diameter was soldered into a stainless steel tube (outer diameter 3 mm; inner diameter 0.6 mm). The thermocouple projected out of this tube by 1 mm. The tube was introduced into a solid stainless steel specimen manipulator with which the tube projection could be adjusted (Figs 7a–c). For measuring in the EMPACT the thermocouple had to protrude 28 mm (Figs 7a and b). Measurements on the conventional machine were performed with a HPM-010 (Baltec, Liechtenstein) installed at the Laboratoire d'Analyse Ultrastructurale of the University of Lausanne, Switzerland. To fit into this machine the thermocouple had to protrude only 8 mm (Fig. 7c).

Figure 7.

 (a) Overview of the thermocouple used for comparative measurements, as used for the measurements in the EMPACT. (b) Enlargement of the thermocouple to measure cooling in the Leica EMPACT. (c) The thermocouple retracted to the correct position to measure temperature drop in the Baltec HPM 010 high-pressure freezing machine (same magnification as in b).

Results

Cooling efficiency

The heat transfer coefficient defines the cooling efficiency of any freezing machine (Shimoni & Müller, 1998). This coefficient can be calculated when a defined thermocouple (material, geometry, size) is used for measuring cooling rate. Assuming that the cooling rate measured with a thermocouple is convection limited (i.e. cooling dependent on coolant properties and coolant application), relative cooling efficiencies of different freezing machines can be compared just by comparing the slope when temperature is recorded vs. time: the steeper the slope is, the more efficient is the machine.

Based on the experience (results not shown) that the measured cooling rate of the thermocouple used is coolant flow dependent, its cooling must be convection limited. Figure 8 shows the change in temperature when measured with the identical thermocouple in the HPM-010 and the EMPACT. The slope of the curve is the same in both machines in the temperature range 0° to −50 °C. At lower temperatures the EMPACT is more efficient.

Figure 8.

 Comparative temperature measurements. The slope following the triangles shows the temperature course in the Leica EMPACT high-pressure freezing machine. The circles represent the Baltec HPM high-pressure freezer. In the range 0 to −50 °C cooling efficiency is the same. However cooling of the EMPACT is better at lower temperatures and reaches liquid nitrogen temperature.

Applications

Yeast cells are vitreous when high-pressure frozen in copper tubes in the EMPACT, as shown by the electron diffraction pattern (Fig. 9). The diffuse rings demonstrate the vitreous state of the cell. However distilled water outside the cell is not vitreous. This is obvious because of the so-called Bragg reflections that show up only when ice crystals are formed during freezing. After freeze-substitution the yeast cells are well preserved (Figs 10a and b). Cell wall and cell membranes, the nucleus and vacuoles are easily recognized. The highly ordered array in Fig. 10(a) seems to be an inclusion in a mitochondrion.

Figure 9.

 A yeast cell high-pressure frozen and investigated as a frozen hydrated section shows the cell surrounded by a well preserved cell wall. The nucleus, a mitochondrion, a vacuole and the cell membrane are shown. The electron diffraction pattern (inset) proves the vitreous state of water in the cell. The water outside is not vitreous; this is obvious because of the Bragg reflections (dark points) revealed. Bar represents 1 µm.

Figure 10.

 Yeast cells high-pressure frozen in a copper tube and freeze-substituted in acetone containing 2% osmium tetroxide. The details in (a) and (b) show the cell wall, the cell membrane with the typical invaginations and organelles (probably mitochondria) containing crystalline arrays (a). Bar in overview represents 1.5 µm; in (a) and (b) 0.3 µm.

Figure 11 demonstrates that the cooling efficiency of the EMPACT is well suited for good cryofixation of a whole ivy leaf sample with a diameter of 1.2 mm and a thickness of 200 µm. Thylakoid membranes in the chloroplasts, the membranes of vacuoles, mitochondria, nuclei and nuclear pores as well as the Golgi apparatus are well preserved (Figs 11a–c). The starch granules in the chloroplasts are represented as almost unstained areas. Also 200 µm thick cartilage samples were well frozen (results shown on http://www.ana.unibe.ch/empact). In our hands, yeast cells and ivy leaves are well frozen with a success rate of close to 100%.

Figure 11.

 A cross-section through an ivy leaf, high-pressure frozen in the type II holder and freeze-substituted (bar represents 25 µm). As the details in (a)–(c) show, the leaf is well frozen throughout; there are no segregation patterns. Membranes, mitochondria, Golgi apparatus, nucleus and nuclear pores as well as chloroplast are well preserved. Bar represents 0.5 µm.

Micrographs of rat liver (Figs 12a and b) and peripheral nerve (Fig. 13) are shown just to illustrate good freezing of different samples. Large parts of liver and peripheral nerve are frozen without detectable damage by ice crystal formation. Liver was frozen in the type II holders and the nerve in the copper tube. Both systems give comparable good results.

Figure 12.

 Liver high-pressure frozen in platelets and processed by freeze-substitution is well preserved (bar represents 2.5 µm). Also at high magnification (a) no segregation patterns are visible. Bar represents 0.4 µm.

Figure 13.

 Peripheral nerve high-pressure frozen in a copper tube and processed by freeze-substitution shows excellent preservation. An unmyelinated axon at the top right and myelinated axons are depicted. The myelin layers are perfectly preserved. Bar represents 0.5 µm.

Discussion

The new EMPACT high-pressure freezing machine freezes well. Vitrification was proven for some biological samples with electron diffraction after cryosectioning. Good freezing was documented for the platelet holder and vitrification for the copper tube holder. The cooling efficiency is as good as in the conventional high-pressure freezing machines. This was shown by the comparative temperature recordings with an identical thermocouple during cooling.

For theoretical reasons, the physics of cooling (i.e. heat conductivity within the sample) limits the sample size in any freezing device. According to published work and our experience, cooling efficiency of all commercially available high-pressure freezing machines (including the EMPACT) seems to be good enough to freeze well large portions of a variety of disc-shaped biological samples up to a thickness of about 200 µm, and rods about 300 µm in diameter (copper tube sample holder EMPACT). At this sample thickness, the cooling rate in the centre of the specimen is conduction limited and is therefore defined by the sample properties (Studer et al., 1995; Shimoni & Müller, 1998). Improving the cooling efficiency of high-pressure freezing machines would only be reasonable if the sample size is reduced. However, to leave at least some cells of a piece of tissue intact, we cannot reduce sample size ad infinitum. An ‘average’ cell sample about 200 µm in thickness consists of about 10 cell layers. When we reduce sample thickness to, e.g. 100 µm, there are only a few cells that are not harmed by excision. With such small samples, we would encounter the same problems as in slam freezing (Escaig, 1982), where only the surface of an excised tissue sample is well frozen and therefore mostly damaged cells are investigated. For this reason we think that sample size should not be reduced when tissues are investigated. We have learned to accept from using conventional machines as well as the EMPACT that there are 200 µm thick samples that are well frozen only partially when high-pressure freezing is applied. Based on our experience, the concentration of the inherent cryoprotectants in the specimens (e.g. carbohydrates, proteins, ions and nucleic acids) seems to be the most important and critical factor for producing well frozen samples.

It was not shown until now that a sample is well frozen when cooled with a jet of liquid nitrogen pressurized to approximately 10 bar. With the EMPACT we have shown that well precooled feed pipes and a relatively small pressure applied to liquid nitrogen are sufficient to allow efficient cooling. The EMPACT is able to freeze efficiently with a jet of liquid nitrogen driven with a pressure of approximately 10 bar. The main reason is, in our opinion, that at 10 bar pressure the boiling point of liquid nitrogen is raised to about −170 °C. The use of a cooled cryogen (e.g. isopentane) instead of liquid nitrogen would improve cooling efficiency. This, however, is under current safety regulations very difficult to build and it would only be reasonable when thin samples (thinner than 100 µm) with a high water content should be well cryofixed.

More important, however, is improving the sample preparation. First attempts in this direction have been made (Hohenberg et al., 1996; Shimoni & Müller, 1998), but in our opinion these improved methods are not fully developed. Future developments have to include a biopsy system that is easy to handle and with which the excised piece of tissue is high-pressure frozen within a short time period (less than 30 s). This is important for investigating freshly excised human tissue close to its native state, which is available during surgery or when biopsies are taken for diagnosis.

In conclusion, we are convinced that the versatility (two sample holders) and handiness (easy transport close to or into an operating theatre) of the EMPACT will help to exploit better the potential of high-pressure freezing in biomedical research.

Acknowledgements

We are most grateful to Dr Matthias Chiquet for revising the manuscript. We wish to thank Mr E. Mühlheim, Mr U. Rohrer (M.E. Müller Institute of Biomechanics) and Mr C. Lehmann (Institute of Anatomy) for technical constructions and Dr G. Hermann and Mrs B. Krieger (both Institute of Anatomy) for technical assistance. We also thank all collaborators of Leica Microsystems (Vienna) involved in this project for support. The first author is grateful for being generously supported by Professor J. Dubochet (LAU, Université de Lausanne) and Professor E. B. Hunziker (M.E. Müller Institute of Biomechanics, University of Bern).

Ancillary