Introduction

Investigation of the fine structure of living cells is important for understanding many intracellular processes. Adequate spatial resolution of structures involved in subcellular processes can be achieved using transmission electron microscopy (TEM). For rapid processes, temporal resolution is also important (Van Harreveld et al. 1974, Heuser et al. 1976, Heuser et al. 1979). However, TEM often fails to preserve the precise structural information about dynamic cellular processes, because conventional processing consists of slow processes such as chemical fixation and dehydration at room temperature. The limitations of chemical fixation can largely be overcome using the cryofixation technique. The advantage of cryofixation over conventional processing lies mainly in the extremely rapid physical fixation of the living specimen. The time required for cryofixation by vitrification (absence of crystalline ice in a specimen) is estimated to be less than 0.1 ms, which is substantially less (by a factor of 10⁴) than that required for chemical fixation by infiltration with aldehydes or heavy metal compounds (Sitte et al. 1987).

It was for a long time difficult to achieve adequate freezing of plant materials, except in the case of single-celled algae (Mita et al. 1986). However, Usukura et al. (1983) described a manually operated liquid helium-cooled cryofixation apparatus, and this apparatus has been subsequently used to achieve high spatial and temporal resolution in several studies of dynamic cellular processes in plants (Shojima et al. 1987, Nishizawa and Mori 1989, Nishizawa et al. 1990, Nishizawa et al. 1994, Nagatani et al. 1993) and for localization of antigens, especially water-soluble substances, by TEM immunocytochemistry (Shojima et al. 1987, Nishizawa et al. 1990, Nishizawa et al. 1994). Besides, the rapid-freeze technique was also used in combination with deep-etching to examine the cell wall architecture of both cultured cells and pea epidermal cells (McCann et al. 1990, Itoh and Ogawa 1993, Fujino and Itoh 1998).

In addition, a method for cryofixation at high pressure (approx. 210 MPa) (Müller and Moore 1984) has been reported to be well suited for plant specimens (Kiss et al. 1990, Staehelin et al. 1990, Galway et al. 1993, Samuels et al. 1995, Lonsdale et al. 1999). Recently, using this technique, a novel kind of cell plate involved in endosperm cellularization was characterized (Otegui and Staehelin 2000), and precise ultrastructural information of nodal endoplasmic reticulum of columella root cap cells was gained (Zheng and Staehelin 2001).

Despite the fact that the cryofixation method at atmospheric pressure using liquid helium has provided excellent preparations, very few studies have so far dealt with material from higher plants. This is mainly because methods using apparatus with a complicated manual setup is extremely time-consuming and gives poor reproducibility. Considerable expertise has therefore been necessary for the preparation of high-quality sections for TEM. To overcome these problems, we have designed an automated device for cryofixation at liquid helium temperature which is simple to operate and therefore of great potential for use by non-specialists in diverse fields of biology.

In this paper, we introduce this new instrument to plant biologists and show how fine structure is preserved in cryofixed samples. Use of this instrument will significantly reduce the labor time and the cost for the preparation of top-quality specimens for TEM.

Results and Discussion

Performance of the automated device

Automation of the rapid freezing device using liquid helium has four main advantages over manual operation, namely (1) simple operation requiring no specific skill, (2) a high reproducibility of cryofixation, (3) reduction of time necessary for preparation of multiple samples, and (4) significant reduction in running costs.

First, we can put a great emphasis on extremely easy operation of this automated cryofixation device. It is possible to carry out cryofixation of living samples without skill or practice, by simply pushing the start button of this device (Fig. 1). The most important feature of the automatic device is the rapid, automated transfer of the specimen from metal-contact at liquid helium temperature (Fig. 1B) to storage in liquid nitrogen (Fig. 1C).

Second, the automatic operation enabled cryofixation to be performed on multiple samples under the same conditions. Considering that one of the most important factors to preserve a fine subcellular structure was the strength of impact at the time of metal contacting (Sitte et al. 1987, Robards 1991), this was controlled mechanically by a spring on the inside of the plunger (Fig. 1B, j) and also by adjusting the amount of using air with electromagnetic valves for the plunger injector (Fig. 1B, q). These control points allow the impact at metal-contact to be properly adjusted according to the properties of the samples. In addition to the samples of plant material reported here, we have also confirmed the high reproducibility of freezing quality in other biological specimens including mutants of Arabidopsis, cultured animal cells, and mice (data not shown). In all cases that the depth of high-quality freezing was essentially the same as that for the plant material described above (Fig. 3, 4). Cells from the liver and small intestine of mice were examined as relatively uniform samples to evaluate the reproducibility of cryofixation. Eight pieces of each organ were frozen using two copper blocks and we estimated the depth of high quality fixation from the contact plane in each specimen. As a result, 100% specimens showed smooth membranes including the plasma membrane and mitochondria in the region within around 10 µm from the contact plane.

Finally, incorporation of an automatic block rotator (Fig. 1C, u; Fig. 2A and 2B, f) also significantly reduces the handling time needed for sequential freezing of multiple samples, thus reducing the cost of the experiment. In order to maintain good thermal conductivity of the contact plane, the surface of the cooled block must be completely clean before each cryofixation. If the block is replaced for each specimen, a large amount of liquid helium is vaporized, resulting in low efficiency of coolant use and loss of time due to the need for more frequent refilling. The block rotator automatically turned the copper block (Fig. 1C, b and c; Fig. 2A and 2B, a and d) by one-eighth after each contact with a specimen, so that a clear surface of the metal became ready immediately for the next specimen. We were thus able to carry out uninterrupted cryofixation of eight specimens using one copper block, avoiding the need for removal of frost and cleaning of the apparatus between samples, and saving a substantial amount of time. Initial setup of the device and filling with coolant took about 20 min, while the automated operation including metal contacting (Fig. 1B), storing the specimen in liquid nitrogen (Fig. 1C) and block rotation (Fig. 1C), was completed within a few s. After freezing the eight specimens, it took 10 min to prepare the subsequent block and to refill the coolant, before being able to resume cryofixation.

In addition to saving time, the use of automation to prepare eight specimens on one block substantially reduced the volume of liquid helium required, compared to processing of individual samples. Initial cooling of the device from room temperature to around –270°C required 6 liters of liquid nitrogen and 20 liters of liquid helium. Processing of each subsequent block required an additional 2 to 3 liters of liquid helium. More than hundred specimens can therefore be prepared with a standard 50-liter tank of liquid helium. Increase the number of specimens frozen on one copper block by eight times allowed relative cost saving per specimen. And it was possible to operate the device continuously throughout the day, which resulted in the most economical use of coolants.

Evaluation of the quality of rapid freezing

We investigated the state of ultrastructure preservation in Arabidopsis root-tip specimens prepared using the automated cryofixation device. Electron micrographs of the resultant samples show that the highest quality preservation of ultrastructure occurred within 5 µm of the contact plane (Fig. 3A and 3B). In this region, the plasma membrane (Fig. 3B, PM) and membranes of other organelles including mitochondria (Fig. 3B, M), endoplasmic reticulum (Fig. 3B, ER), vacuoles (Fig. 3B, V) and Golgi bodies (Fig. 3B, G) were smooth. Vacuoles showed a complicated form and there were amorphous network and several electron-dense spots within the vacuoles (Fig. 3A and 3B, V). The round shaped structures were electron-dense and surrounded by membrane, and ribosomes were parallel to the outer side of this membrane (Fig. 3B, arrows) as also seen in the rough endoplasmic reticulum (Fig. 3B, ER). This structure possibly appeared to be spindle-shape (Fig. 3A, arrows). Several such round- or spindle-shaped structures were seen (Fig. 3A), ranging in diameter from about one-third to one micron. These structures are most likely dilated cisternae of the endoplasmic reticulum, common in members of the Brassicaceae (Gunning 1998). The cryofixation method also provided resolution within the round shaped structure. The region close to the membrane was low in electron density, whereas the central region was significantly more electron-dense (Fig. 3B, arrows) compared to the endoplasmic reticulum (Fig. 3B, ER).

The zone around 10 µm from the contact plane was still well preserved (Fig. 3D), but the zone around 20 µm from the contact plane showed visible damage that indicates the formation of ice crystals (Fig. 3E). High-quality freezing beyond a depth of 15–20 µm has only rarely been reported (Robards 1991) and we therefore conclude that the performance of the automated device is comparable to manual cryofixation.

Application to dynamic subcellular processes

Finally, we examined the efficiency of the automated cryofixation device for capturing rapidly-moving subcellular phenomena in plant cells, using two well-known examples. Subcellular processes in papillar cells of Brassica campestris were previously described from samples prepared using a manual metal-contacting device at liquid helium temperature (Nishizawa et al. 1990). For the purposes of comparison, we therefore examined papillar cells of the stigma from Arabidopsis flowers. Microphotographs of the papillar cells showed clear images in which we observed ribosomes, Golgi bodies, mitochondria, chloroplasts, vesicles, and vacuoles (Fig. 4A), all with well preserved fine structure. Subcellular movements such as secretion, budding, and membrane fusion were also captured (Fig. 4). The Golgi stacks contained cisternae, transfer vesicles and secretory vesicles (Fig. 4B). Vesicles close to the plasma membrane seemed to be exocytosis (Fig. 4C). The quality of these samples prepared using the new automated device was equivalent to that seen in samples prepared by the previously reported manual technique.

We also observed semi-serial thin sections of chloroplasts in papillar cells (Fig. 4E, 4F). The chloroplast was observed as two individual parts (Fig. 4E), joined by a narrow connecting band (Fig. 4F). Within the band, we observed a filamentous structure bounded by an electron-dense layer (Fig. 4G, arrows). These complicated forms were interpreted to be chloroplasts undergoing division (Pyke 1999) and the electron-dense structure in the narrow part (Fig. 4G) may be equivalent to the plastid-dividing ring that was previously reported in single cell algae prepared using the rapid-freezing technique (Mita et al. 1986). The occurrence of a plastid dividing ring during chloroplast division in higher plants has been observed by conventional chemical fixation methods (Robertson et al. 1996, Pyke 1999). However, the rapid freezing technique enabled us to capture the dynamic changes in the shape of the chloroplast with a high degree of temporal resolution.

It has recently been shown by green fluorescent protein (GFP)-fusion (Kircher et al. 1999) and immunocytochemical techniques (Hisada et al. 2000) that phytochrome A translocates quickly from the cytosol to the nucleus upon light exposure. Subcellular movement of phytochrome A has been reported to commence within 1 min of exposure to red light (Hisada et al. 2000). Such a movement is too rapid to be captured by conventional chemical fixation methods. An additional problem is that phytochrome A is a soluble protein (Butler et al. 1959), and its subcellular localization is therefore highly sensitive to disturbance during the fixation process (Pratt and Coleman 1974). We therefore used the rapid-freezing technique to prepare specimens for visualizing the subcellular localization of phytochrome A apoprotein (PHYA) by immunocytochemistry in electron micrographs of pea hook cells.

Endogenous pea PHYA was specifically detected by immunogold-labeling in ultrathin sections, using PHYA-deficient fun1-1 mutant (far-red unresponsive, Weller et al. 1997) as a negative control. When etiolated seedlings were exposed to continuous red light (cR) for 30 min, small aggregations of immuno-gold appeared in the nucleus (Fig. 5A). Using 5 nm immunogold particles, the aggregated form was seen in electron-dense areas of around 50 to 100 nm in diameter (Fig. 5B). This result is consistent with previous reports that PHYA protein (Hisada et al. 2000) and GFP-PHYA fusion protein (Kircher et al. 1999) both show a speckled distribution within the nucleus under cR light. These observations also confirm the intra-nuclear localization of PHYA, since in previous studies using optical microscopy it was difficult to prove that the speckling form of immuno-fluorescence associated with the nucleus was truly appearing inside it. Further TEM studies of the rapid translocation of PHYA will benefit greatly from the high spatial and temporal resolution provided by the automated rapid-freezing device described here.

Metal-contact freezing technique vs. high pressure freezing method

Because of the high pressure, a lower cooling rate was acceptable and this method has therefore made it possible to prepare considerably thicker and larger cryo-specimens (maximum thickness 600 µm by double-sided cooling) without formation of ice crystals (Sitte et al. 1987). A cooling rate can be achieved 5,000 K s^–1 at the surface of a specimen, but is not more than a few hundred in the center region (Moor 1987). Because temporal resolution of cryofixation depends on the cooling rate, high pressure freezing is slower process with poor time resolution than the freezing at atmospheric pressure (Sitte et al. 1987). This method has therefore been widely used for examination of relatively large specimens, where a high degree of temporal resolution is not required. Conversely, the liquid helium-cooled, metal-contact cryofixation at atmospheric pressure was necessarily implemented at the high cooling rates of approx. 10⁶ K s^–1 and therefore achieves vitrification with high temporal resolution but only in a restricted area. The choice of cryofixation method should thus take into account the distinct advantages of these two methods.

Future prospects

In addition to the automation of the process of metal-contact freezing, it will potentially be useful to extend the automation process to include steps before or after cryofixation. For example, an automatic apparatus on the plunger of the cryofixation device could be added, in order to apply specific treatments (such as chemical or light treatments) to living cells on the specimen holder. Use of such an apparatus has been reported in the time-resolved analysis of interaction between myosin subfragment and actin filaments, which employed metal-contact liquid helium cryofixation immediately following photolysis of caged ATP (Funatsu et al. 1993). In these experiments, apparatus for control of the ultraviolet light pulse necessary for photolysis were constructed in the plunger of the cryofixation device (Funatsu et al. 1993). With further modifications, our rapid-freezing device will be valuable for these kinds of time-resolved investigations of various biological responses. More generally, the automatic cryofixation device will be a powerful tool for future analyses of dynamic events, for localization of proteins and water-soluble substances by immunocytochemistry, and for detecting specific elements by X-ray microanalysis.

Materials and Methods

Plant material

Seeds of Arabidopsis thaliana (ecotype Landsberg erecta) were incubated on filter paper saturated with tap-water for 2 d at 23°C under white light (fluorescent tube, FL40SW-B, Hitachi, Tokyo). Root tips were excised from germinating seeds by a razor blade and the root cap was mounted on the sample holder (Fig. 1A, a) which is brought into direct contact with the cooled block for freezing (Fig. 1B, a and b).

Seedlings of Arabidopsis were further cultured for 1 month under continuous white light at 23°C. Stigmas were then excised from flowers by a razor blade and the tip was mounted on the sample holder as above.

Seeds of wild-type pea (Pisum sativum cv. Torsdag) and the isogenic phyA-deficient fun1-1 mutant (for far-red unresponsive, Weller et al. 1997) were imbibed in water for 6 h in darkness and grown on vermiculite saturated with water at 23°C in darkness. Five-day-old seedlings were exposed to continuous red light for 30 min. Red light was obtained from fluorescent tubes (FL-20S Re-66; Toshiba, Tokyo) filtered through 3-mm thick red acrylic plate (Acrylight K5-102; Mitsubishi Rayon, Tokyo), 3-mm thick scattering filter (Acrylight K5-001E; Mitsubishi, Rayon), and 3-mm thick white glass. The light intensity measured at 660 nm was 55 µmol m^–2 s^–1. Immediately after the light irradiation, hook regions were excised from the epicotyl and the epidermis was mounted on the sample holder as above.

Automated cryofixation

Automated operation of the cryofixation device is illustrated in Fig. 1. Prior to automated steps for cryofixation, the contacting surface of the copper block (Fig. 1A, b) was polished, rinsed in acetone, exposed to dry air and settled on the basement block (Fig. 1A, c) in the Dewar flask. When the copper block was handled, the block port (Fig. 1A, d) was positioned right above the copper block (Fig.1A, b). After this operation, the plunger guide pipe (Fig. 1A, e) replaced with the block port, sifting horizontally by 10 mm from the center of copper block (Fig. 1A, b). The outer Dewar flask (Fig. 1A, f) was filled with liquid nitrogen and the inner one (Fig. 1A, g) was filled with liquid helium up to the neck of the basement copper block (Fig. 1A, c) using transfer tube (Fig. 1A, h) connected to the liquid helium tank. While pouring the coolant, evaporating gas was allowed to escape through the valves (Fig. 1A, i).

A fresh specimen mounted on the hollow of the specimen holder (Fig. 1A, a) was placed on the tip of the plunger (Fig. 1A, j) that was inclined (Fig. 1A, arrow*) to enable this manual operation, and then returned to the vertical position (Fig. 1A, arrow**) before the start of the rapid freezing steps. After turning on the switch button (Fig. 1A, k) of the controller (Fig. 1A, l), the following steps proceeded automatically in sequence (Fig. 1B and 1C); namely, (1) the plunger (Fig. 1B, j) was inserted into the guide pipe (Fig. 1B, e) and when the shutter (Fig. 1B, m) of the guide pipe opened, warm air inside the guide pipe (Fig. 1B, e) was ejected out from the device through the air escape (Fig. 1B, n); (2) the sample on the holder (Fig. 1B, a) was rapidly frozen by contacting to the surface of the chilled copper block (Fig. 1B, b); (3) the plunger (Fig. 1C, j) rose to the initial position and the shutter (Fig. 1C, m) closed immediately; (4) the storage bottle (Fig. 1C, o) filled with liquid nitrogen moved just under the plunger (Fig. 1C, j, arrow***) and the specimen on the holder (Fig. 1C, a) was immediately pushed into liquid nitrogen; (5) the copper block (Fig. 1C, b and c) turned by one-eighth (arrows****) so that the clear surface was ready for the next sample (Fig. 2).

After eight samples were consecutively cryofixed, the upper block (Fig. 1C, b) was taken out using the block holder through the block port (Fig. 1C, d). Then, a freshly polished copper block was replaced on the basement block, and the Dewar flasks were refilled with coolant.

Freeze substitution and resin embedding

After cryofixation of fresh specimens at liquid helium temperature using the automated device described above, the frozen samples were kept on the holder (Fig. 1C, a) and transferred from liquid nitrogen into the solution for freeze substitution using forceps. Freeze substitution was performed in acetone containing 4% osmium tetroxide for Arabidopsis specimens, or in acetone containing 0.05% osmium tetroxide for pea specimens at –80°C for 2 d.

For Arabidopsis, specimens were kept at –20°C for 3 h, at 4°C for 3 h and at room temperature for 1 h, washed with acetone, and embedded in the epon-araldite mixed resin. Pea specimens were kept at –20°C for 3 h, at 4°C for 1 h and at room temperature for 1 h, washed with acetone and ethanol, and embedded in LR White resin (London Resin, Berkshire, U.K.).

Electron microscopy

Ultrathin sections were prepared using an ultramicrotome (Sorvall^® MT-6000, Du Pont Company, Wilmington, Delaware, U.S.A.). The sections of Arabidopsis root tip and stigma were stained with uranyl acetate and lead citrate, and examined in a TEM (H-7100, Hitachi Ltd., Tokyo).

Immunolabeling

The ultrathin sections of peas were subjected to immunolabeling using polyclonal anti-pea phytochrome A apoprotein (PHYA) antibody (pAP) and gold-conjugated (particle size 5 or 10 nm) anti-rabbit IgG antibody (AuroProbe EM; Amersham Pharmacia Biotech., Uppsala, Sweden). Immunolabeling was performed as follows: (1) sections were treated in phosphate-buffered saline (PBS) containing 4% bovine serum albumin (BSA) for 30 min to prevent non-specific binding; (2) incubation overnight at 4°C in anti-PHYA-antibody (final concentration: 84 ng ml^–1) in PBS; (3) a rinse in PBS containing 4% BSA, three rinses in PBS containing 0.05% Tween20 (Bio-Rad, Hercules, CA, U.S.A.) every 5 min, and a rinse in PBS containing 4% BSA for 5 min; (4) incubation for 2 h in anti-rabbit IgG antibody; (5) six more rinses in PBS; and (6) treatment with 2% glutaraldehyde in distilled water to fix the immunolabeling. Sections were stained with uranyl acetate.

Acknowledgements

We are grateful to James L. Weller for critical reading of this manuscript, Hisafumi Ohtsuka (Hitachi, Ltd., Instrument Division) for developing the new technology, Shoukichi Matsunami and Wataru Moriya (Hitachi Advanced Research Laboratory, Technical Support Center) for manufacturing of the device, and Sumiko Yabe for technical supporting. We thank to James B. Reid for providing pea seeds, and Akira Nagatani and Hiroko Hanzawa for providing anti-phytochrome A antibodies. This work was supported by Hitachi Advanced Research Laboratory projects (B2023) and a grant from the Program for Promotion of Basic Research Activity for Innovative Biosciences to M. F.